diff --git a/Makefile.in b/Makefile.in index 4adb76ddaa17..c63dc242e79a 100644 --- a/Makefile.in +++ b/Makefile.in @@ -1,644 +1,652 @@ # # SPDX-License-Identifier: BSD-2-Clause # # Copyright (c) 2018-2024 Gavin D. Howard and contributors. # # Redistribution and use in source and binary forms, with or without # modification, are permitted provided that the following conditions are met: # # * Redistributions of source code must retain the above copyright notice, this # list of conditions and the following disclaimer. # # * Redistributions in binary form must reproduce the above copyright notice, # this list of conditions and the following disclaimer in the documentation # and/or other materials provided with the distribution. # # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" # AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE # IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE # ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE # LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR # CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF # SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS # INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN # CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) # ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE # POSSIBILITY OF SUCH DAMAGE. # # %%WARNING%% # .POSIX: ROOTDIR = %%ROOTDIR%% INCDIR = $(ROOTDIR)/include SRCDIR = $(ROOTDIR)/src TESTSDIR = $(ROOTDIR)/tests SCRIPTSDIR = $(ROOTDIR)/scripts GENDIR = $(ROOTDIR)/gen BUILDDIR = %%BUILDDIR%% SRC = %%SRC%% OBJ = %%OBJ%% GCDA = %%GCDA%% GCNO = %%GCNO%% BC_ENABLED_NAME = BC_ENABLED BC_ENABLED = %%BC_ENABLED%% DC_ENABLED_NAME = DC_ENABLED DC_ENABLED = %%DC_ENABLED%% HEADERS = $(INCDIR)/args.h $(INCDIR)/file.h $(INCDIR)/lang.h $(INCDIR)/lex.h $(INCDIR)/num.h $(INCDIR)/opt.h $(INCDIR)/parse.h $(INCDIR)/program.h $(INCDIR)/read.h $(INCDIR)/status.h $(INCDIR)/vector.h $(INCDIR)/vm.h BC_HEADERS = $(INCDIR)/bc.h DC_HEADERS = $(INCDIR)/dc.h HISTORY_HEADERS = $(INCDIR)/history.h EXTRA_MATH_HEADERS = $(INCDIR)/rand.h LIBRARY_HEADERS = $(INCDIR)/bcl.h $(INCDIR)/library.h GEN_DIR = %%GEN_DIR%% GEN = %%GEN%% GEN_EXEC = $(GEN_DIR)/$(GEN) GEN_C = $(GENDIR)/$(GEN).c GEN_EMU = %%GEN_EMU%% BC_LIB = $(GENDIR)/lib.bc BC_LIB_C = $(GEN_DIR)/lib.c BC_LIB_O = %%BC_LIB_O%% BC_LIB_GCDA = $(GEN_DIR)/lib.gcda BC_LIB_GCNO = $(GEN_DIR)/lib.gcno BC_LIB2 = $(GENDIR)/lib2.bc BC_LIB2_C = $(GEN_DIR)/lib2.c BC_LIB2_O = %%BC_LIB2_O%% BC_LIB2_GCDA = $(GEN_DIR)/lib2.gcda BC_LIB2_GCNO = $(GEN_DIR)/lib2.gcno BC_HELP = $(GENDIR)/bc_help.txt BC_HELP_C = $(GEN_DIR)/bc_help.c BC_HELP_O = %%BC_HELP_O%% BC_HELP_GCDA = $(GEN_DIR)/bc_help.gcda BC_HELP_GCNO = $(GEN_DIR)/bc_help.gcno DC_HELP = $(GENDIR)/dc_help.txt DC_HELP_C = $(GEN_DIR)/dc_help.c DC_HELP_O = %%DC_HELP_O%% DC_HELP_GCDA = $(GEN_DIR)/dc_help.gcda DC_HELP_GCNO = $(GEN_DIR)/dc_help.gcno BIN = bin EXEC_SUFFIX = %%EXECSUFFIX%% EXEC_PREFIX = %%EXECPREFIX%% BC = bc DC = dc BC_EXEC = $(BIN)/$(EXEC_PREFIX)$(BC) DC_EXEC = $(BIN)/$(EXEC_PREFIX)$(DC) +BC_FUZZER = $(BIN)/$(BC)_fuzzer_c +BC_FUZZER_C = $(BIN)/$(BC)_fuzzer_C +DC_FUZZER = $(BIN)/$(DC)_fuzzer_c +DC_FUZZER_C = $(BIN)/$(DC)_fuzzer_C BC_TEST_OUTPUTS = tests/bc_outputs BC_FUZZ_OUTPUTS = tests/fuzzing/bc_outputs1 tests/fuzzing/bc_outputs2 tests/fuzzing/bc_outputs3 DC_TEST_OUTPUTS = tests/dc_outputs DC_FUZZ_OUTPUTS = tests/fuzzing/dc_outputs LIB = libbcl LIB_NAME = $(LIB).a LIBBC = $(BIN)/$(LIB_NAME) BCL = bcl BCL_TEST = $(BIN)/$(BCL) BCL_TEST_C = $(TESTSDIR)/$(BCL).c MANUALS = manuals BC_MANPAGE_NAME = $(EXEC_PREFIX)$(BC)$(EXEC_SUFFIX).1 BC_MANPAGE = $(MANUALS)/$(BC).1 BC_MD = $(BC_MANPAGE).md DC_MANPAGE_NAME = $(EXEC_PREFIX)$(DC)$(EXEC_SUFFIX).1 DC_MANPAGE = $(MANUALS)/$(DC).1 DC_MD = $(DC_MANPAGE).md BCL_MANPAGE_NAME = bcl.3 BCL_MANPAGE = $(MANUALS)/$(BCL_MANPAGE_NAME) BCL_MD = $(BCL_MANPAGE).md MANPAGE_INSTALL_ARGS = -Dm644 BINARY_INSTALL_ARGS = -Dm755 PC_INSTALL_ARGS = $(MANPAGE_INSTALL_ARGS) BCL_PC = $(BCL).pc PC_PATH = %%PC_PATH%% BCL_HEADER_NAME = bcl.h BCL_HEADER = $(INCDIR)/$(BCL_HEADER_NAME) %%DESTDIR%% BINDIR = %%BINDIR%% INCLUDEDIR = %%INCLUDEDIR%% LIBDIR = %%LIBDIR%% MAN1DIR = %%MAN1DIR%% MAN3DIR = %%MAN3DIR%% MAIN_EXEC = $(EXEC_PREFIX)$(%%MAIN_EXEC%%)$(EXEC_SUFFIX) EXEC = $(%%EXEC%%) NLSPATH = %%NLSPATH%% BC_BUILD_TYPE = %%BUILD_TYPE%% BC_ENABLE_LIBRARY = %%LIBRARY%% BC_ENABLE_HISTORY = %%HISTORY%% BC_ENABLE_EXTRA_MATH_NAME = BC_ENABLE_EXTRA_MATH BC_ENABLE_EXTRA_MATH = %%EXTRA_MATH%% BC_ENABLE_NLS = %%NLS%% BC_EXCLUDE_EXTRA_MATH = %%EXCLUDE_EXTRA_MATH%% BC_ENABLE_AFL = %%FUZZ%% +BC_ENABLE_OSSFUZZ = %%OSSFUZZ%% BC_ENABLE_MEMCHECK = %%MEMCHECK%% +LIB_FUZZING_ENGINE = %%LIB_FUZZING_ENGINE%% + BC_DEFAULT_BANNER = %%BC_DEFAULT_BANNER%% BC_DEFAULT_SIGINT_RESET = %%BC_DEFAULT_SIGINT_RESET%% DC_DEFAULT_SIGINT_RESET = %%DC_DEFAULT_SIGINT_RESET%% BC_DEFAULT_TTY_MODE = %%BC_DEFAULT_TTY_MODE%% DC_DEFAULT_TTY_MODE = %%DC_DEFAULT_TTY_MODE%% BC_DEFAULT_PROMPT = %%BC_DEFAULT_PROMPT%% DC_DEFAULT_PROMPT = %%DC_DEFAULT_PROMPT%% BC_DEFAULT_EXPR_EXIT = %%BC_DEFAULT_EXPR_EXIT%% DC_DEFAULT_EXPR_EXIT = %%DC_DEFAULT_EXPR_EXIT%% BC_DEFAULT_DIGIT_CLAMP = %%BC_DEFAULT_DIGIT_CLAMP%% DC_DEFAULT_DIGIT_CLAMP = %%DC_DEFAULT_DIGIT_CLAMP%% RM = rm MKDIR = mkdir SCRIPTS = ./scripts MINISTAT = ministat MINISTAT_EXEC = $(SCRIPTS)/$(MINISTAT) BITFUNCGEN = bitfuncgen BITFUNCGEN_EXEC = $(SCRIPTS)/$(BITFUNCGEN) INSTALL = $(SCRIPTSDIR)/exec-install.sh SAFE_INSTALL = $(SCRIPTSDIR)/safe-install.sh LINK = $(SCRIPTSDIR)/link.sh MANPAGE = $(SCRIPTSDIR)/manpage.sh KARATSUBA = $(SCRIPTSDIR)/karatsuba.py LOCALE_INSTALL = $(SCRIPTSDIR)/locale_install.sh LOCALE_UNINSTALL = $(SCRIPTSDIR)/locale_uninstall.sh VALGRIND_ARGS = --error-exitcode=100 --leak-check=full --show-leak-kinds=all --errors-for-leak-kinds=all TEST_STARS = *********************************************************************** BC_NUM_KARATSUBA_LEN = %%KARATSUBA_LEN%% BC_DEFS0 = -DBC_DEFAULT_BANNER=$(BC_DEFAULT_BANNER) BC_DEFS1 = -DBC_DEFAULT_SIGINT_RESET=$(BC_DEFAULT_SIGINT_RESET) BC_DEFS2 = -DBC_DEFAULT_TTY_MODE=$(BC_DEFAULT_TTY_MODE) BC_DEFS3 = -DBC_DEFAULT_PROMPT=$(BC_DEFAULT_PROMPT) BC_DEFS4 = -DBC_DEFAULT_EXPR_EXIT=$(BC_DEFAULT_EXPR_EXIT) BC_DEFS5 = -DBC_DEFAULT_DIGIT_CLAMP=$(BC_DEFAULT_DIGIT_CLAMP) BC_DEFS = $(BC_DEFS0) $(BC_DEFS1) $(BC_DEFS2) $(BC_DEFS3) $(BC_DEFS4) $(BC_DEFS5) DC_DEFS1 = -DDC_DEFAULT_SIGINT_RESET=$(DC_DEFAULT_SIGINT_RESET) DC_DEFS2 = -DDC_DEFAULT_TTY_MODE=$(DC_DEFAULT_TTY_MODE) DC_DEFS3 = -DDC_DEFAULT_PROMPT=$(DC_DEFAULT_PROMPT) DC_DEFS4 = -DDC_DEFAULT_EXPR_EXIT=$(DC_DEFAULT_EXPR_EXIT) DC_DEFS5 = -DDC_DEFAULT_DIGIT_CLAMP=$(DC_DEFAULT_DIGIT_CLAMP) DC_DEFS = $(DC_DEFS1) $(DC_DEFS2) $(DC_DEFS3) $(DC_DEFS4) $(DC_DEFS5) CPPFLAGS1 = -D$(BC_ENABLED_NAME)=$(BC_ENABLED) -D$(DC_ENABLED_NAME)=$(DC_ENABLED) CPPFLAGS2 = $(CPPFLAGS1) -I$(INCDIR)/ -DBUILD_TYPE=$(BC_BUILD_TYPE) %%LONG_BIT_DEFINE%% CPPFLAGS3 = $(CPPFLAGS2) -DEXECPREFIX=$(EXEC_PREFIX) -DMAINEXEC=$(MAIN_EXEC) CPPFLAGS4 = $(CPPFLAGS3) %%BSD%% %%APPLE%% CPPFLAGS5 = $(CPPFLAGS4) -DBC_NUM_KARATSUBA_LEN=$(BC_NUM_KARATSUBA_LEN) CPPFLAGS6 = $(CPPFLAGS5) -DBC_ENABLE_NLS=$(BC_ENABLE_NLS) CPPFLAGS7 = $(CPPFLAGS6) -D$(BC_ENABLE_EXTRA_MATH_NAME)=$(BC_ENABLE_EXTRA_MATH) CPPFLAGS8 = $(CPPFLAGS7) -DBC_ENABLE_HISTORY=$(BC_ENABLE_HISTORY) -DBC_ENABLE_LIBRARY=$(BC_ENABLE_LIBRARY) -CPPFLAGS = $(CPPFLAGS8) -DBC_ENABLE_MEMCHECK=$(BC_ENABLE_MEMCHECK) -DBC_ENABLE_AFL=$(BC_ENABLE_AFL) +CPPFLAGS9 = $(CPPFLAGS8) -DBC_ENABLE_MEMCHECK=$(BC_ENABLE_MEMCHECK) -DBC_ENABLE_AFL=$(BC_ENABLE_AFL) +CPPFLAGS = $(CPPFLAGS9) -DBC_ENABLE_OSSFUZZ=$(BC_ENABLE_OSSFUZZ) CFLAGS = $(CPPFLAGS) $(BC_DEFS) $(DC_DEFS) %%CPPFLAGS%% %%CFLAGS%% LDFLAGS = %%LDFLAGS%% HOSTCFLAGS = %%HOSTCFLAGS%% CC = %%CC%% HOSTCC = %%HOSTCC%% BC_LIB_C_ARGS = bc_lib bc_lib_name $(BC_ENABLED_NAME) 1 BC_LIB2_C_ARGS = bc_lib2 bc_lib2_name "$(BC_ENABLED_NAME) && $(BC_ENABLE_EXTRA_MATH_NAME)" 1 OBJS = $(DC_HELP_O) $(BC_HELP_O) $(BC_LIB_O) $(BC_LIB2_O) $(OBJ) all: %%DEFAULT_TARGET%% %%DEFAULT_TARGET%%: %%DEFAULT_TARGET_PREREQS%% %%DEFAULT_TARGET_CMD%% %%SECOND_TARGET%%: %%SECOND_TARGET_PREREQS%% %%SECOND_TARGET_CMD%% $(GEN_DIR): mkdir -p $(GEN_DIR) $(GEN_EXEC): $(GEN_DIR) %%GEN_EXEC_TARGET%% $(BC_LIB_C): $(GEN_EXEC) $(BC_LIB) $(GEN_EMU) $(GEN_EXEC) $(BC_LIB) $(BC_LIB_C) $(BC_EXCLUDE_EXTRA_MATH) $(BC_LIB_C_ARGS) "" "" 1 $(BC_LIB_O): $(BC_LIB_C) $(CC) $(CFLAGS) -o $@ -c $< $(BC_LIB2_C): $(GEN_EXEC) $(BC_LIB2) $(GEN_EMU) $(GEN_EXEC) $(BC_LIB2) $(BC_LIB2_C) $(BC_EXCLUDE_EXTRA_MATH) $(BC_LIB2_C_ARGS) "" "" 1 $(BC_LIB2_O): $(BC_LIB2_C) $(CC) $(CFLAGS) -o $@ -c $< $(BC_HELP_C): $(GEN_EXEC) $(BC_HELP) $(GEN_EMU) $(GEN_EXEC) $(BC_HELP) $(BC_HELP_C) $(BC_EXCLUDE_EXTRA_MATH) bc_help "" $(BC_ENABLED_NAME) 0 $(BC_HELP_O): $(BC_HELP_C) $(CC) $(CFLAGS) -o $@ -c $< $(DC_HELP_C): $(GEN_EXEC) $(DC_HELP) $(GEN_EMU) $(GEN_EXEC) $(DC_HELP) $(DC_HELP_C) $(BC_EXCLUDE_EXTRA_MATH) dc_help "" $(DC_ENABLED_NAME) 0 $(DC_HELP_O): $(DC_HELP_C) $(CC) $(CFLAGS) -o $@ -c $< $(BIN): $(MKDIR) -p $(BIN) src: $(MKDIR) -p src headers: %%HEADERS%% $(MINISTAT): mkdir -p $(SCRIPTS) $(HOSTCC) $(HOSTCFLAGS) -lm -o $(MINISTAT_EXEC) $(ROOTDIR)/scripts/ministat.c $(BITFUNCGEN): mkdir -p $(SCRIPTS) $(HOSTCC) $(HOSTCFLAGS) -lm -o $(BITFUNCGEN_EXEC) $(ROOTDIR)/scripts/bitfuncgen.c help: @printf 'available targets:\n' @printf '\n' @printf ' all (default) builds %%EXECUTABLES%%\n' @printf ' check alias for `make test`\n' @printf ' clean removes all build files\n' @printf ' clean_config removes all build files as well as the generated Makefile\n' @printf ' clean_tests removes all build files, the generated Makefile,\n' @printf ' and generated tests\n' @printf ' install installs binaries to "%s%s"\n' "$(DESTDIR)" "$(BINDIR)" @printf ' and (if enabled) manpages to "%s%s"\n' "$(DESTDIR)" "$(MAN1DIR)" @printf ' karatsuba runs the karatsuba script (requires Python 3)\n' @printf ' karatsuba_test runs the karatsuba script while running tests\n' @printf ' (requires Python 3)\n' @printf ' uninstall uninstalls binaries from "%s%s"\n' "$(DESTDIR)" "$(BINDIR)" @printf ' and (if enabled) manpages from "%s%s"\n' "$(DESTDIR)" "$(MAN1DIR)" @printf ' test runs the test suite\n' @printf ' test_bc runs the bc test suite, if bc has been built\n' @printf ' test_dc runs the dc test suite, if dc has been built\n' @printf ' time_test runs the test suite, displaying times for some things\n' @printf ' time_test_bc runs the bc test suite, displaying times for some things\n' @printf ' time_test_dc runs the dc test suite, displaying times for some things\n' @printf ' timeconst runs the test on the Linux timeconst.bc script,\n' @printf ' if it exists and bc has been built\n' run_all_tests: bc_all_tests timeconst_all_tests dc_all_tests run_all_tests_np: bc_all_tests_np timeconst_all_tests dc_all_tests_np bc_all_tests: %%BC_ALL_TESTS%% bc_all_tests_np: %%BC_ALL_TESTS_NP%% timeconst_all_tests: %%TIMECONST_ALL_TESTS%% dc_all_tests: %%DC_ALL_TESTS%% dc_all_tests_np: %%DC_ALL_TESTS_NP%% history_all_tests: %%HISTORY_TESTS%% check: test test: %%TESTS%% test_bc: test_bc_header test_bc_tests test_bc_scripts test_bc_errors test_bc_stdin test_bc_read test_bc_other @printf '\nAll bc tests passed.\n\n$(TEST_STARS)\n' test_bc_tests:%%BC_TESTS%% test_bc_scripts:%%BC_SCRIPT_TESTS%% test_bc_stdin: @export BC_TEST_OUTPUT_DIR="$(BUILDDIR)/tests"; sh $(TESTSDIR)/stdin.sh bc %%BC_TEST_EXEC%% test_bc_read: @export BC_TEST_OUTPUT_DIR="$(BUILDDIR)/tests"; sh $(TESTSDIR)/read.sh bc %%BC_TEST_EXEC%% test_bc_errors: test_bc_error_lines%%BC_ERROR_TESTS%% test_bc_error_lines: @export BC_TEST_OUTPUT_DIR="$(BUILDDIR)/tests"; sh $(TESTSDIR)/errors.sh bc %%BC_TEST_EXEC%% test_bc_other: @export BC_TEST_OUTPUT_DIR="$(BUILDDIR)/tests"; sh $(TESTSDIR)/other.sh bc $(BC_ENABLE_EXTRA_MATH) %%BC_TEST_EXEC%% test_bc_header: @printf '$(TEST_STARS)\n\nRunning bc tests...\n\n' test_dc: test_dc_header test_dc_tests test_dc_scripts test_dc_errors test_dc_stdin test_dc_read test_dc_other @printf '\nAll dc tests passed.\n\n$(TEST_STARS)\n' test_dc_tests:%%DC_TESTS%% test_dc_scripts:%%DC_SCRIPT_TESTS%% test_dc_stdin: @export BC_TEST_OUTPUT_DIR="$(BUILDDIR)/tests"; sh $(TESTSDIR)/stdin.sh dc %%DC_TEST_EXEC%% test_dc_read: @export BC_TEST_OUTPUT_DIR="$(BUILDDIR)/tests"; sh $(TESTSDIR)/read.sh dc %%DC_TEST_EXEC%% test_dc_errors: test_dc_error_lines%%DC_ERROR_TESTS%% test_dc_error_lines: @export BC_TEST_OUTPUT_DIR="$(BUILDDIR)/tests"; sh $(TESTSDIR)/errors.sh dc %%DC_TEST_EXEC%% test_dc_other: @export BC_TEST_OUTPUT_DIR="$(BUILDDIR)/tests"; sh $(TESTSDIR)/other.sh dc $(BC_ENABLE_EXTRA_MATH) %%DC_TEST_EXEC%% test_dc_header: @printf '$(TEST_STARS)\n\nRunning dc tests...\n\n' timeconst: %%TIMECONST%% test_history: test_history_header test_bc_history test_dc_history @printf '\nAll history tests passed.\n\n$(TEST_STARS)\n' test_bc_history:%%BC_HISTORY_TEST_PREREQS%% test_bc_history_all: test_bc_history0 test_bc_history1 test_bc_history2 test_bc_history3 test_bc_history4 test_bc_history5 test_bc_history6 test_bc_history7 test_bc_history8 test_bc_history9 test_bc_history10 test_bc_history11 test_bc_history12 test_bc_history13 test_bc_history14 test_bc_history15 test_bc_history16 test_bc_history17 test_bc_history18 test_bc_history19 test_bc_history20 test_bc_history21 test_bc_history_skip: @printf 'No bc history tests to run\n' test_bc_history0: @sh $(TESTSDIR)/history.sh bc 0 %%BC_TEST_EXEC%% test_bc_history1: @sh $(TESTSDIR)/history.sh bc 1 %%BC_TEST_EXEC%% test_bc_history2: @sh $(TESTSDIR)/history.sh bc 2 %%BC_TEST_EXEC%% test_bc_history3: @sh $(TESTSDIR)/history.sh bc 3 %%BC_TEST_EXEC%% test_bc_history4: @sh $(TESTSDIR)/history.sh bc 4 %%BC_TEST_EXEC%% test_bc_history5: @sh $(TESTSDIR)/history.sh bc 5 %%BC_TEST_EXEC%% test_bc_history6: @sh $(TESTSDIR)/history.sh bc 6 %%BC_TEST_EXEC%% test_bc_history7: @sh $(TESTSDIR)/history.sh bc 7 %%BC_TEST_EXEC%% test_bc_history8: @sh $(TESTSDIR)/history.sh bc 8 %%BC_TEST_EXEC%% test_bc_history9: @sh $(TESTSDIR)/history.sh bc 9 %%BC_TEST_EXEC%% test_bc_history10: @sh $(TESTSDIR)/history.sh bc 10 %%BC_TEST_EXEC%% test_bc_history11: @sh $(TESTSDIR)/history.sh bc 11 %%BC_TEST_EXEC%% test_bc_history12: @sh $(TESTSDIR)/history.sh bc 12 %%BC_TEST_EXEC%% test_bc_history13: @sh $(TESTSDIR)/history.sh bc 13 %%BC_TEST_EXEC%% test_bc_history14: @sh $(TESTSDIR)/history.sh bc 14 %%BC_TEST_EXEC%% test_bc_history15: @sh $(TESTSDIR)/history.sh bc 15 %%BC_TEST_EXEC%% test_bc_history16: @sh $(TESTSDIR)/history.sh bc 16 %%BC_TEST_EXEC%% test_bc_history17: @sh $(TESTSDIR)/history.sh bc 17 %%BC_TEST_EXEC%% test_bc_history18: @sh $(TESTSDIR)/history.sh bc 18 %%BC_TEST_EXEC%% test_bc_history19: @sh $(TESTSDIR)/history.sh bc 19 %%BC_TEST_EXEC%% test_bc_history20: @sh $(TESTSDIR)/history.sh bc 20 %%BC_TEST_EXEC%% test_bc_history21: @sh $(TESTSDIR)/history.sh bc 21 %%BC_TEST_EXEC%% test_dc_history:%%DC_HISTORY_TEST_PREREQS%% test_dc_history_all: test_dc_history0 test_dc_history1 test_dc_history2 test_dc_history3 test_dc_history4 test_dc_history5 test_dc_history6 test_dc_history7 test_dc_history8 test_dc_history9 test_dc_history10 test_dc_history_skip: @printf 'No dc history tests to run\n' test_dc_history0: @sh $(TESTSDIR)/history.sh dc 0 %%DC_TEST_EXEC%% test_dc_history1: @sh $(TESTSDIR)/history.sh dc 1 %%DC_TEST_EXEC%% test_dc_history2: @sh $(TESTSDIR)/history.sh dc 2 %%DC_TEST_EXEC%% test_dc_history3: @sh $(TESTSDIR)/history.sh dc 3 %%DC_TEST_EXEC%% test_dc_history4: @sh $(TESTSDIR)/history.sh dc 4 %%DC_TEST_EXEC%% test_dc_history5: @sh $(TESTSDIR)/history.sh dc 5 %%DC_TEST_EXEC%% test_dc_history6: @sh $(TESTSDIR)/history.sh dc 6 %%DC_TEST_EXEC%% test_dc_history7: @sh $(TESTSDIR)/history.sh dc 7 %%DC_TEST_EXEC%% test_dc_history8: @sh $(TESTSDIR)/history.sh dc 8 %%DC_TEST_EXEC%% test_dc_history9: @sh $(TESTSDIR)/history.sh dc 9 %%DC_TEST_EXEC%% test_dc_history10: @sh $(TESTSDIR)/history.sh dc 10 %%DC_TEST_EXEC%% test_history_header: @printf '$(TEST_STARS)\n\nRunning history tests...\n\n' library_test: $(LIBBC) $(CC) $(CFLAGS) -lpthread $(BCL_TEST_C) $(LIBBC) -o $(BCL_TEST) test_library: library_test %%BCL_TEST_EXEC%% karatsuba: %%KARATSUBA%% karatsuba_test: %%KARATSUBA_TEST%% coverage_output: %%COVERAGE_OUTPUT%% coverage:%%COVERAGE_PREREQS%% manpages: $(MANPAGE) bc $(MANPAGE) dc $(MANPAGE) bcl clean_gen: @$(RM) -f $(GEN_EXEC) clean:%%CLEAN_PREREQS%% @printf 'Cleaning files...\n' @$(RM) -f src/*.tmp gen/*.tmp @$(RM) -f $(OBJ) @$(RM) -f $(BC_EXEC) @$(RM) -f $(DC_EXEC) @$(RM) -fr $(BIN) @$(RM) -f $(BC_LIB_C) $(BC_LIB_O) @$(RM) -f $(BC_LIB2_C) $(BC_LIB2_O) @$(RM) -f $(BC_HELP_C) $(BC_HELP_O) @$(RM) -f $(DC_HELP_C) $(DC_HELP_O) @$(RM) -fr vs/bin/ vs/lib/ clean_benchmarks: @printf 'Cleaning benchmarks...\n' @$(RM) -f $(MINISTAT_EXEC) @$(RM) -f $(ROOTDIR)/benchmarks/bc/*.txt @$(RM) -f $(ROOTDIR)/benchmarks/dc/*.txt clean_config: clean clean_benchmarks @printf 'Cleaning config...\n' @$(RM) -f Makefile @$(RM) -f $(BC_MD) $(BC_MANPAGE) @$(RM) -f $(DC_MD) $(DC_MANPAGE) @$(RM) -f compile_commands.json @$(RM) -f $(BCL_PC) clean_coverage: @printf 'Cleaning coverage files...\n' @$(RM) -f *.gcov @$(RM) -f *.html *.css @$(RM) -f *.gcda *.gcno @$(RM) -f *.profraw @$(RM) -f $(GCDA) $(GCNO) @$(RM) -f $(BC_GCDA) $(BC_GCNO) @$(RM) -f $(DC_GCDA) $(DC_GCNO) @$(RM) -f $(HISTORY_GCDA) $(HISTORY_GCNO) @$(RM) -f $(RAND_GCDA) $(RAND_GCNO) @$(RM) -f $(BC_LIB_GCDA) $(BC_LIB_GCNO) @$(RM) -f $(BC_LIB2_GCDA) $(BC_LIB2_GCNO) @$(RM) -f $(BC_HELP_GCDA) $(BC_HELP_GCNO) @$(RM) -f $(DC_HELP_GCDA) $(DC_HELP_GCNO) clean_tests: clean clean_config clean_coverage @printf 'Cleaning test files...\n' @$(RM) -fr $(BC_TEST_OUTPUTS) $(DC_TEST_OUTPUTS) @$(RM) -fr $(BC_FUZZ_OUTPUTS) $(DC_FUZZ_OUTPUTS) @$(RM) -f $(TESTSDIR)/bc/parse.txt $(TESTSDIR)/bc/parse_results.txt @$(RM) -f $(TESTSDIR)/bc/print.txt $(TESTSDIR)/bc/print_results.txt @$(RM) -f $(TESTSDIR)/bc/bessel.txt $(TESTSDIR)/bc/bessel_results.txt @$(RM) -f $(TESTSDIR)/bc/strings2.txt $(TESTSDIR)/bc/strings2_results.txt @$(RM) -f $(TESTSDIR)/bc/scripts/bessel.txt @$(RM) -f $(TESTSDIR)/bc/scripts/parse.txt @$(RM) -f $(TESTSDIR)/bc/scripts/print.txt @$(RM) -f $(TESTSDIR)/bc/scripts/add.txt @$(RM) -f $(TESTSDIR)/bc/scripts/divide.txt @$(RM) -f $(TESTSDIR)/bc/scripts/multiply.txt @$(RM) -f $(TESTSDIR)/bc/scripts/subtract.txt @$(RM) -f $(TESTSDIR)/bc/scripts/strings2.txt @$(RM) -f $(TESTSDIR)/dc/scripts/prime.txt @$(RM) -f .log_*.txt @$(RM) -f .math.txt .results.txt .ops.txt @$(RM) -f .test.txt @$(RM) -f tags .gdbbreakpoints .gdb_history .gdbsetup @$(RM) -f cscope.* @$(RM) -f bc.old @$(RM) -f $(BITFUNCGEN_EXEC) install_locales: %%INSTALL_LOCALES%% install_bc_manpage: $(SAFE_INSTALL) $(MANPAGE_INSTALL_ARGS) $(BC_MANPAGE) $(DESTDIR)$(MAN1DIR)/$(BC_MANPAGE_NAME) install_dc_manpage: $(SAFE_INSTALL) $(MANPAGE_INSTALL_ARGS) $(DC_MANPAGE) $(DESTDIR)$(MAN1DIR)/$(DC_MANPAGE_NAME) install_bcl_manpage: $(SAFE_INSTALL) $(MANPAGE_INSTALL_ARGS) $(BCL_MANPAGE) $(DESTDIR)$(MAN3DIR)/$(BCL_MANPAGE_NAME) install_bcl_header: $(SAFE_INSTALL) $(MANPAGE_INSTALL_ARGS) $(BCL_HEADER) $(DESTDIR)$(INCLUDEDIR)/$(BCL_HEADER_NAME) install_execs: $(INSTALL) $(DESTDIR)$(BINDIR) "$(EXEC_SUFFIX)" "$(BUILDDIR)/bin" install_library: install_bcl_header $(SAFE_INSTALL) $(BINARY_INSTALL_ARGS) $(LIBBC) $(DESTDIR)$(LIBDIR)/$(LIB_NAME) %%PKG_CONFIG_INSTALL%% install:%%INSTALL_LOCALES_PREREQS%%%%INSTALL_MAN_PREREQS%%%%INSTALL_PREREQS%% uninstall_locales: $(LOCALE_UNINSTALL) $(NLSPATH) $(MAIN_EXEC) $(DESTDIR) uninstall_bc_manpage: $(RM) -f $(DESTDIR)$(MAN1DIR)/$(BC_MANPAGE_NAME) uninstall_bc: $(RM) -f $(DESTDIR)$(BINDIR)/$(EXEC_PREFIX)$(BC)$(EXEC_SUFFIX) uninstall_dc_manpage: $(RM) -f $(DESTDIR)$(MAN1DIR)/$(DC_MANPAGE_NAME) uninstall_dc: $(RM) -f $(DESTDIR)$(BINDIR)/$(EXEC_PREFIX)$(DC)$(EXEC_SUFFIX) uninstall_library: uninstall_bcl_header $(RM) -f $(DESTDIR)$(LIBDIR)/$(LIB_NAME) %%PKG_CONFIG_UNINSTALL%% uninstall_bcl_header: $(RM) -f $(DESTDIR)$(INCLUDEDIR)/$(BCL_HEADER_NAME) uninstall_bcl_manpage: $(RM) -f $(DESTDIR)$(MAN3DIR)/$(BCL_MANPAGE_NAME) uninstall:%%UNINSTALL_LOCALES_PREREQS%%%%UNINSTALL_MAN_PREREQS%%%%UNINSTALL_PREREQS%% diff --git a/NEWS.md b/NEWS.md index 95de7e5182c4..1775fa0b6533 100644 --- a/NEWS.md +++ b/NEWS.md @@ -1,1562 +1,1576 @@ # News +## 7.0.0 + +This is a production release to fix three bugs. + +The first bug is that `bc`/`dc` will exit on macOS when the terminal is resized. + +The second bug is that an array, which should only be a function parameter, was +accepted as part of larger expressions. + +The third bug is that value stack for `dc` was cleared on any error. However, +this is not how other `dc` behave. To bring `dc` more in line with other +implementations, this behavior was changed. This change is why this version is a +new major version. + ## 6.7.6 This is a production release to fix one bug. The bug was that `bc` attempted to jump out when flushing `stdout` on exit, but there is no jump buf at that point. ## 6.7.5 This is a production release to fix one small bug. The bug is that sometimes numbers are printed to incorrect line lengths. The number is always correct; the line is just longer than the limit. Users who do not care do not need to update. ## 6.7.4 This is a production release to fix problems in the `bc` manual. Users only need to update if desired. ## 6.7.3 This is a production release to fix the library build on Mac OSX. Users on other platforms do *not* need to update. ## 6.7.2 This is a production release to remove some debugging code that I accidentally committed. ## 6.7.1 This is a production release with a bug fix for `SIGINT` only being handled once. ## 6.7.0 This is a production release with three new functions in the [extended math library][16]: `min()`, `max()`, and `i2rand()`. ## 6.6.1 This is a production release with an improved `p()` function in the [extended math library][16]. Users who don't care do not need to upgrade. ## 6.6.0 This is a production release with two bug fixes and one change. The first bug fix is to fix the build on Mac OSX. The second bug was to remove printing a leading zero in scientific or engineering output modes. The change was that the implementation of `irand()` was improved to call the PRNG less. ## 6.5.0 This is a production release that fixes an infinite loop bug in `root()` and `cbrt()`, fixes a bug with `BC_LINE_LENGTH=0`, and adds the `fib()` function to the extended math library to calculate Fibonacci numbers. ## 6.4.0 This is a production release that fixes a `read()`/`?` bug and adds features to `bcl`. The bug was that multiple read calls could repeat old data. The new features in `bcl` are functions to preserve `BclNumber` arguments and not free them. ***WARNING for `bcl` Users***: The `bcl_rand_seedWithNum()` function used to not consume its arguments. Now it does. This change could have made this version `7.0.0`, but I'm 99.9% confident that there are no `bcl` users, or if there are, they probably don't use the PRNG. So I took a risk and didn't update the major version. `bcl` now includes more capacity to check for invalid numbers when built to run under Valgrind. ## 6.3.1 This is a production release that fixes a `bc` dependency loop for minimal environments and Linux from Scratch. ## 6.3.0 This is a production release with a couple of fixes for manuals and a new feature for `dc`: there is now a command to query whether extended registers are enabled or not. Users who don't care do not need to upgrade. ## 6.2.6 This is a production release that fixes an install bug that affected locale installation of all locales when using `mksh`. Users do ***NOT*** need to upgrade if they don't use `mksh` and/or don't need to install all locales. ## 6.2.5 This is a production release that fixes a test bug that affected Android and `mksh`. Users do ***NOT*** need to upgrade unless they use `mksh` or another affected shell and need to run the test suite. ## 6.2.4 This is a production release that fixes a test failure that happens when `tests/bc/scripts/timeconst.bc` doesn't exist. This should only affect packagers. This bug happened because I forgot something I added in the previous release: better error checking in tests to help packagers. Unfortunately, I was too zealous with the error checking. ## 6.2.3 This is a production release that moves `bc` to . That's all it does: update links. Users do ***NOT*** need to upgrade; there are redirects that will stay in place indefinitely. This release is only for new users. ## 6.2.2 This is a production release that fixes a bug. The bug was that if an array element was used as a parameter, and then a later parameter had the same name as the array whose element was used, `bc` would grab the element from the new array parameter, not the actual element from before the function call. ## 6.2.1 This is a production release with one bug fix for a memory bug in history. ## 6.2.0 This is a production release with a new feature and a few bug fixes. The bug fixes include: * A crash when `bc` and `dc` are built using editline, but history is not activated. * A missing local in the `uint*()` family of functions in the extended math library. * A failure to clear the tail call list in `dc` on error. * A crash when attempting to swap characters in command-line history when no characters exist. * `SIGWINCH` was activated even when history was not. The new feature is that stack traces are now given for runtime errors. In debug mode, the C source file and line of errors are given as well. ## 6.1.1 This is a production release that fixes a build issue with predefined builds and generated tests. ## 6.1.0 This is a production release that fixes a discrepancy from the `bc` standard, a couple of memory bugs, and adds new features. The discrepancy from the `bc` standard was with regards to the behavior of the `quit` command. This `bc` used to quit whenever it encountered `quit` during parsing, even if it was parsing a full file. Now, `bc` only quits when encountering `quit` *after* it has executed all executable statements up to that point. This behavior is slightly different from GNU `bc`, but users will only notice the difference if they put `quit` on the same line as other statements. The first memory bug could be reproduced by assigning a string to a non-local variable in a function, then redefining the function with use of the same non-local variable, which would still refer to a string in the previous version of the function. The second memory bug was caused by passing an array argument to the `asciify()` built-in function. In certain cases, that was wrongly allowed, and the interpreter just assumed everything was correct and accessed memory. Now that arrays are allowed as arguments (see below), this is not an issue. The first feature was the addition of the `is_number()` built-in function (`u` in `dc`) that returns 1 if the runtime argument is a number and 0 otherwise. The second feature was the addition of the `is_string()` built-in function (`t` in `dc`) that returns 1 if the runtime argument is a string and 0 otherwise. These features were added because I realized that type-checking is necessary now that strings can be assigned to variables in `bc` and because they've always been assignable to variables in `dc`. The last added feature is the ability of the `asciify()` built-in function in `bc` to convert a full array of numbers into a string. This means that character-by-character printing will not be necessary, and more strings than just single-character ones will be able to be created. ## 6.0.4 This is a production release that most users will not need to upgrade to. This fixes a build bug for `bcl` only on OpenBSD. Users that do not need `bcl` or have not run into build errors with `bcl` do ***NOT*** need to upgrade. ## 6.0.3 This is a production release that fixes a build bug for cross-compilation. Users that do not need cross-compilation do ***NOT*** need to upgrade. ## 6.0.2 This is a production release that fixes two bugs: * The `-l` option overrode the `-S` option. * A double-free and crash when sending a `SIGINT` while executing expressions given on the command-line. ## 6.0.1 This is a production release that fixes memory bugs and memory leaks in `bcl`. Users that do not use `bcl` (use only `bc` and/or `dc`) do ***NOT*** need to upgrade. These happened because I was unaware that the `bcl` test was not hooked into the Valgrind test infrastructure. Then, when I ran the release script, which tests everything under Valgrind (or so I thought), it caught nothing, and I thought it was safe. But it was not. Nevertheless, I have now run it under Valgrind and fixed all of the memory bugs (caused by not using `memset()` where I should have but previously didn't have to) and memory leaks. ## 6.0.0 This is a production release that fixes an oversight in the `bc` parser (that sometimes caused the wrong error message) and adds a feature for compatibility with the BSD `bc` and `dc`: turning off digit clamping when parsing numbers. The default for clamping can be set during the build (see the [build manual][13]), it can be set with the `BC_DIGIT_CLAMP` and `DC_DIGIT_CLAMP` environment variables, and it can be set with the `-c` and `-C` command-line options. Turning off clamping was also added to the `bcl` library. In addition, signal handling was removed from the `bcl` library in order to add the capability for multi-threading. This required a major version bump. I apologize to all library users (I don't know of any), but signals and threads do not play well together. To help with building, a convenience option (`-p`) to `configure.sh` was added to build a `bc` and `dc` that is by default compatible with either the BSD `bc` and `dc` or the GNU `bc` and `dc`. ## 5.3.3 This is a production release that fixes a build problem in the FreeBSD base system. All other users do **NOT** need to upgrade. ## 5.3.2 This is a production release that fixes prompt bugs with editline and readline where the `BC_PROMPT` environment variable was not being respected. This also fixes editline and readline output on `EOF`. ## 5.3.1 This is a production release that fixes a build problem in the FreeBSD base system, as well as a problem in the `en_US` locale. If you don't have problems with either, you do not need to upgrade. ## 5.3.0 This is a production release that adds features and has a few bug fixes. First, support for editline and readline history has been added. To use editline, pass `-e` to `configure.sh`, and to use readline, pass `-r`. Second, history support for Windows has been fixed and re-enabled. Third, command-line options to set `scale`, `ibase`, `obase`, and `seed` were added. This was requested long ago, and I originally disagreed with the idea. Fourth, the manuals had typos and were missing information. That has been fixed. Fifth, the manuals received different formatting to be more readable as manpages. ## 5.2.5 This is a production release that fixes this `bc`'s behavior on `^D` to match GNU `bc`. ## 5.2.4 This is a production release that fixes two bugs in history: * Without prompt, the cursor could not be placed on the first character in a line. * Home and End key handling in `tmux` was fixed. Any users that do not care about these improvements do not need to upgrade. ## 5.2.3 This is a production release that fixes one bug, a parse error when passing a file to `bc` using `-f` if that file had a multiline comment or string in it. ## 5.2.2 This is a production release that fixes one bug, a segmentation fault if `argv[0]` equals `NULL`. This is not a critical bug; there will be no vulnerability as far as I can tell. There is no need to update if you do not wish to. ## 5.2.1 This is a production release that fixes two parse bugs when in POSIX standard mode. One of these bugs was due to a quirk of the POSIX grammar, and the other was because `bc` was too strict. ## 5.2.0 This is a production release that adds a new feature, fixes some bugs, and adds out-of-source builds and a `pkg-config` file for `bcl`. The new feature is the ability to turn off exiting on expressions. It is also possible to set the default using `configure.sh`. This behavior used to exist with the `BC_EXPR_EXIT` environment variable, which is now used again. Bugs fixed include: * Some possible race conditions with error handling. * Install and uninstall targets for `bcl` did not work. ## 5.1.1 This is a production release that completes a bug fix from `5.1.0`. The bug exists in all versions of `bc`. The bug was that `if` statements without `else` statements would not be handled correctly at the end of files or right before a function definition. ## 5.1.0 This is a production release with some fixes and new features. * Fixed a bug where an `if` statement without an `else` before defining a function caused an error. * Fixed a bug with the `bc` banner and `-q`. * Fixed a bug on Windows where files were not read correctly. * Added a command-line flag (`-z`) to make `bc` and `dc` print leading zeroes on numbers `-1 < x < 1`. * Added four functions to `lib2.bc` (`plz()`, `plznl()`, `pnlz()`, and `pnlznl()`) to allow printing numbers with or without leading zeros, despite the use of `-z` or not. * Added builtin functions to query global state like line length, global stacks, and leading zeroes. * Added a command-line flag (`-L`) to disable wrapping when printing numbers. * Improved builds on Windows. ## 5.0.2 This is a production release with one fix for a flaky test. If you have not experienced problems with the test suite, you do ***NOT*** need to upgrade. The test was one that tested whether `bc` fails gracefully when it can't allocate memory. Unfortunately, there are cases when Linux and FreeBSD lie and pretend to allocate the memory. The reason they do this is because a lot of programs don't use all of the memory they allocate, so those OS's usually get away with it. However, this `bc` uses all of the memory it allocates (at least at page granularity), so when it tries to use the memory, FreeBSD and Linux kill it. This only happens sometimes, however. Other times (on my machine), they do, in fact, refuse the request. So I changed the test to not test for that because I think the graceful failure code won't really change much. ## 5.0.1 This is a production release with two fixes: * Fix for the build on Mac OSX. * Fix for the build on Android. Users that do not use those platforms do ***NOT*** need to update. ## 5.0.0 This is a major production release with several changes: * Added support for OpenBSD's `pledge()` and `unveil()`. * Fixed print bug where a backslash newline combo was printed even if only one digit was left, something I blindly copied from GNU `bc`, like a fool. * Fixed bugs in the manuals. * Fixed a possible multiplication overflow in power. * Temporary numbers are garbage collected if allocation fails, and the allocation is retried. This is to make `bc` and `dc` more resilient to running out of memory. * Limited the number of temporary numbers and made the space for them static so that allocating more space for them cannot fail. * Allowed integers with non-zero `scale` to be used with power, places, and shift operators. * Added greatest common divisor and least common multiple to `lib2.bc`. * Added `SIGQUIT` handling to history. * Added a command to `dc` (`y`) to get the length of register stacks. * Fixed multi-digit bugs in `lib2.bc`. * Removed the no prompt build option. * Created settings that builders can set defaults for and users can set their preferences for. This includes the `bc` banner, resetting on `SIGINT`, TTY mode, and prompt. * Added history support to Windows. * Fixed bugs with the handling of register names in `dc`. * Fixed bugs with multi-line comments and strings in both calculators. * Added a new error type and message for `dc` when register stacks don't have enough items. * Optimized string allocation. * Made `bc` and `dc` UTF-8 capable. * Fixed a bug with `void` functions. * Fixed a misspelled symbol in `bcl`. This is technically a breaking change, which requires this to be `5.0.0`. * Added the ability for users to get the copyright banner back. * Added the ability for users to have `bc` and `dc` quit on `SIGINT`. * Added the ability for users to disable prompt and TTY mode by environment variables. * Added the ability for users to redefine keywords. This is another reason this is `5.0.0`. * Added `dc`'s modular exponentiation and divmod to `bc`. * Added the ability to assign strings to variables and array elements and pass them to functions in `bc`. * Added `dc`'s asciify command and stream printing to `bc`. * Added a command to `dc` (`Y`) to get the length of an array. * Added a command to `dc` (`,`) to get the depth of the execution stack. * Added bitwise and, or, xor, left shift, right shift, reverse, left rotate, right rotate, and mod functions to `lib2.bc`. * Added the functions `s2u(x)` and `s2un(x,n)`, to `lib2.bc`. ## 4.0.2 This is a production release that fixes two bugs: 1. If no files are used and the first statement on `stdin` is invalid, `scale` would not be set to `20` even if `-l` was used. 2. When using history, `bc` failed to respond properly to `SIGSTOP` and `SIGTSTP`. ## 4.0.1 This is a production release that only adds one thing: flushing output when it is printed with a print statement. ## 4.0.0 This is a production release with many fixes, a new command-line option, and a big surprise: * A bug was fixed in `dc`'s `P` command where the item on the stack was *not* popped. * Various bugs in the manuals have been fixed. * A known bug was fixed where history did not interact well with prompts printed by user code without newlines. * A new command-line option, `-R` and `--no-read-prompt` was added to disable just the prompt when using `read()` (`bc`) or `?` (`dc`). * And finally, **official support for Windows was added**. The last item is why this is a major version bump. Currently, only one set of build options (extra math and prompt enabled, history and NLS/locale support disabled, both calculators enabled) is supported on Windows. However, both debug and release builds are supported. In addition, Windows builds are supported for the the library (`bcl`). For more details about how to build on Windows, see the [README][5] or the [build manual][13]. ## 3.3.4 This is a production release that fixes a small bug. The bug was that output was not flushed before a `read()` call, so prompts without a newline on the end were not flushed before the `read()` call. This is such a tiny bug that users only need to upgrade if they are affected. ## 3.3.3 This is a production release with one tweak and fixes for manuals. The tweak is that `length(0)` returns `1` instead of `0`. In `3.3.1`, I changed it so `length(0.x)`, where `x` could be any number of digits, returned the `scale`, but `length(0)` still returned `0` because I believe that `0` has `0` significant digits. After request of FreeBSD and considering the arguments of a mathematician, compatibility with other `bc`'s, and the expectations of users, I decided to make the change. The fixes for manuals fixed a bug where `--` was rendered as `-`. ## 3.3.2 This is a production release that fixes a divide-by-zero bug in `root()` in the [extended math library][16]. All previous versions with `root()` have the bug. ## 3.3.1 This is a production release that fixes a bug. The bug was in the reporting of number length when the value was 0. ## 3.3.0 This is a production release that changes one behavior and fixes documentation bugs. The changed behavior is the treatment of `-e` and `-f` when given through `BC_ENV_ARGS` or `DC_ENV_ARGS`. Now `bc` and `dc` do not exit when those options (or their equivalents) are given through those environment variables. However, `bc` and `dc` still exit when they or their equivalents are given on the command-line. ## 3.2.7 This is a production release that removes a small non-portable shell operation in `configure.sh`. This problem was only noticed on OpenBSD, not FreeBSD or Linux. Non-OpenBSD users do ***NOT*** need to upgrade, although NetBSD users may also need to upgrade. ## 3.2.6 This is a production release that fixes the build on FreeBSD. There was a syntax error in `configure.sh` that the Linux shell did not catch, and FreeBSD depends on the existence of `tests/all.sh`. All users that already upgraded to `3.2.5` should update to this release, with my apologies for the poor release of `3.2.5`. Other users should skip `3.2.5` in favor of this version. ## 3.2.5 This is a production release that fixes several bugs and adds a couple small things. The two most important bugs were bugs that causes `dc` to access memory out-of-bounds (crash in debug builds). This was found by upgrading to `afl++` from `afl`. Both were caused by a failure to distinguish between the same two cases. Another bug was the failure to put all of the licenses in the `LICENSE.md` file. Third, some warnings by `scan-build` were found and eliminated. This needed one big change: `bc` and `dc` now bail out as fast as possible on fatal errors instead of unwinding the stack. Fourth, the pseudo-random number now attempts to seed itself with `/dev/random` if `/dev/urandom` fails. Finally, this release has a few quality-of-life changes to the build system. The usage should not change at all; the only thing that changed was making sure the `Makefile.in` was written to rebuild properly when headers changed and to not rebuild when not necessary. ## 3.2.4 This is a production release that fixes a warning on `gcc` 6 or older, which does not have an attribute that is used. Users do ***NOT*** need to upgrade if they don't use `gcc` 6 or older. ## 3.2.3 This is a production release that fixes a bug in `gen/strgen.sh`. I recently changed `gen/strgen.c`, but I did not change `gen/strgen.sh`. Users that do not use `gen/strgen.sh` do not need to upgrade. ## 3.2.2 This is a production release that fixes a portability bug in `configure.sh`. The bug was using the GNU `find` extension `-wholename`. ## 3.2.1 This is a production release that has one fix for `bcl(3)`. It is technically not a bug fix since the behavior is undefined, but the `BclNumber`s that `bcl_divmod()` returns will be set to `BCL_ERROR_INVALID_NUM` if there is an error. Previously, they were not set. ## 3.2.0 This is a production release that has one bug fix and a major addition. The bug fix was a missing `auto` variable in the bessel `j()` function in the math library. The major addition is a way to build a version of `bc`'s math code as a library. This is done with the `-a` option to `configure.sh`. The API for the library can be read in `./manuals/bcl.3.md` or `man bcl` once the library is installed with `make install`. This library was requested by developers before I even finished version 1.0, but I could not figure out how to do it until now. If the library has API breaking changes, the major version of `bc` will be incremented. ## 3.1.6 This is a production release that fixes a new warning from Clang 12 for FreeBSD and also removes some possible undefined behavior found by UBSan that compilers did not seem to take advantage of. Users do ***NOT*** need to upgrade, if they do not want to. ## 3.1.5 This is a production release that fixes the Chinese locales (which caused `bc` to crash) and a crash caused by `bc` executing code when it should not have been able to. ***ALL USERS SHOULD UPGRADE.*** ## 3.1.4 This is a production release that fixes one bug, changes two behaviors, and removes one environment variable. The bug is like the one in the last release except it applies if files are being executed. I also made the fix more general. The behavior that was changed is that `bc` now exits when given `-e`, `-f`, `--expression` or `--file`. However, if the last one of those is `-f-` (using `stdin` as the file), `bc` does not exit. If `-f-` exists and is not the last of the `-e` and `-f` options (and equivalents), `bc` gives a fatal error and exits. Next, I removed the `BC_EXPR_EXIT` and `DC_EXPR_EXIT` environment variables since their use is not needed with the behavior change. Finally, I made it so `bc` does not print the header, though the `-q` and `--quiet` options were kept for compatibility with GNU `bc`. ## 3.1.3 This is a production release that fixes one minor bug: if `bc` was invoked like the following, it would error: ``` echo "if (1 < 3) 1" | bc ``` Unless users run into this bug, they do not need to upgrade, but it is suggested that they do. ## 3.1.2 This is a production release that adds a way to install *all* locales. Users do ***NOT*** need to upgrade. For package maintainers wishing to make use of the change, just pass `-l` to `configure.sh`. ## 3.1.1 This is a production release that adds two Spanish locales. Users do ***NOT*** need to upgrade, unless they want those locales. ## 3.1.0 This is a production release that adjusts one behavior, fixes eight bugs, and improves manpages for FreeBSD. Because this release fixes bugs, **users and package maintainers should update to this version as soon as possible**. The behavior that was adjusted was how code from the `-e` and `-f` arguments (and equivalents) were executed. They used to be executed as one big chunk, but in this release, they are now executed line-by-line. The first bug fix in how output to `stdout` was handled in `SIGINT`. If a `SIGINT` came in, the `stdout` buffer was not correctly flushed. In fact, a clean-up function was not getting called. This release fixes that bug. The second bug is in how `dc` handled input from `stdin`. This affected `bc` as well since it was a mishandling of the `stdin` buffer. The third fixed bug was that `bc` and `dc` could `abort()` (in debug mode) when receiving a `SIGTERM`. This one was a race condition with pushing and popping items onto and out of vectors. The fourth bug fixed was that `bc` could leave extra items on the stack and thus, not properly clean up some memory. (The memory would still get `free()`'ed, but it would not be `free()`'ed when it could have been.) The next two bugs were bugs in `bc`'s parser that caused crashes when executing the resulting code. The last two bugs were crashes in `dc` that resulted from mishandling of strings. The manpage improvement was done by switching from [ronn][20] to [Pandoc][21] to generate manpages. Pandoc generates much cleaner manpages and doesn't leave blank lines where they shouldn't be. ## 3.0.3 This is a production release that adds one new feature: specific manpages. Before this release, `bc` and `dc` only used one manpage each that referred to various build options. This release changes it so there is one manpage set per relevant build type. Each manual only has information about its particular build, and `configure.sh` selects the correct set for install. ## 3.0.2 This is a production release that adds `utf8` locale symlinks and removes an unused `auto` variable from the `ceil()` function in the [extended math library][16]. Users do ***NOT*** need to update unless they want the locales. ## 3.0.1 This is a production release with two small changes. Users do ***NOT*** need to upgrade to this release; however, if they haven't upgraded to `3.0.0` yet, it may be worthwhile to upgrade to this release. The first change is fixing a compiler warning on FreeBSD with strict warnings on. The second change is to make the new implementation of `ceil()` in `lib2.bc` much more efficient. ## 3.0.0 *Notes for package maintainers:* *First, the `2.7.0` release series saw a change in the option parsing. This made me change one error message and add a few others. The error message that was changed removed one format specifier. This means that `printf()` will seqfault on old locale files. Unfortunately, `bc` cannot use any locale files except the global ones that are already installed, so it will use the previous ones while running tests during install. **If `bc` segfaults while running arg tests when updating, it is because the global locale files have not been replaced. Make sure to either prevent the test suite from running on update or remove the old locale files before updating.** (Removing the locale files can be done with `make uninstall` or by running the [`locale_uninstall.sh`][22] script.) Once this is done, `bc` should install without problems.* *Second, **the option to build without signal support has been removed**. See below for the reasons why.* This is a production release with some small bug fixes, a few improvements, three major bug fixes, and a complete redesign of `bc`'s error and signal handling. **Users and package maintainers should update to this version as soon as possible.** The first major bug fix was in how `bc` executed files. Previously, a whole file was parsed before it was executed, but if a function is defined *after* code, especially if the function definition was actually a redefinition, and the code before the definition referred to the previous function, this `bc` would replace the function before executing any code. The fix was to make sure that all code that existed before a function definition was executed. The second major bug fix was in `bc`'s `lib2.bc`. The `ceil()` function had a bug where a `0` in the decimal place after the truncation position, caused it to output the wrong numbers if there was any non-zero digit after. The third major bug is that when passing parameters to functions, if an expression included an array (not an array element) as a parameter, it was accepted, when it should have been rejected. It is now correctly rejected. Beyond that, this `bc` got several improvements that both sped it up, improved the handling of signals, and improved the error handling. First, the requirements for `bc` were pushed back to POSIX 2008. `bc` uses one function, `strdup()`, which is not in POSIX 2001, and it is in the X/Open System Interfaces group 2001. It is, however, in POSIX 2008, and since POSIX 2008 is old enough to be supported anywhere that I care, that should be the requirement. Second, the `BcVm` global variable was put into `bss`. This actually slightly reduces the size of the executable from a massive code shrink, and it will stop `bc` from allocating a large set of memory when `bc` starts. Third, the default Karatsuba length was updated from 64 to 32 after making the optimization changes below, since 32 is going to be better than 64 after the changes. Fourth, Spanish translations were added. Fifth, the interpreter received a speedup to make performance on non-math-heavy scripts more competitive with GNU `bc`. While improvements did, in fact, get it much closer (see the [benchmarks][19]), it isn't quite there. There were several things done to speed up the interpreter: First, several small inefficiencies were removed. These inefficiencies included calling the function `bc_vec_pop(v)` twice instead of calling `bc_vec_npop(v, 2)`. They also included an extra function call for checking the size of the stack and checking the size of the stack more than once on several operations. Second, since the current `bc` function is the one that stores constants and strings, the program caches pointers to the current function's vectors of constants and strings to prevent needing to grab the current function in order to grab a constant or a string. Third, `bc` tries to reuse `BcNum`'s (the internal representation of arbitary-precision numbers). If a `BcNum` has the default capacity of `BC_NUM_DEF_SIZE` (32 on 64-bit and 16 on 32-bit) when it is freed, it is added to a list of available `BcNum`'s. And then, when a `BcNum` is allocated with a capacity of `BC_NUM_DEF_SIZE` and any `BcNum`'s exist on the list of reusable ones, one of those ones is grabbed instead. In order to support these changes, the `BC_NUM_DEF_SIZE` was changed. It used to be 16 bytes on all systems, but it was changed to more closely align with the minimum allocation size on Linux, which is either 32 bytes (64-bit musl), 24 bytes (64-bit glibc), 16 bytes (32-bit musl), or 12 bytes (32-bit glibc). Since these are the minimum allocation sizes, these are the sizes that would be allocated anyway, making it worth it to just use the whole space, so the value of `BC_NUM_DEF_SIZE` on 64-bit systems was changed to 32 bytes. On top of that, at least on 64-bit, `BC_NUM_DEF_SIZE` supports numbers with either 72 integer digits or 45 integer digits and 27 fractional digits. This should be more than enough for most cases since `bc`'s default `scale` values are 0 or 20, meaning that, by default, it has at most 20 fractional digits. And 45 integer digits are *a lot*; it's enough to calculate the amount of mass in the Milky Way galaxy in kilograms. Also, 72 digits is enough to calculate the diameter of the universe in Planck lengths. (For 32-bit, these numbers are either 32 integer digits or 12 integer digits and 20 fractional digits. These are also quite big, and going much bigger on a 32-bit system seems a little pointless since 12 digits is just under a trillion and 20 fractional digits is still enough for about any use since `10^-20` light years is just under a millimeter.) All of this together means that for ordinary uses, and even uses in scientific work, the default number size will be all that is needed, which means that nearly all, if not all, numbers will be reused, relieving pressure on the system allocator. I did several experiments to find the changes that had the most impact, especially with regard to reusing `BcNum`'s. One was putting `BcNum`'s into buckets according to their capacity in powers of 2 up to 512. That performed worse than `bc` did in `2.7.2`. Another was putting any `BcNum` on the reuse list that had a capacity of `BC_NUM_DEF_SIZE * 2` and reusing them for `BcNum`'s that requested `BC_NUM_DEF_SIZE`. This did reduce the amount of time spent, but it also spent a lot of time in the system allocator for an unknown reason. (When using `strace`, a bunch more `brk` calls showed up.) Just reusing `BcNum`'s that had exactly `BC_NUM_DEF_SIZE` capacity spent the smallest amount of time in both user and system time. This makes sense, especially with the changes to make `BC_NUM_DEF_SIZE` bigger on 64-bit systems, since the vast majority of numbers will only ever use numbers with a size less than or equal to `BC_NUM_DEF_SIZE`. Last of all, `bc`'s signal handling underwent a complete redesign. (This is the reason that this version is `3.0.0` and not `2.8.0`.) The change was to move from a polling approach to signal handling to an interrupt-based approach. Previously, every single loop condition had a check for signals. I suspect that this could be expensive when in tight loops. Now, the signal handler just uses `longjmp()` (actually `siglongjmp()`) to start an unwinding of the stack until it is stopped or the stack is unwound to `main()`, which just returns. If `bc` is currently executing code that cannot be safely interrupted (according to POSIX), then signals are "locked." The signal handler checks if the lock is taken, and if it is, it just sets the status to indicate that a signal arrived. Later, when the signal lock is released, the status is checked to see if a signal came in. If so, the stack unwinding starts. This design eliminates polling in favor of maintaining a stack of `jmp_buf`'s. This has its own performance implications, but it gives better interaction. And the cost of pushing and popping a `jmp_buf` in a function is paid at most twice. Most functions do not pay that price, and most of the rest only pay it once. (There are only some 3 functions in `bc` that push and pop a `jmp_buf` twice.) As a side effect of this change, I had to eliminate the use of `stdio.h` in `bc` because `stdio` does not play nice with signals and `longjmp()`. I implemented custom I/O buffer code that takes a fraction of the size. This means that static builds will be smaller, but non-static builds will be bigger, though they will have less linking time. This change is also good because my history implementation was already bypassing `stdio` for good reasons, and unifying the architecture was a win. Another reason for this change is that my `bc` should *always* behave correctly in the presence of signals like `SIGINT`, `SIGTERM`, and `SIGQUIT`. With the addition of my own I/O buffering, I needed to also make sure that the buffers were correctly flushed even when such signals happened. For this reason, I **removed the option to build without signal support**. As a nice side effect of this change, the error handling code could be changed to take advantage of the stack unwinding that signals used. This means that signals and error handling use the same code paths, which means that the stack unwinding is well-tested. (Errors are tested heavily in the test suite.) It also means that functions do not need to return a status code that ***every*** caller needs to check. This eliminated over 100 branches that simply checked return codes and then passed that return code up the stack if necessary. The code bloat savings from this is at least 1700 bytes on `x86_64`, *before* taking into account the extra code from removing `stdio.h`. ## 2.7.2 This is a production release with one major bug fix. The `length()` built-in function can take either a number or an array. If it takes an array, it returns the length of the array. Arrays can be passed by reference. The bug is that the `length()` function would not properly dereference arrays that were references. This is a bug that affects all users. **ALL USERS SHOULD UPDATE `bc`**. ## 2.7.1 This is a production release with fixes for new locales and fixes for compiler warnings on FreeBSD. ## 2.7.0 This is a production release with a bug fix for Linux, new translations, and new features. Bug fixes: * Option parsing in `BC_ENV_ARGS` was broken on Linux in 2.6.1 because `glibc`'s `getopt_long()` is broken. To get around that, and to support long options on every platform, an adapted version of [`optparse`][17] was added. Now, `bc` does not even use `getopt()`. * Parsing `BC_ENV_ARGS` with quotes now works. It isn't the smartest, but it does the job if there are spaces in file names. The following new languages are supported: * Dutch * Polish * Russian * Japanes * Simplified Chinese All of these translations were generated using [DeepL][18], so improvements are welcome. There is only one new feature: **`bc` now has a built-in pseudo-random number generator** (PRNG). The PRNG is seeded, making it useful for applications where `/dev/urandom` does not work because output needs to be reproducible. However, it also uses `/dev/urandom` to seed itself by default, so it will start with a good seed by default. It also outputs 32 bits on 32-bit platforms and 64 bits on 64-bit platforms, far better than the 15 bits of C's `rand()` and `bash`'s `$RANDOM`. In addition, the PRNG can take a bound, and when it gets a bound, it automatically adjusts to remove bias. It can also generate numbers of arbitrary size. (As of the time of release, the largest pseudo-random number generated by this `bc` was generated with a bound of `2^(2^20)`.) ***IMPORTANT: read the [`bc` manual][9] and the [`dc` manual][10] to find out exactly what guarantees the PRNG provides. The underlying implementation is not guaranteed to stay the same, but the guarantees that it provides are guaranteed to stay the same regardless of the implementation.*** On top of that, four functions were added to `bc`'s [extended math library][16] to make using the PRNG easier: * `frand(p)`: Generates a number between `[0,1)` to `p` decimal places. * `ifrand(i, p)`: Generates an integer with bound `i` and adds it to `frand(p)`. * `srand(x)`: Randomizes the sign of `x`. In other words, it flips the sign of `x` with probability `0.5`. * `brand()`: Returns a random boolean value (either `0` or `1`). ## 2.6.1 This is a production release with a bug fix for FreeBSD. The bug was that when `bc` was built without long options, it would give a fatal error on every run. This was caused by a mishandling of `optind`. ## 2.6.0 This release is a production release ***with no bugfixes***. If you do not want to upgrade, you don't have to. No source code changed; the only thing that changed was `lib2.bc`. This release adds one function to the [extended math library][16]: `p(x, y)`, which calculates `x` to the power of `y`, whether or not `y` is an integer. (The `^` operator can only accept integer powers.) This release also includes a couple of small tweaks to the [extended math library][16], mostly to fix returning numbers with too high of `scale`. ## 2.5.3 This release is a production release which addresses inconsistencies in the Portuguese locales. No `bc` code was changed. The issues were that the ISO files used different naming, and also that the files that should have been symlinks were not. I did not catch that because GitHub rendered them the exact same way. ## 2.5.2 This release is a production release. No code was changed, but the build system was changed to allow `CFLAGS` to be given to `CC`, like this: ``` CC="gcc -O3 -march=native" ./configure.sh ``` If this happens, the flags are automatically put into `CFLAGS`, and the compiler is set appropriately. In the example above this means that `CC` will be "gcc" and `CFLAGS` will be "-O3 -march=native". This behavior was added to conform to GNU autotools practices. ## 2.5.1 This is a production release which addresses portability concerns discovered in the `bc` build system. No `bc` code was changed. * Support for Solaris SPARC and AIX were added. * Minor documentations edits were performed. * An option for `configure.sh` was added to disable long options if `getopt_long()` is missing. ## 2.5.0 This is a production release with new translations. No code changed. The translations were contributed by [bugcrazy][15], and they are for Portuguese, both Portugal and Brazil locales. ## 2.4.0 This is a production release primarily aimed at improving `dc`. * A couple of copy and paste errors in the [`dc` manual][10] were fixed. * `dc` startup was optimized by making sure it didn't have to set up `bc`-only things. * The `bc` `&&` and `||` operators were made available to `dc` through the `M` and `m` commands, respectively. * `dc` macros were changed to be tail call-optimized. The last item, tail call optimization, means that if the last thing in a macro is a call to another macro, then the old macro is popped before executing the new macro. This change was made to stop `dc` from consuming more and more memory as macros are executed in a loop. The `q` and `Q` commands still respect the "hidden" macros by way of recording how many macros were removed by tail call optimization. ## 2.3.2 This is a production release meant to fix warnings in the Gentoo `ebuild` by making it possible to disable binary stripping. Other users do *not* need to upgrade. ## 2.3.1 This is a production release. It fixes a bug that caused `-1000000000 < -1` to return `0`. This only happened with negative numbers and only if the value on the left was more negative by a certain amount. That said, this bug *is* a bad bug, and needs to be fixed. **ALL USERS SHOULD UPDATE `bc`**. ## 2.3.0 This is a production release with changes to the build system. ## 2.2.0 This release is a production release. It only has new features and performance improvements. 1. The performance of `sqrt(x)` was improved. 2. The new function `root(x, n)` was added to the extended math library to calculate `n`th roots. 3. The new function `cbrt(x)` was added to the extended math library to calculate cube roots. ## 2.1.3 This is a non-critical release; it just changes the build system, and in non-breaking ways: 1. Linked locale files were changed to link to their sources with a relative link. 2. A bug in `configure.sh` that caused long option parsing to fail under `bash` was fixed. ## 2.1.2 This release is not a critical release. 1. A few codes were added to history. 2. Multiplication was optimized a bit more. 3. Addition and subtraction were both optimized a bit more. ## 2.1.1 This release contains a fix for the test suite made for Linux from Scratch: now the test suite prints `pass` when a test is passed. Other than that, there is no change in this release, so distros and other users do not need to upgrade. ## 2.1.0 This release is a production release. The following bugs were fixed: 1. A `dc` bug that caused stack mishandling was fixed. 2. A warning on OpenBSD was fixed. 3. Bugs in `ctrl+arrow` operations in history were fixed. 4. The ability to paste multiple lines in history was added. 5. A `bc` bug, mishandling of array arguments to functions, was fixed. 6. A crash caused by freeing the wrong pointer was fixed. 7. A `dc` bug where strings, in a rare case, were mishandled in parsing was fixed. In addition, the following changes were made: 1. Division was slightly optimized. 2. An option was added to the build to disable printing of prompts. 3. The special case of empty arguments is now handled. This is to prevent errors in scripts that end up passing empty arguments. 4. A harmless bug was fixed. This bug was that, with the pop instructions (mostly) removed (see below), `bc` would leave extra values on its stack for `void` functions and in a few other cases. These extra items would not affect anything put on the stack and would not cause any sort of crash or even buggy behavior, but they would cause `bc` to take more memory than it needed. On top of the above changes, the following optimizations were added: 1. The need for pop instructions in `bc` was removed. 2. Extra tests on every iteration of the interpreter loop were removed. 3. Updating function and code pointers on every iteration of the interpreter loop was changed to only updating them when necessary. 4. Extra assignments to pointers were removed. Altogether, these changes sped up the interpreter by around 2x. ***NOTE***: This is the last release with new features because this `bc` is now considered complete. From now on, only bug fixes and new translations will be added to this `bc`. ## 2.0.3 This is a production, bug-fix release. Two bugs were fixed in this release: 1. A rare and subtle signal handling bug was fixed. 2. A misbehavior on `0` to a negative power was fixed. The last bug bears some mentioning. When I originally wrote power, I did not thoroughly check its error cases; instead, I had it check if the first number was `0` and then if so, just return `0`. However, `0` to a negative power means that `1` will be divided by `0`, which is an error. I caught this, but only after I stopped being cocky. You see, sometime later, I had noticed that GNU `bc` returned an error, correctly, but I thought it was wrong simply because that's not what my `bc` did. I saw it again later and had a double take. I checked for real, finally, and found out that my `bc` was wrong all along. That was bad on me. But the bug was easy to fix, so it is fixed now. There are two other things in this release: 1. Subtraction was optimized by [Stefan Eßer][14]. 2. Division was also optimized, also by Stefan Eßer. ## 2.0.2 This release contains a fix for a possible overflow in the signal handling. I would be surprised if any users ran into it because it would only happen after 2 billion (`2^31-1`) `SIGINT`'s, but I saw it and had to fix it. ## 2.0.1 This release contains very few things that will apply to any users. 1. A slight bug in `dc`'s interactive mode was fixed. 2. A bug in the test suite that was only triggered on NetBSD was fixed. 3. **The `-P`/`--no-prompt` option** was added for users that do not want a prompt. 4. A `make check` target was added as an alias for `make test`. 5. `dc` got its own read prompt: `?> `. ## 2.0.0 This release is a production release. This release is also a little different from previous releases. From here on out, I do not plan on adding any more features to this `bc`; I believe that it is complete. However, there may be bug fix releases in the future, if I or any others manage to find bugs. This release has only a few new features: 1. `atan2(y, x)` was added to the extended math library as both `a2(y, x)` and `atan2(y, x)`. 2. Locales were fixed. 3. A **POSIX shell-compatible script was added as an alternative to compiling `gen/strgen.c`** on a host machine. More details about making the choice between the two can be found by running `./configure.sh --help` or reading the [build manual][13]. 4. Multiplication was optimized by using **diagonal multiplication**, rather than straight brute force. 5. The `locale_install.sh` script was fixed. 6. `dc` was given the ability to **use the environment variable `DC_ENV_ARGS`**. 7. `dc` was also given the ability to **use the `-i` or `--interactive`** options. 8. Printing the prompt was fixed so that it did not print when it shouldn't. 9. Signal handling was fixed. 10. **Handling of `SIGTERM` and `SIGQUIT`** was fixed. 11. The **built-in functions `maxibase()`, `maxobase()`, and `maxscale()`** (the commands `T`, `U`, `V` in `dc`, respectively) were added to allow scripts to query for the max allowable values of those globals. 12. Some incompatibilities with POSIX were fixed. In addition, this release is `2.0.0` for a big reason: the internal format for numbers changed. They used to be a `char` array. Now, they are an array of larger integers, packing more decimal digits into each integer. This has delivered ***HUGE*** performance improvements, especially for multiplication, division, and power. This `bc` should now be the fastest `bc` available, but I may be wrong. ## 1.2.8 This release contains a fix for a harmless bug (it is harmless in that it still works, but it just copies extra data) in the [`locale_install.sh`][12] script. ## 1.2.7 This version contains fixes for the build on Arch Linux. ## 1.2.6 This release removes the use of `local` in shell scripts because it's not POSIX shell-compatible, and also updates a man page that should have been updated a long time ago but was missed. ## 1.2.5 This release contains some missing locale `*.msg` files. ## 1.2.4 This release contains a few bug fixes and new French translations. ## 1.2.3 This release contains a fix for a bug: use of uninitialized data. Such data was only used when outputting an error message, but I am striving for perfection. As Michelangelo said, "Trifles make perfection, and perfection is no trifle." ## 1.2.2 This release contains fixes for OpenBSD. ## 1.2.1 This release contains bug fixes for some rare bugs. ## 1.2.0 This is a production release. There have been several changes since `1.1.0`: 1. The build system had some changes. 2. Locale support has been added. (Patches welcome for translations.) 3. **The ability to turn `ibase`, `obase`, and `scale` into stacks** was added with the `-g` command-line option. (See the [`bc` manual][9] for more details.) 4. Support for compiling on Mac OSX out of the box was added. 5. The extended math library got `t(x)`, `ceil(x)`, and some aliases. 6. The extended math library also got `r2d(x)` (for converting from radians to degrees) and `d2r(x)` (for converting from degrees to radians). This is to allow using degrees with the standard library. 7. Both calculators now accept numbers in **scientific notation**. See the [`bc` manual][9] and the [`dc` manual][10] for details. 8. Both calculators can **output in either scientific or engineering notation**. See the [`bc` manual][9] and the [`dc` manual][10] for details. 9. Some inefficiencies were removed. 10. Some bugs were fixed. 11. Some bugs in the extended library were fixed. 12. Some defects from [Coverity Scan][11] were fixed. ## 1.1.4 This release contains a fix to the build system that allows it to build on older versions of `glibc`. ## 1.1.3 This release contains a fix for a bug in the test suite where `bc` tests and `dc` tests could not be run in parallel. ## 1.1.2 This release has a fix for a history bug; the down arrow did not work. ## 1.1.1 This release fixes a bug in the `1.1.0` build system. The source is exactly the same. The bug that was fixed was a failure to install if no `EXECSUFFIX` was used. ## 1.1.0 This is a production release. However, many new features were added since `1.0`. 1. **The build system has been changed** to use a custom, POSIX shell-compatible configure script ([`configure.sh`][6]) to generate a POSIX make-compatible `Makefile`, which means that `bc` and `dc` now build out of the box on any POSIX-compatible system. 2. Out-of-memory and output errors now cause the `bc` to report the error, clean up, and die, rather than just reporting and trying to continue. 3. **Strings and constants are now garbage collected** when possible. 4. Signal handling and checking has been made more simple and more thorough. 5. `BcGlobals` was refactored into `BcVm` and `BcVm` was made global. Some procedure names were changed to reflect its difference to everything else. 6. Addition got a speed improvement. 7. Some common code for addition and multiplication was refactored into its own procedure. 8. A bug was removed where `dc` could have been selected, but the internal `#define` that returned `true` for a query about `dc` would not have returned `true`. 9. Useless calls to `bc_num_zero()` were removed. 10. **History support was added.** The history support is based off of a [UTF-8 aware fork][7] of [`linenoise`][8], which has been customized with `bc`'s own data structures and signal handling. 11. Generating C source from the math library now removes tabs from the library, shrinking the size of the executable. 12. The math library was shrunk. 13. Error handling and reporting was improved. 14. Reallocations were reduced by giving access to the request size for each operation. 15. **`abs()` (`b` command for `dc`) was added as a builtin.** 16. Both calculators were tested on FreeBSD. 17. Many obscure parse bugs were fixed. 18. Markdown and man page manuals were added, and the man pages are installed by `make install`. 19. Executable size was reduced, though the added features probably made the executable end up bigger. 20. **GNU-style array references were added as a supported feature.** 21. Allocations were reduced. 22. **New operators were added**: `$` (`$` for `dc`), `@` (`@` for `dc`), `@=`, `<<` (`H` for `dc`), `<<=`, `>>` (`h` for `dc`), and `>>=`. See the [`bc` manual][9] and the [`dc` manual][10] for more details. 23. **An extended math library was added.** This library contains code that makes it so I can replace my desktop calculator with this `bc`. See the [`bc` manual][3] for more details. 24. Support for all capital letters as numbers was added. 25. **Support for GNU-style void functions was added.** 26. A bug fix for improper handling of function parameters was added. 27. Precedence for the or (`||`) operator was changed to match GNU `bc`. 28. `dc` was given an explicit negation command. 29. `dc` was changed to be able to handle strings in arrays. ## 1.1 Release Candidate 3 This release is the eighth release candidate for 1.1, though it is the third release candidate meant as a general release candidate. The new code has not been tested as thoroughly as it should for release. ## 1.1 Release Candidate 2 This release is the seventh release candidate for 1.1, though it is the second release candidate meant as a general release candidate. The new code has not been tested as thoroughly as it should for release. ## 1.1 FreeBSD Beta 5 This release is the sixth release candidate for 1.1, though it is the fifth release candidate meant specifically to test if `bc` works on FreeBSD. The new code has not been tested as thoroughly as it should for release. ## 1.1 FreeBSD Beta 4 This release is the fifth release candidate for 1.1, though it is the fourth release candidate meant specifically to test if `bc` works on FreeBSD. The new code has not been tested as thoroughly as it should for release. ## 1.1 FreeBSD Beta 3 This release is the fourth release candidate for 1.1, though it is the third release candidate meant specifically to test if `bc` works on FreeBSD. The new code has not been tested as thoroughly as it should for release. ## 1.1 FreeBSD Beta 2 This release is the third release candidate for 1.1, though it is the second release candidate meant specifically to test if `bc` works on FreeBSD. The new code has not been tested as thoroughly as it should for release. ## 1.1 FreeBSD Beta 1 This release is the second release candidate for 1.1, though it is meant specifically to test if `bc` works on FreeBSD. The new code has not been tested as thoroughly as it should for release. ## 1.1 Release Candidate 1 This is the first release candidate for 1.1. The new code has not been tested as thoroughly as it should for release. ## 1.0 This is the first non-beta release. `bc` is ready for production use. As such, a lot has changed since 0.5. 1. `dc` has been added. It has been tested even more thoroughly than `bc` was for `0.5`. It does not have the `!` command, and for security reasons, it never will, so it is complete. 2. `bc` has been more thoroughly tested. An entire section of the test suite (for both programs) has been added to test for errors. 3. A prompt (`>>> `) has been added for interactive mode, making it easier to see inputs and outputs. 4. Interrupt handling has been improved, including elimination of race conditions (as much as possible). 5. MinGW and [Windows Subsystem for Linux][1] support has been added (see [xstatic][2] for binaries). 6. Memory leaks and errors have been eliminated (as far as ASan and Valgrind can tell). 7. Crashes have been eliminated (as far as [afl][3] can tell). 8. Karatsuba multiplication was added (and thoroughly) tested, speeding up multiplication and power by orders of magnitude. 9. Performance was further enhanced by using a "divmod" function to reduce redundant divisions and by removing superfluous `memset()` calls. 10. To switch between Karatsuba and `O(n^2)` multiplication, the config variable `BC_NUM_KARATSUBA_LEN` was added. It is set to a sane default, but the optimal number can be found with [`karatsuba.py`][4] (requires Python 3) and then configured through `make`. 11. The random math test generator script was changed to Python 3 and improved. `bc` and `dc` have together been run through 30+ million random tests. 12. All known math bugs have been fixed, including out of control memory allocations in `sine` and `cosine` (that was actually a parse bug), certain cases of infinite loop on square root, and slight inaccuracies (as much as possible; see the [README][5]) in transcendental functions. 13. Parsing has been fixed as much as possible. 14. Test coverage was improved to 94.8%. The only paths not covered are ones that happen when `malloc()` or `realloc()` fails. 15. An extension to get the length of an array was added. 16. The boolean not (`!`) had its precedence change to match negation. 17. Data input was hardened. 18. `bc` was made fully compliant with POSIX when the `-s` flag is used or `POSIXLY_CORRECT` is defined. 19. Error handling was improved. 20. `bc` now checks that files it is given are not directories. ## 1.0 Release Candidate 7 This is the seventh release candidate for 1.0. It fixes a few bugs in 1.0 Release Candidate 6. ## 1.0 Release Candidate 6 This is the sixth release candidate for 1.0. It fixes a few bugs in 1.0 Release Candidate 5. ## 1.0 Release Candidate 5 This is the fifth release candidate for 1.0. It fixes a few bugs in 1.0 Release Candidate 4. ## 1.0 Release Candidate 4 This is the fourth release candidate for 1.0. It fixes a few bugs in 1.0 Release Candidate 3. ## 1.0 Release Candidate 3 This is the third release candidate for 1.0. It fixes a few bugs in 1.0 Release Candidate 2. ## 1.0 Release Candidate 2 This is the second release candidate for 1.0. It fixes a few bugs in 1.0 Release Candidate 1. ## 1.0 Release Candidate 1 This is the first Release Candidate for 1.0. `bc` is complete, with `dc`, but it is not tested. ## 0.5 This beta release completes more features, but it is still not complete nor tested as thoroughly as necessary. ## 0.4.1 This beta release fixes a few bugs in 0.4. ## 0.4 This is a beta release. It does not have the complete set of features, and it is not thoroughly tested. [1]: https://docs.microsoft.com/en-us/windows/wsl/install-win10 [2]: https://pkg.musl.cc/bc/ [3]: http://lcamtuf.coredump.cx/afl/ [4]: ./scripts/karatsuba.py [5]: ./README.md [6]: ./configure.sh [7]: https://github.com/rain-1/linenoise-mob [8]: https://github.com/antirez/linenoise [9]: ./manuals/bc/A.1.md [10]: ./manuals/dc/A.1.md [11]: https://scan.coverity.com/projects/gavinhoward-bc [12]: ./scripts/locale_install.sh [13]: ./manuals/build.md [14]: https://github.com/stesser [15]: https://github.com/bugcrazy [16]: ./manuals/bc/A.1.md#extended-library [17]: https://github.com/skeeto/optparse [18]: https://www.deepl.com/translator [19]: ./manuals/benchmarks.md [20]: https://github.com/apjanke/ronn-ng [21]: https://pandoc.org/ [22]: ./scripts/locale_uninstall.sh diff --git a/compile_flags.txt b/compile_flags.txt index 7a08c87f3876..3324798013c6 100644 --- a/compile_flags.txt +++ b/compile_flags.txt @@ -1,15 +1,16 @@ -Weverything -pedantic -Wno-unsafe-buffer-usage -D_POSIX_C_SOURCE=200809L -D_XOPEN_SOURCE=700 -D_BSD_SOURCE -D_GNU_SOURCE -D_DEFAULT_SOURCE -Iinclude/ -DBC_DEBUG=1 -DBC_ENABLED=1 -DDC_ENABLED=1 -DBC_ENABLE_EXTRA_MATH=1 -DBC_ENABLE_HISTORY=1 -DBC_ENABLE_NLS=1 +-DBC_ENABLE_OSSFUZZ=0 diff --git a/configure.sh b/configure.sh index 43bb502ea817..442165d15693 100755 --- a/configure.sh +++ b/configure.sh @@ -1,2127 +1,2201 @@ #! /bin/sh # # SPDX-License-Identifier: BSD-2-Clause # # Copyright (c) 2018-2024 Gavin D. Howard and contributors. # # Redistribution and use in source and binary forms, with or without # modification, are permitted provided that the following conditions are met: # # * Redistributions of source code must retain the above copyright notice, this # list of conditions and the following disclaimer. # # * Redistributions in binary form must reproduce the above copyright notice, # this list of conditions and the following disclaimer in the documentation # and/or other materials provided with the distribution. # # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" # AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE # IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE # ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE # LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR # CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF # SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS # INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN # CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) # ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE # POSSIBILITY OF SUCH DAMAGE. # script="$0" scriptdir=$(dirname "$script") script=$(basename "$script") builddir=$(pwd) . "$scriptdir/scripts/functions.sh" # Simply prints the help message and quits based on the argument. # @param msg The help message to print. usage() { if [ $# -gt 0 ]; then _usage_val=1 printf '%s\n\n' "$1" else _usage_val=0 fi printf 'usage:\n' printf ' %s -h\n' "$script" printf ' %s --help\n' "$script" printf ' %s [-a|-bD|-dB|-c] [-CeEfgGHilmMNPrtTvz] [-O OPT_LEVEL] [-k KARATSUBA_LEN]\\\n' "$script" printf ' [-s SETTING] [-S SETTING] [-p TYPE]\n' printf ' %s \\\n' "$script" printf ' [--library|--bc-only --disable-dc|--dc-only --disable-bc|--coverage] \\\n' printf ' [--force --debug --disable-extra-math --disable-generated-tests] \\\n' printf ' [--disable-history --disable-man-pages --disable-nls --disable-strip] \\\n' printf ' [--enable-editline] [--enable-readline] [--enable-internal-history] \\\n' printf ' [--disable-problematic-tests] [--install-all-locales] \\\n' printf ' [--opt=OPT_LEVEL] [--karatsuba-len=KARATSUBA_LEN] \\\n' printf ' [--set-default-on=SETTING] [--set-default-off=SETTING] \\\n' printf ' [--predefined-build-type=TYPE] \\\n' printf ' [--prefix=PREFIX] [--bindir=BINDIR] [--datarootdir=DATAROOTDIR] \\\n' printf ' [--datadir=DATADIR] [--mandir=MANDIR] [--man1dir=MAN1DIR] \\\n' printf ' [--man3dir=MAN3DIR]\n' if [ "$_usage_val" -ne 0 ]; then - exit + exit "$_usage_val" fi printf '\n' printf ' -a, --library\n' printf ' Build the libbcl instead of the programs. This is meant to be used with\n' printf ' Other software like programming languages that want to make use of the\n' printf ' parsing and math capabilities. This option will install headers using\n' printf ' `make install`.\n' printf ' -b, --bc-only\n' printf ' Build bc only. It is an error if "-d", "--dc-only", "-B", or\n' printf ' "--disable-bc" are specified too.\n' printf ' -B, --disable-bc\n' printf ' Disable bc. It is an error if "-b", "--bc-only", "-D", or "--disable-dc"\n' printf ' are specified too.\n' printf ' -c, --coverage\n' printf ' Generate test coverage code. Requires gcov and gcovr.\n' printf ' It is an error if either "-b" ("-D") or "-d" ("-B") is specified.\n' printf ' Requires a compiler that use gcc-compatible coverage options\n' printf ' -C, --disable-clean\n' printf ' Disable the clean that configure.sh does before configure.\n' printf ' -d, --dc-only\n' printf ' Build dc only. It is an error if "-b", "--bc-only", "-D", or\n' printf ' "--disable-dc" are specified too.\n' printf ' -D, --disable-dc\n' printf ' Disable dc. It is an error if "-d", "--dc-only", "-B", or "--disable-bc"\n' printf ' are specified too.\n' printf ' -e, --enable-editline\n' printf ' Enable the use of libedit/editline. This is meant for those users that\n' printf ' want vi-like or Emacs-like behavior in history. This option is ignored\n' printf ' if history is disabled. If the -r or -i options are given with this\n' printf ' option, the last occurrence of all of the three is used.\n' printf ' -E, --disable-extra-math\n' printf ' Disable extra math. This includes: "$" operator (truncate to integer),\n' printf ' "@" operator (set number of decimal places), and r(x, p) (rounding\n' printf ' function). Additionally, this option disables the extra printing\n' printf ' functions in the math library.\n' printf ' -f, --force\n' printf ' Force use of all enabled options, even if they do not work. This\n' printf ' option is to allow the maintainer a way to test that certain options\n' printf ' are not failing invisibly. (Development only.)\n' printf ' -g, --debug\n' printf ' Build in debug mode. Adds the "-g" flag, and if there are no\n' printf ' other CFLAGS, and "-O" was not given, this also adds the "-O0"\n' printf ' flag. If this flag is *not* given, "-DNDEBUG" is added to CPPFLAGS\n' printf ' and a strip flag is added to the link stage.\n' printf ' -G, --disable-generated-tests\n' printf ' Disable generating tests. This is for platforms that do not have a\n' printf ' GNU bc-compatible bc to generate tests.\n' printf ' -h, --help\n' printf ' Print this help message and exit.\n' printf ' -H, --disable-history\n' printf ' Disable history.\n' printf ' -i, --enable-internal-history\n' printf ' Enable the internal history implementation and do not depend on either\n' printf ' editline or readline. This option is ignored if history is disabled.\n' printf ' If this option is given along with -e and -r, the last occurrence of\n' printf ' all of the three is used.\n' printf ' -k KARATSUBA_LEN, --karatsuba-len KARATSUBA_LEN\n' printf ' Set the karatsuba length to KARATSUBA_LEN (default is 32).\n' printf ' It is an error if KARATSUBA_LEN is not a number or is less than 16.\n' printf ' -l, --install-all-locales\n' printf ' Installs all locales, regardless of how many are on the system. This\n' printf ' option is useful for package maintainers who want to make sure that\n' printf ' a package contains all of the locales that end users might need.\n' printf ' -m, --enable-memcheck\n' printf ' Enable memcheck mode, to ensure no memory leaks. For development only.\n' printf ' -M, --disable-man-pages\n' printf ' Disable installing manpages.\n' printf ' -N, --disable-nls\n' printf ' Disable POSIX locale (NLS) support.\n' printf ' ***WARNING***: Locales ignore the prefix because they *must* be\n' printf ' installed at a fixed location to work at all. If you do not want that\n' printf ' to happen, you must disable locales (NLS) completely.\n' printf ' -O OPT_LEVEL, --opt OPT_LEVEL\n' printf ' Set the optimization level. This can also be included in the CFLAGS,\n' printf ' but it is provided, so maintainers can build optimized debug builds.\n' printf ' This is passed through to the compiler, so it must be supported.\n' printf ' -p TYPE, --predefined-build-type=TYPE\n' printf ' Sets a given predefined build type with specific defaults. This is for\n' printf ' easy setting of predefined builds. For example, to get a build that\n' printf ' acts like the GNU bc by default, TYPE should be "GNU" (without the\n' printf ' quotes) This option *must* come before any others that might change the\n' printf ' build options. Currently supported values for TYPE include: "BSD" (for\n' printf ' matching the BSD bc and BSD dc), "GNU" (for matching the GNU bc and\n' printf ' dc), "GDH" (for the preferred build of the author, Gavin D. Howard),\n' printf ' and "DBG" (for the preferred debug build of the author). This will\n' printf ' also automatically enable a release build (except for "DBG").\n' printf ' -P, --disable-problematic-tests\n' printf ' Disables problematic tests. These tests usually include tests that\n' printf ' can cause a SIGKILL because of too much memory usage.\n' printf ' -r, --enable-readline\n' printf ' Enable the use of libreadline/readline. This is meant for those users\n' printf ' that want vi-like or Emacs-like behavior in history. This option is\n' printf ' ignored if history is disabled. If this option is given along with -e\n' printf ' and -i, the last occurrence of all of the three is used.\n' printf ' -s SETTING, --set-default-on SETTING\n' printf ' Set the default named by SETTING to on. See below for possible values\n' printf ' for SETTING. For multiple instances of the -s or -S for the the same\n' printf ' setting, the last one is used.\n' printf ' -S SETTING, --set-default-off SETTING\n' printf ' Set the default named by SETTING to off. See below for possible values\n' printf ' for SETTING. For multiple instances of the -s or -S for the the same\n' printf ' setting, the last one is used.\n' printf ' -t, --enable-test-timing\n' printf ' Enable the timing of tests. This is for development only.\n' printf ' -T, --disable-strip\n' printf ' Disable stripping symbols from the compiled binary or binaries.\n' printf ' Stripping symbols only happens when debug mode is off.\n' printf ' -v, --enable-valgrind\n' printf ' Enable a build appropriate for valgrind. For development only.\n' printf ' -z, --enable-fuzz-mode\n' printf ' Enable fuzzing mode. THIS IS FOR DEVELOPMENT ONLY.\n' + printf ' -Z, --enable-ossfuzz-mode\n' + printf ' Enable fuzzing mode for OSS-Fuzz. THIS IS FOR DEVELOPMENT ONLY.\n' printf ' --prefix PREFIX\n' printf ' The prefix to install to. Overrides "$PREFIX" if it exists.\n' printf ' If PREFIX is "/usr", install path will be "/usr/bin".\n' printf ' Default is "/usr/local".\n' printf ' ***WARNING***: Locales ignore the prefix because they *must* be\n' printf ' installed at a fixed location to work at all. If you do not want that to\n' printf ' happen, you must disable locales (NLS) completely.\n' printf ' --bindir BINDIR\n' printf ' The directory to install binaries in. Overrides "$BINDIR" if it exists.\n' printf ' Default is "$PREFIX/bin".\n' printf ' --includedir INCLUDEDIR\n' printf ' The directory to install headers in. Overrides "$INCLUDEDIR" if it\n' printf ' exists. Default is "$PREFIX/include".\n' printf ' --libdir LIBDIR\n' printf ' The directory to install libraries in. Overrides "$LIBDIR" if it exists.\n' printf ' Default is "$PREFIX/lib".\n' printf ' --datarootdir DATAROOTDIR\n' printf ' The root location for data files. Overrides "$DATAROOTDIR" if it exists.\n' printf ' Default is "$PREFIX/share".\n' printf ' --datadir DATADIR\n' printf ' The location for data files. Overrides "$DATADIR" if it exists.\n' printf ' Default is "$DATAROOTDIR".\n' printf ' --mandir MANDIR\n' printf ' The location to install manpages to. Overrides "$MANDIR" if it exists.\n' printf ' Default is "$DATADIR/man".\n' printf ' --man1dir MAN1DIR\n' printf ' The location to install Section 1 manpages to. Overrides "$MAN1DIR" if\n' printf ' it exists. Default is "$MANDIR/man1".\n' printf ' --man3dir MAN3DIR\n' printf ' The location to install Section 3 manpages to. Overrides "$MAN3DIR" if\n' printf ' it exists. Default is "$MANDIR/man3".\n' printf '\n' printf 'In addition, the following environment variables are used:\n' printf '\n' printf ' CC C compiler. Must be compatible with POSIX c99. If there is a\n' printf ' space in the basename of the compiler, the items after the\n' printf ' first space are assumed to be compiler flags, and in that case,\n' printf ' the flags are automatically moved into CFLAGS. Default is\n' printf ' "c99".\n' printf ' HOSTCC Host C compiler. Must be compatible with POSIX c99. If there is\n' printf ' a space in the basename of the compiler, the items after the\n' printf ' first space are assumed to be compiler flags, and in the case,\n' printf ' the flags are automatically moved into HOSTCFLAGS. Default is\n' printf ' "$CC".\n' printf ' HOST_CC Same as HOSTCC. If HOSTCC also exists, it is used.\n' printf ' CFLAGS C compiler flags.\n' printf ' HOSTCFLAGS CFLAGS for HOSTCC. Default is "$CFLAGS".\n' printf ' HOST_CFLAGS Same as HOST_CFLAGS. If HOST_CFLAGS also exists, it is used.\n' printf ' CPPFLAGS C preprocessor flags. Default is "".\n' printf ' LDFLAGS Linker flags. Default is "".\n' printf ' PREFIX The prefix to install to. Default is "/usr/local".\n' printf ' If PREFIX is "/usr", install path will be "/usr/bin".\n' printf ' ***WARNING***: Locales ignore the prefix because they *must* be\n' printf ' installed at a fixed location to work at all. If you do not\n' printf ' want that to happen, you must disable locales (NLS) completely.\n' printf ' BINDIR The directory to install binaries in. Default is "$PREFIX/bin".\n' printf ' INCLUDEDIR The directory to install header files in. Default is\n' printf ' "$PREFIX/include".\n' printf ' LIBDIR The directory to install libraries in. Default is\n' printf ' "$PREFIX/lib".\n' printf ' DATAROOTDIR The root location for data files. Default is "$PREFIX/share".\n' printf ' DATADIR The location for data files. Default is "$DATAROOTDIR".\n' printf ' MANDIR The location to install manpages to. Default is "$DATADIR/man".\n' printf ' MAN1DIR The location to install Section 1 manpages to. Default is\n' printf ' "$MANDIR/man1".\n' printf ' MAN3DIR The location to install Section 3 manpages to. Default is\n' printf ' "$MANDIR/man3".\n' printf ' NLSPATH The location to install locale catalogs to. Must be an absolute\n' printf ' path (or contain one). This is treated the same as the POSIX\n' printf ' definition of $NLSPATH (see POSIX environment variables for\n' printf ' more information). Default is "/usr/share/locale/%%L/%%N".\n' printf ' PC_PATH The location to install pkg-config files to. Must be an\n' printf ' path or contain one. Default is the first path given by the\n' printf ' output of `pkg-config --variable=pc_path pkg-config`.\n' printf ' EXECSUFFIX The suffix to append to the executable names, used to not\n' printf ' interfere with other installed bc executables. Default is "".\n' printf ' EXECPREFIX The prefix to append to the executable names, used to not\n' printf ' interfere with other installed bc executables. Default is "".\n' printf ' DESTDIR For package creation. Default is "". If it is empty when\n' printf ' `%s` is run, it can also be passed to `make install`\n' "$script" printf ' later as an environment variable. If both are specified,\n' printf ' the one given to `%s` takes precedence.\n' "$script" printf ' LONG_BIT The number of bits in a C `long` type. This is mostly for the\n' printf ' embedded space since this `bc` uses `long`s internally for\n' printf ' overflow checking. In C99, a `long` is required to be 32 bits.\n' printf ' For most normal desktop systems, setting this is unnecessary,\n' printf ' except that 32-bit platforms with 64-bit longs may want to set\n' printf ' it to `32`. Default is the default of `LONG_BIT` for the target\n' printf ' platform. Minimum allowed is `32`. It is a build time error if\n' printf ' the specified value of `LONG_BIT` is greater than the default\n' printf ' value of `LONG_BIT` for the target platform.\n' printf ' GEN_HOST Whether to use `gen/strgen.c`, instead of `gen/strgen.sh`, to\n' printf ' produce the C files that contain the help texts as well as the\n' printf ' math libraries. By default, `gen/strgen.c` is used, compiled by\n' printf ' "$HOSTCC" and run on the host machine. Using `gen/strgen.sh`\n' printf ' removes the need to compile and run an executable on the host\n' printf ' machine since `gen/strgen.sh` is a POSIX shell script. However,\n' printf ' `gen/lib2.bc` is over 4095 characters, the max supported length\n' printf ' of a string literal in C99, and `gen/strgen.sh` generates a\n' printf ' string literal instead of an array, as `gen/strgen.c` does. For\n' printf ' most production-ready compilers, this limit probably is not\n' printf ' enforced, but it could be. Both options are still available for\n' printf ' this reason. If you are sure your compiler does not have the\n' printf ' limit and do not want to compile and run a binary on the host\n' printf ' machine, set this variable to "0". Any other value, or a\n' printf ' non-existent value, will cause the build system to compile and\n' printf ' run `gen/strgen.c`. Default is "".\n' printf ' GEN_EMU Emulator to run string generator code under (leave empty if not\n' printf ' necessary). This is not necessary when using `gen/strgen.sh`.\n' printf ' Default is "".\n' printf '\n' printf 'WARNING: even though `configure.sh` supports both option types, short and\n' printf 'long, it does not support handling both at the same time. Use only one type.\n' printf '\n' printf 'Settings\n' printf '========\n' printf '\n' printf 'bc and dc have some settings that, while they cannot be removed by build time\n' printf 'options, can have their defaults changed at build time by packagers. Users are\n' printf 'also able to change each setting with environment variables.\n' printf '\n' printf 'The following is a table of settings, along with their default values and the\n' printf 'environment variables users can use to change them. (For the defaults, non-zero\n' printf 'means on, and zero means off.)\n' printf '\n' printf '| Setting | Description | Default | Env Variable |\n' printf '| =============== | ==================== | ============ | ==================== |\n' printf '| bc.banner | Whether to display | 0 | BC_BANNER |\n' printf '| | the bc version | | |\n' printf '| | banner when in | | |\n' printf '| | interactive mode. | | |\n' printf '| --------------- | -------------------- | ------------ | -------------------- |\n' printf '| bc.sigint_reset | Whether SIGINT will | 1 | BC_SIGINT_RESET |\n' printf '| | reset bc, instead of | | |\n' printf '| | exiting, when in | | |\n' printf '| | interactive mode. | | |\n' printf '| --------------- | -------------------- | ------------ | -------------------- |\n' printf '| dc.sigint_reset | Whether SIGINT will | 1 | DC_SIGINT_RESET |\n' printf '| | reset dc, instead of | | |\n' printf '| | exiting, when in | | |\n' printf '| | interactive mode. | | |\n' printf '| --------------- | -------------------- | ------------ | -------------------- |\n' printf '| bc.tty_mode | Whether TTY mode for | 1 | BC_TTY_MODE |\n' printf '| | bc should be on when | | |\n' printf '| | available. | | |\n' printf '| --------------- | -------------------- | ------------ | -------------------- |\n' printf '| dc.tty_mode | Whether TTY mode for | 0 | BC_TTY_MODE |\n' printf '| | dc should be on when | | |\n' printf '| | available. | | |\n' printf '| --------------- | -------------------- | ------------ | -------------------- |\n' printf '| bc.prompt | Whether the prompt | $BC_TTY_MODE | BC_PROMPT |\n' printf '| | for bc should be on | | |\n' printf '| | in tty mode. | | |\n' printf '| --------------- | -------------------- | ------------ | -------------------- |\n' printf '| dc.prompt | Whether the prompt | $DC_TTY_MODE | DC_PROMPT |\n' printf '| | for dc should be on | | |\n' printf '| | in tty mode. | | |\n' printf '| --------------- | -------------------- | ------------ | -------------------- |\n' printf '| bc.expr_exit | Whether to exit bc | 1 | BC_EXPR_EXIT |\n' printf '| | if an expression or | | |\n' printf '| | expression file is | | |\n' printf '| | given with the -e or | | |\n' printf '| | -f options. | | |\n' printf '| --------------- | -------------------- | ------------ | -------------------- |\n' printf '| dc.expr_exit | Whether to exit dc | 1 | DC_EXPR_EXIT |\n' printf '| | if an expression or | | |\n' printf '| | expression file is | | |\n' printf '| | given with the -e or | | |\n' printf '| | -f options. | | |\n' printf '| --------------- | -------------------- | ------------ | -------------------- |\n' printf '| bc.digit_clamp | Whether to have bc | 0 | BC_DIGIT_CLAMP |\n' printf '| | clamp digits that | | |\n' printf '| | are greater than or | | |\n' printf '| | equal to the current | | |\n' printf '| | ibase when parsing | | |\n' printf '| | numbers. | | |\n' printf '| --------------- | -------------------- | ------------ | -------------------- |\n' printf '| dc.digit_clamp | Whether to have dc | 0 | DC_DIGIT_CLAMP |\n' printf '| | clamp digits that | | |\n' printf '| | are greater than or | | |\n' printf '| | equal to the current | | |\n' printf '| | ibase when parsing | | |\n' printf '| | numbers. | | |\n' printf '| --------------- | -------------------- | ------------ | -------------------- |\n' printf '\n' printf 'These settings are not meant to be changed on a whim. They are meant to ensure\n' printf 'that this bc and dc will conform to the expectations of the user on each\n' printf 'platform.\n' exit "$_usage_val" } # Replaces a file extension in a filename. This is used mostly to turn filenames # like `src/num.c` into `src/num.o`. In other words, it helps to link targets to # the files they depend on. # # @param file The filename. # @param ext1 The extension to replace. # @param ext2 The new extension. replace_ext() { if [ "$#" -ne 3 ]; then err_exit "Invalid number of args to $0" fi _replace_ext_file="$1" _replace_ext_ext1="$2" _replace_ext_ext2="$3" _replace_ext_result="${_replace_ext_file%.$_replace_ext_ext1}.$_replace_ext_ext2" printf '%s\n' "$_replace_ext_result" } # Replaces a file extension in every filename given in a list. The list is just # a space-separated list of words, so filenames are expected to *not* have # spaces in them. See the documentation for `replace_ext()`. # # @param files The list of space-separated filenames to replace extensions for. # @param ext1 The extension to replace. # @param ext2 The new extension. replace_exts() { if [ "$#" -ne 3 ]; then err_exit "Invalid number of args to $0" fi _replace_exts_files="$1" _replace_exts_ext1="$2" _replace_exts_ext2="$3" for _replace_exts_file in $_replace_exts_files; do _replace_exts_new_name=$(replace_ext "$_replace_exts_file" "$_replace_exts_ext1" "$_replace_exts_ext2") _replace_exts_result="$_replace_exts_result $_replace_exts_new_name" done printf '%s\n' "$_replace_exts_result" } # Finds a placeholder in @a str and replaces it. This is the workhorse of # configure.sh. It's what replaces placeholders in Makefile.in with the data # needed for the chosen build. Below, you will see a lot of calls to this # function. # # Note that needle can never contain an exclamation point. For more information, # see substring_replace() in scripts/functions.sh. # # @param str The string to find and replace placeholders in. # @param needle The placeholder name. # @param replacement The string to use to replace the placeholder. replace() { if [ "$#" -ne 3 ]; then err_exit "Invalid number of args to $0" fi _replace_str="$1" _replace_needle="$2" _replace_replacement="$3" substring_replace "$_replace_str" "%%$_replace_needle%%" "$_replace_replacement" } # This function finds all the source files that need to be built. If there is # only one argument and it is empty, then all source files are built. Otherwise, # the arguments are all assumed to be source files that should *not* be built. find_src_files() { _find_src_files_args="" if [ "$#" -ge 1 ] && [ "$1" != "" ]; then while [ "$#" -ge 1 ]; do _find_src_files_a="${1## }" shift _find_src_files_args=$(printf '%s\n%s/src/%s\n' "$_find_src_files_args" "$scriptdir" "${_find_src_files_a}") done fi _find_src_files_files=$(find "$scriptdir/src" -depth -name "*.c" -print | LC_ALL=C sort) _find_src_files_result="" for _find_src_files_f in $_find_src_files_files; do # If this is true, the file is part of args, and therefore, unneeded. if [ "${_find_src_files_args##*$_find_src_files_f}" != "${_find_src_files_args}" ]; then continue fi _find_src_files_result=$(printf '%s\n%s\n' "$_find_src_files_result" "$_find_src_files_f") done printf '%s\n' "$_find_src_files_result" } # This function generates a list of files to go into the Makefile. It generates # the list of object files, as well as the list of test coverage files. # # @param contents The contents of the Makefile template to put the list of # files into. gen_file_list() { if [ "$#" -lt 1 ]; then err_exit "Invalid number of args to $0" fi _gen_file_list_contents="$1" shift if [ "$#" -ge 1 ]; then _gen_file_list_unneeded="$@" else _gen_file_list_unneeded="" fi _gen_file_list_needle_src="SRC" _gen_file_list_needle_obj="OBJ" _gen_file_list_needle_gcda="GCDA" _gen_file_list_needle_gcno="GCNO" _gen_file_list_replacement=$(find_src_files $_gen_file_list_unneeded | tr '\n' ' ') _gen_file_list_contents=$(replace "$_gen_file_list_contents" \ "$_gen_file_list_needle_src" "$_gen_file_list_replacement") _gen_file_list_cbases="" for _gen_file_list_f in $_gen_file_list_replacement; do _gen_file_list_b=$(basename "$_gen_file_list_f") _gen_file_list_cbases="$_gen_file_list_cbases src/$_gen_file_list_b" done _gen_file_list_replacement=$(replace_exts "$_gen_file_list_cbases" "c" "o") _gen_file_list_contents=$(replace "$_gen_file_list_contents" \ "$_gen_file_list_needle_obj" "$_gen_file_list_replacement") _gen_file_list_replacement=$(replace_exts "$_gen_file_list_replacement" "o" "gcda") _gen_file_list_contents=$(replace "$_gen_file_list_contents" \ "$_gen_file_list_needle_gcda" "$_gen_file_list_replacement") _gen_file_list_replacement=$(replace_exts "$_gen_file_list_replacement" "gcda" "gcno") _gen_file_list_contents=$(replace "$_gen_file_list_contents" \ "$_gen_file_list_needle_gcno" "$_gen_file_list_replacement") printf '%s\n' "$_gen_file_list_contents" } # Generates the proper test targets for each test to have its own target. This # allows `make test` to run in parallel. # # @param name Which calculator to generate tests for. # @param extra_math An integer that, if non-zero, activates extra math tests. # @param time_tests An integer that, if non-zero, tells the test suite to time # the execution of each test. gen_std_tests() { _gen_std_tests_name="$1" shift _gen_std_tests_extra_math="$1" shift _gen_std_tests_time_tests="$1" shift _gen_std_tests_extra_required=$(cat "$scriptdir/tests/extra_required.txt") for _gen_std_tests_t in $(cat "$scriptdir/tests/$_gen_std_tests_name/all.txt"); do if [ "$_gen_std_tests_extra_math" -eq 0 ]; then if [ -z "${_gen_std_tests_extra_required##*$_gen_std_tests_t*}" ]; then printf 'test_%s_%s:\n\t@printf "Skipping %s %s\\n"\n\n' \ "$_gen_std_tests_name" "$_gen_std_tests_t" "$_gen_std_tests_name" \ "$_gen_std_tests_t" >> "Makefile" continue fi fi printf 'test_%s_%s:\n\t@export BC_TEST_OUTPUT_DIR="%s/tests"; sh $(TESTSDIR)/test.sh %s %s %s %s %s\n\n' \ "$_gen_std_tests_name" "$_gen_std_tests_t" "$builddir" "$_gen_std_tests_name" \ "$_gen_std_tests_t" "$generate_tests" "$time_tests" \ "$*" >> "Makefile" done } # Generates a list of test targets that will be used as prerequisites for other # targets. # # @param name The name of the calculator to generate test targets for. gen_std_test_targets() { _gen_std_test_targets_name="$1" shift _gen_std_test_targets_tests=$(cat "$scriptdir/tests/${_gen_std_test_targets_name}/all.txt") for _gen_std_test_targets_t in $_gen_std_test_targets_tests; do printf ' test_%s_%s' "$_gen_std_test_targets_name" "$_gen_std_test_targets_t" done printf '\n' } # Generates the proper test targets for each error test to have its own target. # This allows `make test_bc_errors` and `make test_dc_errors` to run in # parallel. # # @param name Which calculator to generate tests for. gen_err_tests() { _gen_err_tests_name="$1" shift _gen_err_tests_fs=$(ls "$scriptdir/tests/$_gen_err_tests_name/errors/") for _gen_err_tests_t in $_gen_err_tests_fs; do printf 'test_%s_error_%s:\n\t@export BC_TEST_OUTPUT_DIR="%s/tests"; sh $(TESTSDIR)/error.sh %s %s %s %s\n\n' \ "$_gen_err_tests_name" "$_gen_err_tests_t" "$builddir" "$_gen_err_tests_name" \ "$_gen_err_tests_t" "$problematic_tests" "$*" >> "Makefile" done } # Generates a list of error test targets that will be used as prerequisites for # other targets. # # @param name The name of the calculator to generate test targets for. gen_err_test_targets() { _gen_err_test_targets_name="$1" shift _gen_err_test_targets_tests=$(ls "$scriptdir/tests/$_gen_err_test_targets_name/errors/") for _gen_err_test_targets_t in $_gen_err_test_targets_tests; do printf ' test_%s_error_%s' "$_gen_err_test_targets_name" "$_gen_err_test_targets_t" done printf '\n' } # Generates the proper script test targets for each script test to have its own # target. This allows `make test` to run in parallel. # # @param name Which calculator to generate tests for. # @param extra_math An integer that, if non-zero, activates extra math tests. # @param generate An integer that, if non-zero, activates generated tests. # @param time_tests An integer that, if non-zero, tells the test suite to time # the execution of each test. gen_script_tests() { _gen_script_tests_name="$1" shift _gen_script_tests_extra_math="$1" shift _gen_script_tests_generate="$1" shift _gen_script_tests_time="$1" shift _gen_script_tests_tests=$(cat "$scriptdir/tests/$_gen_script_tests_name/scripts/all.txt") for _gen_script_tests_f in $_gen_script_tests_tests; do _gen_script_tests_b=$(basename "$_gen_script_tests_f" ".${_gen_script_tests_name}") printf 'test_%s_script_%s:\n\t@export BC_TEST_OUTPUT_DIR="%s/tests"; sh $(TESTSDIR)/script.sh %s %s %s 1 %s %s %s\n\n' \ "$_gen_script_tests_name" "$_gen_script_tests_b" "$builddir" "$_gen_script_tests_name" \ "$_gen_script_tests_f" "$_gen_script_tests_extra_math" "$_gen_script_tests_generate" \ "$_gen_script_tests_time" "$*" >> "Makefile" done } set_default() { _set_default_on="$1" shift _set_default_name="$1" shift # The reason that the variables that are being set do not have the same # non-collision avoidance that the other variables do is that we *do* want # the settings of these variables to leak out of the function. They adjust # the settings outside of the function. case "$_set_default_name" in bc.banner) bc_default_banner="$_set_default_on" ;; bc.sigint_reset) bc_default_sigint_reset="$_set_default_on" ;; dc.sigint_reset) dc_default_sigint_reset="$_set_default_on" ;; bc.tty_mode) bc_default_tty_mode="$_set_default_on" ;; dc.tty_mode) dc_default_tty_mode="$_set_default_on" ;; bc.prompt) bc_default_prompt="$_set_default_on" ;; dc.prompt) dc_default_prompt="$_set_default_on" ;; bc.expr_exit) bc_default_expr_exit="$_set_default_on";; dc.expr_exit) dc_default_expr_exit="$_set_default_on";; bc.digit_clamp) bc_default_digit_clamp="$_set_default_on";; dc.digit_clamp) dc_default_digit_clamp="$_set_default_on";; ?) usage "Invalid setting: $_set_default_name" ;; esac } predefined_build() { _predefined_build_type="$1" shift # The reason that the variables that are being set do not have the same # non-collision avoidance that the other variables do is that we *do* want # the settings of these variables to leak out of the function. They adjust # the settings outside of the function. case "$_predefined_build_type" in BSD) bc_only=0 dc_only=0 coverage=0 debug=0 optimization="3" hist=1 hist_impl="editline" extra_math=1 generate_tests=$generate_tests install_manpages=0 nls=1 force=0 strip_bin=1 all_locales=0 library=0 fuzz=0 + ossfuzz=0 time_tests=0 vg=0 memcheck=0 clean=1 bc_default_banner=0 bc_default_sigint_reset=1 dc_default_sigint_reset=1 bc_default_tty_mode=1 dc_default_tty_mode=0 bc_default_prompt="" dc_default_prompt="" bc_default_expr_exit=1 dc_default_expr_exit=1 bc_default_digit_clamp=0 dc_default_digit_clamp=0;; GNU) bc_only=0 dc_only=0 coverage=0 debug=0 optimization="3" hist=1 hist_impl="internal" extra_math=1 generate_tests=$generate_tests install_manpages=1 nls=1 force=0 strip_bin=1 all_locales=0 library=0 fuzz=0 + ossfuzz=0 time_tests=0 vg=0 memcheck=0 clean=1 bc_default_banner=1 bc_default_sigint_reset=1 dc_default_sigint_reset=0 bc_default_tty_mode=1 dc_default_tty_mode=0 bc_default_prompt="" dc_default_prompt="" bc_default_expr_exit=1 dc_default_expr_exit=1 bc_default_digit_clamp=1 dc_default_digit_clamp=0;; GDH) - CFLAGS="-flto -Weverything -Wno-padded -Wno-unsafe-buffer-usage -Wno-poison-system-directories -Werror -pedantic -std=c11" + CFLAGS="-Weverything -Wno-padded -Wno-unsafe-buffer-usage -Wno-poison-system-directories" + CFLAGS="$CFLAGS -Wno-switch-default -Werror -pedantic -std=c11" bc_only=0 dc_only=0 coverage=0 debug=0 optimization="3" hist=1 hist_impl="internal" extra_math=1 generate_tests=1 install_manpages=1 nls=0 force=0 strip_bin=1 all_locales=0 library=0 fuzz=0 + ossfuzz=0 time_tests=0 vg=0 memcheck=0 clean=1 bc_default_banner=1 bc_default_sigint_reset=1 dc_default_sigint_reset=1 bc_default_tty_mode=1 dc_default_tty_mode=1 bc_default_prompt="" dc_default_prompt="" bc_default_expr_exit=0 dc_default_expr_exit=0 bc_default_digit_clamp=1 dc_default_digit_clamp=1;; DBG) - CFLAGS="-Weverything -Wno-padded -Wno-unsafe-buffer-usage -Wno-poison-system-directories -Werror -pedantic -std=c11" + CFLAGS="-Weverything -Wno-padded -Wno-unsafe-buffer-usage -Wno-poison-system-directories" + CFLAGS="$CFLAGS -Wno-switch-default -Werror -pedantic -std=c11" bc_only=0 dc_only=0 coverage=0 debug=1 optimization="0" hist=1 hist_impl="internal" extra_math=1 generate_tests=1 install_manpages=1 nls=1 force=0 strip_bin=1 all_locales=0 library=0 fuzz=0 + ossfuzz=0 time_tests=0 vg=0 memcheck=1 clean=1 bc_default_banner=1 bc_default_sigint_reset=1 dc_default_sigint_reset=1 bc_default_tty_mode=1 dc_default_tty_mode=1 bc_default_prompt="" dc_default_prompt="" bc_default_expr_exit=0 dc_default_expr_exit=0 bc_default_digit_clamp=1 dc_default_digit_clamp=1;; ?|'') usage "Invalid user build: \"$_predefined_build_type\". Accepted types are BSD, GNU, GDH, DBG.";; esac } # Generates a list of script test targets that will be used as prerequisites for # other targets. # # @param name The name of the calculator to generate script test targets for. gen_script_test_targets() { _gen_script_test_targets_name="$1" shift _gen_script_test_targets_tests=$(cat "$scriptdir/tests/$_gen_script_test_targets_name/scripts/all.txt") for _gen_script_test_targets_f in $_gen_script_test_targets_tests; do _gen_script_test_targets_b=$(basename "$_gen_script_test_targets_f" \ ".$_gen_script_test_targets_name") printf ' test_%s_script_%s' "$_gen_script_test_targets_name" \ "$_gen_script_test_targets_b" done printf '\n' } # This is a list of defaults, but it is also the list of possible options for # users to change. # # The development options are: force (force options even if they fail), valgrind # (build in a way suitable for valgrind testing), memcheck (same as valgrind), # and fuzzing (build in a way suitable for fuzzing). bc_only=0 dc_only=0 coverage=0 karatsuba_len=32 debug=0 hist=1 hist_impl="internal" extra_math=1 optimization="" generate_tests=1 install_manpages=1 nls=1 force=0 strip_bin=1 all_locales=0 library=0 fuzz=0 +ossfuzz=0 time_tests=0 vg=0 memcheck=0 clean=1 problematic_tests=1 # The empty strings are because they depend on TTY mode. If they are directly # set, though, they will be integers. We test for empty strings later. bc_default_banner=0 bc_default_sigint_reset=1 dc_default_sigint_reset=1 bc_default_tty_mode=1 dc_default_tty_mode=0 bc_default_prompt="" dc_default_prompt="" bc_default_expr_exit=1 dc_default_expr_exit=1 bc_default_digit_clamp=0 dc_default_digit_clamp=0 # getopts is a POSIX utility, but it cannot handle long options. Thus, the # handling of long options is done by hand, and that's the reason that short and # long options cannot be mixed. -while getopts "abBcdDeEfgGhHik:lMmNO:p:PrS:s:tTvz-" opt; do +while getopts "abBcdDeEfgGhHik:lMmNO:p:PrS:s:tTvzZ-" opt; do case "$opt" in a) library=1 ;; b) bc_only=1 ;; B) dc_only=1 ;; c) coverage=1 ;; C) clean=0 ;; d) dc_only=1 ;; D) bc_only=1 ;; e) hist_impl="editline" ;; E) extra_math=0 ;; f) force=1 ;; g) debug=1 ;; G) generate_tests=0 ;; h) usage ;; H) hist=0 ;; i) hist_impl="internal" ;; k) karatsuba_len="$OPTARG" ;; l) all_locales=1 ;; m) memcheck=1 ;; M) install_manpages=0 ;; N) nls=0 ;; O) optimization="$OPTARG" ;; p) predefined_build "$OPTARG" ;; P) problematic_tests=0 ;; r) hist_impl="readline" ;; S) set_default 0 "$OPTARG" ;; s) set_default 1 "$OPTARG" ;; t) time_tests=1 ;; T) strip_bin=0 ;; v) vg=1 ;; z) fuzz=1 ;; + Z) ossfuzz=1 ;; -) arg="$1" arg="${arg#--}" LONG_OPTARG="${arg#*=}" case $arg in help) usage ;; library) library=1 ;; bc-only) bc_only=1 ;; dc-only) dc_only=1 ;; coverage) coverage=1 ;; debug) debug=1 ;; force) force=1 ;; prefix=?*) PREFIX="$LONG_OPTARG" ;; prefix) if [ "$#" -lt 2 ]; then usage "No argument given for '--$arg' option" fi PREFIX="$2" shift ;; bindir=?*) BINDIR="$LONG_OPTARG" ;; bindir) if [ "$#" -lt 2 ]; then usage "No argument given for '--$arg' option" fi BINDIR="$2" shift ;; includedir=?*) INCLUDEDIR="$LONG_OPTARG" ;; includedir) if [ "$#" -lt 2 ]; then usage "No argument given for '--$arg' option" fi INCLUDEDIR="$2" shift ;; libdir=?*) LIBDIR="$LONG_OPTARG" ;; libdir) if [ "$#" -lt 2 ]; then usage "No argument given for '--$arg' option" fi LIBDIR="$2" shift ;; datarootdir=?*) DATAROOTDIR="$LONG_OPTARG" ;; datarootdir) if [ "$#" -lt 2 ]; then usage "No argument given for '--$arg' option" fi DATAROOTDIR="$2" shift ;; datadir=?*) DATADIR="$LONG_OPTARG" ;; datadir) if [ "$#" -lt 2 ]; then usage "No argument given for '--$arg' option" fi DATADIR="$2" shift ;; mandir=?*) MANDIR="$LONG_OPTARG" ;; mandir) if [ "$#" -lt 2 ]; then usage "No argument given for '--$arg' option" fi MANDIR="$2" shift ;; man1dir=?*) MAN1DIR="$LONG_OPTARG" ;; man1dir) if [ "$#" -lt 2 ]; then usage "No argument given for '--$arg' option" fi MAN1DIR="$2" shift ;; man3dir=?*) MAN3DIR="$LONG_OPTARG" ;; man3dir) if [ "$#" -lt 2 ]; then usage "No argument given for '--$arg' option" fi MAN3DIR="$2" shift ;; karatsuba-len=?*) karatsuba_len="$LONG_OPTARG" ;; karatsuba-len) if [ "$#" -lt 2 ]; then usage "No argument given for '--$arg' option" fi karatsuba_len="$1" shift ;; opt=?*) optimization="$LONG_OPTARG" ;; opt) if [ "$#" -lt 2 ]; then usage "No argument given for '--$arg' option" fi optimization="$1" shift ;; set-default-on=?*) set_default 1 "$LONG_OPTARG" ;; set-default-on) if [ "$#" -lt 2 ]; then usage "No argument given for '--$arg' option" fi set_default 1 "$1" shift ;; set-default-off=?*) set_default 0 "$LONG_OPTARG" ;; set-default-off) if [ "$#" -lt 2 ]; then usage "No argument given for '--$arg' option" fi set_default 0 "$1" shift ;; predefined-build-type=?*) predefined_build "$LONG_OPTARG" ;; predefined-build-type) if [ "$#" -lt 2 ]; then usage "No argument given for '--$arg' option" fi predefined_build "$1" shift ;; disable-bc) dc_only=1 ;; disable-dc) bc_only=1 ;; disable-clean) clean=0 ;; disable-extra-math) extra_math=0 ;; disable-generated-tests) generate_tests=0 ;; disable-history) hist=0 ;; disable-man-pages) install_manpages=0 ;; disable-nls) nls=0 ;; disable-strip) strip_bin=0 ;; disable-problematic-tests) problematic_tests=0 ;; enable-editline) hist_impl="editline" ;; enable-readline) hist_impl="readline" ;; enable-internal-history) hist_impl="internal" ;; enable-test-timing) time_tests=1 ;; enable-valgrind) vg=1 ;; enable-fuzz-mode) fuzz=1 ;; + enable-ossfuzz-mode) ossfuzz=1 ;; enable-memcheck) memcheck=1 ;; install-all-locales) all_locales=1 ;; help* | bc-only* | dc-only* | coverage* | debug*) usage "No arg allowed for --$arg option" ;; disable-bc* | disable-dc* | disable-clean*) usage "No arg allowed for --$arg option" ;; disable-extra-math*) usage "No arg allowed for --$arg option" ;; disable-generated-tests* | disable-history*) usage "No arg allowed for --$arg option" ;; disable-man-pages* | disable-nls* | disable-strip*) usage "No arg allowed for --$arg option" ;; disable-problematic-tests*) usage "No arg allowed for --$arg option" ;; enable-fuzz-mode* | enable-test-timing* | enable-valgrind*) usage "No arg allowed for --$arg option" ;; enable-memcheck* | install-all-locales*) usage "No arg allowed for --$arg option" ;; enable-editline* | enable-readline*) usage "No arg allowed for --$arg option" ;; enable-internal-history*) usage "No arg allowed for --$arg option" ;; '') break ;; # "--" terminates argument processing * ) usage "Invalid option $LONG_OPTARG" ;; esac shift OPTIND=1 ;; ?) usage "Invalid option: $opt" ;; esac done # Sometimes, developers don't want configure.sh to do a config clean. But # sometimes they do. if [ "$clean" -ne 0 ]; then if [ -f ./Makefile ]; then make clean_config > /dev/null fi fi # It is an error to say that bc only should be built and likewise for dc. if [ "$bc_only" -eq 1 ] && [ "$dc_only" -eq 1 ]; then usage "Can only specify one of -b(-D) or -d(-B)" fi # The library is mutually exclusive to the calculators, so it's an error to # give an option for either of them. if [ "$library" -ne 0 ]; then if [ "$bc_only" -eq 1 ] || [ "$dc_only" -eq 1 ]; then usage "Must not specify -b(-D) or -d(-B) when building the library" fi fi # KARATSUBA_LEN must be an integer and must be 16 or greater. case $karatsuba_len in (*[!0-9]*|'') usage "KARATSUBA_LEN is not a number" ;; (*) ;; esac if [ "$karatsuba_len" -lt 16 ]; then usage "KARATSUBA_LEN is less than 16" fi set -e if [ -z "${LONG_BIT+set}" ]; then LONG_BIT_DEFINE="" elif [ "$LONG_BIT" -lt 32 ]; then usage "LONG_BIT is less than 32" else LONG_BIT_DEFINE="-DBC_LONG_BIT=$LONG_BIT" fi if [ -z "$CC" ]; then CC="c99" else # I had users complain that, if they gave CFLAGS as part of CC, which # autotools allows in its braindead way, the build would fail with an error. # I don't like adjusting for autotools, but oh well. These lines puts the # stuff after the first space into CFLAGS. ccbase=$(basename "$CC") suffix=" *" prefix="* " if [ "${ccbase%%$suffix}" != "$ccbase" ]; then ccflags="${ccbase#$prefix}" cc="${ccbase%%$suffix}" ccdir=$(dirname "$CC") if [ "$ccdir" = "." ] && [ "${CC#.}" = "$CC" ]; then ccdir="" else ccdir="$ccdir/" fi CC="${ccdir}${cc}" CFLAGS="$CFLAGS $ccflags" fi fi if [ -z "$HOSTCC" ] && [ -z "$HOST_CC" ]; then HOSTCC="$CC" elif [ -z "$HOSTCC" ]; then HOSTCC="$HOST_CC" fi if [ "$HOSTCC" != "$CC" ]; then # Like above, this splits HOSTCC and HOSTCFLAGS. ccbase=$(basename "$HOSTCC") suffix=" *" prefix="* " if [ "${ccbase%%$suffix}" != "$ccbase" ]; then ccflags="${ccbase#$prefix}" cc="${ccbase%%$suffix}" ccdir=$(dirname "$HOSTCC") if [ "$ccdir" = "." ] && [ "${HOSTCC#.}" = "$HOSTCC" ]; then ccdir="" else ccdir="$ccdir/" fi HOSTCC="${ccdir}${cc}" HOSTCFLAGS="$HOSTCFLAGS $ccflags" fi fi if [ -z "${HOSTCFLAGS+set}" ] && [ -z "${HOST_CFLAGS+set}" ]; then HOSTCFLAGS="$CFLAGS" elif [ -z "${HOSTCFLAGS+set}" ]; then HOSTCFLAGS="$HOST_CFLAGS" fi # Store these for the cross compilation detection later. OLDCFLAGS="$CFLAGS" OLDHOSTCFLAGS="$HOSTCFLAGS" link="@printf 'No link necessary\\\\n'" main_exec="BC" executable="BC_EXEC" tests="test_bc timeconst test_dc" bc_test="@export BC_TEST_OUTPUT_DIR=\"$builddir/tests\"; \$(TESTSDIR)/all.sh bc $extra_math 1 $generate_tests $problematic_tests $time_tests \$(BC_EXEC)" bc_test_np="@export BC_TEST_OUTPUT_DIR=\"$builddir/tests\"; \$(TESTSDIR)/all.sh -n bc $extra_math 1 $generate_tests $problematic_tests $time_tests \$(BC_EXEC)" dc_test="@export BC_TEST_OUTPUT_DIR=\"$builddir/tests\"; \$(TESTSDIR)/all.sh dc $extra_math 1 $generate_tests $problematic_tests $time_tests \$(DC_EXEC)" dc_test_np="@export BC_TEST_OUTPUT_DIR=\"$builddir/tests\"; \$(TESTSDIR)/all.sh -n dc $extra_math 1 $generate_tests $problematic_tests $time_tests \$(DC_EXEC)" timeconst="@export BC_TEST_OUTPUT_DIR=\"$builddir/tests\"; \$(TESTSDIR)/bc/timeconst.sh \$(TESTSDIR)/bc/scripts/timeconst.bc \$(BC_EXEC)" # In order to have cleanup at exit, we need to be in # debug mode, so don't run valgrind without that. if [ "$vg" -ne 0 ]; then debug=1 bc_test_exec='valgrind $(VALGRIND_ARGS) $(BC_EXEC)' dc_test_exec='valgrind $(VALGRIND_ARGS) $(DC_EXEC)' bcl_test_exec='valgrind $(VALGRIND_ARGS) $(BCL_TEST)' else bc_test_exec='$(BC_EXEC)' dc_test_exec='$(DC_EXEC)' bcl_test_exec='$(BCL_TEST)' fi test_bc_history_prereqs="test_bc_history_all" test_dc_history_prereqs="test_dc_history_all" karatsuba="@printf 'karatsuba cannot be run because one of bc or dc is not built\\\\n'" karatsuba_test="@printf 'karatsuba cannot be run because one of bc or dc is not built\\\\n'" bc_lib="\$(GEN_DIR)/lib.o" bc_help="\$(GEN_DIR)/bc_help.o" dc_help="\$(GEN_DIR)/dc_help.o" default_target_prereqs="\$(BIN) \$(OBJS)" default_target_cmd="\$(CC) \$(CFLAGS) \$(OBJS) \$(LDFLAGS) -o \$(EXEC)" default_target="\$(DC_EXEC)" second_target_prereqs="" second_target_cmd="$default_target_cmd" second_target="\$(BC_EXEC)" # This if/else if chain is for setting the defaults that change based on whether # the library is being built, bc only, dc only, or both calculators. if [ "$library" -ne 0 ]; then extra_math=1 nls=0 hist=0 bc=1 dc=1 default_target_prereqs="\$(BIN) \$(OBJ)" default_target_cmd="ar -r -cu \$(LIBBC) \$(OBJ)" default_target="\$(LIBBC)" tests="test_library" test_bc_history_prereqs=" test_bc_history_skip" test_dc_history_prereqs=" test_dc_history_skip" install_prereqs=" install_library" uninstall_prereqs=" uninstall_library" install_man_prereqs=" install_bcl_manpage" uninstall_man_prereqs=" uninstall_bcl_manpage" elif [ "$bc_only" -eq 1 ]; then bc=1 dc=0 dc_help="" executables="bc" dc_test="@printf 'No dc tests to run\\\\n'" dc_test_np="@printf 'No dc tests to run\\\\n'" test_dc_history_prereqs=" test_dc_history_skip" install_prereqs=" install_execs" install_man_prereqs=" install_bc_manpage" uninstall_prereqs=" uninstall_bc" uninstall_man_prereqs=" uninstall_bc_manpage" default_target="\$(BC_EXEC)" second_target="\$(DC_EXEC)" tests="test_bc timeconst" elif [ "$dc_only" -eq 1 ]; then bc=0 dc=1 bc_lib="" bc_help="" executables="dc" main_exec="DC" executable="DC_EXEC" bc_test="@printf 'No bc tests to run\\\\n'" bc_test_np="@printf 'No bc tests to run\\\\n'" test_bc_history_prereqs=" test_bc_history_skip" timeconst="@printf 'timeconst cannot be run because bc is not built\\\\n'" install_prereqs=" install_execs" install_man_prereqs=" install_dc_manpage" uninstall_prereqs=" uninstall_dc" uninstall_man_prereqs=" uninstall_dc_manpage" tests="test_dc" +elif [ "$ossfuzz" -eq 1 ]; then + + if [ "$bc_only" -ne 0 ] || [ "$dc_only" -ne 0 ]; then + usage "An OSS-Fuzz build must build both fuzzers." + fi + + bc=1 + dc=1 + + # Expressions *cannot* exit in an OSS-Fuzz build. + bc_default_expr_exit=0 + dc_default_expr_exit=0 + + executables="bc_fuzzer and dc_fuzzer" + + karatsuba="@\$(KARATSUBA) 30 0 \$(BC_EXEC)" + karatsuba_test="@\$(KARATSUBA) 1 100 \$(BC_EXEC)" + + if [ "$library" -eq 0 ]; then + install_prereqs=" install_execs" + install_man_prereqs=" install_bc_manpage install_dc_manpage" + uninstall_prereqs=" uninstall_bc uninstall_dc" + uninstall_man_prereqs=" uninstall_bc_manpage uninstall_dc_manpage" + else + install_prereqs=" install_library install_bcl_header" + install_man_prereqs=" install_bcl_manpage" + uninstall_prereqs=" uninstall_library uninstall_bcl_header" + uninstall_man_prereqs=" uninstall_bcl_manpage" + tests="test_library" + fi + + second_target_prereqs="src/bc_fuzzer.o $default_target_prereqs" + default_target_prereqs="\$(BC_FUZZER) src/dc_fuzzer.o $default_target_prereqs" + default_target_cmd="\$(CXX) \$(CFLAGS) src/dc_fuzzer.o \$(LIB_FUZZING_ENGINE) \$(OBJS) \$(LDFLAGS) -o \$(DC_FUZZER) \&\& ln -sf ./dc_fuzzer_c \$(DC_FUZZER_C)" + second_target_cmd="\$(CXX) \$(CFLAGS) src/bc_fuzzer.o \$(LIB_FUZZING_ENGINE) \$(OBJS) \$(LDFLAGS) -o \$(BC_FUZZER) \&\& ln -sf ./bc_fuzzer_c \$(BC_FUZZER_C)" + + default_target="\$(DC_FUZZER) \$(DC_FUZZER_C)" + second_target="\$(BC_FUZZER) \$(BC_FUZZER_C)" + else bc=1 dc=1 executables="bc and dc" karatsuba="@\$(KARATSUBA) 30 0 \$(BC_EXEC)" karatsuba_test="@\$(KARATSUBA) 1 100 \$(BC_EXEC)" if [ "$library" -eq 0 ]; then install_prereqs=" install_execs" install_man_prereqs=" install_bc_manpage install_dc_manpage" uninstall_prereqs=" uninstall_bc uninstall_dc" uninstall_man_prereqs=" uninstall_bc_manpage uninstall_dc_manpage" else install_prereqs=" install_library install_bcl_header" install_man_prereqs=" install_bcl_manpage" uninstall_prereqs=" uninstall_library uninstall_bcl_header" uninstall_man_prereqs=" uninstall_bcl_manpage" tests="test_library" fi second_target_prereqs="$default_target_prereqs" default_target_prereqs="$second_target" default_target_cmd="\$(LINK) \$(BIN) \$(EXEC_PREFIX)\$(DC)" fi +if [ "$fuzz" -ne 0 ] && [ "$ossfuzz" -ne 0 ]; then + usage "Fuzzing mode and OSS-Fuzz mode are mutually exclusive" +fi + # We need specific stuff for fuzzing. -if [ "$fuzz" -ne 0 ]; then +if [ "$fuzz" -ne 0 ] || [ "$ossfuzz" -ne 0 ]; then debug=1 hist=0 nls=0 optimization="3" fi # This sets some necessary things for debug mode. if [ "$debug" -eq 1 ]; then if [ -z "$CFLAGS" ] && [ -z "$optimization" ]; then CFLAGS="-O0" fi CFLAGS="-g $CFLAGS" else CPPFLAGS="-DNDEBUG $CPPFLAGS" fi # Set optimization CFLAGS. if [ -n "$optimization" ]; then CFLAGS="-O$optimization $CFLAGS" fi # Set test coverage defaults. if [ "$coverage" -eq 1 ]; then if [ "$bc_only" -eq 1 ] || [ "$dc_only" -eq 1 ]; then usage "Can only specify -c without -b or -d" fi CFLAGS="-fprofile-arcs -ftest-coverage -g -O0 $CFLAGS" CPPFLAGS="-DNDEBUG $CPPFLAGS" COVERAGE_OUTPUT="@gcov -pabcdf \$(GCDA) \$(BC_GCDA) \$(DC_GCDA) \$(HISTORY_GCDA) \$(RAND_GCDA)" COVERAGE_OUTPUT="$COVERAGE_OUTPUT;\$(RM) -f \$(GEN)*.gc*" COVERAGE_OUTPUT="$COVERAGE_OUTPUT;gcovr --exclude-unreachable-branches --exclude-throw-branches --html-details --output index.html" COVERAGE_PREREQS=" test coverage_output" else COVERAGE_OUTPUT="@printf 'Coverage not generated\\\\n'" COVERAGE_PREREQS="" fi - # Set some defaults. if [ -z "${DESTDIR+set}" ]; then destdir="" else destdir="DESTDIR = $DESTDIR" fi # defprefix is for a warning about locales later. if [ -z "${PREFIX+set}" ]; then PREFIX="/usr/local" defprefix=1 else defprefix=0 fi if [ -z "${BINDIR+set}" ]; then BINDIR="$PREFIX/bin" fi if [ -z "${INCLUDEDIR+set}" ]; then INCLUDEDIR="$PREFIX/include" fi if [ -z "${LIBDIR+set}" ]; then LIBDIR="$PREFIX/lib" fi if [ -z "${PC_PATH+set}" ]; then set +e command -v pkg-config > /dev/null err=$? set -e if [ "$err" -eq 0 ]; then PC_PATH=$(pkg-config --variable=pc_path pkg-config) PC_PATH="${PC_PATH%%:*}" else PC_PATH="" fi fi # Set a default for the DATAROOTDIR. This is done if either manpages will be # installed, or locales are enabled because that's probably where NLSPATH # points. if [ "$install_manpages" -ne 0 ] || [ "$nls" -ne 0 ]; then if [ -z "${DATAROOTDIR+set}" ]; then DATAROOTDIR="$PREFIX/share" fi fi # Set defaults for manpage environment variables. if [ "$install_manpages" -ne 0 ]; then if [ -z "${DATADIR+set}" ]; then DATADIR="$DATAROOTDIR" fi if [ -z "${MANDIR+set}" ]; then MANDIR="$DATADIR/man" fi if [ -z "${MAN1DIR+set}" ]; then MAN1DIR="$MANDIR/man1" fi if [ -z "${MAN3DIR+set}" ]; then MAN3DIR="$MANDIR/man3" fi else install_man_prereqs="" uninstall_man_prereqs="" fi # Here is where we test NLS (the locale system). This is done by trying to # compile src/vm.c, which has the relevant code. If it fails, then it is # disabled. if [ "$nls" -ne 0 ]; then set +e printf 'Testing NLS...\n' flags="-DBC_ENABLE_NLS=1 -DBC_ENABLED=$bc -DDC_ENABLED=$dc" flags="$flags -DBC_ENABLE_HISTORY=$hist -DBC_ENABLE_LIBRARY=0 -DBC_ENABLE_AFL=0" - flags="$flags -DBC_ENABLE_EXTRA_MATH=$extra_math -I$scriptdir/include/" - flags="$flags -D_POSIX_C_SOURCE=200809L -D_XOPEN_SOURCE=700" + flags="$flags -DBC_ENABLE_EXTRA_MATH=$extra_math -DBC_ENABLE_OSSFUZZ=0" + flags="$flags -I$scriptdir/include/ -D_POSIX_C_SOURCE=200809L -D_XOPEN_SOURCE=700" ccbase=$(basename "$CC") if [ "$ccbase" = "clang" ]; then flags="$flags -Wno-unreachable-code" fi - "$CC" $CPPFLAGS $CFLAGS $flags -c "$scriptdir/src/vm.c" -o "./vm.o" > /dev/null 2>&1 + "$CC" $CPPFLAGS $CFLAGS $flags -c "$scriptdir/src/vm.c" -E > /dev/null err="$?" rm -rf "./vm.o" - # If this errors, it is probably because of building on Windows, - # and NLS is not supported on Windows, so disable it. + # If this errors, it is probably because of building on Windows or musl, + # and NLS is not supported on Windows or musl, so disable it. if [ "$err" -ne 0 ]; then printf 'NLS does not work.\n' if [ $force -eq 0 ]; then printf 'Disabling NLS...\n\n' nls=0 else printf 'Forcing NLS...\n\n' fi else printf 'NLS works.\n\n' printf 'Testing gencat...\n' - gencat "./en_US.cat" "$scriptdir/locales/en_US.msg" > /dev/null 2>&1 + gencat "./en_US.cat" "$scriptdir/locales/en_US.msg" > /dev/null err="$?" rm -rf "./en_US.cat" if [ "$err" -ne 0 ]; then printf 'gencat does not work.\n' if [ $force -eq 0 ]; then printf 'Disabling NLS...\n\n' nls=0 else printf 'Forcing NLS...\n\n' fi else printf 'gencat works.\n\n' # It turns out that POSIX locales are really terrible, and running # gencat on one machine is not guaranteed to make those cat files # portable to another machine, so we had better warn the user here. if [ "$HOSTCC" != "$CC" ] || [ "$OLDHOSTCFLAGS" != "$OLDCFLAGS" ]; then printf 'Cross-compile detected.\n\n' printf 'WARNING: Catalog files generated with gencat may not be portable\n' printf ' across different architectures.\n\n' fi if [ -z "$NLSPATH" ]; then NLSPATH="/usr/share/locale/%L/%N" fi install_locales_prereqs=" install_locales" uninstall_locales_prereqs=" uninstall_locales" fi fi set -e else install_locales_prereqs="" uninstall_locales_prereqs="" all_locales=0 fi if [ "$nls" -ne 0 ] && [ "$all_locales" -ne 0 ]; then install_locales="\$(LOCALE_INSTALL) -l \$(NLSPATH) \$(MAIN_EXEC) \$(DESTDIR)" else install_locales="\$(LOCALE_INSTALL) \$(NLSPATH) \$(MAIN_EXEC) \$(DESTDIR)" fi # Like the above tested locale support, this tests history. if [ "$hist" -eq 1 ]; then if [ "$hist_impl" = "editline" ]; then editline=1 readline=0 elif [ "$hist_impl" = "readline" ]; then editline=0 readline=1 else editline=0 readline=0 fi set +e printf 'Testing history...\n' flags="-DBC_ENABLE_HISTORY=1 -DBC_ENABLED=$bc -DDC_ENABLED=$dc" flags="$flags -DBC_ENABLE_NLS=$nls -DBC_ENABLE_LIBRARY=0 -DBC_ENABLE_AFL=0" flags="$flags -DBC_ENABLE_EDITLINE=$editline -DBC_ENABLE_READLINE=$readline" - flags="$flags -DBC_ENABLE_EXTRA_MATH=$extra_math -I$scriptdir/include/" - flags="$flags -D_POSIX_C_SOURCE=200809L -D_XOPEN_SOURCE=700" + flags="$flags -DBC_ENABLE_EXTRA_MATH=$extra_math -DBC_ENABLE_OSSFUZZ=0" + flags="$flags -I$scriptdir/include/ -D_POSIX_C_SOURCE=200809L -D_XOPEN_SOURCE=700" - "$CC" $CPPFLAGS $CFLAGS $flags -c "$scriptdir/src/history.c" -o "./history.o" > /dev/null 2>&1 + "$CC" $CPPFLAGS $CFLAGS $flags -c "$scriptdir/src/history.c" -E > /dev/null err="$?" rm -rf "./history.o" # If this errors, it is probably because of building on Windows, # and history is not supported on Windows, so disable it. if [ "$err" -ne 0 ]; then printf 'History does not work.\n' if [ $force -eq 0 ]; then printf 'Disabling history...\n\n' hist=0 else printf 'Forcing history...\n\n' fi else printf 'History works.\n\n' fi set -e else editline=0 readline=0 fi # We have to disable the history tests if it is disabled or valgrind is on. Or # if we are using editline or readline. if [ "$hist" -eq 0 ] || [ "$vg" -ne 0 ]; then test_bc_history_prereqs=" test_bc_history_skip" test_dc_history_prereqs=" test_dc_history_skip" history_tests="@printf 'Skipping history tests...\\\\n'" CFLAGS="$CFLAGS -DBC_ENABLE_EDITLINE=0 -DBC_ENABLE_READLINE=0" else if [ "$editline" -eq 0 ] && [ "$readline" -eq 0 ]; then history_tests="@printf '\$(TEST_STARS)\\\\n\\\\nRunning history tests...\\\\n\\\\n'" history_tests="$history_tests \&\& \$(TESTSDIR)/history.sh bc -a \&\&" history_tests="$history_tests \$(TESTSDIR)/history.sh dc -a \&\& printf" history_tests="$history_tests '\\\\nAll history tests passed.\\\\n\\\\n\$(TEST_STARS)\\\\n'" else test_bc_history_prereqs=" test_bc_history_skip" test_dc_history_prereqs=" test_dc_history_skip" history_tests="@printf 'Skipping history tests...\\\\n'" fi # We are also setting the CFLAGS and LDFLAGS here. if [ "$editline" -ne 0 ]; then LDFLAGS="$LDFLAGS -ledit" CPPFLAGS="$CPPFLAGS -DBC_ENABLE_EDITLINE=1 -DBC_ENABLE_READLINE=0" elif [ "$readline" -ne 0 ]; then LDFLAGS="$LDFLAGS -lreadline" CPPFLAGS="$CPPFLAGS -DBC_ENABLE_EDITLINE=0 -DBC_ENABLE_READLINE=1" else CPPFLAGS="$CPPFLAGS -DBC_ENABLE_EDITLINE=0 -DBC_ENABLE_READLINE=0" fi fi # Test FreeBSD. This is not in an if statement because regardless of whatever # the user says, we need to know if we are on FreeBSD. If we are, we cannot set # _POSIX_C_SOURCE and _XOPEN_SOURCE. The FreeBSD headers turn *off* stuff when # that is done. set +e printf 'Testing for FreeBSD...\n' flags="-DBC_TEST_FREEBSD -DBC_ENABLE_AFL=0" -"$CC" $CPPFLAGS $CFLAGS $flags "-I$scriptdir/include" -E "$scriptdir/src/vm.c" > /dev/null 2>&1 +"$CC" $CPPFLAGS $CFLAGS $flags "-I$scriptdir/include" -E "$scriptdir/scripts/os.c" > /dev/null err="$?" if [ "$err" -ne 0 ]; then printf 'On FreeBSD. Not using _POSIX_C_SOURCE and _XOPEN_SOURCE.\n\n' else printf 'Not on FreeBSD. Using _POSIX_C_SOURCE and _XOPEN_SOURCE.\n\n' CPPFLAGS="$CPPFLAGS -D_POSIX_C_SOURCE=200809L -D_XOPEN_SOURCE=700" fi # Test macOS. This is not in an if statement because regardless of whatever the # user says, we need to know if we are on macOS. If we are, we have to set # _DARWIN_C_SOURCE. printf 'Testing for macOS...\n' flags="-DBC_TEST_APPLE -DBC_ENABLE_AFL=0" -"$CC" $CPPFLAGS $CFLAGS $flags "-I$scriptdir/include" -E "$scriptdir/src/vm.c" > /dev/null 2>&1 +"$CC" $CPPFLAGS $CFLAGS $flags "-I$scriptdir/include" -E "$scriptdir/scripts/os.c" > /dev/null err="$?" if [ "$err" -ne 0 ]; then printf 'On macOS. Using _DARWIN_C_SOURCE.\n\n' apple="-D_DARWIN_C_SOURCE" else printf 'Not on macOS.\n\n' apple="" fi # We can't use the linker's strip flag on macOS. if [ "$debug" -eq 0 ] && [ "$apple" = "" ] && [ "$strip_bin" -ne 0 ]; then LDFLAGS="-s $LDFLAGS" fi # Test OpenBSD. This is not in an if statement because regardless of whatever # the user says, we need to know if we are on OpenBSD to activate _BSD_SOURCE. # No, I cannot `#define _BSD_SOURCE` in a header because OpenBSD's patched GCC # and Clang complain that that is only allowed for system headers. Sigh....So we # have to check at configure time and set it on the compiler command-line. And # we have to set it because we also set _POSIX_C_SOURCE, which OpenBSD headers # detect, and when they detect it, they turn off _BSD_SOURCE unless it is # specifically requested. printf 'Testing for OpenBSD...\n' flags="-DBC_TEST_OPENBSD -DBC_ENABLE_AFL=0" -"$CC" $CPPFLAGS $CFLAGS $flags "-I$scriptdir/include" -E "$scriptdir/src/vm.c" > /dev/null 2>&1 +"$CC" $CPPFLAGS $CFLAGS $flags "-I$scriptdir/include" -E "$scriptdir/scripts/os.c" > /dev/null err="$?" if [ "$err" -ne 0 ]; then printf 'On OpenBSD. Using _BSD_SOURCE.\n\n' bsd="-D_BSD_SOURCE" # Readline errors on OpenBSD, for some weird reason. if [ "$readline" -ne 0 ]; then usage "Cannot use readline on OpenBSD" fi else printf 'Not on OpenBSD.\n\n' bsd="" fi set -e if [ "$library" -eq 1 ]; then bc_lib="" fi if [ "$extra_math" -eq 1 ] && [ "$bc" -ne 0 ] && [ "$library" -eq 0 ]; then BC_LIB2_O="\$(GEN_DIR)/lib2.o" else BC_LIB2_O="" fi GEN_DIR="$scriptdir/gen" # These lines set the appropriate targets based on whether `gen/strgen.c` or # `gen/strgen.sh` is used. GEN="strgen" -GEN_EXEC_TARGET="\$(HOSTCC) -DBC_ENABLE_AFL=0 -I$scriptdir/include/ \$(HOSTCFLAGS) -o \$(GEN_EXEC) \$(GEN_C)" +GEN_EXEC_TARGET="\$(HOSTCC) -DBC_ENABLE_AFL=0 -DBC_ENABLE_OSSFUZZ=0 -I$scriptdir/include/ \$(HOSTCFLAGS) -o \$(GEN_EXEC) \$(GEN_C)" CLEAN_PREREQS=" clean_gen clean_coverage" if [ -z "${GEN_HOST+set}" ]; then GEN_HOST=1 else if [ "$GEN_HOST" -eq 0 ]; then GEN="strgen.sh" GEN_EXEC_TARGET="@printf 'Do not need to build gen/strgen.c\\\\n'" CLEAN_PREREQS=" clean_coverage" fi fi +# The fuzzer files are always unneeded because they'll be built separately. manpage_args="" -unneeded="" +unneeded="bc_fuzzer.c dc_fuzzer.c" headers="\$(HEADERS)" # This series of if statements figure out what source files are *not* needed. if [ "$extra_math" -eq 0 ]; then exclude_extra_math=1 manpage_args="E" unneeded="$unneeded rand.c" else exclude_extra_math=0 headers="$headers \$(EXTRA_MATH_HEADERS)" fi # All of these next if statements set the build type and mark certain source # files as unneeded so that they won't have targets generated for them. if [ "$hist" -eq 0 ]; then manpage_args="${manpage_args}H" unneeded="$unneeded history.c" else headers="$headers \$(HISTORY_HEADERS)" fi if [ "$nls" -eq 0 ]; then manpage_args="${manpage_args}N" fi if [ "$bc" -eq 0 ]; then unneeded="$unneeded bc.c bc_lex.c bc_parse.c" else headers="$headers \$(BC_HEADERS)" fi if [ "$dc" -eq 0 ]; then unneeded="$unneeded dc.c dc_lex.c dc_parse.c" else headers="$headers \$(DC_HEADERS)" fi # This convoluted mess does pull the version out. If you change the format of # include/version.h, you may have to change this line. version=$(cat "$scriptdir/include/version.h" | grep "VERSION " - | awk '{ print $3 }' -) if [ "$library" -ne 0 ]; then unneeded="$unneeded args.c opt.c read.c file.c main.c" unneeded="$unneeded lang.c lex.c parse.c program.c" unneeded="$unneeded bc.c bc_lex.c bc_parse.c" unneeded="$unneeded dc.c dc_lex.c dc_parse.c" headers="$headers \$(LIBRARY_HEADERS)" if [ "$PC_PATH" != "" ]; then contents=$(cat "$scriptdir/bcl.pc.in") contents=$(replace "$contents" "INCLUDEDIR" "$INCLUDEDIR") contents=$(replace "$contents" "LIBDIR" "$LIBDIR") contents=$(replace "$contents" "VERSION" "$version") printf '%s\n' "$contents" > "$scriptdir/bcl.pc" pkg_config_install="\$(SAFE_INSTALL) \$(PC_INSTALL_ARGS) \"\$(BCL_PC)\" \"\$(DESTDIR)\$(PC_PATH)/\$(BCL_PC)\"" pkg_config_uninstall="\$(RM) -f \"\$(DESTDIR)\$(PC_PATH)/\$(BCL_PC)\"" else pkg_config_install="" pkg_config_uninstall="" fi +elif [ "$ossfuzz" -ne 0 ]; then + + unneeded="$unneeded library.c main.c" + + PC_PATH="" + pkg_config_install="" + pkg_config_uninstall="" + else unneeded="$unneeded library.c" PC_PATH="" pkg_config_install="" pkg_config_uninstall="" fi -# library.c is not needed under normal circumstances. +# library.c, bc_fuzzer.c, and dc_fuzzer.c are not needed under normal +# circumstances. if [ "$unneeded" = "" ]; then - unneeded="library.c" + unneeded="library.c bc_fuzzer.c dc_fuzzer.c" fi # This sets the appropriate manpage for a full build. if [ "$manpage_args" = "" ]; then manpage_args="A" fi -if [ "$vg" -ne 0 ]; then +if [ "$vg" -ne 0 ] || [ "$ossfuzz" -ne 0 ]; then memcheck=1 fi if [ "$bc_default_prompt" = "" ]; then bc_default_prompt="$bc_default_tty_mode" fi if [ "$dc_default_prompt" = "" ]; then dc_default_prompt="$dc_default_tty_mode" fi # Generate the test targets and prerequisites. bc_tests=$(gen_std_test_targets bc) bc_script_tests=$(gen_script_test_targets bc) bc_err_tests=$(gen_err_test_targets bc) dc_tests=$(gen_std_test_targets dc) dc_script_tests=$(gen_script_test_targets dc) dc_err_tests=$(gen_err_test_targets dc) printf 'unneeded: %s\n' "$unneeded" # Print out the values; this is for debugging. printf 'Version: %s\n' "$version" if [ "$bc" -ne 0 ]; then printf 'Building bc\n' else printf 'Not building bc\n' fi if [ "$dc" -ne 0 ]; then printf 'Building dc\n' else printf 'Not building dc\n' fi printf '\n' printf 'BC_ENABLE_LIBRARY=%s\n\n' "$library" printf 'BC_ENABLE_HISTORY=%s\n' "$hist" printf 'BC_ENABLE_EXTRA_MATH=%s\n' "$extra_math" printf 'BC_ENABLE_NLS=%s\n\n' "$nls" printf 'BC_ENABLE_AFL=%s\n' "$fuzz" printf '\n' printf 'BC_NUM_KARATSUBA_LEN=%s\n' "$karatsuba_len" printf '\n' printf 'CC=%s\n' "$CC" printf 'CFLAGS=%s\n' "$CFLAGS" printf 'HOSTCC=%s\n' "$HOSTCC" printf 'HOSTCFLAGS=%s\n' "$HOSTCFLAGS" printf 'CPPFLAGS=%s\n' "$CPPFLAGS" printf 'LDFLAGS=%s\n' "$LDFLAGS" printf 'PREFIX=%s\n' "$PREFIX" printf 'BINDIR=%s\n' "$BINDIR" printf 'INCLUDEDIR=%s\n' "$INCLUDEDIR" printf 'LIBDIR=%s\n' "$LIBDIR" printf 'DATAROOTDIR=%s\n' "$DATAROOTDIR" printf 'DATADIR=%s\n' "$DATADIR" printf 'MANDIR=%s\n' "$MANDIR" printf 'MAN1DIR=%s\n' "$MAN1DIR" printf 'MAN3DIR=%s\n' "$MAN3DIR" printf 'NLSPATH=%s\n' "$NLSPATH" printf 'PC_PATH=%s\n' "$PC_PATH" printf 'EXECSUFFIX=%s\n' "$EXECSUFFIX" printf 'EXECPREFIX=%s\n' "$EXECPREFIX" printf 'DESTDIR=%s\n' "$DESTDIR" printf 'LONG_BIT=%s\n' "$LONG_BIT" printf 'GEN_HOST=%s\n' "$GEN_HOST" printf 'GEN_EMU=%s\n' "$GEN_EMU" printf '\n' printf 'Setting Defaults\n' printf '================\n' printf 'bc.banner=%s\n' "$bc_default_banner" printf 'bc.sigint_reset=%s\n' "$bc_default_sigint_reset" printf 'dc.sigint_reset=%s\n' "$dc_default_sigint_reset" printf 'bc.tty_mode=%s\n' "$bc_default_tty_mode" printf 'dc.tty_mode=%s\n' "$dc_default_tty_mode" printf 'bc.prompt=%s\n' "$bc_default_prompt" printf 'dc.prompt=%s\n' "$dc_default_prompt" printf 'bc.expr_exit=%s\n' "$bc_default_expr_exit" printf 'dc.expr_exit=%s\n' "$dc_default_expr_exit" printf 'bc.digit_clamp=%s\n' "$bc_default_digit_clamp" printf 'dc.digit_clamp=%s\n' "$dc_default_digit_clamp" # This code outputs a warning. The warning is to not surprise users when locales # are installed outside of the prefix. This warning is suppressed when the # default prefix is used, as well, so as not to panic users just installing by # hand. I believe this will be okay because NLSPATH is usually in /usr and the # default prefix is /usr/local, so they'll be close that way. if [ "$nls" -ne 0 ] && [ "${NLSPATH#$PREFIX}" = "${NLSPATH}" ] && [ "$defprefix" -eq 0 ]; then printf '\n********************************************************************************\n\n' printf 'WARNING: Locales will *NOT* be installed in $PREFIX (%s).\n' "$PREFIX" printf '\n' printf ' This is because they *MUST* be installed at a fixed location to even\n' printf ' work, and that fixed location is $NLSPATH (%s).\n' "$NLSPATH" printf '\n' printf ' This location is *outside* of $PREFIX. If you do not wish to install\n' printf ' locales outside of $PREFIX, you must disable NLS with the -N or the\n' printf ' --disable-nls options.\n' printf '\n' printf ' The author apologizes for the inconvenience, but the need to install\n' printf ' the locales at a fixed location is mandated by POSIX, and it is not\n' printf ' possible for the author to change that requirement.\n' printf '\n********************************************************************************\n' fi # This is where the real work begins. This is the point at which the Makefile.in # template is edited and output to the Makefile. contents=$(cat "$scriptdir/Makefile.in") needle="WARNING" replacement='*** WARNING: Autogenerated from Makefile.in. DO NOT MODIFY ***' contents=$(replace "$contents" "$needle" "$replacement") # The contents are edited to have the list of files to build. contents=$(gen_file_list "$contents" $unneeded) SRC_TARGETS="" # This line and loop generates the individual targets for source files. I used # to just use an implicit target, but that was found to be inadequate when I # added the library. src_files=$(find_src_files $unneeded) for f in $src_files; do o=$(replace_ext "$f" "c" "o") o=$(basename "$o") SRC_TARGETS=$(printf '%s\n\nsrc/%s: src %s %s\n\t$(CC) $(CFLAGS) -o src/%s -c %s\n' \ "$SRC_TARGETS" "$o" "$headers" "$f" "$o" "$f") done # Replace all the placeholders. contents=$(replace "$contents" "ROOTDIR" "$scriptdir") contents=$(replace "$contents" "BUILDDIR" "$builddir") contents=$(replace "$contents" "HEADERS" "$headers") contents=$(replace "$contents" "BC_ENABLED" "$bc") contents=$(replace "$contents" "DC_ENABLED" "$dc") contents=$(replace "$contents" "BC_ALL_TESTS" "$bc_test") contents=$(replace "$contents" "BC_ALL_TESTS_NP" "$bc_test_np") contents=$(replace "$contents" "BC_TESTS" "$bc_tests") contents=$(replace "$contents" "BC_SCRIPT_TESTS" "$bc_script_tests") contents=$(replace "$contents" "BC_ERROR_TESTS" "$bc_err_tests") contents=$(replace "$contents" "BC_TEST_EXEC" "$bc_test_exec") contents=$(replace "$contents" "TIMECONST_ALL_TESTS" "$timeconst") contents=$(replace "$contents" "DC_ALL_TESTS" "$dc_test") contents=$(replace "$contents" "DC_ALL_TESTS_NP" "$dc_test_np") contents=$(replace "$contents" "DC_TESTS" "$dc_tests") contents=$(replace "$contents" "DC_SCRIPT_TESTS" "$dc_script_tests") contents=$(replace "$contents" "DC_ERROR_TESTS" "$dc_err_tests") contents=$(replace "$contents" "DC_TEST_EXEC" "$dc_test_exec") contents=$(replace "$contents" "BCL_TEST_EXEC" "$bcl_test_exec") contents=$(replace "$contents" "BUILD_TYPE" "$manpage_args") contents=$(replace "$contents" "EXCLUDE_EXTRA_MATH" "$exclude_extra_math") contents=$(replace "$contents" "LIBRARY" "$library") contents=$(replace "$contents" "HISTORY" "$hist") contents=$(replace "$contents" "EXTRA_MATH" "$extra_math") contents=$(replace "$contents" "NLS" "$nls") contents=$(replace "$contents" "FUZZ" "$fuzz") +contents=$(replace "$contents" "OSSFUZZ" "$ossfuzz") contents=$(replace "$contents" "MEMCHECK" "$memcheck") +contents=$(replace "$contents" "LIB_FUZZING_ENGINE" "$LIB_FUZZING_ENGINE") contents=$(replace "$contents" "BC_LIB_O" "$bc_lib") contents=$(replace "$contents" "BC_HELP_O" "$bc_help") contents=$(replace "$contents" "DC_HELP_O" "$dc_help") contents=$(replace "$contents" "BC_LIB2_O" "$BC_LIB2_O") contents=$(replace "$contents" "KARATSUBA_LEN" "$karatsuba_len") contents=$(replace "$contents" "NLSPATH" "$NLSPATH") contents=$(replace "$contents" "DESTDIR" "$destdir") contents=$(replace "$contents" "EXECSUFFIX" "$EXECSUFFIX") contents=$(replace "$contents" "EXECPREFIX" "$EXECPREFIX") contents=$(replace "$contents" "BINDIR" "$BINDIR") contents=$(replace "$contents" "INCLUDEDIR" "$INCLUDEDIR") contents=$(replace "$contents" "LIBDIR" "$LIBDIR") contents=$(replace "$contents" "MAN1DIR" "$MAN1DIR") contents=$(replace "$contents" "MAN3DIR" "$MAN3DIR") contents=$(replace "$contents" "CFLAGS" "$CFLAGS") contents=$(replace "$contents" "HOSTCFLAGS" "$HOSTCFLAGS") contents=$(replace "$contents" "CPPFLAGS" "$CPPFLAGS") contents=$(replace "$contents" "LDFLAGS" "$LDFLAGS") contents=$(replace "$contents" "CC" "$CC") contents=$(replace "$contents" "HOSTCC" "$HOSTCC") contents=$(replace "$contents" "COVERAGE_OUTPUT" "$COVERAGE_OUTPUT") contents=$(replace "$contents" "COVERAGE_PREREQS" "$COVERAGE_PREREQS") contents=$(replace "$contents" "INSTALL_PREREQS" "$install_prereqs") contents=$(replace "$contents" "INSTALL_MAN_PREREQS" "$install_man_prereqs") contents=$(replace "$contents" "INSTALL_LOCALES" "$install_locales") contents=$(replace "$contents" "INSTALL_LOCALES_PREREQS" "$install_locales_prereqs") contents=$(replace "$contents" "UNINSTALL_MAN_PREREQS" "$uninstall_man_prereqs") contents=$(replace "$contents" "UNINSTALL_PREREQS" "$uninstall_prereqs") contents=$(replace "$contents" "UNINSTALL_LOCALES_PREREQS" "$uninstall_locales_prereqs") contents=$(replace "$contents" "PC_PATH" "$PC_PATH") contents=$(replace "$contents" "PKG_CONFIG_INSTALL" "$pkg_config_install") contents=$(replace "$contents" "PKG_CONFIG_UNINSTALL" "$pkg_config_uninstall") contents=$(replace "$contents" "DEFAULT_TARGET" "$default_target") contents=$(replace "$contents" "DEFAULT_TARGET_PREREQS" "$default_target_prereqs") contents=$(replace "$contents" "DEFAULT_TARGET_CMD" "$default_target_cmd") contents=$(replace "$contents" "SECOND_TARGET" "$second_target") contents=$(replace "$contents" "SECOND_TARGET_PREREQS" "$second_target_prereqs") contents=$(replace "$contents" "SECOND_TARGET_CMD" "$second_target_cmd") contents=$(replace "$contents" "ALL_PREREQ" "$ALL_PREREQ") contents=$(replace "$contents" "BC_EXEC_PREREQ" "$bc_exec_prereq") contents=$(replace "$contents" "BC_EXEC_CMD" "$bc_exec_cmd") contents=$(replace "$contents" "DC_EXEC_PREREQ" "$dc_exec_prereq") contents=$(replace "$contents" "DC_EXEC_CMD" "$dc_exec_cmd") contents=$(replace "$contents" "EXECUTABLES" "$executables") contents=$(replace "$contents" "MAIN_EXEC" "$main_exec") contents=$(replace "$contents" "EXEC" "$executable") contents=$(replace "$contents" "TESTS" "$tests") contents=$(replace "$contents" "BC_HISTORY_TEST_PREREQS" "$test_bc_history_prereqs") contents=$(replace "$contents" "DC_HISTORY_TEST_PREREQS" "$test_dc_history_prereqs") contents=$(replace "$contents" "HISTORY_TESTS" "$history_tests") contents=$(replace "$contents" "VG_BC_TEST" "$vg_bc_test") contents=$(replace "$contents" "VG_DC_TEST" "$vg_dc_test") contents=$(replace "$contents" "TIMECONST" "$timeconst") contents=$(replace "$contents" "KARATSUBA" "$karatsuba") contents=$(replace "$contents" "KARATSUBA_TEST" "$karatsuba_test") contents=$(replace "$contents" "LONG_BIT_DEFINE" "$LONG_BIT_DEFINE") contents=$(replace "$contents" "GEN_DIR" "$GEN_DIR") contents=$(replace "$contents" "GEN" "$GEN") contents=$(replace "$contents" "GEN_EXEC_TARGET" "$GEN_EXEC_TARGET") contents=$(replace "$contents" "CLEAN_PREREQS" "$CLEAN_PREREQS") contents=$(replace "$contents" "GEN_EMU" "$GEN_EMU") contents=$(replace "$contents" "BSD" "$bsd") contents=$(replace "$contents" "APPLE" "$apple") contents=$(replace "$contents" "BC_DEFAULT_BANNER" "$bc_default_banner") contents=$(replace "$contents" "BC_DEFAULT_SIGINT_RESET" "$bc_default_sigint_reset") contents=$(replace "$contents" "DC_DEFAULT_SIGINT_RESET" "$dc_default_sigint_reset") contents=$(replace "$contents" "BC_DEFAULT_TTY_MODE" "$bc_default_tty_mode") contents=$(replace "$contents" "DC_DEFAULT_TTY_MODE" "$dc_default_tty_mode") contents=$(replace "$contents" "BC_DEFAULT_PROMPT" "$bc_default_prompt") contents=$(replace "$contents" "DC_DEFAULT_PROMPT" "$dc_default_prompt") contents=$(replace "$contents" "BC_DEFAULT_EXPR_EXIT" "$bc_default_expr_exit") contents=$(replace "$contents" "DC_DEFAULT_EXPR_EXIT" "$dc_default_expr_exit") contents=$(replace "$contents" "BC_DEFAULT_DIGIT_CLAMP" "$bc_default_digit_clamp") contents=$(replace "$contents" "DC_DEFAULT_DIGIT_CLAMP" "$dc_default_digit_clamp") # Do the first print to the Makefile. printf '%s\n%s\n\n' "$contents" "$SRC_TARGETS" > "Makefile" # Generate the individual test targets. if [ "$bc" -ne 0 ]; then gen_std_tests bc "$extra_math" "$time_tests" $bc_test_exec gen_script_tests bc "$extra_math" "$generate_tests" "$time_tests" $bc_test_exec gen_err_tests bc $bc_test_exec fi if [ "$dc" -ne 0 ]; then gen_std_tests dc "$extra_math" "$time_tests" $dc_test_exec gen_script_tests dc "$extra_math" "$generate_tests" "$time_tests" $dc_test_exec gen_err_tests dc $dc_test_exec fi +if [ "$ossfuzz" -ne 0 ]; then + + printf 'bc_fuzzer_c: $(BC_FUZZER)\n\tln -sf $(BC_FUZZER) bc_fuzzer_c\n' >> Makefile + printf 'bc_fuzzer_C: $(BC_FUZZER)\n\tln -sf $(BC_FUZZER) bc_fuzzer_C\n' >> Makefile + printf 'dc_fuzzer_c: $(DC_FUZZER)\n\tln -sf $(DC_FUZZER) dc_fuzzer_c\n' >> Makefile + printf 'dc_fuzzer_C: $(DC_FUZZER)\n\tln -sf $(DC_FUZZER) dc_fuzzer_C\n' >> Makefile + +fi + # Copy the correct manuals to the expected places. mkdir -p manuals cp -f "$scriptdir/manuals/bc/$manpage_args.1.md" manuals/bc.1.md cp -f "$scriptdir/manuals/bc/$manpage_args.1" manuals/bc.1 cp -f "$scriptdir/manuals/dc/$manpage_args.1.md" manuals/dc.1.md cp -f "$scriptdir/manuals/dc/$manpage_args.1" manuals/dc.1 make clean > /dev/null diff --git a/include/args.h b/include/args.h index f1e9f007bddf..8f8f00be4630 100644 --- a/include/args.h +++ b/include/args.h @@ -1,84 +1,84 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * Definitions for processing command-line arguments. * */ #ifndef BC_ARGS_H #define BC_ARGS_H #include #include #include /** * Processes command-line arguments. * @param argc How many arguments there are. * @param argv The array of arguments. * @param exit_exprs True if bc/dc should exit when there are expressions, * false otherwise. * @param scale A pointer to return the scale that the arguments set, if * any. * @param ibase A pointer to return the ibase that the arguments set, if * any. * @param obase A pointer to return the obase that the arguments set, if * any. */ void -bc_args(int argc, char* argv[], bool exit_exprs, BcBigDig* scale, +bc_args(int argc, const char* argv[], bool exit_exprs, BcBigDig* scale, BcBigDig* ibase, BcBigDig* obase); #if BC_ENABLED #if DC_ENABLED /// Returns true if the banner should be quieted. #define BC_ARGS_SHOULD_BE_QUIET (BC_IS_DC || vm->exprs.len > 1) #else // DC_ENABLED /// Returns true if the banner should be quieted. #define BC_ARGS_SHOULD_BE_QUIET (vm->exprs.len > 1) #endif // DC_ENABLED #else // BC_ENABLED /// Returns true if the banner should be quieted. #define BC_ARGS_SHOULD_BE_QUIET (BC_IS_DC) #endif // BC_ENABLED // A reference to the list of long options. extern const BcOptLong bc_args_lopt[]; #endif // BC_ARGS_H diff --git a/include/bc.h b/include/bc.h index b25df09a174e..2213278be1da 100644 --- a/include/bc.h +++ b/include/bc.h @@ -1,473 +1,473 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * Definitions for bc only. * */ #ifndef BC_BC_H #define BC_BC_H #if BC_ENABLED #include #include #include #include #include /** * The main function for bc. It just sets variables and passes its arguments * through to @a bc_vm_boot(). * @return A status. */ BcStatus -bc_main(int argc, char* argv[]); +bc_main(int argc, const char* argv[]); // These are references to the help text, the library text, and the "filename" // for the library. extern const char bc_help[]; extern const char bc_lib[]; extern const char* bc_lib_name; // These are references to the second math library and its "filename." #if BC_ENABLE_EXTRA_MATH extern const char bc_lib2[]; extern const char* bc_lib2_name; #endif // BC_ENABLE_EXTRA_MATH /** * A struct containing information about a bc keyword. */ typedef struct BcLexKeyword { /// Holds the length of the keyword along with a bit that, if set, means the /// keyword is used in POSIX bc. uchar data; /// The keyword text. const char name[14]; } BcLexKeyword; /// Sets the most significant bit. Used for setting the POSIX bit in /// BcLexKeyword's data field. #define BC_LEX_CHAR_MSB(bit) ((bit) << (CHAR_BIT - 1)) /// Returns non-zero if the keyword is POSIX, zero otherwise. #define BC_LEX_KW_POSIX(kw) ((kw)->data & (BC_LEX_CHAR_MSB(1))) /// Returns the length of the keyword. #define BC_LEX_KW_LEN(kw) ((size_t) ((kw)->data & ~(BC_LEX_CHAR_MSB(1)))) /// A macro to easily build a keyword entry. See bc_lex_kws in src/data.c. #define BC_LEX_KW_ENTRY(a, b, c) \ { .data = ((b) & ~(BC_LEX_CHAR_MSB(1))) | BC_LEX_CHAR_MSB(c), .name = a } #if BC_ENABLE_EXTRA_MATH /// A macro for the number of keywords bc has. This has to be updated if any are /// added. This is for the redefined_kws field of the BcVm struct. #define BC_LEX_NKWS (37) #else // BC_ENABLE_EXTRA_MATH /// A macro for the number of keywords bc has. This has to be updated if any are /// added. This is for the redefined_kws field of the BcVm struct. #define BC_LEX_NKWS (33) #endif // BC_ENABLE_EXTRA_MATH // The array of keywords and its length. extern const BcLexKeyword bc_lex_kws[]; extern const size_t bc_lex_kws_len; /** * The @a BcLexNext function for bc. (See include/lex.h for a definition of * @a BcLexNext.) * @param l The lexer. */ void bc_lex_token(BcLex* l); // The following section is for flags needed when parsing bc code. These flags // are complicated, but necessary. Why you ask? Because bc's standard is awful. // // If you don't believe me, go read the bc Parsing section of the Development // manual (manuals/development.md). Then come back. // // In other words, these flags are the sign declaring, "Here be dragons." /** * This returns a pointer to the set of flags at the top of the flag stack. * @a p is expected to be a BcParse pointer. * @param p The parser. * @return A pointer to the top flag set. */ #define BC_PARSE_TOP_FLAG_PTR(p) ((uint16_t*) bc_vec_top(&(p)->flags)) /** * This returns the flag set at the top of the flag stack. @a p is expected to * be a BcParse pointer. * @param p The parser. * @return The top flag set. */ #define BC_PARSE_TOP_FLAG(p) (*(BC_PARSE_TOP_FLAG_PTR(p))) // After this point, all flag #defines are in sets of 2: one to define the flag, // and one to define a way to grab the flag from the flag set at the top of the // flag stack. All `p` arguments are pointers to a BcParse. // This flag is set if the parser has seen a left brace. #define BC_PARSE_FLAG_BRACE (UINTMAX_C(1) << 0) #define BC_PARSE_BRACE(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_BRACE) // This flag is set if the parser is parsing inside of the braces of a function // body. #define BC_PARSE_FLAG_FUNC_INNER (UINTMAX_C(1) << 1) #define BC_PARSE_FUNC_INNER(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_FUNC_INNER) // This flag is set if the parser is parsing a function. It is different from // the one above because it is set if it is parsing a function body *or* header, // not just if it's parsing a function body. #define BC_PARSE_FLAG_FUNC (UINTMAX_C(1) << 2) #define BC_PARSE_FUNC(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_FUNC) // This flag is set if the parser is expecting to parse a body, whether of a // function, an if statement, or a loop. #define BC_PARSE_FLAG_BODY (UINTMAX_C(1) << 3) #define BC_PARSE_BODY(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_BODY) // This flag is set if bc is parsing a loop. This is important because the break // and continue keywords are only valid inside of a loop. #define BC_PARSE_FLAG_LOOP (UINTMAX_C(1) << 4) #define BC_PARSE_LOOP(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_LOOP) // This flag is set if bc is parsing the body of a loop. It is different from // the one above the same way @a BC_PARSE_FLAG_FUNC_INNER is different from // @a BC_PARSE_FLAG_FUNC. #define BC_PARSE_FLAG_LOOP_INNER (UINTMAX_C(1) << 5) #define BC_PARSE_LOOP_INNER(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_LOOP_INNER) // This flag is set if bc is parsing an if statement. #define BC_PARSE_FLAG_IF (UINTMAX_C(1) << 6) #define BC_PARSE_IF(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_IF) // This flag is set if bc is parsing an else statement. This is important // because of "else if" constructions, among other things. #define BC_PARSE_FLAG_ELSE (UINTMAX_C(1) << 7) #define BC_PARSE_ELSE(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_ELSE) // This flag is set if bc just finished parsing an if statement and its body. // It tells the parser that it can probably expect an else statement next. This // flag is, thus, one of the most subtle. #define BC_PARSE_FLAG_IF_END (UINTMAX_C(1) << 8) #define BC_PARSE_IF_END(p) (BC_PARSE_TOP_FLAG(p) & BC_PARSE_FLAG_IF_END) /** * This returns true if bc is in a state where it should not execute any code * at all. * @param p The parser. * @return True if execution cannot proceed, false otherwise. */ #define BC_PARSE_NO_EXEC(p) ((p)->flags.len != 1 || BC_PARSE_TOP_FLAG(p) != 0) /** * This returns true if the token @a t is a statement delimiter, which is * either a newline or a semicolon. * @param t The token to check. * @return True if t is a statement delimiter token; false otherwise. */ #define BC_PARSE_DELIMITER(t) \ ((t) == BC_LEX_SCOLON || (t) == BC_LEX_NLINE || (t) == BC_LEX_EOF) /** * This is poorly named, but it basically returns whether or not the current * state is valid for the end of an else statement. * @param f The flag set to be checked. * @return True if the state is valid for the end of an else statement. */ #define BC_PARSE_BLOCK_STMT(f) \ ((f) & (BC_PARSE_FLAG_ELSE | BC_PARSE_FLAG_LOOP_INNER)) /** * This returns the value of the data for an operator with precedence @a p and * associativity @a l (true if left associative, false otherwise). This is used * to construct an array of operators, bc_parse_ops, in src/data.c. * @param p The precedence. * @param l True if the operator is left associative, false otherwise. * @return The data for the operator. */ #define BC_PARSE_OP(p, l) (((p) & ~(BC_LEX_CHAR_MSB(1))) | (BC_LEX_CHAR_MSB(l))) /** * Returns the operator data for the lex token @a t. * @param t The token to return operator data for. * @return The operator data for @a t. */ #define BC_PARSE_OP_DATA(t) bc_parse_ops[((t) - BC_LEX_OP_INC)] /** * Returns non-zero if operator @a op is left associative, zero otherwise. * @param op The operator to test for associativity. * @return Non-zero if the operator is left associative, zero otherwise. */ #define BC_PARSE_OP_LEFT(op) (BC_PARSE_OP_DATA(op) & BC_LEX_CHAR_MSB(1)) /** * Returns the precedence of operator @a op. Lower number means higher * precedence. * @param op The operator to return the precedence of. * @return The precedence of @a op. */ #define BC_PARSE_OP_PREC(op) (BC_PARSE_OP_DATA(op) & ~(BC_LEX_CHAR_MSB(1))) /** * A macro to easily define a series of bits for whether a lex token is an * expression token or not. It takes 8 expression bits, corresponding to the 8 * bits in a uint8_t. You can see this in use for bc_parse_exprs in src/data.c. * @param e1 The first bit. * @param e2 The second bit. * @param e3 The third bit. * @param e4 The fourth bit. * @param e5 The fifth bit. * @param e6 The sixth bit. * @param e7 The seventh bit. * @param e8 The eighth bit. * @return An expression entry for bc_parse_exprs[]. */ #define BC_PARSE_EXPR_ENTRY(e1, e2, e3, e4, e5, e6, e7, e8) \ ((UINTMAX_C(e1) << 7) | (UINTMAX_C(e2) << 6) | (UINTMAX_C(e3) << 5) | \ (UINTMAX_C(e4) << 4) | (UINTMAX_C(e5) << 3) | (UINTMAX_C(e6) << 2) | \ (UINTMAX_C(e7) << 1) | (UINTMAX_C(e8) << 0)) /** * Returns true if token @a i is a token that belongs in an expression. * @param i The token to test. * @return True if i is an expression token, false otherwise. */ #define BC_PARSE_EXPR(i) \ (bc_parse_exprs[(((i) & (uchar) ~(0x07)) >> 3)] & (1 << (7 - ((i) & 0x07)))) /** * Returns the operator (by lex token) that is at the top of the operator * stack. * @param p The parser. * @return The operator that is at the top of the operator stack, as a lex * token. */ #define BC_PARSE_TOP_OP(p) (*((BcLexType*) bc_vec_top(&(p)->ops))) /** * Returns true if bc has a "leaf" token. A "leaf" token is one that can stand * alone in an expression. For example, a number by itself can be an expression, * but a binary operator, while valid for an expression, cannot be alone in the * expression. It must have an expression to the left and right of itself. See * the documentation for @a bc_parse_expr_err() in src/bc_parse.c. * @param prev The previous token as an instruction. * @param bin_last True if that last operator was a binary operator, false * otherwise. * @param rparen True if the last operator was a right paren. * return True if the last token was a leaf token, false otherwise. */ #define BC_PARSE_LEAF(prev, bin_last, rparen) \ (!(bin_last) && ((rparen) || bc_parse_inst_isLeaf(prev))) /** * This returns true if the token @a t should be treated as though it's a * variable. This goes for actual variables, array elements, and globals. * @param t The token to test. * @return True if @a t should be treated as though it's a variable, false * otherwise. */ #if BC_ENABLE_EXTRA_MATH #define BC_PARSE_INST_VAR(t) \ ((t) >= BC_INST_VAR && (t) <= BC_INST_SEED && (t) != BC_INST_ARRAY) #else // BC_ENABLE_EXTRA_MATH #define BC_PARSE_INST_VAR(t) \ ((t) >= BC_INST_VAR && (t) <= BC_INST_SCALE && (t) != BC_INST_ARRAY) #endif // BC_ENABLE_EXTRA_MATH /** * Returns true if the previous token @a p (in the form of a bytecode * instruction) is a prefix operator. The fact that it is for bytecode * instructions is what makes it different from @a BC_PARSE_OP_PREFIX below. * @param p The previous token. * @return True if @a p is a prefix operator. */ #define BC_PARSE_PREV_PREFIX(p) ((p) >= BC_INST_NEG && (p) <= BC_INST_BOOL_NOT) /** * Returns true if token @a t is a prefix operator. * @param t The token to test. * @return True if @a t is a prefix operator, false otherwise. */ #define BC_PARSE_OP_PREFIX(t) ((t) == BC_LEX_OP_BOOL_NOT || (t) == BC_LEX_NEG) /** * We can calculate the conversion between tokens and bytecode instructions by * subtracting the position of the first operator in the lex enum and adding the * position of the first in the instruction enum. Note: This only works for * binary operators. * @param t The token to turn into an instruction. * @return The token as an instruction. */ #define BC_PARSE_TOKEN_INST(t) ((uchar) ((t) - BC_LEX_NEG + BC_INST_NEG)) /** * Returns true if the token is a bc keyword. * @param t The token to check. * @return True if @a t is a bc keyword, false otherwise. */ #define BC_PARSE_IS_KEYWORD(t) ((t) >= BC_LEX_KW_AUTO && (t) <= BC_LEX_KW_ELSE) /// A struct that holds data about what tokens should be expected next. There /// are a few instances of these, all named because they are used in specific /// cases. Basically, in certain situations, it's useful to use the same code, /// but have a list of valid tokens. /// /// Obviously, @a len is the number of tokens in the @a tokens array. If more /// than 4 is needed in the future, @a tokens will have to be changed. typedef struct BcParseNext { /// The number of tokens in the tokens array. uchar len; /// The tokens that can be expected next. uchar tokens[4]; } BcParseNext; /// A macro to construct an array literal of tokens from a parameter list. #define BC_PARSE_NEXT_TOKENS(...) .tokens = { __VA_ARGS__ } /// A macro to generate a BcParseNext literal from BcParseNext data. See /// src/data.c for examples. #define BC_PARSE_NEXT(a, ...) \ { .len = (uchar) (a), BC_PARSE_NEXT_TOKENS(__VA_ARGS__) } /// A status returned by @a bc_parse_expr_err(). It can either return success or /// an error indicating an empty expression. typedef enum BcParseStatus { BC_PARSE_STATUS_SUCCESS, BC_PARSE_STATUS_EMPTY_EXPR, } BcParseStatus; /** * The @a BcParseExpr function for bc. (See include/parse.h for a definition of * @a BcParseExpr.) * @param p The parser. * @param flags Flags that define the requirements that the parsed code must * meet or an error will result. See @a BcParseExpr for more info. */ void bc_parse_expr(BcParse* p, uint8_t flags); /** * The @a BcParseParse function for bc. (See include/parse.h for a definition of * @a BcParseParse.) * @param p The parser. */ void bc_parse_parse(BcParse* p); /** * Ends a series of if statements. This is to ensure that full parses happen * when a file finishes or before defining a function. Without this, bc thinks * that it cannot parse any further. But if we reach the end of a file or a * function definition, we know we can add an empty else clause. * @param p The parser. */ void bc_parse_endif(BcParse* p); /// References to the signal message and its length. extern const char bc_sig_msg[]; extern const uchar bc_sig_msg_len; /// A reference to an array of bits that are set if the corresponding lex token /// is valid in an expression. extern const uint8_t bc_parse_exprs[]; /// A reference to an array of bc operators. extern const uchar bc_parse_ops[]; // References to the various instances of BcParseNext's. /// A reference to what tokens are valid as next tokens when parsing normal /// expressions. More accurately. these are the tokens that are valid for /// *ending* the expression. extern const BcParseNext bc_parse_next_expr; /// A reference to what tokens are valid as next tokens when parsing function /// parameters (well, actually arguments). extern const BcParseNext bc_parse_next_arg; /// A reference to what tokens are valid as next tokens when parsing a print /// statement. extern const BcParseNext bc_parse_next_print; /// A reference to what tokens are valid as next tokens when parsing things like /// loop headers and builtin functions where the only thing expected is a right /// paren. /// /// The name is an artifact of history, and is related to @a BC_PARSE_REL (see /// include/parse.h). It refers to how POSIX only allows some operators as part /// of the conditional of for loops, while loops, and if statements. extern const BcParseNext bc_parse_next_rel; // What tokens are valid as next tokens when parsing an array element // expression. extern const BcParseNext bc_parse_next_elem; /// A reference to what tokens are valid as next tokens when parsing the first /// two parts of a for loop header. extern const BcParseNext bc_parse_next_for; /// A reference to what tokens are valid as next tokens when parsing a read /// expression. extern const BcParseNext bc_parse_next_read; /// A reference to what tokens are valid as next tokens when parsing a builtin /// function with multiple arguments. extern const BcParseNext bc_parse_next_builtin; #else // BC_ENABLED // If bc is not enabled, execution is always possible because dc has strict // rules that ensure execution can always proceed safely. #define BC_PARSE_NO_EXEC(p) (0) #endif // BC_ENABLED #endif // BC_BC_H diff --git a/include/dc.h b/include/dc.h index 1328f1c63b38..63f5ccbd10e3 100644 --- a/include/dc.h +++ b/include/dc.h @@ -1,110 +1,110 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * Definitions for dc only. * */ #ifndef BC_DC_H #define BC_DC_H #if DC_ENABLED #include #include #include /** * The main function for dc. It just sets variables and passes its arguments * through to @a bc_vm_boot(). * @return A status. */ BcStatus -dc_main(int argc, char* argv[]); +dc_main(int argc, const char* argv[]); // A reference to the dc help text. extern const char dc_help[]; /** * The @a BcLexNext function for dc. (See include/lex.h for a definition of * @a BcLexNext.) * @param l The lexer. */ void dc_lex_token(BcLex* l); /** * Returns true if the negative char `_` should be treated as a command or not. * dc considers negative a command if it does *not* immediately proceed a * number. Otherwise, it's just considered a negative. * @param l The lexer. * @return True if a negative should be treated as a command, false if it * should be treated as a negative sign on a number. */ bool dc_lex_negCommand(BcLex* l); // References to the signal message and its length. extern const char dc_sig_msg[]; extern const uchar dc_sig_msg_len; // References to an array and its length. This array is an array of lex tokens // that, when encountered, should be treated as commands that take a register. extern const uint8_t dc_lex_regs[]; extern const size_t dc_lex_regs_len; // References to an array of tokens and its length. This array corresponds to // the ASCII table, starting at double quotes. This makes it easy to look up // tokens for characters. extern const uint8_t dc_lex_tokens[]; extern const uint8_t dc_parse_insts[]; /** * The @a BcParseParse function for dc. (See include/parse.h for a definition of * @a BcParseParse.) * @param p The parser. */ void dc_parse_parse(BcParse* p); /** * The @a BcParseExpr function for dc. (See include/parse.h for a definition of * @a BcParseExpr.) * @param p The parser. * @param flags Flags that define the requirements that the parsed code must * meet or an error will result. See @a BcParseExpr for more info. */ void dc_parse_expr(BcParse* p, uint8_t flags); #endif // DC_ENABLED #endif // BC_DC_H diff --git a/include/opt.h b/include/opt.h index e60328994d8c..41058cb4e29c 100644 --- a/include/opt.h +++ b/include/opt.h @@ -1,142 +1,142 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * Adapted from https://github.com/skeeto/optparse * * ***************************************************************************** * * Definitions for getopt_long() replacement. * */ #ifndef BC_OPT_H #define BC_OPT_H #include #include /// The data required to parse command-line arguments. typedef struct BcOpt { /// The array of arguments. - char** argv; + const char** argv; /// The index of the current argument. size_t optind; /// The actual parse option character. int optopt; /// Where in the option we are for multi-character single-character options. int subopt; /// The option argument. - char* optarg; + const char* optarg; } BcOpt; /// The types of arguments. This is specially adapted for bc. typedef enum BcOptType { /// No argument required. BC_OPT_NONE, /// An argument required. BC_OPT_REQUIRED, /// An option that is bc-only. BC_OPT_BC_ONLY, /// An option that is bc-only that requires an argument. BC_OPT_REQUIRED_BC_ONLY, /// An option that is dc-only. BC_OPT_DC_ONLY, } BcOptType; /// A struct to hold const data for long options. typedef struct BcOptLong { /// The name of the option. const char* name; /// The type of the option. BcOptType type; /// The character to return if the long option was parsed. int val; } BcOptLong; /** * Initialize data for parsing options. * @param o The option data to initialize. * @param argv The array of arguments. */ void -bc_opt_init(BcOpt* o, char** argv); +bc_opt_init(BcOpt* o, const char** argv); /** * Parse an option. This returns a value the same way getopt() and getopt_long() * do, so it returns a character for the parsed option or -1 if done. * @param o The option data. * @param longopts The long options. * @return A character for the parsed option, or -1 if done. */ int bc_opt_parse(BcOpt* o, const BcOptLong* longopts); /** * Returns true if the option is `--` and not a long option. * @param a The argument to parse. * @return True if @a a is the `--` option, false otherwise. */ #define BC_OPT_ISDASHDASH(a) \ ((a) != NULL && (a)[0] == '-' && (a)[1] == '-' && (a)[2] == '\0') /** * Returns true if the option is a short option. * @param a The argument to parse. * @return True if @a a is a short option, false otherwise. */ #define BC_OPT_ISSHORTOPT(a) \ ((a) != NULL && (a)[0] == '-' && (a)[1] != '-' && (a)[1] != '\0') /** * Returns true if the option has `--` at the beginning, i.e., is a long option. * @param a The argument to parse. * @return True if @a a is a long option, false otherwise. */ #define BC_OPT_ISLONGOPT(a) \ ((a) != NULL && (a)[0] == '-' && (a)[1] == '-' && (a)[2] != '\0') #endif // BC_OPT_H diff --git a/src/dc.c b/include/ossfuzz.h similarity index 55% copy from src/dc.c copy to include/ossfuzz.h index 992efe262fd8..5c12a3c9c9fb 100644 --- a/src/dc.c +++ b/include/ossfuzz.h @@ -1,64 +1,79 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * - * The main procedure of dc. + * Declarations for the OSS-Fuzz build of bc and dc. * */ -#if DC_ENABLED +#include +#include -#include +#ifndef BC_OSSFUZZ_H +#define BC_OSSFUZZ_H -#include -#include +/// The number of args in fuzzer arguments, including the NULL terminator. +extern const size_t bc_fuzzer_args_len; + +/// The standard arguments for the bc fuzzer with the -c argument. +extern const char* bc_fuzzer_args_c[]; + +/// The standard arguments for the bc fuzzer with the -C argument. +extern const char* bc_fuzzer_args_C[]; + +/// The standard arguments for the dc fuzzer with the -c argument. +extern const char* dc_fuzzer_args_c[]; + +/// The standard arguments for the dc fuzzer with the -C argument. +extern const char* dc_fuzzer_args_C[]; + +/// The data pointer. +extern uint8_t* bc_fuzzer_data; /** - * The main function for dc. - * @param argc The number of arguments. - * @param argv The arguments. + * The function that the fuzzer runs. + * @param Data The data. + * @param Size The number of bytes in @a Data. + * @return 0 on success, -1 on error. + * @pre @a Data must not be equal to NULL if @a Size > 0. */ -BcStatus -dc_main(int argc, char* argv[]) -{ - // All of these just set dc-specific items in BcVm. - - vm->read_ret = BC_INST_POP_EXEC; - vm->help = dc_help; - vm->sigmsg = dc_sig_msg; - vm->siglen = dc_sig_msg_len; +int +LLVMFuzzerTestOneInput(const uint8_t* Data, size_t Size); - vm->next = dc_lex_token; - vm->parse = dc_parse_parse; - vm->expr = dc_parse_expr; +/** + * The initialization function for the fuzzer. + * @param argc A pointer to the argument count. + * @param argv A pointer to the argument list. + * @return 0 on success, -1 on error. + */ +int +LLVMFuzzerInitialize(int* argc, char*** argv); - return bc_vm_boot(argc, argv); -} -#endif // DC_ENABLED +#endif // BC_OSSFUZZ_H diff --git a/include/status.h b/include/status.h index f579df8c649b..203f09af628b 100644 --- a/include/status.h +++ b/include/status.h @@ -1,1042 +1,1025 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * All bc status codes and cross-platform portability. * */ #ifndef BC_STATUS_H #define BC_STATUS_H #ifdef _WIN32 #include #include #include #include #endif // _WIN32 #include #include -// This is used by configure.sh to test for OpenBSD. -#ifdef BC_TEST_OPENBSD -#ifdef __OpenBSD__ -#error On OpenBSD without _BSD_SOURCE -#endif // __OpenBSD__ -#endif // BC_TEST_OPENBSD - -// This is used by configure.sh to test for FreeBSD. -#ifdef BC_TEST_FREEBSD -#ifdef __FreeBSD__ -#error On FreeBSD with _POSIX_C_SOURCE -#endif // __FreeBSD__ -#endif // BC_TEST_FREEBSD - -// This is used by configure.sh to test for macOS. -#ifdef BC_TEST_APPLE -#ifdef __APPLE__ -#error On macOS without _DARWIN_C_SOURCE -#endif // __APPLE__ -#endif // BC_TEST_APPLE - // Windows has deprecated isatty() and the rest of these. Or doesn't have them. // So these are just fixes for Windows. #ifdef _WIN32 // This one is special. Windows did not like me defining an // inline function that was not given a definition in a header // file. This suppresses that by making inline functions non-inline. #define inline #define restrict __restrict #define strdup _strdup #define write(f, b, s) _write((f), (b), (unsigned int) (s)) #define read(f, b, s) _read((f), (b), (unsigned int) (s)) #define close _close #define open(f, n, m) \ _sopen_s((f), (n), (m) | _O_BINARY, _SH_DENYNO, _S_IREAD | _S_IWRITE) #define sigjmp_buf jmp_buf #define sigsetjmp(j, s) setjmp(j) #define siglongjmp longjmp #define isatty _isatty #define STDIN_FILENO _fileno(stdin) #define STDOUT_FILENO _fileno(stdout) #define STDERR_FILENO _fileno(stderr) #define S_ISDIR(m) ((m) & (_S_IFDIR)) #define O_RDONLY _O_RDONLY #define stat _stat #define fstat _fstat #define BC_FILE_SEP '\\' #else // _WIN32 #define BC_FILE_SEP '/' #endif // _WIN32 #ifndef BC_ENABLED #define BC_ENABLED (1) #endif // BC_ENABLED #ifndef DC_ENABLED #define DC_ENABLED (1) #endif // DC_ENABLED #ifndef BC_ENABLE_EXTRA_MATH #define BC_ENABLE_EXTRA_MATH (1) #endif // BC_ENABLE_EXTRA_MATH #ifndef BC_ENABLE_LIBRARY #define BC_ENABLE_LIBRARY (0) #endif // BC_ENABLE_LIBRARY #ifndef BC_ENABLE_HISTORY #define BC_ENABLE_HISTORY (1) #endif // BC_ENABLE_HISTORY #ifndef BC_ENABLE_EDITLINE #define BC_ENABLE_EDITLINE (0) #endif // BC_ENABLE_EDITLINE #ifndef BC_ENABLE_READLINE #define BC_ENABLE_READLINE (0) #endif // BC_ENABLE_READLINE #ifndef BC_ENABLE_NLS #define BC_ENABLE_NLS (0) #endif // BC_ENABLE_NLS #ifdef __OpenBSD__ #if BC_ENABLE_READLINE #error Cannot use readline on OpenBSD #endif // BC_ENABLE_READLINE #endif // __OpenBSD__ #if BC_ENABLE_EDITLINE && BC_ENABLE_READLINE #error Must enable only one of editline or readline, not both. #endif // BC_ENABLE_EDITLINE && BC_ENABLE_READLINE #if BC_ENABLE_EDITLINE || BC_ENABLE_READLINE #define BC_ENABLE_LINE_LIB (1) #else // BC_ENABLE_EDITLINE || BC_ENABLE_READLINE #define BC_ENABLE_LINE_LIB (0) #endif // BC_ENABLE_EDITLINE || BC_ENABLE_READLINE // This is error checking for fuzz builds. #if BC_ENABLE_AFL #ifndef __AFL_HAVE_MANUAL_CONTROL #error Must compile with afl-clang-fast or afl-clang-lto for fuzzing #endif // __AFL_HAVE_MANUAL_CONTROL #endif // BC_ENABLE_AFL #ifndef BC_ENABLE_MEMCHECK #define BC_ENABLE_MEMCHECK (0) #endif // BC_ENABLE_MEMCHECK /** * Mark a variable as unused. * @param e The variable to mark as unused. */ #define BC_UNUSED(e) ((void) (e)) // If users want, they can define this to something like __builtin_expect(e, 1). // It might give a performance improvement. #ifndef BC_LIKELY /** * Mark a branch expression as likely. * @param e The expression to mark as likely. */ #define BC_LIKELY(e) (e) #endif // BC_LIKELY // If users want, they can define this to something like __builtin_expect(e, 0). // It might give a performance improvement. #ifndef BC_UNLIKELY /** * Mark a branch expression as unlikely. * @param e The expression to mark as unlikely. */ #define BC_UNLIKELY(e) (e) #endif // BC_UNLIKELY /** * Mark a branch expression as an error, if true. * @param e The expression to mark as an error, if true. */ #define BC_ERR(e) BC_UNLIKELY(e) /** * Mark a branch expression as not an error, if true. * @param e The expression to mark as not an error, if true. */ #define BC_NO_ERR(s) BC_LIKELY(s) // Disable extra debug code by default. #ifndef BC_DEBUG_CODE #define BC_DEBUG_CODE (0) #endif // BC_DEBUG_CODE #if defined(__clang__) #define BC_CLANG (1) #else // defined(__clang__) #define BC_CLANG (0) #endif // defined(__clang__) #if defined(__GNUC__) && !BC_CLANG #define BC_GCC (1) #else // defined(__GNUC__) && !BC_CLANG #define BC_GCC (0) #endif // defined(__GNUC__) && !BC_CLANG // We want to be able to use _Noreturn on C11 compilers. #if __STDC_VERSION__ >= 201112L #include #define BC_NORETURN _Noreturn #define BC_C11 (1) #else // __STDC_VERSION__ #if BC_CLANG #if __has_attribute(noreturn) #define BC_NORETURN __attribute((noreturn)) #else // __has_attribute(noreturn) #define BC_NORETURN #endif // __has_attribute(noreturn) #else // BC_CLANG #define BC_NORETURN #endif // BC_CLANG #define BC_MUST_RETURN #define BC_C11 (0) #endif // __STDC_VERSION__ #define BC_HAS_UNREACHABLE (0) #define BC_HAS_COMPUTED_GOTO (0) // GCC and Clang complain if fallthroughs are not marked with their special // attribute. Jerks. This creates a define for marking the fallthroughs that is // nothing on other compilers. #if BC_CLANG || BC_GCC #if defined(__has_attribute) #if __has_attribute(fallthrough) #define BC_FALLTHROUGH __attribute__((fallthrough)); #else // __has_attribute(fallthrough) #define BC_FALLTHROUGH #endif // __has_attribute(fallthrough) #if BC_GCC #if __GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 5) #undef BC_HAS_UNREACHABLE #define BC_HAS_UNREACHABLE (1) #endif // __GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 5) #else // BC_GCC #if __clang_major__ >= 4 #undef BC_HAS_UNREACHABLE #define BC_HAS_UNREACHABLE (1) #endif // __clang_major__ >= 4 #endif // BC_GCC #else // defined(__has_attribute) #define BC_FALLTHROUGH #endif // defined(__has_attribute) #else // BC_CLANG || BC_GCC #define BC_FALLTHROUGH #endif // BC_CLANG || BC_GCC #if BC_HAS_UNREACHABLE #define BC_UNREACHABLE __builtin_unreachable(); #else // BC_HAS_UNREACHABLE #ifdef _WIN32 #define BC_UNREACHABLE __assume(0); #else // _WIN32 #define BC_UNREACHABLE #endif // _WIN32 #endif // BC_HAS_UNREACHABLE #if BC_GCC #if __GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 5) #undef BC_HAS_COMPUTED_GOTO #define BC_HAS_COMPUTED_GOTO (1) #endif // __GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 5) #endif // BC_GCC #if BC_CLANG #if __clang_major__ >= 4 #undef BC_HAS_COMPUTED_GOTO #define BC_HAS_COMPUTED_GOTO (1) #endif // __clang_major__ >= 4 #endif // BC_CLANG #ifdef BC_NO_COMPUTED_GOTO #undef BC_HAS_COMPUTED_GOTO #define BC_HAS_COMPUTED_GOTO (0) #endif // BC_NO_COMPUTED_GOTO #if BC_GCC #ifdef __OpenBSD__ // The OpenBSD GCC doesn't like inline. #define inline #endif // __OpenBSD__ #endif // BC_GCC // Workarounds for AIX's POSIX incompatibility. #ifndef SIZE_MAX #define SIZE_MAX __SIZE_MAX__ #endif // SIZE_MAX #ifndef UINTMAX_C #define UINTMAX_C __UINTMAX_C #endif // UINTMAX_C #ifndef UINT32_C #define UINT32_C __UINT32_C #endif // UINT32_C #ifndef UINT_FAST32_MAX #define UINT_FAST32_MAX __UINT_FAST32_MAX__ #endif // UINT_FAST32_MAX #ifndef UINT16_MAX #define UINT16_MAX __UINT16_MAX__ #endif // UINT16_MAX #ifndef SIG_ATOMIC_MAX #define SIG_ATOMIC_MAX __SIG_ATOMIC_MAX__ #endif // SIG_ATOMIC_MAX // Yes, this has to be here. #include // All of these set defaults for settings. #if BC_ENABLED #ifndef BC_DEFAULT_BANNER #define BC_DEFAULT_BANNER (0) #endif // BC_DEFAULT_BANNER #endif // BC_ENABLED #ifndef BC_DEFAULT_SIGINT_RESET #define BC_DEFAULT_SIGINT_RESET (1) #endif // BC_DEFAULT_SIGINT_RESET #ifndef BC_DEFAULT_TTY_MODE #define BC_DEFAULT_TTY_MODE (1) #endif // BC_DEFAULT_TTY_MODE #ifndef BC_DEFAULT_PROMPT #define BC_DEFAULT_PROMPT BC_DEFAULT_TTY_MODE #endif // BC_DEFAULT_PROMPT #ifndef BC_DEFAULT_EXPR_EXIT #define BC_DEFAULT_EXPR_EXIT (1) #endif // BC_DEFAULT_EXPR_EXIT #ifndef BC_DEFAULT_DIGIT_CLAMP #define BC_DEFAULT_DIGIT_CLAMP (0) #endif // BC_DEFAULT_DIGIT_CLAMP // All of these set defaults for settings. #ifndef DC_DEFAULT_SIGINT_RESET #define DC_DEFAULT_SIGINT_RESET (1) #endif // DC_DEFAULT_SIGINT_RESET #ifndef DC_DEFAULT_TTY_MODE #define DC_DEFAULT_TTY_MODE (0) #endif // DC_DEFAULT_TTY_MODE #ifndef DC_DEFAULT_HISTORY #define DC_DEFAULT_HISTORY DC_DEFAULT_TTY_MODE #endif // DC_DEFAULT_HISTORY #ifndef DC_DEFAULT_PROMPT #define DC_DEFAULT_PROMPT DC_DEFAULT_TTY_MODE #endif // DC_DEFAULT_PROMPT #ifndef DC_DEFAULT_EXPR_EXIT #define DC_DEFAULT_EXPR_EXIT (1) #endif // DC_DEFAULT_EXPR_EXIT #ifndef DC_DEFAULT_DIGIT_CLAMP #define DC_DEFAULT_DIGIT_CLAMP (0) #endif // DC_DEFAULT_DIGIT_CLAMP /// Statuses, which mark either which category of error happened, or some other /// status that matters. typedef enum BcStatus { /// Normal status. BC_STATUS_SUCCESS = 0, /// Math error. BC_STATUS_ERROR_MATH, /// Parse (and lex) error. BC_STATUS_ERROR_PARSE, /// Runtime error. BC_STATUS_ERROR_EXEC, /// Fatal error. BC_STATUS_ERROR_FATAL, /// EOF status. BC_STATUS_EOF, /// Quit status. This means that bc/dc is in the process of quitting. BC_STATUS_QUIT, } BcStatus; /// Errors, which are more specific errors. typedef enum BcErr { // Math errors. /// Negative number used when not allowed. BC_ERR_MATH_NEGATIVE, /// Non-integer used when not allowed. BC_ERR_MATH_NON_INTEGER, /// Conversion to a hardware integer would overflow. BC_ERR_MATH_OVERFLOW, /// Divide by zero. BC_ERR_MATH_DIVIDE_BY_ZERO, // Fatal errors. /// An allocation or reallocation failed. BC_ERR_FATAL_ALLOC_ERR, /// I/O failure. BC_ERR_FATAL_IO_ERR, /// File error, such as permissions or file does not exist. BC_ERR_FATAL_FILE_ERR, /// File is binary, not text, error. BC_ERR_FATAL_BIN_FILE, /// Attempted to read a directory as a file error. BC_ERR_FATAL_PATH_DIR, /// Invalid option error. BC_ERR_FATAL_OPTION, /// Option with required argument not given an argument. BC_ERR_FATAL_OPTION_NO_ARG, /// Option with no argument given an argument. BC_ERR_FATAL_OPTION_ARG, /// Option argument is invalid. BC_ERR_FATAL_ARG, // Runtime errors. /// Invalid ibase value. BC_ERR_EXEC_IBASE, /// Invalid obase value. BC_ERR_EXEC_OBASE, /// Invalid scale value. BC_ERR_EXEC_SCALE, /// Invalid expression parsed by read(). BC_ERR_EXEC_READ_EXPR, /// read() used within an expression given to a read() call. BC_ERR_EXEC_REC_READ, /// Type error. BC_ERR_EXEC_TYPE, /// Stack has too few elements error. BC_ERR_EXEC_STACK, /// Register stack has too few elements error. BC_ERR_EXEC_STACK_REGISTER, /// Wrong number of arguments error. BC_ERR_EXEC_PARAMS, /// Undefined function error. BC_ERR_EXEC_UNDEF_FUNC, /// Void value used in an expression error. BC_ERR_EXEC_VOID_VAL, // Parse (and lex) errors. /// EOF encountered when not expected error. BC_ERR_PARSE_EOF, /// Invalid character error. BC_ERR_PARSE_CHAR, /// Invalid string (no ending quote) error. BC_ERR_PARSE_STRING, /// Invalid comment (no end found) error. BC_ERR_PARSE_COMMENT, /// Invalid token encountered error. BC_ERR_PARSE_TOKEN, #if BC_ENABLED /// Invalid expression error. BC_ERR_PARSE_EXPR, /// Expression is empty error. BC_ERR_PARSE_EMPTY_EXPR, /// Print statement is invalid error. BC_ERR_PARSE_PRINT, /// Function definition is invalid error. BC_ERR_PARSE_FUNC, /// Assignment is invalid error. BC_ERR_PARSE_ASSIGN, /// No auto identifiers given for an auto statement error. BC_ERR_PARSE_NO_AUTO, /// Duplicate local (parameter or auto) error. BC_ERR_PARSE_DUP_LOCAL, /// Invalid block (within braces) error. BC_ERR_PARSE_BLOCK, /// Invalid return statement for void functions. BC_ERR_PARSE_RET_VOID, /// Reference attached to a variable, not an array, error. BC_ERR_PARSE_REF_VAR, // POSIX-only errors. /// Name length greater than 1 error. BC_ERR_POSIX_NAME_LEN, /// Non-POSIX comment used error. BC_ERR_POSIX_COMMENT, /// Non-POSIX keyword error. BC_ERR_POSIX_KW, /// Non-POSIX . (last) error. BC_ERR_POSIX_DOT, /// Non-POSIX return error. BC_ERR_POSIX_RET, /// Non-POSIX boolean operator used error. BC_ERR_POSIX_BOOL, /// POSIX relation operator used outside if, while, or for statements error. BC_ERR_POSIX_REL_POS, /// Multiple POSIX relation operators used in an if, while, or for statement /// error. BC_ERR_POSIX_MULTIREL, /// Empty statements in POSIX for loop error. BC_ERR_POSIX_FOR, /// POSIX's grammar does not allow a function definition right after a /// semicolon. BC_ERR_POSIX_FUNC_AFTER_SEMICOLON, /// Non-POSIX exponential (scientific or engineering) number used error. BC_ERR_POSIX_EXP_NUM, /// Non-POSIX array reference error. BC_ERR_POSIX_REF, /// Non-POSIX void error. BC_ERR_POSIX_VOID, /// Non-POSIX brace position used error. BC_ERR_POSIX_BRACE, /// String used in expression. BC_ERR_POSIX_EXPR_STRING, #endif // BC_ENABLED // Number of elements. BC_ERR_NELEMS, #if BC_ENABLED /// A marker for the start of POSIX errors. BC_ERR_POSIX_START = BC_ERR_POSIX_NAME_LEN, /// A marker for the end of POSIX errors. BC_ERR_POSIX_END = BC_ERR_POSIX_EXPR_STRING, #endif // BC_ENABLED } BcErr; // The indices of each category of error in bc_errs[], and used in bc_err_ids[] // to associate actual errors with their categories. /// Math error category. #define BC_ERR_IDX_MATH (0) /// Parse (and lex) error category. #define BC_ERR_IDX_PARSE (1) /// Runtime error category. #define BC_ERR_IDX_EXEC (2) /// Fatal error category. #define BC_ERR_IDX_FATAL (3) /// Number of categories. #define BC_ERR_IDX_NELEMS (4) // If bc is enabled, we add an extra category for POSIX warnings. #if BC_ENABLED /// POSIX warning category. #define BC_ERR_IDX_WARN (BC_ERR_IDX_NELEMS) #endif // BC_ENABLED /** * The mode bc is in. This is basically what input it is processing. */ typedef enum BcMode { /// Expressions mode. BC_MODE_EXPRS, /// File mode. BC_MODE_FILE, +#if !BC_ENABLE_OSSFUZZ + /// stdin mode. BC_MODE_STDIN, +#endif // !BC_ENABLE_OSSFUZZ + } BcMode; /// Do a longjmp(). This is what to use when activating an "exception", i.e., a /// longjmp(). With debug code, it will print the name of the function it jumped /// from. #if BC_DEBUG_CODE #define BC_JMP bc_vm_jmp(__func__) #else // BC_DEBUG_CODE #define BC_JMP bc_vm_jmp() #endif // BC_DEBUG_CODE #if !BC_ENABLE_LIBRARY /// Returns true if an exception is in flight, false otherwise. #define BC_SIG_EXC(vm) \ BC_UNLIKELY((vm)->status != (sig_atomic_t) BC_STATUS_SUCCESS || (vm)->sig) /// Returns true if there is *no* exception in flight, false otherwise. #define BC_NO_SIG_EXC(vm) \ BC_LIKELY((vm)->status == (sig_atomic_t) BC_STATUS_SUCCESS && !(vm)->sig) #ifndef _WIN32 #define BC_SIG_INTERRUPT(vm) \ BC_UNLIKELY((vm)->sig != 0 && (vm)->sig != SIGWINCH) #else // _WIN32 #define BC_SIG_INTERRUPT(vm) BC_UNLIKELY((vm)->sig != 0) #endif // _WIN32 #if BC_DEBUG /// Assert that signals are locked. There are non-async-signal-safe functions in /// bc, and they *must* have signals locked. Other functions are expected to /// *not* have signals locked, for reasons. So this is a pre-built assert /// (no-op in non-debug mode) that check that signals are locked. #define BC_SIG_ASSERT_LOCKED \ do \ { \ assert(vm->sig_lock); \ } \ while (0) /// Assert that signals are unlocked. There are non-async-signal-safe functions /// in bc, and they *must* have signals locked. Other functions are expected to /// *not* have signals locked, for reasons. So this is a pre-built assert /// (no-op in non-debug mode) that check that signals are unlocked. #define BC_SIG_ASSERT_NOT_LOCKED \ do \ { \ assert(vm->sig_lock == 0); \ } \ while (0) #else // BC_DEBUG /// Assert that signals are locked. There are non-async-signal-safe functions in /// bc, and they *must* have signals locked. Other functions are expected to /// *not* have signals locked, for reasons. So this is a pre-built assert /// (no-op in non-debug mode) that check that signals are locked. #define BC_SIG_ASSERT_LOCKED /// Assert that signals are unlocked. There are non-async-signal-safe functions /// in bc, and they *must* have signals locked. Other functions are expected to /// *not* have signals locked, for reasons. So this is a pre-built assert /// (no-op in non-debug mode) that check that signals are unlocked. #define BC_SIG_ASSERT_NOT_LOCKED #endif // BC_DEBUG /// Locks signals. #define BC_SIG_LOCK \ do \ { \ BC_SIG_ASSERT_NOT_LOCKED; \ vm->sig_lock = 1; \ } \ while (0) /// Unlocks signals. If a signal happened, then this will cause a jump. #define BC_SIG_UNLOCK \ do \ { \ BC_SIG_ASSERT_LOCKED; \ vm->sig_lock = 0; \ if (vm->sig) BC_JMP; \ } \ while (0) /// Locks signals, regardless of if they are already locked. This is really only /// used after labels that longjmp() goes to after the jump because the cleanup /// code must have signals locked, and BC_LONGJMP_CONT will unlock signals if it /// doesn't jump. #define BC_SIG_MAYLOCK \ do \ { \ vm->sig_lock = 1; \ } \ while (0) /// Unlocks signals, regardless of if they were already unlocked. If a signal /// happened, then this will cause a jump. #define BC_SIG_MAYUNLOCK \ do \ { \ vm->sig_lock = 0; \ if (vm->sig) BC_JMP; \ } \ while (0) /** * Locks signals, but stores the old lock state, to be restored later by * BC_SIG_TRYUNLOCK. * @param v The variable to store the old lock state to. */ #define BC_SIG_TRYLOCK(v) \ do \ { \ v = vm->sig_lock; \ vm->sig_lock = 1; \ } \ while (0) /** * Restores the previous state of a signal lock, and if it is now unlocked, * initiates an exception/jump. * @param v The old lock state. */ #define BC_SIG_TRYUNLOCK(v) \ do \ { \ vm->sig_lock = (v); \ if (!(v) && vm->sig) BC_JMP; \ } \ while (0) /// Stops a stack unwinding. Technically, a stack unwinding needs to be done /// manually, but it will always be done unless certain flags are cleared. This /// clears the flags. #define BC_LONGJMP_STOP \ do \ { \ vm->sig_pop = 0; \ vm->sig = 0; \ } \ while (0) /** * Sets a jump like BC_SETJMP, but unlike BC_SETJMP, it assumes signals are * locked and will just set the jump. This does *not* have a call to * bc_vec_grow() because it is assumed that BC_SETJMP_LOCKED(l) is used *after* * the initializations that need the setjmp(). * param l The label to jump to on a longjmp(). */ #define BC_SETJMP_LOCKED(vm, l) \ do \ { \ sigjmp_buf sjb; \ BC_SIG_ASSERT_LOCKED; \ if (sigsetjmp(sjb, 0)) \ { \ assert(BC_SIG_EXC(vm)); \ goto l; \ } \ bc_vec_push(&vm->jmp_bufs, &sjb); \ } \ while (0) /// Used after cleanup labels set by BC_SETJMP and BC_SETJMP_LOCKED to jump to /// the next place. This is what continues the stack unwinding. This basically /// copies BC_SIG_UNLOCK into itself, but that is because its condition for /// jumping is BC_SIG_EXC, not just that a signal happened. #define BC_LONGJMP_CONT(vm) \ do \ { \ BC_SIG_ASSERT_LOCKED; \ if (!vm->sig_pop) bc_vec_pop(&vm->jmp_bufs); \ vm->sig_lock = 0; \ if (BC_SIG_EXC(vm)) BC_JMP; \ } \ while (0) #else // !BC_ENABLE_LIBRARY #define BC_SIG_LOCK #define BC_SIG_UNLOCK #define BC_SIG_MAYLOCK #define BC_SIG_TRYLOCK(lock) #define BC_SIG_TRYUNLOCK(lock) #define BC_SIG_ASSERT_LOCKED /// Returns true if an exception is in flight, false otherwise. #define BC_SIG_EXC(vm) \ BC_UNLIKELY(vm->status != (sig_atomic_t) BC_STATUS_SUCCESS) /// Returns true if there is *no* exception in flight, false otherwise. #define BC_NO_SIG_EXC(vm) \ BC_LIKELY(vm->status == (sig_atomic_t) BC_STATUS_SUCCESS) /// Used after cleanup labels set by BC_SETJMP and BC_SETJMP_LOCKED to jump to /// the next place. This is what continues the stack unwinding. This basically /// copies BC_SIG_UNLOCK into itself, but that is because its condition for /// jumping is BC_SIG_EXC, not just that a signal happened. #define BC_LONGJMP_CONT(vm) \ do \ { \ bc_vec_pop(&vm->jmp_bufs); \ if (BC_SIG_EXC(vm)) BC_JMP; \ } \ while (0) #endif // !BC_ENABLE_LIBRARY /** * Sets a jump, and sets it up as well so that if a longjmp() happens, bc will * immediately goto a label where some cleanup code is. This one assumes that * signals are not locked and will lock them, set the jump, and unlock them. * Setting the jump also includes pushing the jmp_buf onto the jmp_buf stack. * This grows the jmp_bufs vector first to prevent a fatal error from happening * after the setjmp(). This is done because BC_SETJMP(l) is assumed to be used * *before* the actual initialization calls that need the setjmp(). * param l The label to jump to on a longjmp(). */ #define BC_SETJMP(vm, l) \ do \ { \ sigjmp_buf sjb; \ BC_SIG_LOCK; \ bc_vec_grow(&vm->jmp_bufs, 1); \ if (sigsetjmp(sjb, 0)) \ { \ assert(BC_SIG_EXC(vm)); \ goto l; \ } \ bc_vec_push(&vm->jmp_bufs, &sjb); \ BC_SIG_UNLOCK; \ } \ while (0) /// Unsets a jump. It always assumes signals are locked. This basically just /// pops a jmp_buf off of the stack of jmp_bufs, and since the jump mechanism /// always jumps to the location at the top of the stack, this effectively /// undoes a setjmp(). #define BC_UNSETJMP(vm) \ do \ { \ BC_SIG_ASSERT_LOCKED; \ bc_vec_pop(&vm->jmp_bufs); \ } \ while (0) #if BC_ENABLE_LIBRARY #define BC_SETJMP_LOCKED(vm, l) BC_SETJMP(vm, l) // Various convenience macros for calling the bc's error handling routine. /** * Call bc's error handling routine. * @param e The error. * @param l The line of the script that the error happened. * @param ... Extra arguments for error messages as necessary. */ #define bc_error(e, l, ...) (bc_vm_handleError((e))) /** * Call bc's error handling routine. * @param e The error. */ #define bc_err(e) (bc_vm_handleError((e))) /** * Call bc's error handling routine. * @param e The error. */ #define bc_verr(e, ...) (bc_vm_handleError((e))) #else // BC_ENABLE_LIBRARY // Various convenience macros for calling the bc's error handling routine. /** * Call bc's error handling routine. * @param e The error. * @param l The line of the script that the error happened. * @param ... Extra arguments for error messages as necessary. */ #if BC_DEBUG #define bc_error(e, l, ...) \ (bc_vm_handleError((e), __FILE__, __LINE__, (l), __VA_ARGS__)) #else // BC_DEBUG #define bc_error(e, l, ...) (bc_vm_handleError((e), (l), __VA_ARGS__)) #endif // BC_DEBUG /** * Call bc's error handling routine. * @param e The error. */ #if BC_DEBUG #define bc_err(e) (bc_vm_handleError((e), __FILE__, __LINE__, 0)) #else // BC_DEBUG #define bc_err(e) (bc_vm_handleError((e), 0)) #endif // BC_DEBUG /** * Call bc's error handling routine. * @param e The error. */ #if BC_DEBUG #define bc_verr(e, ...) \ (bc_vm_handleError((e), __FILE__, __LINE__, 0, __VA_ARGS__)) #else // BC_DEBUG #define bc_verr(e, ...) (bc_vm_handleError((e), 0, __VA_ARGS__)) #endif // BC_DEBUG #endif // BC_ENABLE_LIBRARY /** * Returns true if status @a s is an error, false otherwise. * @param s The status to test. * @return True if @a s is an error, false otherwise. */ #define BC_STATUS_IS_ERROR(s) \ ((s) >= BC_STATUS_ERROR_MATH && (s) <= BC_STATUS_ERROR_FATAL) // Convenience macros that can be placed at the beginning and exits of functions // for easy marking of where functions are entered and exited. #if BC_DEBUG_CODE #define BC_FUNC_ENTER \ do \ { \ size_t bc_func_enter_i; \ for (bc_func_enter_i = 0; bc_func_enter_i < vm->func_depth; \ ++bc_func_enter_i) \ { \ bc_file_puts(&vm->ferr, bc_flush_none, " "); \ } \ vm->func_depth += 1; \ bc_file_printf(&vm->ferr, "Entering %s\n", __func__); \ bc_file_flush(&vm->ferr, bc_flush_none); \ } \ while (0); #define BC_FUNC_EXIT \ do \ { \ size_t bc_func_enter_i; \ vm->func_depth -= 1; \ for (bc_func_enter_i = 0; bc_func_enter_i < vm->func_depth; \ ++bc_func_enter_i) \ { \ bc_file_puts(&vm->ferr, bc_flush_none, " "); \ } \ bc_file_printf(&vm->ferr, "Leaving %s\n", __func__); \ bc_file_flush(&vm->ferr, bc_flush_none); \ } \ while (0); #else // BC_DEBUG_CODE #define BC_FUNC_ENTER #define BC_FUNC_EXIT #endif // BC_DEBUG_CODE #endif // BC_STATUS_H diff --git a/include/version.h b/include/version.h index 586691a6e7ef..897a19530e3f 100644 --- a/include/version.h +++ b/include/version.h @@ -1,42 +1,42 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * The version of bc. * */ #ifndef BC_VERSION_H #define BC_VERSION_H /// The current version. -#define VERSION 6.7.6 +#define VERSION 7.0.0 #endif // BC_VERSION_H diff --git a/include/vm.h b/include/vm.h index 052c1d14c237..e81206b63871 100644 --- a/include/vm.h +++ b/include/vm.h @@ -1,1092 +1,1092 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * Definitions for bc's VM. * */ #ifndef BC_VM_H #define BC_VM_H #include #include #include #include #if BC_ENABLE_NLS #ifdef _WIN32 #error NLS is not supported on Windows. #endif // _WIN32 #include #endif // BC_ENABLE_NLS #include #include #include #include #include #include #include #include // We don't want to include this file for the library because it's unused. #if !BC_ENABLE_LIBRARY #include #endif // !BC_ENABLE_LIBRARY // This should be obvious. If neither calculator is enabled, barf. #if !BC_ENABLED && !DC_ENABLED #error Must define BC_ENABLED, DC_ENABLED, or both #endif // CHAR_BIT must be at least 6, for various reasons. I might want to bump this // to 8 in the future. #if CHAR_BIT < 6 #error CHAR_BIT must be at least 6. #endif // Set defaults. #ifndef MAINEXEC #define MAINEXEC bc #endif // MAINEXEC #ifndef _WIN32 #ifndef EXECPREFIX #define EXECPREFIX #endif // EXECPREFIX #else // _WIN32 #undef EXECPREFIX #endif // _WIN32 /** * Generate a string from text. * @parm V The text to generate a string for. */ #define GEN_STR(V) #V /** * Help generate a string from text. The preprocessor requires this two-step * process. Trust me. * @parm V The text to generate a string for. */ #define GEN_STR2(V) GEN_STR(V) /// The version as a string. VERSION must be defined previously, usually by the /// build system. #define BC_VERSION GEN_STR2(VERSION) /// The main executable name as a string. MAINEXEC must be defined previously, /// usually by the build system. #define BC_MAINEXEC GEN_STR2(MAINEXEC) /// The build type as a string. BUILD_TYPE must be defined previously, usually /// by the build system. #define BC_BUILD_TYPE GEN_STR2(BUILD_TYPE) // We only allow an empty executable prefix on Windows. #ifndef _WIN32 #define BC_EXECPREFIX GEN_STR2(EXECPREFIX) #else // _WIN32 #define BC_EXECPREFIX "" #endif // _WIN32 #if !BC_ENABLE_LIBRARY #if DC_ENABLED /// The flag for the extended register option. #define DC_FLAG_X (UINTMAX_C(1) << 0) #endif // DC_ENABLED #if BC_ENABLED /// The flag for the POSIX warning option. #define BC_FLAG_W (UINTMAX_C(1) << 1) /// The flag for the POSIX error option. #define BC_FLAG_S (UINTMAX_C(1) << 2) /// The flag for the math library option. #define BC_FLAG_L (UINTMAX_C(1) << 3) /// The flag for the global stacks option. #define BC_FLAG_G (UINTMAX_C(1) << 4) #endif // BC_ENABLED /// The flag for quiet, though this one is reversed; the option clears the flag. #define BC_FLAG_Q (UINTMAX_C(1) << 5) /// The flag for interactive. #define BC_FLAG_I (UINTMAX_C(1) << 6) /// The flag for prompt. This is also reversed; the option clears the flag. #define BC_FLAG_P (UINTMAX_C(1) << 7) /// The flag for read prompt. This is also reversed; the option clears the flag. #define BC_FLAG_R (UINTMAX_C(1) << 8) /// The flag for a leading zero. #define BC_FLAG_Z (UINTMAX_C(1) << 9) /// The flag for stdin being a TTY. #define BC_FLAG_TTYIN (UINTMAX_C(1) << 10) /// The flag for TTY mode. #define BC_FLAG_TTY (UINTMAX_C(1) << 11) /// The flag for reset on SIGINT. #define BC_FLAG_SIGINT (UINTMAX_C(1) << 12) /// The flag for exiting with expressions. #define BC_FLAG_EXPR_EXIT (UINTMAX_C(1) << 13) /// The flag for digit clamping. #define BC_FLAG_DIGIT_CLAMP (UINTMAX_C(1) << 14) /// A convenience macro for getting the TTYIN flag. #define BC_TTYIN (vm->flags & BC_FLAG_TTYIN) /// A convenience macro for getting the TTY flag. #define BC_TTY (vm->flags & BC_FLAG_TTY) /// A convenience macro for getting the SIGINT flag. #define BC_SIGINT (vm->flags & BC_FLAG_SIGINT) #if BC_ENABLED /// A convenience macro for getting the POSIX error flag. #define BC_S (vm->flags & BC_FLAG_S) /// A convenience macro for getting the POSIX warning flag. #define BC_W (vm->flags & BC_FLAG_W) /// A convenience macro for getting the math library flag. #define BC_L (vm->flags & BC_FLAG_L) /// A convenience macro for getting the global stacks flag. #define BC_G (vm->flags & BC_FLAG_G) #endif // BC_ENABLED #if DC_ENABLED /// A convenience macro for getting the extended register flag. #define DC_X (vm->flags & DC_FLAG_X) #endif // DC_ENABLED /// A convenience macro for getting the interactive flag. #define BC_I (vm->flags & BC_FLAG_I) /// A convenience macro for getting the prompt flag. #define BC_P (vm->flags & BC_FLAG_P) /// A convenience macro for getting the read prompt flag. #define BC_R (vm->flags & BC_FLAG_R) /// A convenience macro for getting the leading zero flag. #define BC_Z (vm->flags & BC_FLAG_Z) /// A convenience macro for getting the expression exit flag. #define BC_EXPR_EXIT (vm->flags & BC_FLAG_EXPR_EXIT) /// A convenience macro for getting the digit clamp flag. #define BC_DIGIT_CLAMP (vm->flags & BC_FLAG_DIGIT_CLAMP) #if BC_ENABLED /// A convenience macro for checking if bc is in POSIX mode. #define BC_IS_POSIX (BC_S || BC_W) #if DC_ENABLED /// Returns true if bc is running. #define BC_IS_BC (vm->name[0] != 'd') /// Returns true if dc is running. #define BC_IS_DC (vm->name[0] == 'd') /// Returns the correct read prompt. #define BC_VM_READ_PROMPT (BC_IS_BC ? "read> " : "?> ") /// Returns the string for the line length environment variable. #define BC_VM_LINE_LENGTH_STR (BC_IS_BC ? "BC_LINE_LENGTH" : "DC_LINE_LENGTH") /// Returns the string for the environment args environment variable. #define BC_VM_ENV_ARGS_STR (BC_IS_BC ? "BC_ENV_ARGS" : "DC_ENV_ARGS") /// Returns the string for the expression exit environment variable. #define BC_VM_EXPR_EXIT_STR (BC_IS_BC ? "BC_EXPR_EXIT" : "DC_EXPR_EXIT") /// Returns the default for the expression exit environment variable. #define BC_VM_EXPR_EXIT_DEF \ (BC_IS_BC ? BC_DEFAULT_EXPR_EXIT : DC_DEFAULT_EXPR_EXIT) /// Returns the string for the digit clamp environment variable. #define BC_VM_DIGIT_CLAMP_STR (BC_IS_BC ? "BC_DIGIT_CLAMP" : "DC_DIGIT_CLAMP") /// Returns the default for the digit clamp environment variable. #define BC_VM_DIGIT_CLAMP_DEF \ (BC_IS_BC ? BC_DEFAULT_DIGIT_CLAMP : DC_DEFAULT_DIGIT_CLAMP) /// Returns the string for the TTY mode environment variable. #define BC_VM_TTY_MODE_STR (BC_IS_BC ? "BC_TTY_MODE" : "DC_TTY_MODE") /// Returns the default for the TTY mode environment variable. #define BC_VM_TTY_MODE_DEF \ (BC_IS_BC ? BC_DEFAULT_TTY_MODE : DC_DEFAULT_TTY_MODE) /// Returns the string for the prompt environment variable. #define BC_VM_PROMPT_STR (BC_IS_BC ? "BC_PROMPT" : "DC_PROMPT") /// Returns the default for the prompt environment variable. #define BC_VM_PROMPT_DEF (BC_IS_BC ? BC_DEFAULT_PROMPT : DC_DEFAULT_PROMPT) /// Returns the string for the SIGINT reset environment variable. #define BC_VM_SIGINT_RESET_STR \ (BC_IS_BC ? "BC_SIGINT_RESET" : "DC_SIGINT_RESET") /// Returns the string for the SIGINT reset environment variable. #define BC_VM_SIGINT_RESET_DEF \ (BC_IS_BC ? BC_DEFAULT_SIGINT_RESET : DC_DEFAULT_SIGINT_RESET) /// Returns true if the calculator should run stdin. #define BC_VM_RUN_STDIN(has_file) (BC_IS_BC || !(has_file)) #else // DC_ENABLED /// Returns true if bc is running. #define BC_IS_BC (1) /// Returns true if dc is running. #define BC_IS_DC (0) /// Returns the correct read prompt. #define BC_VM_READ_PROMPT ("read> ") /// Returns the string for the line length environment variable. #define BC_VM_LINE_LENGTH_STR ("BC_LINE_LENGTH") /// Returns the string for the environment args environment variable. #define BC_VM_ENV_ARGS_STR ("BC_ENV_ARGS") /// Returns the string for the expression exit environment variable. #define BC_VM_EXPR_EXIT_STR ("BC_EXPR_EXIT") /// Returns the default for the expression exit environment variable. #define BC_VM_EXPR_EXIT_DEF (BC_DEFAULT_EXPR_EXIT) /// Returns the string for the digit clamp environment variable. #define BC_VM_DIGIT_CLAMP_STR ("BC_DIGIT_CLAMP") /// Returns the default for the digit clamp environment variable. #define BC_VM_DIGIT_CLAMP_DEF (BC_DEFAULT_DIGIT_CLAMP) /// Returns the string for the TTY mode environment variable. #define BC_VM_TTY_MODE_STR ("BC_TTY_MODE") /// Returns the default for the TTY mode environment variable. #define BC_VM_TTY_MODE_DEF (BC_DEFAULT_TTY_MODE) /// Returns the string for the prompt environment variable. #define BC_VM_PROMPT_STR ("BC_PROMPT") /// Returns the default for the SIGINT reset environment variable. #define BC_VM_PROMPT_DEF (BC_DEFAULT_PROMPT) /// Returns the string for the SIGINT reset environment variable. #define BC_VM_SIGINT_RESET_STR ("BC_SIGINT_RESET") /// Returns the string for the SIGINT reset environment variable. #define BC_VM_SIGINT_RESET_DEF (BC_DEFAULT_SIGINT_RESET) /// Returns true if the calculator should run stdin. #define BC_VM_RUN_STDIN(has_file) (BC_IS_BC) #endif // DC_ENABLED #else // BC_ENABLED /// A convenience macro for checking if bc is in POSIX mode. #define BC_IS_POSIX (0) /// Returns true if bc is running. #define BC_IS_BC (0) /// Returns true if dc is running. #define BC_IS_DC (1) /// Returns the correct read prompt. #define BC_VM_READ_PROMPT ("?> ") /// Returns the string for the line length environment variable. #define BC_VM_LINE_LENGTH_STR ("DC_LINE_LENGTH") /// Returns the string for the environment args environment variable. #define BC_VM_ENV_ARGS_STR ("DC_ENV_ARGS") /// Returns the string for the expression exit environment variable. #define BC_VM_EXPR_EXIT_STR ("DC_EXPR_EXIT") /// Returns the default for the expression exit environment variable. #define BC_VM_EXPR_EXIT_DEF (DC_DEFAULT_EXPR_EXIT) /// Returns the string for the digit clamp environment variable. #define BC_VM_DIGIT_CLAMP_STR ("DC_DIGIT_CLAMP") /// Returns the default for the digit clamp environment variable. #define BC_VM_DIGIT_CLAMP_DEF (DC_DEFAULT_DIGIT_CLAMP) /// Returns the string for the TTY mode environment variable. #define BC_VM_TTY_MODE_STR ("DC_TTY_MODE") /// Returns the default for the TTY mode environment variable. #define BC_VM_TTY_MODE_DEF (DC_DEFAULT_TTY_MODE) /// Returns the string for the prompt environment variable. #define BC_VM_PROMPT_STR ("DC_PROMPT") /// Returns the default for the SIGINT reset environment variable. #define BC_VM_PROMPT_DEF (DC_DEFAULT_PROMPT) /// Returns the string for the SIGINT reset environment variable. #define BC_VM_SIGINT_RESET_STR ("DC_SIGINT_RESET") /// Returns the string for the SIGINT reset environment variable. #define BC_VM_SIGINT_RESET_DEF (DC_DEFAULT_SIGINT_RESET) /// Returns true if the calculator should run stdin. #define BC_VM_RUN_STDIN(has_file) (!(has_file)) #endif // BC_ENABLED /// A convenience macro for checking if the prompt is enabled. #define BC_PROMPT (BC_P) #else // !BC_ENABLE_LIBRARY #define BC_Z (vm->leading_zeroes) #define BC_DIGIT_CLAMP (vm->digit_clamp) #endif // !BC_ENABLE_LIBRARY /** * Returns the max of its two arguments. This evaluates arguments twice, so be * careful what args you give it. * @param a The first argument. * @param b The second argument. * @return The max of the two arguments. */ #define BC_MAX(a, b) ((a) > (b) ? (a) : (b)) /** * Returns the min of its two arguments. This evaluates arguments twice, so be * careful what args you give it. * @param a The first argument. * @param b The second argument. * @return The min of the two arguments. */ #define BC_MIN(a, b) ((a) < (b) ? (a) : (b)) /// Returns the max obase that is allowed. #define BC_MAX_OBASE ((BcBigDig) (BC_BASE_POW)) /// Returns the max array size that is allowed. #define BC_MAX_DIM ((BcBigDig) (SIZE_MAX - 1)) /// Returns the max scale that is allowed. #define BC_MAX_SCALE ((BcBigDig) (BC_NUM_BIGDIG_MAX - 1)) /// Returns the max string length that is allowed. #define BC_MAX_STRING ((BcBigDig) (BC_NUM_BIGDIG_MAX - 1)) /// Returns the max identifier length that is allowed. #define BC_MAX_NAME BC_MAX_STRING /// Returns the max number size that is allowed. #define BC_MAX_NUM BC_MAX_SCALE #if BC_ENABLE_EXTRA_MATH /// Returns the max random integer that can be returned. #define BC_MAX_RAND ((BcBigDig) (((BcRand) 0) - 1)) #endif // BC_ENABLE_EXTRA_MATH /// Returns the max exponent that is allowed. #define BC_MAX_EXP ((ulong) (BC_NUM_BIGDIG_MAX)) /// Returns the max number of variables that is allowed. #define BC_MAX_VARS ((ulong) (SIZE_MAX - 1)) #if BC_ENABLE_LINE_LIB /// The size of the global buffer. #define BC_VM_BUF_SIZE (1 << 10) /// The amount of the global buffer allocated to stdin. #define BC_VM_STDIN_BUF_SIZE (BC_VM_BUF_SIZE - 1) #else // BC_ENABLE_LINE_LIB /// The size of the global buffer. #define BC_VM_BUF_SIZE (1 << 12) /// The amount of the global buffer allocated to stdout. #define BC_VM_STDOUT_BUF_SIZE (1 << 11) /// The amount of the global buffer allocated to stderr. #define BC_VM_STDERR_BUF_SIZE (1 << 10) /// The amount of the global buffer allocated to stdin. #define BC_VM_STDIN_BUF_SIZE (BC_VM_STDERR_BUF_SIZE - 1) #endif // BC_ENABLE_LINE_LIB /// The max number of temporary BcNums that can be kept. #define BC_VM_MAX_TEMPS (1 << 9) /// The capacity of the one BcNum, which is a constant. #define BC_VM_ONE_CAP (1) /** * Returns true if a BcResult is safe for garbage collection. * @param r The BcResult to test. * @return True if @a r is safe to garbage collect. */ #define BC_VM_SAFE_RESULT(r) ((r)->t >= BC_RESULT_TEMP) /// The invalid locale catalog return value. #define BC_VM_INVALID_CATALOG ((nl_catd) - 1) /** * Returns true if the *unsigned* multiplication overflows. * @param a The first operand. * @param b The second operand. * @param r The product. * @return True if the multiplication of @a a and @a b overflows. */ #define BC_VM_MUL_OVERFLOW(a, b, r) \ ((r) >= SIZE_MAX || ((a) != 0 && (r) / (a) != (b))) /// The global vm struct. This holds all of the global data besides the file /// buffers. typedef struct BcVm { /// The current status. This is volatile sig_atomic_t because it is also /// used in the signal handler. See the development manual /// (manuals/development.md#async-signal-safe-signal-handling) for more /// information. volatile sig_atomic_t status; /// Non-zero if a jump series is in progress and items should be popped off /// the jmp_bufs vector. This is volatile sig_atomic_t because it is also /// used in the signal handler. See the development manual /// (manuals/development.md#async-signal-safe-signal-handling) for more /// information. volatile sig_atomic_t sig_pop; #if !BC_ENABLE_LIBRARY /// The parser. BcParse prs; /// The program. BcProgram prog; /// A buffer for lines for stdin. BcVec line_buf; /// A buffer to hold a series of lines from stdin. Sometimes, multiple lines /// are necessary for parsing, such as a comment that spans multiple lines. BcVec buffer; /// A parser to parse read expressions. BcParse read_prs; /// A buffer for read expressions. BcVec read_buf; #endif // !BC_ENABLE_LIBRARY /// A vector of jmp_bufs for doing a jump series. This allows exception-type /// error handling, while allowing me to do cleanup on the way. BcVec jmp_bufs; /// The number of temps in the temps array. size_t temps_len; #if BC_ENABLE_LIBRARY /// The vector of contexts for the library. BcVec ctxts; /// The vector for creating strings to pass to the client. BcVec out; #if BC_ENABLE_EXTRA_MATH /// The PRNG. BcRNG rng; #endif // BC_ENABLE_EXTRA_MATH /// The current error. BclError err; /// Whether or not bcl should abort on fatal errors. bool abrt; /// Whether or not to print leading zeros. bool leading_zeroes; /// Whether or not to clamp digits that are greater than or equal to the /// current ibase. bool digit_clamp; /// The number of "references," or times that the library was initialized. unsigned int refs; #else // BC_ENABLE_LIBRARY /// A pointer to the filename of the current file. This is not owned by the /// BcVm struct. const char* file; /// The message printed when SIGINT happens. const char* sigmsg; /// Non-zero when signals are "locked." This is volatile sig_atomic_t /// because it is also used in the signal handler. See the development /// manual (manuals/development.md#async-signal-safe-signal-handling) for /// more information. volatile sig_atomic_t sig_lock; /// Non-zero when a signal has been received, but not acted on. This is /// volatile sig_atomic_t because it is also used in the signal handler. See /// the development manual /// (manuals/development.md#async-signal-safe-signal-handling) for more /// information. volatile sig_atomic_t sig; /// The length of sigmsg. uchar siglen; /// The instruction used for returning from a read() call. uchar read_ret; /// The flags field used by most macros above. uint16_t flags; /// The number of characters printed in the current line. This is used /// because bc has a limit of the number of characters it can print per /// line. uint16_t nchars; /// The length of the line we can print. The user can set this if they wish. uint16_t line_len; /// True if bc should error if expressions are encountered during option /// parsing, false otherwise. bool no_exprs; /// True if bc should exit if expresions are encountered. bool exit_exprs; /// True if EOF was encountered. bool eof; /// The mode that the program is in. uchar mode; #if BC_ENABLED /// True if keywords should not be redefined. This is only true for the /// builtin math libraries for bc. bool no_redefine; #endif // BC_ENABLED /// A vector of filenames to process. BcVec files; /// A vector of expressions to process. BcVec exprs; /// The name of the calculator under use. This is used by BC_IS_BC and /// BC_IS_DC. const char* name; /// The help text for the calculator. const char* help; #if BC_ENABLE_HISTORY /// The history data. BcHistory history; #endif // BC_ENABLE_HISTORY /// The function to call to get the next lex token. BcLexNext next; /// The function to call to parse. BcParseParse parse; /// The function to call to parse expressions. BcParseExpr expr; /// The names of the categories of errors. const char* err_ids[BC_ERR_IDX_NELEMS + BC_ENABLED]; /// The messages for each error. const char* err_msgs[BC_ERR_NELEMS]; #if BC_ENABLE_NLS /// The locale. const char* locale; #endif // BC_ENABLE_NLS #endif // BC_ENABLE_LIBRARY /// An array of maxes for the globals. BcBigDig maxes[BC_PROG_GLOBALS_LEN + BC_ENABLE_EXTRA_MATH]; /// The last base used to parse. BcBigDig last_base; /// The last power of last_base used to parse. BcBigDig last_pow; /// The last exponent of base that equals last_pow. BcBigDig last_exp; /// BC_BASE_POW - last_pow. BcBigDig last_rem; #if !BC_ENABLE_LIBRARY /// A buffer of environment arguments. This is the actual value of the /// environment variable. char* env_args_buffer; /// A vector for environment arguments after parsing. BcVec env_args; /// A BcNum set to constant 0. BcNum zero; #endif // !BC_ENABLE_LIBRARY /// A BcNum set to constant 1. BcNum one; /// A BcNum holding the max number held by a BcBigDig plus 1. BcNum max; /// A BcNum holding the max number held by a BcBigDig times 2 plus 1. BcNum max2; /// The BcDig array for max. BcDig max_num[BC_NUM_BIGDIG_LOG10]; /// The BcDig array for max2. BcDig max2_num[BC_NUM_BIGDIG_LOG10]; // The BcDig array for the one BcNum. BcDig one_num[BC_VM_ONE_CAP]; #if !BC_ENABLE_LIBRARY // The BcDig array for the zero BcNum. BcDig zero_num[BC_VM_ONE_CAP]; /// The stdout file. BcFile fout; /// The stderr file. BcFile ferr; #if BC_ENABLE_NLS /// The locale catalog. nl_catd catalog; #endif // BC_ENABLE_NLS /// A pointer to the stdin buffer. char* buf; /// The number of items in the input buffer. size_t buf_len; /// The slabs vector for constants, strings, function names, and other /// string-like things. BcVec slabs; #if BC_ENABLED /// An array of booleans for which bc keywords have been redefined if /// BC_REDEFINE_KEYWORDS is non-zero. bool redefined_kws[BC_LEX_NKWS]; #endif // BC_ENABLED #endif // !BC_ENABLE_LIBRARY BcDig* temps_buf[BC_VM_MAX_TEMPS]; #if BC_DEBUG_CODE /// The depth for BC_FUNC_ENTER and BC_FUNC_EXIT. size_t func_depth; #endif // BC_DEBUG_CODE } BcVm; /** * Print the copyright banner and help if it's non-NULL. * @param help The help message to print if it's non-NULL. */ void bc_vm_info(const char* const help); /** * The entrance point for bc/dc together. * @param argc The count of arguments. * @param argv The argument array. * @return A status. */ BcStatus -bc_vm_boot(int argc, char* argv[]); +bc_vm_boot(int argc, const char* argv[]); /** * Initializes some of the BcVm global. This is separate to make things easier * on the library code. */ void bc_vm_init(void); /** * Frees the BcVm global. */ void bc_vm_shutdown(void); /** * Add a temp to the temp array. * @param num The BcDig array to add to the temp array. */ void bc_vm_addTemp(BcDig* num); /** * Return the temp on the top of the temp stack, or NULL if there are none. * @return A temp, or NULL if none exist. */ BcDig* bc_vm_takeTemp(void); /** * Gets the top temp of the temp stack. This is separate from bc_vm_takeTemp() * to quiet a GCC warning about longjmp() clobbering in bc_num_init(). * @return A temp, or NULL if none exist. */ BcDig* bc_vm_getTemp(void); /** * Frees all temporaries. */ void bc_vm_freeTemps(void); #if !BC_ENABLE_HISTORY || BC_ENABLE_LINE_LIB || BC_ENABLE_LIBRARY /** * Erases the flush argument if history does not exist because it does not * matter if history does not exist. */ #define bc_vm_putchar(c, t) bc_vm_putchar_impl(c) #else // !BC_ENABLE_HISTORY || BC_ENABLE_LINE_LIB || BC_ENABLE_LIBRARY // This is here to satisfy a clang warning about recursive macros. #define bc_vm_putchar(c, t) bc_vm_putchar_impl(c, t) #endif // !BC_ENABLE_HISTORY || BC_ENABLE_LINE_LIB || BC_ENABLE_LIBRARY /** * Print to stdout with limited formating. * @param fmt The format string. */ void bc_vm_printf(const char* fmt, ...); /** * Puts a char into the stdout buffer. * @param c The character to put on the stdout buffer. * @param type The flush type. */ void bc_vm_putchar(int c, BcFlushType type); /** * Multiplies @a n and @a size and throws an allocation error if overflow * occurs. * @param n The number of elements. * @param size The size of each element. * @return The product of @a n and @a size. */ size_t bc_vm_arraySize(size_t n, size_t size); /** * Adds @a a and @a b and throws an error if overflow occurs. * @param a The first operand. * @param b The second operand. * @return The sum of @a a and @a b. */ size_t bc_vm_growSize(size_t a, size_t b); /** * Allocate @a n bytes and throw an allocation error if allocation fails. * @param n The bytes to allocate. * @return A pointer to the allocated memory. */ void* bc_vm_malloc(size_t n); /** * Reallocate @a ptr to be @a n bytes and throw an allocation error if * reallocation fails. * @param ptr The pointer to a memory allocation to reallocate. * @param n The bytes to allocate. * @return A pointer to the reallocated memory. */ void* bc_vm_realloc(void* ptr, size_t n); /** * Allocates space for, and duplicates, @a str. * @param str The string to allocate. * @return The allocated string. */ char* bc_vm_strdup(const char* str); /** * Reads a line from stdin into BcVm's buffer field. * @param clear True if the buffer should be cleared first, false otherwise. * @return True if a line was read, false otherwise. */ bool bc_vm_readLine(bool clear); /** * Reads a line from the command-line expressions into BcVm's buffer field. * @param clear True if the buffer should be cleared first, false otherwise. * @return True if a line was read, false otherwise. */ bool bc_vm_readBuf(bool clear); /** * A convenience and portability function for OpenBSD's pledge(). * @param promises The promises to pledge(). * @param execpromises The exec promises to pledge(). */ void bc_pledge(const char* promises, const char* execpromises); /** * Returns the value of an environment variable. * @param var The environment variable. * @return The value of the environment variable. */ char* bc_vm_getenv(const char* var); /** * Frees an environment variable value. * @param val The value to free. */ void bc_vm_getenvFree(char* val); #if BC_DEBUG_CODE /** * Start executing a jump series. * @param f The name of the function that started the jump series. */ void bc_vm_jmp(const char* f); #else // BC_DEBUG_CODE /** * Start executing a jump series. */ void bc_vm_jmp(void); #endif // BC_DEBUG_CODE #if BC_ENABLE_LIBRARY /** * Handle an error. This is the true error handler. It will start a jump series * if an error occurred. POSIX errors will not cause jumps when warnings are on * or no POSIX errors are enabled. * @param e The error. */ void bc_vm_handleError(BcErr e); /** * Handle a fatal error. * @param e The error. */ void bc_vm_fatalError(BcErr e); /** * A function to call at exit. */ void bc_vm_atexit(void); #else // BC_ENABLE_LIBRARY /** * Calculates the number of decimal digits in the argument. * @param val The value to calculate the number of decimal digits in. * @return The number of decimal digits in @a val. */ size_t bc_vm_numDigits(size_t val); #if BC_DEBUG /** * Handle an error. This is the true error handler. It will start a jump series * if an error occurred. POSIX errors will not cause jumps when warnings are on * or no POSIX errors are enabled. * @param e The error. * @param file The source file where the error occurred. * @param fline The line in the source file where the error occurred. * @param line The bc source line where the error occurred. */ void bc_vm_handleError(BcErr e, const char* file, int fline, size_t line, ...); #else // BC_DEBUG /** * Handle an error. This is the true error handler. It will start a jump series * if an error occurred. POSIX errors will not cause jumps when warnings are on * or no POSIX errors are enabled. * @param e The error. * @param line The bc source line where the error occurred. */ void bc_vm_handleError(BcErr e, size_t line, ...); #endif // BC_DEBUG /** * Handle a fatal error. * @param e The error. */ #if !BC_ENABLE_MEMCHECK BC_NORETURN #endif // !BC_ENABLE_MEMCHECK void bc_vm_fatalError(BcErr e); /** * A function to call at exit. * @param status The exit status. */ BcStatus bc_vm_atexit(BcStatus status); #endif // BC_ENABLE_LIBRARY /// A reference to the copyright header. extern const char bc_copyright[]; /// A reference to the array of default error category names. extern const char* bc_errs[]; /// A reference to the array of error category indices for each error. extern const uchar bc_err_ids[]; /// A reference to the array of default error messages. extern const char* const bc_err_msgs[]; /// A reference to the pledge() promises at start. extern const char bc_pledge_start[]; #if BC_ENABLE_HISTORY /// A reference to the end pledge() promises when using history. extern const char bc_pledge_end_history[]; #endif // BC_ENABLE_HISTORY /// A reference to the end pledge() promises when *not* using history. extern const char bc_pledge_end[]; #if !BC_ENABLE_LIBRARY /// A reference to the global data. extern BcVm* vm; /// The global data. extern BcVm vm_data; /// A reference to the global output buffers. extern char output_bufs[BC_VM_BUF_SIZE]; #endif // !BC_ENABLE_LIBRARY #endif // BC_VM_H diff --git a/manuals/dc/A.1 b/manuals/dc/A.1 index 33ecb8e2031e..d59e0fa68a58 100644 --- a/manuals/dc/A.1 +++ b/manuals/dc/A.1 @@ -1,1692 +1,1695 @@ .\" .\" SPDX-License-Identifier: BSD-2-Clause .\" .\" Copyright (c) 2018-2024 Gavin D. Howard and contributors. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions are met: .\" .\" * Redistributions of source code must retain the above copyright notice, .\" this list of conditions and the following disclaimer. .\" .\" * Redistributions in binary form must reproduce the above copyright notice, .\" this list of conditions and the following disclaimer in the documentation .\" and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" .\" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE .\" LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF .\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS .\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN .\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) .\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE .\" POSSIBILITY OF SUCH DAMAGE. .\" -.TH "DC" "1" "January 2024" "Gavin D. Howard" "General Commands Manual" +.TH "DC" "1" "August 2024" "Gavin D. Howard" "General Commands Manual" .nh .ad l .SH Name dc \- arbitrary\-precision decimal reverse\-Polish notation calculator .SH SYNOPSIS \f[B]dc\f[R] [\f[B]\-cChiPRvVx\f[R]] [\f[B]\-\-version\f[R]] [\f[B]\-\-help\f[R]] [\f[B]\-\-digit\-clamp\f[R]] [\f[B]\-\-no\-digit\-clamp\f[R]] [\f[B]\-\-interactive\f[R]] [\f[B]\-\-no\-prompt\f[R]] [\f[B]\-\-no\-read\-prompt\f[R]] [\f[B]\-\-extended\-register\f[R]] [\f[B]\-e\f[R] \f[I]expr\f[R]] [\f[B]\-\-expression\f[R]=\f[I]expr\f[R]\&...] [\f[B]\-f\f[R] \f[I]file\f[R]\&...] [\f[B]\-\-file\f[R]=\f[I]file\f[R]\&...] [\f[I]file\f[R]\&...] [\f[B]\-I\f[R] \f[I]ibase\f[R]] [\f[B]\-\-ibase\f[R]=\f[I]ibase\f[R]] [\f[B]\-O\f[R] \f[I]obase\f[R]] [\f[B]\-\-obase\f[R]=\f[I]obase\f[R]] [\f[B]\-S\f[R] \f[I]scale\f[R]] [\f[B]\-\-scale\f[R]=\f[I]scale\f[R]] [\f[B]\-E\f[R] \f[I]seed\f[R]] [\f[B]\-\-seed\f[R]=\f[I]seed\f[R]] .SH DESCRIPTION dc(1) is an arbitrary\-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. .PP If no files are given on the command\-line, then dc(1) reads from \f[B]stdin\f[R] (see the \f[B]STDIN\f[R] section). Otherwise, those files are processed, and dc(1) will then exit. .PP If a user wants to set up a standard environment, they can use \f[B]DC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). For example, if a user wants the \f[B]scale\f[R] always set to \f[B]10\f[R], they can set \f[B]DC_ENV_ARGS\f[R] to \f[B]\-e 10k\f[R], and this dc(1) will always start with a \f[B]scale\f[R] of \f[B]10\f[R]. .SH OPTIONS The following are the options that dc(1) accepts. .TP \f[B]\-C\f[R], \f[B]\-\-no\-digit\-clamp\f[R] Disables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that the value added to a number from a digit is always that digit\[cq]s value multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-c\f[R] or \f[B]\-\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-c\f[R], \f[B]\-\-digit\-clamp\f[R] Enables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-C\f[R] or \f[B]\-\-no\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-E\f[R] \f[I]seed\f[R], \f[B]\-\-seed\f[R]=\f[I]seed\f[R] Sets the builtin variable \f[B]seed\f[R] to the value \f[I]seed\f[R] assuming that \f[I]seed\f[R] is in base 10. It is a fatal error if \f[I]seed\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-e\f[R] \f[I]expr\f[R], \f[B]\-\-expression\f[R]=\f[I]expr\f[R] Evaluates \f[I]expr\f[R]. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R], whether on the command\-line or in \f[B]DC_ENV_ARGS\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-f\f[R] \f[I]file\f[R], \f[B]\-\-file\f[R]=\f[I]file\f[R] Reads in \f[I]file\f[R] and evaluates it, line by line, as though it were read through \f[B]stdin\f[R]. If expressions are also given (see above), the expressions are evaluated in the order given. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-h\f[R], \f[B]\-\-help\f[R] Prints a usage message and exits. .TP \f[B]\-I\f[R] \f[I]ibase\f[R], \f[B]\-\-ibase\f[R]=\f[I]ibase\f[R] Sets the builtin variable \f[B]ibase\f[R] to the value \f[I]ibase\f[R] assuming that \f[I]ibase\f[R] is in base 10. It is a fatal error if \f[I]ibase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-i\f[R], \f[B]\-\-interactive\f[R] Forces interactive mode. (See the \f[B]INTERACTIVE MODE\f[R] section.) .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-L\f[R], \f[B]\-\-no\-line\-length\f[R] Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets \f[B]BC_LINE_LENGTH\f[R] to \f[B]0\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-O\f[R] \f[I]obase\f[R], \f[B]\-\-obase\f[R]=\f[I]obase\f[R] Sets the builtin variable \f[B]obase\f[R] to the value \f[I]obase\f[R] assuming that \f[I]obase\f[R] is in base 10. It is a fatal error if \f[I]obase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-P\f[R], \f[B]\-\-no\-prompt\f[R] Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]DC_ENV_ARGS\f[R]. .RS .PP These options override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-R\f[R], \f[B]\-\-no\-read\-prompt\f[R] Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]BC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. .RS .PP This option does not disable the regular prompt because the read prompt is only used when the \f[B]?\f[R] command is used. .PP These options \f[I]do\f[R] override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), but only for the read prompt. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-S\f[R] \f[I]scale\f[R], \f[B]\-\-scale\f[R]=\f[I]scale\f[R] Sets the builtin variable \f[B]scale\f[R] to the value \f[I]scale\f[R] assuming that \f[I]scale\f[R] is in base 10. It is a fatal error if \f[I]scale\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-v\f[R], \f[B]\-V\f[R], \f[B]\-\-version\f[R] Print the version information (copyright header) and exits. .TP \f[B]\-x\f[R] \f[B]\-\-extended\-register\f[R] Enables extended register mode. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-z\f[R], \f[B]\-\-leading\-zeroes\f[R] Makes dc(1) print all numbers greater than \f[B]\-1\f[R] and less than \f[B]1\f[R], and not equal to \f[B]0\f[R], with a leading zero. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .PP All long options are \f[B]non\-portable extensions\f[R]. .SH STDIN If no files are given on the command\-line and no files or expressions are given by the \f[B]\-f\f[R], \f[B]\-\-file\f[R], \f[B]\-e\f[R], or \f[B]\-\-expression\f[R] options, then dc(1) reads from \f[B]stdin\f[R]. .PP However, there is a caveat to this. .PP First, \f[B]stdin\f[R] is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. .SH STDOUT Any non\-error output is written to \f[B]stdout\f[R]. In addition, if history (see the \f[B]HISTORY\f[R] section) and the prompt (see the \f[B]TTY MODE\f[R] section) are enabled, both are output to \f[B]stdout\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stdout\f[R], so if \f[B]stdout\f[R] is closed, as in \f[B]dc >&\-\f[R], it will quit with an error. This is done so that dc(1) can report problems when \f[B]stdout\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stdout\f[R] to \f[B]/dev/null\f[R]. .SH STDERR Any error output is written to \f[B]stderr\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stderr\f[R], so if \f[B]stderr\f[R] is closed, as in \f[B]dc 2>&\-\f[R], it will quit with an error. This is done so that dc(1) can exit with an error code when \f[B]stderr\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stderr\f[R] to \f[B]/dev/null\f[R]. .SH SYNTAX Each item in the input source code, either a number (see the \f[B]NUMBERS\f[R] section) or a command (see the \f[B]COMMANDS\f[R] section), is processed and executed, in order. Input is processed immediately when entered. .PP \f[B]ibase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to interpret constant numbers. It is the \[lq]input\[rq] base, or the number base used for interpreting input numbers. \f[B]ibase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]ibase\f[R] is \f[B]16\f[R]. The min allowable value for \f[B]ibase\f[R] is \f[B]2\f[R]. The max allowable value for \f[B]ibase\f[R] can be queried in dc(1) programs with the \f[B]T\f[R] command. .PP \f[B]obase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to output results. It is the \[lq]output\[rq] base, or the number base used for outputting numbers. \f[B]obase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]obase\f[R] is \f[B]DC_BASE_MAX\f[R] and can be queried with the \f[B]U\f[R] command. The min allowable value for \f[B]obase\f[R] is \f[B]0\f[R]. If \f[B]obase\f[R] is \f[B]0\f[R], values are output in scientific notation, and if \f[B]obase\f[R] is \f[B]1\f[R], values are output in engineering notation. Otherwise, values are output in the specified base. .PP Outputting in scientific and engineering notations are \f[B]non\-portable extensions\f[R]. .PP The \f[I]scale\f[R] of an expression is the number of digits in the result of the expression right of the decimal point, and \f[B]scale\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that sets the precision of any operations (with exceptions). \f[B]scale\f[R] is initially \f[B]0\f[R]. \f[B]scale\f[R] cannot be negative. The max allowable value for \f[B]scale\f[R] can be queried in dc(1) programs with the \f[B]V\f[R] command. .PP \f[B]seed\f[R] is a register containing the current seed for the pseudo\-random number generator. If the current value of \f[B]seed\f[R] is queried and stored, then if it is assigned to \f[B]seed\f[R] later, the pseudo\-random number generator is guaranteed to produce the same sequence of pseudo\-random numbers that were generated after the value of \f[B]seed\f[R] was first queried. .PP Multiple values assigned to \f[B]seed\f[R] can produce the same sequence of pseudo\-random numbers. Likewise, when a value is assigned to \f[B]seed\f[R], it is not guaranteed that querying \f[B]seed\f[R] immediately after will return the same value. In addition, the value of \f[B]seed\f[R] will change after any call to the \f[B]\[cq]\f[R] command or the \f[B]\[lq]\f[R] command that does not get receive a value of \f[B]0\f[R] or \f[B]1\f[R]. The maximum integer returned by the \f[B]\[cq]\f[R] command can be queried with the \f[B]W\f[R] command. .PP \f[B]Note\f[R]: The values returned by the pseudo\-random number generator with the \f[B]\[cq]\f[R] and \f[B]\[lq]\f[R] commands are guaranteed to \f[B]NOT\f[R] be cryptographically secure. This is a consequence of using a seeded pseudo\-random number generator. However, they \f[I]are\f[R] guaranteed to be reproducible with identical \f[B]seed\f[R] values. This means that the pseudo\-random values from dc(1) should only be used where a reproducible stream of pseudo\-random numbers is \f[I]ESSENTIAL\f[R]. In any other case, use a non\-seeded pseudo\-random number generator. .PP The pseudo\-random number generator, \f[B]seed\f[R], and all associated operations are \f[B]non\-portable extensions\f[R]. .SS Comments Comments go from \f[B]#\f[R] until, and not including, the next newline. This is a \f[B]non\-portable extension\f[R]. .SH NUMBERS Numbers are strings made up of digits, uppercase letters up to \f[B]F\f[R], and at most \f[B]1\f[R] period for a radix. Numbers can have up to \f[B]DC_NUM_MAX\f[R] digits. Uppercase letters are equal to \f[B]9\f[R] plus their position in the alphabet (i.e., \f[B]A\f[R] equals \f[B]10\f[R], or \f[B]9+1\f[R]). .PP If a digit or letter makes no sense with the current value of \f[B]ibase\f[R] (i.e., they are greater than or equal to the current value of \f[B]ibase\f[R]), then the behavior depends on the existence of the \f[B]\-c\f[R]/\f[B]\-\-digit\-clamp\f[R] or \f[B]\-C\f[R]/\f[B]\-\-no\-digit\-clamp\f[R] options (see the \f[B]OPTIONS\f[R] section), the existence and setting of the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or the default, which can be queried with the \f[B]\-h\f[R]/\f[B]\-\-help\f[R] option. .PP If clamping is off, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are not changed. Instead, their given value is multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*A+3\[ha]0*B\f[R], which is \f[B]3\f[R] times \f[B]10\f[R] plus \f[B]11\f[R], or \f[B]41\f[R]. .PP If clamping is on, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are set to the value of the highest valid digit in \f[B]ibase\f[R] before being multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*2+3\[ha]0*2\f[R], which is \f[B]3\f[R] times \f[B]2\f[R] plus \f[B]2\f[R], or \f[B]8\f[R]. .PP There is one exception to clamping: single\-character numbers (i.e., \f[B]A\f[R] alone). Such numbers are never clamped and always take the value they would have in the highest possible \f[B]ibase\f[R]. This means that \f[B]A\f[R] alone always equals decimal \f[B]10\f[R] and \f[B]Z\f[R] alone always equals decimal \f[B]35\f[R]. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current \f[B]ibase\f[R] (with the \f[B]i\f[R] command) regardless of the current value of \f[B]ibase\f[R]. .PP If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for \f[B]A\f[R], use \f[B]0A\f[R]. .PP In addition, dc(1) accepts numbers in scientific notation. These have the form \f[B]e\f[R]. The exponent (the portion after the \f[B]e\f[R]) must be an integer. An example is \f[B]1.89237e9\f[R], which is equal to \f[B]1892370000\f[R]. Negative exponents are also allowed, so \f[B]4.2890e_3\f[R] is equal to \f[B]0.0042890\f[R]. .PP \f[B]WARNING\f[R]: Both the number and the exponent in scientific notation are interpreted according to the current \f[B]ibase\f[R], but the number is still multiplied by \f[B]10\[ha]exponent\f[R] regardless of the current \f[B]ibase\f[R]. For example, if \f[B]ibase\f[R] is \f[B]16\f[R] and dc(1) is given the number string \f[B]FFeA\f[R], the resulting decimal number will be \f[B]2550000000000\f[R], and if dc(1) is given the number string \f[B]10e_4\f[R], the resulting decimal number will be \f[B]0.0016\f[R]. .PP Accepting input as scientific notation is a \f[B]non\-portable extension\f[R]. .SH COMMANDS The valid commands are listed below. .SS Printing These commands are used for printing. .PP Note that both scientific notation and engineering notation are available for printing numbers. Scientific notation is activated by assigning \f[B]0\f[R] to \f[B]obase\f[R] using \f[B]0o\f[R], and engineering notation is activated by assigning \f[B]1\f[R] to \f[B]obase\f[R] using \f[B]1o\f[R]. To deactivate them, just assign a different value to \f[B]obase\f[R]. .PP Printing numbers in scientific notation and/or engineering notation is a \f[B]non\-portable extension\f[R]. .TP \f[B]p\f[R] Prints the value on top of the stack, whether number or string, and prints a newline after. .RS .PP This does not alter the stack. .RE .TP \f[B]n\f[R] Prints the value on top of the stack, whether number or string, and pops it off of the stack. .TP \f[B]P\f[R] Pops a value off the stack. .RS .PP If the value is a number, it is truncated and the absolute value of the result is printed as though \f[B]obase\f[R] is \f[B]256\f[R] and each digit is interpreted as an 8\-bit ASCII character, making it a byte stream. .PP If the value is a string, it is printed without a trailing newline. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]f\f[R] Prints the entire contents of the stack, in order from newest to oldest, without altering anything. .RS .PP Users should use this command when they get lost. .RE .SS Arithmetic These are the commands used for arithmetic. .TP \f[B]+\f[R] The top two values are popped off the stack, added, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]\-\f[R] The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]*\f[R] The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If \f[B]a\f[R] is the \f[I]scale\f[R] of the first expression and \f[B]b\f[R] is the \f[I]scale\f[R] of the second expression, the \f[I]scale\f[R] of the result is equal to \f[B]min(a+b,max(scale,a,b))\f[R] where \f[B]min()\f[R] and \f[B]max()\f[R] return the obvious values. .TP \f[B]/\f[R] The top two values are popped off the stack, divided, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]%\f[R] The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. .RS .PP Remaindering is equivalent to 1) Computing \f[B]a/b\f[R] to current \f[B]scale\f[R], and 2) Using the result of step 1 to calculate \f[B]a\-(a/b)*b\f[R] to \f[I]scale\f[R] \f[B]max(scale+scale(b),scale(a))\f[R]. .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]\[ti]\f[R] The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to \f[B]x y / x y %\f[R] except that \f[B]x\f[R] and \f[B]y\f[R] are only evaluated once. .RS .PP The first value popped off of the stack must be non\-zero. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[ha]\f[R] The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non\-zero. .RE .TP \f[B]v\f[R] The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The value popped off of the stack must be non\-negative. .RE .TP \f[B]_\f[R] If this command \f[I]immediately\f[R] precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. .RS .PP Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]b\f[R] The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]|\f[R] The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. .RS .PP The first value popped is used as the reduction modulus and must be an integer and non\-zero. The second value popped is used as the exponent and must be an integer and non\-negative. The third value popped is the base and must be an integer. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]$\f[R] The top value is popped off the stack and copied, and the copy is truncated and pushed onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[at]\f[R] The top two values are popped off the stack, and the precision of the second is set to the value of the first, whether by truncation or extension. .RS .PP The first value popped off of the stack must be an integer and non\-negative. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]H\f[R] The top two values are popped off the stack, and the second is shifted left (radix shifted right) to the value of the first. .RS .PP The first value popped off of the stack must be an integer and non\-negative. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]h\f[R] The top two values are popped off the stack, and the second is shifted right (radix shifted left) to the value of the first. .RS .PP The first value popped off of the stack must be an integer and non\-negative. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]G\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if they are equal, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]N\f[R] The top value is popped off of the stack, and if it a \f[B]0\f[R], a \f[B]1\f[R] is pushed; otherwise, a \f[B]0\f[R] is pushed. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B](\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]{\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B])\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]}\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]M\f[R] The top two values are popped off of the stack. If they are both non\-zero, a \f[B]1\f[R] is pushed onto the stack. If either of them is zero, or both of them are, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]&&\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]m\f[R] The top two values are popped off of the stack. If at least one of them is non\-zero, a \f[B]1\f[R] is pushed onto the stack. If both of them are zero, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]||\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Pseudo\-Random Number Generator dc(1) has a built\-in pseudo\-random number generator. These commands query the pseudo\-random number generator. (See Parameters for more information about the \f[B]seed\f[R] value that controls the pseudo\-random number generator.) .PP The pseudo\-random number generator is guaranteed to \f[B]NOT\f[R] be cryptographically secure. .TP \f[B]\[cq]\f[R] Generates an integer between 0 and \f[B]DC_RAND_MAX\f[R], inclusive (see the \f[B]LIMITS\f[R] section). .RS .PP The generated integer is made as unbiased as possible, subject to the limitations of the pseudo\-random number generator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[lq]\f[R] Pops a value off of the stack, which is used as an \f[B]exclusive\f[R] upper bound on the integer that will be generated. If the bound is negative or is a non\-integer, an error is raised, and dc(1) resets (see the \f[B]RESET\f[R] section) while \f[B]seed\f[R] remains unchanged. If the bound is larger than \f[B]DC_RAND_MAX\f[R], the higher bound is honored by generating several pseudo\-random integers, multiplying them by appropriate powers of \f[B]DC_RAND_MAX+1\f[R], and adding them together. Thus, the size of integer that can be generated with this command is unbounded. Using this command will change the value of \f[B]seed\f[R], unless the operand is \f[B]0\f[R] or \f[B]1\f[R]. In that case, \f[B]0\f[R] is pushed onto the stack, and \f[B]seed\f[R] is \f[I]not\f[R] changed. .RS .PP The generated integer is made as unbiased as possible, subject to the limitations of the pseudo\-random number generator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Stack Control These commands control the stack. .TP \f[B]c\f[R] Removes all items from (\[lq]clears\[rq]) the stack. .TP \f[B]d\f[R] Copies the item on top of the stack (\[lq]duplicates\[rq]) and pushes the copy onto the stack. .TP \f[B]r\f[R] Swaps (\[lq]reverses\[rq]) the two top items on the stack. .TP \f[B]R\f[R] Pops (\[lq]removes\[rq]) the top value from the stack. .SS Register Control These commands control registers (see the \f[B]REGISTERS\f[R] section). .TP \f[B]s\f[R]\f[I]r\f[R] Pops the value off the top of the stack and stores it into register \f[I]r\f[R]. .TP \f[B]l\f[R]\f[I]r\f[R] Copies the value in register \f[I]r\f[R] and pushes it onto the stack. This does not alter the contents of \f[I]r\f[R]. .TP \f[B]S\f[R]\f[I]r\f[R] Pops the value off the top of the (main) stack and pushes it onto the stack of register \f[I]r\f[R]. The previous value of the register becomes inaccessible. .TP \f[B]L\f[R]\f[I]r\f[R] Pops the value off the top of the stack for register \f[I]r\f[R] and push it onto the main stack. The previous value in the stack for register \f[I]r\f[R], if any, is now accessible via the \f[B]l\f[R]\f[I]r\f[R] command. .SS Parameters These commands control the values of \f[B]ibase\f[R], \f[B]obase\f[R], \f[B]scale\f[R], and \f[B]seed\f[R]. Also see the \f[B]SYNTAX\f[R] section. .TP \f[B]i\f[R] Pops the value off of the top of the stack and uses it to set \f[B]ibase\f[R], which must be between \f[B]2\f[R] and \f[B]16\f[R], inclusive. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]o\f[R] Pops the value off of the top of the stack and uses it to set \f[B]obase\f[R], which must be between \f[B]0\f[R] and \f[B]DC_BASE_MAX\f[R], inclusive (see the \f[B]LIMITS\f[R] section and the \f[B]NUMBERS\f[R] section). .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]k\f[R] Pops the value off of the top of the stack and uses it to set \f[B]scale\f[R], which must be non\-negative. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]j\f[R] Pops the value off of the top of the stack and uses it to set \f[B]seed\f[R]. The meaning of \f[B]seed\f[R] is dependent on the current pseudo\-random number generator but is guaranteed to not change except for new major versions. .RS .PP The \f[I]scale\f[R] and sign of the value may be significant. .PP If a previously used \f[B]seed\f[R] value is used again, the pseudo\-random number generator is guaranteed to produce the same sequence of pseudo\-random numbers as it did when the \f[B]seed\f[R] value was previously used. .PP The exact value assigned to \f[B]seed\f[R] is not guaranteed to be returned if the \f[B]J\f[R] command is used. However, if \f[B]seed\f[R] \f[I]does\f[R] return a different value, both values, when assigned to \f[B]seed\f[R], are guaranteed to produce the same sequence of pseudo\-random numbers. This means that certain values assigned to \f[B]seed\f[R] will not produce unique sequences of pseudo\-random numbers. .PP There is no limit to the length (number of significant decimal digits) or \f[I]scale\f[R] of the value that can be assigned to \f[B]seed\f[R]. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]I\f[R] Pushes the current value of \f[B]ibase\f[R] onto the main stack. .TP \f[B]O\f[R] Pushes the current value of \f[B]obase\f[R] onto the main stack. .TP \f[B]K\f[R] Pushes the current value of \f[B]scale\f[R] onto the main stack. .TP \f[B]J\f[R] Pushes the current value of \f[B]seed\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]T\f[R] Pushes the maximum allowable value of \f[B]ibase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]U\f[R] Pushes the maximum allowable value of \f[B]obase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]V\f[R] Pushes the maximum allowable value of \f[B]scale\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]W\f[R] Pushes the maximum (inclusive) integer that can be generated with the \f[B]\[cq]\f[R] pseudo\-random number generator command. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Strings The following commands control strings. .PP dc(1) can work with both numbers and strings, and registers (see the \f[B]REGISTERS\f[R] section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. .PP While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. .PP Strings can also be executed as macros. For example, if the string \f[B][1pR]\f[R] is executed as a macro, then the code \f[B]1pR\f[R] is executed, meaning that the \f[B]1\f[R] will be printed with a newline after and then popped from the stack. .TP \f[B][\f[R]\f[I]characters\f[R]\f[B]]\f[R] Makes a string containing \f[I]characters\f[R] and pushes it onto the stack. .RS .PP If there are brackets (\f[B][\f[R] and \f[B]]\f[R]) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (\f[B]\[rs]\f[R]) character. .PP If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. .RE .TP \f[B]a\f[R] The value on top of the stack is popped. .RS .PP If it is a number, it is truncated and its absolute value is taken. The result mod \f[B]256\f[R] is calculated. If that result is \f[B]0\f[R], push an empty string; otherwise, push a one\-character string where the character is the result of the mod interpreted as an ASCII character. .PP If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one\-character string. The new string is then pushed onto the stack. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]x\f[R] Pops a value off of the top of the stack. .RS .PP If it is a number, it is pushed back onto the stack. .PP If it is a string, it is executed as a macro. .PP This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. .RE .TP \f[B]>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP For example, \f[B]0 1>a\f[R] will execute the contents of register \f[B]a\f[R], and \f[B]1 0>a\f[R] will not. .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]?\f[R] Reads a line from the \f[B]stdin\f[R] and executes it. This is to allow macros to request input from users. .TP \f[B]q\f[R] During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. .TP \f[B]Q\f[R] Pops a value from the stack which must be non\-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. .TP \f[B],\f[R] Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the \f[B]Q\f[R] command, so the sequence \f[B],Q\f[R] will make dc(1) exit. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Status These commands query status of the stack or its top value. .TP \f[B]Z\f[R] Pops a value off of the stack. .RS .PP If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push \f[B]1\f[R] if the argument is \f[B]0\f[R] with no decimal places. .PP If it is a string, pushes the number of characters the string has. .RE .TP \f[B]X\f[R] Pops a value off of the stack. .RS .PP If it is a number, pushes the \f[I]scale\f[R] of the value onto the stack. .PP If it is a string, pushes \f[B]0\f[R]. .RE .TP \f[B]u\f[R] Pops one value off of the stack. If the value is a number, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a string), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]t\f[R] Pops one value off of the stack. If the value is a string, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a number), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]z\f[R] Pushes the current depth of the stack (before execution of this command) onto the stack. .TP \f[B]y\f[R]\f[I]r\f[R] Pushes the current stack depth of the register \f[I]r\f[R] onto the main stack. .RS .PP Because each register has a depth of \f[B]1\f[R] (with the value \f[B]0\f[R] in the top item) when dc(1) starts, dc(1) requires that each register\[cq]s stack must always have at least one item; dc(1) will give an error and reset otherwise (see the \f[B]RESET\f[R] section). This means that this command will never push \f[B]0\f[R]. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Arrays These commands manipulate arrays. .TP \f[B]:\f[R]\f[I]r\f[R] Pops the top two values off of the stack. The second value will be stored in the array \f[I]r\f[R] (see the \f[B]REGISTERS\f[R] section), indexed by the first value. .TP \f[B];\f[R]\f[I]r\f[R] Pops the value on top of the stack and uses it as an index into the array \f[I]r\f[R]. The selected value is then pushed onto the stack. .TP \f[B]Y\f[R]\f[I]r\f[R] Pushes the length of the array \f[I]r\f[R] onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter \f[B]g\f[R]. Only the characters below are allowed after the character \f[B]g\f[R]; any other character produces a parse error (see the \f[B]ERRORS\f[R] section). .TP \f[B]gl\f[R] Pushes the line length set by \f[B]DC_LINE_LENGTH\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) onto the stack. .TP \f[B]gx\f[R] Pushes \f[B]1\f[R] onto the stack if extended register mode is on, \f[B]0\f[R] otherwise. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .TP \f[B]gz\f[R] Pushes \f[B]0\f[R] onto the stack if the leading zero setting has not been enabled with the \f[B]\-z\f[R] or \f[B]\-\-leading\-zeroes\f[R] options (see the \f[B]OPTIONS\f[R] section), non\-zero otherwise. .SH REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) .PP Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (\f[B]0\f[R]) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. .PP In non\-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (\f[B]`\[rs]n'\f[R]) and a left bracket (\f[B]`['\f[R]); it is a parse error for a newline or a left bracket to be used as a register name. .SS Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. .PP If extended register mode is enabled (\f[B]\-x\f[R] or \f[B]\-\-extended\-register\f[R] command\-line arguments are given), then normal single character registers are used \f[I]unless\f[R] the character immediately following a command that needs a register name is a space (according to \f[B]isspace()\f[R]) and not a newline (\f[B]`\[rs]n'\f[R]). .PP In that case, the register name is found according to the regex \f[B][a\-z][a\-z0\-9_]*\f[R] (like bc(1) identifiers), and it is a parse error if the next non\-space characters do not match that regex. .SH RESET When dc(1) encounters an error or a signal that it has a non\-default handler for, it resets. This means that several things happen. .PP First, any macros that are executing are stopped and popped off the -stack. +execution stack. The behavior is not unlike that of exceptions in programming languages. Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. .PP +However, the stack of values is \f[I]not\f[R] cleared; in interactive +mode, users can inspect the stack and manipulate it. +.PP Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the \f[B]EXIT STATUS\f[R] section), it asks for more input; otherwise, it exits with the appropriate return code. .SH PERFORMANCE Most dc(1) implementations use \f[B]char\f[R] types to calculate the value of \f[B]1\f[R] decimal digit at a time, but that can be slow. This dc(1) does something different. .PP It uses large integers to calculate more than \f[B]1\f[R] decimal digit at a time. If built in a environment where \f[B]DC_LONG_BIT\f[R] (see the \f[B]LIMITS\f[R] section) is \f[B]64\f[R], then each integer has \f[B]9\f[R] decimal digits. If built in an environment where \f[B]DC_LONG_BIT\f[R] is \f[B]32\f[R] then each integer has \f[B]4\f[R] decimal digits. This value (the number of decimal digits per large integer) is called \f[B]DC_BASE_DIGS\f[R]. .PP In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of \f[B]DC_LONG_BIT\f[R], but is always at least twice as large as the integer type used to store digits. .SH LIMITS The following are the limits on dc(1): .TP \f[B]DC_LONG_BIT\f[R] The number of bits in the \f[B]long\f[R] type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the \f[B]PERFORMANCE\f[R] section). .TP \f[B]DC_BASE_DIGS\f[R] The number of decimal digits per large integer (see the \f[B]PERFORMANCE\f[R] section). Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_POW\f[R] The max decimal number that each large integer can store (see \f[B]DC_BASE_DIGS\f[R]) plus \f[B]1\f[R]. Depends on \f[B]DC_BASE_DIGS\f[R]. .TP \f[B]DC_OVERFLOW_MAX\f[R] The max number that the overflow type (see the \f[B]PERFORMANCE\f[R] section) can hold. Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_MAX\f[R] The maximum output base. Set at \f[B]DC_BASE_POW\f[R]. .TP \f[B]DC_DIM_MAX\f[R] The maximum size of arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .TP \f[B]DC_SCALE_MAX\f[R] The maximum \f[B]scale\f[R]. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_STRING_MAX\f[R] The maximum length of strings. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NAME_MAX\f[R] The maximum length of identifiers. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NUM_MAX\f[R] The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_RAND_MAX\f[R] The maximum integer (inclusive) returned by the \f[B]\[cq]\f[R] command, if dc(1). Set at \f[B]2\[ha]DC_LONG_BIT\-1\f[R]. .TP Exponent The maximum allowable exponent (positive or negative). Set at \f[B]DC_OVERFLOW_MAX\f[R]. .TP Number of vars The maximum number of vars/arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .PP These limits are meant to be effectively non\-existent; the limits are so large (at least on 64\-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. .SH ENVIRONMENT VARIABLES As \f[B]non\-portable extensions\f[R], dc(1) recognizes the following environment variables: .TP \f[B]DC_ENV_ARGS\f[R] This is another way to give command\-line arguments to dc(1). They should be in the same format as all other command\-line arguments. These are always processed first, so any files given in \f[B]DC_ENV_ARGS\f[R] will be processed before arguments and files given on the command\-line. This gives the user the ability to set up \[lq]standard\[rq] options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the \f[B]\-e\f[R] option to set \f[B]scale\f[R] to a value other than \f[B]0\f[R]. .RS .PP The code that parses \f[B]DC_ENV_ARGS\f[R] will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string \f[B]\[lq]/home/gavin/some dc file.dc\[rq]\f[R] will be correctly parsed, but the string \f[B]\[lq]/home/gavin/some \[dq]dc\[dq] file.dc\[rq]\f[R] will include the backslashes. .PP The quote parsing will handle either kind of quotes, \f[B]\[cq]\f[R] or \f[B]\[lq]\f[R]. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in \f[B]\[lq]some `dc' file.dc\[rq]\f[R], and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in \f[B]DC_ENV_ARGS\f[R] is not supported due to the complexity of the parsing, though such files are still supported on the command\-line where the parsing is done by the shell. .RE .TP \f[B]DC_LINE_LENGTH\f[R] If this environment variable exists and contains an integer that is greater than \f[B]1\f[R] and is less than \f[B]UINT16_MAX\f[R] (\f[B]2\[ha]16\-1\f[R]), dc(1) will output lines to that length, including the backslash newline combo. The default line length is \f[B]70\f[R]. .RS .PP The special value of \f[B]0\f[R] will disable line length checking and print numbers without regard to line length and without backslashes and newlines. .RE .TP \f[B]DC_SIGINT_RESET\f[R] If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), then this environment variable has no effect because dc(1) exits on \f[B]SIGINT\f[R] when not in interactive mode. .RS .PP However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) reset on \f[B]SIGINT\f[R], rather than exit, and zero makes dc(1) exit. If this environment variable exists and is \f[I]not\f[R] an integer, then dc(1) will exit on \f[B]SIGINT\f[R]. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_TTY_MODE\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non\-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_PROMPT\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) use a prompt, and zero or a non\-integer makes dc(1) not use a prompt. If this environment variable does not exist and \f[B]DC_TTY_MODE\f[R] does, then the value of the \f[B]DC_TTY_MODE\f[R] environment variable is used. .PP This environment variable and the \f[B]DC_TTY_MODE\f[R] environment variable override the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_EXPR_EXIT\f[R] If any expressions or expression files are given on the command\-line with \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R], then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. .RS .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_DIGIT_CLAMP\f[R] When parsing numbers and if this environment variable exists and contains an integer, a non\-zero value makes dc(1) clamp digits that are greater than or equal to the current \f[B]ibase\f[R] so that all such digits are considered equal to the \f[B]ibase\f[R] minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the \f[B]ibase\f[R]. .RS .PP This never applies to single\-digit numbers, as per the bc(1) standard (see the \f[B]STANDARDS\f[R] section). .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .SH EXIT STATUS dc(1) returns the following exit statuses: .TP \f[B]0\f[R] No error. .TP \f[B]1\f[R] A math error occurred. This follows standard practice of using \f[B]1\f[R] for expected errors, since math errors will happen in the process of normal execution. .RS .PP Math errors include divide by \f[B]0\f[R], taking the square root of a negative number, using a negative number as a bound for the pseudo\-random number generator, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non\-integer where an integer is required. .PP Converting to a hardware integer happens for the second operand of the power (\f[B]\[ha]\f[R]), places (\f[B]\[at]\f[R]), left shift (\f[B]H\f[R]), and right shift (\f[B]h\f[R]) operators. .RE .TP \f[B]2\f[R] A parse error occurred. .RS .PP Parse errors include unexpected \f[B]EOF\f[R], using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. .RE .TP \f[B]3\f[R] A runtime error occurred. .RS .PP Runtime errors include assigning an invalid number to any global (\f[B]ibase\f[R], \f[B]obase\f[R], or \f[B]scale\f[R]), giving a bad expression to a \f[B]read()\f[R] call, calling \f[B]read()\f[R] inside of a \f[B]read()\f[R] call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. .RE .TP \f[B]4\f[R] A fatal error occurred. .RS .PP Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command\-line options. .RE .PP The exit status \f[B]4\f[R] is special; when a fatal error occurs, dc(1) always exits and returns \f[B]4\f[R], no matter what mode dc(1) is in. .PP The other statuses will only be returned when dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), since dc(1) resets its state (see the \f[B]RESET\f[R] section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .PP These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .SH INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non\-interactive mode. Interactive mode is turned on automatically when both \f[B]stdin\f[R] and \f[B]stdout\f[R] are hooked to a terminal, but the \f[B]\-i\f[R] flag and \f[B]\-\-interactive\f[R] option can turn it on in other situations. .PP In interactive mode, dc(1) attempts to recover from errors (see the \f[B]RESET\f[R] section), and in normal execution, flushes \f[B]stdout\f[R] as soon as execution is done for the current input. dc(1) may also reset on \f[B]SIGINT\f[R] instead of exit, depending on the contents of, or default for, the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .SH TTY MODE If \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY, then \[lq]TTY mode\[rq] is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. .PP If there is the environment variable \f[B]DC_TTY_MODE\f[R] in the environment (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then if that environment variable contains a non\-zero integer, dc(1) will turn on TTY mode when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY. If the \f[B]DC_TTY_MODE\f[R] environment variable exists but is \f[I]not\f[R] a non\-zero integer, then dc(1) will not turn TTY mode on. .PP If the environment variable \f[B]DC_TTY_MODE\f[R] does \f[I]not\f[R] exist, the default setting is used. The default setting can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the \f[B]STANDARDS\f[R] section), and interactive mode requires only \f[B]stdin\f[R] and \f[B]stdout\f[R] to be connected to a terminal. .SS Command\-Line History Command\-line history is only enabled if TTY mode is, i.e., that \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are connected to a TTY and the \f[B]DC_TTY_MODE\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and its default do not disable TTY mode. See the \f[B]COMMAND LINE HISTORY\f[R] section for more information. .SS Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: \f[B]DC_PROMPT\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP If the environment variable \f[B]DC_PROMPT\f[R] exists and is a non\-zero integer, then the prompt is turned on when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are connected to a TTY and the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options were not used. The read prompt will be turned on under the same conditions, except that the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options must also not be used. .PP However, if \f[B]DC_PROMPT\f[R] does not exist, the prompt can be enabled or disabled with the \f[B]DC_TTY_MODE\f[R] environment variable, the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options, and the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options. See the \f[B]ENVIRONMENT VARIABLES\f[R] and \f[B]OPTIONS\f[R] sections for more details. .SH SIGNAL HANDLING Sending a \f[B]SIGINT\f[R] will cause dc(1) to do one of two things. .PP If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), or the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or its default, is either not an integer or it is zero, dc(1) will exit. .PP However, if dc(1) is in interactive mode, and the \f[B]DC_SIGINT_RESET\f[R] or its default is an integer and non\-zero, then dc(1) will stop executing the current input and reset (see the \f[B]RESET\f[R] section) upon receiving a \f[B]SIGINT\f[R]. .PP Note that \[lq]current input\[rq] can mean one of two things. If dc(1) is processing input from \f[B]stdin\f[R] in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from \f[B]stdin\f[R] if no other file exists. .PP This means that if a \f[B]SIGINT\f[R] is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. .PP \f[B]SIGTERM\f[R] and \f[B]SIGQUIT\f[R] cause dc(1) to clean up and exit, and it uses the default handler for all other signals. The one exception is \f[B]SIGHUP\f[R]; in that case, and only when dc(1) is in TTY mode (see the \f[B]TTY MODE\f[R] section), a \f[B]SIGHUP\f[R] will cause dc(1) to clean up and exit. .SH COMMAND LINE HISTORY dc(1) supports interactive command\-line editing. .PP If dc(1) can be in TTY mode (see the \f[B]TTY MODE\f[R] section), history can be enabled. This means that command\-line history can only be enabled when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY. .PP Like TTY mode itself, it can be turned on or off with the environment variable \f[B]DC_TTY_MODE\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP \f[B]Note\f[R]: tabs are converted to 8 spaces. .SH LOCALES This dc(1) ships with support for adding error messages for different locales and thus, supports \f[B]LC_MESSAGES\f[R]. .SH SEE ALSO bc(1) .SH STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1\-2017 (\[lq]POSIX.1\-2017\[rq]) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . .SH BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . .SH AUTHOR Gavin D. Howard \c .MT gavin@gavinhoward.com .ME \c \ and contributors. diff --git a/manuals/dc/A.1.md b/manuals/dc/A.1.md index 613f98f76814..ad0c59934fd1 100644 --- a/manuals/dc/A.1.md +++ b/manuals/dc/A.1.md @@ -1,1526 +1,1529 @@ # Name dc - arbitrary-precision decimal reverse-Polish notation calculator # SYNOPSIS **dc** [**-cChiPRvVx**] [**-\-version**] [**-\-help**] [**-\-digit-clamp**] [**-\-no-digit-clamp**] [**-\-interactive**] [**-\-no-prompt**] [**-\-no-read-prompt**] [**-\-extended-register**] [**-e** *expr*] [**-\-expression**=*expr*...] [**-f** *file*...] [**-\-file**=*file*...] [*file*...] [**-I** *ibase*] [**-\-ibase**=*ibase*] [**-O** *obase*] [**-\-obase**=*obase*] [**-S** *scale*] [**-\-scale**=*scale*] [**-E** *seed*] [**-\-seed**=*seed*] # DESCRIPTION dc(1) is an arbitrary-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. If no files are given on the command-line, then dc(1) reads from **stdin** (see the **STDIN** section). Otherwise, those files are processed, and dc(1) will then exit. If a user wants to set up a standard environment, they can use **DC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). For example, if a user wants the **scale** always set to **10**, they can set **DC_ENV_ARGS** to **-e 10k**, and this dc(1) will always start with a **scale** of **10**. # OPTIONS The following are the options that dc(1) accepts. **-C**, **-\-no-digit-clamp** : Disables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that the value added to a number from a digit is always that digit's value multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-c** or **-\-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-c**, **-\-digit-clamp** : Enables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-C** or **-\-no-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-E** *seed*, **-\-seed**=*seed* : Sets the builtin variable **seed** to the value *seed* assuming that *seed* is in base 10. It is a fatal error if *seed* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-e** *expr*, **-\-expression**=*expr* : Evaluates *expr*. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**, whether on the command-line or in **DC_ENV_ARGS**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-f** *file*, **-\-file**=*file* : Reads in *file* and evaluates it, line by line, as though it were read through **stdin**. If expressions are also given (see above), the expressions are evaluated in the order given. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-h**, **-\-help** : Prints a usage message and exits. **-I** *ibase*, **-\-ibase**=*ibase* : Sets the builtin variable **ibase** to the value *ibase* assuming that *ibase* is in base 10. It is a fatal error if *ibase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-i**, **-\-interactive** : Forces interactive mode. (See the **INTERACTIVE MODE** section.) This is a **non-portable extension**. **-L**, **-\-no-line-length** : Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets **BC_LINE_LENGTH** to **0** (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-O** *obase*, **-\-obase**=*obase* : Sets the builtin variable **obase** to the value *obase* assuming that *obase* is in base 10. It is a fatal error if *obase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-P**, **-\-no-prompt** : Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in **DC_ENV_ARGS**. These options override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-R**, **-\-no-read-prompt** : Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in **BC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. This option does not disable the regular prompt because the read prompt is only used when the **?** command is used. These options *do* override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section), but only for the read prompt. This is a **non-portable extension**. **-S** *scale*, **-\-scale**=*scale* : Sets the builtin variable **scale** to the value *scale* assuming that *scale* is in base 10. It is a fatal error if *scale* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-v**, **-V**, **-\-version** : Print the version information (copyright header) and exits. **-x** **-\-extended-register** : Enables extended register mode. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. This is a **non-portable extension**. **-z**, **-\-leading-zeroes** : Makes dc(1) print all numbers greater than **-1** and less than **1**, and not equal to **0**, with a leading zero. This is a **non-portable extension**. All long options are **non-portable extensions**. # STDIN If no files are given on the command-line and no files or expressions are given by the **-f**, **-\-file**, **-e**, or **-\-expression** options, then dc(1) reads from **stdin**. However, there is a caveat to this. First, **stdin** is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. # STDOUT Any non-error output is written to **stdout**. In addition, if history (see the **HISTORY** section) and the prompt (see the **TTY MODE** section) are enabled, both are output to **stdout**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stdout**, so if **stdout** is closed, as in **dc >&-**, it will quit with an error. This is done so that dc(1) can report problems when **stdout** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stdout** to **/dev/null**. # STDERR Any error output is written to **stderr**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stderr**, so if **stderr** is closed, as in **dc 2>&-**, it will quit with an error. This is done so that dc(1) can exit with an error code when **stderr** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stderr** to **/dev/null**. # SYNTAX Each item in the input source code, either a number (see the **NUMBERS** section) or a command (see the **COMMANDS** section), is processed and executed, in order. Input is processed immediately when entered. **ibase** is a register (see the **REGISTERS** section) that determines how to interpret constant numbers. It is the "input" base, or the number base used for interpreting input numbers. **ibase** is initially **10**. The max allowable value for **ibase** is **16**. The min allowable value for **ibase** is **2**. The max allowable value for **ibase** can be queried in dc(1) programs with the **T** command. **obase** is a register (see the **REGISTERS** section) that determines how to output results. It is the "output" base, or the number base used for outputting numbers. **obase** is initially **10**. The max allowable value for **obase** is **DC_BASE_MAX** and can be queried with the **U** command. The min allowable value for **obase** is **0**. If **obase** is **0**, values are output in scientific notation, and if **obase** is **1**, values are output in engineering notation. Otherwise, values are output in the specified base. Outputting in scientific and engineering notations are **non-portable extensions**. The *scale* of an expression is the number of digits in the result of the expression right of the decimal point, and **scale** is a register (see the **REGISTERS** section) that sets the precision of any operations (with exceptions). **scale** is initially **0**. **scale** cannot be negative. The max allowable value for **scale** can be queried in dc(1) programs with the **V** command. **seed** is a register containing the current seed for the pseudo-random number generator. If the current value of **seed** is queried and stored, then if it is assigned to **seed** later, the pseudo-random number generator is guaranteed to produce the same sequence of pseudo-random numbers that were generated after the value of **seed** was first queried. Multiple values assigned to **seed** can produce the same sequence of pseudo-random numbers. Likewise, when a value is assigned to **seed**, it is not guaranteed that querying **seed** immediately after will return the same value. In addition, the value of **seed** will change after any call to the **'** command or the **"** command that does not get receive a value of **0** or **1**. The maximum integer returned by the **'** command can be queried with the **W** command. **Note**: The values returned by the pseudo-random number generator with the **'** and **"** commands are guaranteed to **NOT** be cryptographically secure. This is a consequence of using a seeded pseudo-random number generator. However, they *are* guaranteed to be reproducible with identical **seed** values. This means that the pseudo-random values from dc(1) should only be used where a reproducible stream of pseudo-random numbers is *ESSENTIAL*. In any other case, use a non-seeded pseudo-random number generator. The pseudo-random number generator, **seed**, and all associated operations are **non-portable extensions**. ## Comments Comments go from **#** until, and not including, the next newline. This is a **non-portable extension**. # NUMBERS Numbers are strings made up of digits, uppercase letters up to **F**, and at most **1** period for a radix. Numbers can have up to **DC_NUM_MAX** digits. Uppercase letters are equal to **9** plus their position in the alphabet (i.e., **A** equals **10**, or **9+1**). If a digit or letter makes no sense with the current value of **ibase** (i.e., they are greater than or equal to the current value of **ibase**), then the behavior depends on the existence of the **-c**/**-\-digit-clamp** or **-C**/**-\-no-digit-clamp** options (see the **OPTIONS** section), the existence and setting of the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section), or the default, which can be queried with the **-h**/**-\-help** option. If clamping is off, then digits or letters that are greater than or equal to the current value of **ibase** are not changed. Instead, their given value is multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*A+3\^0\*B**, which is **3** times **10** plus **11**, or **41**. If clamping is on, then digits or letters that are greater than or equal to the current value of **ibase** are set to the value of the highest valid digit in **ibase** before being multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*2+3\^0\*2**, which is **3** times **2** plus **2**, or **8**. There is one exception to clamping: single-character numbers (i.e., **A** alone). Such numbers are never clamped and always take the value they would have in the highest possible **ibase**. This means that **A** alone always equals decimal **10** and **Z** alone always equals decimal **35**. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current **ibase** (with the **i** command) regardless of the current value of **ibase**. If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for **A**, use **0A**. In addition, dc(1) accepts numbers in scientific notation. These have the form **\e\**. The exponent (the portion after the **e**) must be an integer. An example is **1.89237e9**, which is equal to **1892370000**. Negative exponents are also allowed, so **4.2890e_3** is equal to **0.0042890**. **WARNING**: Both the number and the exponent in scientific notation are interpreted according to the current **ibase**, but the number is still multiplied by **10\^exponent** regardless of the current **ibase**. For example, if **ibase** is **16** and dc(1) is given the number string **FFeA**, the resulting decimal number will be **2550000000000**, and if dc(1) is given the number string **10e_4**, the resulting decimal number will be **0.0016**. Accepting input as scientific notation is a **non-portable extension**. # COMMANDS The valid commands are listed below. ## Printing These commands are used for printing. Note that both scientific notation and engineering notation are available for printing numbers. Scientific notation is activated by assigning **0** to **obase** using **0o**, and engineering notation is activated by assigning **1** to **obase** using **1o**. To deactivate them, just assign a different value to **obase**. Printing numbers in scientific notation and/or engineering notation is a **non-portable extension**. **p** : Prints the value on top of the stack, whether number or string, and prints a newline after. This does not alter the stack. **n** : Prints the value on top of the stack, whether number or string, and pops it off of the stack. **P** : Pops a value off the stack. If the value is a number, it is truncated and the absolute value of the result is printed as though **obase** is **256** and each digit is interpreted as an 8-bit ASCII character, making it a byte stream. If the value is a string, it is printed without a trailing newline. This is a **non-portable extension**. **f** : Prints the entire contents of the stack, in order from newest to oldest, without altering anything. Users should use this command when they get lost. ## Arithmetic These are the commands used for arithmetic. **+** : The top two values are popped off the stack, added, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **-** : The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **\*** : The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If **a** is the *scale* of the first expression and **b** is the *scale* of the second expression, the *scale* of the result is equal to **min(a+b,max(scale,a,b))** where **min()** and **max()** return the obvious values. **/** : The top two values are popped off the stack, divided, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be non-zero. **%** : The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. Remaindering is equivalent to 1) Computing **a/b** to current **scale**, and 2) Using the result of step 1 to calculate **a-(a/b)\*b** to *scale* **max(scale+scale(b),scale(a))**. The first value popped off of the stack must be non-zero. **~** : The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to **x y / x y %** except that **x** and **y** are only evaluated once. The first value popped off of the stack must be non-zero. This is a **non-portable extension**. **\^** : The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non-zero. **v** : The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The value popped off of the stack must be non-negative. **\_** : If this command *immediately* precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a **non-portable extension**. **b** : The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. This is a **non-portable extension**. **|** : The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. The first value popped is used as the reduction modulus and must be an integer and non-zero. The second value popped is used as the exponent and must be an integer and non-negative. The third value popped is the base and must be an integer. This is a **non-portable extension**. **\$** : The top value is popped off the stack and copied, and the copy is truncated and pushed onto the stack. This is a **non-portable extension**. **\@** : The top two values are popped off the stack, and the precision of the second is set to the value of the first, whether by truncation or extension. The first value popped off of the stack must be an integer and non-negative. This is a **non-portable extension**. **H** : The top two values are popped off the stack, and the second is shifted left (radix shifted right) to the value of the first. The first value popped off of the stack must be an integer and non-negative. This is a **non-portable extension**. **h** : The top two values are popped off the stack, and the second is shifted right (radix shifted left) to the value of the first. The first value popped off of the stack must be an integer and non-negative. This is a **non-portable extension**. **G** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if they are equal, or **0** otherwise. This is a **non-portable extension**. **N** : The top value is popped off of the stack, and if it a **0**, a **1** is pushed; otherwise, a **0** is pushed. This is a **non-portable extension**. **(** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than the second, or **0** otherwise. This is a **non-portable extension**. **{** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **)** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than the second, or **0** otherwise. This is a **non-portable extension**. **}** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **M** : The top two values are popped off of the stack. If they are both non-zero, a **1** is pushed onto the stack. If either of them is zero, or both of them are, then a **0** is pushed onto the stack. This is like the **&&** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. **m** : The top two values are popped off of the stack. If at least one of them is non-zero, a **1** is pushed onto the stack. If both of them are zero, then a **0** is pushed onto the stack. This is like the **||** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. ## Pseudo-Random Number Generator dc(1) has a built-in pseudo-random number generator. These commands query the pseudo-random number generator. (See Parameters for more information about the **seed** value that controls the pseudo-random number generator.) The pseudo-random number generator is guaranteed to **NOT** be cryptographically secure. **'** : Generates an integer between 0 and **DC_RAND_MAX**, inclusive (see the **LIMITS** section). The generated integer is made as unbiased as possible, subject to the limitations of the pseudo-random number generator. This is a **non-portable extension**. **"** : Pops a value off of the stack, which is used as an **exclusive** upper bound on the integer that will be generated. If the bound is negative or is a non-integer, an error is raised, and dc(1) resets (see the **RESET** section) while **seed** remains unchanged. If the bound is larger than **DC_RAND_MAX**, the higher bound is honored by generating several pseudo-random integers, multiplying them by appropriate powers of **DC_RAND_MAX+1**, and adding them together. Thus, the size of integer that can be generated with this command is unbounded. Using this command will change the value of **seed**, unless the operand is **0** or **1**. In that case, **0** is pushed onto the stack, and **seed** is *not* changed. The generated integer is made as unbiased as possible, subject to the limitations of the pseudo-random number generator. This is a **non-portable extension**. ## Stack Control These commands control the stack. **c** : Removes all items from ("clears") the stack. **d** : Copies the item on top of the stack ("duplicates") and pushes the copy onto the stack. **r** : Swaps ("reverses") the two top items on the stack. **R** : Pops ("removes") the top value from the stack. ## Register Control These commands control registers (see the **REGISTERS** section). **s**_r_ : Pops the value off the top of the stack and stores it into register *r*. **l**_r_ : Copies the value in register *r* and pushes it onto the stack. This does not alter the contents of *r*. **S**_r_ : Pops the value off the top of the (main) stack and pushes it onto the stack of register *r*. The previous value of the register becomes inaccessible. **L**_r_ : Pops the value off the top of the stack for register *r* and push it onto the main stack. The previous value in the stack for register *r*, if any, is now accessible via the **l**_r_ command. ## Parameters These commands control the values of **ibase**, **obase**, **scale**, and **seed**. Also see the **SYNTAX** section. **i** : Pops the value off of the top of the stack and uses it to set **ibase**, which must be between **2** and **16**, inclusive. If the value on top of the stack has any *scale*, the *scale* is ignored. **o** : Pops the value off of the top of the stack and uses it to set **obase**, which must be between **0** and **DC_BASE_MAX**, inclusive (see the **LIMITS** section and the **NUMBERS** section). If the value on top of the stack has any *scale*, the *scale* is ignored. **k** : Pops the value off of the top of the stack and uses it to set **scale**, which must be non-negative. If the value on top of the stack has any *scale*, the *scale* is ignored. **j** : Pops the value off of the top of the stack and uses it to set **seed**. The meaning of **seed** is dependent on the current pseudo-random number generator but is guaranteed to not change except for new major versions. The *scale* and sign of the value may be significant. If a previously used **seed** value is used again, the pseudo-random number generator is guaranteed to produce the same sequence of pseudo-random numbers as it did when the **seed** value was previously used. The exact value assigned to **seed** is not guaranteed to be returned if the **J** command is used. However, if **seed** *does* return a different value, both values, when assigned to **seed**, are guaranteed to produce the same sequence of pseudo-random numbers. This means that certain values assigned to **seed** will not produce unique sequences of pseudo-random numbers. There is no limit to the length (number of significant decimal digits) or *scale* of the value that can be assigned to **seed**. This is a **non-portable extension**. **I** : Pushes the current value of **ibase** onto the main stack. **O** : Pushes the current value of **obase** onto the main stack. **K** : Pushes the current value of **scale** onto the main stack. **J** : Pushes the current value of **seed** onto the main stack. This is a **non-portable extension**. **T** : Pushes the maximum allowable value of **ibase** onto the main stack. This is a **non-portable extension**. **U** : Pushes the maximum allowable value of **obase** onto the main stack. This is a **non-portable extension**. **V** : Pushes the maximum allowable value of **scale** onto the main stack. This is a **non-portable extension**. **W** : Pushes the maximum (inclusive) integer that can be generated with the **'** pseudo-random number generator command. This is a **non-portable extension**. ## Strings The following commands control strings. dc(1) can work with both numbers and strings, and registers (see the **REGISTERS** section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. Strings can also be executed as macros. For example, if the string **[1pR]** is executed as a macro, then the code **1pR** is executed, meaning that the **1** will be printed with a newline after and then popped from the stack. **\[**_characters_**\]** : Makes a string containing *characters* and pushes it onto the stack. If there are brackets (**\[** and **\]**) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (**\\**) character. If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. **a** : The value on top of the stack is popped. If it is a number, it is truncated and its absolute value is taken. The result mod **256** is calculated. If that result is **0**, push an empty string; otherwise, push a one-character string where the character is the result of the mod interpreted as an ASCII character. If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one-character string. The new string is then pushed onto the stack. This is a **non-portable extension**. **x** : Pops a value off of the top of the stack. If it is a number, it is pushed back onto the stack. If it is a string, it is executed as a macro. This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. **\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register *r* are executed. For example, **0 1>a** will execute the contents of register **a**, and **1 0>a** will not. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **?** : Reads a line from the **stdin** and executes it. This is to allow macros to request input from users. **q** : During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. **Q** : Pops a value from the stack which must be non-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. **,** : Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the **Q** command, so the sequence **,Q** will make dc(1) exit. This is a **non-portable extension**. ## Status These commands query status of the stack or its top value. **Z** : Pops a value off of the stack. If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push **1** if the argument is **0** with no decimal places. If it is a string, pushes the number of characters the string has. **X** : Pops a value off of the stack. If it is a number, pushes the *scale* of the value onto the stack. If it is a string, pushes **0**. **u** : Pops one value off of the stack. If the value is a number, this pushes **1** onto the stack. Otherwise (if it is a string), it pushes **0**. This is a **non-portable extension**. **t** : Pops one value off of the stack. If the value is a string, this pushes **1** onto the stack. Otherwise (if it is a number), it pushes **0**. This is a **non-portable extension**. **z** : Pushes the current depth of the stack (before execution of this command) onto the stack. **y**_r_ : Pushes the current stack depth of the register *r* onto the main stack. Because each register has a depth of **1** (with the value **0** in the top item) when dc(1) starts, dc(1) requires that each register's stack must always have at least one item; dc(1) will give an error and reset otherwise (see the **RESET** section). This means that this command will never push **0**. This is a **non-portable extension**. ## Arrays These commands manipulate arrays. **:**_r_ : Pops the top two values off of the stack. The second value will be stored in the array *r* (see the **REGISTERS** section), indexed by the first value. **;**_r_ : Pops the value on top of the stack and uses it as an index into the array *r*. The selected value is then pushed onto the stack. **Y**_r_ : Pushes the length of the array *r* onto the stack. This is a **non-portable extension**. ## Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter **g**. Only the characters below are allowed after the character **g**; any other character produces a parse error (see the **ERRORS** section). **gl** : Pushes the line length set by **DC_LINE_LENGTH** (see the **ENVIRONMENT VARIABLES** section) onto the stack. **gx** : Pushes **1** onto the stack if extended register mode is on, **0** otherwise. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. **gz** : Pushes **0** onto the stack if the leading zero setting has not been enabled with the **-z** or **-\-leading-zeroes** options (see the **OPTIONS** section), non-zero otherwise. # REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (**0**) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. In non-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (**'\\n'**) and a left bracket (**'['**); it is a parse error for a newline or a left bracket to be used as a register name. ## Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. If extended register mode is enabled (**-x** or **-\-extended-register** command-line arguments are given), then normal single character registers are used *unless* the character immediately following a command that needs a register name is a space (according to **isspace()**) and not a newline (**'\\n'**). In that case, the register name is found according to the regex **\[a-z\]\[a-z0-9\_\]\*** (like bc(1) identifiers), and it is a parse error if the next non-space characters do not match that regex. # RESET When dc(1) encounters an error or a signal that it has a non-default handler for, it resets. This means that several things happen. -First, any macros that are executing are stopped and popped off the stack. -The behavior is not unlike that of exceptions in programming languages. Then -the execution point is set so that any code waiting to execute (after all +First, any macros that are executing are stopped and popped off the execution +stack. The behavior is not unlike that of exceptions in programming languages. +Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. +However, the stack of values is *not* cleared; in interactive mode, users can +inspect the stack and manipulate it. + Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the **EXIT STATUS** section), it asks for more input; otherwise, it exits with the appropriate return code. # PERFORMANCE Most dc(1) implementations use **char** types to calculate the value of **1** decimal digit at a time, but that can be slow. This dc(1) does something different. It uses large integers to calculate more than **1** decimal digit at a time. If built in a environment where **DC_LONG_BIT** (see the **LIMITS** section) is **64**, then each integer has **9** decimal digits. If built in an environment where **DC_LONG_BIT** is **32** then each integer has **4** decimal digits. This value (the number of decimal digits per large integer) is called **DC_BASE_DIGS**. In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of **DC_LONG_BIT**, but is always at least twice as large as the integer type used to store digits. # LIMITS The following are the limits on dc(1): **DC_LONG_BIT** : The number of bits in the **long** type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the **PERFORMANCE** section). **DC_BASE_DIGS** : The number of decimal digits per large integer (see the **PERFORMANCE** section). Depends on **DC_LONG_BIT**. **DC_BASE_POW** : The max decimal number that each large integer can store (see **DC_BASE_DIGS**) plus **1**. Depends on **DC_BASE_DIGS**. **DC_OVERFLOW_MAX** : The max number that the overflow type (see the **PERFORMANCE** section) can hold. Depends on **DC_LONG_BIT**. **DC_BASE_MAX** : The maximum output base. Set at **DC_BASE_POW**. **DC_DIM_MAX** : The maximum size of arrays. Set at **SIZE_MAX-1**. **DC_SCALE_MAX** : The maximum **scale**. Set at **DC_OVERFLOW_MAX-1**. **DC_STRING_MAX** : The maximum length of strings. Set at **DC_OVERFLOW_MAX-1**. **DC_NAME_MAX** : The maximum length of identifiers. Set at **DC_OVERFLOW_MAX-1**. **DC_NUM_MAX** : The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at **DC_OVERFLOW_MAX-1**. **DC_RAND_MAX** : The maximum integer (inclusive) returned by the **'** command, if dc(1). Set at **2\^DC_LONG_BIT-1**. Exponent : The maximum allowable exponent (positive or negative). Set at **DC_OVERFLOW_MAX**. Number of vars : The maximum number of vars/arrays. Set at **SIZE_MAX-1**. These limits are meant to be effectively non-existent; the limits are so large (at least on 64-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. # ENVIRONMENT VARIABLES As **non-portable extensions**, dc(1) recognizes the following environment variables: **DC_ENV_ARGS** : This is another way to give command-line arguments to dc(1). They should be in the same format as all other command-line arguments. These are always processed first, so any files given in **DC_ENV_ARGS** will be processed before arguments and files given on the command-line. This gives the user the ability to set up "standard" options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the **-e** option to set **scale** to a value other than **0**. The code that parses **DC_ENV_ARGS** will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string **"/home/gavin/some dc file.dc"** will be correctly parsed, but the string **"/home/gavin/some \"dc\" file.dc"** will include the backslashes. The quote parsing will handle either kind of quotes, **'** or **"**. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in **"some 'dc' file.dc"**, and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in **DC_ENV_ARGS** is not supported due to the complexity of the parsing, though such files are still supported on the command-line where the parsing is done by the shell. **DC_LINE_LENGTH** : If this environment variable exists and contains an integer that is greater than **1** and is less than **UINT16_MAX** (**2\^16-1**), dc(1) will output lines to that length, including the backslash newline combo. The default line length is **70**. The special value of **0** will disable line length checking and print numbers without regard to line length and without backslashes and newlines. **DC_SIGINT_RESET** : If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), then this environment variable has no effect because dc(1) exits on **SIGINT** when not in interactive mode. However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) reset on **SIGINT**, rather than exit, and zero makes dc(1) exit. If this environment variable exists and is *not* an integer, then dc(1) will exit on **SIGINT**. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_TTY_MODE** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_PROMPT** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) use a prompt, and zero or a non-integer makes dc(1) not use a prompt. If this environment variable does not exist and **DC_TTY_MODE** does, then the value of the **DC_TTY_MODE** environment variable is used. This environment variable and the **DC_TTY_MODE** environment variable override the default, which can be queried with the **-h** or **-\-help** options. **DC_EXPR_EXIT** : If any expressions or expression files are given on the command-line with **-e**, **-\-expression**, **-f**, or **-\-file**, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_DIGIT_CLAMP** : When parsing numbers and if this environment variable exists and contains an integer, a non-zero value makes dc(1) clamp digits that are greater than or equal to the current **ibase** so that all such digits are considered equal to the **ibase** minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the **ibase**. This never applies to single-digit numbers, as per the bc(1) standard (see the **STANDARDS** section). This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. # EXIT STATUS dc(1) returns the following exit statuses: **0** : No error. **1** : A math error occurred. This follows standard practice of using **1** for expected errors, since math errors will happen in the process of normal execution. Math errors include divide by **0**, taking the square root of a negative number, using a negative number as a bound for the pseudo-random number generator, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non-integer where an integer is required. Converting to a hardware integer happens for the second operand of the power (**\^**), places (**\@**), left shift (**H**), and right shift (**h**) operators. **2** : A parse error occurred. Parse errors include unexpected **EOF**, using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. **3** : A runtime error occurred. Runtime errors include assigning an invalid number to any global (**ibase**, **obase**, or **scale**), giving a bad expression to a **read()** call, calling **read()** inside of a **read()** call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. **4** : A fatal error occurred. Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command-line options. The exit status **4** is special; when a fatal error occurs, dc(1) always exits and returns **4**, no matter what mode dc(1) is in. The other statuses will only be returned when dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), since dc(1) resets its state (see the **RESET** section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the **-i** flag or **-\-interactive** option. These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the **-i** flag or **-\-interactive** option. # INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non-interactive mode. Interactive mode is turned on automatically when both **stdin** and **stdout** are hooked to a terminal, but the **-i** flag and **-\-interactive** option can turn it on in other situations. In interactive mode, dc(1) attempts to recover from errors (see the **RESET** section), and in normal execution, flushes **stdout** as soon as execution is done for the current input. dc(1) may also reset on **SIGINT** instead of exit, depending on the contents of, or default for, the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section). # TTY MODE If **stdin**, **stdout**, and **stderr** are all connected to a TTY, then "TTY mode" is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. If there is the environment variable **DC_TTY_MODE** in the environment (see the **ENVIRONMENT VARIABLES** section), then if that environment variable contains a non-zero integer, dc(1) will turn on TTY mode when **stdin**, **stdout**, and **stderr** are all connected to a TTY. If the **DC_TTY_MODE** environment variable exists but is *not* a non-zero integer, then dc(1) will not turn TTY mode on. If the environment variable **DC_TTY_MODE** does *not* exist, the default setting is used. The default setting can be queried with the **-h** or **-\-help** options. TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the **STANDARDS** section), and interactive mode requires only **stdin** and **stdout** to be connected to a terminal. ## Command-Line History Command-line history is only enabled if TTY mode is, i.e., that **stdin**, **stdout**, and **stderr** are connected to a TTY and the **DC_TTY_MODE** environment variable (see the **ENVIRONMENT VARIABLES** section) and its default do not disable TTY mode. See the **COMMAND LINE HISTORY** section for more information. ## Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: **DC_PROMPT** (see the **ENVIRONMENT VARIABLES** section). If the environment variable **DC_PROMPT** exists and is a non-zero integer, then the prompt is turned on when **stdin**, **stdout**, and **stderr** are connected to a TTY and the **-P** and **-\-no-prompt** options were not used. The read prompt will be turned on under the same conditions, except that the **-R** and **-\-no-read-prompt** options must also not be used. However, if **DC_PROMPT** does not exist, the prompt can be enabled or disabled with the **DC_TTY_MODE** environment variable, the **-P** and **-\-no-prompt** options, and the **-R** and **-\-no-read-prompt** options. See the **ENVIRONMENT VARIABLES** and **OPTIONS** sections for more details. # SIGNAL HANDLING Sending a **SIGINT** will cause dc(1) to do one of two things. If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), or the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section), or its default, is either not an integer or it is zero, dc(1) will exit. However, if dc(1) is in interactive mode, and the **DC_SIGINT_RESET** or its default is an integer and non-zero, then dc(1) will stop executing the current input and reset (see the **RESET** section) upon receiving a **SIGINT**. Note that "current input" can mean one of two things. If dc(1) is processing input from **stdin** in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from **stdin** if no other file exists. This means that if a **SIGINT** is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. **SIGTERM** and **SIGQUIT** cause dc(1) to clean up and exit, and it uses the default handler for all other signals. The one exception is **SIGHUP**; in that case, and only when dc(1) is in TTY mode (see the **TTY MODE** section), a **SIGHUP** will cause dc(1) to clean up and exit. # COMMAND LINE HISTORY dc(1) supports interactive command-line editing. If dc(1) can be in TTY mode (see the **TTY MODE** section), history can be enabled. This means that command-line history can only be enabled when **stdin**, **stdout**, and **stderr** are all connected to a TTY. Like TTY mode itself, it can be turned on or off with the environment variable **DC_TTY_MODE** (see the **ENVIRONMENT VARIABLES** section). **Note**: tabs are converted to 8 spaces. # LOCALES This dc(1) ships with support for adding error messages for different locales and thus, supports **LC_MESSAGES**. # SEE ALSO bc(1) # STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1-2017 (“POSIX.1-2017”) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . # BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . # AUTHOR Gavin D. Howard and contributors. diff --git a/manuals/dc/E.1 b/manuals/dc/E.1 index 91f68dfd7467..a5febe44705f 100644 --- a/manuals/dc/E.1 +++ b/manuals/dc/E.1 @@ -1,1471 +1,1474 @@ .\" .\" SPDX-License-Identifier: BSD-2-Clause .\" .\" Copyright (c) 2018-2024 Gavin D. Howard and contributors. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions are met: .\" .\" * Redistributions of source code must retain the above copyright notice, .\" this list of conditions and the following disclaimer. .\" .\" * Redistributions in binary form must reproduce the above copyright notice, .\" this list of conditions and the following disclaimer in the documentation .\" and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" .\" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE .\" LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF .\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS .\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN .\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) .\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE .\" POSSIBILITY OF SUCH DAMAGE. .\" -.TH "DC" "1" "January 2024" "Gavin D. Howard" "General Commands Manual" +.TH "DC" "1" "August 2024" "Gavin D. Howard" "General Commands Manual" .nh .ad l .SH Name dc \- arbitrary\-precision decimal reverse\-Polish notation calculator .SH SYNOPSIS \f[B]dc\f[R] [\f[B]\-cChiPRvVx\f[R]] [\f[B]\-\-version\f[R]] [\f[B]\-\-help\f[R]] [\f[B]\-\-digit\-clamp\f[R]] [\f[B]\-\-no\-digit\-clamp\f[R]] [\f[B]\-\-interactive\f[R]] [\f[B]\-\-no\-prompt\f[R]] [\f[B]\-\-no\-read\-prompt\f[R]] [\f[B]\-\-extended\-register\f[R]] [\f[B]\-e\f[R] \f[I]expr\f[R]] [\f[B]\-\-expression\f[R]=\f[I]expr\f[R]\&...] [\f[B]\-f\f[R] \f[I]file\f[R]\&...] [\f[B]\-\-file\f[R]=\f[I]file\f[R]\&...] [\f[I]file\f[R]\&...] .SH DESCRIPTION dc(1) is an arbitrary\-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. .PP If no files are given on the command\-line, then dc(1) reads from \f[B]stdin\f[R] (see the \f[B]STDIN\f[R] section). Otherwise, those files are processed, and dc(1) will then exit. .PP If a user wants to set up a standard environment, they can use \f[B]DC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). For example, if a user wants the \f[B]scale\f[R] always set to \f[B]10\f[R], they can set \f[B]DC_ENV_ARGS\f[R] to \f[B]\-e 10k\f[R], and this dc(1) will always start with a \f[B]scale\f[R] of \f[B]10\f[R]. .SH OPTIONS The following are the options that dc(1) accepts. .TP \f[B]\-C\f[R], \f[B]\-\-no\-digit\-clamp\f[R] Disables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that the value added to a number from a digit is always that digit\[cq]s value multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-c\f[R] or \f[B]\-\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-c\f[R], \f[B]\-\-digit\-clamp\f[R] Enables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-C\f[R] or \f[B]\-\-no\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-e\f[R] \f[I]expr\f[R], \f[B]\-\-expression\f[R]=\f[I]expr\f[R] Evaluates \f[I]expr\f[R]. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R], whether on the command\-line or in \f[B]DC_ENV_ARGS\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-f\f[R] \f[I]file\f[R], \f[B]\-\-file\f[R]=\f[I]file\f[R] Reads in \f[I]file\f[R] and evaluates it, line by line, as though it were read through \f[B]stdin\f[R]. If expressions are also given (see above), the expressions are evaluated in the order given. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-h\f[R], \f[B]\-\-help\f[R] Prints a usage message and exits. .TP \f[B]\-I\f[R] \f[I]ibase\f[R], \f[B]\-\-ibase\f[R]=\f[I]ibase\f[R] Sets the builtin variable \f[B]ibase\f[R] to the value \f[I]ibase\f[R] assuming that \f[I]ibase\f[R] is in base 10. It is a fatal error if \f[I]ibase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-i\f[R], \f[B]\-\-interactive\f[R] Forces interactive mode. (See the \f[B]INTERACTIVE MODE\f[R] section.) .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-L\f[R], \f[B]\-\-no\-line\-length\f[R] Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets \f[B]BC_LINE_LENGTH\f[R] to \f[B]0\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-O\f[R] \f[I]obase\f[R], \f[B]\-\-obase\f[R]=\f[I]obase\f[R] Sets the builtin variable \f[B]obase\f[R] to the value \f[I]obase\f[R] assuming that \f[I]obase\f[R] is in base 10. It is a fatal error if \f[I]obase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-P\f[R], \f[B]\-\-no\-prompt\f[R] Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]DC_ENV_ARGS\f[R]. .RS .PP These options override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-R\f[R], \f[B]\-\-no\-read\-prompt\f[R] Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]BC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. .RS .PP This option does not disable the regular prompt because the read prompt is only used when the \f[B]?\f[R] command is used. .PP These options \f[I]do\f[R] override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), but only for the read prompt. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-S\f[R] \f[I]scale\f[R], \f[B]\-\-scale\f[R]=\f[I]scale\f[R] Sets the builtin variable \f[B]scale\f[R] to the value \f[I]scale\f[R] assuming that \f[I]scale\f[R] is in base 10. It is a fatal error if \f[I]scale\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-v\f[R], \f[B]\-V\f[R], \f[B]\-\-version\f[R] Print the version information (copyright header) and exits. .TP \f[B]\-x\f[R] \f[B]\-\-extended\-register\f[R] Enables extended register mode. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-z\f[R], \f[B]\-\-leading\-zeroes\f[R] Makes dc(1) print all numbers greater than \f[B]\-1\f[R] and less than \f[B]1\f[R], and not equal to \f[B]0\f[R], with a leading zero. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .PP All long options are \f[B]non\-portable extensions\f[R]. .SH STDIN If no files are given on the command\-line and no files or expressions are given by the \f[B]\-f\f[R], \f[B]\-\-file\f[R], \f[B]\-e\f[R], or \f[B]\-\-expression\f[R] options, then dc(1) reads from \f[B]stdin\f[R]. .PP However, there is a caveat to this. .PP First, \f[B]stdin\f[R] is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. .SH STDOUT Any non\-error output is written to \f[B]stdout\f[R]. In addition, if history (see the \f[B]HISTORY\f[R] section) and the prompt (see the \f[B]TTY MODE\f[R] section) are enabled, both are output to \f[B]stdout\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stdout\f[R], so if \f[B]stdout\f[R] is closed, as in \f[B]dc >&\-\f[R], it will quit with an error. This is done so that dc(1) can report problems when \f[B]stdout\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stdout\f[R] to \f[B]/dev/null\f[R]. .SH STDERR Any error output is written to \f[B]stderr\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stderr\f[R], so if \f[B]stderr\f[R] is closed, as in \f[B]dc 2>&\-\f[R], it will quit with an error. This is done so that dc(1) can exit with an error code when \f[B]stderr\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stderr\f[R] to \f[B]/dev/null\f[R]. .SH SYNTAX Each item in the input source code, either a number (see the \f[B]NUMBERS\f[R] section) or a command (see the \f[B]COMMANDS\f[R] section), is processed and executed, in order. Input is processed immediately when entered. .PP \f[B]ibase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to interpret constant numbers. It is the \[lq]input\[rq] base, or the number base used for interpreting input numbers. \f[B]ibase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]ibase\f[R] is \f[B]16\f[R]. The min allowable value for \f[B]ibase\f[R] is \f[B]2\f[R]. The max allowable value for \f[B]ibase\f[R] can be queried in dc(1) programs with the \f[B]T\f[R] command. .PP \f[B]obase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to output results. It is the \[lq]output\[rq] base, or the number base used for outputting numbers. \f[B]obase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]obase\f[R] is \f[B]DC_BASE_MAX\f[R] and can be queried with the \f[B]U\f[R] command. The min allowable value for \f[B]obase\f[R] is \f[B]2\f[R]. Values are output in the specified base. .PP The \f[I]scale\f[R] of an expression is the number of digits in the result of the expression right of the decimal point, and \f[B]scale\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that sets the precision of any operations (with exceptions). \f[B]scale\f[R] is initially \f[B]0\f[R]. \f[B]scale\f[R] cannot be negative. The max allowable value for \f[B]scale\f[R] can be queried in dc(1) programs with the \f[B]V\f[R] command. .SS Comments Comments go from \f[B]#\f[R] until, and not including, the next newline. This is a \f[B]non\-portable extension\f[R]. .SH NUMBERS Numbers are strings made up of digits, uppercase letters up to \f[B]F\f[R], and at most \f[B]1\f[R] period for a radix. Numbers can have up to \f[B]DC_NUM_MAX\f[R] digits. Uppercase letters are equal to \f[B]9\f[R] plus their position in the alphabet (i.e., \f[B]A\f[R] equals \f[B]10\f[R], or \f[B]9+1\f[R]). .PP If a digit or letter makes no sense with the current value of \f[B]ibase\f[R] (i.e., they are greater than or equal to the current value of \f[B]ibase\f[R]), then the behavior depends on the existence of the \f[B]\-c\f[R]/\f[B]\-\-digit\-clamp\f[R] or \f[B]\-C\f[R]/\f[B]\-\-no\-digit\-clamp\f[R] options (see the \f[B]OPTIONS\f[R] section), the existence and setting of the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or the default, which can be queried with the \f[B]\-h\f[R]/\f[B]\-\-help\f[R] option. .PP If clamping is off, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are not changed. Instead, their given value is multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*A+3\[ha]0*B\f[R], which is \f[B]3\f[R] times \f[B]10\f[R] plus \f[B]11\f[R], or \f[B]41\f[R]. .PP If clamping is on, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are set to the value of the highest valid digit in \f[B]ibase\f[R] before being multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*2+3\[ha]0*2\f[R], which is \f[B]3\f[R] times \f[B]2\f[R] plus \f[B]2\f[R], or \f[B]8\f[R]. .PP There is one exception to clamping: single\-character numbers (i.e., \f[B]A\f[R] alone). Such numbers are never clamped and always take the value they would have in the highest possible \f[B]ibase\f[R]. This means that \f[B]A\f[R] alone always equals decimal \f[B]10\f[R] and \f[B]Z\f[R] alone always equals decimal \f[B]35\f[R]. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current \f[B]ibase\f[R] (with the \f[B]i\f[R] command) regardless of the current value of \f[B]ibase\f[R]. .PP If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for \f[B]A\f[R], use \f[B]0A\f[R]. .SH COMMANDS The valid commands are listed below. .SS Printing These commands are used for printing. .TP \f[B]p\f[R] Prints the value on top of the stack, whether number or string, and prints a newline after. .RS .PP This does not alter the stack. .RE .TP \f[B]n\f[R] Prints the value on top of the stack, whether number or string, and pops it off of the stack. .TP \f[B]P\f[R] Pops a value off the stack. .RS .PP If the value is a number, it is truncated and the absolute value of the result is printed as though \f[B]obase\f[R] is \f[B]256\f[R] and each digit is interpreted as an 8\-bit ASCII character, making it a byte stream. .PP If the value is a string, it is printed without a trailing newline. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]f\f[R] Prints the entire contents of the stack, in order from newest to oldest, without altering anything. .RS .PP Users should use this command when they get lost. .RE .SS Arithmetic These are the commands used for arithmetic. .TP \f[B]+\f[R] The top two values are popped off the stack, added, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]\-\f[R] The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]*\f[R] The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If \f[B]a\f[R] is the \f[I]scale\f[R] of the first expression and \f[B]b\f[R] is the \f[I]scale\f[R] of the second expression, the \f[I]scale\f[R] of the result is equal to \f[B]min(a+b,max(scale,a,b))\f[R] where \f[B]min()\f[R] and \f[B]max()\f[R] return the obvious values. .TP \f[B]/\f[R] The top two values are popped off the stack, divided, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]%\f[R] The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. .RS .PP Remaindering is equivalent to 1) Computing \f[B]a/b\f[R] to current \f[B]scale\f[R], and 2) Using the result of step 1 to calculate \f[B]a\-(a/b)*b\f[R] to \f[I]scale\f[R] \f[B]max(scale+scale(b),scale(a))\f[R]. .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]\[ti]\f[R] The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to \f[B]x y / x y %\f[R] except that \f[B]x\f[R] and \f[B]y\f[R] are only evaluated once. .RS .PP The first value popped off of the stack must be non\-zero. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[ha]\f[R] The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non\-zero. .RE .TP \f[B]v\f[R] The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The value popped off of the stack must be non\-negative. .RE .TP \f[B]_\f[R] If this command \f[I]immediately\f[R] precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. .RS .PP Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]b\f[R] The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]|\f[R] The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. .RS .PP The first value popped is used as the reduction modulus and must be an integer and non\-zero. The second value popped is used as the exponent and must be an integer and non\-negative. The third value popped is the base and must be an integer. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]G\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if they are equal, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]N\f[R] The top value is popped off of the stack, and if it a \f[B]0\f[R], a \f[B]1\f[R] is pushed; otherwise, a \f[B]0\f[R] is pushed. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B](\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]{\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B])\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]}\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]M\f[R] The top two values are popped off of the stack. If they are both non\-zero, a \f[B]1\f[R] is pushed onto the stack. If either of them is zero, or both of them are, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]&&\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]m\f[R] The top two values are popped off of the stack. If at least one of them is non\-zero, a \f[B]1\f[R] is pushed onto the stack. If both of them are zero, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]||\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Stack Control These commands control the stack. .TP \f[B]c\f[R] Removes all items from (\[lq]clears\[rq]) the stack. .TP \f[B]d\f[R] Copies the item on top of the stack (\[lq]duplicates\[rq]) and pushes the copy onto the stack. .TP \f[B]r\f[R] Swaps (\[lq]reverses\[rq]) the two top items on the stack. .TP \f[B]R\f[R] Pops (\[lq]removes\[rq]) the top value from the stack. .SS Register Control These commands control registers (see the \f[B]REGISTERS\f[R] section). .TP \f[B]s\f[R]\f[I]r\f[R] Pops the value off the top of the stack and stores it into register \f[I]r\f[R]. .TP \f[B]l\f[R]\f[I]r\f[R] Copies the value in register \f[I]r\f[R] and pushes it onto the stack. This does not alter the contents of \f[I]r\f[R]. .TP \f[B]S\f[R]\f[I]r\f[R] Pops the value off the top of the (main) stack and pushes it onto the stack of register \f[I]r\f[R]. The previous value of the register becomes inaccessible. .TP \f[B]L\f[R]\f[I]r\f[R] Pops the value off the top of the stack for register \f[I]r\f[R] and push it onto the main stack. The previous value in the stack for register \f[I]r\f[R], if any, is now accessible via the \f[B]l\f[R]\f[I]r\f[R] command. .SS Parameters These commands control the values of \f[B]ibase\f[R], \f[B]obase\f[R], and \f[B]scale\f[R]. Also see the \f[B]SYNTAX\f[R] section. .TP \f[B]i\f[R] Pops the value off of the top of the stack and uses it to set \f[B]ibase\f[R], which must be between \f[B]2\f[R] and \f[B]16\f[R], inclusive. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]o\f[R] Pops the value off of the top of the stack and uses it to set \f[B]obase\f[R], which must be between \f[B]2\f[R] and \f[B]DC_BASE_MAX\f[R], inclusive (see the \f[B]LIMITS\f[R] section). .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]k\f[R] Pops the value off of the top of the stack and uses it to set \f[B]scale\f[R], which must be non\-negative. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]I\f[R] Pushes the current value of \f[B]ibase\f[R] onto the main stack. .TP \f[B]O\f[R] Pushes the current value of \f[B]obase\f[R] onto the main stack. .TP \f[B]K\f[R] Pushes the current value of \f[B]scale\f[R] onto the main stack. .TP \f[B]T\f[R] Pushes the maximum allowable value of \f[B]ibase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]U\f[R] Pushes the maximum allowable value of \f[B]obase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]V\f[R] Pushes the maximum allowable value of \f[B]scale\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Strings The following commands control strings. .PP dc(1) can work with both numbers and strings, and registers (see the \f[B]REGISTERS\f[R] section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. .PP While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. .PP Strings can also be executed as macros. For example, if the string \f[B][1pR]\f[R] is executed as a macro, then the code \f[B]1pR\f[R] is executed, meaning that the \f[B]1\f[R] will be printed with a newline after and then popped from the stack. .TP \f[B][\f[R]\f[I]characters\f[R]\f[B]]\f[R] Makes a string containing \f[I]characters\f[R] and pushes it onto the stack. .RS .PP If there are brackets (\f[B][\f[R] and \f[B]]\f[R]) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (\f[B]\[rs]\f[R]) character. .PP If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. .RE .TP \f[B]a\f[R] The value on top of the stack is popped. .RS .PP If it is a number, it is truncated and its absolute value is taken. The result mod \f[B]256\f[R] is calculated. If that result is \f[B]0\f[R], push an empty string; otherwise, push a one\-character string where the character is the result of the mod interpreted as an ASCII character. .PP If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one\-character string. The new string is then pushed onto the stack. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]x\f[R] Pops a value off of the top of the stack. .RS .PP If it is a number, it is pushed back onto the stack. .PP If it is a string, it is executed as a macro. .PP This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. .RE .TP \f[B]>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP For example, \f[B]0 1>a\f[R] will execute the contents of register \f[B]a\f[R], and \f[B]1 0>a\f[R] will not. .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]?\f[R] Reads a line from the \f[B]stdin\f[R] and executes it. This is to allow macros to request input from users. .TP \f[B]q\f[R] During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. .TP \f[B]Q\f[R] Pops a value from the stack which must be non\-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. .TP \f[B],\f[R] Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the \f[B]Q\f[R] command, so the sequence \f[B],Q\f[R] will make dc(1) exit. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Status These commands query status of the stack or its top value. .TP \f[B]Z\f[R] Pops a value off of the stack. .RS .PP If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push \f[B]1\f[R] if the argument is \f[B]0\f[R] with no decimal places. .PP If it is a string, pushes the number of characters the string has. .RE .TP \f[B]X\f[R] Pops a value off of the stack. .RS .PP If it is a number, pushes the \f[I]scale\f[R] of the value onto the stack. .PP If it is a string, pushes \f[B]0\f[R]. .RE .TP \f[B]u\f[R] Pops one value off of the stack. If the value is a number, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a string), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]t\f[R] Pops one value off of the stack. If the value is a string, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a number), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]z\f[R] Pushes the current depth of the stack (before execution of this command) onto the stack. .TP \f[B]y\f[R]\f[I]r\f[R] Pushes the current stack depth of the register \f[I]r\f[R] onto the main stack. .RS .PP Because each register has a depth of \f[B]1\f[R] (with the value \f[B]0\f[R] in the top item) when dc(1) starts, dc(1) requires that each register\[cq]s stack must always have at least one item; dc(1) will give an error and reset otherwise (see the \f[B]RESET\f[R] section). This means that this command will never push \f[B]0\f[R]. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Arrays These commands manipulate arrays. .TP \f[B]:\f[R]\f[I]r\f[R] Pops the top two values off of the stack. The second value will be stored in the array \f[I]r\f[R] (see the \f[B]REGISTERS\f[R] section), indexed by the first value. .TP \f[B];\f[R]\f[I]r\f[R] Pops the value on top of the stack and uses it as an index into the array \f[I]r\f[R]. The selected value is then pushed onto the stack. .TP \f[B]Y\f[R]\f[I]r\f[R] Pushes the length of the array \f[I]r\f[R] onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter \f[B]g\f[R]. Only the characters below are allowed after the character \f[B]g\f[R]; any other character produces a parse error (see the \f[B]ERRORS\f[R] section). .TP \f[B]gl\f[R] Pushes the line length set by \f[B]DC_LINE_LENGTH\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) onto the stack. .TP \f[B]gx\f[R] Pushes \f[B]1\f[R] onto the stack if extended register mode is on, \f[B]0\f[R] otherwise. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .TP \f[B]gz\f[R] Pushes \f[B]0\f[R] onto the stack if the leading zero setting has not been enabled with the \f[B]\-z\f[R] or \f[B]\-\-leading\-zeroes\f[R] options (see the \f[B]OPTIONS\f[R] section), non\-zero otherwise. .SH REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) .PP Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (\f[B]0\f[R]) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. .PP In non\-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (\f[B]`\[rs]n'\f[R]) and a left bracket (\f[B]`['\f[R]); it is a parse error for a newline or a left bracket to be used as a register name. .SS Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. .PP If extended register mode is enabled (\f[B]\-x\f[R] or \f[B]\-\-extended\-register\f[R] command\-line arguments are given), then normal single character registers are used \f[I]unless\f[R] the character immediately following a command that needs a register name is a space (according to \f[B]isspace()\f[R]) and not a newline (\f[B]`\[rs]n'\f[R]). .PP In that case, the register name is found according to the regex \f[B][a\-z][a\-z0\-9_]*\f[R] (like bc(1) identifiers), and it is a parse error if the next non\-space characters do not match that regex. .SH RESET When dc(1) encounters an error or a signal that it has a non\-default handler for, it resets. This means that several things happen. .PP First, any macros that are executing are stopped and popped off the -stack. +execution stack. The behavior is not unlike that of exceptions in programming languages. Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. .PP +However, the stack of values is \f[I]not\f[R] cleared; in interactive +mode, users can inspect the stack and manipulate it. +.PP Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the \f[B]EXIT STATUS\f[R] section), it asks for more input; otherwise, it exits with the appropriate return code. .SH PERFORMANCE Most dc(1) implementations use \f[B]char\f[R] types to calculate the value of \f[B]1\f[R] decimal digit at a time, but that can be slow. This dc(1) does something different. .PP It uses large integers to calculate more than \f[B]1\f[R] decimal digit at a time. If built in a environment where \f[B]DC_LONG_BIT\f[R] (see the \f[B]LIMITS\f[R] section) is \f[B]64\f[R], then each integer has \f[B]9\f[R] decimal digits. If built in an environment where \f[B]DC_LONG_BIT\f[R] is \f[B]32\f[R] then each integer has \f[B]4\f[R] decimal digits. This value (the number of decimal digits per large integer) is called \f[B]DC_BASE_DIGS\f[R]. .PP In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of \f[B]DC_LONG_BIT\f[R], but is always at least twice as large as the integer type used to store digits. .SH LIMITS The following are the limits on dc(1): .TP \f[B]DC_LONG_BIT\f[R] The number of bits in the \f[B]long\f[R] type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the \f[B]PERFORMANCE\f[R] section). .TP \f[B]DC_BASE_DIGS\f[R] The number of decimal digits per large integer (see the \f[B]PERFORMANCE\f[R] section). Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_POW\f[R] The max decimal number that each large integer can store (see \f[B]DC_BASE_DIGS\f[R]) plus \f[B]1\f[R]. Depends on \f[B]DC_BASE_DIGS\f[R]. .TP \f[B]DC_OVERFLOW_MAX\f[R] The max number that the overflow type (see the \f[B]PERFORMANCE\f[R] section) can hold. Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_MAX\f[R] The maximum output base. Set at \f[B]DC_BASE_POW\f[R]. .TP \f[B]DC_DIM_MAX\f[R] The maximum size of arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .TP \f[B]DC_SCALE_MAX\f[R] The maximum \f[B]scale\f[R]. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_STRING_MAX\f[R] The maximum length of strings. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NAME_MAX\f[R] The maximum length of identifiers. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NUM_MAX\f[R] The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP Exponent The maximum allowable exponent (positive or negative). Set at \f[B]DC_OVERFLOW_MAX\f[R]. .TP Number of vars The maximum number of vars/arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .PP These limits are meant to be effectively non\-existent; the limits are so large (at least on 64\-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. .SH ENVIRONMENT VARIABLES As \f[B]non\-portable extensions\f[R], dc(1) recognizes the following environment variables: .TP \f[B]DC_ENV_ARGS\f[R] This is another way to give command\-line arguments to dc(1). They should be in the same format as all other command\-line arguments. These are always processed first, so any files given in \f[B]DC_ENV_ARGS\f[R] will be processed before arguments and files given on the command\-line. This gives the user the ability to set up \[lq]standard\[rq] options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the \f[B]\-e\f[R] option to set \f[B]scale\f[R] to a value other than \f[B]0\f[R]. .RS .PP The code that parses \f[B]DC_ENV_ARGS\f[R] will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string \f[B]\[lq]/home/gavin/some dc file.dc\[rq]\f[R] will be correctly parsed, but the string \f[B]\[lq]/home/gavin/some \[dq]dc\[dq] file.dc\[rq]\f[R] will include the backslashes. .PP The quote parsing will handle either kind of quotes, \f[B]\[cq]\f[R] or \f[B]\[lq]\f[R]. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in \f[B]\[lq]some `dc' file.dc\[rq]\f[R], and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in \f[B]DC_ENV_ARGS\f[R] is not supported due to the complexity of the parsing, though such files are still supported on the command\-line where the parsing is done by the shell. .RE .TP \f[B]DC_LINE_LENGTH\f[R] If this environment variable exists and contains an integer that is greater than \f[B]1\f[R] and is less than \f[B]UINT16_MAX\f[R] (\f[B]2\[ha]16\-1\f[R]), dc(1) will output lines to that length, including the backslash newline combo. The default line length is \f[B]70\f[R]. .RS .PP The special value of \f[B]0\f[R] will disable line length checking and print numbers without regard to line length and without backslashes and newlines. .RE .TP \f[B]DC_SIGINT_RESET\f[R] If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), then this environment variable has no effect because dc(1) exits on \f[B]SIGINT\f[R] when not in interactive mode. .RS .PP However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) reset on \f[B]SIGINT\f[R], rather than exit, and zero makes dc(1) exit. If this environment variable exists and is \f[I]not\f[R] an integer, then dc(1) will exit on \f[B]SIGINT\f[R]. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_TTY_MODE\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non\-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_PROMPT\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) use a prompt, and zero or a non\-integer makes dc(1) not use a prompt. If this environment variable does not exist and \f[B]DC_TTY_MODE\f[R] does, then the value of the \f[B]DC_TTY_MODE\f[R] environment variable is used. .PP This environment variable and the \f[B]DC_TTY_MODE\f[R] environment variable override the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_EXPR_EXIT\f[R] If any expressions or expression files are given on the command\-line with \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R], then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. .RS .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_DIGIT_CLAMP\f[R] When parsing numbers and if this environment variable exists and contains an integer, a non\-zero value makes dc(1) clamp digits that are greater than or equal to the current \f[B]ibase\f[R] so that all such digits are considered equal to the \f[B]ibase\f[R] minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the \f[B]ibase\f[R]. .RS .PP This never applies to single\-digit numbers, as per the bc(1) standard (see the \f[B]STANDARDS\f[R] section). .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .SH EXIT STATUS dc(1) returns the following exit statuses: .TP \f[B]0\f[R] No error. .TP \f[B]1\f[R] A math error occurred. This follows standard practice of using \f[B]1\f[R] for expected errors, since math errors will happen in the process of normal execution. .RS .PP Math errors include divide by \f[B]0\f[R], taking the square root of a negative number, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non\-integer where an integer is required. .PP Converting to a hardware integer happens for the second operand of the power (\f[B]\[ha]\f[R]) operator. .RE .TP \f[B]2\f[R] A parse error occurred. .RS .PP Parse errors include unexpected \f[B]EOF\f[R], using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. .RE .TP \f[B]3\f[R] A runtime error occurred. .RS .PP Runtime errors include assigning an invalid number to any global (\f[B]ibase\f[R], \f[B]obase\f[R], or \f[B]scale\f[R]), giving a bad expression to a \f[B]read()\f[R] call, calling \f[B]read()\f[R] inside of a \f[B]read()\f[R] call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. .RE .TP \f[B]4\f[R] A fatal error occurred. .RS .PP Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command\-line options. .RE .PP The exit status \f[B]4\f[R] is special; when a fatal error occurs, dc(1) always exits and returns \f[B]4\f[R], no matter what mode dc(1) is in. .PP The other statuses will only be returned when dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), since dc(1) resets its state (see the \f[B]RESET\f[R] section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .PP These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .SH INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non\-interactive mode. Interactive mode is turned on automatically when both \f[B]stdin\f[R] and \f[B]stdout\f[R] are hooked to a terminal, but the \f[B]\-i\f[R] flag and \f[B]\-\-interactive\f[R] option can turn it on in other situations. .PP In interactive mode, dc(1) attempts to recover from errors (see the \f[B]RESET\f[R] section), and in normal execution, flushes \f[B]stdout\f[R] as soon as execution is done for the current input. dc(1) may also reset on \f[B]SIGINT\f[R] instead of exit, depending on the contents of, or default for, the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .SH TTY MODE If \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY, then \[lq]TTY mode\[rq] is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. .PP If there is the environment variable \f[B]DC_TTY_MODE\f[R] in the environment (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then if that environment variable contains a non\-zero integer, dc(1) will turn on TTY mode when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY. If the \f[B]DC_TTY_MODE\f[R] environment variable exists but is \f[I]not\f[R] a non\-zero integer, then dc(1) will not turn TTY mode on. .PP If the environment variable \f[B]DC_TTY_MODE\f[R] does \f[I]not\f[R] exist, the default setting is used. The default setting can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the \f[B]STANDARDS\f[R] section), and interactive mode requires only \f[B]stdin\f[R] and \f[B]stdout\f[R] to be connected to a terminal. .SS Command\-Line History Command\-line history is only enabled if TTY mode is, i.e., that \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are connected to a TTY and the \f[B]DC_TTY_MODE\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and its default do not disable TTY mode. See the \f[B]COMMAND LINE HISTORY\f[R] section for more information. .SS Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: \f[B]DC_PROMPT\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP If the environment variable \f[B]DC_PROMPT\f[R] exists and is a non\-zero integer, then the prompt is turned on when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are connected to a TTY and the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options were not used. The read prompt will be turned on under the same conditions, except that the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options must also not be used. .PP However, if \f[B]DC_PROMPT\f[R] does not exist, the prompt can be enabled or disabled with the \f[B]DC_TTY_MODE\f[R] environment variable, the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options, and the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options. See the \f[B]ENVIRONMENT VARIABLES\f[R] and \f[B]OPTIONS\f[R] sections for more details. .SH SIGNAL HANDLING Sending a \f[B]SIGINT\f[R] will cause dc(1) to do one of two things. .PP If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), or the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or its default, is either not an integer or it is zero, dc(1) will exit. .PP However, if dc(1) is in interactive mode, and the \f[B]DC_SIGINT_RESET\f[R] or its default is an integer and non\-zero, then dc(1) will stop executing the current input and reset (see the \f[B]RESET\f[R] section) upon receiving a \f[B]SIGINT\f[R]. .PP Note that \[lq]current input\[rq] can mean one of two things. If dc(1) is processing input from \f[B]stdin\f[R] in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from \f[B]stdin\f[R] if no other file exists. .PP This means that if a \f[B]SIGINT\f[R] is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. .PP \f[B]SIGTERM\f[R] and \f[B]SIGQUIT\f[R] cause dc(1) to clean up and exit, and it uses the default handler for all other signals. The one exception is \f[B]SIGHUP\f[R]; in that case, and only when dc(1) is in TTY mode (see the \f[B]TTY MODE\f[R] section), a \f[B]SIGHUP\f[R] will cause dc(1) to clean up and exit. .SH COMMAND LINE HISTORY dc(1) supports interactive command\-line editing. .PP If dc(1) can be in TTY mode (see the \f[B]TTY MODE\f[R] section), history can be enabled. This means that command\-line history can only be enabled when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY. .PP Like TTY mode itself, it can be turned on or off with the environment variable \f[B]DC_TTY_MODE\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP \f[B]Note\f[R]: tabs are converted to 8 spaces. .SH LOCALES This dc(1) ships with support for adding error messages for different locales and thus, supports \f[B]LC_MESSAGES\f[R]. .SH SEE ALSO bc(1) .SH STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1\-2017 (\[lq]POSIX.1\-2017\[rq]) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . .SH BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . .SH AUTHOR Gavin D. Howard \c .MT gavin@gavinhoward.com .ME \c \ and contributors. diff --git a/manuals/dc/E.1.md b/manuals/dc/E.1.md index 3a47f789bd3e..54b877999d0d 100644 --- a/manuals/dc/E.1.md +++ b/manuals/dc/E.1.md @@ -1,1350 +1,1353 @@ # Name dc - arbitrary-precision decimal reverse-Polish notation calculator # SYNOPSIS **dc** [**-cChiPRvVx**] [**-\-version**] [**-\-help**] [**-\-digit-clamp**] [**-\-no-digit-clamp**] [**-\-interactive**] [**-\-no-prompt**] [**-\-no-read-prompt**] [**-\-extended-register**] [**-e** *expr*] [**-\-expression**=*expr*...] [**-f** *file*...] [**-\-file**=*file*...] [*file*...] # DESCRIPTION dc(1) is an arbitrary-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. If no files are given on the command-line, then dc(1) reads from **stdin** (see the **STDIN** section). Otherwise, those files are processed, and dc(1) will then exit. If a user wants to set up a standard environment, they can use **DC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). For example, if a user wants the **scale** always set to **10**, they can set **DC_ENV_ARGS** to **-e 10k**, and this dc(1) will always start with a **scale** of **10**. # OPTIONS The following are the options that dc(1) accepts. **-C**, **-\-no-digit-clamp** : Disables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that the value added to a number from a digit is always that digit's value multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-c** or **-\-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-c**, **-\-digit-clamp** : Enables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-C** or **-\-no-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-e** *expr*, **-\-expression**=*expr* : Evaluates *expr*. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**, whether on the command-line or in **DC_ENV_ARGS**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-f** *file*, **-\-file**=*file* : Reads in *file* and evaluates it, line by line, as though it were read through **stdin**. If expressions are also given (see above), the expressions are evaluated in the order given. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-h**, **-\-help** : Prints a usage message and exits. **-I** *ibase*, **-\-ibase**=*ibase* : Sets the builtin variable **ibase** to the value *ibase* assuming that *ibase* is in base 10. It is a fatal error if *ibase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-i**, **-\-interactive** : Forces interactive mode. (See the **INTERACTIVE MODE** section.) This is a **non-portable extension**. **-L**, **-\-no-line-length** : Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets **BC_LINE_LENGTH** to **0** (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-O** *obase*, **-\-obase**=*obase* : Sets the builtin variable **obase** to the value *obase* assuming that *obase* is in base 10. It is a fatal error if *obase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-P**, **-\-no-prompt** : Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in **DC_ENV_ARGS**. These options override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-R**, **-\-no-read-prompt** : Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in **BC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. This option does not disable the regular prompt because the read prompt is only used when the **?** command is used. These options *do* override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section), but only for the read prompt. This is a **non-portable extension**. **-S** *scale*, **-\-scale**=*scale* : Sets the builtin variable **scale** to the value *scale* assuming that *scale* is in base 10. It is a fatal error if *scale* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-v**, **-V**, **-\-version** : Print the version information (copyright header) and exits. **-x** **-\-extended-register** : Enables extended register mode. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. This is a **non-portable extension**. **-z**, **-\-leading-zeroes** : Makes dc(1) print all numbers greater than **-1** and less than **1**, and not equal to **0**, with a leading zero. This is a **non-portable extension**. All long options are **non-portable extensions**. # STDIN If no files are given on the command-line and no files or expressions are given by the **-f**, **-\-file**, **-e**, or **-\-expression** options, then dc(1) reads from **stdin**. However, there is a caveat to this. First, **stdin** is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. # STDOUT Any non-error output is written to **stdout**. In addition, if history (see the **HISTORY** section) and the prompt (see the **TTY MODE** section) are enabled, both are output to **stdout**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stdout**, so if **stdout** is closed, as in **dc >&-**, it will quit with an error. This is done so that dc(1) can report problems when **stdout** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stdout** to **/dev/null**. # STDERR Any error output is written to **stderr**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stderr**, so if **stderr** is closed, as in **dc 2>&-**, it will quit with an error. This is done so that dc(1) can exit with an error code when **stderr** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stderr** to **/dev/null**. # SYNTAX Each item in the input source code, either a number (see the **NUMBERS** section) or a command (see the **COMMANDS** section), is processed and executed, in order. Input is processed immediately when entered. **ibase** is a register (see the **REGISTERS** section) that determines how to interpret constant numbers. It is the "input" base, or the number base used for interpreting input numbers. **ibase** is initially **10**. The max allowable value for **ibase** is **16**. The min allowable value for **ibase** is **2**. The max allowable value for **ibase** can be queried in dc(1) programs with the **T** command. **obase** is a register (see the **REGISTERS** section) that determines how to output results. It is the "output" base, or the number base used for outputting numbers. **obase** is initially **10**. The max allowable value for **obase** is **DC_BASE_MAX** and can be queried with the **U** command. The min allowable value for **obase** is **2**. Values are output in the specified base. The *scale* of an expression is the number of digits in the result of the expression right of the decimal point, and **scale** is a register (see the **REGISTERS** section) that sets the precision of any operations (with exceptions). **scale** is initially **0**. **scale** cannot be negative. The max allowable value for **scale** can be queried in dc(1) programs with the **V** command. ## Comments Comments go from **#** until, and not including, the next newline. This is a **non-portable extension**. # NUMBERS Numbers are strings made up of digits, uppercase letters up to **F**, and at most **1** period for a radix. Numbers can have up to **DC_NUM_MAX** digits. Uppercase letters are equal to **9** plus their position in the alphabet (i.e., **A** equals **10**, or **9+1**). If a digit or letter makes no sense with the current value of **ibase** (i.e., they are greater than or equal to the current value of **ibase**), then the behavior depends on the existence of the **-c**/**-\-digit-clamp** or **-C**/**-\-no-digit-clamp** options (see the **OPTIONS** section), the existence and setting of the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section), or the default, which can be queried with the **-h**/**-\-help** option. If clamping is off, then digits or letters that are greater than or equal to the current value of **ibase** are not changed. Instead, their given value is multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*A+3\^0\*B**, which is **3** times **10** plus **11**, or **41**. If clamping is on, then digits or letters that are greater than or equal to the current value of **ibase** are set to the value of the highest valid digit in **ibase** before being multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*2+3\^0\*2**, which is **3** times **2** plus **2**, or **8**. There is one exception to clamping: single-character numbers (i.e., **A** alone). Such numbers are never clamped and always take the value they would have in the highest possible **ibase**. This means that **A** alone always equals decimal **10** and **Z** alone always equals decimal **35**. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current **ibase** (with the **i** command) regardless of the current value of **ibase**. If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for **A**, use **0A**. # COMMANDS The valid commands are listed below. ## Printing These commands are used for printing. **p** : Prints the value on top of the stack, whether number or string, and prints a newline after. This does not alter the stack. **n** : Prints the value on top of the stack, whether number or string, and pops it off of the stack. **P** : Pops a value off the stack. If the value is a number, it is truncated and the absolute value of the result is printed as though **obase** is **256** and each digit is interpreted as an 8-bit ASCII character, making it a byte stream. If the value is a string, it is printed without a trailing newline. This is a **non-portable extension**. **f** : Prints the entire contents of the stack, in order from newest to oldest, without altering anything. Users should use this command when they get lost. ## Arithmetic These are the commands used for arithmetic. **+** : The top two values are popped off the stack, added, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **-** : The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **\*** : The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If **a** is the *scale* of the first expression and **b** is the *scale* of the second expression, the *scale* of the result is equal to **min(a+b,max(scale,a,b))** where **min()** and **max()** return the obvious values. **/** : The top two values are popped off the stack, divided, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be non-zero. **%** : The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. Remaindering is equivalent to 1) Computing **a/b** to current **scale**, and 2) Using the result of step 1 to calculate **a-(a/b)\*b** to *scale* **max(scale+scale(b),scale(a))**. The first value popped off of the stack must be non-zero. **~** : The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to **x y / x y %** except that **x** and **y** are only evaluated once. The first value popped off of the stack must be non-zero. This is a **non-portable extension**. **\^** : The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non-zero. **v** : The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The value popped off of the stack must be non-negative. **\_** : If this command *immediately* precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a **non-portable extension**. **b** : The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. This is a **non-portable extension**. **|** : The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. The first value popped is used as the reduction modulus and must be an integer and non-zero. The second value popped is used as the exponent and must be an integer and non-negative. The third value popped is the base and must be an integer. This is a **non-portable extension**. **G** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if they are equal, or **0** otherwise. This is a **non-portable extension**. **N** : The top value is popped off of the stack, and if it a **0**, a **1** is pushed; otherwise, a **0** is pushed. This is a **non-portable extension**. **(** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than the second, or **0** otherwise. This is a **non-portable extension**. **{** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **)** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than the second, or **0** otherwise. This is a **non-portable extension**. **}** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **M** : The top two values are popped off of the stack. If they are both non-zero, a **1** is pushed onto the stack. If either of them is zero, or both of them are, then a **0** is pushed onto the stack. This is like the **&&** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. **m** : The top two values are popped off of the stack. If at least one of them is non-zero, a **1** is pushed onto the stack. If both of them are zero, then a **0** is pushed onto the stack. This is like the **||** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. ## Stack Control These commands control the stack. **c** : Removes all items from ("clears") the stack. **d** : Copies the item on top of the stack ("duplicates") and pushes the copy onto the stack. **r** : Swaps ("reverses") the two top items on the stack. **R** : Pops ("removes") the top value from the stack. ## Register Control These commands control registers (see the **REGISTERS** section). **s**_r_ : Pops the value off the top of the stack and stores it into register *r*. **l**_r_ : Copies the value in register *r* and pushes it onto the stack. This does not alter the contents of *r*. **S**_r_ : Pops the value off the top of the (main) stack and pushes it onto the stack of register *r*. The previous value of the register becomes inaccessible. **L**_r_ : Pops the value off the top of the stack for register *r* and push it onto the main stack. The previous value in the stack for register *r*, if any, is now accessible via the **l**_r_ command. ## Parameters These commands control the values of **ibase**, **obase**, and **scale**. Also see the **SYNTAX** section. **i** : Pops the value off of the top of the stack and uses it to set **ibase**, which must be between **2** and **16**, inclusive. If the value on top of the stack has any *scale*, the *scale* is ignored. **o** : Pops the value off of the top of the stack and uses it to set **obase**, which must be between **2** and **DC_BASE_MAX**, inclusive (see the **LIMITS** section). If the value on top of the stack has any *scale*, the *scale* is ignored. **k** : Pops the value off of the top of the stack and uses it to set **scale**, which must be non-negative. If the value on top of the stack has any *scale*, the *scale* is ignored. **I** : Pushes the current value of **ibase** onto the main stack. **O** : Pushes the current value of **obase** onto the main stack. **K** : Pushes the current value of **scale** onto the main stack. **T** : Pushes the maximum allowable value of **ibase** onto the main stack. This is a **non-portable extension**. **U** : Pushes the maximum allowable value of **obase** onto the main stack. This is a **non-portable extension**. **V** : Pushes the maximum allowable value of **scale** onto the main stack. This is a **non-portable extension**. ## Strings The following commands control strings. dc(1) can work with both numbers and strings, and registers (see the **REGISTERS** section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. Strings can also be executed as macros. For example, if the string **[1pR]** is executed as a macro, then the code **1pR** is executed, meaning that the **1** will be printed with a newline after and then popped from the stack. **\[**_characters_**\]** : Makes a string containing *characters* and pushes it onto the stack. If there are brackets (**\[** and **\]**) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (**\\**) character. If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. **a** : The value on top of the stack is popped. If it is a number, it is truncated and its absolute value is taken. The result mod **256** is calculated. If that result is **0**, push an empty string; otherwise, push a one-character string where the character is the result of the mod interpreted as an ASCII character. If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one-character string. The new string is then pushed onto the stack. This is a **non-portable extension**. **x** : Pops a value off of the top of the stack. If it is a number, it is pushed back onto the stack. If it is a string, it is executed as a macro. This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. **\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register *r* are executed. For example, **0 1>a** will execute the contents of register **a**, and **1 0>a** will not. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **?** : Reads a line from the **stdin** and executes it. This is to allow macros to request input from users. **q** : During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. **Q** : Pops a value from the stack which must be non-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. **,** : Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the **Q** command, so the sequence **,Q** will make dc(1) exit. This is a **non-portable extension**. ## Status These commands query status of the stack or its top value. **Z** : Pops a value off of the stack. If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push **1** if the argument is **0** with no decimal places. If it is a string, pushes the number of characters the string has. **X** : Pops a value off of the stack. If it is a number, pushes the *scale* of the value onto the stack. If it is a string, pushes **0**. **u** : Pops one value off of the stack. If the value is a number, this pushes **1** onto the stack. Otherwise (if it is a string), it pushes **0**. This is a **non-portable extension**. **t** : Pops one value off of the stack. If the value is a string, this pushes **1** onto the stack. Otherwise (if it is a number), it pushes **0**. This is a **non-portable extension**. **z** : Pushes the current depth of the stack (before execution of this command) onto the stack. **y**_r_ : Pushes the current stack depth of the register *r* onto the main stack. Because each register has a depth of **1** (with the value **0** in the top item) when dc(1) starts, dc(1) requires that each register's stack must always have at least one item; dc(1) will give an error and reset otherwise (see the **RESET** section). This means that this command will never push **0**. This is a **non-portable extension**. ## Arrays These commands manipulate arrays. **:**_r_ : Pops the top two values off of the stack. The second value will be stored in the array *r* (see the **REGISTERS** section), indexed by the first value. **;**_r_ : Pops the value on top of the stack and uses it as an index into the array *r*. The selected value is then pushed onto the stack. **Y**_r_ : Pushes the length of the array *r* onto the stack. This is a **non-portable extension**. ## Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter **g**. Only the characters below are allowed after the character **g**; any other character produces a parse error (see the **ERRORS** section). **gl** : Pushes the line length set by **DC_LINE_LENGTH** (see the **ENVIRONMENT VARIABLES** section) onto the stack. **gx** : Pushes **1** onto the stack if extended register mode is on, **0** otherwise. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. **gz** : Pushes **0** onto the stack if the leading zero setting has not been enabled with the **-z** or **-\-leading-zeroes** options (see the **OPTIONS** section), non-zero otherwise. # REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (**0**) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. In non-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (**'\\n'**) and a left bracket (**'['**); it is a parse error for a newline or a left bracket to be used as a register name. ## Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. If extended register mode is enabled (**-x** or **-\-extended-register** command-line arguments are given), then normal single character registers are used *unless* the character immediately following a command that needs a register name is a space (according to **isspace()**) and not a newline (**'\\n'**). In that case, the register name is found according to the regex **\[a-z\]\[a-z0-9\_\]\*** (like bc(1) identifiers), and it is a parse error if the next non-space characters do not match that regex. # RESET When dc(1) encounters an error or a signal that it has a non-default handler for, it resets. This means that several things happen. -First, any macros that are executing are stopped and popped off the stack. -The behavior is not unlike that of exceptions in programming languages. Then -the execution point is set so that any code waiting to execute (after all +First, any macros that are executing are stopped and popped off the execution +stack. The behavior is not unlike that of exceptions in programming languages. +Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. +However, the stack of values is *not* cleared; in interactive mode, users can +inspect the stack and manipulate it. + Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the **EXIT STATUS** section), it asks for more input; otherwise, it exits with the appropriate return code. # PERFORMANCE Most dc(1) implementations use **char** types to calculate the value of **1** decimal digit at a time, but that can be slow. This dc(1) does something different. It uses large integers to calculate more than **1** decimal digit at a time. If built in a environment where **DC_LONG_BIT** (see the **LIMITS** section) is **64**, then each integer has **9** decimal digits. If built in an environment where **DC_LONG_BIT** is **32** then each integer has **4** decimal digits. This value (the number of decimal digits per large integer) is called **DC_BASE_DIGS**. In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of **DC_LONG_BIT**, but is always at least twice as large as the integer type used to store digits. # LIMITS The following are the limits on dc(1): **DC_LONG_BIT** : The number of bits in the **long** type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the **PERFORMANCE** section). **DC_BASE_DIGS** : The number of decimal digits per large integer (see the **PERFORMANCE** section). Depends on **DC_LONG_BIT**. **DC_BASE_POW** : The max decimal number that each large integer can store (see **DC_BASE_DIGS**) plus **1**. Depends on **DC_BASE_DIGS**. **DC_OVERFLOW_MAX** : The max number that the overflow type (see the **PERFORMANCE** section) can hold. Depends on **DC_LONG_BIT**. **DC_BASE_MAX** : The maximum output base. Set at **DC_BASE_POW**. **DC_DIM_MAX** : The maximum size of arrays. Set at **SIZE_MAX-1**. **DC_SCALE_MAX** : The maximum **scale**. Set at **DC_OVERFLOW_MAX-1**. **DC_STRING_MAX** : The maximum length of strings. Set at **DC_OVERFLOW_MAX-1**. **DC_NAME_MAX** : The maximum length of identifiers. Set at **DC_OVERFLOW_MAX-1**. **DC_NUM_MAX** : The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at **DC_OVERFLOW_MAX-1**. Exponent : The maximum allowable exponent (positive or negative). Set at **DC_OVERFLOW_MAX**. Number of vars : The maximum number of vars/arrays. Set at **SIZE_MAX-1**. These limits are meant to be effectively non-existent; the limits are so large (at least on 64-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. # ENVIRONMENT VARIABLES As **non-portable extensions**, dc(1) recognizes the following environment variables: **DC_ENV_ARGS** : This is another way to give command-line arguments to dc(1). They should be in the same format as all other command-line arguments. These are always processed first, so any files given in **DC_ENV_ARGS** will be processed before arguments and files given on the command-line. This gives the user the ability to set up "standard" options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the **-e** option to set **scale** to a value other than **0**. The code that parses **DC_ENV_ARGS** will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string **"/home/gavin/some dc file.dc"** will be correctly parsed, but the string **"/home/gavin/some \"dc\" file.dc"** will include the backslashes. The quote parsing will handle either kind of quotes, **'** or **"**. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in **"some 'dc' file.dc"**, and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in **DC_ENV_ARGS** is not supported due to the complexity of the parsing, though such files are still supported on the command-line where the parsing is done by the shell. **DC_LINE_LENGTH** : If this environment variable exists and contains an integer that is greater than **1** and is less than **UINT16_MAX** (**2\^16-1**), dc(1) will output lines to that length, including the backslash newline combo. The default line length is **70**. The special value of **0** will disable line length checking and print numbers without regard to line length and without backslashes and newlines. **DC_SIGINT_RESET** : If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), then this environment variable has no effect because dc(1) exits on **SIGINT** when not in interactive mode. However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) reset on **SIGINT**, rather than exit, and zero makes dc(1) exit. If this environment variable exists and is *not* an integer, then dc(1) will exit on **SIGINT**. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_TTY_MODE** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_PROMPT** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) use a prompt, and zero or a non-integer makes dc(1) not use a prompt. If this environment variable does not exist and **DC_TTY_MODE** does, then the value of the **DC_TTY_MODE** environment variable is used. This environment variable and the **DC_TTY_MODE** environment variable override the default, which can be queried with the **-h** or **-\-help** options. **DC_EXPR_EXIT** : If any expressions or expression files are given on the command-line with **-e**, **-\-expression**, **-f**, or **-\-file**, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_DIGIT_CLAMP** : When parsing numbers and if this environment variable exists and contains an integer, a non-zero value makes dc(1) clamp digits that are greater than or equal to the current **ibase** so that all such digits are considered equal to the **ibase** minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the **ibase**. This never applies to single-digit numbers, as per the bc(1) standard (see the **STANDARDS** section). This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. # EXIT STATUS dc(1) returns the following exit statuses: **0** : No error. **1** : A math error occurred. This follows standard practice of using **1** for expected errors, since math errors will happen in the process of normal execution. Math errors include divide by **0**, taking the square root of a negative number, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non-integer where an integer is required. Converting to a hardware integer happens for the second operand of the power (**\^**) operator. **2** : A parse error occurred. Parse errors include unexpected **EOF**, using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. **3** : A runtime error occurred. Runtime errors include assigning an invalid number to any global (**ibase**, **obase**, or **scale**), giving a bad expression to a **read()** call, calling **read()** inside of a **read()** call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. **4** : A fatal error occurred. Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command-line options. The exit status **4** is special; when a fatal error occurs, dc(1) always exits and returns **4**, no matter what mode dc(1) is in. The other statuses will only be returned when dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), since dc(1) resets its state (see the **RESET** section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the **-i** flag or **-\-interactive** option. These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the **-i** flag or **-\-interactive** option. # INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non-interactive mode. Interactive mode is turned on automatically when both **stdin** and **stdout** are hooked to a terminal, but the **-i** flag and **-\-interactive** option can turn it on in other situations. In interactive mode, dc(1) attempts to recover from errors (see the **RESET** section), and in normal execution, flushes **stdout** as soon as execution is done for the current input. dc(1) may also reset on **SIGINT** instead of exit, depending on the contents of, or default for, the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section). # TTY MODE If **stdin**, **stdout**, and **stderr** are all connected to a TTY, then "TTY mode" is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. If there is the environment variable **DC_TTY_MODE** in the environment (see the **ENVIRONMENT VARIABLES** section), then if that environment variable contains a non-zero integer, dc(1) will turn on TTY mode when **stdin**, **stdout**, and **stderr** are all connected to a TTY. If the **DC_TTY_MODE** environment variable exists but is *not* a non-zero integer, then dc(1) will not turn TTY mode on. If the environment variable **DC_TTY_MODE** does *not* exist, the default setting is used. The default setting can be queried with the **-h** or **-\-help** options. TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the **STANDARDS** section), and interactive mode requires only **stdin** and **stdout** to be connected to a terminal. ## Command-Line History Command-line history is only enabled if TTY mode is, i.e., that **stdin**, **stdout**, and **stderr** are connected to a TTY and the **DC_TTY_MODE** environment variable (see the **ENVIRONMENT VARIABLES** section) and its default do not disable TTY mode. See the **COMMAND LINE HISTORY** section for more information. ## Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: **DC_PROMPT** (see the **ENVIRONMENT VARIABLES** section). If the environment variable **DC_PROMPT** exists and is a non-zero integer, then the prompt is turned on when **stdin**, **stdout**, and **stderr** are connected to a TTY and the **-P** and **-\-no-prompt** options were not used. The read prompt will be turned on under the same conditions, except that the **-R** and **-\-no-read-prompt** options must also not be used. However, if **DC_PROMPT** does not exist, the prompt can be enabled or disabled with the **DC_TTY_MODE** environment variable, the **-P** and **-\-no-prompt** options, and the **-R** and **-\-no-read-prompt** options. See the **ENVIRONMENT VARIABLES** and **OPTIONS** sections for more details. # SIGNAL HANDLING Sending a **SIGINT** will cause dc(1) to do one of two things. If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), or the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section), or its default, is either not an integer or it is zero, dc(1) will exit. However, if dc(1) is in interactive mode, and the **DC_SIGINT_RESET** or its default is an integer and non-zero, then dc(1) will stop executing the current input and reset (see the **RESET** section) upon receiving a **SIGINT**. Note that "current input" can mean one of two things. If dc(1) is processing input from **stdin** in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from **stdin** if no other file exists. This means that if a **SIGINT** is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. **SIGTERM** and **SIGQUIT** cause dc(1) to clean up and exit, and it uses the default handler for all other signals. The one exception is **SIGHUP**; in that case, and only when dc(1) is in TTY mode (see the **TTY MODE** section), a **SIGHUP** will cause dc(1) to clean up and exit. # COMMAND LINE HISTORY dc(1) supports interactive command-line editing. If dc(1) can be in TTY mode (see the **TTY MODE** section), history can be enabled. This means that command-line history can only be enabled when **stdin**, **stdout**, and **stderr** are all connected to a TTY. Like TTY mode itself, it can be turned on or off with the environment variable **DC_TTY_MODE** (see the **ENVIRONMENT VARIABLES** section). **Note**: tabs are converted to 8 spaces. # LOCALES This dc(1) ships with support for adding error messages for different locales and thus, supports **LC_MESSAGES**. # SEE ALSO bc(1) # STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1-2017 (“POSIX.1-2017”) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . # BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . # AUTHOR Gavin D. Howard and contributors. diff --git a/manuals/dc/EH.1 b/manuals/dc/EH.1 index e60e6e0d8497..61fbaa4efe92 100644 --- a/manuals/dc/EH.1 +++ b/manuals/dc/EH.1 @@ -1,1447 +1,1450 @@ .\" .\" SPDX-License-Identifier: BSD-2-Clause .\" .\" Copyright (c) 2018-2024 Gavin D. Howard and contributors. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions are met: .\" .\" * Redistributions of source code must retain the above copyright notice, .\" this list of conditions and the following disclaimer. .\" .\" * Redistributions in binary form must reproduce the above copyright notice, .\" this list of conditions and the following disclaimer in the documentation .\" and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" .\" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE .\" LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF .\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS .\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN .\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) .\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE .\" POSSIBILITY OF SUCH DAMAGE. .\" -.TH "DC" "1" "January 2024" "Gavin D. Howard" "General Commands Manual" +.TH "DC" "1" "August 2024" "Gavin D. Howard" "General Commands Manual" .nh .ad l .SH Name dc \- arbitrary\-precision decimal reverse\-Polish notation calculator .SH SYNOPSIS \f[B]dc\f[R] [\f[B]\-cChiPRvVx\f[R]] [\f[B]\-\-version\f[R]] [\f[B]\-\-help\f[R]] [\f[B]\-\-digit\-clamp\f[R]] [\f[B]\-\-no\-digit\-clamp\f[R]] [\f[B]\-\-interactive\f[R]] [\f[B]\-\-no\-prompt\f[R]] [\f[B]\-\-no\-read\-prompt\f[R]] [\f[B]\-\-extended\-register\f[R]] [\f[B]\-e\f[R] \f[I]expr\f[R]] [\f[B]\-\-expression\f[R]=\f[I]expr\f[R]\&...] [\f[B]\-f\f[R] \f[I]file\f[R]\&...] [\f[B]\-\-file\f[R]=\f[I]file\f[R]\&...] [\f[I]file\f[R]\&...] .SH DESCRIPTION dc(1) is an arbitrary\-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. .PP If no files are given on the command\-line, then dc(1) reads from \f[B]stdin\f[R] (see the \f[B]STDIN\f[R] section). Otherwise, those files are processed, and dc(1) will then exit. .PP If a user wants to set up a standard environment, they can use \f[B]DC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). For example, if a user wants the \f[B]scale\f[R] always set to \f[B]10\f[R], they can set \f[B]DC_ENV_ARGS\f[R] to \f[B]\-e 10k\f[R], and this dc(1) will always start with a \f[B]scale\f[R] of \f[B]10\f[R]. .SH OPTIONS The following are the options that dc(1) accepts. .TP \f[B]\-C\f[R], \f[B]\-\-no\-digit\-clamp\f[R] Disables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that the value added to a number from a digit is always that digit\[cq]s value multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-c\f[R] or \f[B]\-\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-c\f[R], \f[B]\-\-digit\-clamp\f[R] Enables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-C\f[R] or \f[B]\-\-no\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-e\f[R] \f[I]expr\f[R], \f[B]\-\-expression\f[R]=\f[I]expr\f[R] Evaluates \f[I]expr\f[R]. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R], whether on the command\-line or in \f[B]DC_ENV_ARGS\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-f\f[R] \f[I]file\f[R], \f[B]\-\-file\f[R]=\f[I]file\f[R] Reads in \f[I]file\f[R] and evaluates it, line by line, as though it were read through \f[B]stdin\f[R]. If expressions are also given (see above), the expressions are evaluated in the order given. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-h\f[R], \f[B]\-\-help\f[R] Prints a usage message and exits. .TP \f[B]\-I\f[R] \f[I]ibase\f[R], \f[B]\-\-ibase\f[R]=\f[I]ibase\f[R] Sets the builtin variable \f[B]ibase\f[R] to the value \f[I]ibase\f[R] assuming that \f[I]ibase\f[R] is in base 10. It is a fatal error if \f[I]ibase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-i\f[R], \f[B]\-\-interactive\f[R] Forces interactive mode. (See the \f[B]INTERACTIVE MODE\f[R] section.) .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-L\f[R], \f[B]\-\-no\-line\-length\f[R] Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets \f[B]BC_LINE_LENGTH\f[R] to \f[B]0\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-O\f[R] \f[I]obase\f[R], \f[B]\-\-obase\f[R]=\f[I]obase\f[R] Sets the builtin variable \f[B]obase\f[R] to the value \f[I]obase\f[R] assuming that \f[I]obase\f[R] is in base 10. It is a fatal error if \f[I]obase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-P\f[R], \f[B]\-\-no\-prompt\f[R] Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]DC_ENV_ARGS\f[R]. .RS .PP These options override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-R\f[R], \f[B]\-\-no\-read\-prompt\f[R] Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]BC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. .RS .PP This option does not disable the regular prompt because the read prompt is only used when the \f[B]?\f[R] command is used. .PP These options \f[I]do\f[R] override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), but only for the read prompt. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-S\f[R] \f[I]scale\f[R], \f[B]\-\-scale\f[R]=\f[I]scale\f[R] Sets the builtin variable \f[B]scale\f[R] to the value \f[I]scale\f[R] assuming that \f[I]scale\f[R] is in base 10. It is a fatal error if \f[I]scale\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-v\f[R], \f[B]\-V\f[R], \f[B]\-\-version\f[R] Print the version information (copyright header) and exits. .TP \f[B]\-x\f[R] \f[B]\-\-extended\-register\f[R] Enables extended register mode. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-z\f[R], \f[B]\-\-leading\-zeroes\f[R] Makes dc(1) print all numbers greater than \f[B]\-1\f[R] and less than \f[B]1\f[R], and not equal to \f[B]0\f[R], with a leading zero. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .PP All long options are \f[B]non\-portable extensions\f[R]. .SH STDIN If no files are given on the command\-line and no files or expressions are given by the \f[B]\-f\f[R], \f[B]\-\-file\f[R], \f[B]\-e\f[R], or \f[B]\-\-expression\f[R] options, then dc(1) reads from \f[B]stdin\f[R]. .PP However, there is a caveat to this. .PP First, \f[B]stdin\f[R] is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. .SH STDOUT Any non\-error output is written to \f[B]stdout\f[R]. In addition, if history (see the \f[B]HISTORY\f[R] section) and the prompt (see the \f[B]TTY MODE\f[R] section) are enabled, both are output to \f[B]stdout\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stdout\f[R], so if \f[B]stdout\f[R] is closed, as in \f[B]dc >&\-\f[R], it will quit with an error. This is done so that dc(1) can report problems when \f[B]stdout\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stdout\f[R] to \f[B]/dev/null\f[R]. .SH STDERR Any error output is written to \f[B]stderr\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stderr\f[R], so if \f[B]stderr\f[R] is closed, as in \f[B]dc 2>&\-\f[R], it will quit with an error. This is done so that dc(1) can exit with an error code when \f[B]stderr\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stderr\f[R] to \f[B]/dev/null\f[R]. .SH SYNTAX Each item in the input source code, either a number (see the \f[B]NUMBERS\f[R] section) or a command (see the \f[B]COMMANDS\f[R] section), is processed and executed, in order. Input is processed immediately when entered. .PP \f[B]ibase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to interpret constant numbers. It is the \[lq]input\[rq] base, or the number base used for interpreting input numbers. \f[B]ibase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]ibase\f[R] is \f[B]16\f[R]. The min allowable value for \f[B]ibase\f[R] is \f[B]2\f[R]. The max allowable value for \f[B]ibase\f[R] can be queried in dc(1) programs with the \f[B]T\f[R] command. .PP \f[B]obase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to output results. It is the \[lq]output\[rq] base, or the number base used for outputting numbers. \f[B]obase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]obase\f[R] is \f[B]DC_BASE_MAX\f[R] and can be queried with the \f[B]U\f[R] command. The min allowable value for \f[B]obase\f[R] is \f[B]2\f[R]. Values are output in the specified base. .PP The \f[I]scale\f[R] of an expression is the number of digits in the result of the expression right of the decimal point, and \f[B]scale\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that sets the precision of any operations (with exceptions). \f[B]scale\f[R] is initially \f[B]0\f[R]. \f[B]scale\f[R] cannot be negative. The max allowable value for \f[B]scale\f[R] can be queried in dc(1) programs with the \f[B]V\f[R] command. .SS Comments Comments go from \f[B]#\f[R] until, and not including, the next newline. This is a \f[B]non\-portable extension\f[R]. .SH NUMBERS Numbers are strings made up of digits, uppercase letters up to \f[B]F\f[R], and at most \f[B]1\f[R] period for a radix. Numbers can have up to \f[B]DC_NUM_MAX\f[R] digits. Uppercase letters are equal to \f[B]9\f[R] plus their position in the alphabet (i.e., \f[B]A\f[R] equals \f[B]10\f[R], or \f[B]9+1\f[R]). .PP If a digit or letter makes no sense with the current value of \f[B]ibase\f[R] (i.e., they are greater than or equal to the current value of \f[B]ibase\f[R]), then the behavior depends on the existence of the \f[B]\-c\f[R]/\f[B]\-\-digit\-clamp\f[R] or \f[B]\-C\f[R]/\f[B]\-\-no\-digit\-clamp\f[R] options (see the \f[B]OPTIONS\f[R] section), the existence and setting of the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or the default, which can be queried with the \f[B]\-h\f[R]/\f[B]\-\-help\f[R] option. .PP If clamping is off, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are not changed. Instead, their given value is multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*A+3\[ha]0*B\f[R], which is \f[B]3\f[R] times \f[B]10\f[R] plus \f[B]11\f[R], or \f[B]41\f[R]. .PP If clamping is on, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are set to the value of the highest valid digit in \f[B]ibase\f[R] before being multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*2+3\[ha]0*2\f[R], which is \f[B]3\f[R] times \f[B]2\f[R] plus \f[B]2\f[R], or \f[B]8\f[R]. .PP There is one exception to clamping: single\-character numbers (i.e., \f[B]A\f[R] alone). Such numbers are never clamped and always take the value they would have in the highest possible \f[B]ibase\f[R]. This means that \f[B]A\f[R] alone always equals decimal \f[B]10\f[R] and \f[B]Z\f[R] alone always equals decimal \f[B]35\f[R]. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current \f[B]ibase\f[R] (with the \f[B]i\f[R] command) regardless of the current value of \f[B]ibase\f[R]. .PP If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for \f[B]A\f[R], use \f[B]0A\f[R]. .SH COMMANDS The valid commands are listed below. .SS Printing These commands are used for printing. .TP \f[B]p\f[R] Prints the value on top of the stack, whether number or string, and prints a newline after. .RS .PP This does not alter the stack. .RE .TP \f[B]n\f[R] Prints the value on top of the stack, whether number or string, and pops it off of the stack. .TP \f[B]P\f[R] Pops a value off the stack. .RS .PP If the value is a number, it is truncated and the absolute value of the result is printed as though \f[B]obase\f[R] is \f[B]256\f[R] and each digit is interpreted as an 8\-bit ASCII character, making it a byte stream. .PP If the value is a string, it is printed without a trailing newline. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]f\f[R] Prints the entire contents of the stack, in order from newest to oldest, without altering anything. .RS .PP Users should use this command when they get lost. .RE .SS Arithmetic These are the commands used for arithmetic. .TP \f[B]+\f[R] The top two values are popped off the stack, added, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]\-\f[R] The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]*\f[R] The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If \f[B]a\f[R] is the \f[I]scale\f[R] of the first expression and \f[B]b\f[R] is the \f[I]scale\f[R] of the second expression, the \f[I]scale\f[R] of the result is equal to \f[B]min(a+b,max(scale,a,b))\f[R] where \f[B]min()\f[R] and \f[B]max()\f[R] return the obvious values. .TP \f[B]/\f[R] The top two values are popped off the stack, divided, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]%\f[R] The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. .RS .PP Remaindering is equivalent to 1) Computing \f[B]a/b\f[R] to current \f[B]scale\f[R], and 2) Using the result of step 1 to calculate \f[B]a\-(a/b)*b\f[R] to \f[I]scale\f[R] \f[B]max(scale+scale(b),scale(a))\f[R]. .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]\[ti]\f[R] The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to \f[B]x y / x y %\f[R] except that \f[B]x\f[R] and \f[B]y\f[R] are only evaluated once. .RS .PP The first value popped off of the stack must be non\-zero. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[ha]\f[R] The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non\-zero. .RE .TP \f[B]v\f[R] The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The value popped off of the stack must be non\-negative. .RE .TP \f[B]_\f[R] If this command \f[I]immediately\f[R] precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. .RS .PP Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]b\f[R] The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]|\f[R] The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. .RS .PP The first value popped is used as the reduction modulus and must be an integer and non\-zero. The second value popped is used as the exponent and must be an integer and non\-negative. The third value popped is the base and must be an integer. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]G\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if they are equal, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]N\f[R] The top value is popped off of the stack, and if it a \f[B]0\f[R], a \f[B]1\f[R] is pushed; otherwise, a \f[B]0\f[R] is pushed. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B](\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]{\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B])\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]}\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]M\f[R] The top two values are popped off of the stack. If they are both non\-zero, a \f[B]1\f[R] is pushed onto the stack. If either of them is zero, or both of them are, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]&&\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]m\f[R] The top two values are popped off of the stack. If at least one of them is non\-zero, a \f[B]1\f[R] is pushed onto the stack. If both of them are zero, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]||\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Stack Control These commands control the stack. .TP \f[B]c\f[R] Removes all items from (\[lq]clears\[rq]) the stack. .TP \f[B]d\f[R] Copies the item on top of the stack (\[lq]duplicates\[rq]) and pushes the copy onto the stack. .TP \f[B]r\f[R] Swaps (\[lq]reverses\[rq]) the two top items on the stack. .TP \f[B]R\f[R] Pops (\[lq]removes\[rq]) the top value from the stack. .SS Register Control These commands control registers (see the \f[B]REGISTERS\f[R] section). .TP \f[B]s\f[R]\f[I]r\f[R] Pops the value off the top of the stack and stores it into register \f[I]r\f[R]. .TP \f[B]l\f[R]\f[I]r\f[R] Copies the value in register \f[I]r\f[R] and pushes it onto the stack. This does not alter the contents of \f[I]r\f[R]. .TP \f[B]S\f[R]\f[I]r\f[R] Pops the value off the top of the (main) stack and pushes it onto the stack of register \f[I]r\f[R]. The previous value of the register becomes inaccessible. .TP \f[B]L\f[R]\f[I]r\f[R] Pops the value off the top of the stack for register \f[I]r\f[R] and push it onto the main stack. The previous value in the stack for register \f[I]r\f[R], if any, is now accessible via the \f[B]l\f[R]\f[I]r\f[R] command. .SS Parameters These commands control the values of \f[B]ibase\f[R], \f[B]obase\f[R], and \f[B]scale\f[R]. Also see the \f[B]SYNTAX\f[R] section. .TP \f[B]i\f[R] Pops the value off of the top of the stack and uses it to set \f[B]ibase\f[R], which must be between \f[B]2\f[R] and \f[B]16\f[R], inclusive. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]o\f[R] Pops the value off of the top of the stack and uses it to set \f[B]obase\f[R], which must be between \f[B]2\f[R] and \f[B]DC_BASE_MAX\f[R], inclusive (see the \f[B]LIMITS\f[R] section). .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]k\f[R] Pops the value off of the top of the stack and uses it to set \f[B]scale\f[R], which must be non\-negative. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]I\f[R] Pushes the current value of \f[B]ibase\f[R] onto the main stack. .TP \f[B]O\f[R] Pushes the current value of \f[B]obase\f[R] onto the main stack. .TP \f[B]K\f[R] Pushes the current value of \f[B]scale\f[R] onto the main stack. .TP \f[B]T\f[R] Pushes the maximum allowable value of \f[B]ibase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]U\f[R] Pushes the maximum allowable value of \f[B]obase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]V\f[R] Pushes the maximum allowable value of \f[B]scale\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Strings The following commands control strings. .PP dc(1) can work with both numbers and strings, and registers (see the \f[B]REGISTERS\f[R] section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. .PP While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. .PP Strings can also be executed as macros. For example, if the string \f[B][1pR]\f[R] is executed as a macro, then the code \f[B]1pR\f[R] is executed, meaning that the \f[B]1\f[R] will be printed with a newline after and then popped from the stack. .TP \f[B][\f[R]\f[I]characters\f[R]\f[B]]\f[R] Makes a string containing \f[I]characters\f[R] and pushes it onto the stack. .RS .PP If there are brackets (\f[B][\f[R] and \f[B]]\f[R]) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (\f[B]\[rs]\f[R]) character. .PP If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. .RE .TP \f[B]a\f[R] The value on top of the stack is popped. .RS .PP If it is a number, it is truncated and its absolute value is taken. The result mod \f[B]256\f[R] is calculated. If that result is \f[B]0\f[R], push an empty string; otherwise, push a one\-character string where the character is the result of the mod interpreted as an ASCII character. .PP If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one\-character string. The new string is then pushed onto the stack. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]x\f[R] Pops a value off of the top of the stack. .RS .PP If it is a number, it is pushed back onto the stack. .PP If it is a string, it is executed as a macro. .PP This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. .RE .TP \f[B]>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP For example, \f[B]0 1>a\f[R] will execute the contents of register \f[B]a\f[R], and \f[B]1 0>a\f[R] will not. .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]?\f[R] Reads a line from the \f[B]stdin\f[R] and executes it. This is to allow macros to request input from users. .TP \f[B]q\f[R] During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. .TP \f[B]Q\f[R] Pops a value from the stack which must be non\-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. .TP \f[B],\f[R] Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the \f[B]Q\f[R] command, so the sequence \f[B],Q\f[R] will make dc(1) exit. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Status These commands query status of the stack or its top value. .TP \f[B]Z\f[R] Pops a value off of the stack. .RS .PP If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push \f[B]1\f[R] if the argument is \f[B]0\f[R] with no decimal places. .PP If it is a string, pushes the number of characters the string has. .RE .TP \f[B]X\f[R] Pops a value off of the stack. .RS .PP If it is a number, pushes the \f[I]scale\f[R] of the value onto the stack. .PP If it is a string, pushes \f[B]0\f[R]. .RE .TP \f[B]u\f[R] Pops one value off of the stack. If the value is a number, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a string), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]t\f[R] Pops one value off of the stack. If the value is a string, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a number), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]z\f[R] Pushes the current depth of the stack (before execution of this command) onto the stack. .TP \f[B]y\f[R]\f[I]r\f[R] Pushes the current stack depth of the register \f[I]r\f[R] onto the main stack. .RS .PP Because each register has a depth of \f[B]1\f[R] (with the value \f[B]0\f[R] in the top item) when dc(1) starts, dc(1) requires that each register\[cq]s stack must always have at least one item; dc(1) will give an error and reset otherwise (see the \f[B]RESET\f[R] section). This means that this command will never push \f[B]0\f[R]. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Arrays These commands manipulate arrays. .TP \f[B]:\f[R]\f[I]r\f[R] Pops the top two values off of the stack. The second value will be stored in the array \f[I]r\f[R] (see the \f[B]REGISTERS\f[R] section), indexed by the first value. .TP \f[B];\f[R]\f[I]r\f[R] Pops the value on top of the stack and uses it as an index into the array \f[I]r\f[R]. The selected value is then pushed onto the stack. .TP \f[B]Y\f[R]\f[I]r\f[R] Pushes the length of the array \f[I]r\f[R] onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter \f[B]g\f[R]. Only the characters below are allowed after the character \f[B]g\f[R]; any other character produces a parse error (see the \f[B]ERRORS\f[R] section). .TP \f[B]gl\f[R] Pushes the line length set by \f[B]DC_LINE_LENGTH\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) onto the stack. .TP \f[B]gx\f[R] Pushes \f[B]1\f[R] onto the stack if extended register mode is on, \f[B]0\f[R] otherwise. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .TP \f[B]gz\f[R] Pushes \f[B]0\f[R] onto the stack if the leading zero setting has not been enabled with the \f[B]\-z\f[R] or \f[B]\-\-leading\-zeroes\f[R] options (see the \f[B]OPTIONS\f[R] section), non\-zero otherwise. .SH REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) .PP Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (\f[B]0\f[R]) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. .PP In non\-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (\f[B]`\[rs]n'\f[R]) and a left bracket (\f[B]`['\f[R]); it is a parse error for a newline or a left bracket to be used as a register name. .SS Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. .PP If extended register mode is enabled (\f[B]\-x\f[R] or \f[B]\-\-extended\-register\f[R] command\-line arguments are given), then normal single character registers are used \f[I]unless\f[R] the character immediately following a command that needs a register name is a space (according to \f[B]isspace()\f[R]) and not a newline (\f[B]`\[rs]n'\f[R]). .PP In that case, the register name is found according to the regex \f[B][a\-z][a\-z0\-9_]*\f[R] (like bc(1) identifiers), and it is a parse error if the next non\-space characters do not match that regex. .SH RESET When dc(1) encounters an error or a signal that it has a non\-default handler for, it resets. This means that several things happen. .PP First, any macros that are executing are stopped and popped off the -stack. +execution stack. The behavior is not unlike that of exceptions in programming languages. Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. .PP +However, the stack of values is \f[I]not\f[R] cleared; in interactive +mode, users can inspect the stack and manipulate it. +.PP Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the \f[B]EXIT STATUS\f[R] section), it asks for more input; otherwise, it exits with the appropriate return code. .SH PERFORMANCE Most dc(1) implementations use \f[B]char\f[R] types to calculate the value of \f[B]1\f[R] decimal digit at a time, but that can be slow. This dc(1) does something different. .PP It uses large integers to calculate more than \f[B]1\f[R] decimal digit at a time. If built in a environment where \f[B]DC_LONG_BIT\f[R] (see the \f[B]LIMITS\f[R] section) is \f[B]64\f[R], then each integer has \f[B]9\f[R] decimal digits. If built in an environment where \f[B]DC_LONG_BIT\f[R] is \f[B]32\f[R] then each integer has \f[B]4\f[R] decimal digits. This value (the number of decimal digits per large integer) is called \f[B]DC_BASE_DIGS\f[R]. .PP In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of \f[B]DC_LONG_BIT\f[R], but is always at least twice as large as the integer type used to store digits. .SH LIMITS The following are the limits on dc(1): .TP \f[B]DC_LONG_BIT\f[R] The number of bits in the \f[B]long\f[R] type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the \f[B]PERFORMANCE\f[R] section). .TP \f[B]DC_BASE_DIGS\f[R] The number of decimal digits per large integer (see the \f[B]PERFORMANCE\f[R] section). Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_POW\f[R] The max decimal number that each large integer can store (see \f[B]DC_BASE_DIGS\f[R]) plus \f[B]1\f[R]. Depends on \f[B]DC_BASE_DIGS\f[R]. .TP \f[B]DC_OVERFLOW_MAX\f[R] The max number that the overflow type (see the \f[B]PERFORMANCE\f[R] section) can hold. Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_MAX\f[R] The maximum output base. Set at \f[B]DC_BASE_POW\f[R]. .TP \f[B]DC_DIM_MAX\f[R] The maximum size of arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .TP \f[B]DC_SCALE_MAX\f[R] The maximum \f[B]scale\f[R]. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_STRING_MAX\f[R] The maximum length of strings. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NAME_MAX\f[R] The maximum length of identifiers. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NUM_MAX\f[R] The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP Exponent The maximum allowable exponent (positive or negative). Set at \f[B]DC_OVERFLOW_MAX\f[R]. .TP Number of vars The maximum number of vars/arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .PP These limits are meant to be effectively non\-existent; the limits are so large (at least on 64\-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. .SH ENVIRONMENT VARIABLES As \f[B]non\-portable extensions\f[R], dc(1) recognizes the following environment variables: .TP \f[B]DC_ENV_ARGS\f[R] This is another way to give command\-line arguments to dc(1). They should be in the same format as all other command\-line arguments. These are always processed first, so any files given in \f[B]DC_ENV_ARGS\f[R] will be processed before arguments and files given on the command\-line. This gives the user the ability to set up \[lq]standard\[rq] options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the \f[B]\-e\f[R] option to set \f[B]scale\f[R] to a value other than \f[B]0\f[R]. .RS .PP The code that parses \f[B]DC_ENV_ARGS\f[R] will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string \f[B]\[lq]/home/gavin/some dc file.dc\[rq]\f[R] will be correctly parsed, but the string \f[B]\[lq]/home/gavin/some \[dq]dc\[dq] file.dc\[rq]\f[R] will include the backslashes. .PP The quote parsing will handle either kind of quotes, \f[B]\[cq]\f[R] or \f[B]\[lq]\f[R]. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in \f[B]\[lq]some `dc' file.dc\[rq]\f[R], and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in \f[B]DC_ENV_ARGS\f[R] is not supported due to the complexity of the parsing, though such files are still supported on the command\-line where the parsing is done by the shell. .RE .TP \f[B]DC_LINE_LENGTH\f[R] If this environment variable exists and contains an integer that is greater than \f[B]1\f[R] and is less than \f[B]UINT16_MAX\f[R] (\f[B]2\[ha]16\-1\f[R]), dc(1) will output lines to that length, including the backslash newline combo. The default line length is \f[B]70\f[R]. .RS .PP The special value of \f[B]0\f[R] will disable line length checking and print numbers without regard to line length and without backslashes and newlines. .RE .TP \f[B]DC_SIGINT_RESET\f[R] If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), then this environment variable has no effect because dc(1) exits on \f[B]SIGINT\f[R] when not in interactive mode. .RS .PP However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) reset on \f[B]SIGINT\f[R], rather than exit, and zero makes dc(1) exit. If this environment variable exists and is \f[I]not\f[R] an integer, then dc(1) will exit on \f[B]SIGINT\f[R]. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_TTY_MODE\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non\-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_PROMPT\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) use a prompt, and zero or a non\-integer makes dc(1) not use a prompt. If this environment variable does not exist and \f[B]DC_TTY_MODE\f[R] does, then the value of the \f[B]DC_TTY_MODE\f[R] environment variable is used. .PP This environment variable and the \f[B]DC_TTY_MODE\f[R] environment variable override the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_EXPR_EXIT\f[R] If any expressions or expression files are given on the command\-line with \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R], then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. .RS .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_DIGIT_CLAMP\f[R] When parsing numbers and if this environment variable exists and contains an integer, a non\-zero value makes dc(1) clamp digits that are greater than or equal to the current \f[B]ibase\f[R] so that all such digits are considered equal to the \f[B]ibase\f[R] minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the \f[B]ibase\f[R]. .RS .PP This never applies to single\-digit numbers, as per the bc(1) standard (see the \f[B]STANDARDS\f[R] section). .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .SH EXIT STATUS dc(1) returns the following exit statuses: .TP \f[B]0\f[R] No error. .TP \f[B]1\f[R] A math error occurred. This follows standard practice of using \f[B]1\f[R] for expected errors, since math errors will happen in the process of normal execution. .RS .PP Math errors include divide by \f[B]0\f[R], taking the square root of a negative number, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non\-integer where an integer is required. .PP Converting to a hardware integer happens for the second operand of the power (\f[B]\[ha]\f[R]) operator. .RE .TP \f[B]2\f[R] A parse error occurred. .RS .PP Parse errors include unexpected \f[B]EOF\f[R], using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. .RE .TP \f[B]3\f[R] A runtime error occurred. .RS .PP Runtime errors include assigning an invalid number to any global (\f[B]ibase\f[R], \f[B]obase\f[R], or \f[B]scale\f[R]), giving a bad expression to a \f[B]read()\f[R] call, calling \f[B]read()\f[R] inside of a \f[B]read()\f[R] call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. .RE .TP \f[B]4\f[R] A fatal error occurred. .RS .PP Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command\-line options. .RE .PP The exit status \f[B]4\f[R] is special; when a fatal error occurs, dc(1) always exits and returns \f[B]4\f[R], no matter what mode dc(1) is in. .PP The other statuses will only be returned when dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), since dc(1) resets its state (see the \f[B]RESET\f[R] section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .PP These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .SH INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non\-interactive mode. Interactive mode is turned on automatically when both \f[B]stdin\f[R] and \f[B]stdout\f[R] are hooked to a terminal, but the \f[B]\-i\f[R] flag and \f[B]\-\-interactive\f[R] option can turn it on in other situations. .PP In interactive mode, dc(1) attempts to recover from errors (see the \f[B]RESET\f[R] section), and in normal execution, flushes \f[B]stdout\f[R] as soon as execution is done for the current input. dc(1) may also reset on \f[B]SIGINT\f[R] instead of exit, depending on the contents of, or default for, the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .SH TTY MODE If \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY, then \[lq]TTY mode\[rq] is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. .PP If there is the environment variable \f[B]DC_TTY_MODE\f[R] in the environment (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then if that environment variable contains a non\-zero integer, dc(1) will turn on TTY mode when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY. If the \f[B]DC_TTY_MODE\f[R] environment variable exists but is \f[I]not\f[R] a non\-zero integer, then dc(1) will not turn TTY mode on. .PP If the environment variable \f[B]DC_TTY_MODE\f[R] does \f[I]not\f[R] exist, the default setting is used. The default setting can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the \f[B]STANDARDS\f[R] section), and interactive mode requires only \f[B]stdin\f[R] and \f[B]stdout\f[R] to be connected to a terminal. .SS Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: \f[B]DC_PROMPT\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP If the environment variable \f[B]DC_PROMPT\f[R] exists and is a non\-zero integer, then the prompt is turned on when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are connected to a TTY and the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options were not used. The read prompt will be turned on under the same conditions, except that the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options must also not be used. .PP However, if \f[B]DC_PROMPT\f[R] does not exist, the prompt can be enabled or disabled with the \f[B]DC_TTY_MODE\f[R] environment variable, the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options, and the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options. See the \f[B]ENVIRONMENT VARIABLES\f[R] and \f[B]OPTIONS\f[R] sections for more details. .SH SIGNAL HANDLING Sending a \f[B]SIGINT\f[R] will cause dc(1) to do one of two things. .PP If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), or the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or its default, is either not an integer or it is zero, dc(1) will exit. .PP However, if dc(1) is in interactive mode, and the \f[B]DC_SIGINT_RESET\f[R] or its default is an integer and non\-zero, then dc(1) will stop executing the current input and reset (see the \f[B]RESET\f[R] section) upon receiving a \f[B]SIGINT\f[R]. .PP Note that \[lq]current input\[rq] can mean one of two things. If dc(1) is processing input from \f[B]stdin\f[R] in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from \f[B]stdin\f[R] if no other file exists. .PP This means that if a \f[B]SIGINT\f[R] is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. .PP \f[B]SIGTERM\f[R] and \f[B]SIGQUIT\f[R] cause dc(1) to clean up and exit, and it uses the default handler for all other signals. .SH LOCALES This dc(1) ships with support for adding error messages for different locales and thus, supports \f[B]LC_MESSAGES\f[R]. .SH SEE ALSO bc(1) .SH STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1\-2017 (\[lq]POSIX.1\-2017\[rq]) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . .SH BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . .SH AUTHOR Gavin D. Howard \c .MT gavin@gavinhoward.com .ME \c \ and contributors. diff --git a/manuals/dc/EH.1.md b/manuals/dc/EH.1.md index 761b9a89947b..6398477a84dd 100644 --- a/manuals/dc/EH.1.md +++ b/manuals/dc/EH.1.md @@ -1,1327 +1,1330 @@ # Name dc - arbitrary-precision decimal reverse-Polish notation calculator # SYNOPSIS **dc** [**-cChiPRvVx**] [**-\-version**] [**-\-help**] [**-\-digit-clamp**] [**-\-no-digit-clamp**] [**-\-interactive**] [**-\-no-prompt**] [**-\-no-read-prompt**] [**-\-extended-register**] [**-e** *expr*] [**-\-expression**=*expr*...] [**-f** *file*...] [**-\-file**=*file*...] [*file*...] # DESCRIPTION dc(1) is an arbitrary-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. If no files are given on the command-line, then dc(1) reads from **stdin** (see the **STDIN** section). Otherwise, those files are processed, and dc(1) will then exit. If a user wants to set up a standard environment, they can use **DC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). For example, if a user wants the **scale** always set to **10**, they can set **DC_ENV_ARGS** to **-e 10k**, and this dc(1) will always start with a **scale** of **10**. # OPTIONS The following are the options that dc(1) accepts. **-C**, **-\-no-digit-clamp** : Disables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that the value added to a number from a digit is always that digit's value multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-c** or **-\-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-c**, **-\-digit-clamp** : Enables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-C** or **-\-no-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-e** *expr*, **-\-expression**=*expr* : Evaluates *expr*. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**, whether on the command-line or in **DC_ENV_ARGS**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-f** *file*, **-\-file**=*file* : Reads in *file* and evaluates it, line by line, as though it were read through **stdin**. If expressions are also given (see above), the expressions are evaluated in the order given. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-h**, **-\-help** : Prints a usage message and exits. **-I** *ibase*, **-\-ibase**=*ibase* : Sets the builtin variable **ibase** to the value *ibase* assuming that *ibase* is in base 10. It is a fatal error if *ibase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-i**, **-\-interactive** : Forces interactive mode. (See the **INTERACTIVE MODE** section.) This is a **non-portable extension**. **-L**, **-\-no-line-length** : Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets **BC_LINE_LENGTH** to **0** (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-O** *obase*, **-\-obase**=*obase* : Sets the builtin variable **obase** to the value *obase* assuming that *obase* is in base 10. It is a fatal error if *obase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-P**, **-\-no-prompt** : Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in **DC_ENV_ARGS**. These options override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-R**, **-\-no-read-prompt** : Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in **BC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. This option does not disable the regular prompt because the read prompt is only used when the **?** command is used. These options *do* override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section), but only for the read prompt. This is a **non-portable extension**. **-S** *scale*, **-\-scale**=*scale* : Sets the builtin variable **scale** to the value *scale* assuming that *scale* is in base 10. It is a fatal error if *scale* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-v**, **-V**, **-\-version** : Print the version information (copyright header) and exits. **-x** **-\-extended-register** : Enables extended register mode. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. This is a **non-portable extension**. **-z**, **-\-leading-zeroes** : Makes dc(1) print all numbers greater than **-1** and less than **1**, and not equal to **0**, with a leading zero. This is a **non-portable extension**. All long options are **non-portable extensions**. # STDIN If no files are given on the command-line and no files or expressions are given by the **-f**, **-\-file**, **-e**, or **-\-expression** options, then dc(1) reads from **stdin**. However, there is a caveat to this. First, **stdin** is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. # STDOUT Any non-error output is written to **stdout**. In addition, if history (see the **HISTORY** section) and the prompt (see the **TTY MODE** section) are enabled, both are output to **stdout**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stdout**, so if **stdout** is closed, as in **dc >&-**, it will quit with an error. This is done so that dc(1) can report problems when **stdout** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stdout** to **/dev/null**. # STDERR Any error output is written to **stderr**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stderr**, so if **stderr** is closed, as in **dc 2>&-**, it will quit with an error. This is done so that dc(1) can exit with an error code when **stderr** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stderr** to **/dev/null**. # SYNTAX Each item in the input source code, either a number (see the **NUMBERS** section) or a command (see the **COMMANDS** section), is processed and executed, in order. Input is processed immediately when entered. **ibase** is a register (see the **REGISTERS** section) that determines how to interpret constant numbers. It is the "input" base, or the number base used for interpreting input numbers. **ibase** is initially **10**. The max allowable value for **ibase** is **16**. The min allowable value for **ibase** is **2**. The max allowable value for **ibase** can be queried in dc(1) programs with the **T** command. **obase** is a register (see the **REGISTERS** section) that determines how to output results. It is the "output" base, or the number base used for outputting numbers. **obase** is initially **10**. The max allowable value for **obase** is **DC_BASE_MAX** and can be queried with the **U** command. The min allowable value for **obase** is **2**. Values are output in the specified base. The *scale* of an expression is the number of digits in the result of the expression right of the decimal point, and **scale** is a register (see the **REGISTERS** section) that sets the precision of any operations (with exceptions). **scale** is initially **0**. **scale** cannot be negative. The max allowable value for **scale** can be queried in dc(1) programs with the **V** command. ## Comments Comments go from **#** until, and not including, the next newline. This is a **non-portable extension**. # NUMBERS Numbers are strings made up of digits, uppercase letters up to **F**, and at most **1** period for a radix. Numbers can have up to **DC_NUM_MAX** digits. Uppercase letters are equal to **9** plus their position in the alphabet (i.e., **A** equals **10**, or **9+1**). If a digit or letter makes no sense with the current value of **ibase** (i.e., they are greater than or equal to the current value of **ibase**), then the behavior depends on the existence of the **-c**/**-\-digit-clamp** or **-C**/**-\-no-digit-clamp** options (see the **OPTIONS** section), the existence and setting of the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section), or the default, which can be queried with the **-h**/**-\-help** option. If clamping is off, then digits or letters that are greater than or equal to the current value of **ibase** are not changed. Instead, their given value is multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*A+3\^0\*B**, which is **3** times **10** plus **11**, or **41**. If clamping is on, then digits or letters that are greater than or equal to the current value of **ibase** are set to the value of the highest valid digit in **ibase** before being multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*2+3\^0\*2**, which is **3** times **2** plus **2**, or **8**. There is one exception to clamping: single-character numbers (i.e., **A** alone). Such numbers are never clamped and always take the value they would have in the highest possible **ibase**. This means that **A** alone always equals decimal **10** and **Z** alone always equals decimal **35**. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current **ibase** (with the **i** command) regardless of the current value of **ibase**. If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for **A**, use **0A**. # COMMANDS The valid commands are listed below. ## Printing These commands are used for printing. **p** : Prints the value on top of the stack, whether number or string, and prints a newline after. This does not alter the stack. **n** : Prints the value on top of the stack, whether number or string, and pops it off of the stack. **P** : Pops a value off the stack. If the value is a number, it is truncated and the absolute value of the result is printed as though **obase** is **256** and each digit is interpreted as an 8-bit ASCII character, making it a byte stream. If the value is a string, it is printed without a trailing newline. This is a **non-portable extension**. **f** : Prints the entire contents of the stack, in order from newest to oldest, without altering anything. Users should use this command when they get lost. ## Arithmetic These are the commands used for arithmetic. **+** : The top two values are popped off the stack, added, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **-** : The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **\*** : The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If **a** is the *scale* of the first expression and **b** is the *scale* of the second expression, the *scale* of the result is equal to **min(a+b,max(scale,a,b))** where **min()** and **max()** return the obvious values. **/** : The top two values are popped off the stack, divided, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be non-zero. **%** : The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. Remaindering is equivalent to 1) Computing **a/b** to current **scale**, and 2) Using the result of step 1 to calculate **a-(a/b)\*b** to *scale* **max(scale+scale(b),scale(a))**. The first value popped off of the stack must be non-zero. **~** : The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to **x y / x y %** except that **x** and **y** are only evaluated once. The first value popped off of the stack must be non-zero. This is a **non-portable extension**. **\^** : The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non-zero. **v** : The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The value popped off of the stack must be non-negative. **\_** : If this command *immediately* precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a **non-portable extension**. **b** : The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. This is a **non-portable extension**. **|** : The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. The first value popped is used as the reduction modulus and must be an integer and non-zero. The second value popped is used as the exponent and must be an integer and non-negative. The third value popped is the base and must be an integer. This is a **non-portable extension**. **G** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if they are equal, or **0** otherwise. This is a **non-portable extension**. **N** : The top value is popped off of the stack, and if it a **0**, a **1** is pushed; otherwise, a **0** is pushed. This is a **non-portable extension**. **(** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than the second, or **0** otherwise. This is a **non-portable extension**. **{** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **)** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than the second, or **0** otherwise. This is a **non-portable extension**. **}** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **M** : The top two values are popped off of the stack. If they are both non-zero, a **1** is pushed onto the stack. If either of them is zero, or both of them are, then a **0** is pushed onto the stack. This is like the **&&** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. **m** : The top two values are popped off of the stack. If at least one of them is non-zero, a **1** is pushed onto the stack. If both of them are zero, then a **0** is pushed onto the stack. This is like the **||** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. ## Stack Control These commands control the stack. **c** : Removes all items from ("clears") the stack. **d** : Copies the item on top of the stack ("duplicates") and pushes the copy onto the stack. **r** : Swaps ("reverses") the two top items on the stack. **R** : Pops ("removes") the top value from the stack. ## Register Control These commands control registers (see the **REGISTERS** section). **s**_r_ : Pops the value off the top of the stack and stores it into register *r*. **l**_r_ : Copies the value in register *r* and pushes it onto the stack. This does not alter the contents of *r*. **S**_r_ : Pops the value off the top of the (main) stack and pushes it onto the stack of register *r*. The previous value of the register becomes inaccessible. **L**_r_ : Pops the value off the top of the stack for register *r* and push it onto the main stack. The previous value in the stack for register *r*, if any, is now accessible via the **l**_r_ command. ## Parameters These commands control the values of **ibase**, **obase**, and **scale**. Also see the **SYNTAX** section. **i** : Pops the value off of the top of the stack and uses it to set **ibase**, which must be between **2** and **16**, inclusive. If the value on top of the stack has any *scale*, the *scale* is ignored. **o** : Pops the value off of the top of the stack and uses it to set **obase**, which must be between **2** and **DC_BASE_MAX**, inclusive (see the **LIMITS** section). If the value on top of the stack has any *scale*, the *scale* is ignored. **k** : Pops the value off of the top of the stack and uses it to set **scale**, which must be non-negative. If the value on top of the stack has any *scale*, the *scale* is ignored. **I** : Pushes the current value of **ibase** onto the main stack. **O** : Pushes the current value of **obase** onto the main stack. **K** : Pushes the current value of **scale** onto the main stack. **T** : Pushes the maximum allowable value of **ibase** onto the main stack. This is a **non-portable extension**. **U** : Pushes the maximum allowable value of **obase** onto the main stack. This is a **non-portable extension**. **V** : Pushes the maximum allowable value of **scale** onto the main stack. This is a **non-portable extension**. ## Strings The following commands control strings. dc(1) can work with both numbers and strings, and registers (see the **REGISTERS** section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. Strings can also be executed as macros. For example, if the string **[1pR]** is executed as a macro, then the code **1pR** is executed, meaning that the **1** will be printed with a newline after and then popped from the stack. **\[**_characters_**\]** : Makes a string containing *characters* and pushes it onto the stack. If there are brackets (**\[** and **\]**) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (**\\**) character. If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. **a** : The value on top of the stack is popped. If it is a number, it is truncated and its absolute value is taken. The result mod **256** is calculated. If that result is **0**, push an empty string; otherwise, push a one-character string where the character is the result of the mod interpreted as an ASCII character. If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one-character string. The new string is then pushed onto the stack. This is a **non-portable extension**. **x** : Pops a value off of the top of the stack. If it is a number, it is pushed back onto the stack. If it is a string, it is executed as a macro. This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. **\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register *r* are executed. For example, **0 1>a** will execute the contents of register **a**, and **1 0>a** will not. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **?** : Reads a line from the **stdin** and executes it. This is to allow macros to request input from users. **q** : During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. **Q** : Pops a value from the stack which must be non-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. **,** : Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the **Q** command, so the sequence **,Q** will make dc(1) exit. This is a **non-portable extension**. ## Status These commands query status of the stack or its top value. **Z** : Pops a value off of the stack. If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push **1** if the argument is **0** with no decimal places. If it is a string, pushes the number of characters the string has. **X** : Pops a value off of the stack. If it is a number, pushes the *scale* of the value onto the stack. If it is a string, pushes **0**. **u** : Pops one value off of the stack. If the value is a number, this pushes **1** onto the stack. Otherwise (if it is a string), it pushes **0**. This is a **non-portable extension**. **t** : Pops one value off of the stack. If the value is a string, this pushes **1** onto the stack. Otherwise (if it is a number), it pushes **0**. This is a **non-portable extension**. **z** : Pushes the current depth of the stack (before execution of this command) onto the stack. **y**_r_ : Pushes the current stack depth of the register *r* onto the main stack. Because each register has a depth of **1** (with the value **0** in the top item) when dc(1) starts, dc(1) requires that each register's stack must always have at least one item; dc(1) will give an error and reset otherwise (see the **RESET** section). This means that this command will never push **0**. This is a **non-portable extension**. ## Arrays These commands manipulate arrays. **:**_r_ : Pops the top two values off of the stack. The second value will be stored in the array *r* (see the **REGISTERS** section), indexed by the first value. **;**_r_ : Pops the value on top of the stack and uses it as an index into the array *r*. The selected value is then pushed onto the stack. **Y**_r_ : Pushes the length of the array *r* onto the stack. This is a **non-portable extension**. ## Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter **g**. Only the characters below are allowed after the character **g**; any other character produces a parse error (see the **ERRORS** section). **gl** : Pushes the line length set by **DC_LINE_LENGTH** (see the **ENVIRONMENT VARIABLES** section) onto the stack. **gx** : Pushes **1** onto the stack if extended register mode is on, **0** otherwise. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. **gz** : Pushes **0** onto the stack if the leading zero setting has not been enabled with the **-z** or **-\-leading-zeroes** options (see the **OPTIONS** section), non-zero otherwise. # REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (**0**) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. In non-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (**'\\n'**) and a left bracket (**'['**); it is a parse error for a newline or a left bracket to be used as a register name. ## Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. If extended register mode is enabled (**-x** or **-\-extended-register** command-line arguments are given), then normal single character registers are used *unless* the character immediately following a command that needs a register name is a space (according to **isspace()**) and not a newline (**'\\n'**). In that case, the register name is found according to the regex **\[a-z\]\[a-z0-9\_\]\*** (like bc(1) identifiers), and it is a parse error if the next non-space characters do not match that regex. # RESET When dc(1) encounters an error or a signal that it has a non-default handler for, it resets. This means that several things happen. -First, any macros that are executing are stopped and popped off the stack. -The behavior is not unlike that of exceptions in programming languages. Then -the execution point is set so that any code waiting to execute (after all +First, any macros that are executing are stopped and popped off the execution +stack. The behavior is not unlike that of exceptions in programming languages. +Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. +However, the stack of values is *not* cleared; in interactive mode, users can +inspect the stack and manipulate it. + Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the **EXIT STATUS** section), it asks for more input; otherwise, it exits with the appropriate return code. # PERFORMANCE Most dc(1) implementations use **char** types to calculate the value of **1** decimal digit at a time, but that can be slow. This dc(1) does something different. It uses large integers to calculate more than **1** decimal digit at a time. If built in a environment where **DC_LONG_BIT** (see the **LIMITS** section) is **64**, then each integer has **9** decimal digits. If built in an environment where **DC_LONG_BIT** is **32** then each integer has **4** decimal digits. This value (the number of decimal digits per large integer) is called **DC_BASE_DIGS**. In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of **DC_LONG_BIT**, but is always at least twice as large as the integer type used to store digits. # LIMITS The following are the limits on dc(1): **DC_LONG_BIT** : The number of bits in the **long** type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the **PERFORMANCE** section). **DC_BASE_DIGS** : The number of decimal digits per large integer (see the **PERFORMANCE** section). Depends on **DC_LONG_BIT**. **DC_BASE_POW** : The max decimal number that each large integer can store (see **DC_BASE_DIGS**) plus **1**. Depends on **DC_BASE_DIGS**. **DC_OVERFLOW_MAX** : The max number that the overflow type (see the **PERFORMANCE** section) can hold. Depends on **DC_LONG_BIT**. **DC_BASE_MAX** : The maximum output base. Set at **DC_BASE_POW**. **DC_DIM_MAX** : The maximum size of arrays. Set at **SIZE_MAX-1**. **DC_SCALE_MAX** : The maximum **scale**. Set at **DC_OVERFLOW_MAX-1**. **DC_STRING_MAX** : The maximum length of strings. Set at **DC_OVERFLOW_MAX-1**. **DC_NAME_MAX** : The maximum length of identifiers. Set at **DC_OVERFLOW_MAX-1**. **DC_NUM_MAX** : The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at **DC_OVERFLOW_MAX-1**. Exponent : The maximum allowable exponent (positive or negative). Set at **DC_OVERFLOW_MAX**. Number of vars : The maximum number of vars/arrays. Set at **SIZE_MAX-1**. These limits are meant to be effectively non-existent; the limits are so large (at least on 64-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. # ENVIRONMENT VARIABLES As **non-portable extensions**, dc(1) recognizes the following environment variables: **DC_ENV_ARGS** : This is another way to give command-line arguments to dc(1). They should be in the same format as all other command-line arguments. These are always processed first, so any files given in **DC_ENV_ARGS** will be processed before arguments and files given on the command-line. This gives the user the ability to set up "standard" options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the **-e** option to set **scale** to a value other than **0**. The code that parses **DC_ENV_ARGS** will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string **"/home/gavin/some dc file.dc"** will be correctly parsed, but the string **"/home/gavin/some \"dc\" file.dc"** will include the backslashes. The quote parsing will handle either kind of quotes, **'** or **"**. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in **"some 'dc' file.dc"**, and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in **DC_ENV_ARGS** is not supported due to the complexity of the parsing, though such files are still supported on the command-line where the parsing is done by the shell. **DC_LINE_LENGTH** : If this environment variable exists and contains an integer that is greater than **1** and is less than **UINT16_MAX** (**2\^16-1**), dc(1) will output lines to that length, including the backslash newline combo. The default line length is **70**. The special value of **0** will disable line length checking and print numbers without regard to line length and without backslashes and newlines. **DC_SIGINT_RESET** : If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), then this environment variable has no effect because dc(1) exits on **SIGINT** when not in interactive mode. However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) reset on **SIGINT**, rather than exit, and zero makes dc(1) exit. If this environment variable exists and is *not* an integer, then dc(1) will exit on **SIGINT**. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_TTY_MODE** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_PROMPT** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) use a prompt, and zero or a non-integer makes dc(1) not use a prompt. If this environment variable does not exist and **DC_TTY_MODE** does, then the value of the **DC_TTY_MODE** environment variable is used. This environment variable and the **DC_TTY_MODE** environment variable override the default, which can be queried with the **-h** or **-\-help** options. **DC_EXPR_EXIT** : If any expressions or expression files are given on the command-line with **-e**, **-\-expression**, **-f**, or **-\-file**, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_DIGIT_CLAMP** : When parsing numbers and if this environment variable exists and contains an integer, a non-zero value makes dc(1) clamp digits that are greater than or equal to the current **ibase** so that all such digits are considered equal to the **ibase** minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the **ibase**. This never applies to single-digit numbers, as per the bc(1) standard (see the **STANDARDS** section). This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. # EXIT STATUS dc(1) returns the following exit statuses: **0** : No error. **1** : A math error occurred. This follows standard practice of using **1** for expected errors, since math errors will happen in the process of normal execution. Math errors include divide by **0**, taking the square root of a negative number, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non-integer where an integer is required. Converting to a hardware integer happens for the second operand of the power (**\^**) operator. **2** : A parse error occurred. Parse errors include unexpected **EOF**, using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. **3** : A runtime error occurred. Runtime errors include assigning an invalid number to any global (**ibase**, **obase**, or **scale**), giving a bad expression to a **read()** call, calling **read()** inside of a **read()** call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. **4** : A fatal error occurred. Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command-line options. The exit status **4** is special; when a fatal error occurs, dc(1) always exits and returns **4**, no matter what mode dc(1) is in. The other statuses will only be returned when dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), since dc(1) resets its state (see the **RESET** section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the **-i** flag or **-\-interactive** option. These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the **-i** flag or **-\-interactive** option. # INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non-interactive mode. Interactive mode is turned on automatically when both **stdin** and **stdout** are hooked to a terminal, but the **-i** flag and **-\-interactive** option can turn it on in other situations. In interactive mode, dc(1) attempts to recover from errors (see the **RESET** section), and in normal execution, flushes **stdout** as soon as execution is done for the current input. dc(1) may also reset on **SIGINT** instead of exit, depending on the contents of, or default for, the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section). # TTY MODE If **stdin**, **stdout**, and **stderr** are all connected to a TTY, then "TTY mode" is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. If there is the environment variable **DC_TTY_MODE** in the environment (see the **ENVIRONMENT VARIABLES** section), then if that environment variable contains a non-zero integer, dc(1) will turn on TTY mode when **stdin**, **stdout**, and **stderr** are all connected to a TTY. If the **DC_TTY_MODE** environment variable exists but is *not* a non-zero integer, then dc(1) will not turn TTY mode on. If the environment variable **DC_TTY_MODE** does *not* exist, the default setting is used. The default setting can be queried with the **-h** or **-\-help** options. TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the **STANDARDS** section), and interactive mode requires only **stdin** and **stdout** to be connected to a terminal. ## Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: **DC_PROMPT** (see the **ENVIRONMENT VARIABLES** section). If the environment variable **DC_PROMPT** exists and is a non-zero integer, then the prompt is turned on when **stdin**, **stdout**, and **stderr** are connected to a TTY and the **-P** and **-\-no-prompt** options were not used. The read prompt will be turned on under the same conditions, except that the **-R** and **-\-no-read-prompt** options must also not be used. However, if **DC_PROMPT** does not exist, the prompt can be enabled or disabled with the **DC_TTY_MODE** environment variable, the **-P** and **-\-no-prompt** options, and the **-R** and **-\-no-read-prompt** options. See the **ENVIRONMENT VARIABLES** and **OPTIONS** sections for more details. # SIGNAL HANDLING Sending a **SIGINT** will cause dc(1) to do one of two things. If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), or the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section), or its default, is either not an integer or it is zero, dc(1) will exit. However, if dc(1) is in interactive mode, and the **DC_SIGINT_RESET** or its default is an integer and non-zero, then dc(1) will stop executing the current input and reset (see the **RESET** section) upon receiving a **SIGINT**. Note that "current input" can mean one of two things. If dc(1) is processing input from **stdin** in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from **stdin** if no other file exists. This means that if a **SIGINT** is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. **SIGTERM** and **SIGQUIT** cause dc(1) to clean up and exit, and it uses the default handler for all other signals. # LOCALES This dc(1) ships with support for adding error messages for different locales and thus, supports **LC_MESSAGES**. # SEE ALSO bc(1) # STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1-2017 (“POSIX.1-2017”) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . # BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . # AUTHOR Gavin D. Howard and contributors. diff --git a/manuals/dc/EHN.1 b/manuals/dc/EHN.1 index d26d49c5ce3d..974cb3c86791 100644 --- a/manuals/dc/EHN.1 +++ b/manuals/dc/EHN.1 @@ -1,1444 +1,1447 @@ .\" .\" SPDX-License-Identifier: BSD-2-Clause .\" .\" Copyright (c) 2018-2024 Gavin D. Howard and contributors. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions are met: .\" .\" * Redistributions of source code must retain the above copyright notice, .\" this list of conditions and the following disclaimer. .\" .\" * Redistributions in binary form must reproduce the above copyright notice, .\" this list of conditions and the following disclaimer in the documentation .\" and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" .\" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE .\" LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF .\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS .\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN .\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) .\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE .\" POSSIBILITY OF SUCH DAMAGE. .\" -.TH "DC" "1" "January 2024" "Gavin D. Howard" "General Commands Manual" +.TH "DC" "1" "August 2024" "Gavin D. Howard" "General Commands Manual" .nh .ad l .SH Name dc \- arbitrary\-precision decimal reverse\-Polish notation calculator .SH SYNOPSIS \f[B]dc\f[R] [\f[B]\-cChiPRvVx\f[R]] [\f[B]\-\-version\f[R]] [\f[B]\-\-help\f[R]] [\f[B]\-\-digit\-clamp\f[R]] [\f[B]\-\-no\-digit\-clamp\f[R]] [\f[B]\-\-interactive\f[R]] [\f[B]\-\-no\-prompt\f[R]] [\f[B]\-\-no\-read\-prompt\f[R]] [\f[B]\-\-extended\-register\f[R]] [\f[B]\-e\f[R] \f[I]expr\f[R]] [\f[B]\-\-expression\f[R]=\f[I]expr\f[R]\&...] [\f[B]\-f\f[R] \f[I]file\f[R]\&...] [\f[B]\-\-file\f[R]=\f[I]file\f[R]\&...] [\f[I]file\f[R]\&...] .SH DESCRIPTION dc(1) is an arbitrary\-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. .PP If no files are given on the command\-line, then dc(1) reads from \f[B]stdin\f[R] (see the \f[B]STDIN\f[R] section). Otherwise, those files are processed, and dc(1) will then exit. .PP If a user wants to set up a standard environment, they can use \f[B]DC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). For example, if a user wants the \f[B]scale\f[R] always set to \f[B]10\f[R], they can set \f[B]DC_ENV_ARGS\f[R] to \f[B]\-e 10k\f[R], and this dc(1) will always start with a \f[B]scale\f[R] of \f[B]10\f[R]. .SH OPTIONS The following are the options that dc(1) accepts. .TP \f[B]\-C\f[R], \f[B]\-\-no\-digit\-clamp\f[R] Disables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that the value added to a number from a digit is always that digit\[cq]s value multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-c\f[R] or \f[B]\-\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-c\f[R], \f[B]\-\-digit\-clamp\f[R] Enables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-C\f[R] or \f[B]\-\-no\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-e\f[R] \f[I]expr\f[R], \f[B]\-\-expression\f[R]=\f[I]expr\f[R] Evaluates \f[I]expr\f[R]. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R], whether on the command\-line or in \f[B]DC_ENV_ARGS\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-f\f[R] \f[I]file\f[R], \f[B]\-\-file\f[R]=\f[I]file\f[R] Reads in \f[I]file\f[R] and evaluates it, line by line, as though it were read through \f[B]stdin\f[R]. If expressions are also given (see above), the expressions are evaluated in the order given. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-h\f[R], \f[B]\-\-help\f[R] Prints a usage message and exits. .TP \f[B]\-I\f[R] \f[I]ibase\f[R], \f[B]\-\-ibase\f[R]=\f[I]ibase\f[R] Sets the builtin variable \f[B]ibase\f[R] to the value \f[I]ibase\f[R] assuming that \f[I]ibase\f[R] is in base 10. It is a fatal error if \f[I]ibase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-i\f[R], \f[B]\-\-interactive\f[R] Forces interactive mode. (See the \f[B]INTERACTIVE MODE\f[R] section.) .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-L\f[R], \f[B]\-\-no\-line\-length\f[R] Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets \f[B]BC_LINE_LENGTH\f[R] to \f[B]0\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-O\f[R] \f[I]obase\f[R], \f[B]\-\-obase\f[R]=\f[I]obase\f[R] Sets the builtin variable \f[B]obase\f[R] to the value \f[I]obase\f[R] assuming that \f[I]obase\f[R] is in base 10. It is a fatal error if \f[I]obase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-P\f[R], \f[B]\-\-no\-prompt\f[R] Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]DC_ENV_ARGS\f[R]. .RS .PP These options override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-R\f[R], \f[B]\-\-no\-read\-prompt\f[R] Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]BC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. .RS .PP This option does not disable the regular prompt because the read prompt is only used when the \f[B]?\f[R] command is used. .PP These options \f[I]do\f[R] override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), but only for the read prompt. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-S\f[R] \f[I]scale\f[R], \f[B]\-\-scale\f[R]=\f[I]scale\f[R] Sets the builtin variable \f[B]scale\f[R] to the value \f[I]scale\f[R] assuming that \f[I]scale\f[R] is in base 10. It is a fatal error if \f[I]scale\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-v\f[R], \f[B]\-V\f[R], \f[B]\-\-version\f[R] Print the version information (copyright header) and exits. .TP \f[B]\-x\f[R] \f[B]\-\-extended\-register\f[R] Enables extended register mode. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-z\f[R], \f[B]\-\-leading\-zeroes\f[R] Makes dc(1) print all numbers greater than \f[B]\-1\f[R] and less than \f[B]1\f[R], and not equal to \f[B]0\f[R], with a leading zero. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .PP All long options are \f[B]non\-portable extensions\f[R]. .SH STDIN If no files are given on the command\-line and no files or expressions are given by the \f[B]\-f\f[R], \f[B]\-\-file\f[R], \f[B]\-e\f[R], or \f[B]\-\-expression\f[R] options, then dc(1) reads from \f[B]stdin\f[R]. .PP However, there is a caveat to this. .PP First, \f[B]stdin\f[R] is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. .SH STDOUT Any non\-error output is written to \f[B]stdout\f[R]. In addition, if history (see the \f[B]HISTORY\f[R] section) and the prompt (see the \f[B]TTY MODE\f[R] section) are enabled, both are output to \f[B]stdout\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stdout\f[R], so if \f[B]stdout\f[R] is closed, as in \f[B]dc >&\-\f[R], it will quit with an error. This is done so that dc(1) can report problems when \f[B]stdout\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stdout\f[R] to \f[B]/dev/null\f[R]. .SH STDERR Any error output is written to \f[B]stderr\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stderr\f[R], so if \f[B]stderr\f[R] is closed, as in \f[B]dc 2>&\-\f[R], it will quit with an error. This is done so that dc(1) can exit with an error code when \f[B]stderr\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stderr\f[R] to \f[B]/dev/null\f[R]. .SH SYNTAX Each item in the input source code, either a number (see the \f[B]NUMBERS\f[R] section) or a command (see the \f[B]COMMANDS\f[R] section), is processed and executed, in order. Input is processed immediately when entered. .PP \f[B]ibase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to interpret constant numbers. It is the \[lq]input\[rq] base, or the number base used for interpreting input numbers. \f[B]ibase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]ibase\f[R] is \f[B]16\f[R]. The min allowable value for \f[B]ibase\f[R] is \f[B]2\f[R]. The max allowable value for \f[B]ibase\f[R] can be queried in dc(1) programs with the \f[B]T\f[R] command. .PP \f[B]obase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to output results. It is the \[lq]output\[rq] base, or the number base used for outputting numbers. \f[B]obase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]obase\f[R] is \f[B]DC_BASE_MAX\f[R] and can be queried with the \f[B]U\f[R] command. The min allowable value for \f[B]obase\f[R] is \f[B]2\f[R]. Values are output in the specified base. .PP The \f[I]scale\f[R] of an expression is the number of digits in the result of the expression right of the decimal point, and \f[B]scale\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that sets the precision of any operations (with exceptions). \f[B]scale\f[R] is initially \f[B]0\f[R]. \f[B]scale\f[R] cannot be negative. The max allowable value for \f[B]scale\f[R] can be queried in dc(1) programs with the \f[B]V\f[R] command. .SS Comments Comments go from \f[B]#\f[R] until, and not including, the next newline. This is a \f[B]non\-portable extension\f[R]. .SH NUMBERS Numbers are strings made up of digits, uppercase letters up to \f[B]F\f[R], and at most \f[B]1\f[R] period for a radix. Numbers can have up to \f[B]DC_NUM_MAX\f[R] digits. Uppercase letters are equal to \f[B]9\f[R] plus their position in the alphabet (i.e., \f[B]A\f[R] equals \f[B]10\f[R], or \f[B]9+1\f[R]). .PP If a digit or letter makes no sense with the current value of \f[B]ibase\f[R] (i.e., they are greater than or equal to the current value of \f[B]ibase\f[R]), then the behavior depends on the existence of the \f[B]\-c\f[R]/\f[B]\-\-digit\-clamp\f[R] or \f[B]\-C\f[R]/\f[B]\-\-no\-digit\-clamp\f[R] options (see the \f[B]OPTIONS\f[R] section), the existence and setting of the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or the default, which can be queried with the \f[B]\-h\f[R]/\f[B]\-\-help\f[R] option. .PP If clamping is off, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are not changed. Instead, their given value is multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*A+3\[ha]0*B\f[R], which is \f[B]3\f[R] times \f[B]10\f[R] plus \f[B]11\f[R], or \f[B]41\f[R]. .PP If clamping is on, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are set to the value of the highest valid digit in \f[B]ibase\f[R] before being multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*2+3\[ha]0*2\f[R], which is \f[B]3\f[R] times \f[B]2\f[R] plus \f[B]2\f[R], or \f[B]8\f[R]. .PP There is one exception to clamping: single\-character numbers (i.e., \f[B]A\f[R] alone). Such numbers are never clamped and always take the value they would have in the highest possible \f[B]ibase\f[R]. This means that \f[B]A\f[R] alone always equals decimal \f[B]10\f[R] and \f[B]Z\f[R] alone always equals decimal \f[B]35\f[R]. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current \f[B]ibase\f[R] (with the \f[B]i\f[R] command) regardless of the current value of \f[B]ibase\f[R]. .PP If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for \f[B]A\f[R], use \f[B]0A\f[R]. .SH COMMANDS The valid commands are listed below. .SS Printing These commands are used for printing. .TP \f[B]p\f[R] Prints the value on top of the stack, whether number or string, and prints a newline after. .RS .PP This does not alter the stack. .RE .TP \f[B]n\f[R] Prints the value on top of the stack, whether number or string, and pops it off of the stack. .TP \f[B]P\f[R] Pops a value off the stack. .RS .PP If the value is a number, it is truncated and the absolute value of the result is printed as though \f[B]obase\f[R] is \f[B]256\f[R] and each digit is interpreted as an 8\-bit ASCII character, making it a byte stream. .PP If the value is a string, it is printed without a trailing newline. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]f\f[R] Prints the entire contents of the stack, in order from newest to oldest, without altering anything. .RS .PP Users should use this command when they get lost. .RE .SS Arithmetic These are the commands used for arithmetic. .TP \f[B]+\f[R] The top two values are popped off the stack, added, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]\-\f[R] The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]*\f[R] The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If \f[B]a\f[R] is the \f[I]scale\f[R] of the first expression and \f[B]b\f[R] is the \f[I]scale\f[R] of the second expression, the \f[I]scale\f[R] of the result is equal to \f[B]min(a+b,max(scale,a,b))\f[R] where \f[B]min()\f[R] and \f[B]max()\f[R] return the obvious values. .TP \f[B]/\f[R] The top two values are popped off the stack, divided, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]%\f[R] The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. .RS .PP Remaindering is equivalent to 1) Computing \f[B]a/b\f[R] to current \f[B]scale\f[R], and 2) Using the result of step 1 to calculate \f[B]a\-(a/b)*b\f[R] to \f[I]scale\f[R] \f[B]max(scale+scale(b),scale(a))\f[R]. .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]\[ti]\f[R] The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to \f[B]x y / x y %\f[R] except that \f[B]x\f[R] and \f[B]y\f[R] are only evaluated once. .RS .PP The first value popped off of the stack must be non\-zero. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[ha]\f[R] The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non\-zero. .RE .TP \f[B]v\f[R] The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The value popped off of the stack must be non\-negative. .RE .TP \f[B]_\f[R] If this command \f[I]immediately\f[R] precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. .RS .PP Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]b\f[R] The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]|\f[R] The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. .RS .PP The first value popped is used as the reduction modulus and must be an integer and non\-zero. The second value popped is used as the exponent and must be an integer and non\-negative. The third value popped is the base and must be an integer. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]G\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if they are equal, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]N\f[R] The top value is popped off of the stack, and if it a \f[B]0\f[R], a \f[B]1\f[R] is pushed; otherwise, a \f[B]0\f[R] is pushed. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B](\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]{\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B])\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]}\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]M\f[R] The top two values are popped off of the stack. If they are both non\-zero, a \f[B]1\f[R] is pushed onto the stack. If either of them is zero, or both of them are, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]&&\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]m\f[R] The top two values are popped off of the stack. If at least one of them is non\-zero, a \f[B]1\f[R] is pushed onto the stack. If both of them are zero, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]||\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Stack Control These commands control the stack. .TP \f[B]c\f[R] Removes all items from (\[lq]clears\[rq]) the stack. .TP \f[B]d\f[R] Copies the item on top of the stack (\[lq]duplicates\[rq]) and pushes the copy onto the stack. .TP \f[B]r\f[R] Swaps (\[lq]reverses\[rq]) the two top items on the stack. .TP \f[B]R\f[R] Pops (\[lq]removes\[rq]) the top value from the stack. .SS Register Control These commands control registers (see the \f[B]REGISTERS\f[R] section). .TP \f[B]s\f[R]\f[I]r\f[R] Pops the value off the top of the stack and stores it into register \f[I]r\f[R]. .TP \f[B]l\f[R]\f[I]r\f[R] Copies the value in register \f[I]r\f[R] and pushes it onto the stack. This does not alter the contents of \f[I]r\f[R]. .TP \f[B]S\f[R]\f[I]r\f[R] Pops the value off the top of the (main) stack and pushes it onto the stack of register \f[I]r\f[R]. The previous value of the register becomes inaccessible. .TP \f[B]L\f[R]\f[I]r\f[R] Pops the value off the top of the stack for register \f[I]r\f[R] and push it onto the main stack. The previous value in the stack for register \f[I]r\f[R], if any, is now accessible via the \f[B]l\f[R]\f[I]r\f[R] command. .SS Parameters These commands control the values of \f[B]ibase\f[R], \f[B]obase\f[R], and \f[B]scale\f[R]. Also see the \f[B]SYNTAX\f[R] section. .TP \f[B]i\f[R] Pops the value off of the top of the stack and uses it to set \f[B]ibase\f[R], which must be between \f[B]2\f[R] and \f[B]16\f[R], inclusive. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]o\f[R] Pops the value off of the top of the stack and uses it to set \f[B]obase\f[R], which must be between \f[B]2\f[R] and \f[B]DC_BASE_MAX\f[R], inclusive (see the \f[B]LIMITS\f[R] section). .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]k\f[R] Pops the value off of the top of the stack and uses it to set \f[B]scale\f[R], which must be non\-negative. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]I\f[R] Pushes the current value of \f[B]ibase\f[R] onto the main stack. .TP \f[B]O\f[R] Pushes the current value of \f[B]obase\f[R] onto the main stack. .TP \f[B]K\f[R] Pushes the current value of \f[B]scale\f[R] onto the main stack. .TP \f[B]T\f[R] Pushes the maximum allowable value of \f[B]ibase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]U\f[R] Pushes the maximum allowable value of \f[B]obase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]V\f[R] Pushes the maximum allowable value of \f[B]scale\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Strings The following commands control strings. .PP dc(1) can work with both numbers and strings, and registers (see the \f[B]REGISTERS\f[R] section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. .PP While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. .PP Strings can also be executed as macros. For example, if the string \f[B][1pR]\f[R] is executed as a macro, then the code \f[B]1pR\f[R] is executed, meaning that the \f[B]1\f[R] will be printed with a newline after and then popped from the stack. .TP \f[B][\f[R]\f[I]characters\f[R]\f[B]]\f[R] Makes a string containing \f[I]characters\f[R] and pushes it onto the stack. .RS .PP If there are brackets (\f[B][\f[R] and \f[B]]\f[R]) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (\f[B]\[rs]\f[R]) character. .PP If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. .RE .TP \f[B]a\f[R] The value on top of the stack is popped. .RS .PP If it is a number, it is truncated and its absolute value is taken. The result mod \f[B]256\f[R] is calculated. If that result is \f[B]0\f[R], push an empty string; otherwise, push a one\-character string where the character is the result of the mod interpreted as an ASCII character. .PP If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one\-character string. The new string is then pushed onto the stack. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]x\f[R] Pops a value off of the top of the stack. .RS .PP If it is a number, it is pushed back onto the stack. .PP If it is a string, it is executed as a macro. .PP This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. .RE .TP \f[B]>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP For example, \f[B]0 1>a\f[R] will execute the contents of register \f[B]a\f[R], and \f[B]1 0>a\f[R] will not. .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]?\f[R] Reads a line from the \f[B]stdin\f[R] and executes it. This is to allow macros to request input from users. .TP \f[B]q\f[R] During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. .TP \f[B]Q\f[R] Pops a value from the stack which must be non\-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. .TP \f[B],\f[R] Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the \f[B]Q\f[R] command, so the sequence \f[B],Q\f[R] will make dc(1) exit. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Status These commands query status of the stack or its top value. .TP \f[B]Z\f[R] Pops a value off of the stack. .RS .PP If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push \f[B]1\f[R] if the argument is \f[B]0\f[R] with no decimal places. .PP If it is a string, pushes the number of characters the string has. .RE .TP \f[B]X\f[R] Pops a value off of the stack. .RS .PP If it is a number, pushes the \f[I]scale\f[R] of the value onto the stack. .PP If it is a string, pushes \f[B]0\f[R]. .RE .TP \f[B]u\f[R] Pops one value off of the stack. If the value is a number, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a string), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]t\f[R] Pops one value off of the stack. If the value is a string, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a number), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]z\f[R] Pushes the current depth of the stack (before execution of this command) onto the stack. .TP \f[B]y\f[R]\f[I]r\f[R] Pushes the current stack depth of the register \f[I]r\f[R] onto the main stack. .RS .PP Because each register has a depth of \f[B]1\f[R] (with the value \f[B]0\f[R] in the top item) when dc(1) starts, dc(1) requires that each register\[cq]s stack must always have at least one item; dc(1) will give an error and reset otherwise (see the \f[B]RESET\f[R] section). This means that this command will never push \f[B]0\f[R]. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Arrays These commands manipulate arrays. .TP \f[B]:\f[R]\f[I]r\f[R] Pops the top two values off of the stack. The second value will be stored in the array \f[I]r\f[R] (see the \f[B]REGISTERS\f[R] section), indexed by the first value. .TP \f[B];\f[R]\f[I]r\f[R] Pops the value on top of the stack and uses it as an index into the array \f[I]r\f[R]. The selected value is then pushed onto the stack. .TP \f[B]Y\f[R]\f[I]r\f[R] Pushes the length of the array \f[I]r\f[R] onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter \f[B]g\f[R]. Only the characters below are allowed after the character \f[B]g\f[R]; any other character produces a parse error (see the \f[B]ERRORS\f[R] section). .TP \f[B]gl\f[R] Pushes the line length set by \f[B]DC_LINE_LENGTH\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) onto the stack. .TP \f[B]gx\f[R] Pushes \f[B]1\f[R] onto the stack if extended register mode is on, \f[B]0\f[R] otherwise. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .TP \f[B]gz\f[R] Pushes \f[B]0\f[R] onto the stack if the leading zero setting has not been enabled with the \f[B]\-z\f[R] or \f[B]\-\-leading\-zeroes\f[R] options (see the \f[B]OPTIONS\f[R] section), non\-zero otherwise. .SH REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) .PP Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (\f[B]0\f[R]) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. .PP In non\-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (\f[B]`\[rs]n'\f[R]) and a left bracket (\f[B]`['\f[R]); it is a parse error for a newline or a left bracket to be used as a register name. .SS Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. .PP If extended register mode is enabled (\f[B]\-x\f[R] or \f[B]\-\-extended\-register\f[R] command\-line arguments are given), then normal single character registers are used \f[I]unless\f[R] the character immediately following a command that needs a register name is a space (according to \f[B]isspace()\f[R]) and not a newline (\f[B]`\[rs]n'\f[R]). .PP In that case, the register name is found according to the regex \f[B][a\-z][a\-z0\-9_]*\f[R] (like bc(1) identifiers), and it is a parse error if the next non\-space characters do not match that regex. .SH RESET When dc(1) encounters an error or a signal that it has a non\-default handler for, it resets. This means that several things happen. .PP First, any macros that are executing are stopped and popped off the -stack. +execution stack. The behavior is not unlike that of exceptions in programming languages. Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. .PP +However, the stack of values is \f[I]not\f[R] cleared; in interactive +mode, users can inspect the stack and manipulate it. +.PP Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the \f[B]EXIT STATUS\f[R] section), it asks for more input; otherwise, it exits with the appropriate return code. .SH PERFORMANCE Most dc(1) implementations use \f[B]char\f[R] types to calculate the value of \f[B]1\f[R] decimal digit at a time, but that can be slow. This dc(1) does something different. .PP It uses large integers to calculate more than \f[B]1\f[R] decimal digit at a time. If built in a environment where \f[B]DC_LONG_BIT\f[R] (see the \f[B]LIMITS\f[R] section) is \f[B]64\f[R], then each integer has \f[B]9\f[R] decimal digits. If built in an environment where \f[B]DC_LONG_BIT\f[R] is \f[B]32\f[R] then each integer has \f[B]4\f[R] decimal digits. This value (the number of decimal digits per large integer) is called \f[B]DC_BASE_DIGS\f[R]. .PP In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of \f[B]DC_LONG_BIT\f[R], but is always at least twice as large as the integer type used to store digits. .SH LIMITS The following are the limits on dc(1): .TP \f[B]DC_LONG_BIT\f[R] The number of bits in the \f[B]long\f[R] type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the \f[B]PERFORMANCE\f[R] section). .TP \f[B]DC_BASE_DIGS\f[R] The number of decimal digits per large integer (see the \f[B]PERFORMANCE\f[R] section). Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_POW\f[R] The max decimal number that each large integer can store (see \f[B]DC_BASE_DIGS\f[R]) plus \f[B]1\f[R]. Depends on \f[B]DC_BASE_DIGS\f[R]. .TP \f[B]DC_OVERFLOW_MAX\f[R] The max number that the overflow type (see the \f[B]PERFORMANCE\f[R] section) can hold. Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_MAX\f[R] The maximum output base. Set at \f[B]DC_BASE_POW\f[R]. .TP \f[B]DC_DIM_MAX\f[R] The maximum size of arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .TP \f[B]DC_SCALE_MAX\f[R] The maximum \f[B]scale\f[R]. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_STRING_MAX\f[R] The maximum length of strings. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NAME_MAX\f[R] The maximum length of identifiers. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NUM_MAX\f[R] The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP Exponent The maximum allowable exponent (positive or negative). Set at \f[B]DC_OVERFLOW_MAX\f[R]. .TP Number of vars The maximum number of vars/arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .PP These limits are meant to be effectively non\-existent; the limits are so large (at least on 64\-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. .SH ENVIRONMENT VARIABLES As \f[B]non\-portable extensions\f[R], dc(1) recognizes the following environment variables: .TP \f[B]DC_ENV_ARGS\f[R] This is another way to give command\-line arguments to dc(1). They should be in the same format as all other command\-line arguments. These are always processed first, so any files given in \f[B]DC_ENV_ARGS\f[R] will be processed before arguments and files given on the command\-line. This gives the user the ability to set up \[lq]standard\[rq] options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the \f[B]\-e\f[R] option to set \f[B]scale\f[R] to a value other than \f[B]0\f[R]. .RS .PP The code that parses \f[B]DC_ENV_ARGS\f[R] will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string \f[B]\[lq]/home/gavin/some dc file.dc\[rq]\f[R] will be correctly parsed, but the string \f[B]\[lq]/home/gavin/some \[dq]dc\[dq] file.dc\[rq]\f[R] will include the backslashes. .PP The quote parsing will handle either kind of quotes, \f[B]\[cq]\f[R] or \f[B]\[lq]\f[R]. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in \f[B]\[lq]some `dc' file.dc\[rq]\f[R], and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in \f[B]DC_ENV_ARGS\f[R] is not supported due to the complexity of the parsing, though such files are still supported on the command\-line where the parsing is done by the shell. .RE .TP \f[B]DC_LINE_LENGTH\f[R] If this environment variable exists and contains an integer that is greater than \f[B]1\f[R] and is less than \f[B]UINT16_MAX\f[R] (\f[B]2\[ha]16\-1\f[R]), dc(1) will output lines to that length, including the backslash newline combo. The default line length is \f[B]70\f[R]. .RS .PP The special value of \f[B]0\f[R] will disable line length checking and print numbers without regard to line length and without backslashes and newlines. .RE .TP \f[B]DC_SIGINT_RESET\f[R] If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), then this environment variable has no effect because dc(1) exits on \f[B]SIGINT\f[R] when not in interactive mode. .RS .PP However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) reset on \f[B]SIGINT\f[R], rather than exit, and zero makes dc(1) exit. If this environment variable exists and is \f[I]not\f[R] an integer, then dc(1) will exit on \f[B]SIGINT\f[R]. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_TTY_MODE\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non\-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_PROMPT\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) use a prompt, and zero or a non\-integer makes dc(1) not use a prompt. If this environment variable does not exist and \f[B]DC_TTY_MODE\f[R] does, then the value of the \f[B]DC_TTY_MODE\f[R] environment variable is used. .PP This environment variable and the \f[B]DC_TTY_MODE\f[R] environment variable override the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_EXPR_EXIT\f[R] If any expressions or expression files are given on the command\-line with \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R], then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. .RS .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_DIGIT_CLAMP\f[R] When parsing numbers and if this environment variable exists and contains an integer, a non\-zero value makes dc(1) clamp digits that are greater than or equal to the current \f[B]ibase\f[R] so that all such digits are considered equal to the \f[B]ibase\f[R] minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the \f[B]ibase\f[R]. .RS .PP This never applies to single\-digit numbers, as per the bc(1) standard (see the \f[B]STANDARDS\f[R] section). .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .SH EXIT STATUS dc(1) returns the following exit statuses: .TP \f[B]0\f[R] No error. .TP \f[B]1\f[R] A math error occurred. This follows standard practice of using \f[B]1\f[R] for expected errors, since math errors will happen in the process of normal execution. .RS .PP Math errors include divide by \f[B]0\f[R], taking the square root of a negative number, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non\-integer where an integer is required. .PP Converting to a hardware integer happens for the second operand of the power (\f[B]\[ha]\f[R]) operator. .RE .TP \f[B]2\f[R] A parse error occurred. .RS .PP Parse errors include unexpected \f[B]EOF\f[R], using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. .RE .TP \f[B]3\f[R] A runtime error occurred. .RS .PP Runtime errors include assigning an invalid number to any global (\f[B]ibase\f[R], \f[B]obase\f[R], or \f[B]scale\f[R]), giving a bad expression to a \f[B]read()\f[R] call, calling \f[B]read()\f[R] inside of a \f[B]read()\f[R] call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. .RE .TP \f[B]4\f[R] A fatal error occurred. .RS .PP Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command\-line options. .RE .PP The exit status \f[B]4\f[R] is special; when a fatal error occurs, dc(1) always exits and returns \f[B]4\f[R], no matter what mode dc(1) is in. .PP The other statuses will only be returned when dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), since dc(1) resets its state (see the \f[B]RESET\f[R] section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .PP These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .SH INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non\-interactive mode. Interactive mode is turned on automatically when both \f[B]stdin\f[R] and \f[B]stdout\f[R] are hooked to a terminal, but the \f[B]\-i\f[R] flag and \f[B]\-\-interactive\f[R] option can turn it on in other situations. .PP In interactive mode, dc(1) attempts to recover from errors (see the \f[B]RESET\f[R] section), and in normal execution, flushes \f[B]stdout\f[R] as soon as execution is done for the current input. dc(1) may also reset on \f[B]SIGINT\f[R] instead of exit, depending on the contents of, or default for, the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .SH TTY MODE If \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY, then \[lq]TTY mode\[rq] is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. .PP If there is the environment variable \f[B]DC_TTY_MODE\f[R] in the environment (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then if that environment variable contains a non\-zero integer, dc(1) will turn on TTY mode when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY. If the \f[B]DC_TTY_MODE\f[R] environment variable exists but is \f[I]not\f[R] a non\-zero integer, then dc(1) will not turn TTY mode on. .PP If the environment variable \f[B]DC_TTY_MODE\f[R] does \f[I]not\f[R] exist, the default setting is used. The default setting can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the \f[B]STANDARDS\f[R] section), and interactive mode requires only \f[B]stdin\f[R] and \f[B]stdout\f[R] to be connected to a terminal. .SS Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: \f[B]DC_PROMPT\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP If the environment variable \f[B]DC_PROMPT\f[R] exists and is a non\-zero integer, then the prompt is turned on when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are connected to a TTY and the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options were not used. The read prompt will be turned on under the same conditions, except that the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options must also not be used. .PP However, if \f[B]DC_PROMPT\f[R] does not exist, the prompt can be enabled or disabled with the \f[B]DC_TTY_MODE\f[R] environment variable, the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options, and the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options. See the \f[B]ENVIRONMENT VARIABLES\f[R] and \f[B]OPTIONS\f[R] sections for more details. .SH SIGNAL HANDLING Sending a \f[B]SIGINT\f[R] will cause dc(1) to do one of two things. .PP If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), or the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or its default, is either not an integer or it is zero, dc(1) will exit. .PP However, if dc(1) is in interactive mode, and the \f[B]DC_SIGINT_RESET\f[R] or its default is an integer and non\-zero, then dc(1) will stop executing the current input and reset (see the \f[B]RESET\f[R] section) upon receiving a \f[B]SIGINT\f[R]. .PP Note that \[lq]current input\[rq] can mean one of two things. If dc(1) is processing input from \f[B]stdin\f[R] in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from \f[B]stdin\f[R] if no other file exists. .PP This means that if a \f[B]SIGINT\f[R] is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. .PP \f[B]SIGTERM\f[R] and \f[B]SIGQUIT\f[R] cause dc(1) to clean up and exit, and it uses the default handler for all other signals. .SH SEE ALSO bc(1) .SH STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1\-2017 (\[lq]POSIX.1\-2017\[rq]) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . .SH BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . .SH AUTHOR Gavin D. Howard \c .MT gavin@gavinhoward.com .ME \c \ and contributors. diff --git a/manuals/dc/EHN.1.md b/manuals/dc/EHN.1.md index 58ae149bb686..51e30849996e 100644 --- a/manuals/dc/EHN.1.md +++ b/manuals/dc/EHN.1.md @@ -1,1322 +1,1325 @@ # Name dc - arbitrary-precision decimal reverse-Polish notation calculator # SYNOPSIS **dc** [**-cChiPRvVx**] [**-\-version**] [**-\-help**] [**-\-digit-clamp**] [**-\-no-digit-clamp**] [**-\-interactive**] [**-\-no-prompt**] [**-\-no-read-prompt**] [**-\-extended-register**] [**-e** *expr*] [**-\-expression**=*expr*...] [**-f** *file*...] [**-\-file**=*file*...] [*file*...] # DESCRIPTION dc(1) is an arbitrary-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. If no files are given on the command-line, then dc(1) reads from **stdin** (see the **STDIN** section). Otherwise, those files are processed, and dc(1) will then exit. If a user wants to set up a standard environment, they can use **DC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). For example, if a user wants the **scale** always set to **10**, they can set **DC_ENV_ARGS** to **-e 10k**, and this dc(1) will always start with a **scale** of **10**. # OPTIONS The following are the options that dc(1) accepts. **-C**, **-\-no-digit-clamp** : Disables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that the value added to a number from a digit is always that digit's value multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-c** or **-\-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-c**, **-\-digit-clamp** : Enables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-C** or **-\-no-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-e** *expr*, **-\-expression**=*expr* : Evaluates *expr*. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**, whether on the command-line or in **DC_ENV_ARGS**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-f** *file*, **-\-file**=*file* : Reads in *file* and evaluates it, line by line, as though it were read through **stdin**. If expressions are also given (see above), the expressions are evaluated in the order given. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-h**, **-\-help** : Prints a usage message and exits. **-I** *ibase*, **-\-ibase**=*ibase* : Sets the builtin variable **ibase** to the value *ibase* assuming that *ibase* is in base 10. It is a fatal error if *ibase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-i**, **-\-interactive** : Forces interactive mode. (See the **INTERACTIVE MODE** section.) This is a **non-portable extension**. **-L**, **-\-no-line-length** : Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets **BC_LINE_LENGTH** to **0** (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-O** *obase*, **-\-obase**=*obase* : Sets the builtin variable **obase** to the value *obase* assuming that *obase* is in base 10. It is a fatal error if *obase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-P**, **-\-no-prompt** : Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in **DC_ENV_ARGS**. These options override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-R**, **-\-no-read-prompt** : Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in **BC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. This option does not disable the regular prompt because the read prompt is only used when the **?** command is used. These options *do* override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section), but only for the read prompt. This is a **non-portable extension**. **-S** *scale*, **-\-scale**=*scale* : Sets the builtin variable **scale** to the value *scale* assuming that *scale* is in base 10. It is a fatal error if *scale* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-v**, **-V**, **-\-version** : Print the version information (copyright header) and exits. **-x** **-\-extended-register** : Enables extended register mode. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. This is a **non-portable extension**. **-z**, **-\-leading-zeroes** : Makes dc(1) print all numbers greater than **-1** and less than **1**, and not equal to **0**, with a leading zero. This is a **non-portable extension**. All long options are **non-portable extensions**. # STDIN If no files are given on the command-line and no files or expressions are given by the **-f**, **-\-file**, **-e**, or **-\-expression** options, then dc(1) reads from **stdin**. However, there is a caveat to this. First, **stdin** is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. # STDOUT Any non-error output is written to **stdout**. In addition, if history (see the **HISTORY** section) and the prompt (see the **TTY MODE** section) are enabled, both are output to **stdout**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stdout**, so if **stdout** is closed, as in **dc >&-**, it will quit with an error. This is done so that dc(1) can report problems when **stdout** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stdout** to **/dev/null**. # STDERR Any error output is written to **stderr**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stderr**, so if **stderr** is closed, as in **dc 2>&-**, it will quit with an error. This is done so that dc(1) can exit with an error code when **stderr** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stderr** to **/dev/null**. # SYNTAX Each item in the input source code, either a number (see the **NUMBERS** section) or a command (see the **COMMANDS** section), is processed and executed, in order. Input is processed immediately when entered. **ibase** is a register (see the **REGISTERS** section) that determines how to interpret constant numbers. It is the "input" base, or the number base used for interpreting input numbers. **ibase** is initially **10**. The max allowable value for **ibase** is **16**. The min allowable value for **ibase** is **2**. The max allowable value for **ibase** can be queried in dc(1) programs with the **T** command. **obase** is a register (see the **REGISTERS** section) that determines how to output results. It is the "output" base, or the number base used for outputting numbers. **obase** is initially **10**. The max allowable value for **obase** is **DC_BASE_MAX** and can be queried with the **U** command. The min allowable value for **obase** is **2**. Values are output in the specified base. The *scale* of an expression is the number of digits in the result of the expression right of the decimal point, and **scale** is a register (see the **REGISTERS** section) that sets the precision of any operations (with exceptions). **scale** is initially **0**. **scale** cannot be negative. The max allowable value for **scale** can be queried in dc(1) programs with the **V** command. ## Comments Comments go from **#** until, and not including, the next newline. This is a **non-portable extension**. # NUMBERS Numbers are strings made up of digits, uppercase letters up to **F**, and at most **1** period for a radix. Numbers can have up to **DC_NUM_MAX** digits. Uppercase letters are equal to **9** plus their position in the alphabet (i.e., **A** equals **10**, or **9+1**). If a digit or letter makes no sense with the current value of **ibase** (i.e., they are greater than or equal to the current value of **ibase**), then the behavior depends on the existence of the **-c**/**-\-digit-clamp** or **-C**/**-\-no-digit-clamp** options (see the **OPTIONS** section), the existence and setting of the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section), or the default, which can be queried with the **-h**/**-\-help** option. If clamping is off, then digits or letters that are greater than or equal to the current value of **ibase** are not changed. Instead, their given value is multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*A+3\^0\*B**, which is **3** times **10** plus **11**, or **41**. If clamping is on, then digits or letters that are greater than or equal to the current value of **ibase** are set to the value of the highest valid digit in **ibase** before being multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*2+3\^0\*2**, which is **3** times **2** plus **2**, or **8**. There is one exception to clamping: single-character numbers (i.e., **A** alone). Such numbers are never clamped and always take the value they would have in the highest possible **ibase**. This means that **A** alone always equals decimal **10** and **Z** alone always equals decimal **35**. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current **ibase** (with the **i** command) regardless of the current value of **ibase**. If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for **A**, use **0A**. # COMMANDS The valid commands are listed below. ## Printing These commands are used for printing. **p** : Prints the value on top of the stack, whether number or string, and prints a newline after. This does not alter the stack. **n** : Prints the value on top of the stack, whether number or string, and pops it off of the stack. **P** : Pops a value off the stack. If the value is a number, it is truncated and the absolute value of the result is printed as though **obase** is **256** and each digit is interpreted as an 8-bit ASCII character, making it a byte stream. If the value is a string, it is printed without a trailing newline. This is a **non-portable extension**. **f** : Prints the entire contents of the stack, in order from newest to oldest, without altering anything. Users should use this command when they get lost. ## Arithmetic These are the commands used for arithmetic. **+** : The top two values are popped off the stack, added, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **-** : The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **\*** : The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If **a** is the *scale* of the first expression and **b** is the *scale* of the second expression, the *scale* of the result is equal to **min(a+b,max(scale,a,b))** where **min()** and **max()** return the obvious values. **/** : The top two values are popped off the stack, divided, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be non-zero. **%** : The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. Remaindering is equivalent to 1) Computing **a/b** to current **scale**, and 2) Using the result of step 1 to calculate **a-(a/b)\*b** to *scale* **max(scale+scale(b),scale(a))**. The first value popped off of the stack must be non-zero. **~** : The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to **x y / x y %** except that **x** and **y** are only evaluated once. The first value popped off of the stack must be non-zero. This is a **non-portable extension**. **\^** : The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non-zero. **v** : The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The value popped off of the stack must be non-negative. **\_** : If this command *immediately* precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a **non-portable extension**. **b** : The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. This is a **non-portable extension**. **|** : The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. The first value popped is used as the reduction modulus and must be an integer and non-zero. The second value popped is used as the exponent and must be an integer and non-negative. The third value popped is the base and must be an integer. This is a **non-portable extension**. **G** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if they are equal, or **0** otherwise. This is a **non-portable extension**. **N** : The top value is popped off of the stack, and if it a **0**, a **1** is pushed; otherwise, a **0** is pushed. This is a **non-portable extension**. **(** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than the second, or **0** otherwise. This is a **non-portable extension**. **{** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **)** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than the second, or **0** otherwise. This is a **non-portable extension**. **}** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **M** : The top two values are popped off of the stack. If they are both non-zero, a **1** is pushed onto the stack. If either of them is zero, or both of them are, then a **0** is pushed onto the stack. This is like the **&&** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. **m** : The top two values are popped off of the stack. If at least one of them is non-zero, a **1** is pushed onto the stack. If both of them are zero, then a **0** is pushed onto the stack. This is like the **||** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. ## Stack Control These commands control the stack. **c** : Removes all items from ("clears") the stack. **d** : Copies the item on top of the stack ("duplicates") and pushes the copy onto the stack. **r** : Swaps ("reverses") the two top items on the stack. **R** : Pops ("removes") the top value from the stack. ## Register Control These commands control registers (see the **REGISTERS** section). **s**_r_ : Pops the value off the top of the stack and stores it into register *r*. **l**_r_ : Copies the value in register *r* and pushes it onto the stack. This does not alter the contents of *r*. **S**_r_ : Pops the value off the top of the (main) stack and pushes it onto the stack of register *r*. The previous value of the register becomes inaccessible. **L**_r_ : Pops the value off the top of the stack for register *r* and push it onto the main stack. The previous value in the stack for register *r*, if any, is now accessible via the **l**_r_ command. ## Parameters These commands control the values of **ibase**, **obase**, and **scale**. Also see the **SYNTAX** section. **i** : Pops the value off of the top of the stack and uses it to set **ibase**, which must be between **2** and **16**, inclusive. If the value on top of the stack has any *scale*, the *scale* is ignored. **o** : Pops the value off of the top of the stack and uses it to set **obase**, which must be between **2** and **DC_BASE_MAX**, inclusive (see the **LIMITS** section). If the value on top of the stack has any *scale*, the *scale* is ignored. **k** : Pops the value off of the top of the stack and uses it to set **scale**, which must be non-negative. If the value on top of the stack has any *scale*, the *scale* is ignored. **I** : Pushes the current value of **ibase** onto the main stack. **O** : Pushes the current value of **obase** onto the main stack. **K** : Pushes the current value of **scale** onto the main stack. **T** : Pushes the maximum allowable value of **ibase** onto the main stack. This is a **non-portable extension**. **U** : Pushes the maximum allowable value of **obase** onto the main stack. This is a **non-portable extension**. **V** : Pushes the maximum allowable value of **scale** onto the main stack. This is a **non-portable extension**. ## Strings The following commands control strings. dc(1) can work with both numbers and strings, and registers (see the **REGISTERS** section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. Strings can also be executed as macros. For example, if the string **[1pR]** is executed as a macro, then the code **1pR** is executed, meaning that the **1** will be printed with a newline after and then popped from the stack. **\[**_characters_**\]** : Makes a string containing *characters* and pushes it onto the stack. If there are brackets (**\[** and **\]**) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (**\\**) character. If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. **a** : The value on top of the stack is popped. If it is a number, it is truncated and its absolute value is taken. The result mod **256** is calculated. If that result is **0**, push an empty string; otherwise, push a one-character string where the character is the result of the mod interpreted as an ASCII character. If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one-character string. The new string is then pushed onto the stack. This is a **non-portable extension**. **x** : Pops a value off of the top of the stack. If it is a number, it is pushed back onto the stack. If it is a string, it is executed as a macro. This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. **\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register *r* are executed. For example, **0 1>a** will execute the contents of register **a**, and **1 0>a** will not. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **?** : Reads a line from the **stdin** and executes it. This is to allow macros to request input from users. **q** : During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. **Q** : Pops a value from the stack which must be non-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. **,** : Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the **Q** command, so the sequence **,Q** will make dc(1) exit. This is a **non-portable extension**. ## Status These commands query status of the stack or its top value. **Z** : Pops a value off of the stack. If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push **1** if the argument is **0** with no decimal places. If it is a string, pushes the number of characters the string has. **X** : Pops a value off of the stack. If it is a number, pushes the *scale* of the value onto the stack. If it is a string, pushes **0**. **u** : Pops one value off of the stack. If the value is a number, this pushes **1** onto the stack. Otherwise (if it is a string), it pushes **0**. This is a **non-portable extension**. **t** : Pops one value off of the stack. If the value is a string, this pushes **1** onto the stack. Otherwise (if it is a number), it pushes **0**. This is a **non-portable extension**. **z** : Pushes the current depth of the stack (before execution of this command) onto the stack. **y**_r_ : Pushes the current stack depth of the register *r* onto the main stack. Because each register has a depth of **1** (with the value **0** in the top item) when dc(1) starts, dc(1) requires that each register's stack must always have at least one item; dc(1) will give an error and reset otherwise (see the **RESET** section). This means that this command will never push **0**. This is a **non-portable extension**. ## Arrays These commands manipulate arrays. **:**_r_ : Pops the top two values off of the stack. The second value will be stored in the array *r* (see the **REGISTERS** section), indexed by the first value. **;**_r_ : Pops the value on top of the stack and uses it as an index into the array *r*. The selected value is then pushed onto the stack. **Y**_r_ : Pushes the length of the array *r* onto the stack. This is a **non-portable extension**. ## Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter **g**. Only the characters below are allowed after the character **g**; any other character produces a parse error (see the **ERRORS** section). **gl** : Pushes the line length set by **DC_LINE_LENGTH** (see the **ENVIRONMENT VARIABLES** section) onto the stack. **gx** : Pushes **1** onto the stack if extended register mode is on, **0** otherwise. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. **gz** : Pushes **0** onto the stack if the leading zero setting has not been enabled with the **-z** or **-\-leading-zeroes** options (see the **OPTIONS** section), non-zero otherwise. # REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (**0**) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. In non-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (**'\\n'**) and a left bracket (**'['**); it is a parse error for a newline or a left bracket to be used as a register name. ## Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. If extended register mode is enabled (**-x** or **-\-extended-register** command-line arguments are given), then normal single character registers are used *unless* the character immediately following a command that needs a register name is a space (according to **isspace()**) and not a newline (**'\\n'**). In that case, the register name is found according to the regex **\[a-z\]\[a-z0-9\_\]\*** (like bc(1) identifiers), and it is a parse error if the next non-space characters do not match that regex. # RESET When dc(1) encounters an error or a signal that it has a non-default handler for, it resets. This means that several things happen. -First, any macros that are executing are stopped and popped off the stack. -The behavior is not unlike that of exceptions in programming languages. Then -the execution point is set so that any code waiting to execute (after all +First, any macros that are executing are stopped and popped off the execution +stack. The behavior is not unlike that of exceptions in programming languages. +Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. +However, the stack of values is *not* cleared; in interactive mode, users can +inspect the stack and manipulate it. + Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the **EXIT STATUS** section), it asks for more input; otherwise, it exits with the appropriate return code. # PERFORMANCE Most dc(1) implementations use **char** types to calculate the value of **1** decimal digit at a time, but that can be slow. This dc(1) does something different. It uses large integers to calculate more than **1** decimal digit at a time. If built in a environment where **DC_LONG_BIT** (see the **LIMITS** section) is **64**, then each integer has **9** decimal digits. If built in an environment where **DC_LONG_BIT** is **32** then each integer has **4** decimal digits. This value (the number of decimal digits per large integer) is called **DC_BASE_DIGS**. In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of **DC_LONG_BIT**, but is always at least twice as large as the integer type used to store digits. # LIMITS The following are the limits on dc(1): **DC_LONG_BIT** : The number of bits in the **long** type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the **PERFORMANCE** section). **DC_BASE_DIGS** : The number of decimal digits per large integer (see the **PERFORMANCE** section). Depends on **DC_LONG_BIT**. **DC_BASE_POW** : The max decimal number that each large integer can store (see **DC_BASE_DIGS**) plus **1**. Depends on **DC_BASE_DIGS**. **DC_OVERFLOW_MAX** : The max number that the overflow type (see the **PERFORMANCE** section) can hold. Depends on **DC_LONG_BIT**. **DC_BASE_MAX** : The maximum output base. Set at **DC_BASE_POW**. **DC_DIM_MAX** : The maximum size of arrays. Set at **SIZE_MAX-1**. **DC_SCALE_MAX** : The maximum **scale**. Set at **DC_OVERFLOW_MAX-1**. **DC_STRING_MAX** : The maximum length of strings. Set at **DC_OVERFLOW_MAX-1**. **DC_NAME_MAX** : The maximum length of identifiers. Set at **DC_OVERFLOW_MAX-1**. **DC_NUM_MAX** : The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at **DC_OVERFLOW_MAX-1**. Exponent : The maximum allowable exponent (positive or negative). Set at **DC_OVERFLOW_MAX**. Number of vars : The maximum number of vars/arrays. Set at **SIZE_MAX-1**. These limits are meant to be effectively non-existent; the limits are so large (at least on 64-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. # ENVIRONMENT VARIABLES As **non-portable extensions**, dc(1) recognizes the following environment variables: **DC_ENV_ARGS** : This is another way to give command-line arguments to dc(1). They should be in the same format as all other command-line arguments. These are always processed first, so any files given in **DC_ENV_ARGS** will be processed before arguments and files given on the command-line. This gives the user the ability to set up "standard" options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the **-e** option to set **scale** to a value other than **0**. The code that parses **DC_ENV_ARGS** will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string **"/home/gavin/some dc file.dc"** will be correctly parsed, but the string **"/home/gavin/some \"dc\" file.dc"** will include the backslashes. The quote parsing will handle either kind of quotes, **'** or **"**. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in **"some 'dc' file.dc"**, and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in **DC_ENV_ARGS** is not supported due to the complexity of the parsing, though such files are still supported on the command-line where the parsing is done by the shell. **DC_LINE_LENGTH** : If this environment variable exists and contains an integer that is greater than **1** and is less than **UINT16_MAX** (**2\^16-1**), dc(1) will output lines to that length, including the backslash newline combo. The default line length is **70**. The special value of **0** will disable line length checking and print numbers without regard to line length and without backslashes and newlines. **DC_SIGINT_RESET** : If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), then this environment variable has no effect because dc(1) exits on **SIGINT** when not in interactive mode. However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) reset on **SIGINT**, rather than exit, and zero makes dc(1) exit. If this environment variable exists and is *not* an integer, then dc(1) will exit on **SIGINT**. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_TTY_MODE** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_PROMPT** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) use a prompt, and zero or a non-integer makes dc(1) not use a prompt. If this environment variable does not exist and **DC_TTY_MODE** does, then the value of the **DC_TTY_MODE** environment variable is used. This environment variable and the **DC_TTY_MODE** environment variable override the default, which can be queried with the **-h** or **-\-help** options. **DC_EXPR_EXIT** : If any expressions or expression files are given on the command-line with **-e**, **-\-expression**, **-f**, or **-\-file**, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_DIGIT_CLAMP** : When parsing numbers and if this environment variable exists and contains an integer, a non-zero value makes dc(1) clamp digits that are greater than or equal to the current **ibase** so that all such digits are considered equal to the **ibase** minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the **ibase**. This never applies to single-digit numbers, as per the bc(1) standard (see the **STANDARDS** section). This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. # EXIT STATUS dc(1) returns the following exit statuses: **0** : No error. **1** : A math error occurred. This follows standard practice of using **1** for expected errors, since math errors will happen in the process of normal execution. Math errors include divide by **0**, taking the square root of a negative number, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non-integer where an integer is required. Converting to a hardware integer happens for the second operand of the power (**\^**) operator. **2** : A parse error occurred. Parse errors include unexpected **EOF**, using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. **3** : A runtime error occurred. Runtime errors include assigning an invalid number to any global (**ibase**, **obase**, or **scale**), giving a bad expression to a **read()** call, calling **read()** inside of a **read()** call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. **4** : A fatal error occurred. Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command-line options. The exit status **4** is special; when a fatal error occurs, dc(1) always exits and returns **4**, no matter what mode dc(1) is in. The other statuses will only be returned when dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), since dc(1) resets its state (see the **RESET** section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the **-i** flag or **-\-interactive** option. These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the **-i** flag or **-\-interactive** option. # INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non-interactive mode. Interactive mode is turned on automatically when both **stdin** and **stdout** are hooked to a terminal, but the **-i** flag and **-\-interactive** option can turn it on in other situations. In interactive mode, dc(1) attempts to recover from errors (see the **RESET** section), and in normal execution, flushes **stdout** as soon as execution is done for the current input. dc(1) may also reset on **SIGINT** instead of exit, depending on the contents of, or default for, the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section). # TTY MODE If **stdin**, **stdout**, and **stderr** are all connected to a TTY, then "TTY mode" is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. If there is the environment variable **DC_TTY_MODE** in the environment (see the **ENVIRONMENT VARIABLES** section), then if that environment variable contains a non-zero integer, dc(1) will turn on TTY mode when **stdin**, **stdout**, and **stderr** are all connected to a TTY. If the **DC_TTY_MODE** environment variable exists but is *not* a non-zero integer, then dc(1) will not turn TTY mode on. If the environment variable **DC_TTY_MODE** does *not* exist, the default setting is used. The default setting can be queried with the **-h** or **-\-help** options. TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the **STANDARDS** section), and interactive mode requires only **stdin** and **stdout** to be connected to a terminal. ## Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: **DC_PROMPT** (see the **ENVIRONMENT VARIABLES** section). If the environment variable **DC_PROMPT** exists and is a non-zero integer, then the prompt is turned on when **stdin**, **stdout**, and **stderr** are connected to a TTY and the **-P** and **-\-no-prompt** options were not used. The read prompt will be turned on under the same conditions, except that the **-R** and **-\-no-read-prompt** options must also not be used. However, if **DC_PROMPT** does not exist, the prompt can be enabled or disabled with the **DC_TTY_MODE** environment variable, the **-P** and **-\-no-prompt** options, and the **-R** and **-\-no-read-prompt** options. See the **ENVIRONMENT VARIABLES** and **OPTIONS** sections for more details. # SIGNAL HANDLING Sending a **SIGINT** will cause dc(1) to do one of two things. If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), or the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section), or its default, is either not an integer or it is zero, dc(1) will exit. However, if dc(1) is in interactive mode, and the **DC_SIGINT_RESET** or its default is an integer and non-zero, then dc(1) will stop executing the current input and reset (see the **RESET** section) upon receiving a **SIGINT**. Note that "current input" can mean one of two things. If dc(1) is processing input from **stdin** in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from **stdin** if no other file exists. This means that if a **SIGINT** is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. **SIGTERM** and **SIGQUIT** cause dc(1) to clean up and exit, and it uses the default handler for all other signals. # SEE ALSO bc(1) # STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1-2017 (“POSIX.1-2017”) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . # BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . # AUTHOR Gavin D. Howard and contributors. diff --git a/manuals/dc/EN.1 b/manuals/dc/EN.1 index 03cb5743602c..5ce8defc91c7 100644 --- a/manuals/dc/EN.1 +++ b/manuals/dc/EN.1 @@ -1,1468 +1,1471 @@ .\" .\" SPDX-License-Identifier: BSD-2-Clause .\" .\" Copyright (c) 2018-2024 Gavin D. Howard and contributors. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions are met: .\" .\" * Redistributions of source code must retain the above copyright notice, .\" this list of conditions and the following disclaimer. .\" .\" * Redistributions in binary form must reproduce the above copyright notice, .\" this list of conditions and the following disclaimer in the documentation .\" and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" .\" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE .\" LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF .\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS .\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN .\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) .\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE .\" POSSIBILITY OF SUCH DAMAGE. .\" -.TH "DC" "1" "January 2024" "Gavin D. Howard" "General Commands Manual" +.TH "DC" "1" "August 2024" "Gavin D. Howard" "General Commands Manual" .nh .ad l .SH Name dc \- arbitrary\-precision decimal reverse\-Polish notation calculator .SH SYNOPSIS \f[B]dc\f[R] [\f[B]\-cChiPRvVx\f[R]] [\f[B]\-\-version\f[R]] [\f[B]\-\-help\f[R]] [\f[B]\-\-digit\-clamp\f[R]] [\f[B]\-\-no\-digit\-clamp\f[R]] [\f[B]\-\-interactive\f[R]] [\f[B]\-\-no\-prompt\f[R]] [\f[B]\-\-no\-read\-prompt\f[R]] [\f[B]\-\-extended\-register\f[R]] [\f[B]\-e\f[R] \f[I]expr\f[R]] [\f[B]\-\-expression\f[R]=\f[I]expr\f[R]\&...] [\f[B]\-f\f[R] \f[I]file\f[R]\&...] [\f[B]\-\-file\f[R]=\f[I]file\f[R]\&...] [\f[I]file\f[R]\&...] .SH DESCRIPTION dc(1) is an arbitrary\-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. .PP If no files are given on the command\-line, then dc(1) reads from \f[B]stdin\f[R] (see the \f[B]STDIN\f[R] section). Otherwise, those files are processed, and dc(1) will then exit. .PP If a user wants to set up a standard environment, they can use \f[B]DC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). For example, if a user wants the \f[B]scale\f[R] always set to \f[B]10\f[R], they can set \f[B]DC_ENV_ARGS\f[R] to \f[B]\-e 10k\f[R], and this dc(1) will always start with a \f[B]scale\f[R] of \f[B]10\f[R]. .SH OPTIONS The following are the options that dc(1) accepts. .TP \f[B]\-C\f[R], \f[B]\-\-no\-digit\-clamp\f[R] Disables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that the value added to a number from a digit is always that digit\[cq]s value multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-c\f[R] or \f[B]\-\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-c\f[R], \f[B]\-\-digit\-clamp\f[R] Enables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-C\f[R] or \f[B]\-\-no\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-e\f[R] \f[I]expr\f[R], \f[B]\-\-expression\f[R]=\f[I]expr\f[R] Evaluates \f[I]expr\f[R]. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R], whether on the command\-line or in \f[B]DC_ENV_ARGS\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-f\f[R] \f[I]file\f[R], \f[B]\-\-file\f[R]=\f[I]file\f[R] Reads in \f[I]file\f[R] and evaluates it, line by line, as though it were read through \f[B]stdin\f[R]. If expressions are also given (see above), the expressions are evaluated in the order given. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-h\f[R], \f[B]\-\-help\f[R] Prints a usage message and exits. .TP \f[B]\-I\f[R] \f[I]ibase\f[R], \f[B]\-\-ibase\f[R]=\f[I]ibase\f[R] Sets the builtin variable \f[B]ibase\f[R] to the value \f[I]ibase\f[R] assuming that \f[I]ibase\f[R] is in base 10. It is a fatal error if \f[I]ibase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-i\f[R], \f[B]\-\-interactive\f[R] Forces interactive mode. (See the \f[B]INTERACTIVE MODE\f[R] section.) .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-L\f[R], \f[B]\-\-no\-line\-length\f[R] Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets \f[B]BC_LINE_LENGTH\f[R] to \f[B]0\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-O\f[R] \f[I]obase\f[R], \f[B]\-\-obase\f[R]=\f[I]obase\f[R] Sets the builtin variable \f[B]obase\f[R] to the value \f[I]obase\f[R] assuming that \f[I]obase\f[R] is in base 10. It is a fatal error if \f[I]obase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-P\f[R], \f[B]\-\-no\-prompt\f[R] Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]DC_ENV_ARGS\f[R]. .RS .PP These options override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-R\f[R], \f[B]\-\-no\-read\-prompt\f[R] Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]BC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. .RS .PP This option does not disable the regular prompt because the read prompt is only used when the \f[B]?\f[R] command is used. .PP These options \f[I]do\f[R] override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), but only for the read prompt. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-S\f[R] \f[I]scale\f[R], \f[B]\-\-scale\f[R]=\f[I]scale\f[R] Sets the builtin variable \f[B]scale\f[R] to the value \f[I]scale\f[R] assuming that \f[I]scale\f[R] is in base 10. It is a fatal error if \f[I]scale\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-v\f[R], \f[B]\-V\f[R], \f[B]\-\-version\f[R] Print the version information (copyright header) and exits. .TP \f[B]\-x\f[R] \f[B]\-\-extended\-register\f[R] Enables extended register mode. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-z\f[R], \f[B]\-\-leading\-zeroes\f[R] Makes dc(1) print all numbers greater than \f[B]\-1\f[R] and less than \f[B]1\f[R], and not equal to \f[B]0\f[R], with a leading zero. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .PP All long options are \f[B]non\-portable extensions\f[R]. .SH STDIN If no files are given on the command\-line and no files or expressions are given by the \f[B]\-f\f[R], \f[B]\-\-file\f[R], \f[B]\-e\f[R], or \f[B]\-\-expression\f[R] options, then dc(1) reads from \f[B]stdin\f[R]. .PP However, there is a caveat to this. .PP First, \f[B]stdin\f[R] is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. .SH STDOUT Any non\-error output is written to \f[B]stdout\f[R]. In addition, if history (see the \f[B]HISTORY\f[R] section) and the prompt (see the \f[B]TTY MODE\f[R] section) are enabled, both are output to \f[B]stdout\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stdout\f[R], so if \f[B]stdout\f[R] is closed, as in \f[B]dc >&\-\f[R], it will quit with an error. This is done so that dc(1) can report problems when \f[B]stdout\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stdout\f[R] to \f[B]/dev/null\f[R]. .SH STDERR Any error output is written to \f[B]stderr\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stderr\f[R], so if \f[B]stderr\f[R] is closed, as in \f[B]dc 2>&\-\f[R], it will quit with an error. This is done so that dc(1) can exit with an error code when \f[B]stderr\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stderr\f[R] to \f[B]/dev/null\f[R]. .SH SYNTAX Each item in the input source code, either a number (see the \f[B]NUMBERS\f[R] section) or a command (see the \f[B]COMMANDS\f[R] section), is processed and executed, in order. Input is processed immediately when entered. .PP \f[B]ibase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to interpret constant numbers. It is the \[lq]input\[rq] base, or the number base used for interpreting input numbers. \f[B]ibase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]ibase\f[R] is \f[B]16\f[R]. The min allowable value for \f[B]ibase\f[R] is \f[B]2\f[R]. The max allowable value for \f[B]ibase\f[R] can be queried in dc(1) programs with the \f[B]T\f[R] command. .PP \f[B]obase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to output results. It is the \[lq]output\[rq] base, or the number base used for outputting numbers. \f[B]obase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]obase\f[R] is \f[B]DC_BASE_MAX\f[R] and can be queried with the \f[B]U\f[R] command. The min allowable value for \f[B]obase\f[R] is \f[B]2\f[R]. Values are output in the specified base. .PP The \f[I]scale\f[R] of an expression is the number of digits in the result of the expression right of the decimal point, and \f[B]scale\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that sets the precision of any operations (with exceptions). \f[B]scale\f[R] is initially \f[B]0\f[R]. \f[B]scale\f[R] cannot be negative. The max allowable value for \f[B]scale\f[R] can be queried in dc(1) programs with the \f[B]V\f[R] command. .SS Comments Comments go from \f[B]#\f[R] until, and not including, the next newline. This is a \f[B]non\-portable extension\f[R]. .SH NUMBERS Numbers are strings made up of digits, uppercase letters up to \f[B]F\f[R], and at most \f[B]1\f[R] period for a radix. Numbers can have up to \f[B]DC_NUM_MAX\f[R] digits. Uppercase letters are equal to \f[B]9\f[R] plus their position in the alphabet (i.e., \f[B]A\f[R] equals \f[B]10\f[R], or \f[B]9+1\f[R]). .PP If a digit or letter makes no sense with the current value of \f[B]ibase\f[R] (i.e., they are greater than or equal to the current value of \f[B]ibase\f[R]), then the behavior depends on the existence of the \f[B]\-c\f[R]/\f[B]\-\-digit\-clamp\f[R] or \f[B]\-C\f[R]/\f[B]\-\-no\-digit\-clamp\f[R] options (see the \f[B]OPTIONS\f[R] section), the existence and setting of the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or the default, which can be queried with the \f[B]\-h\f[R]/\f[B]\-\-help\f[R] option. .PP If clamping is off, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are not changed. Instead, their given value is multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*A+3\[ha]0*B\f[R], which is \f[B]3\f[R] times \f[B]10\f[R] plus \f[B]11\f[R], or \f[B]41\f[R]. .PP If clamping is on, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are set to the value of the highest valid digit in \f[B]ibase\f[R] before being multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*2+3\[ha]0*2\f[R], which is \f[B]3\f[R] times \f[B]2\f[R] plus \f[B]2\f[R], or \f[B]8\f[R]. .PP There is one exception to clamping: single\-character numbers (i.e., \f[B]A\f[R] alone). Such numbers are never clamped and always take the value they would have in the highest possible \f[B]ibase\f[R]. This means that \f[B]A\f[R] alone always equals decimal \f[B]10\f[R] and \f[B]Z\f[R] alone always equals decimal \f[B]35\f[R]. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current \f[B]ibase\f[R] (with the \f[B]i\f[R] command) regardless of the current value of \f[B]ibase\f[R]. .PP If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for \f[B]A\f[R], use \f[B]0A\f[R]. .SH COMMANDS The valid commands are listed below. .SS Printing These commands are used for printing. .TP \f[B]p\f[R] Prints the value on top of the stack, whether number or string, and prints a newline after. .RS .PP This does not alter the stack. .RE .TP \f[B]n\f[R] Prints the value on top of the stack, whether number or string, and pops it off of the stack. .TP \f[B]P\f[R] Pops a value off the stack. .RS .PP If the value is a number, it is truncated and the absolute value of the result is printed as though \f[B]obase\f[R] is \f[B]256\f[R] and each digit is interpreted as an 8\-bit ASCII character, making it a byte stream. .PP If the value is a string, it is printed without a trailing newline. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]f\f[R] Prints the entire contents of the stack, in order from newest to oldest, without altering anything. .RS .PP Users should use this command when they get lost. .RE .SS Arithmetic These are the commands used for arithmetic. .TP \f[B]+\f[R] The top two values are popped off the stack, added, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]\-\f[R] The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]*\f[R] The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If \f[B]a\f[R] is the \f[I]scale\f[R] of the first expression and \f[B]b\f[R] is the \f[I]scale\f[R] of the second expression, the \f[I]scale\f[R] of the result is equal to \f[B]min(a+b,max(scale,a,b))\f[R] where \f[B]min()\f[R] and \f[B]max()\f[R] return the obvious values. .TP \f[B]/\f[R] The top two values are popped off the stack, divided, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]%\f[R] The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. .RS .PP Remaindering is equivalent to 1) Computing \f[B]a/b\f[R] to current \f[B]scale\f[R], and 2) Using the result of step 1 to calculate \f[B]a\-(a/b)*b\f[R] to \f[I]scale\f[R] \f[B]max(scale+scale(b),scale(a))\f[R]. .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]\[ti]\f[R] The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to \f[B]x y / x y %\f[R] except that \f[B]x\f[R] and \f[B]y\f[R] are only evaluated once. .RS .PP The first value popped off of the stack must be non\-zero. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[ha]\f[R] The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non\-zero. .RE .TP \f[B]v\f[R] The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The value popped off of the stack must be non\-negative. .RE .TP \f[B]_\f[R] If this command \f[I]immediately\f[R] precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. .RS .PP Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]b\f[R] The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]|\f[R] The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. .RS .PP The first value popped is used as the reduction modulus and must be an integer and non\-zero. The second value popped is used as the exponent and must be an integer and non\-negative. The third value popped is the base and must be an integer. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]G\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if they are equal, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]N\f[R] The top value is popped off of the stack, and if it a \f[B]0\f[R], a \f[B]1\f[R] is pushed; otherwise, a \f[B]0\f[R] is pushed. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B](\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]{\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B])\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]}\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]M\f[R] The top two values are popped off of the stack. If they are both non\-zero, a \f[B]1\f[R] is pushed onto the stack. If either of them is zero, or both of them are, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]&&\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]m\f[R] The top two values are popped off of the stack. If at least one of them is non\-zero, a \f[B]1\f[R] is pushed onto the stack. If both of them are zero, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]||\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Stack Control These commands control the stack. .TP \f[B]c\f[R] Removes all items from (\[lq]clears\[rq]) the stack. .TP \f[B]d\f[R] Copies the item on top of the stack (\[lq]duplicates\[rq]) and pushes the copy onto the stack. .TP \f[B]r\f[R] Swaps (\[lq]reverses\[rq]) the two top items on the stack. .TP \f[B]R\f[R] Pops (\[lq]removes\[rq]) the top value from the stack. .SS Register Control These commands control registers (see the \f[B]REGISTERS\f[R] section). .TP \f[B]s\f[R]\f[I]r\f[R] Pops the value off the top of the stack and stores it into register \f[I]r\f[R]. .TP \f[B]l\f[R]\f[I]r\f[R] Copies the value in register \f[I]r\f[R] and pushes it onto the stack. This does not alter the contents of \f[I]r\f[R]. .TP \f[B]S\f[R]\f[I]r\f[R] Pops the value off the top of the (main) stack and pushes it onto the stack of register \f[I]r\f[R]. The previous value of the register becomes inaccessible. .TP \f[B]L\f[R]\f[I]r\f[R] Pops the value off the top of the stack for register \f[I]r\f[R] and push it onto the main stack. The previous value in the stack for register \f[I]r\f[R], if any, is now accessible via the \f[B]l\f[R]\f[I]r\f[R] command. .SS Parameters These commands control the values of \f[B]ibase\f[R], \f[B]obase\f[R], and \f[B]scale\f[R]. Also see the \f[B]SYNTAX\f[R] section. .TP \f[B]i\f[R] Pops the value off of the top of the stack and uses it to set \f[B]ibase\f[R], which must be between \f[B]2\f[R] and \f[B]16\f[R], inclusive. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]o\f[R] Pops the value off of the top of the stack and uses it to set \f[B]obase\f[R], which must be between \f[B]2\f[R] and \f[B]DC_BASE_MAX\f[R], inclusive (see the \f[B]LIMITS\f[R] section). .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]k\f[R] Pops the value off of the top of the stack and uses it to set \f[B]scale\f[R], which must be non\-negative. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]I\f[R] Pushes the current value of \f[B]ibase\f[R] onto the main stack. .TP \f[B]O\f[R] Pushes the current value of \f[B]obase\f[R] onto the main stack. .TP \f[B]K\f[R] Pushes the current value of \f[B]scale\f[R] onto the main stack. .TP \f[B]T\f[R] Pushes the maximum allowable value of \f[B]ibase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]U\f[R] Pushes the maximum allowable value of \f[B]obase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]V\f[R] Pushes the maximum allowable value of \f[B]scale\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Strings The following commands control strings. .PP dc(1) can work with both numbers and strings, and registers (see the \f[B]REGISTERS\f[R] section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. .PP While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. .PP Strings can also be executed as macros. For example, if the string \f[B][1pR]\f[R] is executed as a macro, then the code \f[B]1pR\f[R] is executed, meaning that the \f[B]1\f[R] will be printed with a newline after and then popped from the stack. .TP \f[B][\f[R]\f[I]characters\f[R]\f[B]]\f[R] Makes a string containing \f[I]characters\f[R] and pushes it onto the stack. .RS .PP If there are brackets (\f[B][\f[R] and \f[B]]\f[R]) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (\f[B]\[rs]\f[R]) character. .PP If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. .RE .TP \f[B]a\f[R] The value on top of the stack is popped. .RS .PP If it is a number, it is truncated and its absolute value is taken. The result mod \f[B]256\f[R] is calculated. If that result is \f[B]0\f[R], push an empty string; otherwise, push a one\-character string where the character is the result of the mod interpreted as an ASCII character. .PP If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one\-character string. The new string is then pushed onto the stack. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]x\f[R] Pops a value off of the top of the stack. .RS .PP If it is a number, it is pushed back onto the stack. .PP If it is a string, it is executed as a macro. .PP This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. .RE .TP \f[B]>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP For example, \f[B]0 1>a\f[R] will execute the contents of register \f[B]a\f[R], and \f[B]1 0>a\f[R] will not. .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]?\f[R] Reads a line from the \f[B]stdin\f[R] and executes it. This is to allow macros to request input from users. .TP \f[B]q\f[R] During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. .TP \f[B]Q\f[R] Pops a value from the stack which must be non\-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. .TP \f[B],\f[R] Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the \f[B]Q\f[R] command, so the sequence \f[B],Q\f[R] will make dc(1) exit. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Status These commands query status of the stack or its top value. .TP \f[B]Z\f[R] Pops a value off of the stack. .RS .PP If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push \f[B]1\f[R] if the argument is \f[B]0\f[R] with no decimal places. .PP If it is a string, pushes the number of characters the string has. .RE .TP \f[B]X\f[R] Pops a value off of the stack. .RS .PP If it is a number, pushes the \f[I]scale\f[R] of the value onto the stack. .PP If it is a string, pushes \f[B]0\f[R]. .RE .TP \f[B]u\f[R] Pops one value off of the stack. If the value is a number, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a string), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]t\f[R] Pops one value off of the stack. If the value is a string, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a number), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]z\f[R] Pushes the current depth of the stack (before execution of this command) onto the stack. .TP \f[B]y\f[R]\f[I]r\f[R] Pushes the current stack depth of the register \f[I]r\f[R] onto the main stack. .RS .PP Because each register has a depth of \f[B]1\f[R] (with the value \f[B]0\f[R] in the top item) when dc(1) starts, dc(1) requires that each register\[cq]s stack must always have at least one item; dc(1) will give an error and reset otherwise (see the \f[B]RESET\f[R] section). This means that this command will never push \f[B]0\f[R]. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Arrays These commands manipulate arrays. .TP \f[B]:\f[R]\f[I]r\f[R] Pops the top two values off of the stack. The second value will be stored in the array \f[I]r\f[R] (see the \f[B]REGISTERS\f[R] section), indexed by the first value. .TP \f[B];\f[R]\f[I]r\f[R] Pops the value on top of the stack and uses it as an index into the array \f[I]r\f[R]. The selected value is then pushed onto the stack. .TP \f[B]Y\f[R]\f[I]r\f[R] Pushes the length of the array \f[I]r\f[R] onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter \f[B]g\f[R]. Only the characters below are allowed after the character \f[B]g\f[R]; any other character produces a parse error (see the \f[B]ERRORS\f[R] section). .TP \f[B]gl\f[R] Pushes the line length set by \f[B]DC_LINE_LENGTH\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) onto the stack. .TP \f[B]gx\f[R] Pushes \f[B]1\f[R] onto the stack if extended register mode is on, \f[B]0\f[R] otherwise. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .TP \f[B]gz\f[R] Pushes \f[B]0\f[R] onto the stack if the leading zero setting has not been enabled with the \f[B]\-z\f[R] or \f[B]\-\-leading\-zeroes\f[R] options (see the \f[B]OPTIONS\f[R] section), non\-zero otherwise. .SH REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) .PP Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (\f[B]0\f[R]) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. .PP In non\-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (\f[B]`\[rs]n'\f[R]) and a left bracket (\f[B]`['\f[R]); it is a parse error for a newline or a left bracket to be used as a register name. .SS Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. .PP If extended register mode is enabled (\f[B]\-x\f[R] or \f[B]\-\-extended\-register\f[R] command\-line arguments are given), then normal single character registers are used \f[I]unless\f[R] the character immediately following a command that needs a register name is a space (according to \f[B]isspace()\f[R]) and not a newline (\f[B]`\[rs]n'\f[R]). .PP In that case, the register name is found according to the regex \f[B][a\-z][a\-z0\-9_]*\f[R] (like bc(1) identifiers), and it is a parse error if the next non\-space characters do not match that regex. .SH RESET When dc(1) encounters an error or a signal that it has a non\-default handler for, it resets. This means that several things happen. .PP First, any macros that are executing are stopped and popped off the -stack. +execution stack. The behavior is not unlike that of exceptions in programming languages. Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. .PP +However, the stack of values is \f[I]not\f[R] cleared; in interactive +mode, users can inspect the stack and manipulate it. +.PP Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the \f[B]EXIT STATUS\f[R] section), it asks for more input; otherwise, it exits with the appropriate return code. .SH PERFORMANCE Most dc(1) implementations use \f[B]char\f[R] types to calculate the value of \f[B]1\f[R] decimal digit at a time, but that can be slow. This dc(1) does something different. .PP It uses large integers to calculate more than \f[B]1\f[R] decimal digit at a time. If built in a environment where \f[B]DC_LONG_BIT\f[R] (see the \f[B]LIMITS\f[R] section) is \f[B]64\f[R], then each integer has \f[B]9\f[R] decimal digits. If built in an environment where \f[B]DC_LONG_BIT\f[R] is \f[B]32\f[R] then each integer has \f[B]4\f[R] decimal digits. This value (the number of decimal digits per large integer) is called \f[B]DC_BASE_DIGS\f[R]. .PP In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of \f[B]DC_LONG_BIT\f[R], but is always at least twice as large as the integer type used to store digits. .SH LIMITS The following are the limits on dc(1): .TP \f[B]DC_LONG_BIT\f[R] The number of bits in the \f[B]long\f[R] type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the \f[B]PERFORMANCE\f[R] section). .TP \f[B]DC_BASE_DIGS\f[R] The number of decimal digits per large integer (see the \f[B]PERFORMANCE\f[R] section). Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_POW\f[R] The max decimal number that each large integer can store (see \f[B]DC_BASE_DIGS\f[R]) plus \f[B]1\f[R]. Depends on \f[B]DC_BASE_DIGS\f[R]. .TP \f[B]DC_OVERFLOW_MAX\f[R] The max number that the overflow type (see the \f[B]PERFORMANCE\f[R] section) can hold. Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_MAX\f[R] The maximum output base. Set at \f[B]DC_BASE_POW\f[R]. .TP \f[B]DC_DIM_MAX\f[R] The maximum size of arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .TP \f[B]DC_SCALE_MAX\f[R] The maximum \f[B]scale\f[R]. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_STRING_MAX\f[R] The maximum length of strings. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NAME_MAX\f[R] The maximum length of identifiers. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NUM_MAX\f[R] The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP Exponent The maximum allowable exponent (positive or negative). Set at \f[B]DC_OVERFLOW_MAX\f[R]. .TP Number of vars The maximum number of vars/arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .PP These limits are meant to be effectively non\-existent; the limits are so large (at least on 64\-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. .SH ENVIRONMENT VARIABLES As \f[B]non\-portable extensions\f[R], dc(1) recognizes the following environment variables: .TP \f[B]DC_ENV_ARGS\f[R] This is another way to give command\-line arguments to dc(1). They should be in the same format as all other command\-line arguments. These are always processed first, so any files given in \f[B]DC_ENV_ARGS\f[R] will be processed before arguments and files given on the command\-line. This gives the user the ability to set up \[lq]standard\[rq] options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the \f[B]\-e\f[R] option to set \f[B]scale\f[R] to a value other than \f[B]0\f[R]. .RS .PP The code that parses \f[B]DC_ENV_ARGS\f[R] will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string \f[B]\[lq]/home/gavin/some dc file.dc\[rq]\f[R] will be correctly parsed, but the string \f[B]\[lq]/home/gavin/some \[dq]dc\[dq] file.dc\[rq]\f[R] will include the backslashes. .PP The quote parsing will handle either kind of quotes, \f[B]\[cq]\f[R] or \f[B]\[lq]\f[R]. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in \f[B]\[lq]some `dc' file.dc\[rq]\f[R], and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in \f[B]DC_ENV_ARGS\f[R] is not supported due to the complexity of the parsing, though such files are still supported on the command\-line where the parsing is done by the shell. .RE .TP \f[B]DC_LINE_LENGTH\f[R] If this environment variable exists and contains an integer that is greater than \f[B]1\f[R] and is less than \f[B]UINT16_MAX\f[R] (\f[B]2\[ha]16\-1\f[R]), dc(1) will output lines to that length, including the backslash newline combo. The default line length is \f[B]70\f[R]. .RS .PP The special value of \f[B]0\f[R] will disable line length checking and print numbers without regard to line length and without backslashes and newlines. .RE .TP \f[B]DC_SIGINT_RESET\f[R] If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), then this environment variable has no effect because dc(1) exits on \f[B]SIGINT\f[R] when not in interactive mode. .RS .PP However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) reset on \f[B]SIGINT\f[R], rather than exit, and zero makes dc(1) exit. If this environment variable exists and is \f[I]not\f[R] an integer, then dc(1) will exit on \f[B]SIGINT\f[R]. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_TTY_MODE\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non\-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_PROMPT\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) use a prompt, and zero or a non\-integer makes dc(1) not use a prompt. If this environment variable does not exist and \f[B]DC_TTY_MODE\f[R] does, then the value of the \f[B]DC_TTY_MODE\f[R] environment variable is used. .PP This environment variable and the \f[B]DC_TTY_MODE\f[R] environment variable override the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_EXPR_EXIT\f[R] If any expressions or expression files are given on the command\-line with \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R], then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. .RS .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_DIGIT_CLAMP\f[R] When parsing numbers and if this environment variable exists and contains an integer, a non\-zero value makes dc(1) clamp digits that are greater than or equal to the current \f[B]ibase\f[R] so that all such digits are considered equal to the \f[B]ibase\f[R] minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the \f[B]ibase\f[R]. .RS .PP This never applies to single\-digit numbers, as per the bc(1) standard (see the \f[B]STANDARDS\f[R] section). .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .SH EXIT STATUS dc(1) returns the following exit statuses: .TP \f[B]0\f[R] No error. .TP \f[B]1\f[R] A math error occurred. This follows standard practice of using \f[B]1\f[R] for expected errors, since math errors will happen in the process of normal execution. .RS .PP Math errors include divide by \f[B]0\f[R], taking the square root of a negative number, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non\-integer where an integer is required. .PP Converting to a hardware integer happens for the second operand of the power (\f[B]\[ha]\f[R]) operator. .RE .TP \f[B]2\f[R] A parse error occurred. .RS .PP Parse errors include unexpected \f[B]EOF\f[R], using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. .RE .TP \f[B]3\f[R] A runtime error occurred. .RS .PP Runtime errors include assigning an invalid number to any global (\f[B]ibase\f[R], \f[B]obase\f[R], or \f[B]scale\f[R]), giving a bad expression to a \f[B]read()\f[R] call, calling \f[B]read()\f[R] inside of a \f[B]read()\f[R] call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. .RE .TP \f[B]4\f[R] A fatal error occurred. .RS .PP Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command\-line options. .RE .PP The exit status \f[B]4\f[R] is special; when a fatal error occurs, dc(1) always exits and returns \f[B]4\f[R], no matter what mode dc(1) is in. .PP The other statuses will only be returned when dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), since dc(1) resets its state (see the \f[B]RESET\f[R] section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .PP These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .SH INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non\-interactive mode. Interactive mode is turned on automatically when both \f[B]stdin\f[R] and \f[B]stdout\f[R] are hooked to a terminal, but the \f[B]\-i\f[R] flag and \f[B]\-\-interactive\f[R] option can turn it on in other situations. .PP In interactive mode, dc(1) attempts to recover from errors (see the \f[B]RESET\f[R] section), and in normal execution, flushes \f[B]stdout\f[R] as soon as execution is done for the current input. dc(1) may also reset on \f[B]SIGINT\f[R] instead of exit, depending on the contents of, or default for, the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .SH TTY MODE If \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY, then \[lq]TTY mode\[rq] is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. .PP If there is the environment variable \f[B]DC_TTY_MODE\f[R] in the environment (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then if that environment variable contains a non\-zero integer, dc(1) will turn on TTY mode when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY. If the \f[B]DC_TTY_MODE\f[R] environment variable exists but is \f[I]not\f[R] a non\-zero integer, then dc(1) will not turn TTY mode on. .PP If the environment variable \f[B]DC_TTY_MODE\f[R] does \f[I]not\f[R] exist, the default setting is used. The default setting can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the \f[B]STANDARDS\f[R] section), and interactive mode requires only \f[B]stdin\f[R] and \f[B]stdout\f[R] to be connected to a terminal. .SS Command\-Line History Command\-line history is only enabled if TTY mode is, i.e., that \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are connected to a TTY and the \f[B]DC_TTY_MODE\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and its default do not disable TTY mode. See the \f[B]COMMAND LINE HISTORY\f[R] section for more information. .SS Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: \f[B]DC_PROMPT\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP If the environment variable \f[B]DC_PROMPT\f[R] exists and is a non\-zero integer, then the prompt is turned on when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are connected to a TTY and the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options were not used. The read prompt will be turned on under the same conditions, except that the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options must also not be used. .PP However, if \f[B]DC_PROMPT\f[R] does not exist, the prompt can be enabled or disabled with the \f[B]DC_TTY_MODE\f[R] environment variable, the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options, and the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options. See the \f[B]ENVIRONMENT VARIABLES\f[R] and \f[B]OPTIONS\f[R] sections for more details. .SH SIGNAL HANDLING Sending a \f[B]SIGINT\f[R] will cause dc(1) to do one of two things. .PP If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), or the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or its default, is either not an integer or it is zero, dc(1) will exit. .PP However, if dc(1) is in interactive mode, and the \f[B]DC_SIGINT_RESET\f[R] or its default is an integer and non\-zero, then dc(1) will stop executing the current input and reset (see the \f[B]RESET\f[R] section) upon receiving a \f[B]SIGINT\f[R]. .PP Note that \[lq]current input\[rq] can mean one of two things. If dc(1) is processing input from \f[B]stdin\f[R] in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from \f[B]stdin\f[R] if no other file exists. .PP This means that if a \f[B]SIGINT\f[R] is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. .PP \f[B]SIGTERM\f[R] and \f[B]SIGQUIT\f[R] cause dc(1) to clean up and exit, and it uses the default handler for all other signals. The one exception is \f[B]SIGHUP\f[R]; in that case, and only when dc(1) is in TTY mode (see the \f[B]TTY MODE\f[R] section), a \f[B]SIGHUP\f[R] will cause dc(1) to clean up and exit. .SH COMMAND LINE HISTORY dc(1) supports interactive command\-line editing. .PP If dc(1) can be in TTY mode (see the \f[B]TTY MODE\f[R] section), history can be enabled. This means that command\-line history can only be enabled when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY. .PP Like TTY mode itself, it can be turned on or off with the environment variable \f[B]DC_TTY_MODE\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP \f[B]Note\f[R]: tabs are converted to 8 spaces. .SH SEE ALSO bc(1) .SH STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1\-2017 (\[lq]POSIX.1\-2017\[rq]) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . .SH BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . .SH AUTHOR Gavin D. Howard \c .MT gavin@gavinhoward.com .ME \c \ and contributors. diff --git a/manuals/dc/EN.1.md b/manuals/dc/EN.1.md index 64c945be8857..ab9647a196be 100644 --- a/manuals/dc/EN.1.md +++ b/manuals/dc/EN.1.md @@ -1,1345 +1,1348 @@ # Name dc - arbitrary-precision decimal reverse-Polish notation calculator # SYNOPSIS **dc** [**-cChiPRvVx**] [**-\-version**] [**-\-help**] [**-\-digit-clamp**] [**-\-no-digit-clamp**] [**-\-interactive**] [**-\-no-prompt**] [**-\-no-read-prompt**] [**-\-extended-register**] [**-e** *expr*] [**-\-expression**=*expr*...] [**-f** *file*...] [**-\-file**=*file*...] [*file*...] # DESCRIPTION dc(1) is an arbitrary-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. If no files are given on the command-line, then dc(1) reads from **stdin** (see the **STDIN** section). Otherwise, those files are processed, and dc(1) will then exit. If a user wants to set up a standard environment, they can use **DC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). For example, if a user wants the **scale** always set to **10**, they can set **DC_ENV_ARGS** to **-e 10k**, and this dc(1) will always start with a **scale** of **10**. # OPTIONS The following are the options that dc(1) accepts. **-C**, **-\-no-digit-clamp** : Disables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that the value added to a number from a digit is always that digit's value multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-c** or **-\-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-c**, **-\-digit-clamp** : Enables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-C** or **-\-no-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-e** *expr*, **-\-expression**=*expr* : Evaluates *expr*. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**, whether on the command-line or in **DC_ENV_ARGS**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-f** *file*, **-\-file**=*file* : Reads in *file* and evaluates it, line by line, as though it were read through **stdin**. If expressions are also given (see above), the expressions are evaluated in the order given. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-h**, **-\-help** : Prints a usage message and exits. **-I** *ibase*, **-\-ibase**=*ibase* : Sets the builtin variable **ibase** to the value *ibase* assuming that *ibase* is in base 10. It is a fatal error if *ibase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-i**, **-\-interactive** : Forces interactive mode. (See the **INTERACTIVE MODE** section.) This is a **non-portable extension**. **-L**, **-\-no-line-length** : Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets **BC_LINE_LENGTH** to **0** (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-O** *obase*, **-\-obase**=*obase* : Sets the builtin variable **obase** to the value *obase* assuming that *obase* is in base 10. It is a fatal error if *obase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-P**, **-\-no-prompt** : Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in **DC_ENV_ARGS**. These options override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-R**, **-\-no-read-prompt** : Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in **BC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. This option does not disable the regular prompt because the read prompt is only used when the **?** command is used. These options *do* override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section), but only for the read prompt. This is a **non-portable extension**. **-S** *scale*, **-\-scale**=*scale* : Sets the builtin variable **scale** to the value *scale* assuming that *scale* is in base 10. It is a fatal error if *scale* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-v**, **-V**, **-\-version** : Print the version information (copyright header) and exits. **-x** **-\-extended-register** : Enables extended register mode. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. This is a **non-portable extension**. **-z**, **-\-leading-zeroes** : Makes dc(1) print all numbers greater than **-1** and less than **1**, and not equal to **0**, with a leading zero. This is a **non-portable extension**. All long options are **non-portable extensions**. # STDIN If no files are given on the command-line and no files or expressions are given by the **-f**, **-\-file**, **-e**, or **-\-expression** options, then dc(1) reads from **stdin**. However, there is a caveat to this. First, **stdin** is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. # STDOUT Any non-error output is written to **stdout**. In addition, if history (see the **HISTORY** section) and the prompt (see the **TTY MODE** section) are enabled, both are output to **stdout**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stdout**, so if **stdout** is closed, as in **dc >&-**, it will quit with an error. This is done so that dc(1) can report problems when **stdout** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stdout** to **/dev/null**. # STDERR Any error output is written to **stderr**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stderr**, so if **stderr** is closed, as in **dc 2>&-**, it will quit with an error. This is done so that dc(1) can exit with an error code when **stderr** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stderr** to **/dev/null**. # SYNTAX Each item in the input source code, either a number (see the **NUMBERS** section) or a command (see the **COMMANDS** section), is processed and executed, in order. Input is processed immediately when entered. **ibase** is a register (see the **REGISTERS** section) that determines how to interpret constant numbers. It is the "input" base, or the number base used for interpreting input numbers. **ibase** is initially **10**. The max allowable value for **ibase** is **16**. The min allowable value for **ibase** is **2**. The max allowable value for **ibase** can be queried in dc(1) programs with the **T** command. **obase** is a register (see the **REGISTERS** section) that determines how to output results. It is the "output" base, or the number base used for outputting numbers. **obase** is initially **10**. The max allowable value for **obase** is **DC_BASE_MAX** and can be queried with the **U** command. The min allowable value for **obase** is **2**. Values are output in the specified base. The *scale* of an expression is the number of digits in the result of the expression right of the decimal point, and **scale** is a register (see the **REGISTERS** section) that sets the precision of any operations (with exceptions). **scale** is initially **0**. **scale** cannot be negative. The max allowable value for **scale** can be queried in dc(1) programs with the **V** command. ## Comments Comments go from **#** until, and not including, the next newline. This is a **non-portable extension**. # NUMBERS Numbers are strings made up of digits, uppercase letters up to **F**, and at most **1** period for a radix. Numbers can have up to **DC_NUM_MAX** digits. Uppercase letters are equal to **9** plus their position in the alphabet (i.e., **A** equals **10**, or **9+1**). If a digit or letter makes no sense with the current value of **ibase** (i.e., they are greater than or equal to the current value of **ibase**), then the behavior depends on the existence of the **-c**/**-\-digit-clamp** or **-C**/**-\-no-digit-clamp** options (see the **OPTIONS** section), the existence and setting of the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section), or the default, which can be queried with the **-h**/**-\-help** option. If clamping is off, then digits or letters that are greater than or equal to the current value of **ibase** are not changed. Instead, their given value is multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*A+3\^0\*B**, which is **3** times **10** plus **11**, or **41**. If clamping is on, then digits or letters that are greater than or equal to the current value of **ibase** are set to the value of the highest valid digit in **ibase** before being multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*2+3\^0\*2**, which is **3** times **2** plus **2**, or **8**. There is one exception to clamping: single-character numbers (i.e., **A** alone). Such numbers are never clamped and always take the value they would have in the highest possible **ibase**. This means that **A** alone always equals decimal **10** and **Z** alone always equals decimal **35**. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current **ibase** (with the **i** command) regardless of the current value of **ibase**. If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for **A**, use **0A**. # COMMANDS The valid commands are listed below. ## Printing These commands are used for printing. **p** : Prints the value on top of the stack, whether number or string, and prints a newline after. This does not alter the stack. **n** : Prints the value on top of the stack, whether number or string, and pops it off of the stack. **P** : Pops a value off the stack. If the value is a number, it is truncated and the absolute value of the result is printed as though **obase** is **256** and each digit is interpreted as an 8-bit ASCII character, making it a byte stream. If the value is a string, it is printed without a trailing newline. This is a **non-portable extension**. **f** : Prints the entire contents of the stack, in order from newest to oldest, without altering anything. Users should use this command when they get lost. ## Arithmetic These are the commands used for arithmetic. **+** : The top two values are popped off the stack, added, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **-** : The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **\*** : The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If **a** is the *scale* of the first expression and **b** is the *scale* of the second expression, the *scale* of the result is equal to **min(a+b,max(scale,a,b))** where **min()** and **max()** return the obvious values. **/** : The top two values are popped off the stack, divided, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be non-zero. **%** : The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. Remaindering is equivalent to 1) Computing **a/b** to current **scale**, and 2) Using the result of step 1 to calculate **a-(a/b)\*b** to *scale* **max(scale+scale(b),scale(a))**. The first value popped off of the stack must be non-zero. **~** : The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to **x y / x y %** except that **x** and **y** are only evaluated once. The first value popped off of the stack must be non-zero. This is a **non-portable extension**. **\^** : The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non-zero. **v** : The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The value popped off of the stack must be non-negative. **\_** : If this command *immediately* precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a **non-portable extension**. **b** : The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. This is a **non-portable extension**. **|** : The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. The first value popped is used as the reduction modulus and must be an integer and non-zero. The second value popped is used as the exponent and must be an integer and non-negative. The third value popped is the base and must be an integer. This is a **non-portable extension**. **G** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if they are equal, or **0** otherwise. This is a **non-portable extension**. **N** : The top value is popped off of the stack, and if it a **0**, a **1** is pushed; otherwise, a **0** is pushed. This is a **non-portable extension**. **(** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than the second, or **0** otherwise. This is a **non-portable extension**. **{** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **)** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than the second, or **0** otherwise. This is a **non-portable extension**. **}** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **M** : The top two values are popped off of the stack. If they are both non-zero, a **1** is pushed onto the stack. If either of them is zero, or both of them are, then a **0** is pushed onto the stack. This is like the **&&** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. **m** : The top two values are popped off of the stack. If at least one of them is non-zero, a **1** is pushed onto the stack. If both of them are zero, then a **0** is pushed onto the stack. This is like the **||** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. ## Stack Control These commands control the stack. **c** : Removes all items from ("clears") the stack. **d** : Copies the item on top of the stack ("duplicates") and pushes the copy onto the stack. **r** : Swaps ("reverses") the two top items on the stack. **R** : Pops ("removes") the top value from the stack. ## Register Control These commands control registers (see the **REGISTERS** section). **s**_r_ : Pops the value off the top of the stack and stores it into register *r*. **l**_r_ : Copies the value in register *r* and pushes it onto the stack. This does not alter the contents of *r*. **S**_r_ : Pops the value off the top of the (main) stack and pushes it onto the stack of register *r*. The previous value of the register becomes inaccessible. **L**_r_ : Pops the value off the top of the stack for register *r* and push it onto the main stack. The previous value in the stack for register *r*, if any, is now accessible via the **l**_r_ command. ## Parameters These commands control the values of **ibase**, **obase**, and **scale**. Also see the **SYNTAX** section. **i** : Pops the value off of the top of the stack and uses it to set **ibase**, which must be between **2** and **16**, inclusive. If the value on top of the stack has any *scale*, the *scale* is ignored. **o** : Pops the value off of the top of the stack and uses it to set **obase**, which must be between **2** and **DC_BASE_MAX**, inclusive (see the **LIMITS** section). If the value on top of the stack has any *scale*, the *scale* is ignored. **k** : Pops the value off of the top of the stack and uses it to set **scale**, which must be non-negative. If the value on top of the stack has any *scale*, the *scale* is ignored. **I** : Pushes the current value of **ibase** onto the main stack. **O** : Pushes the current value of **obase** onto the main stack. **K** : Pushes the current value of **scale** onto the main stack. **T** : Pushes the maximum allowable value of **ibase** onto the main stack. This is a **non-portable extension**. **U** : Pushes the maximum allowable value of **obase** onto the main stack. This is a **non-portable extension**. **V** : Pushes the maximum allowable value of **scale** onto the main stack. This is a **non-portable extension**. ## Strings The following commands control strings. dc(1) can work with both numbers and strings, and registers (see the **REGISTERS** section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. Strings can also be executed as macros. For example, if the string **[1pR]** is executed as a macro, then the code **1pR** is executed, meaning that the **1** will be printed with a newline after and then popped from the stack. **\[**_characters_**\]** : Makes a string containing *characters* and pushes it onto the stack. If there are brackets (**\[** and **\]**) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (**\\**) character. If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. **a** : The value on top of the stack is popped. If it is a number, it is truncated and its absolute value is taken. The result mod **256** is calculated. If that result is **0**, push an empty string; otherwise, push a one-character string where the character is the result of the mod interpreted as an ASCII character. If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one-character string. The new string is then pushed onto the stack. This is a **non-portable extension**. **x** : Pops a value off of the top of the stack. If it is a number, it is pushed back onto the stack. If it is a string, it is executed as a macro. This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. **\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register *r* are executed. For example, **0 1>a** will execute the contents of register **a**, and **1 0>a** will not. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **?** : Reads a line from the **stdin** and executes it. This is to allow macros to request input from users. **q** : During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. **Q** : Pops a value from the stack which must be non-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. **,** : Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the **Q** command, so the sequence **,Q** will make dc(1) exit. This is a **non-portable extension**. ## Status These commands query status of the stack or its top value. **Z** : Pops a value off of the stack. If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push **1** if the argument is **0** with no decimal places. If it is a string, pushes the number of characters the string has. **X** : Pops a value off of the stack. If it is a number, pushes the *scale* of the value onto the stack. If it is a string, pushes **0**. **u** : Pops one value off of the stack. If the value is a number, this pushes **1** onto the stack. Otherwise (if it is a string), it pushes **0**. This is a **non-portable extension**. **t** : Pops one value off of the stack. If the value is a string, this pushes **1** onto the stack. Otherwise (if it is a number), it pushes **0**. This is a **non-portable extension**. **z** : Pushes the current depth of the stack (before execution of this command) onto the stack. **y**_r_ : Pushes the current stack depth of the register *r* onto the main stack. Because each register has a depth of **1** (with the value **0** in the top item) when dc(1) starts, dc(1) requires that each register's stack must always have at least one item; dc(1) will give an error and reset otherwise (see the **RESET** section). This means that this command will never push **0**. This is a **non-portable extension**. ## Arrays These commands manipulate arrays. **:**_r_ : Pops the top two values off of the stack. The second value will be stored in the array *r* (see the **REGISTERS** section), indexed by the first value. **;**_r_ : Pops the value on top of the stack and uses it as an index into the array *r*. The selected value is then pushed onto the stack. **Y**_r_ : Pushes the length of the array *r* onto the stack. This is a **non-portable extension**. ## Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter **g**. Only the characters below are allowed after the character **g**; any other character produces a parse error (see the **ERRORS** section). **gl** : Pushes the line length set by **DC_LINE_LENGTH** (see the **ENVIRONMENT VARIABLES** section) onto the stack. **gx** : Pushes **1** onto the stack if extended register mode is on, **0** otherwise. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. **gz** : Pushes **0** onto the stack if the leading zero setting has not been enabled with the **-z** or **-\-leading-zeroes** options (see the **OPTIONS** section), non-zero otherwise. # REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (**0**) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. In non-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (**'\\n'**) and a left bracket (**'['**); it is a parse error for a newline or a left bracket to be used as a register name. ## Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. If extended register mode is enabled (**-x** or **-\-extended-register** command-line arguments are given), then normal single character registers are used *unless* the character immediately following a command that needs a register name is a space (according to **isspace()**) and not a newline (**'\\n'**). In that case, the register name is found according to the regex **\[a-z\]\[a-z0-9\_\]\*** (like bc(1) identifiers), and it is a parse error if the next non-space characters do not match that regex. # RESET When dc(1) encounters an error or a signal that it has a non-default handler for, it resets. This means that several things happen. -First, any macros that are executing are stopped and popped off the stack. -The behavior is not unlike that of exceptions in programming languages. Then -the execution point is set so that any code waiting to execute (after all +First, any macros that are executing are stopped and popped off the execution +stack. The behavior is not unlike that of exceptions in programming languages. +Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. +However, the stack of values is *not* cleared; in interactive mode, users can +inspect the stack and manipulate it. + Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the **EXIT STATUS** section), it asks for more input; otherwise, it exits with the appropriate return code. # PERFORMANCE Most dc(1) implementations use **char** types to calculate the value of **1** decimal digit at a time, but that can be slow. This dc(1) does something different. It uses large integers to calculate more than **1** decimal digit at a time. If built in a environment where **DC_LONG_BIT** (see the **LIMITS** section) is **64**, then each integer has **9** decimal digits. If built in an environment where **DC_LONG_BIT** is **32** then each integer has **4** decimal digits. This value (the number of decimal digits per large integer) is called **DC_BASE_DIGS**. In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of **DC_LONG_BIT**, but is always at least twice as large as the integer type used to store digits. # LIMITS The following are the limits on dc(1): **DC_LONG_BIT** : The number of bits in the **long** type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the **PERFORMANCE** section). **DC_BASE_DIGS** : The number of decimal digits per large integer (see the **PERFORMANCE** section). Depends on **DC_LONG_BIT**. **DC_BASE_POW** : The max decimal number that each large integer can store (see **DC_BASE_DIGS**) plus **1**. Depends on **DC_BASE_DIGS**. **DC_OVERFLOW_MAX** : The max number that the overflow type (see the **PERFORMANCE** section) can hold. Depends on **DC_LONG_BIT**. **DC_BASE_MAX** : The maximum output base. Set at **DC_BASE_POW**. **DC_DIM_MAX** : The maximum size of arrays. Set at **SIZE_MAX-1**. **DC_SCALE_MAX** : The maximum **scale**. Set at **DC_OVERFLOW_MAX-1**. **DC_STRING_MAX** : The maximum length of strings. Set at **DC_OVERFLOW_MAX-1**. **DC_NAME_MAX** : The maximum length of identifiers. Set at **DC_OVERFLOW_MAX-1**. **DC_NUM_MAX** : The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at **DC_OVERFLOW_MAX-1**. Exponent : The maximum allowable exponent (positive or negative). Set at **DC_OVERFLOW_MAX**. Number of vars : The maximum number of vars/arrays. Set at **SIZE_MAX-1**. These limits are meant to be effectively non-existent; the limits are so large (at least on 64-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. # ENVIRONMENT VARIABLES As **non-portable extensions**, dc(1) recognizes the following environment variables: **DC_ENV_ARGS** : This is another way to give command-line arguments to dc(1). They should be in the same format as all other command-line arguments. These are always processed first, so any files given in **DC_ENV_ARGS** will be processed before arguments and files given on the command-line. This gives the user the ability to set up "standard" options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the **-e** option to set **scale** to a value other than **0**. The code that parses **DC_ENV_ARGS** will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string **"/home/gavin/some dc file.dc"** will be correctly parsed, but the string **"/home/gavin/some \"dc\" file.dc"** will include the backslashes. The quote parsing will handle either kind of quotes, **'** or **"**. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in **"some 'dc' file.dc"**, and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in **DC_ENV_ARGS** is not supported due to the complexity of the parsing, though such files are still supported on the command-line where the parsing is done by the shell. **DC_LINE_LENGTH** : If this environment variable exists and contains an integer that is greater than **1** and is less than **UINT16_MAX** (**2\^16-1**), dc(1) will output lines to that length, including the backslash newline combo. The default line length is **70**. The special value of **0** will disable line length checking and print numbers without regard to line length and without backslashes and newlines. **DC_SIGINT_RESET** : If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), then this environment variable has no effect because dc(1) exits on **SIGINT** when not in interactive mode. However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) reset on **SIGINT**, rather than exit, and zero makes dc(1) exit. If this environment variable exists and is *not* an integer, then dc(1) will exit on **SIGINT**. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_TTY_MODE** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_PROMPT** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) use a prompt, and zero or a non-integer makes dc(1) not use a prompt. If this environment variable does not exist and **DC_TTY_MODE** does, then the value of the **DC_TTY_MODE** environment variable is used. This environment variable and the **DC_TTY_MODE** environment variable override the default, which can be queried with the **-h** or **-\-help** options. **DC_EXPR_EXIT** : If any expressions or expression files are given on the command-line with **-e**, **-\-expression**, **-f**, or **-\-file**, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_DIGIT_CLAMP** : When parsing numbers and if this environment variable exists and contains an integer, a non-zero value makes dc(1) clamp digits that are greater than or equal to the current **ibase** so that all such digits are considered equal to the **ibase** minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the **ibase**. This never applies to single-digit numbers, as per the bc(1) standard (see the **STANDARDS** section). This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. # EXIT STATUS dc(1) returns the following exit statuses: **0** : No error. **1** : A math error occurred. This follows standard practice of using **1** for expected errors, since math errors will happen in the process of normal execution. Math errors include divide by **0**, taking the square root of a negative number, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non-integer where an integer is required. Converting to a hardware integer happens for the second operand of the power (**\^**) operator. **2** : A parse error occurred. Parse errors include unexpected **EOF**, using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. **3** : A runtime error occurred. Runtime errors include assigning an invalid number to any global (**ibase**, **obase**, or **scale**), giving a bad expression to a **read()** call, calling **read()** inside of a **read()** call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. **4** : A fatal error occurred. Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command-line options. The exit status **4** is special; when a fatal error occurs, dc(1) always exits and returns **4**, no matter what mode dc(1) is in. The other statuses will only be returned when dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), since dc(1) resets its state (see the **RESET** section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the **-i** flag or **-\-interactive** option. These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the **-i** flag or **-\-interactive** option. # INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non-interactive mode. Interactive mode is turned on automatically when both **stdin** and **stdout** are hooked to a terminal, but the **-i** flag and **-\-interactive** option can turn it on in other situations. In interactive mode, dc(1) attempts to recover from errors (see the **RESET** section), and in normal execution, flushes **stdout** as soon as execution is done for the current input. dc(1) may also reset on **SIGINT** instead of exit, depending on the contents of, or default for, the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section). # TTY MODE If **stdin**, **stdout**, and **stderr** are all connected to a TTY, then "TTY mode" is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. If there is the environment variable **DC_TTY_MODE** in the environment (see the **ENVIRONMENT VARIABLES** section), then if that environment variable contains a non-zero integer, dc(1) will turn on TTY mode when **stdin**, **stdout**, and **stderr** are all connected to a TTY. If the **DC_TTY_MODE** environment variable exists but is *not* a non-zero integer, then dc(1) will not turn TTY mode on. If the environment variable **DC_TTY_MODE** does *not* exist, the default setting is used. The default setting can be queried with the **-h** or **-\-help** options. TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the **STANDARDS** section), and interactive mode requires only **stdin** and **stdout** to be connected to a terminal. ## Command-Line History Command-line history is only enabled if TTY mode is, i.e., that **stdin**, **stdout**, and **stderr** are connected to a TTY and the **DC_TTY_MODE** environment variable (see the **ENVIRONMENT VARIABLES** section) and its default do not disable TTY mode. See the **COMMAND LINE HISTORY** section for more information. ## Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: **DC_PROMPT** (see the **ENVIRONMENT VARIABLES** section). If the environment variable **DC_PROMPT** exists and is a non-zero integer, then the prompt is turned on when **stdin**, **stdout**, and **stderr** are connected to a TTY and the **-P** and **-\-no-prompt** options were not used. The read prompt will be turned on under the same conditions, except that the **-R** and **-\-no-read-prompt** options must also not be used. However, if **DC_PROMPT** does not exist, the prompt can be enabled or disabled with the **DC_TTY_MODE** environment variable, the **-P** and **-\-no-prompt** options, and the **-R** and **-\-no-read-prompt** options. See the **ENVIRONMENT VARIABLES** and **OPTIONS** sections for more details. # SIGNAL HANDLING Sending a **SIGINT** will cause dc(1) to do one of two things. If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), or the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section), or its default, is either not an integer or it is zero, dc(1) will exit. However, if dc(1) is in interactive mode, and the **DC_SIGINT_RESET** or its default is an integer and non-zero, then dc(1) will stop executing the current input and reset (see the **RESET** section) upon receiving a **SIGINT**. Note that "current input" can mean one of two things. If dc(1) is processing input from **stdin** in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from **stdin** if no other file exists. This means that if a **SIGINT** is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. **SIGTERM** and **SIGQUIT** cause dc(1) to clean up and exit, and it uses the default handler for all other signals. The one exception is **SIGHUP**; in that case, and only when dc(1) is in TTY mode (see the **TTY MODE** section), a **SIGHUP** will cause dc(1) to clean up and exit. # COMMAND LINE HISTORY dc(1) supports interactive command-line editing. If dc(1) can be in TTY mode (see the **TTY MODE** section), history can be enabled. This means that command-line history can only be enabled when **stdin**, **stdout**, and **stderr** are all connected to a TTY. Like TTY mode itself, it can be turned on or off with the environment variable **DC_TTY_MODE** (see the **ENVIRONMENT VARIABLES** section). **Note**: tabs are converted to 8 spaces. # SEE ALSO bc(1) # STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1-2017 (“POSIX.1-2017”) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . # BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . # AUTHOR Gavin D. Howard and contributors. diff --git a/manuals/dc/H.1 b/manuals/dc/H.1 index 36f7458a3316..82c1bbd5c2b9 100644 --- a/manuals/dc/H.1 +++ b/manuals/dc/H.1 @@ -1,1668 +1,1671 @@ .\" .\" SPDX-License-Identifier: BSD-2-Clause .\" .\" Copyright (c) 2018-2024 Gavin D. Howard and contributors. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions are met: .\" .\" * Redistributions of source code must retain the above copyright notice, .\" this list of conditions and the following disclaimer. .\" .\" * Redistributions in binary form must reproduce the above copyright notice, .\" this list of conditions and the following disclaimer in the documentation .\" and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" .\" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE .\" LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF .\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS .\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN .\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) .\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE .\" POSSIBILITY OF SUCH DAMAGE. .\" -.TH "DC" "1" "January 2024" "Gavin D. Howard" "General Commands Manual" +.TH "DC" "1" "August 2024" "Gavin D. Howard" "General Commands Manual" .nh .ad l .SH Name dc \- arbitrary\-precision decimal reverse\-Polish notation calculator .SH SYNOPSIS \f[B]dc\f[R] [\f[B]\-cChiPRvVx\f[R]] [\f[B]\-\-version\f[R]] [\f[B]\-\-help\f[R]] [\f[B]\-\-digit\-clamp\f[R]] [\f[B]\-\-no\-digit\-clamp\f[R]] [\f[B]\-\-interactive\f[R]] [\f[B]\-\-no\-prompt\f[R]] [\f[B]\-\-no\-read\-prompt\f[R]] [\f[B]\-\-extended\-register\f[R]] [\f[B]\-e\f[R] \f[I]expr\f[R]] [\f[B]\-\-expression\f[R]=\f[I]expr\f[R]\&...] [\f[B]\-f\f[R] \f[I]file\f[R]\&...] [\f[B]\-\-file\f[R]=\f[I]file\f[R]\&...] [\f[I]file\f[R]\&...] [\f[B]\-I\f[R] \f[I]ibase\f[R]] [\f[B]\-\-ibase\f[R]=\f[I]ibase\f[R]] [\f[B]\-O\f[R] \f[I]obase\f[R]] [\f[B]\-\-obase\f[R]=\f[I]obase\f[R]] [\f[B]\-S\f[R] \f[I]scale\f[R]] [\f[B]\-\-scale\f[R]=\f[I]scale\f[R]] [\f[B]\-E\f[R] \f[I]seed\f[R]] [\f[B]\-\-seed\f[R]=\f[I]seed\f[R]] .SH DESCRIPTION dc(1) is an arbitrary\-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. .PP If no files are given on the command\-line, then dc(1) reads from \f[B]stdin\f[R] (see the \f[B]STDIN\f[R] section). Otherwise, those files are processed, and dc(1) will then exit. .PP If a user wants to set up a standard environment, they can use \f[B]DC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). For example, if a user wants the \f[B]scale\f[R] always set to \f[B]10\f[R], they can set \f[B]DC_ENV_ARGS\f[R] to \f[B]\-e 10k\f[R], and this dc(1) will always start with a \f[B]scale\f[R] of \f[B]10\f[R]. .SH OPTIONS The following are the options that dc(1) accepts. .TP \f[B]\-C\f[R], \f[B]\-\-no\-digit\-clamp\f[R] Disables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that the value added to a number from a digit is always that digit\[cq]s value multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-c\f[R] or \f[B]\-\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-c\f[R], \f[B]\-\-digit\-clamp\f[R] Enables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-C\f[R] or \f[B]\-\-no\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-E\f[R] \f[I]seed\f[R], \f[B]\-\-seed\f[R]=\f[I]seed\f[R] Sets the builtin variable \f[B]seed\f[R] to the value \f[I]seed\f[R] assuming that \f[I]seed\f[R] is in base 10. It is a fatal error if \f[I]seed\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-e\f[R] \f[I]expr\f[R], \f[B]\-\-expression\f[R]=\f[I]expr\f[R] Evaluates \f[I]expr\f[R]. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R], whether on the command\-line or in \f[B]DC_ENV_ARGS\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-f\f[R] \f[I]file\f[R], \f[B]\-\-file\f[R]=\f[I]file\f[R] Reads in \f[I]file\f[R] and evaluates it, line by line, as though it were read through \f[B]stdin\f[R]. If expressions are also given (see above), the expressions are evaluated in the order given. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-h\f[R], \f[B]\-\-help\f[R] Prints a usage message and exits. .TP \f[B]\-I\f[R] \f[I]ibase\f[R], \f[B]\-\-ibase\f[R]=\f[I]ibase\f[R] Sets the builtin variable \f[B]ibase\f[R] to the value \f[I]ibase\f[R] assuming that \f[I]ibase\f[R] is in base 10. It is a fatal error if \f[I]ibase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-i\f[R], \f[B]\-\-interactive\f[R] Forces interactive mode. (See the \f[B]INTERACTIVE MODE\f[R] section.) .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-L\f[R], \f[B]\-\-no\-line\-length\f[R] Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets \f[B]BC_LINE_LENGTH\f[R] to \f[B]0\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-O\f[R] \f[I]obase\f[R], \f[B]\-\-obase\f[R]=\f[I]obase\f[R] Sets the builtin variable \f[B]obase\f[R] to the value \f[I]obase\f[R] assuming that \f[I]obase\f[R] is in base 10. It is a fatal error if \f[I]obase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-P\f[R], \f[B]\-\-no\-prompt\f[R] Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]DC_ENV_ARGS\f[R]. .RS .PP These options override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-R\f[R], \f[B]\-\-no\-read\-prompt\f[R] Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]BC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. .RS .PP This option does not disable the regular prompt because the read prompt is only used when the \f[B]?\f[R] command is used. .PP These options \f[I]do\f[R] override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), but only for the read prompt. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-S\f[R] \f[I]scale\f[R], \f[B]\-\-scale\f[R]=\f[I]scale\f[R] Sets the builtin variable \f[B]scale\f[R] to the value \f[I]scale\f[R] assuming that \f[I]scale\f[R] is in base 10. It is a fatal error if \f[I]scale\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-v\f[R], \f[B]\-V\f[R], \f[B]\-\-version\f[R] Print the version information (copyright header) and exits. .TP \f[B]\-x\f[R] \f[B]\-\-extended\-register\f[R] Enables extended register mode. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-z\f[R], \f[B]\-\-leading\-zeroes\f[R] Makes dc(1) print all numbers greater than \f[B]\-1\f[R] and less than \f[B]1\f[R], and not equal to \f[B]0\f[R], with a leading zero. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .PP All long options are \f[B]non\-portable extensions\f[R]. .SH STDIN If no files are given on the command\-line and no files or expressions are given by the \f[B]\-f\f[R], \f[B]\-\-file\f[R], \f[B]\-e\f[R], or \f[B]\-\-expression\f[R] options, then dc(1) reads from \f[B]stdin\f[R]. .PP However, there is a caveat to this. .PP First, \f[B]stdin\f[R] is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. .SH STDOUT Any non\-error output is written to \f[B]stdout\f[R]. In addition, if history (see the \f[B]HISTORY\f[R] section) and the prompt (see the \f[B]TTY MODE\f[R] section) are enabled, both are output to \f[B]stdout\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stdout\f[R], so if \f[B]stdout\f[R] is closed, as in \f[B]dc >&\-\f[R], it will quit with an error. This is done so that dc(1) can report problems when \f[B]stdout\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stdout\f[R] to \f[B]/dev/null\f[R]. .SH STDERR Any error output is written to \f[B]stderr\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stderr\f[R], so if \f[B]stderr\f[R] is closed, as in \f[B]dc 2>&\-\f[R], it will quit with an error. This is done so that dc(1) can exit with an error code when \f[B]stderr\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stderr\f[R] to \f[B]/dev/null\f[R]. .SH SYNTAX Each item in the input source code, either a number (see the \f[B]NUMBERS\f[R] section) or a command (see the \f[B]COMMANDS\f[R] section), is processed and executed, in order. Input is processed immediately when entered. .PP \f[B]ibase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to interpret constant numbers. It is the \[lq]input\[rq] base, or the number base used for interpreting input numbers. \f[B]ibase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]ibase\f[R] is \f[B]16\f[R]. The min allowable value for \f[B]ibase\f[R] is \f[B]2\f[R]. The max allowable value for \f[B]ibase\f[R] can be queried in dc(1) programs with the \f[B]T\f[R] command. .PP \f[B]obase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to output results. It is the \[lq]output\[rq] base, or the number base used for outputting numbers. \f[B]obase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]obase\f[R] is \f[B]DC_BASE_MAX\f[R] and can be queried with the \f[B]U\f[R] command. The min allowable value for \f[B]obase\f[R] is \f[B]0\f[R]. If \f[B]obase\f[R] is \f[B]0\f[R], values are output in scientific notation, and if \f[B]obase\f[R] is \f[B]1\f[R], values are output in engineering notation. Otherwise, values are output in the specified base. .PP Outputting in scientific and engineering notations are \f[B]non\-portable extensions\f[R]. .PP The \f[I]scale\f[R] of an expression is the number of digits in the result of the expression right of the decimal point, and \f[B]scale\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that sets the precision of any operations (with exceptions). \f[B]scale\f[R] is initially \f[B]0\f[R]. \f[B]scale\f[R] cannot be negative. The max allowable value for \f[B]scale\f[R] can be queried in dc(1) programs with the \f[B]V\f[R] command. .PP \f[B]seed\f[R] is a register containing the current seed for the pseudo\-random number generator. If the current value of \f[B]seed\f[R] is queried and stored, then if it is assigned to \f[B]seed\f[R] later, the pseudo\-random number generator is guaranteed to produce the same sequence of pseudo\-random numbers that were generated after the value of \f[B]seed\f[R] was first queried. .PP Multiple values assigned to \f[B]seed\f[R] can produce the same sequence of pseudo\-random numbers. Likewise, when a value is assigned to \f[B]seed\f[R], it is not guaranteed that querying \f[B]seed\f[R] immediately after will return the same value. In addition, the value of \f[B]seed\f[R] will change after any call to the \f[B]\[cq]\f[R] command or the \f[B]\[lq]\f[R] command that does not get receive a value of \f[B]0\f[R] or \f[B]1\f[R]. The maximum integer returned by the \f[B]\[cq]\f[R] command can be queried with the \f[B]W\f[R] command. .PP \f[B]Note\f[R]: The values returned by the pseudo\-random number generator with the \f[B]\[cq]\f[R] and \f[B]\[lq]\f[R] commands are guaranteed to \f[B]NOT\f[R] be cryptographically secure. This is a consequence of using a seeded pseudo\-random number generator. However, they \f[I]are\f[R] guaranteed to be reproducible with identical \f[B]seed\f[R] values. This means that the pseudo\-random values from dc(1) should only be used where a reproducible stream of pseudo\-random numbers is \f[I]ESSENTIAL\f[R]. In any other case, use a non\-seeded pseudo\-random number generator. .PP The pseudo\-random number generator, \f[B]seed\f[R], and all associated operations are \f[B]non\-portable extensions\f[R]. .SS Comments Comments go from \f[B]#\f[R] until, and not including, the next newline. This is a \f[B]non\-portable extension\f[R]. .SH NUMBERS Numbers are strings made up of digits, uppercase letters up to \f[B]F\f[R], and at most \f[B]1\f[R] period for a radix. Numbers can have up to \f[B]DC_NUM_MAX\f[R] digits. Uppercase letters are equal to \f[B]9\f[R] plus their position in the alphabet (i.e., \f[B]A\f[R] equals \f[B]10\f[R], or \f[B]9+1\f[R]). .PP If a digit or letter makes no sense with the current value of \f[B]ibase\f[R] (i.e., they are greater than or equal to the current value of \f[B]ibase\f[R]), then the behavior depends on the existence of the \f[B]\-c\f[R]/\f[B]\-\-digit\-clamp\f[R] or \f[B]\-C\f[R]/\f[B]\-\-no\-digit\-clamp\f[R] options (see the \f[B]OPTIONS\f[R] section), the existence and setting of the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or the default, which can be queried with the \f[B]\-h\f[R]/\f[B]\-\-help\f[R] option. .PP If clamping is off, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are not changed. Instead, their given value is multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*A+3\[ha]0*B\f[R], which is \f[B]3\f[R] times \f[B]10\f[R] plus \f[B]11\f[R], or \f[B]41\f[R]. .PP If clamping is on, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are set to the value of the highest valid digit in \f[B]ibase\f[R] before being multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*2+3\[ha]0*2\f[R], which is \f[B]3\f[R] times \f[B]2\f[R] plus \f[B]2\f[R], or \f[B]8\f[R]. .PP There is one exception to clamping: single\-character numbers (i.e., \f[B]A\f[R] alone). Such numbers are never clamped and always take the value they would have in the highest possible \f[B]ibase\f[R]. This means that \f[B]A\f[R] alone always equals decimal \f[B]10\f[R] and \f[B]Z\f[R] alone always equals decimal \f[B]35\f[R]. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current \f[B]ibase\f[R] (with the \f[B]i\f[R] command) regardless of the current value of \f[B]ibase\f[R]. .PP If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for \f[B]A\f[R], use \f[B]0A\f[R]. .PP In addition, dc(1) accepts numbers in scientific notation. These have the form \f[B]e\f[R]. The exponent (the portion after the \f[B]e\f[R]) must be an integer. An example is \f[B]1.89237e9\f[R], which is equal to \f[B]1892370000\f[R]. Negative exponents are also allowed, so \f[B]4.2890e_3\f[R] is equal to \f[B]0.0042890\f[R]. .PP \f[B]WARNING\f[R]: Both the number and the exponent in scientific notation are interpreted according to the current \f[B]ibase\f[R], but the number is still multiplied by \f[B]10\[ha]exponent\f[R] regardless of the current \f[B]ibase\f[R]. For example, if \f[B]ibase\f[R] is \f[B]16\f[R] and dc(1) is given the number string \f[B]FFeA\f[R], the resulting decimal number will be \f[B]2550000000000\f[R], and if dc(1) is given the number string \f[B]10e_4\f[R], the resulting decimal number will be \f[B]0.0016\f[R]. .PP Accepting input as scientific notation is a \f[B]non\-portable extension\f[R]. .SH COMMANDS The valid commands are listed below. .SS Printing These commands are used for printing. .PP Note that both scientific notation and engineering notation are available for printing numbers. Scientific notation is activated by assigning \f[B]0\f[R] to \f[B]obase\f[R] using \f[B]0o\f[R], and engineering notation is activated by assigning \f[B]1\f[R] to \f[B]obase\f[R] using \f[B]1o\f[R]. To deactivate them, just assign a different value to \f[B]obase\f[R]. .PP Printing numbers in scientific notation and/or engineering notation is a \f[B]non\-portable extension\f[R]. .TP \f[B]p\f[R] Prints the value on top of the stack, whether number or string, and prints a newline after. .RS .PP This does not alter the stack. .RE .TP \f[B]n\f[R] Prints the value on top of the stack, whether number or string, and pops it off of the stack. .TP \f[B]P\f[R] Pops a value off the stack. .RS .PP If the value is a number, it is truncated and the absolute value of the result is printed as though \f[B]obase\f[R] is \f[B]256\f[R] and each digit is interpreted as an 8\-bit ASCII character, making it a byte stream. .PP If the value is a string, it is printed without a trailing newline. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]f\f[R] Prints the entire contents of the stack, in order from newest to oldest, without altering anything. .RS .PP Users should use this command when they get lost. .RE .SS Arithmetic These are the commands used for arithmetic. .TP \f[B]+\f[R] The top two values are popped off the stack, added, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]\-\f[R] The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]*\f[R] The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If \f[B]a\f[R] is the \f[I]scale\f[R] of the first expression and \f[B]b\f[R] is the \f[I]scale\f[R] of the second expression, the \f[I]scale\f[R] of the result is equal to \f[B]min(a+b,max(scale,a,b))\f[R] where \f[B]min()\f[R] and \f[B]max()\f[R] return the obvious values. .TP \f[B]/\f[R] The top two values are popped off the stack, divided, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]%\f[R] The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. .RS .PP Remaindering is equivalent to 1) Computing \f[B]a/b\f[R] to current \f[B]scale\f[R], and 2) Using the result of step 1 to calculate \f[B]a\-(a/b)*b\f[R] to \f[I]scale\f[R] \f[B]max(scale+scale(b),scale(a))\f[R]. .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]\[ti]\f[R] The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to \f[B]x y / x y %\f[R] except that \f[B]x\f[R] and \f[B]y\f[R] are only evaluated once. .RS .PP The first value popped off of the stack must be non\-zero. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[ha]\f[R] The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non\-zero. .RE .TP \f[B]v\f[R] The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The value popped off of the stack must be non\-negative. .RE .TP \f[B]_\f[R] If this command \f[I]immediately\f[R] precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. .RS .PP Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]b\f[R] The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]|\f[R] The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. .RS .PP The first value popped is used as the reduction modulus and must be an integer and non\-zero. The second value popped is used as the exponent and must be an integer and non\-negative. The third value popped is the base and must be an integer. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]$\f[R] The top value is popped off the stack and copied, and the copy is truncated and pushed onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[at]\f[R] The top two values are popped off the stack, and the precision of the second is set to the value of the first, whether by truncation or extension. .RS .PP The first value popped off of the stack must be an integer and non\-negative. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]H\f[R] The top two values are popped off the stack, and the second is shifted left (radix shifted right) to the value of the first. .RS .PP The first value popped off of the stack must be an integer and non\-negative. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]h\f[R] The top two values are popped off the stack, and the second is shifted right (radix shifted left) to the value of the first. .RS .PP The first value popped off of the stack must be an integer and non\-negative. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]G\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if they are equal, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]N\f[R] The top value is popped off of the stack, and if it a \f[B]0\f[R], a \f[B]1\f[R] is pushed; otherwise, a \f[B]0\f[R] is pushed. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B](\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]{\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B])\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]}\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]M\f[R] The top two values are popped off of the stack. If they are both non\-zero, a \f[B]1\f[R] is pushed onto the stack. If either of them is zero, or both of them are, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]&&\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]m\f[R] The top two values are popped off of the stack. If at least one of them is non\-zero, a \f[B]1\f[R] is pushed onto the stack. If both of them are zero, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]||\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Pseudo\-Random Number Generator dc(1) has a built\-in pseudo\-random number generator. These commands query the pseudo\-random number generator. (See Parameters for more information about the \f[B]seed\f[R] value that controls the pseudo\-random number generator.) .PP The pseudo\-random number generator is guaranteed to \f[B]NOT\f[R] be cryptographically secure. .TP \f[B]\[cq]\f[R] Generates an integer between 0 and \f[B]DC_RAND_MAX\f[R], inclusive (see the \f[B]LIMITS\f[R] section). .RS .PP The generated integer is made as unbiased as possible, subject to the limitations of the pseudo\-random number generator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[lq]\f[R] Pops a value off of the stack, which is used as an \f[B]exclusive\f[R] upper bound on the integer that will be generated. If the bound is negative or is a non\-integer, an error is raised, and dc(1) resets (see the \f[B]RESET\f[R] section) while \f[B]seed\f[R] remains unchanged. If the bound is larger than \f[B]DC_RAND_MAX\f[R], the higher bound is honored by generating several pseudo\-random integers, multiplying them by appropriate powers of \f[B]DC_RAND_MAX+1\f[R], and adding them together. Thus, the size of integer that can be generated with this command is unbounded. Using this command will change the value of \f[B]seed\f[R], unless the operand is \f[B]0\f[R] or \f[B]1\f[R]. In that case, \f[B]0\f[R] is pushed onto the stack, and \f[B]seed\f[R] is \f[I]not\f[R] changed. .RS .PP The generated integer is made as unbiased as possible, subject to the limitations of the pseudo\-random number generator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Stack Control These commands control the stack. .TP \f[B]c\f[R] Removes all items from (\[lq]clears\[rq]) the stack. .TP \f[B]d\f[R] Copies the item on top of the stack (\[lq]duplicates\[rq]) and pushes the copy onto the stack. .TP \f[B]r\f[R] Swaps (\[lq]reverses\[rq]) the two top items on the stack. .TP \f[B]R\f[R] Pops (\[lq]removes\[rq]) the top value from the stack. .SS Register Control These commands control registers (see the \f[B]REGISTERS\f[R] section). .TP \f[B]s\f[R]\f[I]r\f[R] Pops the value off the top of the stack and stores it into register \f[I]r\f[R]. .TP \f[B]l\f[R]\f[I]r\f[R] Copies the value in register \f[I]r\f[R] and pushes it onto the stack. This does not alter the contents of \f[I]r\f[R]. .TP \f[B]S\f[R]\f[I]r\f[R] Pops the value off the top of the (main) stack and pushes it onto the stack of register \f[I]r\f[R]. The previous value of the register becomes inaccessible. .TP \f[B]L\f[R]\f[I]r\f[R] Pops the value off the top of the stack for register \f[I]r\f[R] and push it onto the main stack. The previous value in the stack for register \f[I]r\f[R], if any, is now accessible via the \f[B]l\f[R]\f[I]r\f[R] command. .SS Parameters These commands control the values of \f[B]ibase\f[R], \f[B]obase\f[R], \f[B]scale\f[R], and \f[B]seed\f[R]. Also see the \f[B]SYNTAX\f[R] section. .TP \f[B]i\f[R] Pops the value off of the top of the stack and uses it to set \f[B]ibase\f[R], which must be between \f[B]2\f[R] and \f[B]16\f[R], inclusive. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]o\f[R] Pops the value off of the top of the stack and uses it to set \f[B]obase\f[R], which must be between \f[B]0\f[R] and \f[B]DC_BASE_MAX\f[R], inclusive (see the \f[B]LIMITS\f[R] section and the \f[B]NUMBERS\f[R] section). .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]k\f[R] Pops the value off of the top of the stack and uses it to set \f[B]scale\f[R], which must be non\-negative. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]j\f[R] Pops the value off of the top of the stack and uses it to set \f[B]seed\f[R]. The meaning of \f[B]seed\f[R] is dependent on the current pseudo\-random number generator but is guaranteed to not change except for new major versions. .RS .PP The \f[I]scale\f[R] and sign of the value may be significant. .PP If a previously used \f[B]seed\f[R] value is used again, the pseudo\-random number generator is guaranteed to produce the same sequence of pseudo\-random numbers as it did when the \f[B]seed\f[R] value was previously used. .PP The exact value assigned to \f[B]seed\f[R] is not guaranteed to be returned if the \f[B]J\f[R] command is used. However, if \f[B]seed\f[R] \f[I]does\f[R] return a different value, both values, when assigned to \f[B]seed\f[R], are guaranteed to produce the same sequence of pseudo\-random numbers. This means that certain values assigned to \f[B]seed\f[R] will not produce unique sequences of pseudo\-random numbers. .PP There is no limit to the length (number of significant decimal digits) or \f[I]scale\f[R] of the value that can be assigned to \f[B]seed\f[R]. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]I\f[R] Pushes the current value of \f[B]ibase\f[R] onto the main stack. .TP \f[B]O\f[R] Pushes the current value of \f[B]obase\f[R] onto the main stack. .TP \f[B]K\f[R] Pushes the current value of \f[B]scale\f[R] onto the main stack. .TP \f[B]J\f[R] Pushes the current value of \f[B]seed\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]T\f[R] Pushes the maximum allowable value of \f[B]ibase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]U\f[R] Pushes the maximum allowable value of \f[B]obase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]V\f[R] Pushes the maximum allowable value of \f[B]scale\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]W\f[R] Pushes the maximum (inclusive) integer that can be generated with the \f[B]\[cq]\f[R] pseudo\-random number generator command. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Strings The following commands control strings. .PP dc(1) can work with both numbers and strings, and registers (see the \f[B]REGISTERS\f[R] section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. .PP While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. .PP Strings can also be executed as macros. For example, if the string \f[B][1pR]\f[R] is executed as a macro, then the code \f[B]1pR\f[R] is executed, meaning that the \f[B]1\f[R] will be printed with a newline after and then popped from the stack. .TP \f[B][\f[R]\f[I]characters\f[R]\f[B]]\f[R] Makes a string containing \f[I]characters\f[R] and pushes it onto the stack. .RS .PP If there are brackets (\f[B][\f[R] and \f[B]]\f[R]) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (\f[B]\[rs]\f[R]) character. .PP If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. .RE .TP \f[B]a\f[R] The value on top of the stack is popped. .RS .PP If it is a number, it is truncated and its absolute value is taken. The result mod \f[B]256\f[R] is calculated. If that result is \f[B]0\f[R], push an empty string; otherwise, push a one\-character string where the character is the result of the mod interpreted as an ASCII character. .PP If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one\-character string. The new string is then pushed onto the stack. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]x\f[R] Pops a value off of the top of the stack. .RS .PP If it is a number, it is pushed back onto the stack. .PP If it is a string, it is executed as a macro. .PP This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. .RE .TP \f[B]>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP For example, \f[B]0 1>a\f[R] will execute the contents of register \f[B]a\f[R], and \f[B]1 0>a\f[R] will not. .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]?\f[R] Reads a line from the \f[B]stdin\f[R] and executes it. This is to allow macros to request input from users. .TP \f[B]q\f[R] During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. .TP \f[B]Q\f[R] Pops a value from the stack which must be non\-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. .TP \f[B],\f[R] Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the \f[B]Q\f[R] command, so the sequence \f[B],Q\f[R] will make dc(1) exit. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Status These commands query status of the stack or its top value. .TP \f[B]Z\f[R] Pops a value off of the stack. .RS .PP If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push \f[B]1\f[R] if the argument is \f[B]0\f[R] with no decimal places. .PP If it is a string, pushes the number of characters the string has. .RE .TP \f[B]X\f[R] Pops a value off of the stack. .RS .PP If it is a number, pushes the \f[I]scale\f[R] of the value onto the stack. .PP If it is a string, pushes \f[B]0\f[R]. .RE .TP \f[B]u\f[R] Pops one value off of the stack. If the value is a number, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a string), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]t\f[R] Pops one value off of the stack. If the value is a string, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a number), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]z\f[R] Pushes the current depth of the stack (before execution of this command) onto the stack. .TP \f[B]y\f[R]\f[I]r\f[R] Pushes the current stack depth of the register \f[I]r\f[R] onto the main stack. .RS .PP Because each register has a depth of \f[B]1\f[R] (with the value \f[B]0\f[R] in the top item) when dc(1) starts, dc(1) requires that each register\[cq]s stack must always have at least one item; dc(1) will give an error and reset otherwise (see the \f[B]RESET\f[R] section). This means that this command will never push \f[B]0\f[R]. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Arrays These commands manipulate arrays. .TP \f[B]:\f[R]\f[I]r\f[R] Pops the top two values off of the stack. The second value will be stored in the array \f[I]r\f[R] (see the \f[B]REGISTERS\f[R] section), indexed by the first value. .TP \f[B];\f[R]\f[I]r\f[R] Pops the value on top of the stack and uses it as an index into the array \f[I]r\f[R]. The selected value is then pushed onto the stack. .TP \f[B]Y\f[R]\f[I]r\f[R] Pushes the length of the array \f[I]r\f[R] onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter \f[B]g\f[R]. Only the characters below are allowed after the character \f[B]g\f[R]; any other character produces a parse error (see the \f[B]ERRORS\f[R] section). .TP \f[B]gl\f[R] Pushes the line length set by \f[B]DC_LINE_LENGTH\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) onto the stack. .TP \f[B]gx\f[R] Pushes \f[B]1\f[R] onto the stack if extended register mode is on, \f[B]0\f[R] otherwise. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .TP \f[B]gz\f[R] Pushes \f[B]0\f[R] onto the stack if the leading zero setting has not been enabled with the \f[B]\-z\f[R] or \f[B]\-\-leading\-zeroes\f[R] options (see the \f[B]OPTIONS\f[R] section), non\-zero otherwise. .SH REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) .PP Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (\f[B]0\f[R]) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. .PP In non\-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (\f[B]`\[rs]n'\f[R]) and a left bracket (\f[B]`['\f[R]); it is a parse error for a newline or a left bracket to be used as a register name. .SS Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. .PP If extended register mode is enabled (\f[B]\-x\f[R] or \f[B]\-\-extended\-register\f[R] command\-line arguments are given), then normal single character registers are used \f[I]unless\f[R] the character immediately following a command that needs a register name is a space (according to \f[B]isspace()\f[R]) and not a newline (\f[B]`\[rs]n'\f[R]). .PP In that case, the register name is found according to the regex \f[B][a\-z][a\-z0\-9_]*\f[R] (like bc(1) identifiers), and it is a parse error if the next non\-space characters do not match that regex. .SH RESET When dc(1) encounters an error or a signal that it has a non\-default handler for, it resets. This means that several things happen. .PP First, any macros that are executing are stopped and popped off the -stack. +execution stack. The behavior is not unlike that of exceptions in programming languages. Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. .PP +However, the stack of values is \f[I]not\f[R] cleared; in interactive +mode, users can inspect the stack and manipulate it. +.PP Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the \f[B]EXIT STATUS\f[R] section), it asks for more input; otherwise, it exits with the appropriate return code. .SH PERFORMANCE Most dc(1) implementations use \f[B]char\f[R] types to calculate the value of \f[B]1\f[R] decimal digit at a time, but that can be slow. This dc(1) does something different. .PP It uses large integers to calculate more than \f[B]1\f[R] decimal digit at a time. If built in a environment where \f[B]DC_LONG_BIT\f[R] (see the \f[B]LIMITS\f[R] section) is \f[B]64\f[R], then each integer has \f[B]9\f[R] decimal digits. If built in an environment where \f[B]DC_LONG_BIT\f[R] is \f[B]32\f[R] then each integer has \f[B]4\f[R] decimal digits. This value (the number of decimal digits per large integer) is called \f[B]DC_BASE_DIGS\f[R]. .PP In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of \f[B]DC_LONG_BIT\f[R], but is always at least twice as large as the integer type used to store digits. .SH LIMITS The following are the limits on dc(1): .TP \f[B]DC_LONG_BIT\f[R] The number of bits in the \f[B]long\f[R] type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the \f[B]PERFORMANCE\f[R] section). .TP \f[B]DC_BASE_DIGS\f[R] The number of decimal digits per large integer (see the \f[B]PERFORMANCE\f[R] section). Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_POW\f[R] The max decimal number that each large integer can store (see \f[B]DC_BASE_DIGS\f[R]) plus \f[B]1\f[R]. Depends on \f[B]DC_BASE_DIGS\f[R]. .TP \f[B]DC_OVERFLOW_MAX\f[R] The max number that the overflow type (see the \f[B]PERFORMANCE\f[R] section) can hold. Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_MAX\f[R] The maximum output base. Set at \f[B]DC_BASE_POW\f[R]. .TP \f[B]DC_DIM_MAX\f[R] The maximum size of arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .TP \f[B]DC_SCALE_MAX\f[R] The maximum \f[B]scale\f[R]. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_STRING_MAX\f[R] The maximum length of strings. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NAME_MAX\f[R] The maximum length of identifiers. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NUM_MAX\f[R] The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_RAND_MAX\f[R] The maximum integer (inclusive) returned by the \f[B]\[cq]\f[R] command, if dc(1). Set at \f[B]2\[ha]DC_LONG_BIT\-1\f[R]. .TP Exponent The maximum allowable exponent (positive or negative). Set at \f[B]DC_OVERFLOW_MAX\f[R]. .TP Number of vars The maximum number of vars/arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .PP These limits are meant to be effectively non\-existent; the limits are so large (at least on 64\-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. .SH ENVIRONMENT VARIABLES As \f[B]non\-portable extensions\f[R], dc(1) recognizes the following environment variables: .TP \f[B]DC_ENV_ARGS\f[R] This is another way to give command\-line arguments to dc(1). They should be in the same format as all other command\-line arguments. These are always processed first, so any files given in \f[B]DC_ENV_ARGS\f[R] will be processed before arguments and files given on the command\-line. This gives the user the ability to set up \[lq]standard\[rq] options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the \f[B]\-e\f[R] option to set \f[B]scale\f[R] to a value other than \f[B]0\f[R]. .RS .PP The code that parses \f[B]DC_ENV_ARGS\f[R] will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string \f[B]\[lq]/home/gavin/some dc file.dc\[rq]\f[R] will be correctly parsed, but the string \f[B]\[lq]/home/gavin/some \[dq]dc\[dq] file.dc\[rq]\f[R] will include the backslashes. .PP The quote parsing will handle either kind of quotes, \f[B]\[cq]\f[R] or \f[B]\[lq]\f[R]. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in \f[B]\[lq]some `dc' file.dc\[rq]\f[R], and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in \f[B]DC_ENV_ARGS\f[R] is not supported due to the complexity of the parsing, though such files are still supported on the command\-line where the parsing is done by the shell. .RE .TP \f[B]DC_LINE_LENGTH\f[R] If this environment variable exists and contains an integer that is greater than \f[B]1\f[R] and is less than \f[B]UINT16_MAX\f[R] (\f[B]2\[ha]16\-1\f[R]), dc(1) will output lines to that length, including the backslash newline combo. The default line length is \f[B]70\f[R]. .RS .PP The special value of \f[B]0\f[R] will disable line length checking and print numbers without regard to line length and without backslashes and newlines. .RE .TP \f[B]DC_SIGINT_RESET\f[R] If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), then this environment variable has no effect because dc(1) exits on \f[B]SIGINT\f[R] when not in interactive mode. .RS .PP However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) reset on \f[B]SIGINT\f[R], rather than exit, and zero makes dc(1) exit. If this environment variable exists and is \f[I]not\f[R] an integer, then dc(1) will exit on \f[B]SIGINT\f[R]. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_TTY_MODE\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non\-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_PROMPT\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) use a prompt, and zero or a non\-integer makes dc(1) not use a prompt. If this environment variable does not exist and \f[B]DC_TTY_MODE\f[R] does, then the value of the \f[B]DC_TTY_MODE\f[R] environment variable is used. .PP This environment variable and the \f[B]DC_TTY_MODE\f[R] environment variable override the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_EXPR_EXIT\f[R] If any expressions or expression files are given on the command\-line with \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R], then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. .RS .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_DIGIT_CLAMP\f[R] When parsing numbers and if this environment variable exists and contains an integer, a non\-zero value makes dc(1) clamp digits that are greater than or equal to the current \f[B]ibase\f[R] so that all such digits are considered equal to the \f[B]ibase\f[R] minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the \f[B]ibase\f[R]. .RS .PP This never applies to single\-digit numbers, as per the bc(1) standard (see the \f[B]STANDARDS\f[R] section). .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .SH EXIT STATUS dc(1) returns the following exit statuses: .TP \f[B]0\f[R] No error. .TP \f[B]1\f[R] A math error occurred. This follows standard practice of using \f[B]1\f[R] for expected errors, since math errors will happen in the process of normal execution. .RS .PP Math errors include divide by \f[B]0\f[R], taking the square root of a negative number, using a negative number as a bound for the pseudo\-random number generator, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non\-integer where an integer is required. .PP Converting to a hardware integer happens for the second operand of the power (\f[B]\[ha]\f[R]), places (\f[B]\[at]\f[R]), left shift (\f[B]H\f[R]), and right shift (\f[B]h\f[R]) operators. .RE .TP \f[B]2\f[R] A parse error occurred. .RS .PP Parse errors include unexpected \f[B]EOF\f[R], using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. .RE .TP \f[B]3\f[R] A runtime error occurred. .RS .PP Runtime errors include assigning an invalid number to any global (\f[B]ibase\f[R], \f[B]obase\f[R], or \f[B]scale\f[R]), giving a bad expression to a \f[B]read()\f[R] call, calling \f[B]read()\f[R] inside of a \f[B]read()\f[R] call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. .RE .TP \f[B]4\f[R] A fatal error occurred. .RS .PP Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command\-line options. .RE .PP The exit status \f[B]4\f[R] is special; when a fatal error occurs, dc(1) always exits and returns \f[B]4\f[R], no matter what mode dc(1) is in. .PP The other statuses will only be returned when dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), since dc(1) resets its state (see the \f[B]RESET\f[R] section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .PP These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .SH INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non\-interactive mode. Interactive mode is turned on automatically when both \f[B]stdin\f[R] and \f[B]stdout\f[R] are hooked to a terminal, but the \f[B]\-i\f[R] flag and \f[B]\-\-interactive\f[R] option can turn it on in other situations. .PP In interactive mode, dc(1) attempts to recover from errors (see the \f[B]RESET\f[R] section), and in normal execution, flushes \f[B]stdout\f[R] as soon as execution is done for the current input. dc(1) may also reset on \f[B]SIGINT\f[R] instead of exit, depending on the contents of, or default for, the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .SH TTY MODE If \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY, then \[lq]TTY mode\[rq] is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. .PP If there is the environment variable \f[B]DC_TTY_MODE\f[R] in the environment (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then if that environment variable contains a non\-zero integer, dc(1) will turn on TTY mode when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY. If the \f[B]DC_TTY_MODE\f[R] environment variable exists but is \f[I]not\f[R] a non\-zero integer, then dc(1) will not turn TTY mode on. .PP If the environment variable \f[B]DC_TTY_MODE\f[R] does \f[I]not\f[R] exist, the default setting is used. The default setting can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the \f[B]STANDARDS\f[R] section), and interactive mode requires only \f[B]stdin\f[R] and \f[B]stdout\f[R] to be connected to a terminal. .SS Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: \f[B]DC_PROMPT\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP If the environment variable \f[B]DC_PROMPT\f[R] exists and is a non\-zero integer, then the prompt is turned on when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are connected to a TTY and the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options were not used. The read prompt will be turned on under the same conditions, except that the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options must also not be used. .PP However, if \f[B]DC_PROMPT\f[R] does not exist, the prompt can be enabled or disabled with the \f[B]DC_TTY_MODE\f[R] environment variable, the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options, and the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options. See the \f[B]ENVIRONMENT VARIABLES\f[R] and \f[B]OPTIONS\f[R] sections for more details. .SH SIGNAL HANDLING Sending a \f[B]SIGINT\f[R] will cause dc(1) to do one of two things. .PP If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), or the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or its default, is either not an integer or it is zero, dc(1) will exit. .PP However, if dc(1) is in interactive mode, and the \f[B]DC_SIGINT_RESET\f[R] or its default is an integer and non\-zero, then dc(1) will stop executing the current input and reset (see the \f[B]RESET\f[R] section) upon receiving a \f[B]SIGINT\f[R]. .PP Note that \[lq]current input\[rq] can mean one of two things. If dc(1) is processing input from \f[B]stdin\f[R] in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from \f[B]stdin\f[R] if no other file exists. .PP This means that if a \f[B]SIGINT\f[R] is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. .PP \f[B]SIGTERM\f[R] and \f[B]SIGQUIT\f[R] cause dc(1) to clean up and exit, and it uses the default handler for all other signals. .SH LOCALES This dc(1) ships with support for adding error messages for different locales and thus, supports \f[B]LC_MESSAGES\f[R]. .SH SEE ALSO bc(1) .SH STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1\-2017 (\[lq]POSIX.1\-2017\[rq]) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . .SH BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . .SH AUTHOR Gavin D. Howard \c .MT gavin@gavinhoward.com .ME \c \ and contributors. diff --git a/manuals/dc/H.1.md b/manuals/dc/H.1.md index cc263eea0db4..64c7142bc4a7 100644 --- a/manuals/dc/H.1.md +++ b/manuals/dc/H.1.md @@ -1,1503 +1,1506 @@ # Name dc - arbitrary-precision decimal reverse-Polish notation calculator # SYNOPSIS **dc** [**-cChiPRvVx**] [**-\-version**] [**-\-help**] [**-\-digit-clamp**] [**-\-no-digit-clamp**] [**-\-interactive**] [**-\-no-prompt**] [**-\-no-read-prompt**] [**-\-extended-register**] [**-e** *expr*] [**-\-expression**=*expr*...] [**-f** *file*...] [**-\-file**=*file*...] [*file*...] [**-I** *ibase*] [**-\-ibase**=*ibase*] [**-O** *obase*] [**-\-obase**=*obase*] [**-S** *scale*] [**-\-scale**=*scale*] [**-E** *seed*] [**-\-seed**=*seed*] # DESCRIPTION dc(1) is an arbitrary-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. If no files are given on the command-line, then dc(1) reads from **stdin** (see the **STDIN** section). Otherwise, those files are processed, and dc(1) will then exit. If a user wants to set up a standard environment, they can use **DC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). For example, if a user wants the **scale** always set to **10**, they can set **DC_ENV_ARGS** to **-e 10k**, and this dc(1) will always start with a **scale** of **10**. # OPTIONS The following are the options that dc(1) accepts. **-C**, **-\-no-digit-clamp** : Disables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that the value added to a number from a digit is always that digit's value multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-c** or **-\-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-c**, **-\-digit-clamp** : Enables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-C** or **-\-no-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-E** *seed*, **-\-seed**=*seed* : Sets the builtin variable **seed** to the value *seed* assuming that *seed* is in base 10. It is a fatal error if *seed* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-e** *expr*, **-\-expression**=*expr* : Evaluates *expr*. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**, whether on the command-line or in **DC_ENV_ARGS**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-f** *file*, **-\-file**=*file* : Reads in *file* and evaluates it, line by line, as though it were read through **stdin**. If expressions are also given (see above), the expressions are evaluated in the order given. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-h**, **-\-help** : Prints a usage message and exits. **-I** *ibase*, **-\-ibase**=*ibase* : Sets the builtin variable **ibase** to the value *ibase* assuming that *ibase* is in base 10. It is a fatal error if *ibase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-i**, **-\-interactive** : Forces interactive mode. (See the **INTERACTIVE MODE** section.) This is a **non-portable extension**. **-L**, **-\-no-line-length** : Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets **BC_LINE_LENGTH** to **0** (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-O** *obase*, **-\-obase**=*obase* : Sets the builtin variable **obase** to the value *obase* assuming that *obase* is in base 10. It is a fatal error if *obase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-P**, **-\-no-prompt** : Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in **DC_ENV_ARGS**. These options override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-R**, **-\-no-read-prompt** : Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in **BC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. This option does not disable the regular prompt because the read prompt is only used when the **?** command is used. These options *do* override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section), but only for the read prompt. This is a **non-portable extension**. **-S** *scale*, **-\-scale**=*scale* : Sets the builtin variable **scale** to the value *scale* assuming that *scale* is in base 10. It is a fatal error if *scale* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-v**, **-V**, **-\-version** : Print the version information (copyright header) and exits. **-x** **-\-extended-register** : Enables extended register mode. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. This is a **non-portable extension**. **-z**, **-\-leading-zeroes** : Makes dc(1) print all numbers greater than **-1** and less than **1**, and not equal to **0**, with a leading zero. This is a **non-portable extension**. All long options are **non-portable extensions**. # STDIN If no files are given on the command-line and no files or expressions are given by the **-f**, **-\-file**, **-e**, or **-\-expression** options, then dc(1) reads from **stdin**. However, there is a caveat to this. First, **stdin** is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. # STDOUT Any non-error output is written to **stdout**. In addition, if history (see the **HISTORY** section) and the prompt (see the **TTY MODE** section) are enabled, both are output to **stdout**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stdout**, so if **stdout** is closed, as in **dc >&-**, it will quit with an error. This is done so that dc(1) can report problems when **stdout** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stdout** to **/dev/null**. # STDERR Any error output is written to **stderr**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stderr**, so if **stderr** is closed, as in **dc 2>&-**, it will quit with an error. This is done so that dc(1) can exit with an error code when **stderr** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stderr** to **/dev/null**. # SYNTAX Each item in the input source code, either a number (see the **NUMBERS** section) or a command (see the **COMMANDS** section), is processed and executed, in order. Input is processed immediately when entered. **ibase** is a register (see the **REGISTERS** section) that determines how to interpret constant numbers. It is the "input" base, or the number base used for interpreting input numbers. **ibase** is initially **10**. The max allowable value for **ibase** is **16**. The min allowable value for **ibase** is **2**. The max allowable value for **ibase** can be queried in dc(1) programs with the **T** command. **obase** is a register (see the **REGISTERS** section) that determines how to output results. It is the "output" base, or the number base used for outputting numbers. **obase** is initially **10**. The max allowable value for **obase** is **DC_BASE_MAX** and can be queried with the **U** command. The min allowable value for **obase** is **0**. If **obase** is **0**, values are output in scientific notation, and if **obase** is **1**, values are output in engineering notation. Otherwise, values are output in the specified base. Outputting in scientific and engineering notations are **non-portable extensions**. The *scale* of an expression is the number of digits in the result of the expression right of the decimal point, and **scale** is a register (see the **REGISTERS** section) that sets the precision of any operations (with exceptions). **scale** is initially **0**. **scale** cannot be negative. The max allowable value for **scale** can be queried in dc(1) programs with the **V** command. **seed** is a register containing the current seed for the pseudo-random number generator. If the current value of **seed** is queried and stored, then if it is assigned to **seed** later, the pseudo-random number generator is guaranteed to produce the same sequence of pseudo-random numbers that were generated after the value of **seed** was first queried. Multiple values assigned to **seed** can produce the same sequence of pseudo-random numbers. Likewise, when a value is assigned to **seed**, it is not guaranteed that querying **seed** immediately after will return the same value. In addition, the value of **seed** will change after any call to the **'** command or the **"** command that does not get receive a value of **0** or **1**. The maximum integer returned by the **'** command can be queried with the **W** command. **Note**: The values returned by the pseudo-random number generator with the **'** and **"** commands are guaranteed to **NOT** be cryptographically secure. This is a consequence of using a seeded pseudo-random number generator. However, they *are* guaranteed to be reproducible with identical **seed** values. This means that the pseudo-random values from dc(1) should only be used where a reproducible stream of pseudo-random numbers is *ESSENTIAL*. In any other case, use a non-seeded pseudo-random number generator. The pseudo-random number generator, **seed**, and all associated operations are **non-portable extensions**. ## Comments Comments go from **#** until, and not including, the next newline. This is a **non-portable extension**. # NUMBERS Numbers are strings made up of digits, uppercase letters up to **F**, and at most **1** period for a radix. Numbers can have up to **DC_NUM_MAX** digits. Uppercase letters are equal to **9** plus their position in the alphabet (i.e., **A** equals **10**, or **9+1**). If a digit or letter makes no sense with the current value of **ibase** (i.e., they are greater than or equal to the current value of **ibase**), then the behavior depends on the existence of the **-c**/**-\-digit-clamp** or **-C**/**-\-no-digit-clamp** options (see the **OPTIONS** section), the existence and setting of the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section), or the default, which can be queried with the **-h**/**-\-help** option. If clamping is off, then digits or letters that are greater than or equal to the current value of **ibase** are not changed. Instead, their given value is multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*A+3\^0\*B**, which is **3** times **10** plus **11**, or **41**. If clamping is on, then digits or letters that are greater than or equal to the current value of **ibase** are set to the value of the highest valid digit in **ibase** before being multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*2+3\^0\*2**, which is **3** times **2** plus **2**, or **8**. There is one exception to clamping: single-character numbers (i.e., **A** alone). Such numbers are never clamped and always take the value they would have in the highest possible **ibase**. This means that **A** alone always equals decimal **10** and **Z** alone always equals decimal **35**. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current **ibase** (with the **i** command) regardless of the current value of **ibase**. If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for **A**, use **0A**. In addition, dc(1) accepts numbers in scientific notation. These have the form **\e\**. The exponent (the portion after the **e**) must be an integer. An example is **1.89237e9**, which is equal to **1892370000**. Negative exponents are also allowed, so **4.2890e_3** is equal to **0.0042890**. **WARNING**: Both the number and the exponent in scientific notation are interpreted according to the current **ibase**, but the number is still multiplied by **10\^exponent** regardless of the current **ibase**. For example, if **ibase** is **16** and dc(1) is given the number string **FFeA**, the resulting decimal number will be **2550000000000**, and if dc(1) is given the number string **10e_4**, the resulting decimal number will be **0.0016**. Accepting input as scientific notation is a **non-portable extension**. # COMMANDS The valid commands are listed below. ## Printing These commands are used for printing. Note that both scientific notation and engineering notation are available for printing numbers. Scientific notation is activated by assigning **0** to **obase** using **0o**, and engineering notation is activated by assigning **1** to **obase** using **1o**. To deactivate them, just assign a different value to **obase**. Printing numbers in scientific notation and/or engineering notation is a **non-portable extension**. **p** : Prints the value on top of the stack, whether number or string, and prints a newline after. This does not alter the stack. **n** : Prints the value on top of the stack, whether number or string, and pops it off of the stack. **P** : Pops a value off the stack. If the value is a number, it is truncated and the absolute value of the result is printed as though **obase** is **256** and each digit is interpreted as an 8-bit ASCII character, making it a byte stream. If the value is a string, it is printed without a trailing newline. This is a **non-portable extension**. **f** : Prints the entire contents of the stack, in order from newest to oldest, without altering anything. Users should use this command when they get lost. ## Arithmetic These are the commands used for arithmetic. **+** : The top two values are popped off the stack, added, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **-** : The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **\*** : The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If **a** is the *scale* of the first expression and **b** is the *scale* of the second expression, the *scale* of the result is equal to **min(a+b,max(scale,a,b))** where **min()** and **max()** return the obvious values. **/** : The top two values are popped off the stack, divided, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be non-zero. **%** : The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. Remaindering is equivalent to 1) Computing **a/b** to current **scale**, and 2) Using the result of step 1 to calculate **a-(a/b)\*b** to *scale* **max(scale+scale(b),scale(a))**. The first value popped off of the stack must be non-zero. **~** : The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to **x y / x y %** except that **x** and **y** are only evaluated once. The first value popped off of the stack must be non-zero. This is a **non-portable extension**. **\^** : The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non-zero. **v** : The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The value popped off of the stack must be non-negative. **\_** : If this command *immediately* precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a **non-portable extension**. **b** : The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. This is a **non-portable extension**. **|** : The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. The first value popped is used as the reduction modulus and must be an integer and non-zero. The second value popped is used as the exponent and must be an integer and non-negative. The third value popped is the base and must be an integer. This is a **non-portable extension**. **\$** : The top value is popped off the stack and copied, and the copy is truncated and pushed onto the stack. This is a **non-portable extension**. **\@** : The top two values are popped off the stack, and the precision of the second is set to the value of the first, whether by truncation or extension. The first value popped off of the stack must be an integer and non-negative. This is a **non-portable extension**. **H** : The top two values are popped off the stack, and the second is shifted left (radix shifted right) to the value of the first. The first value popped off of the stack must be an integer and non-negative. This is a **non-portable extension**. **h** : The top two values are popped off the stack, and the second is shifted right (radix shifted left) to the value of the first. The first value popped off of the stack must be an integer and non-negative. This is a **non-portable extension**. **G** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if they are equal, or **0** otherwise. This is a **non-portable extension**. **N** : The top value is popped off of the stack, and if it a **0**, a **1** is pushed; otherwise, a **0** is pushed. This is a **non-portable extension**. **(** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than the second, or **0** otherwise. This is a **non-portable extension**. **{** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **)** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than the second, or **0** otherwise. This is a **non-portable extension**. **}** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **M** : The top two values are popped off of the stack. If they are both non-zero, a **1** is pushed onto the stack. If either of them is zero, or both of them are, then a **0** is pushed onto the stack. This is like the **&&** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. **m** : The top two values are popped off of the stack. If at least one of them is non-zero, a **1** is pushed onto the stack. If both of them are zero, then a **0** is pushed onto the stack. This is like the **||** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. ## Pseudo-Random Number Generator dc(1) has a built-in pseudo-random number generator. These commands query the pseudo-random number generator. (See Parameters for more information about the **seed** value that controls the pseudo-random number generator.) The pseudo-random number generator is guaranteed to **NOT** be cryptographically secure. **'** : Generates an integer between 0 and **DC_RAND_MAX**, inclusive (see the **LIMITS** section). The generated integer is made as unbiased as possible, subject to the limitations of the pseudo-random number generator. This is a **non-portable extension**. **"** : Pops a value off of the stack, which is used as an **exclusive** upper bound on the integer that will be generated. If the bound is negative or is a non-integer, an error is raised, and dc(1) resets (see the **RESET** section) while **seed** remains unchanged. If the bound is larger than **DC_RAND_MAX**, the higher bound is honored by generating several pseudo-random integers, multiplying them by appropriate powers of **DC_RAND_MAX+1**, and adding them together. Thus, the size of integer that can be generated with this command is unbounded. Using this command will change the value of **seed**, unless the operand is **0** or **1**. In that case, **0** is pushed onto the stack, and **seed** is *not* changed. The generated integer is made as unbiased as possible, subject to the limitations of the pseudo-random number generator. This is a **non-portable extension**. ## Stack Control These commands control the stack. **c** : Removes all items from ("clears") the stack. **d** : Copies the item on top of the stack ("duplicates") and pushes the copy onto the stack. **r** : Swaps ("reverses") the two top items on the stack. **R** : Pops ("removes") the top value from the stack. ## Register Control These commands control registers (see the **REGISTERS** section). **s**_r_ : Pops the value off the top of the stack and stores it into register *r*. **l**_r_ : Copies the value in register *r* and pushes it onto the stack. This does not alter the contents of *r*. **S**_r_ : Pops the value off the top of the (main) stack and pushes it onto the stack of register *r*. The previous value of the register becomes inaccessible. **L**_r_ : Pops the value off the top of the stack for register *r* and push it onto the main stack. The previous value in the stack for register *r*, if any, is now accessible via the **l**_r_ command. ## Parameters These commands control the values of **ibase**, **obase**, **scale**, and **seed**. Also see the **SYNTAX** section. **i** : Pops the value off of the top of the stack and uses it to set **ibase**, which must be between **2** and **16**, inclusive. If the value on top of the stack has any *scale*, the *scale* is ignored. **o** : Pops the value off of the top of the stack and uses it to set **obase**, which must be between **0** and **DC_BASE_MAX**, inclusive (see the **LIMITS** section and the **NUMBERS** section). If the value on top of the stack has any *scale*, the *scale* is ignored. **k** : Pops the value off of the top of the stack and uses it to set **scale**, which must be non-negative. If the value on top of the stack has any *scale*, the *scale* is ignored. **j** : Pops the value off of the top of the stack and uses it to set **seed**. The meaning of **seed** is dependent on the current pseudo-random number generator but is guaranteed to not change except for new major versions. The *scale* and sign of the value may be significant. If a previously used **seed** value is used again, the pseudo-random number generator is guaranteed to produce the same sequence of pseudo-random numbers as it did when the **seed** value was previously used. The exact value assigned to **seed** is not guaranteed to be returned if the **J** command is used. However, if **seed** *does* return a different value, both values, when assigned to **seed**, are guaranteed to produce the same sequence of pseudo-random numbers. This means that certain values assigned to **seed** will not produce unique sequences of pseudo-random numbers. There is no limit to the length (number of significant decimal digits) or *scale* of the value that can be assigned to **seed**. This is a **non-portable extension**. **I** : Pushes the current value of **ibase** onto the main stack. **O** : Pushes the current value of **obase** onto the main stack. **K** : Pushes the current value of **scale** onto the main stack. **J** : Pushes the current value of **seed** onto the main stack. This is a **non-portable extension**. **T** : Pushes the maximum allowable value of **ibase** onto the main stack. This is a **non-portable extension**. **U** : Pushes the maximum allowable value of **obase** onto the main stack. This is a **non-portable extension**. **V** : Pushes the maximum allowable value of **scale** onto the main stack. This is a **non-portable extension**. **W** : Pushes the maximum (inclusive) integer that can be generated with the **'** pseudo-random number generator command. This is a **non-portable extension**. ## Strings The following commands control strings. dc(1) can work with both numbers and strings, and registers (see the **REGISTERS** section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. Strings can also be executed as macros. For example, if the string **[1pR]** is executed as a macro, then the code **1pR** is executed, meaning that the **1** will be printed with a newline after and then popped from the stack. **\[**_characters_**\]** : Makes a string containing *characters* and pushes it onto the stack. If there are brackets (**\[** and **\]**) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (**\\**) character. If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. **a** : The value on top of the stack is popped. If it is a number, it is truncated and its absolute value is taken. The result mod **256** is calculated. If that result is **0**, push an empty string; otherwise, push a one-character string where the character is the result of the mod interpreted as an ASCII character. If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one-character string. The new string is then pushed onto the stack. This is a **non-portable extension**. **x** : Pops a value off of the top of the stack. If it is a number, it is pushed back onto the stack. If it is a string, it is executed as a macro. This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. **\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register *r* are executed. For example, **0 1>a** will execute the contents of register **a**, and **1 0>a** will not. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **?** : Reads a line from the **stdin** and executes it. This is to allow macros to request input from users. **q** : During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. **Q** : Pops a value from the stack which must be non-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. **,** : Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the **Q** command, so the sequence **,Q** will make dc(1) exit. This is a **non-portable extension**. ## Status These commands query status of the stack or its top value. **Z** : Pops a value off of the stack. If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push **1** if the argument is **0** with no decimal places. If it is a string, pushes the number of characters the string has. **X** : Pops a value off of the stack. If it is a number, pushes the *scale* of the value onto the stack. If it is a string, pushes **0**. **u** : Pops one value off of the stack. If the value is a number, this pushes **1** onto the stack. Otherwise (if it is a string), it pushes **0**. This is a **non-portable extension**. **t** : Pops one value off of the stack. If the value is a string, this pushes **1** onto the stack. Otherwise (if it is a number), it pushes **0**. This is a **non-portable extension**. **z** : Pushes the current depth of the stack (before execution of this command) onto the stack. **y**_r_ : Pushes the current stack depth of the register *r* onto the main stack. Because each register has a depth of **1** (with the value **0** in the top item) when dc(1) starts, dc(1) requires that each register's stack must always have at least one item; dc(1) will give an error and reset otherwise (see the **RESET** section). This means that this command will never push **0**. This is a **non-portable extension**. ## Arrays These commands manipulate arrays. **:**_r_ : Pops the top two values off of the stack. The second value will be stored in the array *r* (see the **REGISTERS** section), indexed by the first value. **;**_r_ : Pops the value on top of the stack and uses it as an index into the array *r*. The selected value is then pushed onto the stack. **Y**_r_ : Pushes the length of the array *r* onto the stack. This is a **non-portable extension**. ## Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter **g**. Only the characters below are allowed after the character **g**; any other character produces a parse error (see the **ERRORS** section). **gl** : Pushes the line length set by **DC_LINE_LENGTH** (see the **ENVIRONMENT VARIABLES** section) onto the stack. **gx** : Pushes **1** onto the stack if extended register mode is on, **0** otherwise. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. **gz** : Pushes **0** onto the stack if the leading zero setting has not been enabled with the **-z** or **-\-leading-zeroes** options (see the **OPTIONS** section), non-zero otherwise. # REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (**0**) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. In non-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (**'\\n'**) and a left bracket (**'['**); it is a parse error for a newline or a left bracket to be used as a register name. ## Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. If extended register mode is enabled (**-x** or **-\-extended-register** command-line arguments are given), then normal single character registers are used *unless* the character immediately following a command that needs a register name is a space (according to **isspace()**) and not a newline (**'\\n'**). In that case, the register name is found according to the regex **\[a-z\]\[a-z0-9\_\]\*** (like bc(1) identifiers), and it is a parse error if the next non-space characters do not match that regex. # RESET When dc(1) encounters an error or a signal that it has a non-default handler for, it resets. This means that several things happen. -First, any macros that are executing are stopped and popped off the stack. -The behavior is not unlike that of exceptions in programming languages. Then -the execution point is set so that any code waiting to execute (after all +First, any macros that are executing are stopped and popped off the execution +stack. The behavior is not unlike that of exceptions in programming languages. +Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. +However, the stack of values is *not* cleared; in interactive mode, users can +inspect the stack and manipulate it. + Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the **EXIT STATUS** section), it asks for more input; otherwise, it exits with the appropriate return code. # PERFORMANCE Most dc(1) implementations use **char** types to calculate the value of **1** decimal digit at a time, but that can be slow. This dc(1) does something different. It uses large integers to calculate more than **1** decimal digit at a time. If built in a environment where **DC_LONG_BIT** (see the **LIMITS** section) is **64**, then each integer has **9** decimal digits. If built in an environment where **DC_LONG_BIT** is **32** then each integer has **4** decimal digits. This value (the number of decimal digits per large integer) is called **DC_BASE_DIGS**. In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of **DC_LONG_BIT**, but is always at least twice as large as the integer type used to store digits. # LIMITS The following are the limits on dc(1): **DC_LONG_BIT** : The number of bits in the **long** type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the **PERFORMANCE** section). **DC_BASE_DIGS** : The number of decimal digits per large integer (see the **PERFORMANCE** section). Depends on **DC_LONG_BIT**. **DC_BASE_POW** : The max decimal number that each large integer can store (see **DC_BASE_DIGS**) plus **1**. Depends on **DC_BASE_DIGS**. **DC_OVERFLOW_MAX** : The max number that the overflow type (see the **PERFORMANCE** section) can hold. Depends on **DC_LONG_BIT**. **DC_BASE_MAX** : The maximum output base. Set at **DC_BASE_POW**. **DC_DIM_MAX** : The maximum size of arrays. Set at **SIZE_MAX-1**. **DC_SCALE_MAX** : The maximum **scale**. Set at **DC_OVERFLOW_MAX-1**. **DC_STRING_MAX** : The maximum length of strings. Set at **DC_OVERFLOW_MAX-1**. **DC_NAME_MAX** : The maximum length of identifiers. Set at **DC_OVERFLOW_MAX-1**. **DC_NUM_MAX** : The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at **DC_OVERFLOW_MAX-1**. **DC_RAND_MAX** : The maximum integer (inclusive) returned by the **'** command, if dc(1). Set at **2\^DC_LONG_BIT-1**. Exponent : The maximum allowable exponent (positive or negative). Set at **DC_OVERFLOW_MAX**. Number of vars : The maximum number of vars/arrays. Set at **SIZE_MAX-1**. These limits are meant to be effectively non-existent; the limits are so large (at least on 64-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. # ENVIRONMENT VARIABLES As **non-portable extensions**, dc(1) recognizes the following environment variables: **DC_ENV_ARGS** : This is another way to give command-line arguments to dc(1). They should be in the same format as all other command-line arguments. These are always processed first, so any files given in **DC_ENV_ARGS** will be processed before arguments and files given on the command-line. This gives the user the ability to set up "standard" options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the **-e** option to set **scale** to a value other than **0**. The code that parses **DC_ENV_ARGS** will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string **"/home/gavin/some dc file.dc"** will be correctly parsed, but the string **"/home/gavin/some \"dc\" file.dc"** will include the backslashes. The quote parsing will handle either kind of quotes, **'** or **"**. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in **"some 'dc' file.dc"**, and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in **DC_ENV_ARGS** is not supported due to the complexity of the parsing, though such files are still supported on the command-line where the parsing is done by the shell. **DC_LINE_LENGTH** : If this environment variable exists and contains an integer that is greater than **1** and is less than **UINT16_MAX** (**2\^16-1**), dc(1) will output lines to that length, including the backslash newline combo. The default line length is **70**. The special value of **0** will disable line length checking and print numbers without regard to line length and without backslashes and newlines. **DC_SIGINT_RESET** : If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), then this environment variable has no effect because dc(1) exits on **SIGINT** when not in interactive mode. However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) reset on **SIGINT**, rather than exit, and zero makes dc(1) exit. If this environment variable exists and is *not* an integer, then dc(1) will exit on **SIGINT**. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_TTY_MODE** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_PROMPT** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) use a prompt, and zero or a non-integer makes dc(1) not use a prompt. If this environment variable does not exist and **DC_TTY_MODE** does, then the value of the **DC_TTY_MODE** environment variable is used. This environment variable and the **DC_TTY_MODE** environment variable override the default, which can be queried with the **-h** or **-\-help** options. **DC_EXPR_EXIT** : If any expressions or expression files are given on the command-line with **-e**, **-\-expression**, **-f**, or **-\-file**, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_DIGIT_CLAMP** : When parsing numbers and if this environment variable exists and contains an integer, a non-zero value makes dc(1) clamp digits that are greater than or equal to the current **ibase** so that all such digits are considered equal to the **ibase** minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the **ibase**. This never applies to single-digit numbers, as per the bc(1) standard (see the **STANDARDS** section). This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. # EXIT STATUS dc(1) returns the following exit statuses: **0** : No error. **1** : A math error occurred. This follows standard practice of using **1** for expected errors, since math errors will happen in the process of normal execution. Math errors include divide by **0**, taking the square root of a negative number, using a negative number as a bound for the pseudo-random number generator, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non-integer where an integer is required. Converting to a hardware integer happens for the second operand of the power (**\^**), places (**\@**), left shift (**H**), and right shift (**h**) operators. **2** : A parse error occurred. Parse errors include unexpected **EOF**, using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. **3** : A runtime error occurred. Runtime errors include assigning an invalid number to any global (**ibase**, **obase**, or **scale**), giving a bad expression to a **read()** call, calling **read()** inside of a **read()** call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. **4** : A fatal error occurred. Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command-line options. The exit status **4** is special; when a fatal error occurs, dc(1) always exits and returns **4**, no matter what mode dc(1) is in. The other statuses will only be returned when dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), since dc(1) resets its state (see the **RESET** section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the **-i** flag or **-\-interactive** option. These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the **-i** flag or **-\-interactive** option. # INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non-interactive mode. Interactive mode is turned on automatically when both **stdin** and **stdout** are hooked to a terminal, but the **-i** flag and **-\-interactive** option can turn it on in other situations. In interactive mode, dc(1) attempts to recover from errors (see the **RESET** section), and in normal execution, flushes **stdout** as soon as execution is done for the current input. dc(1) may also reset on **SIGINT** instead of exit, depending on the contents of, or default for, the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section). # TTY MODE If **stdin**, **stdout**, and **stderr** are all connected to a TTY, then "TTY mode" is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. If there is the environment variable **DC_TTY_MODE** in the environment (see the **ENVIRONMENT VARIABLES** section), then if that environment variable contains a non-zero integer, dc(1) will turn on TTY mode when **stdin**, **stdout**, and **stderr** are all connected to a TTY. If the **DC_TTY_MODE** environment variable exists but is *not* a non-zero integer, then dc(1) will not turn TTY mode on. If the environment variable **DC_TTY_MODE** does *not* exist, the default setting is used. The default setting can be queried with the **-h** or **-\-help** options. TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the **STANDARDS** section), and interactive mode requires only **stdin** and **stdout** to be connected to a terminal. ## Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: **DC_PROMPT** (see the **ENVIRONMENT VARIABLES** section). If the environment variable **DC_PROMPT** exists and is a non-zero integer, then the prompt is turned on when **stdin**, **stdout**, and **stderr** are connected to a TTY and the **-P** and **-\-no-prompt** options were not used. The read prompt will be turned on under the same conditions, except that the **-R** and **-\-no-read-prompt** options must also not be used. However, if **DC_PROMPT** does not exist, the prompt can be enabled or disabled with the **DC_TTY_MODE** environment variable, the **-P** and **-\-no-prompt** options, and the **-R** and **-\-no-read-prompt** options. See the **ENVIRONMENT VARIABLES** and **OPTIONS** sections for more details. # SIGNAL HANDLING Sending a **SIGINT** will cause dc(1) to do one of two things. If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), or the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section), or its default, is either not an integer or it is zero, dc(1) will exit. However, if dc(1) is in interactive mode, and the **DC_SIGINT_RESET** or its default is an integer and non-zero, then dc(1) will stop executing the current input and reset (see the **RESET** section) upon receiving a **SIGINT**. Note that "current input" can mean one of two things. If dc(1) is processing input from **stdin** in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from **stdin** if no other file exists. This means that if a **SIGINT** is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. **SIGTERM** and **SIGQUIT** cause dc(1) to clean up and exit, and it uses the default handler for all other signals. # LOCALES This dc(1) ships with support for adding error messages for different locales and thus, supports **LC_MESSAGES**. # SEE ALSO bc(1) # STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1-2017 (“POSIX.1-2017”) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . # BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . # AUTHOR Gavin D. Howard and contributors. diff --git a/manuals/dc/HN.1 b/manuals/dc/HN.1 index eae5cc516f71..c3f8c8ab1ff5 100644 --- a/manuals/dc/HN.1 +++ b/manuals/dc/HN.1 @@ -1,1665 +1,1668 @@ .\" .\" SPDX-License-Identifier: BSD-2-Clause .\" .\" Copyright (c) 2018-2024 Gavin D. Howard and contributors. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions are met: .\" .\" * Redistributions of source code must retain the above copyright notice, .\" this list of conditions and the following disclaimer. .\" .\" * Redistributions in binary form must reproduce the above copyright notice, .\" this list of conditions and the following disclaimer in the documentation .\" and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" .\" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE .\" LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF .\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS .\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN .\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) .\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE .\" POSSIBILITY OF SUCH DAMAGE. .\" -.TH "DC" "1" "January 2024" "Gavin D. Howard" "General Commands Manual" +.TH "DC" "1" "August 2024" "Gavin D. Howard" "General Commands Manual" .nh .ad l .SH Name dc \- arbitrary\-precision decimal reverse\-Polish notation calculator .SH SYNOPSIS \f[B]dc\f[R] [\f[B]\-cChiPRvVx\f[R]] [\f[B]\-\-version\f[R]] [\f[B]\-\-help\f[R]] [\f[B]\-\-digit\-clamp\f[R]] [\f[B]\-\-no\-digit\-clamp\f[R]] [\f[B]\-\-interactive\f[R]] [\f[B]\-\-no\-prompt\f[R]] [\f[B]\-\-no\-read\-prompt\f[R]] [\f[B]\-\-extended\-register\f[R]] [\f[B]\-e\f[R] \f[I]expr\f[R]] [\f[B]\-\-expression\f[R]=\f[I]expr\f[R]\&...] [\f[B]\-f\f[R] \f[I]file\f[R]\&...] [\f[B]\-\-file\f[R]=\f[I]file\f[R]\&...] [\f[I]file\f[R]\&...] [\f[B]\-I\f[R] \f[I]ibase\f[R]] [\f[B]\-\-ibase\f[R]=\f[I]ibase\f[R]] [\f[B]\-O\f[R] \f[I]obase\f[R]] [\f[B]\-\-obase\f[R]=\f[I]obase\f[R]] [\f[B]\-S\f[R] \f[I]scale\f[R]] [\f[B]\-\-scale\f[R]=\f[I]scale\f[R]] [\f[B]\-E\f[R] \f[I]seed\f[R]] [\f[B]\-\-seed\f[R]=\f[I]seed\f[R]] .SH DESCRIPTION dc(1) is an arbitrary\-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. .PP If no files are given on the command\-line, then dc(1) reads from \f[B]stdin\f[R] (see the \f[B]STDIN\f[R] section). Otherwise, those files are processed, and dc(1) will then exit. .PP If a user wants to set up a standard environment, they can use \f[B]DC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). For example, if a user wants the \f[B]scale\f[R] always set to \f[B]10\f[R], they can set \f[B]DC_ENV_ARGS\f[R] to \f[B]\-e 10k\f[R], and this dc(1) will always start with a \f[B]scale\f[R] of \f[B]10\f[R]. .SH OPTIONS The following are the options that dc(1) accepts. .TP \f[B]\-C\f[R], \f[B]\-\-no\-digit\-clamp\f[R] Disables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that the value added to a number from a digit is always that digit\[cq]s value multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-c\f[R] or \f[B]\-\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-c\f[R], \f[B]\-\-digit\-clamp\f[R] Enables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-C\f[R] or \f[B]\-\-no\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-E\f[R] \f[I]seed\f[R], \f[B]\-\-seed\f[R]=\f[I]seed\f[R] Sets the builtin variable \f[B]seed\f[R] to the value \f[I]seed\f[R] assuming that \f[I]seed\f[R] is in base 10. It is a fatal error if \f[I]seed\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-e\f[R] \f[I]expr\f[R], \f[B]\-\-expression\f[R]=\f[I]expr\f[R] Evaluates \f[I]expr\f[R]. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R], whether on the command\-line or in \f[B]DC_ENV_ARGS\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-f\f[R] \f[I]file\f[R], \f[B]\-\-file\f[R]=\f[I]file\f[R] Reads in \f[I]file\f[R] and evaluates it, line by line, as though it were read through \f[B]stdin\f[R]. If expressions are also given (see above), the expressions are evaluated in the order given. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-h\f[R], \f[B]\-\-help\f[R] Prints a usage message and exits. .TP \f[B]\-I\f[R] \f[I]ibase\f[R], \f[B]\-\-ibase\f[R]=\f[I]ibase\f[R] Sets the builtin variable \f[B]ibase\f[R] to the value \f[I]ibase\f[R] assuming that \f[I]ibase\f[R] is in base 10. It is a fatal error if \f[I]ibase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-i\f[R], \f[B]\-\-interactive\f[R] Forces interactive mode. (See the \f[B]INTERACTIVE MODE\f[R] section.) .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-L\f[R], \f[B]\-\-no\-line\-length\f[R] Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets \f[B]BC_LINE_LENGTH\f[R] to \f[B]0\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-O\f[R] \f[I]obase\f[R], \f[B]\-\-obase\f[R]=\f[I]obase\f[R] Sets the builtin variable \f[B]obase\f[R] to the value \f[I]obase\f[R] assuming that \f[I]obase\f[R] is in base 10. It is a fatal error if \f[I]obase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-P\f[R], \f[B]\-\-no\-prompt\f[R] Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]DC_ENV_ARGS\f[R]. .RS .PP These options override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-R\f[R], \f[B]\-\-no\-read\-prompt\f[R] Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]BC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. .RS .PP This option does not disable the regular prompt because the read prompt is only used when the \f[B]?\f[R] command is used. .PP These options \f[I]do\f[R] override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), but only for the read prompt. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-S\f[R] \f[I]scale\f[R], \f[B]\-\-scale\f[R]=\f[I]scale\f[R] Sets the builtin variable \f[B]scale\f[R] to the value \f[I]scale\f[R] assuming that \f[I]scale\f[R] is in base 10. It is a fatal error if \f[I]scale\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-v\f[R], \f[B]\-V\f[R], \f[B]\-\-version\f[R] Print the version information (copyright header) and exits. .TP \f[B]\-x\f[R] \f[B]\-\-extended\-register\f[R] Enables extended register mode. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-z\f[R], \f[B]\-\-leading\-zeroes\f[R] Makes dc(1) print all numbers greater than \f[B]\-1\f[R] and less than \f[B]1\f[R], and not equal to \f[B]0\f[R], with a leading zero. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .PP All long options are \f[B]non\-portable extensions\f[R]. .SH STDIN If no files are given on the command\-line and no files or expressions are given by the \f[B]\-f\f[R], \f[B]\-\-file\f[R], \f[B]\-e\f[R], or \f[B]\-\-expression\f[R] options, then dc(1) reads from \f[B]stdin\f[R]. .PP However, there is a caveat to this. .PP First, \f[B]stdin\f[R] is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. .SH STDOUT Any non\-error output is written to \f[B]stdout\f[R]. In addition, if history (see the \f[B]HISTORY\f[R] section) and the prompt (see the \f[B]TTY MODE\f[R] section) are enabled, both are output to \f[B]stdout\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stdout\f[R], so if \f[B]stdout\f[R] is closed, as in \f[B]dc >&\-\f[R], it will quit with an error. This is done so that dc(1) can report problems when \f[B]stdout\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stdout\f[R] to \f[B]/dev/null\f[R]. .SH STDERR Any error output is written to \f[B]stderr\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stderr\f[R], so if \f[B]stderr\f[R] is closed, as in \f[B]dc 2>&\-\f[R], it will quit with an error. This is done so that dc(1) can exit with an error code when \f[B]stderr\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stderr\f[R] to \f[B]/dev/null\f[R]. .SH SYNTAX Each item in the input source code, either a number (see the \f[B]NUMBERS\f[R] section) or a command (see the \f[B]COMMANDS\f[R] section), is processed and executed, in order. Input is processed immediately when entered. .PP \f[B]ibase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to interpret constant numbers. It is the \[lq]input\[rq] base, or the number base used for interpreting input numbers. \f[B]ibase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]ibase\f[R] is \f[B]16\f[R]. The min allowable value for \f[B]ibase\f[R] is \f[B]2\f[R]. The max allowable value for \f[B]ibase\f[R] can be queried in dc(1) programs with the \f[B]T\f[R] command. .PP \f[B]obase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to output results. It is the \[lq]output\[rq] base, or the number base used for outputting numbers. \f[B]obase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]obase\f[R] is \f[B]DC_BASE_MAX\f[R] and can be queried with the \f[B]U\f[R] command. The min allowable value for \f[B]obase\f[R] is \f[B]0\f[R]. If \f[B]obase\f[R] is \f[B]0\f[R], values are output in scientific notation, and if \f[B]obase\f[R] is \f[B]1\f[R], values are output in engineering notation. Otherwise, values are output in the specified base. .PP Outputting in scientific and engineering notations are \f[B]non\-portable extensions\f[R]. .PP The \f[I]scale\f[R] of an expression is the number of digits in the result of the expression right of the decimal point, and \f[B]scale\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that sets the precision of any operations (with exceptions). \f[B]scale\f[R] is initially \f[B]0\f[R]. \f[B]scale\f[R] cannot be negative. The max allowable value for \f[B]scale\f[R] can be queried in dc(1) programs with the \f[B]V\f[R] command. .PP \f[B]seed\f[R] is a register containing the current seed for the pseudo\-random number generator. If the current value of \f[B]seed\f[R] is queried and stored, then if it is assigned to \f[B]seed\f[R] later, the pseudo\-random number generator is guaranteed to produce the same sequence of pseudo\-random numbers that were generated after the value of \f[B]seed\f[R] was first queried. .PP Multiple values assigned to \f[B]seed\f[R] can produce the same sequence of pseudo\-random numbers. Likewise, when a value is assigned to \f[B]seed\f[R], it is not guaranteed that querying \f[B]seed\f[R] immediately after will return the same value. In addition, the value of \f[B]seed\f[R] will change after any call to the \f[B]\[cq]\f[R] command or the \f[B]\[lq]\f[R] command that does not get receive a value of \f[B]0\f[R] or \f[B]1\f[R]. The maximum integer returned by the \f[B]\[cq]\f[R] command can be queried with the \f[B]W\f[R] command. .PP \f[B]Note\f[R]: The values returned by the pseudo\-random number generator with the \f[B]\[cq]\f[R] and \f[B]\[lq]\f[R] commands are guaranteed to \f[B]NOT\f[R] be cryptographically secure. This is a consequence of using a seeded pseudo\-random number generator. However, they \f[I]are\f[R] guaranteed to be reproducible with identical \f[B]seed\f[R] values. This means that the pseudo\-random values from dc(1) should only be used where a reproducible stream of pseudo\-random numbers is \f[I]ESSENTIAL\f[R]. In any other case, use a non\-seeded pseudo\-random number generator. .PP The pseudo\-random number generator, \f[B]seed\f[R], and all associated operations are \f[B]non\-portable extensions\f[R]. .SS Comments Comments go from \f[B]#\f[R] until, and not including, the next newline. This is a \f[B]non\-portable extension\f[R]. .SH NUMBERS Numbers are strings made up of digits, uppercase letters up to \f[B]F\f[R], and at most \f[B]1\f[R] period for a radix. Numbers can have up to \f[B]DC_NUM_MAX\f[R] digits. Uppercase letters are equal to \f[B]9\f[R] plus their position in the alphabet (i.e., \f[B]A\f[R] equals \f[B]10\f[R], or \f[B]9+1\f[R]). .PP If a digit or letter makes no sense with the current value of \f[B]ibase\f[R] (i.e., they are greater than or equal to the current value of \f[B]ibase\f[R]), then the behavior depends on the existence of the \f[B]\-c\f[R]/\f[B]\-\-digit\-clamp\f[R] or \f[B]\-C\f[R]/\f[B]\-\-no\-digit\-clamp\f[R] options (see the \f[B]OPTIONS\f[R] section), the existence and setting of the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or the default, which can be queried with the \f[B]\-h\f[R]/\f[B]\-\-help\f[R] option. .PP If clamping is off, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are not changed. Instead, their given value is multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*A+3\[ha]0*B\f[R], which is \f[B]3\f[R] times \f[B]10\f[R] plus \f[B]11\f[R], or \f[B]41\f[R]. .PP If clamping is on, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are set to the value of the highest valid digit in \f[B]ibase\f[R] before being multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*2+3\[ha]0*2\f[R], which is \f[B]3\f[R] times \f[B]2\f[R] plus \f[B]2\f[R], or \f[B]8\f[R]. .PP There is one exception to clamping: single\-character numbers (i.e., \f[B]A\f[R] alone). Such numbers are never clamped and always take the value they would have in the highest possible \f[B]ibase\f[R]. This means that \f[B]A\f[R] alone always equals decimal \f[B]10\f[R] and \f[B]Z\f[R] alone always equals decimal \f[B]35\f[R]. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current \f[B]ibase\f[R] (with the \f[B]i\f[R] command) regardless of the current value of \f[B]ibase\f[R]. .PP If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for \f[B]A\f[R], use \f[B]0A\f[R]. .PP In addition, dc(1) accepts numbers in scientific notation. These have the form \f[B]e\f[R]. The exponent (the portion after the \f[B]e\f[R]) must be an integer. An example is \f[B]1.89237e9\f[R], which is equal to \f[B]1892370000\f[R]. Negative exponents are also allowed, so \f[B]4.2890e_3\f[R] is equal to \f[B]0.0042890\f[R]. .PP \f[B]WARNING\f[R]: Both the number and the exponent in scientific notation are interpreted according to the current \f[B]ibase\f[R], but the number is still multiplied by \f[B]10\[ha]exponent\f[R] regardless of the current \f[B]ibase\f[R]. For example, if \f[B]ibase\f[R] is \f[B]16\f[R] and dc(1) is given the number string \f[B]FFeA\f[R], the resulting decimal number will be \f[B]2550000000000\f[R], and if dc(1) is given the number string \f[B]10e_4\f[R], the resulting decimal number will be \f[B]0.0016\f[R]. .PP Accepting input as scientific notation is a \f[B]non\-portable extension\f[R]. .SH COMMANDS The valid commands are listed below. .SS Printing These commands are used for printing. .PP Note that both scientific notation and engineering notation are available for printing numbers. Scientific notation is activated by assigning \f[B]0\f[R] to \f[B]obase\f[R] using \f[B]0o\f[R], and engineering notation is activated by assigning \f[B]1\f[R] to \f[B]obase\f[R] using \f[B]1o\f[R]. To deactivate them, just assign a different value to \f[B]obase\f[R]. .PP Printing numbers in scientific notation and/or engineering notation is a \f[B]non\-portable extension\f[R]. .TP \f[B]p\f[R] Prints the value on top of the stack, whether number or string, and prints a newline after. .RS .PP This does not alter the stack. .RE .TP \f[B]n\f[R] Prints the value on top of the stack, whether number or string, and pops it off of the stack. .TP \f[B]P\f[R] Pops a value off the stack. .RS .PP If the value is a number, it is truncated and the absolute value of the result is printed as though \f[B]obase\f[R] is \f[B]256\f[R] and each digit is interpreted as an 8\-bit ASCII character, making it a byte stream. .PP If the value is a string, it is printed without a trailing newline. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]f\f[R] Prints the entire contents of the stack, in order from newest to oldest, without altering anything. .RS .PP Users should use this command when they get lost. .RE .SS Arithmetic These are the commands used for arithmetic. .TP \f[B]+\f[R] The top two values are popped off the stack, added, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]\-\f[R] The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]*\f[R] The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If \f[B]a\f[R] is the \f[I]scale\f[R] of the first expression and \f[B]b\f[R] is the \f[I]scale\f[R] of the second expression, the \f[I]scale\f[R] of the result is equal to \f[B]min(a+b,max(scale,a,b))\f[R] where \f[B]min()\f[R] and \f[B]max()\f[R] return the obvious values. .TP \f[B]/\f[R] The top two values are popped off the stack, divided, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]%\f[R] The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. .RS .PP Remaindering is equivalent to 1) Computing \f[B]a/b\f[R] to current \f[B]scale\f[R], and 2) Using the result of step 1 to calculate \f[B]a\-(a/b)*b\f[R] to \f[I]scale\f[R] \f[B]max(scale+scale(b),scale(a))\f[R]. .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]\[ti]\f[R] The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to \f[B]x y / x y %\f[R] except that \f[B]x\f[R] and \f[B]y\f[R] are only evaluated once. .RS .PP The first value popped off of the stack must be non\-zero. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[ha]\f[R] The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non\-zero. .RE .TP \f[B]v\f[R] The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The value popped off of the stack must be non\-negative. .RE .TP \f[B]_\f[R] If this command \f[I]immediately\f[R] precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. .RS .PP Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]b\f[R] The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]|\f[R] The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. .RS .PP The first value popped is used as the reduction modulus and must be an integer and non\-zero. The second value popped is used as the exponent and must be an integer and non\-negative. The third value popped is the base and must be an integer. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]$\f[R] The top value is popped off the stack and copied, and the copy is truncated and pushed onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[at]\f[R] The top two values are popped off the stack, and the precision of the second is set to the value of the first, whether by truncation or extension. .RS .PP The first value popped off of the stack must be an integer and non\-negative. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]H\f[R] The top two values are popped off the stack, and the second is shifted left (radix shifted right) to the value of the first. .RS .PP The first value popped off of the stack must be an integer and non\-negative. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]h\f[R] The top two values are popped off the stack, and the second is shifted right (radix shifted left) to the value of the first. .RS .PP The first value popped off of the stack must be an integer and non\-negative. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]G\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if they are equal, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]N\f[R] The top value is popped off of the stack, and if it a \f[B]0\f[R], a \f[B]1\f[R] is pushed; otherwise, a \f[B]0\f[R] is pushed. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B](\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]{\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B])\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]}\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]M\f[R] The top two values are popped off of the stack. If they are both non\-zero, a \f[B]1\f[R] is pushed onto the stack. If either of them is zero, or both of them are, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]&&\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]m\f[R] The top two values are popped off of the stack. If at least one of them is non\-zero, a \f[B]1\f[R] is pushed onto the stack. If both of them are zero, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]||\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Pseudo\-Random Number Generator dc(1) has a built\-in pseudo\-random number generator. These commands query the pseudo\-random number generator. (See Parameters for more information about the \f[B]seed\f[R] value that controls the pseudo\-random number generator.) .PP The pseudo\-random number generator is guaranteed to \f[B]NOT\f[R] be cryptographically secure. .TP \f[B]\[cq]\f[R] Generates an integer between 0 and \f[B]DC_RAND_MAX\f[R], inclusive (see the \f[B]LIMITS\f[R] section). .RS .PP The generated integer is made as unbiased as possible, subject to the limitations of the pseudo\-random number generator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[lq]\f[R] Pops a value off of the stack, which is used as an \f[B]exclusive\f[R] upper bound on the integer that will be generated. If the bound is negative or is a non\-integer, an error is raised, and dc(1) resets (see the \f[B]RESET\f[R] section) while \f[B]seed\f[R] remains unchanged. If the bound is larger than \f[B]DC_RAND_MAX\f[R], the higher bound is honored by generating several pseudo\-random integers, multiplying them by appropriate powers of \f[B]DC_RAND_MAX+1\f[R], and adding them together. Thus, the size of integer that can be generated with this command is unbounded. Using this command will change the value of \f[B]seed\f[R], unless the operand is \f[B]0\f[R] or \f[B]1\f[R]. In that case, \f[B]0\f[R] is pushed onto the stack, and \f[B]seed\f[R] is \f[I]not\f[R] changed. .RS .PP The generated integer is made as unbiased as possible, subject to the limitations of the pseudo\-random number generator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Stack Control These commands control the stack. .TP \f[B]c\f[R] Removes all items from (\[lq]clears\[rq]) the stack. .TP \f[B]d\f[R] Copies the item on top of the stack (\[lq]duplicates\[rq]) and pushes the copy onto the stack. .TP \f[B]r\f[R] Swaps (\[lq]reverses\[rq]) the two top items on the stack. .TP \f[B]R\f[R] Pops (\[lq]removes\[rq]) the top value from the stack. .SS Register Control These commands control registers (see the \f[B]REGISTERS\f[R] section). .TP \f[B]s\f[R]\f[I]r\f[R] Pops the value off the top of the stack and stores it into register \f[I]r\f[R]. .TP \f[B]l\f[R]\f[I]r\f[R] Copies the value in register \f[I]r\f[R] and pushes it onto the stack. This does not alter the contents of \f[I]r\f[R]. .TP \f[B]S\f[R]\f[I]r\f[R] Pops the value off the top of the (main) stack and pushes it onto the stack of register \f[I]r\f[R]. The previous value of the register becomes inaccessible. .TP \f[B]L\f[R]\f[I]r\f[R] Pops the value off the top of the stack for register \f[I]r\f[R] and push it onto the main stack. The previous value in the stack for register \f[I]r\f[R], if any, is now accessible via the \f[B]l\f[R]\f[I]r\f[R] command. .SS Parameters These commands control the values of \f[B]ibase\f[R], \f[B]obase\f[R], \f[B]scale\f[R], and \f[B]seed\f[R]. Also see the \f[B]SYNTAX\f[R] section. .TP \f[B]i\f[R] Pops the value off of the top of the stack and uses it to set \f[B]ibase\f[R], which must be between \f[B]2\f[R] and \f[B]16\f[R], inclusive. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]o\f[R] Pops the value off of the top of the stack and uses it to set \f[B]obase\f[R], which must be between \f[B]0\f[R] and \f[B]DC_BASE_MAX\f[R], inclusive (see the \f[B]LIMITS\f[R] section and the \f[B]NUMBERS\f[R] section). .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]k\f[R] Pops the value off of the top of the stack and uses it to set \f[B]scale\f[R], which must be non\-negative. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]j\f[R] Pops the value off of the top of the stack and uses it to set \f[B]seed\f[R]. The meaning of \f[B]seed\f[R] is dependent on the current pseudo\-random number generator but is guaranteed to not change except for new major versions. .RS .PP The \f[I]scale\f[R] and sign of the value may be significant. .PP If a previously used \f[B]seed\f[R] value is used again, the pseudo\-random number generator is guaranteed to produce the same sequence of pseudo\-random numbers as it did when the \f[B]seed\f[R] value was previously used. .PP The exact value assigned to \f[B]seed\f[R] is not guaranteed to be returned if the \f[B]J\f[R] command is used. However, if \f[B]seed\f[R] \f[I]does\f[R] return a different value, both values, when assigned to \f[B]seed\f[R], are guaranteed to produce the same sequence of pseudo\-random numbers. This means that certain values assigned to \f[B]seed\f[R] will not produce unique sequences of pseudo\-random numbers. .PP There is no limit to the length (number of significant decimal digits) or \f[I]scale\f[R] of the value that can be assigned to \f[B]seed\f[R]. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]I\f[R] Pushes the current value of \f[B]ibase\f[R] onto the main stack. .TP \f[B]O\f[R] Pushes the current value of \f[B]obase\f[R] onto the main stack. .TP \f[B]K\f[R] Pushes the current value of \f[B]scale\f[R] onto the main stack. .TP \f[B]J\f[R] Pushes the current value of \f[B]seed\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]T\f[R] Pushes the maximum allowable value of \f[B]ibase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]U\f[R] Pushes the maximum allowable value of \f[B]obase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]V\f[R] Pushes the maximum allowable value of \f[B]scale\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]W\f[R] Pushes the maximum (inclusive) integer that can be generated with the \f[B]\[cq]\f[R] pseudo\-random number generator command. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Strings The following commands control strings. .PP dc(1) can work with both numbers and strings, and registers (see the \f[B]REGISTERS\f[R] section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. .PP While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. .PP Strings can also be executed as macros. For example, if the string \f[B][1pR]\f[R] is executed as a macro, then the code \f[B]1pR\f[R] is executed, meaning that the \f[B]1\f[R] will be printed with a newline after and then popped from the stack. .TP \f[B][\f[R]\f[I]characters\f[R]\f[B]]\f[R] Makes a string containing \f[I]characters\f[R] and pushes it onto the stack. .RS .PP If there are brackets (\f[B][\f[R] and \f[B]]\f[R]) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (\f[B]\[rs]\f[R]) character. .PP If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. .RE .TP \f[B]a\f[R] The value on top of the stack is popped. .RS .PP If it is a number, it is truncated and its absolute value is taken. The result mod \f[B]256\f[R] is calculated. If that result is \f[B]0\f[R], push an empty string; otherwise, push a one\-character string where the character is the result of the mod interpreted as an ASCII character. .PP If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one\-character string. The new string is then pushed onto the stack. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]x\f[R] Pops a value off of the top of the stack. .RS .PP If it is a number, it is pushed back onto the stack. .PP If it is a string, it is executed as a macro. .PP This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. .RE .TP \f[B]>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP For example, \f[B]0 1>a\f[R] will execute the contents of register \f[B]a\f[R], and \f[B]1 0>a\f[R] will not. .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]?\f[R] Reads a line from the \f[B]stdin\f[R] and executes it. This is to allow macros to request input from users. .TP \f[B]q\f[R] During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. .TP \f[B]Q\f[R] Pops a value from the stack which must be non\-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. .TP \f[B],\f[R] Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the \f[B]Q\f[R] command, so the sequence \f[B],Q\f[R] will make dc(1) exit. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Status These commands query status of the stack or its top value. .TP \f[B]Z\f[R] Pops a value off of the stack. .RS .PP If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push \f[B]1\f[R] if the argument is \f[B]0\f[R] with no decimal places. .PP If it is a string, pushes the number of characters the string has. .RE .TP \f[B]X\f[R] Pops a value off of the stack. .RS .PP If it is a number, pushes the \f[I]scale\f[R] of the value onto the stack. .PP If it is a string, pushes \f[B]0\f[R]. .RE .TP \f[B]u\f[R] Pops one value off of the stack. If the value is a number, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a string), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]t\f[R] Pops one value off of the stack. If the value is a string, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a number), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]z\f[R] Pushes the current depth of the stack (before execution of this command) onto the stack. .TP \f[B]y\f[R]\f[I]r\f[R] Pushes the current stack depth of the register \f[I]r\f[R] onto the main stack. .RS .PP Because each register has a depth of \f[B]1\f[R] (with the value \f[B]0\f[R] in the top item) when dc(1) starts, dc(1) requires that each register\[cq]s stack must always have at least one item; dc(1) will give an error and reset otherwise (see the \f[B]RESET\f[R] section). This means that this command will never push \f[B]0\f[R]. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Arrays These commands manipulate arrays. .TP \f[B]:\f[R]\f[I]r\f[R] Pops the top two values off of the stack. The second value will be stored in the array \f[I]r\f[R] (see the \f[B]REGISTERS\f[R] section), indexed by the first value. .TP \f[B];\f[R]\f[I]r\f[R] Pops the value on top of the stack and uses it as an index into the array \f[I]r\f[R]. The selected value is then pushed onto the stack. .TP \f[B]Y\f[R]\f[I]r\f[R] Pushes the length of the array \f[I]r\f[R] onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter \f[B]g\f[R]. Only the characters below are allowed after the character \f[B]g\f[R]; any other character produces a parse error (see the \f[B]ERRORS\f[R] section). .TP \f[B]gl\f[R] Pushes the line length set by \f[B]DC_LINE_LENGTH\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) onto the stack. .TP \f[B]gx\f[R] Pushes \f[B]1\f[R] onto the stack if extended register mode is on, \f[B]0\f[R] otherwise. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .TP \f[B]gz\f[R] Pushes \f[B]0\f[R] onto the stack if the leading zero setting has not been enabled with the \f[B]\-z\f[R] or \f[B]\-\-leading\-zeroes\f[R] options (see the \f[B]OPTIONS\f[R] section), non\-zero otherwise. .SH REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) .PP Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (\f[B]0\f[R]) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. .PP In non\-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (\f[B]`\[rs]n'\f[R]) and a left bracket (\f[B]`['\f[R]); it is a parse error for a newline or a left bracket to be used as a register name. .SS Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. .PP If extended register mode is enabled (\f[B]\-x\f[R] or \f[B]\-\-extended\-register\f[R] command\-line arguments are given), then normal single character registers are used \f[I]unless\f[R] the character immediately following a command that needs a register name is a space (according to \f[B]isspace()\f[R]) and not a newline (\f[B]`\[rs]n'\f[R]). .PP In that case, the register name is found according to the regex \f[B][a\-z][a\-z0\-9_]*\f[R] (like bc(1) identifiers), and it is a parse error if the next non\-space characters do not match that regex. .SH RESET When dc(1) encounters an error or a signal that it has a non\-default handler for, it resets. This means that several things happen. .PP First, any macros that are executing are stopped and popped off the -stack. +execution stack. The behavior is not unlike that of exceptions in programming languages. Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. .PP +However, the stack of values is \f[I]not\f[R] cleared; in interactive +mode, users can inspect the stack and manipulate it. +.PP Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the \f[B]EXIT STATUS\f[R] section), it asks for more input; otherwise, it exits with the appropriate return code. .SH PERFORMANCE Most dc(1) implementations use \f[B]char\f[R] types to calculate the value of \f[B]1\f[R] decimal digit at a time, but that can be slow. This dc(1) does something different. .PP It uses large integers to calculate more than \f[B]1\f[R] decimal digit at a time. If built in a environment where \f[B]DC_LONG_BIT\f[R] (see the \f[B]LIMITS\f[R] section) is \f[B]64\f[R], then each integer has \f[B]9\f[R] decimal digits. If built in an environment where \f[B]DC_LONG_BIT\f[R] is \f[B]32\f[R] then each integer has \f[B]4\f[R] decimal digits. This value (the number of decimal digits per large integer) is called \f[B]DC_BASE_DIGS\f[R]. .PP In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of \f[B]DC_LONG_BIT\f[R], but is always at least twice as large as the integer type used to store digits. .SH LIMITS The following are the limits on dc(1): .TP \f[B]DC_LONG_BIT\f[R] The number of bits in the \f[B]long\f[R] type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the \f[B]PERFORMANCE\f[R] section). .TP \f[B]DC_BASE_DIGS\f[R] The number of decimal digits per large integer (see the \f[B]PERFORMANCE\f[R] section). Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_POW\f[R] The max decimal number that each large integer can store (see \f[B]DC_BASE_DIGS\f[R]) plus \f[B]1\f[R]. Depends on \f[B]DC_BASE_DIGS\f[R]. .TP \f[B]DC_OVERFLOW_MAX\f[R] The max number that the overflow type (see the \f[B]PERFORMANCE\f[R] section) can hold. Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_MAX\f[R] The maximum output base. Set at \f[B]DC_BASE_POW\f[R]. .TP \f[B]DC_DIM_MAX\f[R] The maximum size of arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .TP \f[B]DC_SCALE_MAX\f[R] The maximum \f[B]scale\f[R]. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_STRING_MAX\f[R] The maximum length of strings. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NAME_MAX\f[R] The maximum length of identifiers. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NUM_MAX\f[R] The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_RAND_MAX\f[R] The maximum integer (inclusive) returned by the \f[B]\[cq]\f[R] command, if dc(1). Set at \f[B]2\[ha]DC_LONG_BIT\-1\f[R]. .TP Exponent The maximum allowable exponent (positive or negative). Set at \f[B]DC_OVERFLOW_MAX\f[R]. .TP Number of vars The maximum number of vars/arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .PP These limits are meant to be effectively non\-existent; the limits are so large (at least on 64\-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. .SH ENVIRONMENT VARIABLES As \f[B]non\-portable extensions\f[R], dc(1) recognizes the following environment variables: .TP \f[B]DC_ENV_ARGS\f[R] This is another way to give command\-line arguments to dc(1). They should be in the same format as all other command\-line arguments. These are always processed first, so any files given in \f[B]DC_ENV_ARGS\f[R] will be processed before arguments and files given on the command\-line. This gives the user the ability to set up \[lq]standard\[rq] options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the \f[B]\-e\f[R] option to set \f[B]scale\f[R] to a value other than \f[B]0\f[R]. .RS .PP The code that parses \f[B]DC_ENV_ARGS\f[R] will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string \f[B]\[lq]/home/gavin/some dc file.dc\[rq]\f[R] will be correctly parsed, but the string \f[B]\[lq]/home/gavin/some \[dq]dc\[dq] file.dc\[rq]\f[R] will include the backslashes. .PP The quote parsing will handle either kind of quotes, \f[B]\[cq]\f[R] or \f[B]\[lq]\f[R]. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in \f[B]\[lq]some `dc' file.dc\[rq]\f[R], and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in \f[B]DC_ENV_ARGS\f[R] is not supported due to the complexity of the parsing, though such files are still supported on the command\-line where the parsing is done by the shell. .RE .TP \f[B]DC_LINE_LENGTH\f[R] If this environment variable exists and contains an integer that is greater than \f[B]1\f[R] and is less than \f[B]UINT16_MAX\f[R] (\f[B]2\[ha]16\-1\f[R]), dc(1) will output lines to that length, including the backslash newline combo. The default line length is \f[B]70\f[R]. .RS .PP The special value of \f[B]0\f[R] will disable line length checking and print numbers without regard to line length and without backslashes and newlines. .RE .TP \f[B]DC_SIGINT_RESET\f[R] If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), then this environment variable has no effect because dc(1) exits on \f[B]SIGINT\f[R] when not in interactive mode. .RS .PP However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) reset on \f[B]SIGINT\f[R], rather than exit, and zero makes dc(1) exit. If this environment variable exists and is \f[I]not\f[R] an integer, then dc(1) will exit on \f[B]SIGINT\f[R]. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_TTY_MODE\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non\-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_PROMPT\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) use a prompt, and zero or a non\-integer makes dc(1) not use a prompt. If this environment variable does not exist and \f[B]DC_TTY_MODE\f[R] does, then the value of the \f[B]DC_TTY_MODE\f[R] environment variable is used. .PP This environment variable and the \f[B]DC_TTY_MODE\f[R] environment variable override the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_EXPR_EXIT\f[R] If any expressions or expression files are given on the command\-line with \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R], then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. .RS .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_DIGIT_CLAMP\f[R] When parsing numbers and if this environment variable exists and contains an integer, a non\-zero value makes dc(1) clamp digits that are greater than or equal to the current \f[B]ibase\f[R] so that all such digits are considered equal to the \f[B]ibase\f[R] minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the \f[B]ibase\f[R]. .RS .PP This never applies to single\-digit numbers, as per the bc(1) standard (see the \f[B]STANDARDS\f[R] section). .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .SH EXIT STATUS dc(1) returns the following exit statuses: .TP \f[B]0\f[R] No error. .TP \f[B]1\f[R] A math error occurred. This follows standard practice of using \f[B]1\f[R] for expected errors, since math errors will happen in the process of normal execution. .RS .PP Math errors include divide by \f[B]0\f[R], taking the square root of a negative number, using a negative number as a bound for the pseudo\-random number generator, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non\-integer where an integer is required. .PP Converting to a hardware integer happens for the second operand of the power (\f[B]\[ha]\f[R]), places (\f[B]\[at]\f[R]), left shift (\f[B]H\f[R]), and right shift (\f[B]h\f[R]) operators. .RE .TP \f[B]2\f[R] A parse error occurred. .RS .PP Parse errors include unexpected \f[B]EOF\f[R], using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. .RE .TP \f[B]3\f[R] A runtime error occurred. .RS .PP Runtime errors include assigning an invalid number to any global (\f[B]ibase\f[R], \f[B]obase\f[R], or \f[B]scale\f[R]), giving a bad expression to a \f[B]read()\f[R] call, calling \f[B]read()\f[R] inside of a \f[B]read()\f[R] call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. .RE .TP \f[B]4\f[R] A fatal error occurred. .RS .PP Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command\-line options. .RE .PP The exit status \f[B]4\f[R] is special; when a fatal error occurs, dc(1) always exits and returns \f[B]4\f[R], no matter what mode dc(1) is in. .PP The other statuses will only be returned when dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), since dc(1) resets its state (see the \f[B]RESET\f[R] section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .PP These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .SH INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non\-interactive mode. Interactive mode is turned on automatically when both \f[B]stdin\f[R] and \f[B]stdout\f[R] are hooked to a terminal, but the \f[B]\-i\f[R] flag and \f[B]\-\-interactive\f[R] option can turn it on in other situations. .PP In interactive mode, dc(1) attempts to recover from errors (see the \f[B]RESET\f[R] section), and in normal execution, flushes \f[B]stdout\f[R] as soon as execution is done for the current input. dc(1) may also reset on \f[B]SIGINT\f[R] instead of exit, depending on the contents of, or default for, the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .SH TTY MODE If \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY, then \[lq]TTY mode\[rq] is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. .PP If there is the environment variable \f[B]DC_TTY_MODE\f[R] in the environment (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then if that environment variable contains a non\-zero integer, dc(1) will turn on TTY mode when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY. If the \f[B]DC_TTY_MODE\f[R] environment variable exists but is \f[I]not\f[R] a non\-zero integer, then dc(1) will not turn TTY mode on. .PP If the environment variable \f[B]DC_TTY_MODE\f[R] does \f[I]not\f[R] exist, the default setting is used. The default setting can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the \f[B]STANDARDS\f[R] section), and interactive mode requires only \f[B]stdin\f[R] and \f[B]stdout\f[R] to be connected to a terminal. .SS Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: \f[B]DC_PROMPT\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP If the environment variable \f[B]DC_PROMPT\f[R] exists and is a non\-zero integer, then the prompt is turned on when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are connected to a TTY and the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options were not used. The read prompt will be turned on under the same conditions, except that the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options must also not be used. .PP However, if \f[B]DC_PROMPT\f[R] does not exist, the prompt can be enabled or disabled with the \f[B]DC_TTY_MODE\f[R] environment variable, the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options, and the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options. See the \f[B]ENVIRONMENT VARIABLES\f[R] and \f[B]OPTIONS\f[R] sections for more details. .SH SIGNAL HANDLING Sending a \f[B]SIGINT\f[R] will cause dc(1) to do one of two things. .PP If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), or the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or its default, is either not an integer or it is zero, dc(1) will exit. .PP However, if dc(1) is in interactive mode, and the \f[B]DC_SIGINT_RESET\f[R] or its default is an integer and non\-zero, then dc(1) will stop executing the current input and reset (see the \f[B]RESET\f[R] section) upon receiving a \f[B]SIGINT\f[R]. .PP Note that \[lq]current input\[rq] can mean one of two things. If dc(1) is processing input from \f[B]stdin\f[R] in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from \f[B]stdin\f[R] if no other file exists. .PP This means that if a \f[B]SIGINT\f[R] is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. .PP \f[B]SIGTERM\f[R] and \f[B]SIGQUIT\f[R] cause dc(1) to clean up and exit, and it uses the default handler for all other signals. .SH SEE ALSO bc(1) .SH STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1\-2017 (\[lq]POSIX.1\-2017\[rq]) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . .SH BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . .SH AUTHOR Gavin D. Howard \c .MT gavin@gavinhoward.com .ME \c \ and contributors. diff --git a/manuals/dc/HN.1.md b/manuals/dc/HN.1.md index b34d80325f73..28b9dadd4b4f 100644 --- a/manuals/dc/HN.1.md +++ b/manuals/dc/HN.1.md @@ -1,1498 +1,1501 @@ # Name dc - arbitrary-precision decimal reverse-Polish notation calculator # SYNOPSIS **dc** [**-cChiPRvVx**] [**-\-version**] [**-\-help**] [**-\-digit-clamp**] [**-\-no-digit-clamp**] [**-\-interactive**] [**-\-no-prompt**] [**-\-no-read-prompt**] [**-\-extended-register**] [**-e** *expr*] [**-\-expression**=*expr*...] [**-f** *file*...] [**-\-file**=*file*...] [*file*...] [**-I** *ibase*] [**-\-ibase**=*ibase*] [**-O** *obase*] [**-\-obase**=*obase*] [**-S** *scale*] [**-\-scale**=*scale*] [**-E** *seed*] [**-\-seed**=*seed*] # DESCRIPTION dc(1) is an arbitrary-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. If no files are given on the command-line, then dc(1) reads from **stdin** (see the **STDIN** section). Otherwise, those files are processed, and dc(1) will then exit. If a user wants to set up a standard environment, they can use **DC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). For example, if a user wants the **scale** always set to **10**, they can set **DC_ENV_ARGS** to **-e 10k**, and this dc(1) will always start with a **scale** of **10**. # OPTIONS The following are the options that dc(1) accepts. **-C**, **-\-no-digit-clamp** : Disables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that the value added to a number from a digit is always that digit's value multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-c** or **-\-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-c**, **-\-digit-clamp** : Enables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-C** or **-\-no-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-E** *seed*, **-\-seed**=*seed* : Sets the builtin variable **seed** to the value *seed* assuming that *seed* is in base 10. It is a fatal error if *seed* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-e** *expr*, **-\-expression**=*expr* : Evaluates *expr*. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**, whether on the command-line or in **DC_ENV_ARGS**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-f** *file*, **-\-file**=*file* : Reads in *file* and evaluates it, line by line, as though it were read through **stdin**. If expressions are also given (see above), the expressions are evaluated in the order given. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-h**, **-\-help** : Prints a usage message and exits. **-I** *ibase*, **-\-ibase**=*ibase* : Sets the builtin variable **ibase** to the value *ibase* assuming that *ibase* is in base 10. It is a fatal error if *ibase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-i**, **-\-interactive** : Forces interactive mode. (See the **INTERACTIVE MODE** section.) This is a **non-portable extension**. **-L**, **-\-no-line-length** : Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets **BC_LINE_LENGTH** to **0** (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-O** *obase*, **-\-obase**=*obase* : Sets the builtin variable **obase** to the value *obase* assuming that *obase* is in base 10. It is a fatal error if *obase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-P**, **-\-no-prompt** : Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in **DC_ENV_ARGS**. These options override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-R**, **-\-no-read-prompt** : Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in **BC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. This option does not disable the regular prompt because the read prompt is only used when the **?** command is used. These options *do* override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section), but only for the read prompt. This is a **non-portable extension**. **-S** *scale*, **-\-scale**=*scale* : Sets the builtin variable **scale** to the value *scale* assuming that *scale* is in base 10. It is a fatal error if *scale* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-v**, **-V**, **-\-version** : Print the version information (copyright header) and exits. **-x** **-\-extended-register** : Enables extended register mode. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. This is a **non-portable extension**. **-z**, **-\-leading-zeroes** : Makes dc(1) print all numbers greater than **-1** and less than **1**, and not equal to **0**, with a leading zero. This is a **non-portable extension**. All long options are **non-portable extensions**. # STDIN If no files are given on the command-line and no files or expressions are given by the **-f**, **-\-file**, **-e**, or **-\-expression** options, then dc(1) reads from **stdin**. However, there is a caveat to this. First, **stdin** is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. # STDOUT Any non-error output is written to **stdout**. In addition, if history (see the **HISTORY** section) and the prompt (see the **TTY MODE** section) are enabled, both are output to **stdout**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stdout**, so if **stdout** is closed, as in **dc >&-**, it will quit with an error. This is done so that dc(1) can report problems when **stdout** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stdout** to **/dev/null**. # STDERR Any error output is written to **stderr**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stderr**, so if **stderr** is closed, as in **dc 2>&-**, it will quit with an error. This is done so that dc(1) can exit with an error code when **stderr** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stderr** to **/dev/null**. # SYNTAX Each item in the input source code, either a number (see the **NUMBERS** section) or a command (see the **COMMANDS** section), is processed and executed, in order. Input is processed immediately when entered. **ibase** is a register (see the **REGISTERS** section) that determines how to interpret constant numbers. It is the "input" base, or the number base used for interpreting input numbers. **ibase** is initially **10**. The max allowable value for **ibase** is **16**. The min allowable value for **ibase** is **2**. The max allowable value for **ibase** can be queried in dc(1) programs with the **T** command. **obase** is a register (see the **REGISTERS** section) that determines how to output results. It is the "output" base, or the number base used for outputting numbers. **obase** is initially **10**. The max allowable value for **obase** is **DC_BASE_MAX** and can be queried with the **U** command. The min allowable value for **obase** is **0**. If **obase** is **0**, values are output in scientific notation, and if **obase** is **1**, values are output in engineering notation. Otherwise, values are output in the specified base. Outputting in scientific and engineering notations are **non-portable extensions**. The *scale* of an expression is the number of digits in the result of the expression right of the decimal point, and **scale** is a register (see the **REGISTERS** section) that sets the precision of any operations (with exceptions). **scale** is initially **0**. **scale** cannot be negative. The max allowable value for **scale** can be queried in dc(1) programs with the **V** command. **seed** is a register containing the current seed for the pseudo-random number generator. If the current value of **seed** is queried and stored, then if it is assigned to **seed** later, the pseudo-random number generator is guaranteed to produce the same sequence of pseudo-random numbers that were generated after the value of **seed** was first queried. Multiple values assigned to **seed** can produce the same sequence of pseudo-random numbers. Likewise, when a value is assigned to **seed**, it is not guaranteed that querying **seed** immediately after will return the same value. In addition, the value of **seed** will change after any call to the **'** command or the **"** command that does not get receive a value of **0** or **1**. The maximum integer returned by the **'** command can be queried with the **W** command. **Note**: The values returned by the pseudo-random number generator with the **'** and **"** commands are guaranteed to **NOT** be cryptographically secure. This is a consequence of using a seeded pseudo-random number generator. However, they *are* guaranteed to be reproducible with identical **seed** values. This means that the pseudo-random values from dc(1) should only be used where a reproducible stream of pseudo-random numbers is *ESSENTIAL*. In any other case, use a non-seeded pseudo-random number generator. The pseudo-random number generator, **seed**, and all associated operations are **non-portable extensions**. ## Comments Comments go from **#** until, and not including, the next newline. This is a **non-portable extension**. # NUMBERS Numbers are strings made up of digits, uppercase letters up to **F**, and at most **1** period for a radix. Numbers can have up to **DC_NUM_MAX** digits. Uppercase letters are equal to **9** plus their position in the alphabet (i.e., **A** equals **10**, or **9+1**). If a digit or letter makes no sense with the current value of **ibase** (i.e., they are greater than or equal to the current value of **ibase**), then the behavior depends on the existence of the **-c**/**-\-digit-clamp** or **-C**/**-\-no-digit-clamp** options (see the **OPTIONS** section), the existence and setting of the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section), or the default, which can be queried with the **-h**/**-\-help** option. If clamping is off, then digits or letters that are greater than or equal to the current value of **ibase** are not changed. Instead, their given value is multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*A+3\^0\*B**, which is **3** times **10** plus **11**, or **41**. If clamping is on, then digits or letters that are greater than or equal to the current value of **ibase** are set to the value of the highest valid digit in **ibase** before being multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*2+3\^0\*2**, which is **3** times **2** plus **2**, or **8**. There is one exception to clamping: single-character numbers (i.e., **A** alone). Such numbers are never clamped and always take the value they would have in the highest possible **ibase**. This means that **A** alone always equals decimal **10** and **Z** alone always equals decimal **35**. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current **ibase** (with the **i** command) regardless of the current value of **ibase**. If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for **A**, use **0A**. In addition, dc(1) accepts numbers in scientific notation. These have the form **\e\**. The exponent (the portion after the **e**) must be an integer. An example is **1.89237e9**, which is equal to **1892370000**. Negative exponents are also allowed, so **4.2890e_3** is equal to **0.0042890**. **WARNING**: Both the number and the exponent in scientific notation are interpreted according to the current **ibase**, but the number is still multiplied by **10\^exponent** regardless of the current **ibase**. For example, if **ibase** is **16** and dc(1) is given the number string **FFeA**, the resulting decimal number will be **2550000000000**, and if dc(1) is given the number string **10e_4**, the resulting decimal number will be **0.0016**. Accepting input as scientific notation is a **non-portable extension**. # COMMANDS The valid commands are listed below. ## Printing These commands are used for printing. Note that both scientific notation and engineering notation are available for printing numbers. Scientific notation is activated by assigning **0** to **obase** using **0o**, and engineering notation is activated by assigning **1** to **obase** using **1o**. To deactivate them, just assign a different value to **obase**. Printing numbers in scientific notation and/or engineering notation is a **non-portable extension**. **p** : Prints the value on top of the stack, whether number or string, and prints a newline after. This does not alter the stack. **n** : Prints the value on top of the stack, whether number or string, and pops it off of the stack. **P** : Pops a value off the stack. If the value is a number, it is truncated and the absolute value of the result is printed as though **obase** is **256** and each digit is interpreted as an 8-bit ASCII character, making it a byte stream. If the value is a string, it is printed without a trailing newline. This is a **non-portable extension**. **f** : Prints the entire contents of the stack, in order from newest to oldest, without altering anything. Users should use this command when they get lost. ## Arithmetic These are the commands used for arithmetic. **+** : The top two values are popped off the stack, added, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **-** : The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **\*** : The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If **a** is the *scale* of the first expression and **b** is the *scale* of the second expression, the *scale* of the result is equal to **min(a+b,max(scale,a,b))** where **min()** and **max()** return the obvious values. **/** : The top two values are popped off the stack, divided, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be non-zero. **%** : The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. Remaindering is equivalent to 1) Computing **a/b** to current **scale**, and 2) Using the result of step 1 to calculate **a-(a/b)\*b** to *scale* **max(scale+scale(b),scale(a))**. The first value popped off of the stack must be non-zero. **~** : The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to **x y / x y %** except that **x** and **y** are only evaluated once. The first value popped off of the stack must be non-zero. This is a **non-portable extension**. **\^** : The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non-zero. **v** : The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The value popped off of the stack must be non-negative. **\_** : If this command *immediately* precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a **non-portable extension**. **b** : The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. This is a **non-portable extension**. **|** : The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. The first value popped is used as the reduction modulus and must be an integer and non-zero. The second value popped is used as the exponent and must be an integer and non-negative. The third value popped is the base and must be an integer. This is a **non-portable extension**. **\$** : The top value is popped off the stack and copied, and the copy is truncated and pushed onto the stack. This is a **non-portable extension**. **\@** : The top two values are popped off the stack, and the precision of the second is set to the value of the first, whether by truncation or extension. The first value popped off of the stack must be an integer and non-negative. This is a **non-portable extension**. **H** : The top two values are popped off the stack, and the second is shifted left (radix shifted right) to the value of the first. The first value popped off of the stack must be an integer and non-negative. This is a **non-portable extension**. **h** : The top two values are popped off the stack, and the second is shifted right (radix shifted left) to the value of the first. The first value popped off of the stack must be an integer and non-negative. This is a **non-portable extension**. **G** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if they are equal, or **0** otherwise. This is a **non-portable extension**. **N** : The top value is popped off of the stack, and if it a **0**, a **1** is pushed; otherwise, a **0** is pushed. This is a **non-portable extension**. **(** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than the second, or **0** otherwise. This is a **non-portable extension**. **{** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **)** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than the second, or **0** otherwise. This is a **non-portable extension**. **}** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **M** : The top two values are popped off of the stack. If they are both non-zero, a **1** is pushed onto the stack. If either of them is zero, or both of them are, then a **0** is pushed onto the stack. This is like the **&&** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. **m** : The top two values are popped off of the stack. If at least one of them is non-zero, a **1** is pushed onto the stack. If both of them are zero, then a **0** is pushed onto the stack. This is like the **||** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. ## Pseudo-Random Number Generator dc(1) has a built-in pseudo-random number generator. These commands query the pseudo-random number generator. (See Parameters for more information about the **seed** value that controls the pseudo-random number generator.) The pseudo-random number generator is guaranteed to **NOT** be cryptographically secure. **'** : Generates an integer between 0 and **DC_RAND_MAX**, inclusive (see the **LIMITS** section). The generated integer is made as unbiased as possible, subject to the limitations of the pseudo-random number generator. This is a **non-portable extension**. **"** : Pops a value off of the stack, which is used as an **exclusive** upper bound on the integer that will be generated. If the bound is negative or is a non-integer, an error is raised, and dc(1) resets (see the **RESET** section) while **seed** remains unchanged. If the bound is larger than **DC_RAND_MAX**, the higher bound is honored by generating several pseudo-random integers, multiplying them by appropriate powers of **DC_RAND_MAX+1**, and adding them together. Thus, the size of integer that can be generated with this command is unbounded. Using this command will change the value of **seed**, unless the operand is **0** or **1**. In that case, **0** is pushed onto the stack, and **seed** is *not* changed. The generated integer is made as unbiased as possible, subject to the limitations of the pseudo-random number generator. This is a **non-portable extension**. ## Stack Control These commands control the stack. **c** : Removes all items from ("clears") the stack. **d** : Copies the item on top of the stack ("duplicates") and pushes the copy onto the stack. **r** : Swaps ("reverses") the two top items on the stack. **R** : Pops ("removes") the top value from the stack. ## Register Control These commands control registers (see the **REGISTERS** section). **s**_r_ : Pops the value off the top of the stack and stores it into register *r*. **l**_r_ : Copies the value in register *r* and pushes it onto the stack. This does not alter the contents of *r*. **S**_r_ : Pops the value off the top of the (main) stack and pushes it onto the stack of register *r*. The previous value of the register becomes inaccessible. **L**_r_ : Pops the value off the top of the stack for register *r* and push it onto the main stack. The previous value in the stack for register *r*, if any, is now accessible via the **l**_r_ command. ## Parameters These commands control the values of **ibase**, **obase**, **scale**, and **seed**. Also see the **SYNTAX** section. **i** : Pops the value off of the top of the stack and uses it to set **ibase**, which must be between **2** and **16**, inclusive. If the value on top of the stack has any *scale*, the *scale* is ignored. **o** : Pops the value off of the top of the stack and uses it to set **obase**, which must be between **0** and **DC_BASE_MAX**, inclusive (see the **LIMITS** section and the **NUMBERS** section). If the value on top of the stack has any *scale*, the *scale* is ignored. **k** : Pops the value off of the top of the stack and uses it to set **scale**, which must be non-negative. If the value on top of the stack has any *scale*, the *scale* is ignored. **j** : Pops the value off of the top of the stack and uses it to set **seed**. The meaning of **seed** is dependent on the current pseudo-random number generator but is guaranteed to not change except for new major versions. The *scale* and sign of the value may be significant. If a previously used **seed** value is used again, the pseudo-random number generator is guaranteed to produce the same sequence of pseudo-random numbers as it did when the **seed** value was previously used. The exact value assigned to **seed** is not guaranteed to be returned if the **J** command is used. However, if **seed** *does* return a different value, both values, when assigned to **seed**, are guaranteed to produce the same sequence of pseudo-random numbers. This means that certain values assigned to **seed** will not produce unique sequences of pseudo-random numbers. There is no limit to the length (number of significant decimal digits) or *scale* of the value that can be assigned to **seed**. This is a **non-portable extension**. **I** : Pushes the current value of **ibase** onto the main stack. **O** : Pushes the current value of **obase** onto the main stack. **K** : Pushes the current value of **scale** onto the main stack. **J** : Pushes the current value of **seed** onto the main stack. This is a **non-portable extension**. **T** : Pushes the maximum allowable value of **ibase** onto the main stack. This is a **non-portable extension**. **U** : Pushes the maximum allowable value of **obase** onto the main stack. This is a **non-portable extension**. **V** : Pushes the maximum allowable value of **scale** onto the main stack. This is a **non-portable extension**. **W** : Pushes the maximum (inclusive) integer that can be generated with the **'** pseudo-random number generator command. This is a **non-portable extension**. ## Strings The following commands control strings. dc(1) can work with both numbers and strings, and registers (see the **REGISTERS** section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. Strings can also be executed as macros. For example, if the string **[1pR]** is executed as a macro, then the code **1pR** is executed, meaning that the **1** will be printed with a newline after and then popped from the stack. **\[**_characters_**\]** : Makes a string containing *characters* and pushes it onto the stack. If there are brackets (**\[** and **\]**) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (**\\**) character. If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. **a** : The value on top of the stack is popped. If it is a number, it is truncated and its absolute value is taken. The result mod **256** is calculated. If that result is **0**, push an empty string; otherwise, push a one-character string where the character is the result of the mod interpreted as an ASCII character. If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one-character string. The new string is then pushed onto the stack. This is a **non-portable extension**. **x** : Pops a value off of the top of the stack. If it is a number, it is pushed back onto the stack. If it is a string, it is executed as a macro. This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. **\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register *r* are executed. For example, **0 1>a** will execute the contents of register **a**, and **1 0>a** will not. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **?** : Reads a line from the **stdin** and executes it. This is to allow macros to request input from users. **q** : During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. **Q** : Pops a value from the stack which must be non-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. **,** : Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the **Q** command, so the sequence **,Q** will make dc(1) exit. This is a **non-portable extension**. ## Status These commands query status of the stack or its top value. **Z** : Pops a value off of the stack. If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push **1** if the argument is **0** with no decimal places. If it is a string, pushes the number of characters the string has. **X** : Pops a value off of the stack. If it is a number, pushes the *scale* of the value onto the stack. If it is a string, pushes **0**. **u** : Pops one value off of the stack. If the value is a number, this pushes **1** onto the stack. Otherwise (if it is a string), it pushes **0**. This is a **non-portable extension**. **t** : Pops one value off of the stack. If the value is a string, this pushes **1** onto the stack. Otherwise (if it is a number), it pushes **0**. This is a **non-portable extension**. **z** : Pushes the current depth of the stack (before execution of this command) onto the stack. **y**_r_ : Pushes the current stack depth of the register *r* onto the main stack. Because each register has a depth of **1** (with the value **0** in the top item) when dc(1) starts, dc(1) requires that each register's stack must always have at least one item; dc(1) will give an error and reset otherwise (see the **RESET** section). This means that this command will never push **0**. This is a **non-portable extension**. ## Arrays These commands manipulate arrays. **:**_r_ : Pops the top two values off of the stack. The second value will be stored in the array *r* (see the **REGISTERS** section), indexed by the first value. **;**_r_ : Pops the value on top of the stack and uses it as an index into the array *r*. The selected value is then pushed onto the stack. **Y**_r_ : Pushes the length of the array *r* onto the stack. This is a **non-portable extension**. ## Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter **g**. Only the characters below are allowed after the character **g**; any other character produces a parse error (see the **ERRORS** section). **gl** : Pushes the line length set by **DC_LINE_LENGTH** (see the **ENVIRONMENT VARIABLES** section) onto the stack. **gx** : Pushes **1** onto the stack if extended register mode is on, **0** otherwise. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. **gz** : Pushes **0** onto the stack if the leading zero setting has not been enabled with the **-z** or **-\-leading-zeroes** options (see the **OPTIONS** section), non-zero otherwise. # REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (**0**) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. In non-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (**'\\n'**) and a left bracket (**'['**); it is a parse error for a newline or a left bracket to be used as a register name. ## Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. If extended register mode is enabled (**-x** or **-\-extended-register** command-line arguments are given), then normal single character registers are used *unless* the character immediately following a command that needs a register name is a space (according to **isspace()**) and not a newline (**'\\n'**). In that case, the register name is found according to the regex **\[a-z\]\[a-z0-9\_\]\*** (like bc(1) identifiers), and it is a parse error if the next non-space characters do not match that regex. # RESET When dc(1) encounters an error or a signal that it has a non-default handler for, it resets. This means that several things happen. -First, any macros that are executing are stopped and popped off the stack. -The behavior is not unlike that of exceptions in programming languages. Then -the execution point is set so that any code waiting to execute (after all +First, any macros that are executing are stopped and popped off the execution +stack. The behavior is not unlike that of exceptions in programming languages. +Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. +However, the stack of values is *not* cleared; in interactive mode, users can +inspect the stack and manipulate it. + Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the **EXIT STATUS** section), it asks for more input; otherwise, it exits with the appropriate return code. # PERFORMANCE Most dc(1) implementations use **char** types to calculate the value of **1** decimal digit at a time, but that can be slow. This dc(1) does something different. It uses large integers to calculate more than **1** decimal digit at a time. If built in a environment where **DC_LONG_BIT** (see the **LIMITS** section) is **64**, then each integer has **9** decimal digits. If built in an environment where **DC_LONG_BIT** is **32** then each integer has **4** decimal digits. This value (the number of decimal digits per large integer) is called **DC_BASE_DIGS**. In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of **DC_LONG_BIT**, but is always at least twice as large as the integer type used to store digits. # LIMITS The following are the limits on dc(1): **DC_LONG_BIT** : The number of bits in the **long** type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the **PERFORMANCE** section). **DC_BASE_DIGS** : The number of decimal digits per large integer (see the **PERFORMANCE** section). Depends on **DC_LONG_BIT**. **DC_BASE_POW** : The max decimal number that each large integer can store (see **DC_BASE_DIGS**) plus **1**. Depends on **DC_BASE_DIGS**. **DC_OVERFLOW_MAX** : The max number that the overflow type (see the **PERFORMANCE** section) can hold. Depends on **DC_LONG_BIT**. **DC_BASE_MAX** : The maximum output base. Set at **DC_BASE_POW**. **DC_DIM_MAX** : The maximum size of arrays. Set at **SIZE_MAX-1**. **DC_SCALE_MAX** : The maximum **scale**. Set at **DC_OVERFLOW_MAX-1**. **DC_STRING_MAX** : The maximum length of strings. Set at **DC_OVERFLOW_MAX-1**. **DC_NAME_MAX** : The maximum length of identifiers. Set at **DC_OVERFLOW_MAX-1**. **DC_NUM_MAX** : The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at **DC_OVERFLOW_MAX-1**. **DC_RAND_MAX** : The maximum integer (inclusive) returned by the **'** command, if dc(1). Set at **2\^DC_LONG_BIT-1**. Exponent : The maximum allowable exponent (positive or negative). Set at **DC_OVERFLOW_MAX**. Number of vars : The maximum number of vars/arrays. Set at **SIZE_MAX-1**. These limits are meant to be effectively non-existent; the limits are so large (at least on 64-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. # ENVIRONMENT VARIABLES As **non-portable extensions**, dc(1) recognizes the following environment variables: **DC_ENV_ARGS** : This is another way to give command-line arguments to dc(1). They should be in the same format as all other command-line arguments. These are always processed first, so any files given in **DC_ENV_ARGS** will be processed before arguments and files given on the command-line. This gives the user the ability to set up "standard" options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the **-e** option to set **scale** to a value other than **0**. The code that parses **DC_ENV_ARGS** will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string **"/home/gavin/some dc file.dc"** will be correctly parsed, but the string **"/home/gavin/some \"dc\" file.dc"** will include the backslashes. The quote parsing will handle either kind of quotes, **'** or **"**. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in **"some 'dc' file.dc"**, and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in **DC_ENV_ARGS** is not supported due to the complexity of the parsing, though such files are still supported on the command-line where the parsing is done by the shell. **DC_LINE_LENGTH** : If this environment variable exists and contains an integer that is greater than **1** and is less than **UINT16_MAX** (**2\^16-1**), dc(1) will output lines to that length, including the backslash newline combo. The default line length is **70**. The special value of **0** will disable line length checking and print numbers without regard to line length and without backslashes and newlines. **DC_SIGINT_RESET** : If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), then this environment variable has no effect because dc(1) exits on **SIGINT** when not in interactive mode. However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) reset on **SIGINT**, rather than exit, and zero makes dc(1) exit. If this environment variable exists and is *not* an integer, then dc(1) will exit on **SIGINT**. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_TTY_MODE** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_PROMPT** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) use a prompt, and zero or a non-integer makes dc(1) not use a prompt. If this environment variable does not exist and **DC_TTY_MODE** does, then the value of the **DC_TTY_MODE** environment variable is used. This environment variable and the **DC_TTY_MODE** environment variable override the default, which can be queried with the **-h** or **-\-help** options. **DC_EXPR_EXIT** : If any expressions or expression files are given on the command-line with **-e**, **-\-expression**, **-f**, or **-\-file**, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_DIGIT_CLAMP** : When parsing numbers and if this environment variable exists and contains an integer, a non-zero value makes dc(1) clamp digits that are greater than or equal to the current **ibase** so that all such digits are considered equal to the **ibase** minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the **ibase**. This never applies to single-digit numbers, as per the bc(1) standard (see the **STANDARDS** section). This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. # EXIT STATUS dc(1) returns the following exit statuses: **0** : No error. **1** : A math error occurred. This follows standard practice of using **1** for expected errors, since math errors will happen in the process of normal execution. Math errors include divide by **0**, taking the square root of a negative number, using a negative number as a bound for the pseudo-random number generator, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non-integer where an integer is required. Converting to a hardware integer happens for the second operand of the power (**\^**), places (**\@**), left shift (**H**), and right shift (**h**) operators. **2** : A parse error occurred. Parse errors include unexpected **EOF**, using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. **3** : A runtime error occurred. Runtime errors include assigning an invalid number to any global (**ibase**, **obase**, or **scale**), giving a bad expression to a **read()** call, calling **read()** inside of a **read()** call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. **4** : A fatal error occurred. Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command-line options. The exit status **4** is special; when a fatal error occurs, dc(1) always exits and returns **4**, no matter what mode dc(1) is in. The other statuses will only be returned when dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), since dc(1) resets its state (see the **RESET** section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the **-i** flag or **-\-interactive** option. These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the **-i** flag or **-\-interactive** option. # INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non-interactive mode. Interactive mode is turned on automatically when both **stdin** and **stdout** are hooked to a terminal, but the **-i** flag and **-\-interactive** option can turn it on in other situations. In interactive mode, dc(1) attempts to recover from errors (see the **RESET** section), and in normal execution, flushes **stdout** as soon as execution is done for the current input. dc(1) may also reset on **SIGINT** instead of exit, depending on the contents of, or default for, the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section). # TTY MODE If **stdin**, **stdout**, and **stderr** are all connected to a TTY, then "TTY mode" is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. If there is the environment variable **DC_TTY_MODE** in the environment (see the **ENVIRONMENT VARIABLES** section), then if that environment variable contains a non-zero integer, dc(1) will turn on TTY mode when **stdin**, **stdout**, and **stderr** are all connected to a TTY. If the **DC_TTY_MODE** environment variable exists but is *not* a non-zero integer, then dc(1) will not turn TTY mode on. If the environment variable **DC_TTY_MODE** does *not* exist, the default setting is used. The default setting can be queried with the **-h** or **-\-help** options. TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the **STANDARDS** section), and interactive mode requires only **stdin** and **stdout** to be connected to a terminal. ## Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: **DC_PROMPT** (see the **ENVIRONMENT VARIABLES** section). If the environment variable **DC_PROMPT** exists and is a non-zero integer, then the prompt is turned on when **stdin**, **stdout**, and **stderr** are connected to a TTY and the **-P** and **-\-no-prompt** options were not used. The read prompt will be turned on under the same conditions, except that the **-R** and **-\-no-read-prompt** options must also not be used. However, if **DC_PROMPT** does not exist, the prompt can be enabled or disabled with the **DC_TTY_MODE** environment variable, the **-P** and **-\-no-prompt** options, and the **-R** and **-\-no-read-prompt** options. See the **ENVIRONMENT VARIABLES** and **OPTIONS** sections for more details. # SIGNAL HANDLING Sending a **SIGINT** will cause dc(1) to do one of two things. If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), or the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section), or its default, is either not an integer or it is zero, dc(1) will exit. However, if dc(1) is in interactive mode, and the **DC_SIGINT_RESET** or its default is an integer and non-zero, then dc(1) will stop executing the current input and reset (see the **RESET** section) upon receiving a **SIGINT**. Note that "current input" can mean one of two things. If dc(1) is processing input from **stdin** in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from **stdin** if no other file exists. This means that if a **SIGINT** is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. **SIGTERM** and **SIGQUIT** cause dc(1) to clean up and exit, and it uses the default handler for all other signals. # SEE ALSO bc(1) # STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1-2017 (“POSIX.1-2017”) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . # BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . # AUTHOR Gavin D. Howard and contributors. diff --git a/manuals/dc/N.1 b/manuals/dc/N.1 index 6233c753dea8..6e2baa587b1c 100644 --- a/manuals/dc/N.1 +++ b/manuals/dc/N.1 @@ -1,1689 +1,1692 @@ .\" .\" SPDX-License-Identifier: BSD-2-Clause .\" .\" Copyright (c) 2018-2024 Gavin D. Howard and contributors. .\" .\" Redistribution and use in source and binary forms, with or without .\" modification, are permitted provided that the following conditions are met: .\" .\" * Redistributions of source code must retain the above copyright notice, .\" this list of conditions and the following disclaimer. .\" .\" * Redistributions in binary form must reproduce the above copyright notice, .\" this list of conditions and the following disclaimer in the documentation .\" and/or other materials provided with the distribution. .\" .\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" .\" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE .\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE .\" ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE .\" LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF .\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS .\" INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN .\" CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) .\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE .\" POSSIBILITY OF SUCH DAMAGE. .\" -.TH "DC" "1" "January 2024" "Gavin D. Howard" "General Commands Manual" +.TH "DC" "1" "August 2024" "Gavin D. Howard" "General Commands Manual" .nh .ad l .SH Name dc \- arbitrary\-precision decimal reverse\-Polish notation calculator .SH SYNOPSIS \f[B]dc\f[R] [\f[B]\-cChiPRvVx\f[R]] [\f[B]\-\-version\f[R]] [\f[B]\-\-help\f[R]] [\f[B]\-\-digit\-clamp\f[R]] [\f[B]\-\-no\-digit\-clamp\f[R]] [\f[B]\-\-interactive\f[R]] [\f[B]\-\-no\-prompt\f[R]] [\f[B]\-\-no\-read\-prompt\f[R]] [\f[B]\-\-extended\-register\f[R]] [\f[B]\-e\f[R] \f[I]expr\f[R]] [\f[B]\-\-expression\f[R]=\f[I]expr\f[R]\&...] [\f[B]\-f\f[R] \f[I]file\f[R]\&...] [\f[B]\-\-file\f[R]=\f[I]file\f[R]\&...] [\f[I]file\f[R]\&...] [\f[B]\-I\f[R] \f[I]ibase\f[R]] [\f[B]\-\-ibase\f[R]=\f[I]ibase\f[R]] [\f[B]\-O\f[R] \f[I]obase\f[R]] [\f[B]\-\-obase\f[R]=\f[I]obase\f[R]] [\f[B]\-S\f[R] \f[I]scale\f[R]] [\f[B]\-\-scale\f[R]=\f[I]scale\f[R]] [\f[B]\-E\f[R] \f[I]seed\f[R]] [\f[B]\-\-seed\f[R]=\f[I]seed\f[R]] .SH DESCRIPTION dc(1) is an arbitrary\-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. .PP If no files are given on the command\-line, then dc(1) reads from \f[B]stdin\f[R] (see the \f[B]STDIN\f[R] section). Otherwise, those files are processed, and dc(1) will then exit. .PP If a user wants to set up a standard environment, they can use \f[B]DC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). For example, if a user wants the \f[B]scale\f[R] always set to \f[B]10\f[R], they can set \f[B]DC_ENV_ARGS\f[R] to \f[B]\-e 10k\f[R], and this dc(1) will always start with a \f[B]scale\f[R] of \f[B]10\f[R]. .SH OPTIONS The following are the options that dc(1) accepts. .TP \f[B]\-C\f[R], \f[B]\-\-no\-digit\-clamp\f[R] Disables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that the value added to a number from a digit is always that digit\[cq]s value multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-c\f[R] or \f[B]\-\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-c\f[R], \f[B]\-\-digit\-clamp\f[R] Enables clamping of digits greater than or equal to the current \f[B]ibase\f[R] when parsing numbers. .RS .PP This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit\[cq]s position, which starts from 0 at the least significant digit. .PP If this and/or the \f[B]\-C\f[R] or \f[B]\-\-no\-digit\-clamp\f[R] options are given multiple times, the last one given is used. .PP This option overrides the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-E\f[R] \f[I]seed\f[R], \f[B]\-\-seed\f[R]=\f[I]seed\f[R] Sets the builtin variable \f[B]seed\f[R] to the value \f[I]seed\f[R] assuming that \f[I]seed\f[R] is in base 10. It is a fatal error if \f[I]seed\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-e\f[R] \f[I]expr\f[R], \f[B]\-\-expression\f[R]=\f[I]expr\f[R] Evaluates \f[I]expr\f[R]. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R], whether on the command\-line or in \f[B]DC_ENV_ARGS\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-f\f[R] \f[I]file\f[R], \f[B]\-\-file\f[R]=\f[I]file\f[R] Reads in \f[I]file\f[R] and evaluates it, line by line, as though it were read through \f[B]stdin\f[R]. If expressions are also given (see above), the expressions are evaluated in the order given. .RS .PP If this option is given on the command\-line (i.e., not in \f[B]DC_ENV_ARGS\f[R], see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then after processing all expressions and files, dc(1) will exit, unless \f[B]\-\f[R] (\f[B]stdin\f[R]) was given as an argument at least once to \f[B]\-f\f[R] or \f[B]\-\-file\f[R]. However, if any other \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R] arguments are given after \f[B]\-f\-\f[R] or equivalent is given, dc(1) will give a fatal error and exit. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-h\f[R], \f[B]\-\-help\f[R] Prints a usage message and exits. .TP \f[B]\-I\f[R] \f[I]ibase\f[R], \f[B]\-\-ibase\f[R]=\f[I]ibase\f[R] Sets the builtin variable \f[B]ibase\f[R] to the value \f[I]ibase\f[R] assuming that \f[I]ibase\f[R] is in base 10. It is a fatal error if \f[I]ibase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-i\f[R], \f[B]\-\-interactive\f[R] Forces interactive mode. (See the \f[B]INTERACTIVE MODE\f[R] section.) .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-L\f[R], \f[B]\-\-no\-line\-length\f[R] Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets \f[B]BC_LINE_LENGTH\f[R] to \f[B]0\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-O\f[R] \f[I]obase\f[R], \f[B]\-\-obase\f[R]=\f[I]obase\f[R] Sets the builtin variable \f[B]obase\f[R] to the value \f[I]obase\f[R] assuming that \f[I]obase\f[R] is in base 10. It is a fatal error if \f[I]obase\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-P\f[R], \f[B]\-\-no\-prompt\f[R] Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]DC_ENV_ARGS\f[R]. .RS .PP These options override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-R\f[R], \f[B]\-\-no\-read\-prompt\f[R] Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the \f[B]TTY MODE\f[R] section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in \f[B]BC_ENV_ARGS\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. .RS .PP This option does not disable the regular prompt because the read prompt is only used when the \f[B]?\f[R] command is used. .PP These options \f[I]do\f[R] override the \f[B]DC_PROMPT\f[R] and \f[B]DC_TTY_MODE\f[R] environment variables (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), but only for the read prompt. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-S\f[R] \f[I]scale\f[R], \f[B]\-\-scale\f[R]=\f[I]scale\f[R] Sets the builtin variable \f[B]scale\f[R] to the value \f[I]scale\f[R] assuming that \f[I]scale\f[R] is in base 10. It is a fatal error if \f[I]scale\f[R] is not a valid number. .RS .PP If multiple instances of this option are given, the last is used. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-v\f[R], \f[B]\-V\f[R], \f[B]\-\-version\f[R] Print the version information (copyright header) and exits. .TP \f[B]\-x\f[R] \f[B]\-\-extended\-register\f[R] Enables extended register mode. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\-z\f[R], \f[B]\-\-leading\-zeroes\f[R] Makes dc(1) print all numbers greater than \f[B]\-1\f[R] and less than \f[B]1\f[R], and not equal to \f[B]0\f[R], with a leading zero. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .PP All long options are \f[B]non\-portable extensions\f[R]. .SH STDIN If no files are given on the command\-line and no files or expressions are given by the \f[B]\-f\f[R], \f[B]\-\-file\f[R], \f[B]\-e\f[R], or \f[B]\-\-expression\f[R] options, then dc(1) reads from \f[B]stdin\f[R]. .PP However, there is a caveat to this. .PP First, \f[B]stdin\f[R] is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. .SH STDOUT Any non\-error output is written to \f[B]stdout\f[R]. In addition, if history (see the \f[B]HISTORY\f[R] section) and the prompt (see the \f[B]TTY MODE\f[R] section) are enabled, both are output to \f[B]stdout\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stdout\f[R], so if \f[B]stdout\f[R] is closed, as in \f[B]dc >&\-\f[R], it will quit with an error. This is done so that dc(1) can report problems when \f[B]stdout\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stdout\f[R] to \f[B]/dev/null\f[R]. .SH STDERR Any error output is written to \f[B]stderr\f[R]. .PP \f[B]Note\f[R]: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the \f[B]EXIT STATUS\f[R] section) if it cannot write to \f[B]stderr\f[R], so if \f[B]stderr\f[R] is closed, as in \f[B]dc 2>&\-\f[R], it will quit with an error. This is done so that dc(1) can exit with an error code when \f[B]stderr\f[R] is redirected to a file. .PP If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect \f[B]stderr\f[R] to \f[B]/dev/null\f[R]. .SH SYNTAX Each item in the input source code, either a number (see the \f[B]NUMBERS\f[R] section) or a command (see the \f[B]COMMANDS\f[R] section), is processed and executed, in order. Input is processed immediately when entered. .PP \f[B]ibase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to interpret constant numbers. It is the \[lq]input\[rq] base, or the number base used for interpreting input numbers. \f[B]ibase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]ibase\f[R] is \f[B]16\f[R]. The min allowable value for \f[B]ibase\f[R] is \f[B]2\f[R]. The max allowable value for \f[B]ibase\f[R] can be queried in dc(1) programs with the \f[B]T\f[R] command. .PP \f[B]obase\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that determines how to output results. It is the \[lq]output\[rq] base, or the number base used for outputting numbers. \f[B]obase\f[R] is initially \f[B]10\f[R]. The max allowable value for \f[B]obase\f[R] is \f[B]DC_BASE_MAX\f[R] and can be queried with the \f[B]U\f[R] command. The min allowable value for \f[B]obase\f[R] is \f[B]0\f[R]. If \f[B]obase\f[R] is \f[B]0\f[R], values are output in scientific notation, and if \f[B]obase\f[R] is \f[B]1\f[R], values are output in engineering notation. Otherwise, values are output in the specified base. .PP Outputting in scientific and engineering notations are \f[B]non\-portable extensions\f[R]. .PP The \f[I]scale\f[R] of an expression is the number of digits in the result of the expression right of the decimal point, and \f[B]scale\f[R] is a register (see the \f[B]REGISTERS\f[R] section) that sets the precision of any operations (with exceptions). \f[B]scale\f[R] is initially \f[B]0\f[R]. \f[B]scale\f[R] cannot be negative. The max allowable value for \f[B]scale\f[R] can be queried in dc(1) programs with the \f[B]V\f[R] command. .PP \f[B]seed\f[R] is a register containing the current seed for the pseudo\-random number generator. If the current value of \f[B]seed\f[R] is queried and stored, then if it is assigned to \f[B]seed\f[R] later, the pseudo\-random number generator is guaranteed to produce the same sequence of pseudo\-random numbers that were generated after the value of \f[B]seed\f[R] was first queried. .PP Multiple values assigned to \f[B]seed\f[R] can produce the same sequence of pseudo\-random numbers. Likewise, when a value is assigned to \f[B]seed\f[R], it is not guaranteed that querying \f[B]seed\f[R] immediately after will return the same value. In addition, the value of \f[B]seed\f[R] will change after any call to the \f[B]\[cq]\f[R] command or the \f[B]\[lq]\f[R] command that does not get receive a value of \f[B]0\f[R] or \f[B]1\f[R]. The maximum integer returned by the \f[B]\[cq]\f[R] command can be queried with the \f[B]W\f[R] command. .PP \f[B]Note\f[R]: The values returned by the pseudo\-random number generator with the \f[B]\[cq]\f[R] and \f[B]\[lq]\f[R] commands are guaranteed to \f[B]NOT\f[R] be cryptographically secure. This is a consequence of using a seeded pseudo\-random number generator. However, they \f[I]are\f[R] guaranteed to be reproducible with identical \f[B]seed\f[R] values. This means that the pseudo\-random values from dc(1) should only be used where a reproducible stream of pseudo\-random numbers is \f[I]ESSENTIAL\f[R]. In any other case, use a non\-seeded pseudo\-random number generator. .PP The pseudo\-random number generator, \f[B]seed\f[R], and all associated operations are \f[B]non\-portable extensions\f[R]. .SS Comments Comments go from \f[B]#\f[R] until, and not including, the next newline. This is a \f[B]non\-portable extension\f[R]. .SH NUMBERS Numbers are strings made up of digits, uppercase letters up to \f[B]F\f[R], and at most \f[B]1\f[R] period for a radix. Numbers can have up to \f[B]DC_NUM_MAX\f[R] digits. Uppercase letters are equal to \f[B]9\f[R] plus their position in the alphabet (i.e., \f[B]A\f[R] equals \f[B]10\f[R], or \f[B]9+1\f[R]). .PP If a digit or letter makes no sense with the current value of \f[B]ibase\f[R] (i.e., they are greater than or equal to the current value of \f[B]ibase\f[R]), then the behavior depends on the existence of the \f[B]\-c\f[R]/\f[B]\-\-digit\-clamp\f[R] or \f[B]\-C\f[R]/\f[B]\-\-no\-digit\-clamp\f[R] options (see the \f[B]OPTIONS\f[R] section), the existence and setting of the \f[B]DC_DIGIT_CLAMP\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or the default, which can be queried with the \f[B]\-h\f[R]/\f[B]\-\-help\f[R] option. .PP If clamping is off, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are not changed. Instead, their given value is multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*A+3\[ha]0*B\f[R], which is \f[B]3\f[R] times \f[B]10\f[R] plus \f[B]11\f[R], or \f[B]41\f[R]. .PP If clamping is on, then digits or letters that are greater than or equal to the current value of \f[B]ibase\f[R] are set to the value of the highest valid digit in \f[B]ibase\f[R] before being multiplied by the appropriate power of \f[B]ibase\f[R] and added into the number. This means that, with an \f[B]ibase\f[R] of \f[B]3\f[R], the number \f[B]AB\f[R] is equal to \f[B]3\[ha]1*2+3\[ha]0*2\f[R], which is \f[B]3\f[R] times \f[B]2\f[R] plus \f[B]2\f[R], or \f[B]8\f[R]. .PP There is one exception to clamping: single\-character numbers (i.e., \f[B]A\f[R] alone). Such numbers are never clamped and always take the value they would have in the highest possible \f[B]ibase\f[R]. This means that \f[B]A\f[R] alone always equals decimal \f[B]10\f[R] and \f[B]Z\f[R] alone always equals decimal \f[B]35\f[R]. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current \f[B]ibase\f[R] (with the \f[B]i\f[R] command) regardless of the current value of \f[B]ibase\f[R]. .PP If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for \f[B]A\f[R], use \f[B]0A\f[R]. .PP In addition, dc(1) accepts numbers in scientific notation. These have the form \f[B]e\f[R]. The exponent (the portion after the \f[B]e\f[R]) must be an integer. An example is \f[B]1.89237e9\f[R], which is equal to \f[B]1892370000\f[R]. Negative exponents are also allowed, so \f[B]4.2890e_3\f[R] is equal to \f[B]0.0042890\f[R]. .PP \f[B]WARNING\f[R]: Both the number and the exponent in scientific notation are interpreted according to the current \f[B]ibase\f[R], but the number is still multiplied by \f[B]10\[ha]exponent\f[R] regardless of the current \f[B]ibase\f[R]. For example, if \f[B]ibase\f[R] is \f[B]16\f[R] and dc(1) is given the number string \f[B]FFeA\f[R], the resulting decimal number will be \f[B]2550000000000\f[R], and if dc(1) is given the number string \f[B]10e_4\f[R], the resulting decimal number will be \f[B]0.0016\f[R]. .PP Accepting input as scientific notation is a \f[B]non\-portable extension\f[R]. .SH COMMANDS The valid commands are listed below. .SS Printing These commands are used for printing. .PP Note that both scientific notation and engineering notation are available for printing numbers. Scientific notation is activated by assigning \f[B]0\f[R] to \f[B]obase\f[R] using \f[B]0o\f[R], and engineering notation is activated by assigning \f[B]1\f[R] to \f[B]obase\f[R] using \f[B]1o\f[R]. To deactivate them, just assign a different value to \f[B]obase\f[R]. .PP Printing numbers in scientific notation and/or engineering notation is a \f[B]non\-portable extension\f[R]. .TP \f[B]p\f[R] Prints the value on top of the stack, whether number or string, and prints a newline after. .RS .PP This does not alter the stack. .RE .TP \f[B]n\f[R] Prints the value on top of the stack, whether number or string, and pops it off of the stack. .TP \f[B]P\f[R] Pops a value off the stack. .RS .PP If the value is a number, it is truncated and the absolute value of the result is printed as though \f[B]obase\f[R] is \f[B]256\f[R] and each digit is interpreted as an 8\-bit ASCII character, making it a byte stream. .PP If the value is a string, it is printed without a trailing newline. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]f\f[R] Prints the entire contents of the stack, in order from newest to oldest, without altering anything. .RS .PP Users should use this command when they get lost. .RE .SS Arithmetic These are the commands used for arithmetic. .TP \f[B]+\f[R] The top two values are popped off the stack, added, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]\-\f[R] The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to the max \f[I]scale\f[R] of both operands. .TP \f[B]*\f[R] The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If \f[B]a\f[R] is the \f[I]scale\f[R] of the first expression and \f[B]b\f[R] is the \f[I]scale\f[R] of the second expression, the \f[I]scale\f[R] of the result is equal to \f[B]min(a+b,max(scale,a,b))\f[R] where \f[B]min()\f[R] and \f[B]max()\f[R] return the obvious values. .TP \f[B]/\f[R] The top two values are popped off the stack, divided, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]%\f[R] The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. .RS .PP Remaindering is equivalent to 1) Computing \f[B]a/b\f[R] to current \f[B]scale\f[R], and 2) Using the result of step 1 to calculate \f[B]a\-(a/b)*b\f[R] to \f[I]scale\f[R] \f[B]max(scale+scale(b),scale(a))\f[R]. .PP The first value popped off of the stack must be non\-zero. .RE .TP \f[B]\[ti]\f[R] The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to \f[B]x y / x y %\f[R] except that \f[B]x\f[R] and \f[B]y\f[R] are only evaluated once. .RS .PP The first value popped off of the stack must be non\-zero. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[ha]\f[R] The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non\-zero. .RE .TP \f[B]v\f[R] The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The \f[I]scale\f[R] of the result is equal to \f[B]scale\f[R]. .RS .PP The value popped off of the stack must be non\-negative. .RE .TP \f[B]_\f[R] If this command \f[I]immediately\f[R] precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. .RS .PP Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]b\f[R] The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]|\f[R] The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. .RS .PP The first value popped is used as the reduction modulus and must be an integer and non\-zero. The second value popped is used as the exponent and must be an integer and non\-negative. The third value popped is the base and must be an integer. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]$\f[R] The top value is popped off the stack and copied, and the copy is truncated and pushed onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[at]\f[R] The top two values are popped off the stack, and the precision of the second is set to the value of the first, whether by truncation or extension. .RS .PP The first value popped off of the stack must be an integer and non\-negative. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]H\f[R] The top two values are popped off the stack, and the second is shifted left (radix shifted right) to the value of the first. .RS .PP The first value popped off of the stack must be an integer and non\-negative. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]h\f[R] The top two values are popped off the stack, and the second is shifted right (radix shifted left) to the value of the first. .RS .PP The first value popped off of the stack must be an integer and non\-negative. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]G\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if they are equal, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]N\f[R] The top value is popped off of the stack, and if it a \f[B]0\f[R], a \f[B]1\f[R] is pushed; otherwise, a \f[B]0\f[R] is pushed. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B](\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]{\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is less than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B])\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]}\f[R] The top two values are popped off of the stack, they are compared, and a \f[B]1\f[R] is pushed if the first is greater than or equal to the second, or \f[B]0\f[R] otherwise. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]M\f[R] The top two values are popped off of the stack. If they are both non\-zero, a \f[B]1\f[R] is pushed onto the stack. If either of them is zero, or both of them are, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]&&\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]m\f[R] The top two values are popped off of the stack. If at least one of them is non\-zero, a \f[B]1\f[R] is pushed onto the stack. If both of them are zero, then a \f[B]0\f[R] is pushed onto the stack. .RS .PP This is like the \f[B]||\f[R] operator in bc(1), and it is \f[I]not\f[R] a short\-circuit operator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Pseudo\-Random Number Generator dc(1) has a built\-in pseudo\-random number generator. These commands query the pseudo\-random number generator. (See Parameters for more information about the \f[B]seed\f[R] value that controls the pseudo\-random number generator.) .PP The pseudo\-random number generator is guaranteed to \f[B]NOT\f[R] be cryptographically secure. .TP \f[B]\[cq]\f[R] Generates an integer between 0 and \f[B]DC_RAND_MAX\f[R], inclusive (see the \f[B]LIMITS\f[R] section). .RS .PP The generated integer is made as unbiased as possible, subject to the limitations of the pseudo\-random number generator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]\[lq]\f[R] Pops a value off of the stack, which is used as an \f[B]exclusive\f[R] upper bound on the integer that will be generated. If the bound is negative or is a non\-integer, an error is raised, and dc(1) resets (see the \f[B]RESET\f[R] section) while \f[B]seed\f[R] remains unchanged. If the bound is larger than \f[B]DC_RAND_MAX\f[R], the higher bound is honored by generating several pseudo\-random integers, multiplying them by appropriate powers of \f[B]DC_RAND_MAX+1\f[R], and adding them together. Thus, the size of integer that can be generated with this command is unbounded. Using this command will change the value of \f[B]seed\f[R], unless the operand is \f[B]0\f[R] or \f[B]1\f[R]. In that case, \f[B]0\f[R] is pushed onto the stack, and \f[B]seed\f[R] is \f[I]not\f[R] changed. .RS .PP The generated integer is made as unbiased as possible, subject to the limitations of the pseudo\-random number generator. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Stack Control These commands control the stack. .TP \f[B]c\f[R] Removes all items from (\[lq]clears\[rq]) the stack. .TP \f[B]d\f[R] Copies the item on top of the stack (\[lq]duplicates\[rq]) and pushes the copy onto the stack. .TP \f[B]r\f[R] Swaps (\[lq]reverses\[rq]) the two top items on the stack. .TP \f[B]R\f[R] Pops (\[lq]removes\[rq]) the top value from the stack. .SS Register Control These commands control registers (see the \f[B]REGISTERS\f[R] section). .TP \f[B]s\f[R]\f[I]r\f[R] Pops the value off the top of the stack and stores it into register \f[I]r\f[R]. .TP \f[B]l\f[R]\f[I]r\f[R] Copies the value in register \f[I]r\f[R] and pushes it onto the stack. This does not alter the contents of \f[I]r\f[R]. .TP \f[B]S\f[R]\f[I]r\f[R] Pops the value off the top of the (main) stack and pushes it onto the stack of register \f[I]r\f[R]. The previous value of the register becomes inaccessible. .TP \f[B]L\f[R]\f[I]r\f[R] Pops the value off the top of the stack for register \f[I]r\f[R] and push it onto the main stack. The previous value in the stack for register \f[I]r\f[R], if any, is now accessible via the \f[B]l\f[R]\f[I]r\f[R] command. .SS Parameters These commands control the values of \f[B]ibase\f[R], \f[B]obase\f[R], \f[B]scale\f[R], and \f[B]seed\f[R]. Also see the \f[B]SYNTAX\f[R] section. .TP \f[B]i\f[R] Pops the value off of the top of the stack and uses it to set \f[B]ibase\f[R], which must be between \f[B]2\f[R] and \f[B]16\f[R], inclusive. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]o\f[R] Pops the value off of the top of the stack and uses it to set \f[B]obase\f[R], which must be between \f[B]0\f[R] and \f[B]DC_BASE_MAX\f[R], inclusive (see the \f[B]LIMITS\f[R] section and the \f[B]NUMBERS\f[R] section). .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]k\f[R] Pops the value off of the top of the stack and uses it to set \f[B]scale\f[R], which must be non\-negative. .RS .PP If the value on top of the stack has any \f[I]scale\f[R], the \f[I]scale\f[R] is ignored. .RE .TP \f[B]j\f[R] Pops the value off of the top of the stack and uses it to set \f[B]seed\f[R]. The meaning of \f[B]seed\f[R] is dependent on the current pseudo\-random number generator but is guaranteed to not change except for new major versions. .RS .PP The \f[I]scale\f[R] and sign of the value may be significant. .PP If a previously used \f[B]seed\f[R] value is used again, the pseudo\-random number generator is guaranteed to produce the same sequence of pseudo\-random numbers as it did when the \f[B]seed\f[R] value was previously used. .PP The exact value assigned to \f[B]seed\f[R] is not guaranteed to be returned if the \f[B]J\f[R] command is used. However, if \f[B]seed\f[R] \f[I]does\f[R] return a different value, both values, when assigned to \f[B]seed\f[R], are guaranteed to produce the same sequence of pseudo\-random numbers. This means that certain values assigned to \f[B]seed\f[R] will not produce unique sequences of pseudo\-random numbers. .PP There is no limit to the length (number of significant decimal digits) or \f[I]scale\f[R] of the value that can be assigned to \f[B]seed\f[R]. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]I\f[R] Pushes the current value of \f[B]ibase\f[R] onto the main stack. .TP \f[B]O\f[R] Pushes the current value of \f[B]obase\f[R] onto the main stack. .TP \f[B]K\f[R] Pushes the current value of \f[B]scale\f[R] onto the main stack. .TP \f[B]J\f[R] Pushes the current value of \f[B]seed\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]T\f[R] Pushes the maximum allowable value of \f[B]ibase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]U\f[R] Pushes the maximum allowable value of \f[B]obase\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]V\f[R] Pushes the maximum allowable value of \f[B]scale\f[R] onto the main stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]W\f[R] Pushes the maximum (inclusive) integer that can be generated with the \f[B]\[cq]\f[R] pseudo\-random number generator command. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Strings The following commands control strings. .PP dc(1) can work with both numbers and strings, and registers (see the \f[B]REGISTERS\f[R] section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. .PP While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. .PP Strings can also be executed as macros. For example, if the string \f[B][1pR]\f[R] is executed as a macro, then the code \f[B]1pR\f[R] is executed, meaning that the \f[B]1\f[R] will be printed with a newline after and then popped from the stack. .TP \f[B][\f[R]\f[I]characters\f[R]\f[B]]\f[R] Makes a string containing \f[I]characters\f[R] and pushes it onto the stack. .RS .PP If there are brackets (\f[B][\f[R] and \f[B]]\f[R]) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (\f[B]\[rs]\f[R]) character. .PP If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. .RE .TP \f[B]a\f[R] The value on top of the stack is popped. .RS .PP If it is a number, it is truncated and its absolute value is taken. The result mod \f[B]256\f[R] is calculated. If that result is \f[B]0\f[R], push an empty string; otherwise, push a one\-character string where the character is the result of the mod interpreted as an ASCII character. .PP If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one\-character string. The new string is then pushed onto the stack. .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]x\f[R] Pops a value off of the top of the stack. .RS .PP If it is a number, it is pushed back onto the stack. .PP If it is a string, it is executed as a macro. .PP This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. .RE .TP \f[B]>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP For example, \f[B]0 1>a\f[R] will execute the contents of register \f[B]a\f[R], and \f[B]1 0>a\f[R] will not. .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!>\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!>\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!<\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!<\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]!=\f[R]\f[I]r\f[R] Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register \f[I]r\f[R] are executed. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .RE .TP \f[B]!=\f[R]\f[I]r\f[R]\f[B]e\f[R]\f[I]s\f[R] Like the above, but will execute register \f[I]s\f[R] if the comparison fails. .RS .PP If either or both of the values are not numbers, dc(1) will raise an error and reset (see the \f[B]RESET\f[R] section). .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]?\f[R] Reads a line from the \f[B]stdin\f[R] and executes it. This is to allow macros to request input from users. .TP \f[B]q\f[R] During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. .TP \f[B]Q\f[R] Pops a value from the stack which must be non\-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. .TP \f[B],\f[R] Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the \f[B]Q\f[R] command, so the sequence \f[B],Q\f[R] will make dc(1) exit. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Status These commands query status of the stack or its top value. .TP \f[B]Z\f[R] Pops a value off of the stack. .RS .PP If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push \f[B]1\f[R] if the argument is \f[B]0\f[R] with no decimal places. .PP If it is a string, pushes the number of characters the string has. .RE .TP \f[B]X\f[R] Pops a value off of the stack. .RS .PP If it is a number, pushes the \f[I]scale\f[R] of the value onto the stack. .PP If it is a string, pushes \f[B]0\f[R]. .RE .TP \f[B]u\f[R] Pops one value off of the stack. If the value is a number, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a string), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]t\f[R] Pops one value off of the stack. If the value is a string, this pushes \f[B]1\f[R] onto the stack. Otherwise (if it is a number), it pushes \f[B]0\f[R]. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .TP \f[B]z\f[R] Pushes the current depth of the stack (before execution of this command) onto the stack. .TP \f[B]y\f[R]\f[I]r\f[R] Pushes the current stack depth of the register \f[I]r\f[R] onto the main stack. .RS .PP Because each register has a depth of \f[B]1\f[R] (with the value \f[B]0\f[R] in the top item) when dc(1) starts, dc(1) requires that each register\[cq]s stack must always have at least one item; dc(1) will give an error and reset otherwise (see the \f[B]RESET\f[R] section). This means that this command will never push \f[B]0\f[R]. .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Arrays These commands manipulate arrays. .TP \f[B]:\f[R]\f[I]r\f[R] Pops the top two values off of the stack. The second value will be stored in the array \f[I]r\f[R] (see the \f[B]REGISTERS\f[R] section), indexed by the first value. .TP \f[B];\f[R]\f[I]r\f[R] Pops the value on top of the stack and uses it as an index into the array \f[I]r\f[R]. The selected value is then pushed onto the stack. .TP \f[B]Y\f[R]\f[I]r\f[R] Pushes the length of the array \f[I]r\f[R] onto the stack. .RS .PP This is a \f[B]non\-portable extension\f[R]. .RE .SS Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter \f[B]g\f[R]. Only the characters below are allowed after the character \f[B]g\f[R]; any other character produces a parse error (see the \f[B]ERRORS\f[R] section). .TP \f[B]gl\f[R] Pushes the line length set by \f[B]DC_LINE_LENGTH\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) onto the stack. .TP \f[B]gx\f[R] Pushes \f[B]1\f[R] onto the stack if extended register mode is on, \f[B]0\f[R] otherwise. See the \f[I]Extended Register Mode\f[R] subsection of the \f[B]REGISTERS\f[R] section for more information. .TP \f[B]gz\f[R] Pushes \f[B]0\f[R] onto the stack if the leading zero setting has not been enabled with the \f[B]\-z\f[R] or \f[B]\-\-leading\-zeroes\f[R] options (see the \f[B]OPTIONS\f[R] section), non\-zero otherwise. .SH REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) .PP Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (\f[B]0\f[R]) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. .PP In non\-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (\f[B]`\[rs]n'\f[R]) and a left bracket (\f[B]`['\f[R]); it is a parse error for a newline or a left bracket to be used as a register name. .SS Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. .PP If extended register mode is enabled (\f[B]\-x\f[R] or \f[B]\-\-extended\-register\f[R] command\-line arguments are given), then normal single character registers are used \f[I]unless\f[R] the character immediately following a command that needs a register name is a space (according to \f[B]isspace()\f[R]) and not a newline (\f[B]`\[rs]n'\f[R]). .PP In that case, the register name is found according to the regex \f[B][a\-z][a\-z0\-9_]*\f[R] (like bc(1) identifiers), and it is a parse error if the next non\-space characters do not match that regex. .SH RESET When dc(1) encounters an error or a signal that it has a non\-default handler for, it resets. This means that several things happen. .PP First, any macros that are executing are stopped and popped off the -stack. +execution stack. The behavior is not unlike that of exceptions in programming languages. Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. .PP +However, the stack of values is \f[I]not\f[R] cleared; in interactive +mode, users can inspect the stack and manipulate it. +.PP Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the \f[B]EXIT STATUS\f[R] section), it asks for more input; otherwise, it exits with the appropriate return code. .SH PERFORMANCE Most dc(1) implementations use \f[B]char\f[R] types to calculate the value of \f[B]1\f[R] decimal digit at a time, but that can be slow. This dc(1) does something different. .PP It uses large integers to calculate more than \f[B]1\f[R] decimal digit at a time. If built in a environment where \f[B]DC_LONG_BIT\f[R] (see the \f[B]LIMITS\f[R] section) is \f[B]64\f[R], then each integer has \f[B]9\f[R] decimal digits. If built in an environment where \f[B]DC_LONG_BIT\f[R] is \f[B]32\f[R] then each integer has \f[B]4\f[R] decimal digits. This value (the number of decimal digits per large integer) is called \f[B]DC_BASE_DIGS\f[R]. .PP In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of \f[B]DC_LONG_BIT\f[R], but is always at least twice as large as the integer type used to store digits. .SH LIMITS The following are the limits on dc(1): .TP \f[B]DC_LONG_BIT\f[R] The number of bits in the \f[B]long\f[R] type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the \f[B]PERFORMANCE\f[R] section). .TP \f[B]DC_BASE_DIGS\f[R] The number of decimal digits per large integer (see the \f[B]PERFORMANCE\f[R] section). Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_POW\f[R] The max decimal number that each large integer can store (see \f[B]DC_BASE_DIGS\f[R]) plus \f[B]1\f[R]. Depends on \f[B]DC_BASE_DIGS\f[R]. .TP \f[B]DC_OVERFLOW_MAX\f[R] The max number that the overflow type (see the \f[B]PERFORMANCE\f[R] section) can hold. Depends on \f[B]DC_LONG_BIT\f[R]. .TP \f[B]DC_BASE_MAX\f[R] The maximum output base. Set at \f[B]DC_BASE_POW\f[R]. .TP \f[B]DC_DIM_MAX\f[R] The maximum size of arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .TP \f[B]DC_SCALE_MAX\f[R] The maximum \f[B]scale\f[R]. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_STRING_MAX\f[R] The maximum length of strings. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NAME_MAX\f[R] The maximum length of identifiers. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_NUM_MAX\f[R] The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at \f[B]DC_OVERFLOW_MAX\-1\f[R]. .TP \f[B]DC_RAND_MAX\f[R] The maximum integer (inclusive) returned by the \f[B]\[cq]\f[R] command, if dc(1). Set at \f[B]2\[ha]DC_LONG_BIT\-1\f[R]. .TP Exponent The maximum allowable exponent (positive or negative). Set at \f[B]DC_OVERFLOW_MAX\f[R]. .TP Number of vars The maximum number of vars/arrays. Set at \f[B]SIZE_MAX\-1\f[R]. .PP These limits are meant to be effectively non\-existent; the limits are so large (at least on 64\-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. .SH ENVIRONMENT VARIABLES As \f[B]non\-portable extensions\f[R], dc(1) recognizes the following environment variables: .TP \f[B]DC_ENV_ARGS\f[R] This is another way to give command\-line arguments to dc(1). They should be in the same format as all other command\-line arguments. These are always processed first, so any files given in \f[B]DC_ENV_ARGS\f[R] will be processed before arguments and files given on the command\-line. This gives the user the ability to set up \[lq]standard\[rq] options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the \f[B]\-e\f[R] option to set \f[B]scale\f[R] to a value other than \f[B]0\f[R]. .RS .PP The code that parses \f[B]DC_ENV_ARGS\f[R] will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string \f[B]\[lq]/home/gavin/some dc file.dc\[rq]\f[R] will be correctly parsed, but the string \f[B]\[lq]/home/gavin/some \[dq]dc\[dq] file.dc\[rq]\f[R] will include the backslashes. .PP The quote parsing will handle either kind of quotes, \f[B]\[cq]\f[R] or \f[B]\[lq]\f[R]. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in \f[B]\[lq]some `dc' file.dc\[rq]\f[R], and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in \f[B]DC_ENV_ARGS\f[R] is not supported due to the complexity of the parsing, though such files are still supported on the command\-line where the parsing is done by the shell. .RE .TP \f[B]DC_LINE_LENGTH\f[R] If this environment variable exists and contains an integer that is greater than \f[B]1\f[R] and is less than \f[B]UINT16_MAX\f[R] (\f[B]2\[ha]16\-1\f[R]), dc(1) will output lines to that length, including the backslash newline combo. The default line length is \f[B]70\f[R]. .RS .PP The special value of \f[B]0\f[R] will disable line length checking and print numbers without regard to line length and without backslashes and newlines. .RE .TP \f[B]DC_SIGINT_RESET\f[R] If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), then this environment variable has no effect because dc(1) exits on \f[B]SIGINT\f[R] when not in interactive mode. .RS .PP However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) reset on \f[B]SIGINT\f[R], rather than exit, and zero makes dc(1) exit. If this environment variable exists and is \f[I]not\f[R] an integer, then dc(1) will exit on \f[B]SIGINT\f[R]. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_TTY_MODE\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non\-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_PROMPT\f[R] If TTY mode is \f[I]not\f[R] available (see the \f[B]TTY MODE\f[R] section), then this environment variable has no effect. .RS .PP However, when TTY mode is available, then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) use a prompt, and zero or a non\-integer makes dc(1) not use a prompt. If this environment variable does not exist and \f[B]DC_TTY_MODE\f[R] does, then the value of the \f[B]DC_TTY_MODE\f[R] environment variable is used. .PP This environment variable and the \f[B]DC_TTY_MODE\f[R] environment variable override the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_EXPR_EXIT\f[R] If any expressions or expression files are given on the command\-line with \f[B]\-e\f[R], \f[B]\-\-expression\f[R], \f[B]\-f\f[R], or \f[B]\-\-file\f[R], then if this environment variable exists and contains an integer, a non\-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. .RS .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .TP \f[B]DC_DIGIT_CLAMP\f[R] When parsing numbers and if this environment variable exists and contains an integer, a non\-zero value makes dc(1) clamp digits that are greater than or equal to the current \f[B]ibase\f[R] so that all such digits are considered equal to the \f[B]ibase\f[R] minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the \f[B]ibase\f[R]. .RS .PP This never applies to single\-digit numbers, as per the bc(1) standard (see the \f[B]STANDARDS\f[R] section). .PP This environment variable overrides the default, which can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .RE .SH EXIT STATUS dc(1) returns the following exit statuses: .TP \f[B]0\f[R] No error. .TP \f[B]1\f[R] A math error occurred. This follows standard practice of using \f[B]1\f[R] for expected errors, since math errors will happen in the process of normal execution. .RS .PP Math errors include divide by \f[B]0\f[R], taking the square root of a negative number, using a negative number as a bound for the pseudo\-random number generator, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non\-integer where an integer is required. .PP Converting to a hardware integer happens for the second operand of the power (\f[B]\[ha]\f[R]), places (\f[B]\[at]\f[R]), left shift (\f[B]H\f[R]), and right shift (\f[B]h\f[R]) operators. .RE .TP \f[B]2\f[R] A parse error occurred. .RS .PP Parse errors include unexpected \f[B]EOF\f[R], using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. .RE .TP \f[B]3\f[R] A runtime error occurred. .RS .PP Runtime errors include assigning an invalid number to any global (\f[B]ibase\f[R], \f[B]obase\f[R], or \f[B]scale\f[R]), giving a bad expression to a \f[B]read()\f[R] call, calling \f[B]read()\f[R] inside of a \f[B]read()\f[R] call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. .RE .TP \f[B]4\f[R] A fatal error occurred. .RS .PP Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command\-line options. .RE .PP The exit status \f[B]4\f[R] is special; when a fatal error occurs, dc(1) always exits and returns \f[B]4\f[R], no matter what mode dc(1) is in. .PP The other statuses will only be returned when dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), since dc(1) resets its state (see the \f[B]RESET\f[R] section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .PP These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the \f[B]\-i\f[R] flag or \f[B]\-\-interactive\f[R] option. .SH INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non\-interactive mode. Interactive mode is turned on automatically when both \f[B]stdin\f[R] and \f[B]stdout\f[R] are hooked to a terminal, but the \f[B]\-i\f[R] flag and \f[B]\-\-interactive\f[R] option can turn it on in other situations. .PP In interactive mode, dc(1) attempts to recover from errors (see the \f[B]RESET\f[R] section), and in normal execution, flushes \f[B]stdout\f[R] as soon as execution is done for the current input. dc(1) may also reset on \f[B]SIGINT\f[R] instead of exit, depending on the contents of, or default for, the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .SH TTY MODE If \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY, then \[lq]TTY mode\[rq] is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. .PP If there is the environment variable \f[B]DC_TTY_MODE\f[R] in the environment (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), then if that environment variable contains a non\-zero integer, dc(1) will turn on TTY mode when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY. If the \f[B]DC_TTY_MODE\f[R] environment variable exists but is \f[I]not\f[R] a non\-zero integer, then dc(1) will not turn TTY mode on. .PP If the environment variable \f[B]DC_TTY_MODE\f[R] does \f[I]not\f[R] exist, the default setting is used. The default setting can be queried with the \f[B]\-h\f[R] or \f[B]\-\-help\f[R] options. .PP TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the \f[B]STANDARDS\f[R] section), and interactive mode requires only \f[B]stdin\f[R] and \f[B]stdout\f[R] to be connected to a terminal. .SS Command\-Line History Command\-line history is only enabled if TTY mode is, i.e., that \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are connected to a TTY and the \f[B]DC_TTY_MODE\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section) and its default do not disable TTY mode. See the \f[B]COMMAND LINE HISTORY\f[R] section for more information. .SS Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: \f[B]DC_PROMPT\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP If the environment variable \f[B]DC_PROMPT\f[R] exists and is a non\-zero integer, then the prompt is turned on when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are connected to a TTY and the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options were not used. The read prompt will be turned on under the same conditions, except that the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options must also not be used. .PP However, if \f[B]DC_PROMPT\f[R] does not exist, the prompt can be enabled or disabled with the \f[B]DC_TTY_MODE\f[R] environment variable, the \f[B]\-P\f[R] and \f[B]\-\-no\-prompt\f[R] options, and the \f[B]\-R\f[R] and \f[B]\-\-no\-read\-prompt\f[R] options. See the \f[B]ENVIRONMENT VARIABLES\f[R] and \f[B]OPTIONS\f[R] sections for more details. .SH SIGNAL HANDLING Sending a \f[B]SIGINT\f[R] will cause dc(1) to do one of two things. .PP If dc(1) is not in interactive mode (see the \f[B]INTERACTIVE MODE\f[R] section), or the \f[B]DC_SIGINT_RESET\f[R] environment variable (see the \f[B]ENVIRONMENT VARIABLES\f[R] section), or its default, is either not an integer or it is zero, dc(1) will exit. .PP However, if dc(1) is in interactive mode, and the \f[B]DC_SIGINT_RESET\f[R] or its default is an integer and non\-zero, then dc(1) will stop executing the current input and reset (see the \f[B]RESET\f[R] section) upon receiving a \f[B]SIGINT\f[R]. .PP Note that \[lq]current input\[rq] can mean one of two things. If dc(1) is processing input from \f[B]stdin\f[R] in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from \f[B]stdin\f[R] if no other file exists. .PP This means that if a \f[B]SIGINT\f[R] is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. .PP \f[B]SIGTERM\f[R] and \f[B]SIGQUIT\f[R] cause dc(1) to clean up and exit, and it uses the default handler for all other signals. The one exception is \f[B]SIGHUP\f[R]; in that case, and only when dc(1) is in TTY mode (see the \f[B]TTY MODE\f[R] section), a \f[B]SIGHUP\f[R] will cause dc(1) to clean up and exit. .SH COMMAND LINE HISTORY dc(1) supports interactive command\-line editing. .PP If dc(1) can be in TTY mode (see the \f[B]TTY MODE\f[R] section), history can be enabled. This means that command\-line history can only be enabled when \f[B]stdin\f[R], \f[B]stdout\f[R], and \f[B]stderr\f[R] are all connected to a TTY. .PP Like TTY mode itself, it can be turned on or off with the environment variable \f[B]DC_TTY_MODE\f[R] (see the \f[B]ENVIRONMENT VARIABLES\f[R] section). .PP \f[B]Note\f[R]: tabs are converted to 8 spaces. .SH SEE ALSO bc(1) .SH STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1\-2017 (\[lq]POSIX.1\-2017\[rq]) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . .SH BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . .SH AUTHOR Gavin D. Howard \c .MT gavin@gavinhoward.com .ME \c \ and contributors. diff --git a/manuals/dc/N.1.md b/manuals/dc/N.1.md index 81933e2160b7..22ea9c96bc80 100644 --- a/manuals/dc/N.1.md +++ b/manuals/dc/N.1.md @@ -1,1521 +1,1524 @@ # Name dc - arbitrary-precision decimal reverse-Polish notation calculator # SYNOPSIS **dc** [**-cChiPRvVx**] [**-\-version**] [**-\-help**] [**-\-digit-clamp**] [**-\-no-digit-clamp**] [**-\-interactive**] [**-\-no-prompt**] [**-\-no-read-prompt**] [**-\-extended-register**] [**-e** *expr*] [**-\-expression**=*expr*...] [**-f** *file*...] [**-\-file**=*file*...] [*file*...] [**-I** *ibase*] [**-\-ibase**=*ibase*] [**-O** *obase*] [**-\-obase**=*obase*] [**-S** *scale*] [**-\-scale**=*scale*] [**-E** *seed*] [**-\-seed**=*seed*] # DESCRIPTION dc(1) is an arbitrary-precision calculator. It uses a stack (reverse Polish notation) to store numbers and results of computations. Arithmetic operations pop arguments off of the stack and push the results. If no files are given on the command-line, then dc(1) reads from **stdin** (see the **STDIN** section). Otherwise, those files are processed, and dc(1) will then exit. If a user wants to set up a standard environment, they can use **DC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). For example, if a user wants the **scale** always set to **10**, they can set **DC_ENV_ARGS** to **-e 10k**, and this dc(1) will always start with a **scale** of **10**. # OPTIONS The following are the options that dc(1) accepts. **-C**, **-\-no-digit-clamp** : Disables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that the value added to a number from a digit is always that digit's value multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-c** or **-\-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-c**, **-\-digit-clamp** : Enables clamping of digits greater than or equal to the current **ibase** when parsing numbers. This means that digits that the value added to a number from a digit that is greater than or equal to the ibase is the value of ibase minus 1 all multiplied by the value of ibase raised to the power of the digit's position, which starts from 0 at the least significant digit. If this and/or the **-C** or **-\-no-digit-clamp** options are given multiple times, the last one given is used. This option overrides the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section) and the default, which can be queried with the **-h** or **-\-help** options. This is a **non-portable extension**. **-E** *seed*, **-\-seed**=*seed* : Sets the builtin variable **seed** to the value *seed* assuming that *seed* is in base 10. It is a fatal error if *seed* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-e** *expr*, **-\-expression**=*expr* : Evaluates *expr*. If multiple expressions are given, they are evaluated in order. If files are given as well (see below), the expressions and files are evaluated in the order given. This means that if a file is given before an expression, the file is read in and evaluated first. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**, whether on the command-line or in **DC_ENV_ARGS**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-f** *file*, **-\-file**=*file* : Reads in *file* and evaluates it, line by line, as though it were read through **stdin**. If expressions are also given (see above), the expressions are evaluated in the order given. If this option is given on the command-line (i.e., not in **DC_ENV_ARGS**, see the **ENVIRONMENT VARIABLES** section), then after processing all expressions and files, dc(1) will exit, unless **-** (**stdin**) was given as an argument at least once to **-f** or **-\-file**. However, if any other **-e**, **-\-expression**, **-f**, or **-\-file** arguments are given after **-f-** or equivalent is given, dc(1) will give a fatal error and exit. This is a **non-portable extension**. **-h**, **-\-help** : Prints a usage message and exits. **-I** *ibase*, **-\-ibase**=*ibase* : Sets the builtin variable **ibase** to the value *ibase* assuming that *ibase* is in base 10. It is a fatal error if *ibase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-i**, **-\-interactive** : Forces interactive mode. (See the **INTERACTIVE MODE** section.) This is a **non-portable extension**. **-L**, **-\-no-line-length** : Disables line length checking and prints numbers without backslashes and newlines. In other words, this option sets **BC_LINE_LENGTH** to **0** (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-O** *obase*, **-\-obase**=*obase* : Sets the builtin variable **obase** to the value *obase* assuming that *obase* is in base 10. It is a fatal error if *obase* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-P**, **-\-no-prompt** : Disables the prompt in TTY mode. (The prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a prompt or are not used to having them in dc(1). Most of those users would want to put this option in **DC_ENV_ARGS**. These options override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section). This is a **non-portable extension**. **-R**, **-\-no-read-prompt** : Disables the read prompt in TTY mode. (The read prompt is only enabled in TTY mode. See the **TTY MODE** section.) This is mostly for those users that do not want a read prompt or are not used to having them in dc(1). Most of those users would want to put this option in **BC_ENV_ARGS** (see the **ENVIRONMENT VARIABLES** section). This option is also useful in hash bang lines of dc(1) scripts that prompt for user input. This option does not disable the regular prompt because the read prompt is only used when the **?** command is used. These options *do* override the **DC_PROMPT** and **DC_TTY_MODE** environment variables (see the **ENVIRONMENT VARIABLES** section), but only for the read prompt. This is a **non-portable extension**. **-S** *scale*, **-\-scale**=*scale* : Sets the builtin variable **scale** to the value *scale* assuming that *scale* is in base 10. It is a fatal error if *scale* is not a valid number. If multiple instances of this option are given, the last is used. This is a **non-portable extension**. **-v**, **-V**, **-\-version** : Print the version information (copyright header) and exits. **-x** **-\-extended-register** : Enables extended register mode. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. This is a **non-portable extension**. **-z**, **-\-leading-zeroes** : Makes dc(1) print all numbers greater than **-1** and less than **1**, and not equal to **0**, with a leading zero. This is a **non-portable extension**. All long options are **non-portable extensions**. # STDIN If no files are given on the command-line and no files or expressions are given by the **-f**, **-\-file**, **-e**, or **-\-expression** options, then dc(1) reads from **stdin**. However, there is a caveat to this. First, **stdin** is evaluated a line at a time. The only exception to this is if a string has been finished, but not ended. This means that, except for escaped brackets, all brackets must be balanced before dc(1) parses and executes. # STDOUT Any non-error output is written to **stdout**. In addition, if history (see the **HISTORY** section) and the prompt (see the **TTY MODE** section) are enabled, both are output to **stdout**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stdout**, so if **stdout** is closed, as in **dc >&-**, it will quit with an error. This is done so that dc(1) can report problems when **stdout** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stdout** to **/dev/null**. # STDERR Any error output is written to **stderr**. **Note**: Unlike other dc(1) implementations, this dc(1) will issue a fatal error (see the **EXIT STATUS** section) if it cannot write to **stderr**, so if **stderr** is closed, as in **dc 2>&-**, it will quit with an error. This is done so that dc(1) can exit with an error code when **stderr** is redirected to a file. If there are scripts that depend on the behavior of other dc(1) implementations, it is recommended that those scripts be changed to redirect **stderr** to **/dev/null**. # SYNTAX Each item in the input source code, either a number (see the **NUMBERS** section) or a command (see the **COMMANDS** section), is processed and executed, in order. Input is processed immediately when entered. **ibase** is a register (see the **REGISTERS** section) that determines how to interpret constant numbers. It is the "input" base, or the number base used for interpreting input numbers. **ibase** is initially **10**. The max allowable value for **ibase** is **16**. The min allowable value for **ibase** is **2**. The max allowable value for **ibase** can be queried in dc(1) programs with the **T** command. **obase** is a register (see the **REGISTERS** section) that determines how to output results. It is the "output" base, or the number base used for outputting numbers. **obase** is initially **10**. The max allowable value for **obase** is **DC_BASE_MAX** and can be queried with the **U** command. The min allowable value for **obase** is **0**. If **obase** is **0**, values are output in scientific notation, and if **obase** is **1**, values are output in engineering notation. Otherwise, values are output in the specified base. Outputting in scientific and engineering notations are **non-portable extensions**. The *scale* of an expression is the number of digits in the result of the expression right of the decimal point, and **scale** is a register (see the **REGISTERS** section) that sets the precision of any operations (with exceptions). **scale** is initially **0**. **scale** cannot be negative. The max allowable value for **scale** can be queried in dc(1) programs with the **V** command. **seed** is a register containing the current seed for the pseudo-random number generator. If the current value of **seed** is queried and stored, then if it is assigned to **seed** later, the pseudo-random number generator is guaranteed to produce the same sequence of pseudo-random numbers that were generated after the value of **seed** was first queried. Multiple values assigned to **seed** can produce the same sequence of pseudo-random numbers. Likewise, when a value is assigned to **seed**, it is not guaranteed that querying **seed** immediately after will return the same value. In addition, the value of **seed** will change after any call to the **'** command or the **"** command that does not get receive a value of **0** or **1**. The maximum integer returned by the **'** command can be queried with the **W** command. **Note**: The values returned by the pseudo-random number generator with the **'** and **"** commands are guaranteed to **NOT** be cryptographically secure. This is a consequence of using a seeded pseudo-random number generator. However, they *are* guaranteed to be reproducible with identical **seed** values. This means that the pseudo-random values from dc(1) should only be used where a reproducible stream of pseudo-random numbers is *ESSENTIAL*. In any other case, use a non-seeded pseudo-random number generator. The pseudo-random number generator, **seed**, and all associated operations are **non-portable extensions**. ## Comments Comments go from **#** until, and not including, the next newline. This is a **non-portable extension**. # NUMBERS Numbers are strings made up of digits, uppercase letters up to **F**, and at most **1** period for a radix. Numbers can have up to **DC_NUM_MAX** digits. Uppercase letters are equal to **9** plus their position in the alphabet (i.e., **A** equals **10**, or **9+1**). If a digit or letter makes no sense with the current value of **ibase** (i.e., they are greater than or equal to the current value of **ibase**), then the behavior depends on the existence of the **-c**/**-\-digit-clamp** or **-C**/**-\-no-digit-clamp** options (see the **OPTIONS** section), the existence and setting of the **DC_DIGIT_CLAMP** environment variable (see the **ENVIRONMENT VARIABLES** section), or the default, which can be queried with the **-h**/**-\-help** option. If clamping is off, then digits or letters that are greater than or equal to the current value of **ibase** are not changed. Instead, their given value is multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*A+3\^0\*B**, which is **3** times **10** plus **11**, or **41**. If clamping is on, then digits or letters that are greater than or equal to the current value of **ibase** are set to the value of the highest valid digit in **ibase** before being multiplied by the appropriate power of **ibase** and added into the number. This means that, with an **ibase** of **3**, the number **AB** is equal to **3\^1\*2+3\^0\*2**, which is **3** times **2** plus **2**, or **8**. There is one exception to clamping: single-character numbers (i.e., **A** alone). Such numbers are never clamped and always take the value they would have in the highest possible **ibase**. This means that **A** alone always equals decimal **10** and **Z** alone always equals decimal **35**. This behavior is mandated by the standard for bc(1) (see the STANDARDS section) and is meant to provide an easy way to set the current **ibase** (with the **i** command) regardless of the current value of **ibase**. If clamping is on, and the clamped value of a character is needed, use a leading zero, i.e., for **A**, use **0A**. In addition, dc(1) accepts numbers in scientific notation. These have the form **\e\**. The exponent (the portion after the **e**) must be an integer. An example is **1.89237e9**, which is equal to **1892370000**. Negative exponents are also allowed, so **4.2890e_3** is equal to **0.0042890**. **WARNING**: Both the number and the exponent in scientific notation are interpreted according to the current **ibase**, but the number is still multiplied by **10\^exponent** regardless of the current **ibase**. For example, if **ibase** is **16** and dc(1) is given the number string **FFeA**, the resulting decimal number will be **2550000000000**, and if dc(1) is given the number string **10e_4**, the resulting decimal number will be **0.0016**. Accepting input as scientific notation is a **non-portable extension**. # COMMANDS The valid commands are listed below. ## Printing These commands are used for printing. Note that both scientific notation and engineering notation are available for printing numbers. Scientific notation is activated by assigning **0** to **obase** using **0o**, and engineering notation is activated by assigning **1** to **obase** using **1o**. To deactivate them, just assign a different value to **obase**. Printing numbers in scientific notation and/or engineering notation is a **non-portable extension**. **p** : Prints the value on top of the stack, whether number or string, and prints a newline after. This does not alter the stack. **n** : Prints the value on top of the stack, whether number or string, and pops it off of the stack. **P** : Pops a value off the stack. If the value is a number, it is truncated and the absolute value of the result is printed as though **obase** is **256** and each digit is interpreted as an 8-bit ASCII character, making it a byte stream. If the value is a string, it is printed without a trailing newline. This is a **non-portable extension**. **f** : Prints the entire contents of the stack, in order from newest to oldest, without altering anything. Users should use this command when they get lost. ## Arithmetic These are the commands used for arithmetic. **+** : The top two values are popped off the stack, added, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **-** : The top two values are popped off the stack, subtracted, and the result is pushed onto the stack. The *scale* of the result is equal to the max *scale* of both operands. **\*** : The top two values are popped off the stack, multiplied, and the result is pushed onto the stack. If **a** is the *scale* of the first expression and **b** is the *scale* of the second expression, the *scale* of the result is equal to **min(a+b,max(scale,a,b))** where **min()** and **max()** return the obvious values. **/** : The top two values are popped off the stack, divided, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be non-zero. **%** : The top two values are popped off the stack, remaindered, and the result is pushed onto the stack. Remaindering is equivalent to 1) Computing **a/b** to current **scale**, and 2) Using the result of step 1 to calculate **a-(a/b)\*b** to *scale* **max(scale+scale(b),scale(a))**. The first value popped off of the stack must be non-zero. **~** : The top two values are popped off the stack, divided and remaindered, and the results (divided first, remainder second) are pushed onto the stack. This is equivalent to **x y / x y %** except that **x** and **y** are only evaluated once. The first value popped off of the stack must be non-zero. This is a **non-portable extension**. **\^** : The top two values are popped off the stack, the second is raised to the power of the first, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The first value popped off of the stack must be an integer, and if that value is negative, the second value popped off of the stack must be non-zero. **v** : The top value is popped off the stack, its square root is computed, and the result is pushed onto the stack. The *scale* of the result is equal to **scale**. The value popped off of the stack must be non-negative. **\_** : If this command *immediately* precedes a number (i.e., no spaces or other commands), then that number is input as a negative number. Otherwise, the top value on the stack is popped and copied, and the copy is negated and pushed onto the stack. This behavior without a number is a **non-portable extension**. **b** : The top value is popped off the stack, and if it is zero, it is pushed back onto the stack. Otherwise, its absolute value is pushed onto the stack. This is a **non-portable extension**. **|** : The top three values are popped off the stack, a modular exponentiation is computed, and the result is pushed onto the stack. The first value popped is used as the reduction modulus and must be an integer and non-zero. The second value popped is used as the exponent and must be an integer and non-negative. The third value popped is the base and must be an integer. This is a **non-portable extension**. **\$** : The top value is popped off the stack and copied, and the copy is truncated and pushed onto the stack. This is a **non-portable extension**. **\@** : The top two values are popped off the stack, and the precision of the second is set to the value of the first, whether by truncation or extension. The first value popped off of the stack must be an integer and non-negative. This is a **non-portable extension**. **H** : The top two values are popped off the stack, and the second is shifted left (radix shifted right) to the value of the first. The first value popped off of the stack must be an integer and non-negative. This is a **non-portable extension**. **h** : The top two values are popped off the stack, and the second is shifted right (radix shifted left) to the value of the first. The first value popped off of the stack must be an integer and non-negative. This is a **non-portable extension**. **G** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if they are equal, or **0** otherwise. This is a **non-portable extension**. **N** : The top value is popped off of the stack, and if it a **0**, a **1** is pushed; otherwise, a **0** is pushed. This is a **non-portable extension**. **(** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than the second, or **0** otherwise. This is a **non-portable extension**. **{** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is less than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **)** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than the second, or **0** otherwise. This is a **non-portable extension**. **}** : The top two values are popped off of the stack, they are compared, and a **1** is pushed if the first is greater than or equal to the second, or **0** otherwise. This is a **non-portable extension**. **M** : The top two values are popped off of the stack. If they are both non-zero, a **1** is pushed onto the stack. If either of them is zero, or both of them are, then a **0** is pushed onto the stack. This is like the **&&** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. **m** : The top two values are popped off of the stack. If at least one of them is non-zero, a **1** is pushed onto the stack. If both of them are zero, then a **0** is pushed onto the stack. This is like the **||** operator in bc(1), and it is *not* a short-circuit operator. This is a **non-portable extension**. ## Pseudo-Random Number Generator dc(1) has a built-in pseudo-random number generator. These commands query the pseudo-random number generator. (See Parameters for more information about the **seed** value that controls the pseudo-random number generator.) The pseudo-random number generator is guaranteed to **NOT** be cryptographically secure. **'** : Generates an integer between 0 and **DC_RAND_MAX**, inclusive (see the **LIMITS** section). The generated integer is made as unbiased as possible, subject to the limitations of the pseudo-random number generator. This is a **non-portable extension**. **"** : Pops a value off of the stack, which is used as an **exclusive** upper bound on the integer that will be generated. If the bound is negative or is a non-integer, an error is raised, and dc(1) resets (see the **RESET** section) while **seed** remains unchanged. If the bound is larger than **DC_RAND_MAX**, the higher bound is honored by generating several pseudo-random integers, multiplying them by appropriate powers of **DC_RAND_MAX+1**, and adding them together. Thus, the size of integer that can be generated with this command is unbounded. Using this command will change the value of **seed**, unless the operand is **0** or **1**. In that case, **0** is pushed onto the stack, and **seed** is *not* changed. The generated integer is made as unbiased as possible, subject to the limitations of the pseudo-random number generator. This is a **non-portable extension**. ## Stack Control These commands control the stack. **c** : Removes all items from ("clears") the stack. **d** : Copies the item on top of the stack ("duplicates") and pushes the copy onto the stack. **r** : Swaps ("reverses") the two top items on the stack. **R** : Pops ("removes") the top value from the stack. ## Register Control These commands control registers (see the **REGISTERS** section). **s**_r_ : Pops the value off the top of the stack and stores it into register *r*. **l**_r_ : Copies the value in register *r* and pushes it onto the stack. This does not alter the contents of *r*. **S**_r_ : Pops the value off the top of the (main) stack and pushes it onto the stack of register *r*. The previous value of the register becomes inaccessible. **L**_r_ : Pops the value off the top of the stack for register *r* and push it onto the main stack. The previous value in the stack for register *r*, if any, is now accessible via the **l**_r_ command. ## Parameters These commands control the values of **ibase**, **obase**, **scale**, and **seed**. Also see the **SYNTAX** section. **i** : Pops the value off of the top of the stack and uses it to set **ibase**, which must be between **2** and **16**, inclusive. If the value on top of the stack has any *scale*, the *scale* is ignored. **o** : Pops the value off of the top of the stack and uses it to set **obase**, which must be between **0** and **DC_BASE_MAX**, inclusive (see the **LIMITS** section and the **NUMBERS** section). If the value on top of the stack has any *scale*, the *scale* is ignored. **k** : Pops the value off of the top of the stack and uses it to set **scale**, which must be non-negative. If the value on top of the stack has any *scale*, the *scale* is ignored. **j** : Pops the value off of the top of the stack and uses it to set **seed**. The meaning of **seed** is dependent on the current pseudo-random number generator but is guaranteed to not change except for new major versions. The *scale* and sign of the value may be significant. If a previously used **seed** value is used again, the pseudo-random number generator is guaranteed to produce the same sequence of pseudo-random numbers as it did when the **seed** value was previously used. The exact value assigned to **seed** is not guaranteed to be returned if the **J** command is used. However, if **seed** *does* return a different value, both values, when assigned to **seed**, are guaranteed to produce the same sequence of pseudo-random numbers. This means that certain values assigned to **seed** will not produce unique sequences of pseudo-random numbers. There is no limit to the length (number of significant decimal digits) or *scale* of the value that can be assigned to **seed**. This is a **non-portable extension**. **I** : Pushes the current value of **ibase** onto the main stack. **O** : Pushes the current value of **obase** onto the main stack. **K** : Pushes the current value of **scale** onto the main stack. **J** : Pushes the current value of **seed** onto the main stack. This is a **non-portable extension**. **T** : Pushes the maximum allowable value of **ibase** onto the main stack. This is a **non-portable extension**. **U** : Pushes the maximum allowable value of **obase** onto the main stack. This is a **non-portable extension**. **V** : Pushes the maximum allowable value of **scale** onto the main stack. This is a **non-portable extension**. **W** : Pushes the maximum (inclusive) integer that can be generated with the **'** pseudo-random number generator command. This is a **non-portable extension**. ## Strings The following commands control strings. dc(1) can work with both numbers and strings, and registers (see the **REGISTERS** section) can hold both strings and numbers. dc(1) always knows whether the contents of a register are a string or a number. While arithmetic operations have to have numbers, and will print an error if given a string, other commands accept strings. Strings can also be executed as macros. For example, if the string **[1pR]** is executed as a macro, then the code **1pR** is executed, meaning that the **1** will be printed with a newline after and then popped from the stack. **\[**_characters_**\]** : Makes a string containing *characters* and pushes it onto the stack. If there are brackets (**\[** and **\]**) in the string, then they must be balanced. Unbalanced brackets can be escaped using a backslash (**\\**) character. If there is a backslash character in the string, the character after it (even another backslash) is put into the string verbatim, but the (first) backslash is not. **a** : The value on top of the stack is popped. If it is a number, it is truncated and its absolute value is taken. The result mod **256** is calculated. If that result is **0**, push an empty string; otherwise, push a one-character string where the character is the result of the mod interpreted as an ASCII character. If it is a string, then a new string is made. If the original string is empty, the new string is empty. If it is not, then the first character of the original string is used to create the new string as a one-character string. The new string is then pushed onto the stack. This is a **non-portable extension**. **x** : Pops a value off of the top of the stack. If it is a number, it is pushed back onto the stack. If it is a string, it is executed as a macro. This behavior is the norm whenever a macro is executed, whether by this command or by the conditional execution commands below. **\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is greater than the second, then the contents of register *r* are executed. For example, **0 1>a** will execute the contents of register **a**, and **1 0>a** will not. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\>**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not greater than the second (less than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\>**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is less than the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!\<**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not less than the second (greater than or equal to), then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!\<**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **!=**_r_ : Pops two values off of the stack that must be numbers and compares them. If the first value is not equal to the second, then the contents of register *r* are executed. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). **!=**_r_**e**_s_ : Like the above, but will execute register *s* if the comparison fails. If either or both of the values are not numbers, dc(1) will raise an error and reset (see the **RESET** section). This is a **non-portable extension**. **?** : Reads a line from the **stdin** and executes it. This is to allow macros to request input from users. **q** : During execution of a macro, this exits the execution of that macro and the execution of the macro that executed it. If there are no macros, or only one macro executing, dc(1) exits. **Q** : Pops a value from the stack which must be non-negative and is used the number of macro executions to pop off of the execution stack. If the number of levels to pop is greater than the number of executing macros, dc(1) exits. **,** : Pushes the depth of the execution stack onto the stack. The execution stack is the stack of string executions. The number that is pushed onto the stack is exactly as many as is needed to make dc(1) exit with the **Q** command, so the sequence **,Q** will make dc(1) exit. This is a **non-portable extension**. ## Status These commands query status of the stack or its top value. **Z** : Pops a value off of the stack. If it is a number, calculates the number of significant decimal digits it has and pushes the result. It will push **1** if the argument is **0** with no decimal places. If it is a string, pushes the number of characters the string has. **X** : Pops a value off of the stack. If it is a number, pushes the *scale* of the value onto the stack. If it is a string, pushes **0**. **u** : Pops one value off of the stack. If the value is a number, this pushes **1** onto the stack. Otherwise (if it is a string), it pushes **0**. This is a **non-portable extension**. **t** : Pops one value off of the stack. If the value is a string, this pushes **1** onto the stack. Otherwise (if it is a number), it pushes **0**. This is a **non-portable extension**. **z** : Pushes the current depth of the stack (before execution of this command) onto the stack. **y**_r_ : Pushes the current stack depth of the register *r* onto the main stack. Because each register has a depth of **1** (with the value **0** in the top item) when dc(1) starts, dc(1) requires that each register's stack must always have at least one item; dc(1) will give an error and reset otherwise (see the **RESET** section). This means that this command will never push **0**. This is a **non-portable extension**. ## Arrays These commands manipulate arrays. **:**_r_ : Pops the top two values off of the stack. The second value will be stored in the array *r* (see the **REGISTERS** section), indexed by the first value. **;**_r_ : Pops the value on top of the stack and uses it as an index into the array *r*. The selected value is then pushed onto the stack. **Y**_r_ : Pushes the length of the array *r* onto the stack. This is a **non-portable extension**. ## Global Settings These commands retrieve global settings. These are the only commands that require multiple specific characters, and all of them begin with the letter **g**. Only the characters below are allowed after the character **g**; any other character produces a parse error (see the **ERRORS** section). **gl** : Pushes the line length set by **DC_LINE_LENGTH** (see the **ENVIRONMENT VARIABLES** section) onto the stack. **gx** : Pushes **1** onto the stack if extended register mode is on, **0** otherwise. See the *Extended Register Mode* subsection of the **REGISTERS** section for more information. **gz** : Pushes **0** onto the stack if the leading zero setting has not been enabled with the **-z** or **-\-leading-zeroes** options (see the **OPTIONS** section), non-zero otherwise. # REGISTERS Registers are names that can store strings, numbers, and arrays. (Number/string registers do not interfere with array registers.) Each register is also its own stack, so the current register value is the top of the stack for the register. All registers, when first referenced, have one value (**0**) in their stack, and it is a runtime error to attempt to pop that item off of the register stack. In non-extended register mode, a register name is just the single character that follows any command that needs a register name. The only exceptions are: a newline (**'\\n'**) and a left bracket (**'['**); it is a parse error for a newline or a left bracket to be used as a register name. ## Extended Register Mode Unlike most other dc(1) implentations, this dc(1) provides nearly unlimited amounts of registers, if extended register mode is enabled. If extended register mode is enabled (**-x** or **-\-extended-register** command-line arguments are given), then normal single character registers are used *unless* the character immediately following a command that needs a register name is a space (according to **isspace()**) and not a newline (**'\\n'**). In that case, the register name is found according to the regex **\[a-z\]\[a-z0-9\_\]\*** (like bc(1) identifiers), and it is a parse error if the next non-space characters do not match that regex. # RESET When dc(1) encounters an error or a signal that it has a non-default handler for, it resets. This means that several things happen. -First, any macros that are executing are stopped and popped off the stack. -The behavior is not unlike that of exceptions in programming languages. Then -the execution point is set so that any code waiting to execute (after all +First, any macros that are executing are stopped and popped off the execution +stack. The behavior is not unlike that of exceptions in programming languages. +Then the execution point is set so that any code waiting to execute (after all macros returned) is skipped. +However, the stack of values is *not* cleared; in interactive mode, users can +inspect the stack and manipulate it. + Thus, when dc(1) resets, it skips any remaining code waiting to be executed. Then, if it is interactive mode, and the error was not a fatal error (see the **EXIT STATUS** section), it asks for more input; otherwise, it exits with the appropriate return code. # PERFORMANCE Most dc(1) implementations use **char** types to calculate the value of **1** decimal digit at a time, but that can be slow. This dc(1) does something different. It uses large integers to calculate more than **1** decimal digit at a time. If built in a environment where **DC_LONG_BIT** (see the **LIMITS** section) is **64**, then each integer has **9** decimal digits. If built in an environment where **DC_LONG_BIT** is **32** then each integer has **4** decimal digits. This value (the number of decimal digits per large integer) is called **DC_BASE_DIGS**. In addition, this dc(1) uses an even larger integer for overflow checking. This integer type depends on the value of **DC_LONG_BIT**, but is always at least twice as large as the integer type used to store digits. # LIMITS The following are the limits on dc(1): **DC_LONG_BIT** : The number of bits in the **long** type in the environment where dc(1) was built. This determines how many decimal digits can be stored in a single large integer (see the **PERFORMANCE** section). **DC_BASE_DIGS** : The number of decimal digits per large integer (see the **PERFORMANCE** section). Depends on **DC_LONG_BIT**. **DC_BASE_POW** : The max decimal number that each large integer can store (see **DC_BASE_DIGS**) plus **1**. Depends on **DC_BASE_DIGS**. **DC_OVERFLOW_MAX** : The max number that the overflow type (see the **PERFORMANCE** section) can hold. Depends on **DC_LONG_BIT**. **DC_BASE_MAX** : The maximum output base. Set at **DC_BASE_POW**. **DC_DIM_MAX** : The maximum size of arrays. Set at **SIZE_MAX-1**. **DC_SCALE_MAX** : The maximum **scale**. Set at **DC_OVERFLOW_MAX-1**. **DC_STRING_MAX** : The maximum length of strings. Set at **DC_OVERFLOW_MAX-1**. **DC_NAME_MAX** : The maximum length of identifiers. Set at **DC_OVERFLOW_MAX-1**. **DC_NUM_MAX** : The maximum length of a number (in decimal digits), which includes digits after the decimal point. Set at **DC_OVERFLOW_MAX-1**. **DC_RAND_MAX** : The maximum integer (inclusive) returned by the **'** command, if dc(1). Set at **2\^DC_LONG_BIT-1**. Exponent : The maximum allowable exponent (positive or negative). Set at **DC_OVERFLOW_MAX**. Number of vars : The maximum number of vars/arrays. Set at **SIZE_MAX-1**. These limits are meant to be effectively non-existent; the limits are so large (at least on 64-bit machines) that there should not be any point at which they become a problem. In fact, memory should be exhausted before these limits should be hit. # ENVIRONMENT VARIABLES As **non-portable extensions**, dc(1) recognizes the following environment variables: **DC_ENV_ARGS** : This is another way to give command-line arguments to dc(1). They should be in the same format as all other command-line arguments. These are always processed first, so any files given in **DC_ENV_ARGS** will be processed before arguments and files given on the command-line. This gives the user the ability to set up "standard" options and files to be used at every invocation. The most useful thing for such files to contain would be useful functions that the user might want every time dc(1) runs. Another use would be to use the **-e** option to set **scale** to a value other than **0**. The code that parses **DC_ENV_ARGS** will correctly handle quoted arguments, but it does not understand escape sequences. For example, the string **"/home/gavin/some dc file.dc"** will be correctly parsed, but the string **"/home/gavin/some \"dc\" file.dc"** will include the backslashes. The quote parsing will handle either kind of quotes, **'** or **"**. Thus, if you have a file with any number of single quotes in the name, you can use double quotes as the outside quotes, as in **"some 'dc' file.dc"**, and vice versa if you have a file with double quotes. However, handling a file with both kinds of quotes in **DC_ENV_ARGS** is not supported due to the complexity of the parsing, though such files are still supported on the command-line where the parsing is done by the shell. **DC_LINE_LENGTH** : If this environment variable exists and contains an integer that is greater than **1** and is less than **UINT16_MAX** (**2\^16-1**), dc(1) will output lines to that length, including the backslash newline combo. The default line length is **70**. The special value of **0** will disable line length checking and print numbers without regard to line length and without backslashes and newlines. **DC_SIGINT_RESET** : If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), then this environment variable has no effect because dc(1) exits on **SIGINT** when not in interactive mode. However, when dc(1) is in interactive mode, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) reset on **SIGINT**, rather than exit, and zero makes dc(1) exit. If this environment variable exists and is *not* an integer, then dc(1) will exit on **SIGINT**. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_TTY_MODE** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, then a non-zero value makes dc(1) use TTY mode, and zero makes dc(1) not use TTY mode. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_PROMPT** : If TTY mode is *not* available (see the **TTY MODE** section), then this environment variable has no effect. However, when TTY mode is available, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) use a prompt, and zero or a non-integer makes dc(1) not use a prompt. If this environment variable does not exist and **DC_TTY_MODE** does, then the value of the **DC_TTY_MODE** environment variable is used. This environment variable and the **DC_TTY_MODE** environment variable override the default, which can be queried with the **-h** or **-\-help** options. **DC_EXPR_EXIT** : If any expressions or expression files are given on the command-line with **-e**, **-\-expression**, **-f**, or **-\-file**, then if this environment variable exists and contains an integer, a non-zero value makes dc(1) exit after executing the expressions and expression files, and a zero value makes dc(1) not exit. This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. **DC_DIGIT_CLAMP** : When parsing numbers and if this environment variable exists and contains an integer, a non-zero value makes dc(1) clamp digits that are greater than or equal to the current **ibase** so that all such digits are considered equal to the **ibase** minus 1, and a zero value disables such clamping so that those digits are always equal to their value, which is multiplied by the power of the **ibase**. This never applies to single-digit numbers, as per the bc(1) standard (see the **STANDARDS** section). This environment variable overrides the default, which can be queried with the **-h** or **-\-help** options. # EXIT STATUS dc(1) returns the following exit statuses: **0** : No error. **1** : A math error occurred. This follows standard practice of using **1** for expected errors, since math errors will happen in the process of normal execution. Math errors include divide by **0**, taking the square root of a negative number, using a negative number as a bound for the pseudo-random number generator, attempting to convert a negative number to a hardware integer, overflow when converting a number to a hardware integer, overflow when calculating the size of a number, and attempting to use a non-integer where an integer is required. Converting to a hardware integer happens for the second operand of the power (**\^**), places (**\@**), left shift (**H**), and right shift (**h**) operators. **2** : A parse error occurred. Parse errors include unexpected **EOF**, using an invalid character, failing to find the end of a string or comment, and using a token where it is invalid. **3** : A runtime error occurred. Runtime errors include assigning an invalid number to any global (**ibase**, **obase**, or **scale**), giving a bad expression to a **read()** call, calling **read()** inside of a **read()** call, type errors (including attempting to execute a number), and attempting an operation when the stack has too few elements. **4** : A fatal error occurred. Fatal errors include memory allocation errors, I/O errors, failing to open files, attempting to use files that do not have only ASCII characters (dc(1) only accepts ASCII characters), attempting to open a directory as a file, and giving invalid command-line options. The exit status **4** is special; when a fatal error occurs, dc(1) always exits and returns **4**, no matter what mode dc(1) is in. The other statuses will only be returned when dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), since dc(1) resets its state (see the **RESET** section) and accepts more input when one of those errors occurs in interactive mode. This is also the case when interactive mode is forced by the **-i** flag or **-\-interactive** option. These exit statuses allow dc(1) to be used in shell scripting with error checking, and its normal behavior can be forced by using the **-i** flag or **-\-interactive** option. # INTERACTIVE MODE Like bc(1), dc(1) has an interactive mode and a non-interactive mode. Interactive mode is turned on automatically when both **stdin** and **stdout** are hooked to a terminal, but the **-i** flag and **-\-interactive** option can turn it on in other situations. In interactive mode, dc(1) attempts to recover from errors (see the **RESET** section), and in normal execution, flushes **stdout** as soon as execution is done for the current input. dc(1) may also reset on **SIGINT** instead of exit, depending on the contents of, or default for, the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section). # TTY MODE If **stdin**, **stdout**, and **stderr** are all connected to a TTY, then "TTY mode" is considered to be available, and thus, dc(1) can turn on TTY mode, subject to some settings. If there is the environment variable **DC_TTY_MODE** in the environment (see the **ENVIRONMENT VARIABLES** section), then if that environment variable contains a non-zero integer, dc(1) will turn on TTY mode when **stdin**, **stdout**, and **stderr** are all connected to a TTY. If the **DC_TTY_MODE** environment variable exists but is *not* a non-zero integer, then dc(1) will not turn TTY mode on. If the environment variable **DC_TTY_MODE** does *not* exist, the default setting is used. The default setting can be queried with the **-h** or **-\-help** options. TTY mode is different from interactive mode because interactive mode is required in the bc(1) specification (see the **STANDARDS** section), and interactive mode requires only **stdin** and **stdout** to be connected to a terminal. ## Command-Line History Command-line history is only enabled if TTY mode is, i.e., that **stdin**, **stdout**, and **stderr** are connected to a TTY and the **DC_TTY_MODE** environment variable (see the **ENVIRONMENT VARIABLES** section) and its default do not disable TTY mode. See the **COMMAND LINE HISTORY** section for more information. ## Prompt If TTY mode is available, then a prompt can be enabled. Like TTY mode itself, it can be turned on or off with an environment variable: **DC_PROMPT** (see the **ENVIRONMENT VARIABLES** section). If the environment variable **DC_PROMPT** exists and is a non-zero integer, then the prompt is turned on when **stdin**, **stdout**, and **stderr** are connected to a TTY and the **-P** and **-\-no-prompt** options were not used. The read prompt will be turned on under the same conditions, except that the **-R** and **-\-no-read-prompt** options must also not be used. However, if **DC_PROMPT** does not exist, the prompt can be enabled or disabled with the **DC_TTY_MODE** environment variable, the **-P** and **-\-no-prompt** options, and the **-R** and **-\-no-read-prompt** options. See the **ENVIRONMENT VARIABLES** and **OPTIONS** sections for more details. # SIGNAL HANDLING Sending a **SIGINT** will cause dc(1) to do one of two things. If dc(1) is not in interactive mode (see the **INTERACTIVE MODE** section), or the **DC_SIGINT_RESET** environment variable (see the **ENVIRONMENT VARIABLES** section), or its default, is either not an integer or it is zero, dc(1) will exit. However, if dc(1) is in interactive mode, and the **DC_SIGINT_RESET** or its default is an integer and non-zero, then dc(1) will stop executing the current input and reset (see the **RESET** section) upon receiving a **SIGINT**. Note that "current input" can mean one of two things. If dc(1) is processing input from **stdin** in interactive mode, it will ask for more input. If dc(1) is processing input from a file in interactive mode, it will stop processing the file and start processing the next file, if one exists, or ask for input from **stdin** if no other file exists. This means that if a **SIGINT** is sent to dc(1) as it is executing a file, it can seem as though dc(1) did not respond to the signal since it will immediately start executing the next file. This is by design; most files that users execute when interacting with dc(1) have function definitions, which are quick to parse. If a file takes a long time to execute, there may be a bug in that file. The rest of the files could still be executed without problem, allowing the user to continue. **SIGTERM** and **SIGQUIT** cause dc(1) to clean up and exit, and it uses the default handler for all other signals. The one exception is **SIGHUP**; in that case, and only when dc(1) is in TTY mode (see the **TTY MODE** section), a **SIGHUP** will cause dc(1) to clean up and exit. # COMMAND LINE HISTORY dc(1) supports interactive command-line editing. If dc(1) can be in TTY mode (see the **TTY MODE** section), history can be enabled. This means that command-line history can only be enabled when **stdin**, **stdout**, and **stderr** are all connected to a TTY. Like TTY mode itself, it can be turned on or off with the environment variable **DC_TTY_MODE** (see the **ENVIRONMENT VARIABLES** section). **Note**: tabs are converted to 8 spaces. # SEE ALSO bc(1) # STANDARDS The dc(1) utility operators and some behavior are compliant with the operators in the IEEE Std 1003.1-2017 (“POSIX.1-2017”) bc(1) specification at https://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html . # BUGS None are known. Report bugs at https://git.gavinhoward.com/gavin/bc . # AUTHOR Gavin D. Howard and contributors. diff --git a/src/dc.c b/scripts/os.c similarity index 71% copy from src/dc.c copy to scripts/os.c index 992efe262fd8..212a61772ccf 100644 --- a/src/dc.c +++ b/scripts/os.c @@ -1,64 +1,59 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * - * The main procedure of dc. + * File for testing compilation on different platforms. * */ -#if DC_ENABLED +// This is used by configure.sh to test for OpenBSD. +#ifdef BC_TEST_OPENBSD +#ifdef __OpenBSD__ +#error On OpenBSD without _BSD_SOURCE +#endif // __OpenBSD__ +#endif // BC_TEST_OPENBSD -#include +// This is used by configure.sh to test for FreeBSD. +#ifdef BC_TEST_FREEBSD +#ifdef __FreeBSD__ +#error On FreeBSD with _POSIX_C_SOURCE +#endif // __FreeBSD__ +#endif // BC_TEST_FREEBSD -#include -#include +// This is used by configure.sh to test for macOS. +#ifdef BC_TEST_APPLE +#ifdef __APPLE__ +#error On macOS without _DARWIN_C_SOURCE +#endif // __APPLE__ +#endif // BC_TEST_APPLE -/** - * The main function for dc. - * @param argc The number of arguments. - * @param argv The arguments. - */ -BcStatus -dc_main(int argc, char* argv[]) -{ - // All of these just set dc-specific items in BcVm. - - vm->read_ret = BC_INST_POP_EXEC; - vm->help = dc_help; - vm->sigmsg = dc_sig_msg; - vm->siglen = dc_sig_msg_len; - - vm->next = dc_lex_token; - vm->parse = dc_parse_parse; - vm->expr = dc_parse_expr; +extern int test; - return bc_vm_boot(argc, argv); -} -#endif // DC_ENABLED +int test; diff --git a/src/args.c b/src/args.c index 635c7227d3de..6eba802d34ac 100644 --- a/src/args.c +++ b/src/args.c @@ -1,405 +1,405 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * Code for processing command-line arguments. * */ #include #include #include #include #include #ifndef _WIN32 #include #endif // _WIN32 #include #include #include #include #include #include /** * Adds @a str to the list of expressions to execute later. * @param str The string to add to the list of expressions. */ static void bc_args_exprs(const char* str) { BC_SIG_ASSERT_LOCKED; if (vm->exprs.v == NULL) { bc_vec_init(&vm->exprs, sizeof(uchar), BC_DTOR_NONE); } bc_vec_concat(&vm->exprs, str); bc_vec_concat(&vm->exprs, "\n"); } /** * Adds the contents of @a file to the list of expressions to execute later. * @param file The name of the file whose contents should be added to the list * of expressions to execute. */ static void bc_args_file(const char* file) { char* buf; BC_SIG_ASSERT_LOCKED; vm->file = file; buf = bc_read_file(file); assert(buf != NULL); bc_args_exprs(buf); free(buf); } static BcBigDig bc_args_builtin(const char* arg) { bool strvalid; BcNum n; BcBigDig res; strvalid = bc_num_strValid(arg); if (BC_ERR(!strvalid)) { bc_verr(BC_ERR_FATAL_ARG, arg); } bc_num_init(&n, 0); bc_num_parse(&n, arg, 10); res = bc_num_bigdig(&n); bc_num_free(&n); return res; } #if BC_ENABLED /** * Redefines a keyword, if it exists and is not a POSIX keyword. Otherwise, it * throws a fatal error. * @param keyword The keyword to redefine. */ static void bc_args_redefine(const char* keyword) { size_t i; BC_SIG_ASSERT_LOCKED; for (i = 0; i < bc_lex_kws_len; ++i) { const BcLexKeyword* kw = bc_lex_kws + i; if (!strcmp(keyword, kw->name)) { if (BC_LEX_KW_POSIX(kw)) break; vm->redefined_kws[i] = true; return; } } bc_error(BC_ERR_FATAL_ARG, 0, keyword); } #endif // BC_ENABLED void -bc_args(int argc, char* argv[], bool exit_exprs, BcBigDig* scale, +bc_args(int argc, const char* argv[], bool exit_exprs, BcBigDig* scale, BcBigDig* ibase, BcBigDig* obase) { int c; size_t i; bool do_exit = false, version = false; BcOpt opts; #if BC_ENABLE_EXTRA_MATH - char* seed = NULL; + const char* seed = NULL; #endif // BC_ENABLE_EXTRA_MATH BC_SIG_ASSERT_LOCKED; bc_opt_init(&opts, argv); // This loop should look familiar to anyone who has used getopt() or // getopt_long() in C. while ((c = bc_opt_parse(&opts, bc_args_lopt)) != -1) { switch (c) { case 'c': { vm->flags |= BC_FLAG_DIGIT_CLAMP; break; } case 'C': { vm->flags &= ~BC_FLAG_DIGIT_CLAMP; break; } case 'e': { // Barf if not allowed. if (vm->no_exprs) { bc_verr(BC_ERR_FATAL_OPTION, "-e (--expression)"); } // Add the expressions and set exit. bc_args_exprs(opts.optarg); vm->exit_exprs = (exit_exprs || vm->exit_exprs); break; } case 'f': { // Figure out if exiting on expressions is disabled. if (!strcmp(opts.optarg, "-")) vm->no_exprs = true; else { // Barf if not allowed. if (vm->no_exprs) { bc_verr(BC_ERR_FATAL_OPTION, "-f (--file)"); } // Add the expressions and set exit. bc_args_file(opts.optarg); vm->exit_exprs = (exit_exprs || vm->exit_exprs); } break; } case 'h': { bc_vm_info(vm->help); do_exit = true; break; } case 'i': { vm->flags |= BC_FLAG_I; break; } case 'I': { *ibase = bc_args_builtin(opts.optarg); break; } case 'z': { vm->flags |= BC_FLAG_Z; break; } case 'L': { vm->line_len = 0; break; } case 'O': { *obase = bc_args_builtin(opts.optarg); break; } case 'P': { vm->flags &= ~(BC_FLAG_P); break; } case 'R': { vm->flags &= ~(BC_FLAG_R); break; } case 'S': { *scale = bc_args_builtin(opts.optarg); break; } #if BC_ENABLE_EXTRA_MATH case 'E': { if (BC_ERR(!bc_num_strValid(opts.optarg))) { bc_verr(BC_ERR_FATAL_ARG, opts.optarg); } seed = opts.optarg; break; } #endif // BC_ENABLE_EXTRA_MATH #if BC_ENABLED case 'g': { assert(BC_IS_BC); vm->flags |= BC_FLAG_G; break; } case 'l': { assert(BC_IS_BC); vm->flags |= BC_FLAG_L; break; } case 'q': { assert(BC_IS_BC); vm->flags &= ~(BC_FLAG_Q); break; } case 'r': { bc_args_redefine(opts.optarg); break; } case 's': { assert(BC_IS_BC); vm->flags |= BC_FLAG_S; break; } case 'w': { assert(BC_IS_BC); vm->flags |= BC_FLAG_W; break; } #endif // BC_ENABLED case 'V': case 'v': { do_exit = version = true; break; } #if DC_ENABLED case 'x': { assert(BC_IS_DC); vm->flags |= DC_FLAG_X; break; } #endif // DC_ENABLED #if BC_DEBUG // We shouldn't get here because bc_opt_error()/bc_error() should // longjmp() out. case '?': case ':': default: { BC_UNREACHABLE #if !BC_CLANG abort(); #endif // !BC_CLANG } #endif // BC_DEBUG } } if (version) bc_vm_info(NULL); if (do_exit) { vm->status = (sig_atomic_t) BC_STATUS_QUIT; BC_JMP; } // We do not print the banner if expressions are used or dc is used. if (BC_ARGS_SHOULD_BE_QUIET) vm->flags &= ~(BC_FLAG_Q); // We need to make sure the files list is initialized. We don't want to // initialize it if there are no files because it's just a waste of memory. if (opts.optind < (size_t) argc && vm->files.v == NULL) { bc_vec_init(&vm->files, sizeof(char*), BC_DTOR_NONE); } // Add all the files to the vector. for (i = opts.optind; i < (size_t) argc; ++i) { bc_vec_push(&vm->files, argv + i); } #if BC_ENABLE_EXTRA_MATH if (seed != NULL) { BcNum n; bc_num_init(&n, strlen(seed)); BC_SIG_UNLOCK; bc_num_parse(&n, seed, BC_BASE); bc_program_assignSeed(&vm->prog, &n); BC_SIG_LOCK; bc_num_free(&n); } #endif // BC_ENABLE_EXTRA_MATH } diff --git a/src/bc.c b/src/bc.c index c5a67f35e109..572e42b1a16d 100644 --- a/src/bc.c +++ b/src/bc.c @@ -1,64 +1,65 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * The main procedure of bc. * */ #if BC_ENABLED #include #include #include /** * The main function for bc. * @param argc The number of arguments. * @param argv The arguments. */ BcStatus -bc_main(int argc, char* argv[]) +bc_main(int argc, const char* argv[]) { // All of these just set bc-specific items in BcVm. vm->read_ret = BC_INST_RET; vm->help = bc_help; vm->sigmsg = bc_sig_msg; vm->siglen = bc_sig_msg_len; vm->next = bc_lex_token; vm->parse = bc_parse_parse; vm->expr = bc_parse_expr; return bc_vm_boot(argc, argv); } + #endif // BC_ENABLED diff --git a/src/main.c b/src/bc_fuzzer.c similarity index 58% copy from src/main.c copy to src/bc_fuzzer.c index a6d50614af57..7d7b3292b727 100644 --- a/src/main.c +++ b/src/bc_fuzzer.c @@ -1,119 +1,112 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * - * The entry point for bc. + * The entry point for libFuzzer when fuzzing bc. * */ -#include -#include -#include - -#if BC_ENABLE_NLS -#include -#endif // BC_ENABLE_NLS - -#ifndef _WIN32 -#include -#endif // _WIN32 - #include +#include #include #include +#include #include #include #include -int -main(int argc, char* argv[]) -{ - BcStatus s; - char* name; - size_t len = strlen(BC_EXECPREFIX); +uint8_t* bc_fuzzer_data; -#if BC_ENABLE_NLS - // Must set the locale properly in order to have the right error messages. - vm->locale = setlocale(LC_ALL, ""); -#endif // BC_ENABLE_NLS +/// A boolean about whether we should use -c (false) or -C (true). +static bool bc_C; - // Set the start pledge(). - bc_pledge(bc_pledge_start, NULL); +int +LLVMFuzzerInitialize(int* argc, char*** argv) +{ + BC_UNUSED(argc); - // Sometimes, argv[0] can be NULL. Better make sure to be robust against it. - if (argv[0] != NULL) + if (argv == NULL || *argv == NULL) { - // Figure out the name of the calculator we are using. We can't use - // basename because it's not portable, but yes, this is stripping off - // the directory. - name = strrchr(argv[0], BC_FILE_SEP); - vm->name = (name == NULL) ? argv[0] : name + 1; + bc_C = false; } else { -#if !DC_ENABLED - vm->name = "bc"; -#elif !BC_ENABLED - vm->name = "dc"; -#else - // Just default to bc in that case. - vm->name = "bc"; -#endif + char* name; + + // Get the basename + name = strrchr((*argv)[0], BC_FILE_SEP); + name = name == NULL ? (*argv)[0] : name + 1; + + // Figure out which to use. + bc_C = (strcmp(name, "bc_fuzzer_C") == 0); } - // If the name is longer than the length of the prefix, skip the prefix. - if (strlen(vm->name) > len) vm->name += len; + return 0; +} + +int +LLVMFuzzerTestOneInput(const uint8_t* Data, size_t Size) +{ + BcStatus s; + + // I've already tested empty input, so just ignore. + if (Size == 0 || Data[0] == '\0') return 0; + + // Clear the global. This is to ensure a clean start. + memset(vm, 0, sizeof(BcVm)); + + // Make sure to set the name. + vm->name = "bc"; BC_SIG_LOCK; // We *must* do this here. Otherwise, other code could not jump out all of // the way. bc_vec_init(&vm->jmp_bufs, sizeof(sigjmp_buf), BC_DTOR_NONE); BC_SETJMP_LOCKED(vm, exit); -#if !DC_ENABLED - s = bc_main(argc, argv); -#elif !BC_ENABLED - s = dc_main(argc, argv); -#else - // BC_IS_BC uses vm->name, which was set above. So we're good. - if (BC_IS_BC) s = bc_main(argc, argv); - else s = dc_main(argc, argv); -#endif + // Create a string with the data. + bc_fuzzer_data = bc_vm_malloc(Size + 1); + memcpy(bc_fuzzer_data, Data, Size); + bc_fuzzer_data[Size] = '\0'; - vm->status = (int) s; + s = bc_main((int) (bc_fuzzer_args_len - 1), + bc_C ? bc_fuzzer_args_C : bc_fuzzer_args_c); exit: + BC_SIG_MAYLOCK; - return vm->status == BC_STATUS_QUIT ? BC_STATUS_SUCCESS : vm->status; + free(bc_fuzzer_data); + + return s == BC_STATUS_SUCCESS || s == BC_STATUS_QUIT ? 0 : -1; } diff --git a/src/bc_lex.c b/src/bc_lex.c index 106bb3ee03c4..f83eaf731622 100644 --- a/src/bc_lex.c +++ b/src/bc_lex.c @@ -1,503 +1,505 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * The lexer for bc. * */ #if BC_ENABLED #include #include #include #include #include /** * Lexes an identifier, which may be a keyword. * @param l The lexer. */ static void bc_lex_identifier(BcLex* l) { // We already passed the first character, so we need to be sure to include // it. const char* buf = l->buf + l->i - 1; size_t i; // This loop is simply checking for keywords. for (i = 0; i < bc_lex_kws_len; ++i) { const BcLexKeyword* kw = bc_lex_kws + i; size_t n = BC_LEX_KW_LEN(kw); if (!strncmp(buf, kw->name, n) && !isalnum(buf[n]) && buf[n] != '_') { // If the keyword has been redefined, and redefinition is allowed // (it is not allowed for builtin libraries), break out of the loop // and use it as a name. This depends on the argument parser to // ensure that only non-POSIX keywords get redefined. if (!vm->no_redefine && vm->redefined_kws[i]) break; l->t = BC_LEX_KW_AUTO + (BcLexType) i; // Warn or error, as appropriate for the mode, if the keyword is not // in the POSIX standard. if (!BC_LEX_KW_POSIX(kw)) bc_lex_verr(l, BC_ERR_POSIX_KW, kw->name); // We minus 1 because the index has already been incremented. l->i += n - 1; // Already have the token; bail. return; } } // If not a keyword, parse the name. bc_lex_name(l); // POSIX doesn't allow identifiers that are more than one character, so we // might have to warn or error here too. if (BC_ERR(l->str.len - 1 > 1)) { bc_lex_verr(l, BC_ERR_POSIX_NAME_LEN, l->str.v); } } /** * Parses a bc string. This is separate from dc strings because dc strings need * to be balanced. * @param l The lexer. */ static void bc_lex_string(BcLex* l) { // We need to keep track of newlines to increment them properly. size_t len, nlines, i; const char* buf; char c; bool got_more; l->t = BC_LEX_STR; do { nlines = 0; buf = l->buf; got_more = false; +#if !BC_ENABLE_OSSFUZZ assert(vm->mode != BC_MODE_STDIN || buf == vm->buffer.v); +#endif // !BC_ENABLE_OSSFUZZ // Fortunately for us, bc doesn't escape quotes. Instead, the equivalent // is '\q', which makes this loop simpler. for (i = l->i; (c = buf[i]) && c != '"'; ++i) { nlines += (c == '\n'); } if (BC_ERR(c == '\0') && !vm->eof && l->mode != BC_MODE_FILE) { got_more = bc_lex_readLine(l); } } while (got_more && c != '"'); // If the string did not end properly, barf. if (c != '"') { l->i = i; bc_lex_err(l, BC_ERR_PARSE_STRING); } // Set the temp string to the parsed string. len = i - l->i; bc_vec_string(&l->str, len, l->buf + l->i); l->i = i + 1; l->line += nlines; } /** * This function takes a lexed operator and checks to see if it's the assignment * version, setting the token appropriately. * @param l The lexer. * @param with The token to assign if it is an assignment operator. * @param without The token to assign if it is not an assignment operator. */ static void bc_lex_assign(BcLex* l, BcLexType with, BcLexType without) { if (l->buf[l->i] == '=') { l->i += 1; l->t = with; } else l->t = without; } void bc_lex_token(BcLex* l) { // We increment here. This means that all lexing needs to take that into // account, such as when parsing an identifier. If we don't, the first // character of every identifier would be missing. char c = l->buf[l->i++], c2; BC_SIG_ASSERT_LOCKED; // This is the workhorse of the lexer. switch (c) { case '\0': case '\n': case '\t': case '\v': case '\f': case '\r': case ' ': { bc_lex_commonTokens(l, c); break; } case '!': { // Even though it's not an assignment, we can use this. bc_lex_assign(l, BC_LEX_OP_REL_NE, BC_LEX_OP_BOOL_NOT); // POSIX doesn't allow boolean not. if (l->t == BC_LEX_OP_BOOL_NOT) { bc_lex_verr(l, BC_ERR_POSIX_BOOL, "!"); } break; } case '"': { bc_lex_string(l); break; } case '#': { // POSIX does not allow line comments. bc_lex_err(l, BC_ERR_POSIX_COMMENT); bc_lex_lineComment(l); break; } case '%': { bc_lex_assign(l, BC_LEX_OP_ASSIGN_MODULUS, BC_LEX_OP_MODULUS); break; } case '&': { c2 = l->buf[l->i]; // Either we have boolean and or an error. And boolean and is not // allowed by POSIX. if (BC_NO_ERR(c2 == '&')) { bc_lex_verr(l, BC_ERR_POSIX_BOOL, "&&"); l->i += 1; l->t = BC_LEX_OP_BOOL_AND; } else bc_lex_invalidChar(l, c); break; } #if BC_ENABLE_EXTRA_MATH case '$': { l->t = BC_LEX_OP_TRUNC; break; } case '@': { bc_lex_assign(l, BC_LEX_OP_ASSIGN_PLACES, BC_LEX_OP_PLACES); break; } #endif // BC_ENABLE_EXTRA_MATH case '(': case ')': { l->t = (BcLexType) (c - '(' + BC_LEX_LPAREN); break; } case '*': { bc_lex_assign(l, BC_LEX_OP_ASSIGN_MULTIPLY, BC_LEX_OP_MULTIPLY); break; } case '+': { c2 = l->buf[l->i]; // Have to check for increment first. if (c2 == '+') { l->i += 1; l->t = BC_LEX_OP_INC; } else bc_lex_assign(l, BC_LEX_OP_ASSIGN_PLUS, BC_LEX_OP_PLUS); break; } case ',': { l->t = BC_LEX_COMMA; break; } case '-': { c2 = l->buf[l->i]; // Have to check for decrement first. if (c2 == '-') { l->i += 1; l->t = BC_LEX_OP_DEC; } else bc_lex_assign(l, BC_LEX_OP_ASSIGN_MINUS, BC_LEX_OP_MINUS); break; } case '.': { c2 = l->buf[l->i]; // If it's alone, it's an alias for last. if (BC_LEX_NUM_CHAR(c2, true, false)) bc_lex_number(l, c); else { l->t = BC_LEX_KW_LAST; bc_lex_err(l, BC_ERR_POSIX_DOT); } break; } case '/': { c2 = l->buf[l->i]; if (c2 == '*') bc_lex_comment(l); else bc_lex_assign(l, BC_LEX_OP_ASSIGN_DIVIDE, BC_LEX_OP_DIVIDE); break; } case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': // Apparently, GNU bc (and maybe others) allows any uppercase letter as // a number. When single digits, they act like the ones above. When // multi-digit, any letter above the input base is automatically set to // the biggest allowable digit in the input base. case 'G': case 'H': case 'I': case 'J': case 'K': case 'L': case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R': case 'S': case 'T': case 'U': case 'V': case 'W': case 'X': case 'Y': case 'Z': { bc_lex_number(l, c); break; } case ';': { l->t = BC_LEX_SCOLON; break; } case '<': { #if BC_ENABLE_EXTRA_MATH c2 = l->buf[l->i]; // Check for shift. if (c2 == '<') { l->i += 1; bc_lex_assign(l, BC_LEX_OP_ASSIGN_LSHIFT, BC_LEX_OP_LSHIFT); break; } #endif // BC_ENABLE_EXTRA_MATH bc_lex_assign(l, BC_LEX_OP_REL_LE, BC_LEX_OP_REL_LT); break; } case '=': { bc_lex_assign(l, BC_LEX_OP_REL_EQ, BC_LEX_OP_ASSIGN); break; } case '>': { #if BC_ENABLE_EXTRA_MATH c2 = l->buf[l->i]; // Check for shift. if (c2 == '>') { l->i += 1; bc_lex_assign(l, BC_LEX_OP_ASSIGN_RSHIFT, BC_LEX_OP_RSHIFT); break; } #endif // BC_ENABLE_EXTRA_MATH bc_lex_assign(l, BC_LEX_OP_REL_GE, BC_LEX_OP_REL_GT); break; } case '[': case ']': { l->t = (BcLexType) (c - '[' + BC_LEX_LBRACKET); break; } case '\\': { // In bc, a backslash+newline is whitespace. if (BC_NO_ERR(l->buf[l->i] == '\n')) { l->i += 1; l->t = BC_LEX_WHITESPACE; } else bc_lex_invalidChar(l, c); break; } case '^': { bc_lex_assign(l, BC_LEX_OP_ASSIGN_POWER, BC_LEX_OP_POWER); break; } case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': case 'g': case 'h': case 'i': case 'j': case 'k': case 'l': case 'm': case 'n': case 'o': case 'p': case 'q': case 'r': case 's': case 't': case 'u': case 'v': case 'w': case 'x': case 'y': case 'z': { bc_lex_identifier(l); break; } case '{': case '}': { l->t = (BcLexType) (c - '{' + BC_LEX_LBRACE); break; } case '|': { c2 = l->buf[l->i]; // Once again, boolean or is not allowed by POSIX. if (BC_NO_ERR(c2 == '|')) { bc_lex_verr(l, BC_ERR_POSIX_BOOL, "||"); l->i += 1; l->t = BC_LEX_OP_BOOL_OR; } else bc_lex_invalidChar(l, c); break; } default: { bc_lex_invalidChar(l, c); } } } #endif // BC_ENABLED diff --git a/src/bc_parse.c b/src/bc_parse.c index 6842885933d6..cf4398709e58 100644 --- a/src/bc_parse.c +++ b/src/bc_parse.c @@ -1,2622 +1,2641 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * The parser for bc. * */ #if BC_ENABLED #include #include #include #include #include #include #include #include // Before you embark on trying to understand this code, have you read the // Development manual (manuals/development.md) and the comment in include/bc.h // yet? No? Do that first. I'm serious. // // The reason is because this file holds the most sensitive and finicky code in // the entire codebase. Even getting history to work on Windows was nothing // compared to this. This is where dreams go to die, where dragons live, and // from which Ken Thompson himself would flee. static void bc_parse_else(BcParse* p); static void bc_parse_stmt(BcParse* p); static BcParseStatus bc_parse_expr_err(BcParse* p, uint8_t flags, BcParseNext next); static void bc_parse_expr_status(BcParse* p, uint8_t flags, BcParseNext next); /** * Returns true if an instruction could only have come from a "leaf" expression. * For more on what leaf expressions are, read the comment for BC_PARSE_LEAF(). * @param t The instruction to test. * @return True if the instruction is a from a leaf expression. */ static bool bc_parse_inst_isLeaf(BcInst t) { return (t >= BC_INST_NUM && t <= BC_INST_LEADING_ZERO) || #if BC_ENABLE_EXTRA_MATH t == BC_INST_TRUNC || #endif // BC_ENABLE_EXTRA_MATH t <= BC_INST_DEC; } /** * Returns true if the *previous* token was a delimiter. A delimiter is anything * that can legally end a statement. In bc's case, it could be a newline, a * semicolon, and a brace in certain cases. * @param p The parser. * @return True if the token is a legal delimiter. */ static bool bc_parse_isDelimiter(const BcParse* p) { BcLexType t = p->l.t; bool good; // If it's an obvious delimiter, say so. if (BC_PARSE_DELIMITER(t)) return true; good = false; // If the current token is a keyword, then...beware. That means that we need // to check for a "dangling" else, where there was no brace-delimited block // on the previous if. if (t == BC_LEX_KW_ELSE) { size_t i; uint16_t *fptr = NULL, flags = BC_PARSE_FLAG_ELSE; // As long as going up the stack is valid for a dangling else, keep on. for (i = 0; i < p->flags.len && BC_PARSE_BLOCK_STMT(flags); ++i) { fptr = bc_vec_item_rev(&p->flags, i); flags = *fptr; // If we need a brace and don't have one, then we don't have a // delimiter. if ((flags & BC_PARSE_FLAG_BRACE) && p->l.last != BC_LEX_RBRACE) { return false; } } // Oh, and we had also better have an if statement somewhere. good = ((flags & BC_PARSE_FLAG_IF) != 0); } else if (t == BC_LEX_RBRACE) { size_t i; // Since we have a brace, we need to just check if a brace was needed. for (i = 0; !good && i < p->flags.len; ++i) { uint16_t* fptr = bc_vec_item_rev(&p->flags, i); good = (((*fptr) & BC_PARSE_FLAG_BRACE) != 0); } } return good; } /** * Returns true if we are in top level of a function body. The POSIX grammar * is defined such that anything is allowed after a function body, so we must * use this function to detect that case when ending a function body. * @param p The parser. * @return True if we are in the top level of parsing a function body. */ static bool bc_parse_TopFunc(const BcParse* p) { bool good = p->flags.len == 2; uint16_t val = BC_PARSE_FLAG_BRACE | BC_PARSE_FLAG_FUNC_INNER; val |= BC_PARSE_FLAG_FUNC; return good && BC_PARSE_TOP_FLAG(p) == val; } /** * Sets a previously defined exit label. What are labels? See the bc Parsing * section of the Development manual (manuals/development.md). * @param p The parser. */ static void bc_parse_setLabel(BcParse* p) { BcFunc* func = p->func; BcInstPtr* ip = bc_vec_top(&p->exits); size_t* label; assert(func == bc_vec_item(&p->prog->fns, p->fidx)); // Set the preallocated label to the correct index. label = bc_vec_item(&func->labels, ip->idx); *label = func->code.len; // Now, we don't need the exit label; it is done. bc_vec_pop(&p->exits); } /** * Creates a label and sets it to idx. If this is an exit label, then idx is * actually invalid, but it doesn't matter because it will be fixed by * bc_parse_setLabel() later. * @param p The parser. * @param idx The index of the label. */ static void bc_parse_createLabel(BcParse* p, size_t idx) { bc_vec_push(&p->func->labels, &idx); } /** * Creates a conditional label. Unlike an exit label, this label is set at * creation time because it comes *before* the code that will target it. * @param p The parser. * @param idx The index of the label. */ static void bc_parse_createCondLabel(BcParse* p, size_t idx) { bc_parse_createLabel(p, p->func->code.len); bc_vec_push(&p->conds, &idx); } /** * Creates an exit label to be filled in later by bc_parse_setLabel(). Also, why * create a label to be filled in later? Because exit labels are meant to be * targeted by code that comes *before* the label. Since we have to parse that * code first, and don't know how long it will be, we need to just make sure to * reserve a slot to be filled in later when we know. * * By the way, this uses BcInstPtr because it was convenient. The field idx * holds the index, and the field func holds the loop boolean. * * @param p The parser. * @param idx The index of the label's position. * @param loop True if the exit label is for a loop or not. */ static void bc_parse_createExitLabel(BcParse* p, size_t idx, bool loop) { BcInstPtr ip; assert(p->func == bc_vec_item(&p->prog->fns, p->fidx)); ip.func = loop; ip.idx = idx; ip.len = 0; bc_vec_push(&p->exits, &ip); bc_parse_createLabel(p, SIZE_MAX); } /** * Pops the correct operators off of the operator stack based on the current * operator. This is because of the Shunting-Yard algorithm. Lower prec means * higher precedence. * @param p The parser. * @param type The operator. * @param start The previous start of the operator stack. For more * information, see the bc Parsing section of the Development * manual (manuals/development.md). * @param nexprs A pointer to the current number of expressions that have not * been consumed yet. This is an IN and OUT parameter. */ static void bc_parse_operator(BcParse* p, BcLexType type, size_t start, size_t* nexprs) { BcLexType t; uchar l, r = BC_PARSE_OP_PREC(type); uchar left = BC_PARSE_OP_LEFT(type); // While we haven't hit the stop point yet... while (p->ops.len > start) { // Get the top operator. t = BC_PARSE_TOP_OP(p); // If it's a left paren, we have reached the end of whatever expression // this is no matter what. We also don't pop the left paren because it // will need to stay for the rest of the subexpression. if (t == BC_LEX_LPAREN) break; // Break for precedence. Precedence operates differently on left and // right associativity, by the way. A left associative operator that // matches the current precedence should take priority, but a right // associative operator should not. // // Also, a lower precedence value means a higher precedence. l = BC_PARSE_OP_PREC(t); if (l >= r && (l != r || !left)) break; // Do the housekeeping. In particular, make sure to note that one // expression was consumed (well, two were, but another was added) if // the operator was not a prefix operator. (Postfix operators are not // handled by this function at all.) bc_parse_push(p, BC_PARSE_TOKEN_INST(t)); bc_vec_pop(&p->ops); *nexprs -= !BC_PARSE_OP_PREFIX(t); } bc_vec_push(&p->ops, &type); } /** * Parses a right paren. In the Shunting-Yard algorithm, it needs to be put on * the operator stack. But before that, it needs to consume whatever operators * there are until it hits a left paren. * @param p The parser. * @param nexprs A pointer to the current number of expressions that have not * been consumed yet. This is an IN and OUT parameter. */ static void bc_parse_rightParen(BcParse* p, size_t* nexprs) { BcLexType top; // Consume operators until a left paren. while ((top = BC_PARSE_TOP_OP(p)) != BC_LEX_LPAREN) { bc_parse_push(p, BC_PARSE_TOKEN_INST(top)); bc_vec_pop(&p->ops); *nexprs -= !BC_PARSE_OP_PREFIX(top); } // We need to pop the left paren as well. bc_vec_pop(&p->ops); // Oh, and we also want the next token. bc_lex_next(&p->l); } /** * Parses function arguments. * @param p The parser. * @param flags Flags restricting what kind of expressions the arguments can * be. */ static void bc_parse_args(BcParse* p, uint8_t flags) { bool comma = false; size_t nargs; bc_lex_next(&p->l); // Print and comparison operators not allowed. Well, comparison operators // only for POSIX. But we do allow arrays, and we *must* get a value. flags &= ~(BC_PARSE_PRINT | BC_PARSE_REL); flags |= (BC_PARSE_ARRAY | BC_PARSE_NEEDVAL); // Count the arguments and parse them. for (nargs = 0; p->l.t != BC_LEX_RPAREN; ++nargs) { bc_parse_expr_status(p, flags, bc_parse_next_arg); comma = (p->l.t == BC_LEX_COMMA); if (comma) bc_lex_next(&p->l); } // An ending comma is FAIL. if (BC_ERR(comma)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); // Now do the call with the number of arguments. bc_parse_push(p, BC_INST_CALL); bc_parse_pushIndex(p, nargs); } /** * Parses a function call. * @param p The parser. * @param flags Flags restricting what kind of expressions the arguments can * be. */ static void bc_parse_call(BcParse* p, const char* name, uint8_t flags) { size_t idx; bc_parse_args(p, flags); // We just assert this because bc_parse_args() should // ensure that the next token is what it should be. assert(p->l.t == BC_LEX_RPAREN); // We cannot use bc_program_insertFunc() here // because it will overwrite an existing function. idx = bc_map_index(&p->prog->fn_map, name); // The function does not exist yet. Create a space for it. If the user does // not define it, it's a *runtime* error, not a parse error. if (idx == BC_VEC_INVALID_IDX) { idx = bc_program_insertFunc(p->prog, name); assert(idx != BC_VEC_INVALID_IDX); // Make sure that this pointer was not invalidated. p->func = bc_vec_item(&p->prog->fns, p->fidx); } // The function exists, so set the right function index. else idx = ((BcId*) bc_vec_item(&p->prog->fn_map, idx))->idx; bc_parse_pushIndex(p, idx); // Make sure to get the next token. bc_lex_next(&p->l); } /** * Parses a name/identifier-based expression. It could be a variable, an array * element, an array itself (for function arguments), a function call, etc. * @param p The parser. * @param type A pointer to return the resulting instruction. * @param can_assign A pointer to return true if the name can be assigned to, * false otherwise. * @param flags Flags restricting what kind of expression the name can be. */ static void bc_parse_name(BcParse* p, BcInst* type, bool* can_assign, uint8_t flags) { char* name; BC_SIG_ASSERT_LOCKED; // We want a copy of the name since the lexer might overwrite its copy. name = bc_vm_strdup(p->l.str.v); BC_SETJMP_LOCKED(vm, err); // We need the next token to see if it's just a variable or something more. bc_lex_next(&p->l); // Array element or array. if (p->l.t == BC_LEX_LBRACKET) { bc_lex_next(&p->l); // Array only. This has to be a function parameter. if (p->l.t == BC_LEX_RBRACKET) { // Error if arrays are not allowed. if (BC_ERR(!(flags & BC_PARSE_ARRAY))) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } *type = BC_INST_ARRAY; *can_assign = false; } else { // If we are here, we have an array element. We need to set the // expression parsing flags. uint8_t flags2 = (flags & ~(BC_PARSE_PRINT | BC_PARSE_REL)) | BC_PARSE_NEEDVAL; bc_parse_expr_status(p, flags2, bc_parse_next_elem); // The next token *must* be a right bracket. if (BC_ERR(p->l.t != BC_LEX_RBRACKET)) { bc_parse_err(p, BC_ERR_PARSE_TOKEN); } *type = BC_INST_ARRAY_ELEM; *can_assign = true; } // Make sure to get the next token. bc_lex_next(&p->l); // Push the instruction and the name of the identifier. bc_parse_push(p, *type); bc_parse_pushName(p, name, false); } else if (p->l.t == BC_LEX_LPAREN) { // We are parsing a function call; error if not allowed. if (BC_ERR(flags & BC_PARSE_NOCALL)) { bc_parse_err(p, BC_ERR_PARSE_TOKEN); } *type = BC_INST_CALL; *can_assign = false; bc_parse_call(p, name, flags); } else { // Just a variable. *type = BC_INST_VAR; *can_assign = true; bc_parse_push(p, BC_INST_VAR); bc_parse_pushName(p, name, true); } err: // Need to make sure to unallocate the name. free(name); BC_LONGJMP_CONT(vm); BC_SIG_MAYLOCK; } /** * Parses a builtin function that takes no arguments. This includes read(), * rand(), maxibase(), maxobase(), maxscale(), and maxrand(). * @param p The parser. * @param inst The instruction corresponding to the builtin. */ static void bc_parse_noArgBuiltin(BcParse* p, BcInst inst) { // Must have a left paren. bc_lex_next(&p->l); if (BC_ERR(p->l.t != BC_LEX_LPAREN)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); // Must have a right paren. bc_lex_next(&p->l); if ((p->l.t != BC_LEX_RPAREN)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); bc_parse_push(p, inst); bc_lex_next(&p->l); } /** * Parses a builtin function that takes 1 argument. This includes length(), * sqrt(), abs(), scale(), and irand(). * @param p The parser. * @param type The lex token. * @param flags The expression parsing flags for parsing the argument. * @param prev An out parameter; the previous instruction pointer. */ static void bc_parse_builtin(BcParse* p, BcLexType type, uint8_t flags, BcInst* prev) { // Must have a left paren. bc_lex_next(&p->l); if (BC_ERR(p->l.t != BC_LEX_LPAREN)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); bc_lex_next(&p->l); // Change the flags as needed for parsing the argument. flags &= ~(BC_PARSE_PRINT | BC_PARSE_REL); flags |= BC_PARSE_NEEDVAL; // Since length can take arrays, we need to specially add that flag. if (type == BC_LEX_KW_LENGTH || type == BC_LEX_KW_ASCIIFY) { flags |= BC_PARSE_ARRAY; } // Otherwise, we need to clear it because it could be set. else flags &= ~(BC_PARSE_ARRAY); bc_parse_expr_status(p, flags, bc_parse_next_rel); // Must have a right paren. if (BC_ERR(p->l.t != BC_LEX_RPAREN)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); // Adjust previous based on the token and push it. *prev = type - BC_LEX_KW_LENGTH + BC_INST_LENGTH; bc_parse_push(p, *prev); bc_lex_next(&p->l); } /** * Parses a builtin function that takes 3 arguments. This includes modexp() and * divmod(). * @param p The parser. * @param type The lex token. * @param flags The expression parsing flags for parsing the argument. * @param prev An out parameter; the previous instruction pointer. */ static void bc_parse_builtin3(BcParse* p, BcLexType type, uint8_t flags, BcInst* prev) { assert(type == BC_LEX_KW_MODEXP || type == BC_LEX_KW_DIVMOD); // Must have a left paren. bc_lex_next(&p->l); if (BC_ERR(p->l.t != BC_LEX_LPAREN)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); bc_lex_next(&p->l); // Change the flags as needed for parsing the argument. flags &= ~(BC_PARSE_PRINT | BC_PARSE_REL); flags |= BC_PARSE_NEEDVAL; bc_parse_expr_status(p, flags, bc_parse_next_builtin); // Must have a comma. if (BC_ERR(p->l.t != BC_LEX_COMMA)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); bc_lex_next(&p->l); bc_parse_expr_status(p, flags, bc_parse_next_builtin); // Must have a comma. if (BC_ERR(p->l.t != BC_LEX_COMMA)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); bc_lex_next(&p->l); // If it is a divmod, parse an array name. Otherwise, just parse another // expression. if (type == BC_LEX_KW_DIVMOD) { // Must have a name. if (BC_ERR(p->l.t != BC_LEX_NAME)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); // This is safe because the next token should not overwrite the name. bc_lex_next(&p->l); // Must have a left bracket. if (BC_ERR(p->l.t != BC_LEX_LBRACKET)) { bc_parse_err(p, BC_ERR_PARSE_TOKEN); } // This is safe because the next token should not overwrite the name. bc_lex_next(&p->l); // Must have a right bracket. if (BC_ERR(p->l.t != BC_LEX_RBRACKET)) { bc_parse_err(p, BC_ERR_PARSE_TOKEN); } // This is safe because the next token should not overwrite the name. bc_lex_next(&p->l); } else bc_parse_expr_status(p, flags, bc_parse_next_rel); // Must have a right paren. if (BC_ERR(p->l.t != BC_LEX_RPAREN)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); // Adjust previous based on the token and push it. *prev = type - BC_LEX_KW_MODEXP + BC_INST_MODEXP; bc_parse_push(p, *prev); // If we have divmod, we need to assign the modulus to the array element, so // we need to push the instructions for doing so. if (type == BC_LEX_KW_DIVMOD) { // The zeroth element. bc_parse_push(p, BC_INST_ZERO); bc_parse_push(p, BC_INST_ARRAY_ELEM); // Push the array. bc_parse_pushName(p, p->l.str.v, false); // Swap them and assign. After this, the top item on the stack should // be the quotient. bc_parse_push(p, BC_INST_SWAP); bc_parse_push(p, BC_INST_ASSIGN_NO_VAL); } bc_lex_next(&p->l); } /** * Parses the scale keyword. This is special because scale can be a value or a * builtin function. * @param p The parser. * @param type An out parameter; the instruction for the parse. * @param can_assign An out parameter; whether the expression can be assigned * to. * @param flags The expression parsing flags for parsing a scale() arg. */ static void bc_parse_scale(BcParse* p, BcInst* type, bool* can_assign, uint8_t flags) { bc_lex_next(&p->l); // Without the left paren, it's just the keyword. if (p->l.t != BC_LEX_LPAREN) { // Set, push, and return. *type = BC_INST_SCALE; *can_assign = true; bc_parse_push(p, BC_INST_SCALE); return; } // Handle the scale function. *type = BC_INST_SCALE_FUNC; *can_assign = false; // Once again, adjust the flags. flags &= ~(BC_PARSE_PRINT | BC_PARSE_REL); flags |= BC_PARSE_NEEDVAL; bc_lex_next(&p->l); bc_parse_expr_status(p, flags, bc_parse_next_rel); // Must have a right paren. if (BC_ERR(p->l.t != BC_LEX_RPAREN)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); bc_parse_push(p, BC_INST_SCALE_FUNC); bc_lex_next(&p->l); } /** * Parses and increment or decrement operator. This is a bit complex. * @param p The parser. * @param prev An out parameter; the previous instruction pointer. * @param can_assign An out parameter; whether the expression can be assigned * to. * @param nexs An in/out parameter; the number of expressions in the * parse tree that are not used. * @param flags The expression parsing flags for parsing a scale() arg. */ static void bc_parse_incdec(BcParse* p, BcInst* prev, bool* can_assign, size_t* nexs, uint8_t flags) { BcLexType type; uchar inst; BcInst etype = *prev; BcLexType last = p->l.last; assert(prev != NULL && can_assign != NULL); // If we can't assign to the previous token, then we have an error. if (BC_ERR(last == BC_LEX_OP_INC || last == BC_LEX_OP_DEC || last == BC_LEX_RPAREN)) { bc_parse_err(p, BC_ERR_PARSE_ASSIGN); } // Is the previous instruction for a variable? if (BC_PARSE_INST_VAR(etype)) { // If so, this is a postfix operator. if (!*can_assign) bc_parse_err(p, BC_ERR_PARSE_ASSIGN); // Only postfix uses BC_INST_INC and BC_INST_DEC. *prev = inst = BC_INST_INC + (p->l.t != BC_LEX_OP_INC); bc_parse_push(p, inst); bc_lex_next(&p->l); *can_assign = false; } else { // This is a prefix operator. In that case, we just convert it to // an assignment instruction. *prev = inst = BC_INST_ASSIGN_PLUS + (p->l.t != BC_LEX_OP_INC); bc_lex_next(&p->l); type = p->l.t; // Because we parse the next part of the expression // right here, we need to increment this. *nexs = *nexs + 1; // Is the next token a normal identifier? if (type == BC_LEX_NAME) { // Parse the name. uint8_t flags2 = flags & ~(BC_PARSE_ARRAY); bc_parse_name(p, prev, can_assign, flags2 | BC_PARSE_NOCALL); } // Is the next token a global? else if (type >= BC_LEX_KW_LAST && type <= BC_LEX_KW_OBASE) { bc_parse_push(p, type - BC_LEX_KW_LAST + BC_INST_LAST); bc_lex_next(&p->l); } // Is the next token specifically scale, which needs special treatment? else if (BC_NO_ERR(type == BC_LEX_KW_SCALE)) { bc_lex_next(&p->l); // Check that scale() was not used. if (BC_ERR(p->l.t == BC_LEX_LPAREN)) { bc_parse_err(p, BC_ERR_PARSE_TOKEN); } else bc_parse_push(p, BC_INST_SCALE); } // Now we know we have an error. else bc_parse_err(p, BC_ERR_PARSE_TOKEN); *can_assign = false; bc_parse_push(p, BC_INST_ONE); bc_parse_push(p, inst); } } /** * Parses the minus operator. This needs special treatment because it is either * subtract or negation. * @param p The parser. * @param prev An in/out parameter; the previous instruction. * @param ops_bgn The size of the operator stack. * @param rparen True if the last token was a right paren. * @param binlast True if the last token was a binary operator. * @param nexprs An in/out parameter; the number of unused expressions. */ static void bc_parse_minus(BcParse* p, BcInst* prev, size_t ops_bgn, bool rparen, bool binlast, size_t* nexprs) { BcLexType type; bc_lex_next(&p->l); // Figure out if it's a minus or a negation. type = BC_PARSE_LEAF(*prev, binlast, rparen) ? BC_LEX_OP_MINUS : BC_LEX_NEG; *prev = BC_PARSE_TOKEN_INST(type); // We can just push onto the op stack because this is the largest // precedence operator that gets pushed. Inc/dec does not. if (type != BC_LEX_OP_MINUS) bc_vec_push(&p->ops, &type); else bc_parse_operator(p, type, ops_bgn, nexprs); } /** * Parses a string. * @param p The parser. * @param inst The instruction corresponding to how the string was found and * how it should be printed. */ static void bc_parse_str(BcParse* p, BcInst inst) { bc_parse_addString(p); bc_parse_push(p, inst); bc_lex_next(&p->l); } /** * Parses a print statement. * @param p The parser. */ static void bc_parse_print(BcParse* p, BcLexType type) { BcLexType t; bool comma = false; BcInst inst = type == BC_LEX_KW_STREAM ? BC_INST_PRINT_STREAM : BC_INST_PRINT_POP; bc_lex_next(&p->l); t = p->l.t; // A print or stream statement has to have *something*. if (bc_parse_isDelimiter(p)) bc_parse_err(p, BC_ERR_PARSE_PRINT); do { // If the token is a string, then print it with escapes. // BC_INST_PRINT_POP plays that role for bc. if (t == BC_LEX_STR) bc_parse_str(p, inst); else { // We have an actual number; parse and add a print instruction. bc_parse_expr_status(p, BC_PARSE_NEEDVAL, bc_parse_next_print); bc_parse_push(p, inst); } // Is the next token a comma? comma = (p->l.t == BC_LEX_COMMA); // Get the next token if we have a comma. if (comma) bc_lex_next(&p->l); else { // If we don't have a comma, the statement needs to end. if (!bc_parse_isDelimiter(p)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); else break; } t = p->l.t; } while (true); // If we have a comma but no token, that's bad. if (BC_ERR(comma)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); } /** * Parses a return statement. * @param p The parser. */ static void bc_parse_return(BcParse* p) { BcLexType t; bool paren; uchar inst = BC_INST_RET0; // If we are not in a function, that's an error. if (BC_ERR(!BC_PARSE_FUNC(p))) bc_parse_err(p, BC_ERR_PARSE_TOKEN); // If we are in a void function, make sure to return void. if (p->func->voidfn) inst = BC_INST_RET_VOID; bc_lex_next(&p->l); t = p->l.t; paren = (t == BC_LEX_LPAREN); // An empty return statement just needs to push the selected instruction. if (bc_parse_isDelimiter(p)) bc_parse_push(p, inst); else { BcParseStatus s; // Need to parse the expression whose value will be returned. s = bc_parse_expr_err(p, BC_PARSE_NEEDVAL, bc_parse_next_expr); // If the expression was empty, just push the selected instruction. if (s == BC_PARSE_STATUS_EMPTY_EXPR) { bc_parse_push(p, inst); bc_lex_next(&p->l); } // POSIX requires parentheses. if (!paren || p->l.last != BC_LEX_RPAREN) { bc_parse_err(p, BC_ERR_POSIX_RET); } // Void functions require an empty expression. if (BC_ERR(p->func->voidfn)) { if (s != BC_PARSE_STATUS_EMPTY_EXPR) { bc_parse_verr(p, BC_ERR_PARSE_RET_VOID, p->func->name); } } // If we got here, we want to be sure to end the function with a real // return instruction, just in case. else bc_parse_push(p, BC_INST_RET); } } /** * Clears flags that indicate the end of an if statement and its block and sets * the jump location. * @param p The parser. */ static void bc_parse_noElse(BcParse* p) { uint16_t* flag_ptr = BC_PARSE_TOP_FLAG_PTR(p); *flag_ptr = (*flag_ptr & ~(BC_PARSE_FLAG_IF_END)); bc_parse_setLabel(p); } /** * Ends (finishes parsing) the body of a control statement or a function. * @param p The parser. * @param brace True if the body was ended by a brace, false otherwise. */ static void bc_parse_endBody(BcParse* p, bool brace) { bool has_brace, new_else = false; // We cannot be ending a body if there are no bodies to end. if (BC_ERR(p->flags.len <= 1)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); if (brace) { // The brace was already gotten; make sure that the caller did not lie. // We check for the requirement of braces later. assert(p->l.t == BC_LEX_RBRACE); bc_lex_next(&p->l); // If the next token is not a delimiter, that is a problem. if (BC_ERR(!bc_parse_isDelimiter(p) && !bc_parse_TopFunc(p))) { bc_parse_err(p, BC_ERR_PARSE_TOKEN); } } // Do we have a brace flag? has_brace = (BC_PARSE_BRACE(p) != 0); do { size_t len = p->flags.len; bool loop; // If we have a brace flag but not a brace, that's a problem. if (has_brace && !brace) bc_parse_err(p, BC_ERR_PARSE_TOKEN); // Are we inside a loop? loop = (BC_PARSE_LOOP_INNER(p) != 0); // If we are ending a loop or an else... if (loop || BC_PARSE_ELSE(p)) { // Loops have condition labels that we have to take care of as well. if (loop) { size_t* label = bc_vec_top(&p->conds); bc_parse_push(p, BC_INST_JUMP); bc_parse_pushIndex(p, *label); bc_vec_pop(&p->conds); } bc_parse_setLabel(p); bc_vec_pop(&p->flags); } // If we are ending a function... else if (BC_PARSE_FUNC_INNER(p)) { BcInst inst = (p->func->voidfn ? BC_INST_RET_VOID : BC_INST_RET0); bc_parse_push(p, inst); bc_parse_updateFunc(p, BC_PROG_MAIN); bc_vec_pop(&p->flags); } // If we have a brace flag and not an if statement, we can pop the top // of the flags stack because they have been taken care of above. else if (has_brace && !BC_PARSE_IF(p)) bc_vec_pop(&p->flags); // This needs to be last to parse nested if's properly. if (BC_PARSE_IF(p) && (len == p->flags.len || !BC_PARSE_BRACE(p))) { // Eat newlines. while (p->l.t == BC_LEX_NLINE) { bc_lex_next(&p->l); } // *Now* we can pop the flags. bc_vec_pop(&p->flags); // If we are allowed non-POSIX stuff... if (!BC_S) { // Have we found yet another dangling else? *(BC_PARSE_TOP_FLAG_PTR(p)) |= BC_PARSE_FLAG_IF_END; new_else = (p->l.t == BC_LEX_KW_ELSE); // Parse the else or end the if statement body. if (new_else) bc_parse_else(p); else if (!has_brace && (!BC_PARSE_IF_END(p) || brace)) { bc_parse_noElse(p); } } // POSIX requires us to do the bare minimum only. else bc_parse_noElse(p); } // If these are both true, we have "used" the braces that we found. if (brace && has_brace) brace = false; } // This condition was perhaps the hardest single part of the parser. If // the flags stack does not have enough, we should stop. If we have a // new else statement, we should stop. If we do have the end of an if // statement and we have eaten the brace, we should stop. If we do have // a brace flag, we should stop. while (p->flags.len > 1 && !new_else && (!BC_PARSE_IF_END(p) || brace) && !(has_brace = (BC_PARSE_BRACE(p) != 0))); // If we have a brace, yet no body for it, that's a problem. if (BC_ERR(p->flags.len == 1 && brace)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); else if (brace && BC_PARSE_BRACE(p)) { // If we make it here, we have a brace and a flag for it. uint16_t flags = BC_PARSE_TOP_FLAG(p); // This condition ensure that the *last* body is correctly finished by // popping its flags. if (!(flags & (BC_PARSE_FLAG_FUNC_INNER | BC_PARSE_FLAG_LOOP_INNER)) && !(flags & (BC_PARSE_FLAG_IF | BC_PARSE_FLAG_ELSE)) && !(flags & (BC_PARSE_FLAG_IF_END))) { bc_vec_pop(&p->flags); } } } /** * Starts the body of a control statement or function. * @param p The parser. * @param flags The current flags (will be edited). */ static void bc_parse_startBody(BcParse* p, uint16_t flags) { assert(flags); flags |= (BC_PARSE_TOP_FLAG(p) & (BC_PARSE_FLAG_FUNC | BC_PARSE_FLAG_LOOP)); flags |= BC_PARSE_FLAG_BODY; bc_vec_push(&p->flags, &flags); } void bc_parse_endif(BcParse* p) { size_t i; bool good; // Not a problem if this is true. if (BC_NO_ERR(!BC_PARSE_NO_EXEC(p))) return; good = true; // Find an instance of a body that needs closing, i.e., a statement that did // not have a right brace when it should have. for (i = 0; good && i < p->flags.len; ++i) { uint16_t flag = *((uint16_t*) bc_vec_item(&p->flags, i)); good = ((flag & BC_PARSE_FLAG_BRACE) != BC_PARSE_FLAG_BRACE); } // If we did not find such an instance... if (good) { // We set this to restore it later. We don't want the parser thinking // that we are on stdin for this one because it will want more. BcMode mode = vm->mode; vm->mode = BC_MODE_FILE; // End all of the if statements and loops. while (p->flags.len > 1 || BC_PARSE_IF_END(p)) { if (BC_PARSE_IF_END(p)) bc_parse_noElse(p); if (p->flags.len > 1) bc_parse_endBody(p, false); } vm->mode = (uchar) mode; } // If we reach here, a block was not properly closed, and we should error. else bc_parse_err(&vm->prs, BC_ERR_PARSE_BLOCK); } /** * Parses an if statement. * @param p The parser. */ static void bc_parse_if(BcParse* p) { // We are allowed relational operators, and we must have a value. size_t idx; uint8_t flags = (BC_PARSE_REL | BC_PARSE_NEEDVAL); // Get the left paren and barf if necessary. bc_lex_next(&p->l); if (BC_ERR(p->l.t != BC_LEX_LPAREN)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); // Parse the condition. bc_lex_next(&p->l); bc_parse_expr_status(p, flags, bc_parse_next_rel); // Must have a right paren. if (BC_ERR(p->l.t != BC_LEX_RPAREN)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); bc_lex_next(&p->l); // Insert the conditional jump instruction. bc_parse_push(p, BC_INST_JUMP_ZERO); idx = p->func->labels.len; // Push the index for the instruction and create an exit label for an else // statement. bc_parse_pushIndex(p, idx); bc_parse_createExitLabel(p, idx, false); bc_parse_startBody(p, BC_PARSE_FLAG_IF); } /** * Parses an else statement. * @param p The parser. */ static void bc_parse_else(BcParse* p) { size_t idx = p->func->labels.len; // We must be at the end of an if statement. if (BC_ERR(!BC_PARSE_IF_END(p))) bc_parse_err(p, BC_ERR_PARSE_TOKEN); // Push an unconditional jump to make bc jump over the else statement if it // executed the original if statement. bc_parse_push(p, BC_INST_JUMP); bc_parse_pushIndex(p, idx); // Clear the else stuff. Yes, that function is misnamed for its use here, // but deal with it. bc_parse_noElse(p); // Create the exit label and parse the body. bc_parse_createExitLabel(p, idx, false); bc_parse_startBody(p, BC_PARSE_FLAG_ELSE); bc_lex_next(&p->l); } /** * Parse a while loop. * @param p The parser. */ static void bc_parse_while(BcParse* p) { // We are allowed relational operators, and we must have a value. size_t idx; uint8_t flags = (BC_PARSE_REL | BC_PARSE_NEEDVAL); // Get the left paren and barf if necessary. bc_lex_next(&p->l); if (BC_ERR(p->l.t != BC_LEX_LPAREN)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); bc_lex_next(&p->l); // Create the labels. Loops need both. bc_parse_createCondLabel(p, p->func->labels.len); idx = p->func->labels.len; bc_parse_createExitLabel(p, idx, true); // Parse the actual condition and barf on non-right paren. bc_parse_expr_status(p, flags, bc_parse_next_rel); if (BC_ERR(p->l.t != BC_LEX_RPAREN)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); bc_lex_next(&p->l); // Now we can push the conditional jump and start the body. bc_parse_push(p, BC_INST_JUMP_ZERO); bc_parse_pushIndex(p, idx); bc_parse_startBody(p, BC_PARSE_FLAG_LOOP | BC_PARSE_FLAG_LOOP_INNER); } /** * Parse a for loop. * @param p The parser. */ static void bc_parse_for(BcParse* p) { size_t cond_idx, exit_idx, body_idx, update_idx; // Barf on the missing left paren. bc_lex_next(&p->l); if (BC_ERR(p->l.t != BC_LEX_LPAREN)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); bc_lex_next(&p->l); // The first statement can be empty, but if it is, check for error in POSIX // mode. Otherwise, parse it. if (p->l.t != BC_LEX_SCOLON) bc_parse_expr_status(p, 0, bc_parse_next_for); else bc_parse_err(p, BC_ERR_POSIX_FOR); // Must have a semicolon. if (BC_ERR(p->l.t != BC_LEX_SCOLON)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); bc_lex_next(&p->l); // These are indices for labels. There are so many of them because the end // of the loop must unconditionally jump to the update code. Then the update // code must unconditionally jump to the condition code. Then the condition // code must *conditionally* jump to the exit. cond_idx = p->func->labels.len; update_idx = cond_idx + 1; body_idx = update_idx + 1; exit_idx = body_idx + 1; // This creates the condition label. bc_parse_createLabel(p, p->func->code.len); // Parse an expression if it exists. if (p->l.t != BC_LEX_SCOLON) { uint8_t flags = (BC_PARSE_REL | BC_PARSE_NEEDVAL); bc_parse_expr_status(p, flags, bc_parse_next_for); } else { // Set this for the next call to bc_parse_number because an empty // condition means that it is an infinite loop, so the condition must be // non-zero. This is safe to set because the current token is a // semicolon, which has no string requirement. bc_vec_string(&p->l.str, sizeof(bc_parse_one) - 1, bc_parse_one); bc_parse_number(p); // An empty condition makes POSIX mad. bc_parse_err(p, BC_ERR_POSIX_FOR); } // Must have a semicolon. if (BC_ERR(p->l.t != BC_LEX_SCOLON)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); bc_lex_next(&p->l); // Now we can set up the conditional jump to the exit and an unconditional // jump to the body right after. The unconditional jump to the body is // because there is update code coming right after the condition, so we need // to skip it to get to the body. bc_parse_push(p, BC_INST_JUMP_ZERO); bc_parse_pushIndex(p, exit_idx); bc_parse_push(p, BC_INST_JUMP); bc_parse_pushIndex(p, body_idx); // Now create the label for the update code. bc_parse_createCondLabel(p, update_idx); // Parse if not empty, and if it is, let POSIX yell if necessary. if (p->l.t != BC_LEX_RPAREN) bc_parse_expr_status(p, 0, bc_parse_next_rel); else bc_parse_err(p, BC_ERR_POSIX_FOR); // Must have a right paren. if (BC_ERR(p->l.t != BC_LEX_RPAREN)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); // Set up a jump to the condition right after the update code. bc_parse_push(p, BC_INST_JUMP); bc_parse_pushIndex(p, cond_idx); bc_parse_createLabel(p, p->func->code.len); // Create an exit label for the body and start the body. bc_parse_createExitLabel(p, exit_idx, true); bc_lex_next(&p->l); bc_parse_startBody(p, BC_PARSE_FLAG_LOOP | BC_PARSE_FLAG_LOOP_INNER); } /** * Parse a statement or token that indicates a loop exit. This includes an * actual loop exit, the break keyword, or the continue keyword. * @param p The parser. * @param type The type of exit. */ static void bc_parse_loopExit(BcParse* p, BcLexType type) { size_t i; BcInstPtr* ip; // Must have a loop. If we don't, that's an error. if (BC_ERR(!BC_PARSE_LOOP(p))) bc_parse_err(p, BC_ERR_PARSE_TOKEN); // If we have a break statement... if (type == BC_LEX_KW_BREAK) { // If there are no exits, something went wrong somewhere. if (BC_ERR(!p->exits.len)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); // Get the exit. i = p->exits.len - 1; ip = bc_vec_item(&p->exits, i); // The condition !ip->func is true if the exit is not for a loop, so we // need to find the first actual loop exit. while (!ip->func && i < p->exits.len) { ip = bc_vec_item(&p->exits, i); i -= 1; } // Make sure everything is hunky dory. assert(ip != NULL && (i < p->exits.len || ip->func)); // Set the index for the exit. i = ip->idx; } // If we have a continue statement or just the loop end, jump to the // condition (or update for a foor loop). else i = *((size_t*) bc_vec_top(&p->conds)); // Add the unconditional jump. bc_parse_push(p, BC_INST_JUMP); bc_parse_pushIndex(p, i); bc_lex_next(&p->l); } /** * Parse a function (header). * @param p The parser. */ static void bc_parse_func(BcParse* p) { bool comma = false, voidfn; uint16_t flags; size_t idx; bc_lex_next(&p->l); // Must have a name. if (BC_ERR(p->l.t != BC_LEX_NAME)) bc_parse_err(p, BC_ERR_PARSE_FUNC); // If the name is "void", and POSIX is not on, mark as void. voidfn = (!BC_IS_POSIX && p->l.t == BC_LEX_NAME && !strcmp(p->l.str.v, "void")); // We can safely do this because the expected token should not overwrite the // function name. bc_lex_next(&p->l); // If we *don't* have another name, then void is the name of the function. voidfn = (voidfn && p->l.t == BC_LEX_NAME); // With a void function, allow POSIX to complain and get a new token. if (voidfn) { bc_parse_err(p, BC_ERR_POSIX_VOID); // We can safely do this because the expected token should not overwrite // the function name. bc_lex_next(&p->l); } // Must have a left paren. if (BC_ERR(p->l.t != BC_LEX_LPAREN)) bc_parse_err(p, BC_ERR_PARSE_FUNC); // Make sure the functions map and vector are synchronized. assert(p->prog->fns.len == p->prog->fn_map.len); // Insert the function by name into the map and vector. idx = bc_program_insertFunc(p->prog, p->l.str.v); // Make sure the insert worked. assert(idx); // Update the function pointer and stuff in the parser and set its void. bc_parse_updateFunc(p, idx); p->func->voidfn = voidfn; bc_lex_next(&p->l); // While we do not have a right paren, we are still parsing arguments. while (p->l.t != BC_LEX_RPAREN) { BcType t = BC_TYPE_VAR; // If we have an asterisk, we are parsing a reference argument. if (p->l.t == BC_LEX_OP_MULTIPLY) { t = BC_TYPE_REF; bc_lex_next(&p->l); // Let POSIX complain if necessary. bc_parse_err(p, BC_ERR_POSIX_REF); } // If we don't have a name, the argument will not have a name. Barf. if (BC_ERR(p->l.t != BC_LEX_NAME)) bc_parse_err(p, BC_ERR_PARSE_FUNC); // Increment the number of parameters. p->func->nparams += 1; // Copy the string in the lexer so that we can use the lexer again. bc_vec_string(&p->buf, p->l.str.len, p->l.str.v); bc_lex_next(&p->l); // We are parsing an array parameter if this is true. if (p->l.t == BC_LEX_LBRACKET) { // Set the array type, unless we are already parsing a reference. if (t == BC_TYPE_VAR) t = BC_TYPE_ARRAY; bc_lex_next(&p->l); // The brackets *must* be empty. if (BC_ERR(p->l.t != BC_LEX_RBRACKET)) { bc_parse_err(p, BC_ERR_PARSE_FUNC); } bc_lex_next(&p->l); } // If we did *not* get a bracket, but we are expecting a reference, we // have a problem. else if (BC_ERR(t == BC_TYPE_REF)) { bc_parse_verr(p, BC_ERR_PARSE_REF_VAR, p->buf.v); } // Test for comma and get the next token if it exists. comma = (p->l.t == BC_LEX_COMMA); if (comma) bc_lex_next(&p->l); // Insert the parameter into the function. bc_func_insert(p->func, p->prog, p->buf.v, t, p->l.line); } // If we have a comma, but no parameter, barf. if (BC_ERR(comma)) bc_parse_err(p, BC_ERR_PARSE_FUNC); // Start the body. flags = BC_PARSE_FLAG_FUNC | BC_PARSE_FLAG_FUNC_INNER; bc_parse_startBody(p, flags); bc_lex_next(&p->l); // POSIX requires that a brace be on the same line as the function header. // If we don't have a brace, let POSIX throw an error. if (p->l.t != BC_LEX_LBRACE) bc_parse_err(p, BC_ERR_POSIX_BRACE); } /** * Parse an auto list. * @param p The parser. */ static void bc_parse_auto(BcParse* p) { bool comma, one; // Error if the auto keyword appeared in the wrong place. if (BC_ERR(!p->auto_part)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); bc_lex_next(&p->l); p->auto_part = comma = false; // We need at least one variable or array. one = (p->l.t == BC_LEX_NAME); // While we have a variable or array. while (p->l.t == BC_LEX_NAME) { BcType t; // Copy the name from the lexer, so we can use it again. bc_vec_string(&p->buf, p->l.str.len - 1, p->l.str.v); bc_lex_next(&p->l); // If we are parsing an array... if (p->l.t == BC_LEX_LBRACKET) { t = BC_TYPE_ARRAY; bc_lex_next(&p->l); // The brackets *must* be empty. if (BC_ERR(p->l.t != BC_LEX_RBRACKET)) { bc_parse_err(p, BC_ERR_PARSE_FUNC); } bc_lex_next(&p->l); } else t = BC_TYPE_VAR; // Test for comma and get the next token if it exists. comma = (p->l.t == BC_LEX_COMMA); if (comma) bc_lex_next(&p->l); // Insert the auto into the function. bc_func_insert(p->func, p->prog, p->buf.v, t, p->l.line); } // If we have a comma, but no auto, barf. if (BC_ERR(comma)) bc_parse_err(p, BC_ERR_PARSE_FUNC); // If we don't have any variables or arrays, barf. if (BC_ERR(!one)) bc_parse_err(p, BC_ERR_PARSE_NO_AUTO); // The auto statement should be all that's in the statement. if (BC_ERR(!bc_parse_isDelimiter(p))) bc_parse_err(p, BC_ERR_PARSE_TOKEN); } /** * Parses a body. * @param p The parser. * @param brace True if a brace was encountered, false otherwise. */ static void bc_parse_body(BcParse* p, bool brace) { uint16_t* flag_ptr = BC_PARSE_TOP_FLAG_PTR(p); assert(flag_ptr != NULL); assert(p->flags.len >= 2); // The body flag is for when we expect a body. We got a body, so clear the // flag. *flag_ptr &= ~(BC_PARSE_FLAG_BODY); // If we are inside a function, that means we just barely entered it, and // we can expect an auto list. if (*flag_ptr & BC_PARSE_FLAG_FUNC_INNER) { // We *must* have a brace in this case. if (BC_ERR(!brace)) bc_parse_err(p, BC_ERR_PARSE_TOKEN); p->auto_part = (p->l.t != BC_LEX_KW_AUTO); if (!p->auto_part) { // Make sure this is true to not get a parse error. p->auto_part = true; // Since we already have the auto keyword, parse. bc_parse_auto(p); } // Eat a newline. if (p->l.t == BC_LEX_NLINE) bc_lex_next(&p->l); } else { // This is the easy part. size_t len = p->flags.len; assert(*flag_ptr); // Parse a statement. bc_parse_stmt(p); // This is a very important condition to get right. If there is no // brace, and no body flag, and the flags len hasn't shrunk, then we // have a body that was not delimited by braces, so we need to end it // now, after just one statement. if (!brace && !BC_PARSE_BODY(p) && len <= p->flags.len) { bc_parse_endBody(p, false); } } } /** * Parses a statement. This is the entry point for just about everything, except * function definitions. * @param p The parser. */ static void bc_parse_stmt(BcParse* p) { size_t len; uint16_t flags; BcLexType type = p->l.t; // Eat newline. if (type == BC_LEX_NLINE) { bc_lex_next(&p->l); return; } // Eat auto list. if (type == BC_LEX_KW_AUTO) { bc_parse_auto(p); return; } // If we reach this point, no auto list is allowed. p->auto_part = false; // Everything but an else needs to be taken care of here, but else is // special. if (type != BC_LEX_KW_ELSE) { // After an if, no else found. if (BC_PARSE_IF_END(p)) { // Clear the expectation for else, end body, and return. Returning // gives us a clean slate for parsing again. bc_parse_noElse(p); if (p->flags.len > 1 && !BC_PARSE_BRACE(p)) { bc_parse_endBody(p, false); } return; } // With a left brace, we are parsing a body. else if (type == BC_LEX_LBRACE) { // We need to start a body if we are not expecting one yet. if (!BC_PARSE_BODY(p)) { bc_parse_startBody(p, BC_PARSE_FLAG_BRACE); bc_lex_next(&p->l); } // If we *are* expecting a body, that body should get a brace. This // takes care of braces being on a different line than if and loop // headers. else { *(BC_PARSE_TOP_FLAG_PTR(p)) |= BC_PARSE_FLAG_BRACE; bc_lex_next(&p->l); bc_parse_body(p, true); } // If we have reached this point, we need to return for a clean // slate. return; } // This happens when we are expecting a body and get a single statement, // i.e., a body with no braces surrounding it. Returns after for a clean // slate. else if (BC_PARSE_BODY(p) && !BC_PARSE_BRACE(p)) { bc_parse_body(p, false); return; } } len = p->flags.len; flags = BC_PARSE_TOP_FLAG(p); switch (type) { // All of these are valid for expressions. case BC_LEX_OP_INC: case BC_LEX_OP_DEC: case BC_LEX_OP_MINUS: case BC_LEX_OP_BOOL_NOT: case BC_LEX_LPAREN: case BC_LEX_NAME: case BC_LEX_NUMBER: case BC_LEX_KW_IBASE: case BC_LEX_KW_LAST: case BC_LEX_KW_LENGTH: case BC_LEX_KW_OBASE: case BC_LEX_KW_SCALE: #if BC_ENABLE_EXTRA_MATH case BC_LEX_KW_SEED: #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_KW_SQRT: case BC_LEX_KW_ABS: case BC_LEX_KW_IS_NUMBER: case BC_LEX_KW_IS_STRING: #if BC_ENABLE_EXTRA_MATH case BC_LEX_KW_IRAND: #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_KW_ASCIIFY: case BC_LEX_KW_MODEXP: case BC_LEX_KW_DIVMOD: case BC_LEX_KW_READ: #if BC_ENABLE_EXTRA_MATH case BC_LEX_KW_RAND: #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_KW_MAXIBASE: case BC_LEX_KW_MAXOBASE: case BC_LEX_KW_MAXSCALE: #if BC_ENABLE_EXTRA_MATH case BC_LEX_KW_MAXRAND: #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_KW_LINE_LENGTH: case BC_LEX_KW_GLOBAL_STACKS: case BC_LEX_KW_LEADING_ZERO: { bc_parse_expr_status(p, BC_PARSE_PRINT, bc_parse_next_expr); break; } case BC_LEX_KW_ELSE: { bc_parse_else(p); break; } // Just eat. case BC_LEX_SCOLON: { // Do nothing. break; } case BC_LEX_RBRACE: { bc_parse_endBody(p, true); break; } case BC_LEX_STR: { bc_parse_str(p, BC_INST_PRINT_STR); break; } case BC_LEX_KW_BREAK: case BC_LEX_KW_CONTINUE: { bc_parse_loopExit(p, p->l.t); break; } case BC_LEX_KW_FOR: { bc_parse_for(p); break; } case BC_LEX_KW_HALT: { bc_parse_push(p, BC_INST_HALT); bc_lex_next(&p->l); break; } case BC_LEX_KW_IF: { bc_parse_if(p); break; } case BC_LEX_KW_LIMITS: { // `limits` is a compile-time command, so execute it right away. bc_vm_printf("BC_LONG_BIT = %lu\n", (ulong) BC_LONG_BIT); bc_vm_printf("BC_BASE_DIGS = %lu\n", (ulong) BC_BASE_DIGS); bc_vm_printf("BC_BASE_POW = %lu\n", (ulong) BC_BASE_POW); bc_vm_printf("BC_OVERFLOW_MAX = %lu\n", (ulong) BC_NUM_BIGDIG_MAX); bc_vm_printf("\n"); bc_vm_printf("BC_BASE_MAX = %lu\n", BC_MAX_OBASE); bc_vm_printf("BC_DIM_MAX = %lu\n", BC_MAX_DIM); bc_vm_printf("BC_SCALE_MAX = %lu\n", BC_MAX_SCALE); bc_vm_printf("BC_STRING_MAX = %lu\n", BC_MAX_STRING); bc_vm_printf("BC_NAME_MAX = %lu\n", BC_MAX_NAME); bc_vm_printf("BC_NUM_MAX = %lu\n", BC_MAX_NUM); #if BC_ENABLE_EXTRA_MATH bc_vm_printf("BC_RAND_MAX = %lu\n", BC_MAX_RAND); #endif // BC_ENABLE_EXTRA_MATH bc_vm_printf("MAX Exponent = %lu\n", BC_MAX_EXP); bc_vm_printf("Number of vars = %lu\n", BC_MAX_VARS); bc_lex_next(&p->l); break; } case BC_LEX_KW_STREAM: case BC_LEX_KW_PRINT: { bc_parse_print(p, type); break; } case BC_LEX_KW_QUIT: { // Quit is a compile-time command. We don't exit directly, so the vm // can clean up. vm->status = BC_STATUS_QUIT; BC_JMP; break; } case BC_LEX_KW_RETURN: { bc_parse_return(p); break; } case BC_LEX_KW_WHILE: { bc_parse_while(p); break; } case BC_LEX_EOF: case BC_LEX_INVALID: case BC_LEX_NEG: #if BC_ENABLE_EXTRA_MATH case BC_LEX_OP_TRUNC: #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_OP_POWER: case BC_LEX_OP_MULTIPLY: case BC_LEX_OP_DIVIDE: case BC_LEX_OP_MODULUS: case BC_LEX_OP_PLUS: #if BC_ENABLE_EXTRA_MATH case BC_LEX_OP_PLACES: case BC_LEX_OP_LSHIFT: case BC_LEX_OP_RSHIFT: #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_OP_REL_EQ: case BC_LEX_OP_REL_LE: case BC_LEX_OP_REL_GE: case BC_LEX_OP_REL_NE: case BC_LEX_OP_REL_LT: case BC_LEX_OP_REL_GT: case BC_LEX_OP_BOOL_OR: case BC_LEX_OP_BOOL_AND: case BC_LEX_OP_ASSIGN_POWER: case BC_LEX_OP_ASSIGN_MULTIPLY: case BC_LEX_OP_ASSIGN_DIVIDE: case BC_LEX_OP_ASSIGN_MODULUS: case BC_LEX_OP_ASSIGN_PLUS: case BC_LEX_OP_ASSIGN_MINUS: #if BC_ENABLE_EXTRA_MATH case BC_LEX_OP_ASSIGN_PLACES: case BC_LEX_OP_ASSIGN_LSHIFT: case BC_LEX_OP_ASSIGN_RSHIFT: #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_OP_ASSIGN: case BC_LEX_NLINE: case BC_LEX_WHITESPACE: case BC_LEX_RPAREN: case BC_LEX_LBRACKET: case BC_LEX_COMMA: case BC_LEX_RBRACKET: case BC_LEX_LBRACE: case BC_LEX_KW_AUTO: case BC_LEX_KW_DEFINE: #if DC_ENABLED case BC_LEX_EXTENDED_REGISTERS: case BC_LEX_EQ_NO_REG: case BC_LEX_COLON: case BC_LEX_EXECUTE: case BC_LEX_PRINT_STACK: case BC_LEX_CLEAR_STACK: case BC_LEX_REG_STACK_LEVEL: case BC_LEX_STACK_LEVEL: case BC_LEX_DUPLICATE: case BC_LEX_SWAP: case BC_LEX_POP: case BC_LEX_STORE_IBASE: case BC_LEX_STORE_OBASE: case BC_LEX_STORE_SCALE: #if BC_ENABLE_EXTRA_MATH case BC_LEX_STORE_SEED: #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_LOAD: case BC_LEX_LOAD_POP: case BC_LEX_STORE_PUSH: case BC_LEX_PRINT_POP: case BC_LEX_NQUIT: case BC_LEX_EXEC_STACK_LENGTH: case BC_LEX_SCALE_FACTOR: case BC_LEX_ARRAY_LENGTH: #endif // DC_ENABLED { bc_parse_err(p, BC_ERR_PARSE_TOKEN); } } // If the flags did not change, we expect a delimiter. if (len == p->flags.len && flags == BC_PARSE_TOP_FLAG(p)) { if (BC_ERR(!bc_parse_isDelimiter(p))) { bc_parse_err(p, BC_ERR_PARSE_TOKEN); } } // Make sure semicolons are eaten. while (p->l.t == BC_LEX_SCOLON || p->l.t == BC_LEX_NLINE) { bc_lex_next(&p->l); } // POSIX's grammar does not allow a function definition after a semicolon // without a newline, so check specifically for that case and error if // the POSIX standard flag is set. if (p->l.last == BC_LEX_SCOLON && p->l.t == BC_LEX_KW_DEFINE && BC_IS_POSIX) { bc_parse_err(p, BC_ERR_POSIX_FUNC_AFTER_SEMICOLON); } } void bc_parse_parse(BcParse* p) { assert(p); BC_SETJMP_LOCKED(vm, exit); // We should not let an EOF get here unless some partial parse was not // completed, in which case, it's the user's fault. if (BC_ERR(p->l.t == BC_LEX_EOF)) bc_parse_err(p, BC_ERR_PARSE_EOF); // Functions need special parsing. else if (p->l.t == BC_LEX_KW_DEFINE) { if (BC_ERR(BC_PARSE_NO_EXEC(p))) { bc_parse_endif(p); if (BC_ERR(BC_PARSE_NO_EXEC(p))) { bc_parse_err(p, BC_ERR_PARSE_TOKEN); } } bc_parse_func(p); } // Otherwise, parse a normal statement. else bc_parse_stmt(p); exit: // We need to reset on error. if (BC_ERR(((vm->status && vm->status != BC_STATUS_QUIT) || vm->sig != 0))) { bc_parse_reset(p); } BC_LONGJMP_CONT(vm); BC_SIG_MAYLOCK; } /** * Parse an expression. This is the actual implementation of the Shunting-Yard * Algorithm. * @param p The parser. * @param flags The flags for what is valid in the expression. * @param next A set of tokens for what is valid *after* the expression. * @return A parse status. In some places, an empty expression is an * error, and sometimes, it is required. This allows this function * to tell the caller if the expression was empty and let the * caller handle it. */ static BcParseStatus bc_parse_expr_err(BcParse* p, uint8_t flags, BcParseNext next) { BcInst prev = BC_INST_PRINT; uchar inst = BC_INST_INVALID; BcLexType top, t; size_t nexprs, ops_bgn; uint32_t i, nparens, nrelops; - bool pfirst, rprn, done, get_token, assign, bin_last, incdec, can_assign; + bool pfirst, rprn, array_last, done, get_token, assign; + bool bin_last, incdec, can_assign; // One of these *must* be true. assert(!(flags & BC_PARSE_PRINT) || !(flags & BC_PARSE_NEEDVAL)); // These are set very carefully. In fact, controlling the values of these // locals is the biggest part of making this work. ops_bgn especially is // important because it marks where the operator stack begins for *this* // invocation of this function. That's because bc_parse_expr_err() is // recursive (the Shunting-Yard Algorithm is most easily expressed // recursively when parsing subexpressions), and each invocation needs to // know where to stop. // // - nparens is the number of left parens without matches. // - nrelops is the number of relational operators that appear in the expr. // - nexprs is the number of unused expressions. // - rprn is a right paren encountered last. + // - array_last is an array item encountered last. // - done means the expression has been fully parsed. // - get_token is true when a token is needed at the end of an iteration. // - assign is true when an assignment statement was parsed last. // - incdec is true when the previous operator was an inc or dec operator. // - can_assign is true when an assignemnt is valid. // - bin_last is true when the previous instruction was a binary operator. t = p->l.t; pfirst = (p->l.t == BC_LEX_LPAREN); nparens = nrelops = 0; nexprs = 0; ops_bgn = p->ops.len; - rprn = done = get_token = assign = incdec = can_assign = false; + rprn = array_last = done = get_token = assign = incdec = can_assign = false; bin_last = true; // We want to eat newlines if newlines are not a valid ending token. // This is for spacing in things like for loop headers. if (!(flags & BC_PARSE_NOREAD)) { while ((t = p->l.t) == BC_LEX_NLINE) { bc_lex_next(&p->l); } } // This is the Shunting-Yard algorithm loop. for (; !done && BC_PARSE_EXPR(t); t = p->l.t) { + // Make sure an array expression is not mixed with any others. However, + // a right parenthesis may end the expression, so we will need to take + // care of that right there. + if (BC_ERR(array_last && t != BC_LEX_RPAREN)) + { + bc_parse_err(p, BC_ERR_PARSE_EXPR); + } + switch (t) { case BC_LEX_OP_INC: case BC_LEX_OP_DEC: { // These operators can only be used with items that can be // assigned to. if (BC_ERR(incdec)) bc_parse_err(p, BC_ERR_PARSE_ASSIGN); bc_parse_incdec(p, &prev, &can_assign, &nexprs, flags); rprn = get_token = bin_last = false; incdec = true; flags &= ~(BC_PARSE_ARRAY); break; } #if BC_ENABLE_EXTRA_MATH case BC_LEX_OP_TRUNC: { // The previous token must have been a leaf expression, or the // operator is in the wrong place. if (BC_ERR(!BC_PARSE_LEAF(prev, bin_last, rprn))) { bc_parse_err(p, BC_ERR_PARSE_TOKEN); } // I can just add the instruction because // negative will already be taken care of. bc_parse_push(p, BC_INST_TRUNC); rprn = can_assign = incdec = false; get_token = true; flags &= ~(BC_PARSE_ARRAY); break; } #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_OP_MINUS: { bc_parse_minus(p, &prev, ops_bgn, rprn, bin_last, &nexprs); rprn = get_token = can_assign = false; // This is true if it was a binary operator last. bin_last = (prev == BC_INST_MINUS); if (bin_last) incdec = false; flags &= ~(BC_PARSE_ARRAY); break; } // All of this group, including the fallthrough, is to parse binary // operators. case BC_LEX_OP_ASSIGN_POWER: case BC_LEX_OP_ASSIGN_MULTIPLY: case BC_LEX_OP_ASSIGN_DIVIDE: case BC_LEX_OP_ASSIGN_MODULUS: case BC_LEX_OP_ASSIGN_PLUS: case BC_LEX_OP_ASSIGN_MINUS: #if BC_ENABLE_EXTRA_MATH case BC_LEX_OP_ASSIGN_PLACES: case BC_LEX_OP_ASSIGN_LSHIFT: case BC_LEX_OP_ASSIGN_RSHIFT: #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_OP_ASSIGN: { // We need to make sure the assignment is valid. if (!BC_PARSE_INST_VAR(prev)) { bc_parse_err(p, BC_ERR_PARSE_ASSIGN); } // Fallthrough. BC_FALLTHROUGH } case BC_LEX_OP_POWER: case BC_LEX_OP_MULTIPLY: case BC_LEX_OP_DIVIDE: case BC_LEX_OP_MODULUS: case BC_LEX_OP_PLUS: #if BC_ENABLE_EXTRA_MATH case BC_LEX_OP_PLACES: case BC_LEX_OP_LSHIFT: case BC_LEX_OP_RSHIFT: #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_OP_REL_EQ: case BC_LEX_OP_REL_LE: case BC_LEX_OP_REL_GE: case BC_LEX_OP_REL_NE: case BC_LEX_OP_REL_LT: case BC_LEX_OP_REL_GT: case BC_LEX_OP_BOOL_NOT: case BC_LEX_OP_BOOL_OR: case BC_LEX_OP_BOOL_AND: { // This is true if the operator if the token is a prefix // operator. This is only for boolean not. if (BC_PARSE_OP_PREFIX(t)) { // Prefix operators are only allowed after binary operators // or prefix operators. if (BC_ERR(!bin_last && !BC_PARSE_OP_PREFIX(p->l.last))) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } } // If we execute the else, that means we have a binary operator. // If the previous operator was a prefix or a binary operator, // then a binary operator is not allowed. else if (BC_ERR(BC_PARSE_PREV_PREFIX(prev) || bin_last)) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } nrelops += (t >= BC_LEX_OP_REL_EQ && t <= BC_LEX_OP_REL_GT); prev = BC_PARSE_TOKEN_INST(t); bc_parse_operator(p, t, ops_bgn, &nexprs); rprn = incdec = can_assign = false; get_token = true; bin_last = !BC_PARSE_OP_PREFIX(t); flags &= ~(BC_PARSE_ARRAY); break; } case BC_LEX_LPAREN: { // A left paren is *not* allowed right after a leaf expr. if (BC_ERR(BC_PARSE_LEAF(prev, bin_last, rprn))) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } nparens += 1; rprn = incdec = can_assign = false; get_token = true; // Push the paren onto the operator stack. bc_vec_push(&p->ops, &t); break; } case BC_LEX_RPAREN: { // This needs to be a status. The error is handled in // bc_parse_expr_status(). if (BC_ERR(p->l.last == BC_LEX_LPAREN)) { return BC_PARSE_STATUS_EMPTY_EXPR; } // The right paren must not come after a prefix or binary // operator. if (BC_ERR(bin_last || BC_PARSE_PREV_PREFIX(prev))) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } // If there are no parens left, we are done, but we need another // token. if (!nparens) { done = true; get_token = false; break; } + // Now that we know the right paren has not ended the + // expression, make sure an array expression is not mixed with + // any others. + if (BC_ERR(array_last)) + { + bc_parse_err(p, BC_ERR_PARSE_EXPR); + } + nparens -= 1; rprn = true; get_token = bin_last = incdec = false; bc_parse_rightParen(p, &nexprs); break; } case BC_LEX_STR: { // POSIX only allows strings alone. if (BC_IS_POSIX) bc_parse_err(p, BC_ERR_POSIX_EXPR_STRING); // A string is a leaf and cannot come right after a leaf. if (BC_ERR(BC_PARSE_LEAF(prev, bin_last, rprn))) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } bc_parse_addString(p); get_token = true; bin_last = rprn = false; nexprs += 1; break; } case BC_LEX_NAME: { // A name is a leaf and cannot come right after a leaf. if (BC_ERR(BC_PARSE_LEAF(prev, bin_last, rprn))) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } get_token = bin_last = false; bc_parse_name(p, &prev, &can_assign, flags & ~BC_PARSE_NOCALL); rprn = (prev == BC_INST_CALL); + array_last = (prev == BC_INST_ARRAY); nexprs += 1; flags &= ~(BC_PARSE_ARRAY); break; } case BC_LEX_NUMBER: { // A number is a leaf and cannot come right after a leaf. if (BC_ERR(BC_PARSE_LEAF(prev, bin_last, rprn))) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } // The number instruction is pushed in here. bc_parse_number(p); nexprs += 1; prev = BC_INST_NUM; get_token = true; rprn = bin_last = can_assign = false; flags &= ~(BC_PARSE_ARRAY); break; } case BC_LEX_KW_IBASE: case BC_LEX_KW_LAST: case BC_LEX_KW_OBASE: #if BC_ENABLE_EXTRA_MATH case BC_LEX_KW_SEED: #endif // BC_ENABLE_EXTRA_MATH { // All of these are leaves and cannot come right after a leaf. if (BC_ERR(BC_PARSE_LEAF(prev, bin_last, rprn))) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } prev = t - BC_LEX_KW_LAST + BC_INST_LAST; bc_parse_push(p, prev); get_token = can_assign = true; rprn = bin_last = false; nexprs += 1; flags &= ~(BC_PARSE_ARRAY); break; } case BC_LEX_KW_LENGTH: case BC_LEX_KW_SQRT: case BC_LEX_KW_ABS: case BC_LEX_KW_IS_NUMBER: case BC_LEX_KW_IS_STRING: #if BC_ENABLE_EXTRA_MATH case BC_LEX_KW_IRAND: #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_KW_ASCIIFY: { // All of these are leaves and cannot come right after a leaf. if (BC_ERR(BC_PARSE_LEAF(prev, bin_last, rprn))) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } bc_parse_builtin(p, t, flags, &prev); rprn = get_token = bin_last = incdec = can_assign = false; nexprs += 1; flags &= ~(BC_PARSE_ARRAY); break; } case BC_LEX_KW_READ: #if BC_ENABLE_EXTRA_MATH case BC_LEX_KW_RAND: #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_KW_MAXIBASE: case BC_LEX_KW_MAXOBASE: case BC_LEX_KW_MAXSCALE: #if BC_ENABLE_EXTRA_MATH case BC_LEX_KW_MAXRAND: #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_KW_LINE_LENGTH: case BC_LEX_KW_GLOBAL_STACKS: case BC_LEX_KW_LEADING_ZERO: { // All of these are leaves and cannot come right after a leaf. if (BC_ERR(BC_PARSE_LEAF(prev, bin_last, rprn))) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } // Error if we have read and it's not allowed. else if (t == BC_LEX_KW_READ && BC_ERR(flags & BC_PARSE_NOREAD)) { bc_parse_err(p, BC_ERR_EXEC_REC_READ); } prev = t - BC_LEX_KW_READ + BC_INST_READ; bc_parse_noArgBuiltin(p, prev); rprn = get_token = bin_last = incdec = can_assign = false; nexprs += 1; flags &= ~(BC_PARSE_ARRAY); break; } case BC_LEX_KW_SCALE: { // This is a leaf and cannot come right after a leaf. if (BC_ERR(BC_PARSE_LEAF(prev, bin_last, rprn))) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } // Scale needs special work because it can be a variable *or* a // function. bc_parse_scale(p, &prev, &can_assign, flags); rprn = get_token = bin_last = false; nexprs += 1; flags &= ~(BC_PARSE_ARRAY); break; } case BC_LEX_KW_MODEXP: case BC_LEX_KW_DIVMOD: { // This is a leaf and cannot come right after a leaf. if (BC_ERR(BC_PARSE_LEAF(prev, bin_last, rprn))) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } bc_parse_builtin3(p, t, flags, &prev); rprn = get_token = bin_last = incdec = can_assign = false; nexprs += 1; flags &= ~(BC_PARSE_ARRAY); break; } case BC_LEX_EOF: case BC_LEX_INVALID: case BC_LEX_NEG: case BC_LEX_NLINE: case BC_LEX_WHITESPACE: case BC_LEX_LBRACKET: case BC_LEX_COMMA: case BC_LEX_RBRACKET: case BC_LEX_LBRACE: case BC_LEX_SCOLON: case BC_LEX_RBRACE: case BC_LEX_KW_AUTO: case BC_LEX_KW_BREAK: case BC_LEX_KW_CONTINUE: case BC_LEX_KW_DEFINE: case BC_LEX_KW_FOR: case BC_LEX_KW_IF: case BC_LEX_KW_LIMITS: case BC_LEX_KW_RETURN: case BC_LEX_KW_WHILE: case BC_LEX_KW_HALT: case BC_LEX_KW_PRINT: case BC_LEX_KW_QUIT: case BC_LEX_KW_STREAM: case BC_LEX_KW_ELSE: #if DC_ENABLED case BC_LEX_EXTENDED_REGISTERS: case BC_LEX_EQ_NO_REG: case BC_LEX_COLON: case BC_LEX_EXECUTE: case BC_LEX_PRINT_STACK: case BC_LEX_CLEAR_STACK: case BC_LEX_REG_STACK_LEVEL: case BC_LEX_STACK_LEVEL: case BC_LEX_DUPLICATE: case BC_LEX_SWAP: case BC_LEX_POP: case BC_LEX_STORE_IBASE: case BC_LEX_STORE_OBASE: case BC_LEX_STORE_SCALE: #if BC_ENABLE_EXTRA_MATH case BC_LEX_STORE_SEED: #endif // BC_ENABLE_EXTRA_MATH case BC_LEX_LOAD: case BC_LEX_LOAD_POP: case BC_LEX_STORE_PUSH: case BC_LEX_PRINT_POP: case BC_LEX_NQUIT: case BC_LEX_EXEC_STACK_LENGTH: case BC_LEX_SCALE_FACTOR: case BC_LEX_ARRAY_LENGTH: #endif // DC_ENABLED { #if BC_DEBUG // We should never get here, even in debug builds. bc_parse_err(p, BC_ERR_PARSE_TOKEN); break; #endif // BC_DEBUG } } if (get_token) bc_lex_next(&p->l); } // Now that we have parsed the expression, we need to empty the operator // stack. while (p->ops.len > ops_bgn) { top = BC_PARSE_TOP_OP(p); assign = top >= BC_LEX_OP_ASSIGN_POWER && top <= BC_LEX_OP_ASSIGN; // There should not be *any* parens on the stack anymore. if (BC_ERR(top == BC_LEX_LPAREN || top == BC_LEX_RPAREN)) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } bc_parse_push(p, BC_PARSE_TOKEN_INST(top)); // Adjust the number of unused expressions. nexprs -= !BC_PARSE_OP_PREFIX(top); bc_vec_pop(&p->ops); incdec = false; } // There must be only one expression at the top. if (BC_ERR(nexprs != 1)) bc_parse_err(p, BC_ERR_PARSE_EXPR); // Check that the next token is correct. for (i = 0; i < next.len && t != next.tokens[i]; ++i) { continue; } if (BC_ERR(i == next.len && !bc_parse_isDelimiter(p))) { bc_parse_err(p, BC_ERR_PARSE_EXPR); } // Check that POSIX would be happy with the number of relational operators. if (!(flags & BC_PARSE_REL) && nrelops) { bc_parse_err(p, BC_ERR_POSIX_REL_POS); } else if ((flags & BC_PARSE_REL) && nrelops > 1) { bc_parse_err(p, BC_ERR_POSIX_MULTIREL); } // If this is true, then we might be in a situation where we don't print. // We would want to have the increment/decrement operator not make an extra // copy if it's not necessary. if (!(flags & BC_PARSE_NEEDVAL) && !pfirst) { // We have the easy case if the last operator was an assignment // operator. if (assign) { inst = *((uchar*) bc_vec_top(&p->func->code)); inst += (BC_INST_ASSIGN_POWER_NO_VAL - BC_INST_ASSIGN_POWER); incdec = false; } // If we have an inc/dec operator and we are *not* printing, implement // the optimization to get rid of the extra copy. else if (incdec && !(flags & BC_PARSE_PRINT)) { inst = *((uchar*) bc_vec_top(&p->func->code)); incdec = (inst <= BC_INST_DEC); inst = BC_INST_ASSIGN_PLUS_NO_VAL + (inst != BC_INST_INC && inst != BC_INST_ASSIGN_PLUS); } // This condition allows us to change the previous assignment // instruction (which does a copy) for a NO_VAL version, which does not. // This condition is set if either of the above if statements ends up // being true. if (inst >= BC_INST_ASSIGN_POWER_NO_VAL && inst <= BC_INST_ASSIGN_NO_VAL) { // Pop the previous assignment instruction and push a new one. // Inc/dec needs the extra instruction because it is now a binary // operator and needs a second operand. bc_vec_pop(&p->func->code); if (incdec) bc_parse_push(p, BC_INST_ONE); bc_parse_push(p, inst); } } // If we might have to print... if ((flags & BC_PARSE_PRINT)) { // With a paren first or the last operator not being an assignment, we // *do* want to print. if (pfirst || !assign) bc_parse_push(p, BC_INST_PRINT); } // We need to make sure to push a pop instruction for assignment statements // that will not print. The print will pop, but without it, we need to pop. else if (!(flags & BC_PARSE_NEEDVAL) && (inst < BC_INST_ASSIGN_POWER_NO_VAL || inst > BC_INST_ASSIGN_NO_VAL)) { bc_parse_push(p, BC_INST_POP); } // We want to eat newlines if newlines are not a valid ending token. // This is for spacing in things like for loop headers. // // Yes, this is one case where I reuse a variable for a different purpose; // in this case, incdec being true now means that newlines are not valid. for (incdec = true, i = 0; i < next.len && incdec; ++i) { incdec = (next.tokens[i] != BC_LEX_NLINE); } if (incdec) { while (p->l.t == BC_LEX_NLINE) { bc_lex_next(&p->l); } } return BC_PARSE_STATUS_SUCCESS; } /** * Parses an expression with bc_parse_expr_err(), but throws an error if it gets * an empty expression. * @param p The parser. * @param flags The flags for what is valid in the expression. * @param next A set of tokens for what is valid *after* the expression. */ static void bc_parse_expr_status(BcParse* p, uint8_t flags, BcParseNext next) { BcParseStatus s = bc_parse_expr_err(p, flags, next); if (BC_ERR(s == BC_PARSE_STATUS_EMPTY_EXPR)) { bc_parse_err(p, BC_ERR_PARSE_EMPTY_EXPR); } } void bc_parse_expr(BcParse* p, uint8_t flags) { assert(p); bc_parse_expr_status(p, flags, bc_parse_next_read); } #endif // BC_ENABLED diff --git a/src/data.c b/src/data.c index 00eda2cc4a5b..bb1a6796f752 100644 --- a/src/data.c +++ b/src/data.c @@ -1,1364 +1,1423 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * Constant data for bc. * */ #include #include #include #include #include #include #include #include #include #include #include #include #include #if !BC_ENABLE_LIBRARY #if BC_ENABLED /// The bc signal message and its length. const char bc_sig_msg[] = "\ninterrupt (type \"quit\" to exit)\n"; const uchar bc_sig_msg_len = (uchar) (sizeof(bc_sig_msg) - 1); #endif // BC_ENABLED #if DC_ENABLED /// The dc signal message and its length. const char dc_sig_msg[] = "\ninterrupt (type \"q\" to exit)\n"; const uchar dc_sig_msg_len = (uchar) (sizeof(dc_sig_msg) - 1); #endif // DC_ENABLED // clang-format off /// The copyright banner. const char bc_copyright[] = "Copyright (c) 2018-2024 Gavin D. Howard and contributors\n" "Report bugs at: https://git.gavinhoward.com/gavin/bc\n\n" "This is free software with ABSOLUTELY NO WARRANTY.\n"; // clang-format on #ifdef __OpenBSD__ #if BC_ENABLE_EXTRA_MATH #if BC_ENABLE_HISTORY /// The pledges for starting bc. const char bc_pledge_start[] = "rpath stdio tty unveil"; /// The final pledges with history enabled. const char bc_pledge_end_history[] = "rpath stdio tty"; #else // BC_ENABLE_HISTORY /// The pledges for starting bc. const char bc_pledge_start[] = "rpath stdio unveil"; #endif // BC_ENABLE_HISTORY /// The final pledges with history history disabled. const char bc_pledge_end[] = "rpath stdio"; #else // BC_ENABLE_EXTRA_MATH #if BC_ENABLE_HISTORY /// The pledges for starting bc. const char bc_pledge_start[] = "rpath stdio tty"; /// The final pledges with history enabled. const char bc_pledge_end_history[] = "stdio tty"; #else // BC_ENABLE_HISTORY /// The pledges for starting bc. const char bc_pledge_start[] = "rpath stdio"; #endif // BC_ENABLE_HISTORY /// The final pledges with history history disabled. const char bc_pledge_end[] = "stdio"; #endif // BC_ENABLE_EXTRA_MATH #else // __OpenBSD__ /// The pledges for starting bc. const char bc_pledge_start[] = ""; #if BC_ENABLE_HISTORY /// The final pledges with history enabled. const char bc_pledge_end_history[] = ""; #endif // BC_ENABLE_HISTORY /// The final pledges with history history disabled. const char bc_pledge_end[] = ""; #endif // __OpenBSD__ /// The list of long options. There is a zero set at the end for detecting the /// end. const BcOptLong bc_args_lopt[] = { { "digit-clamp", BC_OPT_NONE, 'c' }, { "expression", BC_OPT_REQUIRED, 'e' }, { "file", BC_OPT_REQUIRED, 'f' }, { "help", BC_OPT_NONE, 'h' }, { "interactive", BC_OPT_NONE, 'i' }, { "ibase", BC_OPT_REQUIRED, 'I' }, { "leading-zeroes", BC_OPT_NONE, 'z' }, { "no-line-length", BC_OPT_NONE, 'L' }, { "obase", BC_OPT_REQUIRED, 'O' }, { "no-digit-clamp", BC_OPT_NONE, 'C' }, { "no-prompt", BC_OPT_NONE, 'P' }, { "no-read-prompt", BC_OPT_NONE, 'R' }, { "scale", BC_OPT_REQUIRED, 'S' }, #if BC_ENABLE_EXTRA_MATH { "seed", BC_OPT_REQUIRED, 'E' }, #endif // BC_ENABLE_EXTRA_MATH #if BC_ENABLED { "global-stacks", BC_OPT_BC_ONLY, 'g' }, { "mathlib", BC_OPT_BC_ONLY, 'l' }, { "quiet", BC_OPT_BC_ONLY, 'q' }, { "redefine", BC_OPT_REQUIRED_BC_ONLY, 'r' }, { "standard", BC_OPT_BC_ONLY, 's' }, { "warn", BC_OPT_BC_ONLY, 'w' }, #endif // BC_ENABLED { "version", BC_OPT_NONE, 'v' }, { "version", BC_OPT_NONE, 'V' }, #if DC_ENABLED { "extended-register", BC_OPT_DC_ONLY, 'x' }, #endif // DC_ENABLED { NULL, 0, 0 }, }; +#if BC_ENABLE_OSSFUZZ + +const char* bc_fuzzer_args_c[] = { + "bc", + "-lqc", + "-e", + "seed = 82507683022933941343198991100880559238.7080266844215897551270760113" + "4734858017748592704189096562163085637164174146616055338762825421827784" + "566630725748836994171142578125", + NULL, +}; + +const char* dc_fuzzer_args_c[] = { + "dc", + "-xc", + "-e", + "82507683022933941343198991100880559238.7080266844215897551270760113" + "4734858017748592704189096562163085637164174146616055338762825421827784" + "566630725748836994171142578125j", + NULL, +}; + +const char* bc_fuzzer_args_C[] = { + "bc", + "-lqC", + "-e", + "seed = 82507683022933941343198991100880559238.7080266844215897551270760113" + "4734858017748592704189096562163085637164174146616055338762825421827784" + "566630725748836994171142578125", + NULL, +}; + +const char* dc_fuzzer_args_C[] = { + "dc", + "-xC", + "-e", + "82507683022933941343198991100880559238.7080266844215897551270760113" + "4734858017748592704189096562163085637164174146616055338762825421827784" + "566630725748836994171142578125j", + NULL, +}; + +const size_t bc_fuzzer_args_len = sizeof(bc_fuzzer_args_c) / sizeof(char*); + +#if BC_C11 + +_Static_assert(sizeof(bc_fuzzer_args_C) / sizeof(char*) == bc_fuzzer_args_len, + "Wrong number of bc fuzzer args"); + +_Static_assert(sizeof(dc_fuzzer_args_c) / sizeof(char*) == bc_fuzzer_args_len, + "Wrong number of dc fuzzer args"); + +_Static_assert(sizeof(dc_fuzzer_args_C) / sizeof(char*) == bc_fuzzer_args_len, + "Wrong number of dc fuzzer args"); + +#endif // BC_C11 + +#endif // BC_ENABLE_OSSFUZZ + // clang-format off /// The default error category strings. const char *bc_errs[] = { "Math error:", "Parse error:", "Runtime error:", "Fatal error:", #if BC_ENABLED "Warning:", #endif // BC_ENABLED }; // clang-format on /// The error category for each error. const uchar bc_err_ids[] = { BC_ERR_IDX_MATH, BC_ERR_IDX_MATH, BC_ERR_IDX_MATH, BC_ERR_IDX_MATH, BC_ERR_IDX_FATAL, BC_ERR_IDX_FATAL, BC_ERR_IDX_FATAL, BC_ERR_IDX_FATAL, BC_ERR_IDX_FATAL, BC_ERR_IDX_FATAL, BC_ERR_IDX_FATAL, BC_ERR_IDX_FATAL, BC_ERR_IDX_FATAL, BC_ERR_IDX_EXEC, BC_ERR_IDX_EXEC, BC_ERR_IDX_EXEC, BC_ERR_IDX_EXEC, BC_ERR_IDX_EXEC, BC_ERR_IDX_EXEC, BC_ERR_IDX_EXEC, BC_ERR_IDX_EXEC, BC_ERR_IDX_EXEC, BC_ERR_IDX_EXEC, BC_ERR_IDX_EXEC, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, #if BC_ENABLED BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, BC_ERR_IDX_PARSE, #endif // BC_ENABLED }; /// The default error messages. There are NULL pointers because the positions /// must be preserved for the locales. const char* const bc_err_msgs[] = { "negative number", "non-integer number", "overflow: number cannot fit", "divide by 0", "memory allocation failed", "I/O error", "cannot open file: %s", "file is not text: %s", "path is a directory: %s", "bad command-line option: \"%s\"", "option requires an argument: '%c' (\"%s\")", "option takes no arguments: '%c' (\"%s\")", "bad option argument: \"%s\"", "bad ibase: must be [%lu, %lu]", "bad obase: must be [%lu, %lu]", "bad scale: must be [%lu, %lu]", "bad read() expression", "read() call inside of a read() call", "variable or array element is the wrong type", #if DC_ENABLED "stack has too few elements", "stack for register \"%s\" has too few elements", #else // DC_ENABLED NULL, NULL, #endif // DC_ENABLED #if BC_ENABLED "wrong number of parameters; need %zu, have %zu", "undefined function: %s()", "cannot use a void value in an expression", #else NULL, NULL, NULL, #endif // BC_ENABLED "end of file", "bad character '%c'", "string end cannot be found", "comment end cannot be found", "bad token", #if BC_ENABLED "bad expression", "empty expression", "bad print or stream statement", "bad function definition", ("bad assignment: left side must be scale, ibase, " "obase, seed, last, var, or array element"), "no auto variable found", "function parameter or auto \"%s%s\" already exists", "block end cannot be found", "cannot return a value from void function: %s()", "var cannot be a reference: %s", "POSIX does not allow names longer than 1 character: %s", "POSIX does not allow '#' script comments", "POSIX does not allow the following keyword: %s", "POSIX does not allow a period ('.') as a shortcut for the last result", "POSIX requires parentheses around return expressions", "POSIX does not allow the following operator: %s", "POSIX does not allow comparison operators outside if statements or loops", "POSIX requires 0 or 1 comparison operators per condition", "POSIX requires all 3 parts of a for loop to be non-empty", "POSIX requires a newline between a semicolon and a function definition", #if BC_ENABLE_EXTRA_MATH "POSIX does not allow exponential notation", #else NULL, #endif // BC_ENABLE_EXTRA_MATH "POSIX does not allow array references as function parameters", "POSIX does not allow void functions", "POSIX requires the left brace be on the same line as the function header", "POSIX does not allow strings to be assigned to variables or arrays", #endif // BC_ENABLED }; #endif // !BC_ENABLE_LIBRARY /// The destructors corresponding to BcDtorType enum items. const BcVecFree bc_vec_dtors[] = { NULL, bc_vec_free, bc_num_free, #if !BC_ENABLE_LIBRARY #if BC_DEBUG bc_func_free, #endif // BC_DEBUG bc_slab_free, bc_const_free, bc_result_free, #if BC_ENABLE_HISTORY && !BC_ENABLE_LINE_LIB bc_history_string_free, #endif // BC_ENABLE_HISTORY && !BC_ENABLE_LINE_LIB #else // !BC_ENABLE_LIBRARY bcl_num_destruct, #endif // !BC_ENABLE_LIBRARY }; #if !BC_ENABLE_LIBRARY #if BC_ENABLE_EDITLINE /// The normal path to the editrc. const char bc_history_editrc[] = "/.editrc"; /// The length of the normal path to the editrc. const size_t bc_history_editrc_len = sizeof(bc_history_editrc) - 1; #endif // BC_ENABLE_EDITLINE #if BC_ENABLE_HISTORY && !BC_ENABLE_LINE_LIB /// A flush type for not clearing current extras but not saving new ones either. const BcFlushType bc_flush_none = BC_FLUSH_NO_EXTRAS_NO_CLEAR; /// A flush type for clearing extras and not saving new ones. const BcFlushType bc_flush_err = BC_FLUSH_NO_EXTRAS_CLEAR; /// A flush type for clearing previous extras and saving new ones. const BcFlushType bc_flush_save = BC_FLUSH_SAVE_EXTRAS_CLEAR; /// A list of known bad terminals. const char* bc_history_bad_terms[] = { "dumb", "cons25", "emacs", NULL }; /// A constant for tabs and its length. My tab handling is dumb and always /// outputs the entire thing. const char bc_history_tab[] = "\t"; const size_t bc_history_tab_len = sizeof(bc_history_tab) - 1; /// A list of wide chars. These are listed in ascending order for efficiency. const uint32_t bc_history_wchars[][2] = { { 0x1100, 0x115F }, { 0x231A, 0x231B }, { 0x2329, 0x232A }, { 0x23E9, 0x23EC }, { 0x23F0, 0x23F0 }, { 0x23F3, 0x23F3 }, { 0x25FD, 0x25FE }, { 0x2614, 0x2615 }, { 0x2648, 0x2653 }, { 0x267F, 0x267F }, { 0x2693, 0x2693 }, { 0x26A1, 0x26A1 }, { 0x26AA, 0x26AB }, { 0x26BD, 0x26BE }, { 0x26C4, 0x26C5 }, { 0x26CE, 0x26CE }, { 0x26D4, 0x26D4 }, { 0x26EA, 0x26EA }, { 0x26F2, 0x26F3 }, { 0x26F5, 0x26F5 }, { 0x26FA, 0x26FA }, { 0x26FD, 0x26FD }, { 0x2705, 0x2705 }, { 0x270A, 0x270B }, { 0x2728, 0x2728 }, { 0x274C, 0x274C }, { 0x274E, 0x274E }, { 0x2753, 0x2755 }, { 0x2757, 0x2757 }, { 0x2795, 0x2797 }, { 0x27B0, 0x27B0 }, { 0x27BF, 0x27BF }, { 0x2B1B, 0x2B1C }, { 0x2B50, 0x2B50 }, { 0x2B55, 0x2B55 }, { 0x2E80, 0x2E99 }, { 0x2E9B, 0x2EF3 }, { 0x2F00, 0x2FD5 }, { 0x2FF0, 0x2FFB }, { 0x3001, 0x303E }, { 0x3041, 0x3096 }, { 0x3099, 0x30FF }, { 0x3105, 0x312D }, { 0x3131, 0x318E }, { 0x3190, 0x31BA }, { 0x31C0, 0x31E3 }, { 0x31F0, 0x321E }, { 0x3220, 0x3247 }, { 0x3250, 0x32FE }, { 0x3300, 0x4DBF }, { 0x4E00, 0xA48C }, { 0xA490, 0xA4C6 }, { 0xA960, 0xA97C }, { 0xAC00, 0xD7A3 }, { 0xF900, 0xFAFF }, { 0xFE10, 0xFE19 }, { 0xFE30, 0xFE52 }, { 0xFE54, 0xFE66 }, { 0xFE68, 0xFE6B }, { 0x16FE0, 0x16FE0 }, { 0x17000, 0x187EC }, { 0x18800, 0x18AF2 }, { 0x1B000, 0x1B001 }, { 0x1F004, 0x1F004 }, { 0x1F0CF, 0x1F0CF }, { 0x1F18E, 0x1F18E }, { 0x1F191, 0x1F19A }, { 0x1F200, 0x1F202 }, { 0x1F210, 0x1F23B }, { 0x1F240, 0x1F248 }, { 0x1F250, 0x1F251 }, { 0x1F300, 0x1F320 }, { 0x1F32D, 0x1F335 }, { 0x1F337, 0x1F37C }, { 0x1F37E, 0x1F393 }, { 0x1F3A0, 0x1F3CA }, { 0x1F3CF, 0x1F3D3 }, { 0x1F3E0, 0x1F3F0 }, { 0x1F3F4, 0x1F3F4 }, { 0x1F3F8, 0x1F43E }, { 0x1F440, 0x1F440 }, { 0x1F442, 0x1F4FC }, { 0x1F4FF, 0x1F53D }, { 0x1F54B, 0x1F54E }, { 0x1F550, 0x1F567 }, { 0x1F57A, 0x1F57A }, { 0x1F595, 0x1F596 }, { 0x1F5A4, 0x1F5A4 }, { 0x1F5FB, 0x1F64F }, { 0x1F680, 0x1F6C5 }, { 0x1F6CC, 0x1F6CC }, { 0x1F6D0, 0x1F6D2 }, { 0x1F6EB, 0x1F6EC }, { 0x1F6F4, 0x1F6F6 }, { 0x1F910, 0x1F91E }, { 0x1F920, 0x1F927 }, { 0x1F930, 0x1F930 }, { 0x1F933, 0x1F93E }, { 0x1F940, 0x1F94B }, { 0x1F950, 0x1F95E }, { 0x1F980, 0x1F991 }, { 0x1F9C0, 0x1F9C0 }, { 0x20000, 0x2FFFD }, { 0x30000, 0x3FFFD }, }; /// The length of the wide chars list. const size_t bc_history_wchars_len = sizeof(bc_history_wchars) / sizeof(bc_history_wchars[0]); /// A list of combining characters in Unicode. These are listed in ascending /// order for efficiency. const uint32_t bc_history_combo_chars[] = { 0x0300, 0x0301, 0x0302, 0x0303, 0x0304, 0x0305, 0x0306, 0x0307, 0x0308, 0x0309, 0x030A, 0x030B, 0x030C, 0x030D, 0x030E, 0x030F, 0x0310, 0x0311, 0x0312, 0x0313, 0x0314, 0x0315, 0x0316, 0x0317, 0x0318, 0x0319, 0x031A, 0x031B, 0x031C, 0x031D, 0x031E, 0x031F, 0x0320, 0x0321, 0x0322, 0x0323, 0x0324, 0x0325, 0x0326, 0x0327, 0x0328, 0x0329, 0x032A, 0x032B, 0x032C, 0x032D, 0x032E, 0x032F, 0x0330, 0x0331, 0x0332, 0x0333, 0x0334, 0x0335, 0x0336, 0x0337, 0x0338, 0x0339, 0x033A, 0x033B, 0x033C, 0x033D, 0x033E, 0x033F, 0x0340, 0x0341, 0x0342, 0x0343, 0x0344, 0x0345, 0x0346, 0x0347, 0x0348, 0x0349, 0x034A, 0x034B, 0x034C, 0x034D, 0x034E, 0x034F, 0x0350, 0x0351, 0x0352, 0x0353, 0x0354, 0x0355, 0x0356, 0x0357, 0x0358, 0x0359, 0x035A, 0x035B, 0x035C, 0x035D, 0x035E, 0x035F, 0x0360, 0x0361, 0x0362, 0x0363, 0x0364, 0x0365, 0x0366, 0x0367, 0x0368, 0x0369, 0x036A, 0x036B, 0x036C, 0x036D, 0x036E, 0x036F, 0x0483, 0x0484, 0x0485, 0x0486, 0x0487, 0x0591, 0x0592, 0x0593, 0x0594, 0x0595, 0x0596, 0x0597, 0x0598, 0x0599, 0x059A, 0x059B, 0x059C, 0x059D, 0x059E, 0x059F, 0x05A0, 0x05A1, 0x05A2, 0x05A3, 0x05A4, 0x05A5, 0x05A6, 0x05A7, 0x05A8, 0x05A9, 0x05AA, 0x05AB, 0x05AC, 0x05AD, 0x05AE, 0x05AF, 0x05B0, 0x05B1, 0x05B2, 0x05B3, 0x05B4, 0x05B5, 0x05B6, 0x05B7, 0x05B8, 0x05B9, 0x05BA, 0x05BB, 0x05BC, 0x05BD, 0x05BF, 0x05C1, 0x05C2, 0x05C4, 0x05C5, 0x05C7, 0x0610, 0x0611, 0x0612, 0x0613, 0x0614, 0x0615, 0x0616, 0x0617, 0x0618, 0x0619, 0x061A, 0x064B, 0x064C, 0x064D, 0x064E, 0x064F, 0x0650, 0x0651, 0x0652, 0x0653, 0x0654, 0x0655, 0x0656, 0x0657, 0x0658, 0x0659, 0x065A, 0x065B, 0x065C, 0x065D, 0x065E, 0x065F, 0x0670, 0x06D6, 0x06D7, 0x06D8, 0x06D9, 0x06DA, 0x06DB, 0x06DC, 0x06DF, 0x06E0, 0x06E1, 0x06E2, 0x06E3, 0x06E4, 0x06E7, 0x06E8, 0x06EA, 0x06EB, 0x06EC, 0x06ED, 0x0711, 0x0730, 0x0731, 0x0732, 0x0733, 0x0734, 0x0735, 0x0736, 0x0737, 0x0738, 0x0739, 0x073A, 0x073B, 0x073C, 0x073D, 0x073E, 0x073F, 0x0740, 0x0741, 0x0742, 0x0743, 0x0744, 0x0745, 0x0746, 0x0747, 0x0748, 0x0749, 0x074A, 0x07A6, 0x07A7, 0x07A8, 0x07A9, 0x07AA, 0x07AB, 0x07AC, 0x07AD, 0x07AE, 0x07AF, 0x07B0, 0x07EB, 0x07EC, 0x07ED, 0x07EE, 0x07EF, 0x07F0, 0x07F1, 0x07F2, 0x07F3, 0x0816, 0x0817, 0x0818, 0x0819, 0x081B, 0x081C, 0x081D, 0x081E, 0x081F, 0x0820, 0x0821, 0x0822, 0x0823, 0x0825, 0x0826, 0x0827, 0x0829, 0x082A, 0x082B, 0x082C, 0x082D, 0x0859, 0x085A, 0x085B, 0x08D4, 0x08D5, 0x08D6, 0x08D7, 0x08D8, 0x08D9, 0x08DA, 0x08DB, 0x08DC, 0x08DD, 0x08DE, 0x08DF, 0x08E0, 0x08E1, 0x08E3, 0x08E4, 0x08E5, 0x08E6, 0x08E7, 0x08E8, 0x08E9, 0x08EA, 0x08EB, 0x08EC, 0x08ED, 0x08EE, 0x08EF, 0x08F0, 0x08F1, 0x08F2, 0x08F3, 0x08F4, 0x08F5, 0x08F6, 0x08F7, 0x08F8, 0x08F9, 0x08FA, 0x08FB, 0x08FC, 0x08FD, 0x08FE, 0x08FF, 0x0900, 0x0901, 0x0902, 0x093A, 0x093C, 0x0941, 0x0942, 0x0943, 0x0944, 0x0945, 0x0946, 0x0947, 0x0948, 0x094D, 0x0951, 0x0952, 0x0953, 0x0954, 0x0955, 0x0956, 0x0957, 0x0962, 0x0963, 0x0981, 0x09BC, 0x09C1, 0x09C2, 0x09C3, 0x09C4, 0x09CD, 0x09E2, 0x09E3, 0x0A01, 0x0A02, 0x0A3C, 0x0A41, 0x0A42, 0x0A47, 0x0A48, 0x0A4B, 0x0A4C, 0x0A4D, 0x0A51, 0x0A70, 0x0A71, 0x0A75, 0x0A81, 0x0A82, 0x0ABC, 0x0AC1, 0x0AC2, 0x0AC3, 0x0AC4, 0x0AC5, 0x0AC7, 0x0AC8, 0x0ACD, 0x0AE2, 0x0AE3, 0x0B01, 0x0B3C, 0x0B3F, 0x0B41, 0x0B42, 0x0B43, 0x0B44, 0x0B4D, 0x0B56, 0x0B62, 0x0B63, 0x0B82, 0x0BC0, 0x0BCD, 0x0C00, 0x0C3E, 0x0C3F, 0x0C40, 0x0C46, 0x0C47, 0x0C48, 0x0C4A, 0x0C4B, 0x0C4C, 0x0C4D, 0x0C55, 0x0C56, 0x0C62, 0x0C63, 0x0C81, 0x0CBC, 0x0CBF, 0x0CC6, 0x0CCC, 0x0CCD, 0x0CE2, 0x0CE3, 0x0D01, 0x0D41, 0x0D42, 0x0D43, 0x0D44, 0x0D4D, 0x0D62, 0x0D63, 0x0DCA, 0x0DD2, 0x0DD3, 0x0DD4, 0x0DD6, 0x0E31, 0x0E34, 0x0E35, 0x0E36, 0x0E37, 0x0E38, 0x0E39, 0x0E3A, 0x0E47, 0x0E48, 0x0E49, 0x0E4A, 0x0E4B, 0x0E4C, 0x0E4D, 0x0E4E, 0x0EB1, 0x0EB4, 0x0EB5, 0x0EB6, 0x0EB7, 0x0EB8, 0x0EB9, 0x0EBB, 0x0EBC, 0x0EC8, 0x0EC9, 0x0ECA, 0x0ECB, 0x0ECC, 0x0ECD, 0x0F18, 0x0F19, 0x0F35, 0x0F37, 0x0F39, 0x0F71, 0x0F72, 0x0F73, 0x0F74, 0x0F75, 0x0F76, 0x0F77, 0x0F78, 0x0F79, 0x0F7A, 0x0F7B, 0x0F7C, 0x0F7D, 0x0F7E, 0x0F80, 0x0F81, 0x0F82, 0x0F83, 0x0F84, 0x0F86, 0x0F87, 0x0F8D, 0x0F8E, 0x0F8F, 0x0F90, 0x0F91, 0x0F92, 0x0F93, 0x0F94, 0x0F95, 0x0F96, 0x0F97, 0x0F99, 0x0F9A, 0x0F9B, 0x0F9C, 0x0F9D, 0x0F9E, 0x0F9F, 0x0FA0, 0x0FA1, 0x0FA2, 0x0FA3, 0x0FA4, 0x0FA5, 0x0FA6, 0x0FA7, 0x0FA8, 0x0FA9, 0x0FAA, 0x0FAB, 0x0FAC, 0x0FAD, 0x0FAE, 0x0FAF, 0x0FB0, 0x0FB1, 0x0FB2, 0x0FB3, 0x0FB4, 0x0FB5, 0x0FB6, 0x0FB7, 0x0FB8, 0x0FB9, 0x0FBA, 0x0FBB, 0x0FBC, 0x0FC6, 0x102D, 0x102E, 0x102F, 0x1030, 0x1032, 0x1033, 0x1034, 0x1035, 0x1036, 0x1037, 0x1039, 0x103A, 0x103D, 0x103E, 0x1058, 0x1059, 0x105E, 0x105F, 0x1060, 0x1071, 0x1072, 0x1073, 0x1074, 0x1082, 0x1085, 0x1086, 0x108D, 0x109D, 0x135D, 0x135E, 0x135F, 0x1712, 0x1713, 0x1714, 0x1732, 0x1733, 0x1734, 0x1752, 0x1753, 0x1772, 0x1773, 0x17B4, 0x17B5, 0x17B7, 0x17B8, 0x17B9, 0x17BA, 0x17BB, 0x17BC, 0x17BD, 0x17C6, 0x17C9, 0x17CA, 0x17CB, 0x17CC, 0x17CD, 0x17CE, 0x17CF, 0x17D0, 0x17D1, 0x17D2, 0x17D3, 0x17DD, 0x180B, 0x180C, 0x180D, 0x1885, 0x1886, 0x18A9, 0x1920, 0x1921, 0x1922, 0x1927, 0x1928, 0x1932, 0x1939, 0x193A, 0x193B, 0x1A17, 0x1A18, 0x1A1B, 0x1A56, 0x1A58, 0x1A59, 0x1A5A, 0x1A5B, 0x1A5C, 0x1A5D, 0x1A5E, 0x1A60, 0x1A62, 0x1A65, 0x1A66, 0x1A67, 0x1A68, 0x1A69, 0x1A6A, 0x1A6B, 0x1A6C, 0x1A73, 0x1A74, 0x1A75, 0x1A76, 0x1A77, 0x1A78, 0x1A79, 0x1A7A, 0x1A7B, 0x1A7C, 0x1A7F, 0x1AB0, 0x1AB1, 0x1AB2, 0x1AB3, 0x1AB4, 0x1AB5, 0x1AB6, 0x1AB7, 0x1AB8, 0x1AB9, 0x1ABA, 0x1ABB, 0x1ABC, 0x1ABD, 0x1B00, 0x1B01, 0x1B02, 0x1B03, 0x1B34, 0x1B36, 0x1B37, 0x1B38, 0x1B39, 0x1B3A, 0x1B3C, 0x1B42, 0x1B6B, 0x1B6C, 0x1B6D, 0x1B6E, 0x1B6F, 0x1B70, 0x1B71, 0x1B72, 0x1B73, 0x1B80, 0x1B81, 0x1BA2, 0x1BA3, 0x1BA4, 0x1BA5, 0x1BA8, 0x1BA9, 0x1BAB, 0x1BAC, 0x1BAD, 0x1BE6, 0x1BE8, 0x1BE9, 0x1BED, 0x1BEF, 0x1BF0, 0x1BF1, 0x1C2C, 0x1C2D, 0x1C2E, 0x1C2F, 0x1C30, 0x1C31, 0x1C32, 0x1C33, 0x1C36, 0x1C37, 0x1CD0, 0x1CD1, 0x1CD2, 0x1CD4, 0x1CD5, 0x1CD6, 0x1CD7, 0x1CD8, 0x1CD9, 0x1CDA, 0x1CDB, 0x1CDC, 0x1CDD, 0x1CDE, 0x1CDF, 0x1CE0, 0x1CE2, 0x1CE3, 0x1CE4, 0x1CE5, 0x1CE6, 0x1CE7, 0x1CE8, 0x1CED, 0x1CF4, 0x1CF8, 0x1CF9, 0x1DC0, 0x1DC1, 0x1DC2, 0x1DC3, 0x1DC4, 0x1DC5, 0x1DC6, 0x1DC7, 0x1DC8, 0x1DC9, 0x1DCA, 0x1DCB, 0x1DCC, 0x1DCD, 0x1DCE, 0x1DCF, 0x1DD0, 0x1DD1, 0x1DD2, 0x1DD3, 0x1DD4, 0x1DD5, 0x1DD6, 0x1DD7, 0x1DD8, 0x1DD9, 0x1DDA, 0x1DDB, 0x1DDC, 0x1DDD, 0x1DDE, 0x1DDF, 0x1DE0, 0x1DE1, 0x1DE2, 0x1DE3, 0x1DE4, 0x1DE5, 0x1DE6, 0x1DE7, 0x1DE8, 0x1DE9, 0x1DEA, 0x1DEB, 0x1DEC, 0x1DED, 0x1DEE, 0x1DEF, 0x1DF0, 0x1DF1, 0x1DF2, 0x1DF3, 0x1DF4, 0x1DF5, 0x1DFB, 0x1DFC, 0x1DFD, 0x1DFE, 0x1DFF, 0x20D0, 0x20D1, 0x20D2, 0x20D3, 0x20D4, 0x20D5, 0x20D6, 0x20D7, 0x20D8, 0x20D9, 0x20DA, 0x20DB, 0x20DC, 0x20E1, 0x20E5, 0x20E6, 0x20E7, 0x20E8, 0x20E9, 0x20EA, 0x20EB, 0x20EC, 0x20ED, 0x20EE, 0x20EF, 0x20F0, 0x2CEF, 0x2CF0, 0x2CF1, 0x2D7F, 0x2DE0, 0x2DE1, 0x2DE2, 0x2DE3, 0x2DE4, 0x2DE5, 0x2DE6, 0x2DE7, 0x2DE8, 0x2DE9, 0x2DEA, 0x2DEB, 0x2DEC, 0x2DED, 0x2DEE, 0x2DEF, 0x2DF0, 0x2DF1, 0x2DF2, 0x2DF3, 0x2DF4, 0x2DF5, 0x2DF6, 0x2DF7, 0x2DF8, 0x2DF9, 0x2DFA, 0x2DFB, 0x2DFC, 0x2DFD, 0x2DFE, 0x2DFF, 0x302A, 0x302B, 0x302C, 0x302D, 0x3099, 0x309A, 0xA66F, 0xA674, 0xA675, 0xA676, 0xA677, 0xA678, 0xA679, 0xA67A, 0xA67B, 0xA67C, 0xA67D, 0xA69E, 0xA69F, 0xA6F0, 0xA6F1, 0xA802, 0xA806, 0xA80B, 0xA825, 0xA826, 0xA8C4, 0xA8C5, 0xA8E0, 0xA8E1, 0xA8E2, 0xA8E3, 0xA8E4, 0xA8E5, 0xA8E6, 0xA8E7, 0xA8E8, 0xA8E9, 0xA8EA, 0xA8EB, 0xA8EC, 0xA8ED, 0xA8EE, 0xA8EF, 0xA8F0, 0xA8F1, 0xA926, 0xA927, 0xA928, 0xA929, 0xA92A, 0xA92B, 0xA92C, 0xA92D, 0xA947, 0xA948, 0xA949, 0xA94A, 0xA94B, 0xA94C, 0xA94D, 0xA94E, 0xA94F, 0xA950, 0xA951, 0xA980, 0xA981, 0xA982, 0xA9B3, 0xA9B6, 0xA9B7, 0xA9B8, 0xA9B9, 0xA9BC, 0xA9E5, 0xAA29, 0xAA2A, 0xAA2B, 0xAA2C, 0xAA2D, 0xAA2E, 0xAA31, 0xAA32, 0xAA35, 0xAA36, 0xAA43, 0xAA4C, 0xAA7C, 0xAAB0, 0xAAB2, 0xAAB3, 0xAAB4, 0xAAB7, 0xAAB8, 0xAABE, 0xAABF, 0xAAC1, 0xAAEC, 0xAAED, 0xAAF6, 0xABE5, 0xABE8, 0xABED, 0xFB1E, 0xFE00, 0xFE01, 0xFE02, 0xFE03, 0xFE04, 0xFE05, 0xFE06, 0xFE07, 0xFE08, 0xFE09, 0xFE0A, 0xFE0B, 0xFE0C, 0xFE0D, 0xFE0E, 0xFE0F, 0xFE20, 0xFE21, 0xFE22, 0xFE23, 0xFE24, 0xFE25, 0xFE26, 0xFE27, 0xFE28, 0xFE29, 0xFE2A, 0xFE2B, 0xFE2C, 0xFE2D, 0xFE2E, 0xFE2F, 0x101FD, 0x102E0, 0x10376, 0x10377, 0x10378, 0x10379, 0x1037A, 0x10A01, 0x10A02, 0x10A03, 0x10A05, 0x10A06, 0x10A0C, 0x10A0D, 0x10A0E, 0x10A0F, 0x10A38, 0x10A39, 0x10A3A, 0x10A3F, 0x10AE5, 0x10AE6, 0x11001, 0x11038, 0x11039, 0x1103A, 0x1103B, 0x1103C, 0x1103D, 0x1103E, 0x1103F, 0x11040, 0x11041, 0x11042, 0x11043, 0x11044, 0x11045, 0x11046, 0x1107F, 0x11080, 0x11081, 0x110B3, 0x110B4, 0x110B5, 0x110B6, 0x110B9, 0x110BA, 0x11100, 0x11101, 0x11102, 0x11127, 0x11128, 0x11129, 0x1112A, 0x1112B, 0x1112D, 0x1112E, 0x1112F, 0x11130, 0x11131, 0x11132, 0x11133, 0x11134, 0x11173, 0x11180, 0x11181, 0x111B6, 0x111B7, 0x111B8, 0x111B9, 0x111BA, 0x111BB, 0x111BC, 0x111BD, 0x111BE, 0x111CA, 0x111CB, 0x111CC, 0x1122F, 0x11230, 0x11231, 0x11234, 0x11236, 0x11237, 0x1123E, 0x112DF, 0x112E3, 0x112E4, 0x112E5, 0x112E6, 0x112E7, 0x112E8, 0x112E9, 0x112EA, 0x11300, 0x11301, 0x1133C, 0x11340, 0x11366, 0x11367, 0x11368, 0x11369, 0x1136A, 0x1136B, 0x1136C, 0x11370, 0x11371, 0x11372, 0x11373, 0x11374, 0x11438, 0x11439, 0x1143A, 0x1143B, 0x1143C, 0x1143D, 0x1143E, 0x1143F, 0x11442, 0x11443, 0x11444, 0x11446, 0x114B3, 0x114B4, 0x114B5, 0x114B6, 0x114B7, 0x114B8, 0x114BA, 0x114BF, 0x114C0, 0x114C2, 0x114C3, 0x115B2, 0x115B3, 0x115B4, 0x115B5, 0x115BC, 0x115BD, 0x115BF, 0x115C0, 0x115DC, 0x115DD, 0x11633, 0x11634, 0x11635, 0x11636, 0x11637, 0x11638, 0x11639, 0x1163A, 0x1163D, 0x1163F, 0x11640, 0x116AB, 0x116AD, 0x116B0, 0x116B1, 0x116B2, 0x116B3, 0x116B4, 0x116B5, 0x116B7, 0x1171D, 0x1171E, 0x1171F, 0x11722, 0x11723, 0x11724, 0x11725, 0x11727, 0x11728, 0x11729, 0x1172A, 0x1172B, 0x11C30, 0x11C31, 0x11C32, 0x11C33, 0x11C34, 0x11C35, 0x11C36, 0x11C38, 0x11C39, 0x11C3A, 0x11C3B, 0x11C3C, 0x11C3D, 0x11C3F, 0x11C92, 0x11C93, 0x11C94, 0x11C95, 0x11C96, 0x11C97, 0x11C98, 0x11C99, 0x11C9A, 0x11C9B, 0x11C9C, 0x11C9D, 0x11C9E, 0x11C9F, 0x11CA0, 0x11CA1, 0x11CA2, 0x11CA3, 0x11CA4, 0x11CA5, 0x11CA6, 0x11CA7, 0x11CAA, 0x11CAB, 0x11CAC, 0x11CAD, 0x11CAE, 0x11CAF, 0x11CB0, 0x11CB2, 0x11CB3, 0x11CB5, 0x11CB6, 0x16AF0, 0x16AF1, 0x16AF2, 0x16AF3, 0x16AF4, 0x16B30, 0x16B31, 0x16B32, 0x16B33, 0x16B34, 0x16B35, 0x16B36, 0x16F8F, 0x16F90, 0x16F91, 0x16F92, 0x1BC9D, 0x1BC9E, 0x1D167, 0x1D168, 0x1D169, 0x1D17B, 0x1D17C, 0x1D17D, 0x1D17E, 0x1D17F, 0x1D180, 0x1D181, 0x1D182, 0x1D185, 0x1D186, 0x1D187, 0x1D188, 0x1D189, 0x1D18A, 0x1D18B, 0x1D1AA, 0x1D1AB, 0x1D1AC, 0x1D1AD, 0x1D242, 0x1D243, 0x1D244, 0x1DA00, 0x1DA01, 0x1DA02, 0x1DA03, 0x1DA04, 0x1DA05, 0x1DA06, 0x1DA07, 0x1DA08, 0x1DA09, 0x1DA0A, 0x1DA0B, 0x1DA0C, 0x1DA0D, 0x1DA0E, 0x1DA0F, 0x1DA10, 0x1DA11, 0x1DA12, 0x1DA13, 0x1DA14, 0x1DA15, 0x1DA16, 0x1DA17, 0x1DA18, 0x1DA19, 0x1DA1A, 0x1DA1B, 0x1DA1C, 0x1DA1D, 0x1DA1E, 0x1DA1F, 0x1DA20, 0x1DA21, 0x1DA22, 0x1DA23, 0x1DA24, 0x1DA25, 0x1DA26, 0x1DA27, 0x1DA28, 0x1DA29, 0x1DA2A, 0x1DA2B, 0x1DA2C, 0x1DA2D, 0x1DA2E, 0x1DA2F, 0x1DA30, 0x1DA31, 0x1DA32, 0x1DA33, 0x1DA34, 0x1DA35, 0x1DA36, 0x1DA3B, 0x1DA3C, 0x1DA3D, 0x1DA3E, 0x1DA3F, 0x1DA40, 0x1DA41, 0x1DA42, 0x1DA43, 0x1DA44, 0x1DA45, 0x1DA46, 0x1DA47, 0x1DA48, 0x1DA49, 0x1DA4A, 0x1DA4B, 0x1DA4C, 0x1DA4D, 0x1DA4E, 0x1DA4F, 0x1DA50, 0x1DA51, 0x1DA52, 0x1DA53, 0x1DA54, 0x1DA55, 0x1DA56, 0x1DA57, 0x1DA58, 0x1DA59, 0x1DA5A, 0x1DA5B, 0x1DA5C, 0x1DA5D, 0x1DA5E, 0x1DA5F, 0x1DA60, 0x1DA61, 0x1DA62, 0x1DA63, 0x1DA64, 0x1DA65, 0x1DA66, 0x1DA67, 0x1DA68, 0x1DA69, 0x1DA6A, 0x1DA6B, 0x1DA6C, 0x1DA75, 0x1DA84, 0x1DA9B, 0x1DA9C, 0x1DA9D, 0x1DA9E, 0x1DA9F, 0x1DAA1, 0x1DAA2, 0x1DAA3, 0x1DAA4, 0x1DAA5, 0x1DAA6, 0x1DAA7, 0x1DAA8, 0x1DAA9, 0x1DAAA, 0x1DAAB, 0x1DAAC, 0x1DAAD, 0x1DAAE, 0x1DAAF, 0x1E000, 0x1E001, 0x1E002, 0x1E003, 0x1E004, 0x1E005, 0x1E006, 0x1E008, 0x1E009, 0x1E00A, 0x1E00B, 0x1E00C, 0x1E00D, 0x1E00E, 0x1E00F, 0x1E010, 0x1E011, 0x1E012, 0x1E013, 0x1E014, 0x1E015, 0x1E016, 0x1E017, 0x1E018, 0x1E01B, 0x1E01C, 0x1E01D, 0x1E01E, 0x1E01F, 0x1E020, 0x1E021, 0x1E023, 0x1E024, 0x1E026, 0x1E027, 0x1E028, 0x1E029, 0x1E02A, 0x1E8D0, 0x1E8D1, 0x1E8D2, 0x1E8D3, 0x1E8D4, 0x1E8D5, 0x1E8D6, 0x1E944, 0x1E945, 0x1E946, 0x1E947, 0x1E948, 0x1E949, 0x1E94A, 0xE0100, 0xE0101, 0xE0102, 0xE0103, 0xE0104, 0xE0105, 0xE0106, 0xE0107, 0xE0108, 0xE0109, 0xE010A, 0xE010B, 0xE010C, 0xE010D, 0xE010E, 0xE010F, 0xE0110, 0xE0111, 0xE0112, 0xE0113, 0xE0114, 0xE0115, 0xE0116, 0xE0117, 0xE0118, 0xE0119, 0xE011A, 0xE011B, 0xE011C, 0xE011D, 0xE011E, 0xE011F, 0xE0120, 0xE0121, 0xE0122, 0xE0123, 0xE0124, 0xE0125, 0xE0126, 0xE0127, 0xE0128, 0xE0129, 0xE012A, 0xE012B, 0xE012C, 0xE012D, 0xE012E, 0xE012F, 0xE0130, 0xE0131, 0xE0132, 0xE0133, 0xE0134, 0xE0135, 0xE0136, 0xE0137, 0xE0138, 0xE0139, 0xE013A, 0xE013B, 0xE013C, 0xE013D, 0xE013E, 0xE013F, 0xE0140, 0xE0141, 0xE0142, 0xE0143, 0xE0144, 0xE0145, 0xE0146, 0xE0147, 0xE0148, 0xE0149, 0xE014A, 0xE014B, 0xE014C, 0xE014D, 0xE014E, 0xE014F, 0xE0150, 0xE0151, 0xE0152, 0xE0153, 0xE0154, 0xE0155, 0xE0156, 0xE0157, 0xE0158, 0xE0159, 0xE015A, 0xE015B, 0xE015C, 0xE015D, 0xE015E, 0xE015F, 0xE0160, 0xE0161, 0xE0162, 0xE0163, 0xE0164, 0xE0165, 0xE0166, 0xE0167, 0xE0168, 0xE0169, 0xE016A, 0xE016B, 0xE016C, 0xE016D, 0xE016E, 0xE016F, 0xE0170, 0xE0171, 0xE0172, 0xE0173, 0xE0174, 0xE0175, 0xE0176, 0xE0177, 0xE0178, 0xE0179, 0xE017A, 0xE017B, 0xE017C, 0xE017D, 0xE017E, 0xE017F, 0xE0180, 0xE0181, 0xE0182, 0xE0183, 0xE0184, 0xE0185, 0xE0186, 0xE0187, 0xE0188, 0xE0189, 0xE018A, 0xE018B, 0xE018C, 0xE018D, 0xE018E, 0xE018F, 0xE0190, 0xE0191, 0xE0192, 0xE0193, 0xE0194, 0xE0195, 0xE0196, 0xE0197, 0xE0198, 0xE0199, 0xE019A, 0xE019B, 0xE019C, 0xE019D, 0xE019E, 0xE019F, 0xE01A0, 0xE01A1, 0xE01A2, 0xE01A3, 0xE01A4, 0xE01A5, 0xE01A6, 0xE01A7, 0xE01A8, 0xE01A9, 0xE01AA, 0xE01AB, 0xE01AC, 0xE01AD, 0xE01AE, 0xE01AF, 0xE01B0, 0xE01B1, 0xE01B2, 0xE01B3, 0xE01B4, 0xE01B5, 0xE01B6, 0xE01B7, 0xE01B8, 0xE01B9, 0xE01BA, 0xE01BB, 0xE01BC, 0xE01BD, 0xE01BE, 0xE01BF, 0xE01C0, 0xE01C1, 0xE01C2, 0xE01C3, 0xE01C4, 0xE01C5, 0xE01C6, 0xE01C7, 0xE01C8, 0xE01C9, 0xE01CA, 0xE01CB, 0xE01CC, 0xE01CD, 0xE01CE, 0xE01CF, 0xE01D0, 0xE01D1, 0xE01D2, 0xE01D3, 0xE01D4, 0xE01D5, 0xE01D6, 0xE01D7, 0xE01D8, 0xE01D9, 0xE01DA, 0xE01DB, 0xE01DC, 0xE01DD, 0xE01DE, 0xE01DF, 0xE01E0, 0xE01E1, 0xE01E2, 0xE01E3, 0xE01E4, 0xE01E5, 0xE01E6, 0xE01E7, 0xE01E8, 0xE01E9, 0xE01EA, 0xE01EB, 0xE01EC, 0xE01ED, 0xE01EE, 0xE01EF, }; /// The length of the combining characters list. const size_t bc_history_combo_chars_len = sizeof(bc_history_combo_chars) / sizeof(bc_history_combo_chars[0]); #endif // BC_ENABLE_HISTORY && !BC_ENABLE_LINE_LIB /// The human-readable name of the main function in bc source code. const char bc_func_main[] = "(main)"; /// The human-readable name of the read function in bc source code. const char bc_func_read[] = "(read)"; #if BC_DEBUG_CODE /// A list of names of instructions for easy debugging output. const char* bc_inst_names[] = { #if BC_ENABLED "BC_INST_INC", "BC_INST_DEC", #endif // BC_ENABLED "BC_INST_NEG", "BC_INST_BOOL_NOT", #if BC_ENABLE_EXTRA_MATH "BC_INST_TRUNC", #endif // BC_ENABLE_EXTRA_MATH "BC_INST_POWER", "BC_INST_MULTIPLY", "BC_INST_DIVIDE", "BC_INST_MODULUS", "BC_INST_PLUS", "BC_INST_MINUS", #if BC_ENABLE_EXTRA_MATH "BC_INST_PLACES", "BC_INST_LSHIFT", "BC_INST_RSHIFT", #endif // BC_ENABLE_EXTRA_MATH "BC_INST_REL_EQ", "BC_INST_REL_LE", "BC_INST_REL_GE", "BC_INST_REL_NE", "BC_INST_REL_LT", "BC_INST_REL_GT", "BC_INST_BOOL_OR", "BC_INST_BOOL_AND", #if BC_ENABLED "BC_INST_ASSIGN_POWER", "BC_INST_ASSIGN_MULTIPLY", "BC_INST_ASSIGN_DIVIDE", "BC_INST_ASSIGN_MODULUS", "BC_INST_ASSIGN_PLUS", "BC_INST_ASSIGN_MINUS", #if BC_ENABLE_EXTRA_MATH "BC_INST_ASSIGN_PLACES", "BC_INST_ASSIGN_LSHIFT", "BC_INST_ASSIGN_RSHIFT", #endif // BC_ENABLE_EXTRA_MATH "BC_INST_ASSIGN", "BC_INST_ASSIGN_POWER_NO_VAL", "BC_INST_ASSIGN_MULTIPLY_NO_VAL", "BC_INST_ASSIGN_DIVIDE_NO_VAL", "BC_INST_ASSIGN_MODULUS_NO_VAL", "BC_INST_ASSIGN_PLUS_NO_VAL", "BC_INST_ASSIGN_MINUS_NO_VAL", #if BC_ENABLE_EXTRA_MATH "BC_INST_ASSIGN_PLACES_NO_VAL", "BC_INST_ASSIGN_LSHIFT_NO_VAL", "BC_INST_ASSIGN_RSHIFT_NO_VAL", #endif // BC_ENABLE_EXTRA_MATH #endif // BC_ENABLED "BC_INST_ASSIGN_NO_VAL", "BC_INST_NUM", "BC_INST_VAR", "BC_INST_ARRAY_ELEM", "BC_INST_ARRAY", "BC_INST_ZERO", "BC_INST_ONE", #if BC_ENABLED "BC_INST_LAST", #endif // BC_ENABLED "BC_INST_IBASE", "BC_INST_OBASE", "BC_INST_SCALE", #if BC_ENABLE_EXTRA_MATH "BC_INST_SEED", #endif // BC_ENABLE_EXTRA_MATH "BC_INST_LENGTH", "BC_INST_SCALE_FUNC", "BC_INST_SQRT", "BC_INST_ABS", "BC_INST_IS_NUMBER", "BC_INST_IS_STRING", #if BC_ENABLE_EXTRA_MATH "BC_INST_IRAND", #endif // BC_ENABLE_EXTRA_MATH "BC_INST_ASCIIFY", "BC_INST_READ", #if BC_ENABLE_EXTRA_MATH "BC_INST_RAND", #endif // BC_ENABLE_EXTRA_MATH "BC_INST_MAXIBASE", "BC_INST_MAXOBASE", "BC_INST_MAXSCALE", #if BC_ENABLE_EXTRA_MATH "BC_INST_MAXRAND", #endif // BC_ENABLE_EXTRA_MATH "BC_INST_PRINT", "BC_INST_PRINT_POP", "BC_INST_STR", #if BC_ENABLED "BC_INST_PRINT_STR", "BC_INST_JUMP", "BC_INST_JUMP_ZERO", "BC_INST_CALL", "BC_INST_RET", "BC_INST_RET0", "BC_INST_RET_VOID", "BC_INST_HALT", #endif // BC_ENABLED "BC_INST_POP", "BC_INST_SWAP", "BC_INST_MODEXP", "BC_INST_DIVMOD", "BC_INST_PRINT_STREAM", #if DC_ENABLED "BC_INST_POP_EXEC", "BC_INST_EXECUTE", "BC_INST_EXEC_COND", "BC_INST_PRINT_STACK", "BC_INST_CLEAR_STACK", "BC_INST_REG_STACK_LEN", "BC_INST_STACK_LEN", "BC_INST_DUPLICATE", "BC_INST_LOAD", "BC_INST_PUSH_VAR", "BC_INST_PUSH_TO_VAR", "BC_INST_QUIT", "BC_INST_NQUIT", "BC_INST_EXEC_STACK_LEN", #endif // DC_ENABLED "BC_INST_INVALID", }; #endif // BC_DEBUG_CODE /// A constant string for 0. const char bc_parse_zero[2] = "0"; /// A constant string for 1. const char bc_parse_one[2] = "1"; #if BC_ENABLED /// A list of keywords for bc. This needs to be updated if keywords change. const BcLexKeyword bc_lex_kws[] = { BC_LEX_KW_ENTRY("auto", 4, true), BC_LEX_KW_ENTRY("break", 5, true), BC_LEX_KW_ENTRY("continue", 8, false), BC_LEX_KW_ENTRY("define", 6, true), BC_LEX_KW_ENTRY("for", 3, true), BC_LEX_KW_ENTRY("if", 2, true), BC_LEX_KW_ENTRY("limits", 6, false), BC_LEX_KW_ENTRY("return", 6, true), BC_LEX_KW_ENTRY("while", 5, true), BC_LEX_KW_ENTRY("halt", 4, false), BC_LEX_KW_ENTRY("last", 4, false), BC_LEX_KW_ENTRY("ibase", 5, true), BC_LEX_KW_ENTRY("obase", 5, true), BC_LEX_KW_ENTRY("scale", 5, true), #if BC_ENABLE_EXTRA_MATH BC_LEX_KW_ENTRY("seed", 4, false), #endif // BC_ENABLE_EXTRA_MATH BC_LEX_KW_ENTRY("length", 6, true), BC_LEX_KW_ENTRY("print", 5, false), BC_LEX_KW_ENTRY("sqrt", 4, true), BC_LEX_KW_ENTRY("abs", 3, false), BC_LEX_KW_ENTRY("is_number", 9, false), BC_LEX_KW_ENTRY("is_string", 9, false), #if BC_ENABLE_EXTRA_MATH BC_LEX_KW_ENTRY("irand", 5, false), #endif // BC_ENABLE_EXTRA_MATH BC_LEX_KW_ENTRY("asciify", 7, false), BC_LEX_KW_ENTRY("modexp", 6, false), BC_LEX_KW_ENTRY("divmod", 6, false), BC_LEX_KW_ENTRY("quit", 4, true), BC_LEX_KW_ENTRY("read", 4, false), #if BC_ENABLE_EXTRA_MATH BC_LEX_KW_ENTRY("rand", 4, false), #endif // BC_ENABLE_EXTRA_MATH BC_LEX_KW_ENTRY("maxibase", 8, false), BC_LEX_KW_ENTRY("maxobase", 8, false), BC_LEX_KW_ENTRY("maxscale", 8, false), #if BC_ENABLE_EXTRA_MATH BC_LEX_KW_ENTRY("maxrand", 7, false), #endif // BC_ENABLE_EXTRA_MATH BC_LEX_KW_ENTRY("line_length", 11, false), BC_LEX_KW_ENTRY("global_stacks", 13, false), BC_LEX_KW_ENTRY("leading_zero", 12, false), BC_LEX_KW_ENTRY("stream", 6, false), BC_LEX_KW_ENTRY("else", 4, false), }; /// The length of the list of bc keywords. const size_t bc_lex_kws_len = sizeof(bc_lex_kws) / sizeof(BcLexKeyword); #if BC_C11 // This is here to ensure that BC_LEX_NKWS, which is needed for the // redefined_kws in BcVm, is correct. If it's correct under C11, it will be // correct under C99, and I did not know any other way of ensuring they remained // synchronized. _Static_assert(sizeof(bc_lex_kws) / sizeof(BcLexKeyword) == BC_LEX_NKWS, "BC_LEX_NKWS is wrong."); #endif // BC_C11 /// An array of booleans that correspond to token types. An entry is true if the /// token is valid in an expression, false otherwise. This will need to change /// if tokens change. const uint8_t bc_parse_exprs[] = { // Starts with BC_LEX_EOF. BC_PARSE_EXPR_ENTRY(false, false, true, true, true, true, true, true), // Starts with BC_LEX_OP_MULTIPLY if extra math is enabled, BC_LEX_OP_DIVIDE // otherwise. BC_PARSE_EXPR_ENTRY(true, true, true, true, true, true, true, true), // Starts with BC_LEX_OP_REL_EQ if extra math is enabled, BC_LEX_OP_REL_LT // otherwise. BC_PARSE_EXPR_ENTRY(true, true, true, true, true, true, true, true), #if BC_ENABLE_EXTRA_MATH // Starts with BC_LEX_OP_ASSIGN_POWER. BC_PARSE_EXPR_ENTRY(true, true, true, true, true, true, true, true), // Starts with BC_LEX_OP_ASSIGN_RSHIFT. BC_PARSE_EXPR_ENTRY(true, true, false, false, true, true, false, false), // Starts with BC_LEX_RBRACKET. BC_PARSE_EXPR_ENTRY(false, false, false, false, true, true, true, false), // Starts with BC_LEX_KW_BREAK. BC_PARSE_EXPR_ENTRY(false, false, false, false, false, false, false, false), // Starts with BC_LEX_KW_HALT. BC_PARSE_EXPR_ENTRY(false, true, true, true, true, true, true, false), // Starts with BC_LEX_KW_SQRT. BC_PARSE_EXPR_ENTRY(true, true, true, true, true, true, true, true), // Starts with BC_LEX_KW_QUIT. BC_PARSE_EXPR_ENTRY(false, true, true, true, true, true, true, true), // Starts with BC_LEX_KW_GLOBAL_STACKS. BC_PARSE_EXPR_ENTRY(true, true, false, false, 0, 0, 0, 0) #else // BC_ENABLE_EXTRA_MATH // Starts with BC_LEX_OP_ASSIGN_PLUS. BC_PARSE_EXPR_ENTRY(true, true, true, false, false, true, true, false), // Starts with BC_LEX_COMMA. BC_PARSE_EXPR_ENTRY(false, false, false, false, false, true, true, true), // Starts with BC_LEX_KW_AUTO. BC_PARSE_EXPR_ENTRY(false, false, false, false, false, false, false, false), // Starts with BC_LEX_KW_WHILE. BC_PARSE_EXPR_ENTRY(false, false, true, true, true, true, true, false), // Starts with BC_LEX_KW_SQRT. BC_PARSE_EXPR_ENTRY(true, true, true, true, true, true, true, false), // Starts with BC_LEX_KW_MAXIBASE. BC_PARSE_EXPR_ENTRY(true, true, true, true, true, true, true, false), // Starts with BC_LEX_KW_ELSE. BC_PARSE_EXPR_ENTRY(false, 0, 0, 0, 0, 0, 0, 0) #endif // BC_ENABLE_EXTRA_MATH }; /// An array of data for operators that correspond to token types. Note that a /// lower precedence *value* means a higher precedence. const uchar bc_parse_ops[] = { BC_PARSE_OP(0, false), BC_PARSE_OP(0, false), BC_PARSE_OP(1, false), BC_PARSE_OP(1, false), #if BC_ENABLE_EXTRA_MATH BC_PARSE_OP(2, false), #endif // BC_ENABLE_EXTRA_MATH BC_PARSE_OP(4, false), BC_PARSE_OP(5, true), BC_PARSE_OP(5, true), BC_PARSE_OP(5, true), BC_PARSE_OP(6, true), BC_PARSE_OP(6, true), #if BC_ENABLE_EXTRA_MATH BC_PARSE_OP(3, false), BC_PARSE_OP(7, true), BC_PARSE_OP(7, true), #endif // BC_ENABLE_EXTRA_MATH BC_PARSE_OP(9, true), BC_PARSE_OP(9, true), BC_PARSE_OP(9, true), BC_PARSE_OP(9, true), BC_PARSE_OP(9, true), BC_PARSE_OP(9, true), BC_PARSE_OP(11, true), BC_PARSE_OP(10, true), BC_PARSE_OP(8, false), BC_PARSE_OP(8, false), BC_PARSE_OP(8, false), BC_PARSE_OP(8, false), BC_PARSE_OP(8, false), BC_PARSE_OP(8, false), #if BC_ENABLE_EXTRA_MATH BC_PARSE_OP(8, false), BC_PARSE_OP(8, false), BC_PARSE_OP(8, false), #endif // BC_ENABLE_EXTRA_MATH BC_PARSE_OP(8, false), }; // These identify what tokens can come after expressions in certain cases. /// The valid next tokens for normal expressions. const BcParseNext bc_parse_next_expr = BC_PARSE_NEXT(4, BC_LEX_NLINE, BC_LEX_SCOLON, BC_LEX_RBRACE, BC_LEX_EOF); /// The valid next tokens for function argument expressions. const BcParseNext bc_parse_next_arg = BC_PARSE_NEXT(2, BC_LEX_RPAREN, BC_LEX_COMMA); /// The valid next tokens for expressions in print statements. const BcParseNext bc_parse_next_print = BC_PARSE_NEXT(4, BC_LEX_COMMA, BC_LEX_NLINE, BC_LEX_SCOLON, BC_LEX_EOF); /// The valid next tokens for if statement conditions or loop conditions. This /// is used in for loops for the update expression and for builtin function. /// /// The name is an artifact of history, and is related to @a BC_PARSE_REL (see /// include/parse.h). It refers to how POSIX only allows some operators as part /// of the conditional of for loops, while loops, and if statements. const BcParseNext bc_parse_next_rel = BC_PARSE_NEXT(1, BC_LEX_RPAREN); /// The valid next tokens for array element expressions. const BcParseNext bc_parse_next_elem = BC_PARSE_NEXT(1, BC_LEX_RBRACKET); /// The valid next tokens for for loop initialization expressions and condition /// expressions. const BcParseNext bc_parse_next_for = BC_PARSE_NEXT(1, BC_LEX_SCOLON); /// The valid next tokens for read expressions. const BcParseNext bc_parse_next_read = BC_PARSE_NEXT(2, BC_LEX_NLINE, BC_LEX_EOF); /// The valid next tokens for the arguments of a builtin function with multiple /// arguments. const BcParseNext bc_parse_next_builtin = BC_PARSE_NEXT(1, BC_LEX_COMMA); #endif // BC_ENABLED #if DC_ENABLED /// A list of instructions that need register arguments in dc. const uint8_t dc_lex_regs[] = { BC_LEX_OP_REL_EQ, BC_LEX_OP_REL_LE, BC_LEX_OP_REL_GE, BC_LEX_OP_REL_NE, BC_LEX_OP_REL_LT, BC_LEX_OP_REL_GT, BC_LEX_SCOLON, BC_LEX_COLON, BC_LEX_KW_ELSE, BC_LEX_LOAD, BC_LEX_LOAD_POP, BC_LEX_OP_ASSIGN, BC_LEX_STORE_PUSH, BC_LEX_REG_STACK_LEVEL, BC_LEX_ARRAY_LENGTH, }; /// The length of the list of register instructions. const size_t dc_lex_regs_len = sizeof(dc_lex_regs) / sizeof(uint8_t); /// A list corresponding to characters starting at double quote ("). If an entry /// is BC_LEX_INVALID, then that character needs extra lexing in dc. If it does /// not, the character can trivially be replaced by the entry. Positions are /// kept because it corresponds to the ASCII table. This may need to be changed /// if tokens change. const uchar dc_lex_tokens[] = { #if BC_ENABLE_EXTRA_MATH BC_LEX_KW_IRAND, #else // BC_ENABLE_EXTRA_MATH BC_LEX_INVALID, #endif // BC_ENABLE_EXTRA_MATH BC_LEX_INVALID, #if BC_ENABLE_EXTRA_MATH BC_LEX_OP_TRUNC, #else // BC_ENABLE_EXTRA_MATH BC_LEX_INVALID, #endif // BC_ENABLE_EXTRA_MATH BC_LEX_OP_MODULUS, BC_LEX_INVALID, #if BC_ENABLE_EXTRA_MATH BC_LEX_KW_RAND, #else // BC_ENABLE_EXTRA_MATH BC_LEX_INVALID, #endif // BC_ENABLE_EXTRA_MATH BC_LEX_LPAREN, BC_LEX_RPAREN, BC_LEX_OP_MULTIPLY, BC_LEX_OP_PLUS, BC_LEX_EXEC_STACK_LENGTH, BC_LEX_OP_MINUS, BC_LEX_INVALID, BC_LEX_OP_DIVIDE, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_COLON, BC_LEX_SCOLON, BC_LEX_OP_REL_GT, BC_LEX_OP_REL_EQ, BC_LEX_OP_REL_LT, BC_LEX_KW_READ, #if BC_ENABLE_EXTRA_MATH BC_LEX_OP_PLACES, #else // BC_ENABLE_EXTRA_MATH BC_LEX_INVALID, #endif // BC_ENABLE_EXTRA_MATH BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_EQ_NO_REG, #if BC_ENABLE_EXTRA_MATH BC_LEX_OP_LSHIFT, #else // BC_ENABLE_EXTRA_MATH BC_LEX_INVALID, #endif // BC_ENABLE_EXTRA_MATH BC_LEX_KW_IBASE, #if BC_ENABLE_EXTRA_MATH BC_LEX_KW_SEED, #else // BC_ENABLE_EXTRA_MATH BC_LEX_INVALID, #endif // BC_ENABLE_EXTRA_MATH BC_LEX_KW_SCALE, BC_LEX_LOAD_POP, BC_LEX_OP_BOOL_AND, BC_LEX_OP_BOOL_NOT, BC_LEX_KW_OBASE, BC_LEX_KW_STREAM, BC_LEX_NQUIT, BC_LEX_POP, BC_LEX_STORE_PUSH, BC_LEX_KW_MAXIBASE, BC_LEX_KW_MAXOBASE, BC_LEX_KW_MAXSCALE, #if BC_ENABLE_EXTRA_MATH BC_LEX_KW_MAXRAND, #else // BC_ENABLE_EXTRA_MATH BC_LEX_INVALID, #endif // BC_ENABLE_EXTRA_MATH BC_LEX_SCALE_FACTOR, BC_LEX_ARRAY_LENGTH, BC_LEX_KW_LENGTH, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_INVALID, BC_LEX_OP_POWER, BC_LEX_NEG, BC_LEX_INVALID, BC_LEX_KW_ASCIIFY, BC_LEX_KW_ABS, BC_LEX_CLEAR_STACK, BC_LEX_DUPLICATE, BC_LEX_KW_ELSE, BC_LEX_PRINT_STACK, BC_LEX_INVALID, #if BC_ENABLE_EXTRA_MATH BC_LEX_OP_RSHIFT, #else // BC_ENABLE_EXTRA_MATH BC_LEX_INVALID, #endif // BC_ENABLE_EXTRA_MATH BC_LEX_STORE_IBASE, #if BC_ENABLE_EXTRA_MATH BC_LEX_STORE_SEED, #else // BC_ENABLE_EXTRA_MATH BC_LEX_INVALID, #endif // BC_ENABLE_EXTRA_MATH BC_LEX_STORE_SCALE, BC_LEX_LOAD, BC_LEX_OP_BOOL_OR, BC_LEX_PRINT_POP, BC_LEX_STORE_OBASE, BC_LEX_KW_PRINT, BC_LEX_KW_QUIT, BC_LEX_SWAP, BC_LEX_OP_ASSIGN, BC_LEX_KW_IS_STRING, BC_LEX_KW_IS_NUMBER, BC_LEX_KW_SQRT, BC_LEX_INVALID, BC_LEX_EXECUTE, BC_LEX_REG_STACK_LEVEL, BC_LEX_STACK_LEVEL, BC_LEX_LBRACE, BC_LEX_KW_MODEXP, BC_LEX_RBRACE, BC_LEX_KW_DIVMOD, BC_LEX_INVALID }; /// A list of instructions that correspond to lex tokens. If an entry is /// @a BC_INST_INVALID, that lex token needs extra parsing in the dc parser. /// Otherwise, the token can trivially be replaced by the entry. This needs to /// be updated if the tokens change. const uchar dc_parse_insts[] = { BC_INST_INVALID, BC_INST_INVALID, #if BC_ENABLED BC_INST_INVALID, BC_INST_INVALID, #endif // BC_ENABLED BC_INST_INVALID, BC_INST_BOOL_NOT, #if BC_ENABLE_EXTRA_MATH BC_INST_TRUNC, #endif // BC_ENABLE_EXTRA_MATH BC_INST_POWER, BC_INST_MULTIPLY, BC_INST_DIVIDE, BC_INST_MODULUS, BC_INST_PLUS, BC_INST_MINUS, #if BC_ENABLE_EXTRA_MATH BC_INST_PLACES, BC_INST_LSHIFT, BC_INST_RSHIFT, #endif // BC_ENABLE_EXTRA_MATH BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_BOOL_OR, BC_INST_BOOL_AND, #if BC_ENABLED BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, #if BC_ENABLE_EXTRA_MATH BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, #endif // BC_ENABLE_EXTRA_MATH #endif // BC_ENABLED BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_REL_GT, BC_INST_REL_LT, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_REL_GE, BC_INST_INVALID, BC_INST_REL_LE, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, #if BC_ENABLED BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, #endif // BC_ENABLED BC_INST_IBASE, BC_INST_OBASE, BC_INST_SCALE, #if BC_ENABLE_EXTRA_MATH BC_INST_SEED, #endif // BC_ENABLE_EXTRA_MATH BC_INST_LENGTH, BC_INST_PRINT, BC_INST_SQRT, BC_INST_ABS, BC_INST_IS_NUMBER, BC_INST_IS_STRING, #if BC_ENABLE_EXTRA_MATH BC_INST_IRAND, #endif // BC_ENABLE_EXTRA_MATH BC_INST_ASCIIFY, BC_INST_MODEXP, BC_INST_DIVMOD, BC_INST_QUIT, BC_INST_INVALID, #if BC_ENABLE_EXTRA_MATH BC_INST_RAND, #endif // BC_ENABLE_EXTRA_MATH BC_INST_MAXIBASE, BC_INST_MAXOBASE, BC_INST_MAXSCALE, #if BC_ENABLE_EXTRA_MATH BC_INST_MAXRAND, #endif // BC_ENABLE_EXTRA_MATH BC_INST_LINE_LENGTH, #if BC_ENABLED BC_INST_INVALID, #endif // BC_ENABLED BC_INST_LEADING_ZERO, BC_INST_PRINT_STREAM, BC_INST_INVALID, BC_INST_EXTENDED_REGISTERS, BC_INST_REL_EQ, BC_INST_INVALID, BC_INST_EXECUTE, BC_INST_PRINT_STACK, BC_INST_CLEAR_STACK, BC_INST_INVALID, BC_INST_STACK_LEN, BC_INST_DUPLICATE, BC_INST_SWAP, BC_INST_POP, BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, #if BC_ENABLE_EXTRA_MATH BC_INST_INVALID, #endif // BC_ENABLE_EXTRA_MATH BC_INST_INVALID, BC_INST_INVALID, BC_INST_INVALID, BC_INST_PRINT_POP, BC_INST_NQUIT, BC_INST_EXEC_STACK_LEN, BC_INST_SCALE_FUNC, BC_INST_INVALID, }; #endif // DC_ENABLED #endif // !BC_ENABLE_LIBRARY #if BC_ENABLE_EXTRA_MATH /// A constant for the rand multiplier. const BcRandState bc_rand_multiplier = BC_RAND_MULTIPLIER; #endif // BC_ENABLE_EXTRA_MATH // clang-format off #if BC_LONG_BIT >= 64 /// A constant array for the max of a bigdig number as a BcDig array. const BcDig bc_num_bigdigMax[] = { 709551616U, 446744073U, 18U, }; /// A constant array for the max of 2 times a bigdig number as a BcDig array. const BcDig bc_num_bigdigMax2[] = { 768211456U, 374607431U, 938463463U, 282366920U, 340U, }; #else // BC_LONG_BIT >= 64 /// A constant array for the max of a bigdig number as a BcDig array. const BcDig bc_num_bigdigMax[] = { 7296U, 9496U, 42U, }; /// A constant array for the max of 2 times a bigdig number as a BcDig array. const BcDig bc_num_bigdigMax2[] = { 1616U, 955U, 737U, 6744U, 1844U, }; #endif // BC_LONG_BIT >= 64 // clang-format on /// The size of the bigdig max array. const size_t bc_num_bigdigMax_size = sizeof(bc_num_bigdigMax) / sizeof(BcDig); /// The size of the bigdig max times 2 array. const size_t bc_num_bigdigMax2_size = sizeof(bc_num_bigdigMax2) / sizeof(BcDig); /// A string of digits for easy conversion from characters to digits. const char bc_num_hex_digits[] = "0123456789ABCDEF"; // clang-format off /// An array for easy conversion from exponent to power of 10. const BcBigDig bc_num_pow10[BC_BASE_DIGS + 1] = { 1, 10, 100, 1000, 10000, #if BC_BASE_DIGS > 4 100000, 1000000, 10000000, 100000000, 1000000000, #endif // BC_BASE_DIGS > 4 }; // clang-format on #if !BC_ENABLE_LIBRARY /// An array of functions for binary operators corresponding to the order of /// the instructions for the operators. const BcNumBinaryOp bc_program_ops[] = { bc_num_pow, bc_num_mul, bc_num_div, bc_num_mod, bc_num_add, bc_num_sub, #if BC_ENABLE_EXTRA_MATH bc_num_places, bc_num_lshift, bc_num_rshift, #endif // BC_ENABLE_EXTRA_MATH }; /// An array of functions for binary operators allocation requests corresponding /// to the order of the instructions for the operators. const BcNumBinaryOpReq bc_program_opReqs[] = { bc_num_powReq, bc_num_mulReq, bc_num_divReq, bc_num_divReq, bc_num_addReq, bc_num_addReq, #if BC_ENABLE_EXTRA_MATH bc_num_placesReq, bc_num_placesReq, bc_num_placesReq, #endif // BC_ENABLE_EXTRA_MATH }; /// An array of unary operator functions corresponding to the order of the /// instructions. const BcProgramUnary bc_program_unarys[] = { bc_program_negate, bc_program_not, #if BC_ENABLE_EXTRA_MATH bc_program_trunc, #endif // BC_ENABLE_EXTRA_MATH }; /// A filename for when parsing expressions. const char bc_program_exprs_name[] = ""; /// A filename for when parsing stdin.. const char bc_program_stdin_name[] = ""; /// A ready message for SIGINT catching. const char bc_program_ready_msg[] = "ready for more input\n"; /// The length of the ready message. const size_t bc_program_ready_msg_len = sizeof(bc_program_ready_msg) - 1; /// A list of escape characters that a print statement should treat specially. const char bc_program_esc_chars[] = "ab\\efnqrt"; /// A list of characters corresponding to the escape characters above. const char bc_program_esc_seqs[] = "\a\b\\\\\f\n\"\r\t"; #endif // !BC_ENABLE_LIBRARY diff --git a/src/dc.c b/src/dc.c index 992efe262fd8..37419acd4bd4 100644 --- a/src/dc.c +++ b/src/dc.c @@ -1,64 +1,65 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * The main procedure of dc. * */ #if DC_ENABLED #include #include #include /** * The main function for dc. * @param argc The number of arguments. * @param argv The arguments. */ BcStatus -dc_main(int argc, char* argv[]) +dc_main(int argc, const char* argv[]) { // All of these just set dc-specific items in BcVm. vm->read_ret = BC_INST_POP_EXEC; vm->help = dc_help; vm->sigmsg = dc_sig_msg; vm->siglen = dc_sig_msg_len; vm->next = dc_lex_token; vm->parse = dc_parse_parse; vm->expr = dc_parse_expr; return bc_vm_boot(argc, argv); } + #endif // DC_ENABLED diff --git a/src/main.c b/src/dc_fuzzer.c similarity index 58% copy from src/main.c copy to src/dc_fuzzer.c index a6d50614af57..adaf486a668c 100644 --- a/src/main.c +++ b/src/dc_fuzzer.c @@ -1,119 +1,112 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * - * The entry point for bc. + * The entry point for libFuzzer when fuzzing dc. * */ -#include -#include -#include - -#if BC_ENABLE_NLS -#include -#endif // BC_ENABLE_NLS - -#ifndef _WIN32 -#include -#endif // _WIN32 - #include +#include #include #include +#include #include #include #include -int -main(int argc, char* argv[]) -{ - BcStatus s; - char* name; - size_t len = strlen(BC_EXECPREFIX); +uint8_t* bc_fuzzer_data; -#if BC_ENABLE_NLS - // Must set the locale properly in order to have the right error messages. - vm->locale = setlocale(LC_ALL, ""); -#endif // BC_ENABLE_NLS +/// A boolean about whether we should use -c (false) or -C (true). +static bool dc_C; - // Set the start pledge(). - bc_pledge(bc_pledge_start, NULL); +int +LLVMFuzzerInitialize(int* argc, char*** argv) +{ + BC_UNUSED(argc); - // Sometimes, argv[0] can be NULL. Better make sure to be robust against it. - if (argv[0] != NULL) + if (argv == NULL || *argv == NULL) { - // Figure out the name of the calculator we are using. We can't use - // basename because it's not portable, but yes, this is stripping off - // the directory. - name = strrchr(argv[0], BC_FILE_SEP); - vm->name = (name == NULL) ? argv[0] : name + 1; + dc_C = false; } else { -#if !DC_ENABLED - vm->name = "bc"; -#elif !BC_ENABLED - vm->name = "dc"; -#else - // Just default to bc in that case. - vm->name = "bc"; -#endif + char* name; + + // Get the basename + name = strrchr((*argv)[0], BC_FILE_SEP); + name = name == NULL ? (*argv)[0] : name + 1; + + // Figure out which to use. + dc_C = (strcmp(name, "dc_fuzzer_C") == 0); } - // If the name is longer than the length of the prefix, skip the prefix. - if (strlen(vm->name) > len) vm->name += len; + return 0; +} + +int +LLVMFuzzerTestOneInput(const uint8_t* Data, size_t Size) +{ + BcStatus s; + + // I've already tested empty input, so just ignore. + if (Size == 0 || Data[0] == '\0') return 0; + + // Clear the global. This is to ensure a clean start. + memset(vm, 0, sizeof(BcVm)); + + // Make sure to set the name. + vm->name = "dc"; BC_SIG_LOCK; // We *must* do this here. Otherwise, other code could not jump out all of // the way. bc_vec_init(&vm->jmp_bufs, sizeof(sigjmp_buf), BC_DTOR_NONE); BC_SETJMP_LOCKED(vm, exit); -#if !DC_ENABLED - s = bc_main(argc, argv); -#elif !BC_ENABLED - s = dc_main(argc, argv); -#else - // BC_IS_BC uses vm->name, which was set above. So we're good. - if (BC_IS_BC) s = bc_main(argc, argv); - else s = dc_main(argc, argv); -#endif + // Create a string with the data. + bc_fuzzer_data = bc_vm_malloc(Size + 1); + memcpy(bc_fuzzer_data, Data, Size); + bc_fuzzer_data[Size] = '\0'; - vm->status = (int) s; + s = dc_main((int) (bc_fuzzer_args_len - 1), + dc_C ? dc_fuzzer_args_C : dc_fuzzer_args_c); exit: + BC_SIG_MAYLOCK; - return vm->status == BC_STATUS_QUIT ? BC_STATUS_SUCCESS : vm->status; + free(bc_fuzzer_data); + + return s == BC_STATUS_SUCCESS || s == BC_STATUS_QUIT ? 0 : -1; } diff --git a/src/dc_lex.c b/src/dc_lex.c index a58ca8f79cf3..d5131b45331d 100644 --- a/src/dc_lex.c +++ b/src/dc_lex.c @@ -1,302 +1,304 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * The lexer for dc. * */ #if DC_ENABLED #include #include #include bool dc_lex_negCommand(BcLex* l) { char c = l->buf[l->i]; return !BC_LEX_NUM_CHAR(c, false, false); } /** * Processes a dc command that needs a register. This is where the * extended-register extension is implemented. * @param l The lexer. */ static void dc_lex_register(BcLex* l) { // If extended register is enabled and the character is whitespace... if (DC_X && isspace(l->buf[l->i - 1])) { char c; // Eat the whitespace. bc_lex_whitespace(l); c = l->buf[l->i]; // Check for a letter or underscore. if (BC_ERR(!isalpha(c) && c != '_')) { bc_lex_verr(l, BC_ERR_PARSE_CHAR, c); } // Parse a normal identifier. l->i += 1; bc_lex_name(l); } else { // I don't allow newlines because newlines are used for controlling when // execution happens, and allowing newlines would just be complex. if (BC_ERR(l->buf[l->i - 1] == '\n')) { bc_lex_verr(l, BC_ERR_PARSE_CHAR, l->buf[l->i - 1]); } // Set the lexer string and token. bc_vec_popAll(&l->str); bc_vec_pushByte(&l->str, (uchar) l->buf[l->i - 1]); bc_vec_pushByte(&l->str, '\0'); l->t = BC_LEX_NAME; } } /** * Parses a dc string. Since dc's strings need to check for balanced brackets, * we can't just parse bc and dc strings with different start and end * characters. Oh, and dc strings need to check for escaped brackets. * @param l The lexer. */ static void dc_lex_string(BcLex* l) { size_t depth, nls, i; char c; bool got_more; // Set the token and clear the string. l->t = BC_LEX_STR; bc_vec_popAll(&l->str); do { depth = 1; nls = 0; got_more = false; +#if !BC_ENABLE_OSSFUZZ assert(l->mode != BC_MODE_STDIN || l->buf == vm->buffer.v); +#endif // !BC_ENABLE_OSSFUZZ // This is the meat. As long as we don't run into the NUL byte, and we // have "depth", which means we haven't completely balanced brackets // yet, we continue eating the string. for (i = l->i; (c = l->buf[i]) && depth; ++i) { // Check for escaped brackets and set the depths as appropriate. if (c == '\\') { c = l->buf[++i]; if (!c) break; } else { depth += (c == '['); depth -= (c == ']'); } // We want to adjust the line in the lexer as necessary. nls += (c == '\n'); if (depth) bc_vec_push(&l->str, &c); } if (BC_ERR(c == '\0' && depth)) { if (!vm->eof && l->mode != BC_MODE_FILE) { got_more = bc_lex_readLine(l); } if (got_more) { bc_vec_popAll(&l->str); } } } while (got_more && depth); // Obviously, if we didn't balance, that's an error. if (BC_ERR(c == '\0' && depth)) { l->i = i; bc_lex_err(l, BC_ERR_PARSE_STRING); } bc_vec_pushByte(&l->str, '\0'); l->i = i; l->line += nls; } /** * Lexes a dc token. This is the dc implementation of BcLexNext. * @param l The lexer. */ void dc_lex_token(BcLex* l) { char c = l->buf[l->i++], c2; size_t i; BC_SIG_ASSERT_LOCKED; // If the last token was a command that needs a register, we need to parse a // register, so do so. for (i = 0; i < dc_lex_regs_len; ++i) { // If the token is a register token, take care of it and return. if (l->last == dc_lex_regs[i]) { dc_lex_register(l); return; } } // These lines are for tokens that easily correspond to one character. We // just set the token. if (c >= '"' && c <= '~' && (l->t = dc_lex_tokens[(c - '"')]) != BC_LEX_INVALID) { return; } // This is the workhorse of the lexer when more complicated things are // needed. switch (c) { case '\0': case '\n': case '\t': case '\v': case '\f': case '\r': case ' ': { bc_lex_commonTokens(l, c); break; } // We don't have the ! command, so we always expect certain things // after the exclamation point. case '!': { c2 = l->buf[l->i]; if (c2 == '=') l->t = BC_LEX_OP_REL_NE; else if (c2 == '<') l->t = BC_LEX_OP_REL_LE; else if (c2 == '>') l->t = BC_LEX_OP_REL_GE; else bc_lex_invalidChar(l, c); l->i += 1; break; } case '#': { bc_lex_lineComment(l); break; } case '.': { c2 = l->buf[l->i]; // If the character after is a number, this dot is part of a number. // Otherwise, it's the BSD dot (equivalent to last). if (BC_NO_ERR(BC_LEX_NUM_CHAR(c2, true, false))) { bc_lex_number(l, c); } else bc_lex_invalidChar(l, c); break; } case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': { bc_lex_number(l, c); break; } case 'g': { c2 = l->buf[l->i]; if (c2 == 'l') l->t = BC_LEX_KW_LINE_LENGTH; else if (c2 == 'x') l->t = BC_LEX_EXTENDED_REGISTERS; else if (c2 == 'z') l->t = BC_LEX_KW_LEADING_ZERO; else bc_lex_invalidChar(l, c2); l->i += 1; break; } case '[': { dc_lex_string(l); break; } default: { bc_lex_invalidChar(l, c); } } } #endif // DC_ENABLED diff --git a/src/history.c b/src/history.c index 71afe62db879..6ae9785d9a79 100644 --- a/src/history.c +++ b/src/history.c @@ -1,2247 +1,2247 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * Adapted from the following: * * linenoise.c -- guerrilla line editing library against the idea that a * line editing lib needs to be 20,000 lines of C code. * * You can find the original source code at: * http://github.com/antirez/linenoise * * You can find the fork that this code is based on at: * https://github.com/rain-1/linenoise-mob * * ------------------------------------------------------------------------ * * This code is also under the following license: * * Copyright (c) 2010-2016, Salvatore Sanfilippo * Copyright (c) 2010-2013, Pieter Noordhuis * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are * met: * * * Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT * HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * * ------------------------------------------------------------------------ * * Does a number of crazy assumptions that happen to be true in 99.9999% of * the 2010 UNIX computers around. * * References: * - http://invisible-island.net/xterm/ctlseqs/ctlseqs.html * - http://www.3waylabs.com/nw/WWW/products/wizcon/vt220.html * * Todo list: * - Filter bogus Ctrl+ combinations. * - Win32 support * * Bloat: * - History search like Ctrl+r in readline? * * List of escape sequences used by this program, we do everything just * with three sequences. In order to be so cheap we may have some * flickering effect with some slow terminal, but the lesser sequences * the more compatible. * * EL (Erase Line) * Sequence: ESC [ n K * Effect: if n is 0 or missing, clear from cursor to end of line * Effect: if n is 1, clear from beginning of line to cursor * Effect: if n is 2, clear entire line * * CUF (CUrsor Forward) * Sequence: ESC [ n C * Effect: moves cursor forward n chars * * CUB (CUrsor Backward) * Sequence: ESC [ n D * Effect: moves cursor backward n chars * * The following is used to get the terminal width if getting * the width with the TIOCGWINSZ ioctl fails * * DSR (Device Status Report) * Sequence: ESC [ 6 n * Effect: reports the current cusor position as ESC [ n ; m R * where n is the row and m is the column * * When multi line mode is enabled, we also use two additional escape * sequences. However multi line editing is disabled by default. * * CUU (CUrsor Up) * Sequence: ESC [ n A * Effect: moves cursor up of n chars. * * CUD (CUrsor Down) * Sequence: ESC [ n B * Effect: moves cursor down of n chars. * * When bc_history_clearScreen() is called, two additional escape sequences * are used in order to clear the screen and position the cursor at home * position. * * CUP (CUrsor Position) * Sequence: ESC [ H * Effect: moves the cursor to upper left corner * * ED (Erase Display) * Sequence: ESC [ 2 J * Effect: clear the whole screen * * ***************************************************************************** * * Code for line history. * */ #if BC_ENABLE_HISTORY #if BC_ENABLE_EDITLINE #include #include #include #include #include sigjmp_buf bc_history_jmpbuf; volatile sig_atomic_t bc_history_inlinelib; static char* bc_history_prompt; static char bc_history_no_prompt[] = ""; static HistEvent bc_history_event; static bool bc_history_use_prompt; static char* bc_history_promptFunc(EditLine* el) { BC_UNUSED(el); return BC_PROMPT && bc_history_use_prompt ? bc_history_prompt : bc_history_no_prompt; } void bc_history_init(BcHistory* h) { BcVec v; char* home; home = getenv("HOME"); // This will hold the true path to the editrc. bc_vec_init(&v, 1, BC_DTOR_NONE); // Initialize the path to the editrc. This is done manually because the // libedit I used to test was failing with a NULL argument for the path, // which was supposed to automatically do $HOME/.editrc. But it was failing, // so I set it manually. if (home == NULL) { bc_vec_string(&v, bc_history_editrc_len - 1, bc_history_editrc + 1); } else { bc_vec_string(&v, strlen(home), home); bc_vec_concat(&v, bc_history_editrc); } h->hist = history_init(); if (BC_ERR(h->hist == NULL)) bc_vm_fatalError(BC_ERR_FATAL_ALLOC_ERR); h->el = el_init(vm->name, stdin, stdout, stderr); if (BC_ERR(h->el == NULL)) bc_vm_fatalError(BC_ERR_FATAL_ALLOC_ERR); // I want history and a prompt. history(h->hist, &bc_history_event, H_SETSIZE, 100); history(h->hist, &bc_history_event, H_SETUNIQUE, 1); el_set(h->el, EL_EDITOR, "emacs"); el_set(h->el, EL_HIST, history, h->hist); el_set(h->el, EL_PROMPT, bc_history_promptFunc); // I also want to get the user's .editrc. el_source(h->el, v.v); bc_vec_free(&v); h->badTerm = false; bc_history_prompt = NULL; } void bc_history_free(BcHistory* h) { if (BC_PROMPT && bc_history_prompt != NULL) free(bc_history_prompt); el_end(h->el); history_end(h->hist); } BcStatus bc_history_line(BcHistory* h, BcVec* vec, const char* prompt) { BcStatus s = BC_STATUS_SUCCESS; const char* line; int len; BC_SIG_LOCK; // If the jump happens here, then a SIGINT occurred. if (sigsetjmp(bc_history_jmpbuf, 0)) { bc_vec_string(vec, 1, "\n"); goto end; } // This is so the signal handler can handle line libraries properly. bc_history_inlinelib = 1; if (BC_PROMPT) { // Make sure to set the prompt. if (bc_history_prompt != NULL) { if (strcmp(bc_history_prompt, prompt)) { free(bc_history_prompt); bc_history_prompt = bc_vm_strdup(prompt); } } else bc_history_prompt = bc_vm_strdup(prompt); } bc_history_use_prompt = true; line = NULL; len = -1; errno = EINTR; // Get the line. - while (line == NULL && len == -1 && errno == EINTR) + while (line == NULL && (len == -1 || errno == EINTR)) { line = el_gets(h->el, &len); bc_history_use_prompt = false; } // If there is no line... if (BC_ERR(line == NULL)) { // If this is true, there was an error. Otherwise, it's just EOF. if (len == -1) { if (errno == ENOMEM) bc_err(BC_ERR_FATAL_ALLOC_ERR); bc_err(BC_ERR_FATAL_IO_ERR); } else { bc_file_printf(&vm->fout, "\n"); s = BC_STATUS_EOF; } } // If there is a line... else { bc_vec_string(vec, strlen(line), line); if (strcmp(line, "") && strcmp(line, "\n")) { history(h->hist, &bc_history_event, H_ENTER, line); } s = BC_STATUS_SUCCESS; } end: bc_history_inlinelib = 0; BC_SIG_UNLOCK; return s; } #else // BC_ENABLE_EDITLINE #if BC_ENABLE_READLINE #include #include #include #include #include sigjmp_buf bc_history_jmpbuf; volatile sig_atomic_t bc_history_inlinelib; void bc_history_init(BcHistory* h) { h->line = NULL; h->badTerm = false; // I want no tab completion. rl_bind_key('\t', rl_insert); } void bc_history_free(BcHistory* h) { if (h->line != NULL) free(h->line); } BcStatus bc_history_line(BcHistory* h, BcVec* vec, const char* prompt) { BcStatus s = BC_STATUS_SUCCESS; size_t len; BC_SIG_LOCK; // If the jump happens here, then a SIGINT occurred. if (sigsetjmp(bc_history_jmpbuf, 0)) { bc_vec_string(vec, 1, "\n"); goto end; } // This is so the signal handler can handle line libraries properly. bc_history_inlinelib = 1; // Get rid of the last line. if (h->line != NULL) { free(h->line); h->line = NULL; } // Get the line. h->line = readline(BC_PROMPT ? prompt : ""); // If there was a line, add it to the history. Otherwise, just return an // empty line. Oh, and NULL actually means EOF. if (h->line != NULL && h->line[0]) { add_history(h->line); len = strlen(h->line); bc_vec_expand(vec, len + 2); bc_vec_string(vec, len, h->line); bc_vec_concat(vec, "\n"); } else if (h->line == NULL) { bc_file_printf(&vm->fout, "%s\n", "^D"); s = BC_STATUS_EOF; } else bc_vec_string(vec, 1, "\n"); end: bc_history_inlinelib = 0; BC_SIG_UNLOCK; return s; } #else // BC_ENABLE_READLINE #include #include #include #include #include #include #include #include #ifndef _WIN32 #include #include #include #include #include #endif // _WIN32 #include #include #include #include #include #include #if BC_DEBUG_CODE /// A file for outputting to when debugging. BcFile bc_history_debug_fp; /// A buffer for the above file. char* bc_history_debug_buf; #endif // BC_DEBUG_CODE /** * Checks if the code is a wide character. * @param cp The codepoint to check. * @return True if @a cp is a wide character, false otherwise. */ static bool bc_history_wchar(uint32_t cp) { size_t i; for (i = 0; i < bc_history_wchars_len; ++i) { // Ranges are listed in ascending order. Therefore, once the // whole range is higher than the codepoint we're testing, the // codepoint won't be found in any remaining range => bail early. if (bc_history_wchars[i][0] > cp) return false; // Test this range. if (bc_history_wchars[i][0] <= cp && cp <= bc_history_wchars[i][1]) { return true; } } return false; } /** * Checks if the code is a combining character. * @param cp The codepoint to check. * @return True if @a cp is a combining character, false otherwise. */ static bool bc_history_comboChar(uint32_t cp) { size_t i; for (i = 0; i < bc_history_combo_chars_len; ++i) { // Combining chars are listed in ascending order, so once we pass // the codepoint of interest, we know it's not a combining char. if (bc_history_combo_chars[i] > cp) return false; if (bc_history_combo_chars[i] == cp) return true; } return false; } /** * Gets the length of previous UTF8 character. * @param buf The buffer of characters. * @param pos The index into the buffer. */ static size_t bc_history_prevCharLen(const char* buf, size_t pos) { size_t end = pos; for (pos -= 1; pos < end && (buf[pos] & 0xC0) == 0x80; --pos) { continue; } return end - (pos >= end ? 0 : pos); } /** * Converts UTF-8 to a Unicode code point. * @param s The string. * @param len The length of the string. * @param cp An out parameter for the codepoint. * @return The number of bytes eaten by the codepoint. */ static size_t bc_history_codePoint(const char* s, size_t len, uint32_t* cp) { if (len) { uchar byte = (uchar) s[0]; // This is literally the UTF-8 decoding algorithm. Look that up if you // don't understand this. if ((byte & 0x80) == 0) { *cp = byte; return 1; } else if ((byte & 0xE0) == 0xC0) { if (len >= 2) { *cp = (((uint32_t) (s[0] & 0x1F)) << 6) | ((uint32_t) (s[1] & 0x3F)); return 2; } } else if ((byte & 0xF0) == 0xE0) { if (len >= 3) { *cp = (((uint32_t) (s[0] & 0x0F)) << 12) | (((uint32_t) (s[1] & 0x3F)) << 6) | ((uint32_t) (s[2] & 0x3F)); return 3; } } else if ((byte & 0xF8) == 0xF0) { if (len >= 4) { *cp = (((uint32_t) (s[0] & 0x07)) << 18) | (((uint32_t) (s[1] & 0x3F)) << 12) | (((uint32_t) (s[2] & 0x3F)) << 6) | ((uint32_t) (s[3] & 0x3F)); return 4; } } else { *cp = 0xFFFD; return 1; } } *cp = 0; return 1; } /** * Gets the length of next grapheme. * @param buf The buffer. * @param buf_len The length of the buffer. * @param pos The index into the buffer. * @param col_len An out parameter for the length of the grapheme on screen. * @return The number of bytes in the grapheme. */ static size_t bc_history_nextLen(const char* buf, size_t buf_len, size_t pos, size_t* col_len) { uint32_t cp; size_t beg = pos; size_t len = bc_history_codePoint(buf + pos, buf_len - pos, &cp); if (bc_history_comboChar(cp)) { BC_UNREACHABLE #if !BC_CLANG if (col_len != NULL) *col_len = 0; return 0; #endif // !BC_CLANG } // Store the width of the character on screen. if (col_len != NULL) *col_len = bc_history_wchar(cp) ? 2 : 1; pos += len; // Find the first non-combining character. while (pos < buf_len) { len = bc_history_codePoint(buf + pos, buf_len - pos, &cp); if (!bc_history_comboChar(cp)) return pos - beg; pos += len; } return pos - beg; } /** * Gets the length of previous grapheme. * @param buf The buffer. * @param pos The index into the buffer. * @return The number of bytes in the grapheme. */ static size_t bc_history_prevLen(const char* buf, size_t pos) { size_t end = pos; // Find the first non-combining character. while (pos > 0) { uint32_t cp; size_t len = bc_history_prevCharLen(buf, pos); pos -= len; bc_history_codePoint(buf + pos, len, &cp); // The original linenoise-mob had an extra parameter col_len, like // bc_history_nextLen(), which, if not NULL, was set in this if // statement. However, we always passed NULL, so just skip that. if (!bc_history_comboChar(cp)) return end - pos; } BC_UNREACHABLE #if !BC_CLANG return 0; #endif // BC_CLANG } /** * Reads @a n characters from stdin. * @param buf The buffer to read into. The caller is responsible for making * sure this is big enough for @a n. * @param n The number of characters to read. * @return The number of characters read or less than 0 on error. */ static ssize_t bc_history_read(char* buf, size_t n) { ssize_t ret; BC_SIG_ASSERT_LOCKED; #ifndef _WIN32 do { // We don't care about being interrupted. ret = read(STDIN_FILENO, buf, n); } while (ret == EINTR); #else // _WIN32 bool good; DWORD read; HANDLE hn = GetStdHandle(STD_INPUT_HANDLE); good = ReadConsole(hn, buf, (DWORD) n, &read, NULL); ret = (read != n || !good) ? -1 : 1; #endif // _WIN32 return ret; } /** * Reads a Unicode code point into a buffer. * @param buf The buffer to read into. * @param buf_len The length of the buffer. * @param cp An out parameter for the codepoint. * @param nread An out parameter for the number of bytes read. * @return BC_STATUS_EOF or BC_STATUS_SUCCESS. */ static BcStatus bc_history_readCode(char* buf, size_t buf_len, uint32_t* cp, size_t* nread) { ssize_t n; uchar byte; assert(buf_len >= 1); BC_SIG_LOCK; // Read a byte. n = bc_history_read(buf, 1); BC_SIG_UNLOCK; if (BC_ERR(n <= 0)) goto err; // Get the byte. byte = ((uchar*) buf)[0]; // Once again, this is the UTF-8 decoding algorithm, but it has reads // instead of actual decoding. if ((byte & 0x80) != 0) { if ((byte & 0xE0) == 0xC0) { assert(buf_len >= 2); BC_SIG_LOCK; n = bc_history_read(buf + 1, 1); BC_SIG_UNLOCK; if (BC_ERR(n <= 0)) goto err; } else if ((byte & 0xF0) == 0xE0) { assert(buf_len >= 3); BC_SIG_LOCK; n = bc_history_read(buf + 1, 2); BC_SIG_UNLOCK; if (BC_ERR(n <= 0)) goto err; } else if ((byte & 0xF8) == 0xF0) { assert(buf_len >= 3); BC_SIG_LOCK; n = bc_history_read(buf + 1, 3); BC_SIG_UNLOCK; if (BC_ERR(n <= 0)) goto err; } else { n = -1; goto err; } } // Convert to the codepoint. *nread = bc_history_codePoint(buf, buf_len, cp); return BC_STATUS_SUCCESS; err: // If we get here, we either had a fatal error of EOF. if (BC_ERR(n < 0)) bc_vm_fatalError(BC_ERR_FATAL_IO_ERR); else *nread = (size_t) n; return BC_STATUS_EOF; } /** * Gets the column length from beginning of buffer to current byte position. * @param buf The buffer. * @param buf_len The length of the buffer. * @param pos The index into the buffer. * @return The number of columns between the beginning of @a buffer to * @a pos. */ static size_t bc_history_colPos(const char* buf, size_t buf_len, size_t pos) { size_t ret = 0, off = 0; // While we haven't reached the offset, get the length of the next grapheme. while (off < pos && off < buf_len) { size_t col_len, len; len = bc_history_nextLen(buf, buf_len, off, &col_len); off += len; ret += col_len; } return ret; } /** * Returns true if the terminal name is in the list of terminals we know are * not able to understand basic escape sequences. * @return True if the terminal is a bad terminal. */ static inline bool bc_history_isBadTerm(void) { size_t i; bool ret = false; char* term = bc_vm_getenv("TERM"); if (term == NULL) return false; for (i = 0; !ret && bc_history_bad_terms[i]; ++i) { ret = (!strcasecmp(term, bc_history_bad_terms[i])); } bc_vm_getenvFree(term); return ret; } /** * Enables raw mode (1960's black magic). * @param h The history data. */ static void bc_history_enableRaw(BcHistory* h) { // I don't do anything for Windows because in Windows, you set their // equivalent of raw mode and leave it, so I do it in bc_history_init(). #ifndef _WIN32 struct termios raw; int err; assert(BC_TTYIN); if (h->rawMode) return; BC_SIG_LOCK; if (BC_ERR(tcgetattr(STDIN_FILENO, &h->orig_termios) == -1)) { bc_vm_fatalError(BC_ERR_FATAL_IO_ERR); } BC_SIG_UNLOCK; // Modify the original mode. raw = h->orig_termios; // Input modes: no break, no CR to NL, no parity check, no strip char, // no start/stop output control. raw.c_iflag &= (unsigned int) (~(BRKINT | ICRNL | INPCK | ISTRIP | IXON)); // Control modes: set 8 bit chars. raw.c_cflag |= (CS8); // Local modes - choing off, canonical off, no extended functions, // no signal chars (^Z,^C). raw.c_lflag &= (unsigned int) (~(ECHO | ICANON | IEXTEN | ISIG)); // Control chars - set return condition: min number of bytes and timer. // We want read to give every single byte, w/o timeout (1 byte, no timer). raw.c_cc[VMIN] = 1; raw.c_cc[VTIME] = 0; BC_SIG_LOCK; // Put terminal in raw mode after flushing. do { err = tcsetattr(STDIN_FILENO, TCSAFLUSH, &raw); } while (BC_ERR(err < 0) && errno == EINTR); BC_SIG_UNLOCK; if (BC_ERR(err < 0)) bc_vm_fatalError(BC_ERR_FATAL_IO_ERR); #endif // _WIN32 h->rawMode = true; } /** * Disables raw mode. * @param h The history data. */ static void bc_history_disableRaw(BcHistory* h) { sig_atomic_t lock; if (!h->rawMode) return; BC_SIG_TRYLOCK(lock); #ifndef _WIN32 if (BC_ERR(tcsetattr(STDIN_FILENO, TCSAFLUSH, &h->orig_termios) != -1)) { h->rawMode = false; } #endif // _WIN32 BC_SIG_TRYUNLOCK(lock); } /** * Uses the ESC [6n escape sequence to query the horizontal cursor position * and return it. On error -1 is returned, on success the position of the * cursor. * @return The horizontal cursor position. */ static size_t bc_history_cursorPos(void) { char buf[BC_HIST_SEQ_SIZE]; char* ptr; char* ptr2; size_t cols, rows, i; BC_SIG_ASSERT_LOCKED; // Report cursor location. bc_file_write(&vm->fout, bc_flush_none, "\x1b[6n", 4); bc_file_flush(&vm->fout, bc_flush_none); // Read the response: ESC [ rows ; cols R. for (i = 0; i < sizeof(buf) - 1; ++i) { if (bc_history_read(buf + i, 1) != 1 || buf[i] == 'R') break; } buf[i] = '\0'; // This is basically an error; we didn't get what we were expecting. if (BC_ERR(buf[0] != BC_ACTION_ESC || buf[1] != '[')) return SIZE_MAX; // Parse the rows. ptr = buf + 2; rows = strtoul(ptr, &ptr2, 10); // Here we also didn't get what we were expecting. if (BC_ERR(!rows || ptr2[0] != ';')) return SIZE_MAX; // Parse the columns. ptr = ptr2 + 1; cols = strtoul(ptr, NULL, 10); if (BC_ERR(!cols)) return SIZE_MAX; return cols <= UINT16_MAX ? cols : 0; } /** * Tries to get the number of columns in the current terminal, or assume 80 * if it fails. * @return The number of columns in the terminal. */ static size_t bc_history_columns(void) { #ifndef _WIN32 struct winsize ws; int ret; ret = ioctl(vm->fout.fd, TIOCGWINSZ, &ws); if (BC_ERR(ret == -1 || !ws.ws_col)) { // Calling ioctl() failed. Try to query the terminal itself. size_t start, cols; // Get the initial position so we can restore it later. start = bc_history_cursorPos(); if (BC_ERR(start == SIZE_MAX)) return BC_HIST_DEF_COLS; // Go to right margin and get position. bc_file_write(&vm->fout, bc_flush_none, "\x1b[999C", 6); bc_file_flush(&vm->fout, bc_flush_none); cols = bc_history_cursorPos(); if (BC_ERR(cols == SIZE_MAX)) return BC_HIST_DEF_COLS; // Restore position. if (cols > start) { bc_file_printf(&vm->fout, "\x1b[%zuD", cols - start); bc_file_flush(&vm->fout, bc_flush_none); } return cols; } return ws.ws_col; #else // _WIN32 CONSOLE_SCREEN_BUFFER_INFO csbi; if (!GetConsoleScreenBufferInfo(GetStdHandle(STD_OUTPUT_HANDLE), &csbi)) { return 80; } return ((size_t) (csbi.srWindow.Right)) - csbi.srWindow.Left + 1; #endif // _WIN32 } /** * Gets the column length of prompt text. This is probably unnecessary because * the prompts that I use are ASCII, but I kept it just in case. * @param prompt The prompt. * @param plen The length of the prompt. * @return The column length of the prompt. */ static size_t bc_history_promptColLen(const char* prompt, size_t plen) { char buf[BC_HIST_MAX_LINE + 1]; size_t buf_len = 0, off = 0; // The original linenoise-mob checked for ANSI escapes here on the prompt. I // know the prompts do not have ANSI escapes. I deleted the code. while (off < plen) { buf[buf_len++] = prompt[off++]; } return bc_history_colPos(buf, buf_len, buf_len); } /** * Rewrites the currently edited line accordingly to the buffer content, * cursor position, and number of columns of the terminal. * @param h The history data. */ static void bc_history_refresh(BcHistory* h) { char* buf = h->buf.v; size_t colpos, len = BC_HIST_BUF_LEN(h), pos = h->pos, extras_len = 0; BC_SIG_ASSERT_LOCKED; bc_file_flush(&vm->fout, bc_flush_none); // Get to the prompt column position from the left. while (h->pcol + bc_history_colPos(buf, len, pos) >= h->cols) { size_t chlen = bc_history_nextLen(buf, len, 0, NULL); buf += chlen; len -= chlen; pos -= chlen; } // Get to the prompt column position from the right. while (h->pcol + bc_history_colPos(buf, len, len) > h->cols) { len -= bc_history_prevLen(buf, len); } // Cursor to left edge. bc_file_write(&vm->fout, bc_flush_none, "\r", 1); // Take the extra stuff into account. This is where history makes sure to // preserve stuff that was printed without a newline. if (h->extras.len > 1) { extras_len = h->extras.len - 1; bc_vec_grow(&h->buf, extras_len); len += extras_len; pos += extras_len; bc_file_write(&vm->fout, bc_flush_none, h->extras.v, extras_len); } // Write the prompt, if desired. if (BC_PROMPT) bc_file_write(&vm->fout, bc_flush_none, h->prompt, h->plen); bc_file_write(&vm->fout, bc_flush_none, h->buf.v, len - extras_len); // Erase to right. bc_file_write(&vm->fout, bc_flush_none, "\x1b[0K", 4); // We need to be sure to grow this. if (pos >= h->buf.len - extras_len) bc_vec_grow(&h->buf, pos + extras_len); // Move cursor to original position. Do NOT move the putchar of '\r' to the // printf with colpos. That causes a bug where the cursor will go to the end // of the line when there is no prompt. bc_file_putchar(&vm->fout, bc_flush_none, '\r'); colpos = bc_history_colPos(h->buf.v, len - extras_len, pos) + h->pcol; // Set the cursor position again. if (colpos) bc_file_printf(&vm->fout, "\x1b[%zuC", colpos); bc_file_flush(&vm->fout, bc_flush_none); } /** * Inserts the character(s) 'c' at cursor current position. * @param h The history data. * @param cbuf The character buffer to copy from. * @param clen The number of characters to copy. */ static void bc_history_edit_insert(BcHistory* h, const char* cbuf, size_t clen) { BC_SIG_ASSERT_LOCKED; bc_vec_grow(&h->buf, clen); // If we are at the end of the line... if (h->pos == BC_HIST_BUF_LEN(h)) { size_t colpos = 0, len; // Copy into the buffer. memcpy(bc_vec_item(&h->buf, h->pos), cbuf, clen); // Adjust the buffer. h->pos += clen; h->buf.len += clen - 1; bc_vec_pushByte(&h->buf, '\0'); // Set the length and column position. len = BC_HIST_BUF_LEN(h) + h->extras.len - 1; colpos = bc_history_promptColLen(h->prompt, h->plen); colpos += bc_history_colPos(h->buf.v, len, len); // Do we have the trivial case? if (colpos < h->cols) { // Avoid a full update of the line in the trivial case. bc_file_write(&vm->fout, bc_flush_none, cbuf, clen); bc_file_flush(&vm->fout, bc_flush_none); } else bc_history_refresh(h); } else { // Amount that we need to move. size_t amt = BC_HIST_BUF_LEN(h) - h->pos; // Move the stuff. memmove(h->buf.v + h->pos + clen, h->buf.v + h->pos, amt); memcpy(h->buf.v + h->pos, cbuf, clen); // Adjust the buffer. h->pos += clen; h->buf.len += clen; h->buf.v[BC_HIST_BUF_LEN(h)] = '\0'; bc_history_refresh(h); } } /** * Moves the cursor to the left. * @param h The history data. */ static void bc_history_edit_left(BcHistory* h) { BC_SIG_ASSERT_LOCKED; // Stop at the left end. if (h->pos <= 0) return; h->pos -= bc_history_prevLen(h->buf.v, h->pos); bc_history_refresh(h); } /** * Moves the cursor to the right. * @param h The history data. */ static void bc_history_edit_right(BcHistory* h) { BC_SIG_ASSERT_LOCKED; // Stop at the right end. if (h->pos == BC_HIST_BUF_LEN(h)) return; h->pos += bc_history_nextLen(h->buf.v, BC_HIST_BUF_LEN(h), h->pos, NULL); bc_history_refresh(h); } /** * Moves the cursor to the end of the current word. * @param h The history data. */ static void bc_history_edit_wordEnd(BcHistory* h) { size_t len = BC_HIST_BUF_LEN(h); BC_SIG_ASSERT_LOCKED; // Don't overflow. if (!len || h->pos >= len) return; // Find the word, then find the end of it. while (h->pos < len && isspace(h->buf.v[h->pos])) { h->pos += 1; } while (h->pos < len && !isspace(h->buf.v[h->pos])) { h->pos += 1; } bc_history_refresh(h); } /** * Moves the cursor to the start of the current word. * @param h The history data. */ static void bc_history_edit_wordStart(BcHistory* h) { size_t len = BC_HIST_BUF_LEN(h); BC_SIG_ASSERT_LOCKED; // Stop with no data. if (!len) return; // Find the word, the find the beginning of the word. while (h->pos > 0 && isspace(h->buf.v[h->pos - 1])) { h->pos -= 1; } while (h->pos > 0 && !isspace(h->buf.v[h->pos - 1])) { h->pos -= 1; } bc_history_refresh(h); } /** * Moves the cursor to the start of the line. * @param h The history data. */ static void bc_history_edit_home(BcHistory* h) { BC_SIG_ASSERT_LOCKED; // Stop at the beginning. if (!h->pos) return; h->pos = 0; bc_history_refresh(h); } /** * Moves the cursor to the end of the line. * @param h The history data. */ static void bc_history_edit_end(BcHistory* h) { BC_SIG_ASSERT_LOCKED; // Stop at the end of the line. if (h->pos == BC_HIST_BUF_LEN(h)) return; h->pos = BC_HIST_BUF_LEN(h); bc_history_refresh(h); } /** * Substitutes the currently edited line with the next or previous history * entry as specified by 'dir' (direction). * @param h The history data. * @param dir The direction to substitute; true means previous, false next. */ static void bc_history_edit_next(BcHistory* h, bool dir) { const char* dup; const char* str; BC_SIG_ASSERT_LOCKED; // Stop if there is no history. if (h->history.len <= 1) return; // Duplicate the buffer. if (h->buf.v[0]) dup = bc_vm_strdup(h->buf.v); else dup = ""; // Update the current history entry before overwriting it with the next one. bc_vec_replaceAt(&h->history, h->history.len - 1 - h->idx, &dup); // Show the new entry. h->idx += (dir == BC_HIST_PREV ? 1 : SIZE_MAX); // Se the index appropriately at the ends. if (h->idx == SIZE_MAX) { h->idx = 0; return; } else if (h->idx >= h->history.len) { h->idx = h->history.len - 1; return; } // Get the string. str = *((char**) bc_vec_item(&h->history, h->history.len - 1 - h->idx)); bc_vec_string(&h->buf, strlen(str), str); assert(h->buf.len > 0); // Set the position at the end. h->pos = BC_HIST_BUF_LEN(h); bc_history_refresh(h); } /** * Deletes the character at the right of the cursor without altering the cursor * position. Basically, this is what happens with the "Delete" keyboard key. * @param h The history data. */ static void bc_history_edit_delete(BcHistory* h) { size_t chlen, len = BC_HIST_BUF_LEN(h); BC_SIG_ASSERT_LOCKED; // If there is no character, skip. if (!len || h->pos >= len) return; // Get the length of the character. chlen = bc_history_nextLen(h->buf.v, len, h->pos, NULL); // Move characters after it into its place. memmove(h->buf.v + h->pos, h->buf.v + h->pos + chlen, len - h->pos - chlen); // Make the buffer valid again. h->buf.len -= chlen; h->buf.v[BC_HIST_BUF_LEN(h)] = '\0'; bc_history_refresh(h); } /** * Deletes the character to the left of the cursor and moves the cursor back one * space. Basically, this is what happens with the "Backspace" keyboard key. * @param h The history data. */ static void bc_history_edit_backspace(BcHistory* h) { size_t chlen, len = BC_HIST_BUF_LEN(h); BC_SIG_ASSERT_LOCKED; // If there are no characters, skip. if (!h->pos || !len) return; // Get the length of the previous character. chlen = bc_history_prevLen(h->buf.v, h->pos); // Move everything back one. memmove(h->buf.v + h->pos - chlen, h->buf.v + h->pos, len - h->pos); // Make the buffer valid again. h->pos -= chlen; h->buf.len -= chlen; h->buf.v[BC_HIST_BUF_LEN(h)] = '\0'; bc_history_refresh(h); } /** * Deletes the previous word, maintaining the cursor at the start of the * current word. * @param h The history data. */ static void bc_history_edit_deletePrevWord(BcHistory* h) { size_t diff, old_pos = h->pos; BC_SIG_ASSERT_LOCKED; // If at the beginning of the line, skip. if (!old_pos) return; // Find the word, then the beginning of the word. while (h->pos > 0 && isspace(h->buf.v[h->pos - 1])) { h->pos -= 1; } while (h->pos > 0 && !isspace(h->buf.v[h->pos - 1])) { h->pos -= 1; } // Get the difference in position. diff = old_pos - h->pos; // Move the data back. memmove(h->buf.v + h->pos, h->buf.v + old_pos, BC_HIST_BUF_LEN(h) - old_pos + 1); // Make the buffer valid again. h->buf.len -= diff; bc_history_refresh(h); } /** * Deletes the next word, maintaining the cursor at the same position. * @param h The history data. */ static void bc_history_edit_deleteNextWord(BcHistory* h) { size_t next_end = h->pos, len = BC_HIST_BUF_LEN(h); BC_SIG_ASSERT_LOCKED; // If at the end of the line, skip. if (next_end == len) return; // Find the word, then the end of the word. while (next_end < len && isspace(h->buf.v[next_end])) { next_end += 1; } while (next_end < len && !isspace(h->buf.v[next_end])) { next_end += 1; } // Move the stuff into position. memmove(h->buf.v + h->pos, h->buf.v + next_end, len - next_end); // Make the buffer valid again. h->buf.len -= next_end - h->pos; bc_history_refresh(h); } /** * Swaps two characters, the one under the cursor and the one to the left. * @param h The history data. */ static void bc_history_swap(BcHistory* h) { size_t pcl, ncl; char auxb[5]; BC_SIG_ASSERT_LOCKED; // If there are no characters, skip. if (!h->pos) return; // Get the length of the previous and next characters. pcl = bc_history_prevLen(h->buf.v, h->pos); ncl = bc_history_nextLen(h->buf.v, BC_HIST_BUF_LEN(h), h->pos, NULL); // To perform a swap we need: // * Nonzero char length to the left. // * To not be at the end of the line. if (pcl && h->pos != BC_HIST_BUF_LEN(h) && pcl < 5 && ncl < 5) { // Swap. memcpy(auxb, h->buf.v + h->pos - pcl, pcl); memcpy(h->buf.v + h->pos - pcl, h->buf.v + h->pos, ncl); memcpy(h->buf.v + h->pos - pcl + ncl, auxb, pcl); // Reset the position. h->pos += ((~pcl) + 1) + ncl; bc_history_refresh(h); } } /** * Raises the specified signal. This is a convenience function. * @param h The history data. * @param sig The signal to raise. */ static void bc_history_raise(BcHistory* h, int sig) { // We really don't want to be in raw mode when longjmp()'s are flying. bc_history_disableRaw(h); raise(sig); } /** * Handles escape sequences. This function will make sense if you know VT100 * escape codes; otherwise, it will be confusing. * @param h The history data. */ static void bc_history_escape(BcHistory* h) { char c, seq[3]; BC_SIG_ASSERT_LOCKED; // Read a character into seq. if (BC_ERR(BC_HIST_READ(seq, 1))) return; c = seq[0]; // ESC ? sequences. if (c != '[' && c != 'O') { if (c == 'f') bc_history_edit_wordEnd(h); else if (c == 'b') bc_history_edit_wordStart(h); else if (c == 'd') bc_history_edit_deleteNextWord(h); } else { // Read a character into seq. if (BC_ERR(BC_HIST_READ(seq + 1, 1))) { bc_vm_fatalError(BC_ERR_FATAL_IO_ERR); } // ESC [ sequences. if (c == '[') { c = seq[1]; if (c >= '0' && c <= '9') { // Extended escape, read additional byte. if (BC_ERR(BC_HIST_READ(seq + 2, 1))) { bc_vm_fatalError(BC_ERR_FATAL_IO_ERR); } if (seq[2] == '~') { switch (c) { case '1': { bc_history_edit_home(h); break; } case '3': { bc_history_edit_delete(h); break; } case '4': { bc_history_edit_end(h); break; } default: { break; } } } else if (seq[2] == ';') { // Read two characters into seq. if (BC_ERR(BC_HIST_READ(seq, 2))) { bc_vm_fatalError(BC_ERR_FATAL_IO_ERR); } if (seq[0] != '5') return; else if (seq[1] == 'C') bc_history_edit_wordEnd(h); else if (seq[1] == 'D') bc_history_edit_wordStart(h); } } else { switch (c) { // Up. case 'A': { bc_history_edit_next(h, BC_HIST_PREV); break; } // Down. case 'B': { bc_history_edit_next(h, BC_HIST_NEXT); break; } // Right. case 'C': { bc_history_edit_right(h); break; } // Left. case 'D': { bc_history_edit_left(h); break; } // Home. case 'H': case '1': { bc_history_edit_home(h); break; } // End. case 'F': case '4': { bc_history_edit_end(h); break; } case 'd': { bc_history_edit_deleteNextWord(h); break; } } } } // ESC O sequences. else { switch (seq[1]) { case 'A': { bc_history_edit_next(h, BC_HIST_PREV); break; } case 'B': { bc_history_edit_next(h, BC_HIST_NEXT); break; } case 'C': { bc_history_edit_right(h); break; } case 'D': { bc_history_edit_left(h); break; } case 'F': { bc_history_edit_end(h); break; } case 'H': { bc_history_edit_home(h); break; } } } } } /** * Adds a line to the history. * @param h The history data. * @param line The line to add. */ static void bc_history_add(BcHistory* h, char* line) { BC_SIG_ASSERT_LOCKED; // If there is something already there... if (h->history.len) { // Get the previous. char* s = *((char**) bc_vec_item_rev(&h->history, 0)); // Check for, and discard, duplicates. if (!strcmp(s, line)) { free(line); return; } } bc_vec_push(&h->history, &line); } /** * Adds an empty line to the history. This is separate from bc_history_add() * because we don't want it allocating. * @param h The history data. */ static void bc_history_add_empty(BcHistory* h) { const char* line = ""; BC_SIG_ASSERT_LOCKED; // If there is something already there... if (h->history.len) { // Get the previous. char* s = *((char**) bc_vec_item_rev(&h->history, 0)); // Check for, and discard, duplicates. if (!s[0]) return; } bc_vec_push(&h->history, &line); } /** * Resets the history state to nothing. * @param h The history data. */ static void bc_history_reset(BcHistory* h) { BC_SIG_ASSERT_LOCKED; h->oldcolpos = h->pos = h->idx = 0; h->cols = bc_history_columns(); // The latest history entry is always our current buffer, that // initially is just an empty string. bc_history_add_empty(h); // Buffer starts empty. bc_vec_empty(&h->buf); } /** * Prints a control character. * @param h The history data. * @param c The control character to print. */ static void bc_history_printCtrl(BcHistory* h, unsigned int c) { char str[3] = { '^', 'A', '\0' }; const char newline[2] = { '\n', '\0' }; BC_SIG_ASSERT_LOCKED; // Set the correct character. str[1] = (char) (c + 'A' - BC_ACTION_CTRL_A); // Concatenate the string. bc_vec_concat(&h->buf, str); h->pos = BC_HIST_BUF_LEN(h); bc_history_refresh(h); // Pop the string. bc_vec_npop(&h->buf, sizeof(str)); bc_vec_pushByte(&h->buf, '\0'); h->pos = 0; if (c != BC_ACTION_CTRL_C && c != BC_ACTION_CTRL_D) { // We sometimes want to print a newline; for the times we don't; it's // because newlines are taken care of elsewhere. bc_file_write(&vm->fout, bc_flush_none, newline, sizeof(newline) - 1); bc_history_refresh(h); } } /** * Edits a line of history. This function is the core of the line editing * capability of bc history. It expects 'fd' to be already in "raw mode" so that * every key pressed will be returned ASAP to read(). * @param h The history data. * @param prompt The prompt. * @return BC_STATUS_SUCCESS or BC_STATUS_EOF. */ static BcStatus bc_history_edit(BcHistory* h, const char* prompt) { BC_SIG_LOCK; bc_history_reset(h); // Don't write the saved output the first time. This is because it has // already been written to output. In other words, don't uncomment the // line below or add anything like it. // bc_file_write(&vm->fout, bc_flush_none, h->extras.v, h->extras.len - 1); // Write the prompt if desired. if (BC_PROMPT) { h->prompt = prompt; h->plen = strlen(prompt); h->pcol = bc_history_promptColLen(prompt, h->plen); bc_file_write(&vm->fout, bc_flush_none, prompt, h->plen); bc_file_flush(&vm->fout, bc_flush_none); } // This is the input loop. for (;;) { BcStatus s; char cbuf[32]; unsigned int c = 0; size_t nread = 0; BC_SIG_UNLOCK; // Read a code. s = bc_history_readCode(cbuf, sizeof(cbuf), &c, &nread); if (BC_ERR(s)) return s; BC_SIG_LOCK; switch (c) { case BC_ACTION_LINE_FEED: case BC_ACTION_ENTER: { // Return the line. bc_vec_pop(&h->history); BC_SIG_UNLOCK; return s; } case BC_ACTION_TAB: { // My tab handling is dumb; it just prints 8 spaces every time. memcpy(cbuf, bc_history_tab, bc_history_tab_len + 1); bc_history_edit_insert(h, cbuf, bc_history_tab_len); break; } case BC_ACTION_CTRL_C: { bc_history_printCtrl(h, c); // Quit if the user wants it. if (!BC_SIGINT) { vm->status = BC_STATUS_QUIT; BC_SIG_UNLOCK; BC_JMP; } // Print the ready message. bc_file_write(&vm->fout, bc_flush_none, vm->sigmsg, vm->siglen); bc_file_write(&vm->fout, bc_flush_none, bc_program_ready_msg, bc_program_ready_msg_len); bc_history_reset(h); bc_history_refresh(h); break; } case BC_ACTION_BACKSPACE: case BC_ACTION_CTRL_H: { bc_history_edit_backspace(h); break; } // Act as end-of-file or delete-forward-char. case BC_ACTION_CTRL_D: { // Act as EOF if there's no chacters, otherwise emulate Emacs // delete next character to match historical gnu bc behavior. if (BC_HIST_BUF_LEN(h) == 0) { bc_history_printCtrl(h, c); BC_SIG_UNLOCK; return BC_STATUS_EOF; } bc_history_edit_delete(h); break; } // Swaps current character with previous. case BC_ACTION_CTRL_T: { bc_history_swap(h); break; } case BC_ACTION_CTRL_B: { bc_history_edit_left(h); break; } case BC_ACTION_CTRL_F: { bc_history_edit_right(h); break; } case BC_ACTION_CTRL_P: { bc_history_edit_next(h, BC_HIST_PREV); break; } case BC_ACTION_CTRL_N: { bc_history_edit_next(h, BC_HIST_NEXT); break; } case BC_ACTION_ESC: { bc_history_escape(h); break; } // Delete the whole line. case BC_ACTION_CTRL_U: { bc_vec_string(&h->buf, 0, ""); h->pos = 0; bc_history_refresh(h); break; } // Delete from current to end of line. case BC_ACTION_CTRL_K: { bc_vec_npop(&h->buf, h->buf.len - h->pos); bc_vec_pushByte(&h->buf, '\0'); bc_history_refresh(h); break; } // Go to the start of the line. case BC_ACTION_CTRL_A: { bc_history_edit_home(h); break; } // Go to the end of the line. case BC_ACTION_CTRL_E: { bc_history_edit_end(h); break; } // Clear screen. case BC_ACTION_CTRL_L: { bc_file_write(&vm->fout, bc_flush_none, "\x1b[H\x1b[2J", 7); bc_history_refresh(h); break; } // Delete previous word. case BC_ACTION_CTRL_W: { bc_history_edit_deletePrevWord(h); break; } default: { // If we have a control character, print it and raise signals as // needed. if ((c >= BC_ACTION_CTRL_A && c <= BC_ACTION_CTRL_Z) || c == BC_ACTION_CTRL_BSLASH) { bc_history_printCtrl(h, c); #ifndef _WIN32 if (c == BC_ACTION_CTRL_Z) bc_history_raise(h, SIGTSTP); if (c == BC_ACTION_CTRL_S) bc_history_raise(h, SIGSTOP); if (c == BC_ACTION_CTRL_BSLASH) { bc_history_raise(h, SIGQUIT); } #else // _WIN32 vm->status = BC_STATUS_QUIT; BC_SIG_UNLOCK; BC_JMP; #endif // _WIN32 } // Otherwise, just insert. else bc_history_edit_insert(h, cbuf, nread); break; } } } BC_SIG_UNLOCK; return BC_STATUS_SUCCESS; } /** * Returns true if stdin has more data. This is for multi-line pasting, and it * does not work on Windows. * @param h The history data. */ static inline bool bc_history_stdinHasData(BcHistory* h) { #ifndef _WIN32 int n; return pselect(1, &h->rdset, NULL, NULL, &h->ts, &h->sigmask) > 0 || (ioctl(STDIN_FILENO, FIONREAD, &n) >= 0 && n > 0); #else // _WIN32 return false; #endif // _WIN32 } BcStatus bc_history_line(BcHistory* h, BcVec* vec, const char* prompt) { BcStatus s; char* line; assert(vm->fout.len == 0); bc_history_enableRaw(h); do { // Do the edit. s = bc_history_edit(h, prompt); // Print a newline and flush. bc_file_write(&vm->fout, bc_flush_none, "\n", 1); bc_file_flush(&vm->fout, bc_flush_none); BC_SIG_LOCK; // If we actually have data... if (h->buf.v[0]) { // Duplicate it. line = bc_vm_strdup(h->buf.v); // Store it. bc_history_add(h, line); } // Add an empty string. else bc_history_add_empty(h); BC_SIG_UNLOCK; // Concatenate the line to the return vector. bc_vec_concat(vec, h->buf.v); bc_vec_concat(vec, "\n"); } while (!s && bc_history_stdinHasData(h)); assert(!s || s == BC_STATUS_EOF); bc_history_disableRaw(h); return s; } void bc_history_string_free(void* str) { char* s = *((char**) str); BC_SIG_ASSERT_LOCKED; if (s[0]) free(s); } void bc_history_init(BcHistory* h) { #ifdef _WIN32 HANDLE out, in; #endif // _WIN32 BC_SIG_ASSERT_LOCKED; h->rawMode = false; h->badTerm = bc_history_isBadTerm(); // Just don't initialize with a bad terminal. if (h->badTerm) return; #ifdef _WIN32 h->orig_in = 0; h->orig_out = 0; in = GetStdHandle(STD_INPUT_HANDLE); out = GetStdHandle(STD_OUTPUT_HANDLE); // Set the code pages. SetConsoleCP(CP_UTF8); SetConsoleOutputCP(CP_UTF8); // Get the original modes. if (!GetConsoleMode(in, &h->orig_in) || !GetConsoleMode(out, &h->orig_out)) { // Just mark it as a bad terminal on error. h->badTerm = true; return; } else { // Set the new modes. DWORD reqOut = h->orig_out | ENABLE_VIRTUAL_TERMINAL_PROCESSING; DWORD reqIn = h->orig_in | ENABLE_VIRTUAL_TERMINAL_INPUT; // The input handle requires turning *off* some modes. That's why // history didn't work before; I didn't read the documentation // closely enough to see that most modes were automaticall enabled, // and they need to be turned off. reqOut |= DISABLE_NEWLINE_AUTO_RETURN | ENABLE_PROCESSED_OUTPUT; reqIn &= ~(ENABLE_LINE_INPUT | ENABLE_ECHO_INPUT); reqIn &= ~(ENABLE_PROCESSED_INPUT); // Set the modes; if there was an error, assume a bad terminal and // quit. if (!SetConsoleMode(in, reqIn) || !SetConsoleMode(out, reqOut)) { h->badTerm = true; return; } } #endif // _WIN32 bc_vec_init(&h->buf, sizeof(char), BC_DTOR_NONE); bc_vec_init(&h->history, sizeof(char*), BC_DTOR_HISTORY_STRING); bc_vec_init(&h->extras, sizeof(char), BC_DTOR_NONE); #ifndef _WIN32 FD_ZERO(&h->rdset); FD_SET(STDIN_FILENO, &h->rdset); h->ts.tv_sec = 0; h->ts.tv_nsec = 0; sigemptyset(&h->sigmask); sigaddset(&h->sigmask, SIGINT); #endif // _WIN32 } void bc_history_free(BcHistory* h) { BC_SIG_ASSERT_LOCKED; #ifndef _WIN32 bc_history_disableRaw(h); #else // _WIN32 SetConsoleMode(GetStdHandle(STD_INPUT_HANDLE), h->orig_in); SetConsoleMode(GetStdHandle(STD_OUTPUT_HANDLE), h->orig_out); #endif // _WIN32 #if BC_DEBUG bc_vec_free(&h->buf); bc_vec_free(&h->history); bc_vec_free(&h->extras); #endif // BC_DEBUG } #if BC_DEBUG_CODE /** * Prints scan codes. This special mode is used by bc history in order to print * scan codes on screen for debugging / development purposes. * @param h The history data. */ void bc_history_printKeyCodes(BcHistory* h) { char quit[4]; bc_vm_printf("Linenoise key codes debugging mode.\n" "Press keys to see scan codes. " "Type 'quit' at any time to exit.\n"); bc_history_enableRaw(h); memset(quit, ' ', 4); while (true) { char c; ssize_t nread; nread = bc_history_read(&c, 1); if (nread <= 0) continue; // Shift string to left. memmove(quit, quit + 1, sizeof(quit) - 1); // Insert current char on the right. quit[sizeof(quit) - 1] = c; if (!memcmp(quit, "quit", sizeof(quit))) break; bc_vm_printf("'%c' %lu (type quit to exit)\n", isprint(c) ? c : '?', (unsigned long) c); // Go left edge manually, we are in raw mode. bc_vm_putchar('\r', bc_flush_none); bc_file_flush(&vm->fout, bc_flush_none); } bc_history_disableRaw(h); } #endif // BC_DEBUG_CODE #endif // BC_ENABLE_HISTORY #endif // BC_ENABLE_READLINE #endif // BC_ENABLE_EDITLINE diff --git a/src/lang.c b/src/lang.c index bb147fc60d0b..7968bcbd9dfd 100644 --- a/src/lang.c +++ b/src/lang.c @@ -1,353 +1,353 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * Code to manipulate data structures in programs. * */ #include #include #include #include #include #include void bc_const_free(void* constant) { BcConst* c = constant; BC_SIG_ASSERT_LOCKED; assert(c->val != NULL); bc_num_free(&c->num); } #if BC_ENABLED void bc_func_insert(BcFunc* f, BcProgram* p, char* name, BcType type, size_t line) { BcAuto a; size_t i, idx; // The function must *always* be valid. assert(f != NULL); // Get the index of the variable. idx = bc_program_search(p, name, type == BC_TYPE_VAR); // Search through all of the other autos/parameters. for (i = 0; i < f->autos.len; ++i) { // Get the auto. BcAuto* aptr = bc_vec_item(&f->autos, i); // If they match, barf. if (BC_ERR(idx == aptr->idx && type == aptr->type)) { const char* array = type == BC_TYPE_ARRAY ? "[]" : ""; bc_error(BC_ERR_PARSE_DUP_LOCAL, line, name, array); } } // Set the auto. a.idx = idx; a.type = type; // Push it. bc_vec_push(&f->autos, &a); } #endif // BC_ENABLED void bc_func_init(BcFunc* f, const char* name) { BC_SIG_ASSERT_LOCKED; assert(f != NULL && name != NULL); bc_vec_init(&f->code, sizeof(uchar), BC_DTOR_NONE); #if BC_ENABLED // Only bc needs these things. if (BC_IS_BC) { bc_vec_init(&f->autos, sizeof(BcAuto), BC_DTOR_NONE); bc_vec_init(&f->labels, sizeof(size_t), BC_DTOR_NONE); f->nparams = 0; f->voidfn = false; } #endif // BC_ENABLED f->name = name; } void bc_func_reset(BcFunc* f) { BC_SIG_ASSERT_LOCKED; assert(f != NULL); bc_vec_popAll(&f->code); #if BC_ENABLED if (BC_IS_BC) { bc_vec_popAll(&f->autos); bc_vec_popAll(&f->labels); f->nparams = 0; f->voidfn = false; } #endif // BC_ENABLED } -#if BC_DEBUG +#if BC_DEBUG || BC_ENABLE_MEMCHECK void bc_func_free(void* func) { BcFunc* f = (BcFunc*) func; BC_SIG_ASSERT_LOCKED; assert(f != NULL); bc_vec_free(&f->code); #if BC_ENABLED if (BC_IS_BC) { bc_vec_free(&f->autos); bc_vec_free(&f->labels); } #endif // BC_ENABLED } -#endif // BC_DEBUG +#endif // BC_DEBUG || BC_ENABLE_MEMCHECK void bc_array_init(BcVec* a, bool nums) { BC_SIG_ASSERT_LOCKED; // Set the proper vector. if (nums) bc_vec_init(a, sizeof(BcNum), BC_DTOR_NUM); else bc_vec_init(a, sizeof(BcVec), BC_DTOR_VEC); // We always want at least one item in the array. bc_array_expand(a, 1); } void bc_array_copy(BcVec* d, const BcVec* s) { size_t i; BC_SIG_ASSERT_LOCKED; assert(d != NULL && s != NULL); assert(d != s && d->size == s->size && d->dtor == s->dtor); // Make sure to destroy everything currently in d. This will put a lot of // temps on the reuse list, so allocating later is not going to be as // expensive as it seems. Also, it makes it easier to copy numbers that are // strings. bc_vec_popAll(d); // Preexpand. bc_vec_expand(d, s->cap); d->len = s->len; for (i = 0; i < s->len; ++i) { BcNum* dnum; BcNum* snum; dnum = bc_vec_item(d, i); snum = bc_vec_item(s, i); // We have to create a copy of the number as well. if (BC_PROG_STR(snum)) { // NOLINTNEXTLINE memcpy(dnum, snum, sizeof(BcNum)); } else bc_num_createCopy(dnum, snum); } } void bc_array_expand(BcVec* a, size_t len) { assert(a != NULL); BC_SIG_ASSERT_LOCKED; bc_vec_expand(a, len); // If this is true, then we have a num array. if (a->size == sizeof(BcNum) && a->dtor == BC_DTOR_NUM) { // Initialize numbers until we reach the target. while (len > a->len) { BcNum* n = bc_vec_pushEmpty(a); bc_num_init(n, BC_NUM_DEF_SIZE); } } else { assert(a->size == sizeof(BcVec) && a->dtor == BC_DTOR_VEC); // Recursively initialize arrays until we reach the target. Having the // second argument of bc_array_init() be true will activate the base // case, so we're safe. while (len > a->len) { BcVec* v = bc_vec_pushEmpty(a); bc_array_init(v, true); } } } void bc_result_clear(BcResult* r) { r->t = BC_RESULT_TEMP; bc_num_clear(&r->d.n); } #if DC_ENABLED void bc_result_copy(BcResult* d, BcResult* src) { assert(d != NULL && src != NULL); BC_SIG_ASSERT_LOCKED; // d is assumed to not be valid yet. d->t = src->t; // Yes, it depends on what type. switch (d->t) { case BC_RESULT_TEMP: case BC_RESULT_IBASE: case BC_RESULT_SCALE: case BC_RESULT_OBASE: #if BC_ENABLE_EXTRA_MATH case BC_RESULT_SEED: #endif // BC_ENABLE_EXTRA_MATH { bc_num_createCopy(&d->d.n, &src->d.n); break; } case BC_RESULT_VAR: case BC_RESULT_ARRAY: case BC_RESULT_ARRAY_ELEM: { // NOLINTNEXTLINE memcpy(&d->d.loc, &src->d.loc, sizeof(BcLoc)); break; } case BC_RESULT_STR: { // NOLINTNEXTLINE memcpy(&d->d.n, &src->d.n, sizeof(BcNum)); break; } case BC_RESULT_ZERO: case BC_RESULT_ONE: { // Do nothing. break; } #if BC_ENABLED case BC_RESULT_VOID: case BC_RESULT_LAST: { #if BC_DEBUG // We should *never* try copying either of these. abort(); #endif // BC_DEBUG } #endif // BC_ENABLED } } #endif // DC_ENABLED void bc_result_free(void* result) { BcResult* r = (BcResult*) result; BC_SIG_ASSERT_LOCKED; assert(r != NULL); switch (r->t) { case BC_RESULT_TEMP: case BC_RESULT_IBASE: case BC_RESULT_SCALE: case BC_RESULT_OBASE: #if BC_ENABLE_EXTRA_MATH case BC_RESULT_SEED: #endif // BC_ENABLE_EXTRA_MATH { bc_num_free(&r->d.n); break; } case BC_RESULT_VAR: case BC_RESULT_ARRAY: case BC_RESULT_ARRAY_ELEM: case BC_RESULT_STR: case BC_RESULT_ZERO: case BC_RESULT_ONE: #if BC_ENABLED case BC_RESULT_VOID: case BC_RESULT_LAST: #endif // BC_ENABLED { // Do nothing. break; } } } diff --git a/src/lex.c b/src/lex.c index d01e327e2939..37e52c33fffd 100644 --- a/src/lex.c +++ b/src/lex.c @@ -1,403 +1,411 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * Common code for the lexers. * */ #include #include #include #include #include #include #include void bc_lex_invalidChar(BcLex* l, char c) { l->t = BC_LEX_INVALID; bc_lex_verr(l, BC_ERR_PARSE_CHAR, c); } void bc_lex_lineComment(BcLex* l) { l->t = BC_LEX_WHITESPACE; while (l->i < l->len && l->buf[l->i] != '\n') { l->i += 1; } } void bc_lex_comment(BcLex* l) { size_t i, nlines = 0; const char* buf; bool end = false, got_more; char c; l->i += 1; l->t = BC_LEX_WHITESPACE; // This loop is complex because it might need to request more data from // stdin if the comment is not ended. This loop is taken until the comment // is finished or we have EOF. do { buf = l->buf; got_more = false; // If we are in stdin mode, the buffer must be the one used for stdin. +#if !BC_ENABLE_OSSFUZZ assert(vm->mode != BC_MODE_STDIN || buf == vm->buffer.v); +#endif // !BC_ENABLE_OSSFUZZ // Find the end of the comment. for (i = l->i; !end; i += !end) { // While we don't have an asterisk, eat, but increment nlines. for (; (c = buf[i]) && c != '*'; ++i) { nlines += (c == '\n'); } // If this is true, we need to request more data. if (BC_ERR(!c || buf[i + 1] == '\0')) { +#if !BC_ENABLE_OSSFUZZ // Read more, if possible. if (!vm->eof && l->mode != BC_MODE_FILE) { got_more = bc_lex_readLine(l); } +#endif // !BC_ENABLE_OSSFUZZ break; } // If this turns true, we found the end. Yay! end = (buf[i + 1] == '/'); } } while (got_more && !end); // If we didn't find the end, barf. if (!end) { l->i = i; bc_lex_err(l, BC_ERR_PARSE_COMMENT); } l->i = i + 2; l->line += nlines; } void bc_lex_whitespace(BcLex* l) { char c; l->t = BC_LEX_WHITESPACE; // Eat. We don't eat newlines because they can be special. for (c = l->buf[l->i]; c != '\n' && isspace(c); c = l->buf[++l->i]) { continue; } } void bc_lex_commonTokens(BcLex* l, char c) { if (!c) l->t = BC_LEX_EOF; else if (c == '\n') l->t = BC_LEX_NLINE; else bc_lex_whitespace(l); } /** * Parses a number. * @param l The lexer. * @param start The start character. * @param int_only Whether this function should only look for an integer. This * is used to implement the exponent of scientific notation. */ static size_t bc_lex_num(BcLex* l, char start, bool int_only) { const char* buf = l->buf + l->i; size_t i; char c; bool last_pt, pt = (start == '.'); // This loop looks complex. It is not. It is asking if the character is not // a nul byte and it if it a valid num character based on what we have found // thus far, or whether it is a backslash followed by a newline. I can do // i+1 on the buffer because the buffer must have a nul byte. for (i = 0; (c = buf[i]) && (BC_LEX_NUM_CHAR(c, pt, int_only) || (c == '\\' && buf[i + 1] == '\n')); ++i) { // I don't need to test that the next character is a newline because // the loop condition above ensures that. if (c == '\\') { i += 2; // Make sure to eat whitespace at the beginning of the line. while (isspace(buf[i]) && buf[i] != '\n') { i += 1; } c = buf[i]; // If the next character is not a number character, bail. if (!BC_LEX_NUM_CHAR(c, pt, int_only)) break; } // Did we find the radix point? last_pt = (c == '.'); // If we did, and we already have one, then break because it's not part // of this number. if (pt && last_pt) break; // Set whether we have found a radix point. pt = pt || last_pt; bc_vec_push(&l->str, &c); } return i; } void bc_lex_number(BcLex* l, char start) { l->t = BC_LEX_NUMBER; // Make sure the string is clear. bc_vec_popAll(&l->str); bc_vec_push(&l->str, &start); // Parse the number. l->i += bc_lex_num(l, start, false); #if BC_ENABLE_EXTRA_MATH { char c = l->buf[l->i]; // Do we have a number in scientific notation? if (c == 'e') { #if BC_ENABLED // Barf for POSIX. if (BC_IS_POSIX) bc_lex_err(l, BC_ERR_POSIX_EXP_NUM); #endif // BC_ENABLED // Push the e. bc_vec_push(&l->str, &c); l->i += 1; c = l->buf[l->i]; // Check for negative specifically because bc_lex_num() does not. if (c == BC_LEX_NEG_CHAR) { bc_vec_push(&l->str, &c); l->i += 1; c = l->buf[l->i]; } // We must have a number character, so barf if not. if (BC_ERR(!BC_LEX_NUM_CHAR(c, false, true))) { bc_lex_verr(l, BC_ERR_PARSE_CHAR, c); } // Parse the exponent. l->i += bc_lex_num(l, 0, true); } } #endif // BC_ENABLE_EXTRA_MATH bc_vec_pushByte(&l->str, '\0'); } void bc_lex_name(BcLex* l) { size_t i = 0; const char* buf = l->buf + l->i - 1; char c = buf[i]; l->t = BC_LEX_NAME; // Should be obvious. It's looking for valid characters. while ((c >= 'a' && c <= 'z') || isdigit(c) || c == '_') { c = buf[++i]; } // Set the string to the identifier. bc_vec_string(&l->str, i, buf); // Increment the index. We minus 1 because it has already been incremented. l->i += i - 1; } void bc_lex_init(BcLex* l) { BC_SIG_ASSERT_LOCKED; assert(l != NULL); bc_vec_init(&l->str, sizeof(char), BC_DTOR_NONE); } void bc_lex_free(BcLex* l) { BC_SIG_ASSERT_LOCKED; assert(l != NULL); bc_vec_free(&l->str); } void bc_lex_file(BcLex* l, const char* file) { assert(l != NULL && file != NULL); l->line = 1; vm->file = file; } void bc_lex_next(BcLex* l) { BC_SIG_ASSERT_LOCKED; assert(l != NULL); l->last = l->t; // If this wasn't here, the line number would be off. l->line += (l->i != 0 && l->buf[l->i - 1] == '\n'); // If the last token was EOF, someone called this one too many times. if (BC_ERR(l->last == BC_LEX_EOF)) bc_lex_err(l, BC_ERR_PARSE_EOF); l->t = BC_LEX_EOF; // We are done if this is true. if (l->i == l->len) return; // Loop until failure or we don't have whitespace. This // is so the parser doesn't get inundated with whitespace. do { vm->next(l); } while (l->t == BC_LEX_WHITESPACE); } /** * Updates the buffer and len so that they are not invalidated when the stdin * buffer grows. * @param l The lexer. * @param text The text. * @param len The length of the text. */ static void bc_lex_fixText(BcLex* l, const char* text, size_t len) { l->buf = text; l->len = len; } bool bc_lex_readLine(BcLex* l) { bool good; // These are reversed because they should be already locked, but // bc_vm_readLine() needs them to be unlocked. BC_SIG_UNLOCK; // Make sure we read from the appropriate place. switch (l->mode) { case BC_MODE_EXPRS: { good = bc_vm_readBuf(false); break; } case BC_MODE_FILE: { good = false; break; } +#if !BC_ENABLE_OSSFUZZ + case BC_MODE_STDIN: { good = bc_vm_readLine(false); break; } +#endif // !BC_ENABLE_OSSFUZZ + #ifdef __GNUC__ #ifndef __clang__ default: { // We should never get here. abort(); } #endif // __clang__ #endif // __GNUC__ } BC_SIG_LOCK; bc_lex_fixText(l, vm->buffer.v, vm->buffer.len - 1); return good; } void bc_lex_text(BcLex* l, const char* text, BcMode mode) { BC_SIG_ASSERT_LOCKED; assert(l != NULL && text != NULL); bc_lex_fixText(l, text, strlen(text)); l->i = 0; l->t = l->last = BC_LEX_INVALID; l->mode = mode; bc_lex_next(l); } diff --git a/src/main.c b/src/main.c index a6d50614af57..e4a1f7399bb4 100644 --- a/src/main.c +++ b/src/main.c @@ -1,119 +1,128 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * The entry point for bc. * */ #include #include #include #if BC_ENABLE_NLS #include #endif // BC_ENABLE_NLS #ifndef _WIN32 #include #endif // _WIN32 #include #include #include #include #include #include int main(int argc, char* argv[]) { BcStatus s; char* name; size_t len = strlen(BC_EXECPREFIX); #if BC_ENABLE_NLS // Must set the locale properly in order to have the right error messages. vm->locale = setlocale(LC_ALL, ""); #endif // BC_ENABLE_NLS // Set the start pledge(). bc_pledge(bc_pledge_start, NULL); // Sometimes, argv[0] can be NULL. Better make sure to be robust against it. if (argv[0] != NULL) { // Figure out the name of the calculator we are using. We can't use // basename because it's not portable, but yes, this is stripping off // the directory. name = strrchr(argv[0], BC_FILE_SEP); vm->name = (name == NULL) ? argv[0] : name + 1; } else { #if !DC_ENABLED vm->name = "bc"; #elif !BC_ENABLED vm->name = "dc"; #else // Just default to bc in that case. vm->name = "bc"; #endif } // If the name is longer than the length of the prefix, skip the prefix. if (strlen(vm->name) > len) vm->name += len; BC_SIG_LOCK; // We *must* do this here. Otherwise, other code could not jump out all of // the way. bc_vec_init(&vm->jmp_bufs, sizeof(sigjmp_buf), BC_DTOR_NONE); BC_SETJMP_LOCKED(vm, exit); +#if BC_CLANG +#pragma clang diagnostic push +#pragma clang diagnostic ignored "-Wcast-qual" +#endif // BC_CLANG #if !DC_ENABLED - s = bc_main(argc, argv); + s = bc_main(argc, (const char**) argv); #elif !BC_ENABLED - s = dc_main(argc, argv); + s = dc_main(argc, (const char**) argv); #else // BC_IS_BC uses vm->name, which was set above. So we're good. - if (BC_IS_BC) s = bc_main(argc, argv); - else s = dc_main(argc, argv); + if (BC_IS_BC) s = bc_main(argc, (const char**) argv); + else s = dc_main(argc, (const char**) argv); #endif +#if BC_CLANG +#pragma clang diagnostic pop +#endif // BC_CLANG - vm->status = (int) s; + vm->status = (sig_atomic_t) s; exit: BC_SIG_MAYLOCK; - return vm->status == BC_STATUS_QUIT ? BC_STATUS_SUCCESS : vm->status; + s = bc_vm_atexit((BcStatus) vm->status); + + return (int) s; } diff --git a/src/num.c b/src/num.c index 5420183c1e1a..83f84edb91fc 100644 --- a/src/num.c +++ b/src/num.c @@ -1,4516 +1,4520 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * Code for the number type. * */ #include #include #include #include #include #include #include #include #include #include #if BC_ENABLE_LIBRARY #include #endif // BC_ENABLE_LIBRARY // Before you try to understand this code, see the development manual // (manuals/development.md#numbers). static void bc_num_m(BcNum* a, BcNum* b, BcNum* restrict c, size_t scale); /** * Multiply two numbers and throw a math error if they overflow. * @param a The first operand. * @param b The second operand. * @return The product of the two operands. */ static inline size_t bc_num_mulOverflow(size_t a, size_t b) { size_t res = a * b; if (BC_ERR(BC_VM_MUL_OVERFLOW(a, b, res))) bc_err(BC_ERR_MATH_OVERFLOW); return res; } /** * Conditionally negate @a n based on @a neg. Algorithm taken from * https://graphics.stanford.edu/~seander/bithacks.html#ConditionalNegate . * @param n The value to turn into a signed value and negate. * @param neg The condition to negate or not. */ static inline ssize_t bc_num_neg(size_t n, bool neg) { return (((ssize_t) n) ^ -((ssize_t) neg)) + neg; } /** * Compare a BcNum against zero. * @param n The number to compare. * @return -1 if the number is less than 0, 1 if greater, and 0 if equal. */ ssize_t bc_num_cmpZero(const BcNum* n) { return bc_num_neg((n)->len != 0, BC_NUM_NEG(n)); } /** * Return the number of integer limbs in a BcNum. This is the opposite of rdx. * @param n The number to return the amount of integer limbs for. * @return The amount of integer limbs in @a n. */ static inline size_t bc_num_int(const BcNum* n) { return n->len ? n->len - BC_NUM_RDX_VAL(n) : 0; } /** * Expand a number's allocation capacity to at least req limbs. * @param n The number to expand. * @param req The number limbs to expand the allocation capacity to. */ static void bc_num_expand(BcNum* restrict n, size_t req) { assert(n != NULL); req = req >= BC_NUM_DEF_SIZE ? req : BC_NUM_DEF_SIZE; if (req > n->cap) { BC_SIG_LOCK; n->num = bc_vm_realloc(n->num, BC_NUM_SIZE(req)); n->cap = req; BC_SIG_UNLOCK; } } /** * Set a number to 0 with the specified scale. * @param n The number to set to zero. * @param scale The scale to set the number to. */ static inline void bc_num_setToZero(BcNum* restrict n, size_t scale) { assert(n != NULL); n->scale = scale; n->len = n->rdx = 0; } void bc_num_zero(BcNum* restrict n) { bc_num_setToZero(n, 0); } void bc_num_one(BcNum* restrict n) { bc_num_zero(n); n->len = 1; n->num[0] = 1; } /** * "Cleans" a number, which means reducing the length if the most significant * limbs are zero. * @param n The number to clean. */ static void bc_num_clean(BcNum* restrict n) { // Reduce the length. while (BC_NUM_NONZERO(n) && !n->num[n->len - 1]) { n->len -= 1; } // Special cases. if (BC_NUM_ZERO(n)) n->rdx = 0; else { // len must be at least as much as rdx. size_t rdx = BC_NUM_RDX_VAL(n); if (n->len < rdx) n->len = rdx; } } /** * Returns the log base 10 of @a i. I could have done this with floating-point * math, and in fact, I originally did. However, that was the only * floating-point code in the entire codebase, and I decided I didn't want any. * This is fast enough. Also, it might handle larger numbers better. * @param i The number to return the log base 10 of. * @return The log base 10 of @a i. */ static size_t bc_num_log10(size_t i) { size_t len; for (len = 1; i; i /= BC_BASE, ++len) { continue; } assert(len - 1 <= BC_BASE_DIGS + 1); return len - 1; } /** * Returns the number of decimal digits in a limb that are zero starting at the * most significant digits. This basically returns how much of the limb is used. * @param n The number. * @return The number of decimal digits that are 0 starting at the most * significant digits. */ static inline size_t bc_num_zeroDigits(const BcDig* n) { assert(*n >= 0); assert(((size_t) *n) < BC_BASE_POW); return BC_BASE_DIGS - bc_num_log10((size_t) *n); } /** * Returns the power of 10 that the least significant limb should be multiplied * by to put its digits in the right place. For example, if the scale only * reaches 8 places into the limb, this will return 1 (because it should be * multiplied by 10^1) to put the number in the correct place. * @param scale The scale. * @return The power of 10 that the least significant limb should be * multiplied by */ static inline size_t bc_num_leastSigPow(size_t scale) { size_t digs; digs = scale % BC_BASE_DIGS; digs = digs != 0 ? BC_BASE_DIGS - digs : 0; return bc_num_pow10[digs]; } /** * Return the total number of integer digits in a number. This is the opposite * of scale, like bc_num_int() is the opposite of rdx. * @param n The number. * @return The number of integer digits in @a n. */ static size_t bc_num_intDigits(const BcNum* n) { size_t digits = bc_num_int(n) * BC_BASE_DIGS; if (digits > 0) digits -= bc_num_zeroDigits(n->num + n->len - 1); return digits; } /** * Returns the number of limbs of a number that are non-zero starting at the * most significant limbs. This expects that there are *no* integer limbs in the * number because it is specifically to figure out how many zero limbs after the * decimal place to ignore. If there are zero limbs after non-zero limbs, they * are counted as non-zero limbs. * @param n The number. * @return The number of non-zero limbs after the decimal point. */ static size_t bc_num_nonZeroLen(const BcNum* restrict n) { size_t i, len = n->len; assert(len == BC_NUM_RDX_VAL(n)); for (i = len - 1; i < len && !n->num[i]; --i) { continue; } assert(i + 1 > 0); return i + 1; } +#if BC_ENABLE_EXTRA_MATH + /** * Returns the power of 10 that a number with an absolute value less than 1 * needs to be multiplied by in order to be greater than 1 or less than -1. * @param n The number. * @return The power of 10 that a number greater than 1 and less than -1 must * be multiplied by to be greater than 1 or less than -1. */ static size_t bc_num_negPow10(const BcNum* restrict n) { // Figure out how many limbs after the decimal point is zero. size_t i, places, idx = bc_num_nonZeroLen(n) - 1; places = 1; // Figure out how much in the last limb is zero. for (i = BC_BASE_DIGS - 1; i < BC_BASE_DIGS; --i) { if (bc_num_pow10[i] > (BcBigDig) n->num[idx]) places += 1; else break; } // Calculate the combination of zero limbs and zero digits in the last // limb. return places + (BC_NUM_RDX_VAL(n) - (idx + 1)) * BC_BASE_DIGS; } +#endif // BC_ENABLE_EXTRA_MATH + /** * Performs a one-limb add with a carry. * @param a The first limb. * @param b The second limb. * @param carry An in/out parameter; the carry in from the previous add and the * carry out from this add. * @return The resulting limb sum. */ static BcDig bc_num_addDigits(BcDig a, BcDig b, bool* carry) { assert(((BcBigDig) BC_BASE_POW) * 2 == ((BcDig) BC_BASE_POW) * 2); assert(a < BC_BASE_POW && a >= 0); assert(b < BC_BASE_POW && b >= 0); a += b + *carry; *carry = (a >= BC_BASE_POW); if (*carry) a -= BC_BASE_POW; assert(a >= 0); assert(a < BC_BASE_POW); return a; } /** * Performs a one-limb subtract with a carry. * @param a The first limb. * @param b The second limb. * @param carry An in/out parameter; the carry in from the previous subtract * and the carry out from this subtract. * @return The resulting limb difference. */ static BcDig bc_num_subDigits(BcDig a, BcDig b, bool* carry) { assert(a < BC_BASE_POW && a >= 0); assert(b < BC_BASE_POW && b >= 0); b += *carry; *carry = (a < b); if (*carry) a += BC_BASE_POW; assert(a - b >= 0); assert(a - b < BC_BASE_POW); return a - b; } /** * Add two BcDig arrays and store the result in the first array. * @param a The first operand and out array. * @param b The second operand. * @param len The length of @a b. */ static void bc_num_addArrays(BcDig* restrict a, const BcDig* restrict b, size_t len) { size_t i; bool carry = false; for (i = 0; i < len; ++i) { a[i] = bc_num_addDigits(a[i], b[i], &carry); } // Take care of the extra limbs in the bigger array. for (; carry; ++i) { a[i] = bc_num_addDigits(a[i], 0, &carry); } } /** * Subtract two BcDig arrays and store the result in the first array. * @param a The first operand and out array. * @param b The second operand. * @param len The length of @a b. */ static void bc_num_subArrays(BcDig* restrict a, const BcDig* restrict b, size_t len) { size_t i; bool carry = false; for (i = 0; i < len; ++i) { a[i] = bc_num_subDigits(a[i], b[i], &carry); } // Take care of the extra limbs in the bigger array. for (; carry; ++i) { a[i] = bc_num_subDigits(a[i], 0, &carry); } } /** * Multiply a BcNum array by a one-limb number. This is a faster version of * multiplication for when we can use it. * @param a The BcNum to multiply by the one-limb number. * @param b The one limb of the one-limb number. * @param c The return parameter. */ static void bc_num_mulArray(const BcNum* restrict a, BcBigDig b, BcNum* restrict c) { size_t i; BcBigDig carry = 0; assert(b <= BC_BASE_POW); // Make sure the return parameter is big enough. if (a->len + 1 > c->cap) bc_num_expand(c, a->len + 1); // We want the entire return parameter to be zero for cleaning later. // NOLINTNEXTLINE memset(c->num, 0, BC_NUM_SIZE(c->cap)); // Actual multiplication loop. for (i = 0; i < a->len; ++i) { BcBigDig in = ((BcBigDig) a->num[i]) * b + carry; c->num[i] = in % BC_BASE_POW; carry = in / BC_BASE_POW; } assert(carry < BC_BASE_POW); // Finishing touches. c->num[i] = (BcDig) carry; assert(c->num[i] >= 0 && c->num[i] < BC_BASE_POW); c->len = a->len; c->len += (carry != 0); bc_num_clean(c); // Postconditions. assert(!BC_NUM_NEG(c) || BC_NUM_NONZERO(c)); assert(BC_NUM_RDX_VAL(c) <= c->len || !c->len); assert(!c->len || c->num[c->len - 1] || BC_NUM_RDX_VAL(c) == c->len); } /** * Divide a BcNum array by a one-limb number. This is a faster version of divide * for when we can use it. * @param a The BcNum to multiply by the one-limb number. * @param b The one limb of the one-limb number. * @param c The return parameter for the quotient. * @param rem The return parameter for the remainder. */ static void bc_num_divArray(const BcNum* restrict a, BcBigDig b, BcNum* restrict c, BcBigDig* rem) { size_t i; BcBigDig carry = 0; assert(c->cap >= a->len); // Actual division loop. for (i = a->len - 1; i < a->len; --i) { BcBigDig in = ((BcBigDig) a->num[i]) + carry * BC_BASE_POW; assert(in / b < BC_BASE_POW); c->num[i] = (BcDig) (in / b); assert(c->num[i] >= 0 && c->num[i] < BC_BASE_POW); carry = in % b; } // Finishing touches. c->len = a->len; bc_num_clean(c); *rem = carry; // Postconditions. assert(!BC_NUM_NEG(c) || BC_NUM_NONZERO(c)); assert(BC_NUM_RDX_VAL(c) <= c->len || !c->len); assert(!c->len || c->num[c->len - 1] || BC_NUM_RDX_VAL(c) == c->len); } /** * Compare two BcDig arrays and return >0 if @a b is greater, <0 if @a b is * less, and 0 if equal. Both @a a and @a b must have the same length. * @param a The first array. * @param b The second array. * @param len The minimum length of the arrays. */ static ssize_t bc_num_compare(const BcDig* restrict a, const BcDig* restrict b, size_t len) { size_t i; BcDig c = 0; for (i = len - 1; i < len && !(c = a[i] - b[i]); --i) { continue; } return bc_num_neg(i + 1, c < 0); } ssize_t bc_num_cmp(const BcNum* a, const BcNum* b) { size_t i, min, a_int, b_int, diff, ardx, brdx; BcDig* max_num; BcDig* min_num; bool a_max, neg = false; ssize_t cmp; assert(a != NULL && b != NULL); // Same num? Equal. if (a == b) return 0; // Easy cases. if (BC_NUM_ZERO(a)) return bc_num_neg(b->len != 0, !BC_NUM_NEG(b)); if (BC_NUM_ZERO(b)) return bc_num_cmpZero(a); if (BC_NUM_NEG(a)) { if (BC_NUM_NEG(b)) neg = true; else return -1; } else if (BC_NUM_NEG(b)) return 1; // Get the number of int limbs in each number and get the difference. a_int = bc_num_int(a); b_int = bc_num_int(b); a_int -= b_int; // If there's a difference, then just return the comparison. if (a_int) return neg ? -((ssize_t) a_int) : (ssize_t) a_int; // Get the rdx's and figure out the max. ardx = BC_NUM_RDX_VAL(a); brdx = BC_NUM_RDX_VAL(b); a_max = (ardx > brdx); // Set variables based on the above. if (a_max) { min = brdx; diff = ardx - brdx; max_num = a->num + diff; min_num = b->num; } else { min = ardx; diff = brdx - ardx; max_num = b->num + diff; min_num = a->num; } // Do a full limb-by-limb comparison. cmp = bc_num_compare(max_num, min_num, b_int + min); // If we found a difference, return it based on state. if (cmp) return bc_num_neg((size_t) cmp, !a_max == !neg); // If there was no difference, then the final step is to check which number // has greater or lesser limbs beyond the other's. for (max_num -= diff, i = diff - 1; i < diff; --i) { if (max_num[i]) return bc_num_neg(1, !a_max == !neg); } return 0; } void bc_num_truncate(BcNum* restrict n, size_t places) { size_t nrdx, places_rdx; if (!places) return; // Grab these needed values; places_rdx is the rdx equivalent to places like // rdx is to scale. nrdx = BC_NUM_RDX_VAL(n); places_rdx = nrdx ? nrdx - BC_NUM_RDX(n->scale - places) : 0; // We cannot truncate more places than we have. assert(places <= n->scale && (BC_NUM_ZERO(n) || places_rdx <= n->len)); n->scale -= places; BC_NUM_RDX_SET(n, nrdx - places_rdx); // Only when the number is nonzero do we need to do the hard stuff. if (BC_NUM_NONZERO(n)) { size_t pow; // This calculates how many decimal digits are in the least significant // limb, then gets the power for that. pow = bc_num_leastSigPow(n->scale); n->len -= places_rdx; // We have to move limbs to maintain invariants. The limbs must begin at // the beginning of the BcNum array. // NOLINTNEXTLINE memmove(n->num, n->num + places_rdx, BC_NUM_SIZE(n->len)); // Clear the lower part of the last digit. if (BC_NUM_NONZERO(n)) n->num[0] -= n->num[0] % (BcDig) pow; bc_num_clean(n); } } void bc_num_extend(BcNum* restrict n, size_t places) { size_t nrdx, places_rdx; if (!places) return; // Easy case with zero; set the scale. if (BC_NUM_ZERO(n)) { n->scale += places; return; } // Grab these needed values; places_rdx is the rdx equivalent to places like // rdx is to scale. nrdx = BC_NUM_RDX_VAL(n); places_rdx = BC_NUM_RDX(places + n->scale) - nrdx; // This is the hard case. We need to expand the number, move the limbs, and // set the limbs that were just cleared. if (places_rdx) { bc_num_expand(n, bc_vm_growSize(n->len, places_rdx)); // NOLINTNEXTLINE memmove(n->num + places_rdx, n->num, BC_NUM_SIZE(n->len)); // NOLINTNEXTLINE memset(n->num, 0, BC_NUM_SIZE(places_rdx)); } // Finally, set scale and rdx. BC_NUM_RDX_SET(n, nrdx + places_rdx); n->scale += places; n->len += places_rdx; assert(BC_NUM_RDX_VAL(n) == BC_NUM_RDX(n->scale)); } /** * Retires (finishes) a multiplication or division operation. */ static void bc_num_retireMul(BcNum* restrict n, size_t scale, bool neg1, bool neg2) { // Make sure scale is correct. if (n->scale < scale) bc_num_extend(n, scale - n->scale); else bc_num_truncate(n, n->scale - scale); bc_num_clean(n); // We need to ensure rdx is correct. if (BC_NUM_NONZERO(n)) n->rdx = BC_NUM_NEG_VAL(n, !neg1 != !neg2); } /** * Splits a number into two BcNum's. This is used in Karatsuba. * @param n The number to split. * @param idx The index to split at. * @param a An out parameter; the low part of @a n. * @param b An out parameter; the high part of @a n. */ static void bc_num_split(const BcNum* restrict n, size_t idx, BcNum* restrict a, BcNum* restrict b) { // We want a and b to be clear. assert(BC_NUM_ZERO(a)); assert(BC_NUM_ZERO(b)); // The usual case. if (idx < n->len) { // Set the fields first. b->len = n->len - idx; a->len = idx; a->scale = b->scale = 0; BC_NUM_RDX_SET(a, 0); BC_NUM_RDX_SET(b, 0); assert(a->cap >= a->len); assert(b->cap >= b->len); // Copy the arrays. This is not necessary for safety, but it is faster, // for some reason. // NOLINTNEXTLINE memcpy(b->num, n->num + idx, BC_NUM_SIZE(b->len)); // NOLINTNEXTLINE memcpy(a->num, n->num, BC_NUM_SIZE(idx)); bc_num_clean(b); } // If the index is weird, just skip the split. else bc_num_copy(a, n); bc_num_clean(a); } /** * Copies a number into another, but shifts the rdx so that the result number * only sees the integer part of the source number. * @param n The number to copy. * @param r The result number with a shifted rdx, len, and num. */ static void bc_num_shiftRdx(const BcNum* restrict n, BcNum* restrict r) { size_t rdx = BC_NUM_RDX_VAL(n); r->len = n->len - rdx; r->cap = n->cap - rdx; r->num = n->num + rdx; BC_NUM_RDX_SET_NEG(r, 0, BC_NUM_NEG(n)); r->scale = 0; } /** * Shifts a number so that all of the least significant limbs of the number are * skipped. This must be undone by bc_num_unshiftZero(). * @param n The number to shift. */ static size_t bc_num_shiftZero(BcNum* restrict n) { // This is volatile to quiet a GCC warning about longjmp() clobbering. volatile size_t i; // If we don't have an integer, that is a problem, but it's also a bug // because the caller should have set everything up right. assert(!BC_NUM_RDX_VAL(n) || BC_NUM_ZERO(n)); for (i = 0; i < n->len && !n->num[i]; ++i) { continue; } n->len -= i; n->num += i; return i; } /** * Undo the damage done by bc_num_unshiftZero(). This must be called like all * cleanup functions: after a label used by setjmp() and longjmp(). * @param n The number to unshift. * @param places_rdx The amount the number was originally shift. */ static void bc_num_unshiftZero(BcNum* restrict n, size_t places_rdx) { n->len += places_rdx; n->num -= places_rdx; } /** * Shifts the digits right within a number by no more than BC_BASE_DIGS - 1. * This is the final step on shifting numbers with the shift operators. It * depends on the caller to shift the limbs properly because all it does is * ensure that the rdx point is realigned to be between limbs. * @param n The number to shift digits in. * @param dig The number of places to shift right. */ static void bc_num_shift(BcNum* restrict n, BcBigDig dig) { size_t i, len = n->len; BcBigDig carry = 0, pow; BcDig* ptr = n->num; assert(dig < BC_BASE_DIGS); // Figure out the parameters for division. pow = bc_num_pow10[dig]; dig = bc_num_pow10[BC_BASE_DIGS - dig]; // Run a series of divisions and mods with carries across the entire number // array. This effectively shifts everything over. for (i = len - 1; i < len; --i) { BcBigDig in, temp; in = ((BcBigDig) ptr[i]); temp = carry * dig; carry = in % pow; ptr[i] = ((BcDig) (in / pow)) + (BcDig) temp; assert(ptr[i] >= 0 && ptr[i] < BC_BASE_POW); } assert(!carry); } /** * Shift a number left by a certain number of places. This is the workhorse of * the left shift operator. * @param n The number to shift left. * @param places The amount of places to shift @a n left by. */ static void bc_num_shiftLeft(BcNum* restrict n, size_t places) { BcBigDig dig; size_t places_rdx; bool shift; if (!places) return; // Make sure to grow the number if necessary. if (places > n->scale) { size_t size = bc_vm_growSize(BC_NUM_RDX(places - n->scale), n->len); if (size > SIZE_MAX - 1) bc_err(BC_ERR_MATH_OVERFLOW); } // If zero, we can just set the scale and bail. if (BC_NUM_ZERO(n)) { if (n->scale >= places) n->scale -= places; else n->scale = 0; return; } // When I changed bc to have multiple digits per limb, this was the hardest // code to change. This and shift right. Make sure you understand this // before attempting anything. // This is how many limbs we will shift. dig = (BcBigDig) (places % BC_BASE_DIGS); shift = (dig != 0); // Convert places to a rdx value. places_rdx = BC_NUM_RDX(places); // If the number is not an integer, we need special care. The reason an // integer doesn't is because left shift would only extend the integer, // whereas a non-integer might have its fractional part eliminated or only // partially eliminated. if (n->scale) { size_t nrdx = BC_NUM_RDX_VAL(n); // If the number's rdx is bigger, that's the hard case. if (nrdx >= places_rdx) { size_t mod = n->scale % BC_BASE_DIGS, revdig; // We want mod to be in the range [1, BC_BASE_DIGS], not // [0, BC_BASE_DIGS). mod = mod ? mod : BC_BASE_DIGS; // We need to reverse dig to get the actual number of digits. revdig = dig ? BC_BASE_DIGS - dig : 0; // If the two overflow BC_BASE_DIGS, we need to move an extra place. if (mod + revdig > BC_BASE_DIGS) places_rdx = 1; else places_rdx = 0; } else places_rdx -= nrdx; } // If this is non-zero, we need an extra place, so expand, move, and set. if (places_rdx) { bc_num_expand(n, bc_vm_growSize(n->len, places_rdx)); // NOLINTNEXTLINE memmove(n->num + places_rdx, n->num, BC_NUM_SIZE(n->len)); // NOLINTNEXTLINE memset(n->num, 0, BC_NUM_SIZE(places_rdx)); n->len += places_rdx; } // Set the scale appropriately. if (places > n->scale) { n->scale = 0; BC_NUM_RDX_SET(n, 0); } else { n->scale -= places; BC_NUM_RDX_SET(n, BC_NUM_RDX(n->scale)); } // Finally, shift within limbs. if (shift) bc_num_shift(n, BC_BASE_DIGS - dig); bc_num_clean(n); } void bc_num_shiftRight(BcNum* restrict n, size_t places) { BcBigDig dig; size_t places_rdx, scale, scale_mod, int_len, expand; bool shift; if (!places) return; // If zero, we can just set the scale and bail. if (BC_NUM_ZERO(n)) { n->scale += places; bc_num_expand(n, BC_NUM_RDX(n->scale)); return; } // Amount within a limb we have to shift by. dig = (BcBigDig) (places % BC_BASE_DIGS); shift = (dig != 0); scale = n->scale; // Figure out how the scale is affected. scale_mod = scale % BC_BASE_DIGS; scale_mod = scale_mod ? scale_mod : BC_BASE_DIGS; // We need to know the int length and rdx for places. int_len = bc_num_int(n); places_rdx = BC_NUM_RDX(places); // If we are going to shift past a limb boundary or not, set accordingly. if (scale_mod + dig > BC_BASE_DIGS) { expand = places_rdx - 1; places_rdx = 1; } else { expand = places_rdx; places_rdx = 0; } // Clamp expanding. if (expand > int_len) expand -= int_len; else expand = 0; // Extend, expand, and zero. bc_num_extend(n, places_rdx * BC_BASE_DIGS); bc_num_expand(n, bc_vm_growSize(expand, n->len)); // NOLINTNEXTLINE memset(n->num + n->len, 0, BC_NUM_SIZE(expand)); // Set the fields. n->len += expand; n->scale = 0; BC_NUM_RDX_SET(n, 0); // Finally, shift within limbs. if (shift) bc_num_shift(n, dig); n->scale = scale + places; BC_NUM_RDX_SET(n, BC_NUM_RDX(n->scale)); bc_num_clean(n); assert(BC_NUM_RDX_VAL(n) <= n->len && n->len <= n->cap); assert(BC_NUM_RDX_VAL(n) == BC_NUM_RDX(n->scale)); } /** * Tests if a number is a integer with scale or not. Returns true if the number * is not an integer. If it is, its integer shifted form is copied into the * result parameter for use where only integers are allowed. * @param n The integer to test and shift. * @param r The number to store the shifted result into. This number should * *not* be allocated. * @return True if the number is a non-integer, false otherwise. */ static bool bc_num_nonInt(const BcNum* restrict n, BcNum* restrict r) { bool zero; size_t i, rdx = BC_NUM_RDX_VAL(n); if (!rdx) { // NOLINTNEXTLINE memcpy(r, n, sizeof(BcNum)); return false; } zero = true; for (i = 0; zero && i < rdx; ++i) { zero = (n->num[i] == 0); } if (BC_ERR(!zero)) return true; bc_num_shiftRdx(n, r); return false; } #if BC_ENABLE_EXTRA_MATH /** * Execute common code for an operater that needs an integer for the second * operand and return the integer operand as a BcBigDig. * @param a The first operand. * @param b The second operand. * @param c The result operand. * @return The second operand as a hardware integer. */ static BcBigDig bc_num_intop(const BcNum* a, const BcNum* b, BcNum* restrict c) { BcNum temp; #if BC_GCC temp.len = 0; temp.rdx = 0; temp.num = NULL; #endif // BC_GCC if (BC_ERR(bc_num_nonInt(b, &temp))) bc_err(BC_ERR_MATH_NON_INTEGER); bc_num_copy(c, a); return bc_num_bigdig(&temp); } #endif // BC_ENABLE_EXTRA_MATH /** * This is the actual implementation of add *and* subtract. Since this function * doesn't need to use scale (per the bc spec), I am hijacking it to say whether * it's doing an add or a subtract. And then I convert substraction to addition * of negative second operand. This is a BcNumBinOp function. * @param a The first operand. * @param b The second operand. * @param c The return parameter. * @param sub Non-zero for a subtract, zero for an add. */ static void bc_num_as(BcNum* a, BcNum* b, BcNum* restrict c, size_t sub) { BcDig* ptr_c; BcDig* ptr_l; BcDig* ptr_r; size_t i, min_rdx, max_rdx, diff, a_int, b_int, min_len, max_len, max_int; size_t len_l, len_r, ardx, brdx; bool b_neg, do_sub, do_rev_sub, carry, c_neg; if (BC_NUM_ZERO(b)) { bc_num_copy(c, a); return; } if (BC_NUM_ZERO(a)) { bc_num_copy(c, b); c->rdx = BC_NUM_NEG_VAL(c, BC_NUM_NEG(b) != sub); return; } // Invert sign of b if it is to be subtracted. This operation must // precede the tests for any of the operands being zero. b_neg = (BC_NUM_NEG(b) != sub); // Figure out if we will actually add the numbers if their signs are equal // or subtract. do_sub = (BC_NUM_NEG(a) != b_neg); a_int = bc_num_int(a); b_int = bc_num_int(b); max_int = BC_MAX(a_int, b_int); // Figure out which number will have its last limbs copied (for addition) or // subtracted (for subtraction). ardx = BC_NUM_RDX_VAL(a); brdx = BC_NUM_RDX_VAL(b); min_rdx = BC_MIN(ardx, brdx); max_rdx = BC_MAX(ardx, brdx); diff = max_rdx - min_rdx; max_len = max_int + max_rdx; if (do_sub) { // Check whether b has to be subtracted from a or a from b. if (a_int != b_int) do_rev_sub = (a_int < b_int); else if (ardx > brdx) { do_rev_sub = (bc_num_compare(a->num + diff, b->num, b->len) < 0); } else do_rev_sub = (bc_num_compare(a->num, b->num + diff, a->len) <= 0); } else { // The result array of the addition might come out one element // longer than the bigger of the operand arrays. max_len += 1; do_rev_sub = (a_int < b_int); } assert(max_len <= c->cap); // Cache values for simple code later. if (do_rev_sub) { ptr_l = b->num; ptr_r = a->num; len_l = b->len; len_r = a->len; } else { ptr_l = a->num; ptr_r = b->num; len_l = a->len; len_r = b->len; } ptr_c = c->num; carry = false; // This is true if the numbers have a different number of limbs after the // decimal point. if (diff) { // If the rdx values of the operands do not match, the result will // have low end elements that are the positive or negative trailing // elements of the operand with higher rdx value. if ((ardx > brdx) != do_rev_sub) { // !do_rev_sub && ardx > brdx || do_rev_sub && brdx > ardx // The left operand has BcDig values that need to be copied, // either from a or from b (in case of a reversed subtraction). // NOLINTNEXTLINE memcpy(ptr_c, ptr_l, BC_NUM_SIZE(diff)); ptr_l += diff; len_l -= diff; } else { // The right operand has BcDig values that need to be copied // or subtracted from zero (in case of a subtraction). if (do_sub) { // do_sub (do_rev_sub && ardx > brdx || // !do_rev_sub && brdx > ardx) for (i = 0; i < diff; i++) { ptr_c[i] = bc_num_subDigits(0, ptr_r[i], &carry); } } else { // !do_sub && brdx > ardx // NOLINTNEXTLINE memcpy(ptr_c, ptr_r, BC_NUM_SIZE(diff)); } // Future code needs to ignore the limbs we just did. ptr_r += diff; len_r -= diff; } // The return value pointer needs to ignore what we just did. ptr_c += diff; } // This is the length that can be directly added/subtracted. min_len = BC_MIN(len_l, len_r); // After dealing with possible low array elements that depend on only one // operand above, the actual add or subtract can be performed as if the rdx // of both operands was the same. // // Inlining takes care of eliminating constant zero arguments to // addDigit/subDigit (checked in disassembly of resulting bc binary // compiled with gcc and clang). if (do_sub) { // Actual subtraction. for (i = 0; i < min_len; ++i) { ptr_c[i] = bc_num_subDigits(ptr_l[i], ptr_r[i], &carry); } // Finishing the limbs beyond the direct subtraction. for (; i < len_l; ++i) { ptr_c[i] = bc_num_subDigits(ptr_l[i], 0, &carry); } } else { // Actual addition. for (i = 0; i < min_len; ++i) { ptr_c[i] = bc_num_addDigits(ptr_l[i], ptr_r[i], &carry); } // Finishing the limbs beyond the direct addition. for (; i < len_l; ++i) { ptr_c[i] = bc_num_addDigits(ptr_l[i], 0, &carry); } // Addition can create an extra limb. We take care of that here. ptr_c[i] = bc_num_addDigits(0, 0, &carry); } assert(carry == false); // The result has the same sign as a, unless the operation was a // reverse subtraction (b - a). c_neg = BC_NUM_NEG(a) != (do_sub && do_rev_sub); BC_NUM_RDX_SET_NEG(c, max_rdx, c_neg); c->len = max_len; c->scale = BC_MAX(a->scale, b->scale); bc_num_clean(c); } /** * The simple multiplication that karatsuba dishes out to when the length of the * numbers gets low enough. This doesn't use scale because it treats the * operands as though they are integers. * @param a The first operand. * @param b The second operand. * @param c The return parameter. */ static void bc_num_m_simp(const BcNum* a, const BcNum* b, BcNum* restrict c) { size_t i, alen = a->len, blen = b->len, clen; BcDig* ptr_a = a->num; BcDig* ptr_b = b->num; BcDig* ptr_c; BcBigDig sum = 0, carry = 0; assert(sizeof(sum) >= sizeof(BcDig) * 2); assert(!BC_NUM_RDX_VAL(a) && !BC_NUM_RDX_VAL(b)); // Make sure c is big enough. clen = bc_vm_growSize(alen, blen); bc_num_expand(c, bc_vm_growSize(clen, 1)); // If we don't memset, then we might have uninitialized data use later. ptr_c = c->num; // NOLINTNEXTLINE memset(ptr_c, 0, BC_NUM_SIZE(c->cap)); // This is the actual multiplication loop. It uses the lattice form of long // multiplication (see the explanation on the web page at // https://knilt.arcc.albany.edu/What_is_Lattice_Multiplication or the // explanation at Wikipedia). for (i = 0; i < clen; ++i) { ssize_t sidx = (ssize_t) (i - blen + 1); size_t j, k; // These are the start indices. j = (size_t) BC_MAX(0, sidx); k = BC_MIN(i, blen - 1); // On every iteration of this loop, a multiplication happens, then the // sum is automatically calculated. for (; j < alen && k < blen; ++j, --k) { sum += ((BcBigDig) ptr_a[j]) * ((BcBigDig) ptr_b[k]); if (sum >= ((BcBigDig) BC_BASE_POW) * BC_BASE_POW) { carry += sum / BC_BASE_POW; sum %= BC_BASE_POW; } } // Calculate the carry. if (sum >= BC_BASE_POW) { carry += sum / BC_BASE_POW; sum %= BC_BASE_POW; } // Store and set up for next iteration. ptr_c[i] = (BcDig) sum; assert(ptr_c[i] < BC_BASE_POW); sum = carry; carry = 0; } // This should always be true because there should be no carry on the last // digit; multiplication never goes above the sum of both lengths. assert(!sum); c->len = clen; } /** * Does a shifted add or subtract for Karatsuba below. This calls either * bc_num_addArrays() or bc_num_subArrays(). * @param n An in/out parameter; the first operand and return parameter. * @param a The second operand. * @param shift The amount to shift @a n by when adding/subtracting. * @param op The function to call, either bc_num_addArrays() or * bc_num_subArrays(). */ static void bc_num_shiftAddSub(BcNum* restrict n, const BcNum* restrict a, size_t shift, BcNumShiftAddOp op) { assert(n->len >= shift + a->len); assert(!BC_NUM_RDX_VAL(n) && !BC_NUM_RDX_VAL(a)); op(n->num + shift, a->num, a->len); } /** * Implements the Karatsuba algorithm. */ static void bc_num_k(const BcNum* a, const BcNum* b, BcNum* restrict c) { size_t max, max2, total; BcNum l1, h1, l2, h2, m2, m1, z0, z1, z2, temp; BcDig* digs; BcDig* dig_ptr; BcNumShiftAddOp op; bool aone = BC_NUM_ONE(a); #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY assert(BC_NUM_ZERO(c)); if (BC_NUM_ZERO(a) || BC_NUM_ZERO(b)) return; if (aone || BC_NUM_ONE(b)) { bc_num_copy(c, aone ? b : a); if ((aone && BC_NUM_NEG(a)) || BC_NUM_NEG(b)) BC_NUM_NEG_TGL(c); return; } // Shell out to the simple algorithm with certain conditions. if (a->len < BC_NUM_KARATSUBA_LEN || b->len < BC_NUM_KARATSUBA_LEN) { bc_num_m_simp(a, b, c); return; } // We need to calculate the max size of the numbers that can result from the // operations. max = BC_MAX(a->len, b->len); max = BC_MAX(max, BC_NUM_DEF_SIZE); max2 = (max + 1) / 2; // Calculate the space needed for all of the temporary allocations. We do // this to just allocate once. total = bc_vm_arraySize(BC_NUM_KARATSUBA_ALLOCS, max); BC_SIG_LOCK; // Allocate space for all of the temporaries. digs = dig_ptr = bc_vm_malloc(BC_NUM_SIZE(total)); // Set up the temporaries. bc_num_setup(&l1, dig_ptr, max); dig_ptr += max; bc_num_setup(&h1, dig_ptr, max); dig_ptr += max; bc_num_setup(&l2, dig_ptr, max); dig_ptr += max; bc_num_setup(&h2, dig_ptr, max); dig_ptr += max; bc_num_setup(&m1, dig_ptr, max); dig_ptr += max; bc_num_setup(&m2, dig_ptr, max); // Some temporaries need the ability to grow, so we allocate them // separately. max = bc_vm_growSize(max, 1); bc_num_init(&z0, max); bc_num_init(&z1, max); bc_num_init(&z2, max); max = bc_vm_growSize(max, max) + 1; bc_num_init(&temp, max); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; // First, set up c. bc_num_expand(c, max); c->len = max; // NOLINTNEXTLINE memset(c->num, 0, BC_NUM_SIZE(c->len)); // Split the parameters. bc_num_split(a, max2, &l1, &h1); bc_num_split(b, max2, &l2, &h2); // Do the subtraction. bc_num_sub(&h1, &l1, &m1, 0); bc_num_sub(&l2, &h2, &m2, 0); // The if statements below are there for efficiency reasons. The best way to // understand them is to understand the Karatsuba algorithm because now that // the ollocations and splits are done, the algorithm is pretty // straightforward. if (BC_NUM_NONZERO(&h1) && BC_NUM_NONZERO(&h2)) { assert(BC_NUM_RDX_VALID_NP(h1)); assert(BC_NUM_RDX_VALID_NP(h2)); bc_num_m(&h1, &h2, &z2, 0); bc_num_clean(&z2); bc_num_shiftAddSub(c, &z2, max2 * 2, bc_num_addArrays); bc_num_shiftAddSub(c, &z2, max2, bc_num_addArrays); } if (BC_NUM_NONZERO(&l1) && BC_NUM_NONZERO(&l2)) { assert(BC_NUM_RDX_VALID_NP(l1)); assert(BC_NUM_RDX_VALID_NP(l2)); bc_num_m(&l1, &l2, &z0, 0); bc_num_clean(&z0); bc_num_shiftAddSub(c, &z0, max2, bc_num_addArrays); bc_num_shiftAddSub(c, &z0, 0, bc_num_addArrays); } if (BC_NUM_NONZERO(&m1) && BC_NUM_NONZERO(&m2)) { assert(BC_NUM_RDX_VALID_NP(m1)); assert(BC_NUM_RDX_VALID_NP(m1)); bc_num_m(&m1, &m2, &z1, 0); bc_num_clean(&z1); op = (BC_NUM_NEG_NP(m1) != BC_NUM_NEG_NP(m2)) ? bc_num_subArrays : bc_num_addArrays; bc_num_shiftAddSub(c, &z1, max2, op); } err: BC_SIG_MAYLOCK; free(digs); bc_num_free(&temp); bc_num_free(&z2); bc_num_free(&z1); bc_num_free(&z0); BC_LONGJMP_CONT(vm); } /** * Does checks for Karatsuba. It also changes things to ensure that the * Karatsuba and simple multiplication can treat the numbers as integers. This * is a BcNumBinOp function. * @param a The first operand. * @param b The second operand. * @param c The return parameter. * @param scale The current scale. */ static void bc_num_m(BcNum* a, BcNum* b, BcNum* restrict c, size_t scale) { BcNum cpa, cpb; size_t ascale, bscale, ardx, brdx, zero, len, rscale; // These are meant to quiet warnings on GCC about longjmp() clobbering. // The problem is real here. size_t scale1, scale2, realscale; // These are meant to quiet the GCC longjmp() clobbering, even though it // does not apply here. volatile size_t azero; volatile size_t bzero; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY assert(BC_NUM_RDX_VALID(a)); assert(BC_NUM_RDX_VALID(b)); bc_num_zero(c); ascale = a->scale; bscale = b->scale; // This sets the final scale according to the bc spec. scale1 = BC_MAX(scale, ascale); scale2 = BC_MAX(scale1, bscale); rscale = ascale + bscale; realscale = BC_MIN(rscale, scale2); // If this condition is true, we can use bc_num_mulArray(), which would be // much faster. if ((a->len == 1 || b->len == 1) && !a->rdx && !b->rdx) { BcNum* operand; BcBigDig dig; // Set the correct operands. if (a->len == 1) { dig = (BcBigDig) a->num[0]; operand = b; } else { dig = (BcBigDig) b->num[0]; operand = a; } bc_num_mulArray(operand, dig, c); // Need to make sure the sign is correct. if (BC_NUM_NONZERO(c)) { c->rdx = BC_NUM_NEG_VAL(c, BC_NUM_NEG(a) != BC_NUM_NEG(b)); } return; } assert(BC_NUM_RDX_VALID(a)); assert(BC_NUM_RDX_VALID(b)); BC_SIG_LOCK; // We need copies because of all of the mutation needed to make Karatsuba // think the numbers are integers. bc_num_init(&cpa, a->len + BC_NUM_RDX_VAL(a)); bc_num_init(&cpb, b->len + BC_NUM_RDX_VAL(b)); BC_SETJMP_LOCKED(vm, init_err); BC_SIG_UNLOCK; bc_num_copy(&cpa, a); bc_num_copy(&cpb, b); assert(BC_NUM_RDX_VALID_NP(cpa)); assert(BC_NUM_RDX_VALID_NP(cpb)); BC_NUM_NEG_CLR_NP(cpa); BC_NUM_NEG_CLR_NP(cpb); assert(BC_NUM_RDX_VALID_NP(cpa)); assert(BC_NUM_RDX_VALID_NP(cpb)); // These are what makes them appear like integers. ardx = BC_NUM_RDX_VAL_NP(cpa) * BC_BASE_DIGS; bc_num_shiftLeft(&cpa, ardx); brdx = BC_NUM_RDX_VAL_NP(cpb) * BC_BASE_DIGS; bc_num_shiftLeft(&cpb, brdx); // We need to reset the jump here because azero and bzero are used in the // cleanup, and local variables are not guaranteed to be the same after a // jump. BC_SIG_LOCK; BC_UNSETJMP(vm); // We want to ignore zero limbs. azero = bc_num_shiftZero(&cpa); bzero = bc_num_shiftZero(&cpb); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; bc_num_clean(&cpa); bc_num_clean(&cpb); bc_num_k(&cpa, &cpb, c); // The return parameter needs to have its scale set. This is the start. It // also needs to be shifted by the same amount as a and b have limbs after // the decimal point. zero = bc_vm_growSize(azero, bzero); len = bc_vm_growSize(c->len, zero); bc_num_expand(c, len); // Shift c based on the limbs after the decimal point in a and b. bc_num_shiftLeft(c, (len - c->len) * BC_BASE_DIGS); bc_num_shiftRight(c, ardx + brdx); bc_num_retireMul(c, realscale, BC_NUM_NEG(a), BC_NUM_NEG(b)); err: BC_SIG_MAYLOCK; bc_num_unshiftZero(&cpb, bzero); bc_num_unshiftZero(&cpa, azero); init_err: BC_SIG_MAYLOCK; bc_num_free(&cpb); bc_num_free(&cpa); BC_LONGJMP_CONT(vm); } /** * Returns true if the BcDig array has non-zero limbs, false otherwise. * @param a The array to test. * @param len The length of the array. * @return True if @a has any non-zero limbs, false otherwise. */ static bool bc_num_nonZeroDig(BcDig* restrict a, size_t len) { size_t i; for (i = len - 1; i < len; --i) { if (a[i] != 0) return true; } return false; } /** * Compares a BcDig array against a BcNum. This is especially suited for * division. Returns >0 if @a a is greater than @a b, <0 if it is less, and =0 * if they are equal. * @param a The array. * @param b The number. * @param len The length to assume the arrays are. This is always less than the * actual length because of how this is implemented. */ static ssize_t bc_num_divCmp(const BcDig* a, const BcNum* b, size_t len) { ssize_t cmp; if (b->len > len && a[len]) cmp = bc_num_compare(a, b->num, len + 1); else if (b->len <= len) { if (a[len]) cmp = 1; else cmp = bc_num_compare(a, b->num, len); } else cmp = -1; return cmp; } /** * Extends the two operands of a division by BC_BASE_DIGS minus the number of * digits in the divisor estimate. In other words, it is shifting the numbers in * order to force the divisor estimate to fill the limb. * @param a The first operand. * @param b The second operand. * @param divisor The divisor estimate. */ static void bc_num_divExtend(BcNum* restrict a, BcNum* restrict b, BcBigDig divisor) { size_t pow; assert(divisor < BC_BASE_POW); pow = BC_BASE_DIGS - bc_num_log10((size_t) divisor); bc_num_shiftLeft(a, pow); bc_num_shiftLeft(b, pow); } /** * Actually does division. This is a rewrite of my original code by Stefan Esser * from FreeBSD. * @param a The first operand. * @param b The second operand. * @param c The return parameter. * @param scale The current scale. */ static void bc_num_d_long(BcNum* restrict a, BcNum* restrict b, BcNum* restrict c, size_t scale) { BcBigDig divisor; size_t i, rdx; // This is volatile and len 2 and reallen exist to quiet the GCC warning // about clobbering on longjmp(). This one is possible, I think. volatile size_t len; size_t len2, reallen; // This is volatile and realend exists to quiet the GCC warning about // clobbering on longjmp(). This one is possible, I think. volatile size_t end; size_t realend; BcNum cpb; // This is volatile and realnonzero exists to quiet the GCC warning about // clobbering on longjmp(). This one is possible, I think. volatile bool nonzero; bool realnonzero; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY assert(b->len < a->len); len = b->len; end = a->len - len; assert(len >= 1); // This is a final time to make sure c is big enough and that its array is // properly zeroed. bc_num_expand(c, a->len); // NOLINTNEXTLINE memset(c->num, 0, c->cap * sizeof(BcDig)); // Setup. BC_NUM_RDX_SET(c, BC_NUM_RDX_VAL(a)); c->scale = a->scale; c->len = a->len; // This is pulling the most significant limb of b in order to establish a // good "estimate" for the actual divisor. divisor = (BcBigDig) b->num[len - 1]; // The entire bit of code in this if statement is to tighten the estimate of // the divisor. The condition asks if b has any other non-zero limbs. if (len > 1 && bc_num_nonZeroDig(b->num, len - 1)) { // This takes a little bit of understanding. The "10*BC_BASE_DIGS/6+1" // results in either 16 for 64-bit 9-digit limbs or 7 for 32-bit 4-digit // limbs. Then it shifts a 1 by that many, which in both cases, puts the // result above *half* of the max value a limb can store. Basically, // this quickly calculates if the divisor is greater than half the max // of a limb. nonzero = (divisor > 1 << ((10 * BC_BASE_DIGS) / 6 + 1)); // If the divisor is *not* greater than half the limb... if (!nonzero) { // Extend the parameters by the number of missing digits in the // divisor. bc_num_divExtend(a, b, divisor); // Check bc_num_d(). In there, we grow a again and again. We do it // again here; we *always* want to be sure it is big enough. len2 = BC_MAX(a->len, b->len); bc_num_expand(a, len2 + 1); // Make a have a zero most significant limb to match the len. if (len2 + 1 > a->len) a->len = len2 + 1; // Grab the new divisor estimate, new because the shift has made it // different. reallen = b->len; realend = a->len - reallen; divisor = (BcBigDig) b->num[reallen - 1]; realnonzero = bc_num_nonZeroDig(b->num, reallen - 1); } else { realend = end; realnonzero = nonzero; } } else { realend = end; realnonzero = false; } // If b has other nonzero limbs, we want the divisor to be one higher, so // that it is an upper bound. divisor += realnonzero; // Make sure c can fit the new length. bc_num_expand(c, a->len); // NOLINTNEXTLINE memset(c->num, 0, BC_NUM_SIZE(c->cap)); assert(c->scale >= scale); rdx = BC_NUM_RDX_VAL(c) - BC_NUM_RDX(scale); BC_SIG_LOCK; bc_num_init(&cpb, len + 1); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; // This is the actual division loop. for (i = realend - 1; i < realend && i >= rdx && BC_NUM_NONZERO(a); --i) { ssize_t cmp; BcDig* n; BcBigDig result; n = a->num + i; assert(n >= a->num); result = 0; cmp = bc_num_divCmp(n, b, len); // This is true if n is greater than b, which means that division can // proceed, so this inner loop is the part that implements one instance // of the division. while (cmp >= 0) { BcBigDig n1, dividend, quotient; // These should be named obviously enough. Just imagine that it's a // division of one limb. Because that's what it is. n1 = (BcBigDig) n[len]; dividend = n1 * BC_BASE_POW + (BcBigDig) n[len - 1]; quotient = (dividend / divisor); // If this is true, then we can just subtract. Remember: setting // quotient to 1 is not bad because we already know that n is // greater than b. if (quotient <= 1) { quotient = 1; bc_num_subArrays(n, b->num, len); } else { assert(quotient <= BC_BASE_POW); // We need to multiply and subtract for a quotient above 1. bc_num_mulArray(b, (BcBigDig) quotient, &cpb); bc_num_subArrays(n, cpb.num, cpb.len); } // The result is the *real* quotient, by the way, but it might take // multiple trips around this loop to get it. result += quotient; assert(result <= BC_BASE_POW); // And here's why it might take multiple trips: n might *still* be // greater than b. So we have to loop again. That's what this is // setting up for: the condition of the while loop. if (realnonzero) cmp = bc_num_divCmp(n, b, len); else cmp = -1; } assert(result < BC_BASE_POW); // Store the actual limb quotient. c->num[i] = (BcDig) result; } err: BC_SIG_MAYLOCK; bc_num_free(&cpb); BC_LONGJMP_CONT(vm); } /** * Implements division. This is a BcNumBinOp function. * @param a The first operand. * @param b The second operand. * @param c The return parameter. * @param scale The current scale. */ static void bc_num_d(BcNum* a, BcNum* b, BcNum* restrict c, size_t scale) { size_t len, cpardx; BcNum cpa, cpb; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY if (BC_NUM_ZERO(b)) bc_err(BC_ERR_MATH_DIVIDE_BY_ZERO); if (BC_NUM_ZERO(a)) { bc_num_setToZero(c, scale); return; } if (BC_NUM_ONE(b)) { bc_num_copy(c, a); bc_num_retireMul(c, scale, BC_NUM_NEG(a), BC_NUM_NEG(b)); return; } // If this is true, we can use bc_num_divArray(), which would be faster. if (!BC_NUM_RDX_VAL(a) && !BC_NUM_RDX_VAL(b) && b->len == 1 && !scale) { BcBigDig rem; bc_num_divArray(a, (BcBigDig) b->num[0], c, &rem); bc_num_retireMul(c, scale, BC_NUM_NEG(a), BC_NUM_NEG(b)); return; } len = bc_num_divReq(a, b, scale); BC_SIG_LOCK; // Initialize copies of the parameters. We want the length of the first // operand copy to be as big as the result because of the way the division // is implemented. bc_num_init(&cpa, len); bc_num_copy(&cpa, a); bc_num_createCopy(&cpb, b); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; len = b->len; // Like the above comment, we want the copy of the first parameter to be // larger than the second parameter. if (len > cpa.len) { bc_num_expand(&cpa, bc_vm_growSize(len, 2)); bc_num_extend(&cpa, (len - cpa.len) * BC_BASE_DIGS); } cpardx = BC_NUM_RDX_VAL_NP(cpa); cpa.scale = cpardx * BC_BASE_DIGS; // This is just setting up the scale in preparation for the division. bc_num_extend(&cpa, b->scale); cpardx = BC_NUM_RDX_VAL_NP(cpa) - BC_NUM_RDX(b->scale); BC_NUM_RDX_SET_NP(cpa, cpardx); cpa.scale = cpardx * BC_BASE_DIGS; // Once again, just setting things up, this time to match scale. if (scale > cpa.scale) { bc_num_extend(&cpa, scale); cpardx = BC_NUM_RDX_VAL_NP(cpa); cpa.scale = cpardx * BC_BASE_DIGS; } // Grow if necessary. if (cpa.cap == cpa.len) bc_num_expand(&cpa, bc_vm_growSize(cpa.len, 1)); // We want an extra zero in front to make things simpler. cpa.num[cpa.len++] = 0; // Still setting things up. Why all of these things are needed is not // something that can be easily explained, but it has to do with making the // actual algorithm easier to understand because it can assume a lot of // things. Thus, you should view all of this setup code as establishing // assumptions for bc_num_d_long(), where the actual division happens. // // But in short, this setup makes it so bc_num_d_long() can pretend the // numbers are integers. if (cpardx == cpa.len) cpa.len = bc_num_nonZeroLen(&cpa); if (BC_NUM_RDX_VAL_NP(cpb) == cpb.len) cpb.len = bc_num_nonZeroLen(&cpb); cpb.scale = 0; BC_NUM_RDX_SET_NP(cpb, 0); bc_num_d_long(&cpa, &cpb, c, scale); bc_num_retireMul(c, scale, BC_NUM_NEG(a), BC_NUM_NEG(b)); err: BC_SIG_MAYLOCK; bc_num_free(&cpb); bc_num_free(&cpa); BC_LONGJMP_CONT(vm); } /** * Implements divmod. This is the actual modulus function; since modulus * requires a division anyway, this returns the quotient and modulus. Either can * be thrown out as desired. * @param a The first operand. * @param b The second operand. * @param c The return parameter for the quotient. * @param d The return parameter for the modulus. * @param scale The current scale. * @param ts The scale that the operation should be done to. Yes, it's not * necessarily the same as scale, per the bc spec. */ static void bc_num_r(BcNum* a, BcNum* b, BcNum* restrict c, BcNum* restrict d, size_t scale, size_t ts) { BcNum temp; // realscale is meant to quiet a warning on GCC about longjmp() clobbering. // This one is real. size_t realscale; bool neg; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY if (BC_NUM_ZERO(b)) bc_err(BC_ERR_MATH_DIVIDE_BY_ZERO); if (BC_NUM_ZERO(a)) { bc_num_setToZero(c, ts); bc_num_setToZero(d, ts); return; } BC_SIG_LOCK; bc_num_init(&temp, d->cap); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; // Division. bc_num_d(a, b, c, scale); // We want an extra digit so we can safely truncate. if (scale) realscale = ts + 1; else realscale = scale; assert(BC_NUM_RDX_VALID(c)); assert(BC_NUM_RDX_VALID(b)); // Implement the rest of the (a - (a / b) * b) formula. bc_num_m(c, b, &temp, realscale); bc_num_sub(a, &temp, d, realscale); // Extend if necessary. if (ts > d->scale && BC_NUM_NONZERO(d)) bc_num_extend(d, ts - d->scale); neg = BC_NUM_NEG(d); bc_num_retireMul(d, ts, BC_NUM_NEG(a), BC_NUM_NEG(b)); d->rdx = BC_NUM_NEG_VAL(d, BC_NUM_NONZERO(d) ? neg : false); err: BC_SIG_MAYLOCK; bc_num_free(&temp); BC_LONGJMP_CONT(vm); } /** * Implements modulus/remainder. (Yes, I know they are different, but not in the * context of bc.) This is a BcNumBinOp function. * @param a The first operand. * @param b The second operand. * @param c The return parameter. * @param scale The current scale. */ static void bc_num_rem(BcNum* a, BcNum* b, BcNum* restrict c, size_t scale) { BcNum c1; size_t ts; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY ts = bc_vm_growSize(scale, b->scale); ts = BC_MAX(ts, a->scale); BC_SIG_LOCK; // Need a temp for the quotient. bc_num_init(&c1, bc_num_mulReq(a, b, ts)); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; bc_num_r(a, b, &c1, c, scale, ts); err: BC_SIG_MAYLOCK; bc_num_free(&c1); BC_LONGJMP_CONT(vm); } /** * Implements power (exponentiation). This is a BcNumBinOp function. * @param a The first operand. * @param b The second operand. * @param c The return parameter. * @param scale The current scale. */ static void bc_num_p(BcNum* a, BcNum* b, BcNum* restrict c, size_t scale) { BcNum copy, btemp; BcBigDig exp; // realscale is meant to quiet a warning on GCC about longjmp() clobbering. // This one is real. size_t powrdx, resrdx, realscale; bool neg; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY // This is here to silence a warning from GCC. #if BC_GCC btemp.len = 0; btemp.rdx = 0; btemp.num = NULL; #endif // BC_GCC if (BC_ERR(bc_num_nonInt(b, &btemp))) bc_err(BC_ERR_MATH_NON_INTEGER); assert(btemp.len == 0 || btemp.num != NULL); if (BC_NUM_ZERO(&btemp)) { bc_num_one(c); return; } if (BC_NUM_ZERO(a)) { if (BC_NUM_NEG_NP(btemp)) bc_err(BC_ERR_MATH_DIVIDE_BY_ZERO); bc_num_setToZero(c, scale); return; } if (BC_NUM_ONE(&btemp)) { if (!BC_NUM_NEG_NP(btemp)) bc_num_copy(c, a); else bc_num_inv(a, c, scale); return; } neg = BC_NUM_NEG_NP(btemp); BC_NUM_NEG_CLR_NP(btemp); exp = bc_num_bigdig(&btemp); BC_SIG_LOCK; bc_num_createCopy(©, a); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; // If this is true, then we do not have to do a division, and we need to // set scale accordingly. if (!neg) { size_t max = BC_MAX(scale, a->scale), scalepow; scalepow = bc_num_mulOverflow(a->scale, exp); realscale = BC_MIN(scalepow, max); } else realscale = scale; // This is only implementing the first exponentiation by squaring, until it // reaches the first time where the square is actually used. for (powrdx = a->scale; !(exp & 1); exp >>= 1) { powrdx <<= 1; assert(BC_NUM_RDX_VALID_NP(copy)); bc_num_mul(©, ©, ©, powrdx); } // Make c a copy of copy for the purpose of saving the squares that should // be saved. bc_num_copy(c, ©); resrdx = powrdx; // Now finish the exponentiation by squaring, this time saving the squares // as necessary. while (exp >>= 1) { powrdx <<= 1; assert(BC_NUM_RDX_VALID_NP(copy)); bc_num_mul(©, ©, ©, powrdx); // If this is true, we want to save that particular square. This does // that by multiplying c with copy. if (exp & 1) { resrdx += powrdx; assert(BC_NUM_RDX_VALID(c)); assert(BC_NUM_RDX_VALID_NP(copy)); bc_num_mul(c, ©, c, resrdx); } } // Invert if necessary. if (neg) bc_num_inv(c, c, realscale); // Truncate if necessary. if (c->scale > realscale) bc_num_truncate(c, c->scale - realscale); bc_num_clean(c); err: BC_SIG_MAYLOCK; bc_num_free(©); BC_LONGJMP_CONT(vm); } #if BC_ENABLE_EXTRA_MATH /** * Implements the places operator. This is a BcNumBinOp function. * @param a The first operand. * @param b The second operand. * @param c The return parameter. * @param scale The current scale. */ static void bc_num_place(BcNum* a, BcNum* b, BcNum* restrict c, size_t scale) { BcBigDig val; BC_UNUSED(scale); val = bc_num_intop(a, b, c); // Just truncate or extend as appropriate. if (val < c->scale) bc_num_truncate(c, c->scale - val); else if (val > c->scale) bc_num_extend(c, val - c->scale); } /** * Implements the left shift operator. This is a BcNumBinOp function. */ static void bc_num_left(BcNum* a, BcNum* b, BcNum* restrict c, size_t scale) { BcBigDig val; BC_UNUSED(scale); val = bc_num_intop(a, b, c); bc_num_shiftLeft(c, (size_t) val); } /** * Implements the right shift operator. This is a BcNumBinOp function. */ static void bc_num_right(BcNum* a, BcNum* b, BcNum* restrict c, size_t scale) { BcBigDig val; BC_UNUSED(scale); val = bc_num_intop(a, b, c); if (BC_NUM_ZERO(c)) return; bc_num_shiftRight(c, (size_t) val); } #endif // BC_ENABLE_EXTRA_MATH /** * Prepares for, and calls, a binary operator function. This is probably the * most important function in the entire file because it establishes assumptions * that make the rest of the code so easy. Those assumptions include: * * - a is not the same pointer as c. * - b is not the same pointer as c. * - there is enough room in c for the result. * * Without these, this whole function would basically have to be duplicated for * *all* binary operators. * * @param a The first operand. * @param b The second operand. * @param c The return parameter. * @param scale The current scale. * @param req The number of limbs needed to fit the result. */ static void bc_num_binary(BcNum* a, BcNum* b, BcNum* c, size_t scale, BcNumBinOp op, size_t req) { BcNum* ptr_a; BcNum* ptr_b; BcNum num2; #if BC_ENABLE_LIBRARY BcVm* vm = NULL; #endif // BC_ENABLE_LIBRARY assert(a != NULL && b != NULL && c != NULL && op != NULL); assert(BC_NUM_RDX_VALID(a)); assert(BC_NUM_RDX_VALID(b)); BC_SIG_LOCK; ptr_a = c == a ? &num2 : a; ptr_b = c == b ? &num2 : b; // Actually reallocate. If we don't reallocate, we want to expand at the // very least. if (c == a || c == b) { #if BC_ENABLE_LIBRARY vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY // NOLINTNEXTLINE memcpy(&num2, c, sizeof(BcNum)); bc_num_init(c, req); // Must prepare for cleanup. We want this here so that locals that got // set stay set since a longjmp() is not guaranteed to preserve locals. BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; } else { BC_SIG_UNLOCK; bc_num_expand(c, req); } // It is okay for a and b to be the same. If a binary operator function does // need them to be different, the binary operator function is responsible // for that. // Call the actual binary operator function. op(ptr_a, ptr_b, c, scale); assert(!BC_NUM_NEG(c) || BC_NUM_NONZERO(c)); assert(BC_NUM_RDX_VAL(c) <= c->len || !c->len); assert(BC_NUM_RDX_VALID(c)); assert(!c->len || c->num[c->len - 1] || BC_NUM_RDX_VAL(c) == c->len); err: // Cleanup only needed if we initialized c to a new number. if (c == a || c == b) { BC_SIG_MAYLOCK; bc_num_free(&num2); BC_LONGJMP_CONT(vm); } } /** * Tests a number string for validity. This function has a history; I originally * wrote it because I did not trust my parser. Over time, however, I came to * trust it, so I was able to relegate this function to debug builds only, and I * used it in assert()'s. But then I created the library, and well, I can't * trust users, so I reused this for yelling at users. * @param val The string to check to see if it's a valid number string. * @return True if the string is a valid number string, false otherwise. */ bool bc_num_strValid(const char* restrict val) { bool radix = false; size_t i, len = strlen(val); // Notice that I don't check if there is a negative sign. That is not part // of a valid number, except in the library. The library-specific code takes // care of that part. // Nothing in the string is okay. if (!len) return true; // Loop through the characters. for (i = 0; i < len; ++i) { BcDig c = val[i]; // If we have found a radix point... if (c == '.') { // We don't allow two radices. if (radix) return false; radix = true; continue; } // We only allow digits and uppercase letters. if (!(isdigit(c) || isupper(c))) return false; } return true; } /** * Parses one character and returns the digit that corresponds to that * character according to the base. * @param c The character to parse. * @param base The base. * @return The character as a digit. */ static BcBigDig bc_num_parseChar(char c, size_t base) { assert(isupper(c) || isdigit(c)); // If a letter... if (isupper(c)) { #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY // This returns the digit that directly corresponds with the letter. c = BC_NUM_NUM_LETTER(c); // If the digit is greater than the base, we clamp. if (BC_DIGIT_CLAMP) { c = ((size_t) c) >= base ? (char) base - 1 : c; } } // Straight convert the digit to a number. else c -= '0'; return (BcBigDig) (uchar) c; } /** * Parses a string as a decimal number. This is separate because it's going to * be the most used, and it can be heavily optimized for decimal only. * @param n The number to parse into and return. Must be preallocated. * @param val The string to parse. */ static void bc_num_parseDecimal(BcNum* restrict n, const char* restrict val) { size_t len, i, temp, mod; const char* ptr; bool zero = true, rdx; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY // Eat leading zeroes. for (i = 0; val[i] == '0'; ++i) { continue; } val += i; assert(!val[0] || isalnum(val[0]) || val[0] == '.'); // All 0's. We can just return, since this procedure expects a virgin // (already 0) BcNum. if (!val[0]) return; // The length of the string is the length of the number, except it might be // one bigger because of a decimal point. len = strlen(val); // Find the location of the decimal point. ptr = strchr(val, '.'); rdx = (ptr != NULL); // We eat leading zeroes again. These leading zeroes are different because // they will come after the decimal point if they exist, and since that's // the case, they must be preserved. for (i = 0; i < len && (zero = (val[i] == '0' || val[i] == '.')); ++i) { continue; } // Set the scale of the number based on the location of the decimal point. // The casts to uintptr_t is to ensure that bc does not hit undefined // behavior when doing math on the values. n->scale = (size_t) (rdx * (((uintptr_t) (val + len)) - (((uintptr_t) ptr) + 1))); // Set rdx. BC_NUM_RDX_SET(n, BC_NUM_RDX(n->scale)); // Calculate length. First, the length of the integer, then the number of // digits in the last limb, then the length. i = len - (ptr == val ? 0 : i) - rdx; temp = BC_NUM_ROUND_POW(i); mod = n->scale % BC_BASE_DIGS; i = mod ? BC_BASE_DIGS - mod : 0; n->len = ((temp + i) / BC_BASE_DIGS); // Expand and zero. The plus extra is in case the lack of clamping causes // the number to overflow the original bounds. bc_num_expand(n, n->len + !BC_DIGIT_CLAMP); // NOLINTNEXTLINE memset(n->num, 0, BC_NUM_SIZE(n->len + !BC_DIGIT_CLAMP)); if (zero) { // I think I can set rdx directly to zero here because n should be a // new number with sign set to false. n->len = n->rdx = 0; } else { // There is actually stuff to parse if we make it here. Yay... BcBigDig exp, pow; assert(i <= BC_NUM_BIGDIG_MAX); // The exponent and power. exp = (BcBigDig) i; pow = bc_num_pow10[exp]; // Parse loop. We parse backwards because numbers are stored little // endian. for (i = len - 1; i < len; --i, ++exp) { char c = val[i]; // Skip the decimal point. if (c == '.') exp -= 1; else { // The index of the limb. size_t idx = exp / BC_BASE_DIGS; BcBigDig dig; if (isupper(c)) { // Clamp for the base. if (!BC_DIGIT_CLAMP) c = BC_NUM_NUM_LETTER(c); else c = 9; } else c -= '0'; // Add the digit to the limb. This takes care of overflow from // lack of clamping. dig = ((BcBigDig) n->num[idx]) + ((BcBigDig) c) * pow; if (dig >= BC_BASE_POW) { // We cannot go over BC_BASE_POW with clamping. assert(!BC_DIGIT_CLAMP); n->num[idx + 1] = (BcDig) (dig / BC_BASE_POW); n->num[idx] = (BcDig) (dig % BC_BASE_POW); assert(n->num[idx] >= 0 && n->num[idx] < BC_BASE_POW); assert(n->num[idx + 1] >= 0 && n->num[idx + 1] < BC_BASE_POW); } else { n->num[idx] = (BcDig) dig; assert(n->num[idx] >= 0 && n->num[idx] < BC_BASE_POW); } // Adjust the power and exponent. if ((exp + 1) % BC_BASE_DIGS == 0) pow = 1; else pow *= BC_BASE; } } } // Make sure to add one to the length if needed from lack of clamping. n->len += (!BC_DIGIT_CLAMP && n->num[n->len] != 0); } /** * Parse a number in any base (besides decimal). * @param n The number to parse into and return. Must be preallocated. * @param val The string to parse. * @param base The base to parse as. */ static void bc_num_parseBase(BcNum* restrict n, const char* restrict val, BcBigDig base) { BcNum temp, mult1, mult2, result1, result2; BcNum* m1; BcNum* m2; BcNum* ptr; char c = 0; bool zero = true; BcBigDig v; size_t digs, len = strlen(val); // This is volatile to quiet a warning on GCC about longjmp() clobbering. volatile size_t i; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY // If zero, just return because the number should be virgin (already 0). for (i = 0; zero && i < len; ++i) { zero = (val[i] == '.' || val[i] == '0'); } if (zero) return; BC_SIG_LOCK; bc_num_init(&temp, BC_NUM_BIGDIG_LOG10); bc_num_init(&mult1, BC_NUM_BIGDIG_LOG10); BC_SETJMP_LOCKED(vm, int_err); BC_SIG_UNLOCK; // We split parsing into parsing the integer and parsing the fractional // part. // Parse the integer part. This is the easy part because we just multiply // the number by the base, then add the digit. for (i = 0; i < len && (c = val[i]) && c != '.'; ++i) { // Convert the character to a digit. v = bc_num_parseChar(c, base); // Multiply the number. bc_num_mulArray(n, base, &mult1); // Convert the digit to a number and add. bc_num_bigdig2num(&temp, v); bc_num_add(&mult1, &temp, n, 0); } // If this condition is true, then we are done. We still need to do cleanup // though. if (i == len && !val[i]) goto int_err; // If we get here, we *must* be at the radix point. assert(val[i] == '.'); BC_SIG_LOCK; // Unset the jump to reset in for these new initializations. BC_UNSETJMP(vm); bc_num_init(&mult2, BC_NUM_BIGDIG_LOG10); bc_num_init(&result1, BC_NUM_DEF_SIZE); bc_num_init(&result2, BC_NUM_DEF_SIZE); bc_num_one(&mult1); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; // Pointers for easy switching. m1 = &mult1; m2 = &mult2; // Parse the fractional part. This is the hard part. for (i += 1, digs = 0; i < len && (c = val[i]); ++i, ++digs) { size_t rdx; // Convert the character to a digit. v = bc_num_parseChar(c, base); // We keep growing result2 according to the base because the more digits // after the radix, the more significant the digits close to the radix // should be. bc_num_mulArray(&result1, base, &result2); // Convert the digit to a number. bc_num_bigdig2num(&temp, v); // Add the digit into the fraction part. bc_num_add(&result2, &temp, &result1, 0); // Keep growing m1 and m2 for use after the loop. bc_num_mulArray(m1, base, m2); rdx = BC_NUM_RDX_VAL(m2); if (m2->len < rdx) m2->len = rdx; // Switch. ptr = m1; m1 = m2; m2 = ptr; } // This one cannot be a divide by 0 because mult starts out at 1, then is // multiplied by base, and base cannot be 0, so mult cannot be 0. And this // is the reason we keep growing m1 and m2; this division is what converts // the parsed fractional part from an integer to a fractional part. bc_num_div(&result1, m1, &result2, digs * 2); // Pretruncate. bc_num_truncate(&result2, digs); // The final add of the integer part to the fractional part. bc_num_add(n, &result2, n, digs); // Basic cleanup. if (BC_NUM_NONZERO(n)) { if (n->scale < digs) bc_num_extend(n, digs - n->scale); } else bc_num_zero(n); err: BC_SIG_MAYLOCK; bc_num_free(&result2); bc_num_free(&result1); bc_num_free(&mult2); int_err: BC_SIG_MAYLOCK; bc_num_free(&mult1); bc_num_free(&temp); BC_LONGJMP_CONT(vm); } /** * Prints a backslash+newline combo if the number of characters needs it. This * is really a convenience function. */ static inline void bc_num_printNewline(void) { #if !BC_ENABLE_LIBRARY if (vm->nchars >= vm->line_len - 1 && vm->line_len) { bc_vm_putchar('\\', bc_flush_none); bc_vm_putchar('\n', bc_flush_err); } #endif // !BC_ENABLE_LIBRARY } /** * Prints a character after a backslash+newline, if needed. * @param c The character to print. * @param bslash Whether to print a backslash+newline. */ static void bc_num_putchar(int c, bool bslash) { if (c != '\n' && bslash) bc_num_printNewline(); bc_vm_putchar(c, bc_flush_save); } #if !BC_ENABLE_LIBRARY /** * Prints a character for a number's digit. This is for printing for dc's P * command. This function does not need to worry about radix points. This is a * BcNumDigitOp. * @param n The "digit" to print. * @param len The "length" of the digit, or number of characters that will * need to be printed for the digit. * @param rdx True if a decimal (radix) point should be printed. * @param bslash True if a backslash+newline should be printed if the character * limit for the line is reached, false otherwise. */ static void bc_num_printChar(size_t n, size_t len, bool rdx, bool bslash) { BC_UNUSED(rdx); BC_UNUSED(len); BC_UNUSED(bslash); assert(len == 1); bc_vm_putchar((uchar) n, bc_flush_save); } #endif // !BC_ENABLE_LIBRARY /** * Prints a series of characters for large bases. This is for printing in bases * above hexadecimal. This is a BcNumDigitOp. * @param n The "digit" to print. * @param len The "length" of the digit, or number of characters that will * need to be printed for the digit. * @param rdx True if a decimal (radix) point should be printed. * @param bslash True if a backslash+newline should be printed if the character * limit for the line is reached, false otherwise. */ static void bc_num_printDigits(size_t n, size_t len, bool rdx, bool bslash) { size_t exp, pow; // If needed, print the radix; otherwise, print a space to separate digits. bc_num_putchar(rdx ? '.' : ' ', true); // Calculate the exponent and power. for (exp = 0, pow = 1; exp < len - 1; ++exp, pow *= BC_BASE) { continue; } // Print each character individually. for (exp = 0; exp < len; pow /= BC_BASE, ++exp) { // The individual subdigit. size_t dig = n / pow; // Take the subdigit away. n -= dig * pow; // Print the subdigit. bc_num_putchar(((uchar) dig) + '0', bslash || exp != len - 1); } } /** * Prints a character for a number's digit. This is for printing in bases for * hexadecimal and below because they always print only one character at a time. * This is a BcNumDigitOp. * @param n The "digit" to print. * @param len The "length" of the digit, or number of characters that will * need to be printed for the digit. * @param rdx True if a decimal (radix) point should be printed. * @param bslash True if a backslash+newline should be printed if the character * limit for the line is reached, false otherwise. */ static void bc_num_printHex(size_t n, size_t len, bool rdx, bool bslash) { BC_UNUSED(len); BC_UNUSED(bslash); assert(len == 1); if (rdx) bc_num_putchar('.', true); bc_num_putchar(bc_num_hex_digits[n], bslash); } /** * Prints a decimal number. This is specially written for optimization since * this will be used the most and because bc's numbers are already in decimal. * @param n The number to print. * @param newline Whether to print backslash+newlines on long enough lines. */ static void bc_num_printDecimal(const BcNum* restrict n, bool newline) { size_t i, j, rdx = BC_NUM_RDX_VAL(n); bool zero = true; size_t buffer[BC_BASE_DIGS]; // Print loop. for (i = n->len - 1; i < n->len; --i) { BcDig n9 = n->num[i]; size_t temp; bool irdx = (i == rdx - 1); // Calculate the number of digits in the limb. zero = (zero & !irdx); temp = n->scale % BC_BASE_DIGS; temp = i || !temp ? 0 : BC_BASE_DIGS - temp; // NOLINTNEXTLINE memset(buffer, 0, BC_BASE_DIGS * sizeof(size_t)); // Fill the buffer with individual digits. for (j = 0; n9 && j < BC_BASE_DIGS; ++j) { buffer[j] = ((size_t) n9) % BC_BASE; n9 /= BC_BASE; } // Print the digits in the buffer. for (j = BC_BASE_DIGS - 1; j < BC_BASE_DIGS && j >= temp; --j) { // Figure out whether to print the decimal point. bool print_rdx = (irdx & (j == BC_BASE_DIGS - 1)); // The zero variable helps us skip leading zero digits in the limb. zero = (zero && buffer[j] == 0); if (!zero) { // While the first three arguments should be self-explanatory, // the last needs explaining. I don't want to print a newline // when the last digit to be printed could take the place of the // backslash rather than being pushed, as a single character, to // the next line. That's what that last argument does for bc. bc_num_printHex(buffer[j], 1, print_rdx, !newline || (j > temp || i != 0)); } } } } #if BC_ENABLE_EXTRA_MATH /** * Prints a number in scientific or engineering format. When doing this, we are * always printing in decimal. * @param n The number to print. * @param eng True if we are in engineering mode. * @param newline Whether to print backslash+newlines on long enough lines. */ static void bc_num_printExponent(const BcNum* restrict n, bool eng, bool newline) { size_t places, mod, nrdx = BC_NUM_RDX_VAL(n); bool neg = (n->len <= nrdx); BcNum temp, exp; BcDig digs[BC_NUM_BIGDIG_LOG10]; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY BC_SIG_LOCK; bc_num_createCopy(&temp, n); BC_SETJMP_LOCKED(vm, exit); BC_SIG_UNLOCK; // We need to calculate the exponents, and they change based on whether the // number is all fractional or not, obviously. if (neg) { // Figure out the negative power of 10. places = bc_num_negPow10(n); // Figure out how many digits mod 3 there are (important for // engineering mode). mod = places % 3; // Calculate places if we are in engineering mode. if (eng && mod != 0) places += 3 - mod; // Shift the temp to the right place. bc_num_shiftLeft(&temp, places); } else { // This is the number of digits that we are supposed to put behind the // decimal point. places = bc_num_intDigits(n) - 1; // Calculate the true number based on whether engineering mode is // activated. mod = places % 3; if (eng && mod != 0) places -= 3 - (3 - mod); // Shift the temp to the right place. bc_num_shiftRight(&temp, places); } // Print the shifted number. bc_num_printDecimal(&temp, newline); // Print the e. bc_num_putchar('e', !newline); // Need to explicitly print a zero exponent. if (!places) { bc_num_printHex(0, 1, false, !newline); goto exit; } // Need to print sign for the exponent. if (neg) bc_num_putchar('-', true); // Create a temporary for the exponent... bc_num_setup(&exp, digs, BC_NUM_BIGDIG_LOG10); bc_num_bigdig2num(&exp, (BcBigDig) places); /// ..and print it. bc_num_printDecimal(&exp, newline); exit: BC_SIG_MAYLOCK; bc_num_free(&temp); BC_LONGJMP_CONT(vm); } #endif // BC_ENABLE_EXTRA_MATH /** * Takes a number with limbs with base BC_BASE_POW and converts the limb at the * given index to base @a pow, where @a pow is obase^N. * @param n The number to convert. * @param rem BC_BASE_POW - @a pow. * @param pow The power of obase we will convert the number to. * @param idx The index of the number to start converting at. Doing the * conversion is O(n^2); we have to sweep through starting at the * least significant limb. */ static void bc_num_printFixup(BcNum* restrict n, BcBigDig rem, BcBigDig pow, size_t idx) { size_t i, len = n->len - idx; BcBigDig acc; BcDig* a = n->num + idx; // Ignore if there's just one limb left. This is the part that requires the // extra loop after the one calling this function in bc_num_printPrepare(). if (len < 2) return; // Loop through the remaining limbs and convert. We start at the second limb // because we pull the value from the previous one as well. for (i = len - 1; i > 0; --i) { // Get the limb and add it to the previous, along with multiplying by // the remainder because that's the proper overflow. "acc" means // "accumulator," by the way. acc = ((BcBigDig) a[i]) * rem + ((BcBigDig) a[i - 1]); // Store a value in base pow in the previous limb. a[i - 1] = (BcDig) (acc % pow); // Divide by the base and accumulate the remaining value in the limb. acc /= pow; acc += (BcBigDig) a[i]; // If the accumulator is greater than the base... if (acc >= BC_BASE_POW) { // Do we need to grow? if (i == len - 1) { // Grow. len = bc_vm_growSize(len, 1); bc_num_expand(n, bc_vm_growSize(len, idx)); // Update the pointer because it may have moved. a = n->num + idx; // Zero out the last limb. a[len - 1] = 0; } // Overflow into the next limb since we are over the base. a[i + 1] += acc / BC_BASE_POW; acc %= BC_BASE_POW; } assert(acc < BC_BASE_POW); // Set the limb. a[i] = (BcDig) acc; } // We may have grown the number, so adjust the length. n->len = len + idx; } /** * Prepares a number for printing in a base that does not have BC_BASE_POW as a * power. This basically converts the number from having limbs of base * BC_BASE_POW to limbs of pow, where pow is obase^N. * @param n The number to prepare for printing. * @param rem The remainder of BC_BASE_POW when divided by a power of the base. * @param pow The power of the base. */ static void bc_num_printPrepare(BcNum* restrict n, BcBigDig rem, BcBigDig pow) { size_t i; // Loop from the least significant limb to the most significant limb and // convert limbs in each pass. for (i = 0; i < n->len; ++i) { bc_num_printFixup(n, rem, pow, i); } // bc_num_printFixup() does not do everything it is supposed to, so we do // the last bit of cleanup here. That cleanup is to ensure that each limb // is less than pow and to expand the number to fit new limbs as necessary. for (i = 0; i < n->len; ++i) { assert(pow == ((BcBigDig) ((BcDig) pow))); // If the limb needs fixing... if (n->num[i] >= (BcDig) pow) { // Do we need to grow? if (i + 1 == n->len) { // Grow the number. n->len = bc_vm_growSize(n->len, 1); bc_num_expand(n, n->len); // Without this, we might use uninitialized data. n->num[i + 1] = 0; } assert(pow < BC_BASE_POW); // Overflow into the next limb. n->num[i + 1] += n->num[i] / ((BcDig) pow); n->num[i] %= (BcDig) pow; } } } static void bc_num_printNum(BcNum* restrict n, BcBigDig base, size_t len, BcNumDigitOp print, bool newline) { BcVec stack; BcNum intp, fracp1, fracp2, digit, flen1, flen2; BcNum* n1; BcNum* n2; BcNum* temp; BcBigDig dig = 0, acc, exp; BcBigDig* ptr; size_t i, j, nrdx, idigits; bool radix; BcDig digit_digs[BC_NUM_BIGDIG_LOG10 + 1]; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY assert(base > 1); // Easy case. Even with scale, we just print this. if (BC_NUM_ZERO(n)) { print(0, len, false, !newline); return; } // This function uses an algorithm that Stefan Esser came // up with to print the integer part of a number. What it does is convert // intp into a number of the specified base, but it does it directly, // instead of just doing a series of divisions and printing the remainders // in reverse order. // // Let me explain in a bit more detail: // // The algorithm takes the current least significant limb (after intp has // been converted to an integer) and the next to least significant limb, and // it converts the least significant limb into one of the specified base, // putting any overflow into the next to least significant limb. It iterates // through the whole number, from least significant to most significant, // doing this conversion. At the end of that iteration, the least // significant limb is converted, but the others are not, so it iterates // again, starting at the next to least significant limb. It keeps doing // that conversion, skipping one more limb than the last time, until all // limbs have been converted. Then it prints them in reverse order. // // That is the gist of the algorithm. It leaves out several things, such as // the fact that limbs are not always converted into the specified base, but // into something close, basically a power of the specified base. In // Stefan's words, "You could consider BcDigs to be of base 10^BC_BASE_DIGS // in the normal case and obase^N for the largest value of N that satisfies // obase^N <= 10^BC_BASE_DIGS. [This means that] the result is not in base // "obase", but in base "obase^N", which happens to be printable as a number // of base "obase" without consideration for neighbouring BcDigs." This fact // is what necessitates the existence of the loop later in this function. // // The conversion happens in bc_num_printPrepare() where the outer loop // happens and bc_num_printFixup() where the inner loop, or actual // conversion, happens. In other words, bc_num_printPrepare() is where the // loop that starts at the least significant limb and goes to the most // significant limb. Then, on every iteration of its loop, it calls // bc_num_printFixup(), which has the inner loop of actually converting // the limbs it passes into limbs of base obase^N rather than base // BC_BASE_POW. nrdx = BC_NUM_RDX_VAL(n); BC_SIG_LOCK; // The stack is what allows us to reverse the digits for printing. bc_vec_init(&stack, sizeof(BcBigDig), BC_DTOR_NONE); bc_num_init(&fracp1, nrdx); // intp will be the "integer part" of the number, so copy it. bc_num_createCopy(&intp, n); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; // Make intp an integer. bc_num_truncate(&intp, intp.scale); // Get the fractional part out. bc_num_sub(n, &intp, &fracp1, 0); // If the base is not the same as the last base used for printing, we need // to update the cached exponent and power. Yes, we cache the values of the // exponent and power. That is to prevent us from calculating them every // time because printing will probably happen multiple times on the same // base. if (base != vm->last_base) { vm->last_pow = 1; vm->last_exp = 0; // Calculate the exponent and power. while (vm->last_pow * base <= BC_BASE_POW) { vm->last_pow *= base; vm->last_exp += 1; } // Also, the remainder and base itself. vm->last_rem = BC_BASE_POW - vm->last_pow; vm->last_base = base; } exp = vm->last_exp; // If vm->last_rem is 0, then the base we are printing in is a divisor of // BC_BASE_POW, which is the easy case because it means that BC_BASE_POW is // a power of obase, and no conversion is needed. If it *is* 0, then we have // the hard case, and we have to prepare the number for the base. if (vm->last_rem != 0) { bc_num_printPrepare(&intp, vm->last_rem, vm->last_pow); } // After the conversion comes the surprisingly easy part. From here on out, // this is basically naive code that I wrote, adjusted for the larger bases. // Fill the stack of digits for the integer part. for (i = 0; i < intp.len; ++i) { // Get the limb. acc = (BcBigDig) intp.num[i]; // Turn the limb into digits of base obase. for (j = 0; j < exp && (i < intp.len - 1 || acc != 0); ++j) { // This condition is true if we are not at the last digit. if (j != exp - 1) { dig = acc % base; acc /= base; } else { dig = acc; acc = 0; } assert(dig < base); // Push the digit onto the stack. bc_vec_push(&stack, &dig); } assert(acc == 0); } // Go through the stack backwards and print each digit. for (i = 0; i < stack.len; ++i) { ptr = bc_vec_item_rev(&stack, i); assert(ptr != NULL); // While the first three arguments should be self-explanatory, the last // needs explaining. I don't want to print a backslash+newline when the // last digit to be printed could take the place of the backslash rather // than being pushed, as a single character, to the next line. That's // what that last argument does for bc. // // First, it needs to check if newlines are completely disabled. If they // are not disabled, it needs to check the next part. // // If the number has a scale, then because we are printing just the // integer part, there will be at least two more characters (a radix // point plus at least one digit). So if there is a scale, a backslash // is necessary. // // Finally, the last condition checks to see if we are at the end of the // stack. If we are *not* (i.e., the index is not one less than the // stack length), then a backslash is necessary because there is at // least one more character for at least one more digit). Otherwise, if // the index is equal to one less than the stack length, we want to // disable backslash printing. // // The function that prints bases 17 and above will take care of not // printing a backslash in the right case. print(*ptr, len, false, !newline || (n->scale != 0 || i < stack.len - 1)); } // We are done if there is no fractional part. if (!n->scale) goto err; BC_SIG_LOCK; // Reset the jump because some locals are changing. BC_UNSETJMP(vm); bc_num_init(&fracp2, nrdx); bc_num_setup(&digit, digit_digs, sizeof(digit_digs) / sizeof(BcDig)); bc_num_init(&flen1, BC_NUM_BIGDIG_LOG10); bc_num_init(&flen2, BC_NUM_BIGDIG_LOG10); BC_SETJMP_LOCKED(vm, frac_err); BC_SIG_UNLOCK; bc_num_one(&flen1); radix = true; // Pointers for easy switching. n1 = &flen1; n2 = &flen2; fracp2.scale = n->scale; BC_NUM_RDX_SET_NP(fracp2, BC_NUM_RDX(fracp2.scale)); // As long as we have not reached the scale of the number, keep printing. while ((idigits = bc_num_intDigits(n1)) <= n->scale) { // These numbers will keep growing. bc_num_expand(&fracp2, fracp1.len + 1); bc_num_mulArray(&fracp1, base, &fracp2); nrdx = BC_NUM_RDX_VAL_NP(fracp2); // Ensure an invariant. if (fracp2.len < nrdx) fracp2.len = nrdx; // fracp is guaranteed to be non-negative and small enough. dig = bc_num_bigdig2(&fracp2); // Convert the digit to a number and subtract it from the number. bc_num_bigdig2num(&digit, dig); bc_num_sub(&fracp2, &digit, &fracp1, 0); // While the first three arguments should be self-explanatory, the last // needs explaining. I don't want to print a newline when the last digit // to be printed could take the place of the backslash rather than being // pushed, as a single character, to the next line. That's what that // last argument does for bc. print(dig, len, radix, !newline || idigits != n->scale); // Update the multipliers. bc_num_mulArray(n1, base, n2); radix = false; // Switch. temp = n1; n1 = n2; n2 = temp; } frac_err: BC_SIG_MAYLOCK; bc_num_free(&flen2); bc_num_free(&flen1); bc_num_free(&fracp2); err: BC_SIG_MAYLOCK; bc_num_free(&fracp1); bc_num_free(&intp); bc_vec_free(&stack); BC_LONGJMP_CONT(vm); } /** * Prints a number in the specified base, or rather, figures out which function * to call to print the number in the specified base and calls it. * @param n The number to print. * @param base The base to print in. * @param newline Whether to print backslash+newlines on long enough lines. */ static void bc_num_printBase(BcNum* restrict n, BcBigDig base, bool newline) { size_t width; BcNumDigitOp print; bool neg = BC_NUM_NEG(n); // Clear the sign because it makes the actual printing easier when we have // to do math. BC_NUM_NEG_CLR(n); // Bases at hexadecimal and below are printed as one character, larger bases // are printed as a series of digits separated by spaces. if (base <= BC_NUM_MAX_POSIX_IBASE) { width = 1; print = bc_num_printHex; } else { assert(base <= BC_BASE_POW); width = bc_num_log10(base - 1); print = bc_num_printDigits; } // Print. bc_num_printNum(n, base, width, print, newline); // Reset the sign. n->rdx = BC_NUM_NEG_VAL(n, neg); } #if !BC_ENABLE_LIBRARY void bc_num_stream(BcNum* restrict n) { bc_num_printNum(n, BC_NUM_STREAM_BASE, 1, bc_num_printChar, false); } #endif // !BC_ENABLE_LIBRARY void bc_num_setup(BcNum* restrict n, BcDig* restrict num, size_t cap) { assert(n != NULL); n->num = num; n->cap = cap; bc_num_zero(n); } void bc_num_init(BcNum* restrict n, size_t req) { BcDig* num; BC_SIG_ASSERT_LOCKED; assert(n != NULL); // BC_NUM_DEF_SIZE is set to be about the smallest allocation size that // malloc() returns in practice, so just use it. req = req >= BC_NUM_DEF_SIZE ? req : BC_NUM_DEF_SIZE; // If we can't use a temp, allocate. if (req != BC_NUM_DEF_SIZE) num = bc_vm_malloc(BC_NUM_SIZE(req)); else { num = bc_vm_getTemp() == NULL ? bc_vm_malloc(BC_NUM_SIZE(req)) : bc_vm_takeTemp(); } bc_num_setup(n, num, req); } void bc_num_clear(BcNum* restrict n) { n->num = NULL; n->cap = 0; } void bc_num_free(void* num) { BcNum* n = (BcNum*) num; BC_SIG_ASSERT_LOCKED; assert(n != NULL); if (n->cap == BC_NUM_DEF_SIZE) bc_vm_addTemp(n->num); else free(n->num); } void bc_num_copy(BcNum* d, const BcNum* s) { assert(d != NULL && s != NULL); if (d == s) return; bc_num_expand(d, s->len); d->len = s->len; // I can just copy directly here because the sign *and* rdx will be // properly preserved. d->rdx = s->rdx; d->scale = s->scale; // NOLINTNEXTLINE memcpy(d->num, s->num, BC_NUM_SIZE(d->len)); } void bc_num_createCopy(BcNum* d, const BcNum* s) { BC_SIG_ASSERT_LOCKED; bc_num_init(d, s->len); bc_num_copy(d, s); } void bc_num_createFromBigdig(BcNum* restrict n, BcBigDig val) { BC_SIG_ASSERT_LOCKED; bc_num_init(n, BC_NUM_BIGDIG_LOG10); bc_num_bigdig2num(n, val); } size_t bc_num_scale(const BcNum* restrict n) { return n->scale; } size_t bc_num_len(const BcNum* restrict n) { size_t len = n->len; // Always return at least 1. if (BC_NUM_ZERO(n)) return n->scale ? n->scale : 1; // If this is true, there is no integer portion of the number. if (BC_NUM_RDX_VAL(n) == len) { // We have to take into account the fact that some of the digits right // after the decimal could be zero. If that is the case, we need to // ignore them until we hit the first non-zero digit. size_t zero, scale; // The number of limbs with non-zero digits. len = bc_num_nonZeroLen(n); // Get the number of digits in the last limb. scale = n->scale % BC_BASE_DIGS; scale = scale ? scale : BC_BASE_DIGS; // Get the number of zero digits. zero = bc_num_zeroDigits(n->num + len - 1); // Calculate the true length. len = len * BC_BASE_DIGS - zero - (BC_BASE_DIGS - scale); } // Otherwise, count the number of int digits and return that plus the scale. else len = bc_num_intDigits(n) + n->scale; return len; } void bc_num_parse(BcNum* restrict n, const char* restrict val, BcBigDig base) { #if BC_DEBUG #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY #endif // BC_DEBUG assert(n != NULL && val != NULL && base); assert(base >= BC_NUM_MIN_BASE && base <= vm->maxes[BC_PROG_GLOBALS_IBASE]); assert(bc_num_strValid(val)); // A one character number is *always* parsed as though the base was the // maximum allowed ibase, per the bc spec. if (!val[1]) { BcBigDig dig = bc_num_parseChar(val[0], BC_NUM_MAX_LBASE); bc_num_bigdig2num(n, dig); } else if (base == BC_BASE) bc_num_parseDecimal(n, val); else bc_num_parseBase(n, val, base); assert(BC_NUM_RDX_VALID(n)); } void bc_num_print(BcNum* restrict n, BcBigDig base, bool newline) { assert(n != NULL); assert(BC_ENABLE_EXTRA_MATH || base >= BC_NUM_MIN_BASE); // We may need a newline, just to start. bc_num_printNewline(); if (BC_NUM_NONZERO(n)) { #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY // Print the sign. if (BC_NUM_NEG(n)) bc_num_putchar('-', true); // Print the leading zero if necessary. We don't print when using // scientific or engineering modes. if (BC_Z && BC_NUM_RDX_VAL(n) == n->len && base != 0 && base != 1) { bc_num_printHex(0, 1, false, !newline); } } // Short-circuit 0. if (BC_NUM_ZERO(n)) bc_num_printHex(0, 1, false, !newline); else if (base == BC_BASE) bc_num_printDecimal(n, newline); #if BC_ENABLE_EXTRA_MATH else if (base == 0 || base == 1) { bc_num_printExponent(n, base != 0, newline); } #endif // BC_ENABLE_EXTRA_MATH else bc_num_printBase(n, base, newline); if (newline) bc_num_putchar('\n', false); } BcBigDig bc_num_bigdig2(const BcNum* restrict n) { #if BC_DEBUG #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY #endif // BC_DEBUG // This function returns no errors because it's guaranteed to succeed if // its preconditions are met. Those preconditions include both n needs to // be non-NULL, n being non-negative, and n being less than vm->max. If all // of that is true, then we can just convert without worrying about negative // errors or overflow. BcBigDig r = 0; size_t nrdx = BC_NUM_RDX_VAL(n); assert(n != NULL); assert(!BC_NUM_NEG(n)); assert(bc_num_cmp(n, &vm->max) < 0); assert(n->len - nrdx <= 3); // There is a small speed win from unrolling the loop here, and since it // only adds 53 bytes, I decided that it was worth it. switch (n->len - nrdx) { case 3: { r = (BcBigDig) n->num[nrdx + 2]; // Fallthrough. BC_FALLTHROUGH } case 2: { r = r * BC_BASE_POW + (BcBigDig) n->num[nrdx + 1]; // Fallthrough. BC_FALLTHROUGH } case 1: { r = r * BC_BASE_POW + (BcBigDig) n->num[nrdx]; } } return r; } BcBigDig bc_num_bigdig(const BcNum* restrict n) { #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY assert(n != NULL); // This error checking is extremely important, and if you do not have a // guarantee that converting a number will always succeed in a particular // case, you *must* call this function to get these error checks. This // includes all instances of numbers inputted by the user or calculated by // the user. Otherwise, you can call the faster bc_num_bigdig2(). if (BC_ERR(BC_NUM_NEG(n))) bc_err(BC_ERR_MATH_NEGATIVE); if (BC_ERR(bc_num_cmp(n, &vm->max) >= 0)) bc_err(BC_ERR_MATH_OVERFLOW); return bc_num_bigdig2(n); } void bc_num_bigdig2num(BcNum* restrict n, BcBigDig val) { BcDig* ptr; size_t i; assert(n != NULL); bc_num_zero(n); // Already 0. if (!val) return; // Expand first. This is the only way this function can fail, and it's a // fatal error. bc_num_expand(n, BC_NUM_BIGDIG_LOG10); // The conversion is easy because numbers are laid out in little-endian // order. for (ptr = n->num, i = 0; val; ++i, val /= BC_BASE_POW) { ptr[i] = val % BC_BASE_POW; } n->len = i; } #if BC_ENABLE_EXTRA_MATH void bc_num_rng(const BcNum* restrict n, BcRNG* rng) { BcNum temp, temp2, intn, frac; BcRand state1, state2, inc1, inc2; size_t nrdx = BC_NUM_RDX_VAL(n); #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY // This function holds the secret of how I interpret a seed number for the // PRNG. Well, it's actually in the development manual // (manuals/development.md#pseudo-random-number-generator), so look there // before you try to understand this. BC_SIG_LOCK; bc_num_init(&temp, n->len); bc_num_init(&temp2, n->len); bc_num_init(&frac, nrdx); bc_num_init(&intn, bc_num_int(n)); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; assert(BC_NUM_RDX_VALID_NP(vm->max)); // NOLINTNEXTLINE memcpy(frac.num, n->num, BC_NUM_SIZE(nrdx)); frac.len = nrdx; BC_NUM_RDX_SET_NP(frac, nrdx); frac.scale = n->scale; assert(BC_NUM_RDX_VALID_NP(frac)); assert(BC_NUM_RDX_VALID_NP(vm->max2)); // Multiply the fraction and truncate so that it's an integer. The // truncation is what clamps it, by the way. bc_num_mul(&frac, &vm->max2, &temp, 0); bc_num_truncate(&temp, temp.scale); bc_num_copy(&frac, &temp); // Get the integer. // NOLINTNEXTLINE memcpy(intn.num, n->num + nrdx, BC_NUM_SIZE(bc_num_int(n))); intn.len = bc_num_int(n); // This assert is here because it has to be true. It is also here to justify // some optimizations. assert(BC_NUM_NONZERO(&vm->max)); // If there *was* a fractional part... if (BC_NUM_NONZERO(&frac)) { // This divmod splits frac into the two state parts. bc_num_divmod(&frac, &vm->max, &temp, &temp2, 0); // frac is guaranteed to be smaller than vm->max * vm->max (pow). // This means that when dividing frac by vm->max, as above, the // quotient and remainder are both guaranteed to be less than vm->max, // which means we can use bc_num_bigdig2() here and not worry about // overflow. state1 = (BcRand) bc_num_bigdig2(&temp2); state2 = (BcRand) bc_num_bigdig2(&temp); } else state1 = state2 = 0; // If there *was* an integer part... if (BC_NUM_NONZERO(&intn)) { // This divmod splits intn into the two inc parts. bc_num_divmod(&intn, &vm->max, &temp, &temp2, 0); // Because temp2 is the mod of vm->max, from above, it is guaranteed // to be small enough to use bc_num_bigdig2(). inc1 = (BcRand) bc_num_bigdig2(&temp2); // Clamp the second inc part. if (bc_num_cmp(&temp, &vm->max) >= 0) { bc_num_copy(&temp2, &temp); bc_num_mod(&temp2, &vm->max, &temp, 0); } // The if statement above ensures that temp is less than vm->max, which // means that we can use bc_num_bigdig2() here. inc2 = (BcRand) bc_num_bigdig2(&temp); } else inc1 = inc2 = 0; bc_rand_seed(rng, state1, state2, inc1, inc2); err: BC_SIG_MAYLOCK; bc_num_free(&intn); bc_num_free(&frac); bc_num_free(&temp2); bc_num_free(&temp); BC_LONGJMP_CONT(vm); } void bc_num_createFromRNG(BcNum* restrict n, BcRNG* rng) { BcRand s1, s2, i1, i2; BcNum conv, temp1, temp2, temp3; BcDig temp1_num[BC_RAND_NUM_SIZE], temp2_num[BC_RAND_NUM_SIZE]; BcDig conv_num[BC_NUM_BIGDIG_LOG10]; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY BC_SIG_LOCK; bc_num_init(&temp3, 2 * BC_RAND_NUM_SIZE); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; bc_num_setup(&temp1, temp1_num, sizeof(temp1_num) / sizeof(BcDig)); bc_num_setup(&temp2, temp2_num, sizeof(temp2_num) / sizeof(BcDig)); bc_num_setup(&conv, conv_num, sizeof(conv_num) / sizeof(BcDig)); // This assert is here because it has to be true. It is also here to justify // the assumption that vm->max is not zero. assert(BC_NUM_NONZERO(&vm->max)); // Because this is true, we can just ignore math errors that would happen // otherwise. assert(BC_NUM_NONZERO(&vm->max2)); bc_rand_getRands(rng, &s1, &s2, &i1, &i2); // Put the second piece of state into a number. bc_num_bigdig2num(&conv, (BcBigDig) s2); assert(BC_NUM_RDX_VALID_NP(conv)); // Multiply by max to make room for the first piece of state. bc_num_mul(&conv, &vm->max, &temp1, 0); // Add in the first piece of state. bc_num_bigdig2num(&conv, (BcBigDig) s1); bc_num_add(&conv, &temp1, &temp2, 0); // Divide to make it an entirely fractional part. bc_num_div(&temp2, &vm->max2, &temp3, BC_RAND_STATE_BITS); // Now start on the increment parts. It's the same process without the // divide, so put the second piece of increment into a number. bc_num_bigdig2num(&conv, (BcBigDig) i2); assert(BC_NUM_RDX_VALID_NP(conv)); // Multiply by max to make room for the first piece of increment. bc_num_mul(&conv, &vm->max, &temp1, 0); // Add in the first piece of increment. bc_num_bigdig2num(&conv, (BcBigDig) i1); bc_num_add(&conv, &temp1, &temp2, 0); // Now add the two together. bc_num_add(&temp2, &temp3, n, 0); assert(BC_NUM_RDX_VALID(n)); err: BC_SIG_MAYLOCK; bc_num_free(&temp3); BC_LONGJMP_CONT(vm); } void bc_num_irand(BcNum* restrict a, BcNum* restrict b, BcRNG* restrict rng) { BcNum atemp; size_t i; assert(a != b); if (BC_ERR(BC_NUM_NEG(a))) bc_err(BC_ERR_MATH_NEGATIVE); // If either of these are true, then the numbers are integers. if (BC_NUM_ZERO(a) || BC_NUM_ONE(a)) return; #if BC_GCC // This is here in GCC to quiet the "maybe-uninitialized" warning. atemp.num = NULL; atemp.len = 0; #endif // BC_GCC if (BC_ERR(bc_num_nonInt(a, &atemp))) bc_err(BC_ERR_MATH_NON_INTEGER); assert(atemp.num != NULL); assert(atemp.len); if (atemp.len > 2) { size_t len; len = atemp.len - 2; // Just generate a random number for each limb. for (i = 0; i < len; i += 2) { BcRand dig; dig = bc_rand_bounded(rng, BC_BASE_RAND_POW); b->num[i] = (BcDig) (dig % BC_BASE_POW); b->num[i + 1] = (BcDig) (dig / BC_BASE_POW); } } else { // We need this set. i = 0; } // This will be true if there's one full limb after the two limb groups. if (i == atemp.len - 2) { // Increment this for easy use. i += 1; // If the last digit is not one, we need to set a bound for it // explicitly. Since there's still an empty limb, we need to fill that. if (atemp.num[i] != 1) { BcRand dig; BcRand bound; // Set the bound to the bound of the last limb times the amount // needed to fill the second-to-last limb as well. bound = ((BcRand) atemp.num[i]) * BC_BASE_POW; dig = bc_rand_bounded(rng, bound); // Fill the last two. b->num[i - 1] = (BcDig) (dig % BC_BASE_POW); b->num[i] = (BcDig) (dig / BC_BASE_POW); // Ensure that the length will be correct. If the last limb is zero, // then the length needs to be one less than the bound. b->len = atemp.len - (b->num[i] == 0); } // Here the last limb *is* one, which means the last limb does *not* // need to be filled. Also, the length needs to be one less because the // last limb is 0. else { b->num[i - 1] = (BcDig) bc_rand_bounded(rng, BC_BASE_POW); b->len = atemp.len - 1; } } // Here, there is only one limb to fill. else { // See above for how this works. if (atemp.num[i] != 1) { b->num[i] = (BcDig) bc_rand_bounded(rng, (BcRand) atemp.num[i]); b->len = atemp.len - (b->num[i] == 0); } else b->len = atemp.len - 1; } bc_num_clean(b); assert(BC_NUM_RDX_VALID(b)); } #endif // BC_ENABLE_EXTRA_MATH size_t bc_num_addReq(const BcNum* a, const BcNum* b, size_t scale) { size_t aint, bint, ardx, brdx; // Addition and subtraction require the max of the length of the two numbers // plus 1. BC_UNUSED(scale); ardx = BC_NUM_RDX_VAL(a); aint = bc_num_int(a); assert(aint <= a->len && ardx <= a->len); brdx = BC_NUM_RDX_VAL(b); bint = bc_num_int(b); assert(bint <= b->len && brdx <= b->len); ardx = BC_MAX(ardx, brdx); aint = BC_MAX(aint, bint); return bc_vm_growSize(bc_vm_growSize(ardx, aint), 1); } size_t bc_num_mulReq(const BcNum* a, const BcNum* b, size_t scale) { size_t max, rdx; // Multiplication requires the sum of the lengths of the numbers. rdx = bc_vm_growSize(BC_NUM_RDX_VAL(a), BC_NUM_RDX_VAL(b)); max = BC_NUM_RDX(scale); max = bc_vm_growSize(BC_MAX(max, rdx), 1); rdx = bc_vm_growSize(bc_vm_growSize(bc_num_int(a), bc_num_int(b)), max); return rdx; } size_t bc_num_divReq(const BcNum* a, const BcNum* b, size_t scale) { size_t max, rdx; // Division requires the length of the dividend plus the scale. rdx = bc_vm_growSize(BC_NUM_RDX_VAL(a), BC_NUM_RDX_VAL(b)); max = BC_NUM_RDX(scale); max = bc_vm_growSize(BC_MAX(max, rdx), 1); rdx = bc_vm_growSize(bc_num_int(a), max); return rdx; } size_t bc_num_powReq(const BcNum* a, const BcNum* b, size_t scale) { BC_UNUSED(scale); return bc_vm_growSize(bc_vm_growSize(a->len, b->len), 1); } #if BC_ENABLE_EXTRA_MATH size_t bc_num_placesReq(const BcNum* a, const BcNum* b, size_t scale) { BC_UNUSED(scale); return a->len + b->len - BC_NUM_RDX_VAL(a) - BC_NUM_RDX_VAL(b); } #endif // BC_ENABLE_EXTRA_MATH void bc_num_add(BcNum* a, BcNum* b, BcNum* c, size_t scale) { assert(BC_NUM_RDX_VALID(a)); assert(BC_NUM_RDX_VALID(b)); bc_num_binary(a, b, c, false, bc_num_as, bc_num_addReq(a, b, scale)); } void bc_num_sub(BcNum* a, BcNum* b, BcNum* c, size_t scale) { assert(BC_NUM_RDX_VALID(a)); assert(BC_NUM_RDX_VALID(b)); bc_num_binary(a, b, c, true, bc_num_as, bc_num_addReq(a, b, scale)); } void bc_num_mul(BcNum* a, BcNum* b, BcNum* c, size_t scale) { assert(BC_NUM_RDX_VALID(a)); assert(BC_NUM_RDX_VALID(b)); bc_num_binary(a, b, c, scale, bc_num_m, bc_num_mulReq(a, b, scale)); } void bc_num_div(BcNum* a, BcNum* b, BcNum* c, size_t scale) { assert(BC_NUM_RDX_VALID(a)); assert(BC_NUM_RDX_VALID(b)); bc_num_binary(a, b, c, scale, bc_num_d, bc_num_divReq(a, b, scale)); } void bc_num_mod(BcNum* a, BcNum* b, BcNum* c, size_t scale) { assert(BC_NUM_RDX_VALID(a)); assert(BC_NUM_RDX_VALID(b)); bc_num_binary(a, b, c, scale, bc_num_rem, bc_num_divReq(a, b, scale)); } void bc_num_pow(BcNum* a, BcNum* b, BcNum* c, size_t scale) { assert(BC_NUM_RDX_VALID(a)); assert(BC_NUM_RDX_VALID(b)); bc_num_binary(a, b, c, scale, bc_num_p, bc_num_powReq(a, b, scale)); } #if BC_ENABLE_EXTRA_MATH void bc_num_places(BcNum* a, BcNum* b, BcNum* c, size_t scale) { assert(BC_NUM_RDX_VALID(a)); assert(BC_NUM_RDX_VALID(b)); bc_num_binary(a, b, c, scale, bc_num_place, bc_num_placesReq(a, b, scale)); } void bc_num_lshift(BcNum* a, BcNum* b, BcNum* c, size_t scale) { assert(BC_NUM_RDX_VALID(a)); assert(BC_NUM_RDX_VALID(b)); bc_num_binary(a, b, c, scale, bc_num_left, bc_num_placesReq(a, b, scale)); } void bc_num_rshift(BcNum* a, BcNum* b, BcNum* c, size_t scale) { assert(BC_NUM_RDX_VALID(a)); assert(BC_NUM_RDX_VALID(b)); bc_num_binary(a, b, c, scale, bc_num_right, bc_num_placesReq(a, b, scale)); } #endif // BC_ENABLE_EXTRA_MATH void bc_num_sqrt(BcNum* restrict a, BcNum* restrict b, size_t scale) { BcNum num1, num2, half, f, fprime; BcNum* x0; BcNum* x1; BcNum* temp; // realscale is meant to quiet a warning on GCC about longjmp() clobbering. // This one is real. size_t pow, len, rdx, req, resscale, realscale; BcDig half_digs[1]; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY assert(a != NULL && b != NULL && a != b); if (BC_ERR(BC_NUM_NEG(a))) bc_err(BC_ERR_MATH_NEGATIVE); // We want to calculate to a's scale if it is bigger so that the result will // truncate properly. if (a->scale > scale) realscale = a->scale; else realscale = scale; // Set parameters for the result. len = bc_vm_growSize(bc_num_intDigits(a), 1); rdx = BC_NUM_RDX(realscale); // Square root needs half of the length of the parameter. req = bc_vm_growSize(BC_MAX(rdx, BC_NUM_RDX_VAL(a)), len >> 1); req = bc_vm_growSize(req, 1); BC_SIG_LOCK; // Unlike the binary operators, this function is the only single parameter // function and is expected to initialize the result. This means that it // expects that b is *NOT* preallocated. We allocate it here. bc_num_init(b, req); BC_SIG_UNLOCK; assert(a != NULL && b != NULL && a != b); assert(a->num != NULL && b->num != NULL); // Easy case. if (BC_NUM_ZERO(a)) { bc_num_setToZero(b, realscale); return; } // Another easy case. if (BC_NUM_ONE(a)) { bc_num_one(b); bc_num_extend(b, realscale); return; } // Set the parameters again. rdx = BC_NUM_RDX(realscale); rdx = BC_MAX(rdx, BC_NUM_RDX_VAL(a)); len = bc_vm_growSize(a->len, rdx); BC_SIG_LOCK; bc_num_init(&num1, len); bc_num_init(&num2, len); bc_num_setup(&half, half_digs, sizeof(half_digs) / sizeof(BcDig)); // There is a division by two in the formula. We set up a number that's 1/2 // so that we can use multiplication instead of heavy division. bc_num_setToZero(&half, 1); half.num[0] = BC_BASE_POW / 2; half.len = 1; BC_NUM_RDX_SET_NP(half, 1); bc_num_init(&f, len); bc_num_init(&fprime, len); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; // Pointers for easy switching. x0 = &num1; x1 = &num2; // Start with 1. bc_num_one(x0); // The power of the operand is needed for the estimate. pow = bc_num_intDigits(a); // The code in this if statement calculates the initial estimate. First, if // a is less than 1, then 0 is a good estimate. Otherwise, we want something // in the same ballpark. That ballpark is half of pow because the result // will have half the digits. if (pow) { // An odd number is served by starting with 2^((pow-1)/2), and an even // number is served by starting with 6^((pow-2)/2). Why? Because math. if (pow & 1) x0->num[0] = 2; else x0->num[0] = 6; pow -= 2 - (pow & 1); bc_num_shiftLeft(x0, pow / 2); } // I can set the rdx here directly because neg should be false. x0->scale = x0->rdx = 0; resscale = (realscale + BC_BASE_DIGS) + 2; // This is the calculation loop. This compare goes to 0 eventually as the // difference between the two numbers gets smaller than resscale. while (bc_num_cmp(x1, x0)) { assert(BC_NUM_NONZERO(x0)); // This loop directly corresponds to the iteration in Newton's method. // If you know the formula, this loop makes sense. Go study the formula. bc_num_div(a, x0, &f, resscale); bc_num_add(x0, &f, &fprime, resscale); assert(BC_NUM_RDX_VALID_NP(fprime)); assert(BC_NUM_RDX_VALID_NP(half)); bc_num_mul(&fprime, &half, x1, resscale); // Switch. temp = x0; x0 = x1; x1 = temp; } // Copy to the result and truncate. bc_num_copy(b, x0); if (b->scale > realscale) bc_num_truncate(b, b->scale - realscale); assert(!BC_NUM_NEG(b) || BC_NUM_NONZERO(b)); assert(BC_NUM_RDX_VALID(b)); assert(BC_NUM_RDX_VAL(b) <= b->len || !b->len); assert(!b->len || b->num[b->len - 1] || BC_NUM_RDX_VAL(b) == b->len); err: BC_SIG_MAYLOCK; bc_num_free(&fprime); bc_num_free(&f); bc_num_free(&num2); bc_num_free(&num1); BC_LONGJMP_CONT(vm); } void bc_num_divmod(BcNum* a, BcNum* b, BcNum* c, BcNum* d, size_t scale) { size_t ts, len; BcNum *ptr_a, num2; // This is volatile to quiet a warning on GCC about clobbering with // longjmp(). volatile bool init = false; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY // The bulk of this function is just doing what bc_num_binary() does for the // binary operators. However, it assumes that only c and a can be equal. // Set up the parameters. ts = BC_MAX(scale + b->scale, a->scale); len = bc_num_mulReq(a, b, ts); assert(a != NULL && b != NULL && c != NULL && d != NULL); assert(c != d && a != d && b != d && b != c); // Initialize or expand as necessary. if (c == a) { // NOLINTNEXTLINE memcpy(&num2, c, sizeof(BcNum)); ptr_a = &num2; BC_SIG_LOCK; bc_num_init(c, len); init = true; BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; } else { ptr_a = a; bc_num_expand(c, len); } // Do the quick version if possible. if (BC_NUM_NONZERO(a) && !BC_NUM_RDX_VAL(a) && !BC_NUM_RDX_VAL(b) && b->len == 1 && !scale) { BcBigDig rem; bc_num_divArray(ptr_a, (BcBigDig) b->num[0], c, &rem); assert(rem < BC_BASE_POW); d->num[0] = (BcDig) rem; d->len = (rem != 0); } // Do the slow method. else bc_num_r(ptr_a, b, c, d, scale, ts); assert(!BC_NUM_NEG(c) || BC_NUM_NONZERO(c)); assert(BC_NUM_RDX_VALID(c)); assert(BC_NUM_RDX_VAL(c) <= c->len || !c->len); assert(!c->len || c->num[c->len - 1] || BC_NUM_RDX_VAL(c) == c->len); assert(!BC_NUM_NEG(d) || BC_NUM_NONZERO(d)); assert(BC_NUM_RDX_VALID(d)); assert(BC_NUM_RDX_VAL(d) <= d->len || !d->len); assert(!d->len || d->num[d->len - 1] || BC_NUM_RDX_VAL(d) == d->len); err: // Only cleanup if we initialized. if (init) { BC_SIG_MAYLOCK; bc_num_free(&num2); BC_LONGJMP_CONT(vm); } } void bc_num_modexp(BcNum* a, BcNum* b, BcNum* c, BcNum* restrict d) { BcNum base, exp, two, temp, atemp, btemp, ctemp; BcDig two_digs[2]; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY assert(a != NULL && b != NULL && c != NULL && d != NULL); assert(a != d && b != d && c != d); if (BC_ERR(BC_NUM_ZERO(c))) bc_err(BC_ERR_MATH_DIVIDE_BY_ZERO); if (BC_ERR(BC_NUM_NEG(b))) bc_err(BC_ERR_MATH_NEGATIVE); #if BC_DEBUG || BC_GCC // This is entirely for quieting a useless scan-build error. btemp.len = 0; ctemp.len = 0; #endif // BC_DEBUG || BC_GCC // Eliminate fractional parts that are zero or error if they are not zero. if (BC_ERR(bc_num_nonInt(a, &atemp) || bc_num_nonInt(b, &btemp) || bc_num_nonInt(c, &ctemp))) { bc_err(BC_ERR_MATH_NON_INTEGER); } bc_num_expand(d, ctemp.len); BC_SIG_LOCK; bc_num_init(&base, ctemp.len); bc_num_setup(&two, two_digs, sizeof(two_digs) / sizeof(BcDig)); bc_num_init(&temp, btemp.len + 1); bc_num_createCopy(&exp, &btemp); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; bc_num_one(&two); two.num[0] = 2; bc_num_one(d); // We already checked for 0. bc_num_rem(&atemp, &ctemp, &base, 0); // If you know the algorithm I used, the memory-efficient method, then this // loop should be self-explanatory because it is the calculation loop. while (BC_NUM_NONZERO(&exp)) { // Num two cannot be 0, so no errors. bc_num_divmod(&exp, &two, &exp, &temp, 0); if (BC_NUM_ONE(&temp) && !BC_NUM_NEG_NP(temp)) { assert(BC_NUM_RDX_VALID(d)); assert(BC_NUM_RDX_VALID_NP(base)); bc_num_mul(d, &base, &temp, 0); // We already checked for 0. bc_num_rem(&temp, &ctemp, d, 0); } assert(BC_NUM_RDX_VALID_NP(base)); bc_num_mul(&base, &base, &temp, 0); // We already checked for 0. bc_num_rem(&temp, &ctemp, &base, 0); } err: BC_SIG_MAYLOCK; bc_num_free(&exp); bc_num_free(&temp); bc_num_free(&base); BC_LONGJMP_CONT(vm); assert(!BC_NUM_NEG(d) || d->len); assert(BC_NUM_RDX_VALID(d)); assert(!d->len || d->num[d->len - 1] || BC_NUM_RDX_VAL(d) == d->len); } #if BC_DEBUG_CODE void bc_num_printDebug(const BcNum* n, const char* name, bool emptyline) { bc_file_puts(&vm->fout, bc_flush_none, name); bc_file_puts(&vm->fout, bc_flush_none, ": "); bc_num_printDecimal(n, true); bc_file_putchar(&vm->fout, bc_flush_err, '\n'); if (emptyline) bc_file_putchar(&vm->fout, bc_flush_err, '\n'); vm->nchars = 0; } void bc_num_printDigs(const BcDig* n, size_t len, bool emptyline) { size_t i; for (i = len - 1; i < len; --i) { bc_file_printf(&vm->fout, " %lu", (unsigned long) n[i]); } bc_file_putchar(&vm->fout, bc_flush_err, '\n'); if (emptyline) bc_file_putchar(&vm->fout, bc_flush_err, '\n'); vm->nchars = 0; } void bc_num_printWithDigs(const BcNum* n, const char* name, bool emptyline) { bc_file_puts(&vm->fout, bc_flush_none, name); bc_file_printf(&vm->fout, " len: %zu, rdx: %zu, scale: %zu\n", name, n->len, BC_NUM_RDX_VAL(n), n->scale); bc_num_printDigs(n->num, n->len, emptyline); } void bc_num_dump(const char* varname, const BcNum* n) { ulong i, scale = n->scale; bc_file_printf(&vm->ferr, "\n%s = %s", varname, n->len ? (BC_NUM_NEG(n) ? "-" : "+") : "0 "); for (i = n->len - 1; i < n->len; --i) { if (i + 1 == BC_NUM_RDX_VAL(n)) { bc_file_puts(&vm->ferr, bc_flush_none, ". "); } if (scale / BC_BASE_DIGS != BC_NUM_RDX_VAL(n) - i - 1) { bc_file_printf(&vm->ferr, "%lu ", (unsigned long) n->num[i]); } else { int mod = scale % BC_BASE_DIGS; int d = BC_BASE_DIGS - mod; BcDig div; if (mod != 0) { div = n->num[i] / ((BcDig) bc_num_pow10[(ulong) d]); bc_file_printf(&vm->ferr, "%lu", (unsigned long) div); } div = n->num[i] % ((BcDig) bc_num_pow10[(ulong) d]); bc_file_printf(&vm->ferr, " ' %lu ", (unsigned long) div); } } bc_file_printf(&vm->ferr, "(%zu | %zu.%zu / %zu) %lu\n", n->scale, n->len, BC_NUM_RDX_VAL(n), n->cap, (unsigned long) (void*) n->num); bc_file_flush(&vm->ferr, bc_flush_err); } #endif // BC_DEBUG_CODE diff --git a/src/opt.c b/src/opt.c index f01d86e18830..a1c8e813b1ea 100644 --- a/src/opt.c +++ b/src/opt.c @@ -1,394 +1,394 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * Adapted from https://github.com/skeeto/optparse * * ***************************************************************************** * * Code for getopt_long() replacement. It turns out that getopt_long() has * different behavior on different platforms. * */ #include #include #include #include #include #include #include /** * Returns true if index @a i is the end of the longopts array. * @param longopts The long options array. * @param i The index to test. * @return True if @a i is the last index, false otherwise. */ static inline bool bc_opt_longoptsEnd(const BcOptLong* longopts, size_t i) { return !longopts[i].name && !longopts[i].val; } /** * Returns the name of the long option that matches the character @a c. * @param longopts The long options array. * @param c The character to match against. * @return The name of the long option that matches @a c, or "NULL". */ static const char* bc_opt_longopt(const BcOptLong* longopts, int c) { size_t i; for (i = 0; !bc_opt_longoptsEnd(longopts, i); ++i) { if (longopts[i].val == c) return longopts[i].name; } BC_UNREACHABLE #if !BC_CLANG return "NULL"; #endif // !BC_CLANG } /** * Issues a fatal error for an option parsing failure. * @param err The error. * @param c The character for the failing option. * @param str Either the string for the failing option, or the invalid * option. * @param use_short True if the short option should be used for error printing, * false otherwise. */ static void bc_opt_error(BcErr err, int c, const char* str, bool use_short) { if (err == BC_ERR_FATAL_OPTION) { if (use_short) { char short_str[2]; short_str[0] = (char) c; short_str[1] = '\0'; bc_error(err, 0, short_str); } else bc_error(err, 0, str); } else bc_error(err, 0, (int) c, str); } /** * Returns the type of the long option that matches @a c. * @param longopts The long options array. * @param c The character to match against. * @return The type of the long option as an integer, or -1 if none. */ static int bc_opt_type(const BcOptLong* longopts, char c) { size_t i; if (c == ':') return -1; for (i = 0; !bc_opt_longoptsEnd(longopts, i) && longopts[i].val != c; ++i) { continue; } if (bc_opt_longoptsEnd(longopts, i)) return -1; return (int) longopts[i].type; } /** * Parses a short option. * @param o The option parser. * @param longopts The long options array. * @return The character for the short option, or -1 if none left. */ static int bc_opt_parseShort(BcOpt* o, const BcOptLong* longopts) { int type; - char* next; - char* option = o->argv[o->optind]; + const char* next; + const char* option = o->argv[o->optind]; int ret = -1; // Make sure to clear these. o->optopt = 0; o->optarg = NULL; // Get the next option. option += o->subopt + 1; o->optopt = option[0]; // Get the type and the next data. type = bc_opt_type(longopts, option[0]); next = o->argv[o->optind + 1]; switch (type) { case -1: case BC_OPT_BC_ONLY: case BC_OPT_DC_ONLY: { // Check for invalid option and barf if so. if (type == -1 || (type == BC_OPT_BC_ONLY && BC_IS_DC) || (type == BC_OPT_DC_ONLY && BC_IS_BC)) { char str[2] = { 0, 0 }; str[0] = option[0]; o->optind += 1; bc_opt_error(BC_ERR_FATAL_OPTION, option[0], str, true); } // Fallthrough. BC_FALLTHROUGH } case BC_OPT_NONE: { // If there is something else, update the suboption. if (option[1]) o->subopt += 1; else { // Go to the next argument. o->subopt = 0; o->optind += 1; } ret = (int) option[0]; break; } case BC_OPT_REQUIRED_BC_ONLY: { #if DC_ENABLED if (BC_IS_DC) { bc_opt_error(BC_ERR_FATAL_OPTION, option[0], bc_opt_longopt(longopts, option[0]), true); } #endif // DC_ENABLED // Fallthrough BC_FALLTHROUGH } case BC_OPT_REQUIRED: { // Always go to the next argument. o->subopt = 0; o->optind += 1; // Use the next characters, if they exist. if (option[1]) o->optarg = option + 1; else if (next != NULL) { // USe the next. o->optarg = next; o->optind += 1; } // No argument, barf. else { bc_opt_error(BC_ERR_FATAL_OPTION_NO_ARG, option[0], bc_opt_longopt(longopts, option[0]), true); } ret = (int) option[0]; break; } } return ret; } /** * Ensures that a long option argument matches a long option name, regardless of * "=" at the end. * @param name The name to match. * @param option The command-line argument. * @return True if @a option matches @a name, false otherwise. */ static bool bc_opt_longoptsMatch(const char* name, const char* option) { const char* a = option; const char* n = name; // Can never match a NULL name. if (name == NULL) return false; // Loop through. for (; *a && *n && *a != '='; ++a, ++n) { if (*a != *n) return false; } // Ensure they both end at the same place. return (*n == '\0' && (*a == '\0' || *a == '=')); } /** * Returns a pointer to the argument of a long option, or NULL if it not in the * same argument. * @param option The option to find the argument of. * @return A pointer to the argument of the option, or NULL if none. */ -static char* -bc_opt_longoptsArg(char* option) +static const char* +bc_opt_longoptsArg(const char* option) { // Find the end or equals sign. for (; *option && *option != '='; ++option) { continue; } if (*option == '=') return option + 1; else return NULL; } int bc_opt_parse(BcOpt* o, const BcOptLong* longopts) { size_t i; - char* option; + const char* option; bool empty; // This just eats empty options. do { option = o->argv[o->optind]; if (option == NULL) return -1; empty = !strcmp(option, ""); o->optind += empty; } while (empty); // If the option is just a "--". if (BC_OPT_ISDASHDASH(option)) { // Consume "--". o->optind += 1; return -1; } // Parse a short option. else if (BC_OPT_ISSHORTOPT(option)) return bc_opt_parseShort(o, longopts); // If the option is not long at this point, we are done. else if (!BC_OPT_ISLONGOPT(option)) return -1; // Clear these. o->optopt = 0; o->optarg = NULL; // Skip "--" at beginning of the option. option += 2; o->optind += 1; // Loop through the valid long options. for (i = 0; !bc_opt_longoptsEnd(longopts, i); i++) { const char* name = longopts[i].name; // If we have a match... if (bc_opt_longoptsMatch(name, option)) { - char* arg; + const char* arg; // Get the option char and the argument. o->optopt = longopts[i].val; arg = bc_opt_longoptsArg(option); // Error if the option is invalid.. if ((longopts[i].type == BC_OPT_BC_ONLY && BC_IS_DC) || (longopts[i].type == BC_OPT_REQUIRED_BC_ONLY && BC_IS_DC) || (longopts[i].type == BC_OPT_DC_ONLY && BC_IS_BC)) { bc_opt_error(BC_ERR_FATAL_OPTION, o->optopt, name, false); } // Error if we have an argument and should not. if (longopts[i].type == BC_OPT_NONE && arg != NULL) { bc_opt_error(BC_ERR_FATAL_OPTION_ARG, o->optopt, name, false); } // Set the argument, or check the next argument if we don't have // one. if (arg != NULL) o->optarg = arg; else if (longopts[i].type == BC_OPT_REQUIRED || longopts[i].type == BC_OPT_REQUIRED_BC_ONLY) { // Get the next argument. o->optarg = o->argv[o->optind]; // All's good if it exists; otherwise, barf. if (o->optarg != NULL) o->optind += 1; else { bc_opt_error(BC_ERR_FATAL_OPTION_NO_ARG, o->optopt, name, false); } } return o->optopt; } } // If we reach this point, the option is invalid. bc_opt_error(BC_ERR_FATAL_OPTION, 0, option, false); BC_UNREACHABLE #if !BC_CLANG return -1; #endif // !BC_CLANG } void -bc_opt_init(BcOpt* o, char* argv[]) +bc_opt_init(BcOpt* o, const char* argv[]) { o->argv = argv; o->optind = 1; o->subopt = 0; o->optarg = NULL; } diff --git a/src/program.c b/src/program.c index f30be26f2141..3b6ebc003a3e 100644 --- a/src/program.c +++ b/src/program.c @@ -1,3844 +1,3849 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * Code to execute bc programs. * */ #include #include #include #include #include #include #include #include #include #include /** * Does a type check for something that expects a number. * @param r The result that will be checked. * @param n The result's number. */ static inline void bc_program_type_num(BcResult* r, BcNum* n) { #if BC_ENABLED // This should have already been taken care of. assert(r->t != BC_RESULT_VOID); #endif // BC_ENABLED if (BC_ERR(!BC_PROG_NUM(r, n))) bc_err(BC_ERR_EXEC_TYPE); } #if BC_ENABLED /** * Does a type check. * @param r The result to check. * @param t The type that the result should be. */ static void bc_program_type_match(BcResult* r, BcType t) { if (BC_ERR((r->t != BC_RESULT_ARRAY) != (!t))) bc_err(BC_ERR_EXEC_TYPE); } #endif // BC_ENABLED /** * Pulls an index out of a bytecode vector and updates the index into the vector * to point to the spot after the index. For more details on bytecode indices, * see the development manual (manuals/development.md#bytecode-indices). * @param code The bytecode vector. * @param bgn An in/out parameter; the index into the vector that will be * updated. * @return The index at @a bgn in the bytecode vector. */ static size_t bc_program_index(const char* restrict code, size_t* restrict bgn) { uchar amt = (uchar) code[(*bgn)++], i = 0; size_t res = 0; for (; i < amt; ++i, ++(*bgn)) { size_t temp = ((size_t) ((int) (uchar) code[*bgn]) & UCHAR_MAX); res |= (temp << (i * CHAR_BIT)); } return res; } /** * Returns a string from a result and its number. * @param p The program. * @param n The number tied to the result. * @return The string corresponding to the result and number. */ static inline char* bc_program_string(BcProgram* p, const BcNum* n) { return *((char**) bc_vec_item(&p->strs, n->scale)); } #if BC_ENABLED /** * Prepares the globals for a function call. This is only called when global * stacks are on because it pushes a copy of the current globals onto each of * their respective stacks. * @param p The program. */ static void bc_program_prepGlobals(BcProgram* p) { size_t i; for (i = 0; i < BC_PROG_GLOBALS_LEN; ++i) { bc_vec_push(p->globals_v + i, p->globals + i); } #if BC_ENABLE_EXTRA_MATH bc_rand_push(&p->rng); #endif // BC_ENABLE_EXTRA_MATH } /** * Pops globals stacks on returning from a function, or in the case of reset, * pops all but one item on each global stack. * @param p The program. * @param reset True if all but one item on each stack should be popped, false * otherwise. */ static void bc_program_popGlobals(BcProgram* p, bool reset) { size_t i; BC_SIG_ASSERT_LOCKED; for (i = 0; i < BC_PROG_GLOBALS_LEN; ++i) { BcVec* v = p->globals_v + i; bc_vec_npop(v, reset ? v->len - 1 : 1); p->globals[i] = BC_PROG_GLOBAL(v); } #if BC_ENABLE_EXTRA_MATH bc_rand_pop(&p->rng, reset); #endif // BC_ENABLE_EXTRA_MATH } /** * Derefeneces an array reference and returns a pointer to the real array. * @param p The program. * @param vec The reference vector. * @return A pointer to the desired array. */ static BcVec* bc_program_dereference(const BcProgram* p, BcVec* vec) { BcVec* v; size_t vidx, nidx, i = 0; // We want to be sure we have a reference vector. assert(vec->size == sizeof(uchar)); // Get the index of the vector in arrs, then the index of the original // referenced vector. vidx = bc_program_index(vec->v, &i); nidx = bc_program_index(vec->v, &i); v = bc_vec_item(bc_vec_item(&p->arrs, vidx), nidx); // We want to be sure we do *not* have a reference vector. assert(v->size != sizeof(uchar)); return v; } #endif // BC_ENABLED /** * Creates a BcNum from a BcBigDig and pushes onto the results stack. This is a * convenience function. * @param p The program. * @param dig The BcBigDig to push onto the results stack. * @param type The type that the pushed result should be. */ static void bc_program_pushBigdig(BcProgram* p, BcBigDig dig, BcResultType type) { BcResult res; res.t = type; BC_SIG_LOCK; bc_num_createFromBigdig(&res.d.n, dig); bc_vec_push(&p->results, &res); BC_SIG_UNLOCK; } size_t bc_program_addString(BcProgram* p, const char* str) { size_t idx; BC_SIG_ASSERT_LOCKED; if (bc_map_insert(&p->str_map, str, p->strs.len, &idx)) { char** str_ptr; BcId* id = bc_vec_item(&p->str_map, idx); // Get the index. idx = id->idx; // Push an empty string on the proper vector. str_ptr = bc_vec_pushEmpty(&p->strs); // We reuse the string in the ID (allocated by bc_map_insert()), because // why not? *str_ptr = id->name; } else { BcId* id = bc_vec_item(&p->str_map, idx); idx = id->idx; } return idx; } size_t bc_program_search(BcProgram* p, const char* name, bool var) { BcVec* v; BcVec* map; size_t i; BC_SIG_ASSERT_LOCKED; // Grab the right vector and map. v = var ? &p->vars : &p->arrs; map = var ? &p->var_map : &p->arr_map; // We do an insert because the variable might not exist yet. This is because // the parser calls this function. If the insert succeeds, we create a stack // for the variable/array. But regardless, bc_map_insert() gives us the // index of the item in i. if (bc_map_insert(map, name, v->len, &i)) { BcVec* temp = bc_vec_pushEmpty(v); bc_array_init(temp, var); } return ((BcId*) bc_vec_item(map, i))->idx; } /** * Returns the correct variable or array stack for the type. * @param p The program. * @param idx The index of the variable or array in the variable or array * vector. * @param type The type of vector to return. * @return A pointer to the variable or array stack. */ static inline BcVec* bc_program_vec(const BcProgram* p, size_t idx, BcType type) { const BcVec* v = (type == BC_TYPE_VAR) ? &p->vars : &p->arrs; return bc_vec_item(v, idx); } /** * Returns a pointer to the BcNum corresponding to the result. There is one * case, however, where this returns a pointer to a BcVec: if the type of the * result is array. In that case, the pointer is casted to a pointer to BcNum, * but is never used. The function that calls this expecting an array casts the * pointer back. This function is called a lot and needs to be as fast as * possible. * @param p The program. * @param r The result whose number will be returned. * @return The BcNum corresponding to the result. */ static BcNum* bc_program_num(BcProgram* p, BcResult* r) { BcNum* n; #ifdef _WIN32 // Windows made it an error to not initialize this, so shut it up. // I don't want to do this on other platforms because this procedure // is one of the most heavily-used, and eliminating the initialization // is a performance win. n = NULL; #endif // _WIN32 switch (r->t) { case BC_RESULT_STR: case BC_RESULT_TEMP: case BC_RESULT_IBASE: case BC_RESULT_SCALE: case BC_RESULT_OBASE: #if BC_ENABLE_EXTRA_MATH case BC_RESULT_SEED: #endif // BC_ENABLE_EXTRA_MATH { n = &r->d.n; break; } case BC_RESULT_VAR: case BC_RESULT_ARRAY: case BC_RESULT_ARRAY_ELEM: { BcVec* v; BcType type = (r->t == BC_RESULT_VAR) ? BC_TYPE_VAR : BC_TYPE_ARRAY; // Get the correct variable or array vector. v = bc_program_vec(p, r->d.loc.loc, type); // Surprisingly enough, the hard case is *not* returning an array; // it's returning an array element. This is because we have to dig // deeper to get *to* the element. That's what the code inside this // if statement does. if (r->t == BC_RESULT_ARRAY_ELEM) { size_t idx = r->d.loc.idx; v = bc_vec_item(v, r->d.loc.stack_idx); #if BC_ENABLED // If this is true, we have a reference vector, so dereference // it. The reason we don't need to worry about it for returning // a straight array is because we only care about references // when we access elements of an array that is a reference. That // is this code, so in essence, this line takes care of arrays // as well. if (v->size == sizeof(uchar)) v = bc_program_dereference(p, v); #endif // BC_ENABLED // We want to be sure we got a valid array of numbers. assert(v->size == sizeof(BcNum)); // The bc spec says that if an element is accessed that does not // exist, it should be preinitialized to 0. Well, if we access // an element *way* out there, we have to preinitialize all // elements between the current last element and the actual // accessed element. if (v->len <= idx) { BC_SIG_LOCK; bc_array_expand(v, bc_vm_growSize(idx, 1)); BC_SIG_UNLOCK; } n = bc_vec_item(v, idx); } // This is either a number (for a var) or an array (for an array). // Because bc_vec_top() and bc_vec_item() return a void*, we don't // need to cast. else { #if BC_ENABLED if (BC_IS_BC) { n = bc_vec_item(v, r->d.loc.stack_idx); } else #endif // BC_ENABLED { n = bc_vec_top(v); } } break; } case BC_RESULT_ZERO: { n = &vm->zero; break; } case BC_RESULT_ONE: { n = &vm->one; break; } #if BC_ENABLED // We should never get here; this is taken care of earlier because a // result is expected. case BC_RESULT_VOID: #if BC_DEBUG { abort(); // Fallthrough } #endif // BC_DEBUG case BC_RESULT_LAST: { n = &p->last; break; } #endif // BC_ENABLED #if BC_GCC // This is here in GCC to quiet the "maybe-uninitialized" warning. default: { abort(); } #endif // BC_GCC } return n; } /** * Prepares an operand for use. * @param p The program. * @param r An out parameter; this is set to the pointer to the result that * we care about. * @param n An out parameter; this is set to the pointer to the number that * we care about. * @param idx The index of the result from the top of the results stack. */ static void bc_program_operand(BcProgram* p, BcResult** r, BcNum** n, size_t idx) { *r = bc_vec_item_rev(&p->results, idx); #if BC_ENABLED if (BC_ERR((*r)->t == BC_RESULT_VOID)) bc_err(BC_ERR_EXEC_VOID_VAL); #endif // BC_ENABLED *n = bc_program_num(p, *r); } /** * Prepares the operands of a binary operator. * @param p The program. * @param l An out parameter; this is set to the pointer to the result for * the left operand. * @param ln An out parameter; this is set to the pointer to the number for * the left operand. * @param r An out parameter; this is set to the pointer to the result for * the right operand. * @param rn An out parameter; this is set to the pointer to the number for * the right operand. * @param idx The starting index where the operands are in the results stack, * starting from the top. */ static void bc_program_binPrep(BcProgram* p, BcResult** l, BcNum** ln, BcResult** r, BcNum** rn, size_t idx) { BcResultType lt; assert(p != NULL && l != NULL && ln != NULL && r != NULL && rn != NULL); #ifndef BC_PROG_NO_STACK_CHECK // Check the stack for dc. if (BC_IS_DC) { if (BC_ERR(!BC_PROG_STACK(&p->results, idx + 2))) { bc_err(BC_ERR_EXEC_STACK); } } #endif // BC_PROG_NO_STACK_CHECK assert(BC_PROG_STACK(&p->results, idx + 2)); // Get the operands. bc_program_operand(p, l, ln, idx + 1); bc_program_operand(p, r, rn, idx); lt = (*l)->t; #if BC_ENABLED // bc_program_operand() checked these for us. assert(lt != BC_RESULT_VOID && (*r)->t != BC_RESULT_VOID); #endif // BC_ENABLED // We run this again under these conditions in case any vector has been // reallocated out from under the BcNums or arrays we had. In other words, // this is to fix pointer invalidation. if (lt == (*r)->t && (lt == BC_RESULT_VAR || lt == BC_RESULT_ARRAY_ELEM)) { *ln = bc_program_num(p, *l); } if (BC_ERR(lt == BC_RESULT_STR)) bc_err(BC_ERR_EXEC_TYPE); } /** * Prepares the operands of a binary operator and type checks them. This is * separate from bc_program_binPrep() because some places want this, others want * bc_program_binPrep(). * @param p The program. * @param l An out parameter; this is set to the pointer to the result for * the left operand. * @param ln An out parameter; this is set to the pointer to the number for * the left operand. * @param r An out parameter; this is set to the pointer to the result for * the right operand. * @param rn An out parameter; this is set to the pointer to the number for * the right operand. * @param idx The starting index where the operands are in the results stack, * starting from the top. */ static void bc_program_binOpPrep(BcProgram* p, BcResult** l, BcNum** ln, BcResult** r, BcNum** rn, size_t idx) { bc_program_binPrep(p, l, ln, r, rn, idx); bc_program_type_num(*l, *ln); bc_program_type_num(*r, *rn); } /** * Prepares the operands of an assignment operator. * @param p The program. * @param l An out parameter; this is set to the pointer to the result for the * left operand. * @param ln An out parameter; this is set to the pointer to the number for the * left operand. * @param r An out parameter; this is set to the pointer to the result for the * right operand. * @param rn An out parameter; this is set to the pointer to the number for the * right operand. */ static void bc_program_assignPrep(BcProgram* p, BcResult** l, BcNum** ln, BcResult** r, BcNum** rn) { BcResultType lt, min; bool good; // This is the min non-allowable result type. dc allows strings. min = BC_RESULT_TEMP - ((unsigned int) (BC_IS_BC)); // Prepare the operands. bc_program_binPrep(p, l, ln, r, rn, 0); lt = (*l)->t; // Typecheck the left. if (BC_ERR(lt >= min && lt <= BC_RESULT_ONE)) bc_err(BC_ERR_EXEC_TYPE); // Strings can be assigned to variables. We are already good if we are // assigning a string. good = ((*r)->t == BC_RESULT_STR && lt <= BC_RESULT_ARRAY_ELEM); assert(BC_PROG_STR(*rn) || (*r)->t != BC_RESULT_STR); // If not, type check for a number. if (!good) bc_program_type_num(*r, *rn); } /** * Prepares a single operand and type checks it. This is separate from * bc_program_operand() because different places want one or the other. * @param p The program. * @param r An out parameter; this is set to the pointer to the result that * we care about. * @param n An out parameter; this is set to the pointer to the number that * we care about. * @param idx The index of the result from the top of the results stack. */ static void bc_program_prep(BcProgram* p, BcResult** r, BcNum** n, size_t idx) { assert(p != NULL && r != NULL && n != NULL); #ifndef BC_PROG_NO_STACK_CHECK // Check the stack for dc. if (BC_IS_DC) { if (BC_ERR(!BC_PROG_STACK(&p->results, idx + 1))) { bc_err(BC_ERR_EXEC_STACK); } } #endif // BC_PROG_NO_STACK_CHECK assert(BC_PROG_STACK(&p->results, idx + 1)); bc_program_operand(p, r, n, idx); // dc does not allow strings in this case. bc_program_type_num(*r, *n); } /** * Prepares and returns a clean result for the result of an operation. * @param p The program. * @return A clean result. */ static BcResult* bc_program_prepResult(BcProgram* p) { BcResult* res = bc_vec_pushEmpty(&p->results); bc_result_clear(res); return res; } /** * Prepares a constant for use. This parses the constant into a number and then * pushes that number onto the results stack. * @param p The program. * @param code The bytecode vector that we will pull the index of the constant * from. * @param bgn An in/out parameter; marks the start of the index in the * bytecode vector and will be updated to point to after the index. */ static void bc_program_const(BcProgram* p, const char* code, size_t* bgn) { // I lied. I actually push the result first. I can do this because the // result will be popped on error. I also get the constant itself. BcResult* r = bc_program_prepResult(p); BcConst* c = bc_vec_item(&p->consts, bc_program_index(code, bgn)); BcBigDig base = BC_PROG_IBASE(p); // Only reparse if the base changed. if (c->base != base) { // Allocate if we haven't yet. if (c->num.num == NULL) { // The plus 1 is in case of overflow with lack of clamping. size_t len = strlen(c->val) + (BC_DIGIT_CLAMP == 0); BC_SIG_LOCK; bc_num_init(&c->num, BC_NUM_RDX(len)); BC_SIG_UNLOCK; } // We need to zero an already existing number. else bc_num_zero(&c->num); // bc_num_parse() should only do operations that cannot fail. bc_num_parse(&c->num, c->val, base); c->base = base; } BC_SIG_LOCK; bc_num_createCopy(&r->d.n, &c->num); BC_SIG_UNLOCK; } /** * Executes a binary operator operation. * @param p The program. * @param inst The instruction corresponding to the binary operator to execute. */ static void bc_program_op(BcProgram* p, uchar inst) { BcResult* opd1; BcResult* opd2; BcResult* res; BcNum* n1; BcNum* n2; size_t idx = inst - BC_INST_POWER; res = bc_program_prepResult(p); bc_program_binOpPrep(p, &opd1, &n1, &opd2, &n2, 1); BC_SIG_LOCK; // Initialize the number with enough space, using the correct // BcNumBinaryOpReq function. This looks weird because it is executing an // item of an array. Rest assured that item is a function. bc_num_init(&res->d.n, bc_program_opReqs[idx](n1, n2, BC_PROG_SCALE(p))); BC_SIG_UNLOCK; assert(BC_NUM_RDX_VALID(n1)); assert(BC_NUM_RDX_VALID(n2)); // Run the operation. This also executes an item of an array. bc_program_ops[idx](n1, n2, &res->d.n, BC_PROG_SCALE(p)); bc_program_retire(p, 1, 2); } /** * Executes a read() or ? command. * @param p The program. */ static void bc_program_read(BcProgram* p) { BcStatus s; BcInstPtr ip; size_t i; const char* file; BcMode mode; BcFunc* f = bc_vec_item(&p->fns, BC_PROG_READ); // If we are already executing a read, that is an error. So look for a read // and barf. for (i = 0; i < p->stack.len; ++i) { BcInstPtr* ip_ptr = bc_vec_item(&p->stack, i); if (ip_ptr->func == BC_PROG_READ) bc_err(BC_ERR_EXEC_REC_READ); } BC_SIG_LOCK; // Save the filename because we are going to overwrite it. file = vm->file; mode = vm->mode; // It is a parse error if there needs to be more than one line, so we unset // this to tell the lexer to not request more. We set it back later. vm->mode = BC_MODE_FILE; if (!BC_PARSE_IS_INITED(&vm->read_prs, p)) { // We need to parse, but we don't want to use the existing parser // because it has state it needs to keep. (It could have a partial parse // state.) So we create a new parser. This parser is in the BcVm struct // so that it is not local, which means that a longjmp() could change // it. bc_parse_init(&vm->read_prs, p, BC_PROG_READ); // We need a separate input buffer; that's why it is also in the BcVm // struct. bc_vec_init(&vm->read_buf, sizeof(char), BC_DTOR_NONE); } else { // This needs to be updated because the parser could have been used // somewhere else. bc_parse_updateFunc(&vm->read_prs, BC_PROG_READ); // The read buffer also needs to be emptied or else it will still // contain previous read expressions. bc_vec_empty(&vm->read_buf); } BC_SETJMP_LOCKED(vm, exec_err); BC_SIG_UNLOCK; // Set up the lexer and the read function. bc_lex_file(&vm->read_prs.l, bc_program_stdin_name); bc_vec_popAll(&f->code); // Read a line. if (!BC_R) s = bc_read_line(&vm->read_buf, ""); else s = bc_read_line(&vm->read_buf, BC_VM_READ_PROMPT); // We should *not* have run into EOF. if (s == BC_STATUS_EOF) bc_err(BC_ERR_EXEC_READ_EXPR); // Parse *one* expression, so mode should not be stdin. bc_parse_text(&vm->read_prs, vm->read_buf.v, BC_MODE_FILE); BC_SIG_LOCK; vm->expr(&vm->read_prs, BC_PARSE_NOREAD | BC_PARSE_NEEDVAL); BC_SIG_UNLOCK; // We *must* have a valid expression. A semicolon cannot end an expression, // although EOF can. if (BC_ERR(vm->read_prs.l.t != BC_LEX_NLINE && vm->read_prs.l.t != BC_LEX_EOF)) { bc_err(BC_ERR_EXEC_READ_EXPR); } #if BC_ENABLED // Push on the globals stack if necessary. if (BC_G) bc_program_prepGlobals(p); #endif // BC_ENABLED // Set up a new BcInstPtr. ip.func = BC_PROG_READ; ip.idx = 0; ip.len = p->results.len; // Update this pointer, just in case. f = bc_vec_item(&p->fns, BC_PROG_READ); // We want a return instruction to simplify things. bc_vec_pushByte(&f->code, vm->read_ret); // This lock is here to make sure dc's tail calls are the same length. BC_SIG_LOCK; bc_vec_push(&p->stack, &ip); #if DC_ENABLED // We need a new tail call entry for dc. if (BC_IS_DC) { size_t temp = 0; bc_vec_push(&p->tail_calls, &temp); } #endif // DC_ENABLED exec_err: BC_SIG_MAYLOCK; vm->mode = (uchar) mode; vm->file = file; BC_LONGJMP_CONT(vm); } #if BC_ENABLE_EXTRA_MATH /** * Execute a rand(). * @param p The program. */ static void bc_program_rand(BcProgram* p) { BcRand rand = bc_rand_int(&p->rng); bc_program_pushBigdig(p, (BcBigDig) rand, BC_RESULT_TEMP); #if BC_DEBUG // This is just to ensure that the generated number is correct. I also use // braces because I declare every local at the top of the scope. { BcResult* r = bc_vec_top(&p->results); assert(BC_NUM_RDX_VALID_NP(r->d.n)); } #endif // BC_DEBUG } #endif // BC_ENABLE_EXTRA_MATH /** * Prints a series of characters, without escapes. * @param str The string (series of characters). */ static void bc_program_printChars(const char* str) { const char* nl; size_t len = vm->nchars + strlen(str); sig_atomic_t lock; BC_SIG_TRYLOCK(lock); bc_file_puts(&vm->fout, bc_flush_save, str); // We need to update the number of characters, so we find the last newline // and set the characters accordingly. nl = strrchr(str, '\n'); if (nl != NULL) len = strlen(nl + 1); vm->nchars = len > UINT16_MAX ? UINT16_MAX : (uint16_t) len; BC_SIG_TRYUNLOCK(lock); } /** * Prints a string with escapes. * @param str The string. */ static void bc_program_printString(const char* restrict str) { size_t i, len = strlen(str); #if DC_ENABLED // This is to ensure a nul byte is printed for dc's stream operation. if (!len && BC_IS_DC) { bc_vm_putchar('\0', bc_flush_save); return; } #endif // DC_ENABLED // Loop over the characters, processing escapes and printing the rest. for (i = 0; i < len; ++i) { int c = str[i]; // If we have an escape... if (c == '\\' && i != len - 1) { const char* ptr; // Get the escape character and its companion. c = str[++i]; ptr = strchr(bc_program_esc_chars, c); // If we have a companion character... if (ptr != NULL) { // We need to specially handle a newline. if (c == 'n') { BC_SIG_LOCK; vm->nchars = UINT16_MAX; BC_SIG_UNLOCK; } // Grab the actual character. c = bc_program_esc_seqs[(size_t) (ptr - bc_program_esc_chars)]; } else { // Just print the backslash if there is no companion character. // The following character will be printed later after the outer // if statement. bc_vm_putchar('\\', bc_flush_save); } } bc_vm_putchar(c, bc_flush_save); } } /** * Executes a print. This function handles all printing except streaming. * @param p The program. * @param inst The instruction for the type of print we are doing. * @param idx The index of the result that we are printing. */ static void bc_program_print(BcProgram* p, uchar inst, size_t idx) { BcResult* r; char* str; BcNum* n; bool pop = (inst != BC_INST_PRINT); assert(p != NULL); #ifndef BC_PROG_NO_STACK_CHECK if (BC_IS_DC) { if (BC_ERR(!BC_PROG_STACK(&p->results, idx + 1))) { bc_err(BC_ERR_EXEC_STACK); } } #endif // BC_PROG_NO_STACK_CHECK assert(BC_PROG_STACK(&p->results, idx + 1)); r = bc_vec_item_rev(&p->results, idx); #if BC_ENABLED // If we have a void value, that's not necessarily an error. It is if pop is // true because that means that we are executing a print statement, but // attempting to do a print on a lone void value is allowed because that's // exactly how we want void values used. if (r->t == BC_RESULT_VOID) { if (BC_ERR(pop)) bc_err(BC_ERR_EXEC_VOID_VAL); bc_vec_pop(&p->results); return; } #endif // BC_ENABLED n = bc_program_num(p, r); // If we have a number... if (BC_PROG_NUM(r, n)) { #if BC_ENABLED assert(inst != BC_INST_PRINT_STR); #endif // BC_ENABLED // Print the number. bc_num_print(n, BC_PROG_OBASE(p), !pop); #if BC_ENABLED // Need to store the number in last. if (BC_IS_BC) bc_num_copy(&p->last, n); #endif // BC_ENABLED } else { // We want to flush any stuff in the stdout buffer first. bc_file_flush(&vm->fout, bc_flush_save); str = bc_program_string(p, n); #if BC_ENABLED if (inst == BC_INST_PRINT_STR) bc_program_printChars(str); else #endif // BC_ENABLED { bc_program_printString(str); // Need to print a newline only in this case. if (inst == BC_INST_PRINT) bc_vm_putchar('\n', bc_flush_err); } } // bc always pops. This macro makes sure that happens. if (BC_PROGRAM_POP(pop)) bc_vec_pop(&p->results); } void bc_program_negate(BcResult* r, BcNum* n) { bc_num_copy(&r->d.n, n); if (BC_NUM_NONZERO(&r->d.n)) BC_NUM_NEG_TGL_NP(r->d.n); } void bc_program_not(BcResult* r, BcNum* n) { if (!bc_num_cmpZero(n)) bc_num_one(&r->d.n); } #if BC_ENABLE_EXTRA_MATH void bc_program_trunc(BcResult* r, BcNum* n) { bc_num_copy(&r->d.n, n); bc_num_truncate(&r->d.n, n->scale); } #endif // BC_ENABLE_EXTRA_MATH /** * Runs a unary operation. * @param p The program. * @param inst The unary operation. */ static void bc_program_unary(BcProgram* p, uchar inst) { BcResult* res; BcResult* ptr; BcNum* num; res = bc_program_prepResult(p); bc_program_prep(p, &ptr, &num, 1); BC_SIG_LOCK; bc_num_init(&res->d.n, num->len); BC_SIG_UNLOCK; // This calls a function that is in an array. bc_program_unarys[inst - BC_INST_NEG](res, num); bc_program_retire(p, 1, 1); } /** * Executes a logical operator. * @param p The program. * @param inst The operator. */ static void bc_program_logical(BcProgram* p, uchar inst) { BcResult* opd1; BcResult* opd2; BcResult* res; BcNum* n1; BcNum* n2; bool cond = 0; ssize_t cmp; res = bc_program_prepResult(p); // All logical operators (except boolean not, which is taken care of by // bc_program_unary()), are binary operators. bc_program_binOpPrep(p, &opd1, &n1, &opd2, &n2, 1); // Boolean and and or are not short circuiting. This is why; they can be // implemented much easier this way. if (inst == BC_INST_BOOL_AND) { cond = (bc_num_cmpZero(n1) && bc_num_cmpZero(n2)); } else if (inst == BC_INST_BOOL_OR) { cond = (bc_num_cmpZero(n1) || bc_num_cmpZero(n2)); } else { // We have a relational operator, so do a comparison. cmp = bc_num_cmp(n1, n2); switch (inst) { case BC_INST_REL_EQ: { cond = (cmp == 0); break; } case BC_INST_REL_LE: { cond = (cmp <= 0); break; } case BC_INST_REL_GE: { cond = (cmp >= 0); break; } case BC_INST_REL_NE: { cond = (cmp != 0); break; } case BC_INST_REL_LT: { cond = (cmp < 0); break; } case BC_INST_REL_GT: { cond = (cmp > 0); break; } #if BC_DEBUG default: { // There is a bug if we get here. abort(); } #endif // BC_DEBUG } } BC_SIG_LOCK; bc_num_init(&res->d.n, BC_NUM_DEF_SIZE); BC_SIG_UNLOCK; if (cond) bc_num_one(&res->d.n); bc_program_retire(p, 1, 2); } /** * Assigns a string to a variable. * @param p The program. * @param num The location of the string as a BcNum. * @param v The stack for the variable. * @param push Whether to push the string or not. To push means to move the * string from the results stack and push it onto the variable * stack. */ static void bc_program_assignStr(BcProgram* p, BcNum* num, BcVec* v, bool push) { BcNum* n; assert(BC_PROG_STACK(&p->results, 1 + !push)); assert(num != NULL && num->num == NULL && num->cap == 0); // If we are not pushing onto the variable stack, we need to replace the // top of the variable stack. if (!push) bc_vec_pop(v); bc_vec_npop(&p->results, 1 + !push); n = bc_vec_pushEmpty(v); // We can just copy because the num should not have allocated anything. // NOLINTNEXTLINE memcpy(n, num, sizeof(BcNum)); } /** * Copies a value to a variable. This is used for storing in dc as well as to * set function parameters to arguments in bc. * @param p The program. * @param idx The index of the variable or array to copy to. * @param t The type to copy to. This could be a variable or an array. */ static void bc_program_copyToVar(BcProgram* p, size_t idx, BcType t) { BcResult *ptr = NULL, r; BcVec* vec; BcNum* n = NULL; bool var = (t == BC_TYPE_VAR); #if DC_ENABLED // Check the stack for dc. if (BC_IS_DC) { if (BC_ERR(!BC_PROG_STACK(&p->results, 1))) bc_err(BC_ERR_EXEC_STACK); } #endif assert(BC_PROG_STACK(&p->results, 1)); bc_program_operand(p, &ptr, &n, 0); #if BC_ENABLED // Get the variable for a bc function call. if (BC_IS_BC) { // Type match the result. bc_program_type_match(ptr, t); } #endif // BC_ENABLED vec = bc_program_vec(p, idx, t); // We can shortcut in dc if it's assigning a string by using // bc_program_assignStr(). if (ptr->t == BC_RESULT_STR) { assert(BC_PROG_STR(n)); if (BC_ERR(!var)) bc_err(BC_ERR_EXEC_TYPE); bc_program_assignStr(p, n, vec, true); return; } BC_SIG_LOCK; // Just create and copy for a normal variable. if (var) { if (BC_PROG_STR(n)) { // NOLINTNEXTLINE memcpy(&r.d.n, n, sizeof(BcNum)); } else bc_num_createCopy(&r.d.n, n); } else { // If we get here, we are handling an array. This is one place we need // to cast the number from bc_program_num() to a vector. BcVec* v = (BcVec*) n; BcVec* rv = &r.d.v; #if BC_ENABLED if (BC_IS_BC) { bool ref, ref_size; // True if we are using a reference. ref = (v->size == sizeof(BcNum) && t == BC_TYPE_REF); // True if we already have a reference vector. This is slightly // (okay, a lot; it just doesn't look that way) different from // above. The above means that we need to construct a reference // vector, whereas this means that we have one and we might have to // *dereference* it. ref_size = (v->size == sizeof(uchar)); // If we *should* have a reference. if (ref || (ref_size && t == BC_TYPE_REF)) { // Create a new reference vector. bc_vec_init(rv, sizeof(uchar), BC_DTOR_NONE); // If this is true, then we need to construct a reference. if (ref) { // Make sure the pointer was not invalidated. vec = bc_program_vec(p, idx, t); // Push the indices onto the reference vector. This takes // care of last; it ensures the reference goes to the right // place. bc_vec_pushIndex(rv, ptr->d.loc.loc); bc_vec_pushIndex(rv, ptr->d.loc.stack_idx); } // If we get here, we are copying a ref to a ref. Just push a // copy of all of the bytes. else bc_vec_npush(rv, v->len * sizeof(uchar), v->v); // Push the reference vector onto the array stack and pop the // source. bc_vec_push(vec, &r.d); bc_vec_pop(&p->results); // We need to return early to avoid executing code that we must // not touch. BC_SIG_UNLOCK; return; } // If we get here, we have a reference, but we need an array, so // dereference the array. else if (ref_size && t != BC_TYPE_REF) { v = bc_program_dereference(p, v); } } #endif // BC_ENABLED // If we get here, we need to copy the array because in bc, all // arguments are passed by value. Yes, this is expensive. bc_array_init(rv, true); bc_array_copy(rv, v); } // Push the vector onto the array stack and pop the source. bc_vec_push(vec, &r.d); bc_vec_pop(&p->results); BC_SIG_UNLOCK; } void bc_program_assignBuiltin(BcProgram* p, bool scale, bool obase, BcBigDig val) { BcBigDig* ptr_t; BcBigDig max, min; #if BC_ENABLED BcVec* v; BcBigDig* ptr; #endif // BC_ENABLED assert(!scale || !obase); // Scale needs handling separate from ibase and obase. if (scale) { // Set the min and max. min = 0; max = vm->maxes[BC_PROG_GLOBALS_SCALE]; #if BC_ENABLED // Get a pointer to the stack. v = p->globals_v + BC_PROG_GLOBALS_SCALE; #endif // BC_ENABLED // Get a pointer to the current value. ptr_t = p->globals + BC_PROG_GLOBALS_SCALE; } else { // Set the min and max. min = BC_NUM_MIN_BASE; if (BC_ENABLE_EXTRA_MATH && obase && (BC_IS_DC || !BC_IS_POSIX)) { min = 0; } max = vm->maxes[obase + BC_PROG_GLOBALS_IBASE]; #if BC_ENABLED // Get a pointer to the stack. v = p->globals_v + BC_PROG_GLOBALS_IBASE + obase; #endif // BC_ENABLED // Get a pointer to the current value. ptr_t = p->globals + BC_PROG_GLOBALS_IBASE + obase; } // Check for error. if (BC_ERR(val > max || val < min)) { BcErr e; // This grabs the right error. if (scale) e = BC_ERR_EXEC_SCALE; else if (obase) e = BC_ERR_EXEC_OBASE; else e = BC_ERR_EXEC_IBASE; bc_verr(e, min, max); } #if BC_ENABLED // Set the top of the stack. ptr = bc_vec_top(v); *ptr = val; #endif // BC_ENABLED // Set the actual global variable. *ptr_t = val; } #if BC_ENABLE_EXTRA_MATH void bc_program_assignSeed(BcProgram* p, BcNum* val) { bc_num_rng(val, &p->rng); } #endif // BC_ENABLE_EXTRA_MATH /** * Executes an assignment operator. * @param p The program. * @param inst The assignment operator to execute. */ static void bc_program_assign(BcProgram* p, uchar inst) { // The local use_val is true when the assigned value needs to be copied. BcResult* left; BcResult* right; BcResult res; BcNum* l; BcNum* r; bool ob, sc, use_val = BC_INST_USE_VAL(inst); bc_program_assignPrep(p, &left, &l, &right, &r); // Assigning to a string should be impossible simply because of the parse. assert(left->t != BC_RESULT_STR); // If we are assigning a string... if (right->t == BC_RESULT_STR) { assert(BC_PROG_STR(r)); #if BC_ENABLED if (inst != BC_INST_ASSIGN && inst != BC_INST_ASSIGN_NO_VAL) { bc_err(BC_ERR_EXEC_TYPE); } #endif // BC_ENABLED // If we are assigning to an array element... if (left->t == BC_RESULT_ARRAY_ELEM) { BC_SIG_LOCK; // We need to free the number and clear it. bc_num_free(l); // NOLINTNEXTLINE memcpy(l, r, sizeof(BcNum)); // Now we can pop the results. bc_vec_npop(&p->results, 2); BC_SIG_UNLOCK; } else { // If we get here, we are assigning to a variable, which we can use // bc_program_assignStr() for. BcVec* v = bc_program_vec(p, left->d.loc.loc, BC_TYPE_VAR); bc_program_assignStr(p, r, v, false); } #if BC_ENABLED // If this is true, the value is going to be used again, so we want to // push a temporary with the string. if (inst == BC_INST_ASSIGN) { res.t = BC_RESULT_STR; // NOLINTNEXTLINE memcpy(&res.d.n, r, sizeof(BcNum)); bc_vec_push(&p->results, &res); } #endif // BC_ENABLED // By using bc_program_assignStr(), we short-circuited this, so return. return; } // If we have a normal assignment operator, not a math one... if (BC_INST_IS_ASSIGN(inst)) { // Assigning to a variable that has a string here is fine because there // is no math done on it. // BC_RESULT_TEMP, BC_RESULT_IBASE, BC_RESULT_OBASE, BC_RESULT_SCALE, // and BC_RESULT_SEED all have temporary copies. Because that's the // case, we can free the left and just move the value over. We set the // type of right to BC_RESULT_ZERO in order to prevent it from being // freed. We also don't have to worry about BC_RESULT_STR because it's // take care of above. if (right->t == BC_RESULT_TEMP || right->t >= BC_RESULT_IBASE) { BC_SIG_LOCK; bc_num_free(l); // NOLINTNEXTLINE memcpy(l, r, sizeof(BcNum)); right->t = BC_RESULT_ZERO; BC_SIG_UNLOCK; } // Copy over. else bc_num_copy(l, r); } #if BC_ENABLED else { // If we get here, we are doing a math assignment (+=, -=, etc.). So // we need to prepare for a binary operator. BcBigDig scale = BC_PROG_SCALE(p); // At this point, the left side could still be a string because it could // be a variable that has the string. If that's the case, we have a type // error. if (BC_PROG_STR(l)) bc_err(BC_ERR_EXEC_TYPE); // Get the right type of assignment operator, whether val is used or // NO_VAL for performance. if (!use_val) { inst -= (BC_INST_ASSIGN_POWER_NO_VAL - BC_INST_ASSIGN_POWER); } assert(BC_NUM_RDX_VALID(l)); assert(BC_NUM_RDX_VALID(r)); // Run the actual operation. We do not need worry about reallocating l // because bc_num_binary() does that behind the scenes for us. bc_program_ops[inst - BC_INST_ASSIGN_POWER](l, r, l, scale); } #endif // BC_ENABLED ob = (left->t == BC_RESULT_OBASE); sc = (left->t == BC_RESULT_SCALE); // The globals need special handling, especially the non-seed ones. The // first part of the if statement handles them. if (ob || sc || left->t == BC_RESULT_IBASE) { // Get the actual value. BcBigDig val = bc_num_bigdig(l); bc_program_assignBuiltin(p, sc, ob, val); } #if BC_ENABLE_EXTRA_MATH // To assign to steed, let bc_num_rng() do its magic. else if (left->t == BC_RESULT_SEED) bc_program_assignSeed(p, l); #endif // BC_ENABLE_EXTRA_MATH BC_SIG_LOCK; // If we needed to use the value, then we need to copy it. Otherwise, we can // pop indiscriminately. Oh, and the copy should be a BC_RESULT_TEMP. if (use_val) { bc_num_createCopy(&res.d.n, l); res.t = BC_RESULT_TEMP; bc_vec_npop(&p->results, 2); bc_vec_push(&p->results, &res); } else bc_vec_npop(&p->results, 2); BC_SIG_UNLOCK; } /** * Pushes a variable's value onto the results stack. * @param p The program. * @param code The bytecode vector to pull the variable's index out of. * @param bgn An in/out parameter; the start of the index in the bytecode * vector, and will be updated to point after the index on return. * @param pop True if the variable's value should be popped off its stack. * This is only used in dc. * @param copy True if the variable's value should be copied to the results * stack. This is only used in dc. */ static void bc_program_pushVar(BcProgram* p, const char* restrict code, size_t* restrict bgn, bool pop, bool copy) { BcResult r; size_t idx = bc_program_index(code, bgn); BcVec* v; // Set the result appropriately. r.t = BC_RESULT_VAR; r.d.loc.loc = idx; // Get the stack for the variable. This is used in both bc and dc. v = bc_program_vec(p, idx, BC_TYPE_VAR); r.d.loc.stack_idx = v->len - 1; #if DC_ENABLED // If this condition is true, then we have the hard case, where we have to // adjust dc registers. if (BC_IS_DC && (pop || copy)) { // Get the number at the top at the top of the stack. BcNum* num = bc_vec_top(v); // Ensure there are enough elements on the stack. if (BC_ERR(!BC_PROG_STACK(v, 2 - copy))) { const char* name = bc_map_name(&p->var_map, idx); bc_verr(BC_ERR_EXEC_STACK_REGISTER, name); } assert(BC_PROG_STACK(v, 2 - copy)); // If the top of the stack is actually a number... if (!BC_PROG_STR(num)) { BC_SIG_LOCK; // Create a copy to go onto the results stack as appropriate. r.t = BC_RESULT_TEMP; bc_num_createCopy(&r.d.n, num); // If we are not actually copying, we need to do a replace, so pop. if (!copy) bc_vec_pop(v); bc_vec_push(&p->results, &r); BC_SIG_UNLOCK; return; } else { // Set the string result. We can just memcpy because all of the // fields in the num should be cleared. // NOLINTNEXTLINE memcpy(&r.d.n, num, sizeof(BcNum)); r.t = BC_RESULT_STR; } // If we are not actually copying, we need to do a replace, so pop. if (!copy) bc_vec_pop(v); } #endif // DC_ENABLED bc_vec_push(&p->results, &r); } /** * Pushes an array or an array element onto the results stack. * @param p The program. * @param code The bytecode vector to pull the variable's index out of. * @param bgn An in/out parameter; the start of the index in the bytecode * vector, and will be updated to point after the index on return. * @param inst The instruction; whether to push an array or an array element. */ static void bc_program_pushArray(BcProgram* p, const char* restrict code, size_t* restrict bgn, uchar inst) { BcResult r; BcResult* operand; BcNum* num; BcBigDig temp; BcVec* v; // Get the index of the array. r.d.loc.loc = bc_program_index(code, bgn); // We need the array to get its length. v = bc_program_vec(p, r.d.loc.loc, BC_TYPE_ARRAY); assert(v != NULL); r.d.loc.stack_idx = v->len - 1; // Doing an array is easy; just set the result type and finish. if (inst == BC_INST_ARRAY) { r.t = BC_RESULT_ARRAY; bc_vec_push(&p->results, &r); return; } // Grab the top element of the results stack for the array index. bc_program_prep(p, &operand, &num, 0); temp = bc_num_bigdig(num); // Set the result. r.t = BC_RESULT_ARRAY_ELEM; r.d.loc.idx = (size_t) temp; BC_SIG_LOCK; // Pop the index and push the element. bc_vec_pop(&p->results); bc_vec_push(&p->results, &r); BC_SIG_UNLOCK; } #if BC_ENABLED /** * Executes an increment or decrement operator. This only handles postfix * inc/dec because the parser translates prefix inc/dec into an assignment where * the value is used. * @param p The program. * @param inst The instruction; whether to do an increment or decrement. */ static void bc_program_incdec(BcProgram* p, uchar inst) { BcResult *ptr, res, copy; BcNum* num; uchar inst2; bc_program_prep(p, &ptr, &num, 0); BC_SIG_LOCK; // We need a copy from *before* the operation. copy.t = BC_RESULT_TEMP; bc_num_createCopy(©.d.n, num); BC_SETJMP_LOCKED(vm, exit); BC_SIG_UNLOCK; // Create the proper assignment. res.t = BC_RESULT_ONE; inst2 = BC_INST_ASSIGN_PLUS_NO_VAL + (inst & 0x01); bc_vec_push(&p->results, &res); bc_program_assign(p, inst2); BC_SIG_LOCK; bc_vec_push(&p->results, ©); BC_UNSETJMP(vm); BC_SIG_UNLOCK; // No need to free the copy here because we pushed it onto the stack. return; exit: BC_SIG_MAYLOCK; bc_num_free(©.d.n); BC_LONGJMP_CONT(vm); } /** * Executes a function call for bc. * @param p The program. * @param code The bytecode vector to pull the number of arguments and the * function index out of. * @param bgn An in/out parameter; the start of the indices in the bytecode * vector, and will be updated to point after the indices on * return. */ static void bc_program_call(BcProgram* p, const char* restrict code, size_t* restrict bgn) { BcInstPtr ip; size_t i, nargs; BcFunc* f; BcVec* v; BcAuto* a; BcResult* arg; // Pull the number of arguments out of the bytecode vector. nargs = bc_program_index(code, bgn); // Set up instruction pointer. ip.idx = 0; ip.func = bc_program_index(code, bgn); f = bc_vec_item(&p->fns, ip.func); // Error checking. if (BC_ERR(!f->code.len)) bc_verr(BC_ERR_EXEC_UNDEF_FUNC, f->name); if (BC_ERR(nargs != f->nparams)) { bc_verr(BC_ERR_EXEC_PARAMS, f->nparams, nargs); } // Set the length of the results stack. We discount the argument, of course. ip.len = p->results.len - nargs; assert(BC_PROG_STACK(&p->results, nargs)); // Prepare the globals' stacks. if (BC_G) bc_program_prepGlobals(p); // Push the arguments onto the stacks of their respective parameters. for (i = 0; i < nargs; ++i) { arg = bc_vec_top(&p->results); if (BC_ERR(arg->t == BC_RESULT_VOID)) bc_err(BC_ERR_EXEC_VOID_VAL); // Get the corresponding parameter. a = bc_vec_item(&f->autos, nargs - 1 - i); // Actually push the value onto the parameter's stack. bc_program_copyToVar(p, a->idx, a->type); } BC_SIG_LOCK; // Push zeroes onto the stacks of the auto variables. for (; i < f->autos.len; ++i) { // Get the auto and its stack. a = bc_vec_item(&f->autos, i); v = bc_program_vec(p, a->idx, a->type); // If a variable, just push a 0; otherwise, push an array. if (a->type == BC_TYPE_VAR) { BcNum* n = bc_vec_pushEmpty(v); bc_num_init(n, BC_NUM_DEF_SIZE); } else { BcVec* v2; assert(a->type == BC_TYPE_ARRAY); v2 = bc_vec_pushEmpty(v); bc_array_init(v2, true); } } // Push the instruction pointer onto the execution stack. bc_vec_push(&p->stack, &ip); BC_SIG_UNLOCK; } /** * Executes a return instruction. * @param p The program. * @param inst The return instruction. bc can return void, and we need to know * if it is. */ static void bc_program_return(BcProgram* p, uchar inst) { BcResult* res; BcFunc* f; BcInstPtr* ip; size_t i, nresults; // Get the instruction pointer. ip = bc_vec_top(&p->stack); // Get the difference between the actual number of results and the number of // results the caller expects. nresults = p->results.len - ip->len; // If this isn't true, there was a missing call somewhere. assert(BC_PROG_STACK(&p->stack, 2)); // If this isn't true, the parser screwed by giving us no value when we // expected one, or giving us a value when we expected none. assert(BC_PROG_STACK(&p->results, ip->len + (inst == BC_INST_RET))); // Get the function we are returning from. f = bc_vec_item(&p->fns, ip->func); res = bc_program_prepResult(p); // If we are returning normally... if (inst == BC_INST_RET) { BcNum* num; BcResult* operand; // Prepare and copy the return value. bc_program_operand(p, &operand, &num, 1); if (BC_PROG_STR(num)) { // We need to set this because otherwise, it will be a // BC_RESULT_TEMP, and BC_RESULT_TEMP needs an actual number to make // it easier to do type checking. res->t = BC_RESULT_STR; // NOLINTNEXTLINE memcpy(&res->d.n, num, sizeof(BcNum)); } else { BC_SIG_LOCK; bc_num_createCopy(&res->d.n, num); } } // Void is easy; set the result. else if (inst == BC_INST_RET_VOID) res->t = BC_RESULT_VOID; else { BC_SIG_LOCK; // If we get here, the instruction is for returning a zero, so do that. bc_num_init(&res->d.n, BC_NUM_DEF_SIZE); } BC_SIG_MAYUNLOCK; // We need to pop items off of the stacks of arguments and autos as well. for (i = 0; i < f->autos.len; ++i) { BcAuto* a = bc_vec_item(&f->autos, i); BcVec* v = bc_program_vec(p, a->idx, a->type); bc_vec_pop(v); } BC_SIG_LOCK; // When we retire, pop all of the unused results. bc_program_retire(p, 1, nresults); // Pop the globals, if necessary. if (BC_G) bc_program_popGlobals(p, false); // Pop the stack. This is what causes the function to actually "return." bc_vec_pop(&p->stack); BC_SIG_UNLOCK; } #endif // BC_ENABLED /** * Executes a builtin function. * @param p The program. * @param inst The builtin to execute. */ static void bc_program_builtin(BcProgram* p, uchar inst) { BcResult* opd; BcResult* res; BcNum* num; bool len = (inst == BC_INST_LENGTH); // Ensure we have a valid builtin. #if BC_ENABLE_EXTRA_MATH assert(inst >= BC_INST_LENGTH && inst <= BC_INST_IRAND); #else // BC_ENABLE_EXTRA_MATH assert(inst >= BC_INST_LENGTH && inst <= BC_INST_IS_STRING); #endif // BC_ENABLE_EXTRA_MATH #ifndef BC_PROG_NO_STACK_CHECK // Check stack for dc. if (BC_IS_DC && BC_ERR(!BC_PROG_STACK(&p->results, 1))) { bc_err(BC_ERR_EXEC_STACK); } #endif // BC_PROG_NO_STACK_CHECK assert(BC_PROG_STACK(&p->results, 1)); res = bc_program_prepResult(p); bc_program_operand(p, &opd, &num, 1); assert(num != NULL); // We need to ensure that strings and arrays aren't passed to most builtins. // The scale function can take strings in dc. if (!len && (inst != BC_INST_SCALE_FUNC || BC_IS_BC) && inst != BC_INST_IS_NUMBER && inst != BC_INST_IS_STRING) { bc_program_type_num(opd, num); } // Square root is easy. if (inst == BC_INST_SQRT) bc_num_sqrt(num, &res->d.n, BC_PROG_SCALE(p)); // Absolute value is easy. else if (inst == BC_INST_ABS) { BC_SIG_LOCK; bc_num_createCopy(&res->d.n, num); BC_SIG_UNLOCK; BC_NUM_NEG_CLR_NP(res->d.n); } // Testing for number or string is easy. else if (inst == BC_INST_IS_NUMBER || inst == BC_INST_IS_STRING) { bool cond; bool is_str; BC_SIG_LOCK; bc_num_init(&res->d.n, BC_NUM_DEF_SIZE); BC_SIG_UNLOCK; // Test if the number is a string. is_str = BC_PROG_STR(num); // This confusing condition simply means that the instruction must be // true if is_str is, or it must be false if is_str is. Otherwise, the // returned value is false (0). cond = ((inst == BC_INST_IS_STRING) == is_str); if (cond) bc_num_one(&res->d.n); } #if BC_ENABLE_EXTRA_MATH // irand() is easy. else if (inst == BC_INST_IRAND) { BC_SIG_LOCK; bc_num_init(&res->d.n, num->len - BC_NUM_RDX_VAL(num)); BC_SIG_UNLOCK; bc_num_irand(num, &res->d.n, &p->rng); } #endif // BC_ENABLE_EXTRA_MATH // Everything else is...not easy. else { BcBigDig val = 0; // Well, scale() is easy, but length() is not. if (len) { // If we are bc and we have an array... if (opd->t == BC_RESULT_ARRAY) { // Yes, this is one place where we need to cast the number from // bc_program_num() to a vector. BcVec* v = (BcVec*) num; // XXX: If this is changed, you should also change the similar // code in bc_program_asciify(). #if BC_ENABLED // Dereference the array, if necessary. if (BC_IS_BC && v->size == sizeof(uchar)) { v = bc_program_dereference(p, v); } #endif // BC_ENABLED assert(v->size == sizeof(BcNum)); val = (BcBigDig) v->len; } else { // If the item is a string... if (!BC_PROG_NUM(opd, num)) { char* str; // Get the string, then get the length. str = bc_program_string(p, num); val = (BcBigDig) strlen(str); } else { // Calculate the length of the number. val = (BcBigDig) bc_num_len(num); } } } // Like I said; scale() is actually easy. It just also needs the integer // conversion that length() does. else if (BC_IS_BC || BC_PROG_NUM(opd, num)) { val = (BcBigDig) bc_num_scale(num); } BC_SIG_LOCK; // Create the result. bc_num_createFromBigdig(&res->d.n, val); BC_SIG_UNLOCK; } bc_program_retire(p, 1, 1); } /** * Executes a divmod. * @param p The program. */ static void bc_program_divmod(BcProgram* p) { BcResult* opd1; BcResult* opd2; BcResult* res; BcResult* res2; BcNum* n1; BcNum* n2; size_t req; // We grow first to avoid pointer invalidation. bc_vec_grow(&p->results, 2); // We don't need to update the pointer because // the capacity is enough due to the line above. res2 = bc_program_prepResult(p); res = bc_program_prepResult(p); // Prepare the operands. bc_program_binOpPrep(p, &opd1, &n1, &opd2, &n2, 2); req = bc_num_mulReq(n1, n2, BC_PROG_SCALE(p)); BC_SIG_LOCK; // Initialize the results. bc_num_init(&res->d.n, req); bc_num_init(&res2->d.n, req); BC_SIG_UNLOCK; // Execute. bc_num_divmod(n1, n2, &res2->d.n, &res->d.n, BC_PROG_SCALE(p)); bc_program_retire(p, 2, 2); } /** * Executes modular exponentiation. * @param p The program. */ static void bc_program_modexp(BcProgram* p) { BcResult* r1; BcResult* r2; BcResult* r3; BcResult* res; BcNum* n1; BcNum* n2; BcNum* n3; #if DC_ENABLED // Check the stack. if (BC_IS_DC && BC_ERR(!BC_PROG_STACK(&p->results, 3))) { bc_err(BC_ERR_EXEC_STACK); } #endif // DC_ENABLED assert(BC_PROG_STACK(&p->results, 3)); res = bc_program_prepResult(p); // Get the first operand and typecheck. bc_program_operand(p, &r1, &n1, 3); bc_program_type_num(r1, n1); // Get the last two operands. bc_program_binOpPrep(p, &r2, &n2, &r3, &n3, 1); // Make sure that the values have their pointers updated, if necessary. // Only array elements are possible because this is dc. if (r1->t == BC_RESULT_ARRAY_ELEM && (r1->t == r2->t || r1->t == r3->t)) { n1 = bc_program_num(p, r1); } BC_SIG_LOCK; bc_num_init(&res->d.n, n3->len); BC_SIG_UNLOCK; bc_num_modexp(n1, n2, n3, &res->d.n); bc_program_retire(p, 1, 3); } /** * Asciifies a number for dc. This is a helper for bc_program_asciify(). * @param p The program. * @param n The number to asciify. */ static uchar bc_program_asciifyNum(BcProgram* p, BcNum* n) { bc_num_copy(&p->asciify, n); // We want to clear the scale and sign for easy mod later. bc_num_truncate(&p->asciify, p->asciify.scale); BC_NUM_NEG_CLR(&p->asciify); // This is guaranteed to not have a divide by 0 // because strmb is equal to 256. bc_num_mod(&p->asciify, &p->strmb, &p->asciify, 0); // This is also guaranteed to not error because num is in the range // [0, UCHAR_MAX], which is definitely in range for a BcBigDig. And // it is not negative. return (uchar) bc_num_bigdig2(&p->asciify); } /** * Executes the "asciify" command in bc and dc. * @param p The program. */ static void bc_program_asciify(BcProgram* p) { BcResult *r, res; BcNum* n; uchar c; size_t idx; #if BC_ENABLED // This is in the outer scope because it has to be freed after a jump. char* temp_str; #endif // BC_ENABLED // Check the stack. if (BC_ERR(!BC_PROG_STACK(&p->results, 1))) bc_err(BC_ERR_EXEC_STACK); assert(BC_PROG_STACK(&p->results, 1)); // Get the top of the results stack. bc_program_operand(p, &r, &n, 0); assert(n != NULL); assert(BC_IS_BC || r->t != BC_RESULT_ARRAY); #if BC_ENABLED // Handle arrays in bc specially. if (r->t == BC_RESULT_ARRAY) { // Yes, this is one place where we need to cast the number from // bc_program_num() to a vector. BcVec* v = (BcVec*) n; size_t i; // XXX: If this is changed, you should also change the similar code in // bc_program_builtin(). // Dereference the array, if necessary. if (v->size == sizeof(uchar)) { v = bc_program_dereference(p, v); } assert(v->size == sizeof(BcNum)); // Allocate the string and set the jump for it. BC_SIG_LOCK; temp_str = bc_vm_malloc(v->len + 1); BC_SETJMP_LOCKED(vm, exit); BC_SIG_UNLOCK; // Convert the array. for (i = 0; i < v->len; ++i) { BcNum* num = (BcNum*) bc_vec_item(v, i); if (BC_PROG_STR(num)) { temp_str[i] = (bc_program_string(p, num))[0]; } else { temp_str[i] = (char) bc_program_asciifyNum(p, num); } } temp_str[v->len] = '\0'; // Store the string in the slab and map, and free the temp string. BC_SIG_LOCK; idx = bc_program_addString(p, temp_str); free(temp_str); BC_UNSETJMP(vm); BC_SIG_UNLOCK; } else #endif // BC_ENABLED { char str[2]; char* str2; // Asciify. if (BC_PROG_NUM(r, n)) c = bc_program_asciifyNum(p, n); else { // Get the string itself, then the first character. str2 = bc_program_string(p, n); c = (uchar) str2[0]; } // Fill the resulting string. str[0] = (char) c; str[1] = '\0'; // Add the string to the data structures. BC_SIG_LOCK; idx = bc_program_addString(p, str); BC_SIG_UNLOCK; } // Set the result res.t = BC_RESULT_STR; bc_num_clear(&res.d.n); res.d.n.scale = idx; // Pop and push. bc_vec_pop(&p->results); bc_vec_push(&p->results, &res); return; #if BC_ENABLED exit: free(temp_str); #endif // BC_ENABLED } /** * Streams a number or a string to stdout. * @param p The program. */ static void bc_program_printStream(BcProgram* p) { BcResult* r; BcNum* n; // Check the stack. if (BC_ERR(!BC_PROG_STACK(&p->results, 1))) bc_err(BC_ERR_EXEC_STACK); assert(BC_PROG_STACK(&p->results, 1)); // Get the top of the results stack. bc_program_operand(p, &r, &n, 0); assert(n != NULL); // Stream appropriately. if (BC_PROG_NUM(r, n)) bc_num_stream(n); else bc_program_printChars(bc_program_string(p, n)); // Pop the operand. bc_vec_pop(&p->results); } #if DC_ENABLED /** * Gets the length of a register in dc and pushes it onto the results stack. * @param p The program. * @param code The bytecode vector to pull the register's index out of. * @param bgn An in/out parameter; the start of the index in the bytecode * vector, and will be updated to point after the index on return. */ static void bc_program_regStackLen(BcProgram* p, const char* restrict code, size_t* restrict bgn) { size_t idx = bc_program_index(code, bgn); BcVec* v = bc_program_vec(p, idx, BC_TYPE_VAR); bc_program_pushBigdig(p, (BcBigDig) v->len, BC_RESULT_TEMP); } /** * Pushes the length of the results stack onto the results stack. * @param p The program. */ static void bc_program_stackLen(BcProgram* p) { bc_program_pushBigdig(p, (BcBigDig) p->results.len, BC_RESULT_TEMP); } /** * Pops a certain number of elements off the execution stack. * @param p The program. * @param inst The instruction to tell us how many. There is one to pop up to * 2, and one to pop the amount equal to the number at the top of * the results stack. */ static void bc_program_nquit(BcProgram* p, uchar inst) { BcResult* opnd; BcNum* num; BcBigDig val; size_t i; // Ensure that the tail calls stack is correct. assert(p->stack.len == p->tail_calls.len); // Get the number of executions to pop. if (inst == BC_INST_QUIT) val = 2; else { bc_program_prep(p, &opnd, &num, 0); val = bc_num_bigdig(num); bc_vec_pop(&p->results); } // Loop over the tail call stack and adjust the quit value appropriately. for (i = 0; val && i < p->tail_calls.len; ++i) { // Get the number of tail calls for this one. size_t calls = *((size_t*) bc_vec_item_rev(&p->tail_calls, i)) + 1; // Adjust the value. if (calls >= val) val = 0; else val -= (BcBigDig) calls; } // If we don't have enough executions, just quit. if (i == p->stack.len) { vm->status = BC_STATUS_QUIT; BC_JMP; } else { // We can always pop the last item we reached on the tail call stack // because these are for tail calls. That means that any executions that // we would not have quit in that position on the stack would have quit // anyway. BC_SIG_LOCK; bc_vec_npop(&p->stack, i); bc_vec_npop(&p->tail_calls, i); BC_SIG_UNLOCK; } } /** * Pushes the depth of the execution stack onto the stack. * @param p The program. */ static void bc_program_execStackLen(BcProgram* p) { size_t i, amt, len = p->tail_calls.len; amt = len; for (i = 0; i < len; ++i) { amt += *((size_t*) bc_vec_item(&p->tail_calls, i)); } bc_program_pushBigdig(p, (BcBigDig) amt, BC_RESULT_TEMP); } /** * * @param p The program. * @param code The bytecode vector to pull the register's index out of. * @param bgn An in/out parameter; the start of the index in the bytecode * vector, and will be updated to point after the index on return. * @param cond True if the execution is conditional, false otherwise. * @param len The number of bytes in the bytecode vector. */ static void bc_program_execStr(BcProgram* p, const char* restrict code, size_t* restrict bgn, bool cond, size_t len) { BcResult* r; char* str; BcFunc* f; BcInstPtr ip; size_t fidx; BcNum* n; assert(p->stack.len == p->tail_calls.len); // Check the stack. if (BC_ERR(!BC_PROG_STACK(&p->results, 1))) bc_err(BC_ERR_EXEC_STACK); assert(BC_PROG_STACK(&p->results, 1)); // Get the operand. bc_program_operand(p, &r, &n, 0); // If execution is conditional... if (cond) { bool exec; size_t then_idx; // These are volatile to quiet warnings on GCC about clobbering with // longjmp(). volatile size_t else_idx; volatile size_t idx; // Get the index of the "then" var and "else" var. then_idx = bc_program_index(code, bgn); else_idx = bc_program_index(code, bgn); // Figure out if we should execute. exec = (r->d.n.len != 0); idx = exec ? then_idx : else_idx; BC_SIG_LOCK; BC_SETJMP_LOCKED(vm, exit); // If we are supposed to execute, execute. If else_idx == SIZE_MAX, that // means there was no else clause, so if execute is false and else does // not exist, we don't execute. The goto skips all of the setup for the // execution. if (exec || (else_idx != SIZE_MAX)) { n = bc_vec_top(bc_program_vec(p, idx, BC_TYPE_VAR)); } else goto exit; if (BC_ERR(!BC_PROG_STR(n))) bc_err(BC_ERR_EXEC_TYPE); BC_UNSETJMP(vm); BC_SIG_UNLOCK; } else { // In non-conditional situations, only the top of stack can be executed, // and in those cases, variables are not allowed to be "on the stack"; // they are only put on the stack to be assigned to. assert(r->t != BC_RESULT_VAR); if (r->t != BC_RESULT_STR) return; } assert(BC_PROG_STR(n)); // Get the string. str = bc_program_string(p, n); // Get the function index and function. BC_SIG_LOCK; fidx = bc_program_insertFunc(p, str); BC_SIG_UNLOCK; f = bc_vec_item(&p->fns, fidx); // If the function has not been parsed yet... if (!f->code.len) { BC_SIG_LOCK; if (!BC_PARSE_IS_INITED(&vm->read_prs, p)) { bc_parse_init(&vm->read_prs, p, fidx); // Initialize this too because bc_vm_shutdown() expects them to be // initialized togther. bc_vec_init(&vm->read_buf, sizeof(char), BC_DTOR_NONE); } // This needs to be updated because the parser could have been used // somewhere else else bc_parse_updateFunc(&vm->read_prs, fidx); bc_lex_file(&vm->read_prs.l, vm->file); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; // Parse. Only one expression is needed, so stdin isn't used. bc_parse_text(&vm->read_prs, str, BC_MODE_FILE); BC_SIG_LOCK; vm->expr(&vm->read_prs, BC_PARSE_NOCALL); BC_UNSETJMP(vm); // We can just assert this here because // dc should parse everything until EOF. assert(vm->read_prs.l.t == BC_LEX_EOF); BC_SIG_UNLOCK; } // Set the instruction pointer. ip.idx = 0; ip.len = p->results.len; ip.func = fidx; BC_SIG_LOCK; // Pop the operand. bc_vec_pop(&p->results); // Tail call processing. This condition means that there is more on the // execution stack, and we are at the end of the bytecode vector, and the // last instruction is just a BC_INST_POP_EXEC, which would return. if (p->stack.len > 1 && *bgn == len - 1 && code[*bgn] == BC_INST_POP_EXEC) { size_t* call_ptr = bc_vec_top(&p->tail_calls); // Add one to the tail call. *call_ptr += 1; // Pop the execution stack before pushing the new instruction pointer // on. bc_vec_pop(&p->stack); } // If not a tail call, just push a new one. else bc_vec_push(&p->tail_calls, &ip.idx); // Push the new function onto the execution stack and return. bc_vec_push(&p->stack, &ip); BC_SIG_UNLOCK; return; err: BC_SIG_MAYLOCK; f = bc_vec_item(&p->fns, fidx); // Make sure to erase the bytecode vector so dc knows it is not parsed. bc_vec_popAll(&f->code); exit: bc_vec_pop(&p->results); BC_LONGJMP_CONT(vm); } /** * Prints every item on the results stack, one per line. * @param p The program. */ static void bc_program_printStack(BcProgram* p) { size_t idx; for (idx = 0; idx < p->results.len; ++idx) { bc_program_print(p, BC_INST_PRINT, idx); } } #endif // DC_ENABLED /** * Pushes the value of a global onto the results stack. * @param p The program. * @param inst Which global to push, as an instruction. */ static void bc_program_pushGlobal(BcProgram* p, uchar inst) { BcResultType t; // Make sure the instruction is valid. assert(inst >= BC_INST_IBASE && inst <= BC_INST_SCALE); // Push the global. t = inst - BC_INST_IBASE + BC_RESULT_IBASE; bc_program_pushBigdig(p, p->globals[inst - BC_INST_IBASE], t); } /** * Pushes the value of a global setting onto the stack. * @param p The program. * @param inst Which global setting to push, as an instruction. */ static void bc_program_globalSetting(BcProgram* p, uchar inst) { BcBigDig val; // Make sure the instruction is valid. #if DC_ENABLED assert((inst >= BC_INST_LINE_LENGTH && inst <= BC_INST_LEADING_ZERO) || (BC_IS_DC && inst == BC_INST_EXTENDED_REGISTERS)); #else // DC_ENABLED assert(inst >= BC_INST_LINE_LENGTH && inst <= BC_INST_LEADING_ZERO); #endif // DC_ENABLED if (inst == BC_INST_LINE_LENGTH) { val = (BcBigDig) vm->line_len; } #if BC_ENABLED else if (inst == BC_INST_GLOBAL_STACKS) { val = (BC_G != 0); } #endif // BC_ENABLED #if DC_ENABLED else if (inst == BC_INST_EXTENDED_REGISTERS) { val = (DC_X != 0); } #endif // DC_ENABLED else val = (BC_Z != 0); // Push the global. bc_program_pushBigdig(p, val, BC_RESULT_TEMP); } #if BC_ENABLE_EXTRA_MATH /** * Pushes the value of seed on the stack. * @param p The program. */ static void bc_program_pushSeed(BcProgram* p) { BcResult* res; res = bc_program_prepResult(p); res->t = BC_RESULT_SEED; BC_SIG_LOCK; // We need 2*BC_RAND_NUM_SIZE because of the size of the state. bc_num_init(&res->d.n, 2 * BC_RAND_NUM_SIZE); BC_SIG_UNLOCK; bc_num_createFromRNG(&res->d.n, &p->rng); } #endif // BC_ENABLE_EXTRA_MATH /** * Adds a function to the fns array. The function's ID must have already been * inserted into the map. * @param p The program. * @param id_ptr The ID of the function as inserted into the map. */ static void bc_program_addFunc(BcProgram* p, BcId* id_ptr) { BcFunc* f; BC_SIG_ASSERT_LOCKED; // Push and init. f = bc_vec_pushEmpty(&p->fns); bc_func_init(f, id_ptr->name); } size_t bc_program_insertFunc(BcProgram* p, const char* name) { BcId* id_ptr; bool new; size_t idx; BC_SIG_ASSERT_LOCKED; assert(p != NULL && name != NULL); // Insert into the map and get the resulting ID. new = bc_map_insert(&p->fn_map, name, p->fns.len, &idx); id_ptr = (BcId*) bc_vec_item(&p->fn_map, idx); idx = id_ptr->idx; // If the function is new... if (new) { // Add the function to the fns array. bc_program_addFunc(p, id_ptr); } #if BC_ENABLED // bc has to reset the function because it's about to be redefined. else if (BC_IS_BC) { BcFunc* func = bc_vec_item(&p->fns, idx); bc_func_reset(func); } #endif // BC_ENABLED return idx; } -#if BC_DEBUG +#if BC_DEBUG || BC_ENABLE_MEMCHECK void bc_program_free(BcProgram* p) { #if BC_ENABLED size_t i; #endif // BC_ENABLED BC_SIG_ASSERT_LOCKED; assert(p != NULL); #if BC_ENABLED // Free the globals stacks. for (i = 0; i < BC_PROG_GLOBALS_LEN; ++i) { bc_vec_free(p->globals_v + i); } #endif // BC_ENABLED bc_vec_free(&p->fns); bc_vec_free(&p->fn_map); bc_vec_free(&p->vars); bc_vec_free(&p->var_map); bc_vec_free(&p->arrs); bc_vec_free(&p->arr_map); bc_vec_free(&p->results); bc_vec_free(&p->stack); bc_vec_free(&p->consts); bc_vec_free(&p->const_map); bc_vec_free(&p->strs); bc_vec_free(&p->str_map); bc_num_free(&p->asciify); #if BC_ENABLED if (BC_IS_BC) bc_num_free(&p->last); #endif // BC_ENABLED #if BC_ENABLE_EXTRA_MATH bc_rand_free(&p->rng); #endif // BC_ENABLE_EXTRA_MATH #if DC_ENABLED if (BC_IS_DC) bc_vec_free(&p->tail_calls); #endif // DC_ENABLED } -#endif // BC_DEBUG +#endif // BC_DEBUG || BC_ENABLE_MEMCHECK void bc_program_init(BcProgram* p) { BcInstPtr ip; size_t i; BC_SIG_ASSERT_LOCKED; assert(p != NULL); // We want this clear. // NOLINTNEXTLINE memset(&ip, 0, sizeof(BcInstPtr)); // Setup the globals stacks and the current values. for (i = 0; i < BC_PROG_GLOBALS_LEN; ++i) { BcBigDig val = i == BC_PROG_GLOBALS_SCALE ? 0 : BC_BASE; #if BC_ENABLED bc_vec_init(p->globals_v + i, sizeof(BcBigDig), BC_DTOR_NONE); bc_vec_push(p->globals_v + i, &val); #endif // BC_ENABLED p->globals[i] = val; } #if DC_ENABLED // dc-only setup. if (BC_IS_DC) { bc_vec_init(&p->tail_calls, sizeof(size_t), BC_DTOR_NONE); // We want an item for the main function on the tail call stack. i = 0; bc_vec_push(&p->tail_calls, &i); } #endif // DC_ENABLED bc_num_setup(&p->strmb, p->strmb_num, BC_NUM_BIGDIG_LOG10); bc_num_bigdig2num(&p->strmb, BC_NUM_STREAM_BASE); bc_num_init(&p->asciify, BC_NUM_DEF_SIZE); #if BC_ENABLE_EXTRA_MATH // We need to initialize srand() just in case /dev/urandom and /dev/random // are not available. srand((unsigned int) time(NULL)); bc_rand_init(&p->rng); #endif // BC_ENABLE_EXTRA_MATH #if BC_ENABLED if (BC_IS_BC) bc_num_init(&p->last, BC_NUM_DEF_SIZE); #endif // BC_ENABLED #if BC_DEBUG bc_vec_init(&p->fns, sizeof(BcFunc), BC_DTOR_FUNC); #else // BC_DEBUG bc_vec_init(&p->fns, sizeof(BcFunc), BC_DTOR_NONE); #endif // BC_DEBUG bc_map_init(&p->fn_map); bc_program_insertFunc(p, bc_func_main); bc_program_insertFunc(p, bc_func_read); bc_vec_init(&p->vars, sizeof(BcVec), BC_DTOR_VEC); bc_map_init(&p->var_map); bc_vec_init(&p->arrs, sizeof(BcVec), BC_DTOR_VEC); bc_map_init(&p->arr_map); bc_vec_init(&p->results, sizeof(BcResult), BC_DTOR_RESULT); // Push the first instruction pointer onto the execution stack. bc_vec_init(&p->stack, sizeof(BcInstPtr), BC_DTOR_NONE); bc_vec_push(&p->stack, &ip); bc_vec_init(&p->consts, sizeof(BcConst), BC_DTOR_CONST); bc_map_init(&p->const_map); bc_vec_init(&p->strs, sizeof(char*), BC_DTOR_NONE); bc_map_init(&p->str_map); } void bc_program_printStackTrace(BcProgram* p) { size_t i, max_digits; max_digits = bc_vm_numDigits(p->stack.len - 1); for (i = 0; i < p->stack.len; ++i) { BcInstPtr* ip = bc_vec_item_rev(&p->stack, i); BcFunc* f = bc_vec_item(&p->fns, ip->func); size_t j, digits; digits = bc_vm_numDigits(i); bc_file_puts(&vm->ferr, bc_flush_none, " "); for (j = 0; j < max_digits - digits; ++j) { bc_file_putchar(&vm->ferr, bc_flush_none, ' '); } bc_file_printf(&vm->ferr, "%zu: %s", i, f->name); #if BC_ENABLED if (BC_IS_BC && ip->func != BC_PROG_MAIN && ip->func != BC_PROG_READ) { bc_file_puts(&vm->ferr, bc_flush_none, "()"); } #endif // BC_ENABLED bc_file_putchar(&vm->ferr, bc_flush_none, '\n'); } } void bc_program_reset(BcProgram* p) { BcFunc* f; BcInstPtr* ip; BC_SIG_ASSERT_LOCKED; - // Pop all but the last execution and all results. + // Pop all but the last execution. bc_vec_npop(&p->stack, p->stack.len - 1); - bc_vec_popAll(&p->results); #if DC_ENABLED // We need to pop tail calls too. if (BC_IS_DC) bc_vec_npop(&p->tail_calls, p->tail_calls.len - 1); #endif // DC_ENABLED #if BC_ENABLED + // Clear the stack if we are in bc. We have to do this in bc because bc's + // stack is implicit. + // + // XXX: We don't do this in dc because other dc implementations don't. + if (BC_IS_BC || !BC_I) bc_vec_popAll(&p->results); + // Clear the globals' stacks. if (BC_G) bc_program_popGlobals(p, true); #endif // BC_ENABLED // Clear the bytecode vector of the main function. f = bc_vec_item(&p->fns, BC_PROG_MAIN); bc_vec_npop(&f->code, f->code.len); // Reset the instruction pointer. ip = bc_vec_top(&p->stack); // NOLINTNEXTLINE memset(ip, 0, sizeof(BcInstPtr)); if (BC_SIG_INTERRUPT(vm)) { // Write the ready message for a signal. bc_file_printf(&vm->fout, "%s", bc_program_ready_msg); bc_file_flush(&vm->fout, bc_flush_err); } // Clear the signal. vm->sig = 0; } void bc_program_exec(BcProgram* p) { size_t idx; BcResult r; BcResult* ptr; BcInstPtr* ip; BcFunc* func; char* code; bool cond = false; uchar inst; #if BC_ENABLED BcNum* num; #endif // BC_ENABLED #if !BC_HAS_COMPUTED_GOTO #if BC_DEBUG size_t jmp_bufs_len; #endif // BC_DEBUG #endif // !BC_HAS_COMPUTED_GOTO #if BC_HAS_COMPUTED_GOTO #if BC_GCC #pragma GCC diagnostic ignored "-Wpedantic" #endif // BC_GCC #if BC_CLANG #pragma clang diagnostic ignored "-Wgnu-label-as-value" #endif // BC_CLANG BC_PROG_LBLS; BC_PROG_LBLS_ASSERT; #if BC_CLANG #pragma clang diagnostic warning "-Wgnu-label-as-value" #endif // BC_CLANG #if BC_GCC #pragma GCC diagnostic warning "-Wpedantic" #endif // BC_GCC // BC_INST_INVALID is a marker for the end so that we don't have to have an // execution loop. func = (BcFunc*) bc_vec_item(&p->fns, BC_PROG_MAIN); bc_vec_pushByte(&func->code, BC_INST_INVALID); #endif // BC_HAS_COMPUTED_GOTO BC_SETJMP(vm, end); ip = bc_vec_top(&p->stack); func = (BcFunc*) bc_vec_item(&p->fns, ip->func); code = func->code.v; #if !BC_HAS_COMPUTED_GOTO #if BC_DEBUG jmp_bufs_len = vm->jmp_bufs.len; #endif // BC_DEBUG // This loop is the heart of the execution engine. It *is* the engine. For // computed goto, it is ignored. while (ip->idx < func->code.len) #endif // !BC_HAS_COMPUTED_GOTO { BC_SIG_ASSERT_NOT_LOCKED; #if BC_HAS_COMPUTED_GOTO #if BC_GCC #pragma GCC diagnostic ignored "-Wpedantic" #endif // BC_GCC #if BC_CLANG #pragma clang diagnostic ignored "-Wgnu-label-as-value" #endif // BC_CLANG BC_PROG_JUMP(inst, code, ip); #else // BC_HAS_COMPUTED_GOTO // Get the next instruction and increment the index. inst = (uchar) code[(ip->idx)++]; #endif // BC_HAS_COMPUTED_GOTO #if BC_DEBUG_CODE bc_file_printf(&vm->ferr, "inst: %s\n", bc_inst_names[inst]); bc_file_flush(&vm->ferr, bc_flush_none); #endif // BC_DEBUG_CODE #if !BC_HAS_COMPUTED_GOTO switch (inst) #endif // !BC_HAS_COMPUTED_GOTO { #if BC_ENABLED // This just sets up the condition for the unconditional jump below, // which checks the condition, if necessary. // clang-format off BC_PROG_LBL(BC_INST_JUMP_ZERO): // clang-format on { bc_program_prep(p, &ptr, &num, 0); cond = !bc_num_cmpZero(num); bc_vec_pop(&p->results); BC_PROG_DIRECT_JUMP(BC_INST_JUMP) } // Fallthrough. BC_PROG_FALLTHROUGH // clang-format off BC_PROG_LBL(BC_INST_JUMP): // clang-format on { idx = bc_program_index(code, &ip->idx); // If a jump is required... if (inst == BC_INST_JUMP || cond) { // Get the address to jump to. size_t* addr = bc_vec_item(&func->labels, idx); // If this fails, then the parser failed to set up the // labels correctly. assert(*addr != SIZE_MAX); // Set the new address. ip->idx = *addr; } BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_CALL): // clang-format on { assert(BC_IS_BC); bc_program_call(p, code, &ip->idx); // Because we changed the execution stack and where we are // executing, we have to update all of this. BC_SIG_LOCK; ip = bc_vec_top(&p->stack); func = bc_vec_item(&p->fns, ip->func); code = func->code.v; BC_SIG_UNLOCK; BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_INC): BC_PROG_LBL(BC_INST_DEC): // clang-format on { bc_program_incdec(p, inst); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_HALT): // clang-format on { vm->status = BC_STATUS_QUIT; // Just jump out. The jump series will take care of everything. BC_JMP; BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_RET): BC_PROG_LBL(BC_INST_RET0): BC_PROG_LBL(BC_INST_RET_VOID): // clang-format on { bc_program_return(p, inst); // Because we changed the execution stack and where we are // executing, we have to update all of this. BC_SIG_LOCK; ip = bc_vec_top(&p->stack); func = bc_vec_item(&p->fns, ip->func); code = func->code.v; BC_SIG_UNLOCK; BC_PROG_JUMP(inst, code, ip); } #endif // BC_ENABLED // clang-format off BC_PROG_LBL(BC_INST_BOOL_OR): BC_PROG_LBL(BC_INST_BOOL_AND): BC_PROG_LBL(BC_INST_REL_EQ): BC_PROG_LBL(BC_INST_REL_LE): BC_PROG_LBL(BC_INST_REL_GE): BC_PROG_LBL(BC_INST_REL_NE): BC_PROG_LBL(BC_INST_REL_LT): BC_PROG_LBL(BC_INST_REL_GT): // clang-format on { bc_program_logical(p, inst); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_READ): // clang-format on { // We want to flush output before // this in case there is a prompt. bc_file_flush(&vm->fout, bc_flush_save); bc_program_read(p); // Because we changed the execution stack and where we are // executing, we have to update all of this. BC_SIG_LOCK; ip = bc_vec_top(&p->stack); func = bc_vec_item(&p->fns, ip->func); code = func->code.v; BC_SIG_UNLOCK; BC_PROG_JUMP(inst, code, ip); } #if BC_ENABLE_EXTRA_MATH // clang-format off BC_PROG_LBL(BC_INST_RAND): // clang-format on { bc_program_rand(p); BC_PROG_JUMP(inst, code, ip); } #endif // BC_ENABLE_EXTRA_MATH // clang-format off BC_PROG_LBL(BC_INST_MAXIBASE): BC_PROG_LBL(BC_INST_MAXOBASE): BC_PROG_LBL(BC_INST_MAXSCALE): #if BC_ENABLE_EXTRA_MATH BC_PROG_LBL(BC_INST_MAXRAND): #endif // BC_ENABLE_EXTRA_MATH // clang-format on { BcBigDig dig = vm->maxes[inst - BC_INST_MAXIBASE]; bc_program_pushBigdig(p, dig, BC_RESULT_TEMP); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_LINE_LENGTH): #if BC_ENABLED BC_PROG_LBL(BC_INST_GLOBAL_STACKS): #endif // BC_ENABLED #if DC_ENABLED BC_PROG_LBL(BC_INST_EXTENDED_REGISTERS): #endif // DC_ENABLE BC_PROG_LBL(BC_INST_LEADING_ZERO): // clang-format on { bc_program_globalSetting(p, inst); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_VAR): // clang-format on { bc_program_pushVar(p, code, &ip->idx, false, false); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_ARRAY_ELEM): BC_PROG_LBL(BC_INST_ARRAY): // clang-format on { bc_program_pushArray(p, code, &ip->idx, inst); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_IBASE): BC_PROG_LBL(BC_INST_SCALE): BC_PROG_LBL(BC_INST_OBASE): // clang-format on { bc_program_pushGlobal(p, inst); BC_PROG_JUMP(inst, code, ip); } #if BC_ENABLE_EXTRA_MATH // clang-format off BC_PROG_LBL(BC_INST_SEED): // clang-format on { bc_program_pushSeed(p); BC_PROG_JUMP(inst, code, ip); } #endif // BC_ENABLE_EXTRA_MATH // clang-format off BC_PROG_LBL(BC_INST_LENGTH): BC_PROG_LBL(BC_INST_SCALE_FUNC): BC_PROG_LBL(BC_INST_SQRT): BC_PROG_LBL(BC_INST_ABS): BC_PROG_LBL(BC_INST_IS_NUMBER): BC_PROG_LBL(BC_INST_IS_STRING): #if BC_ENABLE_EXTRA_MATH BC_PROG_LBL(BC_INST_IRAND): #endif // BC_ENABLE_EXTRA_MATH // clang-format on { bc_program_builtin(p, inst); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_ASCIIFY): // clang-format on { bc_program_asciify(p); // Because we changed the execution stack and where we are // executing, we have to update all of this. BC_SIG_LOCK; ip = bc_vec_top(&p->stack); func = bc_vec_item(&p->fns, ip->func); code = func->code.v; BC_SIG_UNLOCK; BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_NUM): // clang-format on { bc_program_const(p, code, &ip->idx); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_ZERO): BC_PROG_LBL(BC_INST_ONE): #if BC_ENABLED BC_PROG_LBL(BC_INST_LAST): #endif // BC_ENABLED // clang-format on { r.t = BC_RESULT_ZERO + (inst - BC_INST_ZERO); bc_vec_push(&p->results, &r); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_PRINT): BC_PROG_LBL(BC_INST_PRINT_POP): #if BC_ENABLED BC_PROG_LBL(BC_INST_PRINT_STR): #endif // BC_ENABLED // clang-format on { bc_program_print(p, inst, 0); // We want to flush right away to save the output for history, // if history must preserve it when taking input. bc_file_flush(&vm->fout, bc_flush_save); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_STR): // clang-format on { // Set up the result and push. r.t = BC_RESULT_STR; bc_num_clear(&r.d.n); r.d.n.scale = bc_program_index(code, &ip->idx); bc_vec_push(&p->results, &r); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_POWER): BC_PROG_LBL(BC_INST_MULTIPLY): BC_PROG_LBL(BC_INST_DIVIDE): BC_PROG_LBL(BC_INST_MODULUS): BC_PROG_LBL(BC_INST_PLUS): BC_PROG_LBL(BC_INST_MINUS): #if BC_ENABLE_EXTRA_MATH BC_PROG_LBL(BC_INST_PLACES): BC_PROG_LBL(BC_INST_LSHIFT): BC_PROG_LBL(BC_INST_RSHIFT): #endif // BC_ENABLE_EXTRA_MATH // clang-format on { bc_program_op(p, inst); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_NEG): BC_PROG_LBL(BC_INST_BOOL_NOT): #if BC_ENABLE_EXTRA_MATH BC_PROG_LBL(BC_INST_TRUNC): #endif // BC_ENABLE_EXTRA_MATH // clang-format on { bc_program_unary(p, inst); BC_PROG_JUMP(inst, code, ip); } // clang-format off #if BC_ENABLED BC_PROG_LBL(BC_INST_ASSIGN_POWER): BC_PROG_LBL(BC_INST_ASSIGN_MULTIPLY): BC_PROG_LBL(BC_INST_ASSIGN_DIVIDE): BC_PROG_LBL(BC_INST_ASSIGN_MODULUS): BC_PROG_LBL(BC_INST_ASSIGN_PLUS): BC_PROG_LBL(BC_INST_ASSIGN_MINUS): #if BC_ENABLE_EXTRA_MATH BC_PROG_LBL(BC_INST_ASSIGN_PLACES): BC_PROG_LBL(BC_INST_ASSIGN_LSHIFT): BC_PROG_LBL(BC_INST_ASSIGN_RSHIFT): #endif // BC_ENABLE_EXTRA_MATH BC_PROG_LBL(BC_INST_ASSIGN): BC_PROG_LBL(BC_INST_ASSIGN_POWER_NO_VAL): BC_PROG_LBL(BC_INST_ASSIGN_MULTIPLY_NO_VAL): BC_PROG_LBL(BC_INST_ASSIGN_DIVIDE_NO_VAL): BC_PROG_LBL(BC_INST_ASSIGN_MODULUS_NO_VAL): BC_PROG_LBL(BC_INST_ASSIGN_PLUS_NO_VAL): BC_PROG_LBL(BC_INST_ASSIGN_MINUS_NO_VAL): #if BC_ENABLE_EXTRA_MATH BC_PROG_LBL(BC_INST_ASSIGN_PLACES_NO_VAL): BC_PROG_LBL(BC_INST_ASSIGN_LSHIFT_NO_VAL): BC_PROG_LBL(BC_INST_ASSIGN_RSHIFT_NO_VAL): #endif // BC_ENABLE_EXTRA_MATH #endif // BC_ENABLED BC_PROG_LBL(BC_INST_ASSIGN_NO_VAL): // clang-format on { bc_program_assign(p, inst); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_POP): // clang-format on { #ifndef BC_PROG_NO_STACK_CHECK // dc must do a stack check, but bc does not. if (BC_IS_DC) { if (BC_ERR(!BC_PROG_STACK(&p->results, 1))) { bc_err(BC_ERR_EXEC_STACK); } } #endif // BC_PROG_NO_STACK_CHECK assert(BC_PROG_STACK(&p->results, 1)); bc_vec_pop(&p->results); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_SWAP): // clang-format on { BcResult* ptr2; // Check the stack. if (BC_ERR(!BC_PROG_STACK(&p->results, 2))) { bc_err(BC_ERR_EXEC_STACK); } assert(BC_PROG_STACK(&p->results, 2)); // Get the two items. ptr = bc_vec_item_rev(&p->results, 0); ptr2 = bc_vec_item_rev(&p->results, 1); // Swap. It's just easiest to do it this way. // NOLINTNEXTLINE memcpy(&r, ptr, sizeof(BcResult)); // NOLINTNEXTLINE memcpy(ptr, ptr2, sizeof(BcResult)); // NOLINTNEXTLINE memcpy(ptr2, &r, sizeof(BcResult)); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_MODEXP): // clang-format on { bc_program_modexp(p); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_DIVMOD): // clang-format on { bc_program_divmod(p); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_PRINT_STREAM): // clang-format on { bc_program_printStream(p); BC_PROG_JUMP(inst, code, ip); } #if DC_ENABLED // clang-format off BC_PROG_LBL(BC_INST_POP_EXEC): // clang-format on { // If this fails, the dc parser got something wrong. assert(BC_PROG_STACK(&p->stack, 2)); // Pop the execution stack and tail call stack. bc_vec_pop(&p->stack); bc_vec_pop(&p->tail_calls); // Because we changed the execution stack and where we are // executing, we have to update all of this. BC_SIG_LOCK; ip = bc_vec_top(&p->stack); func = bc_vec_item(&p->fns, ip->func); code = func->code.v; BC_SIG_UNLOCK; BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_EXECUTE): BC_PROG_LBL(BC_INST_EXEC_COND): // clang-format on { cond = (inst == BC_INST_EXEC_COND); bc_program_execStr(p, code, &ip->idx, cond, func->code.len); // Because we changed the execution stack and where we are // executing, we have to update all of this. BC_SIG_LOCK; ip = bc_vec_top(&p->stack); func = bc_vec_item(&p->fns, ip->func); code = func->code.v; BC_SIG_UNLOCK; BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_PRINT_STACK): // clang-format on { bc_program_printStack(p); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_CLEAR_STACK): // clang-format on { bc_vec_popAll(&p->results); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_REG_STACK_LEN): // clang-format on { bc_program_regStackLen(p, code, &ip->idx); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_STACK_LEN): // clang-format on { bc_program_stackLen(p); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_DUPLICATE): // clang-format on { // Check the stack. if (BC_ERR(!BC_PROG_STACK(&p->results, 1))) { bc_err(BC_ERR_EXEC_STACK); } assert(BC_PROG_STACK(&p->results, 1)); // Get the top of the stack. ptr = bc_vec_top(&p->results); BC_SIG_LOCK; // Copy and push. bc_result_copy(&r, ptr); bc_vec_push(&p->results, &r); BC_SIG_UNLOCK; BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_LOAD): BC_PROG_LBL(BC_INST_PUSH_VAR): // clang-format on { bool copy = (inst == BC_INST_LOAD); bc_program_pushVar(p, code, &ip->idx, true, copy); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_PUSH_TO_VAR): // clang-format on { idx = bc_program_index(code, &ip->idx); bc_program_copyToVar(p, idx, BC_TYPE_VAR); BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_QUIT): BC_PROG_LBL(BC_INST_NQUIT): // clang-format on { bc_program_nquit(p, inst); // Because we changed the execution stack and where we are // executing, we have to update all of this. BC_SIG_LOCK; ip = bc_vec_top(&p->stack); func = bc_vec_item(&p->fns, ip->func); code = func->code.v; BC_SIG_UNLOCK; BC_PROG_JUMP(inst, code, ip); } // clang-format off BC_PROG_LBL(BC_INST_EXEC_STACK_LEN): // clang-format on { bc_program_execStackLen(p); BC_PROG_JUMP(inst, code, ip); } #endif // DC_ENABLED #if BC_HAS_COMPUTED_GOTO // clang-format off BC_PROG_LBL(BC_INST_INVALID): // clang-format on { goto end; } #else // BC_HAS_COMPUTED_GOTO default: { BC_UNREACHABLE #if BC_DEBUG && !BC_CLANG abort(); #endif // BC_DEBUG && !BC_CLANG } #endif // BC_HAS_COMPUTED_GOTO } #if BC_HAS_COMPUTED_GOTO #if BC_CLANG #pragma clang diagnostic warning "-Wgnu-label-as-value" #endif // BC_CLANG #if BC_GCC #pragma GCC diagnostic warning "-Wpedantic" #endif // BC_GCC #else // BC_HAS_COMPUTED_GOTO #if BC_DEBUG // This is to allow me to use a debugger to see the last instruction, // which will point to which function was the problem. But it's also a // good smoke test for error handling changes. assert(jmp_bufs_len == vm->jmp_bufs.len); #endif // BC_DEBUG #endif // BC_HAS_COMPUTED_GOTO } end: BC_SIG_MAYLOCK; // This is here just to print a stack trace on interrupts. This is for // finding infinite loops. if (BC_SIG_INTERRUPT(vm)) { BcStatus s; bc_file_putchar(&vm->ferr, bc_flush_none, '\n'); bc_program_printStackTrace(p); s = bc_file_flushErr(&vm->ferr, bc_flush_err); if (BC_ERR(s != BC_STATUS_SUCCESS && vm->status == BC_STATUS_SUCCESS)) { vm->status = (sig_atomic_t) s; } } BC_LONGJMP_CONT(vm); } #if BC_DEBUG_CODE #if BC_ENABLED && DC_ENABLED void bc_program_printStackDebug(BcProgram* p) { bc_file_puts(&vm->fout, bc_flush_err, "-------------- Stack ----------\n"); bc_program_printStack(p); bc_file_puts(&vm->fout, bc_flush_err, "-------------- Stack End ------\n"); } static void bc_program_printIndex(const char* restrict code, size_t* restrict bgn) { uchar byte, i, bytes = (uchar) code[(*bgn)++]; ulong val = 0; for (byte = 1, i = 0; byte && i < bytes; ++i) { byte = (uchar) code[(*bgn)++]; if (byte) val |= ((ulong) byte) << (CHAR_BIT * i); } bc_vm_printf(" (%lu) ", val); } static void bc_program_printStr(const BcProgram* p, const char* restrict code, size_t* restrict bgn) { size_t idx = bc_program_index(code, bgn); char* s; s = *((char**) bc_vec_item(&p->strs, idx)); bc_vm_printf(" (\"%s\") ", s); } void bc_program_printInst(const BcProgram* p, const char* restrict code, size_t* restrict bgn) { uchar inst = (uchar) code[(*bgn)++]; bc_vm_printf("Inst[%zu]: %s [%lu]; ", *bgn - 1, bc_inst_names[inst], (unsigned long) inst); if (inst == BC_INST_VAR || inst == BC_INST_ARRAY_ELEM || inst == BC_INST_ARRAY) { bc_program_printIndex(code, bgn); } else if (inst == BC_INST_STR) bc_program_printStr(p, code, bgn); else if (inst == BC_INST_NUM) { size_t idx = bc_program_index(code, bgn); BcConst* c = bc_vec_item(&p->consts, idx); bc_vm_printf("(%s)", c->val); } else if (inst == BC_INST_CALL || (inst > BC_INST_STR && inst <= BC_INST_JUMP_ZERO)) { bc_program_printIndex(code, bgn); if (inst == BC_INST_CALL) bc_program_printIndex(code, bgn); } bc_vm_putchar('\n', bc_flush_err); } void bc_program_code(const BcProgram* p) { BcFunc* f; char* code; BcInstPtr ip; size_t i; for (i = 0; i < p->fns.len; ++i) { ip.idx = ip.len = 0; ip.func = i; f = bc_vec_item(&p->fns, ip.func); code = f->code.v; bc_vm_printf("func[%zu]:\n", ip.func); while (ip.idx < f->code.len) { bc_program_printInst(p, code, &ip.idx); } bc_file_puts(&vm->fout, bc_flush_err, "\n\n"); } } #endif // BC_ENABLED && DC_ENABLED #endif // BC_DEBUG_CODE diff --git a/src/vm.c b/src/vm.c index 1a93e965a3f1..636cd4ba0c1b 100644 --- a/src/vm.c +++ b/src/vm.c @@ -1,1827 +1,1855 @@ /* * ***************************************************************************** * * SPDX-License-Identifier: BSD-2-Clause * * Copyright (c) 2018-2024 Gavin D. Howard and contributors. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions are met: * * * Redistributions of source code must retain the above copyright notice, this * list of conditions and the following disclaimer. * * * Redistributions in binary form must reproduce the above copyright notice, * this list of conditions and the following disclaimer in the documentation * and/or other materials provided with the distribution. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE * POSSIBILITY OF SUCH DAMAGE. * * ***************************************************************************** * * Code common to all of bc and dc. * */ #include #include #include #include #include #include #include #ifndef _WIN32 #include #include #include #else // _WIN32 #define WIN32_LEAN_AND_MEAN #include #include #endif // _WIN32 #include #include #include #include #include #include #if BC_ENABLE_LIBRARY #include #endif // BC_ENABLE_LIBRARY +#if BC_ENABLE_OSSFUZZ +#include +#endif // BC_ENABLE_OSSFUZZ #if !BC_ENABLE_LIBRARY // The actual globals. char output_bufs[BC_VM_BUF_SIZE]; BcVm vm_data; BcVm* vm = &vm_data; #endif // !BC_ENABLE_LIBRARY #if BC_DEBUG_CODE BC_NORETURN void bc_vm_jmp(const char* f) { #else // BC_DEBUG_CODE BC_NORETURN void bc_vm_jmp(void) { #endif #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY assert(BC_SIG_EXC(vm)); BC_SIG_MAYLOCK; #if BC_DEBUG_CODE bc_file_puts(&vm->ferr, bc_flush_none, "Longjmp: "); bc_file_puts(&vm->ferr, bc_flush_none, f); bc_file_putchar(&vm->ferr, bc_flush_none, '\n'); bc_file_flush(&vm->ferr, bc_flush_none); #endif // BC_DEBUG_CODE #if BC_DEBUG assert(vm->jmp_bufs.len - (size_t) vm->sig_pop); #endif // BC_DEBUG if (vm->jmp_bufs.len == 0) abort(); if (vm->sig_pop) bc_vec_pop(&vm->jmp_bufs); else vm->sig_pop = 1; siglongjmp(*((sigjmp_buf*) bc_vec_top(&vm->jmp_bufs)), 1); } #if !BC_ENABLE_LIBRARY /** * Handles signals. This is the signal handler. * @param sig The signal to handle. */ static void bc_vm_sig(int sig) { #if BC_ENABLE_EDITLINE // Editline needs this to resize the terminal. This also needs to come first // because a resize always needs to happen. if (sig == SIGWINCH) { if (BC_TTY) { el_resize(vm->history.el); // If the signal was a SIGWINCH, clear it because we don't need to // print a stack trace in that case. if (vm->sig == SIGWINCH) { vm->sig = 0; } } return; } #endif // BC_ENABLE_EDITLINE // There is already a signal in flight if this is true. if (vm->status == (sig_atomic_t) BC_STATUS_QUIT || vm->sig != 0) { if (!BC_I || sig != SIGINT) vm->status = BC_STATUS_QUIT; return; } // We always want to set this because a stack trace can be printed if we do. vm->sig = sig; // Only reset under these conditions; otherwise, quit. if (sig == SIGINT && BC_SIGINT && BC_I) { int err = errno; #if BC_ENABLE_EDITLINE // Editline needs this, for some unknown reason. if (write(STDOUT_FILENO, "^C", 2) != (ssize_t) 2) { vm->status = BC_STATUS_ERROR_FATAL; } #endif // BC_ENABLE_EDITLINE // Write the message. if (write(STDOUT_FILENO, vm->sigmsg, vm->siglen) != (ssize_t) vm->siglen) { vm->status = BC_STATUS_ERROR_FATAL; } errno = err; } else { #if BC_ENABLE_EDITLINE if (write(STDOUT_FILENO, "^C", 2) != (ssize_t) 2) { vm->status = BC_STATUS_ERROR_FATAL; return; } #endif // BC_ENABLE_EDITLINE vm->status = BC_STATUS_QUIT; } #if BC_ENABLE_LINE_LIB // Readline and Editline need this to actually handle sigints correctly. if (sig == SIGINT && bc_history_inlinelib) { bc_history_inlinelib = 0; siglongjmp(bc_history_jmpbuf, 1); } #endif // BC_ENABLE_LINE_LIB assert(vm->jmp_bufs.len); // Only jump if signals are not locked. The jump will happen by whoever // unlocks signals. if (!vm->sig_lock) BC_JMP; } /** * Sets up signal handling. */ static void bc_vm_sigaction(void) { #ifndef _WIN32 struct sigaction sa; sigemptyset(&sa.sa_mask); sa.sa_flags = BC_ENABLE_EDITLINE ? 0 : SA_NODEFER; // This mess is to silence a warning on Clang with regards to glibc's // sigaction handler, which activates the warning here. #if BC_CLANG #pragma clang diagnostic ignored "-Wdisabled-macro-expansion" #endif // BC_CLANG sa.sa_handler = bc_vm_sig; #if BC_CLANG #pragma clang diagnostic warning "-Wdisabled-macro-expansion" #endif // BC_CLANG sigaction(SIGTERM, &sa, NULL); sigaction(SIGQUIT, &sa, NULL); sigaction(SIGINT, &sa, NULL); #if BC_ENABLE_EDITLINE // Editline needs this to resize the terminal. if (BC_TTY) sigaction(SIGWINCH, &sa, NULL); #endif // BC_ENABLE_EDITLINE #if BC_ENABLE_HISTORY if (BC_TTY) sigaction(SIGHUP, &sa, NULL); #endif // BC_ENABLE_HISTORY #else // _WIN32 signal(SIGTERM, bc_vm_sig); signal(SIGINT, bc_vm_sig); #endif // _WIN32 } void bc_vm_info(const char* const help) { BC_SIG_ASSERT_LOCKED; // Print the banner. bc_file_printf(&vm->fout, "%s %s\n%s", vm->name, BC_VERSION, bc_copyright); // Print the help. if (help != NULL) { bc_file_putchar(&vm->fout, bc_flush_none, '\n'); #if BC_ENABLED if (BC_IS_BC) { const char* const banner = BC_DEFAULT_BANNER ? "to" : "to not"; const char* const sigint = BC_DEFAULT_SIGINT_RESET ? "to reset" : "to exit"; const char* const tty = BC_DEFAULT_TTY_MODE ? "enabled" : "disabled"; const char* const prompt = BC_DEFAULT_PROMPT ? "enabled" : "disabled"; const char* const expr = BC_DEFAULT_EXPR_EXIT ? "to exit" : "to not exit"; const char* const clamp = BC_DEFAULT_DIGIT_CLAMP ? "to clamp" : "to not clamp"; bc_file_printf(&vm->fout, help, vm->name, vm->name, BC_VERSION, BC_BUILD_TYPE, banner, sigint, tty, prompt, expr, clamp); } #endif // BC_ENABLED #if DC_ENABLED if (BC_IS_DC) { const char* const sigint = DC_DEFAULT_SIGINT_RESET ? "to reset" : "to exit"; const char* const tty = DC_DEFAULT_TTY_MODE ? "enabled" : "disabled"; const char* const prompt = DC_DEFAULT_PROMPT ? "enabled" : "disabled"; const char* const expr = DC_DEFAULT_EXPR_EXIT ? "to exit" : "to not exit"; const char* const clamp = DC_DEFAULT_DIGIT_CLAMP ? "to clamp" : "to not clamp"; bc_file_printf(&vm->fout, help, vm->name, vm->name, BC_VERSION, BC_BUILD_TYPE, sigint, tty, prompt, expr, clamp); } #endif // DC_ENABLED } // Flush. bc_file_flush(&vm->fout, bc_flush_none); } #endif // !BC_ENABLE_LIBRARY #if !BC_ENABLE_LIBRARY && !BC_ENABLE_MEMCHECK BC_NORETURN #endif // !BC_ENABLE_LIBRARY && !BC_ENABLE_MEMCHECK void bc_vm_fatalError(BcErr e) { bc_err(e); #if !BC_ENABLE_LIBRARY && !BC_ENABLE_MEMCHECK BC_UNREACHABLE #if !BC_CLANG abort(); #endif // !BC_CLANG #endif // !BC_ENABLE_LIBRARY && !BC_ENABLE_MEMCHECK } #if BC_ENABLE_LIBRARY BC_NORETURN void bc_vm_handleError(BcErr e) { #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY assert(e < BC_ERR_NELEMS); assert(!vm->sig_pop); BC_SIG_LOCK; // If we have a normal error... if (e <= BC_ERR_MATH_DIVIDE_BY_ZERO) { // Set the error. vm->err = (BclError) (e - BC_ERR_MATH_NEGATIVE + BCL_ERROR_MATH_NEGATIVE); } // Abort if we should. else if (vm->abrt) abort(); else if (e == BC_ERR_FATAL_ALLOC_ERR) vm->err = BCL_ERROR_FATAL_ALLOC_ERR; else vm->err = BCL_ERROR_FATAL_UNKNOWN_ERR; BC_JMP; } #else // BC_ENABLE_LIBRARY #if BC_DEBUG void bc_vm_handleError(BcErr e, const char* file, int fline, size_t line, ...) #else // BC_DEBUG void bc_vm_handleError(BcErr e, size_t line, ...) #endif // BC_DEBUG { BcStatus s; BcStatus fout_s; va_list args; uchar id = bc_err_ids[e]; const char* err_type = vm->err_ids[id]; sig_atomic_t lock; assert(e < BC_ERR_NELEMS); assert(!vm->sig_pop); #if BC_ENABLED // Figure out if the POSIX error should be an error, a warning, or nothing. if (!BC_S && e >= BC_ERR_POSIX_START) { if (BC_W) { // Make sure to not return an error. id = UCHAR_MAX; err_type = vm->err_ids[BC_ERR_IDX_WARN]; } else return; } #endif // BC_ENABLED BC_SIG_TRYLOCK(lock); // Make sure all of stdout is written first. fout_s = bc_file_flushErr(&vm->fout, bc_flush_err); // XXX: Keep the status for later. // Print the error message. va_start(args, line); bc_file_putchar(&vm->ferr, bc_flush_none, '\n'); bc_file_puts(&vm->ferr, bc_flush_none, err_type); bc_file_putchar(&vm->ferr, bc_flush_none, ' '); bc_file_vprintf(&vm->ferr, vm->err_msgs[e], args); va_end(args); // Print the extra information if we have it. if (BC_NO_ERR(vm->file != NULL)) { // This is the condition for parsing vs runtime. // If line is not 0, it is parsing. if (line) { bc_file_puts(&vm->ferr, bc_flush_none, "\n "); bc_file_puts(&vm->ferr, bc_flush_none, vm->file); bc_file_printf(&vm->ferr, ":%zu\n", line); } else { // Print a stack trace. bc_file_putchar(&vm->ferr, bc_flush_none, '\n'); bc_program_printStackTrace(&vm->prog); } } else { bc_file_putchar(&vm->ferr, bc_flush_none, '\n'); } #if BC_DEBUG bc_file_printf(&vm->ferr, "\n %s:%d\n", file, fline); #endif // BC_DEBUG bc_file_puts(&vm->ferr, bc_flush_none, "\n"); // If flushing to stdout failed, try to print *that* error, as long as that // was not the error already. if (fout_s == BC_STATUS_ERROR_FATAL && e != BC_ERR_FATAL_IO_ERR) { bc_file_putchar(&vm->ferr, bc_flush_none, '\n'); bc_file_puts(&vm->ferr, bc_flush_none, vm->err_ids[bc_err_ids[BC_ERR_FATAL_IO_ERR]]); bc_file_putchar(&vm->ferr, bc_flush_none, ' '); bc_file_puts(&vm->ferr, bc_flush_none, vm->err_msgs[BC_ERR_FATAL_IO_ERR]); } s = bc_file_flushErr(&vm->ferr, bc_flush_err); #if !BC_ENABLE_MEMCHECK // Because this function is called by a BC_NORETURN function when fatal // errors happen, we need to make sure to exit on fatal errors. This will // be faster anyway. This function *cannot jump when a fatal error occurs!* if (BC_ERR(id == BC_ERR_IDX_FATAL || fout_s == BC_STATUS_ERROR_FATAL || s == BC_STATUS_ERROR_FATAL)) { exit((int) BC_STATUS_ERROR_FATAL); } #else // !BC_ENABLE_MEMCHECK if (BC_ERR(fout_s == BC_STATUS_ERROR_FATAL)) { vm->status = (sig_atomic_t) fout_s; } else if (BC_ERR(s == BC_STATUS_ERROR_FATAL)) { vm->status = (sig_atomic_t) s; } else #endif // !BC_ENABLE_MEMCHECK { vm->status = (sig_atomic_t) (uchar) (id + 1); } // Only jump if there is an error. if (BC_ERR(vm->status)) BC_JMP; BC_SIG_TRYUNLOCK(lock); } char* bc_vm_getenv(const char* var) { char* ret; #ifndef _WIN32 ret = getenv(var); #else // _WIN32 _dupenv_s(&ret, NULL, var); #endif // _WIN32 return ret; } void bc_vm_getenvFree(char* val) { BC_UNUSED(val); #ifdef _WIN32 free(val); #endif // _WIN32 } /** * Sets a flag from an environment variable and the default. * @param var The environment variable. * @param def The default. * @param flag The flag to set. */ static void bc_vm_setenvFlag(const char* const var, int def, uint16_t flag) { // Get the value. char* val = bc_vm_getenv(var); // If there is no value... if (val == NULL) { // Set the default. if (def) vm->flags |= flag; else vm->flags &= ~(flag); } // Parse the value. else if (strtoul(val, NULL, 0)) vm->flags |= flag; else vm->flags &= ~(flag); bc_vm_getenvFree(val); } /** * Parses the arguments in {B,D]C_ENV_ARGS. * @param env_args_name The environment variable to use. * @param scale A pointer to return the scale that the arguments set, * if any. * @param ibase A pointer to return the ibase that the arguments set, * if any. * @param obase A pointer to return the obase that the arguments set, * if any. */ static void bc_vm_envArgs(const char* const env_args_name, BcBigDig* scale, BcBigDig* ibase, BcBigDig* obase) { char *env_args = bc_vm_getenv(env_args_name), *buf, *start; char instr = '\0'; BC_SIG_ASSERT_LOCKED; if (env_args == NULL) return; // Windows already allocates, so we don't need to. #ifndef _WIN32 start = buf = vm->env_args_buffer = bc_vm_strdup(env_args); #else // _WIN32 start = buf = vm->env_args_buffer = env_args; #endif // _WIN32 assert(buf != NULL); // Create two buffers for parsing. These need to stay throughout the entire // execution of bc, unfortunately, because of filenames that might be in // there. bc_vec_init(&vm->env_args, sizeof(char*), BC_DTOR_NONE); bc_vec_push(&vm->env_args, &env_args_name); // While we haven't reached the end of the args... while (*buf) { // If we don't have whitespace... if (!isspace(*buf)) { // If we have the start of a string... if (*buf == '"' || *buf == '\'') { // Set stuff appropriately. instr = *buf; buf += 1; // Check for the empty string. if (*buf == instr) { instr = '\0'; buf += 1; continue; } } // Push the pointer to the args buffer. bc_vec_push(&vm->env_args, &buf); // Parse the string. while (*buf && ((!instr && !isspace(*buf)) || (instr && *buf != instr))) { buf += 1; } // If we did find the end of the string... if (*buf) { if (instr) instr = '\0'; // Reset stuff. *buf = '\0'; buf += 1; start = buf; } else if (instr) bc_error(BC_ERR_FATAL_OPTION, 0, start); } // If we have whitespace, eat it. else buf += 1; } // Make sure to push a NULL pointer at the end. buf = NULL; bc_vec_push(&vm->env_args, &buf); // Parse the arguments. bc_args((int) vm->env_args.len - 1, bc_vec_item(&vm->env_args, 0), false, scale, ibase, obase); } /** * Gets the {B,D}C_LINE_LENGTH. * @param var The environment variable to pull it from. * @return The line length. */ static size_t bc_vm_envLen(const char* var) { char* lenv = bc_vm_getenv(var); size_t i, len = BC_NUM_PRINT_WIDTH; int num; // Return the default with none. if (lenv == NULL) return len; len = strlen(lenv); // Figure out if it's a number. for (num = 1, i = 0; num && i < len; ++i) { num = isdigit(lenv[i]); } // If it is a number... if (num) { // Parse it and clamp it if needed. len = (size_t) strtol(lenv, NULL, 10); if (len != 0) { len -= 1; if (len < 2 || len >= UINT16_MAX) len = BC_NUM_PRINT_WIDTH; } } // Set the default. else len = BC_NUM_PRINT_WIDTH; bc_vm_getenvFree(lenv); return len; } #endif // BC_ENABLE_LIBRARY void bc_vm_shutdown(void) { BC_SIG_ASSERT_LOCKED; #if BC_ENABLE_NLS if (vm->catalog != BC_VM_INVALID_CATALOG) catclose(vm->catalog); #endif // BC_ENABLE_NLS #if !BC_ENABLE_LIBRARY #if BC_ENABLE_HISTORY // This must always run to ensure that the terminal is back to normal, i.e., // has raw mode disabled. But we should only do it if we did not have a bad // terminal because history was not initialized if it is a bad terminal. if (BC_TTY && !vm->history.badTerm) bc_history_free(&vm->history); #endif // BC_ENABLE_HISTORY #endif // !BC_ENABLE_LIBRARY -#if BC_DEBUG +#if BC_DEBUG || BC_ENABLE_MEMCHECK #if !BC_ENABLE_LIBRARY bc_vec_free(&vm->env_args); free(vm->env_args_buffer); bc_vec_free(&vm->files); bc_vec_free(&vm->exprs); if (BC_PARSE_IS_INITED(&vm->read_prs, &vm->prog)) { bc_vec_free(&vm->read_buf); bc_parse_free(&vm->read_prs); } bc_parse_free(&vm->prs); bc_program_free(&vm->prog); bc_slabvec_free(&vm->slabs); #endif // !BC_ENABLE_LIBRARY bc_vm_freeTemps(); -#endif // BC_DEBUG +#endif // BC_DEBUG || BC_ENABLE_MEMCHECK #if !BC_ENABLE_LIBRARY // We always want to flush. bc_file_free(&vm->fout); bc_file_free(&vm->ferr); #endif // !BC_ENABLE_LIBRARY } void bc_vm_addTemp(BcDig* num) { #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY BC_SIG_ASSERT_LOCKED; // If we don't have room, just free. if (vm->temps_len == BC_VM_MAX_TEMPS) free(num); else { // Add to the buffer and length. vm->temps_buf[vm->temps_len] = num; vm->temps_len += 1; } } BcDig* bc_vm_takeTemp(void) { #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY BC_SIG_ASSERT_LOCKED; if (!vm->temps_len) return NULL; vm->temps_len -= 1; return vm->temps_buf[vm->temps_len]; } BcDig* bc_vm_getTemp(void) { #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY BC_SIG_ASSERT_LOCKED; if (!vm->temps_len) return NULL; return vm->temps_buf[vm->temps_len - 1]; } void bc_vm_freeTemps(void) { size_t i; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY BC_SIG_ASSERT_LOCKED; if (!vm->temps_len) return; // Free them all... for (i = 0; i < vm->temps_len; ++i) { free(vm->temps_buf[i]); } vm->temps_len = 0; } #if !BC_ENABLE_LIBRARY size_t bc_vm_numDigits(size_t val) { size_t digits = 0; do { digits += 1; val /= 10; } while (val != 0); return digits; } #endif // !BC_ENABLE_LIBRARY inline size_t bc_vm_arraySize(size_t n, size_t size) { size_t res = n * size; if (BC_ERR(BC_VM_MUL_OVERFLOW(n, size, res))) { bc_vm_fatalError(BC_ERR_FATAL_ALLOC_ERR); } return res; } inline size_t bc_vm_growSize(size_t a, size_t b) { size_t res = a + b; if (BC_ERR(res >= SIZE_MAX || res < a)) { bc_vm_fatalError(BC_ERR_FATAL_ALLOC_ERR); } return res; } void* bc_vm_malloc(size_t n) { void* ptr; BC_SIG_ASSERT_LOCKED; ptr = malloc(n); if (BC_ERR(ptr == NULL)) { bc_vm_freeTemps(); ptr = malloc(n); if (BC_ERR(ptr == NULL)) bc_vm_fatalError(BC_ERR_FATAL_ALLOC_ERR); } return ptr; } void* bc_vm_realloc(void* ptr, size_t n) { void* temp; BC_SIG_ASSERT_LOCKED; temp = realloc(ptr, n); if (BC_ERR(temp == NULL)) { bc_vm_freeTemps(); temp = realloc(ptr, n); if (BC_ERR(temp == NULL)) bc_vm_fatalError(BC_ERR_FATAL_ALLOC_ERR); } return temp; } char* bc_vm_strdup(const char* str) { char* s; BC_SIG_ASSERT_LOCKED; s = strdup(str); if (BC_ERR(s == NULL)) { bc_vm_freeTemps(); s = strdup(str); if (BC_ERR(s == NULL)) bc_vm_fatalError(BC_ERR_FATAL_ALLOC_ERR); } return s; } #if !BC_ENABLE_LIBRARY void bc_vm_printf(const char* fmt, ...) { va_list args; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #else // BC_ENABLE_LIBRARY sig_atomic_t lock; #endif // BC_ENABLE_LIBRARY BC_SIG_TRYLOCK(lock); va_start(args, fmt); bc_file_vprintf(&vm->fout, fmt, args); va_end(args); vm->nchars = 0; BC_SIG_TRYUNLOCK(lock); } #endif // !BC_ENABLE_LIBRARY void bc_vm_putchar(int c, BcFlushType type) { #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); bc_vec_pushByte(&vm->out, (uchar) c); #else // BC_ENABLE_LIBRARY bc_file_putchar(&vm->fout, type, (uchar) c); vm->nchars = (c == '\n' ? 0 : vm->nchars + 1); #endif // BC_ENABLE_LIBRARY } #if !BC_ENABLE_LIBRARY #ifdef __OpenBSD__ /** * Aborts with a message. This should never be called because I have carefully * made sure that the calls to pledge() and unveil() are correct, but it's here * just in case. * @param msg The message to print. */ BC_NORETURN static void bc_abortm(const char* msg) { bc_file_puts(&vm->ferr, bc_flush_none, msg); bc_file_puts(&vm->ferr, bc_flush_none, "; this is a bug"); bc_file_flush(&vm->ferr, bc_flush_none); abort(); } void bc_pledge(const char* promises, const char* execpromises) { int r = pledge(promises, execpromises); if (r) bc_abortm("pledge() failed"); } #if BC_ENABLE_EXTRA_MATH /** * A convenience and portability function for OpenBSD's unveil(). * @param path The path. * @param permissions The permissions for the path. */ static void bc_unveil(const char* path, const char* permissions) { int r = unveil(path, permissions); if (r) bc_abortm("unveil() failed"); } #endif // BC_ENABLE_EXTRA_MATH #else // __OpenBSD__ void bc_pledge(const char* promises, const char* execpromises) { BC_UNUSED(promises); BC_UNUSED(execpromises); } #if BC_ENABLE_EXTRA_MATH static void bc_unveil(const char* path, const char* permissions) { BC_UNUSED(path); BC_UNUSED(permissions); } #endif // BC_ENABLE_EXTRA_MATH #endif // __OpenBSD__ /** * Cleans unneeded variables, arrays, functions, strings, and constants when * done executing a line of stdin. This is to prevent memory usage growing * without bound. This is an idea from busybox. */ static void bc_vm_clean(void) { BcVec* fns = &vm->prog.fns; BcFunc* f = bc_vec_item(fns, BC_PROG_MAIN); BcInstPtr* ip = bc_vec_item(&vm->prog.stack, 0); bool good = ((vm->status && vm->status != BC_STATUS_QUIT) || vm->sig != 0); BC_SIG_ASSERT_LOCKED; // If all is good, go ahead and reset. if (good) bc_program_reset(&vm->prog); #if BC_ENABLED // bc has this extra condition. If it not satisfied, it is in the middle of // a parse. if (good && BC_IS_BC) good = !BC_PARSE_NO_EXEC(&vm->prs); #endif // BC_ENABLED #if DC_ENABLED // For dc, it is safe only when all of the results on the results stack are // safe, which means that they are temporaries or other things that don't // need strings or constants. if (BC_IS_DC) { size_t i; good = true; for (i = 0; good && i < vm->prog.results.len; ++i) { BcResult* r = (BcResult*) bc_vec_item(&vm->prog.results, i); good = BC_VM_SAFE_RESULT(r); } } #endif // DC_ENABLED // If this condition is true, we can get rid of strings, // constants, and code. if (good && vm->prog.stack.len == 1 && ip->idx == f->code.len) { // XXX: Nothing can be popped in dc. Deal with it. #if BC_ENABLED if (BC_IS_BC) { // XXX: you cannot delete strings, functions, or constants in bc. // Deal with it. bc_vec_popAll(&f->labels); } #endif // BC_ENABLED bc_vec_popAll(&f->code); ip->idx = 0; } } /** * Process a bunch of text. * @param text The text to process. * @param mode The mode to process in. */ static void bc_vm_process(const char* text, BcMode mode) { // Set up the parser. bc_parse_text(&vm->prs, text, mode); while (vm->prs.l.t != BC_LEX_EOF) { // Parsing requires a signal lock. We also don't parse everything; we // want to execute as soon as possible for *everything*. BC_SIG_LOCK; vm->parse(&vm->prs); BC_SIG_UNLOCK; // Execute if possible. if (BC_IS_DC || !BC_PARSE_NO_EXEC(&vm->prs)) bc_program_exec(&vm->prog); assert(BC_IS_DC || vm->prog.results.len == 0); // Flush in interactive mode. if (BC_I) bc_file_flush(&vm->fout, bc_flush_save); } } #if BC_ENABLED /** * Ends a series of if statements. This is to ensure that full parses happen * when a file finishes or stdin has no more data. Without this, bc thinks that * it cannot parse any further. But if we reach the end of a file or stdin has * no more data, we know we can add an empty else clause. */ static void bc_vm_endif(void) { bc_parse_endif(&vm->prs); bc_program_exec(&vm->prog); } #endif // BC_ENABLED /** * Processes a file. * @param file The filename. */ static void bc_vm_file(const char* file) { char* data = NULL; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY assert(!vm->sig_pop); vm->mode = BC_MODE_FILE; // Set up the lexer. bc_lex_file(&vm->prs.l, file); BC_SIG_LOCK; // Read the file. data = bc_read_file(file); assert(data != NULL); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; // Process it. bc_vm_process(data, BC_MODE_FILE); #if BC_ENABLED // Make sure to end any open if statements. if (BC_IS_BC) bc_vm_endif(); #endif // BC_ENABLED err: BC_SIG_MAYLOCK; // Cleanup. free(data); bc_vm_clean(); // bc_program_reset(), called by bc_vm_clean(), resets the status. // We want it to clear the sig_pop variable in case it was set. if (vm->status == (sig_atomic_t) BC_STATUS_SUCCESS) BC_LONGJMP_STOP; BC_LONGJMP_CONT(vm); } +#if !BC_ENABLE_OSSFUZZ + bool bc_vm_readLine(bool clear) { BcStatus s; bool good; BC_SIG_ASSERT_NOT_LOCKED; // Clear the buffer if desired. if (clear) bc_vec_empty(&vm->buffer); // Empty the line buffer. bc_vec_empty(&vm->line_buf); if (vm->eof) return false; do { // bc_read_line() must always return either BC_STATUS_SUCCESS or // BC_STATUS_EOF. Everything else, it and whatever it calls, must jump // out instead. s = bc_read_line(&vm->line_buf, ">>> "); vm->eof = (s == BC_STATUS_EOF); } while (s == BC_STATUS_SUCCESS && !vm->eof && vm->line_buf.len < 1); good = (vm->line_buf.len > 1); // Concat if we found something. if (good) bc_vec_concat(&vm->buffer, vm->line_buf.v); return good; } /** * Processes text from stdin. */ static void bc_vm_stdin(void) { bool clear; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY clear = true; vm->mode = BC_MODE_STDIN; // Set up the lexer. bc_lex_file(&vm->prs.l, bc_program_stdin_name); // These are global so that the lexers can access them, but they are // allocated and freed in this function because they should only be used for // stdin and expressions (they are used in bc_vm_exprs() as well). So they // are tied to this function, really. Well, this and bc_vm_readLine(). These // are the reasons that we have vm->is_stdin to tell the lexers if we are // reading from stdin. Well, both lexers care. And the reason they care is // so that if a comment or a string goes across multiple lines, the lexer // can request more data from stdin until the comment or string is ended. BC_SIG_LOCK; bc_vec_init(&vm->buffer, sizeof(uchar), BC_DTOR_NONE); bc_vec_init(&vm->line_buf, sizeof(uchar), BC_DTOR_NONE); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; // This label exists because errors can cause jumps to end up at the err label // below. If that happens, and the error should be cleared and execution // continue, then we need to jump back. restart: // While we still read data from stdin. while (bc_vm_readLine(clear)) { size_t len = vm->buffer.len - 1; const char* str = vm->buffer.v; // We don't want to clear the buffer when the line ends with a backslash // because a backslash newline is special in bc. clear = (len < 2 || str[len - 2] != '\\' || str[len - 1] != '\n'); if (!clear) continue; // Process the data. bc_vm_process(vm->buffer.v, BC_MODE_STDIN); if (vm->eof) break; else { BC_SIG_LOCK; bc_vm_clean(); BC_SIG_UNLOCK; } } #if BC_ENABLED // End the if statements. if (BC_IS_BC) bc_vm_endif(); #endif // BC_ENABLED err: BC_SIG_MAYLOCK; // Cleanup. bc_vm_clean(); #if !BC_ENABLE_MEMCHECK assert(vm->status != BC_STATUS_ERROR_FATAL); vm->status = vm->status == BC_STATUS_QUIT || !BC_I ? vm->status : BC_STATUS_SUCCESS; #else // !BC_ENABLE_MEMCHECK vm->status = vm->status == BC_STATUS_ERROR_FATAL || vm->status == BC_STATUS_QUIT || !BC_I ? vm->status : BC_STATUS_SUCCESS; #endif // !BC_ENABLE_MEMCHECK if (!vm->status && !vm->eof) { bc_vec_empty(&vm->buffer); BC_LONGJMP_STOP; BC_SIG_UNLOCK; goto restart; } #if BC_DEBUG // Since these are tied to this function, free them here. We only free in // debug mode because stdin is always the last thing read. bc_vec_free(&vm->line_buf); bc_vec_free(&vm->buffer); #endif // BC_DEBUG BC_LONGJMP_CONT(vm); } +#endif // BC_ENABLE_OSSFUZZ + bool bc_vm_readBuf(bool clear) { size_t len = vm->exprs.len - 1; bool more; BC_SIG_ASSERT_NOT_LOCKED; // Clear the buffer if desired. if (clear) bc_vec_empty(&vm->buffer); // We want to pop the nul byte off because that's what bc_read_buf() // expects. bc_vec_pop(&vm->buffer); // Read one line of expressions. more = bc_read_buf(&vm->buffer, vm->exprs.v, &len); bc_vec_pushByte(&vm->buffer, '\0'); return more; } static void bc_vm_exprs(void) { bool clear; #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY clear = true; vm->mode = BC_MODE_EXPRS; // Prepare the lexer. bc_lex_file(&vm->prs.l, bc_program_exprs_name); // We initialize this so that the lexer can access it in the case that it // needs more data for expressions, such as for a multiline string or // comment. See the comment on the allocation of vm->buffer above in // bc_vm_stdin() for more information. BC_SIG_LOCK; bc_vec_init(&vm->buffer, sizeof(uchar), BC_DTOR_NONE); BC_SETJMP_LOCKED(vm, err); BC_SIG_UNLOCK; while (bc_vm_readBuf(clear)) { size_t len = vm->buffer.len - 1; const char* str = vm->buffer.v; // We don't want to clear the buffer when the line ends with a backslash // because a backslash newline is special in bc. clear = (len < 2 || str[len - 2] != '\\' || str[len - 1] != '\n'); if (!clear) continue; // Process the data. bc_vm_process(vm->buffer.v, BC_MODE_EXPRS); } // If we were not supposed to clear, then we should process everything. This // makes sure that errors get reported. if (!clear) bc_vm_process(vm->buffer.v, BC_MODE_EXPRS); err: BC_SIG_MAYLOCK; // Cleanup. bc_vm_clean(); // bc_program_reset(), called by bc_vm_clean(), resets the status. // We want it to clear the sig_pop variable in case it was set. if (vm->status == (sig_atomic_t) BC_STATUS_SUCCESS) BC_LONGJMP_STOP; // Since this is tied to this function, free it here. We always free it here // because bc_vm_stdin() may or may not use it later. bc_vec_free(&vm->buffer); BC_LONGJMP_CONT(vm); } #if BC_ENABLED /** * Loads a math library. * @param name The name of the library. * @param text The text of the source code. */ static void bc_vm_load(const char* name, const char* text) { bc_lex_file(&vm->prs.l, name); bc_parse_text(&vm->prs, text, BC_MODE_FILE); BC_SIG_LOCK; while (vm->prs.l.t != BC_LEX_EOF) { vm->parse(&vm->prs); } BC_SIG_UNLOCK; } #endif // BC_ENABLED /** * Loads the default error messages. */ static void bc_vm_defaultMsgs(void) { size_t i; // Load the error categories. for (i = 0; i < BC_ERR_IDX_NELEMS + BC_ENABLED; ++i) { vm->err_ids[i] = bc_errs[i]; } // Load the error messages. for (i = 0; i < BC_ERR_NELEMS; ++i) { vm->err_msgs[i] = bc_err_msgs[i]; } } /** * Loads the error messages for the locale. If NLS is disabled, this just loads * the default messages. */ static void bc_vm_gettext(void) { #if BC_ENABLE_NLS uchar id = 0; int set, msg = 1; size_t i; // If no locale, load the defaults. if (vm->locale == NULL) { vm->catalog = BC_VM_INVALID_CATALOG; bc_vm_defaultMsgs(); return; } vm->catalog = catopen(BC_MAINEXEC, NL_CAT_LOCALE); // If no catalog, load the defaults. if (vm->catalog == BC_VM_INVALID_CATALOG) { bc_vm_defaultMsgs(); return; } // Load the error categories. for (set = 1; msg <= BC_ERR_IDX_NELEMS + BC_ENABLED; ++msg) { vm->err_ids[msg - 1] = catgets(vm->catalog, set, msg, bc_errs[msg - 1]); } i = 0; id = bc_err_ids[i]; // Load the error messages. In order to understand this loop, you must know // the order of messages and categories in the enum and in the locale files. for (set = id + 2, msg = 1; i < BC_ERR_NELEMS; ++i, ++msg) { if (id != bc_err_ids[i]) { msg = 1; id = bc_err_ids[i]; set = id + 2; } vm->err_msgs[i] = catgets(vm->catalog, set, msg, bc_err_msgs[i]); } #else // BC_ENABLE_NLS bc_vm_defaultMsgs(); #endif // BC_ENABLE_NLS } /** * Starts execution. Really, this is a function of historical accident; it could * probably be combined with bc_vm_boot(), but I don't care enough. Really, this * function starts when execution of bc or dc source code starts. */ static void bc_vm_exec(void) { size_t i; #if DC_ENABLED bool has_file = false; #endif // DC_ENABLED #if BC_ENABLED // Load the math libraries. if (BC_IS_BC && (vm->flags & BC_FLAG_L)) { // Can't allow redefinitions in the builtin library. vm->no_redefine = true; bc_vm_load(bc_lib_name, bc_lib); #if BC_ENABLE_EXTRA_MATH if (!BC_IS_POSIX) bc_vm_load(bc_lib2_name, bc_lib2); #endif // BC_ENABLE_EXTRA_MATH // Make sure to clear this. vm->no_redefine = false; // Execute to ensure that all is hunky dory. Without this, scale can be // set improperly. bc_program_exec(&vm->prog); } #endif // BC_ENABLED + assert(!BC_ENABLE_OSSFUZZ || BC_EXPR_EXIT == 0); + // If there are expressions to execute... if (vm->exprs.len) { // Process the expressions. bc_vm_exprs(); // Sometimes, executing expressions means we need to quit. - if (!vm->no_exprs && vm->exit_exprs && BC_EXPR_EXIT) return; + if (vm->status != BC_STATUS_SUCCESS || + (!vm->no_exprs && vm->exit_exprs && BC_EXPR_EXIT)) + { + return; + } } // Process files. for (i = 0; i < vm->files.len; ++i) { char* path = *((char**) bc_vec_item(&vm->files, i)); if (!strcmp(path, "")) continue; #if DC_ENABLED has_file = true; #endif // DC_ENABLED bc_vm_file(path); + + if (vm->status != BC_STATUS_SUCCESS) return; } #if BC_ENABLE_EXTRA_MATH // These are needed for the pseudo-random number generator. bc_unveil("/dev/urandom", "r"); bc_unveil("/dev/random", "r"); bc_unveil(NULL, NULL); #endif // BC_ENABLE_EXTRA_MATH #if BC_ENABLE_HISTORY // We need to keep tty if history is enabled, and we need to keep rpath for // the times when we read from /dev/urandom. if (BC_TTY && !vm->history.badTerm) bc_pledge(bc_pledge_end_history, NULL); else #endif // BC_ENABLE_HISTORY { bc_pledge(bc_pledge_end, NULL); } #if BC_ENABLE_AFL // This is the thing that makes fuzzing with AFL++ so fast. If you move this // back, you won't cause any problems, but fuzzing will slow down. If you // move this forward, you won't fuzz anything because you will be skipping // the reading from stdin. __AFL_INIT(); #endif // BC_ENABLE_AFL +#if BC_ENABLE_OSSFUZZ + + if (BC_VM_RUN_STDIN(has_file)) + { + // XXX: Yes, this is a hack to run the fuzzer for OSS-Fuzz, but it + // works. + bc_vm_load("", (const char*) bc_fuzzer_data); + } + +#else // BC_ENABLE_OSSFUZZ + // Execute from stdin. bc always does. if (BC_VM_RUN_STDIN(has_file)) bc_vm_stdin(); + +#endif // BC_ENABLE_OSSFUZZ } BcStatus -bc_vm_boot(int argc, char* argv[]) +bc_vm_boot(int argc, const char* argv[]) { int ttyin, ttyout, ttyerr; bool tty; const char* const env_len = BC_VM_LINE_LENGTH_STR; const char* const env_args = BC_VM_ENV_ARGS_STR; const char* const env_exit = BC_VM_EXPR_EXIT_STR; const char* const env_clamp = BC_VM_DIGIT_CLAMP_STR; int env_exit_def = BC_VM_EXPR_EXIT_DEF; int env_clamp_def = BC_VM_DIGIT_CLAMP_DEF; BcBigDig scale = BC_NUM_BIGDIG_MAX; BcBigDig env_scale = BC_NUM_BIGDIG_MAX; BcBigDig ibase = BC_NUM_BIGDIG_MAX; BcBigDig env_ibase = BC_NUM_BIGDIG_MAX; BcBigDig obase = BC_NUM_BIGDIG_MAX; BcBigDig env_obase = BC_NUM_BIGDIG_MAX; // We need to know which of stdin, stdout, and stderr are tty's. ttyin = isatty(STDIN_FILENO); ttyout = isatty(STDOUT_FILENO); ttyerr = isatty(STDERR_FILENO); tty = (ttyin != 0 && ttyout != 0 && ttyerr != 0); vm->flags |= ttyin ? BC_FLAG_TTYIN : 0; vm->flags |= tty ? BC_FLAG_TTY : 0; vm->flags |= ttyin && ttyout ? BC_FLAG_I : 0; // Set up signals. bc_vm_sigaction(); // Initialize some vm stuff. This is separate to make things easier for the // library. bc_vm_init(); // Explicitly set this in case NULL isn't all zeroes. vm->file = NULL; // Set the error messages. bc_vm_gettext(); #if BC_ENABLE_LINE_LIB // Initialize the output file buffers. bc_file_init(&vm->ferr, stderr, true); bc_file_init(&vm->fout, stdout, false); // Set the input buffer. vm->buf = output_bufs; #else // BC_ENABLE_LINE_LIB // Initialize the output file buffers. They each take portions of the global // buffer. stdout gets more because it will probably have more data. bc_file_init(&vm->ferr, STDERR_FILENO, output_bufs + BC_VM_STDOUT_BUF_SIZE, BC_VM_STDERR_BUF_SIZE, true); bc_file_init(&vm->fout, STDOUT_FILENO, output_bufs, BC_VM_STDOUT_BUF_SIZE, false); // Set the input buffer to the rest of the global buffer. vm->buf = output_bufs + BC_VM_STDOUT_BUF_SIZE + BC_VM_STDERR_BUF_SIZE; #endif // BC_ENABLE_LINE_LIB // Set the line length by environment variable. vm->line_len = (uint16_t) bc_vm_envLen(env_len); bc_vm_setenvFlag(env_exit, env_exit_def, BC_FLAG_EXPR_EXIT); bc_vm_setenvFlag(env_clamp, env_clamp_def, BC_FLAG_DIGIT_CLAMP); // Clear the files and expressions vectors, just in case. This marks them as // *not* allocated. bc_vec_clear(&vm->files); bc_vec_clear(&vm->exprs); #if !BC_ENABLE_LIBRARY // Initialize the slab vector. bc_slabvec_init(&vm->slabs); #endif // !BC_ENABLE_LIBRARY // Initialize the program and main parser. These have to be in this order // because the program has to be initialized first, since a pointer to it is // passed to the parser. bc_program_init(&vm->prog); bc_parse_init(&vm->prs, &vm->prog, BC_PROG_MAIN); // Set defaults. vm->flags |= BC_TTY ? BC_FLAG_P | BC_FLAG_R : 0; vm->flags |= BC_I ? BC_FLAG_Q : 0; #if BC_ENABLED if (BC_IS_BC) { // bc checks this environment variable to see if it should run in // standard mode. char* var = bc_vm_getenv("POSIXLY_CORRECT"); vm->flags |= BC_FLAG_S * (var != NULL); bc_vm_getenvFree(var); // Set whether we print the banner or not. if (BC_I) bc_vm_setenvFlag("BC_BANNER", BC_DEFAULT_BANNER, BC_FLAG_Q); } #endif // BC_ENABLED // Are we in TTY mode? if (BC_TTY) { const char* const env_tty = BC_VM_TTY_MODE_STR; int env_tty_def = BC_VM_TTY_MODE_DEF; const char* const env_prompt = BC_VM_PROMPT_STR; int env_prompt_def = BC_VM_PROMPT_DEF; // Set flags for TTY mode and prompt. bc_vm_setenvFlag(env_tty, env_tty_def, BC_FLAG_TTY); bc_vm_setenvFlag(env_prompt, tty ? env_prompt_def : 0, BC_FLAG_P); #if BC_ENABLE_HISTORY // If TTY mode is used, activate history. if (BC_TTY) bc_history_init(&vm->history); #endif // BC_ENABLE_HISTORY } // Process environment and command-line arguments. bc_vm_envArgs(env_args, &env_scale, &env_ibase, &env_obase); bc_args(argc, argv, true, &scale, &ibase, &obase); // This section is here because we don't want the math library to stomp on // the user's given value for scale. And we don't want ibase affecting how // the scale is interpreted. Also, it's sectioned off just for this comment. { BC_SIG_UNLOCK; scale = scale == BC_NUM_BIGDIG_MAX ? env_scale : scale; #if BC_ENABLED // Assign the library value only if it is used and no value was set. scale = scale == BC_NUM_BIGDIG_MAX && BC_L ? 20 : scale; #endif // BC_ENABLED obase = obase == BC_NUM_BIGDIG_MAX ? env_obase : obase; ibase = ibase == BC_NUM_BIGDIG_MAX ? env_ibase : ibase; if (scale != BC_NUM_BIGDIG_MAX) { bc_program_assignBuiltin(&vm->prog, true, false, scale); } if (obase != BC_NUM_BIGDIG_MAX) { bc_program_assignBuiltin(&vm->prog, false, true, obase); } // This is last to avoid it affecting the value of the others. if (ibase != BC_NUM_BIGDIG_MAX) { bc_program_assignBuiltin(&vm->prog, false, false, ibase); } BC_SIG_LOCK; } // If we are in interactive mode... if (BC_I) { const char* const env_sigint = BC_VM_SIGINT_RESET_STR; int env_sigint_def = BC_VM_SIGINT_RESET_DEF; // Set whether we reset on SIGINT or not. bc_vm_setenvFlag(env_sigint, env_sigint_def, BC_FLAG_SIGINT); } #if BC_ENABLED // Disable global stacks in POSIX mode. if (BC_IS_POSIX) vm->flags &= ~(BC_FLAG_G); // Print the banner if allowed. We have to be in bc, in interactive mode, // and not be quieted by command-line option or environment variable. if (BC_IS_BC && BC_I && (vm->flags & BC_FLAG_Q)) { bc_vm_info(NULL); bc_file_putchar(&vm->fout, bc_flush_none, '\n'); bc_file_flush(&vm->fout, bc_flush_none); } #endif // BC_ENABLED BC_SIG_UNLOCK; // Start executing. bc_vm_exec(); BC_SIG_LOCK; // Exit. - return bc_vm_atexit((BcStatus) vm->status); + return (BcStatus) vm->status; } #endif // !BC_ENABLE_LIBRARY void bc_vm_init(void) { #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY BC_SIG_ASSERT_LOCKED; #if !BC_ENABLE_LIBRARY // Set up the constant zero. bc_num_setup(&vm->zero, vm->zero_num, BC_VM_ONE_CAP); #endif // !BC_ENABLE_LIBRARY // Set up more constant BcNum's. bc_num_setup(&vm->one, vm->one_num, BC_VM_ONE_CAP); bc_num_one(&vm->one); // Set up more constant BcNum's. // NOLINTNEXTLINE memcpy(vm->max_num, bc_num_bigdigMax, bc_num_bigdigMax_size * sizeof(BcDig)); // NOLINTNEXTLINE memcpy(vm->max2_num, bc_num_bigdigMax2, bc_num_bigdigMax2_size * sizeof(BcDig)); bc_num_setup(&vm->max, vm->max_num, BC_NUM_BIGDIG_LOG10); bc_num_setup(&vm->max2, vm->max2_num, BC_NUM_BIGDIG_LOG10); vm->max.len = bc_num_bigdigMax_size; vm->max2.len = bc_num_bigdigMax2_size; // Set up the maxes for the globals. vm->maxes[BC_PROG_GLOBALS_IBASE] = BC_NUM_MAX_POSIX_IBASE; vm->maxes[BC_PROG_GLOBALS_OBASE] = BC_MAX_OBASE; vm->maxes[BC_PROG_GLOBALS_SCALE] = BC_MAX_SCALE; #if BC_ENABLE_EXTRA_MATH vm->maxes[BC_PROG_MAX_RAND] = ((BcRand) 0) - 1; #endif // BC_ENABLE_EXTRA_MATH #if BC_ENABLED #if !BC_ENABLE_LIBRARY // bc has a higher max ibase when it's not in POSIX mode. if (BC_IS_BC && !BC_IS_POSIX) #endif // !BC_ENABLE_LIBRARY { vm->maxes[BC_PROG_GLOBALS_IBASE] = BC_NUM_MAX_IBASE; } #endif // BC_ENABLED } #if BC_ENABLE_LIBRARY void bc_vm_atexit(void) { #if BC_DEBUG #if BC_ENABLE_LIBRARY BcVm* vm = bcl_getspecific(); #endif // BC_ENABLE_LIBRARY #endif // BC_DEBUG bc_vm_shutdown(); #if BC_DEBUG bc_vec_free(&vm->jmp_bufs); #endif // BC_DEBUG } #else // BC_ENABLE_LIBRARY BcStatus bc_vm_atexit(BcStatus status) { // Set the status correctly. BcStatus s = BC_STATUS_IS_ERROR(status) ? status : BC_STATUS_SUCCESS; bc_vm_shutdown(); #if BC_DEBUG bc_vec_free(&vm->jmp_bufs); #endif // BC_DEBUG return s; } #endif // BC_ENABLE_LIBRARY diff --git a/tests/bc/errors/37.txt b/tests/bc/errors/37.txt new file mode 100644 index 000000000000..e7c504dcdb88 --- /dev/null +++ b/tests/bc/errors/37.txt @@ -0,0 +1,37 @@ +print f +if(6)H +if(6)streafoob#! /q + +define printarray(a[], len) { + + auto i + + for (i = 0; i < len; ++i) { + m[i] + } +} + +define a2(a[], len) { + + auto i + + for (i = 0; i < len; ++i) { + a[i] = a[i] * a[i] + } + + printarray(a[], len) +} +define a1(*a[], len) { + + auto i + + for (i = 0; i < len; ++i) { + a[i] = i + } + + a2(a[], len) + + printarray(a[], len) +} +len = 16 +a1(b[] ++ase^= , len) diff --git a/tests/bc/errors/38.txt b/tests/bc/errors/38.txt new file mode 100644 index 000000000000..b0f9eb22f7a3 --- /dev/null +++ b/tests/bc/errors/38.txt @@ -0,0 +1,37 @@ +print f +if(6)H +if(6)streafoob#! /q + +define printarray(a[], len) { + + auto i + + for (i = 0; i < len; ++i) { + m[i] + } +} + +define a2(a[], len) { + + auto i + + for (i = 0; i < len; ++i) { + a[i] = a[i] * a[i] + } + + printarray(a[], len) +} +define a1(*a[], len) { + + auto i + + for (i = 0; i < len; ++i) { + a[i] = i + } + + a2(a[], len) + + printarray(a[], len) +} +len = 16 +a1((b[]) + ++ase^= , len) diff --git a/vs/bc.vcxproj b/vs/bc.vcxproj index 377eb8645a71..c98ebc6eee53 100644 --- a/vs/bc.vcxproj +++ b/vs/bc.vcxproj @@ -1,302 +1,302 @@ Debug Win32 Release Win32 Debug x64 Release x64 16.0 Win32Proj {4450d61f-2535-4085-b1b1-f96acd23cc9f} bc 10.0 Application true - v142 + v143 Unicode Application false - v142 + v143 true Unicode Application true - v142 + v143 Unicode Application false - v142 + v143 true Unicode true ClCompile false bin\$(Platform)\$(Configuration)\ bin\$(Platform)\$(Configuration)\ false ClCompile false bin\$(Platform)\$(Configuration)\ bin\$(Platform)\$(Configuration)\ true ClCompile false bin\$(Platform)\$(Configuration)\ bin\$(Platform)\$(Configuration)\ false ClCompile false bin\$(Platform)\$(Configuration)\ bin\$(Platform)\$(Configuration)\ /std:c17 /MP $(AdditionalOptions) Level3 true - BC_ENABLED=1;DC_ENABLED=1;BC_ENABLE_EXTRA_MATH=1;BC_ENABLE_HISTORY=1;BC_ENABLE_NLS=0;BC_DEBUG_CODE=0;BC_ENABLE_LIBRARY=0;BC_ENABLE_EDITLINE=0;BC_ENABLE_READLINE=0;BUILD_TYPE=N;BC_DEFAULT_BANNER=1;BC_DEFAULT_SIGINT_RESET=0;DC_DEFAULT_SIGINT_RESET=0;BC_DEFAULT_TTY_MODE=1;DC_DEFAULT_TTY_MODE=1;BC_DEFAULT_PROMPT=1;DC_DEFAULT_PROMPT=1;BC_DEFAULT_EXPR_EXIT=1;DC_DEFAULT_EXPR_EXIT=1;BC_DEFAULT_DIGIT_CLAMP=1;DC_DEFAULT_DIGIT_CLAMP=1;WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions) + BC_ENABLED=1;DC_ENABLED=1;BC_ENABLE_EXTRA_MATH=1;BC_ENABLE_HISTORY=1;BC_ENABLE_NLS=0;BC_DEBUG_CODE=0;BC_ENABLE_LIBRARY=0;BC_ENABLE_EDITLINE=0;BC_ENABLE_READLINE=0;BC_ENABLE_OSSFUZZ=0;BUILD_TYPE=N;BC_DEFAULT_BANNER=1;BC_DEFAULT_SIGINT_RESET=0;DC_DEFAULT_SIGINT_RESET=0;BC_DEFAULT_TTY_MODE=1;DC_DEFAULT_TTY_MODE=1;BC_DEFAULT_PROMPT=1;DC_DEFAULT_PROMPT=1;BC_DEFAULT_EXPR_EXIT=1;DC_DEFAULT_EXPR_EXIT=1;BC_DEFAULT_DIGIT_CLAMP=1;DC_DEFAULT_DIGIT_CLAMP=1;WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions) ..\include MultiThreadedDebug true Console true bcrypt.lib;%(AdditionalDependencies) copy /b /y $(OutDir)bc.exe $(OutDir)dc.exe /std:c17 /MP $(AdditionalOptions) Level3 true true true - BC_ENABLED=1;DC_ENABLED=1;BC_ENABLE_EXTRA_MATH=1;BC_ENABLE_HISTORY=1;BC_ENABLE_NLS=0;BC_DEBUG_CODE=0;BC_ENABLE_LIBRARY=0;BC_ENABLE_EDITLINE=0;BC_ENABLE_READLINE=0;BUILD_TYPE=N;BC_DEFAULT_BANNER=1;BC_DEFAULT_SIGINT_RESET=0;DC_DEFAULT_SIGINT_RESET=0;BC_DEFAULT_TTY_MODE=1;DC_DEFAULT_TTY_MODE=1;BC_DEFAULT_PROMPT=1;DC_DEFAULT_PROMPT=1;BC_DEFAULT_EXPR_EXIT=1;DC_DEFAULT_EXPR_EXIT=1;BC_DEFAULT_DIGIT_CLAMP=1;DC_DEFAULT_DIGIT_CLAMP=1;WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions) + BC_ENABLED=1;DC_ENABLED=1;BC_ENABLE_EXTRA_MATH=1;BC_ENABLE_HISTORY=1;BC_ENABLE_NLS=0;BC_DEBUG_CODE=0;BC_ENABLE_LIBRARY=0;BC_ENABLE_EDITLINE=0;BC_ENABLE_READLINE=0;BC_ENABLE_OSSFUZZ=0;BUILD_TYPE=N;BC_DEFAULT_BANNER=1;BC_DEFAULT_SIGINT_RESET=0;DC_DEFAULT_SIGINT_RESET=0;BC_DEFAULT_TTY_MODE=1;DC_DEFAULT_TTY_MODE=1;BC_DEFAULT_PROMPT=1;DC_DEFAULT_PROMPT=1;BC_DEFAULT_EXPR_EXIT=1;DC_DEFAULT_EXPR_EXIT=1;BC_DEFAULT_DIGIT_CLAMP=1;DC_DEFAULT_DIGIT_CLAMP=1;WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions) ..\include MultiThreaded true Console true true false bcrypt.lib;%(AdditionalDependencies) copy /b /y $(OutDir)bc.exe $(OutDir)dc.exe /std:c17 /MP $(AdditionalOptions) Level3 true - BC_ENABLED=1;DC_ENABLED=1;BC_ENABLE_EXTRA_MATH=1;BC_ENABLE_HISTORY=1;BC_ENABLE_NLS=0;BC_DEBUG_CODE=0;BC_ENABLE_LIBRARY=0;BC_ENABLE_EDITLINE=0;BC_ENABLE_READLINE=0;BUILD_TYPE=N;BC_DEFAULT_BANNER=1;BC_DEFAULT_SIGINT_RESET=0;DC_DEFAULT_SIGINT_RESET=0;BC_DEFAULT_TTY_MODE=1;DC_DEFAULT_TTY_MODE=1;BC_DEFAULT_PROMPT=1;DC_DEFAULT_PROMPT=1;BC_DEFAULT_EXPR_EXIT=1;DC_DEFAULT_EXPR_EXIT=1;BC_DEFAULT_DIGIT_CLAMP=1;DC_DEFAULT_DIGIT_CLAMP=1;_DEBUG;_CONSOLE;%(PreprocessorDefinitions) + BC_ENABLED=1;DC_ENABLED=1;BC_ENABLE_EXTRA_MATH=1;BC_ENABLE_HISTORY=1;BC_ENABLE_NLS=0;BC_DEBUG_CODE=0;BC_ENABLE_LIBRARY=0;BC_ENABLE_EDITLINE=0;BC_ENABLE_READLINE=0;BC_ENABLE_OSSFUZZ=0;BUILD_TYPE=N;BC_DEFAULT_BANNER=1;BC_DEFAULT_SIGINT_RESET=0;DC_DEFAULT_SIGINT_RESET=0;BC_DEFAULT_TTY_MODE=1;DC_DEFAULT_TTY_MODE=1;BC_DEFAULT_PROMPT=1;DC_DEFAULT_PROMPT=1;BC_DEFAULT_EXPR_EXIT=1;DC_DEFAULT_EXPR_EXIT=1;BC_DEFAULT_DIGIT_CLAMP=1;DC_DEFAULT_DIGIT_CLAMP=1;_DEBUG;_CONSOLE;%(PreprocessorDefinitions) ..\include MultiThreadedDebug true Console true bcrypt.lib;%(AdditionalDependencies) copy /b /y $(OutDir)bc.exe $(OutDir)dc.exe /std:c17 /MP $(AdditionalOptions) Level3 true true - BC_ENABLED=1;DC_ENABLED=1;BC_ENABLE_EXTRA_MATH=1;BC_ENABLE_HISTORY=1;BC_ENABLE_NLS=0;BC_DEBUG_CODE=0;BC_ENABLE_LIBRARY=0;BC_ENABLE_EDITLINE=0;BC_ENABLE_READLINE=0;BUILD_TYPE=N;BC_DEFAULT_BANNER=1;BC_DEFAULT_SIGINT_RESET=0;DC_DEFAULT_SIGINT_RESET=0;BC_DEFAULT_TTY_MODE=1;DC_DEFAULT_TTY_MODE=1;BC_DEFAULT_PROMPT=1;DC_DEFAULT_PROMPT=1;BC_DEFAULT_EXPR_EXIT=1;DC_DEFAULT_EXPR_EXIT=1;BC_DEFAULT_DIGIT_CLAMP=1;DC_DEFAULT_DIGIT_CLAMP=1;NDEBUG;_CONSOLE;%(PreprocessorDefinitions) + BC_ENABLED=1;DC_ENABLED=1;BC_ENABLE_EXTRA_MATH=1;BC_ENABLE_HISTORY=1;BC_ENABLE_NLS=0;BC_DEBUG_CODE=0;BC_ENABLE_LIBRARY=0;BC_ENABLE_EDITLINE=0;BC_ENABLE_READLINE=0;BC_ENABLE_OSSFUZZ=0;BUILD_TYPE=N;BC_DEFAULT_BANNER=1;BC_DEFAULT_SIGINT_RESET=0;DC_DEFAULT_SIGINT_RESET=0;BC_DEFAULT_TTY_MODE=1;DC_DEFAULT_TTY_MODE=1;BC_DEFAULT_PROMPT=1;DC_DEFAULT_PROMPT=1;BC_DEFAULT_EXPR_EXIT=1;DC_DEFAULT_EXPR_EXIT=1;BC_DEFAULT_DIGIT_CLAMP=1;DC_DEFAULT_DIGIT_CLAMP=1;NDEBUG;_CONSOLE;%(PreprocessorDefinitions) ..\include MultiThreaded true true Console true true false bcrypt.lib;%(AdditionalDependencies) copy /b /y $(OutDir)bc.exe $(OutDir)dc.exe CppCode cl.exe /I..\include /DBC_ENABLE_LIBRARY=0 /Fo:$(OutDir)strgen.obj /Fe:$(OutDir)strgen.exe %(Identity) $(OutDir)strgen.exe cl.exe /I..\include /DBC_ENABLE_LIBRARY=0 /Fo:$(OutDir)strgen.obj /Fe:$(OutDir)strgen.exe %(Identity) $(OutDir)strgen.exe cl.exe /I..\include /DBC_ENABLE_LIBRARY=0 /Fo:$(OutDir)strgen.obj /Fe:$(OutDir)strgen.exe %(Identity) $(OutDir)strgen.exe cl.exe /I..\include /DBC_ENABLE_LIBRARY=0 /Fo:$(OutDir)strgen.obj /Fe:$(OutDir)strgen.exe %(Identity) $(OutDir)strgen.exe Document $(OutDir)strgen.exe %(Identity) src2\lib.c 0 bc_lib bc_lib_name BC_ENABLED 1 src2\lib.c $(OutDir)strgen.exe %(Identity) src2\lib.c 0 bc_lib bc_lib_name BC_ENABLED 1 src2\lib.c $(OutDir)strgen.exe %(Identity) src2\lib.c 0 bc_lib bc_lib_name BC_ENABLED 1 src2\lib.c $(OutDir)strgen.exe %(Identity) src2\lib.c 0 bc_lib bc_lib_name BC_ENABLED 1 src2\lib.c Document $(OutDir)strgen.exe %(Identity) src2\lib2.c 0 bc_lib2 bc_lib2_name BC_ENABLED 1 src2\lib2.c $(OutDir)strgen.exe %(Identity) src2\lib2.c 0 bc_lib2 bc_lib2_name BC_ENABLED 1 src2\lib2.c $(OutDir)strgen.exe %(Identity) src2\lib2.c 0 bc_lib2 bc_lib2_name BC_ENABLED 1 src2\lib2.c $(OutDir)strgen.exe %(Identity) src2\lib2.c 0 bc_lib2 bc_lib2_name BC_ENABLED 1 src2\lib2.c $(OutDir)strgen.exe %(Identity) src2\dc_help.c 0 dc_help "" DC_ENABLED src2\dc_help.c $(OutDir)strgen.exe %(Identity) src2\dc_help.c 0 dc_help "" DC_ENABLED src2\dc_help.c $(OutDir)strgen.exe %(Identity) src2\dc_help.c 0 dc_help "" DC_ENABLED src2\dc_help.c $(OutDir)strgen.exe %(Identity) src2\dc_help.c 0 dc_help "" DC_ENABLED src2\dc_help.c $(OutDir)strgen.exe %(Identity) src2\bc_help.c 0 bc_help "" BC_ENABLED src2\bc_help.c $(OutDir)strgen.exe %(Identity) src2\bc_help.c 0 bc_help "" BC_ENABLED src2\bc_help.c $(OutDir)strgen.exe %(Identity) src2\bc_help.c 0 bc_help "" BC_ENABLED src2\bc_help.c $(OutDir)strgen.exe %(Identity) src2\bc_help.c 0 bc_help "" BC_ENABLED src2\bc_help.c - + \ No newline at end of file