Page MenuHomeFreeBSD
Files F136931713

D18297.id50940.diff
No OneTemporary
Actions

D18297.id50940.diff
View Options

Index: lib/libc/regex/engine.c
===================================================================
--- lib/libc/regex/engine.c
+++ lib/libc/regex/engine.c
@@ -48,6 +48,7 @@
*/
#ifdef SNAMES
+#define stepback sstepback
#define matcher smatcher
#define walk swalk
#define dissect sdissect
@@ -58,6 +59,7 @@
#define match smat
#endif
#ifdef LNAMES
+#define stepback lstepback
#define matcher lmatcher
#define walk lwalk
#define dissect ldissect
@@ -68,6 +70,7 @@
#define match lmat
#endif
#ifdef MNAMES
+#define stepback mstepback
#define matcher mmatcher
#define walk mwalk
#define dissect mdissect
@@ -141,6 +144,39 @@
#define NOTE(s) /* nothing */
#endif
+/*
+ * Given a multibyte string pointed to by start, step back nchar characters
+ * from current position pointed to by cur.
+ */
+static const char *
+stepback(const char *start, const char *cur, int nchar)
+{
+ const char *ret;
+ int wc, mbc;
+ mbstate_t mbs;
+ size_t clen;
+
+ if (MB_CUR_MAX == 1)
+ return ((cur - nchar) > start ? cur - nchar : NULL);
+
+ ret = cur;
+ for (wc = nchar; wc > 0; wc--) {
+ for (mbc = 1; mbc <= MB_CUR_MAX; mbc++) {
+ if ((ret - mbc) < start)
+ return (NULL);
+ memset(&mbs, 0, sizeof(mbs));
+ clen = mbrtowc(NULL, ret - mbc, mbc, &mbs);
+ if (clen != (size_t)-1 && clen != (size_t)-2)
+ break;
+ }
+ if (mbc > MB_CUR_MAX)
+ return (NULL);
+ ret -= mbc;
+ }
+
+ return (ret);
+}
+
/*
- matcher - the actual matching engine
== static int matcher(struct re_guts *g, const char *string, \
@@ -244,8 +280,13 @@
ZAPSTATE(&m->mbs);
/* Adjust start according to moffset, to speed things up */
- if (dp != NULL && g->moffset > -1)
- start = ((dp - g->moffset) < start) ? start : dp - g->moffset;
+ if (dp != NULL && g->moffset > -1) {
+ const char *nstart;
+
+ nstart = stepback(start, dp, g->moffset);
+ if (nstart != NULL)
+ start = nstart;
+ }
SP("mloop", m->st, *start);
@@ -1083,6 +1124,7 @@
#endif
#endif
+#undef stepback
#undef matcher
#undef walk
#undef dissect
Index: lib/libc/tests/regex/Makefile.inc
===================================================================
--- lib/libc/tests/regex/Makefile.inc
+++ lib/libc/tests/regex/Makefile.inc
@@ -4,6 +4,9 @@
BINDIR?= ${TESTSDIR}
+# local test cases
+ATF_TESTS_SH+= multibyte
+
# SKIP_LEFTASSOC -> these testcases fail on FreeBSD.
IMPLEMENTATION?= -DREGEX_SPENCER -DSKIP_LEFTASSOC
Index: lib/libc/tests/regex/multibyte.sh
===================================================================
--- /dev/null
+++ lib/libc/tests/regex/multibyte.sh
@@ -0,0 +1,35 @@
+# $FreeBSD$
+
+atf_test_case multibyte
+multibyte_head()
+{
+ atf_set "descr" "Check matching multibyte characters (PR153502)"
+}
+multibyte_body()
+{
+ export LC_CTYPE="C.UTF-8"
+
+ printf 'é' | atf_check -o "inline:é" \
+ sed -ne '/^.$/p'
+ printf 'éé' | atf_check -o "inline:éé" \
+ sed -ne '/^..$/p'
+ printf 'aéa' | atf_check -o "inline:aéa" \
+ sed -ne '/a.a/p'
+ printf 'aéa'| atf_check -o "inline:aéa" \
+ sed -ne '/a.*a/p'
+ printf 'aaéaa' | atf_check -o "inline:aaéaa" \
+ sed -ne '/aa.aa/p'
+ printf 'aéaéa' | atf_check -o "inline:aéaéa" \
+ sed -ne '/a.a.a/p'
+ printf 'éa' | atf_check -o "inline:éa" \
+ sed -ne '/.a/p'
+ printf 'aéaa' | atf_check -o "inline:aéaa" \
+ sed -ne '/a.aa/p'
+ printf 'éaé' | atf_check -o "inline:éaé" \
+ sed -ne '/.a./p'
+}
+
+atf_init_test_cases()
+{
+ atf_add_test_case multibyte
+}

File Metadata

Mime Type
text/plain
Expires
Fri, Nov 21, 5:36 PM (3 h, 49 m)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
25782535
Default Alt Text
D18297.id50940.diff (3 KB)

Event Timeline