Index: head/contrib/libarchive/libarchive/archive_acl.c =================================================================== --- head/contrib/libarchive/libarchive/archive_acl.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_acl.c (revision 311903) @@ -1,2063 +1,2067 @@ /*- * Copyright (c) 2003-2010 Tim Kientzle * Copyright (c) 2016 Martin Matuska * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_WCHAR_H #include #endif #include "archive_acl_private.h" #include "archive_entry.h" #include "archive_private.h" #undef max #define max(a, b) ((a)>(b)?(a):(b)) #ifndef HAVE_WMEMCMP /* Good enough for simple equality testing, but not for sorting. */ #define wmemcmp(a,b,i) memcmp((a), (b), (i) * sizeof(wchar_t)) #endif static int acl_special(struct archive_acl *acl, int type, int permset, int tag); static struct archive_acl_entry *acl_new_entry(struct archive_acl *acl, int type, int permset, int tag, int id); static int archive_acl_add_entry_len_l(struct archive_acl *acl, int type, int permset, int tag, int id, const char *name, size_t len, struct archive_string_conv *sc); static int archive_acl_text_want_type(struct archive_acl *acl, int flags); static ssize_t archive_acl_text_len(struct archive_acl *acl, int want_type, int flags, int wide, struct archive *a, struct archive_string_conv *sc); static int isint_w(const wchar_t *start, const wchar_t *end, int *result); static int ismode_w(const wchar_t *start, const wchar_t *end, int *result); static int is_nfs4_flags_w(const wchar_t *start, const wchar_t *end, int *result); static int is_nfs4_perms_w(const wchar_t *start, const wchar_t *end, int *result); static void next_field_w(const wchar_t **wp, const wchar_t **start, const wchar_t **end, wchar_t *sep); static void append_entry_w(wchar_t **wp, const wchar_t *prefix, int type, int tag, int flags, const wchar_t *wname, int perm, int id); static void append_id_w(wchar_t **wp, int id); static int isint(const char *start, const char *end, int *result); static int ismode(const char *start, const char *end, int *result); static int is_nfs4_flags(const char *start, const char *end, int *result); static int is_nfs4_perms(const char *start, const char *end, int *result); static void next_field(const char **p, const char **start, const char **end, char *sep); static void append_entry(char **p, const char *prefix, int type, int tag, int flags, const char *name, int perm, int id); static void append_id(char **p, int id); void archive_acl_clear(struct archive_acl *acl) { struct archive_acl_entry *ap; while (acl->acl_head != NULL) { ap = acl->acl_head->next; archive_mstring_clean(&acl->acl_head->name); free(acl->acl_head); acl->acl_head = ap; } if (acl->acl_text_w != NULL) { free(acl->acl_text_w); acl->acl_text_w = NULL; } if (acl->acl_text != NULL) { free(acl->acl_text); acl->acl_text = NULL; } acl->acl_p = NULL; acl->acl_types = 0; acl->acl_state = 0; /* Not counting. */ } void archive_acl_copy(struct archive_acl *dest, struct archive_acl *src) { struct archive_acl_entry *ap, *ap2; archive_acl_clear(dest); dest->mode = src->mode; ap = src->acl_head; while (ap != NULL) { ap2 = acl_new_entry(dest, ap->type, ap->permset, ap->tag, ap->id); if (ap2 != NULL) archive_mstring_copy(&ap2->name, &ap->name); ap = ap->next; } } int archive_acl_add_entry(struct archive_acl *acl, int type, int permset, int tag, int id, const char *name) { struct archive_acl_entry *ap; if (acl_special(acl, type, permset, tag) == 0) return ARCHIVE_OK; ap = acl_new_entry(acl, type, permset, tag, id); if (ap == NULL) { /* XXX Error XXX */ return ARCHIVE_FAILED; } if (name != NULL && *name != '\0') archive_mstring_copy_mbs(&ap->name, name); else archive_mstring_clean(&ap->name); return ARCHIVE_OK; } int archive_acl_add_entry_w_len(struct archive_acl *acl, int type, int permset, int tag, int id, const wchar_t *name, size_t len) { struct archive_acl_entry *ap; if (acl_special(acl, type, permset, tag) == 0) return ARCHIVE_OK; ap = acl_new_entry(acl, type, permset, tag, id); if (ap == NULL) { /* XXX Error XXX */ return ARCHIVE_FAILED; } if (name != NULL && *name != L'\0' && len > 0) archive_mstring_copy_wcs_len(&ap->name, name, len); else archive_mstring_clean(&ap->name); return ARCHIVE_OK; } static int archive_acl_add_entry_len_l(struct archive_acl *acl, int type, int permset, int tag, int id, const char *name, size_t len, struct archive_string_conv *sc) { struct archive_acl_entry *ap; int r; if (acl_special(acl, type, permset, tag) == 0) return ARCHIVE_OK; ap = acl_new_entry(acl, type, permset, tag, id); if (ap == NULL) { /* XXX Error XXX */ return ARCHIVE_FAILED; } if (name != NULL && *name != '\0' && len > 0) { r = archive_mstring_copy_mbs_len_l(&ap->name, name, len, sc); } else { r = 0; archive_mstring_clean(&ap->name); } if (r == 0) return (ARCHIVE_OK); else if (errno == ENOMEM) return (ARCHIVE_FATAL); else return (ARCHIVE_WARN); } /* * If this ACL entry is part of the standard POSIX permissions set, * store the permissions in the stat structure and return zero. */ static int acl_special(struct archive_acl *acl, int type, int permset, int tag) { if (type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS && ((permset & ~007) == 0)) { switch (tag) { case ARCHIVE_ENTRY_ACL_USER_OBJ: acl->mode &= ~0700; acl->mode |= (permset & 7) << 6; return (0); case ARCHIVE_ENTRY_ACL_GROUP_OBJ: acl->mode &= ~0070; acl->mode |= (permset & 7) << 3; return (0); case ARCHIVE_ENTRY_ACL_OTHER: acl->mode &= ~0007; acl->mode |= permset & 7; return (0); } } return (1); } /* * Allocate and populate a new ACL entry with everything but the * name. */ static struct archive_acl_entry * acl_new_entry(struct archive_acl *acl, int type, int permset, int tag, int id) { struct archive_acl_entry *ap, *aq; /* Type argument must be a valid NFS4 or POSIX.1e type. * The type must agree with anything already set and * the permset must be compatible. */ if (type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) { if (acl->acl_types & ~ARCHIVE_ENTRY_ACL_TYPE_NFS4) { return (NULL); } if (permset & ~(ARCHIVE_ENTRY_ACL_PERMS_NFS4 | ARCHIVE_ENTRY_ACL_INHERITANCE_NFS4)) { return (NULL); } } else if (type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) { if (acl->acl_types & ~ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) { return (NULL); } if (permset & ~ARCHIVE_ENTRY_ACL_PERMS_POSIX1E) { return (NULL); } } else { return (NULL); } /* Verify the tag is valid and compatible with NFS4 or POSIX.1e. */ switch (tag) { case ARCHIVE_ENTRY_ACL_USER: case ARCHIVE_ENTRY_ACL_USER_OBJ: case ARCHIVE_ENTRY_ACL_GROUP: case ARCHIVE_ENTRY_ACL_GROUP_OBJ: /* Tags valid in both NFS4 and POSIX.1e */ break; case ARCHIVE_ENTRY_ACL_MASK: case ARCHIVE_ENTRY_ACL_OTHER: /* Tags valid only in POSIX.1e. */ if (type & ~ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) { return (NULL); } break; case ARCHIVE_ENTRY_ACL_EVERYONE: /* Tags valid only in NFS4. */ if (type & ~ARCHIVE_ENTRY_ACL_TYPE_NFS4) { return (NULL); } break; default: /* No other values are valid. */ return (NULL); } if (acl->acl_text_w != NULL) { free(acl->acl_text_w); acl->acl_text_w = NULL; } if (acl->acl_text != NULL) { free(acl->acl_text); acl->acl_text = NULL; } /* * If there's a matching entry already in the list, overwrite it. * NFSv4 entries may be repeated and are not overwritten. * * TODO: compare names of no id is provided (needs more rework) */ ap = acl->acl_head; aq = NULL; while (ap != NULL) { if (((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) == 0) && ap->type == type && ap->tag == tag && ap->id == id) { if (id != -1 || (tag != ARCHIVE_ENTRY_ACL_USER && tag != ARCHIVE_ENTRY_ACL_GROUP)) { ap->permset = permset; return (ap); } } aq = ap; ap = ap->next; } /* Add a new entry to the end of the list. */ ap = (struct archive_acl_entry *)calloc(1, sizeof(*ap)); if (ap == NULL) return (NULL); if (aq == NULL) acl->acl_head = ap; else aq->next = ap; ap->type = type; ap->tag = tag; ap->id = id; ap->permset = permset; acl->acl_types |= type; return (ap); } /* * Return a count of entries matching "want_type". */ int archive_acl_count(struct archive_acl *acl, int want_type) { int count; struct archive_acl_entry *ap; count = 0; ap = acl->acl_head; while (ap != NULL) { if ((ap->type & want_type) != 0) count++; ap = ap->next; } if (count > 0 && ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0)) count += 3; return (count); } /* * Prepare for reading entries from the ACL data. Returns a count * of entries matching "want_type", or zero if there are no * non-extended ACL entries of that type. */ int archive_acl_reset(struct archive_acl *acl, int want_type) { int count, cutoff; count = archive_acl_count(acl, want_type); /* * If the only entries are the three standard ones, * then don't return any ACL data. (In this case, * client can just use chmod(2) to set permissions.) */ if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) cutoff = 3; else cutoff = 0; if (count > cutoff) acl->acl_state = ARCHIVE_ENTRY_ACL_USER_OBJ; else acl->acl_state = 0; acl->acl_p = acl->acl_head; return (count); } /* * Return the next ACL entry in the list. Fake entries for the * standard permissions and include them in the returned list. */ int archive_acl_next(struct archive *a, struct archive_acl *acl, int want_type, int *type, int *permset, int *tag, int *id, const char **name) { *name = NULL; *id = -1; /* * The acl_state is either zero (no entries available), -1 * (reading from list), or an entry type (retrieve that type * from ae_stat.aest_mode). */ if (acl->acl_state == 0) return (ARCHIVE_WARN); /* The first three access entries are special. */ if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) { switch (acl->acl_state) { case ARCHIVE_ENTRY_ACL_USER_OBJ: *permset = (acl->mode >> 6) & 7; *type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; *tag = ARCHIVE_ENTRY_ACL_USER_OBJ; acl->acl_state = ARCHIVE_ENTRY_ACL_GROUP_OBJ; return (ARCHIVE_OK); case ARCHIVE_ENTRY_ACL_GROUP_OBJ: *permset = (acl->mode >> 3) & 7; *type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; *tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; acl->acl_state = ARCHIVE_ENTRY_ACL_OTHER; return (ARCHIVE_OK); case ARCHIVE_ENTRY_ACL_OTHER: *permset = acl->mode & 7; *type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; *tag = ARCHIVE_ENTRY_ACL_OTHER; acl->acl_state = -1; acl->acl_p = acl->acl_head; return (ARCHIVE_OK); default: break; } } while (acl->acl_p != NULL && (acl->acl_p->type & want_type) == 0) acl->acl_p = acl->acl_p->next; if (acl->acl_p == NULL) { acl->acl_state = 0; *type = 0; *permset = 0; *tag = 0; *id = -1; *name = NULL; return (ARCHIVE_EOF); /* End of ACL entries. */ } *type = acl->acl_p->type; *permset = acl->acl_p->permset; *tag = acl->acl_p->tag; *id = acl->acl_p->id; if (archive_mstring_get_mbs(a, &acl->acl_p->name, name) != 0) { if (errno == ENOMEM) return (ARCHIVE_FATAL); *name = NULL; } acl->acl_p = acl->acl_p->next; return (ARCHIVE_OK); } /* * Determine what type of ACL do we want */ static int archive_acl_text_want_type(struct archive_acl *acl, int flags) { int want_type; /* Check if ACL is NFSv4 */ if ((acl->acl_types & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { /* NFSv4 should never mix with POSIX.1e */ if ((acl->acl_types & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) return (0); else return (ARCHIVE_ENTRY_ACL_TYPE_NFS4); } /* Now deal with POSIX.1e ACLs */ want_type = 0; if ((flags & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) want_type |= ARCHIVE_ENTRY_ACL_TYPE_ACCESS; if ((flags & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0) want_type |= ARCHIVE_ENTRY_ACL_TYPE_DEFAULT; /* By default we want both access and default ACLs */ if (want_type == 0) return (ARCHIVE_ENTRY_ACL_TYPE_POSIX1E); return (want_type); } /* * Calculate ACL text string length */ static ssize_t archive_acl_text_len(struct archive_acl *acl, int want_type, int flags, int wide, struct archive *a, struct archive_string_conv *sc) { struct archive_acl_entry *ap; const char *name; const wchar_t *wname; int count, idlen, tmp, r; ssize_t length; size_t len; count = 0; length = 0; for (ap = acl->acl_head; ap != NULL; ap = ap->next) { if ((ap->type & want_type) == 0) continue; /* * Filemode-mapping ACL entries are stored exclusively in * ap->mode so they should not be in the list */ if ((ap->type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS) && (ap->tag == ARCHIVE_ENTRY_ACL_USER_OBJ || ap->tag == ARCHIVE_ENTRY_ACL_GROUP_OBJ || ap->tag == ARCHIVE_ENTRY_ACL_OTHER)) continue; count++; if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0 && (ap->type & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0) length += 8; /* "default:" */ switch (ap->tag) { case ARCHIVE_ENTRY_ACL_USER_OBJ: if (want_type == ARCHIVE_ENTRY_ACL_TYPE_NFS4) { length += 6; /* "owner@" */ break; } /* FALLTHROUGH */ case ARCHIVE_ENTRY_ACL_USER: case ARCHIVE_ENTRY_ACL_MASK: length += 4; /* "user", "mask" */ break; case ARCHIVE_ENTRY_ACL_GROUP_OBJ: if (want_type == ARCHIVE_ENTRY_ACL_TYPE_NFS4) { length += 6; /* "group@" */ break; } /* FALLTHROUGH */ case ARCHIVE_ENTRY_ACL_GROUP: case ARCHIVE_ENTRY_ACL_OTHER: length += 5; /* "group", "other" */ break; case ARCHIVE_ENTRY_ACL_EVERYONE: length += 9; /* "everyone@" */ break; } length += 1; /* colon after tag */ if (ap->tag == ARCHIVE_ENTRY_ACL_USER || ap->tag == ARCHIVE_ENTRY_ACL_GROUP) { if (wide) { r = archive_mstring_get_wcs(a, &ap->name, &wname); if (r == 0 && wname != NULL) length += wcslen(wname); else if (r < 0 && errno == ENOMEM) return (0); else length += sizeof(uid_t) * 3 + 1; } else { r = archive_mstring_get_mbs_l(&ap->name, &name, &len, sc); if (r != 0) return (0); if (len > 0 && name != NULL) length += len; else length += sizeof(uid_t) * 3 + 1; } length += 1; /* colon after user or group name */ } else if (want_type != ARCHIVE_ENTRY_ACL_TYPE_NFS4) length += 1; /* 2nd colon empty user,group or other */ if (((flags & ARCHIVE_ENTRY_ACL_STYLE_SOLARIS) != 0) && ((want_type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) && (ap->tag == ARCHIVE_ENTRY_ACL_OTHER || ap->tag == ARCHIVE_ENTRY_ACL_MASK)) { /* Solaris has no colon after other: and mask: */ length = length - 1; } if (want_type == ARCHIVE_ENTRY_ACL_TYPE_NFS4) { /* rwxpdDaARWcCos:fdinSFI:deny */ length += 27; if ((ap->type & ARCHIVE_ENTRY_ACL_TYPE_DENY) == 0) length += 1; /* allow, alarm, audit */ } else length += 3; /* rwx */ if ((ap->tag == ARCHIVE_ENTRY_ACL_USER || ap->tag == ARCHIVE_ENTRY_ACL_GROUP) && (flags & ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID) != 0) { length += 1; /* colon */ /* ID digit count */ idlen = 1; tmp = ap->id; while (tmp > 9) { tmp = tmp / 10; idlen++; } length += idlen; } length ++; /* entry separator */ } /* Add filemode-mapping access entries to the length */ if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) { if ((flags & ARCHIVE_ENTRY_ACL_STYLE_SOLARIS) != 0) { /* "user::rwx\ngroup::rwx\nother:rwx\n" */ length += 31; } else { /* "user::rwx\ngroup::rwx\nother::rwx\n" */ length += 32; } } else if (count == 0) return (0); /* The terminating character is included in count */ return (length); } /* * Generate a wide text version of the ACL. The flags parameter controls * the type and style of the generated ACL. */ wchar_t * archive_acl_to_text_w(struct archive_acl *acl, ssize_t *text_len, int flags, struct archive *a) { int count; ssize_t length; size_t len; const wchar_t *wname; const wchar_t *prefix; wchar_t separator; struct archive_acl_entry *ap; int id, r, want_type; wchar_t *wp, *ws; want_type = archive_acl_text_want_type(acl, flags); /* Both NFSv4 and POSIX.1 types found */ if (want_type == 0) return (NULL); if (want_type == ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) flags |= ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT; length = archive_acl_text_len(acl, want_type, flags, 1, a, NULL); if (length == 0) return (NULL); if (flags & ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA) separator = L','; else separator = L'\n'; /* Now, allocate the string and actually populate it. */ wp = ws = (wchar_t *)malloc(length * sizeof(wchar_t)); if (wp == NULL) { if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } count = 0; if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) { append_entry_w(&wp, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_USER_OBJ, flags, NULL, acl->mode & 0700, -1); *wp++ = separator; append_entry_w(&wp, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_GROUP_OBJ, flags, NULL, acl->mode & 0070, -1); *wp++ = separator; append_entry_w(&wp, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_OTHER, flags, NULL, acl->mode & 0007, -1); count += 3; } for (ap = acl->acl_head; ap != NULL; ap = ap->next) { if ((ap->type & want_type) == 0) continue; /* * Filemode-mapping ACL entries are stored exclusively in * ap->mode so they should not be in the list */ if ((ap->type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS) && (ap->tag == ARCHIVE_ENTRY_ACL_USER_OBJ || ap->tag == ARCHIVE_ENTRY_ACL_GROUP_OBJ || ap->tag == ARCHIVE_ENTRY_ACL_OTHER)) continue; if (ap->type == ARCHIVE_ENTRY_ACL_TYPE_DEFAULT && (flags & ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT) != 0) prefix = L"default:"; else prefix = NULL; r = archive_mstring_get_wcs(a, &ap->name, &wname); if (r == 0) { if (count > 0) *wp++ = separator; if (flags & ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID) id = ap->id; else id = -1; append_entry_w(&wp, prefix, ap->type, ap->tag, flags, wname, ap->permset, id); count++; } else if (r < 0 && errno == ENOMEM) return (NULL); } /* Add terminating character */ *wp++ = L'\0'; len = wcslen(ws); if ((ssize_t)len > (length - 1)) __archive_errx(1, "Buffer overrun"); if (text_len != NULL) *text_len = len; return (ws); } static void append_id_w(wchar_t **wp, int id) { if (id < 0) id = 0; if (id > 9) append_id_w(wp, id / 10); *(*wp)++ = L"0123456789"[id % 10]; } static void append_entry_w(wchar_t **wp, const wchar_t *prefix, int type, int tag, int flags, const wchar_t *wname, int perm, int id) { if (prefix != NULL) { wcscpy(*wp, prefix); *wp += wcslen(*wp); } switch (tag) { case ARCHIVE_ENTRY_ACL_USER_OBJ: wname = NULL; id = -1; if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { wcscpy(*wp, L"owner@"); break; } /* FALLTHROUGH */ case ARCHIVE_ENTRY_ACL_USER: wcscpy(*wp, L"user"); break; case ARCHIVE_ENTRY_ACL_GROUP_OBJ: wname = NULL; id = -1; if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { wcscpy(*wp, L"group@"); break; } /* FALLTHROUGH */ case ARCHIVE_ENTRY_ACL_GROUP: wcscpy(*wp, L"group"); break; case ARCHIVE_ENTRY_ACL_MASK: wcscpy(*wp, L"mask"); wname = NULL; id = -1; break; case ARCHIVE_ENTRY_ACL_OTHER: wcscpy(*wp, L"other"); wname = NULL; id = -1; break; case ARCHIVE_ENTRY_ACL_EVERYONE: wcscpy(*wp, L"everyone@"); wname = NULL; id = -1; break; } *wp += wcslen(*wp); *(*wp)++ = L':'; if (((type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) || tag == ARCHIVE_ENTRY_ACL_USER || tag == ARCHIVE_ENTRY_ACL_GROUP) { if (wname != NULL) { wcscpy(*wp, wname); *wp += wcslen(*wp); } else if (tag == ARCHIVE_ENTRY_ACL_USER || tag == ARCHIVE_ENTRY_ACL_GROUP) { append_id_w(wp, id); - id = -1; + if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) == 0) + id = -1; } /* Solaris style has no second colon after other and mask */ if (((flags & ARCHIVE_ENTRY_ACL_STYLE_SOLARIS) == 0) || (tag != ARCHIVE_ENTRY_ACL_OTHER && tag != ARCHIVE_ENTRY_ACL_MASK)) *(*wp)++ = L':'; } if ((type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) { /* POSIX.1e ACL perms */ *(*wp)++ = (perm & 0444) ? L'r' : L'-'; *(*wp)++ = (perm & 0222) ? L'w' : L'-'; *(*wp)++ = (perm & 0111) ? L'x' : L'-'; } else { /* NFS4 ACL perms */ *(*wp)++ = (perm & (ARCHIVE_ENTRY_ACL_READ_DATA | ARCHIVE_ENTRY_ACL_LIST_DIRECTORY)) ? L'r' : L'-'; *(*wp)++ = (perm & (ARCHIVE_ENTRY_ACL_WRITE_DATA | ARCHIVE_ENTRY_ACL_ADD_FILE)) ? L'w' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_EXECUTE) ? L'x' : L'-'; *(*wp)++ = (perm & (ARCHIVE_ENTRY_ACL_APPEND_DATA | ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY)) ? L'p' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_DELETE) ? L'd' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_DELETE_CHILD) ? L'D' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES) ? L'a' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES) ? L'A' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS) ? L'R' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS) ? L'W' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_READ_ACL) ? L'c' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_WRITE_ACL) ? L'C' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_WRITE_OWNER) ? L'o' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_SYNCHRONIZE) ? L's' : L'-'; *(*wp)++ = L':'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT) ? L'f' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT) ? L'd' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY) ? L'i' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT) ? L'n' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS) ? L'S' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS) ? L'F' : L'-'; *(*wp)++ = (perm & ARCHIVE_ENTRY_ACL_ENTRY_INHERITED) ? L'I' : L'-'; *(*wp)++ = L':'; switch (type) { case ARCHIVE_ENTRY_ACL_TYPE_ALLOW: wcscpy(*wp, L"allow"); break; case ARCHIVE_ENTRY_ACL_TYPE_DENY: wcscpy(*wp, L"deny"); break; case ARCHIVE_ENTRY_ACL_TYPE_AUDIT: wcscpy(*wp, L"audit"); break; case ARCHIVE_ENTRY_ACL_TYPE_ALARM: wcscpy(*wp, L"alarm"); break; default: break; } *wp += wcslen(*wp); } if (id != -1) { *(*wp)++ = L':'; append_id_w(wp, id); } } /* * Generate a text version of the ACL. The flags parameter controls * the type and style of the generated ACL. */ char * archive_acl_to_text_l(struct archive_acl *acl, ssize_t *text_len, int flags, struct archive_string_conv *sc) { int count; ssize_t length; size_t len; const char *name; const char *prefix; char separator; struct archive_acl_entry *ap; int id, r, want_type; char *p, *s; want_type = archive_acl_text_want_type(acl, flags); /* Both NFSv4 and POSIX.1 types found */ if (want_type == 0) return (NULL); if (want_type == ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) flags |= ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT; length = archive_acl_text_len(acl, want_type, flags, 0, NULL, sc); if (length == 0) return (NULL); if (flags & ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA) separator = ','; else separator = '\n'; /* Now, allocate the string and actually populate it. */ p = s = (char *)malloc(length * sizeof(char)); if (p == NULL) { if (errno == ENOMEM) __archive_errx(1, "No memory"); return (NULL); } count = 0; if ((want_type & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) { append_entry(&p, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_USER_OBJ, flags, NULL, acl->mode & 0700, -1); *p++ = separator; append_entry(&p, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_GROUP_OBJ, flags, NULL, acl->mode & 0070, -1); *p++ = separator; append_entry(&p, NULL, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_OTHER, flags, NULL, acl->mode & 0007, -1); count += 3; } for (ap = acl->acl_head; ap != NULL; ap = ap->next) { if ((ap->type & want_type) == 0) continue; /* * Filemode-mapping ACL entries are stored exclusively in * ap->mode so they should not be in the list */ if ((ap->type == ARCHIVE_ENTRY_ACL_TYPE_ACCESS) && (ap->tag == ARCHIVE_ENTRY_ACL_USER_OBJ || ap->tag == ARCHIVE_ENTRY_ACL_GROUP_OBJ || ap->tag == ARCHIVE_ENTRY_ACL_OTHER)) continue; if (ap->type == ARCHIVE_ENTRY_ACL_TYPE_DEFAULT && (flags & ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT) != 0) prefix = "default:"; else prefix = NULL; r = archive_mstring_get_mbs_l( &ap->name, &name, &len, sc); if (r != 0) return (NULL); if (count > 0) *p++ = separator; if (name == NULL || (flags & ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID)) { id = ap->id; } else { id = -1; } append_entry(&p, prefix, ap->type, ap->tag, flags, name, ap->permset, id); count++; } /* Add terminating character */ *p++ = '\0'; len = strlen(s); if ((ssize_t)len > (length - 1)) __archive_errx(1, "Buffer overrun"); if (text_len != NULL) *text_len = len; return (s); } static void append_id(char **p, int id) { if (id < 0) id = 0; if (id > 9) append_id(p, id / 10); *(*p)++ = "0123456789"[id % 10]; } static void append_entry(char **p, const char *prefix, int type, int tag, int flags, const char *name, int perm, int id) { if (prefix != NULL) { strcpy(*p, prefix); *p += strlen(*p); } switch (tag) { case ARCHIVE_ENTRY_ACL_USER_OBJ: name = NULL; id = -1; if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { strcpy(*p, "owner@"); break; } /* FALLTHROUGH */ case ARCHIVE_ENTRY_ACL_USER: strcpy(*p, "user"); break; case ARCHIVE_ENTRY_ACL_GROUP_OBJ: name = NULL; id = -1; if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { strcpy(*p, "group@"); break; } /* FALLTHROUGH */ case ARCHIVE_ENTRY_ACL_GROUP: strcpy(*p, "group"); break; case ARCHIVE_ENTRY_ACL_MASK: strcpy(*p, "mask"); name = NULL; id = -1; break; case ARCHIVE_ENTRY_ACL_OTHER: strcpy(*p, "other"); name = NULL; id = -1; break; case ARCHIVE_ENTRY_ACL_EVERYONE: strcpy(*p, "everyone@"); name = NULL; id = -1; break; } *p += strlen(*p); *(*p)++ = ':'; if (((type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) || tag == ARCHIVE_ENTRY_ACL_USER || tag == ARCHIVE_ENTRY_ACL_GROUP) { if (name != NULL) { strcpy(*p, name); *p += strlen(*p); } else if (tag == ARCHIVE_ENTRY_ACL_USER || tag == ARCHIVE_ENTRY_ACL_GROUP) { append_id(p, id); - id = -1; + if ((type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) == 0) + id = -1; } /* Solaris style has no second colon after other and mask */ if (((flags & ARCHIVE_ENTRY_ACL_STYLE_SOLARIS) == 0) || (tag != ARCHIVE_ENTRY_ACL_OTHER && tag != ARCHIVE_ENTRY_ACL_MASK)) *(*p)++ = ':'; } if ((type & ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) != 0) { /* POSIX.1e ACL perms */ *(*p)++ = (perm & 0444) ? 'r' : '-'; *(*p)++ = (perm & 0222) ? 'w' : '-'; *(*p)++ = (perm & 0111) ? 'x' : '-'; } else { /* NFS4 ACL perms */ *(*p)++ = (perm & (ARCHIVE_ENTRY_ACL_READ_DATA | ARCHIVE_ENTRY_ACL_LIST_DIRECTORY)) ? 'r' : '-'; *(*p)++ = (perm & (ARCHIVE_ENTRY_ACL_WRITE_DATA | ARCHIVE_ENTRY_ACL_ADD_FILE)) ? 'w' : '-'; *(*p)++ = (perm & (ARCHIVE_ENTRY_ACL_EXECUTE)) ? 'x' : '-'; *(*p)++ = (perm & (ARCHIVE_ENTRY_ACL_APPEND_DATA | ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY)) ? 'p' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_DELETE) ? 'd' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_DELETE_CHILD) ? 'D' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES) ? 'a' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES) ? 'A' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS) ? 'R' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS) ? 'W' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_READ_ACL) ? 'c' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_WRITE_ACL) ? 'C' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_WRITE_OWNER) ? 'o' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_SYNCHRONIZE) ? 's' : '-'; *(*p)++ = ':'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT) ? 'f' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT) ? 'd' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY) ? 'i' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT) ? 'n' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS) ? 'S' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS) ? 'F' : '-'; *(*p)++ = (perm & ARCHIVE_ENTRY_ACL_ENTRY_INHERITED) ? 'I' : '-'; *(*p)++ = ':'; switch (type) { case ARCHIVE_ENTRY_ACL_TYPE_ALLOW: strcpy(*p, "allow"); break; case ARCHIVE_ENTRY_ACL_TYPE_DENY: strcpy(*p, "deny"); break; case ARCHIVE_ENTRY_ACL_TYPE_AUDIT: strcpy(*p, "audit"); break; case ARCHIVE_ENTRY_ACL_TYPE_ALARM: strcpy(*p, "alarm"); break; } *p += strlen(*p); } if (id != -1) { *(*p)++ = ':'; append_id(p, id); } } /* * Parse a wide ACL text string. * * The want_type argument may be one of the following: * ARCHIVE_ENTRY_ACL_TYPE_ACCESS - text is a POSIX.1e ACL of type ACCESS * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT - text is a POSIX.1e ACL of type DEFAULT * ARCHIVE_ENTRY_ACL_TYPE_NFS4 - text is as a NFSv4 ACL * * POSIX.1e ACL entries prefixed with "default:" are treated as * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT unless type is ARCHIVE_ENTRY_ACL_TYPE_NFS4 */ int archive_acl_from_text_w(struct archive_acl *acl, const wchar_t *text, int want_type) { struct { const wchar_t *start; const wchar_t *end; } field[6], name; const wchar_t *s, *st; int numfields, fields, n, r, ret; int type, types, tag, permset, id; size_t len; wchar_t sep; ret = ARCHIVE_OK; types = 0; switch (want_type) { case ARCHIVE_ENTRY_ACL_TYPE_POSIX1E: want_type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: numfields = 5; break; case ARCHIVE_ENTRY_ACL_TYPE_NFS4: numfields = 6; break; default: return (ARCHIVE_FATAL); } while (text != NULL && *text != L'\0') { /* * Parse the fields out of the next entry, * advance 'text' to start of next entry. */ fields = 0; do { const wchar_t *start, *end; next_field_w(&text, &start, &end, &sep); if (fields < numfields) { field[fields].start = start; field[fields].end = end; } ++fields; } while (sep == L':'); /* Set remaining fields to blank. */ for (n = fields; n < numfields; ++n) field[n].start = field[n].end = NULL; if (field[0].start != NULL && *(field[0].start) == L'#') { /* Comment, skip entry */ continue; } n = 0; id = -1; permset = 0; name.start = name.end = NULL; if (want_type != ARCHIVE_ENTRY_ACL_TYPE_NFS4) { /* POSIX.1e ACLs */ /* * Default keyword "default:user::rwx" * if found, we have one more field * * We also support old Solaris extension: * "defaultuser::rwx" is the default ACL corresponding * to "user::rwx", etc. valid only for first field */ s = field[0].start; len = field[0].end - field[0].start; if (*s == L'd' && (len == 1 || (len >= 7 && wmemcmp((s + 1), L"efault", 6) == 0))) { type = ARCHIVE_ENTRY_ACL_TYPE_DEFAULT; if (len > 7) field[0].start += 7; else n = 1; } else type = want_type; /* Check for a numeric ID in field n+1 or n+3. */ isint_w(field[n + 1].start, field[n + 1].end, &id); /* Field n+3 is optional. */ if (id == -1 && fields > n+3) isint_w(field[n + 3].start, field[n + 3].end, &id); tag = 0; s = field[n].start; st = field[n].start + 1; len = field[n].end - field[n].start; switch (*s) { case L'u': if (len == 1 || (len == 4 && wmemcmp(st, L"ser", 3) == 0)) tag = ARCHIVE_ENTRY_ACL_USER_OBJ; break; case L'g': if (len == 1 || (len == 5 && wmemcmp(st, L"roup", 4) == 0)) tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; break; case L'o': if (len == 1 || (len == 5 && wmemcmp(st, L"ther", 4) == 0)) tag = ARCHIVE_ENTRY_ACL_OTHER; break; case L'm': if (len == 1 || (len == 4 && wmemcmp(st, L"ask", 3) == 0)) tag = ARCHIVE_ENTRY_ACL_MASK; break; default: break; } switch (tag) { case ARCHIVE_ENTRY_ACL_OTHER: case ARCHIVE_ENTRY_ACL_MASK: if (fields == (n + 2) && field[n + 1].start < field[n + 1].end && ismode_w(field[n + 1].start, field[n + 1].end, &permset)) { /* This is Solaris-style "other:rwx" */ } else if (fields == (n + 3) && field[n + 1].start < field[n + 1].end) { /* Invalid mask or other field */ ret = ARCHIVE_WARN; continue; } break; case ARCHIVE_ENTRY_ACL_USER_OBJ: case ARCHIVE_ENTRY_ACL_GROUP_OBJ: if (id != -1 || field[n + 1].start < field[n + 1].end) { name = field[n + 1]; if (tag == ARCHIVE_ENTRY_ACL_USER_OBJ) tag = ARCHIVE_ENTRY_ACL_USER; else tag = ARCHIVE_ENTRY_ACL_GROUP; } break; default: /* Invalid tag, skip entry */ ret = ARCHIVE_WARN; continue; } /* Without "default:" we expect mode in field 2 */ if (permset == 0 && !ismode_w(field[n + 2].start, field[n + 2].end, &permset)) { /* Invalid mode, skip entry */ ret = ARCHIVE_WARN; continue; } } else { /* NFS4 ACLs */ s = field[0].start; len = field[0].end - field[0].start; tag = 0; switch (len) { case 4: if (wmemcmp(s, L"user", 4) == 0) tag = ARCHIVE_ENTRY_ACL_USER; break; case 5: if (wmemcmp(s, L"group", 5) == 0) tag = ARCHIVE_ENTRY_ACL_GROUP; break; case 6: if (wmemcmp(s, L"owner@", 6) == 0) tag = ARCHIVE_ENTRY_ACL_USER_OBJ; else if (wmemcmp(s, L"group@", len) == 0) tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; break; case 9: if (wmemcmp(s, L"everyone@", 9) == 0) tag = ARCHIVE_ENTRY_ACL_EVERYONE; default: break; } if (tag == 0) { /* Invalid tag, skip entry */ ret = ARCHIVE_WARN; continue; } else if (tag == ARCHIVE_ENTRY_ACL_USER || tag == ARCHIVE_ENTRY_ACL_GROUP) { n = 1; name = field[1]; + isint_w(name.start, name.end, &id); } else n = 0; if (!is_nfs4_perms_w(field[1 + n].start, field[1 + n].end, &permset)) { /* Invalid NFSv4 perms, skip entry */ ret = ARCHIVE_WARN; continue; } if (!is_nfs4_flags_w(field[2 + n].start, field[2 + n].end, &permset)) { /* Invalid NFSv4 flags, skip entry */ ret = ARCHIVE_WARN; continue; } s = field[3 + n].start; len = field[3 + n].end - field[3 + n].start; type = 0; if (len == 4) { if (wmemcmp(s, L"deny", 4) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_DENY; } else if (len == 5) { if (wmemcmp(s, L"allow", 5) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_ALLOW; else if (wmemcmp(s, L"audit", 5) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_AUDIT; else if (wmemcmp(s, L"alarm", 5) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_ALARM; } if (type == 0) { /* Invalid entry type, skip entry */ ret = ARCHIVE_WARN; continue; } isint_w(field[4 + n].start, field[4 + n].end, &id); } /* Add entry to the internal list. */ r = archive_acl_add_entry_w_len(acl, type, permset, tag, id, name.start, name.end - name.start); if (r < ARCHIVE_WARN) return (r); if (r != ARCHIVE_OK) ret = ARCHIVE_WARN; types |= type; } /* Reset ACL */ archive_acl_reset(acl, types); return (ret); } /* * Parse a string to a positive decimal integer. Returns true if * the string is non-empty and consists only of decimal digits, * false otherwise. */ static int isint_w(const wchar_t *start, const wchar_t *end, int *result) { int n = 0; if (start >= end) return (0); while (start < end) { if (*start < '0' || *start > '9') return (0); if (n > (INT_MAX / 10) || (n == INT_MAX / 10 && (*start - '0') > INT_MAX % 10)) { n = INT_MAX; } else { n *= 10; n += *start - '0'; } start++; } *result = n; return (1); } /* * Parse a string as a mode field. Returns true if * the string is non-empty and consists only of mode characters, * false otherwise. */ static int ismode_w(const wchar_t *start, const wchar_t *end, int *permset) { const wchar_t *p; if (start >= end) return (0); p = start; *permset = 0; while (p < end) { switch (*p++) { case L'r': case L'R': *permset |= ARCHIVE_ENTRY_ACL_READ; break; case L'w': case L'W': *permset |= ARCHIVE_ENTRY_ACL_WRITE; break; case L'x': case L'X': *permset |= ARCHIVE_ENTRY_ACL_EXECUTE; break; case L'-': break; default: return (0); } } return (1); } /* * Parse a string as a NFS4 ACL permission field. * Returns true if the string is non-empty and consists only of NFS4 ACL * permission characters, false otherwise */ static int is_nfs4_perms_w(const wchar_t *start, const wchar_t *end, int *permset) { const wchar_t *p; if (start >= end) return (0); p = start; while (p < end) { switch (*p++) { case L'r': *permset |= ARCHIVE_ENTRY_ACL_READ_DATA; break; case L'w': *permset |= ARCHIVE_ENTRY_ACL_WRITE_DATA; break; case L'x': *permset |= ARCHIVE_ENTRY_ACL_EXECUTE; break; case L'p': *permset |= ARCHIVE_ENTRY_ACL_APPEND_DATA; break; case L'D': *permset |= ARCHIVE_ENTRY_ACL_DELETE_CHILD; break; case L'd': *permset |= ARCHIVE_ENTRY_ACL_DELETE; break; case L'a': *permset |= ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES; break; case L'A': *permset |= ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES; break; case L'R': *permset |= ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS; break; case L'W': *permset |= ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS; break; case L'c': *permset |= ARCHIVE_ENTRY_ACL_READ_ACL; break; case L'C': *permset |= ARCHIVE_ENTRY_ACL_WRITE_ACL; break; case L'o': *permset |= ARCHIVE_ENTRY_ACL_WRITE_OWNER; break; case L's': *permset |= ARCHIVE_ENTRY_ACL_SYNCHRONIZE; break; case L'-': break; default: return(0); } } return (1); } /* * Parse a string as a NFS4 ACL flags field. * Returns true if the string is non-empty and consists only of NFS4 ACL * flag characters, false otherwise */ static int is_nfs4_flags_w(const wchar_t *start, const wchar_t *end, int *permset) { const wchar_t *p; if (start >= end) return (0); p = start; while (p < end) { switch(*p++) { case L'f': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT; break; case L'd': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT; break; case L'i': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY; break; case L'n': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT; break; case L'S': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS; break; case L'F': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS; break; case L'I': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_INHERITED; break; case L'-': break; default: return (0); } } return (1); } /* * Match "[:whitespace:]*(.*)[:whitespace:]*[:,\n]". *wp is updated * to point to just after the separator. *start points to the first * character of the matched text and *end just after the last * character of the matched identifier. In particular *end - *start * is the length of the field body, not including leading or trailing * whitespace. */ static void next_field_w(const wchar_t **wp, const wchar_t **start, const wchar_t **end, wchar_t *sep) { /* Skip leading whitespace to find start of field. */ while (**wp == L' ' || **wp == L'\t' || **wp == L'\n') { (*wp)++; } *start = *wp; /* Scan for the separator. */ while (**wp != L'\0' && **wp != L',' && **wp != L':' && **wp != L'\n') { (*wp)++; } *sep = **wp; /* Trim trailing whitespace to locate end of field. */ *end = *wp - 1; while (**end == L' ' || **end == L'\t' || **end == L'\n') { (*end)--; } (*end)++; /* Adjust scanner location. */ if (**wp != L'\0') (*wp)++; } /* * Parse an ACL text string. * * The want_type argument may be one of the following: * ARCHIVE_ENTRY_ACL_TYPE_ACCESS - text is a POSIX.1e ACL of type ACCESS * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT - text is a POSIX.1e ACL of type DEFAULT * ARCHIVE_ENTRY_ACL_TYPE_NFS4 - text is as a NFSv4 ACL * * POSIX.1e ACL entries prefixed with "default:" are treated as * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT unless type is ARCHIVE_ENTRY_ACL_TYPE_NFS4 */ int archive_acl_from_text_l(struct archive_acl *acl, const char *text, int want_type, struct archive_string_conv *sc) { struct { const char *start; const char *end; } field[6], name; const char *s, *st; int numfields, fields, n, r, ret; int type, types, tag, permset, id; size_t len; char sep; switch (want_type) { case ARCHIVE_ENTRY_ACL_TYPE_POSIX1E: want_type = ARCHIVE_ENTRY_ACL_TYPE_ACCESS; case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: numfields = 5; break; case ARCHIVE_ENTRY_ACL_TYPE_NFS4: numfields = 6; break; default: return (ARCHIVE_FATAL); } ret = ARCHIVE_OK; types = 0; while (text != NULL && *text != '\0') { /* * Parse the fields out of the next entry, * advance 'text' to start of next entry. */ fields = 0; do { const char *start, *end; next_field(&text, &start, &end, &sep); if (fields < numfields) { field[fields].start = start; field[fields].end = end; } ++fields; } while (sep == ':'); /* Set remaining fields to blank. */ for (n = fields; n < numfields; ++n) field[n].start = field[n].end = NULL; if (field[0].start != NULL && *(field[0].start) == '#') { /* Comment, skip entry */ continue; } n = 0; id = -1; permset = 0; name.start = name.end = NULL; if (want_type != ARCHIVE_ENTRY_ACL_TYPE_NFS4) { /* POSIX.1e ACLs */ /* * Default keyword "default:user::rwx" * if found, we have one more field * * We also support old Solaris extension: * "defaultuser::rwx" is the default ACL corresponding * to "user::rwx", etc. valid only for first field */ s = field[0].start; len = field[0].end - field[0].start; if (*s == 'd' && (len == 1 || (len >= 7 && memcmp((s + 1), "efault", 6) == 0))) { type = ARCHIVE_ENTRY_ACL_TYPE_DEFAULT; if (len > 7) field[0].start += 7; else n = 1; } else type = want_type; /* Check for a numeric ID in field n+1 or n+3. */ isint(field[n + 1].start, field[n + 1].end, &id); /* Field n+3 is optional. */ if (id == -1 && fields > (n + 3)) isint(field[n + 3].start, field[n + 3].end, &id); tag = 0; s = field[n].start; st = field[n].start + 1; len = field[n].end - field[n].start; switch (*s) { case 'u': if (len == 1 || (len == 4 && memcmp(st, "ser", 3) == 0)) tag = ARCHIVE_ENTRY_ACL_USER_OBJ; break; case 'g': if (len == 1 || (len == 5 && memcmp(st, "roup", 4) == 0)) tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; break; case 'o': if (len == 1 || (len == 5 && memcmp(st, "ther", 4) == 0)) tag = ARCHIVE_ENTRY_ACL_OTHER; break; case 'm': if (len == 1 || (len == 4 && memcmp(st, "ask", 3) == 0)) tag = ARCHIVE_ENTRY_ACL_MASK; break; default: break; } switch (tag) { case ARCHIVE_ENTRY_ACL_OTHER: case ARCHIVE_ENTRY_ACL_MASK: if (fields == (n + 2) && field[n + 1].start < field[n + 1].end && ismode(field[n + 1].start, field[n + 1].end, &permset)) { /* This is Solaris-style "other:rwx" */ } else if (fields == (n + 3) && field[n + 1].start < field[n + 1].end) { /* Invalid mask or other field */ ret = ARCHIVE_WARN; continue; } break; case ARCHIVE_ENTRY_ACL_USER_OBJ: case ARCHIVE_ENTRY_ACL_GROUP_OBJ: if (id != -1 || field[n + 1].start < field[n + 1].end) { name = field[n + 1]; if (tag == ARCHIVE_ENTRY_ACL_USER_OBJ) tag = ARCHIVE_ENTRY_ACL_USER; else tag = ARCHIVE_ENTRY_ACL_GROUP; } break; default: /* Invalid tag, skip entry */ ret = ARCHIVE_WARN; continue; } /* Without "default:" we expect mode in field 2 */ if (permset == 0 && !ismode(field[n + 2].start, field[n + 2].end, &permset)) { /* Invalid mode, skip entry */ ret = ARCHIVE_WARN; continue; } } else { /* NFS4 ACLs */ s = field[0].start; len = field[0].end - field[0].start; tag = 0; switch (len) { case 4: if (memcmp(s, "user", 4) == 0) tag = ARCHIVE_ENTRY_ACL_USER; break; case 5: if (memcmp(s, "group", 5) == 0) tag = ARCHIVE_ENTRY_ACL_GROUP; break; case 6: if (memcmp(s, "owner@", 6) == 0) tag = ARCHIVE_ENTRY_ACL_USER_OBJ; else if (memcmp(s, "group@", 6) == 0) tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; break; case 9: if (memcmp(s, "everyone@", 9) == 0) tag = ARCHIVE_ENTRY_ACL_EVERYONE; break; default: break; } if (tag == 0) { /* Invalid tag, skip entry */ ret = ARCHIVE_WARN; continue; } else if (tag == ARCHIVE_ENTRY_ACL_USER || tag == ARCHIVE_ENTRY_ACL_GROUP) { n = 1; name = field[1]; + isint(name.start, name.end, &id); } else n = 0; if (!is_nfs4_perms(field[1 + n].start, field[1 + n].end, &permset)) { /* Invalid NFSv4 perms, skip entry */ ret = ARCHIVE_WARN; continue; } if (!is_nfs4_flags(field[2 + n].start, field[2 + n].end, &permset)) { /* Invalid NFSv4 flags, skip entry */ ret = ARCHIVE_WARN; continue; } s = field[3 + n].start; len = field[3 + n].end - field[3 + n].start; type = 0; if (len == 4) { if (memcmp(s, "deny", 4) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_DENY; } else if (len == 5) { if (memcmp(s, "allow", 5) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_ALLOW; else if (memcmp(s, "audit", 5) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_AUDIT; else if (memcmp(s, "alarm", 5) == 0) type = ARCHIVE_ENTRY_ACL_TYPE_ALARM; } if (type == 0) { /* Invalid entry type, skip entry */ ret = ARCHIVE_WARN; continue; } isint(field[4 + n].start, field[4 + n].end, &id); } /* Add entry to the internal list. */ r = archive_acl_add_entry_len_l(acl, type, permset, tag, id, name.start, name.end - name.start, sc); if (r < ARCHIVE_WARN) return (r); if (r != ARCHIVE_OK) ret = ARCHIVE_WARN; types |= type; } /* Reset ACL */ archive_acl_reset(acl, types); return (ret); } /* * Parse a string to a positive decimal integer. Returns true if * the string is non-empty and consists only of decimal digits, * false otherwise. */ static int isint(const char *start, const char *end, int *result) { int n = 0; if (start >= end) return (0); while (start < end) { if (*start < '0' || *start > '9') return (0); if (n > (INT_MAX / 10) || (n == INT_MAX / 10 && (*start - '0') > INT_MAX % 10)) { n = INT_MAX; } else { n *= 10; n += *start - '0'; } start++; } *result = n; return (1); } /* * Parse a string as a mode field. Returns true if * the string is non-empty and consists only of mode characters, * false otherwise. */ static int ismode(const char *start, const char *end, int *permset) { const char *p; if (start >= end) return (0); p = start; *permset = 0; while (p < end) { switch (*p++) { case 'r': case 'R': *permset |= ARCHIVE_ENTRY_ACL_READ; break; case 'w': case 'W': *permset |= ARCHIVE_ENTRY_ACL_WRITE; break; case 'x': case 'X': *permset |= ARCHIVE_ENTRY_ACL_EXECUTE; break; case '-': break; default: return (0); } } return (1); } /* * Parse a string as a NFS4 ACL permission field. * Returns true if the string is non-empty and consists only of NFS4 ACL * permission characters, false otherwise */ static int is_nfs4_perms(const char *start, const char *end, int *permset) { const char *p; if (start >= end) return (0); p = start; while (p < end) { switch (*p++) { case 'r': *permset |= ARCHIVE_ENTRY_ACL_READ_DATA; break; case 'w': *permset |= ARCHIVE_ENTRY_ACL_WRITE_DATA; break; case 'x': *permset |= ARCHIVE_ENTRY_ACL_EXECUTE; break; case 'p': *permset |= ARCHIVE_ENTRY_ACL_APPEND_DATA; break; case 'D': *permset |= ARCHIVE_ENTRY_ACL_DELETE_CHILD; break; case 'd': *permset |= ARCHIVE_ENTRY_ACL_DELETE; break; case 'a': *permset |= ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES; break; case 'A': *permset |= ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES; break; case 'R': *permset |= ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS; break; case 'W': *permset |= ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS; break; case 'c': *permset |= ARCHIVE_ENTRY_ACL_READ_ACL; break; case 'C': *permset |= ARCHIVE_ENTRY_ACL_WRITE_ACL; break; case 'o': *permset |= ARCHIVE_ENTRY_ACL_WRITE_OWNER; break; case 's': *permset |= ARCHIVE_ENTRY_ACL_SYNCHRONIZE; break; case '-': break; default: return(0); } } return (1); } /* * Parse a string as a NFS4 ACL flags field. * Returns true if the string is non-empty and consists only of NFS4 ACL * flag characters, false otherwise */ static int is_nfs4_flags(const char *start, const char *end, int *permset) { const char *p; if (start >= end) return (0); p = start; while (p < end) { switch(*p++) { case 'f': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT; break; case 'd': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT; break; case 'i': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY; break; case 'n': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT; break; case 'S': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS; break; case 'F': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS; break; case 'I': *permset |= ARCHIVE_ENTRY_ACL_ENTRY_INHERITED; break; case '-': break; default: return (0); } } return (1); } /* * Match "[:whitespace:]*(.*)[:whitespace:]*[:,\n]". *wp is updated * to point to just after the separator. *start points to the first * character of the matched text and *end just after the last * character of the matched identifier. In particular *end - *start * is the length of the field body, not including leading or trailing * whitespace. */ static void next_field(const char **p, const char **start, const char **end, char *sep) { /* Skip leading whitespace to find start of field. */ while (**p == ' ' || **p == '\t' || **p == '\n') { (*p)++; } *start = *p; /* Scan for the separator. */ while (**p != '\0' && **p != ',' && **p != ':' && **p != '\n') { (*p)++; } *sep = **p; /* Trim trailing whitespace to locate end of field. */ *end = *p - 1; while (**end == ' ' || **end == '\t' || **end == '\n') { (*end)--; } (*end)++; /* Adjust scanner location. */ if (**p != '\0') (*p)++; } Index: head/contrib/libarchive/libarchive/archive_read_disk_entry_from_file.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_disk_entry_from_file.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_read_disk_entry_from_file.c (revision 311903) @@ -1,1394 +1,1402 @@ /*- * Copyright (c) 2003-2009 Tim Kientzle * Copyright (c) 2010-2012 Michihiro NAKAJIMA * Copyright (c) 2016 Martin Matuska * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); /* This is the tree-walking code for POSIX systems. */ #if !defined(_WIN32) || defined(__CYGWIN__) #ifdef HAVE_SYS_TYPES_H /* Mac OSX requires sys/types.h before sys/acl.h. */ #include #endif #ifdef HAVE_SYS_ACL_H #include #endif #ifdef HAVE_SYS_EXTATTR_H #include #endif #ifdef HAVE_SYS_IOCTL_H #include #endif #ifdef HAVE_SYS_PARAM_H #include #endif #ifdef HAVE_SYS_STAT_H #include #endif #if defined(HAVE_SYS_XATTR_H) #include #elif defined(HAVE_ATTR_XATTR_H) #include #endif #ifdef HAVE_SYS_EA_H #include #endif #ifdef HAVE_ACL_LIBACL_H #include #endif #ifdef HAVE_COPYFILE_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_LINUX_TYPES_H #include #endif #ifdef HAVE_LINUX_FIEMAP_H #include #endif #ifdef HAVE_LINUX_FS_H #include #endif /* * Some Linux distributions have both linux/ext2_fs.h and ext2fs/ext2_fs.h. * As the include guards don't agree, the order of include is important. */ #ifdef HAVE_LINUX_EXT2_FS_H #include /* for Linux file flags */ #endif #if defined(HAVE_EXT2FS_EXT2_FS_H) && !defined(__CYGWIN__) #include /* Linux file flags, broken on Cygwin */ #endif #ifdef HAVE_PATHS_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_private.h" #include "archive_read_disk_private.h" #ifndef O_CLOEXEC #define O_CLOEXEC 0 #endif /* * Linux and FreeBSD plug this obvious hole in POSIX.1e in * different ways. */ #if HAVE_ACL_GET_PERM #define ACL_GET_PERM acl_get_perm #elif HAVE_ACL_GET_PERM_NP #define ACL_GET_PERM acl_get_perm_np #endif static int setup_acls(struct archive_read_disk *, struct archive_entry *, int *fd); static int setup_mac_metadata(struct archive_read_disk *, struct archive_entry *, int *fd); static int setup_xattrs(struct archive_read_disk *, struct archive_entry *, int *fd); static int setup_sparse(struct archive_read_disk *, struct archive_entry *, int *fd); #if defined(HAVE_LINUX_FIEMAP_H) static int setup_sparse_fiemap(struct archive_read_disk *, struct archive_entry *, int *fd); #endif int archive_read_disk_entry_from_file(struct archive *_a, struct archive_entry *entry, int fd, const struct stat *st) { struct archive_read_disk *a = (struct archive_read_disk *)_a; const char *path, *name; struct stat s; int initial_fd = fd; int r, r1; archive_clear_error(_a); path = archive_entry_sourcepath(entry); if (path == NULL) path = archive_entry_pathname(entry); if (a->tree == NULL) { if (st == NULL) { #if HAVE_FSTAT if (fd >= 0) { if (fstat(fd, &s) != 0) { archive_set_error(&a->archive, errno, "Can't fstat"); return (ARCHIVE_FAILED); } } else #endif #if HAVE_LSTAT if (!a->follow_symlinks) { if (lstat(path, &s) != 0) { archive_set_error(&a->archive, errno, "Can't lstat %s", path); return (ARCHIVE_FAILED); } } else #endif if (stat(path, &s) != 0) { archive_set_error(&a->archive, errno, "Can't stat %s", path); return (ARCHIVE_FAILED); } st = &s; } archive_entry_copy_stat(entry, st); } /* Lookup uname/gname */ name = archive_read_disk_uname(_a, archive_entry_uid(entry)); if (name != NULL) archive_entry_copy_uname(entry, name); name = archive_read_disk_gname(_a, archive_entry_gid(entry)); if (name != NULL) archive_entry_copy_gname(entry, name); #ifdef HAVE_STRUCT_STAT_ST_FLAGS /* On FreeBSD, we get flags for free with the stat. */ /* TODO: Does this belong in copy_stat()? */ if (st->st_flags != 0) archive_entry_set_fflags(entry, st->st_flags, 0); #endif #if defined(EXT2_IOC_GETFLAGS) && defined(HAVE_WORKING_EXT2_IOC_GETFLAGS) /* Linux requires an extra ioctl to pull the flags. Although * this is an extra step, it has a nice side-effect: We get an * open file descriptor which we can use in the subsequent lookups. */ if ((S_ISREG(st->st_mode) || S_ISDIR(st->st_mode))) { if (fd < 0) { if (a->tree != NULL) fd = a->open_on_current_dir(a->tree, path, O_RDONLY | O_NONBLOCK | O_CLOEXEC); else fd = open(path, O_RDONLY | O_NONBLOCK | O_CLOEXEC); __archive_ensure_cloexec_flag(fd); } if (fd >= 0) { int stflags; r = ioctl(fd, EXT2_IOC_GETFLAGS, &stflags); if (r == 0 && stflags != 0) archive_entry_set_fflags(entry, stflags, 0); } } #endif #if defined(HAVE_READLINK) || defined(HAVE_READLINKAT) if (S_ISLNK(st->st_mode)) { size_t linkbuffer_len = st->st_size + 1; char *linkbuffer; int lnklen; linkbuffer = malloc(linkbuffer_len); if (linkbuffer == NULL) { archive_set_error(&a->archive, ENOMEM, "Couldn't read link data"); return (ARCHIVE_FAILED); } if (a->tree != NULL) { #ifdef HAVE_READLINKAT lnklen = readlinkat(a->tree_current_dir_fd(a->tree), path, linkbuffer, linkbuffer_len); #else if (a->tree_enter_working_dir(a->tree) != 0) { archive_set_error(&a->archive, errno, "Couldn't read link data"); free(linkbuffer); return (ARCHIVE_FAILED); } lnklen = readlink(path, linkbuffer, linkbuffer_len); #endif /* HAVE_READLINKAT */ } else lnklen = readlink(path, linkbuffer, linkbuffer_len); if (lnklen < 0) { archive_set_error(&a->archive, errno, "Couldn't read link data"); free(linkbuffer); return (ARCHIVE_FAILED); } linkbuffer[lnklen] = 0; archive_entry_set_symlink(entry, linkbuffer); free(linkbuffer); } #endif /* HAVE_READLINK || HAVE_READLINKAT */ r = setup_acls(a, entry, &fd); if (!a->suppress_xattr) { r1 = setup_xattrs(a, entry, &fd); if (r1 < r) r = r1; } if (a->enable_copyfile) { r1 = setup_mac_metadata(a, entry, &fd); if (r1 < r) r = r1; } r1 = setup_sparse(a, entry, &fd); if (r1 < r) r = r1; /* If we opened the file earlier in this function, close it. */ if (initial_fd != fd) close(fd); return (r); } #if defined(__APPLE__) && defined(HAVE_COPYFILE_H) /* * The Mac OS "copyfile()" API copies the extended metadata for a * file into a separate file in AppleDouble format (see RFC 1740). * * Mac OS tar and cpio implementations store this extended * metadata as a separate entry just before the regular entry * with a "._" prefix added to the filename. * * Note that this is currently done unconditionally; the tar program has * an option to discard this information before the archive is written. * * TODO: If there's a failure, report it and return ARCHIVE_WARN. */ static int setup_mac_metadata(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { int tempfd = -1; int copyfile_flags = COPYFILE_NOFOLLOW | COPYFILE_ACL | COPYFILE_XATTR; struct stat copyfile_stat; int ret = ARCHIVE_OK; void *buff = NULL; int have_attrs; const char *name, *tempdir; struct archive_string tempfile; (void)fd; /* UNUSED */ name = archive_entry_sourcepath(entry); if (name == NULL) name = archive_entry_pathname(entry); if (name == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't open file to read extended attributes: No name"); return (ARCHIVE_WARN); } if (a->tree != NULL) { if (a->tree_enter_working_dir(a->tree) != 0) { archive_set_error(&a->archive, errno, "Couldn't change dir"); return (ARCHIVE_FAILED); } } /* Short-circuit if there's nothing to do. */ have_attrs = copyfile(name, NULL, 0, copyfile_flags | COPYFILE_CHECK); if (have_attrs == -1) { archive_set_error(&a->archive, errno, "Could not check extended attributes"); return (ARCHIVE_WARN); } if (have_attrs == 0) return (ARCHIVE_OK); tempdir = NULL; if (issetugid() == 0) tempdir = getenv("TMPDIR"); if (tempdir == NULL) tempdir = _PATH_TMP; archive_string_init(&tempfile); archive_strcpy(&tempfile, tempdir); archive_strcat(&tempfile, "tar.md.XXXXXX"); tempfd = mkstemp(tempfile.s); if (tempfd < 0) { archive_set_error(&a->archive, errno, "Could not open extended attribute file"); ret = ARCHIVE_WARN; goto cleanup; } __archive_ensure_cloexec_flag(tempfd); /* XXX I wish copyfile() could pack directly to a memory * buffer; that would avoid the temp file here. For that * matter, it would be nice if fcopyfile() actually worked, * that would reduce the many open/close races here. */ if (copyfile(name, tempfile.s, 0, copyfile_flags | COPYFILE_PACK)) { archive_set_error(&a->archive, errno, "Could not pack extended attributes"); ret = ARCHIVE_WARN; goto cleanup; } if (fstat(tempfd, ©file_stat)) { archive_set_error(&a->archive, errno, "Could not check size of extended attributes"); ret = ARCHIVE_WARN; goto cleanup; } buff = malloc(copyfile_stat.st_size); if (buff == NULL) { archive_set_error(&a->archive, errno, "Could not allocate memory for extended attributes"); ret = ARCHIVE_WARN; goto cleanup; } if (copyfile_stat.st_size != read(tempfd, buff, copyfile_stat.st_size)) { archive_set_error(&a->archive, errno, "Could not read extended attributes into memory"); ret = ARCHIVE_WARN; goto cleanup; } archive_entry_copy_mac_metadata(entry, buff, copyfile_stat.st_size); cleanup: if (tempfd >= 0) { close(tempfd); unlink(tempfile.s); } archive_string_free(&tempfile); free(buff); return (ret); } #else /* * Stub implementation for non-Mac systems. */ static int setup_mac_metadata(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { (void)a; /* UNUSED */ (void)entry; /* UNUSED */ (void)fd; /* UNUSED */ return (ARCHIVE_OK); } #endif #ifdef HAVE_POSIX_ACL static int translate_acl(struct archive_read_disk *a, struct archive_entry *entry, acl_t acl, int archive_entry_acl_type); static int setup_acls(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { const char *accpath; acl_t acl; int r; accpath = archive_entry_sourcepath(entry); if (accpath == NULL) accpath = archive_entry_pathname(entry); if (*fd < 0 && a->tree != NULL) { if (a->follow_symlinks || archive_entry_filetype(entry) != AE_IFLNK) *fd = a->open_on_current_dir(a->tree, accpath, O_RDONLY | O_NONBLOCK); if (*fd < 0) { if (a->tree_enter_working_dir(a->tree) != 0) { archive_set_error(&a->archive, errno, "Couldn't access %s", accpath); return (ARCHIVE_FAILED); } } } archive_entry_acl_clear(entry); acl = NULL; #ifdef ACL_TYPE_NFS4 /* Try NFSv4 ACL first. */ if (*fd >= 0) #if HAVE_ACL_GET_FD_NP acl = acl_get_fd_np(*fd, ACL_TYPE_NFS4); #else acl = acl_get_fd(*fd); #endif #if HAVE_ACL_GET_LINK_NP else if (!a->follow_symlinks) acl = acl_get_link_np(accpath, ACL_TYPE_NFS4); #else else if ((!a->follow_symlinks) && (archive_entry_filetype(entry) == AE_IFLNK)) /* We can't get the ACL of a symlink, so we assume it can't have one. */ acl = NULL; #endif else acl = acl_get_file(accpath, ACL_TYPE_NFS4); #if HAVE_ACL_IS_TRIVIAL_NP if (acl != NULL && acl_is_trivial_np(acl, &r) == 0) { /* Ignore "trivial" ACLs that just mirror the file mode. */ if (r) { acl_free(acl); acl = NULL; /* * Simultaneous NFSv4 and POSIX.1e ACLs for the same * entry are not allowed, so we should return here */ return (ARCHIVE_OK); } } #endif if (acl != NULL) { r = translate_acl(a, entry, acl, ARCHIVE_ENTRY_ACL_TYPE_NFS4); acl_free(acl); if (r != ARCHIVE_OK) { archive_set_error(&a->archive, errno, "Couldn't translate NFSv4 ACLs: %s", accpath); } return (r); } #endif /* ACL_TYPE_NFS4 */ /* Retrieve access ACL from file. */ if (*fd >= 0) acl = acl_get_fd(*fd); #if HAVE_ACL_GET_LINK_NP else if (!a->follow_symlinks) acl = acl_get_link_np(accpath, ACL_TYPE_ACCESS); #else else if ((!a->follow_symlinks) && (archive_entry_filetype(entry) == AE_IFLNK)) /* We can't get the ACL of a symlink, so we assume it can't have one. */ acl = NULL; #endif else acl = acl_get_file(accpath, ACL_TYPE_ACCESS); #if HAVE_ACL_IS_TRIVIAL_NP /* Ignore "trivial" ACLs that just mirror the file mode. */ if (acl != NULL && acl_is_trivial_np(acl, &r) == 0) { if (r) { acl_free(acl); acl = NULL; } } #endif if (acl != NULL) { r = translate_acl(a, entry, acl, ARCHIVE_ENTRY_ACL_TYPE_ACCESS); acl_free(acl); acl = NULL; if (r != ARCHIVE_OK) { archive_set_error(&a->archive, errno, "Couldn't translate access ACLs: %s", accpath); return (r); } } /* Only directories can have default ACLs. */ if (S_ISDIR(archive_entry_mode(entry))) { +#if HAVE_ACL_GET_FD_NP + if (*fd >= 0) + acl = acl_get_fd_np(*fd, ACL_TYPE_DEFAULT); + else +#endif acl = acl_get_file(accpath, ACL_TYPE_DEFAULT); if (acl != NULL) { r = translate_acl(a, entry, acl, ARCHIVE_ENTRY_ACL_TYPE_DEFAULT); acl_free(acl); if (r != ARCHIVE_OK) { archive_set_error(&a->archive, errno, "Couldn't translate default ACLs: %s", accpath); return (r); } } } return (ARCHIVE_OK); } /* * Translate system ACL into libarchive internal structure. */ static struct { int archive_perm; int platform_perm; } acl_perm_map[] = { {ARCHIVE_ENTRY_ACL_EXECUTE, ACL_EXECUTE}, {ARCHIVE_ENTRY_ACL_WRITE, ACL_WRITE}, {ARCHIVE_ENTRY_ACL_READ, ACL_READ}, #ifdef ACL_TYPE_NFS4 {ARCHIVE_ENTRY_ACL_READ_DATA, ACL_READ_DATA}, {ARCHIVE_ENTRY_ACL_LIST_DIRECTORY, ACL_LIST_DIRECTORY}, {ARCHIVE_ENTRY_ACL_WRITE_DATA, ACL_WRITE_DATA}, {ARCHIVE_ENTRY_ACL_ADD_FILE, ACL_ADD_FILE}, {ARCHIVE_ENTRY_ACL_APPEND_DATA, ACL_APPEND_DATA}, {ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY, ACL_ADD_SUBDIRECTORY}, {ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS, ACL_READ_NAMED_ATTRS}, {ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS, ACL_WRITE_NAMED_ATTRS}, {ARCHIVE_ENTRY_ACL_DELETE_CHILD, ACL_DELETE_CHILD}, {ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES, ACL_READ_ATTRIBUTES}, {ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES, ACL_WRITE_ATTRIBUTES}, {ARCHIVE_ENTRY_ACL_DELETE, ACL_DELETE}, {ARCHIVE_ENTRY_ACL_READ_ACL, ACL_READ_ACL}, {ARCHIVE_ENTRY_ACL_WRITE_ACL, ACL_WRITE_ACL}, {ARCHIVE_ENTRY_ACL_WRITE_OWNER, ACL_WRITE_OWNER}, {ARCHIVE_ENTRY_ACL_SYNCHRONIZE, ACL_SYNCHRONIZE} #endif }; #ifdef ACL_TYPE_NFS4 static struct { int archive_inherit; int platform_inherit; } acl_inherit_map[] = { {ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT, ACL_ENTRY_FILE_INHERIT}, {ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT, ACL_ENTRY_DIRECTORY_INHERIT}, {ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT, ACL_ENTRY_NO_PROPAGATE_INHERIT}, - {ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, ACL_ENTRY_INHERIT_ONLY} + {ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, ACL_ENTRY_INHERIT_ONLY}, + {ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS, ACL_ENTRY_SUCCESSFUL_ACCESS}, + {ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS, ACL_ENTRY_FAILED_ACCESS}, + {ARCHIVE_ENTRY_ACL_ENTRY_INHERITED, ACL_ENTRY_INHERITED} }; #endif static int translate_acl(struct archive_read_disk *a, struct archive_entry *entry, acl_t acl, int default_entry_acl_type) { acl_tag_t acl_tag; #ifdef ACL_TYPE_NFS4 acl_entry_type_t acl_type; acl_flagset_t acl_flagset; int brand; #endif acl_entry_t acl_entry; acl_permset_t acl_permset; int i, entry_acl_type; int r, s, ae_id, ae_tag, ae_perm; const char *ae_name; #ifdef ACL_TYPE_NFS4 // FreeBSD "brands" ACLs as POSIX.1e or NFSv4 // Make sure the "brand" on this ACL is consistent // with the default_entry_acl_type bits provided. if (acl_get_brand_np(acl, &brand) != 0) { archive_set_error(&a->archive, errno, "Failed to read ACL brand"); return (ARCHIVE_WARN); } switch (brand) { case ACL_BRAND_POSIX: switch (default_entry_acl_type) { case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid ACL entry type for POSIX.1e ACL"); return (ARCHIVE_WARN); } break; case ACL_BRAND_NFS4: if (default_entry_acl_type & ~ARCHIVE_ENTRY_ACL_TYPE_NFS4) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid ACL entry type for NFSv4 ACL"); return (ARCHIVE_WARN); } break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unknown ACL brand"); return (ARCHIVE_WARN); } #endif s = acl_get_entry(acl, ACL_FIRST_ENTRY, &acl_entry); if (s == -1) { archive_set_error(&a->archive, errno, "Failed to get first ACL entry"); return (ARCHIVE_WARN); } while (s == 1) { ae_id = -1; ae_name = NULL; ae_perm = 0; if (acl_get_tag_type(acl_entry, &acl_tag) != 0) { archive_set_error(&a->archive, errno, "Failed to get ACL tag type"); return (ARCHIVE_WARN); } switch (acl_tag) { case ACL_USER: ae_id = (int)*(uid_t *)acl_get_qualifier(acl_entry); ae_name = archive_read_disk_uname(&a->archive, ae_id); ae_tag = ARCHIVE_ENTRY_ACL_USER; break; case ACL_GROUP: ae_id = (int)*(gid_t *)acl_get_qualifier(acl_entry); ae_name = archive_read_disk_gname(&a->archive, ae_id); ae_tag = ARCHIVE_ENTRY_ACL_GROUP; break; case ACL_MASK: ae_tag = ARCHIVE_ENTRY_ACL_MASK; break; case ACL_USER_OBJ: ae_tag = ARCHIVE_ENTRY_ACL_USER_OBJ; break; case ACL_GROUP_OBJ: ae_tag = ARCHIVE_ENTRY_ACL_GROUP_OBJ; break; case ACL_OTHER: ae_tag = ARCHIVE_ENTRY_ACL_OTHER; break; #ifdef ACL_TYPE_NFS4 case ACL_EVERYONE: ae_tag = ARCHIVE_ENTRY_ACL_EVERYONE; break; #endif default: /* Skip types that libarchive can't support. */ s = acl_get_entry(acl, ACL_NEXT_ENTRY, &acl_entry); continue; } // XXX acl_type maps to allow/deny/audit/YYYY bits entry_acl_type = default_entry_acl_type; #ifdef ACL_TYPE_NFS4 if (default_entry_acl_type & ARCHIVE_ENTRY_ACL_TYPE_NFS4) { /* * acl_get_entry_type_np() fails with non-NFSv4 ACLs */ if (acl_get_entry_type_np(acl_entry, &acl_type) != 0) { archive_set_error(&a->archive, errno, "Failed " "to get ACL type from a NFSv4 ACL entry"); return (ARCHIVE_WARN); } switch (acl_type) { case ACL_ENTRY_TYPE_ALLOW: entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_ALLOW; break; case ACL_ENTRY_TYPE_DENY: entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_DENY; break; case ACL_ENTRY_TYPE_AUDIT: entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_AUDIT; break; case ACL_ENTRY_TYPE_ALARM: entry_acl_type = ARCHIVE_ENTRY_ACL_TYPE_ALARM; break; default: archive_set_error(&a->archive, errno, "Invalid NFSv4 ACL entry type"); return (ARCHIVE_WARN); } /* * Libarchive stores "flag" (NFSv4 inheritance bits) * in the ae_perm bitmap. * * acl_get_flagset_np() fails with non-NFSv4 ACLs */ if (acl_get_flagset_np(acl_entry, &acl_flagset) != 0) { archive_set_error(&a->archive, errno, "Failed to get flagset from a NFSv4 ACL entry"); return (ARCHIVE_WARN); } for (i = 0; i < (int)(sizeof(acl_inherit_map) / sizeof(acl_inherit_map[0])); ++i) { r = acl_get_flag_np(acl_flagset, acl_inherit_map[i].platform_inherit); if (r == -1) { archive_set_error(&a->archive, errno, "Failed to check flag in a NFSv4 " "ACL flagset"); return (ARCHIVE_WARN); } else if (r) ae_perm |= acl_inherit_map[i].archive_inherit; } } #endif if (acl_get_permset(acl_entry, &acl_permset) != 0) { archive_set_error(&a->archive, errno, "Failed to get ACL permission set"); return (ARCHIVE_WARN); } for (i = 0; i < (int)(sizeof(acl_perm_map) / sizeof(acl_perm_map[0])); ++i) { /* * acl_get_perm() is spelled differently on different * platforms; see above. */ r = ACL_GET_PERM(acl_permset, acl_perm_map[i].platform_perm); if (r == -1) { archive_set_error(&a->archive, errno, "Failed to check permission in an ACL permission set"); return (ARCHIVE_WARN); } else if (r) ae_perm |= acl_perm_map[i].archive_perm; } archive_entry_acl_add_entry(entry, entry_acl_type, ae_perm, ae_tag, ae_id, ae_name); s = acl_get_entry(acl, ACL_NEXT_ENTRY, &acl_entry); if (s == -1) { archive_set_error(&a->archive, errno, "Failed to get next ACL entry"); return (ARCHIVE_WARN); } } return (ARCHIVE_OK); } #else static int setup_acls(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { (void)a; /* UNUSED */ (void)entry; /* UNUSED */ (void)fd; /* UNUSED */ return (ARCHIVE_OK); } #endif #if (HAVE_FGETXATTR && HAVE_FLISTXATTR && HAVE_LISTXATTR && \ HAVE_LLISTXATTR && HAVE_GETXATTR && HAVE_LGETXATTR) || \ (HAVE_FGETEA && HAVE_FLISTEA && HAVE_LISTEA) /* * Linux and AIX extended attribute support. * * TODO: By using a stack-allocated buffer for the first * call to getxattr(), we might be able to avoid the second * call entirely. We only need the second call if the * stack-allocated buffer is too small. But a modest buffer * of 1024 bytes or so will often be big enough. Same applies * to listxattr(). */ static int setup_xattr(struct archive_read_disk *a, struct archive_entry *entry, const char *name, int fd) { ssize_t size; void *value = NULL; const char *accpath; accpath = archive_entry_sourcepath(entry); if (accpath == NULL) accpath = archive_entry_pathname(entry); #if HAVE_FGETXATTR if (fd >= 0) size = fgetxattr(fd, name, NULL, 0); else if (!a->follow_symlinks) size = lgetxattr(accpath, name, NULL, 0); else size = getxattr(accpath, name, NULL, 0); #elif HAVE_FGETEA if (fd >= 0) size = fgetea(fd, name, NULL, 0); else if (!a->follow_symlinks) size = lgetea(accpath, name, NULL, 0); else size = getea(accpath, name, NULL, 0); #endif if (size == -1) { archive_set_error(&a->archive, errno, "Couldn't query extended attribute"); return (ARCHIVE_WARN); } if (size > 0 && (value = malloc(size)) == NULL) { archive_set_error(&a->archive, errno, "Out of memory"); return (ARCHIVE_FATAL); } #if HAVE_FGETXATTR if (fd >= 0) size = fgetxattr(fd, name, value, size); else if (!a->follow_symlinks) size = lgetxattr(accpath, name, value, size); else size = getxattr(accpath, name, value, size); #elif HAVE_FGETEA if (fd >= 0) size = fgetea(fd, name, value, size); else if (!a->follow_symlinks) size = lgetea(accpath, name, value, size); else size = getea(accpath, name, value, size); #endif if (size == -1) { archive_set_error(&a->archive, errno, "Couldn't read extended attribute"); return (ARCHIVE_WARN); } archive_entry_xattr_add_entry(entry, name, value, size); free(value); return (ARCHIVE_OK); } static int setup_xattrs(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { char *list, *p; const char *path; ssize_t list_size; path = archive_entry_sourcepath(entry); if (path == NULL) path = archive_entry_pathname(entry); if (*fd < 0 && a->tree != NULL) { if (a->follow_symlinks || archive_entry_filetype(entry) != AE_IFLNK) *fd = a->open_on_current_dir(a->tree, path, O_RDONLY | O_NONBLOCK); if (*fd < 0) { if (a->tree_enter_working_dir(a->tree) != 0) { archive_set_error(&a->archive, errno, "Couldn't access %s", path); return (ARCHIVE_FAILED); } } } #if HAVE_FLISTXATTR if (*fd >= 0) list_size = flistxattr(*fd, NULL, 0); else if (!a->follow_symlinks) list_size = llistxattr(path, NULL, 0); else list_size = listxattr(path, NULL, 0); #elif HAVE_FLISTEA if (*fd >= 0) list_size = flistea(*fd, NULL, 0); else if (!a->follow_symlinks) list_size = llistea(path, NULL, 0); else list_size = listea(path, NULL, 0); #endif if (list_size == -1) { if (errno == ENOTSUP || errno == ENOSYS) return (ARCHIVE_OK); archive_set_error(&a->archive, errno, "Couldn't list extended attributes"); return (ARCHIVE_WARN); } if (list_size == 0) return (ARCHIVE_OK); if ((list = malloc(list_size)) == NULL) { archive_set_error(&a->archive, errno, "Out of memory"); return (ARCHIVE_FATAL); } #if HAVE_FLISTXATTR if (*fd >= 0) list_size = flistxattr(*fd, list, list_size); else if (!a->follow_symlinks) list_size = llistxattr(path, list, list_size); else list_size = listxattr(path, list, list_size); #elif HAVE_FLISTEA if (*fd >= 0) list_size = flistea(*fd, list, list_size); else if (!a->follow_symlinks) list_size = llistea(path, list, list_size); else list_size = listea(path, list, list_size); #endif if (list_size == -1) { archive_set_error(&a->archive, errno, "Couldn't retrieve extended attributes"); free(list); return (ARCHIVE_WARN); } for (p = list; (p - list) < list_size; p += strlen(p) + 1) { if (strncmp(p, "system.", 7) == 0 || strncmp(p, "xfsroot.", 8) == 0) continue; setup_xattr(a, entry, p, *fd); } free(list); return (ARCHIVE_OK); } #elif HAVE_EXTATTR_GET_FILE && HAVE_EXTATTR_LIST_FILE && \ HAVE_DECL_EXTATTR_NAMESPACE_USER /* * FreeBSD extattr interface. */ /* TODO: Implement this. Follow the Linux model above, but * with FreeBSD-specific system calls, of course. Be careful * to not include the system extattrs that hold ACLs; we handle * those separately. */ static int setup_xattr(struct archive_read_disk *a, struct archive_entry *entry, int namespace, const char *name, const char *fullname, int fd); static int setup_xattr(struct archive_read_disk *a, struct archive_entry *entry, int namespace, const char *name, const char *fullname, int fd) { ssize_t size; void *value = NULL; const char *accpath; accpath = archive_entry_sourcepath(entry); if (accpath == NULL) accpath = archive_entry_pathname(entry); if (fd >= 0) size = extattr_get_fd(fd, namespace, name, NULL, 0); else if (!a->follow_symlinks) size = extattr_get_link(accpath, namespace, name, NULL, 0); else size = extattr_get_file(accpath, namespace, name, NULL, 0); if (size == -1) { archive_set_error(&a->archive, errno, "Couldn't query extended attribute"); return (ARCHIVE_WARN); } if (size > 0 && (value = malloc(size)) == NULL) { archive_set_error(&a->archive, errno, "Out of memory"); return (ARCHIVE_FATAL); } if (fd >= 0) size = extattr_get_fd(fd, namespace, name, value, size); else if (!a->follow_symlinks) size = extattr_get_link(accpath, namespace, name, value, size); else size = extattr_get_file(accpath, namespace, name, value, size); if (size == -1) { free(value); archive_set_error(&a->archive, errno, "Couldn't read extended attribute"); return (ARCHIVE_WARN); } archive_entry_xattr_add_entry(entry, fullname, value, size); free(value); return (ARCHIVE_OK); } static int setup_xattrs(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { char buff[512]; char *list, *p; ssize_t list_size; const char *path; int namespace = EXTATTR_NAMESPACE_USER; path = archive_entry_sourcepath(entry); if (path == NULL) path = archive_entry_pathname(entry); if (*fd < 0 && a->tree != NULL) { if (a->follow_symlinks || archive_entry_filetype(entry) != AE_IFLNK) *fd = a->open_on_current_dir(a->tree, path, O_RDONLY | O_NONBLOCK); if (*fd < 0) { if (a->tree_enter_working_dir(a->tree) != 0) { archive_set_error(&a->archive, errno, "Couldn't access %s", path); return (ARCHIVE_FAILED); } } } if (*fd >= 0) list_size = extattr_list_fd(*fd, namespace, NULL, 0); else if (!a->follow_symlinks) list_size = extattr_list_link(path, namespace, NULL, 0); else list_size = extattr_list_file(path, namespace, NULL, 0); if (list_size == -1 && errno == EOPNOTSUPP) return (ARCHIVE_OK); if (list_size == -1) { archive_set_error(&a->archive, errno, "Couldn't list extended attributes"); return (ARCHIVE_WARN); } if (list_size == 0) return (ARCHIVE_OK); if ((list = malloc(list_size)) == NULL) { archive_set_error(&a->archive, errno, "Out of memory"); return (ARCHIVE_FATAL); } if (*fd >= 0) list_size = extattr_list_fd(*fd, namespace, list, list_size); else if (!a->follow_symlinks) list_size = extattr_list_link(path, namespace, list, list_size); else list_size = extattr_list_file(path, namespace, list, list_size); if (list_size == -1) { archive_set_error(&a->archive, errno, "Couldn't retrieve extended attributes"); free(list); return (ARCHIVE_WARN); } p = list; while ((p - list) < list_size) { size_t len = 255 & (int)*p; char *name; strcpy(buff, "user."); name = buff + strlen(buff); memcpy(name, p + 1, len); name[len] = '\0'; setup_xattr(a, entry, namespace, name, buff, *fd); p += 1 + len; } free(list); return (ARCHIVE_OK); } #else /* * Generic (stub) extended attribute support. */ static int setup_xattrs(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { (void)a; /* UNUSED */ (void)entry; /* UNUSED */ (void)fd; /* UNUSED */ return (ARCHIVE_OK); } #endif #if defined(HAVE_LINUX_FIEMAP_H) /* * Linux FIEMAP sparse interface. * * The FIEMAP ioctl returns an "extent" for each physical allocation * on disk. We need to process those to generate a more compact list * of logical file blocks. We also need to be very careful to use * FIEMAP_FLAG_SYNC here, since there are reports that Linux sometimes * does not report allocations for newly-written data that hasn't * been synced to disk. * * It's important to return a minimal sparse file list because we want * to not trigger sparse file extensions if we don't have to, since * not all readers support them. */ static int setup_sparse_fiemap(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { char buff[4096]; struct fiemap *fm; struct fiemap_extent *fe; int64_t size; int count, do_fiemap, iters; int exit_sts = ARCHIVE_OK; if (archive_entry_filetype(entry) != AE_IFREG || archive_entry_size(entry) <= 0 || archive_entry_hardlink(entry) != NULL) return (ARCHIVE_OK); if (*fd < 0) { const char *path; path = archive_entry_sourcepath(entry); if (path == NULL) path = archive_entry_pathname(entry); if (a->tree != NULL) *fd = a->open_on_current_dir(a->tree, path, O_RDONLY | O_NONBLOCK | O_CLOEXEC); else *fd = open(path, O_RDONLY | O_NONBLOCK | O_CLOEXEC); if (*fd < 0) { archive_set_error(&a->archive, errno, "Can't open `%s'", path); return (ARCHIVE_FAILED); } __archive_ensure_cloexec_flag(*fd); } /* Initialize buffer to avoid the error valgrind complains about. */ memset(buff, 0, sizeof(buff)); count = (sizeof(buff) - sizeof(*fm))/sizeof(*fe); fm = (struct fiemap *)buff; fm->fm_start = 0; fm->fm_length = ~0ULL;; fm->fm_flags = FIEMAP_FLAG_SYNC; fm->fm_extent_count = count; do_fiemap = 1; size = archive_entry_size(entry); for (iters = 0; ; ++iters) { int i, r; r = ioctl(*fd, FS_IOC_FIEMAP, fm); if (r < 0) { /* When something error happens, it is better we * should return ARCHIVE_OK because an earlier * version(<2.6.28) cannot perform FS_IOC_FIEMAP. */ goto exit_setup_sparse_fiemap; } if (fm->fm_mapped_extents == 0) { if (iters == 0) { /* Fully sparse file; insert a zero-length "data" entry */ archive_entry_sparse_add_entry(entry, 0, 0); } break; } fe = fm->fm_extents; for (i = 0; i < (int)fm->fm_mapped_extents; i++, fe++) { if (!(fe->fe_flags & FIEMAP_EXTENT_UNWRITTEN)) { /* The fe_length of the last block does not * adjust itself to its size files. */ int64_t length = fe->fe_length; if (fe->fe_logical + length > (uint64_t)size) length -= fe->fe_logical + length - size; if (fe->fe_logical == 0 && length == size) { /* This is not sparse. */ do_fiemap = 0; break; } if (length > 0) archive_entry_sparse_add_entry(entry, fe->fe_logical, length); } if (fe->fe_flags & FIEMAP_EXTENT_LAST) do_fiemap = 0; } if (do_fiemap) { fe = fm->fm_extents + fm->fm_mapped_extents -1; fm->fm_start = fe->fe_logical + fe->fe_length; } else break; } exit_setup_sparse_fiemap: return (exit_sts); } #if !defined(SEEK_HOLE) || !defined(SEEK_DATA) static int setup_sparse(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { return setup_sparse_fiemap(a, entry, fd); } #endif #endif /* defined(HAVE_LINUX_FIEMAP_H) */ #if defined(SEEK_HOLE) && defined(SEEK_DATA) /* * SEEK_HOLE sparse interface (FreeBSD, Linux, Solaris) */ static int setup_sparse(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { int64_t size; off_t initial_off; off_t off_s, off_e; int exit_sts = ARCHIVE_OK; int check_fully_sparse = 0; if (archive_entry_filetype(entry) != AE_IFREG || archive_entry_size(entry) <= 0 || archive_entry_hardlink(entry) != NULL) return (ARCHIVE_OK); /* Does filesystem support the reporting of hole ? */ if (*fd < 0 && a->tree != NULL) { const char *path; path = archive_entry_sourcepath(entry); if (path == NULL) path = archive_entry_pathname(entry); *fd = a->open_on_current_dir(a->tree, path, O_RDONLY | O_NONBLOCK); if (*fd < 0) { archive_set_error(&a->archive, errno, "Can't open `%s'", path); return (ARCHIVE_FAILED); } } if (*fd >= 0) { #ifdef _PC_MIN_HOLE_SIZE if (fpathconf(*fd, _PC_MIN_HOLE_SIZE) <= 0) return (ARCHIVE_OK); #endif initial_off = lseek(*fd, 0, SEEK_CUR); if (initial_off != 0) lseek(*fd, 0, SEEK_SET); } else { const char *path; path = archive_entry_sourcepath(entry); if (path == NULL) path = archive_entry_pathname(entry); #ifdef _PC_MIN_HOLE_SIZE if (pathconf(path, _PC_MIN_HOLE_SIZE) <= 0) return (ARCHIVE_OK); #endif *fd = open(path, O_RDONLY | O_NONBLOCK | O_CLOEXEC); if (*fd < 0) { archive_set_error(&a->archive, errno, "Can't open `%s'", path); return (ARCHIVE_FAILED); } __archive_ensure_cloexec_flag(*fd); initial_off = 0; } #ifndef _PC_MIN_HOLE_SIZE /* Check if the underlying filesystem supports seek hole */ off_s = lseek(*fd, 0, SEEK_HOLE); if (off_s < 0) #if defined(HAVE_LINUX_FIEMAP_H) return setup_sparse_fiemap(a, entry, fd); #else goto exit_setup_sparse; #endif else if (off_s > 0) lseek(*fd, 0, SEEK_SET); #endif off_s = 0; size = archive_entry_size(entry); while (off_s < size) { off_s = lseek(*fd, off_s, SEEK_DATA); if (off_s == (off_t)-1) { if (errno == ENXIO) { /* no more hole */ if (archive_entry_sparse_count(entry) == 0) { /* Potentially a fully-sparse file. */ check_fully_sparse = 1; } break; } archive_set_error(&a->archive, errno, "lseek(SEEK_HOLE) failed"); exit_sts = ARCHIVE_FAILED; goto exit_setup_sparse; } off_e = lseek(*fd, off_s, SEEK_HOLE); if (off_e == (off_t)-1) { if (errno == ENXIO) { off_e = lseek(*fd, 0, SEEK_END); if (off_e != (off_t)-1) break;/* no more data */ } archive_set_error(&a->archive, errno, "lseek(SEEK_DATA) failed"); exit_sts = ARCHIVE_FAILED; goto exit_setup_sparse; } if (off_s == 0 && off_e == size) break;/* This is not sparse. */ archive_entry_sparse_add_entry(entry, off_s, off_e - off_s); off_s = off_e; } if (check_fully_sparse) { if (lseek(*fd, 0, SEEK_HOLE) == 0 && lseek(*fd, 0, SEEK_END) == size) { /* Fully sparse file; insert a zero-length "data" entry */ archive_entry_sparse_add_entry(entry, 0, 0); } } exit_setup_sparse: lseek(*fd, initial_off, SEEK_SET); return (exit_sts); } #elif !defined(HAVE_LINUX_FIEMAP_H) /* * Generic (stub) sparse support. */ static int setup_sparse(struct archive_read_disk *a, struct archive_entry *entry, int *fd) { (void)a; /* UNUSED */ (void)entry; /* UNUSED */ (void)fd; /* UNUSED */ return (ARCHIVE_OK); } #endif #endif /* !defined(_WIN32) || defined(__CYGWIN__) */ Index: head/contrib/libarchive/libarchive/archive_read_disk_posix.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_disk_posix.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_read_disk_posix.c (revision 311903) @@ -1,2671 +1,2671 @@ /*- * Copyright (c) 2003-2009 Tim Kientzle * Copyright (c) 2010-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ /* This is the tree-walking code for POSIX systems. */ #if !defined(_WIN32) || defined(__CYGWIN__) #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_SYS_PARAM_H #include #endif #ifdef HAVE_SYS_MOUNT_H #include #endif #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_SYS_STATFS_H #include #endif #ifdef HAVE_SYS_STATVFS_H #include #endif #ifdef HAVE_SYS_TIME_H #include #endif #ifdef HAVE_LINUX_MAGIC_H #include #endif #ifdef HAVE_LINUX_FS_H #include #endif /* * Some Linux distributions have both linux/ext2_fs.h and ext2fs/ext2_fs.h. * As the include guards don't agree, the order of include is important. */ #ifdef HAVE_LINUX_EXT2_FS_H #include /* for Linux file flags */ #endif #if defined(HAVE_EXT2FS_EXT2_FS_H) && !defined(__CYGWIN__) #include /* Linux file flags, broken on Cygwin */ #endif #ifdef HAVE_DIRECT_H #include #endif #ifdef HAVE_DIRENT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_SYS_IOCTL_H #include #endif #include "archive.h" #include "archive_string.h" #include "archive_entry.h" #include "archive_private.h" #include "archive_read_disk_private.h" #ifndef HAVE_FCHDIR #error fchdir function required. #endif #ifndef O_BINARY #define O_BINARY 0 #endif #ifndef O_CLOEXEC #define O_CLOEXEC 0 #endif /*- * This is a new directory-walking system that addresses a number * of problems I've had with fts(3). In particular, it has no * pathname-length limits (other than the size of 'int'), handles * deep logical traversals, uses considerably less memory, and has * an opaque interface (easier to modify in the future). * * Internally, it keeps a single list of "tree_entry" items that * represent filesystem objects that require further attention. * Non-directories are not kept in memory: they are pulled from * readdir(), returned to the client, then freed as soon as possible. * Any directory entry to be traversed gets pushed onto the stack. * * There is surprisingly little information that needs to be kept for * each item on the stack. Just the name, depth (represented here as the * string length of the parent directory's pathname), and some markers * indicating how to get back to the parent (via chdir("..") for a * regular dir or via fchdir(2) for a symlink). */ /* * TODO: * 1) Loop checking. * 3) Arbitrary logical traversals by closing/reopening intermediate fds. */ struct restore_time { const char *name; time_t mtime; long mtime_nsec; time_t atime; long atime_nsec; mode_t filetype; int noatime; }; struct tree_entry { int depth; struct tree_entry *next; struct tree_entry *parent; struct archive_string name; size_t dirname_length; int64_t dev; int64_t ino; int flags; int filesystem_id; /* How to return back to the parent of a symlink. */ int symlink_parent_fd; /* How to restore time of a directory. */ struct restore_time restore_time; }; struct filesystem { int64_t dev; int synthetic; int remote; int noatime; #if defined(USE_READDIR_R) size_t name_max; #endif long incr_xfer_size; long max_xfer_size; long min_xfer_size; long xfer_align; /* * Buffer used for reading file contents. */ /* Exactly allocated memory pointer. */ unsigned char *allocation_ptr; /* Pointer adjusted to the filesystem alignment . */ unsigned char *buff; size_t buff_size; }; /* Definitions for tree_entry.flags bitmap. */ #define isDir 1 /* This entry is a regular directory. */ #define isDirLink 2 /* This entry is a symbolic link to a directory. */ #define needsFirstVisit 4 /* This is an initial entry. */ #define needsDescent 8 /* This entry needs to be previsited. */ #define needsOpen 16 /* This is a directory that needs to be opened. */ #define needsAscent 32 /* This entry needs to be postvisited. */ /* * Local data for this package. */ struct tree { struct tree_entry *stack; struct tree_entry *current; DIR *d; #define INVALID_DIR_HANDLE NULL struct dirent *de; #if defined(USE_READDIR_R) struct dirent *dirent; size_t dirent_allocated; #endif int flags; int visit_type; /* Error code from last failed operation. */ int tree_errno; /* Dynamically-sized buffer for holding path */ struct archive_string path; /* Last path element */ const char *basename; /* Leading dir length */ size_t dirname_length; int depth; int openCount; int maxOpenCount; int initial_dir_fd; int working_dir_fd; struct stat lst; struct stat st; int descend; int nlink; /* How to restore time of a file. */ struct restore_time restore_time; struct entry_sparse { int64_t length; int64_t offset; } *sparse_list, *current_sparse; int sparse_count; int sparse_list_size; char initial_symlink_mode; char symlink_mode; struct filesystem *current_filesystem; struct filesystem *filesystem_table; int initial_filesystem_id; int current_filesystem_id; int max_filesystem_id; int allocated_filesystem; int entry_fd; int entry_eof; int64_t entry_remaining_bytes; int64_t entry_total; unsigned char *entry_buff; size_t entry_buff_size; }; /* Definitions for tree.flags bitmap. */ #define hasStat 16 /* The st entry is valid. */ #define hasLstat 32 /* The lst entry is valid. */ #define onWorkingDir 64 /* We are on the working dir where we are * reading directory entry at this time. */ #define needsRestoreTimes 128 #define onInitialDir 256 /* We are on the initial dir. */ static int tree_dir_next_posix(struct tree *t); #ifdef HAVE_DIRENT_D_NAMLEN /* BSD extension; avoids need for a strlen() call. */ #define D_NAMELEN(dp) (dp)->d_namlen #else #define D_NAMELEN(dp) (strlen((dp)->d_name)) #endif /* Initiate/terminate a tree traversal. */ static struct tree *tree_open(const char *, int, int); static struct tree *tree_reopen(struct tree *, const char *, int); static void tree_close(struct tree *); static void tree_free(struct tree *); static void tree_push(struct tree *, const char *, int, int64_t, int64_t, struct restore_time *); static int tree_enter_initial_dir(struct tree *); static int tree_enter_working_dir(struct tree *); static int tree_current_dir_fd(struct tree *); /* * tree_next() returns Zero if there is no next entry, non-zero if * there is. Note that directories are visited three times. * Directories are always visited first as part of enumerating their * parent; that is a "regular" visit. If tree_descend() is invoked at * that time, the directory is added to a work list and will * subsequently be visited two more times: once just after descending * into the directory ("postdescent") and again just after ascending * back to the parent ("postascent"). * * TREE_ERROR_DIR is returned if the descent failed (because the * directory couldn't be opened, for instance). This is returned * instead of TREE_POSTDESCENT/TREE_POSTASCENT. TREE_ERROR_DIR is not a * fatal error, but it does imply that the relevant subtree won't be * visited. TREE_ERROR_FATAL is returned for an error that left the * traversal completely hosed. Right now, this is only returned for * chdir() failures during ascent. */ #define TREE_REGULAR 1 #define TREE_POSTDESCENT 2 #define TREE_POSTASCENT 3 #define TREE_ERROR_DIR -1 #define TREE_ERROR_FATAL -2 static int tree_next(struct tree *); /* * Return information about the current entry. */ /* * The current full pathname, length of the full pathname, and a name * that can be used to access the file. Because tree does use chdir * extensively, the access path is almost never the same as the full * current path. * * TODO: On platforms that support it, use openat()-style operations * to eliminate the chdir() operations entirely while still supporting * arbitrarily deep traversals. This makes access_path troublesome to * support, of course, which means we'll need a rich enough interface * that clients can function without it. (In particular, we'll need * tree_current_open() that returns an open file descriptor.) * */ static const char *tree_current_path(struct tree *); static const char *tree_current_access_path(struct tree *); /* * Request the lstat() or stat() data for the current path. Since the * tree package needs to do some of this anyway, and caches the * results, you should take advantage of it here if you need it rather * than make a redundant stat() or lstat() call of your own. */ static const struct stat *tree_current_stat(struct tree *); static const struct stat *tree_current_lstat(struct tree *); static int tree_current_is_symblic_link_target(struct tree *); /* The following functions use tricks to avoid a certain number of * stat()/lstat() calls. */ /* "is_physical_dir" is equivalent to S_ISDIR(tree_current_lstat()->st_mode) */ static int tree_current_is_physical_dir(struct tree *); /* "is_dir" is equivalent to S_ISDIR(tree_current_stat()->st_mode) */ static int tree_current_is_dir(struct tree *); static int update_current_filesystem(struct archive_read_disk *a, int64_t dev); static int setup_current_filesystem(struct archive_read_disk *); static int tree_target_is_same_as_parent(struct tree *, const struct stat *); static int _archive_read_disk_open(struct archive *, const char *); static int _archive_read_free(struct archive *); static int _archive_read_close(struct archive *); static int _archive_read_data_block(struct archive *, const void **, size_t *, int64_t *); static int _archive_read_next_header(struct archive *, struct archive_entry **); static int _archive_read_next_header2(struct archive *, struct archive_entry *); static const char *trivial_lookup_gname(void *, int64_t gid); static const char *trivial_lookup_uname(void *, int64_t uid); static int setup_sparse(struct archive_read_disk *, struct archive_entry *); static int close_and_restore_time(int fd, struct tree *, struct restore_time *); static int open_on_current_dir(struct tree *, const char *, int); static int tree_dup(int); static struct archive_vtable * archive_read_disk_vtable(void) { static struct archive_vtable av; static int inited = 0; if (!inited) { av.archive_free = _archive_read_free; av.archive_close = _archive_read_close; av.archive_read_data_block = _archive_read_data_block; av.archive_read_next_header = _archive_read_next_header; av.archive_read_next_header2 = _archive_read_next_header2; inited = 1; } return (&av); } const char * archive_read_disk_gname(struct archive *_a, int64_t gid) { struct archive_read_disk *a = (struct archive_read_disk *)_a; if (ARCHIVE_OK != __archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_gname")) return (NULL); if (a->lookup_gname == NULL) return (NULL); return ((*a->lookup_gname)(a->lookup_gname_data, gid)); } const char * archive_read_disk_uname(struct archive *_a, int64_t uid) { struct archive_read_disk *a = (struct archive_read_disk *)_a; if (ARCHIVE_OK != __archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_uname")) return (NULL); if (a->lookup_uname == NULL) return (NULL); return ((*a->lookup_uname)(a->lookup_uname_data, uid)); } int archive_read_disk_set_gname_lookup(struct archive *_a, void *private_data, const char * (*lookup_gname)(void *private, int64_t gid), void (*cleanup_gname)(void *private)) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(&a->archive, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_set_gname_lookup"); if (a->cleanup_gname != NULL && a->lookup_gname_data != NULL) (a->cleanup_gname)(a->lookup_gname_data); a->lookup_gname = lookup_gname; a->cleanup_gname = cleanup_gname; a->lookup_gname_data = private_data; return (ARCHIVE_OK); } int archive_read_disk_set_uname_lookup(struct archive *_a, void *private_data, const char * (*lookup_uname)(void *private, int64_t uid), void (*cleanup_uname)(void *private)) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(&a->archive, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_set_uname_lookup"); if (a->cleanup_uname != NULL && a->lookup_uname_data != NULL) (a->cleanup_uname)(a->lookup_uname_data); a->lookup_uname = lookup_uname; a->cleanup_uname = cleanup_uname; a->lookup_uname_data = private_data; return (ARCHIVE_OK); } /* * Create a new archive_read_disk object and initialize it with global state. */ struct archive * archive_read_disk_new(void) { struct archive_read_disk *a; a = (struct archive_read_disk *)calloc(1, sizeof(*a)); if (a == NULL) return (NULL); a->archive.magic = ARCHIVE_READ_DISK_MAGIC; a->archive.state = ARCHIVE_STATE_NEW; a->archive.vtable = archive_read_disk_vtable(); a->entry = archive_entry_new2(&a->archive); a->lookup_uname = trivial_lookup_uname; a->lookup_gname = trivial_lookup_gname; a->enable_copyfile = 1; a->traverse_mount_points = 1; a->open_on_current_dir = open_on_current_dir; a->tree_current_dir_fd = tree_current_dir_fd; a->tree_enter_working_dir = tree_enter_working_dir; return (&a->archive); } static int _archive_read_free(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; int r; if (_a == NULL) return (ARCHIVE_OK); archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_read_free"); if (a->archive.state != ARCHIVE_STATE_CLOSED) r = _archive_read_close(&a->archive); else r = ARCHIVE_OK; tree_free(a->tree); if (a->cleanup_gname != NULL && a->lookup_gname_data != NULL) (a->cleanup_gname)(a->lookup_gname_data); if (a->cleanup_uname != NULL && a->lookup_uname_data != NULL) (a->cleanup_uname)(a->lookup_uname_data); archive_string_free(&a->archive.error_string); archive_entry_free(a->entry); a->archive.magic = 0; __archive_clean(&a->archive); free(a); return (r); } static int _archive_read_close(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_read_close"); if (a->archive.state != ARCHIVE_STATE_FATAL) a->archive.state = ARCHIVE_STATE_CLOSED; tree_close(a->tree); return (ARCHIVE_OK); } static void setup_symlink_mode(struct archive_read_disk *a, char symlink_mode, int follow_symlinks) { a->symlink_mode = symlink_mode; a->follow_symlinks = follow_symlinks; if (a->tree != NULL) { a->tree->initial_symlink_mode = a->symlink_mode; a->tree->symlink_mode = a->symlink_mode; } } int archive_read_disk_set_symlink_logical(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_set_symlink_logical"); setup_symlink_mode(a, 'L', 1); return (ARCHIVE_OK); } int archive_read_disk_set_symlink_physical(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_set_symlink_physical"); setup_symlink_mode(a, 'P', 0); return (ARCHIVE_OK); } int archive_read_disk_set_symlink_hybrid(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_set_symlink_hybrid"); setup_symlink_mode(a, 'H', 1);/* Follow symlinks initially. */ return (ARCHIVE_OK); } int archive_read_disk_set_atime_restored(struct archive *_a) { #ifndef HAVE_UTIMES static int warning_done = 0; #endif struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_restore_atime"); #ifdef HAVE_UTIMES a->restore_time = 1; if (a->tree != NULL) a->tree->flags |= needsRestoreTimes; return (ARCHIVE_OK); #else if (warning_done) /* Warning was already emitted; suppress further warnings. */ return (ARCHIVE_OK); archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Cannot restore access time on this system"); warning_done = 1; return (ARCHIVE_WARN); #endif } int archive_read_disk_set_behavior(struct archive *_a, int flags) { struct archive_read_disk *a = (struct archive_read_disk *)_a; int r = ARCHIVE_OK; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_honor_nodump"); if (flags & ARCHIVE_READDISK_RESTORE_ATIME) r = archive_read_disk_set_atime_restored(_a); else { a->restore_time = 0; if (a->tree != NULL) a->tree->flags &= ~needsRestoreTimes; } if (flags & ARCHIVE_READDISK_HONOR_NODUMP) a->honor_nodump = 1; else a->honor_nodump = 0; if (flags & ARCHIVE_READDISK_MAC_COPYFILE) a->enable_copyfile = 1; else a->enable_copyfile = 0; if (flags & ARCHIVE_READDISK_NO_TRAVERSE_MOUNTS) a->traverse_mount_points = 0; else a->traverse_mount_points = 1; if (flags & ARCHIVE_READDISK_NO_XATTR) a->suppress_xattr = 1; else a->suppress_xattr = 0; return (r); } /* * Trivial implementations of gname/uname lookup functions. * These are normally overridden by the client, but these stub * versions ensure that we always have something that works. */ static const char * trivial_lookup_gname(void *private_data, int64_t gid) { (void)private_data; /* UNUSED */ (void)gid; /* UNUSED */ return (NULL); } static const char * trivial_lookup_uname(void *private_data, int64_t uid) { (void)private_data; /* UNUSED */ (void)uid; /* UNUSED */ return (NULL); } /* * Allocate memory for the reading buffer adjusted to the filesystem * alignment. */ static int setup_suitable_read_buffer(struct archive_read_disk *a) { struct tree *t = a->tree; struct filesystem *cf = t->current_filesystem; size_t asize; size_t s; if (cf->allocation_ptr == NULL) { /* If we couldn't get a filesystem alignment, * we use 4096 as default value but we won't use * O_DIRECT to open() and openat() operations. */ long xfer_align = (cf->xfer_align == -1)?4096:cf->xfer_align; if (cf->max_xfer_size != -1) asize = cf->max_xfer_size + xfer_align; else { long incr = cf->incr_xfer_size; /* Some platform does not set a proper value to * incr_xfer_size.*/ if (incr < 0) incr = cf->min_xfer_size; if (cf->min_xfer_size < 0) { incr = xfer_align; asize = xfer_align; } else asize = cf->min_xfer_size; /* Increase a buffer size up to 64K bytes in - * a proper incremant size. */ + * a proper increment size. */ while (asize < 1024*64) asize += incr; /* Take a margin to adjust to the filesystem * alignment. */ asize += xfer_align; } cf->allocation_ptr = malloc(asize); if (cf->allocation_ptr == NULL) { archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory"); a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } /* * Calculate proper address for the filesystem. */ s = (uintptr_t)cf->allocation_ptr; s %= xfer_align; if (s > 0) s = xfer_align - s; /* * Set a read buffer pointer in the proper alignment of * the current filesystem. */ cf->buff = cf->allocation_ptr + s; cf->buff_size = asize - xfer_align; } return (ARCHIVE_OK); } static int _archive_read_data_block(struct archive *_a, const void **buff, size_t *size, int64_t *offset) { struct archive_read_disk *a = (struct archive_read_disk *)_a; struct tree *t = a->tree; int r; ssize_t bytes; size_t buffbytes; int empty_sparse_region = 0; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, "archive_read_data_block"); if (t->entry_eof || t->entry_remaining_bytes <= 0) { r = ARCHIVE_EOF; goto abort_read_data; } /* * Open the current file. */ if (t->entry_fd < 0) { int flags = O_RDONLY | O_BINARY | O_CLOEXEC; /* * Eliminate or reduce cache effects if we can. * * Carefully consider this to be enabled. */ #if defined(O_DIRECT) && 0/* Disabled for now */ if (t->current_filesystem->xfer_align != -1 && t->nlink == 1) flags |= O_DIRECT; #endif #if defined(O_NOATIME) /* * Linux has O_NOATIME flag; use it if we need. */ if ((t->flags & needsRestoreTimes) != 0 && t->restore_time.noatime == 0) flags |= O_NOATIME; do { #endif t->entry_fd = open_on_current_dir(t, tree_current_access_path(t), flags); __archive_ensure_cloexec_flag(t->entry_fd); #if defined(O_NOATIME) /* * When we did open the file with O_NOATIME flag, * if successful, set 1 to t->restore_time.noatime * not to restore an atime of the file later. * if failed by EPERM, retry it without O_NOATIME flag. */ if (flags & O_NOATIME) { if (t->entry_fd >= 0) t->restore_time.noatime = 1; else if (errno == EPERM) { flags &= ~O_NOATIME; continue; } } } while (0); #endif if (t->entry_fd < 0) { archive_set_error(&a->archive, errno, "Couldn't open %s", tree_current_path(t)); r = ARCHIVE_FAILED; tree_enter_initial_dir(t); goto abort_read_data; } tree_enter_initial_dir(t); } /* * Allocate read buffer if not allocated. */ if (t->current_filesystem->allocation_ptr == NULL) { r = setup_suitable_read_buffer(a); if (r != ARCHIVE_OK) { a->archive.state = ARCHIVE_STATE_FATAL; goto abort_read_data; } } t->entry_buff = t->current_filesystem->buff; t->entry_buff_size = t->current_filesystem->buff_size; buffbytes = t->entry_buff_size; if ((int64_t)buffbytes > t->current_sparse->length) buffbytes = t->current_sparse->length; if (t->current_sparse->length == 0) empty_sparse_region = 1; /* * Skip hole. * TODO: Should we consider t->current_filesystem->xfer_align? */ if (t->current_sparse->offset > t->entry_total) { if (lseek(t->entry_fd, (off_t)t->current_sparse->offset, SEEK_SET) < 0) { archive_set_error(&a->archive, errno, "Seek error"); r = ARCHIVE_FATAL; a->archive.state = ARCHIVE_STATE_FATAL; goto abort_read_data; } bytes = t->current_sparse->offset - t->entry_total; t->entry_remaining_bytes -= bytes; t->entry_total += bytes; } /* * Read file contents. */ if (buffbytes > 0) { bytes = read(t->entry_fd, t->entry_buff, buffbytes); if (bytes < 0) { archive_set_error(&a->archive, errno, "Read error"); r = ARCHIVE_FATAL; a->archive.state = ARCHIVE_STATE_FATAL; goto abort_read_data; } } else bytes = 0; /* * Return an EOF unless we've read a leading empty sparse region, which * is used to represent fully-sparse files. */ if (bytes == 0 && !empty_sparse_region) { /* Get EOF */ t->entry_eof = 1; r = ARCHIVE_EOF; goto abort_read_data; } *buff = t->entry_buff; *size = bytes; *offset = t->entry_total; t->entry_total += bytes; t->entry_remaining_bytes -= bytes; if (t->entry_remaining_bytes == 0) { /* Close the current file descriptor */ close_and_restore_time(t->entry_fd, t, &t->restore_time); t->entry_fd = -1; t->entry_eof = 1; } t->current_sparse->offset += bytes; t->current_sparse->length -= bytes; if (t->current_sparse->length == 0 && !t->entry_eof) t->current_sparse++; return (ARCHIVE_OK); abort_read_data: *buff = NULL; *size = 0; *offset = t->entry_total; if (t->entry_fd >= 0) { /* Close the current file descriptor */ close_and_restore_time(t->entry_fd, t, &t->restore_time); t->entry_fd = -1; } return (r); } static int next_entry(struct archive_read_disk *a, struct tree *t, struct archive_entry *entry) { const struct stat *st; /* info to use for this entry */ const struct stat *lst;/* lstat() information */ const char *name; int descend, r; st = NULL; lst = NULL; t->descend = 0; do { switch (tree_next(t)) { case TREE_ERROR_FATAL: archive_set_error(&a->archive, t->tree_errno, "%s: Unable to continue traversing directory tree", tree_current_path(t)); a->archive.state = ARCHIVE_STATE_FATAL; tree_enter_initial_dir(t); return (ARCHIVE_FATAL); case TREE_ERROR_DIR: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: Couldn't visit directory", tree_current_path(t)); tree_enter_initial_dir(t); return (ARCHIVE_FAILED); case 0: tree_enter_initial_dir(t); return (ARCHIVE_EOF); case TREE_POSTDESCENT: case TREE_POSTASCENT: break; case TREE_REGULAR: lst = tree_current_lstat(t); if (lst == NULL) { archive_set_error(&a->archive, errno, "%s: Cannot stat", tree_current_path(t)); tree_enter_initial_dir(t); return (ARCHIVE_FAILED); } break; } } while (lst == NULL); #ifdef __APPLE__ if (a->enable_copyfile) { /* If we're using copyfile(), ignore "._XXX" files. */ const char *bname = strrchr(tree_current_path(t), '/'); if (bname == NULL) bname = tree_current_path(t); else ++bname; if (bname[0] == '.' && bname[1] == '_') return (ARCHIVE_RETRY); } #endif archive_entry_copy_pathname(entry, tree_current_path(t)); /* * Perform path matching. */ if (a->matching) { r = archive_match_path_excluded(a->matching, entry); if (r < 0) { archive_set_error(&(a->archive), errno, "Failed : %s", archive_error_string(a->matching)); return (r); } if (r) { if (a->excluded_cb_func) a->excluded_cb_func(&(a->archive), a->excluded_cb_data, entry); return (ARCHIVE_RETRY); } } /* * Distinguish 'L'/'P'/'H' symlink following. */ switch(t->symlink_mode) { case 'H': /* 'H': After the first item, rest like 'P'. */ t->symlink_mode = 'P'; /* 'H': First item (from command line) like 'L'. */ /* FALLTHROUGH */ case 'L': /* 'L': Do descend through a symlink to dir. */ descend = tree_current_is_dir(t); /* 'L': Follow symlinks to files. */ a->symlink_mode = 'L'; a->follow_symlinks = 1; /* 'L': Archive symlinks as targets, if we can. */ st = tree_current_stat(t); if (st != NULL && !tree_target_is_same_as_parent(t, st)) break; /* If stat fails, we have a broken symlink; * in that case, don't follow the link. */ /* FALLTHROUGH */ default: /* 'P': Don't descend through a symlink to dir. */ descend = tree_current_is_physical_dir(t); /* 'P': Don't follow symlinks to files. */ a->symlink_mode = 'P'; a->follow_symlinks = 0; /* 'P': Archive symlinks as symlinks. */ st = lst; break; } if (update_current_filesystem(a, st->st_dev) != ARCHIVE_OK) { a->archive.state = ARCHIVE_STATE_FATAL; tree_enter_initial_dir(t); return (ARCHIVE_FATAL); } if (t->initial_filesystem_id == -1) t->initial_filesystem_id = t->current_filesystem_id; if (!a->traverse_mount_points) { if (t->initial_filesystem_id != t->current_filesystem_id) descend = 0; } t->descend = descend; /* * Honor nodump flag. * If the file is marked with nodump flag, do not return this entry. */ if (a->honor_nodump) { #if defined(HAVE_STRUCT_STAT_ST_FLAGS) && defined(UF_NODUMP) if (st->st_flags & UF_NODUMP) return (ARCHIVE_RETRY); #elif defined(EXT2_IOC_GETFLAGS) && defined(EXT2_NODUMP_FL) &&\ defined(HAVE_WORKING_EXT2_IOC_GETFLAGS) if (S_ISREG(st->st_mode) || S_ISDIR(st->st_mode)) { int stflags; t->entry_fd = open_on_current_dir(t, tree_current_access_path(t), O_RDONLY | O_NONBLOCK | O_CLOEXEC); __archive_ensure_cloexec_flag(t->entry_fd); if (t->entry_fd >= 0) { r = ioctl(t->entry_fd, EXT2_IOC_GETFLAGS, &stflags); if (r == 0 && (stflags & EXT2_NODUMP_FL) != 0) return (ARCHIVE_RETRY); } } #endif } archive_entry_copy_stat(entry, st); /* Save the times to be restored. This must be in before * calling archive_read_disk_descend() or any chance of it, * especially, invoking a callback. */ t->restore_time.mtime = archive_entry_mtime(entry); t->restore_time.mtime_nsec = archive_entry_mtime_nsec(entry); t->restore_time.atime = archive_entry_atime(entry); t->restore_time.atime_nsec = archive_entry_atime_nsec(entry); t->restore_time.filetype = archive_entry_filetype(entry); t->restore_time.noatime = t->current_filesystem->noatime; /* * Perform time matching. */ if (a->matching) { r = archive_match_time_excluded(a->matching, entry); if (r < 0) { archive_set_error(&(a->archive), errno, "Failed : %s", archive_error_string(a->matching)); return (r); } if (r) { if (a->excluded_cb_func) a->excluded_cb_func(&(a->archive), a->excluded_cb_data, entry); return (ARCHIVE_RETRY); } } /* Lookup uname/gname */ name = archive_read_disk_uname(&(a->archive), archive_entry_uid(entry)); if (name != NULL) archive_entry_copy_uname(entry, name); name = archive_read_disk_gname(&(a->archive), archive_entry_gid(entry)); if (name != NULL) archive_entry_copy_gname(entry, name); /* * Perform owner matching. */ if (a->matching) { r = archive_match_owner_excluded(a->matching, entry); if (r < 0) { archive_set_error(&(a->archive), errno, "Failed : %s", archive_error_string(a->matching)); return (r); } if (r) { if (a->excluded_cb_func) a->excluded_cb_func(&(a->archive), a->excluded_cb_data, entry); return (ARCHIVE_RETRY); } } /* * Invoke a meta data filter callback. */ if (a->metadata_filter_func) { if (!a->metadata_filter_func(&(a->archive), a->metadata_filter_data, entry)) return (ARCHIVE_RETRY); } /* * Populate the archive_entry with metadata from the disk. */ archive_entry_copy_sourcepath(entry, tree_current_access_path(t)); r = archive_read_disk_entry_from_file(&(a->archive), entry, t->entry_fd, st); return (r); } static int _archive_read_next_header(struct archive *_a, struct archive_entry **entryp) { int ret; struct archive_read_disk *a = (struct archive_read_disk *)_a; *entryp = NULL; ret = _archive_read_next_header2(_a, a->entry); *entryp = a->entry; return ret; } static int _archive_read_next_header2(struct archive *_a, struct archive_entry *entry) { struct archive_read_disk *a = (struct archive_read_disk *)_a; struct tree *t; int r; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_read_next_header2"); t = a->tree; if (t->entry_fd >= 0) { close_and_restore_time(t->entry_fd, t, &t->restore_time); t->entry_fd = -1; } for (;;) { r = next_entry(a, t, entry); if (t->entry_fd >= 0) { close(t->entry_fd); t->entry_fd = -1; } if (r == ARCHIVE_RETRY) { archive_entry_clear(entry); continue; } break; } /* Return to the initial directory. */ tree_enter_initial_dir(t); /* * EOF and FATAL are persistent at this layer. By * modifying the state, we guarantee that future calls to * read a header or read data will fail. */ switch (r) { case ARCHIVE_EOF: a->archive.state = ARCHIVE_STATE_EOF; break; case ARCHIVE_OK: case ARCHIVE_WARN: /* Overwrite the sourcepath based on the initial directory. */ archive_entry_copy_sourcepath(entry, tree_current_path(t)); t->entry_total = 0; if (archive_entry_filetype(entry) == AE_IFREG) { t->nlink = archive_entry_nlink(entry); t->entry_remaining_bytes = archive_entry_size(entry); t->entry_eof = (t->entry_remaining_bytes == 0)? 1: 0; if (!t->entry_eof && setup_sparse(a, entry) != ARCHIVE_OK) return (ARCHIVE_FATAL); } else { t->entry_remaining_bytes = 0; t->entry_eof = 1; } a->archive.state = ARCHIVE_STATE_DATA; break; case ARCHIVE_RETRY: break; case ARCHIVE_FATAL: a->archive.state = ARCHIVE_STATE_FATAL; break; } __archive_reset_read_data(&a->archive); return (r); } static int setup_sparse(struct archive_read_disk *a, struct archive_entry *entry) { struct tree *t = a->tree; int64_t length, offset; int i; t->sparse_count = archive_entry_sparse_reset(entry); if (t->sparse_count+1 > t->sparse_list_size) { free(t->sparse_list); t->sparse_list_size = t->sparse_count + 1; t->sparse_list = malloc(sizeof(t->sparse_list[0]) * t->sparse_list_size); if (t->sparse_list == NULL) { t->sparse_list_size = 0; archive_set_error(&a->archive, ENOMEM, "Can't allocate data"); a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } } for (i = 0; i < t->sparse_count; i++) { archive_entry_sparse_next(entry, &offset, &length); t->sparse_list[i].offset = offset; t->sparse_list[i].length = length; } if (i == 0) { t->sparse_list[i].offset = 0; t->sparse_list[i].length = archive_entry_size(entry); } else { t->sparse_list[i].offset = archive_entry_size(entry); t->sparse_list[i].length = 0; } t->current_sparse = t->sparse_list; return (ARCHIVE_OK); } int archive_read_disk_set_matching(struct archive *_a, struct archive *_ma, void (*_excluded_func)(struct archive *, void *, struct archive_entry *), void *_client_data) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_set_matching"); a->matching = _ma; a->excluded_cb_func = _excluded_func; a->excluded_cb_data = _client_data; return (ARCHIVE_OK); } int archive_read_disk_set_metadata_filter_callback(struct archive *_a, int (*_metadata_filter_func)(struct archive *, void *, struct archive_entry *), void *_client_data) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_ANY, "archive_read_disk_set_metadata_filter_callback"); a->metadata_filter_func = _metadata_filter_func; a->metadata_filter_data = _client_data; return (ARCHIVE_OK); } int archive_read_disk_can_descend(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; struct tree *t = a->tree; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_read_disk_can_descend"); return (t->visit_type == TREE_REGULAR && t->descend); } /* * Called by the client to mark the directory just returned from * tree_next() as needing to be visited. */ int archive_read_disk_descend(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; struct tree *t = a->tree; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_read_disk_descend"); if (t->visit_type != TREE_REGULAR || !t->descend) return (ARCHIVE_OK); if (tree_current_is_physical_dir(t)) { tree_push(t, t->basename, t->current_filesystem_id, t->lst.st_dev, t->lst.st_ino, &t->restore_time); t->stack->flags |= isDir; } else if (tree_current_is_dir(t)) { tree_push(t, t->basename, t->current_filesystem_id, t->st.st_dev, t->st.st_ino, &t->restore_time); t->stack->flags |= isDirLink; } t->descend = 0; return (ARCHIVE_OK); } int archive_read_disk_open(struct archive *_a, const char *pathname) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_NEW | ARCHIVE_STATE_CLOSED, "archive_read_disk_open"); archive_clear_error(&a->archive); return (_archive_read_disk_open(_a, pathname)); } int archive_read_disk_open_w(struct archive *_a, const wchar_t *pathname) { struct archive_read_disk *a = (struct archive_read_disk *)_a; struct archive_string path; int ret; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_NEW | ARCHIVE_STATE_CLOSED, "archive_read_disk_open_w"); archive_clear_error(&a->archive); /* Make a char string from a wchar_t string. */ archive_string_init(&path); if (archive_string_append_from_wcs(&path, pathname, wcslen(pathname)) != 0) { if (errno == ENOMEM) archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't convert a path to a char string"); a->archive.state = ARCHIVE_STATE_FATAL; ret = ARCHIVE_FATAL; } else ret = _archive_read_disk_open(_a, path.s); archive_string_free(&path); return (ret); } static int _archive_read_disk_open(struct archive *_a, const char *pathname) { struct archive_read_disk *a = (struct archive_read_disk *)_a; if (a->tree != NULL) a->tree = tree_reopen(a->tree, pathname, a->restore_time); else a->tree = tree_open(pathname, a->symlink_mode, a->restore_time); if (a->tree == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate tar data"); a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } a->archive.state = ARCHIVE_STATE_HEADER; return (ARCHIVE_OK); } /* * Return a current filesystem ID which is index of the filesystem entry * you've visited through archive_read_disk. */ int archive_read_disk_current_filesystem(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, "archive_read_disk_current_filesystem"); return (a->tree->current_filesystem_id); } static int update_current_filesystem(struct archive_read_disk *a, int64_t dev) { struct tree *t = a->tree; int i, fid; if (t->current_filesystem != NULL && t->current_filesystem->dev == dev) return (ARCHIVE_OK); for (i = 0; i < t->max_filesystem_id; i++) { if (t->filesystem_table[i].dev == dev) { /* There is the filesystem ID we've already generated. */ t->current_filesystem_id = i; t->current_filesystem = &(t->filesystem_table[i]); return (ARCHIVE_OK); } } /* * This is the new filesystem which we have to generate a new ID for. */ fid = t->max_filesystem_id++; if (t->max_filesystem_id > t->allocated_filesystem) { size_t s; void *p; s = t->max_filesystem_id * 2; p = realloc(t->filesystem_table, s * sizeof(*t->filesystem_table)); if (p == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate tar data"); return (ARCHIVE_FATAL); } t->filesystem_table = (struct filesystem *)p; t->allocated_filesystem = s; } t->current_filesystem_id = fid; t->current_filesystem = &(t->filesystem_table[fid]); t->current_filesystem->dev = dev; t->current_filesystem->allocation_ptr = NULL; t->current_filesystem->buff = NULL; /* Setup the current filesystem properties which depend on * platform specific. */ return (setup_current_filesystem(a)); } /* * Returns 1 if current filesystem is generated filesystem, 0 if it is not * or -1 if it is unknown. */ int archive_read_disk_current_filesystem_is_synthetic(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, "archive_read_disk_current_filesystem"); return (a->tree->current_filesystem->synthetic); } /* * Returns 1 if current filesystem is remote filesystem, 0 if it is not * or -1 if it is unknown. */ int archive_read_disk_current_filesystem_is_remote(struct archive *_a) { struct archive_read_disk *a = (struct archive_read_disk *)_a; archive_check_magic(_a, ARCHIVE_READ_DISK_MAGIC, ARCHIVE_STATE_DATA, "archive_read_disk_current_filesystem"); return (a->tree->current_filesystem->remote); } #if defined(_PC_REC_INCR_XFER_SIZE) && defined(_PC_REC_MAX_XFER_SIZE) &&\ defined(_PC_REC_MIN_XFER_SIZE) && defined(_PC_REC_XFER_ALIGN) static int get_xfer_size(struct tree *t, int fd, const char *path) { t->current_filesystem->xfer_align = -1; errno = 0; if (fd >= 0) { t->current_filesystem->incr_xfer_size = fpathconf(fd, _PC_REC_INCR_XFER_SIZE); t->current_filesystem->max_xfer_size = fpathconf(fd, _PC_REC_MAX_XFER_SIZE); t->current_filesystem->min_xfer_size = fpathconf(fd, _PC_REC_MIN_XFER_SIZE); t->current_filesystem->xfer_align = fpathconf(fd, _PC_REC_XFER_ALIGN); } else if (path != NULL) { t->current_filesystem->incr_xfer_size = pathconf(path, _PC_REC_INCR_XFER_SIZE); t->current_filesystem->max_xfer_size = pathconf(path, _PC_REC_MAX_XFER_SIZE); t->current_filesystem->min_xfer_size = pathconf(path, _PC_REC_MIN_XFER_SIZE); t->current_filesystem->xfer_align = pathconf(path, _PC_REC_XFER_ALIGN); } /* At least we need an alignment size. */ if (t->current_filesystem->xfer_align == -1) return ((errno == EINVAL)?1:-1); else return (0); } #else static int get_xfer_size(struct tree *t, int fd, const char *path) { (void)t; /* UNUSED */ (void)fd; /* UNUSED */ (void)path; /* UNUSED */ return (1);/* Not supported */ } #endif #if defined(HAVE_STATFS) && defined(HAVE_FSTATFS) && defined(MNT_LOCAL) \ && !defined(ST_LOCAL) /* * Gather current filesystem properties on FreeBSD, OpenBSD and Mac OS X. */ static int setup_current_filesystem(struct archive_read_disk *a) { struct tree *t = a->tree; struct statfs sfs; #if defined(HAVE_GETVFSBYNAME) && defined(VFCF_SYNTHETIC) /* TODO: configure should set GETVFSBYNAME_ARG_TYPE to make * this accurate; some platforms have both and we need the one that's * used by getvfsbyname() * * Then the following would become: * #if defined(GETVFSBYNAME_ARG_TYPE) * GETVFSBYNAME_ARG_TYPE vfc; * #endif */ # if defined(HAVE_STRUCT_XVFSCONF) struct xvfsconf vfc; # else struct vfsconf vfc; # endif #endif int r, xr = 0; #if !defined(HAVE_STRUCT_STATFS_F_NAMEMAX) long nm; #endif t->current_filesystem->synthetic = -1; t->current_filesystem->remote = -1; if (tree_current_is_symblic_link_target(t)) { #if defined(HAVE_OPENAT) /* * Get file system statistics on any directory * where current is. */ int fd = openat(tree_current_dir_fd(t), tree_current_access_path(t), O_RDONLY | O_CLOEXEC); __archive_ensure_cloexec_flag(fd); if (fd < 0) { archive_set_error(&a->archive, errno, "openat failed"); return (ARCHIVE_FAILED); } r = fstatfs(fd, &sfs); if (r == 0) xr = get_xfer_size(t, fd, NULL); close(fd); #else if (tree_enter_working_dir(t) != 0) { archive_set_error(&a->archive, errno, "fchdir failed"); return (ARCHIVE_FAILED); } r = statfs(tree_current_access_path(t), &sfs); if (r == 0) xr = get_xfer_size(t, -1, tree_current_access_path(t)); #endif } else { r = fstatfs(tree_current_dir_fd(t), &sfs); if (r == 0) xr = get_xfer_size(t, tree_current_dir_fd(t), NULL); } if (r == -1 || xr == -1) { archive_set_error(&a->archive, errno, "statfs failed"); return (ARCHIVE_FAILED); } else if (xr == 1) { /* pathconf(_PC_REX_*) operations are not supported. */ t->current_filesystem->xfer_align = sfs.f_bsize; t->current_filesystem->max_xfer_size = -1; t->current_filesystem->min_xfer_size = sfs.f_iosize; t->current_filesystem->incr_xfer_size = sfs.f_iosize; } if (sfs.f_flags & MNT_LOCAL) t->current_filesystem->remote = 0; else t->current_filesystem->remote = 1; #if defined(HAVE_GETVFSBYNAME) && defined(VFCF_SYNTHETIC) r = getvfsbyname(sfs.f_fstypename, &vfc); if (r == -1) { archive_set_error(&a->archive, errno, "getvfsbyname failed"); return (ARCHIVE_FAILED); } if (vfc.vfc_flags & VFCF_SYNTHETIC) t->current_filesystem->synthetic = 1; else t->current_filesystem->synthetic = 0; #endif #if defined(MNT_NOATIME) if (sfs.f_flags & MNT_NOATIME) t->current_filesystem->noatime = 1; else #endif t->current_filesystem->noatime = 0; #if defined(USE_READDIR_R) /* Set maximum filename length. */ #if defined(HAVE_STRUCT_STATFS_F_NAMEMAX) t->current_filesystem->name_max = sfs.f_namemax; #else # if defined(_PC_NAME_MAX) /* Mac OS X does not have f_namemax in struct statfs. */ if (tree_current_is_symblic_link_target(t)) { if (tree_enter_working_dir(t) != 0) { archive_set_error(&a->archive, errno, "fchdir failed"); return (ARCHIVE_FAILED); } nm = pathconf(tree_current_access_path(t), _PC_NAME_MAX); } else nm = fpathconf(tree_current_dir_fd(t), _PC_NAME_MAX); # else nm = -1; # endif if (nm == -1) t->current_filesystem->name_max = NAME_MAX; else t->current_filesystem->name_max = nm; #endif #endif /* USE_READDIR_R */ return (ARCHIVE_OK); } #elif (defined(HAVE_STATVFS) || defined(HAVE_FSTATVFS)) && defined(ST_LOCAL) /* * Gather current filesystem properties on NetBSD */ static int setup_current_filesystem(struct archive_read_disk *a) { struct tree *t = a->tree; struct statvfs sfs; int r, xr = 0; t->current_filesystem->synthetic = -1; if (tree_enter_working_dir(t) != 0) { archive_set_error(&a->archive, errno, "fchdir failed"); return (ARCHIVE_FAILED); } if (tree_current_is_symblic_link_target(t)) { r = statvfs(tree_current_access_path(t), &sfs); if (r == 0) xr = get_xfer_size(t, -1, tree_current_access_path(t)); } else { #ifdef HAVE_FSTATVFS r = fstatvfs(tree_current_dir_fd(t), &sfs); if (r == 0) xr = get_xfer_size(t, tree_current_dir_fd(t), NULL); #else r = statvfs(".", &sfs); if (r == 0) xr = get_xfer_size(t, -1, "."); #endif } if (r == -1 || xr == -1) { t->current_filesystem->remote = -1; archive_set_error(&a->archive, errno, "statvfs failed"); return (ARCHIVE_FAILED); } else if (xr == 1) { - /* Usuall come here unless NetBSD supports _PC_REC_XFER_ALIGN + /* Usually come here unless NetBSD supports _PC_REC_XFER_ALIGN * for pathconf() function. */ t->current_filesystem->xfer_align = sfs.f_frsize; t->current_filesystem->max_xfer_size = -1; #if defined(HAVE_STRUCT_STATVFS_F_IOSIZE) t->current_filesystem->min_xfer_size = sfs.f_iosize; t->current_filesystem->incr_xfer_size = sfs.f_iosize; #else t->current_filesystem->min_xfer_size = sfs.f_bsize; t->current_filesystem->incr_xfer_size = sfs.f_bsize; #endif } if (sfs.f_flag & ST_LOCAL) t->current_filesystem->remote = 0; else t->current_filesystem->remote = 1; #if defined(ST_NOATIME) if (sfs.f_flag & ST_NOATIME) t->current_filesystem->noatime = 1; else #endif t->current_filesystem->noatime = 0; /* Set maximum filename length. */ t->current_filesystem->name_max = sfs.f_namemax; return (ARCHIVE_OK); } #elif defined(HAVE_SYS_STATFS_H) && defined(HAVE_LINUX_MAGIC_H) &&\ defined(HAVE_STATFS) && defined(HAVE_FSTATFS) /* * Note: statfs is deprecated since LSB 3.2 */ #ifndef CIFS_SUPER_MAGIC #define CIFS_SUPER_MAGIC 0xFF534D42 #endif #ifndef DEVFS_SUPER_MAGIC #define DEVFS_SUPER_MAGIC 0x1373 #endif /* * Gather current filesystem properties on Linux */ static int setup_current_filesystem(struct archive_read_disk *a) { struct tree *t = a->tree; struct statfs sfs; #if defined(HAVE_STATVFS) struct statvfs svfs; #endif int r, vr = 0, xr = 0; if (tree_current_is_symblic_link_target(t)) { #if defined(HAVE_OPENAT) /* * Get file system statistics on any directory * where current is. */ int fd = openat(tree_current_dir_fd(t), tree_current_access_path(t), O_RDONLY | O_CLOEXEC); __archive_ensure_cloexec_flag(fd); if (fd < 0) { archive_set_error(&a->archive, errno, "openat failed"); return (ARCHIVE_FAILED); } #if defined(HAVE_FSTATVFS) vr = fstatvfs(fd, &svfs);/* for f_flag, mount flags */ #endif r = fstatfs(fd, &sfs); if (r == 0) xr = get_xfer_size(t, fd, NULL); close(fd); #else if (tree_enter_working_dir(t) != 0) { archive_set_error(&a->archive, errno, "fchdir failed"); return (ARCHIVE_FAILED); } #if defined(HAVE_STATVFS) vr = statvfs(tree_current_access_path(t), &svfs); #endif r = statfs(tree_current_access_path(t), &sfs); if (r == 0) xr = get_xfer_size(t, -1, tree_current_access_path(t)); #endif } else { #ifdef HAVE_FSTATFS #if defined(HAVE_FSTATVFS) vr = fstatvfs(tree_current_dir_fd(t), &svfs); #endif r = fstatfs(tree_current_dir_fd(t), &sfs); if (r == 0) xr = get_xfer_size(t, tree_current_dir_fd(t), NULL); #else if (tree_enter_working_dir(t) != 0) { archive_set_error(&a->archive, errno, "fchdir failed"); return (ARCHIVE_FAILED); } #if defined(HAVE_STATVFS) vr = statvfs(".", &svfs); #endif r = statfs(".", &sfs); if (r == 0) xr = get_xfer_size(t, -1, "."); #endif } if (r == -1 || xr == -1 || vr == -1) { t->current_filesystem->synthetic = -1; t->current_filesystem->remote = -1; archive_set_error(&a->archive, errno, "statfs failed"); return (ARCHIVE_FAILED); } else if (xr == 1) { /* pathconf(_PC_REX_*) operations are not supported. */ #if defined(HAVE_STATVFS) t->current_filesystem->xfer_align = svfs.f_frsize; t->current_filesystem->max_xfer_size = -1; t->current_filesystem->min_xfer_size = svfs.f_bsize; t->current_filesystem->incr_xfer_size = svfs.f_bsize; #else t->current_filesystem->xfer_align = sfs.f_frsize; t->current_filesystem->max_xfer_size = -1; t->current_filesystem->min_xfer_size = sfs.f_bsize; t->current_filesystem->incr_xfer_size = sfs.f_bsize; #endif } switch (sfs.f_type) { case AFS_SUPER_MAGIC: case CIFS_SUPER_MAGIC: case CODA_SUPER_MAGIC: case NCP_SUPER_MAGIC:/* NetWare */ case NFS_SUPER_MAGIC: case SMB_SUPER_MAGIC: t->current_filesystem->remote = 1; t->current_filesystem->synthetic = 0; break; case DEVFS_SUPER_MAGIC: case PROC_SUPER_MAGIC: case USBDEVICE_SUPER_MAGIC: t->current_filesystem->remote = 0; t->current_filesystem->synthetic = 1; break; default: t->current_filesystem->remote = 0; t->current_filesystem->synthetic = 0; break; } #if defined(ST_NOATIME) #if defined(HAVE_STATVFS) if (svfs.f_flag & ST_NOATIME) #else if (sfs.f_flag & ST_NOATIME) #endif t->current_filesystem->noatime = 1; else #endif t->current_filesystem->noatime = 0; #if defined(USE_READDIR_R) /* Set maximum filename length. */ t->current_filesystem->name_max = sfs.f_namelen; #endif return (ARCHIVE_OK); } #elif defined(HAVE_SYS_STATVFS_H) &&\ (defined(HAVE_STATVFS) || defined(HAVE_FSTATVFS)) /* * Gather current filesystem properties on other posix platform. */ static int setup_current_filesystem(struct archive_read_disk *a) { struct tree *t = a->tree; struct statvfs sfs; int r, xr = 0; t->current_filesystem->synthetic = -1;/* Not supported */ t->current_filesystem->remote = -1;/* Not supported */ if (tree_current_is_symblic_link_target(t)) { #if defined(HAVE_OPENAT) /* * Get file system statistics on any directory * where current is. */ int fd = openat(tree_current_dir_fd(t), tree_current_access_path(t), O_RDONLY | O_CLOEXEC); __archive_ensure_cloexec_flag(fd); if (fd < 0) { archive_set_error(&a->archive, errno, "openat failed"); return (ARCHIVE_FAILED); } r = fstatvfs(fd, &sfs); if (r == 0) xr = get_xfer_size(t, fd, NULL); close(fd); #else if (tree_enter_working_dir(t) != 0) { archive_set_error(&a->archive, errno, "fchdir failed"); return (ARCHIVE_FAILED); } r = statvfs(tree_current_access_path(t), &sfs); if (r == 0) xr = get_xfer_size(t, -1, tree_current_access_path(t)); #endif } else { #ifdef HAVE_FSTATVFS r = fstatvfs(tree_current_dir_fd(t), &sfs); if (r == 0) xr = get_xfer_size(t, tree_current_dir_fd(t), NULL); #else if (tree_enter_working_dir(t) != 0) { archive_set_error(&a->archive, errno, "fchdir failed"); return (ARCHIVE_FAILED); } r = statvfs(".", &sfs); if (r == 0) xr = get_xfer_size(t, -1, "."); #endif } if (r == -1 || xr == -1) { t->current_filesystem->synthetic = -1; t->current_filesystem->remote = -1; archive_set_error(&a->archive, errno, "statvfs failed"); return (ARCHIVE_FAILED); } else if (xr == 1) { /* pathconf(_PC_REX_*) operations are not supported. */ t->current_filesystem->xfer_align = sfs.f_frsize; t->current_filesystem->max_xfer_size = -1; t->current_filesystem->min_xfer_size = sfs.f_bsize; t->current_filesystem->incr_xfer_size = sfs.f_bsize; } #if defined(ST_NOATIME) if (sfs.f_flag & ST_NOATIME) t->current_filesystem->noatime = 1; else #endif t->current_filesystem->noatime = 0; #if defined(USE_READDIR_R) /* Set maximum filename length. */ t->current_filesystem->name_max = sfs.f_namemax; #endif return (ARCHIVE_OK); } #else /* * Generic: Gather current filesystem properties. * TODO: Is this generic function really needed? */ static int setup_current_filesystem(struct archive_read_disk *a) { struct tree *t = a->tree; #if defined(_PC_NAME_MAX) && defined(USE_READDIR_R) long nm; #endif t->current_filesystem->synthetic = -1;/* Not supported */ t->current_filesystem->remote = -1;/* Not supported */ t->current_filesystem->noatime = 0; (void)get_xfer_size(t, -1, ".");/* Dummy call to avoid build error. */ t->current_filesystem->xfer_align = -1;/* Unknown */ t->current_filesystem->max_xfer_size = -1; t->current_filesystem->min_xfer_size = -1; t->current_filesystem->incr_xfer_size = -1; #if defined(USE_READDIR_R) /* Set maximum filename length. */ # if defined(_PC_NAME_MAX) if (tree_current_is_symblic_link_target(t)) { if (tree_enter_working_dir(t) != 0) { archive_set_error(&a->archive, errno, "fchdir failed"); return (ARCHIVE_FAILED); } nm = pathconf(tree_current_access_path(t), _PC_NAME_MAX); } else nm = fpathconf(tree_current_dir_fd(t), _PC_NAME_MAX); if (nm == -1) # endif /* _PC_NAME_MAX */ /* - * Some sysmtes (HP-UX or others?) incorrectly defined + * Some systems (HP-UX or others?) incorrectly defined * NAME_MAX macro to be a smaller value. */ # if defined(NAME_MAX) && NAME_MAX >= 255 t->current_filesystem->name_max = NAME_MAX; # else /* No way to get a trusted value of maximum filename * length. */ t->current_filesystem->name_max = PATH_MAX; # endif /* NAME_MAX */ # if defined(_PC_NAME_MAX) else t->current_filesystem->name_max = nm; # endif /* _PC_NAME_MAX */ #endif /* USE_READDIR_R */ return (ARCHIVE_OK); } #endif static int close_and_restore_time(int fd, struct tree *t, struct restore_time *rt) { #ifndef HAVE_UTIMES (void)t; /* UNUSED */ (void)rt; /* UNUSED */ return (close(fd)); #else #if defined(HAVE_FUTIMENS) && !defined(__CYGWIN__) struct timespec timespecs[2]; #endif struct timeval times[2]; if ((t->flags & needsRestoreTimes) == 0 || rt->noatime) { if (fd >= 0) return (close(fd)); else return (0); } #if defined(HAVE_FUTIMENS) && !defined(__CYGWIN__) timespecs[1].tv_sec = rt->mtime; timespecs[1].tv_nsec = rt->mtime_nsec; timespecs[0].tv_sec = rt->atime; timespecs[0].tv_nsec = rt->atime_nsec; /* futimens() is defined in POSIX.1-2008. */ if (futimens(fd, timespecs) == 0) return (close(fd)); #endif times[1].tv_sec = rt->mtime; times[1].tv_usec = rt->mtime_nsec / 1000; times[0].tv_sec = rt->atime; times[0].tv_usec = rt->atime_nsec / 1000; #if !defined(HAVE_FUTIMENS) && defined(HAVE_FUTIMES) && !defined(__CYGWIN__) if (futimes(fd, times) == 0) return (close(fd)); #endif close(fd); #if defined(HAVE_FUTIMESAT) if (futimesat(tree_current_dir_fd(t), rt->name, times) == 0) return (0); #endif #ifdef HAVE_LUTIMES if (lutimes(rt->name, times) != 0) #else if (AE_IFLNK != rt->filetype && utimes(rt->name, times) != 0) #endif return (-1); #endif return (0); } static int open_on_current_dir(struct tree *t, const char *path, int flags) { #ifdef HAVE_OPENAT return (openat(tree_current_dir_fd(t), path, flags)); #else if (tree_enter_working_dir(t) != 0) return (-1); return (open(path, flags)); #endif } static int tree_dup(int fd) { int new_fd; #ifdef F_DUPFD_CLOEXEC static volatile int can_dupfd_cloexec = 1; if (can_dupfd_cloexec) { new_fd = fcntl(fd, F_DUPFD_CLOEXEC, 0); if (new_fd != -1) return (new_fd); /* Linux 2.6.18 - 2.6.23 declare F_DUPFD_CLOEXEC, * but it cannot be used. So we have to try dup(). */ /* We won't try F_DUPFD_CLOEXEC. */ can_dupfd_cloexec = 0; } #endif /* F_DUPFD_CLOEXEC */ new_fd = dup(fd); __archive_ensure_cloexec_flag(new_fd); return (new_fd); } /* * Add a directory path to the current stack. */ static void tree_push(struct tree *t, const char *path, int filesystem_id, int64_t dev, int64_t ino, struct restore_time *rt) { struct tree_entry *te; te = calloc(1, sizeof(*te)); te->next = t->stack; te->parent = t->current; if (te->parent) te->depth = te->parent->depth + 1; t->stack = te; archive_string_init(&te->name); te->symlink_parent_fd = -1; archive_strcpy(&te->name, path); te->flags = needsDescent | needsOpen | needsAscent; te->filesystem_id = filesystem_id; te->dev = dev; te->ino = ino; te->dirname_length = t->dirname_length; te->restore_time.name = te->name.s; if (rt != NULL) { te->restore_time.mtime = rt->mtime; te->restore_time.mtime_nsec = rt->mtime_nsec; te->restore_time.atime = rt->atime; te->restore_time.atime_nsec = rt->atime_nsec; te->restore_time.filetype = rt->filetype; te->restore_time.noatime = rt->noatime; } } /* * Append a name to the current dir path. */ static void tree_append(struct tree *t, const char *name, size_t name_length) { size_t size_needed; t->path.s[t->dirname_length] = '\0'; t->path.length = t->dirname_length; /* Strip trailing '/' from name, unless entire name is "/". */ while (name_length > 1 && name[name_length - 1] == '/') name_length--; /* Resize pathname buffer as needed. */ size_needed = name_length + t->dirname_length + 2; archive_string_ensure(&t->path, size_needed); /* Add a separating '/' if it's needed. */ if (t->dirname_length > 0 && t->path.s[archive_strlen(&t->path)-1] != '/') archive_strappend_char(&t->path, '/'); t->basename = t->path.s + archive_strlen(&t->path); archive_strncat(&t->path, name, name_length); t->restore_time.name = t->basename; } /* * Open a directory tree for traversal. */ static struct tree * tree_open(const char *path, int symlink_mode, int restore_time) { struct tree *t; if ((t = calloc(1, sizeof(*t))) == NULL) return (NULL); archive_string_init(&t->path); archive_string_ensure(&t->path, 31); t->initial_symlink_mode = symlink_mode; return (tree_reopen(t, path, restore_time)); } static struct tree * tree_reopen(struct tree *t, const char *path, int restore_time) { t->flags = (restore_time)?needsRestoreTimes:0; t->flags |= onInitialDir; t->visit_type = 0; t->tree_errno = 0; t->dirname_length = 0; t->depth = 0; t->descend = 0; t->current = NULL; t->d = INVALID_DIR_HANDLE; t->symlink_mode = t->initial_symlink_mode; archive_string_empty(&t->path); t->entry_fd = -1; t->entry_eof = 0; t->entry_remaining_bytes = 0; t->initial_filesystem_id = -1; /* First item is set up a lot like a symlink traversal. */ tree_push(t, path, 0, 0, 0, NULL); t->stack->flags = needsFirstVisit; t->maxOpenCount = t->openCount = 1; t->initial_dir_fd = open(".", O_RDONLY | O_CLOEXEC); __archive_ensure_cloexec_flag(t->initial_dir_fd); t->working_dir_fd = tree_dup(t->initial_dir_fd); return (t); } static int tree_descent(struct tree *t) { int flag, new_fd, r = 0; t->dirname_length = archive_strlen(&t->path); flag = O_RDONLY | O_CLOEXEC; #if defined(O_DIRECTORY) flag |= O_DIRECTORY; #endif new_fd = open_on_current_dir(t, t->stack->name.s, flag); __archive_ensure_cloexec_flag(new_fd); if (new_fd < 0) { t->tree_errno = errno; r = TREE_ERROR_DIR; } else { t->depth++; /* If it is a link, set up fd for the ascent. */ if (t->stack->flags & isDirLink) { t->stack->symlink_parent_fd = t->working_dir_fd; t->openCount++; if (t->openCount > t->maxOpenCount) t->maxOpenCount = t->openCount; } else close(t->working_dir_fd); /* Renew the current working directory. */ t->working_dir_fd = new_fd; t->flags &= ~onWorkingDir; } return (r); } /* * We've finished a directory; ascend back to the parent. */ static int tree_ascend(struct tree *t) { struct tree_entry *te; int new_fd, r = 0, prev_dir_fd; te = t->stack; prev_dir_fd = t->working_dir_fd; if (te->flags & isDirLink) new_fd = te->symlink_parent_fd; else { new_fd = open_on_current_dir(t, "..", O_RDONLY | O_CLOEXEC); __archive_ensure_cloexec_flag(new_fd); } if (new_fd < 0) { t->tree_errno = errno; r = TREE_ERROR_FATAL; } else { /* Renew the current working directory. */ t->working_dir_fd = new_fd; t->flags &= ~onWorkingDir; /* Current directory has been changed, we should * close an fd of previous working directory. */ close_and_restore_time(prev_dir_fd, t, &te->restore_time); if (te->flags & isDirLink) { t->openCount--; te->symlink_parent_fd = -1; } t->depth--; } return (r); } /* * Return to the initial directory where tree_open() was performed. */ static int tree_enter_initial_dir(struct tree *t) { int r = 0; if ((t->flags & onInitialDir) == 0) { r = fchdir(t->initial_dir_fd); if (r == 0) { t->flags &= ~onWorkingDir; t->flags |= onInitialDir; } } return (r); } /* * Restore working directory of directory traversals. */ static int tree_enter_working_dir(struct tree *t) { int r = 0; /* * Change the current directory if really needed. * Sometimes this is unneeded when we did not do * descent. */ if (t->depth > 0 && (t->flags & onWorkingDir) == 0) { r = fchdir(t->working_dir_fd); if (r == 0) { t->flags &= ~onInitialDir; t->flags |= onWorkingDir; } } return (r); } static int tree_current_dir_fd(struct tree *t) { return (t->working_dir_fd); } /* * Pop the working stack. */ static void tree_pop(struct tree *t) { struct tree_entry *te; t->path.s[t->dirname_length] = '\0'; t->path.length = t->dirname_length; if (t->stack == t->current && t->current != NULL) t->current = t->current->parent; te = t->stack; t->stack = te->next; t->dirname_length = te->dirname_length; t->basename = t->path.s + t->dirname_length; while (t->basename[0] == '/') t->basename++; archive_string_free(&te->name); free(te); } /* * Get the next item in the tree traversal. */ static int tree_next(struct tree *t) { int r; while (t->stack != NULL) { /* If there's an open dir, get the next entry from there. */ if (t->d != INVALID_DIR_HANDLE) { r = tree_dir_next_posix(t); if (r == 0) continue; return (r); } if (t->stack->flags & needsFirstVisit) { /* Top stack item needs a regular visit. */ t->current = t->stack; tree_append(t, t->stack->name.s, archive_strlen(&(t->stack->name))); /* t->dirname_length = t->path_length; */ /* tree_pop(t); */ t->stack->flags &= ~needsFirstVisit; return (t->visit_type = TREE_REGULAR); } else if (t->stack->flags & needsDescent) { /* Top stack item is dir to descend into. */ t->current = t->stack; tree_append(t, t->stack->name.s, archive_strlen(&(t->stack->name))); t->stack->flags &= ~needsDescent; r = tree_descent(t); if (r != 0) { tree_pop(t); t->visit_type = r; } else t->visit_type = TREE_POSTDESCENT; return (t->visit_type); } else if (t->stack->flags & needsOpen) { t->stack->flags &= ~needsOpen; r = tree_dir_next_posix(t); if (r == 0) continue; return (r); } else if (t->stack->flags & needsAscent) { /* Top stack item is dir and we're done with it. */ r = tree_ascend(t); tree_pop(t); t->visit_type = r != 0 ? r : TREE_POSTASCENT; return (t->visit_type); } else { /* Top item on stack is dead. */ tree_pop(t); t->flags &= ~hasLstat; t->flags &= ~hasStat; } } return (t->visit_type = 0); } static int tree_dir_next_posix(struct tree *t) { int r; const char *name; size_t namelen; if (t->d == NULL) { #if defined(USE_READDIR_R) size_t dirent_size; #endif #if defined(HAVE_FDOPENDIR) t->d = fdopendir(tree_dup(t->working_dir_fd)); #else /* HAVE_FDOPENDIR */ if (tree_enter_working_dir(t) == 0) { t->d = opendir("."); #if HAVE_DIRFD || defined(dirfd) __archive_ensure_cloexec_flag(dirfd(t->d)); #endif } #endif /* HAVE_FDOPENDIR */ if (t->d == NULL) { r = tree_ascend(t); /* Undo "chdir" */ tree_pop(t); t->tree_errno = errno; t->visit_type = r != 0 ? r : TREE_ERROR_DIR; return (t->visit_type); } #if defined(USE_READDIR_R) dirent_size = offsetof(struct dirent, d_name) + t->filesystem_table[t->current->filesystem_id].name_max + 1; if (t->dirent == NULL || t->dirent_allocated < dirent_size) { free(t->dirent); t->dirent = malloc(dirent_size); if (t->dirent == NULL) { closedir(t->d); t->d = INVALID_DIR_HANDLE; (void)tree_ascend(t); tree_pop(t); t->tree_errno = ENOMEM; t->visit_type = TREE_ERROR_DIR; return (t->visit_type); } t->dirent_allocated = dirent_size; } #endif /* USE_READDIR_R */ } for (;;) { errno = 0; #if defined(USE_READDIR_R) r = readdir_r(t->d, t->dirent, &t->de); #ifdef _AIX /* Note: According to the man page, return value 9 indicates * that the readdir_r was not successful and the error code * is set to the global errno variable. And then if the end * of directory entries was reached, the return value is 9 * and the third parameter is set to NULL and errno is * unchanged. */ if (r == 9) r = errno; #endif /* _AIX */ if (r != 0 || t->de == NULL) { #else t->de = readdir(t->d); if (t->de == NULL) { r = errno; #endif closedir(t->d); t->d = INVALID_DIR_HANDLE; if (r != 0) { t->tree_errno = r; t->visit_type = TREE_ERROR_DIR; return (t->visit_type); } else return (0); } name = t->de->d_name; namelen = D_NAMELEN(t->de); t->flags &= ~hasLstat; t->flags &= ~hasStat; if (name[0] == '.' && name[1] == '\0') continue; if (name[0] == '.' && name[1] == '.' && name[2] == '\0') continue; tree_append(t, name, namelen); return (t->visit_type = TREE_REGULAR); } } /* * Get the stat() data for the entry just returned from tree_next(). */ static const struct stat * tree_current_stat(struct tree *t) { if (!(t->flags & hasStat)) { #ifdef HAVE_FSTATAT if (fstatat(tree_current_dir_fd(t), tree_current_access_path(t), &t->st, 0) != 0) #else if (tree_enter_working_dir(t) != 0) return NULL; if (stat(tree_current_access_path(t), &t->st) != 0) #endif return NULL; t->flags |= hasStat; } return (&t->st); } /* * Get the lstat() data for the entry just returned from tree_next(). */ static const struct stat * tree_current_lstat(struct tree *t) { if (!(t->flags & hasLstat)) { #ifdef HAVE_FSTATAT if (fstatat(tree_current_dir_fd(t), tree_current_access_path(t), &t->lst, AT_SYMLINK_NOFOLLOW) != 0) #else if (tree_enter_working_dir(t) != 0) return NULL; if (lstat(tree_current_access_path(t), &t->lst) != 0) #endif return NULL; t->flags |= hasLstat; } return (&t->lst); } /* * Test whether current entry is a dir or link to a dir. */ static int tree_current_is_dir(struct tree *t) { const struct stat *st; /* * If we already have lstat() info, then try some * cheap tests to determine if this is a dir. */ if (t->flags & hasLstat) { /* If lstat() says it's a dir, it must be a dir. */ st = tree_current_lstat(t); if (st == NULL) return 0; if (S_ISDIR(st->st_mode)) return 1; /* Not a dir; might be a link to a dir. */ /* If it's not a link, then it's not a link to a dir. */ if (!S_ISLNK(st->st_mode)) return 0; /* * It's a link, but we don't know what it's a link to, * so we'll have to use stat(). */ } st = tree_current_stat(t); /* If we can't stat it, it's not a dir. */ if (st == NULL) return 0; /* Use the definitive test. Hopefully this is cached. */ return (S_ISDIR(st->st_mode)); } /* * Test whether current entry is a physical directory. Usually, we * already have at least one of stat() or lstat() in memory, so we * use tricks to try to avoid an extra trip to the disk. */ static int tree_current_is_physical_dir(struct tree *t) { const struct stat *st; /* * If stat() says it isn't a dir, then it's not a dir. * If stat() data is cached, this check is free, so do it first. */ if (t->flags & hasStat) { st = tree_current_stat(t); if (st == NULL) return (0); if (!S_ISDIR(st->st_mode)) return (0); } /* * Either stat() said it was a dir (in which case, we have * to determine whether it's really a link to a dir) or * stat() info wasn't available. So we use lstat(), which * hopefully is already cached. */ st = tree_current_lstat(t); /* If we can't stat it, it's not a dir. */ if (st == NULL) return 0; /* Use the definitive test. Hopefully this is cached. */ return (S_ISDIR(st->st_mode)); } /* * Test whether the same file has been in the tree as its parent. */ static int tree_target_is_same_as_parent(struct tree *t, const struct stat *st) { struct tree_entry *te; for (te = t->current->parent; te != NULL; te = te->parent) { if (te->dev == (int64_t)st->st_dev && te->ino == (int64_t)st->st_ino) return (1); } return (0); } /* * Test whether the current file is symbolic link target and * on the other filesystem. */ static int tree_current_is_symblic_link_target(struct tree *t) { static const struct stat *lst, *st; lst = tree_current_lstat(t); st = tree_current_stat(t); return (st != NULL && lst != NULL && (int64_t)st->st_dev == t->current_filesystem->dev && st->st_dev != lst->st_dev); } /* * Return the access path for the entry just returned from tree_next(). */ static const char * tree_current_access_path(struct tree *t) { return (t->basename); } /* * Return the full path for the entry just returned from tree_next(). */ static const char * tree_current_path(struct tree *t) { return (t->path.s); } /* * Terminate the traversal. */ static void tree_close(struct tree *t) { if (t == NULL) return; if (t->entry_fd >= 0) { close_and_restore_time(t->entry_fd, t, &t->restore_time); t->entry_fd = -1; } /* Close the handle of readdir(). */ if (t->d != INVALID_DIR_HANDLE) { closedir(t->d); t->d = INVALID_DIR_HANDLE; } /* Release anything remaining in the stack. */ while (t->stack != NULL) { if (t->stack->flags & isDirLink) close(t->stack->symlink_parent_fd); tree_pop(t); } if (t->working_dir_fd >= 0) { close(t->working_dir_fd); t->working_dir_fd = -1; } if (t->initial_dir_fd >= 0) { close(t->initial_dir_fd); t->initial_dir_fd = -1; } } /* * Release any resources. */ static void tree_free(struct tree *t) { int i; if (t == NULL) return; archive_string_free(&t->path); #if defined(USE_READDIR_R) free(t->dirent); #endif free(t->sparse_list); for (i = 0; i < t->max_filesystem_id; i++) free(t->filesystem_table[i].allocation_ptr); free(t->filesystem_table); free(t); } #endif Index: head/contrib/libarchive/libarchive/archive_read_support_filter_lz4.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_filter_lz4.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_read_support_filter_lz4.c (revision 311903) @@ -1,737 +1,737 @@ /*- * Copyright (c) 2014 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_LZ4_H #include #endif #include "archive.h" #include "archive_endian.h" #include "archive_private.h" #include "archive_read_private.h" #include "archive_xxhash.h" #define LZ4_MAGICNUMBER 0x184d2204 #define LZ4_SKIPPABLED 0x184d2a50 #define LZ4_LEGACY 0x184c2102 #if defined(HAVE_LIBLZ4) struct private_data { enum { SELECT_STREAM, READ_DEFAULT_STREAM, READ_DEFAULT_BLOCK, READ_LEGACY_STREAM, READ_LEGACY_BLOCK, } stage; struct { unsigned block_independence:1; unsigned block_checksum:3; unsigned stream_size:1; unsigned stream_checksum:1; unsigned preset_dictionary:1; int block_maximum_size; } flags; int64_t stream_size; uint32_t dict_id; char *out_block; size_t out_block_size; /* Bytes read but not yet consumed via __archive_read_consume() */ size_t unconsumed; size_t decoded_size; void *xxh32_state; char valid; /* True = decompressor is initialized */ char eof; /* True = found end of compressed data. */ }; #define LEGACY_BLOCK_SIZE (8 * 1024 * 1024) /* Lz4 filter */ static ssize_t lz4_filter_read(struct archive_read_filter *, const void **); static int lz4_filter_close(struct archive_read_filter *); #endif /* * Note that we can detect lz4 archives even if we can't decompress * them. (In fact, we like detecting them because we can give better * error messages.) So the bid framework here gets compiled even * if liblz4 is unavailable. */ static int lz4_reader_bid(struct archive_read_filter_bidder *, struct archive_read_filter *); static int lz4_reader_init(struct archive_read_filter *); static int lz4_reader_free(struct archive_read_filter_bidder *); #if defined(HAVE_LIBLZ4) static ssize_t lz4_filter_read_default_stream(struct archive_read_filter *, const void **); static ssize_t lz4_filter_read_legacy_stream(struct archive_read_filter *, const void **); #endif int archive_read_support_filter_lz4(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct archive_read_filter_bidder *reader; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_filter_lz4"); if (__archive_read_get_bidder(a, &reader) != ARCHIVE_OK) return (ARCHIVE_FATAL); reader->data = NULL; reader->name = "lz4"; reader->bid = lz4_reader_bid; reader->init = lz4_reader_init; reader->options = NULL; reader->free = lz4_reader_free; #if defined(HAVE_LIBLZ4) return (ARCHIVE_OK); #else archive_set_error(_a, ARCHIVE_ERRNO_MISC, "Using external lz4 program"); return (ARCHIVE_WARN); #endif } static int lz4_reader_free(struct archive_read_filter_bidder *self){ (void)self; /* UNUSED */ return (ARCHIVE_OK); } /* * Test whether we can handle this data. * * This logic returns zero if any part of the signature fails. It * also tries to Do The Right Thing if a very short buffer prevents us * from verifying as much as we would like. */ static int lz4_reader_bid(struct archive_read_filter_bidder *self, struct archive_read_filter *filter) { const unsigned char *buffer; ssize_t avail; int bits_checked; uint32_t number; (void)self; /* UNUSED */ /* Minimal lz4 archive is 11 bytes. */ buffer = __archive_read_filter_ahead(filter, 11, &avail); if (buffer == NULL) return (0); /* First four bytes must be LZ4 magic numbers. */ bits_checked = 0; if ((number = archive_le32dec(buffer)) == LZ4_MAGICNUMBER) { unsigned char flag, BD; bits_checked += 32; /* Next follows a stream descriptor. */ /* Descriptor Flags. */ flag = buffer[4]; /* A version number must be "01". */ if (((flag & 0xc0) >> 6) != 1) return (0); /* A reserved bit must be "0". */ if (flag & 2) return (0); bits_checked += 8; BD = buffer[5]; - /* A block maximum size shuld be more than 3. */ + /* A block maximum size should be more than 3. */ if (((BD & 0x70) >> 4) < 4) return (0); /* Reserved bits must be "0". */ if (BD & ~0x70) return (0); bits_checked += 8; } else if (number == LZ4_LEGACY) { bits_checked += 32; } return (bits_checked); } #if !defined(HAVE_LIBLZ4) /* * If we don't have the library on this system, we can't actually do the * decompression. We can, however, still detect compressed archives * and emit a useful message. */ static int lz4_reader_init(struct archive_read_filter *self) { int r; r = __archive_read_program(self, "lz4 -d -q"); /* Note: We set the format here even if __archive_read_program() * above fails. We do, after all, know what the format is * even if we weren't able to read it. */ self->code = ARCHIVE_FILTER_LZ4; self->name = "lz4"; return (r); } #else /* * Setup the callbacks. */ static int lz4_reader_init(struct archive_read_filter *self) { struct private_data *state; self->code = ARCHIVE_FILTER_LZ4; self->name = "lz4"; state = (struct private_data *)calloc(sizeof(*state), 1); if (state == NULL) { archive_set_error(&self->archive->archive, ENOMEM, "Can't allocate data for lz4 decompression"); return (ARCHIVE_FATAL); } self->data = state; state->stage = SELECT_STREAM; self->read = lz4_filter_read; self->skip = NULL; /* not supported */ self->close = lz4_filter_close; return (ARCHIVE_OK); } static int lz4_allocate_out_block(struct archive_read_filter *self) { struct private_data *state = (struct private_data *)self->data; size_t out_block_size = state->flags.block_maximum_size; void *out_block; if (!state->flags.block_independence) out_block_size += 64 * 1024; if (state->out_block_size < out_block_size) { free(state->out_block); out_block = (unsigned char *)malloc(out_block_size); state->out_block_size = out_block_size; if (out_block == NULL) { archive_set_error(&self->archive->archive, ENOMEM, "Can't allocate data for lz4 decompression"); return (ARCHIVE_FATAL); } state->out_block = out_block; } if (!state->flags.block_independence) memset(state->out_block, 0, 64 * 1024); return (ARCHIVE_OK); } static int lz4_allocate_out_block_for_legacy(struct archive_read_filter *self) { struct private_data *state = (struct private_data *)self->data; size_t out_block_size = LEGACY_BLOCK_SIZE; void *out_block; if (state->out_block_size < out_block_size) { free(state->out_block); out_block = (unsigned char *)malloc(out_block_size); state->out_block_size = out_block_size; if (out_block == NULL) { archive_set_error(&self->archive->archive, ENOMEM, "Can't allocate data for lz4 decompression"); return (ARCHIVE_FATAL); } state->out_block = out_block; } return (ARCHIVE_OK); } /* * Return the next block of decompressed data. */ static ssize_t lz4_filter_read(struct archive_read_filter *self, const void **p) { struct private_data *state = (struct private_data *)self->data; ssize_t ret; if (state->eof) { *p = NULL; return (0); } __archive_read_filter_consume(self->upstream, state->unconsumed); state->unconsumed = 0; switch (state->stage) { case SELECT_STREAM: break; case READ_DEFAULT_STREAM: case READ_LEGACY_STREAM: /* Reading a lz4 stream already failed. */ archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Invalid sequence."); return (ARCHIVE_FATAL); case READ_DEFAULT_BLOCK: ret = lz4_filter_read_default_stream(self, p); if (ret != 0 || state->stage != SELECT_STREAM) return ret; break; case READ_LEGACY_BLOCK: ret = lz4_filter_read_legacy_stream(self, p); if (ret != 0 || state->stage != SELECT_STREAM) return ret; break; default: archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Program error."); return (ARCHIVE_FATAL); break; } while (state->stage == SELECT_STREAM) { const char *read_buf; /* Read a magic number. */ read_buf = __archive_read_filter_ahead(self->upstream, 4, NULL); if (read_buf == NULL) { state->eof = 1; *p = NULL; return (0); } uint32_t number = archive_le32dec(read_buf); __archive_read_filter_consume(self->upstream, 4); if (number == LZ4_MAGICNUMBER) return lz4_filter_read_default_stream(self, p); else if (number == LZ4_LEGACY) return lz4_filter_read_legacy_stream(self, p); else if ((number & ~0xF) == LZ4_SKIPPABLED) { read_buf = __archive_read_filter_ahead( self->upstream, 4, NULL); if (read_buf == NULL) { archive_set_error( &self->archive->archive, ARCHIVE_ERRNO_MISC, "Malformed lz4 data"); return (ARCHIVE_FATAL); } uint32_t skip_bytes = archive_le32dec(read_buf); __archive_read_filter_consume(self->upstream, 4 + skip_bytes); } else { /* Ignore following unrecognized data. */ state->eof = 1; *p = NULL; return (0); } } state->eof = 1; *p = NULL; return (0); } static int lz4_filter_read_descriptor(struct archive_read_filter *self) { struct private_data *state = (struct private_data *)self->data; const char *read_buf; ssize_t bytes_remaining; ssize_t descriptor_bytes; unsigned char flag, bd; unsigned int chsum, chsum_verifier; /* Make sure we have 2 bytes for flags. */ read_buf = __archive_read_filter_ahead(self->upstream, 2, &bytes_remaining); if (read_buf == NULL) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "truncated lz4 input"); return (ARCHIVE_FATAL); } /* Parse flags. */ flag = (unsigned char)read_buf[0]; /* Verify version number. */ if ((flag & 0xc0) != 1<<6) goto malformed_error; /* A reserved bit must be zero. */ if (flag & 0x02) goto malformed_error; state->flags.block_independence = (flag & 0x20) != 0; state->flags.block_checksum = (flag & 0x10)?4:0; state->flags.stream_size = (flag & 0x08) != 0; state->flags.stream_checksum = (flag & 0x04) != 0; state->flags.preset_dictionary = (flag & 0x01) != 0; /* BD */ bd = (unsigned char)read_buf[1]; /* Reserved bits must be zero. */ if (bd & 0x8f) goto malformed_error; - /* Get a maxinum block size. */ + /* Get a maximum block size. */ switch (read_buf[1] >> 4) { case 4: /* 64 KB */ state->flags.block_maximum_size = 64 * 1024; break; case 5: /* 256 KB */ state->flags.block_maximum_size = 256 * 1024; break; case 6: /* 1 MB */ state->flags.block_maximum_size = 1024 * 1024; break; case 7: /* 4 MB */ state->flags.block_maximum_size = 4 * 1024 * 1024; break; default: goto malformed_error; } /* Read the whole descriptor in a stream block. */ descriptor_bytes = 3; if (state->flags.stream_size) descriptor_bytes += 8; if (state->flags.preset_dictionary) descriptor_bytes += 4; if (bytes_remaining < descriptor_bytes) { read_buf = __archive_read_filter_ahead(self->upstream, descriptor_bytes, &bytes_remaining); if (read_buf == NULL) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "truncated lz4 input"); return (ARCHIVE_FATAL); } } /* Check if a descriptor is corrupted */ chsum = __archive_xxhash.XXH32(read_buf, (int)descriptor_bytes -1, 0); chsum = (chsum >> 8) & 0xff; chsum_verifier = read_buf[descriptor_bytes-1] & 0xff; if (chsum != chsum_verifier) goto malformed_error; __archive_read_filter_consume(self->upstream, descriptor_bytes); /* Make sure we have an enough buffer for uncompressed data. */ if (lz4_allocate_out_block(self) != ARCHIVE_OK) return (ARCHIVE_FATAL); if (state->flags.stream_checksum) state->xxh32_state = __archive_xxhash.XXH32_init(0); state->decoded_size = 0; /* Success */ return (ARCHIVE_OK); malformed_error: archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "malformed lz4 data"); return (ARCHIVE_FATAL); } static ssize_t lz4_filter_read_data_block(struct archive_read_filter *self, const void **p) { struct private_data *state = (struct private_data *)self->data; ssize_t compressed_size; const char *read_buf; ssize_t bytes_remaining; int checksum_size; ssize_t uncompressed_size; size_t prefix64k; *p = NULL; /* Make sure we have 4 bytes for a block size. */ read_buf = __archive_read_filter_ahead(self->upstream, 4, &bytes_remaining); if (read_buf == NULL) goto truncated_error; compressed_size = archive_le32dec(read_buf); if ((compressed_size & ~(1 << 31)) > state->flags.block_maximum_size) goto malformed_error; /* A compressed size == 0 means the end of stream blocks. */ if (compressed_size == 0) { __archive_read_filter_consume(self->upstream, 4); return 0; } checksum_size = state->flags.block_checksum; /* Check if the block is uncompressed. */ if (compressed_size & (1 << 31)) { compressed_size &= ~(1 << 31); uncompressed_size = compressed_size; } else uncompressed_size = 0;/* Unknown yet. */ /* Unfortunately, lz4 decompression API requires a whole block for its decompression speed, so we read a whole block and allocate a huge buffer used for decoded data. */ read_buf = __archive_read_filter_ahead(self->upstream, 4 + compressed_size + checksum_size, &bytes_remaining); if (read_buf == NULL) goto truncated_error; /* Optional process, checking a block sum. */ if (checksum_size) { unsigned int chsum = __archive_xxhash.XXH32( read_buf + 4, (int)compressed_size, 0); unsigned int chsum_block = archive_le32dec(read_buf + 4 + compressed_size); if (chsum != chsum_block) goto malformed_error; } /* If the block is uncompressed, there is nothing to do. */ if (uncompressed_size) { /* Prepare a prefix 64k block for next block. */ if (!state->flags.block_independence) { prefix64k = 64 * 1024; if (uncompressed_size < (ssize_t)prefix64k) { memcpy(state->out_block + prefix64k - uncompressed_size, read_buf + 4, uncompressed_size); memset(state->out_block, 0, prefix64k - uncompressed_size); } else { memcpy(state->out_block, read_buf + 4 + uncompressed_size - prefix64k, prefix64k); } state->decoded_size = 0; } state->unconsumed = 4 + uncompressed_size + checksum_size; *p = read_buf + 4; return uncompressed_size; } /* Decompress a block data. */ if (state->flags.block_independence) { prefix64k = 0; uncompressed_size = LZ4_decompress_safe(read_buf + 4, state->out_block, (int)compressed_size, state->flags.block_maximum_size); } else { prefix64k = 64 * 1024; if (state->decoded_size) { if (state->decoded_size < prefix64k) { memmove(state->out_block + prefix64k - state->decoded_size, state->out_block + prefix64k, state->decoded_size); memset(state->out_block, 0, prefix64k - state->decoded_size); } else { memmove(state->out_block, state->out_block + state->decoded_size, prefix64k); } } #if LZ4_VERSION_MAJOR >= 1 && LZ4_VERSION_MINOR >= 7 uncompressed_size = LZ4_decompress_safe_usingDict( read_buf + 4, state->out_block + prefix64k, (int)compressed_size, state->flags.block_maximum_size, state->out_block, prefix64k); #else uncompressed_size = LZ4_decompress_safe_withPrefix64k( read_buf + 4, state->out_block + prefix64k, (int)compressed_size, state->flags.block_maximum_size); #endif } /* Check if an error occurred in the decompression process. */ if (uncompressed_size < 0) { archive_set_error(&(self->archive->archive), ARCHIVE_ERRNO_MISC, "lz4 decompression failed"); return (ARCHIVE_FATAL); } state->unconsumed = 4 + compressed_size + checksum_size; *p = state->out_block + prefix64k; state->decoded_size = uncompressed_size; return uncompressed_size; malformed_error: archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "malformed lz4 data"); return (ARCHIVE_FATAL); truncated_error: archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "truncated lz4 input"); return (ARCHIVE_FATAL); } static ssize_t lz4_filter_read_default_stream(struct archive_read_filter *self, const void **p) { struct private_data *state = (struct private_data *)self->data; const char *read_buf; ssize_t bytes_remaining; ssize_t ret; if (state->stage == SELECT_STREAM) { state->stage = READ_DEFAULT_STREAM; - /* First, read a desciprtor. */ + /* First, read a descriptor. */ if((ret = lz4_filter_read_descriptor(self)) != ARCHIVE_OK) return (ret); state->stage = READ_DEFAULT_BLOCK; } /* Decompress a block. */ ret = lz4_filter_read_data_block(self, p); /* If the end of block is detected, change the filter status to read next stream. */ if (ret == 0 && *p == NULL) state->stage = SELECT_STREAM; /* Optional process, checking a stream sum. */ if (state->flags.stream_checksum) { if (state->stage == SELECT_STREAM) { unsigned int checksum; unsigned int checksum_stream; read_buf = __archive_read_filter_ahead(self->upstream, 4, &bytes_remaining); if (read_buf == NULL) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "truncated lz4 input"); return (ARCHIVE_FATAL); } checksum = archive_le32dec(read_buf); __archive_read_filter_consume(self->upstream, 4); checksum_stream = __archive_xxhash.XXH32_digest( state->xxh32_state); state->xxh32_state = NULL; if (checksum != checksum_stream) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "lz4 stream cheksum error"); return (ARCHIVE_FATAL); } } else if (ret > 0) __archive_xxhash.XXH32_update(state->xxh32_state, *p, (int)ret); } return (ret); } static ssize_t lz4_filter_read_legacy_stream(struct archive_read_filter *self, const void **p) { struct private_data *state = (struct private_data *)self->data; int compressed; const char *read_buf; ssize_t ret; *p = NULL; ret = lz4_allocate_out_block_for_legacy(self); if (ret != ARCHIVE_OK) return ret; /* Make sure we have 4 bytes for a block size. */ read_buf = __archive_read_filter_ahead(self->upstream, 4, NULL); if (read_buf == NULL) { if (state->stage == SELECT_STREAM) { state->stage = READ_LEGACY_STREAM; archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "truncated lz4 input"); return (ARCHIVE_FATAL); } state->stage = SELECT_STREAM; return 0; } state->stage = READ_LEGACY_BLOCK; compressed = archive_le32dec(read_buf); if (compressed > LZ4_COMPRESSBOUND(LEGACY_BLOCK_SIZE)) { state->stage = SELECT_STREAM; return 0; } /* Make sure we have a whole block. */ read_buf = __archive_read_filter_ahead(self->upstream, 4 + compressed, NULL); ret = LZ4_decompress_safe(read_buf + 4, state->out_block, compressed, (int)state->out_block_size); if (ret < 0) { archive_set_error(&(self->archive->archive), ARCHIVE_ERRNO_MISC, "lz4 decompression failed"); return (ARCHIVE_FATAL); } *p = state->out_block; state->unconsumed = 4 + compressed; return ret; } /* * Clean up the decompressor. */ static int lz4_filter_close(struct archive_read_filter *self) { struct private_data *state; int ret = ARCHIVE_OK; state = (struct private_data *)self->data; free(state->xxh32_state); free(state->out_block); free(state); return (ret); } #endif /* HAVE_LIBLZ4 */ Index: head/contrib/libarchive/libarchive/archive_read_support_filter_lzop.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_filter_lzop.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_read_support_filter_lzop.c (revision 311903) @@ -1,494 +1,494 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * Copyright (c) 2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_LZO_LZOCONF_H #include #endif #ifdef HAVE_LZO_LZO1X_H #include #endif #ifdef HAVE_ZLIB_H #include /* for crc32 and adler32 */ #endif #include "archive.h" #if !defined(HAVE_ZLIB_H) &&\ defined(HAVE_LZO_LZOCONF_H) && defined(HAVE_LZO_LZO1X_H) #include "archive_crc32.h" #endif #include "archive_endian.h" #include "archive_private.h" #include "archive_read_private.h" #ifndef HAVE_ZLIB_H #define adler32 lzo_adler32 #endif #define LZOP_HEADER_MAGIC "\x89\x4c\x5a\x4f\x00\x0d\x0a\x1a\x0a" #define LZOP_HEADER_MAGIC_LEN 9 #if defined(HAVE_LZO_LZOCONF_H) && defined(HAVE_LZO_LZO1X_H) struct read_lzop { unsigned char *out_block; size_t out_block_size; int64_t total_out; int flags; uint32_t compressed_cksum; uint32_t uncompressed_cksum; size_t compressed_size; size_t uncompressed_size; size_t unconsumed_bytes; char in_stream; char eof; /* True = found end of compressed data. */ }; #define FILTER 0x0800 #define CRC32_HEADER 0x1000 #define EXTRA_FIELD 0x0040 #define ADLER32_UNCOMPRESSED 0x0001 #define ADLER32_COMPRESSED 0x0002 #define CRC32_UNCOMPRESSED 0x0100 #define CRC32_COMPRESSED 0x0200 #define MAX_BLOCK_SIZE (64 * 1024 * 1024) static ssize_t lzop_filter_read(struct archive_read_filter *, const void **); static int lzop_filter_close(struct archive_read_filter *); #endif static int lzop_bidder_bid(struct archive_read_filter_bidder *, struct archive_read_filter *); static int lzop_bidder_init(struct archive_read_filter *); int archive_read_support_filter_lzop(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct archive_read_filter_bidder *reader; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_filter_lzop"); if (__archive_read_get_bidder(a, &reader) != ARCHIVE_OK) return (ARCHIVE_FATAL); reader->data = NULL; reader->bid = lzop_bidder_bid; reader->init = lzop_bidder_init; reader->options = NULL; reader->free = NULL; /* Signal the extent of lzop support with the return value here. */ #if defined(HAVE_LZO_LZOCONF_H) && defined(HAVE_LZO_LZO1X_H) return (ARCHIVE_OK); #else /* Return ARCHIVE_WARN since this always uses an external program. */ archive_set_error(_a, ARCHIVE_ERRNO_MISC, "Using external lzop program for lzop decompression"); return (ARCHIVE_WARN); #endif } /* * Bidder just verifies the header and returns the number of verified bits. */ static int lzop_bidder_bid(struct archive_read_filter_bidder *self, struct archive_read_filter *filter) { const unsigned char *p; ssize_t avail; (void)self; /* UNUSED */ p = __archive_read_filter_ahead(filter, LZOP_HEADER_MAGIC_LEN, &avail); if (p == NULL || avail == 0) return (0); if (memcmp(p, LZOP_HEADER_MAGIC, LZOP_HEADER_MAGIC_LEN)) return (0); return (LZOP_HEADER_MAGIC_LEN * 8); } #if !defined(HAVE_LZO_LZOCONF_H) || !defined(HAVE_LZO_LZO1X_H) /* * If we don't have the library on this system, we can't do the * decompression directly. We can, however, try to run "lzop -d" * in case that's available. */ static int lzop_bidder_init(struct archive_read_filter *self) { int r; r = __archive_read_program(self, "lzop -d"); /* Note: We set the format here even if __archive_read_program() * above fails. We do, after all, know what the format is * even if we weren't able to read it. */ self->code = ARCHIVE_FILTER_LZOP; self->name = "lzop"; return (r); } #else /* * Initialize the filter object. */ static int lzop_bidder_init(struct archive_read_filter *self) { struct read_lzop *state; self->code = ARCHIVE_FILTER_LZOP; self->name = "lzop"; state = (struct read_lzop *)calloc(sizeof(*state), 1); if (state == NULL) { archive_set_error(&self->archive->archive, ENOMEM, "Can't allocate data for lzop decompression"); return (ARCHIVE_FATAL); } self->data = state; self->read = lzop_filter_read; self->skip = NULL; /* not supported */ self->close = lzop_filter_close; return (ARCHIVE_OK); } static int consume_header(struct archive_read_filter *self) { struct read_lzop *state = (struct read_lzop *)self->data; const unsigned char *p, *_p; unsigned checksum, flags, len, method, version; /* * Check LZOP magic code. */ p = __archive_read_filter_ahead(self->upstream, LZOP_HEADER_MAGIC_LEN, NULL); if (p == NULL) return (ARCHIVE_EOF); if (memcmp(p, LZOP_HEADER_MAGIC, LZOP_HEADER_MAGIC_LEN)) return (ARCHIVE_EOF); __archive_read_filter_consume(self->upstream, LZOP_HEADER_MAGIC_LEN); p = __archive_read_filter_ahead(self->upstream, 29, NULL); if (p == NULL) goto truncated; _p = p; version = archive_be16dec(p); p += 4;/* version(2 bytes) + library version(2 bytes) */ if (version >= 0x940) { unsigned reqversion = archive_be16dec(p); p += 2; if (reqversion < 0x900) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Invalid required version"); return (ARCHIVE_FAILED); } } method = *p++; if (method < 1 || method > 3) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Unsupported method"); return (ARCHIVE_FAILED); } if (version >= 0x940) { unsigned level = *p++; #if 0 unsigned default_level[] = {0, 3, 1, 9}; #endif if (level == 0) /* Method is 1..3 here due to check above. */ #if 0 /* Avoid an error Clang Static Analyzer claims "Value stored to 'level' is never read". */ level = default_level[method]; #else ;/* NOP */ #endif else if (level > 9) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Invalid level"); return (ARCHIVE_FAILED); } } flags = archive_be32dec(p); p += 4; if (flags & FILTER) p += 4; /* Skip filter */ p += 4; /* Skip mode */ if (version >= 0x940) p += 8; /* Skip mtime */ else p += 4; /* Skip mtime */ len = *p++; /* Read filename length */ len += p - _p; /* Make sure we have all bytes we need to calculate checksum. */ p = __archive_read_filter_ahead(self->upstream, len + 4, NULL); if (p == NULL) goto truncated; if (flags & CRC32_HEADER) checksum = crc32(crc32(0, NULL, 0), p, len); else checksum = adler32(adler32(0, NULL, 0), p, len); if (archive_be32dec(p + len) != checksum) goto corrupted; __archive_read_filter_consume(self->upstream, len + 4); if (flags & EXTRA_FIELD) { /* Skip extra field */ p = __archive_read_filter_ahead(self->upstream, 4, NULL); if (p == NULL) goto truncated; len = archive_be32dec(p); __archive_read_filter_consume(self->upstream, len + 4 + 4); } state->flags = flags; state->in_stream = 1; return (ARCHIVE_OK); truncated: archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated lzop data"); return (ARCHIVE_FAILED); corrupted: archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Corrupted lzop header"); return (ARCHIVE_FAILED); } static int consume_block_info(struct archive_read_filter *self) { struct read_lzop *state = (struct read_lzop *)self->data; const unsigned char *p; unsigned flags = state->flags; p = __archive_read_filter_ahead(self->upstream, 4, NULL); if (p == NULL) goto truncated; state->uncompressed_size = archive_be32dec(p); __archive_read_filter_consume(self->upstream, 4); if (state->uncompressed_size == 0) return (ARCHIVE_EOF); if (state->uncompressed_size > MAX_BLOCK_SIZE) goto corrupted; p = __archive_read_filter_ahead(self->upstream, 4, NULL); if (p == NULL) goto truncated; state->compressed_size = archive_be32dec(p); __archive_read_filter_consume(self->upstream, 4); if (state->compressed_size > state->uncompressed_size) goto corrupted; if (flags & (CRC32_UNCOMPRESSED | ADLER32_UNCOMPRESSED)) { p = __archive_read_filter_ahead(self->upstream, 4, NULL); if (p == NULL) goto truncated; state->compressed_cksum = state->uncompressed_cksum = archive_be32dec(p); __archive_read_filter_consume(self->upstream, 4); } if ((flags & (CRC32_COMPRESSED | ADLER32_COMPRESSED)) && state->compressed_size < state->uncompressed_size) { p = __archive_read_filter_ahead(self->upstream, 4, NULL); if (p == NULL) goto truncated; state->compressed_cksum = archive_be32dec(p); __archive_read_filter_consume(self->upstream, 4); } return (ARCHIVE_OK); truncated: archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated lzop data"); return (ARCHIVE_FAILED); corrupted: archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Corrupted lzop header"); return (ARCHIVE_FAILED); } static ssize_t lzop_filter_read(struct archive_read_filter *self, const void **p) { struct read_lzop *state = (struct read_lzop *)self->data; const void *b; lzo_uint out_size; uint32_t cksum; int ret, r; if (state->unconsumed_bytes) { __archive_read_filter_consume(self->upstream, state->unconsumed_bytes); state->unconsumed_bytes = 0; } if (state->eof) return (0); for (;;) { if (!state->in_stream) { ret = consume_header(self); if (ret < ARCHIVE_OK) return (ret); if (ret == ARCHIVE_EOF) { state->eof = 1; return (0); } } ret = consume_block_info(self); if (ret < ARCHIVE_OK) return (ret); if (ret == ARCHIVE_EOF) state->in_stream = 0; else break; } if (state->out_block == NULL || state->out_block_size < state->uncompressed_size) { void *new_block; new_block = realloc(state->out_block, state->uncompressed_size); if (new_block == NULL) { archive_set_error(&self->archive->archive, ENOMEM, "Can't allocate data for lzop decompression"); return (ARCHIVE_FATAL); } state->out_block = new_block; state->out_block_size = state->uncompressed_size; } b = __archive_read_filter_ahead(self->upstream, state->compressed_size, NULL); if (b == NULL) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated lzop data"); return (ARCHIVE_FATAL); } if (state->flags & CRC32_COMPRESSED) cksum = crc32(crc32(0, NULL, 0), b, state->compressed_size); else if (state->flags & ADLER32_COMPRESSED) cksum = adler32(adler32(0, NULL, 0), b, state->compressed_size); else cksum = state->compressed_cksum; if (cksum != state->compressed_cksum) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Corrupted data"); return (ARCHIVE_FATAL); } /* * If the both uncompressed size and compressed size are the same, * we do not decompress this block. */ if (state->uncompressed_size == state->compressed_size) { *p = b; state->total_out += state->compressed_size; state->unconsumed_bytes = state->compressed_size; return ((ssize_t)state->uncompressed_size); } /* - * Drive lzo uncompresison. + * Drive lzo uncompression. */ out_size = (lzo_uint)state->uncompressed_size; r = lzo1x_decompress_safe(b, (lzo_uint)state->compressed_size, state->out_block, &out_size, NULL); switch (r) { case LZO_E_OK: if (out_size == state->uncompressed_size) break; archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Corrupted data"); return (ARCHIVE_FATAL); case LZO_E_OUT_OF_MEMORY: archive_set_error(&self->archive->archive, ENOMEM, "lzop decompression failed: out of memory"); return (ARCHIVE_FATAL); default: archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "lzop decompression failed: %d", r); return (ARCHIVE_FATAL); } if (state->flags & CRC32_UNCOMPRESSED) cksum = crc32(crc32(0, NULL, 0), state->out_block, state->uncompressed_size); else if (state->flags & ADLER32_UNCOMPRESSED) cksum = adler32(adler32(0, NULL, 0), state->out_block, state->uncompressed_size); else cksum = state->uncompressed_cksum; if (cksum != state->uncompressed_cksum) { archive_set_error(&self->archive->archive, ARCHIVE_ERRNO_MISC, "Corrupted data"); return (ARCHIVE_FATAL); } __archive_read_filter_consume(self->upstream, state->compressed_size); *p = state->out_block; state->total_out += out_size; return ((ssize_t)out_size); } /* * Clean up the decompressor. */ static int lzop_filter_close(struct archive_read_filter *self) { struct read_lzop *state = (struct read_lzop *)self->data; free(state->out_block); free(state); return (ARCHIVE_OK); } #endif Index: head/contrib/libarchive/libarchive/archive_read_support_format_7zip.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_format_7zip.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_read_support_format_7zip.c (revision 311903) @@ -1,3886 +1,3886 @@ /*- * Copyright (c) 2011 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_BZLIB_H #include #endif #ifdef HAVE_LZMA_H #include #endif #ifdef HAVE_ZLIB_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_ppmd7_private.h" #include "archive_private.h" #include "archive_read_private.h" #include "archive_endian.h" #ifndef HAVE_ZLIB_H #include "archive_crc32.h" #endif #define _7ZIP_SIGNATURE "7z\xBC\xAF\x27\x1C" #define SFX_MIN_ADDR 0x27000 #define SFX_MAX_ADDR 0x60000 /* * Codec ID */ #define _7Z_COPY 0 #define _7Z_LZMA 0x030101 #define _7Z_LZMA2 0x21 #define _7Z_DEFLATE 0x040108 #define _7Z_BZ2 0x040202 #define _7Z_PPMD 0x030401 #define _7Z_DELTA 0x03 #define _7Z_CRYPTO_MAIN_ZIP 0x06F10101 /* Main Zip crypto algo */ #define _7Z_CRYPTO_RAR_29 0x06F10303 /* Rar29 AES-128 + (modified SHA-1) */ #define _7Z_CRYPTO_AES_256_SHA_256 0x06F10701 /* AES-256 + SHA-256 */ #define _7Z_X86 0x03030103 #define _7Z_X86_BCJ2 0x0303011B #define _7Z_POWERPC 0x03030205 #define _7Z_IA64 0x03030401 #define _7Z_ARM 0x03030501 #define _7Z_ARMTHUMB 0x03030701 #define _7Z_SPARC 0x03030805 /* * 7-Zip header property IDs. */ #define kEnd 0x00 #define kHeader 0x01 #define kArchiveProperties 0x02 #define kAdditionalStreamsInfo 0x03 #define kMainStreamsInfo 0x04 #define kFilesInfo 0x05 #define kPackInfo 0x06 #define kUnPackInfo 0x07 #define kSubStreamsInfo 0x08 #define kSize 0x09 #define kCRC 0x0A #define kFolder 0x0B #define kCodersUnPackSize 0x0C #define kNumUnPackStream 0x0D #define kEmptyStream 0x0E #define kEmptyFile 0x0F #define kAnti 0x10 #define kName 0x11 #define kCTime 0x12 #define kATime 0x13 #define kMTime 0x14 #define kAttributes 0x15 #define kEncodedHeader 0x17 #define kDummy 0x19 struct _7z_digests { unsigned char *defineds; uint32_t *digests; }; struct _7z_folder { uint64_t numCoders; struct _7z_coder { unsigned long codec; uint64_t numInStreams; uint64_t numOutStreams; uint64_t propertiesSize; unsigned char *properties; } *coders; uint64_t numBindPairs; struct { uint64_t inIndex; uint64_t outIndex; } *bindPairs; uint64_t numPackedStreams; uint64_t *packedStreams; uint64_t numInStreams; uint64_t numOutStreams; uint64_t *unPackSize; unsigned char digest_defined; uint32_t digest; uint64_t numUnpackStreams; uint32_t packIndex; /* Unoperated bytes. */ uint64_t skipped_bytes; }; struct _7z_coders_info { uint64_t numFolders; struct _7z_folder *folders; uint64_t dataStreamIndex; }; struct _7z_pack_info { uint64_t pos; uint64_t numPackStreams; uint64_t *sizes; struct _7z_digests digest; /* Calculated from pos and numPackStreams. */ uint64_t *positions; }; struct _7z_substream_info { size_t unpack_streams; uint64_t *unpackSizes; unsigned char *digestsDefined; uint32_t *digests; }; struct _7z_stream_info { struct _7z_pack_info pi; struct _7z_coders_info ci; struct _7z_substream_info ss; }; struct _7z_header_info { uint64_t dataIndex; unsigned char *emptyStreamBools; unsigned char *emptyFileBools; unsigned char *antiBools; unsigned char *attrBools; }; struct _7zip_entry { size_t name_len; unsigned char *utf16name; #if defined(_WIN32) && !defined(__CYGWIN__) && defined(_DEBUG) const wchar_t *wname; #endif uint32_t folderIndex; uint32_t ssIndex; unsigned flg; #define MTIME_IS_SET (1<<0) #define ATIME_IS_SET (1<<1) #define CTIME_IS_SET (1<<2) #define CRC32_IS_SET (1<<3) #define HAS_STREAM (1<<4) time_t mtime; time_t atime; time_t ctime; long mtime_ns; long atime_ns; long ctime_ns; uint32_t mode; uint32_t attr; }; struct _7zip { /* Structural information about the archive. */ struct _7z_stream_info si; int header_is_being_read; int header_is_encoded; uint64_t header_bytes_remaining; unsigned long header_crc32; /* Header offset to check that reading points of the file contents * will not exceed the header. */ uint64_t header_offset; /* Base offset of the archive file for a seek in case reading SFX. */ uint64_t seek_base; /* List of entries */ size_t entries_remaining; uint64_t numFiles; struct _7zip_entry *entries; struct _7zip_entry *entry; unsigned char *entry_names; /* entry_bytes_remaining is the number of bytes we expect. */ int64_t entry_offset; uint64_t entry_bytes_remaining; /* Running CRC32 of the decompressed data */ unsigned long entry_crc32; /* Flags to mark progress of decompression. */ char end_of_entry; /* Uncompressed buffer control. */ #define UBUFF_SIZE (64 * 1024) unsigned char *uncompressed_buffer; unsigned char *uncompressed_buffer_pointer; size_t uncompressed_buffer_size; size_t uncompressed_buffer_bytes_remaining; /* Offset of the compressed data. */ int64_t stream_offset; /* * Decompressing control data. */ unsigned folder_index; uint64_t folder_outbytes_remaining; unsigned pack_stream_index; unsigned pack_stream_remaining; uint64_t pack_stream_inbytes_remaining; size_t pack_stream_bytes_unconsumed; /* The codec information of a folder. */ unsigned long codec; unsigned long codec2; /* * Decompressor controllers. */ /* Decoding LZMA1 and LZMA2 data. */ #ifdef HAVE_LZMA_H lzma_stream lzstream; int lzstream_valid; #endif /* Decoding bzip2 data. */ #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) bz_stream bzstream; int bzstream_valid; #endif /* Decoding deflate data. */ #ifdef HAVE_ZLIB_H z_stream stream; int stream_valid; #endif /* Decoding PPMd data. */ int ppmd7_stat; CPpmd7 ppmd7_context; CPpmd7z_RangeDec range_dec; IByteIn bytein; struct { const unsigned char *next_in; int64_t avail_in; int64_t total_in; unsigned char *next_out; int64_t avail_out; int64_t total_out; int overconsumed; } ppstream; int ppmd7_valid; /* Decoding BCJ and BCJ2 data. */ uint32_t bcj_state; size_t odd_bcj_size; unsigned char odd_bcj[4]; /* Decoding BCJ data. */ size_t bcj_prevPosT; uint32_t bcj_prevMask; uint32_t bcj_ip; /* Decoding BCJ2 data. */ size_t main_stream_bytes_remaining; unsigned char *sub_stream_buff[3]; size_t sub_stream_size[3]; size_t sub_stream_bytes_remaining[3]; unsigned char *tmp_stream_buff; size_t tmp_stream_buff_size; size_t tmp_stream_bytes_avail; size_t tmp_stream_bytes_remaining; #ifdef _LZMA_PROB32 #define CProb uint32_t #else #define CProb uint16_t #endif CProb bcj2_p[256 + 2]; uint8_t bcj2_prevByte; uint32_t bcj2_range; uint32_t bcj2_code; uint64_t bcj2_outPos; /* Filename character-set conversion data. */ struct archive_string_conv *sconv; char format_name[64]; /* Custom value that is non-zero if this archive contains encrypted entries. */ int has_encrypted_entries; }; /* Maximum entry size. This limitation prevents reading intentional * corrupted 7-zip files on assuming there are not so many entries in * the files. */ #define UMAX_ENTRY ARCHIVE_LITERAL_ULL(100000000) static int archive_read_format_7zip_has_encrypted_entries(struct archive_read *); static int archive_read_support_format_7zip_capabilities(struct archive_read *a); static int archive_read_format_7zip_bid(struct archive_read *, int); static int archive_read_format_7zip_cleanup(struct archive_read *); static int archive_read_format_7zip_read_data(struct archive_read *, const void **, size_t *, int64_t *); static int archive_read_format_7zip_read_data_skip(struct archive_read *); static int archive_read_format_7zip_read_header(struct archive_read *, struct archive_entry *); static int check_7zip_header_in_sfx(const char *); static unsigned long decode_codec_id(const unsigned char *, size_t); static int decode_encoded_header_info(struct archive_read *, struct _7z_stream_info *); static int decompress(struct archive_read *, struct _7zip *, void *, size_t *, const void *, size_t *); static ssize_t extract_pack_stream(struct archive_read *, size_t); static void fileTimeToUtc(uint64_t, time_t *, long *); static uint64_t folder_uncompressed_size(struct _7z_folder *); static void free_CodersInfo(struct _7z_coders_info *); static void free_Digest(struct _7z_digests *); static void free_Folder(struct _7z_folder *); static void free_Header(struct _7z_header_info *); static void free_PackInfo(struct _7z_pack_info *); static void free_StreamsInfo(struct _7z_stream_info *); static void free_SubStreamsInfo(struct _7z_substream_info *); static int free_decompression(struct archive_read *, struct _7zip *); static ssize_t get_uncompressed_data(struct archive_read *, const void **, size_t, size_t); static const unsigned char * header_bytes(struct archive_read *, size_t); static int init_decompression(struct archive_read *, struct _7zip *, const struct _7z_coder *, const struct _7z_coder *); static int parse_7zip_uint64(struct archive_read *, uint64_t *); static int read_Bools(struct archive_read *, unsigned char *, size_t); static int read_CodersInfo(struct archive_read *, struct _7z_coders_info *); static int read_Digests(struct archive_read *, struct _7z_digests *, size_t); static int read_Folder(struct archive_read *, struct _7z_folder *); static int read_Header(struct archive_read *, struct _7z_header_info *, int); static int read_PackInfo(struct archive_read *, struct _7z_pack_info *); static int read_StreamsInfo(struct archive_read *, struct _7z_stream_info *); static int read_SubStreamsInfo(struct archive_read *, struct _7z_substream_info *, struct _7z_folder *, size_t); static int read_Times(struct archive_read *, struct _7z_header_info *, int); static void read_consume(struct archive_read *); static ssize_t read_stream(struct archive_read *, const void **, size_t, size_t); static int seek_pack(struct archive_read *); static int64_t skip_stream(struct archive_read *, size_t); static int skip_sfx(struct archive_read *, ssize_t); static int slurp_central_directory(struct archive_read *, struct _7zip *, struct _7z_header_info *); static int setup_decode_folder(struct archive_read *, struct _7z_folder *, int); static void x86_Init(struct _7zip *); static size_t x86_Convert(struct _7zip *, uint8_t *, size_t); static ssize_t Bcj2_Decode(struct _7zip *, uint8_t *, size_t); int archive_read_support_format_7zip(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct _7zip *zip; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_7zip"); zip = calloc(1, sizeof(*zip)); if (zip == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate 7zip data"); return (ARCHIVE_FATAL); } /* * Until enough data has been read, we cannot tell about * any encrypted entries yet. */ zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; r = __archive_read_register_format(a, zip, "7zip", archive_read_format_7zip_bid, NULL, archive_read_format_7zip_read_header, archive_read_format_7zip_read_data, archive_read_format_7zip_read_data_skip, NULL, archive_read_format_7zip_cleanup, archive_read_support_format_7zip_capabilities, archive_read_format_7zip_has_encrypted_entries); if (r != ARCHIVE_OK) free(zip); return (ARCHIVE_OK); } static int archive_read_support_format_7zip_capabilities(struct archive_read * a) { (void)a; /* UNUSED */ return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); } static int archive_read_format_7zip_has_encrypted_entries(struct archive_read *_a) { if (_a && _a->format) { struct _7zip * zip = (struct _7zip *)_a->format->data; if (zip) { return zip->has_encrypted_entries; } } return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; } static int archive_read_format_7zip_bid(struct archive_read *a, int best_bid) { const char *p; /* If someone has already bid more than 32, then avoid trashing the look-ahead buffers with a seek. */ if (best_bid > 32) return (-1); if ((p = __archive_read_ahead(a, 6, NULL)) == NULL) return (0); /* If first six bytes are the 7-Zip signature, * return the bid right now. */ if (memcmp(p, _7ZIP_SIGNATURE, 6) == 0) return (48); /* * It may a 7-Zip SFX archive file. If first two bytes are * 'M' and 'Z' available on Windows or first four bytes are * "\x7F\x45LF" available on posix like system, seek the 7-Zip * signature. Although we will perform a seek when reading * a header, what we do not use __archive_read_seek() here is * due to a bidding performance. */ if ((p[0] == 'M' && p[1] == 'Z') || memcmp(p, "\x7F\x45LF", 4) == 0) { ssize_t offset = SFX_MIN_ADDR; ssize_t window = 4096; ssize_t bytes_avail; while (offset + window <= (SFX_MAX_ADDR)) { const char *buff = __archive_read_ahead(a, offset + window, &bytes_avail); if (buff == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < 0x40) return (0); continue; } p = buff + offset; while (p + 32 < buff + bytes_avail) { int step = check_7zip_header_in_sfx(p); if (step == 0) return (48); p += step; } offset = p - buff; } } return (0); } static int check_7zip_header_in_sfx(const char *p) { switch ((unsigned char)p[5]) { case 0x1C: if (memcmp(p, _7ZIP_SIGNATURE, 6) != 0) return (6); /* * Test the CRC because its extraction code has 7-Zip * Magic Code, so we should do this in order not to * make a mis-detection. */ if (crc32(0, (const unsigned char *)p + 12, 20) != archive_le32dec(p + 8)) return (6); /* Hit the header! */ return (0); case 0x37: return (5); case 0x7A: return (4); case 0xBC: return (3); case 0xAF: return (2); case 0x27: return (1); default: return (6); } } static int skip_sfx(struct archive_read *a, ssize_t bytes_avail) { const void *h; const char *p, *q; size_t skip, offset; ssize_t bytes, window; /* * If bytes_avail > SFX_MIN_ADDR we do not have to call * __archive_read_seek() at this time since we have - * alredy had enough data. + * already had enough data. */ if (bytes_avail > SFX_MIN_ADDR) __archive_read_consume(a, SFX_MIN_ADDR); else if (__archive_read_seek(a, SFX_MIN_ADDR, SEEK_SET) < 0) return (ARCHIVE_FATAL); offset = 0; window = 1; while (offset + window <= SFX_MAX_ADDR - SFX_MIN_ADDR) { h = __archive_read_ahead(a, window, &bytes); if (h == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < 0x40) goto fatal; continue; } if (bytes < 6) { /* This case might happen when window == 1. */ window = 4096; continue; } p = (const char *)h; q = p + bytes; /* * Scan ahead until we find something that looks * like the 7-Zip header. */ while (p + 32 < q) { int step = check_7zip_header_in_sfx(p); if (step == 0) { struct _7zip *zip = (struct _7zip *)a->format->data; skip = p - (const char *)h; __archive_read_consume(a, skip); zip->seek_base = SFX_MIN_ADDR + offset + skip; return (ARCHIVE_OK); } p += step; } skip = p - (const char *)h; __archive_read_consume(a, skip); offset += skip; if (window == 1) window = 4096; } fatal: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Couldn't find out 7-Zip header"); return (ARCHIVE_FATAL); } static int archive_read_format_7zip_read_header(struct archive_read *a, struct archive_entry *entry) { struct _7zip *zip = (struct _7zip *)a->format->data; struct _7zip_entry *zip_entry; int r, ret = ARCHIVE_OK; struct _7z_folder *folder = 0; uint64_t fidx = 0; /* * It should be sufficient to call archive_read_next_header() for * a reader to determine if an entry is encrypted or not. If the * encryption of an entry is only detectable when calling * archive_read_data(), so be it. We'll do the same check there * as well. */ if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { zip->has_encrypted_entries = 0; } a->archive.archive_format = ARCHIVE_FORMAT_7ZIP; if (a->archive.archive_format_name == NULL) a->archive.archive_format_name = "7-Zip"; if (zip->entries == NULL) { struct _7z_header_info header; memset(&header, 0, sizeof(header)); r = slurp_central_directory(a, zip, &header); free_Header(&header); if (r != ARCHIVE_OK) return (r); zip->entries_remaining = (size_t)zip->numFiles; zip->entry = zip->entries; } else { ++zip->entry; } zip_entry = zip->entry; if (zip->entries_remaining <= 0 || zip_entry == NULL) return ARCHIVE_EOF; --zip->entries_remaining; zip->entry_offset = 0; zip->end_of_entry = 0; zip->entry_crc32 = crc32(0, NULL, 0); /* Setup a string conversion for a filename. */ if (zip->sconv == NULL) { zip->sconv = archive_string_conversion_from_charset( &a->archive, "UTF-16LE", 1); if (zip->sconv == NULL) return (ARCHIVE_FATAL); } /* Figure out if the entry is encrypted by looking at the folder that is associated to the current 7zip entry. If the folder has a coder with a _7Z_CRYPTO codec then the folder is encrypted. Hence the entry must also be encrypted. */ if (zip_entry && zip_entry->folderIndex < zip->si.ci.numFolders) { folder = &(zip->si.ci.folders[zip_entry->folderIndex]); for (fidx=0; folder && fidxnumCoders; fidx++) { switch(folder->coders[fidx].codec) { case _7Z_CRYPTO_MAIN_ZIP: case _7Z_CRYPTO_RAR_29: case _7Z_CRYPTO_AES_256_SHA_256: { archive_entry_set_is_data_encrypted(entry, 1); zip->has_encrypted_entries = 1; break; } } } } /* Now that we've checked for encryption, if there were still no * encrypted entries found we can say for sure that there are none. */ if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { zip->has_encrypted_entries = 0; } if (archive_entry_copy_pathname_l(entry, (const char *)zip_entry->utf16name, zip_entry->name_len, zip->sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Pathname cannot be converted " "from %s to current locale.", archive_string_conversion_charset_name(zip->sconv)); ret = ARCHIVE_WARN; } /* Populate some additional entry fields: */ archive_entry_set_mode(entry, zip_entry->mode); if (zip_entry->flg & MTIME_IS_SET) archive_entry_set_mtime(entry, zip_entry->mtime, zip_entry->mtime_ns); if (zip_entry->flg & CTIME_IS_SET) archive_entry_set_ctime(entry, zip_entry->ctime, zip_entry->ctime_ns); if (zip_entry->flg & ATIME_IS_SET) archive_entry_set_atime(entry, zip_entry->atime, zip_entry->atime_ns); if (zip_entry->ssIndex != (uint32_t)-1) { zip->entry_bytes_remaining = zip->si.ss.unpackSizes[zip_entry->ssIndex]; archive_entry_set_size(entry, zip->entry_bytes_remaining); } else { zip->entry_bytes_remaining = 0; archive_entry_set_size(entry, 0); } /* If there's no body, force read_data() to return EOF immediately. */ if (zip->entry_bytes_remaining < 1) zip->end_of_entry = 1; if ((zip_entry->mode & AE_IFMT) == AE_IFLNK) { unsigned char *symname = NULL; size_t symsize = 0; /* * Symbolic-name is recorded as its contents. We have to * read the contents at this time. */ while (zip->entry_bytes_remaining > 0) { const void *buff; unsigned char *mem; size_t size; int64_t offset; r = archive_read_format_7zip_read_data(a, &buff, &size, &offset); if (r < ARCHIVE_WARN) { free(symname); return (r); } mem = realloc(symname, symsize + size + 1); if (mem == NULL) { free(symname); archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Symname"); return (ARCHIVE_FATAL); } symname = mem; memcpy(symname+symsize, buff, size); symsize += size; } if (symsize == 0) { - /* If there is no synname, handle it as a regular + /* If there is no symname, handle it as a regular * file. */ zip_entry->mode &= ~AE_IFMT; zip_entry->mode |= AE_IFREG; archive_entry_set_mode(entry, zip_entry->mode); } else { symname[symsize] = '\0'; archive_entry_copy_symlink(entry, (const char *)symname); } free(symname); archive_entry_set_size(entry, 0); } /* Set up a more descriptive format name. */ sprintf(zip->format_name, "7-Zip"); a->archive.archive_format_name = zip->format_name; return (ret); } static int archive_read_format_7zip_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct _7zip *zip; ssize_t bytes; int ret = ARCHIVE_OK; zip = (struct _7zip *)(a->format->data); if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { zip->has_encrypted_entries = 0; } if (zip->pack_stream_bytes_unconsumed) read_consume(a); *offset = zip->entry_offset; *size = 0; *buff = NULL; /* * If we hit end-of-entry last time, clean up and return * ARCHIVE_EOF this time. */ if (zip->end_of_entry) return (ARCHIVE_EOF); bytes = read_stream(a, buff, (size_t)zip->entry_bytes_remaining, 0); if (bytes < 0) return ((int)bytes); if (bytes == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7-Zip file body"); return (ARCHIVE_FATAL); } zip->entry_bytes_remaining -= bytes; if (zip->entry_bytes_remaining == 0) zip->end_of_entry = 1; /* Update checksum */ if ((zip->entry->flg & CRC32_IS_SET) && bytes) zip->entry_crc32 = crc32(zip->entry_crc32, *buff, (unsigned)bytes); /* If we hit the end, swallow any end-of-data marker. */ if (zip->end_of_entry) { /* Check computed CRC against file contents. */ if ((zip->entry->flg & CRC32_IS_SET) && zip->si.ss.digests[zip->entry->ssIndex] != zip->entry_crc32) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "7-Zip bad CRC: 0x%lx should be 0x%lx", (unsigned long)zip->entry_crc32, (unsigned long)zip->si.ss.digests[ zip->entry->ssIndex]); ret = ARCHIVE_WARN; } } *size = bytes; *offset = zip->entry_offset; zip->entry_offset += bytes; return (ret); } static int archive_read_format_7zip_read_data_skip(struct archive_read *a) { struct _7zip *zip; int64_t bytes_skipped; zip = (struct _7zip *)(a->format->data); if (zip->pack_stream_bytes_unconsumed) read_consume(a); /* If we've already read to end of data, we're done. */ if (zip->end_of_entry) return (ARCHIVE_OK); /* * If the length is at the beginning, we can skip the * compressed data much more quickly. */ bytes_skipped = skip_stream(a, (size_t)zip->entry_bytes_remaining); if (bytes_skipped < 0) return (ARCHIVE_FATAL); zip->entry_bytes_remaining = 0; /* This entry is finished and done. */ zip->end_of_entry = 1; return (ARCHIVE_OK); } static int archive_read_format_7zip_cleanup(struct archive_read *a) { struct _7zip *zip; zip = (struct _7zip *)(a->format->data); free_StreamsInfo(&(zip->si)); free(zip->entries); free(zip->entry_names); free_decompression(a, zip); free(zip->uncompressed_buffer); free(zip->sub_stream_buff[0]); free(zip->sub_stream_buff[1]); free(zip->sub_stream_buff[2]); free(zip->tmp_stream_buff); free(zip); (a->format->data) = NULL; return (ARCHIVE_OK); } static void read_consume(struct archive_read *a) { struct _7zip *zip = (struct _7zip *)a->format->data; if (zip->pack_stream_bytes_unconsumed) { __archive_read_consume(a, zip->pack_stream_bytes_unconsumed); zip->stream_offset += zip->pack_stream_bytes_unconsumed; zip->pack_stream_bytes_unconsumed = 0; } } #ifdef HAVE_LZMA_H /* * Set an error code and choose an error message for liblzma. */ static void set_error(struct archive_read *a, int ret) { switch (ret) { case LZMA_STREAM_END: /* Found end of stream. */ case LZMA_OK: /* Decompressor made some progress. */ break; case LZMA_MEM_ERROR: archive_set_error(&a->archive, ENOMEM, "Lzma library error: Cannot allocate memory"); break; case LZMA_MEMLIMIT_ERROR: archive_set_error(&a->archive, ENOMEM, "Lzma library error: Out of memory"); break; case LZMA_FORMAT_ERROR: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Lzma library error: format not recognized"); break; case LZMA_OPTIONS_ERROR: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Lzma library error: Invalid options"); break; case LZMA_DATA_ERROR: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Lzma library error: Corrupted input data"); break; case LZMA_BUF_ERROR: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Lzma library error: No progress is possible"); break; default: /* Return an error. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Lzma decompression failed: Unknown error"); break; } } #endif static unsigned long decode_codec_id(const unsigned char *codecId, size_t id_size) { unsigned i; unsigned long id = 0; for (i = 0; i < id_size; i++) { id <<= 8; id += codecId[i]; } return (id); } static void * ppmd_alloc(void *p, size_t size) { (void)p; return malloc(size); } static void ppmd_free(void *p, void *address) { (void)p; free(address); } static Byte ppmd_read(void *p) { struct archive_read *a = ((IByteIn*)p)->a; struct _7zip *zip = (struct _7zip *)(a->format->data); Byte b; if (zip->ppstream.avail_in == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); zip->ppstream.overconsumed = 1; return (0); } b = *zip->ppstream.next_in++; zip->ppstream.avail_in--; zip->ppstream.total_in++; return (b); } static ISzAlloc g_szalloc = { ppmd_alloc, ppmd_free }; static int init_decompression(struct archive_read *a, struct _7zip *zip, const struct _7z_coder *coder1, const struct _7z_coder *coder2) { int r; zip->codec = coder1->codec; zip->codec2 = -1; switch (zip->codec) { case _7Z_COPY: case _7Z_BZ2: case _7Z_DEFLATE: case _7Z_PPMD: if (coder2 != NULL) { if (coder2->codec != _7Z_X86 && coder2->codec != _7Z_X86_BCJ2) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unsupported filter %lx for %lx", coder2->codec, coder1->codec); return (ARCHIVE_FAILED); } zip->codec2 = coder2->codec; zip->bcj_state = 0; if (coder2->codec == _7Z_X86) x86_Init(zip); } break; default: break; } switch (zip->codec) { case _7Z_COPY: break; case _7Z_LZMA: case _7Z_LZMA2: #ifdef HAVE_LZMA_H #if LZMA_VERSION_MAJOR >= 5 /* Effectively disable the limiter. */ #define LZMA_MEMLIMIT UINT64_MAX #else /* NOTE: This needs to check memory size which running system has. */ #define LZMA_MEMLIMIT (1U << 30) #endif { lzma_options_delta delta_opt; lzma_filter filters[LZMA_FILTERS_MAX], *ff; int fi = 0; if (zip->lzstream_valid) { lzma_end(&(zip->lzstream)); zip->lzstream_valid = 0; } /* * NOTE: liblzma incompletely handle the BCJ+LZMA compressed * data made by 7-Zip because 7-Zip does not add End-Of- * Payload Marker(EOPM) at the end of LZMA compressed data, * and so liblzma cannot know the end of the compressed data * without EOPM. So consequently liblzma will not return last * three or four bytes of uncompressed data because * LZMA_FILTER_X86 filter does not handle input data if its * data size is less than five bytes. If liblzma detect EOPM * or know the uncompressed data size, liblzma will flush out * the remaining that three or four bytes of uncompressed * data. That is why we have to use our converting program * for BCJ+LZMA. If we were able to tell the uncompressed * size to liblzma when using lzma_raw_decoder() liblzma * could correctly deal with BCJ+LZMA. But unfortunately * there is no way to do that. * Discussion about this can be found at XZ Utils forum. */ if (coder2 != NULL) { zip->codec2 = coder2->codec; filters[fi].options = NULL; switch (zip->codec2) { case _7Z_X86: if (zip->codec == _7Z_LZMA2) { filters[fi].id = LZMA_FILTER_X86; fi++; } else /* Use our filter. */ x86_Init(zip); break; case _7Z_X86_BCJ2: /* Use our filter. */ zip->bcj_state = 0; break; case _7Z_DELTA: filters[fi].id = LZMA_FILTER_DELTA; memset(&delta_opt, 0, sizeof(delta_opt)); delta_opt.type = LZMA_DELTA_TYPE_BYTE; delta_opt.dist = 1; filters[fi].options = &delta_opt; fi++; break; /* Following filters have not been tested yet. */ case _7Z_POWERPC: filters[fi].id = LZMA_FILTER_POWERPC; fi++; break; case _7Z_IA64: filters[fi].id = LZMA_FILTER_IA64; fi++; break; case _7Z_ARM: filters[fi].id = LZMA_FILTER_ARM; fi++; break; case _7Z_ARMTHUMB: filters[fi].id = LZMA_FILTER_ARMTHUMB; fi++; break; case _7Z_SPARC: filters[fi].id = LZMA_FILTER_SPARC; fi++; break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unexpected codec ID: %lX", zip->codec2); return (ARCHIVE_FAILED); } } if (zip->codec == _7Z_LZMA2) filters[fi].id = LZMA_FILTER_LZMA2; else filters[fi].id = LZMA_FILTER_LZMA1; filters[fi].options = NULL; ff = &filters[fi]; r = lzma_properties_decode(&filters[fi], NULL, coder1->properties, (size_t)coder1->propertiesSize); if (r != LZMA_OK) { set_error(a, r); return (ARCHIVE_FAILED); } fi++; filters[fi].id = LZMA_VLI_UNKNOWN; filters[fi].options = NULL; r = lzma_raw_decoder(&(zip->lzstream), filters); free(ff->options); if (r != LZMA_OK) { set_error(a, r); return (ARCHIVE_FAILED); } zip->lzstream_valid = 1; zip->lzstream.total_in = 0; zip->lzstream.total_out = 0; break; } #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "LZMA codec is unsupported"); return (ARCHIVE_FAILED); #endif case _7Z_BZ2: #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) if (zip->bzstream_valid) { BZ2_bzDecompressEnd(&(zip->bzstream)); zip->bzstream_valid = 0; } r = BZ2_bzDecompressInit(&(zip->bzstream), 0, 0); if (r == BZ_MEM_ERROR) r = BZ2_bzDecompressInit(&(zip->bzstream), 0, 1); if (r != BZ_OK) { int err = ARCHIVE_ERRNO_MISC; const char *detail = NULL; switch (r) { case BZ_PARAM_ERROR: detail = "invalid setup parameter"; break; case BZ_MEM_ERROR: err = ENOMEM; detail = "out of memory"; break; case BZ_CONFIG_ERROR: detail = "mis-compiled library"; break; } archive_set_error(&a->archive, err, "Internal error initializing decompressor: %s", detail != NULL ? detail : "??"); zip->bzstream_valid = 0; return (ARCHIVE_FAILED); } zip->bzstream_valid = 1; zip->bzstream.total_in_lo32 = 0; zip->bzstream.total_in_hi32 = 0; zip->bzstream.total_out_lo32 = 0; zip->bzstream.total_out_hi32 = 0; break; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "BZ2 codec is unsupported"); return (ARCHIVE_FAILED); #endif case _7Z_DEFLATE: #ifdef HAVE_ZLIB_H if (zip->stream_valid) r = inflateReset(&(zip->stream)); else r = inflateInit2(&(zip->stream), -15 /* Don't check for zlib header */); if (r != Z_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Couldn't initialize zlib stream."); return (ARCHIVE_FAILED); } zip->stream_valid = 1; zip->stream.total_in = 0; zip->stream.total_out = 0; break; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "DEFLATE codec is unsupported"); return (ARCHIVE_FAILED); #endif case _7Z_PPMD: { unsigned order; uint32_t msize; if (zip->ppmd7_valid) { __archive_ppmd7_functions.Ppmd7_Free( &zip->ppmd7_context, &g_szalloc); zip->ppmd7_valid = 0; } if (coder1->propertiesSize < 5) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed PPMd parameter"); return (ARCHIVE_FAILED); } order = coder1->properties[0]; msize = archive_le32dec(&(coder1->properties[1])); if (order < PPMD7_MIN_ORDER || order > PPMD7_MAX_ORDER || msize < PPMD7_MIN_MEM_SIZE || msize > PPMD7_MAX_MEM_SIZE) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed PPMd parameter"); return (ARCHIVE_FAILED); } __archive_ppmd7_functions.Ppmd7_Construct(&zip->ppmd7_context); r = __archive_ppmd7_functions.Ppmd7_Alloc( &zip->ppmd7_context, msize, &g_szalloc); if (r == 0) { archive_set_error(&a->archive, ENOMEM, "Coludn't allocate memory for PPMd"); return (ARCHIVE_FATAL); } __archive_ppmd7_functions.Ppmd7_Init( &zip->ppmd7_context, order); __archive_ppmd7_functions.Ppmd7z_RangeDec_CreateVTable( &zip->range_dec); zip->ppmd7_valid = 1; zip->ppmd7_stat = 0; zip->ppstream.overconsumed = 0; zip->ppstream.total_in = 0; zip->ppstream.total_out = 0; break; } case _7Z_X86: case _7Z_X86_BCJ2: case _7Z_POWERPC: case _7Z_IA64: case _7Z_ARM: case _7Z_ARMTHUMB: case _7Z_SPARC: case _7Z_DELTA: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unexpected codec ID: %lX", zip->codec); return (ARCHIVE_FAILED); case _7Z_CRYPTO_MAIN_ZIP: case _7Z_CRYPTO_RAR_29: case _7Z_CRYPTO_AES_256_SHA_256: if (a->entry) { archive_entry_set_is_metadata_encrypted(a->entry, 1); archive_entry_set_is_data_encrypted(a->entry, 1); zip->has_encrypted_entries = 1; } archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Crypto codec not supported yet (ID: 0x%lX)", zip->codec); return (ARCHIVE_FAILED); default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unknown codec ID: %lX", zip->codec); return (ARCHIVE_FAILED); } return (ARCHIVE_OK); } static int decompress(struct archive_read *a, struct _7zip *zip, void *buff, size_t *outbytes, const void *b, size_t *used) { const uint8_t *t_next_in; uint8_t *t_next_out; size_t o_avail_in, o_avail_out; size_t t_avail_in, t_avail_out; uint8_t *bcj2_next_out; size_t bcj2_avail_out; int r, ret = ARCHIVE_OK; t_avail_in = o_avail_in = *used; t_avail_out = o_avail_out = *outbytes; t_next_in = b; t_next_out = buff; if (zip->codec != _7Z_LZMA2 && zip->codec2 == _7Z_X86) { int i; /* Do not copy out the BCJ remaining bytes when the output * buffer size is less than five bytes. */ if (o_avail_in != 0 && t_avail_out < 5 && zip->odd_bcj_size) { *used = 0; *outbytes = 0; return (ret); } for (i = 0; zip->odd_bcj_size > 0 && t_avail_out; i++) { *t_next_out++ = zip->odd_bcj[i]; t_avail_out--; zip->odd_bcj_size--; } if (o_avail_in == 0 || t_avail_out == 0) { *used = o_avail_in - t_avail_in; *outbytes = o_avail_out - t_avail_out; if (o_avail_in == 0) ret = ARCHIVE_EOF; return (ret); } } bcj2_next_out = t_next_out; bcj2_avail_out = t_avail_out; if (zip->codec2 == _7Z_X86_BCJ2) { /* * Decord a remaining decompressed main stream for BCJ2. */ if (zip->tmp_stream_bytes_remaining) { ssize_t bytes; size_t remaining = zip->tmp_stream_bytes_remaining; bytes = Bcj2_Decode(zip, t_next_out, t_avail_out); if (bytes < 0) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "BCJ2 conversion Failed"); return (ARCHIVE_FAILED); } zip->main_stream_bytes_remaining -= remaining - zip->tmp_stream_bytes_remaining; t_avail_out -= bytes; if (o_avail_in == 0 || t_avail_out == 0) { *used = 0; *outbytes = o_avail_out - t_avail_out; if (o_avail_in == 0 && zip->tmp_stream_bytes_remaining) ret = ARCHIVE_EOF; return (ret); } t_next_out += bytes; bcj2_next_out = t_next_out; bcj2_avail_out = t_avail_out; } t_next_out = zip->tmp_stream_buff; t_avail_out = zip->tmp_stream_buff_size; } switch (zip->codec) { case _7Z_COPY: { size_t bytes = (t_avail_in > t_avail_out)?t_avail_out:t_avail_in; memcpy(t_next_out, t_next_in, bytes); t_avail_in -= bytes; t_avail_out -= bytes; if (o_avail_in == 0) ret = ARCHIVE_EOF; break; } #ifdef HAVE_LZMA_H case _7Z_LZMA: case _7Z_LZMA2: zip->lzstream.next_in = t_next_in; zip->lzstream.avail_in = t_avail_in; zip->lzstream.next_out = t_next_out; zip->lzstream.avail_out = t_avail_out; r = lzma_code(&(zip->lzstream), LZMA_RUN); switch (r) { case LZMA_STREAM_END: /* Found end of stream. */ lzma_end(&(zip->lzstream)); zip->lzstream_valid = 0; ret = ARCHIVE_EOF; break; case LZMA_OK: /* Decompressor made some progress. */ break; default: archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Decompression failed(%d)", r); return (ARCHIVE_FAILED); } t_avail_in = zip->lzstream.avail_in; t_avail_out = zip->lzstream.avail_out; break; #endif #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) case _7Z_BZ2: zip->bzstream.next_in = (char *)(uintptr_t)t_next_in; zip->bzstream.avail_in = t_avail_in; zip->bzstream.next_out = (char *)(uintptr_t)t_next_out; zip->bzstream.avail_out = t_avail_out; r = BZ2_bzDecompress(&(zip->bzstream)); switch (r) { case BZ_STREAM_END: /* Found end of stream. */ switch (BZ2_bzDecompressEnd(&(zip->bzstream))) { case BZ_OK: break; default: archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Failed to clean up decompressor"); return (ARCHIVE_FAILED); } zip->bzstream_valid = 0; ret = ARCHIVE_EOF; break; case BZ_OK: /* Decompressor made some progress. */ break; default: archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "bzip decompression failed"); return (ARCHIVE_FAILED); } t_avail_in = zip->bzstream.avail_in; t_avail_out = zip->bzstream.avail_out; break; #endif #ifdef HAVE_ZLIB_H case _7Z_DEFLATE: zip->stream.next_in = (Bytef *)(uintptr_t)t_next_in; zip->stream.avail_in = (uInt)t_avail_in; zip->stream.next_out = t_next_out; zip->stream.avail_out = (uInt)t_avail_out; r = inflate(&(zip->stream), 0); switch (r) { case Z_STREAM_END: /* Found end of stream. */ ret = ARCHIVE_EOF; break; case Z_OK: /* Decompressor made some progress.*/ break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "File decompression failed (%d)", r); return (ARCHIVE_FAILED); } t_avail_in = zip->stream.avail_in; t_avail_out = zip->stream.avail_out; break; #endif case _7Z_PPMD: { uint64_t flush_bytes; if (!zip->ppmd7_valid || zip->ppmd7_stat < 0 || t_avail_out <= 0) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Decompression internal error"); return (ARCHIVE_FAILED); } zip->ppstream.next_in = t_next_in; zip->ppstream.avail_in = t_avail_in; zip->ppstream.next_out = t_next_out; zip->ppstream.avail_out = t_avail_out; if (zip->ppmd7_stat == 0) { zip->bytein.a = a; zip->bytein.Read = &ppmd_read; zip->range_dec.Stream = &zip->bytein; r = __archive_ppmd7_functions.Ppmd7z_RangeDec_Init( &(zip->range_dec)); if (r == 0) { zip->ppmd7_stat = -1; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to initialize PPMd range decorder"); return (ARCHIVE_FAILED); } if (zip->ppstream.overconsumed) { zip->ppmd7_stat = -1; return (ARCHIVE_FAILED); } zip->ppmd7_stat = 1; } if (t_avail_in == 0) /* XXX Flush out remaining decoded data XXX */ flush_bytes = zip->folder_outbytes_remaining; else flush_bytes = 0; do { int sym; sym = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( &(zip->ppmd7_context), &(zip->range_dec.p)); if (sym < 0) { zip->ppmd7_stat = -1; archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Failed to decode PPMd"); return (ARCHIVE_FAILED); } if (zip->ppstream.overconsumed) { zip->ppmd7_stat = -1; return (ARCHIVE_FAILED); } *zip->ppstream.next_out++ = (unsigned char)sym; zip->ppstream.avail_out--; zip->ppstream.total_out++; if (flush_bytes) flush_bytes--; } while (zip->ppstream.avail_out && (zip->ppstream.avail_in || flush_bytes)); t_avail_in = (size_t)zip->ppstream.avail_in; t_avail_out = (size_t)zip->ppstream.avail_out; break; } default: archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Decompression internal error"); return (ARCHIVE_FAILED); } if (ret != ARCHIVE_OK && ret != ARCHIVE_EOF) return (ret); *used = o_avail_in - t_avail_in; *outbytes = o_avail_out - t_avail_out; /* * Decord BCJ. */ if (zip->codec != _7Z_LZMA2 && zip->codec2 == _7Z_X86) { size_t l = x86_Convert(zip, buff, *outbytes); zip->odd_bcj_size = *outbytes - l; if (zip->odd_bcj_size > 0 && zip->odd_bcj_size <= 4 && o_avail_in && ret != ARCHIVE_EOF) { memcpy(zip->odd_bcj, ((unsigned char *)buff) + l, zip->odd_bcj_size); *outbytes = l; } else zip->odd_bcj_size = 0; } /* * Decord BCJ2 with a decompressed main stream. */ if (zip->codec2 == _7Z_X86_BCJ2) { ssize_t bytes; zip->tmp_stream_bytes_avail = zip->tmp_stream_buff_size - t_avail_out; if (zip->tmp_stream_bytes_avail > zip->main_stream_bytes_remaining) zip->tmp_stream_bytes_avail = zip->main_stream_bytes_remaining; zip->tmp_stream_bytes_remaining = zip->tmp_stream_bytes_avail; bytes = Bcj2_Decode(zip, bcj2_next_out, bcj2_avail_out); if (bytes < 0) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "BCJ2 conversion Failed"); return (ARCHIVE_FAILED); } zip->main_stream_bytes_remaining -= zip->tmp_stream_bytes_avail - zip->tmp_stream_bytes_remaining; bcj2_avail_out -= bytes; *outbytes = o_avail_out - bcj2_avail_out; } return (ret); } static int free_decompression(struct archive_read *a, struct _7zip *zip) { int r = ARCHIVE_OK; #if !defined(HAVE_ZLIB_H) &&\ !(defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR)) (void)a;/* UNUSED */ #endif #ifdef HAVE_LZMA_H if (zip->lzstream_valid) lzma_end(&(zip->lzstream)); #endif #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) if (zip->bzstream_valid) { if (BZ2_bzDecompressEnd(&(zip->bzstream)) != BZ_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to clean up bzip2 decompressor"); r = ARCHIVE_FATAL; } zip->bzstream_valid = 0; } #endif #ifdef HAVE_ZLIB_H if (zip->stream_valid) { if (inflateEnd(&(zip->stream)) != Z_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to clean up zlib decompressor"); r = ARCHIVE_FATAL; } zip->stream_valid = 0; } #endif if (zip->ppmd7_valid) { __archive_ppmd7_functions.Ppmd7_Free( &zip->ppmd7_context, &g_szalloc); zip->ppmd7_valid = 0; } return (r); } static int parse_7zip_uint64(struct archive_read *a, uint64_t *val) { const unsigned char *p; unsigned char avail, mask; int i; if ((p = header_bytes(a, 1)) == NULL) return (-1); avail = *p; mask = 0x80; *val = 0; for (i = 0; i < 8; i++) { if (avail & mask) { if ((p = header_bytes(a, 1)) == NULL) return (-1); *val |= ((uint64_t)*p) << (8 * i); mask >>= 1; continue; } *val += ((uint64_t)(avail & (mask -1))) << (8 * i); break; } return (0); } static int read_Bools(struct archive_read *a, unsigned char *data, size_t num) { const unsigned char *p; unsigned i, mask = 0, avail = 0; for (i = 0; i < num; i++) { if (mask == 0) { if ((p = header_bytes(a, 1)) == NULL) return (-1); avail = *p; mask = 0x80; } data[i] = (avail & mask)?1:0; mask >>= 1; } return (0); } static void free_Digest(struct _7z_digests *d) { free(d->defineds); free(d->digests); } static int read_Digests(struct archive_read *a, struct _7z_digests *d, size_t num) { const unsigned char *p; unsigned i; if (num == 0) return (-1); memset(d, 0, sizeof(*d)); d->defineds = malloc(num); if (d->defineds == NULL) return (-1); /* * Read Bools. */ if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p == 0) { if (read_Bools(a, d->defineds, num) < 0) return (-1); } else /* All are defined */ memset(d->defineds, 1, num); d->digests = calloc(num, sizeof(*d->digests)); if (d->digests == NULL) return (-1); for (i = 0; i < num; i++) { if (d->defineds[i]) { if ((p = header_bytes(a, 4)) == NULL) return (-1); d->digests[i] = archive_le32dec(p); } } return (0); } static void free_PackInfo(struct _7z_pack_info *pi) { free(pi->sizes); free(pi->positions); free_Digest(&(pi->digest)); } static int read_PackInfo(struct archive_read *a, struct _7z_pack_info *pi) { const unsigned char *p; unsigned i; memset(pi, 0, sizeof(*pi)); /* * Read PackPos. */ if (parse_7zip_uint64(a, &(pi->pos)) < 0) return (-1); /* * Read NumPackStreams. */ if (parse_7zip_uint64(a, &(pi->numPackStreams)) < 0) return (-1); if (pi->numPackStreams == 0) return (-1); if (UMAX_ENTRY < pi->numPackStreams) return (-1); /* * Read PackSizes[num] */ if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p == kEnd) /* PackSizes[num] are not present. */ return (0); if (*p != kSize) return (-1); pi->sizes = calloc((size_t)pi->numPackStreams, sizeof(uint64_t)); pi->positions = calloc((size_t)pi->numPackStreams, sizeof(uint64_t)); if (pi->sizes == NULL || pi->positions == NULL) return (-1); for (i = 0; i < pi->numPackStreams; i++) { if (parse_7zip_uint64(a, &(pi->sizes[i])) < 0) return (-1); } /* * Read PackStreamDigests[num] */ if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p == kEnd) { /* PackStreamDigests[num] are not present. */ pi->digest.defineds = calloc((size_t)pi->numPackStreams, sizeof(*pi->digest.defineds)); pi->digest.digests = calloc((size_t)pi->numPackStreams, sizeof(*pi->digest.digests)); if (pi->digest.defineds == NULL || pi->digest.digests == NULL) return (-1); return (0); } if (*p != kSize) return (-1); if (read_Digests(a, &(pi->digest), (size_t)pi->numPackStreams) < 0) return (-1); /* * Must be marked by kEnd. */ if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p != kEnd) return (-1); return (0); } static void free_Folder(struct _7z_folder *f) { unsigned i; if (f->coders) { for (i = 0; i< f->numCoders; i++) { free(f->coders[i].properties); } free(f->coders); } free(f->bindPairs); free(f->packedStreams); free(f->unPackSize); } static int read_Folder(struct archive_read *a, struct _7z_folder *f) { struct _7zip *zip = (struct _7zip *)a->format->data; const unsigned char *p; uint64_t numInStreamsTotal = 0; uint64_t numOutStreamsTotal = 0; unsigned i; memset(f, 0, sizeof(*f)); /* * Read NumCoders. */ if (parse_7zip_uint64(a, &(f->numCoders)) < 0) return (-1); if (f->numCoders > 4) /* Too many coders. */ return (-1); f->coders = calloc((size_t)f->numCoders, sizeof(*f->coders)); if (f->coders == NULL) return (-1); for (i = 0; i< f->numCoders; i++) { size_t codec_size; int simple, attr; if ((p = header_bytes(a, 1)) == NULL) return (-1); /* * 0:3 CodecIdSize * 4: 0 - IsSimple * 1 - Is not Simple * 5: 0 - No Attributes * 1 - There are Attributes; * 7: Must be zero. */ codec_size = *p & 0xf; simple = (*p & 0x10)?0:1; attr = *p & 0x20; if (*p & 0x80) return (-1);/* Not supported. */ /* * Read Decompression Method IDs. */ if ((p = header_bytes(a, codec_size)) == NULL) return (-1); f->coders[i].codec = decode_codec_id(p, codec_size); if (simple) { f->coders[i].numInStreams = 1; f->coders[i].numOutStreams = 1; } else { if (parse_7zip_uint64( a, &(f->coders[i].numInStreams)) < 0) return (-1); if (UMAX_ENTRY < f->coders[i].numInStreams) return (-1); if (parse_7zip_uint64( a, &(f->coders[i].numOutStreams)) < 0) return (-1); if (UMAX_ENTRY < f->coders[i].numOutStreams) return (-1); } if (attr) { if (parse_7zip_uint64( a, &(f->coders[i].propertiesSize)) < 0) return (-1); if ((p = header_bytes( a, (size_t)f->coders[i].propertiesSize)) == NULL) return (-1); f->coders[i].properties = malloc((size_t)f->coders[i].propertiesSize); if (f->coders[i].properties == NULL) return (-1); memcpy(f->coders[i].properties, p, (size_t)f->coders[i].propertiesSize); } numInStreamsTotal += f->coders[i].numInStreams; numOutStreamsTotal += f->coders[i].numOutStreams; } if (numOutStreamsTotal == 0 || numInStreamsTotal < numOutStreamsTotal-1) return (-1); f->numBindPairs = numOutStreamsTotal - 1; if (zip->header_bytes_remaining < f->numBindPairs) return (-1); if (f->numBindPairs > 0) { f->bindPairs = calloc((size_t)f->numBindPairs, sizeof(*f->bindPairs)); if (f->bindPairs == NULL) return (-1); } else f->bindPairs = NULL; for (i = 0; i < f->numBindPairs; i++) { if (parse_7zip_uint64(a, &(f->bindPairs[i].inIndex)) < 0) return (-1); if (UMAX_ENTRY < f->bindPairs[i].inIndex) return (-1); if (parse_7zip_uint64(a, &(f->bindPairs[i].outIndex)) < 0) return (-1); if (UMAX_ENTRY < f->bindPairs[i].outIndex) return (-1); } f->numPackedStreams = numInStreamsTotal - f->numBindPairs; f->packedStreams = calloc((size_t)f->numPackedStreams, sizeof(*f->packedStreams)); if (f->packedStreams == NULL) return (-1); if (f->numPackedStreams == 1) { for (i = 0; i < numInStreamsTotal; i++) { unsigned j; for (j = 0; j < f->numBindPairs; j++) { if (f->bindPairs[j].inIndex == i) break; } if (j == f->numBindPairs) break; } if (i == numInStreamsTotal) return (-1); f->packedStreams[0] = i; } else { for (i = 0; i < f->numPackedStreams; i++) { if (parse_7zip_uint64(a, &(f->packedStreams[i])) < 0) return (-1); if (UMAX_ENTRY < f->packedStreams[i]) return (-1); } } f->numInStreams = numInStreamsTotal; f->numOutStreams = numOutStreamsTotal; return (0); } static void free_CodersInfo(struct _7z_coders_info *ci) { unsigned i; if (ci->folders) { for (i = 0; i < ci->numFolders; i++) free_Folder(&(ci->folders[i])); free(ci->folders); } } static int read_CodersInfo(struct archive_read *a, struct _7z_coders_info *ci) { const unsigned char *p; struct _7z_digests digest; unsigned i; memset(ci, 0, sizeof(*ci)); memset(&digest, 0, sizeof(digest)); if ((p = header_bytes(a, 1)) == NULL) goto failed; if (*p != kFolder) goto failed; /* * Read NumFolders. */ if (parse_7zip_uint64(a, &(ci->numFolders)) < 0) goto failed; if (UMAX_ENTRY < ci->numFolders) return (-1); /* * Read External. */ if ((p = header_bytes(a, 1)) == NULL) goto failed; switch (*p) { case 0: ci->folders = calloc((size_t)ci->numFolders, sizeof(*ci->folders)); if (ci->folders == NULL) return (-1); for (i = 0; i < ci->numFolders; i++) { if (read_Folder(a, &(ci->folders[i])) < 0) goto failed; } break; case 1: if (parse_7zip_uint64(a, &(ci->dataStreamIndex)) < 0) return (-1); if (UMAX_ENTRY < ci->dataStreamIndex) return (-1); if (ci->numFolders > 0) { archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); goto failed; } break; default: archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); goto failed; } if ((p = header_bytes(a, 1)) == NULL) goto failed; if (*p != kCodersUnPackSize) goto failed; for (i = 0; i < ci->numFolders; i++) { struct _7z_folder *folder = &(ci->folders[i]); unsigned j; folder->unPackSize = calloc((size_t)folder->numOutStreams, sizeof(*folder->unPackSize)); if (folder->unPackSize == NULL) goto failed; for (j = 0; j < folder->numOutStreams; j++) { if (parse_7zip_uint64(a, &(folder->unPackSize[j])) < 0) goto failed; } } /* * Read CRCs. */ if ((p = header_bytes(a, 1)) == NULL) goto failed; if (*p == kEnd) return (0); if (*p != kCRC) goto failed; if (read_Digests(a, &digest, (size_t)ci->numFolders) < 0) goto failed; for (i = 0; i < ci->numFolders; i++) { ci->folders[i].digest_defined = digest.defineds[i]; ci->folders[i].digest = digest.digests[i]; } /* * Must be kEnd. */ if ((p = header_bytes(a, 1)) == NULL) goto failed; if (*p != kEnd) goto failed; free_Digest(&digest); return (0); failed: free_Digest(&digest); return (-1); } static uint64_t folder_uncompressed_size(struct _7z_folder *f) { int n = (int)f->numOutStreams; unsigned pairs = (unsigned)f->numBindPairs; while (--n >= 0) { unsigned i; for (i = 0; i < pairs; i++) { if (f->bindPairs[i].outIndex == (uint64_t)n) break; } if (i >= pairs) return (f->unPackSize[n]); } return (0); } static void free_SubStreamsInfo(struct _7z_substream_info *ss) { free(ss->unpackSizes); free(ss->digestsDefined); free(ss->digests); } static int read_SubStreamsInfo(struct archive_read *a, struct _7z_substream_info *ss, struct _7z_folder *f, size_t numFolders) { const unsigned char *p; uint64_t *usizes; size_t unpack_streams; int type; unsigned i; uint32_t numDigests; memset(ss, 0, sizeof(*ss)); for (i = 0; i < numFolders; i++) f[i].numUnpackStreams = 1; if ((p = header_bytes(a, 1)) == NULL) return (-1); type = *p; if (type == kNumUnPackStream) { unpack_streams = 0; for (i = 0; i < numFolders; i++) { if (parse_7zip_uint64(a, &(f[i].numUnpackStreams)) < 0) return (-1); if (UMAX_ENTRY < f[i].numUnpackStreams) return (-1); if (unpack_streams > SIZE_MAX - UMAX_ENTRY) { return (-1); } unpack_streams += (size_t)f[i].numUnpackStreams; } if ((p = header_bytes(a, 1)) == NULL) return (-1); type = *p; } else unpack_streams = numFolders; ss->unpack_streams = unpack_streams; if (unpack_streams) { ss->unpackSizes = calloc(unpack_streams, sizeof(*ss->unpackSizes)); ss->digestsDefined = calloc(unpack_streams, sizeof(*ss->digestsDefined)); ss->digests = calloc(unpack_streams, sizeof(*ss->digests)); if (ss->unpackSizes == NULL || ss->digestsDefined == NULL || ss->digests == NULL) return (-1); } usizes = ss->unpackSizes; for (i = 0; i < numFolders; i++) { unsigned pack; uint64_t sum; if (f[i].numUnpackStreams == 0) continue; sum = 0; if (type == kSize) { for (pack = 1; pack < f[i].numUnpackStreams; pack++) { if (parse_7zip_uint64(a, usizes) < 0) return (-1); sum += *usizes++; } } *usizes++ = folder_uncompressed_size(&f[i]) - sum; } if (type == kSize) { if ((p = header_bytes(a, 1)) == NULL) return (-1); type = *p; } for (i = 0; i < unpack_streams; i++) { ss->digestsDefined[i] = 0; ss->digests[i] = 0; } numDigests = 0; for (i = 0; i < numFolders; i++) { if (f[i].numUnpackStreams != 1 || !f[i].digest_defined) numDigests += (uint32_t)f[i].numUnpackStreams; } if (type == kCRC) { struct _7z_digests tmpDigests; unsigned char *digestsDefined = ss->digestsDefined; uint32_t * digests = ss->digests; int di = 0; memset(&tmpDigests, 0, sizeof(tmpDigests)); if (read_Digests(a, &(tmpDigests), numDigests) < 0) { free_Digest(&tmpDigests); return (-1); } for (i = 0; i < numFolders; i++) { if (f[i].numUnpackStreams == 1 && f[i].digest_defined) { *digestsDefined++ = 1; *digests++ = f[i].digest; } else { unsigned j; for (j = 0; j < f[i].numUnpackStreams; j++, di++) { *digestsDefined++ = tmpDigests.defineds[di]; *digests++ = tmpDigests.digests[di]; } } } free_Digest(&tmpDigests); if ((p = header_bytes(a, 1)) == NULL) return (-1); type = *p; } /* * Must be kEnd. */ if (type != kEnd) return (-1); return (0); } static void free_StreamsInfo(struct _7z_stream_info *si) { free_PackInfo(&(si->pi)); free_CodersInfo(&(si->ci)); free_SubStreamsInfo(&(si->ss)); } static int read_StreamsInfo(struct archive_read *a, struct _7z_stream_info *si) { struct _7zip *zip = (struct _7zip *)a->format->data; const unsigned char *p; unsigned i; memset(si, 0, sizeof(*si)); if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p == kPackInfo) { uint64_t packPos; if (read_PackInfo(a, &(si->pi)) < 0) return (-1); if (si->pi.positions == NULL || si->pi.sizes == NULL) return (-1); /* * Calculate packed stream positions. */ packPos = si->pi.pos; for (i = 0; i < si->pi.numPackStreams; i++) { si->pi.positions[i] = packPos; packPos += si->pi.sizes[i]; if (packPos > zip->header_offset) return (-1); } if ((p = header_bytes(a, 1)) == NULL) return (-1); } if (*p == kUnPackInfo) { uint32_t packIndex; struct _7z_folder *f; if (read_CodersInfo(a, &(si->ci)) < 0) return (-1); /* * Calculate packed stream indexes. */ packIndex = 0; f = si->ci.folders; for (i = 0; i < si->ci.numFolders; i++) { f[i].packIndex = packIndex; packIndex += (uint32_t)f[i].numPackedStreams; if (packIndex > si->pi.numPackStreams) return (-1); } if ((p = header_bytes(a, 1)) == NULL) return (-1); } if (*p == kSubStreamsInfo) { if (read_SubStreamsInfo(a, &(si->ss), si->ci.folders, (size_t)si->ci.numFolders) < 0) return (-1); if ((p = header_bytes(a, 1)) == NULL) return (-1); } /* * Must be kEnd. */ if (*p != kEnd) return (-1); return (0); } static void free_Header(struct _7z_header_info *h) { free(h->emptyStreamBools); free(h->emptyFileBools); free(h->antiBools); free(h->attrBools); } static int read_Header(struct archive_read *a, struct _7z_header_info *h, int check_header_id) { struct _7zip *zip = (struct _7zip *)a->format->data; const unsigned char *p; struct _7z_folder *folders; struct _7z_stream_info *si = &(zip->si); struct _7zip_entry *entries; uint32_t folderIndex, indexInFolder; unsigned i; int eindex, empty_streams, sindex; if (check_header_id) { /* * Read Header. */ if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p != kHeader) return (-1); } /* * Read ArchiveProperties. */ if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p == kArchiveProperties) { for (;;) { uint64_t size; if ((p = header_bytes(a, 1)) == NULL) return (-1); if (*p == 0) break; if (parse_7zip_uint64(a, &size) < 0) return (-1); } if ((p = header_bytes(a, 1)) == NULL) return (-1); } /* * Read MainStreamsInfo. */ if (*p == kMainStreamsInfo) { if (read_StreamsInfo(a, &(zip->si)) < 0) return (-1); if ((p = header_bytes(a, 1)) == NULL) return (-1); } if (*p == kEnd) return (0); /* * Read FilesInfo. */ if (*p != kFilesInfo) return (-1); if (parse_7zip_uint64(a, &(zip->numFiles)) < 0) return (-1); if (UMAX_ENTRY < zip->numFiles) return (-1); zip->entries = calloc((size_t)zip->numFiles, sizeof(*zip->entries)); if (zip->entries == NULL) return (-1); entries = zip->entries; empty_streams = 0; for (;;) { int type; uint64_t size; size_t ll; if ((p = header_bytes(a, 1)) == NULL) return (-1); type = *p; if (type == kEnd) break; if (parse_7zip_uint64(a, &size) < 0) return (-1); if (zip->header_bytes_remaining < size) return (-1); ll = (size_t)size; switch (type) { case kEmptyStream: if (h->emptyStreamBools != NULL) return (-1); h->emptyStreamBools = calloc((size_t)zip->numFiles, sizeof(*h->emptyStreamBools)); if (h->emptyStreamBools == NULL) return (-1); if (read_Bools( a, h->emptyStreamBools, (size_t)zip->numFiles) < 0) return (-1); empty_streams = 0; for (i = 0; i < zip->numFiles; i++) { if (h->emptyStreamBools[i]) empty_streams++; } break; case kEmptyFile: if (empty_streams <= 0) { /* Unexcepted sequence. Skip this. */ if (header_bytes(a, ll) == NULL) return (-1); break; } if (h->emptyFileBools != NULL) return (-1); h->emptyFileBools = calloc(empty_streams, sizeof(*h->emptyFileBools)); if (h->emptyFileBools == NULL) return (-1); if (read_Bools(a, h->emptyFileBools, empty_streams) < 0) return (-1); break; case kAnti: if (empty_streams <= 0) { /* Unexcepted sequence. Skip this. */ if (header_bytes(a, ll) == NULL) return (-1); break; } if (h->antiBools != NULL) return (-1); h->antiBools = calloc(empty_streams, sizeof(*h->antiBools)); if (h->antiBools == NULL) return (-1); if (read_Bools(a, h->antiBools, empty_streams) < 0) return (-1); break; case kCTime: case kATime: case kMTime: if (read_Times(a, h, type) < 0) return (-1); break; case kName: { unsigned char *np; size_t nl, nb; /* Skip one byte. */ if ((p = header_bytes(a, 1)) == NULL) return (-1); ll--; if ((ll & 1) || ll < zip->numFiles * 4) return (-1); if (zip->entry_names != NULL) return (-1); zip->entry_names = malloc(ll); if (zip->entry_names == NULL) return (-1); np = zip->entry_names; nb = ll; /* * Copy whole file names. * NOTE: This loop prevents from expanding * the uncompressed buffer in order not to * use extra memory resource. */ while (nb) { size_t b; if (nb > UBUFF_SIZE) b = UBUFF_SIZE; else b = nb; if ((p = header_bytes(a, b)) == NULL) return (-1); memcpy(np, p, b); np += b; nb -= b; } np = zip->entry_names; nl = ll; for (i = 0; i < zip->numFiles; i++) { entries[i].utf16name = np; #if defined(_WIN32) && !defined(__CYGWIN__) && defined(_DEBUG) entries[i].wname = (wchar_t *)np; #endif /* Find a terminator. */ while (nl >= 2 && (np[0] || np[1])) { np += 2; nl -= 2; } if (nl < 2) return (-1);/* Terminator not found */ entries[i].name_len = np - entries[i].utf16name; np += 2; nl -= 2; } break; } case kAttributes: { int allAreDefined; if ((p = header_bytes(a, 2)) == NULL) return (-1); allAreDefined = *p; if (h->attrBools != NULL) return (-1); h->attrBools = calloc((size_t)zip->numFiles, sizeof(*h->attrBools)); if (h->attrBools == NULL) return (-1); if (allAreDefined) memset(h->attrBools, 1, (size_t)zip->numFiles); else { if (read_Bools(a, h->attrBools, (size_t)zip->numFiles) < 0) return (-1); } for (i = 0; i < zip->numFiles; i++) { if (h->attrBools[i]) { if ((p = header_bytes(a, 4)) == NULL) return (-1); entries[i].attr = archive_le32dec(p); } } break; } case kDummy: if (ll == 0) break; default: if (header_bytes(a, ll) == NULL) return (-1); break; } } /* * Set up entry's attributes. */ folders = si->ci.folders; eindex = sindex = 0; folderIndex = indexInFolder = 0; for (i = 0; i < zip->numFiles; i++) { if (h->emptyStreamBools == NULL || h->emptyStreamBools[i] == 0) entries[i].flg |= HAS_STREAM; /* The high 16 bits of attributes is a posix file mode. */ entries[i].mode = entries[i].attr >> 16; if (entries[i].flg & HAS_STREAM) { if ((size_t)sindex >= si->ss.unpack_streams) return (-1); if (entries[i].mode == 0) entries[i].mode = AE_IFREG | 0666; if (si->ss.digestsDefined[sindex]) entries[i].flg |= CRC32_IS_SET; entries[i].ssIndex = sindex; sindex++; } else { int dir; if (h->emptyFileBools == NULL) dir = 1; else { if (h->emptyFileBools[eindex]) dir = 0; else dir = 1; eindex++; } if (entries[i].mode == 0) { if (dir) entries[i].mode = AE_IFDIR | 0777; else entries[i].mode = AE_IFREG | 0666; } else if (dir && (entries[i].mode & AE_IFMT) != AE_IFDIR) { entries[i].mode &= ~AE_IFMT; entries[i].mode |= AE_IFDIR; } if ((entries[i].mode & AE_IFMT) == AE_IFDIR && entries[i].name_len >= 2 && (entries[i].utf16name[entries[i].name_len-2] != '/' || entries[i].utf16name[entries[i].name_len-1] != 0)) { entries[i].utf16name[entries[i].name_len] = '/'; entries[i].utf16name[entries[i].name_len+1] = 0; entries[i].name_len += 2; } entries[i].ssIndex = -1; } if (entries[i].attr & 0x01) entries[i].mode &= ~0222;/* Read only. */ if ((entries[i].flg & HAS_STREAM) == 0 && indexInFolder == 0) { /* * The entry is an empty file or a directory file, * those both have no contents. */ entries[i].folderIndex = -1; continue; } if (indexInFolder == 0) { for (;;) { if (folderIndex >= si->ci.numFolders) return (-1); if (folders[folderIndex].numUnpackStreams) break; folderIndex++; } } entries[i].folderIndex = folderIndex; if ((entries[i].flg & HAS_STREAM) == 0) continue; indexInFolder++; if (indexInFolder >= folders[folderIndex].numUnpackStreams) { folderIndex++; indexInFolder = 0; } } return (0); } #define EPOC_TIME ARCHIVE_LITERAL_ULL(116444736000000000) static void fileTimeToUtc(uint64_t fileTime, time_t *timep, long *ns) { if (fileTime >= EPOC_TIME) { fileTime -= EPOC_TIME; /* milli seconds base */ *timep = (time_t)(fileTime / 10000000); /* nano seconds base */ *ns = (long)(fileTime % 10000000) * 100; } else { *timep = 0; *ns = 0; } } static int read_Times(struct archive_read *a, struct _7z_header_info *h, int type) { struct _7zip *zip = (struct _7zip *)a->format->data; const unsigned char *p; struct _7zip_entry *entries = zip->entries; unsigned char *timeBools; int allAreDefined; unsigned i; timeBools = calloc((size_t)zip->numFiles, sizeof(*timeBools)); if (timeBools == NULL) return (-1); /* Read allAreDefined. */ if ((p = header_bytes(a, 1)) == NULL) goto failed; allAreDefined = *p; if (allAreDefined) memset(timeBools, 1, (size_t)zip->numFiles); else { if (read_Bools(a, timeBools, (size_t)zip->numFiles) < 0) goto failed; } /* Read external. */ if ((p = header_bytes(a, 1)) == NULL) goto failed; if (*p) { if (parse_7zip_uint64(a, &(h->dataIndex)) < 0) goto failed; if (UMAX_ENTRY < h->dataIndex) goto failed; } for (i = 0; i < zip->numFiles; i++) { if (!timeBools[i]) continue; if ((p = header_bytes(a, 8)) == NULL) goto failed; switch (type) { case kCTime: fileTimeToUtc(archive_le64dec(p), &(entries[i].ctime), &(entries[i].ctime_ns)); entries[i].flg |= CTIME_IS_SET; break; case kATime: fileTimeToUtc(archive_le64dec(p), &(entries[i].atime), &(entries[i].atime_ns)); entries[i].flg |= ATIME_IS_SET; break; case kMTime: fileTimeToUtc(archive_le64dec(p), &(entries[i].mtime), &(entries[i].mtime_ns)); entries[i].flg |= MTIME_IS_SET; break; } } free(timeBools); return (0); failed: free(timeBools); return (-1); } static int decode_encoded_header_info(struct archive_read *a, struct _7z_stream_info *si) { struct _7zip *zip = (struct _7zip *)a->format->data; errno = 0; if (read_StreamsInfo(a, si) < 0) { if (errno == ENOMEM) archive_set_error(&a->archive, -1, "Couldn't allocate memory"); else archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); return (ARCHIVE_FATAL); } if (si->pi.numPackStreams == 0 || si->ci.numFolders == 0) { archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); return (ARCHIVE_FATAL); } if (zip->header_offset < si->pi.pos + si->pi.sizes[0] || (int64_t)(si->pi.pos + si->pi.sizes[0]) < 0 || si->pi.sizes[0] == 0 || (int64_t)si->pi.pos < 0) { archive_set_error(&a->archive, -1, "Malformed Header offset"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } static const unsigned char * header_bytes(struct archive_read *a, size_t rbytes) { struct _7zip *zip = (struct _7zip *)a->format->data; const unsigned char *p; if (zip->header_bytes_remaining < rbytes) return (NULL); if (zip->pack_stream_bytes_unconsumed) read_consume(a); if (zip->header_is_encoded == 0) { p = __archive_read_ahead(a, rbytes, NULL); if (p == NULL) return (NULL); zip->header_bytes_remaining -= rbytes; zip->pack_stream_bytes_unconsumed = rbytes; } else { const void *buff; ssize_t bytes; bytes = read_stream(a, &buff, rbytes, rbytes); if (bytes <= 0) return (NULL); zip->header_bytes_remaining -= bytes; p = buff; } /* Update checksum */ zip->header_crc32 = crc32(zip->header_crc32, p, (unsigned)rbytes); return (p); } static int slurp_central_directory(struct archive_read *a, struct _7zip *zip, struct _7z_header_info *header) { const unsigned char *p; uint64_t next_header_offset; uint64_t next_header_size; uint32_t next_header_crc; ssize_t bytes_avail; int check_header_crc, r; if ((p = __archive_read_ahead(a, 32, &bytes_avail)) == NULL) return (ARCHIVE_FATAL); if ((p[0] == 'M' && p[1] == 'Z') || memcmp(p, "\x7F\x45LF", 4) == 0) { /* This is an executable ? Must be self-extracting... */ r = skip_sfx(a, bytes_avail); if (r < ARCHIVE_WARN) return (r); if ((p = __archive_read_ahead(a, 32, &bytes_avail)) == NULL) return (ARCHIVE_FATAL); } zip->seek_base += 32; if (memcmp(p, _7ZIP_SIGNATURE, 6) != 0) { archive_set_error(&a->archive, -1, "Not 7-Zip archive file"); return (ARCHIVE_FATAL); } /* CRC check. */ if (crc32(0, (const unsigned char *)p + 12, 20) != archive_le32dec(p + 8)) { archive_set_error(&a->archive, -1, "Header CRC error"); return (ARCHIVE_FATAL); } next_header_offset = archive_le64dec(p + 12); next_header_size = archive_le64dec(p + 20); next_header_crc = archive_le32dec(p + 28); if (next_header_size == 0) /* There is no entry in an archive file. */ return (ARCHIVE_EOF); if (((int64_t)next_header_offset) < 0) { archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); return (ARCHIVE_FATAL); } __archive_read_consume(a, 32); if (next_header_offset != 0) { if (bytes_avail >= (ssize_t)next_header_offset) __archive_read_consume(a, next_header_offset); else if (__archive_read_seek(a, next_header_offset + zip->seek_base, SEEK_SET) < 0) return (ARCHIVE_FATAL); } zip->stream_offset = next_header_offset; zip->header_offset = next_header_offset; zip->header_bytes_remaining = next_header_size; zip->header_crc32 = 0; zip->header_is_encoded = 0; zip->header_is_being_read = 1; zip->has_encrypted_entries = 0; check_header_crc = 1; if ((p = header_bytes(a, 1)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7-Zip file body"); return (ARCHIVE_FATAL); } /* Parse ArchiveProperties. */ switch (p[0]) { case kEncodedHeader: /* * The archive has an encoded header and we have to decode it * in order to parse the header correctly. */ r = decode_encoded_header_info(a, &(zip->si)); /* Check the EncodedHeader CRC.*/ if (r == 0 && zip->header_crc32 != next_header_crc) { archive_set_error(&a->archive, -1, "Damaged 7-Zip archive"); r = -1; } if (r == 0) { if (zip->si.ci.folders[0].digest_defined) next_header_crc = zip->si.ci.folders[0].digest; else check_header_crc = 0; if (zip->pack_stream_bytes_unconsumed) read_consume(a); r = setup_decode_folder(a, zip->si.ci.folders, 1); if (r == 0) { zip->header_bytes_remaining = zip->folder_outbytes_remaining; r = seek_pack(a); } } /* Clean up StreamsInfo. */ free_StreamsInfo(&(zip->si)); memset(&(zip->si), 0, sizeof(zip->si)); if (r < 0) return (ARCHIVE_FATAL); zip->header_is_encoded = 1; zip->header_crc32 = 0; /* FALL THROUGH */ case kHeader: /* * Parse the header. */ errno = 0; r = read_Header(a, header, zip->header_is_encoded); if (r < 0) { if (errno == ENOMEM) archive_set_error(&a->archive, -1, "Couldn't allocate memory"); else archive_set_error(&a->archive, -1, "Damaged 7-Zip archive"); return (ARCHIVE_FATAL); } /* * Must be kEnd. */ if ((p = header_bytes(a, 1)) == NULL ||*p != kEnd) { archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); return (ARCHIVE_FATAL); } /* Check the Header CRC.*/ if (check_header_crc && zip->header_crc32 != next_header_crc) { archive_set_error(&a->archive, -1, "Malformed 7-Zip archive"); return (ARCHIVE_FATAL); } break; default: archive_set_error(&a->archive, -1, "Unexpected Property ID = %X", p[0]); return (ARCHIVE_FATAL); } /* Clean up variables be used for decoding the archive header */ zip->pack_stream_remaining = 0; zip->pack_stream_index = 0; zip->folder_outbytes_remaining = 0; zip->uncompressed_buffer_bytes_remaining = 0; zip->pack_stream_bytes_unconsumed = 0; zip->header_is_being_read = 0; return (ARCHIVE_OK); } static ssize_t get_uncompressed_data(struct archive_read *a, const void **buff, size_t size, size_t minimum) { struct _7zip *zip = (struct _7zip *)a->format->data; ssize_t bytes_avail; if (zip->codec == _7Z_COPY && zip->codec2 == (unsigned long)-1) { /* Copy mode. */ /* * Note: '1' here is a performance optimization. * Recall that the decompression layer returns a count of * available bytes; asking for more than that forces the * decompressor to combine reads by copying data. */ *buff = __archive_read_ahead(a, 1, &bytes_avail); if (bytes_avail <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7-Zip file data"); return (ARCHIVE_FATAL); } if ((size_t)bytes_avail > zip->uncompressed_buffer_bytes_remaining) bytes_avail = (ssize_t) zip->uncompressed_buffer_bytes_remaining; if ((size_t)bytes_avail > size) bytes_avail = (ssize_t)size; zip->pack_stream_bytes_unconsumed = bytes_avail; } else if (zip->uncompressed_buffer_pointer == NULL) { /* Decompression has failed. */ archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Damaged 7-Zip archive"); return (ARCHIVE_FATAL); } else { /* Packed mode. */ if (minimum > zip->uncompressed_buffer_bytes_remaining) { /* * If remaining uncompressed data size is less than * the minimum size, fill the buffer up to the * minimum size. */ if (extract_pack_stream(a, minimum) < 0) return (ARCHIVE_FATAL); } if (size > zip->uncompressed_buffer_bytes_remaining) bytes_avail = (ssize_t) zip->uncompressed_buffer_bytes_remaining; else bytes_avail = (ssize_t)size; *buff = zip->uncompressed_buffer_pointer; zip->uncompressed_buffer_pointer += bytes_avail; } zip->uncompressed_buffer_bytes_remaining -= bytes_avail; return (bytes_avail); } static ssize_t extract_pack_stream(struct archive_read *a, size_t minimum) { struct _7zip *zip = (struct _7zip *)a->format->data; ssize_t bytes_avail; int r; if (zip->codec == _7Z_COPY && zip->codec2 == (unsigned long)-1) { if (minimum == 0) minimum = 1; if (__archive_read_ahead(a, minimum, &bytes_avail) == NULL || bytes_avail <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7-Zip file body"); return (ARCHIVE_FATAL); } if (bytes_avail > (ssize_t)zip->pack_stream_inbytes_remaining) bytes_avail = (ssize_t)zip->pack_stream_inbytes_remaining; zip->pack_stream_inbytes_remaining -= bytes_avail; if (bytes_avail > (ssize_t)zip->folder_outbytes_remaining) bytes_avail = (ssize_t)zip->folder_outbytes_remaining; zip->folder_outbytes_remaining -= bytes_avail; zip->uncompressed_buffer_bytes_remaining = bytes_avail; return (ARCHIVE_OK); } /* If the buffer hasn't been allocated, allocate it now. */ if (zip->uncompressed_buffer == NULL) { zip->uncompressed_buffer_size = UBUFF_SIZE; if (zip->uncompressed_buffer_size < minimum) { zip->uncompressed_buffer_size = minimum + 1023; zip->uncompressed_buffer_size &= ~0x3ff; } zip->uncompressed_buffer = malloc(zip->uncompressed_buffer_size); if (zip->uncompressed_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for 7-Zip decompression"); return (ARCHIVE_FATAL); } zip->uncompressed_buffer_bytes_remaining = 0; } else if (zip->uncompressed_buffer_size < minimum || zip->uncompressed_buffer_bytes_remaining < minimum) { /* * Make sure the uncompressed buffer can have bytes * at least `minimum' bytes. * NOTE: This case happen when reading the header. */ size_t used; if (zip->uncompressed_buffer_pointer != 0) used = zip->uncompressed_buffer_pointer - zip->uncompressed_buffer; else used = 0; if (zip->uncompressed_buffer_size < minimum) { /* * Expand the uncompressed buffer up to * the minimum size. */ void *p; size_t new_size; new_size = minimum + 1023; new_size &= ~0x3ff; p = realloc(zip->uncompressed_buffer, new_size); if (p == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for 7-Zip decompression"); return (ARCHIVE_FATAL); } zip->uncompressed_buffer = (unsigned char *)p; zip->uncompressed_buffer_size = new_size; } /* * Move unconsumed bytes to the head. */ if (used) { memmove(zip->uncompressed_buffer, zip->uncompressed_buffer + used, zip->uncompressed_buffer_bytes_remaining); } } else zip->uncompressed_buffer_bytes_remaining = 0; zip->uncompressed_buffer_pointer = NULL; for (;;) { size_t bytes_in, bytes_out; const void *buff_in; unsigned char *buff_out; int end_of_data; /* * Note: '1' here is a performance optimization. * Recall that the decompression layer returns a count of * available bytes; asking for more than that forces the * decompressor to combine reads by copying data. */ buff_in = __archive_read_ahead(a, 1, &bytes_avail); if (bytes_avail <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7-Zip file body"); return (ARCHIVE_FATAL); } buff_out = zip->uncompressed_buffer + zip->uncompressed_buffer_bytes_remaining; bytes_out = zip->uncompressed_buffer_size - zip->uncompressed_buffer_bytes_remaining; bytes_in = bytes_avail; if (bytes_in > zip->pack_stream_inbytes_remaining) bytes_in = (size_t)zip->pack_stream_inbytes_remaining; /* Drive decompression. */ r = decompress(a, zip, buff_out, &bytes_out, buff_in, &bytes_in); switch (r) { case ARCHIVE_OK: end_of_data = 0; break; case ARCHIVE_EOF: end_of_data = 1; break; default: return (ARCHIVE_FATAL); } zip->pack_stream_inbytes_remaining -= bytes_in; if (bytes_out > zip->folder_outbytes_remaining) bytes_out = (size_t)zip->folder_outbytes_remaining; zip->folder_outbytes_remaining -= bytes_out; zip->uncompressed_buffer_bytes_remaining += bytes_out; zip->pack_stream_bytes_unconsumed = bytes_in; /* * Continue decompression until uncompressed_buffer is full. */ if (zip->uncompressed_buffer_bytes_remaining == zip->uncompressed_buffer_size) break; if (zip->codec2 == _7Z_X86 && zip->odd_bcj_size && zip->uncompressed_buffer_bytes_remaining + 5 > zip->uncompressed_buffer_size) break; if (zip->pack_stream_inbytes_remaining == 0 && zip->folder_outbytes_remaining == 0) break; if (end_of_data || (bytes_in == 0 && bytes_out == 0)) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Damaged 7-Zip archive"); return (ARCHIVE_FATAL); } read_consume(a); } if (zip->uncompressed_buffer_bytes_remaining < minimum) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Damaged 7-Zip archive"); return (ARCHIVE_FATAL); } zip->uncompressed_buffer_pointer = zip->uncompressed_buffer; return (ARCHIVE_OK); } static int seek_pack(struct archive_read *a) { struct _7zip *zip = (struct _7zip *)a->format->data; int64_t pack_offset; if (zip->pack_stream_remaining <= 0) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Damaged 7-Zip archive"); return (ARCHIVE_FATAL); } zip->pack_stream_inbytes_remaining = zip->si.pi.sizes[zip->pack_stream_index]; pack_offset = zip->si.pi.positions[zip->pack_stream_index]; if (zip->stream_offset != pack_offset) { if (0 > __archive_read_seek(a, pack_offset + zip->seek_base, SEEK_SET)) return (ARCHIVE_FATAL); zip->stream_offset = pack_offset; } zip->pack_stream_index++; zip->pack_stream_remaining--; return (ARCHIVE_OK); } static ssize_t read_stream(struct archive_read *a, const void **buff, size_t size, size_t minimum) { struct _7zip *zip = (struct _7zip *)a->format->data; uint64_t skip_bytes = 0; ssize_t r; if (zip->uncompressed_buffer_bytes_remaining == 0) { if (zip->pack_stream_inbytes_remaining > 0) { r = extract_pack_stream(a, 0); if (r < 0) return (r); return (get_uncompressed_data(a, buff, size, minimum)); } else if (zip->folder_outbytes_remaining > 0) { /* Extract a remaining pack stream. */ r = extract_pack_stream(a, 0); if (r < 0) return (r); return (get_uncompressed_data(a, buff, size, minimum)); } } else return (get_uncompressed_data(a, buff, size, minimum)); /* * Current pack stream has been consumed. */ if (zip->pack_stream_remaining == 0) { if (zip->header_is_being_read) { /* Invalid sequence. This might happen when * reading a malformed archive. */ archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Malformed 7-Zip archive"); return (ARCHIVE_FATAL); } /* * All current folder's pack streams have been * consumed. Switch to next folder. */ if (zip->folder_index == 0 && (zip->si.ci.folders[zip->entry->folderIndex].skipped_bytes || zip->folder_index != zip->entry->folderIndex)) { zip->folder_index = zip->entry->folderIndex; skip_bytes = zip->si.ci.folders[zip->folder_index].skipped_bytes; } if (zip->folder_index >= zip->si.ci.numFolders) { /* * We have consumed all folders and its pack streams. */ *buff = NULL; return (0); } r = setup_decode_folder(a, &(zip->si.ci.folders[zip->folder_index]), 0); if (r != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->folder_index++; } /* * Switch to next pack stream. */ r = seek_pack(a); if (r < 0) return (r); /* Extract a new pack stream. */ r = extract_pack_stream(a, 0); if (r < 0) return (r); /* - * Skip the bytes we alrady has skipped in skip_stream(). + * Skip the bytes we already has skipped in skip_stream(). */ while (skip_bytes) { ssize_t skipped; if (zip->uncompressed_buffer_bytes_remaining == 0) { if (zip->pack_stream_inbytes_remaining > 0) { r = extract_pack_stream(a, 0); if (r < 0) return (r); } else if (zip->folder_outbytes_remaining > 0) { /* Extract a remaining pack stream. */ r = extract_pack_stream(a, 0); if (r < 0) return (r); } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7-Zip file body"); return (ARCHIVE_FATAL); } } skipped = get_uncompressed_data( a, buff, (size_t)skip_bytes, 0); if (skipped < 0) return (skipped); skip_bytes -= skipped; if (zip->pack_stream_bytes_unconsumed) read_consume(a); } return (get_uncompressed_data(a, buff, size, minimum)); } static int setup_decode_folder(struct archive_read *a, struct _7z_folder *folder, int header) { struct _7zip *zip = (struct _7zip *)a->format->data; const struct _7z_coder *coder1, *coder2; const char *cname = (header)?"archive header":"file content"; unsigned i; int r, found_bcj2 = 0; /* * Release the memory which the previous folder used for BCJ2. */ for (i = 0; i < 3; i++) { if (zip->sub_stream_buff[i] != NULL) free(zip->sub_stream_buff[i]); zip->sub_stream_buff[i] = NULL; } /* * Initialize a stream reader. */ zip->pack_stream_remaining = (unsigned)folder->numPackedStreams; zip->pack_stream_index = (unsigned)folder->packIndex; zip->folder_outbytes_remaining = folder_uncompressed_size(folder); zip->uncompressed_buffer_bytes_remaining = 0; /* * Check coder types. */ for (i = 0; i < folder->numCoders; i++) { switch(folder->coders[i].codec) { case _7Z_CRYPTO_MAIN_ZIP: case _7Z_CRYPTO_RAR_29: case _7Z_CRYPTO_AES_256_SHA_256: { /* For entry that is associated with this folder, mark it as encrypted (data+metadata). */ zip->has_encrypted_entries = 1; if (a->entry) { archive_entry_set_is_data_encrypted(a->entry, 1); archive_entry_set_is_metadata_encrypted(a->entry, 1); } archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "The %s is encrypted, " "but currently not supported", cname); return (ARCHIVE_FATAL); } case _7Z_X86_BCJ2: { found_bcj2++; break; } } } /* Now that we've checked for encryption, if there were still no * encrypted entries found we can say for sure that there are none. */ if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { zip->has_encrypted_entries = 0; } if ((folder->numCoders > 2 && !found_bcj2) || found_bcj2 > 1) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "The %s is encoded with many filters, " "but currently not supported", cname); return (ARCHIVE_FATAL); } coder1 = &(folder->coders[0]); if (folder->numCoders == 2) coder2 = &(folder->coders[1]); else coder2 = NULL; if (found_bcj2) { /* * Preparation to decode BCJ2. * Decoding BCJ2 requires four sources. Those are at least, * as far as I know, two types of the storage form. */ const struct _7z_coder *fc = folder->coders; static const struct _7z_coder coder_copy = {0, 1, 1, 0, NULL}; const struct _7z_coder *scoder[3] = {&coder_copy, &coder_copy, &coder_copy}; const void *buff; ssize_t bytes; unsigned char *b[3] = {NULL, NULL, NULL}; uint64_t sunpack[3] ={-1, -1, -1}; size_t s[3] = {0, 0, 0}; int idx[3] = {0, 1, 2}; if (folder->numCoders == 4 && fc[3].codec == _7Z_X86_BCJ2 && folder->numInStreams == 7 && folder->numOutStreams == 4 && zip->pack_stream_remaining == 4) { /* Source type 1 made by 7zr or 7z with -m options. */ if (folder->bindPairs[0].inIndex == 5) { /* The form made by 7zr */ idx[0] = 1; idx[1] = 2; idx[2] = 0; scoder[1] = &(fc[1]); scoder[2] = &(fc[0]); sunpack[1] = folder->unPackSize[1]; sunpack[2] = folder->unPackSize[0]; coder1 = &(fc[2]); } else { /* * NOTE: Some patterns do not work. * work: * 7z a -m0=BCJ2 -m1=COPY -m2=COPY * -m3=(any) * 7z a -m0=BCJ2 -m1=COPY -m2=(any) * -m3=COPY * 7z a -m0=BCJ2 -m1=(any) -m2=COPY * -m3=COPY * not work: * other patterns. * * We have to handle this like `pipe' or * our libarchive7s filter frame work, * decoding the BCJ2 main stream sequentially, * m3 -> m2 -> m1 -> BCJ2. * */ if (fc[0].codec == _7Z_COPY && fc[1].codec == _7Z_COPY) coder1 = &(folder->coders[2]); else if (fc[0].codec == _7Z_COPY && fc[2].codec == _7Z_COPY) coder1 = &(folder->coders[1]); else if (fc[1].codec == _7Z_COPY && fc[2].codec == _7Z_COPY) coder1 = &(folder->coders[0]); else { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Unsupported form of " "BCJ2 streams"); return (ARCHIVE_FATAL); } } coder2 = &(fc[3]); zip->main_stream_bytes_remaining = (size_t)folder->unPackSize[2]; } else if (coder2 != NULL && coder2->codec == _7Z_X86_BCJ2 && zip->pack_stream_remaining == 4 && folder->numInStreams == 5 && folder->numOutStreams == 2) { /* Source type 0 made by 7z */ zip->main_stream_bytes_remaining = (size_t)folder->unPackSize[0]; } else { /* We got an unexpected form. */ archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Unsupported form of BCJ2 streams"); return (ARCHIVE_FATAL); } /* Skip the main stream at this time. */ if ((r = seek_pack(a)) < 0) return (r); zip->pack_stream_bytes_unconsumed = (size_t)zip->pack_stream_inbytes_remaining; read_consume(a); /* Read following three sub streams. */ for (i = 0; i < 3; i++) { const struct _7z_coder *coder = scoder[i]; if ((r = seek_pack(a)) < 0) { free(b[0]); free(b[1]); free(b[2]); return (r); } if (sunpack[i] == (uint64_t)-1) zip->folder_outbytes_remaining = zip->pack_stream_inbytes_remaining; else zip->folder_outbytes_remaining = sunpack[i]; r = init_decompression(a, zip, coder, NULL); if (r != ARCHIVE_OK) { free(b[0]); free(b[1]); free(b[2]); return (ARCHIVE_FATAL); } - /* Allocate memory for the decorded data of a sub + /* Allocate memory for the decoded data of a sub * stream. */ b[i] = malloc((size_t)zip->folder_outbytes_remaining); if (b[i] == NULL) { free(b[0]); free(b[1]); free(b[2]); archive_set_error(&a->archive, ENOMEM, "No memory for 7-Zip decompression"); return (ARCHIVE_FATAL); } /* Extract a sub stream. */ while (zip->pack_stream_inbytes_remaining > 0) { r = (int)extract_pack_stream(a, 0); if (r < 0) { free(b[0]); free(b[1]); free(b[2]); return (r); } bytes = get_uncompressed_data(a, &buff, zip->uncompressed_buffer_bytes_remaining, 0); if (bytes < 0) { free(b[0]); free(b[1]); free(b[2]); return ((int)bytes); } memcpy(b[i]+s[i], buff, bytes); s[i] += bytes; if (zip->pack_stream_bytes_unconsumed) read_consume(a); } } /* Set the sub streams to the right place. */ for (i = 0; i < 3; i++) { zip->sub_stream_buff[i] = b[idx[i]]; zip->sub_stream_size[i] = s[idx[i]]; zip->sub_stream_bytes_remaining[i] = s[idx[i]]; } /* Allocate memory used for decoded main stream bytes. */ if (zip->tmp_stream_buff == NULL) { zip->tmp_stream_buff_size = 32 * 1024; zip->tmp_stream_buff = malloc(zip->tmp_stream_buff_size); if (zip->tmp_stream_buff == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for 7-Zip decompression"); return (ARCHIVE_FATAL); } } zip->tmp_stream_bytes_avail = 0; zip->tmp_stream_bytes_remaining = 0; zip->odd_bcj_size = 0; zip->bcj2_outPos = 0; /* * Reset a stream reader in order to read the main stream * of BCJ2. */ zip->pack_stream_remaining = 1; zip->pack_stream_index = (unsigned)folder->packIndex; zip->folder_outbytes_remaining = folder_uncompressed_size(folder); zip->uncompressed_buffer_bytes_remaining = 0; } /* * Initialize the decompressor for the new folder's pack streams. */ r = init_decompression(a, zip, coder1, coder2); if (r != ARCHIVE_OK) return (ARCHIVE_FATAL); return (ARCHIVE_OK); } static int64_t skip_stream(struct archive_read *a, size_t skip_bytes) { struct _7zip *zip = (struct _7zip *)a->format->data; const void *p; int64_t skipped_bytes; size_t bytes = skip_bytes; if (zip->folder_index == 0) { /* * Optimization for a list mode. - * Avoid unncecessary decoding operations. + * Avoid unnecessary decoding operations. */ zip->si.ci.folders[zip->entry->folderIndex].skipped_bytes += skip_bytes; return (skip_bytes); } while (bytes) { skipped_bytes = read_stream(a, &p, bytes, 0); if (skipped_bytes < 0) return (skipped_bytes); if (skipped_bytes == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated 7-Zip file body"); return (ARCHIVE_FATAL); } bytes -= (size_t)skipped_bytes; if (zip->pack_stream_bytes_unconsumed) read_consume(a); } return (skip_bytes); } /* * Brought from LZMA SDK. * * Bra86.c -- Converter for x86 code (BCJ) * 2008-10-04 : Igor Pavlov : Public domain * */ #define Test86MSByte(b) ((b) == 0 || (b) == 0xFF) static void x86_Init(struct _7zip *zip) { zip->bcj_state = 0; zip->bcj_prevPosT = (size_t)0 - 1; zip->bcj_prevMask = 0; zip->bcj_ip = 5; } static size_t x86_Convert(struct _7zip *zip, uint8_t *data, size_t size) { static const uint8_t kMaskToAllowedStatus[8] = {1, 1, 1, 0, 1, 0, 0, 0}; static const uint8_t kMaskToBitNumber[8] = {0, 1, 2, 2, 3, 3, 3, 3}; size_t bufferPos, prevPosT; uint32_t ip, prevMask; if (size < 5) return 0; bufferPos = 0; prevPosT = zip->bcj_prevPosT; prevMask = zip->bcj_prevMask; ip = zip->bcj_ip; for (;;) { uint8_t *p = data + bufferPos; uint8_t *limit = data + size - 4; for (; p < limit; p++) if ((*p & 0xFE) == 0xE8) break; bufferPos = (size_t)(p - data); if (p >= limit) break; prevPosT = bufferPos - prevPosT; if (prevPosT > 3) prevMask = 0; else { prevMask = (prevMask << ((int)prevPosT - 1)) & 0x7; if (prevMask != 0) { unsigned char b = p[4 - kMaskToBitNumber[prevMask]]; if (!kMaskToAllowedStatus[prevMask] || Test86MSByte(b)) { prevPosT = bufferPos; prevMask = ((prevMask << 1) & 0x7) | 1; bufferPos++; continue; } } } prevPosT = bufferPos; if (Test86MSByte(p[4])) { uint32_t src = ((uint32_t)p[4] << 24) | ((uint32_t)p[3] << 16) | ((uint32_t)p[2] << 8) | ((uint32_t)p[1]); uint32_t dest; for (;;) { uint8_t b; int b_index; dest = src - (ip + (uint32_t)bufferPos); if (prevMask == 0) break; b_index = kMaskToBitNumber[prevMask] * 8; b = (uint8_t)(dest >> (24 - b_index)); if (!Test86MSByte(b)) break; src = dest ^ ((1 << (32 - b_index)) - 1); } p[4] = (uint8_t)(~(((dest >> 24) & 1) - 1)); p[3] = (uint8_t)(dest >> 16); p[2] = (uint8_t)(dest >> 8); p[1] = (uint8_t)dest; bufferPos += 5; } else { prevMask = ((prevMask << 1) & 0x7) | 1; bufferPos++; } } zip->bcj_prevPosT = prevPosT; zip->bcj_prevMask = prevMask; zip->bcj_ip += (uint32_t)bufferPos; return (bufferPos); } /* * Brought from LZMA SDK. * * Bcj2.c -- Converter for x86 code (BCJ2) * 2008-10-04 : Igor Pavlov : Public domain * */ #define SZ_ERROR_DATA ARCHIVE_FAILED #define IsJcc(b0, b1) ((b0) == 0x0F && ((b1) & 0xF0) == 0x80) #define IsJ(b0, b1) ((b1 & 0xFE) == 0xE8 || IsJcc(b0, b1)) #define kNumTopBits 24 #define kTopValue ((uint32_t)1 << kNumTopBits) #define kNumBitModelTotalBits 11 #define kBitModelTotal (1 << kNumBitModelTotalBits) #define kNumMoveBits 5 #define RC_READ_BYTE (*buffer++) #define RC_TEST { if (buffer == bufferLim) return SZ_ERROR_DATA; } #define RC_INIT2 zip->bcj2_code = 0; zip->bcj2_range = 0xFFFFFFFF; \ { int ii; for (ii = 0; ii < 5; ii++) { RC_TEST; zip->bcj2_code = (zip->bcj2_code << 8) | RC_READ_BYTE; }} #define NORMALIZE if (zip->bcj2_range < kTopValue) { RC_TEST; zip->bcj2_range <<= 8; zip->bcj2_code = (zip->bcj2_code << 8) | RC_READ_BYTE; } #define IF_BIT_0(p) ttt = *(p); bound = (zip->bcj2_range >> kNumBitModelTotalBits) * ttt; if (zip->bcj2_code < bound) #define UPDATE_0(p) zip->bcj2_range = bound; *(p) = (CProb)(ttt + ((kBitModelTotal - ttt) >> kNumMoveBits)); NORMALIZE; #define UPDATE_1(p) zip->bcj2_range -= bound; zip->bcj2_code -= bound; *(p) = (CProb)(ttt - (ttt >> kNumMoveBits)); NORMALIZE; static ssize_t Bcj2_Decode(struct _7zip *zip, uint8_t *outBuf, size_t outSize) { size_t inPos = 0, outPos = 0; const uint8_t *buf0, *buf1, *buf2, *buf3; size_t size0, size1, size2, size3; const uint8_t *buffer, *bufferLim; unsigned int i, j; size0 = zip->tmp_stream_bytes_remaining; buf0 = zip->tmp_stream_buff + zip->tmp_stream_bytes_avail - size0; size1 = zip->sub_stream_bytes_remaining[0]; buf1 = zip->sub_stream_buff[0] + zip->sub_stream_size[0] - size1; size2 = zip->sub_stream_bytes_remaining[1]; buf2 = zip->sub_stream_buff[1] + zip->sub_stream_size[1] - size2; size3 = zip->sub_stream_bytes_remaining[2]; buf3 = zip->sub_stream_buff[2] + zip->sub_stream_size[2] - size3; buffer = buf3; bufferLim = buffer + size3; if (zip->bcj_state == 0) { /* * Initialize. */ zip->bcj2_prevByte = 0; for (i = 0; i < sizeof(zip->bcj2_p) / sizeof(zip->bcj2_p[0]); i++) zip->bcj2_p[i] = kBitModelTotal >> 1; RC_INIT2; zip->bcj_state = 1; } /* * Gather the odd bytes of a previous call. */ for (i = 0; zip->odd_bcj_size > 0 && outPos < outSize; i++) { outBuf[outPos++] = zip->odd_bcj[i]; zip->odd_bcj_size--; } if (outSize == 0) { zip->bcj2_outPos += outPos; return (outPos); } for (;;) { uint8_t b; CProb *prob; uint32_t bound; uint32_t ttt; size_t limit = size0 - inPos; if (outSize - outPos < limit) limit = outSize - outPos; if (zip->bcj_state == 1) { while (limit != 0) { uint8_t bb = buf0[inPos]; outBuf[outPos++] = bb; if (IsJ(zip->bcj2_prevByte, bb)) { zip->bcj_state = 2; break; } inPos++; zip->bcj2_prevByte = bb; limit--; } } if (limit == 0 || outPos == outSize) break; zip->bcj_state = 1; b = buf0[inPos++]; if (b == 0xE8) prob = zip->bcj2_p + zip->bcj2_prevByte; else if (b == 0xE9) prob = zip->bcj2_p + 256; else prob = zip->bcj2_p + 257; IF_BIT_0(prob) { UPDATE_0(prob) zip->bcj2_prevByte = b; } else { uint32_t dest; const uint8_t *v; uint8_t out[4]; UPDATE_1(prob) if (b == 0xE8) { v = buf1; if (size1 < 4) return SZ_ERROR_DATA; buf1 += 4; size1 -= 4; } else { v = buf2; if (size2 < 4) return SZ_ERROR_DATA; buf2 += 4; size2 -= 4; } dest = (((uint32_t)v[0] << 24) | ((uint32_t)v[1] << 16) | ((uint32_t)v[2] << 8) | ((uint32_t)v[3])) - ((uint32_t)zip->bcj2_outPos + (uint32_t)outPos + 4); out[0] = (uint8_t)dest; out[1] = (uint8_t)(dest >> 8); out[2] = (uint8_t)(dest >> 16); out[3] = zip->bcj2_prevByte = (uint8_t)(dest >> 24); for (i = 0; i < 4 && outPos < outSize; i++) outBuf[outPos++] = out[i]; if (i < 4) { /* * Save odd bytes which we could not add into * the output buffer because of out of space. */ zip->odd_bcj_size = 4 -i; for (; i < 4; i++) { j = i - 4 + (unsigned)zip->odd_bcj_size; zip->odd_bcj[j] = out[i]; } break; } } } zip->tmp_stream_bytes_remaining -= inPos; zip->sub_stream_bytes_remaining[0] = size1; zip->sub_stream_bytes_remaining[1] = size2; zip->sub_stream_bytes_remaining[2] = bufferLim - buffer; zip->bcj2_outPos += outPos; return ((ssize_t)outPos); } Index: head/contrib/libarchive/libarchive/archive_read_support_format_iso9660.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_format_iso9660.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_read_support_format_iso9660.c (revision 311903) @@ -1,3263 +1,3263 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * Copyright (c) 2009 Andreas Henriksson * Copyright (c) 2009-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif /* #include */ /* See archive_platform.h */ #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include #ifdef HAVE_ZLIB_H #include #endif #include "archive.h" #include "archive_endian.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_read_private.h" #include "archive_string.h" /* * An overview of ISO 9660 format: * * Each disk is laid out as follows: * * 32k reserved for private use * * Volume descriptor table. Each volume descriptor * is 2k and specifies basic format information. * The "Primary Volume Descriptor" (PVD) is defined by the * standard and should always be present; other volume * descriptors include various vendor-specific extensions. * * Files and directories. Each file/dir is specified by * an "extent" (starting sector and length in bytes). * Dirs are just files with directory records packed one * after another. The PVD contains a single dir entry * specifying the location of the root directory. Everything * else follows from there. * * This module works by first reading the volume descriptors, then * building a list of directory entries, sorted by starting * sector. At each step, I look for the earliest dir entry that * hasn't yet been read, seek forward to that location and read * that entry. If it's a dir, I slurp in the new dir entries and * add them to the heap; if it's a regular file, I return the * corresponding archive_entry and wait for the client to request * the file body. This strategy allows us to read most compliant * CDs with a single pass through the data, as required by libarchive. */ #define LOGICAL_BLOCK_SIZE 2048 #define SYSTEM_AREA_BLOCK 16 /* Structure of on-disk primary volume descriptor. */ #define PVD_type_offset 0 #define PVD_type_size 1 #define PVD_id_offset (PVD_type_offset + PVD_type_size) #define PVD_id_size 5 #define PVD_version_offset (PVD_id_offset + PVD_id_size) #define PVD_version_size 1 #define PVD_reserved1_offset (PVD_version_offset + PVD_version_size) #define PVD_reserved1_size 1 #define PVD_system_id_offset (PVD_reserved1_offset + PVD_reserved1_size) #define PVD_system_id_size 32 #define PVD_volume_id_offset (PVD_system_id_offset + PVD_system_id_size) #define PVD_volume_id_size 32 #define PVD_reserved2_offset (PVD_volume_id_offset + PVD_volume_id_size) #define PVD_reserved2_size 8 #define PVD_volume_space_size_offset (PVD_reserved2_offset + PVD_reserved2_size) #define PVD_volume_space_size_size 8 #define PVD_reserved3_offset (PVD_volume_space_size_offset + PVD_volume_space_size_size) #define PVD_reserved3_size 32 #define PVD_volume_set_size_offset (PVD_reserved3_offset + PVD_reserved3_size) #define PVD_volume_set_size_size 4 #define PVD_volume_sequence_number_offset (PVD_volume_set_size_offset + PVD_volume_set_size_size) #define PVD_volume_sequence_number_size 4 #define PVD_logical_block_size_offset (PVD_volume_sequence_number_offset + PVD_volume_sequence_number_size) #define PVD_logical_block_size_size 4 #define PVD_path_table_size_offset (PVD_logical_block_size_offset + PVD_logical_block_size_size) #define PVD_path_table_size_size 8 #define PVD_type_1_path_table_offset (PVD_path_table_size_offset + PVD_path_table_size_size) #define PVD_type_1_path_table_size 4 #define PVD_opt_type_1_path_table_offset (PVD_type_1_path_table_offset + PVD_type_1_path_table_size) #define PVD_opt_type_1_path_table_size 4 #define PVD_type_m_path_table_offset (PVD_opt_type_1_path_table_offset + PVD_opt_type_1_path_table_size) #define PVD_type_m_path_table_size 4 #define PVD_opt_type_m_path_table_offset (PVD_type_m_path_table_offset + PVD_type_m_path_table_size) #define PVD_opt_type_m_path_table_size 4 #define PVD_root_directory_record_offset (PVD_opt_type_m_path_table_offset + PVD_opt_type_m_path_table_size) #define PVD_root_directory_record_size 34 #define PVD_volume_set_id_offset (PVD_root_directory_record_offset + PVD_root_directory_record_size) #define PVD_volume_set_id_size 128 #define PVD_publisher_id_offset (PVD_volume_set_id_offset + PVD_volume_set_id_size) #define PVD_publisher_id_size 128 #define PVD_preparer_id_offset (PVD_publisher_id_offset + PVD_publisher_id_size) #define PVD_preparer_id_size 128 #define PVD_application_id_offset (PVD_preparer_id_offset + PVD_preparer_id_size) #define PVD_application_id_size 128 #define PVD_copyright_file_id_offset (PVD_application_id_offset + PVD_application_id_size) #define PVD_copyright_file_id_size 37 #define PVD_abstract_file_id_offset (PVD_copyright_file_id_offset + PVD_copyright_file_id_size) #define PVD_abstract_file_id_size 37 #define PVD_bibliographic_file_id_offset (PVD_abstract_file_id_offset + PVD_abstract_file_id_size) #define PVD_bibliographic_file_id_size 37 #define PVD_creation_date_offset (PVD_bibliographic_file_id_offset + PVD_bibliographic_file_id_size) #define PVD_creation_date_size 17 #define PVD_modification_date_offset (PVD_creation_date_offset + PVD_creation_date_size) #define PVD_modification_date_size 17 #define PVD_expiration_date_offset (PVD_modification_date_offset + PVD_modification_date_size) #define PVD_expiration_date_size 17 #define PVD_effective_date_offset (PVD_expiration_date_offset + PVD_expiration_date_size) #define PVD_effective_date_size 17 #define PVD_file_structure_version_offset (PVD_effective_date_offset + PVD_effective_date_size) #define PVD_file_structure_version_size 1 #define PVD_reserved4_offset (PVD_file_structure_version_offset + PVD_file_structure_version_size) #define PVD_reserved4_size 1 #define PVD_application_data_offset (PVD_reserved4_offset + PVD_reserved4_size) #define PVD_application_data_size 512 #define PVD_reserved5_offset (PVD_application_data_offset + PVD_application_data_size) #define PVD_reserved5_size (2048 - PVD_reserved5_offset) /* TODO: It would make future maintenance easier to just hardcode the * above values. In particular, ECMA119 states the offsets as part of * the standard. That would eliminate the need for the following check.*/ #if PVD_reserved5_offset != 1395 #error PVD offset and size definitions are wrong. #endif /* Structure of optional on-disk supplementary volume descriptor. */ #define SVD_type_offset 0 #define SVD_type_size 1 #define SVD_id_offset (SVD_type_offset + SVD_type_size) #define SVD_id_size 5 #define SVD_version_offset (SVD_id_offset + SVD_id_size) #define SVD_version_size 1 /* ... */ #define SVD_reserved1_offset 72 #define SVD_reserved1_size 8 #define SVD_volume_space_size_offset 80 #define SVD_volume_space_size_size 8 #define SVD_escape_sequences_offset (SVD_volume_space_size_offset + SVD_volume_space_size_size) #define SVD_escape_sequences_size 32 /* ... */ #define SVD_logical_block_size_offset 128 #define SVD_logical_block_size_size 4 #define SVD_type_L_path_table_offset 140 #define SVD_type_M_path_table_offset 148 /* ... */ #define SVD_root_directory_record_offset 156 #define SVD_root_directory_record_size 34 #define SVD_file_structure_version_offset 881 #define SVD_reserved2_offset 882 #define SVD_reserved2_size 1 #define SVD_reserved3_offset 1395 #define SVD_reserved3_size 653 /* ... */ /* FIXME: validate correctness of last SVD entry offset. */ /* Structure of an on-disk directory record. */ /* Note: ISO9660 stores each multi-byte integer twice, once in * each byte order. The sizes here are the size of just one * of the two integers. (This is why the offset of a field isn't * the same as the offset+size of the previous field.) */ #define DR_length_offset 0 #define DR_length_size 1 #define DR_ext_attr_length_offset 1 #define DR_ext_attr_length_size 1 #define DR_extent_offset 2 #define DR_extent_size 4 #define DR_size_offset 10 #define DR_size_size 4 #define DR_date_offset 18 #define DR_date_size 7 #define DR_flags_offset 25 #define DR_flags_size 1 #define DR_file_unit_size_offset 26 #define DR_file_unit_size_size 1 #define DR_interleave_offset 27 #define DR_interleave_size 1 #define DR_volume_sequence_number_offset 28 #define DR_volume_sequence_number_size 2 #define DR_name_len_offset 32 #define DR_name_len_size 1 #define DR_name_offset 33 #ifdef HAVE_ZLIB_H static const unsigned char zisofs_magic[8] = { 0x37, 0xE4, 0x53, 0x96, 0xC9, 0xDB, 0xD6, 0x07 }; struct zisofs { /* Set 1 if this file compressed by paged zlib */ int pz; int pz_log2_bs; /* Log2 of block size */ uint64_t pz_uncompressed_size; int initialized; unsigned char *uncompressed_buffer; size_t uncompressed_buffer_size; uint32_t pz_offset; unsigned char header[16]; size_t header_avail; int header_passed; unsigned char *block_pointers; size_t block_pointers_alloc; size_t block_pointers_size; size_t block_pointers_avail; size_t block_off; uint32_t block_avail; z_stream stream; int stream_valid; }; #else struct zisofs { /* Set 1 if this file compressed by paged zlib */ int pz; }; #endif struct content { uint64_t offset;/* Offset on disk. */ uint64_t size; /* File size in bytes. */ struct content *next; }; /* In-memory storage for a directory record. */ struct file_info { struct file_info *use_next; struct file_info *parent; struct file_info *next; struct file_info *re_next; int subdirs; uint64_t key; /* Heap Key. */ uint64_t offset; /* Offset on disk. */ uint64_t size; /* File size in bytes. */ uint32_t ce_offset; /* Offset of CE. */ uint32_t ce_size; /* Size of CE. */ char rr_moved; /* Flag to rr_moved. */ char rr_moved_has_re_only; char re; /* Having RRIP "RE" extension. */ char re_descendant; uint64_t cl_offset; /* Having RRIP "CL" extension. */ int birthtime_is_set; time_t birthtime; /* File created time. */ time_t mtime; /* File last modified time. */ time_t atime; /* File last accessed time. */ time_t ctime; /* File attribute change time. */ uint64_t rdev; /* Device number. */ mode_t mode; uid_t uid; gid_t gid; int64_t number; int nlinks; struct archive_string name; /* Pathname */ unsigned char *utf16be_name; size_t utf16be_bytes; char name_continues; /* Non-zero if name continues */ struct archive_string symlink; char symlink_continues; /* Non-zero if link continues */ /* Set 1 if this file compressed by paged zlib(zisofs) */ int pz; int pz_log2_bs; /* Log2 of block size */ uint64_t pz_uncompressed_size; /* Set 1 if this file is multi extent. */ int multi_extent; struct { struct content *first; struct content **last; } contents; struct { struct file_info *first; struct file_info **last; } rede_files; }; struct heap_queue { struct file_info **files; int allocated; int used; }; struct iso9660 { int magic; #define ISO9660_MAGIC 0x96609660 int opt_support_joliet; int opt_support_rockridge; struct archive_string pathname; char seenRockridge; /* Set true if RR extensions are used. */ - char seenSUSP; /* Set true if SUSP is beging used. */ + char seenSUSP; /* Set true if SUSP is being used. */ char seenJoliet; unsigned char suspOffset; struct file_info *rr_moved; struct read_ce_queue { struct read_ce_req { uint64_t offset;/* Offset of CE on disk. */ struct file_info *file; } *reqs; int cnt; int allocated; } read_ce_req; int64_t previous_number; struct archive_string previous_pathname; struct file_info *use_files; struct heap_queue pending_files; struct { struct file_info *first; struct file_info **last; } cache_files; struct { struct file_info *first; struct file_info **last; } re_files; uint64_t current_position; ssize_t logical_block_size; uint64_t volume_size; /* Total size of volume in bytes. */ int32_t volume_block;/* Total size of volume in logical blocks. */ struct vd { int location; /* Location of Extent. */ uint32_t size; } primary, joliet; int64_t entry_sparse_offset; int64_t entry_bytes_remaining; size_t entry_bytes_unconsumed; struct zisofs entry_zisofs; struct content *entry_content; struct archive_string_conv *sconv_utf16be; /* * Buffers for a full pathname in UTF-16BE in Joliet extensions. */ #define UTF16_NAME_MAX 1024 unsigned char *utf16be_path; size_t utf16be_path_len; unsigned char *utf16be_previous_path; size_t utf16be_previous_path_len; /* Null buffer used in bidder to improve its performance. */ unsigned char null[2048]; }; static int archive_read_format_iso9660_bid(struct archive_read *, int); static int archive_read_format_iso9660_options(struct archive_read *, const char *, const char *); static int archive_read_format_iso9660_cleanup(struct archive_read *); static int archive_read_format_iso9660_read_data(struct archive_read *, const void **, size_t *, int64_t *); static int archive_read_format_iso9660_read_data_skip(struct archive_read *); static int archive_read_format_iso9660_read_header(struct archive_read *, struct archive_entry *); static const char *build_pathname(struct archive_string *, struct file_info *, int); static int build_pathname_utf16be(unsigned char *, size_t, size_t *, struct file_info *); #if DEBUG static void dump_isodirrec(FILE *, const unsigned char *isodirrec); #endif static time_t time_from_tm(struct tm *); static time_t isodate17(const unsigned char *); static time_t isodate7(const unsigned char *); static int isBootRecord(struct iso9660 *, const unsigned char *); static int isVolumePartition(struct iso9660 *, const unsigned char *); static int isVDSetTerminator(struct iso9660 *, const unsigned char *); static int isJolietSVD(struct iso9660 *, const unsigned char *); static int isSVD(struct iso9660 *, const unsigned char *); static int isEVD(struct iso9660 *, const unsigned char *); static int isPVD(struct iso9660 *, const unsigned char *); static int next_cache_entry(struct archive_read *, struct iso9660 *, struct file_info **); static int next_entry_seek(struct archive_read *, struct iso9660 *, struct file_info **); static struct file_info * parse_file_info(struct archive_read *a, struct file_info *parent, const unsigned char *isodirrec); static int parse_rockridge(struct archive_read *a, struct file_info *file, const unsigned char *start, const unsigned char *end); static int register_CE(struct archive_read *a, int32_t location, struct file_info *file); static int read_CE(struct archive_read *a, struct iso9660 *iso9660); static void parse_rockridge_NM1(struct file_info *, const unsigned char *, int); static void parse_rockridge_SL1(struct file_info *, const unsigned char *, int); static void parse_rockridge_TF1(struct file_info *, const unsigned char *, int); static void parse_rockridge_ZF1(struct file_info *, const unsigned char *, int); static void register_file(struct iso9660 *, struct file_info *); static void release_files(struct iso9660 *); static unsigned toi(const void *p, int n); static inline void re_add_entry(struct iso9660 *, struct file_info *); static inline struct file_info * re_get_entry(struct iso9660 *); static inline int rede_add_entry(struct file_info *); static inline struct file_info * rede_get_entry(struct file_info *); static inline void cache_add_entry(struct iso9660 *iso9660, struct file_info *file); static inline struct file_info *cache_get_entry(struct iso9660 *iso9660); static int heap_add_entry(struct archive_read *a, struct heap_queue *heap, struct file_info *file, uint64_t key); static struct file_info *heap_get_entry(struct heap_queue *heap); #define add_entry(arch, iso9660, file) \ heap_add_entry(arch, &((iso9660)->pending_files), file, file->offset) #define next_entry(iso9660) \ heap_get_entry(&((iso9660)->pending_files)) int archive_read_support_format_iso9660(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct iso9660 *iso9660; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_iso9660"); iso9660 = (struct iso9660 *)calloc(1, sizeof(*iso9660)); if (iso9660 == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate iso9660 data"); return (ARCHIVE_FATAL); } iso9660->magic = ISO9660_MAGIC; iso9660->cache_files.first = NULL; iso9660->cache_files.last = &(iso9660->cache_files.first); iso9660->re_files.first = NULL; iso9660->re_files.last = &(iso9660->re_files.first); /* Enable to support Joliet extensions by default. */ iso9660->opt_support_joliet = 1; /* Enable to support Rock Ridge extensions by default. */ iso9660->opt_support_rockridge = 1; r = __archive_read_register_format(a, iso9660, "iso9660", archive_read_format_iso9660_bid, archive_read_format_iso9660_options, archive_read_format_iso9660_read_header, archive_read_format_iso9660_read_data, archive_read_format_iso9660_read_data_skip, NULL, archive_read_format_iso9660_cleanup, NULL, NULL); if (r != ARCHIVE_OK) { free(iso9660); return (r); } return (ARCHIVE_OK); } static int archive_read_format_iso9660_bid(struct archive_read *a, int best_bid) { struct iso9660 *iso9660; ssize_t bytes_read; const unsigned char *p; int seenTerminator; /* If there's already a better bid than we can ever make, don't bother testing. */ if (best_bid > 48) return (-1); iso9660 = (struct iso9660 *)(a->format->data); /* * Skip the first 32k (reserved area) and get the first * 8 sectors of the volume descriptor table. Of course, * if the I/O layer gives us more, we'll take it. */ #define RESERVED_AREA (SYSTEM_AREA_BLOCK * LOGICAL_BLOCK_SIZE) p = __archive_read_ahead(a, RESERVED_AREA + 8 * LOGICAL_BLOCK_SIZE, &bytes_read); if (p == NULL) return (-1); /* Skip the reserved area. */ bytes_read -= RESERVED_AREA; p += RESERVED_AREA; /* Check each volume descriptor. */ seenTerminator = 0; for (; bytes_read > LOGICAL_BLOCK_SIZE; bytes_read -= LOGICAL_BLOCK_SIZE, p += LOGICAL_BLOCK_SIZE) { /* Do not handle undefined Volume Descriptor Type. */ if (p[0] >= 4 && p[0] <= 254) return (0); /* Standard Identifier must be "CD001" */ if (memcmp(p + 1, "CD001", 5) != 0) return (0); if (isPVD(iso9660, p)) continue; if (!iso9660->joliet.location) { if (isJolietSVD(iso9660, p)) continue; } if (isBootRecord(iso9660, p)) continue; if (isEVD(iso9660, p)) continue; if (isSVD(iso9660, p)) continue; if (isVolumePartition(iso9660, p)) continue; if (isVDSetTerminator(iso9660, p)) { seenTerminator = 1; break; } return (0); } /* * ISO 9660 format must have Primary Volume Descriptor and * Volume Descriptor Set Terminator. */ if (seenTerminator && iso9660->primary.location > 16) return (48); /* We didn't find a valid PVD; return a bid of zero. */ return (0); } static int archive_read_format_iso9660_options(struct archive_read *a, const char *key, const char *val) { struct iso9660 *iso9660; iso9660 = (struct iso9660 *)(a->format->data); if (strcmp(key, "joliet") == 0) { if (val == NULL || strcmp(val, "off") == 0 || strcmp(val, "ignore") == 0 || strcmp(val, "disable") == 0 || strcmp(val, "0") == 0) iso9660->opt_support_joliet = 0; else iso9660->opt_support_joliet = 1; return (ARCHIVE_OK); } if (strcmp(key, "rockridge") == 0 || strcmp(key, "Rockridge") == 0) { iso9660->opt_support_rockridge = val != NULL; return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int isNull(struct iso9660 *iso9660, const unsigned char *h, unsigned offset, unsigned bytes) { while (bytes >= sizeof(iso9660->null)) { if (!memcmp(iso9660->null, h + offset, sizeof(iso9660->null))) return (0); offset += sizeof(iso9660->null); bytes -= sizeof(iso9660->null); } if (bytes) return memcmp(iso9660->null, h + offset, bytes) == 0; else return (1); } static int isBootRecord(struct iso9660 *iso9660, const unsigned char *h) { (void)iso9660; /* UNUSED */ /* Type of the Volume Descriptor Boot Record must be 0. */ if (h[0] != 0) return (0); /* Volume Descriptor Version must be 1. */ if (h[6] != 1) return (0); return (1); } static int isVolumePartition(struct iso9660 *iso9660, const unsigned char *h) { int32_t location; /* Type of the Volume Partition Descriptor must be 3. */ if (h[0] != 3) return (0); /* Volume Descriptor Version must be 1. */ if (h[6] != 1) return (0); /* Unused Field */ if (h[7] != 0) return (0); location = archive_le32dec(h + 72); if (location <= SYSTEM_AREA_BLOCK || location >= iso9660->volume_block) return (0); if ((uint32_t)location != archive_be32dec(h + 76)) return (0); return (1); } static int isVDSetTerminator(struct iso9660 *iso9660, const unsigned char *h) { (void)iso9660; /* UNUSED */ /* Type of the Volume Descriptor Set Terminator must be 255. */ if (h[0] != 255) return (0); /* Volume Descriptor Version must be 1. */ if (h[6] != 1) return (0); /* Reserved field must be 0. */ if (!isNull(iso9660, h, 7, 2048-7)) return (0); return (1); } static int isJolietSVD(struct iso9660 *iso9660, const unsigned char *h) { const unsigned char *p; ssize_t logical_block_size; int32_t volume_block; /* Check if current sector is a kind of Supplementary Volume * Descriptor. */ if (!isSVD(iso9660, h)) return (0); /* FIXME: do more validations according to joliet spec. */ /* check if this SVD contains joliet extension! */ p = h + SVD_escape_sequences_offset; /* N.B. Joliet spec says p[1] == '\\', but.... */ if (p[0] == '%' && p[1] == '/') { int level = 0; if (p[2] == '@') level = 1; else if (p[2] == 'C') level = 2; else if (p[2] == 'E') level = 3; else /* not joliet */ return (0); iso9660->seenJoliet = level; } else /* not joliet */ return (0); logical_block_size = archive_le16dec(h + SVD_logical_block_size_offset); volume_block = archive_le32dec(h + SVD_volume_space_size_offset); iso9660->logical_block_size = logical_block_size; iso9660->volume_block = volume_block; iso9660->volume_size = logical_block_size * (uint64_t)volume_block; /* Read Root Directory Record in Volume Descriptor. */ p = h + SVD_root_directory_record_offset; iso9660->joliet.location = archive_le32dec(p + DR_extent_offset); iso9660->joliet.size = archive_le32dec(p + DR_size_offset); return (48); } static int isSVD(struct iso9660 *iso9660, const unsigned char *h) { const unsigned char *p; ssize_t logical_block_size; int32_t volume_block; int32_t location; (void)iso9660; /* UNUSED */ /* Type 2 means it's a SVD. */ if (h[SVD_type_offset] != 2) return (0); /* Reserved field must be 0. */ if (!isNull(iso9660, h, SVD_reserved1_offset, SVD_reserved1_size)) return (0); if (!isNull(iso9660, h, SVD_reserved2_offset, SVD_reserved2_size)) return (0); if (!isNull(iso9660, h, SVD_reserved3_offset, SVD_reserved3_size)) return (0); /* File structure version must be 1 for ISO9660/ECMA119. */ if (h[SVD_file_structure_version_offset] != 1) return (0); logical_block_size = archive_le16dec(h + SVD_logical_block_size_offset); if (logical_block_size <= 0) return (0); volume_block = archive_le32dec(h + SVD_volume_space_size_offset); if (volume_block <= SYSTEM_AREA_BLOCK+4) return (0); /* Location of Occurrence of Type L Path Table must be * available location, * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ location = archive_le32dec(h+SVD_type_L_path_table_offset); if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) return (0); /* The Type M Path Table must be at a valid location (WinISO * and probably other programs omit this, so we allow zero) * * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ location = archive_be32dec(h+SVD_type_M_path_table_offset); if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) || location >= volume_block) return (0); /* Read Root Directory Record in Volume Descriptor. */ p = h + SVD_root_directory_record_offset; if (p[DR_length_offset] != 34) return (0); return (48); } static int isEVD(struct iso9660 *iso9660, const unsigned char *h) { const unsigned char *p; ssize_t logical_block_size; int32_t volume_block; int32_t location; (void)iso9660; /* UNUSED */ /* Type of the Enhanced Volume Descriptor must be 2. */ if (h[PVD_type_offset] != 2) return (0); /* EVD version must be 2. */ if (h[PVD_version_offset] != 2) return (0); /* Reserved field must be 0. */ if (h[PVD_reserved1_offset] != 0) return (0); /* Reserved field must be 0. */ if (!isNull(iso9660, h, PVD_reserved2_offset, PVD_reserved2_size)) return (0); /* Reserved field must be 0. */ if (!isNull(iso9660, h, PVD_reserved3_offset, PVD_reserved3_size)) return (0); /* Logical block size must be > 0. */ /* I've looked at Ecma 119 and can't find any stronger * restriction on this field. */ logical_block_size = archive_le16dec(h + PVD_logical_block_size_offset); if (logical_block_size <= 0) return (0); volume_block = archive_le32dec(h + PVD_volume_space_size_offset); if (volume_block <= SYSTEM_AREA_BLOCK+4) return (0); /* File structure version must be 2 for ISO9660:1999. */ if (h[PVD_file_structure_version_offset] != 2) return (0); /* Location of Occurrence of Type L Path Table must be * available location, * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ location = archive_le32dec(h+PVD_type_1_path_table_offset); if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) return (0); /* Location of Occurrence of Type M Path Table must be * available location, * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ location = archive_be32dec(h+PVD_type_m_path_table_offset); if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) || location >= volume_block) return (0); /* Reserved field must be 0. */ if (!isNull(iso9660, h, PVD_reserved4_offset, PVD_reserved4_size)) return (0); /* Reserved field must be 0. */ if (!isNull(iso9660, h, PVD_reserved5_offset, PVD_reserved5_size)) return (0); /* Read Root Directory Record in Volume Descriptor. */ p = h + PVD_root_directory_record_offset; if (p[DR_length_offset] != 34) return (0); return (48); } static int isPVD(struct iso9660 *iso9660, const unsigned char *h) { const unsigned char *p; ssize_t logical_block_size; int32_t volume_block; int32_t location; int i; /* Type of the Primary Volume Descriptor must be 1. */ if (h[PVD_type_offset] != 1) return (0); /* PVD version must be 1. */ if (h[PVD_version_offset] != 1) return (0); /* Reserved field must be 0. */ if (h[PVD_reserved1_offset] != 0) return (0); /* Reserved field must be 0. */ if (!isNull(iso9660, h, PVD_reserved2_offset, PVD_reserved2_size)) return (0); /* Reserved field must be 0. */ if (!isNull(iso9660, h, PVD_reserved3_offset, PVD_reserved3_size)) return (0); /* Logical block size must be > 0. */ /* I've looked at Ecma 119 and can't find any stronger * restriction on this field. */ logical_block_size = archive_le16dec(h + PVD_logical_block_size_offset); if (logical_block_size <= 0) return (0); volume_block = archive_le32dec(h + PVD_volume_space_size_offset); if (volume_block <= SYSTEM_AREA_BLOCK+4) return (0); /* File structure version must be 1 for ISO9660/ECMA119. */ if (h[PVD_file_structure_version_offset] != 1) return (0); /* Location of Occurrence of Type L Path Table must be * available location, * > SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ location = archive_le32dec(h+PVD_type_1_path_table_offset); if (location < SYSTEM_AREA_BLOCK+2 || location >= volume_block) return (0); /* The Type M Path Table must also be at a valid location * (although ECMA 119 requires a Type M Path Table, WinISO and * probably other programs omit it, so we permit a zero here) * * >= SYSTEM_AREA_BLOCK(16) + 2 and < Volume Space Size. */ location = archive_be32dec(h+PVD_type_m_path_table_offset); if ((location > 0 && location < SYSTEM_AREA_BLOCK+2) || location >= volume_block) return (0); /* Reserved field must be 0. */ /* But accept NetBSD/FreeBSD "makefs" images with 0x20 here. */ for (i = 0; i < PVD_reserved4_size; ++i) if (h[PVD_reserved4_offset + i] != 0 && h[PVD_reserved4_offset + i] != 0x20) return (0); /* Reserved field must be 0. */ if (!isNull(iso9660, h, PVD_reserved5_offset, PVD_reserved5_size)) return (0); /* XXX TODO: Check other values for sanity; reject more * malformed PVDs. XXX */ /* Read Root Directory Record in Volume Descriptor. */ p = h + PVD_root_directory_record_offset; if (p[DR_length_offset] != 34) return (0); if (!iso9660->primary.location) { iso9660->logical_block_size = logical_block_size; iso9660->volume_block = volume_block; iso9660->volume_size = logical_block_size * (uint64_t)volume_block; iso9660->primary.location = archive_le32dec(p + DR_extent_offset); iso9660->primary.size = archive_le32dec(p + DR_size_offset); } return (48); } static int read_children(struct archive_read *a, struct file_info *parent) { struct iso9660 *iso9660; const unsigned char *b, *p; struct file_info *multi; size_t step, skip_size; iso9660 = (struct iso9660 *)(a->format->data); /* flush any remaining bytes from the last round to ensure * we're positioned */ if (iso9660->entry_bytes_unconsumed) { __archive_read_consume(a, iso9660->entry_bytes_unconsumed); iso9660->entry_bytes_unconsumed = 0; } if (iso9660->current_position > parent->offset) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Ignoring out-of-order directory (%s) %jd > %jd", parent->name.s, (intmax_t)iso9660->current_position, (intmax_t)parent->offset); return (ARCHIVE_WARN); } if (parent->offset + parent->size > iso9660->volume_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Directory is beyond end-of-media: %s", parent->name.s); return (ARCHIVE_WARN); } if (iso9660->current_position < parent->offset) { int64_t skipsize; skipsize = parent->offset - iso9660->current_position; skipsize = __archive_read_consume(a, skipsize); if (skipsize < 0) return ((int)skipsize); iso9660->current_position = parent->offset; } step = (size_t)(((parent->size + iso9660->logical_block_size -1) / iso9660->logical_block_size) * iso9660->logical_block_size); b = __archive_read_ahead(a, step, NULL); if (b == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to read full block when scanning " "ISO9660 directory list"); return (ARCHIVE_FATAL); } iso9660->current_position += step; multi = NULL; skip_size = step; while (step) { p = b; b += iso9660->logical_block_size; step -= iso9660->logical_block_size; for (; *p != 0 && p < b && p + *p <= b; p += *p) { struct file_info *child; /* N.B.: these special directory identifiers * are 8 bit "values" even on a * Joliet CD with UCS-2 (16bit) encoding. */ /* Skip '.' entry. */ if (*(p + DR_name_len_offset) == 1 && *(p + DR_name_offset) == '\0') continue; /* Skip '..' entry. */ if (*(p + DR_name_len_offset) == 1 && *(p + DR_name_offset) == '\001') continue; child = parse_file_info(a, parent, p); if (child == NULL) { __archive_read_consume(a, skip_size); return (ARCHIVE_FATAL); } if (child->cl_offset == 0 && (child->multi_extent || multi != NULL)) { struct content *con; if (multi == NULL) { multi = child; multi->contents.first = NULL; multi->contents.last = &(multi->contents.first); } con = malloc(sizeof(struct content)); if (con == NULL) { archive_set_error( &a->archive, ENOMEM, "No memory for multi extent"); __archive_read_consume(a, skip_size); return (ARCHIVE_FATAL); } con->offset = child->offset; con->size = child->size; con->next = NULL; *multi->contents.last = con; multi->contents.last = &(con->next); if (multi == child) { if (add_entry(a, iso9660, child) != ARCHIVE_OK) return (ARCHIVE_FATAL); } else { multi->size += child->size; if (!child->multi_extent) multi = NULL; } } else if (add_entry(a, iso9660, child) != ARCHIVE_OK) return (ARCHIVE_FATAL); } } __archive_read_consume(a, skip_size); /* Read data which recorded by RRIP "CE" extension. */ if (read_CE(a, iso9660) != ARCHIVE_OK) return (ARCHIVE_FATAL); return (ARCHIVE_OK); } static int choose_volume(struct archive_read *a, struct iso9660 *iso9660) { struct file_info *file; int64_t skipsize; struct vd *vd; const void *block; char seenJoliet; vd = &(iso9660->primary); if (!iso9660->opt_support_joliet) iso9660->seenJoliet = 0; if (iso9660->seenJoliet && vd->location > iso9660->joliet.location) /* This condition is unlikely; by way of caution. */ vd = &(iso9660->joliet); skipsize = LOGICAL_BLOCK_SIZE * (int64_t)vd->location; skipsize = __archive_read_consume(a, skipsize); if (skipsize < 0) return ((int)skipsize); iso9660->current_position = skipsize; block = __archive_read_ahead(a, vd->size, NULL); if (block == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to read full block when scanning " "ISO9660 directory list"); return (ARCHIVE_FATAL); } /* * While reading Root Directory, flag seenJoliet must be zero to * avoid converting special name 0x00(Current Directory) and * next byte to UCS2. */ seenJoliet = iso9660->seenJoliet;/* Save flag. */ iso9660->seenJoliet = 0; file = parse_file_info(a, NULL, block); if (file == NULL) return (ARCHIVE_FATAL); iso9660->seenJoliet = seenJoliet; /* * If the iso image has both RockRidge and Joliet, we preferentially * use RockRidge Extensions rather than Joliet ones. */ if (vd == &(iso9660->primary) && iso9660->seenRockridge && iso9660->seenJoliet) iso9660->seenJoliet = 0; if (vd == &(iso9660->primary) && !iso9660->seenRockridge && iso9660->seenJoliet) { /* Switch reading data from primary to joliet. */ vd = &(iso9660->joliet); skipsize = LOGICAL_BLOCK_SIZE * (int64_t)vd->location; skipsize -= iso9660->current_position; skipsize = __archive_read_consume(a, skipsize); if (skipsize < 0) return ((int)skipsize); iso9660->current_position += skipsize; block = __archive_read_ahead(a, vd->size, NULL); if (block == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to read full block when scanning " "ISO9660 directory list"); return (ARCHIVE_FATAL); } iso9660->seenJoliet = 0; file = parse_file_info(a, NULL, block); if (file == NULL) return (ARCHIVE_FATAL); iso9660->seenJoliet = seenJoliet; } /* Store the root directory in the pending list. */ if (add_entry(a, iso9660, file) != ARCHIVE_OK) return (ARCHIVE_FATAL); if (iso9660->seenRockridge) { a->archive.archive_format = ARCHIVE_FORMAT_ISO9660_ROCKRIDGE; a->archive.archive_format_name = "ISO9660 with Rockridge extensions"; } return (ARCHIVE_OK); } static int archive_read_format_iso9660_read_header(struct archive_read *a, struct archive_entry *entry) { struct iso9660 *iso9660; struct file_info *file; int r, rd_r = ARCHIVE_OK; iso9660 = (struct iso9660 *)(a->format->data); if (!a->archive.archive_format) { a->archive.archive_format = ARCHIVE_FORMAT_ISO9660; a->archive.archive_format_name = "ISO9660"; } if (iso9660->current_position == 0) { r = choose_volume(a, iso9660); if (r != ARCHIVE_OK) return (r); } file = NULL;/* Eliminate a warning. */ /* Get the next entry that appears after the current offset. */ r = next_entry_seek(a, iso9660, &file); if (r != ARCHIVE_OK) return (r); if (iso9660->seenJoliet) { /* * Convert UTF-16BE of a filename to local locale MBS * and store the result into a filename field. */ if (iso9660->sconv_utf16be == NULL) { iso9660->sconv_utf16be = archive_string_conversion_from_charset( &(a->archive), "UTF-16BE", 1); if (iso9660->sconv_utf16be == NULL) /* Couldn't allocate memory */ return (ARCHIVE_FATAL); } if (iso9660->utf16be_path == NULL) { iso9660->utf16be_path = malloc(UTF16_NAME_MAX); if (iso9660->utf16be_path == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory"); return (ARCHIVE_FATAL); } } if (iso9660->utf16be_previous_path == NULL) { iso9660->utf16be_previous_path = malloc(UTF16_NAME_MAX); if (iso9660->utf16be_previous_path == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory"); return (ARCHIVE_FATAL); } } iso9660->utf16be_path_len = 0; if (build_pathname_utf16be(iso9660->utf16be_path, UTF16_NAME_MAX, &(iso9660->utf16be_path_len), file) != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Pathname is too long"); return (ARCHIVE_FATAL); } r = archive_entry_copy_pathname_l(entry, (const char *)iso9660->utf16be_path, iso9660->utf16be_path_len, iso9660->sconv_utf16be); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "No memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Pathname cannot be converted " "from %s to current locale.", archive_string_conversion_charset_name( iso9660->sconv_utf16be)); rd_r = ARCHIVE_WARN; } } else { const char *path = build_pathname(&iso9660->pathname, file, 0); if (path == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Pathname is too long"); return (ARCHIVE_FATAL); } else { archive_string_empty(&iso9660->pathname); archive_entry_set_pathname(entry, path); } } iso9660->entry_bytes_remaining = file->size; /* Offset for sparse-file-aware clients. */ iso9660->entry_sparse_offset = 0; if (file->offset + file->size > iso9660->volume_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "File is beyond end-of-media: %s", archive_entry_pathname(entry)); iso9660->entry_bytes_remaining = 0; return (ARCHIVE_WARN); } /* Set up the entry structure with information about this entry. */ archive_entry_set_mode(entry, file->mode); archive_entry_set_uid(entry, file->uid); archive_entry_set_gid(entry, file->gid); archive_entry_set_nlink(entry, file->nlinks); if (file->birthtime_is_set) archive_entry_set_birthtime(entry, file->birthtime, 0); else archive_entry_unset_birthtime(entry); archive_entry_set_mtime(entry, file->mtime, 0); archive_entry_set_ctime(entry, file->ctime, 0); archive_entry_set_atime(entry, file->atime, 0); /* N.B.: Rock Ridge supports 64-bit device numbers. */ archive_entry_set_rdev(entry, (dev_t)file->rdev); archive_entry_set_size(entry, iso9660->entry_bytes_remaining); if (file->symlink.s != NULL) archive_entry_copy_symlink(entry, file->symlink.s); /* Note: If the input isn't seekable, we can't rewind to * return the same body again, so if the next entry refers to * the same data, we have to return it as a hardlink to the * original entry. */ if (file->number != -1 && file->number == iso9660->previous_number) { if (iso9660->seenJoliet) { r = archive_entry_copy_hardlink_l(entry, (const char *)iso9660->utf16be_previous_path, iso9660->utf16be_previous_path_len, iso9660->sconv_utf16be); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "No memory for Linkname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Linkname cannot be converted " "from %s to current locale.", archive_string_conversion_charset_name( iso9660->sconv_utf16be)); rd_r = ARCHIVE_WARN; } } else archive_entry_set_hardlink(entry, iso9660->previous_pathname.s); archive_entry_unset_size(entry); iso9660->entry_bytes_remaining = 0; return (rd_r); } if ((file->mode & AE_IFMT) != AE_IFDIR && file->offset < iso9660->current_position) { int64_t r64; r64 = __archive_read_seek(a, file->offset, SEEK_SET); if (r64 != (int64_t)file->offset) { /* We can't seek backwards to extract it, so issue * a warning. Note that this can only happen if * this entry was added to the heap after we passed * this offset, that is, only if the directory * mentioning this entry is later than the body of * the entry. Such layouts are very unusual; most * ISO9660 writers lay out and record all directory * information first, then store all file bodies. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Ignoring out-of-order file @%jx (%s) %jd < %jd", (intmax_t)file->number, iso9660->pathname.s, (intmax_t)file->offset, (intmax_t)iso9660->current_position); iso9660->entry_bytes_remaining = 0; return (ARCHIVE_WARN); } iso9660->current_position = (uint64_t)r64; } /* Initialize zisofs variables. */ iso9660->entry_zisofs.pz = file->pz; if (file->pz) { #ifdef HAVE_ZLIB_H struct zisofs *zisofs; zisofs = &iso9660->entry_zisofs; zisofs->initialized = 0; zisofs->pz_log2_bs = file->pz_log2_bs; zisofs->pz_uncompressed_size = file->pz_uncompressed_size; zisofs->pz_offset = 0; zisofs->header_avail = 0; zisofs->header_passed = 0; zisofs->block_pointers_avail = 0; #endif archive_entry_set_size(entry, file->pz_uncompressed_size); } iso9660->previous_number = file->number; if (iso9660->seenJoliet) { memcpy(iso9660->utf16be_previous_path, iso9660->utf16be_path, iso9660->utf16be_path_len); iso9660->utf16be_previous_path_len = iso9660->utf16be_path_len; } else archive_strcpy( &iso9660->previous_pathname, iso9660->pathname.s); /* Reset entry_bytes_remaining if the file is multi extent. */ iso9660->entry_content = file->contents.first; if (iso9660->entry_content != NULL) iso9660->entry_bytes_remaining = iso9660->entry_content->size; if (archive_entry_filetype(entry) == AE_IFDIR) { /* Overwrite nlinks by proper link number which is * calculated from number of sub directories. */ archive_entry_set_nlink(entry, 2 + file->subdirs); /* Directory data has been read completely. */ iso9660->entry_bytes_remaining = 0; } if (rd_r != ARCHIVE_OK) return (rd_r); return (ARCHIVE_OK); } static int archive_read_format_iso9660_read_data_skip(struct archive_read *a) { /* Because read_next_header always does an explicit skip * to the next entry, we don't need to do anything here. */ (void)a; /* UNUSED */ return (ARCHIVE_OK); } #ifdef HAVE_ZLIB_H static int zisofs_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct iso9660 *iso9660; struct zisofs *zisofs; const unsigned char *p; size_t avail; ssize_t bytes_read; size_t uncompressed_size; int r; iso9660 = (struct iso9660 *)(a->format->data); zisofs = &iso9660->entry_zisofs; p = __archive_read_ahead(a, 1, &bytes_read); if (bytes_read <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated zisofs file body"); return (ARCHIVE_FATAL); } if (bytes_read > iso9660->entry_bytes_remaining) bytes_read = (ssize_t)iso9660->entry_bytes_remaining; avail = bytes_read; uncompressed_size = 0; if (!zisofs->initialized) { size_t ceil, xsize; /* Allocate block pointers buffer. */ ceil = (size_t)((zisofs->pz_uncompressed_size + (((int64_t)1) << zisofs->pz_log2_bs) - 1) >> zisofs->pz_log2_bs); xsize = (ceil + 1) * 4; if (zisofs->block_pointers_alloc < xsize) { size_t alloc; if (zisofs->block_pointers != NULL) free(zisofs->block_pointers); alloc = ((xsize >> 10) + 1) << 10; zisofs->block_pointers = malloc(alloc); if (zisofs->block_pointers == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for zisofs decompression"); return (ARCHIVE_FATAL); } zisofs->block_pointers_alloc = alloc; } zisofs->block_pointers_size = xsize; /* Allocate uncompressed data buffer. */ xsize = (size_t)1UL << zisofs->pz_log2_bs; if (zisofs->uncompressed_buffer_size < xsize) { if (zisofs->uncompressed_buffer != NULL) free(zisofs->uncompressed_buffer); zisofs->uncompressed_buffer = malloc(xsize); if (zisofs->uncompressed_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for zisofs decompression"); return (ARCHIVE_FATAL); } } zisofs->uncompressed_buffer_size = xsize; /* * Read the file header, and check the magic code of zisofs. */ if (zisofs->header_avail < sizeof(zisofs->header)) { xsize = sizeof(zisofs->header) - zisofs->header_avail; if (avail < xsize) xsize = avail; memcpy(zisofs->header + zisofs->header_avail, p, xsize); zisofs->header_avail += xsize; avail -= xsize; p += xsize; } if (!zisofs->header_passed && zisofs->header_avail == sizeof(zisofs->header)) { int err = 0; if (memcmp(zisofs->header, zisofs_magic, sizeof(zisofs_magic)) != 0) err = 1; if (archive_le32dec(zisofs->header + 8) != zisofs->pz_uncompressed_size) err = 1; if (zisofs->header[12] != 4) err = 1; if (zisofs->header[13] != zisofs->pz_log2_bs) err = 1; if (err) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Illegal zisofs file body"); return (ARCHIVE_FATAL); } zisofs->header_passed = 1; } /* * Read block pointers. */ if (zisofs->header_passed && zisofs->block_pointers_avail < zisofs->block_pointers_size) { xsize = zisofs->block_pointers_size - zisofs->block_pointers_avail; if (avail < xsize) xsize = avail; memcpy(zisofs->block_pointers + zisofs->block_pointers_avail, p, xsize); zisofs->block_pointers_avail += xsize; avail -= xsize; p += xsize; if (zisofs->block_pointers_avail == zisofs->block_pointers_size) { /* We've got all block pointers and initialize * related variables. */ zisofs->block_off = 0; zisofs->block_avail = 0; /* Complete a initialization */ zisofs->initialized = 1; } } if (!zisofs->initialized) goto next_data; /* We need more data. */ } /* * Get block offsets from block pointers. */ if (zisofs->block_avail == 0) { uint32_t bst, bed; if (zisofs->block_off + 4 >= zisofs->block_pointers_size) { /* There isn't a pair of offsets. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Illegal zisofs block pointers"); return (ARCHIVE_FATAL); } bst = archive_le32dec( zisofs->block_pointers + zisofs->block_off); if (bst != zisofs->pz_offset + (bytes_read - avail)) { /* TODO: Should we seek offset of current file * by bst ? */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Illegal zisofs block pointers(cannot seek)"); return (ARCHIVE_FATAL); } bed = archive_le32dec( zisofs->block_pointers + zisofs->block_off + 4); if (bed < bst) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Illegal zisofs block pointers"); return (ARCHIVE_FATAL); } zisofs->block_avail = bed - bst; zisofs->block_off += 4; /* Initialize compression library for new block. */ if (zisofs->stream_valid) r = inflateReset(&zisofs->stream); else r = inflateInit(&zisofs->stream); if (r != Z_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't initialize zisofs decompression."); return (ARCHIVE_FATAL); } zisofs->stream_valid = 1; zisofs->stream.total_in = 0; zisofs->stream.total_out = 0; } /* * Make uncompressed data. */ if (zisofs->block_avail == 0) { memset(zisofs->uncompressed_buffer, 0, zisofs->uncompressed_buffer_size); uncompressed_size = zisofs->uncompressed_buffer_size; } else { zisofs->stream.next_in = (Bytef *)(uintptr_t)(const void *)p; if (avail > zisofs->block_avail) zisofs->stream.avail_in = zisofs->block_avail; else zisofs->stream.avail_in = (uInt)avail; zisofs->stream.next_out = zisofs->uncompressed_buffer; zisofs->stream.avail_out = (uInt)zisofs->uncompressed_buffer_size; r = inflate(&zisofs->stream, 0); switch (r) { case Z_OK: /* Decompressor made some progress.*/ case Z_STREAM_END: /* Found end of stream. */ break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "zisofs decompression failed (%d)", r); return (ARCHIVE_FATAL); } uncompressed_size = zisofs->uncompressed_buffer_size - zisofs->stream.avail_out; avail -= zisofs->stream.next_in - p; zisofs->block_avail -= (uint32_t)(zisofs->stream.next_in - p); } next_data: bytes_read -= avail; *buff = zisofs->uncompressed_buffer; *size = uncompressed_size; *offset = iso9660->entry_sparse_offset; iso9660->entry_sparse_offset += uncompressed_size; iso9660->entry_bytes_remaining -= bytes_read; iso9660->current_position += bytes_read; zisofs->pz_offset += (uint32_t)bytes_read; iso9660->entry_bytes_unconsumed += bytes_read; return (ARCHIVE_OK); } #else /* HAVE_ZLIB_H */ static int zisofs_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { (void)buff;/* UNUSED */ (void)size;/* UNUSED */ (void)offset;/* UNUSED */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "zisofs is not supported on this platform."); return (ARCHIVE_FAILED); } #endif /* HAVE_ZLIB_H */ static int archive_read_format_iso9660_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { ssize_t bytes_read; struct iso9660 *iso9660; iso9660 = (struct iso9660 *)(a->format->data); if (iso9660->entry_bytes_unconsumed) { __archive_read_consume(a, iso9660->entry_bytes_unconsumed); iso9660->entry_bytes_unconsumed = 0; } if (iso9660->entry_bytes_remaining <= 0) { if (iso9660->entry_content != NULL) iso9660->entry_content = iso9660->entry_content->next; if (iso9660->entry_content == NULL) { *buff = NULL; *size = 0; *offset = iso9660->entry_sparse_offset; return (ARCHIVE_EOF); } /* Seek forward to the start of the entry. */ if (iso9660->current_position < iso9660->entry_content->offset) { int64_t step; step = iso9660->entry_content->offset - iso9660->current_position; step = __archive_read_consume(a, step); if (step < 0) return ((int)step); iso9660->current_position = iso9660->entry_content->offset; } if (iso9660->entry_content->offset < iso9660->current_position) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Ignoring out-of-order file (%s) %jd < %jd", iso9660->pathname.s, (intmax_t)iso9660->entry_content->offset, (intmax_t)iso9660->current_position); *buff = NULL; *size = 0; *offset = iso9660->entry_sparse_offset; return (ARCHIVE_WARN); } iso9660->entry_bytes_remaining = iso9660->entry_content->size; } if (iso9660->entry_zisofs.pz) return (zisofs_read_data(a, buff, size, offset)); *buff = __archive_read_ahead(a, 1, &bytes_read); if (bytes_read == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Truncated input file"); if (*buff == NULL) return (ARCHIVE_FATAL); if (bytes_read > iso9660->entry_bytes_remaining) bytes_read = (ssize_t)iso9660->entry_bytes_remaining; *size = bytes_read; *offset = iso9660->entry_sparse_offset; iso9660->entry_sparse_offset += bytes_read; iso9660->entry_bytes_remaining -= bytes_read; iso9660->entry_bytes_unconsumed = bytes_read; iso9660->current_position += bytes_read; return (ARCHIVE_OK); } static int archive_read_format_iso9660_cleanup(struct archive_read *a) { struct iso9660 *iso9660; int r = ARCHIVE_OK; iso9660 = (struct iso9660 *)(a->format->data); release_files(iso9660); free(iso9660->read_ce_req.reqs); archive_string_free(&iso9660->pathname); archive_string_free(&iso9660->previous_pathname); if (iso9660->pending_files.files) free(iso9660->pending_files.files); #ifdef HAVE_ZLIB_H free(iso9660->entry_zisofs.uncompressed_buffer); free(iso9660->entry_zisofs.block_pointers); if (iso9660->entry_zisofs.stream_valid) { if (inflateEnd(&iso9660->entry_zisofs.stream) != Z_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to clean up zlib decompressor"); r = ARCHIVE_FATAL; } } #endif free(iso9660->utf16be_path); free(iso9660->utf16be_previous_path); free(iso9660); (a->format->data) = NULL; return (r); } /* * This routine parses a single ISO directory record, makes sense * of any extensions, and stores the result in memory. */ static struct file_info * parse_file_info(struct archive_read *a, struct file_info *parent, const unsigned char *isodirrec) { struct iso9660 *iso9660; struct file_info *file, *filep; size_t name_len; const unsigned char *rr_start, *rr_end; const unsigned char *p; size_t dr_len; uint64_t fsize, offset; int32_t location; int flags; iso9660 = (struct iso9660 *)(a->format->data); dr_len = (size_t)isodirrec[DR_length_offset]; name_len = (size_t)isodirrec[DR_name_len_offset]; location = archive_le32dec(isodirrec + DR_extent_offset); fsize = toi(isodirrec + DR_size_offset, DR_size_size); /* Sanity check that dr_len needs at least 34. */ if (dr_len < 34) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid length of directory record"); return (NULL); } /* Sanity check that name_len doesn't exceed dr_len. */ if (dr_len - 33 < name_len || name_len == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid length of file identifier"); return (NULL); } /* Sanity check that location doesn't exceed volume block. * Don't check lower limit of location; it's possibility * the location has negative value when file type is symbolic * link or file size is zero. As far as I know latest mkisofs * do that. */ if (location > 0 && (location + ((fsize + iso9660->logical_block_size -1) / iso9660->logical_block_size)) > (uint32_t)iso9660->volume_block) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid location of extent of file"); return (NULL); } /* Sanity check that location doesn't have a negative value * when the file is not empty. it's too large. */ if (fsize != 0 && location < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid location of extent of file"); return (NULL); } /* Sanity check that this entry does not create a cycle. */ offset = iso9660->logical_block_size * (uint64_t)location; for (filep = parent; filep != NULL; filep = filep->parent) { if (filep->offset == offset) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Directory structure contains loop"); return (NULL); } } /* Create a new file entry and copy data from the ISO dir record. */ file = (struct file_info *)calloc(1, sizeof(*file)); if (file == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for file entry"); return (NULL); } file->parent = parent; file->offset = offset; file->size = fsize; file->mtime = isodate7(isodirrec + DR_date_offset); file->ctime = file->atime = file->mtime; file->rede_files.first = NULL; file->rede_files.last = &(file->rede_files.first); p = isodirrec + DR_name_offset; /* Rockridge extensions (if any) follow name. Compute this * before fidgeting the name_len below. */ rr_start = p + name_len + (name_len & 1 ? 0 : 1); rr_end = isodirrec + dr_len; if (iso9660->seenJoliet) { /* Joliet names are max 64 chars (128 bytes) according to spec, * but genisoimage/mkisofs allows recording longer Joliet * names which are 103 UCS2 characters(206 bytes) by their * option '-joliet-long'. */ if (name_len > 206) name_len = 206; name_len &= ~1; /* trim trailing first version and dot from filename. * * Remember we were in UTF-16BE land! * SEPARATOR 1 (.) and SEPARATOR 2 (;) are both * 16 bits big endian characters on Joliet. * * TODO: sanitize filename? * Joliet allows any UCS-2 char except: * *, /, :, ;, ? and \. */ /* Chop off trailing ';1' from files. */ if (name_len > 4 && p[name_len-4] == 0 && p[name_len-3] == ';' && p[name_len-2] == 0 && p[name_len-1] == '1') name_len -= 4; #if 0 /* XXX: this somehow manages to strip of single-character file extensions, like '.c'. */ /* Chop off trailing '.' from filenames. */ if (name_len > 2 && p[name_len-2] == 0 && p[name_len-1] == '.') name_len -= 2; #endif if ((file->utf16be_name = malloc(name_len)) == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for file name"); return (NULL); } memcpy(file->utf16be_name, p, name_len); file->utf16be_bytes = name_len; } else { /* Chop off trailing ';1' from files. */ if (name_len > 2 && p[name_len - 2] == ';' && p[name_len - 1] == '1') name_len -= 2; /* Chop off trailing '.' from filenames. */ if (name_len > 1 && p[name_len - 1] == '.') --name_len; archive_strncpy(&file->name, (const char *)p, name_len); } flags = isodirrec[DR_flags_offset]; if (flags & 0x02) file->mode = AE_IFDIR | 0700; else file->mode = AE_IFREG | 0400; if (flags & 0x80) file->multi_extent = 1; else file->multi_extent = 0; /* * Use a location for the file number, which is treated as an inode * number to find out hardlink target. If Rockridge extensions is * being used, the file number will be overwritten by FILE SERIAL * NUMBER of RRIP "PX" extension. * Note: Old mkisofs did not record that FILE SERIAL NUMBER * in ISO images. * Note2: xorriso set 0 to the location of a symlink file. */ if (file->size == 0 && location >= 0) { /* If file->size is zero, its location points wrong place, * and so we should not use it for the file number. * When the location has negative value, it can be used * for the file number. */ file->number = -1; /* Do not appear before any directory entries. */ file->offset = -1; } else file->number = (int64_t)(uint32_t)location; /* Rockridge extensions overwrite information from above. */ if (iso9660->opt_support_rockridge) { if (parent == NULL && rr_end - rr_start >= 7) { p = rr_start; if (memcmp(p, "SP\x07\x01\xbe\xef", 6) == 0) { /* * SP extension stores the suspOffset * (Number of bytes to skip between * filename and SUSP records.) * It is mandatory by the SUSP standard * (IEEE 1281). * * It allows SUSP to coexist with * non-SUSP uses of the System * Use Area by placing non-SUSP data * before SUSP data. * * SP extension must be in the root * directory entry, disable all SUSP * processing if not found. */ iso9660->suspOffset = p[6]; iso9660->seenSUSP = 1; rr_start += 7; } } if (iso9660->seenSUSP) { int r; file->name_continues = 0; file->symlink_continues = 0; rr_start += iso9660->suspOffset; r = parse_rockridge(a, file, rr_start, rr_end); if (r != ARCHIVE_OK) { free(file); return (NULL); } /* * A file size of symbolic link files in ISO images * made by makefs is not zero and its location is * the same as those of next regular file. That is * the same as hard like file and it causes unexpected * error. */ if (file->size > 0 && (file->mode & AE_IFMT) == AE_IFLNK) { file->size = 0; file->number = -1; file->offset = -1; } } else /* If there isn't SUSP, disable parsing * rock ridge extensions. */ iso9660->opt_support_rockridge = 0; } file->nlinks = 1;/* Reset nlink. we'll calculate it later. */ /* Tell file's parent how many children that parent has. */ if (parent != NULL && (flags & 0x02)) parent->subdirs++; if (iso9660->seenRockridge) { if (parent != NULL && parent->parent == NULL && (flags & 0x02) && iso9660->rr_moved == NULL && file->name.s && (strcmp(file->name.s, "rr_moved") == 0 || strcmp(file->name.s, ".rr_moved") == 0)) { iso9660->rr_moved = file; file->rr_moved = 1; file->rr_moved_has_re_only = 1; file->re = 0; parent->subdirs--; } else if (file->re) { /* * Sanity check: file's parent is rr_moved. */ if (parent == NULL || parent->rr_moved == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid Rockridge RE"); return (NULL); } /* * Sanity check: file does not have "CL" extension. */ if (file->cl_offset) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid Rockridge RE and CL"); return (NULL); } /* * Sanity check: The file type must be a directory. */ if ((flags & 0x02) == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid Rockridge RE"); return (NULL); } } else if (parent != NULL && parent->rr_moved) file->rr_moved_has_re_only = 0; else if (parent != NULL && (flags & 0x02) && (parent->re || parent->re_descendant)) file->re_descendant = 1; if (file->cl_offset) { struct file_info *r; if (parent == NULL || parent->parent == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid Rockridge CL"); return (NULL); } /* * Sanity check: The file type must be a regular file. */ if ((flags & 0x02) != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid Rockridge CL"); return (NULL); } parent->subdirs++; /* Overwrite an offset and a number of this "CL" entry * to appear before other dirs. "+1" to those is to * make sure to appear after "RE" entry which this * "CL" entry should be connected with. */ file->offset = file->number = file->cl_offset + 1; /* * Sanity check: cl_offset does not point at its * the parents or itself. */ for (r = parent; r; r = r->parent) { if (r->offset == file->cl_offset) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid Rockridge CL"); return (NULL); } } if (file->cl_offset == file->offset || parent->rr_moved) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid Rockridge CL"); return (NULL); } } } #if DEBUG /* DEBUGGING: Warn about attributes I don't yet fully support. */ if ((flags & ~0x02) != 0) { fprintf(stderr, "\n ** Unrecognized flag: "); dump_isodirrec(stderr, isodirrec); fprintf(stderr, "\n"); } else if (toi(isodirrec + DR_volume_sequence_number_offset, 2) != 1) { fprintf(stderr, "\n ** Unrecognized sequence number: "); dump_isodirrec(stderr, isodirrec); fprintf(stderr, "\n"); } else if (*(isodirrec + DR_file_unit_size_offset) != 0) { fprintf(stderr, "\n ** Unexpected file unit size: "); dump_isodirrec(stderr, isodirrec); fprintf(stderr, "\n"); } else if (*(isodirrec + DR_interleave_offset) != 0) { fprintf(stderr, "\n ** Unexpected interleave: "); dump_isodirrec(stderr, isodirrec); fprintf(stderr, "\n"); } else if (*(isodirrec + DR_ext_attr_length_offset) != 0) { fprintf(stderr, "\n ** Unexpected extended attribute length: "); dump_isodirrec(stderr, isodirrec); fprintf(stderr, "\n"); } #endif register_file(iso9660, file); return (file); } static int parse_rockridge(struct archive_read *a, struct file_info *file, const unsigned char *p, const unsigned char *end) { struct iso9660 *iso9660; iso9660 = (struct iso9660 *)(a->format->data); while (p + 4 <= end /* Enough space for another entry. */ && p[0] >= 'A' && p[0] <= 'Z' /* Sanity-check 1st char of name. */ && p[1] >= 'A' && p[1] <= 'Z' /* Sanity-check 2nd char of name. */ && p[2] >= 4 /* Sanity-check length. */ && p + p[2] <= end) { /* Sanity-check length. */ const unsigned char *data = p + 4; int data_length = p[2] - 4; int version = p[3]; switch(p[0]) { case 'C': if (p[1] == 'E') { if (version == 1 && data_length == 24) { /* * CE extension comprises: * 8 byte sector containing extension * 8 byte offset w/in above sector * 8 byte length of continuation */ int32_t location = archive_le32dec(data); file->ce_offset = archive_le32dec(data+8); file->ce_size = archive_le32dec(data+16); if (register_CE(a, location, file) != ARCHIVE_OK) return (ARCHIVE_FATAL); } } else if (p[1] == 'L') { if (version == 1 && data_length == 8) { file->cl_offset = (uint64_t) iso9660->logical_block_size * (uint64_t)archive_le32dec(data); iso9660->seenRockridge = 1; } } break; case 'N': if (p[1] == 'M') { if (version == 1) { parse_rockridge_NM1(file, data, data_length); iso9660->seenRockridge = 1; } } break; case 'P': /* * PD extension is padding; * contents are always ignored. * * PL extension won't appear; * contents are always ignored. */ if (p[1] == 'N') { if (version == 1 && data_length == 16) { file->rdev = toi(data,4); file->rdev <<= 32; file->rdev |= toi(data + 8, 4); iso9660->seenRockridge = 1; } } else if (p[1] == 'X') { /* * PX extension comprises: * 8 bytes for mode, * 8 bytes for nlinks, * 8 bytes for uid, * 8 bytes for gid, * 8 bytes for inode. */ if (version == 1) { if (data_length >= 8) file->mode = toi(data, 4); if (data_length >= 16) file->nlinks = toi(data + 8, 4); if (data_length >= 24) file->uid = toi(data + 16, 4); if (data_length >= 32) file->gid = toi(data + 24, 4); if (data_length >= 40) file->number = toi(data + 32, 4); iso9660->seenRockridge = 1; } } break; case 'R': if (p[1] == 'E' && version == 1) { file->re = 1; iso9660->seenRockridge = 1; } else if (p[1] == 'R' && version == 1) { /* * RR extension comprises: * one byte flag value * This extension is obsolete, * so contents are always ignored. */ } break; case 'S': if (p[1] == 'L') { if (version == 1) { parse_rockridge_SL1(file, data, data_length); iso9660->seenRockridge = 1; } } else if (p[1] == 'T' && data_length == 0 && version == 1) { /* * ST extension marks end of this * block of SUSP entries. * * It allows SUSP to coexist with * non-SUSP uses of the System * Use Area by placing non-SUSP data * after SUSP data. */ iso9660->seenSUSP = 0; iso9660->seenRockridge = 0; return (ARCHIVE_OK); } break; case 'T': if (p[1] == 'F') { if (version == 1) { parse_rockridge_TF1(file, data, data_length); iso9660->seenRockridge = 1; } } break; case 'Z': if (p[1] == 'F') { if (version == 1) parse_rockridge_ZF1(file, data, data_length); } break; default: break; } p += p[2]; } return (ARCHIVE_OK); } static int register_CE(struct archive_read *a, int32_t location, struct file_info *file) { struct iso9660 *iso9660; struct read_ce_queue *heap; struct read_ce_req *p; uint64_t offset, parent_offset; int hole, parent; iso9660 = (struct iso9660 *)(a->format->data); offset = ((uint64_t)location) * (uint64_t)iso9660->logical_block_size; if (((file->mode & AE_IFMT) == AE_IFREG && offset >= file->offset) || offset < iso9660->current_position || (((uint64_t)file->ce_offset) + file->ce_size) > (uint64_t)iso9660->logical_block_size || offset + file->ce_offset + file->ce_size > iso9660->volume_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid parameter in SUSP \"CE\" extension"); return (ARCHIVE_FATAL); } /* Expand our CE list as necessary. */ heap = &(iso9660->read_ce_req); if (heap->cnt >= heap->allocated) { int new_size; if (heap->allocated < 16) new_size = 16; else new_size = heap->allocated * 2; /* Overflow might keep us from growing the list. */ if (new_size <= heap->allocated) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } p = calloc(new_size, sizeof(p[0])); if (p == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } if (heap->reqs != NULL) { memcpy(p, heap->reqs, heap->cnt * sizeof(*p)); free(heap->reqs); } heap->reqs = p; heap->allocated = new_size; } /* * Start with hole at end, walk it up tree to find insertion point. */ hole = heap->cnt++; while (hole > 0) { parent = (hole - 1)/2; parent_offset = heap->reqs[parent].offset; if (offset >= parent_offset) { heap->reqs[hole].offset = offset; heap->reqs[hole].file = file; return (ARCHIVE_OK); } /* Move parent into hole <==> move hole up tree. */ heap->reqs[hole] = heap->reqs[parent]; hole = parent; } heap->reqs[0].offset = offset; heap->reqs[0].file = file; return (ARCHIVE_OK); } static void next_CE(struct read_ce_queue *heap) { uint64_t a_offset, b_offset, c_offset; int a, b, c; struct read_ce_req tmp; if (heap->cnt < 1) return; /* * Move the last item in the heap to the root of the tree */ heap->reqs[0] = heap->reqs[--(heap->cnt)]; /* * Rebalance the heap. */ a = 0; /* Starting element and its offset */ a_offset = heap->reqs[a].offset; for (;;) { b = a + a + 1; /* First child */ if (b >= heap->cnt) return; b_offset = heap->reqs[b].offset; c = b + 1; /* Use second child if it is smaller. */ if (c < heap->cnt) { c_offset = heap->reqs[c].offset; if (c_offset < b_offset) { b = c; b_offset = c_offset; } } if (a_offset <= b_offset) return; tmp = heap->reqs[a]; heap->reqs[a] = heap->reqs[b]; heap->reqs[b] = tmp; a = b; } } static int read_CE(struct archive_read *a, struct iso9660 *iso9660) { struct read_ce_queue *heap; const unsigned char *b, *p, *end; struct file_info *file; size_t step; int r; /* Read data which RRIP "CE" extension points. */ heap = &(iso9660->read_ce_req); step = iso9660->logical_block_size; while (heap->cnt && heap->reqs[0].offset == iso9660->current_position) { b = __archive_read_ahead(a, step, NULL); if (b == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to read full block when scanning " "ISO9660 directory list"); return (ARCHIVE_FATAL); } do { file = heap->reqs[0].file; if (file->ce_offset + file->ce_size > step) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Malformed CE information"); return (ARCHIVE_FATAL); } p = b + file->ce_offset; end = p + file->ce_size; next_CE(heap); r = parse_rockridge(a, file, p, end); if (r != ARCHIVE_OK) return (ARCHIVE_FATAL); } while (heap->cnt && heap->reqs[0].offset == iso9660->current_position); /* NOTE: Do not move this consume's code to front of * do-while loop. Registration of nested CE extension * might cause error because of current position. */ __archive_read_consume(a, step); iso9660->current_position += step; } return (ARCHIVE_OK); } static void parse_rockridge_NM1(struct file_info *file, const unsigned char *data, int data_length) { if (!file->name_continues) archive_string_empty(&file->name); file->name_continues = 0; if (data_length < 1) return; /* * NM version 1 extension comprises: * 1 byte flag, value is one of: * = 0: remainder is name * = 1: remainder is name, next NM entry continues name * = 2: "." * = 4: ".." * = 32: Implementation specific * All other values are reserved. */ switch(data[0]) { case 0: if (data_length < 2) return; archive_strncat(&file->name, (const char *)data + 1, data_length - 1); break; case 1: if (data_length < 2) return; archive_strncat(&file->name, (const char *)data + 1, data_length - 1); file->name_continues = 1; break; case 2: archive_strcat(&file->name, "."); break; case 4: archive_strcat(&file->name, ".."); break; default: return; } } static void parse_rockridge_TF1(struct file_info *file, const unsigned char *data, int data_length) { char flag; /* * TF extension comprises: * one byte flag * create time (optional) * modify time (optional) * access time (optional) * attribute time (optional) * Time format and presence of fields * is controlled by flag bits. */ if (data_length < 1) return; flag = data[0]; ++data; --data_length; if (flag & 0x80) { /* Use 17-byte time format. */ if ((flag & 1) && data_length >= 17) { /* Create time. */ file->birthtime_is_set = 1; file->birthtime = isodate17(data); data += 17; data_length -= 17; } if ((flag & 2) && data_length >= 17) { /* Modify time. */ file->mtime = isodate17(data); data += 17; data_length -= 17; } if ((flag & 4) && data_length >= 17) { /* Access time. */ file->atime = isodate17(data); data += 17; data_length -= 17; } if ((flag & 8) && data_length >= 17) { /* Attribute change time. */ file->ctime = isodate17(data); } } else { /* Use 7-byte time format. */ if ((flag & 1) && data_length >= 7) { /* Create time. */ file->birthtime_is_set = 1; file->birthtime = isodate7(data); data += 7; data_length -= 7; } if ((flag & 2) && data_length >= 7) { /* Modify time. */ file->mtime = isodate7(data); data += 7; data_length -= 7; } if ((flag & 4) && data_length >= 7) { /* Access time. */ file->atime = isodate7(data); data += 7; data_length -= 7; } if ((flag & 8) && data_length >= 7) { /* Attribute change time. */ file->ctime = isodate7(data); } } } static void parse_rockridge_SL1(struct file_info *file, const unsigned char *data, int data_length) { const char *separator = ""; if (!file->symlink_continues || file->symlink.length < 1) archive_string_empty(&file->symlink); file->symlink_continues = 0; /* * Defined flag values: * 0: This is the last SL record for this symbolic link * 1: this symbolic link field continues in next SL entry * All other values are reserved. */ if (data_length < 1) return; switch(*data) { case 0: break; case 1: file->symlink_continues = 1; break; default: return; } ++data; /* Skip flag byte. */ --data_length; /* * SL extension body stores "components". * Basically, this is a complicated way of storing * a POSIX path. It also interferes with using * symlinks for storing non-path data. * * Each component is 2 bytes (flag and length) * possibly followed by name data. */ while (data_length >= 2) { unsigned char flag = *data++; unsigned char nlen = *data++; data_length -= 2; archive_strcat(&file->symlink, separator); separator = "/"; switch(flag) { case 0: /* Usual case, this is text. */ if (data_length < nlen) return; archive_strncat(&file->symlink, (const char *)data, nlen); break; case 0x01: /* Text continues in next component. */ if (data_length < nlen) return; archive_strncat(&file->symlink, (const char *)data, nlen); separator = ""; break; case 0x02: /* Current dir. */ archive_strcat(&file->symlink, "."); break; case 0x04: /* Parent dir. */ archive_strcat(&file->symlink, ".."); break; case 0x08: /* Root of filesystem. */ archive_strcat(&file->symlink, "/"); separator = ""; break; case 0x10: /* Undefined (historically "volume root" */ archive_string_empty(&file->symlink); archive_strcat(&file->symlink, "ROOT"); break; case 0x20: /* Undefined (historically "hostname") */ archive_strcat(&file->symlink, "hostname"); break; default: /* TODO: issue a warning ? */ return; } data += nlen; data_length -= nlen; } } static void parse_rockridge_ZF1(struct file_info *file, const unsigned char *data, int data_length) { if (data[0] == 0x70 && data[1] == 0x7a && data_length == 12) { /* paged zlib */ file->pz = 1; file->pz_log2_bs = data[3]; file->pz_uncompressed_size = archive_le32dec(&data[4]); } } static void register_file(struct iso9660 *iso9660, struct file_info *file) { file->use_next = iso9660->use_files; iso9660->use_files = file; } static void release_files(struct iso9660 *iso9660) { struct content *con, *connext; struct file_info *file; file = iso9660->use_files; while (file != NULL) { struct file_info *next = file->use_next; archive_string_free(&file->name); archive_string_free(&file->symlink); free(file->utf16be_name); con = file->contents.first; while (con != NULL) { connext = con->next; free(con); con = connext; } free(file); file = next; } } static int next_entry_seek(struct archive_read *a, struct iso9660 *iso9660, struct file_info **pfile) { struct file_info *file; int r; r = next_cache_entry(a, iso9660, pfile); if (r != ARCHIVE_OK) return (r); file = *pfile; /* Don't waste time seeking for zero-length bodies. */ if (file->size == 0) file->offset = iso9660->current_position; /* flush any remaining bytes from the last round to ensure * we're positioned */ if (iso9660->entry_bytes_unconsumed) { __archive_read_consume(a, iso9660->entry_bytes_unconsumed); iso9660->entry_bytes_unconsumed = 0; } /* Seek forward to the start of the entry. */ if (iso9660->current_position < file->offset) { int64_t step; step = file->offset - iso9660->current_position; step = __archive_read_consume(a, step); if (step < 0) return ((int)step); iso9660->current_position = file->offset; } /* We found body of file; handle it now. */ return (ARCHIVE_OK); } static int next_cache_entry(struct archive_read *a, struct iso9660 *iso9660, struct file_info **pfile) { struct file_info *file; struct { struct file_info *first; struct file_info **last; } empty_files; int64_t number; int count; file = cache_get_entry(iso9660); if (file != NULL) { *pfile = file; return (ARCHIVE_OK); } for (;;) { struct file_info *re, *d; *pfile = file = next_entry(iso9660); if (file == NULL) { /* * If directory entries all which are descendant of * rr_moved are still remaining, expose their. */ if (iso9660->re_files.first != NULL && iso9660->rr_moved != NULL && iso9660->rr_moved->rr_moved_has_re_only) /* Expose "rr_moved" entry. */ cache_add_entry(iso9660, iso9660->rr_moved); while ((re = re_get_entry(iso9660)) != NULL) { /* Expose its descendant dirs. */ while ((d = rede_get_entry(re)) != NULL) cache_add_entry(iso9660, d); } if (iso9660->cache_files.first != NULL) return (next_cache_entry(a, iso9660, pfile)); return (ARCHIVE_EOF); } if (file->cl_offset) { struct file_info *first_re = NULL; int nexted_re = 0; /* * Find "RE" dir for the current file, which * has "CL" flag. */ while ((re = re_get_entry(iso9660)) != first_re) { if (first_re == NULL) first_re = re; if (re->offset == file->cl_offset) { re->parent->subdirs--; re->parent = file->parent; re->re = 0; if (re->parent->re_descendant) { nexted_re = 1; re->re_descendant = 1; if (rede_add_entry(re) < 0) goto fatal_rr; /* Move a list of descendants * to a new ancestor. */ while ((d = rede_get_entry( re)) != NULL) if (rede_add_entry(d) < 0) goto fatal_rr; break; } /* Replace the current file * with "RE" dir */ *pfile = file = re; /* Expose its descendant */ while ((d = rede_get_entry( file)) != NULL) cache_add_entry( iso9660, d); break; } else re_add_entry(iso9660, re); } if (nexted_re) { /* * Do not expose this at this time * because we have not gotten its full-path * name yet. */ continue; } } else if ((file->mode & AE_IFMT) == AE_IFDIR) { int r; /* Read file entries in this dir. */ r = read_children(a, file); if (r != ARCHIVE_OK) return (r); /* * Handle a special dir of Rockridge extensions, * "rr_moved". */ if (file->rr_moved) { /* * If this has only the subdirectories which * have "RE" flags, do not expose at this time. */ if (file->rr_moved_has_re_only) continue; /* Otherwise expose "rr_moved" entry. */ } else if (file->re) { /* * Do not expose this at this time * because we have not gotten its full-path * name yet. */ re_add_entry(iso9660, file); continue; } else if (file->re_descendant) { /* * If the top level "RE" entry of this entry * is not exposed, we, accordingly, should not * expose this entry at this time because * we cannot make its proper full-path name. */ if (rede_add_entry(file) == 0) continue; /* Otherwise we can expose this entry because * it seems its top level "RE" has already been * exposed. */ } } break; } if ((file->mode & AE_IFMT) != AE_IFREG || file->number == -1) return (ARCHIVE_OK); count = 0; number = file->number; iso9660->cache_files.first = NULL; iso9660->cache_files.last = &(iso9660->cache_files.first); empty_files.first = NULL; empty_files.last = &empty_files.first; /* Collect files which has the same file serial number. * Peek pending_files so that file which number is different * is not put back. */ while (iso9660->pending_files.used > 0 && (iso9660->pending_files.files[0]->number == -1 || iso9660->pending_files.files[0]->number == number)) { if (file->number == -1) { /* This file has the same offset * but it's wrong offset which empty files * and symlink files have. * NOTE: This wrong offset was recorded by * old mkisofs utility. If ISO images is * created by latest mkisofs, this does not * happen. */ file->next = NULL; *empty_files.last = file; empty_files.last = &(file->next); } else { count++; cache_add_entry(iso9660, file); } file = next_entry(iso9660); } if (count == 0) { *pfile = file; return ((file == NULL)?ARCHIVE_EOF:ARCHIVE_OK); } if (file->number == -1) { file->next = NULL; *empty_files.last = file; empty_files.last = &(file->next); } else { count++; cache_add_entry(iso9660, file); } if (count > 1) { /* The count is the same as number of hardlink, * so much so that each nlinks of files in cache_file * is overwritten by value of the count. */ for (file = iso9660->cache_files.first; file != NULL; file = file->next) file->nlinks = count; } /* If there are empty files, that files are added * to the tail of the cache_files. */ if (empty_files.first != NULL) { *iso9660->cache_files.last = empty_files.first; iso9660->cache_files.last = empty_files.last; } *pfile = cache_get_entry(iso9660); return ((*pfile == NULL)?ARCHIVE_EOF:ARCHIVE_OK); fatal_rr: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to connect 'CL' pointer to 'RE' rr_moved pointer of " "Rockridge extensions: current position = %jd, CL offset = %jd", (intmax_t)iso9660->current_position, (intmax_t)file->cl_offset); return (ARCHIVE_FATAL); } static inline void re_add_entry(struct iso9660 *iso9660, struct file_info *file) { file->re_next = NULL; *iso9660->re_files.last = file; iso9660->re_files.last = &(file->re_next); } static inline struct file_info * re_get_entry(struct iso9660 *iso9660) { struct file_info *file; if ((file = iso9660->re_files.first) != NULL) { iso9660->re_files.first = file->re_next; if (iso9660->re_files.first == NULL) iso9660->re_files.last = &(iso9660->re_files.first); } return (file); } static inline int rede_add_entry(struct file_info *file) { struct file_info *re; /* * Find "RE" entry. */ re = file->parent; while (re != NULL && !re->re) re = re->parent; if (re == NULL) return (-1); file->re_next = NULL; *re->rede_files.last = file; re->rede_files.last = &(file->re_next); return (0); } static inline struct file_info * rede_get_entry(struct file_info *re) { struct file_info *file; if ((file = re->rede_files.first) != NULL) { re->rede_files.first = file->re_next; if (re->rede_files.first == NULL) re->rede_files.last = &(re->rede_files.first); } return (file); } static inline void cache_add_entry(struct iso9660 *iso9660, struct file_info *file) { file->next = NULL; *iso9660->cache_files.last = file; iso9660->cache_files.last = &(file->next); } static inline struct file_info * cache_get_entry(struct iso9660 *iso9660) { struct file_info *file; if ((file = iso9660->cache_files.first) != NULL) { iso9660->cache_files.first = file->next; if (iso9660->cache_files.first == NULL) iso9660->cache_files.last = &(iso9660->cache_files.first); } return (file); } static int heap_add_entry(struct archive_read *a, struct heap_queue *heap, struct file_info *file, uint64_t key) { uint64_t file_key, parent_key; int hole, parent; /* Expand our pending files list as necessary. */ if (heap->used >= heap->allocated) { struct file_info **new_pending_files; int new_size = heap->allocated * 2; if (heap->allocated < 1024) new_size = 1024; /* Overflow might keep us from growing the list. */ if (new_size <= heap->allocated) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } new_pending_files = (struct file_info **) malloc(new_size * sizeof(new_pending_files[0])); if (new_pending_files == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } memcpy(new_pending_files, heap->files, heap->allocated * sizeof(new_pending_files[0])); if (heap->files != NULL) free(heap->files); heap->files = new_pending_files; heap->allocated = new_size; } file_key = file->key = key; /* * Start with hole at end, walk it up tree to find insertion point. */ hole = heap->used++; while (hole > 0) { parent = (hole - 1)/2; parent_key = heap->files[parent]->key; if (file_key >= parent_key) { heap->files[hole] = file; return (ARCHIVE_OK); } /* Move parent into hole <==> move hole up tree. */ heap->files[hole] = heap->files[parent]; hole = parent; } heap->files[0] = file; return (ARCHIVE_OK); } static struct file_info * heap_get_entry(struct heap_queue *heap) { uint64_t a_key, b_key, c_key; int a, b, c; struct file_info *r, *tmp; if (heap->used < 1) return (NULL); /* * The first file in the list is the earliest; we'll return this. */ r = heap->files[0]; /* * Move the last item in the heap to the root of the tree */ heap->files[0] = heap->files[--(heap->used)]; /* * Rebalance the heap. */ a = 0; /* Starting element and its heap key */ a_key = heap->files[a]->key; for (;;) { b = a + a + 1; /* First child */ if (b >= heap->used) return (r); b_key = heap->files[b]->key; c = b + 1; /* Use second child if it is smaller. */ if (c < heap->used) { c_key = heap->files[c]->key; if (c_key < b_key) { b = c; b_key = c_key; } } if (a_key <= b_key) return (r); tmp = heap->files[a]; heap->files[a] = heap->files[b]; heap->files[b] = tmp; a = b; } } static unsigned int toi(const void *p, int n) { const unsigned char *v = (const unsigned char *)p; if (n > 1) return v[0] + 256 * toi(v + 1, n - 1); if (n == 1) return v[0]; return (0); } static time_t isodate7(const unsigned char *v) { struct tm tm; int offset; time_t t; memset(&tm, 0, sizeof(tm)); tm.tm_year = v[0]; tm.tm_mon = v[1] - 1; tm.tm_mday = v[2]; tm.tm_hour = v[3]; tm.tm_min = v[4]; tm.tm_sec = v[5]; /* v[6] is the signed timezone offset, in 1/4-hour increments. */ offset = ((const signed char *)v)[6]; if (offset > -48 && offset < 52) { tm.tm_hour -= offset / 4; tm.tm_min -= (offset % 4) * 15; } t = time_from_tm(&tm); if (t == (time_t)-1) return ((time_t)0); return (t); } static time_t isodate17(const unsigned char *v) { struct tm tm; int offset; time_t t; memset(&tm, 0, sizeof(tm)); tm.tm_year = (v[0] - '0') * 1000 + (v[1] - '0') * 100 + (v[2] - '0') * 10 + (v[3] - '0') - 1900; tm.tm_mon = (v[4] - '0') * 10 + (v[5] - '0'); tm.tm_mday = (v[6] - '0') * 10 + (v[7] - '0'); tm.tm_hour = (v[8] - '0') * 10 + (v[9] - '0'); tm.tm_min = (v[10] - '0') * 10 + (v[11] - '0'); tm.tm_sec = (v[12] - '0') * 10 + (v[13] - '0'); /* v[16] is the signed timezone offset, in 1/4-hour increments. */ offset = ((const signed char *)v)[16]; if (offset > -48 && offset < 52) { tm.tm_hour -= offset / 4; tm.tm_min -= (offset % 4) * 15; } t = time_from_tm(&tm); if (t == (time_t)-1) return ((time_t)0); return (t); } static time_t time_from_tm(struct tm *t) { #if HAVE_TIMEGM /* Use platform timegm() if available. */ return (timegm(t)); #elif HAVE__MKGMTIME64 return (_mkgmtime64(t)); #else /* Else use direct calculation using POSIX assumptions. */ /* First, fix up tm_yday based on the year/month/day. */ if (mktime(t) == (time_t)-1) return ((time_t)-1); /* Then we can compute timegm() from first principles. */ return (t->tm_sec + t->tm_min * 60 + t->tm_hour * 3600 + t->tm_yday * 86400 + (t->tm_year - 70) * 31536000 + ((t->tm_year - 69) / 4) * 86400 - ((t->tm_year - 1) / 100) * 86400 + ((t->tm_year + 299) / 400) * 86400); #endif } static const char * build_pathname(struct archive_string *as, struct file_info *file, int depth) { // Plain ISO9660 only allows 8 dir levels; if we get // to 1000, then something is very, very wrong. if (depth > 1000) { return NULL; } if (file->parent != NULL && archive_strlen(&file->parent->name) > 0) { if (build_pathname(as, file->parent, depth + 1) == NULL) { return NULL; } archive_strcat(as, "/"); } if (archive_strlen(&file->name) == 0) archive_strcat(as, "."); else archive_string_concat(as, &file->name); return (as->s); } static int build_pathname_utf16be(unsigned char *p, size_t max, size_t *len, struct file_info *file) { if (file->parent != NULL && file->parent->utf16be_bytes > 0) { if (build_pathname_utf16be(p, max, len, file->parent) != 0) return (-1); p[*len] = 0; p[*len + 1] = '/'; *len += 2; } if (file->utf16be_bytes == 0) { if (*len + 2 > max) return (-1);/* Path is too long! */ p[*len] = 0; p[*len + 1] = '.'; *len += 2; } else { if (*len + file->utf16be_bytes > max) return (-1);/* Path is too long! */ memcpy(p + *len, file->utf16be_name, file->utf16be_bytes); *len += file->utf16be_bytes; } return (0); } #if DEBUG static void dump_isodirrec(FILE *out, const unsigned char *isodirrec) { fprintf(out, " l %d,", toi(isodirrec + DR_length_offset, DR_length_size)); fprintf(out, " a %d,", toi(isodirrec + DR_ext_attr_length_offset, DR_ext_attr_length_size)); fprintf(out, " ext 0x%x,", toi(isodirrec + DR_extent_offset, DR_extent_size)); fprintf(out, " s %d,", toi(isodirrec + DR_size_offset, DR_extent_size)); fprintf(out, " f 0x%x,", toi(isodirrec + DR_flags_offset, DR_flags_size)); fprintf(out, " u %d,", toi(isodirrec + DR_file_unit_size_offset, DR_file_unit_size_size)); fprintf(out, " ilv %d,", toi(isodirrec + DR_interleave_offset, DR_interleave_size)); fprintf(out, " seq %d,", toi(isodirrec + DR_volume_sequence_number_offset, DR_volume_sequence_number_size)); fprintf(out, " nl %d:", toi(isodirrec + DR_name_len_offset, DR_name_len_size)); fprintf(out, " `%.*s'", toi(isodirrec + DR_name_len_offset, DR_name_len_size), isodirrec + DR_name_offset); } #endif Index: head/contrib/libarchive/libarchive/archive_read_support_format_lha.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_format_lha.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_read_support_format_lha.c (revision 311903) @@ -1,2811 +1,2811 @@ /*- * Copyright (c) 2008-2014 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_read_private.h" #include "archive_endian.h" #define MAXMATCH 256 /* Maximum match length. */ #define MINMATCH 3 /* Minimum match length. */ /* * Literal table format: * +0 +256 +510 * +---------------+-------------------------+ * | literal code | match length | * | 0 ... 255 | MINMATCH ... MAXMATCH | * +---------------+-------------------------+ * <--- LT_BITLEN_SIZE ---> */ /* Literal table size. */ #define LT_BITLEN_SIZE (UCHAR_MAX + 1 + MAXMATCH - MINMATCH + 1) /* Position table size. * Note: this used for both position table and pre literal table.*/ #define PT_BITLEN_SIZE (3 + 16) struct lzh_dec { /* Decoding status. */ int state; /* * Window to see last 8Ki(lh5),32Ki(lh6),64Ki(lh7) bytes of decoded * data. */ int w_size; int w_mask; /* Window buffer, which is a loop buffer. */ unsigned char *w_buff; /* The insert position to the window. */ int w_pos; /* The position where we can copy decoded code from the window. */ int copy_pos; /* The length how many bytes we can copy decoded code from * the window. */ int copy_len; /* * Bit stream reader. */ struct lzh_br { #define CACHE_TYPE uint64_t #define CACHE_BITS (8 * sizeof(CACHE_TYPE)) /* Cache buffer. */ CACHE_TYPE cache_buffer; /* Indicates how many bits avail in cache_buffer. */ int cache_avail; } br; /* * Huffman coding. */ struct huffman { int len_size; int len_avail; int len_bits; int freq[17]; unsigned char *bitlen; /* * Use a index table. It's faster than searching a huffman * coding tree, which is a binary tree. But a use of a large * index table causes L1 cache read miss many times. */ #define HTBL_BITS 10 int max_bits; int shift_bits; int tbl_bits; int tree_used; int tree_avail; /* Direct access table. */ uint16_t *tbl; /* Binary tree table for extra bits over the direct access. */ struct htree_t { uint16_t left; uint16_t right; } *tree; } lt, pt; int blocks_avail; int pos_pt_len_size; int pos_pt_len_bits; int literal_pt_len_size; int literal_pt_len_bits; int reading_position; int loop; int error; }; struct lzh_stream { const unsigned char *next_in; int avail_in; int64_t total_in; const unsigned char *ref_ptr; int avail_out; int64_t total_out; struct lzh_dec *ds; }; struct lha { /* entry_bytes_remaining is the number of bytes we expect. */ int64_t entry_offset; int64_t entry_bytes_remaining; int64_t entry_unconsumed; uint16_t entry_crc_calculated; size_t header_size; /* header size */ unsigned char level; /* header level */ char method[3]; /* compress type */ int64_t compsize; /* compressed data size */ int64_t origsize; /* original file size */ int setflag; #define BIRTHTIME_IS_SET 1 #define ATIME_IS_SET 2 #define UNIX_MODE_IS_SET 4 #define CRC_IS_SET 8 time_t birthtime; long birthtime_tv_nsec; time_t mtime; long mtime_tv_nsec; time_t atime; long atime_tv_nsec; mode_t mode; int64_t uid; int64_t gid; struct archive_string uname; struct archive_string gname; uint16_t header_crc; uint16_t crc; struct archive_string_conv *sconv; struct archive_string_conv *opt_sconv; struct archive_string dirname; struct archive_string filename; struct archive_wstring ws; unsigned char dos_attr; /* Flag to mark progress that an archive was read their first header.*/ char found_first_header; /* Flag to mark that indicates an empty directory. */ char directory; /* Flags to mark progress of decompression. */ char decompress_init; char end_of_entry; char end_of_entry_cleanup; char entry_is_compressed; char format_name[64]; struct lzh_stream strm; }; /* * LHA header common member offset. */ #define H_METHOD_OFFSET 2 /* Compress type. */ #define H_ATTR_OFFSET 19 /* DOS attribute. */ #define H_LEVEL_OFFSET 20 /* Header Level. */ #define H_SIZE 22 /* Minimum header size. */ static int archive_read_format_lha_bid(struct archive_read *, int); static int archive_read_format_lha_options(struct archive_read *, const char *, const char *); static int archive_read_format_lha_read_header(struct archive_read *, struct archive_entry *); static int archive_read_format_lha_read_data(struct archive_read *, const void **, size_t *, int64_t *); static int archive_read_format_lha_read_data_skip(struct archive_read *); static int archive_read_format_lha_cleanup(struct archive_read *); static void lha_replace_path_separator(struct lha *, struct archive_entry *); static int lha_read_file_header_0(struct archive_read *, struct lha *); static int lha_read_file_header_1(struct archive_read *, struct lha *); static int lha_read_file_header_2(struct archive_read *, struct lha *); static int lha_read_file_header_3(struct archive_read *, struct lha *); static int lha_read_file_extended_header(struct archive_read *, struct lha *, uint16_t *, int, size_t, size_t *); static size_t lha_check_header_format(const void *); static int lha_skip_sfx(struct archive_read *); static time_t lha_dos_time(const unsigned char *); static time_t lha_win_time(uint64_t, long *); static unsigned char lha_calcsum(unsigned char, const void *, int, size_t); static int lha_parse_linkname(struct archive_string *, struct archive_string *); static int lha_read_data_none(struct archive_read *, const void **, size_t *, int64_t *); static int lha_read_data_lzh(struct archive_read *, const void **, size_t *, int64_t *); static void lha_crc16_init(void); static uint16_t lha_crc16(uint16_t, const void *, size_t); static int lzh_decode_init(struct lzh_stream *, const char *); static void lzh_decode_free(struct lzh_stream *); static int lzh_decode(struct lzh_stream *, int); static int lzh_br_fillup(struct lzh_stream *, struct lzh_br *); static int lzh_huffman_init(struct huffman *, size_t, int); static void lzh_huffman_free(struct huffman *); static int lzh_read_pt_bitlen(struct lzh_stream *, int start, int end); static int lzh_make_fake_table(struct huffman *, uint16_t); static int lzh_make_huffman_table(struct huffman *); static inline int lzh_decode_huffman(struct huffman *, unsigned); static int lzh_decode_huffman_tree(struct huffman *, unsigned, int); int archive_read_support_format_lha(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct lha *lha; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_lha"); lha = (struct lha *)calloc(1, sizeof(*lha)); if (lha == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate lha data"); return (ARCHIVE_FATAL); } archive_string_init(&lha->ws); r = __archive_read_register_format(a, lha, "lha", archive_read_format_lha_bid, archive_read_format_lha_options, archive_read_format_lha_read_header, archive_read_format_lha_read_data, archive_read_format_lha_read_data_skip, NULL, archive_read_format_lha_cleanup, NULL, NULL); if (r != ARCHIVE_OK) free(lha); return (ARCHIVE_OK); } static size_t lha_check_header_format(const void *h) { const unsigned char *p = h; size_t next_skip_bytes; switch (p[H_METHOD_OFFSET+3]) { /* * "-lh0-" ... "-lh7-" "-lhd-" * "-lzs-" "-lz5-" */ case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case 'd': case 's': next_skip_bytes = 4; /* b0 == 0 means the end of an LHa archive file. */ if (p[0] == 0) break; if (p[H_METHOD_OFFSET] != '-' || p[H_METHOD_OFFSET+1] != 'l' || p[H_METHOD_OFFSET+4] != '-') break; if (p[H_METHOD_OFFSET+2] == 'h') { /* "-lh?-" */ if (p[H_METHOD_OFFSET+3] == 's') break; if (p[H_LEVEL_OFFSET] == 0) return (0); if (p[H_LEVEL_OFFSET] <= 3 && p[H_ATTR_OFFSET] == 0x20) return (0); } if (p[H_METHOD_OFFSET+2] == 'z') { /* LArc extensions: -lzs-,-lz4- and -lz5- */ if (p[H_LEVEL_OFFSET] != 0) break; if (p[H_METHOD_OFFSET+3] == 's' || p[H_METHOD_OFFSET+3] == '4' || p[H_METHOD_OFFSET+3] == '5') return (0); } break; case 'h': next_skip_bytes = 1; break; case 'z': next_skip_bytes = 1; break; case 'l': next_skip_bytes = 2; break; case '-': next_skip_bytes = 3; break; default : next_skip_bytes = 4; break; } return (next_skip_bytes); } static int archive_read_format_lha_bid(struct archive_read *a, int best_bid) { const char *p; const void *buff; ssize_t bytes_avail, offset, window; size_t next; /* If there's already a better bid than we can ever make, don't bother testing. */ if (best_bid > 30) return (-1); if ((p = __archive_read_ahead(a, H_SIZE, NULL)) == NULL) return (-1); if (lha_check_header_format(p) == 0) return (30); if (p[0] == 'M' && p[1] == 'Z') { /* PE file */ offset = 0; window = 4096; while (offset < (1024 * 20)) { buff = __archive_read_ahead(a, offset + window, &bytes_avail); if (buff == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < (H_SIZE + 3)) return (0); continue; } p = (const char *)buff + offset; while (p + H_SIZE < (const char *)buff + bytes_avail) { if ((next = lha_check_header_format(p)) == 0) return (30); p += next; } offset = p - (const char *)buff; } } return (0); } static int archive_read_format_lha_options(struct archive_read *a, const char *key, const char *val) { struct lha *lha; int ret = ARCHIVE_FAILED; lha = (struct lha *)(a->format->data); if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "lha: hdrcharset option needs a character-set name"); else { lha->opt_sconv = archive_string_conversion_from_charset( &a->archive, val, 0); if (lha->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int lha_skip_sfx(struct archive_read *a) { const void *h; const char *p, *q; size_t next, skip; ssize_t bytes, window; window = 4096; for (;;) { h = __archive_read_ahead(a, window, &bytes); if (h == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < (H_SIZE + 3)) goto fatal; continue; } if (bytes < H_SIZE) goto fatal; p = h; q = p + bytes; /* * Scan ahead until we find something that looks * like the lha header. */ while (p + H_SIZE < q) { if ((next = lha_check_header_format(p)) == 0) { skip = p - (const char *)h; __archive_read_consume(a, skip); return (ARCHIVE_OK); } p += next; } skip = p - (const char *)h; __archive_read_consume(a, skip); } fatal: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Couldn't find out LHa header"); return (ARCHIVE_FATAL); } static int truncated_error(struct archive_read *a) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated LHa header"); return (ARCHIVE_FATAL); } static int archive_read_format_lha_read_header(struct archive_read *a, struct archive_entry *entry) { struct archive_string linkname; struct archive_string pathname; struct lha *lha; const unsigned char *p; const char *signature; int err; lha_crc16_init(); a->archive.archive_format = ARCHIVE_FORMAT_LHA; if (a->archive.archive_format_name == NULL) a->archive.archive_format_name = "lha"; lha = (struct lha *)(a->format->data); lha->decompress_init = 0; lha->end_of_entry = 0; lha->end_of_entry_cleanup = 0; lha->entry_unconsumed = 0; if ((p = __archive_read_ahead(a, H_SIZE, NULL)) == NULL) { /* * LHa archiver added 0 to the tail of its archive file as * the mark of the end of the archive. */ signature = __archive_read_ahead(a, sizeof(signature[0]), NULL); if (signature == NULL || signature[0] == 0) return (ARCHIVE_EOF); return (truncated_error(a)); } signature = (const char *)p; if (lha->found_first_header == 0 && signature[0] == 'M' && signature[1] == 'Z') { /* This is an executable? Must be self-extracting... */ err = lha_skip_sfx(a); if (err < ARCHIVE_WARN) return (err); if ((p = __archive_read_ahead(a, sizeof(*p), NULL)) == NULL) return (truncated_error(a)); signature = (const char *)p; } /* signature[0] == 0 means the end of an LHa archive file. */ if (signature[0] == 0) return (ARCHIVE_EOF); /* * Check the header format and method type. */ if (lha_check_header_format(p) != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Bad LHa file"); return (ARCHIVE_FATAL); } /* We've found the first header. */ lha->found_first_header = 1; /* Set a default value and common data */ lha->header_size = 0; lha->level = p[H_LEVEL_OFFSET]; lha->method[0] = p[H_METHOD_OFFSET+1]; lha->method[1] = p[H_METHOD_OFFSET+2]; lha->method[2] = p[H_METHOD_OFFSET+3]; if (memcmp(lha->method, "lhd", 3) == 0) lha->directory = 1; else lha->directory = 0; if (memcmp(lha->method, "lh0", 3) == 0 || memcmp(lha->method, "lz4", 3) == 0) lha->entry_is_compressed = 0; else lha->entry_is_compressed = 1; lha->compsize = 0; lha->origsize = 0; lha->setflag = 0; lha->birthtime = 0; lha->birthtime_tv_nsec = 0; lha->mtime = 0; lha->mtime_tv_nsec = 0; lha->atime = 0; lha->atime_tv_nsec = 0; lha->mode = (lha->directory)? 0777 : 0666; lha->uid = 0; lha->gid = 0; archive_string_empty(&lha->dirname); archive_string_empty(&lha->filename); lha->dos_attr = 0; if (lha->opt_sconv != NULL) lha->sconv = lha->opt_sconv; else lha->sconv = NULL; switch (p[H_LEVEL_OFFSET]) { case 0: err = lha_read_file_header_0(a, lha); break; case 1: err = lha_read_file_header_1(a, lha); break; case 2: err = lha_read_file_header_2(a, lha); break; case 3: err = lha_read_file_header_3(a, lha); break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported LHa header level %d", p[H_LEVEL_OFFSET]); err = ARCHIVE_FATAL; break; } if (err < ARCHIVE_WARN) return (err); if (!lha->directory && archive_strlen(&lha->filename) == 0) /* The filename has not been set */ return (truncated_error(a)); /* * Make a pathname from a dirname and a filename. */ archive_string_concat(&lha->dirname, &lha->filename); archive_string_init(&pathname); archive_string_init(&linkname); archive_string_copy(&pathname, &lha->dirname); if ((lha->mode & AE_IFMT) == AE_IFLNK) { /* * Extract the symlink-name if it's included in the pathname. */ if (!lha_parse_linkname(&linkname, &pathname)) { /* We couldn't get the symlink-name. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unknown symlink-name"); archive_string_free(&pathname); archive_string_free(&linkname); return (ARCHIVE_FAILED); } } else { /* * Make sure a file-type is set. * The mode has been overridden if it is in the extended data. */ lha->mode = (lha->mode & ~AE_IFMT) | ((lha->directory)? AE_IFDIR: AE_IFREG); } if ((lha->setflag & UNIX_MODE_IS_SET) == 0 && (lha->dos_attr & 1) != 0) lha->mode &= ~(0222);/* read only. */ /* * Set basic file parameters. */ if (archive_entry_copy_pathname_l(entry, pathname.s, pathname.length, lha->sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Pathname cannot be converted " "from %s to current locale.", archive_string_conversion_charset_name(lha->sconv)); err = ARCHIVE_WARN; } archive_string_free(&pathname); if (archive_strlen(&linkname) > 0) { if (archive_entry_copy_symlink_l(entry, linkname.s, linkname.length, lha->sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Linkname cannot be converted " "from %s to current locale.", archive_string_conversion_charset_name(lha->sconv)); err = ARCHIVE_WARN; } } else archive_entry_set_symlink(entry, NULL); archive_string_free(&linkname); /* * When a header level is 0, there is a possibility that * a pathname and a symlink has '\' character, a directory * separator in DOS/Windows. So we should convert it to '/'. */ if (p[H_LEVEL_OFFSET] == 0) lha_replace_path_separator(lha, entry); archive_entry_set_mode(entry, lha->mode); archive_entry_set_uid(entry, lha->uid); archive_entry_set_gid(entry, lha->gid); if (archive_strlen(&lha->uname) > 0) archive_entry_set_uname(entry, lha->uname.s); if (archive_strlen(&lha->gname) > 0) archive_entry_set_gname(entry, lha->gname.s); if (lha->setflag & BIRTHTIME_IS_SET) { archive_entry_set_birthtime(entry, lha->birthtime, lha->birthtime_tv_nsec); archive_entry_set_ctime(entry, lha->birthtime, lha->birthtime_tv_nsec); } else { archive_entry_unset_birthtime(entry); archive_entry_unset_ctime(entry); } archive_entry_set_mtime(entry, lha->mtime, lha->mtime_tv_nsec); if (lha->setflag & ATIME_IS_SET) archive_entry_set_atime(entry, lha->atime, lha->atime_tv_nsec); else archive_entry_unset_atime(entry); if (lha->directory || archive_entry_symlink(entry) != NULL) archive_entry_unset_size(entry); else archive_entry_set_size(entry, lha->origsize); /* * Prepare variables used to read a file content. */ lha->entry_bytes_remaining = lha->compsize; lha->entry_offset = 0; lha->entry_crc_calculated = 0; /* * This file does not have a content. */ if (lha->directory || lha->compsize == 0) lha->end_of_entry = 1; sprintf(lha->format_name, "lha -%c%c%c-", lha->method[0], lha->method[1], lha->method[2]); a->archive.archive_format_name = lha->format_name; return (err); } /* * Replace a DOS path separator '\' by a character '/'. * Some multi-byte character set have a character '\' in its second byte. */ static void lha_replace_path_separator(struct lha *lha, struct archive_entry *entry) { const wchar_t *wp; size_t i; if ((wp = archive_entry_pathname_w(entry)) != NULL) { archive_wstrcpy(&(lha->ws), wp); for (i = 0; i < archive_strlen(&(lha->ws)); i++) { if (lha->ws.s[i] == L'\\') lha->ws.s[i] = L'/'; } archive_entry_copy_pathname_w(entry, lha->ws.s); } if ((wp = archive_entry_symlink_w(entry)) != NULL) { archive_wstrcpy(&(lha->ws), wp); for (i = 0; i < archive_strlen(&(lha->ws)); i++) { if (lha->ws.s[i] == L'\\') lha->ws.s[i] = L'/'; } archive_entry_copy_symlink_w(entry, lha->ws.s); } } /* * Header 0 format * * +0 +1 +2 +7 +11 * +---------------+----------+----------------+-------------------+ * |header size(*1)|header sum|compression type|compressed size(*2)| * +---------------+----------+----------------+-------------------+ * <---------------------(*1)----------* * * +11 +15 +17 +19 +20 +21 * +-----------------+---------+---------+--------------+----------------+ * |uncompressed size|time(DOS)|date(DOS)|attribute(DOS)|header level(=0)| * +-----------------+---------+---------+--------------+----------------+ * *--------------------------------(*1)---------------------------------* * * +21 +22 +22+(*3) +22+(*3)+2 +22+(*3)+2+(*4) * +---------------+---------+----------+----------------+------------------+ * |name length(*3)|file name|file CRC16|extra header(*4)| compressed data | * +---------------+---------+----------+----------------+------------------+ * <--(*3)-> <------(*2)------> * *----------------------(*1)--------------------------> * */ #define H0_HEADER_SIZE_OFFSET 0 #define H0_HEADER_SUM_OFFSET 1 #define H0_COMP_SIZE_OFFSET 7 #define H0_ORIG_SIZE_OFFSET 11 #define H0_DOS_TIME_OFFSET 15 #define H0_NAME_LEN_OFFSET 21 #define H0_FILE_NAME_OFFSET 22 #define H0_FIXED_SIZE 24 static int lha_read_file_header_0(struct archive_read *a, struct lha *lha) { const unsigned char *p; int extdsize, namelen; unsigned char headersum, sum_calculated; if ((p = __archive_read_ahead(a, H0_FIXED_SIZE, NULL)) == NULL) return (truncated_error(a)); lha->header_size = p[H0_HEADER_SIZE_OFFSET] + 2; headersum = p[H0_HEADER_SUM_OFFSET]; lha->compsize = archive_le32dec(p + H0_COMP_SIZE_OFFSET); lha->origsize = archive_le32dec(p + H0_ORIG_SIZE_OFFSET); lha->mtime = lha_dos_time(p + H0_DOS_TIME_OFFSET); namelen = p[H0_NAME_LEN_OFFSET]; extdsize = (int)lha->header_size - H0_FIXED_SIZE - namelen; if ((namelen > 221 || extdsize < 0) && extdsize != -2) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid LHa header"); return (ARCHIVE_FATAL); } if ((p = __archive_read_ahead(a, lha->header_size, NULL)) == NULL) return (truncated_error(a)); archive_strncpy(&lha->filename, p + H0_FILE_NAME_OFFSET, namelen); /* When extdsize == -2, A CRC16 value is not present in the header. */ if (extdsize >= 0) { lha->crc = archive_le16dec(p + H0_FILE_NAME_OFFSET + namelen); lha->setflag |= CRC_IS_SET; } sum_calculated = lha_calcsum(0, p, 2, lha->header_size - 2); /* Read an extended header */ if (extdsize > 0) { /* This extended data is set by 'LHa for UNIX' only. * Maybe fixed size. */ p += H0_FILE_NAME_OFFSET + namelen + 2; if (p[0] == 'U' && extdsize == 12) { /* p[1] is a minor version. */ lha->mtime = archive_le32dec(&p[2]); lha->mode = archive_le16dec(&p[6]); lha->uid = archive_le16dec(&p[8]); lha->gid = archive_le16dec(&p[10]); lha->setflag |= UNIX_MODE_IS_SET; } } __archive_read_consume(a, lha->header_size); if (sum_calculated != headersum) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "LHa header sum error"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } /* * Header 1 format * * +0 +1 +2 +7 +11 * +---------------+----------+----------------+-------------+ * |header size(*1)|header sum|compression type|skip size(*2)| * +---------------+----------+----------------+-------------+ * <---------------(*1)----------* * * +11 +15 +17 +19 +20 +21 * +-----------------+---------+---------+--------------+----------------+ * |uncompressed size|time(DOS)|date(DOS)|attribute(DOS)|header level(=1)| * +-----------------+---------+---------+--------------+----------------+ * *-------------------------------(*1)----------------------------------* * * +21 +22 +22+(*3) +22+(*3)+2 +22+(*3)+3 +22+(*3)+3+(*4) * +---------------+---------+----------+-----------+-----------+ * |name length(*3)|file name|file CRC16| creator |padding(*4)| * +---------------+---------+----------+-----------+-----------+ * <--(*3)-> * *----------------------------(*1)----------------------------* * * +22+(*3)+3+(*4) +22+(*3)+3+(*4)+2 +22+(*3)+3+(*4)+2+(*5) * +----------------+---------------------+------------------------+ * |next header size| extended header(*5) | compressed data | * +----------------+---------------------+------------------------+ * *------(*1)-----> <--------------------(*2)--------------------> */ #define H1_HEADER_SIZE_OFFSET 0 #define H1_HEADER_SUM_OFFSET 1 #define H1_COMP_SIZE_OFFSET 7 #define H1_ORIG_SIZE_OFFSET 11 #define H1_DOS_TIME_OFFSET 15 #define H1_NAME_LEN_OFFSET 21 #define H1_FILE_NAME_OFFSET 22 #define H1_FIXED_SIZE 27 static int lha_read_file_header_1(struct archive_read *a, struct lha *lha) { const unsigned char *p; size_t extdsize; int i, err, err2; int namelen, padding; unsigned char headersum, sum_calculated; err = ARCHIVE_OK; if ((p = __archive_read_ahead(a, H1_FIXED_SIZE, NULL)) == NULL) return (truncated_error(a)); lha->header_size = p[H1_HEADER_SIZE_OFFSET] + 2; headersum = p[H1_HEADER_SUM_OFFSET]; /* Note: An extended header size is included in a compsize. */ lha->compsize = archive_le32dec(p + H1_COMP_SIZE_OFFSET); lha->origsize = archive_le32dec(p + H1_ORIG_SIZE_OFFSET); lha->mtime = lha_dos_time(p + H1_DOS_TIME_OFFSET); namelen = p[H1_NAME_LEN_OFFSET]; /* Calculate a padding size. The result will be normally 0 only(?) */ padding = ((int)lha->header_size) - H1_FIXED_SIZE - namelen; if (namelen > 230 || padding < 0) goto invalid; if ((p = __archive_read_ahead(a, lha->header_size, NULL)) == NULL) return (truncated_error(a)); for (i = 0; i < namelen; i++) { if (p[i + H1_FILE_NAME_OFFSET] == 0xff) goto invalid;/* Invalid filename. */ } archive_strncpy(&lha->filename, p + H1_FILE_NAME_OFFSET, namelen); lha->crc = archive_le16dec(p + H1_FILE_NAME_OFFSET + namelen); lha->setflag |= CRC_IS_SET; sum_calculated = lha_calcsum(0, p, 2, lha->header_size - 2); /* Consume used bytes but not include `next header size' data * since it will be consumed in lha_read_file_extended_header(). */ __archive_read_consume(a, lha->header_size - 2); /* Read extended headers */ err2 = lha_read_file_extended_header(a, lha, NULL, 2, (size_t)(lha->compsize + 2), &extdsize); if (err2 < ARCHIVE_WARN) return (err2); if (err2 < err) err = err2; /* Get a real compressed file size. */ lha->compsize -= extdsize - 2; if (sum_calculated != headersum) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "LHa header sum error"); return (ARCHIVE_FATAL); } return (err); invalid: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid LHa header"); return (ARCHIVE_FATAL); } /* * Header 2 format * * +0 +2 +7 +11 +15 * +---------------+----------------+-------------------+-----------------+ * |header size(*1)|compression type|compressed size(*2)|uncompressed size| * +---------------+----------------+-------------------+-----------------+ * <--------------------------------(*1)---------------------------------* * * +15 +19 +20 +21 +23 +24 * +-----------------+------------+----------------+----------+-----------+ * |data/time(time_t)| 0x20 fixed |header level(=2)|file CRC16| creator | * +-----------------+------------+----------------+----------+-----------+ * *---------------------------------(*1)---------------------------------* * * +24 +26 +26+(*3) +26+(*3)+(*4) * +----------------+-------------------+-------------+-------------------+ * |next header size|extended header(*3)| padding(*4) | compressed data | * +----------------+-------------------+-------------+-------------------+ * *--------------------------(*1)-------------------> <------(*2)-------> * */ #define H2_HEADER_SIZE_OFFSET 0 #define H2_COMP_SIZE_OFFSET 7 #define H2_ORIG_SIZE_OFFSET 11 #define H2_TIME_OFFSET 15 #define H2_CRC_OFFSET 21 #define H2_FIXED_SIZE 24 static int lha_read_file_header_2(struct archive_read *a, struct lha *lha) { const unsigned char *p; size_t extdsize; int err, padding; uint16_t header_crc; if ((p = __archive_read_ahead(a, H2_FIXED_SIZE, NULL)) == NULL) return (truncated_error(a)); lha->header_size =archive_le16dec(p + H2_HEADER_SIZE_OFFSET); lha->compsize = archive_le32dec(p + H2_COMP_SIZE_OFFSET); lha->origsize = archive_le32dec(p + H2_ORIG_SIZE_OFFSET); lha->mtime = archive_le32dec(p + H2_TIME_OFFSET); lha->crc = archive_le16dec(p + H2_CRC_OFFSET); lha->setflag |= CRC_IS_SET; if (lha->header_size < H2_FIXED_SIZE) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid LHa header size"); return (ARCHIVE_FATAL); } header_crc = lha_crc16(0, p, H2_FIXED_SIZE); __archive_read_consume(a, H2_FIXED_SIZE); /* Read extended headers */ err = lha_read_file_extended_header(a, lha, &header_crc, 2, lha->header_size - H2_FIXED_SIZE, &extdsize); if (err < ARCHIVE_WARN) return (err); /* Calculate a padding size. The result will be normally 0 or 1. */ padding = (int)lha->header_size - (int)(H2_FIXED_SIZE + extdsize); if (padding > 0) { if ((p = __archive_read_ahead(a, padding, NULL)) == NULL) return (truncated_error(a)); header_crc = lha_crc16(header_crc, p, padding); __archive_read_consume(a, padding); } if (header_crc != lha->header_crc) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "LHa header CRC error"); return (ARCHIVE_FATAL); } return (err); } /* * Header 3 format * * +0 +2 +7 +11 +15 * +------------+----------------+-------------------+-----------------+ * | 0x04 fixed |compression type|compressed size(*2)|uncompressed size| * +------------+----------------+-------------------+-----------------+ * <-------------------------------(*1)-------------------------------* * * +15 +19 +20 +21 +23 +24 * +-----------------+------------+----------------+----------+-----------+ * |date/time(time_t)| 0x20 fixed |header level(=3)|file CRC16| creator | * +-----------------+------------+----------------+----------+-----------+ * *--------------------------------(*1)----------------------------------* * * +24 +28 +32 +32+(*3) * +---------------+----------------+-------------------+-----------------+ * |header size(*1)|next header size|extended header(*3)| compressed data | * +---------------+----------------+-------------------+-----------------+ * *------------------------(*1)-----------------------> <------(*2)-----> * */ #define H3_FIELD_LEN_OFFSET 0 #define H3_COMP_SIZE_OFFSET 7 #define H3_ORIG_SIZE_OFFSET 11 #define H3_TIME_OFFSET 15 #define H3_CRC_OFFSET 21 #define H3_HEADER_SIZE_OFFSET 24 #define H3_FIXED_SIZE 28 static int lha_read_file_header_3(struct archive_read *a, struct lha *lha) { const unsigned char *p; size_t extdsize; int err; uint16_t header_crc; if ((p = __archive_read_ahead(a, H3_FIXED_SIZE, NULL)) == NULL) return (truncated_error(a)); if (archive_le16dec(p + H3_FIELD_LEN_OFFSET) != 4) goto invalid; lha->header_size =archive_le32dec(p + H3_HEADER_SIZE_OFFSET); lha->compsize = archive_le32dec(p + H3_COMP_SIZE_OFFSET); lha->origsize = archive_le32dec(p + H3_ORIG_SIZE_OFFSET); lha->mtime = archive_le32dec(p + H3_TIME_OFFSET); lha->crc = archive_le16dec(p + H3_CRC_OFFSET); lha->setflag |= CRC_IS_SET; if (lha->header_size < H3_FIXED_SIZE + 4) goto invalid; header_crc = lha_crc16(0, p, H3_FIXED_SIZE); __archive_read_consume(a, H3_FIXED_SIZE); /* Read extended headers */ err = lha_read_file_extended_header(a, lha, &header_crc, 4, lha->header_size - H3_FIXED_SIZE, &extdsize); if (err < ARCHIVE_WARN) return (err); if (header_crc != lha->header_crc) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "LHa header CRC error"); return (ARCHIVE_FATAL); } return (err); invalid: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid LHa header"); return (ARCHIVE_FATAL); } /* * Extended header format * * +0 +2 +3 -- used in header 1 and 2 * +0 +4 +5 -- used in header 3 * +--------------+---------+-------------------+--------------+-- * |ex-header size|header id| data |ex-header size| ....... * +--------------+---------+-------------------+--------------+-- * <-------------( ex-header size)------------> <-- next extended header --* * * If the ex-header size is zero, it is the make of the end of extended * headers. * */ static int lha_read_file_extended_header(struct archive_read *a, struct lha *lha, uint16_t *crc, int sizefield_length, size_t limitsize, size_t *total_size) { const void *h; const unsigned char *extdheader; size_t extdsize; size_t datasize; unsigned int i; unsigned char extdtype; #define EXT_HEADER_CRC 0x00 /* Header CRC and information*/ #define EXT_FILENAME 0x01 /* Filename */ #define EXT_DIRECTORY 0x02 /* Directory name */ #define EXT_DOS_ATTR 0x40 /* MS-DOS attribute */ #define EXT_TIMESTAMP 0x41 /* Windows time stamp */ #define EXT_FILESIZE 0x42 /* Large file size */ #define EXT_TIMEZONE 0x43 /* Time zone */ #define EXT_UTF16_FILENAME 0x44 /* UTF-16 filename */ #define EXT_UTF16_DIRECTORY 0x45 /* UTF-16 directory name */ #define EXT_CODEPAGE 0x46 /* Codepage */ #define EXT_UNIX_MODE 0x50 /* File permission */ #define EXT_UNIX_GID_UID 0x51 /* gid,uid */ #define EXT_UNIX_GNAME 0x52 /* Group name */ #define EXT_UNIX_UNAME 0x53 /* User name */ #define EXT_UNIX_MTIME 0x54 /* Modified time */ #define EXT_OS2_NEW_ATTR 0x7f /* new attribute(OS/2 only) */ #define EXT_NEW_ATTR 0xff /* new attribute */ *total_size = sizefield_length; for (;;) { /* Read an extended header size. */ if ((h = __archive_read_ahead(a, sizefield_length, NULL)) == NULL) return (truncated_error(a)); /* Check if the size is the zero indicates the end of the * extended header. */ if (sizefield_length == sizeof(uint16_t)) extdsize = archive_le16dec(h); else extdsize = archive_le32dec(h); if (extdsize == 0) { /* End of extended header */ if (crc != NULL) *crc = lha_crc16(*crc, h, sizefield_length); __archive_read_consume(a, sizefield_length); return (ARCHIVE_OK); } /* Sanity check to the extended header size. */ if (((uint64_t)*total_size + extdsize) > (uint64_t)limitsize || extdsize <= (size_t)sizefield_length) goto invalid; /* Read the extended header. */ if ((h = __archive_read_ahead(a, extdsize, NULL)) == NULL) return (truncated_error(a)); *total_size += extdsize; extdheader = (const unsigned char *)h; /* Get the extended header type. */ extdtype = extdheader[sizefield_length]; /* Calculate an extended data size. */ datasize = extdsize - (1 + sizefield_length); /* Skip an extended header size field and type field. */ extdheader += sizefield_length + 1; if (crc != NULL && extdtype != EXT_HEADER_CRC) *crc = lha_crc16(*crc, h, extdsize); switch (extdtype) { case EXT_HEADER_CRC: /* We only use a header CRC. Following data will not * be used. */ if (datasize >= 2) { lha->header_crc = archive_le16dec(extdheader); if (crc != NULL) { static const char zeros[2] = {0, 0}; *crc = lha_crc16(*crc, h, extdsize - datasize); /* CRC value itself as zero */ *crc = lha_crc16(*crc, zeros, 2); *crc = lha_crc16(*crc, extdheader+2, datasize - 2); } } break; case EXT_FILENAME: if (datasize == 0) { /* maybe directory header */ archive_string_empty(&lha->filename); break; } if (extdheader[0] == '\0') goto invalid; archive_strncpy(&lha->filename, (const char *)extdheader, datasize); break; case EXT_DIRECTORY: if (datasize == 0 || extdheader[0] == '\0') /* no directory name data. exit this case. */ goto invalid; archive_strncpy(&lha->dirname, (const char *)extdheader, datasize); /* * Convert directory delimiter from 0xFF * to '/' for local system. */ for (i = 0; i < lha->dirname.length; i++) { if ((unsigned char)lha->dirname.s[i] == 0xFF) lha->dirname.s[i] = '/'; } /* Is last character directory separator? */ if (lha->dirname.s[lha->dirname.length-1] != '/') /* invalid directory data */ goto invalid; break; case EXT_DOS_ATTR: if (datasize == 2) lha->dos_attr = (unsigned char) (archive_le16dec(extdheader) & 0xff); break; case EXT_TIMESTAMP: if (datasize == (sizeof(uint64_t) * 3)) { lha->birthtime = lha_win_time( archive_le64dec(extdheader), &lha->birthtime_tv_nsec); extdheader += sizeof(uint64_t); lha->mtime = lha_win_time( archive_le64dec(extdheader), &lha->mtime_tv_nsec); extdheader += sizeof(uint64_t); lha->atime = lha_win_time( archive_le64dec(extdheader), &lha->atime_tv_nsec); lha->setflag |= BIRTHTIME_IS_SET | ATIME_IS_SET; } break; case EXT_FILESIZE: if (datasize == sizeof(uint64_t) * 2) { lha->compsize = archive_le64dec(extdheader); extdheader += sizeof(uint64_t); lha->origsize = archive_le64dec(extdheader); } break; case EXT_CODEPAGE: /* Get an archived filename charset from codepage. * This overwrites the charset specified by * hdrcharset option. */ if (datasize == sizeof(uint32_t)) { struct archive_string cp; const char *charset; archive_string_init(&cp); switch (archive_le32dec(extdheader)) { case 65001: /* UTF-8 */ charset = "UTF-8"; break; default: archive_string_sprintf(&cp, "CP%d", (int)archive_le32dec(extdheader)); charset = cp.s; break; } lha->sconv = archive_string_conversion_from_charset( &(a->archive), charset, 1); archive_string_free(&cp); if (lha->sconv == NULL) return (ARCHIVE_FATAL); } break; case EXT_UNIX_MODE: if (datasize == sizeof(uint16_t)) { lha->mode = archive_le16dec(extdheader); lha->setflag |= UNIX_MODE_IS_SET; } break; case EXT_UNIX_GID_UID: if (datasize == (sizeof(uint16_t) * 2)) { lha->gid = archive_le16dec(extdheader); lha->uid = archive_le16dec(extdheader+2); } break; case EXT_UNIX_GNAME: if (datasize > 0) archive_strncpy(&lha->gname, (const char *)extdheader, datasize); break; case EXT_UNIX_UNAME: if (datasize > 0) archive_strncpy(&lha->uname, (const char *)extdheader, datasize); break; case EXT_UNIX_MTIME: if (datasize == sizeof(uint32_t)) lha->mtime = archive_le32dec(extdheader); break; case EXT_OS2_NEW_ATTR: /* This extended header is OS/2 depend. */ if (datasize == 16) { lha->dos_attr = (unsigned char) (archive_le16dec(extdheader) & 0xff); lha->mode = archive_le16dec(extdheader+2); lha->gid = archive_le16dec(extdheader+4); lha->uid = archive_le16dec(extdheader+6); lha->birthtime = archive_le32dec(extdheader+8); lha->atime = archive_le32dec(extdheader+12); lha->setflag |= UNIX_MODE_IS_SET | BIRTHTIME_IS_SET | ATIME_IS_SET; } break; case EXT_NEW_ATTR: if (datasize == 20) { lha->mode = (mode_t)archive_le32dec(extdheader); lha->gid = archive_le32dec(extdheader+4); lha->uid = archive_le32dec(extdheader+8); lha->birthtime = archive_le32dec(extdheader+12); lha->atime = archive_le32dec(extdheader+16); lha->setflag |= UNIX_MODE_IS_SET | BIRTHTIME_IS_SET | ATIME_IS_SET; } break; case EXT_TIMEZONE: /* Not supported */ case EXT_UTF16_FILENAME: /* Not supported */ case EXT_UTF16_DIRECTORY: /* Not supported */ default: break; } __archive_read_consume(a, extdsize); } invalid: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid extended LHa header"); return (ARCHIVE_FATAL); } static int lha_end_of_entry(struct archive_read *a) { struct lha *lha = (struct lha *)(a->format->data); int r = ARCHIVE_EOF; if (!lha->end_of_entry_cleanup) { if ((lha->setflag & CRC_IS_SET) && lha->crc != lha->entry_crc_calculated) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "LHa data CRC error"); r = ARCHIVE_WARN; } /* End-of-entry cleanup done. */ lha->end_of_entry_cleanup = 1; } return (r); } static int archive_read_format_lha_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct lha *lha = (struct lha *)(a->format->data); int r; if (lha->entry_unconsumed) { /* Consume as much as the decompressor actually used. */ __archive_read_consume(a, lha->entry_unconsumed); lha->entry_unconsumed = 0; } if (lha->end_of_entry) { *offset = lha->entry_offset; *size = 0; *buff = NULL; return (lha_end_of_entry(a)); } if (lha->entry_is_compressed) r = lha_read_data_lzh(a, buff, size, offset); else /* No compression. */ r = lha_read_data_none(a, buff, size, offset); return (r); } /* * Read a file content in no compression. * * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets * lha->end_of_entry if it consumes all of the data. */ static int lha_read_data_none(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct lha *lha = (struct lha *)(a->format->data); ssize_t bytes_avail; if (lha->entry_bytes_remaining == 0) { *buff = NULL; *size = 0; *offset = lha->entry_offset; lha->end_of_entry = 1; return (ARCHIVE_OK); } /* * Note: '1' here is a performance optimization. * Recall that the decompression layer returns a count of * available bytes; asking for more than that forces the * decompressor to combine reads by copying data. */ *buff = __archive_read_ahead(a, 1, &bytes_avail); if (bytes_avail <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated LHa file data"); return (ARCHIVE_FATAL); } if (bytes_avail > lha->entry_bytes_remaining) bytes_avail = (ssize_t)lha->entry_bytes_remaining; lha->entry_crc_calculated = lha_crc16(lha->entry_crc_calculated, *buff, bytes_avail); *size = bytes_avail; *offset = lha->entry_offset; lha->entry_offset += bytes_avail; lha->entry_bytes_remaining -= bytes_avail; if (lha->entry_bytes_remaining == 0) lha->end_of_entry = 1; lha->entry_unconsumed = bytes_avail; return (ARCHIVE_OK); } /* * Read a file content in LZHUFF encoding. * * Returns ARCHIVE_OK if successful, returns ARCHIVE_WARN if compression is * unsupported, ARCHIVE_FATAL otherwise, sets lha->end_of_entry if it consumes * all of the data. */ static int lha_read_data_lzh(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct lha *lha = (struct lha *)(a->format->data); ssize_t bytes_avail; int r; /* If we haven't yet read any data, initialize the decompressor. */ if (!lha->decompress_init) { r = lzh_decode_init(&(lha->strm), lha->method); switch (r) { case ARCHIVE_OK: break; case ARCHIVE_FAILED: /* Unsupported compression. */ *buff = NULL; *size = 0; *offset = 0; archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported lzh compression method -%c%c%c-", lha->method[0], lha->method[1], lha->method[2]); /* We know compressed size; just skip it. */ archive_read_format_lha_read_data_skip(a); return (ARCHIVE_WARN); default: archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory " "for lzh decompression"); return (ARCHIVE_FATAL); } /* We've initialized decompression for this stream. */ lha->decompress_init = 1; lha->strm.avail_out = 0; lha->strm.total_out = 0; } /* * Note: '1' here is a performance optimization. * Recall that the decompression layer returns a count of * available bytes; asking for more than that forces the * decompressor to combine reads by copying data. */ lha->strm.next_in = __archive_read_ahead(a, 1, &bytes_avail); if (bytes_avail <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated LHa file body"); return (ARCHIVE_FATAL); } if (bytes_avail > lha->entry_bytes_remaining) bytes_avail = (ssize_t)lha->entry_bytes_remaining; lha->strm.avail_in = (int)bytes_avail; lha->strm.total_in = 0; lha->strm.avail_out = 0; r = lzh_decode(&(lha->strm), bytes_avail == lha->entry_bytes_remaining); switch (r) { case ARCHIVE_OK: break; case ARCHIVE_EOF: lha->end_of_entry = 1; break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Bad lzh data"); return (ARCHIVE_FAILED); } lha->entry_unconsumed = lha->strm.total_in; lha->entry_bytes_remaining -= lha->strm.total_in; if (lha->strm.avail_out) { *offset = lha->entry_offset; *size = lha->strm.avail_out; *buff = lha->strm.ref_ptr; lha->entry_crc_calculated = lha_crc16(lha->entry_crc_calculated, *buff, *size); lha->entry_offset += *size; } else { *offset = lha->entry_offset; *size = 0; *buff = NULL; if (lha->end_of_entry) return (lha_end_of_entry(a)); } return (ARCHIVE_OK); } /* * Skip a file content. */ static int archive_read_format_lha_read_data_skip(struct archive_read *a) { struct lha *lha; int64_t bytes_skipped; lha = (struct lha *)(a->format->data); if (lha->entry_unconsumed) { /* Consume as much as the decompressor actually used. */ __archive_read_consume(a, lha->entry_unconsumed); lha->entry_unconsumed = 0; } /* if we've already read to end of data, we're done. */ if (lha->end_of_entry_cleanup) return (ARCHIVE_OK); /* * If the length is at the beginning, we can skip the * compressed data much more quickly. */ bytes_skipped = __archive_read_consume(a, lha->entry_bytes_remaining); if (bytes_skipped < 0) return (ARCHIVE_FATAL); /* This entry is finished and done. */ lha->end_of_entry_cleanup = lha->end_of_entry = 1; return (ARCHIVE_OK); } static int archive_read_format_lha_cleanup(struct archive_read *a) { struct lha *lha = (struct lha *)(a->format->data); lzh_decode_free(&(lha->strm)); archive_string_free(&(lha->dirname)); archive_string_free(&(lha->filename)); archive_string_free(&(lha->uname)); archive_string_free(&(lha->gname)); archive_wstring_free(&(lha->ws)); free(lha); (a->format->data) = NULL; return (ARCHIVE_OK); } /* * 'LHa for UNIX' utility has archived a symbolic-link name after * a pathname with '|' character. * This function extracts the symbolic-link name from the pathname. * * example. * 1. a symbolic-name is 'aaa/bb/cc' * 2. a filename is 'xxx/bbb' * then a archived pathname is 'xxx/bbb|aaa/bb/cc' */ static int lha_parse_linkname(struct archive_string *linkname, struct archive_string *pathname) { char * linkptr; size_t symlen; linkptr = strchr(pathname->s, '|'); if (linkptr != NULL) { symlen = strlen(linkptr + 1); archive_strncpy(linkname, linkptr+1, symlen); *linkptr = 0; pathname->length = strlen(pathname->s); return (1); } return (0); } /* Convert an MSDOS-style date/time into Unix-style time. */ static time_t lha_dos_time(const unsigned char *p) { int msTime, msDate; struct tm ts; msTime = archive_le16dec(p); msDate = archive_le16dec(p+2); memset(&ts, 0, sizeof(ts)); ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ ts.tm_mday = msDate & 0x1f; /* Day of month. */ ts.tm_hour = (msTime >> 11) & 0x1f; ts.tm_min = (msTime >> 5) & 0x3f; ts.tm_sec = (msTime << 1) & 0x3e; ts.tm_isdst = -1; return (mktime(&ts)); } /* Convert an MS-Windows-style date/time into Unix-style time. */ static time_t lha_win_time(uint64_t wintime, long *ns) { #define EPOC_TIME ARCHIVE_LITERAL_ULL(116444736000000000) if (wintime >= EPOC_TIME) { wintime -= EPOC_TIME; /* 1970-01-01 00:00:00 (UTC) */ if (ns != NULL) *ns = (long)(wintime % 10000000) * 100; return (wintime / 10000000); } else { if (ns != NULL) *ns = 0; return (0); } } static unsigned char lha_calcsum(unsigned char sum, const void *pp, int offset, size_t size) { unsigned char const *p = (unsigned char const *)pp; p += offset; for (;size > 0; --size) sum += *p++; return (sum); } static uint16_t crc16tbl[2][256]; static void lha_crc16_init(void) { unsigned int i; static int crc16init = 0; if (crc16init) return; crc16init = 1; for (i = 0; i < 256; i++) { unsigned int j; uint16_t crc = (uint16_t)i; for (j = 8; j; j--) crc = (crc >> 1) ^ ((crc & 1) * 0xA001); crc16tbl[0][i] = crc; } for (i = 0; i < 256; i++) { crc16tbl[1][i] = (crc16tbl[0][i] >> 8) ^ crc16tbl[0][crc16tbl[0][i] & 0xff]; } } static uint16_t lha_crc16(uint16_t crc, const void *pp, size_t len) { const unsigned char *p = (const unsigned char *)pp; const uint16_t *buff; const union { uint32_t i; char c[4]; } u = { 0x01020304 }; if (len == 0) return crc; /* Process unaligned address. */ if (((uintptr_t)p) & (uintptr_t)0x1) { crc = (crc >> 8) ^ crc16tbl[0][(crc ^ *p++) & 0xff]; len--; } buff = (const uint16_t *)p; /* * Modern C compiler such as GCC does not unroll automatically yet * without unrolling pragma, and Clang is so. So we should * unroll this loop for its performance. */ for (;len >= 8; len -= 8) { /* This if statement expects compiler optimization will - * remove the stament which will not be executed. */ + * remove the statement which will not be executed. */ #undef bswap16 #if defined(_MSC_VER) && _MSC_VER >= 1400 /* Visual Studio */ # define bswap16(x) _byteswap_ushort(x) #elif defined(__GNUC__) && ((__GNUC__ == 4 && __GNUC_MINOR__ >= 8) || __GNUC__ > 4) /* GCC 4.8 and later has __builtin_bswap16() */ # define bswap16(x) __builtin_bswap16(x) #elif defined(__clang__) /* All clang versions have __builtin_bswap16() */ # define bswap16(x) __builtin_bswap16(x) #else # define bswap16(x) ((((x) >> 8) & 0xff) | ((x) << 8)) #endif #define CRC16W do { \ if(u.c[0] == 1) { /* Big endian */ \ crc ^= bswap16(*buff); buff++; \ } else \ crc ^= *buff++; \ crc = crc16tbl[1][crc & 0xff] ^ crc16tbl[0][crc >> 8];\ } while (0) CRC16W; CRC16W; CRC16W; CRC16W; #undef CRC16W #undef bswap16 } p = (const unsigned char *)buff; for (;len; len--) { crc = (crc >> 8) ^ crc16tbl[0][(crc ^ *p++) & 0xff]; } return crc; } /* * Initialize LZHUF decoder. * * Returns ARCHIVE_OK if initialization was successful. * Returns ARCHIVE_FAILED if method is unsupported. * Returns ARCHIVE_FATAL if initialization failed; memory allocation * error occurred. */ static int lzh_decode_init(struct lzh_stream *strm, const char *method) { struct lzh_dec *ds; int w_bits, w_size; if (strm->ds == NULL) { strm->ds = calloc(1, sizeof(*strm->ds)); if (strm->ds == NULL) return (ARCHIVE_FATAL); } ds = strm->ds; ds->error = ARCHIVE_FAILED; if (method == NULL || method[0] != 'l' || method[1] != 'h') return (ARCHIVE_FAILED); switch (method[2]) { case '5': w_bits = 13;/* 8KiB for window */ break; case '6': w_bits = 15;/* 32KiB for window */ break; case '7': w_bits = 16;/* 64KiB for window */ break; default: return (ARCHIVE_FAILED);/* Not supported. */ } ds->error = ARCHIVE_FATAL; /* Expand a window size up to 128 KiB for decompressing process * performance whatever its original window size is. */ ds->w_size = 1U << 17; ds->w_mask = ds->w_size -1; if (ds->w_buff == NULL) { ds->w_buff = malloc(ds->w_size); if (ds->w_buff == NULL) return (ARCHIVE_FATAL); } w_size = 1U << w_bits; memset(ds->w_buff + ds->w_size - w_size, 0x20, w_size); ds->w_pos = 0; ds->state = 0; ds->pos_pt_len_size = w_bits + 1; ds->pos_pt_len_bits = (w_bits == 15 || w_bits == 16)? 5: 4; ds->literal_pt_len_size = PT_BITLEN_SIZE; ds->literal_pt_len_bits = 5; ds->br.cache_buffer = 0; ds->br.cache_avail = 0; if (lzh_huffman_init(&(ds->lt), LT_BITLEN_SIZE, 16) != ARCHIVE_OK) return (ARCHIVE_FATAL); ds->lt.len_bits = 9; if (lzh_huffman_init(&(ds->pt), PT_BITLEN_SIZE, 16) != ARCHIVE_OK) return (ARCHIVE_FATAL); ds->error = 0; return (ARCHIVE_OK); } /* * Release LZHUF decoder. */ static void lzh_decode_free(struct lzh_stream *strm) { if (strm->ds == NULL) return; free(strm->ds->w_buff); lzh_huffman_free(&(strm->ds->lt)); lzh_huffman_free(&(strm->ds->pt)); free(strm->ds); strm->ds = NULL; } /* * Bit stream reader. */ /* Check that the cache buffer has enough bits. */ #define lzh_br_has(br, n) ((br)->cache_avail >= n) /* Get compressed data by bit. */ #define lzh_br_bits(br, n) \ (((uint16_t)((br)->cache_buffer >> \ ((br)->cache_avail - (n)))) & cache_masks[n]) #define lzh_br_bits_forced(br, n) \ (((uint16_t)((br)->cache_buffer << \ ((n) - (br)->cache_avail))) & cache_masks[n]) /* Read ahead to make sure the cache buffer has enough compressed data we * will use. * True : completed, there is enough data in the cache buffer. * False : we met that strm->next_in is empty, we have to get following * bytes. */ #define lzh_br_read_ahead_0(strm, br, n) \ (lzh_br_has(br, (n)) || lzh_br_fillup(strm, br)) /* True : the cache buffer has some bits as much as we need. * False : there are no enough bits in the cache buffer to be used, * we have to get following bytes if we could. */ #define lzh_br_read_ahead(strm, br, n) \ (lzh_br_read_ahead_0((strm), (br), (n)) || lzh_br_has((br), (n))) /* Notify how many bits we consumed. */ #define lzh_br_consume(br, n) ((br)->cache_avail -= (n)) #define lzh_br_unconsume(br, n) ((br)->cache_avail += (n)) static const uint16_t cache_masks[] = { 0x0000, 0x0001, 0x0003, 0x0007, 0x000F, 0x001F, 0x003F, 0x007F, 0x00FF, 0x01FF, 0x03FF, 0x07FF, 0x0FFF, 0x1FFF, 0x3FFF, 0x7FFF, 0xFFFF, 0xFFFF, 0xFFFF, 0xFFFF }; /* * Shift away used bits in the cache data and fill it up with following bits. * Call this when cache buffer does not have enough bits you need. * * Returns 1 if the cache buffer is full. * Returns 0 if the cache buffer is not full; input buffer is empty. */ static int lzh_br_fillup(struct lzh_stream *strm, struct lzh_br *br) { int n = CACHE_BITS - br->cache_avail; for (;;) { const int x = n >> 3; if (strm->avail_in >= x) { switch (x) { case 8: br->cache_buffer = ((uint64_t)strm->next_in[0]) << 56 | ((uint64_t)strm->next_in[1]) << 48 | ((uint64_t)strm->next_in[2]) << 40 | ((uint64_t)strm->next_in[3]) << 32 | ((uint32_t)strm->next_in[4]) << 24 | ((uint32_t)strm->next_in[5]) << 16 | ((uint32_t)strm->next_in[6]) << 8 | (uint32_t)strm->next_in[7]; strm->next_in += 8; strm->avail_in -= 8; br->cache_avail += 8 * 8; return (1); case 7: br->cache_buffer = (br->cache_buffer << 56) | ((uint64_t)strm->next_in[0]) << 48 | ((uint64_t)strm->next_in[1]) << 40 | ((uint64_t)strm->next_in[2]) << 32 | ((uint32_t)strm->next_in[3]) << 24 | ((uint32_t)strm->next_in[4]) << 16 | ((uint32_t)strm->next_in[5]) << 8 | (uint32_t)strm->next_in[6]; strm->next_in += 7; strm->avail_in -= 7; br->cache_avail += 7 * 8; return (1); case 6: br->cache_buffer = (br->cache_buffer << 48) | ((uint64_t)strm->next_in[0]) << 40 | ((uint64_t)strm->next_in[1]) << 32 | ((uint32_t)strm->next_in[2]) << 24 | ((uint32_t)strm->next_in[3]) << 16 | ((uint32_t)strm->next_in[4]) << 8 | (uint32_t)strm->next_in[5]; strm->next_in += 6; strm->avail_in -= 6; br->cache_avail += 6 * 8; return (1); case 0: /* We have enough compressed data in * the cache buffer.*/ return (1); default: break; } } if (strm->avail_in == 0) { /* There is not enough compressed data to fill up the * cache buffer. */ return (0); } br->cache_buffer = (br->cache_buffer << 8) | *strm->next_in++; strm->avail_in--; br->cache_avail += 8; n -= 8; } } /* * Decode LZHUF. * * 1. Returns ARCHIVE_OK if output buffer or input buffer are empty. * Please set available buffer and call this function again. * 2. Returns ARCHIVE_EOF if decompression has been completed. * 3. Returns ARCHIVE_FAILED if an error occurred; compressed data * is broken or you do not set 'last' flag properly. * 4. 'last' flag is very important, you must set 1 to the flag if there * is no input data. The lha compressed data format does not provide how * to know the compressed data is really finished. * Note: lha command utility check if the total size of output bytes is * reached the uncompressed size recorded in its header. it does not mind * that the decoding process is properly finished. * GNU ZIP can decompress another compressed file made by SCO LZH compress. * it handles EOF as null to fill read buffer with zero until the decoding * process meet 2 bytes of zeros at reading a size of a next chunk, so the * zeros are treated as the mark of the end of the data although the zeros * is dummy, not the file data. */ static int lzh_read_blocks(struct lzh_stream *, int); static int lzh_decode_blocks(struct lzh_stream *, int); #define ST_RD_BLOCK 0 #define ST_RD_PT_1 1 #define ST_RD_PT_2 2 #define ST_RD_PT_3 3 #define ST_RD_PT_4 4 #define ST_RD_LITERAL_1 5 #define ST_RD_LITERAL_2 6 #define ST_RD_LITERAL_3 7 #define ST_RD_POS_DATA_1 8 #define ST_GET_LITERAL 9 #define ST_GET_POS_1 10 #define ST_GET_POS_2 11 #define ST_COPY_DATA 12 static int lzh_decode(struct lzh_stream *strm, int last) { struct lzh_dec *ds = strm->ds; int avail_in; int r; if (ds->error) return (ds->error); avail_in = strm->avail_in; do { if (ds->state < ST_GET_LITERAL) r = lzh_read_blocks(strm, last); else r = lzh_decode_blocks(strm, last); } while (r == 100); strm->total_in += avail_in - strm->avail_in; return (r); } static void lzh_emit_window(struct lzh_stream *strm, size_t s) { strm->ref_ptr = strm->ds->w_buff; strm->avail_out = (int)s; strm->total_out += s; } static int lzh_read_blocks(struct lzh_stream *strm, int last) { struct lzh_dec *ds = strm->ds; struct lzh_br *br = &(ds->br); int c = 0, i; unsigned rbits; for (;;) { switch (ds->state) { case ST_RD_BLOCK: /* * Read a block number indicates how many blocks * we will handle. The block is composed of a * literal and a match, sometimes a literal only * in particular, there are no reference data at * the beginning of the decompression. */ if (!lzh_br_read_ahead_0(strm, br, 16)) { if (!last) /* We need following data. */ return (ARCHIVE_OK); if (lzh_br_has(br, 8)) { /* * It seems there are extra bits. * 1. Compressed data is broken. * 2. `last' flag does not properly * set. */ goto failed; } if (ds->w_pos > 0) { lzh_emit_window(strm, ds->w_pos); ds->w_pos = 0; return (ARCHIVE_OK); } /* End of compressed data; we have completely * handled all compressed data. */ return (ARCHIVE_EOF); } ds->blocks_avail = lzh_br_bits(br, 16); if (ds->blocks_avail == 0) goto failed; lzh_br_consume(br, 16); /* * Read a literal table compressed in huffman * coding. */ ds->pt.len_size = ds->literal_pt_len_size; ds->pt.len_bits = ds->literal_pt_len_bits; ds->reading_position = 0; /* FALL THROUGH */ case ST_RD_PT_1: /* Note: ST_RD_PT_1, ST_RD_PT_2 and ST_RD_PT_4 are * used in reading both a literal table and a * position table. */ if (!lzh_br_read_ahead(strm, br, ds->pt.len_bits)) { if (last) goto failed;/* Truncated data. */ ds->state = ST_RD_PT_1; return (ARCHIVE_OK); } ds->pt.len_avail = lzh_br_bits(br, ds->pt.len_bits); lzh_br_consume(br, ds->pt.len_bits); /* FALL THROUGH */ case ST_RD_PT_2: if (ds->pt.len_avail == 0) { /* There is no bitlen. */ if (!lzh_br_read_ahead(strm, br, ds->pt.len_bits)) { if (last) goto failed;/* Truncated data.*/ ds->state = ST_RD_PT_2; return (ARCHIVE_OK); } if (!lzh_make_fake_table(&(ds->pt), lzh_br_bits(br, ds->pt.len_bits))) goto failed;/* Invalid data. */ lzh_br_consume(br, ds->pt.len_bits); if (ds->reading_position) ds->state = ST_GET_LITERAL; else ds->state = ST_RD_LITERAL_1; break; } else if (ds->pt.len_avail > ds->pt.len_size) goto failed;/* Invalid data. */ ds->loop = 0; memset(ds->pt.freq, 0, sizeof(ds->pt.freq)); if (ds->pt.len_avail < 3 || ds->pt.len_size == ds->pos_pt_len_size) { ds->state = ST_RD_PT_4; break; } /* FALL THROUGH */ case ST_RD_PT_3: ds->loop = lzh_read_pt_bitlen(strm, ds->loop, 3); if (ds->loop < 3) { if (ds->loop < 0 || last) goto failed;/* Invalid data. */ /* Not completed, get following data. */ ds->state = ST_RD_PT_3; return (ARCHIVE_OK); } /* There are some null in bitlen of the literal. */ if (!lzh_br_read_ahead(strm, br, 2)) { if (last) goto failed;/* Truncated data. */ ds->state = ST_RD_PT_3; return (ARCHIVE_OK); } c = lzh_br_bits(br, 2); lzh_br_consume(br, 2); if (c > ds->pt.len_avail - 3) goto failed;/* Invalid data. */ for (i = 3; c-- > 0 ;) ds->pt.bitlen[i++] = 0; ds->loop = i; /* FALL THROUGH */ case ST_RD_PT_4: ds->loop = lzh_read_pt_bitlen(strm, ds->loop, ds->pt.len_avail); if (ds->loop < ds->pt.len_avail) { if (ds->loop < 0 || last) goto failed;/* Invalid data. */ /* Not completed, get following data. */ ds->state = ST_RD_PT_4; return (ARCHIVE_OK); } if (!lzh_make_huffman_table(&(ds->pt))) goto failed;/* Invalid data */ if (ds->reading_position) { ds->state = ST_GET_LITERAL; break; } /* FALL THROUGH */ case ST_RD_LITERAL_1: if (!lzh_br_read_ahead(strm, br, ds->lt.len_bits)) { if (last) goto failed;/* Truncated data. */ ds->state = ST_RD_LITERAL_1; return (ARCHIVE_OK); } ds->lt.len_avail = lzh_br_bits(br, ds->lt.len_bits); lzh_br_consume(br, ds->lt.len_bits); /* FALL THROUGH */ case ST_RD_LITERAL_2: if (ds->lt.len_avail == 0) { /* There is no bitlen. */ if (!lzh_br_read_ahead(strm, br, ds->lt.len_bits)) { if (last) goto failed;/* Truncated data.*/ ds->state = ST_RD_LITERAL_2; return (ARCHIVE_OK); } if (!lzh_make_fake_table(&(ds->lt), lzh_br_bits(br, ds->lt.len_bits))) goto failed;/* Invalid data */ lzh_br_consume(br, ds->lt.len_bits); ds->state = ST_RD_POS_DATA_1; break; } else if (ds->lt.len_avail > ds->lt.len_size) goto failed;/* Invalid data */ ds->loop = 0; memset(ds->lt.freq, 0, sizeof(ds->lt.freq)); /* FALL THROUGH */ case ST_RD_LITERAL_3: i = ds->loop; while (i < ds->lt.len_avail) { if (!lzh_br_read_ahead(strm, br, ds->pt.max_bits)) { if (last) goto failed;/* Truncated data.*/ ds->loop = i; ds->state = ST_RD_LITERAL_3; return (ARCHIVE_OK); } rbits = lzh_br_bits(br, ds->pt.max_bits); c = lzh_decode_huffman(&(ds->pt), rbits); if (c > 2) { /* Note: 'c' will never be more than * eighteen since it's limited by * PT_BITLEN_SIZE, which is being set * to ds->pt.len_size through * ds->literal_pt_len_size. */ lzh_br_consume(br, ds->pt.bitlen[c]); c -= 2; ds->lt.freq[c]++; ds->lt.bitlen[i++] = c; } else if (c == 0) { lzh_br_consume(br, ds->pt.bitlen[c]); ds->lt.bitlen[i++] = 0; } else { /* c == 1 or c == 2 */ int n = (c == 1)?4:9; if (!lzh_br_read_ahead(strm, br, ds->pt.bitlen[c] + n)) { if (last) /* Truncated data. */ goto failed; ds->loop = i; ds->state = ST_RD_LITERAL_3; return (ARCHIVE_OK); } lzh_br_consume(br, ds->pt.bitlen[c]); c = lzh_br_bits(br, n); lzh_br_consume(br, n); c += (n == 4)?3:20; if (i + c > ds->lt.len_avail) goto failed;/* Invalid data */ memset(&(ds->lt.bitlen[i]), 0, c); i += c; } } if (i > ds->lt.len_avail || !lzh_make_huffman_table(&(ds->lt))) goto failed;/* Invalid data */ /* FALL THROUGH */ case ST_RD_POS_DATA_1: /* * Read a position table compressed in huffman * coding. */ ds->pt.len_size = ds->pos_pt_len_size; ds->pt.len_bits = ds->pos_pt_len_bits; ds->reading_position = 1; ds->state = ST_RD_PT_1; break; case ST_GET_LITERAL: return (100); } } failed: return (ds->error = ARCHIVE_FAILED); } static int lzh_decode_blocks(struct lzh_stream *strm, int last) { struct lzh_dec *ds = strm->ds; struct lzh_br bre = ds->br; struct huffman *lt = &(ds->lt); struct huffman *pt = &(ds->pt); unsigned char *w_buff = ds->w_buff; unsigned char *lt_bitlen = lt->bitlen; unsigned char *pt_bitlen = pt->bitlen; int blocks_avail = ds->blocks_avail, c = 0; int copy_len = ds->copy_len, copy_pos = ds->copy_pos; int w_pos = ds->w_pos, w_mask = ds->w_mask, w_size = ds->w_size; int lt_max_bits = lt->max_bits, pt_max_bits = pt->max_bits; int state = ds->state; for (;;) { switch (state) { case ST_GET_LITERAL: for (;;) { if (blocks_avail == 0) { /* We have decoded all blocks. * Let's handle next blocks. */ ds->state = ST_RD_BLOCK; ds->br = bre; ds->blocks_avail = 0; ds->w_pos = w_pos; ds->copy_pos = 0; return (100); } /* lzh_br_read_ahead() always try to fill the * cache buffer up. In specific situation we * are close to the end of the data, the cache * buffer will not be full and thus we have to * determine if the cache buffer has some bits * as much as we need after lzh_br_read_ahead() * failed. */ if (!lzh_br_read_ahead(strm, &bre, lt_max_bits)) { if (!last) goto next_data; /* Remaining bits are less than * maximum bits(lt.max_bits) but maybe * it still remains as much as we need, * so we should try to use it with * dummy bits. */ c = lzh_decode_huffman(lt, lzh_br_bits_forced(&bre, lt_max_bits)); lzh_br_consume(&bre, lt_bitlen[c]); if (!lzh_br_has(&bre, 0)) goto failed;/* Over read. */ } else { c = lzh_decode_huffman(lt, lzh_br_bits(&bre, lt_max_bits)); lzh_br_consume(&bre, lt_bitlen[c]); } blocks_avail--; if (c > UCHAR_MAX) /* Current block is a match data. */ break; /* * 'c' is exactly a literal code. */ /* Save a decoded code to reference it * afterward. */ w_buff[w_pos] = c; if (++w_pos >= w_size) { w_pos = 0; lzh_emit_window(strm, w_size); goto next_data; } } /* 'c' is the length of a match pattern we have * already extracted, which has be stored in * window(ds->w_buff). */ copy_len = c - (UCHAR_MAX + 1) + MINMATCH; /* FALL THROUGH */ case ST_GET_POS_1: /* * Get a reference position. */ if (!lzh_br_read_ahead(strm, &bre, pt_max_bits)) { if (!last) { state = ST_GET_POS_1; ds->copy_len = copy_len; goto next_data; } copy_pos = lzh_decode_huffman(pt, lzh_br_bits_forced(&bre, pt_max_bits)); lzh_br_consume(&bre, pt_bitlen[copy_pos]); if (!lzh_br_has(&bre, 0)) goto failed;/* Over read. */ } else { copy_pos = lzh_decode_huffman(pt, lzh_br_bits(&bre, pt_max_bits)); lzh_br_consume(&bre, pt_bitlen[copy_pos]); } /* FALL THROUGH */ case ST_GET_POS_2: if (copy_pos > 1) { /* We need an additional adjustment number to * the position. */ int p = copy_pos - 1; if (!lzh_br_read_ahead(strm, &bre, p)) { if (last) goto failed;/* Truncated data.*/ state = ST_GET_POS_2; ds->copy_len = copy_len; ds->copy_pos = copy_pos; goto next_data; } copy_pos = (1 << p) + lzh_br_bits(&bre, p); lzh_br_consume(&bre, p); } /* The position is actually a distance from the last * code we had extracted and thus we have to convert * it to a position of the window. */ copy_pos = (w_pos - copy_pos - 1) & w_mask; /* FALL THROUGH */ case ST_COPY_DATA: /* * Copy `copy_len' bytes as extracted data from * the window into the output buffer. */ for (;;) { int l; l = copy_len; if (copy_pos > w_pos) { if (l > w_size - copy_pos) l = w_size - copy_pos; } else { if (l > w_size - w_pos) l = w_size - w_pos; } if ((copy_pos + l < w_pos) || (w_pos + l < copy_pos)) { /* No overlap. */ memcpy(w_buff + w_pos, w_buff + copy_pos, l); } else { const unsigned char *s; unsigned char *d; int li; d = w_buff + w_pos; s = w_buff + copy_pos; for (li = 0; li < l-1;) { d[li] = s[li];li++; d[li] = s[li];li++; } if (li < l) d[li] = s[li]; } w_pos += l; if (w_pos == w_size) { w_pos = 0; lzh_emit_window(strm, w_size); if (copy_len <= l) state = ST_GET_LITERAL; else { state = ST_COPY_DATA; ds->copy_len = copy_len - l; ds->copy_pos = (copy_pos + l) & w_mask; } goto next_data; } if (copy_len <= l) /* A copy of current pattern ended. */ break; copy_len -= l; copy_pos = (copy_pos + l) & w_mask; } state = ST_GET_LITERAL; break; } } failed: return (ds->error = ARCHIVE_FAILED); next_data: ds->br = bre; ds->blocks_avail = blocks_avail; ds->state = state; ds->w_pos = w_pos; return (ARCHIVE_OK); } static int lzh_huffman_init(struct huffman *hf, size_t len_size, int tbl_bits) { int bits; if (hf->bitlen == NULL) { hf->bitlen = malloc(len_size * sizeof(hf->bitlen[0])); if (hf->bitlen == NULL) return (ARCHIVE_FATAL); } if (hf->tbl == NULL) { if (tbl_bits < HTBL_BITS) bits = tbl_bits; else bits = HTBL_BITS; hf->tbl = malloc(((size_t)1 << bits) * sizeof(hf->tbl[0])); if (hf->tbl == NULL) return (ARCHIVE_FATAL); } if (hf->tree == NULL && tbl_bits > HTBL_BITS) { hf->tree_avail = 1 << (tbl_bits - HTBL_BITS + 4); hf->tree = malloc(hf->tree_avail * sizeof(hf->tree[0])); if (hf->tree == NULL) return (ARCHIVE_FATAL); } hf->len_size = (int)len_size; hf->tbl_bits = tbl_bits; return (ARCHIVE_OK); } static void lzh_huffman_free(struct huffman *hf) { free(hf->bitlen); free(hf->tbl); free(hf->tree); } static char bitlen_tbl[0x400] = { 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 9, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 10, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 11, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 12, 13, 13, 13, 13, 13, 13, 13, 13, 14, 14, 14, 14, 15, 15, 16, 0 }; static int lzh_read_pt_bitlen(struct lzh_stream *strm, int start, int end) { struct lzh_dec *ds = strm->ds; struct lzh_br *br = &(ds->br); int c, i; for (i = start; i < end; ) { /* * bit pattern the number we need * 000 -> 0 * 001 -> 1 * 010 -> 2 * ... * 110 -> 6 * 1110 -> 7 * 11110 -> 8 * ... * 1111111111110 -> 16 */ if (!lzh_br_read_ahead(strm, br, 3)) return (i); if ((c = lzh_br_bits(br, 3)) == 7) { if (!lzh_br_read_ahead(strm, br, 13)) return (i); c = bitlen_tbl[lzh_br_bits(br, 13) & 0x3FF]; if (c) lzh_br_consume(br, c - 3); else return (-1);/* Invalid data. */ } else lzh_br_consume(br, 3); ds->pt.bitlen[i++] = c; ds->pt.freq[c]++; } return (i); } static int lzh_make_fake_table(struct huffman *hf, uint16_t c) { if (c >= hf->len_size) return (0); hf->tbl[0] = c; hf->max_bits = 0; hf->shift_bits = 0; hf->bitlen[hf->tbl[0]] = 0; return (1); } /* * Make a huffman coding table. */ static int lzh_make_huffman_table(struct huffman *hf) { uint16_t *tbl; const unsigned char *bitlen; int bitptn[17], weight[17]; int i, maxbits = 0, ptn, tbl_size, w; int diffbits, len_avail; /* * Initialize bit patterns. */ ptn = 0; for (i = 1, w = 1 << 15; i <= 16; i++, w >>= 1) { bitptn[i] = ptn; weight[i] = w; if (hf->freq[i]) { ptn += hf->freq[i] * w; maxbits = i; } } if (ptn != 0x10000 || maxbits > hf->tbl_bits) return (0);/* Invalid */ hf->max_bits = maxbits; /* * Cut out extra bits which we won't house in the table. * This preparation reduces the same calculation in the for-loop * making the table. */ if (maxbits < 16) { int ebits = 16 - maxbits; for (i = 1; i <= maxbits; i++) { bitptn[i] >>= ebits; weight[i] >>= ebits; } } if (maxbits > HTBL_BITS) { unsigned htbl_max; uint16_t *p; diffbits = maxbits - HTBL_BITS; for (i = 1; i <= HTBL_BITS; i++) { bitptn[i] >>= diffbits; weight[i] >>= diffbits; } htbl_max = bitptn[HTBL_BITS] + weight[HTBL_BITS] * hf->freq[HTBL_BITS]; p = &(hf->tbl[htbl_max]); while (p < &hf->tbl[1U<shift_bits = diffbits; /* * Make the table. */ tbl_size = 1 << HTBL_BITS; tbl = hf->tbl; bitlen = hf->bitlen; len_avail = hf->len_avail; hf->tree_used = 0; for (i = 0; i < len_avail; i++) { uint16_t *p; int len, cnt; uint16_t bit; int extlen; struct htree_t *ht; if (bitlen[i] == 0) continue; /* Get a bit pattern */ len = bitlen[i]; ptn = bitptn[len]; cnt = weight[len]; if (len <= HTBL_BITS) { /* Calculate next bit pattern */ if ((bitptn[len] = ptn + cnt) > tbl_size) return (0);/* Invalid */ /* Update the table */ p = &(tbl[ptn]); if (cnt > 7) { uint16_t *pc; cnt -= 8; pc = &p[cnt]; pc[0] = (uint16_t)i; pc[1] = (uint16_t)i; pc[2] = (uint16_t)i; pc[3] = (uint16_t)i; pc[4] = (uint16_t)i; pc[5] = (uint16_t)i; pc[6] = (uint16_t)i; pc[7] = (uint16_t)i; if (cnt > 7) { cnt -= 8; memcpy(&p[cnt], pc, 8 * sizeof(uint16_t)); pc = &p[cnt]; while (cnt > 15) { cnt -= 16; memcpy(&p[cnt], pc, 16 * sizeof(uint16_t)); } } if (cnt) memcpy(p, pc, cnt * sizeof(uint16_t)); } else { while (cnt > 1) { p[--cnt] = (uint16_t)i; p[--cnt] = (uint16_t)i; } if (cnt) p[--cnt] = (uint16_t)i; } continue; } /* * A bit length is too big to be housed to a direct table, * so we use a tree model for its extra bits. */ bitptn[len] = ptn + cnt; bit = 1U << (diffbits -1); extlen = len - HTBL_BITS; p = &(tbl[ptn >> diffbits]); if (*p == 0) { *p = len_avail + hf->tree_used; ht = &(hf->tree[hf->tree_used++]); if (hf->tree_used > hf->tree_avail) return (0);/* Invalid */ ht->left = 0; ht->right = 0; } else { if (*p < len_avail || *p >= (len_avail + hf->tree_used)) return (0);/* Invalid */ ht = &(hf->tree[*p - len_avail]); } while (--extlen > 0) { if (ptn & bit) { if (ht->left < len_avail) { ht->left = len_avail + hf->tree_used; ht = &(hf->tree[hf->tree_used++]); if (hf->tree_used > hf->tree_avail) return (0);/* Invalid */ ht->left = 0; ht->right = 0; } else { ht = &(hf->tree[ht->left - len_avail]); } } else { if (ht->right < len_avail) { ht->right = len_avail + hf->tree_used; ht = &(hf->tree[hf->tree_used++]); if (hf->tree_used > hf->tree_avail) return (0);/* Invalid */ ht->left = 0; ht->right = 0; } else { ht = &(hf->tree[ht->right - len_avail]); } } bit >>= 1; } if (ptn & bit) { if (ht->left != 0) return (0);/* Invalid */ ht->left = (uint16_t)i; } else { if (ht->right != 0) return (0);/* Invalid */ ht->right = (uint16_t)i; } } return (1); } static int lzh_decode_huffman_tree(struct huffman *hf, unsigned rbits, int c) { struct htree_t *ht; int extlen; ht = hf->tree; extlen = hf->shift_bits; while (c >= hf->len_avail) { c -= hf->len_avail; if (extlen-- <= 0 || c >= hf->tree_used) return (0); if (rbits & (1U << extlen)) c = ht[c].left; else c = ht[c].right; } return (c); } static inline int lzh_decode_huffman(struct huffman *hf, unsigned rbits) { int c; /* * At first search an index table for a bit pattern. * If it fails, search a huffman tree for. */ c = hf->tbl[rbits >> hf->shift_bits]; if (c < hf->len_avail || hf->len_avail == 0) return (c); /* This bit pattern needs to be found out at a huffman tree. */ return (lzh_decode_huffman_tree(hf, rbits, c)); } Index: head/contrib/libarchive/libarchive/archive_read_support_format_rar.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_format_rar.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_read_support_format_rar.c (revision 311903) @@ -1,2953 +1,2953 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * Copyright (c) 2011 Andres Mejia * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" #ifdef HAVE_ERRNO_H #include #endif #include #include #ifdef HAVE_ZLIB_H #include /* crc32 */ #endif #include "archive.h" #ifndef HAVE_ZLIB_H #include "archive_crc32.h" #endif #include "archive_endian.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_ppmd7_private.h" #include "archive_private.h" #include "archive_read_private.h" /* RAR signature, also known as the mark header */ #define RAR_SIGNATURE "\x52\x61\x72\x21\x1A\x07\x00" /* Header types */ #define MARK_HEAD 0x72 #define MAIN_HEAD 0x73 #define FILE_HEAD 0x74 #define COMM_HEAD 0x75 #define AV_HEAD 0x76 #define SUB_HEAD 0x77 #define PROTECT_HEAD 0x78 #define SIGN_HEAD 0x79 #define NEWSUB_HEAD 0x7a #define ENDARC_HEAD 0x7b /* Main Header Flags */ #define MHD_VOLUME 0x0001 #define MHD_COMMENT 0x0002 #define MHD_LOCK 0x0004 #define MHD_SOLID 0x0008 #define MHD_NEWNUMBERING 0x0010 #define MHD_AV 0x0020 #define MHD_PROTECT 0x0040 #define MHD_PASSWORD 0x0080 #define MHD_FIRSTVOLUME 0x0100 #define MHD_ENCRYPTVER 0x0200 /* Flags common to all headers */ #define HD_MARKDELETION 0x4000 #define HD_ADD_SIZE_PRESENT 0x8000 /* File Header Flags */ #define FHD_SPLIT_BEFORE 0x0001 #define FHD_SPLIT_AFTER 0x0002 #define FHD_PASSWORD 0x0004 #define FHD_COMMENT 0x0008 #define FHD_SOLID 0x0010 #define FHD_LARGE 0x0100 #define FHD_UNICODE 0x0200 #define FHD_SALT 0x0400 #define FHD_VERSION 0x0800 #define FHD_EXTTIME 0x1000 #define FHD_EXTFLAGS 0x2000 /* File dictionary sizes */ #define DICTIONARY_SIZE_64 0x00 #define DICTIONARY_SIZE_128 0x20 #define DICTIONARY_SIZE_256 0x40 #define DICTIONARY_SIZE_512 0x60 #define DICTIONARY_SIZE_1024 0x80 #define DICTIONARY_SIZE_2048 0xA0 #define DICTIONARY_SIZE_4096 0xC0 #define FILE_IS_DIRECTORY 0xE0 #define DICTIONARY_MASK FILE_IS_DIRECTORY /* OS Flags */ #define OS_MSDOS 0 #define OS_OS2 1 #define OS_WIN32 2 #define OS_UNIX 3 #define OS_MAC_OS 4 #define OS_BEOS 5 /* Compression Methods */ #define COMPRESS_METHOD_STORE 0x30 /* LZSS */ #define COMPRESS_METHOD_FASTEST 0x31 #define COMPRESS_METHOD_FAST 0x32 #define COMPRESS_METHOD_NORMAL 0x33 /* PPMd Variant H */ #define COMPRESS_METHOD_GOOD 0x34 #define COMPRESS_METHOD_BEST 0x35 #define CRC_POLYNOMIAL 0xEDB88320 #define NS_UNIT 10000000 #define DICTIONARY_MAX_SIZE 0x400000 #define MAINCODE_SIZE 299 #define OFFSETCODE_SIZE 60 #define LOWOFFSETCODE_SIZE 17 #define LENGTHCODE_SIZE 28 #define HUFFMAN_TABLE_SIZE \ MAINCODE_SIZE + OFFSETCODE_SIZE + LOWOFFSETCODE_SIZE + LENGTHCODE_SIZE #define MAX_SYMBOL_LENGTH 0xF #define MAX_SYMBOLS 20 /* * Considering L1,L2 cache miss and a calling of write system-call, * the best size of the output buffer(uncompressed buffer) is 128K. * If the structure of extracting process is changed, this value * might be researched again. */ #define UNP_BUFFER_SIZE (128 * 1024) /* Define this here for non-Windows platforms */ #if !((defined(__WIN32__) || defined(_WIN32) || defined(__WIN32)) && !defined(__CYGWIN__)) #define FILE_ATTRIBUTE_DIRECTORY 0x10 #endif /* Fields common to all headers */ struct rar_header { char crc[2]; char type; char flags[2]; char size[2]; }; /* Fields common to all file headers */ struct rar_file_header { char pack_size[4]; char unp_size[4]; char host_os; char file_crc[4]; char file_time[4]; char unp_ver; char method; char name_size[2]; char file_attr[4]; }; struct huffman_tree_node { int branches[2]; }; struct huffman_table_entry { unsigned int length; int value; }; struct huffman_code { struct huffman_tree_node *tree; int numentries; int numallocatedentries; int minlength; int maxlength; int tablesize; struct huffman_table_entry *table; }; struct lzss { unsigned char *window; int mask; int64_t position; }; struct data_block_offsets { int64_t header_size; int64_t start_offset; int64_t end_offset; }; struct rar { /* Entries from main RAR header */ unsigned main_flags; unsigned long file_crc; char reserved1[2]; char reserved2[4]; char encryptver; /* File header entries */ char compression_method; unsigned file_flags; int64_t packed_size; int64_t unp_size; time_t mtime; long mnsec; mode_t mode; char *filename; char *filename_save; size_t filename_save_size; size_t filename_allocated; /* File header optional entries */ char salt[8]; time_t atime; long ansec; time_t ctime; long cnsec; time_t arctime; long arcnsec; /* Fields to help with tracking decompression of files. */ int64_t bytes_unconsumed; int64_t bytes_remaining; int64_t bytes_uncopied; int64_t offset; int64_t offset_outgoing; int64_t offset_seek; char valid; unsigned int unp_offset; unsigned int unp_buffer_size; unsigned char *unp_buffer; unsigned int dictionary_size; char start_new_block; char entry_eof; unsigned long crc_calculated; int found_first_header; char has_endarc_header; struct data_block_offsets *dbo; unsigned int cursor; unsigned int nodes; /* LZSS members */ struct huffman_code maincode; struct huffman_code offsetcode; struct huffman_code lowoffsetcode; struct huffman_code lengthcode; unsigned char lengthtable[HUFFMAN_TABLE_SIZE]; struct lzss lzss; char output_last_match; unsigned int lastlength; unsigned int lastoffset; unsigned int oldoffset[4]; unsigned int lastlowoffset; unsigned int numlowoffsetrepeats; int64_t filterstart; char start_new_table; /* PPMd Variant H members */ char ppmd_valid; char ppmd_eod; char is_ppmd_block; int ppmd_escape; CPpmd7 ppmd7_context; CPpmd7z_RangeDec range_dec; IByteIn bytein; /* * String conversion object. */ int init_default_conversion; struct archive_string_conv *sconv_default; struct archive_string_conv *opt_sconv; struct archive_string_conv *sconv_utf8; struct archive_string_conv *sconv_utf16be; /* * Bit stream reader. */ struct rar_br { #define CACHE_TYPE uint64_t #define CACHE_BITS (8 * sizeof(CACHE_TYPE)) /* Cache buffer. */ CACHE_TYPE cache_buffer; /* Indicates how many bits avail in cache_buffer. */ int cache_avail; ssize_t avail_in; const unsigned char *next_in; } br; /* * Custom field to denote that this archive contains encrypted entries */ int has_encrypted_entries; }; static int archive_read_support_format_rar_capabilities(struct archive_read *); static int archive_read_format_rar_has_encrypted_entries(struct archive_read *); static int archive_read_format_rar_bid(struct archive_read *, int); static int archive_read_format_rar_options(struct archive_read *, const char *, const char *); static int archive_read_format_rar_read_header(struct archive_read *, struct archive_entry *); static int archive_read_format_rar_read_data(struct archive_read *, const void **, size_t *, int64_t *); static int archive_read_format_rar_read_data_skip(struct archive_read *a); static int64_t archive_read_format_rar_seek_data(struct archive_read *, int64_t, int); static int archive_read_format_rar_cleanup(struct archive_read *); /* Support functions */ static int read_header(struct archive_read *, struct archive_entry *, char); static time_t get_time(int); static int read_exttime(const char *, struct rar *, const char *); static int read_symlink_stored(struct archive_read *, struct archive_entry *, struct archive_string_conv *); static int read_data_stored(struct archive_read *, const void **, size_t *, int64_t *); static int read_data_compressed(struct archive_read *, const void **, size_t *, int64_t *); static int rar_br_preparation(struct archive_read *, struct rar_br *); static int parse_codes(struct archive_read *); static void free_codes(struct archive_read *); static int read_next_symbol(struct archive_read *, struct huffman_code *); static int create_code(struct archive_read *, struct huffman_code *, unsigned char *, int, char); static int add_value(struct archive_read *, struct huffman_code *, int, int, int); static int new_node(struct huffman_code *); static int make_table(struct archive_read *, struct huffman_code *); static int make_table_recurse(struct archive_read *, struct huffman_code *, int, struct huffman_table_entry *, int, int); static int64_t expand(struct archive_read *, int64_t); static int copy_from_lzss_window(struct archive_read *, const void **, int64_t, int); static const void *rar_read_ahead(struct archive_read *, size_t, ssize_t *); /* * Bit stream reader. */ /* Check that the cache buffer has enough bits. */ #define rar_br_has(br, n) ((br)->cache_avail >= n) /* Get compressed data by bit. */ #define rar_br_bits(br, n) \ (((uint32_t)((br)->cache_buffer >> \ ((br)->cache_avail - (n)))) & cache_masks[n]) #define rar_br_bits_forced(br, n) \ (((uint32_t)((br)->cache_buffer << \ ((n) - (br)->cache_avail))) & cache_masks[n]) /* Read ahead to make sure the cache buffer has enough compressed data we * will use. * True : completed, there is enough data in the cache buffer. * False : there is no data in the stream. */ #define rar_br_read_ahead(a, br, n) \ ((rar_br_has(br, (n)) || rar_br_fillup(a, br)) || rar_br_has(br, (n))) /* Notify how many bits we consumed. */ #define rar_br_consume(br, n) ((br)->cache_avail -= (n)) #define rar_br_consume_unalined_bits(br) ((br)->cache_avail &= ~7) static const uint32_t cache_masks[] = { 0x00000000, 0x00000001, 0x00000003, 0x00000007, 0x0000000F, 0x0000001F, 0x0000003F, 0x0000007F, 0x000000FF, 0x000001FF, 0x000003FF, 0x000007FF, 0x00000FFF, 0x00001FFF, 0x00003FFF, 0x00007FFF, 0x0000FFFF, 0x0001FFFF, 0x0003FFFF, 0x0007FFFF, 0x000FFFFF, 0x001FFFFF, 0x003FFFFF, 0x007FFFFF, 0x00FFFFFF, 0x01FFFFFF, 0x03FFFFFF, 0x07FFFFFF, 0x0FFFFFFF, 0x1FFFFFFF, 0x3FFFFFFF, 0x7FFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF }; /* * Shift away used bits in the cache data and fill it up with following bits. * Call this when cache buffer does not have enough bits you need. * * Returns 1 if the cache buffer is full. * Returns 0 if the cache buffer is not full; input buffer is empty. */ static int rar_br_fillup(struct archive_read *a, struct rar_br *br) { struct rar *rar = (struct rar *)(a->format->data); int n = CACHE_BITS - br->cache_avail; for (;;) { switch (n >> 3) { case 8: if (br->avail_in >= 8) { br->cache_buffer = ((uint64_t)br->next_in[0]) << 56 | ((uint64_t)br->next_in[1]) << 48 | ((uint64_t)br->next_in[2]) << 40 | ((uint64_t)br->next_in[3]) << 32 | ((uint32_t)br->next_in[4]) << 24 | ((uint32_t)br->next_in[5]) << 16 | ((uint32_t)br->next_in[6]) << 8 | (uint32_t)br->next_in[7]; br->next_in += 8; br->avail_in -= 8; br->cache_avail += 8 * 8; rar->bytes_unconsumed += 8; rar->bytes_remaining -= 8; return (1); } break; case 7: if (br->avail_in >= 7) { br->cache_buffer = (br->cache_buffer << 56) | ((uint64_t)br->next_in[0]) << 48 | ((uint64_t)br->next_in[1]) << 40 | ((uint64_t)br->next_in[2]) << 32 | ((uint32_t)br->next_in[3]) << 24 | ((uint32_t)br->next_in[4]) << 16 | ((uint32_t)br->next_in[5]) << 8 | (uint32_t)br->next_in[6]; br->next_in += 7; br->avail_in -= 7; br->cache_avail += 7 * 8; rar->bytes_unconsumed += 7; rar->bytes_remaining -= 7; return (1); } break; case 6: if (br->avail_in >= 6) { br->cache_buffer = (br->cache_buffer << 48) | ((uint64_t)br->next_in[0]) << 40 | ((uint64_t)br->next_in[1]) << 32 | ((uint32_t)br->next_in[2]) << 24 | ((uint32_t)br->next_in[3]) << 16 | ((uint32_t)br->next_in[4]) << 8 | (uint32_t)br->next_in[5]; br->next_in += 6; br->avail_in -= 6; br->cache_avail += 6 * 8; rar->bytes_unconsumed += 6; rar->bytes_remaining -= 6; return (1); } break; case 0: /* We have enough compressed data in * the cache buffer.*/ return (1); default: break; } if (br->avail_in <= 0) { if (rar->bytes_unconsumed > 0) { /* Consume as much as the decompressor * actually used. */ __archive_read_consume(a, rar->bytes_unconsumed); rar->bytes_unconsumed = 0; } br->next_in = rar_read_ahead(a, 1, &(br->avail_in)); if (br->next_in == NULL) return (0); if (br->avail_in == 0) return (0); } br->cache_buffer = (br->cache_buffer << 8) | *br->next_in++; br->avail_in--; br->cache_avail += 8; n -= 8; rar->bytes_unconsumed++; rar->bytes_remaining--; } } static int rar_br_preparation(struct archive_read *a, struct rar_br *br) { struct rar *rar = (struct rar *)(a->format->data); if (rar->bytes_remaining > 0) { br->next_in = rar_read_ahead(a, 1, &(br->avail_in)); if (br->next_in == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); return (ARCHIVE_FATAL); } if (br->cache_avail == 0) (void)rar_br_fillup(a, br); } return (ARCHIVE_OK); } /* Find last bit set */ static inline int rar_fls(unsigned int word) { word |= (word >> 1); word |= (word >> 2); word |= (word >> 4); word |= (word >> 8); word |= (word >> 16); return word - (word >> 1); } /* LZSS functions */ static inline int64_t lzss_position(struct lzss *lzss) { return lzss->position; } static inline int lzss_mask(struct lzss *lzss) { return lzss->mask; } static inline int lzss_size(struct lzss *lzss) { return lzss->mask + 1; } static inline int lzss_offset_for_position(struct lzss *lzss, int64_t pos) { return (int)(pos & lzss->mask); } static inline unsigned char * lzss_pointer_for_position(struct lzss *lzss, int64_t pos) { return &lzss->window[lzss_offset_for_position(lzss, pos)]; } static inline int lzss_current_offset(struct lzss *lzss) { return lzss_offset_for_position(lzss, lzss->position); } static inline uint8_t * lzss_current_pointer(struct lzss *lzss) { return lzss_pointer_for_position(lzss, lzss->position); } static inline void lzss_emit_literal(struct rar *rar, uint8_t literal) { *lzss_current_pointer(&rar->lzss) = literal; rar->lzss.position++; } static inline void lzss_emit_match(struct rar *rar, int offset, int length) { int dstoffs = lzss_current_offset(&rar->lzss); int srcoffs = (dstoffs - offset) & lzss_mask(&rar->lzss); int l, li, remaining; unsigned char *d, *s; remaining = length; while (remaining > 0) { l = remaining; if (dstoffs > srcoffs) { if (l > lzss_size(&rar->lzss) - dstoffs) l = lzss_size(&rar->lzss) - dstoffs; } else { if (l > lzss_size(&rar->lzss) - srcoffs) l = lzss_size(&rar->lzss) - srcoffs; } d = &(rar->lzss.window[dstoffs]); s = &(rar->lzss.window[srcoffs]); if ((dstoffs + l < srcoffs) || (srcoffs + l < dstoffs)) memcpy(d, s, l); else { for (li = 0; li < l; li++) d[li] = s[li]; } remaining -= l; dstoffs = (dstoffs + l) & lzss_mask(&(rar->lzss)); srcoffs = (srcoffs + l) & lzss_mask(&(rar->lzss)); } rar->lzss.position += length; } static void * ppmd_alloc(void *p, size_t size) { (void)p; return malloc(size); } static void ppmd_free(void *p, void *address) { (void)p; free(address); } static ISzAlloc g_szalloc = { ppmd_alloc, ppmd_free }; static Byte ppmd_read(void *p) { struct archive_read *a = ((IByteIn*)p)->a; struct rar *rar = (struct rar *)(a->format->data); struct rar_br *br = &(rar->br); Byte b; if (!rar_br_read_ahead(a, br, 8)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); rar->valid = 0; return 0; } b = rar_br_bits(br, 8); rar_br_consume(br, 8); return b; } int archive_read_support_format_rar(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct rar *rar; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_rar"); rar = (struct rar *)calloc(sizeof(*rar), 1); if (rar == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate rar data"); return (ARCHIVE_FATAL); } /* * Until enough data has been read, we cannot tell about * any encrypted entries yet. */ rar->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; r = __archive_read_register_format(a, rar, "rar", archive_read_format_rar_bid, archive_read_format_rar_options, archive_read_format_rar_read_header, archive_read_format_rar_read_data, archive_read_format_rar_read_data_skip, archive_read_format_rar_seek_data, archive_read_format_rar_cleanup, archive_read_support_format_rar_capabilities, archive_read_format_rar_has_encrypted_entries); if (r != ARCHIVE_OK) free(rar); return (r); } static int archive_read_support_format_rar_capabilities(struct archive_read * a) { (void)a; /* UNUSED */ return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); } static int archive_read_format_rar_has_encrypted_entries(struct archive_read *_a) { if (_a && _a->format) { struct rar * rar = (struct rar *)_a->format->data; if (rar) { return rar->has_encrypted_entries; } } return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; } static int archive_read_format_rar_bid(struct archive_read *a, int best_bid) { const char *p; /* If there's already a bid > 30, we'll never win. */ if (best_bid > 30) return (-1); if ((p = __archive_read_ahead(a, 7, NULL)) == NULL) return (-1); if (memcmp(p, RAR_SIGNATURE, 7) == 0) return (30); if ((p[0] == 'M' && p[1] == 'Z') || memcmp(p, "\x7F\x45LF", 4) == 0) { /* This is a PE file */ ssize_t offset = 0x10000; ssize_t window = 4096; ssize_t bytes_avail; while (offset + window <= (1024 * 128)) { const char *buff = __archive_read_ahead(a, offset + window, &bytes_avail); if (buff == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < 0x40) return (0); continue; } p = buff + offset; while (p + 7 < buff + bytes_avail) { if (memcmp(p, RAR_SIGNATURE, 7) == 0) return (30); p += 0x10; } offset = p - buff; } } return (0); } static int skip_sfx(struct archive_read *a) { const void *h; const char *p, *q; size_t skip, total; ssize_t bytes, window; total = 0; window = 4096; while (total + window <= (1024 * 128)) { h = __archive_read_ahead(a, window, &bytes); if (h == NULL) { /* Remaining bytes are less than window. */ window >>= 1; if (window < 0x40) goto fatal; continue; } if (bytes < 0x40) goto fatal; p = h; q = p + bytes; /* * Scan ahead until we find something that looks * like the RAR header. */ while (p + 7 < q) { if (memcmp(p, RAR_SIGNATURE, 7) == 0) { skip = p - (const char *)h; __archive_read_consume(a, skip); return (ARCHIVE_OK); } p += 0x10; } skip = p - (const char *)h; __archive_read_consume(a, skip); total += skip; } fatal: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Couldn't find out RAR header"); return (ARCHIVE_FATAL); } static int archive_read_format_rar_options(struct archive_read *a, const char *key, const char *val) { struct rar *rar; int ret = ARCHIVE_FAILED; rar = (struct rar *)(a->format->data); if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "rar: hdrcharset option needs a character-set name"); else { rar->opt_sconv = archive_string_conversion_from_charset( &a->archive, val, 0); if (rar->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int archive_read_format_rar_read_header(struct archive_read *a, struct archive_entry *entry) { const void *h; const char *p; struct rar *rar; size_t skip; char head_type; int ret; unsigned flags; unsigned long crc32_expected; a->archive.archive_format = ARCHIVE_FORMAT_RAR; if (a->archive.archive_format_name == NULL) a->archive.archive_format_name = "RAR"; rar = (struct rar *)(a->format->data); /* * It should be sufficient to call archive_read_next_header() for * a reader to determine if an entry is encrypted or not. If the * encryption of an entry is only detectable when calling * archive_read_data(), so be it. We'll do the same check there * as well. */ if (rar->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { rar->has_encrypted_entries = 0; } /* RAR files can be generated without EOF headers, so return ARCHIVE_EOF if * this fails. */ if ((h = __archive_read_ahead(a, 7, NULL)) == NULL) return (ARCHIVE_EOF); p = h; if (rar->found_first_header == 0 && ((p[0] == 'M' && p[1] == 'Z') || memcmp(p, "\x7F\x45LF", 4) == 0)) { /* This is an executable ? Must be self-extracting... */ ret = skip_sfx(a); if (ret < ARCHIVE_WARN) return (ret); } rar->found_first_header = 1; while (1) { unsigned long crc32_val; if ((h = __archive_read_ahead(a, 7, NULL)) == NULL) return (ARCHIVE_FATAL); p = h; head_type = p[2]; switch(head_type) { case MARK_HEAD: if (memcmp(p, RAR_SIGNATURE, 7) != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid marker header"); return (ARCHIVE_FATAL); } __archive_read_consume(a, 7); break; case MAIN_HEAD: rar->main_flags = archive_le16dec(p + 3); skip = archive_le16dec(p + 5); if (skip < 7 + sizeof(rar->reserved1) + sizeof(rar->reserved2)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size"); return (ARCHIVE_FATAL); } if ((h = __archive_read_ahead(a, skip, NULL)) == NULL) return (ARCHIVE_FATAL); p = h; memcpy(rar->reserved1, p + 7, sizeof(rar->reserved1)); memcpy(rar->reserved2, p + 7 + sizeof(rar->reserved1), sizeof(rar->reserved2)); if (rar->main_flags & MHD_ENCRYPTVER) { if (skip < 7 + sizeof(rar->reserved1) + sizeof(rar->reserved2)+1) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size"); return (ARCHIVE_FATAL); } rar->encryptver = *(p + 7 + sizeof(rar->reserved1) + sizeof(rar->reserved2)); } - /* Main header is password encrytped, so we cannot read any + /* Main header is password encrypted, so we cannot read any file names or any other info about files from the header. */ if (rar->main_flags & MHD_PASSWORD) { archive_entry_set_is_metadata_encrypted(entry, 1); archive_entry_set_is_data_encrypted(entry, 1); rar->has_encrypted_entries = 1; archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "RAR encryption support unavailable."); return (ARCHIVE_FATAL); } crc32_val = crc32(0, (const unsigned char *)p + 2, (unsigned)skip - 2); if ((crc32_val & 0xffff) != archive_le16dec(p)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Header CRC error"); return (ARCHIVE_FATAL); } __archive_read_consume(a, skip); break; case FILE_HEAD: return read_header(a, entry, head_type); case COMM_HEAD: case AV_HEAD: case SUB_HEAD: case PROTECT_HEAD: case SIGN_HEAD: case ENDARC_HEAD: flags = archive_le16dec(p + 3); skip = archive_le16dec(p + 5); if (skip < 7) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size too small"); return (ARCHIVE_FATAL); } if (flags & HD_ADD_SIZE_PRESENT) { if (skip < 7 + 4) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size too small"); return (ARCHIVE_FATAL); } if ((h = __archive_read_ahead(a, skip, NULL)) == NULL) return (ARCHIVE_FATAL); p = h; skip += archive_le32dec(p + 7); } /* Skip over the 2-byte CRC at the beginning of the header. */ crc32_expected = archive_le16dec(p); __archive_read_consume(a, 2); skip -= 2; /* Skim the entire header and compute the CRC. */ crc32_val = 0; while (skip > 0) { size_t to_read = skip; ssize_t did_read; if (to_read > 32 * 1024) { to_read = 32 * 1024; } if ((h = __archive_read_ahead(a, to_read, &did_read)) == NULL) { return (ARCHIVE_FATAL); } p = h; crc32_val = crc32(crc32_val, (const unsigned char *)p, (unsigned)did_read); __archive_read_consume(a, did_read); skip -= did_read; } if ((crc32_val & 0xffff) != crc32_expected) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Header CRC error"); return (ARCHIVE_FATAL); } if (head_type == ENDARC_HEAD) return (ARCHIVE_EOF); break; case NEWSUB_HEAD: if ((ret = read_header(a, entry, head_type)) < ARCHIVE_WARN) return ret; break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Bad RAR file"); return (ARCHIVE_FATAL); } } } static int archive_read_format_rar_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct rar *rar = (struct rar *)(a->format->data); int ret; if (rar->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { rar->has_encrypted_entries = 0; } if (rar->bytes_unconsumed > 0) { /* Consume as much as the decompressor actually used. */ __archive_read_consume(a, rar->bytes_unconsumed); rar->bytes_unconsumed = 0; } *buff = NULL; if (rar->entry_eof || rar->offset_seek >= rar->unp_size) { *size = 0; *offset = rar->offset; if (*offset < rar->unp_size) *offset = rar->unp_size; return (ARCHIVE_EOF); } switch (rar->compression_method) { case COMPRESS_METHOD_STORE: ret = read_data_stored(a, buff, size, offset); break; case COMPRESS_METHOD_FASTEST: case COMPRESS_METHOD_FAST: case COMPRESS_METHOD_NORMAL: case COMPRESS_METHOD_GOOD: case COMPRESS_METHOD_BEST: ret = read_data_compressed(a, buff, size, offset); if (ret != ARCHIVE_OK && ret != ARCHIVE_WARN) __archive_ppmd7_functions.Ppmd7_Free(&rar->ppmd7_context, &g_szalloc); break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported compression method for RAR file."); ret = ARCHIVE_FATAL; break; } return (ret); } static int archive_read_format_rar_read_data_skip(struct archive_read *a) { struct rar *rar; int64_t bytes_skipped; int ret; rar = (struct rar *)(a->format->data); if (rar->bytes_unconsumed > 0) { /* Consume as much as the decompressor actually used. */ __archive_read_consume(a, rar->bytes_unconsumed); rar->bytes_unconsumed = 0; } if (rar->bytes_remaining > 0) { bytes_skipped = __archive_read_consume(a, rar->bytes_remaining); if (bytes_skipped < 0) return (ARCHIVE_FATAL); } /* Compressed data to skip must be read from each header in a multivolume * archive. */ if (rar->main_flags & MHD_VOLUME && rar->file_flags & FHD_SPLIT_AFTER) { ret = archive_read_format_rar_read_header(a, a->entry); if (ret == (ARCHIVE_EOF)) ret = archive_read_format_rar_read_header(a, a->entry); if (ret != (ARCHIVE_OK)) return ret; return archive_read_format_rar_read_data_skip(a); } return (ARCHIVE_OK); } static int64_t archive_read_format_rar_seek_data(struct archive_read *a, int64_t offset, int whence) { int64_t client_offset, ret; unsigned int i; struct rar *rar = (struct rar *)(a->format->data); if (rar->compression_method == COMPRESS_METHOD_STORE) { /* Modify the offset for use with SEEK_SET */ switch (whence) { case SEEK_CUR: client_offset = rar->offset_seek; break; case SEEK_END: client_offset = rar->unp_size; break; case SEEK_SET: default: client_offset = 0; } client_offset += offset; if (client_offset < 0) { /* Can't seek past beginning of data block */ return -1; } else if (client_offset > rar->unp_size) { /* * Set the returned offset but only seek to the end of * the data block. */ rar->offset_seek = client_offset; client_offset = rar->unp_size; } client_offset += rar->dbo[0].start_offset; i = 0; while (i < rar->cursor) { i++; client_offset += rar->dbo[i].start_offset - rar->dbo[i-1].end_offset; } if (rar->main_flags & MHD_VOLUME) { /* Find the appropriate offset among the multivolume archive */ while (1) { if (client_offset < rar->dbo[rar->cursor].start_offset && rar->file_flags & FHD_SPLIT_BEFORE) { /* Search backwards for the correct data block */ if (rar->cursor == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Attempt to seek past beginning of RAR data block"); return (ARCHIVE_FAILED); } rar->cursor--; client_offset -= rar->dbo[rar->cursor+1].start_offset - rar->dbo[rar->cursor].end_offset; if (client_offset < rar->dbo[rar->cursor].start_offset) continue; ret = __archive_read_seek(a, rar->dbo[rar->cursor].start_offset - rar->dbo[rar->cursor].header_size, SEEK_SET); if (ret < (ARCHIVE_OK)) return ret; ret = archive_read_format_rar_read_header(a, a->entry); if (ret != (ARCHIVE_OK)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Error during seek of RAR file"); return (ARCHIVE_FAILED); } rar->cursor--; break; } else if (client_offset > rar->dbo[rar->cursor].end_offset && rar->file_flags & FHD_SPLIT_AFTER) { /* Search forward for the correct data block */ rar->cursor++; if (rar->cursor < rar->nodes && client_offset > rar->dbo[rar->cursor].end_offset) { client_offset += rar->dbo[rar->cursor].start_offset - rar->dbo[rar->cursor-1].end_offset; continue; } rar->cursor--; ret = __archive_read_seek(a, rar->dbo[rar->cursor].end_offset, SEEK_SET); if (ret < (ARCHIVE_OK)) return ret; ret = archive_read_format_rar_read_header(a, a->entry); if (ret == (ARCHIVE_EOF)) { rar->has_endarc_header = 1; ret = archive_read_format_rar_read_header(a, a->entry); } if (ret != (ARCHIVE_OK)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Error during seek of RAR file"); return (ARCHIVE_FAILED); } client_offset += rar->dbo[rar->cursor].start_offset - rar->dbo[rar->cursor-1].end_offset; continue; } break; } } ret = __archive_read_seek(a, client_offset, SEEK_SET); if (ret < (ARCHIVE_OK)) return ret; rar->bytes_remaining = rar->dbo[rar->cursor].end_offset - ret; i = rar->cursor; while (i > 0) { i--; ret -= rar->dbo[i+1].start_offset - rar->dbo[i].end_offset; } ret -= rar->dbo[0].start_offset; /* Always restart reading the file after a seek */ __archive_reset_read_data(&a->archive); rar->bytes_unconsumed = 0; rar->offset = 0; /* * If a seek past the end of file was requested, return the requested * offset. */ if (ret == rar->unp_size && rar->offset_seek > rar->unp_size) return rar->offset_seek; /* Return the new offset */ rar->offset_seek = ret; return rar->offset_seek; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Seeking of compressed RAR files is unsupported"); } return (ARCHIVE_FAILED); } static int archive_read_format_rar_cleanup(struct archive_read *a) { struct rar *rar; rar = (struct rar *)(a->format->data); free_codes(a); free(rar->filename); free(rar->filename_save); free(rar->dbo); free(rar->unp_buffer); free(rar->lzss.window); __archive_ppmd7_functions.Ppmd7_Free(&rar->ppmd7_context, &g_szalloc); free(rar); (a->format->data) = NULL; return (ARCHIVE_OK); } static int read_header(struct archive_read *a, struct archive_entry *entry, char head_type) { const void *h; const char *p, *endp; struct rar *rar; struct rar_header rar_header; struct rar_file_header file_header; int64_t header_size; unsigned filename_size, end; char *filename; char *strp; char packed_size[8]; char unp_size[8]; int ttime; struct archive_string_conv *sconv, *fn_sconv; unsigned long crc32_val; int ret = (ARCHIVE_OK), ret2; rar = (struct rar *)(a->format->data); /* Setup a string conversion object for non-rar-unicode filenames. */ sconv = rar->opt_sconv; if (sconv == NULL) { if (!rar->init_default_conversion) { rar->sconv_default = archive_string_default_conversion_for_read( &(a->archive)); rar->init_default_conversion = 1; } sconv = rar->sconv_default; } if ((h = __archive_read_ahead(a, 7, NULL)) == NULL) return (ARCHIVE_FATAL); p = h; memcpy(&rar_header, p, sizeof(rar_header)); rar->file_flags = archive_le16dec(rar_header.flags); header_size = archive_le16dec(rar_header.size); if (header_size < (int64_t)sizeof(file_header) + 7) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size"); return (ARCHIVE_FATAL); } crc32_val = crc32(0, (const unsigned char *)p + 2, 7 - 2); __archive_read_consume(a, 7); if (!(rar->file_flags & FHD_SOLID)) { rar->compression_method = 0; rar->packed_size = 0; rar->unp_size = 0; rar->mtime = 0; rar->ctime = 0; rar->atime = 0; rar->arctime = 0; rar->mode = 0; memset(&rar->salt, 0, sizeof(rar->salt)); rar->atime = 0; rar->ansec = 0; rar->ctime = 0; rar->cnsec = 0; rar->mtime = 0; rar->mnsec = 0; rar->arctime = 0; rar->arcnsec = 0; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "RAR solid archive support unavailable."); return (ARCHIVE_FATAL); } if ((h = __archive_read_ahead(a, (size_t)header_size - 7, NULL)) == NULL) return (ARCHIVE_FATAL); /* File Header CRC check. */ crc32_val = crc32(crc32_val, h, (unsigned)(header_size - 7)); if ((crc32_val & 0xffff) != archive_le16dec(rar_header.crc)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Header CRC error"); return (ARCHIVE_FATAL); } /* If no CRC error, Go on parsing File Header. */ p = h; endp = p + header_size - 7; memcpy(&file_header, p, sizeof(file_header)); p += sizeof(file_header); rar->compression_method = file_header.method; ttime = archive_le32dec(file_header.file_time); rar->mtime = get_time(ttime); rar->file_crc = archive_le32dec(file_header.file_crc); if (rar->file_flags & FHD_PASSWORD) { archive_entry_set_is_data_encrypted(entry, 1); rar->has_encrypted_entries = 1; archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "RAR encryption support unavailable."); /* Since it is only the data part itself that is encrypted we can at least extract information about the currently processed entry and don't need to return ARCHIVE_FATAL here. */ /*return (ARCHIVE_FATAL);*/ } if (rar->file_flags & FHD_LARGE) { memcpy(packed_size, file_header.pack_size, 4); memcpy(packed_size + 4, p, 4); /* High pack size */ p += 4; memcpy(unp_size, file_header.unp_size, 4); memcpy(unp_size + 4, p, 4); /* High unpack size */ p += 4; rar->packed_size = archive_le64dec(&packed_size); rar->unp_size = archive_le64dec(&unp_size); } else { rar->packed_size = archive_le32dec(file_header.pack_size); rar->unp_size = archive_le32dec(file_header.unp_size); } if (rar->packed_size < 0 || rar->unp_size < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid sizes specified."); return (ARCHIVE_FATAL); } rar->bytes_remaining = rar->packed_size; /* TODO: RARv3 subblocks contain comments. For now the complete block is * consumed at the end. */ if (head_type == NEWSUB_HEAD) { size_t distance = p - (const char *)h; header_size += rar->packed_size; /* Make sure we have the extended data. */ if ((h = __archive_read_ahead(a, (size_t)header_size - 7, NULL)) == NULL) return (ARCHIVE_FATAL); p = h; endp = p + header_size - 7; p += distance; } filename_size = archive_le16dec(file_header.name_size); if (p + filename_size > endp) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid filename size"); return (ARCHIVE_FATAL); } if (rar->filename_allocated < filename_size * 2 + 2) { char *newptr; size_t newsize = filename_size * 2 + 2; newptr = realloc(rar->filename, newsize); if (newptr == NULL) { archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory."); return (ARCHIVE_FATAL); } rar->filename = newptr; rar->filename_allocated = newsize; } filename = rar->filename; memcpy(filename, p, filename_size); filename[filename_size] = '\0'; if (rar->file_flags & FHD_UNICODE) { if (filename_size != strlen(filename)) { unsigned char highbyte, flagbits, flagbyte; unsigned fn_end, offset; end = filename_size; fn_end = filename_size * 2; filename_size = 0; offset = (unsigned)strlen(filename) + 1; highbyte = *(p + offset++); flagbits = 0; flagbyte = 0; while (offset < end && filename_size < fn_end) { if (!flagbits) { flagbyte = *(p + offset++); flagbits = 8; } flagbits -= 2; switch((flagbyte >> flagbits) & 3) { case 0: filename[filename_size++] = '\0'; filename[filename_size++] = *(p + offset++); break; case 1: filename[filename_size++] = highbyte; filename[filename_size++] = *(p + offset++); break; case 2: filename[filename_size++] = *(p + offset + 1); filename[filename_size++] = *(p + offset); offset += 2; break; case 3: { char extra, high; uint8_t length = *(p + offset++); if (length & 0x80) { extra = *(p + offset++); high = (char)highbyte; } else extra = high = 0; length = (length & 0x7f) + 2; while (length && filename_size < fn_end) { unsigned cp = filename_size >> 1; filename[filename_size++] = high; filename[filename_size++] = p[cp] + extra; length--; } } break; } } if (filename_size > fn_end) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid filename"); return (ARCHIVE_FATAL); } filename[filename_size++] = '\0'; filename[filename_size++] = '\0'; /* Decoded unicode form is UTF-16BE, so we have to update a string * conversion object for it. */ if (rar->sconv_utf16be == NULL) { rar->sconv_utf16be = archive_string_conversion_from_charset( &a->archive, "UTF-16BE", 1); if (rar->sconv_utf16be == NULL) return (ARCHIVE_FATAL); } fn_sconv = rar->sconv_utf16be; strp = filename; while (memcmp(strp, "\x00\x00", 2)) { if (!memcmp(strp, "\x00\\", 2)) *(strp + 1) = '/'; strp += 2; } p += offset; } else { /* * If FHD_UNICODE is set but no unicode data, this file name form * is UTF-8, so we have to update a string conversion object for * it accordingly. */ if (rar->sconv_utf8 == NULL) { rar->sconv_utf8 = archive_string_conversion_from_charset( &a->archive, "UTF-8", 1); if (rar->sconv_utf8 == NULL) return (ARCHIVE_FATAL); } fn_sconv = rar->sconv_utf8; while ((strp = strchr(filename, '\\')) != NULL) *strp = '/'; p += filename_size; } } else { fn_sconv = sconv; while ((strp = strchr(filename, '\\')) != NULL) *strp = '/'; p += filename_size; } /* Split file in multivolume RAR. No more need to process header. */ if (rar->filename_save && filename_size == rar->filename_save_size && !memcmp(rar->filename, rar->filename_save, filename_size + 1)) { __archive_read_consume(a, header_size - 7); rar->cursor++; if (rar->cursor >= rar->nodes) { rar->nodes++; if ((rar->dbo = realloc(rar->dbo, sizeof(*rar->dbo) * rar->nodes)) == NULL) { archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory."); return (ARCHIVE_FATAL); } rar->dbo[rar->cursor].header_size = header_size; rar->dbo[rar->cursor].start_offset = -1; rar->dbo[rar->cursor].end_offset = -1; } if (rar->dbo[rar->cursor].start_offset < 0) { rar->dbo[rar->cursor].start_offset = a->filter->position; rar->dbo[rar->cursor].end_offset = rar->dbo[rar->cursor].start_offset + rar->packed_size; } return ret; } rar->filename_save = (char*)realloc(rar->filename_save, filename_size + 1); memcpy(rar->filename_save, rar->filename, filename_size + 1); rar->filename_save_size = filename_size; /* Set info for seeking */ free(rar->dbo); if ((rar->dbo = calloc(1, sizeof(*rar->dbo))) == NULL) { archive_set_error(&a->archive, ENOMEM, "Couldn't allocate memory."); return (ARCHIVE_FATAL); } rar->dbo[0].header_size = header_size; rar->dbo[0].start_offset = -1; rar->dbo[0].end_offset = -1; rar->cursor = 0; rar->nodes = 1; if (rar->file_flags & FHD_SALT) { if (p + 8 > endp) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size"); return (ARCHIVE_FATAL); } memcpy(rar->salt, p, 8); p += 8; } if (rar->file_flags & FHD_EXTTIME) { if (read_exttime(p, rar, endp) < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid header size"); return (ARCHIVE_FATAL); } } __archive_read_consume(a, header_size - 7); rar->dbo[0].start_offset = a->filter->position; rar->dbo[0].end_offset = rar->dbo[0].start_offset + rar->packed_size; switch(file_header.host_os) { case OS_MSDOS: case OS_OS2: case OS_WIN32: rar->mode = archive_le32dec(file_header.file_attr); if (rar->mode & FILE_ATTRIBUTE_DIRECTORY) rar->mode = AE_IFDIR | S_IXUSR | S_IXGRP | S_IXOTH; else rar->mode = AE_IFREG; rar->mode |= S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH; break; case OS_UNIX: case OS_MAC_OS: case OS_BEOS: rar->mode = archive_le32dec(file_header.file_attr); break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unknown file attributes from RAR file's host OS"); return (ARCHIVE_FATAL); } rar->bytes_uncopied = rar->bytes_unconsumed = 0; rar->lzss.position = rar->offset = 0; rar->offset_seek = 0; rar->dictionary_size = 0; rar->offset_outgoing = 0; rar->br.cache_avail = 0; rar->br.avail_in = 0; rar->crc_calculated = 0; rar->entry_eof = 0; rar->valid = 1; rar->is_ppmd_block = 0; rar->start_new_table = 1; free(rar->unp_buffer); rar->unp_buffer = NULL; rar->unp_offset = 0; rar->unp_buffer_size = UNP_BUFFER_SIZE; memset(rar->lengthtable, 0, sizeof(rar->lengthtable)); __archive_ppmd7_functions.Ppmd7_Free(&rar->ppmd7_context, &g_szalloc); rar->ppmd_valid = rar->ppmd_eod = 0; /* Don't set any archive entries for non-file header types */ if (head_type == NEWSUB_HEAD) return ret; archive_entry_set_mtime(entry, rar->mtime, rar->mnsec); archive_entry_set_ctime(entry, rar->ctime, rar->cnsec); archive_entry_set_atime(entry, rar->atime, rar->ansec); archive_entry_set_size(entry, rar->unp_size); archive_entry_set_mode(entry, rar->mode); if (archive_entry_copy_pathname_l(entry, filename, filename_size, fn_sconv)) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Pathname cannot be converted from %s to current locale.", archive_string_conversion_charset_name(fn_sconv)); ret = (ARCHIVE_WARN); } if (((rar->mode) & AE_IFMT) == AE_IFLNK) { /* Make sure a symbolic-link file does not have its body. */ rar->bytes_remaining = 0; archive_entry_set_size(entry, 0); /* Read a symbolic-link name. */ if ((ret2 = read_symlink_stored(a, entry, sconv)) < (ARCHIVE_WARN)) return ret2; if (ret > ret2) ret = ret2; } if (rar->bytes_remaining == 0) rar->entry_eof = 1; return ret; } static time_t get_time(int ttime) { struct tm tm; tm.tm_sec = 2 * (ttime & 0x1f); tm.tm_min = (ttime >> 5) & 0x3f; tm.tm_hour = (ttime >> 11) & 0x1f; tm.tm_mday = (ttime >> 16) & 0x1f; tm.tm_mon = ((ttime >> 21) & 0x0f) - 1; tm.tm_year = ((ttime >> 25) & 0x7f) + 80; tm.tm_isdst = -1; return mktime(&tm); } static int read_exttime(const char *p, struct rar *rar, const char *endp) { unsigned rmode, flags, rem, j, count; int ttime, i; struct tm *tm; time_t t; long nsec; if (p + 2 > endp) return (-1); flags = archive_le16dec(p); p += 2; for (i = 3; i >= 0; i--) { t = 0; if (i == 3) t = rar->mtime; rmode = flags >> i * 4; if (rmode & 8) { if (!t) { if (p + 4 > endp) return (-1); ttime = archive_le32dec(p); t = get_time(ttime); p += 4; } rem = 0; count = rmode & 3; if (p + count > endp) return (-1); for (j = 0; j < count; j++) { rem = ((*p) << 16) | (rem >> 8); p++; } tm = localtime(&t); nsec = tm->tm_sec + rem / NS_UNIT; if (rmode & 4) { tm->tm_sec++; t = mktime(tm); } if (i == 3) { rar->mtime = t; rar->mnsec = nsec; } else if (i == 2) { rar->ctime = t; rar->cnsec = nsec; } else if (i == 1) { rar->atime = t; rar->ansec = nsec; } else { rar->arctime = t; rar->arcnsec = nsec; } } } return (0); } static int read_symlink_stored(struct archive_read *a, struct archive_entry *entry, struct archive_string_conv *sconv) { const void *h; const char *p; struct rar *rar; int ret = (ARCHIVE_OK); rar = (struct rar *)(a->format->data); if ((h = rar_read_ahead(a, (size_t)rar->packed_size, NULL)) == NULL) return (ARCHIVE_FATAL); p = h; if (archive_entry_copy_symlink_l(entry, p, (size_t)rar->packed_size, sconv)) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for link"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "link cannot be converted from %s to current locale.", archive_string_conversion_charset_name(sconv)); ret = (ARCHIVE_WARN); } __archive_read_consume(a, rar->packed_size); return ret; } static int read_data_stored(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct rar *rar; ssize_t bytes_avail; rar = (struct rar *)(a->format->data); if (rar->bytes_remaining == 0 && !(rar->main_flags & MHD_VOLUME && rar->file_flags & FHD_SPLIT_AFTER)) { *buff = NULL; *size = 0; *offset = rar->offset; if (rar->file_crc != rar->crc_calculated) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "File CRC error"); return (ARCHIVE_FATAL); } rar->entry_eof = 1; return (ARCHIVE_EOF); } *buff = rar_read_ahead(a, 1, &bytes_avail); if (bytes_avail <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); return (ARCHIVE_FATAL); } *size = bytes_avail; *offset = rar->offset; rar->offset += bytes_avail; rar->offset_seek += bytes_avail; rar->bytes_remaining -= bytes_avail; rar->bytes_unconsumed = bytes_avail; /* Calculate File CRC. */ rar->crc_calculated = crc32(rar->crc_calculated, *buff, (unsigned)bytes_avail); return (ARCHIVE_OK); } static int read_data_compressed(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct rar *rar; int64_t start, end, actualend; size_t bs; int ret = (ARCHIVE_OK), sym, code, lzss_offset, length, i; rar = (struct rar *)(a->format->data); do { if (!rar->valid) return (ARCHIVE_FATAL); if (rar->ppmd_eod || (rar->dictionary_size && rar->offset >= rar->unp_size)) { if (rar->unp_offset > 0) { /* * We have unprocessed extracted data. write it out. */ *buff = rar->unp_buffer; *size = rar->unp_offset; *offset = rar->offset_outgoing; rar->offset_outgoing += *size; /* Calculate File CRC. */ rar->crc_calculated = crc32(rar->crc_calculated, *buff, (unsigned)*size); rar->unp_offset = 0; return (ARCHIVE_OK); } *buff = NULL; *size = 0; *offset = rar->offset; if (rar->file_crc != rar->crc_calculated) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "File CRC error"); return (ARCHIVE_FATAL); } rar->entry_eof = 1; return (ARCHIVE_EOF); } if (!rar->is_ppmd_block && rar->dictionary_size && rar->bytes_uncopied > 0) { if (rar->bytes_uncopied > (rar->unp_buffer_size - rar->unp_offset)) bs = rar->unp_buffer_size - rar->unp_offset; else bs = (size_t)rar->bytes_uncopied; ret = copy_from_lzss_window(a, buff, rar->offset, (int)bs); if (ret != ARCHIVE_OK) return (ret); rar->offset += bs; rar->bytes_uncopied -= bs; if (*buff != NULL) { rar->unp_offset = 0; *size = rar->unp_buffer_size; *offset = rar->offset_outgoing; rar->offset_outgoing += *size; /* Calculate File CRC. */ rar->crc_calculated = crc32(rar->crc_calculated, *buff, (unsigned)*size); return (ret); } continue; } if (!rar->br.next_in && (ret = rar_br_preparation(a, &(rar->br))) < ARCHIVE_WARN) return (ret); if (rar->start_new_table && ((ret = parse_codes(a)) < (ARCHIVE_WARN))) return (ret); if (rar->is_ppmd_block) { if ((sym = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( &rar->ppmd7_context, &rar->range_dec.p)) < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid symbol"); return (ARCHIVE_FATAL); } if(sym != rar->ppmd_escape) { lzss_emit_literal(rar, sym); rar->bytes_uncopied++; } else { if ((code = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( &rar->ppmd7_context, &rar->range_dec.p)) < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid symbol"); return (ARCHIVE_FATAL); } switch(code) { case 0: rar->start_new_table = 1; return read_data_compressed(a, buff, size, offset); case 2: rar->ppmd_eod = 1;/* End Of ppmd Data. */ continue; case 3: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Parsing filters is unsupported."); return (ARCHIVE_FAILED); case 4: lzss_offset = 0; for (i = 2; i >= 0; i--) { if ((code = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( &rar->ppmd7_context, &rar->range_dec.p)) < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid symbol"); return (ARCHIVE_FATAL); } lzss_offset |= code << (i * 8); } if ((length = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( &rar->ppmd7_context, &rar->range_dec.p)) < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid symbol"); return (ARCHIVE_FATAL); } lzss_emit_match(rar, lzss_offset + 2, length + 32); rar->bytes_uncopied += length + 32; break; case 5: if ((length = __archive_ppmd7_functions.Ppmd7_DecodeSymbol( &rar->ppmd7_context, &rar->range_dec.p)) < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid symbol"); return (ARCHIVE_FATAL); } lzss_emit_match(rar, 1, length + 4); rar->bytes_uncopied += length + 4; break; default: lzss_emit_literal(rar, sym); rar->bytes_uncopied++; } } } else { start = rar->offset; end = start + rar->dictionary_size; rar->filterstart = INT64_MAX; if ((actualend = expand(a, end)) < 0) return ((int)actualend); rar->bytes_uncopied = actualend - start; if (rar->bytes_uncopied == 0) { /* Broken RAR files cause this case. * NOTE: If this case were possible on a normal RAR file * we would find out where it was actually bad and * what we would do to solve it. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Internal error extracting RAR file"); return (ARCHIVE_FATAL); } } if (rar->bytes_uncopied > (rar->unp_buffer_size - rar->unp_offset)) bs = rar->unp_buffer_size - rar->unp_offset; else bs = (size_t)rar->bytes_uncopied; ret = copy_from_lzss_window(a, buff, rar->offset, (int)bs); if (ret != ARCHIVE_OK) return (ret); rar->offset += bs; rar->bytes_uncopied -= bs; /* * If *buff is NULL, it means unp_buffer is not full. * So we have to continue extracting a RAR file. */ } while (*buff == NULL); rar->unp_offset = 0; *size = rar->unp_buffer_size; *offset = rar->offset_outgoing; rar->offset_outgoing += *size; /* Calculate File CRC. */ rar->crc_calculated = crc32(rar->crc_calculated, *buff, (unsigned)*size); return ret; } static int parse_codes(struct archive_read *a) { int i, j, val, n, r; unsigned char bitlengths[MAX_SYMBOLS], zerocount, ppmd_flags; unsigned int maxorder; struct huffman_code precode; struct rar *rar = (struct rar *)(a->format->data); struct rar_br *br = &(rar->br); free_codes(a); /* Skip to the next byte */ rar_br_consume_unalined_bits(br); /* PPMd block flag */ if (!rar_br_read_ahead(a, br, 1)) goto truncated_data; if ((rar->is_ppmd_block = rar_br_bits(br, 1)) != 0) { rar_br_consume(br, 1); if (!rar_br_read_ahead(a, br, 7)) goto truncated_data; ppmd_flags = rar_br_bits(br, 7); rar_br_consume(br, 7); /* Memory is allocated in MB */ if (ppmd_flags & 0x20) { if (!rar_br_read_ahead(a, br, 8)) goto truncated_data; rar->dictionary_size = (rar_br_bits(br, 8) + 1) << 20; rar_br_consume(br, 8); } if (ppmd_flags & 0x40) { if (!rar_br_read_ahead(a, br, 8)) goto truncated_data; rar->ppmd_escape = rar->ppmd7_context.InitEsc = rar_br_bits(br, 8); rar_br_consume(br, 8); } else rar->ppmd_escape = 2; if (ppmd_flags & 0x20) { maxorder = (ppmd_flags & 0x1F) + 1; if(maxorder > 16) maxorder = 16 + (maxorder - 16) * 3; if (maxorder == 1) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); return (ARCHIVE_FATAL); } /* Make sure ppmd7_contest is freed before Ppmd7_Construct * because reading a broken file cause this abnormal sequence. */ __archive_ppmd7_functions.Ppmd7_Free(&rar->ppmd7_context, &g_szalloc); rar->bytein.a = a; rar->bytein.Read = &ppmd_read; __archive_ppmd7_functions.PpmdRAR_RangeDec_CreateVTable(&rar->range_dec); rar->range_dec.Stream = &rar->bytein; __archive_ppmd7_functions.Ppmd7_Construct(&rar->ppmd7_context); if (rar->dictionary_size == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid zero dictionary size"); return (ARCHIVE_FATAL); } if (!__archive_ppmd7_functions.Ppmd7_Alloc(&rar->ppmd7_context, rar->dictionary_size, &g_szalloc)) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } if (!__archive_ppmd7_functions.PpmdRAR_RangeDec_Init(&rar->range_dec)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unable to initialize PPMd range decoder"); return (ARCHIVE_FATAL); } __archive_ppmd7_functions.Ppmd7_Init(&rar->ppmd7_context, maxorder); rar->ppmd_valid = 1; } else { if (!rar->ppmd_valid) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid PPMd sequence"); return (ARCHIVE_FATAL); } if (!__archive_ppmd7_functions.PpmdRAR_RangeDec_Init(&rar->range_dec)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unable to initialize PPMd range decoder"); return (ARCHIVE_FATAL); } } } else { rar_br_consume(br, 1); /* Keep existing table flag */ if (!rar_br_read_ahead(a, br, 1)) goto truncated_data; if (!rar_br_bits(br, 1)) memset(rar->lengthtable, 0, sizeof(rar->lengthtable)); rar_br_consume(br, 1); memset(&bitlengths, 0, sizeof(bitlengths)); for (i = 0; i < MAX_SYMBOLS;) { if (!rar_br_read_ahead(a, br, 4)) goto truncated_data; bitlengths[i++] = rar_br_bits(br, 4); rar_br_consume(br, 4); if (bitlengths[i-1] == 0xF) { if (!rar_br_read_ahead(a, br, 4)) goto truncated_data; zerocount = rar_br_bits(br, 4); rar_br_consume(br, 4); if (zerocount) { i--; for (j = 0; j < zerocount + 2 && i < MAX_SYMBOLS; j++) bitlengths[i++] = 0; } } } memset(&precode, 0, sizeof(precode)); r = create_code(a, &precode, bitlengths, MAX_SYMBOLS, MAX_SYMBOL_LENGTH); if (r != ARCHIVE_OK) { free(precode.tree); free(precode.table); return (r); } for (i = 0; i < HUFFMAN_TABLE_SIZE;) { if ((val = read_next_symbol(a, &precode)) < 0) { free(precode.tree); free(precode.table); return (ARCHIVE_FATAL); } if (val < 16) { rar->lengthtable[i] = (rar->lengthtable[i] + val) & 0xF; i++; } else if (val < 18) { if (i == 0) { free(precode.tree); free(precode.table); archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Internal error extracting RAR file."); return (ARCHIVE_FATAL); } if(val == 16) { if (!rar_br_read_ahead(a, br, 3)) { free(precode.tree); free(precode.table); goto truncated_data; } n = rar_br_bits(br, 3) + 3; rar_br_consume(br, 3); } else { if (!rar_br_read_ahead(a, br, 7)) { free(precode.tree); free(precode.table); goto truncated_data; } n = rar_br_bits(br, 7) + 11; rar_br_consume(br, 7); } for (j = 0; j < n && i < HUFFMAN_TABLE_SIZE; j++) { rar->lengthtable[i] = rar->lengthtable[i-1]; i++; } } else { if(val == 18) { if (!rar_br_read_ahead(a, br, 3)) { free(precode.tree); free(precode.table); goto truncated_data; } n = rar_br_bits(br, 3) + 3; rar_br_consume(br, 3); } else { if (!rar_br_read_ahead(a, br, 7)) { free(precode.tree); free(precode.table); goto truncated_data; } n = rar_br_bits(br, 7) + 11; rar_br_consume(br, 7); } for(j = 0; j < n && i < HUFFMAN_TABLE_SIZE; j++) rar->lengthtable[i++] = 0; } } free(precode.tree); free(precode.table); r = create_code(a, &rar->maincode, &rar->lengthtable[0], MAINCODE_SIZE, MAX_SYMBOL_LENGTH); if (r != ARCHIVE_OK) return (r); r = create_code(a, &rar->offsetcode, &rar->lengthtable[MAINCODE_SIZE], OFFSETCODE_SIZE, MAX_SYMBOL_LENGTH); if (r != ARCHIVE_OK) return (r); r = create_code(a, &rar->lowoffsetcode, &rar->lengthtable[MAINCODE_SIZE + OFFSETCODE_SIZE], LOWOFFSETCODE_SIZE, MAX_SYMBOL_LENGTH); if (r != ARCHIVE_OK) return (r); r = create_code(a, &rar->lengthcode, &rar->lengthtable[MAINCODE_SIZE + OFFSETCODE_SIZE + LOWOFFSETCODE_SIZE], LENGTHCODE_SIZE, MAX_SYMBOL_LENGTH); if (r != ARCHIVE_OK) return (r); } if (!rar->dictionary_size || !rar->lzss.window) { /* Seems as though dictionary sizes are not used. Even so, minimize * memory usage as much as possible. */ void *new_window; unsigned int new_size; if (rar->unp_size >= DICTIONARY_MAX_SIZE) new_size = DICTIONARY_MAX_SIZE; else new_size = rar_fls((unsigned int)rar->unp_size) << 1; new_window = realloc(rar->lzss.window, new_size); if (new_window == NULL) { archive_set_error(&a->archive, ENOMEM, "Unable to allocate memory for uncompressed data."); return (ARCHIVE_FATAL); } rar->lzss.window = (unsigned char *)new_window; rar->dictionary_size = new_size; memset(rar->lzss.window, 0, rar->dictionary_size); rar->lzss.mask = rar->dictionary_size - 1; } rar->start_new_table = 0; return (ARCHIVE_OK); truncated_data: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); rar->valid = 0; return (ARCHIVE_FATAL); } static void free_codes(struct archive_read *a) { struct rar *rar = (struct rar *)(a->format->data); free(rar->maincode.tree); free(rar->offsetcode.tree); free(rar->lowoffsetcode.tree); free(rar->lengthcode.tree); free(rar->maincode.table); free(rar->offsetcode.table); free(rar->lowoffsetcode.table); free(rar->lengthcode.table); memset(&rar->maincode, 0, sizeof(rar->maincode)); memset(&rar->offsetcode, 0, sizeof(rar->offsetcode)); memset(&rar->lowoffsetcode, 0, sizeof(rar->lowoffsetcode)); memset(&rar->lengthcode, 0, sizeof(rar->lengthcode)); } static int read_next_symbol(struct archive_read *a, struct huffman_code *code) { unsigned char bit; unsigned int bits; int length, value, node; struct rar *rar; struct rar_br *br; if (!code->table) { if (make_table(a, code) != (ARCHIVE_OK)) return -1; } rar = (struct rar *)(a->format->data); br = &(rar->br); /* Look ahead (peek) at bits */ if (!rar_br_read_ahead(a, br, code->tablesize)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); rar->valid = 0; return -1; } bits = rar_br_bits(br, code->tablesize); length = code->table[bits].length; value = code->table[bits].value; if (length < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid prefix code in bitstream"); return -1; } if (length <= code->tablesize) { /* Skip length bits */ rar_br_consume(br, length); return value; } /* Skip tablesize bits */ rar_br_consume(br, code->tablesize); node = value; while (!(code->tree[node].branches[0] == code->tree[node].branches[1])) { if (!rar_br_read_ahead(a, br, 1)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); rar->valid = 0; return -1; } bit = rar_br_bits(br, 1); rar_br_consume(br, 1); if (code->tree[node].branches[bit] < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid prefix code in bitstream"); return -1; } node = code->tree[node].branches[bit]; } return code->tree[node].branches[0]; } static int create_code(struct archive_read *a, struct huffman_code *code, unsigned char *lengths, int numsymbols, char maxlength) { int i, j, codebits = 0, symbolsleft = numsymbols; code->numentries = 0; code->numallocatedentries = 0; if (new_node(code) < 0) { archive_set_error(&a->archive, ENOMEM, "Unable to allocate memory for node data."); return (ARCHIVE_FATAL); } code->numentries = 1; code->minlength = INT_MAX; code->maxlength = INT_MIN; codebits = 0; for(i = 1; i <= maxlength; i++) { for(j = 0; j < numsymbols; j++) { if (lengths[j] != i) continue; if (add_value(a, code, j, codebits, i) != ARCHIVE_OK) return (ARCHIVE_FATAL); codebits++; if (--symbolsleft <= 0) { break; break; } } codebits <<= 1; } return (ARCHIVE_OK); } static int add_value(struct archive_read *a, struct huffman_code *code, int value, int codebits, int length) { int repeatpos, lastnode, bitpos, bit, repeatnode, nextnode; free(code->table); code->table = NULL; if(length > code->maxlength) code->maxlength = length; if(length < code->minlength) code->minlength = length; repeatpos = -1; if (repeatpos == 0 || (repeatpos >= 0 && (((codebits >> (repeatpos - 1)) & 3) == 0 || ((codebits >> (repeatpos - 1)) & 3) == 3))) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid repeat position"); return (ARCHIVE_FATAL); } lastnode = 0; for (bitpos = length - 1; bitpos >= 0; bitpos--) { bit = (codebits >> bitpos) & 1; /* Leaf node check */ if (code->tree[lastnode].branches[0] == code->tree[lastnode].branches[1]) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Prefix found"); return (ARCHIVE_FATAL); } if (bitpos == repeatpos) { /* Open branch check */ if (!(code->tree[lastnode].branches[bit] < 0)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid repeating code"); return (ARCHIVE_FATAL); } if ((repeatnode = new_node(code)) < 0) { archive_set_error(&a->archive, ENOMEM, "Unable to allocate memory for node data."); return (ARCHIVE_FATAL); } if ((nextnode = new_node(code)) < 0) { archive_set_error(&a->archive, ENOMEM, "Unable to allocate memory for node data."); return (ARCHIVE_FATAL); } /* Set branches */ code->tree[lastnode].branches[bit] = repeatnode; code->tree[repeatnode].branches[bit] = repeatnode; code->tree[repeatnode].branches[bit^1] = nextnode; lastnode = nextnode; bitpos++; /* terminating bit already handled, skip it */ } else { /* Open branch check */ if (code->tree[lastnode].branches[bit] < 0) { if (new_node(code) < 0) { archive_set_error(&a->archive, ENOMEM, "Unable to allocate memory for node data."); return (ARCHIVE_FATAL); } code->tree[lastnode].branches[bit] = code->numentries++; } /* set to branch */ lastnode = code->tree[lastnode].branches[bit]; } } if (!(code->tree[lastnode].branches[0] == -1 && code->tree[lastnode].branches[1] == -2)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Prefix found"); return (ARCHIVE_FATAL); } /* Set leaf value */ code->tree[lastnode].branches[0] = value; code->tree[lastnode].branches[1] = value; return (ARCHIVE_OK); } static int new_node(struct huffman_code *code) { void *new_tree; if (code->numallocatedentries == code->numentries) { int new_num_entries = 256; if (code->numentries > 0) { new_num_entries = code->numentries * 2; } new_tree = realloc(code->tree, new_num_entries * sizeof(*code->tree)); if (new_tree == NULL) return (-1); code->tree = (struct huffman_tree_node *)new_tree; code->numallocatedentries = new_num_entries; } code->tree[code->numentries].branches[0] = -1; code->tree[code->numentries].branches[1] = -2; return 1; } static int make_table(struct archive_read *a, struct huffman_code *code) { if (code->maxlength < code->minlength || code->maxlength > 10) code->tablesize = 10; else code->tablesize = code->maxlength; code->table = (struct huffman_table_entry *)calloc(1, sizeof(*code->table) * ((size_t)1 << code->tablesize)); return make_table_recurse(a, code, 0, code->table, 0, code->tablesize); } static int make_table_recurse(struct archive_read *a, struct huffman_code *code, int node, struct huffman_table_entry *table, int depth, int maxdepth) { int currtablesize, i, ret = (ARCHIVE_OK); if (!code->tree) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Huffman tree was not created."); return (ARCHIVE_FATAL); } if (node < 0 || node >= code->numentries) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Invalid location to Huffman tree specified."); return (ARCHIVE_FATAL); } currtablesize = 1 << (maxdepth - depth); if (code->tree[node].branches[0] == code->tree[node].branches[1]) { for(i = 0; i < currtablesize; i++) { table[i].length = depth; table[i].value = code->tree[node].branches[0]; } } else if (node < 0) { for(i = 0; i < currtablesize; i++) table[i].length = -1; } else { if(depth == maxdepth) { table[0].length = maxdepth + 1; table[0].value = node; } else { ret |= make_table_recurse(a, code, code->tree[node].branches[0], table, depth + 1, maxdepth); ret |= make_table_recurse(a, code, code->tree[node].branches[1], table + currtablesize / 2, depth + 1, maxdepth); } } return ret; } static int64_t expand(struct archive_read *a, int64_t end) { static const unsigned char lengthbases[] = { 0, 1, 2, 3, 4, 5, 6, 7, 8, 10, 12, 14, 16, 20, 24, 28, 32, 40, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224 }; static const unsigned char lengthbits[] = { 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5 }; static const unsigned int offsetbases[] = { 0, 1, 2, 3, 4, 6, 8, 12, 16, 24, 32, 48, 64, 96, 128, 192, 256, 384, 512, 768, 1024, 1536, 2048, 3072, 4096, 6144, 8192, 12288, 16384, 24576, 32768, 49152, 65536, 98304, 131072, 196608, 262144, 327680, 393216, 458752, 524288, 589824, 655360, 720896, 786432, 851968, 917504, 983040, 1048576, 1310720, 1572864, 1835008, 2097152, 2359296, 2621440, 2883584, 3145728, 3407872, 3670016, 3932160 }; static const unsigned char offsetbits[] = { 0, 0, 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 8, 9, 9, 10, 10, 11, 11, 12, 12, 13, 13, 14, 14, 15, 15, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 16, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18, 18 }; static const unsigned char shortbases[] = { 0, 4, 8, 16, 32, 64, 128, 192 }; static const unsigned char shortbits[] = { 2, 2, 3, 4, 5, 6, 6, 6 }; int symbol, offs, len, offsindex, lensymbol, i, offssymbol, lowoffsetsymbol; unsigned char newfile; struct rar *rar = (struct rar *)(a->format->data); struct rar_br *br = &(rar->br); if (rar->filterstart < end) end = rar->filterstart; while (1) { if (rar->output_last_match && lzss_position(&rar->lzss) + rar->lastlength <= end) { lzss_emit_match(rar, rar->lastoffset, rar->lastlength); rar->output_last_match = 0; } if(rar->is_ppmd_block || rar->output_last_match || lzss_position(&rar->lzss) >= end) return lzss_position(&rar->lzss); if ((symbol = read_next_symbol(a, &rar->maincode)) < 0) return (ARCHIVE_FATAL); rar->output_last_match = 0; if (symbol < 256) { lzss_emit_literal(rar, symbol); continue; } else if (symbol == 256) { if (!rar_br_read_ahead(a, br, 1)) goto truncated_data; newfile = !rar_br_bits(br, 1); rar_br_consume(br, 1); if(newfile) { rar->start_new_block = 1; if (!rar_br_read_ahead(a, br, 1)) goto truncated_data; rar->start_new_table = rar_br_bits(br, 1); rar_br_consume(br, 1); return lzss_position(&rar->lzss); } else { if (parse_codes(a) != ARCHIVE_OK) return (ARCHIVE_FATAL); continue; } } else if(symbol==257) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Parsing filters is unsupported."); return (ARCHIVE_FAILED); } else if(symbol==258) { if(rar->lastlength == 0) continue; offs = rar->lastoffset; len = rar->lastlength; } else if (symbol <= 262) { offsindex = symbol - 259; offs = rar->oldoffset[offsindex]; if ((lensymbol = read_next_symbol(a, &rar->lengthcode)) < 0) goto bad_data; if (lensymbol > (int)(sizeof(lengthbases)/sizeof(lengthbases[0]))) goto bad_data; if (lensymbol > (int)(sizeof(lengthbits)/sizeof(lengthbits[0]))) goto bad_data; len = lengthbases[lensymbol] + 2; if (lengthbits[lensymbol] > 0) { if (!rar_br_read_ahead(a, br, lengthbits[lensymbol])) goto truncated_data; len += rar_br_bits(br, lengthbits[lensymbol]); rar_br_consume(br, lengthbits[lensymbol]); } for (i = offsindex; i > 0; i--) rar->oldoffset[i] = rar->oldoffset[i-1]; rar->oldoffset[0] = offs; } else if(symbol<=270) { offs = shortbases[symbol-263] + 1; if(shortbits[symbol-263] > 0) { if (!rar_br_read_ahead(a, br, shortbits[symbol-263])) goto truncated_data; offs += rar_br_bits(br, shortbits[symbol-263]); rar_br_consume(br, shortbits[symbol-263]); } len = 2; for(i = 3; i > 0; i--) rar->oldoffset[i] = rar->oldoffset[i-1]; rar->oldoffset[0] = offs; } else { if (symbol-271 > (int)(sizeof(lengthbases)/sizeof(lengthbases[0]))) goto bad_data; if (symbol-271 > (int)(sizeof(lengthbits)/sizeof(lengthbits[0]))) goto bad_data; len = lengthbases[symbol-271]+3; if(lengthbits[symbol-271] > 0) { if (!rar_br_read_ahead(a, br, lengthbits[symbol-271])) goto truncated_data; len += rar_br_bits(br, lengthbits[symbol-271]); rar_br_consume(br, lengthbits[symbol-271]); } if ((offssymbol = read_next_symbol(a, &rar->offsetcode)) < 0) goto bad_data; if (offssymbol > (int)(sizeof(offsetbases)/sizeof(offsetbases[0]))) goto bad_data; if (offssymbol > (int)(sizeof(offsetbits)/sizeof(offsetbits[0]))) goto bad_data; offs = offsetbases[offssymbol]+1; if(offsetbits[offssymbol] > 0) { if(offssymbol > 9) { if(offsetbits[offssymbol] > 4) { if (!rar_br_read_ahead(a, br, offsetbits[offssymbol] - 4)) goto truncated_data; offs += rar_br_bits(br, offsetbits[offssymbol] - 4) << 4; rar_br_consume(br, offsetbits[offssymbol] - 4); } if(rar->numlowoffsetrepeats > 0) { rar->numlowoffsetrepeats--; offs += rar->lastlowoffset; } else { if ((lowoffsetsymbol = read_next_symbol(a, &rar->lowoffsetcode)) < 0) return (ARCHIVE_FATAL); if(lowoffsetsymbol == 16) { rar->numlowoffsetrepeats = 15; offs += rar->lastlowoffset; } else { offs += lowoffsetsymbol; rar->lastlowoffset = lowoffsetsymbol; } } } else { if (!rar_br_read_ahead(a, br, offsetbits[offssymbol])) goto truncated_data; offs += rar_br_bits(br, offsetbits[offssymbol]); rar_br_consume(br, offsetbits[offssymbol]); } } if (offs >= 0x40000) len++; if (offs >= 0x2000) len++; for(i = 3; i > 0; i--) rar->oldoffset[i] = rar->oldoffset[i-1]; rar->oldoffset[0] = offs; } rar->lastoffset = offs; rar->lastlength = len; rar->output_last_match = 1; } truncated_data: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated RAR file data"); rar->valid = 0; return (ARCHIVE_FATAL); bad_data: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Bad RAR file data"); return (ARCHIVE_FATAL); } static int copy_from_lzss_window(struct archive_read *a, const void **buffer, int64_t startpos, int length) { int windowoffs, firstpart; struct rar *rar = (struct rar *)(a->format->data); if (!rar->unp_buffer) { if ((rar->unp_buffer = malloc(rar->unp_buffer_size)) == NULL) { archive_set_error(&a->archive, ENOMEM, "Unable to allocate memory for uncompressed data."); return (ARCHIVE_FATAL); } } windowoffs = lzss_offset_for_position(&rar->lzss, startpos); if(windowoffs + length <= lzss_size(&rar->lzss)) { memcpy(&rar->unp_buffer[rar->unp_offset], &rar->lzss.window[windowoffs], length); } else if (length <= lzss_size(&rar->lzss)) { firstpart = lzss_size(&rar->lzss) - windowoffs; if (firstpart < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Bad RAR file data"); return (ARCHIVE_FATAL); } if (firstpart < length) { memcpy(&rar->unp_buffer[rar->unp_offset], &rar->lzss.window[windowoffs], firstpart); memcpy(&rar->unp_buffer[rar->unp_offset + firstpart], &rar->lzss.window[0], length - firstpart); } else { memcpy(&rar->unp_buffer[rar->unp_offset], &rar->lzss.window[windowoffs], length); } } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Bad RAR file data"); return (ARCHIVE_FATAL); } rar->unp_offset += length; if (rar->unp_offset >= rar->unp_buffer_size) *buffer = rar->unp_buffer; else *buffer = NULL; return (ARCHIVE_OK); } static const void * rar_read_ahead(struct archive_read *a, size_t min, ssize_t *avail) { struct rar *rar = (struct rar *)(a->format->data); const void *h = __archive_read_ahead(a, min, avail); int ret; if (avail) { if (a->archive.read_data_is_posix_read && *avail > (ssize_t)a->archive.read_data_requested) *avail = a->archive.read_data_requested; if (*avail > rar->bytes_remaining) *avail = (ssize_t)rar->bytes_remaining; if (*avail < 0) return NULL; else if (*avail == 0 && rar->main_flags & MHD_VOLUME && rar->file_flags & FHD_SPLIT_AFTER) { ret = archive_read_format_rar_read_header(a, a->entry); if (ret == (ARCHIVE_EOF)) { rar->has_endarc_header = 1; ret = archive_read_format_rar_read_header(a, a->entry); } if (ret != (ARCHIVE_OK)) return NULL; return rar_read_ahead(a, min, avail); } } return h; } Index: head/contrib/libarchive/libarchive/archive_read_support_format_tar.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_format_tar.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_read_support_format_tar.c (revision 311903) @@ -1,2856 +1,2877 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * Copyright (c) 2011-2012 Michihiro NAKAJIMA * Copyright (c) 2016 Martin Matuska * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "archive.h" #include "archive_acl_private.h" /* For ACL parsing routines. */ #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_read_private.h" #define tar_min(a,b) ((a) < (b) ? (a) : (b)) /* * Layout of POSIX 'ustar' tar header. */ struct archive_entry_header_ustar { char name[100]; char mode[8]; char uid[8]; char gid[8]; char size[12]; char mtime[12]; char checksum[8]; char typeflag[1]; char linkname[100]; /* "old format" header ends here */ char magic[6]; /* For POSIX: "ustar\0" */ char version[2]; /* For POSIX: "00" */ char uname[32]; char gname[32]; char rdevmajor[8]; char rdevminor[8]; char prefix[155]; }; /* * Structure of GNU tar header */ struct gnu_sparse { char offset[12]; char numbytes[12]; }; struct archive_entry_header_gnutar { char name[100]; char mode[8]; char uid[8]; char gid[8]; char size[12]; char mtime[12]; char checksum[8]; char typeflag[1]; char linkname[100]; char magic[8]; /* "ustar \0" (note blank/blank/null at end) */ char uname[32]; char gname[32]; char rdevmajor[8]; char rdevminor[8]; char atime[12]; char ctime[12]; char offset[12]; char longnames[4]; char unused[1]; struct gnu_sparse sparse[4]; char isextended[1]; char realsize[12]; /* * Old GNU format doesn't use POSIX 'prefix' field; they use * the 'L' (longname) entry instead. */ }; /* * Data specific to this format. */ struct sparse_block { struct sparse_block *next; int64_t offset; int64_t remaining; int hole; }; struct tar { struct archive_string acl_text; struct archive_string entry_pathname; /* For "GNU.sparse.name" and other similar path extensions. */ struct archive_string entry_pathname_override; struct archive_string entry_linkpath; struct archive_string entry_uname; struct archive_string entry_gname; struct archive_string longlink; struct archive_string longname; struct archive_string pax_header; struct archive_string pax_global; struct archive_string line; int pax_hdrcharset_binary; int header_recursion_depth; int64_t entry_bytes_remaining; int64_t entry_offset; int64_t entry_padding; int64_t entry_bytes_unconsumed; int64_t realsize; int sparse_allowed; struct sparse_block *sparse_list; struct sparse_block *sparse_last; int64_t sparse_offset; int64_t sparse_numbytes; int sparse_gnu_major; int sparse_gnu_minor; char sparse_gnu_pending; struct archive_string localname; struct archive_string_conv *opt_sconv; struct archive_string_conv *sconv; struct archive_string_conv *sconv_acl; struct archive_string_conv *sconv_default; int init_default_conversion; int compat_2x; int process_mac_extensions; int read_concatenated_archives; }; static int archive_block_is_null(const char *p); static char *base64_decode(const char *, size_t, size_t *); static int gnu_add_sparse_entry(struct archive_read *, struct tar *, int64_t offset, int64_t remaining); static void gnu_clear_sparse_list(struct tar *); static int gnu_sparse_old_read(struct archive_read *, struct tar *, const struct archive_entry_header_gnutar *header, size_t *); static int gnu_sparse_old_parse(struct archive_read *, struct tar *, const struct gnu_sparse *sparse, int length); static int gnu_sparse_01_parse(struct archive_read *, struct tar *, const char *); static ssize_t gnu_sparse_10_read(struct archive_read *, struct tar *, size_t *); static int header_Solaris_ACL(struct archive_read *, struct tar *, struct archive_entry *, const void *, size_t *); static int header_common(struct archive_read *, struct tar *, struct archive_entry *, const void *); static int header_old_tar(struct archive_read *, struct tar *, struct archive_entry *, const void *); static int header_pax_extensions(struct archive_read *, struct tar *, struct archive_entry *, const void *, size_t *); static int header_pax_global(struct archive_read *, struct tar *, struct archive_entry *, const void *h, size_t *); static int header_longlink(struct archive_read *, struct tar *, struct archive_entry *, const void *h, size_t *); static int header_longname(struct archive_read *, struct tar *, struct archive_entry *, const void *h, size_t *); static int read_mac_metadata_blob(struct archive_read *, struct tar *, struct archive_entry *, const void *h, size_t *); static int header_volume(struct archive_read *, struct tar *, struct archive_entry *, const void *h, size_t *); static int header_ustar(struct archive_read *, struct tar *, struct archive_entry *, const void *h); static int header_gnutar(struct archive_read *, struct tar *, struct archive_entry *, const void *h, size_t *); static int archive_read_format_tar_bid(struct archive_read *, int); static int archive_read_format_tar_options(struct archive_read *, const char *, const char *); static int archive_read_format_tar_cleanup(struct archive_read *); static int archive_read_format_tar_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset); static int archive_read_format_tar_skip(struct archive_read *a); static int archive_read_format_tar_read_header(struct archive_read *, struct archive_entry *); static int checksum(struct archive_read *, const void *); static int pax_attribute(struct archive_read *, struct tar *, - struct archive_entry *, const char *key, const char *value); + struct archive_entry *, const char *key, const char *value, + size_t value_length); static int pax_attribute_acl(struct archive_read *, struct tar *, struct archive_entry *, const char *, int); static int pax_attribute_xattr(struct archive_entry *, const char *, const char *); static int pax_header(struct archive_read *, struct tar *, - struct archive_entry *, char *attr); + struct archive_entry *, struct archive_string *); static void pax_time(const char *, int64_t *sec, long *nanos); static ssize_t readline(struct archive_read *, struct tar *, const char **, ssize_t limit, size_t *); static int read_body_to_string(struct archive_read *, struct tar *, struct archive_string *, const void *h, size_t *); static int solaris_sparse_parse(struct archive_read *, struct tar *, struct archive_entry *, const char *); static int64_t tar_atol(const char *, size_t); static int64_t tar_atol10(const char *, size_t); static int64_t tar_atol256(const char *, size_t); static int64_t tar_atol8(const char *, size_t); static int tar_read_header(struct archive_read *, struct tar *, struct archive_entry *, size_t *); static int tohex(int c); static char *url_decode(const char *); static void tar_flush_unconsumed(struct archive_read *, size_t *); int archive_read_support_format_gnutar(struct archive *a) { archive_check_magic(a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_gnutar"); return (archive_read_support_format_tar(a)); } int archive_read_support_format_tar(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct tar *tar; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_tar"); tar = (struct tar *)calloc(1, sizeof(*tar)); #ifdef HAVE_COPYFILE_H /* Set this by default on Mac OS. */ tar->process_mac_extensions = 1; #endif if (tar == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate tar data"); return (ARCHIVE_FATAL); } r = __archive_read_register_format(a, tar, "tar", archive_read_format_tar_bid, archive_read_format_tar_options, archive_read_format_tar_read_header, archive_read_format_tar_read_data, archive_read_format_tar_skip, NULL, archive_read_format_tar_cleanup, NULL, NULL); if (r != ARCHIVE_OK) free(tar); return (ARCHIVE_OK); } static int archive_read_format_tar_cleanup(struct archive_read *a) { struct tar *tar; tar = (struct tar *)(a->format->data); gnu_clear_sparse_list(tar); archive_string_free(&tar->acl_text); archive_string_free(&tar->entry_pathname); archive_string_free(&tar->entry_pathname_override); archive_string_free(&tar->entry_linkpath); archive_string_free(&tar->entry_uname); archive_string_free(&tar->entry_gname); archive_string_free(&tar->line); archive_string_free(&tar->pax_global); archive_string_free(&tar->pax_header); archive_string_free(&tar->longname); archive_string_free(&tar->longlink); archive_string_free(&tar->localname); free(tar); (a->format->data) = NULL; return (ARCHIVE_OK); } /* * Validate number field * * This has to be pretty lenient in order to accommodate the enormous * variety of tar writers in the world: * = POSIX (IEEE Std 1003.1-1988) ustar requires octal values with leading * zeros and allows fields to be terminated with space or null characters * = Many writers use different termination (in particular, libarchive * omits terminator bytes to squeeze one or two more digits) * = Many writers pad with space and omit leading zeros * = GNU tar and star write base-256 values if numbers are too * big to be represented in octal * * Examples of specific tar headers that we should support: * = Perl Archive::Tar terminates uid, gid, devminor and devmajor with two * null bytes, pads size with spaces and other numeric fields with zeroes * = plexus-archiver prior to 2.6.3 (before switching to commons-compress) * may have uid and gid fields filled with spaces without any octal digits * at all and pads all numeric fields with spaces * * This should tolerate all variants in use. It will reject a field * where the writer just left garbage after a trailing NUL. */ static int validate_number_field(const char* p_field, size_t i_size) { unsigned char marker = (unsigned char)p_field[0]; if (marker == 128 || marker == 255 || marker == 0) { /* Base-256 marker, there's nothing we can check. */ return 1; } else { /* Must be octal */ size_t i = 0; /* Skip any leading spaces */ while (i < i_size && p_field[i] == ' ') { ++i; } /* Skip octal digits. */ while (i < i_size && p_field[i] >= '0' && p_field[i] <= '7') { ++i; } /* Any remaining characters must be space or NUL padding. */ while (i < i_size) { if (p_field[i] != ' ' && p_field[i] != 0) { return 0; } ++i; } return 1; } } static int archive_read_format_tar_bid(struct archive_read *a, int best_bid) { int bid; const char *h; const struct archive_entry_header_ustar *header; (void)best_bid; /* UNUSED */ bid = 0; /* Now let's look at the actual header and see if it matches. */ h = __archive_read_ahead(a, 512, NULL); if (h == NULL) return (-1); /* If it's an end-of-archive mark, we can handle it. */ if (h[0] == 0 && archive_block_is_null(h)) { /* * Usually, I bid the number of bits verified, but * in this case, 4096 seems excessive so I picked 10 as * an arbitrary but reasonable-seeming value. */ return (10); } /* If it's not an end-of-archive mark, it must have a valid checksum.*/ if (!checksum(a, h)) return (0); bid += 48; /* Checksum is usually 6 octal digits. */ header = (const struct archive_entry_header_ustar *)h; /* Recognize POSIX formats. */ if ((memcmp(header->magic, "ustar\0", 6) == 0) && (memcmp(header->version, "00", 2) == 0)) bid += 56; /* Recognize GNU tar format. */ if ((memcmp(header->magic, "ustar ", 6) == 0) && (memcmp(header->version, " \0", 2) == 0)) bid += 56; /* Type flag must be null, digit or A-Z, a-z. */ if (header->typeflag[0] != 0 && !( header->typeflag[0] >= '0' && header->typeflag[0] <= '9') && !( header->typeflag[0] >= 'A' && header->typeflag[0] <= 'Z') && !( header->typeflag[0] >= 'a' && header->typeflag[0] <= 'z') ) return (0); bid += 2; /* 6 bits of variation in an 8-bit field leaves 2 bits. */ /* * Check format of mode/uid/gid/mtime/size/rdevmajor/rdevminor fields. */ if (bid > 0 && ( validate_number_field(header->mode, sizeof(header->mode)) == 0 || validate_number_field(header->uid, sizeof(header->uid)) == 0 || validate_number_field(header->gid, sizeof(header->gid)) == 0 || validate_number_field(header->mtime, sizeof(header->mtime)) == 0 || validate_number_field(header->size, sizeof(header->size)) == 0 || validate_number_field(header->rdevmajor, sizeof(header->rdevmajor)) == 0 || validate_number_field(header->rdevminor, sizeof(header->rdevminor)) == 0)) { bid = 0; } return (bid); } static int archive_read_format_tar_options(struct archive_read *a, const char *key, const char *val) { struct tar *tar; int ret = ARCHIVE_FAILED; tar = (struct tar *)(a->format->data); if (strcmp(key, "compat-2x") == 0) { /* Handle UTF-8 filenames as libarchive 2.x */ tar->compat_2x = (val != NULL && val[0] != 0); tar->init_default_conversion = tar->compat_2x; return (ARCHIVE_OK); } else if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "tar: hdrcharset option needs a character-set name"); else { tar->opt_sconv = archive_string_conversion_from_charset( &a->archive, val, 0); if (tar->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } else if (strcmp(key, "mac-ext") == 0) { tar->process_mac_extensions = (val != NULL && val[0] != 0); return (ARCHIVE_OK); } else if (strcmp(key, "read_concatenated_archives") == 0) { tar->read_concatenated_archives = (val != NULL && val[0] != 0); return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } /* utility function- this exists to centralize the logic of tracking * how much unconsumed data we have floating around, and to consume * anything outstanding since we're going to do read_aheads */ static void tar_flush_unconsumed(struct archive_read *a, size_t *unconsumed) { if (*unconsumed) { /* void *data = (void *)__archive_read_ahead(a, *unconsumed, NULL); * this block of code is to poison claimed unconsumed space, ensuring * things break if it is in use still. * currently it WILL break things, so enable it only for debugging this issue if (data) { memset(data, 0xff, *unconsumed); } */ __archive_read_consume(a, *unconsumed); *unconsumed = 0; } } /* * The function invoked by archive_read_next_header(). This * just sets up a few things and then calls the internal * tar_read_header() function below. */ static int archive_read_format_tar_read_header(struct archive_read *a, struct archive_entry *entry) { /* * When converting tar archives to cpio archives, it is * essential that each distinct file have a distinct inode * number. To simplify this, we keep a static count here to * assign fake dev/inode numbers to each tar entry. Note that * pax format archives may overwrite this with something more * useful. * * Ideally, we would track every file read from the archive so * that we could assign the same dev/ino pair to hardlinks, * but the memory required to store a complete lookup table is * probably not worthwhile just to support the relatively * obscure tar->cpio conversion case. */ static int default_inode; static int default_dev; struct tar *tar; const char *p; const wchar_t *wp; int r; size_t l, unconsumed = 0; /* Assign default device/inode values. */ archive_entry_set_dev(entry, 1 + default_dev); /* Don't use zero. */ archive_entry_set_ino(entry, ++default_inode); /* Don't use zero. */ /* Limit generated st_ino number to 16 bits. */ if (default_inode >= 0xffff) { ++default_dev; default_inode = 0; } tar = (struct tar *)(a->format->data); tar->entry_offset = 0; gnu_clear_sparse_list(tar); tar->realsize = -1; /* Mark this as "unset" */ /* Setup default string conversion. */ tar->sconv = tar->opt_sconv; if (tar->sconv == NULL) { if (!tar->init_default_conversion) { tar->sconv_default = archive_string_default_conversion_for_read(&(a->archive)); tar->init_default_conversion = 1; } tar->sconv = tar->sconv_default; } r = tar_read_header(a, tar, entry, &unconsumed); tar_flush_unconsumed(a, &unconsumed); /* * "non-sparse" files are really just sparse files with * a single block. */ if (tar->sparse_list == NULL) { if (gnu_add_sparse_entry(a, tar, 0, tar->entry_bytes_remaining) != ARCHIVE_OK) return (ARCHIVE_FATAL); } else { struct sparse_block *sb; for (sb = tar->sparse_list; sb != NULL; sb = sb->next) { if (!sb->hole) archive_entry_sparse_add_entry(entry, sb->offset, sb->remaining); } } if (r == ARCHIVE_OK && archive_entry_filetype(entry) == AE_IFREG) { /* * "Regular" entry with trailing '/' is really * directory: This is needed for certain old tar * variants and even for some broken newer ones. */ if ((wp = archive_entry_pathname_w(entry)) != NULL) { l = wcslen(wp); if (l > 0 && wp[l - 1] == L'/') { archive_entry_set_filetype(entry, AE_IFDIR); } } else if ((p = archive_entry_pathname(entry)) != NULL) { l = strlen(p); if (l > 0 && p[l - 1] == '/') { archive_entry_set_filetype(entry, AE_IFDIR); } } } return (r); } static int archive_read_format_tar_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { ssize_t bytes_read; struct tar *tar; struct sparse_block *p; tar = (struct tar *)(a->format->data); for (;;) { /* Remove exhausted entries from sparse list. */ while (tar->sparse_list != NULL && tar->sparse_list->remaining == 0) { p = tar->sparse_list; tar->sparse_list = p->next; free(p); } if (tar->entry_bytes_unconsumed) { __archive_read_consume(a, tar->entry_bytes_unconsumed); tar->entry_bytes_unconsumed = 0; } /* If we're at end of file, return EOF. */ if (tar->sparse_list == NULL || tar->entry_bytes_remaining == 0) { if (__archive_read_consume(a, tar->entry_padding) < 0) return (ARCHIVE_FATAL); tar->entry_padding = 0; *buff = NULL; *size = 0; *offset = tar->realsize; return (ARCHIVE_EOF); } *buff = __archive_read_ahead(a, 1, &bytes_read); if (bytes_read < 0) return (ARCHIVE_FATAL); if (*buff == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Truncated tar archive"); return (ARCHIVE_FATAL); } if (bytes_read > tar->entry_bytes_remaining) bytes_read = (ssize_t)tar->entry_bytes_remaining; /* Don't read more than is available in the * current sparse block. */ if (tar->sparse_list->remaining < bytes_read) bytes_read = (ssize_t)tar->sparse_list->remaining; *size = bytes_read; *offset = tar->sparse_list->offset; tar->sparse_list->remaining -= bytes_read; tar->sparse_list->offset += bytes_read; tar->entry_bytes_remaining -= bytes_read; tar->entry_bytes_unconsumed = bytes_read; if (!tar->sparse_list->hole) return (ARCHIVE_OK); /* Current is hole data and skip this. */ } } static int archive_read_format_tar_skip(struct archive_read *a) { int64_t bytes_skipped; int64_t request; struct sparse_block *p; struct tar* tar; tar = (struct tar *)(a->format->data); /* Do not consume the hole of a sparse file. */ request = 0; for (p = tar->sparse_list; p != NULL; p = p->next) { if (!p->hole) { if (p->remaining >= INT64_MAX - request) { return ARCHIVE_FATAL; } request += p->remaining; } } if (request > tar->entry_bytes_remaining) request = tar->entry_bytes_remaining; request += tar->entry_padding + tar->entry_bytes_unconsumed; bytes_skipped = __archive_read_consume(a, request); if (bytes_skipped < 0) return (ARCHIVE_FATAL); tar->entry_bytes_remaining = 0; tar->entry_bytes_unconsumed = 0; tar->entry_padding = 0; /* Free the sparse list. */ gnu_clear_sparse_list(tar); return (ARCHIVE_OK); } /* * This function recursively interprets all of the headers associated * with a single entry. */ static int tar_read_header(struct archive_read *a, struct tar *tar, struct archive_entry *entry, size_t *unconsumed) { ssize_t bytes; int err; const char *h; const struct archive_entry_header_ustar *header; const struct archive_entry_header_gnutar *gnuheader; /* Loop until we find a workable header record. */ for (;;) { tar_flush_unconsumed(a, unconsumed); /* Read 512-byte header record */ h = __archive_read_ahead(a, 512, &bytes); if (bytes < 0) return ((int)bytes); if (bytes == 0) { /* EOF at a block boundary. */ /* Some writers do omit the block of nulls. */ return (ARCHIVE_EOF); } if (bytes < 512) { /* Short block at EOF; this is bad. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated tar archive"); return (ARCHIVE_FATAL); } *unconsumed = 512; /* Header is workable if it's not an end-of-archive mark. */ if (h[0] != 0 || !archive_block_is_null(h)) break; /* Ensure format is set for archives with only null blocks. */ if (a->archive.archive_format_name == NULL) { a->archive.archive_format = ARCHIVE_FORMAT_TAR; a->archive.archive_format_name = "tar"; } if (!tar->read_concatenated_archives) { /* Try to consume a second all-null record, as well. */ tar_flush_unconsumed(a, unconsumed); h = __archive_read_ahead(a, 512, NULL); if (h != NULL && h[0] == 0 && archive_block_is_null(h)) __archive_read_consume(a, 512); archive_clear_error(&a->archive); return (ARCHIVE_EOF); } /* * We're reading concatenated archives, ignore this block and * loop to get the next. */ } /* * Note: If the checksum fails and we return ARCHIVE_RETRY, * then the client is likely to just retry. This is a very * crude way to search for the next valid header! * * TODO: Improve this by implementing a real header scan. */ if (!checksum(a, h)) { tar_flush_unconsumed(a, unconsumed); archive_set_error(&a->archive, EINVAL, "Damaged tar archive"); return (ARCHIVE_RETRY); /* Retryable: Invalid header */ } if (++tar->header_recursion_depth > 32) { tar_flush_unconsumed(a, unconsumed); archive_set_error(&a->archive, EINVAL, "Too many special headers"); return (ARCHIVE_WARN); } /* Determine the format variant. */ header = (const struct archive_entry_header_ustar *)h; switch(header->typeflag[0]) { case 'A': /* Solaris tar ACL */ a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; a->archive.archive_format_name = "Solaris tar"; err = header_Solaris_ACL(a, tar, entry, h, unconsumed); break; case 'g': /* POSIX-standard 'g' header. */ a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; a->archive.archive_format_name = "POSIX pax interchange format"; err = header_pax_global(a, tar, entry, h, unconsumed); if (err == ARCHIVE_EOF) return (err); break; case 'K': /* Long link name (GNU tar, others) */ err = header_longlink(a, tar, entry, h, unconsumed); break; case 'L': /* Long filename (GNU tar, others) */ err = header_longname(a, tar, entry, h, unconsumed); break; case 'V': /* GNU volume header */ err = header_volume(a, tar, entry, h, unconsumed); break; case 'X': /* Used by SUN tar; same as 'x'. */ a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; a->archive.archive_format_name = "POSIX pax interchange format (Sun variant)"; err = header_pax_extensions(a, tar, entry, h, unconsumed); break; case 'x': /* POSIX-standard 'x' header. */ a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; a->archive.archive_format_name = "POSIX pax interchange format"; err = header_pax_extensions(a, tar, entry, h, unconsumed); break; default: gnuheader = (const struct archive_entry_header_gnutar *)h; if (memcmp(gnuheader->magic, "ustar \0", 8) == 0) { a->archive.archive_format = ARCHIVE_FORMAT_TAR_GNUTAR; a->archive.archive_format_name = "GNU tar format"; err = header_gnutar(a, tar, entry, h, unconsumed); } else if (memcmp(header->magic, "ustar", 5) == 0) { if (a->archive.archive_format != ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE) { a->archive.archive_format = ARCHIVE_FORMAT_TAR_USTAR; a->archive.archive_format_name = "POSIX ustar format"; } err = header_ustar(a, tar, entry, h); } else { a->archive.archive_format = ARCHIVE_FORMAT_TAR; a->archive.archive_format_name = "tar (non-POSIX)"; err = header_old_tar(a, tar, entry, h); } } if (err == ARCHIVE_FATAL) return (err); tar_flush_unconsumed(a, unconsumed); h = NULL; header = NULL; --tar->header_recursion_depth; /* Yuck. Apple's design here ends up storing long pathname * extensions for both the AppleDouble extension entry and the * regular entry. */ if ((err == ARCHIVE_WARN || err == ARCHIVE_OK) && tar->header_recursion_depth == 0 && tar->process_mac_extensions) { int err2 = read_mac_metadata_blob(a, tar, entry, h, unconsumed); if (err2 < err) err = err2; } /* We return warnings or success as-is. Anything else is fatal. */ if (err == ARCHIVE_WARN || err == ARCHIVE_OK) { if (tar->sparse_gnu_pending) { if (tar->sparse_gnu_major == 1 && tar->sparse_gnu_minor == 0) { ssize_t bytes_read; tar->sparse_gnu_pending = 0; /* Read initial sparse map. */ bytes_read = gnu_sparse_10_read(a, tar, unconsumed); tar->entry_bytes_remaining -= bytes_read; if (bytes_read < 0) return ((int)bytes_read); } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unrecognized GNU sparse file format"); return (ARCHIVE_WARN); } tar->sparse_gnu_pending = 0; } return (err); } if (err == ARCHIVE_EOF) /* EOF when recursively reading a header is bad. */ archive_set_error(&a->archive, EINVAL, "Damaged tar archive"); return (ARCHIVE_FATAL); } /* * Return true if block checksum is correct. */ static int checksum(struct archive_read *a, const void *h) { const unsigned char *bytes; const struct archive_entry_header_ustar *header; int check, sum; size_t i; (void)a; /* UNUSED */ bytes = (const unsigned char *)h; header = (const struct archive_entry_header_ustar *)h; /* Checksum field must hold an octal number */ for (i = 0; i < sizeof(header->checksum); ++i) { char c = header->checksum[i]; if (c != ' ' && c != '\0' && (c < '0' || c > '7')) return 0; } /* * Test the checksum. Note that POSIX specifies _unsigned_ * bytes for this calculation. */ sum = (int)tar_atol(header->checksum, sizeof(header->checksum)); check = 0; for (i = 0; i < 148; i++) check += (unsigned char)bytes[i]; for (; i < 156; i++) check += 32; for (; i < 512; i++) check += (unsigned char)bytes[i]; if (sum == check) return (1); /* * Repeat test with _signed_ bytes, just in case this archive * was created by an old BSD, Solaris, or HP-UX tar with a * broken checksum calculation. */ check = 0; for (i = 0; i < 148; i++) check += (signed char)bytes[i]; for (; i < 156; i++) check += 32; for (; i < 512; i++) check += (signed char)bytes[i]; if (sum == check) return (1); return (0); } /* * Return true if this block contains only nulls. */ static int archive_block_is_null(const char *p) { unsigned i; for (i = 0; i < 512; i++) if (*p++) return (0); return (1); } /* * Interpret 'A' Solaris ACL header */ static int header_Solaris_ACL(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { const struct archive_entry_header_ustar *header; size_t size; int err; int64_t type; char *acl, *p; /* * read_body_to_string adds a NUL terminator, but we need a little * more to make sure that we don't overrun acl_text later. */ header = (const struct archive_entry_header_ustar *)h; size = (size_t)tar_atol(header->size, sizeof(header->size)); err = read_body_to_string(a, tar, &(tar->acl_text), h, unconsumed); if (err != ARCHIVE_OK) return (err); /* Recursively read next header */ err = tar_read_header(a, tar, entry, unconsumed); if ((err != ARCHIVE_OK) && (err != ARCHIVE_WARN)) return (err); /* TODO: Examine the first characters to see if this * is an AIX ACL descriptor. We'll likely never support * them, but it would be polite to recognize and warn when * we do see them. */ /* Leading octal number indicates ACL type and number of entries. */ p = acl = tar->acl_text.s; type = 0; while (*p != '\0' && p < acl + size) { if (*p < '0' || *p > '7') { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (invalid digit)"); return(ARCHIVE_WARN); } type <<= 3; type += *p - '0'; if (type > 077777777) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (count too large)"); return (ARCHIVE_WARN); } p++; } switch ((int)type & ~0777777) { case 01000000: /* POSIX.1e ACL */ break; case 03000000: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Solaris NFSv4 ACLs not supported"); return (ARCHIVE_WARN); default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (unsupported type %o)", (int)type); return (ARCHIVE_WARN); } p++; if (p >= acl + size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (body overflow)"); return(ARCHIVE_WARN); } /* ACL text is null-terminated; find the end. */ size -= (p - acl); acl = p; while (*p != '\0' && p < acl + size) p++; if (tar->sconv_acl == NULL) { tar->sconv_acl = archive_string_conversion_from_charset( &(a->archive), "UTF-8", 1); if (tar->sconv_acl == NULL) return (ARCHIVE_FATAL); } archive_strncpy(&(tar->localname), acl, p - acl); err = archive_acl_from_text_l(archive_entry_acl(entry), tar->localname.s, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, tar->sconv_acl); if (err != ARCHIVE_OK) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for ACL"); } else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed Solaris ACL attribute (unparsable)"); } return (err); } /* * Interpret 'K' long linkname header. */ static int header_longlink(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { int err; err = read_body_to_string(a, tar, &(tar->longlink), h, unconsumed); if (err != ARCHIVE_OK) return (err); err = tar_read_header(a, tar, entry, unconsumed); if ((err != ARCHIVE_OK) && (err != ARCHIVE_WARN)) return (err); /* Set symlink if symlink already set, else hardlink. */ archive_entry_copy_link(entry, tar->longlink.s); return (ARCHIVE_OK); } static int set_conversion_failed_error(struct archive_read *a, struct archive_string_conv *sconv, const char *name) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for %s", name); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "%s can't be converted from %s to current locale.", name, archive_string_conversion_charset_name(sconv)); return (ARCHIVE_WARN); } /* * Interpret 'L' long filename header. */ static int header_longname(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { int err; err = read_body_to_string(a, tar, &(tar->longname), h, unconsumed); if (err != ARCHIVE_OK) return (err); /* Read and parse "real" header, then override name. */ err = tar_read_header(a, tar, entry, unconsumed); if ((err != ARCHIVE_OK) && (err != ARCHIVE_WARN)) return (err); if (archive_entry_copy_pathname_l(entry, tar->longname.s, archive_strlen(&(tar->longname)), tar->sconv) != 0) err = set_conversion_failed_error(a, tar->sconv, "Pathname"); return (err); } /* * Interpret 'V' GNU tar volume header. */ static int header_volume(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { (void)h; /* Just skip this and read the next header. */ return (tar_read_header(a, tar, entry, unconsumed)); } /* * Read body of an archive entry into an archive_string object. */ static int read_body_to_string(struct archive_read *a, struct tar *tar, struct archive_string *as, const void *h, size_t *unconsumed) { int64_t size; const struct archive_entry_header_ustar *header; const void *src; (void)tar; /* UNUSED */ header = (const struct archive_entry_header_ustar *)h; size = tar_atol(header->size, sizeof(header->size)); if ((size > 1048576) || (size < 0)) { archive_set_error(&a->archive, EINVAL, "Special header too large"); return (ARCHIVE_FATAL); } /* Fail if we can't make our buffer big enough. */ if (archive_string_ensure(as, (size_t)size+1) == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory"); return (ARCHIVE_FATAL); } tar_flush_unconsumed(a, unconsumed); /* Read the body into the string. */ *unconsumed = (size_t)((size + 511) & ~ 511); src = __archive_read_ahead(a, *unconsumed, NULL); if (src == NULL) { *unconsumed = 0; return (ARCHIVE_FATAL); } memcpy(as->s, src, (size_t)size); as->s[size] = '\0'; as->length = (size_t)size; return (ARCHIVE_OK); } /* * Parse out common header elements. * * This would be the same as header_old_tar, except that the * filename is handled slightly differently for old and POSIX * entries (POSIX entries support a 'prefix'). This factoring * allows header_old_tar and header_ustar * to handle filenames differently, while still putting most of the * common parsing into one place. */ static int header_common(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h) { const struct archive_entry_header_ustar *header; char tartype; int err = ARCHIVE_OK; header = (const struct archive_entry_header_ustar *)h; if (header->linkname[0]) archive_strncpy(&(tar->entry_linkpath), header->linkname, sizeof(header->linkname)); else archive_string_empty(&(tar->entry_linkpath)); /* Parse out the numeric fields (all are octal) */ archive_entry_set_mode(entry, (mode_t)tar_atol(header->mode, sizeof(header->mode))); archive_entry_set_uid(entry, tar_atol(header->uid, sizeof(header->uid))); archive_entry_set_gid(entry, tar_atol(header->gid, sizeof(header->gid))); tar->entry_bytes_remaining = tar_atol(header->size, sizeof(header->size)); if (tar->entry_bytes_remaining < 0) { tar->entry_bytes_remaining = 0; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Tar entry has negative size"); return (ARCHIVE_FATAL); } if (tar->entry_bytes_remaining == INT64_MAX) { /* Note: tar_atol returns INT64_MAX on overflow */ tar->entry_bytes_remaining = 0; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Tar entry size overflow"); return (ARCHIVE_FATAL); } tar->realsize = tar->entry_bytes_remaining; archive_entry_set_size(entry, tar->entry_bytes_remaining); archive_entry_set_mtime(entry, tar_atol(header->mtime, sizeof(header->mtime)), 0); /* Handle the tar type flag appropriately. */ tartype = header->typeflag[0]; switch (tartype) { case '1': /* Hard link */ if (archive_entry_copy_hardlink_l(entry, tar->entry_linkpath.s, archive_strlen(&(tar->entry_linkpath)), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Linkname"); if (err == ARCHIVE_FATAL) return (err); } /* * The following may seem odd, but: Technically, tar * does not store the file type for a "hard link" * entry, only the fact that it is a hard link. So, I * leave the type zero normally. But, pax interchange * format allows hard links to have data, which * implies that the underlying entry is a regular * file. */ if (archive_entry_size(entry) > 0) archive_entry_set_filetype(entry, AE_IFREG); /* * A tricky point: Traditionally, tar readers have * ignored the size field when reading hardlink * entries, and some writers put non-zero sizes even * though the body is empty. POSIX blessed this * convention in the 1988 standard, but broke with * this tradition in 2001 by permitting hardlink * entries to store valid bodies in pax interchange * format, but not in ustar format. Since there is no * hard and fast way to distinguish pax interchange * from earlier archives (the 'x' and 'g' entries are * optional, after all), we need a heuristic. */ if (archive_entry_size(entry) == 0) { /* If the size is already zero, we're done. */ } else if (a->archive.archive_format == ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE) { /* Definitely pax extended; must obey hardlink size. */ } else if (a->archive.archive_format == ARCHIVE_FORMAT_TAR || a->archive.archive_format == ARCHIVE_FORMAT_TAR_GNUTAR) { /* Old-style or GNU tar: we must ignore the size. */ archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; } else if (archive_read_format_tar_bid(a, 50) > 50) { /* * We don't know if it's pax: If the bid * function sees a valid ustar header * immediately following, then let's ignore * the hardlink size. */ archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; } /* * TODO: There are still two cases I'd like to handle: * = a ustar non-pax archive with a hardlink entry at * end-of-archive. (Look for block of nulls following?) * = a pax archive that has not seen any pax headers * and has an entry which is a hardlink entry storing * a body containing an uncompressed tar archive. * The first is worth addressing; I don't see any reliable * way to deal with the second possibility. */ break; case '2': /* Symlink */ archive_entry_set_filetype(entry, AE_IFLNK); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; if (archive_entry_copy_symlink_l(entry, tar->entry_linkpath.s, archive_strlen(&(tar->entry_linkpath)), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Linkname"); if (err == ARCHIVE_FATAL) return (err); } break; case '3': /* Character device */ archive_entry_set_filetype(entry, AE_IFCHR); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; break; case '4': /* Block device */ archive_entry_set_filetype(entry, AE_IFBLK); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; break; case '5': /* Dir */ archive_entry_set_filetype(entry, AE_IFDIR); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; break; case '6': /* FIFO device */ archive_entry_set_filetype(entry, AE_IFIFO); archive_entry_set_size(entry, 0); tar->entry_bytes_remaining = 0; break; case 'D': /* GNU incremental directory type */ /* * No special handling is actually required here. * It might be nice someday to preprocess the file list and * provide it to the client, though. */ archive_entry_set_filetype(entry, AE_IFDIR); break; case 'M': /* GNU "Multi-volume" (remainder of file from last archive)*/ /* * As far as I can tell, this is just like a regular file * entry, except that the contents should be _appended_ to * the indicated file at the indicated offset. This may * require some API work to fully support. */ break; case 'N': /* Old GNU "long filename" entry. */ /* The body of this entry is a script for renaming * previously-extracted entries. Ugh. It will never * be supported by libarchive. */ archive_entry_set_filetype(entry, AE_IFREG); break; case 'S': /* GNU sparse files */ /* * Sparse files are really just regular files with * sparse information in the extended area. */ /* FALLTHROUGH */ case '0': /* * Enable sparse file "read" support only for regular * files and explicit GNU sparse files. However, we * don't allow non-standard file types to be sparse. */ tar->sparse_allowed = 1; /* FALLTHROUGH */ default: /* Regular file and non-standard types */ /* * Per POSIX: non-recognized types should always be * treated as regular files. */ archive_entry_set_filetype(entry, AE_IFREG); break; } return (err); } /* * Parse out header elements for "old-style" tar archives. */ static int header_old_tar(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h) { const struct archive_entry_header_ustar *header; int err = ARCHIVE_OK, err2; /* Copy filename over (to ensure null termination). */ header = (const struct archive_entry_header_ustar *)h; if (archive_entry_copy_pathname_l(entry, header->name, sizeof(header->name), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Pathname"); if (err == ARCHIVE_FATAL) return (err); } /* Grab rest of common fields */ err2 = header_common(a, tar, entry, h); if (err > err2) err = err2; tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); return (err); } /* * Read a Mac AppleDouble-encoded blob of file metadata, * if there is one. */ static int read_mac_metadata_blob(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { int64_t size; const void *data; const char *p, *name; const wchar_t *wp, *wname; (void)h; /* UNUSED */ wname = wp = archive_entry_pathname_w(entry); if (wp != NULL) { /* Find the last path element. */ for (; *wp != L'\0'; ++wp) { if (wp[0] == '/' && wp[1] != L'\0') wname = wp + 1; } /* * If last path element starts with "._", then * this is a Mac extension. */ if (wname[0] != L'.' || wname[1] != L'_' || wname[2] == L'\0') return ARCHIVE_OK; } else { /* Find the last path element. */ name = p = archive_entry_pathname(entry); if (p == NULL) return (ARCHIVE_FAILED); for (; *p != '\0'; ++p) { if (p[0] == '/' && p[1] != '\0') name = p + 1; } /* * If last path element starts with "._", then * this is a Mac extension. */ if (name[0] != '.' || name[1] != '_' || name[2] == '\0') return ARCHIVE_OK; } /* Read the body as a Mac OS metadata blob. */ size = archive_entry_size(entry); /* * TODO: Look beyond the body here to peek at the next header. * If it's a regular header (not an extension header) * that has the wrong name, just return the current * entry as-is, without consuming the body here. * That would reduce the risk of us mis-identifying * an ordinary file that just happened to have * a name starting with "._". * * Q: Is the above idea really possible? Even * when there are GNU or pax extension entries? */ data = __archive_read_ahead(a, (size_t)size, NULL); if (data == NULL) { *unconsumed = 0; return (ARCHIVE_FATAL); } archive_entry_copy_mac_metadata(entry, data, (size_t)size); *unconsumed = (size_t)((size + 511) & ~ 511); tar_flush_unconsumed(a, unconsumed); return (tar_read_header(a, tar, entry, unconsumed)); } /* * Parse a file header for a pax extended archive entry. */ static int header_pax_global(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { int err; err = read_body_to_string(a, tar, &(tar->pax_global), h, unconsumed); if (err != ARCHIVE_OK) return (err); err = tar_read_header(a, tar, entry, unconsumed); return (err); } static int header_pax_extensions(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { int err, err2; err = read_body_to_string(a, tar, &(tar->pax_header), h, unconsumed); if (err != ARCHIVE_OK) return (err); /* Parse the next header. */ err = tar_read_header(a, tar, entry, unconsumed); if ((err != ARCHIVE_OK) && (err != ARCHIVE_WARN)) return (err); /* * TODO: Parse global/default options into 'entry' struct here * before handling file-specific options. * * This design (parse standard header, then overwrite with pax * extended attribute data) usually works well, but isn't ideal; * it would be better to parse the pax extended attributes first * and then skip any fields in the standard header that were * defined in the pax header. */ - err2 = pax_header(a, tar, entry, tar->pax_header.s); + err2 = pax_header(a, tar, entry, &tar->pax_header); err = err_combine(err, err2); tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); return (err); } /* * Parse a file header for a Posix "ustar" archive entry. This also * handles "pax" or "extended ustar" entries. */ static int header_ustar(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h) { const struct archive_entry_header_ustar *header; struct archive_string *as; int err = ARCHIVE_OK, r; header = (const struct archive_entry_header_ustar *)h; /* Copy name into an internal buffer to ensure null-termination. */ as = &(tar->entry_pathname); if (header->prefix[0]) { archive_strncpy(as, header->prefix, sizeof(header->prefix)); if (as->s[archive_strlen(as) - 1] != '/') archive_strappend_char(as, '/'); archive_strncat(as, header->name, sizeof(header->name)); } else { archive_strncpy(as, header->name, sizeof(header->name)); } if (archive_entry_copy_pathname_l(entry, as->s, archive_strlen(as), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Pathname"); if (err == ARCHIVE_FATAL) return (err); } /* Handle rest of common fields. */ r = header_common(a, tar, entry, h); if (r == ARCHIVE_FATAL) return (r); if (r < err) err = r; /* Handle POSIX ustar fields. */ if (archive_entry_copy_uname_l(entry, header->uname, sizeof(header->uname), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Uname"); if (err == ARCHIVE_FATAL) return (err); } if (archive_entry_copy_gname_l(entry, header->gname, sizeof(header->gname), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Gname"); if (err == ARCHIVE_FATAL) return (err); } /* Parse out device numbers only for char and block specials. */ if (header->typeflag[0] == '3' || header->typeflag[0] == '4') { archive_entry_set_rdevmajor(entry, (dev_t) tar_atol(header->rdevmajor, sizeof(header->rdevmajor))); archive_entry_set_rdevminor(entry, (dev_t) tar_atol(header->rdevminor, sizeof(header->rdevminor))); } tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); return (err); } /* * Parse the pax extended attributes record. * * Returns non-zero if there's an error in the data. */ static int pax_header(struct archive_read *a, struct tar *tar, - struct archive_entry *entry, char *attr) + struct archive_entry *entry, struct archive_string *in_as) { - size_t attr_length, l, line_length; + size_t attr_length, l, line_length, value_length; char *p; char *key, *value; struct archive_string *as; struct archive_string_conv *sconv; int err, err2; + char *attr = in_as->s; - attr_length = strlen(attr); + attr_length = in_as->length; tar->pax_hdrcharset_binary = 0; archive_string_empty(&(tar->entry_gname)); archive_string_empty(&(tar->entry_linkpath)); archive_string_empty(&(tar->entry_pathname)); archive_string_empty(&(tar->entry_pathname_override)); archive_string_empty(&(tar->entry_uname)); err = ARCHIVE_OK; while (attr_length > 0) { /* Parse decimal length field at start of line. */ line_length = 0; l = attr_length; p = attr; /* Record start of line. */ while (l>0) { if (*p == ' ') { p++; l--; break; } if (*p < '0' || *p > '9') { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Ignoring malformed pax extended attributes"); return (ARCHIVE_WARN); } line_length *= 10; line_length += *p - '0'; if (line_length > 999999) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Rejecting pax extended attribute > 1MB"); return (ARCHIVE_WARN); } p++; l--; } /* * Parsed length must be no bigger than available data, * at least 1, and the last character of the line must * be '\n'. */ if (line_length > attr_length || line_length < 1 || attr[line_length - 1] != '\n') { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Ignoring malformed pax extended attribute"); return (ARCHIVE_WARN); } /* Null-terminate the line. */ attr[line_length - 1] = '\0'; /* Find end of key and null terminate it. */ key = p; if (key[0] == '=') return (-1); while (*p && *p != '=') ++p; if (*p == '\0') { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid pax extended attributes"); return (ARCHIVE_WARN); } *p = '\0'; - /* Identify null-terminated 'value' portion. */ value = p + 1; + /* Some values may be binary data */ + value_length = attr + line_length - 1 - value; + /* Identify this attribute and set it in the entry. */ - err2 = pax_attribute(a, tar, entry, key, value); + err2 = pax_attribute(a, tar, entry, key, value, value_length); if (err2 == ARCHIVE_FATAL) return (err2); err = err_combine(err, err2); /* Skip to next line */ attr += line_length; attr_length -= line_length; } /* * PAX format uses UTF-8 as default charset for its metadata * unless hdrcharset=BINARY is present in its header. * We apply the charset specified by the hdrcharset option only * when the hdrcharset attribute(in PAX header) is BINARY because * we respect the charset described in PAX header and BINARY also * means that metadata(filename,uname and gname) character-set * is unknown. */ if (tar->pax_hdrcharset_binary) sconv = tar->opt_sconv; else { sconv = archive_string_conversion_from_charset( &(a->archive), "UTF-8", 1); if (sconv == NULL) return (ARCHIVE_FATAL); if (tar->compat_2x) archive_string_conversion_set_opt(sconv, SCONV_SET_OPT_UTF8_LIBARCHIVE2X); } if (archive_strlen(&(tar->entry_gname)) > 0) { if (archive_entry_copy_gname_l(entry, tar->entry_gname.s, archive_strlen(&(tar->entry_gname)), sconv) != 0) { err = set_conversion_failed_error(a, sconv, "Gname"); if (err == ARCHIVE_FATAL) return (err); /* Use a converted an original name. */ archive_entry_copy_gname(entry, tar->entry_gname.s); } } if (archive_strlen(&(tar->entry_linkpath)) > 0) { if (archive_entry_copy_link_l(entry, tar->entry_linkpath.s, archive_strlen(&(tar->entry_linkpath)), sconv) != 0) { err = set_conversion_failed_error(a, sconv, "Linkname"); if (err == ARCHIVE_FATAL) return (err); /* Use a converted an original name. */ archive_entry_copy_link(entry, tar->entry_linkpath.s); } } /* * Some extensions (such as the GNU sparse file extensions) * deliberately store a synthetic name under the regular 'path' * attribute and the real file name under a different attribute. * Since we're supposed to not care about the order, we * have no choice but to store all of the various filenames * we find and figure it all out afterwards. This is the * figuring out part. */ as = NULL; if (archive_strlen(&(tar->entry_pathname_override)) > 0) as = &(tar->entry_pathname_override); else if (archive_strlen(&(tar->entry_pathname)) > 0) as = &(tar->entry_pathname); if (as != NULL) { if (archive_entry_copy_pathname_l(entry, as->s, archive_strlen(as), sconv) != 0) { err = set_conversion_failed_error(a, sconv, "Pathname"); if (err == ARCHIVE_FATAL) return (err); /* Use a converted an original name. */ archive_entry_copy_pathname(entry, as->s); } } if (archive_strlen(&(tar->entry_uname)) > 0) { if (archive_entry_copy_uname_l(entry, tar->entry_uname.s, archive_strlen(&(tar->entry_uname)), sconv) != 0) { err = set_conversion_failed_error(a, sconv, "Uname"); if (err == ARCHIVE_FATAL) return (err); /* Use a converted an original name. */ archive_entry_copy_uname(entry, tar->entry_uname.s); } } return (err); } static int pax_attribute_xattr(struct archive_entry *entry, const char *name, const char *value) { char *name_decoded; void *value_decoded; size_t value_len; if (strlen(name) < 18 || (memcmp(name, "LIBARCHIVE.xattr.", 17)) != 0) return 3; name += 17; /* URL-decode name */ name_decoded = url_decode(name); if (name_decoded == NULL) return 2; /* Base-64 decode value */ value_decoded = base64_decode(value, strlen(value), &value_len); if (value_decoded == NULL) { free(name_decoded); return 1; } archive_entry_xattr_add_entry(entry, name_decoded, value_decoded, value_len); free(name_decoded); free(value_decoded); return 0; } static int +pax_attribute_schily_xattr(struct archive_entry *entry, + const char *name, const char *value, size_t value_length) +{ + if (strlen(name) < 14 || (memcmp(name, "SCHILY.xattr.", 13)) != 0) + return 1; + + name += 13; + + archive_entry_xattr_add_entry(entry, name, value, value_length); + + return 0; +} + +static int pax_attribute_acl(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const char *value, int type) { int r; const char* errstr; switch (type) { case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: errstr = "SCHILY.acl.access"; break; case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: errstr = "SCHILY.acl.default"; break; case ARCHIVE_ENTRY_ACL_TYPE_NFS4: errstr = "SCHILY.acl.ace"; break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Unknown ACL type: %d", type); return(ARCHIVE_FATAL); } if (tar->sconv_acl == NULL) { tar->sconv_acl = archive_string_conversion_from_charset( &(a->archive), "UTF-8", 1); if (tar->sconv_acl == NULL) return (ARCHIVE_FATAL); } r = archive_acl_from_text_l(archive_entry_acl(entry), value, type, tar->sconv_acl); if (r != ARCHIVE_OK) { if (r == ARCHIVE_FATAL) { archive_set_error(&a->archive, ENOMEM, "%s %s", "Can't allocate memory for ", errstr); return (r); } archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s %s", "Parse error: ", errstr); } return (r); } /* * Parse a single key=value attribute. key/value pointers are * assumed to point into reasonably long-lived storage. * * Note that POSIX reserves all-lowercase keywords. Vendor-specific * extensions should always have keywords of the form "VENDOR.attribute" * In particular, it's quite feasible to support many different * vendor extensions here. I'm using "LIBARCHIVE" for extensions * unique to this library. * * Investigate other vendor-specific extensions and see if * any of them look useful. */ static int pax_attribute(struct archive_read *a, struct tar *tar, - struct archive_entry *entry, const char *key, const char *value) + struct archive_entry *entry, const char *key, const char *value, size_t value_length) { int64_t s; long n; int err = ARCHIVE_OK, r; #ifndef __FreeBSD__ if (value == NULL) value = ""; /* Disable compiler warning; do not pass * NULL pointer to strlen(). */ #endif switch (key[0]) { case 'G': /* Reject GNU.sparse.* headers on non-regular files. */ if (strncmp(key, "GNU.sparse", 10) == 0 && !tar->sparse_allowed) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Non-regular file cannot be sparse"); return (ARCHIVE_FATAL); } /* GNU "0.0" sparse pax format. */ if (strcmp(key, "GNU.sparse.numblocks") == 0) { tar->sparse_offset = -1; tar->sparse_numbytes = -1; tar->sparse_gnu_major = 0; tar->sparse_gnu_minor = 0; } if (strcmp(key, "GNU.sparse.offset") == 0) { tar->sparse_offset = tar_atol10(value, strlen(value)); if (tar->sparse_numbytes != -1) { if (gnu_add_sparse_entry(a, tar, tar->sparse_offset, tar->sparse_numbytes) != ARCHIVE_OK) return (ARCHIVE_FATAL); tar->sparse_offset = -1; tar->sparse_numbytes = -1; } } if (strcmp(key, "GNU.sparse.numbytes") == 0) { tar->sparse_numbytes = tar_atol10(value, strlen(value)); if (tar->sparse_numbytes != -1) { if (gnu_add_sparse_entry(a, tar, tar->sparse_offset, tar->sparse_numbytes) != ARCHIVE_OK) return (ARCHIVE_FATAL); tar->sparse_offset = -1; tar->sparse_numbytes = -1; } } if (strcmp(key, "GNU.sparse.size") == 0) { tar->realsize = tar_atol10(value, strlen(value)); archive_entry_set_size(entry, tar->realsize); } /* GNU "0.1" sparse pax format. */ if (strcmp(key, "GNU.sparse.map") == 0) { tar->sparse_gnu_major = 0; tar->sparse_gnu_minor = 1; if (gnu_sparse_01_parse(a, tar, value) != ARCHIVE_OK) return (ARCHIVE_WARN); } /* GNU "1.0" sparse pax format */ if (strcmp(key, "GNU.sparse.major") == 0) { tar->sparse_gnu_major = (int)tar_atol10(value, strlen(value)); tar->sparse_gnu_pending = 1; } if (strcmp(key, "GNU.sparse.minor") == 0) { tar->sparse_gnu_minor = (int)tar_atol10(value, strlen(value)); tar->sparse_gnu_pending = 1; } if (strcmp(key, "GNU.sparse.name") == 0) { /* * The real filename; when storing sparse * files, GNU tar puts a synthesized name into * the regular 'path' attribute in an attempt * to limit confusion. ;-) */ archive_strcpy(&(tar->entry_pathname_override), value); } if (strcmp(key, "GNU.sparse.realsize") == 0) { tar->realsize = tar_atol10(value, strlen(value)); archive_entry_set_size(entry, tar->realsize); } break; case 'L': /* Our extensions */ /* TODO: Handle arbitrary extended attributes... */ /* if (strcmp(key, "LIBARCHIVE.xxxxxxx") == 0) archive_entry_set_xxxxxx(entry, value); */ if (strcmp(key, "LIBARCHIVE.creationtime") == 0) { pax_time(value, &s, &n); archive_entry_set_birthtime(entry, s, n); } if (memcmp(key, "LIBARCHIVE.xattr.", 17) == 0) pax_attribute_xattr(entry, key, value); break; case 'S': /* We support some keys used by the "star" archiver */ if (strcmp(key, "SCHILY.acl.access") == 0) { r = pax_attribute_acl(a, tar, entry, value, ARCHIVE_ENTRY_ACL_TYPE_ACCESS); if (r == ARCHIVE_FATAL) return (r); } else if (strcmp(key, "SCHILY.acl.default") == 0) { r = pax_attribute_acl(a, tar, entry, value, ARCHIVE_ENTRY_ACL_TYPE_DEFAULT); if (r == ARCHIVE_FATAL) return (r); } else if (strcmp(key, "SCHILY.acl.ace") == 0) { r = pax_attribute_acl(a, tar, entry, value, ARCHIVE_ENTRY_ACL_TYPE_NFS4); if (r == ARCHIVE_FATAL) return (r); } else if (strcmp(key, "SCHILY.devmajor") == 0) { archive_entry_set_rdevmajor(entry, (dev_t)tar_atol10(value, strlen(value))); } else if (strcmp(key, "SCHILY.devminor") == 0) { archive_entry_set_rdevminor(entry, (dev_t)tar_atol10(value, strlen(value))); } else if (strcmp(key, "SCHILY.fflags") == 0) { archive_entry_copy_fflags_text(entry, value); } else if (strcmp(key, "SCHILY.dev") == 0) { archive_entry_set_dev(entry, (dev_t)tar_atol10(value, strlen(value))); } else if (strcmp(key, "SCHILY.ino") == 0) { archive_entry_set_ino(entry, tar_atol10(value, strlen(value))); } else if (strcmp(key, "SCHILY.nlink") == 0) { archive_entry_set_nlink(entry, (unsigned) tar_atol10(value, strlen(value))); } else if (strcmp(key, "SCHILY.realsize") == 0) { tar->realsize = tar_atol10(value, strlen(value)); archive_entry_set_size(entry, tar->realsize); + } else if (strncmp(key, "SCHILY.xattr.", 13) == 0) { + pax_attribute_schily_xattr(entry, key, value, + value_length); } else if (strcmp(key, "SUN.holesdata") == 0) { /* A Solaris extension for sparse. */ r = solaris_sparse_parse(a, tar, entry, value); if (r < err) { if (r == ARCHIVE_FATAL) return (r); err = r; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Parse error: SUN.holesdata"); } } break; case 'a': if (strcmp(key, "atime") == 0) { pax_time(value, &s, &n); archive_entry_set_atime(entry, s, n); } break; case 'c': if (strcmp(key, "ctime") == 0) { pax_time(value, &s, &n); archive_entry_set_ctime(entry, s, n); } else if (strcmp(key, "charset") == 0) { /* TODO: Publish charset information in entry. */ } else if (strcmp(key, "comment") == 0) { /* TODO: Publish comment in entry. */ } break; case 'g': if (strcmp(key, "gid") == 0) { archive_entry_set_gid(entry, tar_atol10(value, strlen(value))); } else if (strcmp(key, "gname") == 0) { archive_strcpy(&(tar->entry_gname), value); } break; case 'h': if (strcmp(key, "hdrcharset") == 0) { if (strcmp(value, "BINARY") == 0) /* Binary mode. */ tar->pax_hdrcharset_binary = 1; else if (strcmp(value, "ISO-IR 10646 2000 UTF-8") == 0) tar->pax_hdrcharset_binary = 0; } break; case 'l': /* pax interchange doesn't distinguish hardlink vs. symlink. */ if (strcmp(key, "linkpath") == 0) { archive_strcpy(&(tar->entry_linkpath), value); } break; case 'm': if (strcmp(key, "mtime") == 0) { pax_time(value, &s, &n); archive_entry_set_mtime(entry, s, n); } break; case 'p': if (strcmp(key, "path") == 0) { archive_strcpy(&(tar->entry_pathname), value); } break; case 'r': /* POSIX has reserved 'realtime.*' */ break; case 's': /* POSIX has reserved 'security.*' */ /* Someday: if (strcmp(key, "security.acl") == 0) { ... } */ if (strcmp(key, "size") == 0) { /* "size" is the size of the data in the entry. */ tar->entry_bytes_remaining = tar_atol10(value, strlen(value)); /* * But, "size" is not necessarily the size of * the file on disk; if this is a sparse file, * the disk size may have already been set from * GNU.sparse.realsize or GNU.sparse.size or * an old GNU header field or SCHILY.realsize * or .... */ if (tar->realsize < 0) { archive_entry_set_size(entry, tar->entry_bytes_remaining); tar->realsize = tar->entry_bytes_remaining; } } break; case 'u': if (strcmp(key, "uid") == 0) { archive_entry_set_uid(entry, tar_atol10(value, strlen(value))); } else if (strcmp(key, "uname") == 0) { archive_strcpy(&(tar->entry_uname), value); } break; } return (err); } /* * parse a decimal time value, which may include a fractional portion */ static void pax_time(const char *p, int64_t *ps, long *pn) { char digit; int64_t s; unsigned long l; int sign; int64_t limit, last_digit_limit; limit = INT64_MAX / 10; last_digit_limit = INT64_MAX % 10; s = 0; sign = 1; if (*p == '-') { sign = -1; p++; } while (*p >= '0' && *p <= '9') { digit = *p - '0'; if (s > limit || (s == limit && digit > last_digit_limit)) { s = INT64_MAX; break; } s = (s * 10) + digit; ++p; } *ps = s * sign; /* Calculate nanoseconds. */ *pn = 0; if (*p != '.') return; l = 100000000UL; do { ++p; if (*p >= '0' && *p <= '9') *pn += (*p - '0') * l; else break; } while (l /= 10); } /* * Parse GNU tar header */ static int header_gnutar(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const void *h, size_t *unconsumed) { const struct archive_entry_header_gnutar *header; int64_t t; int err = ARCHIVE_OK; /* * GNU header is like POSIX ustar, except 'prefix' is * replaced with some other fields. This also means the * filename is stored as in old-style archives. */ /* Grab fields common to all tar variants. */ err = header_common(a, tar, entry, h); if (err == ARCHIVE_FATAL) return (err); /* Copy filename over (to ensure null termination). */ header = (const struct archive_entry_header_gnutar *)h; if (archive_entry_copy_pathname_l(entry, header->name, sizeof(header->name), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Pathname"); if (err == ARCHIVE_FATAL) return (err); } /* Fields common to ustar and GNU */ /* XXX Can the following be factored out since it's common * to ustar and gnu tar? Is it okay to move it down into * header_common, perhaps? */ if (archive_entry_copy_uname_l(entry, header->uname, sizeof(header->uname), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Uname"); if (err == ARCHIVE_FATAL) return (err); } if (archive_entry_copy_gname_l(entry, header->gname, sizeof(header->gname), tar->sconv) != 0) { err = set_conversion_failed_error(a, tar->sconv, "Gname"); if (err == ARCHIVE_FATAL) return (err); } /* Parse out device numbers only for char and block specials */ if (header->typeflag[0] == '3' || header->typeflag[0] == '4') { archive_entry_set_rdevmajor(entry, (dev_t) tar_atol(header->rdevmajor, sizeof(header->rdevmajor))); archive_entry_set_rdevminor(entry, (dev_t) tar_atol(header->rdevminor, sizeof(header->rdevminor))); } else archive_entry_set_rdev(entry, 0); tar->entry_padding = 0x1ff & (-tar->entry_bytes_remaining); /* Grab GNU-specific fields. */ t = tar_atol(header->atime, sizeof(header->atime)); if (t > 0) archive_entry_set_atime(entry, t, 0); t = tar_atol(header->ctime, sizeof(header->ctime)); if (t > 0) archive_entry_set_ctime(entry, t, 0); if (header->realsize[0] != 0) { tar->realsize = tar_atol(header->realsize, sizeof(header->realsize)); archive_entry_set_size(entry, tar->realsize); } if (header->sparse[0].offset[0] != 0) { if (gnu_sparse_old_read(a, tar, header, unconsumed) != ARCHIVE_OK) return (ARCHIVE_FATAL); } else { if (header->isextended[0] != 0) { /* XXX WTF? XXX */ } } return (err); } static int gnu_add_sparse_entry(struct archive_read *a, struct tar *tar, int64_t offset, int64_t remaining) { struct sparse_block *p; p = (struct sparse_block *)calloc(1, sizeof(*p)); if (p == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } if (tar->sparse_last != NULL) tar->sparse_last->next = p; else tar->sparse_list = p; tar->sparse_last = p; if (remaining < 0 || offset < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Malformed sparse map data"); return (ARCHIVE_FATAL); } p->offset = offset; p->remaining = remaining; return (ARCHIVE_OK); } static void gnu_clear_sparse_list(struct tar *tar) { struct sparse_block *p; while (tar->sparse_list != NULL) { p = tar->sparse_list; tar->sparse_list = p->next; free(p); } tar->sparse_last = NULL; } /* * GNU tar old-format sparse data. * * GNU old-format sparse data is stored in a fixed-field * format. Offset/size values are 11-byte octal fields (same * format as 'size' field in ustart header). These are * stored in the header, allocating subsequent header blocks * as needed. Extending the header in this way is a pretty * severe POSIX violation; this design has earned GNU tar a * lot of criticism. */ static int gnu_sparse_old_read(struct archive_read *a, struct tar *tar, const struct archive_entry_header_gnutar *header, size_t *unconsumed) { ssize_t bytes_read; const void *data; struct extended { struct gnu_sparse sparse[21]; char isextended[1]; char padding[7]; }; const struct extended *ext; if (gnu_sparse_old_parse(a, tar, header->sparse, 4) != ARCHIVE_OK) return (ARCHIVE_FATAL); if (header->isextended[0] == 0) return (ARCHIVE_OK); do { tar_flush_unconsumed(a, unconsumed); data = __archive_read_ahead(a, 512, &bytes_read); if (bytes_read < 0) return (ARCHIVE_FATAL); if (bytes_read < 512) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated tar archive " "detected while reading sparse file data"); return (ARCHIVE_FATAL); } *unconsumed = 512; ext = (const struct extended *)data; if (gnu_sparse_old_parse(a, tar, ext->sparse, 21) != ARCHIVE_OK) return (ARCHIVE_FATAL); } while (ext->isextended[0] != 0); if (tar->sparse_list != NULL) tar->entry_offset = tar->sparse_list->offset; return (ARCHIVE_OK); } static int gnu_sparse_old_parse(struct archive_read *a, struct tar *tar, const struct gnu_sparse *sparse, int length) { while (length > 0 && sparse->offset[0] != 0) { if (gnu_add_sparse_entry(a, tar, tar_atol(sparse->offset, sizeof(sparse->offset)), tar_atol(sparse->numbytes, sizeof(sparse->numbytes))) != ARCHIVE_OK) return (ARCHIVE_FATAL); sparse++; length--; } return (ARCHIVE_OK); } /* * GNU tar sparse format 0.0 * * Beginning with GNU tar 1.15, sparse files are stored using * information in the pax extended header. The GNU tar maintainers * have gone through a number of variations in the process of working * out this scheme; fortunately, they're all numbered. * * Sparse format 0.0 uses attribute GNU.sparse.numblocks to store the * number of blocks, and GNU.sparse.offset/GNU.sparse.numbytes to * store offset/size for each block. The repeated instances of these * latter fields violate the pax specification (which frowns on * duplicate keys), so this format was quickly replaced. */ /* * GNU tar sparse format 0.1 * * This version replaced the offset/numbytes attributes with * a single "map" attribute that stored a list of integers. This * format had two problems: First, the "map" attribute could be very * long, which caused problems for some implementations. More * importantly, the sparse data was lost when extracted by archivers * that didn't recognize this extension. */ static int gnu_sparse_01_parse(struct archive_read *a, struct tar *tar, const char *p) { const char *e; int64_t offset = -1, size = -1; for (;;) { e = p; while (*e != '\0' && *e != ',') { if (*e < '0' || *e > '9') return (ARCHIVE_WARN); e++; } if (offset < 0) { offset = tar_atol10(p, e - p); if (offset < 0) return (ARCHIVE_WARN); } else { size = tar_atol10(p, e - p); if (size < 0) return (ARCHIVE_WARN); if (gnu_add_sparse_entry(a, tar, offset, size) != ARCHIVE_OK) return (ARCHIVE_FATAL); offset = -1; } if (*e == '\0') return (ARCHIVE_OK); p = e + 1; } } /* * GNU tar sparse format 1.0 * * The idea: The offset/size data is stored as a series of base-10 * ASCII numbers prepended to the file data, so that dearchivers that * don't support this format will extract the block map along with the * data and a separate post-process can restore the sparseness. * * Unfortunately, GNU tar 1.16 had a bug that added unnecessary * padding to the body of the file when using this format. GNU tar * 1.17 corrected this bug without bumping the version number, so * it's not possible to support both variants. This code supports * the later variant at the expense of not supporting the former. * * This variant also replaced GNU.sparse.size with GNU.sparse.realsize * and introduced the GNU.sparse.major/GNU.sparse.minor attributes. */ /* * Read the next line from the input, and parse it as a decimal * integer followed by '\n'. Returns positive integer value or * negative on error. */ static int64_t gnu_sparse_10_atol(struct archive_read *a, struct tar *tar, int64_t *remaining, size_t *unconsumed) { int64_t l, limit, last_digit_limit; const char *p; ssize_t bytes_read; int base, digit; base = 10; limit = INT64_MAX / base; last_digit_limit = INT64_MAX % base; /* * Skip any lines starting with '#'; GNU tar specs * don't require this, but they should. */ do { bytes_read = readline(a, tar, &p, (ssize_t)tar_min(*remaining, 100), unconsumed); if (bytes_read <= 0) return (ARCHIVE_FATAL); *remaining -= bytes_read; } while (p[0] == '#'); l = 0; while (bytes_read > 0) { if (*p == '\n') return (l); if (*p < '0' || *p >= '0' + base) return (ARCHIVE_WARN); digit = *p - '0'; if (l > limit || (l == limit && digit > last_digit_limit)) l = INT64_MAX; /* Truncate on overflow. */ else l = (l * base) + digit; p++; bytes_read--; } /* TODO: Error message. */ return (ARCHIVE_WARN); } /* * Returns length (in bytes) of the sparse data description * that was read. */ static ssize_t gnu_sparse_10_read(struct archive_read *a, struct tar *tar, size_t *unconsumed) { ssize_t bytes_read; int entries; int64_t offset, size, to_skip, remaining; /* Clear out the existing sparse list. */ gnu_clear_sparse_list(tar); remaining = tar->entry_bytes_remaining; /* Parse entries. */ entries = (int)gnu_sparse_10_atol(a, tar, &remaining, unconsumed); if (entries < 0) return (ARCHIVE_FATAL); /* Parse the individual entries. */ while (entries-- > 0) { /* Parse offset/size */ offset = gnu_sparse_10_atol(a, tar, &remaining, unconsumed); if (offset < 0) return (ARCHIVE_FATAL); size = gnu_sparse_10_atol(a, tar, &remaining, unconsumed); if (size < 0) return (ARCHIVE_FATAL); /* Add a new sparse entry. */ if (gnu_add_sparse_entry(a, tar, offset, size) != ARCHIVE_OK) return (ARCHIVE_FATAL); } /* Skip rest of block... */ tar_flush_unconsumed(a, unconsumed); bytes_read = (ssize_t)(tar->entry_bytes_remaining - remaining); to_skip = 0x1ff & -bytes_read; if (to_skip != __archive_read_consume(a, to_skip)) return (ARCHIVE_FATAL); return ((ssize_t)(bytes_read + to_skip)); } /* * Solaris pax extension for a sparse file. This is recorded with the * data and hole pairs. The way recording sparse information by Solaris' * pax simply indicates where data and sparse are, so the stored contents * consist of both data and hole. */ static int solaris_sparse_parse(struct archive_read *a, struct tar *tar, struct archive_entry *entry, const char *p) { const char *e; int64_t start, end; int hole = 1; (void)entry; /* UNUSED */ end = 0; if (*p == ' ') p++; else return (ARCHIVE_WARN); for (;;) { e = p; while (*e != '\0' && *e != ' ') { if (*e < '0' || *e > '9') return (ARCHIVE_WARN); e++; } start = end; end = tar_atol10(p, e - p); if (end < 0) return (ARCHIVE_WARN); if (start < end) { if (gnu_add_sparse_entry(a, tar, start, end - start) != ARCHIVE_OK) return (ARCHIVE_FATAL); tar->sparse_last->hole = hole; } if (*e == '\0') return (ARCHIVE_OK); p = e + 1; hole = hole == 0; } } /*- * Convert text->integer. * * Traditional tar formats (including POSIX) specify base-8 for * all of the standard numeric fields. This is a significant limitation * in practice: * = file size is limited to 8GB * = rdevmajor and rdevminor are limited to 21 bits * = uid/gid are limited to 21 bits * * There are two workarounds for this: * = pax extended headers, which use variable-length string fields * = GNU tar and STAR both allow either base-8 or base-256 in * most fields. The high bit is set to indicate base-256. * * On read, this implementation supports both extensions. */ static int64_t tar_atol(const char *p, size_t char_cnt) { /* * Technically, GNU tar considers a field to be in base-256 * only if the first byte is 0xff or 0x80. */ if (*p & 0x80) return (tar_atol256(p, char_cnt)); return (tar_atol8(p, char_cnt)); } /* * Note that this implementation does not (and should not!) obey * locale settings; you cannot simply substitute strtol here, since * it does obey locale. */ static int64_t tar_atol_base_n(const char *p, size_t char_cnt, int base) { int64_t l, maxval, limit, last_digit_limit; int digit, sign; maxval = INT64_MAX; limit = INT64_MAX / base; last_digit_limit = INT64_MAX % base; /* the pointer will not be dereferenced if char_cnt is zero * due to the way the && operator is evaluated. */ while (char_cnt != 0 && (*p == ' ' || *p == '\t')) { p++; char_cnt--; } sign = 1; if (char_cnt != 0 && *p == '-') { sign = -1; p++; char_cnt--; maxval = INT64_MIN; limit = -(INT64_MIN / base); last_digit_limit = INT64_MIN % base; } l = 0; if (char_cnt != 0) { digit = *p - '0'; while (digit >= 0 && digit < base && char_cnt != 0) { if (l>limit || (l == limit && digit > last_digit_limit)) { return maxval; /* Truncate on overflow. */ } l = (l * base) + digit; digit = *++p - '0'; char_cnt--; } } return (sign < 0) ? -l : l; } static int64_t tar_atol8(const char *p, size_t char_cnt) { return tar_atol_base_n(p, char_cnt, 8); } static int64_t tar_atol10(const char *p, size_t char_cnt) { return tar_atol_base_n(p, char_cnt, 10); } /* * Parse a base-256 integer. This is just a variable-length * twos-complement signed binary value in big-endian order, except * that the high-order bit is ignored. The values here can be up to * 12 bytes, so we need to be careful about overflowing 64-bit * (8-byte) integers. * * This code unashamedly assumes that the local machine uses 8-bit * bytes and twos-complement arithmetic. */ static int64_t tar_atol256(const char *_p, size_t char_cnt) { uint64_t l; const unsigned char *p = (const unsigned char *)_p; unsigned char c, neg; /* Extend 7-bit 2s-comp to 8-bit 2s-comp, decide sign. */ c = *p; if (c & 0x40) { neg = 0xff; c |= 0x80; l = ~ARCHIVE_LITERAL_ULL(0); } else { neg = 0; c &= 0x7f; l = 0; } /* If more than 8 bytes, check that we can ignore * high-order bits without overflow. */ while (char_cnt > sizeof(int64_t)) { --char_cnt; if (c != neg) return neg ? INT64_MIN : INT64_MAX; c = *++p; } /* c is first byte that fits; if sign mismatch, return overflow */ if ((c ^ neg) & 0x80) { return neg ? INT64_MIN : INT64_MAX; } /* Accumulate remaining bytes. */ while (--char_cnt > 0) { l = (l << 8) | c; c = *++p; } l = (l << 8) | c; /* Return signed twos-complement value. */ return (int64_t)(l); } /* * Returns length of line (including trailing newline) * or negative on error. 'start' argument is updated to * point to first character of line. This avoids copying * when possible. */ static ssize_t readline(struct archive_read *a, struct tar *tar, const char **start, ssize_t limit, size_t *unconsumed) { ssize_t bytes_read; ssize_t total_size = 0; const void *t; const char *s; void *p; tar_flush_unconsumed(a, unconsumed); t = __archive_read_ahead(a, 1, &bytes_read); if (bytes_read <= 0) return (ARCHIVE_FATAL); s = t; /* Start of line? */ p = memchr(t, '\n', bytes_read); /* If we found '\n' in the read buffer, return pointer to that. */ if (p != NULL) { bytes_read = 1 + ((const char *)p) - s; if (bytes_read > limit) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Line too long"); return (ARCHIVE_FATAL); } *unconsumed = bytes_read; *start = s; return (bytes_read); } *unconsumed = bytes_read; /* Otherwise, we need to accumulate in a line buffer. */ for (;;) { if (total_size + bytes_read > limit) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Line too long"); return (ARCHIVE_FATAL); } if (archive_string_ensure(&tar->line, total_size + bytes_read) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate working buffer"); return (ARCHIVE_FATAL); } memcpy(tar->line.s + total_size, t, bytes_read); tar_flush_unconsumed(a, unconsumed); total_size += bytes_read; /* If we found '\n', clean up and return. */ if (p != NULL) { *start = tar->line.s; return (total_size); } /* Read some more. */ t = __archive_read_ahead(a, 1, &bytes_read); if (bytes_read <= 0) return (ARCHIVE_FATAL); s = t; /* Start of line? */ p = memchr(t, '\n', bytes_read); /* If we found '\n', trim the read. */ if (p != NULL) { bytes_read = 1 + ((const char *)p) - s; } *unconsumed = bytes_read; } } /* * base64_decode - Base64 decode * * This accepts most variations of base-64 encoding, including: * * with or without line breaks * * with or without the final group padded with '=' or '_' characters * (The most economical Base-64 variant does not pad the last group and * omits line breaks; RFC1341 used for MIME requires both.) */ static char * base64_decode(const char *s, size_t len, size_t *out_len) { static const unsigned char digits[64] = { 'A','B','C','D','E','F','G','H','I','J','K','L','M','N', 'O','P','Q','R','S','T','U','V','W','X','Y','Z','a','b', 'c','d','e','f','g','h','i','j','k','l','m','n','o','p', 'q','r','s','t','u','v','w','x','y','z','0','1','2','3', '4','5','6','7','8','9','+','/' }; static unsigned char decode_table[128]; char *out, *d; const unsigned char *src = (const unsigned char *)s; /* If the decode table is not yet initialized, prepare it. */ if (decode_table[digits[1]] != 1) { unsigned i; memset(decode_table, 0xff, sizeof(decode_table)); for (i = 0; i < sizeof(digits); i++) decode_table[digits[i]] = i; } /* Allocate enough space to hold the entire output. */ /* Note that we may not use all of this... */ out = (char *)malloc(len - len / 4 + 1); if (out == NULL) { *out_len = 0; return (NULL); } d = out; while (len > 0) { /* Collect the next group of (up to) four characters. */ int v = 0; int group_size = 0; while (group_size < 4 && len > 0) { /* '=' or '_' padding indicates final group. */ if (*src == '=' || *src == '_') { len = 0; break; } /* Skip illegal characters (including line breaks) */ if (*src > 127 || *src < 32 || decode_table[*src] == 0xff) { len--; src++; continue; } v <<= 6; v |= decode_table[*src++]; len --; group_size++; } /* Align a short group properly. */ v <<= 6 * (4 - group_size); /* Unpack the group we just collected. */ switch (group_size) { case 4: d[2] = v & 0xff; /* FALLTHROUGH */ case 3: d[1] = (v >> 8) & 0xff; /* FALLTHROUGH */ case 2: d[0] = (v >> 16) & 0xff; break; case 1: /* this is invalid! */ break; } d += group_size * 3 / 4; } *out_len = d - out; return (out); } static char * url_decode(const char *in) { char *out, *d; const char *s; out = (char *)malloc(strlen(in) + 1); if (out == NULL) return (NULL); for (s = in, d = out; *s != '\0'; ) { if (s[0] == '%' && s[1] != '\0' && s[2] != '\0') { /* Try to convert % escape */ int digit1 = tohex(s[1]); int digit2 = tohex(s[2]); if (digit1 >= 0 && digit2 >= 0) { /* Looks good, consume three chars */ s += 3; /* Convert output */ *d++ = ((digit1 << 4) | digit2); continue; } /* Else fall through and treat '%' as normal char */ } *d++ = *s++; } *d = '\0'; return (out); } static int tohex(int c) { if (c >= '0' && c <= '9') return (c - '0'); else if (c >= 'A' && c <= 'F') return (c - 'A' + 10); else if (c >= 'a' && c <= 'f') return (c - 'a' + 10); else return (-1); } Index: head/contrib/libarchive/libarchive/archive_read_support_format_warc.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_format_warc.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_read_support_format_warc.c (revision 311903) @@ -1,800 +1,800 @@ /*- * Copyright (c) 2014 Sebastian Freundt * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); /** * WARC is standardised by ISO TC46/SC4/WG12 and currently available as * ISO 28500:2009. * For the purposes of this file we used the final draft from: * http://bibnum.bnf.fr/warc/WARC_ISO_28500_version1_latestdraft.pdf * * Todo: * [ ] real-world warcs can contain resources at endpoints ending in / * e.g. http://bibnum.bnf.fr/warc/ * if you're lucky their response contains a Content-Location: header * pointing to a unix-compliant filename, in the example above it's * Content-Location: http://bibnum.bnf.fr/warc/index.html * however, that's not mandated and github for example doesn't follow * this convention. * We need a set of archive options to control what to do with * entries like these, at the moment care is taken to skip them. * **/ #ifdef HAVE_SYS_STAT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_CTYPE_H #include #endif #ifdef HAVE_TIME_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_private.h" #include "archive_read_private.h" typedef enum { WT_NONE, /* warcinfo */ WT_INFO, /* metadata */ WT_META, /* resource */ WT_RSRC, /* request, unsupported */ WT_REQ, /* response, unsupported */ WT_RSP, /* revisit, unsupported */ WT_RVIS, /* conversion, unsupported */ WT_CONV, - /* continutation, unsupported at the moment */ + /* continuation, unsupported at the moment */ WT_CONT, /* invalid type */ LAST_WT } warc_type_t; typedef struct { size_t len; const char *str; } warc_string_t; typedef struct { size_t len; char *str; } warc_strbuf_t; struct warc_s { /* content length ahead */ size_t cntlen; /* and how much we've processed so far */ size_t cntoff; /* and how much we need to consume between calls */ size_t unconsumed; /* string pool */ warc_strbuf_t pool; /* previous version */ unsigned int pver; /* stringified format name */ struct archive_string sver; }; static int _warc_bid(struct archive_read *a, int); static int _warc_cleanup(struct archive_read *a); static int _warc_read(struct archive_read*, const void**, size_t*, int64_t*); static int _warc_skip(struct archive_read *a); static int _warc_rdhdr(struct archive_read *a, struct archive_entry *e); /* private routines */ static unsigned int _warc_rdver(const char buf[10], size_t bsz); static unsigned int _warc_rdtyp(const char *buf, size_t bsz); static warc_string_t _warc_rduri(const char *buf, size_t bsz); static ssize_t _warc_rdlen(const char *buf, size_t bsz); static time_t _warc_rdrtm(const char *buf, size_t bsz); static time_t _warc_rdmtm(const char *buf, size_t bsz); static const char *_warc_find_eoh(const char *buf, size_t bsz); int archive_read_support_format_warc(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct warc_s *w; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_warc"); if ((w = calloc(1, sizeof(*w))) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate warc data"); return (ARCHIVE_FATAL); } r = __archive_read_register_format( a, w, "warc", _warc_bid, NULL, _warc_rdhdr, _warc_read, _warc_skip, NULL, _warc_cleanup, NULL, NULL); if (r != ARCHIVE_OK) { free(w); return (r); } return (ARCHIVE_OK); } static int _warc_cleanup(struct archive_read *a) { struct warc_s *w = a->format->data; if (w->pool.len > 0U) { free(w->pool.str); } archive_string_free(&w->sver); free(w); a->format->data = NULL; return (ARCHIVE_OK); } static int _warc_bid(struct archive_read *a, int best_bid) { const char *hdr; ssize_t nrd; unsigned int ver; (void)best_bid; /* UNUSED */ /* check first line of file, it should be a record already */ if ((hdr = __archive_read_ahead(a, 12U, &nrd)) == NULL) { /* no idea what to do */ return -1; } else if (nrd < 12) { /* nah, not for us, our magic cookie is at least 12 bytes */ return -1; } /* otherwise snarf the record's version number */ ver = _warc_rdver(hdr, nrd); if (ver == 0U || ver > 10000U) { /* oh oh oh, best not to wager ... */ return -1; } /* otherwise be confident */ return (64); } static int _warc_rdhdr(struct archive_read *a, struct archive_entry *entry) { #define HDR_PROBE_LEN (12U) struct warc_s *w = a->format->data; unsigned int ver; const char *buf; ssize_t nrd; const char *eoh; /* for the file name, saves some strndup()'ing */ warc_string_t fnam; /* warc record type, not that we really use it a lot */ warc_type_t ftyp; /* content-length+error monad */ ssize_t cntlen; /* record time is the WARC-Date time we reinterpret it as ctime */ time_t rtime; /* mtime is the Last-Modified time which will be the entry's mtime */ time_t mtime; start_over: /* just use read_ahead() they keep track of unconsumed * bits and bobs for us; no need to put an extra shift in * and reproduce that functionality here */ buf = __archive_read_ahead(a, HDR_PROBE_LEN, &nrd); if (nrd < 0) { /* no good */ archive_set_error( &a->archive, ARCHIVE_ERRNO_MISC, "Bad record header"); return (ARCHIVE_FATAL); } else if (buf == NULL) { /* there should be room for at least WARC/bla\r\n * must be EOF therefore */ return (ARCHIVE_EOF); } /* looks good so far, try and find the end of the header now */ eoh = _warc_find_eoh(buf, nrd); if (eoh == NULL) { /* still no good, the header end might be beyond the * probe we've requested, but then again who'd cram * so much stuff into the header *and* be 28500-compliant */ archive_set_error( &a->archive, ARCHIVE_ERRNO_MISC, "Bad record header"); return (ARCHIVE_FATAL); } else if ((ver = _warc_rdver(buf, eoh - buf)) > 10000U) { /* nawww, I wish they promised backward compatibility * anyhoo, in their infinite wisdom the 28500 guys might * come up with something we can't possibly handle so * best end things here */ archive_set_error( &a->archive, ARCHIVE_ERRNO_MISC, "Unsupported record version"); return (ARCHIVE_FATAL); } else if ((cntlen = _warc_rdlen(buf, eoh - buf)) < 0) { /* nightmare! the specs say content-length is mandatory * so I don't feel overly bad stopping the reader here */ archive_set_error( &a->archive, EINVAL, "Bad content length"); return (ARCHIVE_FATAL); } else if ((rtime = _warc_rdrtm(buf, eoh - buf)) == (time_t)-1) { /* record time is mandatory as per WARC/1.0, * so just barf here, fast and loud */ archive_set_error( &a->archive, EINVAL, "Bad record time"); return (ARCHIVE_FATAL); } /* let the world know we're a WARC archive */ a->archive.archive_format = ARCHIVE_FORMAT_WARC; if (ver != w->pver) { /* stringify this entry's version */ archive_string_sprintf(&w->sver, "WARC/%u.%u", ver / 10000, ver % 10000); /* remember the version */ w->pver = ver; } /* start off with the type */ ftyp = _warc_rdtyp(buf, eoh - buf); /* and let future calls know about the content */ w->cntlen = cntlen; w->cntoff = 0U; mtime = 0;/* Avoid compiling error on some platform. */ switch (ftyp) { case WT_RSRC: case WT_RSP: /* only try and read the filename in the cases that are * guaranteed to have one */ fnam = _warc_rduri(buf, eoh - buf); /* check the last character in the URI to avoid creating * directory endpoints as files, see Todo above */ if (fnam.len == 0 || fnam.str[fnam.len - 1] == '/') { /* break here for now */ fnam.len = 0U; fnam.str = NULL; break; } /* bang to our string pool, so we save a * malloc()+free() roundtrip */ if (fnam.len + 1U > w->pool.len) { w->pool.len = ((fnam.len + 64U) / 64U) * 64U; w->pool.str = realloc(w->pool.str, w->pool.len); } memcpy(w->pool.str, fnam.str, fnam.len); w->pool.str[fnam.len] = '\0'; /* let no one else know about the pool, it's a secret, shhh */ fnam.str = w->pool.str; /* snarf mtime or deduce from rtime * this is a custom header added by our writer, it's quite * hard to believe anyone else would go through with it * (apart from being part of some http responses of course) */ if ((mtime = _warc_rdmtm(buf, eoh - buf)) == (time_t)-1) { mtime = rtime; } break; default: fnam.len = 0U; fnam.str = NULL; break; } /* now eat some of those delicious buffer bits */ __archive_read_consume(a, eoh - buf); switch (ftyp) { case WT_RSRC: case WT_RSP: if (fnam.len > 0U) { /* populate entry object */ archive_entry_set_filetype(entry, AE_IFREG); archive_entry_copy_pathname(entry, fnam.str); archive_entry_set_size(entry, cntlen); archive_entry_set_perm(entry, 0644); /* rtime is the new ctime, mtime stays mtime */ archive_entry_set_ctime(entry, rtime, 0L); archive_entry_set_mtime(entry, mtime, 0L); break; } /* FALLTHROUGH */ default: /* consume the content and start over */ _warc_skip(a); goto start_over; } return (ARCHIVE_OK); } static int _warc_read(struct archive_read *a, const void **buf, size_t *bsz, int64_t *off) { struct warc_s *w = a->format->data; const char *rab; ssize_t nrd; if (w->cntoff >= w->cntlen) { eof: /* it's our lucky day, no work, we can leave early */ *buf = NULL; *bsz = 0U; *off = w->cntoff + 4U/*for \r\n\r\n separator*/; w->unconsumed = 0U; return (ARCHIVE_EOF); } rab = __archive_read_ahead(a, 1U, &nrd); if (nrd < 0) { *bsz = 0U; /* big catastrophe */ return (int)nrd; } else if (nrd == 0) { goto eof; } else if ((size_t)nrd > w->cntlen - w->cntoff) { /* clamp to content-length */ nrd = w->cntlen - w->cntoff; } *off = w->cntoff; *bsz = nrd; *buf = rab; w->cntoff += nrd; w->unconsumed = (size_t)nrd; return (ARCHIVE_OK); } static int _warc_skip(struct archive_read *a) { struct warc_s *w = a->format->data; __archive_read_consume(a, w->cntlen + 4U/*\r\n\r\n separator*/); w->cntlen = 0U; w->cntoff = 0U; return (ARCHIVE_OK); } /* private routines */ static void* deconst(const void *c) { return (char *)0x1 + (((const char *)c) - (const char *)0x1); } static char* xmemmem(const char *hay, const size_t haysize, const char *needle, const size_t needlesize) { const char *const eoh = hay + haysize; const char *const eon = needle + needlesize; const char *hp; const char *np; const char *cand; unsigned int hsum; unsigned int nsum; unsigned int eqp; /* trivial checks first * a 0-sized needle is defined to be found anywhere in haystack * then run strchr() to find a candidate in HAYSTACK (i.e. a portion * that happens to begin with *NEEDLE) */ if (needlesize == 0UL) { return deconst(hay); } else if ((hay = memchr(hay, *needle, haysize)) == NULL) { /* trivial */ return NULL; } /* First characters of haystack and needle are the same now. Both are * guaranteed to be at least one character long. Now computes the sum * of characters values of needle together with the sum of the first * needle_len characters of haystack. */ for (hp = hay + 1U, np = needle + 1U, hsum = *hay, nsum = *hay, eqp = 1U; hp < eoh && np < eon; hsum ^= *hp, nsum ^= *np, eqp &= *hp == *np, hp++, np++); /* HP now references the (NEEDLESIZE + 1)-th character. */ if (np < eon) { /* haystack is smaller than needle, :O */ return NULL; } else if (eqp) { /* found a match */ return deconst(hay); } /* now loop through the rest of haystack, * updating the sum iteratively */ for (cand = hay; hp < eoh; hp++) { hsum ^= *cand++; hsum ^= *hp; /* Since the sum of the characters is already known to be * equal at that point, it is enough to check just NEEDLESIZE - 1 * characters for equality, * also CAND is by design < HP, so no need for range checks */ if (hsum == nsum && memcmp(cand, needle, needlesize - 1U) == 0) { return deconst(cand); } } return NULL; } static int strtoi_lim(const char *str, const char **ep, int llim, int ulim) { int res = 0; const char *sp; /* we keep track of the number of digits via rulim */ int rulim; for (sp = str, rulim = ulim > 10 ? ulim : 10; res * 10 <= ulim && rulim && *sp >= '0' && *sp <= '9'; sp++, rulim /= 10) { res *= 10; res += *sp - '0'; } if (sp == str) { res = -1; } else if (res < llim || res > ulim) { res = -2; } *ep = (const char*)sp; return res; } static time_t time_from_tm(struct tm *t) { #if HAVE_TIMEGM /* Use platform timegm() if available. */ return (timegm(t)); #elif HAVE__MKGMTIME64 return (_mkgmtime64(t)); #else /* Else use direct calculation using POSIX assumptions. */ /* First, fix up tm_yday based on the year/month/day. */ if (mktime(t) == (time_t)-1) return ((time_t)-1); /* Then we can compute timegm() from first principles. */ return (t->tm_sec + t->tm_min * 60 + t->tm_hour * 3600 + t->tm_yday * 86400 + (t->tm_year - 70) * 31536000 + ((t->tm_year - 69) / 4) * 86400 - ((t->tm_year - 1) / 100) * 86400 + ((t->tm_year + 299) / 400) * 86400); #endif } static time_t xstrpisotime(const char *s, char **endptr) { /** like strptime() but strictly for ISO 8601 Zulu strings */ struct tm tm; time_t res = (time_t)-1; /* make sure tm is clean */ memset(&tm, 0, sizeof(tm)); /* as a courtesy to our callers, and since this is a non-standard * routine, we skip leading whitespace */ while (isspace((unsigned char)*s)) ++s; /* read year */ if ((tm.tm_year = strtoi_lim(s, &s, 1583, 4095)) < 0 || *s++ != '-') { goto out; } /* read month */ if ((tm.tm_mon = strtoi_lim(s, &s, 1, 12)) < 0 || *s++ != '-') { goto out; } /* read day-of-month */ if ((tm.tm_mday = strtoi_lim(s, &s, 1, 31)) < 0 || *s++ != 'T') { goto out; } /* read hour */ if ((tm.tm_hour = strtoi_lim(s, &s, 0, 23)) < 0 || *s++ != ':') { goto out; } /* read minute */ if ((tm.tm_min = strtoi_lim(s, &s, 0, 59)) < 0 || *s++ != ':') { goto out; } /* read second */ if ((tm.tm_sec = strtoi_lim(s, &s, 0, 60)) < 0 || *s++ != 'Z') { goto out; } - /* massage TM to fulfill some of POSIX' contraints */ + /* massage TM to fulfill some of POSIX' constraints */ tm.tm_year -= 1900; tm.tm_mon--; /* now convert our custom tm struct to a unix stamp using UTC */ res = time_from_tm(&tm); out: if (endptr != NULL) { *endptr = deconst(s); } return res; } static unsigned int _warc_rdver(const char buf[10], size_t bsz) { static const char magic[] = "WARC/"; unsigned int ver; (void)bsz; /* UNUSED */ if (memcmp(buf, magic, sizeof(magic) - 1U) != 0) { /* nope */ return 99999U; } /* looks good so far, read the version number for a laugh */ buf += sizeof(magic) - 1U; /* most common case gets a quick-check here */ if (memcmp(buf, "1.0\r\n", 5U) == 0) { ver = 10000U; } else { switch (*buf) { case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': if (buf[1U] == '.') { char *on; /* set up major version */ ver = (buf[0U] - '0') * 10000U; /* minor version, anyone? */ ver += (strtol(buf + 2U, &on, 10)) * 100U; /* don't parse anything else */ if (on > buf + 2U) { break; } } /* FALLTHROUGH */ case '9': default: /* just make the version ridiculously high */ ver = 999999U; break; } } return ver; } static unsigned int _warc_rdtyp(const char *buf, size_t bsz) { static const char _key[] = "\r\nWARC-Type:"; const char *const eob = buf + bsz; const char *val; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return WT_NONE; } /* overread whitespace */ val += sizeof(_key) - 1U; while (val < eob && isspace((unsigned char)*val)) ++val; if (val + 8U > eob) { ; } else if (memcmp(val, "resource", 8U) == 0) { return WT_RSRC; } else if (memcmp(val, "warcinfo", 8U) == 0) { return WT_INFO; } else if (memcmp(val, "metadata", 8U) == 0) { return WT_META; } else if (memcmp(val, "request", 7U) == 0) { return WT_REQ; } else if (memcmp(val, "response", 8U) == 0) { return WT_RSP; } else if (memcmp(val, "conversi", 8U) == 0) { return WT_CONV; } else if (memcmp(val, "continua", 8U) == 0) { return WT_CONT; } return WT_NONE; } static warc_string_t _warc_rduri(const char *buf, size_t bsz) { static const char _key[] = "\r\nWARC-Target-URI:"; const char *const eob = buf + bsz; const char *val; const char *uri; const char *eol; warc_string_t res = {0U, NULL}; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return res; } /* overread whitespace */ val += sizeof(_key) - 1U; while (val < eob && isspace((unsigned char)*val)) ++val; /* overread URL designators */ if ((uri = xmemmem(val, eob - val, "://", 3U)) == NULL) { /* not touching that! */ return res; } else if ((eol = memchr(uri, '\n', eob - uri)) == NULL) { /* no end of line? :O */ return res; } /* massage uri to point to after :// */ uri += 3U; /* also massage eol to point to the first whitespace * after the last non-whitespace character before * the end of the line */ while (eol > uri && isspace((unsigned char)eol[-1])) --eol; /* now then, inspect the URI */ if (memcmp(val, "file", 4U) == 0) { /* perfect, nothing left to do here */ } else if (memcmp(val, "http", 4U) == 0 || memcmp(val, "ftp", 3U) == 0) { /* overread domain, and the first / */ while (uri < eol && *uri++ != '/'); } else { /* not sure what to do? best to bugger off */ return res; } res.str = uri; res.len = eol - uri; return res; } static ssize_t _warc_rdlen(const char *buf, size_t bsz) { static const char _key[] = "\r\nContent-Length:"; const char *val; char *on = NULL; long int len; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return -1; } /* strtol kindly overreads whitespace for us, so use that */ val += sizeof(_key) - 1U; len = strtol(val, &on, 10); if (on == NULL || !isspace((unsigned char)*on)) { /* hm, can we trust that number? Best not. */ return -1; } return (size_t)len; } static time_t _warc_rdrtm(const char *buf, size_t bsz) { static const char _key[] = "\r\nWARC-Date:"; const char *val; char *on = NULL; time_t res; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return (time_t)-1; } /* xstrpisotime() kindly overreads whitespace for us, so use that */ val += sizeof(_key) - 1U; res = xstrpisotime(val, &on); if (on == NULL || !isspace((unsigned char)*on)) { /* hm, can we trust that number? Best not. */ return (time_t)-1; } return res; } static time_t _warc_rdmtm(const char *buf, size_t bsz) { static const char _key[] = "\r\nLast-Modified:"; const char *val; char *on = NULL; time_t res; if ((val = xmemmem(buf, bsz, _key, sizeof(_key) - 1U)) == NULL) { /* no bother */ return (time_t)-1; } /* xstrpisotime() kindly overreads whitespace for us, so use that */ val += sizeof(_key) - 1U; res = xstrpisotime(val, &on); if (on == NULL || !isspace((unsigned char)*on)) { /* hm, can we trust that number? Best not. */ return (time_t)-1; } return res; } static const char* _warc_find_eoh(const char *buf, size_t bsz) { static const char _marker[] = "\r\n\r\n"; const char *hit = xmemmem(buf, bsz, _marker, sizeof(_marker) - 1U); if (hit != NULL) { hit += sizeof(_marker) - 1U; } return hit; } /* archive_read_support_format_warc.c ends here */ Index: head/contrib/libarchive/libarchive/archive_read_support_format_zip.c =================================================================== --- head/contrib/libarchive/libarchive/archive_read_support_format_zip.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_read_support_format_zip.c (revision 311903) @@ -1,3082 +1,3082 @@ /*- * Copyright (c) 2004-2013 Tim Kientzle * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA * Copyright (c) 2013 Konrad Kleine * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); /* * The definitive documentation of the Zip file format is: * http://www.pkware.com/documents/casestudies/APPNOTE.TXT * * The Info-Zip project has pioneered various extensions to better * support Zip on Unix, including the 0x5455 "UT", 0x5855 "UX", 0x7855 * "Ux", and 0x7875 "ux" extensions for time and ownership * information. * * History of this code: The streaming Zip reader was first added to * libarchive in January 2005. Support for seekable input sources was * added in Nov 2011. Zip64 support (including a significant code * refactoring) was added in 2014. */ #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_ZLIB_H #include #endif #include "archive.h" #include "archive_digest_private.h" #include "archive_cryptor_private.h" #include "archive_endian.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_hmac_private.h" #include "archive_private.h" #include "archive_rb.h" #include "archive_read_private.h" #ifndef HAVE_ZLIB_H #include "archive_crc32.h" #endif struct zip_entry { struct archive_rb_node node; struct zip_entry *next; int64_t local_header_offset; int64_t compressed_size; int64_t uncompressed_size; int64_t gid; int64_t uid; struct archive_string rsrcname; time_t mtime; time_t atime; time_t ctime; uint32_t crc32; uint16_t mode; uint16_t zip_flags; /* From GP Flags Field */ unsigned char compression; unsigned char system; /* From "version written by" */ unsigned char flags; /* Our extra markers. */ unsigned char decdat;/* Used for Decryption check */ /* WinZip AES encryption extra field should be available * when compression is 99. */ struct { /* Vendor version: AE-1 - 0x0001, AE-2 - 0x0002 */ unsigned vendor; #define AES_VENDOR_AE_1 0x0001 #define AES_VENDOR_AE_2 0x0002 /* AES encryption strength: * 1 - 128 bits, 2 - 192 bits, 2 - 256 bits. */ unsigned strength; /* Actual compression method. */ unsigned char compression; } aes_extra; }; struct trad_enc_ctx { uint32_t keys[3]; }; /* Bits used in zip_flags. */ #define ZIP_ENCRYPTED (1 << 0) #define ZIP_LENGTH_AT_END (1 << 3) #define ZIP_STRONG_ENCRYPTED (1 << 6) #define ZIP_UTF8_NAME (1 << 11) /* See "7.2 Single Password Symmetric Encryption Method" in http://www.pkware.com/documents/casestudies/APPNOTE.TXT */ #define ZIP_CENTRAL_DIRECTORY_ENCRYPTED (1 << 13) /* Bits used in flags. */ #define LA_USED_ZIP64 (1 << 0) #define LA_FROM_CENTRAL_DIRECTORY (1 << 1) /* * See "WinZip - AES Encryption Information" * http://www.winzip.com/aes_info.htm */ /* Value used in compression method. */ #define WINZIP_AES_ENCRYPTION 99 /* Authentication code size. */ #define AUTH_CODE_SIZE 10 /**/ #define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2) struct zip { /* Structural information about the archive. */ struct archive_string format_name; int64_t central_directory_offset; size_t central_directory_entries_total; size_t central_directory_entries_on_this_disk; int has_encrypted_entries; /* List of entries (seekable Zip only) */ struct zip_entry *zip_entries; struct archive_rb_tree tree; struct archive_rb_tree tree_rsrc; /* Bytes read but not yet consumed via __archive_read_consume() */ size_t unconsumed; /* Information about entry we're currently reading. */ struct zip_entry *entry; int64_t entry_bytes_remaining; /* These count the number of bytes actually read for the entry. */ int64_t entry_compressed_bytes_read; int64_t entry_uncompressed_bytes_read; /* Running CRC32 of the decompressed data */ unsigned long entry_crc32; unsigned long (*crc32func)(unsigned long, const void *, size_t); char ignore_crc32; /* Flags to mark progress of decompression. */ char decompress_init; char end_of_entry; #ifdef HAVE_ZLIB_H unsigned char *uncompressed_buffer; size_t uncompressed_buffer_size; z_stream stream; char stream_valid; #endif struct archive_string_conv *sconv; struct archive_string_conv *sconv_default; struct archive_string_conv *sconv_utf8; int init_default_conversion; int process_mac_extensions; char init_decryption; /* Decryption buffer. */ /* * The decrypted data starts at decrypted_ptr and * extends for decrypted_bytes_remaining. Decryption * adds new data to the end of this block, data is returned * to clients from the beginning. When the block hits the * end of decrypted_buffer, it has to be shuffled back to * the beginning of the buffer. */ unsigned char *decrypted_buffer; unsigned char *decrypted_ptr; size_t decrypted_buffer_size; size_t decrypted_bytes_remaining; size_t decrypted_unconsumed_bytes; /* Traditional PKWARE decryption. */ struct trad_enc_ctx tctx; char tctx_valid; - /* WinZip AES decyption. */ + /* WinZip AES decryption. */ /* Contexts used for AES decryption. */ archive_crypto_ctx cctx; char cctx_valid; archive_hmac_sha1_ctx hctx; char hctx_valid; /* Strong encryption's decryption header information. */ unsigned iv_size; unsigned alg_id; unsigned bit_len; unsigned flags; unsigned erd_size; unsigned v_size; unsigned v_crc32; uint8_t *iv; uint8_t *erd; uint8_t *v_data; }; /* Many systems define min or MIN, but not all. */ #define zipmin(a,b) ((a) < (b) ? (a) : (b)) /* ------------------------------------------------------------------------ */ /* Traditional PKWARE Decryption functions. */ static void trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) { uint8_t t; #define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) ctx->keys[0] = CRC32(ctx->keys[0], c); ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; t = (ctx->keys[1] >> 24) & 0xff; ctx->keys[2] = CRC32(ctx->keys[2], t); #undef CRC32 } static uint8_t -trad_enc_decypt_byte(struct trad_enc_ctx *ctx) +trad_enc_decrypt_byte(struct trad_enc_ctx *ctx) { unsigned temp = ctx->keys[2] | 2; return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; } static void trad_enc_decrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, size_t in_len, uint8_t *out, size_t out_len) { unsigned i, max; max = (unsigned)((in_len < out_len)? in_len: out_len); for (i = 0; i < max; i++) { - uint8_t t = in[i] ^ trad_enc_decypt_byte(ctx); + uint8_t t = in[i] ^ trad_enc_decrypt_byte(ctx); out[i] = t; trad_enc_update_keys(ctx, t); } } static int trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len, const uint8_t *key, size_t key_len, uint8_t *crcchk) { uint8_t header[12]; if (key_len < 12) { *crcchk = 0xff; return -1; } ctx->keys[0] = 305419896L; ctx->keys[1] = 591751049L; ctx->keys[2] = 878082192L; for (;pw_len; --pw_len) trad_enc_update_keys(ctx, *pw++); trad_enc_decrypt_update(ctx, key, 12, header, 12); /* Return the last byte for CRC check. */ *crcchk = header[11]; return 0; } #if 0 static void crypt_derive_key_sha1(const void *p, int size, unsigned char *key, int key_size) { #define MD_SIZE 20 archive_sha1_ctx ctx; unsigned char md1[MD_SIZE]; unsigned char md2[MD_SIZE * 2]; unsigned char mkb[64]; int i; archive_sha1_init(&ctx); archive_sha1_update(&ctx, p, size); archive_sha1_final(&ctx, md1); memset(mkb, 0x36, sizeof(mkb)); for (i = 0; i < MD_SIZE; i++) mkb[i] ^= md1[i]; archive_sha1_init(&ctx); archive_sha1_update(&ctx, mkb, sizeof(mkb)); archive_sha1_final(&ctx, md2); memset(mkb, 0x5C, sizeof(mkb)); for (i = 0; i < MD_SIZE; i++) mkb[i] ^= md1[i]; archive_sha1_init(&ctx); archive_sha1_update(&ctx, mkb, sizeof(mkb)); archive_sha1_final(&ctx, md2 + MD_SIZE); if (key_size > 32) key_size = 32; memcpy(key, md2, key_size); #undef MD_SIZE } #endif /* * Common code for streaming or seeking modes. * * Includes code to read local file headers, decompress data * from entry bodies, and common API. */ static unsigned long real_crc32(unsigned long crc, const void *buff, size_t len) { return crc32(crc, buff, (unsigned int)len); } /* Used by "ignorecrc32" option to speed up tests. */ static unsigned long fake_crc32(unsigned long crc, const void *buff, size_t len) { (void)crc; /* UNUSED */ (void)buff; /* UNUSED */ (void)len; /* UNUSED */ return 0; } static struct { int id; const char * name; } compression_methods[] = { {0, "uncompressed"}, /* The file is stored (no compression) */ {1, "shrinking"}, /* The file is Shrunk */ {2, "reduced-1"}, /* The file is Reduced with compression factor 1 */ {3, "reduced-2"}, /* The file is Reduced with compression factor 2 */ {4, "reduced-3"}, /* The file is Reduced with compression factor 3 */ {5, "reduced-4"}, /* The file is Reduced with compression factor 4 */ {6, "imploded"}, /* The file is Imploded */ {7, "reserved"}, /* Reserved for Tokenizing compression algorithm */ {8, "deflation"}, /* The file is Deflated */ {9, "deflation-64-bit"}, /* Enhanced Deflating using Deflate64(tm) */ {10, "ibm-terse"},/* PKWARE Data Compression Library Imploding * (old IBM TERSE) */ {11, "reserved"}, /* Reserved by PKWARE */ {12, "bzip"}, /* File is compressed using BZIP2 algorithm */ {13, "reserved"}, /* Reserved by PKWARE */ {14, "lzma"}, /* LZMA (EFS) */ {15, "reserved"}, /* Reserved by PKWARE */ {16, "reserved"}, /* Reserved by PKWARE */ {17, "reserved"}, /* Reserved by PKWARE */ {18, "ibm-terse-new"}, /* File is compressed using IBM TERSE (new) */ {19, "ibm-lz777"},/* IBM LZ77 z Architecture (PFS) */ {97, "wav-pack"}, /* WavPack compressed data */ {98, "ppmd-1"}, /* PPMd version I, Rev 1 */ {99, "aes"} /* WinZip AES encryption */ }; static const char * compression_name(const int compression) { static const int num_compression_methods = sizeof(compression_methods)/sizeof(compression_methods[0]); int i=0; while(compression >= 0 && i < num_compression_methods) { if (compression_methods[i].id == compression) return compression_methods[i].name; i++; } return "??"; } /* Convert an MSDOS-style date/time into Unix-style time. */ static time_t zip_time(const char *p) { int msTime, msDate; struct tm ts; msTime = (0xff & (unsigned)p[0]) + 256 * (0xff & (unsigned)p[1]); msDate = (0xff & (unsigned)p[2]) + 256 * (0xff & (unsigned)p[3]); memset(&ts, 0, sizeof(ts)); ts.tm_year = ((msDate >> 9) & 0x7f) + 80; /* Years since 1900. */ ts.tm_mon = ((msDate >> 5) & 0x0f) - 1; /* Month number. */ ts.tm_mday = msDate & 0x1f; /* Day of month. */ ts.tm_hour = (msTime >> 11) & 0x1f; ts.tm_min = (msTime >> 5) & 0x3f; ts.tm_sec = (msTime << 1) & 0x3e; ts.tm_isdst = -1; return mktime(&ts); } /* * The extra data is stored as a list of * id1+size1+data1 + id2+size2+data2 ... * triplets. id and size are 2 bytes each. */ static int process_extra(struct archive_read *a, const char *p, size_t extra_length, struct zip_entry* zip_entry) { unsigned offset = 0; if (extra_length == 0) { return ARCHIVE_OK; } if (extra_length < 4) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Too-small extra data: Need at least 4 bytes, but only found %d bytes", (int)extra_length); return ARCHIVE_FAILED; } while (offset <= extra_length - 4) { unsigned short headerid = archive_le16dec(p + offset); unsigned short datasize = archive_le16dec(p + offset + 2); offset += 4; if (offset + datasize > extra_length) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Extra data overflow: Need %d bytes but only found %d bytes", (int)datasize, (int)(extra_length - offset)); return ARCHIVE_FAILED; } #ifdef DEBUG fprintf(stderr, "Header id 0x%04x, length %d\n", headerid, datasize); #endif switch (headerid) { case 0x0001: /* Zip64 extended information extra field. */ zip_entry->flags |= LA_USED_ZIP64; if (zip_entry->uncompressed_size == 0xffffffff) { if (datasize < 8) break; zip_entry->uncompressed_size = archive_le64dec(p + offset); offset += 8; datasize -= 8; } if (zip_entry->compressed_size == 0xffffffff) { if (datasize < 8) break; zip_entry->compressed_size = archive_le64dec(p + offset); offset += 8; datasize -= 8; } if (zip_entry->local_header_offset == 0xffffffff) { if (datasize < 8) break; zip_entry->local_header_offset = archive_le64dec(p + offset); offset += 8; datasize -= 8; } /* archive_le32dec(p + offset) gives disk * on which file starts, but we don't handle * multi-volume Zip files. */ break; #ifdef DEBUG case 0x0017: { /* Strong encryption field. */ if (archive_le16dec(p + offset) == 2) { unsigned algId = archive_le16dec(p + offset + 2); unsigned bitLen = archive_le16dec(p + offset + 4); int flags = archive_le16dec(p + offset + 6); fprintf(stderr, "algId=0x%04x, bitLen=%u, " "flgas=%d\n", algId, bitLen,flags); } break; } #endif case 0x5455: { /* Extended time field "UT". */ int flags = p[offset]; offset++; datasize--; /* Flag bits indicate which dates are present. */ if (flags & 0x01) { #ifdef DEBUG fprintf(stderr, "mtime: %lld -> %d\n", (long long)zip_entry->mtime, archive_le32dec(p + offset)); #endif if (datasize < 4) break; zip_entry->mtime = archive_le32dec(p + offset); offset += 4; datasize -= 4; } if (flags & 0x02) { if (datasize < 4) break; zip_entry->atime = archive_le32dec(p + offset); offset += 4; datasize -= 4; } if (flags & 0x04) { if (datasize < 4) break; zip_entry->ctime = archive_le32dec(p + offset); offset += 4; datasize -= 4; } break; } case 0x5855: { /* Info-ZIP Unix Extra Field (old version) "UX". */ if (datasize >= 8) { zip_entry->atime = archive_le32dec(p + offset); zip_entry->mtime = archive_le32dec(p + offset + 4); } if (datasize >= 12) { zip_entry->uid = archive_le16dec(p + offset + 8); zip_entry->gid = archive_le16dec(p + offset + 10); } break; } case 0x6c78: { /* Experimental 'xl' field */ /* * Introduced Dec 2013 to provide a way to * include external file attributes (and other * fields that ordinarily appear only in * central directory) in local file header. * This provides file type and permission * information necessary to support full * streaming extraction. Currently being * discussed with other Zip developers * ... subject to change. * * Format: * The field starts with a bitmap that specifies * which additional fields are included. The * bitmap is variable length and can be extended in * the future. * * n bytes - feature bitmap: first byte has low-order * 7 bits. If high-order bit is set, a subsequent * byte holds the next 7 bits, etc. * * if bitmap & 1, 2 byte "version made by" * if bitmap & 2, 2 byte "internal file attributes" * if bitmap & 4, 4 byte "external file attributes" * if bitmap & 8, 2 byte comment length + n byte comment */ int bitmap, bitmap_last; if (datasize < 1) break; bitmap_last = bitmap = 0xff & p[offset]; offset += 1; datasize -= 1; /* We only support first 7 bits of bitmap; skip rest. */ while ((bitmap_last & 0x80) != 0 && datasize >= 1) { bitmap_last = p[offset]; offset += 1; datasize -= 1; } if (bitmap & 1) { /* 2 byte "version made by" */ if (datasize < 2) break; zip_entry->system = archive_le16dec(p + offset) >> 8; offset += 2; datasize -= 2; } if (bitmap & 2) { /* 2 byte "internal file attributes" */ uint32_t internal_attributes; if (datasize < 2) break; internal_attributes = archive_le16dec(p + offset); /* Not used by libarchive at present. */ (void)internal_attributes; /* UNUSED */ offset += 2; datasize -= 2; } if (bitmap & 4) { /* 4 byte "external file attributes" */ uint32_t external_attributes; if (datasize < 4) break; external_attributes = archive_le32dec(p + offset); if (zip_entry->system == 3) { zip_entry->mode = external_attributes >> 16; } else if (zip_entry->system == 0) { // Interpret MSDOS directory bit if (0x10 == (external_attributes & 0x10)) { zip_entry->mode = AE_IFDIR | 0775; } else { zip_entry->mode = AE_IFREG | 0664; } if (0x01 == (external_attributes & 0x01)) { // Read-only bit; strip write permissions zip_entry->mode &= 0555; } } else { zip_entry->mode = 0; } offset += 4; datasize -= 4; } if (bitmap & 8) { /* 2 byte comment length + comment */ uint32_t comment_length; if (datasize < 2) break; comment_length = archive_le16dec(p + offset); offset += 2; datasize -= 2; if (datasize < comment_length) break; /* Comment is not supported by libarchive */ offset += comment_length; datasize -= comment_length; } break; } case 0x7855: /* Info-ZIP Unix Extra Field (type 2) "Ux". */ #ifdef DEBUG fprintf(stderr, "uid %d gid %d\n", archive_le16dec(p + offset), archive_le16dec(p + offset + 2)); #endif if (datasize >= 2) zip_entry->uid = archive_le16dec(p + offset); if (datasize >= 4) zip_entry->gid = archive_le16dec(p + offset + 2); break; case 0x7875: { /* Info-Zip Unix Extra Field (type 3) "ux". */ int uidsize = 0, gidsize = 0; /* TODO: support arbitrary uidsize/gidsize. */ if (datasize >= 1 && p[offset] == 1) {/* version=1 */ if (datasize >= 4) { /* get a uid size. */ uidsize = 0xff & (int)p[offset+1]; if (uidsize == 2) zip_entry->uid = archive_le16dec( p + offset + 2); else if (uidsize == 4 && datasize >= 6) zip_entry->uid = archive_le32dec( p + offset + 2); } if (datasize >= (2 + uidsize + 3)) { /* get a gid size. */ gidsize = 0xff & (int)p[offset+2+uidsize]; if (gidsize == 2) zip_entry->gid = archive_le16dec( p+offset+2+uidsize+1); else if (gidsize == 4 && datasize >= (2 + uidsize + 5)) zip_entry->gid = archive_le32dec( p+offset+2+uidsize+1); } } break; } case 0x9901: - /* WinZIp AES extra data field. */ + /* WinZip AES extra data field. */ if (p[offset + 2] == 'A' && p[offset + 3] == 'E') { /* Vendor version. */ zip_entry->aes_extra.vendor = archive_le16dec(p + offset); /* AES encryption strength. */ zip_entry->aes_extra.strength = p[offset + 4]; /* Actual compression method. */ zip_entry->aes_extra.compression = p[offset + 5]; } break; default: break; } offset += datasize; } if (offset != extra_length) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Malformed extra data: Consumed %d bytes of %d bytes", (int)offset, (int)extra_length); return ARCHIVE_FAILED; } return ARCHIVE_OK; } /* * Assumes file pointer is at beginning of local file header. */ static int zip_read_local_file_header(struct archive_read *a, struct archive_entry *entry, struct zip *zip) { const char *p; const void *h; const wchar_t *wp; const char *cp; size_t len, filename_length, extra_length; struct archive_string_conv *sconv; struct zip_entry *zip_entry = zip->entry; struct zip_entry zip_entry_central_dir; int ret = ARCHIVE_OK; char version; /* Save a copy of the original for consistency checks. */ zip_entry_central_dir = *zip_entry; zip->decompress_init = 0; zip->end_of_entry = 0; zip->entry_uncompressed_bytes_read = 0; zip->entry_compressed_bytes_read = 0; zip->entry_crc32 = zip->crc32func(0, NULL, 0); /* Setup default conversion. */ if (zip->sconv == NULL && !zip->init_default_conversion) { zip->sconv_default = archive_string_default_conversion_for_read(&(a->archive)); zip->init_default_conversion = 1; } if ((p = __archive_read_ahead(a, 30, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file header"); return (ARCHIVE_FATAL); } if (memcmp(p, "PK\003\004", 4) != 0) { archive_set_error(&a->archive, -1, "Damaged Zip archive"); return ARCHIVE_FATAL; } version = p[4]; zip_entry->system = p[5]; zip_entry->zip_flags = archive_le16dec(p + 6); if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)) { zip->has_encrypted_entries = 1; archive_entry_set_is_data_encrypted(entry, 1); if (zip_entry->zip_flags & ZIP_CENTRAL_DIRECTORY_ENCRYPTED && zip_entry->zip_flags & ZIP_ENCRYPTED && zip_entry->zip_flags & ZIP_STRONG_ENCRYPTED) { archive_entry_set_is_metadata_encrypted(entry, 1); return ARCHIVE_FATAL; } } zip->init_decryption = (zip_entry->zip_flags & ZIP_ENCRYPTED); zip_entry->compression = (char)archive_le16dec(p + 8); zip_entry->mtime = zip_time(p + 10); zip_entry->crc32 = archive_le32dec(p + 14); if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) zip_entry->decdat = p[11]; else zip_entry->decdat = p[17]; zip_entry->compressed_size = archive_le32dec(p + 18); zip_entry->uncompressed_size = archive_le32dec(p + 22); filename_length = archive_le16dec(p + 26); extra_length = archive_le16dec(p + 28); __archive_read_consume(a, 30); /* Read the filename. */ if ((h = __archive_read_ahead(a, filename_length, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file header"); return (ARCHIVE_FATAL); } if (zip_entry->zip_flags & ZIP_UTF8_NAME) { /* The filename is stored to be UTF-8. */ if (zip->sconv_utf8 == NULL) { zip->sconv_utf8 = archive_string_conversion_from_charset( &a->archive, "UTF-8", 1); if (zip->sconv_utf8 == NULL) return (ARCHIVE_FATAL); } sconv = zip->sconv_utf8; } else if (zip->sconv != NULL) sconv = zip->sconv; else sconv = zip->sconv_default; if (archive_entry_copy_pathname_l(entry, h, filename_length, sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Pathname cannot be converted " "from %s to current locale.", archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; } __archive_read_consume(a, filename_length); /* Read the extra data. */ if ((h = __archive_read_ahead(a, extra_length, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file header"); return (ARCHIVE_FATAL); } if (ARCHIVE_OK != process_extra(a, h, extra_length, zip_entry)) { return ARCHIVE_FATAL; } __archive_read_consume(a, extra_length); /* Work around a bug in Info-Zip: When reading from a pipe, it * stats the pipe instead of synthesizing a file entry. */ if ((zip_entry->mode & AE_IFMT) == AE_IFIFO) { zip_entry->mode &= ~ AE_IFMT; zip_entry->mode |= AE_IFREG; } /* If the mode is totally empty, set some sane default. */ if (zip_entry->mode == 0) { zip_entry->mode |= 0664; } /* Make sure that entries with a trailing '/' are marked as directories * even if the External File Attributes contains bogus values. If this * is not a directory and there is no type, assume regularfile. */ if ((zip_entry->mode & AE_IFMT) != AE_IFDIR) { int has_slash; wp = archive_entry_pathname_w(entry); if (wp != NULL) { len = wcslen(wp); has_slash = len > 0 && wp[len - 1] == L'/'; } else { cp = archive_entry_pathname(entry); len = (cp != NULL)?strlen(cp):0; has_slash = len > 0 && cp[len - 1] == '/'; } /* Correct file type as needed. */ if (has_slash) { zip_entry->mode &= ~AE_IFMT; zip_entry->mode |= AE_IFDIR; zip_entry->mode |= 0111; } else if ((zip_entry->mode & AE_IFMT) == 0) { zip_entry->mode |= AE_IFREG; } } /* Make sure directories end in '/' */ if ((zip_entry->mode & AE_IFMT) == AE_IFDIR) { wp = archive_entry_pathname_w(entry); if (wp != NULL) { len = wcslen(wp); if (len > 0 && wp[len - 1] != L'/') { struct archive_wstring s; archive_string_init(&s); archive_wstrcat(&s, wp); archive_wstrappend_wchar(&s, L'/'); archive_entry_copy_pathname_w(entry, s.s); } } else { cp = archive_entry_pathname(entry); len = (cp != NULL)?strlen(cp):0; if (len > 0 && cp[len - 1] != '/') { struct archive_string s; archive_string_init(&s); archive_strcat(&s, cp); archive_strappend_char(&s, '/'); archive_entry_set_pathname(entry, s.s); } } } if (zip_entry->flags & LA_FROM_CENTRAL_DIRECTORY) { /* If this came from the central dir, it's size info * is definitive, so ignore the length-at-end flag. */ zip_entry->zip_flags &= ~ZIP_LENGTH_AT_END; /* If local header is missing a value, use the one from the central directory. If both have it, warn about mismatches. */ if (zip_entry->crc32 == 0) { zip_entry->crc32 = zip_entry_central_dir.crc32; } else if (!zip->ignore_crc32 && zip_entry->crc32 != zip_entry_central_dir.crc32) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Inconsistent CRC32 values"); ret = ARCHIVE_WARN; } if (zip_entry->compressed_size == 0) { zip_entry->compressed_size = zip_entry_central_dir.compressed_size; } else if (zip_entry->compressed_size != zip_entry_central_dir.compressed_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Inconsistent compressed size: " "%jd in central directory, %jd in local header", (intmax_t)zip_entry_central_dir.compressed_size, (intmax_t)zip_entry->compressed_size); ret = ARCHIVE_WARN; } if (zip_entry->uncompressed_size == 0) { zip_entry->uncompressed_size = zip_entry_central_dir.uncompressed_size; } else if (zip_entry->uncompressed_size != zip_entry_central_dir.uncompressed_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Inconsistent uncompressed size: " "%jd in central directory, %jd in local header", (intmax_t)zip_entry_central_dir.uncompressed_size, (intmax_t)zip_entry->uncompressed_size); ret = ARCHIVE_WARN; } } /* Populate some additional entry fields: */ archive_entry_set_mode(entry, zip_entry->mode); archive_entry_set_uid(entry, zip_entry->uid); archive_entry_set_gid(entry, zip_entry->gid); archive_entry_set_mtime(entry, zip_entry->mtime, 0); archive_entry_set_ctime(entry, zip_entry->ctime, 0); archive_entry_set_atime(entry, zip_entry->atime, 0); if ((zip->entry->mode & AE_IFMT) == AE_IFLNK) { size_t linkname_length; if (zip_entry->compressed_size > 64 * 1024) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Zip file with oversized link entry"); return ARCHIVE_FATAL; } linkname_length = (size_t)zip_entry->compressed_size; archive_entry_set_size(entry, 0); p = __archive_read_ahead(a, linkname_length, NULL); if (p == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Truncated Zip file"); return ARCHIVE_FATAL; } sconv = zip->sconv; if (sconv == NULL && (zip->entry->zip_flags & ZIP_UTF8_NAME)) sconv = zip->sconv_utf8; if (sconv == NULL) sconv = zip->sconv_default; if (archive_entry_copy_symlink_l(entry, p, linkname_length, sconv) != 0) { if (errno != ENOMEM && sconv == zip->sconv_utf8 && (zip->entry->zip_flags & ZIP_UTF8_NAME)) archive_entry_copy_symlink_l(entry, p, linkname_length, NULL); if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Symlink"); return (ARCHIVE_FATAL); } /* * Since there is no character-set regulation for * symlink name, do not report the conversion error * in an automatic conversion. */ if (sconv != zip->sconv_utf8 || (zip->entry->zip_flags & ZIP_UTF8_NAME) == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Symlink cannot be converted " "from %s to current locale.", archive_string_conversion_charset_name( sconv)); ret = ARCHIVE_WARN; } } zip_entry->uncompressed_size = zip_entry->compressed_size = 0; if (__archive_read_consume(a, linkname_length) < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Read error skipping symlink target name"); return ARCHIVE_FATAL; } } else if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) || zip_entry->uncompressed_size > 0) { /* Set the size only if it's meaningful. */ archive_entry_set_size(entry, zip_entry->uncompressed_size); } zip->entry_bytes_remaining = zip_entry->compressed_size; /* If there's no body, force read_data() to return EOF immediately. */ if (0 == (zip_entry->zip_flags & ZIP_LENGTH_AT_END) && zip->entry_bytes_remaining < 1) zip->end_of_entry = 1; /* Set up a more descriptive format name. */ archive_string_sprintf(&zip->format_name, "ZIP %d.%d (%s)", version / 10, version % 10, compression_name(zip->entry->compression)); a->archive.archive_format_name = zip->format_name.s; return (ret); } static int check_authentication_code(struct archive_read *a, const void *_p) { struct zip *zip = (struct zip *)(a->format->data); /* Check authentication code. */ if (zip->hctx_valid) { const void *p; uint8_t hmac[20]; size_t hmac_len = 20; int cmp; archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); if (_p == NULL) { /* Read authentication code. */ p = __archive_read_ahead(a, AUTH_CODE_SIZE, NULL); if (p == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file data"); return (ARCHIVE_FATAL); } } else { p = _p; } cmp = memcmp(hmac, p, AUTH_CODE_SIZE); __archive_read_consume(a, AUTH_CODE_SIZE); if (cmp != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "ZIP bad Authentication code"); return (ARCHIVE_WARN); } } return (ARCHIVE_OK); } /* * Read "uncompressed" data. There are three cases: * 1) We know the size of the data. This is always true for the * seeking reader (we've examined the Central Directory already). * 2) ZIP_LENGTH_AT_END was set, but only the CRC was deferred. * Info-ZIP seems to do this; we know the size but have to grab * the CRC from the data descriptor afterwards. * 3) We're streaming and ZIP_LENGTH_AT_END was specified and * we have no size information. In this case, we can do pretty * well by watching for the data descriptor record. The data * descriptor is 16 bytes and includes a computed CRC that should * provide a strong check. * * TODO: Technically, the PK\007\010 signature is optional. * In the original spec, the data descriptor contained CRC * and size fields but had no leading signature. In practice, * newer writers seem to provide the signature pretty consistently. * * For uncompressed data, the PK\007\010 marker seems essential * to be sure we've actually seen the end of the entry. * * Returns ARCHIVE_OK if successful, ARCHIVE_FATAL otherwise, sets * zip->end_of_entry if it consumes all of the data. */ static int zip_read_data_none(struct archive_read *a, const void **_buff, size_t *size, int64_t *offset) { struct zip *zip; const char *buff; ssize_t bytes_avail; int r; (void)offset; /* UNUSED */ zip = (struct zip *)(a->format->data); if (zip->entry->zip_flags & ZIP_LENGTH_AT_END) { const char *p; ssize_t grabbing_bytes = 24; if (zip->hctx_valid) grabbing_bytes += AUTH_CODE_SIZE; /* Grab at least 24 bytes. */ buff = __archive_read_ahead(a, grabbing_bytes, &bytes_avail); if (bytes_avail < grabbing_bytes) { /* Zip archives have end-of-archive markers that are longer than this, so a failure to get at least 24 bytes really does indicate a truncated file. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file data"); return (ARCHIVE_FATAL); } /* Check for a complete PK\007\010 signature, followed * by the correct 4-byte CRC. */ p = buff; if (zip->hctx_valid) p += AUTH_CODE_SIZE; if (p[0] == 'P' && p[1] == 'K' && p[2] == '\007' && p[3] == '\010' && (archive_le32dec(p + 4) == zip->entry_crc32 || zip->ignore_crc32 || (zip->hctx_valid && zip->entry->aes_extra.vendor == AES_VENDOR_AE_2))) { if (zip->entry->flags & LA_USED_ZIP64) { zip->entry->crc32 = archive_le32dec(p + 4); zip->entry->compressed_size = archive_le64dec(p + 8); zip->entry->uncompressed_size = archive_le64dec(p + 16); zip->unconsumed = 24; } else { zip->entry->crc32 = archive_le32dec(p + 4); zip->entry->compressed_size = archive_le32dec(p + 8); zip->entry->uncompressed_size = archive_le32dec(p + 12); zip->unconsumed = 16; } if (zip->hctx_valid) { r = check_authentication_code(a, buff); if (r != ARCHIVE_OK) return (r); } zip->end_of_entry = 1; return (ARCHIVE_OK); } /* If not at EOF, ensure we consume at least one byte. */ ++p; /* Scan forward until we see where a PK\007\010 signature * might be. */ /* Return bytes up until that point. On the next call, * the code above will verify the data descriptor. */ while (p < buff + bytes_avail - 4) { if (p[3] == 'P') { p += 3; } else if (p[3] == 'K') { p += 2; } else if (p[3] == '\007') { p += 1; } else if (p[3] == '\010' && p[2] == '\007' && p[1] == 'K' && p[0] == 'P') { if (zip->hctx_valid) p -= AUTH_CODE_SIZE; break; } else { p += 4; } } bytes_avail = p - buff; } else { if (zip->entry_bytes_remaining == 0) { zip->end_of_entry = 1; if (zip->hctx_valid) { r = check_authentication_code(a, NULL); if (r != ARCHIVE_OK) return (r); } return (ARCHIVE_OK); } /* Grab a bunch of bytes. */ buff = __archive_read_ahead(a, 1, &bytes_avail); if (bytes_avail <= 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file data"); return (ARCHIVE_FATAL); } if (bytes_avail > zip->entry_bytes_remaining) bytes_avail = (ssize_t)zip->entry_bytes_remaining; } if (zip->tctx_valid || zip->cctx_valid) { size_t dec_size = bytes_avail; if (dec_size > zip->decrypted_buffer_size) dec_size = zip->decrypted_buffer_size; if (zip->tctx_valid) { trad_enc_decrypt_update(&zip->tctx, (const uint8_t *)buff, dec_size, zip->decrypted_buffer, dec_size); } else { size_t dsize = dec_size; archive_hmac_sha1_update(&zip->hctx, (const uint8_t *)buff, dec_size); archive_decrypto_aes_ctr_update(&zip->cctx, (const uint8_t *)buff, dec_size, zip->decrypted_buffer, &dsize); } bytes_avail = dec_size; buff = (const char *)zip->decrypted_buffer; } *size = bytes_avail; zip->entry_bytes_remaining -= bytes_avail; zip->entry_uncompressed_bytes_read += bytes_avail; zip->entry_compressed_bytes_read += bytes_avail; zip->unconsumed += bytes_avail; *_buff = buff; return (ARCHIVE_OK); } #ifdef HAVE_ZLIB_H static int zip_deflate_init(struct archive_read *a, struct zip *zip) { int r; /* If we haven't yet read any data, initialize the decompressor. */ if (!zip->decompress_init) { if (zip->stream_valid) r = inflateReset(&zip->stream); else r = inflateInit2(&zip->stream, -15 /* Don't check for zlib header */); if (r != Z_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't initialize ZIP decompression."); return (ARCHIVE_FATAL); } /* Stream structure has been set up. */ zip->stream_valid = 1; /* We've initialized decompression for this stream. */ zip->decompress_init = 1; } return (ARCHIVE_OK); } static int zip_read_data_deflate(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { struct zip *zip; ssize_t bytes_avail; const void *compressed_buff, *sp; int r; (void)offset; /* UNUSED */ zip = (struct zip *)(a->format->data); /* If the buffer hasn't been allocated, allocate it now. */ if (zip->uncompressed_buffer == NULL) { zip->uncompressed_buffer_size = 256 * 1024; zip->uncompressed_buffer = (unsigned char *)malloc(zip->uncompressed_buffer_size); if (zip->uncompressed_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for ZIP decompression"); return (ARCHIVE_FATAL); } } r = zip_deflate_init(a, zip); if (r != ARCHIVE_OK) return (r); /* * Note: '1' here is a performance optimization. * Recall that the decompression layer returns a count of * available bytes; asking for more than that forces the * decompressor to combine reads by copying data. */ compressed_buff = sp = __archive_read_ahead(a, 1, &bytes_avail); if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && bytes_avail > zip->entry_bytes_remaining) { bytes_avail = (ssize_t)zip->entry_bytes_remaining; } if (bytes_avail < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file body"); return (ARCHIVE_FATAL); } if (zip->tctx_valid || zip->cctx_valid) { if (zip->decrypted_bytes_remaining < (size_t)bytes_avail) { size_t buff_remaining = (zip->decrypted_buffer + zip->decrypted_buffer_size) - (zip->decrypted_ptr + zip->decrypted_bytes_remaining); if (buff_remaining > (size_t)bytes_avail) buff_remaining = (size_t)bytes_avail; if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && zip->entry_bytes_remaining > 0) { if ((int64_t)(zip->decrypted_bytes_remaining + buff_remaining) > zip->entry_bytes_remaining) { if (zip->entry_bytes_remaining < (int64_t)zip->decrypted_bytes_remaining) buff_remaining = 0; else buff_remaining = (size_t)zip->entry_bytes_remaining - zip->decrypted_bytes_remaining; } } if (buff_remaining > 0) { if (zip->tctx_valid) { trad_enc_decrypt_update(&zip->tctx, compressed_buff, buff_remaining, zip->decrypted_ptr + zip->decrypted_bytes_remaining, buff_remaining); } else { size_t dsize = buff_remaining; archive_decrypto_aes_ctr_update( &zip->cctx, compressed_buff, buff_remaining, zip->decrypted_ptr + zip->decrypted_bytes_remaining, &dsize); } zip->decrypted_bytes_remaining += buff_remaining; } } bytes_avail = zip->decrypted_bytes_remaining; compressed_buff = (const char *)zip->decrypted_ptr; } /* * A bug in zlib.h: stream.next_in should be marked 'const' * but isn't (the library never alters data through the * next_in pointer, only reads it). The result: this ugly * cast to remove 'const'. */ zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)compressed_buff; zip->stream.avail_in = (uInt)bytes_avail; zip->stream.total_in = 0; zip->stream.next_out = zip->uncompressed_buffer; zip->stream.avail_out = (uInt)zip->uncompressed_buffer_size; zip->stream.total_out = 0; r = inflate(&zip->stream, 0); switch (r) { case Z_OK: break; case Z_STREAM_END: zip->end_of_entry = 1; break; case Z_MEM_ERROR: archive_set_error(&a->archive, ENOMEM, "Out of memory for ZIP decompression"); return (ARCHIVE_FATAL); default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "ZIP decompression failed (%d)", r); return (ARCHIVE_FATAL); } /* Consume as much as the compressor actually used. */ bytes_avail = zip->stream.total_in; if (zip->tctx_valid || zip->cctx_valid) { zip->decrypted_bytes_remaining -= bytes_avail; if (zip->decrypted_bytes_remaining == 0) zip->decrypted_ptr = zip->decrypted_buffer; else zip->decrypted_ptr += bytes_avail; } /* Calculate compressed data as much as we used.*/ if (zip->hctx_valid) archive_hmac_sha1_update(&zip->hctx, sp, bytes_avail); __archive_read_consume(a, bytes_avail); zip->entry_bytes_remaining -= bytes_avail; zip->entry_compressed_bytes_read += bytes_avail; *size = zip->stream.total_out; zip->entry_uncompressed_bytes_read += zip->stream.total_out; *buff = zip->uncompressed_buffer; if (zip->end_of_entry && zip->hctx_valid) { r = check_authentication_code(a, NULL); if (r != ARCHIVE_OK) return (r); } if (zip->end_of_entry && (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { const char *p; if (NULL == (p = __archive_read_ahead(a, 24, NULL))) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP end-of-file record"); return (ARCHIVE_FATAL); } /* Consume the optional PK\007\010 marker. */ if (p[0] == 'P' && p[1] == 'K' && p[2] == '\007' && p[3] == '\010') { p += 4; zip->unconsumed = 4; } if (zip->entry->flags & LA_USED_ZIP64) { zip->entry->crc32 = archive_le32dec(p); zip->entry->compressed_size = archive_le64dec(p + 4); zip->entry->uncompressed_size = archive_le64dec(p + 12); zip->unconsumed += 20; } else { zip->entry->crc32 = archive_le32dec(p); zip->entry->compressed_size = archive_le32dec(p + 4); zip->entry->uncompressed_size = archive_le32dec(p + 8); zip->unconsumed += 12; } } return (ARCHIVE_OK); } #endif static int read_decryption_header(struct archive_read *a) { struct zip *zip = (struct zip *)(a->format->data); const char *p; unsigned int remaining_size; unsigned int ts; /* * Read an initialization vector data field. */ p = __archive_read_ahead(a, 2, NULL); if (p == NULL) goto truncated; ts = zip->iv_size; zip->iv_size = archive_le16dec(p); __archive_read_consume(a, 2); if (ts < zip->iv_size) { free(zip->iv); zip->iv = NULL; } p = __archive_read_ahead(a, zip->iv_size, NULL); if (p == NULL) goto truncated; if (zip->iv == NULL) { zip->iv = malloc(zip->iv_size); if (zip->iv == NULL) goto nomem; } memcpy(zip->iv, p, zip->iv_size); __archive_read_consume(a, zip->iv_size); /* * Read a size of remaining decryption header field. */ p = __archive_read_ahead(a, 14, NULL); if (p == NULL) goto truncated; remaining_size = archive_le32dec(p); if (remaining_size < 16 || remaining_size > (1 << 18)) goto corrupted; /* Check if format version is supported. */ if (archive_le16dec(p+4) != 3) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported encryption format version: %u", archive_le16dec(p+4)); return (ARCHIVE_FAILED); } /* * Read an encryption algorithm field. */ zip->alg_id = archive_le16dec(p+6); switch (zip->alg_id) { case 0x6601:/* DES */ case 0x6602:/* RC2 */ case 0x6603:/* 3DES 168 */ case 0x6609:/* 3DES 112 */ case 0x660E:/* AES 128 */ case 0x660F:/* AES 192 */ case 0x6610:/* AES 256 */ case 0x6702:/* RC2 (version >= 5.2) */ case 0x6720:/* Blowfish */ case 0x6721:/* Twofish */ case 0x6801:/* RC4 */ - /* Suuported encryption algorithm. */ + /* Supported encryption algorithm. */ break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unknown encryption algorithm: %u", zip->alg_id); return (ARCHIVE_FAILED); } /* * Read a bit length field. */ zip->bit_len = archive_le16dec(p+8); /* * Read a flags field. */ zip->flags = archive_le16dec(p+10); switch (zip->flags & 0xf000) { case 0x0001: /* Password is required to decrypt. */ case 0x0002: /* Certificates only. */ case 0x0003: /* Password or certificate required to decrypt. */ break; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unknown encryption flag: %u", zip->flags); return (ARCHIVE_FAILED); } if ((zip->flags & 0xf000) == 0 || (zip->flags & 0xf000) == 0x4000) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unknown encryption flag: %u", zip->flags); return (ARCHIVE_FAILED); } /* * Read an encrypted random data field. */ ts = zip->erd_size; zip->erd_size = archive_le16dec(p+12); __archive_read_consume(a, 14); if ((zip->erd_size & 0xf) != 0 || (zip->erd_size + 16) > remaining_size || (zip->erd_size + 16) < zip->erd_size) goto corrupted; if (ts < zip->erd_size) { free(zip->erd); zip->erd = NULL; } p = __archive_read_ahead(a, zip->erd_size, NULL); if (p == NULL) goto truncated; if (zip->erd == NULL) { zip->erd = malloc(zip->erd_size); if (zip->erd == NULL) goto nomem; } memcpy(zip->erd, p, zip->erd_size); __archive_read_consume(a, zip->erd_size); /* * Read a reserved data field. */ p = __archive_read_ahead(a, 4, NULL); if (p == NULL) goto truncated; /* Reserved data size should be zero. */ if (archive_le32dec(p) != 0) goto corrupted; __archive_read_consume(a, 4); /* * Read a password validation data field. */ p = __archive_read_ahead(a, 2, NULL); if (p == NULL) goto truncated; ts = zip->v_size; zip->v_size = archive_le16dec(p); __archive_read_consume(a, 2); if ((zip->v_size & 0x0f) != 0 || (zip->erd_size + zip->v_size + 16) > remaining_size || (zip->erd_size + zip->v_size + 16) < (zip->erd_size + zip->v_size)) goto corrupted; if (ts < zip->v_size) { free(zip->v_data); zip->v_data = NULL; } p = __archive_read_ahead(a, zip->v_size, NULL); if (p == NULL) goto truncated; if (zip->v_data == NULL) { zip->v_data = malloc(zip->v_size); if (zip->v_data == NULL) goto nomem; } memcpy(zip->v_data, p, zip->v_size); __archive_read_consume(a, zip->v_size); p = __archive_read_ahead(a, 4, NULL); if (p == NULL) goto truncated; zip->v_crc32 = archive_le32dec(p); __archive_read_consume(a, 4); /*return (ARCHIVE_OK); - * This is not fully implemnted yet.*/ + * This is not fully implemented yet.*/ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Encrypted file is unsupported"); return (ARCHIVE_FAILED); truncated: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file data"); return (ARCHIVE_FATAL); corrupted: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Corrupted ZIP file data"); return (ARCHIVE_FATAL); nomem: archive_set_error(&a->archive, ENOMEM, "No memory for ZIP decryption"); return (ARCHIVE_FATAL); } static int zip_alloc_decryption_buffer(struct archive_read *a) { struct zip *zip = (struct zip *)(a->format->data); size_t bs = 256 * 1024; if (zip->decrypted_buffer == NULL) { zip->decrypted_buffer_size = bs; zip->decrypted_buffer = malloc(bs); if (zip->decrypted_buffer == NULL) { archive_set_error(&a->archive, ENOMEM, "No memory for ZIP decryption"); return (ARCHIVE_FATAL); } } zip->decrypted_ptr = zip->decrypted_buffer; return (ARCHIVE_OK); } static int init_traditional_PKWARE_decryption(struct archive_read *a) { struct zip *zip = (struct zip *)(a->format->data); const void *p; int retry; int r; if (zip->tctx_valid) return (ARCHIVE_OK); /* Read the 12 bytes encryption header stored at the start of the data area. */ #define ENC_HEADER_SIZE 12 if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && zip->entry_bytes_remaining < ENC_HEADER_SIZE) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated Zip encrypted body: only %jd bytes available", (intmax_t)zip->entry_bytes_remaining); return (ARCHIVE_FATAL); } p = __archive_read_ahead(a, ENC_HEADER_SIZE, NULL); if (p == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file data"); return (ARCHIVE_FATAL); } for (retry = 0;; retry++) { const char *passphrase; uint8_t crcchk; passphrase = __archive_read_next_passphrase(a); if (passphrase == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, (retry > 0)? "Incorrect passphrase": "Passphrase required for this entry"); return (ARCHIVE_FAILED); } /* - * Initialize ctx for Traditional PKWARE Decyption. + * Initialize ctx for Traditional PKWARE Decryption. */ r = trad_enc_init(&zip->tctx, passphrase, strlen(passphrase), p, ENC_HEADER_SIZE, &crcchk); if (r == 0 && crcchk == zip->entry->decdat) break;/* The passphrase is OK. */ if (retry > 10000) { /* Avoid infinity loop. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Too many incorrect passphrases"); return (ARCHIVE_FAILED); } } __archive_read_consume(a, ENC_HEADER_SIZE); zip->tctx_valid = 1; if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END)) { zip->entry_bytes_remaining -= ENC_HEADER_SIZE; } /*zip->entry_uncompressed_bytes_read += ENC_HEADER_SIZE;*/ zip->entry_compressed_bytes_read += ENC_HEADER_SIZE; zip->decrypted_bytes_remaining = 0; return (zip_alloc_decryption_buffer(a)); #undef ENC_HEADER_SIZE } static int init_WinZip_AES_decryption(struct archive_read *a) { struct zip *zip = (struct zip *)(a->format->data); const void *p; const uint8_t *pv; size_t key_len, salt_len; uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; int retry; int r; if (zip->cctx_valid || zip->hctx_valid) return (ARCHIVE_OK); switch (zip->entry->aes_extra.strength) { case 1: salt_len = 8; key_len = 16; break; case 2: salt_len = 12; key_len = 24; break; case 3: salt_len = 16; key_len = 32; break; default: goto corrupted; } p = __archive_read_ahead(a, salt_len + 2, NULL); if (p == NULL) goto truncated; for (retry = 0;; retry++) { const char *passphrase; passphrase = __archive_read_next_passphrase(a); if (passphrase == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, (retry > 0)? "Incorrect passphrase": "Passphrase required for this entry"); return (ARCHIVE_FAILED); } memset(derived_key, 0, sizeof(derived_key)); r = archive_pbkdf2_sha1(passphrase, strlen(passphrase), p, salt_len, 1000, derived_key, key_len * 2 + 2); if (r != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Decryption is unsupported due to lack of " "crypto library"); return (ARCHIVE_FAILED); } /* Check password verification value. */ pv = ((const uint8_t *)p) + salt_len; if (derived_key[key_len * 2] == pv[0] && derived_key[key_len * 2 + 1] == pv[1]) break;/* The passphrase is OK. */ if (retry > 10000) { /* Avoid infinity loop. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Too many incorrect passphrases"); return (ARCHIVE_FAILED); } } r = archive_decrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); if (r != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Decryption is unsupported due to lack of crypto library"); return (ARCHIVE_FAILED); } r = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); if (r != 0) { archive_decrypto_aes_ctr_release(&zip->cctx); archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to initialize HMAC-SHA1"); return (ARCHIVE_FAILED); } zip->cctx_valid = zip->hctx_valid = 1; __archive_read_consume(a, salt_len + 2); zip->entry_bytes_remaining -= salt_len + 2 + AUTH_CODE_SIZE; if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) && zip->entry_bytes_remaining < 0) goto corrupted; zip->entry_compressed_bytes_read += salt_len + 2 + AUTH_CODE_SIZE; zip->decrypted_bytes_remaining = 0; zip->entry->compression = zip->entry->aes_extra.compression; return (zip_alloc_decryption_buffer(a)); truncated: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file data"); return (ARCHIVE_FATAL); corrupted: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Corrupted ZIP file data"); return (ARCHIVE_FATAL); } static int archive_read_format_zip_read_data(struct archive_read *a, const void **buff, size_t *size, int64_t *offset) { int r; struct zip *zip = (struct zip *)(a->format->data); if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) { zip->has_encrypted_entries = 0; } *offset = zip->entry_uncompressed_bytes_read; *size = 0; *buff = NULL; /* If we hit end-of-entry last time, return ARCHIVE_EOF. */ if (zip->end_of_entry) return (ARCHIVE_EOF); /* Return EOF immediately if this is a non-regular file. */ if (AE_IFREG != (zip->entry->mode & AE_IFMT)) return (ARCHIVE_EOF); __archive_read_consume(a, zip->unconsumed); zip->unconsumed = 0; if (zip->init_decryption) { zip->has_encrypted_entries = 1; if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) r = read_decryption_header(a); else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) r = init_WinZip_AES_decryption(a); else r = init_traditional_PKWARE_decryption(a); if (r != ARCHIVE_OK) return (r); zip->init_decryption = 0; } switch(zip->entry->compression) { case 0: /* No compression. */ r = zip_read_data_none(a, buff, size, offset); break; #ifdef HAVE_ZLIB_H case 8: /* Deflate compression. */ r = zip_read_data_deflate(a, buff, size, offset); break; #endif default: /* Unsupported compression. */ /* Return a warning. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported ZIP compression method (%s)", compression_name(zip->entry->compression)); /* We can't decompress this entry, but we will * be able to skip() it and try the next entry. */ return (ARCHIVE_FAILED); break; } if (r != ARCHIVE_OK) return (r); /* Update checksum */ if (*size) zip->entry_crc32 = zip->crc32func(zip->entry_crc32, *buff, (unsigned)*size); /* If we hit the end, swallow any end-of-data marker. */ if (zip->end_of_entry) { /* Check file size, CRC against these values. */ if (zip->entry->compressed_size != zip->entry_compressed_bytes_read) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "ZIP compressed data is wrong size " "(read %jd, expected %jd)", (intmax_t)zip->entry_compressed_bytes_read, (intmax_t)zip->entry->compressed_size); return (ARCHIVE_WARN); } /* Size field only stores the lower 32 bits of the actual * size. */ if ((zip->entry->uncompressed_size & UINT32_MAX) != (zip->entry_uncompressed_bytes_read & UINT32_MAX)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "ZIP uncompressed data is wrong size " "(read %jd, expected %jd)\n", (intmax_t)zip->entry_uncompressed_bytes_read, (intmax_t)zip->entry->uncompressed_size); return (ARCHIVE_WARN); } /* Check computed CRC against header */ if ((!zip->hctx_valid || zip->entry->aes_extra.vendor != AES_VENDOR_AE_2) && zip->entry->crc32 != zip->entry_crc32 && !zip->ignore_crc32) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "ZIP bad CRC: 0x%lx should be 0x%lx", (unsigned long)zip->entry_crc32, (unsigned long)zip->entry->crc32); return (ARCHIVE_WARN); } } return (ARCHIVE_OK); } static int archive_read_format_zip_cleanup(struct archive_read *a) { struct zip *zip; struct zip_entry *zip_entry, *next_zip_entry; zip = (struct zip *)(a->format->data); #ifdef HAVE_ZLIB_H if (zip->stream_valid) inflateEnd(&zip->stream); free(zip->uncompressed_buffer); #endif if (zip->zip_entries) { zip_entry = zip->zip_entries; while (zip_entry != NULL) { next_zip_entry = zip_entry->next; archive_string_free(&zip_entry->rsrcname); free(zip_entry); zip_entry = next_zip_entry; } } free(zip->decrypted_buffer); if (zip->cctx_valid) archive_decrypto_aes_ctr_release(&zip->cctx); if (zip->hctx_valid) archive_hmac_sha1_cleanup(&zip->hctx); free(zip->iv); free(zip->erd); free(zip->v_data); archive_string_free(&zip->format_name); free(zip); (a->format->data) = NULL; return (ARCHIVE_OK); } static int archive_read_format_zip_has_encrypted_entries(struct archive_read *_a) { if (_a && _a->format) { struct zip * zip = (struct zip *)_a->format->data; if (zip) { return zip->has_encrypted_entries; } } return ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; } static int archive_read_format_zip_options(struct archive_read *a, const char *key, const char *val) { struct zip *zip; int ret = ARCHIVE_FAILED; zip = (struct zip *)(a->format->data); if (strcmp(key, "compat-2x") == 0) { /* Handle filenames as libarchive 2.x */ zip->init_default_conversion = (val != NULL) ? 1 : 0; return (ARCHIVE_OK); } else if (strcmp(key, "hdrcharset") == 0) { if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "zip: hdrcharset option needs a character-set name" ); else { zip->sconv = archive_string_conversion_from_charset( &a->archive, val, 0); if (zip->sconv != NULL) { if (strcmp(val, "UTF-8") == 0) zip->sconv_utf8 = zip->sconv; ret = ARCHIVE_OK; } else ret = ARCHIVE_FATAL; } return (ret); } else if (strcmp(key, "ignorecrc32") == 0) { /* Mostly useful for testing. */ if (val == NULL || val[0] == 0) { zip->crc32func = real_crc32; zip->ignore_crc32 = 0; } else { zip->crc32func = fake_crc32; zip->ignore_crc32 = 1; } return (ARCHIVE_OK); } else if (strcmp(key, "mac-ext") == 0) { zip->process_mac_extensions = (val != NULL && val[0] != 0); return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } int archive_read_support_format_zip(struct archive *a) { int r; r = archive_read_support_format_zip_streamable(a); if (r != ARCHIVE_OK) return r; return (archive_read_support_format_zip_seekable(a)); } /* ------------------------------------------------------------------------ */ /* * Streaming-mode support */ static int archive_read_support_format_zip_capabilities_streamable(struct archive_read * a) { (void)a; /* UNUSED */ return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); } static int archive_read_format_zip_streamable_bid(struct archive_read *a, int best_bid) { const char *p; (void)best_bid; /* UNUSED */ if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) return (-1); /* * Bid of 29 here comes from: * + 16 bits for "PK", * + next 16-bit field has 6 options so contributes * about 16 - log_2(6) ~= 16 - 2.6 ~= 13 bits * * So we've effectively verified ~29 total bits of check data. */ if (p[0] == 'P' && p[1] == 'K') { if ((p[2] == '\001' && p[3] == '\002') || (p[2] == '\003' && p[3] == '\004') || (p[2] == '\005' && p[3] == '\006') || (p[2] == '\006' && p[3] == '\006') || (p[2] == '\007' && p[3] == '\010') || (p[2] == '0' && p[3] == '0')) return (29); } /* TODO: It's worth looking ahead a little bit for a valid * PK signature. In particular, that would make it possible * to read some UUEncoded SFX files or SFX files coming from * a network socket. */ return (0); } static int archive_read_format_zip_streamable_read_header(struct archive_read *a, struct archive_entry *entry) { struct zip *zip; a->archive.archive_format = ARCHIVE_FORMAT_ZIP; if (a->archive.archive_format_name == NULL) a->archive.archive_format_name = "ZIP"; zip = (struct zip *)(a->format->data); /* * It should be sufficient to call archive_read_next_header() for * a reader to determine if an entry is encrypted or not. If the * encryption of an entry is only detectable when calling * archive_read_data(), so be it. We'll do the same check there * as well. */ if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) zip->has_encrypted_entries = 0; /* Make sure we have a zip_entry structure to use. */ if (zip->zip_entries == NULL) { zip->zip_entries = malloc(sizeof(struct zip_entry)); if (zip->zip_entries == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return ARCHIVE_FATAL; } } zip->entry = zip->zip_entries; memset(zip->entry, 0, sizeof(struct zip_entry)); if (zip->cctx_valid) archive_decrypto_aes_ctr_release(&zip->cctx); if (zip->hctx_valid) archive_hmac_sha1_cleanup(&zip->hctx); zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; __archive_read_reset_passphrase(a); /* Search ahead for the next local file header. */ __archive_read_consume(a, zip->unconsumed); zip->unconsumed = 0; for (;;) { int64_t skipped = 0; const char *p, *end; ssize_t bytes; p = __archive_read_ahead(a, 4, &bytes); if (p == NULL) return (ARCHIVE_FATAL); end = p + bytes; while (p + 4 <= end) { if (p[0] == 'P' && p[1] == 'K') { if (p[2] == '\003' && p[3] == '\004') { /* Regular file entry. */ __archive_read_consume(a, skipped); return zip_read_local_file_header(a, entry, zip); } /* * TODO: We cannot restore permissions * based only on the local file headers. * Consider scanning the central * directory and returning additional * entries for at least directories. * This would allow us to properly set * directory permissions. * * This won't help us fix symlinks * and may not help with regular file * permissions, either. */ if (p[2] == '\001' && p[3] == '\002') { return (ARCHIVE_EOF); } /* End of central directory? Must be an * empty archive. */ if ((p[2] == '\005' && p[3] == '\006') || (p[2] == '\006' && p[3] == '\006')) return (ARCHIVE_EOF); } ++p; ++skipped; } __archive_read_consume(a, skipped); } } static int archive_read_format_zip_read_data_skip_streamable(struct archive_read *a) { struct zip *zip; int64_t bytes_skipped; zip = (struct zip *)(a->format->data); bytes_skipped = __archive_read_consume(a, zip->unconsumed); zip->unconsumed = 0; if (bytes_skipped < 0) return (ARCHIVE_FATAL); /* If we've already read to end of data, we're done. */ if (zip->end_of_entry) return (ARCHIVE_OK); /* So we know we're streaming... */ if (0 == (zip->entry->zip_flags & ZIP_LENGTH_AT_END) || zip->entry->compressed_size > 0) { /* We know the compressed length, so we can just skip. */ bytes_skipped = __archive_read_consume(a, zip->entry_bytes_remaining); if (bytes_skipped < 0) return (ARCHIVE_FATAL); return (ARCHIVE_OK); } if (zip->init_decryption) { int r; zip->has_encrypted_entries = 1; if (zip->entry->zip_flags & ZIP_STRONG_ENCRYPTED) r = read_decryption_header(a); else if (zip->entry->compression == WINZIP_AES_ENCRYPTION) r = init_WinZip_AES_decryption(a); else r = init_traditional_PKWARE_decryption(a); if (r != ARCHIVE_OK) return (r); zip->init_decryption = 0; } /* We're streaming and we don't know the length. */ /* If the body is compressed and we know the format, we can * find an exact end-of-entry by decompressing it. */ switch (zip->entry->compression) { #ifdef HAVE_ZLIB_H case 8: /* Deflate compression. */ while (!zip->end_of_entry) { int64_t offset = 0; const void *buff = NULL; size_t size = 0; int r; r = zip_read_data_deflate(a, &buff, &size, &offset); if (r != ARCHIVE_OK) return (r); } return ARCHIVE_OK; #endif default: /* Uncompressed or unknown. */ /* Scan for a PK\007\010 signature. */ for (;;) { const char *p, *buff; ssize_t bytes_avail; buff = __archive_read_ahead(a, 16, &bytes_avail); if (bytes_avail < 16) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file data"); return (ARCHIVE_FATAL); } p = buff; while (p <= buff + bytes_avail - 16) { if (p[3] == 'P') { p += 3; } else if (p[3] == 'K') { p += 2; } else if (p[3] == '\007') { p += 1; } else if (p[3] == '\010' && p[2] == '\007' && p[1] == 'K' && p[0] == 'P') { if (zip->entry->flags & LA_USED_ZIP64) __archive_read_consume(a, p - buff + 24); else __archive_read_consume(a, p - buff + 16); return ARCHIVE_OK; } else { p += 4; } } __archive_read_consume(a, p - buff); } } } int archive_read_support_format_zip_streamable(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct zip *zip; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_zip"); zip = (struct zip *)calloc(1, sizeof(*zip)); if (zip == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip data"); return (ARCHIVE_FATAL); } /* Streamable reader doesn't support mac extensions. */ zip->process_mac_extensions = 0; /* * Until enough data has been read, we cannot tell about * any encrypted entries yet. */ zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; zip->crc32func = real_crc32; r = __archive_read_register_format(a, zip, "zip", archive_read_format_zip_streamable_bid, archive_read_format_zip_options, archive_read_format_zip_streamable_read_header, archive_read_format_zip_read_data, archive_read_format_zip_read_data_skip_streamable, NULL, archive_read_format_zip_cleanup, archive_read_support_format_zip_capabilities_streamable, archive_read_format_zip_has_encrypted_entries); if (r != ARCHIVE_OK) free(zip); return (ARCHIVE_OK); } /* ------------------------------------------------------------------------ */ /* * Seeking-mode support */ static int archive_read_support_format_zip_capabilities_seekable(struct archive_read * a) { (void)a; /* UNUSED */ return (ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_DATA | ARCHIVE_READ_FORMAT_CAPS_ENCRYPT_METADATA); } /* * TODO: This is a performance sink because it forces the read core to * drop buffered data from the start of file, which will then have to * be re-read again if this bidder loses. * * We workaround this a little by passing in the best bid so far so * that later bidders can do nothing if they know they'll never * outbid. But we can certainly do better... */ static int read_eocd(struct zip *zip, const char *p, int64_t current_offset) { /* Sanity-check the EOCD we've found. */ /* This must be the first volume. */ if (archive_le16dec(p + 4) != 0) return 0; /* Central directory must be on this volume. */ if (archive_le16dec(p + 4) != archive_le16dec(p + 6)) return 0; /* All central directory entries must be on this volume. */ if (archive_le16dec(p + 10) != archive_le16dec(p + 8)) return 0; /* Central directory can't extend beyond start of EOCD record. */ if (archive_le32dec(p + 16) + archive_le32dec(p + 12) > current_offset) return 0; /* Save the central directory location for later use. */ zip->central_directory_offset = archive_le32dec(p + 16); /* This is just a tiny bit higher than the maximum returned by the streaming Zip bidder. This ensures that the more accurate seeking Zip parser wins whenever seek is available. */ return 32; } /* * Examine Zip64 EOCD locator: If it's valid, store the information * from it. */ static void read_zip64_eocd(struct archive_read *a, struct zip *zip, const char *p) { int64_t eocd64_offset; int64_t eocd64_size; /* Sanity-check the locator record. */ /* Central dir must be on first volume. */ if (archive_le32dec(p + 4) != 0) return; /* Must be only a single volume. */ if (archive_le32dec(p + 16) != 1) return; /* Find the Zip64 EOCD record. */ eocd64_offset = archive_le64dec(p + 8); if (__archive_read_seek(a, eocd64_offset, SEEK_SET) < 0) return; if ((p = __archive_read_ahead(a, 56, NULL)) == NULL) return; /* Make sure we can read all of it. */ eocd64_size = archive_le64dec(p + 4) + 12; if (eocd64_size < 56 || eocd64_size > 16384) return; if ((p = __archive_read_ahead(a, (size_t)eocd64_size, NULL)) == NULL) return; /* Sanity-check the EOCD64 */ if (archive_le32dec(p + 16) != 0) /* Must be disk #0 */ return; if (archive_le32dec(p + 20) != 0) /* CD must be on disk #0 */ return; /* CD can't be split. */ if (archive_le64dec(p + 24) != archive_le64dec(p + 32)) return; /* Save the central directory offset for later use. */ zip->central_directory_offset = archive_le64dec(p + 48); } static int archive_read_format_zip_seekable_bid(struct archive_read *a, int best_bid) { struct zip *zip = (struct zip *)a->format->data; int64_t file_size, current_offset; const char *p; int i, tail; /* If someone has already bid more than 32, then avoid trashing the look-ahead buffers with a seek. */ if (best_bid > 32) return (-1); file_size = __archive_read_seek(a, 0, SEEK_END); if (file_size <= 0) return 0; /* Search last 16k of file for end-of-central-directory * record (which starts with PK\005\006) */ tail = (int)zipmin(1024 * 16, file_size); current_offset = __archive_read_seek(a, -tail, SEEK_END); if (current_offset < 0) return 0; if ((p = __archive_read_ahead(a, (size_t)tail, NULL)) == NULL) return 0; /* Boyer-Moore search backwards from the end, since we want * to match the last EOCD in the file (there can be more than * one if there is an uncompressed Zip archive as a member * within this Zip archive). */ for (i = tail - 22; i > 0;) { switch (p[i]) { case 'P': if (memcmp(p + i, "PK\005\006", 4) == 0) { int ret = read_eocd(zip, p + i, current_offset + i); if (ret > 0) { /* Zip64 EOCD locator precedes * regular EOCD if present. */ if (i >= 20 && memcmp(p + i - 20, "PK\006\007", 4) == 0) { read_zip64_eocd(a, zip, p + i - 20); } return (ret); } } i -= 4; break; case 'K': i -= 1; break; case 005: i -= 2; break; case 006: i -= 3; break; default: i -= 4; break; } } return 0; } /* The red-black trees are only used in seeking mode to manage * the in-memory copy of the central directory. */ static int cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) { const struct zip_entry *e1 = (const struct zip_entry *)n1; const struct zip_entry *e2 = (const struct zip_entry *)n2; if (e1->local_header_offset > e2->local_header_offset) return -1; if (e1->local_header_offset < e2->local_header_offset) return 1; return 0; } static int cmp_key(const struct archive_rb_node *n, const void *key) { /* This function won't be called */ (void)n; /* UNUSED */ (void)key; /* UNUSED */ return 1; } static const struct archive_rb_tree_ops rb_ops = { &cmp_node, &cmp_key }; static int rsrc_cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) { const struct zip_entry *e1 = (const struct zip_entry *)n1; const struct zip_entry *e2 = (const struct zip_entry *)n2; return (strcmp(e2->rsrcname.s, e1->rsrcname.s)); } static int rsrc_cmp_key(const struct archive_rb_node *n, const void *key) { const struct zip_entry *e = (const struct zip_entry *)n; return (strcmp((const char *)key, e->rsrcname.s)); } static const struct archive_rb_tree_ops rb_rsrc_ops = { &rsrc_cmp_node, &rsrc_cmp_key }; static const char * rsrc_basename(const char *name, size_t name_length) { const char *s, *r; r = s = name; for (;;) { s = memchr(s, '/', name_length - (s - name)); if (s == NULL) break; r = ++s; } return (r); } static void expose_parent_dirs(struct zip *zip, const char *name, size_t name_length) { struct archive_string str; struct zip_entry *dir; char *s; archive_string_init(&str); archive_strncpy(&str, name, name_length); for (;;) { s = strrchr(str.s, '/'); if (s == NULL) break; *s = '\0'; /* Transfer the parent directory from zip->tree_rsrc RB * tree to zip->tree RB tree to expose. */ dir = (struct zip_entry *) __archive_rb_tree_find_node(&zip->tree_rsrc, str.s); if (dir == NULL) break; __archive_rb_tree_remove_node(&zip->tree_rsrc, &dir->node); archive_string_free(&dir->rsrcname); __archive_rb_tree_insert_node(&zip->tree, &dir->node); } archive_string_free(&str); } static int slurp_central_directory(struct archive_read *a, struct zip *zip) { ssize_t i; unsigned found; int64_t correction; ssize_t bytes_avail; const char *p; /* * Find the start of the central directory. The end-of-CD * record has our starting point, but there are lots of * Zip archives which have had other data prepended to the * file, which makes the recorded offsets all too small. * So we search forward from the specified offset until we * find the real start of the central directory. Then we * know the correction we need to apply to account for leading * padding. */ if (__archive_read_seek(a, zip->central_directory_offset, SEEK_SET) < 0) return ARCHIVE_FATAL; found = 0; while (!found) { if ((p = __archive_read_ahead(a, 20, &bytes_avail)) == NULL) return ARCHIVE_FATAL; for (found = 0, i = 0; !found && i < bytes_avail - 4;) { switch (p[i + 3]) { case 'P': i += 3; break; case 'K': i += 2; break; case 001: i += 1; break; case 002: if (memcmp(p + i, "PK\001\002", 4) == 0) { p += i; found = 1; } else i += 4; break; case 005: i += 1; break; case 006: if (memcmp(p + i, "PK\005\006", 4) == 0) { p += i; found = 1; } else if (memcmp(p + i, "PK\006\006", 4) == 0) { p += i; found = 1; } else i += 1; break; default: i += 4; break; } } __archive_read_consume(a, i); } correction = archive_filter_bytes(&a->archive, 0) - zip->central_directory_offset; __archive_rb_tree_init(&zip->tree, &rb_ops); __archive_rb_tree_init(&zip->tree_rsrc, &rb_rsrc_ops); zip->central_directory_entries_total = 0; while (1) { struct zip_entry *zip_entry; size_t filename_length, extra_length, comment_length; uint32_t external_attributes; const char *name, *r; if ((p = __archive_read_ahead(a, 4, NULL)) == NULL) return ARCHIVE_FATAL; if (memcmp(p, "PK\006\006", 4) == 0 || memcmp(p, "PK\005\006", 4) == 0) { break; } else if (memcmp(p, "PK\001\002", 4) != 0) { archive_set_error(&a->archive, -1, "Invalid central directory signature"); return ARCHIVE_FATAL; } if ((p = __archive_read_ahead(a, 46, NULL)) == NULL) return ARCHIVE_FATAL; zip_entry = calloc(1, sizeof(struct zip_entry)); zip_entry->next = zip->zip_entries; zip_entry->flags |= LA_FROM_CENTRAL_DIRECTORY; zip->zip_entries = zip_entry; zip->central_directory_entries_total++; /* version = p[4]; */ zip_entry->system = p[5]; /* version_required = archive_le16dec(p + 6); */ zip_entry->zip_flags = archive_le16dec(p + 8); if (zip_entry->zip_flags & (ZIP_ENCRYPTED | ZIP_STRONG_ENCRYPTED)){ zip->has_encrypted_entries = 1; } zip_entry->compression = (char)archive_le16dec(p + 10); zip_entry->mtime = zip_time(p + 12); zip_entry->crc32 = archive_le32dec(p + 16); if (zip_entry->zip_flags & ZIP_LENGTH_AT_END) zip_entry->decdat = p[13]; else zip_entry->decdat = p[19]; zip_entry->compressed_size = archive_le32dec(p + 20); zip_entry->uncompressed_size = archive_le32dec(p + 24); filename_length = archive_le16dec(p + 28); extra_length = archive_le16dec(p + 30); comment_length = archive_le16dec(p + 32); /* disk_start = archive_le16dec(p + 34); */ /* Better be zero. */ /* internal_attributes = archive_le16dec(p + 36); */ /* text bit */ external_attributes = archive_le32dec(p + 38); zip_entry->local_header_offset = archive_le32dec(p + 42) + correction; /* If we can't guess the mode, leave it zero here; when we read the local file header we might get more information. */ if (zip_entry->system == 3) { zip_entry->mode = external_attributes >> 16; } else if (zip_entry->system == 0) { // Interpret MSDOS directory bit if (0x10 == (external_attributes & 0x10)) { zip_entry->mode = AE_IFDIR | 0775; } else { zip_entry->mode = AE_IFREG | 0664; } if (0x01 == (external_attributes & 0x01)) { // Read-only bit; strip write permissions zip_entry->mode &= 0555; } } else { zip_entry->mode = 0; } /* We're done with the regular data; get the filename and * extra data. */ __archive_read_consume(a, 46); p = __archive_read_ahead(a, filename_length + extra_length, NULL); if (p == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file header"); return ARCHIVE_FATAL; } if (ARCHIVE_OK != process_extra(a, p + filename_length, extra_length, zip_entry)) { return ARCHIVE_FATAL; } /* * Mac resource fork files are stored under the * "__MACOSX/" directory, so we should check if * it is. */ if (!zip->process_mac_extensions) { /* Treat every entry as a regular entry. */ __archive_rb_tree_insert_node(&zip->tree, &zip_entry->node); } else { name = p; r = rsrc_basename(name, filename_length); if (filename_length >= 9 && strncmp("__MACOSX/", name, 9) == 0) { /* If this file is not a resource fork nor * a directory. We should treat it as a non * resource fork file to expose it. */ if (name[filename_length-1] != '/' && (r - name < 3 || r[0] != '.' || r[1] != '_')) { __archive_rb_tree_insert_node( &zip->tree, &zip_entry->node); /* Expose its parent directories. */ expose_parent_dirs(zip, name, filename_length); } else { /* This file is a resource fork file or * a directory. */ archive_strncpy(&(zip_entry->rsrcname), name, filename_length); __archive_rb_tree_insert_node( &zip->tree_rsrc, &zip_entry->node); } } else { /* Generate resource fork name to find its * resource file at zip->tree_rsrc. */ archive_strcpy(&(zip_entry->rsrcname), "__MACOSX/"); archive_strncat(&(zip_entry->rsrcname), name, r - name); archive_strcat(&(zip_entry->rsrcname), "._"); archive_strncat(&(zip_entry->rsrcname), name + (r - name), filename_length - (r - name)); /* Register an entry to RB tree to sort it by * file offset. */ __archive_rb_tree_insert_node(&zip->tree, &zip_entry->node); } } /* Skip the comment too ... */ __archive_read_consume(a, filename_length + extra_length + comment_length); } return ARCHIVE_OK; } static ssize_t zip_get_local_file_header_size(struct archive_read *a, size_t extra) { const char *p; ssize_t filename_length, extra_length; if ((p = __archive_read_ahead(a, extra + 30, NULL)) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file header"); return (ARCHIVE_WARN); } p += extra; if (memcmp(p, "PK\003\004", 4) != 0) { archive_set_error(&a->archive, -1, "Damaged Zip archive"); return ARCHIVE_WARN; } filename_length = archive_le16dec(p + 26); extra_length = archive_le16dec(p + 28); return (30 + filename_length + extra_length); } static int zip_read_mac_metadata(struct archive_read *a, struct archive_entry *entry, struct zip_entry *rsrc) { struct zip *zip = (struct zip *)a->format->data; unsigned char *metadata, *mp; int64_t offset = archive_filter_bytes(&a->archive, 0); size_t remaining_bytes, metadata_bytes; ssize_t hsize; int ret = ARCHIVE_OK, eof; switch(rsrc->compression) { case 0: /* No compression. */ if (rsrc->uncompressed_size != rsrc->compressed_size) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Malformed OS X metadata entry: inconsistent size"); return (ARCHIVE_FATAL); } #ifdef HAVE_ZLIB_H case 8: /* Deflate compression. */ #endif break; default: /* Unsupported compression. */ /* Return a warning. */ archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Unsupported ZIP compression method (%s)", compression_name(rsrc->compression)); /* We can't decompress this entry, but we will * be able to skip() it and try the next entry. */ return (ARCHIVE_WARN); } if (rsrc->uncompressed_size > (4 * 1024 * 1024)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Mac metadata is too large: %jd > 4M bytes", (intmax_t)rsrc->uncompressed_size); return (ARCHIVE_WARN); } if (rsrc->compressed_size > (4 * 1024 * 1024)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Mac metadata is too large: %jd > 4M bytes", (intmax_t)rsrc->compressed_size); return (ARCHIVE_WARN); } metadata = malloc((size_t)rsrc->uncompressed_size); if (metadata == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Mac metadata"); return (ARCHIVE_FATAL); } if (offset < rsrc->local_header_offset) __archive_read_consume(a, rsrc->local_header_offset - offset); else if (offset != rsrc->local_header_offset) { __archive_read_seek(a, rsrc->local_header_offset, SEEK_SET); } hsize = zip_get_local_file_header_size(a, 0); __archive_read_consume(a, hsize); remaining_bytes = (size_t)rsrc->compressed_size; metadata_bytes = (size_t)rsrc->uncompressed_size; mp = metadata; eof = 0; while (!eof && remaining_bytes) { const unsigned char *p; ssize_t bytes_avail; size_t bytes_used; p = __archive_read_ahead(a, 1, &bytes_avail); if (p == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Truncated ZIP file header"); ret = ARCHIVE_WARN; goto exit_mac_metadata; } if ((size_t)bytes_avail > remaining_bytes) bytes_avail = remaining_bytes; switch(rsrc->compression) { case 0: /* No compression. */ if ((size_t)bytes_avail > metadata_bytes) bytes_avail = metadata_bytes; memcpy(mp, p, bytes_avail); bytes_used = (size_t)bytes_avail; metadata_bytes -= bytes_used; mp += bytes_used; if (metadata_bytes == 0) eof = 1; break; #ifdef HAVE_ZLIB_H case 8: /* Deflate compression. */ { int r; ret = zip_deflate_init(a, zip); if (ret != ARCHIVE_OK) goto exit_mac_metadata; zip->stream.next_in = (Bytef *)(uintptr_t)(const void *)p; zip->stream.avail_in = (uInt)bytes_avail; zip->stream.total_in = 0; zip->stream.next_out = mp; zip->stream.avail_out = (uInt)metadata_bytes; zip->stream.total_out = 0; r = inflate(&zip->stream, 0); switch (r) { case Z_OK: break; case Z_STREAM_END: eof = 1; break; case Z_MEM_ERROR: archive_set_error(&a->archive, ENOMEM, "Out of memory for ZIP decompression"); ret = ARCHIVE_FATAL; goto exit_mac_metadata; default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "ZIP decompression failed (%d)", r); ret = ARCHIVE_FATAL; goto exit_mac_metadata; } bytes_used = zip->stream.total_in; metadata_bytes -= zip->stream.total_out; mp += zip->stream.total_out; break; } #endif default: bytes_used = 0; break; } __archive_read_consume(a, bytes_used); remaining_bytes -= bytes_used; } archive_entry_copy_mac_metadata(entry, metadata, (size_t)rsrc->uncompressed_size - metadata_bytes); exit_mac_metadata: __archive_read_seek(a, offset, SEEK_SET); zip->decompress_init = 0; free(metadata); return (ret); } static int archive_read_format_zip_seekable_read_header(struct archive_read *a, struct archive_entry *entry) { struct zip *zip = (struct zip *)a->format->data; struct zip_entry *rsrc; int64_t offset; int r, ret = ARCHIVE_OK; /* * It should be sufficient to call archive_read_next_header() for * a reader to determine if an entry is encrypted or not. If the * encryption of an entry is only detectable when calling * archive_read_data(), so be it. We'll do the same check there * as well. */ if (zip->has_encrypted_entries == ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW) zip->has_encrypted_entries = 0; a->archive.archive_format = ARCHIVE_FORMAT_ZIP; if (a->archive.archive_format_name == NULL) a->archive.archive_format_name = "ZIP"; if (zip->zip_entries == NULL) { r = slurp_central_directory(a, zip); if (r != ARCHIVE_OK) return r; /* Get first entry whose local header offset is lower than * other entries in the archive file. */ zip->entry = (struct zip_entry *)ARCHIVE_RB_TREE_MIN(&zip->tree); } else if (zip->entry != NULL) { /* Get next entry in local header offset order. */ zip->entry = (struct zip_entry *)__archive_rb_tree_iterate( &zip->tree, &zip->entry->node, ARCHIVE_RB_DIR_RIGHT); } if (zip->entry == NULL) return ARCHIVE_EOF; if (zip->entry->rsrcname.s) rsrc = (struct zip_entry *)__archive_rb_tree_find_node( &zip->tree_rsrc, zip->entry->rsrcname.s); else rsrc = NULL; if (zip->cctx_valid) archive_decrypto_aes_ctr_release(&zip->cctx); if (zip->hctx_valid) archive_hmac_sha1_cleanup(&zip->hctx); zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; __archive_read_reset_passphrase(a); /* File entries are sorted by the header offset, we should mostly * use __archive_read_consume to advance a read point to avoid redundant * data reading. */ offset = archive_filter_bytes(&a->archive, 0); if (offset < zip->entry->local_header_offset) __archive_read_consume(a, zip->entry->local_header_offset - offset); else if (offset != zip->entry->local_header_offset) { __archive_read_seek(a, zip->entry->local_header_offset, SEEK_SET); } zip->unconsumed = 0; r = zip_read_local_file_header(a, entry, zip); if (r != ARCHIVE_OK) return r; if (rsrc) { int ret2 = zip_read_mac_metadata(a, entry, rsrc); if (ret2 < ret) ret = ret2; } return (ret); } /* * We're going to seek for the next header anyway, so we don't * need to bother doing anything here. */ static int archive_read_format_zip_read_data_skip_seekable(struct archive_read *a) { struct zip *zip; zip = (struct zip *)(a->format->data); zip->unconsumed = 0; return (ARCHIVE_OK); } int archive_read_support_format_zip_seekable(struct archive *_a) { struct archive_read *a = (struct archive_read *)_a; struct zip *zip; int r; archive_check_magic(_a, ARCHIVE_READ_MAGIC, ARCHIVE_STATE_NEW, "archive_read_support_format_zip_seekable"); zip = (struct zip *)calloc(1, sizeof(*zip)); if (zip == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip data"); return (ARCHIVE_FATAL); } #ifdef HAVE_COPYFILE_H /* Set this by default on Mac OS. */ zip->process_mac_extensions = 1; #endif /* * Until enough data has been read, we cannot tell about * any encrypted entries yet. */ zip->has_encrypted_entries = ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW; zip->crc32func = real_crc32; r = __archive_read_register_format(a, zip, "zip", archive_read_format_zip_seekable_bid, archive_read_format_zip_options, archive_read_format_zip_seekable_read_header, archive_read_format_zip_read_data, archive_read_format_zip_read_data_skip_seekable, NULL, archive_read_format_zip_cleanup, archive_read_support_format_zip_capabilities_seekable, archive_read_format_zip_has_encrypted_entries); if (r != ARCHIVE_OK) free(zip); return (ARCHIVE_OK); } Index: head/contrib/libarchive/libarchive/archive_string.c =================================================================== --- head/contrib/libarchive/libarchive/archive_string.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_string.c (revision 311903) @@ -1,4200 +1,4206 @@ /*- * Copyright (c) 2003-2011 Tim Kientzle * Copyright (c) 2011-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); /* * Basic resizable string support, to simplify manipulating arbitrary-sized * strings while minimizing heap activity. * * In particular, the buffer used by a string object is only grown, it * never shrinks, so you can clear and reuse the same string object * without incurring additional memory allocations. */ #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_ICONV_H #include #endif #ifdef HAVE_LANGINFO_H #include #endif #ifdef HAVE_LOCALCHARSET_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_WCHAR_H #include #endif #if defined(_WIN32) && !defined(__CYGWIN__) #include #include #endif #include "archive_endian.h" #include "archive_private.h" #include "archive_string.h" #include "archive_string_composition.h" #if !defined(HAVE_WMEMCPY) && !defined(wmemcpy) #define wmemcpy(a,b,i) (wchar_t *)memcpy((a), (b), (i) * sizeof(wchar_t)) #endif #if !defined(HAVE_WMEMMOVE) && !defined(wmemmove) #define wmemmove(a,b,i) (wchar_t *)memmove((a), (b), (i) * sizeof(wchar_t)) #endif struct archive_string_conv { struct archive_string_conv *next; char *from_charset; char *to_charset; unsigned from_cp; unsigned to_cp; /* Set 1 if from_charset and to_charset are the same. */ int same; int flag; #define SCONV_TO_CHARSET 1 /* MBS is being converted to specified * charset. */ #define SCONV_FROM_CHARSET (1<<1) /* MBS is being converted from * specified charset. */ #define SCONV_BEST_EFFORT (1<<2) /* Copy at least ASCII code. */ #define SCONV_WIN_CP (1<<3) /* Use Windows API for converting * MBS. */ #define SCONV_UTF8_LIBARCHIVE_2 (1<<4) /* Incorrect UTF-8 made by libarchive * 2.x in the wrong assumption. */ #define SCONV_NORMALIZATION_C (1<<6) /* Need normalization to be Form C. * Before UTF-8 characters are actually * processed. */ #define SCONV_NORMALIZATION_D (1<<7) /* Need normalization to be Form D. * Before UTF-8 characters are actually * processed. * Currently this only for MAC OS X. */ #define SCONV_TO_UTF8 (1<<8) /* "to charset" side is UTF-8. */ #define SCONV_FROM_UTF8 (1<<9) /* "from charset" side is UTF-8. */ #define SCONV_TO_UTF16BE (1<<10) /* "to charset" side is UTF-16BE. */ #define SCONV_FROM_UTF16BE (1<<11) /* "from charset" side is UTF-16BE. */ #define SCONV_TO_UTF16LE (1<<12) /* "to charset" side is UTF-16LE. */ #define SCONV_FROM_UTF16LE (1<<13) /* "from charset" side is UTF-16LE. */ #define SCONV_TO_UTF16 (SCONV_TO_UTF16BE | SCONV_TO_UTF16LE) #define SCONV_FROM_UTF16 (SCONV_FROM_UTF16BE | SCONV_FROM_UTF16LE) #if HAVE_ICONV iconv_t cd; iconv_t cd_w;/* Use at archive_mstring on * Windows. */ #endif /* A temporary buffer for normalization. */ struct archive_string utftmp; int (*converter[2])(struct archive_string *, const void *, size_t, struct archive_string_conv *); int nconverter; }; #define CP_C_LOCALE 0 /* "C" locale only for this file. */ #define CP_UTF16LE 1200 #define CP_UTF16BE 1201 #define IS_HIGH_SURROGATE_LA(uc) ((uc) >= 0xD800 && (uc) <= 0xDBFF) #define IS_LOW_SURROGATE_LA(uc) ((uc) >= 0xDC00 && (uc) <= 0xDFFF) #define IS_SURROGATE_PAIR_LA(uc) ((uc) >= 0xD800 && (uc) <= 0xDFFF) #define UNICODE_MAX 0x10FFFF #define UNICODE_R_CHAR 0xFFFD /* Replacement character. */ /* Set U+FFFD(Replacement character) in UTF-8. */ static const char utf8_replacement_char[] = {0xef, 0xbf, 0xbd}; static struct archive_string_conv *find_sconv_object(struct archive *, const char *, const char *); static void add_sconv_object(struct archive *, struct archive_string_conv *); static struct archive_string_conv *create_sconv_object(const char *, const char *, unsigned, int); static void free_sconv_object(struct archive_string_conv *); static struct archive_string_conv *get_sconv_object(struct archive *, const char *, const char *, int); static unsigned make_codepage_from_charset(const char *); static unsigned get_current_codepage(void); static unsigned get_current_oemcp(void); static size_t mbsnbytes(const void *, size_t); static size_t utf16nbytes(const void *, size_t); #if defined(_WIN32) && !defined(__CYGWIN__) static int archive_wstring_append_from_mbs_in_codepage( struct archive_wstring *, const char *, size_t, struct archive_string_conv *); static int archive_string_append_from_wcs_in_codepage(struct archive_string *, const wchar_t *, size_t, struct archive_string_conv *); static int is_big_endian(void); static int strncat_in_codepage(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int win_strncat_from_utf16be(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int win_strncat_from_utf16le(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int win_strncat_to_utf16be(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int win_strncat_to_utf16le(struct archive_string *, const void *, size_t, struct archive_string_conv *); #endif static int best_effort_strncat_from_utf16be(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int best_effort_strncat_from_utf16le(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int best_effort_strncat_to_utf16be(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int best_effort_strncat_to_utf16le(struct archive_string *, const void *, size_t, struct archive_string_conv *); #if defined(HAVE_ICONV) static int iconv_strncat_in_locale(struct archive_string *, const void *, size_t, struct archive_string_conv *); #endif static int best_effort_strncat_in_locale(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int _utf8_to_unicode(uint32_t *, const char *, size_t); static int utf8_to_unicode(uint32_t *, const char *, size_t); static inline uint32_t combine_surrogate_pair(uint32_t, uint32_t); static int cesu8_to_unicode(uint32_t *, const char *, size_t); static size_t unicode_to_utf8(char *, size_t, uint32_t); static int utf16_to_unicode(uint32_t *, const char *, size_t, int); static size_t unicode_to_utf16be(char *, size_t, uint32_t); static size_t unicode_to_utf16le(char *, size_t, uint32_t); static int strncat_from_utf8_libarchive2(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int strncat_from_utf8_to_utf8(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int archive_string_normalize_C(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int archive_string_normalize_D(struct archive_string *, const void *, size_t, struct archive_string_conv *); static int archive_string_append_unicode(struct archive_string *, const void *, size_t, struct archive_string_conv *); static struct archive_string * archive_string_append(struct archive_string *as, const char *p, size_t s) { if (archive_string_ensure(as, as->length + s + 1) == NULL) return (NULL); memmove(as->s + as->length, p, s); as->length += s; as->s[as->length] = 0; return (as); } static struct archive_wstring * archive_wstring_append(struct archive_wstring *as, const wchar_t *p, size_t s) { if (archive_wstring_ensure(as, as->length + s + 1) == NULL) return (NULL); wmemmove(as->s + as->length, p, s); as->length += s; as->s[as->length] = 0; return (as); } +struct archive_string * +archive_array_append(struct archive_string *as, const char *p, size_t s) +{ + return archive_string_append(as, p, s); +} + void archive_string_concat(struct archive_string *dest, struct archive_string *src) { if (archive_string_append(dest, src->s, src->length) == NULL) __archive_errx(1, "Out of memory"); } void archive_wstring_concat(struct archive_wstring *dest, struct archive_wstring *src) { if (archive_wstring_append(dest, src->s, src->length) == NULL) __archive_errx(1, "Out of memory"); } void archive_string_free(struct archive_string *as) { as->length = 0; as->buffer_length = 0; free(as->s); as->s = NULL; } void archive_wstring_free(struct archive_wstring *as) { as->length = 0; as->buffer_length = 0; free(as->s); as->s = NULL; } struct archive_wstring * archive_wstring_ensure(struct archive_wstring *as, size_t s) { return (struct archive_wstring *) archive_string_ensure((struct archive_string *)as, s * sizeof(wchar_t)); } /* Returns NULL on any allocation failure. */ struct archive_string * archive_string_ensure(struct archive_string *as, size_t s) { char *p; size_t new_length; /* If buffer is already big enough, don't reallocate. */ if (as->s && (s <= as->buffer_length)) return (as); /* * Growing the buffer at least exponentially ensures that * append operations are always linear in the number of * characters appended. Using a smaller growth rate for * larger buffers reduces memory waste somewhat at the cost of * a larger constant factor. */ if (as->buffer_length < 32) /* Start with a minimum 32-character buffer. */ new_length = 32; else if (as->buffer_length < 8192) /* Buffers under 8k are doubled for speed. */ new_length = as->buffer_length + as->buffer_length; else { /* Buffers 8k and over grow by at least 25% each time. */ new_length = as->buffer_length + as->buffer_length / 4; /* Be safe: If size wraps, fail. */ if (new_length < as->buffer_length) { /* On failure, wipe the string and return NULL. */ archive_string_free(as); errno = ENOMEM;/* Make sure errno has ENOMEM. */ return (NULL); } } /* * The computation above is a lower limit to how much we'll * grow the buffer. In any case, we have to grow it enough to * hold the request. */ if (new_length < s) new_length = s; /* Now we can reallocate the buffer. */ p = (char *)realloc(as->s, new_length); if (p == NULL) { /* On failure, wipe the string and return NULL. */ archive_string_free(as); errno = ENOMEM;/* Make sure errno has ENOMEM. */ return (NULL); } as->s = p; as->buffer_length = new_length; return (as); } /* * TODO: See if there's a way to avoid scanning * the source string twice. Then test to see * if it actually helps (remember that we're almost * always called with pretty short arguments, so * such an optimization might not help). */ struct archive_string * archive_strncat(struct archive_string *as, const void *_p, size_t n) { size_t s; const char *p, *pp; p = (const char *)_p; /* Like strlen(p), except won't examine positions beyond p[n]. */ s = 0; pp = p; while (s < n && *pp) { pp++; s++; } if ((as = archive_string_append(as, p, s)) == NULL) __archive_errx(1, "Out of memory"); return (as); } struct archive_wstring * archive_wstrncat(struct archive_wstring *as, const wchar_t *p, size_t n) { size_t s; const wchar_t *pp; /* Like strlen(p), except won't examine positions beyond p[n]. */ s = 0; pp = p; while (s < n && *pp) { pp++; s++; } if ((as = archive_wstring_append(as, p, s)) == NULL) __archive_errx(1, "Out of memory"); return (as); } struct archive_string * archive_strcat(struct archive_string *as, const void *p) { /* strcat is just strncat without an effective limit. * Assert that we'll never get called with a source * string over 16MB. * TODO: Review all uses of strcat in the source * and try to replace them with strncat(). */ return archive_strncat(as, p, 0x1000000); } struct archive_wstring * archive_wstrcat(struct archive_wstring *as, const wchar_t *p) { /* Ditto. */ return archive_wstrncat(as, p, 0x1000000); } struct archive_string * archive_strappend_char(struct archive_string *as, char c) { if ((as = archive_string_append(as, &c, 1)) == NULL) __archive_errx(1, "Out of memory"); return (as); } struct archive_wstring * archive_wstrappend_wchar(struct archive_wstring *as, wchar_t c) { if ((as = archive_wstring_append(as, &c, 1)) == NULL) __archive_errx(1, "Out of memory"); return (as); } /* * Get the "current character set" name to use with iconv. * On FreeBSD, the empty character set name "" chooses * the correct character encoding for the current locale, * so this isn't necessary. * But iconv on Mac OS 10.6 doesn't seem to handle this correctly; * on that system, we have to explicitly call nl_langinfo() * to get the right name. Not sure about other platforms. * * NOTE: GNU libiconv does not recognize the character-set name * which some platform nl_langinfo(CODESET) returns, so we should * use locale_charset() instead of nl_langinfo(CODESET) for GNU libiconv. */ static const char * default_iconv_charset(const char *charset) { if (charset != NULL && charset[0] != '\0') return charset; #if HAVE_LOCALE_CHARSET && !defined(__APPLE__) /* locale_charset() is broken on Mac OS */ return locale_charset(); #elif HAVE_NL_LANGINFO return nl_langinfo(CODESET); #else return ""; #endif } #if defined(_WIN32) && !defined(__CYGWIN__) /* * Convert MBS to WCS. * Note: returns -1 if conversion fails. */ int archive_wstring_append_from_mbs(struct archive_wstring *dest, const char *p, size_t len) { return archive_wstring_append_from_mbs_in_codepage(dest, p, len, NULL); } static int archive_wstring_append_from_mbs_in_codepage(struct archive_wstring *dest, const char *s, size_t length, struct archive_string_conv *sc) { int count, ret = 0; UINT from_cp; if (sc != NULL) from_cp = sc->from_cp; else from_cp = get_current_codepage(); if (from_cp == CP_C_LOCALE) { /* * "C" locale special process. */ wchar_t *ws; const unsigned char *mp; if (NULL == archive_wstring_ensure(dest, dest->length + length + 1)) return (-1); ws = dest->s + dest->length; mp = (const unsigned char *)s; count = 0; while (count < (int)length && *mp) { *ws++ = (wchar_t)*mp++; count++; } } else if (sc != NULL && (sc->flag & (SCONV_NORMALIZATION_C | SCONV_NORMALIZATION_D))) { /* * Normalize UTF-8 and UTF-16BE and convert it directly * to UTF-16 as wchar_t. */ struct archive_string u16; int saved_flag = sc->flag;/* save current flag. */ if (is_big_endian()) sc->flag |= SCONV_TO_UTF16BE; else sc->flag |= SCONV_TO_UTF16LE; if (sc->flag & SCONV_FROM_UTF16) { /* * UTF-16BE/LE NFD ===> UTF-16 NFC * UTF-16BE/LE NFC ===> UTF-16 NFD */ count = (int)utf16nbytes(s, length); } else { /* * UTF-8 NFD ===> UTF-16 NFC * UTF-8 NFC ===> UTF-16 NFD */ count = (int)mbsnbytes(s, length); } u16.s = (char *)dest->s; u16.length = dest->length << 1;; u16.buffer_length = dest->buffer_length; if (sc->flag & SCONV_NORMALIZATION_C) ret = archive_string_normalize_C(&u16, s, count, sc); else ret = archive_string_normalize_D(&u16, s, count, sc); dest->s = (wchar_t *)u16.s; dest->length = u16.length >> 1; dest->buffer_length = u16.buffer_length; sc->flag = saved_flag;/* restore the saved flag. */ return (ret); } else if (sc != NULL && (sc->flag & SCONV_FROM_UTF16)) { count = (int)utf16nbytes(s, length); count >>= 1; /* to be WCS length */ /* Allocate memory for WCS. */ if (NULL == archive_wstring_ensure(dest, dest->length + count + 1)) return (-1); wmemcpy(dest->s + dest->length, (const wchar_t *)s, count); if ((sc->flag & SCONV_FROM_UTF16BE) && !is_big_endian()) { uint16_t *u16 = (uint16_t *)(dest->s + dest->length); int b; for (b = 0; b < count; b++) { uint16_t val = archive_le16dec(u16+b); archive_be16enc(u16+b, val); } } else if ((sc->flag & SCONV_FROM_UTF16LE) && is_big_endian()) { uint16_t *u16 = (uint16_t *)(dest->s + dest->length); int b; for (b = 0; b < count; b++) { uint16_t val = archive_be16dec(u16+b); archive_le16enc(u16+b, val); } } } else { DWORD mbflag; size_t buffsize; if (sc == NULL) mbflag = 0; else if (sc->flag & SCONV_FROM_CHARSET) { /* Do not trust the length which comes from * an archive file. */ length = mbsnbytes(s, length); mbflag = 0; } else mbflag = MB_PRECOMPOSED; buffsize = dest->length + length + 1; do { /* Allocate memory for WCS. */ if (NULL == archive_wstring_ensure(dest, buffsize)) return (-1); /* Convert MBS to WCS. */ count = MultiByteToWideChar(from_cp, mbflag, s, (int)length, dest->s + dest->length, (int)(dest->buffer_length >> 1) -1); if (count == 0 && GetLastError() == ERROR_INSUFFICIENT_BUFFER) { /* Expand the WCS buffer. */ buffsize = dest->buffer_length << 1; continue; } if (count == 0 && length != 0) ret = -1; break; } while (1); } dest->length += count; dest->s[dest->length] = L'\0'; return (ret); } #else /* * Convert MBS to WCS. * Note: returns -1 if conversion fails. */ int archive_wstring_append_from_mbs(struct archive_wstring *dest, const char *p, size_t len) { size_t r; int ret_val = 0; /* * No single byte will be more than one wide character, * so this length estimate will always be big enough. */ size_t wcs_length = len; size_t mbs_length = len; const char *mbs = p; wchar_t *wcs; #if HAVE_MBRTOWC mbstate_t shift_state; memset(&shift_state, 0, sizeof(shift_state)); #endif if (NULL == archive_wstring_ensure(dest, dest->length + wcs_length + 1)) return (-1); wcs = dest->s + dest->length; /* * We cannot use mbsrtowcs/mbstowcs here because those may convert - * extra MBS when strlen(p) > len and one wide character consis of + * extra MBS when strlen(p) > len and one wide character consists of * multi bytes. */ while (*mbs && mbs_length > 0) { if (wcs_length == 0) { dest->length = wcs - dest->s; dest->s[dest->length] = L'\0'; wcs_length = mbs_length; if (NULL == archive_wstring_ensure(dest, dest->length + wcs_length + 1)) return (-1); wcs = dest->s + dest->length; } #if HAVE_MBRTOWC r = mbrtowc(wcs, mbs, wcs_length, &shift_state); #else r = mbtowc(wcs, mbs, wcs_length); #endif if (r == (size_t)-1 || r == (size_t)-2) { ret_val = -1; if (errno == EILSEQ) { ++mbs; --mbs_length; continue; } else break; } if (r == 0 || r > mbs_length) break; wcs++; wcs_length--; mbs += r; mbs_length -= r; } dest->length = wcs - dest->s; dest->s[dest->length] = L'\0'; return (ret_val); } #endif #if defined(_WIN32) && !defined(__CYGWIN__) /* * WCS ==> MBS. * Note: returns -1 if conversion fails. * * Win32 builds use WideCharToMultiByte from the Windows API. * (Maybe Cygwin should too? WideCharToMultiByte will know a * lot more about local character encodings than the wcrtomb() * wrapper is going to know.) */ int archive_string_append_from_wcs(struct archive_string *as, const wchar_t *w, size_t len) { return archive_string_append_from_wcs_in_codepage(as, w, len, NULL); } static int archive_string_append_from_wcs_in_codepage(struct archive_string *as, const wchar_t *ws, size_t len, struct archive_string_conv *sc) { BOOL defchar_used, *dp; int count, ret = 0; UINT to_cp; int wslen = (int)len; if (sc != NULL) to_cp = sc->to_cp; else to_cp = get_current_codepage(); if (to_cp == CP_C_LOCALE) { /* * "C" locale special process. */ const wchar_t *wp = ws; char *p; if (NULL == archive_string_ensure(as, as->length + wslen +1)) return (-1); p = as->s + as->length; count = 0; defchar_used = 0; while (count < wslen && *wp) { if (*wp > 255) { *p++ = '?'; wp++; defchar_used = 1; } else *p++ = (char)*wp++; count++; } } else if (sc != NULL && (sc->flag & SCONV_TO_UTF16)) { uint16_t *u16; if (NULL == archive_string_ensure(as, as->length + len * 2 + 2)) return (-1); u16 = (uint16_t *)(as->s + as->length); count = 0; defchar_used = 0; if (sc->flag & SCONV_TO_UTF16BE) { while (count < (int)len && *ws) { archive_be16enc(u16+count, *ws); ws++; count++; } } else { while (count < (int)len && *ws) { archive_le16enc(u16+count, *ws); ws++; count++; } } count <<= 1; /* to be byte size */ } else { /* Make sure the MBS buffer has plenty to set. */ if (NULL == archive_string_ensure(as, as->length + len * 2 + 1)) return (-1); do { defchar_used = 0; if (to_cp == CP_UTF8 || sc == NULL) dp = NULL; else dp = &defchar_used; count = WideCharToMultiByte(to_cp, 0, ws, wslen, as->s + as->length, (int)as->buffer_length-1, NULL, dp); if (count == 0 && GetLastError() == ERROR_INSUFFICIENT_BUFFER) { /* Expand the MBS buffer and retry. */ if (NULL == archive_string_ensure(as, as->buffer_length + len)) return (-1); continue; } if (count == 0) ret = -1; break; } while (1); } as->length += count; as->s[as->length] = '\0'; return (defchar_used?-1:ret); } #elif defined(HAVE_WCTOMB) || defined(HAVE_WCRTOMB) /* * Translates a wide character string into current locale character set * and appends to the archive_string. Note: returns -1 if conversion * fails. */ int archive_string_append_from_wcs(struct archive_string *as, const wchar_t *w, size_t len) { /* We cannot use the standard wcstombs() here because it * cannot tell us how big the output buffer should be. So * I've built a loop around wcrtomb() or wctomb() that * converts a character at a time and resizes the string as * needed. We prefer wcrtomb() when it's available because * it's thread-safe. */ int n, ret_val = 0; char *p; char *end; #if HAVE_WCRTOMB mbstate_t shift_state; memset(&shift_state, 0, sizeof(shift_state)); #else /* Clear the shift state before starting. */ wctomb(NULL, L'\0'); #endif /* * Allocate buffer for MBS. * We need this allocation here since it is possible that * as->s is still NULL. */ if (archive_string_ensure(as, as->length + len + 1) == NULL) return (-1); p = as->s + as->length; end = as->s + as->buffer_length - MB_CUR_MAX -1; while (*w != L'\0' && len > 0) { if (p >= end) { as->length = p - as->s; as->s[as->length] = '\0'; /* Re-allocate buffer for MBS. */ if (archive_string_ensure(as, as->length + len * 2 + 1) == NULL) return (-1); p = as->s + as->length; end = as->s + as->buffer_length - MB_CUR_MAX -1; } #if HAVE_WCRTOMB n = wcrtomb(p, *w++, &shift_state); #else n = wctomb(p, *w++); #endif if (n == -1) { if (errno == EILSEQ) { /* Skip an illegal wide char. */ *p++ = '?'; ret_val = -1; } else { ret_val = -1; break; } } else p += n; len--; } as->length = p - as->s; as->s[as->length] = '\0'; return (ret_val); } #else /* HAVE_WCTOMB || HAVE_WCRTOMB */ /* * TODO: Test if __STDC_ISO_10646__ is defined. * Non-Windows uses ISO C wcrtomb() or wctomb() to perform the conversion * one character at a time. If a non-Windows platform doesn't have * either of these, fall back to the built-in UTF8 conversion. */ int archive_string_append_from_wcs(struct archive_string *as, const wchar_t *w, size_t len) { (void)as;/* UNUSED */ (void)w;/* UNUSED */ (void)len;/* UNUSED */ errno = ENOSYS; return (-1); } #endif /* HAVE_WCTOMB || HAVE_WCRTOMB */ /* * Find a string conversion object by a pair of 'from' charset name * and 'to' charset name from an archive object. * Return NULL if not found. */ static struct archive_string_conv * find_sconv_object(struct archive *a, const char *fc, const char *tc) { struct archive_string_conv *sc; if (a == NULL) return (NULL); for (sc = a->sconv; sc != NULL; sc = sc->next) { if (strcmp(sc->from_charset, fc) == 0 && strcmp(sc->to_charset, tc) == 0) break; } return (sc); } /* * Register a string object to an archive object. */ static void add_sconv_object(struct archive *a, struct archive_string_conv *sc) { struct archive_string_conv **psc; /* Add a new sconv to sconv list. */ psc = &(a->sconv); while (*psc != NULL) psc = &((*psc)->next); *psc = sc; } static void add_converter(struct archive_string_conv *sc, int (*converter) (struct archive_string *, const void *, size_t, struct archive_string_conv *)) { if (sc == NULL || sc->nconverter >= 2) __archive_errx(1, "Programing error"); sc->converter[sc->nconverter++] = converter; } static void setup_converter(struct archive_string_conv *sc) { /* Reset. */ sc->nconverter = 0; /* * Perform special sequence for the incorrect UTF-8 filenames * made by libarchive2.x. */ if (sc->flag & SCONV_UTF8_LIBARCHIVE_2) { add_converter(sc, strncat_from_utf8_libarchive2); return; } /* * Convert a string to UTF-16BE/LE. */ if (sc->flag & SCONV_TO_UTF16) { /* * If the current locale is UTF-8, we can translate * a UTF-8 string into a UTF-16BE string. */ if (sc->flag & SCONV_FROM_UTF8) { add_converter(sc, archive_string_append_unicode); return; } #if defined(_WIN32) && !defined(__CYGWIN__) if (sc->flag & SCONV_WIN_CP) { if (sc->flag & SCONV_TO_UTF16BE) add_converter(sc, win_strncat_to_utf16be); else add_converter(sc, win_strncat_to_utf16le); return; } #endif #if defined(HAVE_ICONV) if (sc->cd != (iconv_t)-1) { add_converter(sc, iconv_strncat_in_locale); return; } #endif if (sc->flag & SCONV_BEST_EFFORT) { if (sc->flag & SCONV_TO_UTF16BE) add_converter(sc, best_effort_strncat_to_utf16be); else add_converter(sc, best_effort_strncat_to_utf16le); } else /* Make sure we have no converter. */ sc->nconverter = 0; return; } /* * Convert a string from UTF-16BE/LE. */ if (sc->flag & SCONV_FROM_UTF16) { /* * At least we should normalize a UTF-16BE string. */ if (sc->flag & SCONV_NORMALIZATION_D) add_converter(sc,archive_string_normalize_D); else if (sc->flag & SCONV_NORMALIZATION_C) add_converter(sc, archive_string_normalize_C); if (sc->flag & SCONV_TO_UTF8) { /* * If the current locale is UTF-8, we can translate * a UTF-16BE/LE string into a UTF-8 string directly. */ if (!(sc->flag & (SCONV_NORMALIZATION_D |SCONV_NORMALIZATION_C))) add_converter(sc, archive_string_append_unicode); return; } #if defined(_WIN32) && !defined(__CYGWIN__) if (sc->flag & SCONV_WIN_CP) { if (sc->flag & SCONV_FROM_UTF16BE) add_converter(sc, win_strncat_from_utf16be); else add_converter(sc, win_strncat_from_utf16le); return; } #endif #if defined(HAVE_ICONV) if (sc->cd != (iconv_t)-1) { add_converter(sc, iconv_strncat_in_locale); return; } #endif if ((sc->flag & (SCONV_BEST_EFFORT | SCONV_FROM_UTF16BE)) == (SCONV_BEST_EFFORT | SCONV_FROM_UTF16BE)) add_converter(sc, best_effort_strncat_from_utf16be); else if ((sc->flag & (SCONV_BEST_EFFORT | SCONV_FROM_UTF16LE)) == (SCONV_BEST_EFFORT | SCONV_FROM_UTF16LE)) add_converter(sc, best_effort_strncat_from_utf16le); else /* Make sure we have no converter. */ sc->nconverter = 0; return; } if (sc->flag & SCONV_FROM_UTF8) { /* * At least we should normalize a UTF-8 string. */ if (sc->flag & SCONV_NORMALIZATION_D) add_converter(sc,archive_string_normalize_D); else if (sc->flag & SCONV_NORMALIZATION_C) add_converter(sc, archive_string_normalize_C); /* * Copy UTF-8 string with a check of CESU-8. * Apparently, iconv does not check surrogate pairs in UTF-8 * when both from-charset and to-charset are UTF-8, and then * we use our UTF-8 copy code. */ if (sc->flag & SCONV_TO_UTF8) { /* * If the current locale is UTF-8, we can translate * a UTF-16BE string into a UTF-8 string directly. */ if (!(sc->flag & (SCONV_NORMALIZATION_D |SCONV_NORMALIZATION_C))) add_converter(sc, strncat_from_utf8_to_utf8); return; } } #if defined(_WIN32) && !defined(__CYGWIN__) /* * On Windows we can use Windows API for a string conversion. */ if (sc->flag & SCONV_WIN_CP) { add_converter(sc, strncat_in_codepage); return; } #endif #if HAVE_ICONV if (sc->cd != (iconv_t)-1) { add_converter(sc, iconv_strncat_in_locale); /* * iconv generally does not support UTF-8-MAC and so * we have to the output of iconv from NFC to NFD if * need. */ if ((sc->flag & SCONV_FROM_CHARSET) && (sc->flag & SCONV_TO_UTF8)) { if (sc->flag & SCONV_NORMALIZATION_D) add_converter(sc, archive_string_normalize_D); } return; } #endif /* * Try conversion in the best effort or no conversion. */ if ((sc->flag & SCONV_BEST_EFFORT) || sc->same) add_converter(sc, best_effort_strncat_in_locale); else /* Make sure we have no converter. */ sc->nconverter = 0; } /* * Return canonicalized charset-name but this supports just UTF-8, UTF-16BE * and CP932 which are referenced in create_sconv_object(). */ static const char * canonical_charset_name(const char *charset) { char cs[16]; char *p; const char *s; if (charset == NULL || charset[0] == '\0' || strlen(charset) > 15) return (charset); /* Copy name to uppercase. */ p = cs; s = charset; while (*s) { char c = *s++; if (c >= 'a' && c <= 'z') c -= 'a' - 'A'; *p++ = c; } *p++ = '\0'; if (strcmp(cs, "UTF-8") == 0 || strcmp(cs, "UTF8") == 0) return ("UTF-8"); if (strcmp(cs, "UTF-16BE") == 0 || strcmp(cs, "UTF16BE") == 0) return ("UTF-16BE"); if (strcmp(cs, "UTF-16LE") == 0 || strcmp(cs, "UTF16LE") == 0) return ("UTF-16LE"); if (strcmp(cs, "CP932") == 0) return ("CP932"); return (charset); } /* * Create a string conversion object. */ static struct archive_string_conv * create_sconv_object(const char *fc, const char *tc, unsigned current_codepage, int flag) { struct archive_string_conv *sc; sc = calloc(1, sizeof(*sc)); if (sc == NULL) return (NULL); sc->next = NULL; sc->from_charset = strdup(fc); if (sc->from_charset == NULL) { free(sc); return (NULL); } sc->to_charset = strdup(tc); if (sc->to_charset == NULL) { free(sc->from_charset); free(sc); return (NULL); } archive_string_init(&sc->utftmp); if (flag & SCONV_TO_CHARSET) { /* * Convert characters from the current locale charset to * a specified charset. */ sc->from_cp = current_codepage; sc->to_cp = make_codepage_from_charset(tc); #if defined(_WIN32) && !defined(__CYGWIN__) if (IsValidCodePage(sc->to_cp)) flag |= SCONV_WIN_CP; #endif } else if (flag & SCONV_FROM_CHARSET) { /* * Convert characters from a specified charset to * the current locale charset. */ sc->to_cp = current_codepage; sc->from_cp = make_codepage_from_charset(fc); #if defined(_WIN32) && !defined(__CYGWIN__) if (IsValidCodePage(sc->from_cp)) flag |= SCONV_WIN_CP; #endif } /* * Check if "from charset" and "to charset" are the same. */ if (strcmp(fc, tc) == 0 || (sc->from_cp != (unsigned)-1 && sc->from_cp == sc->to_cp)) sc->same = 1; else sc->same = 0; /* * Mark if "from charset" or "to charset" are UTF-8 or UTF-16BE/LE. */ if (strcmp(tc, "UTF-8") == 0) flag |= SCONV_TO_UTF8; else if (strcmp(tc, "UTF-16BE") == 0) flag |= SCONV_TO_UTF16BE; else if (strcmp(tc, "UTF-16LE") == 0) flag |= SCONV_TO_UTF16LE; if (strcmp(fc, "UTF-8") == 0) flag |= SCONV_FROM_UTF8; else if (strcmp(fc, "UTF-16BE") == 0) flag |= SCONV_FROM_UTF16BE; else if (strcmp(fc, "UTF-16LE") == 0) flag |= SCONV_FROM_UTF16LE; #if defined(_WIN32) && !defined(__CYGWIN__) if (sc->to_cp == CP_UTF8) flag |= SCONV_TO_UTF8; else if (sc->to_cp == CP_UTF16BE) flag |= SCONV_TO_UTF16BE | SCONV_WIN_CP; else if (sc->to_cp == CP_UTF16LE) flag |= SCONV_TO_UTF16LE | SCONV_WIN_CP; if (sc->from_cp == CP_UTF8) flag |= SCONV_FROM_UTF8; else if (sc->from_cp == CP_UTF16BE) flag |= SCONV_FROM_UTF16BE | SCONV_WIN_CP; else if (sc->from_cp == CP_UTF16LE) flag |= SCONV_FROM_UTF16LE | SCONV_WIN_CP; #endif /* * Set a flag for Unicode NFD. Usually iconv cannot correctly * handle it. So we have to translate NFD characters to NFC ones * ourselves before iconv handles. Another reason is to prevent * that the same sight of two filenames, one is NFC and other * is NFD, would be in its directory. * On Mac OS X, although its filesystem layer automatically * convert filenames to NFD, it would be useful for filename * comparing to find out the same filenames that we normalize * that to be NFD ourselves. */ if ((flag & SCONV_FROM_CHARSET) && (flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8))) { #if defined(__APPLE__) if (flag & SCONV_TO_UTF8) flag |= SCONV_NORMALIZATION_D; else #endif flag |= SCONV_NORMALIZATION_C; } #if defined(__APPLE__) /* * In case writing an archive file, make sure that a filename * going to be passed to iconv is a Unicode NFC string since * a filename in HFS Plus filesystem is a Unicode NFD one and * iconv cannot handle it with "UTF-8" charset. It is simpler * than a use of "UTF-8-MAC" charset. */ if ((flag & SCONV_TO_CHARSET) && (flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8)) && !(flag & (SCONV_TO_UTF16 | SCONV_TO_UTF8))) flag |= SCONV_NORMALIZATION_C; /* * In case reading an archive file. make sure that a filename * will be passed to users is a Unicode NFD string in order to * correctly compare the filename with other one which comes * from HFS Plus filesystem. */ if ((flag & SCONV_FROM_CHARSET) && !(flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8)) && (flag & SCONV_TO_UTF8)) flag |= SCONV_NORMALIZATION_D; #endif #if defined(HAVE_ICONV) sc->cd_w = (iconv_t)-1; /* * Create an iconv object. */ if (((flag & (SCONV_TO_UTF8 | SCONV_TO_UTF16)) && (flag & (SCONV_FROM_UTF8 | SCONV_FROM_UTF16))) || (flag & SCONV_WIN_CP)) { /* This case we won't use iconv. */ sc->cd = (iconv_t)-1; } else { sc->cd = iconv_open(tc, fc); if (sc->cd == (iconv_t)-1 && (sc->flag & SCONV_BEST_EFFORT)) { /* - * Unfortunaly, all of iconv implements do support + * Unfortunately, all of iconv implements do support * "CP932" character-set, so we should use "SJIS" * instead if iconv_open failed. */ if (strcmp(tc, "CP932") == 0) sc->cd = iconv_open("SJIS", fc); else if (strcmp(fc, "CP932") == 0) sc->cd = iconv_open(tc, "SJIS"); } #if defined(_WIN32) && !defined(__CYGWIN__) /* * archive_mstring on Windows directly convert multi-bytes * into archive_wstring in order not to depend on locale - * so that you can do a I18N programing. This will be + * so that you can do a I18N programming. This will be * used only in archive_mstring_copy_mbs_len_l so far. */ if (flag & SCONV_FROM_CHARSET) { sc->cd_w = iconv_open("UTF-8", fc); if (sc->cd_w == (iconv_t)-1 && (sc->flag & SCONV_BEST_EFFORT)) { if (strcmp(fc, "CP932") == 0) sc->cd_w = iconv_open("UTF-8", "SJIS"); } } #endif /* _WIN32 && !__CYGWIN__ */ } #endif /* HAVE_ICONV */ sc->flag = flag; /* * Set up converters. */ setup_converter(sc); return (sc); } /* * Free a string conversion object. */ static void free_sconv_object(struct archive_string_conv *sc) { free(sc->from_charset); free(sc->to_charset); archive_string_free(&sc->utftmp); #if HAVE_ICONV if (sc->cd != (iconv_t)-1) iconv_close(sc->cd); if (sc->cd_w != (iconv_t)-1) iconv_close(sc->cd_w); #endif free(sc); } #if defined(_WIN32) && !defined(__CYGWIN__) static unsigned my_atoi(const char *p) { unsigned cp; cp = 0; while (*p) { if (*p >= '0' && *p <= '9') cp = cp * 10 + (*p - '0'); else return (-1); p++; } return (cp); } /* * Translate Charset name (as used by iconv) into CodePage (as used by Windows) * Return -1 if failed. * * Note: This translation code may be insufficient. */ static struct charset { const char *name; unsigned cp; } charsets[] = { /* MUST BE SORTED! */ {"ASCII", 1252}, {"ASMO-708", 708}, {"BIG5", 950}, {"CHINESE", 936}, {"CP367", 1252}, {"CP819", 1252}, {"CP1025", 21025}, {"DOS-720", 720}, {"DOS-862", 862}, {"EUC-CN", 51936}, {"EUC-JP", 51932}, {"EUC-KR", 949}, {"EUCCN", 51936}, {"EUCJP", 51932}, {"EUCKR", 949}, {"GB18030", 54936}, {"GB2312", 936}, {"HEBREW", 1255}, {"HZ-GB-2312", 52936}, {"IBM273", 20273}, {"IBM277", 20277}, {"IBM278", 20278}, {"IBM280", 20280}, {"IBM284", 20284}, {"IBM285", 20285}, {"IBM290", 20290}, {"IBM297", 20297}, {"IBM367", 1252}, {"IBM420", 20420}, {"IBM423", 20423}, {"IBM424", 20424}, {"IBM819", 1252}, {"IBM871", 20871}, {"IBM880", 20880}, {"IBM905", 20905}, {"IBM924", 20924}, {"ISO-8859-1", 28591}, {"ISO-8859-13", 28603}, {"ISO-8859-15", 28605}, {"ISO-8859-2", 28592}, {"ISO-8859-3", 28593}, {"ISO-8859-4", 28594}, {"ISO-8859-5", 28595}, {"ISO-8859-6", 28596}, {"ISO-8859-7", 28597}, {"ISO-8859-8", 28598}, {"ISO-8859-9", 28599}, {"ISO8859-1", 28591}, {"ISO8859-13", 28603}, {"ISO8859-15", 28605}, {"ISO8859-2", 28592}, {"ISO8859-3", 28593}, {"ISO8859-4", 28594}, {"ISO8859-5", 28595}, {"ISO8859-6", 28596}, {"ISO8859-7", 28597}, {"ISO8859-8", 28598}, {"ISO8859-9", 28599}, {"JOHAB", 1361}, {"KOI8-R", 20866}, {"KOI8-U", 21866}, {"KS_C_5601-1987", 949}, {"LATIN1", 1252}, {"LATIN2", 28592}, {"MACINTOSH", 10000}, {"SHIFT-JIS", 932}, {"SHIFT_JIS", 932}, {"SJIS", 932}, {"US", 1252}, {"US-ASCII", 1252}, {"UTF-16", 1200}, {"UTF-16BE", 1201}, {"UTF-16LE", 1200}, {"UTF-8", CP_UTF8}, {"X-EUROPA", 29001}, {"X-MAC-ARABIC", 10004}, {"X-MAC-CE", 10029}, {"X-MAC-CHINESEIMP", 10008}, {"X-MAC-CHINESETRAD", 10002}, {"X-MAC-CROATIAN", 10082}, {"X-MAC-CYRILLIC", 10007}, {"X-MAC-GREEK", 10006}, {"X-MAC-HEBREW", 10005}, {"X-MAC-ICELANDIC", 10079}, {"X-MAC-JAPANESE", 10001}, {"X-MAC-KOREAN", 10003}, {"X-MAC-ROMANIAN", 10010}, {"X-MAC-THAI", 10021}, {"X-MAC-TURKISH", 10081}, {"X-MAC-UKRAINIAN", 10017}, }; static unsigned make_codepage_from_charset(const char *charset) { char cs[16]; char *p; unsigned cp; int a, b; if (charset == NULL || strlen(charset) > 15) return -1; /* Copy name to uppercase. */ p = cs; while (*charset) { char c = *charset++; if (c >= 'a' && c <= 'z') c -= 'a' - 'A'; *p++ = c; } *p++ = '\0'; cp = -1; /* Look it up in the table first, so that we can easily * override CP367, which we map to 1252 instead of 367. */ a = 0; b = sizeof(charsets)/sizeof(charsets[0]); while (b > a) { int c = (b + a) / 2; int r = strcmp(charsets[c].name, cs); if (r < 0) a = c + 1; else if (r > 0) b = c; else return charsets[c].cp; } /* If it's not in the table, try to parse it. */ switch (*cs) { case 'C': if (cs[1] == 'P' && cs[2] >= '0' && cs[2] <= '9') { cp = my_atoi(cs + 2); } else if (strcmp(cs, "CP_ACP") == 0) cp = get_current_codepage(); else if (strcmp(cs, "CP_OEMCP") == 0) cp = get_current_oemcp(); break; case 'I': if (cs[1] == 'B' && cs[2] == 'M' && cs[3] >= '0' && cs[3] <= '9') { cp = my_atoi(cs + 3); } break; case 'W': if (strncmp(cs, "WINDOWS-", 8) == 0) { cp = my_atoi(cs + 8); if (cp != 874 && (cp < 1250 || cp > 1258)) cp = -1;/* This may invalid code. */ } break; } return (cp); } /* * Return ANSI Code Page of current locale set by setlocale(). */ static unsigned get_current_codepage(void) { char *locale, *p; unsigned cp; locale = setlocale(LC_CTYPE, NULL); if (locale == NULL) return (GetACP()); if (locale[0] == 'C' && locale[1] == '\0') return (CP_C_LOCALE); p = strrchr(locale, '.'); if (p == NULL) return (GetACP()); cp = my_atoi(p+1); if (cp <= 0) return (GetACP()); return (cp); } /* * Translation table between Locale Name and ACP/OEMCP. */ static struct { unsigned acp; unsigned ocp; const char *locale; } acp_ocp_map[] = { { 950, 950, "Chinese_Taiwan" }, { 936, 936, "Chinese_People's Republic of China" }, { 950, 950, "Chinese_Taiwan" }, { 1250, 852, "Czech_Czech Republic" }, { 1252, 850, "Danish_Denmark" }, { 1252, 850, "Dutch_Netherlands" }, { 1252, 850, "Dutch_Belgium" }, { 1252, 437, "English_United States" }, { 1252, 850, "English_Australia" }, { 1252, 850, "English_Canada" }, { 1252, 850, "English_New Zealand" }, { 1252, 850, "English_United Kingdom" }, { 1252, 437, "English_United States" }, { 1252, 850, "Finnish_Finland" }, { 1252, 850, "French_France" }, { 1252, 850, "French_Belgium" }, { 1252, 850, "French_Canada" }, { 1252, 850, "French_Switzerland" }, { 1252, 850, "German_Germany" }, { 1252, 850, "German_Austria" }, { 1252, 850, "German_Switzerland" }, { 1253, 737, "Greek_Greece" }, { 1250, 852, "Hungarian_Hungary" }, { 1252, 850, "Icelandic_Iceland" }, { 1252, 850, "Italian_Italy" }, { 1252, 850, "Italian_Switzerland" }, { 932, 932, "Japanese_Japan" }, { 949, 949, "Korean_Korea" }, { 1252, 850, "Norwegian (BokmOl)_Norway" }, { 1252, 850, "Norwegian (BokmOl)_Norway" }, { 1252, 850, "Norwegian-Nynorsk_Norway" }, { 1250, 852, "Polish_Poland" }, { 1252, 850, "Portuguese_Portugal" }, { 1252, 850, "Portuguese_Brazil" }, { 1251, 866, "Russian_Russia" }, { 1250, 852, "Slovak_Slovakia" }, { 1252, 850, "Spanish_Spain" }, { 1252, 850, "Spanish_Mexico" }, { 1252, 850, "Spanish_Spain" }, { 1252, 850, "Swedish_Sweden" }, { 1254, 857, "Turkish_Turkey" }, { 0, 0, NULL} }; /* * Return OEM Code Page of current locale set by setlocale(). */ static unsigned get_current_oemcp(void) { int i; char *locale, *p; size_t len; locale = setlocale(LC_CTYPE, NULL); if (locale == NULL) return (GetOEMCP()); if (locale[0] == 'C' && locale[1] == '\0') return (CP_C_LOCALE); p = strrchr(locale, '.'); if (p == NULL) return (GetOEMCP()); len = p - locale; for (i = 0; acp_ocp_map[i].acp; i++) { if (strncmp(acp_ocp_map[i].locale, locale, len) == 0) return (acp_ocp_map[i].ocp); } return (GetOEMCP()); } #else /* * POSIX platform does not use CodePage. */ static unsigned get_current_codepage(void) { return (-1);/* Unknown */ } static unsigned make_codepage_from_charset(const char *charset) { (void)charset; /* UNUSED */ return (-1);/* Unknown */ } static unsigned get_current_oemcp(void) { return (-1);/* Unknown */ } #endif /* defined(_WIN32) && !defined(__CYGWIN__) */ /* * Return a string conversion object. */ static struct archive_string_conv * get_sconv_object(struct archive *a, const char *fc, const char *tc, int flag) { struct archive_string_conv *sc; unsigned current_codepage; /* Check if we have made the sconv object. */ sc = find_sconv_object(a, fc, tc); if (sc != NULL) return (sc); if (a == NULL) current_codepage = get_current_codepage(); else current_codepage = a->current_codepage; sc = create_sconv_object(canonical_charset_name(fc), canonical_charset_name(tc), current_codepage, flag); if (sc == NULL) { if (a != NULL) archive_set_error(a, ENOMEM, "Could not allocate memory for " "a string conversion object"); return (NULL); } /* * If there is no converter for current string conversion object, * we cannot handle this conversion. */ if (sc->nconverter == 0) { if (a != NULL) { #if HAVE_ICONV archive_set_error(a, ARCHIVE_ERRNO_MISC, "iconv_open failed : Cannot handle ``%s''", (flag & SCONV_TO_CHARSET)?tc:fc); #else archive_set_error(a, ARCHIVE_ERRNO_MISC, "A character-set conversion not fully supported " "on this platform"); #endif } /* Failed; free a sconv object. */ free_sconv_object(sc); return (NULL); } /* * Success! */ if (a != NULL) add_sconv_object(a, sc); return (sc); } static const char * get_current_charset(struct archive *a) { const char *cur_charset; if (a == NULL) cur_charset = default_iconv_charset(""); else { cur_charset = default_iconv_charset(a->current_code); if (a->current_code == NULL) { a->current_code = strdup(cur_charset); a->current_codepage = get_current_codepage(); a->current_oemcp = get_current_oemcp(); } } return (cur_charset); } /* * Make and Return a string conversion object. * Return NULL if the platform does not support the specified conversion * and best_effort is 0. * If best_effort is set, A string conversion object must be returned * unless memory allocation for the object fails, but the conversion * might fail when non-ASCII code is found. */ struct archive_string_conv * archive_string_conversion_to_charset(struct archive *a, const char *charset, int best_effort) { int flag = SCONV_TO_CHARSET; if (best_effort) flag |= SCONV_BEST_EFFORT; return (get_sconv_object(a, get_current_charset(a), charset, flag)); } struct archive_string_conv * archive_string_conversion_from_charset(struct archive *a, const char *charset, int best_effort) { int flag = SCONV_FROM_CHARSET; if (best_effort) flag |= SCONV_BEST_EFFORT; return (get_sconv_object(a, charset, get_current_charset(a), flag)); } /* * archive_string_default_conversion_*_archive() are provided for Windows * platform because other archiver application use CP_OEMCP for * MultiByteToWideChar() and WideCharToMultiByte() for the filenames * in tar or zip files. But mbstowcs/wcstombs(CRT) usually use CP_ACP * unless you use setlocale(LC_ALL, ".OCP")(specify CP_OEMCP). * So we should make a string conversion between CP_ACP and CP_OEMCP - * for compatibillty. + * for compatibility. */ #if defined(_WIN32) && !defined(__CYGWIN__) struct archive_string_conv * archive_string_default_conversion_for_read(struct archive *a) { const char *cur_charset = get_current_charset(a); char oemcp[16]; /* NOTE: a check of cur_charset is unneeded but we need * that get_current_charset() has been surely called at * this time whatever C compiler optimized. */ if (cur_charset != NULL && (a->current_codepage == CP_C_LOCALE || a->current_codepage == a->current_oemcp)) return (NULL);/* no conversion. */ _snprintf(oemcp, sizeof(oemcp)-1, "CP%d", a->current_oemcp); /* Make sure a null termination must be set. */ oemcp[sizeof(oemcp)-1] = '\0'; return (get_sconv_object(a, oemcp, cur_charset, SCONV_FROM_CHARSET)); } struct archive_string_conv * archive_string_default_conversion_for_write(struct archive *a) { const char *cur_charset = get_current_charset(a); char oemcp[16]; /* NOTE: a check of cur_charset is unneeded but we need * that get_current_charset() has been surely called at * this time whatever C compiler optimized. */ if (cur_charset != NULL && (a->current_codepage == CP_C_LOCALE || a->current_codepage == a->current_oemcp)) return (NULL);/* no conversion. */ _snprintf(oemcp, sizeof(oemcp)-1, "CP%d", a->current_oemcp); /* Make sure a null termination must be set. */ oemcp[sizeof(oemcp)-1] = '\0'; return (get_sconv_object(a, cur_charset, oemcp, SCONV_TO_CHARSET)); } #else struct archive_string_conv * archive_string_default_conversion_for_read(struct archive *a) { (void)a; /* UNUSED */ return (NULL); } struct archive_string_conv * archive_string_default_conversion_for_write(struct archive *a) { (void)a; /* UNUSED */ return (NULL); } #endif /* * Dispose of all character conversion objects in the archive object. */ void archive_string_conversion_free(struct archive *a) { struct archive_string_conv *sc; struct archive_string_conv *sc_next; for (sc = a->sconv; sc != NULL; sc = sc_next) { sc_next = sc->next; free_sconv_object(sc); } a->sconv = NULL; free(a->current_code); a->current_code = NULL; } /* * Return a conversion charset name. */ const char * archive_string_conversion_charset_name(struct archive_string_conv *sc) { if (sc->flag & SCONV_TO_CHARSET) return (sc->to_charset); else return (sc->from_charset); } /* * Change the behavior of a string conversion. */ void archive_string_conversion_set_opt(struct archive_string_conv *sc, int opt) { switch (opt) { /* * A filename in UTF-8 was made with libarchive 2.x in a wrong * assumption that wchar_t was Unicode. * This option enables simulating the assumption in order to read * that filename correctly. */ case SCONV_SET_OPT_UTF8_LIBARCHIVE2X: #if (defined(_WIN32) && !defined(__CYGWIN__)) \ || defined(__STDC_ISO_10646__) || defined(__APPLE__) /* * Nothing to do for it since wchar_t on these platforms * is really Unicode. */ (void)sc; /* UNUSED */ #else if ((sc->flag & SCONV_UTF8_LIBARCHIVE_2) == 0) { sc->flag |= SCONV_UTF8_LIBARCHIVE_2; /* Set up string converters. */ setup_converter(sc); } #endif break; case SCONV_SET_OPT_NORMALIZATION_C: if ((sc->flag & SCONV_NORMALIZATION_C) == 0) { sc->flag |= SCONV_NORMALIZATION_C; sc->flag &= ~SCONV_NORMALIZATION_D; /* Set up string converters. */ setup_converter(sc); } break; case SCONV_SET_OPT_NORMALIZATION_D: #if defined(HAVE_ICONV) /* * If iconv will take the string, do not change the * setting of the normalization. */ if (!(sc->flag & SCONV_WIN_CP) && (sc->flag & (SCONV_FROM_UTF16 | SCONV_FROM_UTF8)) && !(sc->flag & (SCONV_TO_UTF16 | SCONV_TO_UTF8))) break; #endif if ((sc->flag & SCONV_NORMALIZATION_D) == 0) { sc->flag |= SCONV_NORMALIZATION_D; sc->flag &= ~SCONV_NORMALIZATION_C; /* Set up string converters. */ setup_converter(sc); } break; default: break; } } /* * * Copy one archive_string to another in locale conversion. * * archive_strncat_l(); * archive_strncpy_l(); * */ static size_t mbsnbytes(const void *_p, size_t n) { size_t s; const char *p, *pp; if (_p == NULL) return (0); p = (const char *)_p; /* Like strlen(p), except won't examine positions beyond p[n]. */ s = 0; pp = p; while (s < n && *pp) { pp++; s++; } return (s); } static size_t utf16nbytes(const void *_p, size_t n) { size_t s; const char *p, *pp; if (_p == NULL) return (0); p = (const char *)_p; /* Like strlen(p), except won't examine positions beyond p[n]. */ s = 0; pp = p; n >>= 1; while (s < n && (pp[0] || pp[1])) { pp += 2; s++; } return (s<<1); } int archive_strncpy_l(struct archive_string *as, const void *_p, size_t n, struct archive_string_conv *sc) { as->length = 0; return (archive_strncat_l(as, _p, n, sc)); } int archive_strncat_l(struct archive_string *as, const void *_p, size_t n, struct archive_string_conv *sc) { const void *s; size_t length = 0; int i, r = 0, r2; if (_p != NULL && n > 0) { if (sc != NULL && (sc->flag & SCONV_FROM_UTF16)) length = utf16nbytes(_p, n); else length = mbsnbytes(_p, n); } /* We must allocate memory even if there is no data for conversion * or copy. This simulates archive_string_append behavior. */ if (length == 0) { int tn = 1; if (sc != NULL && (sc->flag & SCONV_TO_UTF16)) tn = 2; if (archive_string_ensure(as, as->length + tn) == NULL) return (-1); as->s[as->length] = 0; if (tn == 2) as->s[as->length+1] = 0; return (0); } /* * If sc is NULL, we just make a copy. */ if (sc == NULL) { if (archive_string_append(as, _p, length) == NULL) return (-1);/* No memory */ return (0); } s = _p; i = 0; if (sc->nconverter > 1) { sc->utftmp.length = 0; r2 = sc->converter[0](&(sc->utftmp), s, length, sc); if (r2 != 0 && errno == ENOMEM) return (r2); if (r > r2) r = r2; s = sc->utftmp.s; length = sc->utftmp.length; ++i; } r2 = sc->converter[i](as, s, length, sc); if (r > r2) r = r2; return (r); } #if HAVE_ICONV /* * Return -1 if conversion fails. */ static int iconv_strncat_in_locale(struct archive_string *as, const void *_p, size_t length, struct archive_string_conv *sc) { ICONV_CONST char *itp; size_t remaining; iconv_t cd; char *outp; size_t avail, bs; int return_value = 0; /* success */ int to_size, from_size; if (sc->flag & SCONV_TO_UTF16) to_size = 2; else to_size = 1; if (sc->flag & SCONV_FROM_UTF16) from_size = 2; else from_size = 1; if (archive_string_ensure(as, as->length + length*2+to_size) == NULL) return (-1); cd = sc->cd; itp = (char *)(uintptr_t)_p; remaining = length; outp = as->s + as->length; avail = as->buffer_length - as->length - to_size; while (remaining >= (size_t)from_size) { size_t result = iconv(cd, &itp, &remaining, &outp, &avail); if (result != (size_t)-1) break; /* Conversion completed. */ if (errno == EILSEQ || errno == EINVAL) { /* * If an output charset is UTF-8 or UTF-16BE/LE, * unknown character should be U+FFFD * (replacement character). */ if (sc->flag & (SCONV_TO_UTF8 | SCONV_TO_UTF16)) { size_t rbytes; if (sc->flag & SCONV_TO_UTF8) rbytes = sizeof(utf8_replacement_char); else rbytes = 2; if (avail < rbytes) { as->length = outp - as->s; bs = as->buffer_length + (remaining * to_size) + rbytes; if (NULL == archive_string_ensure(as, bs)) return (-1); outp = as->s + as->length; avail = as->buffer_length - as->length - to_size; } if (sc->flag & SCONV_TO_UTF8) memcpy(outp, utf8_replacement_char, sizeof(utf8_replacement_char)); else if (sc->flag & SCONV_TO_UTF16BE) archive_be16enc(outp, UNICODE_R_CHAR); else archive_le16enc(outp, UNICODE_R_CHAR); outp += rbytes; avail -= rbytes; } else { /* Skip the illegal input bytes. */ *outp++ = '?'; avail--; } itp += from_size; remaining -= from_size; return_value = -1; /* failure */ } else { /* E2BIG no output buffer, * Increase an output buffer. */ as->length = outp - as->s; bs = as->buffer_length + remaining * 2; if (NULL == archive_string_ensure(as, bs)) return (-1); outp = as->s + as->length; avail = as->buffer_length - as->length - to_size; } } as->length = outp - as->s; as->s[as->length] = 0; if (to_size == 2) as->s[as->length+1] = 0; return (return_value); } #endif /* HAVE_ICONV */ #if defined(_WIN32) && !defined(__CYGWIN__) /* * Translate a string from a some CodePage to an another CodePage by * Windows APIs, and copy the result. Return -1 if conversion fails. */ static int strncat_in_codepage(struct archive_string *as, const void *_p, size_t length, struct archive_string_conv *sc) { const char *s = (const char *)_p; struct archive_wstring aws; size_t l; int r, saved_flag; archive_string_init(&aws); saved_flag = sc->flag; sc->flag &= ~(SCONV_NORMALIZATION_D | SCONV_NORMALIZATION_C); r = archive_wstring_append_from_mbs_in_codepage(&aws, s, length, sc); sc->flag = saved_flag; if (r != 0) { archive_wstring_free(&aws); if (errno != ENOMEM) archive_string_append(as, s, length); return (-1); } l = as->length; r = archive_string_append_from_wcs_in_codepage( as, aws.s, aws.length, sc); if (r != 0 && errno != ENOMEM && l == as->length) archive_string_append(as, s, length); archive_wstring_free(&aws); return (r); } /* * Test whether MBS ==> WCS is okay. */ static int invalid_mbs(const void *_p, size_t n, struct archive_string_conv *sc) { const char *p = (const char *)_p; unsigned codepage; DWORD mbflag = MB_ERR_INVALID_CHARS; if (sc->flag & SCONV_FROM_CHARSET) codepage = sc->to_cp; else codepage = sc->from_cp; if (codepage == CP_C_LOCALE) return (0); if (codepage != CP_UTF8) mbflag |= MB_PRECOMPOSED; if (MultiByteToWideChar(codepage, mbflag, p, (int)n, NULL, 0) == 0) return (-1); /* Invalid */ return (0); /* Okay */ } #else /* * Test whether MBS ==> WCS is okay. */ static int invalid_mbs(const void *_p, size_t n, struct archive_string_conv *sc) { const char *p = (const char *)_p; size_t r; #if HAVE_MBRTOWC mbstate_t shift_state; memset(&shift_state, 0, sizeof(shift_state)); #else /* Clear the shift state before starting. */ mbtowc(NULL, NULL, 0); #endif while (n) { wchar_t wc; #if HAVE_MBRTOWC r = mbrtowc(&wc, p, n, &shift_state); #else r = mbtowc(&wc, p, n); #endif if (r == (size_t)-1 || r == (size_t)-2) return (-1);/* Invalid. */ if (r == 0) break; p += r; n -= r; } (void)sc; /* UNUSED */ return (0); /* All Okey. */ } #endif /* defined(_WIN32) && !defined(__CYGWIN__) */ /* * Basically returns -1 because we cannot make a conversion of charset * without iconv but in some cases this would return 0. * Returns 0 if all copied characters are ASCII. * Returns 0 if both from-locale and to-locale are the same and those * can be WCS with no error. */ static int best_effort_strncat_in_locale(struct archive_string *as, const void *_p, size_t length, struct archive_string_conv *sc) { size_t remaining; const uint8_t *itp; int return_value = 0; /* success */ /* * If both from-locale and to-locale is the same, this makes a copy. * And then this checks all copied MBS can be WCS if so returns 0. */ if (sc->same) { if (archive_string_append(as, _p, length) == NULL) return (-1);/* No memory */ return (invalid_mbs(_p, length, sc)); } /* * If a character is ASCII, this just copies it. If not, this - * assigns '?' charater instead but in UTF-8 locale this assigns + * assigns '?' character instead but in UTF-8 locale this assigns * byte sequence 0xEF 0xBD 0xBD, which are code point U+FFFD, * a Replacement Character in Unicode. */ remaining = length; itp = (const uint8_t *)_p; while (*itp && remaining > 0) { if (*itp > 127) { // Non-ASCII: Substitute with suitable replacement if (sc->flag & SCONV_TO_UTF8) { if (archive_string_append(as, utf8_replacement_char, sizeof(utf8_replacement_char)) == NULL) { __archive_errx(1, "Out of memory"); } } else { archive_strappend_char(as, '?'); } return_value = -1; } else { archive_strappend_char(as, *itp); } ++itp; } return (return_value); } /* * Unicode conversion functions. * - UTF-8 <===> UTF-8 in removing surrogate pairs. * - UTF-8 NFD ===> UTF-8 NFC in removing surrogate pairs. * - UTF-8 made by libarchive 2.x ===> UTF-8. * - UTF-16BE <===> UTF-8. * */ /* * Utility to convert a single UTF-8 sequence. * * Usually return used bytes, return used byte in negative value when * a unicode character is replaced with U+FFFD. * See also http://unicode.org/review/pr-121.html Public Review Issue #121 * Recommended Practice for Replacement Characters. */ static int _utf8_to_unicode(uint32_t *pwc, const char *s, size_t n) { static const char utf8_count[256] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 00 - 0F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 10 - 1F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 20 - 2F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 30 - 3F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 40 - 4F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 50 - 5F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 60 - 6F */ 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,/* 70 - 7F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 80 - 8F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* 90 - 9F */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* A0 - AF */ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,/* B0 - BF */ 0, 0, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* C0 - CF */ 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,/* D0 - DF */ 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,/* E0 - EF */ 4, 4, 4, 4, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 /* F0 - FF */ }; int ch, i; int cnt; uint32_t wc; /* Sanity check. */ if (n == 0) return (0); /* * Decode 1-4 bytes depending on the value of the first byte. */ ch = (unsigned char)*s; if (ch == 0) return (0); /* Standard: return 0 for end-of-string. */ cnt = utf8_count[ch]; /* Invalid sequence or there are not plenty bytes. */ if ((int)n < cnt) { cnt = (int)n; for (i = 1; i < cnt; i++) { if ((s[i] & 0xc0) != 0x80) { cnt = i; break; } } goto invalid_sequence; } /* Make a Unicode code point from a single UTF-8 sequence. */ switch (cnt) { case 1: /* 1 byte sequence. */ *pwc = ch & 0x7f; return (cnt); case 2: /* 2 bytes sequence. */ if ((s[1] & 0xc0) != 0x80) { cnt = 1; goto invalid_sequence; } *pwc = ((ch & 0x1f) << 6) | (s[1] & 0x3f); return (cnt); case 3: /* 3 bytes sequence. */ if ((s[1] & 0xc0) != 0x80) { cnt = 1; goto invalid_sequence; } if ((s[2] & 0xc0) != 0x80) { cnt = 2; goto invalid_sequence; } wc = ((ch & 0x0f) << 12) | ((s[1] & 0x3f) << 6) | (s[2] & 0x3f); if (wc < 0x800) goto invalid_sequence;/* Overlong sequence. */ break; case 4: /* 4 bytes sequence. */ if ((s[1] & 0xc0) != 0x80) { cnt = 1; goto invalid_sequence; } if ((s[2] & 0xc0) != 0x80) { cnt = 2; goto invalid_sequence; } if ((s[3] & 0xc0) != 0x80) { cnt = 3; goto invalid_sequence; } wc = ((ch & 0x07) << 18) | ((s[1] & 0x3f) << 12) | ((s[2] & 0x3f) << 6) | (s[3] & 0x3f); if (wc < 0x10000) goto invalid_sequence;/* Overlong sequence. */ break; default: /* Others are all invalid sequence. */ if (ch == 0xc0 || ch == 0xc1) cnt = 2; else if (ch >= 0xf5 && ch <= 0xf7) cnt = 4; else if (ch >= 0xf8 && ch <= 0xfb) cnt = 5; else if (ch == 0xfc || ch == 0xfd) cnt = 6; else cnt = 1; if ((int)n < cnt) cnt = (int)n; for (i = 1; i < cnt; i++) { if ((s[i] & 0xc0) != 0x80) { cnt = i; break; } } goto invalid_sequence; } /* The code point larger than 0x10FFFF is not legal * Unicode values. */ if (wc > UNICODE_MAX) goto invalid_sequence; /* Correctly gets a Unicode, returns used bytes. */ *pwc = wc; return (cnt); invalid_sequence: *pwc = UNICODE_R_CHAR;/* set the Replacement Character instead. */ return (cnt * -1); } static int utf8_to_unicode(uint32_t *pwc, const char *s, size_t n) { int cnt; cnt = _utf8_to_unicode(pwc, s, n); /* Any of Surrogate pair is not legal Unicode values. */ if (cnt == 3 && IS_SURROGATE_PAIR_LA(*pwc)) return (-3); return (cnt); } static inline uint32_t combine_surrogate_pair(uint32_t uc, uint32_t uc2) { uc -= 0xD800; uc *= 0x400; uc += uc2 - 0xDC00; uc += 0x10000; return (uc); } /* * Convert a single UTF-8/CESU-8 sequence to a Unicode code point in * removing surrogate pairs. * * CESU-8: The Compatibility Encoding Scheme for UTF-16. * * Usually return used bytes, return used byte in negative value when * a unicode character is replaced with U+FFFD. */ static int cesu8_to_unicode(uint32_t *pwc, const char *s, size_t n) { uint32_t wc = 0; int cnt; cnt = _utf8_to_unicode(&wc, s, n); if (cnt == 3 && IS_HIGH_SURROGATE_LA(wc)) { uint32_t wc2 = 0; if (n - 3 < 3) { /* Invalid byte sequence. */ goto invalid_sequence; } cnt = _utf8_to_unicode(&wc2, s+3, n-3); if (cnt != 3 || !IS_LOW_SURROGATE_LA(wc2)) { /* Invalid byte sequence. */ goto invalid_sequence; } wc = combine_surrogate_pair(wc, wc2); cnt = 6; } else if (cnt == 3 && IS_LOW_SURROGATE_LA(wc)) { /* Invalid byte sequence. */ goto invalid_sequence; } *pwc = wc; return (cnt); invalid_sequence: *pwc = UNICODE_R_CHAR;/* set the Replacement Character instead. */ if (cnt > 0) cnt *= -1; return (cnt); } /* * Convert a Unicode code point to a single UTF-8 sequence. * * NOTE:This function does not check if the Unicode is legal or not. * Please you definitely check it before calling this. */ static size_t unicode_to_utf8(char *p, size_t remaining, uint32_t uc) { char *_p = p; /* Invalid Unicode char maps to Replacement character */ if (uc > UNICODE_MAX) uc = UNICODE_R_CHAR; /* Translate code point to UTF8 */ if (uc <= 0x7f) { if (remaining == 0) return (0); *p++ = (char)uc; } else if (uc <= 0x7ff) { if (remaining < 2) return (0); *p++ = 0xc0 | ((uc >> 6) & 0x1f); *p++ = 0x80 | (uc & 0x3f); } else if (uc <= 0xffff) { if (remaining < 3) return (0); *p++ = 0xe0 | ((uc >> 12) & 0x0f); *p++ = 0x80 | ((uc >> 6) & 0x3f); *p++ = 0x80 | (uc & 0x3f); } else { if (remaining < 4) return (0); *p++ = 0xf0 | ((uc >> 18) & 0x07); *p++ = 0x80 | ((uc >> 12) & 0x3f); *p++ = 0x80 | ((uc >> 6) & 0x3f); *p++ = 0x80 | (uc & 0x3f); } return (p - _p); } static int utf16be_to_unicode(uint32_t *pwc, const char *s, size_t n) { return (utf16_to_unicode(pwc, s, n, 1)); } static int utf16le_to_unicode(uint32_t *pwc, const char *s, size_t n) { return (utf16_to_unicode(pwc, s, n, 0)); } static int utf16_to_unicode(uint32_t *pwc, const char *s, size_t n, int be) { const char *utf16 = s; unsigned uc; if (n == 0) return (0); if (n == 1) { /* set the Replacement Character instead. */ *pwc = UNICODE_R_CHAR; return (-1); } if (be) uc = archive_be16dec(utf16); else uc = archive_le16dec(utf16); utf16 += 2; /* If this is a surrogate pair, assemble the full code point.*/ if (IS_HIGH_SURROGATE_LA(uc)) { unsigned uc2; if (n >= 4) { if (be) uc2 = archive_be16dec(utf16); else uc2 = archive_le16dec(utf16); } else uc2 = 0; if (IS_LOW_SURROGATE_LA(uc2)) { uc = combine_surrogate_pair(uc, uc2); utf16 += 2; } else { /* Undescribed code point should be U+FFFD * (replacement character). */ *pwc = UNICODE_R_CHAR; return (-2); } } /* * Surrogate pair values(0xd800 through 0xdfff) are only - * used by UTF-16, so, after above culculation, the code + * used by UTF-16, so, after above calculation, the code * must not be surrogate values, and Unicode has no codes * larger than 0x10ffff. Thus, those are not legal Unicode * values. */ if (IS_SURROGATE_PAIR_LA(uc) || uc > UNICODE_MAX) { /* Undescribed code point should be U+FFFD * (replacement character). */ *pwc = UNICODE_R_CHAR; return (((int)(utf16 - s)) * -1); } *pwc = uc; return ((int)(utf16 - s)); } static size_t unicode_to_utf16be(char *p, size_t remaining, uint32_t uc) { char *utf16 = p; if (uc > 0xffff) { /* We have a code point that won't fit into a * wchar_t; convert it to a surrogate pair. */ if (remaining < 4) return (0); uc -= 0x10000; archive_be16enc(utf16, ((uc >> 10) & 0x3ff) + 0xD800); archive_be16enc(utf16+2, (uc & 0x3ff) + 0xDC00); return (4); } else { if (remaining < 2) return (0); archive_be16enc(utf16, uc); return (2); } } static size_t unicode_to_utf16le(char *p, size_t remaining, uint32_t uc) { char *utf16 = p; if (uc > 0xffff) { /* We have a code point that won't fit into a * wchar_t; convert it to a surrogate pair. */ if (remaining < 4) return (0); uc -= 0x10000; archive_le16enc(utf16, ((uc >> 10) & 0x3ff) + 0xD800); archive_le16enc(utf16+2, (uc & 0x3ff) + 0xDC00); return (4); } else { if (remaining < 2) return (0); archive_le16enc(utf16, uc); return (2); } } /* * Copy UTF-8 string in checking surrogate pair. * If any surrogate pair are found, it would be canonicalized. */ static int strncat_from_utf8_to_utf8(struct archive_string *as, const void *_p, size_t len, struct archive_string_conv *sc) { const char *s; char *p, *endp; int n, ret = 0; (void)sc; /* UNUSED */ if (archive_string_ensure(as, as->length + len + 1) == NULL) return (-1); s = (const char *)_p; p = as->s + as->length; endp = as->s + as->buffer_length -1; do { uint32_t uc; const char *ss = s; size_t w; /* * Forward byte sequence until a conversion of that is needed. */ while ((n = utf8_to_unicode(&uc, s, len)) > 0) { s += n; len -= n; } if (ss < s) { if (p + (s - ss) > endp) { as->length = p - as->s; if (archive_string_ensure(as, as->buffer_length + len + 1) == NULL) return (-1); p = as->s + as->length; endp = as->s + as->buffer_length -1; } memcpy(p, ss, s - ss); p += s - ss; } /* * If n is negative, current byte sequence needs a replacement. */ if (n < 0) { if (n == -3 && IS_SURROGATE_PAIR_LA(uc)) { /* Current byte sequence may be CESU-8. */ n = cesu8_to_unicode(&uc, s, len); } if (n < 0) { ret = -1; n *= -1;/* Use a replaced unicode character. */ } /* Rebuild UTF-8 byte sequence. */ while ((w = unicode_to_utf8(p, endp - p, uc)) == 0) { as->length = p - as->s; if (archive_string_ensure(as, as->buffer_length + len + 1) == NULL) return (-1); p = as->s + as->length; endp = as->s + as->buffer_length -1; } p += w; s += n; len -= n; } } while (n > 0); as->length = p - as->s; as->s[as->length] = '\0'; return (ret); } static int archive_string_append_unicode(struct archive_string *as, const void *_p, size_t len, struct archive_string_conv *sc) { const char *s; char *p, *endp; uint32_t uc; size_t w; int n, ret = 0, ts, tm; int (*parse)(uint32_t *, const char *, size_t); size_t (*unparse)(char *, size_t, uint32_t); if (sc->flag & SCONV_TO_UTF16BE) { unparse = unicode_to_utf16be; ts = 2; } else if (sc->flag & SCONV_TO_UTF16LE) { unparse = unicode_to_utf16le; ts = 2; } else if (sc->flag & SCONV_TO_UTF8) { unparse = unicode_to_utf8; ts = 1; } else { /* * This case is going to be converted to another * character-set through iconv. */ if (sc->flag & SCONV_FROM_UTF16BE) { unparse = unicode_to_utf16be; ts = 2; } else if (sc->flag & SCONV_FROM_UTF16LE) { unparse = unicode_to_utf16le; ts = 2; } else { unparse = unicode_to_utf8; ts = 1; } } if (sc->flag & SCONV_FROM_UTF16BE) { parse = utf16be_to_unicode; tm = 1; } else if (sc->flag & SCONV_FROM_UTF16LE) { parse = utf16le_to_unicode; tm = 1; } else { parse = cesu8_to_unicode; tm = ts; } if (archive_string_ensure(as, as->length + len * tm + ts) == NULL) return (-1); s = (const char *)_p; p = as->s + as->length; endp = as->s + as->buffer_length - ts; while ((n = parse(&uc, s, len)) != 0) { if (n < 0) { /* Use a replaced unicode character. */ n *= -1; ret = -1; } s += n; len -= n; while ((w = unparse(p, endp - p, uc)) == 0) { /* There is not enough output buffer so * we have to expand it. */ as->length = p - as->s; if (archive_string_ensure(as, as->buffer_length + len * tm + ts) == NULL) return (-1); p = as->s + as->length; endp = as->s + as->buffer_length - ts; } p += w; } as->length = p - as->s; as->s[as->length] = '\0'; if (ts == 2) as->s[as->length+1] = '\0'; return (ret); } /* * Following Constants for Hangul compositions this information comes from * Unicode Standard Annex #15 http://unicode.org/reports/tr15/ */ #define HC_SBASE 0xAC00 #define HC_LBASE 0x1100 #define HC_VBASE 0x1161 #define HC_TBASE 0x11A7 #define HC_LCOUNT 19 #define HC_VCOUNT 21 #define HC_TCOUNT 28 #define HC_NCOUNT (HC_VCOUNT * HC_TCOUNT) #define HC_SCOUNT (HC_LCOUNT * HC_NCOUNT) static uint32_t get_nfc(uint32_t uc, uint32_t uc2) { int t, b; t = 0; b = sizeof(u_composition_table)/sizeof(u_composition_table[0]) -1; while (b >= t) { int m = (t + b) / 2; if (u_composition_table[m].cp1 < uc) t = m + 1; else if (u_composition_table[m].cp1 > uc) b = m - 1; else if (u_composition_table[m].cp2 < uc2) t = m + 1; else if (u_composition_table[m].cp2 > uc2) b = m - 1; else return (u_composition_table[m].nfc); } return (0); } #define FDC_MAX 10 /* The maximum number of Following Decomposable * Characters. */ /* * Update first code point. */ #define UPDATE_UC(new_uc) do { \ uc = new_uc; \ ucptr = NULL; \ } while (0) /* * Replace first code point with second code point. */ #define REPLACE_UC_WITH_UC2() do { \ uc = uc2; \ ucptr = uc2ptr; \ n = n2; \ } while (0) #define EXPAND_BUFFER() do { \ as->length = p - as->s; \ if (archive_string_ensure(as, \ as->buffer_length + len * tm + ts) == NULL)\ return (-1); \ p = as->s + as->length; \ endp = as->s + as->buffer_length - ts; \ } while (0) #define UNPARSE(p, endp, uc) do { \ while ((w = unparse(p, (endp) - (p), uc)) == 0) {\ EXPAND_BUFFER(); \ } \ p += w; \ } while (0) /* * Write first code point. * If the code point has not be changed from its original code, * this just copies it from its original buffer pointer. * If not, this converts it to UTF-8 byte sequence and copies it. */ #define WRITE_UC() do { \ if (ucptr) { \ if (p + n > endp) \ EXPAND_BUFFER(); \ switch (n) { \ case 4: \ *p++ = *ucptr++; \ /* FALL THROUGH */ \ case 3: \ *p++ = *ucptr++; \ /* FALL THROUGH */ \ case 2: \ *p++ = *ucptr++; \ /* FALL THROUGH */ \ case 1: \ *p++ = *ucptr; \ break; \ } \ ucptr = NULL; \ } else { \ UNPARSE(p, endp, uc); \ } \ } while (0) /* * Collect following decomposable code points. */ #define COLLECT_CPS(start) do { \ int _i; \ for (_i = start; _i < FDC_MAX ; _i++) { \ nx = parse(&ucx[_i], s, len); \ if (nx <= 0) \ break; \ cx = CCC(ucx[_i]); \ if (cl >= cx && cl != 228 && cx != 228)\ break; \ s += nx; \ len -= nx; \ cl = cx; \ ccx[_i] = cx; \ } \ if (_i >= FDC_MAX) { \ ret = -1; \ ucx_size = FDC_MAX; \ } else \ ucx_size = _i; \ } while (0) /* * Normalize UTF-8/UTF-16BE characters to Form C and copy the result. * - * TODO: Convert composition exclusions,which are never converted + * TODO: Convert composition exclusions, which are never converted * from NFC,NFD,NFKC and NFKD, to Form C. */ static int archive_string_normalize_C(struct archive_string *as, const void *_p, size_t len, struct archive_string_conv *sc) { const char *s = (const char *)_p; char *p, *endp; uint32_t uc, uc2; size_t w; int always_replace, n, n2, ret = 0, spair, ts, tm; int (*parse)(uint32_t *, const char *, size_t); size_t (*unparse)(char *, size_t, uint32_t); always_replace = 1; ts = 1;/* text size. */ if (sc->flag & SCONV_TO_UTF16BE) { unparse = unicode_to_utf16be; ts = 2; if (sc->flag & SCONV_FROM_UTF16BE) always_replace = 0; } else if (sc->flag & SCONV_TO_UTF16LE) { unparse = unicode_to_utf16le; ts = 2; if (sc->flag & SCONV_FROM_UTF16LE) always_replace = 0; } else if (sc->flag & SCONV_TO_UTF8) { unparse = unicode_to_utf8; if (sc->flag & SCONV_FROM_UTF8) always_replace = 0; } else { /* * This case is going to be converted to another * character-set through iconv. */ always_replace = 0; if (sc->flag & SCONV_FROM_UTF16BE) { unparse = unicode_to_utf16be; ts = 2; } else if (sc->flag & SCONV_FROM_UTF16LE) { unparse = unicode_to_utf16le; ts = 2; } else { unparse = unicode_to_utf8; } } if (sc->flag & SCONV_FROM_UTF16BE) { parse = utf16be_to_unicode; tm = 1; spair = 4;/* surrogate pair size in UTF-16. */ } else if (sc->flag & SCONV_FROM_UTF16LE) { parse = utf16le_to_unicode; tm = 1; spair = 4;/* surrogate pair size in UTF-16. */ } else { parse = cesu8_to_unicode; tm = ts; spair = 6;/* surrogate pair size in UTF-8. */ } if (archive_string_ensure(as, as->length + len * tm + ts) == NULL) return (-1); p = as->s + as->length; endp = as->s + as->buffer_length - ts; while ((n = parse(&uc, s, len)) != 0) { const char *ucptr, *uc2ptr; if (n < 0) { /* Use a replaced unicode character. */ UNPARSE(p, endp, uc); s += n*-1; len -= n*-1; ret = -1; continue; } else if (n == spair || always_replace) /* uc is converted from a surrogate pair. * this should be treated as a changed code. */ ucptr = NULL; else ucptr = s; s += n; len -= n; /* Read second code point. */ while ((n2 = parse(&uc2, s, len)) > 0) { uint32_t ucx[FDC_MAX]; int ccx[FDC_MAX]; int cl, cx, i, nx, ucx_size; int LIndex,SIndex; uint32_t nfc; if (n2 == spair || always_replace) /* uc2 is converted from a surrogate pair. * this should be treated as a changed code. */ uc2ptr = NULL; else uc2ptr = s; s += n2; len -= n2; /* * If current second code point is out of decomposable * code points, finding compositions is unneeded. */ if (!IS_DECOMPOSABLE_BLOCK(uc2)) { WRITE_UC(); REPLACE_UC_WITH_UC2(); continue; } /* * Try to combine current code points. */ /* * We have to combine Hangul characters according to * http://uniicode.org/reports/tr15/#Hangul */ if (0 <= (LIndex = uc - HC_LBASE) && LIndex < HC_LCOUNT) { /* * Hangul Composition. * 1. Two current code points are L and V. */ int VIndex = uc2 - HC_VBASE; if (0 <= VIndex && VIndex < HC_VCOUNT) { /* Make syllable of form LV. */ UPDATE_UC(HC_SBASE + (LIndex * HC_VCOUNT + VIndex) * HC_TCOUNT); } else { WRITE_UC(); REPLACE_UC_WITH_UC2(); } continue; } else if (0 <= (SIndex = uc - HC_SBASE) && SIndex < HC_SCOUNT && (SIndex % HC_TCOUNT) == 0) { /* * Hangul Composition. * 2. Two current code points are LV and T. */ int TIndex = uc2 - HC_TBASE; if (0 < TIndex && TIndex < HC_TCOUNT) { /* Make syllable of form LVT. */ UPDATE_UC(uc + TIndex); } else { WRITE_UC(); REPLACE_UC_WITH_UC2(); } continue; } else if ((nfc = get_nfc(uc, uc2)) != 0) { /* A composition to current code points * is found. */ UPDATE_UC(nfc); continue; } else if ((cl = CCC(uc2)) == 0) { /* Clearly 'uc2' the second code point is not * a decomposable code. */ WRITE_UC(); REPLACE_UC_WITH_UC2(); continue; } /* * Collect following decomposable code points. */ cx = 0; ucx[0] = uc2; ccx[0] = cl; COLLECT_CPS(1); /* * Find a composed code in the collected code points. */ i = 1; while (i < ucx_size) { int j; if ((nfc = get_nfc(uc, ucx[i])) == 0) { i++; continue; } /* * nfc is composed of uc and ucx[i]. */ UPDATE_UC(nfc); /* * Remove ucx[i] by shifting * following code points. */ for (j = i; j+1 < ucx_size; j++) { ucx[j] = ucx[j+1]; ccx[j] = ccx[j+1]; } ucx_size --; /* * Collect following code points blocked * by ucx[i] the removed code point. */ if (ucx_size > 0 && i == ucx_size && nx > 0 && cx == cl) { cl = ccx[ucx_size-1]; COLLECT_CPS(ucx_size); } /* * Restart finding a composed code with * the updated uc from the top of the * collected code points. */ i = 0; } /* * Apparently the current code points are not * decomposed characters or already composed. */ WRITE_UC(); for (i = 0; i < ucx_size; i++) UNPARSE(p, endp, ucx[i]); /* * Flush out remaining canonical combining characters. */ if (nx > 0 && cx == cl && len > 0) { while ((nx = parse(&ucx[0], s, len)) > 0) { cx = CCC(ucx[0]); if (cl > cx) break; s += nx; len -= nx; cl = cx; UNPARSE(p, endp, ucx[0]); } } break; } if (n2 < 0) { WRITE_UC(); /* Use a replaced unicode character. */ UNPARSE(p, endp, uc2); s += n2*-1; len -= n2*-1; ret = -1; continue; } else if (n2 == 0) { WRITE_UC(); break; } } as->length = p - as->s; as->s[as->length] = '\0'; if (ts == 2) as->s[as->length+1] = '\0'; return (ret); } static int get_nfd(uint32_t *cp1, uint32_t *cp2, uint32_t uc) { int t, b; /* * These are not converted to NFD on Mac OS. */ if ((uc >= 0x2000 && uc <= 0x2FFF) || (uc >= 0xF900 && uc <= 0xFAFF) || (uc >= 0x2F800 && uc <= 0x2FAFF)) return (0); /* * Those code points are not converted to NFD on Mac OS. * I do not know the reason because it is undocumented. * NFC NFD * 1109A ==> 11099 110BA * 1109C ==> 1109B 110BA * 110AB ==> 110A5 110BA */ if (uc == 0x1109A || uc == 0x1109C || uc == 0x110AB) return (0); t = 0; b = sizeof(u_decomposition_table)/sizeof(u_decomposition_table[0]) -1; while (b >= t) { int m = (t + b) / 2; if (u_decomposition_table[m].nfc < uc) t = m + 1; else if (u_decomposition_table[m].nfc > uc) b = m - 1; else { *cp1 = u_decomposition_table[m].cp1; *cp2 = u_decomposition_table[m].cp2; return (1); } } return (0); } #define REPLACE_UC_WITH(cp) do { \ uc = cp; \ ucptr = NULL; \ } while (0) /* * Normalize UTF-8 characters to Form D and copy the result. */ static int archive_string_normalize_D(struct archive_string *as, const void *_p, size_t len, struct archive_string_conv *sc) { const char *s = (const char *)_p; char *p, *endp; uint32_t uc, uc2; size_t w; int always_replace, n, n2, ret = 0, spair, ts, tm; int (*parse)(uint32_t *, const char *, size_t); size_t (*unparse)(char *, size_t, uint32_t); always_replace = 1; ts = 1;/* text size. */ if (sc->flag & SCONV_TO_UTF16BE) { unparse = unicode_to_utf16be; ts = 2; if (sc->flag & SCONV_FROM_UTF16BE) always_replace = 0; } else if (sc->flag & SCONV_TO_UTF16LE) { unparse = unicode_to_utf16le; ts = 2; if (sc->flag & SCONV_FROM_UTF16LE) always_replace = 0; } else if (sc->flag & SCONV_TO_UTF8) { unparse = unicode_to_utf8; if (sc->flag & SCONV_FROM_UTF8) always_replace = 0; } else { /* * This case is going to be converted to another * character-set through iconv. */ always_replace = 0; if (sc->flag & SCONV_FROM_UTF16BE) { unparse = unicode_to_utf16be; ts = 2; } else if (sc->flag & SCONV_FROM_UTF16LE) { unparse = unicode_to_utf16le; ts = 2; } else { unparse = unicode_to_utf8; } } if (sc->flag & SCONV_FROM_UTF16BE) { parse = utf16be_to_unicode; tm = 1; spair = 4;/* surrogate pair size in UTF-16. */ } else if (sc->flag & SCONV_FROM_UTF16LE) { parse = utf16le_to_unicode; tm = 1; spair = 4;/* surrogate pair size in UTF-16. */ } else { parse = cesu8_to_unicode; tm = ts; spair = 6;/* surrogate pair size in UTF-8. */ } if (archive_string_ensure(as, as->length + len * tm + ts) == NULL) return (-1); p = as->s + as->length; endp = as->s + as->buffer_length - ts; while ((n = parse(&uc, s, len)) != 0) { const char *ucptr; uint32_t cp1, cp2; int SIndex; struct { uint32_t uc; int ccc; } fdc[FDC_MAX]; int fdi, fdj; int ccc; check_first_code: if (n < 0) { /* Use a replaced unicode character. */ UNPARSE(p, endp, uc); s += n*-1; len -= n*-1; ret = -1; continue; } else if (n == spair || always_replace) /* uc is converted from a surrogate pair. * this should be treated as a changed code. */ ucptr = NULL; else ucptr = s; s += n; len -= n; /* Hangul Decomposition. */ if ((SIndex = uc - HC_SBASE) >= 0 && SIndex < HC_SCOUNT) { int L = HC_LBASE + SIndex / HC_NCOUNT; int V = HC_VBASE + (SIndex % HC_NCOUNT) / HC_TCOUNT; int T = HC_TBASE + SIndex % HC_TCOUNT; REPLACE_UC_WITH(L); WRITE_UC(); REPLACE_UC_WITH(V); WRITE_UC(); if (T != HC_TBASE) { REPLACE_UC_WITH(T); WRITE_UC(); } continue; } if (IS_DECOMPOSABLE_BLOCK(uc) && CCC(uc) != 0) { WRITE_UC(); continue; } fdi = 0; while (get_nfd(&cp1, &cp2, uc) && fdi < FDC_MAX) { int k; for (k = fdi; k > 0; k--) fdc[k] = fdc[k-1]; fdc[0].ccc = CCC(cp2); fdc[0].uc = cp2; fdi++; REPLACE_UC_WITH(cp1); } /* Read following code points. */ while ((n2 = parse(&uc2, s, len)) > 0 && (ccc = CCC(uc2)) != 0 && fdi < FDC_MAX) { int j, k; s += n2; len -= n2; for (j = 0; j < fdi; j++) { if (fdc[j].ccc > ccc) break; } if (j < fdi) { for (k = fdi; k > j; k--) fdc[k] = fdc[k-1]; fdc[j].ccc = ccc; fdc[j].uc = uc2; } else { fdc[fdi].ccc = ccc; fdc[fdi].uc = uc2; } fdi++; } WRITE_UC(); for (fdj = 0; fdj < fdi; fdj++) { REPLACE_UC_WITH(fdc[fdj].uc); WRITE_UC(); } if (n2 == 0) break; REPLACE_UC_WITH(uc2); n = n2; goto check_first_code; } as->length = p - as->s; as->s[as->length] = '\0'; if (ts == 2) as->s[as->length+1] = '\0'; return (ret); } /* * libarchive 2.x made incorrect UTF-8 strings in the wrong assumption * that WCS is Unicode. It is true for several platforms but some are false. * And then people who did not use UTF-8 locale on the non Unicode WCS * platform and made a tar file with libarchive(mostly bsdtar) 2.x. Those * now cannot get right filename from libarchive 3.x and later since we * fixed the wrong assumption and it is incompatible to older its versions. * So we provide special option, "compat-2x.x", for resolving it. * That option enable the string conversion of libarchive 2.x. * * Translates the wrong UTF-8 string made by libarchive 2.x into current * locale character set and appends to the archive_string. * Note: returns -1 if conversion fails. */ static int strncat_from_utf8_libarchive2(struct archive_string *as, const void *_p, size_t len, struct archive_string_conv *sc) { const char *s; int n; char *p; char *end; uint32_t unicode; #if HAVE_WCRTOMB mbstate_t shift_state; memset(&shift_state, 0, sizeof(shift_state)); #else /* Clear the shift state before starting. */ wctomb(NULL, L'\0'); #endif (void)sc; /* UNUSED */ /* * Allocate buffer for MBS. * We need this allocation here since it is possible that * as->s is still NULL. */ if (archive_string_ensure(as, as->length + len + 1) == NULL) return (-1); s = (const char *)_p; p = as->s + as->length; end = as->s + as->buffer_length - MB_CUR_MAX -1; while ((n = _utf8_to_unicode(&unicode, s, len)) != 0) { wchar_t wc; if (p >= end) { as->length = p - as->s; /* Re-allocate buffer for MBS. */ if (archive_string_ensure(as, as->length + len * 2 + 1) == NULL) return (-1); p = as->s + as->length; end = as->s + as->buffer_length - MB_CUR_MAX -1; } /* - * As libarchie 2.x, translates the UTF-8 characters into + * As libarchive 2.x, translates the UTF-8 characters into * wide-characters in the assumption that WCS is Unicode. */ if (n < 0) { n *= -1; wc = L'?'; } else wc = (wchar_t)unicode; s += n; len -= n; /* * Translates the wide-character into the current locale MBS. */ #if HAVE_WCRTOMB n = (int)wcrtomb(p, wc, &shift_state); #else n = (int)wctomb(p, wc); #endif if (n == -1) return (-1); p += n; } as->length = p - as->s; as->s[as->length] = '\0'; return (0); } /* * Conversion functions between current locale dependent MBS and UTF-16BE. * strncat_from_utf16be() : UTF-16BE --> MBS * strncat_to_utf16be() : MBS --> UTF16BE */ #if defined(_WIN32) && !defined(__CYGWIN__) /* * Convert a UTF-16BE/LE string to current locale and copy the result. * Return -1 if conversion fails. */ static int win_strncat_from_utf16(struct archive_string *as, const void *_p, size_t bytes, struct archive_string_conv *sc, int be) { struct archive_string tmp; const char *u16; int ll; BOOL defchar; char *mbs; size_t mbs_size, b; int ret = 0; bytes &= ~1; if (archive_string_ensure(as, as->length + bytes +1) == NULL) return (-1); mbs = as->s + as->length; mbs_size = as->buffer_length - as->length -1; if (sc->to_cp == CP_C_LOCALE) { /* * "C" locale special process. */ u16 = _p; ll = 0; for (b = 0; b < bytes; b += 2) { uint16_t val; if (be) val = archive_be16dec(u16+b); else val = archive_le16dec(u16+b); if (val > 255) { *mbs++ = '?'; ret = -1; } else *mbs++ = (char)(val&0xff); ll++; } as->length += ll; as->s[as->length] = '\0'; return (ret); } archive_string_init(&tmp); if (be) { if (is_big_endian()) { u16 = _p; } else { if (archive_string_ensure(&tmp, bytes+2) == NULL) return (-1); memcpy(tmp.s, _p, bytes); for (b = 0; b < bytes; b += 2) { uint16_t val = archive_be16dec(tmp.s+b); archive_le16enc(tmp.s+b, val); } u16 = tmp.s; } } else { if (!is_big_endian()) { u16 = _p; } else { if (archive_string_ensure(&tmp, bytes+2) == NULL) return (-1); memcpy(tmp.s, _p, bytes); for (b = 0; b < bytes; b += 2) { uint16_t val = archive_le16dec(tmp.s+b); archive_be16enc(tmp.s+b, val); } u16 = tmp.s; } } do { defchar = 0; ll = WideCharToMultiByte(sc->to_cp, 0, (LPCWSTR)u16, (int)bytes>>1, mbs, (int)mbs_size, NULL, &defchar); /* Exit loop if we succeeded */ if (ll != 0 || GetLastError() != ERROR_INSUFFICIENT_BUFFER) { break; } /* Else expand buffer and loop to try again. */ ll = WideCharToMultiByte(sc->to_cp, 0, (LPCWSTR)u16, (int)bytes, NULL, 0, NULL, NULL); if (archive_string_ensure(as, ll +1) == NULL) return (-1); mbs = as->s + as->length; mbs_size = as->buffer_length - as->length -1; } while (1); archive_string_free(&tmp); as->length += ll; as->s[as->length] = '\0'; if (ll == 0 || defchar) ret = -1; return (ret); } static int win_strncat_from_utf16be(struct archive_string *as, const void *_p, size_t bytes, struct archive_string_conv *sc) { return (win_strncat_from_utf16(as, _p, bytes, sc, 1)); } static int win_strncat_from_utf16le(struct archive_string *as, const void *_p, size_t bytes, struct archive_string_conv *sc) { return (win_strncat_from_utf16(as, _p, bytes, sc, 0)); } static int is_big_endian(void) { uint16_t d = 1; return (archive_be16dec(&d) == 1); } /* * Convert a current locale string to UTF-16BE/LE and copy the result. * Return -1 if conversion fails. */ static int win_strncat_to_utf16(struct archive_string *as16, const void *_p, size_t length, struct archive_string_conv *sc, int bigendian) { const char *s = (const char *)_p; char *u16; size_t count, avail; if (archive_string_ensure(as16, as16->length + (length + 1) * 2) == NULL) return (-1); u16 = as16->s + as16->length; avail = as16->buffer_length - 2; if (sc->from_cp == CP_C_LOCALE) { /* * "C" locale special process. */ count = 0; while (count < length && *s) { if (bigendian) archive_be16enc(u16, *s); else archive_le16enc(u16, *s); u16 += 2; s++; count++; } as16->length += count << 1; as16->s[as16->length] = 0; as16->s[as16->length+1] = 0; return (0); } do { count = MultiByteToWideChar(sc->from_cp, MB_PRECOMPOSED, s, (int)length, (LPWSTR)u16, (int)avail>>1); /* Exit loop if we succeeded */ if (count != 0 || GetLastError() != ERROR_INSUFFICIENT_BUFFER) { break; } /* Expand buffer and try again */ count = MultiByteToWideChar(sc->from_cp, MB_PRECOMPOSED, s, (int)length, NULL, 0); if (archive_string_ensure(as16, (count +1) * 2) == NULL) return (-1); u16 = as16->s + as16->length; avail = as16->buffer_length - 2; } while (1); as16->length += count * 2; as16->s[as16->length] = 0; as16->s[as16->length+1] = 0; if (count == 0) return (-1); if (is_big_endian()) { if (!bigendian) { while (count > 0) { uint16_t v = archive_be16dec(u16); archive_le16enc(u16, v); u16 += 2; count--; } } } else { if (bigendian) { while (count > 0) { uint16_t v = archive_le16dec(u16); archive_be16enc(u16, v); u16 += 2; count--; } } } return (0); } static int win_strncat_to_utf16be(struct archive_string *as16, const void *_p, size_t length, struct archive_string_conv *sc) { return (win_strncat_to_utf16(as16, _p, length, sc, 1)); } static int win_strncat_to_utf16le(struct archive_string *as16, const void *_p, size_t length, struct archive_string_conv *sc) { return (win_strncat_to_utf16(as16, _p, length, sc, 0)); } #endif /* _WIN32 && !__CYGWIN__ */ /* * Do the best effort for conversions. * We cannot handle UTF-16BE character-set without such iconv, * but there is a chance if a string consists just ASCII code or * a current locale is UTF-8. */ /* * Convert a UTF-16BE string to current locale and copy the result. * Return -1 if conversion fails. */ static int best_effort_strncat_from_utf16(struct archive_string *as, const void *_p, size_t bytes, struct archive_string_conv *sc, int be) { const char *utf16 = (const char *)_p; char *mbs; uint32_t uc; int n, ret; (void)sc; /* UNUSED */ /* * Other case, we should do the best effort. * If all character are ASCII(<0x7f), we can convert it. * if not , we set a alternative character and return -1. */ ret = 0; if (archive_string_ensure(as, as->length + bytes +1) == NULL) return (-1); mbs = as->s + as->length; while ((n = utf16_to_unicode(&uc, utf16, bytes, be)) != 0) { if (n < 0) { n *= -1; ret = -1; } bytes -= n; utf16 += n; if (uc > 127) { /* We cannot handle it. */ *mbs++ = '?'; ret = -1; } else *mbs++ = (char)uc; } as->length = mbs - as->s; as->s[as->length] = '\0'; return (ret); } static int best_effort_strncat_from_utf16be(struct archive_string *as, const void *_p, size_t bytes, struct archive_string_conv *sc) { return (best_effort_strncat_from_utf16(as, _p, bytes, sc, 1)); } static int best_effort_strncat_from_utf16le(struct archive_string *as, const void *_p, size_t bytes, struct archive_string_conv *sc) { return (best_effort_strncat_from_utf16(as, _p, bytes, sc, 0)); } /* * Convert a current locale string to UTF-16BE/LE and copy the result. * Return -1 if conversion fails. */ static int best_effort_strncat_to_utf16(struct archive_string *as16, const void *_p, size_t length, struct archive_string_conv *sc, int bigendian) { const char *s = (const char *)_p; char *utf16; size_t remaining; int ret; (void)sc; /* UNUSED */ /* * Other case, we should do the best effort. * If all character are ASCII(<0x7f), we can convert it. * if not , we set a alternative character and return -1. */ ret = 0; remaining = length; if (archive_string_ensure(as16, as16->length + (length + 1) * 2) == NULL) return (-1); utf16 = as16->s + as16->length; while (remaining--) { unsigned c = *s++; if (c > 127) { /* We cannot handle it. */ c = UNICODE_R_CHAR; ret = -1; } if (bigendian) archive_be16enc(utf16, c); else archive_le16enc(utf16, c); utf16 += 2; } as16->length = utf16 - as16->s; as16->s[as16->length] = 0; as16->s[as16->length+1] = 0; return (ret); } static int best_effort_strncat_to_utf16be(struct archive_string *as16, const void *_p, size_t length, struct archive_string_conv *sc) { return (best_effort_strncat_to_utf16(as16, _p, length, sc, 1)); } static int best_effort_strncat_to_utf16le(struct archive_string *as16, const void *_p, size_t length, struct archive_string_conv *sc) { return (best_effort_strncat_to_utf16(as16, _p, length, sc, 0)); } /* * Multistring operations. */ void archive_mstring_clean(struct archive_mstring *aes) { archive_wstring_free(&(aes->aes_wcs)); archive_string_free(&(aes->aes_mbs)); archive_string_free(&(aes->aes_utf8)); archive_string_free(&(aes->aes_mbs_in_locale)); aes->aes_set = 0; } void archive_mstring_copy(struct archive_mstring *dest, struct archive_mstring *src) { dest->aes_set = src->aes_set; archive_string_copy(&(dest->aes_mbs), &(src->aes_mbs)); archive_string_copy(&(dest->aes_utf8), &(src->aes_utf8)); archive_wstring_copy(&(dest->aes_wcs), &(src->aes_wcs)); } int archive_mstring_get_utf8(struct archive *a, struct archive_mstring *aes, const char **p) { struct archive_string_conv *sc; int r; /* If we already have a UTF8 form, return that immediately. */ if (aes->aes_set & AES_SET_UTF8) { *p = aes->aes_utf8.s; return (0); } *p = NULL; if (aes->aes_set & AES_SET_MBS) { sc = archive_string_conversion_to_charset(a, "UTF-8", 1); if (sc == NULL) return (-1);/* Couldn't allocate memory for sc. */ r = archive_strncpy_l(&(aes->aes_utf8), aes->aes_mbs.s, aes->aes_mbs.length, sc); if (a == NULL) free_sconv_object(sc); if (r == 0) { aes->aes_set |= AES_SET_UTF8; *p = aes->aes_utf8.s; return (0);/* success. */ } else return (-1);/* failure. */ } return (0);/* success. */ } int archive_mstring_get_mbs(struct archive *a, struct archive_mstring *aes, const char **p) { int r, ret = 0; (void)a; /* UNUSED */ /* If we already have an MBS form, return that immediately. */ if (aes->aes_set & AES_SET_MBS) { *p = aes->aes_mbs.s; return (ret); } *p = NULL; /* If there's a WCS form, try converting with the native locale. */ if (aes->aes_set & AES_SET_WCS) { archive_string_empty(&(aes->aes_mbs)); r = archive_string_append_from_wcs(&(aes->aes_mbs), aes->aes_wcs.s, aes->aes_wcs.length); *p = aes->aes_mbs.s; if (r == 0) { aes->aes_set |= AES_SET_MBS; return (ret); } else ret = -1; } /* * Only a UTF-8 form cannot avail because its conversion already * failed at archive_mstring_update_utf8(). */ return (ret); } int archive_mstring_get_wcs(struct archive *a, struct archive_mstring *aes, const wchar_t **wp) { int r, ret = 0; (void)a;/* UNUSED */ /* Return WCS form if we already have it. */ if (aes->aes_set & AES_SET_WCS) { *wp = aes->aes_wcs.s; return (ret); } *wp = NULL; /* Try converting MBS to WCS using native locale. */ if (aes->aes_set & AES_SET_MBS) { archive_wstring_empty(&(aes->aes_wcs)); r = archive_wstring_append_from_mbs(&(aes->aes_wcs), aes->aes_mbs.s, aes->aes_mbs.length); if (r == 0) { aes->aes_set |= AES_SET_WCS; *wp = aes->aes_wcs.s; } else ret = -1;/* failure. */ } return (ret); } int archive_mstring_get_mbs_l(struct archive_mstring *aes, const char **p, size_t *length, struct archive_string_conv *sc) { int r, ret = 0; #if defined(_WIN32) && !defined(__CYGWIN__) /* - * Internationalization programing on Windows must use Wide + * Internationalization programming on Windows must use Wide * characters because Windows platform cannot make locale UTF-8. */ if (sc != NULL && (aes->aes_set & AES_SET_WCS) != 0) { archive_string_empty(&(aes->aes_mbs_in_locale)); r = archive_string_append_from_wcs_in_codepage( &(aes->aes_mbs_in_locale), aes->aes_wcs.s, aes->aes_wcs.length, sc); if (r == 0) { *p = aes->aes_mbs_in_locale.s; if (length != NULL) *length = aes->aes_mbs_in_locale.length; return (0); } else if (errno == ENOMEM) return (-1); else ret = -1; } #endif /* If there is not an MBS form but is a WCS form, try converting * with the native locale to be used for translating it to specified * character-set. */ if ((aes->aes_set & AES_SET_MBS) == 0 && (aes->aes_set & AES_SET_WCS) != 0) { archive_string_empty(&(aes->aes_mbs)); r = archive_string_append_from_wcs(&(aes->aes_mbs), aes->aes_wcs.s, aes->aes_wcs.length); if (r == 0) aes->aes_set |= AES_SET_MBS; else if (errno == ENOMEM) return (-1); else ret = -1; } /* If we already have an MBS form, use it to be translated to * specified character-set. */ if (aes->aes_set & AES_SET_MBS) { if (sc == NULL) { /* Conversion is unneeded. */ *p = aes->aes_mbs.s; if (length != NULL) *length = aes->aes_mbs.length; return (0); } ret = archive_strncpy_l(&(aes->aes_mbs_in_locale), aes->aes_mbs.s, aes->aes_mbs.length, sc); *p = aes->aes_mbs_in_locale.s; if (length != NULL) *length = aes->aes_mbs_in_locale.length; } else { *p = NULL; if (length != NULL) *length = 0; } return (ret); } int archive_mstring_copy_mbs(struct archive_mstring *aes, const char *mbs) { if (mbs == NULL) { aes->aes_set = 0; return (0); } return (archive_mstring_copy_mbs_len(aes, mbs, strlen(mbs))); } int archive_mstring_copy_mbs_len(struct archive_mstring *aes, const char *mbs, size_t len) { if (mbs == NULL) { aes->aes_set = 0; return (0); } aes->aes_set = AES_SET_MBS; /* Only MBS form is set now. */ archive_strncpy(&(aes->aes_mbs), mbs, len); archive_string_empty(&(aes->aes_utf8)); archive_wstring_empty(&(aes->aes_wcs)); return (0); } int archive_mstring_copy_wcs(struct archive_mstring *aes, const wchar_t *wcs) { return archive_mstring_copy_wcs_len(aes, wcs, wcs == NULL ? 0 : wcslen(wcs)); } int archive_mstring_copy_utf8(struct archive_mstring *aes, const char *utf8) { if (utf8 == NULL) { aes->aes_set = 0; } aes->aes_set = AES_SET_UTF8; archive_string_empty(&(aes->aes_mbs)); archive_string_empty(&(aes->aes_wcs)); archive_strncpy(&(aes->aes_utf8), utf8, strlen(utf8)); return (int)strlen(utf8); } int archive_mstring_copy_wcs_len(struct archive_mstring *aes, const wchar_t *wcs, size_t len) { if (wcs == NULL) { aes->aes_set = 0; } aes->aes_set = AES_SET_WCS; /* Only WCS form set. */ archive_string_empty(&(aes->aes_mbs)); archive_string_empty(&(aes->aes_utf8)); archive_wstrncpy(&(aes->aes_wcs), wcs, len); return (0); } int archive_mstring_copy_mbs_len_l(struct archive_mstring *aes, const char *mbs, size_t len, struct archive_string_conv *sc) { int r; if (mbs == NULL) { aes->aes_set = 0; return (0); } archive_string_empty(&(aes->aes_mbs)); archive_wstring_empty(&(aes->aes_wcs)); archive_string_empty(&(aes->aes_utf8)); #if defined(_WIN32) && !defined(__CYGWIN__) /* - * Internationalization programing on Windows must use Wide + * Internationalization programming on Windows must use Wide * characters because Windows platform cannot make locale UTF-8. */ if (sc == NULL) { if (archive_string_append(&(aes->aes_mbs), mbs, mbsnbytes(mbs, len)) == NULL) { aes->aes_set = 0; r = -1; } else { aes->aes_set = AES_SET_MBS; r = 0; } #if defined(HAVE_ICONV) } else if (sc != NULL && sc->cd_w != (iconv_t)-1) { /* * This case happens only when MultiByteToWideChar() cannot * handle sc->from_cp, and we have to iconv in order to * translate character-set to wchar_t,UTF-16. */ iconv_t cd = sc->cd; unsigned from_cp; int flag; /* * Translate multi-bytes from some character-set to UTF-8. */ sc->cd = sc->cd_w; r = archive_strncpy_l(&(aes->aes_utf8), mbs, len, sc); sc->cd = cd; if (r != 0) { aes->aes_set = 0; return (r); } aes->aes_set = AES_SET_UTF8; /* * Append the UTF-8 string into wstring. */ flag = sc->flag; sc->flag &= ~(SCONV_NORMALIZATION_C | SCONV_TO_UTF16| SCONV_FROM_UTF16); from_cp = sc->from_cp; sc->from_cp = CP_UTF8; r = archive_wstring_append_from_mbs_in_codepage(&(aes->aes_wcs), aes->aes_utf8.s, aes->aes_utf8.length, sc); sc->flag = flag; sc->from_cp = from_cp; if (r == 0) aes->aes_set |= AES_SET_WCS; #endif } else { r = archive_wstring_append_from_mbs_in_codepage( &(aes->aes_wcs), mbs, len, sc); if (r == 0) aes->aes_set = AES_SET_WCS; else aes->aes_set = 0; } #else r = archive_strncpy_l(&(aes->aes_mbs), mbs, len, sc); if (r == 0) aes->aes_set = AES_SET_MBS; /* Only MBS form is set now. */ else aes->aes_set = 0; #endif return (r); } /* * The 'update' form tries to proactively update all forms of * this string (WCS and MBS) and returns an error if any of * them fail. This is used by the 'pax' handler, for instance, * to detect and report character-conversion failures early while * still allowing clients to get potentially useful values from * the more tolerant lazy conversions. (get_mbs and get_wcs will * strive to give the user something useful, so you can get hopefully * usable values even if some of the character conversions are failing.) */ int archive_mstring_update_utf8(struct archive *a, struct archive_mstring *aes, const char *utf8) { struct archive_string_conv *sc; int r; if (utf8 == NULL) { aes->aes_set = 0; return (0); /* Succeeded in clearing everything. */ } /* Save the UTF8 string. */ archive_strcpy(&(aes->aes_utf8), utf8); /* Empty the mbs and wcs strings. */ archive_string_empty(&(aes->aes_mbs)); archive_wstring_empty(&(aes->aes_wcs)); aes->aes_set = AES_SET_UTF8; /* Only UTF8 is set now. */ /* Try converting UTF-8 to MBS, return false on failure. */ sc = archive_string_conversion_from_charset(a, "UTF-8", 1); if (sc == NULL) return (-1);/* Couldn't allocate memory for sc. */ r = archive_strcpy_l(&(aes->aes_mbs), utf8, sc); if (a == NULL) free_sconv_object(sc); if (r != 0) return (-1); aes->aes_set = AES_SET_UTF8 | AES_SET_MBS; /* Both UTF8 and MBS set. */ /* Try converting MBS to WCS, return false on failure. */ if (archive_wstring_append_from_mbs(&(aes->aes_wcs), aes->aes_mbs.s, aes->aes_mbs.length)) return (-1); aes->aes_set = AES_SET_UTF8 | AES_SET_WCS | AES_SET_MBS; /* All conversions succeeded. */ return (0); } Index: head/contrib/libarchive/libarchive/archive_string.h =================================================================== --- head/contrib/libarchive/libarchive/archive_string.h (revision 311902) +++ head/contrib/libarchive/libarchive/archive_string.h (revision 311903) @@ -1,239 +1,243 @@ /*- * Copyright (c) 2003-2010 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. * * $FreeBSD$ * */ #ifndef __LIBARCHIVE_BUILD #ifndef __LIBARCHIVE_TEST #error This header is only to be used internally to libarchive. #endif #endif #ifndef ARCHIVE_STRING_H_INCLUDED #define ARCHIVE_STRING_H_INCLUDED #include #ifdef HAVE_STDLIB_H #include /* required for wchar_t on some systems */ #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_WCHAR_H #include #endif #include "archive.h" /* * Basic resizable/reusable string support similar to Java's "StringBuffer." * * Unlike sbuf(9), the buffers here are fully reusable and track the * length throughout. */ struct archive_string { char *s; /* Pointer to the storage */ size_t length; /* Length of 's' in characters */ size_t buffer_length; /* Length of malloc-ed storage in bytes. */ }; struct archive_wstring { wchar_t *s; /* Pointer to the storage */ size_t length; /* Length of 's' in characters */ size_t buffer_length; /* Length of malloc-ed storage in bytes. */ }; struct archive_string_conv; /* Initialize an archive_string object on the stack or elsewhere. */ #define archive_string_init(a) \ do { (a)->s = NULL; (a)->length = 0; (a)->buffer_length = 0; } while(0) /* Append a C char to an archive_string, resizing as necessary. */ struct archive_string * archive_strappend_char(struct archive_string *, char); /* Ditto for a wchar_t and an archive_wstring. */ struct archive_wstring * archive_wstrappend_wchar(struct archive_wstring *, wchar_t); +/* Append a raw array to an archive_string, resizing as necessary */ +struct archive_string * +archive_array_append(struct archive_string *, const char *, size_t); + /* Convert a Unicode string to current locale and append the result. */ /* Returns -1 if conversion fails. */ int archive_string_append_from_wcs(struct archive_string *, const wchar_t *, size_t); /* Create a string conversion object. * Return NULL and set a error message if the conversion is not supported * on the platform. */ struct archive_string_conv * archive_string_conversion_to_charset(struct archive *, const char *, int); struct archive_string_conv * archive_string_conversion_from_charset(struct archive *, const char *, int); /* Create the default string conversion object for reading/writing an archive. * Return NULL if the conversion is unneeded. * Note: On non Windows platform this always returns NULL. */ struct archive_string_conv * archive_string_default_conversion_for_read(struct archive *); struct archive_string_conv * archive_string_default_conversion_for_write(struct archive *); /* Dispose of a string conversion object. */ void archive_string_conversion_free(struct archive *); const char * archive_string_conversion_charset_name(struct archive_string_conv *); void archive_string_conversion_set_opt(struct archive_string_conv *, int); #define SCONV_SET_OPT_UTF8_LIBARCHIVE2X 1 #define SCONV_SET_OPT_NORMALIZATION_C 2 #define SCONV_SET_OPT_NORMALIZATION_D 4 /* Copy one archive_string to another in locale conversion. * Return -1 if conversion fails. */ int archive_strncpy_l(struct archive_string *, const void *, size_t, struct archive_string_conv *); /* Copy one archive_string to another in locale conversion. * Return -1 if conversion fails. */ int archive_strncat_l(struct archive_string *, const void *, size_t, struct archive_string_conv *); /* Copy one archive_string to another */ #define archive_string_copy(dest, src) \ ((dest)->length = 0, archive_string_concat((dest), (src))) #define archive_wstring_copy(dest, src) \ ((dest)->length = 0, archive_wstring_concat((dest), (src))) /* Concatenate one archive_string to another */ void archive_string_concat(struct archive_string *dest, struct archive_string *src); void archive_wstring_concat(struct archive_wstring *dest, struct archive_wstring *src); /* Ensure that the underlying buffer is at least as large as the request. */ struct archive_string * archive_string_ensure(struct archive_string *, size_t); struct archive_wstring * archive_wstring_ensure(struct archive_wstring *, size_t); /* Append C string, which may lack trailing \0. */ /* The source is declared void * here because this gets used with * "signed char *", "unsigned char *" and "char *" arguments. * Declaring it "char *" as with some of the other functions just * leads to a lot of extra casts. */ struct archive_string * archive_strncat(struct archive_string *, const void *, size_t); struct archive_wstring * archive_wstrncat(struct archive_wstring *, const wchar_t *, size_t); /* Append a C string to an archive_string, resizing as necessary. */ struct archive_string * archive_strcat(struct archive_string *, const void *); struct archive_wstring * archive_wstrcat(struct archive_wstring *, const wchar_t *); /* Copy a C string to an archive_string, resizing as necessary. */ #define archive_strcpy(as,p) \ archive_strncpy((as), (p), ((p) == NULL ? 0 : strlen(p))) #define archive_wstrcpy(as,p) \ archive_wstrncpy((as), (p), ((p) == NULL ? 0 : wcslen(p))) #define archive_strcpy_l(as,p,lo) \ archive_strncpy_l((as), (p), ((p) == NULL ? 0 : strlen(p)), (lo)) /* Copy a C string to an archive_string with limit, resizing as necessary. */ #define archive_strncpy(as,p,l) \ ((as)->length=0, archive_strncat((as), (p), (l))) #define archive_wstrncpy(as,p,l) \ ((as)->length = 0, archive_wstrncat((as), (p), (l))) /* Return length of string. */ #define archive_strlen(a) ((a)->length) /* Set string length to zero. */ #define archive_string_empty(a) ((a)->length = 0) #define archive_wstring_empty(a) ((a)->length = 0) /* Release any allocated storage resources. */ void archive_string_free(struct archive_string *); void archive_wstring_free(struct archive_wstring *); /* Like 'vsprintf', but resizes the underlying string as necessary. */ /* Note: This only implements a small subset of standard printf functionality. */ void archive_string_vsprintf(struct archive_string *, const char *, va_list) __LA_PRINTF(2, 0); void archive_string_sprintf(struct archive_string *, const char *, ...) __LA_PRINTF(2, 3); /* Translates from MBS to Unicode. */ /* Returns non-zero if conversion failed in any way. */ int archive_wstring_append_from_mbs(struct archive_wstring *dest, const char *, size_t); /* A "multistring" can hold Unicode, UTF8, or MBS versions of * the string. If you set and read the same version, no translation * is done. If you set and read different versions, the library * will attempt to transparently convert. */ struct archive_mstring { struct archive_string aes_mbs; struct archive_string aes_utf8; struct archive_wstring aes_wcs; struct archive_string aes_mbs_in_locale; /* Bitmap of which of the above are valid. Because we're lazy * about malloc-ing and reusing the underlying storage, we * can't rely on NULL pointers to indicate whether a string * has been set. */ int aes_set; #define AES_SET_MBS 1 #define AES_SET_UTF8 2 #define AES_SET_WCS 4 }; void archive_mstring_clean(struct archive_mstring *); void archive_mstring_copy(struct archive_mstring *dest, struct archive_mstring *src); int archive_mstring_get_mbs(struct archive *, struct archive_mstring *, const char **); int archive_mstring_get_utf8(struct archive *, struct archive_mstring *, const char **); int archive_mstring_get_wcs(struct archive *, struct archive_mstring *, const wchar_t **); int archive_mstring_get_mbs_l(struct archive_mstring *, const char **, size_t *, struct archive_string_conv *); int archive_mstring_copy_mbs(struct archive_mstring *, const char *mbs); int archive_mstring_copy_mbs_len(struct archive_mstring *, const char *mbs, size_t); int archive_mstring_copy_utf8(struct archive_mstring *, const char *utf8); int archive_mstring_copy_wcs(struct archive_mstring *, const wchar_t *wcs); int archive_mstring_copy_wcs_len(struct archive_mstring *, const wchar_t *wcs, size_t); int archive_mstring_copy_mbs_len_l(struct archive_mstring *, const char *mbs, size_t, struct archive_string_conv *); int archive_mstring_update_utf8(struct archive *, struct archive_mstring *aes, const char *utf8); #endif Index: head/contrib/libarchive/libarchive/archive_string_composition.h =================================================================== --- head/contrib/libarchive/libarchive/archive_string_composition.h (revision 311902) +++ head/contrib/libarchive/libarchive/archive_string_composition.h (revision 311903) @@ -1,2292 +1,2292 @@ /*- * Copyright (c) 2011-2012 libarchive Project * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) 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 AUTHOR(S) 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. * * $FreeBSD$ * */ /* * ATTENTION! * This file is generated by build/utils/gen_archive_string_composition_h.sh * from http://unicode.org/Public/6.0.0/ucd/UnicodeData.txt * * See also http://unicode.org/report/tr15/ */ #ifndef __LIBARCHIVE_BUILD #error This header is only to be used internally to libarchive. #endif #ifndef ARCHIVE_STRING_COMPOSITION_H_INCLUDED #define ARCHIVE_STRING_COMPOSITION_H_INCLUDED struct unicode_composition_table { uint32_t cp1; uint32_t cp2; uint32_t nfc; }; static const struct unicode_composition_table u_composition_table[] = { { 0x0003C , 0x00338 , 0x0226E }, { 0x0003D , 0x00338 , 0x02260 }, { 0x0003E , 0x00338 , 0x0226F }, { 0x00041 , 0x00300 , 0x000C0 }, { 0x00041 , 0x00301 , 0x000C1 }, { 0x00041 , 0x00302 , 0x000C2 }, { 0x00041 , 0x00303 , 0x000C3 }, { 0x00041 , 0x00304 , 0x00100 }, { 0x00041 , 0x00306 , 0x00102 }, { 0x00041 , 0x00307 , 0x00226 }, { 0x00041 , 0x00308 , 0x000C4 }, { 0x00041 , 0x00309 , 0x01EA2 }, { 0x00041 , 0x0030A , 0x000C5 }, { 0x00041 , 0x0030C , 0x001CD }, { 0x00041 , 0x0030F , 0x00200 }, { 0x00041 , 0x00311 , 0x00202 }, { 0x00041 , 0x00323 , 0x01EA0 }, { 0x00041 , 0x00325 , 0x01E00 }, { 0x00041 , 0x00328 , 0x00104 }, { 0x00042 , 0x00307 , 0x01E02 }, { 0x00042 , 0x00323 , 0x01E04 }, { 0x00042 , 0x00331 , 0x01E06 }, { 0x00043 , 0x00301 , 0x00106 }, { 0x00043 , 0x00302 , 0x00108 }, { 0x00043 , 0x00307 , 0x0010A }, { 0x00043 , 0x0030C , 0x0010C }, { 0x00043 , 0x00327 , 0x000C7 }, { 0x00044 , 0x00307 , 0x01E0A }, { 0x00044 , 0x0030C , 0x0010E }, { 0x00044 , 0x00323 , 0x01E0C }, { 0x00044 , 0x00327 , 0x01E10 }, { 0x00044 , 0x0032D , 0x01E12 }, { 0x00044 , 0x00331 , 0x01E0E }, { 0x00045 , 0x00300 , 0x000C8 }, { 0x00045 , 0x00301 , 0x000C9 }, { 0x00045 , 0x00302 , 0x000CA }, { 0x00045 , 0x00303 , 0x01EBC }, { 0x00045 , 0x00304 , 0x00112 }, { 0x00045 , 0x00306 , 0x00114 }, { 0x00045 , 0x00307 , 0x00116 }, { 0x00045 , 0x00308 , 0x000CB }, { 0x00045 , 0x00309 , 0x01EBA }, { 0x00045 , 0x0030C , 0x0011A }, { 0x00045 , 0x0030F , 0x00204 }, { 0x00045 , 0x00311 , 0x00206 }, { 0x00045 , 0x00323 , 0x01EB8 }, { 0x00045 , 0x00327 , 0x00228 }, { 0x00045 , 0x00328 , 0x00118 }, { 0x00045 , 0x0032D , 0x01E18 }, { 0x00045 , 0x00330 , 0x01E1A }, { 0x00046 , 0x00307 , 0x01E1E }, { 0x00047 , 0x00301 , 0x001F4 }, { 0x00047 , 0x00302 , 0x0011C }, { 0x00047 , 0x00304 , 0x01E20 }, { 0x00047 , 0x00306 , 0x0011E }, { 0x00047 , 0x00307 , 0x00120 }, { 0x00047 , 0x0030C , 0x001E6 }, { 0x00047 , 0x00327 , 0x00122 }, { 0x00048 , 0x00302 , 0x00124 }, { 0x00048 , 0x00307 , 0x01E22 }, { 0x00048 , 0x00308 , 0x01E26 }, { 0x00048 , 0x0030C , 0x0021E }, { 0x00048 , 0x00323 , 0x01E24 }, { 0x00048 , 0x00327 , 0x01E28 }, { 0x00048 , 0x0032E , 0x01E2A }, { 0x00049 , 0x00300 , 0x000CC }, { 0x00049 , 0x00301 , 0x000CD }, { 0x00049 , 0x00302 , 0x000CE }, { 0x00049 , 0x00303 , 0x00128 }, { 0x00049 , 0x00304 , 0x0012A }, { 0x00049 , 0x00306 , 0x0012C }, { 0x00049 , 0x00307 , 0x00130 }, { 0x00049 , 0x00308 , 0x000CF }, { 0x00049 , 0x00309 , 0x01EC8 }, { 0x00049 , 0x0030C , 0x001CF }, { 0x00049 , 0x0030F , 0x00208 }, { 0x00049 , 0x00311 , 0x0020A }, { 0x00049 , 0x00323 , 0x01ECA }, { 0x00049 , 0x00328 , 0x0012E }, { 0x00049 , 0x00330 , 0x01E2C }, { 0x0004A , 0x00302 , 0x00134 }, { 0x0004B , 0x00301 , 0x01E30 }, { 0x0004B , 0x0030C , 0x001E8 }, { 0x0004B , 0x00323 , 0x01E32 }, { 0x0004B , 0x00327 , 0x00136 }, { 0x0004B , 0x00331 , 0x01E34 }, { 0x0004C , 0x00301 , 0x00139 }, { 0x0004C , 0x0030C , 0x0013D }, { 0x0004C , 0x00323 , 0x01E36 }, { 0x0004C , 0x00327 , 0x0013B }, { 0x0004C , 0x0032D , 0x01E3C }, { 0x0004C , 0x00331 , 0x01E3A }, { 0x0004D , 0x00301 , 0x01E3E }, { 0x0004D , 0x00307 , 0x01E40 }, { 0x0004D , 0x00323 , 0x01E42 }, { 0x0004E , 0x00300 , 0x001F8 }, { 0x0004E , 0x00301 , 0x00143 }, { 0x0004E , 0x00303 , 0x000D1 }, { 0x0004E , 0x00307 , 0x01E44 }, { 0x0004E , 0x0030C , 0x00147 }, { 0x0004E , 0x00323 , 0x01E46 }, { 0x0004E , 0x00327 , 0x00145 }, { 0x0004E , 0x0032D , 0x01E4A }, { 0x0004E , 0x00331 , 0x01E48 }, { 0x0004F , 0x00300 , 0x000D2 }, { 0x0004F , 0x00301 , 0x000D3 }, { 0x0004F , 0x00302 , 0x000D4 }, { 0x0004F , 0x00303 , 0x000D5 }, { 0x0004F , 0x00304 , 0x0014C }, { 0x0004F , 0x00306 , 0x0014E }, { 0x0004F , 0x00307 , 0x0022E }, { 0x0004F , 0x00308 , 0x000D6 }, { 0x0004F , 0x00309 , 0x01ECE }, { 0x0004F , 0x0030B , 0x00150 }, { 0x0004F , 0x0030C , 0x001D1 }, { 0x0004F , 0x0030F , 0x0020C }, { 0x0004F , 0x00311 , 0x0020E }, { 0x0004F , 0x0031B , 0x001A0 }, { 0x0004F , 0x00323 , 0x01ECC }, { 0x0004F , 0x00328 , 0x001EA }, { 0x00050 , 0x00301 , 0x01E54 }, { 0x00050 , 0x00307 , 0x01E56 }, { 0x00052 , 0x00301 , 0x00154 }, { 0x00052 , 0x00307 , 0x01E58 }, { 0x00052 , 0x0030C , 0x00158 }, { 0x00052 , 0x0030F , 0x00210 }, { 0x00052 , 0x00311 , 0x00212 }, { 0x00052 , 0x00323 , 0x01E5A }, { 0x00052 , 0x00327 , 0x00156 }, { 0x00052 , 0x00331 , 0x01E5E }, { 0x00053 , 0x00301 , 0x0015A }, { 0x00053 , 0x00302 , 0x0015C }, { 0x00053 , 0x00307 , 0x01E60 }, { 0x00053 , 0x0030C , 0x00160 }, { 0x00053 , 0x00323 , 0x01E62 }, { 0x00053 , 0x00326 , 0x00218 }, { 0x00053 , 0x00327 , 0x0015E }, { 0x00054 , 0x00307 , 0x01E6A }, { 0x00054 , 0x0030C , 0x00164 }, { 0x00054 , 0x00323 , 0x01E6C }, { 0x00054 , 0x00326 , 0x0021A }, { 0x00054 , 0x00327 , 0x00162 }, { 0x00054 , 0x0032D , 0x01E70 }, { 0x00054 , 0x00331 , 0x01E6E }, { 0x00055 , 0x00300 , 0x000D9 }, { 0x00055 , 0x00301 , 0x000DA }, { 0x00055 , 0x00302 , 0x000DB }, { 0x00055 , 0x00303 , 0x00168 }, { 0x00055 , 0x00304 , 0x0016A }, { 0x00055 , 0x00306 , 0x0016C }, { 0x00055 , 0x00308 , 0x000DC }, { 0x00055 , 0x00309 , 0x01EE6 }, { 0x00055 , 0x0030A , 0x0016E }, { 0x00055 , 0x0030B , 0x00170 }, { 0x00055 , 0x0030C , 0x001D3 }, { 0x00055 , 0x0030F , 0x00214 }, { 0x00055 , 0x00311 , 0x00216 }, { 0x00055 , 0x0031B , 0x001AF }, { 0x00055 , 0x00323 , 0x01EE4 }, { 0x00055 , 0x00324 , 0x01E72 }, { 0x00055 , 0x00328 , 0x00172 }, { 0x00055 , 0x0032D , 0x01E76 }, { 0x00055 , 0x00330 , 0x01E74 }, { 0x00056 , 0x00303 , 0x01E7C }, { 0x00056 , 0x00323 , 0x01E7E }, { 0x00057 , 0x00300 , 0x01E80 }, { 0x00057 , 0x00301 , 0x01E82 }, { 0x00057 , 0x00302 , 0x00174 }, { 0x00057 , 0x00307 , 0x01E86 }, { 0x00057 , 0x00308 , 0x01E84 }, { 0x00057 , 0x00323 , 0x01E88 }, { 0x00058 , 0x00307 , 0x01E8A }, { 0x00058 , 0x00308 , 0x01E8C }, { 0x00059 , 0x00300 , 0x01EF2 }, { 0x00059 , 0x00301 , 0x000DD }, { 0x00059 , 0x00302 , 0x00176 }, { 0x00059 , 0x00303 , 0x01EF8 }, { 0x00059 , 0x00304 , 0x00232 }, { 0x00059 , 0x00307 , 0x01E8E }, { 0x00059 , 0x00308 , 0x00178 }, { 0x00059 , 0x00309 , 0x01EF6 }, { 0x00059 , 0x00323 , 0x01EF4 }, { 0x0005A , 0x00301 , 0x00179 }, { 0x0005A , 0x00302 , 0x01E90 }, { 0x0005A , 0x00307 , 0x0017B }, { 0x0005A , 0x0030C , 0x0017D }, { 0x0005A , 0x00323 , 0x01E92 }, { 0x0005A , 0x00331 , 0x01E94 }, { 0x00061 , 0x00300 , 0x000E0 }, { 0x00061 , 0x00301 , 0x000E1 }, { 0x00061 , 0x00302 , 0x000E2 }, { 0x00061 , 0x00303 , 0x000E3 }, { 0x00061 , 0x00304 , 0x00101 }, { 0x00061 , 0x00306 , 0x00103 }, { 0x00061 , 0x00307 , 0x00227 }, { 0x00061 , 0x00308 , 0x000E4 }, { 0x00061 , 0x00309 , 0x01EA3 }, { 0x00061 , 0x0030A , 0x000E5 }, { 0x00061 , 0x0030C , 0x001CE }, { 0x00061 , 0x0030F , 0x00201 }, { 0x00061 , 0x00311 , 0x00203 }, { 0x00061 , 0x00323 , 0x01EA1 }, { 0x00061 , 0x00325 , 0x01E01 }, { 0x00061 , 0x00328 , 0x00105 }, { 0x00062 , 0x00307 , 0x01E03 }, { 0x00062 , 0x00323 , 0x01E05 }, { 0x00062 , 0x00331 , 0x01E07 }, { 0x00063 , 0x00301 , 0x00107 }, { 0x00063 , 0x00302 , 0x00109 }, { 0x00063 , 0x00307 , 0x0010B }, { 0x00063 , 0x0030C , 0x0010D }, { 0x00063 , 0x00327 , 0x000E7 }, { 0x00064 , 0x00307 , 0x01E0B }, { 0x00064 , 0x0030C , 0x0010F }, { 0x00064 , 0x00323 , 0x01E0D }, { 0x00064 , 0x00327 , 0x01E11 }, { 0x00064 , 0x0032D , 0x01E13 }, { 0x00064 , 0x00331 , 0x01E0F }, { 0x00065 , 0x00300 , 0x000E8 }, { 0x00065 , 0x00301 , 0x000E9 }, { 0x00065 , 0x00302 , 0x000EA }, { 0x00065 , 0x00303 , 0x01EBD }, { 0x00065 , 0x00304 , 0x00113 }, { 0x00065 , 0x00306 , 0x00115 }, { 0x00065 , 0x00307 , 0x00117 }, { 0x00065 , 0x00308 , 0x000EB }, { 0x00065 , 0x00309 , 0x01EBB }, { 0x00065 , 0x0030C , 0x0011B }, { 0x00065 , 0x0030F , 0x00205 }, { 0x00065 , 0x00311 , 0x00207 }, { 0x00065 , 0x00323 , 0x01EB9 }, { 0x00065 , 0x00327 , 0x00229 }, { 0x00065 , 0x00328 , 0x00119 }, { 0x00065 , 0x0032D , 0x01E19 }, { 0x00065 , 0x00330 , 0x01E1B }, { 0x00066 , 0x00307 , 0x01E1F }, { 0x00067 , 0x00301 , 0x001F5 }, { 0x00067 , 0x00302 , 0x0011D }, { 0x00067 , 0x00304 , 0x01E21 }, { 0x00067 , 0x00306 , 0x0011F }, { 0x00067 , 0x00307 , 0x00121 }, { 0x00067 , 0x0030C , 0x001E7 }, { 0x00067 , 0x00327 , 0x00123 }, { 0x00068 , 0x00302 , 0x00125 }, { 0x00068 , 0x00307 , 0x01E23 }, { 0x00068 , 0x00308 , 0x01E27 }, { 0x00068 , 0x0030C , 0x0021F }, { 0x00068 , 0x00323 , 0x01E25 }, { 0x00068 , 0x00327 , 0x01E29 }, { 0x00068 , 0x0032E , 0x01E2B }, { 0x00068 , 0x00331 , 0x01E96 }, { 0x00069 , 0x00300 , 0x000EC }, { 0x00069 , 0x00301 , 0x000ED }, { 0x00069 , 0x00302 , 0x000EE }, { 0x00069 , 0x00303 , 0x00129 }, { 0x00069 , 0x00304 , 0x0012B }, { 0x00069 , 0x00306 , 0x0012D }, { 0x00069 , 0x00308 , 0x000EF }, { 0x00069 , 0x00309 , 0x01EC9 }, { 0x00069 , 0x0030C , 0x001D0 }, { 0x00069 , 0x0030F , 0x00209 }, { 0x00069 , 0x00311 , 0x0020B }, { 0x00069 , 0x00323 , 0x01ECB }, { 0x00069 , 0x00328 , 0x0012F }, { 0x00069 , 0x00330 , 0x01E2D }, { 0x0006A , 0x00302 , 0x00135 }, { 0x0006A , 0x0030C , 0x001F0 }, { 0x0006B , 0x00301 , 0x01E31 }, { 0x0006B , 0x0030C , 0x001E9 }, { 0x0006B , 0x00323 , 0x01E33 }, { 0x0006B , 0x00327 , 0x00137 }, { 0x0006B , 0x00331 , 0x01E35 }, { 0x0006C , 0x00301 , 0x0013A }, { 0x0006C , 0x0030C , 0x0013E }, { 0x0006C , 0x00323 , 0x01E37 }, { 0x0006C , 0x00327 , 0x0013C }, { 0x0006C , 0x0032D , 0x01E3D }, { 0x0006C , 0x00331 , 0x01E3B }, { 0x0006D , 0x00301 , 0x01E3F }, { 0x0006D , 0x00307 , 0x01E41 }, { 0x0006D , 0x00323 , 0x01E43 }, { 0x0006E , 0x00300 , 0x001F9 }, { 0x0006E , 0x00301 , 0x00144 }, { 0x0006E , 0x00303 , 0x000F1 }, { 0x0006E , 0x00307 , 0x01E45 }, { 0x0006E , 0x0030C , 0x00148 }, { 0x0006E , 0x00323 , 0x01E47 }, { 0x0006E , 0x00327 , 0x00146 }, { 0x0006E , 0x0032D , 0x01E4B }, { 0x0006E , 0x00331 , 0x01E49 }, { 0x0006F , 0x00300 , 0x000F2 }, { 0x0006F , 0x00301 , 0x000F3 }, { 0x0006F , 0x00302 , 0x000F4 }, { 0x0006F , 0x00303 , 0x000F5 }, { 0x0006F , 0x00304 , 0x0014D }, { 0x0006F , 0x00306 , 0x0014F }, { 0x0006F , 0x00307 , 0x0022F }, { 0x0006F , 0x00308 , 0x000F6 }, { 0x0006F , 0x00309 , 0x01ECF }, { 0x0006F , 0x0030B , 0x00151 }, { 0x0006F , 0x0030C , 0x001D2 }, { 0x0006F , 0x0030F , 0x0020D }, { 0x0006F , 0x00311 , 0x0020F }, { 0x0006F , 0x0031B , 0x001A1 }, { 0x0006F , 0x00323 , 0x01ECD }, { 0x0006F , 0x00328 , 0x001EB }, { 0x00070 , 0x00301 , 0x01E55 }, { 0x00070 , 0x00307 , 0x01E57 }, { 0x00072 , 0x00301 , 0x00155 }, { 0x00072 , 0x00307 , 0x01E59 }, { 0x00072 , 0x0030C , 0x00159 }, { 0x00072 , 0x0030F , 0x00211 }, { 0x00072 , 0x00311 , 0x00213 }, { 0x00072 , 0x00323 , 0x01E5B }, { 0x00072 , 0x00327 , 0x00157 }, { 0x00072 , 0x00331 , 0x01E5F }, { 0x00073 , 0x00301 , 0x0015B }, { 0x00073 , 0x00302 , 0x0015D }, { 0x00073 , 0x00307 , 0x01E61 }, { 0x00073 , 0x0030C , 0x00161 }, { 0x00073 , 0x00323 , 0x01E63 }, { 0x00073 , 0x00326 , 0x00219 }, { 0x00073 , 0x00327 , 0x0015F }, { 0x00074 , 0x00307 , 0x01E6B }, { 0x00074 , 0x00308 , 0x01E97 }, { 0x00074 , 0x0030C , 0x00165 }, { 0x00074 , 0x00323 , 0x01E6D }, { 0x00074 , 0x00326 , 0x0021B }, { 0x00074 , 0x00327 , 0x00163 }, { 0x00074 , 0x0032D , 0x01E71 }, { 0x00074 , 0x00331 , 0x01E6F }, { 0x00075 , 0x00300 , 0x000F9 }, { 0x00075 , 0x00301 , 0x000FA }, { 0x00075 , 0x00302 , 0x000FB }, { 0x00075 , 0x00303 , 0x00169 }, { 0x00075 , 0x00304 , 0x0016B }, { 0x00075 , 0x00306 , 0x0016D }, { 0x00075 , 0x00308 , 0x000FC }, { 0x00075 , 0x00309 , 0x01EE7 }, { 0x00075 , 0x0030A , 0x0016F }, { 0x00075 , 0x0030B , 0x00171 }, { 0x00075 , 0x0030C , 0x001D4 }, { 0x00075 , 0x0030F , 0x00215 }, { 0x00075 , 0x00311 , 0x00217 }, { 0x00075 , 0x0031B , 0x001B0 }, { 0x00075 , 0x00323 , 0x01EE5 }, { 0x00075 , 0x00324 , 0x01E73 }, { 0x00075 , 0x00328 , 0x00173 }, { 0x00075 , 0x0032D , 0x01E77 }, { 0x00075 , 0x00330 , 0x01E75 }, { 0x00076 , 0x00303 , 0x01E7D }, { 0x00076 , 0x00323 , 0x01E7F }, { 0x00077 , 0x00300 , 0x01E81 }, { 0x00077 , 0x00301 , 0x01E83 }, { 0x00077 , 0x00302 , 0x00175 }, { 0x00077 , 0x00307 , 0x01E87 }, { 0x00077 , 0x00308 , 0x01E85 }, { 0x00077 , 0x0030A , 0x01E98 }, { 0x00077 , 0x00323 , 0x01E89 }, { 0x00078 , 0x00307 , 0x01E8B }, { 0x00078 , 0x00308 , 0x01E8D }, { 0x00079 , 0x00300 , 0x01EF3 }, { 0x00079 , 0x00301 , 0x000FD }, { 0x00079 , 0x00302 , 0x00177 }, { 0x00079 , 0x00303 , 0x01EF9 }, { 0x00079 , 0x00304 , 0x00233 }, { 0x00079 , 0x00307 , 0x01E8F }, { 0x00079 , 0x00308 , 0x000FF }, { 0x00079 , 0x00309 , 0x01EF7 }, { 0x00079 , 0x0030A , 0x01E99 }, { 0x00079 , 0x00323 , 0x01EF5 }, { 0x0007A , 0x00301 , 0x0017A }, { 0x0007A , 0x00302 , 0x01E91 }, { 0x0007A , 0x00307 , 0x0017C }, { 0x0007A , 0x0030C , 0x0017E }, { 0x0007A , 0x00323 , 0x01E93 }, { 0x0007A , 0x00331 , 0x01E95 }, { 0x000A8 , 0x00300 , 0x01FED }, { 0x000A8 , 0x00301 , 0x00385 }, { 0x000A8 , 0x00342 , 0x01FC1 }, { 0x000C2 , 0x00300 , 0x01EA6 }, { 0x000C2 , 0x00301 , 0x01EA4 }, { 0x000C2 , 0x00303 , 0x01EAA }, { 0x000C2 , 0x00309 , 0x01EA8 }, { 0x000C4 , 0x00304 , 0x001DE }, { 0x000C5 , 0x00301 , 0x001FA }, { 0x000C6 , 0x00301 , 0x001FC }, { 0x000C6 , 0x00304 , 0x001E2 }, { 0x000C7 , 0x00301 , 0x01E08 }, { 0x000CA , 0x00300 , 0x01EC0 }, { 0x000CA , 0x00301 , 0x01EBE }, { 0x000CA , 0x00303 , 0x01EC4 }, { 0x000CA , 0x00309 , 0x01EC2 }, { 0x000CF , 0x00301 , 0x01E2E }, { 0x000D4 , 0x00300 , 0x01ED2 }, { 0x000D4 , 0x00301 , 0x01ED0 }, { 0x000D4 , 0x00303 , 0x01ED6 }, { 0x000D4 , 0x00309 , 0x01ED4 }, { 0x000D5 , 0x00301 , 0x01E4C }, { 0x000D5 , 0x00304 , 0x0022C }, { 0x000D5 , 0x00308 , 0x01E4E }, { 0x000D6 , 0x00304 , 0x0022A }, { 0x000D8 , 0x00301 , 0x001FE }, { 0x000DC , 0x00300 , 0x001DB }, { 0x000DC , 0x00301 , 0x001D7 }, { 0x000DC , 0x00304 , 0x001D5 }, { 0x000DC , 0x0030C , 0x001D9 }, { 0x000E2 , 0x00300 , 0x01EA7 }, { 0x000E2 , 0x00301 , 0x01EA5 }, { 0x000E2 , 0x00303 , 0x01EAB }, { 0x000E2 , 0x00309 , 0x01EA9 }, { 0x000E4 , 0x00304 , 0x001DF }, { 0x000E5 , 0x00301 , 0x001FB }, { 0x000E6 , 0x00301 , 0x001FD }, { 0x000E6 , 0x00304 , 0x001E3 }, { 0x000E7 , 0x00301 , 0x01E09 }, { 0x000EA , 0x00300 , 0x01EC1 }, { 0x000EA , 0x00301 , 0x01EBF }, { 0x000EA , 0x00303 , 0x01EC5 }, { 0x000EA , 0x00309 , 0x01EC3 }, { 0x000EF , 0x00301 , 0x01E2F }, { 0x000F4 , 0x00300 , 0x01ED3 }, { 0x000F4 , 0x00301 , 0x01ED1 }, { 0x000F4 , 0x00303 , 0x01ED7 }, { 0x000F4 , 0x00309 , 0x01ED5 }, { 0x000F5 , 0x00301 , 0x01E4D }, { 0x000F5 , 0x00304 , 0x0022D }, { 0x000F5 , 0x00308 , 0x01E4F }, { 0x000F6 , 0x00304 , 0x0022B }, { 0x000F8 , 0x00301 , 0x001FF }, { 0x000FC , 0x00300 , 0x001DC }, { 0x000FC , 0x00301 , 0x001D8 }, { 0x000FC , 0x00304 , 0x001D6 }, { 0x000FC , 0x0030C , 0x001DA }, { 0x00102 , 0x00300 , 0x01EB0 }, { 0x00102 , 0x00301 , 0x01EAE }, { 0x00102 , 0x00303 , 0x01EB4 }, { 0x00102 , 0x00309 , 0x01EB2 }, { 0x00103 , 0x00300 , 0x01EB1 }, { 0x00103 , 0x00301 , 0x01EAF }, { 0x00103 , 0x00303 , 0x01EB5 }, { 0x00103 , 0x00309 , 0x01EB3 }, { 0x00112 , 0x00300 , 0x01E14 }, { 0x00112 , 0x00301 , 0x01E16 }, { 0x00113 , 0x00300 , 0x01E15 }, { 0x00113 , 0x00301 , 0x01E17 }, { 0x0014C , 0x00300 , 0x01E50 }, { 0x0014C , 0x00301 , 0x01E52 }, { 0x0014D , 0x00300 , 0x01E51 }, { 0x0014D , 0x00301 , 0x01E53 }, { 0x0015A , 0x00307 , 0x01E64 }, { 0x0015B , 0x00307 , 0x01E65 }, { 0x00160 , 0x00307 , 0x01E66 }, { 0x00161 , 0x00307 , 0x01E67 }, { 0x00168 , 0x00301 , 0x01E78 }, { 0x00169 , 0x00301 , 0x01E79 }, { 0x0016A , 0x00308 , 0x01E7A }, { 0x0016B , 0x00308 , 0x01E7B }, { 0x0017F , 0x00307 , 0x01E9B }, { 0x001A0 , 0x00300 , 0x01EDC }, { 0x001A0 , 0x00301 , 0x01EDA }, { 0x001A0 , 0x00303 , 0x01EE0 }, { 0x001A0 , 0x00309 , 0x01EDE }, { 0x001A0 , 0x00323 , 0x01EE2 }, { 0x001A1 , 0x00300 , 0x01EDD }, { 0x001A1 , 0x00301 , 0x01EDB }, { 0x001A1 , 0x00303 , 0x01EE1 }, { 0x001A1 , 0x00309 , 0x01EDF }, { 0x001A1 , 0x00323 , 0x01EE3 }, { 0x001AF , 0x00300 , 0x01EEA }, { 0x001AF , 0x00301 , 0x01EE8 }, { 0x001AF , 0x00303 , 0x01EEE }, { 0x001AF , 0x00309 , 0x01EEC }, { 0x001AF , 0x00323 , 0x01EF0 }, { 0x001B0 , 0x00300 , 0x01EEB }, { 0x001B0 , 0x00301 , 0x01EE9 }, { 0x001B0 , 0x00303 , 0x01EEF }, { 0x001B0 , 0x00309 , 0x01EED }, { 0x001B0 , 0x00323 , 0x01EF1 }, { 0x001B7 , 0x0030C , 0x001EE }, { 0x001EA , 0x00304 , 0x001EC }, { 0x001EB , 0x00304 , 0x001ED }, { 0x00226 , 0x00304 , 0x001E0 }, { 0x00227 , 0x00304 , 0x001E1 }, { 0x00228 , 0x00306 , 0x01E1C }, { 0x00229 , 0x00306 , 0x01E1D }, { 0x0022E , 0x00304 , 0x00230 }, { 0x0022F , 0x00304 , 0x00231 }, { 0x00292 , 0x0030C , 0x001EF }, { 0x00391 , 0x00300 , 0x01FBA }, { 0x00391 , 0x00301 , 0x00386 }, { 0x00391 , 0x00304 , 0x01FB9 }, { 0x00391 , 0x00306 , 0x01FB8 }, { 0x00391 , 0x00313 , 0x01F08 }, { 0x00391 , 0x00314 , 0x01F09 }, { 0x00391 , 0x00345 , 0x01FBC }, { 0x00395 , 0x00300 , 0x01FC8 }, { 0x00395 , 0x00301 , 0x00388 }, { 0x00395 , 0x00313 , 0x01F18 }, { 0x00395 , 0x00314 , 0x01F19 }, { 0x00397 , 0x00300 , 0x01FCA }, { 0x00397 , 0x00301 , 0x00389 }, { 0x00397 , 0x00313 , 0x01F28 }, { 0x00397 , 0x00314 , 0x01F29 }, { 0x00397 , 0x00345 , 0x01FCC }, { 0x00399 , 0x00300 , 0x01FDA }, { 0x00399 , 0x00301 , 0x0038A }, { 0x00399 , 0x00304 , 0x01FD9 }, { 0x00399 , 0x00306 , 0x01FD8 }, { 0x00399 , 0x00308 , 0x003AA }, { 0x00399 , 0x00313 , 0x01F38 }, { 0x00399 , 0x00314 , 0x01F39 }, { 0x0039F , 0x00300 , 0x01FF8 }, { 0x0039F , 0x00301 , 0x0038C }, { 0x0039F , 0x00313 , 0x01F48 }, { 0x0039F , 0x00314 , 0x01F49 }, { 0x003A1 , 0x00314 , 0x01FEC }, { 0x003A5 , 0x00300 , 0x01FEA }, { 0x003A5 , 0x00301 , 0x0038E }, { 0x003A5 , 0x00304 , 0x01FE9 }, { 0x003A5 , 0x00306 , 0x01FE8 }, { 0x003A5 , 0x00308 , 0x003AB }, { 0x003A5 , 0x00314 , 0x01F59 }, { 0x003A9 , 0x00300 , 0x01FFA }, { 0x003A9 , 0x00301 , 0x0038F }, { 0x003A9 , 0x00313 , 0x01F68 }, { 0x003A9 , 0x00314 , 0x01F69 }, { 0x003A9 , 0x00345 , 0x01FFC }, { 0x003AC , 0x00345 , 0x01FB4 }, { 0x003AE , 0x00345 , 0x01FC4 }, { 0x003B1 , 0x00300 , 0x01F70 }, { 0x003B1 , 0x00301 , 0x003AC }, { 0x003B1 , 0x00304 , 0x01FB1 }, { 0x003B1 , 0x00306 , 0x01FB0 }, { 0x003B1 , 0x00313 , 0x01F00 }, { 0x003B1 , 0x00314 , 0x01F01 }, { 0x003B1 , 0x00342 , 0x01FB6 }, { 0x003B1 , 0x00345 , 0x01FB3 }, { 0x003B5 , 0x00300 , 0x01F72 }, { 0x003B5 , 0x00301 , 0x003AD }, { 0x003B5 , 0x00313 , 0x01F10 }, { 0x003B5 , 0x00314 , 0x01F11 }, { 0x003B7 , 0x00300 , 0x01F74 }, { 0x003B7 , 0x00301 , 0x003AE }, { 0x003B7 , 0x00313 , 0x01F20 }, { 0x003B7 , 0x00314 , 0x01F21 }, { 0x003B7 , 0x00342 , 0x01FC6 }, { 0x003B7 , 0x00345 , 0x01FC3 }, { 0x003B9 , 0x00300 , 0x01F76 }, { 0x003B9 , 0x00301 , 0x003AF }, { 0x003B9 , 0x00304 , 0x01FD1 }, { 0x003B9 , 0x00306 , 0x01FD0 }, { 0x003B9 , 0x00308 , 0x003CA }, { 0x003B9 , 0x00313 , 0x01F30 }, { 0x003B9 , 0x00314 , 0x01F31 }, { 0x003B9 , 0x00342 , 0x01FD6 }, { 0x003BF , 0x00300 , 0x01F78 }, { 0x003BF , 0x00301 , 0x003CC }, { 0x003BF , 0x00313 , 0x01F40 }, { 0x003BF , 0x00314 , 0x01F41 }, { 0x003C1 , 0x00313 , 0x01FE4 }, { 0x003C1 , 0x00314 , 0x01FE5 }, { 0x003C5 , 0x00300 , 0x01F7A }, { 0x003C5 , 0x00301 , 0x003CD }, { 0x003C5 , 0x00304 , 0x01FE1 }, { 0x003C5 , 0x00306 , 0x01FE0 }, { 0x003C5 , 0x00308 , 0x003CB }, { 0x003C5 , 0x00313 , 0x01F50 }, { 0x003C5 , 0x00314 , 0x01F51 }, { 0x003C5 , 0x00342 , 0x01FE6 }, { 0x003C9 , 0x00300 , 0x01F7C }, { 0x003C9 , 0x00301 , 0x003CE }, { 0x003C9 , 0x00313 , 0x01F60 }, { 0x003C9 , 0x00314 , 0x01F61 }, { 0x003C9 , 0x00342 , 0x01FF6 }, { 0x003C9 , 0x00345 , 0x01FF3 }, { 0x003CA , 0x00300 , 0x01FD2 }, { 0x003CA , 0x00301 , 0x00390 }, { 0x003CA , 0x00342 , 0x01FD7 }, { 0x003CB , 0x00300 , 0x01FE2 }, { 0x003CB , 0x00301 , 0x003B0 }, { 0x003CB , 0x00342 , 0x01FE7 }, { 0x003CE , 0x00345 , 0x01FF4 }, { 0x003D2 , 0x00301 , 0x003D3 }, { 0x003D2 , 0x00308 , 0x003D4 }, { 0x00406 , 0x00308 , 0x00407 }, { 0x00410 , 0x00306 , 0x004D0 }, { 0x00410 , 0x00308 , 0x004D2 }, { 0x00413 , 0x00301 , 0x00403 }, { 0x00415 , 0x00300 , 0x00400 }, { 0x00415 , 0x00306 , 0x004D6 }, { 0x00415 , 0x00308 , 0x00401 }, { 0x00416 , 0x00306 , 0x004C1 }, { 0x00416 , 0x00308 , 0x004DC }, { 0x00417 , 0x00308 , 0x004DE }, { 0x00418 , 0x00300 , 0x0040D }, { 0x00418 , 0x00304 , 0x004E2 }, { 0x00418 , 0x00306 , 0x00419 }, { 0x00418 , 0x00308 , 0x004E4 }, { 0x0041A , 0x00301 , 0x0040C }, { 0x0041E , 0x00308 , 0x004E6 }, { 0x00423 , 0x00304 , 0x004EE }, { 0x00423 , 0x00306 , 0x0040E }, { 0x00423 , 0x00308 , 0x004F0 }, { 0x00423 , 0x0030B , 0x004F2 }, { 0x00427 , 0x00308 , 0x004F4 }, { 0x0042B , 0x00308 , 0x004F8 }, { 0x0042D , 0x00308 , 0x004EC }, { 0x00430 , 0x00306 , 0x004D1 }, { 0x00430 , 0x00308 , 0x004D3 }, { 0x00433 , 0x00301 , 0x00453 }, { 0x00435 , 0x00300 , 0x00450 }, { 0x00435 , 0x00306 , 0x004D7 }, { 0x00435 , 0x00308 , 0x00451 }, { 0x00436 , 0x00306 , 0x004C2 }, { 0x00436 , 0x00308 , 0x004DD }, { 0x00437 , 0x00308 , 0x004DF }, { 0x00438 , 0x00300 , 0x0045D }, { 0x00438 , 0x00304 , 0x004E3 }, { 0x00438 , 0x00306 , 0x00439 }, { 0x00438 , 0x00308 , 0x004E5 }, { 0x0043A , 0x00301 , 0x0045C }, { 0x0043E , 0x00308 , 0x004E7 }, { 0x00443 , 0x00304 , 0x004EF }, { 0x00443 , 0x00306 , 0x0045E }, { 0x00443 , 0x00308 , 0x004F1 }, { 0x00443 , 0x0030B , 0x004F3 }, { 0x00447 , 0x00308 , 0x004F5 }, { 0x0044B , 0x00308 , 0x004F9 }, { 0x0044D , 0x00308 , 0x004ED }, { 0x00456 , 0x00308 , 0x00457 }, { 0x00474 , 0x0030F , 0x00476 }, { 0x00475 , 0x0030F , 0x00477 }, { 0x004D8 , 0x00308 , 0x004DA }, { 0x004D9 , 0x00308 , 0x004DB }, { 0x004E8 , 0x00308 , 0x004EA }, { 0x004E9 , 0x00308 , 0x004EB }, { 0x00627 , 0x00653 , 0x00622 }, { 0x00627 , 0x00654 , 0x00623 }, { 0x00627 , 0x00655 , 0x00625 }, { 0x00648 , 0x00654 , 0x00624 }, { 0x0064A , 0x00654 , 0x00626 }, { 0x006C1 , 0x00654 , 0x006C2 }, { 0x006D2 , 0x00654 , 0x006D3 }, { 0x006D5 , 0x00654 , 0x006C0 }, { 0x00928 , 0x0093C , 0x00929 }, { 0x00930 , 0x0093C , 0x00931 }, { 0x00933 , 0x0093C , 0x00934 }, { 0x009C7 , 0x009BE , 0x009CB }, { 0x009C7 , 0x009D7 , 0x009CC }, { 0x00B47 , 0x00B3E , 0x00B4B }, { 0x00B47 , 0x00B56 , 0x00B48 }, { 0x00B47 , 0x00B57 , 0x00B4C }, { 0x00B92 , 0x00BD7 , 0x00B94 }, { 0x00BC6 , 0x00BBE , 0x00BCA }, { 0x00BC6 , 0x00BD7 , 0x00BCC }, { 0x00BC7 , 0x00BBE , 0x00BCB }, { 0x00C46 , 0x00C56 , 0x00C48 }, { 0x00CBF , 0x00CD5 , 0x00CC0 }, { 0x00CC6 , 0x00CC2 , 0x00CCA }, { 0x00CC6 , 0x00CD5 , 0x00CC7 }, { 0x00CC6 , 0x00CD6 , 0x00CC8 }, { 0x00CCA , 0x00CD5 , 0x00CCB }, { 0x00D46 , 0x00D3E , 0x00D4A }, { 0x00D46 , 0x00D57 , 0x00D4C }, { 0x00D47 , 0x00D3E , 0x00D4B }, { 0x00DD9 , 0x00DCA , 0x00DDA }, { 0x00DD9 , 0x00DCF , 0x00DDC }, { 0x00DD9 , 0x00DDF , 0x00DDE }, { 0x00DDC , 0x00DCA , 0x00DDD }, { 0x01025 , 0x0102E , 0x01026 }, { 0x01B05 , 0x01B35 , 0x01B06 }, { 0x01B07 , 0x01B35 , 0x01B08 }, { 0x01B09 , 0x01B35 , 0x01B0A }, { 0x01B0B , 0x01B35 , 0x01B0C }, { 0x01B0D , 0x01B35 , 0x01B0E }, { 0x01B11 , 0x01B35 , 0x01B12 }, { 0x01B3A , 0x01B35 , 0x01B3B }, { 0x01B3C , 0x01B35 , 0x01B3D }, { 0x01B3E , 0x01B35 , 0x01B40 }, { 0x01B3F , 0x01B35 , 0x01B41 }, { 0x01B42 , 0x01B35 , 0x01B43 }, { 0x01E36 , 0x00304 , 0x01E38 }, { 0x01E37 , 0x00304 , 0x01E39 }, { 0x01E5A , 0x00304 , 0x01E5C }, { 0x01E5B , 0x00304 , 0x01E5D }, { 0x01E62 , 0x00307 , 0x01E68 }, { 0x01E63 , 0x00307 , 0x01E69 }, { 0x01EA0 , 0x00302 , 0x01EAC }, { 0x01EA0 , 0x00306 , 0x01EB6 }, { 0x01EA1 , 0x00302 , 0x01EAD }, { 0x01EA1 , 0x00306 , 0x01EB7 }, { 0x01EB8 , 0x00302 , 0x01EC6 }, { 0x01EB9 , 0x00302 , 0x01EC7 }, { 0x01ECC , 0x00302 , 0x01ED8 }, { 0x01ECD , 0x00302 , 0x01ED9 }, { 0x01F00 , 0x00300 , 0x01F02 }, { 0x01F00 , 0x00301 , 0x01F04 }, { 0x01F00 , 0x00342 , 0x01F06 }, { 0x01F00 , 0x00345 , 0x01F80 }, { 0x01F01 , 0x00300 , 0x01F03 }, { 0x01F01 , 0x00301 , 0x01F05 }, { 0x01F01 , 0x00342 , 0x01F07 }, { 0x01F01 , 0x00345 , 0x01F81 }, { 0x01F02 , 0x00345 , 0x01F82 }, { 0x01F03 , 0x00345 , 0x01F83 }, { 0x01F04 , 0x00345 , 0x01F84 }, { 0x01F05 , 0x00345 , 0x01F85 }, { 0x01F06 , 0x00345 , 0x01F86 }, { 0x01F07 , 0x00345 , 0x01F87 }, { 0x01F08 , 0x00300 , 0x01F0A }, { 0x01F08 , 0x00301 , 0x01F0C }, { 0x01F08 , 0x00342 , 0x01F0E }, { 0x01F08 , 0x00345 , 0x01F88 }, { 0x01F09 , 0x00300 , 0x01F0B }, { 0x01F09 , 0x00301 , 0x01F0D }, { 0x01F09 , 0x00342 , 0x01F0F }, { 0x01F09 , 0x00345 , 0x01F89 }, { 0x01F0A , 0x00345 , 0x01F8A }, { 0x01F0B , 0x00345 , 0x01F8B }, { 0x01F0C , 0x00345 , 0x01F8C }, { 0x01F0D , 0x00345 , 0x01F8D }, { 0x01F0E , 0x00345 , 0x01F8E }, { 0x01F0F , 0x00345 , 0x01F8F }, { 0x01F10 , 0x00300 , 0x01F12 }, { 0x01F10 , 0x00301 , 0x01F14 }, { 0x01F11 , 0x00300 , 0x01F13 }, { 0x01F11 , 0x00301 , 0x01F15 }, { 0x01F18 , 0x00300 , 0x01F1A }, { 0x01F18 , 0x00301 , 0x01F1C }, { 0x01F19 , 0x00300 , 0x01F1B }, { 0x01F19 , 0x00301 , 0x01F1D }, { 0x01F20 , 0x00300 , 0x01F22 }, { 0x01F20 , 0x00301 , 0x01F24 }, { 0x01F20 , 0x00342 , 0x01F26 }, { 0x01F20 , 0x00345 , 0x01F90 }, { 0x01F21 , 0x00300 , 0x01F23 }, { 0x01F21 , 0x00301 , 0x01F25 }, { 0x01F21 , 0x00342 , 0x01F27 }, { 0x01F21 , 0x00345 , 0x01F91 }, { 0x01F22 , 0x00345 , 0x01F92 }, { 0x01F23 , 0x00345 , 0x01F93 }, { 0x01F24 , 0x00345 , 0x01F94 }, { 0x01F25 , 0x00345 , 0x01F95 }, { 0x01F26 , 0x00345 , 0x01F96 }, { 0x01F27 , 0x00345 , 0x01F97 }, { 0x01F28 , 0x00300 , 0x01F2A }, { 0x01F28 , 0x00301 , 0x01F2C }, { 0x01F28 , 0x00342 , 0x01F2E }, { 0x01F28 , 0x00345 , 0x01F98 }, { 0x01F29 , 0x00300 , 0x01F2B }, { 0x01F29 , 0x00301 , 0x01F2D }, { 0x01F29 , 0x00342 , 0x01F2F }, { 0x01F29 , 0x00345 , 0x01F99 }, { 0x01F2A , 0x00345 , 0x01F9A }, { 0x01F2B , 0x00345 , 0x01F9B }, { 0x01F2C , 0x00345 , 0x01F9C }, { 0x01F2D , 0x00345 , 0x01F9D }, { 0x01F2E , 0x00345 , 0x01F9E }, { 0x01F2F , 0x00345 , 0x01F9F }, { 0x01F30 , 0x00300 , 0x01F32 }, { 0x01F30 , 0x00301 , 0x01F34 }, { 0x01F30 , 0x00342 , 0x01F36 }, { 0x01F31 , 0x00300 , 0x01F33 }, { 0x01F31 , 0x00301 , 0x01F35 }, { 0x01F31 , 0x00342 , 0x01F37 }, { 0x01F38 , 0x00300 , 0x01F3A }, { 0x01F38 , 0x00301 , 0x01F3C }, { 0x01F38 , 0x00342 , 0x01F3E }, { 0x01F39 , 0x00300 , 0x01F3B }, { 0x01F39 , 0x00301 , 0x01F3D }, { 0x01F39 , 0x00342 , 0x01F3F }, { 0x01F40 , 0x00300 , 0x01F42 }, { 0x01F40 , 0x00301 , 0x01F44 }, { 0x01F41 , 0x00300 , 0x01F43 }, { 0x01F41 , 0x00301 , 0x01F45 }, { 0x01F48 , 0x00300 , 0x01F4A }, { 0x01F48 , 0x00301 , 0x01F4C }, { 0x01F49 , 0x00300 , 0x01F4B }, { 0x01F49 , 0x00301 , 0x01F4D }, { 0x01F50 , 0x00300 , 0x01F52 }, { 0x01F50 , 0x00301 , 0x01F54 }, { 0x01F50 , 0x00342 , 0x01F56 }, { 0x01F51 , 0x00300 , 0x01F53 }, { 0x01F51 , 0x00301 , 0x01F55 }, { 0x01F51 , 0x00342 , 0x01F57 }, { 0x01F59 , 0x00300 , 0x01F5B }, { 0x01F59 , 0x00301 , 0x01F5D }, { 0x01F59 , 0x00342 , 0x01F5F }, { 0x01F60 , 0x00300 , 0x01F62 }, { 0x01F60 , 0x00301 , 0x01F64 }, { 0x01F60 , 0x00342 , 0x01F66 }, { 0x01F60 , 0x00345 , 0x01FA0 }, { 0x01F61 , 0x00300 , 0x01F63 }, { 0x01F61 , 0x00301 , 0x01F65 }, { 0x01F61 , 0x00342 , 0x01F67 }, { 0x01F61 , 0x00345 , 0x01FA1 }, { 0x01F62 , 0x00345 , 0x01FA2 }, { 0x01F63 , 0x00345 , 0x01FA3 }, { 0x01F64 , 0x00345 , 0x01FA4 }, { 0x01F65 , 0x00345 , 0x01FA5 }, { 0x01F66 , 0x00345 , 0x01FA6 }, { 0x01F67 , 0x00345 , 0x01FA7 }, { 0x01F68 , 0x00300 , 0x01F6A }, { 0x01F68 , 0x00301 , 0x01F6C }, { 0x01F68 , 0x00342 , 0x01F6E }, { 0x01F68 , 0x00345 , 0x01FA8 }, { 0x01F69 , 0x00300 , 0x01F6B }, { 0x01F69 , 0x00301 , 0x01F6D }, { 0x01F69 , 0x00342 , 0x01F6F }, { 0x01F69 , 0x00345 , 0x01FA9 }, { 0x01F6A , 0x00345 , 0x01FAA }, { 0x01F6B , 0x00345 , 0x01FAB }, { 0x01F6C , 0x00345 , 0x01FAC }, { 0x01F6D , 0x00345 , 0x01FAD }, { 0x01F6E , 0x00345 , 0x01FAE }, { 0x01F6F , 0x00345 , 0x01FAF }, { 0x01F70 , 0x00345 , 0x01FB2 }, { 0x01F74 , 0x00345 , 0x01FC2 }, { 0x01F7C , 0x00345 , 0x01FF2 }, { 0x01FB6 , 0x00345 , 0x01FB7 }, { 0x01FBF , 0x00300 , 0x01FCD }, { 0x01FBF , 0x00301 , 0x01FCE }, { 0x01FBF , 0x00342 , 0x01FCF }, { 0x01FC6 , 0x00345 , 0x01FC7 }, { 0x01FF6 , 0x00345 , 0x01FF7 }, { 0x01FFE , 0x00300 , 0x01FDD }, { 0x01FFE , 0x00301 , 0x01FDE }, { 0x01FFE , 0x00342 , 0x01FDF }, { 0x02190 , 0x00338 , 0x0219A }, { 0x02192 , 0x00338 , 0x0219B }, { 0x02194 , 0x00338 , 0x021AE }, { 0x021D0 , 0x00338 , 0x021CD }, { 0x021D2 , 0x00338 , 0x021CF }, { 0x021D4 , 0x00338 , 0x021CE }, { 0x02203 , 0x00338 , 0x02204 }, { 0x02208 , 0x00338 , 0x02209 }, { 0x0220B , 0x00338 , 0x0220C }, { 0x02223 , 0x00338 , 0x02224 }, { 0x02225 , 0x00338 , 0x02226 }, { 0x0223C , 0x00338 , 0x02241 }, { 0x02243 , 0x00338 , 0x02244 }, { 0x02245 , 0x00338 , 0x02247 }, { 0x02248 , 0x00338 , 0x02249 }, { 0x0224D , 0x00338 , 0x0226D }, { 0x02261 , 0x00338 , 0x02262 }, { 0x02264 , 0x00338 , 0x02270 }, { 0x02265 , 0x00338 , 0x02271 }, { 0x02272 , 0x00338 , 0x02274 }, { 0x02273 , 0x00338 , 0x02275 }, { 0x02276 , 0x00338 , 0x02278 }, { 0x02277 , 0x00338 , 0x02279 }, { 0x0227A , 0x00338 , 0x02280 }, { 0x0227B , 0x00338 , 0x02281 }, { 0x0227C , 0x00338 , 0x022E0 }, { 0x0227D , 0x00338 , 0x022E1 }, { 0x02282 , 0x00338 , 0x02284 }, { 0x02283 , 0x00338 , 0x02285 }, { 0x02286 , 0x00338 , 0x02288 }, { 0x02287 , 0x00338 , 0x02289 }, { 0x02291 , 0x00338 , 0x022E2 }, { 0x02292 , 0x00338 , 0x022E3 }, { 0x022A2 , 0x00338 , 0x022AC }, { 0x022A8 , 0x00338 , 0x022AD }, { 0x022A9 , 0x00338 , 0x022AE }, { 0x022AB , 0x00338 , 0x022AF }, { 0x022B2 , 0x00338 , 0x022EA }, { 0x022B3 , 0x00338 , 0x022EB }, { 0x022B4 , 0x00338 , 0x022EC }, { 0x022B5 , 0x00338 , 0x022ED }, { 0x03046 , 0x03099 , 0x03094 }, { 0x0304B , 0x03099 , 0x0304C }, { 0x0304D , 0x03099 , 0x0304E }, { 0x0304F , 0x03099 , 0x03050 }, { 0x03051 , 0x03099 , 0x03052 }, { 0x03053 , 0x03099 , 0x03054 }, { 0x03055 , 0x03099 , 0x03056 }, { 0x03057 , 0x03099 , 0x03058 }, { 0x03059 , 0x03099 , 0x0305A }, { 0x0305B , 0x03099 , 0x0305C }, { 0x0305D , 0x03099 , 0x0305E }, { 0x0305F , 0x03099 , 0x03060 }, { 0x03061 , 0x03099 , 0x03062 }, { 0x03064 , 0x03099 , 0x03065 }, { 0x03066 , 0x03099 , 0x03067 }, { 0x03068 , 0x03099 , 0x03069 }, { 0x0306F , 0x03099 , 0x03070 }, { 0x0306F , 0x0309A , 0x03071 }, { 0x03072 , 0x03099 , 0x03073 }, { 0x03072 , 0x0309A , 0x03074 }, { 0x03075 , 0x03099 , 0x03076 }, { 0x03075 , 0x0309A , 0x03077 }, { 0x03078 , 0x03099 , 0x03079 }, { 0x03078 , 0x0309A , 0x0307A }, { 0x0307B , 0x03099 , 0x0307C }, { 0x0307B , 0x0309A , 0x0307D }, { 0x0309D , 0x03099 , 0x0309E }, { 0x030A6 , 0x03099 , 0x030F4 }, { 0x030AB , 0x03099 , 0x030AC }, { 0x030AD , 0x03099 , 0x030AE }, { 0x030AF , 0x03099 , 0x030B0 }, { 0x030B1 , 0x03099 , 0x030B2 }, { 0x030B3 , 0x03099 , 0x030B4 }, { 0x030B5 , 0x03099 , 0x030B6 }, { 0x030B7 , 0x03099 , 0x030B8 }, { 0x030B9 , 0x03099 , 0x030BA }, { 0x030BB , 0x03099 , 0x030BC }, { 0x030BD , 0x03099 , 0x030BE }, { 0x030BF , 0x03099 , 0x030C0 }, { 0x030C1 , 0x03099 , 0x030C2 }, { 0x030C4 , 0x03099 , 0x030C5 }, { 0x030C6 , 0x03099 , 0x030C7 }, { 0x030C8 , 0x03099 , 0x030C9 }, { 0x030CF , 0x03099 , 0x030D0 }, { 0x030CF , 0x0309A , 0x030D1 }, { 0x030D2 , 0x03099 , 0x030D3 }, { 0x030D2 , 0x0309A , 0x030D4 }, { 0x030D5 , 0x03099 , 0x030D6 }, { 0x030D5 , 0x0309A , 0x030D7 }, { 0x030D8 , 0x03099 , 0x030D9 }, { 0x030D8 , 0x0309A , 0x030DA }, { 0x030DB , 0x03099 , 0x030DC }, { 0x030DB , 0x0309A , 0x030DD }, { 0x030EF , 0x03099 , 0x030F7 }, { 0x030F0 , 0x03099 , 0x030F8 }, { 0x030F1 , 0x03099 , 0x030F9 }, { 0x030F2 , 0x03099 , 0x030FA }, { 0x030FD , 0x03099 , 0x030FE }, { 0x11099 , 0x110BA , 0x1109A }, { 0x1109B , 0x110BA , 0x1109C }, { 0x110A5 , 0x110BA , 0x110AB }, }; #define CANONICAL_CLASS_MIN 0x0300 #define CANONICAL_CLASS_MAX 0x1D244 #define IS_DECOMPOSABLE_BLOCK(uc) \ (((uc)>>8) <= 0x1D2 && u_decomposable_blocks[(uc)>>8]) static const char u_decomposable_blocks[0x1D2+1] = { 0,0,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,1,0,0,0,1,1,1,1,1,1,1,0,0, 1,0,0,0,0,0,0,0,0,0,0,0,1,1,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,1,0,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,1,0, 0,1,0,0,0,0,0,0,0,0,1,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1, }; /* Get Canonical Combining Class(CCC). */ #define CCC(uc) \ (((uc) > 0x1D244)?0:\ ccc_val[ccc_val_index[ccc_index[(uc)>>8]][((uc)>>4)&0x0F]][(uc)&0x0F]) -/* The table of the value of Canonical Cimbining Class */ +/* The table of the value of Canonical Combining Class */ static const unsigned char ccc_val[][16] = { /* idx=0: XXXX0 - XXXXF */ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=1: 00300 - 0030F */ {230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230 }, /* idx=2: 00310 - 0031F */ {230, 230, 230, 230, 230, 232, 220, 220, 220, 220, 232, 216, 220, 220, 220, 220 }, /* idx=3: 00320 - 0032F */ {220, 202, 202, 220, 220, 220, 220, 202, 202, 220, 220, 220, 220, 220, 220, 220 }, /* idx=4: 00330 - 0033F */ {220, 220, 220, 220, 1, 1, 1, 1, 1, 220, 220, 220, 220, 230, 230, 230 }, /* idx=5: 00340 - 0034F */ {230, 230, 230, 230, 230, 240, 230, 220, 220, 220, 230, 230, 230, 220, 220, 0 }, /* idx=6: 00350 - 0035F */ {230, 230, 230, 220, 220, 220, 220, 230, 232, 220, 220, 230, 233, 234, 234, 233 }, /* idx=7: 00360 - 0036F */ {234, 234, 233, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230 }, /* idx=8: 00480 - 0048F */ {0, 0, 0, 230, 230, 230, 230, 230, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=9: 00590 - 0059F */ {0, 220, 230, 230, 230, 230, 220, 230, 230, 230, 222, 220, 230, 230, 230, 230 }, /* idx=10: 005A0 - 005AF */ {230, 230, 220, 220, 220, 220, 220, 220, 230, 230, 220, 230, 230, 222, 228, 230 }, /* idx=11: 005B0 - 005BF */ {10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 19, 20, 21, 22, 0, 23 }, /* idx=12: 005C0 - 005CF */ {0, 24, 25, 0, 230, 220, 0, 18, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=13: 00610 - 0061F */ {230, 230, 230, 230, 230, 230, 230, 230, 30, 31, 32, 0, 0, 0, 0, 0 }, /* idx=14: 00640 - 0064F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 27, 28, 29, 30, 31 }, /* idx=15: 00650 - 0065F */ {32, 33, 34, 230, 230, 220, 220, 230, 230, 230, 230, 230, 220, 230, 230, 220 }, /* idx=16: 00670 - 0067F */ {35, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=17: 006D0 - 006DF */ {0, 0, 0, 0, 0, 0, 230, 230, 230, 230, 230, 230, 230, 0, 0, 230 }, /* idx=18: 006E0 - 006EF */ {230, 230, 230, 220, 230, 0, 0, 230, 230, 0, 220, 230, 230, 220, 0, 0 }, /* idx=19: 00710 - 0071F */ {0, 36, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=20: 00730 - 0073F */ {230, 220, 230, 230, 220, 230, 230, 220, 220, 220, 230, 220, 220, 230, 220, 230 }, /* idx=21: 00740 - 0074F */ {230, 230, 220, 230, 220, 230, 220, 230, 220, 230, 230, 0, 0, 0, 0, 0 }, /* idx=22: 007E0 - 007EF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230, 230, 230, 230, 230 }, /* idx=23: 007F0 - 007FF */ {230, 230, 220, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=24: 00810 - 0081F */ {0, 0, 0, 0, 0, 0, 230, 230, 230, 230, 0, 230, 230, 230, 230, 230 }, /* idx=25: 00820 - 0082F */ {230, 230, 230, 230, 0, 230, 230, 230, 0, 230, 230, 230, 230, 230, 0, 0 }, /* idx=26: 00850 - 0085F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 220, 220, 220, 0, 0, 0, 0 }, /* idx=27: 00930 - 0093F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0 }, /* idx=28: 00940 - 0094F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, /* idx=29: 00950 - 0095F */ {0, 230, 220, 230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=30: 009B0 - 009BF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0 }, /* idx=31: 009C0 - 009CF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, /* idx=32: 00A30 - 00A3F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0 }, /* idx=33: 00A40 - 00A4F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, /* idx=34: 00AB0 - 00ABF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0 }, /* idx=35: 00AC0 - 00ACF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, /* idx=36: 00B30 - 00B3F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0 }, /* idx=37: 00B40 - 00B4F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, /* idx=38: 00BC0 - 00BCF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, /* idx=39: 00C40 - 00C4F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, /* idx=40: 00C50 - 00C5F */ {0, 0, 0, 0, 0, 84, 91, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=41: 00CB0 - 00CBF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0 }, /* idx=42: 00CC0 - 00CCF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, /* idx=43: 00D40 - 00D4F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, /* idx=44: 00DC0 - 00DCF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0, 0, 0, 0 }, /* idx=45: 00E30 - 00E3F */ {0, 0, 0, 0, 0, 0, 0, 0, 103, 103, 9, 0, 0, 0, 0, 0 }, /* idx=46: 00E40 - 00E4F */ {0, 0, 0, 0, 0, 0, 0, 0, 107, 107, 107, 107, 0, 0, 0, 0 }, /* idx=47: 00EB0 - 00EBF */ {0, 0, 0, 0, 0, 0, 0, 0, 118, 118, 0, 0, 0, 0, 0, 0 }, /* idx=48: 00EC0 - 00ECF */ {0, 0, 0, 0, 0, 0, 0, 0, 122, 122, 122, 122, 0, 0, 0, 0 }, /* idx=49: 00F10 - 00F1F */ {0, 0, 0, 0, 0, 0, 0, 0, 220, 220, 0, 0, 0, 0, 0, 0 }, /* idx=50: 00F30 - 00F3F */ {0, 0, 0, 0, 0, 220, 0, 220, 0, 216, 0, 0, 0, 0, 0, 0 }, /* idx=51: 00F70 - 00F7F */ {0, 129, 130, 0, 132, 0, 0, 0, 0, 0, 130, 130, 130, 130, 0, 0 }, /* idx=52: 00F80 - 00F8F */ {130, 0, 230, 230, 9, 0, 230, 230, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=53: 00FC0 - 00FCF */ {0, 0, 0, 0, 0, 0, 220, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=54: 01030 - 0103F */ {0, 0, 0, 0, 0, 0, 0, 7, 0, 9, 9, 0, 0, 0, 0, 0 }, /* idx=55: 01080 - 0108F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 220, 0, 0 }, /* idx=56: 01350 - 0135F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230, 230, 230 }, /* idx=57: 01710 - 0171F */ {0, 0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=58: 01730 - 0173F */ {0, 0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=59: 017D0 - 017DF */ {0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230, 0, 0 }, /* idx=60: 018A0 - 018AF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 228, 0, 0, 0, 0, 0, 0 }, /* idx=61: 01930 - 0193F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 222, 230, 220, 0, 0, 0, 0 }, /* idx=62: 01A10 - 01A1F */ {0, 0, 0, 0, 0, 0, 0, 230, 220, 0, 0, 0, 0, 0, 0, 0 }, /* idx=63: 01A60 - 01A6F */ {9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=64: 01A70 - 01A7F */ {0, 0, 0, 0, 0, 230, 230, 230, 230, 230, 230, 230, 230, 0, 0, 220 }, /* idx=65: 01B30 - 01B3F */ {0, 0, 0, 0, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=66: 01B40 - 01B4F */ {0, 0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=67: 01B60 - 01B6F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230, 220, 230, 230, 230 }, /* idx=68: 01B70 - 01B7F */ {230, 230, 230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=69: 01BA0 - 01BAF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0, 0, 0, 0 }, /* idx=70: 01BE0 - 01BEF */ {0, 0, 0, 0, 0, 0, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=71: 01BF0 - 01BFF */ {0, 0, 9, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=72: 01C30 - 01C3F */ {0, 0, 0, 0, 0, 0, 0, 7, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=73: 01CD0 - 01CDF */ {230, 230, 230, 0, 1, 220, 220, 220, 220, 220, 230, 230, 220, 220, 220, 220 }, /* idx=74: 01CE0 - 01CEF */ {230, 0, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 220, 0, 0 }, /* idx=75: 01DC0 - 01DCF */ {230, 230, 220, 230, 230, 230, 230, 230, 230, 230, 220, 230, 230, 234, 214, 220 }, /* idx=76: 01DD0 - 01DDF */ {202, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230 }, /* idx=77: 01DE0 - 01DEF */ {230, 230, 230, 230, 230, 230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=78: 01DF0 - 01DFF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 233, 220, 230, 220 }, /* idx=79: 020D0 - 020DF */ {230, 230, 1, 1, 230, 230, 230, 230, 1, 1, 1, 230, 230, 0, 0, 0 }, /* idx=80: 020E0 - 020EF */ {0, 230, 0, 0, 0, 1, 1, 230, 220, 230, 1, 1, 220, 220, 220, 220 }, /* idx=81: 020F0 - 020FF */ {230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=82: 02CE0 - 02CEF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230 }, /* idx=83: 02CF0 - 02CFF */ {230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=84: 02D70 - 02D7F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9 }, /* idx=85: 02DE0 - 02DEF */ {230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230 }, /* idx=86: 02DF0 - 02DFF */ {230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230 }, /* idx=87: 03020 - 0302F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 218, 228, 232, 222, 224, 224 }, /* idx=88: 03090 - 0309F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 8, 8, 0, 0, 0, 0, 0 }, /* idx=89: 0A660 - 0A66F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230 }, /* idx=90: 0A670 - 0A67F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230, 230, 0, 0 }, /* idx=91: 0A6F0 - 0A6FF */ {230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=92: 0A800 - 0A80F */ {0, 0, 0, 0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=93: 0A8C0 - 0A8CF */ {0, 0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=94: 0A8E0 - 0A8EF */ {230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230, 230 }, /* idx=95: 0A8F0 - 0A8FF */ {230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=96: 0A920 - 0A92F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 220, 220, 220, 0, 0 }, /* idx=97: 0A950 - 0A95F */ {0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=98: 0A9B0 - 0A9BF */ {0, 0, 0, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=99: 0A9C0 - 0A9CF */ {9, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=100: 0AAB0 - 0AABF */ {230, 0, 230, 230, 220, 0, 0, 230, 230, 0, 0, 0, 0, 0, 230, 230 }, /* idx=101: 0AAC0 - 0AACF */ {0, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=102: 0ABE0 - 0ABEF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 0, 0 }, /* idx=103: 0FB10 - 0FB1F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 26, 0 }, /* idx=104: 0FE20 - 0FE2F */ {230, 230, 230, 230, 230, 230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=105: 101F0 - 101FF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 220, 0, 0 }, /* idx=106: 10A00 - 10A0F */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 220, 0, 230 }, /* idx=107: 10A30 - 10A3F */ {0, 0, 0, 0, 0, 0, 0, 0, 230, 1, 220, 0, 0, 0, 0, 9 }, /* idx=108: 11040 - 1104F */ {0, 0, 0, 0, 0, 0, 9, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=109: 110B0 - 110BF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 9, 7, 0, 0, 0, 0, 0 }, /* idx=110: 1D160 - 1D16F */ {0, 0, 0, 0, 0, 216, 216, 1, 1, 1, 0, 0, 0, 226, 216, 216 }, /* idx=111: 1D170 - 1D17F */ {216, 216, 216, 0, 0, 0, 0, 0, 0, 0, 0, 220, 220, 220, 220, 220 }, /* idx=112: 1D180 - 1D18F */ {220, 220, 220, 0, 0, 230, 230, 230, 230, 230, 220, 220, 0, 0, 0, 0 }, /* idx=113: 1D1A0 - 1D1AF */ {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 230, 230, 230, 230, 0, 0 }, /* idx=114: 1D240 - 1D24F */ {0, 0, 230, 230, 230, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, }; /* The index table to ccc_val[*][16] */ static const unsigned char ccc_val_index[][16] = { /* idx=0: XXX00 - XXXFF */ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=1: 00300 - 003FF */ { 1, 2, 3, 4, 5, 6, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=2: 00400 - 004FF */ { 0, 0, 0, 0, 0, 0, 0, 0, 8, 0, 0, 0, 0, 0, 0, 0 }, /* idx=3: 00500 - 005FF */ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 9,10,11,12, 0, 0, 0 }, /* idx=4: 00600 - 006FF */ { 0,13, 0, 0,14,15, 0,16, 0, 0, 0, 0, 0,17,18, 0 }, /* idx=5: 00700 - 007FF */ { 0,19, 0,20,21, 0, 0, 0, 0, 0, 0, 0, 0, 0,22,23 }, /* idx=6: 00800 - 008FF */ { 0,24,25, 0, 0,26, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=7: 00900 - 009FF */ { 0, 0, 0,27,28,29, 0, 0, 0, 0, 0,30,31, 0, 0, 0 }, /* idx=8: 00A00 - 00AFF */ { 0, 0, 0,32,33, 0, 0, 0, 0, 0, 0,34,35, 0, 0, 0 }, /* idx=9: 00B00 - 00BFF */ { 0, 0, 0,36,37, 0, 0, 0, 0, 0, 0, 0,38, 0, 0, 0 }, /* idx=10: 00C00 - 00CFF */ { 0, 0, 0, 0,39,40, 0, 0, 0, 0, 0,41,42, 0, 0, 0 }, /* idx=11: 00D00 - 00DFF */ { 0, 0, 0, 0,43, 0, 0, 0, 0, 0, 0, 0,44, 0, 0, 0 }, /* idx=12: 00E00 - 00EFF */ { 0, 0, 0,45,46, 0, 0, 0, 0, 0, 0,47,48, 0, 0, 0 }, /* idx=13: 00F00 - 00FFF */ { 0,49, 0,50, 0, 0, 0,51,52, 0, 0, 0,53, 0, 0, 0 }, /* idx=14: 01000 - 010FF */ { 0, 0, 0,54, 0, 0, 0, 0,55, 0, 0, 0, 0, 0, 0, 0 }, /* idx=15: 01300 - 013FF */ { 0, 0, 0, 0, 0,56, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=16: 01700 - 017FF */ { 0,57, 0,58, 0, 0, 0, 0, 0, 0, 0, 0, 0,59, 0, 0 }, /* idx=17: 01800 - 018FF */ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,60, 0, 0, 0, 0, 0 }, /* idx=18: 01900 - 019FF */ { 0, 0, 0,61, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=19: 01A00 - 01AFF */ { 0,62, 0, 0, 0, 0,63,64, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=20: 01B00 - 01BFF */ { 0, 0, 0,65,66, 0,67,68, 0, 0,69, 0, 0, 0,70,71 }, /* idx=21: 01C00 - 01CFF */ { 0, 0, 0,72, 0, 0, 0, 0, 0, 0, 0, 0, 0,73,74, 0 }, /* idx=22: 01D00 - 01DFF */ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,75,76,77,78 }, /* idx=23: 02000 - 020FF */ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,79,80,81 }, /* idx=24: 02C00 - 02CFF */ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,82,83 }, /* idx=25: 02D00 - 02DFF */ { 0, 0, 0, 0, 0, 0, 0,84, 0, 0, 0, 0, 0, 0,85,86 }, /* idx=26: 03000 - 030FF */ { 0, 0,87, 0, 0, 0, 0, 0, 0,88, 0, 0, 0, 0, 0, 0 }, /* idx=27: 0A600 - 0A6FF */ { 0, 0, 0, 0, 0, 0,89,90, 0, 0, 0, 0, 0, 0, 0,91 }, /* idx=28: 0A800 - 0A8FF */ {92, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,93, 0,94,95 }, /* idx=29: 0A900 - 0A9FF */ { 0, 0,96, 0, 0,97, 0, 0, 0, 0, 0,98,99, 0, 0, 0 }, /* idx=30: 0AA00 - 0AAFF */ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,100,101, 0, 0, 0 }, /* idx=31: 0AB00 - 0ABFF */ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,102, 0 }, /* idx=32: 0FB00 - 0FBFF */ { 0,103, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=33: 0FE00 - 0FEFF */ { 0, 0,104, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=34: 10100 - 101FF */ { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,105 }, /* idx=35: 10A00 - 10AFF */ {106, 0, 0,107, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, /* idx=36: 11000 - 110FF */ { 0, 0, 0, 0,108, 0, 0, 0, 0, 0, 0,109, 0, 0, 0, 0 }, /* idx=37: 1D100 - 1D1FF */ { 0, 0, 0, 0, 0, 0,110,111,112, 0,113, 0, 0, 0, 0, 0 }, /* idx=38: 1D200 - 1D2FF */ { 0, 0, 0, 0,114, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, }; /* The index table to ccc_val_index[*][16] */ static const unsigned char ccc_index[] = { 0, 0, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9,10,11,12,13,14, 0, 0,15, 0, 0, 0,16, 17,18,19,20,21,22, 0, 0,23, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,24,25, 0, 0, 26, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,27, 0, 28,29,30,31, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,32, 0, 0,33, 0, 0,34, 0, 0, 0, 0, 0, 0, 0, 0,35, 0, 0, 0, 0, 0,36, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,37,38,}; struct unicode_decomposition_table { uint32_t nfc; uint32_t cp1; uint32_t cp2; }; static const struct unicode_decomposition_table u_decomposition_table[] = { { 0x000C0 , 0x00041 , 0x00300 }, { 0x000C1 , 0x00041 , 0x00301 }, { 0x000C2 , 0x00041 , 0x00302 }, { 0x000C3 , 0x00041 , 0x00303 }, { 0x000C4 , 0x00041 , 0x00308 }, { 0x000C5 , 0x00041 , 0x0030A }, { 0x000C7 , 0x00043 , 0x00327 }, { 0x000C8 , 0x00045 , 0x00300 }, { 0x000C9 , 0x00045 , 0x00301 }, { 0x000CA , 0x00045 , 0x00302 }, { 0x000CB , 0x00045 , 0x00308 }, { 0x000CC , 0x00049 , 0x00300 }, { 0x000CD , 0x00049 , 0x00301 }, { 0x000CE , 0x00049 , 0x00302 }, { 0x000CF , 0x00049 , 0x00308 }, { 0x000D1 , 0x0004E , 0x00303 }, { 0x000D2 , 0x0004F , 0x00300 }, { 0x000D3 , 0x0004F , 0x00301 }, { 0x000D4 , 0x0004F , 0x00302 }, { 0x000D5 , 0x0004F , 0x00303 }, { 0x000D6 , 0x0004F , 0x00308 }, { 0x000D9 , 0x00055 , 0x00300 }, { 0x000DA , 0x00055 , 0x00301 }, { 0x000DB , 0x00055 , 0x00302 }, { 0x000DC , 0x00055 , 0x00308 }, { 0x000DD , 0x00059 , 0x00301 }, { 0x000E0 , 0x00061 , 0x00300 }, { 0x000E1 , 0x00061 , 0x00301 }, { 0x000E2 , 0x00061 , 0x00302 }, { 0x000E3 , 0x00061 , 0x00303 }, { 0x000E4 , 0x00061 , 0x00308 }, { 0x000E5 , 0x00061 , 0x0030A }, { 0x000E7 , 0x00063 , 0x00327 }, { 0x000E8 , 0x00065 , 0x00300 }, { 0x000E9 , 0x00065 , 0x00301 }, { 0x000EA , 0x00065 , 0x00302 }, { 0x000EB , 0x00065 , 0x00308 }, { 0x000EC , 0x00069 , 0x00300 }, { 0x000ED , 0x00069 , 0x00301 }, { 0x000EE , 0x00069 , 0x00302 }, { 0x000EF , 0x00069 , 0x00308 }, { 0x000F1 , 0x0006E , 0x00303 }, { 0x000F2 , 0x0006F , 0x00300 }, { 0x000F3 , 0x0006F , 0x00301 }, { 0x000F4 , 0x0006F , 0x00302 }, { 0x000F5 , 0x0006F , 0x00303 }, { 0x000F6 , 0x0006F , 0x00308 }, { 0x000F9 , 0x00075 , 0x00300 }, { 0x000FA , 0x00075 , 0x00301 }, { 0x000FB , 0x00075 , 0x00302 }, { 0x000FC , 0x00075 , 0x00308 }, { 0x000FD , 0x00079 , 0x00301 }, { 0x000FF , 0x00079 , 0x00308 }, { 0x00100 , 0x00041 , 0x00304 }, { 0x00101 , 0x00061 , 0x00304 }, { 0x00102 , 0x00041 , 0x00306 }, { 0x00103 , 0x00061 , 0x00306 }, { 0x00104 , 0x00041 , 0x00328 }, { 0x00105 , 0x00061 , 0x00328 }, { 0x00106 , 0x00043 , 0x00301 }, { 0x00107 , 0x00063 , 0x00301 }, { 0x00108 , 0x00043 , 0x00302 }, { 0x00109 , 0x00063 , 0x00302 }, { 0x0010A , 0x00043 , 0x00307 }, { 0x0010B , 0x00063 , 0x00307 }, { 0x0010C , 0x00043 , 0x0030C }, { 0x0010D , 0x00063 , 0x0030C }, { 0x0010E , 0x00044 , 0x0030C }, { 0x0010F , 0x00064 , 0x0030C }, { 0x00112 , 0x00045 , 0x00304 }, { 0x00113 , 0x00065 , 0x00304 }, { 0x00114 , 0x00045 , 0x00306 }, { 0x00115 , 0x00065 , 0x00306 }, { 0x00116 , 0x00045 , 0x00307 }, { 0x00117 , 0x00065 , 0x00307 }, { 0x00118 , 0x00045 , 0x00328 }, { 0x00119 , 0x00065 , 0x00328 }, { 0x0011A , 0x00045 , 0x0030C }, { 0x0011B , 0x00065 , 0x0030C }, { 0x0011C , 0x00047 , 0x00302 }, { 0x0011D , 0x00067 , 0x00302 }, { 0x0011E , 0x00047 , 0x00306 }, { 0x0011F , 0x00067 , 0x00306 }, { 0x00120 , 0x00047 , 0x00307 }, { 0x00121 , 0x00067 , 0x00307 }, { 0x00122 , 0x00047 , 0x00327 }, { 0x00123 , 0x00067 , 0x00327 }, { 0x00124 , 0x00048 , 0x00302 }, { 0x00125 , 0x00068 , 0x00302 }, { 0x00128 , 0x00049 , 0x00303 }, { 0x00129 , 0x00069 , 0x00303 }, { 0x0012A , 0x00049 , 0x00304 }, { 0x0012B , 0x00069 , 0x00304 }, { 0x0012C , 0x00049 , 0x00306 }, { 0x0012D , 0x00069 , 0x00306 }, { 0x0012E , 0x00049 , 0x00328 }, { 0x0012F , 0x00069 , 0x00328 }, { 0x00130 , 0x00049 , 0x00307 }, { 0x00134 , 0x0004A , 0x00302 }, { 0x00135 , 0x0006A , 0x00302 }, { 0x00136 , 0x0004B , 0x00327 }, { 0x00137 , 0x0006B , 0x00327 }, { 0x00139 , 0x0004C , 0x00301 }, { 0x0013A , 0x0006C , 0x00301 }, { 0x0013B , 0x0004C , 0x00327 }, { 0x0013C , 0x0006C , 0x00327 }, { 0x0013D , 0x0004C , 0x0030C }, { 0x0013E , 0x0006C , 0x0030C }, { 0x00143 , 0x0004E , 0x00301 }, { 0x00144 , 0x0006E , 0x00301 }, { 0x00145 , 0x0004E , 0x00327 }, { 0x00146 , 0x0006E , 0x00327 }, { 0x00147 , 0x0004E , 0x0030C }, { 0x00148 , 0x0006E , 0x0030C }, { 0x0014C , 0x0004F , 0x00304 }, { 0x0014D , 0x0006F , 0x00304 }, { 0x0014E , 0x0004F , 0x00306 }, { 0x0014F , 0x0006F , 0x00306 }, { 0x00150 , 0x0004F , 0x0030B }, { 0x00151 , 0x0006F , 0x0030B }, { 0x00154 , 0x00052 , 0x00301 }, { 0x00155 , 0x00072 , 0x00301 }, { 0x00156 , 0x00052 , 0x00327 }, { 0x00157 , 0x00072 , 0x00327 }, { 0x00158 , 0x00052 , 0x0030C }, { 0x00159 , 0x00072 , 0x0030C }, { 0x0015A , 0x00053 , 0x00301 }, { 0x0015B , 0x00073 , 0x00301 }, { 0x0015C , 0x00053 , 0x00302 }, { 0x0015D , 0x00073 , 0x00302 }, { 0x0015E , 0x00053 , 0x00327 }, { 0x0015F , 0x00073 , 0x00327 }, { 0x00160 , 0x00053 , 0x0030C }, { 0x00161 , 0x00073 , 0x0030C }, { 0x00162 , 0x00054 , 0x00327 }, { 0x00163 , 0x00074 , 0x00327 }, { 0x00164 , 0x00054 , 0x0030C }, { 0x00165 , 0x00074 , 0x0030C }, { 0x00168 , 0x00055 , 0x00303 }, { 0x00169 , 0x00075 , 0x00303 }, { 0x0016A , 0x00055 , 0x00304 }, { 0x0016B , 0x00075 , 0x00304 }, { 0x0016C , 0x00055 , 0x00306 }, { 0x0016D , 0x00075 , 0x00306 }, { 0x0016E , 0x00055 , 0x0030A }, { 0x0016F , 0x00075 , 0x0030A }, { 0x00170 , 0x00055 , 0x0030B }, { 0x00171 , 0x00075 , 0x0030B }, { 0x00172 , 0x00055 , 0x00328 }, { 0x00173 , 0x00075 , 0x00328 }, { 0x00174 , 0x00057 , 0x00302 }, { 0x00175 , 0x00077 , 0x00302 }, { 0x00176 , 0x00059 , 0x00302 }, { 0x00177 , 0x00079 , 0x00302 }, { 0x00178 , 0x00059 , 0x00308 }, { 0x00179 , 0x0005A , 0x00301 }, { 0x0017A , 0x0007A , 0x00301 }, { 0x0017B , 0x0005A , 0x00307 }, { 0x0017C , 0x0007A , 0x00307 }, { 0x0017D , 0x0005A , 0x0030C }, { 0x0017E , 0x0007A , 0x0030C }, { 0x001A0 , 0x0004F , 0x0031B }, { 0x001A1 , 0x0006F , 0x0031B }, { 0x001AF , 0x00055 , 0x0031B }, { 0x001B0 , 0x00075 , 0x0031B }, { 0x001CD , 0x00041 , 0x0030C }, { 0x001CE , 0x00061 , 0x0030C }, { 0x001CF , 0x00049 , 0x0030C }, { 0x001D0 , 0x00069 , 0x0030C }, { 0x001D1 , 0x0004F , 0x0030C }, { 0x001D2 , 0x0006F , 0x0030C }, { 0x001D3 , 0x00055 , 0x0030C }, { 0x001D4 , 0x00075 , 0x0030C }, { 0x001D5 , 0x000DC , 0x00304 }, { 0x001D6 , 0x000FC , 0x00304 }, { 0x001D7 , 0x000DC , 0x00301 }, { 0x001D8 , 0x000FC , 0x00301 }, { 0x001D9 , 0x000DC , 0x0030C }, { 0x001DA , 0x000FC , 0x0030C }, { 0x001DB , 0x000DC , 0x00300 }, { 0x001DC , 0x000FC , 0x00300 }, { 0x001DE , 0x000C4 , 0x00304 }, { 0x001DF , 0x000E4 , 0x00304 }, { 0x001E0 , 0x00226 , 0x00304 }, { 0x001E1 , 0x00227 , 0x00304 }, { 0x001E2 , 0x000C6 , 0x00304 }, { 0x001E3 , 0x000E6 , 0x00304 }, { 0x001E6 , 0x00047 , 0x0030C }, { 0x001E7 , 0x00067 , 0x0030C }, { 0x001E8 , 0x0004B , 0x0030C }, { 0x001E9 , 0x0006B , 0x0030C }, { 0x001EA , 0x0004F , 0x00328 }, { 0x001EB , 0x0006F , 0x00328 }, { 0x001EC , 0x001EA , 0x00304 }, { 0x001ED , 0x001EB , 0x00304 }, { 0x001EE , 0x001B7 , 0x0030C }, { 0x001EF , 0x00292 , 0x0030C }, { 0x001F0 , 0x0006A , 0x0030C }, { 0x001F4 , 0x00047 , 0x00301 }, { 0x001F5 , 0x00067 , 0x00301 }, { 0x001F8 , 0x0004E , 0x00300 }, { 0x001F9 , 0x0006E , 0x00300 }, { 0x001FA , 0x000C5 , 0x00301 }, { 0x001FB , 0x000E5 , 0x00301 }, { 0x001FC , 0x000C6 , 0x00301 }, { 0x001FD , 0x000E6 , 0x00301 }, { 0x001FE , 0x000D8 , 0x00301 }, { 0x001FF , 0x000F8 , 0x00301 }, { 0x00200 , 0x00041 , 0x0030F }, { 0x00201 , 0x00061 , 0x0030F }, { 0x00202 , 0x00041 , 0x00311 }, { 0x00203 , 0x00061 , 0x00311 }, { 0x00204 , 0x00045 , 0x0030F }, { 0x00205 , 0x00065 , 0x0030F }, { 0x00206 , 0x00045 , 0x00311 }, { 0x00207 , 0x00065 , 0x00311 }, { 0x00208 , 0x00049 , 0x0030F }, { 0x00209 , 0x00069 , 0x0030F }, { 0x0020A , 0x00049 , 0x00311 }, { 0x0020B , 0x00069 , 0x00311 }, { 0x0020C , 0x0004F , 0x0030F }, { 0x0020D , 0x0006F , 0x0030F }, { 0x0020E , 0x0004F , 0x00311 }, { 0x0020F , 0x0006F , 0x00311 }, { 0x00210 , 0x00052 , 0x0030F }, { 0x00211 , 0x00072 , 0x0030F }, { 0x00212 , 0x00052 , 0x00311 }, { 0x00213 , 0x00072 , 0x00311 }, { 0x00214 , 0x00055 , 0x0030F }, { 0x00215 , 0x00075 , 0x0030F }, { 0x00216 , 0x00055 , 0x00311 }, { 0x00217 , 0x00075 , 0x00311 }, { 0x00218 , 0x00053 , 0x00326 }, { 0x00219 , 0x00073 , 0x00326 }, { 0x0021A , 0x00054 , 0x00326 }, { 0x0021B , 0x00074 , 0x00326 }, { 0x0021E , 0x00048 , 0x0030C }, { 0x0021F , 0x00068 , 0x0030C }, { 0x00226 , 0x00041 , 0x00307 }, { 0x00227 , 0x00061 , 0x00307 }, { 0x00228 , 0x00045 , 0x00327 }, { 0x00229 , 0x00065 , 0x00327 }, { 0x0022A , 0x000D6 , 0x00304 }, { 0x0022B , 0x000F6 , 0x00304 }, { 0x0022C , 0x000D5 , 0x00304 }, { 0x0022D , 0x000F5 , 0x00304 }, { 0x0022E , 0x0004F , 0x00307 }, { 0x0022F , 0x0006F , 0x00307 }, { 0x00230 , 0x0022E , 0x00304 }, { 0x00231 , 0x0022F , 0x00304 }, { 0x00232 , 0x00059 , 0x00304 }, { 0x00233 , 0x00079 , 0x00304 }, { 0x00385 , 0x000A8 , 0x00301 }, { 0x00386 , 0x00391 , 0x00301 }, { 0x00388 , 0x00395 , 0x00301 }, { 0x00389 , 0x00397 , 0x00301 }, { 0x0038A , 0x00399 , 0x00301 }, { 0x0038C , 0x0039F , 0x00301 }, { 0x0038E , 0x003A5 , 0x00301 }, { 0x0038F , 0x003A9 , 0x00301 }, { 0x00390 , 0x003CA , 0x00301 }, { 0x003AA , 0x00399 , 0x00308 }, { 0x003AB , 0x003A5 , 0x00308 }, { 0x003AC , 0x003B1 , 0x00301 }, { 0x003AD , 0x003B5 , 0x00301 }, { 0x003AE , 0x003B7 , 0x00301 }, { 0x003AF , 0x003B9 , 0x00301 }, { 0x003B0 , 0x003CB , 0x00301 }, { 0x003CA , 0x003B9 , 0x00308 }, { 0x003CB , 0x003C5 , 0x00308 }, { 0x003CC , 0x003BF , 0x00301 }, { 0x003CD , 0x003C5 , 0x00301 }, { 0x003CE , 0x003C9 , 0x00301 }, { 0x003D3 , 0x003D2 , 0x00301 }, { 0x003D4 , 0x003D2 , 0x00308 }, { 0x00400 , 0x00415 , 0x00300 }, { 0x00401 , 0x00415 , 0x00308 }, { 0x00403 , 0x00413 , 0x00301 }, { 0x00407 , 0x00406 , 0x00308 }, { 0x0040C , 0x0041A , 0x00301 }, { 0x0040D , 0x00418 , 0x00300 }, { 0x0040E , 0x00423 , 0x00306 }, { 0x00419 , 0x00418 , 0x00306 }, { 0x00439 , 0x00438 , 0x00306 }, { 0x00450 , 0x00435 , 0x00300 }, { 0x00451 , 0x00435 , 0x00308 }, { 0x00453 , 0x00433 , 0x00301 }, { 0x00457 , 0x00456 , 0x00308 }, { 0x0045C , 0x0043A , 0x00301 }, { 0x0045D , 0x00438 , 0x00300 }, { 0x0045E , 0x00443 , 0x00306 }, { 0x00476 , 0x00474 , 0x0030F }, { 0x00477 , 0x00475 , 0x0030F }, { 0x004C1 , 0x00416 , 0x00306 }, { 0x004C2 , 0x00436 , 0x00306 }, { 0x004D0 , 0x00410 , 0x00306 }, { 0x004D1 , 0x00430 , 0x00306 }, { 0x004D2 , 0x00410 , 0x00308 }, { 0x004D3 , 0x00430 , 0x00308 }, { 0x004D6 , 0x00415 , 0x00306 }, { 0x004D7 , 0x00435 , 0x00306 }, { 0x004DA , 0x004D8 , 0x00308 }, { 0x004DB , 0x004D9 , 0x00308 }, { 0x004DC , 0x00416 , 0x00308 }, { 0x004DD , 0x00436 , 0x00308 }, { 0x004DE , 0x00417 , 0x00308 }, { 0x004DF , 0x00437 , 0x00308 }, { 0x004E2 , 0x00418 , 0x00304 }, { 0x004E3 , 0x00438 , 0x00304 }, { 0x004E4 , 0x00418 , 0x00308 }, { 0x004E5 , 0x00438 , 0x00308 }, { 0x004E6 , 0x0041E , 0x00308 }, { 0x004E7 , 0x0043E , 0x00308 }, { 0x004EA , 0x004E8 , 0x00308 }, { 0x004EB , 0x004E9 , 0x00308 }, { 0x004EC , 0x0042D , 0x00308 }, { 0x004ED , 0x0044D , 0x00308 }, { 0x004EE , 0x00423 , 0x00304 }, { 0x004EF , 0x00443 , 0x00304 }, { 0x004F0 , 0x00423 , 0x00308 }, { 0x004F1 , 0x00443 , 0x00308 }, { 0x004F2 , 0x00423 , 0x0030B }, { 0x004F3 , 0x00443 , 0x0030B }, { 0x004F4 , 0x00427 , 0x00308 }, { 0x004F5 , 0x00447 , 0x00308 }, { 0x004F8 , 0x0042B , 0x00308 }, { 0x004F9 , 0x0044B , 0x00308 }, { 0x00622 , 0x00627 , 0x00653 }, { 0x00623 , 0x00627 , 0x00654 }, { 0x00624 , 0x00648 , 0x00654 }, { 0x00625 , 0x00627 , 0x00655 }, { 0x00626 , 0x0064A , 0x00654 }, { 0x006C0 , 0x006D5 , 0x00654 }, { 0x006C2 , 0x006C1 , 0x00654 }, { 0x006D3 , 0x006D2 , 0x00654 }, { 0x00929 , 0x00928 , 0x0093C }, { 0x00931 , 0x00930 , 0x0093C }, { 0x00934 , 0x00933 , 0x0093C }, { 0x009CB , 0x009C7 , 0x009BE }, { 0x009CC , 0x009C7 , 0x009D7 }, { 0x00B48 , 0x00B47 , 0x00B56 }, { 0x00B4B , 0x00B47 , 0x00B3E }, { 0x00B4C , 0x00B47 , 0x00B57 }, { 0x00B94 , 0x00B92 , 0x00BD7 }, { 0x00BCA , 0x00BC6 , 0x00BBE }, { 0x00BCB , 0x00BC7 , 0x00BBE }, { 0x00BCC , 0x00BC6 , 0x00BD7 }, { 0x00C48 , 0x00C46 , 0x00C56 }, { 0x00CC0 , 0x00CBF , 0x00CD5 }, { 0x00CC7 , 0x00CC6 , 0x00CD5 }, { 0x00CC8 , 0x00CC6 , 0x00CD6 }, { 0x00CCA , 0x00CC6 , 0x00CC2 }, { 0x00CCB , 0x00CCA , 0x00CD5 }, { 0x00D4A , 0x00D46 , 0x00D3E }, { 0x00D4B , 0x00D47 , 0x00D3E }, { 0x00D4C , 0x00D46 , 0x00D57 }, { 0x00DDA , 0x00DD9 , 0x00DCA }, { 0x00DDC , 0x00DD9 , 0x00DCF }, { 0x00DDD , 0x00DDC , 0x00DCA }, { 0x00DDE , 0x00DD9 , 0x00DDF }, { 0x01026 , 0x01025 , 0x0102E }, { 0x01B06 , 0x01B05 , 0x01B35 }, { 0x01B08 , 0x01B07 , 0x01B35 }, { 0x01B0A , 0x01B09 , 0x01B35 }, { 0x01B0C , 0x01B0B , 0x01B35 }, { 0x01B0E , 0x01B0D , 0x01B35 }, { 0x01B12 , 0x01B11 , 0x01B35 }, { 0x01B3B , 0x01B3A , 0x01B35 }, { 0x01B3D , 0x01B3C , 0x01B35 }, { 0x01B40 , 0x01B3E , 0x01B35 }, { 0x01B41 , 0x01B3F , 0x01B35 }, { 0x01B43 , 0x01B42 , 0x01B35 }, { 0x01E00 , 0x00041 , 0x00325 }, { 0x01E01 , 0x00061 , 0x00325 }, { 0x01E02 , 0x00042 , 0x00307 }, { 0x01E03 , 0x00062 , 0x00307 }, { 0x01E04 , 0x00042 , 0x00323 }, { 0x01E05 , 0x00062 , 0x00323 }, { 0x01E06 , 0x00042 , 0x00331 }, { 0x01E07 , 0x00062 , 0x00331 }, { 0x01E08 , 0x000C7 , 0x00301 }, { 0x01E09 , 0x000E7 , 0x00301 }, { 0x01E0A , 0x00044 , 0x00307 }, { 0x01E0B , 0x00064 , 0x00307 }, { 0x01E0C , 0x00044 , 0x00323 }, { 0x01E0D , 0x00064 , 0x00323 }, { 0x01E0E , 0x00044 , 0x00331 }, { 0x01E0F , 0x00064 , 0x00331 }, { 0x01E10 , 0x00044 , 0x00327 }, { 0x01E11 , 0x00064 , 0x00327 }, { 0x01E12 , 0x00044 , 0x0032D }, { 0x01E13 , 0x00064 , 0x0032D }, { 0x01E14 , 0x00112 , 0x00300 }, { 0x01E15 , 0x00113 , 0x00300 }, { 0x01E16 , 0x00112 , 0x00301 }, { 0x01E17 , 0x00113 , 0x00301 }, { 0x01E18 , 0x00045 , 0x0032D }, { 0x01E19 , 0x00065 , 0x0032D }, { 0x01E1A , 0x00045 , 0x00330 }, { 0x01E1B , 0x00065 , 0x00330 }, { 0x01E1C , 0x00228 , 0x00306 }, { 0x01E1D , 0x00229 , 0x00306 }, { 0x01E1E , 0x00046 , 0x00307 }, { 0x01E1F , 0x00066 , 0x00307 }, { 0x01E20 , 0x00047 , 0x00304 }, { 0x01E21 , 0x00067 , 0x00304 }, { 0x01E22 , 0x00048 , 0x00307 }, { 0x01E23 , 0x00068 , 0x00307 }, { 0x01E24 , 0x00048 , 0x00323 }, { 0x01E25 , 0x00068 , 0x00323 }, { 0x01E26 , 0x00048 , 0x00308 }, { 0x01E27 , 0x00068 , 0x00308 }, { 0x01E28 , 0x00048 , 0x00327 }, { 0x01E29 , 0x00068 , 0x00327 }, { 0x01E2A , 0x00048 , 0x0032E }, { 0x01E2B , 0x00068 , 0x0032E }, { 0x01E2C , 0x00049 , 0x00330 }, { 0x01E2D , 0x00069 , 0x00330 }, { 0x01E2E , 0x000CF , 0x00301 }, { 0x01E2F , 0x000EF , 0x00301 }, { 0x01E30 , 0x0004B , 0x00301 }, { 0x01E31 , 0x0006B , 0x00301 }, { 0x01E32 , 0x0004B , 0x00323 }, { 0x01E33 , 0x0006B , 0x00323 }, { 0x01E34 , 0x0004B , 0x00331 }, { 0x01E35 , 0x0006B , 0x00331 }, { 0x01E36 , 0x0004C , 0x00323 }, { 0x01E37 , 0x0006C , 0x00323 }, { 0x01E38 , 0x01E36 , 0x00304 }, { 0x01E39 , 0x01E37 , 0x00304 }, { 0x01E3A , 0x0004C , 0x00331 }, { 0x01E3B , 0x0006C , 0x00331 }, { 0x01E3C , 0x0004C , 0x0032D }, { 0x01E3D , 0x0006C , 0x0032D }, { 0x01E3E , 0x0004D , 0x00301 }, { 0x01E3F , 0x0006D , 0x00301 }, { 0x01E40 , 0x0004D , 0x00307 }, { 0x01E41 , 0x0006D , 0x00307 }, { 0x01E42 , 0x0004D , 0x00323 }, { 0x01E43 , 0x0006D , 0x00323 }, { 0x01E44 , 0x0004E , 0x00307 }, { 0x01E45 , 0x0006E , 0x00307 }, { 0x01E46 , 0x0004E , 0x00323 }, { 0x01E47 , 0x0006E , 0x00323 }, { 0x01E48 , 0x0004E , 0x00331 }, { 0x01E49 , 0x0006E , 0x00331 }, { 0x01E4A , 0x0004E , 0x0032D }, { 0x01E4B , 0x0006E , 0x0032D }, { 0x01E4C , 0x000D5 , 0x00301 }, { 0x01E4D , 0x000F5 , 0x00301 }, { 0x01E4E , 0x000D5 , 0x00308 }, { 0x01E4F , 0x000F5 , 0x00308 }, { 0x01E50 , 0x0014C , 0x00300 }, { 0x01E51 , 0x0014D , 0x00300 }, { 0x01E52 , 0x0014C , 0x00301 }, { 0x01E53 , 0x0014D , 0x00301 }, { 0x01E54 , 0x00050 , 0x00301 }, { 0x01E55 , 0x00070 , 0x00301 }, { 0x01E56 , 0x00050 , 0x00307 }, { 0x01E57 , 0x00070 , 0x00307 }, { 0x01E58 , 0x00052 , 0x00307 }, { 0x01E59 , 0x00072 , 0x00307 }, { 0x01E5A , 0x00052 , 0x00323 }, { 0x01E5B , 0x00072 , 0x00323 }, { 0x01E5C , 0x01E5A , 0x00304 }, { 0x01E5D , 0x01E5B , 0x00304 }, { 0x01E5E , 0x00052 , 0x00331 }, { 0x01E5F , 0x00072 , 0x00331 }, { 0x01E60 , 0x00053 , 0x00307 }, { 0x01E61 , 0x00073 , 0x00307 }, { 0x01E62 , 0x00053 , 0x00323 }, { 0x01E63 , 0x00073 , 0x00323 }, { 0x01E64 , 0x0015A , 0x00307 }, { 0x01E65 , 0x0015B , 0x00307 }, { 0x01E66 , 0x00160 , 0x00307 }, { 0x01E67 , 0x00161 , 0x00307 }, { 0x01E68 , 0x01E62 , 0x00307 }, { 0x01E69 , 0x01E63 , 0x00307 }, { 0x01E6A , 0x00054 , 0x00307 }, { 0x01E6B , 0x00074 , 0x00307 }, { 0x01E6C , 0x00054 , 0x00323 }, { 0x01E6D , 0x00074 , 0x00323 }, { 0x01E6E , 0x00054 , 0x00331 }, { 0x01E6F , 0x00074 , 0x00331 }, { 0x01E70 , 0x00054 , 0x0032D }, { 0x01E71 , 0x00074 , 0x0032D }, { 0x01E72 , 0x00055 , 0x00324 }, { 0x01E73 , 0x00075 , 0x00324 }, { 0x01E74 , 0x00055 , 0x00330 }, { 0x01E75 , 0x00075 , 0x00330 }, { 0x01E76 , 0x00055 , 0x0032D }, { 0x01E77 , 0x00075 , 0x0032D }, { 0x01E78 , 0x00168 , 0x00301 }, { 0x01E79 , 0x00169 , 0x00301 }, { 0x01E7A , 0x0016A , 0x00308 }, { 0x01E7B , 0x0016B , 0x00308 }, { 0x01E7C , 0x00056 , 0x00303 }, { 0x01E7D , 0x00076 , 0x00303 }, { 0x01E7E , 0x00056 , 0x00323 }, { 0x01E7F , 0x00076 , 0x00323 }, { 0x01E80 , 0x00057 , 0x00300 }, { 0x01E81 , 0x00077 , 0x00300 }, { 0x01E82 , 0x00057 , 0x00301 }, { 0x01E83 , 0x00077 , 0x00301 }, { 0x01E84 , 0x00057 , 0x00308 }, { 0x01E85 , 0x00077 , 0x00308 }, { 0x01E86 , 0x00057 , 0x00307 }, { 0x01E87 , 0x00077 , 0x00307 }, { 0x01E88 , 0x00057 , 0x00323 }, { 0x01E89 , 0x00077 , 0x00323 }, { 0x01E8A , 0x00058 , 0x00307 }, { 0x01E8B , 0x00078 , 0x00307 }, { 0x01E8C , 0x00058 , 0x00308 }, { 0x01E8D , 0x00078 , 0x00308 }, { 0x01E8E , 0x00059 , 0x00307 }, { 0x01E8F , 0x00079 , 0x00307 }, { 0x01E90 , 0x0005A , 0x00302 }, { 0x01E91 , 0x0007A , 0x00302 }, { 0x01E92 , 0x0005A , 0x00323 }, { 0x01E93 , 0x0007A , 0x00323 }, { 0x01E94 , 0x0005A , 0x00331 }, { 0x01E95 , 0x0007A , 0x00331 }, { 0x01E96 , 0x00068 , 0x00331 }, { 0x01E97 , 0x00074 , 0x00308 }, { 0x01E98 , 0x00077 , 0x0030A }, { 0x01E99 , 0x00079 , 0x0030A }, { 0x01E9B , 0x0017F , 0x00307 }, { 0x01EA0 , 0x00041 , 0x00323 }, { 0x01EA1 , 0x00061 , 0x00323 }, { 0x01EA2 , 0x00041 , 0x00309 }, { 0x01EA3 , 0x00061 , 0x00309 }, { 0x01EA4 , 0x000C2 , 0x00301 }, { 0x01EA5 , 0x000E2 , 0x00301 }, { 0x01EA6 , 0x000C2 , 0x00300 }, { 0x01EA7 , 0x000E2 , 0x00300 }, { 0x01EA8 , 0x000C2 , 0x00309 }, { 0x01EA9 , 0x000E2 , 0x00309 }, { 0x01EAA , 0x000C2 , 0x00303 }, { 0x01EAB , 0x000E2 , 0x00303 }, { 0x01EAC , 0x01EA0 , 0x00302 }, { 0x01EAD , 0x01EA1 , 0x00302 }, { 0x01EAE , 0x00102 , 0x00301 }, { 0x01EAF , 0x00103 , 0x00301 }, { 0x01EB0 , 0x00102 , 0x00300 }, { 0x01EB1 , 0x00103 , 0x00300 }, { 0x01EB2 , 0x00102 , 0x00309 }, { 0x01EB3 , 0x00103 , 0x00309 }, { 0x01EB4 , 0x00102 , 0x00303 }, { 0x01EB5 , 0x00103 , 0x00303 }, { 0x01EB6 , 0x01EA0 , 0x00306 }, { 0x01EB7 , 0x01EA1 , 0x00306 }, { 0x01EB8 , 0x00045 , 0x00323 }, { 0x01EB9 , 0x00065 , 0x00323 }, { 0x01EBA , 0x00045 , 0x00309 }, { 0x01EBB , 0x00065 , 0x00309 }, { 0x01EBC , 0x00045 , 0x00303 }, { 0x01EBD , 0x00065 , 0x00303 }, { 0x01EBE , 0x000CA , 0x00301 }, { 0x01EBF , 0x000EA , 0x00301 }, { 0x01EC0 , 0x000CA , 0x00300 }, { 0x01EC1 , 0x000EA , 0x00300 }, { 0x01EC2 , 0x000CA , 0x00309 }, { 0x01EC3 , 0x000EA , 0x00309 }, { 0x01EC4 , 0x000CA , 0x00303 }, { 0x01EC5 , 0x000EA , 0x00303 }, { 0x01EC6 , 0x01EB8 , 0x00302 }, { 0x01EC7 , 0x01EB9 , 0x00302 }, { 0x01EC8 , 0x00049 , 0x00309 }, { 0x01EC9 , 0x00069 , 0x00309 }, { 0x01ECA , 0x00049 , 0x00323 }, { 0x01ECB , 0x00069 , 0x00323 }, { 0x01ECC , 0x0004F , 0x00323 }, { 0x01ECD , 0x0006F , 0x00323 }, { 0x01ECE , 0x0004F , 0x00309 }, { 0x01ECF , 0x0006F , 0x00309 }, { 0x01ED0 , 0x000D4 , 0x00301 }, { 0x01ED1 , 0x000F4 , 0x00301 }, { 0x01ED2 , 0x000D4 , 0x00300 }, { 0x01ED3 , 0x000F4 , 0x00300 }, { 0x01ED4 , 0x000D4 , 0x00309 }, { 0x01ED5 , 0x000F4 , 0x00309 }, { 0x01ED6 , 0x000D4 , 0x00303 }, { 0x01ED7 , 0x000F4 , 0x00303 }, { 0x01ED8 , 0x01ECC , 0x00302 }, { 0x01ED9 , 0x01ECD , 0x00302 }, { 0x01EDA , 0x001A0 , 0x00301 }, { 0x01EDB , 0x001A1 , 0x00301 }, { 0x01EDC , 0x001A0 , 0x00300 }, { 0x01EDD , 0x001A1 , 0x00300 }, { 0x01EDE , 0x001A0 , 0x00309 }, { 0x01EDF , 0x001A1 , 0x00309 }, { 0x01EE0 , 0x001A0 , 0x00303 }, { 0x01EE1 , 0x001A1 , 0x00303 }, { 0x01EE2 , 0x001A0 , 0x00323 }, { 0x01EE3 , 0x001A1 , 0x00323 }, { 0x01EE4 , 0x00055 , 0x00323 }, { 0x01EE5 , 0x00075 , 0x00323 }, { 0x01EE6 , 0x00055 , 0x00309 }, { 0x01EE7 , 0x00075 , 0x00309 }, { 0x01EE8 , 0x001AF , 0x00301 }, { 0x01EE9 , 0x001B0 , 0x00301 }, { 0x01EEA , 0x001AF , 0x00300 }, { 0x01EEB , 0x001B0 , 0x00300 }, { 0x01EEC , 0x001AF , 0x00309 }, { 0x01EED , 0x001B0 , 0x00309 }, { 0x01EEE , 0x001AF , 0x00303 }, { 0x01EEF , 0x001B0 , 0x00303 }, { 0x01EF0 , 0x001AF , 0x00323 }, { 0x01EF1 , 0x001B0 , 0x00323 }, { 0x01EF2 , 0x00059 , 0x00300 }, { 0x01EF3 , 0x00079 , 0x00300 }, { 0x01EF4 , 0x00059 , 0x00323 }, { 0x01EF5 , 0x00079 , 0x00323 }, { 0x01EF6 , 0x00059 , 0x00309 }, { 0x01EF7 , 0x00079 , 0x00309 }, { 0x01EF8 , 0x00059 , 0x00303 }, { 0x01EF9 , 0x00079 , 0x00303 }, { 0x01F00 , 0x003B1 , 0x00313 }, { 0x01F01 , 0x003B1 , 0x00314 }, { 0x01F02 , 0x01F00 , 0x00300 }, { 0x01F03 , 0x01F01 , 0x00300 }, { 0x01F04 , 0x01F00 , 0x00301 }, { 0x01F05 , 0x01F01 , 0x00301 }, { 0x01F06 , 0x01F00 , 0x00342 }, { 0x01F07 , 0x01F01 , 0x00342 }, { 0x01F08 , 0x00391 , 0x00313 }, { 0x01F09 , 0x00391 , 0x00314 }, { 0x01F0A , 0x01F08 , 0x00300 }, { 0x01F0B , 0x01F09 , 0x00300 }, { 0x01F0C , 0x01F08 , 0x00301 }, { 0x01F0D , 0x01F09 , 0x00301 }, { 0x01F0E , 0x01F08 , 0x00342 }, { 0x01F0F , 0x01F09 , 0x00342 }, { 0x01F10 , 0x003B5 , 0x00313 }, { 0x01F11 , 0x003B5 , 0x00314 }, { 0x01F12 , 0x01F10 , 0x00300 }, { 0x01F13 , 0x01F11 , 0x00300 }, { 0x01F14 , 0x01F10 , 0x00301 }, { 0x01F15 , 0x01F11 , 0x00301 }, { 0x01F18 , 0x00395 , 0x00313 }, { 0x01F19 , 0x00395 , 0x00314 }, { 0x01F1A , 0x01F18 , 0x00300 }, { 0x01F1B , 0x01F19 , 0x00300 }, { 0x01F1C , 0x01F18 , 0x00301 }, { 0x01F1D , 0x01F19 , 0x00301 }, { 0x01F20 , 0x003B7 , 0x00313 }, { 0x01F21 , 0x003B7 , 0x00314 }, { 0x01F22 , 0x01F20 , 0x00300 }, { 0x01F23 , 0x01F21 , 0x00300 }, { 0x01F24 , 0x01F20 , 0x00301 }, { 0x01F25 , 0x01F21 , 0x00301 }, { 0x01F26 , 0x01F20 , 0x00342 }, { 0x01F27 , 0x01F21 , 0x00342 }, { 0x01F28 , 0x00397 , 0x00313 }, { 0x01F29 , 0x00397 , 0x00314 }, { 0x01F2A , 0x01F28 , 0x00300 }, { 0x01F2B , 0x01F29 , 0x00300 }, { 0x01F2C , 0x01F28 , 0x00301 }, { 0x01F2D , 0x01F29 , 0x00301 }, { 0x01F2E , 0x01F28 , 0x00342 }, { 0x01F2F , 0x01F29 , 0x00342 }, { 0x01F30 , 0x003B9 , 0x00313 }, { 0x01F31 , 0x003B9 , 0x00314 }, { 0x01F32 , 0x01F30 , 0x00300 }, { 0x01F33 , 0x01F31 , 0x00300 }, { 0x01F34 , 0x01F30 , 0x00301 }, { 0x01F35 , 0x01F31 , 0x00301 }, { 0x01F36 , 0x01F30 , 0x00342 }, { 0x01F37 , 0x01F31 , 0x00342 }, { 0x01F38 , 0x00399 , 0x00313 }, { 0x01F39 , 0x00399 , 0x00314 }, { 0x01F3A , 0x01F38 , 0x00300 }, { 0x01F3B , 0x01F39 , 0x00300 }, { 0x01F3C , 0x01F38 , 0x00301 }, { 0x01F3D , 0x01F39 , 0x00301 }, { 0x01F3E , 0x01F38 , 0x00342 }, { 0x01F3F , 0x01F39 , 0x00342 }, { 0x01F40 , 0x003BF , 0x00313 }, { 0x01F41 , 0x003BF , 0x00314 }, { 0x01F42 , 0x01F40 , 0x00300 }, { 0x01F43 , 0x01F41 , 0x00300 }, { 0x01F44 , 0x01F40 , 0x00301 }, { 0x01F45 , 0x01F41 , 0x00301 }, { 0x01F48 , 0x0039F , 0x00313 }, { 0x01F49 , 0x0039F , 0x00314 }, { 0x01F4A , 0x01F48 , 0x00300 }, { 0x01F4B , 0x01F49 , 0x00300 }, { 0x01F4C , 0x01F48 , 0x00301 }, { 0x01F4D , 0x01F49 , 0x00301 }, { 0x01F50 , 0x003C5 , 0x00313 }, { 0x01F51 , 0x003C5 , 0x00314 }, { 0x01F52 , 0x01F50 , 0x00300 }, { 0x01F53 , 0x01F51 , 0x00300 }, { 0x01F54 , 0x01F50 , 0x00301 }, { 0x01F55 , 0x01F51 , 0x00301 }, { 0x01F56 , 0x01F50 , 0x00342 }, { 0x01F57 , 0x01F51 , 0x00342 }, { 0x01F59 , 0x003A5 , 0x00314 }, { 0x01F5B , 0x01F59 , 0x00300 }, { 0x01F5D , 0x01F59 , 0x00301 }, { 0x01F5F , 0x01F59 , 0x00342 }, { 0x01F60 , 0x003C9 , 0x00313 }, { 0x01F61 , 0x003C9 , 0x00314 }, { 0x01F62 , 0x01F60 , 0x00300 }, { 0x01F63 , 0x01F61 , 0x00300 }, { 0x01F64 , 0x01F60 , 0x00301 }, { 0x01F65 , 0x01F61 , 0x00301 }, { 0x01F66 , 0x01F60 , 0x00342 }, { 0x01F67 , 0x01F61 , 0x00342 }, { 0x01F68 , 0x003A9 , 0x00313 }, { 0x01F69 , 0x003A9 , 0x00314 }, { 0x01F6A , 0x01F68 , 0x00300 }, { 0x01F6B , 0x01F69 , 0x00300 }, { 0x01F6C , 0x01F68 , 0x00301 }, { 0x01F6D , 0x01F69 , 0x00301 }, { 0x01F6E , 0x01F68 , 0x00342 }, { 0x01F6F , 0x01F69 , 0x00342 }, { 0x01F70 , 0x003B1 , 0x00300 }, { 0x01F72 , 0x003B5 , 0x00300 }, { 0x01F74 , 0x003B7 , 0x00300 }, { 0x01F76 , 0x003B9 , 0x00300 }, { 0x01F78 , 0x003BF , 0x00300 }, { 0x01F7A , 0x003C5 , 0x00300 }, { 0x01F7C , 0x003C9 , 0x00300 }, { 0x01F80 , 0x01F00 , 0x00345 }, { 0x01F81 , 0x01F01 , 0x00345 }, { 0x01F82 , 0x01F02 , 0x00345 }, { 0x01F83 , 0x01F03 , 0x00345 }, { 0x01F84 , 0x01F04 , 0x00345 }, { 0x01F85 , 0x01F05 , 0x00345 }, { 0x01F86 , 0x01F06 , 0x00345 }, { 0x01F87 , 0x01F07 , 0x00345 }, { 0x01F88 , 0x01F08 , 0x00345 }, { 0x01F89 , 0x01F09 , 0x00345 }, { 0x01F8A , 0x01F0A , 0x00345 }, { 0x01F8B , 0x01F0B , 0x00345 }, { 0x01F8C , 0x01F0C , 0x00345 }, { 0x01F8D , 0x01F0D , 0x00345 }, { 0x01F8E , 0x01F0E , 0x00345 }, { 0x01F8F , 0x01F0F , 0x00345 }, { 0x01F90 , 0x01F20 , 0x00345 }, { 0x01F91 , 0x01F21 , 0x00345 }, { 0x01F92 , 0x01F22 , 0x00345 }, { 0x01F93 , 0x01F23 , 0x00345 }, { 0x01F94 , 0x01F24 , 0x00345 }, { 0x01F95 , 0x01F25 , 0x00345 }, { 0x01F96 , 0x01F26 , 0x00345 }, { 0x01F97 , 0x01F27 , 0x00345 }, { 0x01F98 , 0x01F28 , 0x00345 }, { 0x01F99 , 0x01F29 , 0x00345 }, { 0x01F9A , 0x01F2A , 0x00345 }, { 0x01F9B , 0x01F2B , 0x00345 }, { 0x01F9C , 0x01F2C , 0x00345 }, { 0x01F9D , 0x01F2D , 0x00345 }, { 0x01F9E , 0x01F2E , 0x00345 }, { 0x01F9F , 0x01F2F , 0x00345 }, { 0x01FA0 , 0x01F60 , 0x00345 }, { 0x01FA1 , 0x01F61 , 0x00345 }, { 0x01FA2 , 0x01F62 , 0x00345 }, { 0x01FA3 , 0x01F63 , 0x00345 }, { 0x01FA4 , 0x01F64 , 0x00345 }, { 0x01FA5 , 0x01F65 , 0x00345 }, { 0x01FA6 , 0x01F66 , 0x00345 }, { 0x01FA7 , 0x01F67 , 0x00345 }, { 0x01FA8 , 0x01F68 , 0x00345 }, { 0x01FA9 , 0x01F69 , 0x00345 }, { 0x01FAA , 0x01F6A , 0x00345 }, { 0x01FAB , 0x01F6B , 0x00345 }, { 0x01FAC , 0x01F6C , 0x00345 }, { 0x01FAD , 0x01F6D , 0x00345 }, { 0x01FAE , 0x01F6E , 0x00345 }, { 0x01FAF , 0x01F6F , 0x00345 }, { 0x01FB0 , 0x003B1 , 0x00306 }, { 0x01FB1 , 0x003B1 , 0x00304 }, { 0x01FB2 , 0x01F70 , 0x00345 }, { 0x01FB3 , 0x003B1 , 0x00345 }, { 0x01FB4 , 0x003AC , 0x00345 }, { 0x01FB6 , 0x003B1 , 0x00342 }, { 0x01FB7 , 0x01FB6 , 0x00345 }, { 0x01FB8 , 0x00391 , 0x00306 }, { 0x01FB9 , 0x00391 , 0x00304 }, { 0x01FBA , 0x00391 , 0x00300 }, { 0x01FBC , 0x00391 , 0x00345 }, { 0x01FC1 , 0x000A8 , 0x00342 }, { 0x01FC2 , 0x01F74 , 0x00345 }, { 0x01FC3 , 0x003B7 , 0x00345 }, { 0x01FC4 , 0x003AE , 0x00345 }, { 0x01FC6 , 0x003B7 , 0x00342 }, { 0x01FC7 , 0x01FC6 , 0x00345 }, { 0x01FC8 , 0x00395 , 0x00300 }, { 0x01FCA , 0x00397 , 0x00300 }, { 0x01FCC , 0x00397 , 0x00345 }, { 0x01FCD , 0x01FBF , 0x00300 }, { 0x01FCE , 0x01FBF , 0x00301 }, { 0x01FCF , 0x01FBF , 0x00342 }, { 0x01FD0 , 0x003B9 , 0x00306 }, { 0x01FD1 , 0x003B9 , 0x00304 }, { 0x01FD2 , 0x003CA , 0x00300 }, { 0x01FD6 , 0x003B9 , 0x00342 }, { 0x01FD7 , 0x003CA , 0x00342 }, { 0x01FD8 , 0x00399 , 0x00306 }, { 0x01FD9 , 0x00399 , 0x00304 }, { 0x01FDA , 0x00399 , 0x00300 }, { 0x01FDD , 0x01FFE , 0x00300 }, { 0x01FDE , 0x01FFE , 0x00301 }, { 0x01FDF , 0x01FFE , 0x00342 }, { 0x01FE0 , 0x003C5 , 0x00306 }, { 0x01FE1 , 0x003C5 , 0x00304 }, { 0x01FE2 , 0x003CB , 0x00300 }, { 0x01FE4 , 0x003C1 , 0x00313 }, { 0x01FE5 , 0x003C1 , 0x00314 }, { 0x01FE6 , 0x003C5 , 0x00342 }, { 0x01FE7 , 0x003CB , 0x00342 }, { 0x01FE8 , 0x003A5 , 0x00306 }, { 0x01FE9 , 0x003A5 , 0x00304 }, { 0x01FEA , 0x003A5 , 0x00300 }, { 0x01FEC , 0x003A1 , 0x00314 }, { 0x01FED , 0x000A8 , 0x00300 }, { 0x01FF2 , 0x01F7C , 0x00345 }, { 0x01FF3 , 0x003C9 , 0x00345 }, { 0x01FF4 , 0x003CE , 0x00345 }, { 0x01FF6 , 0x003C9 , 0x00342 }, { 0x01FF7 , 0x01FF6 , 0x00345 }, { 0x01FF8 , 0x0039F , 0x00300 }, { 0x01FFA , 0x003A9 , 0x00300 }, { 0x01FFC , 0x003A9 , 0x00345 }, { 0x0219A , 0x02190 , 0x00338 }, { 0x0219B , 0x02192 , 0x00338 }, { 0x021AE , 0x02194 , 0x00338 }, { 0x021CD , 0x021D0 , 0x00338 }, { 0x021CE , 0x021D4 , 0x00338 }, { 0x021CF , 0x021D2 , 0x00338 }, { 0x02204 , 0x02203 , 0x00338 }, { 0x02209 , 0x02208 , 0x00338 }, { 0x0220C , 0x0220B , 0x00338 }, { 0x02224 , 0x02223 , 0x00338 }, { 0x02226 , 0x02225 , 0x00338 }, { 0x02241 , 0x0223C , 0x00338 }, { 0x02244 , 0x02243 , 0x00338 }, { 0x02247 , 0x02245 , 0x00338 }, { 0x02249 , 0x02248 , 0x00338 }, { 0x02260 , 0x0003D , 0x00338 }, { 0x02262 , 0x02261 , 0x00338 }, { 0x0226D , 0x0224D , 0x00338 }, { 0x0226E , 0x0003C , 0x00338 }, { 0x0226F , 0x0003E , 0x00338 }, { 0x02270 , 0x02264 , 0x00338 }, { 0x02271 , 0x02265 , 0x00338 }, { 0x02274 , 0x02272 , 0x00338 }, { 0x02275 , 0x02273 , 0x00338 }, { 0x02278 , 0x02276 , 0x00338 }, { 0x02279 , 0x02277 , 0x00338 }, { 0x02280 , 0x0227A , 0x00338 }, { 0x02281 , 0x0227B , 0x00338 }, { 0x02284 , 0x02282 , 0x00338 }, { 0x02285 , 0x02283 , 0x00338 }, { 0x02288 , 0x02286 , 0x00338 }, { 0x02289 , 0x02287 , 0x00338 }, { 0x022AC , 0x022A2 , 0x00338 }, { 0x022AD , 0x022A8 , 0x00338 }, { 0x022AE , 0x022A9 , 0x00338 }, { 0x022AF , 0x022AB , 0x00338 }, { 0x022E0 , 0x0227C , 0x00338 }, { 0x022E1 , 0x0227D , 0x00338 }, { 0x022E2 , 0x02291 , 0x00338 }, { 0x022E3 , 0x02292 , 0x00338 }, { 0x022EA , 0x022B2 , 0x00338 }, { 0x022EB , 0x022B3 , 0x00338 }, { 0x022EC , 0x022B4 , 0x00338 }, { 0x022ED , 0x022B5 , 0x00338 }, { 0x0304C , 0x0304B , 0x03099 }, { 0x0304E , 0x0304D , 0x03099 }, { 0x03050 , 0x0304F , 0x03099 }, { 0x03052 , 0x03051 , 0x03099 }, { 0x03054 , 0x03053 , 0x03099 }, { 0x03056 , 0x03055 , 0x03099 }, { 0x03058 , 0x03057 , 0x03099 }, { 0x0305A , 0x03059 , 0x03099 }, { 0x0305C , 0x0305B , 0x03099 }, { 0x0305E , 0x0305D , 0x03099 }, { 0x03060 , 0x0305F , 0x03099 }, { 0x03062 , 0x03061 , 0x03099 }, { 0x03065 , 0x03064 , 0x03099 }, { 0x03067 , 0x03066 , 0x03099 }, { 0x03069 , 0x03068 , 0x03099 }, { 0x03070 , 0x0306F , 0x03099 }, { 0x03071 , 0x0306F , 0x0309A }, { 0x03073 , 0x03072 , 0x03099 }, { 0x03074 , 0x03072 , 0x0309A }, { 0x03076 , 0x03075 , 0x03099 }, { 0x03077 , 0x03075 , 0x0309A }, { 0x03079 , 0x03078 , 0x03099 }, { 0x0307A , 0x03078 , 0x0309A }, { 0x0307C , 0x0307B , 0x03099 }, { 0x0307D , 0x0307B , 0x0309A }, { 0x03094 , 0x03046 , 0x03099 }, { 0x0309E , 0x0309D , 0x03099 }, { 0x030AC , 0x030AB , 0x03099 }, { 0x030AE , 0x030AD , 0x03099 }, { 0x030B0 , 0x030AF , 0x03099 }, { 0x030B2 , 0x030B1 , 0x03099 }, { 0x030B4 , 0x030B3 , 0x03099 }, { 0x030B6 , 0x030B5 , 0x03099 }, { 0x030B8 , 0x030B7 , 0x03099 }, { 0x030BA , 0x030B9 , 0x03099 }, { 0x030BC , 0x030BB , 0x03099 }, { 0x030BE , 0x030BD , 0x03099 }, { 0x030C0 , 0x030BF , 0x03099 }, { 0x030C2 , 0x030C1 , 0x03099 }, { 0x030C5 , 0x030C4 , 0x03099 }, { 0x030C7 , 0x030C6 , 0x03099 }, { 0x030C9 , 0x030C8 , 0x03099 }, { 0x030D0 , 0x030CF , 0x03099 }, { 0x030D1 , 0x030CF , 0x0309A }, { 0x030D3 , 0x030D2 , 0x03099 }, { 0x030D4 , 0x030D2 , 0x0309A }, { 0x030D6 , 0x030D5 , 0x03099 }, { 0x030D7 , 0x030D5 , 0x0309A }, { 0x030D9 , 0x030D8 , 0x03099 }, { 0x030DA , 0x030D8 , 0x0309A }, { 0x030DC , 0x030DB , 0x03099 }, { 0x030DD , 0x030DB , 0x0309A }, { 0x030F4 , 0x030A6 , 0x03099 }, { 0x030F7 , 0x030EF , 0x03099 }, { 0x030F8 , 0x030F0 , 0x03099 }, { 0x030F9 , 0x030F1 , 0x03099 }, { 0x030FA , 0x030F2 , 0x03099 }, { 0x030FE , 0x030FD , 0x03099 }, { 0x1109A , 0x11099 , 0x110BA }, { 0x1109C , 0x1109B , 0x110BA }, { 0x110AB , 0x110A5 , 0x110BA }, }; #endif /* ARCHIVE_STRING_COMPOSITION_H_INCLUDED */ Index: head/contrib/libarchive/libarchive/archive_write.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_write.c (revision 311903) @@ -1,734 +1,734 @@ /*- * Copyright (c) 2003-2010 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); /* * This file contains the "essential" portions of the write API, that * is, stuff that will essentially always be used by any client that * actually needs to write an archive. Optional pieces have been, as * far as possible, separated out into separate files to reduce * needlessly bloating statically-linked clients. */ #ifdef HAVE_SYS_WAIT_H #include #endif #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #include #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include #ifdef HAVE_UNISTD_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_private.h" #include "archive_write_private.h" static struct archive_vtable *archive_write_vtable(void); static int _archive_filter_code(struct archive *, int); static const char *_archive_filter_name(struct archive *, int); static int64_t _archive_filter_bytes(struct archive *, int); static int _archive_write_filter_count(struct archive *); static int _archive_write_close(struct archive *); static int _archive_write_free(struct archive *); static int _archive_write_header(struct archive *, struct archive_entry *); static int _archive_write_finish_entry(struct archive *); static ssize_t _archive_write_data(struct archive *, const void *, size_t); struct archive_none { size_t buffer_size; size_t avail; char *buffer; char *next; }; static struct archive_vtable * archive_write_vtable(void) { static struct archive_vtable av; static int inited = 0; if (!inited) { av.archive_close = _archive_write_close; av.archive_filter_bytes = _archive_filter_bytes; av.archive_filter_code = _archive_filter_code; av.archive_filter_name = _archive_filter_name; av.archive_filter_count = _archive_write_filter_count; av.archive_free = _archive_write_free; av.archive_write_header = _archive_write_header; av.archive_write_finish_entry = _archive_write_finish_entry; av.archive_write_data = _archive_write_data; inited = 1; } return (&av); } /* * Allocate, initialize and return an archive object. */ struct archive * archive_write_new(void) { struct archive_write *a; unsigned char *nulls; a = (struct archive_write *)calloc(1, sizeof(*a)); if (a == NULL) return (NULL); a->archive.magic = ARCHIVE_WRITE_MAGIC; a->archive.state = ARCHIVE_STATE_NEW; a->archive.vtable = archive_write_vtable(); /* * The value 10240 here matches the traditional tar default, * but is otherwise arbitrary. * TODO: Set the default block size from the format selected. */ a->bytes_per_block = 10240; a->bytes_in_last_block = -1; /* Default */ /* Initialize a block of nulls for padding purposes. */ a->null_length = 1024; nulls = (unsigned char *)calloc(1, a->null_length); if (nulls == NULL) { free(a); return (NULL); } a->nulls = nulls; return (&a->archive); } /* * Set the block size. Returns 0 if successful. */ int archive_write_set_bytes_per_block(struct archive *_a, int bytes_per_block) { struct archive_write *a = (struct archive_write *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_bytes_per_block"); a->bytes_per_block = bytes_per_block; return (ARCHIVE_OK); } /* * Get the current block size. -1 if it has never been set. */ int archive_write_get_bytes_per_block(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_ANY, "archive_write_get_bytes_per_block"); return (a->bytes_per_block); } /* * Set the size for the last block. * Returns 0 if successful. */ int archive_write_set_bytes_in_last_block(struct archive *_a, int bytes) { struct archive_write *a = (struct archive_write *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_ANY, "archive_write_set_bytes_in_last_block"); a->bytes_in_last_block = bytes; return (ARCHIVE_OK); } /* * Return the value set above. -1 indicates it has not been set. */ int archive_write_get_bytes_in_last_block(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_ANY, "archive_write_get_bytes_in_last_block"); return (a->bytes_in_last_block); } /* * dev/ino of a file to be rejected. Used to prevent adding * an archive to itself recursively. */ int archive_write_set_skip_file(struct archive *_a, int64_t d, int64_t i) { struct archive_write *a = (struct archive_write *)_a; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_ANY, "archive_write_set_skip_file"); a->skip_file_set = 1; a->skip_file_dev = d; a->skip_file_ino = i; return (ARCHIVE_OK); } /* * Allocate and return the next filter structure. */ struct archive_write_filter * __archive_write_allocate_filter(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct archive_write_filter *f; f = calloc(1, sizeof(*f)); f->archive = _a; if (a->filter_first == NULL) a->filter_first = f; else a->filter_last->next_filter = f; a->filter_last = f; return f; } /* * Write data to a particular filter. */ int __archive_write_filter(struct archive_write_filter *f, const void *buff, size_t length) { int r; if (length == 0) return(ARCHIVE_OK); if (f->write == NULL) - /* If unset, a fatal error has already ocuured, so this filter + /* If unset, a fatal error has already occurred, so this filter * didn't open. We cannot write anything. */ return(ARCHIVE_FATAL); r = (f->write)(f, buff, length); f->bytes_written += length; return (r); } /* * Open a filter. */ int __archive_write_open_filter(struct archive_write_filter *f) { if (f->open == NULL) return (ARCHIVE_OK); return (f->open)(f); } /* * Close a filter. */ int __archive_write_close_filter(struct archive_write_filter *f) { if (f->close != NULL) return (f->close)(f); if (f->next_filter != NULL) return (__archive_write_close_filter(f->next_filter)); return (ARCHIVE_OK); } int __archive_write_output(struct archive_write *a, const void *buff, size_t length) { return (__archive_write_filter(a->filter_first, buff, length)); } int __archive_write_nulls(struct archive_write *a, size_t length) { if (length == 0) return (ARCHIVE_OK); while (length > 0) { size_t to_write = length < a->null_length ? length : a->null_length; int r = __archive_write_output(a, a->nulls, to_write); if (r < ARCHIVE_OK) return (r); length -= to_write; } return (ARCHIVE_OK); } static int archive_write_client_open(struct archive_write_filter *f) { struct archive_write *a = (struct archive_write *)f->archive; struct archive_none *state; void *buffer; size_t buffer_size; f->bytes_per_block = archive_write_get_bytes_per_block(f->archive); f->bytes_in_last_block = archive_write_get_bytes_in_last_block(f->archive); buffer_size = f->bytes_per_block; state = (struct archive_none *)calloc(1, sizeof(*state)); buffer = (char *)malloc(buffer_size); if (state == NULL || buffer == NULL) { free(state); free(buffer); archive_set_error(f->archive, ENOMEM, "Can't allocate data for output buffering"); return (ARCHIVE_FATAL); } state->buffer_size = buffer_size; state->buffer = buffer; state->next = state->buffer; state->avail = state->buffer_size; f->data = state; if (a->client_opener == NULL) return (ARCHIVE_OK); return (a->client_opener(f->archive, a->client_data)); } static int archive_write_client_write(struct archive_write_filter *f, const void *_buff, size_t length) { struct archive_write *a = (struct archive_write *)f->archive; struct archive_none *state = (struct archive_none *)f->data; const char *buff = (const char *)_buff; ssize_t remaining, to_copy; ssize_t bytes_written; remaining = length; /* * If there is no buffer for blocking, just pass the data * straight through to the client write callback. In * particular, this supports "no write delay" operation for * special applications. Just set the block size to zero. */ if (state->buffer_size == 0) { while (remaining > 0) { bytes_written = (a->client_writer)(&a->archive, a->client_data, buff, remaining); if (bytes_written <= 0) return (ARCHIVE_FATAL); remaining -= bytes_written; buff += bytes_written; } return (ARCHIVE_OK); } /* If the copy buffer isn't empty, try to fill it. */ if (state->avail < state->buffer_size) { /* If buffer is not empty... */ /* ... copy data into buffer ... */ to_copy = ((size_t)remaining > state->avail) ? state->avail : (size_t)remaining; memcpy(state->next, buff, to_copy); state->next += to_copy; state->avail -= to_copy; buff += to_copy; remaining -= to_copy; /* ... if it's full, write it out. */ if (state->avail == 0) { char *p = state->buffer; size_t to_write = state->buffer_size; while (to_write > 0) { bytes_written = (a->client_writer)(&a->archive, a->client_data, p, to_write); if (bytes_written <= 0) return (ARCHIVE_FATAL); if ((size_t)bytes_written > to_write) { archive_set_error(&(a->archive), -1, "write overrun"); return (ARCHIVE_FATAL); } p += bytes_written; to_write -= bytes_written; } state->next = state->buffer; state->avail = state->buffer_size; } } while ((size_t)remaining >= state->buffer_size) { /* Write out full blocks directly to client. */ bytes_written = (a->client_writer)(&a->archive, a->client_data, buff, state->buffer_size); if (bytes_written <= 0) return (ARCHIVE_FATAL); buff += bytes_written; remaining -= bytes_written; } if (remaining > 0) { /* Copy last bit into copy buffer. */ memcpy(state->next, buff, remaining); state->next += remaining; state->avail -= remaining; } return (ARCHIVE_OK); } static int archive_write_client_close(struct archive_write_filter *f) { struct archive_write *a = (struct archive_write *)f->archive; struct archive_none *state = (struct archive_none *)f->data; ssize_t block_length; ssize_t target_block_length; ssize_t bytes_written; int ret = ARCHIVE_OK; /* If there's pending data, pad and write the last block */ if (state->next != state->buffer) { block_length = state->buffer_size - state->avail; /* Tricky calculation to determine size of last block */ if (a->bytes_in_last_block <= 0) /* Default or Zero: pad to full block */ target_block_length = a->bytes_per_block; else /* Round to next multiple of bytes_in_last_block. */ target_block_length = a->bytes_in_last_block * ( (block_length + a->bytes_in_last_block - 1) / a->bytes_in_last_block); if (target_block_length > a->bytes_per_block) target_block_length = a->bytes_per_block; if (block_length < target_block_length) { memset(state->next, 0, target_block_length - block_length); block_length = target_block_length; } bytes_written = (a->client_writer)(&a->archive, a->client_data, state->buffer, block_length); ret = bytes_written <= 0 ? ARCHIVE_FATAL : ARCHIVE_OK; } if (a->client_closer) (*a->client_closer)(&a->archive, a->client_data); free(state->buffer); free(state); /* Clear the close handler myself not to be called again. */ f->close = NULL; a->client_data = NULL; /* Clear passphrase. */ if (a->passphrase != NULL) { memset(a->passphrase, 0, strlen(a->passphrase)); free(a->passphrase); a->passphrase = NULL; } return (ret); } /* * Open the archive using the current settings. */ int archive_write_open(struct archive *_a, void *client_data, archive_open_callback *opener, archive_write_callback *writer, archive_close_callback *closer) { struct archive_write *a = (struct archive_write *)_a; struct archive_write_filter *client_filter; int ret, r1; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_open"); archive_clear_error(&a->archive); a->client_writer = writer; a->client_opener = opener; a->client_closer = closer; a->client_data = client_data; client_filter = __archive_write_allocate_filter(_a); client_filter->open = archive_write_client_open; client_filter->write = archive_write_client_write; client_filter->close = archive_write_client_close; ret = __archive_write_open_filter(a->filter_first); if (ret < ARCHIVE_WARN) { r1 = __archive_write_close_filter(a->filter_first); return (r1 < ret ? r1 : ret); } a->archive.state = ARCHIVE_STATE_HEADER; if (a->format_init) ret = (a->format_init)(a); return (ret); } /* * Close out the archive. */ static int _archive_write_close(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int r = ARCHIVE_OK, r1 = ARCHIVE_OK; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_write_close"); if (a->archive.state == ARCHIVE_STATE_NEW || a->archive.state == ARCHIVE_STATE_CLOSED) return (ARCHIVE_OK); /* Okay to close() when not open. */ archive_clear_error(&a->archive); /* Finish the last entry if a finish callback is specified */ if (a->archive.state == ARCHIVE_STATE_DATA && a->format_finish_entry != NULL) r = ((a->format_finish_entry)(a)); /* Finish off the archive. */ /* TODO: have format closers invoke compression close. */ if (a->format_close != NULL) { r1 = (a->format_close)(a); if (r1 < r) r = r1; } /* Finish the compression and close the stream. */ r1 = __archive_write_close_filter(a->filter_first); if (r1 < r) r = r1; if (a->archive.state != ARCHIVE_STATE_FATAL) a->archive.state = ARCHIVE_STATE_CLOSED; return (r); } static int _archive_write_filter_count(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct archive_write_filter *p = a->filter_first; int count = 0; while(p) { count++; p = p->next_filter; } return count; } void __archive_write_filters_free(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int r = ARCHIVE_OK, r1; while (a->filter_first != NULL) { struct archive_write_filter *next = a->filter_first->next_filter; if (a->filter_first->free != NULL) { r1 = (*a->filter_first->free)(a->filter_first); if (r > r1) r = r1; } free(a->filter_first); a->filter_first = next; } a->filter_last = NULL; } /* * Destroy the archive structure. * * Be careful: user might just call write_new and then write_free. * Don't assume we actually wrote anything or performed any non-trivial * initialization. */ static int _archive_write_free(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int r = ARCHIVE_OK, r1; if (_a == NULL) return (ARCHIVE_OK); /* It is okay to call free() in state FATAL. */ archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_ANY | ARCHIVE_STATE_FATAL, "archive_write_free"); if (a->archive.state != ARCHIVE_STATE_FATAL) r = archive_write_close(&a->archive); /* Release format resources. */ if (a->format_free != NULL) { r1 = (a->format_free)(a); if (r1 < r) r = r1; } __archive_write_filters_free(_a); /* Release various dynamic buffers. */ free((void *)(uintptr_t)(const void *)a->nulls); archive_string_free(&a->archive.error_string); if (a->passphrase != NULL) { /* A passphrase should be cleaned. */ memset(a->passphrase, 0, strlen(a->passphrase)); free(a->passphrase); } a->archive.magic = 0; __archive_clean(&a->archive); free(a); return (r); } /* * Write the appropriate header. */ static int _archive_write_header(struct archive *_a, struct archive_entry *entry) { struct archive_write *a = (struct archive_write *)_a; int ret, r2; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_DATA | ARCHIVE_STATE_HEADER, "archive_write_header"); archive_clear_error(&a->archive); if (a->format_write_header == NULL) { archive_set_error(&(a->archive), -1, "Format must be set before you can write to an archive."); a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } /* In particular, "retry" and "fatal" get returned immediately. */ ret = archive_write_finish_entry(&a->archive); if (ret == ARCHIVE_FATAL) { a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } if (ret < ARCHIVE_OK && ret != ARCHIVE_WARN) return (ret); if (a->skip_file_set && archive_entry_dev_is_set(entry) && archive_entry_ino_is_set(entry) && archive_entry_dev(entry) == (dev_t)a->skip_file_dev && archive_entry_ino64(entry) == a->skip_file_ino) { archive_set_error(&a->archive, 0, "Can't add archive to itself"); return (ARCHIVE_FAILED); } /* Format and write header. */ r2 = ((a->format_write_header)(a, entry)); if (r2 == ARCHIVE_FAILED) { return (ARCHIVE_FAILED); } if (r2 == ARCHIVE_FATAL) { a->archive.state = ARCHIVE_STATE_FATAL; return (ARCHIVE_FATAL); } if (r2 < ret) ret = r2; a->archive.state = ARCHIVE_STATE_DATA; return (ret); } static int _archive_write_finish_entry(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int ret = ARCHIVE_OK; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_finish_entry"); if (a->archive.state & ARCHIVE_STATE_DATA && a->format_finish_entry != NULL) ret = (a->format_finish_entry)(a); a->archive.state = ARCHIVE_STATE_HEADER; return (ret); } /* * Note that the compressor is responsible for blocking. */ static ssize_t _archive_write_data(struct archive *_a, const void *buff, size_t s) { struct archive_write *a = (struct archive_write *)_a; const size_t max_write = INT_MAX; archive_check_magic(&a->archive, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_DATA, "archive_write_data"); /* In particular, this catches attempts to pass negative values. */ if (s > max_write) s = max_write; archive_clear_error(&a->archive); return ((a->format_write_data)(a, buff, s)); } static struct archive_write_filter * filter_lookup(struct archive *_a, int n) { struct archive_write *a = (struct archive_write *)_a; struct archive_write_filter *f = a->filter_first; if (n == -1) return a->filter_last; if (n < 0) return NULL; while (n > 0 && f != NULL) { f = f->next_filter; --n; } return f; } static int _archive_filter_code(struct archive *_a, int n) { struct archive_write_filter *f = filter_lookup(_a, n); return f == NULL ? -1 : f->code; } static const char * _archive_filter_name(struct archive *_a, int n) { struct archive_write_filter *f = filter_lookup(_a, n); return f != NULL ? f->name : NULL; } static int64_t _archive_filter_bytes(struct archive *_a, int n) { struct archive_write_filter *f = filter_lookup(_a, n); return f == NULL ? -1 : f->bytes_written; } Index: head/contrib/libarchive/libarchive/archive_write_add_filter_xz.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_add_filter_xz.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_write_add_filter_xz.c (revision 311903) @@ -1,547 +1,547 @@ /*- * Copyright (c) 2003-2010 Tim Kientzle * Copyright (c) 2009-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD: head/lib/libarchive/archive_write_set_compression_xz.c 201108 2009-12-28 03:28:21Z kientzle $"); #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include #ifdef HAVE_LZMA_H #include #endif #include "archive.h" #include "archive_endian.h" #include "archive_private.h" #include "archive_write_private.h" #if ARCHIVE_VERSION_NUMBER < 4000000 int archive_write_set_compression_lzip(struct archive *a) { __archive_write_filters_free(a); return (archive_write_add_filter_lzip(a)); } int archive_write_set_compression_lzma(struct archive *a) { __archive_write_filters_free(a); return (archive_write_add_filter_lzma(a)); } int archive_write_set_compression_xz(struct archive *a) { __archive_write_filters_free(a); return (archive_write_add_filter_xz(a)); } #endif #ifndef HAVE_LZMA_H int archive_write_add_filter_xz(struct archive *a) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "xz compression not supported on this platform"); return (ARCHIVE_FATAL); } int archive_write_add_filter_lzma(struct archive *a) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "lzma compression not supported on this platform"); return (ARCHIVE_FATAL); } int archive_write_add_filter_lzip(struct archive *a) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "lzma compression not supported on this platform"); return (ARCHIVE_FATAL); } #else /* Don't compile this if we don't have liblzma. */ struct private_data { int compression_level; uint32_t threads; lzma_stream stream; lzma_filter lzmafilters[2]; lzma_options_lzma lzma_opt; int64_t total_in; unsigned char *compressed; size_t compressed_buffer_size; int64_t total_out; /* the CRC32 value of uncompressed data for lzip */ uint32_t crc32; }; static int archive_compressor_xz_options(struct archive_write_filter *, const char *, const char *); static int archive_compressor_xz_open(struct archive_write_filter *); static int archive_compressor_xz_write(struct archive_write_filter *, const void *, size_t); static int archive_compressor_xz_close(struct archive_write_filter *); static int archive_compressor_xz_free(struct archive_write_filter *); static int drive_compressor(struct archive_write_filter *, struct private_data *, int finishing); struct option_value { uint32_t dict_size; uint32_t nice_len; lzma_match_finder mf; }; static const struct option_value option_values[] = { { 1 << 16, 32, LZMA_MF_HC3}, { 1 << 20, 32, LZMA_MF_HC3}, { 3 << 19, 32, LZMA_MF_HC4}, { 1 << 21, 32, LZMA_MF_BT4}, { 3 << 20, 32, LZMA_MF_BT4}, { 1 << 22, 32, LZMA_MF_BT4}, { 1 << 23, 64, LZMA_MF_BT4}, { 1 << 24, 64, LZMA_MF_BT4}, { 3 << 23, 64, LZMA_MF_BT4}, { 1 << 25, 64, LZMA_MF_BT4} }; static int common_setup(struct archive_write_filter *f) { struct private_data *data; struct archive_write *a = (struct archive_write *)f->archive; data = calloc(1, sizeof(*data)); if (data == NULL) { archive_set_error(&a->archive, ENOMEM, "Out of memory"); return (ARCHIVE_FATAL); } f->data = data; data->compression_level = LZMA_PRESET_DEFAULT; data->threads = 1; f->open = &archive_compressor_xz_open; f->close = archive_compressor_xz_close; f->free = archive_compressor_xz_free; f->options = &archive_compressor_xz_options; return (ARCHIVE_OK); } /* * Add an xz compression filter to this write handle. */ int archive_write_add_filter_xz(struct archive *_a) { struct archive_write_filter *f; int r; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_add_filter_xz"); f = __archive_write_allocate_filter(_a); r = common_setup(f); if (r == ARCHIVE_OK) { f->code = ARCHIVE_FILTER_XZ; f->name = "xz"; } return (r); } /* LZMA is handled identically, we just need a different compression * code set. (The liblzma setup looks at the code to determine * the one place that XZ and LZMA require different handling.) */ int archive_write_add_filter_lzma(struct archive *_a) { struct archive_write_filter *f; int r; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_add_filter_lzma"); f = __archive_write_allocate_filter(_a); r = common_setup(f); if (r == ARCHIVE_OK) { f->code = ARCHIVE_FILTER_LZMA; f->name = "lzma"; } return (r); } int archive_write_add_filter_lzip(struct archive *_a) { struct archive_write_filter *f; int r; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_add_filter_lzip"); f = __archive_write_allocate_filter(_a); r = common_setup(f); if (r == ARCHIVE_OK) { f->code = ARCHIVE_FILTER_LZIP; f->name = "lzip"; } return (r); } static int archive_compressor_xz_init_stream(struct archive_write_filter *f, struct private_data *data) { static const lzma_stream lzma_stream_init_data = LZMA_STREAM_INIT; int ret; #ifdef HAVE_LZMA_STREAM_ENCODER_MT lzma_mt mt_options; #endif data->stream = lzma_stream_init_data; data->stream.next_out = data->compressed; data->stream.avail_out = data->compressed_buffer_size; if (f->code == ARCHIVE_FILTER_XZ) { #ifdef HAVE_LZMA_STREAM_ENCODER_MT if (data->threads != 1) { - bzero(&mt_options, sizeof(mt_options)); + memset(&mt_options, 0, sizeof(mt_options)); mt_options.threads = data->threads; mt_options.timeout = 300; mt_options.filters = data->lzmafilters; mt_options.check = LZMA_CHECK_CRC64; ret = lzma_stream_encoder_mt(&(data->stream), &mt_options); } else #endif ret = lzma_stream_encoder(&(data->stream), data->lzmafilters, LZMA_CHECK_CRC64); } else if (f->code == ARCHIVE_FILTER_LZMA) { ret = lzma_alone_encoder(&(data->stream), &data->lzma_opt); } else { /* ARCHIVE_FILTER_LZIP */ int dict_size = data->lzma_opt.dict_size; int ds, log2dic, wedges; /* Calculate a coded dictionary size */ if (dict_size < (1 << 12) || dict_size > (1 << 27)) { archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "Unacceptable dictionary size for lzip: %d", dict_size); return (ARCHIVE_FATAL); } for (log2dic = 27; log2dic >= 12; log2dic--) { if (dict_size & (1 << log2dic)) break; } if (dict_size > (1 << log2dic)) { log2dic++; wedges = ((1 << log2dic) - dict_size) / (1 << (log2dic - 4)); } else wedges = 0; ds = ((wedges << 5) & 0xe0) | (log2dic & 0x1f); data->crc32 = 0; /* Make a header */ data->compressed[0] = 0x4C; data->compressed[1] = 0x5A; data->compressed[2] = 0x49; data->compressed[3] = 0x50; data->compressed[4] = 1;/* Version */ data->compressed[5] = (unsigned char)ds; data->stream.next_out += 6; data->stream.avail_out -= 6; ret = lzma_raw_encoder(&(data->stream), data->lzmafilters); } if (ret == LZMA_OK) return (ARCHIVE_OK); switch (ret) { case LZMA_MEM_ERROR: archive_set_error(f->archive, ENOMEM, "Internal error initializing compression library: " "Cannot allocate memory"); break; default: archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library: " "It's a bug in liblzma"); break; } return (ARCHIVE_FATAL); } /* * Setup callback. */ static int archive_compressor_xz_open(struct archive_write_filter *f) { struct private_data *data = f->data; int ret; ret = __archive_write_open_filter(f->next_filter); if (ret != ARCHIVE_OK) return (ret); if (data->compressed == NULL) { size_t bs = 65536, bpb; if (f->archive->magic == ARCHIVE_WRITE_MAGIC) { /* Buffer size should be a multiple number of the of bytes * per block for performance. */ bpb = archive_write_get_bytes_per_block(f->archive); if (bpb > bs) bs = bpb; else if (bpb != 0) bs -= bs % bpb; } data->compressed_buffer_size = bs; data->compressed = (unsigned char *)malloc(data->compressed_buffer_size); if (data->compressed == NULL) { archive_set_error(f->archive, ENOMEM, "Can't allocate data for compression buffer"); return (ARCHIVE_FATAL); } } f->write = archive_compressor_xz_write; /* Initialize compression library. */ if (f->code == ARCHIVE_FILTER_LZIP) { const struct option_value *val = &option_values[data->compression_level]; data->lzma_opt.dict_size = val->dict_size; data->lzma_opt.preset_dict = NULL; data->lzma_opt.preset_dict_size = 0; data->lzma_opt.lc = LZMA_LC_DEFAULT; data->lzma_opt.lp = LZMA_LP_DEFAULT; data->lzma_opt.pb = LZMA_PB_DEFAULT; data->lzma_opt.mode = data->compression_level<= 2? LZMA_MODE_FAST:LZMA_MODE_NORMAL; data->lzma_opt.nice_len = val->nice_len; data->lzma_opt.mf = val->mf; data->lzma_opt.depth = 0; data->lzmafilters[0].id = LZMA_FILTER_LZMA1; data->lzmafilters[0].options = &data->lzma_opt; data->lzmafilters[1].id = LZMA_VLI_UNKNOWN;/* Terminate */ } else { if (lzma_lzma_preset(&data->lzma_opt, data->compression_level)) { archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library"); } data->lzmafilters[0].id = LZMA_FILTER_LZMA2; data->lzmafilters[0].options = &data->lzma_opt; data->lzmafilters[1].id = LZMA_VLI_UNKNOWN;/* Terminate */ } ret = archive_compressor_xz_init_stream(f, data); if (ret == LZMA_OK) { f->data = data; return (0); } return (ARCHIVE_FATAL); } /* * Set write options. */ static int archive_compressor_xz_options(struct archive_write_filter *f, const char *key, const char *value) { struct private_data *data = (struct private_data *)f->data; if (strcmp(key, "compression-level") == 0) { if (value == NULL || !(value[0] >= '0' && value[0] <= '9') || value[1] != '\0') return (ARCHIVE_WARN); data->compression_level = value[0] - '0'; if (data->compression_level > 6) data->compression_level = 6; return (ARCHIVE_OK); } else if (strcmp(key, "threads") == 0) { if (value == NULL) return (ARCHIVE_WARN); data->threads = (int)strtoul(value, NULL, 10); if (data->threads == 0 && errno != 0) { data->threads = 1; return (ARCHIVE_WARN); } if (data->threads == 0) { #ifdef HAVE_LZMA_STREAM_ENCODER_MT data->threads = lzma_cputhreads(); #else data->threads = 1; #endif } return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } /* * Write data to the compressed stream. */ static int archive_compressor_xz_write(struct archive_write_filter *f, const void *buff, size_t length) { struct private_data *data = (struct private_data *)f->data; int ret; /* Update statistics */ data->total_in += length; if (f->code == ARCHIVE_FILTER_LZIP) data->crc32 = lzma_crc32(buff, length, data->crc32); /* Compress input data to output buffer */ data->stream.next_in = buff; data->stream.avail_in = length; if ((ret = drive_compressor(f, data, 0)) != ARCHIVE_OK) return (ret); return (ARCHIVE_OK); } /* * Finish the compression... */ static int archive_compressor_xz_close(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; int ret, r1; ret = drive_compressor(f, data, 1); if (ret == ARCHIVE_OK) { data->total_out += data->compressed_buffer_size - data->stream.avail_out; ret = __archive_write_filter(f->next_filter, data->compressed, data->compressed_buffer_size - data->stream.avail_out); if (f->code == ARCHIVE_FILTER_LZIP && ret == ARCHIVE_OK) { archive_le32enc(data->compressed, data->crc32); archive_le64enc(data->compressed+4, data->total_in); archive_le64enc(data->compressed+12, data->total_out + 20); ret = __archive_write_filter(f->next_filter, data->compressed, 20); } } lzma_end(&(data->stream)); r1 = __archive_write_close_filter(f->next_filter); return (r1 < ret ? r1 : ret); } static int archive_compressor_xz_free(struct archive_write_filter *f) { struct private_data *data = (struct private_data *)f->data; free(data->compressed); free(data); f->data = NULL; return (ARCHIVE_OK); } /* * Utility function to push input data through compressor, * writing full output blocks as necessary. * * Note that this handles both the regular write case (finishing == * false) and the end-of-archive case (finishing == true). */ static int drive_compressor(struct archive_write_filter *f, struct private_data *data, int finishing) { int ret; for (;;) { if (data->stream.avail_out == 0) { data->total_out += data->compressed_buffer_size; ret = __archive_write_filter(f->next_filter, data->compressed, data->compressed_buffer_size); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); data->stream.next_out = data->compressed; data->stream.avail_out = data->compressed_buffer_size; } /* If there's nothing to do, we're done. */ if (!finishing && data->stream.avail_in == 0) return (ARCHIVE_OK); ret = lzma_code(&(data->stream), finishing ? LZMA_FINISH : LZMA_RUN ); switch (ret) { case LZMA_OK: /* In non-finishing case, check if compressor * consumed everything */ if (!finishing && data->stream.avail_in == 0) return (ARCHIVE_OK); /* In finishing case, this return always means * there's more work */ break; case LZMA_STREAM_END: /* This return can only occur in finishing case. */ if (finishing) return (ARCHIVE_OK); archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "lzma compression data error"); return (ARCHIVE_FATAL); case LZMA_MEMLIMIT_ERROR: archive_set_error(f->archive, ENOMEM, "lzma compression error: " "%ju MiB would have been needed", (uintmax_t)((lzma_memusage(&(data->stream)) + 1024 * 1024 -1) / (1024 * 1024))); return (ARCHIVE_FATAL); default: /* Any other return value indicates an error. */ archive_set_error(f->archive, ARCHIVE_ERRNO_MISC, "lzma compression failed:" " lzma_code() call returned status %d", ret); return (ARCHIVE_FATAL); } } } #endif /* HAVE_LZMA_H */ Index: head/contrib/libarchive/libarchive/archive_write_disk_acl.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_disk_acl.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_write_disk_acl.c (revision 311903) @@ -1,322 +1,337 @@ /*- * Copyright (c) 2003-2010 Tim Kientzle * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer * in this position and unchanged. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD: head/lib/libarchive/archive_write_disk.c 201159 2009-12-29 05:35:40Z kientzle $"); #ifdef HAVE_SYS_TYPES_H #include #endif #ifdef HAVE_SYS_ACL_H #define _ACL_PRIVATE /* For debugging */ #include #endif #ifdef HAVE_ERRNO_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_acl_private.h" #include "archive_write_disk_private.h" #ifndef HAVE_POSIX_ACL /* Default empty function body to satisfy mainline code. */ int archive_write_disk_set_acls(struct archive *a, int fd, const char *name, struct archive_acl *abstract_acl) { (void)a; /* UNUSED */ (void)fd; /* UNUSED */ (void)name; /* UNUSED */ (void)abstract_acl; /* UNUSED */ return (ARCHIVE_OK); } #else static int set_acl(struct archive *, int fd, const char *, struct archive_acl *, acl_type_t, int archive_entry_acl_type, const char *tn); /* * XXX TODO: What about ACL types other than ACCESS and DEFAULT? */ int archive_write_disk_set_acls(struct archive *a, int fd, const char *name, struct archive_acl *abstract_acl) { int ret; if (archive_acl_count(abstract_acl, ARCHIVE_ENTRY_ACL_TYPE_POSIX1E) > 0) { ret = set_acl(a, fd, name, abstract_acl, ACL_TYPE_ACCESS, ARCHIVE_ENTRY_ACL_TYPE_ACCESS, "access"); if (ret != ARCHIVE_OK) return (ret); ret = set_acl(a, fd, name, abstract_acl, ACL_TYPE_DEFAULT, ARCHIVE_ENTRY_ACL_TYPE_DEFAULT, "default"); return (ret); #ifdef ACL_TYPE_NFS4 } else if (archive_acl_count(abstract_acl, ARCHIVE_ENTRY_ACL_TYPE_NFS4) > 0) { ret = set_acl(a, fd, name, abstract_acl, ACL_TYPE_NFS4, ARCHIVE_ENTRY_ACL_TYPE_NFS4, "nfs4"); return (ret); #endif } else return ARCHIVE_OK; } static struct { int archive_perm; int platform_perm; } acl_perm_map[] = { {ARCHIVE_ENTRY_ACL_EXECUTE, ACL_EXECUTE}, {ARCHIVE_ENTRY_ACL_WRITE, ACL_WRITE}, {ARCHIVE_ENTRY_ACL_READ, ACL_READ}, #ifdef ACL_TYPE_NFS4 {ARCHIVE_ENTRY_ACL_READ_DATA, ACL_READ_DATA}, {ARCHIVE_ENTRY_ACL_LIST_DIRECTORY, ACL_LIST_DIRECTORY}, {ARCHIVE_ENTRY_ACL_WRITE_DATA, ACL_WRITE_DATA}, {ARCHIVE_ENTRY_ACL_ADD_FILE, ACL_ADD_FILE}, {ARCHIVE_ENTRY_ACL_APPEND_DATA, ACL_APPEND_DATA}, {ARCHIVE_ENTRY_ACL_ADD_SUBDIRECTORY, ACL_ADD_SUBDIRECTORY}, {ARCHIVE_ENTRY_ACL_READ_NAMED_ATTRS, ACL_READ_NAMED_ATTRS}, {ARCHIVE_ENTRY_ACL_WRITE_NAMED_ATTRS, ACL_WRITE_NAMED_ATTRS}, {ARCHIVE_ENTRY_ACL_DELETE_CHILD, ACL_DELETE_CHILD}, {ARCHIVE_ENTRY_ACL_READ_ATTRIBUTES, ACL_READ_ATTRIBUTES}, {ARCHIVE_ENTRY_ACL_WRITE_ATTRIBUTES, ACL_WRITE_ATTRIBUTES}, {ARCHIVE_ENTRY_ACL_DELETE, ACL_DELETE}, {ARCHIVE_ENTRY_ACL_READ_ACL, ACL_READ_ACL}, {ARCHIVE_ENTRY_ACL_WRITE_ACL, ACL_WRITE_ACL}, {ARCHIVE_ENTRY_ACL_WRITE_OWNER, ACL_WRITE_OWNER}, {ARCHIVE_ENTRY_ACL_SYNCHRONIZE, ACL_SYNCHRONIZE} #endif }; #ifdef ACL_TYPE_NFS4 static struct { int archive_inherit; int platform_inherit; } acl_inherit_map[] = { {ARCHIVE_ENTRY_ACL_ENTRY_FILE_INHERIT, ACL_ENTRY_FILE_INHERIT}, {ARCHIVE_ENTRY_ACL_ENTRY_DIRECTORY_INHERIT, ACL_ENTRY_DIRECTORY_INHERIT}, {ARCHIVE_ENTRY_ACL_ENTRY_NO_PROPAGATE_INHERIT, ACL_ENTRY_NO_PROPAGATE_INHERIT}, - {ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, ACL_ENTRY_INHERIT_ONLY} + {ARCHIVE_ENTRY_ACL_ENTRY_INHERIT_ONLY, ACL_ENTRY_INHERIT_ONLY}, + {ARCHIVE_ENTRY_ACL_ENTRY_SUCCESSFUL_ACCESS, ACL_ENTRY_SUCCESSFUL_ACCESS}, + {ARCHIVE_ENTRY_ACL_ENTRY_FAILED_ACCESS, ACL_ENTRY_FAILED_ACCESS}, + {ARCHIVE_ENTRY_ACL_ENTRY_INHERITED, ACL_ENTRY_INHERITED} }; #endif static int set_acl(struct archive *a, int fd, const char *name, struct archive_acl *abstract_acl, acl_type_t acl_type, int ae_requested_type, const char *tname) { acl_t acl; acl_entry_t acl_entry; acl_permset_t acl_permset; #ifdef ACL_TYPE_NFS4 acl_flagset_t acl_flagset; int r; #endif int ret; int ae_type, ae_permset, ae_tag, ae_id; uid_t ae_uid; gid_t ae_gid; const char *ae_name; int entries; int i; ret = ARCHIVE_OK; entries = archive_acl_reset(abstract_acl, ae_requested_type); if (entries == 0) return (ARCHIVE_OK); acl = acl_init(entries); if (acl == (acl_t)NULL) { archive_set_error(a, errno, "Failed to initialize ACL working storage"); return (ARCHIVE_FAILED); } while (archive_acl_next(a, abstract_acl, ae_requested_type, &ae_type, &ae_permset, &ae_tag, &ae_id, &ae_name) == ARCHIVE_OK) { if (acl_create_entry(&acl, &acl_entry) != 0) { archive_set_error(a, errno, "Failed to create a new ACL entry"); ret = ARCHIVE_FAILED; goto exit_free; } switch (ae_tag) { case ARCHIVE_ENTRY_ACL_USER: acl_set_tag_type(acl_entry, ACL_USER); ae_uid = archive_write_disk_uid(a, ae_name, ae_id); acl_set_qualifier(acl_entry, &ae_uid); break; case ARCHIVE_ENTRY_ACL_GROUP: acl_set_tag_type(acl_entry, ACL_GROUP); ae_gid = archive_write_disk_gid(a, ae_name, ae_id); acl_set_qualifier(acl_entry, &ae_gid); break; case ARCHIVE_ENTRY_ACL_USER_OBJ: acl_set_tag_type(acl_entry, ACL_USER_OBJ); break; case ARCHIVE_ENTRY_ACL_GROUP_OBJ: acl_set_tag_type(acl_entry, ACL_GROUP_OBJ); break; case ARCHIVE_ENTRY_ACL_MASK: acl_set_tag_type(acl_entry, ACL_MASK); break; case ARCHIVE_ENTRY_ACL_OTHER: acl_set_tag_type(acl_entry, ACL_OTHER); break; #ifdef ACL_TYPE_NFS4 case ARCHIVE_ENTRY_ACL_EVERYONE: acl_set_tag_type(acl_entry, ACL_EVERYONE); break; #endif default: archive_set_error(a, ARCHIVE_ERRNO_MISC, "Unknown ACL tag"); ret = ARCHIVE_FAILED; goto exit_free; } #ifdef ACL_TYPE_NFS4 r = 0; switch (ae_type) { case ARCHIVE_ENTRY_ACL_TYPE_ALLOW: r = acl_set_entry_type_np(acl_entry, ACL_ENTRY_TYPE_ALLOW); break; case ARCHIVE_ENTRY_ACL_TYPE_DENY: r = acl_set_entry_type_np(acl_entry, ACL_ENTRY_TYPE_DENY); break; case ARCHIVE_ENTRY_ACL_TYPE_AUDIT: r = acl_set_entry_type_np(acl_entry, ACL_ENTRY_TYPE_AUDIT); break; case ARCHIVE_ENTRY_ACL_TYPE_ALARM: r = acl_set_entry_type_np(acl_entry, ACL_ENTRY_TYPE_ALARM); break; case ARCHIVE_ENTRY_ACL_TYPE_ACCESS: case ARCHIVE_ENTRY_ACL_TYPE_DEFAULT: // These don't translate directly into the system ACL. break; default: archive_set_error(a, ARCHIVE_ERRNO_MISC, "Unknown ACL entry type"); ret = ARCHIVE_FAILED; goto exit_free; } if (r != 0) { archive_set_error(a, errno, "Failed to set ACL entry type"); ret = ARCHIVE_FAILED; goto exit_free; } #endif if (acl_get_permset(acl_entry, &acl_permset) != 0) { archive_set_error(a, errno, "Failed to get ACL permission set"); ret = ARCHIVE_FAILED; goto exit_free; } if (acl_clear_perms(acl_permset) != 0) { archive_set_error(a, errno, "Failed to clear ACL permissions"); ret = ARCHIVE_FAILED; goto exit_free; } for (i = 0; i < (int)(sizeof(acl_perm_map) / sizeof(acl_perm_map[0])); ++i) { if (ae_permset & acl_perm_map[i].archive_perm) if (acl_add_perm(acl_permset, acl_perm_map[i].platform_perm) != 0) { archive_set_error(a, errno, "Failed to add ACL permission"); ret = ARCHIVE_FAILED; goto exit_free; } } #ifdef ACL_TYPE_NFS4 if (acl_type == ACL_TYPE_NFS4) { /* * acl_get_flagset_np() fails with non-NFSv4 ACLs */ if (acl_get_flagset_np(acl_entry, &acl_flagset) != 0) { archive_set_error(a, errno, "Failed to get flagset from an NFSv4 ACL entry"); ret = ARCHIVE_FAILED; goto exit_free; } if (acl_clear_flags_np(acl_flagset) != 0) { archive_set_error(a, errno, "Failed to clear flags from an NFSv4 ACL flagset"); ret = ARCHIVE_FAILED; goto exit_free; } for (i = 0; i < (int)(sizeof(acl_inherit_map) / sizeof(acl_inherit_map[0])); ++i) { if (ae_permset & acl_inherit_map[i].archive_inherit) { if (acl_add_flag_np(acl_flagset, acl_inherit_map[i].platform_inherit) != 0) { archive_set_error(a, errno, "Failed to add flag to NFSv4 ACL flagset"); ret = ARCHIVE_FAILED; goto exit_free; } } } } #endif } /* Try restoring the ACL through 'fd' if we can. */ -#if HAVE_ACL_SET_FD - if (fd >= 0 && acl_type == ACL_TYPE_ACCESS && acl_set_fd(fd, acl) == 0) - ret = ARCHIVE_OK; - else -#else +#if HAVE_ACL_SET_FD_NP || HAVE_ACL_SET_FD #if HAVE_ACL_SET_FD_NP - if (fd >= 0 && acl_set_fd_np(fd, acl, acl_type) == 0) - ret = ARCHIVE_OK; - else + if (fd >= 0) { + if (acl_set_fd_np(fd, acl, acl_type) == 0) +#else /* HAVE_ACL_SET_FD */ + if (fd >= 0 && acl_type == ACL_TYPE_ACCESS) { + if (acl_set_fd(fd, acl) == 0) #endif -#endif + ret = ARCHIVE_OK; + else { + if (errno == EOPNOTSUPP) { + /* Filesystem doesn't support ACLs */ + ret = ARCHIVE_OK; + } else { + archive_set_error(a, errno, + "Failed to set %s acl on fd", tname); + } + } + } else +#endif /* HAVE_ACL_SET_FD_NP || HAVE_ACL_SET_FD */ #if HAVE_ACL_SET_LINK_NP - if (acl_set_link_np(name, acl_type, acl) != 0) { - archive_set_error(a, errno, "Failed to set %s acl", tname); - ret = ARCHIVE_WARN; - } + if (acl_set_link_np(name, acl_type, acl) != 0) { #else /* TODO: Skip this if 'name' is a symlink. */ if (acl_set_file(name, acl_type, acl) != 0) { - archive_set_error(a, errno, "Failed to set %s acl", tname); - ret = ARCHIVE_WARN; - } #endif + if (errno == EOPNOTSUPP) { + /* Filesystem doesn't support ACLs */ + ret = ARCHIVE_OK; + } else { + archive_set_error(a, errno, "Failed to set %s acl", + tname); + ret = ARCHIVE_WARN; + } + } exit_free: acl_free(acl); return (ret); } #endif Index: head/contrib/libarchive/libarchive/archive_write_set_format_7zip.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_set_format_7zip.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_write_set_format_7zip.c (revision 311903) @@ -1,2328 +1,2328 @@ /*- * Copyright (c) 2011-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #include #ifdef HAVE_BZLIB_H #include #endif #if HAVE_LZMA_H #include #endif #ifdef HAVE_ZLIB_H #include #endif #include "archive.h" #ifndef HAVE_ZLIB_H #include "archive_crc32.h" #endif #include "archive_endian.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_ppmd7_private.h" #include "archive_private.h" #include "archive_rb.h" #include "archive_string.h" #include "archive_write_private.h" /* * Codec ID */ #define _7Z_COPY 0 #define _7Z_LZMA1 0x030101 #define _7Z_LZMA2 0x21 #define _7Z_DEFLATE 0x040108 #define _7Z_BZIP2 0x040202 #define _7Z_PPMD 0x030401 /* * 7-Zip header property IDs. */ #define kEnd 0x00 #define kHeader 0x01 #define kArchiveProperties 0x02 #define kAdditionalStreamsInfo 0x03 #define kMainStreamsInfo 0x04 #define kFilesInfo 0x05 #define kPackInfo 0x06 #define kUnPackInfo 0x07 #define kSubStreamsInfo 0x08 #define kSize 0x09 #define kCRC 0x0A #define kFolder 0x0B #define kCodersUnPackSize 0x0C #define kNumUnPackStream 0x0D #define kEmptyStream 0x0E #define kEmptyFile 0x0F #define kAnti 0x10 #define kName 0x11 #define kCTime 0x12 #define kATime 0x13 #define kMTime 0x14 #define kAttributes 0x15 #define kEncodedHeader 0x17 enum la_zaction { ARCHIVE_Z_FINISH, ARCHIVE_Z_RUN }; /* * A stream object of universal compressor. */ struct la_zstream { const uint8_t *next_in; size_t avail_in; uint64_t total_in; uint8_t *next_out; size_t avail_out; uint64_t total_out; uint32_t prop_size; uint8_t *props; int valid; void *real_stream; int (*code) (struct archive *a, struct la_zstream *lastrm, enum la_zaction action); int (*end)(struct archive *a, struct la_zstream *lastrm); }; #define PPMD7_DEFAULT_ORDER 6 #define PPMD7_DEFAULT_MEM_SIZE (1 << 24) struct ppmd_stream { int stat; CPpmd7 ppmd7_context; CPpmd7z_RangeEnc range_enc; IByteOut byteout; uint8_t *buff; uint8_t *buff_ptr; uint8_t *buff_end; size_t buff_bytes; }; struct coder { unsigned codec; size_t prop_size; uint8_t *props; }; struct file { struct archive_rb_node rbnode; struct file *next; unsigned name_len; uint8_t *utf16name;/* UTF16-LE name. */ uint64_t size; unsigned flg; #define MTIME_IS_SET (1<<0) #define ATIME_IS_SET (1<<1) #define CTIME_IS_SET (1<<2) #define CRC32_IS_SET (1<<3) #define HAS_STREAM (1<<4) struct { time_t time; long time_ns; } times[3]; #define MTIME 0 #define ATIME 1 #define CTIME 2 mode_t mode; uint32_t crc32; int dir:1; }; struct _7zip { int temp_fd; uint64_t temp_offset; struct file *cur_file; size_t total_number_entry; size_t total_number_nonempty_entry; size_t total_number_empty_entry; size_t total_number_dir_entry; size_t total_bytes_entry_name; size_t total_number_time_defined[3]; uint64_t total_bytes_compressed; uint64_t total_bytes_uncompressed; uint64_t entry_bytes_remaining; uint32_t entry_crc32; uint32_t precode_crc32; uint32_t encoded_crc32; int crc32flg; #define PRECODE_CRC32 1 #define ENCODED_CRC32 2 unsigned opt_compression; int opt_compression_level; struct la_zstream stream; struct coder coder; struct archive_string_conv *sconv; /* * Compressed data buffer. */ unsigned char wbuff[512 * 20 * 6]; size_t wbuff_remaining; /* * The list of the file entries which has its contents is used to * manage struct file objects. * We use 'next' (a member of struct file) to chain. */ struct { struct file *first; struct file **last; } file_list, empty_list; struct archive_rb_tree rbtree;/* for empty files */ }; static int _7z_options(struct archive_write *, const char *, const char *); static int _7z_write_header(struct archive_write *, struct archive_entry *); static ssize_t _7z_write_data(struct archive_write *, const void *, size_t); static int _7z_finish_entry(struct archive_write *); static int _7z_close(struct archive_write *); static int _7z_free(struct archive_write *); static int file_cmp_node(const struct archive_rb_node *, const struct archive_rb_node *); static int file_cmp_key(const struct archive_rb_node *, const void *); static int file_new(struct archive_write *a, struct archive_entry *, struct file **); static void file_free(struct file *); static void file_register(struct _7zip *, struct file *); static void file_register_empty(struct _7zip *, struct file *); static void file_init_register(struct _7zip *); static void file_init_register_empty(struct _7zip *); static void file_free_register(struct _7zip *); static ssize_t compress_out(struct archive_write *, const void *, size_t , enum la_zaction); static int compression_init_encoder_copy(struct archive *, struct la_zstream *); static int compression_code_copy(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_copy(struct archive *, struct la_zstream *); static int compression_init_encoder_deflate(struct archive *, struct la_zstream *, int, int); #ifdef HAVE_ZLIB_H static int compression_code_deflate(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_deflate(struct archive *, struct la_zstream *); #endif static int compression_init_encoder_bzip2(struct archive *, struct la_zstream *, int); #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) static int compression_code_bzip2(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_bzip2(struct archive *, struct la_zstream *); #endif static int compression_init_encoder_lzma1(struct archive *, struct la_zstream *, int); static int compression_init_encoder_lzma2(struct archive *, struct la_zstream *, int); #if defined(HAVE_LZMA_H) static int compression_code_lzma(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_lzma(struct archive *, struct la_zstream *); #endif static int compression_init_encoder_ppmd(struct archive *, struct la_zstream *, unsigned, uint32_t); static int compression_code_ppmd(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_ppmd(struct archive *, struct la_zstream *); static int _7z_compression_init_encoder(struct archive_write *, unsigned, int); static int compression_code(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end(struct archive *, struct la_zstream *); static int enc_uint64(struct archive_write *, uint64_t); static int make_header(struct archive_write *, uint64_t, uint64_t, uint64_t, int, struct coder *); static int make_streamsInfo(struct archive_write *, uint64_t, uint64_t, uint64_t, int, struct coder *, int, uint32_t); int archive_write_set_format_7zip(struct archive *_a) { static const struct archive_rb_tree_ops rb_ops = { file_cmp_node, file_cmp_key }; struct archive_write *a = (struct archive_write *)_a; struct _7zip *zip; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_7zip"); /* If another format was already registered, unregister it. */ if (a->format_free != NULL) (a->format_free)(a); zip = calloc(1, sizeof(*zip)); if (zip == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate 7-Zip data"); return (ARCHIVE_FATAL); } zip->temp_fd = -1; __archive_rb_tree_init(&(zip->rbtree), &rb_ops); file_init_register(zip); file_init_register_empty(zip); /* Set default compression type and its level. */ #if HAVE_LZMA_H zip->opt_compression = _7Z_LZMA1; #elif defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) zip->opt_compression = _7Z_BZIP2; #elif defined(HAVE_ZLIB_H) zip->opt_compression = _7Z_DEFLATE; #else zip->opt_compression = _7Z_COPY; #endif zip->opt_compression_level = 6; a->format_data = zip; a->format_name = "7zip"; a->format_options = _7z_options; a->format_write_header = _7z_write_header; a->format_write_data = _7z_write_data; a->format_finish_entry = _7z_finish_entry; a->format_close = _7z_close; a->format_free = _7z_free; a->archive.archive_format = ARCHIVE_FORMAT_7ZIP; a->archive.archive_format_name = "7zip"; return (ARCHIVE_OK); } static int _7z_options(struct archive_write *a, const char *key, const char *value) { struct _7zip *zip; zip = (struct _7zip *)a->format_data; if (strcmp(key, "compression") == 0) { const char *name = NULL; if (value == NULL || strcmp(value, "copy") == 0 || strcmp(value, "COPY") == 0 || strcmp(value, "store") == 0 || strcmp(value, "STORE") == 0) zip->opt_compression = _7Z_COPY; else if (strcmp(value, "deflate") == 0 || strcmp(value, "DEFLATE") == 0) #if HAVE_ZLIB_H zip->opt_compression = _7Z_DEFLATE; #else name = "deflate"; #endif else if (strcmp(value, "bzip2") == 0 || strcmp(value, "BZIP2") == 0) #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) zip->opt_compression = _7Z_BZIP2; #else name = "bzip2"; #endif else if (strcmp(value, "lzma1") == 0 || strcmp(value, "LZMA1") == 0) #if HAVE_LZMA_H zip->opt_compression = _7Z_LZMA1; #else name = "lzma1"; #endif else if (strcmp(value, "lzma2") == 0 || strcmp(value, "LZMA2") == 0) #if HAVE_LZMA_H zip->opt_compression = _7Z_LZMA2; #else name = "lzma2"; #endif else if (strcmp(value, "ppmd") == 0 || strcmp(value, "PPMD") == 0 || strcmp(value, "PPMd") == 0) zip->opt_compression = _7Z_PPMD; else { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Unknown compression name: `%s'", value); return (ARCHIVE_FAILED); } if (name != NULL) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "`%s' compression not supported " "on this platform", name); return (ARCHIVE_FAILED); } return (ARCHIVE_OK); } if (strcmp(key, "compression-level") == 0) { if (value == NULL || !(value[0] >= '0' && value[0] <= '9') || value[1] != '\0') { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Illegal value `%s'", value); return (ARCHIVE_FAILED); } zip->opt_compression_level = value[0] - '0'; return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int _7z_write_header(struct archive_write *a, struct archive_entry *entry) { struct _7zip *zip; struct file *file; int r; zip = (struct _7zip *)a->format_data; zip->cur_file = NULL; zip->entry_bytes_remaining = 0; if (zip->sconv == NULL) { zip->sconv = archive_string_conversion_to_charset( &a->archive, "UTF-16LE", 1); if (zip->sconv == NULL) return (ARCHIVE_FATAL); } r = file_new(a, entry, &file); if (r < ARCHIVE_WARN) { file_free(file); return (r); } if (file->size == 0 && file->dir) { if (!__archive_rb_tree_insert_node(&(zip->rbtree), (struct archive_rb_node *)file)) { /* We have already had the same file. */ file_free(file); return (ARCHIVE_OK); } } if (file->flg & MTIME_IS_SET) zip->total_number_time_defined[MTIME]++; if (file->flg & CTIME_IS_SET) zip->total_number_time_defined[CTIME]++; if (file->flg & ATIME_IS_SET) zip->total_number_time_defined[ATIME]++; zip->total_number_entry++; zip->total_bytes_entry_name += file->name_len + 2; if (file->size == 0) { /* Count up the number of empty files. */ zip->total_number_empty_entry++; if (file->dir) zip->total_number_dir_entry++; else file_register_empty(zip, file); return (r); } /* * Init compression. */ if ((zip->total_number_entry - zip->total_number_empty_entry) == 1) { r = _7z_compression_init_encoder(a, zip->opt_compression, zip->opt_compression_level); if (r < 0) { file_free(file); return (ARCHIVE_FATAL); } } /* Register a non-empty file. */ file_register(zip, file); /* * Set the current file to cur_file to read its contents. */ zip->cur_file = file; /* Save a offset of current file in temporary file. */ zip->entry_bytes_remaining = file->size; zip->entry_crc32 = 0; /* * Store a symbolic link name as file contents. */ if (archive_entry_filetype(entry) == AE_IFLNK) { ssize_t bytes; const void *p = (const void *)archive_entry_symlink(entry); bytes = compress_out(a, p, (size_t)file->size, ARCHIVE_Z_RUN); if (bytes < 0) return ((int)bytes); zip->entry_crc32 = crc32(zip->entry_crc32, p, (unsigned)bytes); zip->entry_bytes_remaining -= bytes; } return (r); } /* * Write data to a temporary file. */ static int write_to_temp(struct archive_write *a, const void *buff, size_t s) { struct _7zip *zip; const unsigned char *p; ssize_t ws; zip = (struct _7zip *)a->format_data; /* * Open a temporary file. */ if (zip->temp_fd == -1) { zip->temp_offset = 0; zip->temp_fd = __archive_mktemp(NULL); if (zip->temp_fd < 0) { archive_set_error(&a->archive, errno, "Couldn't create temporary file"); return (ARCHIVE_FATAL); } } p = (const unsigned char *)buff; while (s) { ws = write(zip->temp_fd, p, s); if (ws < 0) { archive_set_error(&(a->archive), errno, "fwrite function failed"); return (ARCHIVE_FATAL); } s -= ws; p += ws; zip->temp_offset += ws; } return (ARCHIVE_OK); } static ssize_t compress_out(struct archive_write *a, const void *buff, size_t s, enum la_zaction run) { struct _7zip *zip = (struct _7zip *)a->format_data; int r; if (run == ARCHIVE_Z_FINISH && zip->stream.total_in == 0 && s == 0) return (0); if ((zip->crc32flg & PRECODE_CRC32) && s) zip->precode_crc32 = crc32(zip->precode_crc32, buff, (unsigned)s); zip->stream.next_in = (const unsigned char *)buff; zip->stream.avail_in = s; for (;;) { /* Compress file data. */ r = compression_code(&(a->archive), &(zip->stream), run); if (r != ARCHIVE_OK && r != ARCHIVE_EOF) return (ARCHIVE_FATAL); if (zip->stream.avail_out == 0) { if (write_to_temp(a, zip->wbuff, sizeof(zip->wbuff)) != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->stream.next_out = zip->wbuff; zip->stream.avail_out = sizeof(zip->wbuff); if (zip->crc32flg & ENCODED_CRC32) zip->encoded_crc32 = crc32(zip->encoded_crc32, zip->wbuff, sizeof(zip->wbuff)); if (run == ARCHIVE_Z_FINISH && r != ARCHIVE_EOF) continue; } if (zip->stream.avail_in == 0) break; } if (run == ARCHIVE_Z_FINISH) { uint64_t bytes = sizeof(zip->wbuff) - zip->stream.avail_out; if (write_to_temp(a, zip->wbuff, (size_t)bytes) != ARCHIVE_OK) return (ARCHIVE_FATAL); if ((zip->crc32flg & ENCODED_CRC32) && bytes) zip->encoded_crc32 = crc32(zip->encoded_crc32, zip->wbuff, (unsigned)bytes); } return (s); } static ssize_t _7z_write_data(struct archive_write *a, const void *buff, size_t s) { struct _7zip *zip; ssize_t bytes; zip = (struct _7zip *)a->format_data; if (s > zip->entry_bytes_remaining) s = (size_t)zip->entry_bytes_remaining; if (s == 0 || zip->cur_file == NULL) return (0); bytes = compress_out(a, buff, s, ARCHIVE_Z_RUN); if (bytes < 0) return (bytes); zip->entry_crc32 = crc32(zip->entry_crc32, buff, (unsigned)bytes); zip->entry_bytes_remaining -= bytes; return (bytes); } static int _7z_finish_entry(struct archive_write *a) { struct _7zip *zip; size_t s; ssize_t r; zip = (struct _7zip *)a->format_data; if (zip->cur_file == NULL) return (ARCHIVE_OK); while (zip->entry_bytes_remaining > 0) { s = (size_t)zip->entry_bytes_remaining; if (s > a->null_length) s = a->null_length; r = _7z_write_data(a, a->nulls, s); if (r < 0) return ((int)r); } zip->total_bytes_compressed += zip->stream.total_in; zip->total_bytes_uncompressed += zip->stream.total_out; zip->cur_file->crc32 = zip->entry_crc32; zip->cur_file = NULL; return (ARCHIVE_OK); } static int flush_wbuff(struct archive_write *a) { struct _7zip *zip; int r; size_t s; zip = (struct _7zip *)a->format_data; s = sizeof(zip->wbuff) - zip->wbuff_remaining; r = __archive_write_output(a, zip->wbuff, s); if (r != ARCHIVE_OK) return (r); zip->wbuff_remaining = sizeof(zip->wbuff); return (r); } static int copy_out(struct archive_write *a, uint64_t offset, uint64_t length) { struct _7zip *zip; int r; zip = (struct _7zip *)a->format_data; if (zip->temp_offset > 0 && lseek(zip->temp_fd, offset, SEEK_SET) < 0) { archive_set_error(&(a->archive), errno, "lseek failed"); return (ARCHIVE_FATAL); } while (length) { size_t rsize; ssize_t rs; unsigned char *wb; if (length > zip->wbuff_remaining) rsize = zip->wbuff_remaining; else rsize = (size_t)length; wb = zip->wbuff + (sizeof(zip->wbuff) - zip->wbuff_remaining); rs = read(zip->temp_fd, wb, rsize); if (rs < 0) { archive_set_error(&(a->archive), errno, "Can't read temporary file(%jd)", (intmax_t)rs); return (ARCHIVE_FATAL); } if (rs == 0) { archive_set_error(&(a->archive), 0, "Truncated 7-Zip archive"); return (ARCHIVE_FATAL); } zip->wbuff_remaining -= rs; length -= rs; if (zip->wbuff_remaining == 0) { r = flush_wbuff(a); if (r != ARCHIVE_OK) return (r); } } return (ARCHIVE_OK); } static int _7z_close(struct archive_write *a) { struct _7zip *zip; unsigned char *wb; uint64_t header_offset, header_size, header_unpacksize; uint64_t length; uint32_t header_crc32; int r; zip = (struct _7zip *)a->format_data; if (zip->total_number_entry > 0) { struct archive_rb_node *n; uint64_t data_offset, data_size, data_unpacksize; unsigned header_compression; r = (int)compress_out(a, NULL, 0, ARCHIVE_Z_FINISH); if (r < 0) return (r); data_offset = 0; data_size = zip->stream.total_out; data_unpacksize = zip->stream.total_in; zip->coder.codec = zip->opt_compression; zip->coder.prop_size = zip->stream.prop_size; zip->coder.props = zip->stream.props; zip->stream.prop_size = 0; zip->stream.props = NULL; zip->total_number_nonempty_entry = zip->total_number_entry - zip->total_number_empty_entry; /* Connect an empty file list. */ if (zip->empty_list.first != NULL) { *zip->file_list.last = zip->empty_list.first; zip->file_list.last = zip->empty_list.last; } /* Connect a directory file list. */ ARCHIVE_RB_TREE_FOREACH(n, &(zip->rbtree)) { file_register(zip, (struct file *)n); } /* * NOTE: 7z command supports just LZMA1, LZMA2 and COPY for * the compression type for encoding the header. */ #if HAVE_LZMA_H header_compression = _7Z_LZMA1; /* If the stored file is only one, do not encode the header. * This is the same way 7z command does. */ if (zip->total_number_entry == 1) header_compression = _7Z_COPY; #else header_compression = _7Z_COPY; #endif r = _7z_compression_init_encoder(a, header_compression, 6); if (r < 0) return (r); zip->crc32flg = PRECODE_CRC32; zip->precode_crc32 = 0; r = make_header(a, data_offset, data_size, data_unpacksize, 1, &(zip->coder)); if (r < 0) return (r); r = (int)compress_out(a, NULL, 0, ARCHIVE_Z_FINISH); if (r < 0) return (r); header_offset = data_offset + data_size; header_size = zip->stream.total_out; header_crc32 = zip->precode_crc32; header_unpacksize = zip->stream.total_in; if (header_compression != _7Z_COPY) { /* * Encode the header in order to reduce the size * of the archive. */ free(zip->coder.props); zip->coder.codec = header_compression; zip->coder.prop_size = zip->stream.prop_size; zip->coder.props = zip->stream.props; zip->stream.prop_size = 0; zip->stream.props = NULL; r = _7z_compression_init_encoder(a, _7Z_COPY, 0); if (r < 0) return (r); zip->crc32flg = ENCODED_CRC32; zip->encoded_crc32 = 0; /* * Make EncodedHeader. */ r = enc_uint64(a, kEncodedHeader); if (r < 0) return (r); r = make_streamsInfo(a, header_offset, header_size, header_unpacksize, 1, &(zip->coder), 0, header_crc32); if (r < 0) return (r); r = (int)compress_out(a, NULL, 0, ARCHIVE_Z_FINISH); if (r < 0) return (r); header_offset = header_offset + header_size; header_size = zip->stream.total_out; header_crc32 = zip->encoded_crc32; } zip->crc32flg = 0; } else { header_offset = header_size = 0; header_crc32 = 0; } length = zip->temp_offset; /* * Make the zip header on wbuff(write buffer). */ wb = zip->wbuff; zip->wbuff_remaining = sizeof(zip->wbuff); memcpy(&wb[0], "7z\xBC\xAF\x27\x1C", 6); wb[6] = 0;/* Major version. */ wb[7] = 3;/* Minor version. */ archive_le64enc(&wb[12], header_offset);/* Next Header Offset */ archive_le64enc(&wb[20], header_size);/* Next Header Size */ archive_le32enc(&wb[28], header_crc32);/* Next Header CRC */ archive_le32enc(&wb[8], crc32(0, &wb[12], 20));/* Start Header CRC */ zip->wbuff_remaining -= 32; /* * Read all file contents and an encoded header from the temporary * file and write out it. */ r = copy_out(a, 0, length); if (r != ARCHIVE_OK) return (r); r = flush_wbuff(a); return (r); } /* * Encode 64 bits value into 7-Zip's encoded UINT64 value. */ static int enc_uint64(struct archive_write *a, uint64_t val) { unsigned mask = 0x80; uint8_t numdata[9]; int i; numdata[0] = 0; for (i = 1; i < (int)sizeof(numdata); i++) { if (val < mask) { numdata[0] |= (uint8_t)val; break; } numdata[i] = (uint8_t)val; val >>= 8; numdata[0] |= mask; mask >>= 1; } return ((int)compress_out(a, numdata, i, ARCHIVE_Z_RUN)); } static int make_substreamsInfo(struct archive_write *a, struct coder *coders) { struct _7zip *zip = (struct _7zip *)a->format_data; struct file *file; int r; /* * Make SubStreamsInfo. */ r = enc_uint64(a, kSubStreamsInfo); if (r < 0) return (r); if (zip->total_number_nonempty_entry > 1 && coders->codec != _7Z_COPY) { /* * Make NumUnPackStream. */ r = enc_uint64(a, kNumUnPackStream); if (r < 0) return (r); /* Write numUnpackStreams */ r = enc_uint64(a, zip->total_number_nonempty_entry); if (r < 0) return (r); /* * Make kSize. */ r = enc_uint64(a, kSize); if (r < 0) return (r); file = zip->file_list.first; for (;file != NULL; file = file->next) { if (file->next == NULL || file->next->size == 0) break; r = enc_uint64(a, file->size); if (r < 0) return (r); } } /* * Make CRC. */ r = enc_uint64(a, kCRC); if (r < 0) return (r); /* All are defined */ r = enc_uint64(a, 1); if (r < 0) return (r); file = zip->file_list.first; for (;file != NULL; file = file->next) { uint8_t crc[4]; if (file->size == 0) break; archive_le32enc(crc, file->crc32); r = (int)compress_out(a, crc, 4, ARCHIVE_Z_RUN); if (r < 0) return (r); } /* Write End. */ r = enc_uint64(a, kEnd); if (r < 0) return (r); return (ARCHIVE_OK); } static int make_streamsInfo(struct archive_write *a, uint64_t offset, uint64_t pack_size, uint64_t unpack_size, int num_coder, struct coder *coders, int substrm, uint32_t header_crc) { struct _7zip *zip = (struct _7zip *)a->format_data; uint8_t codec_buff[8]; int numFolders, fi; int codec_size; int i, r; if (coders->codec == _7Z_COPY) numFolders = (int)zip->total_number_nonempty_entry; else numFolders = 1; /* * Make PackInfo. */ r = enc_uint64(a, kPackInfo); if (r < 0) return (r); /* Write PackPos. */ r = enc_uint64(a, offset); if (r < 0) return (r); /* Write NumPackStreams. */ r = enc_uint64(a, numFolders); if (r < 0) return (r); /* Make Size. */ r = enc_uint64(a, kSize); if (r < 0) return (r); if (numFolders > 1) { struct file *file = zip->file_list.first; for (;file != NULL; file = file->next) { if (file->size == 0) break; r = enc_uint64(a, file->size); if (r < 0) return (r); } } else { /* Write size. */ r = enc_uint64(a, pack_size); if (r < 0) return (r); } r = enc_uint64(a, kEnd); if (r < 0) return (r); /* * Make UnPackInfo. */ r = enc_uint64(a, kUnPackInfo); if (r < 0) return (r); /* * Make Folder. */ r = enc_uint64(a, kFolder); if (r < 0) return (r); /* Write NumFolders. */ r = enc_uint64(a, numFolders); if (r < 0) return (r); /* Write External. */ r = enc_uint64(a, 0); if (r < 0) return (r); for (fi = 0; fi < numFolders; fi++) { /* Write NumCoders. */ r = enc_uint64(a, num_coder); if (r < 0) return (r); for (i = 0; i < num_coder; i++) { unsigned codec_id = coders[i].codec; /* Write Codec flag. */ archive_be64enc(codec_buff, codec_id); for (codec_size = 8; codec_size > 0; codec_size--) { if (codec_buff[8 - codec_size]) break; } if (codec_size == 0) codec_size = 1; if (coders[i].prop_size) r = enc_uint64(a, codec_size | 0x20); else r = enc_uint64(a, codec_size); if (r < 0) return (r); /* Write Codec ID. */ codec_size &= 0x0f; r = (int)compress_out(a, &codec_buff[8-codec_size], codec_size, ARCHIVE_Z_RUN); if (r < 0) return (r); if (coders[i].prop_size) { /* Write Codec property size. */ r = enc_uint64(a, coders[i].prop_size); if (r < 0) return (r); /* Write Codec properties. */ r = (int)compress_out(a, coders[i].props, coders[i].prop_size, ARCHIVE_Z_RUN); if (r < 0) return (r); } } } /* * Make CodersUnPackSize. */ r = enc_uint64(a, kCodersUnPackSize); if (r < 0) return (r); if (numFolders > 1) { struct file *file = zip->file_list.first; for (;file != NULL; file = file->next) { if (file->size == 0) break; r = enc_uint64(a, file->size); if (r < 0) return (r); } } else { /* Write UnPackSize. */ r = enc_uint64(a, unpack_size); if (r < 0) return (r); } if (!substrm) { uint8_t crc[4]; /* * Make CRC. */ r = enc_uint64(a, kCRC); if (r < 0) return (r); /* All are defined */ r = enc_uint64(a, 1); if (r < 0) return (r); archive_le32enc(crc, header_crc); r = (int)compress_out(a, crc, 4, ARCHIVE_Z_RUN); if (r < 0) return (r); } /* Write End. */ r = enc_uint64(a, kEnd); if (r < 0) return (r); if (substrm) { /* * Make SubStreamsInfo. */ r = make_substreamsInfo(a, coders); if (r < 0) return (r); } /* Write End. */ r = enc_uint64(a, kEnd); if (r < 0) return (r); return (ARCHIVE_OK); } #define EPOC_TIME ARCHIVE_LITERAL_ULL(116444736000000000) static uint64_t utcToFiletime(time_t t, long ns) { uint64_t fileTime; fileTime = t; fileTime *= 10000000; fileTime += ns / 100; fileTime += EPOC_TIME; return (fileTime); } static int make_time(struct archive_write *a, uint8_t type, unsigned flg, int ti) { uint8_t filetime[8]; struct _7zip *zip = (struct _7zip *)a->format_data; struct file *file; int r; uint8_t b, mask; /* * Make Time Bools. */ if (zip->total_number_time_defined[ti] == zip->total_number_entry) { /* Write Time Type. */ r = enc_uint64(a, type); if (r < 0) return (r); /* Write EmptyStream Size. */ r = enc_uint64(a, 2 + zip->total_number_entry * 8); if (r < 0) return (r); /* All are defined. */ r = enc_uint64(a, 1); if (r < 0) return (r); } else { if (zip->total_number_time_defined[ti] == 0) return (ARCHIVE_OK); /* Write Time Type. */ r = enc_uint64(a, type); if (r < 0) return (r); /* Write EmptyStream Size. */ r = enc_uint64(a, 2 + ((zip->total_number_entry + 7) >> 3) + zip->total_number_time_defined[ti] * 8); if (r < 0) return (r); /* All are not defined. */ r = enc_uint64(a, 0); if (r < 0) return (r); b = 0; mask = 0x80; file = zip->file_list.first; for (;file != NULL; file = file->next) { if (file->flg & flg) b |= mask; mask >>= 1; if (mask == 0) { r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); if (r < 0) return (r); mask = 0x80; b = 0; } } if (mask != 0x80) { r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); if (r < 0) return (r); } } /* External. */ r = enc_uint64(a, 0); if (r < 0) return (r); /* * Make Times. */ file = zip->file_list.first; for (;file != NULL; file = file->next) { if ((file->flg & flg) == 0) continue; archive_le64enc(filetime, utcToFiletime(file->times[ti].time, file->times[ti].time_ns)); r = (int)compress_out(a, filetime, 8, ARCHIVE_Z_RUN); if (r < 0) return (r); } return (ARCHIVE_OK); } static int make_header(struct archive_write *a, uint64_t offset, uint64_t pack_size, uint64_t unpack_size, int codernum, struct coder *coders) { struct _7zip *zip = (struct _7zip *)a->format_data; struct file *file; int r; uint8_t b, mask; /* * Make FilesInfo. */ r = enc_uint64(a, kHeader); if (r < 0) return (r); /* * If there are empty files only, do not write MainStreamInfo. */ if (zip->total_number_nonempty_entry) { /* * Make MainStreamInfo. */ r = enc_uint64(a, kMainStreamsInfo); if (r < 0) return (r); r = make_streamsInfo(a, offset, pack_size, unpack_size, codernum, coders, 1, 0); if (r < 0) return (r); } /* * Make FilesInfo. */ r = enc_uint64(a, kFilesInfo); if (r < 0) return (r); /* Write numFiles. */ r = enc_uint64(a, zip->total_number_entry); if (r < 0) return (r); if (zip->total_number_empty_entry > 0) { /* Make EmptyStream. */ r = enc_uint64(a, kEmptyStream); if (r < 0) return (r); /* Write EmptyStream Size. */ r = enc_uint64(a, (zip->total_number_entry+7)>>3); if (r < 0) return (r); b = 0; mask = 0x80; file = zip->file_list.first; for (;file != NULL; file = file->next) { if (file->size == 0) b |= mask; mask >>= 1; if (mask == 0) { r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); if (r < 0) return (r); mask = 0x80; b = 0; } } if (mask != 0x80) { r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); if (r < 0) return (r); } } if (zip->total_number_empty_entry > zip->total_number_dir_entry) { /* Make EmptyFile. */ r = enc_uint64(a, kEmptyFile); if (r < 0) return (r); /* Write EmptyFile Size. */ r = enc_uint64(a, (zip->total_number_empty_entry + 7) >> 3); if (r < 0) return (r); b = 0; mask = 0x80; file = zip->file_list.first; for (;file != NULL; file = file->next) { if (file->size) continue; if (!file->dir) b |= mask; mask >>= 1; if (mask == 0) { r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); if (r < 0) return (r); mask = 0x80; b = 0; } } if (mask != 0x80) { r = (int)compress_out(a, &b, 1, ARCHIVE_Z_RUN); if (r < 0) return (r); } } /* Make Name. */ r = enc_uint64(a, kName); if (r < 0) return (r); - /* Write Nume size. */ + /* Write Name size. */ r = enc_uint64(a, zip->total_bytes_entry_name+1); if (r < 0) return (r); /* Write dmy byte. */ r = enc_uint64(a, 0); if (r < 0) return (r); file = zip->file_list.first; for (;file != NULL; file = file->next) { r = (int)compress_out(a, file->utf16name, file->name_len+2, ARCHIVE_Z_RUN); if (r < 0) return (r); } /* Make MTime. */ r = make_time(a, kMTime, MTIME_IS_SET, MTIME); if (r < 0) return (r); /* Make CTime. */ r = make_time(a, kCTime, CTIME_IS_SET, CTIME); if (r < 0) return (r); /* Make ATime. */ r = make_time(a, kATime, ATIME_IS_SET, ATIME); if (r < 0) return (r); /* Make Attributes. */ r = enc_uint64(a, kAttributes); if (r < 0) return (r); /* Write Attributes size. */ r = enc_uint64(a, 2 + zip->total_number_entry * 4); if (r < 0) return (r); /* Write "All Are Defined". */ r = enc_uint64(a, 1); if (r < 0) return (r); /* Write dmy byte. */ r = enc_uint64(a, 0); if (r < 0) return (r); file = zip->file_list.first; for (;file != NULL; file = file->next) { /* * High 16bits is unix mode. * Low 16bits is Windows attributes. */ uint32_t encattr, attr; if (file->dir) attr = 0x8010; else attr = 0x8020; if ((file->mode & 0222) == 0) attr |= 1;/* Read Only. */ attr |= ((uint32_t)file->mode) << 16; archive_le32enc(&encattr, attr); r = (int)compress_out(a, &encattr, 4, ARCHIVE_Z_RUN); if (r < 0) return (r); } /* Write End. */ r = enc_uint64(a, kEnd); if (r < 0) return (r); /* Write End. */ r = enc_uint64(a, kEnd); if (r < 0) return (r); return (ARCHIVE_OK); } static int _7z_free(struct archive_write *a) { struct _7zip *zip = (struct _7zip *)a->format_data; /* Close the temporary file. */ if (zip->temp_fd >= 0) close(zip->temp_fd); file_free_register(zip); compression_end(&(a->archive), &(zip->stream)); free(zip->coder.props); free(zip); return (ARCHIVE_OK); } static int file_cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) { const struct file *f1 = (const struct file *)n1; const struct file *f2 = (const struct file *)n2; if (f1->name_len == f2->name_len) return (memcmp(f1->utf16name, f2->utf16name, f1->name_len)); return (f1->name_len > f2->name_len)?1:-1; } static int file_cmp_key(const struct archive_rb_node *n, const void *key) { const struct file *f = (const struct file *)n; return (f->name_len - *(const char *)key); } static int file_new(struct archive_write *a, struct archive_entry *entry, struct file **newfile) { struct _7zip *zip; struct file *file; const char *u16; size_t u16len; int ret = ARCHIVE_OK; zip = (struct _7zip *)a->format_data; *newfile = NULL; file = calloc(1, sizeof(*file)); if (file == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } if (0 > archive_entry_pathname_l(entry, &u16, &u16len, zip->sconv)) { if (errno == ENOMEM) { free(file); archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for UTF-16LE"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "A filename cannot be converted to UTF-16LE;" "You should disable making Joliet extension"); ret = ARCHIVE_WARN; } file->utf16name = malloc(u16len + 2); if (file->utf16name == NULL) { free(file); archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Name"); return (ARCHIVE_FATAL); } memcpy(file->utf16name, u16, u16len); file->utf16name[u16len+0] = 0; file->utf16name[u16len+1] = 0; file->name_len = (unsigned)u16len; file->mode = archive_entry_mode(entry); if (archive_entry_filetype(entry) == AE_IFREG) file->size = archive_entry_size(entry); else archive_entry_set_size(entry, 0); if (archive_entry_filetype(entry) == AE_IFDIR) file->dir = 1; else if (archive_entry_filetype(entry) == AE_IFLNK) file->size = strlen(archive_entry_symlink(entry)); if (archive_entry_mtime_is_set(entry)) { file->flg |= MTIME_IS_SET; file->times[MTIME].time = archive_entry_mtime(entry); file->times[MTIME].time_ns = archive_entry_mtime_nsec(entry); } if (archive_entry_atime_is_set(entry)) { file->flg |= ATIME_IS_SET; file->times[ATIME].time = archive_entry_atime(entry); file->times[ATIME].time_ns = archive_entry_atime_nsec(entry); } if (archive_entry_ctime_is_set(entry)) { file->flg |= CTIME_IS_SET; file->times[CTIME].time = archive_entry_ctime(entry); file->times[CTIME].time_ns = archive_entry_ctime_nsec(entry); } *newfile = file; return (ret); } static void file_free(struct file *file) { free(file->utf16name); free(file); } static void file_register(struct _7zip *zip, struct file *file) { file->next = NULL; *zip->file_list.last = file; zip->file_list.last = &(file->next); } static void file_init_register(struct _7zip *zip) { zip->file_list.first = NULL; zip->file_list.last = &(zip->file_list.first); } static void file_free_register(struct _7zip *zip) { struct file *file, *file_next; file = zip->file_list.first; while (file != NULL) { file_next = file->next; file_free(file); file = file_next; } } static void file_register_empty(struct _7zip *zip, struct file *file) { file->next = NULL; *zip->empty_list.last = file; zip->empty_list.last = &(file->next); } static void file_init_register_empty(struct _7zip *zip) { zip->empty_list.first = NULL; zip->empty_list.last = &(zip->empty_list.first); } #if !defined(HAVE_ZLIB_H) || !defined(HAVE_BZLIB_H) ||\ !defined(BZ_CONFIG_ERROR) || !defined(HAVE_LZMA_H) static int compression_unsupported_encoder(struct archive *a, struct la_zstream *lastrm, const char *name) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "%s compression not supported on this platform", name); lastrm->valid = 0; lastrm->real_stream = NULL; return (ARCHIVE_FAILED); } #endif /* * _7_COPY compressor. */ static int compression_init_encoder_copy(struct archive *a, struct la_zstream *lastrm) { if (lastrm->valid) compression_end(a, lastrm); lastrm->valid = 1; lastrm->code = compression_code_copy; lastrm->end = compression_end_copy; return (ARCHIVE_OK); } static int compression_code_copy(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { size_t bytes; (void)a; /* UNUSED */ if (lastrm->avail_out > lastrm->avail_in) bytes = lastrm->avail_in; else bytes = lastrm->avail_out; if (bytes) { memcpy(lastrm->next_out, lastrm->next_in, bytes); lastrm->next_in += bytes; lastrm->avail_in -= bytes; lastrm->total_in += bytes; lastrm->next_out += bytes; lastrm->avail_out -= bytes; lastrm->total_out += bytes; } if (action == ARCHIVE_Z_FINISH && lastrm->avail_in == 0) return (ARCHIVE_EOF); return (ARCHIVE_OK); } static int compression_end_copy(struct archive *a, struct la_zstream *lastrm) { (void)a; /* UNUSED */ lastrm->valid = 0; return (ARCHIVE_OK); } /* * _7_DEFLATE compressor. */ #ifdef HAVE_ZLIB_H static int compression_init_encoder_deflate(struct archive *a, struct la_zstream *lastrm, int level, int withheader) { z_stream *strm; if (lastrm->valid) compression_end(a, lastrm); strm = calloc(1, sizeof(*strm)); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for gzip stream"); return (ARCHIVE_FATAL); } /* zlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (Bytef *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = (uInt)lastrm->avail_in; strm->total_in = (uLong)lastrm->total_in; strm->next_out = lastrm->next_out; strm->avail_out = (uInt)lastrm->avail_out; strm->total_out = (uLong)lastrm->total_out; if (deflateInit2(strm, level, Z_DEFLATED, (withheader)?15:-15, 8, Z_DEFAULT_STRATEGY) != Z_OK) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library"); return (ARCHIVE_FATAL); } lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_deflate; lastrm->end = compression_end_deflate; return (ARCHIVE_OK); } static int compression_code_deflate(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { z_stream *strm; int r; strm = (z_stream *)lastrm->real_stream; /* zlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (Bytef *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = (uInt)lastrm->avail_in; strm->total_in = (uLong)lastrm->total_in; strm->next_out = lastrm->next_out; strm->avail_out = (uInt)lastrm->avail_out; strm->total_out = (uLong)lastrm->total_out; r = deflate(strm, (action == ARCHIVE_Z_FINISH)? Z_FINISH: Z_NO_FLUSH); lastrm->next_in = strm->next_in; lastrm->avail_in = strm->avail_in; lastrm->total_in = strm->total_in; lastrm->next_out = strm->next_out; lastrm->avail_out = strm->avail_out; lastrm->total_out = strm->total_out; switch (r) { case Z_OK: return (ARCHIVE_OK); case Z_STREAM_END: return (ARCHIVE_EOF); default: archive_set_error(a, ARCHIVE_ERRNO_MISC, "GZip compression failed:" " deflate() call returned status %d", r); return (ARCHIVE_FATAL); } } static int compression_end_deflate(struct archive *a, struct la_zstream *lastrm) { z_stream *strm; int r; strm = (z_stream *)lastrm->real_stream; r = deflateEnd(strm); free(strm); lastrm->real_stream = NULL; lastrm->valid = 0; if (r != Z_OK) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "Failed to clean up compressor"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } #else static int compression_init_encoder_deflate(struct archive *a, struct la_zstream *lastrm, int level, int withheader) { (void) level; /* UNUSED */ (void) withheader; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "deflate")); } #endif /* * _7_BZIP2 compressor. */ #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) static int compression_init_encoder_bzip2(struct archive *a, struct la_zstream *lastrm, int level) { bz_stream *strm; if (lastrm->valid) compression_end(a, lastrm); strm = calloc(1, sizeof(*strm)); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for bzip2 stream"); return (ARCHIVE_FATAL); } /* bzlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (char *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = lastrm->avail_in; strm->total_in_lo32 = (uint32_t)(lastrm->total_in & 0xffffffff); strm->total_in_hi32 = (uint32_t)(lastrm->total_in >> 32); strm->next_out = (char *)lastrm->next_out; strm->avail_out = lastrm->avail_out; strm->total_out_lo32 = (uint32_t)(lastrm->total_out & 0xffffffff); strm->total_out_hi32 = (uint32_t)(lastrm->total_out >> 32); if (BZ2_bzCompressInit(strm, level, 0, 30) != BZ_OK) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library"); return (ARCHIVE_FATAL); } lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_bzip2; lastrm->end = compression_end_bzip2; return (ARCHIVE_OK); } static int compression_code_bzip2(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { bz_stream *strm; int r; strm = (bz_stream *)lastrm->real_stream; /* bzlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (char *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = lastrm->avail_in; strm->total_in_lo32 = (uint32_t)(lastrm->total_in & 0xffffffff); strm->total_in_hi32 = (uint32_t)(lastrm->total_in >> 32); strm->next_out = (char *)lastrm->next_out; strm->avail_out = lastrm->avail_out; strm->total_out_lo32 = (uint32_t)(lastrm->total_out & 0xffffffff); strm->total_out_hi32 = (uint32_t)(lastrm->total_out >> 32); r = BZ2_bzCompress(strm, (action == ARCHIVE_Z_FINISH)? BZ_FINISH: BZ_RUN); lastrm->next_in = (const unsigned char *)strm->next_in; lastrm->avail_in = strm->avail_in; lastrm->total_in = (((uint64_t)(uint32_t)strm->total_in_hi32) << 32) + (uint64_t)(uint32_t)strm->total_in_lo32; lastrm->next_out = (unsigned char *)strm->next_out; lastrm->avail_out = strm->avail_out; lastrm->total_out = (((uint64_t)(uint32_t)strm->total_out_hi32) << 32) + (uint64_t)(uint32_t)strm->total_out_lo32; switch (r) { case BZ_RUN_OK: /* Non-finishing */ case BZ_FINISH_OK: /* Finishing: There's more work to do */ return (ARCHIVE_OK); case BZ_STREAM_END: /* Finishing: all done */ /* Only occurs in finishing case */ return (ARCHIVE_EOF); default: /* Any other return value indicates an error */ archive_set_error(a, ARCHIVE_ERRNO_MISC, "Bzip2 compression failed:" " BZ2_bzCompress() call returned status %d", r); return (ARCHIVE_FATAL); } } static int compression_end_bzip2(struct archive *a, struct la_zstream *lastrm) { bz_stream *strm; int r; strm = (bz_stream *)lastrm->real_stream; r = BZ2_bzCompressEnd(strm); free(strm); lastrm->real_stream = NULL; lastrm->valid = 0; if (r != BZ_OK) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "Failed to clean up compressor"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } #else static int compression_init_encoder_bzip2(struct archive *a, struct la_zstream *lastrm, int level) { (void) level; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "bzip2")); } #endif /* * _7_LZMA1, _7_LZMA2 compressor. */ #if defined(HAVE_LZMA_H) static int compression_init_encoder_lzma(struct archive *a, struct la_zstream *lastrm, int level, uint64_t filter_id) { static const lzma_stream lzma_init_data = LZMA_STREAM_INIT; lzma_stream *strm; lzma_filter *lzmafilters; lzma_options_lzma lzma_opt; int r; if (lastrm->valid) compression_end(a, lastrm); strm = calloc(1, sizeof(*strm) + sizeof(*lzmafilters) * 2); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for lzma stream"); return (ARCHIVE_FATAL); } lzmafilters = (lzma_filter *)(strm+1); if (level > 6) level = 6; if (lzma_lzma_preset(&lzma_opt, level)) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ENOMEM, "Internal error initializing compression library"); return (ARCHIVE_FATAL); } lzmafilters[0].id = filter_id; lzmafilters[0].options = &lzma_opt; lzmafilters[1].id = LZMA_VLI_UNKNOWN;/* Terminate */ r = lzma_properties_size(&(lastrm->prop_size), lzmafilters); if (r != LZMA_OK) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "lzma_properties_size failed"); return (ARCHIVE_FATAL); } if (lastrm->prop_size) { lastrm->props = malloc(lastrm->prop_size); if (lastrm->props == NULL) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ENOMEM, "Cannot allocate memory"); return (ARCHIVE_FATAL); } r = lzma_properties_encode(lzmafilters, lastrm->props); if (r != LZMA_OK) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "lzma_properties_encode failed"); return (ARCHIVE_FATAL); } } *strm = lzma_init_data; r = lzma_raw_encoder(strm, lzmafilters); switch (r) { case LZMA_OK: lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_lzma; lastrm->end = compression_end_lzma; r = ARCHIVE_OK; break; case LZMA_MEM_ERROR: free(strm); lastrm->real_stream = NULL; archive_set_error(a, ENOMEM, "Internal error initializing compression library: " "Cannot allocate memory"); r = ARCHIVE_FATAL; break; default: free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library: " "It's a bug in liblzma"); r = ARCHIVE_FATAL; break; } return (r); } static int compression_init_encoder_lzma1(struct archive *a, struct la_zstream *lastrm, int level) { return compression_init_encoder_lzma(a, lastrm, level, LZMA_FILTER_LZMA1); } static int compression_init_encoder_lzma2(struct archive *a, struct la_zstream *lastrm, int level) { return compression_init_encoder_lzma(a, lastrm, level, LZMA_FILTER_LZMA2); } static int compression_code_lzma(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { lzma_stream *strm; int r; strm = (lzma_stream *)lastrm->real_stream; strm->next_in = lastrm->next_in; strm->avail_in = lastrm->avail_in; strm->total_in = lastrm->total_in; strm->next_out = lastrm->next_out; strm->avail_out = lastrm->avail_out; strm->total_out = lastrm->total_out; r = lzma_code(strm, (action == ARCHIVE_Z_FINISH)? LZMA_FINISH: LZMA_RUN); lastrm->next_in = strm->next_in; lastrm->avail_in = strm->avail_in; lastrm->total_in = strm->total_in; lastrm->next_out = strm->next_out; lastrm->avail_out = strm->avail_out; lastrm->total_out = strm->total_out; switch (r) { case LZMA_OK: /* Non-finishing case */ return (ARCHIVE_OK); case LZMA_STREAM_END: /* This return can only occur in finishing case. */ return (ARCHIVE_EOF); case LZMA_MEMLIMIT_ERROR: archive_set_error(a, ENOMEM, "lzma compression error:" " %ju MiB would have been needed", (uintmax_t)((lzma_memusage(strm) + 1024 * 1024 -1) / (1024 * 1024))); return (ARCHIVE_FATAL); default: /* Any other return value indicates an error */ archive_set_error(a, ARCHIVE_ERRNO_MISC, "lzma compression failed:" " lzma_code() call returned status %d", r); return (ARCHIVE_FATAL); } } static int compression_end_lzma(struct archive *a, struct la_zstream *lastrm) { lzma_stream *strm; (void)a; /* UNUSED */ strm = (lzma_stream *)lastrm->real_stream; lzma_end(strm); free(strm); lastrm->valid = 0; lastrm->real_stream = NULL; return (ARCHIVE_OK); } #else static int compression_init_encoder_lzma1(struct archive *a, struct la_zstream *lastrm, int level) { (void) level; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "lzma")); } static int compression_init_encoder_lzma2(struct archive *a, struct la_zstream *lastrm, int level) { (void) level; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "lzma")); } #endif /* * _7_PPMD compressor. */ static void * ppmd_alloc(void *p, size_t size) { (void)p; return malloc(size); } static void ppmd_free(void *p, void *address) { (void)p; free(address); } static ISzAlloc g_szalloc = { ppmd_alloc, ppmd_free }; static void ppmd_write(void *p, Byte b) { struct archive_write *a = ((IByteOut *)p)->a; struct _7zip *zip = (struct _7zip *)(a->format_data); struct la_zstream *lastrm = &(zip->stream); struct ppmd_stream *strm; if (lastrm->avail_out) { *lastrm->next_out++ = b; lastrm->avail_out--; lastrm->total_out++; return; } strm = (struct ppmd_stream *)lastrm->real_stream; if (strm->buff_ptr < strm->buff_end) { *strm->buff_ptr++ = b; strm->buff_bytes++; } } static int compression_init_encoder_ppmd(struct archive *a, struct la_zstream *lastrm, unsigned maxOrder, uint32_t msize) { struct ppmd_stream *strm; uint8_t *props; int r; if (lastrm->valid) compression_end(a, lastrm); strm = calloc(1, sizeof(*strm)); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for PPMd"); return (ARCHIVE_FATAL); } strm->buff = malloc(32); if (strm->buff == NULL) { free(strm); archive_set_error(a, ENOMEM, "Can't allocate memory for PPMd"); return (ARCHIVE_FATAL); } strm->buff_ptr = strm->buff; strm->buff_end = strm->buff + 32; props = malloc(1+4); if (props == NULL) { free(strm->buff); free(strm); archive_set_error(a, ENOMEM, "Coludn't allocate memory for PPMd"); return (ARCHIVE_FATAL); } props[0] = maxOrder; archive_le32enc(props+1, msize); __archive_ppmd7_functions.Ppmd7_Construct(&strm->ppmd7_context); r = __archive_ppmd7_functions.Ppmd7_Alloc( &strm->ppmd7_context, msize, &g_szalloc); if (r == 0) { free(strm->buff); free(strm); free(props); archive_set_error(a, ENOMEM, "Coludn't allocate memory for PPMd"); return (ARCHIVE_FATAL); } __archive_ppmd7_functions.Ppmd7_Init(&(strm->ppmd7_context), maxOrder); strm->byteout.a = (struct archive_write *)a; strm->byteout.Write = ppmd_write; strm->range_enc.Stream = &(strm->byteout); __archive_ppmd7_functions.Ppmd7z_RangeEnc_Init(&(strm->range_enc)); strm->stat = 0; lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_ppmd; lastrm->end = compression_end_ppmd; lastrm->prop_size = 5; lastrm->props = props; return (ARCHIVE_OK); } static int compression_code_ppmd(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { struct ppmd_stream *strm; (void)a; /* UNUSED */ strm = (struct ppmd_stream *)lastrm->real_stream; /* Copy encoded data if there are remaining bytes from previous call. */ if (strm->buff_bytes) { uint8_t *p = strm->buff_ptr - strm->buff_bytes; while (lastrm->avail_out && strm->buff_bytes) { *lastrm->next_out++ = *p++; lastrm->avail_out--; lastrm->total_out++; strm->buff_bytes--; } if (strm->buff_bytes) return (ARCHIVE_OK); if (strm->stat == 1) return (ARCHIVE_EOF); strm->buff_ptr = strm->buff; } while (lastrm->avail_in && lastrm->avail_out) { __archive_ppmd7_functions.Ppmd7_EncodeSymbol( &(strm->ppmd7_context), &(strm->range_enc), *lastrm->next_in++); lastrm->avail_in--; lastrm->total_in++; } if (lastrm->avail_in == 0 && action == ARCHIVE_Z_FINISH) { __archive_ppmd7_functions.Ppmd7z_RangeEnc_FlushData( &(strm->range_enc)); strm->stat = 1; /* Return EOF if there are no remaining bytes. */ if (strm->buff_bytes == 0) return (ARCHIVE_EOF); } return (ARCHIVE_OK); } static int compression_end_ppmd(struct archive *a, struct la_zstream *lastrm) { struct ppmd_stream *strm; (void)a; /* UNUSED */ strm = (struct ppmd_stream *)lastrm->real_stream; __archive_ppmd7_functions.Ppmd7_Free(&strm->ppmd7_context, &g_szalloc); free(strm->buff); free(strm); lastrm->real_stream = NULL; lastrm->valid = 0; return (ARCHIVE_OK); } /* * Universal compressor initializer. */ static int _7z_compression_init_encoder(struct archive_write *a, unsigned compression, int compression_level) { struct _7zip *zip; int r; zip = (struct _7zip *)a->format_data; switch (compression) { case _7Z_DEFLATE: r = compression_init_encoder_deflate( &(a->archive), &(zip->stream), compression_level, 0); break; case _7Z_BZIP2: r = compression_init_encoder_bzip2( &(a->archive), &(zip->stream), compression_level); break; case _7Z_LZMA1: r = compression_init_encoder_lzma1( &(a->archive), &(zip->stream), compression_level); break; case _7Z_LZMA2: r = compression_init_encoder_lzma2( &(a->archive), &(zip->stream), compression_level); break; case _7Z_PPMD: r = compression_init_encoder_ppmd( &(a->archive), &(zip->stream), PPMD7_DEFAULT_ORDER, PPMD7_DEFAULT_MEM_SIZE); break; case _7Z_COPY: default: r = compression_init_encoder_copy( &(a->archive), &(zip->stream)); break; } if (r == ARCHIVE_OK) { zip->stream.total_in = 0; zip->stream.next_out = zip->wbuff; zip->stream.avail_out = sizeof(zip->wbuff); zip->stream.total_out = 0; } return (r); } static int compression_code(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { if (lastrm->valid) return (lastrm->code(a, lastrm, action)); return (ARCHIVE_OK); } static int compression_end(struct archive *a, struct la_zstream *lastrm) { if (lastrm->valid) { lastrm->prop_size = 0; free(lastrm->props); lastrm->props = NULL; return (lastrm->end(a, lastrm)); } return (ARCHIVE_OK); } Index: head/contrib/libarchive/libarchive/archive_write_set_format_pax.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_set_format_pax.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_write_set_format_pax.c (revision 311903) @@ -1,1930 +1,1969 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * Copyright (c) 2010-2012 Michihiro NAKAJIMA * Copyright (c) 2016 Martin Matuska * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #include "archive.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_write_private.h" struct sparse_block { struct sparse_block *next; int is_hole; uint64_t offset; uint64_t remaining; }; struct pax { uint64_t entry_bytes_remaining; uint64_t entry_padding; struct archive_string l_url_encoded_name; struct archive_string pax_header; struct archive_string sparse_map; size_t sparse_map_padding; struct sparse_block *sparse_list; struct sparse_block *sparse_tail; struct archive_string_conv *sconv_utf8; int opt_binary; + + unsigned flags; +#define WRITE_SCHILY_XATTR (1 << 0) +#define WRITE_LIBARCHIVE_XATTR (1 << 1) }; static void add_pax_attr(struct archive_string *, const char *key, const char *value); +static void add_pax_attr_binary(struct archive_string *, + const char *key, + const char *value, size_t value_len); static void add_pax_attr_int(struct archive_string *, const char *key, int64_t value); static void add_pax_attr_time(struct archive_string *, const char *key, int64_t sec, unsigned long nanos); static int add_pax_acl(struct archive_write *, struct archive_entry *, struct pax *, int); static ssize_t archive_write_pax_data(struct archive_write *, const void *, size_t); static int archive_write_pax_close(struct archive_write *); static int archive_write_pax_free(struct archive_write *); static int archive_write_pax_finish_entry(struct archive_write *); static int archive_write_pax_header(struct archive_write *, struct archive_entry *); static int archive_write_pax_options(struct archive_write *, const char *, const char *); static char *base64_encode(const char *src, size_t len); static char *build_gnu_sparse_name(char *dest, const char *src); static char *build_pax_attribute_name(char *dest, const char *src); static char *build_ustar_entry_name(char *dest, const char *src, size_t src_length, const char *insert); static char *format_int(char *dest, int64_t); static int has_non_ASCII(const char *); static void sparse_list_clear(struct pax *); static int sparse_list_add(struct pax *, int64_t, int64_t); static char *url_encode(const char *in); /* * Set output format to 'restricted pax' format. * * This is the same as normal 'pax', but tries to suppress * the pax header whenever possible. This is the default for * bsdtar, for instance. */ int archive_write_set_format_pax_restricted(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int r; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_pax_restricted"); r = archive_write_set_format_pax(&a->archive); a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_RESTRICTED; a->archive.archive_format_name = "restricted POSIX pax interchange"; return (r); } /* * Set output format to 'pax' format. */ int archive_write_set_format_pax(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct pax *pax; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_pax"); if (a->format_free != NULL) (a->format_free)(a); pax = (struct pax *)calloc(1, sizeof(*pax)); if (pax == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate pax data"); return (ARCHIVE_FATAL); } + pax->flags = WRITE_LIBARCHIVE_XATTR | WRITE_SCHILY_XATTR; + a->format_data = pax; a->format_name = "pax"; a->format_options = archive_write_pax_options; a->format_write_header = archive_write_pax_header; a->format_write_data = archive_write_pax_data; a->format_close = archive_write_pax_close; a->format_free = archive_write_pax_free; a->format_finish_entry = archive_write_pax_finish_entry; a->archive.archive_format = ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE; a->archive.archive_format_name = "POSIX pax interchange"; return (ARCHIVE_OK); } static int archive_write_pax_options(struct archive_write *a, const char *key, const char *val) { struct pax *pax = (struct pax *)a->format_data; int ret = ARCHIVE_FAILED; if (strcmp(key, "hdrcharset") == 0) { /* * The character-set we can use are defined in * IEEE Std 1003.1-2001 */ if (val == NULL || val[0] == 0) archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "pax: hdrcharset option needs a character-set name"); else if (strcmp(val, "BINARY") == 0 || strcmp(val, "binary") == 0) { /* * Specify binary mode. We will not convert * filenames, uname and gname to any charsets. */ pax->opt_binary = 1; ret = ARCHIVE_OK; } else if (strcmp(val, "UTF-8") == 0) { /* * Specify UTF-8 character-set to be used for * filenames. This is almost the test that * running platform supports the string conversion. * Especially libarchive_test needs this trick for * its test. */ pax->sconv_utf8 = archive_string_conversion_to_charset( &(a->archive), "UTF-8", 0); if (pax->sconv_utf8 == NULL) ret = ARCHIVE_FATAL; else ret = ARCHIVE_OK; } else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "pax: invalid charset name"); return (ret); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } /* * Note: This code assumes that 'nanos' has the same sign as 'sec', * which implies that sec=-1, nanos=200000000 represents -1.2 seconds * and not -0.8 seconds. This is a pretty pedantic point, as we're * unlikely to encounter many real files created before Jan 1, 1970, * much less ones with timestamps recorded to sub-second resolution. */ static void add_pax_attr_time(struct archive_string *as, const char *key, int64_t sec, unsigned long nanos) { int digit, i; char *t; /* * Note that each byte contributes fewer than 3 base-10 * digits, so this will always be big enough. */ char tmp[1 + 3*sizeof(sec) + 1 + 3*sizeof(nanos)]; tmp[sizeof(tmp) - 1] = 0; t = tmp + sizeof(tmp) - 1; /* Skip trailing zeros in the fractional part. */ for (digit = 0, i = 10; i > 0 && digit == 0; i--) { digit = nanos % 10; nanos /= 10; } /* Only format the fraction if it's non-zero. */ if (i > 0) { while (i > 0) { *--t = "0123456789"[digit]; digit = nanos % 10; nanos /= 10; i--; } *--t = '.'; } t = format_int(t, sec); add_pax_attr(as, key, t); } static char * format_int(char *t, int64_t i) { uint64_t ui; if (i < 0) ui = (i == INT64_MIN) ? (uint64_t)(INT64_MAX) + 1 : (uint64_t)(-i); else ui = i; do { *--t = "0123456789"[ui % 10]; } while (ui /= 10); if (i < 0) *--t = '-'; return (t); } static void add_pax_attr_int(struct archive_string *as, const char *key, int64_t value) { char tmp[1 + 3 * sizeof(value)]; tmp[sizeof(tmp) - 1] = 0; add_pax_attr(as, key, format_int(tmp + sizeof(tmp) - 1, value)); } /* * Add a key/value attribute to the pax header. This function handles * the length field and various other syntactic requirements. */ static void add_pax_attr(struct archive_string *as, const char *key, const char *value) { + add_pax_attr_binary(as, key, value, strlen(value)); +} + +/* + * Add a key/value attribute to the pax header. This function handles + * binary values. + */ +static void +add_pax_attr_binary(struct archive_string *as, const char *key, + const char *value, size_t value_len) +{ int digits, i, len, next_ten; char tmp[1 + 3 * sizeof(int)]; /* < 3 base-10 digits per byte */ /*- * PAX attributes have the following layout: * <=> */ - len = 1 + (int)strlen(key) + 1 + (int)strlen(value) + 1; + len = 1 + (int)strlen(key) + 1 + (int)value_len + 1; /* * The field includes the length of the field, so * computing the correct length is tricky. I start by * counting the number of base-10 digits in 'len' and * computing the next higher power of 10. */ next_ten = 1; digits = 0; i = len; while (i > 0) { i = i / 10; digits++; next_ten = next_ten * 10; } /* * For example, if string without the length field is 99 * chars, then adding the 2 digit length "99" will force the * total length past 100, requiring an extra digit. The next * statement adjusts for this effect. */ if (len + digits >= next_ten) digits++; /* Now, we have the right length so we can build the line. */ tmp[sizeof(tmp) - 1] = 0; /* Null-terminate the work area. */ archive_strcat(as, format_int(tmp + sizeof(tmp) - 1, len + digits)); archive_strappend_char(as, ' '); archive_strcat(as, key); archive_strappend_char(as, '='); - archive_strcat(as, value); + archive_array_append(as, value, value_len); archive_strappend_char(as, '\n'); } +static void +archive_write_pax_header_xattr(struct pax *pax, const char *encoded_name, + const void *value, size_t value_len) +{ + struct archive_string s; + char *encoded_value; + + if (pax->flags & WRITE_LIBARCHIVE_XATTR) { + encoded_value = base64_encode((const char *)value, value_len); + + if (encoded_name != NULL && encoded_value != NULL) { + archive_string_init(&s); + archive_strcpy(&s, "LIBARCHIVE.xattr."); + archive_strcat(&s, encoded_name); + add_pax_attr(&(pax->pax_header), s.s, encoded_value); + archive_string_free(&s); + } + free(encoded_value); + } + if (pax->flags & WRITE_SCHILY_XATTR) { + archive_string_init(&s); + archive_strcpy(&s, "SCHILY.xattr."); + archive_strcat(&s, encoded_name); + add_pax_attr_binary(&(pax->pax_header), s.s, value, value_len); + archive_string_free(&s); + } +} + static int archive_write_pax_header_xattrs(struct archive_write *a, struct pax *pax, struct archive_entry *entry) { - struct archive_string s; int i = archive_entry_xattr_reset(entry); while (i--) { const char *name; const void *value; - char *encoded_value; char *url_encoded_name = NULL, *encoded_name = NULL; size_t size; int r; archive_entry_xattr_next(entry, &name, &value, &size); url_encoded_name = url_encode(name); if (url_encoded_name != NULL) { /* Convert narrow-character to UTF-8. */ r = archive_strcpy_l(&(pax->l_url_encoded_name), url_encoded_name, pax->sconv_utf8); free(url_encoded_name); /* Done with this. */ if (r == 0) encoded_name = pax->l_url_encoded_name.s; else if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } } - encoded_value = base64_encode((const char *)value, size); + archive_write_pax_header_xattr(pax, encoded_name, + value, size); - if (encoded_name != NULL && encoded_value != NULL) { - archive_string_init(&s); - archive_strcpy(&s, "LIBARCHIVE.xattr."); - archive_strcat(&s, encoded_name); - add_pax_attr(&(pax->pax_header), s.s, encoded_value); - archive_string_free(&s); - } - free(encoded_value); } return (ARCHIVE_OK); } static int get_entry_hardlink(struct archive_write *a, struct archive_entry *entry, const char **name, size_t *length, struct archive_string_conv *sc) { int r; r = archive_entry_hardlink_l(entry, name, length, sc); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } return (ARCHIVE_WARN); } return (ARCHIVE_OK); } static int get_entry_pathname(struct archive_write *a, struct archive_entry *entry, const char **name, size_t *length, struct archive_string_conv *sc) { int r; r = archive_entry_pathname_l(entry, name, length, sc); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } return (ARCHIVE_WARN); } return (ARCHIVE_OK); } static int get_entry_uname(struct archive_write *a, struct archive_entry *entry, const char **name, size_t *length, struct archive_string_conv *sc) { int r; r = archive_entry_uname_l(entry, name, length, sc); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Uname"); return (ARCHIVE_FATAL); } return (ARCHIVE_WARN); } return (ARCHIVE_OK); } static int get_entry_gname(struct archive_write *a, struct archive_entry *entry, const char **name, size_t *length, struct archive_string_conv *sc) { int r; r = archive_entry_gname_l(entry, name, length, sc); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Gname"); return (ARCHIVE_FATAL); } return (ARCHIVE_WARN); } return (ARCHIVE_OK); } static int get_entry_symlink(struct archive_write *a, struct archive_entry *entry, const char **name, size_t *length, struct archive_string_conv *sc) { int r; r = archive_entry_symlink_l(entry, name, length, sc); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } return (ARCHIVE_WARN); } return (ARCHIVE_OK); } /* Add ACL to pax header */ static int add_pax_acl(struct archive_write *a, struct archive_entry *entry, struct pax *pax, int flags) { char *p; const char *attr; int acl_types; acl_types = archive_entry_acl_types(entry); if ((acl_types & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) attr = "SCHILY.acl.ace"; else if ((flags & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) != 0) attr = "SCHILY.acl.access"; else if ((flags & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) != 0) attr = "SCHILY.acl.default"; else return (ARCHIVE_FATAL); p = archive_entry_acl_to_text_l(entry, NULL, flags, pax->sconv_utf8); if (p == NULL) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "%s %s", "Can't allocate memory for ", attr); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "%s %s %s", "Can't translate ", attr, " to UTF-8"); return(ARCHIVE_WARN); } else if (*p != '\0') { add_pax_attr(&(pax->pax_header), attr, p); free(p); } return(ARCHIVE_OK); } /* * TODO: Consider adding 'comment' and 'charset' fields to * archive_entry so that clients can specify them. Also, consider * adding generic key/value tags so clients can add arbitrary * key/value data. * * TODO: Break up this 700-line function!!!! Yowza! */ static int archive_write_pax_header(struct archive_write *a, struct archive_entry *entry_original) { struct archive_entry *entry_main; const char *p; const char *suffix; int need_extension, r, ret; int acl_types; int sparse_count; uint64_t sparse_total, real_size; struct pax *pax; const char *hardlink; const char *path = NULL, *linkpath = NULL; const char *uname = NULL, *gname = NULL; const void *mac_metadata; size_t mac_metadata_size; struct archive_string_conv *sconv; size_t hardlink_length, path_length, linkpath_length; size_t uname_length, gname_length; char paxbuff[512]; char ustarbuff[512]; char ustar_entry_name[256]; char pax_entry_name[256]; char gnu_sparse_name[256]; struct archive_string entry_name; ret = ARCHIVE_OK; need_extension = 0; pax = (struct pax *)a->format_data; /* Sanity check. */ if (archive_entry_pathname(entry_original) == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't record entry in tar file without pathname"); return (ARCHIVE_FAILED); } /* * Choose a header encoding. */ if (pax->opt_binary) sconv = NULL;/* Binary mode. */ else { /* Header encoding is UTF-8. */ if (pax->sconv_utf8 == NULL) { /* Initialize the string conversion object * we must need */ pax->sconv_utf8 = archive_string_conversion_to_charset( &(a->archive), "UTF-8", 1); if (pax->sconv_utf8 == NULL) /* Couldn't allocate memory */ return (ARCHIVE_FAILED); } sconv = pax->sconv_utf8; } r = get_entry_hardlink(a, entry_original, &hardlink, &hardlink_length, sconv); if (r == ARCHIVE_FATAL) return (r); else if (r != ARCHIVE_OK) { r = get_entry_hardlink(a, entry_original, &hardlink, &hardlink_length, NULL); if (r == ARCHIVE_FATAL) return (r); archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate linkname '%s' to %s", hardlink, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; sconv = NULL;/* The header charset switches to binary mode. */ } /* Make sure this is a type of entry that we can handle here */ if (hardlink == NULL) { switch (archive_entry_filetype(entry_original)) { case AE_IFBLK: case AE_IFCHR: case AE_IFIFO: case AE_IFLNK: case AE_IFREG: break; case AE_IFDIR: { /* * Ensure a trailing '/'. Modify the original * entry so the client sees the change. */ #if defined(_WIN32) && !defined(__CYGWIN__) const wchar_t *wp; wp = archive_entry_pathname_w(entry_original); if (wp != NULL && wp[wcslen(wp) -1] != L'/') { struct archive_wstring ws; archive_string_init(&ws); path_length = wcslen(wp); if (archive_wstring_ensure(&ws, path_length + 2) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate pax data"); archive_wstring_free(&ws); return(ARCHIVE_FATAL); } /* Should we keep '\' ? */ if (wp[path_length -1] == L'\\') path_length--; archive_wstrncpy(&ws, wp, path_length); archive_wstrappend_wchar(&ws, L'/'); archive_entry_copy_pathname_w( entry_original, ws.s); archive_wstring_free(&ws); p = NULL; } else #endif p = archive_entry_pathname(entry_original); /* * On Windows, this is a backup operation just in * case getting WCS failed. On POSIX, this is a * normal operation. */ if (p != NULL && p[strlen(p) - 1] != '/') { struct archive_string as; archive_string_init(&as); path_length = strlen(p); if (archive_string_ensure(&as, path_length + 2) == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate pax data"); archive_string_free(&as); return(ARCHIVE_FATAL); } #if defined(_WIN32) && !defined(__CYGWIN__) /* NOTE: This might break the pathname * if the current code page is CP932 and * the pathname includes a character '\' * as a part of its multibyte pathname. */ if (p[strlen(p) -1] == '\\') path_length--; else #endif archive_strncpy(&as, p, path_length); archive_strappend_char(&as, '/'); archive_entry_copy_pathname( entry_original, as.s); archive_string_free(&as); } break; } case AE_IFSOCK: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "tar format cannot archive socket"); return (ARCHIVE_FAILED); default: archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "tar format cannot archive this (type=0%lo)", (unsigned long) archive_entry_filetype(entry_original)); return (ARCHIVE_FAILED); } } /* * If Mac OS metadata blob is here, recurse to write that * as a separate entry. This is really a pretty poor design: * In particular, it doubles the overhead for long filenames. * TODO: Help Apple folks design something better and figure * out how to transition from this legacy format. * * Note that this code is present on every platform; clients * on non-Mac are unlikely to ever provide this data, but * applications that copy entries from one archive to another * should not lose data just because the local filesystem * can't store it. */ mac_metadata = archive_entry_mac_metadata(entry_original, &mac_metadata_size); if (mac_metadata != NULL) { const char *oname; char *name, *bname; size_t name_length; struct archive_entry *extra = archive_entry_new2(&a->archive); oname = archive_entry_pathname(entry_original); name_length = strlen(oname); name = malloc(name_length + 3); if (name == NULL || extra == NULL) { /* XXX error message */ archive_entry_free(extra); free(name); return (ARCHIVE_FAILED); } strcpy(name, oname); /* Find last '/'; strip trailing '/' characters */ bname = strrchr(name, '/'); while (bname != NULL && bname[1] == '\0') { *bname = '\0'; bname = strrchr(name, '/'); } if (bname == NULL) { memmove(name + 2, name, name_length + 1); memmove(name, "._", 2); } else { bname += 1; memmove(bname + 2, bname, strlen(bname) + 1); memmove(bname, "._", 2); } archive_entry_copy_pathname(extra, name); free(name); archive_entry_set_size(extra, mac_metadata_size); archive_entry_set_filetype(extra, AE_IFREG); archive_entry_set_perm(extra, archive_entry_perm(entry_original)); archive_entry_set_mtime(extra, archive_entry_mtime(entry_original), archive_entry_mtime_nsec(entry_original)); archive_entry_set_gid(extra, archive_entry_gid(entry_original)); archive_entry_set_gname(extra, archive_entry_gname(entry_original)); archive_entry_set_uid(extra, archive_entry_uid(entry_original)); archive_entry_set_uname(extra, archive_entry_uname(entry_original)); /* Recurse to write the special copyfile entry. */ r = archive_write_pax_header(a, extra); archive_entry_free(extra); if (r < ARCHIVE_WARN) return (r); if (r < ret) ret = r; r = (int)archive_write_pax_data(a, mac_metadata, mac_metadata_size); if (r < ARCHIVE_WARN) return (r); if (r < ret) ret = r; r = archive_write_pax_finish_entry(a); if (r < ARCHIVE_WARN) return (r); if (r < ret) ret = r; } /* Copy entry so we can modify it as needed. */ #if defined(_WIN32) && !defined(__CYGWIN__) /* Make sure the path separators in pathname, hardlink and symlink * are all slash '/', not the Windows path separator '\'. */ entry_main = __la_win_entry_in_posix_pathseparator(entry_original); if (entry_main == entry_original) entry_main = archive_entry_clone(entry_original); #else entry_main = archive_entry_clone(entry_original); #endif if (entry_main == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate pax data"); return(ARCHIVE_FATAL); } archive_string_empty(&(pax->pax_header)); /* Blank our work area. */ archive_string_empty(&(pax->sparse_map)); sparse_total = 0; sparse_list_clear(pax); if (hardlink == NULL && archive_entry_filetype(entry_main) == AE_IFREG) sparse_count = archive_entry_sparse_reset(entry_main); else sparse_count = 0; if (sparse_count) { int64_t offset, length, last_offset = 0; /* Get the last entry of sparse block. */ while (archive_entry_sparse_next( entry_main, &offset, &length) == ARCHIVE_OK) last_offset = offset + length; /* If the last sparse block does not reach the end of file, * We have to add a empty sparse block as the last entry to * manage storing file data. */ if (last_offset < archive_entry_size(entry_main)) archive_entry_sparse_add_entry(entry_main, archive_entry_size(entry_main), 0); sparse_count = archive_entry_sparse_reset(entry_main); } /* * First, check the name fields and see if any of them * require binary coding. If any of them does, then all of * them do. */ r = get_entry_pathname(a, entry_main, &path, &path_length, sconv); if (r == ARCHIVE_FATAL) return (r); else if (r != ARCHIVE_OK) { r = get_entry_pathname(a, entry_main, &path, &path_length, NULL); if (r == ARCHIVE_FATAL) return (r); archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate pathname '%s' to %s", path, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; sconv = NULL;/* The header charset switches to binary mode. */ } r = get_entry_uname(a, entry_main, &uname, &uname_length, sconv); if (r == ARCHIVE_FATAL) return (r); else if (r != ARCHIVE_OK) { r = get_entry_uname(a, entry_main, &uname, &uname_length, NULL); if (r == ARCHIVE_FATAL) return (r); archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate uname '%s' to %s", uname, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; sconv = NULL;/* The header charset switches to binary mode. */ } r = get_entry_gname(a, entry_main, &gname, &gname_length, sconv); if (r == ARCHIVE_FATAL) return (r); else if (r != ARCHIVE_OK) { r = get_entry_gname(a, entry_main, &gname, &gname_length, NULL); if (r == ARCHIVE_FATAL) return (r); archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate gname '%s' to %s", gname, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; sconv = NULL;/* The header charset switches to binary mode. */ } linkpath = hardlink; linkpath_length = hardlink_length; if (linkpath == NULL) { r = get_entry_symlink(a, entry_main, &linkpath, &linkpath_length, sconv); if (r == ARCHIVE_FATAL) return (r); else if (r != ARCHIVE_OK) { r = get_entry_symlink(a, entry_main, &linkpath, &linkpath_length, NULL); if (r == ARCHIVE_FATAL) return (r); archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate linkname '%s' to %s", linkpath, archive_string_conversion_charset_name(sconv)); ret = ARCHIVE_WARN; sconv = NULL; } } /* If any string conversions failed, get all attributes * in binary-mode. */ if (sconv == NULL && !pax->opt_binary) { if (hardlink != NULL) { r = get_entry_hardlink(a, entry_main, &hardlink, &hardlink_length, NULL); if (r == ARCHIVE_FATAL) return (r); linkpath = hardlink; linkpath_length = hardlink_length; } r = get_entry_pathname(a, entry_main, &path, &path_length, NULL); if (r == ARCHIVE_FATAL) return (r); r = get_entry_uname(a, entry_main, &uname, &uname_length, NULL); if (r == ARCHIVE_FATAL) return (r); r = get_entry_gname(a, entry_main, &gname, &gname_length, NULL); if (r == ARCHIVE_FATAL) return (r); } /* Store the header encoding first, to be nice to readers. */ if (sconv == NULL) add_pax_attr(&(pax->pax_header), "hdrcharset", "BINARY"); /* * If name is too long, or has non-ASCII characters, add * 'path' to pax extended attrs. (Note that an unconvertible * name must have non-ASCII characters.) */ if (has_non_ASCII(path)) { /* We have non-ASCII characters. */ add_pax_attr(&(pax->pax_header), "path", path); archive_entry_set_pathname(entry_main, build_ustar_entry_name(ustar_entry_name, path, path_length, NULL)); need_extension = 1; } else { /* We have an all-ASCII path; we'd like to just store * it in the ustar header if it will fit. Yes, this * duplicates some of the logic in * archive_write_set_format_ustar.c */ if (path_length <= 100) { /* Fits in the old 100-char tar name field. */ } else { /* Find largest suffix that will fit. */ /* Note: strlen() > 100, so strlen() - 100 - 1 >= 0 */ suffix = strchr(path + path_length - 100 - 1, '/'); /* Don't attempt an empty prefix. */ if (suffix == path) suffix = strchr(suffix + 1, '/'); /* We can put it in the ustar header if it's * all ASCII and it's either <= 100 characters * or can be split at a '/' into a prefix <= * 155 chars and a suffix <= 100 chars. (Note * the strchr() above will return NULL exactly * when the path can't be split.) */ if (suffix == NULL /* Suffix > 100 chars. */ || suffix[1] == '\0' /* empty suffix */ || suffix - path > 155) /* Prefix > 155 chars */ { add_pax_attr(&(pax->pax_header), "path", path); archive_entry_set_pathname(entry_main, build_ustar_entry_name(ustar_entry_name, path, path_length, NULL)); need_extension = 1; } } } if (linkpath != NULL) { /* If link name is too long or has non-ASCII characters, add * 'linkpath' to pax extended attrs. */ if (linkpath_length > 100 || has_non_ASCII(linkpath)) { add_pax_attr(&(pax->pax_header), "linkpath", linkpath); if (linkpath_length > 100) { if (hardlink != NULL) archive_entry_set_hardlink(entry_main, "././@LongHardLink"); else archive_entry_set_symlink(entry_main, "././@LongSymLink"); } need_extension = 1; } } /* Save a pathname since it will be renamed if `entry_main` has * sparse blocks. */ archive_string_init(&entry_name); archive_strcpy(&entry_name, archive_entry_pathname(entry_main)); /* If file size is too large, add 'size' to pax extended attrs. */ if (archive_entry_size(entry_main) >= (((int64_t)1) << 33)) { add_pax_attr_int(&(pax->pax_header), "size", archive_entry_size(entry_main)); need_extension = 1; } /* If numeric GID is too large, add 'gid' to pax extended attrs. */ if ((unsigned int)archive_entry_gid(entry_main) >= (1 << 18)) { add_pax_attr_int(&(pax->pax_header), "gid", archive_entry_gid(entry_main)); need_extension = 1; } /* If group name is too large or has non-ASCII characters, add * 'gname' to pax extended attrs. */ if (gname != NULL) { if (gname_length > 31 || has_non_ASCII(gname)) { add_pax_attr(&(pax->pax_header), "gname", gname); need_extension = 1; } } /* If numeric UID is too large, add 'uid' to pax extended attrs. */ if ((unsigned int)archive_entry_uid(entry_main) >= (1 << 18)) { add_pax_attr_int(&(pax->pax_header), "uid", archive_entry_uid(entry_main)); need_extension = 1; } /* Add 'uname' to pax extended attrs if necessary. */ if (uname != NULL) { if (uname_length > 31 || has_non_ASCII(uname)) { add_pax_attr(&(pax->pax_header), "uname", uname); need_extension = 1; } } /* * POSIX/SUSv3 doesn't provide a standard key for large device * numbers. I use the same keys here that Joerg Schilling * used for 'star.' (Which, somewhat confusingly, are called * "devXXX" even though they code "rdev" values.) No doubt, * other implementations use other keys. Note that there's no * reason we can't write the same information into a number of * different keys. * * Of course, this is only needed for block or char device entries. */ if (archive_entry_filetype(entry_main) == AE_IFBLK || archive_entry_filetype(entry_main) == AE_IFCHR) { /* * If rdevmajor is too large, add 'SCHILY.devmajor' to * extended attributes. */ int rdevmajor, rdevminor; rdevmajor = archive_entry_rdevmajor(entry_main); rdevminor = archive_entry_rdevminor(entry_main); if (rdevmajor >= (1 << 18)) { add_pax_attr_int(&(pax->pax_header), "SCHILY.devmajor", rdevmajor); /* * Non-strict formatting below means we don't * have to truncate here. Not truncating improves * the chance that some more modern tar archivers * (such as GNU tar 1.13) can restore the full * value even if they don't understand the pax * extended attributes. See my rant below about * file size fields for additional details. */ /* archive_entry_set_rdevmajor(entry_main, rdevmajor & ((1 << 18) - 1)); */ need_extension = 1; } /* * If devminor is too large, add 'SCHILY.devminor' to * extended attributes. */ if (rdevminor >= (1 << 18)) { add_pax_attr_int(&(pax->pax_header), "SCHILY.devminor", rdevminor); /* Truncation is not necessary here, either. */ /* archive_entry_set_rdevminor(entry_main, rdevminor & ((1 << 18) - 1)); */ need_extension = 1; } } /* * Technically, the mtime field in the ustar header can * support 33 bits, but many platforms use signed 32-bit time * values. The cutoff of 0x7fffffff here is a compromise. * Yes, this check is duplicated just below; this helps to * avoid writing an mtime attribute just to handle a * high-resolution timestamp in "restricted pax" mode. */ if (!need_extension && ((archive_entry_mtime(entry_main) < 0) || (archive_entry_mtime(entry_main) >= 0x7fffffff))) need_extension = 1; /* I use a star-compatible file flag attribute. */ p = archive_entry_fflags_text(entry_main); if (!need_extension && p != NULL && *p != '\0') need_extension = 1; /* If there are extended attributes, we need an extension */ if (!need_extension && archive_entry_xattr_count(entry_original) > 0) need_extension = 1; /* If there are sparse info, we need an extension */ if (!need_extension && sparse_count > 0) need_extension = 1; acl_types = archive_entry_acl_types(entry_original); /* If there are any ACL entries, we need an extension */ if (!need_extension && acl_types != 0) need_extension = 1; /* * Libarchive used to include these in extended headers for * restricted pax format, but that confused people who * expected ustar-like time semantics. So now we only include * them in full pax format. */ if (a->archive.archive_format != ARCHIVE_FORMAT_TAR_PAX_RESTRICTED) { if (archive_entry_ctime(entry_main) != 0 || archive_entry_ctime_nsec(entry_main) != 0) add_pax_attr_time(&(pax->pax_header), "ctime", archive_entry_ctime(entry_main), archive_entry_ctime_nsec(entry_main)); if (archive_entry_atime(entry_main) != 0 || archive_entry_atime_nsec(entry_main) != 0) add_pax_attr_time(&(pax->pax_header), "atime", archive_entry_atime(entry_main), archive_entry_atime_nsec(entry_main)); /* Store birth/creationtime only if it's earlier than mtime */ if (archive_entry_birthtime_is_set(entry_main) && archive_entry_birthtime(entry_main) < archive_entry_mtime(entry_main)) add_pax_attr_time(&(pax->pax_header), "LIBARCHIVE.creationtime", archive_entry_birthtime(entry_main), archive_entry_birthtime_nsec(entry_main)); } /* * The following items are handled differently in "pax * restricted" format. In particular, in "pax restricted" * format they won't be added unless need_extension is * already set (we're already generating an extended header, so * may as well include these). */ if (a->archive.archive_format != ARCHIVE_FORMAT_TAR_PAX_RESTRICTED || need_extension) { if (archive_entry_mtime(entry_main) < 0 || archive_entry_mtime(entry_main) >= 0x7fffffff || archive_entry_mtime_nsec(entry_main) != 0) add_pax_attr_time(&(pax->pax_header), "mtime", archive_entry_mtime(entry_main), archive_entry_mtime_nsec(entry_main)); /* I use a star-compatible file flag attribute. */ p = archive_entry_fflags_text(entry_main); if (p != NULL && *p != '\0') add_pax_attr(&(pax->pax_header), "SCHILY.fflags", p); /* I use star-compatible ACL attributes. */ if ((acl_types & ARCHIVE_ENTRY_ACL_TYPE_NFS4) != 0) { ret = add_pax_acl(a, entry_original, pax, ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID | ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA); if (ret == ARCHIVE_FATAL) return (ARCHIVE_FATAL); } if (acl_types & ARCHIVE_ENTRY_ACL_TYPE_ACCESS) { ret = add_pax_acl(a, entry_original, pax, ARCHIVE_ENTRY_ACL_TYPE_ACCESS | ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID | ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA); if (ret == ARCHIVE_FATAL) return (ARCHIVE_FATAL); } if (acl_types & ARCHIVE_ENTRY_ACL_TYPE_DEFAULT) { ret = add_pax_acl(a, entry_original, pax, ARCHIVE_ENTRY_ACL_TYPE_DEFAULT | ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID | ARCHIVE_ENTRY_ACL_STYLE_SEPARATOR_COMMA); if (ret == ARCHIVE_FATAL) return (ARCHIVE_FATAL); } /* We use GNU-tar-compatible sparse attributes. */ if (sparse_count > 0) { int64_t soffset, slength; add_pax_attr_int(&(pax->pax_header), "GNU.sparse.major", 1); add_pax_attr_int(&(pax->pax_header), "GNU.sparse.minor", 0); add_pax_attr(&(pax->pax_header), "GNU.sparse.name", entry_name.s); add_pax_attr_int(&(pax->pax_header), "GNU.sparse.realsize", archive_entry_size(entry_main)); /* Rename the file name which will be used for * ustar header to a special name, which GNU * PAX Format 1.0 requires */ archive_entry_set_pathname(entry_main, build_gnu_sparse_name(gnu_sparse_name, entry_name.s)); /* * - Make a sparse map, which will precede a file data. * - Get the total size of available data of sparse. */ archive_string_sprintf(&(pax->sparse_map), "%d\n", sparse_count); while (archive_entry_sparse_next(entry_main, &soffset, &slength) == ARCHIVE_OK) { archive_string_sprintf(&(pax->sparse_map), "%jd\n%jd\n", (intmax_t)soffset, (intmax_t)slength); sparse_total += slength; if (sparse_list_add(pax, soffset, slength) != ARCHIVE_OK) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); archive_entry_free(entry_main); archive_string_free(&entry_name); return (ARCHIVE_FATAL); } } } /* Store extended attributes */ if (archive_write_pax_header_xattrs(a, pax, entry_original) == ARCHIVE_FATAL) { archive_entry_free(entry_main); archive_string_free(&entry_name); return (ARCHIVE_FATAL); } } /* Only regular files have data. */ if (archive_entry_filetype(entry_main) != AE_IFREG) archive_entry_set_size(entry_main, 0); /* * Pax-restricted does not store data for hardlinks, in order * to improve compatibility with ustar. */ if (a->archive.archive_format != ARCHIVE_FORMAT_TAR_PAX_INTERCHANGE && hardlink != NULL) archive_entry_set_size(entry_main, 0); /* * XXX Full pax interchange format does permit a hardlink * entry to have data associated with it. I'm not supporting * that here because the client expects me to tell them whether * or not this format expects data for hardlinks. If I * don't check here, then every pax archive will end up with * duplicated data for hardlinks. Someday, there may be * need to select this behavior, in which case the following * will need to be revisited. XXX */ if (hardlink != NULL) archive_entry_set_size(entry_main, 0); /* Save a real file size. */ real_size = archive_entry_size(entry_main); /* * Overwrite a file size by the total size of sparse blocks and * the size of sparse map info. That file size is the length of * the data, which we will exactly store into an archive file. */ if (archive_strlen(&(pax->sparse_map))) { size_t mapsize = archive_strlen(&(pax->sparse_map)); pax->sparse_map_padding = 0x1ff & (-(ssize_t)mapsize); archive_entry_set_size(entry_main, mapsize + pax->sparse_map_padding + sparse_total); } /* Format 'ustar' header for main entry. * * The trouble with file size: If the reader can't understand * the file size, they may not be able to locate the next * entry and the rest of the archive is toast. Pax-compliant * readers are supposed to ignore the file size in the main * header, so the question becomes how to maximize portability * for readers that don't support pax attribute extensions. * For maximum compatibility, I permit numeric extensions in * the main header so that the file size stored will always be * correct, even if it's in a format that only some * implementations understand. The technique used here is: * * a) If possible, follow the standard exactly. This handles * files up to 8 gigabytes minus 1. * * b) If that fails, try octal but omit the field terminator. * That handles files up to 64 gigabytes minus 1. * * c) Otherwise, use base-256 extensions. That handles files * up to 2^63 in this implementation, with the potential to * go up to 2^94. That should hold us for a while. ;-) * * The non-strict formatter uses similar logic for other * numeric fields, though they're less critical. */ if (__archive_write_format_header_ustar(a, ustarbuff, entry_main, -1, 0, NULL) == ARCHIVE_FATAL) return (ARCHIVE_FATAL); /* If we built any extended attributes, write that entry first. */ if (archive_strlen(&(pax->pax_header)) > 0) { struct archive_entry *pax_attr_entry; time_t s; int64_t uid, gid; int mode; pax_attr_entry = archive_entry_new2(&a->archive); p = entry_name.s; archive_entry_set_pathname(pax_attr_entry, build_pax_attribute_name(pax_entry_name, p)); archive_entry_set_size(pax_attr_entry, archive_strlen(&(pax->pax_header))); /* Copy uid/gid (but clip to ustar limits). */ uid = archive_entry_uid(entry_main); if (uid >= 1 << 18) uid = (1 << 18) - 1; archive_entry_set_uid(pax_attr_entry, uid); gid = archive_entry_gid(entry_main); if (gid >= 1 << 18) gid = (1 << 18) - 1; archive_entry_set_gid(pax_attr_entry, gid); /* Copy mode over (but not setuid/setgid bits) */ mode = archive_entry_mode(entry_main); #ifdef S_ISUID mode &= ~S_ISUID; #endif #ifdef S_ISGID mode &= ~S_ISGID; #endif #ifdef S_ISVTX mode &= ~S_ISVTX; #endif archive_entry_set_mode(pax_attr_entry, mode); /* Copy uname/gname. */ archive_entry_set_uname(pax_attr_entry, archive_entry_uname(entry_main)); archive_entry_set_gname(pax_attr_entry, archive_entry_gname(entry_main)); /* Copy mtime, but clip to ustar limits. */ s = archive_entry_mtime(entry_main); if (s < 0) { s = 0; } if (s >= 0x7fffffff) { s = 0x7fffffff; } archive_entry_set_mtime(pax_attr_entry, s, 0); /* Standard ustar doesn't support atime. */ archive_entry_set_atime(pax_attr_entry, 0, 0); /* Standard ustar doesn't support ctime. */ archive_entry_set_ctime(pax_attr_entry, 0, 0); r = __archive_write_format_header_ustar(a, paxbuff, pax_attr_entry, 'x', 1, NULL); archive_entry_free(pax_attr_entry); /* Note that the 'x' header shouldn't ever fail to format */ if (r < ARCHIVE_WARN) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "archive_write_pax_header: " "'x' header failed?! This can't happen.\n"); return (ARCHIVE_FATAL); } else if (r < ret) ret = r; r = __archive_write_output(a, paxbuff, 512); if (r != ARCHIVE_OK) { sparse_list_clear(pax); pax->entry_bytes_remaining = 0; pax->entry_padding = 0; return (ARCHIVE_FATAL); } pax->entry_bytes_remaining = archive_strlen(&(pax->pax_header)); pax->entry_padding = 0x1ff & (-(int64_t)pax->entry_bytes_remaining); r = __archive_write_output(a, pax->pax_header.s, archive_strlen(&(pax->pax_header))); if (r != ARCHIVE_OK) { /* If a write fails, we're pretty much toast. */ return (ARCHIVE_FATAL); } /* Pad out the end of the entry. */ r = __archive_write_nulls(a, (size_t)pax->entry_padding); if (r != ARCHIVE_OK) { /* If a write fails, we're pretty much toast. */ return (ARCHIVE_FATAL); } pax->entry_bytes_remaining = pax->entry_padding = 0; } /* Write the header for main entry. */ r = __archive_write_output(a, ustarbuff, 512); if (r != ARCHIVE_OK) return (r); /* * Inform the client of the on-disk size we're using, so * they can avoid unnecessarily writing a body for something * that we're just going to ignore. */ archive_entry_set_size(entry_original, real_size); if (pax->sparse_list == NULL && real_size > 0) { /* This is not a sparse file but we handle its data as * a sparse block. */ sparse_list_add(pax, 0, real_size); sparse_total = real_size; } pax->entry_padding = 0x1ff & (-(int64_t)sparse_total); archive_entry_free(entry_main); archive_string_free(&entry_name); return (ret); } /* * We need a valid name for the regular 'ustar' entry. This routine * tries to hack something more-or-less reasonable. * * The approach here tries to preserve leading dir names. We do so by * working with four sections: * 1) "prefix" directory names, * 2) "suffix" directory names, * 3) inserted dir name (optional), * 4) filename. * * These sections must satisfy the following requirements: * * Parts 1 & 2 together form an initial portion of the dir name. * * Part 3 is specified by the caller. (It should not contain a leading * or trailing '/'.) * * Part 4 forms an initial portion of the base filename. * * The filename must be <= 99 chars to fit the ustar 'name' field. * * Parts 2, 3, 4 together must be <= 99 chars to fit the ustar 'name' fld. * * Part 1 must be <= 155 chars to fit the ustar 'prefix' field. * * If the original name ends in a '/', the new name must also end in a '/' * * Trailing '/.' sequences may be stripped. * * Note: Recall that the ustar format does not store the '/' separating * parts 1 & 2, but does store the '/' separating parts 2 & 3. */ static char * build_ustar_entry_name(char *dest, const char *src, size_t src_length, const char *insert) { const char *prefix, *prefix_end; const char *suffix, *suffix_end; const char *filename, *filename_end; char *p; int need_slash = 0; /* Was there a trailing slash? */ size_t suffix_length = 99; size_t insert_length; /* Length of additional dir element to be added. */ if (insert == NULL) insert_length = 0; else /* +2 here allows for '/' before and after the insert. */ insert_length = strlen(insert) + 2; /* Step 0: Quick bailout in a common case. */ if (src_length < 100 && insert == NULL) { strncpy(dest, src, src_length); dest[src_length] = '\0'; return (dest); } /* Step 1: Locate filename and enforce the length restriction. */ filename_end = src + src_length; /* Remove trailing '/' chars and '/.' pairs. */ for (;;) { if (filename_end > src && filename_end[-1] == '/') { filename_end --; need_slash = 1; /* Remember to restore trailing '/'. */ continue; } if (filename_end > src + 1 && filename_end[-1] == '.' && filename_end[-2] == '/') { filename_end -= 2; need_slash = 1; /* "foo/." will become "foo/" */ continue; } break; } if (need_slash) suffix_length--; /* Find start of filename. */ filename = filename_end - 1; while ((filename > src) && (*filename != '/')) filename --; if ((*filename == '/') && (filename < filename_end - 1)) filename ++; /* Adjust filename_end so that filename + insert fits in 99 chars. */ suffix_length -= insert_length; if (filename_end > filename + suffix_length) filename_end = filename + suffix_length; /* Calculate max size for "suffix" section (#3 above). */ suffix_length -= filename_end - filename; /* Step 2: Locate the "prefix" section of the dirname, including * trailing '/'. */ prefix = src; prefix_end = prefix + 155; if (prefix_end > filename) prefix_end = filename; while (prefix_end > prefix && *prefix_end != '/') prefix_end--; if ((prefix_end < filename) && (*prefix_end == '/')) prefix_end++; /* Step 3: Locate the "suffix" section of the dirname, * including trailing '/'. */ suffix = prefix_end; suffix_end = suffix + suffix_length; /* Enforce limit. */ if (suffix_end > filename) suffix_end = filename; if (suffix_end < suffix) suffix_end = suffix; while (suffix_end > suffix && *suffix_end != '/') suffix_end--; if ((suffix_end < filename) && (*suffix_end == '/')) suffix_end++; /* Step 4: Build the new name. */ /* The OpenBSD strlcpy function is safer, but less portable. */ /* Rather than maintain two versions, just use the strncpy version. */ p = dest; if (prefix_end > prefix) { strncpy(p, prefix, prefix_end - prefix); p += prefix_end - prefix; } if (suffix_end > suffix) { strncpy(p, suffix, suffix_end - suffix); p += suffix_end - suffix; } if (insert != NULL) { /* Note: assume insert does not have leading or trailing '/' */ strcpy(p, insert); p += strlen(insert); *p++ = '/'; } strncpy(p, filename, filename_end - filename); p += filename_end - filename; if (need_slash) *p++ = '/'; *p = '\0'; return (dest); } /* * The ustar header for the pax extended attributes must have a * reasonable name: SUSv3 requires 'dirname'/PaxHeader.'pid'/'filename' * where 'pid' is the PID of the archiving process. Unfortunately, * that makes testing a pain since the output varies for each run, * so I'm sticking with the simpler 'dirname'/PaxHeader/'filename' * for now. (Someday, I'll make this settable. Then I can use the * SUS recommendation as default and test harnesses can override it * to get predictable results.) * * Joerg Schilling has argued that this is unnecessary because, in * practice, if the pax extended attributes get extracted as regular * files, no one is going to bother reading those attributes to * manually restore them. Based on this, 'star' uses * /tmp/PaxHeader/'basename' as the ustar header name. This is a * tempting argument, in part because it's simpler than the SUSv3 * recommendation, but I'm not entirely convinced. I'm also * uncomfortable with the fact that "/tmp" is a Unix-ism. * * The following routine leverages build_ustar_entry_name() above and * so is simpler than you might think. It just needs to provide the * additional path element and handle a few pathological cases). */ static char * build_pax_attribute_name(char *dest, const char *src) { char buff[64]; const char *p; /* Handle the null filename case. */ if (src == NULL || *src == '\0') { strcpy(dest, "PaxHeader/blank"); return (dest); } /* Prune final '/' and other unwanted final elements. */ p = src + strlen(src); for (;;) { /* Ends in "/", remove the '/' */ if (p > src && p[-1] == '/') { --p; continue; } /* Ends in "/.", remove the '.' */ if (p > src + 1 && p[-1] == '.' && p[-2] == '/') { --p; continue; } break; } /* Pathological case: After above, there was nothing left. * This includes "/." "/./." "/.//./." etc. */ if (p == src) { strcpy(dest, "/PaxHeader/rootdir"); return (dest); } /* Convert unadorned "." into a suitable filename. */ if (*src == '.' && p == src + 1) { strcpy(dest, "PaxHeader/currentdir"); return (dest); } /* * TODO: Push this string into the 'pax' structure to avoid * recomputing it every time. That will also open the door * to having clients override it. */ #if HAVE_GETPID && 0 /* Disable this for now; see above comment. */ sprintf(buff, "PaxHeader.%d", getpid()); #else /* If the platform can't fetch the pid, don't include it. */ strcpy(buff, "PaxHeader"); #endif /* General case: build a ustar-compatible name adding * "/PaxHeader/". */ build_ustar_entry_name(dest, src, p - src, buff); return (dest); } /* * GNU PAX Format 1.0 requires the special name, which pattern is: * /GNUSparseFile./ * * This function is used for only Sparse file, a file type of which * is regular file. */ static char * build_gnu_sparse_name(char *dest, const char *src) { char buff[64]; const char *p; /* Handle the null filename case. */ if (src == NULL || *src == '\0') { strcpy(dest, "GNUSparseFile/blank"); return (dest); } /* Prune final '/' and other unwanted final elements. */ p = src + strlen(src); for (;;) { /* Ends in "/", remove the '/' */ if (p > src && p[-1] == '/') { --p; continue; } /* Ends in "/.", remove the '.' */ if (p > src + 1 && p[-1] == '.' && p[-2] == '/') { --p; continue; } break; } #if HAVE_GETPID && 0 /* Disable this as pax attribute name. */ sprintf(buff, "GNUSparseFile.%d", getpid()); #else /* If the platform can't fetch the pid, don't include it. */ strcpy(buff, "GNUSparseFile"); #endif /* General case: build a ustar-compatible name adding * "/GNUSparseFile/". */ build_ustar_entry_name(dest, src, p - src, buff); return (dest); } /* Write two null blocks for the end of archive */ static int archive_write_pax_close(struct archive_write *a) { return (__archive_write_nulls(a, 512 * 2)); } static int archive_write_pax_free(struct archive_write *a) { struct pax *pax; pax = (struct pax *)a->format_data; if (pax == NULL) return (ARCHIVE_OK); archive_string_free(&pax->pax_header); archive_string_free(&pax->sparse_map); archive_string_free(&pax->l_url_encoded_name); sparse_list_clear(pax); free(pax); a->format_data = NULL; return (ARCHIVE_OK); } static int archive_write_pax_finish_entry(struct archive_write *a) { struct pax *pax; uint64_t remaining; int ret; pax = (struct pax *)a->format_data; remaining = pax->entry_bytes_remaining; if (remaining == 0) { while (pax->sparse_list) { struct sparse_block *sb; if (!pax->sparse_list->is_hole) remaining += pax->sparse_list->remaining; sb = pax->sparse_list->next; free(pax->sparse_list); pax->sparse_list = sb; } } ret = __archive_write_nulls(a, (size_t)(remaining + pax->entry_padding)); pax->entry_bytes_remaining = pax->entry_padding = 0; return (ret); } static ssize_t archive_write_pax_data(struct archive_write *a, const void *buff, size_t s) { struct pax *pax; size_t ws; size_t total; int ret; pax = (struct pax *)a->format_data; /* * According to GNU PAX format 1.0, write a sparse map * before the body. */ if (archive_strlen(&(pax->sparse_map))) { ret = __archive_write_output(a, pax->sparse_map.s, archive_strlen(&(pax->sparse_map))); if (ret != ARCHIVE_OK) return (ret); ret = __archive_write_nulls(a, pax->sparse_map_padding); if (ret != ARCHIVE_OK) return (ret); archive_string_empty(&(pax->sparse_map)); } total = 0; while (total < s) { const unsigned char *p; while (pax->sparse_list != NULL && pax->sparse_list->remaining == 0) { struct sparse_block *sb = pax->sparse_list->next; free(pax->sparse_list); pax->sparse_list = sb; } if (pax->sparse_list == NULL) return (total); p = ((const unsigned char *)buff) + total; ws = s - total; if (ws > pax->sparse_list->remaining) ws = (size_t)pax->sparse_list->remaining; if (pax->sparse_list->is_hole) { /* Current block is hole thus we do not write * the body. */ pax->sparse_list->remaining -= ws; total += ws; continue; } ret = __archive_write_output(a, p, ws); pax->sparse_list->remaining -= ws; total += ws; if (ret != ARCHIVE_OK) return (ret); } return (total); } static int has_non_ASCII(const char *_p) { const unsigned char *p = (const unsigned char *)_p; if (p == NULL) return (1); while (*p != '\0' && *p < 128) p++; return (*p != '\0'); } /* * Used by extended attribute support; encodes the name * so that there will be no '=' characters in the result. */ static char * url_encode(const char *in) { const char *s; char *d; int out_len = 0; char *out; for (s = in; *s != '\0'; s++) { if (*s < 33 || *s > 126 || *s == '%' || *s == '=') out_len += 3; else out_len++; } out = (char *)malloc(out_len + 1); if (out == NULL) return (NULL); for (s = in, d = out; *s != '\0'; s++) { /* encode any non-printable ASCII character or '%' or '=' */ if (*s < 33 || *s > 126 || *s == '%' || *s == '=') { /* URL encoding is '%' followed by two hex digits */ *d++ = '%'; *d++ = "0123456789ABCDEF"[0x0f & (*s >> 4)]; *d++ = "0123456789ABCDEF"[0x0f & *s]; } else { *d++ = *s; } } *d = '\0'; return (out); } /* * Encode a sequence of bytes into a C string using base-64 encoding. * * Returns a null-terminated C string allocated with malloc(); caller * is responsible for freeing the result. */ static char * base64_encode(const char *s, size_t len) { static const char digits[64] = { 'A','B','C','D','E','F','G','H','I','J','K','L','M','N','O', 'P','Q','R','S','T','U','V','W','X','Y','Z','a','b','c','d', 'e','f','g','h','i','j','k','l','m','n','o','p','q','r','s', 't','u','v','w','x','y','z','0','1','2','3','4','5','6','7', '8','9','+','/' }; int v; char *d, *out; /* 3 bytes becomes 4 chars, but round up and allow for trailing NUL */ out = (char *)malloc((len * 4 + 2) / 3 + 1); if (out == NULL) return (NULL); d = out; /* Convert each group of 3 bytes into 4 characters. */ while (len >= 3) { v = (((int)s[0] << 16) & 0xff0000) | (((int)s[1] << 8) & 0xff00) | (((int)s[2]) & 0x00ff); s += 3; len -= 3; *d++ = digits[(v >> 18) & 0x3f]; *d++ = digits[(v >> 12) & 0x3f]; *d++ = digits[(v >> 6) & 0x3f]; *d++ = digits[(v) & 0x3f]; } /* Handle final group of 1 byte (2 chars) or 2 bytes (3 chars). */ switch (len) { case 0: break; case 1: v = (((int)s[0] << 16) & 0xff0000); *d++ = digits[(v >> 18) & 0x3f]; *d++ = digits[(v >> 12) & 0x3f]; break; case 2: v = (((int)s[0] << 16) & 0xff0000) | (((int)s[1] << 8) & 0xff00); *d++ = digits[(v >> 18) & 0x3f]; *d++ = digits[(v >> 12) & 0x3f]; *d++ = digits[(v >> 6) & 0x3f]; break; } /* Add trailing NUL character so output is a valid C string. */ *d = '\0'; return (out); } static void sparse_list_clear(struct pax *pax) { while (pax->sparse_list != NULL) { struct sparse_block *sb = pax->sparse_list; pax->sparse_list = sb->next; free(sb); } pax->sparse_tail = NULL; } static int _sparse_list_add_block(struct pax *pax, int64_t offset, int64_t length, int is_hole) { struct sparse_block *sb; sb = (struct sparse_block *)malloc(sizeof(*sb)); if (sb == NULL) return (ARCHIVE_FATAL); sb->next = NULL; sb->is_hole = is_hole; sb->offset = offset; sb->remaining = length; if (pax->sparse_list == NULL || pax->sparse_tail == NULL) pax->sparse_list = pax->sparse_tail = sb; else { pax->sparse_tail->next = sb; pax->sparse_tail = sb; } return (ARCHIVE_OK); } static int sparse_list_add(struct pax *pax, int64_t offset, int64_t length) { int64_t last_offset; int r; if (pax->sparse_tail == NULL) last_offset = 0; else { last_offset = pax->sparse_tail->offset + pax->sparse_tail->remaining; } if (last_offset < offset) { /* Add a hole block. */ r = _sparse_list_add_block(pax, last_offset, offset - last_offset, 1); if (r != ARCHIVE_OK) return (r); } /* Add data block. */ return (_sparse_list_add_block(pax, offset, length, 0)); } Index: head/contrib/libarchive/libarchive/archive_write_set_format_xar.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_set_format_xar.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_write_set_format_xar.c (revision 311903) @@ -1,3223 +1,3223 @@ /*- * Copyright (c) 2010-2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #include #if HAVE_LIBXML_XMLWRITER_H #include #endif #ifdef HAVE_BZLIB_H #include #endif #if HAVE_LZMA_H #include #endif #ifdef HAVE_ZLIB_H #include #endif #include "archive.h" #include "archive_digest_private.h" #include "archive_endian.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_private.h" #include "archive_rb.h" #include "archive_string.h" #include "archive_write_private.h" /* * Differences to xar utility. * - Subdocument is not supported yet. * - ACL is not supported yet. * - When writing an XML element , * which is a file type a symbolic link is referencing is always marked * as "broken". Xar utility uses stat(2) to get the file type, but, in * libarchive format writer, we should not use it; if it is needed, we * should get about it at archive_read_disk.c. * - It is possible to appear both and elements. * Xar utility generates on BSD platform and on Linux * platform. * */ #if !(defined(HAVE_LIBXML_XMLWRITER_H) && defined(LIBXML_VERSION) &&\ LIBXML_VERSION >= 20703) ||\ !defined(HAVE_ZLIB_H) || \ !defined(ARCHIVE_HAS_MD5) || !defined(ARCHIVE_HAS_SHA1) /* * xar needs several external libraries. * o libxml2 * o openssl or MD5/SHA1 hash function * o zlib * o bzlib2 (option) * o liblzma (option) */ int archive_write_set_format_xar(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Xar not supported on this platform"); return (ARCHIVE_WARN); } #else /* Support xar format */ /*#define DEBUG_PRINT_TOC 1 */ #define BAD_CAST_CONST (const xmlChar *) #define HEADER_MAGIC 0x78617221 #define HEADER_SIZE 28 #define HEADER_VERSION 1 enum sumalg { CKSUM_NONE = 0, CKSUM_SHA1 = 1, CKSUM_MD5 = 2 }; #define MD5_SIZE 16 #define SHA1_SIZE 20 #define MAX_SUM_SIZE 20 #define MD5_NAME "md5" #define SHA1_NAME "sha1" enum enctype { NONE, GZIP, BZIP2, LZMA, XZ, }; struct chksumwork { enum sumalg alg; #ifdef ARCHIVE_HAS_MD5 archive_md5_ctx md5ctx; #endif #ifdef ARCHIVE_HAS_SHA1 archive_sha1_ctx sha1ctx; #endif }; enum la_zaction { ARCHIVE_Z_FINISH, ARCHIVE_Z_RUN }; /* * Universal zstream. */ struct la_zstream { const unsigned char *next_in; size_t avail_in; uint64_t total_in; unsigned char *next_out; size_t avail_out; uint64_t total_out; int valid; void *real_stream; int (*code) (struct archive *a, struct la_zstream *lastrm, enum la_zaction action); int (*end)(struct archive *a, struct la_zstream *lastrm); }; struct chksumval { enum sumalg alg; size_t len; unsigned char val[MAX_SUM_SIZE]; }; struct heap_data { int id; struct heap_data *next; uint64_t temp_offset; uint64_t length; /* archived size. */ uint64_t size; /* extracted size. */ enum enctype compression; struct chksumval a_sum; /* archived checksum. */ struct chksumval e_sum; /* extracted checksum. */ }; struct file { struct archive_rb_node rbnode; int id; struct archive_entry *entry; struct archive_rb_tree rbtree; struct file *next; struct file *chnext; struct file *hlnext; /* For hardlinked files. * Use only when archive_entry_nlink() > 1 */ struct file *hardlink_target; struct file *parent; /* parent directory entry */ /* * To manage sub directory files. * We use 'chnext' (a member of struct file) to chain. */ struct { struct file *first; struct file **last; } children; /* For making a directory tree. */ struct archive_string parentdir; struct archive_string basename; struct archive_string symlink; int ea_idx; struct { struct heap_data *first; struct heap_data **last; } xattr; struct heap_data data; struct archive_string script; int virtual:1; int dir:1; }; struct hardlink { struct archive_rb_node rbnode; int nlink; struct { struct file *first; struct file **last; } file_list; }; struct xar { int temp_fd; uint64_t temp_offset; int file_idx; struct file *root; struct file *cur_dirent; struct archive_string cur_dirstr; struct file *cur_file; uint64_t bytes_remaining; struct archive_string tstr; struct archive_string vstr; enum sumalg opt_toc_sumalg; enum sumalg opt_sumalg; enum enctype opt_compression; int opt_compression_level; uint32_t opt_threads; struct chksumwork a_sumwrk; /* archived checksum. */ struct chksumwork e_sumwrk; /* extracted checksum. */ struct la_zstream stream; struct archive_string_conv *sconv; /* * Compressed data buffer. */ unsigned char wbuff[1024 * 64]; size_t wbuff_remaining; struct heap_data toc; /* * The list of all file entries is used to manage struct file * objects. * We use 'next' (a member of struct file) to chain. */ struct { struct file *first; struct file **last; } file_list; /* * The list of hard-linked file entries. * We use 'hlnext' (a member of struct file) to chain. */ struct archive_rb_tree hardlink_rbtree; }; static int xar_options(struct archive_write *, const char *, const char *); static int xar_write_header(struct archive_write *, struct archive_entry *); static ssize_t xar_write_data(struct archive_write *, const void *, size_t); static int xar_finish_entry(struct archive_write *); static int xar_close(struct archive_write *); static int xar_free(struct archive_write *); static struct file *file_new(struct archive_write *a, struct archive_entry *); static void file_free(struct file *); static struct file *file_create_virtual_dir(struct archive_write *a, struct xar *, const char *); static int file_add_child_tail(struct file *, struct file *); static struct file *file_find_child(struct file *, const char *); static int file_gen_utility_names(struct archive_write *, struct file *); static int get_path_component(char *, int, const char *); static int file_tree(struct archive_write *, struct file **); static void file_register(struct xar *, struct file *); static void file_init_register(struct xar *); static void file_free_register(struct xar *); static int file_register_hardlink(struct archive_write *, struct file *); static void file_connect_hardlink_files(struct xar *); static void file_init_hardlinks(struct xar *); static void file_free_hardlinks(struct xar *); static void checksum_init(struct chksumwork *, enum sumalg); static void checksum_update(struct chksumwork *, const void *, size_t); static void checksum_final(struct chksumwork *, struct chksumval *); static int compression_init_encoder_gzip(struct archive *, struct la_zstream *, int, int); static int compression_code_gzip(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_gzip(struct archive *, struct la_zstream *); static int compression_init_encoder_bzip2(struct archive *, struct la_zstream *, int); #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) static int compression_code_bzip2(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_bzip2(struct archive *, struct la_zstream *); #endif static int compression_init_encoder_lzma(struct archive *, struct la_zstream *, int); static int compression_init_encoder_xz(struct archive *, struct la_zstream *, int, int); #if defined(HAVE_LZMA_H) static int compression_code_lzma(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end_lzma(struct archive *, struct la_zstream *); #endif static int xar_compression_init_encoder(struct archive_write *); static int compression_code(struct archive *, struct la_zstream *, enum la_zaction); static int compression_end(struct archive *, struct la_zstream *); static int save_xattrs(struct archive_write *, struct file *); static int getalgsize(enum sumalg); static const char *getalgname(enum sumalg); int archive_write_set_format_xar(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct xar *xar; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_xar"); /* If another format was already registered, unregister it. */ if (a->format_free != NULL) (a->format_free)(a); xar = calloc(1, sizeof(*xar)); if (xar == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate xar data"); return (ARCHIVE_FATAL); } xar->temp_fd = -1; file_init_register(xar); file_init_hardlinks(xar); archive_string_init(&(xar->tstr)); archive_string_init(&(xar->vstr)); /* * Create the root directory. */ xar->root = file_create_virtual_dir(a, xar, ""); if (xar->root == NULL) { free(xar); archive_set_error(&a->archive, ENOMEM, "Can't allocate xar data"); return (ARCHIVE_FATAL); } xar->root->parent = xar->root; file_register(xar, xar->root); xar->cur_dirent = xar->root; archive_string_init(&(xar->cur_dirstr)); archive_string_ensure(&(xar->cur_dirstr), 1); xar->cur_dirstr.s[0] = 0; /* * Initialize option. */ /* Set default checksum type. */ xar->opt_toc_sumalg = CKSUM_SHA1; xar->opt_sumalg = CKSUM_SHA1; /* Set default compression type, level, and number of threads. */ xar->opt_compression = GZIP; xar->opt_compression_level = 6; xar->opt_threads = 1; a->format_data = xar; a->format_name = "xar"; a->format_options = xar_options; a->format_write_header = xar_write_header; a->format_write_data = xar_write_data; a->format_finish_entry = xar_finish_entry; a->format_close = xar_close; a->format_free = xar_free; a->archive.archive_format = ARCHIVE_FORMAT_XAR; a->archive.archive_format_name = "xar"; return (ARCHIVE_OK); } static int xar_options(struct archive_write *a, const char *key, const char *value) { struct xar *xar; xar = (struct xar *)a->format_data; if (strcmp(key, "checksum") == 0) { if (value == NULL) xar->opt_sumalg = CKSUM_NONE; else if (strcmp(value, "sha1") == 0) xar->opt_sumalg = CKSUM_SHA1; else if (strcmp(value, "md5") == 0) xar->opt_sumalg = CKSUM_MD5; else { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Unknown checksum name: `%s'", value); return (ARCHIVE_FAILED); } return (ARCHIVE_OK); } if (strcmp(key, "compression") == 0) { const char *name = NULL; if (value == NULL) xar->opt_compression = NONE; else if (strcmp(value, "gzip") == 0) xar->opt_compression = GZIP; else if (strcmp(value, "bzip2") == 0) #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) xar->opt_compression = BZIP2; #else name = "bzip2"; #endif else if (strcmp(value, "lzma") == 0) #if HAVE_LZMA_H xar->opt_compression = LZMA; #else name = "lzma"; #endif else if (strcmp(value, "xz") == 0) #if HAVE_LZMA_H xar->opt_compression = XZ; #else name = "xz"; #endif else { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Unknown compression name: `%s'", value); return (ARCHIVE_FAILED); } if (name != NULL) { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "`%s' compression not supported " "on this platform", name); return (ARCHIVE_FAILED); } return (ARCHIVE_OK); } if (strcmp(key, "compression-level") == 0) { if (value == NULL || !(value[0] >= '0' && value[0] <= '9') || value[1] != '\0') { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Illegal value `%s'", value); return (ARCHIVE_FAILED); } xar->opt_compression_level = value[0] - '0'; return (ARCHIVE_OK); } if (strcmp(key, "toc-checksum") == 0) { if (value == NULL) xar->opt_toc_sumalg = CKSUM_NONE; else if (strcmp(value, "sha1") == 0) xar->opt_toc_sumalg = CKSUM_SHA1; else if (strcmp(value, "md5") == 0) xar->opt_toc_sumalg = CKSUM_MD5; else { archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Unknown checksum name: `%s'", value); return (ARCHIVE_FAILED); } return (ARCHIVE_OK); } if (strcmp(key, "threads") == 0) { if (value == NULL) return (ARCHIVE_FAILED); xar->opt_threads = (int)strtoul(value, NULL, 10); if (xar->opt_threads == 0 && errno != 0) { xar->opt_threads = 1; archive_set_error(&(a->archive), ARCHIVE_ERRNO_MISC, "Illegal value `%s'", value); return (ARCHIVE_FAILED); } if (xar->opt_threads == 0) { #ifdef HAVE_LZMA_STREAM_ENCODER_MT xar->opt_threads = lzma_cputhreads(); #else xar->opt_threads = 1; #endif } } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } static int xar_write_header(struct archive_write *a, struct archive_entry *entry) { struct xar *xar; struct file *file; struct archive_entry *file_entry; int r, r2; xar = (struct xar *)a->format_data; xar->cur_file = NULL; xar->bytes_remaining = 0; if (xar->sconv == NULL) { xar->sconv = archive_string_conversion_to_charset( &a->archive, "UTF-8", 1); if (xar->sconv == NULL) return (ARCHIVE_FATAL); } file = file_new(a, entry); if (file == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate data"); return (ARCHIVE_FATAL); } r2 = file_gen_utility_names(a, file); if (r2 < ARCHIVE_WARN) return (r2); /* * Ignore a path which looks like the top of directory name * since we have already made the root directory of an Xar archive. */ if (archive_strlen(&(file->parentdir)) == 0 && archive_strlen(&(file->basename)) == 0) { file_free(file); return (r2); } /* Add entry into tree */ file_entry = file->entry; r = file_tree(a, &file); if (r != ARCHIVE_OK) return (r); /* There is the same file in tree and * the current file is older than the file in tree. * So we don't need the current file data anymore. */ if (file->entry != file_entry) return (r2); if (file->id == 0) file_register(xar, file); /* A virtual file, which is a directory, does not have * any contents and we won't store it into a archive * file other than its name. */ if (file->virtual) return (r2); /* * Prepare to save the contents of the file. */ if (xar->temp_fd == -1) { int algsize; xar->temp_offset = 0; xar->temp_fd = __archive_mktemp(NULL); if (xar->temp_fd < 0) { archive_set_error(&a->archive, errno, "Couldn't create temporary file"); return (ARCHIVE_FATAL); } algsize = getalgsize(xar->opt_toc_sumalg); if (algsize > 0) { if (lseek(xar->temp_fd, algsize, SEEK_SET) < 0) { archive_set_error(&(a->archive), errno, "lseek failed"); return (ARCHIVE_FATAL); } xar->temp_offset = algsize; } } if (archive_entry_hardlink(file->entry) == NULL) { r = save_xattrs(a, file); if (r != ARCHIVE_OK) return (ARCHIVE_FATAL); } /* Non regular files contents are unneeded to be saved to * a temporary file. */ if (archive_entry_filetype(file->entry) != AE_IFREG) return (r2); /* * Set the current file to cur_file to read its contents. */ xar->cur_file = file; if (archive_entry_nlink(file->entry) > 1) { r = file_register_hardlink(a, file); if (r != ARCHIVE_OK) return (r); if (archive_entry_hardlink(file->entry) != NULL) { archive_entry_unset_size(file->entry); return (r2); } } /* Save a offset of current file in temporary file. */ file->data.temp_offset = xar->temp_offset; file->data.size = archive_entry_size(file->entry); file->data.compression = xar->opt_compression; xar->bytes_remaining = archive_entry_size(file->entry); checksum_init(&(xar->a_sumwrk), xar->opt_sumalg); checksum_init(&(xar->e_sumwrk), xar->opt_sumalg); r = xar_compression_init_encoder(a); if (r != ARCHIVE_OK) return (r); else return (r2); } static int write_to_temp(struct archive_write *a, const void *buff, size_t s) { struct xar *xar; const unsigned char *p; ssize_t ws; xar = (struct xar *)a->format_data; p = (const unsigned char *)buff; while (s) { ws = write(xar->temp_fd, p, s); if (ws < 0) { archive_set_error(&(a->archive), errno, "fwrite function failed"); return (ARCHIVE_FATAL); } s -= ws; p += ws; xar->temp_offset += ws; } return (ARCHIVE_OK); } static ssize_t xar_write_data(struct archive_write *a, const void *buff, size_t s) { struct xar *xar; enum la_zaction run; size_t size, rsize; int r; xar = (struct xar *)a->format_data; if (s > xar->bytes_remaining) s = (size_t)xar->bytes_remaining; if (s == 0 || xar->cur_file == NULL) return (0); if (xar->cur_file->data.compression == NONE) { checksum_update(&(xar->e_sumwrk), buff, s); checksum_update(&(xar->a_sumwrk), buff, s); size = rsize = s; } else { xar->stream.next_in = (const unsigned char *)buff; xar->stream.avail_in = s; if (xar->bytes_remaining > s) run = ARCHIVE_Z_RUN; else run = ARCHIVE_Z_FINISH; /* Compress file data. */ r = compression_code(&(a->archive), &(xar->stream), run); if (r != ARCHIVE_OK && r != ARCHIVE_EOF) return (ARCHIVE_FATAL); rsize = s - xar->stream.avail_in; checksum_update(&(xar->e_sumwrk), buff, rsize); size = sizeof(xar->wbuff) - xar->stream.avail_out; checksum_update(&(xar->a_sumwrk), xar->wbuff, size); } #if !defined(_WIN32) || defined(__CYGWIN__) if (xar->bytes_remaining == (uint64_t)archive_entry_size(xar->cur_file->entry)) { /* * Get the path of a shell script if so. */ const unsigned char *b = (const unsigned char *)buff; archive_string_empty(&(xar->cur_file->script)); if (rsize > 2 && b[0] == '#' && b[1] == '!') { size_t i, end, off; off = 2; if (b[off] == ' ') off++; #ifdef PATH_MAX if ((rsize - off) > PATH_MAX) end = off + PATH_MAX; else #endif end = rsize; /* Find the end of a script path. */ for (i = off; i < end && b[i] != '\0' && b[i] != '\n' && b[i] != '\r' && b[i] != ' ' && b[i] != '\t'; i++) ; archive_strncpy(&(xar->cur_file->script), b + off, i - off); } } #endif if (xar->cur_file->data.compression == NONE) { if (write_to_temp(a, buff, size) != ARCHIVE_OK) return (ARCHIVE_FATAL); } else { if (write_to_temp(a, xar->wbuff, size) != ARCHIVE_OK) return (ARCHIVE_FATAL); } xar->bytes_remaining -= rsize; xar->cur_file->data.length += size; return (rsize); } static int xar_finish_entry(struct archive_write *a) { struct xar *xar; struct file *file; size_t s; ssize_t w; xar = (struct xar *)a->format_data; if (xar->cur_file == NULL) return (ARCHIVE_OK); while (xar->bytes_remaining > 0) { s = (size_t)xar->bytes_remaining; if (s > a->null_length) s = a->null_length; w = xar_write_data(a, a->nulls, s); if (w > 0) xar->bytes_remaining -= w; else return (w); } file = xar->cur_file; checksum_final(&(xar->e_sumwrk), &(file->data.e_sum)); checksum_final(&(xar->a_sumwrk), &(file->data.a_sum)); xar->cur_file = NULL; return (ARCHIVE_OK); } static int xmlwrite_string_attr(struct archive_write *a, xmlTextWriterPtr writer, const char *key, const char *value, const char *attrkey, const char *attrvalue) { int r; r = xmlTextWriterStartElement(writer, BAD_CAST_CONST(key)); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterStartElement() failed: %d", r); return (ARCHIVE_FATAL); } if (attrkey != NULL && attrvalue != NULL) { r = xmlTextWriterWriteAttribute(writer, BAD_CAST_CONST(attrkey), BAD_CAST_CONST(attrvalue)); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterWriteAttribute() failed: %d", r); return (ARCHIVE_FATAL); } } if (value != NULL) { r = xmlTextWriterWriteString(writer, BAD_CAST_CONST(value)); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterWriteString() failed: %d", r); return (ARCHIVE_FATAL); } } r = xmlTextWriterEndElement(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterEndElement() failed: %d", r); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } static int xmlwrite_string(struct archive_write *a, xmlTextWriterPtr writer, const char *key, const char *value) { int r; if (value == NULL) return (ARCHIVE_OK); r = xmlTextWriterStartElement(writer, BAD_CAST_CONST(key)); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterStartElement() failed: %d", r); return (ARCHIVE_FATAL); } if (value != NULL) { r = xmlTextWriterWriteString(writer, BAD_CAST_CONST(value)); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterWriteString() failed: %d", r); return (ARCHIVE_FATAL); } } r = xmlTextWriterEndElement(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterEndElement() failed: %d", r); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } static int xmlwrite_fstring(struct archive_write *a, xmlTextWriterPtr writer, const char *key, const char *fmt, ...) { struct xar *xar; va_list ap; xar = (struct xar *)a->format_data; va_start(ap, fmt); archive_string_empty(&xar->vstr); archive_string_vsprintf(&xar->vstr, fmt, ap); va_end(ap); return (xmlwrite_string(a, writer, key, xar->vstr.s)); } static int xmlwrite_time(struct archive_write *a, xmlTextWriterPtr writer, const char *key, time_t t, int z) { char timestr[100]; struct tm tm; #if defined(HAVE_GMTIME_R) gmtime_r(&t, &tm); #elif defined(HAVE__GMTIME64_S) _gmtime64_s(&tm, &t); #else memcpy(&tm, gmtime(&t), sizeof(tm)); #endif memset(×tr, 0, sizeof(timestr)); /* Do not use %F and %T for portability. */ strftime(timestr, sizeof(timestr), "%Y-%m-%dT%H:%M:%S", &tm); if (z) strcat(timestr, "Z"); return (xmlwrite_string(a, writer, key, timestr)); } static int xmlwrite_mode(struct archive_write *a, xmlTextWriterPtr writer, const char *key, mode_t mode) { char ms[5]; ms[0] = '0'; ms[1] = '0' + ((mode >> 6) & 07); ms[2] = '0' + ((mode >> 3) & 07); ms[3] = '0' + (mode & 07); ms[4] = '\0'; return (xmlwrite_string(a, writer, key, ms)); } static int xmlwrite_sum(struct archive_write *a, xmlTextWriterPtr writer, const char *key, struct chksumval *sum) { const char *algname; int algsize; char buff[MAX_SUM_SIZE*2 + 1]; char *p; unsigned char *s; int i, r; if (sum->len > 0) { algname = getalgname(sum->alg); algsize = getalgsize(sum->alg); if (algname != NULL) { const char *hex = "0123456789abcdef"; p = buff; s = sum->val; for (i = 0; i < algsize; i++) { *p++ = hex[(*s >> 4)]; *p++ = hex[(*s & 0x0f)]; s++; } *p = '\0'; r = xmlwrite_string_attr(a, writer, key, buff, "style", algname); if (r < 0) return (ARCHIVE_FATAL); } } return (ARCHIVE_OK); } static int xmlwrite_heap(struct archive_write *a, xmlTextWriterPtr writer, struct heap_data *heap) { const char *encname; int r; r = xmlwrite_fstring(a, writer, "length", "%ju", heap->length); if (r < 0) return (ARCHIVE_FATAL); r = xmlwrite_fstring(a, writer, "offset", "%ju", heap->temp_offset); if (r < 0) return (ARCHIVE_FATAL); r = xmlwrite_fstring(a, writer, "size", "%ju", heap->size); if (r < 0) return (ARCHIVE_FATAL); switch (heap->compression) { case GZIP: encname = "application/x-gzip"; break; case BZIP2: encname = "application/x-bzip2"; break; case LZMA: encname = "application/x-lzma"; break; case XZ: encname = "application/x-xz"; break; default: encname = "application/octet-stream"; break; } r = xmlwrite_string_attr(a, writer, "encoding", NULL, "style", encname); if (r < 0) return (ARCHIVE_FATAL); r = xmlwrite_sum(a, writer, "archived-checksum", &(heap->a_sum)); if (r < 0) return (ARCHIVE_FATAL); r = xmlwrite_sum(a, writer, "extracted-checksum", &(heap->e_sum)); if (r < 0) return (ARCHIVE_FATAL); return (ARCHIVE_OK); } /* * xar utility records fflags as following xml elements: * * * ..... * * or * * * ..... * * If xar is running on BSD platform, records ..; * if xar is running on linux platform, records ..; * otherwise does not record. * * Our implements records both and if it's necessary. */ static int make_fflags_entry(struct archive_write *a, xmlTextWriterPtr writer, const char *element, const char *fflags_text) { static const struct flagentry { const char *name; const char *xarname; } flagbsd[] = { { "sappnd", "SystemAppend"}, { "sappend", "SystemAppend"}, { "arch", "SystemArchived"}, { "archived", "SystemArchived"}, { "schg", "SystemImmutable"}, { "schange", "SystemImmutable"}, { "simmutable", "SystemImmutable"}, { "nosunlnk", "SystemNoUnlink"}, { "nosunlink", "SystemNoUnlink"}, { "snapshot", "SystemSnapshot"}, { "uappnd", "UserAppend"}, { "uappend", "UserAppend"}, { "uchg", "UserImmutable"}, { "uchange", "UserImmutable"}, { "uimmutable", "UserImmutable"}, { "nodump", "UserNoDump"}, { "noopaque", "UserOpaque"}, { "nouunlnk", "UserNoUnlink"}, { "nouunlink", "UserNoUnlink"}, { NULL, NULL} }, flagext2[] = { { "sappnd", "AppendOnly"}, { "sappend", "AppendOnly"}, { "schg", "Immutable"}, { "schange", "Immutable"}, { "simmutable", "Immutable"}, { "nodump", "NoDump"}, { "nouunlnk", "Undelete"}, { "nouunlink", "Undelete"}, { "btree", "BTree"}, { "comperr", "CompError"}, { "compress", "Compress"}, { "noatime", "NoAtime"}, { "compdirty", "CompDirty"}, { "comprblk", "CompBlock"}, { "dirsync", "DirSync"}, { "hashidx", "HashIndexed"}, { "imagic", "iMagic"}, { "journal", "Journaled"}, { "securedeletion", "SecureDeletion"}, { "sync", "Synchronous"}, { "notail", "NoTail"}, { "topdir", "TopDir"}, { "reserved", "Reserved"}, { NULL, NULL} }; const struct flagentry *fe, *flagentry; #define FLAGENTRY_MAXSIZE ((sizeof(flagbsd)+sizeof(flagext2))/sizeof(flagbsd)) const struct flagentry *avail[FLAGENTRY_MAXSIZE]; const char *p; int i, n, r; if (strcmp(element, "ext2") == 0) flagentry = flagext2; else flagentry = flagbsd; n = 0; p = fflags_text; do { const char *cp; cp = strchr(p, ','); if (cp == NULL) cp = p + strlen(p); for (fe = flagentry; fe->name != NULL; fe++) { if (fe->name[cp - p] != '\0' || p[0] != fe->name[0]) continue; if (strncmp(p, fe->name, cp - p) == 0) { avail[n++] = fe; break; } } if (*cp == ',') p = cp + 1; else p = NULL; } while (p != NULL); if (n > 0) { r = xmlTextWriterStartElement(writer, BAD_CAST_CONST(element)); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterStartElement() failed: %d", r); return (ARCHIVE_FATAL); } for (i = 0; i < n; i++) { r = xmlwrite_string(a, writer, avail[i]->xarname, NULL); if (r != ARCHIVE_OK) return (r); } r = xmlTextWriterEndElement(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterEndElement() failed: %d", r); return (ARCHIVE_FATAL); } } return (ARCHIVE_OK); } static int make_file_entry(struct archive_write *a, xmlTextWriterPtr writer, struct file *file) { struct xar *xar; const char *filetype, *filelink, *fflags; struct archive_string linkto; struct heap_data *heap; unsigned char *tmp; const char *p; size_t len; int r, r2, l, ll; xar = (struct xar *)a->format_data; r2 = ARCHIVE_OK; /* * Make a file name entry, "". */ l = ll = archive_strlen(&(file->basename)); tmp = malloc(l); if (tmp == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } r = UTF8Toisolat1(tmp, &l, BAD_CAST(file->basename.s), &ll); free(tmp); if (r < 0) { r = xmlTextWriterStartElement(writer, BAD_CAST("name")); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterStartElement() failed: %d", r); return (ARCHIVE_FATAL); } r = xmlTextWriterWriteAttribute(writer, BAD_CAST("enctype"), BAD_CAST("base64")); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterWriteAttribute() failed: %d", r); return (ARCHIVE_FATAL); } r = xmlTextWriterWriteBase64(writer, file->basename.s, 0, archive_strlen(&(file->basename))); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterWriteBase64() failed: %d", r); return (ARCHIVE_FATAL); } r = xmlTextWriterEndElement(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterEndElement() failed: %d", r); return (ARCHIVE_FATAL); } } else { r = xmlwrite_string(a, writer, "name", file->basename.s); if (r < 0) return (ARCHIVE_FATAL); } /* * Make a file type entry, "". */ filelink = NULL; archive_string_init(&linkto); switch (archive_entry_filetype(file->entry)) { case AE_IFDIR: filetype = "directory"; break; case AE_IFLNK: filetype = "symlink"; break; case AE_IFCHR: filetype = "character special"; break; case AE_IFBLK: filetype = "block special"; break; case AE_IFSOCK: filetype = "socket"; break; case AE_IFIFO: filetype = "fifo"; break; case AE_IFREG: default: if (file->hardlink_target != NULL) { filetype = "hardlink"; filelink = "link"; if (file->hardlink_target == file) archive_strcpy(&linkto, "original"); else archive_string_sprintf(&linkto, "%d", file->hardlink_target->id); } else filetype = "file"; break; } r = xmlwrite_string_attr(a, writer, "type", filetype, filelink, linkto.s); archive_string_free(&linkto); if (r < 0) return (ARCHIVE_FATAL); /* * On a virtual directory, we record "name" and "type" only. */ if (file->virtual) return (ARCHIVE_OK); switch (archive_entry_filetype(file->entry)) { case AE_IFLNK: /* * xar utility has checked a file type, which * a symbolic-link file has referenced. * For example: * ../ref/ * The symlink target file is "../ref/" and its * file type is a directory. * * ../f * The symlink target file is "../f" and its * file type is a regular file. * * But our implementation cannot do it, and then we * always record that a attribute "type" is "broken", * for example: * foo/bar * It means "foo/bar" is not reachable. */ r = xmlwrite_string_attr(a, writer, "link", file->symlink.s, "type", "broken"); if (r < 0) return (ARCHIVE_FATAL); break; case AE_IFCHR: case AE_IFBLK: r = xmlTextWriterStartElement(writer, BAD_CAST("device")); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterStartElement() failed: %d", r); return (ARCHIVE_FATAL); } r = xmlwrite_fstring(a, writer, "major", "%d", archive_entry_rdevmajor(file->entry)); if (r < 0) return (ARCHIVE_FATAL); r = xmlwrite_fstring(a, writer, "minor", "%d", archive_entry_rdevminor(file->entry)); if (r < 0) return (ARCHIVE_FATAL); r = xmlTextWriterEndElement(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterEndElement() failed: %d", r); return (ARCHIVE_FATAL); } break; default: break; } /* * Make a inode entry, "". */ r = xmlwrite_fstring(a, writer, "inode", "%jd", archive_entry_ino64(file->entry)); if (r < 0) return (ARCHIVE_FATAL); if (archive_entry_dev(file->entry) != 0) { r = xmlwrite_fstring(a, writer, "deviceno", "%d", archive_entry_dev(file->entry)); if (r < 0) return (ARCHIVE_FATAL); } /* * Make a file mode entry, "". */ r = xmlwrite_mode(a, writer, "mode", archive_entry_mode(file->entry)); if (r < 0) return (ARCHIVE_FATAL); /* * Make a user entry, "" and ". */ r = xmlwrite_fstring(a, writer, "uid", "%d", archive_entry_uid(file->entry)); if (r < 0) return (ARCHIVE_FATAL); r = archive_entry_uname_l(file->entry, &p, &len, xar->sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Uname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate uname '%s' to UTF-8", archive_entry_uname(file->entry)); r2 = ARCHIVE_WARN; } if (len > 0) { r = xmlwrite_string(a, writer, "user", p); if (r < 0) return (ARCHIVE_FATAL); } /* * Make a group entry, "" and ". */ r = xmlwrite_fstring(a, writer, "gid", "%d", archive_entry_gid(file->entry)); if (r < 0) return (ARCHIVE_FATAL); r = archive_entry_gname_l(file->entry, &p, &len, xar->sconv); if (r != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Gname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate gname '%s' to UTF-8", archive_entry_gname(file->entry)); r2 = ARCHIVE_WARN; } if (len > 0) { r = xmlwrite_string(a, writer, "group", p); if (r < 0) return (ARCHIVE_FATAL); } /* * Make a ctime entry, "". */ if (archive_entry_ctime_is_set(file->entry)) { r = xmlwrite_time(a, writer, "ctime", archive_entry_ctime(file->entry), 1); if (r < 0) return (ARCHIVE_FATAL); } /* * Make a mtime entry, "". */ if (archive_entry_mtime_is_set(file->entry)) { r = xmlwrite_time(a, writer, "mtime", archive_entry_mtime(file->entry), 1); if (r < 0) return (ARCHIVE_FATAL); } /* * Make a atime entry, "". */ if (archive_entry_atime_is_set(file->entry)) { r = xmlwrite_time(a, writer, "atime", archive_entry_atime(file->entry), 1); if (r < 0) return (ARCHIVE_FATAL); } /* * Make fflags entries, "" and "". */ fflags = archive_entry_fflags_text(file->entry); if (fflags != NULL) { r = make_fflags_entry(a, writer, "flags", fflags); if (r < 0) return (r); r = make_fflags_entry(a, writer, "ext2", fflags); if (r < 0) return (r); } /* * Make extended attribute entries, "". */ archive_entry_xattr_reset(file->entry); for (heap = file->xattr.first; heap != NULL; heap = heap->next) { const char *name; const void *value; size_t size; archive_entry_xattr_next(file->entry, &name, &value, &size); r = xmlTextWriterStartElement(writer, BAD_CAST("ea")); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterStartElement() failed: %d", r); return (ARCHIVE_FATAL); } r = xmlTextWriterWriteFormatAttribute(writer, BAD_CAST("id"), "%d", heap->id); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterWriteAttribute() failed: %d", r); return (ARCHIVE_FATAL); } r = xmlwrite_heap(a, writer, heap); if (r < 0) return (ARCHIVE_FATAL); r = xmlwrite_string(a, writer, "name", name); if (r < 0) return (ARCHIVE_FATAL); r = xmlTextWriterEndElement(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterEndElement() failed: %d", r); return (ARCHIVE_FATAL); } } /* * Make a file data entry, "". */ if (file->data.length > 0) { r = xmlTextWriterStartElement(writer, BAD_CAST("data")); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterStartElement() failed: %d", r); return (ARCHIVE_FATAL); } r = xmlwrite_heap(a, writer, &(file->data)); if (r < 0) return (ARCHIVE_FATAL); r = xmlTextWriterEndElement(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterEndElement() failed: %d", r); return (ARCHIVE_FATAL); } } if (archive_strlen(&file->script) > 0) { r = xmlTextWriterStartElement(writer, BAD_CAST("content")); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterStartElement() failed: %d", r); return (ARCHIVE_FATAL); } r = xmlwrite_string(a, writer, "interpreter", file->script.s); if (r < 0) return (ARCHIVE_FATAL); r = xmlwrite_string(a, writer, "type", "script"); if (r < 0) return (ARCHIVE_FATAL); r = xmlTextWriterEndElement(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterEndElement() failed: %d", r); return (ARCHIVE_FATAL); } } return (r2); } /* * Make the TOC */ static int make_toc(struct archive_write *a) { struct xar *xar; struct file *np; xmlBufferPtr bp; xmlTextWriterPtr writer; int algsize; int r, ret; xar = (struct xar *)a->format_data; ret = ARCHIVE_FATAL; /* * Initialize xml writer. */ writer = NULL; bp = xmlBufferCreate(); if (bp == NULL) { archive_set_error(&a->archive, ENOMEM, "xmlBufferCreate() " "couldn't create xml buffer"); goto exit_toc; } writer = xmlNewTextWriterMemory(bp, 0); if (writer == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlNewTextWriterMemory() " "couldn't create xml writer"); goto exit_toc; } r = xmlTextWriterStartDocument(writer, "1.0", "UTF-8", NULL); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterStartDocument() failed: %d", r); goto exit_toc; } r = xmlTextWriterSetIndent(writer, 4); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterSetIndent() failed: %d", r); goto exit_toc; } /* * Start recording TOC */ r = xmlTextWriterStartElement(writer, BAD_CAST("xar")); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterStartElement() failed: %d", r); goto exit_toc; } r = xmlTextWriterStartElement(writer, BAD_CAST("toc")); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterStartDocument() failed: %d", r); goto exit_toc; } /* * Record the creation time of the archive file. */ r = xmlwrite_time(a, writer, "creation-time", time(NULL), 0); if (r < 0) goto exit_toc; /* * Record the checksum value of TOC */ algsize = getalgsize(xar->opt_toc_sumalg); if (algsize) { /* * Record TOC checksum */ r = xmlTextWriterStartElement(writer, BAD_CAST("checksum")); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterStartElement() failed: %d", r); goto exit_toc; } r = xmlTextWriterWriteAttribute(writer, BAD_CAST("style"), BAD_CAST_CONST(getalgname(xar->opt_toc_sumalg))); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterWriteAttribute() failed: %d", r); goto exit_toc; } /* * Record the offset of the value of checksum of TOC */ r = xmlwrite_string(a, writer, "offset", "0"); if (r < 0) goto exit_toc; /* * Record the size of the value of checksum of TOC */ r = xmlwrite_fstring(a, writer, "size", "%d", algsize); if (r < 0) goto exit_toc; r = xmlTextWriterEndElement(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterEndElement() failed: %d", r); goto exit_toc; } } np = xar->root; do { if (np != np->parent) { r = make_file_entry(a, writer, np); if (r != ARCHIVE_OK) goto exit_toc; } if (np->dir && np->children.first != NULL) { /* Enter to sub directories. */ np = np->children.first; r = xmlTextWriterStartElement(writer, BAD_CAST("file")); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterStartElement() " "failed: %d", r); goto exit_toc; } r = xmlTextWriterWriteFormatAttribute( writer, BAD_CAST("id"), "%d", np->id); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterWriteAttribute() " "failed: %d", r); goto exit_toc; } continue; } while (np != np->parent) { r = xmlTextWriterEndElement(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterEndElement() " "failed: %d", r); goto exit_toc; } if (np->chnext == NULL) { /* Return to the parent directory. */ np = np->parent; } else { np = np->chnext; r = xmlTextWriterStartElement(writer, BAD_CAST("file")); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterStartElement() " "failed: %d", r); goto exit_toc; } r = xmlTextWriterWriteFormatAttribute( writer, BAD_CAST("id"), "%d", np->id); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterWriteAttribute() " "failed: %d", r); goto exit_toc; } break; } } } while (np != np->parent); r = xmlTextWriterEndDocument(writer); if (r < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "xmlTextWriterEndDocument() failed: %d", r); goto exit_toc; } #if DEBUG_PRINT_TOC fprintf(stderr, "\n---TOC-- %d bytes --\n%s\n", strlen((const char *)bp->content), bp->content); #endif /* * Compress the TOC and calculate the sum of the TOC. */ xar->toc.temp_offset = xar->temp_offset; xar->toc.size = bp->use; checksum_init(&(xar->a_sumwrk), xar->opt_toc_sumalg); r = compression_init_encoder_gzip(&(a->archive), &(xar->stream), 6, 1); if (r != ARCHIVE_OK) goto exit_toc; xar->stream.next_in = bp->content; xar->stream.avail_in = bp->use; xar->stream.total_in = 0; xar->stream.next_out = xar->wbuff; xar->stream.avail_out = sizeof(xar->wbuff); xar->stream.total_out = 0; for (;;) { size_t size; r = compression_code(&(a->archive), &(xar->stream), ARCHIVE_Z_FINISH); if (r != ARCHIVE_OK && r != ARCHIVE_EOF) goto exit_toc; size = sizeof(xar->wbuff) - xar->stream.avail_out; checksum_update(&(xar->a_sumwrk), xar->wbuff, size); if (write_to_temp(a, xar->wbuff, size) != ARCHIVE_OK) goto exit_toc; if (r == ARCHIVE_EOF) break; xar->stream.next_out = xar->wbuff; xar->stream.avail_out = sizeof(xar->wbuff); } r = compression_end(&(a->archive), &(xar->stream)); if (r != ARCHIVE_OK) goto exit_toc; xar->toc.length = xar->stream.total_out; xar->toc.compression = GZIP; checksum_final(&(xar->a_sumwrk), &(xar->toc.a_sum)); ret = ARCHIVE_OK; exit_toc: if (writer) xmlFreeTextWriter(writer); if (bp) xmlBufferFree(bp); return (ret); } static int flush_wbuff(struct archive_write *a) { struct xar *xar; int r; size_t s; xar = (struct xar *)a->format_data; s = sizeof(xar->wbuff) - xar->wbuff_remaining; r = __archive_write_output(a, xar->wbuff, s); if (r != ARCHIVE_OK) return (r); xar->wbuff_remaining = sizeof(xar->wbuff); return (r); } static int copy_out(struct archive_write *a, uint64_t offset, uint64_t length) { struct xar *xar; int r; xar = (struct xar *)a->format_data; if (lseek(xar->temp_fd, offset, SEEK_SET) < 0) { archive_set_error(&(a->archive), errno, "lseek failed"); return (ARCHIVE_FATAL); } while (length) { size_t rsize; ssize_t rs; unsigned char *wb; if (length > xar->wbuff_remaining) rsize = xar->wbuff_remaining; else rsize = (size_t)length; wb = xar->wbuff + (sizeof(xar->wbuff) - xar->wbuff_remaining); rs = read(xar->temp_fd, wb, rsize); if (rs < 0) { archive_set_error(&(a->archive), errno, "Can't read temporary file(%jd)", (intmax_t)rs); return (ARCHIVE_FATAL); } if (rs == 0) { archive_set_error(&(a->archive), 0, "Truncated xar archive"); return (ARCHIVE_FATAL); } xar->wbuff_remaining -= rs; length -= rs; if (xar->wbuff_remaining == 0) { r = flush_wbuff(a); if (r != ARCHIVE_OK) return (r); } } return (ARCHIVE_OK); } static int xar_close(struct archive_write *a) { struct xar *xar; unsigned char *wb; uint64_t length; int r; xar = (struct xar *)a->format_data; /* Empty! */ if (xar->root->children.first == NULL) return (ARCHIVE_OK); /* Save the length of all file extended attributes and contents. */ length = xar->temp_offset; /* Connect hardlinked files */ file_connect_hardlink_files(xar); /* Make the TOC */ r = make_toc(a); if (r != ARCHIVE_OK) return (r); /* * Make the xar header on wbuff(write buffer). */ wb = xar->wbuff; xar->wbuff_remaining = sizeof(xar->wbuff); archive_be32enc(&wb[0], HEADER_MAGIC); archive_be16enc(&wb[4], HEADER_SIZE); archive_be16enc(&wb[6], HEADER_VERSION); archive_be64enc(&wb[8], xar->toc.length); archive_be64enc(&wb[16], xar->toc.size); archive_be32enc(&wb[24], xar->toc.a_sum.alg); xar->wbuff_remaining -= HEADER_SIZE; /* * Write the TOC */ r = copy_out(a, xar->toc.temp_offset, xar->toc.length); if (r != ARCHIVE_OK) return (r); /* Write the checksum value of the TOC. */ if (xar->toc.a_sum.len) { if (xar->wbuff_remaining < xar->toc.a_sum.len) { r = flush_wbuff(a); if (r != ARCHIVE_OK) return (r); } wb = xar->wbuff + (sizeof(xar->wbuff) - xar->wbuff_remaining); memcpy(wb, xar->toc.a_sum.val, xar->toc.a_sum.len); xar->wbuff_remaining -= xar->toc.a_sum.len; } /* * Write all file extended attributes and contents. */ r = copy_out(a, xar->toc.a_sum.len, length); if (r != ARCHIVE_OK) return (r); r = flush_wbuff(a); return (r); } static int xar_free(struct archive_write *a) { struct xar *xar; xar = (struct xar *)a->format_data; /* Close the temporary file. */ if (xar->temp_fd >= 0) close(xar->temp_fd); archive_string_free(&(xar->cur_dirstr)); archive_string_free(&(xar->tstr)); archive_string_free(&(xar->vstr)); file_free_hardlinks(xar); file_free_register(xar); compression_end(&(a->archive), &(xar->stream)); free(xar); return (ARCHIVE_OK); } static int file_cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) { const struct file *f1 = (const struct file *)n1; const struct file *f2 = (const struct file *)n2; return (strcmp(f1->basename.s, f2->basename.s)); } static int file_cmp_key(const struct archive_rb_node *n, const void *key) { const struct file *f = (const struct file *)n; return (strcmp(f->basename.s, (const char *)key)); } static struct file * file_new(struct archive_write *a, struct archive_entry *entry) { struct file *file; static const struct archive_rb_tree_ops rb_ops = { file_cmp_node, file_cmp_key }; file = calloc(1, sizeof(*file)); if (file == NULL) return (NULL); if (entry != NULL) file->entry = archive_entry_clone(entry); else file->entry = archive_entry_new2(&a->archive); if (file->entry == NULL) { free(file); return (NULL); } __archive_rb_tree_init(&(file->rbtree), &rb_ops); file->children.first = NULL; file->children.last = &(file->children.first); file->xattr.first = NULL; file->xattr.last = &(file->xattr.first); archive_string_init(&(file->parentdir)); archive_string_init(&(file->basename)); archive_string_init(&(file->symlink)); archive_string_init(&(file->script)); if (entry != NULL && archive_entry_filetype(entry) == AE_IFDIR) file->dir = 1; return (file); } static void file_free(struct file *file) { struct heap_data *heap, *next_heap; heap = file->xattr.first; while (heap != NULL) { next_heap = heap->next; free(heap); heap = next_heap; } archive_string_free(&(file->parentdir)); archive_string_free(&(file->basename)); archive_string_free(&(file->symlink)); archive_string_free(&(file->script)); free(file); } static struct file * file_create_virtual_dir(struct archive_write *a, struct xar *xar, const char *pathname) { struct file *file; (void)xar; /* UNUSED */ file = file_new(a, NULL); if (file == NULL) return (NULL); archive_entry_set_pathname(file->entry, pathname); archive_entry_set_mode(file->entry, 0555 | AE_IFDIR); file->dir = 1; file->virtual = 1; return (file); } static int file_add_child_tail(struct file *parent, struct file *child) { if (!__archive_rb_tree_insert_node( &(parent->rbtree), (struct archive_rb_node *)child)) return (0); child->chnext = NULL; *parent->children.last = child; parent->children.last = &(child->chnext); child->parent = parent; return (1); } /* * Find a entry from `parent' */ static struct file * file_find_child(struct file *parent, const char *child_name) { struct file *np; np = (struct file *)__archive_rb_tree_find_node( &(parent->rbtree), child_name); return (np); } #if defined(_WIN32) || defined(__CYGWIN__) static void cleanup_backslash(char *utf8, size_t len) { /* Convert a path-separator from '\' to '/' */ while (*utf8 != '\0' && len) { if (*utf8 == '\\') *utf8 = '/'; ++utf8; --len; } } #else #define cleanup_backslash(p, len) /* nop */ #endif /* * Generate a parent directory name and a base name from a pathname. */ static int file_gen_utility_names(struct archive_write *a, struct file *file) { struct xar *xar; const char *pp; char *p, *dirname, *slash; size_t len; int r = ARCHIVE_OK; xar = (struct xar *)a->format_data; archive_string_empty(&(file->parentdir)); archive_string_empty(&(file->basename)); archive_string_empty(&(file->symlink)); if (file->parent == file)/* virtual root */ return (ARCHIVE_OK); if (archive_entry_pathname_l(file->entry, &pp, &len, xar->sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate pathname '%s' to UTF-8", archive_entry_pathname(file->entry)); r = ARCHIVE_WARN; } archive_strncpy(&(file->parentdir), pp, len); len = file->parentdir.length; p = dirname = file->parentdir.s; /* * Convert a path-separator from '\' to '/' */ cleanup_backslash(p, len); /* * Remove leading '/', '../' and './' elements */ while (*p) { if (p[0] == '/') { p++; len--; } else if (p[0] != '.') break; else if (p[1] == '.' && p[2] == '/') { p += 3; len -= 3; } else if (p[1] == '/' || (p[1] == '.' && p[2] == '\0')) { p += 2; len -= 2; } else if (p[1] == '\0') { p++; len--; } else break; } if (p != dirname) { memmove(dirname, p, len+1); p = dirname; } /* * Remove "/","/." and "/.." elements from tail. */ while (len > 0) { size_t ll = len; if (len > 0 && p[len-1] == '/') { p[len-1] = '\0'; len--; } if (len > 1 && p[len-2] == '/' && p[len-1] == '.') { p[len-2] = '\0'; len -= 2; } if (len > 2 && p[len-3] == '/' && p[len-2] == '.' && p[len-1] == '.') { p[len-3] = '\0'; len -= 3; } if (ll == len) break; } while (*p) { if (p[0] == '/') { if (p[1] == '/') /* Convert '//' --> '/' */ strcpy(p, p+1); else if (p[1] == '.' && p[2] == '/') /* Convert '/./' --> '/' */ strcpy(p, p+2); else if (p[1] == '.' && p[2] == '.' && p[3] == '/') { /* Convert 'dir/dir1/../dir2/' * --> 'dir/dir2/' */ char *rp = p -1; while (rp >= dirname) { if (*rp == '/') break; --rp; } if (rp > dirname) { strcpy(rp, p+3); p = rp; } else { strcpy(dirname, p+4); p = dirname; } } else p++; } else p++; } p = dirname; len = strlen(p); if (archive_entry_filetype(file->entry) == AE_IFLNK) { size_t len2; /* Convert symlink name too. */ if (archive_entry_symlink_l(file->entry, &pp, &len2, xar->sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Linkname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate symlink '%s' to UTF-8", archive_entry_symlink(file->entry)); r = ARCHIVE_WARN; } archive_strncpy(&(file->symlink), pp, len2); cleanup_backslash(file->symlink.s, file->symlink.length); } /* * - Count up directory elements. * - Find out the position which points the last position of * path separator('/'). */ slash = NULL; for (; *p != '\0'; p++) if (*p == '/') slash = p; if (slash == NULL) { /* The pathname doesn't have a parent directory. */ file->parentdir.length = len; archive_string_copy(&(file->basename), &(file->parentdir)); archive_string_empty(&(file->parentdir)); *file->parentdir.s = '\0'; return (r); } /* Make a basename from dirname and slash */ *slash = '\0'; file->parentdir.length = slash - dirname; archive_strcpy(&(file->basename), slash + 1); return (r); } static int get_path_component(char *name, int n, const char *fn) { char *p; int l; p = strchr(fn, '/'); if (p == NULL) { if ((l = strlen(fn)) == 0) return (0); } else l = p - fn; if (l > n -1) return (-1); memcpy(name, fn, l); name[l] = '\0'; return (l); } /* * Add a new entry into the tree. */ static int file_tree(struct archive_write *a, struct file **filepp) { #if defined(_WIN32) && !defined(__CYGWIN__) char name[_MAX_FNAME];/* Included null terminator size. */ #elif defined(NAME_MAX) && NAME_MAX >= 255 char name[NAME_MAX+1]; #else char name[256]; #endif struct xar *xar = (struct xar *)a->format_data; struct file *dent, *file, *np; struct archive_entry *ent; const char *fn, *p; int l; file = *filepp; dent = xar->root; if (file->parentdir.length > 0) fn = p = file->parentdir.s; else fn = p = ""; /* * If the path of the parent directory of `file' entry is * the same as the path of `cur_dirent', add isoent to * `cur_dirent'. */ if (archive_strlen(&(xar->cur_dirstr)) == archive_strlen(&(file->parentdir)) && strcmp(xar->cur_dirstr.s, fn) == 0) { if (!file_add_child_tail(xar->cur_dirent, file)) { np = (struct file *)__archive_rb_tree_find_node( &(xar->cur_dirent->rbtree), file->basename.s); goto same_entry; } return (ARCHIVE_OK); } for (;;) { l = get_path_component(name, sizeof(name), fn); if (l == 0) { np = NULL; break; } if (l < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "A name buffer is too small"); file_free(file); *filepp = NULL; return (ARCHIVE_FATAL); } np = file_find_child(dent, name); if (np == NULL || fn[0] == '\0') break; /* Find next subdirectory. */ if (!np->dir) { /* NOT Directory! */ archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "`%s' is not directory, we cannot insert `%s' ", archive_entry_pathname(np->entry), archive_entry_pathname(file->entry)); file_free(file); *filepp = NULL; return (ARCHIVE_FAILED); } fn += l; if (fn[0] == '/') fn++; dent = np; } if (np == NULL) { /* * Create virtual parent directories. */ while (fn[0] != '\0') { struct file *vp; struct archive_string as; archive_string_init(&as); archive_strncat(&as, p, fn - p + l); if (as.s[as.length-1] == '/') { as.s[as.length-1] = '\0'; as.length--; } vp = file_create_virtual_dir(a, xar, as.s); if (vp == NULL) { archive_string_free(&as); archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); file_free(file); *filepp = NULL; return (ARCHIVE_FATAL); } archive_string_free(&as); if (file_gen_utility_names(a, vp) <= ARCHIVE_FAILED) return (ARCHIVE_FATAL); file_add_child_tail(dent, vp); file_register(xar, vp); np = vp; fn += l; if (fn[0] == '/') fn++; l = get_path_component(name, sizeof(name), fn); if (l < 0) { archive_string_free(&as); archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "A name buffer is too small"); file_free(file); *filepp = NULL; return (ARCHIVE_FATAL); } dent = np; } /* Found out the parent directory where isoent can be * inserted. */ xar->cur_dirent = dent; archive_string_empty(&(xar->cur_dirstr)); archive_string_ensure(&(xar->cur_dirstr), archive_strlen(&(dent->parentdir)) + archive_strlen(&(dent->basename)) + 2); if (archive_strlen(&(dent->parentdir)) + archive_strlen(&(dent->basename)) == 0) xar->cur_dirstr.s[0] = 0; else { if (archive_strlen(&(dent->parentdir)) > 0) { archive_string_copy(&(xar->cur_dirstr), &(dent->parentdir)); archive_strappend_char(&(xar->cur_dirstr), '/'); } archive_string_concat(&(xar->cur_dirstr), &(dent->basename)); } if (!file_add_child_tail(dent, file)) { np = (struct file *)__archive_rb_tree_find_node( &(dent->rbtree), file->basename.s); goto same_entry; } return (ARCHIVE_OK); } same_entry: /* * We have already has the entry the filename of which is * the same. */ if (archive_entry_filetype(np->entry) != archive_entry_filetype(file->entry)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Found duplicate entries `%s' and its file type is " "different", archive_entry_pathname(np->entry)); file_free(file); *filepp = NULL; return (ARCHIVE_FAILED); } /* Swap files. */ ent = np->entry; np->entry = file->entry; file->entry = ent; np->virtual = 0; file_free(file); *filepp = np; return (ARCHIVE_OK); } static void file_register(struct xar *xar, struct file *file) { file->id = xar->file_idx++; file->next = NULL; *xar->file_list.last = file; xar->file_list.last = &(file->next); } static void file_init_register(struct xar *xar) { xar->file_list.first = NULL; xar->file_list.last = &(xar->file_list.first); } static void file_free_register(struct xar *xar) { struct file *file, *file_next; file = xar->file_list.first; while (file != NULL) { file_next = file->next; file_free(file); file = file_next; } } /* * Register entry to get a hardlink target. */ static int file_register_hardlink(struct archive_write *a, struct file *file) { struct xar *xar = (struct xar *)a->format_data; struct hardlink *hl; const char *pathname; archive_entry_set_nlink(file->entry, 1); pathname = archive_entry_hardlink(file->entry); if (pathname == NULL) { /* This `file` is a hardlink target. */ hl = malloc(sizeof(*hl)); if (hl == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory"); return (ARCHIVE_FATAL); } hl->nlink = 1; /* A hardlink target must be the first position. */ file->hlnext = NULL; hl->file_list.first = file; hl->file_list.last = &(file->hlnext); __archive_rb_tree_insert_node(&(xar->hardlink_rbtree), (struct archive_rb_node *)hl); } else { hl = (struct hardlink *)__archive_rb_tree_find_node( &(xar->hardlink_rbtree), pathname); if (hl != NULL) { /* Insert `file` entry into the tail. */ file->hlnext = NULL; *hl->file_list.last = file; hl->file_list.last = &(file->hlnext); hl->nlink++; } archive_entry_unset_size(file->entry); } return (ARCHIVE_OK); } /* * Hardlinked files have to have the same location of extent. * We have to find out hardlink target entries for entries which * have a hardlink target name. */ static void file_connect_hardlink_files(struct xar *xar) { struct archive_rb_node *n; struct hardlink *hl; struct file *target, *nf; ARCHIVE_RB_TREE_FOREACH(n, &(xar->hardlink_rbtree)) { hl = (struct hardlink *)n; /* The first entry must be a hardlink target. */ target = hl->file_list.first; archive_entry_set_nlink(target->entry, hl->nlink); if (hl->nlink > 1) /* It means this file is a hardlink * target itself. */ target->hardlink_target = target; for (nf = target->hlnext; nf != NULL; nf = nf->hlnext) { nf->hardlink_target = target; archive_entry_set_nlink(nf->entry, hl->nlink); } } } static int file_hd_cmp_node(const struct archive_rb_node *n1, const struct archive_rb_node *n2) { const struct hardlink *h1 = (const struct hardlink *)n1; const struct hardlink *h2 = (const struct hardlink *)n2; return (strcmp(archive_entry_pathname(h1->file_list.first->entry), archive_entry_pathname(h2->file_list.first->entry))); } static int file_hd_cmp_key(const struct archive_rb_node *n, const void *key) { const struct hardlink *h = (const struct hardlink *)n; return (strcmp(archive_entry_pathname(h->file_list.first->entry), (const char *)key)); } static void file_init_hardlinks(struct xar *xar) { static const struct archive_rb_tree_ops rb_ops = { file_hd_cmp_node, file_hd_cmp_key, }; __archive_rb_tree_init(&(xar->hardlink_rbtree), &rb_ops); } static void file_free_hardlinks(struct xar *xar) { struct archive_rb_node *n, *next; for (n = ARCHIVE_RB_TREE_MIN(&(xar->hardlink_rbtree)); n;) { next = __archive_rb_tree_iterate(&(xar->hardlink_rbtree), n, ARCHIVE_RB_DIR_RIGHT); free(n); n = next; } } static void checksum_init(struct chksumwork *sumwrk, enum sumalg sum_alg) { sumwrk->alg = sum_alg; switch (sum_alg) { case CKSUM_NONE: break; case CKSUM_SHA1: archive_sha1_init(&(sumwrk->sha1ctx)); break; case CKSUM_MD5: archive_md5_init(&(sumwrk->md5ctx)); break; } } static void checksum_update(struct chksumwork *sumwrk, const void *buff, size_t size) { switch (sumwrk->alg) { case CKSUM_NONE: break; case CKSUM_SHA1: archive_sha1_update(&(sumwrk->sha1ctx), buff, size); break; case CKSUM_MD5: archive_md5_update(&(sumwrk->md5ctx), buff, size); break; } } static void checksum_final(struct chksumwork *sumwrk, struct chksumval *sumval) { switch (sumwrk->alg) { case CKSUM_NONE: sumval->len = 0; break; case CKSUM_SHA1: archive_sha1_final(&(sumwrk->sha1ctx), sumval->val); sumval->len = SHA1_SIZE; break; case CKSUM_MD5: archive_md5_final(&(sumwrk->md5ctx), sumval->val); sumval->len = MD5_SIZE; break; } sumval->alg = sumwrk->alg; } #if !defined(HAVE_BZLIB_H) || !defined(BZ_CONFIG_ERROR) || !defined(HAVE_LZMA_H) static int compression_unsupported_encoder(struct archive *a, struct la_zstream *lastrm, const char *name) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "%s compression not supported on this platform", name); lastrm->valid = 0; lastrm->real_stream = NULL; return (ARCHIVE_FAILED); } #endif static int compression_init_encoder_gzip(struct archive *a, struct la_zstream *lastrm, int level, int withheader) { z_stream *strm; if (lastrm->valid) compression_end(a, lastrm); strm = calloc(1, sizeof(*strm)); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for gzip stream"); return (ARCHIVE_FATAL); } /* zlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (Bytef *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = lastrm->avail_in; strm->total_in = (uLong)lastrm->total_in; strm->next_out = lastrm->next_out; strm->avail_out = lastrm->avail_out; strm->total_out = (uLong)lastrm->total_out; if (deflateInit2(strm, level, Z_DEFLATED, (withheader)?15:-15, 8, Z_DEFAULT_STRATEGY) != Z_OK) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library"); return (ARCHIVE_FATAL); } lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_gzip; lastrm->end = compression_end_gzip; return (ARCHIVE_OK); } static int compression_code_gzip(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { z_stream *strm; int r; strm = (z_stream *)lastrm->real_stream; /* zlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (Bytef *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = lastrm->avail_in; strm->total_in = (uLong)lastrm->total_in; strm->next_out = lastrm->next_out; strm->avail_out = lastrm->avail_out; strm->total_out = (uLong)lastrm->total_out; r = deflate(strm, (action == ARCHIVE_Z_FINISH)? Z_FINISH: Z_NO_FLUSH); lastrm->next_in = strm->next_in; lastrm->avail_in = strm->avail_in; lastrm->total_in = strm->total_in; lastrm->next_out = strm->next_out; lastrm->avail_out = strm->avail_out; lastrm->total_out = strm->total_out; switch (r) { case Z_OK: return (ARCHIVE_OK); case Z_STREAM_END: return (ARCHIVE_EOF); default: archive_set_error(a, ARCHIVE_ERRNO_MISC, "GZip compression failed:" " deflate() call returned status %d", r); return (ARCHIVE_FATAL); } } static int compression_end_gzip(struct archive *a, struct la_zstream *lastrm) { z_stream *strm; int r; strm = (z_stream *)lastrm->real_stream; r = deflateEnd(strm); free(strm); lastrm->real_stream = NULL; lastrm->valid = 0; if (r != Z_OK) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "Failed to clean up compressor"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } #if defined(HAVE_BZLIB_H) && defined(BZ_CONFIG_ERROR) static int compression_init_encoder_bzip2(struct archive *a, struct la_zstream *lastrm, int level) { bz_stream *strm; if (lastrm->valid) compression_end(a, lastrm); strm = calloc(1, sizeof(*strm)); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for bzip2 stream"); return (ARCHIVE_FATAL); } /* bzlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (char *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = lastrm->avail_in; strm->total_in_lo32 = (uint32_t)(lastrm->total_in & 0xffffffff); strm->total_in_hi32 = (uint32_t)(lastrm->total_in >> 32); strm->next_out = (char *)lastrm->next_out; strm->avail_out = lastrm->avail_out; strm->total_out_lo32 = (uint32_t)(lastrm->total_out & 0xffffffff); strm->total_out_hi32 = (uint32_t)(lastrm->total_out >> 32); if (BZ2_bzCompressInit(strm, level, 0, 30) != BZ_OK) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library"); return (ARCHIVE_FATAL); } lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_bzip2; lastrm->end = compression_end_bzip2; return (ARCHIVE_OK); } static int compression_code_bzip2(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { bz_stream *strm; int r; strm = (bz_stream *)lastrm->real_stream; /* bzlib.h is not const-correct, so we need this one bit * of ugly hackery to convert a const * pointer to * a non-const pointer. */ strm->next_in = (char *)(uintptr_t)(const void *)lastrm->next_in; strm->avail_in = lastrm->avail_in; strm->total_in_lo32 = (uint32_t)(lastrm->total_in & 0xffffffff); strm->total_in_hi32 = (uint32_t)(lastrm->total_in >> 32); strm->next_out = (char *)lastrm->next_out; strm->avail_out = lastrm->avail_out; strm->total_out_lo32 = (uint32_t)(lastrm->total_out & 0xffffffff); strm->total_out_hi32 = (uint32_t)(lastrm->total_out >> 32); r = BZ2_bzCompress(strm, (action == ARCHIVE_Z_FINISH)? BZ_FINISH: BZ_RUN); lastrm->next_in = (const unsigned char *)strm->next_in; lastrm->avail_in = strm->avail_in; lastrm->total_in = (((uint64_t)(uint32_t)strm->total_in_hi32) << 32) + (uint64_t)(uint32_t)strm->total_in_lo32; lastrm->next_out = (unsigned char *)strm->next_out; lastrm->avail_out = strm->avail_out; lastrm->total_out = (((uint64_t)(uint32_t)strm->total_out_hi32) << 32) + (uint64_t)(uint32_t)strm->total_out_lo32; switch (r) { case BZ_RUN_OK: /* Non-finishing */ case BZ_FINISH_OK: /* Finishing: There's more work to do */ return (ARCHIVE_OK); case BZ_STREAM_END: /* Finishing: all done */ /* Only occurs in finishing case */ return (ARCHIVE_EOF); default: /* Any other return value indicates an error */ archive_set_error(a, ARCHIVE_ERRNO_MISC, "Bzip2 compression failed:" " BZ2_bzCompress() call returned status %d", r); return (ARCHIVE_FATAL); } } static int compression_end_bzip2(struct archive *a, struct la_zstream *lastrm) { bz_stream *strm; int r; strm = (bz_stream *)lastrm->real_stream; r = BZ2_bzCompressEnd(strm); free(strm); lastrm->real_stream = NULL; lastrm->valid = 0; if (r != BZ_OK) { archive_set_error(a, ARCHIVE_ERRNO_MISC, "Failed to clean up compressor"); return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } #else static int compression_init_encoder_bzip2(struct archive *a, struct la_zstream *lastrm, int level) { (void) level; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "bzip2")); } #endif #if defined(HAVE_LZMA_H) static int compression_init_encoder_lzma(struct archive *a, struct la_zstream *lastrm, int level) { static const lzma_stream lzma_init_data = LZMA_STREAM_INIT; lzma_stream *strm; lzma_options_lzma lzma_opt; int r; if (lastrm->valid) compression_end(a, lastrm); if (lzma_lzma_preset(&lzma_opt, level)) { lastrm->real_stream = NULL; archive_set_error(a, ENOMEM, "Internal error initializing compression library"); return (ARCHIVE_FATAL); } strm = calloc(1, sizeof(*strm)); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for lzma stream"); return (ARCHIVE_FATAL); } *strm = lzma_init_data; r = lzma_alone_encoder(strm, &lzma_opt); switch (r) { case LZMA_OK: lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_lzma; lastrm->end = compression_end_lzma; r = ARCHIVE_OK; break; case LZMA_MEM_ERROR: free(strm); lastrm->real_stream = NULL; archive_set_error(a, ENOMEM, "Internal error initializing compression library: " "Cannot allocate memory"); r = ARCHIVE_FATAL; break; default: free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library: " "It's a bug in liblzma"); r = ARCHIVE_FATAL; break; } return (r); } static int compression_init_encoder_xz(struct archive *a, struct la_zstream *lastrm, int level, int threads) { static const lzma_stream lzma_init_data = LZMA_STREAM_INIT; lzma_stream *strm; lzma_filter *lzmafilters; lzma_options_lzma lzma_opt; int r; #ifdef HAVE_LZMA_STREAM_ENCODER_MT lzma_mt mt_options; #endif (void)threads; /* UNUSED (if multi-threaded LZMA library not avail) */ if (lastrm->valid) compression_end(a, lastrm); strm = calloc(1, sizeof(*strm) + sizeof(*lzmafilters) * 2); if (strm == NULL) { archive_set_error(a, ENOMEM, "Can't allocate memory for xz stream"); return (ARCHIVE_FATAL); } lzmafilters = (lzma_filter *)(strm+1); if (level > 6) level = 6; if (lzma_lzma_preset(&lzma_opt, level)) { free(strm); lastrm->real_stream = NULL; archive_set_error(a, ENOMEM, "Internal error initializing compression library"); return (ARCHIVE_FATAL); } lzmafilters[0].id = LZMA_FILTER_LZMA2; lzmafilters[0].options = &lzma_opt; lzmafilters[1].id = LZMA_VLI_UNKNOWN;/* Terminate */ *strm = lzma_init_data; #ifdef HAVE_LZMA_STREAM_ENCODER_MT if (threads > 1) { - bzero(&mt_options, sizeof(mt_options)); + memset(&mt_options, 0, sizeof(mt_options)); mt_options.threads = threads; mt_options.timeout = 300; mt_options.filters = lzmafilters; mt_options.check = LZMA_CHECK_CRC64; r = lzma_stream_encoder_mt(strm, &mt_options); } else #endif r = lzma_stream_encoder(strm, lzmafilters, LZMA_CHECK_CRC64); switch (r) { case LZMA_OK: lastrm->real_stream = strm; lastrm->valid = 1; lastrm->code = compression_code_lzma; lastrm->end = compression_end_lzma; r = ARCHIVE_OK; break; case LZMA_MEM_ERROR: free(strm); lastrm->real_stream = NULL; archive_set_error(a, ENOMEM, "Internal error initializing compression library: " "Cannot allocate memory"); r = ARCHIVE_FATAL; break; default: free(strm); lastrm->real_stream = NULL; archive_set_error(a, ARCHIVE_ERRNO_MISC, "Internal error initializing compression library: " "It's a bug in liblzma"); r = ARCHIVE_FATAL; break; } return (r); } static int compression_code_lzma(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { lzma_stream *strm; int r; strm = (lzma_stream *)lastrm->real_stream; strm->next_in = lastrm->next_in; strm->avail_in = lastrm->avail_in; strm->total_in = lastrm->total_in; strm->next_out = lastrm->next_out; strm->avail_out = lastrm->avail_out; strm->total_out = lastrm->total_out; r = lzma_code(strm, (action == ARCHIVE_Z_FINISH)? LZMA_FINISH: LZMA_RUN); lastrm->next_in = strm->next_in; lastrm->avail_in = strm->avail_in; lastrm->total_in = strm->total_in; lastrm->next_out = strm->next_out; lastrm->avail_out = strm->avail_out; lastrm->total_out = strm->total_out; switch (r) { case LZMA_OK: /* Non-finishing case */ return (ARCHIVE_OK); case LZMA_STREAM_END: /* This return can only occur in finishing case. */ return (ARCHIVE_EOF); case LZMA_MEMLIMIT_ERROR: archive_set_error(a, ENOMEM, "lzma compression error:" " %ju MiB would have been needed", (uintmax_t)((lzma_memusage(strm) + 1024 * 1024 -1) / (1024 * 1024))); return (ARCHIVE_FATAL); default: /* Any other return value indicates an error */ archive_set_error(a, ARCHIVE_ERRNO_MISC, "lzma compression failed:" " lzma_code() call returned status %d", r); return (ARCHIVE_FATAL); } } static int compression_end_lzma(struct archive *a, struct la_zstream *lastrm) { lzma_stream *strm; (void)a; /* UNUSED */ strm = (lzma_stream *)lastrm->real_stream; lzma_end(strm); free(strm); lastrm->valid = 0; lastrm->real_stream = NULL; return (ARCHIVE_OK); } #else static int compression_init_encoder_lzma(struct archive *a, struct la_zstream *lastrm, int level) { (void) level; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "lzma")); } static int compression_init_encoder_xz(struct archive *a, struct la_zstream *lastrm, int level, int threads) { (void) level; /* UNUSED */ (void) threads; /* UNUSED */ if (lastrm->valid) compression_end(a, lastrm); return (compression_unsupported_encoder(a, lastrm, "xz")); } #endif static int xar_compression_init_encoder(struct archive_write *a) { struct xar *xar; int r; xar = (struct xar *)a->format_data; switch (xar->opt_compression) { case GZIP: r = compression_init_encoder_gzip( &(a->archive), &(xar->stream), xar->opt_compression_level, 1); break; case BZIP2: r = compression_init_encoder_bzip2( &(a->archive), &(xar->stream), xar->opt_compression_level); break; case LZMA: r = compression_init_encoder_lzma( &(a->archive), &(xar->stream), xar->opt_compression_level); break; case XZ: r = compression_init_encoder_xz( &(a->archive), &(xar->stream), xar->opt_compression_level, xar->opt_threads); break; default: r = ARCHIVE_OK; break; } if (r == ARCHIVE_OK) { xar->stream.total_in = 0; xar->stream.next_out = xar->wbuff; xar->stream.avail_out = sizeof(xar->wbuff); xar->stream.total_out = 0; } return (r); } static int compression_code(struct archive *a, struct la_zstream *lastrm, enum la_zaction action) { if (lastrm->valid) return (lastrm->code(a, lastrm, action)); return (ARCHIVE_OK); } static int compression_end(struct archive *a, struct la_zstream *lastrm) { if (lastrm->valid) return (lastrm->end(a, lastrm)); return (ARCHIVE_OK); } static int save_xattrs(struct archive_write *a, struct file *file) { struct xar *xar; const char *name; const void *value; struct heap_data *heap; size_t size; int count, r; xar = (struct xar *)a->format_data; count = archive_entry_xattr_reset(file->entry); if (count == 0) return (ARCHIVE_OK); while (count--) { archive_entry_xattr_next(file->entry, &name, &value, &size); checksum_init(&(xar->a_sumwrk), xar->opt_sumalg); checksum_init(&(xar->e_sumwrk), xar->opt_sumalg); heap = calloc(1, sizeof(*heap)); if (heap == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for xattr"); return (ARCHIVE_FATAL); } heap->id = file->ea_idx++; heap->temp_offset = xar->temp_offset; heap->size = size;/* save a extracted size */ heap->compression = xar->opt_compression; /* Get a extracted sumcheck value. */ checksum_update(&(xar->e_sumwrk), value, size); checksum_final(&(xar->e_sumwrk), &(heap->e_sum)); /* * Not compression to xattr is simple way. */ if (heap->compression == NONE) { checksum_update(&(xar->a_sumwrk), value, size); checksum_final(&(xar->a_sumwrk), &(heap->a_sum)); if (write_to_temp(a, value, size) != ARCHIVE_OK) { free(heap); return (ARCHIVE_FATAL); } heap->length = size; /* Add heap to the tail of file->xattr. */ heap->next = NULL; *file->xattr.last = heap; file->xattr.last = &(heap->next); /* Next xattr */ continue; } /* * Init compression library. */ r = xar_compression_init_encoder(a); if (r != ARCHIVE_OK) { free(heap); return (ARCHIVE_FATAL); } xar->stream.next_in = (const unsigned char *)value; xar->stream.avail_in = size; for (;;) { r = compression_code(&(a->archive), &(xar->stream), ARCHIVE_Z_FINISH); if (r != ARCHIVE_OK && r != ARCHIVE_EOF) { free(heap); return (ARCHIVE_FATAL); } size = sizeof(xar->wbuff) - xar->stream.avail_out; checksum_update(&(xar->a_sumwrk), xar->wbuff, size); if (write_to_temp(a, xar->wbuff, size) != ARCHIVE_OK) return (ARCHIVE_FATAL); if (r == ARCHIVE_OK) { xar->stream.next_out = xar->wbuff; xar->stream.avail_out = sizeof(xar->wbuff); } else { checksum_final(&(xar->a_sumwrk), &(heap->a_sum)); heap->length = xar->stream.total_out; /* Add heap to the tail of file->xattr. */ heap->next = NULL; *file->xattr.last = heap; file->xattr.last = &(heap->next); break; } } /* Clean up compression library. */ r = compression_end(&(a->archive), &(xar->stream)); if (r != ARCHIVE_OK) return (ARCHIVE_FATAL); } return (ARCHIVE_OK); } static int getalgsize(enum sumalg sumalg) { switch (sumalg) { default: case CKSUM_NONE: return (0); case CKSUM_SHA1: return (SHA1_SIZE); case CKSUM_MD5: return (MD5_SIZE); } } static const char * getalgname(enum sumalg sumalg) { switch (sumalg) { default: case CKSUM_NONE: return (NULL); case CKSUM_SHA1: return (SHA1_NAME); case CKSUM_MD5: return (MD5_NAME); } } #endif /* Support xar format */ Index: head/contrib/libarchive/libarchive/archive_write_set_format_zip.c =================================================================== --- head/contrib/libarchive/libarchive/archive_write_set_format_zip.c (revision 311902) +++ head/contrib/libarchive/libarchive/archive_write_set_format_zip.c (revision 311903) @@ -1,1678 +1,1678 @@ /*- * Copyright (c) 2008 Anselm Strauss * Copyright (c) 2009 Joerg Sonnenberger * Copyright (c) 2011-2012,2014 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ /* * Development supported by Google Summer of Code 2008. */ #include "archive_platform.h" __FBSDID("$FreeBSD$"); #ifdef HAVE_ERRNO_H #include #endif #ifdef HAVE_LANGINFO_H #include #endif #ifdef HAVE_STDLIB_H #include #endif #ifdef HAVE_STRING_H #include #endif #ifdef HAVE_ZLIB_H #include #endif #include "archive.h" #include "archive_cryptor_private.h" #include "archive_endian.h" #include "archive_entry.h" #include "archive_entry_locale.h" #include "archive_hmac_private.h" #include "archive_private.h" #include "archive_random_private.h" #include "archive_write_private.h" #ifndef HAVE_ZLIB_H #include "archive_crc32.h" #endif #define ZIP_ENTRY_FLAG_ENCRYPTED (1<<0) #define ZIP_ENTRY_FLAG_LENGTH_AT_END (1<<3) #define ZIP_ENTRY_FLAG_UTF8_NAME (1 << 11) #define ZIP_4GB_MAX ARCHIVE_LITERAL_LL(0xffffffff) #define ZIP_4GB_MAX_UNCOMPRESSED ARCHIVE_LITERAL_LL(0xff000000) enum compression { COMPRESSION_UNSPECIFIED = -1, COMPRESSION_STORE = 0, COMPRESSION_DEFLATE = 8 }; #ifdef HAVE_ZLIB_H #define COMPRESSION_DEFAULT COMPRESSION_DEFLATE #else #define COMPRESSION_DEFAULT COMPRESSION_STORE #endif enum encryption { ENCRYPTION_NONE = 0, ENCRYPTION_TRADITIONAL, /* Traditional PKWARE encryption. */ ENCRYPTION_WINZIP_AES128, /* WinZIP AES-128 encryption. */ ENCRYPTION_WINZIP_AES256, /* WinZIP AES-256 encryption. */ }; #define TRAD_HEADER_SIZE 12 /* * See "WinZip - AES Encryption Information" * http://www.winzip.com/aes_info.htm */ /* Value used in compression method. */ #define WINZIP_AES_ENCRYPTION 99 /* A WinZip AES header size which is stored at the beginning of * file contents. */ #define WINZIP_AES128_HEADER_SIZE (8 + 2) #define WINZIP_AES256_HEADER_SIZE (16 + 2) /* AES vendor version. */ #define AES_VENDOR_AE_1 0x0001 #define AES_VENDOR_AE_2 0x0002 /* Authentication code size. */ #define AUTH_CODE_SIZE 10 /**/ #define MAX_DERIVED_KEY_BUF_SIZE (AES_MAX_KEY_SIZE * 2 + 2) struct cd_segment { struct cd_segment *next; size_t buff_size; unsigned char *buff; unsigned char *p; }; struct trad_enc_ctx { uint32_t keys[3]; }; struct zip { int64_t entry_offset; int64_t entry_compressed_size; int64_t entry_uncompressed_size; int64_t entry_compressed_written; int64_t entry_uncompressed_written; int64_t entry_uncompressed_limit; struct archive_entry *entry; uint32_t entry_crc32; enum compression entry_compression; enum encryption entry_encryption; int entry_flags; int entry_uses_zip64; int experiments; struct trad_enc_ctx tctx; char tctx_valid; unsigned char trad_chkdat; unsigned aes_vendor; archive_crypto_ctx cctx; char cctx_valid; archive_hmac_sha1_ctx hctx; char hctx_valid; unsigned char *file_header; size_t file_header_extra_offset; unsigned long (*crc32func)(unsigned long crc, const void *buff, size_t len); struct cd_segment *central_directory; struct cd_segment *central_directory_last; size_t central_directory_bytes; size_t central_directory_entries; int64_t written_bytes; /* Overall position in file. */ struct archive_string_conv *opt_sconv; struct archive_string_conv *sconv_default; enum compression requested_compression; int deflate_compression_level; int init_default_conversion; enum encryption encryption_type; #define ZIP_FLAG_AVOID_ZIP64 1 #define ZIP_FLAG_FORCE_ZIP64 2 #define ZIP_FLAG_EXPERIMENT_xl 4 int flags; #ifdef HAVE_ZLIB_H z_stream stream; #endif size_t len_buf; unsigned char *buf; }; /* Don't call this min or MIN, since those are already defined on lots of platforms (but not all). */ #define zipmin(a, b) ((a) > (b) ? (b) : (a)) static ssize_t archive_write_zip_data(struct archive_write *, const void *buff, size_t s); static int archive_write_zip_close(struct archive_write *); static int archive_write_zip_free(struct archive_write *); static int archive_write_zip_finish_entry(struct archive_write *); static int archive_write_zip_header(struct archive_write *, struct archive_entry *); static int archive_write_zip_options(struct archive_write *, const char *, const char *); static unsigned int dos_time(const time_t); static size_t path_length(struct archive_entry *); static int write_path(struct archive_entry *, struct archive_write *); static void copy_path(struct archive_entry *, unsigned char *); static struct archive_string_conv *get_sconv(struct archive_write *, struct zip *); static int trad_enc_init(struct trad_enc_ctx *, const char *, size_t); static unsigned trad_enc_encrypt_update(struct trad_enc_ctx *, const uint8_t *, size_t, uint8_t *, size_t); static int init_traditional_pkware_encryption(struct archive_write *); static int is_traditional_pkware_encryption_supported(void); static int init_winzip_aes_encryption(struct archive_write *); static int is_winzip_aes_encryption_supported(int encryption); static unsigned char * cd_alloc(struct zip *zip, size_t length) { unsigned char *p; if (zip->central_directory == NULL || (zip->central_directory_last->p + length > zip->central_directory_last->buff + zip->central_directory_last->buff_size)) { struct cd_segment *segment = calloc(1, sizeof(*segment)); if (segment == NULL) return NULL; segment->buff_size = 64 * 1024; segment->buff = malloc(segment->buff_size); if (segment->buff == NULL) { free(segment); return NULL; } segment->p = segment->buff; if (zip->central_directory == NULL) { zip->central_directory = zip->central_directory_last = segment; } else { zip->central_directory_last->next = segment; zip->central_directory_last = segment; } } p = zip->central_directory_last->p; zip->central_directory_last->p += length; zip->central_directory_bytes += length; return (p); } static unsigned long real_crc32(unsigned long crc, const void *buff, size_t len) { return crc32(crc, buff, (unsigned int)len); } static unsigned long fake_crc32(unsigned long crc, const void *buff, size_t len) { (void)crc; /* UNUSED */ (void)buff; /* UNUSED */ (void)len; /* UNUSED */ return 0; } static int archive_write_zip_options(struct archive_write *a, const char *key, const char *val) { struct zip *zip = a->format_data; int ret = ARCHIVE_FAILED; if (strcmp(key, "compression") == 0) { /* * Set compression to use on all future entries. * This only affects regular files. */ if (val == NULL || val[0] == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: compression option needs a compression name", a->format_name); } else if (strcmp(val, "deflate") == 0) { #ifdef HAVE_ZLIB_H zip->requested_compression = COMPRESSION_DEFLATE; ret = ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "deflate compression not supported"); #endif } else if (strcmp(val, "store") == 0) { zip->requested_compression = COMPRESSION_STORE; ret = ARCHIVE_OK; } return (ret); } else if (strcmp(key, "compression-level") == 0) { if (val == NULL || !(val[0] >= '0' && val[0] <= '9') || val[1] != '\0') { return ARCHIVE_WARN; } if (val[0] == '0') { zip->requested_compression = COMPRESSION_STORE; return ARCHIVE_OK; } else { #ifdef HAVE_ZLIB_H zip->requested_compression = COMPRESSION_DEFLATE; zip->deflate_compression_level = val[0] - '0'; return ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "deflate compression not supported"); #endif } } else if (strcmp(key, "encryption") == 0) { if (val == NULL) { zip->encryption_type = ENCRYPTION_NONE; ret = ARCHIVE_OK; } else if (val[0] == '1' || strcmp(val, "traditional") == 0 || strcmp(val, "zipcrypt") == 0 || strcmp(val, "ZipCrypt") == 0) { if (is_traditional_pkware_encryption_supported()) { zip->encryption_type = ENCRYPTION_TRADITIONAL; ret = ARCHIVE_OK; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "encryption not supported"); } } else if (strcmp(val, "aes128") == 0) { if (is_winzip_aes_encryption_supported( ENCRYPTION_WINZIP_AES128)) { zip->encryption_type = ENCRYPTION_WINZIP_AES128; ret = ARCHIVE_OK; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "encryption not supported"); } } else if (strcmp(val, "aes256") == 0) { if (is_winzip_aes_encryption_supported( ENCRYPTION_WINZIP_AES256)) { zip->encryption_type = ENCRYPTION_WINZIP_AES256; ret = ARCHIVE_OK; } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "encryption not supported"); } } else { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: unknown encryption '%s'", a->format_name, val); } return (ret); } else if (strcmp(key, "experimental") == 0) { if (val == NULL || val[0] == 0) { zip->flags &= ~ ZIP_FLAG_EXPERIMENT_xl; } else { zip->flags |= ZIP_FLAG_EXPERIMENT_xl; } return (ARCHIVE_OK); } else if (strcmp(key, "fakecrc32") == 0) { /* * FOR TESTING ONLY: disable CRC calculation to speed up * certain complex tests. */ if (val == NULL || val[0] == 0) { zip->crc32func = real_crc32; } else { zip->crc32func = fake_crc32; } return (ARCHIVE_OK); } else if (strcmp(key, "hdrcharset") == 0) { /* * Set the character set used in translating filenames. */ if (val == NULL || val[0] == 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "%s: hdrcharset option needs a character-set name", a->format_name); } else { zip->opt_sconv = archive_string_conversion_to_charset( &a->archive, val, 0); if (zip->opt_sconv != NULL) ret = ARCHIVE_OK; else ret = ARCHIVE_FATAL; } return (ret); } else if (strcmp(key, "zip64") == 0) { /* * Bias decisions about Zip64: force them to be * generated in certain cases where they are not * forbidden or avoid them in certain cases where they * are not strictly required. */ if (val != NULL && *val != '\0') { zip->flags |= ZIP_FLAG_FORCE_ZIP64; zip->flags &= ~ZIP_FLAG_AVOID_ZIP64; } else { zip->flags &= ~ZIP_FLAG_FORCE_ZIP64; zip->flags |= ZIP_FLAG_AVOID_ZIP64; } return (ARCHIVE_OK); } /* Note: The "warn" return is just to inform the options * supervisor that we didn't handle it. It will generate * a suitable error if no one used this option. */ return (ARCHIVE_WARN); } int archive_write_zip_set_compression_deflate(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; int ret = ARCHIVE_FAILED; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW | ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_zip_set_compression_deflate"); if (a->archive.archive_format != ARCHIVE_FORMAT_ZIP) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can only use archive_write_zip_set_compression_deflate" " with zip format"); ret = ARCHIVE_FATAL; } else { #ifdef HAVE_ZLIB_H struct zip *zip = a->format_data; zip->requested_compression = COMPRESSION_DEFLATE; ret = ARCHIVE_OK; #else archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "deflate compression not supported"); ret = ARCHIVE_FAILED; #endif } return (ret); } int archive_write_zip_set_compression_store(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct zip *zip = a->format_data; int ret = ARCHIVE_FAILED; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW | ARCHIVE_STATE_HEADER | ARCHIVE_STATE_DATA, "archive_write_zip_set_compression_deflate"); if (a->archive.archive_format != ARCHIVE_FORMAT_ZIP) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can only use archive_write_zip_set_compression_store" " with zip format"); ret = ARCHIVE_FATAL; } else { zip->requested_compression = COMPRESSION_STORE; ret = ARCHIVE_OK; } return (ret); } int archive_write_set_format_zip(struct archive *_a) { struct archive_write *a = (struct archive_write *)_a; struct zip *zip; archive_check_magic(_a, ARCHIVE_WRITE_MAGIC, ARCHIVE_STATE_NEW, "archive_write_set_format_zip"); /* If another format was already registered, unregister it. */ if (a->format_free != NULL) (a->format_free)(a); zip = (struct zip *) calloc(1, sizeof(*zip)); if (zip == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip data"); return (ARCHIVE_FATAL); } /* "Unspecified" lets us choose the appropriate compression. */ zip->requested_compression = COMPRESSION_UNSPECIFIED; #ifdef HAVE_ZLIB_H zip->deflate_compression_level = Z_DEFAULT_COMPRESSION; #endif zip->crc32func = real_crc32; /* A buffer used for both compression and encryption. */ zip->len_buf = 65536; zip->buf = malloc(zip->len_buf); if (zip->buf == NULL) { free(zip); archive_set_error(&a->archive, ENOMEM, "Can't allocate compression buffer"); return (ARCHIVE_FATAL); } a->format_data = zip; a->format_name = "zip"; a->format_options = archive_write_zip_options; a->format_write_header = archive_write_zip_header; a->format_write_data = archive_write_zip_data; a->format_finish_entry = archive_write_zip_finish_entry; a->format_close = archive_write_zip_close; a->format_free = archive_write_zip_free; a->archive.archive_format = ARCHIVE_FORMAT_ZIP; a->archive.archive_format_name = "ZIP"; return (ARCHIVE_OK); } static int is_all_ascii(const char *p) { const unsigned char *pp = (const unsigned char *)p; while (*pp) { if (*pp++ > 127) return (0); } return (1); } static int archive_write_zip_header(struct archive_write *a, struct archive_entry *entry) { unsigned char local_header[32]; unsigned char local_extra[144]; struct zip *zip = a->format_data; unsigned char *e; unsigned char *cd_extra; size_t filename_length; const char *slink = NULL; size_t slink_size = 0; struct archive_string_conv *sconv = get_sconv(a, zip); int ret, ret2 = ARCHIVE_OK; mode_t type; int version_needed = 10; /* Ignore types of entries that we don't support. */ type = archive_entry_filetype(entry); if (type != AE_IFREG && type != AE_IFDIR && type != AE_IFLNK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Filetype not supported"); return ARCHIVE_FAILED; }; /* If we're not using Zip64, reject large files. */ if (zip->flags & ZIP_FLAG_AVOID_ZIP64) { /* Reject entries over 4GB. */ if (archive_entry_size_is_set(entry) && (archive_entry_size(entry) > ZIP_4GB_MAX)) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Files > 4GB require Zip64 extensions"); return ARCHIVE_FAILED; } /* Reject entries if archive is > 4GB. */ if (zip->written_bytes > ZIP_4GB_MAX) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Archives > 4GB require Zip64 extensions"); return ARCHIVE_FAILED; } } /* Only regular files can have size > 0. */ if (type != AE_IFREG) archive_entry_set_size(entry, 0); /* Reset information from last entry. */ zip->entry_offset = zip->written_bytes; zip->entry_uncompressed_limit = INT64_MAX; zip->entry_compressed_size = 0; zip->entry_uncompressed_size = 0; zip->entry_compressed_written = 0; zip->entry_uncompressed_written = 0; zip->entry_flags = 0; zip->entry_uses_zip64 = 0; zip->entry_crc32 = zip->crc32func(0, NULL, 0); zip->entry_encryption = 0; if (zip->entry != NULL) { archive_entry_free(zip->entry); zip->entry = NULL; } if (zip->cctx_valid) archive_encrypto_aes_ctr_release(&zip->cctx); if (zip->hctx_valid) archive_hmac_sha1_cleanup(&zip->hctx); zip->tctx_valid = zip->cctx_valid = zip->hctx_valid = 0; if (type == AE_IFREG &&(!archive_entry_size_is_set(entry) || archive_entry_size(entry) > 0)) { switch (zip->encryption_type) { case ENCRYPTION_TRADITIONAL: case ENCRYPTION_WINZIP_AES128: case ENCRYPTION_WINZIP_AES256: zip->entry_flags |= ZIP_ENTRY_FLAG_ENCRYPTED; zip->entry_encryption = zip->encryption_type; break; default: break; } } #if defined(_WIN32) && !defined(__CYGWIN__) /* Make sure the path separators in pathname, hardlink and symlink * are all slash '/', not the Windows path separator '\'. */ zip->entry = __la_win_entry_in_posix_pathseparator(entry); if (zip->entry == entry) zip->entry = archive_entry_clone(entry); #else zip->entry = archive_entry_clone(entry); #endif if (zip->entry == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip header data"); return (ARCHIVE_FATAL); } if (sconv != NULL) { const char *p; size_t len; if (archive_entry_pathname_l(entry, &p, &len, sconv) != 0) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory for Pathname"); return (ARCHIVE_FATAL); } archive_set_error(&a->archive, ARCHIVE_ERRNO_FILE_FORMAT, "Can't translate Pathname '%s' to %s", archive_entry_pathname(entry), archive_string_conversion_charset_name(sconv)); ret2 = ARCHIVE_WARN; } if (len > 0) archive_entry_set_pathname(zip->entry, p); /* * There is no standard for symlink handling; we convert * it using the same character-set translation that we use * for filename. */ if (type == AE_IFLNK) { if (archive_entry_symlink_l(entry, &p, &len, sconv)) { if (errno == ENOMEM) { archive_set_error(&a->archive, ENOMEM, "Can't allocate memory " " for Symlink"); return (ARCHIVE_FATAL); } /* No error if we can't convert. */ } else if (len > 0) archive_entry_set_symlink(zip->entry, p); } } /* If filename isn't ASCII and we can use UTF-8, set the UTF-8 flag. */ if (!is_all_ascii(archive_entry_pathname(zip->entry))) { if (zip->opt_sconv != NULL) { if (strcmp(archive_string_conversion_charset_name( zip->opt_sconv), "UTF-8") == 0) zip->entry_flags |= ZIP_ENTRY_FLAG_UTF8_NAME; #if HAVE_NL_LANGINFO } else if (strcmp(nl_langinfo(CODESET), "UTF-8") == 0) { zip->entry_flags |= ZIP_ENTRY_FLAG_UTF8_NAME; #endif } } filename_length = path_length(zip->entry); /* Determine appropriate compression and size for this entry. */ if (type == AE_IFLNK) { slink = archive_entry_symlink(zip->entry); if (slink != NULL) slink_size = strlen(slink); else slink_size = 0; zip->entry_uncompressed_limit = slink_size; zip->entry_compressed_size = slink_size; zip->entry_uncompressed_size = slink_size; zip->entry_crc32 = zip->crc32func(zip->entry_crc32, (const unsigned char *)slink, slink_size); zip->entry_compression = COMPRESSION_STORE; version_needed = 20; } else if (type != AE_IFREG) { zip->entry_compression = COMPRESSION_STORE; zip->entry_uncompressed_limit = 0; version_needed = 20; } else if (archive_entry_size_is_set(zip->entry)) { int64_t size = archive_entry_size(zip->entry); int64_t additional_size = 0; zip->entry_uncompressed_limit = size; zip->entry_compression = zip->requested_compression; if (zip->entry_compression == COMPRESSION_UNSPECIFIED) { zip->entry_compression = COMPRESSION_DEFAULT; } if (zip->entry_compression == COMPRESSION_STORE) { zip->entry_compressed_size = size; zip->entry_uncompressed_size = size; version_needed = 10; } else { zip->entry_uncompressed_size = size; version_needed = 20; } if (zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) { switch (zip->entry_encryption) { case ENCRYPTION_TRADITIONAL: additional_size = TRAD_HEADER_SIZE; version_needed = 20; break; case ENCRYPTION_WINZIP_AES128: additional_size = WINZIP_AES128_HEADER_SIZE + AUTH_CODE_SIZE; version_needed = 20; break; case ENCRYPTION_WINZIP_AES256: additional_size = WINZIP_AES256_HEADER_SIZE + AUTH_CODE_SIZE; version_needed = 20; break; default: break; } if (zip->entry_compression == COMPRESSION_STORE) zip->entry_compressed_size += additional_size; } /* * Set Zip64 extension in any of the following cases * (this was suggested by discussion on info-zip-dev * mailing list): * = Zip64 is being forced by user * = File is over 4GiB uncompressed * (including encryption header, if any) * = File is close to 4GiB and is being compressed * (compression might make file larger) */ if ((zip->flags & ZIP_FLAG_FORCE_ZIP64) || (zip->entry_uncompressed_size + additional_size > ZIP_4GB_MAX) || (zip->entry_uncompressed_size > ZIP_4GB_MAX_UNCOMPRESSED && zip->entry_compression != COMPRESSION_STORE)) { zip->entry_uses_zip64 = 1; version_needed = 45; } /* We may know the size, but never the CRC. */ zip->entry_flags |= ZIP_ENTRY_FLAG_LENGTH_AT_END; } else { /* We don't know the size. In this case, we prefer * deflate (it has a clear end-of-data marker which * makes length-at-end more reliable) and will * enable Zip64 extensions unless we're told not to. */ zip->entry_compression = COMPRESSION_DEFAULT; zip->entry_flags |= ZIP_ENTRY_FLAG_LENGTH_AT_END; if ((zip->flags & ZIP_FLAG_AVOID_ZIP64) == 0) { zip->entry_uses_zip64 = 1; version_needed = 45; } else if (zip->entry_compression == COMPRESSION_STORE) { version_needed = 10; } else { version_needed = 20; } if (zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) { switch (zip->entry_encryption) { case ENCRYPTION_TRADITIONAL: case ENCRYPTION_WINZIP_AES128: case ENCRYPTION_WINZIP_AES256: if (version_needed < 20) version_needed = 20; break; default: break; } } } /* Format the local header. */ memset(local_header, 0, sizeof(local_header)); memcpy(local_header, "PK\003\004", 4); archive_le16enc(local_header + 4, version_needed); archive_le16enc(local_header + 6, zip->entry_flags); if (zip->entry_encryption == ENCRYPTION_WINZIP_AES128 || zip->entry_encryption == ENCRYPTION_WINZIP_AES256) archive_le16enc(local_header + 8, WINZIP_AES_ENCRYPTION); else archive_le16enc(local_header + 8, zip->entry_compression); archive_le32enc(local_header + 10, dos_time(archive_entry_mtime(zip->entry))); archive_le32enc(local_header + 14, zip->entry_crc32); if (zip->entry_uses_zip64) { /* Zip64 data in the local header "must" include both * compressed and uncompressed sizes AND those fields * are included only if these are 0xffffffff; * THEREFORE these must be set this way, even if we * know one of them is smaller. */ archive_le32enc(local_header + 18, ZIP_4GB_MAX); archive_le32enc(local_header + 22, ZIP_4GB_MAX); } else { archive_le32enc(local_header + 18, (uint32_t)zip->entry_compressed_size); archive_le32enc(local_header + 22, (uint32_t)zip->entry_uncompressed_size); } archive_le16enc(local_header + 26, (uint16_t)filename_length); if (zip->entry_encryption == ENCRYPTION_TRADITIONAL) { if (zip->entry_flags & ZIP_ENTRY_FLAG_LENGTH_AT_END) zip->trad_chkdat = local_header[11]; else zip->trad_chkdat = local_header[17]; } /* Format as much of central directory file header as we can: */ zip->file_header = cd_alloc(zip, 46); /* If (zip->file_header == NULL) XXXX */ ++zip->central_directory_entries; memset(zip->file_header, 0, 46); memcpy(zip->file_header, "PK\001\002", 4); /* "Made by PKZip 2.0 on Unix." */ archive_le16enc(zip->file_header + 4, 3 * 256 + version_needed); archive_le16enc(zip->file_header + 6, version_needed); archive_le16enc(zip->file_header + 8, zip->entry_flags); if (zip->entry_encryption == ENCRYPTION_WINZIP_AES128 || zip->entry_encryption == ENCRYPTION_WINZIP_AES256) archive_le16enc(zip->file_header + 10, WINZIP_AES_ENCRYPTION); else archive_le16enc(zip->file_header + 10, zip->entry_compression); archive_le32enc(zip->file_header + 12, dos_time(archive_entry_mtime(zip->entry))); archive_le16enc(zip->file_header + 28, (uint16_t)filename_length); /* Following Info-Zip, store mode in the "external attributes" field. */ archive_le32enc(zip->file_header + 38, ((uint32_t)archive_entry_mode(zip->entry)) << 16); e = cd_alloc(zip, filename_length); /* If (e == NULL) XXXX */ copy_path(zip->entry, e); /* Format extra data. */ memset(local_extra, 0, sizeof(local_extra)); e = local_extra; /* First, extra blocks that are the same between * the local file header and the central directory. * We format them once and then duplicate them. */ /* UT timestamp, length depends on what timestamps are set. */ memcpy(e, "UT", 2); archive_le16enc(e + 2, 1 + (archive_entry_mtime_is_set(entry) ? 4 : 0) + (archive_entry_atime_is_set(entry) ? 4 : 0) + (archive_entry_ctime_is_set(entry) ? 4 : 0)); e += 4; *e++ = (archive_entry_mtime_is_set(entry) ? 1 : 0) | (archive_entry_atime_is_set(entry) ? 2 : 0) | (archive_entry_ctime_is_set(entry) ? 4 : 0); if (archive_entry_mtime_is_set(entry)) { archive_le32enc(e, (uint32_t)archive_entry_mtime(entry)); e += 4; } if (archive_entry_atime_is_set(entry)) { archive_le32enc(e, (uint32_t)archive_entry_atime(entry)); e += 4; } if (archive_entry_ctime_is_set(entry)) { archive_le32enc(e, (uint32_t)archive_entry_ctime(entry)); e += 4; } /* ux Unix extra data, length 11, version 1 */ /* TODO: If uid < 64k, use 2 bytes, ditto for gid. */ memcpy(e, "ux\013\000\001", 5); e += 5; *e++ = 4; /* Length of following UID */ archive_le32enc(e, (uint32_t)archive_entry_uid(entry)); e += 4; *e++ = 4; /* Length of following GID */ archive_le32enc(e, (uint32_t)archive_entry_gid(entry)); e += 4; /* AES extra data field: WinZIP AES information, ID=0x9901 */ if ((zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) && (zip->entry_encryption == ENCRYPTION_WINZIP_AES128 || zip->entry_encryption == ENCRYPTION_WINZIP_AES256)) { memcpy(e, "\001\231\007\000\001\000AE", 8); - /* AES vendoer version AE-2 does not store a CRC. + /* AES vendor version AE-2 does not store a CRC. * WinZip 11 uses AE-1, which does store the CRC, * but it does not store the CRC when the file size * is less than 20 bytes. So we simulate what * WinZip 11 does. * NOTE: WinZip 9.0 and 10.0 uses AE-2 by default. */ if (archive_entry_size_is_set(zip->entry) && archive_entry_size(zip->entry) < 20) { archive_le16enc(e+4, AES_VENDOR_AE_2); zip->aes_vendor = AES_VENDOR_AE_2;/* no CRC. */ } else zip->aes_vendor = AES_VENDOR_AE_1; e += 8; /* AES encryption strength. */ *e++ = (zip->entry_encryption == ENCRYPTION_WINZIP_AES128)?1:3; /* Actual compression method. */ archive_le16enc(e, zip->entry_compression); e += 2; } /* Copy UT ,ux, and AES-extra into central directory as well. */ zip->file_header_extra_offset = zip->central_directory_bytes; cd_extra = cd_alloc(zip, e - local_extra); memcpy(cd_extra, local_extra, e - local_extra); /* * Following extra blocks vary between local header and * central directory. These are the local header versions. * Central directory versions get formatted in * archive_write_zip_finish_entry() below. */ /* "[Zip64 entry] in the local header MUST include BOTH * original [uncompressed] and compressed size fields." */ if (zip->entry_uses_zip64) { unsigned char *zip64_start = e; memcpy(e, "\001\000\020\000", 4); e += 4; archive_le64enc(e, zip->entry_uncompressed_size); e += 8; archive_le64enc(e, zip->entry_compressed_size); e += 8; archive_le16enc(zip64_start + 2, (uint16_t)(e - (zip64_start + 4))); } if (zip->flags & ZIP_FLAG_EXPERIMENT_xl) { /* Experimental 'xl' extension to improve streaming. */ unsigned char *external_info = e; int included = 7; memcpy(e, "xl\000\000", 4); // 0x6c65 + 2-byte length e += 4; e[0] = included; /* bitmap of included fields */ e += 1; if (included & 1) { archive_le16enc(e, /* "Version created by" */ 3 * 256 + version_needed); e += 2; } if (included & 2) { archive_le16enc(e, 0); /* internal file attributes */ e += 2; } if (included & 4) { archive_le32enc(e, /* external file attributes */ ((uint32_t)archive_entry_mode(zip->entry)) << 16); e += 4; } if (included & 8) { // Libarchive does not currently support file comments. } archive_le16enc(external_info + 2, (uint16_t)(e - (external_info + 4))); } /* Update local header with size of extra data and write it all out: */ archive_le16enc(local_header + 28, (uint16_t)(e - local_extra)); ret = __archive_write_output(a, local_header, 30); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += 30; ret = write_path(zip->entry, a); if (ret <= ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += ret; ret = __archive_write_output(a, local_extra, e - local_extra); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += e - local_extra; /* For symlinks, write the body now. */ if (slink != NULL) { ret = __archive_write_output(a, slink, slink_size); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->entry_compressed_written += slink_size; zip->entry_uncompressed_written += slink_size; zip->written_bytes += slink_size; } #ifdef HAVE_ZLIB_H if (zip->entry_compression == COMPRESSION_DEFLATE) { zip->stream.zalloc = Z_NULL; zip->stream.zfree = Z_NULL; zip->stream.opaque = Z_NULL; zip->stream.next_out = zip->buf; zip->stream.avail_out = (uInt)zip->len_buf; if (deflateInit2(&zip->stream, zip->deflate_compression_level, Z_DEFLATED, -15, 8, Z_DEFAULT_STRATEGY) != Z_OK) { archive_set_error(&a->archive, ENOMEM, "Can't init deflate compressor"); return (ARCHIVE_FATAL); } } #endif return (ret2); } static ssize_t archive_write_zip_data(struct archive_write *a, const void *buff, size_t s) { int ret; struct zip *zip = a->format_data; if ((int64_t)s > zip->entry_uncompressed_limit) s = (size_t)zip->entry_uncompressed_limit; zip->entry_uncompressed_written += s; if (s == 0) return 0; if (zip->entry_flags & ZIP_ENTRY_FLAG_ENCRYPTED) { switch (zip->entry_encryption) { case ENCRYPTION_TRADITIONAL: - /* Initialize traditoinal PKWARE encryption context. */ + /* Initialize traditional PKWARE encryption context. */ if (!zip->tctx_valid) { ret = init_traditional_pkware_encryption(a); if (ret != ARCHIVE_OK) return (ret); zip->tctx_valid = 1; } break; case ENCRYPTION_WINZIP_AES128: case ENCRYPTION_WINZIP_AES256: if (!zip->cctx_valid) { ret = init_winzip_aes_encryption(a); if (ret != ARCHIVE_OK) return (ret); zip->cctx_valid = zip->hctx_valid = 1; } break; default: break; } } switch (zip->entry_compression) { case COMPRESSION_STORE: if (zip->tctx_valid || zip->cctx_valid) { const uint8_t *rb = (const uint8_t *)buff; const uint8_t * const re = rb + s; while (rb < re) { size_t l; if (zip->tctx_valid) { l = trad_enc_encrypt_update(&zip->tctx, rb, re - rb, zip->buf, zip->len_buf); } else { l = zip->len_buf; ret = archive_encrypto_aes_ctr_update( &zip->cctx, rb, re - rb, zip->buf, &l); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, l); } ret = __archive_write_output(a, zip->buf, l); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += l; zip->written_bytes += l; rb += l; } } else { ret = __archive_write_output(a, buff, s); if (ret != ARCHIVE_OK) return (ret); zip->written_bytes += s; zip->entry_compressed_written += s; } break; #if HAVE_ZLIB_H case COMPRESSION_DEFLATE: zip->stream.next_in = (unsigned char*)(uintptr_t)buff; zip->stream.avail_in = (uInt)s; do { ret = deflate(&zip->stream, Z_NO_FLUSH); if (ret == Z_STREAM_ERROR) return (ARCHIVE_FATAL); if (zip->stream.avail_out == 0) { if (zip->tctx_valid) { trad_enc_encrypt_update(&zip->tctx, zip->buf, zip->len_buf, zip->buf, zip->len_buf); } else if (zip->cctx_valid) { size_t outl = zip->len_buf; ret = archive_encrypto_aes_ctr_update( &zip->cctx, zip->buf, zip->len_buf, zip->buf, &outl); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, zip->len_buf); } ret = __archive_write_output(a, zip->buf, zip->len_buf); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += zip->len_buf; zip->written_bytes += zip->len_buf; zip->stream.next_out = zip->buf; zip->stream.avail_out = (uInt)zip->len_buf; } } while (zip->stream.avail_in != 0); break; #endif default: archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Invalid ZIP compression type"); return ARCHIVE_FATAL; } zip->entry_uncompressed_limit -= s; if (!zip->cctx_valid || zip->aes_vendor != AES_VENDOR_AE_2) zip->entry_crc32 = zip->crc32func(zip->entry_crc32, buff, (unsigned)s); return (s); } static int archive_write_zip_finish_entry(struct archive_write *a) { struct zip *zip = a->format_data; int ret; #if HAVE_ZLIB_H if (zip->entry_compression == COMPRESSION_DEFLATE) { for (;;) { size_t remainder; ret = deflate(&zip->stream, Z_FINISH); if (ret == Z_STREAM_ERROR) return (ARCHIVE_FATAL); remainder = zip->len_buf - zip->stream.avail_out; if (zip->tctx_valid) { trad_enc_encrypt_update(&zip->tctx, zip->buf, remainder, zip->buf, remainder); } else if (zip->cctx_valid) { size_t outl = remainder; ret = archive_encrypto_aes_ctr_update( &zip->cctx, zip->buf, remainder, zip->buf, &outl); if (ret < 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to encrypt file"); return (ARCHIVE_FAILED); } archive_hmac_sha1_update(&zip->hctx, zip->buf, remainder); } ret = __archive_write_output(a, zip->buf, remainder); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += remainder; zip->written_bytes += remainder; zip->stream.next_out = zip->buf; if (zip->stream.avail_out != 0) break; zip->stream.avail_out = (uInt)zip->len_buf; } deflateEnd(&zip->stream); } #endif if (zip->hctx_valid) { uint8_t hmac[20]; size_t hmac_len = 20; archive_hmac_sha1_final(&zip->hctx, hmac, &hmac_len); ret = __archive_write_output(a, hmac, AUTH_CODE_SIZE); if (ret != ARCHIVE_OK) return (ret); zip->entry_compressed_written += AUTH_CODE_SIZE; zip->written_bytes += AUTH_CODE_SIZE; } /* Write trailing data descriptor. */ if ((zip->entry_flags & ZIP_ENTRY_FLAG_LENGTH_AT_END) != 0) { char d[24]; memcpy(d, "PK\007\010", 4); if (zip->cctx_valid && zip->aes_vendor == AES_VENDOR_AE_2) archive_le32enc(d + 4, 0);/* no CRC.*/ else archive_le32enc(d + 4, zip->entry_crc32); if (zip->entry_uses_zip64) { archive_le64enc(d + 8, (uint64_t)zip->entry_compressed_written); archive_le64enc(d + 16, (uint64_t)zip->entry_uncompressed_written); ret = __archive_write_output(a, d, 24); zip->written_bytes += 24; } else { archive_le32enc(d + 8, (uint32_t)zip->entry_compressed_written); archive_le32enc(d + 12, (uint32_t)zip->entry_uncompressed_written); ret = __archive_write_output(a, d, 16); zip->written_bytes += 16; } if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); } /* Append Zip64 extra data to central directory information. */ if (zip->entry_compressed_written > ZIP_4GB_MAX || zip->entry_uncompressed_written > ZIP_4GB_MAX || zip->entry_offset > ZIP_4GB_MAX) { unsigned char zip64[32]; unsigned char *z = zip64, *zd; memcpy(z, "\001\000\000\000", 4); z += 4; if (zip->entry_uncompressed_written >= ZIP_4GB_MAX) { archive_le64enc(z, zip->entry_uncompressed_written); z += 8; } if (zip->entry_compressed_written >= ZIP_4GB_MAX) { archive_le64enc(z, zip->entry_compressed_written); z += 8; } if (zip->entry_offset >= ZIP_4GB_MAX) { archive_le64enc(z, zip->entry_offset); z += 8; } archive_le16enc(zip64 + 2, (uint16_t)(z - (zip64 + 4))); zd = cd_alloc(zip, z - zip64); if (zd == NULL) { archive_set_error(&a->archive, ENOMEM, "Can't allocate zip data"); return (ARCHIVE_FATAL); } memcpy(zd, zip64, z - zip64); /* Zip64 means version needs to be set to at least 4.5 */ if (archive_le16dec(zip->file_header + 6) < 45) archive_le16enc(zip->file_header + 6, 45); } /* Fix up central directory file header. */ if (zip->cctx_valid && zip->aes_vendor == AES_VENDOR_AE_2) archive_le32enc(zip->file_header + 16, 0);/* no CRC.*/ else archive_le32enc(zip->file_header + 16, zip->entry_crc32); archive_le32enc(zip->file_header + 20, (uint32_t)zipmin(zip->entry_compressed_written, ZIP_4GB_MAX)); archive_le32enc(zip->file_header + 24, (uint32_t)zipmin(zip->entry_uncompressed_written, ZIP_4GB_MAX)); archive_le16enc(zip->file_header + 30, (uint16_t)(zip->central_directory_bytes - zip->file_header_extra_offset)); archive_le32enc(zip->file_header + 42, (uint32_t)zipmin(zip->entry_offset, ZIP_4GB_MAX)); return (ARCHIVE_OK); } static int archive_write_zip_close(struct archive_write *a) { uint8_t buff[64]; int64_t offset_start, offset_end; struct zip *zip = a->format_data; struct cd_segment *segment; int ret; offset_start = zip->written_bytes; segment = zip->central_directory; while (segment != NULL) { ret = __archive_write_output(a, segment->buff, segment->p - segment->buff); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += segment->p - segment->buff; segment = segment->next; } offset_end = zip->written_bytes; /* If central dir info is too large, write Zip64 end-of-cd */ if (offset_end - offset_start > ZIP_4GB_MAX || offset_start > ZIP_4GB_MAX || zip->central_directory_entries > 0xffffUL || (zip->flags & ZIP_FLAG_FORCE_ZIP64)) { /* Zip64 end-of-cd record */ memset(buff, 0, 56); memcpy(buff, "PK\006\006", 4); archive_le64enc(buff + 4, 44); archive_le16enc(buff + 12, 45); archive_le16enc(buff + 14, 45); /* This is disk 0 of 0. */ archive_le64enc(buff + 24, zip->central_directory_entries); archive_le64enc(buff + 32, zip->central_directory_entries); archive_le64enc(buff + 40, offset_end - offset_start); archive_le64enc(buff + 48, offset_start); ret = __archive_write_output(a, buff, 56); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += 56; /* Zip64 end-of-cd locator record. */ memset(buff, 0, 20); memcpy(buff, "PK\006\007", 4); archive_le32enc(buff + 4, 0); archive_le64enc(buff + 8, offset_end); archive_le32enc(buff + 16, 1); ret = __archive_write_output(a, buff, 20); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += 20; } /* Format and write end of central directory. */ memset(buff, 0, sizeof(buff)); memcpy(buff, "PK\005\006", 4); archive_le16enc(buff + 8, (uint16_t)zipmin(0xffffU, zip->central_directory_entries)); archive_le16enc(buff + 10, (uint16_t)zipmin(0xffffU, zip->central_directory_entries)); archive_le32enc(buff + 12, (uint32_t)zipmin(ZIP_4GB_MAX, (offset_end - offset_start))); archive_le32enc(buff + 16, (uint32_t)zipmin(ZIP_4GB_MAX, offset_start)); ret = __archive_write_output(a, buff, 22); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); zip->written_bytes += 22; return (ARCHIVE_OK); } static int archive_write_zip_free(struct archive_write *a) { struct zip *zip; struct cd_segment *segment; zip = a->format_data; while (zip->central_directory != NULL) { segment = zip->central_directory; zip->central_directory = segment->next; free(segment->buff); free(segment); } free(zip->buf); archive_entry_free(zip->entry); if (zip->cctx_valid) archive_encrypto_aes_ctr_release(&zip->cctx); if (zip->hctx_valid) archive_hmac_sha1_cleanup(&zip->hctx); /* TODO: Free opt_sconv, sconv_default */ free(zip); a->format_data = NULL; return (ARCHIVE_OK); } /* Convert into MSDOS-style date/time. */ static unsigned int dos_time(const time_t unix_time) { struct tm *t; unsigned int dt; /* This will not preserve time when creating/extracting the archive * on two systems with different time zones. */ t = localtime(&unix_time); /* MSDOS-style date/time is only between 1980-01-01 and 2107-12-31 */ if (t->tm_year < 1980 - 1900) /* Set minimum date/time '1980-01-01 00:00:00'. */ dt = 0x00210000U; else if (t->tm_year > 2107 - 1900) /* Set maximum date/time '2107-12-31 23:59:58'. */ dt = 0xff9fbf7dU; else { dt = 0; dt += ((t->tm_year - 80) & 0x7f) << 9; dt += ((t->tm_mon + 1) & 0x0f) << 5; dt += (t->tm_mday & 0x1f); dt <<= 16; dt += (t->tm_hour & 0x1f) << 11; dt += (t->tm_min & 0x3f) << 5; dt += (t->tm_sec & 0x3e) >> 1; /* Only counting every 2 seconds. */ } return dt; } static size_t path_length(struct archive_entry *entry) { mode_t type; const char *path; type = archive_entry_filetype(entry); path = archive_entry_pathname(entry); if (path == NULL) return (0); if (type == AE_IFDIR && (path[0] == '\0' || path[strlen(path) - 1] != '/')) { return strlen(path) + 1; } else { return strlen(path); } } static int write_path(struct archive_entry *entry, struct archive_write *archive) { int ret; const char *path; mode_t type; size_t written_bytes; path = archive_entry_pathname(entry); type = archive_entry_filetype(entry); written_bytes = 0; ret = __archive_write_output(archive, path, strlen(path)); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); written_bytes += strlen(path); /* Folders are recognized by a trailing slash. */ if ((type == AE_IFDIR) & (path[strlen(path) - 1] != '/')) { ret = __archive_write_output(archive, "/", 1); if (ret != ARCHIVE_OK) return (ARCHIVE_FATAL); written_bytes += 1; } return ((int)written_bytes); } static void copy_path(struct archive_entry *entry, unsigned char *p) { const char *path; size_t pathlen; mode_t type; path = archive_entry_pathname(entry); pathlen = strlen(path); type = archive_entry_filetype(entry); memcpy(p, path, pathlen); /* Folders are recognized by a trailing slash. */ if ((type == AE_IFDIR) & (path[pathlen - 1] != '/')) { p[pathlen] = '/'; p[pathlen + 1] = '\0'; } } static struct archive_string_conv * get_sconv(struct archive_write *a, struct zip *zip) { if (zip->opt_sconv != NULL) return (zip->opt_sconv); if (!zip->init_default_conversion) { zip->sconv_default = archive_string_default_conversion_for_write(&(a->archive)); zip->init_default_conversion = 1; } return (zip->sconv_default); } /* Traditional PKWARE Decryption functions. */ static void trad_enc_update_keys(struct trad_enc_ctx *ctx, uint8_t c) { uint8_t t; #define CRC32(c, b) (crc32(c ^ 0xffffffffUL, &b, 1) ^ 0xffffffffUL) ctx->keys[0] = CRC32(ctx->keys[0], c); ctx->keys[1] = (ctx->keys[1] + (ctx->keys[0] & 0xff)) * 134775813L + 1; t = (ctx->keys[1] >> 24) & 0xff; ctx->keys[2] = CRC32(ctx->keys[2], t); #undef CRC32 } static uint8_t -trad_enc_decypt_byte(struct trad_enc_ctx *ctx) +trad_enc_decrypt_byte(struct trad_enc_ctx *ctx) { unsigned temp = ctx->keys[2] | 2; return (uint8_t)((temp * (temp ^ 1)) >> 8) & 0xff; } static unsigned trad_enc_encrypt_update(struct trad_enc_ctx *ctx, const uint8_t *in, size_t in_len, uint8_t *out, size_t out_len) { unsigned i, max; max = (unsigned)((in_len < out_len)? in_len: out_len); for (i = 0; i < max; i++) { uint8_t t = in[i]; - out[i] = t ^ trad_enc_decypt_byte(ctx); + out[i] = t ^ trad_enc_decrypt_byte(ctx); trad_enc_update_keys(ctx, t); } return i; } static int trad_enc_init(struct trad_enc_ctx *ctx, const char *pw, size_t pw_len) { ctx->keys[0] = 305419896L; ctx->keys[1] = 591751049L; ctx->keys[2] = 878082192L; for (;pw_len; --pw_len) trad_enc_update_keys(ctx, *pw++); return 0; } static int is_traditional_pkware_encryption_supported(void) { uint8_t key[TRAD_HEADER_SIZE]; if (archive_random(key, sizeof(key)-1) != ARCHIVE_OK) return (0); return (1); } static int init_traditional_pkware_encryption(struct archive_write *a) { struct zip *zip = a->format_data; const char *passphrase; uint8_t key[TRAD_HEADER_SIZE]; uint8_t key_encrypted[TRAD_HEADER_SIZE]; int ret; passphrase = __archive_write_get_passphrase(a); if (passphrase == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Encryption needs passphrase"); return ARCHIVE_FAILED; } if (archive_random(key, sizeof(key)-1) != ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't generate random number for encryption"); return ARCHIVE_FATAL; } trad_enc_init(&zip->tctx, passphrase, strlen(passphrase)); /* Set the last key code which will be used as a check code * for verifying passphrase in decryption. */ key[TRAD_HEADER_SIZE-1] = zip->trad_chkdat; trad_enc_encrypt_update(&zip->tctx, key, TRAD_HEADER_SIZE, key_encrypted, TRAD_HEADER_SIZE); /* Write encrypted keys in the top of the file content. */ ret = __archive_write_output(a, key_encrypted, TRAD_HEADER_SIZE); if (ret != ARCHIVE_OK) return (ret); zip->written_bytes += TRAD_HEADER_SIZE; zip->entry_compressed_written += TRAD_HEADER_SIZE; return (ret); } static int init_winzip_aes_encryption(struct archive_write *a) { struct zip *zip = a->format_data; const char *passphrase; size_t key_len, salt_len; uint8_t salt[16 + 2]; uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; int ret; passphrase = __archive_write_get_passphrase(a); if (passphrase == NULL) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Encryption needs passphrase"); return (ARCHIVE_FAILED); } if (zip->entry_encryption == ENCRYPTION_WINZIP_AES128) { salt_len = 8; key_len = 16; } else { /* AES 256 */ salt_len = 16; key_len = 32; } if (archive_random(salt, salt_len) != ARCHIVE_OK) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Can't generate random number for encryption"); return (ARCHIVE_FATAL); } archive_pbkdf2_sha1(passphrase, strlen(passphrase), salt, salt_len, 1000, derived_key, key_len * 2 + 2); ret = archive_encrypto_aes_ctr_init(&zip->cctx, derived_key, key_len); if (ret != 0) { archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Decryption is unsupported due to lack of crypto library"); return (ARCHIVE_FAILED); } ret = archive_hmac_sha1_init(&zip->hctx, derived_key + key_len, key_len); if (ret != 0) { archive_encrypto_aes_ctr_release(&zip->cctx); archive_set_error(&a->archive, ARCHIVE_ERRNO_MISC, "Failed to initialize HMAC-SHA1"); return (ARCHIVE_FAILED); } - /* Set a passowrd verification value after the 'salt'. */ + /* Set a password verification value after the 'salt'. */ salt[salt_len] = derived_key[key_len * 2]; salt[salt_len + 1] = derived_key[key_len * 2 + 1]; /* Write encrypted keys in the top of the file content. */ ret = __archive_write_output(a, salt, salt_len + 2); if (ret != ARCHIVE_OK) return (ret); zip->written_bytes += salt_len + 2; zip->entry_compressed_written += salt_len + 2; return (ARCHIVE_OK); } static int is_winzip_aes_encryption_supported(int encryption) { size_t key_len, salt_len; uint8_t salt[16 + 2]; uint8_t derived_key[MAX_DERIVED_KEY_BUF_SIZE]; archive_crypto_ctx cctx; archive_hmac_sha1_ctx hctx; int ret; if (encryption == ENCRYPTION_WINZIP_AES128) { salt_len = 8; key_len = 16; } else { /* AES 256 */ salt_len = 16; key_len = 32; } if (archive_random(salt, salt_len) != ARCHIVE_OK) return (0); ret = archive_pbkdf2_sha1("p", 1, salt, salt_len, 1000, derived_key, key_len * 2 + 2); if (ret != 0) return (0); ret = archive_encrypto_aes_ctr_init(&cctx, derived_key, key_len); if (ret != 0) return (0); ret = archive_hmac_sha1_init(&hctx, derived_key + key_len, key_len); archive_encrypto_aes_ctr_release(&cctx); if (ret != 0) return (0); archive_hmac_sha1_cleanup(&hctx); return (1); } Index: head/contrib/libarchive/libarchive/test/test_archive_read_add_passphrase.c =================================================================== --- head/contrib/libarchive/libarchive/test/test_archive_read_add_passphrase.c (revision 311902) +++ head/contrib/libarchive/libarchive/test/test_archive_read_add_passphrase.c (revision 311903) @@ -1,260 +1,260 @@ /*- * Copyright (c) 2011 Tim Kientzle * Copyright (c) 2014 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "test.h" __FBSDID("$FreeBSD$"); struct archive_read; extern void __archive_read_reset_passphrase(struct archive_read *); extern const char * __archive_read_next_passphrase(struct archive_read *); static void test(int pristine) { struct archive* a = archive_read_new(); if (!pristine) { archive_read_support_filter_all(a); archive_read_support_format_all(a); } assertEqualInt(ARCHIVE_OK, archive_read_add_passphrase(a, "pass1")); /* An empty passphrase cannot be accepted. */ assertEqualInt(ARCHIVE_FAILED, archive_read_add_passphrase(a, "")); /* NULL passphrases cannot be accepted. */ assertEqualInt(ARCHIVE_FAILED, archive_read_add_passphrase(a, NULL)); archive_read_free(a); } DEFINE_TEST(test_archive_read_add_passphrase) { test(1); test(0); } DEFINE_TEST(test_archive_read_add_passphrase_incorrect_sequance) { struct archive* a = archive_read_new(); struct archive_read *ar = (struct archive_read *)a; assertEqualInt(ARCHIVE_OK, archive_read_add_passphrase(a, "pass1")); /* No call of __archive_read_reset_passphrase() leads to * get NULL even if a user has passed a passphrases. */ assertEqualString(NULL, __archive_read_next_passphrase(ar)); archive_read_free(a); } DEFINE_TEST(test_archive_read_add_passphrase_single) { struct archive* a = archive_read_new(); struct archive_read *ar = (struct archive_read *)a; assertEqualInt(ARCHIVE_OK, archive_read_add_passphrase(a, "pass1")); __archive_read_reset_passphrase(ar); /* Fist call, we should get "pass1" as a passphrase. */ assertEqualString("pass1", __archive_read_next_passphrase(ar)); /* Second call, we should get NULL which means all the passphrases * are passed already. */ assertEqualString(NULL, __archive_read_next_passphrase(ar)); archive_read_free(a); } DEFINE_TEST(test_archive_read_add_passphrase_multiple) { struct archive* a = archive_read_new(); struct archive_read *ar = (struct archive_read *)a; assertEqualInt(ARCHIVE_OK, archive_read_add_passphrase(a, "pass1")); assertEqualInt(ARCHIVE_OK, archive_read_add_passphrase(a, "pass2")); __archive_read_reset_passphrase(ar); /* Fist call, we should get "pass1" as a passphrase. */ assertEqualString("pass1", __archive_read_next_passphrase(ar)); /* Second call, we should get "pass2" as a passphrase. */ assertEqualString("pass2", __archive_read_next_passphrase(ar)); /* Third call, we should get NULL which means all the passphrases * are passed already. */ assertEqualString(NULL, __archive_read_next_passphrase(ar)); archive_read_free(a); } static const char * callback1(struct archive *a, void *_client_data) { (void)a; /* UNUSED */ (void)_client_data; /* UNUSED */ return ("passCallBack"); } DEFINE_TEST(test_archive_read_add_passphrase_set_callback1) { struct archive* a = archive_read_new(); struct archive_read *ar = (struct archive_read *)a; assertEqualInt(ARCHIVE_OK, archive_read_set_passphrase_callback(a, NULL, callback1)); __archive_read_reset_passphrase(ar); /* Fist call, we should get "passCallBack" as a passphrase. */ assertEqualString("passCallBack", __archive_read_next_passphrase(ar)); /* Second call, we still get "passCallBack" as a passphrase. */ assertEqualString("passCallBack", __archive_read_next_passphrase(ar)); archive_read_free(a); /* Without __archive_read_reset_passphrase call, the callback * should work fine. */ a = archive_read_new(); ar = (struct archive_read *)a; assertEqualInt(ARCHIVE_OK, archive_read_set_passphrase_callback(a, NULL, callback1)); /* Fist call, we should get "passCallBack" as a passphrase. */ assertEqualString("passCallBack", __archive_read_next_passphrase(ar)); /* Second call, we still get "passCallBack" as a passphrase. */ assertEqualString("passCallBack", __archive_read_next_passphrase(ar)); archive_read_free(a); } static const char * callback2(struct archive *a, void *_client_data) { int *cd = (int *)_client_data; (void)a; /* UNUSED */ if (*cd == 0) { *cd = 1; return ("passCallBack"); } return (NULL); } DEFINE_TEST(test_archive_read_add_passphrase_set_callback2) { struct archive* a = archive_read_new(); struct archive_read *ar = (struct archive_read *)a; int client_data = 0; assertEqualInt(ARCHIVE_OK, archive_read_set_passphrase_callback(a, &client_data, callback2)); __archive_read_reset_passphrase(ar); /* Fist call, we should get "passCallBack" as a passphrase. */ assertEqualString("passCallBack", __archive_read_next_passphrase(ar)); /* Second call, we should get NULL which means all the passphrases * are passed already. */ assertEqualString(NULL, __archive_read_next_passphrase(ar)); archive_read_free(a); } DEFINE_TEST(test_archive_read_add_passphrase_set_callback3) { struct archive* a = archive_read_new(); struct archive_read *ar = (struct archive_read *)a; int client_data = 0; assertEqualInt(ARCHIVE_OK, archive_read_set_passphrase_callback(a, &client_data, callback2)); __archive_read_reset_passphrase(ar); /* Fist call, we should get "passCallBack" as a passphrase. */ assertEqualString("passCallBack", __archive_read_next_passphrase(ar)); __archive_read_reset_passphrase(ar); - /* After reset passphrase, we should get "passCallBack"passphrase. */ + /* After reset passphrase, we should get "passCallBack" passphrase. */ assertEqualString("passCallBack", __archive_read_next_passphrase(ar)); /* Second call, we should get NULL which means all the passphrases * are passed already. */ assertEqualString(NULL, __archive_read_next_passphrase(ar)); archive_read_free(a); } DEFINE_TEST(test_archive_read_add_passphrase_multiple_with_callback) { struct archive* a = archive_read_new(); struct archive_read *ar = (struct archive_read *)a; int client_data = 0; assertEqualInt(ARCHIVE_OK, archive_read_add_passphrase(a, "pass1")); assertEqualInt(ARCHIVE_OK, archive_read_add_passphrase(a, "pass2")); assertEqualInt(ARCHIVE_OK, archive_read_set_passphrase_callback(a, &client_data, callback2)); __archive_read_reset_passphrase(ar); /* Fist call, we should get "pass1" as a passphrase. */ assertEqualString("pass1", __archive_read_next_passphrase(ar)); /* Second call, we should get "pass2" as a passphrase. */ assertEqualString("pass2", __archive_read_next_passphrase(ar)); /* Third call, we should get "passCallBack" as a passphrase. */ assertEqualString("passCallBack", __archive_read_next_passphrase(ar)); /* Fourth call, we should get NULL which means all the passphrases * are passed already. */ assertEqualString(NULL, __archive_read_next_passphrase(ar)); archive_read_free(a); } DEFINE_TEST(test_archive_read_add_passphrase_multiple_with_callback2) { struct archive* a = archive_read_new(); struct archive_read *ar = (struct archive_read *)a; int client_data = 0; assertEqualInt(ARCHIVE_OK, archive_read_add_passphrase(a, "pass1")); assertEqualInt(ARCHIVE_OK, archive_read_add_passphrase(a, "pass2")); assertEqualInt(ARCHIVE_OK, archive_read_set_passphrase_callback(a, &client_data, callback2)); __archive_read_reset_passphrase(ar); /* Fist call, we should get "pass1" as a passphrase. */ assertEqualString("pass1", __archive_read_next_passphrase(ar)); /* Second call, we should get "pass2" as a passphrase. */ assertEqualString("pass2", __archive_read_next_passphrase(ar)); /* Third call, we should get "passCallBack" as a passphrase. */ assertEqualString("passCallBack", __archive_read_next_passphrase(ar)); __archive_read_reset_passphrase(ar); /* After reset passphrase, we should get "passCallBack" passphrase. */ assertEqualString("passCallBack", __archive_read_next_passphrase(ar)); /* Second call, we should get "pass1" as a passphrase. */ assertEqualString("pass1", __archive_read_next_passphrase(ar)); /* Third call, we should get "passCallBack" as a passphrase. */ assertEqualString("pass2", __archive_read_next_passphrase(ar)); /* Fourth call, we should get NULL which means all the passphrases * are passed already. */ assertEqualString(NULL, __archive_read_next_passphrase(ar)); archive_read_free(a); } Index: head/contrib/libarchive/libarchive/test/test_compat_uudecode.c =================================================================== --- head/contrib/libarchive/libarchive/test/test_compat_uudecode.c (revision 311902) +++ head/contrib/libarchive/libarchive/test/test_compat_uudecode.c (revision 311903) @@ -1,63 +1,63 @@ /*- * Copyright (c) 2003-2007 Tim Kientzle * Copyright (c) 2012 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "test.h" __FBSDID("$FreeBSD$"); static char archive_data[] = { "begin 644 test_read_uu.Z\n" "M'YV0+@`('$BPH,&#\"!,J7,BP(4(8$&_4J`$\"`,08$F%4O)AQ(\\2/(#7&@#%C\n" "M!@T8-##.L`$\"QL@:-F(``%'#H/1A2\n" "IHU\"0%9=*G4JUJM6K6+-JWO8,.*'4NVK-FS:-.J7 file1 echo "0123456789abcdef" > file2 # make afio use a large ASCII header sudo chown 65536 file2 find . -name "file[12]" | afio -o sample od -c sample | sed -E -e "s/^0[0-9]+//;s/^ //;s/( +)([^ ]{1,2})/'\2',/g;s/'\\0'/0/g;/^[*]/d" > test_read_format_cpio_afio.sample.txt rm -f file1 file2 sample exit1 */ static unsigned char archive[] = { '0','7','0','7','0','7','0','0','0','1','4','3','1','2','5','3', '2','1','1','0','0','6','4','4','0','0','1','7','5','1','0','0', '1','7','5','1','0','0','0','0','0','1','0','0','0','0','0','0', '1','1','3','3','2','2','4','5','0','2','0','0','0','0','0','0', '6','0','0','0','0','0','0','0','0','0','2','1','f','i','l','e', '1',0,'0','1','2','3','4','5','6','7','8','9','a','b','c','d', 'e','f','\n','0','7','0','7','2','7','0','0','0','0','0','0','6', '3','0','0','0','0','0','0','0','0','0','0','0','D','A','A','E', '6','m','1','0','0','6','4','4','0','0','0','1','0','0','0','0', '0','0','0','0','0','3','E','9','0','0','0','0','0','0','0','1', '0','0','0','0','0','0','0','0','0','0','0','0','0','0','0','0', '4','B','6','9','4','A','1','0','n','0','0','0','6','0','0','0', '0','0','0','0','0','s','0','0','0','0','0','0','0','0','0','0', '0','0','0','0','1','1',':','f','i','l','e','2',0,'0','1','2', '3','4','5','6','7','8','9','a','b','c','d','e','f','\n','0','7', '0','7','0','7','0','0','0','0','0','0','0','0','0','0','0','0', '0','0','0','0','0','0','0','0','0','0','0','0','0','0','0','0', '0','0','0','0','0','0','0','1','0','0','0','0','0','0','0','0', '0','0','0','0','0','0','0','0','0','0','0','0','0','1','3','0', '0','0','0','0','0','1','1','2','7','3','T','R','A','I','L','E', 'R','!','!','!',0,0,0,0,0,0,0,0,0,0,0,0, 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, }; /* * XXX This must be removed when we use int64_t for uid. */ static int uid_size(void) { return (sizeof(uid_t)); } DEFINE_TEST(test_read_format_cpio_afio) { unsigned char *p; size_t size; struct archive_entry *ae; struct archive *a; /* The default block size of afio is 5120. we simulate it */ size = (sizeof(archive) + 5120 -1 / 5120) * 5120; assert((p = malloc(size)) != NULL); if (p == NULL) return; memset(p, 0, size); memcpy(p, archive, sizeof(archive)); assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_memory(a, p, size)); /* * First entry is odc format. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt(17, archive_entry_size(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED); assertA(archive_filter_code(a, 0) == ARCHIVE_FILTER_NONE); assertA(archive_format(a) == ARCHIVE_FORMAT_CPIO_POSIX); /* * Second entry is afio large ASCII format. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt(17, archive_entry_size(ae)); if (uid_size() > 4) assertEqualInt(65536, archive_entry_uid(ae)); assertEqualInt(archive_entry_is_encrypted(ae), 0); assertEqualIntA(a, archive_read_has_encrypted_entries(a), ARCHIVE_READ_FORMAT_ENCRYPTION_UNSUPPORTED); assertA(archive_filter_code(a, 0) == ARCHIVE_FILTER_NONE); assertA(archive_format(a) == ARCHIVE_FORMAT_CPIO_AFIO_LARGE); assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); free(p); } Index: head/contrib/libarchive/libarchive/test/test_read_format_zip_traditional_encryption_data.c =================================================================== --- head/contrib/libarchive/libarchive/test/test_read_format_zip_traditional_encryption_data.c (revision 311902) +++ head/contrib/libarchive/libarchive/test/test_read_format_zip_traditional_encryption_data.c (revision 311903) @@ -1,168 +1,168 @@ /*- * Copyright (c) 2013 Konrad Kleine * Copyright (c) 2014 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "test.h" __FBSDID("$FreeBSD$"); DEFINE_TEST(test_read_format_zip_traditional_encryption_data) { - /* This file is password protected (Traditional PKWARE Enctypted). + /* This file is password protected (Traditional PKWARE Encrypted). The headers are NOT encrypted. Password is "12345678". */ const char *refname = "test_read_format_zip_traditional_encryption_data.zip"; struct archive_entry *ae; struct archive *a; char buff[512]; - /* Check if running system has cryptographic functionarity. */ + /* Check if running system has cryptographic functionality. */ assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_write_add_filter_none(a)); if (ARCHIVE_OK != archive_write_set_options(a, "zip:encryption=traditional")) { skipping("This system does not have cryptographic liberary"); archive_write_free(a); return; } archive_write_free(a); extract_reference_file(refname); /* * Extract a zip file without password. */ assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); assertEqualIntA(a, ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW, archive_read_has_encrypted_entries(a)); /* Verify encrypted file "bar.txt" */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("bar.txt", archive_entry_pathname(ae)); assertEqualInt(495, archive_entry_size(ae)); assertEqualInt(1, archive_entry_is_data_encrypted(ae)); assertEqualInt(0, archive_entry_is_metadata_encrypted(ae)); assertEqualIntA(a, 1, archive_read_has_encrypted_entries(a)); assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, sizeof(buff))); /* Verify encrypted file "foo.txt" */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("foo.txt", archive_entry_pathname(ae)); assertEqualInt(495, archive_entry_size(ae)); assertEqualInt(1, archive_entry_is_data_encrypted(ae)); assertEqualInt(0, archive_entry_is_metadata_encrypted(ae)); assertEqualIntA(a, 1, archive_read_has_encrypted_entries(a)); assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, sizeof(buff))); assertEqualInt(2, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); /* * Extract a zip file with password. */ assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); /* Pass three passphrases to decrypt a file content. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_add_passphrase(a, "invalid_pass")); assertEqualIntA(a, ARCHIVE_OK, archive_read_add_passphrase(a, "invalid_phrase")); assertEqualIntA(a, ARCHIVE_OK, archive_read_add_passphrase(a, "12345678")); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); assertEqualIntA(a, ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW, archive_read_has_encrypted_entries(a)); /* Verify encrypted file "bar.txt" */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("bar.txt", archive_entry_pathname(ae)); assertEqualInt(495, archive_entry_size(ae)); assertEqualInt(1, archive_entry_is_data_encrypted(ae)); assertEqualInt(0, archive_entry_is_metadata_encrypted(ae)); assertEqualIntA(a, 1, archive_read_has_encrypted_entries(a)); if (archive_zlib_version() != NULL) { assertEqualInt(495, archive_read_data(a, buff, sizeof(buff))); } else { assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, sizeof(buff))); assertEqualString(archive_error_string(a), "Unsupported ZIP compression method (deflation)"); assert(archive_errno(a) != 0); } /* Verify encrypted file "foo.txt" */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("foo.txt", archive_entry_pathname(ae)); assertEqualInt(495, archive_entry_size(ae)); assertEqualInt(1, archive_entry_is_data_encrypted(ae)); assertEqualInt(0, archive_entry_is_metadata_encrypted(ae)); assertEqualIntA(a, 1, archive_read_has_encrypted_entries(a)); if (archive_zlib_version() != NULL) { assertEqualInt(495, archive_read_data(a, buff, sizeof(buff))); } else { assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, sizeof(buff))); assertEqualString(archive_error_string(a), "Unsupported ZIP compression method (deflation)"); assert(archive_errno(a) != 0); } assertEqualInt(2, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } Index: head/contrib/libarchive/libarchive/test/test_read_format_zip_winzip_aes.c =================================================================== --- head/contrib/libarchive/libarchive/test/test_read_format_zip_winzip_aes.c (revision 311902) +++ head/contrib/libarchive/libarchive/test/test_read_format_zip_winzip_aes.c (revision 311903) @@ -1,152 +1,152 @@ /*- * Copyright (c) 2013 Konrad Kleine * Copyright (c) 2014 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "test.h" __FBSDID("$FreeBSD$"); static void test_winzip_aes(const char *refname, int need_libz) { struct archive_entry *ae; struct archive *a; char buff[512]; - /* Check if running system has cryptographic functionarity. */ + /* Check if running system has cryptographic functionality. */ assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_write_add_filter_none(a)); if (ARCHIVE_OK != archive_write_set_options(a, "zip:encryption=aes256")) { skipping("This system does not have cryptographic liberary"); archive_write_free(a); return; } archive_write_free(a); extract_reference_file(refname); /* * Extract a zip file without password. */ assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); assertEqualIntA(a, ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW, archive_read_has_encrypted_entries(a)); /* Verify encrypted file "README" */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("README", archive_entry_pathname(ae)); assertEqualInt(6818, archive_entry_size(ae)); assertEqualInt(1, archive_entry_is_data_encrypted(ae)); assertEqualInt(0, archive_entry_is_metadata_encrypted(ae)); assertEqualIntA(a, 1, archive_read_has_encrypted_entries(a)); assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, sizeof(buff))); assertEqualInt(1, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); /* * Extract a zip file with password. */ assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); /* Pass three passphrases to decrypt a file content. */ assertEqualIntA(a, ARCHIVE_OK, archive_read_add_passphrase(a, "invalid_pass")); assertEqualIntA(a, ARCHIVE_OK, archive_read_add_passphrase(a, "invalid_phrase")); assertEqualIntA(a, ARCHIVE_OK, archive_read_add_passphrase(a, "password")); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); assertEqualIntA(a, ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW, archive_read_has_encrypted_entries(a)); /* Verify encrypted file "README" */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("README", archive_entry_pathname(ae)); assertEqualInt(6818, archive_entry_size(ae)); assertEqualInt(1, archive_entry_is_data_encrypted(ae)); assertEqualInt(0, archive_entry_is_metadata_encrypted(ae)); assertEqualIntA(a, 1, archive_read_has_encrypted_entries(a)); if (!need_libz || archive_zlib_version() != NULL) { assertEqualInt(512, archive_read_data(a, buff, sizeof(buff))); } else { assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, 19)); assertEqualString(archive_error_string(a), "Unsupported ZIP compression method (deflation)"); assert(archive_errno(a) != 0); } assertEqualInt(1, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } DEFINE_TEST(test_read_format_zip_winzip_aes128) { /* WinZip AES-128 encryption. */ test_winzip_aes("test_read_format_zip_winzip_aes128.zip", 1); } DEFINE_TEST(test_read_format_zip_winzip_aes256) { /* WinZip AES-256 encryption. */ test_winzip_aes("test_read_format_zip_winzip_aes256.zip", 1); } DEFINE_TEST(test_read_format_zip_winzip_aes256_stored) { /* WinZip AES-256 encryption with stored data. */ test_winzip_aes("test_read_format_zip_winzip_aes256_stored.zip", 0); } Index: head/contrib/libarchive/libarchive/test/test_read_format_zip_winzip_aes_large.c =================================================================== --- head/contrib/libarchive/libarchive/test/test_read_format_zip_winzip_aes_large.c (revision 311902) +++ head/contrib/libarchive/libarchive/test/test_read_format_zip_winzip_aes_large.c (revision 311903) @@ -1,217 +1,217 @@ /*- * Copyright (c) 2013 Konrad Kleine * Copyright (c) 2014 Michihiro NAKAJIMA * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. */ #include "test.h" __FBSDID("$FreeBSD$"); DEFINE_TEST(test_read_format_zip_winzip_aes256_large) { const char *refname = "test_read_format_zip_winzip_aes256_large.zip"; struct archive_entry *ae; struct archive *a; char buff[512]; - /* Check if running system has cryptographic functionarity. */ + /* Check if running system has cryptographic functionality. */ assert((a = archive_write_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_write_set_format_zip(a)); assertEqualIntA(a, ARCHIVE_OK, archive_write_add_filter_none(a)); if (ARCHIVE_OK != archive_write_set_options(a, "zip:encryption=aes256")) { skipping("This system does not have cryptographic liberary"); archive_write_free(a); return; } archive_write_free(a); extract_reference_file(refname); /* * Extract a zip file without password. */ assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); assertEqualIntA(a, ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW, archive_read_has_encrypted_entries(a)); /* Verify encrypted file "Makefile" */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("Makefile", archive_entry_pathname(ae)); assertEqualInt(1456747, archive_entry_size(ae)); assertEqualInt(1, archive_entry_is_data_encrypted(ae)); assertEqualInt(0, archive_entry_is_metadata_encrypted(ae)); assertEqualIntA(a, 1, archive_read_has_encrypted_entries(a)); assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, sizeof(buff))); /* Verify encrypted file "NEWS" */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("NEWS", archive_entry_pathname(ae)); assertEqualInt(29357, archive_entry_size(ae)); assertEqualInt(1, archive_entry_is_data_encrypted(ae)); assertEqualInt(0, archive_entry_is_metadata_encrypted(ae)); assertEqualIntA(a, 1, archive_read_has_encrypted_entries(a)); assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, sizeof(buff))); /* Verify encrypted file "README" */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("README", archive_entry_pathname(ae)); assertEqualInt(6818, archive_entry_size(ae)); assertEqualInt(1, archive_entry_is_data_encrypted(ae)); assertEqualInt(0, archive_entry_is_metadata_encrypted(ae)); assertEqualIntA(a, 1, archive_read_has_encrypted_entries(a)); assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, sizeof(buff))); /* Verify encrypted file "config.h" */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("config.h", archive_entry_pathname(ae)); assertEqualInt(32667, archive_entry_size(ae)); assertEqualInt(1, archive_entry_is_data_encrypted(ae)); assertEqualInt(0, archive_entry_is_metadata_encrypted(ae)); assertEqualIntA(a, 1, archive_read_has_encrypted_entries(a)); assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, sizeof(buff))); assertEqualInt(4, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); /* * Extract a zip file with password. */ assert((a = archive_read_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); assertEqualIntA(a, ARCHIVE_OK, archive_read_add_passphrase(a, "password")); assertEqualIntA(a, ARCHIVE_OK, archive_read_open_filename(a, refname, 10240)); assertEqualIntA(a, ARCHIVE_READ_FORMAT_ENCRYPTION_DONT_KNOW, archive_read_has_encrypted_entries(a)); /* Verify encrypted file "Makefile" */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("Makefile", archive_entry_pathname(ae)); assertEqualInt(1456747, archive_entry_size(ae)); assertEqualInt(1, archive_entry_is_data_encrypted(ae)); assertEqualInt(0, archive_entry_is_metadata_encrypted(ae)); assertEqualIntA(a, 1, archive_read_has_encrypted_entries(a)); if (archive_zlib_version() != NULL) { assertEqualInt(512, archive_read_data(a, buff, sizeof(buff))); } else { assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, sizeof(buff))); assertEqualString(archive_error_string(a), "Unsupported ZIP compression method (deflation)"); assert(archive_errno(a) != 0); } /* Verify encrypted file "NEWS" */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("NEWS", archive_entry_pathname(ae)); assertEqualInt(29357, archive_entry_size(ae)); assertEqualInt(1, archive_entry_is_data_encrypted(ae)); assertEqualInt(0, archive_entry_is_metadata_encrypted(ae)); assertEqualIntA(a, 1, archive_read_has_encrypted_entries(a)); if (archive_zlib_version() != NULL) { assertEqualInt(512, archive_read_data(a, buff, sizeof(buff))); } else { assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, sizeof(buff))); assertEqualString(archive_error_string(a), "Unsupported ZIP compression method (deflation)"); assert(archive_errno(a) != 0); } /* Verify encrypted file "README" */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("README", archive_entry_pathname(ae)); assertEqualInt(6818, archive_entry_size(ae)); assertEqualInt(1, archive_entry_is_data_encrypted(ae)); assertEqualInt(0, archive_entry_is_metadata_encrypted(ae)); assertEqualIntA(a, 1, archive_read_has_encrypted_entries(a)); if (archive_zlib_version() != NULL) { assertEqualInt(512, archive_read_data(a, buff, sizeof(buff))); } else { assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, sizeof(buff))); assertEqualString(archive_error_string(a), "Unsupported ZIP compression method (deflation)"); assert(archive_errno(a) != 0); } /* Verify encrypted file "config.h" */ assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header(a, &ae)); assertEqualInt((AE_IFREG | 0644), archive_entry_mode(ae)); assertEqualString("config.h", archive_entry_pathname(ae)); assertEqualInt(32667, archive_entry_size(ae)); assertEqualInt(1, archive_entry_is_data_encrypted(ae)); assertEqualInt(0, archive_entry_is_metadata_encrypted(ae)); assertEqualIntA(a, 1, archive_read_has_encrypted_entries(a)); if (archive_zlib_version() != NULL) { assertEqualInt(512, archive_read_data(a, buff, sizeof(buff))); } else { assertEqualInt(ARCHIVE_FAILED, archive_read_data(a, buff, sizeof(buff))); assertEqualString(archive_error_string(a), "Unsupported ZIP compression method (deflation)"); assert(archive_errno(a) != 0); } assertEqualInt(4, archive_file_count(a)); /* End of archive. */ assertEqualIntA(a, ARCHIVE_EOF, archive_read_next_header(a, &ae)); /* Verify archive format. */ assertEqualIntA(a, ARCHIVE_FILTER_NONE, archive_filter_code(a, 0)); assertEqualIntA(a, ARCHIVE_FORMAT_ZIP, archive_format(a)); /* Close the archive. */ assertEqualInt(ARCHIVE_OK, archive_read_close(a)); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); } Index: head/contrib/libarchive/libarchive/test/test_read_pax_schily_xattr.c =================================================================== --- head/contrib/libarchive/libarchive/test/test_read_pax_schily_xattr.c (nonexistent) +++ head/contrib/libarchive/libarchive/test/test_read_pax_schily_xattr.c (revision 311903) @@ -0,0 +1,70 @@ +/*- + * Copyright (c) 2016 IBM Corporation + * Copyright (c) 2003-2007 Tim Kientzle + * + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. 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 AUTHOR(S) ``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 AUTHOR(S) 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. + * + * This test case's code has been derived from test_entry.c + */ +#include "test.h" + +DEFINE_TEST(test_schily_xattr_pax) +{ + struct archive *a; + struct archive_entry *ae; + const char *refname = "test_read_pax_schily_xattr.tar"; + const char *xname; /* For xattr tests. */ + const void *xval; /* For xattr tests. */ + size_t xsize; /* For xattr tests. */ + const char *string, *array; + + assert((a = archive_read_new()) != NULL); + assertEqualIntA(a, ARCHIVE_OK, archive_read_support_filter_all(a)); + assertEqualIntA(a, ARCHIVE_OK, archive_read_support_format_all(a)); + + extract_reference_file(refname); + assertEqualIntA(a, ARCHIVE_OK, + archive_read_open_filename(a, refname, 10240)); + + assertEqualInt(ARCHIVE_OK, archive_read_next_header(a, &ae)); + assertEqualInt(2, archive_entry_xattr_count(ae)); + assertEqualInt(2, archive_entry_xattr_reset(ae)); + + assertEqualInt(0, archive_entry_xattr_next(ae, &xname, &xval, &xsize)); + assertEqualString(xname, "security.selinux"); + string = "system_u:object_r:unlabeled_t:s0"; + assertEqualString(xval, string); + /* the xattr's value also contains the terminating \0 */ + assertEqualInt((int)xsize, strlen(string) + 1); + + assertEqualInt(0, archive_entry_xattr_next(ae, &xname, &xval, &xsize)); + assertEqualString(xname, "security.ima"); + assertEqualInt((int)xsize, 265); + /* we only compare the first 12 bytes */ + array = "\x03\x02\x04\xb0\xe9\xd6\x79\x01\x00\x2b\xad\x1e"; + assertEqualMem(xval, array, 12); + + /* Close the archive. */ + assertEqualInt(ARCHIVE_OK, archive_read_close(a)); + assertEqualInt(ARCHIVE_OK, archive_read_free(a)); +} Property changes on: head/contrib/libarchive/libarchive/test/test_read_pax_schily_xattr.c ___________________________________________________________________ Added: svn:eol-style ## -0,0 +1 ## +native \ No newline at end of property Added: svn:keywords ## -0,0 +1 ## +FreeBSD=%H \ No newline at end of property Added: svn:mime-type ## -0,0 +1 ## +text/plain \ No newline at end of property Index: head/contrib/libarchive/libarchive/test/test_read_pax_schily_xattr.tar.uu =================================================================== --- head/contrib/libarchive/libarchive/test/test_read_pax_schily_xattr.tar.uu (nonexistent) +++ head/contrib/libarchive/libarchive/test/test_read_pax_schily_xattr.tar.uu (revision 311903) @@ -0,0 +1,231 @@ +begin 644 test_schily_xattr_pax.tar +M+B]087A(96%D97)S+C$U,C4O8V]N9F9I;&5S```````````````````````` +M```````````````````````````````````````````````````````````` +M`````````````#`P,#`V-#0`,#`P,#`P,``P,#`P,#`P`#`P,#`P,#`P-C0W +M`#$R-S$R,C$P-3`V`#`Q,C4V-@`@>``````````````````````````````` +M```````````````````````````````````````````````````````````` +M``````````````````````````````````````````!UW6X5O?Y6: +M9^':P2MZR[4)$@W?)B6GX0U@<,0M%6YNMO%OG+IS%/.< +M,"A(N&S.F9]=!*5=\).X."2$GUGJ,0C:@+G#$M_E8UQP,LU-G(8IKW^K^<8* +M*3_.N0'%8.^$8S$`D9XOF+DK<<)U34U'_"O5/22YS96QI;G5X +M/7-Y #endif #ifdef HAVE_SYS_PARAM_H #include #endif #ifdef HAVE_FCNTL_H #include #endif #ifdef HAVE_LIMITS_H #include #endif #ifdef HAVE_UNISTD_H #include #endif #ifdef HAVE_LINUX_TYPES_H #include #endif #ifdef HAVE_LINUX_FIEMAP_H #include #endif #ifdef HAVE_LINUX_FS_H #include #endif /* The logic to compare sparse file data read from disk with the * specification is a little involved. Set to 1 to have the progress * dumped. */ #define DEBUG 0 /* * NOTE: On FreeBSD and Solaris, this test needs ZFS. - * You may should perfom this test as + * You may perform this test as * 'TMPDIR= libarchive_test'. */ struct sparse { enum { DATA, HOLE, END } type; size_t size; }; static void create_sparse_file(const char *, const struct sparse *); #if defined(_WIN32) && !defined(__CYGWIN__) #include /* * Create a sparse file on Windows. */ #if !defined(PATH_MAX) #define PATH_MAX MAX_PATH #endif #if !defined(__BORLANDC__) #define getcwd _getcwd #endif static int is_sparse_supported(const char *path) { char root[MAX_PATH+1]; char vol[MAX_PATH+1]; char sys[MAX_PATH+1]; DWORD flags; BOOL r; strncpy(root, path, sizeof(root)-1); if (((root[0] >= 'c' && root[0] <= 'z') || (root[0] >= 'C' && root[0] <= 'Z')) && root[1] == ':' && (root[2] == '\\' || root[2] == '/')) root[3] = '\0'; else return (0); assertEqualInt((r = GetVolumeInformation(root, vol, sizeof(vol), NULL, NULL, &flags, sys, sizeof(sys))), 1); return (r != 0 && (flags & FILE_SUPPORTS_SPARSE_FILES) != 0); } static void create_sparse_file(const char *path, const struct sparse *s) { char buff[1024]; HANDLE handle; DWORD dmy; memset(buff, ' ', sizeof(buff)); handle = CreateFileA(path, GENERIC_WRITE, 0, NULL, CREATE_NEW, FILE_ATTRIBUTE_NORMAL, NULL); assert(handle != INVALID_HANDLE_VALUE); assert(DeviceIoControl(handle, FSCTL_SET_SPARSE, NULL, 0, NULL, 0, &dmy, NULL) != 0); while (s->type != END) { if (s->type == HOLE) { LARGE_INTEGER distance; distance.QuadPart = s->size; assert(SetFilePointerEx(handle, distance, NULL, FILE_CURRENT) != 0); } else { DWORD w, wr; size_t size; size = s->size; while (size) { if (size > sizeof(buff)) w = sizeof(buff); else w = (DWORD)size; assert(WriteFile(handle, buff, w, &wr, NULL) != 0); size -= wr; } } s++; } assertEqualInt(CloseHandle(handle), 1); } #else #if defined(HAVE_LINUX_FIEMAP_H) /* * FIEMAP, which can detect 'hole' of a sparse file, has * been supported from 2.6.28 */ static int is_sparse_supported_fiemap(const char *path) { const struct sparse sparse_file[] = { /* This hole size is too small to create a sparse * files for almost filesystem. */ { HOLE, 1024 }, { DATA, 10240 }, { END, 0 } }; int fd, r; struct fiemap *fm; char buff[1024]; const char *testfile = "can_sparse"; (void)path; /* UNUSED */ memset(buff, 0, sizeof(buff)); create_sparse_file(testfile, sparse_file); fd = open(testfile, O_RDWR); if (fd < 0) return (0); fm = (struct fiemap *)buff; fm->fm_start = 0; fm->fm_length = ~0ULL;; fm->fm_flags = FIEMAP_FLAG_SYNC; fm->fm_extent_count = (sizeof(buff) - sizeof(*fm))/ sizeof(struct fiemap_extent); r = ioctl(fd, FS_IOC_FIEMAP, fm); close(fd); unlink(testfile); return (r >= 0); } #if !defined(SEEK_HOLE) || !defined(SEEK_DATA) static int is_sparse_supported(const char *path) { return is_sparse_supported_fiemap(path); } #endif #endif #if defined(_PC_MIN_HOLE_SIZE) /* * FreeBSD and Solaris can detect 'hole' of a sparse file * through lseek(HOLE) on ZFS. (UFS does not support yet) */ static int is_sparse_supported(const char *path) { return (pathconf(path, _PC_MIN_HOLE_SIZE) > 0); } #elif defined(SEEK_HOLE) && defined(SEEK_DATA) static int is_sparse_supported(const char *path) { const struct sparse sparse_file[] = { /* This hole size is too small to create a sparse * files for almost filesystem. */ { HOLE, 1024 }, { DATA, 10240 }, { END, 0 } }; int fd, r; const char *testfile = "can_sparse"; (void)path; /* UNUSED */ create_sparse_file(testfile, sparse_file); fd = open(testfile, O_RDWR); if (fd < 0) return (0); r = lseek(fd, 0, SEEK_HOLE); close(fd); unlink(testfile); #if defined(HAVE_LINUX_FIEMAP_H) if (r < 0) return (is_sparse_supported_fiemap(path)); #endif return (r >= 0); } #elif !defined(HAVE_LINUX_FIEMAP_H) /* * Other system may do not have the API such as lseek(HOLE), * which detect 'hole' of a sparse file. */ static int is_sparse_supported(const char *path) { (void)path; /* UNUSED */ return (0); } #endif /* * Create a sparse file on POSIX like system. */ static void create_sparse_file(const char *path, const struct sparse *s) { char buff[1024]; int fd; size_t total_size = 0; const struct sparse *cur = s; memset(buff, ' ', sizeof(buff)); assert((fd = open(path, O_CREAT | O_WRONLY, 0600)) != -1); /* Handle holes at the end by extending the file */ while (cur->type != END) { total_size += cur->size; ++cur; } assert(ftruncate(fd, total_size) != -1); while (s->type != END) { if (s->type == HOLE) { assert(lseek(fd, s->size, SEEK_CUR) != (off_t)-1); } else { size_t w, size; size = s->size; while (size) { if (size > sizeof(buff)) w = sizeof(buff); else w = size; assert(write(fd, buff, w) != (ssize_t)-1); size -= w; } } s++; } close(fd); } #endif /* * Sparse test with directory traversals. */ static void verify_sparse_file(struct archive *a, const char *path, const struct sparse *sparse, int expected_holes) { struct archive_entry *ae; const void *buff; size_t bytes_read; int64_t offset, expected_offset, last_offset; int holes_seen = 0; create_sparse_file(path, sparse); assert((ae = archive_entry_new()) != NULL); assertEqualIntA(a, ARCHIVE_OK, archive_read_disk_open(a, path)); assertEqualIntA(a, ARCHIVE_OK, archive_read_next_header2(a, ae)); expected_offset = 0; last_offset = 0; while (ARCHIVE_OK == archive_read_data_block(a, &buff, &bytes_read, &offset)) { const char *start = buff; #if DEBUG fprintf(stderr, "%s: bytes_read=%d offset=%d\n", path, (int)bytes_read, (int)offset); #endif if (offset > last_offset) { ++holes_seen; } /* Blocks entirely before the data we just read. */ while (expected_offset + (int64_t)sparse->size < offset) { #if DEBUG fprintf(stderr, " skipping expected_offset=%d, size=%d\n", (int)expected_offset, (int)sparse->size); #endif /* Must be holes. */ assert(sparse->type == HOLE); expected_offset += sparse->size; ++sparse; } /* Block that overlaps beginning of data */ if (expected_offset < offset && expected_offset + (int64_t)sparse->size <= offset + (int64_t)bytes_read) { const char *end = (const char *)buff + (expected_offset - offset) + (size_t)sparse->size; #if DEBUG fprintf(stderr, " overlapping hole expected_offset=%d, size=%d\n", (int)expected_offset, (int)sparse->size); #endif /* Must be a hole, overlap must be filled with '\0' */ if (assert(sparse->type == HOLE)) { assertMemoryFilledWith(start, end - start, '\0'); } start = end; expected_offset += sparse->size; ++sparse; } /* Blocks completely contained in data we just read. */ while (expected_offset + (int64_t)sparse->size <= offset + (int64_t)bytes_read) { const char *end = (const char *)buff + (expected_offset - offset) + (size_t)sparse->size; if (sparse->type == HOLE) { #if DEBUG fprintf(stderr, " contained hole expected_offset=%d, size=%d\n", (int)expected_offset, (int)sparse->size); #endif /* verify data corresponding to hole is '\0' */ if (end > (const char *)buff + bytes_read) { end = (const char *)buff + bytes_read; } assertMemoryFilledWith(start, end - start, '\0'); start = end; expected_offset += sparse->size; ++sparse; } else if (sparse->type == DATA) { #if DEBUG fprintf(stderr, " contained data expected_offset=%d, size=%d\n", (int)expected_offset, (int)sparse->size); #endif /* verify data corresponding to hole is ' ' */ if (assert(expected_offset + sparse->size <= offset + bytes_read)) { assert(start == (const char *)buff + (size_t)(expected_offset - offset)); assertMemoryFilledWith(start, end - start, ' '); } start = end; expected_offset += sparse->size; ++sparse; } else { break; } } /* Block that overlaps end of data */ if (expected_offset < offset + (int64_t)bytes_read) { const char *end = (const char *)buff + bytes_read; #if DEBUG fprintf(stderr, " trailing overlap expected_offset=%d, size=%d\n", (int)expected_offset, (int)sparse->size); #endif /* Must be a hole, overlap must be filled with '\0' */ if (assert(sparse->type == HOLE)) { assertMemoryFilledWith(start, end - start, '\0'); } } last_offset = offset + bytes_read; } /* Count a hole at EOF? */ if (last_offset < archive_entry_size(ae)) { ++holes_seen; } /* Verify blocks after last read */ while (sparse->type == HOLE) { expected_offset += sparse->size; ++sparse; } assert(sparse->type == END); assertEqualInt(expected_offset, archive_entry_size(ae)); assertEqualInt(holes_seen, expected_holes); assertEqualIntA(a, ARCHIVE_OK, archive_read_close(a)); archive_entry_free(ae); } #if defined(_WIN32) && !defined(__CYGWIN__) #define close _close #define open _open #endif /* * Sparse test without directory traversals. */ static void verify_sparse_file2(struct archive *a, const char *path, const struct sparse *sparse, int blocks, int preopen) { struct archive_entry *ae; int fd; (void)sparse; /* UNUSED */ assert((ae = archive_entry_new()) != NULL); archive_entry_set_pathname(ae, path); if (preopen) fd = open(path, O_RDONLY | O_BINARY); else fd = -1; assertEqualIntA(a, ARCHIVE_OK, archive_read_disk_entry_from_file(a, ae, fd, NULL)); if (fd >= 0) close(fd); /* Verify the number of holes only, not its offset nor its * length because those alignments are deeply dependence on * its filesystem. */ assertEqualInt(blocks, archive_entry_sparse_count(ae)); archive_entry_free(ae); } static void test_sparse_whole_file_data() { struct archive_entry *ae; int64_t offset; int i; assert((ae = archive_entry_new()) != NULL); archive_entry_set_size(ae, 1024*10); /* * Add sparse block data up to the file size. */ offset = 0; for (i = 0; i < 10; i++) { archive_entry_sparse_add_entry(ae, offset, 1024); offset += 1024; } failure("There should be no sparse"); assertEqualInt(0, archive_entry_sparse_count(ae)); archive_entry_free(ae); } DEFINE_TEST(test_sparse_basic) { char *cwd; struct archive *a; /* * The alignment of the hole of sparse files deeply depends * on filesystem. In my experience, sparse_file2 test with * 204800 bytes hole size did not pass on ZFS and the result * of that test seemed the size was too small, thus you should * keep a hole size more than 409600 bytes to pass this test * on all platform. */ const struct sparse sparse_file0[] = { { DATA, 1024 }, { HOLE, 2048000 }, { DATA, 2048 }, { HOLE, 2048000 }, { DATA, 4096 }, { HOLE, 20480000 }, { DATA, 8192 }, { HOLE, 204800000 }, { DATA, 1 }, { END, 0 } }; const struct sparse sparse_file1[] = { { HOLE, 409600 }, { DATA, 1 }, { HOLE, 409600 }, { DATA, 1 }, { HOLE, 409600 }, { END, 0 } }; const struct sparse sparse_file2[] = { { HOLE, 409600 * 1 }, { DATA, 1024 }, { HOLE, 409600 * 2 }, { DATA, 1024 }, { HOLE, 409600 * 3 }, { DATA, 1024 }, { HOLE, 409600 * 4 }, { DATA, 1024 }, { HOLE, 409600 * 5 }, { DATA, 1024 }, { HOLE, 409600 * 6 }, { DATA, 1024 }, { HOLE, 409600 * 7 }, { DATA, 1024 }, { HOLE, 409600 * 8 }, { DATA, 1024 }, { HOLE, 409600 * 9 }, { DATA, 1024 }, { HOLE, 409600 * 10}, { DATA, 1024 },/* 10 */ { HOLE, 409600 * 1 }, { DATA, 1024 * 1 }, { HOLE, 409600 * 2 }, { DATA, 1024 * 2 }, { HOLE, 409600 * 3 }, { DATA, 1024 * 3 }, { HOLE, 409600 * 4 }, { DATA, 1024 * 4 }, { HOLE, 409600 * 5 }, { DATA, 1024 * 5 }, { HOLE, 409600 * 6 }, { DATA, 1024 * 6 }, { HOLE, 409600 * 7 }, { DATA, 1024 * 7 }, { HOLE, 409600 * 8 }, { DATA, 1024 * 8 }, { HOLE, 409600 * 9 }, { DATA, 1024 * 9 }, { HOLE, 409600 * 10}, { DATA, 1024 * 10},/* 20 */ { END, 0 } }; const struct sparse sparse_file3[] = { /* This hole size is too small to create a sparse file */ { HOLE, 1 }, { DATA, 10240 }, { HOLE, 1 }, { DATA, 10240 }, { HOLE, 1 }, { DATA, 10240 }, { END, 0 } }; /* * Test for the case that sparse data indicates just the whole file * data. */ test_sparse_whole_file_data(); /* Check if the filesystem where CWD on can * report the number of the holes of a sparse file. */ #ifdef PATH_MAX cwd = getcwd(NULL, PATH_MAX);/* Solaris getcwd needs the size. */ #else cwd = getcwd(NULL, 0); #endif if (!assert(cwd != NULL)) return; if (!is_sparse_supported(cwd)) { free(cwd); skipping("This filesystem or platform do not support " "the reporting of the holes of a sparse file through " "API such as lseek(HOLE)"); return; } /* * Get sparse data through directory traversals. */ assert((a = archive_read_disk_new()) != NULL); verify_sparse_file(a, "file0", sparse_file0, 4); verify_sparse_file(a, "file1", sparse_file1, 3); verify_sparse_file(a, "file2", sparse_file2, 20); /* Encoded non sparse; expect a data block but no sparse entries. */ verify_sparse_file(a, "file3", sparse_file3, 0); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); /* * Get sparse data through archive_read_disk_entry_from_file(). */ assert((a = archive_read_disk_new()) != NULL); verify_sparse_file2(a, "file0", sparse_file0, 5, 0); verify_sparse_file2(a, "file0", sparse_file0, 5, 1); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); free(cwd); } DEFINE_TEST(test_fully_sparse_files) { char *cwd; struct archive *a; const struct sparse sparse_file[] = { { HOLE, 409600 }, { END, 0 } }; /* Check if the filesystem where CWD on can * report the number of the holes of a sparse file. */ #ifdef PATH_MAX cwd = getcwd(NULL, PATH_MAX);/* Solaris getcwd needs the size. */ #else cwd = getcwd(NULL, 0); #endif if (!assert(cwd != NULL)) return; if (!is_sparse_supported(cwd)) { free(cwd); skipping("This filesystem or platform do not support " "the reporting of the holes of a sparse file through " "API such as lseek(HOLE)"); return; } assert((a = archive_read_disk_new()) != NULL); /* Fully sparse files are encoded with a zero-length "data" block. */ verify_sparse_file(a, "file0", sparse_file, 1); assertEqualInt(ARCHIVE_OK, archive_read_free(a)); free(cwd); } Index: head/contrib/libarchive/libarchive/xxhash.c =================================================================== --- head/contrib/libarchive/libarchive/xxhash.c (revision 311902) +++ head/contrib/libarchive/libarchive/xxhash.c (revision 311903) @@ -1,514 +1,515 @@ /* xxHash - Fast Hash algorithm Copyright (C) 2012-2014, Yann Collet. BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php) 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 OWNER 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. You can contact the author at : - xxHash source repository : http://code.google.com/p/xxhash/ */ +#include "archive_platform.h" + #include #include -#include "archive_platform.h" #include "archive_xxhash.h" #ifdef HAVE_LIBLZ4 /*************************************** ** Tuning parameters ****************************************/ /* Unaligned memory access is automatically enabled for "common" CPU, such as x86. ** For others CPU, the compiler will be more cautious, and insert extra code to ensure aligned access is respected. ** If you know your target CPU supports unaligned memory access, you want to force this option manually to improve performance. ** You can also enable this parameter if you know your input data will always be aligned (boundaries of 4, for U32). */ #if defined(__ARM_FEATURE_UNALIGNED) || defined(__i386) || defined(_M_IX86) || defined(__x86_64__) || defined(_M_X64) # define XXH_USE_UNALIGNED_ACCESS 1 #endif /* XXH_ACCEPT_NULL_INPUT_POINTER : ** If the input pointer is a null pointer, xxHash default behavior is to trigger a memory access error, since it is a bad pointer. ** When this option is enabled, xxHash output for null input pointers will be the same as a null-length input. ** This option has a very small performance cost (only measurable on small inputs). ** By default, this option is disabled. To enable it, uncomment below define : ** #define XXH_ACCEPT_NULL_INPUT_POINTER 1 ** XXH_FORCE_NATIVE_FORMAT : ** By default, xxHash library provides endian-independent Hash values, based on little-endian convention. ** Results are therefore identical for little-endian and big-endian CPU. ** This comes at a performance cost for big-endian CPU, since some swapping is required to emulate little-endian format. -** Should endian-independance be of no importance for your application, you may set the #define below to 1. +** Should endian-independence be of no importance for your application, you may set the #define below to 1. ** It will improve speed for Big-endian CPU. ** This option has no impact on Little_Endian CPU. */ #define XXH_FORCE_NATIVE_FORMAT 0 /*************************************** ** Compiler Specific Options ****************************************/ /* Disable some Visual warning messages */ #ifdef _MSC_VER /* Visual Studio */ # pragma warning(disable : 4127) /* disable: C4127: conditional expression is constant */ #endif #ifdef _MSC_VER /* Visual Studio */ # define FORCE_INLINE __forceinline #else # ifdef __GNUC__ # define FORCE_INLINE inline __attribute__((always_inline)) # else # define FORCE_INLINE inline # endif #endif /*************************************** ** Includes & Memory related functions ****************************************/ #define XXH_malloc malloc #define XXH_free free #define XXH_memcpy memcpy static unsigned int XXH32 (const void*, unsigned int, unsigned int); static void* XXH32_init (unsigned int); static XXH_errorcode XXH32_update (void*, const void*, unsigned int); static unsigned int XXH32_digest (void*); /*static int XXH32_sizeofState(void);*/ static XXH_errorcode XXH32_resetState(void*, unsigned int); #define XXH32_SIZEOFSTATE 48 typedef struct { long long ll[(XXH32_SIZEOFSTATE+(sizeof(long long)-1))/sizeof(long long)]; } XXH32_stateSpace_t; static unsigned int XXH32_intermediateDigest (void*); /*************************************** ** Basic Types ****************************************/ #if defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L /* C99 */ # include typedef uint8_t BYTE; typedef uint16_t U16; typedef uint32_t U32; typedef int32_t S32; typedef uint64_t U64; #else typedef unsigned char BYTE; typedef unsigned short U16; typedef unsigned int U32; typedef signed int S32; typedef unsigned long long U64; #endif #if defined(__GNUC__) && !defined(XXH_USE_UNALIGNED_ACCESS) # define _PACKED __attribute__ ((packed)) #else # define _PACKED #endif #if !defined(XXH_USE_UNALIGNED_ACCESS) && !defined(__GNUC__) # ifdef __IBMC__ # pragma pack(1) # else # pragma pack(push, 1) # endif #endif typedef struct _U32_S { U32 v; } _PACKED U32_S; #if !defined(XXH_USE_UNALIGNED_ACCESS) && !defined(__GNUC__) # pragma pack(pop) #endif #define A32(x) (((const U32_S *)(x))->v) /**************************************** ** Compiler-specific Functions and Macros *****************************************/ #define GCC_VERSION (__GNUC__ * 100 + __GNUC_MINOR__) /* Note : although _rotl exists for minGW (GCC under windows), performance seems poor */ #if defined(_MSC_VER) # define XXH_rotl32(x,r) _rotl(x,r) #else # define XXH_rotl32(x,r) ((x << r) | (x >> (32 - r))) #endif #if defined(_MSC_VER) /* Visual Studio */ # define XXH_swap32 _byteswap_ulong #elif GCC_VERSION >= 403 # define XXH_swap32 __builtin_bswap32 #else static inline U32 XXH_swap32 (U32 x) { return ((x << 24) & 0xff000000 ) | ((x << 8) & 0x00ff0000 ) | ((x >> 8) & 0x0000ff00 ) | ((x >> 24) & 0x000000ff );} #endif /*************************************** ** Constants ****************************************/ #define PRIME32_1 2654435761U #define PRIME32_2 2246822519U #define PRIME32_3 3266489917U #define PRIME32_4 668265263U #define PRIME32_5 374761393U /*************************************** ** Architecture Macros ****************************************/ typedef enum { XXH_bigEndian=0, XXH_littleEndian=1 } XXH_endianess; #ifndef XXH_CPU_LITTLE_ENDIAN /* It is possible to define XXH_CPU_LITTLE_ENDIAN externally, for example using a compiler switch */ static const int one = 1; # define XXH_CPU_LITTLE_ENDIAN (*(const char*)(&one)) #endif /*************************************** ** Macros ****************************************/ #define XXH_STATIC_ASSERT(c) { enum { XXH_static_assert = 1/(!!(c)) }; } /* use only *after* variable declarations */ /***************************** ** Memory reads ******************************/ typedef enum { XXH_aligned, XXH_unaligned } XXH_alignment; static FORCE_INLINE U32 XXH_readLE32_align(const U32* ptr, XXH_endianess endian, XXH_alignment align) { if (align==XXH_unaligned) return endian==XXH_littleEndian ? A32(ptr) : XXH_swap32(A32(ptr)); else return endian==XXH_littleEndian ? *ptr : XXH_swap32(*ptr); } static FORCE_INLINE U32 XXH_readLE32(const U32* ptr, XXH_endianess endian) { return XXH_readLE32_align(ptr, endian, XXH_unaligned); } /***************************** ** Simple Hash Functions ******************************/ static FORCE_INLINE U32 XXH32_endian_align(const void* input, unsigned int len, U32 seed, XXH_endianess endian, XXH_alignment align) { const BYTE* p = (const BYTE*)input; const BYTE* bEnd = p + len; U32 h32; #define XXH_get32bits(p) XXH_readLE32_align((const U32*)p, endian, align) #ifdef XXH_ACCEPT_NULL_INPUT_POINTER if (p==NULL) { len=0; bEnd=p=(const BYTE*)(size_t)16; } #endif if (len>=16) { const BYTE* const limit = bEnd - 16; U32 v1 = seed + PRIME32_1 + PRIME32_2; U32 v2 = seed + PRIME32_2; U32 v3 = seed + 0; U32 v4 = seed - PRIME32_1; do { v1 += XXH_get32bits(p) * PRIME32_2; v1 = XXH_rotl32(v1, 13); v1 *= PRIME32_1; p+=4; v2 += XXH_get32bits(p) * PRIME32_2; v2 = XXH_rotl32(v2, 13); v2 *= PRIME32_1; p+=4; v3 += XXH_get32bits(p) * PRIME32_2; v3 = XXH_rotl32(v3, 13); v3 *= PRIME32_1; p+=4; v4 += XXH_get32bits(p) * PRIME32_2; v4 = XXH_rotl32(v4, 13); v4 *= PRIME32_1; p+=4; } while (p<=limit); h32 = XXH_rotl32(v1, 1) + XXH_rotl32(v2, 7) + XXH_rotl32(v3, 12) + XXH_rotl32(v4, 18); } else { h32 = seed + PRIME32_5; } h32 += (U32) len; while (p<=bEnd-4) { h32 += XXH_get32bits(p) * PRIME32_3; h32 = XXH_rotl32(h32, 17) * PRIME32_4 ; p+=4; } while (p> 15; h32 *= PRIME32_2; h32 ^= h32 >> 13; h32 *= PRIME32_3; h32 ^= h32 >> 16; return h32; } U32 XXH32(const void* input, unsigned int len, U32 seed) { #if 0 // Simple version, good for code maintenance, but unfortunately slow for small inputs void* state = XXH32_init(seed); XXH32_update(state, input, len); return XXH32_digest(state); #else XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; # if !defined(XXH_USE_UNALIGNED_ACCESS) if ((((size_t)input) & 3) == 0) /* Input is aligned, let's leverage the speed advantage */ { if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) return XXH32_endian_align(input, len, seed, XXH_littleEndian, XXH_aligned); else return XXH32_endian_align(input, len, seed, XXH_bigEndian, XXH_aligned); } # endif if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) return XXH32_endian_align(input, len, seed, XXH_littleEndian, XXH_unaligned); else return XXH32_endian_align(input, len, seed, XXH_bigEndian, XXH_unaligned); #endif } /***************************** ** Advanced Hash Functions ******************************/ struct XXH_state32_t { U64 total_len; U32 seed; U32 v1; U32 v2; U32 v3; U32 v4; int memsize; char memory[16]; }; #if 0 static int XXH32_sizeofState(void) { XXH_STATIC_ASSERT(XXH32_SIZEOFSTATE >= sizeof(struct XXH_state32_t)); /* A compilation error here means XXH32_SIZEOFSTATE is not large enough */ return sizeof(struct XXH_state32_t); } #endif static XXH_errorcode XXH32_resetState(void* state_in, U32 seed) { struct XXH_state32_t * state = (struct XXH_state32_t *) state_in; state->seed = seed; state->v1 = seed + PRIME32_1 + PRIME32_2; state->v2 = seed + PRIME32_2; state->v3 = seed + 0; state->v4 = seed - PRIME32_1; state->total_len = 0; state->memsize = 0; return XXH_OK; } static void* XXH32_init (U32 seed) { void* state = XXH_malloc (sizeof(struct XXH_state32_t)); XXH32_resetState(state, seed); return state; } static FORCE_INLINE XXH_errorcode XXH32_update_endian (void* state_in, const void* input, int len, XXH_endianess endian) { struct XXH_state32_t * state = (struct XXH_state32_t *) state_in; const BYTE* p = (const BYTE*)input; const BYTE* const bEnd = p + len; #ifdef XXH_ACCEPT_NULL_INPUT_POINTER if (input==NULL) return XXH_ERROR; #endif state->total_len += len; if (state->memsize + len < 16) /* fill in tmp buffer */ { XXH_memcpy(state->memory + state->memsize, input, len); state->memsize += len; return XXH_OK; } if (state->memsize) /* some data left from previous update */ { XXH_memcpy(state->memory + state->memsize, input, 16-state->memsize); { const U32* p32 = (const U32*)state->memory; state->v1 += XXH_readLE32(p32, endian) * PRIME32_2; state->v1 = XXH_rotl32(state->v1, 13); state->v1 *= PRIME32_1; p32++; state->v2 += XXH_readLE32(p32, endian) * PRIME32_2; state->v2 = XXH_rotl32(state->v2, 13); state->v2 *= PRIME32_1; p32++; state->v3 += XXH_readLE32(p32, endian) * PRIME32_2; state->v3 = XXH_rotl32(state->v3, 13); state->v3 *= PRIME32_1; p32++; state->v4 += XXH_readLE32(p32, endian) * PRIME32_2; state->v4 = XXH_rotl32(state->v4, 13); state->v4 *= PRIME32_1; p32++; } p += 16-state->memsize; state->memsize = 0; } if (p <= bEnd-16) { const BYTE* const limit = bEnd - 16; U32 v1 = state->v1; U32 v2 = state->v2; U32 v3 = state->v3; U32 v4 = state->v4; do { v1 += XXH_readLE32((const U32*)p, endian) * PRIME32_2; v1 = XXH_rotl32(v1, 13); v1 *= PRIME32_1; p+=4; v2 += XXH_readLE32((const U32*)p, endian) * PRIME32_2; v2 = XXH_rotl32(v2, 13); v2 *= PRIME32_1; p+=4; v3 += XXH_readLE32((const U32*)p, endian) * PRIME32_2; v3 = XXH_rotl32(v3, 13); v3 *= PRIME32_1; p+=4; v4 += XXH_readLE32((const U32*)p, endian) * PRIME32_2; v4 = XXH_rotl32(v4, 13); v4 *= PRIME32_1; p+=4; } while (p<=limit); state->v1 = v1; state->v2 = v2; state->v3 = v3; state->v4 = v4; } if (p < bEnd) { XXH_memcpy(state->memory, p, bEnd-p); state->memsize = (int)(bEnd-p); } return XXH_OK; } static XXH_errorcode XXH32_update (void* state_in, const void* input, unsigned int len) { XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) return XXH32_update_endian(state_in, input, len, XXH_littleEndian); else return XXH32_update_endian(state_in, input, len, XXH_bigEndian); } static FORCE_INLINE U32 XXH32_intermediateDigest_endian (void* state_in, XXH_endianess endian) { struct XXH_state32_t * state = (struct XXH_state32_t *) state_in; const BYTE * p = (const BYTE*)state->memory; BYTE* bEnd = (BYTE*)state->memory + state->memsize; U32 h32; if (state->total_len >= 16) { h32 = XXH_rotl32(state->v1, 1) + XXH_rotl32(state->v2, 7) + XXH_rotl32(state->v3, 12) + XXH_rotl32(state->v4, 18); } else { h32 = state->seed + PRIME32_5; } h32 += (U32) state->total_len; while (p<=bEnd-4) { h32 += XXH_readLE32((const U32*)p, endian) * PRIME32_3; h32 = XXH_rotl32(h32, 17) * PRIME32_4; p+=4; } while (p> 15; h32 *= PRIME32_2; h32 ^= h32 >> 13; h32 *= PRIME32_3; h32 ^= h32 >> 16; return h32; } static U32 XXH32_intermediateDigest (void* state_in) { XXH_endianess endian_detected = (XXH_endianess)XXH_CPU_LITTLE_ENDIAN; if ((endian_detected==XXH_littleEndian) || XXH_FORCE_NATIVE_FORMAT) return XXH32_intermediateDigest_endian(state_in, XXH_littleEndian); else return XXH32_intermediateDigest_endian(state_in, XXH_bigEndian); } static U32 XXH32_digest (void* state_in) { U32 h32 = XXH32_intermediateDigest(state_in); XXH_free(state_in); return h32; } const struct archive_xxhash __archive_xxhash = { XXH32, XXH32_init, XXH32_update, XXH32_digest }; #else /* * Define an empty version of the struct if we aren't using the LZ4 library. */ const struct archive_xxhash __archive_xxhash = { NULL, NULL, NULL, NULL }; #endif /* HAVE_LIBLZ4 */ Index: head/contrib/libarchive =================================================================== --- head/contrib/libarchive (revision 311902) +++ head/contrib/libarchive (revision 311903) Property changes on: head/contrib/libarchive ___________________________________________________________________ Modified: svn:mergeinfo ## -0,0 +0,1 ## Merged /vendor/libarchive/dist:r311899