2 * This handles recursive filename detection with exclude
3 * files, index knowledge etc..
5 * Copyright (C) Linus Torvalds, 2005-2006
6 * Junio Hamano, 2005-2006
12 struct path_simplify {
17 static int read_directory_recursive(struct dir_struct *dir, const char *path, int len,
18 int check_only, const struct path_simplify *simplify);
19 static int get_dtype(struct dirent *de, const char *path, int len);
21 /* helper string functions with support for the ignore_case flag */
22 int strcmp_icase(const char *a, const char *b)
24 return ignore_case ? strcasecmp(a, b) : strcmp(a, b);
27 int strncmp_icase(const char *a, const char *b, size_t count)
29 return ignore_case ? strncasecmp(a, b, count) : strncmp(a, b, count);
32 int fnmatch_icase(const char *pattern, const char *string, int flags)
34 return fnmatch(pattern, string, flags | (ignore_case ? FNM_CASEFOLD : 0));
37 inline int git_fnmatch(const char *pattern, const char *string,
38 int flags, int prefix)
41 if (flags & GFNM_PATHNAME)
42 fnm_flags |= FNM_PATHNAME;
44 if (strncmp(pattern, string, prefix))
49 if (flags & GFNM_ONESTAR) {
50 int pattern_len = strlen(++pattern);
51 int string_len = strlen(string);
52 return string_len < pattern_len ||
54 string + string_len - pattern_len);
56 return fnmatch(pattern, string, fnm_flags);
59 static size_t common_prefix_len(const char **pathspec)
61 const char *n, *first;
63 int literal = limit_pathspec_to_literal();
69 while ((n = *pathspec++)) {
71 for (i = 0; first == n || i < max; i++) {
73 if (!c || c != first[i] || (!literal && is_glob_special(c)))
78 if (first == n || len < max) {
88 * Returns a copy of the longest leading path common among all
91 char *common_prefix(const char **pathspec)
93 unsigned long len = common_prefix_len(pathspec);
95 return len ? xmemdupz(*pathspec, len) : NULL;
98 int fill_directory(struct dir_struct *dir, const char **pathspec)
103 * Calculate common prefix for the pathspec, and
104 * use that to optimize the directory walk
106 len = common_prefix_len(pathspec);
108 /* Read the directory and prune it */
109 read_directory(dir, pathspec ? *pathspec : "", len, pathspec);
113 int within_depth(const char *name, int namelen,
114 int depth, int max_depth)
116 const char *cp = name, *cpe = name + namelen;
122 if (depth > max_depth)
129 * Does 'match' match the given name?
130 * A match is found if
132 * (1) the 'match' string is leading directory of 'name', or
133 * (2) the 'match' string is a wildcard and matches 'name', or
134 * (3) the 'match' string is exactly the same as 'name'.
136 * and the return value tells which case it was.
138 * It returns 0 when there is no match.
140 static int match_one(const char *match, const char *name, int namelen)
143 int literal = limit_pathspec_to_literal();
145 /* If the match was just the prefix, we matched */
147 return MATCHED_RECURSIVELY;
151 unsigned char c1 = tolower(*match);
152 unsigned char c2 = tolower(*name);
153 if (c1 == '\0' || (!literal && is_glob_special(c1)))
163 unsigned char c1 = *match;
164 unsigned char c2 = *name;
165 if (c1 == '\0' || (!literal && is_glob_special(c1)))
176 * If we don't match the matchstring exactly,
177 * we need to match by fnmatch
179 matchlen = strlen(match);
180 if (strncmp_icase(match, name, matchlen)) {
183 return !fnmatch_icase(match, name, 0) ? MATCHED_FNMATCH : 0;
186 if (namelen == matchlen)
187 return MATCHED_EXACTLY;
188 if (match[matchlen-1] == '/' || name[matchlen] == '/')
189 return MATCHED_RECURSIVELY;
194 * Given a name and a list of pathspecs, see if the name matches
195 * any of the pathspecs. The caller is also interested in seeing
196 * all pathspec matches some names it calls this function with
197 * (otherwise the user could have mistyped the unmatched pathspec),
198 * and a mark is left in seen[] array for pathspec element that
199 * actually matched anything.
201 int match_pathspec(const char **pathspec, const char *name, int namelen,
202 int prefix, char *seen)
212 for (i = 0; pathspec[i] != NULL; i++) {
214 const char *match = pathspec[i] + prefix;
215 if (seen && seen[i] == MATCHED_EXACTLY)
217 how = match_one(match, name, namelen);
221 if (seen && seen[i] < how)
229 * Does 'match' match the given name?
230 * A match is found if
232 * (1) the 'match' string is leading directory of 'name', or
233 * (2) the 'match' string is a wildcard and matches 'name', or
234 * (3) the 'match' string is exactly the same as 'name'.
236 * and the return value tells which case it was.
238 * It returns 0 when there is no match.
240 static int match_pathspec_item(const struct pathspec_item *item, int prefix,
241 const char *name, int namelen)
243 /* name/namelen has prefix cut off by caller */
244 const char *match = item->match + prefix;
245 int matchlen = item->len - prefix;
247 /* If the match was just the prefix, we matched */
249 return MATCHED_RECURSIVELY;
251 if (matchlen <= namelen && !strncmp(match, name, matchlen)) {
252 if (matchlen == namelen)
253 return MATCHED_EXACTLY;
255 if (match[matchlen-1] == '/' || name[matchlen] == '/')
256 return MATCHED_RECURSIVELY;
259 if (item->nowildcard_len < item->len &&
260 !git_fnmatch(match, name,
261 item->flags & PATHSPEC_ONESTAR ? GFNM_ONESTAR : 0,
262 item->nowildcard_len - prefix))
263 return MATCHED_FNMATCH;
269 * Given a name and a list of pathspecs, see if the name matches
270 * any of the pathspecs. The caller is also interested in seeing
271 * all pathspec matches some names it calls this function with
272 * (otherwise the user could have mistyped the unmatched pathspec),
273 * and a mark is left in seen[] array for pathspec element that
274 * actually matched anything.
276 int match_pathspec_depth(const struct pathspec *ps,
277 const char *name, int namelen,
278 int prefix, char *seen)
283 if (!ps->recursive || ps->max_depth == -1)
284 return MATCHED_RECURSIVELY;
286 if (within_depth(name, namelen, 0, ps->max_depth))
287 return MATCHED_EXACTLY;
295 for (i = ps->nr - 1; i >= 0; i--) {
297 if (seen && seen[i] == MATCHED_EXACTLY)
299 how = match_pathspec_item(ps->items+i, prefix, name, namelen);
300 if (ps->recursive && ps->max_depth != -1 &&
301 how && how != MATCHED_FNMATCH) {
302 int len = ps->items[i].len;
303 if (name[len] == '/')
305 if (within_depth(name+len, namelen-len, 0, ps->max_depth))
306 how = MATCHED_EXACTLY;
313 if (seen && seen[i] < how)
321 * Return the length of the "simple" part of a path match limiter.
323 static int simple_length(const char *match)
328 unsigned char c = *match++;
330 if (c == '\0' || is_glob_special(c))
335 static int no_wildcard(const char *string)
337 return string[simple_length(string)] == '\0';
340 void parse_exclude_pattern(const char **pattern,
345 const char *p = *pattern;
350 *flags |= EXC_FLAG_NEGATIVE;
354 if (len && p[len - 1] == '/') {
356 *flags |= EXC_FLAG_MUSTBEDIR;
358 for (i = 0; i < len; i++) {
363 *flags |= EXC_FLAG_NODIR;
364 *nowildcardlen = simple_length(p);
366 * we should have excluded the trailing slash from 'p' too,
367 * but that's one more allocation. Instead just make sure
368 * nowildcardlen does not exceed real patternlen
370 if (*nowildcardlen > len)
371 *nowildcardlen = len;
372 if (*p == '*' && no_wildcard(p + 1))
373 *flags |= EXC_FLAG_ENDSWITH;
378 void add_exclude(const char *string, const char *base,
379 int baselen, struct exclude_list *which)
386 parse_exclude_pattern(&string, &patternlen, &flags, &nowildcardlen);
387 if (flags & EXC_FLAG_MUSTBEDIR) {
389 x = xmalloc(sizeof(*x) + patternlen + 1);
391 memcpy(s, string, patternlen);
392 s[patternlen] = '\0';
395 x = xmalloc(sizeof(*x));
398 x->patternlen = patternlen;
399 x->nowildcardlen = nowildcardlen;
401 x->baselen = baselen;
403 ALLOC_GROW(which->excludes, which->nr + 1, which->alloc);
404 which->excludes[which->nr++] = x;
407 static void *read_skip_worktree_file_from_index(const char *path, size_t *size)
411 enum object_type type;
413 struct index_state *istate = &the_index;
416 pos = index_name_pos(istate, path, len);
419 if (!ce_skip_worktree(istate->cache[pos]))
421 data = read_sha1_file(istate->cache[pos]->sha1, &type, &sz);
422 if (!data || type != OBJ_BLOB) {
430 void free_excludes(struct exclude_list *el)
434 for (i = 0; i < el->nr; i++)
435 free(el->excludes[i]);
442 int add_excludes_from_file_to_list(const char *fname,
446 struct exclude_list *which,
454 fd = open(fname, O_RDONLY);
455 if (fd < 0 || fstat(fd, &st) < 0) {
457 warn_on_inaccessible(fname);
461 (buf = read_skip_worktree_file_from_index(fname, &size)) == NULL)
467 if (buf[size-1] != '\n') {
468 buf = xrealloc(buf, size+1);
473 size = xsize_t(st.st_size);
478 buf = xmalloc(size+1);
479 if (read_in_full(fd, buf, size) != size) {
491 for (i = 0; i < size; i++) {
492 if (buf[i] == '\n') {
493 if (entry != buf + i && entry[0] != '#') {
494 buf[i - (i && buf[i-1] == '\r')] = 0;
495 add_exclude(entry, base, baselen, which);
503 void add_excludes_from_file(struct dir_struct *dir, const char *fname)
505 if (add_excludes_from_file_to_list(fname, "", 0, NULL,
506 &dir->exclude_list[EXC_FILE], 0) < 0)
507 die("cannot use %s as an exclude file", fname);
510 static void prep_exclude(struct dir_struct *dir, const char *base, int baselen)
512 struct exclude_list *el;
513 struct exclude_stack *stk = NULL;
516 if ((!dir->exclude_per_dir) ||
517 (baselen + strlen(dir->exclude_per_dir) >= PATH_MAX))
518 return; /* too long a path -- ignore */
520 /* Pop the ones that are not the prefix of the path being checked. */
521 el = &dir->exclude_list[EXC_DIRS];
522 while ((stk = dir->exclude_stack) != NULL) {
523 if (stk->baselen <= baselen &&
524 !strncmp(dir->basebuf, base, stk->baselen))
526 dir->exclude_stack = stk->prev;
527 while (stk->exclude_ix < el->nr)
528 free(el->excludes[--el->nr]);
533 /* Read from the parent directories and push them down. */
534 current = stk ? stk->baselen : -1;
535 while (current < baselen) {
536 struct exclude_stack *stk = xcalloc(1, sizeof(*stk));
544 cp = strchr(base + current + 1, '/');
546 die("oops in prep_exclude");
549 stk->prev = dir->exclude_stack;
550 stk->baselen = cp - base;
551 stk->exclude_ix = el->nr;
552 memcpy(dir->basebuf + current, base + current,
553 stk->baselen - current);
554 strcpy(dir->basebuf + stk->baselen, dir->exclude_per_dir);
555 add_excludes_from_file_to_list(dir->basebuf,
556 dir->basebuf, stk->baselen,
557 &stk->filebuf, el, 1);
558 dir->exclude_stack = stk;
559 current = stk->baselen;
561 dir->basebuf[baselen] = '\0';
564 int match_basename(const char *basename, int basenamelen,
565 const char *pattern, int prefix, int patternlen,
568 if (prefix == patternlen) {
569 if (!strcmp_icase(pattern, basename))
571 } else if (flags & EXC_FLAG_ENDSWITH) {
572 if (patternlen - 1 <= basenamelen &&
573 !strcmp_icase(pattern + 1,
574 basename + basenamelen - patternlen + 1))
577 if (fnmatch_icase(pattern, basename, 0) == 0)
583 int match_pathname(const char *pathname, int pathlen,
584 const char *base, int baselen,
585 const char *pattern, int prefix, int patternlen,
592 * match with FNM_PATHNAME; the pattern has base implicitly
595 if (*pattern == '/') {
601 * baselen does not count the trailing slash. base[] may or
602 * may not end with a trailing slash though.
604 if (pathlen < baselen + 1 ||
605 (baselen && pathname[baselen] != '/') ||
606 strncmp_icase(pathname, base, baselen))
609 namelen = baselen ? pathlen - baselen - 1 : pathlen;
610 name = pathname + pathlen - namelen;
614 * if the non-wildcard part is longer than the
615 * remaining pathname, surely it cannot match.
617 if (prefix > namelen)
620 if (strncmp_icase(pattern, name, prefix))
627 return fnmatch_icase(pattern, name, FNM_PATHNAME) == 0;
630 /* Scan the list and let the last match determine the fate.
631 * Return 1 for exclude, 0 for include and -1 for undecided.
633 int excluded_from_list(const char *pathname,
634 int pathlen, const char *basename, int *dtype,
635 struct exclude_list *el)
640 return -1; /* undefined */
642 for (i = el->nr - 1; 0 <= i; i--) {
643 struct exclude *x = el->excludes[i];
644 const char *exclude = x->pattern;
645 int to_exclude = x->flags & EXC_FLAG_NEGATIVE ? 0 : 1;
646 int prefix = x->nowildcardlen;
648 if (x->flags & EXC_FLAG_MUSTBEDIR) {
649 if (*dtype == DT_UNKNOWN)
650 *dtype = get_dtype(NULL, pathname, pathlen);
651 if (*dtype != DT_DIR)
655 if (x->flags & EXC_FLAG_NODIR) {
656 if (match_basename(basename,
657 pathlen - (basename - pathname),
658 exclude, prefix, x->patternlen,
664 assert(x->baselen == 0 || x->base[x->baselen - 1] == '/');
665 if (match_pathname(pathname, pathlen,
666 x->base, x->baselen ? x->baselen - 1 : 0,
667 exclude, prefix, x->patternlen, x->flags))
670 return -1; /* undecided */
673 static int excluded(struct dir_struct *dir, const char *pathname, int *dtype_p)
675 int pathlen = strlen(pathname);
677 const char *basename = strrchr(pathname, '/');
678 basename = (basename) ? basename+1 : pathname;
680 prep_exclude(dir, pathname, basename-pathname);
681 for (st = EXC_CMDL; st <= EXC_FILE; st++) {
682 switch (excluded_from_list(pathname, pathlen, basename,
683 dtype_p, &dir->exclude_list[st])) {
693 void path_exclude_check_init(struct path_exclude_check *check,
694 struct dir_struct *dir)
697 strbuf_init(&check->path, 256);
700 void path_exclude_check_clear(struct path_exclude_check *check)
702 strbuf_release(&check->path);
706 * Is this name excluded? This is for a caller like show_files() that
707 * do not honor directory hierarchy and iterate through paths that are
708 * possibly in an ignored directory.
710 * A path to a directory known to be excluded is left in check->path to
711 * optimize for repeated checks for files in the same excluded directory.
713 int path_excluded(struct path_exclude_check *check,
714 const char *name, int namelen, int *dtype)
717 struct strbuf *path = &check->path;
720 * we allow the caller to pass namelen as an optimization; it
721 * must match the length of the name, as we eventually call
722 * excluded() on the whole name string.
725 namelen = strlen(name);
728 path->len <= namelen &&
729 !memcmp(name, path->buf, path->len) &&
730 (!name[path->len] || name[path->len] == '/'))
733 strbuf_setlen(path, 0);
734 for (i = 0; name[i]; i++) {
739 if (excluded(check->dir, path->buf, &dt))
742 strbuf_addch(path, ch);
745 /* An entry in the index; cannot be a directory with subentries */
746 strbuf_setlen(path, 0);
748 return excluded(check->dir, name, dtype);
751 static struct dir_entry *dir_entry_new(const char *pathname, int len)
753 struct dir_entry *ent;
755 ent = xmalloc(sizeof(*ent) + len + 1);
757 memcpy(ent->name, pathname, len);
762 static struct dir_entry *dir_add_name(struct dir_struct *dir, const char *pathname, int len)
764 if (cache_name_exists(pathname, len, ignore_case))
767 ALLOC_GROW(dir->entries, dir->nr+1, dir->alloc);
768 return dir->entries[dir->nr++] = dir_entry_new(pathname, len);
771 struct dir_entry *dir_add_ignored(struct dir_struct *dir, const char *pathname, int len)
773 if (!cache_name_is_other(pathname, len))
776 ALLOC_GROW(dir->ignored, dir->ignored_nr+1, dir->ignored_alloc);
777 return dir->ignored[dir->ignored_nr++] = dir_entry_new(pathname, len);
781 index_nonexistent = 0,
787 * Do not use the alphabetically stored index to look up
788 * the directory name; instead, use the case insensitive
791 static enum exist_status directory_exists_in_index_icase(const char *dirname, int len)
793 struct cache_entry *ce = index_name_exists(&the_index, dirname, len + 1, ignore_case);
794 unsigned char endchar;
797 return index_nonexistent;
798 endchar = ce->name[len];
801 * The cache_entry structure returned will contain this dirname
802 * and possibly additional path components.
805 return index_directory;
808 * If there are no additional path components, then this cache_entry
809 * represents a submodule. Submodules, despite being directories,
810 * are stored in the cache without a closing slash.
812 if (!endchar && S_ISGITLINK(ce->ce_mode))
815 /* This should never be hit, but it exists just in case. */
816 return index_nonexistent;
820 * The index sorts alphabetically by entry name, which
821 * means that a gitlink sorts as '\0' at the end, while
822 * a directory (which is defined not as an entry, but as
823 * the files it contains) will sort with the '/' at the
826 static enum exist_status directory_exists_in_index(const char *dirname, int len)
831 return directory_exists_in_index_icase(dirname, len);
833 pos = cache_name_pos(dirname, len);
836 while (pos < active_nr) {
837 struct cache_entry *ce = active_cache[pos++];
838 unsigned char endchar;
840 if (strncmp(ce->name, dirname, len))
842 endchar = ce->name[len];
846 return index_directory;
847 if (!endchar && S_ISGITLINK(ce->ce_mode))
850 return index_nonexistent;
854 * When we find a directory when traversing the filesystem, we
855 * have three distinct cases:
858 * - see it as a directory
861 * and which one we choose depends on a combination of existing
862 * git index contents and the flags passed into the directory
865 * Case 1: If we *already* have entries in the index under that
866 * directory name, we always recurse into the directory to see
869 * Case 2: If we *already* have that directory name as a gitlink,
870 * we always continue to see it as a gitlink, regardless of whether
871 * there is an actual git directory there or not (it might not
872 * be checked out as a subproject!)
874 * Case 3: if we didn't have it in the index previously, we
875 * have a few sub-cases:
877 * (a) if "show_other_directories" is true, we show it as
878 * just a directory, unless "hide_empty_directories" is
879 * also true and the directory is empty, in which case
880 * we just ignore it entirely.
881 * (b) if it looks like a git directory, and we don't have
882 * 'no_gitlinks' set we treat it as a gitlink, and show it
884 * (c) otherwise, we recurse into it.
886 enum directory_treatment {
889 recurse_into_directory
892 static enum directory_treatment treat_directory(struct dir_struct *dir,
893 const char *dirname, int len,
894 const struct path_simplify *simplify)
896 /* The "len-1" is to strip the final '/' */
897 switch (directory_exists_in_index(dirname, len-1)) {
898 case index_directory:
899 return recurse_into_directory;
902 if (dir->flags & DIR_SHOW_OTHER_DIRECTORIES)
903 return ignore_directory;
904 return show_directory;
906 case index_nonexistent:
907 if (dir->flags & DIR_SHOW_OTHER_DIRECTORIES)
909 if (!(dir->flags & DIR_NO_GITLINKS)) {
910 unsigned char sha1[20];
911 if (resolve_gitlink_ref(dirname, "HEAD", sha1) == 0)
912 return show_directory;
914 return recurse_into_directory;
917 /* This is the "show_other_directories" case */
918 if (!(dir->flags & DIR_HIDE_EMPTY_DIRECTORIES))
919 return show_directory;
920 if (!read_directory_recursive(dir, dirname, len, 1, simplify))
921 return ignore_directory;
922 return show_directory;
926 * This is an inexact early pruning of any recursive directory
927 * reading - if the path cannot possibly be in the pathspec,
928 * return true, and we'll skip it early.
930 static int simplify_away(const char *path, int pathlen, const struct path_simplify *simplify)
934 const char *match = simplify->path;
935 int len = simplify->len;
941 if (!memcmp(path, match, len))
951 * This function tells us whether an excluded path matches a
952 * list of "interesting" pathspecs. That is, whether a path matched
953 * by any of the pathspecs could possibly be ignored by excluding
954 * the specified path. This can happen if:
956 * 1. the path is mentioned explicitly in the pathspec
958 * 2. the path is a directory prefix of some element in the
961 static int exclude_matches_pathspec(const char *path, int len,
962 const struct path_simplify *simplify)
965 for (; simplify->path; simplify++) {
966 if (len == simplify->len
967 && !memcmp(path, simplify->path, len))
969 if (len < simplify->len
970 && simplify->path[len] == '/'
971 && !memcmp(path, simplify->path, len))
978 static int get_index_dtype(const char *path, int len)
981 struct cache_entry *ce;
983 ce = cache_name_exists(path, len, 0);
985 if (!ce_uptodate(ce))
987 if (S_ISGITLINK(ce->ce_mode))
990 * Nobody actually cares about the
991 * difference between DT_LNK and DT_REG
996 /* Try to look it up as a directory */
997 pos = cache_name_pos(path, len);
1001 while (pos < active_nr) {
1002 ce = active_cache[pos++];
1003 if (strncmp(ce->name, path, len))
1005 if (ce->name[len] > '/')
1007 if (ce->name[len] < '/')
1009 if (!ce_uptodate(ce))
1010 break; /* continue? */
1016 static int get_dtype(struct dirent *de, const char *path, int len)
1018 int dtype = de ? DTYPE(de) : DT_UNKNOWN;
1021 if (dtype != DT_UNKNOWN)
1023 dtype = get_index_dtype(path, len);
1024 if (dtype != DT_UNKNOWN)
1026 if (lstat(path, &st))
1028 if (S_ISREG(st.st_mode))
1030 if (S_ISDIR(st.st_mode))
1032 if (S_ISLNK(st.st_mode))
1037 enum path_treatment {
1043 static enum path_treatment treat_one_path(struct dir_struct *dir,
1044 struct strbuf *path,
1045 const struct path_simplify *simplify,
1046 int dtype, struct dirent *de)
1048 int exclude = excluded(dir, path->buf, &dtype);
1049 if (exclude && (dir->flags & DIR_COLLECT_IGNORED)
1050 && exclude_matches_pathspec(path->buf, path->len, simplify))
1051 dir_add_ignored(dir, path->buf, path->len);
1054 * Excluded? If we don't explicitly want to show
1055 * ignored files, ignore it
1057 if (exclude && !(dir->flags & DIR_SHOW_IGNORED))
1058 return path_ignored;
1060 if (dtype == DT_UNKNOWN)
1061 dtype = get_dtype(de, path->buf, path->len);
1064 * Do we want to see just the ignored files?
1065 * We still need to recurse into directories,
1066 * even if we don't ignore them, since the
1067 * directory may contain files that we do..
1069 if (!exclude && (dir->flags & DIR_SHOW_IGNORED)) {
1070 if (dtype != DT_DIR)
1071 return path_ignored;
1076 return path_ignored;
1078 strbuf_addch(path, '/');
1079 switch (treat_directory(dir, path->buf, path->len, simplify)) {
1080 case show_directory:
1081 if (exclude != !!(dir->flags
1082 & DIR_SHOW_IGNORED))
1083 return path_ignored;
1085 case recurse_into_directory:
1086 return path_recurse;
1087 case ignore_directory:
1088 return path_ignored;
1095 return path_handled;
1098 static enum path_treatment treat_path(struct dir_struct *dir,
1100 struct strbuf *path,
1102 const struct path_simplify *simplify)
1106 if (is_dot_or_dotdot(de->d_name) || !strcmp(de->d_name, ".git"))
1107 return path_ignored;
1108 strbuf_setlen(path, baselen);
1109 strbuf_addstr(path, de->d_name);
1110 if (simplify_away(path->buf, path->len, simplify))
1111 return path_ignored;
1114 return treat_one_path(dir, path, simplify, dtype, de);
1118 * Read a directory tree. We currently ignore anything but
1119 * directories, regular files and symlinks. That's because git
1120 * doesn't handle them at all yet. Maybe that will change some
1123 * Also, we ignore the name ".git" (even if it is not a directory).
1124 * That likely will not change.
1126 static int read_directory_recursive(struct dir_struct *dir,
1127 const char *base, int baselen,
1129 const struct path_simplify *simplify)
1134 struct strbuf path = STRBUF_INIT;
1136 strbuf_add(&path, base, baselen);
1138 fdir = opendir(path.len ? path.buf : ".");
1142 while ((de = readdir(fdir)) != NULL) {
1143 switch (treat_path(dir, de, &path, baselen, simplify)) {
1145 contents += read_directory_recursive(dir, path.buf,
1157 dir_add_name(dir, path.buf, path.len);
1161 strbuf_release(&path);
1166 static int cmp_name(const void *p1, const void *p2)
1168 const struct dir_entry *e1 = *(const struct dir_entry **)p1;
1169 const struct dir_entry *e2 = *(const struct dir_entry **)p2;
1171 return cache_name_compare(e1->name, e1->len,
1175 static struct path_simplify *create_simplify(const char **pathspec)
1178 struct path_simplify *simplify = NULL;
1183 for (nr = 0 ; ; nr++) {
1186 alloc = alloc_nr(alloc);
1187 simplify = xrealloc(simplify, alloc * sizeof(*simplify));
1189 match = *pathspec++;
1192 simplify[nr].path = match;
1193 simplify[nr].len = simple_length(match);
1195 simplify[nr].path = NULL;
1196 simplify[nr].len = 0;
1200 static void free_simplify(struct path_simplify *simplify)
1205 static int treat_leading_path(struct dir_struct *dir,
1206 const char *path, int len,
1207 const struct path_simplify *simplify)
1209 struct strbuf sb = STRBUF_INIT;
1210 int baselen, rc = 0;
1213 while (len && path[len - 1] == '/')
1219 cp = path + baselen + !!baselen;
1220 cp = memchr(cp, '/', path + len - cp);
1224 baselen = cp - path;
1225 strbuf_setlen(&sb, 0);
1226 strbuf_add(&sb, path, baselen);
1227 if (!is_directory(sb.buf))
1229 if (simplify_away(sb.buf, sb.len, simplify))
1231 if (treat_one_path(dir, &sb, simplify,
1232 DT_DIR, NULL) == path_ignored)
1233 break; /* do not recurse into it */
1234 if (len <= baselen) {
1236 break; /* finished checking */
1239 strbuf_release(&sb);
1243 int read_directory(struct dir_struct *dir, const char *path, int len, const char **pathspec)
1245 struct path_simplify *simplify;
1247 if (has_symlink_leading_path(path, len))
1250 simplify = create_simplify(pathspec);
1251 if (!len || treat_leading_path(dir, path, len, simplify))
1252 read_directory_recursive(dir, path, len, 0, simplify);
1253 free_simplify(simplify);
1254 qsort(dir->entries, dir->nr, sizeof(struct dir_entry *), cmp_name);
1255 qsort(dir->ignored, dir->ignored_nr, sizeof(struct dir_entry *), cmp_name);
1259 int file_exists(const char *f)
1262 return lstat(f, &sb) == 0;
1266 * Given two normalized paths (a trailing slash is ok), if subdir is
1267 * outside dir, return -1. Otherwise return the offset in subdir that
1268 * can be used as relative path to dir.
1270 int dir_inside_of(const char *subdir, const char *dir)
1274 assert(dir && subdir && *dir && *subdir);
1276 while (*dir && *subdir && *dir == *subdir) {
1282 /* hel[p]/me vs hel[l]/yeah */
1283 if (*dir && *subdir)
1287 return !*dir ? offset : -1; /* same dir */
1289 /* foo/[b]ar vs foo/[] */
1290 if (is_dir_sep(dir[-1]))
1291 return is_dir_sep(subdir[-1]) ? offset : -1;
1293 /* foo[/]bar vs foo[] */
1294 return is_dir_sep(*subdir) ? offset + 1 : -1;
1297 int is_inside_dir(const char *dir)
1302 if (!getcwd(cwd, sizeof(cwd)))
1303 die_errno("can't find the current directory");
1304 return dir_inside_of(cwd, dir) >= 0;
1307 int is_empty_dir(const char *path)
1309 DIR *dir = opendir(path);
1316 while ((e = readdir(dir)) != NULL)
1317 if (!is_dot_or_dotdot(e->d_name)) {
1326 static int remove_dir_recurse(struct strbuf *path, int flag, int *kept_up)
1330 int ret = 0, original_len = path->len, len, kept_down = 0;
1331 int only_empty = (flag & REMOVE_DIR_EMPTY_ONLY);
1332 int keep_toplevel = (flag & REMOVE_DIR_KEEP_TOPLEVEL);
1333 unsigned char submodule_head[20];
1335 if ((flag & REMOVE_DIR_KEEP_NESTED_GIT) &&
1336 !resolve_gitlink_ref(path->buf, "HEAD", submodule_head)) {
1337 /* Do not descend and nuke a nested git work tree. */
1343 flag &= ~REMOVE_DIR_KEEP_TOPLEVEL;
1344 dir = opendir(path->buf);
1346 /* an empty dir could be removed even if it is unreadble */
1348 return rmdir(path->buf);
1352 if (path->buf[original_len - 1] != '/')
1353 strbuf_addch(path, '/');
1356 while ((e = readdir(dir)) != NULL) {
1358 if (is_dot_or_dotdot(e->d_name))
1361 strbuf_setlen(path, len);
1362 strbuf_addstr(path, e->d_name);
1363 if (lstat(path->buf, &st))
1365 else if (S_ISDIR(st.st_mode)) {
1366 if (!remove_dir_recurse(path, flag, &kept_down))
1367 continue; /* happy */
1368 } else if (!only_empty && !unlink(path->buf))
1369 continue; /* happy, too */
1371 /* path too long, stat fails, or non-directory still exists */
1377 strbuf_setlen(path, original_len);
1378 if (!ret && !keep_toplevel && !kept_down)
1379 ret = rmdir(path->buf);
1382 * report the uplevel that it is not an error that we
1383 * did not rmdir() our directory.
1389 int remove_dir_recursively(struct strbuf *path, int flag)
1391 return remove_dir_recurse(path, flag, NULL);
1394 void setup_standard_excludes(struct dir_struct *dir)
1399 dir->exclude_per_dir = ".gitignore";
1400 path = git_path("info/exclude");
1401 if (!excludes_file) {
1402 home_config_paths(NULL, &xdg_path, "ignore");
1403 excludes_file = xdg_path;
1405 if (!access_or_warn(path, R_OK))
1406 add_excludes_from_file(dir, path);
1407 if (excludes_file && !access_or_warn(excludes_file, R_OK))
1408 add_excludes_from_file(dir, excludes_file);
1411 int remove_path(const char *name)
1415 if (unlink(name) && errno != ENOENT)
1418 slash = strrchr(name, '/');
1420 char *dirs = xstrdup(name);
1421 slash = dirs + (slash - name);
1424 } while (rmdir(dirs) == 0 && (slash = strrchr(dirs, '/')));
1430 static int pathspec_item_cmp(const void *a_, const void *b_)
1432 struct pathspec_item *a, *b;
1434 a = (struct pathspec_item *)a_;
1435 b = (struct pathspec_item *)b_;
1436 return strcmp(a->match, b->match);
1439 int init_pathspec(struct pathspec *pathspec, const char **paths)
1441 const char **p = paths;
1444 memset(pathspec, 0, sizeof(*pathspec));
1449 pathspec->raw = paths;
1450 pathspec->nr = p - paths;
1454 pathspec->items = xmalloc(sizeof(struct pathspec_item)*pathspec->nr);
1455 for (i = 0; i < pathspec->nr; i++) {
1456 struct pathspec_item *item = pathspec->items+i;
1457 const char *path = paths[i];
1460 item->len = strlen(path);
1462 if (limit_pathspec_to_literal()) {
1463 item->nowildcard_len = item->len;
1465 item->nowildcard_len = simple_length(path);
1466 if (item->nowildcard_len < item->len) {
1467 pathspec->has_wildcard = 1;
1468 if (path[item->nowildcard_len] == '*' &&
1469 no_wildcard(path + item->nowildcard_len + 1))
1470 item->flags |= PATHSPEC_ONESTAR;
1475 qsort(pathspec->items, pathspec->nr,
1476 sizeof(struct pathspec_item), pathspec_item_cmp);
1481 void free_pathspec(struct pathspec *pathspec)
1483 free(pathspec->items);
1484 pathspec->items = NULL;
1487 int limit_pathspec_to_literal(void)
1489 static int flag = -1;
1491 flag = git_env_bool(GIT_LITERAL_PATHSPECS_ENVIRONMENT, 0);