dir.c: make excluded() file scope static
[alt-git.git] / dir.c
blob79e43377ad6a1abcfa21e740522d15e4a0b3c566
1 /*
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
7 */
8 #include "cache.h"
9 #include "dir.h"
10 #include "refs.h"
12 struct path_simplify {
13 int len;
14 const char *path;
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 static size_t common_prefix_len(const char **pathspec)
39 const char *n, *first;
40 size_t max = 0;
42 if (!pathspec)
43 return max;
45 first = *pathspec;
46 while ((n = *pathspec++)) {
47 size_t i, len = 0;
48 for (i = 0; first == n || i < max; i++) {
49 char c = n[i];
50 if (!c || c != first[i] || is_glob_special(c))
51 break;
52 if (c == '/')
53 len = i + 1;
55 if (first == n || len < max) {
56 max = len;
57 if (!max)
58 break;
61 return max;
65 * Returns a copy of the longest leading path common among all
66 * pathspecs.
68 char *common_prefix(const char **pathspec)
70 unsigned long len = common_prefix_len(pathspec);
72 return len ? xmemdupz(*pathspec, len) : NULL;
75 int fill_directory(struct dir_struct *dir, const char **pathspec)
77 const char *path;
78 size_t len;
81 * Calculate common prefix for the pathspec, and
82 * use that to optimize the directory walk
84 len = common_prefix_len(pathspec);
85 path = "";
87 if (len)
88 path = xmemdupz(*pathspec, len);
90 /* Read the directory and prune it */
91 read_directory(dir, path, len, pathspec);
92 if (*path)
93 free((char *)path);
94 return len;
97 int within_depth(const char *name, int namelen,
98 int depth, int max_depth)
100 const char *cp = name, *cpe = name + namelen;
102 while (cp < cpe) {
103 if (*cp++ != '/')
104 continue;
105 depth++;
106 if (depth > max_depth)
107 return 0;
109 return 1;
113 * Does 'match' match the given name?
114 * A match is found if
116 * (1) the 'match' string is leading directory of 'name', or
117 * (2) the 'match' string is a wildcard and matches 'name', or
118 * (3) the 'match' string is exactly the same as 'name'.
120 * and the return value tells which case it was.
122 * It returns 0 when there is no match.
124 static int match_one(const char *match, const char *name, int namelen)
126 int matchlen;
128 /* If the match was just the prefix, we matched */
129 if (!*match)
130 return MATCHED_RECURSIVELY;
132 if (ignore_case) {
133 for (;;) {
134 unsigned char c1 = tolower(*match);
135 unsigned char c2 = tolower(*name);
136 if (c1 == '\0' || is_glob_special(c1))
137 break;
138 if (c1 != c2)
139 return 0;
140 match++;
141 name++;
142 namelen--;
144 } else {
145 for (;;) {
146 unsigned char c1 = *match;
147 unsigned char c2 = *name;
148 if (c1 == '\0' || is_glob_special(c1))
149 break;
150 if (c1 != c2)
151 return 0;
152 match++;
153 name++;
154 namelen--;
160 * If we don't match the matchstring exactly,
161 * we need to match by fnmatch
163 matchlen = strlen(match);
164 if (strncmp_icase(match, name, matchlen))
165 return !fnmatch_icase(match, name, 0) ? MATCHED_FNMATCH : 0;
167 if (namelen == matchlen)
168 return MATCHED_EXACTLY;
169 if (match[matchlen-1] == '/' || name[matchlen] == '/')
170 return MATCHED_RECURSIVELY;
171 return 0;
175 * Given a name and a list of pathspecs, see if the name matches
176 * any of the pathspecs. The caller is also interested in seeing
177 * all pathspec matches some names it calls this function with
178 * (otherwise the user could have mistyped the unmatched pathspec),
179 * and a mark is left in seen[] array for pathspec element that
180 * actually matched anything.
182 int match_pathspec(const char **pathspec, const char *name, int namelen,
183 int prefix, char *seen)
185 int i, retval = 0;
187 if (!pathspec)
188 return 1;
190 name += prefix;
191 namelen -= prefix;
193 for (i = 0; pathspec[i] != NULL; i++) {
194 int how;
195 const char *match = pathspec[i] + prefix;
196 if (seen && seen[i] == MATCHED_EXACTLY)
197 continue;
198 how = match_one(match, name, namelen);
199 if (how) {
200 if (retval < how)
201 retval = how;
202 if (seen && seen[i] < how)
203 seen[i] = how;
206 return retval;
210 * Does 'match' match the given name?
211 * A match is found if
213 * (1) the 'match' string is leading directory of 'name', or
214 * (2) the 'match' string is a wildcard and matches 'name', or
215 * (3) the 'match' string is exactly the same as 'name'.
217 * and the return value tells which case it was.
219 * It returns 0 when there is no match.
221 static int match_pathspec_item(const struct pathspec_item *item, int prefix,
222 const char *name, int namelen)
224 /* name/namelen has prefix cut off by caller */
225 const char *match = item->match + prefix;
226 int matchlen = item->len - prefix;
228 /* If the match was just the prefix, we matched */
229 if (!*match)
230 return MATCHED_RECURSIVELY;
232 if (matchlen <= namelen && !strncmp(match, name, matchlen)) {
233 if (matchlen == namelen)
234 return MATCHED_EXACTLY;
236 if (match[matchlen-1] == '/' || name[matchlen] == '/')
237 return MATCHED_RECURSIVELY;
240 if (item->use_wildcard && !fnmatch(match, name, 0))
241 return MATCHED_FNMATCH;
243 return 0;
247 * Given a name and a list of pathspecs, see if the name matches
248 * any of the pathspecs. The caller is also interested in seeing
249 * all pathspec matches some names it calls this function with
250 * (otherwise the user could have mistyped the unmatched pathspec),
251 * and a mark is left in seen[] array for pathspec element that
252 * actually matched anything.
254 int match_pathspec_depth(const struct pathspec *ps,
255 const char *name, int namelen,
256 int prefix, char *seen)
258 int i, retval = 0;
260 if (!ps->nr) {
261 if (!ps->recursive || ps->max_depth == -1)
262 return MATCHED_RECURSIVELY;
264 if (within_depth(name, namelen, 0, ps->max_depth))
265 return MATCHED_EXACTLY;
266 else
267 return 0;
270 name += prefix;
271 namelen -= prefix;
273 for (i = ps->nr - 1; i >= 0; i--) {
274 int how;
275 if (seen && seen[i] == MATCHED_EXACTLY)
276 continue;
277 how = match_pathspec_item(ps->items+i, prefix, name, namelen);
278 if (ps->recursive && ps->max_depth != -1 &&
279 how && how != MATCHED_FNMATCH) {
280 int len = ps->items[i].len;
281 if (name[len] == '/')
282 len++;
283 if (within_depth(name+len, namelen-len, 0, ps->max_depth))
284 how = MATCHED_EXACTLY;
285 else
286 how = 0;
288 if (how) {
289 if (retval < how)
290 retval = how;
291 if (seen && seen[i] < how)
292 seen[i] = how;
295 return retval;
298 static int no_wildcard(const char *string)
300 return string[strcspn(string, "*?[{\\")] == '\0';
303 void add_exclude(const char *string, const char *base,
304 int baselen, struct exclude_list *which)
306 struct exclude *x;
307 size_t len;
308 int to_exclude = 1;
309 int flags = 0;
311 if (*string == '!') {
312 to_exclude = 0;
313 string++;
315 len = strlen(string);
316 if (len && string[len - 1] == '/') {
317 char *s;
318 x = xmalloc(sizeof(*x) + len);
319 s = (char *)(x+1);
320 memcpy(s, string, len - 1);
321 s[len - 1] = '\0';
322 string = s;
323 x->pattern = s;
324 flags = EXC_FLAG_MUSTBEDIR;
325 } else {
326 x = xmalloc(sizeof(*x));
327 x->pattern = string;
329 x->to_exclude = to_exclude;
330 x->patternlen = strlen(string);
331 x->base = base;
332 x->baselen = baselen;
333 x->flags = flags;
334 if (!strchr(string, '/'))
335 x->flags |= EXC_FLAG_NODIR;
336 if (no_wildcard(string))
337 x->flags |= EXC_FLAG_NOWILDCARD;
338 if (*string == '*' && no_wildcard(string+1))
339 x->flags |= EXC_FLAG_ENDSWITH;
340 ALLOC_GROW(which->excludes, which->nr + 1, which->alloc);
341 which->excludes[which->nr++] = x;
344 static void *read_skip_worktree_file_from_index(const char *path, size_t *size)
346 int pos, len;
347 unsigned long sz;
348 enum object_type type;
349 void *data;
350 struct index_state *istate = &the_index;
352 len = strlen(path);
353 pos = index_name_pos(istate, path, len);
354 if (pos < 0)
355 return NULL;
356 if (!ce_skip_worktree(istate->cache[pos]))
357 return NULL;
358 data = read_sha1_file(istate->cache[pos]->sha1, &type, &sz);
359 if (!data || type != OBJ_BLOB) {
360 free(data);
361 return NULL;
363 *size = xsize_t(sz);
364 return data;
367 void free_excludes(struct exclude_list *el)
369 int i;
371 for (i = 0; i < el->nr; i++)
372 free(el->excludes[i]);
373 free(el->excludes);
375 el->nr = 0;
376 el->excludes = NULL;
379 int add_excludes_from_file_to_list(const char *fname,
380 const char *base,
381 int baselen,
382 char **buf_p,
383 struct exclude_list *which,
384 int check_index)
386 struct stat st;
387 int fd, i;
388 size_t size = 0;
389 char *buf, *entry;
391 fd = open(fname, O_RDONLY);
392 if (fd < 0 || fstat(fd, &st) < 0) {
393 if (0 <= fd)
394 close(fd);
395 if (!check_index ||
396 (buf = read_skip_worktree_file_from_index(fname, &size)) == NULL)
397 return -1;
398 if (size == 0) {
399 free(buf);
400 return 0;
402 if (buf[size-1] != '\n') {
403 buf = xrealloc(buf, size+1);
404 buf[size++] = '\n';
407 else {
408 size = xsize_t(st.st_size);
409 if (size == 0) {
410 close(fd);
411 return 0;
413 buf = xmalloc(size+1);
414 if (read_in_full(fd, buf, size) != size) {
415 free(buf);
416 close(fd);
417 return -1;
419 buf[size++] = '\n';
420 close(fd);
423 if (buf_p)
424 *buf_p = buf;
425 entry = buf;
426 for (i = 0; i < size; i++) {
427 if (buf[i] == '\n') {
428 if (entry != buf + i && entry[0] != '#') {
429 buf[i - (i && buf[i-1] == '\r')] = 0;
430 add_exclude(entry, base, baselen, which);
432 entry = buf + i + 1;
435 return 0;
438 void add_excludes_from_file(struct dir_struct *dir, const char *fname)
440 if (add_excludes_from_file_to_list(fname, "", 0, NULL,
441 &dir->exclude_list[EXC_FILE], 0) < 0)
442 die("cannot use %s as an exclude file", fname);
445 static void prep_exclude(struct dir_struct *dir, const char *base, int baselen)
447 struct exclude_list *el;
448 struct exclude_stack *stk = NULL;
449 int current;
451 if ((!dir->exclude_per_dir) ||
452 (baselen + strlen(dir->exclude_per_dir) >= PATH_MAX))
453 return; /* too long a path -- ignore */
455 /* Pop the ones that are not the prefix of the path being checked. */
456 el = &dir->exclude_list[EXC_DIRS];
457 while ((stk = dir->exclude_stack) != NULL) {
458 if (stk->baselen <= baselen &&
459 !strncmp(dir->basebuf, base, stk->baselen))
460 break;
461 dir->exclude_stack = stk->prev;
462 while (stk->exclude_ix < el->nr)
463 free(el->excludes[--el->nr]);
464 free(stk->filebuf);
465 free(stk);
468 /* Read from the parent directories and push them down. */
469 current = stk ? stk->baselen : -1;
470 while (current < baselen) {
471 struct exclude_stack *stk = xcalloc(1, sizeof(*stk));
472 const char *cp;
474 if (current < 0) {
475 cp = base;
476 current = 0;
478 else {
479 cp = strchr(base + current + 1, '/');
480 if (!cp)
481 die("oops in prep_exclude");
482 cp++;
484 stk->prev = dir->exclude_stack;
485 stk->baselen = cp - base;
486 stk->exclude_ix = el->nr;
487 memcpy(dir->basebuf + current, base + current,
488 stk->baselen - current);
489 strcpy(dir->basebuf + stk->baselen, dir->exclude_per_dir);
490 add_excludes_from_file_to_list(dir->basebuf,
491 dir->basebuf, stk->baselen,
492 &stk->filebuf, el, 1);
493 dir->exclude_stack = stk;
494 current = stk->baselen;
496 dir->basebuf[baselen] = '\0';
499 /* Scan the list and let the last match determine the fate.
500 * Return 1 for exclude, 0 for include and -1 for undecided.
502 int excluded_from_list(const char *pathname,
503 int pathlen, const char *basename, int *dtype,
504 struct exclude_list *el)
506 int i;
508 if (el->nr) {
509 for (i = el->nr - 1; 0 <= i; i--) {
510 struct exclude *x = el->excludes[i];
511 const char *exclude = x->pattern;
512 int to_exclude = x->to_exclude;
514 if (x->flags & EXC_FLAG_MUSTBEDIR) {
515 if (*dtype == DT_UNKNOWN)
516 *dtype = get_dtype(NULL, pathname, pathlen);
517 if (*dtype != DT_DIR)
518 continue;
521 if (x->flags & EXC_FLAG_NODIR) {
522 /* match basename */
523 if (x->flags & EXC_FLAG_NOWILDCARD) {
524 if (!strcmp_icase(exclude, basename))
525 return to_exclude;
526 } else if (x->flags & EXC_FLAG_ENDSWITH) {
527 if (x->patternlen - 1 <= pathlen &&
528 !strcmp_icase(exclude + 1, pathname + pathlen - x->patternlen + 1))
529 return to_exclude;
530 } else {
531 if (fnmatch_icase(exclude, basename, 0) == 0)
532 return to_exclude;
535 else {
536 /* match with FNM_PATHNAME:
537 * exclude has base (baselen long) implicitly
538 * in front of it.
540 int baselen = x->baselen;
541 if (*exclude == '/')
542 exclude++;
544 if (pathlen < baselen ||
545 (baselen && pathname[baselen-1] != '/') ||
546 strncmp_icase(pathname, x->base, baselen))
547 continue;
549 if (x->flags & EXC_FLAG_NOWILDCARD) {
550 if (!strcmp_icase(exclude, pathname + baselen))
551 return to_exclude;
552 } else {
553 if (fnmatch_icase(exclude, pathname+baselen,
554 FNM_PATHNAME) == 0)
555 return to_exclude;
560 return -1; /* undecided */
563 static int excluded(struct dir_struct *dir, const char *pathname, int *dtype_p)
565 int pathlen = strlen(pathname);
566 int st;
567 const char *basename = strrchr(pathname, '/');
568 basename = (basename) ? basename+1 : pathname;
570 prep_exclude(dir, pathname, basename-pathname);
571 for (st = EXC_CMDL; st <= EXC_FILE; st++) {
572 switch (excluded_from_list(pathname, pathlen, basename,
573 dtype_p, &dir->exclude_list[st])) {
574 case 0:
575 return 0;
576 case 1:
577 return 1;
580 return 0;
583 void path_exclude_check_init(struct path_exclude_check *check,
584 struct dir_struct *dir)
586 check->dir = dir;
587 strbuf_init(&check->path, 256);
590 void path_exclude_check_clear(struct path_exclude_check *check)
592 strbuf_release(&check->path);
596 * Is this name excluded? This is for a caller like show_files() that
597 * do not honor directory hierarchy and iterate through paths that are
598 * possibly in an ignored directory.
600 * A path to a directory known to be excluded is left in check->path to
601 * optimize for repeated checks for files in the same excluded directory.
603 int path_excluded(struct path_exclude_check *check,
604 const char *name, int namelen, int *dtype)
606 int i;
607 struct strbuf *path = &check->path;
610 * we allow the caller to pass namelen as an optimization; it
611 * must match the length of the name, as we eventually call
612 * excluded() on the whole name string.
614 if (namelen < 0)
615 namelen = strlen(name);
617 if (path->len &&
618 path->len <= namelen &&
619 !memcmp(name, path->buf, path->len) &&
620 (!name[path->len] || name[path->len] == '/'))
621 return 1;
623 strbuf_setlen(path, 0);
624 for (i = 0; name[i]; i++) {
625 int ch = name[i];
627 if (ch == '/') {
628 int dt = DT_DIR;
629 if (excluded(check->dir, path->buf, &dt))
630 return 1;
632 strbuf_addch(path, ch);
635 /* An entry in the index; cannot be a directory with subentries */
636 strbuf_setlen(path, 0);
638 return excluded(check->dir, name, dtype);
641 static struct dir_entry *dir_entry_new(const char *pathname, int len)
643 struct dir_entry *ent;
645 ent = xmalloc(sizeof(*ent) + len + 1);
646 ent->len = len;
647 memcpy(ent->name, pathname, len);
648 ent->name[len] = 0;
649 return ent;
652 static struct dir_entry *dir_add_name(struct dir_struct *dir, const char *pathname, int len)
654 if (cache_name_exists(pathname, len, ignore_case))
655 return NULL;
657 ALLOC_GROW(dir->entries, dir->nr+1, dir->alloc);
658 return dir->entries[dir->nr++] = dir_entry_new(pathname, len);
661 struct dir_entry *dir_add_ignored(struct dir_struct *dir, const char *pathname, int len)
663 if (!cache_name_is_other(pathname, len))
664 return NULL;
666 ALLOC_GROW(dir->ignored, dir->ignored_nr+1, dir->ignored_alloc);
667 return dir->ignored[dir->ignored_nr++] = dir_entry_new(pathname, len);
670 enum exist_status {
671 index_nonexistent = 0,
672 index_directory,
673 index_gitdir
677 * Do not use the alphabetically stored index to look up
678 * the directory name; instead, use the case insensitive
679 * name hash.
681 static enum exist_status directory_exists_in_index_icase(const char *dirname, int len)
683 struct cache_entry *ce = index_name_exists(&the_index, dirname, len + 1, ignore_case);
684 unsigned char endchar;
686 if (!ce)
687 return index_nonexistent;
688 endchar = ce->name[len];
691 * The cache_entry structure returned will contain this dirname
692 * and possibly additional path components.
694 if (endchar == '/')
695 return index_directory;
698 * If there are no additional path components, then this cache_entry
699 * represents a submodule. Submodules, despite being directories,
700 * are stored in the cache without a closing slash.
702 if (!endchar && S_ISGITLINK(ce->ce_mode))
703 return index_gitdir;
705 /* This should never be hit, but it exists just in case. */
706 return index_nonexistent;
710 * The index sorts alphabetically by entry name, which
711 * means that a gitlink sorts as '\0' at the end, while
712 * a directory (which is defined not as an entry, but as
713 * the files it contains) will sort with the '/' at the
714 * end.
716 static enum exist_status directory_exists_in_index(const char *dirname, int len)
718 int pos;
720 if (ignore_case)
721 return directory_exists_in_index_icase(dirname, len);
723 pos = cache_name_pos(dirname, len);
724 if (pos < 0)
725 pos = -pos-1;
726 while (pos < active_nr) {
727 struct cache_entry *ce = active_cache[pos++];
728 unsigned char endchar;
730 if (strncmp(ce->name, dirname, len))
731 break;
732 endchar = ce->name[len];
733 if (endchar > '/')
734 break;
735 if (endchar == '/')
736 return index_directory;
737 if (!endchar && S_ISGITLINK(ce->ce_mode))
738 return index_gitdir;
740 return index_nonexistent;
744 * When we find a directory when traversing the filesystem, we
745 * have three distinct cases:
747 * - ignore it
748 * - see it as a directory
749 * - recurse into it
751 * and which one we choose depends on a combination of existing
752 * git index contents and the flags passed into the directory
753 * traversal routine.
755 * Case 1: If we *already* have entries in the index under that
756 * directory name, we always recurse into the directory to see
757 * all the files.
759 * Case 2: If we *already* have that directory name as a gitlink,
760 * we always continue to see it as a gitlink, regardless of whether
761 * there is an actual git directory there or not (it might not
762 * be checked out as a subproject!)
764 * Case 3: if we didn't have it in the index previously, we
765 * have a few sub-cases:
767 * (a) if "show_other_directories" is true, we show it as
768 * just a directory, unless "hide_empty_directories" is
769 * also true and the directory is empty, in which case
770 * we just ignore it entirely.
771 * (b) if it looks like a git directory, and we don't have
772 * 'no_gitlinks' set we treat it as a gitlink, and show it
773 * as a directory.
774 * (c) otherwise, we recurse into it.
776 enum directory_treatment {
777 show_directory,
778 ignore_directory,
779 recurse_into_directory
782 static enum directory_treatment treat_directory(struct dir_struct *dir,
783 const char *dirname, int len,
784 const struct path_simplify *simplify)
786 /* The "len-1" is to strip the final '/' */
787 switch (directory_exists_in_index(dirname, len-1)) {
788 case index_directory:
789 return recurse_into_directory;
791 case index_gitdir:
792 if (dir->flags & DIR_SHOW_OTHER_DIRECTORIES)
793 return ignore_directory;
794 return show_directory;
796 case index_nonexistent:
797 if (dir->flags & DIR_SHOW_OTHER_DIRECTORIES)
798 break;
799 if (!(dir->flags & DIR_NO_GITLINKS)) {
800 unsigned char sha1[20];
801 if (resolve_gitlink_ref(dirname, "HEAD", sha1) == 0)
802 return show_directory;
804 return recurse_into_directory;
807 /* This is the "show_other_directories" case */
808 if (!(dir->flags & DIR_HIDE_EMPTY_DIRECTORIES))
809 return show_directory;
810 if (!read_directory_recursive(dir, dirname, len, 1, simplify))
811 return ignore_directory;
812 return show_directory;
816 * This is an inexact early pruning of any recursive directory
817 * reading - if the path cannot possibly be in the pathspec,
818 * return true, and we'll skip it early.
820 static int simplify_away(const char *path, int pathlen, const struct path_simplify *simplify)
822 if (simplify) {
823 for (;;) {
824 const char *match = simplify->path;
825 int len = simplify->len;
827 if (!match)
828 break;
829 if (len > pathlen)
830 len = pathlen;
831 if (!memcmp(path, match, len))
832 return 0;
833 simplify++;
835 return 1;
837 return 0;
841 * This function tells us whether an excluded path matches a
842 * list of "interesting" pathspecs. That is, whether a path matched
843 * by any of the pathspecs could possibly be ignored by excluding
844 * the specified path. This can happen if:
846 * 1. the path is mentioned explicitly in the pathspec
848 * 2. the path is a directory prefix of some element in the
849 * pathspec
851 static int exclude_matches_pathspec(const char *path, int len,
852 const struct path_simplify *simplify)
854 if (simplify) {
855 for (; simplify->path; simplify++) {
856 if (len == simplify->len
857 && !memcmp(path, simplify->path, len))
858 return 1;
859 if (len < simplify->len
860 && simplify->path[len] == '/'
861 && !memcmp(path, simplify->path, len))
862 return 1;
865 return 0;
868 static int get_index_dtype(const char *path, int len)
870 int pos;
871 struct cache_entry *ce;
873 ce = cache_name_exists(path, len, 0);
874 if (ce) {
875 if (!ce_uptodate(ce))
876 return DT_UNKNOWN;
877 if (S_ISGITLINK(ce->ce_mode))
878 return DT_DIR;
880 * Nobody actually cares about the
881 * difference between DT_LNK and DT_REG
883 return DT_REG;
886 /* Try to look it up as a directory */
887 pos = cache_name_pos(path, len);
888 if (pos >= 0)
889 return DT_UNKNOWN;
890 pos = -pos-1;
891 while (pos < active_nr) {
892 ce = active_cache[pos++];
893 if (strncmp(ce->name, path, len))
894 break;
895 if (ce->name[len] > '/')
896 break;
897 if (ce->name[len] < '/')
898 continue;
899 if (!ce_uptodate(ce))
900 break; /* continue? */
901 return DT_DIR;
903 return DT_UNKNOWN;
906 static int get_dtype(struct dirent *de, const char *path, int len)
908 int dtype = de ? DTYPE(de) : DT_UNKNOWN;
909 struct stat st;
911 if (dtype != DT_UNKNOWN)
912 return dtype;
913 dtype = get_index_dtype(path, len);
914 if (dtype != DT_UNKNOWN)
915 return dtype;
916 if (lstat(path, &st))
917 return dtype;
918 if (S_ISREG(st.st_mode))
919 return DT_REG;
920 if (S_ISDIR(st.st_mode))
921 return DT_DIR;
922 if (S_ISLNK(st.st_mode))
923 return DT_LNK;
924 return dtype;
927 enum path_treatment {
928 path_ignored,
929 path_handled,
930 path_recurse
933 static enum path_treatment treat_one_path(struct dir_struct *dir,
934 char *path, int *len,
935 const struct path_simplify *simplify,
936 int dtype, struct dirent *de)
938 int exclude = excluded(dir, path, &dtype);
939 if (exclude && (dir->flags & DIR_COLLECT_IGNORED)
940 && exclude_matches_pathspec(path, *len, simplify))
941 dir_add_ignored(dir, path, *len);
944 * Excluded? If we don't explicitly want to show
945 * ignored files, ignore it
947 if (exclude && !(dir->flags & DIR_SHOW_IGNORED))
948 return path_ignored;
950 if (dtype == DT_UNKNOWN)
951 dtype = get_dtype(de, path, *len);
954 * Do we want to see just the ignored files?
955 * We still need to recurse into directories,
956 * even if we don't ignore them, since the
957 * directory may contain files that we do..
959 if (!exclude && (dir->flags & DIR_SHOW_IGNORED)) {
960 if (dtype != DT_DIR)
961 return path_ignored;
964 switch (dtype) {
965 default:
966 return path_ignored;
967 case DT_DIR:
968 memcpy(path + *len, "/", 2);
969 (*len)++;
970 switch (treat_directory(dir, path, *len, simplify)) {
971 case show_directory:
972 if (exclude != !!(dir->flags
973 & DIR_SHOW_IGNORED))
974 return path_ignored;
975 break;
976 case recurse_into_directory:
977 return path_recurse;
978 case ignore_directory:
979 return path_ignored;
981 break;
982 case DT_REG:
983 case DT_LNK:
984 break;
986 return path_handled;
989 static enum path_treatment treat_path(struct dir_struct *dir,
990 struct dirent *de,
991 char *path, int path_max,
992 int baselen,
993 const struct path_simplify *simplify,
994 int *len)
996 int dtype;
998 if (is_dot_or_dotdot(de->d_name) || !strcmp(de->d_name, ".git"))
999 return path_ignored;
1000 *len = strlen(de->d_name);
1001 /* Ignore overly long pathnames! */
1002 if (*len + baselen + 8 > path_max)
1003 return path_ignored;
1004 memcpy(path + baselen, de->d_name, *len + 1);
1005 *len += baselen;
1006 if (simplify_away(path, *len, simplify))
1007 return path_ignored;
1009 dtype = DTYPE(de);
1010 return treat_one_path(dir, path, len, simplify, dtype, de);
1014 * Read a directory tree. We currently ignore anything but
1015 * directories, regular files and symlinks. That's because git
1016 * doesn't handle them at all yet. Maybe that will change some
1017 * day.
1019 * Also, we ignore the name ".git" (even if it is not a directory).
1020 * That likely will not change.
1022 static int read_directory_recursive(struct dir_struct *dir,
1023 const char *base, int baselen,
1024 int check_only,
1025 const struct path_simplify *simplify)
1027 DIR *fdir = opendir(*base ? base : ".");
1028 int contents = 0;
1029 struct dirent *de;
1030 char path[PATH_MAX + 1];
1032 if (!fdir)
1033 return 0;
1035 memcpy(path, base, baselen);
1037 while ((de = readdir(fdir)) != NULL) {
1038 int len;
1039 switch (treat_path(dir, de, path, sizeof(path),
1040 baselen, simplify, &len)) {
1041 case path_recurse:
1042 contents += read_directory_recursive(dir, path, len, 0, simplify);
1043 continue;
1044 case path_ignored:
1045 continue;
1046 case path_handled:
1047 break;
1049 contents++;
1050 if (check_only)
1051 goto exit_early;
1052 else
1053 dir_add_name(dir, path, len);
1055 exit_early:
1056 closedir(fdir);
1058 return contents;
1061 static int cmp_name(const void *p1, const void *p2)
1063 const struct dir_entry *e1 = *(const struct dir_entry **)p1;
1064 const struct dir_entry *e2 = *(const struct dir_entry **)p2;
1066 return cache_name_compare(e1->name, e1->len,
1067 e2->name, e2->len);
1071 * Return the length of the "simple" part of a path match limiter.
1073 static int simple_length(const char *match)
1075 int len = -1;
1077 for (;;) {
1078 unsigned char c = *match++;
1079 len++;
1080 if (c == '\0' || is_glob_special(c))
1081 return len;
1085 static struct path_simplify *create_simplify(const char **pathspec)
1087 int nr, alloc = 0;
1088 struct path_simplify *simplify = NULL;
1090 if (!pathspec)
1091 return NULL;
1093 for (nr = 0 ; ; nr++) {
1094 const char *match;
1095 if (nr >= alloc) {
1096 alloc = alloc_nr(alloc);
1097 simplify = xrealloc(simplify, alloc * sizeof(*simplify));
1099 match = *pathspec++;
1100 if (!match)
1101 break;
1102 simplify[nr].path = match;
1103 simplify[nr].len = simple_length(match);
1105 simplify[nr].path = NULL;
1106 simplify[nr].len = 0;
1107 return simplify;
1110 static void free_simplify(struct path_simplify *simplify)
1112 free(simplify);
1115 static int treat_leading_path(struct dir_struct *dir,
1116 const char *path, int len,
1117 const struct path_simplify *simplify)
1119 char pathbuf[PATH_MAX];
1120 int baselen, blen;
1121 const char *cp;
1123 while (len && path[len - 1] == '/')
1124 len--;
1125 if (!len)
1126 return 1;
1127 baselen = 0;
1128 while (1) {
1129 cp = path + baselen + !!baselen;
1130 cp = memchr(cp, '/', path + len - cp);
1131 if (!cp)
1132 baselen = len;
1133 else
1134 baselen = cp - path;
1135 memcpy(pathbuf, path, baselen);
1136 pathbuf[baselen] = '\0';
1137 if (!is_directory(pathbuf))
1138 return 0;
1139 if (simplify_away(pathbuf, baselen, simplify))
1140 return 0;
1141 blen = baselen;
1142 if (treat_one_path(dir, pathbuf, &blen, simplify,
1143 DT_DIR, NULL) == path_ignored)
1144 return 0; /* do not recurse into it */
1145 if (len <= baselen)
1146 return 1; /* finished checking */
1150 int read_directory(struct dir_struct *dir, const char *path, int len, const char **pathspec)
1152 struct path_simplify *simplify;
1154 if (has_symlink_leading_path(path, len))
1155 return dir->nr;
1157 simplify = create_simplify(pathspec);
1158 if (!len || treat_leading_path(dir, path, len, simplify))
1159 read_directory_recursive(dir, path, len, 0, simplify);
1160 free_simplify(simplify);
1161 qsort(dir->entries, dir->nr, sizeof(struct dir_entry *), cmp_name);
1162 qsort(dir->ignored, dir->ignored_nr, sizeof(struct dir_entry *), cmp_name);
1163 return dir->nr;
1166 int file_exists(const char *f)
1168 struct stat sb;
1169 return lstat(f, &sb) == 0;
1173 * Given two normalized paths (a trailing slash is ok), if subdir is
1174 * outside dir, return -1. Otherwise return the offset in subdir that
1175 * can be used as relative path to dir.
1177 int dir_inside_of(const char *subdir, const char *dir)
1179 int offset = 0;
1181 assert(dir && subdir && *dir && *subdir);
1183 while (*dir && *subdir && *dir == *subdir) {
1184 dir++;
1185 subdir++;
1186 offset++;
1189 /* hel[p]/me vs hel[l]/yeah */
1190 if (*dir && *subdir)
1191 return -1;
1193 if (!*subdir)
1194 return !*dir ? offset : -1; /* same dir */
1196 /* foo/[b]ar vs foo/[] */
1197 if (is_dir_sep(dir[-1]))
1198 return is_dir_sep(subdir[-1]) ? offset : -1;
1200 /* foo[/]bar vs foo[] */
1201 return is_dir_sep(*subdir) ? offset + 1 : -1;
1204 int is_inside_dir(const char *dir)
1206 char cwd[PATH_MAX];
1207 if (!dir)
1208 return 0;
1209 if (!getcwd(cwd, sizeof(cwd)))
1210 die_errno("can't find the current directory");
1211 return dir_inside_of(cwd, dir) >= 0;
1214 int is_empty_dir(const char *path)
1216 DIR *dir = opendir(path);
1217 struct dirent *e;
1218 int ret = 1;
1220 if (!dir)
1221 return 0;
1223 while ((e = readdir(dir)) != NULL)
1224 if (!is_dot_or_dotdot(e->d_name)) {
1225 ret = 0;
1226 break;
1229 closedir(dir);
1230 return ret;
1233 int remove_dir_recursively(struct strbuf *path, int flag)
1235 DIR *dir;
1236 struct dirent *e;
1237 int ret = 0, original_len = path->len, len;
1238 int only_empty = (flag & REMOVE_DIR_EMPTY_ONLY);
1239 unsigned char submodule_head[20];
1241 if ((flag & REMOVE_DIR_KEEP_NESTED_GIT) &&
1242 !resolve_gitlink_ref(path->buf, "HEAD", submodule_head))
1243 /* Do not descend and nuke a nested git work tree. */
1244 return 0;
1246 dir = opendir(path->buf);
1247 if (!dir)
1248 return rmdir(path->buf);
1249 if (path->buf[original_len - 1] != '/')
1250 strbuf_addch(path, '/');
1252 len = path->len;
1253 while ((e = readdir(dir)) != NULL) {
1254 struct stat st;
1255 if (is_dot_or_dotdot(e->d_name))
1256 continue;
1258 strbuf_setlen(path, len);
1259 strbuf_addstr(path, e->d_name);
1260 if (lstat(path->buf, &st))
1261 ; /* fall thru */
1262 else if (S_ISDIR(st.st_mode)) {
1263 if (!remove_dir_recursively(path, only_empty))
1264 continue; /* happy */
1265 } else if (!only_empty && !unlink(path->buf))
1266 continue; /* happy, too */
1268 /* path too long, stat fails, or non-directory still exists */
1269 ret = -1;
1270 break;
1272 closedir(dir);
1274 strbuf_setlen(path, original_len);
1275 if (!ret)
1276 ret = rmdir(path->buf);
1277 return ret;
1280 void setup_standard_excludes(struct dir_struct *dir)
1282 const char *path;
1284 dir->exclude_per_dir = ".gitignore";
1285 path = git_path("info/exclude");
1286 if (!access(path, R_OK))
1287 add_excludes_from_file(dir, path);
1288 if (excludes_file && !access(excludes_file, R_OK))
1289 add_excludes_from_file(dir, excludes_file);
1292 int remove_path(const char *name)
1294 char *slash;
1296 if (unlink(name) && errno != ENOENT)
1297 return -1;
1299 slash = strrchr(name, '/');
1300 if (slash) {
1301 char *dirs = xstrdup(name);
1302 slash = dirs + (slash - name);
1303 do {
1304 *slash = '\0';
1305 } while (rmdir(dirs) == 0 && (slash = strrchr(dirs, '/')));
1306 free(dirs);
1308 return 0;
1311 static int pathspec_item_cmp(const void *a_, const void *b_)
1313 struct pathspec_item *a, *b;
1315 a = (struct pathspec_item *)a_;
1316 b = (struct pathspec_item *)b_;
1317 return strcmp(a->match, b->match);
1320 int init_pathspec(struct pathspec *pathspec, const char **paths)
1322 const char **p = paths;
1323 int i;
1325 memset(pathspec, 0, sizeof(*pathspec));
1326 if (!p)
1327 return 0;
1328 while (*p)
1329 p++;
1330 pathspec->raw = paths;
1331 pathspec->nr = p - paths;
1332 if (!pathspec->nr)
1333 return 0;
1335 pathspec->items = xmalloc(sizeof(struct pathspec_item)*pathspec->nr);
1336 for (i = 0; i < pathspec->nr; i++) {
1337 struct pathspec_item *item = pathspec->items+i;
1338 const char *path = paths[i];
1340 item->match = path;
1341 item->len = strlen(path);
1342 item->use_wildcard = !no_wildcard(path);
1343 if (item->use_wildcard)
1344 pathspec->has_wildcard = 1;
1347 qsort(pathspec->items, pathspec->nr,
1348 sizeof(struct pathspec_item), pathspec_item_cmp);
1350 return 0;
1353 void free_pathspec(struct pathspec *pathspec)
1355 free(pathspec->items);
1356 pathspec->items = NULL;