match_pathspec: match pathspec "foo/" against directory "foo"
[git/git.git] / dir.c
CommitLineData
453ec4bd
LT
1/*
2 * This handles recursive filename detection with exclude
3 * files, index knowledge etc..
4 *
95a68344
AS
5 * See Documentation/technical/api-directory-listing.txt
6 *
453ec4bd
LT
7 * Copyright (C) Linus Torvalds, 2005-2006
8 * Junio Hamano, 2005-2006
9 */
453ec4bd
LT
10#include "cache.h"
11#include "dir.h"
09595258 12#include "refs.h"
237ec6e4 13#include "wildmatch.h"
64acde94 14#include "pathspec.h"
453ec4bd 15
9fc42d60
LT
16struct path_simplify {
17 int len;
18 const char *path;
19};
20
defd7c7b
KB
21/*
22 * Tells read_directory_recursive how a file or directory should be treated.
23 * Values are ordered by significance, e.g. if a directory contains both
24 * excluded and untracked files, it is listed as untracked because
25 * path_untracked > path_excluded.
26 */
27enum path_treatment {
28 path_none = 0,
29 path_recurse,
30 path_excluded,
31 path_untracked
32};
33
34static enum path_treatment read_directory_recursive(struct dir_struct *dir,
35 const char *path, int len,
09595258 36 int check_only, const struct path_simplify *simplify);
caa6b782 37static int get_dtype(struct dirent *de, const char *path, int len);
09595258 38
8cf2a84e
JJ
39/* helper string functions with support for the ignore_case flag */
40int strcmp_icase(const char *a, const char *b)
41{
42 return ignore_case ? strcasecmp(a, b) : strcmp(a, b);
43}
44
45int strncmp_icase(const char *a, const char *b, size_t count)
46{
47 return ignore_case ? strncasecmp(a, b, count) : strncmp(a, b, count);
48}
49
50int fnmatch_icase(const char *pattern, const char *string, int flags)
51{
52 return fnmatch(pattern, string, flags | (ignore_case ? FNM_CASEFOLD : 0));
53}
54
bd30c2e4
NTND
55inline int git_fnmatch(const struct pathspec_item *item,
56 const char *pattern, const char *string,
57 int prefix)
5d74762d 58{
5d74762d 59 if (prefix > 0) {
93d93537 60 if (ps_strncmp(item, pattern, string, prefix))
5d74762d
NTND
61 return FNM_NOMATCH;
62 pattern += prefix;
63 string += prefix;
64 }
bd30c2e4 65 if (item->flags & PATHSPEC_ONESTAR) {
8c6abbcd
NTND
66 int pattern_len = strlen(++pattern);
67 int string_len = strlen(string);
68 return string_len < pattern_len ||
93d93537
NTND
69 ps_strcmp(item, pattern,
70 string + string_len - pattern_len);
8c6abbcd 71 }
bd30c2e4 72 if (item->magic & PATHSPEC_GLOB)
93d93537
NTND
73 return wildmatch(pattern, string,
74 WM_PATHNAME |
75 (item->magic & PATHSPEC_ICASE ? WM_CASEFOLD : 0),
76 NULL);
bd30c2e4
NTND
77 else
78 /* wildmatch has not learned no FNM_PATHNAME mode yet */
93d93537
NTND
79 return fnmatch(pattern, string,
80 item->magic & PATHSPEC_ICASE ? FNM_CASEFOLD : 0);
5d74762d
NTND
81}
82
0b6e56df
JH
83static int fnmatch_icase_mem(const char *pattern, int patternlen,
84 const char *string, int stringlen,
85 int flags)
86{
87 int match_status;
88 struct strbuf pat_buf = STRBUF_INIT;
89 struct strbuf str_buf = STRBUF_INIT;
90 const char *use_pat = pattern;
91 const char *use_str = string;
92
93 if (pattern[patternlen]) {
94 strbuf_add(&pat_buf, pattern, patternlen);
95 use_pat = pat_buf.buf;
96 }
97 if (string[stringlen]) {
98 strbuf_add(&str_buf, string, stringlen);
99 use_str = str_buf.buf;
100 }
101
f30366b2
JH
102 if (ignore_case)
103 flags |= WM_CASEFOLD;
104 match_status = wildmatch(use_pat, use_str, flags, NULL);
0b6e56df
JH
105
106 strbuf_release(&pat_buf);
107 strbuf_release(&str_buf);
108
109 return match_status;
110}
111
827f4d6c 112static size_t common_prefix_len(const struct pathspec *pathspec)
3c6a370b 113{
827f4d6c 114 int n;
4a085b16 115 size_t max = 0;
3c6a370b 116
93d93537
NTND
117 /*
118 * ":(icase)path" is treated as a pathspec full of
119 * wildcard. In other words, only prefix is considered common
120 * prefix. If the pathspec is abc/foo abc/bar, running in
121 * subdir xyz, the common prefix is still xyz, not xuz/abc as
122 * in non-:(icase).
123 */
5c6933d2
NTND
124 GUARD_PATHSPEC(pathspec,
125 PATHSPEC_FROMTOP |
126 PATHSPEC_MAXDEPTH |
bd30c2e4 127 PATHSPEC_LITERAL |
93d93537 128 PATHSPEC_GLOB |
ef79b1f8
NTND
129 PATHSPEC_ICASE |
130 PATHSPEC_EXCLUDE);
4a085b16 131
827f4d6c 132 for (n = 0; n < pathspec->nr; n++) {
93d93537 133 size_t i = 0, len = 0, item_len;
ef79b1f8
NTND
134 if (pathspec->items[n].magic & PATHSPEC_EXCLUDE)
135 continue;
93d93537
NTND
136 if (pathspec->items[n].magic & PATHSPEC_ICASE)
137 item_len = pathspec->items[n].prefix;
138 else
139 item_len = pathspec->items[n].nowildcard_len;
140 while (i < item_len && (n == 0 || i < max)) {
827f4d6c
NTND
141 char c = pathspec->items[n].match[i];
142 if (c != pathspec->items[0].match[i])
4a085b16
JH
143 break;
144 if (c == '/')
145 len = i + 1;
827f4d6c 146 i++;
4a085b16 147 }
827f4d6c 148 if (n == 0 || len < max) {
4a085b16
JH
149 max = len;
150 if (!max)
151 break;
152 }
3c6a370b 153 }
4a085b16 154 return max;
3c6a370b
LT
155}
156
f950eb95
CB
157/*
158 * Returns a copy of the longest leading path common among all
159 * pathspecs.
160 */
827f4d6c 161char *common_prefix(const struct pathspec *pathspec)
f950eb95
CB
162{
163 unsigned long len = common_prefix_len(pathspec);
164
827f4d6c 165 return len ? xmemdupz(pathspec->items[0].match, len) : NULL;
f950eb95
CB
166}
167
7327d3d1 168int fill_directory(struct dir_struct *dir, const struct pathspec *pathspec)
1d8842d9 169{
4a085b16 170 size_t len;
1d8842d9
LT
171
172 /*
173 * Calculate common prefix for the pathspec, and
174 * use that to optimize the directory walk
175 */
4a085b16 176 len = common_prefix_len(pathspec);
1d8842d9
LT
177
178 /* Read the directory and prune it */
b3920bbd 179 read_directory(dir, pathspec->nr ? pathspec->_raw[0] : "", len, pathspec);
dba2e203 180 return len;
1d8842d9
LT
181}
182
bc96cc87
NTND
183int within_depth(const char *name, int namelen,
184 int depth, int max_depth)
185{
186 const char *cp = name, *cpe = name + namelen;
187
188 while (cp < cpe) {
189 if (*cp++ != '/')
190 continue;
191 depth++;
192 if (depth > max_depth)
193 return 0;
194 }
195 return 1;
196}
197
42b0874a 198#define DO_MATCH_EXCLUDE 1
68690fdd 199#define DO_MATCH_DIRECTORY 2
42b0874a 200
61cf2820
NTND
201/*
202 * Does 'match' match the given name?
203 * A match is found if
204 *
205 * (1) the 'match' string is leading directory of 'name', or
206 * (2) the 'match' string is a wildcard and matches 'name', or
207 * (3) the 'match' string is exactly the same as 'name'.
208 *
209 * and the return value tells which case it was.
210 *
211 * It returns 0 when there is no match.
212 */
213static int match_pathspec_item(const struct pathspec_item *item, int prefix,
42b0874a 214 const char *name, int namelen, unsigned flags)
61cf2820
NTND
215{
216 /* name/namelen has prefix cut off by caller */
217 const char *match = item->match + prefix;
218 int matchlen = item->len - prefix;
219
93d93537
NTND
220 /*
221 * The normal call pattern is:
222 * 1. prefix = common_prefix_len(ps);
223 * 2. prune something, or fill_directory
854b0959 224 * 3. match_pathspec()
93d93537
NTND
225 *
226 * 'prefix' at #1 may be shorter than the command's prefix and
227 * it's ok for #2 to match extra files. Those extras will be
228 * trimmed at #3.
229 *
230 * Suppose the pathspec is 'foo' and '../bar' running from
231 * subdir 'xyz'. The common prefix at #1 will be empty, thanks
232 * to "../". We may have xyz/foo _and_ XYZ/foo after #2. The
233 * user does not want XYZ/foo, only the "foo" part should be
234 * case-insensitive. We need to filter out XYZ/foo here. In
235 * other words, we do not trust the caller on comparing the
236 * prefix part when :(icase) is involved. We do exact
237 * comparison ourselves.
238 *
239 * Normally the caller (common_prefix_len() in fact) does
240 * _exact_ matching on name[-prefix+1..-1] and we do not need
241 * to check that part. Be defensive and check it anyway, in
242 * case common_prefix_len is changed, or a new caller is
243 * introduced that does not use common_prefix_len.
244 *
245 * If the penalty turns out too high when prefix is really
246 * long, maybe change it to
247 * strncmp(match, name, item->prefix - prefix)
248 */
249 if (item->prefix && (item->magic & PATHSPEC_ICASE) &&
250 strncmp(item->match, name - prefix, item->prefix))
251 return 0;
252
61cf2820
NTND
253 /* If the match was just the prefix, we matched */
254 if (!*match)
255 return MATCHED_RECURSIVELY;
256
93d93537 257 if (matchlen <= namelen && !ps_strncmp(item, match, name, matchlen)) {
61cf2820
NTND
258 if (matchlen == namelen)
259 return MATCHED_EXACTLY;
260
261 if (match[matchlen-1] == '/' || name[matchlen] == '/')
262 return MATCHED_RECURSIVELY;
68690fdd
NTND
263 } else if ((flags & DO_MATCH_DIRECTORY) &&
264 match[matchlen - 1] == '/' &&
265 namelen == matchlen - 1 &&
266 !ps_strncmp(item, match, name, namelen))
267 return MATCHED_EXACTLY;
61cf2820 268
5d74762d 269 if (item->nowildcard_len < item->len &&
bd30c2e4 270 !git_fnmatch(item, match, name,
8c6abbcd 271 item->nowildcard_len - prefix))
61cf2820
NTND
272 return MATCHED_FNMATCH;
273
274 return 0;
275}
276
277/*
52ed1894
AS
278 * Given a name and a list of pathspecs, returns the nature of the
279 * closest (i.e. most specific) match of the name to any of the
280 * pathspecs.
281 *
282 * The caller typically calls this multiple times with the same
283 * pathspec and seen[] array but with different name/namelen
284 * (e.g. entries from the index) and is interested in seeing if and
285 * how each pathspec matches all the names it calls this function
286 * with. A mark is left in the seen[] array for each pathspec element
287 * indicating the closest type of match that element achieved, so if
288 * seen[n] remains zero after multiple invocations, that means the nth
289 * pathspec did not match any names, which could indicate that the
290 * user mistyped the nth pathspec.
61cf2820 291 */
854b0959
NTND
292static int do_match_pathspec(const struct pathspec *ps,
293 const char *name, int namelen,
294 int prefix, char *seen,
42b0874a 295 unsigned flags)
61cf2820 296{
42b0874a 297 int i, retval = 0, exclude = flags & DO_MATCH_EXCLUDE;
61cf2820 298
5c6933d2
NTND
299 GUARD_PATHSPEC(ps,
300 PATHSPEC_FROMTOP |
301 PATHSPEC_MAXDEPTH |
bd30c2e4 302 PATHSPEC_LITERAL |
93d93537 303 PATHSPEC_GLOB |
ef79b1f8
NTND
304 PATHSPEC_ICASE |
305 PATHSPEC_EXCLUDE);
8f4f8f45 306
61cf2820 307 if (!ps->nr) {
6330a171
NTND
308 if (!ps->recursive ||
309 !(ps->magic & PATHSPEC_MAXDEPTH) ||
310 ps->max_depth == -1)
61cf2820
NTND
311 return MATCHED_RECURSIVELY;
312
313 if (within_depth(name, namelen, 0, ps->max_depth))
314 return MATCHED_EXACTLY;
315 else
316 return 0;
317 }
318
319 name += prefix;
320 namelen -= prefix;
321
322 for (i = ps->nr - 1; i >= 0; i--) {
323 int how;
ef79b1f8
NTND
324
325 if ((!exclude && ps->items[i].magic & PATHSPEC_EXCLUDE) ||
326 ( exclude && !(ps->items[i].magic & PATHSPEC_EXCLUDE)))
327 continue;
328
61cf2820
NTND
329 if (seen && seen[i] == MATCHED_EXACTLY)
330 continue;
ef79b1f8
NTND
331 /*
332 * Make exclude patterns optional and never report
333 * "pathspec ':(exclude)foo' matches no files"
334 */
335 if (seen && ps->items[i].magic & PATHSPEC_EXCLUDE)
336 seen[i] = MATCHED_FNMATCH;
42b0874a
NTND
337 how = match_pathspec_item(ps->items+i, prefix, name,
338 namelen, flags);
6330a171
NTND
339 if (ps->recursive &&
340 (ps->magic & PATHSPEC_MAXDEPTH) &&
341 ps->max_depth != -1 &&
61cf2820
NTND
342 how && how != MATCHED_FNMATCH) {
343 int len = ps->items[i].len;
344 if (name[len] == '/')
345 len++;
346 if (within_depth(name+len, namelen-len, 0, ps->max_depth))
347 how = MATCHED_EXACTLY;
348 else
349 how = 0;
350 }
351 if (how) {
352 if (retval < how)
353 retval = how;
354 if (seen && seen[i] < how)
355 seen[i] = how;
356 }
357 }
358 return retval;
359}
360
854b0959
NTND
361int match_pathspec(const struct pathspec *ps,
362 const char *name, int namelen,
363 int prefix, char *seen)
ef79b1f8
NTND
364{
365 int positive, negative;
42b0874a
NTND
366 unsigned flags = 0;
367 positive = do_match_pathspec(ps, name, namelen,
368 prefix, seen, flags);
ef79b1f8
NTND
369 if (!(ps->magic & PATHSPEC_EXCLUDE) || !positive)
370 return positive;
42b0874a
NTND
371 negative = do_match_pathspec(ps, name, namelen,
372 prefix, seen,
373 flags | DO_MATCH_EXCLUDE);
ef79b1f8
NTND
374 return negative ? 0 : positive;
375}
376
fcd631ed
NTND
377/*
378 * Return the length of the "simple" part of a path match limiter.
379 */
87323bda 380int simple_length(const char *match)
fcd631ed
NTND
381{
382 int len = -1;
383
384 for (;;) {
385 unsigned char c = *match++;
386 len++;
387 if (c == '\0' || is_glob_special(c))
388 return len;
389 }
390}
391
87323bda 392int no_wildcard(const char *string)
68492fc7 393{
fcd631ed 394 return string[simple_length(string)] == '\0';
68492fc7
LK
395}
396
82dce998
NTND
397void parse_exclude_pattern(const char **pattern,
398 int *patternlen,
399 int *flags,
400 int *nowildcardlen)
84460eec
NTND
401{
402 const char *p = *pattern;
403 size_t i, len;
404
405 *flags = 0;
406 if (*p == '!') {
407 *flags |= EXC_FLAG_NEGATIVE;
408 p++;
409 }
410 len = strlen(p);
411 if (len && p[len - 1] == '/') {
412 len--;
413 *flags |= EXC_FLAG_MUSTBEDIR;
414 }
415 for (i = 0; i < len; i++) {
416 if (p[i] == '/')
417 break;
418 }
419 if (i == len)
420 *flags |= EXC_FLAG_NODIR;
421 *nowildcardlen = simple_length(p);
422 /*
423 * we should have excluded the trailing slash from 'p' too,
424 * but that's one more allocation. Instead just make sure
425 * nowildcardlen does not exceed real patternlen
426 */
427 if (*nowildcardlen > len)
428 *nowildcardlen = len;
429 if (*p == '*' && no_wildcard(p + 1))
430 *flags |= EXC_FLAG_ENDSWITH;
431 *pattern = p;
432 *patternlen = len;
433}
434
453ec4bd 435void add_exclude(const char *string, const char *base,
c04318e4 436 int baselen, struct exclude_list *el, int srcpos)
453ec4bd 437{
d6b8fc30 438 struct exclude *x;
84460eec
NTND
439 int patternlen;
440 int flags;
441 int nowildcardlen;
453ec4bd 442
84460eec
NTND
443 parse_exclude_pattern(&string, &patternlen, &flags, &nowildcardlen);
444 if (flags & EXC_FLAG_MUSTBEDIR) {
d6b8fc30 445 char *s;
84460eec 446 x = xmalloc(sizeof(*x) + patternlen + 1);
4b25d091 447 s = (char *)(x+1);
84460eec
NTND
448 memcpy(s, string, patternlen);
449 s[patternlen] = '\0';
d6b8fc30 450 x->pattern = s;
d6b8fc30
JH
451 } else {
452 x = xmalloc(sizeof(*x));
453 x->pattern = string;
454 }
84460eec
NTND
455 x->patternlen = patternlen;
456 x->nowildcardlen = nowildcardlen;
453ec4bd
LT
457 x->base = base;
458 x->baselen = baselen;
d6b8fc30 459 x->flags = flags;
c04318e4 460 x->srcpos = srcpos;
840fc334
AS
461 ALLOC_GROW(el->excludes, el->nr + 1, el->alloc);
462 el->excludes[el->nr++] = x;
c04318e4 463 x->el = el;
453ec4bd
LT
464}
465
c28b3d6e
NTND
466static void *read_skip_worktree_file_from_index(const char *path, size_t *size)
467{
468 int pos, len;
469 unsigned long sz;
470 enum object_type type;
471 void *data;
c28b3d6e
NTND
472
473 len = strlen(path);
71261027 474 pos = cache_name_pos(path, len);
c28b3d6e
NTND
475 if (pos < 0)
476 return NULL;
71261027 477 if (!ce_skip_worktree(active_cache[pos]))
c28b3d6e 478 return NULL;
71261027 479 data = read_sha1_file(active_cache[pos]->sha1, &type, &sz);
c28b3d6e
NTND
480 if (!data || type != OBJ_BLOB) {
481 free(data);
482 return NULL;
483 }
484 *size = xsize_t(sz);
485 return data;
486}
487
f6198812
AS
488/*
489 * Frees memory within el which was allocated for exclude patterns and
490 * the file buffer. Does not free el itself.
491 */
492void clear_exclude_list(struct exclude_list *el)
0fd0e241
NTND
493{
494 int i;
495
496 for (i = 0; i < el->nr; i++)
497 free(el->excludes[i]);
498 free(el->excludes);
c082df24 499 free(el->filebuf);
0fd0e241
NTND
500
501 el->nr = 0;
502 el->excludes = NULL;
c082df24 503 el->filebuf = NULL;
0fd0e241
NTND
504}
505
cb097534
NTND
506int add_excludes_from_file_to_list(const char *fname,
507 const char *base,
508 int baselen,
840fc334 509 struct exclude_list *el,
cb097534 510 int check_index)
453ec4bd 511{
c470701a 512 struct stat st;
c04318e4 513 int fd, i, lineno = 1;
9d14017a 514 size_t size = 0;
453ec4bd
LT
515 char *buf, *entry;
516
517 fd = open(fname, O_RDONLY);
c28b3d6e 518 if (fd < 0 || fstat(fd, &st) < 0) {
69660731 519 if (errno != ENOENT)
55b38a48 520 warn_on_inaccessible(fname);
c28b3d6e
NTND
521 if (0 <= fd)
522 close(fd);
523 if (!check_index ||
524 (buf = read_skip_worktree_file_from_index(fname, &size)) == NULL)
525 return -1;
45d76f17
NTND
526 if (size == 0) {
527 free(buf);
528 return 0;
529 }
530 if (buf[size-1] != '\n') {
531 buf = xrealloc(buf, size+1);
532 buf[size++] = '\n';
533 }
453ec4bd 534 }
c28b3d6e
NTND
535 else {
536 size = xsize_t(st.st_size);
537 if (size == 0) {
538 close(fd);
539 return 0;
540 }
45d76f17 541 buf = xmalloc(size+1);
c28b3d6e 542 if (read_in_full(fd, buf, size) != size) {
45d76f17 543 free(buf);
c28b3d6e
NTND
544 close(fd);
545 return -1;
546 }
45d76f17 547 buf[size++] = '\n';
c28b3d6e 548 close(fd);
6ba78238 549 }
453ec4bd 550
c082df24 551 el->filebuf = buf;
453ec4bd 552 entry = buf;
45d76f17
NTND
553 for (i = 0; i < size; i++) {
554 if (buf[i] == '\n') {
453ec4bd
LT
555 if (entry != buf + i && entry[0] != '#') {
556 buf[i - (i && buf[i-1] == '\r')] = 0;
c04318e4 557 add_exclude(entry, base, baselen, el, lineno);
453ec4bd 558 }
c04318e4 559 lineno++;
453ec4bd
LT
560 entry = buf + i + 1;
561 }
562 }
563 return 0;
453ec4bd
LT
564}
565
c04318e4
AS
566struct exclude_list *add_exclude_list(struct dir_struct *dir,
567 int group_type, const char *src)
c082df24
AS
568{
569 struct exclude_list *el;
570 struct exclude_list_group *group;
571
572 group = &dir->exclude_list_group[group_type];
573 ALLOC_GROW(group->el, group->nr + 1, group->alloc);
574 el = &group->el[group->nr++];
575 memset(el, 0, sizeof(*el));
c04318e4 576 el->src = src;
c082df24
AS
577 return el;
578}
579
580/*
581 * Used to set up core.excludesfile and .git/info/exclude lists.
582 */
453ec4bd
LT
583void add_excludes_from_file(struct dir_struct *dir, const char *fname)
584{
c082df24 585 struct exclude_list *el;
c04318e4 586 el = add_exclude_list(dir, EXC_FILE, fname);
c082df24 587 if (add_excludes_from_file_to_list(fname, "", 0, el, 0) < 0)
453ec4bd
LT
588 die("cannot use %s as an exclude file", fname);
589}
590
82dce998
NTND
591int match_basename(const char *basename, int basenamelen,
592 const char *pattern, int prefix, int patternlen,
593 int flags)
593cb880
NTND
594{
595 if (prefix == patternlen) {
0b6e56df
JH
596 if (patternlen == basenamelen &&
597 !strncmp_icase(pattern, basename, basenamelen))
593cb880
NTND
598 return 1;
599 } else if (flags & EXC_FLAG_ENDSWITH) {
0b6e56df 600 /* "*literal" matching against "fooliteral" */
593cb880 601 if (patternlen - 1 <= basenamelen &&
0b6e56df
JH
602 !strncmp_icase(pattern + 1,
603 basename + basenamelen - (patternlen - 1),
604 patternlen - 1))
593cb880
NTND
605 return 1;
606 } else {
0b6e56df
JH
607 if (fnmatch_icase_mem(pattern, patternlen,
608 basename, basenamelen,
609 0) == 0)
593cb880
NTND
610 return 1;
611 }
612 return 0;
613}
614
82dce998
NTND
615int match_pathname(const char *pathname, int pathlen,
616 const char *base, int baselen,
617 const char *pattern, int prefix, int patternlen,
618 int flags)
b5592632
NTND
619{
620 const char *name;
621 int namelen;
622
623 /*
624 * match with FNM_PATHNAME; the pattern has base implicitly
625 * in front of it.
626 */
627 if (*pattern == '/') {
628 pattern++;
982ac873 629 patternlen--;
b5592632
NTND
630 prefix--;
631 }
632
633 /*
634 * baselen does not count the trailing slash. base[] may or
635 * may not end with a trailing slash though.
636 */
637 if (pathlen < baselen + 1 ||
638 (baselen && pathname[baselen] != '/') ||
639 strncmp_icase(pathname, base, baselen))
640 return 0;
641
642 namelen = baselen ? pathlen - baselen - 1 : pathlen;
643 name = pathname + pathlen - namelen;
644
645 if (prefix) {
646 /*
647 * if the non-wildcard part is longer than the
648 * remaining pathname, surely it cannot match.
649 */
650 if (prefix > namelen)
651 return 0;
652
653 if (strncmp_icase(pattern, name, prefix))
654 return 0;
655 pattern += prefix;
ab3aebc1 656 patternlen -= prefix;
b5592632
NTND
657 name += prefix;
658 namelen -= prefix;
ab3aebc1
JK
659
660 /*
661 * If the whole pattern did not have a wildcard,
662 * then our prefix match is all we need; we
663 * do not need to call fnmatch at all.
664 */
665 if (!patternlen && !namelen)
666 return 1;
b5592632
NTND
667 }
668
ab3aebc1
JK
669 return fnmatch_icase_mem(pattern, patternlen,
670 name, namelen,
f30366b2 671 WM_PATHNAME) == 0;
b5592632
NTND
672}
673
578cd7c3
AS
674/*
675 * Scan the given exclude list in reverse to see whether pathname
676 * should be ignored. The first match (i.e. the last on the list), if
677 * any, determines the fate. Returns the exclude_list element which
678 * matched, or NULL for undecided.
453ec4bd 679 */
578cd7c3
AS
680static struct exclude *last_exclude_matching_from_list(const char *pathname,
681 int pathlen,
682 const char *basename,
683 int *dtype,
684 struct exclude_list *el)
453ec4bd
LT
685{
686 int i;
687
35a94d44 688 if (!el->nr)
578cd7c3 689 return NULL; /* undefined */
d6b8fc30 690
35a94d44
NTND
691 for (i = el->nr - 1; 0 <= i; i--) {
692 struct exclude *x = el->excludes[i];
b5592632 693 const char *exclude = x->pattern;
b5592632 694 int prefix = x->nowildcardlen;
35a94d44
NTND
695
696 if (x->flags & EXC_FLAG_MUSTBEDIR) {
697 if (*dtype == DT_UNKNOWN)
698 *dtype = get_dtype(NULL, pathname, pathlen);
699 if (*dtype != DT_DIR)
700 continue;
701 }
d6b8fc30 702
35a94d44 703 if (x->flags & EXC_FLAG_NODIR) {
593cb880
NTND
704 if (match_basename(basename,
705 pathlen - (basename - pathname),
706 exclude, prefix, x->patternlen,
707 x->flags))
578cd7c3 708 return x;
35a94d44
NTND
709 continue;
710 }
711
b5592632
NTND
712 assert(x->baselen == 0 || x->base[x->baselen - 1] == '/');
713 if (match_pathname(pathname, pathlen,
714 x->base, x->baselen ? x->baselen - 1 : 0,
715 exclude, prefix, x->patternlen, x->flags))
578cd7c3 716 return x;
453ec4bd 717 }
578cd7c3
AS
718 return NULL; /* undecided */
719}
720
721/*
722 * Scan the list and let the last match determine the fate.
723 * Return 1 for exclude, 0 for include and -1 for undecided.
724 */
725int is_excluded_from_list(const char *pathname,
726 int pathlen, const char *basename, int *dtype,
727 struct exclude_list *el)
728{
729 struct exclude *exclude;
730 exclude = last_exclude_matching_from_list(pathname, pathlen, basename, dtype, el);
731 if (exclude)
732 return exclude->flags & EXC_FLAG_NEGATIVE ? 0 : 1;
453ec4bd
LT
733 return -1; /* undecided */
734}
735
46aa2f95
KB
736static struct exclude *last_exclude_matching_from_lists(struct dir_struct *dir,
737 const char *pathname, int pathlen, const char *basename,
738 int *dtype_p)
739{
740 int i, j;
741 struct exclude_list_group *group;
742 struct exclude *exclude;
743 for (i = EXC_CMDL; i <= EXC_FILE; i++) {
744 group = &dir->exclude_list_group[i];
745 for (j = group->nr - 1; j >= 0; j--) {
746 exclude = last_exclude_matching_from_list(
747 pathname, pathlen, basename, dtype_p,
748 &group->el[j]);
749 if (exclude)
750 return exclude;
751 }
752 }
753 return NULL;
754}
755
6cd5c582
KB
756/*
757 * Loads the per-directory exclude list for the substring of base
758 * which has a char length of baselen.
759 */
760static void prep_exclude(struct dir_struct *dir, const char *base, int baselen)
761{
762 struct exclude_list_group *group;
763 struct exclude_list *el;
764 struct exclude_stack *stk = NULL;
765 int current;
766
6cd5c582
KB
767 group = &dir->exclude_list_group[EXC_DIRS];
768
769 /* Pop the exclude lists from the EXCL_DIRS exclude_list_group
770 * which originate from directories not in the prefix of the
771 * path being checked. */
772 while ((stk = dir->exclude_stack) != NULL) {
773 if (stk->baselen <= baselen &&
774 !strncmp(dir->basebuf, base, stk->baselen))
775 break;
776 el = &group->el[dir->exclude_stack->exclude_ix];
777 dir->exclude_stack = stk->prev;
95c6f271 778 dir->exclude = NULL;
6cd5c582
KB
779 free((char *)el->src); /* see strdup() below */
780 clear_exclude_list(el);
781 free(stk);
782 group->nr--;
783 }
784
95c6f271
KB
785 /* Skip traversing into sub directories if the parent is excluded */
786 if (dir->exclude)
787 return;
788
6cd5c582
KB
789 /* Read from the parent directories and push them down. */
790 current = stk ? stk->baselen : -1;
791 while (current < baselen) {
792 struct exclude_stack *stk = xcalloc(1, sizeof(*stk));
793 const char *cp;
794
795 if (current < 0) {
796 cp = base;
797 current = 0;
798 }
799 else {
800 cp = strchr(base + current + 1, '/');
801 if (!cp)
802 die("oops in prep_exclude");
803 cp++;
804 }
805 stk->prev = dir->exclude_stack;
806 stk->baselen = cp - base;
95c6f271
KB
807 stk->exclude_ix = group->nr;
808 el = add_exclude_list(dir, EXC_DIRS, NULL);
6cd5c582
KB
809 memcpy(dir->basebuf + current, base + current,
810 stk->baselen - current);
95c6f271
KB
811
812 /* Abort if the directory is excluded */
813 if (stk->baselen) {
814 int dt = DT_DIR;
815 dir->basebuf[stk->baselen - 1] = 0;
816 dir->exclude = last_exclude_matching_from_lists(dir,
817 dir->basebuf, stk->baselen - 1,
818 dir->basebuf + current, &dt);
819 dir->basebuf[stk->baselen - 1] = '/';
c3c327de
KB
820 if (dir->exclude &&
821 dir->exclude->flags & EXC_FLAG_NEGATIVE)
822 dir->exclude = NULL;
95c6f271
KB
823 if (dir->exclude) {
824 dir->basebuf[stk->baselen] = 0;
825 dir->exclude_stack = stk;
826 return;
827 }
828 }
829
830 /* Try to read per-directory file unless path is too long */
831 if (dir->exclude_per_dir &&
832 stk->baselen + strlen(dir->exclude_per_dir) < PATH_MAX) {
833 strcpy(dir->basebuf + stk->baselen,
834 dir->exclude_per_dir);
835 /*
836 * dir->basebuf gets reused by the traversal, but we
837 * need fname to remain unchanged to ensure the src
838 * member of each struct exclude correctly
839 * back-references its source file. Other invocations
840 * of add_exclude_list provide stable strings, so we
841 * strdup() and free() here in the caller.
842 */
843 el->src = strdup(dir->basebuf);
844 add_excludes_from_file_to_list(dir->basebuf,
845 dir->basebuf, stk->baselen, el, 1);
846 }
6cd5c582
KB
847 dir->exclude_stack = stk;
848 current = stk->baselen;
849 }
850 dir->basebuf[baselen] = '\0';
851}
852
f4cd69a6
AS
853/*
854 * Loads the exclude lists for the directory containing pathname, then
855 * scans all exclude lists to determine whether pathname is excluded.
856 * Returns the exclude_list element which matched, or NULL for
857 * undecided.
858 */
b07bc8c8 859struct exclude *last_exclude_matching(struct dir_struct *dir,
f4cd69a6
AS
860 const char *pathname,
861 int *dtype_p)
453ec4bd
LT
862{
863 int pathlen = strlen(pathname);
68492fc7
LK
864 const char *basename = strrchr(pathname, '/');
865 basename = (basename) ? basename+1 : pathname;
453ec4bd 866
63d285c8 867 prep_exclude(dir, pathname, basename-pathname);
c082df24 868
95c6f271
KB
869 if (dir->exclude)
870 return dir->exclude;
871
46aa2f95
KB
872 return last_exclude_matching_from_lists(dir, pathname, pathlen,
873 basename, dtype_p);
f4cd69a6
AS
874}
875
876/*
877 * Loads the exclude lists for the directory containing pathname, then
878 * scans all exclude lists to determine whether pathname is excluded.
879 * Returns 1 if true, otherwise 0.
880 */
b07bc8c8 881int is_excluded(struct dir_struct *dir, const char *pathname, int *dtype_p)
f4cd69a6
AS
882{
883 struct exclude *exclude =
884 last_exclude_matching(dir, pathname, dtype_p);
885 if (exclude)
886 return exclude->flags & EXC_FLAG_NEGATIVE ? 0 : 1;
453ec4bd
LT
887 return 0;
888}
889
f3fa1838
JH
890static struct dir_entry *dir_entry_new(const char *pathname, int len)
891{
453ec4bd
LT
892 struct dir_entry *ent;
893
453ec4bd
LT
894 ent = xmalloc(sizeof(*ent) + len + 1);
895 ent->len = len;
896 memcpy(ent->name, pathname, len);
897 ent->name[len] = 0;
4d06f8ac 898 return ent;
453ec4bd
LT
899}
900
159b3212 901static struct dir_entry *dir_add_name(struct dir_struct *dir, const char *pathname, int len)
6815e569 902{
ebbd7439 903 if (cache_file_exists(pathname, len, ignore_case))
6815e569
JK
904 return NULL;
905
25fd2f7a 906 ALLOC_GROW(dir->entries, dir->nr+1, dir->alloc);
6815e569
JK
907 return dir->entries[dir->nr++] = dir_entry_new(pathname, len);
908}
909
108da0db 910struct dir_entry *dir_add_ignored(struct dir_struct *dir, const char *pathname, int len)
2abd31b0 911{
6e4f981f 912 if (!cache_name_is_other(pathname, len))
2abd31b0
JK
913 return NULL;
914
25fd2f7a 915 ALLOC_GROW(dir->ignored, dir->ignored_nr+1, dir->ignored_alloc);
2abd31b0
JK
916 return dir->ignored[dir->ignored_nr++] = dir_entry_new(pathname, len);
917}
918
09595258
LT
919enum exist_status {
920 index_nonexistent = 0,
921 index_directory,
4b05548f 922 index_gitdir
09595258
LT
923};
924
5102c617 925/*
71261027 926 * Do not use the alphabetically sorted index to look up
5102c617 927 * the directory name; instead, use the case insensitive
ebbd7439 928 * directory hash.
5102c617
JJ
929 */
930static enum exist_status directory_exists_in_index_icase(const char *dirname, int len)
931{
d28eec26 932 const struct cache_entry *ce = cache_dir_exists(dirname, len);
5102c617
JJ
933 unsigned char endchar;
934
935 if (!ce)
936 return index_nonexistent;
937 endchar = ce->name[len];
938
939 /*
940 * The cache_entry structure returned will contain this dirname
941 * and possibly additional path components.
942 */
943 if (endchar == '/')
944 return index_directory;
945
946 /*
947 * If there are no additional path components, then this cache_entry
948 * represents a submodule. Submodules, despite being directories,
949 * are stored in the cache without a closing slash.
950 */
951 if (!endchar && S_ISGITLINK(ce->ce_mode))
952 return index_gitdir;
953
954 /* This should never be hit, but it exists just in case. */
955 return index_nonexistent;
956}
957
09595258
LT
958/*
959 * The index sorts alphabetically by entry name, which
960 * means that a gitlink sorts as '\0' at the end, while
961 * a directory (which is defined not as an entry, but as
962 * the files it contains) will sort with the '/' at the
963 * end.
964 */
965static enum exist_status directory_exists_in_index(const char *dirname, int len)
453ec4bd 966{
5102c617
JJ
967 int pos;
968
969 if (ignore_case)
970 return directory_exists_in_index_icase(dirname, len);
971
972 pos = cache_name_pos(dirname, len);
09595258
LT
973 if (pos < 0)
974 pos = -pos-1;
975 while (pos < active_nr) {
9c5e6c80 976 const struct cache_entry *ce = active_cache[pos++];
09595258
LT
977 unsigned char endchar;
978
979 if (strncmp(ce->name, dirname, len))
980 break;
981 endchar = ce->name[len];
982 if (endchar > '/')
983 break;
984 if (endchar == '/')
985 return index_directory;
7a51ed66 986 if (!endchar && S_ISGITLINK(ce->ce_mode))
09595258
LT
987 return index_gitdir;
988 }
989 return index_nonexistent;
990}
991
992/*
993 * When we find a directory when traversing the filesystem, we
994 * have three distinct cases:
995 *
996 * - ignore it
997 * - see it as a directory
998 * - recurse into it
999 *
1000 * and which one we choose depends on a combination of existing
1001 * git index contents and the flags passed into the directory
1002 * traversal routine.
1003 *
1004 * Case 1: If we *already* have entries in the index under that
5bd8e2d8
KB
1005 * directory name, we always recurse into the directory to see
1006 * all the files.
09595258
LT
1007 *
1008 * Case 2: If we *already* have that directory name as a gitlink,
1009 * we always continue to see it as a gitlink, regardless of whether
1010 * there is an actual git directory there or not (it might not
1011 * be checked out as a subproject!)
1012 *
1013 * Case 3: if we didn't have it in the index previously, we
1014 * have a few sub-cases:
1015 *
1016 * (a) if "show_other_directories" is true, we show it as
1017 * just a directory, unless "hide_empty_directories" is
defd7c7b
KB
1018 * also true, in which case we need to check if it contains any
1019 * untracked and / or ignored files.
09595258 1020 * (b) if it looks like a git directory, and we don't have
302b9282 1021 * 'no_gitlinks' set we treat it as a gitlink, and show it
09595258
LT
1022 * as a directory.
1023 * (c) otherwise, we recurse into it.
1024 */
defd7c7b 1025static enum path_treatment treat_directory(struct dir_struct *dir,
721ac4ed 1026 const char *dirname, int len, int exclude,
09595258
LT
1027 const struct path_simplify *simplify)
1028{
1029 /* The "len-1" is to strip the final '/' */
1030 switch (directory_exists_in_index(dirname, len-1)) {
1031 case index_directory:
defd7c7b 1032 return path_recurse;
09595258
LT
1033
1034 case index_gitdir:
26c986e1 1035 return path_none;
09595258
LT
1036
1037 case index_nonexistent:
7c4c97c0 1038 if (dir->flags & DIR_SHOW_OTHER_DIRECTORIES)
09595258 1039 break;
7c4c97c0 1040 if (!(dir->flags & DIR_NO_GITLINKS)) {
09595258
LT
1041 unsigned char sha1[20];
1042 if (resolve_gitlink_ref(dirname, "HEAD", sha1) == 0)
defd7c7b 1043 return path_untracked;
09595258 1044 }
defd7c7b 1045 return path_recurse;
09595258
LT
1046 }
1047
1048 /* This is the "show_other_directories" case */
721ac4ed 1049
184d2a8e 1050 if (!(dir->flags & DIR_HIDE_EMPTY_DIRECTORIES))
defd7c7b
KB
1051 return exclude ? path_excluded : path_untracked;
1052
1053 return read_directory_recursive(dir, dirname, len, 1, simplify);
453ec4bd
LT
1054}
1055
9fc42d60
LT
1056/*
1057 * This is an inexact early pruning of any recursive directory
1058 * reading - if the path cannot possibly be in the pathspec,
1059 * return true, and we'll skip it early.
1060 */
1061static int simplify_away(const char *path, int pathlen, const struct path_simplify *simplify)
1062{
1063 if (simplify) {
1064 for (;;) {
1065 const char *match = simplify->path;
1066 int len = simplify->len;
1067
1068 if (!match)
1069 break;
1070 if (len > pathlen)
1071 len = pathlen;
1072 if (!memcmp(path, match, len))
1073 return 0;
1074 simplify++;
1075 }
1076 return 1;
1077 }
1078 return 0;
1079}
1080
29209cbe
JK
1081/*
1082 * This function tells us whether an excluded path matches a
1083 * list of "interesting" pathspecs. That is, whether a path matched
1084 * by any of the pathspecs could possibly be ignored by excluding
1085 * the specified path. This can happen if:
1086 *
1087 * 1. the path is mentioned explicitly in the pathspec
1088 *
1089 * 2. the path is a directory prefix of some element in the
1090 * pathspec
1091 */
1092static int exclude_matches_pathspec(const char *path, int len,
1093 const struct path_simplify *simplify)
e96980ef
JK
1094{
1095 if (simplify) {
1096 for (; simplify->path; simplify++) {
1097 if (len == simplify->len
1098 && !memcmp(path, simplify->path, len))
1099 return 1;
29209cbe
JK
1100 if (len < simplify->len
1101 && simplify->path[len] == '/'
1102 && !memcmp(path, simplify->path, len))
1103 return 1;
e96980ef
JK
1104 }
1105 }
1106 return 0;
1107}
1108
443e061a
LT
1109static int get_index_dtype(const char *path, int len)
1110{
1111 int pos;
9c5e6c80 1112 const struct cache_entry *ce;
443e061a 1113
ebbd7439 1114 ce = cache_file_exists(path, len, 0);
443e061a
LT
1115 if (ce) {
1116 if (!ce_uptodate(ce))
1117 return DT_UNKNOWN;
1118 if (S_ISGITLINK(ce->ce_mode))
1119 return DT_DIR;
1120 /*
1121 * Nobody actually cares about the
1122 * difference between DT_LNK and DT_REG
1123 */
1124 return DT_REG;
1125 }
1126
1127 /* Try to look it up as a directory */
1128 pos = cache_name_pos(path, len);
1129 if (pos >= 0)
1130 return DT_UNKNOWN;
1131 pos = -pos-1;
1132 while (pos < active_nr) {
1133 ce = active_cache[pos++];
1134 if (strncmp(ce->name, path, len))
1135 break;
1136 if (ce->name[len] > '/')
1137 break;
1138 if (ce->name[len] < '/')
1139 continue;
1140 if (!ce_uptodate(ce))
1141 break; /* continue? */
1142 return DT_DIR;
1143 }
1144 return DT_UNKNOWN;
1145}
1146
caa6b782 1147static int get_dtype(struct dirent *de, const char *path, int len)
07134421 1148{
6831a88a 1149 int dtype = de ? DTYPE(de) : DT_UNKNOWN;
07134421
LT
1150 struct stat st;
1151
1152 if (dtype != DT_UNKNOWN)
1153 return dtype;
443e061a
LT
1154 dtype = get_index_dtype(path, len);
1155 if (dtype != DT_UNKNOWN)
1156 return dtype;
1157 if (lstat(path, &st))
07134421
LT
1158 return dtype;
1159 if (S_ISREG(st.st_mode))
1160 return DT_REG;
1161 if (S_ISDIR(st.st_mode))
1162 return DT_DIR;
1163 if (S_ISLNK(st.st_mode))
1164 return DT_LNK;
1165 return dtype;
1166}
1167
16e2cfa9 1168static enum path_treatment treat_one_path(struct dir_struct *dir,
49dc2cc2 1169 struct strbuf *path,
16e2cfa9
JH
1170 const struct path_simplify *simplify,
1171 int dtype, struct dirent *de)
53cc5356 1172{
8aaf8d77 1173 int exclude;
ebbd7439 1174 int has_path_in_index = !!cache_file_exists(path->buf, path->len, ignore_case);
2eac2a4c 1175
8aaf8d77
KB
1176 if (dtype == DT_UNKNOWN)
1177 dtype = get_dtype(de, path->buf, path->len);
1178
1179 /* Always exclude indexed files */
2eac2a4c 1180 if (dtype != DT_DIR && has_path_in_index)
defd7c7b 1181 return path_none;
8aaf8d77 1182
2eac2a4c
JH
1183 /*
1184 * When we are looking at a directory P in the working tree,
1185 * there are three cases:
1186 *
1187 * (1) P exists in the index. Everything inside the directory P in
1188 * the working tree needs to go when P is checked out from the
1189 * index.
1190 *
1191 * (2) P does not exist in the index, but there is P/Q in the index.
1192 * We know P will stay a directory when we check out the contents
1193 * of the index, but we do not know yet if there is a directory
1194 * P/Q in the working tree to be killed, so we need to recurse.
1195 *
1196 * (3) P does not exist in the index, and there is no P/Q in the index
1197 * to require P to be a directory, either. Only in this case, we
1198 * know that everything inside P will not be killed without
1199 * recursing.
1200 */
1201 if ((dir->flags & DIR_COLLECT_KILLED_ONLY) &&
1202 (dtype == DT_DIR) &&
de372b1b
ES
1203 !has_path_in_index &&
1204 (directory_exists_in_index(path->buf, path->len) == index_nonexistent))
1205 return path_none;
8aaf8d77
KB
1206
1207 exclude = is_excluded(dir, path->buf, &dtype);
53cc5356
JH
1208
1209 /*
1210 * Excluded? If we don't explicitly want to show
1211 * ignored files, ignore it
1212 */
0aaf62b6 1213 if (exclude && !(dir->flags & (DIR_SHOW_IGNORED|DIR_SHOW_IGNORED_TOO)))
defd7c7b 1214 return path_excluded;
53cc5356 1215
53cc5356
JH
1216 switch (dtype) {
1217 default:
defd7c7b 1218 return path_none;
53cc5356 1219 case DT_DIR:
49dc2cc2 1220 strbuf_addch(path, '/');
defd7c7b
KB
1221 return treat_directory(dir, path->buf, path->len, exclude,
1222 simplify);
53cc5356
JH
1223 case DT_REG:
1224 case DT_LNK:
defd7c7b 1225 return exclude ? path_excluded : path_untracked;
53cc5356 1226 }
53cc5356
JH
1227}
1228
16e2cfa9
JH
1229static enum path_treatment treat_path(struct dir_struct *dir,
1230 struct dirent *de,
49dc2cc2 1231 struct strbuf *path,
16e2cfa9 1232 int baselen,
49dc2cc2 1233 const struct path_simplify *simplify)
16e2cfa9
JH
1234{
1235 int dtype;
1236
1237 if (is_dot_or_dotdot(de->d_name) || !strcmp(de->d_name, ".git"))
defd7c7b 1238 return path_none;
49dc2cc2
RS
1239 strbuf_setlen(path, baselen);
1240 strbuf_addstr(path, de->d_name);
1241 if (simplify_away(path->buf, path->len, simplify))
defd7c7b 1242 return path_none;
16e2cfa9
JH
1243
1244 dtype = DTYPE(de);
49dc2cc2 1245 return treat_one_path(dir, path, simplify, dtype, de);
16e2cfa9
JH
1246}
1247
453ec4bd
LT
1248/*
1249 * Read a directory tree. We currently ignore anything but
1250 * directories, regular files and symlinks. That's because git
1251 * doesn't handle them at all yet. Maybe that will change some
1252 * day.
1253 *
1254 * Also, we ignore the name ".git" (even if it is not a directory).
1255 * That likely will not change.
defd7c7b
KB
1256 *
1257 * Returns the most significant path_treatment value encountered in the scan.
453ec4bd 1258 */
defd7c7b 1259static enum path_treatment read_directory_recursive(struct dir_struct *dir,
53cc5356
JH
1260 const char *base, int baselen,
1261 int check_only,
1262 const struct path_simplify *simplify)
453ec4bd 1263{
1528d247 1264 DIR *fdir;
defd7c7b 1265 enum path_treatment state, subdir_state, dir_state = path_none;
02cb6753 1266 struct dirent *de;
bef36921 1267 struct strbuf path = STRBUF_INIT;
453ec4bd 1268
bef36921 1269 strbuf_add(&path, base, baselen);
02cb6753 1270
1528d247
RS
1271 fdir = opendir(path.len ? path.buf : ".");
1272 if (!fdir)
1273 goto out;
1274
02cb6753 1275 while ((de = readdir(fdir)) != NULL) {
defd7c7b
KB
1276 /* check how the file or directory should be treated */
1277 state = treat_path(dir, de, &path, baselen, simplify);
1278 if (state > dir_state)
1279 dir_state = state;
1280
1281 /* recurse into subdir if instructed by treat_path */
1282 if (state == path_recurse) {
1283 subdir_state = read_directory_recursive(dir, path.buf,
289ff559 1284 path.len, check_only, simplify);
defd7c7b
KB
1285 if (subdir_state > dir_state)
1286 dir_state = subdir_state;
1287 }
1288
1289 if (check_only) {
1290 /* abort early if maximum state has been reached */
1291 if (dir_state == path_untracked)
1292 break;
1293 /* skip the dir_add_* part */
02cb6753 1294 continue;
453ec4bd 1295 }
defd7c7b
KB
1296
1297 /* add the path to the appropriate result list */
1298 switch (state) {
1299 case path_excluded:
1300 if (dir->flags & DIR_SHOW_IGNORED)
1301 dir_add_name(dir, path.buf, path.len);
0aaf62b6
KB
1302 else if ((dir->flags & DIR_SHOW_IGNORED_TOO) ||
1303 ((dir->flags & DIR_COLLECT_IGNORED) &&
1304 exclude_matches_pathspec(path.buf, path.len,
1305 simplify)))
1306 dir_add_ignored(dir, path.buf, path.len);
defd7c7b
KB
1307 break;
1308
1309 case path_untracked:
1310 if (!(dir->flags & DIR_SHOW_IGNORED))
1311 dir_add_name(dir, path.buf, path.len);
1528d247 1312 break;
defd7c7b
KB
1313
1314 default:
1315 break;
1316 }
453ec4bd 1317 }
02cb6753 1318 closedir(fdir);
1528d247 1319 out:
bef36921 1320 strbuf_release(&path);
453ec4bd 1321
defd7c7b 1322 return dir_state;
453ec4bd
LT
1323}
1324
1325static int cmp_name(const void *p1, const void *p2)
1326{
1327 const struct dir_entry *e1 = *(const struct dir_entry **)p1;
1328 const struct dir_entry *e2 = *(const struct dir_entry **)p2;
1329
1330 return cache_name_compare(e1->name, e1->len,
1331 e2->name, e2->len);
1332}
1333
9fc42d60
LT
1334static struct path_simplify *create_simplify(const char **pathspec)
1335{
1336 int nr, alloc = 0;
1337 struct path_simplify *simplify = NULL;
1338
1339 if (!pathspec)
1340 return NULL;
1341
1342 for (nr = 0 ; ; nr++) {
1343 const char *match;
1344 if (nr >= alloc) {
1345 alloc = alloc_nr(alloc);
1346 simplify = xrealloc(simplify, alloc * sizeof(*simplify));
1347 }
1348 match = *pathspec++;
1349 if (!match)
1350 break;
1351 simplify[nr].path = match;
1352 simplify[nr].len = simple_length(match);
1353 }
1354 simplify[nr].path = NULL;
1355 simplify[nr].len = 0;
1356 return simplify;
1357}
1358
1359static void free_simplify(struct path_simplify *simplify)
1360{
8e0f7003 1361 free(simplify);
9fc42d60
LT
1362}
1363
48ffef96
JH
1364static int treat_leading_path(struct dir_struct *dir,
1365 const char *path, int len,
1366 const struct path_simplify *simplify)
1367{
49dc2cc2
RS
1368 struct strbuf sb = STRBUF_INIT;
1369 int baselen, rc = 0;
48ffef96 1370 const char *cp;
be8a84c5 1371 int old_flags = dir->flags;
48ffef96
JH
1372
1373 while (len && path[len - 1] == '/')
1374 len--;
1375 if (!len)
1376 return 1;
1377 baselen = 0;
be8a84c5 1378 dir->flags &= ~DIR_SHOW_OTHER_DIRECTORIES;
48ffef96
JH
1379 while (1) {
1380 cp = path + baselen + !!baselen;
1381 cp = memchr(cp, '/', path + len - cp);
1382 if (!cp)
1383 baselen = len;
1384 else
1385 baselen = cp - path;
49dc2cc2
RS
1386 strbuf_setlen(&sb, 0);
1387 strbuf_add(&sb, path, baselen);
1388 if (!is_directory(sb.buf))
1389 break;
1390 if (simplify_away(sb.buf, sb.len, simplify))
1391 break;
1392 if (treat_one_path(dir, &sb, simplify,
defd7c7b 1393 DT_DIR, NULL) == path_none)
49dc2cc2
RS
1394 break; /* do not recurse into it */
1395 if (len <= baselen) {
1396 rc = 1;
1397 break; /* finished checking */
1398 }
48ffef96 1399 }
49dc2cc2 1400 strbuf_release(&sb);
be8a84c5 1401 dir->flags = old_flags;
49dc2cc2 1402 return rc;
48ffef96
JH
1403}
1404
7327d3d1 1405int read_directory(struct dir_struct *dir, const char *path, int len, const struct pathspec *pathspec)
9fc42d60 1406{
725b0605 1407 struct path_simplify *simplify;
b4189aa8 1408
7327d3d1
NTND
1409 /*
1410 * Check out create_simplify()
1411 */
1412 if (pathspec)
5c6933d2
NTND
1413 GUARD_PATHSPEC(pathspec,
1414 PATHSPEC_FROMTOP |
1415 PATHSPEC_MAXDEPTH |
bd30c2e4 1416 PATHSPEC_LITERAL |
93d93537 1417 PATHSPEC_GLOB |
ef79b1f8
NTND
1418 PATHSPEC_ICASE |
1419 PATHSPEC_EXCLUDE);
7327d3d1 1420
dba2e203 1421 if (has_symlink_leading_path(path, len))
725b0605
JH
1422 return dir->nr;
1423
ef79b1f8
NTND
1424 /*
1425 * exclude patterns are treated like positive ones in
1426 * create_simplify. Usually exclude patterns should be a
1427 * subset of positive ones, which has no impacts on
1428 * create_simplify().
1429 */
b3920bbd 1430 simplify = create_simplify(pathspec ? pathspec->_raw : NULL);
48ffef96
JH
1431 if (!len || treat_leading_path(dir, path, len, simplify))
1432 read_directory_recursive(dir, path, len, 0, simplify);
9fc42d60 1433 free_simplify(simplify);
453ec4bd 1434 qsort(dir->entries, dir->nr, sizeof(struct dir_entry *), cmp_name);
2abd31b0 1435 qsort(dir->ignored, dir->ignored_nr, sizeof(struct dir_entry *), cmp_name);
453ec4bd
LT
1436 return dir->nr;
1437}
c91f0d92 1438
686a4a06 1439int file_exists(const char *f)
c91f0d92 1440{
686a4a06 1441 struct stat sb;
a50f9fc5 1442 return lstat(f, &sb) == 0;
c91f0d92 1443}
e6636747
JS
1444
1445/*
9b125da4
NTND
1446 * Given two normalized paths (a trailing slash is ok), if subdir is
1447 * outside dir, return -1. Otherwise return the offset in subdir that
1448 * can be used as relative path to dir.
e6636747 1449 */
9b125da4 1450int dir_inside_of(const char *subdir, const char *dir)
e6636747 1451{
9b125da4 1452 int offset = 0;
e6636747 1453
9b125da4 1454 assert(dir && subdir && *dir && *subdir);
e6636747 1455
9b125da4 1456 while (*dir && *subdir && *dir == *subdir) {
e6636747 1457 dir++;
9b125da4
NTND
1458 subdir++;
1459 offset++;
490544b1 1460 }
9b125da4
NTND
1461
1462 /* hel[p]/me vs hel[l]/yeah */
1463 if (*dir && *subdir)
1464 return -1;
1465
1466 if (!*subdir)
1467 return !*dir ? offset : -1; /* same dir */
1468
1469 /* foo/[b]ar vs foo/[] */
1470 if (is_dir_sep(dir[-1]))
1471 return is_dir_sep(subdir[-1]) ? offset : -1;
1472
1473 /* foo[/]bar vs foo[] */
1474 return is_dir_sep(*subdir) ? offset + 1 : -1;
e6636747
JS
1475}
1476
1477int is_inside_dir(const char *dir)
1478{
b892913d
NTND
1479 char cwd[PATH_MAX];
1480 if (!dir)
1481 return 0;
1482 if (!getcwd(cwd, sizeof(cwd)))
1483 die_errno("can't find the current directory");
1484 return dir_inside_of(cwd, dir) >= 0;
e6636747 1485}
7155b727 1486
55892d23
AP
1487int is_empty_dir(const char *path)
1488{
1489 DIR *dir = opendir(path);
1490 struct dirent *e;
1491 int ret = 1;
1492
1493 if (!dir)
1494 return 0;
1495
1496 while ((e = readdir(dir)) != NULL)
1497 if (!is_dot_or_dotdot(e->d_name)) {
1498 ret = 0;
1499 break;
1500 }
1501
1502 closedir(dir);
1503 return ret;
1504}
1505
ae2f203e 1506static int remove_dir_recurse(struct strbuf *path, int flag, int *kept_up)
7155b727 1507{
a0f4afbe 1508 DIR *dir;
7155b727 1509 struct dirent *e;
ae2f203e 1510 int ret = 0, original_len = path->len, len, kept_down = 0;
a0f4afbe 1511 int only_empty = (flag & REMOVE_DIR_EMPTY_ONLY);
c844a803 1512 int keep_toplevel = (flag & REMOVE_DIR_KEEP_TOPLEVEL);
a0f4afbe 1513 unsigned char submodule_head[20];
7155b727 1514
a0f4afbe 1515 if ((flag & REMOVE_DIR_KEEP_NESTED_GIT) &&
ae2f203e 1516 !resolve_gitlink_ref(path->buf, "HEAD", submodule_head)) {
a0f4afbe 1517 /* Do not descend and nuke a nested git work tree. */
ae2f203e
JH
1518 if (kept_up)
1519 *kept_up = 1;
a0f4afbe 1520 return 0;
ae2f203e 1521 }
a0f4afbe 1522
ae2f203e 1523 flag &= ~REMOVE_DIR_KEEP_TOPLEVEL;
a0f4afbe 1524 dir = opendir(path->buf);
c844a803 1525 if (!dir) {
863808cd
MH
1526 if (errno == ENOENT)
1527 return keep_toplevel ? -1 : 0;
1528 else if (errno == EACCES && !keep_toplevel)
ecb2c282
MH
1529 /*
1530 * An empty dir could be removable even if it
1531 * is unreadable:
1532 */
c844a803
JH
1533 return rmdir(path->buf);
1534 else
1535 return -1;
1536 }
7155b727
JS
1537 if (path->buf[original_len - 1] != '/')
1538 strbuf_addch(path, '/');
1539
1540 len = path->len;
1541 while ((e = readdir(dir)) != NULL) {
1542 struct stat st;
8ca12c0d
AP
1543 if (is_dot_or_dotdot(e->d_name))
1544 continue;
7155b727
JS
1545
1546 strbuf_setlen(path, len);
1547 strbuf_addstr(path, e->d_name);
863808cd
MH
1548 if (lstat(path->buf, &st)) {
1549 if (errno == ENOENT)
1550 /*
1551 * file disappeared, which is what we
1552 * wanted anyway
1553 */
1554 continue;
1555 /* fall thru */
1556 } else if (S_ISDIR(st.st_mode)) {
ae2f203e 1557 if (!remove_dir_recurse(path, flag, &kept_down))
7155b727 1558 continue; /* happy */
863808cd
MH
1559 } else if (!only_empty &&
1560 (!unlink(path->buf) || errno == ENOENT)) {
7155b727 1561 continue; /* happy, too */
863808cd 1562 }
7155b727
JS
1563
1564 /* path too long, stat fails, or non-directory still exists */
1565 ret = -1;
1566 break;
1567 }
1568 closedir(dir);
1569
1570 strbuf_setlen(path, original_len);
ae2f203e 1571 if (!ret && !keep_toplevel && !kept_down)
863808cd 1572 ret = (!rmdir(path->buf) || errno == ENOENT) ? 0 : -1;
ae2f203e
JH
1573 else if (kept_up)
1574 /*
1575 * report the uplevel that it is not an error that we
1576 * did not rmdir() our directory.
1577 */
1578 *kept_up = !ret;
7155b727
JS
1579 return ret;
1580}
039bc64e 1581
ae2f203e
JH
1582int remove_dir_recursively(struct strbuf *path, int flag)
1583{
1584 return remove_dir_recurse(path, flag, NULL);
1585}
1586
039bc64e
JH
1587void setup_standard_excludes(struct dir_struct *dir)
1588{
1589 const char *path;
dc79687e 1590 char *xdg_path;
039bc64e
JH
1591
1592 dir->exclude_per_dir = ".gitignore";
1593 path = git_path("info/exclude");
dc79687e
HKNN
1594 if (!excludes_file) {
1595 home_config_paths(NULL, &xdg_path, "ignore");
1596 excludes_file = xdg_path;
1597 }
4698c8fe 1598 if (!access_or_warn(path, R_OK, 0))
039bc64e 1599 add_excludes_from_file(dir, path);
4698c8fe 1600 if (excludes_file && !access_or_warn(excludes_file, R_OK, 0))
039bc64e
JH
1601 add_excludes_from_file(dir, excludes_file);
1602}
4a92d1bf
AR
1603
1604int remove_path(const char *name)
1605{
1606 char *slash;
1607
9a6728d4 1608 if (unlink(name) && errno != ENOENT && errno != ENOTDIR)
4a92d1bf
AR
1609 return -1;
1610
1611 slash = strrchr(name, '/');
1612 if (slash) {
1613 char *dirs = xstrdup(name);
1614 slash = dirs + (slash - name);
1615 do {
1616 *slash = '\0';
3fc0d131 1617 } while (rmdir(dirs) == 0 && (slash = strrchr(dirs, '/')));
4a92d1bf
AR
1618 free(dirs);
1619 }
1620 return 0;
1621}
1622
270be816
AS
1623/*
1624 * Frees memory within dir which was allocated for exclude lists and
1625 * the exclude_stack. Does not free dir itself.
1626 */
1627void clear_directory(struct dir_struct *dir)
1628{
1629 int i, j;
1630 struct exclude_list_group *group;
1631 struct exclude_list *el;
1632 struct exclude_stack *stk;
1633
1634 for (i = EXC_CMDL; i <= EXC_FILE; i++) {
1635 group = &dir->exclude_list_group[i];
1636 for (j = 0; j < group->nr; j++) {
1637 el = &group->el[j];
1638 if (i == EXC_DIRS)
1639 free((char *)el->src);
1640 clear_exclude_list(el);
1641 }
1642 free(group->el);
1643 }
1644
1645 stk = dir->exclude_stack;
1646 while (stk) {
1647 struct exclude_stack *prev = stk->prev;
1648 free(stk);
1649 stk = prev;
1650 }
1651}