name-hash: refactor polymorphic index_name_exists()
[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
NTND
128 PATHSPEC_GLOB |
129 PATHSPEC_ICASE);
4a085b16 130
827f4d6c 131 for (n = 0; n < pathspec->nr; n++) {
93d93537
NTND
132 size_t i = 0, len = 0, item_len;
133 if (pathspec->items[n].magic & PATHSPEC_ICASE)
134 item_len = pathspec->items[n].prefix;
135 else
136 item_len = pathspec->items[n].nowildcard_len;
137 while (i < item_len && (n == 0 || i < max)) {
827f4d6c
NTND
138 char c = pathspec->items[n].match[i];
139 if (c != pathspec->items[0].match[i])
4a085b16
JH
140 break;
141 if (c == '/')
142 len = i + 1;
827f4d6c 143 i++;
4a085b16 144 }
827f4d6c 145 if (n == 0 || len < max) {
4a085b16
JH
146 max = len;
147 if (!max)
148 break;
149 }
3c6a370b 150 }
4a085b16 151 return max;
3c6a370b
LT
152}
153
f950eb95
CB
154/*
155 * Returns a copy of the longest leading path common among all
156 * pathspecs.
157 */
827f4d6c 158char *common_prefix(const struct pathspec *pathspec)
f950eb95
CB
159{
160 unsigned long len = common_prefix_len(pathspec);
161
827f4d6c 162 return len ? xmemdupz(pathspec->items[0].match, len) : NULL;
f950eb95
CB
163}
164
7327d3d1 165int fill_directory(struct dir_struct *dir, const struct pathspec *pathspec)
1d8842d9 166{
4a085b16 167 size_t len;
1d8842d9
LT
168
169 /*
170 * Calculate common prefix for the pathspec, and
171 * use that to optimize the directory walk
172 */
4a085b16 173 len = common_prefix_len(pathspec);
1d8842d9
LT
174
175 /* Read the directory and prune it */
b3920bbd 176 read_directory(dir, pathspec->nr ? pathspec->_raw[0] : "", len, pathspec);
dba2e203 177 return len;
1d8842d9
LT
178}
179
bc96cc87
NTND
180int within_depth(const char *name, int namelen,
181 int depth, int max_depth)
182{
183 const char *cp = name, *cpe = name + namelen;
184
185 while (cp < cpe) {
186 if (*cp++ != '/')
187 continue;
188 depth++;
189 if (depth > max_depth)
190 return 0;
191 }
192 return 1;
193}
194
61cf2820
NTND
195/*
196 * Does 'match' match the given name?
197 * A match is found if
198 *
199 * (1) the 'match' string is leading directory of 'name', or
200 * (2) the 'match' string is a wildcard and matches 'name', or
201 * (3) the 'match' string is exactly the same as 'name'.
202 *
203 * and the return value tells which case it was.
204 *
205 * It returns 0 when there is no match.
206 */
207static int match_pathspec_item(const struct pathspec_item *item, int prefix,
208 const char *name, int namelen)
209{
210 /* name/namelen has prefix cut off by caller */
211 const char *match = item->match + prefix;
212 int matchlen = item->len - prefix;
213
93d93537
NTND
214 /*
215 * The normal call pattern is:
216 * 1. prefix = common_prefix_len(ps);
217 * 2. prune something, or fill_directory
218 * 3. match_pathspec_depth()
219 *
220 * 'prefix' at #1 may be shorter than the command's prefix and
221 * it's ok for #2 to match extra files. Those extras will be
222 * trimmed at #3.
223 *
224 * Suppose the pathspec is 'foo' and '../bar' running from
225 * subdir 'xyz'. The common prefix at #1 will be empty, thanks
226 * to "../". We may have xyz/foo _and_ XYZ/foo after #2. The
227 * user does not want XYZ/foo, only the "foo" part should be
228 * case-insensitive. We need to filter out XYZ/foo here. In
229 * other words, we do not trust the caller on comparing the
230 * prefix part when :(icase) is involved. We do exact
231 * comparison ourselves.
232 *
233 * Normally the caller (common_prefix_len() in fact) does
234 * _exact_ matching on name[-prefix+1..-1] and we do not need
235 * to check that part. Be defensive and check it anyway, in
236 * case common_prefix_len is changed, or a new caller is
237 * introduced that does not use common_prefix_len.
238 *
239 * If the penalty turns out too high when prefix is really
240 * long, maybe change it to
241 * strncmp(match, name, item->prefix - prefix)
242 */
243 if (item->prefix && (item->magic & PATHSPEC_ICASE) &&
244 strncmp(item->match, name - prefix, item->prefix))
245 return 0;
246
61cf2820
NTND
247 /* If the match was just the prefix, we matched */
248 if (!*match)
249 return MATCHED_RECURSIVELY;
250
93d93537 251 if (matchlen <= namelen && !ps_strncmp(item, match, name, matchlen)) {
61cf2820
NTND
252 if (matchlen == namelen)
253 return MATCHED_EXACTLY;
254
255 if (match[matchlen-1] == '/' || name[matchlen] == '/')
256 return MATCHED_RECURSIVELY;
257 }
258
5d74762d 259 if (item->nowildcard_len < item->len &&
bd30c2e4 260 !git_fnmatch(item, match, name,
8c6abbcd 261 item->nowildcard_len - prefix))
61cf2820
NTND
262 return MATCHED_FNMATCH;
263
264 return 0;
265}
266
267/*
52ed1894
AS
268 * Given a name and a list of pathspecs, returns the nature of the
269 * closest (i.e. most specific) match of the name to any of the
270 * pathspecs.
271 *
272 * The caller typically calls this multiple times with the same
273 * pathspec and seen[] array but with different name/namelen
274 * (e.g. entries from the index) and is interested in seeing if and
275 * how each pathspec matches all the names it calls this function
276 * with. A mark is left in the seen[] array for each pathspec element
277 * indicating the closest type of match that element achieved, so if
278 * seen[n] remains zero after multiple invocations, that means the nth
279 * pathspec did not match any names, which could indicate that the
280 * user mistyped the nth pathspec.
61cf2820
NTND
281 */
282int match_pathspec_depth(const struct pathspec *ps,
283 const char *name, int namelen,
284 int prefix, char *seen)
285{
286 int i, retval = 0;
287
5c6933d2
NTND
288 GUARD_PATHSPEC(ps,
289 PATHSPEC_FROMTOP |
290 PATHSPEC_MAXDEPTH |
bd30c2e4 291 PATHSPEC_LITERAL |
93d93537
NTND
292 PATHSPEC_GLOB |
293 PATHSPEC_ICASE);
8f4f8f45 294
61cf2820 295 if (!ps->nr) {
6330a171
NTND
296 if (!ps->recursive ||
297 !(ps->magic & PATHSPEC_MAXDEPTH) ||
298 ps->max_depth == -1)
61cf2820
NTND
299 return MATCHED_RECURSIVELY;
300
301 if (within_depth(name, namelen, 0, ps->max_depth))
302 return MATCHED_EXACTLY;
303 else
304 return 0;
305 }
306
307 name += prefix;
308 namelen -= prefix;
309
310 for (i = ps->nr - 1; i >= 0; i--) {
311 int how;
312 if (seen && seen[i] == MATCHED_EXACTLY)
313 continue;
314 how = match_pathspec_item(ps->items+i, prefix, name, namelen);
6330a171
NTND
315 if (ps->recursive &&
316 (ps->magic & PATHSPEC_MAXDEPTH) &&
317 ps->max_depth != -1 &&
61cf2820
NTND
318 how && how != MATCHED_FNMATCH) {
319 int len = ps->items[i].len;
320 if (name[len] == '/')
321 len++;
322 if (within_depth(name+len, namelen-len, 0, ps->max_depth))
323 how = MATCHED_EXACTLY;
324 else
325 how = 0;
326 }
327 if (how) {
328 if (retval < how)
329 retval = how;
330 if (seen && seen[i] < how)
331 seen[i] = how;
332 }
333 }
334 return retval;
335}
336
fcd631ed
NTND
337/*
338 * Return the length of the "simple" part of a path match limiter.
339 */
87323bda 340int simple_length(const char *match)
fcd631ed
NTND
341{
342 int len = -1;
343
344 for (;;) {
345 unsigned char c = *match++;
346 len++;
347 if (c == '\0' || is_glob_special(c))
348 return len;
349 }
350}
351
87323bda 352int no_wildcard(const char *string)
68492fc7 353{
fcd631ed 354 return string[simple_length(string)] == '\0';
68492fc7
LK
355}
356
82dce998
NTND
357void parse_exclude_pattern(const char **pattern,
358 int *patternlen,
359 int *flags,
360 int *nowildcardlen)
84460eec
NTND
361{
362 const char *p = *pattern;
363 size_t i, len;
364
365 *flags = 0;
366 if (*p == '!') {
367 *flags |= EXC_FLAG_NEGATIVE;
368 p++;
369 }
370 len = strlen(p);
371 if (len && p[len - 1] == '/') {
372 len--;
373 *flags |= EXC_FLAG_MUSTBEDIR;
374 }
375 for (i = 0; i < len; i++) {
376 if (p[i] == '/')
377 break;
378 }
379 if (i == len)
380 *flags |= EXC_FLAG_NODIR;
381 *nowildcardlen = simple_length(p);
382 /*
383 * we should have excluded the trailing slash from 'p' too,
384 * but that's one more allocation. Instead just make sure
385 * nowildcardlen does not exceed real patternlen
386 */
387 if (*nowildcardlen > len)
388 *nowildcardlen = len;
389 if (*p == '*' && no_wildcard(p + 1))
390 *flags |= EXC_FLAG_ENDSWITH;
391 *pattern = p;
392 *patternlen = len;
393}
394
453ec4bd 395void add_exclude(const char *string, const char *base,
c04318e4 396 int baselen, struct exclude_list *el, int srcpos)
453ec4bd 397{
d6b8fc30 398 struct exclude *x;
84460eec
NTND
399 int patternlen;
400 int flags;
401 int nowildcardlen;
453ec4bd 402
84460eec
NTND
403 parse_exclude_pattern(&string, &patternlen, &flags, &nowildcardlen);
404 if (flags & EXC_FLAG_MUSTBEDIR) {
d6b8fc30 405 char *s;
84460eec 406 x = xmalloc(sizeof(*x) + patternlen + 1);
4b25d091 407 s = (char *)(x+1);
84460eec
NTND
408 memcpy(s, string, patternlen);
409 s[patternlen] = '\0';
d6b8fc30 410 x->pattern = s;
d6b8fc30
JH
411 } else {
412 x = xmalloc(sizeof(*x));
413 x->pattern = string;
414 }
84460eec
NTND
415 x->patternlen = patternlen;
416 x->nowildcardlen = nowildcardlen;
453ec4bd
LT
417 x->base = base;
418 x->baselen = baselen;
d6b8fc30 419 x->flags = flags;
c04318e4 420 x->srcpos = srcpos;
840fc334
AS
421 ALLOC_GROW(el->excludes, el->nr + 1, el->alloc);
422 el->excludes[el->nr++] = x;
c04318e4 423 x->el = el;
453ec4bd
LT
424}
425
c28b3d6e
NTND
426static void *read_skip_worktree_file_from_index(const char *path, size_t *size)
427{
428 int pos, len;
429 unsigned long sz;
430 enum object_type type;
431 void *data;
c28b3d6e
NTND
432
433 len = strlen(path);
71261027 434 pos = cache_name_pos(path, len);
c28b3d6e
NTND
435 if (pos < 0)
436 return NULL;
71261027 437 if (!ce_skip_worktree(active_cache[pos]))
c28b3d6e 438 return NULL;
71261027 439 data = read_sha1_file(active_cache[pos]->sha1, &type, &sz);
c28b3d6e
NTND
440 if (!data || type != OBJ_BLOB) {
441 free(data);
442 return NULL;
443 }
444 *size = xsize_t(sz);
445 return data;
446}
447
f6198812
AS
448/*
449 * Frees memory within el which was allocated for exclude patterns and
450 * the file buffer. Does not free el itself.
451 */
452void clear_exclude_list(struct exclude_list *el)
0fd0e241
NTND
453{
454 int i;
455
456 for (i = 0; i < el->nr; i++)
457 free(el->excludes[i]);
458 free(el->excludes);
c082df24 459 free(el->filebuf);
0fd0e241
NTND
460
461 el->nr = 0;
462 el->excludes = NULL;
c082df24 463 el->filebuf = NULL;
0fd0e241
NTND
464}
465
cb097534
NTND
466int add_excludes_from_file_to_list(const char *fname,
467 const char *base,
468 int baselen,
840fc334 469 struct exclude_list *el,
cb097534 470 int check_index)
453ec4bd 471{
c470701a 472 struct stat st;
c04318e4 473 int fd, i, lineno = 1;
9d14017a 474 size_t size = 0;
453ec4bd
LT
475 char *buf, *entry;
476
477 fd = open(fname, O_RDONLY);
c28b3d6e 478 if (fd < 0 || fstat(fd, &st) < 0) {
69660731 479 if (errno != ENOENT)
55b38a48 480 warn_on_inaccessible(fname);
c28b3d6e
NTND
481 if (0 <= fd)
482 close(fd);
483 if (!check_index ||
484 (buf = read_skip_worktree_file_from_index(fname, &size)) == NULL)
485 return -1;
45d76f17
NTND
486 if (size == 0) {
487 free(buf);
488 return 0;
489 }
490 if (buf[size-1] != '\n') {
491 buf = xrealloc(buf, size+1);
492 buf[size++] = '\n';
493 }
453ec4bd 494 }
c28b3d6e
NTND
495 else {
496 size = xsize_t(st.st_size);
497 if (size == 0) {
498 close(fd);
499 return 0;
500 }
45d76f17 501 buf = xmalloc(size+1);
c28b3d6e 502 if (read_in_full(fd, buf, size) != size) {
45d76f17 503 free(buf);
c28b3d6e
NTND
504 close(fd);
505 return -1;
506 }
45d76f17 507 buf[size++] = '\n';
c28b3d6e 508 close(fd);
6ba78238 509 }
453ec4bd 510
c082df24 511 el->filebuf = buf;
453ec4bd 512 entry = buf;
45d76f17
NTND
513 for (i = 0; i < size; i++) {
514 if (buf[i] == '\n') {
453ec4bd
LT
515 if (entry != buf + i && entry[0] != '#') {
516 buf[i - (i && buf[i-1] == '\r')] = 0;
c04318e4 517 add_exclude(entry, base, baselen, el, lineno);
453ec4bd 518 }
c04318e4 519 lineno++;
453ec4bd
LT
520 entry = buf + i + 1;
521 }
522 }
523 return 0;
453ec4bd
LT
524}
525
c04318e4
AS
526struct exclude_list *add_exclude_list(struct dir_struct *dir,
527 int group_type, const char *src)
c082df24
AS
528{
529 struct exclude_list *el;
530 struct exclude_list_group *group;
531
532 group = &dir->exclude_list_group[group_type];
533 ALLOC_GROW(group->el, group->nr + 1, group->alloc);
534 el = &group->el[group->nr++];
535 memset(el, 0, sizeof(*el));
c04318e4 536 el->src = src;
c082df24
AS
537 return el;
538}
539
540/*
541 * Used to set up core.excludesfile and .git/info/exclude lists.
542 */
453ec4bd
LT
543void add_excludes_from_file(struct dir_struct *dir, const char *fname)
544{
c082df24 545 struct exclude_list *el;
c04318e4 546 el = add_exclude_list(dir, EXC_FILE, fname);
c082df24 547 if (add_excludes_from_file_to_list(fname, "", 0, el, 0) < 0)
453ec4bd
LT
548 die("cannot use %s as an exclude file", fname);
549}
550
82dce998
NTND
551int match_basename(const char *basename, int basenamelen,
552 const char *pattern, int prefix, int patternlen,
553 int flags)
593cb880
NTND
554{
555 if (prefix == patternlen) {
0b6e56df
JH
556 if (patternlen == basenamelen &&
557 !strncmp_icase(pattern, basename, basenamelen))
593cb880
NTND
558 return 1;
559 } else if (flags & EXC_FLAG_ENDSWITH) {
0b6e56df 560 /* "*literal" matching against "fooliteral" */
593cb880 561 if (patternlen - 1 <= basenamelen &&
0b6e56df
JH
562 !strncmp_icase(pattern + 1,
563 basename + basenamelen - (patternlen - 1),
564 patternlen - 1))
593cb880
NTND
565 return 1;
566 } else {
0b6e56df
JH
567 if (fnmatch_icase_mem(pattern, patternlen,
568 basename, basenamelen,
569 0) == 0)
593cb880
NTND
570 return 1;
571 }
572 return 0;
573}
574
82dce998
NTND
575int match_pathname(const char *pathname, int pathlen,
576 const char *base, int baselen,
577 const char *pattern, int prefix, int patternlen,
578 int flags)
b5592632
NTND
579{
580 const char *name;
581 int namelen;
582
583 /*
584 * match with FNM_PATHNAME; the pattern has base implicitly
585 * in front of it.
586 */
587 if (*pattern == '/') {
588 pattern++;
982ac873 589 patternlen--;
b5592632
NTND
590 prefix--;
591 }
592
593 /*
594 * baselen does not count the trailing slash. base[] may or
595 * may not end with a trailing slash though.
596 */
597 if (pathlen < baselen + 1 ||
598 (baselen && pathname[baselen] != '/') ||
599 strncmp_icase(pathname, base, baselen))
600 return 0;
601
602 namelen = baselen ? pathlen - baselen - 1 : pathlen;
603 name = pathname + pathlen - namelen;
604
605 if (prefix) {
606 /*
607 * if the non-wildcard part is longer than the
608 * remaining pathname, surely it cannot match.
609 */
610 if (prefix > namelen)
611 return 0;
612
613 if (strncmp_icase(pattern, name, prefix))
614 return 0;
615 pattern += prefix;
ab3aebc1 616 patternlen -= prefix;
b5592632
NTND
617 name += prefix;
618 namelen -= prefix;
ab3aebc1
JK
619
620 /*
621 * If the whole pattern did not have a wildcard,
622 * then our prefix match is all we need; we
623 * do not need to call fnmatch at all.
624 */
625 if (!patternlen && !namelen)
626 return 1;
b5592632
NTND
627 }
628
ab3aebc1
JK
629 return fnmatch_icase_mem(pattern, patternlen,
630 name, namelen,
f30366b2 631 WM_PATHNAME) == 0;
b5592632
NTND
632}
633
578cd7c3
AS
634/*
635 * Scan the given exclude list in reverse to see whether pathname
636 * should be ignored. The first match (i.e. the last on the list), if
637 * any, determines the fate. Returns the exclude_list element which
638 * matched, or NULL for undecided.
453ec4bd 639 */
578cd7c3
AS
640static struct exclude *last_exclude_matching_from_list(const char *pathname,
641 int pathlen,
642 const char *basename,
643 int *dtype,
644 struct exclude_list *el)
453ec4bd
LT
645{
646 int i;
647
35a94d44 648 if (!el->nr)
578cd7c3 649 return NULL; /* undefined */
d6b8fc30 650
35a94d44
NTND
651 for (i = el->nr - 1; 0 <= i; i--) {
652 struct exclude *x = el->excludes[i];
b5592632 653 const char *exclude = x->pattern;
b5592632 654 int prefix = x->nowildcardlen;
35a94d44
NTND
655
656 if (x->flags & EXC_FLAG_MUSTBEDIR) {
657 if (*dtype == DT_UNKNOWN)
658 *dtype = get_dtype(NULL, pathname, pathlen);
659 if (*dtype != DT_DIR)
660 continue;
661 }
d6b8fc30 662
35a94d44 663 if (x->flags & EXC_FLAG_NODIR) {
593cb880
NTND
664 if (match_basename(basename,
665 pathlen - (basename - pathname),
666 exclude, prefix, x->patternlen,
667 x->flags))
578cd7c3 668 return x;
35a94d44
NTND
669 continue;
670 }
671
b5592632
NTND
672 assert(x->baselen == 0 || x->base[x->baselen - 1] == '/');
673 if (match_pathname(pathname, pathlen,
674 x->base, x->baselen ? x->baselen - 1 : 0,
675 exclude, prefix, x->patternlen, x->flags))
578cd7c3 676 return x;
453ec4bd 677 }
578cd7c3
AS
678 return NULL; /* undecided */
679}
680
681/*
682 * Scan the list and let the last match determine the fate.
683 * Return 1 for exclude, 0 for include and -1 for undecided.
684 */
685int is_excluded_from_list(const char *pathname,
686 int pathlen, const char *basename, int *dtype,
687 struct exclude_list *el)
688{
689 struct exclude *exclude;
690 exclude = last_exclude_matching_from_list(pathname, pathlen, basename, dtype, el);
691 if (exclude)
692 return exclude->flags & EXC_FLAG_NEGATIVE ? 0 : 1;
453ec4bd
LT
693 return -1; /* undecided */
694}
695
46aa2f95
KB
696static struct exclude *last_exclude_matching_from_lists(struct dir_struct *dir,
697 const char *pathname, int pathlen, const char *basename,
698 int *dtype_p)
699{
700 int i, j;
701 struct exclude_list_group *group;
702 struct exclude *exclude;
703 for (i = EXC_CMDL; i <= EXC_FILE; i++) {
704 group = &dir->exclude_list_group[i];
705 for (j = group->nr - 1; j >= 0; j--) {
706 exclude = last_exclude_matching_from_list(
707 pathname, pathlen, basename, dtype_p,
708 &group->el[j]);
709 if (exclude)
710 return exclude;
711 }
712 }
713 return NULL;
714}
715
6cd5c582
KB
716/*
717 * Loads the per-directory exclude list for the substring of base
718 * which has a char length of baselen.
719 */
720static void prep_exclude(struct dir_struct *dir, const char *base, int baselen)
721{
722 struct exclude_list_group *group;
723 struct exclude_list *el;
724 struct exclude_stack *stk = NULL;
725 int current;
726
6cd5c582
KB
727 group = &dir->exclude_list_group[EXC_DIRS];
728
729 /* Pop the exclude lists from the EXCL_DIRS exclude_list_group
730 * which originate from directories not in the prefix of the
731 * path being checked. */
732 while ((stk = dir->exclude_stack) != NULL) {
733 if (stk->baselen <= baselen &&
734 !strncmp(dir->basebuf, base, stk->baselen))
735 break;
736 el = &group->el[dir->exclude_stack->exclude_ix];
737 dir->exclude_stack = stk->prev;
95c6f271 738 dir->exclude = NULL;
6cd5c582
KB
739 free((char *)el->src); /* see strdup() below */
740 clear_exclude_list(el);
741 free(stk);
742 group->nr--;
743 }
744
95c6f271
KB
745 /* Skip traversing into sub directories if the parent is excluded */
746 if (dir->exclude)
747 return;
748
6cd5c582
KB
749 /* Read from the parent directories and push them down. */
750 current = stk ? stk->baselen : -1;
751 while (current < baselen) {
752 struct exclude_stack *stk = xcalloc(1, sizeof(*stk));
753 const char *cp;
754
755 if (current < 0) {
756 cp = base;
757 current = 0;
758 }
759 else {
760 cp = strchr(base + current + 1, '/');
761 if (!cp)
762 die("oops in prep_exclude");
763 cp++;
764 }
765 stk->prev = dir->exclude_stack;
766 stk->baselen = cp - base;
95c6f271
KB
767 stk->exclude_ix = group->nr;
768 el = add_exclude_list(dir, EXC_DIRS, NULL);
6cd5c582
KB
769 memcpy(dir->basebuf + current, base + current,
770 stk->baselen - current);
95c6f271
KB
771
772 /* Abort if the directory is excluded */
773 if (stk->baselen) {
774 int dt = DT_DIR;
775 dir->basebuf[stk->baselen - 1] = 0;
776 dir->exclude = last_exclude_matching_from_lists(dir,
777 dir->basebuf, stk->baselen - 1,
778 dir->basebuf + current, &dt);
779 dir->basebuf[stk->baselen - 1] = '/';
c3c327de
KB
780 if (dir->exclude &&
781 dir->exclude->flags & EXC_FLAG_NEGATIVE)
782 dir->exclude = NULL;
95c6f271
KB
783 if (dir->exclude) {
784 dir->basebuf[stk->baselen] = 0;
785 dir->exclude_stack = stk;
786 return;
787 }
788 }
789
790 /* Try to read per-directory file unless path is too long */
791 if (dir->exclude_per_dir &&
792 stk->baselen + strlen(dir->exclude_per_dir) < PATH_MAX) {
793 strcpy(dir->basebuf + stk->baselen,
794 dir->exclude_per_dir);
795 /*
796 * dir->basebuf gets reused by the traversal, but we
797 * need fname to remain unchanged to ensure the src
798 * member of each struct exclude correctly
799 * back-references its source file. Other invocations
800 * of add_exclude_list provide stable strings, so we
801 * strdup() and free() here in the caller.
802 */
803 el->src = strdup(dir->basebuf);
804 add_excludes_from_file_to_list(dir->basebuf,
805 dir->basebuf, stk->baselen, el, 1);
806 }
6cd5c582
KB
807 dir->exclude_stack = stk;
808 current = stk->baselen;
809 }
810 dir->basebuf[baselen] = '\0';
811}
812
f4cd69a6
AS
813/*
814 * Loads the exclude lists for the directory containing pathname, then
815 * scans all exclude lists to determine whether pathname is excluded.
816 * Returns the exclude_list element which matched, or NULL for
817 * undecided.
818 */
b07bc8c8 819struct exclude *last_exclude_matching(struct dir_struct *dir,
f4cd69a6
AS
820 const char *pathname,
821 int *dtype_p)
453ec4bd
LT
822{
823 int pathlen = strlen(pathname);
68492fc7
LK
824 const char *basename = strrchr(pathname, '/');
825 basename = (basename) ? basename+1 : pathname;
453ec4bd 826
63d285c8 827 prep_exclude(dir, pathname, basename-pathname);
c082df24 828
95c6f271
KB
829 if (dir->exclude)
830 return dir->exclude;
831
46aa2f95
KB
832 return last_exclude_matching_from_lists(dir, pathname, pathlen,
833 basename, dtype_p);
f4cd69a6
AS
834}
835
836/*
837 * Loads the exclude lists for the directory containing pathname, then
838 * scans all exclude lists to determine whether pathname is excluded.
839 * Returns 1 if true, otherwise 0.
840 */
b07bc8c8 841int is_excluded(struct dir_struct *dir, const char *pathname, int *dtype_p)
f4cd69a6
AS
842{
843 struct exclude *exclude =
844 last_exclude_matching(dir, pathname, dtype_p);
845 if (exclude)
846 return exclude->flags & EXC_FLAG_NEGATIVE ? 0 : 1;
453ec4bd
LT
847 return 0;
848}
849
f3fa1838
JH
850static struct dir_entry *dir_entry_new(const char *pathname, int len)
851{
453ec4bd
LT
852 struct dir_entry *ent;
853
453ec4bd
LT
854 ent = xmalloc(sizeof(*ent) + len + 1);
855 ent->len = len;
856 memcpy(ent->name, pathname, len);
857 ent->name[len] = 0;
4d06f8ac 858 return ent;
453ec4bd
LT
859}
860
159b3212 861static struct dir_entry *dir_add_name(struct dir_struct *dir, const char *pathname, int len)
6815e569 862{
5bd8e2d8 863 if (cache_name_exists(pathname, len, ignore_case))
6815e569
JK
864 return NULL;
865
25fd2f7a 866 ALLOC_GROW(dir->entries, dir->nr+1, dir->alloc);
6815e569
JK
867 return dir->entries[dir->nr++] = dir_entry_new(pathname, len);
868}
869
108da0db 870struct dir_entry *dir_add_ignored(struct dir_struct *dir, const char *pathname, int len)
2abd31b0 871{
6e4f981f 872 if (!cache_name_is_other(pathname, len))
2abd31b0
JK
873 return NULL;
874
25fd2f7a 875 ALLOC_GROW(dir->ignored, dir->ignored_nr+1, dir->ignored_alloc);
2abd31b0
JK
876 return dir->ignored[dir->ignored_nr++] = dir_entry_new(pathname, len);
877}
878
09595258
LT
879enum exist_status {
880 index_nonexistent = 0,
881 index_directory,
4b05548f 882 index_gitdir
09595258
LT
883};
884
5102c617 885/*
71261027 886 * Do not use the alphabetically sorted index to look up
5102c617
JJ
887 * the directory name; instead, use the case insensitive
888 * name hash.
889 */
890static enum exist_status directory_exists_in_index_icase(const char *dirname, int len)
891{
4c4d9d9b 892 const struct cache_entry *ce = cache_name_exists(dirname, len + 1, ignore_case);
5102c617
JJ
893 unsigned char endchar;
894
895 if (!ce)
896 return index_nonexistent;
897 endchar = ce->name[len];
898
899 /*
900 * The cache_entry structure returned will contain this dirname
901 * and possibly additional path components.
902 */
903 if (endchar == '/')
904 return index_directory;
905
906 /*
907 * If there are no additional path components, then this cache_entry
908 * represents a submodule. Submodules, despite being directories,
909 * are stored in the cache without a closing slash.
910 */
911 if (!endchar && S_ISGITLINK(ce->ce_mode))
912 return index_gitdir;
913
914 /* This should never be hit, but it exists just in case. */
915 return index_nonexistent;
916}
917
09595258
LT
918/*
919 * The index sorts alphabetically by entry name, which
920 * means that a gitlink sorts as '\0' at the end, while
921 * a directory (which is defined not as an entry, but as
922 * the files it contains) will sort with the '/' at the
923 * end.
924 */
925static enum exist_status directory_exists_in_index(const char *dirname, int len)
453ec4bd 926{
5102c617
JJ
927 int pos;
928
929 if (ignore_case)
930 return directory_exists_in_index_icase(dirname, len);
931
932 pos = cache_name_pos(dirname, len);
09595258
LT
933 if (pos < 0)
934 pos = -pos-1;
935 while (pos < active_nr) {
9c5e6c80 936 const struct cache_entry *ce = active_cache[pos++];
09595258
LT
937 unsigned char endchar;
938
939 if (strncmp(ce->name, dirname, len))
940 break;
941 endchar = ce->name[len];
942 if (endchar > '/')
943 break;
944 if (endchar == '/')
945 return index_directory;
7a51ed66 946 if (!endchar && S_ISGITLINK(ce->ce_mode))
09595258
LT
947 return index_gitdir;
948 }
949 return index_nonexistent;
950}
951
952/*
953 * When we find a directory when traversing the filesystem, we
954 * have three distinct cases:
955 *
956 * - ignore it
957 * - see it as a directory
958 * - recurse into it
959 *
960 * and which one we choose depends on a combination of existing
961 * git index contents and the flags passed into the directory
962 * traversal routine.
963 *
964 * Case 1: If we *already* have entries in the index under that
5bd8e2d8
KB
965 * directory name, we always recurse into the directory to see
966 * all the files.
09595258
LT
967 *
968 * Case 2: If we *already* have that directory name as a gitlink,
969 * we always continue to see it as a gitlink, regardless of whether
970 * there is an actual git directory there or not (it might not
971 * be checked out as a subproject!)
972 *
973 * Case 3: if we didn't have it in the index previously, we
974 * have a few sub-cases:
975 *
976 * (a) if "show_other_directories" is true, we show it as
977 * just a directory, unless "hide_empty_directories" is
defd7c7b
KB
978 * also true, in which case we need to check if it contains any
979 * untracked and / or ignored files.
09595258 980 * (b) if it looks like a git directory, and we don't have
302b9282 981 * 'no_gitlinks' set we treat it as a gitlink, and show it
09595258
LT
982 * as a directory.
983 * (c) otherwise, we recurse into it.
984 */
defd7c7b 985static enum path_treatment treat_directory(struct dir_struct *dir,
721ac4ed 986 const char *dirname, int len, int exclude,
09595258
LT
987 const struct path_simplify *simplify)
988{
989 /* The "len-1" is to strip the final '/' */
990 switch (directory_exists_in_index(dirname, len-1)) {
991 case index_directory:
defd7c7b 992 return path_recurse;
09595258
LT
993
994 case index_gitdir:
26c986e1 995 return path_none;
09595258
LT
996
997 case index_nonexistent:
7c4c97c0 998 if (dir->flags & DIR_SHOW_OTHER_DIRECTORIES)
09595258 999 break;
7c4c97c0 1000 if (!(dir->flags & DIR_NO_GITLINKS)) {
09595258
LT
1001 unsigned char sha1[20];
1002 if (resolve_gitlink_ref(dirname, "HEAD", sha1) == 0)
defd7c7b 1003 return path_untracked;
09595258 1004 }
defd7c7b 1005 return path_recurse;
09595258
LT
1006 }
1007
1008 /* This is the "show_other_directories" case */
721ac4ed 1009
184d2a8e 1010 if (!(dir->flags & DIR_HIDE_EMPTY_DIRECTORIES))
defd7c7b
KB
1011 return exclude ? path_excluded : path_untracked;
1012
1013 return read_directory_recursive(dir, dirname, len, 1, simplify);
453ec4bd
LT
1014}
1015
9fc42d60
LT
1016/*
1017 * This is an inexact early pruning of any recursive directory
1018 * reading - if the path cannot possibly be in the pathspec,
1019 * return true, and we'll skip it early.
1020 */
1021static int simplify_away(const char *path, int pathlen, const struct path_simplify *simplify)
1022{
1023 if (simplify) {
1024 for (;;) {
1025 const char *match = simplify->path;
1026 int len = simplify->len;
1027
1028 if (!match)
1029 break;
1030 if (len > pathlen)
1031 len = pathlen;
1032 if (!memcmp(path, match, len))
1033 return 0;
1034 simplify++;
1035 }
1036 return 1;
1037 }
1038 return 0;
1039}
1040
29209cbe
JK
1041/*
1042 * This function tells us whether an excluded path matches a
1043 * list of "interesting" pathspecs. That is, whether a path matched
1044 * by any of the pathspecs could possibly be ignored by excluding
1045 * the specified path. This can happen if:
1046 *
1047 * 1. the path is mentioned explicitly in the pathspec
1048 *
1049 * 2. the path is a directory prefix of some element in the
1050 * pathspec
1051 */
1052static int exclude_matches_pathspec(const char *path, int len,
1053 const struct path_simplify *simplify)
e96980ef
JK
1054{
1055 if (simplify) {
1056 for (; simplify->path; simplify++) {
1057 if (len == simplify->len
1058 && !memcmp(path, simplify->path, len))
1059 return 1;
29209cbe
JK
1060 if (len < simplify->len
1061 && simplify->path[len] == '/'
1062 && !memcmp(path, simplify->path, len))
1063 return 1;
e96980ef
JK
1064 }
1065 }
1066 return 0;
1067}
1068
443e061a
LT
1069static int get_index_dtype(const char *path, int len)
1070{
1071 int pos;
9c5e6c80 1072 const struct cache_entry *ce;
443e061a
LT
1073
1074 ce = cache_name_exists(path, len, 0);
1075 if (ce) {
1076 if (!ce_uptodate(ce))
1077 return DT_UNKNOWN;
1078 if (S_ISGITLINK(ce->ce_mode))
1079 return DT_DIR;
1080 /*
1081 * Nobody actually cares about the
1082 * difference between DT_LNK and DT_REG
1083 */
1084 return DT_REG;
1085 }
1086
1087 /* Try to look it up as a directory */
1088 pos = cache_name_pos(path, len);
1089 if (pos >= 0)
1090 return DT_UNKNOWN;
1091 pos = -pos-1;
1092 while (pos < active_nr) {
1093 ce = active_cache[pos++];
1094 if (strncmp(ce->name, path, len))
1095 break;
1096 if (ce->name[len] > '/')
1097 break;
1098 if (ce->name[len] < '/')
1099 continue;
1100 if (!ce_uptodate(ce))
1101 break; /* continue? */
1102 return DT_DIR;
1103 }
1104 return DT_UNKNOWN;
1105}
1106
caa6b782 1107static int get_dtype(struct dirent *de, const char *path, int len)
07134421 1108{
6831a88a 1109 int dtype = de ? DTYPE(de) : DT_UNKNOWN;
07134421
LT
1110 struct stat st;
1111
1112 if (dtype != DT_UNKNOWN)
1113 return dtype;
443e061a
LT
1114 dtype = get_index_dtype(path, len);
1115 if (dtype != DT_UNKNOWN)
1116 return dtype;
1117 if (lstat(path, &st))
07134421
LT
1118 return dtype;
1119 if (S_ISREG(st.st_mode))
1120 return DT_REG;
1121 if (S_ISDIR(st.st_mode))
1122 return DT_DIR;
1123 if (S_ISLNK(st.st_mode))
1124 return DT_LNK;
1125 return dtype;
1126}
1127
16e2cfa9 1128static enum path_treatment treat_one_path(struct dir_struct *dir,
49dc2cc2 1129 struct strbuf *path,
16e2cfa9
JH
1130 const struct path_simplify *simplify,
1131 int dtype, struct dirent *de)
53cc5356 1132{
8aaf8d77 1133 int exclude;
2eac2a4c
JH
1134 int has_path_in_index = !!cache_name_exists(path->buf, path->len, ignore_case);
1135
8aaf8d77
KB
1136 if (dtype == DT_UNKNOWN)
1137 dtype = get_dtype(de, path->buf, path->len);
1138
1139 /* Always exclude indexed files */
2eac2a4c 1140 if (dtype != DT_DIR && has_path_in_index)
defd7c7b 1141 return path_none;
8aaf8d77 1142
2eac2a4c
JH
1143 /*
1144 * When we are looking at a directory P in the working tree,
1145 * there are three cases:
1146 *
1147 * (1) P exists in the index. Everything inside the directory P in
1148 * the working tree needs to go when P is checked out from the
1149 * index.
1150 *
1151 * (2) P does not exist in the index, but there is P/Q in the index.
1152 * We know P will stay a directory when we check out the contents
1153 * of the index, but we do not know yet if there is a directory
1154 * P/Q in the working tree to be killed, so we need to recurse.
1155 *
1156 * (3) P does not exist in the index, and there is no P/Q in the index
1157 * to require P to be a directory, either. Only in this case, we
1158 * know that everything inside P will not be killed without
1159 * recursing.
1160 */
1161 if ((dir->flags & DIR_COLLECT_KILLED_ONLY) &&
1162 (dtype == DT_DIR) &&
680be044
ES
1163 !has_path_in_index) {
1164 /*
1165 * NEEDSWORK: directory_exists_in_index_icase()
1166 * assumes that one byte past the given path is
1167 * readable and has '/', which needs to be fixed, but
1168 * until then, work it around in the caller.
1169 */
1170 strbuf_addch(path, '/');
1171 if (directory_exists_in_index(path->buf, path->len - 1) ==
1172 index_nonexistent) {
1173 strbuf_setlen(path, path->len - 1);
1174 return path_none;
1175 }
1176 strbuf_setlen(path, path->len - 1);
1177 }
8aaf8d77
KB
1178
1179 exclude = is_excluded(dir, path->buf, &dtype);
53cc5356
JH
1180
1181 /*
1182 * Excluded? If we don't explicitly want to show
1183 * ignored files, ignore it
1184 */
0aaf62b6 1185 if (exclude && !(dir->flags & (DIR_SHOW_IGNORED|DIR_SHOW_IGNORED_TOO)))
defd7c7b 1186 return path_excluded;
53cc5356 1187
53cc5356
JH
1188 switch (dtype) {
1189 default:
defd7c7b 1190 return path_none;
53cc5356 1191 case DT_DIR:
49dc2cc2 1192 strbuf_addch(path, '/');
defd7c7b
KB
1193 return treat_directory(dir, path->buf, path->len, exclude,
1194 simplify);
53cc5356
JH
1195 case DT_REG:
1196 case DT_LNK:
defd7c7b 1197 return exclude ? path_excluded : path_untracked;
53cc5356 1198 }
53cc5356
JH
1199}
1200
16e2cfa9
JH
1201static enum path_treatment treat_path(struct dir_struct *dir,
1202 struct dirent *de,
49dc2cc2 1203 struct strbuf *path,
16e2cfa9 1204 int baselen,
49dc2cc2 1205 const struct path_simplify *simplify)
16e2cfa9
JH
1206{
1207 int dtype;
1208
1209 if (is_dot_or_dotdot(de->d_name) || !strcmp(de->d_name, ".git"))
defd7c7b 1210 return path_none;
49dc2cc2
RS
1211 strbuf_setlen(path, baselen);
1212 strbuf_addstr(path, de->d_name);
1213 if (simplify_away(path->buf, path->len, simplify))
defd7c7b 1214 return path_none;
16e2cfa9
JH
1215
1216 dtype = DTYPE(de);
49dc2cc2 1217 return treat_one_path(dir, path, simplify, dtype, de);
16e2cfa9
JH
1218}
1219
453ec4bd
LT
1220/*
1221 * Read a directory tree. We currently ignore anything but
1222 * directories, regular files and symlinks. That's because git
1223 * doesn't handle them at all yet. Maybe that will change some
1224 * day.
1225 *
1226 * Also, we ignore the name ".git" (even if it is not a directory).
1227 * That likely will not change.
defd7c7b
KB
1228 *
1229 * Returns the most significant path_treatment value encountered in the scan.
453ec4bd 1230 */
defd7c7b 1231static enum path_treatment read_directory_recursive(struct dir_struct *dir,
53cc5356
JH
1232 const char *base, int baselen,
1233 int check_only,
1234 const struct path_simplify *simplify)
453ec4bd 1235{
1528d247 1236 DIR *fdir;
defd7c7b 1237 enum path_treatment state, subdir_state, dir_state = path_none;
02cb6753 1238 struct dirent *de;
bef36921 1239 struct strbuf path = STRBUF_INIT;
453ec4bd 1240
bef36921 1241 strbuf_add(&path, base, baselen);
02cb6753 1242
1528d247
RS
1243 fdir = opendir(path.len ? path.buf : ".");
1244 if (!fdir)
1245 goto out;
1246
02cb6753 1247 while ((de = readdir(fdir)) != NULL) {
defd7c7b
KB
1248 /* check how the file or directory should be treated */
1249 state = treat_path(dir, de, &path, baselen, simplify);
1250 if (state > dir_state)
1251 dir_state = state;
1252
1253 /* recurse into subdir if instructed by treat_path */
1254 if (state == path_recurse) {
1255 subdir_state = read_directory_recursive(dir, path.buf,
289ff559 1256 path.len, check_only, simplify);
defd7c7b
KB
1257 if (subdir_state > dir_state)
1258 dir_state = subdir_state;
1259 }
1260
1261 if (check_only) {
1262 /* abort early if maximum state has been reached */
1263 if (dir_state == path_untracked)
1264 break;
1265 /* skip the dir_add_* part */
02cb6753 1266 continue;
453ec4bd 1267 }
defd7c7b
KB
1268
1269 /* add the path to the appropriate result list */
1270 switch (state) {
1271 case path_excluded:
1272 if (dir->flags & DIR_SHOW_IGNORED)
1273 dir_add_name(dir, path.buf, path.len);
0aaf62b6
KB
1274 else if ((dir->flags & DIR_SHOW_IGNORED_TOO) ||
1275 ((dir->flags & DIR_COLLECT_IGNORED) &&
1276 exclude_matches_pathspec(path.buf, path.len,
1277 simplify)))
1278 dir_add_ignored(dir, path.buf, path.len);
defd7c7b
KB
1279 break;
1280
1281 case path_untracked:
1282 if (!(dir->flags & DIR_SHOW_IGNORED))
1283 dir_add_name(dir, path.buf, path.len);
1528d247 1284 break;
defd7c7b
KB
1285
1286 default:
1287 break;
1288 }
453ec4bd 1289 }
02cb6753 1290 closedir(fdir);
1528d247 1291 out:
bef36921 1292 strbuf_release(&path);
453ec4bd 1293
defd7c7b 1294 return dir_state;
453ec4bd
LT
1295}
1296
1297static int cmp_name(const void *p1, const void *p2)
1298{
1299 const struct dir_entry *e1 = *(const struct dir_entry **)p1;
1300 const struct dir_entry *e2 = *(const struct dir_entry **)p2;
1301
1302 return cache_name_compare(e1->name, e1->len,
1303 e2->name, e2->len);
1304}
1305
9fc42d60
LT
1306static struct path_simplify *create_simplify(const char **pathspec)
1307{
1308 int nr, alloc = 0;
1309 struct path_simplify *simplify = NULL;
1310
1311 if (!pathspec)
1312 return NULL;
1313
1314 for (nr = 0 ; ; nr++) {
1315 const char *match;
1316 if (nr >= alloc) {
1317 alloc = alloc_nr(alloc);
1318 simplify = xrealloc(simplify, alloc * sizeof(*simplify));
1319 }
1320 match = *pathspec++;
1321 if (!match)
1322 break;
1323 simplify[nr].path = match;
1324 simplify[nr].len = simple_length(match);
1325 }
1326 simplify[nr].path = NULL;
1327 simplify[nr].len = 0;
1328 return simplify;
1329}
1330
1331static void free_simplify(struct path_simplify *simplify)
1332{
8e0f7003 1333 free(simplify);
9fc42d60
LT
1334}
1335
48ffef96
JH
1336static int treat_leading_path(struct dir_struct *dir,
1337 const char *path, int len,
1338 const struct path_simplify *simplify)
1339{
49dc2cc2
RS
1340 struct strbuf sb = STRBUF_INIT;
1341 int baselen, rc = 0;
48ffef96 1342 const char *cp;
be8a84c5 1343 int old_flags = dir->flags;
48ffef96
JH
1344
1345 while (len && path[len - 1] == '/')
1346 len--;
1347 if (!len)
1348 return 1;
1349 baselen = 0;
be8a84c5 1350 dir->flags &= ~DIR_SHOW_OTHER_DIRECTORIES;
48ffef96
JH
1351 while (1) {
1352 cp = path + baselen + !!baselen;
1353 cp = memchr(cp, '/', path + len - cp);
1354 if (!cp)
1355 baselen = len;
1356 else
1357 baselen = cp - path;
49dc2cc2
RS
1358 strbuf_setlen(&sb, 0);
1359 strbuf_add(&sb, path, baselen);
1360 if (!is_directory(sb.buf))
1361 break;
1362 if (simplify_away(sb.buf, sb.len, simplify))
1363 break;
1364 if (treat_one_path(dir, &sb, simplify,
defd7c7b 1365 DT_DIR, NULL) == path_none)
49dc2cc2
RS
1366 break; /* do not recurse into it */
1367 if (len <= baselen) {
1368 rc = 1;
1369 break; /* finished checking */
1370 }
48ffef96 1371 }
49dc2cc2 1372 strbuf_release(&sb);
be8a84c5 1373 dir->flags = old_flags;
49dc2cc2 1374 return rc;
48ffef96
JH
1375}
1376
7327d3d1 1377int read_directory(struct dir_struct *dir, const char *path, int len, const struct pathspec *pathspec)
9fc42d60 1378{
725b0605 1379 struct path_simplify *simplify;
b4189aa8 1380
7327d3d1
NTND
1381 /*
1382 * Check out create_simplify()
1383 */
1384 if (pathspec)
5c6933d2
NTND
1385 GUARD_PATHSPEC(pathspec,
1386 PATHSPEC_FROMTOP |
1387 PATHSPEC_MAXDEPTH |
bd30c2e4 1388 PATHSPEC_LITERAL |
93d93537
NTND
1389 PATHSPEC_GLOB |
1390 PATHSPEC_ICASE);
7327d3d1 1391
dba2e203 1392 if (has_symlink_leading_path(path, len))
725b0605
JH
1393 return dir->nr;
1394
b3920bbd 1395 simplify = create_simplify(pathspec ? pathspec->_raw : NULL);
48ffef96
JH
1396 if (!len || treat_leading_path(dir, path, len, simplify))
1397 read_directory_recursive(dir, path, len, 0, simplify);
9fc42d60 1398 free_simplify(simplify);
453ec4bd 1399 qsort(dir->entries, dir->nr, sizeof(struct dir_entry *), cmp_name);
2abd31b0 1400 qsort(dir->ignored, dir->ignored_nr, sizeof(struct dir_entry *), cmp_name);
453ec4bd
LT
1401 return dir->nr;
1402}
c91f0d92 1403
686a4a06 1404int file_exists(const char *f)
c91f0d92 1405{
686a4a06 1406 struct stat sb;
a50f9fc5 1407 return lstat(f, &sb) == 0;
c91f0d92 1408}
e6636747
JS
1409
1410/*
9b125da4
NTND
1411 * Given two normalized paths (a trailing slash is ok), if subdir is
1412 * outside dir, return -1. Otherwise return the offset in subdir that
1413 * can be used as relative path to dir.
e6636747 1414 */
9b125da4 1415int dir_inside_of(const char *subdir, const char *dir)
e6636747 1416{
9b125da4 1417 int offset = 0;
e6636747 1418
9b125da4 1419 assert(dir && subdir && *dir && *subdir);
e6636747 1420
9b125da4 1421 while (*dir && *subdir && *dir == *subdir) {
e6636747 1422 dir++;
9b125da4
NTND
1423 subdir++;
1424 offset++;
490544b1 1425 }
9b125da4
NTND
1426
1427 /* hel[p]/me vs hel[l]/yeah */
1428 if (*dir && *subdir)
1429 return -1;
1430
1431 if (!*subdir)
1432 return !*dir ? offset : -1; /* same dir */
1433
1434 /* foo/[b]ar vs foo/[] */
1435 if (is_dir_sep(dir[-1]))
1436 return is_dir_sep(subdir[-1]) ? offset : -1;
1437
1438 /* foo[/]bar vs foo[] */
1439 return is_dir_sep(*subdir) ? offset + 1 : -1;
e6636747
JS
1440}
1441
1442int is_inside_dir(const char *dir)
1443{
b892913d
NTND
1444 char cwd[PATH_MAX];
1445 if (!dir)
1446 return 0;
1447 if (!getcwd(cwd, sizeof(cwd)))
1448 die_errno("can't find the current directory");
1449 return dir_inside_of(cwd, dir) >= 0;
e6636747 1450}
7155b727 1451
55892d23
AP
1452int is_empty_dir(const char *path)
1453{
1454 DIR *dir = opendir(path);
1455 struct dirent *e;
1456 int ret = 1;
1457
1458 if (!dir)
1459 return 0;
1460
1461 while ((e = readdir(dir)) != NULL)
1462 if (!is_dot_or_dotdot(e->d_name)) {
1463 ret = 0;
1464 break;
1465 }
1466
1467 closedir(dir);
1468 return ret;
1469}
1470
ae2f203e 1471static int remove_dir_recurse(struct strbuf *path, int flag, int *kept_up)
7155b727 1472{
a0f4afbe 1473 DIR *dir;
7155b727 1474 struct dirent *e;
ae2f203e 1475 int ret = 0, original_len = path->len, len, kept_down = 0;
a0f4afbe 1476 int only_empty = (flag & REMOVE_DIR_EMPTY_ONLY);
c844a803 1477 int keep_toplevel = (flag & REMOVE_DIR_KEEP_TOPLEVEL);
a0f4afbe 1478 unsigned char submodule_head[20];
7155b727 1479
a0f4afbe 1480 if ((flag & REMOVE_DIR_KEEP_NESTED_GIT) &&
ae2f203e 1481 !resolve_gitlink_ref(path->buf, "HEAD", submodule_head)) {
a0f4afbe 1482 /* Do not descend and nuke a nested git work tree. */
ae2f203e
JH
1483 if (kept_up)
1484 *kept_up = 1;
a0f4afbe 1485 return 0;
ae2f203e 1486 }
a0f4afbe 1487
ae2f203e 1488 flag &= ~REMOVE_DIR_KEEP_TOPLEVEL;
a0f4afbe 1489 dir = opendir(path->buf);
c844a803 1490 if (!dir) {
ae2f203e 1491 /* an empty dir could be removed even if it is unreadble */
c844a803
JH
1492 if (!keep_toplevel)
1493 return rmdir(path->buf);
1494 else
1495 return -1;
1496 }
7155b727
JS
1497 if (path->buf[original_len - 1] != '/')
1498 strbuf_addch(path, '/');
1499
1500 len = path->len;
1501 while ((e = readdir(dir)) != NULL) {
1502 struct stat st;
8ca12c0d
AP
1503 if (is_dot_or_dotdot(e->d_name))
1504 continue;
7155b727
JS
1505
1506 strbuf_setlen(path, len);
1507 strbuf_addstr(path, e->d_name);
1508 if (lstat(path->buf, &st))
1509 ; /* fall thru */
1510 else if (S_ISDIR(st.st_mode)) {
ae2f203e 1511 if (!remove_dir_recurse(path, flag, &kept_down))
7155b727
JS
1512 continue; /* happy */
1513 } else if (!only_empty && !unlink(path->buf))
1514 continue; /* happy, too */
1515
1516 /* path too long, stat fails, or non-directory still exists */
1517 ret = -1;
1518 break;
1519 }
1520 closedir(dir);
1521
1522 strbuf_setlen(path, original_len);
ae2f203e 1523 if (!ret && !keep_toplevel && !kept_down)
7155b727 1524 ret = rmdir(path->buf);
ae2f203e
JH
1525 else if (kept_up)
1526 /*
1527 * report the uplevel that it is not an error that we
1528 * did not rmdir() our directory.
1529 */
1530 *kept_up = !ret;
7155b727
JS
1531 return ret;
1532}
039bc64e 1533
ae2f203e
JH
1534int remove_dir_recursively(struct strbuf *path, int flag)
1535{
1536 return remove_dir_recurse(path, flag, NULL);
1537}
1538
039bc64e
JH
1539void setup_standard_excludes(struct dir_struct *dir)
1540{
1541 const char *path;
dc79687e 1542 char *xdg_path;
039bc64e
JH
1543
1544 dir->exclude_per_dir = ".gitignore";
1545 path = git_path("info/exclude");
dc79687e
HKNN
1546 if (!excludes_file) {
1547 home_config_paths(NULL, &xdg_path, "ignore");
1548 excludes_file = xdg_path;
1549 }
4698c8fe 1550 if (!access_or_warn(path, R_OK, 0))
039bc64e 1551 add_excludes_from_file(dir, path);
4698c8fe 1552 if (excludes_file && !access_or_warn(excludes_file, R_OK, 0))
039bc64e
JH
1553 add_excludes_from_file(dir, excludes_file);
1554}
4a92d1bf
AR
1555
1556int remove_path(const char *name)
1557{
1558 char *slash;
1559
9a6728d4 1560 if (unlink(name) && errno != ENOENT && errno != ENOTDIR)
4a92d1bf
AR
1561 return -1;
1562
1563 slash = strrchr(name, '/');
1564 if (slash) {
1565 char *dirs = xstrdup(name);
1566 slash = dirs + (slash - name);
1567 do {
1568 *slash = '\0';
3fc0d131 1569 } while (rmdir(dirs) == 0 && (slash = strrchr(dirs, '/')));
4a92d1bf
AR
1570 free(dirs);
1571 }
1572 return 0;
1573}
1574
270be816
AS
1575/*
1576 * Frees memory within dir which was allocated for exclude lists and
1577 * the exclude_stack. Does not free dir itself.
1578 */
1579void clear_directory(struct dir_struct *dir)
1580{
1581 int i, j;
1582 struct exclude_list_group *group;
1583 struct exclude_list *el;
1584 struct exclude_stack *stk;
1585
1586 for (i = EXC_CMDL; i <= EXC_FILE; i++) {
1587 group = &dir->exclude_list_group[i];
1588 for (j = 0; j < group->nr; j++) {
1589 el = &group->el[j];
1590 if (i == EXC_DIRS)
1591 free((char *)el->src);
1592 clear_exclude_list(el);
1593 }
1594 free(group->el);
1595 }
1596
1597 stk = dir->exclude_stack;
1598 while (stk) {
1599 struct exclude_stack *prev = stk->prev;
1600 free(stk);
1601 stk = prev;
1602 }
1603}