Improve tree_entry_interesting() handling code
[git/git.git] / dir.c
CommitLineData
453ec4bd
LT
1/*
2 * This handles recursive filename detection with exclude
3 * files, index knowledge etc..
4 *
5 * Copyright (C) Linus Torvalds, 2005-2006
6 * Junio Hamano, 2005-2006
7 */
453ec4bd
LT
8#include "cache.h"
9#include "dir.h"
09595258 10#include "refs.h"
453ec4bd 11
9fc42d60
LT
12struct path_simplify {
13 int len;
14 const char *path;
15};
16
dba2e203 17static int read_directory_recursive(struct dir_struct *dir, const char *path, int len,
09595258 18 int check_only, const struct path_simplify *simplify);
caa6b782 19static int get_dtype(struct dirent *de, const char *path, int len);
09595258 20
8cf2a84e
JJ
21/* helper string functions with support for the ignore_case flag */
22int strcmp_icase(const char *a, const char *b)
23{
24 return ignore_case ? strcasecmp(a, b) : strcmp(a, b);
25}
26
27int strncmp_icase(const char *a, const char *b, size_t count)
28{
29 return ignore_case ? strncasecmp(a, b, count) : strncmp(a, b, count);
30}
31
32int fnmatch_icase(const char *pattern, const char *string, int flags)
33{
34 return fnmatch(pattern, string, flags | (ignore_case ? FNM_CASEFOLD : 0));
35}
36
1d8842d9 37static int common_prefix(const char **pathspec)
3c6a370b
LT
38{
39 const char *path, *slash, *next;
40 int prefix;
41
42 if (!pathspec)
43 return 0;
44
45 path = *pathspec;
46 slash = strrchr(path, '/');
47 if (!slash)
48 return 0;
49
42f9852f
JH
50 /*
51 * The first 'prefix' characters of 'path' are common leading
52 * path components among the pathspecs we have seen so far,
53 * including the trailing slash.
54 */
3c6a370b
LT
55 prefix = slash - path + 1;
56 while ((next = *++pathspec) != NULL) {
42f9852f
JH
57 int len, last_matching_slash = -1;
58 for (len = 0; len < prefix && next[len] == path[len]; len++)
59 if (next[len] == '/')
60 last_matching_slash = len;
61 if (len == prefix)
3c6a370b 62 continue;
42f9852f
JH
63 if (last_matching_slash < 0)
64 return 0;
65 prefix = last_matching_slash + 1;
3c6a370b
LT
66 }
67 return prefix;
68}
69
1d8842d9
LT
70int fill_directory(struct dir_struct *dir, const char **pathspec)
71{
dba2e203
LT
72 const char *path;
73 int len;
1d8842d9
LT
74
75 /*
76 * Calculate common prefix for the pathspec, and
77 * use that to optimize the directory walk
78 */
dba2e203 79 len = common_prefix(pathspec);
1d8842d9 80 path = "";
1d8842d9 81
dba2e203
LT
82 if (len)
83 path = xmemdupz(*pathspec, len);
1d8842d9
LT
84
85 /* Read the directory and prune it */
dba2e203
LT
86 read_directory(dir, path, len, pathspec);
87 return len;
1d8842d9
LT
88}
89
bc96cc87
NTND
90int within_depth(const char *name, int namelen,
91 int depth, int max_depth)
92{
93 const char *cp = name, *cpe = name + namelen;
94
95 while (cp < cpe) {
96 if (*cp++ != '/')
97 continue;
98 depth++;
99 if (depth > max_depth)
100 return 0;
101 }
102 return 1;
103}
104
e813d50e 105/*
2c5b0115 106 * Does 'match' match the given name?
e813d50e
JH
107 * A match is found if
108 *
109 * (1) the 'match' string is leading directory of 'name', or
110 * (2) the 'match' string is a wildcard and matches 'name', or
111 * (3) the 'match' string is exactly the same as 'name'.
112 *
113 * and the return value tells which case it was.
114 *
115 * It returns 0 when there is no match.
116 */
3c6a370b
LT
117static int match_one(const char *match, const char *name, int namelen)
118{
119 int matchlen;
120
121 /* If the match was just the prefix, we matched */
88ea8112 122 if (!*match)
e813d50e 123 return MATCHED_RECURSIVELY;
3c6a370b 124
21444f18
JJ
125 if (ignore_case) {
126 for (;;) {
127 unsigned char c1 = tolower(*match);
128 unsigned char c2 = tolower(*name);
129 if (c1 == '\0' || is_glob_special(c1))
130 break;
131 if (c1 != c2)
132 return 0;
133 match++;
134 name++;
135 namelen--;
136 }
137 } else {
138 for (;;) {
139 unsigned char c1 = *match;
140 unsigned char c2 = *name;
141 if (c1 == '\0' || is_glob_special(c1))
142 break;
143 if (c1 != c2)
144 return 0;
145 match++;
146 name++;
147 namelen--;
148 }
88ea8112
LT
149 }
150
151
3c6a370b
LT
152 /*
153 * If we don't match the matchstring exactly,
154 * we need to match by fnmatch
155 */
88ea8112 156 matchlen = strlen(match);
21444f18
JJ
157 if (strncmp_icase(match, name, matchlen))
158 return !fnmatch_icase(match, name, 0) ? MATCHED_FNMATCH : 0;
3c6a370b 159
f2d0df71 160 if (namelen == matchlen)
e813d50e
JH
161 return MATCHED_EXACTLY;
162 if (match[matchlen-1] == '/' || name[matchlen] == '/')
163 return MATCHED_RECURSIVELY;
164 return 0;
3c6a370b
LT
165}
166
e813d50e
JH
167/*
168 * Given a name and a list of pathspecs, see if the name matches
169 * any of the pathspecs. The caller is also interested in seeing
170 * all pathspec matches some names it calls this function with
171 * (otherwise the user could have mistyped the unmatched pathspec),
172 * and a mark is left in seen[] array for pathspec element that
173 * actually matched anything.
174 */
0b50922a
CB
175int match_pathspec(const char **pathspec, const char *name, int namelen,
176 int prefix, char *seen)
3c6a370b 177{
0b50922a
CB
178 int i, retval = 0;
179
180 if (!pathspec)
181 return 1;
3c6a370b
LT
182
183 name += prefix;
184 namelen -= prefix;
185
0b50922a 186 for (i = 0; pathspec[i] != NULL; i++) {
e813d50e 187 int how;
0b50922a
CB
188 const char *match = pathspec[i] + prefix;
189 if (seen && seen[i] == MATCHED_EXACTLY)
3c6a370b 190 continue;
e813d50e
JH
191 how = match_one(match, name, namelen);
192 if (how) {
193 if (retval < how)
194 retval = how;
0b50922a
CB
195 if (seen && seen[i] < how)
196 seen[i] = how;
3c6a370b
LT
197 }
198 }
199 return retval;
200}
201
61cf2820
NTND
202/*
203 * Does 'match' match the given name?
204 * A match is found if
205 *
206 * (1) the 'match' string is leading directory of 'name', or
207 * (2) the 'match' string is a wildcard and matches 'name', or
208 * (3) the 'match' string is exactly the same as 'name'.
209 *
210 * and the return value tells which case it was.
211 *
212 * It returns 0 when there is no match.
213 */
214static int match_pathspec_item(const struct pathspec_item *item, int prefix,
215 const char *name, int namelen)
216{
217 /* name/namelen has prefix cut off by caller */
218 const char *match = item->match + prefix;
219 int matchlen = item->len - prefix;
220
221 /* If the match was just the prefix, we matched */
222 if (!*match)
223 return MATCHED_RECURSIVELY;
224
225 if (matchlen <= namelen && !strncmp(match, name, matchlen)) {
226 if (matchlen == namelen)
227 return MATCHED_EXACTLY;
228
229 if (match[matchlen-1] == '/' || name[matchlen] == '/')
230 return MATCHED_RECURSIVELY;
231 }
232
233 if (item->has_wildcard && !fnmatch(match, name, 0))
234 return MATCHED_FNMATCH;
235
236 return 0;
237}
238
239/*
240 * Given a name and a list of pathspecs, see if the name matches
241 * any of the pathspecs. The caller is also interested in seeing
242 * all pathspec matches some names it calls this function with
243 * (otherwise the user could have mistyped the unmatched pathspec),
244 * and a mark is left in seen[] array for pathspec element that
245 * actually matched anything.
246 */
247int match_pathspec_depth(const struct pathspec *ps,
248 const char *name, int namelen,
249 int prefix, char *seen)
250{
251 int i, retval = 0;
252
253 if (!ps->nr) {
254 if (!ps->recursive || ps->max_depth == -1)
255 return MATCHED_RECURSIVELY;
256
257 if (within_depth(name, namelen, 0, ps->max_depth))
258 return MATCHED_EXACTLY;
259 else
260 return 0;
261 }
262
263 name += prefix;
264 namelen -= prefix;
265
266 for (i = ps->nr - 1; i >= 0; i--) {
267 int how;
268 if (seen && seen[i] == MATCHED_EXACTLY)
269 continue;
270 how = match_pathspec_item(ps->items+i, prefix, name, namelen);
271 if (ps->recursive && ps->max_depth != -1 &&
272 how && how != MATCHED_FNMATCH) {
273 int len = ps->items[i].len;
274 if (name[len] == '/')
275 len++;
276 if (within_depth(name+len, namelen-len, 0, ps->max_depth))
277 how = MATCHED_EXACTLY;
278 else
279 how = 0;
280 }
281 if (how) {
282 if (retval < how)
283 retval = how;
284 if (seen && seen[i] < how)
285 seen[i] = how;
286 }
287 }
288 return retval;
289}
290
68492fc7
LK
291static int no_wildcard(const char *string)
292{
dd482eea 293 return string[strcspn(string, "*?[{\\")] == '\0';
68492fc7
LK
294}
295
453ec4bd
LT
296void add_exclude(const char *string, const char *base,
297 int baselen, struct exclude_list *which)
298{
d6b8fc30
JH
299 struct exclude *x;
300 size_t len;
301 int to_exclude = 1;
302 int flags = 0;
453ec4bd 303
68492fc7 304 if (*string == '!') {
d6b8fc30 305 to_exclude = 0;
68492fc7
LK
306 string++;
307 }
d6b8fc30
JH
308 len = strlen(string);
309 if (len && string[len - 1] == '/') {
310 char *s;
311 x = xmalloc(sizeof(*x) + len);
4b25d091 312 s = (char *)(x+1);
d6b8fc30
JH
313 memcpy(s, string, len - 1);
314 s[len - 1] = '\0';
315 string = s;
316 x->pattern = s;
317 flags = EXC_FLAG_MUSTBEDIR;
318 } else {
319 x = xmalloc(sizeof(*x));
320 x->pattern = string;
321 }
322 x->to_exclude = to_exclude;
68492fc7 323 x->patternlen = strlen(string);
453ec4bd
LT
324 x->base = base;
325 x->baselen = baselen;
d6b8fc30 326 x->flags = flags;
68492fc7
LK
327 if (!strchr(string, '/'))
328 x->flags |= EXC_FLAG_NODIR;
329 if (no_wildcard(string))
330 x->flags |= EXC_FLAG_NOWILDCARD;
331 if (*string == '*' && no_wildcard(string+1))
332 x->flags |= EXC_FLAG_ENDSWITH;
686a4a06 333 ALLOC_GROW(which->excludes, which->nr + 1, which->alloc);
453ec4bd
LT
334 which->excludes[which->nr++] = x;
335}
336
c28b3d6e
NTND
337static void *read_skip_worktree_file_from_index(const char *path, size_t *size)
338{
339 int pos, len;
340 unsigned long sz;
341 enum object_type type;
342 void *data;
343 struct index_state *istate = &the_index;
344
345 len = strlen(path);
346 pos = index_name_pos(istate, path, len);
347 if (pos < 0)
348 return NULL;
349 if (!ce_skip_worktree(istate->cache[pos]))
350 return NULL;
351 data = read_sha1_file(istate->cache[pos]->sha1, &type, &sz);
352 if (!data || type != OBJ_BLOB) {
353 free(data);
354 return NULL;
355 }
356 *size = xsize_t(sz);
357 return data;
358}
359
0fd0e241
NTND
360void free_excludes(struct exclude_list *el)
361{
362 int i;
363
364 for (i = 0; i < el->nr; i++)
365 free(el->excludes[i]);
366 free(el->excludes);
367
368 el->nr = 0;
369 el->excludes = NULL;
370}
371
cb097534
NTND
372int add_excludes_from_file_to_list(const char *fname,
373 const char *base,
374 int baselen,
375 char **buf_p,
376 struct exclude_list *which,
377 int check_index)
453ec4bd 378{
c470701a 379 struct stat st;
453ec4bd 380 int fd, i;
9d14017a 381 size_t size = 0;
453ec4bd
LT
382 char *buf, *entry;
383
384 fd = open(fname, O_RDONLY);
c28b3d6e
NTND
385 if (fd < 0 || fstat(fd, &st) < 0) {
386 if (0 <= fd)
387 close(fd);
388 if (!check_index ||
389 (buf = read_skip_worktree_file_from_index(fname, &size)) == NULL)
390 return -1;
45d76f17
NTND
391 if (size == 0) {
392 free(buf);
393 return 0;
394 }
395 if (buf[size-1] != '\n') {
396 buf = xrealloc(buf, size+1);
397 buf[size++] = '\n';
398 }
453ec4bd 399 }
c28b3d6e
NTND
400 else {
401 size = xsize_t(st.st_size);
402 if (size == 0) {
403 close(fd);
404 return 0;
405 }
45d76f17 406 buf = xmalloc(size+1);
c28b3d6e 407 if (read_in_full(fd, buf, size) != size) {
45d76f17 408 free(buf);
c28b3d6e
NTND
409 close(fd);
410 return -1;
411 }
45d76f17 412 buf[size++] = '\n';
c28b3d6e 413 close(fd);
6ba78238 414 }
453ec4bd 415
63d285c8
JH
416 if (buf_p)
417 *buf_p = buf;
453ec4bd 418 entry = buf;
45d76f17
NTND
419 for (i = 0; i < size; i++) {
420 if (buf[i] == '\n') {
453ec4bd
LT
421 if (entry != buf + i && entry[0] != '#') {
422 buf[i - (i && buf[i-1] == '\r')] = 0;
423 add_exclude(entry, base, baselen, which);
424 }
425 entry = buf + i + 1;
426 }
427 }
428 return 0;
453ec4bd
LT
429}
430
431void add_excludes_from_file(struct dir_struct *dir, const char *fname)
432{
cb097534
NTND
433 if (add_excludes_from_file_to_list(fname, "", 0, NULL,
434 &dir->exclude_list[EXC_FILE], 0) < 0)
453ec4bd
LT
435 die("cannot use %s as an exclude file", fname);
436}
437
63d285c8 438static void prep_exclude(struct dir_struct *dir, const char *base, int baselen)
453ec4bd 439{
63d285c8
JH
440 struct exclude_list *el;
441 struct exclude_stack *stk = NULL;
442 int current;
443
444 if ((!dir->exclude_per_dir) ||
445 (baselen + strlen(dir->exclude_per_dir) >= PATH_MAX))
446 return; /* too long a path -- ignore */
447
448 /* Pop the ones that are not the prefix of the path being checked. */
449 el = &dir->exclude_list[EXC_DIRS];
450 while ((stk = dir->exclude_stack) != NULL) {
451 if (stk->baselen <= baselen &&
452 !strncmp(dir->basebuf, base, stk->baselen))
453 break;
454 dir->exclude_stack = stk->prev;
455 while (stk->exclude_ix < el->nr)
456 free(el->excludes[--el->nr]);
457 free(stk->filebuf);
458 free(stk);
453ec4bd 459 }
453ec4bd 460
63d285c8
JH
461 /* Read from the parent directories and push them down. */
462 current = stk ? stk->baselen : -1;
463 while (current < baselen) {
464 struct exclude_stack *stk = xcalloc(1, sizeof(*stk));
465 const char *cp;
453ec4bd 466
63d285c8
JH
467 if (current < 0) {
468 cp = base;
469 current = 0;
470 }
471 else {
472 cp = strchr(base + current + 1, '/');
473 if (!cp)
474 die("oops in prep_exclude");
475 cp++;
476 }
477 stk->prev = dir->exclude_stack;
478 stk->baselen = cp - base;
479 stk->exclude_ix = el->nr;
480 memcpy(dir->basebuf + current, base + current,
481 stk->baselen - current);
482 strcpy(dir->basebuf + stk->baselen, dir->exclude_per_dir);
cb097534
NTND
483 add_excludes_from_file_to_list(dir->basebuf,
484 dir->basebuf, stk->baselen,
485 &stk->filebuf, el, 1);
63d285c8
JH
486 dir->exclude_stack = stk;
487 current = stk->baselen;
488 }
489 dir->basebuf[baselen] = '\0';
453ec4bd
LT
490}
491
2c5b0115 492/* Scan the list and let the last match determine the fate.
453ec4bd
LT
493 * Return 1 for exclude, 0 for include and -1 for undecided.
494 */
cb097534
NTND
495int excluded_from_list(const char *pathname,
496 int pathlen, const char *basename, int *dtype,
497 struct exclude_list *el)
453ec4bd
LT
498{
499 int i;
500
501 if (el->nr) {
502 for (i = el->nr - 1; 0 <= i; i--) {
503 struct exclude *x = el->excludes[i];
504 const char *exclude = x->pattern;
68492fc7 505 int to_exclude = x->to_exclude;
453ec4bd 506
6831a88a
JH
507 if (x->flags & EXC_FLAG_MUSTBEDIR) {
508 if (*dtype == DT_UNKNOWN)
caa6b782 509 *dtype = get_dtype(NULL, pathname, pathlen);
6831a88a
JH
510 if (*dtype != DT_DIR)
511 continue;
512 }
d6b8fc30 513
68492fc7 514 if (x->flags & EXC_FLAG_NODIR) {
453ec4bd 515 /* match basename */
68492fc7 516 if (x->flags & EXC_FLAG_NOWILDCARD) {
10d4b02b 517 if (!strcmp_icase(exclude, basename))
68492fc7
LK
518 return to_exclude;
519 } else if (x->flags & EXC_FLAG_ENDSWITH) {
520 if (x->patternlen - 1 <= pathlen &&
10d4b02b 521 !strcmp_icase(exclude + 1, pathname + pathlen - x->patternlen + 1))
68492fc7
LK
522 return to_exclude;
523 } else {
10d4b02b 524 if (fnmatch_icase(exclude, basename, 0) == 0)
68492fc7
LK
525 return to_exclude;
526 }
453ec4bd
LT
527 }
528 else {
529 /* match with FNM_PATHNAME:
530 * exclude has base (baselen long) implicitly
531 * in front of it.
532 */
533 int baselen = x->baselen;
534 if (*exclude == '/')
535 exclude++;
536
537 if (pathlen < baselen ||
538 (baselen && pathname[baselen-1] != '/') ||
10d4b02b 539 strncmp_icase(pathname, x->base, baselen))
453ec4bd
LT
540 continue;
541
68492fc7 542 if (x->flags & EXC_FLAG_NOWILDCARD) {
10d4b02b 543 if (!strcmp_icase(exclude, pathname + baselen))
68492fc7
LK
544 return to_exclude;
545 } else {
10d4b02b 546 if (fnmatch_icase(exclude, pathname+baselen,
68492fc7
LK
547 FNM_PATHNAME) == 0)
548 return to_exclude;
549 }
453ec4bd
LT
550 }
551 }
552 }
553 return -1; /* undecided */
554}
555
6831a88a 556int excluded(struct dir_struct *dir, const char *pathname, int *dtype_p)
453ec4bd
LT
557{
558 int pathlen = strlen(pathname);
559 int st;
68492fc7
LK
560 const char *basename = strrchr(pathname, '/');
561 basename = (basename) ? basename+1 : pathname;
453ec4bd 562
63d285c8 563 prep_exclude(dir, pathname, basename-pathname);
453ec4bd 564 for (st = EXC_CMDL; st <= EXC_FILE; st++) {
cb097534
NTND
565 switch (excluded_from_list(pathname, pathlen, basename,
566 dtype_p, &dir->exclude_list[st])) {
453ec4bd
LT
567 case 0:
568 return 0;
569 case 1:
570 return 1;
571 }
572 }
573 return 0;
574}
575
f3fa1838
JH
576static struct dir_entry *dir_entry_new(const char *pathname, int len)
577{
453ec4bd
LT
578 struct dir_entry *ent;
579
453ec4bd
LT
580 ent = xmalloc(sizeof(*ent) + len + 1);
581 ent->len = len;
582 memcpy(ent->name, pathname, len);
583 ent->name[len] = 0;
4d06f8ac 584 return ent;
453ec4bd
LT
585}
586
159b3212 587static struct dir_entry *dir_add_name(struct dir_struct *dir, const char *pathname, int len)
6815e569 588{
0a9b88b7 589 if (cache_name_exists(pathname, len, ignore_case))
6815e569
JK
590 return NULL;
591
25fd2f7a 592 ALLOC_GROW(dir->entries, dir->nr+1, dir->alloc);
6815e569
JK
593 return dir->entries[dir->nr++] = dir_entry_new(pathname, len);
594}
595
108da0db 596struct dir_entry *dir_add_ignored(struct dir_struct *dir, const char *pathname, int len)
2abd31b0 597{
6e4f981f 598 if (!cache_name_is_other(pathname, len))
2abd31b0
JK
599 return NULL;
600
25fd2f7a 601 ALLOC_GROW(dir->ignored, dir->ignored_nr+1, dir->ignored_alloc);
2abd31b0
JK
602 return dir->ignored[dir->ignored_nr++] = dir_entry_new(pathname, len);
603}
604
09595258
LT
605enum exist_status {
606 index_nonexistent = 0,
607 index_directory,
4b05548f 608 index_gitdir
09595258
LT
609};
610
5102c617
JJ
611/*
612 * Do not use the alphabetically stored index to look up
613 * the directory name; instead, use the case insensitive
614 * name hash.
615 */
616static enum exist_status directory_exists_in_index_icase(const char *dirname, int len)
617{
618 struct cache_entry *ce = index_name_exists(&the_index, dirname, len + 1, ignore_case);
619 unsigned char endchar;
620
621 if (!ce)
622 return index_nonexistent;
623 endchar = ce->name[len];
624
625 /*
626 * The cache_entry structure returned will contain this dirname
627 * and possibly additional path components.
628 */
629 if (endchar == '/')
630 return index_directory;
631
632 /*
633 * If there are no additional path components, then this cache_entry
634 * represents a submodule. Submodules, despite being directories,
635 * are stored in the cache without a closing slash.
636 */
637 if (!endchar && S_ISGITLINK(ce->ce_mode))
638 return index_gitdir;
639
640 /* This should never be hit, but it exists just in case. */
641 return index_nonexistent;
642}
643
09595258
LT
644/*
645 * The index sorts alphabetically by entry name, which
646 * means that a gitlink sorts as '\0' at the end, while
647 * a directory (which is defined not as an entry, but as
648 * the files it contains) will sort with the '/' at the
649 * end.
650 */
651static enum exist_status directory_exists_in_index(const char *dirname, int len)
453ec4bd 652{
5102c617
JJ
653 int pos;
654
655 if (ignore_case)
656 return directory_exists_in_index_icase(dirname, len);
657
658 pos = cache_name_pos(dirname, len);
09595258
LT
659 if (pos < 0)
660 pos = -pos-1;
661 while (pos < active_nr) {
662 struct cache_entry *ce = active_cache[pos++];
663 unsigned char endchar;
664
665 if (strncmp(ce->name, dirname, len))
666 break;
667 endchar = ce->name[len];
668 if (endchar > '/')
669 break;
670 if (endchar == '/')
671 return index_directory;
7a51ed66 672 if (!endchar && S_ISGITLINK(ce->ce_mode))
09595258
LT
673 return index_gitdir;
674 }
675 return index_nonexistent;
676}
677
678/*
679 * When we find a directory when traversing the filesystem, we
680 * have three distinct cases:
681 *
682 * - ignore it
683 * - see it as a directory
684 * - recurse into it
685 *
686 * and which one we choose depends on a combination of existing
687 * git index contents and the flags passed into the directory
688 * traversal routine.
689 *
690 * Case 1: If we *already* have entries in the index under that
691 * directory name, we always recurse into the directory to see
692 * all the files.
693 *
694 * Case 2: If we *already* have that directory name as a gitlink,
695 * we always continue to see it as a gitlink, regardless of whether
696 * there is an actual git directory there or not (it might not
697 * be checked out as a subproject!)
698 *
699 * Case 3: if we didn't have it in the index previously, we
700 * have a few sub-cases:
701 *
702 * (a) if "show_other_directories" is true, we show it as
703 * just a directory, unless "hide_empty_directories" is
704 * also true and the directory is empty, in which case
705 * we just ignore it entirely.
706 * (b) if it looks like a git directory, and we don't have
302b9282 707 * 'no_gitlinks' set we treat it as a gitlink, and show it
09595258
LT
708 * as a directory.
709 * (c) otherwise, we recurse into it.
710 */
711enum directory_treatment {
712 show_directory,
713 ignore_directory,
4b05548f 714 recurse_into_directory
09595258
LT
715};
716
717static enum directory_treatment treat_directory(struct dir_struct *dir,
718 const char *dirname, int len,
719 const struct path_simplify *simplify)
720{
721 /* The "len-1" is to strip the final '/' */
722 switch (directory_exists_in_index(dirname, len-1)) {
723 case index_directory:
724 return recurse_into_directory;
725
726 case index_gitdir:
7c4c97c0 727 if (dir->flags & DIR_SHOW_OTHER_DIRECTORIES)
ab22aed3 728 return ignore_directory;
09595258
LT
729 return show_directory;
730
731 case index_nonexistent:
7c4c97c0 732 if (dir->flags & DIR_SHOW_OTHER_DIRECTORIES)
09595258 733 break;
7c4c97c0 734 if (!(dir->flags & DIR_NO_GITLINKS)) {
09595258
LT
735 unsigned char sha1[20];
736 if (resolve_gitlink_ref(dirname, "HEAD", sha1) == 0)
737 return show_directory;
738 }
739 return recurse_into_directory;
740 }
741
742 /* This is the "show_other_directories" case */
7c4c97c0 743 if (!(dir->flags & DIR_HIDE_EMPTY_DIRECTORIES))
09595258 744 return show_directory;
dba2e203 745 if (!read_directory_recursive(dir, dirname, len, 1, simplify))
09595258
LT
746 return ignore_directory;
747 return show_directory;
453ec4bd
LT
748}
749
9fc42d60
LT
750/*
751 * This is an inexact early pruning of any recursive directory
752 * reading - if the path cannot possibly be in the pathspec,
753 * return true, and we'll skip it early.
754 */
755static int simplify_away(const char *path, int pathlen, const struct path_simplify *simplify)
756{
757 if (simplify) {
758 for (;;) {
759 const char *match = simplify->path;
760 int len = simplify->len;
761
762 if (!match)
763 break;
764 if (len > pathlen)
765 len = pathlen;
766 if (!memcmp(path, match, len))
767 return 0;
768 simplify++;
769 }
770 return 1;
771 }
772 return 0;
773}
774
29209cbe
JK
775/*
776 * This function tells us whether an excluded path matches a
777 * list of "interesting" pathspecs. That is, whether a path matched
778 * by any of the pathspecs could possibly be ignored by excluding
779 * the specified path. This can happen if:
780 *
781 * 1. the path is mentioned explicitly in the pathspec
782 *
783 * 2. the path is a directory prefix of some element in the
784 * pathspec
785 */
786static int exclude_matches_pathspec(const char *path, int len,
787 const struct path_simplify *simplify)
e96980ef
JK
788{
789 if (simplify) {
790 for (; simplify->path; simplify++) {
791 if (len == simplify->len
792 && !memcmp(path, simplify->path, len))
793 return 1;
29209cbe
JK
794 if (len < simplify->len
795 && simplify->path[len] == '/'
796 && !memcmp(path, simplify->path, len))
797 return 1;
e96980ef
JK
798 }
799 }
800 return 0;
801}
802
443e061a
LT
803static int get_index_dtype(const char *path, int len)
804{
805 int pos;
806 struct cache_entry *ce;
807
808 ce = cache_name_exists(path, len, 0);
809 if (ce) {
810 if (!ce_uptodate(ce))
811 return DT_UNKNOWN;
812 if (S_ISGITLINK(ce->ce_mode))
813 return DT_DIR;
814 /*
815 * Nobody actually cares about the
816 * difference between DT_LNK and DT_REG
817 */
818 return DT_REG;
819 }
820
821 /* Try to look it up as a directory */
822 pos = cache_name_pos(path, len);
823 if (pos >= 0)
824 return DT_UNKNOWN;
825 pos = -pos-1;
826 while (pos < active_nr) {
827 ce = active_cache[pos++];
828 if (strncmp(ce->name, path, len))
829 break;
830 if (ce->name[len] > '/')
831 break;
832 if (ce->name[len] < '/')
833 continue;
834 if (!ce_uptodate(ce))
835 break; /* continue? */
836 return DT_DIR;
837 }
838 return DT_UNKNOWN;
839}
840
caa6b782 841static int get_dtype(struct dirent *de, const char *path, int len)
07134421 842{
6831a88a 843 int dtype = de ? DTYPE(de) : DT_UNKNOWN;
07134421
LT
844 struct stat st;
845
846 if (dtype != DT_UNKNOWN)
847 return dtype;
443e061a
LT
848 dtype = get_index_dtype(path, len);
849 if (dtype != DT_UNKNOWN)
850 return dtype;
851 if (lstat(path, &st))
07134421
LT
852 return dtype;
853 if (S_ISREG(st.st_mode))
854 return DT_REG;
855 if (S_ISDIR(st.st_mode))
856 return DT_DIR;
857 if (S_ISLNK(st.st_mode))
858 return DT_LNK;
859 return dtype;
860}
861
53cc5356
JH
862enum path_treatment {
863 path_ignored,
864 path_handled,
4b05548f 865 path_recurse
53cc5356
JH
866};
867
16e2cfa9
JH
868static enum path_treatment treat_one_path(struct dir_struct *dir,
869 char *path, int *len,
870 const struct path_simplify *simplify,
871 int dtype, struct dirent *de)
53cc5356 872{
16e2cfa9 873 int exclude = excluded(dir, path, &dtype);
53cc5356 874 if (exclude && (dir->flags & DIR_COLLECT_IGNORED)
29209cbe 875 && exclude_matches_pathspec(path, *len, simplify))
53cc5356
JH
876 dir_add_ignored(dir, path, *len);
877
878 /*
879 * Excluded? If we don't explicitly want to show
880 * ignored files, ignore it
881 */
882 if (exclude && !(dir->flags & DIR_SHOW_IGNORED))
883 return path_ignored;
884
885 if (dtype == DT_UNKNOWN)
886 dtype = get_dtype(de, path, *len);
887
888 /*
889 * Do we want to see just the ignored files?
890 * We still need to recurse into directories,
891 * even if we don't ignore them, since the
892 * directory may contain files that we do..
893 */
894 if (!exclude && (dir->flags & DIR_SHOW_IGNORED)) {
895 if (dtype != DT_DIR)
896 return path_ignored;
897 }
898
899 switch (dtype) {
900 default:
901 return path_ignored;
902 case DT_DIR:
903 memcpy(path + *len, "/", 2);
904 (*len)++;
905 switch (treat_directory(dir, path, *len, simplify)) {
906 case show_directory:
907 if (exclude != !!(dir->flags
908 & DIR_SHOW_IGNORED))
909 return path_ignored;
910 break;
911 case recurse_into_directory:
912 return path_recurse;
913 case ignore_directory:
914 return path_ignored;
915 }
916 break;
917 case DT_REG:
918 case DT_LNK:
919 break;
920 }
921 return path_handled;
922}
923
16e2cfa9
JH
924static enum path_treatment treat_path(struct dir_struct *dir,
925 struct dirent *de,
926 char *path, int path_max,
927 int baselen,
928 const struct path_simplify *simplify,
929 int *len)
930{
931 int dtype;
932
933 if (is_dot_or_dotdot(de->d_name) || !strcmp(de->d_name, ".git"))
934 return path_ignored;
935 *len = strlen(de->d_name);
936 /* Ignore overly long pathnames! */
937 if (*len + baselen + 8 > path_max)
938 return path_ignored;
939 memcpy(path + baselen, de->d_name, *len + 1);
940 *len += baselen;
941 if (simplify_away(path, *len, simplify))
942 return path_ignored;
943
944 dtype = DTYPE(de);
945 return treat_one_path(dir, path, len, simplify, dtype, de);
946}
947
453ec4bd
LT
948/*
949 * Read a directory tree. We currently ignore anything but
950 * directories, regular files and symlinks. That's because git
951 * doesn't handle them at all yet. Maybe that will change some
952 * day.
953 *
954 * Also, we ignore the name ".git" (even if it is not a directory).
955 * That likely will not change.
956 */
53cc5356
JH
957static int read_directory_recursive(struct dir_struct *dir,
958 const char *base, int baselen,
959 int check_only,
960 const struct path_simplify *simplify)
453ec4bd 961{
dba2e203 962 DIR *fdir = opendir(*base ? base : ".");
453ec4bd
LT
963 int contents = 0;
964
965 if (fdir) {
453ec4bd 966 struct dirent *de;
dba2e203
LT
967 char path[PATH_MAX + 1];
968 memcpy(path, base, baselen);
453ec4bd 969
453ec4bd 970 while ((de = readdir(fdir)) != NULL) {
53cc5356
JH
971 int len;
972 switch (treat_path(dir, de, path, sizeof(path),
973 baselen, simplify, &len)) {
974 case path_recurse:
975 contents += read_directory_recursive
976 (dir, path, len, 0, simplify);
453ec4bd 977 continue;
53cc5356 978 case path_ignored:
5d5cea67 979 continue;
53cc5356 980 case path_handled:
453ec4bd
LT
981 break;
982 }
453ec4bd 983 contents++;
07ccbff8
JS
984 if (check_only)
985 goto exit_early;
986 else
dba2e203 987 dir_add_name(dir, path, len);
453ec4bd 988 }
07ccbff8 989exit_early:
453ec4bd 990 closedir(fdir);
453ec4bd
LT
991 }
992
993 return contents;
994}
995
996static int cmp_name(const void *p1, const void *p2)
997{
998 const struct dir_entry *e1 = *(const struct dir_entry **)p1;
999 const struct dir_entry *e2 = *(const struct dir_entry **)p2;
1000
1001 return cache_name_compare(e1->name, e1->len,
1002 e2->name, e2->len);
1003}
1004
9fc42d60
LT
1005/*
1006 * Return the length of the "simple" part of a path match limiter.
1007 */
1008static int simple_length(const char *match)
453ec4bd 1009{
9fc42d60
LT
1010 int len = -1;
1011
1012 for (;;) {
1013 unsigned char c = *match++;
1014 len++;
8cc32992 1015 if (c == '\0' || is_glob_special(c))
9fc42d60
LT
1016 return len;
1017 }
1018}
1019
1020static struct path_simplify *create_simplify(const char **pathspec)
1021{
1022 int nr, alloc = 0;
1023 struct path_simplify *simplify = NULL;
1024
1025 if (!pathspec)
1026 return NULL;
1027
1028 for (nr = 0 ; ; nr++) {
1029 const char *match;
1030 if (nr >= alloc) {
1031 alloc = alloc_nr(alloc);
1032 simplify = xrealloc(simplify, alloc * sizeof(*simplify));
1033 }
1034 match = *pathspec++;
1035 if (!match)
1036 break;
1037 simplify[nr].path = match;
1038 simplify[nr].len = simple_length(match);
1039 }
1040 simplify[nr].path = NULL;
1041 simplify[nr].len = 0;
1042 return simplify;
1043}
1044
1045static void free_simplify(struct path_simplify *simplify)
1046{
8e0f7003 1047 free(simplify);
9fc42d60
LT
1048}
1049
48ffef96
JH
1050static int treat_leading_path(struct dir_struct *dir,
1051 const char *path, int len,
1052 const struct path_simplify *simplify)
1053{
1054 char pathbuf[PATH_MAX];
1055 int baselen, blen;
1056 const char *cp;
1057
1058 while (len && path[len - 1] == '/')
1059 len--;
1060 if (!len)
1061 return 1;
1062 baselen = 0;
1063 while (1) {
1064 cp = path + baselen + !!baselen;
1065 cp = memchr(cp, '/', path + len - cp);
1066 if (!cp)
1067 baselen = len;
1068 else
1069 baselen = cp - path;
1070 memcpy(pathbuf, path, baselen);
1071 pathbuf[baselen] = '\0';
1072 if (!is_directory(pathbuf))
1073 return 0;
1074 if (simplify_away(pathbuf, baselen, simplify))
1075 return 0;
1076 blen = baselen;
1077 if (treat_one_path(dir, pathbuf, &blen, simplify,
1078 DT_DIR, NULL) == path_ignored)
1079 return 0; /* do not recurse into it */
1080 if (len <= baselen)
1081 return 1; /* finished checking */
1082 }
1083}
1084
dba2e203 1085int read_directory(struct dir_struct *dir, const char *path, int len, const char **pathspec)
9fc42d60 1086{
725b0605 1087 struct path_simplify *simplify;
b4189aa8 1088
dba2e203 1089 if (has_symlink_leading_path(path, len))
725b0605
JH
1090 return dir->nr;
1091
1092 simplify = create_simplify(pathspec);
48ffef96
JH
1093 if (!len || treat_leading_path(dir, path, len, simplify))
1094 read_directory_recursive(dir, path, len, 0, simplify);
9fc42d60 1095 free_simplify(simplify);
453ec4bd 1096 qsort(dir->entries, dir->nr, sizeof(struct dir_entry *), cmp_name);
2abd31b0 1097 qsort(dir->ignored, dir->ignored_nr, sizeof(struct dir_entry *), cmp_name);
453ec4bd
LT
1098 return dir->nr;
1099}
c91f0d92 1100
686a4a06 1101int file_exists(const char *f)
c91f0d92 1102{
686a4a06 1103 struct stat sb;
a50f9fc5 1104 return lstat(f, &sb) == 0;
c91f0d92 1105}
e6636747
JS
1106
1107/*
1108 * get_relative_cwd() gets the prefix of the current working directory
1109 * relative to 'dir'. If we are not inside 'dir', it returns NULL.
420acb31
JS
1110 *
1111 * As a convenience, it also returns NULL if 'dir' is already NULL. The
1112 * reason for this behaviour is that it is natural for functions returning
1113 * directory names to return NULL to say "this directory does not exist"
1114 * or "this directory is invalid". These cases are usually handled the
1115 * same as if the cwd is not inside 'dir' at all, so get_relative_cwd()
1116 * returns NULL for both of them.
1117 *
1118 * Most notably, get_relative_cwd(buffer, size, get_git_work_tree())
1119 * unifies the handling of "outside work tree" with "no work tree at all".
e6636747
JS
1120 */
1121char *get_relative_cwd(char *buffer, int size, const char *dir)
1122{
1123 char *cwd = buffer;
1124
e6636747
JS
1125 if (!dir)
1126 return NULL;
1127 if (!getcwd(buffer, size))
d824cbba 1128 die_errno("can't find the current directory");
e6636747
JS
1129
1130 if (!is_absolute_path(dir))
1131 dir = make_absolute_path(dir);
1132
1133 while (*dir && *dir == *cwd) {
1134 dir++;
1135 cwd++;
1136 }
1137 if (*dir)
1138 return NULL;
490544b1
CB
1139 switch (*cwd) {
1140 case '\0':
1141 return cwd;
1142 case '/':
e6636747 1143 return cwd + 1;
490544b1 1144 default:
fbbb4e19
NTND
1145 /*
1146 * dir can end with a path separator when it's root
1147 * directory. Return proper prefix in that case.
1148 */
1149 if (dir[-1] == '/')
1150 return cwd;
490544b1
CB
1151 return NULL;
1152 }
e6636747
JS
1153}
1154
1155int is_inside_dir(const char *dir)
1156{
1157 char buffer[PATH_MAX];
1158 return get_relative_cwd(buffer, sizeof(buffer), dir) != NULL;
1159}
7155b727 1160
55892d23
AP
1161int is_empty_dir(const char *path)
1162{
1163 DIR *dir = opendir(path);
1164 struct dirent *e;
1165 int ret = 1;
1166
1167 if (!dir)
1168 return 0;
1169
1170 while ((e = readdir(dir)) != NULL)
1171 if (!is_dot_or_dotdot(e->d_name)) {
1172 ret = 0;
1173 break;
1174 }
1175
1176 closedir(dir);
1177 return ret;
1178}
1179
a0f4afbe 1180int remove_dir_recursively(struct strbuf *path, int flag)
7155b727 1181{
a0f4afbe 1182 DIR *dir;
7155b727
JS
1183 struct dirent *e;
1184 int ret = 0, original_len = path->len, len;
a0f4afbe
JH
1185 int only_empty = (flag & REMOVE_DIR_EMPTY_ONLY);
1186 unsigned char submodule_head[20];
7155b727 1187
a0f4afbe
JH
1188 if ((flag & REMOVE_DIR_KEEP_NESTED_GIT) &&
1189 !resolve_gitlink_ref(path->buf, "HEAD", submodule_head))
1190 /* Do not descend and nuke a nested git work tree. */
1191 return 0;
1192
1193 dir = opendir(path->buf);
7155b727
JS
1194 if (!dir)
1195 return -1;
1196 if (path->buf[original_len - 1] != '/')
1197 strbuf_addch(path, '/');
1198
1199 len = path->len;
1200 while ((e = readdir(dir)) != NULL) {
1201 struct stat st;
8ca12c0d
AP
1202 if (is_dot_or_dotdot(e->d_name))
1203 continue;
7155b727
JS
1204
1205 strbuf_setlen(path, len);
1206 strbuf_addstr(path, e->d_name);
1207 if (lstat(path->buf, &st))
1208 ; /* fall thru */
1209 else if (S_ISDIR(st.st_mode)) {
1210 if (!remove_dir_recursively(path, only_empty))
1211 continue; /* happy */
1212 } else if (!only_empty && !unlink(path->buf))
1213 continue; /* happy, too */
1214
1215 /* path too long, stat fails, or non-directory still exists */
1216 ret = -1;
1217 break;
1218 }
1219 closedir(dir);
1220
1221 strbuf_setlen(path, original_len);
1222 if (!ret)
1223 ret = rmdir(path->buf);
1224 return ret;
1225}
039bc64e
JH
1226
1227void setup_standard_excludes(struct dir_struct *dir)
1228{
1229 const char *path;
1230
1231 dir->exclude_per_dir = ".gitignore";
1232 path = git_path("info/exclude");
1233 if (!access(path, R_OK))
1234 add_excludes_from_file(dir, path);
1235 if (excludes_file && !access(excludes_file, R_OK))
1236 add_excludes_from_file(dir, excludes_file);
1237}
4a92d1bf
AR
1238
1239int remove_path(const char *name)
1240{
1241 char *slash;
1242
1243 if (unlink(name) && errno != ENOENT)
1244 return -1;
1245
1246 slash = strrchr(name, '/');
1247 if (slash) {
1248 char *dirs = xstrdup(name);
1249 slash = dirs + (slash - name);
1250 do {
1251 *slash = '\0';
3fc0d131 1252 } while (rmdir(dirs) == 0 && (slash = strrchr(dirs, '/')));
4a92d1bf
AR
1253 free(dirs);
1254 }
1255 return 0;
1256}
1257
86e4ca69
NTND
1258static int pathspec_item_cmp(const void *a_, const void *b_)
1259{
1260 struct pathspec_item *a, *b;
1261
1262 a = (struct pathspec_item *)a_;
1263 b = (struct pathspec_item *)b_;
1264 return strcmp(a->match, b->match);
1265}
1266
0602f3e9
NTND
1267int init_pathspec(struct pathspec *pathspec, const char **paths)
1268{
1269 const char **p = paths;
1270 int i;
1271
1272 memset(pathspec, 0, sizeof(*pathspec));
1273 if (!p)
1274 return 0;
1275 while (*p)
1276 p++;
1277 pathspec->raw = paths;
1278 pathspec->nr = p - paths;
1279 if (!pathspec->nr)
1280 return 0;
1281
1282 pathspec->items = xmalloc(sizeof(struct pathspec_item)*pathspec->nr);
1283 for (i = 0; i < pathspec->nr; i++) {
1284 struct pathspec_item *item = pathspec->items+i;
1285 const char *path = paths[i];
1286
1287 item->match = path;
1288 item->len = strlen(path);
d38f2809
NTND
1289 item->has_wildcard = !no_wildcard(path);
1290 if (item->has_wildcard)
1291 pathspec->has_wildcard = 1;
0602f3e9 1292 }
86e4ca69
NTND
1293
1294 qsort(pathspec->items, pathspec->nr,
1295 sizeof(struct pathspec_item), pathspec_item_cmp);
1296
0602f3e9
NTND
1297 return 0;
1298}
1299
1300void free_pathspec(struct pathspec *pathspec)
1301{
1302 free(pathspec->items);
1303 pathspec->items = NULL;
1304}