]>
Commit | Line | Data |
---|---|---|
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" |
453ec4bd | 14 | |
9fc42d60 LT |
15 | struct path_simplify { |
16 | int len; | |
17 | const char *path; | |
18 | }; | |
19 | ||
dba2e203 | 20 | static int read_directory_recursive(struct dir_struct *dir, const char *path, int len, |
09595258 | 21 | int check_only, const struct path_simplify *simplify); |
caa6b782 | 22 | static int get_dtype(struct dirent *de, const char *path, int len); |
09595258 | 23 | |
8cf2a84e JJ |
24 | /* helper string functions with support for the ignore_case flag */ |
25 | int strcmp_icase(const char *a, const char *b) | |
26 | { | |
27 | return ignore_case ? strcasecmp(a, b) : strcmp(a, b); | |
28 | } | |
29 | ||
30 | int strncmp_icase(const char *a, const char *b, size_t count) | |
31 | { | |
32 | return ignore_case ? strncasecmp(a, b, count) : strncmp(a, b, count); | |
33 | } | |
34 | ||
35 | int fnmatch_icase(const char *pattern, const char *string, int flags) | |
36 | { | |
37 | return fnmatch(pattern, string, flags | (ignore_case ? FNM_CASEFOLD : 0)); | |
38 | } | |
39 | ||
5d74762d NTND |
40 | inline int git_fnmatch(const char *pattern, const char *string, |
41 | int flags, int prefix) | |
42 | { | |
43 | int fnm_flags = 0; | |
44 | if (flags & GFNM_PATHNAME) | |
45 | fnm_flags |= FNM_PATHNAME; | |
46 | if (prefix > 0) { | |
47 | if (strncmp(pattern, string, prefix)) | |
48 | return FNM_NOMATCH; | |
49 | pattern += prefix; | |
50 | string += prefix; | |
51 | } | |
8c6abbcd NTND |
52 | if (flags & GFNM_ONESTAR) { |
53 | int pattern_len = strlen(++pattern); | |
54 | int string_len = strlen(string); | |
55 | return string_len < pattern_len || | |
56 | strcmp(pattern, | |
57 | string + string_len - pattern_len); | |
58 | } | |
5d74762d NTND |
59 | return fnmatch(pattern, string, fnm_flags); |
60 | } | |
61 | ||
0b6e56df JH |
62 | static int fnmatch_icase_mem(const char *pattern, int patternlen, |
63 | const char *string, int stringlen, | |
64 | int flags) | |
65 | { | |
66 | int match_status; | |
67 | struct strbuf pat_buf = STRBUF_INIT; | |
68 | struct strbuf str_buf = STRBUF_INIT; | |
69 | const char *use_pat = pattern; | |
70 | const char *use_str = string; | |
71 | ||
72 | if (pattern[patternlen]) { | |
73 | strbuf_add(&pat_buf, pattern, patternlen); | |
74 | use_pat = pat_buf.buf; | |
75 | } | |
76 | if (string[stringlen]) { | |
77 | strbuf_add(&str_buf, string, stringlen); | |
78 | use_str = str_buf.buf; | |
79 | } | |
80 | ||
f30366b2 JH |
81 | if (ignore_case) |
82 | flags |= WM_CASEFOLD; | |
83 | match_status = wildmatch(use_pat, use_str, flags, NULL); | |
0b6e56df JH |
84 | |
85 | strbuf_release(&pat_buf); | |
86 | strbuf_release(&str_buf); | |
87 | ||
88 | return match_status; | |
89 | } | |
90 | ||
f950eb95 | 91 | static size_t common_prefix_len(const char **pathspec) |
3c6a370b | 92 | { |
4a085b16 JH |
93 | const char *n, *first; |
94 | size_t max = 0; | |
823ab40f | 95 | int literal = limit_pathspec_to_literal(); |
3c6a370b LT |
96 | |
97 | if (!pathspec) | |
4a085b16 JH |
98 | return max; |
99 | ||
100 | first = *pathspec; | |
101 | while ((n = *pathspec++)) { | |
102 | size_t i, len = 0; | |
103 | for (i = 0; first == n || i < max; i++) { | |
104 | char c = n[i]; | |
823ab40f | 105 | if (!c || c != first[i] || (!literal && is_glob_special(c))) |
4a085b16 JH |
106 | break; |
107 | if (c == '/') | |
108 | len = i + 1; | |
109 | } | |
110 | if (first == n || len < max) { | |
111 | max = len; | |
112 | if (!max) | |
113 | break; | |
114 | } | |
3c6a370b | 115 | } |
4a085b16 | 116 | return max; |
3c6a370b LT |
117 | } |
118 | ||
f950eb95 CB |
119 | /* |
120 | * Returns a copy of the longest leading path common among all | |
121 | * pathspecs. | |
122 | */ | |
123 | char *common_prefix(const char **pathspec) | |
124 | { | |
125 | unsigned long len = common_prefix_len(pathspec); | |
126 | ||
127 | return len ? xmemdupz(*pathspec, len) : NULL; | |
128 | } | |
129 | ||
1d8842d9 LT |
130 | int fill_directory(struct dir_struct *dir, const char **pathspec) |
131 | { | |
4a085b16 | 132 | size_t len; |
1d8842d9 LT |
133 | |
134 | /* | |
135 | * Calculate common prefix for the pathspec, and | |
136 | * use that to optimize the directory walk | |
137 | */ | |
4a085b16 | 138 | len = common_prefix_len(pathspec); |
1d8842d9 LT |
139 | |
140 | /* Read the directory and prune it */ | |
2b189435 | 141 | read_directory(dir, pathspec ? *pathspec : "", len, pathspec); |
dba2e203 | 142 | return len; |
1d8842d9 LT |
143 | } |
144 | ||
bc96cc87 NTND |
145 | int within_depth(const char *name, int namelen, |
146 | int depth, int max_depth) | |
147 | { | |
148 | const char *cp = name, *cpe = name + namelen; | |
149 | ||
150 | while (cp < cpe) { | |
151 | if (*cp++ != '/') | |
152 | continue; | |
153 | depth++; | |
154 | if (depth > max_depth) | |
155 | return 0; | |
156 | } | |
157 | return 1; | |
158 | } | |
159 | ||
e813d50e | 160 | /* |
2c5b0115 | 161 | * Does 'match' match the given name? |
e813d50e JH |
162 | * A match is found if |
163 | * | |
164 | * (1) the 'match' string is leading directory of 'name', or | |
165 | * (2) the 'match' string is a wildcard and matches 'name', or | |
166 | * (3) the 'match' string is exactly the same as 'name'. | |
167 | * | |
168 | * and the return value tells which case it was. | |
169 | * | |
170 | * It returns 0 when there is no match. | |
171 | */ | |
3c6a370b LT |
172 | static int match_one(const char *match, const char *name, int namelen) |
173 | { | |
174 | int matchlen; | |
823ab40f | 175 | int literal = limit_pathspec_to_literal(); |
3c6a370b LT |
176 | |
177 | /* If the match was just the prefix, we matched */ | |
88ea8112 | 178 | if (!*match) |
e813d50e | 179 | return MATCHED_RECURSIVELY; |
3c6a370b | 180 | |
21444f18 JJ |
181 | if (ignore_case) { |
182 | for (;;) { | |
183 | unsigned char c1 = tolower(*match); | |
184 | unsigned char c2 = tolower(*name); | |
823ab40f | 185 | if (c1 == '\0' || (!literal && is_glob_special(c1))) |
21444f18 JJ |
186 | break; |
187 | if (c1 != c2) | |
188 | return 0; | |
189 | match++; | |
190 | name++; | |
191 | namelen--; | |
192 | } | |
193 | } else { | |
194 | for (;;) { | |
195 | unsigned char c1 = *match; | |
196 | unsigned char c2 = *name; | |
823ab40f | 197 | if (c1 == '\0' || (!literal && is_glob_special(c1))) |
21444f18 JJ |
198 | break; |
199 | if (c1 != c2) | |
200 | return 0; | |
201 | match++; | |
202 | name++; | |
203 | namelen--; | |
204 | } | |
88ea8112 LT |
205 | } |
206 | ||
3c6a370b LT |
207 | /* |
208 | * If we don't match the matchstring exactly, | |
209 | * we need to match by fnmatch | |
210 | */ | |
88ea8112 | 211 | matchlen = strlen(match); |
823ab40f JK |
212 | if (strncmp_icase(match, name, matchlen)) { |
213 | if (literal) | |
214 | return 0; | |
21444f18 | 215 | return !fnmatch_icase(match, name, 0) ? MATCHED_FNMATCH : 0; |
823ab40f | 216 | } |
3c6a370b | 217 | |
f2d0df71 | 218 | if (namelen == matchlen) |
e813d50e JH |
219 | return MATCHED_EXACTLY; |
220 | if (match[matchlen-1] == '/' || name[matchlen] == '/') | |
221 | return MATCHED_RECURSIVELY; | |
222 | return 0; | |
3c6a370b LT |
223 | } |
224 | ||
e813d50e | 225 | /* |
52ed1894 AS |
226 | * Given a name and a list of pathspecs, returns the nature of the |
227 | * closest (i.e. most specific) match of the name to any of the | |
228 | * pathspecs. | |
229 | * | |
230 | * The caller typically calls this multiple times with the same | |
231 | * pathspec and seen[] array but with different name/namelen | |
232 | * (e.g. entries from the index) and is interested in seeing if and | |
233 | * how each pathspec matches all the names it calls this function | |
234 | * with. A mark is left in the seen[] array for each pathspec element | |
235 | * indicating the closest type of match that element achieved, so if | |
236 | * seen[n] remains zero after multiple invocations, that means the nth | |
237 | * pathspec did not match any names, which could indicate that the | |
238 | * user mistyped the nth pathspec. | |
e813d50e | 239 | */ |
0b50922a CB |
240 | int match_pathspec(const char **pathspec, const char *name, int namelen, |
241 | int prefix, char *seen) | |
3c6a370b | 242 | { |
0b50922a CB |
243 | int i, retval = 0; |
244 | ||
245 | if (!pathspec) | |
246 | return 1; | |
3c6a370b LT |
247 | |
248 | name += prefix; | |
249 | namelen -= prefix; | |
250 | ||
0b50922a | 251 | for (i = 0; pathspec[i] != NULL; i++) { |
e813d50e | 252 | int how; |
0b50922a CB |
253 | const char *match = pathspec[i] + prefix; |
254 | if (seen && seen[i] == MATCHED_EXACTLY) | |
3c6a370b | 255 | continue; |
e813d50e JH |
256 | how = match_one(match, name, namelen); |
257 | if (how) { | |
258 | if (retval < how) | |
259 | retval = how; | |
0b50922a CB |
260 | if (seen && seen[i] < how) |
261 | seen[i] = how; | |
3c6a370b LT |
262 | } |
263 | } | |
264 | return retval; | |
265 | } | |
266 | ||
61cf2820 NTND |
267 | /* |
268 | * Does 'match' match the given name? | |
269 | * A match is found if | |
270 | * | |
271 | * (1) the 'match' string is leading directory of 'name', or | |
272 | * (2) the 'match' string is a wildcard and matches 'name', or | |
273 | * (3) the 'match' string is exactly the same as 'name'. | |
274 | * | |
275 | * and the return value tells which case it was. | |
276 | * | |
277 | * It returns 0 when there is no match. | |
278 | */ | |
279 | static int match_pathspec_item(const struct pathspec_item *item, int prefix, | |
280 | const char *name, int namelen) | |
281 | { | |
282 | /* name/namelen has prefix cut off by caller */ | |
283 | const char *match = item->match + prefix; | |
284 | int matchlen = item->len - prefix; | |
285 | ||
286 | /* If the match was just the prefix, we matched */ | |
287 | if (!*match) | |
288 | return MATCHED_RECURSIVELY; | |
289 | ||
290 | if (matchlen <= namelen && !strncmp(match, name, matchlen)) { | |
291 | if (matchlen == namelen) | |
292 | return MATCHED_EXACTLY; | |
293 | ||
294 | if (match[matchlen-1] == '/' || name[matchlen] == '/') | |
295 | return MATCHED_RECURSIVELY; | |
296 | } | |
297 | ||
5d74762d | 298 | if (item->nowildcard_len < item->len && |
8c6abbcd NTND |
299 | !git_fnmatch(match, name, |
300 | item->flags & PATHSPEC_ONESTAR ? GFNM_ONESTAR : 0, | |
301 | item->nowildcard_len - prefix)) | |
61cf2820 NTND |
302 | return MATCHED_FNMATCH; |
303 | ||
304 | return 0; | |
305 | } | |
306 | ||
307 | /* | |
52ed1894 AS |
308 | * Given a name and a list of pathspecs, returns the nature of the |
309 | * closest (i.e. most specific) match of the name to any of the | |
310 | * pathspecs. | |
311 | * | |
312 | * The caller typically calls this multiple times with the same | |
313 | * pathspec and seen[] array but with different name/namelen | |
314 | * (e.g. entries from the index) and is interested in seeing if and | |
315 | * how each pathspec matches all the names it calls this function | |
316 | * with. A mark is left in the seen[] array for each pathspec element | |
317 | * indicating the closest type of match that element achieved, so if | |
318 | * seen[n] remains zero after multiple invocations, that means the nth | |
319 | * pathspec did not match any names, which could indicate that the | |
320 | * user mistyped the nth pathspec. | |
61cf2820 NTND |
321 | */ |
322 | int match_pathspec_depth(const struct pathspec *ps, | |
323 | const char *name, int namelen, | |
324 | int prefix, char *seen) | |
325 | { | |
326 | int i, retval = 0; | |
327 | ||
328 | if (!ps->nr) { | |
329 | if (!ps->recursive || ps->max_depth == -1) | |
330 | return MATCHED_RECURSIVELY; | |
331 | ||
332 | if (within_depth(name, namelen, 0, ps->max_depth)) | |
333 | return MATCHED_EXACTLY; | |
334 | else | |
335 | return 0; | |
336 | } | |
337 | ||
338 | name += prefix; | |
339 | namelen -= prefix; | |
340 | ||
341 | for (i = ps->nr - 1; i >= 0; i--) { | |
342 | int how; | |
343 | if (seen && seen[i] == MATCHED_EXACTLY) | |
344 | continue; | |
345 | how = match_pathspec_item(ps->items+i, prefix, name, namelen); | |
346 | if (ps->recursive && ps->max_depth != -1 && | |
347 | how && how != MATCHED_FNMATCH) { | |
348 | int len = ps->items[i].len; | |
349 | if (name[len] == '/') | |
350 | len++; | |
351 | if (within_depth(name+len, namelen-len, 0, ps->max_depth)) | |
352 | how = MATCHED_EXACTLY; | |
353 | else | |
354 | how = 0; | |
355 | } | |
356 | if (how) { | |
357 | if (retval < how) | |
358 | retval = how; | |
359 | if (seen && seen[i] < how) | |
360 | seen[i] = how; | |
361 | } | |
362 | } | |
363 | return retval; | |
364 | } | |
365 | ||
fcd631ed NTND |
366 | /* |
367 | * Return the length of the "simple" part of a path match limiter. | |
368 | */ | |
369 | static int simple_length(const char *match) | |
370 | { | |
371 | int len = -1; | |
372 | ||
373 | for (;;) { | |
374 | unsigned char c = *match++; | |
375 | len++; | |
376 | if (c == '\0' || is_glob_special(c)) | |
377 | return len; | |
378 | } | |
379 | } | |
380 | ||
68492fc7 LK |
381 | static int no_wildcard(const char *string) |
382 | { | |
fcd631ed | 383 | return string[simple_length(string)] == '\0'; |
68492fc7 LK |
384 | } |
385 | ||
82dce998 NTND |
386 | void parse_exclude_pattern(const char **pattern, |
387 | int *patternlen, | |
388 | int *flags, | |
389 | int *nowildcardlen) | |
84460eec NTND |
390 | { |
391 | const char *p = *pattern; | |
392 | size_t i, len; | |
393 | ||
394 | *flags = 0; | |
395 | if (*p == '!') { | |
396 | *flags |= EXC_FLAG_NEGATIVE; | |
397 | p++; | |
398 | } | |
399 | len = strlen(p); | |
400 | if (len && p[len - 1] == '/') { | |
401 | len--; | |
402 | *flags |= EXC_FLAG_MUSTBEDIR; | |
403 | } | |
404 | for (i = 0; i < len; i++) { | |
405 | if (p[i] == '/') | |
406 | break; | |
407 | } | |
408 | if (i == len) | |
409 | *flags |= EXC_FLAG_NODIR; | |
410 | *nowildcardlen = simple_length(p); | |
411 | /* | |
412 | * we should have excluded the trailing slash from 'p' too, | |
413 | * but that's one more allocation. Instead just make sure | |
414 | * nowildcardlen does not exceed real patternlen | |
415 | */ | |
416 | if (*nowildcardlen > len) | |
417 | *nowildcardlen = len; | |
418 | if (*p == '*' && no_wildcard(p + 1)) | |
419 | *flags |= EXC_FLAG_ENDSWITH; | |
420 | *pattern = p; | |
421 | *patternlen = len; | |
422 | } | |
423 | ||
453ec4bd | 424 | void add_exclude(const char *string, const char *base, |
c04318e4 | 425 | int baselen, struct exclude_list *el, int srcpos) |
453ec4bd | 426 | { |
d6b8fc30 | 427 | struct exclude *x; |
84460eec NTND |
428 | int patternlen; |
429 | int flags; | |
430 | int nowildcardlen; | |
453ec4bd | 431 | |
84460eec NTND |
432 | parse_exclude_pattern(&string, &patternlen, &flags, &nowildcardlen); |
433 | if (flags & EXC_FLAG_MUSTBEDIR) { | |
d6b8fc30 | 434 | char *s; |
84460eec | 435 | x = xmalloc(sizeof(*x) + patternlen + 1); |
4b25d091 | 436 | s = (char *)(x+1); |
84460eec NTND |
437 | memcpy(s, string, patternlen); |
438 | s[patternlen] = '\0'; | |
d6b8fc30 | 439 | x->pattern = s; |
d6b8fc30 JH |
440 | } else { |
441 | x = xmalloc(sizeof(*x)); | |
442 | x->pattern = string; | |
443 | } | |
84460eec NTND |
444 | x->patternlen = patternlen; |
445 | x->nowildcardlen = nowildcardlen; | |
453ec4bd LT |
446 | x->base = base; |
447 | x->baselen = baselen; | |
d6b8fc30 | 448 | x->flags = flags; |
c04318e4 | 449 | x->srcpos = srcpos; |
840fc334 AS |
450 | ALLOC_GROW(el->excludes, el->nr + 1, el->alloc); |
451 | el->excludes[el->nr++] = x; | |
c04318e4 | 452 | x->el = el; |
453ec4bd LT |
453 | } |
454 | ||
c28b3d6e NTND |
455 | static void *read_skip_worktree_file_from_index(const char *path, size_t *size) |
456 | { | |
457 | int pos, len; | |
458 | unsigned long sz; | |
459 | enum object_type type; | |
460 | void *data; | |
461 | struct index_state *istate = &the_index; | |
462 | ||
463 | len = strlen(path); | |
464 | pos = index_name_pos(istate, path, len); | |
465 | if (pos < 0) | |
466 | return NULL; | |
467 | if (!ce_skip_worktree(istate->cache[pos])) | |
468 | return NULL; | |
469 | data = read_sha1_file(istate->cache[pos]->sha1, &type, &sz); | |
470 | if (!data || type != OBJ_BLOB) { | |
471 | free(data); | |
472 | return NULL; | |
473 | } | |
474 | *size = xsize_t(sz); | |
475 | return data; | |
476 | } | |
477 | ||
f6198812 AS |
478 | /* |
479 | * Frees memory within el which was allocated for exclude patterns and | |
480 | * the file buffer. Does not free el itself. | |
481 | */ | |
482 | void clear_exclude_list(struct exclude_list *el) | |
0fd0e241 NTND |
483 | { |
484 | int i; | |
485 | ||
486 | for (i = 0; i < el->nr; i++) | |
487 | free(el->excludes[i]); | |
488 | free(el->excludes); | |
c082df24 | 489 | free(el->filebuf); |
0fd0e241 NTND |
490 | |
491 | el->nr = 0; | |
492 | el->excludes = NULL; | |
c082df24 | 493 | el->filebuf = NULL; |
0fd0e241 NTND |
494 | } |
495 | ||
cb097534 NTND |
496 | int add_excludes_from_file_to_list(const char *fname, |
497 | const char *base, | |
498 | int baselen, | |
840fc334 | 499 | struct exclude_list *el, |
cb097534 | 500 | int check_index) |
453ec4bd | 501 | { |
c470701a | 502 | struct stat st; |
c04318e4 | 503 | int fd, i, lineno = 1; |
9d14017a | 504 | size_t size = 0; |
453ec4bd LT |
505 | char *buf, *entry; |
506 | ||
507 | fd = open(fname, O_RDONLY); | |
c28b3d6e | 508 | if (fd < 0 || fstat(fd, &st) < 0) { |
69660731 | 509 | if (errno != ENOENT) |
55b38a48 | 510 | warn_on_inaccessible(fname); |
c28b3d6e NTND |
511 | if (0 <= fd) |
512 | close(fd); | |
513 | if (!check_index || | |
514 | (buf = read_skip_worktree_file_from_index(fname, &size)) == NULL) | |
515 | return -1; | |
45d76f17 NTND |
516 | if (size == 0) { |
517 | free(buf); | |
518 | return 0; | |
519 | } | |
520 | if (buf[size-1] != '\n') { | |
521 | buf = xrealloc(buf, size+1); | |
522 | buf[size++] = '\n'; | |
523 | } | |
453ec4bd | 524 | } |
c28b3d6e NTND |
525 | else { |
526 | size = xsize_t(st.st_size); | |
527 | if (size == 0) { | |
528 | close(fd); | |
529 | return 0; | |
530 | } | |
45d76f17 | 531 | buf = xmalloc(size+1); |
c28b3d6e | 532 | if (read_in_full(fd, buf, size) != size) { |
45d76f17 | 533 | free(buf); |
c28b3d6e NTND |
534 | close(fd); |
535 | return -1; | |
536 | } | |
45d76f17 | 537 | buf[size++] = '\n'; |
c28b3d6e | 538 | close(fd); |
6ba78238 | 539 | } |
453ec4bd | 540 | |
c082df24 | 541 | el->filebuf = buf; |
453ec4bd | 542 | entry = buf; |
45d76f17 NTND |
543 | for (i = 0; i < size; i++) { |
544 | if (buf[i] == '\n') { | |
453ec4bd LT |
545 | if (entry != buf + i && entry[0] != '#') { |
546 | buf[i - (i && buf[i-1] == '\r')] = 0; | |
c04318e4 | 547 | add_exclude(entry, base, baselen, el, lineno); |
453ec4bd | 548 | } |
c04318e4 | 549 | lineno++; |
453ec4bd LT |
550 | entry = buf + i + 1; |
551 | } | |
552 | } | |
553 | return 0; | |
453ec4bd LT |
554 | } |
555 | ||
c04318e4 AS |
556 | struct exclude_list *add_exclude_list(struct dir_struct *dir, |
557 | int group_type, const char *src) | |
c082df24 AS |
558 | { |
559 | struct exclude_list *el; | |
560 | struct exclude_list_group *group; | |
561 | ||
562 | group = &dir->exclude_list_group[group_type]; | |
563 | ALLOC_GROW(group->el, group->nr + 1, group->alloc); | |
564 | el = &group->el[group->nr++]; | |
565 | memset(el, 0, sizeof(*el)); | |
c04318e4 | 566 | el->src = src; |
c082df24 AS |
567 | return el; |
568 | } | |
569 | ||
570 | /* | |
571 | * Used to set up core.excludesfile and .git/info/exclude lists. | |
572 | */ | |
453ec4bd LT |
573 | void add_excludes_from_file(struct dir_struct *dir, const char *fname) |
574 | { | |
c082df24 | 575 | struct exclude_list *el; |
c04318e4 | 576 | el = add_exclude_list(dir, EXC_FILE, fname); |
c082df24 | 577 | if (add_excludes_from_file_to_list(fname, "", 0, el, 0) < 0) |
453ec4bd LT |
578 | die("cannot use %s as an exclude file", fname); |
579 | } | |
580 | ||
82dce998 NTND |
581 | int match_basename(const char *basename, int basenamelen, |
582 | const char *pattern, int prefix, int patternlen, | |
583 | int flags) | |
593cb880 NTND |
584 | { |
585 | if (prefix == patternlen) { | |
0b6e56df JH |
586 | if (patternlen == basenamelen && |
587 | !strncmp_icase(pattern, basename, basenamelen)) | |
593cb880 NTND |
588 | return 1; |
589 | } else if (flags & EXC_FLAG_ENDSWITH) { | |
0b6e56df | 590 | /* "*literal" matching against "fooliteral" */ |
593cb880 | 591 | if (patternlen - 1 <= basenamelen && |
0b6e56df JH |
592 | !strncmp_icase(pattern + 1, |
593 | basename + basenamelen - (patternlen - 1), | |
594 | patternlen - 1)) | |
593cb880 NTND |
595 | return 1; |
596 | } else { | |
0b6e56df JH |
597 | if (fnmatch_icase_mem(pattern, patternlen, |
598 | basename, basenamelen, | |
599 | 0) == 0) | |
593cb880 NTND |
600 | return 1; |
601 | } | |
602 | return 0; | |
603 | } | |
604 | ||
82dce998 NTND |
605 | int match_pathname(const char *pathname, int pathlen, |
606 | const char *base, int baselen, | |
607 | const char *pattern, int prefix, int patternlen, | |
608 | int flags) | |
b5592632 NTND |
609 | { |
610 | const char *name; | |
611 | int namelen; | |
612 | ||
613 | /* | |
614 | * match with FNM_PATHNAME; the pattern has base implicitly | |
615 | * in front of it. | |
616 | */ | |
617 | if (*pattern == '/') { | |
618 | pattern++; | |
982ac873 | 619 | patternlen--; |
b5592632 NTND |
620 | prefix--; |
621 | } | |
622 | ||
623 | /* | |
624 | * baselen does not count the trailing slash. base[] may or | |
625 | * may not end with a trailing slash though. | |
626 | */ | |
627 | if (pathlen < baselen + 1 || | |
628 | (baselen && pathname[baselen] != '/') || | |
629 | strncmp_icase(pathname, base, baselen)) | |
630 | return 0; | |
631 | ||
632 | namelen = baselen ? pathlen - baselen - 1 : pathlen; | |
633 | name = pathname + pathlen - namelen; | |
634 | ||
635 | if (prefix) { | |
636 | /* | |
637 | * if the non-wildcard part is longer than the | |
638 | * remaining pathname, surely it cannot match. | |
639 | */ | |
640 | if (prefix > namelen) | |
641 | return 0; | |
642 | ||
643 | if (strncmp_icase(pattern, name, prefix)) | |
644 | return 0; | |
645 | pattern += prefix; | |
ab3aebc1 | 646 | patternlen -= prefix; |
b5592632 NTND |
647 | name += prefix; |
648 | namelen -= prefix; | |
ab3aebc1 JK |
649 | |
650 | /* | |
651 | * If the whole pattern did not have a wildcard, | |
652 | * then our prefix match is all we need; we | |
653 | * do not need to call fnmatch at all. | |
654 | */ | |
655 | if (!patternlen && !namelen) | |
656 | return 1; | |
b5592632 NTND |
657 | } |
658 | ||
ab3aebc1 JK |
659 | return fnmatch_icase_mem(pattern, patternlen, |
660 | name, namelen, | |
f30366b2 | 661 | WM_PATHNAME) == 0; |
b5592632 NTND |
662 | } |
663 | ||
578cd7c3 AS |
664 | /* |
665 | * Scan the given exclude list in reverse to see whether pathname | |
666 | * should be ignored. The first match (i.e. the last on the list), if | |
667 | * any, determines the fate. Returns the exclude_list element which | |
668 | * matched, or NULL for undecided. | |
453ec4bd | 669 | */ |
578cd7c3 AS |
670 | static struct exclude *last_exclude_matching_from_list(const char *pathname, |
671 | int pathlen, | |
672 | const char *basename, | |
673 | int *dtype, | |
674 | struct exclude_list *el) | |
453ec4bd LT |
675 | { |
676 | int i; | |
677 | ||
35a94d44 | 678 | if (!el->nr) |
578cd7c3 | 679 | return NULL; /* undefined */ |
d6b8fc30 | 680 | |
35a94d44 NTND |
681 | for (i = el->nr - 1; 0 <= i; i--) { |
682 | struct exclude *x = el->excludes[i]; | |
b5592632 | 683 | const char *exclude = x->pattern; |
b5592632 | 684 | int prefix = x->nowildcardlen; |
35a94d44 NTND |
685 | |
686 | if (x->flags & EXC_FLAG_MUSTBEDIR) { | |
687 | if (*dtype == DT_UNKNOWN) | |
688 | *dtype = get_dtype(NULL, pathname, pathlen); | |
689 | if (*dtype != DT_DIR) | |
690 | continue; | |
691 | } | |
d6b8fc30 | 692 | |
35a94d44 | 693 | if (x->flags & EXC_FLAG_NODIR) { |
593cb880 NTND |
694 | if (match_basename(basename, |
695 | pathlen - (basename - pathname), | |
696 | exclude, prefix, x->patternlen, | |
697 | x->flags)) | |
578cd7c3 | 698 | return x; |
35a94d44 NTND |
699 | continue; |
700 | } | |
701 | ||
b5592632 NTND |
702 | assert(x->baselen == 0 || x->base[x->baselen - 1] == '/'); |
703 | if (match_pathname(pathname, pathlen, | |
704 | x->base, x->baselen ? x->baselen - 1 : 0, | |
705 | exclude, prefix, x->patternlen, x->flags)) | |
578cd7c3 | 706 | return x; |
453ec4bd | 707 | } |
578cd7c3 AS |
708 | return NULL; /* undecided */ |
709 | } | |
710 | ||
711 | /* | |
712 | * Scan the list and let the last match determine the fate. | |
713 | * Return 1 for exclude, 0 for include and -1 for undecided. | |
714 | */ | |
715 | int is_excluded_from_list(const char *pathname, | |
716 | int pathlen, const char *basename, int *dtype, | |
717 | struct exclude_list *el) | |
718 | { | |
719 | struct exclude *exclude; | |
720 | exclude = last_exclude_matching_from_list(pathname, pathlen, basename, dtype, el); | |
721 | if (exclude) | |
722 | return exclude->flags & EXC_FLAG_NEGATIVE ? 0 : 1; | |
453ec4bd LT |
723 | return -1; /* undecided */ |
724 | } | |
725 | ||
46aa2f95 KB |
726 | static struct exclude *last_exclude_matching_from_lists(struct dir_struct *dir, |
727 | const char *pathname, int pathlen, const char *basename, | |
728 | int *dtype_p) | |
729 | { | |
730 | int i, j; | |
731 | struct exclude_list_group *group; | |
732 | struct exclude *exclude; | |
733 | for (i = EXC_CMDL; i <= EXC_FILE; i++) { | |
734 | group = &dir->exclude_list_group[i]; | |
735 | for (j = group->nr - 1; j >= 0; j--) { | |
736 | exclude = last_exclude_matching_from_list( | |
737 | pathname, pathlen, basename, dtype_p, | |
738 | &group->el[j]); | |
739 | if (exclude) | |
740 | return exclude; | |
741 | } | |
742 | } | |
743 | return NULL; | |
744 | } | |
745 | ||
6cd5c582 KB |
746 | /* |
747 | * Loads the per-directory exclude list for the substring of base | |
748 | * which has a char length of baselen. | |
749 | */ | |
750 | static void prep_exclude(struct dir_struct *dir, const char *base, int baselen) | |
751 | { | |
752 | struct exclude_list_group *group; | |
753 | struct exclude_list *el; | |
754 | struct exclude_stack *stk = NULL; | |
755 | int current; | |
756 | ||
6cd5c582 KB |
757 | group = &dir->exclude_list_group[EXC_DIRS]; |
758 | ||
759 | /* Pop the exclude lists from the EXCL_DIRS exclude_list_group | |
760 | * which originate from directories not in the prefix of the | |
761 | * path being checked. */ | |
762 | while ((stk = dir->exclude_stack) != NULL) { | |
763 | if (stk->baselen <= baselen && | |
764 | !strncmp(dir->basebuf, base, stk->baselen)) | |
765 | break; | |
766 | el = &group->el[dir->exclude_stack->exclude_ix]; | |
767 | dir->exclude_stack = stk->prev; | |
95c6f271 | 768 | dir->exclude = NULL; |
6cd5c582 KB |
769 | free((char *)el->src); /* see strdup() below */ |
770 | clear_exclude_list(el); | |
771 | free(stk); | |
772 | group->nr--; | |
773 | } | |
774 | ||
95c6f271 KB |
775 | /* Skip traversing into sub directories if the parent is excluded */ |
776 | if (dir->exclude) | |
777 | return; | |
778 | ||
6cd5c582 KB |
779 | /* Read from the parent directories and push them down. */ |
780 | current = stk ? stk->baselen : -1; | |
781 | while (current < baselen) { | |
782 | struct exclude_stack *stk = xcalloc(1, sizeof(*stk)); | |
783 | const char *cp; | |
784 | ||
785 | if (current < 0) { | |
786 | cp = base; | |
787 | current = 0; | |
788 | } | |
789 | else { | |
790 | cp = strchr(base + current + 1, '/'); | |
791 | if (!cp) | |
792 | die("oops in prep_exclude"); | |
793 | cp++; | |
794 | } | |
795 | stk->prev = dir->exclude_stack; | |
796 | stk->baselen = cp - base; | |
95c6f271 KB |
797 | stk->exclude_ix = group->nr; |
798 | el = add_exclude_list(dir, EXC_DIRS, NULL); | |
6cd5c582 KB |
799 | memcpy(dir->basebuf + current, base + current, |
800 | stk->baselen - current); | |
95c6f271 KB |
801 | |
802 | /* Abort if the directory is excluded */ | |
803 | if (stk->baselen) { | |
804 | int dt = DT_DIR; | |
805 | dir->basebuf[stk->baselen - 1] = 0; | |
806 | dir->exclude = last_exclude_matching_from_lists(dir, | |
807 | dir->basebuf, stk->baselen - 1, | |
808 | dir->basebuf + current, &dt); | |
809 | dir->basebuf[stk->baselen - 1] = '/'; | |
810 | if (dir->exclude) { | |
811 | dir->basebuf[stk->baselen] = 0; | |
812 | dir->exclude_stack = stk; | |
813 | return; | |
814 | } | |
815 | } | |
816 | ||
817 | /* Try to read per-directory file unless path is too long */ | |
818 | if (dir->exclude_per_dir && | |
819 | stk->baselen + strlen(dir->exclude_per_dir) < PATH_MAX) { | |
820 | strcpy(dir->basebuf + stk->baselen, | |
821 | dir->exclude_per_dir); | |
822 | /* | |
823 | * dir->basebuf gets reused by the traversal, but we | |
824 | * need fname to remain unchanged to ensure the src | |
825 | * member of each struct exclude correctly | |
826 | * back-references its source file. Other invocations | |
827 | * of add_exclude_list provide stable strings, so we | |
828 | * strdup() and free() here in the caller. | |
829 | */ | |
830 | el->src = strdup(dir->basebuf); | |
831 | add_excludes_from_file_to_list(dir->basebuf, | |
832 | dir->basebuf, stk->baselen, el, 1); | |
833 | } | |
6cd5c582 KB |
834 | dir->exclude_stack = stk; |
835 | current = stk->baselen; | |
836 | } | |
837 | dir->basebuf[baselen] = '\0'; | |
838 | } | |
839 | ||
f4cd69a6 AS |
840 | /* |
841 | * Loads the exclude lists for the directory containing pathname, then | |
842 | * scans all exclude lists to determine whether pathname is excluded. | |
843 | * Returns the exclude_list element which matched, or NULL for | |
844 | * undecided. | |
845 | */ | |
b07bc8c8 | 846 | struct exclude *last_exclude_matching(struct dir_struct *dir, |
f4cd69a6 AS |
847 | const char *pathname, |
848 | int *dtype_p) | |
453ec4bd LT |
849 | { |
850 | int pathlen = strlen(pathname); | |
68492fc7 LK |
851 | const char *basename = strrchr(pathname, '/'); |
852 | basename = (basename) ? basename+1 : pathname; | |
453ec4bd | 853 | |
63d285c8 | 854 | prep_exclude(dir, pathname, basename-pathname); |
c082df24 | 855 | |
95c6f271 KB |
856 | if (dir->exclude) |
857 | return dir->exclude; | |
858 | ||
46aa2f95 KB |
859 | return last_exclude_matching_from_lists(dir, pathname, pathlen, |
860 | basename, dtype_p); | |
f4cd69a6 AS |
861 | } |
862 | ||
863 | /* | |
864 | * Loads the exclude lists for the directory containing pathname, then | |
865 | * scans all exclude lists to determine whether pathname is excluded. | |
866 | * Returns 1 if true, otherwise 0. | |
867 | */ | |
b07bc8c8 | 868 | int is_excluded(struct dir_struct *dir, const char *pathname, int *dtype_p) |
f4cd69a6 AS |
869 | { |
870 | struct exclude *exclude = | |
871 | last_exclude_matching(dir, pathname, dtype_p); | |
872 | if (exclude) | |
873 | return exclude->flags & EXC_FLAG_NEGATIVE ? 0 : 1; | |
453ec4bd LT |
874 | return 0; |
875 | } | |
876 | ||
f3fa1838 JH |
877 | static struct dir_entry *dir_entry_new(const char *pathname, int len) |
878 | { | |
453ec4bd LT |
879 | struct dir_entry *ent; |
880 | ||
453ec4bd LT |
881 | ent = xmalloc(sizeof(*ent) + len + 1); |
882 | ent->len = len; | |
883 | memcpy(ent->name, pathname, len); | |
884 | ent->name[len] = 0; | |
4d06f8ac | 885 | return ent; |
453ec4bd LT |
886 | } |
887 | ||
159b3212 | 888 | static struct dir_entry *dir_add_name(struct dir_struct *dir, const char *pathname, int len) |
6815e569 | 889 | { |
5bd8e2d8 | 890 | if (cache_name_exists(pathname, len, ignore_case)) |
6815e569 JK |
891 | return NULL; |
892 | ||
25fd2f7a | 893 | ALLOC_GROW(dir->entries, dir->nr+1, dir->alloc); |
6815e569 JK |
894 | return dir->entries[dir->nr++] = dir_entry_new(pathname, len); |
895 | } | |
896 | ||
108da0db | 897 | struct dir_entry *dir_add_ignored(struct dir_struct *dir, const char *pathname, int len) |
2abd31b0 | 898 | { |
6e4f981f | 899 | if (!cache_name_is_other(pathname, len)) |
2abd31b0 JK |
900 | return NULL; |
901 | ||
25fd2f7a | 902 | ALLOC_GROW(dir->ignored, dir->ignored_nr+1, dir->ignored_alloc); |
2abd31b0 JK |
903 | return dir->ignored[dir->ignored_nr++] = dir_entry_new(pathname, len); |
904 | } | |
905 | ||
09595258 LT |
906 | enum exist_status { |
907 | index_nonexistent = 0, | |
908 | index_directory, | |
4b05548f | 909 | index_gitdir |
09595258 LT |
910 | }; |
911 | ||
5102c617 JJ |
912 | /* |
913 | * Do not use the alphabetically stored index to look up | |
914 | * the directory name; instead, use the case insensitive | |
915 | * name hash. | |
916 | */ | |
917 | static enum exist_status directory_exists_in_index_icase(const char *dirname, int len) | |
918 | { | |
919 | struct cache_entry *ce = index_name_exists(&the_index, dirname, len + 1, ignore_case); | |
920 | unsigned char endchar; | |
921 | ||
922 | if (!ce) | |
923 | return index_nonexistent; | |
924 | endchar = ce->name[len]; | |
925 | ||
926 | /* | |
927 | * The cache_entry structure returned will contain this dirname | |
928 | * and possibly additional path components. | |
929 | */ | |
930 | if (endchar == '/') | |
931 | return index_directory; | |
932 | ||
933 | /* | |
934 | * If there are no additional path components, then this cache_entry | |
935 | * represents a submodule. Submodules, despite being directories, | |
936 | * are stored in the cache without a closing slash. | |
937 | */ | |
938 | if (!endchar && S_ISGITLINK(ce->ce_mode)) | |
939 | return index_gitdir; | |
940 | ||
941 | /* This should never be hit, but it exists just in case. */ | |
942 | return index_nonexistent; | |
943 | } | |
944 | ||
09595258 LT |
945 | /* |
946 | * The index sorts alphabetically by entry name, which | |
947 | * means that a gitlink sorts as '\0' at the end, while | |
948 | * a directory (which is defined not as an entry, but as | |
949 | * the files it contains) will sort with the '/' at the | |
950 | * end. | |
951 | */ | |
952 | static enum exist_status directory_exists_in_index(const char *dirname, int len) | |
453ec4bd | 953 | { |
5102c617 JJ |
954 | int pos; |
955 | ||
956 | if (ignore_case) | |
957 | return directory_exists_in_index_icase(dirname, len); | |
958 | ||
959 | pos = cache_name_pos(dirname, len); | |
09595258 LT |
960 | if (pos < 0) |
961 | pos = -pos-1; | |
962 | while (pos < active_nr) { | |
963 | struct cache_entry *ce = active_cache[pos++]; | |
964 | unsigned char endchar; | |
965 | ||
966 | if (strncmp(ce->name, dirname, len)) | |
967 | break; | |
968 | endchar = ce->name[len]; | |
969 | if (endchar > '/') | |
970 | break; | |
971 | if (endchar == '/') | |
972 | return index_directory; | |
7a51ed66 | 973 | if (!endchar && S_ISGITLINK(ce->ce_mode)) |
09595258 LT |
974 | return index_gitdir; |
975 | } | |
976 | return index_nonexistent; | |
977 | } | |
978 | ||
979 | /* | |
980 | * When we find a directory when traversing the filesystem, we | |
981 | * have three distinct cases: | |
982 | * | |
983 | * - ignore it | |
984 | * - see it as a directory | |
985 | * - recurse into it | |
986 | * | |
987 | * and which one we choose depends on a combination of existing | |
988 | * git index contents and the flags passed into the directory | |
989 | * traversal routine. | |
990 | * | |
991 | * Case 1: If we *already* have entries in the index under that | |
5bd8e2d8 KB |
992 | * directory name, we always recurse into the directory to see |
993 | * all the files. | |
09595258 LT |
994 | * |
995 | * Case 2: If we *already* have that directory name as a gitlink, | |
996 | * we always continue to see it as a gitlink, regardless of whether | |
997 | * there is an actual git directory there or not (it might not | |
998 | * be checked out as a subproject!) | |
999 | * | |
1000 | * Case 3: if we didn't have it in the index previously, we | |
1001 | * have a few sub-cases: | |
1002 | * | |
1003 | * (a) if "show_other_directories" is true, we show it as | |
1004 | * just a directory, unless "hide_empty_directories" is | |
1005 | * also true and the directory is empty, in which case | |
1006 | * we just ignore it entirely. | |
721ac4ed AP |
1007 | * if we are looking for ignored directories, look if it |
1008 | * contains only ignored files to decide if it must be shown as | |
1009 | * ignored or not. | |
09595258 | 1010 | * (b) if it looks like a git directory, and we don't have |
302b9282 | 1011 | * 'no_gitlinks' set we treat it as a gitlink, and show it |
09595258 LT |
1012 | * as a directory. |
1013 | * (c) otherwise, we recurse into it. | |
1014 | */ | |
1015 | enum directory_treatment { | |
1016 | show_directory, | |
1017 | ignore_directory, | |
4b05548f | 1018 | recurse_into_directory |
09595258 LT |
1019 | }; |
1020 | ||
1021 | static enum directory_treatment treat_directory(struct dir_struct *dir, | |
721ac4ed | 1022 | const char *dirname, int len, int exclude, |
09595258 LT |
1023 | const struct path_simplify *simplify) |
1024 | { | |
1025 | /* The "len-1" is to strip the final '/' */ | |
1026 | switch (directory_exists_in_index(dirname, len-1)) { | |
1027 | case index_directory: | |
1028 | return recurse_into_directory; | |
1029 | ||
1030 | case index_gitdir: | |
7c4c97c0 | 1031 | if (dir->flags & DIR_SHOW_OTHER_DIRECTORIES) |
ab22aed3 | 1032 | return ignore_directory; |
09595258 LT |
1033 | return show_directory; |
1034 | ||
1035 | case index_nonexistent: | |
7c4c97c0 | 1036 | if (dir->flags & DIR_SHOW_OTHER_DIRECTORIES) |
09595258 | 1037 | break; |
7c4c97c0 | 1038 | if (!(dir->flags & DIR_NO_GITLINKS)) { |
09595258 LT |
1039 | unsigned char sha1[20]; |
1040 | if (resolve_gitlink_ref(dirname, "HEAD", sha1) == 0) | |
1041 | return show_directory; | |
1042 | } | |
1043 | return recurse_into_directory; | |
1044 | } | |
1045 | ||
1046 | /* This is the "show_other_directories" case */ | |
721ac4ed AP |
1047 | |
1048 | /* | |
1049 | * We are looking for ignored files and our directory is not ignored, | |
c94ab010 | 1050 | * check if it contains untracked files (i.e. is listed as untracked) |
721ac4ed AP |
1051 | */ |
1052 | if ((dir->flags & DIR_SHOW_IGNORED) && !exclude) { | |
1053 | int ignored; | |
1054 | dir->flags &= ~DIR_SHOW_IGNORED; | |
721ac4ed | 1055 | ignored = read_directory_recursive(dir, dirname, len, 1, simplify); |
721ac4ed AP |
1056 | dir->flags |= DIR_SHOW_IGNORED; |
1057 | ||
c94ab010 KB |
1058 | if (ignored) |
1059 | return ignore_directory; | |
721ac4ed | 1060 | } |
184d2a8e KB |
1061 | |
1062 | if (!(dir->flags & DIR_HIDE_EMPTY_DIRECTORIES)) | |
09595258 | 1063 | return show_directory; |
dba2e203 | 1064 | if (!read_directory_recursive(dir, dirname, len, 1, simplify)) |
09595258 LT |
1065 | return ignore_directory; |
1066 | return show_directory; | |
453ec4bd LT |
1067 | } |
1068 | ||
9fc42d60 LT |
1069 | /* |
1070 | * This is an inexact early pruning of any recursive directory | |
1071 | * reading - if the path cannot possibly be in the pathspec, | |
1072 | * return true, and we'll skip it early. | |
1073 | */ | |
1074 | static int simplify_away(const char *path, int pathlen, const struct path_simplify *simplify) | |
1075 | { | |
1076 | if (simplify) { | |
1077 | for (;;) { | |
1078 | const char *match = simplify->path; | |
1079 | int len = simplify->len; | |
1080 | ||
1081 | if (!match) | |
1082 | break; | |
1083 | if (len > pathlen) | |
1084 | len = pathlen; | |
1085 | if (!memcmp(path, match, len)) | |
1086 | return 0; | |
1087 | simplify++; | |
1088 | } | |
1089 | return 1; | |
1090 | } | |
1091 | return 0; | |
1092 | } | |
1093 | ||
29209cbe JK |
1094 | /* |
1095 | * This function tells us whether an excluded path matches a | |
1096 | * list of "interesting" pathspecs. That is, whether a path matched | |
1097 | * by any of the pathspecs could possibly be ignored by excluding | |
1098 | * the specified path. This can happen if: | |
1099 | * | |
1100 | * 1. the path is mentioned explicitly in the pathspec | |
1101 | * | |
1102 | * 2. the path is a directory prefix of some element in the | |
1103 | * pathspec | |
1104 | */ | |
1105 | static int exclude_matches_pathspec(const char *path, int len, | |
1106 | const struct path_simplify *simplify) | |
e96980ef JK |
1107 | { |
1108 | if (simplify) { | |
1109 | for (; simplify->path; simplify++) { | |
1110 | if (len == simplify->len | |
1111 | && !memcmp(path, simplify->path, len)) | |
1112 | return 1; | |
29209cbe JK |
1113 | if (len < simplify->len |
1114 | && simplify->path[len] == '/' | |
1115 | && !memcmp(path, simplify->path, len)) | |
1116 | return 1; | |
e96980ef JK |
1117 | } |
1118 | } | |
1119 | return 0; | |
1120 | } | |
1121 | ||
443e061a LT |
1122 | static int get_index_dtype(const char *path, int len) |
1123 | { | |
1124 | int pos; | |
1125 | struct cache_entry *ce; | |
1126 | ||
1127 | ce = cache_name_exists(path, len, 0); | |
1128 | if (ce) { | |
1129 | if (!ce_uptodate(ce)) | |
1130 | return DT_UNKNOWN; | |
1131 | if (S_ISGITLINK(ce->ce_mode)) | |
1132 | return DT_DIR; | |
1133 | /* | |
1134 | * Nobody actually cares about the | |
1135 | * difference between DT_LNK and DT_REG | |
1136 | */ | |
1137 | return DT_REG; | |
1138 | } | |
1139 | ||
1140 | /* Try to look it up as a directory */ | |
1141 | pos = cache_name_pos(path, len); | |
1142 | if (pos >= 0) | |
1143 | return DT_UNKNOWN; | |
1144 | pos = -pos-1; | |
1145 | while (pos < active_nr) { | |
1146 | ce = active_cache[pos++]; | |
1147 | if (strncmp(ce->name, path, len)) | |
1148 | break; | |
1149 | if (ce->name[len] > '/') | |
1150 | break; | |
1151 | if (ce->name[len] < '/') | |
1152 | continue; | |
1153 | if (!ce_uptodate(ce)) | |
1154 | break; /* continue? */ | |
1155 | return DT_DIR; | |
1156 | } | |
1157 | return DT_UNKNOWN; | |
1158 | } | |
1159 | ||
caa6b782 | 1160 | static int get_dtype(struct dirent *de, const char *path, int len) |
07134421 | 1161 | { |
6831a88a | 1162 | int dtype = de ? DTYPE(de) : DT_UNKNOWN; |
07134421 LT |
1163 | struct stat st; |
1164 | ||
1165 | if (dtype != DT_UNKNOWN) | |
1166 | return dtype; | |
443e061a LT |
1167 | dtype = get_index_dtype(path, len); |
1168 | if (dtype != DT_UNKNOWN) | |
1169 | return dtype; | |
1170 | if (lstat(path, &st)) | |
07134421 LT |
1171 | return dtype; |
1172 | if (S_ISREG(st.st_mode)) | |
1173 | return DT_REG; | |
1174 | if (S_ISDIR(st.st_mode)) | |
1175 | return DT_DIR; | |
1176 | if (S_ISLNK(st.st_mode)) | |
1177 | return DT_LNK; | |
1178 | return dtype; | |
1179 | } | |
1180 | ||
53cc5356 JH |
1181 | enum path_treatment { |
1182 | path_ignored, | |
1183 | path_handled, | |
4b05548f | 1184 | path_recurse |
53cc5356 JH |
1185 | }; |
1186 | ||
16e2cfa9 | 1187 | static enum path_treatment treat_one_path(struct dir_struct *dir, |
49dc2cc2 | 1188 | struct strbuf *path, |
16e2cfa9 JH |
1189 | const struct path_simplify *simplify, |
1190 | int dtype, struct dirent *de) | |
53cc5356 | 1191 | { |
8aaf8d77 KB |
1192 | int exclude; |
1193 | if (dtype == DT_UNKNOWN) | |
1194 | dtype = get_dtype(de, path->buf, path->len); | |
1195 | ||
1196 | /* Always exclude indexed files */ | |
1197 | if (dtype != DT_DIR && | |
1198 | cache_name_exists(path->buf, path->len, ignore_case)) | |
1199 | return path_ignored; | |
1200 | ||
1201 | exclude = is_excluded(dir, path->buf, &dtype); | |
53cc5356 | 1202 | if (exclude && (dir->flags & DIR_COLLECT_IGNORED) |
49dc2cc2 RS |
1203 | && exclude_matches_pathspec(path->buf, path->len, simplify)) |
1204 | dir_add_ignored(dir, path->buf, path->len); | |
53cc5356 JH |
1205 | |
1206 | /* | |
1207 | * Excluded? If we don't explicitly want to show | |
1208 | * ignored files, ignore it | |
1209 | */ | |
1210 | if (exclude && !(dir->flags & DIR_SHOW_IGNORED)) | |
1211 | return path_ignored; | |
1212 | ||
53cc5356 JH |
1213 | switch (dtype) { |
1214 | default: | |
1215 | return path_ignored; | |
1216 | case DT_DIR: | |
49dc2cc2 | 1217 | strbuf_addch(path, '/'); |
721ac4ed | 1218 | switch (treat_directory(dir, path->buf, path->len, exclude, simplify)) { |
53cc5356 | 1219 | case show_directory: |
53cc5356 JH |
1220 | break; |
1221 | case recurse_into_directory: | |
1222 | return path_recurse; | |
1223 | case ignore_directory: | |
1224 | return path_ignored; | |
1225 | } | |
1226 | break; | |
1227 | case DT_REG: | |
1228 | case DT_LNK: | |
8aaf8d77 | 1229 | if (exclude == !(dir->flags & DIR_SHOW_IGNORED)) |
721ac4ed | 1230 | return path_ignored; |
b07bc8c8 | 1231 | break; |
53cc5356 JH |
1232 | } |
1233 | return path_handled; | |
1234 | } | |
1235 | ||
16e2cfa9 JH |
1236 | static enum path_treatment treat_path(struct dir_struct *dir, |
1237 | struct dirent *de, | |
49dc2cc2 | 1238 | struct strbuf *path, |
16e2cfa9 | 1239 | int baselen, |
49dc2cc2 | 1240 | const struct path_simplify *simplify) |
16e2cfa9 JH |
1241 | { |
1242 | int dtype; | |
1243 | ||
1244 | if (is_dot_or_dotdot(de->d_name) || !strcmp(de->d_name, ".git")) | |
1245 | return path_ignored; | |
49dc2cc2 RS |
1246 | strbuf_setlen(path, baselen); |
1247 | strbuf_addstr(path, de->d_name); | |
1248 | if (simplify_away(path->buf, path->len, simplify)) | |
16e2cfa9 JH |
1249 | return path_ignored; |
1250 | ||
1251 | dtype = DTYPE(de); | |
49dc2cc2 | 1252 | return treat_one_path(dir, path, simplify, dtype, de); |
16e2cfa9 JH |
1253 | } |
1254 | ||
453ec4bd LT |
1255 | /* |
1256 | * Read a directory tree. We currently ignore anything but | |
1257 | * directories, regular files and symlinks. That's because git | |
1258 | * doesn't handle them at all yet. Maybe that will change some | |
1259 | * day. | |
1260 | * | |
1261 | * Also, we ignore the name ".git" (even if it is not a directory). | |
1262 | * That likely will not change. | |
1263 | */ | |
53cc5356 JH |
1264 | static int read_directory_recursive(struct dir_struct *dir, |
1265 | const char *base, int baselen, | |
1266 | int check_only, | |
1267 | const struct path_simplify *simplify) | |
453ec4bd | 1268 | { |
1528d247 | 1269 | DIR *fdir; |
453ec4bd | 1270 | int contents = 0; |
02cb6753 | 1271 | struct dirent *de; |
bef36921 | 1272 | struct strbuf path = STRBUF_INIT; |
453ec4bd | 1273 | |
bef36921 | 1274 | strbuf_add(&path, base, baselen); |
02cb6753 | 1275 | |
1528d247 RS |
1276 | fdir = opendir(path.len ? path.buf : "."); |
1277 | if (!fdir) | |
1278 | goto out; | |
1279 | ||
02cb6753 | 1280 | while ((de = readdir(fdir)) != NULL) { |
bef36921 | 1281 | switch (treat_path(dir, de, &path, baselen, simplify)) { |
02cb6753 | 1282 | case path_recurse: |
bef36921 | 1283 | contents += read_directory_recursive(dir, path.buf, |
289ff559 | 1284 | path.len, check_only, simplify); |
02cb6753 NTND |
1285 | continue; |
1286 | case path_ignored: | |
1287 | continue; | |
1288 | case path_handled: | |
1289 | break; | |
453ec4bd | 1290 | } |
02cb6753 NTND |
1291 | contents++; |
1292 | if (check_only) | |
1528d247 RS |
1293 | break; |
1294 | dir_add_name(dir, path.buf, path.len); | |
453ec4bd | 1295 | } |
02cb6753 | 1296 | closedir(fdir); |
1528d247 | 1297 | out: |
bef36921 | 1298 | strbuf_release(&path); |
453ec4bd LT |
1299 | |
1300 | return contents; | |
1301 | } | |
1302 | ||
1303 | static int cmp_name(const void *p1, const void *p2) | |
1304 | { | |
1305 | const struct dir_entry *e1 = *(const struct dir_entry **)p1; | |
1306 | const struct dir_entry *e2 = *(const struct dir_entry **)p2; | |
1307 | ||
1308 | return cache_name_compare(e1->name, e1->len, | |
1309 | e2->name, e2->len); | |
1310 | } | |
1311 | ||
9fc42d60 LT |
1312 | static struct path_simplify *create_simplify(const char **pathspec) |
1313 | { | |
1314 | int nr, alloc = 0; | |
1315 | struct path_simplify *simplify = NULL; | |
1316 | ||
1317 | if (!pathspec) | |
1318 | return NULL; | |
1319 | ||
1320 | for (nr = 0 ; ; nr++) { | |
1321 | const char *match; | |
1322 | if (nr >= alloc) { | |
1323 | alloc = alloc_nr(alloc); | |
1324 | simplify = xrealloc(simplify, alloc * sizeof(*simplify)); | |
1325 | } | |
1326 | match = *pathspec++; | |
1327 | if (!match) | |
1328 | break; | |
1329 | simplify[nr].path = match; | |
1330 | simplify[nr].len = simple_length(match); | |
1331 | } | |
1332 | simplify[nr].path = NULL; | |
1333 | simplify[nr].len = 0; | |
1334 | return simplify; | |
1335 | } | |
1336 | ||
1337 | static void free_simplify(struct path_simplify *simplify) | |
1338 | { | |
8e0f7003 | 1339 | free(simplify); |
9fc42d60 LT |
1340 | } |
1341 | ||
48ffef96 JH |
1342 | static int treat_leading_path(struct dir_struct *dir, |
1343 | const char *path, int len, | |
1344 | const struct path_simplify *simplify) | |
1345 | { | |
49dc2cc2 RS |
1346 | struct strbuf sb = STRBUF_INIT; |
1347 | int baselen, rc = 0; | |
48ffef96 | 1348 | const char *cp; |
be8a84c5 | 1349 | int old_flags = dir->flags; |
48ffef96 JH |
1350 | |
1351 | while (len && path[len - 1] == '/') | |
1352 | len--; | |
1353 | if (!len) | |
1354 | return 1; | |
1355 | baselen = 0; | |
be8a84c5 | 1356 | dir->flags &= ~DIR_SHOW_OTHER_DIRECTORIES; |
48ffef96 JH |
1357 | while (1) { |
1358 | cp = path + baselen + !!baselen; | |
1359 | cp = memchr(cp, '/', path + len - cp); | |
1360 | if (!cp) | |
1361 | baselen = len; | |
1362 | else | |
1363 | baselen = cp - path; | |
49dc2cc2 RS |
1364 | strbuf_setlen(&sb, 0); |
1365 | strbuf_add(&sb, path, baselen); | |
1366 | if (!is_directory(sb.buf)) | |
1367 | break; | |
1368 | if (simplify_away(sb.buf, sb.len, simplify)) | |
1369 | break; | |
1370 | if (treat_one_path(dir, &sb, simplify, | |
48ffef96 | 1371 | DT_DIR, NULL) == path_ignored) |
49dc2cc2 RS |
1372 | break; /* do not recurse into it */ |
1373 | if (len <= baselen) { | |
1374 | rc = 1; | |
1375 | break; /* finished checking */ | |
1376 | } | |
48ffef96 | 1377 | } |
49dc2cc2 | 1378 | strbuf_release(&sb); |
be8a84c5 | 1379 | dir->flags = old_flags; |
49dc2cc2 | 1380 | return rc; |
48ffef96 JH |
1381 | } |
1382 | ||
dba2e203 | 1383 | int read_directory(struct dir_struct *dir, const char *path, int len, const char **pathspec) |
9fc42d60 | 1384 | { |
725b0605 | 1385 | struct path_simplify *simplify; |
b4189aa8 | 1386 | |
dba2e203 | 1387 | if (has_symlink_leading_path(path, len)) |
725b0605 JH |
1388 | return dir->nr; |
1389 | ||
1390 | simplify = create_simplify(pathspec); | |
48ffef96 JH |
1391 | if (!len || treat_leading_path(dir, path, len, simplify)) |
1392 | read_directory_recursive(dir, path, len, 0, simplify); | |
9fc42d60 | 1393 | free_simplify(simplify); |
453ec4bd | 1394 | qsort(dir->entries, dir->nr, sizeof(struct dir_entry *), cmp_name); |
2abd31b0 | 1395 | qsort(dir->ignored, dir->ignored_nr, sizeof(struct dir_entry *), cmp_name); |
453ec4bd LT |
1396 | return dir->nr; |
1397 | } | |
c91f0d92 | 1398 | |
686a4a06 | 1399 | int file_exists(const char *f) |
c91f0d92 | 1400 | { |
686a4a06 | 1401 | struct stat sb; |
a50f9fc5 | 1402 | return lstat(f, &sb) == 0; |
c91f0d92 | 1403 | } |
e6636747 JS |
1404 | |
1405 | /* | |
9b125da4 NTND |
1406 | * Given two normalized paths (a trailing slash is ok), if subdir is |
1407 | * outside dir, return -1. Otherwise return the offset in subdir that | |
1408 | * can be used as relative path to dir. | |
e6636747 | 1409 | */ |
9b125da4 | 1410 | int dir_inside_of(const char *subdir, const char *dir) |
e6636747 | 1411 | { |
9b125da4 | 1412 | int offset = 0; |
e6636747 | 1413 | |
9b125da4 | 1414 | assert(dir && subdir && *dir && *subdir); |
e6636747 | 1415 | |
9b125da4 | 1416 | while (*dir && *subdir && *dir == *subdir) { |
e6636747 | 1417 | dir++; |
9b125da4 NTND |
1418 | subdir++; |
1419 | offset++; | |
490544b1 | 1420 | } |
9b125da4 NTND |
1421 | |
1422 | /* hel[p]/me vs hel[l]/yeah */ | |
1423 | if (*dir && *subdir) | |
1424 | return -1; | |
1425 | ||
1426 | if (!*subdir) | |
1427 | return !*dir ? offset : -1; /* same dir */ | |
1428 | ||
1429 | /* foo/[b]ar vs foo/[] */ | |
1430 | if (is_dir_sep(dir[-1])) | |
1431 | return is_dir_sep(subdir[-1]) ? offset : -1; | |
1432 | ||
1433 | /* foo[/]bar vs foo[] */ | |
1434 | return is_dir_sep(*subdir) ? offset + 1 : -1; | |
e6636747 JS |
1435 | } |
1436 | ||
1437 | int is_inside_dir(const char *dir) | |
1438 | { | |
b892913d NTND |
1439 | char cwd[PATH_MAX]; |
1440 | if (!dir) | |
1441 | return 0; | |
1442 | if (!getcwd(cwd, sizeof(cwd))) | |
1443 | die_errno("can't find the current directory"); | |
1444 | return dir_inside_of(cwd, dir) >= 0; | |
e6636747 | 1445 | } |
7155b727 | 1446 | |
55892d23 AP |
1447 | int is_empty_dir(const char *path) |
1448 | { | |
1449 | DIR *dir = opendir(path); | |
1450 | struct dirent *e; | |
1451 | int ret = 1; | |
1452 | ||
1453 | if (!dir) | |
1454 | return 0; | |
1455 | ||
1456 | while ((e = readdir(dir)) != NULL) | |
1457 | if (!is_dot_or_dotdot(e->d_name)) { | |
1458 | ret = 0; | |
1459 | break; | |
1460 | } | |
1461 | ||
1462 | closedir(dir); | |
1463 | return ret; | |
1464 | } | |
1465 | ||
ae2f203e | 1466 | static int remove_dir_recurse(struct strbuf *path, int flag, int *kept_up) |
7155b727 | 1467 | { |
a0f4afbe | 1468 | DIR *dir; |
7155b727 | 1469 | struct dirent *e; |
ae2f203e | 1470 | int ret = 0, original_len = path->len, len, kept_down = 0; |
a0f4afbe | 1471 | int only_empty = (flag & REMOVE_DIR_EMPTY_ONLY); |
c844a803 | 1472 | int keep_toplevel = (flag & REMOVE_DIR_KEEP_TOPLEVEL); |
a0f4afbe | 1473 | unsigned char submodule_head[20]; |
7155b727 | 1474 | |
a0f4afbe | 1475 | if ((flag & REMOVE_DIR_KEEP_NESTED_GIT) && |
ae2f203e | 1476 | !resolve_gitlink_ref(path->buf, "HEAD", submodule_head)) { |
a0f4afbe | 1477 | /* Do not descend and nuke a nested git work tree. */ |
ae2f203e JH |
1478 | if (kept_up) |
1479 | *kept_up = 1; | |
a0f4afbe | 1480 | return 0; |
ae2f203e | 1481 | } |
a0f4afbe | 1482 | |
ae2f203e | 1483 | flag &= ~REMOVE_DIR_KEEP_TOPLEVEL; |
a0f4afbe | 1484 | dir = opendir(path->buf); |
c844a803 | 1485 | if (!dir) { |
ae2f203e | 1486 | /* an empty dir could be removed even if it is unreadble */ |
c844a803 JH |
1487 | if (!keep_toplevel) |
1488 | return rmdir(path->buf); | |
1489 | else | |
1490 | return -1; | |
1491 | } | |
7155b727 JS |
1492 | if (path->buf[original_len - 1] != '/') |
1493 | strbuf_addch(path, '/'); | |
1494 | ||
1495 | len = path->len; | |
1496 | while ((e = readdir(dir)) != NULL) { | |
1497 | struct stat st; | |
8ca12c0d AP |
1498 | if (is_dot_or_dotdot(e->d_name)) |
1499 | continue; | |
7155b727 JS |
1500 | |
1501 | strbuf_setlen(path, len); | |
1502 | strbuf_addstr(path, e->d_name); | |
1503 | if (lstat(path->buf, &st)) | |
1504 | ; /* fall thru */ | |
1505 | else if (S_ISDIR(st.st_mode)) { | |
ae2f203e | 1506 | if (!remove_dir_recurse(path, flag, &kept_down)) |
7155b727 JS |
1507 | continue; /* happy */ |
1508 | } else if (!only_empty && !unlink(path->buf)) | |
1509 | continue; /* happy, too */ | |
1510 | ||
1511 | /* path too long, stat fails, or non-directory still exists */ | |
1512 | ret = -1; | |
1513 | break; | |
1514 | } | |
1515 | closedir(dir); | |
1516 | ||
1517 | strbuf_setlen(path, original_len); | |
ae2f203e | 1518 | if (!ret && !keep_toplevel && !kept_down) |
7155b727 | 1519 | ret = rmdir(path->buf); |
ae2f203e JH |
1520 | else if (kept_up) |
1521 | /* | |
1522 | * report the uplevel that it is not an error that we | |
1523 | * did not rmdir() our directory. | |
1524 | */ | |
1525 | *kept_up = !ret; | |
7155b727 JS |
1526 | return ret; |
1527 | } | |
039bc64e | 1528 | |
ae2f203e JH |
1529 | int remove_dir_recursively(struct strbuf *path, int flag) |
1530 | { | |
1531 | return remove_dir_recurse(path, flag, NULL); | |
1532 | } | |
1533 | ||
039bc64e JH |
1534 | void setup_standard_excludes(struct dir_struct *dir) |
1535 | { | |
1536 | const char *path; | |
dc79687e | 1537 | char *xdg_path; |
039bc64e JH |
1538 | |
1539 | dir->exclude_per_dir = ".gitignore"; | |
1540 | path = git_path("info/exclude"); | |
dc79687e HKNN |
1541 | if (!excludes_file) { |
1542 | home_config_paths(NULL, &xdg_path, "ignore"); | |
1543 | excludes_file = xdg_path; | |
1544 | } | |
69660731 | 1545 | if (!access_or_warn(path, R_OK)) |
039bc64e | 1546 | add_excludes_from_file(dir, path); |
69660731 | 1547 | if (excludes_file && !access_or_warn(excludes_file, R_OK)) |
039bc64e JH |
1548 | add_excludes_from_file(dir, excludes_file); |
1549 | } | |
4a92d1bf AR |
1550 | |
1551 | int remove_path(const char *name) | |
1552 | { | |
1553 | char *slash; | |
1554 | ||
9a6728d4 | 1555 | if (unlink(name) && errno != ENOENT && errno != ENOTDIR) |
4a92d1bf AR |
1556 | return -1; |
1557 | ||
1558 | slash = strrchr(name, '/'); | |
1559 | if (slash) { | |
1560 | char *dirs = xstrdup(name); | |
1561 | slash = dirs + (slash - name); | |
1562 | do { | |
1563 | *slash = '\0'; | |
3fc0d131 | 1564 | } while (rmdir(dirs) == 0 && (slash = strrchr(dirs, '/'))); |
4a92d1bf AR |
1565 | free(dirs); |
1566 | } | |
1567 | return 0; | |
1568 | } | |
1569 | ||
86e4ca69 NTND |
1570 | static int pathspec_item_cmp(const void *a_, const void *b_) |
1571 | { | |
1572 | struct pathspec_item *a, *b; | |
1573 | ||
1574 | a = (struct pathspec_item *)a_; | |
1575 | b = (struct pathspec_item *)b_; | |
1576 | return strcmp(a->match, b->match); | |
1577 | } | |
1578 | ||
0602f3e9 NTND |
1579 | int init_pathspec(struct pathspec *pathspec, const char **paths) |
1580 | { | |
1581 | const char **p = paths; | |
1582 | int i; | |
1583 | ||
1584 | memset(pathspec, 0, sizeof(*pathspec)); | |
1585 | if (!p) | |
1586 | return 0; | |
1587 | while (*p) | |
1588 | p++; | |
1589 | pathspec->raw = paths; | |
1590 | pathspec->nr = p - paths; | |
1591 | if (!pathspec->nr) | |
1592 | return 0; | |
1593 | ||
1594 | pathspec->items = xmalloc(sizeof(struct pathspec_item)*pathspec->nr); | |
1595 | for (i = 0; i < pathspec->nr; i++) { | |
1596 | struct pathspec_item *item = pathspec->items+i; | |
1597 | const char *path = paths[i]; | |
1598 | ||
1599 | item->match = path; | |
1600 | item->len = strlen(path); | |
8c6abbcd | 1601 | item->flags = 0; |
971e829c JH |
1602 | if (limit_pathspec_to_literal()) { |
1603 | item->nowildcard_len = item->len; | |
1604 | } else { | |
1605 | item->nowildcard_len = simple_length(path); | |
1606 | if (item->nowildcard_len < item->len) { | |
1607 | pathspec->has_wildcard = 1; | |
1608 | if (path[item->nowildcard_len] == '*' && | |
1609 | no_wildcard(path + item->nowildcard_len + 1)) | |
1610 | item->flags |= PATHSPEC_ONESTAR; | |
1611 | } | |
8c6abbcd | 1612 | } |
0602f3e9 | 1613 | } |
86e4ca69 NTND |
1614 | |
1615 | qsort(pathspec->items, pathspec->nr, | |
1616 | sizeof(struct pathspec_item), pathspec_item_cmp); | |
1617 | ||
0602f3e9 NTND |
1618 | return 0; |
1619 | } | |
1620 | ||
1621 | void free_pathspec(struct pathspec *pathspec) | |
1622 | { | |
1623 | free(pathspec->items); | |
1624 | pathspec->items = NULL; | |
1625 | } | |
823ab40f JK |
1626 | |
1627 | int limit_pathspec_to_literal(void) | |
1628 | { | |
1629 | static int flag = -1; | |
1630 | if (flag < 0) | |
1631 | flag = git_env_bool(GIT_LITERAL_PATHSPECS_ENVIRONMENT, 0); | |
1632 | return flag; | |
1633 | } | |
a39b15b4 | 1634 | |
270be816 AS |
1635 | /* |
1636 | * Frees memory within dir which was allocated for exclude lists and | |
1637 | * the exclude_stack. Does not free dir itself. | |
1638 | */ | |
1639 | void clear_directory(struct dir_struct *dir) | |
1640 | { | |
1641 | int i, j; | |
1642 | struct exclude_list_group *group; | |
1643 | struct exclude_list *el; | |
1644 | struct exclude_stack *stk; | |
1645 | ||
1646 | for (i = EXC_CMDL; i <= EXC_FILE; i++) { | |
1647 | group = &dir->exclude_list_group[i]; | |
1648 | for (j = 0; j < group->nr; j++) { | |
1649 | el = &group->el[j]; | |
1650 | if (i == EXC_DIRS) | |
1651 | free((char *)el->src); | |
1652 | clear_exclude_list(el); | |
1653 | } | |
1654 | free(group->el); | |
1655 | } | |
1656 | ||
1657 | stk = dir->exclude_stack; | |
1658 | while (stk) { | |
1659 | struct exclude_stack *prev = stk->prev; | |
1660 | free(stk); | |
1661 | stk = prev; | |
1662 | } | |
1663 | } |