]>
Commit | Line | Data |
---|---|---|
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 |
12 | struct path_simplify { |
13 | int len; | |
14 | const char *path; | |
15 | }; | |
16 | ||
09595258 LT |
17 | static int read_directory_recursive(struct dir_struct *dir, |
18 | const char *path, const char *base, int baselen, | |
19 | int check_only, const struct path_simplify *simplify); | |
20 | ||
3c6a370b LT |
21 | int common_prefix(const char **pathspec) |
22 | { | |
23 | const char *path, *slash, *next; | |
24 | int prefix; | |
25 | ||
26 | if (!pathspec) | |
27 | return 0; | |
28 | ||
29 | path = *pathspec; | |
30 | slash = strrchr(path, '/'); | |
31 | if (!slash) | |
32 | return 0; | |
33 | ||
34 | prefix = slash - path + 1; | |
35 | while ((next = *++pathspec) != NULL) { | |
36 | int len = strlen(next); | |
37 | if (len >= prefix && !memcmp(path, next, len)) | |
38 | continue; | |
39 | for (;;) { | |
40 | if (!len) | |
41 | return 0; | |
42 | if (next[--len] != '/') | |
43 | continue; | |
44 | if (memcmp(path, next, len+1)) | |
45 | continue; | |
46 | prefix = len + 1; | |
47 | break; | |
48 | } | |
49 | } | |
50 | return prefix; | |
51 | } | |
52 | ||
e813d50e JH |
53 | /* |
54 | * Does 'match' matches the given name? | |
55 | * A match is found if | |
56 | * | |
57 | * (1) the 'match' string is leading directory of 'name', or | |
58 | * (2) the 'match' string is a wildcard and matches 'name', or | |
59 | * (3) the 'match' string is exactly the same as 'name'. | |
60 | * | |
61 | * and the return value tells which case it was. | |
62 | * | |
63 | * It returns 0 when there is no match. | |
64 | */ | |
3c6a370b LT |
65 | static int match_one(const char *match, const char *name, int namelen) |
66 | { | |
67 | int matchlen; | |
68 | ||
69 | /* If the match was just the prefix, we matched */ | |
70 | matchlen = strlen(match); | |
71 | if (!matchlen) | |
e813d50e | 72 | return MATCHED_RECURSIVELY; |
3c6a370b LT |
73 | |
74 | /* | |
75 | * If we don't match the matchstring exactly, | |
76 | * we need to match by fnmatch | |
77 | */ | |
78 | if (strncmp(match, name, matchlen)) | |
e813d50e | 79 | return !fnmatch(match, name, 0) ? MATCHED_FNMATCH : 0; |
3c6a370b | 80 | |
e813d50e JH |
81 | if (!name[matchlen]) |
82 | return MATCHED_EXACTLY; | |
83 | if (match[matchlen-1] == '/' || name[matchlen] == '/') | |
84 | return MATCHED_RECURSIVELY; | |
85 | return 0; | |
3c6a370b LT |
86 | } |
87 | ||
e813d50e JH |
88 | /* |
89 | * Given a name and a list of pathspecs, see if the name matches | |
90 | * any of the pathspecs. The caller is also interested in seeing | |
91 | * all pathspec matches some names it calls this function with | |
92 | * (otherwise the user could have mistyped the unmatched pathspec), | |
93 | * and a mark is left in seen[] array for pathspec element that | |
94 | * actually matched anything. | |
95 | */ | |
3c6a370b LT |
96 | int match_pathspec(const char **pathspec, const char *name, int namelen, int prefix, char *seen) |
97 | { | |
98 | int retval; | |
99 | const char *match; | |
100 | ||
101 | name += prefix; | |
102 | namelen -= prefix; | |
103 | ||
104 | for (retval = 0; (match = *pathspec++) != NULL; seen++) { | |
e813d50e JH |
105 | int how; |
106 | if (retval && *seen == MATCHED_EXACTLY) | |
3c6a370b LT |
107 | continue; |
108 | match += prefix; | |
e813d50e JH |
109 | how = match_one(match, name, namelen); |
110 | if (how) { | |
111 | if (retval < how) | |
112 | retval = how; | |
113 | if (*seen < how) | |
114 | *seen = how; | |
3c6a370b LT |
115 | } |
116 | } | |
117 | return retval; | |
118 | } | |
119 | ||
453ec4bd LT |
120 | void add_exclude(const char *string, const char *base, |
121 | int baselen, struct exclude_list *which) | |
122 | { | |
123 | struct exclude *x = xmalloc(sizeof (*x)); | |
124 | ||
125 | x->pattern = string; | |
126 | x->base = base; | |
127 | x->baselen = baselen; | |
128 | if (which->nr == which->alloc) { | |
129 | which->alloc = alloc_nr(which->alloc); | |
83572c1a JF |
130 | which->excludes = xrealloc(which->excludes, |
131 | which->alloc * sizeof(x)); | |
453ec4bd LT |
132 | } |
133 | which->excludes[which->nr++] = x; | |
134 | } | |
135 | ||
136 | static int add_excludes_from_file_1(const char *fname, | |
137 | const char *base, | |
138 | int baselen, | |
139 | struct exclude_list *which) | |
140 | { | |
c470701a | 141 | struct stat st; |
453ec4bd | 142 | int fd, i; |
dc49cd76 | 143 | size_t size; |
453ec4bd LT |
144 | char *buf, *entry; |
145 | ||
146 | fd = open(fname, O_RDONLY); | |
c470701a | 147 | if (fd < 0 || fstat(fd, &st) < 0) |
453ec4bd | 148 | goto err; |
dc49cd76 | 149 | size = xsize_t(st.st_size); |
453ec4bd LT |
150 | if (size == 0) { |
151 | close(fd); | |
152 | return 0; | |
153 | } | |
154 | buf = xmalloc(size+1); | |
93d26e4c | 155 | if (read_in_full(fd, buf, size) != size) |
453ec4bd LT |
156 | goto err; |
157 | close(fd); | |
158 | ||
159 | buf[size++] = '\n'; | |
160 | entry = buf; | |
161 | for (i = 0; i < size; i++) { | |
162 | if (buf[i] == '\n') { | |
163 | if (entry != buf + i && entry[0] != '#') { | |
164 | buf[i - (i && buf[i-1] == '\r')] = 0; | |
165 | add_exclude(entry, base, baselen, which); | |
166 | } | |
167 | entry = buf + i + 1; | |
168 | } | |
169 | } | |
170 | return 0; | |
171 | ||
172 | err: | |
173 | if (0 <= fd) | |
174 | close(fd); | |
175 | return -1; | |
176 | } | |
177 | ||
178 | void add_excludes_from_file(struct dir_struct *dir, const char *fname) | |
179 | { | |
180 | if (add_excludes_from_file_1(fname, "", 0, | |
181 | &dir->exclude_list[EXC_FILE]) < 0) | |
182 | die("cannot use %s as an exclude file", fname); | |
183 | } | |
184 | ||
f8a9d428 | 185 | int push_exclude_per_directory(struct dir_struct *dir, const char *base, int baselen) |
453ec4bd LT |
186 | { |
187 | char exclude_file[PATH_MAX]; | |
188 | struct exclude_list *el = &dir->exclude_list[EXC_DIRS]; | |
189 | int current_nr = el->nr; | |
190 | ||
191 | if (dir->exclude_per_dir) { | |
192 | memcpy(exclude_file, base, baselen); | |
193 | strcpy(exclude_file + baselen, dir->exclude_per_dir); | |
194 | add_excludes_from_file_1(exclude_file, base, baselen, el); | |
195 | } | |
196 | return current_nr; | |
197 | } | |
198 | ||
f8a9d428 | 199 | void pop_exclude_per_directory(struct dir_struct *dir, int stk) |
453ec4bd LT |
200 | { |
201 | struct exclude_list *el = &dir->exclude_list[EXC_DIRS]; | |
202 | ||
203 | while (stk < el->nr) | |
204 | free(el->excludes[--el->nr]); | |
205 | } | |
206 | ||
207 | /* Scan the list and let the last match determines the fate. | |
208 | * Return 1 for exclude, 0 for include and -1 for undecided. | |
209 | */ | |
210 | static int excluded_1(const char *pathname, | |
211 | int pathlen, | |
212 | struct exclude_list *el) | |
213 | { | |
214 | int i; | |
215 | ||
216 | if (el->nr) { | |
217 | for (i = el->nr - 1; 0 <= i; i--) { | |
218 | struct exclude *x = el->excludes[i]; | |
219 | const char *exclude = x->pattern; | |
220 | int to_exclude = 1; | |
221 | ||
222 | if (*exclude == '!') { | |
223 | to_exclude = 0; | |
224 | exclude++; | |
225 | } | |
226 | ||
227 | if (!strchr(exclude, '/')) { | |
228 | /* match basename */ | |
229 | const char *basename = strrchr(pathname, '/'); | |
230 | basename = (basename) ? basename+1 : pathname; | |
231 | if (fnmatch(exclude, basename, 0) == 0) | |
232 | return to_exclude; | |
233 | } | |
234 | else { | |
235 | /* match with FNM_PATHNAME: | |
236 | * exclude has base (baselen long) implicitly | |
237 | * in front of it. | |
238 | */ | |
239 | int baselen = x->baselen; | |
240 | if (*exclude == '/') | |
241 | exclude++; | |
242 | ||
243 | if (pathlen < baselen || | |
244 | (baselen && pathname[baselen-1] != '/') || | |
245 | strncmp(pathname, x->base, baselen)) | |
246 | continue; | |
247 | ||
248 | if (fnmatch(exclude, pathname+baselen, | |
249 | FNM_PATHNAME) == 0) | |
250 | return to_exclude; | |
251 | } | |
252 | } | |
253 | } | |
254 | return -1; /* undecided */ | |
255 | } | |
256 | ||
257 | int excluded(struct dir_struct *dir, const char *pathname) | |
258 | { | |
259 | int pathlen = strlen(pathname); | |
260 | int st; | |
261 | ||
262 | for (st = EXC_CMDL; st <= EXC_FILE; st++) { | |
263 | switch (excluded_1(pathname, pathlen, &dir->exclude_list[st])) { | |
264 | case 0: | |
265 | return 0; | |
266 | case 1: | |
267 | return 1; | |
268 | } | |
269 | } | |
270 | return 0; | |
271 | } | |
272 | ||
4d06f8ac | 273 | struct dir_entry *dir_add_name(struct dir_struct *dir, const char *pathname, int len) |
453ec4bd LT |
274 | { |
275 | struct dir_entry *ent; | |
276 | ||
277 | if (cache_name_pos(pathname, len) >= 0) | |
4d06f8ac | 278 | return NULL; |
453ec4bd LT |
279 | |
280 | if (dir->nr == dir->alloc) { | |
281 | int alloc = alloc_nr(dir->alloc); | |
282 | dir->alloc = alloc; | |
283 | dir->entries = xrealloc(dir->entries, alloc*sizeof(ent)); | |
284 | } | |
285 | ent = xmalloc(sizeof(*ent) + len + 1); | |
4d06f8ac | 286 | ent->ignored = ent->ignored_dir = 0; |
453ec4bd LT |
287 | ent->len = len; |
288 | memcpy(ent->name, pathname, len); | |
289 | ent->name[len] = 0; | |
290 | dir->entries[dir->nr++] = ent; | |
4d06f8ac | 291 | return ent; |
453ec4bd LT |
292 | } |
293 | ||
09595258 LT |
294 | enum exist_status { |
295 | index_nonexistent = 0, | |
296 | index_directory, | |
297 | index_gitdir, | |
298 | }; | |
299 | ||
300 | /* | |
301 | * The index sorts alphabetically by entry name, which | |
302 | * means that a gitlink sorts as '\0' at the end, while | |
303 | * a directory (which is defined not as an entry, but as | |
304 | * the files it contains) will sort with the '/' at the | |
305 | * end. | |
306 | */ | |
307 | static enum exist_status directory_exists_in_index(const char *dirname, int len) | |
453ec4bd LT |
308 | { |
309 | int pos = cache_name_pos(dirname, len); | |
09595258 LT |
310 | if (pos < 0) |
311 | pos = -pos-1; | |
312 | while (pos < active_nr) { | |
313 | struct cache_entry *ce = active_cache[pos++]; | |
314 | unsigned char endchar; | |
315 | ||
316 | if (strncmp(ce->name, dirname, len)) | |
317 | break; | |
318 | endchar = ce->name[len]; | |
319 | if (endchar > '/') | |
320 | break; | |
321 | if (endchar == '/') | |
322 | return index_directory; | |
323 | if (!endchar && S_ISDIRLNK(ntohl(ce->ce_mode))) | |
324 | return index_gitdir; | |
325 | } | |
326 | return index_nonexistent; | |
327 | } | |
328 | ||
329 | /* | |
330 | * When we find a directory when traversing the filesystem, we | |
331 | * have three distinct cases: | |
332 | * | |
333 | * - ignore it | |
334 | * - see it as a directory | |
335 | * - recurse into it | |
336 | * | |
337 | * and which one we choose depends on a combination of existing | |
338 | * git index contents and the flags passed into the directory | |
339 | * traversal routine. | |
340 | * | |
341 | * Case 1: If we *already* have entries in the index under that | |
342 | * directory name, we always recurse into the directory to see | |
343 | * all the files. | |
344 | * | |
345 | * Case 2: If we *already* have that directory name as a gitlink, | |
346 | * we always continue to see it as a gitlink, regardless of whether | |
347 | * there is an actual git directory there or not (it might not | |
348 | * be checked out as a subproject!) | |
349 | * | |
350 | * Case 3: if we didn't have it in the index previously, we | |
351 | * have a few sub-cases: | |
352 | * | |
353 | * (a) if "show_other_directories" is true, we show it as | |
354 | * just a directory, unless "hide_empty_directories" is | |
355 | * also true and the directory is empty, in which case | |
356 | * we just ignore it entirely. | |
357 | * (b) if it looks like a git directory, and we don't have | |
358 | * 'no_dirlinks' set we treat it as a gitlink, and show it | |
359 | * as a directory. | |
360 | * (c) otherwise, we recurse into it. | |
361 | */ | |
362 | enum directory_treatment { | |
363 | show_directory, | |
364 | ignore_directory, | |
365 | recurse_into_directory, | |
366 | }; | |
367 | ||
368 | static enum directory_treatment treat_directory(struct dir_struct *dir, | |
369 | const char *dirname, int len, | |
370 | const struct path_simplify *simplify) | |
371 | { | |
372 | /* The "len-1" is to strip the final '/' */ | |
373 | switch (directory_exists_in_index(dirname, len-1)) { | |
374 | case index_directory: | |
375 | return recurse_into_directory; | |
376 | ||
377 | case index_gitdir: | |
378 | return show_directory; | |
379 | ||
380 | case index_nonexistent: | |
381 | if (dir->show_other_directories) | |
382 | break; | |
383 | if (!dir->no_dirlinks) { | |
384 | unsigned char sha1[20]; | |
385 | if (resolve_gitlink_ref(dirname, "HEAD", sha1) == 0) | |
386 | return show_directory; | |
387 | } | |
388 | return recurse_into_directory; | |
389 | } | |
390 | ||
391 | /* This is the "show_other_directories" case */ | |
392 | if (!dir->hide_empty_directories) | |
393 | return show_directory; | |
394 | if (!read_directory_recursive(dir, dirname, dirname, len, 1, simplify)) | |
395 | return ignore_directory; | |
396 | return show_directory; | |
453ec4bd LT |
397 | } |
398 | ||
9fc42d60 LT |
399 | /* |
400 | * This is an inexact early pruning of any recursive directory | |
401 | * reading - if the path cannot possibly be in the pathspec, | |
402 | * return true, and we'll skip it early. | |
403 | */ | |
404 | static int simplify_away(const char *path, int pathlen, const struct path_simplify *simplify) | |
405 | { | |
406 | if (simplify) { | |
407 | for (;;) { | |
408 | const char *match = simplify->path; | |
409 | int len = simplify->len; | |
410 | ||
411 | if (!match) | |
412 | break; | |
413 | if (len > pathlen) | |
414 | len = pathlen; | |
415 | if (!memcmp(path, match, len)) | |
416 | return 0; | |
417 | simplify++; | |
418 | } | |
419 | return 1; | |
420 | } | |
421 | return 0; | |
422 | } | |
423 | ||
453ec4bd LT |
424 | /* |
425 | * Read a directory tree. We currently ignore anything but | |
426 | * directories, regular files and symlinks. That's because git | |
427 | * doesn't handle them at all yet. Maybe that will change some | |
428 | * day. | |
429 | * | |
430 | * Also, we ignore the name ".git" (even if it is not a directory). | |
431 | * That likely will not change. | |
432 | */ | |
9fc42d60 | 433 | static int read_directory_recursive(struct dir_struct *dir, const char *path, const char *base, int baselen, int check_only, const struct path_simplify *simplify) |
453ec4bd LT |
434 | { |
435 | DIR *fdir = opendir(path); | |
436 | int contents = 0; | |
437 | ||
438 | if (fdir) { | |
439 | int exclude_stk; | |
440 | struct dirent *de; | |
095c424d | 441 | char fullname[PATH_MAX + 1]; |
453ec4bd LT |
442 | memcpy(fullname, base, baselen); |
443 | ||
444 | exclude_stk = push_exclude_per_directory(dir, base, baselen); | |
445 | ||
446 | while ((de = readdir(fdir)) != NULL) { | |
447 | int len; | |
448 | ||
449 | if ((de->d_name[0] == '.') && | |
450 | (de->d_name[1] == 0 || | |
451 | !strcmp(de->d_name + 1, ".") || | |
452 | !strcmp(de->d_name + 1, "git"))) | |
453 | continue; | |
454 | len = strlen(de->d_name); | |
5d5cea67 LT |
455 | /* Ignore overly long pathnames! */ |
456 | if (len + baselen + 8 > sizeof(fullname)) | |
457 | continue; | |
453ec4bd | 458 | memcpy(fullname + baselen, de->d_name, len+1); |
9fc42d60 LT |
459 | if (simplify_away(fullname, baselen + len, simplify)) |
460 | continue; | |
c889763b JH |
461 | if (excluded(dir, fullname) != dir->show_ignored) { |
462 | if (!dir->show_ignored || DTYPE(de) != DT_DIR) { | |
463 | continue; | |
464 | } | |
465 | } | |
453ec4bd LT |
466 | |
467 | switch (DTYPE(de)) { | |
468 | struct stat st; | |
453ec4bd LT |
469 | default: |
470 | continue; | |
471 | case DT_UNKNOWN: | |
472 | if (lstat(fullname, &st)) | |
473 | continue; | |
474 | if (S_ISREG(st.st_mode) || S_ISLNK(st.st_mode)) | |
475 | break; | |
476 | if (!S_ISDIR(st.st_mode)) | |
477 | continue; | |
478 | /* fallthrough */ | |
479 | case DT_DIR: | |
480 | memcpy(fullname + baselen + len, "/", 2); | |
481 | len++; | |
09595258 LT |
482 | switch (treat_directory(dir, fullname, baselen + len, simplify)) { |
483 | case show_directory: | |
453ec4bd | 484 | break; |
09595258 LT |
485 | case recurse_into_directory: |
486 | contents += read_directory_recursive(dir, | |
487 | fullname, fullname, baselen + len, 0, simplify); | |
488 | continue; | |
489 | case ignore_directory: | |
490 | continue; | |
453ec4bd | 491 | } |
09595258 | 492 | break; |
453ec4bd LT |
493 | case DT_REG: |
494 | case DT_LNK: | |
495 | break; | |
496 | } | |
453ec4bd | 497 | contents++; |
07ccbff8 JS |
498 | if (check_only) |
499 | goto exit_early; | |
500 | else | |
4d06f8ac | 501 | dir_add_name(dir, fullname, baselen + len); |
453ec4bd | 502 | } |
07ccbff8 | 503 | exit_early: |
453ec4bd LT |
504 | closedir(fdir); |
505 | ||
506 | pop_exclude_per_directory(dir, exclude_stk); | |
507 | } | |
508 | ||
509 | return contents; | |
510 | } | |
511 | ||
512 | static int cmp_name(const void *p1, const void *p2) | |
513 | { | |
514 | const struct dir_entry *e1 = *(const struct dir_entry **)p1; | |
515 | const struct dir_entry *e2 = *(const struct dir_entry **)p2; | |
516 | ||
517 | return cache_name_compare(e1->name, e1->len, | |
518 | e2->name, e2->len); | |
519 | } | |
520 | ||
9fc42d60 LT |
521 | /* |
522 | * Return the length of the "simple" part of a path match limiter. | |
523 | */ | |
524 | static int simple_length(const char *match) | |
453ec4bd | 525 | { |
9fc42d60 LT |
526 | const char special[256] = { |
527 | [0] = 1, ['?'] = 1, | |
528 | ['\\'] = 1, ['*'] = 1, | |
529 | ['['] = 1 | |
530 | }; | |
531 | int len = -1; | |
532 | ||
533 | for (;;) { | |
534 | unsigned char c = *match++; | |
535 | len++; | |
536 | if (special[c]) | |
537 | return len; | |
538 | } | |
539 | } | |
540 | ||
541 | static struct path_simplify *create_simplify(const char **pathspec) | |
542 | { | |
543 | int nr, alloc = 0; | |
544 | struct path_simplify *simplify = NULL; | |
545 | ||
546 | if (!pathspec) | |
547 | return NULL; | |
548 | ||
549 | for (nr = 0 ; ; nr++) { | |
550 | const char *match; | |
551 | if (nr >= alloc) { | |
552 | alloc = alloc_nr(alloc); | |
553 | simplify = xrealloc(simplify, alloc * sizeof(*simplify)); | |
554 | } | |
555 | match = *pathspec++; | |
556 | if (!match) | |
557 | break; | |
558 | simplify[nr].path = match; | |
559 | simplify[nr].len = simple_length(match); | |
560 | } | |
561 | simplify[nr].path = NULL; | |
562 | simplify[nr].len = 0; | |
563 | return simplify; | |
564 | } | |
565 | ||
566 | static void free_simplify(struct path_simplify *simplify) | |
567 | { | |
568 | if (simplify) | |
569 | free(simplify); | |
570 | } | |
571 | ||
572 | int read_directory(struct dir_struct *dir, const char *path, const char *base, int baselen, const char **pathspec) | |
573 | { | |
574 | struct path_simplify *simplify = create_simplify(pathspec); | |
575 | ||
b4189aa8 LT |
576 | /* |
577 | * Make sure to do the per-directory exclude for all the | |
578 | * directories leading up to our base. | |
579 | */ | |
580 | if (baselen) { | |
581 | if (dir->exclude_per_dir) { | |
582 | char *p, *pp = xmalloc(baselen+1); | |
583 | memcpy(pp, base, baselen+1); | |
584 | p = pp; | |
585 | while (1) { | |
586 | char save = *p; | |
587 | *p = 0; | |
588 | push_exclude_per_directory(dir, pp, p-pp); | |
589 | *p++ = save; | |
590 | if (!save) | |
591 | break; | |
592 | p = strchr(p, '/'); | |
593 | if (p) | |
594 | p++; | |
595 | else | |
596 | p = pp + baselen; | |
597 | } | |
598 | free(pp); | |
599 | } | |
600 | } | |
601 | ||
9fc42d60 LT |
602 | read_directory_recursive(dir, path, base, baselen, 0, simplify); |
603 | free_simplify(simplify); | |
453ec4bd LT |
604 | qsort(dir->entries, dir->nr, sizeof(struct dir_entry *), cmp_name); |
605 | return dir->nr; | |
606 | } | |
c91f0d92 JK |
607 | |
608 | int | |
609 | file_exists(const char *f) | |
610 | { | |
611 | struct stat sb; | |
612 | return stat(f, &sb) == 0; | |
613 | } |