]>
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 | 8 | #include "cache.h" |
b2141fc1 | 9 | #include "config.h" |
453ec4bd | 10 | #include "dir.h" |
cbd53a21 | 11 | #include "object-store.h" |
b0db7046 | 12 | #include "attr.h" |
09595258 | 13 | #include "refs.h" |
237ec6e4 | 14 | #include "wildmatch.h" |
64acde94 | 15 | #include "pathspec.h" |
dde843e7 | 16 | #include "utf8.h" |
83c094ad NTND |
17 | #include "varint.h" |
18 | #include "ewah/ewok.h" | |
883e248b | 19 | #include "fsmonitor.h" |
da62f786 | 20 | #include "submodule-config.h" |
453ec4bd | 21 | |
defd7c7b KB |
22 | /* |
23 | * Tells read_directory_recursive how a file or directory should be treated. | |
24 | * Values are ordered by significance, e.g. if a directory contains both | |
25 | * excluded and untracked files, it is listed as untracked because | |
26 | * path_untracked > path_excluded. | |
27 | */ | |
28 | enum path_treatment { | |
29 | path_none = 0, | |
30 | path_recurse, | |
31 | path_excluded, | |
32 | path_untracked | |
33 | }; | |
34 | ||
cf7c6148 NTND |
35 | /* |
36 | * Support data structure for our opendir/readdir/closedir wrappers | |
37 | */ | |
38 | struct cached_dir { | |
39 | DIR *fdir; | |
40 | struct untracked_cache_dir *untracked; | |
91a2288b NTND |
41 | int nr_files; |
42 | int nr_dirs; | |
43 | ||
ad6f2157 JK |
44 | const char *d_name; |
45 | int d_type; | |
91a2288b NTND |
46 | const char *file; |
47 | struct untracked_cache_dir *ucd; | |
cf7c6148 NTND |
48 | }; |
49 | ||
defd7c7b | 50 | static enum path_treatment read_directory_recursive(struct dir_struct *dir, |
0ef8e169 BW |
51 | struct index_state *istate, const char *path, int len, |
52 | struct untracked_cache_dir *untracked, | |
5aaa7fd3 | 53 | int check_only, int stop_at_first_file, const struct pathspec *pathspec); |
ad6f2157 JK |
54 | static int resolve_dtype(int dtype, struct index_state *istate, |
55 | const char *path, int len); | |
906fc557 | 56 | struct dirent *readdir_skip_dot_and_dotdot(DIR *dirp) |
b548f0f1 EN |
57 | { |
58 | struct dirent *e; | |
59 | ||
60 | while ((e = readdir(dirp)) != NULL) { | |
61 | if (!is_dot_or_dotdot(e->d_name)) | |
62 | break; | |
63 | } | |
64 | return e; | |
65 | } | |
66 | ||
e0556a92 PC |
67 | int count_slashes(const char *s) |
68 | { | |
69 | int cnt = 0; | |
70 | while (*s) | |
71 | if (*s++ == '/') | |
72 | cnt++; | |
73 | return cnt; | |
74 | } | |
75 | ||
ba0897e6 | 76 | int fspathcmp(const char *a, const char *b) |
8cf2a84e JJ |
77 | { |
78 | return ignore_case ? strcasecmp(a, b) : strcmp(a, b); | |
79 | } | |
80 | ||
cf2dc1c2 EW |
81 | int fspatheq(const char *a, const char *b) |
82 | { | |
83 | return !fspathcmp(a, b); | |
84 | } | |
85 | ||
ba0897e6 | 86 | int fspathncmp(const char *a, const char *b, size_t count) |
8cf2a84e JJ |
87 | { |
88 | return ignore_case ? strncasecmp(a, b, count) : strncmp(a, b, count); | |
89 | } | |
90 | ||
cf2dc1c2 EW |
91 | unsigned int fspathhash(const char *str) |
92 | { | |
93 | return ignore_case ? strihash(str) : strhash(str); | |
94 | } | |
95 | ||
1f26ce61 CB |
96 | int git_fnmatch(const struct pathspec_item *item, |
97 | const char *pattern, const char *string, | |
98 | int prefix) | |
5d74762d | 99 | { |
5d74762d | 100 | if (prefix > 0) { |
93d93537 | 101 | if (ps_strncmp(item, pattern, string, prefix)) |
eb07894f | 102 | return WM_NOMATCH; |
5d74762d NTND |
103 | pattern += prefix; |
104 | string += prefix; | |
105 | } | |
bd30c2e4 | 106 | if (item->flags & PATHSPEC_ONESTAR) { |
8c6abbcd NTND |
107 | int pattern_len = strlen(++pattern); |
108 | int string_len = strlen(string); | |
109 | return string_len < pattern_len || | |
93d93537 NTND |
110 | ps_strcmp(item, pattern, |
111 | string + string_len - pattern_len); | |
8c6abbcd | 112 | } |
bd30c2e4 | 113 | if (item->magic & PATHSPEC_GLOB) |
93d93537 NTND |
114 | return wildmatch(pattern, string, |
115 | WM_PATHNAME | | |
55d34269 | 116 | (item->magic & PATHSPEC_ICASE ? WM_CASEFOLD : 0)); |
bd30c2e4 NTND |
117 | else |
118 | /* wildmatch has not learned no FNM_PATHNAME mode yet */ | |
eb07894f | 119 | return wildmatch(pattern, string, |
55d34269 | 120 | item->magic & PATHSPEC_ICASE ? WM_CASEFOLD : 0); |
5d74762d NTND |
121 | } |
122 | ||
0b6e56df JH |
123 | static int fnmatch_icase_mem(const char *pattern, int patternlen, |
124 | const char *string, int stringlen, | |
125 | int flags) | |
126 | { | |
127 | int match_status; | |
128 | struct strbuf pat_buf = STRBUF_INIT; | |
129 | struct strbuf str_buf = STRBUF_INIT; | |
130 | const char *use_pat = pattern; | |
131 | const char *use_str = string; | |
132 | ||
133 | if (pattern[patternlen]) { | |
134 | strbuf_add(&pat_buf, pattern, patternlen); | |
135 | use_pat = pat_buf.buf; | |
136 | } | |
137 | if (string[stringlen]) { | |
138 | strbuf_add(&str_buf, string, stringlen); | |
139 | use_str = str_buf.buf; | |
140 | } | |
141 | ||
f30366b2 JH |
142 | if (ignore_case) |
143 | flags |= WM_CASEFOLD; | |
55d34269 | 144 | match_status = wildmatch(use_pat, use_str, flags); |
0b6e56df JH |
145 | |
146 | strbuf_release(&pat_buf); | |
147 | strbuf_release(&str_buf); | |
148 | ||
149 | return match_status; | |
150 | } | |
151 | ||
827f4d6c | 152 | static size_t common_prefix_len(const struct pathspec *pathspec) |
3c6a370b | 153 | { |
827f4d6c | 154 | int n; |
4a085b16 | 155 | size_t max = 0; |
3c6a370b | 156 | |
93d93537 NTND |
157 | /* |
158 | * ":(icase)path" is treated as a pathspec full of | |
159 | * wildcard. In other words, only prefix is considered common | |
160 | * prefix. If the pathspec is abc/foo abc/bar, running in | |
bbbb6b0b | 161 | * subdir xyz, the common prefix is still xyz, not xyz/abc as |
93d93537 NTND |
162 | * in non-:(icase). |
163 | */ | |
5c6933d2 NTND |
164 | GUARD_PATHSPEC(pathspec, |
165 | PATHSPEC_FROMTOP | | |
166 | PATHSPEC_MAXDEPTH | | |
bd30c2e4 | 167 | PATHSPEC_LITERAL | |
93d93537 | 168 | PATHSPEC_GLOB | |
ef79b1f8 | 169 | PATHSPEC_ICASE | |
b0db7046 BW |
170 | PATHSPEC_EXCLUDE | |
171 | PATHSPEC_ATTR); | |
4a085b16 | 172 | |
827f4d6c | 173 | for (n = 0; n < pathspec->nr; n++) { |
93d93537 | 174 | size_t i = 0, len = 0, item_len; |
ef79b1f8 NTND |
175 | if (pathspec->items[n].magic & PATHSPEC_EXCLUDE) |
176 | continue; | |
93d93537 NTND |
177 | if (pathspec->items[n].magic & PATHSPEC_ICASE) |
178 | item_len = pathspec->items[n].prefix; | |
179 | else | |
180 | item_len = pathspec->items[n].nowildcard_len; | |
181 | while (i < item_len && (n == 0 || i < max)) { | |
827f4d6c NTND |
182 | char c = pathspec->items[n].match[i]; |
183 | if (c != pathspec->items[0].match[i]) | |
4a085b16 JH |
184 | break; |
185 | if (c == '/') | |
186 | len = i + 1; | |
827f4d6c | 187 | i++; |
4a085b16 | 188 | } |
827f4d6c | 189 | if (n == 0 || len < max) { |
4a085b16 JH |
190 | max = len; |
191 | if (!max) | |
192 | break; | |
193 | } | |
3c6a370b | 194 | } |
4a085b16 | 195 | return max; |
3c6a370b LT |
196 | } |
197 | ||
f950eb95 CB |
198 | /* |
199 | * Returns a copy of the longest leading path common among all | |
200 | * pathspecs. | |
201 | */ | |
827f4d6c | 202 | char *common_prefix(const struct pathspec *pathspec) |
f950eb95 CB |
203 | { |
204 | unsigned long len = common_prefix_len(pathspec); | |
205 | ||
827f4d6c | 206 | return len ? xmemdupz(pathspec->items[0].match, len) : NULL; |
f950eb95 CB |
207 | } |
208 | ||
0d32c183 BW |
209 | int fill_directory(struct dir_struct *dir, |
210 | struct index_state *istate, | |
211 | const struct pathspec *pathspec) | |
1d8842d9 | 212 | { |
bec5ab89 | 213 | const char *prefix; |
966de302 | 214 | size_t prefix_len; |
1d8842d9 | 215 | |
351ea1c3 EN |
216 | unsigned exclusive_flags = DIR_SHOW_IGNORED | DIR_SHOW_IGNORED_TOO; |
217 | if ((dir->flags & exclusive_flags) == exclusive_flags) | |
218 | BUG("DIR_SHOW_IGNORED and DIR_SHOW_IGNORED_TOO are exclusive"); | |
219 | ||
1d8842d9 LT |
220 | /* |
221 | * Calculate common prefix for the pathspec, and | |
222 | * use that to optimize the directory walk | |
223 | */ | |
bec5ab89 RS |
224 | prefix_len = common_prefix_len(pathspec); |
225 | prefix = prefix_len ? pathspec->items[0].match : ""; | |
1d8842d9 LT |
226 | |
227 | /* Read the directory and prune it */ | |
0d32c183 | 228 | read_directory(dir, istate, prefix, prefix_len, pathspec); |
966de302 | 229 | |
966de302 | 230 | return prefix_len; |
1d8842d9 LT |
231 | } |
232 | ||
bc96cc87 NTND |
233 | int within_depth(const char *name, int namelen, |
234 | int depth, int max_depth) | |
235 | { | |
236 | const char *cp = name, *cpe = name + namelen; | |
237 | ||
238 | while (cp < cpe) { | |
239 | if (*cp++ != '/') | |
240 | continue; | |
241 | depth++; | |
242 | if (depth > max_depth) | |
243 | return 0; | |
244 | } | |
245 | return 1; | |
246 | } | |
247 | ||
578d81d0 JH |
248 | /* |
249 | * Read the contents of the blob with the given OID into a buffer. | |
250 | * Append a trailing LF to the end if the last line doesn't have one. | |
251 | * | |
252 | * Returns: | |
253 | * -1 when the OID is invalid or unknown or does not refer to a blob. | |
254 | * 0 when the blob is empty. | |
255 | * 1 along with { data, size } of the (possibly augmented) buffer | |
256 | * when successful. | |
257 | * | |
4b33e602 | 258 | * Optionally updates the given oid_stat with the given OID (when valid). |
578d81d0 | 259 | */ |
4b33e602 PO |
260 | static int do_read_blob(const struct object_id *oid, struct oid_stat *oid_stat, |
261 | size_t *size_out, char **data_out) | |
578d81d0 JH |
262 | { |
263 | enum object_type type; | |
264 | unsigned long sz; | |
265 | char *data; | |
266 | ||
267 | *size_out = 0; | |
268 | *data_out = NULL; | |
269 | ||
b4f5aca4 | 270 | data = read_object_file(oid, &type, &sz); |
578d81d0 JH |
271 | if (!data || type != OBJ_BLOB) { |
272 | free(data); | |
273 | return -1; | |
274 | } | |
275 | ||
4b33e602 PO |
276 | if (oid_stat) { |
277 | memset(&oid_stat->stat, 0, sizeof(oid_stat->stat)); | |
278 | oidcpy(&oid_stat->oid, oid); | |
578d81d0 JH |
279 | } |
280 | ||
281 | if (sz == 0) { | |
282 | free(data); | |
283 | return 0; | |
284 | } | |
285 | ||
286 | if (data[sz - 1] != '\n') { | |
287 | data = xrealloc(data, st_add(sz, 1)); | |
288 | data[sz++] = '\n'; | |
289 | } | |
290 | ||
291 | *size_out = xsize_t(sz); | |
292 | *data_out = data; | |
293 | ||
294 | return 1; | |
295 | } | |
296 | ||
75a6315f BW |
297 | #define DO_MATCH_EXCLUDE (1<<0) |
298 | #define DO_MATCH_DIRECTORY (1<<1) | |
a3d89d8f | 299 | #define DO_MATCH_LEADING_PATHSPEC (1<<2) |
42b0874a | 300 | |
61cf2820 | 301 | /* |
29b577b9 | 302 | * Does the given pathspec match the given name? A match is found if |
61cf2820 | 303 | * |
29b577b9 EN |
304 | * (1) the pathspec string is leading directory of 'name' ("RECURSIVELY"), or |
305 | * (2) the pathspec string has a leading part matching 'name' ("LEADING"), or | |
306 | * (3) the pathspec string is a wildcard and matches 'name' ("WILDCARD"), or | |
307 | * (4) the pathspec string is exactly the same as 'name' ("EXACT"). | |
61cf2820 | 308 | * |
29b577b9 | 309 | * Return value tells which case it was (1-4), or 0 when there is no match. |
61cf2820 | 310 | * |
29b577b9 EN |
311 | * It may be instructive to look at a small table of concrete examples |
312 | * to understand the differences between 1, 2, and 4: | |
313 | * | |
314 | * Pathspecs | |
315 | * | a/b | a/b/ | a/b/c | |
316 | * ------+-----------+-----------+------------ | |
317 | * a/b | EXACT | EXACT[1] | LEADING[2] | |
318 | * Names a/b/ | RECURSIVE | EXACT | LEADING[2] | |
319 | * a/b/c | RECURSIVE | RECURSIVE | EXACT | |
320 | * | |
321 | * [1] Only if DO_MATCH_DIRECTORY is passed; otherwise, this is NOT a match. | |
322 | * [2] Only if DO_MATCH_LEADING_PATHSPEC is passed; otherwise, not a match. | |
61cf2820 | 323 | */ |
847a9e5d | 324 | static int match_pathspec_item(struct index_state *istate, |
6d2df284 | 325 | const struct pathspec_item *item, int prefix, |
42b0874a | 326 | const char *name, int namelen, unsigned flags) |
61cf2820 NTND |
327 | { |
328 | /* name/namelen has prefix cut off by caller */ | |
329 | const char *match = item->match + prefix; | |
330 | int matchlen = item->len - prefix; | |
331 | ||
93d93537 NTND |
332 | /* |
333 | * The normal call pattern is: | |
334 | * 1. prefix = common_prefix_len(ps); | |
335 | * 2. prune something, or fill_directory | |
854b0959 | 336 | * 3. match_pathspec() |
93d93537 NTND |
337 | * |
338 | * 'prefix' at #1 may be shorter than the command's prefix and | |
339 | * it's ok for #2 to match extra files. Those extras will be | |
340 | * trimmed at #3. | |
341 | * | |
342 | * Suppose the pathspec is 'foo' and '../bar' running from | |
343 | * subdir 'xyz'. The common prefix at #1 will be empty, thanks | |
344 | * to "../". We may have xyz/foo _and_ XYZ/foo after #2. The | |
345 | * user does not want XYZ/foo, only the "foo" part should be | |
346 | * case-insensitive. We need to filter out XYZ/foo here. In | |
347 | * other words, we do not trust the caller on comparing the | |
348 | * prefix part when :(icase) is involved. We do exact | |
349 | * comparison ourselves. | |
350 | * | |
351 | * Normally the caller (common_prefix_len() in fact) does | |
352 | * _exact_ matching on name[-prefix+1..-1] and we do not need | |
353 | * to check that part. Be defensive and check it anyway, in | |
354 | * case common_prefix_len is changed, or a new caller is | |
355 | * introduced that does not use common_prefix_len. | |
356 | * | |
357 | * If the penalty turns out too high when prefix is really | |
358 | * long, maybe change it to | |
359 | * strncmp(match, name, item->prefix - prefix) | |
360 | */ | |
361 | if (item->prefix && (item->magic & PATHSPEC_ICASE) && | |
362 | strncmp(item->match, name - prefix, item->prefix)) | |
363 | return 0; | |
364 | ||
22af33be NTND |
365 | if (item->attr_match_nr && |
366 | !match_pathspec_attrs(istate, name, namelen, item)) | |
b0db7046 BW |
367 | return 0; |
368 | ||
61cf2820 NTND |
369 | /* If the match was just the prefix, we matched */ |
370 | if (!*match) | |
371 | return MATCHED_RECURSIVELY; | |
372 | ||
93d93537 | 373 | if (matchlen <= namelen && !ps_strncmp(item, match, name, matchlen)) { |
61cf2820 NTND |
374 | if (matchlen == namelen) |
375 | return MATCHED_EXACTLY; | |
376 | ||
377 | if (match[matchlen-1] == '/' || name[matchlen] == '/') | |
378 | return MATCHED_RECURSIVELY; | |
68690fdd NTND |
379 | } else if ((flags & DO_MATCH_DIRECTORY) && |
380 | match[matchlen - 1] == '/' && | |
381 | namelen == matchlen - 1 && | |
382 | !ps_strncmp(item, match, name, namelen)) | |
383 | return MATCHED_EXACTLY; | |
61cf2820 | 384 | |
5d74762d | 385 | if (item->nowildcard_len < item->len && |
bd30c2e4 | 386 | !git_fnmatch(item, match, name, |
8c6abbcd | 387 | item->nowildcard_len - prefix)) |
61cf2820 NTND |
388 | return MATCHED_FNMATCH; |
389 | ||
29b577b9 | 390 | /* Perform checks to see if "name" is a leading string of the pathspec */ |
f1f061e1 EN |
391 | if ( (flags & DO_MATCH_LEADING_PATHSPEC) && |
392 | !(flags & DO_MATCH_EXCLUDE)) { | |
75a6315f | 393 | /* name is a literal prefix of the pathspec */ |
a5e916c7 | 394 | int offset = name[namelen-1] == '/' ? 1 : 0; |
75a6315f | 395 | if ((namelen < matchlen) && |
a5e916c7 | 396 | (match[namelen-offset] == '/') && |
75a6315f | 397 | !ps_strncmp(item, match, name, namelen)) |
89a1f4aa | 398 | return MATCHED_RECURSIVELY_LEADING_PATHSPEC; |
75a6315f | 399 | |
2f5d3847 | 400 | /* name doesn't match up to the first wild character */ |
75a6315f BW |
401 | if (item->nowildcard_len < item->len && |
402 | ps_strncmp(item, match, name, | |
403 | item->nowildcard_len - prefix)) | |
404 | return 0; | |
405 | ||
072a2310 EN |
406 | /* |
407 | * name has no wildcard, and it didn't match as a leading | |
408 | * pathspec so return. | |
409 | */ | |
410 | if (item->nowildcard_len == item->len) | |
411 | return 0; | |
412 | ||
75a6315f BW |
413 | /* |
414 | * Here is where we would perform a wildmatch to check if | |
415 | * "name" can be matched as a directory (or a prefix) against | |
416 | * the pathspec. Since wildmatch doesn't have this capability | |
417 | * at the present we have to punt and say that it is a match, | |
418 | * potentially returning a false positive | |
419 | * The submodules themselves will be able to perform more | |
420 | * accurate matching to determine if the pathspec matches. | |
421 | */ | |
89a1f4aa | 422 | return MATCHED_RECURSIVELY_LEADING_PATHSPEC; |
75a6315f BW |
423 | } |
424 | ||
61cf2820 NTND |
425 | return 0; |
426 | } | |
427 | ||
428 | /* | |
f1f061e1 EN |
429 | * do_match_pathspec() is meant to ONLY be called by |
430 | * match_pathspec_with_flags(); calling it directly risks pathspecs | |
431 | * like ':!unwanted_path' being ignored. | |
432 | * | |
52ed1894 AS |
433 | * Given a name and a list of pathspecs, returns the nature of the |
434 | * closest (i.e. most specific) match of the name to any of the | |
435 | * pathspecs. | |
436 | * | |
437 | * The caller typically calls this multiple times with the same | |
438 | * pathspec and seen[] array but with different name/namelen | |
439 | * (e.g. entries from the index) and is interested in seeing if and | |
440 | * how each pathspec matches all the names it calls this function | |
441 | * with. A mark is left in the seen[] array for each pathspec element | |
442 | * indicating the closest type of match that element achieved, so if | |
443 | * seen[n] remains zero after multiple invocations, that means the nth | |
444 | * pathspec did not match any names, which could indicate that the | |
445 | * user mistyped the nth pathspec. | |
61cf2820 | 446 | */ |
847a9e5d | 447 | static int do_match_pathspec(struct index_state *istate, |
6d2df284 | 448 | const struct pathspec *ps, |
854b0959 NTND |
449 | const char *name, int namelen, |
450 | int prefix, char *seen, | |
42b0874a | 451 | unsigned flags) |
61cf2820 | 452 | { |
42b0874a | 453 | int i, retval = 0, exclude = flags & DO_MATCH_EXCLUDE; |
61cf2820 | 454 | |
5c6933d2 NTND |
455 | GUARD_PATHSPEC(ps, |
456 | PATHSPEC_FROMTOP | | |
457 | PATHSPEC_MAXDEPTH | | |
bd30c2e4 | 458 | PATHSPEC_LITERAL | |
93d93537 | 459 | PATHSPEC_GLOB | |
ef79b1f8 | 460 | PATHSPEC_ICASE | |
b0db7046 BW |
461 | PATHSPEC_EXCLUDE | |
462 | PATHSPEC_ATTR); | |
8f4f8f45 | 463 | |
61cf2820 | 464 | if (!ps->nr) { |
6330a171 NTND |
465 | if (!ps->recursive || |
466 | !(ps->magic & PATHSPEC_MAXDEPTH) || | |
467 | ps->max_depth == -1) | |
61cf2820 NTND |
468 | return MATCHED_RECURSIVELY; |
469 | ||
470 | if (within_depth(name, namelen, 0, ps->max_depth)) | |
471 | return MATCHED_EXACTLY; | |
472 | else | |
473 | return 0; | |
474 | } | |
475 | ||
476 | name += prefix; | |
477 | namelen -= prefix; | |
478 | ||
479 | for (i = ps->nr - 1; i >= 0; i--) { | |
480 | int how; | |
ef79b1f8 NTND |
481 | |
482 | if ((!exclude && ps->items[i].magic & PATHSPEC_EXCLUDE) || | |
483 | ( exclude && !(ps->items[i].magic & PATHSPEC_EXCLUDE))) | |
484 | continue; | |
485 | ||
61cf2820 NTND |
486 | if (seen && seen[i] == MATCHED_EXACTLY) |
487 | continue; | |
ef79b1f8 NTND |
488 | /* |
489 | * Make exclude patterns optional and never report | |
490 | * "pathspec ':(exclude)foo' matches no files" | |
491 | */ | |
492 | if (seen && ps->items[i].magic & PATHSPEC_EXCLUDE) | |
493 | seen[i] = MATCHED_FNMATCH; | |
6d2df284 | 494 | how = match_pathspec_item(istate, ps->items+i, prefix, name, |
42b0874a | 495 | namelen, flags); |
6330a171 NTND |
496 | if (ps->recursive && |
497 | (ps->magic & PATHSPEC_MAXDEPTH) && | |
498 | ps->max_depth != -1 && | |
61cf2820 NTND |
499 | how && how != MATCHED_FNMATCH) { |
500 | int len = ps->items[i].len; | |
501 | if (name[len] == '/') | |
502 | len++; | |
503 | if (within_depth(name+len, namelen-len, 0, ps->max_depth)) | |
504 | how = MATCHED_EXACTLY; | |
505 | else | |
506 | how = 0; | |
507 | } | |
508 | if (how) { | |
509 | if (retval < how) | |
510 | retval = how; | |
511 | if (seen && seen[i] < how) | |
512 | seen[i] = how; | |
513 | } | |
514 | } | |
515 | return retval; | |
516 | } | |
517 | ||
847a9e5d | 518 | static int match_pathspec_with_flags(struct index_state *istate, |
f1f061e1 EN |
519 | const struct pathspec *ps, |
520 | const char *name, int namelen, | |
521 | int prefix, char *seen, unsigned flags) | |
ef79b1f8 NTND |
522 | { |
523 | int positive, negative; | |
6d2df284 | 524 | positive = do_match_pathspec(istate, ps, name, namelen, |
42b0874a | 525 | prefix, seen, flags); |
ef79b1f8 NTND |
526 | if (!(ps->magic & PATHSPEC_EXCLUDE) || !positive) |
527 | return positive; | |
6d2df284 | 528 | negative = do_match_pathspec(istate, ps, name, namelen, |
42b0874a NTND |
529 | prefix, seen, |
530 | flags | DO_MATCH_EXCLUDE); | |
ef79b1f8 NTND |
531 | return negative ? 0 : positive; |
532 | } | |
533 | ||
847a9e5d | 534 | int match_pathspec(struct index_state *istate, |
f1f061e1 EN |
535 | const struct pathspec *ps, |
536 | const char *name, int namelen, | |
537 | int prefix, char *seen, int is_dir) | |
538 | { | |
539 | unsigned flags = is_dir ? DO_MATCH_DIRECTORY : 0; | |
540 | return match_pathspec_with_flags(istate, ps, name, namelen, | |
541 | prefix, seen, flags); | |
542 | } | |
543 | ||
75a6315f BW |
544 | /** |
545 | * Check if a submodule is a superset of the pathspec | |
546 | */ | |
847a9e5d | 547 | int submodule_path_match(struct index_state *istate, |
6d2df284 | 548 | const struct pathspec *ps, |
75a6315f BW |
549 | const char *submodule_name, |
550 | char *seen) | |
551 | { | |
f1f061e1 EN |
552 | int matched = match_pathspec_with_flags(istate, ps, submodule_name, |
553 | strlen(submodule_name), | |
554 | 0, seen, | |
555 | DO_MATCH_DIRECTORY | | |
556 | DO_MATCH_LEADING_PATHSPEC); | |
75a6315f BW |
557 | return matched; |
558 | } | |
559 | ||
777c55a6 | 560 | int report_path_error(const char *ps_matched, |
c5c33504 | 561 | const struct pathspec *pathspec) |
777c55a6 JH |
562 | { |
563 | /* | |
564 | * Make sure all pathspec matched; otherwise it is an error. | |
565 | */ | |
777c55a6 JH |
566 | int num, errors = 0; |
567 | for (num = 0; num < pathspec->nr; num++) { | |
568 | int other, found_dup; | |
569 | ||
570 | if (ps_matched[num]) | |
571 | continue; | |
572 | /* | |
573 | * The caller might have fed identical pathspec | |
574 | * twice. Do not barf on such a mistake. | |
575 | * FIXME: parse_pathspec should have eliminated | |
576 | * duplicate pathspec. | |
577 | */ | |
578 | for (found_dup = other = 0; | |
579 | !found_dup && other < pathspec->nr; | |
580 | other++) { | |
581 | if (other == num || !ps_matched[other]) | |
582 | continue; | |
583 | if (!strcmp(pathspec->items[other].original, | |
584 | pathspec->items[num].original)) | |
585 | /* | |
586 | * Ok, we have a match already. | |
587 | */ | |
588 | found_dup = 1; | |
589 | } | |
590 | if (found_dup) | |
591 | continue; | |
592 | ||
a80897c1 | 593 | error(_("pathspec '%s' did not match any file(s) known to git"), |
777c55a6 JH |
594 | pathspec->items[num].original); |
595 | errors++; | |
596 | } | |
777c55a6 JH |
597 | return errors; |
598 | } | |
599 | ||
fcd631ed NTND |
600 | /* |
601 | * Return the length of the "simple" part of a path match limiter. | |
602 | */ | |
87323bda | 603 | int simple_length(const char *match) |
fcd631ed NTND |
604 | { |
605 | int len = -1; | |
606 | ||
607 | for (;;) { | |
608 | unsigned char c = *match++; | |
609 | len++; | |
610 | if (c == '\0' || is_glob_special(c)) | |
611 | return len; | |
612 | } | |
613 | } | |
614 | ||
87323bda | 615 | int no_wildcard(const char *string) |
68492fc7 | 616 | { |
fcd631ed | 617 | return string[simple_length(string)] == '\0'; |
68492fc7 LK |
618 | } |
619 | ||
65edd96a | 620 | void parse_path_pattern(const char **pattern, |
82dce998 | 621 | int *patternlen, |
f8708998 | 622 | unsigned *flags, |
82dce998 | 623 | int *nowildcardlen) |
84460eec NTND |
624 | { |
625 | const char *p = *pattern; | |
626 | size_t i, len; | |
627 | ||
628 | *flags = 0; | |
629 | if (*p == '!') { | |
4ff89ee5 | 630 | *flags |= PATTERN_FLAG_NEGATIVE; |
84460eec NTND |
631 | p++; |
632 | } | |
633 | len = strlen(p); | |
634 | if (len && p[len - 1] == '/') { | |
635 | len--; | |
4ff89ee5 | 636 | *flags |= PATTERN_FLAG_MUSTBEDIR; |
84460eec NTND |
637 | } |
638 | for (i = 0; i < len; i++) { | |
639 | if (p[i] == '/') | |
640 | break; | |
641 | } | |
642 | if (i == len) | |
4ff89ee5 | 643 | *flags |= PATTERN_FLAG_NODIR; |
84460eec NTND |
644 | *nowildcardlen = simple_length(p); |
645 | /* | |
646 | * we should have excluded the trailing slash from 'p' too, | |
647 | * but that's one more allocation. Instead just make sure | |
648 | * nowildcardlen does not exceed real patternlen | |
649 | */ | |
650 | if (*nowildcardlen > len) | |
651 | *nowildcardlen = len; | |
652 | if (*p == '*' && no_wildcard(p + 1)) | |
4ff89ee5 | 653 | *flags |= PATTERN_FLAG_ENDSWITH; |
84460eec NTND |
654 | *pattern = p; |
655 | *patternlen = len; | |
656 | } | |
657 | ||
af09ce24 DS |
658 | int pl_hashmap_cmp(const void *unused_cmp_data, |
659 | const struct hashmap_entry *a, | |
660 | const struct hashmap_entry *b, | |
661 | const void *key) | |
96cc8ab5 DS |
662 | { |
663 | const struct pattern_entry *ee1 = | |
664 | container_of(a, struct pattern_entry, ent); | |
665 | const struct pattern_entry *ee2 = | |
666 | container_of(b, struct pattern_entry, ent); | |
667 | ||
668 | size_t min_len = ee1->patternlen <= ee2->patternlen | |
669 | ? ee1->patternlen | |
670 | : ee2->patternlen; | |
671 | ||
190a65f9 DS |
672 | if (ignore_case) |
673 | return strncasecmp(ee1->pattern, ee2->pattern, min_len); | |
96cc8ab5 DS |
674 | return strncmp(ee1->pattern, ee2->pattern, min_len); |
675 | } | |
676 | ||
4f52c2ce DS |
677 | static char *dup_and_filter_pattern(const char *pattern) |
678 | { | |
679 | char *set, *read; | |
680 | size_t count = 0; | |
681 | char *result = xstrdup(pattern); | |
682 | ||
683 | set = result; | |
684 | read = result; | |
685 | ||
686 | while (*read) { | |
687 | /* skip escape characters (once) */ | |
688 | if (*read == '\\') | |
689 | read++; | |
690 | ||
691 | *set = *read; | |
692 | ||
693 | set++; | |
694 | read++; | |
695 | count++; | |
696 | } | |
697 | *set = 0; | |
698 | ||
699 | if (count > 2 && | |
700 | *(set - 1) == '*' && | |
701 | *(set - 2) == '/') | |
702 | *(set - 2) = 0; | |
703 | ||
704 | return result; | |
705 | } | |
706 | ||
96cc8ab5 DS |
707 | static void add_pattern_to_hashsets(struct pattern_list *pl, struct path_pattern *given) |
708 | { | |
709 | struct pattern_entry *translated; | |
710 | char *truncated; | |
711 | char *data = NULL; | |
9abc60f8 | 712 | const char *prev, *cur, *next; |
96cc8ab5 DS |
713 | |
714 | if (!pl->use_cone_patterns) | |
715 | return; | |
716 | ||
717 | if (given->flags & PATTERN_FLAG_NEGATIVE && | |
718 | given->flags & PATTERN_FLAG_MUSTBEDIR && | |
719 | !strcmp(given->pattern, "/*")) { | |
720 | pl->full_cone = 0; | |
721 | return; | |
722 | } | |
723 | ||
724 | if (!given->flags && !strcmp(given->pattern, "/*")) { | |
725 | pl->full_cone = 1; | |
726 | return; | |
727 | } | |
728 | ||
6c11c6a1 | 729 | if (given->patternlen < 2 || |
9abc60f8 | 730 | *given->pattern == '*' || |
9e6d3e64 | 731 | strstr(given->pattern, "**")) { |
41de0c6f DS |
732 | /* Not a cone pattern. */ |
733 | warning(_("unrecognized pattern: '%s'"), given->pattern); | |
734 | goto clear_hashmaps; | |
735 | } | |
736 | ||
9abc60f8 DS |
737 | prev = given->pattern; |
738 | cur = given->pattern + 1; | |
739 | next = given->pattern + 2; | |
740 | ||
741 | while (*cur) { | |
742 | /* Watch for glob characters '*', '\', '[', '?' */ | |
743 | if (!is_glob_special(*cur)) | |
744 | goto increment; | |
745 | ||
746 | /* But only if *prev != '\\' */ | |
747 | if (*prev == '\\') | |
748 | goto increment; | |
749 | ||
750 | /* But allow the initial '\' */ | |
751 | if (*cur == '\\' && | |
752 | is_glob_special(*next)) | |
753 | goto increment; | |
754 | ||
755 | /* But a trailing '/' then '*' is fine */ | |
756 | if (*prev == '/' && | |
757 | *cur == '*' && | |
758 | *next == 0) | |
759 | goto increment; | |
760 | ||
761 | /* Not a cone pattern. */ | |
762 | warning(_("unrecognized pattern: '%s'"), given->pattern); | |
763 | goto clear_hashmaps; | |
764 | ||
765 | increment: | |
766 | prev++; | |
767 | cur++; | |
768 | next++; | |
769 | } | |
770 | ||
96cc8ab5 DS |
771 | if (given->patternlen > 2 && |
772 | !strcmp(given->pattern + given->patternlen - 2, "/*")) { | |
773 | if (!(given->flags & PATTERN_FLAG_NEGATIVE)) { | |
774 | /* Not a cone pattern. */ | |
96cc8ab5 DS |
775 | warning(_("unrecognized pattern: '%s'"), given->pattern); |
776 | goto clear_hashmaps; | |
777 | } | |
778 | ||
4f52c2ce | 779 | truncated = dup_and_filter_pattern(given->pattern); |
96cc8ab5 DS |
780 | |
781 | translated = xmalloc(sizeof(struct pattern_entry)); | |
782 | translated->pattern = truncated; | |
783 | translated->patternlen = given->patternlen - 2; | |
784 | hashmap_entry_init(&translated->ent, | |
74318423 | 785 | fspathhash(translated->pattern)); |
96cc8ab5 DS |
786 | |
787 | if (!hashmap_get_entry(&pl->recursive_hashmap, | |
788 | translated, ent, NULL)) { | |
789 | /* We did not see the "parent" included */ | |
790 | warning(_("unrecognized negative pattern: '%s'"), | |
791 | given->pattern); | |
792 | free(truncated); | |
793 | free(translated); | |
794 | goto clear_hashmaps; | |
795 | } | |
796 | ||
797 | hashmap_add(&pl->parent_hashmap, &translated->ent); | |
798 | hashmap_remove(&pl->recursive_hashmap, &translated->ent, &data); | |
799 | free(data); | |
800 | return; | |
801 | } | |
802 | ||
803 | if (given->flags & PATTERN_FLAG_NEGATIVE) { | |
804 | warning(_("unrecognized negative pattern: '%s'"), | |
805 | given->pattern); | |
806 | goto clear_hashmaps; | |
807 | } | |
808 | ||
809 | translated = xmalloc(sizeof(struct pattern_entry)); | |
810 | ||
4f52c2ce | 811 | translated->pattern = dup_and_filter_pattern(given->pattern); |
96cc8ab5 DS |
812 | translated->patternlen = given->patternlen; |
813 | hashmap_entry_init(&translated->ent, | |
74318423 | 814 | fspathhash(translated->pattern)); |
96cc8ab5 DS |
815 | |
816 | hashmap_add(&pl->recursive_hashmap, &translated->ent); | |
817 | ||
818 | if (hashmap_get_entry(&pl->parent_hashmap, translated, ent, NULL)) { | |
819 | /* we already included this at the parent level */ | |
820 | warning(_("your sparse-checkout file may have issues: pattern '%s' is repeated"), | |
821 | given->pattern); | |
822 | hashmap_remove(&pl->parent_hashmap, &translated->ent, &data); | |
823 | free(data); | |
824 | free(translated); | |
825 | } | |
826 | ||
827 | return; | |
828 | ||
829 | clear_hashmaps: | |
830 | warning(_("disabling cone pattern matching")); | |
6da1a258 EN |
831 | hashmap_clear_and_free(&pl->parent_hashmap, struct pattern_entry, ent); |
832 | hashmap_clear_and_free(&pl->recursive_hashmap, struct pattern_entry, ent); | |
96cc8ab5 DS |
833 | pl->use_cone_patterns = 0; |
834 | } | |
835 | ||
836 | static int hashmap_contains_path(struct hashmap *map, | |
837 | struct strbuf *pattern) | |
838 | { | |
839 | struct pattern_entry p; | |
840 | ||
841 | /* Check straight mapping */ | |
842 | p.pattern = pattern->buf; | |
843 | p.patternlen = pattern->len; | |
74318423 | 844 | hashmap_entry_init(&p.ent, fspathhash(p.pattern)); |
96cc8ab5 DS |
845 | return !!hashmap_get_entry(map, &p, ent, NULL); |
846 | } | |
847 | ||
848 | int hashmap_contains_parent(struct hashmap *map, | |
849 | const char *path, | |
850 | struct strbuf *buffer) | |
851 | { | |
852 | char *slash_pos; | |
853 | ||
854 | strbuf_setlen(buffer, 0); | |
855 | ||
856 | if (path[0] != '/') | |
857 | strbuf_addch(buffer, '/'); | |
858 | ||
859 | strbuf_addstr(buffer, path); | |
860 | ||
861 | slash_pos = strrchr(buffer->buf, '/'); | |
862 | ||
863 | while (slash_pos > buffer->buf) { | |
864 | strbuf_setlen(buffer, slash_pos - buffer->buf); | |
865 | ||
866 | if (hashmap_contains_path(map, buffer)) | |
867 | return 1; | |
868 | ||
869 | slash_pos = strrchr(buffer->buf, '/'); | |
870 | } | |
871 | ||
872 | return 0; | |
873 | } | |
874 | ||
65edd96a | 875 | void add_pattern(const char *string, const char *base, |
caa3d554 | 876 | int baselen, struct pattern_list *pl, int srcpos) |
453ec4bd | 877 | { |
ab8db613 | 878 | struct path_pattern *pattern; |
84460eec | 879 | int patternlen; |
f8708998 | 880 | unsigned flags; |
84460eec | 881 | int nowildcardlen; |
453ec4bd | 882 | |
65edd96a | 883 | parse_path_pattern(&string, &patternlen, &flags, &nowildcardlen); |
4ff89ee5 | 884 | if (flags & PATTERN_FLAG_MUSTBEDIR) { |
ab8db613 | 885 | FLEXPTR_ALLOC_MEM(pattern, pattern, string, patternlen); |
d6b8fc30 | 886 | } else { |
ab8db613 DS |
887 | pattern = xmalloc(sizeof(*pattern)); |
888 | pattern->pattern = string; | |
d6b8fc30 | 889 | } |
ab8db613 DS |
890 | pattern->patternlen = patternlen; |
891 | pattern->nowildcardlen = nowildcardlen; | |
892 | pattern->base = base; | |
893 | pattern->baselen = baselen; | |
894 | pattern->flags = flags; | |
895 | pattern->srcpos = srcpos; | |
caa3d554 DS |
896 | ALLOC_GROW(pl->patterns, pl->nr + 1, pl->alloc); |
897 | pl->patterns[pl->nr++] = pattern; | |
898 | pattern->pl = pl; | |
96cc8ab5 DS |
899 | |
900 | add_pattern_to_hashsets(pl, pattern); | |
453ec4bd LT |
901 | } |
902 | ||
847a9e5d | 903 | static int read_skip_worktree_file_from_index(struct index_state *istate, |
578d81d0 | 904 | const char *path, |
4b33e602 PO |
905 | size_t *size_out, char **data_out, |
906 | struct oid_stat *oid_stat) | |
c28b3d6e NTND |
907 | { |
908 | int pos, len; | |
c28b3d6e NTND |
909 | |
910 | len = strlen(path); | |
6f52b741 | 911 | pos = index_name_pos(istate, path, len); |
c28b3d6e | 912 | if (pos < 0) |
578d81d0 | 913 | return -1; |
6f52b741 | 914 | if (!ce_skip_worktree(istate->cache[pos])) |
578d81d0 JH |
915 | return -1; |
916 | ||
4b33e602 | 917 | return do_read_blob(&istate->cache[pos]->oid, oid_stat, size_out, data_out); |
c28b3d6e NTND |
918 | } |
919 | ||
f6198812 | 920 | /* |
caa3d554 DS |
921 | * Frees memory within pl which was allocated for exclude patterns and |
922 | * the file buffer. Does not free pl itself. | |
f6198812 | 923 | */ |
65edd96a | 924 | void clear_pattern_list(struct pattern_list *pl) |
0fd0e241 NTND |
925 | { |
926 | int i; | |
927 | ||
caa3d554 DS |
928 | for (i = 0; i < pl->nr; i++) |
929 | free(pl->patterns[i]); | |
930 | free(pl->patterns); | |
931 | free(pl->filebuf); | |
6da1a258 EN |
932 | hashmap_clear_and_free(&pl->recursive_hashmap, struct pattern_entry, ent); |
933 | hashmap_clear_and_free(&pl->parent_hashmap, struct pattern_entry, ent); | |
0fd0e241 | 934 | |
caa3d554 | 935 | memset(pl, 0, sizeof(*pl)); |
0fd0e241 NTND |
936 | } |
937 | ||
7e2e4b37 | 938 | static void trim_trailing_spaces(char *buf) |
16402b99 | 939 | { |
e61a6c1d PB |
940 | char *p, *last_space = NULL; |
941 | ||
942 | for (p = buf; *p; p++) | |
943 | switch (*p) { | |
944 | case ' ': | |
945 | if (!last_space) | |
946 | last_space = p; | |
947 | break; | |
948 | case '\\': | |
949 | p++; | |
950 | if (!*p) | |
951 | return; | |
952 | /* fallthrough */ | |
953 | default: | |
954 | last_space = NULL; | |
955 | } | |
956 | ||
957 | if (last_space) | |
958 | *last_space = '\0'; | |
16402b99 NTND |
959 | } |
960 | ||
0dcb8d7f NTND |
961 | /* |
962 | * Given a subdirectory name and "dir" of the current directory, | |
963 | * search the subdir in "dir" and return it, or create a new one if it | |
964 | * does not exist in "dir". | |
965 | * | |
966 | * If "name" has the trailing slash, it'll be excluded in the search. | |
967 | */ | |
968 | static struct untracked_cache_dir *lookup_untracked(struct untracked_cache *uc, | |
969 | struct untracked_cache_dir *dir, | |
970 | const char *name, int len) | |
971 | { | |
972 | int first, last; | |
973 | struct untracked_cache_dir *d; | |
974 | if (!dir) | |
975 | return NULL; | |
976 | if (len && name[len - 1] == '/') | |
977 | len--; | |
978 | first = 0; | |
979 | last = dir->dirs_nr; | |
980 | while (last > first) { | |
568a05c5 | 981 | int cmp, next = first + ((last - first) >> 1); |
0dcb8d7f NTND |
982 | d = dir->dirs[next]; |
983 | cmp = strncmp(name, d->name, len); | |
984 | if (!cmp && strlen(d->name) > len) | |
985 | cmp = -1; | |
986 | if (!cmp) | |
987 | return d; | |
988 | if (cmp < 0) { | |
989 | last = next; | |
990 | continue; | |
991 | } | |
992 | first = next+1; | |
993 | } | |
994 | ||
995 | uc->dir_created++; | |
96ffc06f | 996 | FLEX_ALLOC_MEM(d, name, name, len); |
0dcb8d7f NTND |
997 | |
998 | ALLOC_GROW(dir->dirs, dir->dirs_nr + 1, dir->dirs_alloc); | |
f919ffeb SG |
999 | MOVE_ARRAY(dir->dirs + first + 1, dir->dirs + first, |
1000 | dir->dirs_nr - first); | |
0dcb8d7f NTND |
1001 | dir->dirs_nr++; |
1002 | dir->dirs[first] = d; | |
1003 | return d; | |
1004 | } | |
1005 | ||
ccad261f NTND |
1006 | static void do_invalidate_gitignore(struct untracked_cache_dir *dir) |
1007 | { | |
1008 | int i; | |
1009 | dir->valid = 0; | |
1010 | dir->untracked_nr = 0; | |
1011 | for (i = 0; i < dir->dirs_nr; i++) | |
1012 | do_invalidate_gitignore(dir->dirs[i]); | |
1013 | } | |
1014 | ||
1015 | static void invalidate_gitignore(struct untracked_cache *uc, | |
1016 | struct untracked_cache_dir *dir) | |
1017 | { | |
1018 | uc->gitignore_invalidated++; | |
1019 | do_invalidate_gitignore(dir); | |
1020 | } | |
1021 | ||
91a2288b NTND |
1022 | static void invalidate_directory(struct untracked_cache *uc, |
1023 | struct untracked_cache_dir *dir) | |
1024 | { | |
26cb0182 | 1025 | int i; |
b6403131 NTND |
1026 | |
1027 | /* | |
1028 | * Invalidation increment here is just roughly correct. If | |
1029 | * untracked_nr or any of dirs[].recurse is non-zero, we | |
1030 | * should increment dir_invalidated too. But that's more | |
1031 | * expensive to do. | |
1032 | */ | |
1033 | if (dir->valid) | |
1034 | uc->dir_invalidated++; | |
1035 | ||
91a2288b NTND |
1036 | dir->valid = 0; |
1037 | dir->untracked_nr = 0; | |
26cb0182 NTND |
1038 | for (i = 0; i < dir->dirs_nr; i++) |
1039 | dir->dirs[i]->recurse = 0; | |
91a2288b NTND |
1040 | } |
1041 | ||
65edd96a | 1042 | static int add_patterns_from_buffer(char *buf, size_t size, |
578d81d0 | 1043 | const char *base, int baselen, |
caa3d554 | 1044 | struct pattern_list *pl); |
578d81d0 | 1045 | |
feb9b779 JK |
1046 | /* Flags for add_patterns() */ |
1047 | #define PATTERN_NOFOLLOW (1<<0) | |
1048 | ||
55fe6f51 NTND |
1049 | /* |
1050 | * Given a file with name "fname", read it (either from disk, or from | |
473e3930 | 1051 | * an index if 'istate' is non-null), parse it and store the |
caa3d554 | 1052 | * exclude rules in "pl". |
55fe6f51 | 1053 | * |
e5cf6d3d | 1054 | * If "oid_stat" is not NULL, compute oid of the exclude file and fill |
65edd96a | 1055 | * stat data from disk (only valid if add_patterns returns zero). If |
e5cf6d3d | 1056 | * oid_stat.valid is non-zero, "oid_stat" must contain good value as input. |
55fe6f51 | 1057 | */ |
65edd96a | 1058 | static int add_patterns(const char *fname, const char *base, int baselen, |
caa3d554 | 1059 | struct pattern_list *pl, struct index_state *istate, |
1679d60b | 1060 | unsigned flags, struct oid_stat *oid_stat) |
453ec4bd | 1061 | { |
c470701a | 1062 | struct stat st; |
578d81d0 JH |
1063 | int r; |
1064 | int fd; | |
9d14017a | 1065 | size_t size = 0; |
578d81d0 | 1066 | char *buf; |
453ec4bd | 1067 | |
feb9b779 JK |
1068 | if (flags & PATTERN_NOFOLLOW) |
1069 | fd = open_nofollow(fname, O_RDONLY); | |
1070 | else | |
1071 | fd = open(fname, O_RDONLY); | |
1072 | ||
c28b3d6e | 1073 | if (fd < 0 || fstat(fd, &st) < 0) { |
11dc1fcb NTND |
1074 | if (fd < 0) |
1075 | warn_on_fopen_errors(fname); | |
1076 | else | |
c28b3d6e | 1077 | close(fd); |
578d81d0 | 1078 | if (!istate) |
c28b3d6e | 1079 | return -1; |
578d81d0 JH |
1080 | r = read_skip_worktree_file_from_index(istate, fname, |
1081 | &size, &buf, | |
4b33e602 | 1082 | oid_stat); |
578d81d0 JH |
1083 | if (r != 1) |
1084 | return r; | |
d961baa8 | 1085 | } else { |
c28b3d6e NTND |
1086 | size = xsize_t(st.st_size); |
1087 | if (size == 0) { | |
4b33e602 PO |
1088 | if (oid_stat) { |
1089 | fill_stat_data(&oid_stat->stat, &st); | |
ba2df751 | 1090 | oidcpy(&oid_stat->oid, the_hash_algo->empty_blob); |
4b33e602 | 1091 | oid_stat->valid = 1; |
55fe6f51 | 1092 | } |
c28b3d6e NTND |
1093 | close(fd); |
1094 | return 0; | |
1095 | } | |
3733e694 | 1096 | buf = xmallocz(size); |
c28b3d6e | 1097 | if (read_in_full(fd, buf, size) != size) { |
45d76f17 | 1098 | free(buf); |
c28b3d6e NTND |
1099 | close(fd); |
1100 | return -1; | |
1101 | } | |
45d76f17 | 1102 | buf[size++] = '\n'; |
c28b3d6e | 1103 | close(fd); |
4b33e602 | 1104 | if (oid_stat) { |
55fe6f51 | 1105 | int pos; |
4b33e602 PO |
1106 | if (oid_stat->valid && |
1107 | !match_stat_data_racy(istate, &oid_stat->stat, &st)) | |
e5cf6d3d | 1108 | ; /* no content change, oid_stat->oid still good */ |
473e3930 BW |
1109 | else if (istate && |
1110 | (pos = index_name_pos(istate, fname, strlen(fname))) >= 0 && | |
1111 | !ce_stage(istate->cache[pos]) && | |
1112 | ce_uptodate(istate->cache[pos]) && | |
82b474e0 | 1113 | !would_convert_to_git(istate, fname)) |
4b33e602 PO |
1114 | oidcpy(&oid_stat->oid, |
1115 | &istate->cache[pos]->oid); | |
55fe6f51 | 1116 | else |
2dcde20e MT |
1117 | hash_object_file(the_hash_algo, buf, size, |
1118 | "blob", &oid_stat->oid); | |
4b33e602 PO |
1119 | fill_stat_data(&oid_stat->stat, &st); |
1120 | oid_stat->valid = 1; | |
55fe6f51 | 1121 | } |
6ba78238 | 1122 | } |
453ec4bd | 1123 | |
65edd96a | 1124 | add_patterns_from_buffer(buf, size, base, baselen, pl); |
578d81d0 JH |
1125 | return 0; |
1126 | } | |
1127 | ||
65edd96a | 1128 | static int add_patterns_from_buffer(char *buf, size_t size, |
578d81d0 | 1129 | const char *base, int baselen, |
caa3d554 | 1130 | struct pattern_list *pl) |
578d81d0 JH |
1131 | { |
1132 | int i, lineno = 1; | |
1133 | char *entry; | |
1134 | ||
96cc8ab5 DS |
1135 | hashmap_init(&pl->recursive_hashmap, pl_hashmap_cmp, NULL, 0); |
1136 | hashmap_init(&pl->parent_hashmap, pl_hashmap_cmp, NULL, 0); | |
1137 | ||
caa3d554 | 1138 | pl->filebuf = buf; |
245e1c19 | 1139 | |
dde843e7 | 1140 | if (skip_utf8_bom(&buf, size)) |
caa3d554 | 1141 | size -= buf - pl->filebuf; |
dde843e7 | 1142 | |
453ec4bd | 1143 | entry = buf; |
245e1c19 | 1144 | |
45d76f17 NTND |
1145 | for (i = 0; i < size; i++) { |
1146 | if (buf[i] == '\n') { | |
453ec4bd LT |
1147 | if (entry != buf + i && entry[0] != '#') { |
1148 | buf[i - (i && buf[i-1] == '\r')] = 0; | |
7e2e4b37 | 1149 | trim_trailing_spaces(entry); |
65edd96a | 1150 | add_pattern(entry, base, baselen, pl, lineno); |
453ec4bd | 1151 | } |
c04318e4 | 1152 | lineno++; |
453ec4bd LT |
1153 | entry = buf + i + 1; |
1154 | } | |
1155 | } | |
1156 | return 0; | |
453ec4bd LT |
1157 | } |
1158 | ||
65edd96a | 1159 | int add_patterns_from_file_to_list(const char *fname, const char *base, |
caa3d554 | 1160 | int baselen, struct pattern_list *pl, |
1679d60b JK |
1161 | struct index_state *istate, |
1162 | unsigned flags) | |
55fe6f51 | 1163 | { |
1679d60b | 1164 | return add_patterns(fname, base, baselen, pl, istate, flags, NULL); |
55fe6f51 NTND |
1165 | } |
1166 | ||
65edd96a | 1167 | int add_patterns_from_blob_to_list( |
578d81d0 JH |
1168 | struct object_id *oid, |
1169 | const char *base, int baselen, | |
caa3d554 | 1170 | struct pattern_list *pl) |
578d81d0 JH |
1171 | { |
1172 | char *buf; | |
1173 | size_t size; | |
1174 | int r; | |
1175 | ||
1176 | r = do_read_blob(oid, NULL, &size, &buf); | |
1177 | if (r != 1) | |
1178 | return r; | |
1179 | ||
65edd96a | 1180 | add_patterns_from_buffer(buf, size, base, baselen, pl); |
578d81d0 JH |
1181 | return 0; |
1182 | } | |
1183 | ||
65edd96a | 1184 | struct pattern_list *add_pattern_list(struct dir_struct *dir, |
c04318e4 | 1185 | int group_type, const char *src) |
c082df24 | 1186 | { |
caa3d554 | 1187 | struct pattern_list *pl; |
c082df24 AS |
1188 | struct exclude_list_group *group; |
1189 | ||
1190 | group = &dir->exclude_list_group[group_type]; | |
caa3d554 DS |
1191 | ALLOC_GROW(group->pl, group->nr + 1, group->alloc); |
1192 | pl = &group->pl[group->nr++]; | |
1193 | memset(pl, 0, sizeof(*pl)); | |
1194 | pl->src = src; | |
1195 | return pl; | |
c082df24 AS |
1196 | } |
1197 | ||
1198 | /* | |
1199 | * Used to set up core.excludesfile and .git/info/exclude lists. | |
1200 | */ | |
65edd96a | 1201 | static void add_patterns_from_file_1(struct dir_struct *dir, const char *fname, |
4b33e602 | 1202 | struct oid_stat *oid_stat) |
453ec4bd | 1203 | { |
caa3d554 | 1204 | struct pattern_list *pl; |
ccad261f NTND |
1205 | /* |
1206 | * catch setup_standard_excludes() that's called before | |
1207 | * dir->untracked is assigned. That function behaves | |
1208 | * differently when dir->untracked is non-NULL. | |
1209 | */ | |
1210 | if (!dir->untracked) | |
1211 | dir->unmanaged_exclude_files++; | |
65edd96a | 1212 | pl = add_pattern_list(dir, EXC_FILE, fname); |
1679d60b | 1213 | if (add_patterns(fname, "", 0, pl, NULL, 0, oid_stat) < 0) |
a80897c1 | 1214 | die(_("cannot use %s as an exclude file"), fname); |
453ec4bd LT |
1215 | } |
1216 | ||
65edd96a | 1217 | void add_patterns_from_file(struct dir_struct *dir, const char *fname) |
0dcb8d7f | 1218 | { |
ccad261f | 1219 | dir->unmanaged_exclude_files++; /* see validate_untracked_cache() */ |
65edd96a | 1220 | add_patterns_from_file_1(dir, fname, NULL); |
0dcb8d7f NTND |
1221 | } |
1222 | ||
82dce998 NTND |
1223 | int match_basename(const char *basename, int basenamelen, |
1224 | const char *pattern, int prefix, int patternlen, | |
f8708998 | 1225 | unsigned flags) |
593cb880 NTND |
1226 | { |
1227 | if (prefix == patternlen) { | |
0b6e56df | 1228 | if (patternlen == basenamelen && |
ba0897e6 | 1229 | !fspathncmp(pattern, basename, basenamelen)) |
593cb880 | 1230 | return 1; |
4ff89ee5 | 1231 | } else if (flags & PATTERN_FLAG_ENDSWITH) { |
0b6e56df | 1232 | /* "*literal" matching against "fooliteral" */ |
593cb880 | 1233 | if (patternlen - 1 <= basenamelen && |
ba0897e6 | 1234 | !fspathncmp(pattern + 1, |
0b6e56df JH |
1235 | basename + basenamelen - (patternlen - 1), |
1236 | patternlen - 1)) | |
593cb880 NTND |
1237 | return 1; |
1238 | } else { | |
0b6e56df JH |
1239 | if (fnmatch_icase_mem(pattern, patternlen, |
1240 | basename, basenamelen, | |
1241 | 0) == 0) | |
593cb880 NTND |
1242 | return 1; |
1243 | } | |
1244 | return 0; | |
1245 | } | |
1246 | ||
82dce998 NTND |
1247 | int match_pathname(const char *pathname, int pathlen, |
1248 | const char *base, int baselen, | |
1249 | const char *pattern, int prefix, int patternlen, | |
f8708998 | 1250 | unsigned flags) |
b5592632 NTND |
1251 | { |
1252 | const char *name; | |
1253 | int namelen; | |
1254 | ||
1255 | /* | |
1256 | * match with FNM_PATHNAME; the pattern has base implicitly | |
1257 | * in front of it. | |
1258 | */ | |
1259 | if (*pattern == '/') { | |
1260 | pattern++; | |
982ac873 | 1261 | patternlen--; |
b5592632 NTND |
1262 | prefix--; |
1263 | } | |
1264 | ||
1265 | /* | |
1266 | * baselen does not count the trailing slash. base[] may or | |
1267 | * may not end with a trailing slash though. | |
1268 | */ | |
1269 | if (pathlen < baselen + 1 || | |
1270 | (baselen && pathname[baselen] != '/') || | |
ba0897e6 | 1271 | fspathncmp(pathname, base, baselen)) |
b5592632 NTND |
1272 | return 0; |
1273 | ||
1274 | namelen = baselen ? pathlen - baselen - 1 : pathlen; | |
1275 | name = pathname + pathlen - namelen; | |
1276 | ||
1277 | if (prefix) { | |
1278 | /* | |
1279 | * if the non-wildcard part is longer than the | |
1280 | * remaining pathname, surely it cannot match. | |
1281 | */ | |
1282 | if (prefix > namelen) | |
1283 | return 0; | |
1284 | ||
ba0897e6 | 1285 | if (fspathncmp(pattern, name, prefix)) |
b5592632 NTND |
1286 | return 0; |
1287 | pattern += prefix; | |
ab3aebc1 | 1288 | patternlen -= prefix; |
b5592632 NTND |
1289 | name += prefix; |
1290 | namelen -= prefix; | |
ab3aebc1 JK |
1291 | |
1292 | /* | |
1293 | * If the whole pattern did not have a wildcard, | |
1294 | * then our prefix match is all we need; we | |
1295 | * do not need to call fnmatch at all. | |
1296 | */ | |
ed495847 | 1297 | if (!patternlen && (!namelen || (flags & PATTERN_FLAG_MUSTBEDIR))) |
ab3aebc1 | 1298 | return 1; |
b5592632 NTND |
1299 | } |
1300 | ||
ab3aebc1 JK |
1301 | return fnmatch_icase_mem(pattern, patternlen, |
1302 | name, namelen, | |
f30366b2 | 1303 | WM_PATHNAME) == 0; |
b5592632 NTND |
1304 | } |
1305 | ||
f6526728 DS |
1306 | static int path_matches_dir_pattern(const char *pathname, |
1307 | int pathlen, | |
1308 | struct strbuf **path_parent, | |
1309 | int *dtype, | |
1310 | struct path_pattern *pattern, | |
1311 | struct index_state *istate) | |
1312 | { | |
1313 | if (!*path_parent) { | |
1314 | char *slash; | |
1315 | CALLOC_ARRAY(*path_parent, 1); | |
1316 | strbuf_add(*path_parent, pathname, pathlen); | |
1317 | slash = find_last_dir_sep((*path_parent)->buf); | |
1318 | ||
1319 | if (slash) | |
1320 | strbuf_setlen(*path_parent, slash - (*path_parent)->buf); | |
1321 | else | |
1322 | strbuf_setlen(*path_parent, 0); | |
1323 | } | |
1324 | ||
1325 | /* | |
1326 | * If the parent directory matches the pattern, then we do not | |
1327 | * need to check for dtype. | |
1328 | */ | |
1329 | if ((*path_parent)->len && | |
1330 | match_pathname((*path_parent)->buf, (*path_parent)->len, | |
1331 | pattern->base, | |
1332 | pattern->baselen ? pattern->baselen - 1 : 0, | |
1333 | pattern->pattern, pattern->nowildcardlen, | |
1334 | pattern->patternlen, pattern->flags)) | |
1335 | return 1; | |
1336 | ||
1337 | *dtype = resolve_dtype(*dtype, istate, pathname, pathlen); | |
1338 | if (*dtype != DT_DIR) | |
1339 | return 0; | |
1340 | ||
1341 | return 1; | |
1342 | } | |
1343 | ||
578cd7c3 AS |
1344 | /* |
1345 | * Scan the given exclude list in reverse to see whether pathname | |
1346 | * should be ignored. The first match (i.e. the last on the list), if | |
1347 | * any, determines the fate. Returns the exclude_list element which | |
1348 | * matched, or NULL for undecided. | |
453ec4bd | 1349 | */ |
65edd96a | 1350 | static struct path_pattern *last_matching_pattern_from_list(const char *pathname, |
578cd7c3 AS |
1351 | int pathlen, |
1352 | const char *basename, | |
1353 | int *dtype, | |
caa3d554 | 1354 | struct pattern_list *pl, |
2b70e88d | 1355 | struct index_state *istate) |
453ec4bd | 1356 | { |
ab8db613 | 1357 | struct path_pattern *res = NULL; /* undecided */ |
5cee3493 | 1358 | int i; |
f6526728 | 1359 | struct strbuf *path_parent = NULL; |
453ec4bd | 1360 | |
caa3d554 | 1361 | if (!pl->nr) |
578cd7c3 | 1362 | return NULL; /* undefined */ |
d6b8fc30 | 1363 | |
caa3d554 DS |
1364 | for (i = pl->nr - 1; 0 <= i; i--) { |
1365 | struct path_pattern *pattern = pl->patterns[i]; | |
ab8db613 DS |
1366 | const char *exclude = pattern->pattern; |
1367 | int prefix = pattern->nowildcardlen; | |
35a94d44 | 1368 | |
f6526728 DS |
1369 | if (pattern->flags & PATTERN_FLAG_MUSTBEDIR && |
1370 | !path_matches_dir_pattern(pathname, pathlen, &path_parent, | |
1371 | dtype, pattern, istate)) | |
1372 | continue; | |
d6b8fc30 | 1373 | |
4ff89ee5 | 1374 | if (pattern->flags & PATTERN_FLAG_NODIR) { |
593cb880 NTND |
1375 | if (match_basename(basename, |
1376 | pathlen - (basename - pathname), | |
ab8db613 DS |
1377 | exclude, prefix, pattern->patternlen, |
1378 | pattern->flags)) { | |
1379 | res = pattern; | |
e6efecc4 NTND |
1380 | break; |
1381 | } | |
35a94d44 NTND |
1382 | continue; |
1383 | } | |
1384 | ||
ab8db613 DS |
1385 | assert(pattern->baselen == 0 || |
1386 | pattern->base[pattern->baselen - 1] == '/'); | |
b5592632 | 1387 | if (match_pathname(pathname, pathlen, |
ab8db613 DS |
1388 | pattern->base, |
1389 | pattern->baselen ? pattern->baselen - 1 : 0, | |
1390 | exclude, prefix, pattern->patternlen, | |
1391 | pattern->flags)) { | |
1392 | res = pattern; | |
e6efecc4 NTND |
1393 | break; |
1394 | } | |
453ec4bd | 1395 | } |
f6526728 DS |
1396 | |
1397 | if (path_parent) { | |
1398 | strbuf_release(path_parent); | |
1399 | free(path_parent); | |
1400 | } | |
1401 | ||
ab8db613 | 1402 | return res; |
578cd7c3 AS |
1403 | } |
1404 | ||
1405 | /* | |
468ce99b DS |
1406 | * Scan the list of patterns to determine if the ordered list |
1407 | * of patterns matches on 'pathname'. | |
1408 | * | |
1409 | * Return 1 for a match, 0 for not matched and -1 for undecided. | |
578cd7c3 | 1410 | */ |
468ce99b DS |
1411 | enum pattern_match_result path_matches_pattern_list( |
1412 | const char *pathname, int pathlen, | |
1413 | const char *basename, int *dtype, | |
1414 | struct pattern_list *pl, | |
1415 | struct index_state *istate) | |
578cd7c3 | 1416 | { |
ab8db613 | 1417 | struct path_pattern *pattern; |
96cc8ab5 DS |
1418 | struct strbuf parent_pathname = STRBUF_INIT; |
1419 | int result = NOT_MATCHED; | |
69bdbdb0 | 1420 | size_t slash_pos; |
96cc8ab5 DS |
1421 | |
1422 | if (!pl->use_cone_patterns) { | |
1423 | pattern = last_matching_pattern_from_list(pathname, pathlen, basename, | |
1424 | dtype, pl, istate); | |
1425 | if (pattern) { | |
1426 | if (pattern->flags & PATTERN_FLAG_NEGATIVE) | |
1427 | return NOT_MATCHED; | |
1428 | else | |
1429 | return MATCHED; | |
1430 | } | |
1431 | ||
1432 | return UNDECIDED; | |
1433 | } | |
1434 | ||
1435 | if (pl->full_cone) | |
1436 | return MATCHED; | |
1437 | ||
1438 | strbuf_addch(&parent_pathname, '/'); | |
1439 | strbuf_add(&parent_pathname, pathname, pathlen); | |
1440 | ||
69bdbdb0 DS |
1441 | /* |
1442 | * Directory entries are matched if and only if a file | |
1443 | * contained immediately within them is matched. For the | |
1444 | * case of a directory entry, modify the path to create | |
1445 | * a fake filename within this directory, allowing us to | |
1446 | * use the file-base matching logic in an equivalent way. | |
1447 | */ | |
1448 | if (parent_pathname.len > 0 && | |
1449 | parent_pathname.buf[parent_pathname.len - 1] == '/') { | |
1450 | slash_pos = parent_pathname.len - 1; | |
1451 | strbuf_add(&parent_pathname, "-", 1); | |
1452 | } else { | |
1453 | const char *slash_ptr = strrchr(parent_pathname.buf, '/'); | |
1454 | slash_pos = slash_ptr ? slash_ptr - parent_pathname.buf : 0; | |
1455 | } | |
1456 | ||
96cc8ab5 DS |
1457 | if (hashmap_contains_path(&pl->recursive_hashmap, |
1458 | &parent_pathname)) { | |
eb42feca | 1459 | result = MATCHED_RECURSIVE; |
96cc8ab5 | 1460 | goto done; |
468ce99b DS |
1461 | } |
1462 | ||
69bdbdb0 | 1463 | if (!slash_pos) { |
96cc8ab5 DS |
1464 | /* include every file in root */ |
1465 | result = MATCHED; | |
1466 | goto done; | |
1467 | } | |
1468 | ||
69bdbdb0 | 1469 | strbuf_setlen(&parent_pathname, slash_pos); |
96cc8ab5 DS |
1470 | |
1471 | if (hashmap_contains_path(&pl->parent_hashmap, &parent_pathname)) { | |
1472 | result = MATCHED; | |
1473 | goto done; | |
1474 | } | |
1475 | ||
1476 | if (hashmap_contains_parent(&pl->recursive_hashmap, | |
1477 | pathname, | |
1478 | &parent_pathname)) | |
eb42feca | 1479 | result = MATCHED_RECURSIVE; |
96cc8ab5 DS |
1480 | |
1481 | done: | |
1482 | strbuf_release(&parent_pathname); | |
1483 | return result; | |
453ec4bd LT |
1484 | } |
1485 | ||
02155c8c DS |
1486 | int init_sparse_checkout_patterns(struct index_state *istate) |
1487 | { | |
1488 | if (!core_apply_sparse_checkout) | |
1489 | return 1; | |
1490 | if (istate->sparse_checkout_patterns) | |
1491 | return 0; | |
1492 | ||
1493 | CALLOC_ARRAY(istate->sparse_checkout_patterns, 1); | |
1494 | ||
1495 | if (get_sparse_checkout_patterns(istate->sparse_checkout_patterns) < 0) { | |
1496 | FREE_AND_NULL(istate->sparse_checkout_patterns); | |
1497 | return -1; | |
1498 | } | |
1499 | ||
1500 | return 0; | |
1501 | } | |
1502 | ||
1503 | static int path_in_sparse_checkout_1(const char *path, | |
1504 | struct index_state *istate, | |
1505 | int require_cone_mode) | |
1506 | { | |
02155c8c | 1507 | int dtype = DT_REG; |
20141e32 MT |
1508 | enum pattern_match_result match = UNDECIDED; |
1509 | const char *end, *slash; | |
02155c8c DS |
1510 | |
1511 | /* | |
1512 | * We default to accepting a path if there are no patterns or | |
1513 | * they are of the wrong type. | |
1514 | */ | |
1515 | if (init_sparse_checkout_patterns(istate) || | |
1516 | (require_cone_mode && | |
1517 | !istate->sparse_checkout_patterns->use_cone_patterns)) | |
1518 | return 1; | |
1519 | ||
20141e32 MT |
1520 | /* |
1521 | * If UNDECIDED, use the match from the parent dir (recursively), or | |
1522 | * fall back to NOT_MATCHED at the topmost level. Note that cone mode | |
1523 | * never returns UNDECIDED, so we will execute only one iteration in | |
1524 | * this case. | |
1525 | */ | |
1526 | for (end = path + strlen(path); | |
1527 | end > path && match == UNDECIDED; | |
1528 | end = slash) { | |
1529 | ||
1530 | for (slash = end - 1; slash > path && *slash != '/'; slash--) | |
1531 | ; /* do nothing */ | |
1532 | ||
1533 | match = path_matches_pattern_list(path, end - path, | |
1534 | slash > path ? slash + 1 : path, &dtype, | |
1535 | istate->sparse_checkout_patterns, istate); | |
1536 | ||
1537 | /* We are going to match the parent dir now */ | |
1538 | dtype = DT_DIR; | |
1539 | } | |
1540 | return match > 0; | |
02155c8c DS |
1541 | } |
1542 | ||
1543 | int path_in_sparse_checkout(const char *path, | |
1544 | struct index_state *istate) | |
1545 | { | |
1546 | return path_in_sparse_checkout_1(path, istate, 0); | |
1547 | } | |
1548 | ||
1549 | int path_in_cone_mode_sparse_checkout(const char *path, | |
1550 | struct index_state *istate) | |
1551 | { | |
1552 | return path_in_sparse_checkout_1(path, istate, 1); | |
1553 | } | |
1554 | ||
65edd96a | 1555 | static struct path_pattern *last_matching_pattern_from_lists( |
ab8db613 DS |
1556 | struct dir_struct *dir, struct index_state *istate, |
1557 | const char *pathname, int pathlen, | |
1558 | const char *basename, int *dtype_p) | |
46aa2f95 KB |
1559 | { |
1560 | int i, j; | |
1561 | struct exclude_list_group *group; | |
ab8db613 | 1562 | struct path_pattern *pattern; |
46aa2f95 KB |
1563 | for (i = EXC_CMDL; i <= EXC_FILE; i++) { |
1564 | group = &dir->exclude_list_group[i]; | |
1565 | for (j = group->nr - 1; j >= 0; j--) { | |
65edd96a | 1566 | pattern = last_matching_pattern_from_list( |
46aa2f95 | 1567 | pathname, pathlen, basename, dtype_p, |
caa3d554 | 1568 | &group->pl[j], istate); |
ab8db613 DS |
1569 | if (pattern) |
1570 | return pattern; | |
46aa2f95 KB |
1571 | } |
1572 | } | |
1573 | return NULL; | |
1574 | } | |
1575 | ||
6cd5c582 KB |
1576 | /* |
1577 | * Loads the per-directory exclude list for the substring of base | |
1578 | * which has a char length of baselen. | |
1579 | */ | |
e799ed40 BW |
1580 | static void prep_exclude(struct dir_struct *dir, |
1581 | struct index_state *istate, | |
1582 | const char *base, int baselen) | |
6cd5c582 KB |
1583 | { |
1584 | struct exclude_list_group *group; | |
caa3d554 | 1585 | struct pattern_list *pl; |
6cd5c582 | 1586 | struct exclude_stack *stk = NULL; |
0dcb8d7f | 1587 | struct untracked_cache_dir *untracked; |
6cd5c582 KB |
1588 | int current; |
1589 | ||
6cd5c582 KB |
1590 | group = &dir->exclude_list_group[EXC_DIRS]; |
1591 | ||
d961baa8 NTND |
1592 | /* |
1593 | * Pop the exclude lists from the EXCL_DIRS exclude_list_group | |
6cd5c582 | 1594 | * which originate from directories not in the prefix of the |
d961baa8 NTND |
1595 | * path being checked. |
1596 | */ | |
6cd5c582 KB |
1597 | while ((stk = dir->exclude_stack) != NULL) { |
1598 | if (stk->baselen <= baselen && | |
aceb9429 | 1599 | !strncmp(dir->basebuf.buf, base, stk->baselen)) |
6cd5c582 | 1600 | break; |
caa3d554 | 1601 | pl = &group->pl[dir->exclude_stack->exclude_ix]; |
6cd5c582 | 1602 | dir->exclude_stack = stk->prev; |
ab8db613 | 1603 | dir->pattern = NULL; |
caa3d554 | 1604 | free((char *)pl->src); /* see strbuf_detach() below */ |
65edd96a | 1605 | clear_pattern_list(pl); |
6cd5c582 KB |
1606 | free(stk); |
1607 | group->nr--; | |
1608 | } | |
1609 | ||
95c6f271 | 1610 | /* Skip traversing into sub directories if the parent is excluded */ |
ab8db613 | 1611 | if (dir->pattern) |
95c6f271 KB |
1612 | return; |
1613 | ||
aceb9429 NTND |
1614 | /* |
1615 | * Lazy initialization. All call sites currently just | |
1616 | * memset(dir, 0, sizeof(*dir)) before use. Changing all of | |
1617 | * them seems lots of work for little benefit. | |
1618 | */ | |
1619 | if (!dir->basebuf.buf) | |
1620 | strbuf_init(&dir->basebuf, PATH_MAX); | |
1621 | ||
6cd5c582 KB |
1622 | /* Read from the parent directories and push them down. */ |
1623 | current = stk ? stk->baselen : -1; | |
aceb9429 | 1624 | strbuf_setlen(&dir->basebuf, current < 0 ? 0 : current); |
0dcb8d7f NTND |
1625 | if (dir->untracked) |
1626 | untracked = stk ? stk->ucd : dir->untracked->root; | |
1627 | else | |
1628 | untracked = NULL; | |
1629 | ||
6cd5c582 | 1630 | while (current < baselen) { |
6cd5c582 | 1631 | const char *cp; |
4b33e602 | 1632 | struct oid_stat oid_stat; |
6cd5c582 | 1633 | |
ca56dadb | 1634 | CALLOC_ARRAY(stk, 1); |
6cd5c582 KB |
1635 | if (current < 0) { |
1636 | cp = base; | |
1637 | current = 0; | |
d961baa8 | 1638 | } else { |
6cd5c582 KB |
1639 | cp = strchr(base + current + 1, '/'); |
1640 | if (!cp) | |
1641 | die("oops in prep_exclude"); | |
1642 | cp++; | |
0dcb8d7f NTND |
1643 | untracked = |
1644 | lookup_untracked(dir->untracked, untracked, | |
1645 | base + current, | |
1646 | cp - base - current); | |
6cd5c582 KB |
1647 | } |
1648 | stk->prev = dir->exclude_stack; | |
1649 | stk->baselen = cp - base; | |
95c6f271 | 1650 | stk->exclude_ix = group->nr; |
0dcb8d7f | 1651 | stk->ucd = untracked; |
65edd96a | 1652 | pl = add_pattern_list(dir, EXC_DIRS, NULL); |
aceb9429 NTND |
1653 | strbuf_add(&dir->basebuf, base + current, stk->baselen - current); |
1654 | assert(stk->baselen == dir->basebuf.len); | |
95c6f271 KB |
1655 | |
1656 | /* Abort if the directory is excluded */ | |
1657 | if (stk->baselen) { | |
1658 | int dt = DT_DIR; | |
aceb9429 | 1659 | dir->basebuf.buf[stk->baselen - 1] = 0; |
65edd96a | 1660 | dir->pattern = last_matching_pattern_from_lists(dir, |
e799ed40 | 1661 | istate, |
aceb9429 NTND |
1662 | dir->basebuf.buf, stk->baselen - 1, |
1663 | dir->basebuf.buf + current, &dt); | |
1664 | dir->basebuf.buf[stk->baselen - 1] = '/'; | |
ab8db613 | 1665 | if (dir->pattern && |
4ff89ee5 | 1666 | dir->pattern->flags & PATTERN_FLAG_NEGATIVE) |
ab8db613 DS |
1667 | dir->pattern = NULL; |
1668 | if (dir->pattern) { | |
95c6f271 KB |
1669 | dir->exclude_stack = stk; |
1670 | return; | |
1671 | } | |
1672 | } | |
1673 | ||
aceb9429 | 1674 | /* Try to read per-directory file */ |
4b33e602 PO |
1675 | oidclr(&oid_stat.oid); |
1676 | oid_stat.valid = 0; | |
27b099ae NTND |
1677 | if (dir->exclude_per_dir && |
1678 | /* | |
1679 | * If we know that no files have been added in | |
1680 | * this directory (i.e. valid_cached_dir() has | |
1681 | * been executed and set untracked->valid) .. | |
1682 | */ | |
1683 | (!untracked || !untracked->valid || | |
1684 | /* | |
1685 | * .. and .gitignore does not exist before | |
70c369cd | 1686 | * (i.e. null exclude_oid). Then we can skip |
7687252f DT |
1687 | * loading .gitignore, which would result in |
1688 | * ENOENT anyway. | |
27b099ae | 1689 | */ |
70c369cd | 1690 | !is_null_oid(&untracked->exclude_oid))) { |
95c6f271 KB |
1691 | /* |
1692 | * dir->basebuf gets reused by the traversal, but we | |
1693 | * need fname to remain unchanged to ensure the src | |
ab8db613 | 1694 | * member of each struct path_pattern correctly |
95c6f271 | 1695 | * back-references its source file. Other invocations |
65edd96a | 1696 | * of add_pattern_list provide stable strings, so we |
aceb9429 | 1697 | * strbuf_detach() and free() here in the caller. |
95c6f271 | 1698 | */ |
aceb9429 NTND |
1699 | struct strbuf sb = STRBUF_INIT; |
1700 | strbuf_addbuf(&sb, &dir->basebuf); | |
1701 | strbuf_addstr(&sb, dir->exclude_per_dir); | |
caa3d554 | 1702 | pl->src = strbuf_detach(&sb, NULL); |
65edd96a | 1703 | add_patterns(pl->src, pl->src, stk->baselen, pl, istate, |
feb9b779 | 1704 | PATTERN_NOFOLLOW, |
4b33e602 | 1705 | untracked ? &oid_stat : NULL); |
0dcb8d7f | 1706 | } |
5ebf79ad NTND |
1707 | /* |
1708 | * NEEDSWORK: when untracked cache is enabled, prep_exclude() | |
1709 | * will first be called in valid_cached_dir() then maybe many | |
65edd96a DS |
1710 | * times more in last_matching_pattern(). When the cache is |
1711 | * used, last_matching_pattern() will not be called and | |
5ebf79ad NTND |
1712 | * reading .gitignore content will be a waste. |
1713 | * | |
1714 | * So when it's called by valid_cached_dir() and we can get | |
1715 | * .gitignore SHA-1 from the index (i.e. .gitignore is not | |
1716 | * modified on work tree), we could delay reading the | |
1717 | * .gitignore content until we absolutely need it in | |
65edd96a | 1718 | * last_matching_pattern(). Be careful about ignore rule |
5ebf79ad NTND |
1719 | * order, though, if you do that. |
1720 | */ | |
1721 | if (untracked && | |
9001dc2a | 1722 | !oideq(&oid_stat.oid, &untracked->exclude_oid)) { |
5ebf79ad | 1723 | invalidate_gitignore(dir->untracked, untracked); |
70c369cd | 1724 | oidcpy(&untracked->exclude_oid, &oid_stat.oid); |
95c6f271 | 1725 | } |
6cd5c582 KB |
1726 | dir->exclude_stack = stk; |
1727 | current = stk->baselen; | |
1728 | } | |
aceb9429 | 1729 | strbuf_setlen(&dir->basebuf, baselen); |
6cd5c582 KB |
1730 | } |
1731 | ||
f4cd69a6 AS |
1732 | /* |
1733 | * Loads the exclude lists for the directory containing pathname, then | |
1734 | * scans all exclude lists to determine whether pathname is excluded. | |
1735 | * Returns the exclude_list element which matched, or NULL for | |
1736 | * undecided. | |
1737 | */ | |
65edd96a | 1738 | struct path_pattern *last_matching_pattern(struct dir_struct *dir, |
a0bba65b BW |
1739 | struct index_state *istate, |
1740 | const char *pathname, | |
1741 | int *dtype_p) | |
453ec4bd LT |
1742 | { |
1743 | int pathlen = strlen(pathname); | |
68492fc7 LK |
1744 | const char *basename = strrchr(pathname, '/'); |
1745 | basename = (basename) ? basename+1 : pathname; | |
453ec4bd | 1746 | |
a0bba65b | 1747 | prep_exclude(dir, istate, pathname, basename-pathname); |
c082df24 | 1748 | |
ab8db613 DS |
1749 | if (dir->pattern) |
1750 | return dir->pattern; | |
95c6f271 | 1751 | |
65edd96a | 1752 | return last_matching_pattern_from_lists(dir, istate, pathname, pathlen, |
46aa2f95 | 1753 | basename, dtype_p); |
f4cd69a6 AS |
1754 | } |
1755 | ||
1756 | /* | |
1757 | * Loads the exclude lists for the directory containing pathname, then | |
1758 | * scans all exclude lists to determine whether pathname is excluded. | |
1759 | * Returns 1 if true, otherwise 0. | |
1760 | */ | |
a0bba65b BW |
1761 | int is_excluded(struct dir_struct *dir, struct index_state *istate, |
1762 | const char *pathname, int *dtype_p) | |
f4cd69a6 | 1763 | { |
ab8db613 | 1764 | struct path_pattern *pattern = |
65edd96a | 1765 | last_matching_pattern(dir, istate, pathname, dtype_p); |
ab8db613 | 1766 | if (pattern) |
4ff89ee5 | 1767 | return pattern->flags & PATTERN_FLAG_NEGATIVE ? 0 : 1; |
453ec4bd LT |
1768 | return 0; |
1769 | } | |
1770 | ||
f3fa1838 JH |
1771 | static struct dir_entry *dir_entry_new(const char *pathname, int len) |
1772 | { | |
453ec4bd LT |
1773 | struct dir_entry *ent; |
1774 | ||
96ffc06f | 1775 | FLEX_ALLOC_MEM(ent, name, pathname, len); |
453ec4bd | 1776 | ent->len = len; |
4d06f8ac | 1777 | return ent; |
453ec4bd LT |
1778 | } |
1779 | ||
9e58beca BW |
1780 | static struct dir_entry *dir_add_name(struct dir_struct *dir, |
1781 | struct index_state *istate, | |
1782 | const char *pathname, int len) | |
6815e569 | 1783 | { |
9e58beca | 1784 | if (index_file_exists(istate, pathname, len, ignore_case)) |
6815e569 JK |
1785 | return NULL; |
1786 | ||
25fd2f7a | 1787 | ALLOC_GROW(dir->entries, dir->nr+1, dir->alloc); |
6815e569 JK |
1788 | return dir->entries[dir->nr++] = dir_entry_new(pathname, len); |
1789 | } | |
1790 | ||
9e58beca BW |
1791 | struct dir_entry *dir_add_ignored(struct dir_struct *dir, |
1792 | struct index_state *istate, | |
1793 | const char *pathname, int len) | |
2abd31b0 | 1794 | { |
9e58beca | 1795 | if (!index_name_is_other(istate, pathname, len)) |
2abd31b0 JK |
1796 | return NULL; |
1797 | ||
25fd2f7a | 1798 | ALLOC_GROW(dir->ignored, dir->ignored_nr+1, dir->ignored_alloc); |
2abd31b0 JK |
1799 | return dir->ignored[dir->ignored_nr++] = dir_entry_new(pathname, len); |
1800 | } | |
1801 | ||
09595258 LT |
1802 | enum exist_status { |
1803 | index_nonexistent = 0, | |
1804 | index_directory, | |
4b05548f | 1805 | index_gitdir |
09595258 LT |
1806 | }; |
1807 | ||
5102c617 | 1808 | /* |
71261027 | 1809 | * Do not use the alphabetically sorted index to look up |
5102c617 | 1810 | * the directory name; instead, use the case insensitive |
ebbd7439 | 1811 | * directory hash. |
5102c617 | 1812 | */ |
ae520e36 BW |
1813 | static enum exist_status directory_exists_in_index_icase(struct index_state *istate, |
1814 | const char *dirname, int len) | |
5102c617 | 1815 | { |
41284eb0 | 1816 | struct cache_entry *ce; |
5102c617 | 1817 | |
ae520e36 | 1818 | if (index_dir_exists(istate, dirname, len)) |
5102c617 JJ |
1819 | return index_directory; |
1820 | ||
ae520e36 | 1821 | ce = index_file_exists(istate, dirname, len, ignore_case); |
41284eb0 | 1822 | if (ce && S_ISGITLINK(ce->ce_mode)) |
5102c617 JJ |
1823 | return index_gitdir; |
1824 | ||
5102c617 JJ |
1825 | return index_nonexistent; |
1826 | } | |
1827 | ||
09595258 LT |
1828 | /* |
1829 | * The index sorts alphabetically by entry name, which | |
1830 | * means that a gitlink sorts as '\0' at the end, while | |
1831 | * a directory (which is defined not as an entry, but as | |
1832 | * the files it contains) will sort with the '/' at the | |
1833 | * end. | |
1834 | */ | |
ae520e36 BW |
1835 | static enum exist_status directory_exists_in_index(struct index_state *istate, |
1836 | const char *dirname, int len) | |
453ec4bd | 1837 | { |
5102c617 JJ |
1838 | int pos; |
1839 | ||
1840 | if (ignore_case) | |
ae520e36 | 1841 | return directory_exists_in_index_icase(istate, dirname, len); |
5102c617 | 1842 | |
ae520e36 | 1843 | pos = index_name_pos(istate, dirname, len); |
09595258 LT |
1844 | if (pos < 0) |
1845 | pos = -pos-1; | |
ae520e36 BW |
1846 | while (pos < istate->cache_nr) { |
1847 | const struct cache_entry *ce = istate->cache[pos++]; | |
09595258 LT |
1848 | unsigned char endchar; |
1849 | ||
1850 | if (strncmp(ce->name, dirname, len)) | |
1851 | break; | |
1852 | endchar = ce->name[len]; | |
1853 | if (endchar > '/') | |
1854 | break; | |
1855 | if (endchar == '/') | |
1856 | return index_directory; | |
7a51ed66 | 1857 | if (!endchar && S_ISGITLINK(ce->ce_mode)) |
09595258 LT |
1858 | return index_gitdir; |
1859 | } | |
1860 | return index_nonexistent; | |
1861 | } | |
1862 | ||
1863 | /* | |
1864 | * When we find a directory when traversing the filesystem, we | |
1865 | * have three distinct cases: | |
1866 | * | |
1867 | * - ignore it | |
1868 | * - see it as a directory | |
1869 | * - recurse into it | |
1870 | * | |
1871 | * and which one we choose depends on a combination of existing | |
1872 | * git index contents and the flags passed into the directory | |
1873 | * traversal routine. | |
1874 | * | |
1875 | * Case 1: If we *already* have entries in the index under that | |
5bd8e2d8 KB |
1876 | * directory name, we always recurse into the directory to see |
1877 | * all the files. | |
09595258 LT |
1878 | * |
1879 | * Case 2: If we *already* have that directory name as a gitlink, | |
1880 | * we always continue to see it as a gitlink, regardless of whether | |
1881 | * there is an actual git directory there or not (it might not | |
1882 | * be checked out as a subproject!) | |
1883 | * | |
1884 | * Case 3: if we didn't have it in the index previously, we | |
1885 | * have a few sub-cases: | |
1886 | * | |
4e689d81 DS |
1887 | * (a) if DIR_SHOW_OTHER_DIRECTORIES flag is set, we show it as |
1888 | * just a directory, unless DIR_HIDE_EMPTY_DIRECTORIES is | |
defd7c7b KB |
1889 | * also true, in which case we need to check if it contains any |
1890 | * untracked and / or ignored files. | |
4e689d81 DS |
1891 | * (b) if it looks like a git directory and we don't have the |
1892 | * DIR_NO_GITLINKS flag, then we treat it as a gitlink, and | |
1893 | * show it as a directory. | |
09595258 LT |
1894 | * (c) otherwise, we recurse into it. |
1895 | */ | |
defd7c7b | 1896 | static enum path_treatment treat_directory(struct dir_struct *dir, |
0ef8e169 | 1897 | struct index_state *istate, |
0dcb8d7f | 1898 | struct untracked_cache_dir *untracked, |
2df179d3 | 1899 | const char *dirname, int len, int baselen, int excluded, |
e1b8c7bd | 1900 | const struct pathspec *pathspec) |
09595258 | 1901 | { |
8d92fb29 EN |
1902 | /* |
1903 | * WARNING: From this function, you can return path_recurse or you | |
1904 | * can call read_directory_recursive() (or neither), but | |
1905 | * you CAN'T DO BOTH. | |
1906 | */ | |
1907 | enum path_treatment state; | |
7f45ab2d | 1908 | int matches_how = 0; |
16846444 EN |
1909 | int nested_repo = 0, check_only, stop_early; |
1910 | int old_ignored_nr, old_untracked_nr; | |
09595258 | 1911 | /* The "len-1" is to strip the final '/' */ |
0bbd0e8b | 1912 | enum exist_status status = directory_exists_in_index(istate, dirname, len-1); |
09595258 | 1913 | |
0bbd0e8b DS |
1914 | if (status == index_directory) |
1915 | return path_recurse; | |
1916 | if (status == index_gitdir) | |
26c986e1 | 1917 | return path_none; |
0bbd0e8b DS |
1918 | if (status != index_nonexistent) |
1919 | BUG("Unhandled value for directory_exists_in_index: %d\n", status); | |
09595258 | 1920 | |
7f45ab2d EN |
1921 | /* |
1922 | * We don't want to descend into paths that don't match the necessary | |
1923 | * patterns. Clearly, if we don't have a pathspec, then we can't check | |
1924 | * for matching patterns. Also, if (excluded) then we know we matched | |
1925 | * the exclusion patterns so as an optimization we can skip checking | |
1926 | * for matching patterns. | |
1927 | */ | |
1928 | if (pathspec && !excluded) { | |
f1f061e1 EN |
1929 | matches_how = match_pathspec_with_flags(istate, pathspec, |
1930 | dirname, len, | |
1931 | 0 /* prefix */, | |
1932 | NULL /* seen */, | |
1933 | DO_MATCH_LEADING_PATHSPEC); | |
7f45ab2d EN |
1934 | if (!matches_how) |
1935 | return path_none; | |
1936 | } | |
09487f2c | 1937 | |
7f45ab2d | 1938 | |
0bbd0e8b DS |
1939 | if ((dir->flags & DIR_SKIP_NESTED_GIT) || |
1940 | !(dir->flags & DIR_NO_GITLINKS)) { | |
1941 | struct strbuf sb = STRBUF_INIT; | |
1942 | strbuf_addstr(&sb, dirname); | |
1943 | nested_repo = is_nonbare_repository_dir(&sb); | |
1944 | strbuf_release(&sb); | |
1945 | } | |
ab282aa5 EN |
1946 | if (nested_repo) { |
1947 | if ((dir->flags & DIR_SKIP_NESTED_GIT) || | |
1948 | (matches_how == MATCHED_RECURSIVELY_LEADING_PATHSPEC)) | |
1949 | return path_none; | |
1950 | return excluded ? path_excluded : path_untracked; | |
1951 | } | |
09595258 | 1952 | |
0bbd0e8b | 1953 | if (!(dir->flags & DIR_SHOW_OTHER_DIRECTORIES)) { |
2df179d3 EN |
1954 | if (excluded && |
1955 | (dir->flags & DIR_SHOW_IGNORED_TOO) && | |
1956 | (dir->flags & DIR_SHOW_IGNORED_TOO_MODE_MATCHING)) { | |
eec0f7f2 JM |
1957 | |
1958 | /* | |
1959 | * This is an excluded directory and we are | |
1960 | * showing ignored paths that match an exclude | |
1961 | * pattern. (e.g. show directory as ignored | |
1962 | * only if it matches an exclude pattern). | |
1963 | * This path will either be 'path_excluded` | |
1964 | * (if we are showing empty directories or if | |
1965 | * the directory is not empty), or will be | |
1966 | * 'path_none' (empty directory, and we are | |
1967 | * not showing empty directories). | |
1968 | */ | |
1969 | if (!(dir->flags & DIR_HIDE_EMPTY_DIRECTORIES)) | |
1970 | return path_excluded; | |
1971 | ||
1972 | if (read_directory_recursive(dir, istate, dirname, len, | |
1973 | untracked, 1, 1, pathspec) == path_excluded) | |
1974 | return path_excluded; | |
1975 | ||
1976 | return path_none; | |
1977 | } | |
defd7c7b | 1978 | return path_recurse; |
09595258 LT |
1979 | } |
1980 | ||
aa6e1b21 | 1981 | assert(dir->flags & DIR_SHOW_OTHER_DIRECTORIES); |
721ac4ed | 1982 | |
8d92fb29 EN |
1983 | /* |
1984 | * If we have a pathspec which could match something _below_ this | |
1985 | * directory (e.g. when checking 'subdir/' having a pathspec like | |
1986 | * 'subdir/some/deep/path/file' or 'subdir/widget-*.c'), then we | |
1987 | * need to recurse. | |
1988 | */ | |
7f45ab2d EN |
1989 | if (matches_how == MATCHED_RECURSIVELY_LEADING_PATHSPEC) |
1990 | return path_recurse; | |
8d92fb29 | 1991 | |
aa6e1b21 EN |
1992 | /* Special cases for where this directory is excluded/ignored */ |
1993 | if (excluded) { | |
1994 | /* | |
4e689d81 | 1995 | * If DIR_SHOW_OTHER_DIRECTORIES is set and we're not |
aa6e1b21 EN |
1996 | * hiding empty directories, there is no need to |
1997 | * recurse into an ignored directory. | |
1998 | */ | |
1999 | if (!(dir->flags & DIR_HIDE_EMPTY_DIRECTORIES)) | |
2000 | return path_excluded; | |
2001 | ||
2002 | /* | |
2003 | * Even if we are hiding empty directories, we can still avoid | |
2004 | * recursing into ignored directories for DIR_SHOW_IGNORED_TOO | |
2005 | * if DIR_SHOW_IGNORED_TOO_MODE_MATCHING is also set. | |
2006 | */ | |
2007 | if ((dir->flags & DIR_SHOW_IGNORED_TOO) && | |
2008 | (dir->flags & DIR_SHOW_IGNORED_TOO_MODE_MATCHING)) | |
2009 | return path_excluded; | |
2010 | } | |
2011 | ||
8d92fb29 | 2012 | /* |
aa6e1b21 | 2013 | * Other than the path_recurse case above, we only need to |
dd55fc0d | 2014 | * recurse into untracked directories if any of the following |
aa6e1b21 | 2015 | * bits is set: |
dd55fc0d EN |
2016 | * - DIR_SHOW_IGNORED (because then we need to determine if |
2017 | * there are ignored entries below) | |
2018 | * - DIR_SHOW_IGNORED_TOO (same as above) | |
8d92fb29 EN |
2019 | * - DIR_HIDE_EMPTY_DIRECTORIES (because we have to determine if |
2020 | * the directory is empty) | |
2021 | */ | |
aa6e1b21 | 2022 | if (!excluded && |
dd55fc0d EN |
2023 | !(dir->flags & (DIR_SHOW_IGNORED | |
2024 | DIR_SHOW_IGNORED_TOO | | |
aa6e1b21 EN |
2025 | DIR_HIDE_EMPTY_DIRECTORIES))) { |
2026 | return path_untracked; | |
2027 | } | |
8d92fb29 EN |
2028 | |
2029 | /* | |
e6c0be92 EN |
2030 | * Even if we don't want to know all the paths under an untracked or |
2031 | * ignored directory, we may still need to go into the directory to | |
2032 | * determine if it is empty (because with DIR_HIDE_EMPTY_DIRECTORIES, | |
2033 | * an empty directory should be path_none instead of path_excluded or | |
2034 | * path_untracked). | |
8d92fb29 EN |
2035 | */ |
2036 | check_only = ((dir->flags & DIR_HIDE_EMPTY_DIRECTORIES) && | |
2037 | !(dir->flags & DIR_SHOW_IGNORED_TOO)); | |
2038 | ||
2039 | /* | |
2040 | * However, there's another optimization possible as a subset of | |
2041 | * check_only, based on the cases we have to consider: | |
2042 | * A) Directory matches no exclude patterns: | |
2043 | * * Directory is empty => path_none | |
2044 | * * Directory has an untracked file under it => path_untracked | |
2045 | * * Directory has only ignored files under it => path_excluded | |
2046 | * B) Directory matches an exclude pattern: | |
2047 | * * Directory is empty => path_none | |
2048 | * * Directory has an untracked file under it => path_excluded | |
2049 | * * Directory has only ignored files under it => path_excluded | |
2050 | * In case A, we can exit as soon as we've found an untracked | |
2051 | * file but otherwise have to walk all files. In case B, though, | |
2052 | * we can stop at the first file we find under the directory. | |
2053 | */ | |
2054 | stop_early = check_only && excluded; | |
2055 | ||
2056 | /* | |
2057 | * If /every/ file within an untracked directory is ignored, then | |
2058 | * we want to treat the directory as ignored (for e.g. status | |
2059 | * --porcelain), without listing the individual ignored files | |
2060 | * underneath. To do so, we'll save the current ignored_nr, and | |
2061 | * pop all the ones added after it if it turns out the entire | |
16846444 EN |
2062 | * directory is ignored. Also, when DIR_SHOW_IGNORED_TOO and |
2063 | * !DIR_KEEP_UNTRACKED_CONTENTS then we don't want to show | |
2064 | * untracked paths so will need to pop all those off the last | |
2065 | * after we traverse. | |
8d92fb29 EN |
2066 | */ |
2067 | old_ignored_nr = dir->ignored_nr; | |
16846444 | 2068 | old_untracked_nr = dir->nr; |
8d92fb29 EN |
2069 | |
2070 | /* Actually recurse into dirname now, we'll fixup the state later. */ | |
2e5910f2 DT |
2071 | untracked = lookup_untracked(dir->untracked, untracked, |
2072 | dirname + baselen, len - baselen); | |
8d92fb29 EN |
2073 | state = read_directory_recursive(dir, istate, dirname, len, untracked, |
2074 | check_only, stop_early, pathspec); | |
2075 | ||
2076 | /* There are a variety of reasons we may need to fixup the state... */ | |
2077 | if (state == path_excluded) { | |
2078 | /* state == path_excluded implies all paths under | |
2079 | * dirname were ignored... | |
2080 | * | |
2081 | * if running e.g. `git status --porcelain --ignored=matching`, | |
2082 | * then we want to see the subpaths that are ignored. | |
2083 | * | |
2084 | * if running e.g. just `git status --porcelain`, then | |
2085 | * we just want the directory itself to be listed as ignored | |
2086 | * and not the individual paths underneath. | |
2087 | */ | |
2088 | int want_ignored_subpaths = | |
2089 | ((dir->flags & DIR_SHOW_IGNORED_TOO) && | |
2090 | (dir->flags & DIR_SHOW_IGNORED_TOO_MODE_MATCHING)); | |
2091 | ||
2092 | if (want_ignored_subpaths) { | |
2093 | /* | |
2094 | * with --ignored=matching, we want the subpaths | |
2095 | * INSTEAD of the directory itself. | |
2096 | */ | |
2097 | state = path_none; | |
2098 | } else { | |
2099 | int i; | |
2100 | for (i = old_ignored_nr + 1; i<dir->ignored_nr; ++i) | |
2101 | FREE_AND_NULL(dir->ignored[i]); | |
2102 | dir->ignored_nr = old_ignored_nr; | |
2103 | } | |
2104 | } | |
5aaa7fd3 JM |
2105 | |
2106 | /* | |
16846444 EN |
2107 | * We may need to ignore some of the untracked paths we found while |
2108 | * traversing subdirectories. | |
5aaa7fd3 | 2109 | */ |
16846444 EN |
2110 | if ((dir->flags & DIR_SHOW_IGNORED_TOO) && |
2111 | !(dir->flags & DIR_KEEP_UNTRACKED_CONTENTS)) { | |
2112 | int i; | |
2113 | for (i = old_untracked_nr + 1; i<dir->nr; ++i) | |
2114 | FREE_AND_NULL(dir->entries[i]); | |
2115 | dir->nr = old_untracked_nr; | |
2116 | } | |
2117 | ||
5aaa7fd3 | 2118 | /* |
8d92fb29 EN |
2119 | * If there is nothing under the current directory and we are not |
2120 | * hiding empty directories, then we need to report on the | |
2121 | * untracked or ignored status of the directory itself. | |
5aaa7fd3 | 2122 | */ |
8d92fb29 EN |
2123 | if (state == path_none && !(dir->flags & DIR_HIDE_EMPTY_DIRECTORIES)) |
2124 | state = excluded ? path_excluded : path_untracked; | |
2125 | ||
8d92fb29 | 2126 | return state; |
453ec4bd LT |
2127 | } |
2128 | ||
9fc42d60 LT |
2129 | /* |
2130 | * This is an inexact early pruning of any recursive directory | |
2131 | * reading - if the path cannot possibly be in the pathspec, | |
2132 | * return true, and we'll skip it early. | |
2133 | */ | |
e1b8c7bd BW |
2134 | static int simplify_away(const char *path, int pathlen, |
2135 | const struct pathspec *pathspec) | |
9fc42d60 | 2136 | { |
e1b8c7bd | 2137 | int i; |
9fc42d60 | 2138 | |
e1b8c7bd BW |
2139 | if (!pathspec || !pathspec->nr) |
2140 | return 0; | |
2141 | ||
2142 | GUARD_PATHSPEC(pathspec, | |
2143 | PATHSPEC_FROMTOP | | |
2144 | PATHSPEC_MAXDEPTH | | |
2145 | PATHSPEC_LITERAL | | |
2146 | PATHSPEC_GLOB | | |
2147 | PATHSPEC_ICASE | | |
b0db7046 BW |
2148 | PATHSPEC_EXCLUDE | |
2149 | PATHSPEC_ATTR); | |
e1b8c7bd BW |
2150 | |
2151 | for (i = 0; i < pathspec->nr; i++) { | |
2152 | const struct pathspec_item *item = &pathspec->items[i]; | |
2153 | int len = item->nowildcard_len; | |
2154 | ||
2155 | if (len > pathlen) | |
2156 | len = pathlen; | |
2157 | if (!ps_strncmp(item, item->match, path, len)) | |
2158 | return 0; | |
9fc42d60 | 2159 | } |
e1b8c7bd BW |
2160 | |
2161 | return 1; | |
9fc42d60 LT |
2162 | } |
2163 | ||
29209cbe JK |
2164 | /* |
2165 | * This function tells us whether an excluded path matches a | |
2166 | * list of "interesting" pathspecs. That is, whether a path matched | |
2167 | * by any of the pathspecs could possibly be ignored by excluding | |
2168 | * the specified path. This can happen if: | |
2169 | * | |
2170 | * 1. the path is mentioned explicitly in the pathspec | |
2171 | * | |
2172 | * 2. the path is a directory prefix of some element in the | |
2173 | * pathspec | |
2174 | */ | |
e1b8c7bd BW |
2175 | static int exclude_matches_pathspec(const char *path, int pathlen, |
2176 | const struct pathspec *pathspec) | |
2177 | { | |
2178 | int i; | |
2179 | ||
2180 | if (!pathspec || !pathspec->nr) | |
2181 | return 0; | |
2182 | ||
2183 | GUARD_PATHSPEC(pathspec, | |
2184 | PATHSPEC_FROMTOP | | |
2185 | PATHSPEC_MAXDEPTH | | |
2186 | PATHSPEC_LITERAL | | |
2187 | PATHSPEC_GLOB | | |
2188 | PATHSPEC_ICASE | | |
2189 | PATHSPEC_EXCLUDE); | |
2190 | ||
2191 | for (i = 0; i < pathspec->nr; i++) { | |
2192 | const struct pathspec_item *item = &pathspec->items[i]; | |
2193 | int len = item->nowildcard_len; | |
2194 | ||
2195 | if (len == pathlen && | |
2196 | !ps_strncmp(item, item->match, path, pathlen)) | |
2197 | return 1; | |
2198 | if (len > pathlen && | |
2199 | item->match[pathlen] == '/' && | |
2200 | !ps_strncmp(item, item->match, path, pathlen)) | |
2201 | return 1; | |
e96980ef JK |
2202 | } |
2203 | return 0; | |
2204 | } | |
2205 | ||
98f2a687 BW |
2206 | static int get_index_dtype(struct index_state *istate, |
2207 | const char *path, int len) | |
443e061a LT |
2208 | { |
2209 | int pos; | |
9c5e6c80 | 2210 | const struct cache_entry *ce; |
443e061a | 2211 | |
98f2a687 | 2212 | ce = index_file_exists(istate, path, len, 0); |
443e061a LT |
2213 | if (ce) { |
2214 | if (!ce_uptodate(ce)) | |
2215 | return DT_UNKNOWN; | |
2216 | if (S_ISGITLINK(ce->ce_mode)) | |
2217 | return DT_DIR; | |
2218 | /* | |
2219 | * Nobody actually cares about the | |
2220 | * difference between DT_LNK and DT_REG | |
2221 | */ | |
2222 | return DT_REG; | |
2223 | } | |
2224 | ||
2225 | /* Try to look it up as a directory */ | |
98f2a687 | 2226 | pos = index_name_pos(istate, path, len); |
443e061a LT |
2227 | if (pos >= 0) |
2228 | return DT_UNKNOWN; | |
2229 | pos = -pos-1; | |
98f2a687 BW |
2230 | while (pos < istate->cache_nr) { |
2231 | ce = istate->cache[pos++]; | |
443e061a LT |
2232 | if (strncmp(ce->name, path, len)) |
2233 | break; | |
2234 | if (ce->name[len] > '/') | |
2235 | break; | |
2236 | if (ce->name[len] < '/') | |
2237 | continue; | |
2238 | if (!ce_uptodate(ce)) | |
2239 | break; /* continue? */ | |
2240 | return DT_DIR; | |
2241 | } | |
2242 | return DT_UNKNOWN; | |
2243 | } | |
2244 | ||
ad6f2157 JK |
2245 | static int resolve_dtype(int dtype, struct index_state *istate, |
2246 | const char *path, int len) | |
07134421 | 2247 | { |
07134421 LT |
2248 | struct stat st; |
2249 | ||
2250 | if (dtype != DT_UNKNOWN) | |
2251 | return dtype; | |
98f2a687 | 2252 | dtype = get_index_dtype(istate, path, len); |
443e061a LT |
2253 | if (dtype != DT_UNKNOWN) |
2254 | return dtype; | |
2255 | if (lstat(path, &st)) | |
07134421 LT |
2256 | return dtype; |
2257 | if (S_ISREG(st.st_mode)) | |
2258 | return DT_REG; | |
2259 | if (S_ISDIR(st.st_mode)) | |
2260 | return DT_DIR; | |
2261 | if (S_ISLNK(st.st_mode)) | |
2262 | return DT_LNK; | |
2263 | return dtype; | |
2264 | } | |
2265 | ||
91a2288b | 2266 | static enum path_treatment treat_path_fast(struct dir_struct *dir, |
91a2288b | 2267 | struct cached_dir *cdir, |
0ef8e169 | 2268 | struct index_state *istate, |
91a2288b NTND |
2269 | struct strbuf *path, |
2270 | int baselen, | |
e1b8c7bd | 2271 | const struct pathspec *pathspec) |
91a2288b | 2272 | { |
8d92fb29 EN |
2273 | /* |
2274 | * WARNING: From this function, you can return path_recurse or you | |
2275 | * can call read_directory_recursive() (or neither), but | |
2276 | * you CAN'T DO BOTH. | |
2277 | */ | |
91a2288b NTND |
2278 | strbuf_setlen(path, baselen); |
2279 | if (!cdir->ucd) { | |
2280 | strbuf_addstr(path, cdir->file); | |
2281 | return path_untracked; | |
2282 | } | |
2283 | strbuf_addstr(path, cdir->ucd->name); | |
2284 | /* treat_one_path() does this before it calls treat_directory() */ | |
00b6c178 | 2285 | strbuf_complete(path, '/'); |
91a2288b NTND |
2286 | if (cdir->ucd->check_only) |
2287 | /* | |
2288 | * check_only is set as a result of treat_directory() getting | |
2289 | * to its bottom. Verify again the same set of directories | |
2290 | * with check_only set. | |
2291 | */ | |
0ef8e169 | 2292 | return read_directory_recursive(dir, istate, path->buf, path->len, |
5aaa7fd3 | 2293 | cdir->ucd, 1, 0, pathspec); |
91a2288b NTND |
2294 | /* |
2295 | * We get path_recurse in the first run when | |
2296 | * directory_exists_in_index() returns index_nonexistent. We | |
2297 | * are sure that new changes in the index does not impact the | |
2298 | * outcome. Return now. | |
2299 | */ | |
2300 | return path_recurse; | |
2301 | } | |
2302 | ||
16e2cfa9 | 2303 | static enum path_treatment treat_path(struct dir_struct *dir, |
0dcb8d7f | 2304 | struct untracked_cache_dir *untracked, |
cf7c6148 | 2305 | struct cached_dir *cdir, |
0ef8e169 | 2306 | struct index_state *istate, |
49dc2cc2 | 2307 | struct strbuf *path, |
16e2cfa9 | 2308 | int baselen, |
e1b8c7bd | 2309 | const struct pathspec *pathspec) |
16e2cfa9 | 2310 | { |
2df179d3 | 2311 | int has_path_in_index, dtype, excluded; |
2eac2a4c | 2312 | |
ad6f2157 | 2313 | if (!cdir->d_name) |
842385b8 | 2314 | return treat_path_fast(dir, cdir, istate, path, |
e1b8c7bd | 2315 | baselen, pathspec); |
ad6f2157 | 2316 | if (is_dot_or_dotdot(cdir->d_name) || !fspathcmp(cdir->d_name, ".git")) |
defd7c7b | 2317 | return path_none; |
49dc2cc2 | 2318 | strbuf_setlen(path, baselen); |
ad6f2157 | 2319 | strbuf_addstr(path, cdir->d_name); |
e1b8c7bd | 2320 | if (simplify_away(path->buf, path->len, pathspec)) |
defd7c7b | 2321 | return path_none; |
16e2cfa9 | 2322 | |
cd129eed | 2323 | dtype = resolve_dtype(cdir->d_type, istate, path->buf, path->len); |
8aaf8d77 KB |
2324 | |
2325 | /* Always exclude indexed files */ | |
cd129eed EN |
2326 | has_path_in_index = !!index_file_exists(istate, path->buf, path->len, |
2327 | ignore_case); | |
2eac2a4c | 2328 | if (dtype != DT_DIR && has_path_in_index) |
defd7c7b | 2329 | return path_none; |
8aaf8d77 | 2330 | |
2eac2a4c JH |
2331 | /* |
2332 | * When we are looking at a directory P in the working tree, | |
2333 | * there are three cases: | |
2334 | * | |
2335 | * (1) P exists in the index. Everything inside the directory P in | |
2336 | * the working tree needs to go when P is checked out from the | |
2337 | * index. | |
2338 | * | |
2339 | * (2) P does not exist in the index, but there is P/Q in the index. | |
2340 | * We know P will stay a directory when we check out the contents | |
2341 | * of the index, but we do not know yet if there is a directory | |
2342 | * P/Q in the working tree to be killed, so we need to recurse. | |
2343 | * | |
2344 | * (3) P does not exist in the index, and there is no P/Q in the index | |
2345 | * to require P to be a directory, either. Only in this case, we | |
2346 | * know that everything inside P will not be killed without | |
2347 | * recursing. | |
2348 | */ | |
2349 | if ((dir->flags & DIR_COLLECT_KILLED_ONLY) && | |
2350 | (dtype == DT_DIR) && | |
de372b1b | 2351 | !has_path_in_index && |
0ef8e169 | 2352 | (directory_exists_in_index(istate, path->buf, path->len) == index_nonexistent)) |
de372b1b | 2353 | return path_none; |
8aaf8d77 | 2354 | |
2df179d3 | 2355 | excluded = is_excluded(dir, istate, path->buf, &dtype); |
53cc5356 JH |
2356 | |
2357 | /* | |
2358 | * Excluded? If we don't explicitly want to show | |
2359 | * ignored files, ignore it | |
2360 | */ | |
2df179d3 | 2361 | if (excluded && !(dir->flags & (DIR_SHOW_IGNORED|DIR_SHOW_IGNORED_TOO))) |
defd7c7b | 2362 | return path_excluded; |
53cc5356 | 2363 | |
53cc5356 JH |
2364 | switch (dtype) { |
2365 | default: | |
defd7c7b | 2366 | return path_none; |
53cc5356 | 2367 | case DT_DIR: |
07966ed1 | 2368 | /* |
8d92fb29 EN |
2369 | * WARNING: Do not ignore/amend the return value from |
2370 | * treat_directory(), and especially do not change it to return | |
2371 | * path_recurse as that can cause exponential slowdown. | |
2372 | * Instead, modify treat_directory() to return the right value. | |
07966ed1 | 2373 | */ |
8d92fb29 EN |
2374 | strbuf_addch(path, '/'); |
2375 | return treat_directory(dir, istate, untracked, | |
2376 | path->buf, path->len, | |
2377 | baselen, excluded, pathspec); | |
53cc5356 JH |
2378 | case DT_REG: |
2379 | case DT_LNK: | |
95c11ecc | 2380 | if (pathspec && |
f1f061e1 EN |
2381 | !match_pathspec(istate, pathspec, path->buf, path->len, |
2382 | 0 /* prefix */, NULL /* seen */, | |
2383 | 0 /* is_dir */)) | |
95c11ecc | 2384 | return path_none; |
cada7308 MÅ |
2385 | if (excluded) |
2386 | return path_excluded; | |
95c11ecc | 2387 | return path_untracked; |
53cc5356 | 2388 | } |
0dcb8d7f NTND |
2389 | } |
2390 | ||
2391 | static void add_untracked(struct untracked_cache_dir *dir, const char *name) | |
2392 | { | |
2393 | if (!dir) | |
2394 | return; | |
2395 | ALLOC_GROW(dir->untracked, dir->untracked_nr + 1, | |
2396 | dir->untracked_alloc); | |
2397 | dir->untracked[dir->untracked_nr++] = xstrdup(name); | |
16e2cfa9 JH |
2398 | } |
2399 | ||
91a2288b NTND |
2400 | static int valid_cached_dir(struct dir_struct *dir, |
2401 | struct untracked_cache_dir *untracked, | |
207a06ce | 2402 | struct index_state *istate, |
91a2288b NTND |
2403 | struct strbuf *path, |
2404 | int check_only) | |
2405 | { | |
2406 | struct stat st; | |
2407 | ||
2408 | if (!untracked) | |
2409 | return 0; | |
2410 | ||
883e248b BP |
2411 | /* |
2412 | * With fsmonitor, we can trust the untracked cache's valid field. | |
2413 | */ | |
2414 | refresh_fsmonitor(istate); | |
2415 | if (!(dir->untracked->use_fsmonitor && untracked->valid)) { | |
2523c4be | 2416 | if (lstat(path->len ? path->buf : ".", &st)) { |
883e248b BP |
2417 | memset(&untracked->stat_data, 0, sizeof(untracked->stat_data)); |
2418 | return 0; | |
2419 | } | |
2420 | if (!untracked->valid || | |
2421 | match_stat_data_racy(istate, &untracked->stat_data, &st)) { | |
883e248b BP |
2422 | fill_stat_data(&untracked->stat_data, &st); |
2423 | return 0; | |
2424 | } | |
91a2288b NTND |
2425 | } |
2426 | ||
b6403131 | 2427 | if (untracked->check_only != !!check_only) |
91a2288b | 2428 | return 0; |
91a2288b NTND |
2429 | |
2430 | /* | |
2431 | * prep_exclude will be called eventually on this directory, | |
65edd96a | 2432 | * but it's called much later in last_matching_pattern(). We |
91a2288b NTND |
2433 | * need it now to determine the validity of the cache for this |
2434 | * path. The next calls will be nearly no-op, the way | |
2435 | * prep_exclude() is designed. | |
2436 | */ | |
2437 | if (path->len && path->buf[path->len - 1] != '/') { | |
2438 | strbuf_addch(path, '/'); | |
207a06ce | 2439 | prep_exclude(dir, istate, path->buf, path->len); |
91a2288b NTND |
2440 | strbuf_setlen(path, path->len - 1); |
2441 | } else | |
207a06ce | 2442 | prep_exclude(dir, istate, path->buf, path->len); |
91a2288b NTND |
2443 | |
2444 | /* hopefully prep_exclude() haven't invalidated this entry... */ | |
2445 | return untracked->valid; | |
2446 | } | |
2447 | ||
cf7c6148 NTND |
2448 | static int open_cached_dir(struct cached_dir *cdir, |
2449 | struct dir_struct *dir, | |
2450 | struct untracked_cache_dir *untracked, | |
207a06ce | 2451 | struct index_state *istate, |
cf7c6148 NTND |
2452 | struct strbuf *path, |
2453 | int check_only) | |
2454 | { | |
b6731550 NTND |
2455 | const char *c_path; |
2456 | ||
cf7c6148 NTND |
2457 | memset(cdir, 0, sizeof(*cdir)); |
2458 | cdir->untracked = untracked; | |
207a06ce | 2459 | if (valid_cached_dir(dir, untracked, istate, path, check_only)) |
91a2288b | 2460 | return 0; |
b6731550 NTND |
2461 | c_path = path->len ? path->buf : "."; |
2462 | cdir->fdir = opendir(c_path); | |
2463 | if (!cdir->fdir) | |
2464 | warning_errno(_("could not open directory '%s'"), c_path); | |
b6403131 NTND |
2465 | if (dir->untracked) { |
2466 | invalidate_directory(dir->untracked, untracked); | |
91a2288b | 2467 | dir->untracked->dir_opened++; |
b6403131 | 2468 | } |
cf7c6148 NTND |
2469 | if (!cdir->fdir) |
2470 | return -1; | |
2471 | return 0; | |
2472 | } | |
2473 | ||
2474 | static int read_cached_dir(struct cached_dir *cdir) | |
2475 | { | |
ad6f2157 JK |
2476 | struct dirent *de; |
2477 | ||
cf7c6148 | 2478 | if (cdir->fdir) { |
b548f0f1 | 2479 | de = readdir_skip_dot_and_dotdot(cdir->fdir); |
ad6f2157 JK |
2480 | if (!de) { |
2481 | cdir->d_name = NULL; | |
2482 | cdir->d_type = DT_UNKNOWN; | |
cf7c6148 | 2483 | return -1; |
ad6f2157 JK |
2484 | } |
2485 | cdir->d_name = de->d_name; | |
2486 | cdir->d_type = DTYPE(de); | |
cf7c6148 NTND |
2487 | return 0; |
2488 | } | |
91a2288b NTND |
2489 | while (cdir->nr_dirs < cdir->untracked->dirs_nr) { |
2490 | struct untracked_cache_dir *d = cdir->untracked->dirs[cdir->nr_dirs]; | |
26cb0182 NTND |
2491 | if (!d->recurse) { |
2492 | cdir->nr_dirs++; | |
2493 | continue; | |
2494 | } | |
91a2288b NTND |
2495 | cdir->ucd = d; |
2496 | cdir->nr_dirs++; | |
2497 | return 0; | |
2498 | } | |
2499 | cdir->ucd = NULL; | |
2500 | if (cdir->nr_files < cdir->untracked->untracked_nr) { | |
2501 | struct untracked_cache_dir *d = cdir->untracked; | |
2502 | cdir->file = d->untracked[cdir->nr_files++]; | |
2503 | return 0; | |
2504 | } | |
cf7c6148 NTND |
2505 | return -1; |
2506 | } | |
2507 | ||
2508 | static void close_cached_dir(struct cached_dir *cdir) | |
2509 | { | |
2510 | if (cdir->fdir) | |
2511 | closedir(cdir->fdir); | |
91a2288b NTND |
2512 | /* |
2513 | * We have gone through this directory and found no untracked | |
2514 | * entries. Mark it valid. | |
2515 | */ | |
26cb0182 | 2516 | if (cdir->untracked) { |
91a2288b | 2517 | cdir->untracked->valid = 1; |
26cb0182 NTND |
2518 | cdir->untracked->recurse = 1; |
2519 | } | |
16e2cfa9 JH |
2520 | } |
2521 | ||
c5c4eddd EN |
2522 | static void add_path_to_appropriate_result_list(struct dir_struct *dir, |
2523 | struct untracked_cache_dir *untracked, | |
2524 | struct cached_dir *cdir, | |
2525 | struct index_state *istate, | |
2526 | struct strbuf *path, | |
2527 | int baselen, | |
2528 | const struct pathspec *pathspec, | |
2529 | enum path_treatment state) | |
2530 | { | |
2531 | /* add the path to the appropriate result list */ | |
2532 | switch (state) { | |
2533 | case path_excluded: | |
2534 | if (dir->flags & DIR_SHOW_IGNORED) | |
2535 | dir_add_name(dir, istate, path->buf, path->len); | |
2536 | else if ((dir->flags & DIR_SHOW_IGNORED_TOO) || | |
2537 | ((dir->flags & DIR_COLLECT_IGNORED) && | |
2538 | exclude_matches_pathspec(path->buf, path->len, | |
2539 | pathspec))) | |
2540 | dir_add_ignored(dir, istate, path->buf, path->len); | |
2541 | break; | |
2542 | ||
2543 | case path_untracked: | |
2544 | if (dir->flags & DIR_SHOW_IGNORED) | |
2545 | break; | |
2546 | dir_add_name(dir, istate, path->buf, path->len); | |
2547 | if (cdir->fdir) | |
2548 | add_untracked(untracked, path->buf + baselen); | |
2549 | break; | |
2550 | ||
2551 | default: | |
2552 | break; | |
2553 | } | |
2554 | } | |
2555 | ||
453ec4bd LT |
2556 | /* |
2557 | * Read a directory tree. We currently ignore anything but | |
2558 | * directories, regular files and symlinks. That's because git | |
2559 | * doesn't handle them at all yet. Maybe that will change some | |
2560 | * day. | |
2561 | * | |
2562 | * Also, we ignore the name ".git" (even if it is not a directory). | |
2563 | * That likely will not change. | |
defd7c7b | 2564 | * |
5aaa7fd3 JM |
2565 | * If 'stop_at_first_file' is specified, 'path_excluded' is returned |
2566 | * to signal that a file was found. This is the least significant value that | |
2567 | * indicates that a file was encountered that does not depend on the order of | |
446f46d8 | 2568 | * whether an untracked or excluded path was encountered first. |
5aaa7fd3 | 2569 | * |
defd7c7b | 2570 | * Returns the most significant path_treatment value encountered in the scan. |
5aaa7fd3 JM |
2571 | * If 'stop_at_first_file' is specified, `path_excluded` is the most |
2572 | * significant path_treatment value that will be returned. | |
453ec4bd | 2573 | */ |
5aaa7fd3 | 2574 | |
defd7c7b | 2575 | static enum path_treatment read_directory_recursive(struct dir_struct *dir, |
0ef8e169 BW |
2576 | struct index_state *istate, const char *base, int baselen, |
2577 | struct untracked_cache_dir *untracked, int check_only, | |
5aaa7fd3 | 2578 | int stop_at_first_file, const struct pathspec *pathspec) |
453ec4bd | 2579 | { |
777b4203 | 2580 | /* |
8d92fb29 EN |
2581 | * WARNING: Do NOT recurse unless path_recurse is returned from |
2582 | * treat_path(). Recursing on any other return value | |
2583 | * can result in exponential slowdown. | |
777b4203 | 2584 | */ |
cf7c6148 | 2585 | struct cached_dir cdir; |
defd7c7b | 2586 | enum path_treatment state, subdir_state, dir_state = path_none; |
bef36921 | 2587 | struct strbuf path = STRBUF_INIT; |
453ec4bd | 2588 | |
bef36921 | 2589 | strbuf_add(&path, base, baselen); |
02cb6753 | 2590 | |
0ef8e169 | 2591 | if (open_cached_dir(&cdir, dir, untracked, istate, &path, check_only)) |
1528d247 | 2592 | goto out; |
7fe1ffda | 2593 | dir->visited_directories++; |
1528d247 | 2594 | |
0dcb8d7f NTND |
2595 | if (untracked) |
2596 | untracked->check_only = !!check_only; | |
2597 | ||
cf7c6148 | 2598 | while (!read_cached_dir(&cdir)) { |
defd7c7b | 2599 | /* check how the file or directory should be treated */ |
0ef8e169 | 2600 | state = treat_path(dir, untracked, &cdir, istate, &path, |
e1b8c7bd | 2601 | baselen, pathspec); |
7fe1ffda | 2602 | dir->visited_paths++; |
0dcb8d7f | 2603 | |
defd7c7b KB |
2604 | if (state > dir_state) |
2605 | dir_state = state; | |
2606 | ||
2607 | /* recurse into subdir if instructed by treat_path */ | |
8d92fb29 | 2608 | if (state == path_recurse) { |
0dcb8d7f NTND |
2609 | struct untracked_cache_dir *ud; |
2610 | ud = lookup_untracked(dir->untracked, untracked, | |
2611 | path.buf + baselen, | |
2612 | path.len - baselen); | |
2613 | subdir_state = | |
0ef8e169 | 2614 | read_directory_recursive(dir, istate, path.buf, |
e1b8c7bd | 2615 | path.len, ud, |
5aaa7fd3 | 2616 | check_only, stop_at_first_file, pathspec); |
defd7c7b KB |
2617 | if (subdir_state > dir_state) |
2618 | dir_state = subdir_state; | |
404ebced | 2619 | |
69f272b9 EN |
2620 | if (pathspec && |
2621 | !match_pathspec(istate, pathspec, path.buf, path.len, | |
404ebced EN |
2622 | 0 /* prefix */, NULL, |
2623 | 0 /* do NOT special case dirs */)) | |
2624 | state = path_none; | |
defd7c7b KB |
2625 | } |
2626 | ||
2627 | if (check_only) { | |
5aaa7fd3 JM |
2628 | if (stop_at_first_file) { |
2629 | /* | |
2630 | * If stopping at first file, then | |
2631 | * signal that a file was found by | |
2632 | * returning `path_excluded`. This is | |
2633 | * to return a consistent value | |
2634 | * regardless of whether an ignored or | |
2635 | * excluded file happened to be | |
2636 | * encountered 1st. | |
2637 | * | |
2638 | * In current usage, the | |
2639 | * `stop_at_first_file` is passed when | |
2640 | * an ancestor directory has matched | |
2641 | * an exclude pattern, so any found | |
2642 | * files will be excluded. | |
2643 | */ | |
2644 | if (dir_state >= path_excluded) { | |
2645 | dir_state = path_excluded; | |
2646 | break; | |
2647 | } | |
2648 | } | |
2649 | ||
defd7c7b | 2650 | /* abort early if maximum state has been reached */ |
0dcb8d7f | 2651 | if (dir_state == path_untracked) { |
91a2288b | 2652 | if (cdir.fdir) |
0dcb8d7f | 2653 | add_untracked(untracked, path.buf + baselen); |
defd7c7b | 2654 | break; |
0dcb8d7f | 2655 | } |
0126d141 | 2656 | /* skip the add_path_to_appropriate_result_list() */ |
02cb6753 | 2657 | continue; |
453ec4bd | 2658 | } |
defd7c7b | 2659 | |
c5c4eddd EN |
2660 | add_path_to_appropriate_result_list(dir, untracked, &cdir, |
2661 | istate, &path, baselen, | |
2662 | pathspec, state); | |
453ec4bd | 2663 | } |
cf7c6148 | 2664 | close_cached_dir(&cdir); |
1528d247 | 2665 | out: |
bef36921 | 2666 | strbuf_release(&path); |
453ec4bd | 2667 | |
defd7c7b | 2668 | return dir_state; |
453ec4bd LT |
2669 | } |
2670 | ||
bbf504a9 | 2671 | int cmp_dir_entry(const void *p1, const void *p2) |
453ec4bd LT |
2672 | { |
2673 | const struct dir_entry *e1 = *(const struct dir_entry **)p1; | |
2674 | const struct dir_entry *e2 = *(const struct dir_entry **)p2; | |
2675 | ||
ccdd4a0f | 2676 | return name_compare(e1->name, e1->len, e2->name, e2->len); |
453ec4bd LT |
2677 | } |
2678 | ||
fb898888 | 2679 | /* check if *out lexically strictly contains *in */ |
bbf504a9 | 2680 | int check_dir_entry_contains(const struct dir_entry *out, const struct dir_entry *in) |
fb898888 SL |
2681 | { |
2682 | return (out->len < in->len) && | |
2683 | (out->name[out->len - 1] == '/') && | |
2684 | !memcmp(out->name, in->name, out->len); | |
2685 | } | |
2686 | ||
48ffef96 | 2687 | static int treat_leading_path(struct dir_struct *dir, |
0ef8e169 | 2688 | struct index_state *istate, |
48ffef96 | 2689 | const char *path, int len, |
e1b8c7bd | 2690 | const struct pathspec *pathspec) |
48ffef96 | 2691 | { |
49dc2cc2 | 2692 | struct strbuf sb = STRBUF_INIT; |
ad6f2157 | 2693 | struct strbuf subdir = STRBUF_INIT; |
b9670c1f | 2694 | int prevlen, baselen; |
48ffef96 | 2695 | const char *cp; |
b9670c1f | 2696 | struct cached_dir cdir; |
b9670c1f EN |
2697 | enum path_treatment state = path_none; |
2698 | ||
2699 | /* | |
2700 | * For each directory component of path, we are going to check whether | |
2701 | * that path is relevant given the pathspec. For example, if path is | |
2702 | * foo/bar/baz/ | |
2703 | * then we will ask treat_path() whether we should go into foo, then | |
2704 | * whether we should go into bar, then whether baz is relevant. | |
2705 | * Checking each is important because e.g. if path is | |
2706 | * .git/info/ | |
2707 | * then we need to check .git to know we shouldn't traverse it. | |
2708 | * If the return from treat_path() is: | |
2709 | * * path_none, for any path, we return false. | |
2710 | * * path_recurse, for all path components, we return true | |
2711 | * * <anything else> for some intermediate component, we make sure | |
2712 | * to add that path to the relevant list but return false | |
2713 | * signifying that we shouldn't recurse into it. | |
2714 | */ | |
48ffef96 JH |
2715 | |
2716 | while (len && path[len - 1] == '/') | |
2717 | len--; | |
2718 | if (!len) | |
2719 | return 1; | |
b9670c1f | 2720 | |
b9670c1f | 2721 | memset(&cdir, 0, sizeof(cdir)); |
ad6f2157 | 2722 | cdir.d_type = DT_DIR; |
48ffef96 | 2723 | baselen = 0; |
b9670c1f | 2724 | prevlen = 0; |
48ffef96 | 2725 | while (1) { |
b9670c1f EN |
2726 | prevlen = baselen + !!baselen; |
2727 | cp = path + prevlen; | |
48ffef96 JH |
2728 | cp = memchr(cp, '/', path + len - cp); |
2729 | if (!cp) | |
2730 | baselen = len; | |
2731 | else | |
2732 | baselen = cp - path; | |
b9670c1f | 2733 | strbuf_reset(&sb); |
49dc2cc2 RS |
2734 | strbuf_add(&sb, path, baselen); |
2735 | if (!is_directory(sb.buf)) | |
2736 | break; | |
b9670c1f EN |
2737 | strbuf_reset(&sb); |
2738 | strbuf_add(&sb, path, prevlen); | |
ad6f2157 JK |
2739 | strbuf_reset(&subdir); |
2740 | strbuf_add(&subdir, path+prevlen, baselen-prevlen); | |
2741 | cdir.d_name = subdir.buf; | |
8d92fb29 | 2742 | state = treat_path(dir, NULL, &cdir, istate, &sb, prevlen, pathspec); |
777b4203 | 2743 | |
b9670c1f | 2744 | if (state != path_recurse) |
49dc2cc2 | 2745 | break; /* do not recurse into it */ |
b9670c1f | 2746 | if (len <= baselen) |
49dc2cc2 | 2747 | break; /* finished checking */ |
48ffef96 | 2748 | } |
b9670c1f EN |
2749 | add_path_to_appropriate_result_list(dir, NULL, &cdir, istate, |
2750 | &sb, baselen, pathspec, | |
2751 | state); | |
2752 | ||
ad6f2157 | 2753 | strbuf_release(&subdir); |
49dc2cc2 | 2754 | strbuf_release(&sb); |
b9670c1f | 2755 | return state == path_recurse; |
48ffef96 JH |
2756 | } |
2757 | ||
1e8fef60 NTND |
2758 | static const char *get_ident_string(void) |
2759 | { | |
2760 | static struct strbuf sb = STRBUF_INIT; | |
2761 | struct utsname uts; | |
2762 | ||
2763 | if (sb.len) | |
2764 | return sb.buf; | |
100e4337 | 2765 | if (uname(&uts) < 0) |
1e8fef60 | 2766 | die_errno(_("failed to get kernel name and information")); |
0e0f7618 CC |
2767 | strbuf_addf(&sb, "Location %s, system %s", get_git_work_tree(), |
2768 | uts.sysname); | |
1e8fef60 NTND |
2769 | return sb.buf; |
2770 | } | |
2771 | ||
2772 | static int ident_in_untracked(const struct untracked_cache *uc) | |
2773 | { | |
0e0f7618 CC |
2774 | /* |
2775 | * Previous git versions may have saved many NUL separated | |
2776 | * strings in the "ident" field, but it is insane to manage | |
2777 | * many locations, so just take care of the first one. | |
2778 | */ | |
1e8fef60 | 2779 | |
0e0f7618 | 2780 | return !strcmp(uc->ident.buf, get_ident_string()); |
1e8fef60 NTND |
2781 | } |
2782 | ||
0e0f7618 | 2783 | static void set_untracked_ident(struct untracked_cache *uc) |
1e8fef60 | 2784 | { |
0e0f7618 | 2785 | strbuf_reset(&uc->ident); |
1e8fef60 | 2786 | strbuf_addstr(&uc->ident, get_ident_string()); |
0e0f7618 CC |
2787 | |
2788 | /* | |
2789 | * This strbuf used to contain a list of NUL separated | |
2790 | * strings, so save NUL too for backward compatibility. | |
2791 | */ | |
1e8fef60 NTND |
2792 | strbuf_addch(&uc->ident, 0); |
2793 | } | |
2794 | ||
4a4ca479 CC |
2795 | static void new_untracked_cache(struct index_state *istate) |
2796 | { | |
2797 | struct untracked_cache *uc = xcalloc(1, sizeof(*uc)); | |
2798 | strbuf_init(&uc->ident, 100); | |
2799 | uc->exclude_per_dir = ".gitignore"; | |
2800 | /* should be the same flags used by git-status */ | |
2801 | uc->dir_flags = DIR_SHOW_OTHER_DIRECTORIES | DIR_HIDE_EMPTY_DIRECTORIES; | |
0e0f7618 | 2802 | set_untracked_ident(uc); |
4a4ca479 | 2803 | istate->untracked = uc; |
0e0f7618 | 2804 | istate->cache_changed |= UNTRACKED_CHANGED; |
4a4ca479 CC |
2805 | } |
2806 | ||
2807 | void add_untracked_cache(struct index_state *istate) | |
2808 | { | |
2809 | if (!istate->untracked) { | |
2810 | new_untracked_cache(istate); | |
0e0f7618 CC |
2811 | } else { |
2812 | if (!ident_in_untracked(istate->untracked)) { | |
2813 | free_untracked_cache(istate->untracked); | |
2814 | new_untracked_cache(istate); | |
2815 | } | |
2816 | } | |
4a4ca479 CC |
2817 | } |
2818 | ||
07b29bfd CC |
2819 | void remove_untracked_cache(struct index_state *istate) |
2820 | { | |
2821 | if (istate->untracked) { | |
2822 | free_untracked_cache(istate->untracked); | |
2823 | istate->untracked = NULL; | |
2824 | istate->cache_changed |= UNTRACKED_CHANGED; | |
2825 | } | |
2826 | } | |
2827 | ||
ccad261f NTND |
2828 | static struct untracked_cache_dir *validate_untracked_cache(struct dir_struct *dir, |
2829 | int base_len, | |
2830 | const struct pathspec *pathspec) | |
2831 | { | |
2832 | struct untracked_cache_dir *root; | |
026336cb | 2833 | static int untracked_cache_disabled = -1; |
ccad261f | 2834 | |
026336cb JH |
2835 | if (!dir->untracked) |
2836 | return NULL; | |
2837 | if (untracked_cache_disabled < 0) | |
2838 | untracked_cache_disabled = git_env_bool("GIT_DISABLE_UNTRACKED_CACHE", 0); | |
2839 | if (untracked_cache_disabled) | |
ccad261f NTND |
2840 | return NULL; |
2841 | ||
2842 | /* | |
2843 | * We only support $GIT_DIR/info/exclude and core.excludesfile | |
2844 | * as the global ignore rule files. Any other additions | |
2845 | * (e.g. from command line) invalidate the cache. This | |
2846 | * condition also catches running setup_standard_excludes() | |
2847 | * before setting dir->untracked! | |
2848 | */ | |
2849 | if (dir->unmanaged_exclude_files) | |
2850 | return NULL; | |
2851 | ||
2852 | /* | |
2853 | * Optimize for the main use case only: whole-tree git | |
2854 | * status. More work involved in treat_leading_path() if we | |
2855 | * use cache on just a subset of the worktree. pathspec | |
2856 | * support could make the matter even worse. | |
2857 | */ | |
2858 | if (base_len || (pathspec && pathspec->nr)) | |
2859 | return NULL; | |
2860 | ||
2861 | /* Different set of flags may produce different results */ | |
2862 | if (dir->flags != dir->untracked->dir_flags || | |
2863 | /* | |
2864 | * See treat_directory(), case index_nonexistent. Without | |
2865 | * this flag, we may need to also cache .git file content | |
2866 | * for the resolve_gitlink_ref() call, which we don't. | |
2867 | */ | |
2868 | !(dir->flags & DIR_SHOW_OTHER_DIRECTORIES) || | |
2869 | /* We don't support collecting ignore files */ | |
2870 | (dir->flags & (DIR_SHOW_IGNORED | DIR_SHOW_IGNORED_TOO | | |
2871 | DIR_COLLECT_IGNORED))) | |
2872 | return NULL; | |
2873 | ||
2874 | /* | |
2875 | * If we use .gitignore in the cache and now you change it to | |
2876 | * .gitexclude, everything will go wrong. | |
2877 | */ | |
2878 | if (dir->exclude_per_dir != dir->untracked->exclude_per_dir && | |
2879 | strcmp(dir->exclude_per_dir, dir->untracked->exclude_per_dir)) | |
2880 | return NULL; | |
2881 | ||
2882 | /* | |
2883 | * EXC_CMDL is not considered in the cache. If people set it, | |
2884 | * skip the cache. | |
2885 | */ | |
2886 | if (dir->exclude_list_group[EXC_CMDL].nr) | |
2887 | return NULL; | |
2888 | ||
1e8fef60 | 2889 | if (!ident_in_untracked(dir->untracked)) { |
1a07e59c | 2890 | warning(_("untracked cache is disabled on this system or location")); |
1e8fef60 NTND |
2891 | return NULL; |
2892 | } | |
2893 | ||
6347d649 JH |
2894 | if (!dir->untracked->root) |
2895 | FLEX_ALLOC_STR(dir->untracked->root, name, ""); | |
ccad261f NTND |
2896 | |
2897 | /* Validate $GIT_DIR/info/exclude and core.excludesfile */ | |
2898 | root = dir->untracked->root; | |
9001dc2a | 2899 | if (!oideq(&dir->ss_info_exclude.oid, |
4b33e602 | 2900 | &dir->untracked->ss_info_exclude.oid)) { |
ccad261f NTND |
2901 | invalidate_gitignore(dir->untracked, root); |
2902 | dir->untracked->ss_info_exclude = dir->ss_info_exclude; | |
2903 | } | |
9001dc2a | 2904 | if (!oideq(&dir->ss_excludes_file.oid, |
4b33e602 | 2905 | &dir->untracked->ss_excludes_file.oid)) { |
ccad261f NTND |
2906 | invalidate_gitignore(dir->untracked, root); |
2907 | dir->untracked->ss_excludes_file = dir->ss_excludes_file; | |
2908 | } | |
26cb0182 NTND |
2909 | |
2910 | /* Make sure this directory is not dropped out at saving phase */ | |
2911 | root->recurse = 1; | |
ccad261f NTND |
2912 | return root; |
2913 | } | |
2914 | ||
7f9dd879 EN |
2915 | static void emit_traversal_statistics(struct dir_struct *dir, |
2916 | struct repository *repo, | |
2917 | const char *path, | |
2918 | int path_len) | |
2919 | { | |
2920 | if (!trace2_is_enabled()) | |
2921 | return; | |
2922 | ||
2923 | if (!path_len) { | |
2924 | trace2_data_string("read_directory", repo, "path", ""); | |
2925 | } else { | |
2926 | struct strbuf tmp = STRBUF_INIT; | |
2927 | strbuf_add(&tmp, path, path_len); | |
2928 | trace2_data_string("read_directory", repo, "path", tmp.buf); | |
2929 | strbuf_release(&tmp); | |
2930 | } | |
2931 | ||
7fe1ffda EN |
2932 | trace2_data_intmax("read_directory", repo, |
2933 | "directories-visited", dir->visited_directories); | |
2934 | trace2_data_intmax("read_directory", repo, | |
2935 | "paths-visited", dir->visited_paths); | |
2936 | ||
7f9dd879 EN |
2937 | if (!dir->untracked) |
2938 | return; | |
2939 | trace2_data_intmax("read_directory", repo, | |
2940 | "node-creation", dir->untracked->dir_created); | |
2941 | trace2_data_intmax("read_directory", repo, | |
2942 | "gitignore-invalidation", | |
2943 | dir->untracked->gitignore_invalidated); | |
2944 | trace2_data_intmax("read_directory", repo, | |
2945 | "directory-invalidation", | |
2946 | dir->untracked->dir_invalidated); | |
2947 | trace2_data_intmax("read_directory", repo, | |
2948 | "opendir", dir->untracked->dir_opened); | |
2949 | } | |
2950 | ||
2c1eb104 BW |
2951 | int read_directory(struct dir_struct *dir, struct index_state *istate, |
2952 | const char *path, int len, const struct pathspec *pathspec) | |
9fc42d60 | 2953 | { |
ccad261f | 2954 | struct untracked_cache_dir *untracked; |
b4189aa8 | 2955 | |
7f9dd879 | 2956 | trace2_region_enter("dir", "read_directory", istate->repo); |
7fe1ffda EN |
2957 | dir->visited_paths = 0; |
2958 | dir->visited_directories = 0; | |
c46c406a NTND |
2959 | |
2960 | if (has_symlink_leading_path(path, len)) { | |
7f9dd879 | 2961 | trace2_region_leave("dir", "read_directory", istate->repo); |
725b0605 | 2962 | return dir->nr; |
c46c406a | 2963 | } |
725b0605 | 2964 | |
ccad261f NTND |
2965 | untracked = validate_untracked_cache(dir, len, pathspec); |
2966 | if (!untracked) | |
2967 | /* | |
2968 | * make sure untracked cache code path is disabled, | |
2969 | * e.g. prep_exclude() | |
2970 | */ | |
2971 | dir->untracked = NULL; | |
2c1eb104 | 2972 | if (!len || treat_leading_path(dir, istate, path, len, pathspec)) |
5aaa7fd3 | 2973 | read_directory_recursive(dir, istate, path, len, untracked, 0, 0, pathspec); |
bbf504a9 SL |
2974 | QSORT(dir->entries, dir->nr, cmp_dir_entry); |
2975 | QSORT(dir->ignored, dir->ignored_nr, cmp_dir_entry); | |
fb898888 | 2976 | |
7f9dd879 EN |
2977 | emit_traversal_statistics(dir, istate->repo, path, len); |
2978 | ||
2979 | trace2_region_leave("dir", "read_directory", istate->repo); | |
c9ccb5d3 | 2980 | if (dir->untracked) { |
026336cb | 2981 | static int force_untracked_cache = -1; |
026336cb JH |
2982 | |
2983 | if (force_untracked_cache < 0) | |
2984 | force_untracked_cache = | |
2985 | git_env_bool("GIT_FORCE_UNTRACKED_CACHE", 0); | |
026336cb | 2986 | if (force_untracked_cache && |
fc9ecbeb | 2987 | dir->untracked == istate->untracked && |
1bbb3dba NTND |
2988 | (dir->untracked->dir_opened || |
2989 | dir->untracked->gitignore_invalidated || | |
2990 | dir->untracked->dir_invalidated)) | |
2c1eb104 BW |
2991 | istate->cache_changed |= UNTRACKED_CHANGED; |
2992 | if (dir->untracked != istate->untracked) { | |
6a83d902 | 2993 | FREE_AND_NULL(dir->untracked); |
1bbb3dba | 2994 | } |
c9ccb5d3 | 2995 | } |
7f9dd879 | 2996 | |
453ec4bd LT |
2997 | return dir->nr; |
2998 | } | |
c91f0d92 | 2999 | |
686a4a06 | 3000 | int file_exists(const char *f) |
c91f0d92 | 3001 | { |
686a4a06 | 3002 | struct stat sb; |
a50f9fc5 | 3003 | return lstat(f, &sb) == 0; |
c91f0d92 | 3004 | } |
e6636747 | 3005 | |
0488481e NTND |
3006 | int repo_file_exists(struct repository *repo, const char *path) |
3007 | { | |
3008 | if (repo != the_repository) | |
3009 | BUG("do not know how to check file existence in arbitrary repo"); | |
3010 | ||
3011 | return file_exists(path); | |
3012 | } | |
3013 | ||
63ec5e1f JS |
3014 | static int cmp_icase(char a, char b) |
3015 | { | |
3016 | if (a == b) | |
3017 | return 0; | |
3018 | if (ignore_case) | |
3019 | return toupper(a) - toupper(b); | |
3020 | return a - b; | |
3021 | } | |
3022 | ||
e6636747 | 3023 | /* |
9b125da4 NTND |
3024 | * Given two normalized paths (a trailing slash is ok), if subdir is |
3025 | * outside dir, return -1. Otherwise return the offset in subdir that | |
3026 | * can be used as relative path to dir. | |
e6636747 | 3027 | */ |
9b125da4 | 3028 | int dir_inside_of(const char *subdir, const char *dir) |
e6636747 | 3029 | { |
9b125da4 | 3030 | int offset = 0; |
e6636747 | 3031 | |
9b125da4 | 3032 | assert(dir && subdir && *dir && *subdir); |
e6636747 | 3033 | |
63ec5e1f | 3034 | while (*dir && *subdir && !cmp_icase(*dir, *subdir)) { |
e6636747 | 3035 | dir++; |
9b125da4 NTND |
3036 | subdir++; |
3037 | offset++; | |
490544b1 | 3038 | } |
9b125da4 NTND |
3039 | |
3040 | /* hel[p]/me vs hel[l]/yeah */ | |
3041 | if (*dir && *subdir) | |
3042 | return -1; | |
3043 | ||
3044 | if (!*subdir) | |
3045 | return !*dir ? offset : -1; /* same dir */ | |
3046 | ||
3047 | /* foo/[b]ar vs foo/[] */ | |
3048 | if (is_dir_sep(dir[-1])) | |
3049 | return is_dir_sep(subdir[-1]) ? offset : -1; | |
3050 | ||
3051 | /* foo[/]bar vs foo[] */ | |
3052 | return is_dir_sep(*subdir) ? offset + 1 : -1; | |
e6636747 JS |
3053 | } |
3054 | ||
3055 | int is_inside_dir(const char *dir) | |
3056 | { | |
56b9f6e7 RS |
3057 | char *cwd; |
3058 | int rc; | |
3059 | ||
b892913d NTND |
3060 | if (!dir) |
3061 | return 0; | |
56b9f6e7 RS |
3062 | |
3063 | cwd = xgetcwd(); | |
3064 | rc = (dir_inside_of(cwd, dir) >= 0); | |
3065 | free(cwd); | |
3066 | return rc; | |
e6636747 | 3067 | } |
7155b727 | 3068 | |
55892d23 AP |
3069 | int is_empty_dir(const char *path) |
3070 | { | |
3071 | DIR *dir = opendir(path); | |
3072 | struct dirent *e; | |
3073 | int ret = 1; | |
3074 | ||
3075 | if (!dir) | |
3076 | return 0; | |
3077 | ||
b548f0f1 EN |
3078 | e = readdir_skip_dot_and_dotdot(dir); |
3079 | if (e) | |
3080 | ret = 0; | |
55892d23 AP |
3081 | |
3082 | closedir(dir); | |
3083 | return ret; | |
3084 | } | |
3085 | ||
ed86301f AR |
3086 | char *git_url_basename(const char *repo, int is_bundle, int is_bare) |
3087 | { | |
3088 | const char *end = repo + strlen(repo), *start, *ptr; | |
3089 | size_t len; | |
3090 | char *dir; | |
3091 | ||
3092 | /* | |
3093 | * Skip scheme. | |
3094 | */ | |
3095 | start = strstr(repo, "://"); | |
3096 | if (start == NULL) | |
3097 | start = repo; | |
3098 | else | |
3099 | start += 3; | |
3100 | ||
3101 | /* | |
3102 | * Skip authentication data. The stripping does happen | |
3103 | * greedily, such that we strip up to the last '@' inside | |
3104 | * the host part. | |
3105 | */ | |
3106 | for (ptr = start; ptr < end && !is_dir_sep(*ptr); ptr++) { | |
3107 | if (*ptr == '@') | |
3108 | start = ptr + 1; | |
3109 | } | |
3110 | ||
3111 | /* | |
3112 | * Strip trailing spaces, slashes and /.git | |
3113 | */ | |
3114 | while (start < end && (is_dir_sep(end[-1]) || isspace(end[-1]))) | |
3115 | end--; | |
3116 | if (end - start > 5 && is_dir_sep(end[-5]) && | |
3117 | !strncmp(end - 4, ".git", 4)) { | |
3118 | end -= 5; | |
3119 | while (start < end && is_dir_sep(end[-1])) | |
3120 | end--; | |
3121 | } | |
3122 | ||
3123 | /* | |
3124 | * Strip trailing port number if we've got only a | |
3125 | * hostname (that is, there is no dir separator but a | |
3126 | * colon). This check is required such that we do not | |
3127 | * strip URI's like '/foo/bar:2222.git', which should | |
3128 | * result in a dir '2222' being guessed due to backwards | |
3129 | * compatibility. | |
3130 | */ | |
3131 | if (memchr(start, '/', end - start) == NULL | |
3132 | && memchr(start, ':', end - start) != NULL) { | |
3133 | ptr = end; | |
3134 | while (start < ptr && isdigit(ptr[-1]) && ptr[-1] != ':') | |
3135 | ptr--; | |
3136 | if (start < ptr && ptr[-1] == ':') | |
3137 | end = ptr - 1; | |
3138 | } | |
3139 | ||
3140 | /* | |
3141 | * Find last component. To remain backwards compatible we | |
3142 | * also regard colons as path separators, such that | |
3143 | * cloning a repository 'foo:bar.git' would result in a | |
3144 | * directory 'bar' being guessed. | |
3145 | */ | |
3146 | ptr = end; | |
3147 | while (start < ptr && !is_dir_sep(ptr[-1]) && ptr[-1] != ':') | |
3148 | ptr--; | |
3149 | start = ptr; | |
3150 | ||
3151 | /* | |
3152 | * Strip .{bundle,git}. | |
3153 | */ | |
3154 | len = end - start; | |
3155 | strip_suffix_mem(start, &len, is_bundle ? ".bundle" : ".git"); | |
3156 | ||
3157 | if (!len || (len == 1 && *start == '/')) | |
3158 | die(_("No directory name could be guessed.\n" | |
3159 | "Please specify a directory on the command line")); | |
3160 | ||
3161 | if (is_bare) | |
3162 | dir = xstrfmt("%.*s.git", (int)len, start); | |
3163 | else | |
3164 | dir = xstrndup(start, len); | |
3165 | /* | |
3166 | * Replace sequences of 'control' characters and whitespace | |
3167 | * with one ascii space, remove leading and trailing spaces. | |
3168 | */ | |
3169 | if (*dir) { | |
3170 | char *out = dir; | |
3171 | int prev_space = 1 /* strip leading whitespace */; | |
3172 | for (end = dir; *end; ++end) { | |
3173 | char ch = *end; | |
3174 | if ((unsigned char)ch < '\x20') | |
3175 | ch = '\x20'; | |
3176 | if (isspace(ch)) { | |
3177 | if (prev_space) | |
3178 | continue; | |
3179 | prev_space = 1; | |
3180 | } else | |
3181 | prev_space = 0; | |
3182 | *out++ = ch; | |
3183 | } | |
3184 | *out = '\0'; | |
3185 | if (out > dir && prev_space) | |
3186 | out[-1] = '\0'; | |
3187 | } | |
3188 | return dir; | |
3189 | } | |
3190 | ||
3191 | void strip_dir_trailing_slashes(char *dir) | |
3192 | { | |
3193 | char *end = dir + strlen(dir); | |
3194 | ||
3195 | while (dir < end - 1 && is_dir_sep(end[-1])) | |
3196 | end--; | |
3197 | *end = '\0'; | |
3198 | } | |
3199 | ||
ae2f203e | 3200 | static int remove_dir_recurse(struct strbuf *path, int flag, int *kept_up) |
7155b727 | 3201 | { |
a0f4afbe | 3202 | DIR *dir; |
7155b727 | 3203 | struct dirent *e; |
ae2f203e | 3204 | int ret = 0, original_len = path->len, len, kept_down = 0; |
a0f4afbe | 3205 | int only_empty = (flag & REMOVE_DIR_EMPTY_ONLY); |
c844a803 | 3206 | int keep_toplevel = (flag & REMOVE_DIR_KEEP_TOPLEVEL); |
1053fe82 | 3207 | struct object_id submodule_head; |
7155b727 | 3208 | |
a0f4afbe | 3209 | if ((flag & REMOVE_DIR_KEEP_NESTED_GIT) && |
a98e6101 | 3210 | !resolve_gitlink_ref(path->buf, "HEAD", &submodule_head)) { |
a0f4afbe | 3211 | /* Do not descend and nuke a nested git work tree. */ |
ae2f203e JH |
3212 | if (kept_up) |
3213 | *kept_up = 1; | |
a0f4afbe | 3214 | return 0; |
ae2f203e | 3215 | } |
a0f4afbe | 3216 | |
ae2f203e | 3217 | flag &= ~REMOVE_DIR_KEEP_TOPLEVEL; |
a0f4afbe | 3218 | dir = opendir(path->buf); |
c844a803 | 3219 | if (!dir) { |
863808cd MH |
3220 | if (errno == ENOENT) |
3221 | return keep_toplevel ? -1 : 0; | |
3222 | else if (errno == EACCES && !keep_toplevel) | |
ecb2c282 MH |
3223 | /* |
3224 | * An empty dir could be removable even if it | |
3225 | * is unreadable: | |
3226 | */ | |
c844a803 JH |
3227 | return rmdir(path->buf); |
3228 | else | |
3229 | return -1; | |
3230 | } | |
00b6c178 | 3231 | strbuf_complete(path, '/'); |
7155b727 JS |
3232 | |
3233 | len = path->len; | |
b548f0f1 | 3234 | while ((e = readdir_skip_dot_and_dotdot(dir)) != NULL) { |
7155b727 | 3235 | struct stat st; |
7155b727 JS |
3236 | |
3237 | strbuf_setlen(path, len); | |
3238 | strbuf_addstr(path, e->d_name); | |
863808cd MH |
3239 | if (lstat(path->buf, &st)) { |
3240 | if (errno == ENOENT) | |
3241 | /* | |
3242 | * file disappeared, which is what we | |
3243 | * wanted anyway | |
3244 | */ | |
3245 | continue; | |
15beaaa3 | 3246 | /* fall through */ |
863808cd | 3247 | } else if (S_ISDIR(st.st_mode)) { |
ae2f203e | 3248 | if (!remove_dir_recurse(path, flag, &kept_down)) |
7155b727 | 3249 | continue; /* happy */ |
863808cd MH |
3250 | } else if (!only_empty && |
3251 | (!unlink(path->buf) || errno == ENOENT)) { | |
7155b727 | 3252 | continue; /* happy, too */ |
863808cd | 3253 | } |
7155b727 JS |
3254 | |
3255 | /* path too long, stat fails, or non-directory still exists */ | |
3256 | ret = -1; | |
3257 | break; | |
3258 | } | |
3259 | closedir(dir); | |
3260 | ||
3261 | strbuf_setlen(path, original_len); | |
ae2f203e | 3262 | if (!ret && !keep_toplevel && !kept_down) |
863808cd | 3263 | ret = (!rmdir(path->buf) || errno == ENOENT) ? 0 : -1; |
ae2f203e JH |
3264 | else if (kept_up) |
3265 | /* | |
3266 | * report the uplevel that it is not an error that we | |
3267 | * did not rmdir() our directory. | |
3268 | */ | |
3269 | *kept_up = !ret; | |
7155b727 JS |
3270 | return ret; |
3271 | } | |
039bc64e | 3272 | |
ae2f203e JH |
3273 | int remove_dir_recursively(struct strbuf *path, int flag) |
3274 | { | |
3275 | return remove_dir_recurse(path, flag, NULL); | |
3276 | } | |
3277 | ||
f932729c JK |
3278 | static GIT_PATH_FUNC(git_path_info_exclude, "info/exclude") |
3279 | ||
039bc64e JH |
3280 | void setup_standard_excludes(struct dir_struct *dir) |
3281 | { | |
039bc64e | 3282 | dir->exclude_per_dir = ".gitignore"; |
099d2d86 | 3283 | |
51d18631 | 3284 | /* core.excludesfile defaulting to $XDG_CONFIG_HOME/git/ignore */ |
2845ce7f PT |
3285 | if (!excludes_file) |
3286 | excludes_file = xdg_config_home("ignore"); | |
4698c8fe | 3287 | if (excludes_file && !access_or_warn(excludes_file, R_OK, 0)) |
65edd96a | 3288 | add_patterns_from_file_1(dir, excludes_file, |
38ccaf93 | 3289 | dir->untracked ? &dir->ss_excludes_file : NULL); |
099d2d86 JH |
3290 | |
3291 | /* per repository user preference */ | |
f0056f64 JK |
3292 | if (startup_info->have_repository) { |
3293 | const char *path = git_path_info_exclude(); | |
3294 | if (!access_or_warn(path, R_OK, 0)) | |
65edd96a | 3295 | add_patterns_from_file_1(dir, path, |
f0056f64 JK |
3296 | dir->untracked ? &dir->ss_info_exclude : NULL); |
3297 | } | |
039bc64e | 3298 | } |
4a92d1bf | 3299 | |
dd23022a DS |
3300 | char *get_sparse_checkout_filename(void) |
3301 | { | |
3302 | return git_pathdup("info/sparse-checkout"); | |
3303 | } | |
3304 | ||
3305 | int get_sparse_checkout_patterns(struct pattern_list *pl) | |
3306 | { | |
3307 | int res; | |
3308 | char *sparse_filename = get_sparse_checkout_filename(); | |
3309 | ||
3310 | pl->use_cone_patterns = core_sparse_checkout_cone; | |
1679d60b | 3311 | res = add_patterns_from_file_to_list(sparse_filename, "", 0, pl, NULL, 0); |
dd23022a DS |
3312 | |
3313 | free(sparse_filename); | |
3314 | return res; | |
3315 | } | |
3316 | ||
4a92d1bf AR |
3317 | int remove_path(const char *name) |
3318 | { | |
3319 | char *slash; | |
3320 | ||
c7054209 | 3321 | if (unlink(name) && !is_missing_file_error(errno)) |
4a92d1bf AR |
3322 | return -1; |
3323 | ||
3324 | slash = strrchr(name, '/'); | |
3325 | if (slash) { | |
3326 | char *dirs = xstrdup(name); | |
3327 | slash = dirs + (slash - name); | |
3328 | do { | |
3329 | *slash = '\0'; | |
3fc0d131 | 3330 | } while (rmdir(dirs) == 0 && (slash = strrchr(dirs, '/'))); |
4a92d1bf AR |
3331 | free(dirs); |
3332 | } | |
3333 | return 0; | |
3334 | } | |
3335 | ||
270be816 | 3336 | /* |
dad4f23c EN |
3337 | * Frees memory within dir which was allocated, and resets fields for further |
3338 | * use. Does not free dir itself. | |
270be816 | 3339 | */ |
eceba532 | 3340 | void dir_clear(struct dir_struct *dir) |
270be816 AS |
3341 | { |
3342 | int i, j; | |
3343 | struct exclude_list_group *group; | |
caa3d554 | 3344 | struct pattern_list *pl; |
270be816 | 3345 | struct exclude_stack *stk; |
ce93a4c6 | 3346 | struct dir_struct new = DIR_INIT; |
270be816 AS |
3347 | |
3348 | for (i = EXC_CMDL; i <= EXC_FILE; i++) { | |
3349 | group = &dir->exclude_list_group[i]; | |
3350 | for (j = 0; j < group->nr; j++) { | |
caa3d554 | 3351 | pl = &group->pl[j]; |
270be816 | 3352 | if (i == EXC_DIRS) |
caa3d554 | 3353 | free((char *)pl->src); |
65edd96a | 3354 | clear_pattern_list(pl); |
270be816 | 3355 | } |
caa3d554 | 3356 | free(group->pl); |
270be816 AS |
3357 | } |
3358 | ||
dad4f23c EN |
3359 | for (i = 0; i < dir->ignored_nr; i++) |
3360 | free(dir->ignored[i]); | |
3361 | for (i = 0; i < dir->nr; i++) | |
3362 | free(dir->entries[i]); | |
3363 | free(dir->ignored); | |
3364 | free(dir->entries); | |
3365 | ||
270be816 AS |
3366 | stk = dir->exclude_stack; |
3367 | while (stk) { | |
3368 | struct exclude_stack *prev = stk->prev; | |
3369 | free(stk); | |
3370 | stk = prev; | |
3371 | } | |
aceb9429 | 3372 | strbuf_release(&dir->basebuf); |
dad4f23c | 3373 | |
ce93a4c6 | 3374 | memcpy(dir, &new, sizeof(*dir)); |
270be816 | 3375 | } |
83c094ad NTND |
3376 | |
3377 | struct ondisk_untracked_cache { | |
3378 | struct stat_data info_exclude_stat; | |
3379 | struct stat_data excludes_file_stat; | |
3380 | uint32_t dir_flags; | |
83c094ad NTND |
3381 | }; |
3382 | ||
268ba201 | 3383 | #define ouc_offset(x) offsetof(struct ondisk_untracked_cache, x) |
83c094ad NTND |
3384 | |
3385 | struct write_data { | |
3386 | int index; /* number of written untracked_cache_dir */ | |
3387 | struct ewah_bitmap *check_only; /* from untracked_cache_dir */ | |
3388 | struct ewah_bitmap *valid; /* from untracked_cache_dir */ | |
3389 | struct ewah_bitmap *sha1_valid; /* set if exclude_sha1 is not null */ | |
3390 | struct strbuf out; | |
3391 | struct strbuf sb_stat; | |
3392 | struct strbuf sb_sha1; | |
3393 | }; | |
3394 | ||
3395 | static void stat_data_to_disk(struct stat_data *to, const struct stat_data *from) | |
3396 | { | |
3397 | to->sd_ctime.sec = htonl(from->sd_ctime.sec); | |
3398 | to->sd_ctime.nsec = htonl(from->sd_ctime.nsec); | |
3399 | to->sd_mtime.sec = htonl(from->sd_mtime.sec); | |
3400 | to->sd_mtime.nsec = htonl(from->sd_mtime.nsec); | |
3401 | to->sd_dev = htonl(from->sd_dev); | |
3402 | to->sd_ino = htonl(from->sd_ino); | |
3403 | to->sd_uid = htonl(from->sd_uid); | |
3404 | to->sd_gid = htonl(from->sd_gid); | |
3405 | to->sd_size = htonl(from->sd_size); | |
3406 | } | |
3407 | ||
3408 | static void write_one_dir(struct untracked_cache_dir *untracked, | |
3409 | struct write_data *wd) | |
3410 | { | |
3411 | struct stat_data stat_data; | |
3412 | struct strbuf *out = &wd->out; | |
3413 | unsigned char intbuf[16]; | |
3414 | unsigned int intlen, value; | |
3415 | int i = wd->index++; | |
3416 | ||
3417 | /* | |
3418 | * untracked_nr should be reset whenever valid is clear, but | |
3419 | * for safety.. | |
3420 | */ | |
3421 | if (!untracked->valid) { | |
3422 | untracked->untracked_nr = 0; | |
3423 | untracked->check_only = 0; | |
3424 | } | |
3425 | ||
3426 | if (untracked->check_only) | |
3427 | ewah_set(wd->check_only, i); | |
3428 | if (untracked->valid) { | |
3429 | ewah_set(wd->valid, i); | |
3430 | stat_data_to_disk(&stat_data, &untracked->stat_data); | |
3431 | strbuf_add(&wd->sb_stat, &stat_data, sizeof(stat_data)); | |
3432 | } | |
70c369cd | 3433 | if (!is_null_oid(&untracked->exclude_oid)) { |
83c094ad | 3434 | ewah_set(wd->sha1_valid, i); |
70c369cd | 3435 | strbuf_add(&wd->sb_sha1, untracked->exclude_oid.hash, |
3436 | the_hash_algo->rawsz); | |
83c094ad NTND |
3437 | } |
3438 | ||
3439 | intlen = encode_varint(untracked->untracked_nr, intbuf); | |
3440 | strbuf_add(out, intbuf, intlen); | |
3441 | ||
3442 | /* skip non-recurse directories */ | |
3443 | for (i = 0, value = 0; i < untracked->dirs_nr; i++) | |
3444 | if (untracked->dirs[i]->recurse) | |
3445 | value++; | |
3446 | intlen = encode_varint(value, intbuf); | |
3447 | strbuf_add(out, intbuf, intlen); | |
3448 | ||
3449 | strbuf_add(out, untracked->name, strlen(untracked->name) + 1); | |
3450 | ||
3451 | for (i = 0; i < untracked->untracked_nr; i++) | |
3452 | strbuf_add(out, untracked->untracked[i], | |
3453 | strlen(untracked->untracked[i]) + 1); | |
3454 | ||
3455 | for (i = 0; i < untracked->dirs_nr; i++) | |
3456 | if (untracked->dirs[i]->recurse) | |
3457 | write_one_dir(untracked->dirs[i], wd); | |
3458 | } | |
3459 | ||
3460 | void write_untracked_extension(struct strbuf *out, struct untracked_cache *untracked) | |
3461 | { | |
3462 | struct ondisk_untracked_cache *ouc; | |
3463 | struct write_data wd; | |
3464 | unsigned char varbuf[16]; | |
e0b83735 | 3465 | int varint_len; |
3899b88b | 3466 | const unsigned hashsz = the_hash_algo->rawsz; |
e0b83735 | 3467 | |
ca56dadb | 3468 | CALLOC_ARRAY(ouc, 1); |
83c094ad NTND |
3469 | stat_data_to_disk(&ouc->info_exclude_stat, &untracked->ss_info_exclude.stat); |
3470 | stat_data_to_disk(&ouc->excludes_file_stat, &untracked->ss_excludes_file.stat); | |
83c094ad | 3471 | ouc->dir_flags = htonl(untracked->dir_flags); |
1e8fef60 NTND |
3472 | |
3473 | varint_len = encode_varint(untracked->ident.len, varbuf); | |
3474 | strbuf_add(out, varbuf, varint_len); | |
8109984d | 3475 | strbuf_addbuf(out, &untracked->ident); |
1e8fef60 | 3476 | |
3899b88b | 3477 | strbuf_add(out, ouc, sizeof(*ouc)); |
3478 | strbuf_add(out, untracked->ss_info_exclude.oid.hash, hashsz); | |
3479 | strbuf_add(out, untracked->ss_excludes_file.oid.hash, hashsz); | |
3480 | strbuf_add(out, untracked->exclude_per_dir, strlen(untracked->exclude_per_dir) + 1); | |
6a83d902 | 3481 | FREE_AND_NULL(ouc); |
83c094ad NTND |
3482 | |
3483 | if (!untracked->root) { | |
3484 | varint_len = encode_varint(0, varbuf); | |
3485 | strbuf_add(out, varbuf, varint_len); | |
3486 | return; | |
3487 | } | |
3488 | ||
3489 | wd.index = 0; | |
3490 | wd.check_only = ewah_new(); | |
3491 | wd.valid = ewah_new(); | |
3492 | wd.sha1_valid = ewah_new(); | |
3493 | strbuf_init(&wd.out, 1024); | |
3494 | strbuf_init(&wd.sb_stat, 1024); | |
3495 | strbuf_init(&wd.sb_sha1, 1024); | |
3496 | write_one_dir(untracked->root, &wd); | |
3497 | ||
3498 | varint_len = encode_varint(wd.index, varbuf); | |
3499 | strbuf_add(out, varbuf, varint_len); | |
3500 | strbuf_addbuf(out, &wd.out); | |
3501 | ewah_serialize_strbuf(wd.valid, out); | |
3502 | ewah_serialize_strbuf(wd.check_only, out); | |
3503 | ewah_serialize_strbuf(wd.sha1_valid, out); | |
3504 | strbuf_addbuf(out, &wd.sb_stat); | |
3505 | strbuf_addbuf(out, &wd.sb_sha1); | |
3506 | strbuf_addch(out, '\0'); /* safe guard for string lists */ | |
3507 | ||
3508 | ewah_free(wd.valid); | |
3509 | ewah_free(wd.check_only); | |
3510 | ewah_free(wd.sha1_valid); | |
3511 | strbuf_release(&wd.out); | |
3512 | strbuf_release(&wd.sb_stat); | |
3513 | strbuf_release(&wd.sb_sha1); | |
3514 | } | |
f9e6c649 NTND |
3515 | |
3516 | static void free_untracked(struct untracked_cache_dir *ucd) | |
3517 | { | |
3518 | int i; | |
3519 | if (!ucd) | |
3520 | return; | |
3521 | for (i = 0; i < ucd->dirs_nr; i++) | |
3522 | free_untracked(ucd->dirs[i]); | |
3523 | for (i = 0; i < ucd->untracked_nr; i++) | |
3524 | free(ucd->untracked[i]); | |
3525 | free(ucd->untracked); | |
3526 | free(ucd->dirs); | |
3527 | free(ucd); | |
3528 | } | |
3529 | ||
3530 | void free_untracked_cache(struct untracked_cache *uc) | |
3531 | { | |
3532 | if (uc) | |
3533 | free_untracked(uc->root); | |
3534 | free(uc); | |
3535 | } | |
3536 | ||
3537 | struct read_data { | |
3538 | int index; | |
3539 | struct untracked_cache_dir **ucd; | |
3540 | struct ewah_bitmap *check_only; | |
3541 | struct ewah_bitmap *valid; | |
3542 | struct ewah_bitmap *sha1_valid; | |
3543 | const unsigned char *data; | |
3544 | const unsigned char *end; | |
3545 | }; | |
3546 | ||
268ba201 | 3547 | static void stat_data_from_disk(struct stat_data *to, const unsigned char *data) |
f9e6c649 | 3548 | { |
268ba201 RS |
3549 | memcpy(to, data, sizeof(*to)); |
3550 | to->sd_ctime.sec = ntohl(to->sd_ctime.sec); | |
3551 | to->sd_ctime.nsec = ntohl(to->sd_ctime.nsec); | |
3552 | to->sd_mtime.sec = ntohl(to->sd_mtime.sec); | |
3553 | to->sd_mtime.nsec = ntohl(to->sd_mtime.nsec); | |
3554 | to->sd_dev = ntohl(to->sd_dev); | |
3555 | to->sd_ino = ntohl(to->sd_ino); | |
3556 | to->sd_uid = ntohl(to->sd_uid); | |
3557 | to->sd_gid = ntohl(to->sd_gid); | |
3558 | to->sd_size = ntohl(to->sd_size); | |
f9e6c649 NTND |
3559 | } |
3560 | ||
3561 | static int read_one_dir(struct untracked_cache_dir **untracked_, | |
3562 | struct read_data *rd) | |
3563 | { | |
3564 | struct untracked_cache_dir ud, *untracked; | |
b511d6d5 | 3565 | const unsigned char *data = rd->data, *end = rd->end; |
c6909f99 | 3566 | const unsigned char *eos; |
f9e6c649 | 3567 | unsigned int value; |
08bf354d | 3568 | int i; |
f9e6c649 NTND |
3569 | |
3570 | memset(&ud, 0, sizeof(ud)); | |
3571 | ||
b511d6d5 JK |
3572 | value = decode_varint(&data); |
3573 | if (data > end) | |
f9e6c649 NTND |
3574 | return -1; |
3575 | ud.recurse = 1; | |
3576 | ud.untracked_alloc = value; | |
3577 | ud.untracked_nr = value; | |
3578 | if (ud.untracked_nr) | |
b32fa95f | 3579 | ALLOC_ARRAY(ud.untracked, ud.untracked_nr); |
f9e6c649 | 3580 | |
b511d6d5 JK |
3581 | ud.dirs_alloc = ud.dirs_nr = decode_varint(&data); |
3582 | if (data > end) | |
f9e6c649 | 3583 | return -1; |
b32fa95f | 3584 | ALLOC_ARRAY(ud.dirs, ud.dirs_nr); |
f9e6c649 | 3585 | |
c6909f99 JK |
3586 | eos = memchr(data, '\0', end - data); |
3587 | if (!eos || eos == end) | |
f9e6c649 | 3588 | return -1; |
c6909f99 | 3589 | |
08bf354d | 3590 | *untracked_ = untracked = xmalloc(st_add3(sizeof(*untracked), eos - data, 1)); |
f9e6c649 | 3591 | memcpy(untracked, &ud, sizeof(ud)); |
08bf354d | 3592 | memcpy(untracked->name, data, eos - data + 1); |
b511d6d5 | 3593 | data = eos + 1; |
f9e6c649 NTND |
3594 | |
3595 | for (i = 0; i < untracked->untracked_nr; i++) { | |
c6909f99 JK |
3596 | eos = memchr(data, '\0', end - data); |
3597 | if (!eos || eos == end) | |
f9e6c649 | 3598 | return -1; |
08bf354d | 3599 | untracked->untracked[i] = xmemdupz(data, eos - data); |
b511d6d5 | 3600 | data = eos + 1; |
f9e6c649 NTND |
3601 | } |
3602 | ||
3603 | rd->ucd[rd->index++] = untracked; | |
3604 | rd->data = data; | |
3605 | ||
3606 | for (i = 0; i < untracked->dirs_nr; i++) { | |
08bf354d | 3607 | if (read_one_dir(untracked->dirs + i, rd) < 0) |
f9e6c649 NTND |
3608 | return -1; |
3609 | } | |
3610 | return 0; | |
3611 | } | |
3612 | ||
3613 | static void set_check_only(size_t pos, void *cb) | |
3614 | { | |
3615 | struct read_data *rd = cb; | |
3616 | struct untracked_cache_dir *ud = rd->ucd[pos]; | |
3617 | ud->check_only = 1; | |
3618 | } | |
3619 | ||
3620 | static void read_stat(size_t pos, void *cb) | |
3621 | { | |
3622 | struct read_data *rd = cb; | |
3623 | struct untracked_cache_dir *ud = rd->ucd[pos]; | |
3624 | if (rd->data + sizeof(struct stat_data) > rd->end) { | |
3625 | rd->data = rd->end + 1; | |
3626 | return; | |
3627 | } | |
268ba201 | 3628 | stat_data_from_disk(&ud->stat_data, rd->data); |
f9e6c649 NTND |
3629 | rd->data += sizeof(struct stat_data); |
3630 | ud->valid = 1; | |
3631 | } | |
3632 | ||
70c369cd | 3633 | static void read_oid(size_t pos, void *cb) |
f9e6c649 NTND |
3634 | { |
3635 | struct read_data *rd = cb; | |
3636 | struct untracked_cache_dir *ud = rd->ucd[pos]; | |
70c369cd | 3637 | if (rd->data + the_hash_algo->rawsz > rd->end) { |
f9e6c649 NTND |
3638 | rd->data = rd->end + 1; |
3639 | return; | |
3640 | } | |
92e2cab9 | 3641 | oidread(&ud->exclude_oid, rd->data); |
70c369cd | 3642 | rd->data += the_hash_algo->rawsz; |
f9e6c649 NTND |
3643 | } |
3644 | ||
4b33e602 PO |
3645 | static void load_oid_stat(struct oid_stat *oid_stat, const unsigned char *data, |
3646 | const unsigned char *sha1) | |
f9e6c649 | 3647 | { |
4b33e602 | 3648 | stat_data_from_disk(&oid_stat->stat, data); |
92e2cab9 | 3649 | oidread(&oid_stat->oid, sha1); |
4b33e602 | 3650 | oid_stat->valid = 1; |
f9e6c649 NTND |
3651 | } |
3652 | ||
3653 | struct untracked_cache *read_untracked_extension(const void *data, unsigned long sz) | |
3654 | { | |
f9e6c649 NTND |
3655 | struct untracked_cache *uc; |
3656 | struct read_data rd; | |
3657 | const unsigned char *next = data, *end = (const unsigned char *)data + sz; | |
1e8fef60 | 3658 | const char *ident; |
1140bf01 JK |
3659 | int ident_len; |
3660 | ssize_t len; | |
268ba201 | 3661 | const char *exclude_per_dir; |
3899b88b | 3662 | const unsigned hashsz = the_hash_algo->rawsz; |
3663 | const unsigned offset = sizeof(struct ondisk_untracked_cache); | |
3664 | const unsigned exclude_per_dir_offset = offset + 2 * hashsz; | |
f9e6c649 NTND |
3665 | |
3666 | if (sz <= 1 || end[-1] != '\0') | |
3667 | return NULL; | |
3668 | end--; | |
3669 | ||
1e8fef60 NTND |
3670 | ident_len = decode_varint(&next); |
3671 | if (next + ident_len > end) | |
3672 | return NULL; | |
3673 | ident = (const char *)next; | |
3674 | next += ident_len; | |
3675 | ||
3899b88b | 3676 | if (next + exclude_per_dir_offset + 1 > end) |
f9e6c649 NTND |
3677 | return NULL; |
3678 | ||
ca56dadb | 3679 | CALLOC_ARRAY(uc, 1); |
1e8fef60 NTND |
3680 | strbuf_init(&uc->ident, ident_len); |
3681 | strbuf_add(&uc->ident, ident, ident_len); | |
4b33e602 PO |
3682 | load_oid_stat(&uc->ss_info_exclude, |
3683 | next + ouc_offset(info_exclude_stat), | |
3899b88b | 3684 | next + offset); |
4b33e602 PO |
3685 | load_oid_stat(&uc->ss_excludes_file, |
3686 | next + ouc_offset(excludes_file_stat), | |
3899b88b | 3687 | next + offset + hashsz); |
268ba201 | 3688 | uc->dir_flags = get_be32(next + ouc_offset(dir_flags)); |
3899b88b | 3689 | exclude_per_dir = (const char *)next + exclude_per_dir_offset; |
268ba201 | 3690 | uc->exclude_per_dir = xstrdup(exclude_per_dir); |
f9e6c649 | 3691 | /* NUL after exclude_per_dir is covered by sizeof(*ouc) */ |
3899b88b | 3692 | next += exclude_per_dir_offset + strlen(exclude_per_dir) + 1; |
f9e6c649 NTND |
3693 | if (next >= end) |
3694 | goto done2; | |
3695 | ||
3696 | len = decode_varint(&next); | |
3697 | if (next > end || len == 0) | |
3698 | goto done2; | |
3699 | ||
3700 | rd.valid = ewah_new(); | |
3701 | rd.check_only = ewah_new(); | |
3702 | rd.sha1_valid = ewah_new(); | |
3703 | rd.data = next; | |
3704 | rd.end = end; | |
3705 | rd.index = 0; | |
b32fa95f | 3706 | ALLOC_ARRAY(rd.ucd, len); |
f9e6c649 NTND |
3707 | |
3708 | if (read_one_dir(&uc->root, &rd) || rd.index != len) | |
3709 | goto done; | |
3710 | ||
3711 | next = rd.data; | |
3712 | len = ewah_read_mmap(rd.valid, next, end - next); | |
3713 | if (len < 0) | |
3714 | goto done; | |
3715 | ||
3716 | next += len; | |
3717 | len = ewah_read_mmap(rd.check_only, next, end - next); | |
3718 | if (len < 0) | |
3719 | goto done; | |
3720 | ||
3721 | next += len; | |
3722 | len = ewah_read_mmap(rd.sha1_valid, next, end - next); | |
3723 | if (len < 0) | |
3724 | goto done; | |
3725 | ||
3726 | ewah_each_bit(rd.check_only, set_check_only, &rd); | |
3727 | rd.data = next + len; | |
3728 | ewah_each_bit(rd.valid, read_stat, &rd); | |
70c369cd | 3729 | ewah_each_bit(rd.sha1_valid, read_oid, &rd); |
f9e6c649 NTND |
3730 | next = rd.data; |
3731 | ||
3732 | done: | |
3733 | free(rd.ucd); | |
3734 | ewah_free(rd.valid); | |
3735 | ewah_free(rd.check_only); | |
3736 | ewah_free(rd.sha1_valid); | |
3737 | done2: | |
3738 | if (next != end) { | |
3739 | free_untracked_cache(uc); | |
3740 | uc = NULL; | |
3741 | } | |
3742 | return uc; | |
3743 | } | |
e931371a | 3744 | |
73f9145f NTND |
3745 | static void invalidate_one_directory(struct untracked_cache *uc, |
3746 | struct untracked_cache_dir *ucd) | |
3747 | { | |
3748 | uc->dir_invalidated++; | |
3749 | ucd->valid = 0; | |
3750 | ucd->untracked_nr = 0; | |
3751 | } | |
3752 | ||
3753 | /* | |
3754 | * Normally when an entry is added or removed from a directory, | |
3755 | * invalidating that directory is enough. No need to touch its | |
3756 | * ancestors. When a directory is shown as "foo/bar/" in git-status | |
3757 | * however, deleting or adding an entry may have cascading effect. | |
3758 | * | |
3759 | * Say the "foo/bar/file" has become untracked, we need to tell the | |
3760 | * untracked_cache_dir of "foo" that "bar/" is not an untracked | |
3761 | * directory any more (because "bar" is managed by foo as an untracked | |
3762 | * "file"). | |
3763 | * | |
3764 | * Similarly, if "foo/bar/file" moves from untracked to tracked and it | |
3765 | * was the last untracked entry in the entire "foo", we should show | |
3766 | * "foo/" instead. Which means we have to invalidate past "bar" up to | |
3767 | * "foo". | |
3768 | * | |
3769 | * This function traverses all directories from root to leaf. If there | |
3770 | * is a chance of one of the above cases happening, we invalidate back | |
3771 | * to root. Otherwise we just invalidate the leaf. There may be a more | |
3772 | * sophisticated way than checking for SHOW_OTHER_DIRECTORIES to | |
3773 | * detect these cases and avoid unnecessary invalidation, for example, | |
3774 | * checking for the untracked entry named "bar/" in "foo", but for now | |
3775 | * stick to something safe and simple. | |
3776 | */ | |
3777 | static int invalidate_one_component(struct untracked_cache *uc, | |
3778 | struct untracked_cache_dir *dir, | |
3779 | const char *path, int len) | |
3780 | { | |
3781 | const char *rest = strchr(path, '/'); | |
3782 | ||
3783 | if (rest) { | |
3784 | int component_len = rest - path; | |
3785 | struct untracked_cache_dir *d = | |
3786 | lookup_untracked(uc, dir, path, component_len); | |
3787 | int ret = | |
3788 | invalidate_one_component(uc, d, rest + 1, | |
3789 | len - (component_len + 1)); | |
3790 | if (ret) | |
3791 | invalidate_one_directory(uc, dir); | |
3792 | return ret; | |
3793 | } | |
3794 | ||
3795 | invalidate_one_directory(uc, dir); | |
3796 | return uc->dir_flags & DIR_SHOW_OTHER_DIRECTORIES; | |
3797 | } | |
3798 | ||
e931371a | 3799 | void untracked_cache_invalidate_path(struct index_state *istate, |
0cacebf0 | 3800 | const char *path, int safe_path) |
e931371a | 3801 | { |
e931371a NTND |
3802 | if (!istate->untracked || !istate->untracked->root) |
3803 | return; | |
68f95b26 | 3804 | if (!safe_path && !verify_path(path, 0)) |
0cacebf0 | 3805 | return; |
73f9145f NTND |
3806 | invalidate_one_component(istate->untracked, istate->untracked->root, |
3807 | path, strlen(path)); | |
e931371a NTND |
3808 | } |
3809 | ||
3810 | void untracked_cache_remove_from_index(struct index_state *istate, | |
3811 | const char *path) | |
3812 | { | |
0cacebf0 | 3813 | untracked_cache_invalidate_path(istate, path, 1); |
e931371a NTND |
3814 | } |
3815 | ||
3816 | void untracked_cache_add_to_index(struct index_state *istate, | |
3817 | const char *path) | |
3818 | { | |
0cacebf0 | 3819 | untracked_cache_invalidate_path(istate, path, 1); |
e931371a | 3820 | } |
47e83eb3 | 3821 | |
da62f786 SB |
3822 | static void connect_wt_gitdir_in_nested(const char *sub_worktree, |
3823 | const char *sub_gitdir) | |
3824 | { | |
3825 | int i; | |
3826 | struct repository subrepo; | |
3827 | struct strbuf sub_wt = STRBUF_INIT; | |
3828 | struct strbuf sub_gd = STRBUF_INIT; | |
3829 | ||
3830 | const struct submodule *sub; | |
3831 | ||
3832 | /* If the submodule has no working tree, we can ignore it. */ | |
3833 | if (repo_init(&subrepo, sub_gitdir, sub_worktree)) | |
3834 | return; | |
3835 | ||
3836 | if (repo_read_index(&subrepo) < 0) | |
a80897c1 | 3837 | die(_("index file corrupt in repo %s"), subrepo.gitdir); |
da62f786 | 3838 | |
d425f651 DS |
3839 | /* TODO: audit for interaction with sparse-index. */ |
3840 | ensure_full_index(subrepo.index); | |
da62f786 SB |
3841 | for (i = 0; i < subrepo.index->cache_nr; i++) { |
3842 | const struct cache_entry *ce = subrepo.index->cache[i]; | |
3843 | ||
3844 | if (!S_ISGITLINK(ce->ce_mode)) | |
3845 | continue; | |
3846 | ||
3847 | while (i + 1 < subrepo.index->cache_nr && | |
3848 | !strcmp(ce->name, subrepo.index->cache[i + 1]->name)) | |
3849 | /* | |
3850 | * Skip entries with the same name in different stages | |
3851 | * to make sure an entry is returned only once. | |
3852 | */ | |
3853 | i++; | |
3854 | ||
14228447 | 3855 | sub = submodule_from_path(&subrepo, null_oid(), ce->name); |
da62f786 SB |
3856 | if (!sub || !is_submodule_active(&subrepo, ce->name)) |
3857 | /* .gitmodules broken or inactive sub */ | |
3858 | continue; | |
3859 | ||
3860 | strbuf_reset(&sub_wt); | |
3861 | strbuf_reset(&sub_gd); | |
3862 | strbuf_addf(&sub_wt, "%s/%s", sub_worktree, sub->path); | |
ce125d43 | 3863 | submodule_name_to_gitdir(&sub_gd, &subrepo, sub->name); |
da62f786 SB |
3864 | |
3865 | connect_work_tree_and_git_dir(sub_wt.buf, sub_gd.buf, 1); | |
3866 | } | |
3867 | strbuf_release(&sub_wt); | |
3868 | strbuf_release(&sub_gd); | |
3869 | repo_clear(&subrepo); | |
3870 | } | |
3871 | ||
3872 | void connect_work_tree_and_git_dir(const char *work_tree_, | |
3873 | const char *git_dir_, | |
3874 | int recurse_into_nested) | |
47e83eb3 | 3875 | { |
365444a6 SB |
3876 | struct strbuf gitfile_sb = STRBUF_INIT; |
3877 | struct strbuf cfg_sb = STRBUF_INIT; | |
47e83eb3 | 3878 | struct strbuf rel_path = STRBUF_INIT; |
365444a6 | 3879 | char *git_dir, *work_tree; |
47e83eb3 | 3880 | |
365444a6 SB |
3881 | /* Prepare .git file */ |
3882 | strbuf_addf(&gitfile_sb, "%s/.git", work_tree_); | |
3883 | if (safe_create_leading_directories_const(gitfile_sb.buf)) | |
3884 | die(_("could not create directories for %s"), gitfile_sb.buf); | |
3885 | ||
3886 | /* Prepare config file */ | |
3887 | strbuf_addf(&cfg_sb, "%s/config", git_dir_); | |
3888 | if (safe_create_leading_directories_const(cfg_sb.buf)) | |
3889 | die(_("could not create directories for %s"), cfg_sb.buf); | |
47e83eb3 | 3890 | |
e394fa01 JH |
3891 | git_dir = real_pathdup(git_dir_, 1); |
3892 | work_tree = real_pathdup(work_tree_, 1); | |
365444a6 SB |
3893 | |
3894 | /* Write .git file */ | |
3895 | write_file(gitfile_sb.buf, "gitdir: %s", | |
3896 | relative_path(git_dir, work_tree, &rel_path)); | |
47e83eb3 | 3897 | /* Update core.worktree setting */ |
365444a6 | 3898 | git_config_set_in_file(cfg_sb.buf, "core.worktree", |
47e83eb3 SB |
3899 | relative_path(work_tree, git_dir, &rel_path)); |
3900 | ||
365444a6 SB |
3901 | strbuf_release(&gitfile_sb); |
3902 | strbuf_release(&cfg_sb); | |
47e83eb3 | 3903 | strbuf_release(&rel_path); |
da62f786 SB |
3904 | |
3905 | if (recurse_into_nested) | |
3906 | connect_wt_gitdir_in_nested(work_tree, git_dir); | |
3907 | ||
47e83eb3 SB |
3908 | free(work_tree); |
3909 | free(git_dir); | |
3910 | } | |
f6f85861 SB |
3911 | |
3912 | /* | |
3913 | * Migrate the git directory of the given path from old_git_dir to new_git_dir. | |
3914 | */ | |
3915 | void relocate_gitdir(const char *path, const char *old_git_dir, const char *new_git_dir) | |
3916 | { | |
3917 | if (rename(old_git_dir, new_git_dir) < 0) | |
3918 | die_errno(_("could not migrate git directory from '%s' to '%s'"), | |
3919 | old_git_dir, new_git_dir); | |
3920 | ||
da62f786 | 3921 | connect_work_tree_and_git_dir(path, new_git_dir, 0); |
f6f85861 | 3922 | } |