]>
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 | ||
cf7c6148 | 44 | struct dirent *de; |
91a2288b NTND |
45 | const char *file; |
46 | struct untracked_cache_dir *ucd; | |
cf7c6148 NTND |
47 | }; |
48 | ||
defd7c7b | 49 | static enum path_treatment read_directory_recursive(struct dir_struct *dir, |
0ef8e169 BW |
50 | struct index_state *istate, const char *path, int len, |
51 | struct untracked_cache_dir *untracked, | |
5aaa7fd3 | 52 | int check_only, int stop_at_first_file, const struct pathspec *pathspec); |
98f2a687 BW |
53 | static int get_dtype(struct dirent *de, struct index_state *istate, |
54 | const char *path, int len); | |
09595258 | 55 | |
e0556a92 PC |
56 | int count_slashes(const char *s) |
57 | { | |
58 | int cnt = 0; | |
59 | while (*s) | |
60 | if (*s++ == '/') | |
61 | cnt++; | |
62 | return cnt; | |
63 | } | |
64 | ||
ba0897e6 | 65 | int fspathcmp(const char *a, const char *b) |
8cf2a84e JJ |
66 | { |
67 | return ignore_case ? strcasecmp(a, b) : strcmp(a, b); | |
68 | } | |
69 | ||
ba0897e6 | 70 | int fspathncmp(const char *a, const char *b, size_t count) |
8cf2a84e JJ |
71 | { |
72 | return ignore_case ? strncasecmp(a, b, count) : strncmp(a, b, count); | |
73 | } | |
74 | ||
1f26ce61 CB |
75 | int git_fnmatch(const struct pathspec_item *item, |
76 | const char *pattern, const char *string, | |
77 | int prefix) | |
5d74762d | 78 | { |
5d74762d | 79 | if (prefix > 0) { |
93d93537 | 80 | if (ps_strncmp(item, pattern, string, prefix)) |
eb07894f | 81 | return WM_NOMATCH; |
5d74762d NTND |
82 | pattern += prefix; |
83 | string += prefix; | |
84 | } | |
bd30c2e4 | 85 | if (item->flags & PATHSPEC_ONESTAR) { |
8c6abbcd NTND |
86 | int pattern_len = strlen(++pattern); |
87 | int string_len = strlen(string); | |
88 | return string_len < pattern_len || | |
93d93537 NTND |
89 | ps_strcmp(item, pattern, |
90 | string + string_len - pattern_len); | |
8c6abbcd | 91 | } |
bd30c2e4 | 92 | if (item->magic & PATHSPEC_GLOB) |
93d93537 NTND |
93 | return wildmatch(pattern, string, |
94 | WM_PATHNAME | | |
55d34269 | 95 | (item->magic & PATHSPEC_ICASE ? WM_CASEFOLD : 0)); |
bd30c2e4 NTND |
96 | else |
97 | /* wildmatch has not learned no FNM_PATHNAME mode yet */ | |
eb07894f | 98 | return wildmatch(pattern, string, |
55d34269 | 99 | item->magic & PATHSPEC_ICASE ? WM_CASEFOLD : 0); |
5d74762d NTND |
100 | } |
101 | ||
0b6e56df JH |
102 | static int fnmatch_icase_mem(const char *pattern, int patternlen, |
103 | const char *string, int stringlen, | |
104 | int flags) | |
105 | { | |
106 | int match_status; | |
107 | struct strbuf pat_buf = STRBUF_INIT; | |
108 | struct strbuf str_buf = STRBUF_INIT; | |
109 | const char *use_pat = pattern; | |
110 | const char *use_str = string; | |
111 | ||
112 | if (pattern[patternlen]) { | |
113 | strbuf_add(&pat_buf, pattern, patternlen); | |
114 | use_pat = pat_buf.buf; | |
115 | } | |
116 | if (string[stringlen]) { | |
117 | strbuf_add(&str_buf, string, stringlen); | |
118 | use_str = str_buf.buf; | |
119 | } | |
120 | ||
f30366b2 JH |
121 | if (ignore_case) |
122 | flags |= WM_CASEFOLD; | |
55d34269 | 123 | match_status = wildmatch(use_pat, use_str, flags); |
0b6e56df JH |
124 | |
125 | strbuf_release(&pat_buf); | |
126 | strbuf_release(&str_buf); | |
127 | ||
128 | return match_status; | |
129 | } | |
130 | ||
827f4d6c | 131 | static size_t common_prefix_len(const struct pathspec *pathspec) |
3c6a370b | 132 | { |
827f4d6c | 133 | int n; |
4a085b16 | 134 | size_t max = 0; |
3c6a370b | 135 | |
93d93537 NTND |
136 | /* |
137 | * ":(icase)path" is treated as a pathspec full of | |
138 | * wildcard. In other words, only prefix is considered common | |
139 | * prefix. If the pathspec is abc/foo abc/bar, running in | |
bbbb6b0b | 140 | * subdir xyz, the common prefix is still xyz, not xyz/abc as |
93d93537 NTND |
141 | * in non-:(icase). |
142 | */ | |
5c6933d2 NTND |
143 | GUARD_PATHSPEC(pathspec, |
144 | PATHSPEC_FROMTOP | | |
145 | PATHSPEC_MAXDEPTH | | |
bd30c2e4 | 146 | PATHSPEC_LITERAL | |
93d93537 | 147 | PATHSPEC_GLOB | |
ef79b1f8 | 148 | PATHSPEC_ICASE | |
b0db7046 BW |
149 | PATHSPEC_EXCLUDE | |
150 | PATHSPEC_ATTR); | |
4a085b16 | 151 | |
827f4d6c | 152 | for (n = 0; n < pathspec->nr; n++) { |
93d93537 | 153 | size_t i = 0, len = 0, item_len; |
ef79b1f8 NTND |
154 | if (pathspec->items[n].magic & PATHSPEC_EXCLUDE) |
155 | continue; | |
93d93537 NTND |
156 | if (pathspec->items[n].magic & PATHSPEC_ICASE) |
157 | item_len = pathspec->items[n].prefix; | |
158 | else | |
159 | item_len = pathspec->items[n].nowildcard_len; | |
160 | while (i < item_len && (n == 0 || i < max)) { | |
827f4d6c NTND |
161 | char c = pathspec->items[n].match[i]; |
162 | if (c != pathspec->items[0].match[i]) | |
4a085b16 JH |
163 | break; |
164 | if (c == '/') | |
165 | len = i + 1; | |
827f4d6c | 166 | i++; |
4a085b16 | 167 | } |
827f4d6c | 168 | if (n == 0 || len < max) { |
4a085b16 JH |
169 | max = len; |
170 | if (!max) | |
171 | break; | |
172 | } | |
3c6a370b | 173 | } |
4a085b16 | 174 | return max; |
3c6a370b LT |
175 | } |
176 | ||
f950eb95 CB |
177 | /* |
178 | * Returns a copy of the longest leading path common among all | |
179 | * pathspecs. | |
180 | */ | |
827f4d6c | 181 | char *common_prefix(const struct pathspec *pathspec) |
f950eb95 CB |
182 | { |
183 | unsigned long len = common_prefix_len(pathspec); | |
184 | ||
827f4d6c | 185 | return len ? xmemdupz(pathspec->items[0].match, len) : NULL; |
f950eb95 CB |
186 | } |
187 | ||
0d32c183 BW |
188 | int fill_directory(struct dir_struct *dir, |
189 | struct index_state *istate, | |
190 | const struct pathspec *pathspec) | |
1d8842d9 | 191 | { |
bec5ab89 | 192 | const char *prefix; |
966de302 | 193 | size_t prefix_len; |
1d8842d9 LT |
194 | |
195 | /* | |
196 | * Calculate common prefix for the pathspec, and | |
197 | * use that to optimize the directory walk | |
198 | */ | |
bec5ab89 RS |
199 | prefix_len = common_prefix_len(pathspec); |
200 | prefix = prefix_len ? pathspec->items[0].match : ""; | |
1d8842d9 LT |
201 | |
202 | /* Read the directory and prune it */ | |
0d32c183 | 203 | read_directory(dir, istate, prefix, prefix_len, pathspec); |
966de302 | 204 | |
966de302 | 205 | return prefix_len; |
1d8842d9 LT |
206 | } |
207 | ||
bc96cc87 NTND |
208 | int within_depth(const char *name, int namelen, |
209 | int depth, int max_depth) | |
210 | { | |
211 | const char *cp = name, *cpe = name + namelen; | |
212 | ||
213 | while (cp < cpe) { | |
214 | if (*cp++ != '/') | |
215 | continue; | |
216 | depth++; | |
217 | if (depth > max_depth) | |
218 | return 0; | |
219 | } | |
220 | return 1; | |
221 | } | |
222 | ||
578d81d0 JH |
223 | /* |
224 | * Read the contents of the blob with the given OID into a buffer. | |
225 | * Append a trailing LF to the end if the last line doesn't have one. | |
226 | * | |
227 | * Returns: | |
228 | * -1 when the OID is invalid or unknown or does not refer to a blob. | |
229 | * 0 when the blob is empty. | |
230 | * 1 along with { data, size } of the (possibly augmented) buffer | |
231 | * when successful. | |
232 | * | |
4b33e602 | 233 | * Optionally updates the given oid_stat with the given OID (when valid). |
578d81d0 | 234 | */ |
4b33e602 PO |
235 | static int do_read_blob(const struct object_id *oid, struct oid_stat *oid_stat, |
236 | size_t *size_out, char **data_out) | |
578d81d0 JH |
237 | { |
238 | enum object_type type; | |
239 | unsigned long sz; | |
240 | char *data; | |
241 | ||
242 | *size_out = 0; | |
243 | *data_out = NULL; | |
244 | ||
b4f5aca4 | 245 | data = read_object_file(oid, &type, &sz); |
578d81d0 JH |
246 | if (!data || type != OBJ_BLOB) { |
247 | free(data); | |
248 | return -1; | |
249 | } | |
250 | ||
4b33e602 PO |
251 | if (oid_stat) { |
252 | memset(&oid_stat->stat, 0, sizeof(oid_stat->stat)); | |
253 | oidcpy(&oid_stat->oid, oid); | |
578d81d0 JH |
254 | } |
255 | ||
256 | if (sz == 0) { | |
257 | free(data); | |
258 | return 0; | |
259 | } | |
260 | ||
261 | if (data[sz - 1] != '\n') { | |
262 | data = xrealloc(data, st_add(sz, 1)); | |
263 | data[sz++] = '\n'; | |
264 | } | |
265 | ||
266 | *size_out = xsize_t(sz); | |
267 | *data_out = data; | |
268 | ||
269 | return 1; | |
270 | } | |
271 | ||
75a6315f BW |
272 | #define DO_MATCH_EXCLUDE (1<<0) |
273 | #define DO_MATCH_DIRECTORY (1<<1) | |
a3d89d8f | 274 | #define DO_MATCH_LEADING_PATHSPEC (1<<2) |
42b0874a | 275 | |
61cf2820 | 276 | /* |
29b577b9 | 277 | * Does the given pathspec match the given name? A match is found if |
61cf2820 | 278 | * |
29b577b9 EN |
279 | * (1) the pathspec string is leading directory of 'name' ("RECURSIVELY"), or |
280 | * (2) the pathspec string has a leading part matching 'name' ("LEADING"), or | |
281 | * (3) the pathspec string is a wildcard and matches 'name' ("WILDCARD"), or | |
282 | * (4) the pathspec string is exactly the same as 'name' ("EXACT"). | |
61cf2820 | 283 | * |
29b577b9 | 284 | * Return value tells which case it was (1-4), or 0 when there is no match. |
61cf2820 | 285 | * |
29b577b9 EN |
286 | * It may be instructive to look at a small table of concrete examples |
287 | * to understand the differences between 1, 2, and 4: | |
288 | * | |
289 | * Pathspecs | |
290 | * | a/b | a/b/ | a/b/c | |
291 | * ------+-----------+-----------+------------ | |
292 | * a/b | EXACT | EXACT[1] | LEADING[2] | |
293 | * Names a/b/ | RECURSIVE | EXACT | LEADING[2] | |
294 | * a/b/c | RECURSIVE | RECURSIVE | EXACT | |
295 | * | |
296 | * [1] Only if DO_MATCH_DIRECTORY is passed; otherwise, this is NOT a match. | |
297 | * [2] Only if DO_MATCH_LEADING_PATHSPEC is passed; otherwise, not a match. | |
61cf2820 | 298 | */ |
6d2df284 NTND |
299 | static int match_pathspec_item(const struct index_state *istate, |
300 | const struct pathspec_item *item, int prefix, | |
42b0874a | 301 | const char *name, int namelen, unsigned flags) |
61cf2820 NTND |
302 | { |
303 | /* name/namelen has prefix cut off by caller */ | |
304 | const char *match = item->match + prefix; | |
305 | int matchlen = item->len - prefix; | |
306 | ||
93d93537 NTND |
307 | /* |
308 | * The normal call pattern is: | |
309 | * 1. prefix = common_prefix_len(ps); | |
310 | * 2. prune something, or fill_directory | |
854b0959 | 311 | * 3. match_pathspec() |
93d93537 NTND |
312 | * |
313 | * 'prefix' at #1 may be shorter than the command's prefix and | |
314 | * it's ok for #2 to match extra files. Those extras will be | |
315 | * trimmed at #3. | |
316 | * | |
317 | * Suppose the pathspec is 'foo' and '../bar' running from | |
318 | * subdir 'xyz'. The common prefix at #1 will be empty, thanks | |
319 | * to "../". We may have xyz/foo _and_ XYZ/foo after #2. The | |
320 | * user does not want XYZ/foo, only the "foo" part should be | |
321 | * case-insensitive. We need to filter out XYZ/foo here. In | |
322 | * other words, we do not trust the caller on comparing the | |
323 | * prefix part when :(icase) is involved. We do exact | |
324 | * comparison ourselves. | |
325 | * | |
326 | * Normally the caller (common_prefix_len() in fact) does | |
327 | * _exact_ matching on name[-prefix+1..-1] and we do not need | |
328 | * to check that part. Be defensive and check it anyway, in | |
329 | * case common_prefix_len is changed, or a new caller is | |
330 | * introduced that does not use common_prefix_len. | |
331 | * | |
332 | * If the penalty turns out too high when prefix is really | |
333 | * long, maybe change it to | |
334 | * strncmp(match, name, item->prefix - prefix) | |
335 | */ | |
336 | if (item->prefix && (item->magic & PATHSPEC_ICASE) && | |
337 | strncmp(item->match, name - prefix, item->prefix)) | |
338 | return 0; | |
339 | ||
22af33be NTND |
340 | if (item->attr_match_nr && |
341 | !match_pathspec_attrs(istate, name, namelen, item)) | |
b0db7046 BW |
342 | return 0; |
343 | ||
61cf2820 NTND |
344 | /* If the match was just the prefix, we matched */ |
345 | if (!*match) | |
346 | return MATCHED_RECURSIVELY; | |
347 | ||
93d93537 | 348 | if (matchlen <= namelen && !ps_strncmp(item, match, name, matchlen)) { |
61cf2820 NTND |
349 | if (matchlen == namelen) |
350 | return MATCHED_EXACTLY; | |
351 | ||
352 | if (match[matchlen-1] == '/' || name[matchlen] == '/') | |
353 | return MATCHED_RECURSIVELY; | |
68690fdd NTND |
354 | } else if ((flags & DO_MATCH_DIRECTORY) && |
355 | match[matchlen - 1] == '/' && | |
356 | namelen == matchlen - 1 && | |
357 | !ps_strncmp(item, match, name, namelen)) | |
358 | return MATCHED_EXACTLY; | |
61cf2820 | 359 | |
5d74762d | 360 | if (item->nowildcard_len < item->len && |
bd30c2e4 | 361 | !git_fnmatch(item, match, name, |
8c6abbcd | 362 | item->nowildcard_len - prefix)) |
61cf2820 NTND |
363 | return MATCHED_FNMATCH; |
364 | ||
29b577b9 | 365 | /* Perform checks to see if "name" is a leading string of the pathspec */ |
a3d89d8f | 366 | if (flags & DO_MATCH_LEADING_PATHSPEC) { |
75a6315f | 367 | /* name is a literal prefix of the pathspec */ |
a5e916c7 | 368 | int offset = name[namelen-1] == '/' ? 1 : 0; |
75a6315f | 369 | if ((namelen < matchlen) && |
a5e916c7 | 370 | (match[namelen-offset] == '/') && |
75a6315f | 371 | !ps_strncmp(item, match, name, namelen)) |
89a1f4aa | 372 | return MATCHED_RECURSIVELY_LEADING_PATHSPEC; |
75a6315f | 373 | |
2f5d3847 | 374 | /* name doesn't match up to the first wild character */ |
75a6315f BW |
375 | if (item->nowildcard_len < item->len && |
376 | ps_strncmp(item, match, name, | |
377 | item->nowildcard_len - prefix)) | |
378 | return 0; | |
379 | ||
072a2310 EN |
380 | /* |
381 | * name has no wildcard, and it didn't match as a leading | |
382 | * pathspec so return. | |
383 | */ | |
384 | if (item->nowildcard_len == item->len) | |
385 | return 0; | |
386 | ||
75a6315f BW |
387 | /* |
388 | * Here is where we would perform a wildmatch to check if | |
389 | * "name" can be matched as a directory (or a prefix) against | |
390 | * the pathspec. Since wildmatch doesn't have this capability | |
391 | * at the present we have to punt and say that it is a match, | |
392 | * potentially returning a false positive | |
393 | * The submodules themselves will be able to perform more | |
394 | * accurate matching to determine if the pathspec matches. | |
395 | */ | |
89a1f4aa | 396 | return MATCHED_RECURSIVELY_LEADING_PATHSPEC; |
75a6315f BW |
397 | } |
398 | ||
61cf2820 NTND |
399 | return 0; |
400 | } | |
401 | ||
402 | /* | |
52ed1894 AS |
403 | * Given a name and a list of pathspecs, returns the nature of the |
404 | * closest (i.e. most specific) match of the name to any of the | |
405 | * pathspecs. | |
406 | * | |
407 | * The caller typically calls this multiple times with the same | |
408 | * pathspec and seen[] array but with different name/namelen | |
409 | * (e.g. entries from the index) and is interested in seeing if and | |
410 | * how each pathspec matches all the names it calls this function | |
411 | * with. A mark is left in the seen[] array for each pathspec element | |
412 | * indicating the closest type of match that element achieved, so if | |
413 | * seen[n] remains zero after multiple invocations, that means the nth | |
414 | * pathspec did not match any names, which could indicate that the | |
415 | * user mistyped the nth pathspec. | |
61cf2820 | 416 | */ |
6d2df284 NTND |
417 | static int do_match_pathspec(const struct index_state *istate, |
418 | const struct pathspec *ps, | |
854b0959 NTND |
419 | const char *name, int namelen, |
420 | int prefix, char *seen, | |
42b0874a | 421 | unsigned flags) |
61cf2820 | 422 | { |
42b0874a | 423 | int i, retval = 0, exclude = flags & DO_MATCH_EXCLUDE; |
61cf2820 | 424 | |
5c6933d2 NTND |
425 | GUARD_PATHSPEC(ps, |
426 | PATHSPEC_FROMTOP | | |
427 | PATHSPEC_MAXDEPTH | | |
bd30c2e4 | 428 | PATHSPEC_LITERAL | |
93d93537 | 429 | PATHSPEC_GLOB | |
ef79b1f8 | 430 | PATHSPEC_ICASE | |
b0db7046 BW |
431 | PATHSPEC_EXCLUDE | |
432 | PATHSPEC_ATTR); | |
8f4f8f45 | 433 | |
61cf2820 | 434 | if (!ps->nr) { |
6330a171 NTND |
435 | if (!ps->recursive || |
436 | !(ps->magic & PATHSPEC_MAXDEPTH) || | |
437 | ps->max_depth == -1) | |
61cf2820 NTND |
438 | return MATCHED_RECURSIVELY; |
439 | ||
440 | if (within_depth(name, namelen, 0, ps->max_depth)) | |
441 | return MATCHED_EXACTLY; | |
442 | else | |
443 | return 0; | |
444 | } | |
445 | ||
446 | name += prefix; | |
447 | namelen -= prefix; | |
448 | ||
449 | for (i = ps->nr - 1; i >= 0; i--) { | |
450 | int how; | |
ef79b1f8 NTND |
451 | |
452 | if ((!exclude && ps->items[i].magic & PATHSPEC_EXCLUDE) || | |
453 | ( exclude && !(ps->items[i].magic & PATHSPEC_EXCLUDE))) | |
454 | continue; | |
455 | ||
61cf2820 NTND |
456 | if (seen && seen[i] == MATCHED_EXACTLY) |
457 | continue; | |
ef79b1f8 NTND |
458 | /* |
459 | * Make exclude patterns optional and never report | |
460 | * "pathspec ':(exclude)foo' matches no files" | |
461 | */ | |
462 | if (seen && ps->items[i].magic & PATHSPEC_EXCLUDE) | |
463 | seen[i] = MATCHED_FNMATCH; | |
6d2df284 | 464 | how = match_pathspec_item(istate, ps->items+i, prefix, name, |
42b0874a | 465 | namelen, flags); |
6330a171 NTND |
466 | if (ps->recursive && |
467 | (ps->magic & PATHSPEC_MAXDEPTH) && | |
468 | ps->max_depth != -1 && | |
61cf2820 NTND |
469 | how && how != MATCHED_FNMATCH) { |
470 | int len = ps->items[i].len; | |
471 | if (name[len] == '/') | |
472 | len++; | |
473 | if (within_depth(name+len, namelen-len, 0, ps->max_depth)) | |
474 | how = MATCHED_EXACTLY; | |
475 | else | |
476 | how = 0; | |
477 | } | |
478 | if (how) { | |
479 | if (retval < how) | |
480 | retval = how; | |
481 | if (seen && seen[i] < how) | |
482 | seen[i] = how; | |
483 | } | |
484 | } | |
485 | return retval; | |
486 | } | |
487 | ||
6d2df284 NTND |
488 | int match_pathspec(const struct index_state *istate, |
489 | const struct pathspec *ps, | |
854b0959 | 490 | const char *name, int namelen, |
ae8d0824 | 491 | int prefix, char *seen, int is_dir) |
ef79b1f8 NTND |
492 | { |
493 | int positive, negative; | |
ae8d0824 | 494 | unsigned flags = is_dir ? DO_MATCH_DIRECTORY : 0; |
6d2df284 | 495 | positive = do_match_pathspec(istate, ps, name, namelen, |
42b0874a | 496 | prefix, seen, flags); |
ef79b1f8 NTND |
497 | if (!(ps->magic & PATHSPEC_EXCLUDE) || !positive) |
498 | return positive; | |
6d2df284 | 499 | negative = do_match_pathspec(istate, ps, name, namelen, |
42b0874a NTND |
500 | prefix, seen, |
501 | flags | DO_MATCH_EXCLUDE); | |
ef79b1f8 NTND |
502 | return negative ? 0 : positive; |
503 | } | |
504 | ||
75a6315f BW |
505 | /** |
506 | * Check if a submodule is a superset of the pathspec | |
507 | */ | |
6d2df284 NTND |
508 | int submodule_path_match(const struct index_state *istate, |
509 | const struct pathspec *ps, | |
75a6315f BW |
510 | const char *submodule_name, |
511 | char *seen) | |
512 | { | |
6d2df284 | 513 | int matched = do_match_pathspec(istate, ps, submodule_name, |
75a6315f BW |
514 | strlen(submodule_name), |
515 | 0, seen, | |
516 | DO_MATCH_DIRECTORY | | |
a3d89d8f | 517 | DO_MATCH_LEADING_PATHSPEC); |
75a6315f BW |
518 | return matched; |
519 | } | |
520 | ||
777c55a6 | 521 | int report_path_error(const char *ps_matched, |
c5c33504 | 522 | const struct pathspec *pathspec) |
777c55a6 JH |
523 | { |
524 | /* | |
525 | * Make sure all pathspec matched; otherwise it is an error. | |
526 | */ | |
777c55a6 JH |
527 | int num, errors = 0; |
528 | for (num = 0; num < pathspec->nr; num++) { | |
529 | int other, found_dup; | |
530 | ||
531 | if (ps_matched[num]) | |
532 | continue; | |
533 | /* | |
534 | * The caller might have fed identical pathspec | |
535 | * twice. Do not barf on such a mistake. | |
536 | * FIXME: parse_pathspec should have eliminated | |
537 | * duplicate pathspec. | |
538 | */ | |
539 | for (found_dup = other = 0; | |
540 | !found_dup && other < pathspec->nr; | |
541 | other++) { | |
542 | if (other == num || !ps_matched[other]) | |
543 | continue; | |
544 | if (!strcmp(pathspec->items[other].original, | |
545 | pathspec->items[num].original)) | |
546 | /* | |
547 | * Ok, we have a match already. | |
548 | */ | |
549 | found_dup = 1; | |
550 | } | |
551 | if (found_dup) | |
552 | continue; | |
553 | ||
a80897c1 | 554 | error(_("pathspec '%s' did not match any file(s) known to git"), |
777c55a6 JH |
555 | pathspec->items[num].original); |
556 | errors++; | |
557 | } | |
777c55a6 JH |
558 | return errors; |
559 | } | |
560 | ||
fcd631ed NTND |
561 | /* |
562 | * Return the length of the "simple" part of a path match limiter. | |
563 | */ | |
87323bda | 564 | int simple_length(const char *match) |
fcd631ed NTND |
565 | { |
566 | int len = -1; | |
567 | ||
568 | for (;;) { | |
569 | unsigned char c = *match++; | |
570 | len++; | |
571 | if (c == '\0' || is_glob_special(c)) | |
572 | return len; | |
573 | } | |
574 | } | |
575 | ||
87323bda | 576 | int no_wildcard(const char *string) |
68492fc7 | 577 | { |
fcd631ed | 578 | return string[simple_length(string)] == '\0'; |
68492fc7 LK |
579 | } |
580 | ||
65edd96a | 581 | void parse_path_pattern(const char **pattern, |
82dce998 | 582 | int *patternlen, |
f8708998 | 583 | unsigned *flags, |
82dce998 | 584 | int *nowildcardlen) |
84460eec NTND |
585 | { |
586 | const char *p = *pattern; | |
587 | size_t i, len; | |
588 | ||
589 | *flags = 0; | |
590 | if (*p == '!') { | |
4ff89ee5 | 591 | *flags |= PATTERN_FLAG_NEGATIVE; |
84460eec NTND |
592 | p++; |
593 | } | |
594 | len = strlen(p); | |
595 | if (len && p[len - 1] == '/') { | |
596 | len--; | |
4ff89ee5 | 597 | *flags |= PATTERN_FLAG_MUSTBEDIR; |
84460eec NTND |
598 | } |
599 | for (i = 0; i < len; i++) { | |
600 | if (p[i] == '/') | |
601 | break; | |
602 | } | |
603 | if (i == len) | |
4ff89ee5 | 604 | *flags |= PATTERN_FLAG_NODIR; |
84460eec NTND |
605 | *nowildcardlen = simple_length(p); |
606 | /* | |
607 | * we should have excluded the trailing slash from 'p' too, | |
608 | * but that's one more allocation. Instead just make sure | |
609 | * nowildcardlen does not exceed real patternlen | |
610 | */ | |
611 | if (*nowildcardlen > len) | |
612 | *nowildcardlen = len; | |
613 | if (*p == '*' && no_wildcard(p + 1)) | |
4ff89ee5 | 614 | *flags |= PATTERN_FLAG_ENDSWITH; |
84460eec NTND |
615 | *pattern = p; |
616 | *patternlen = len; | |
617 | } | |
618 | ||
af09ce24 DS |
619 | int pl_hashmap_cmp(const void *unused_cmp_data, |
620 | const struct hashmap_entry *a, | |
621 | const struct hashmap_entry *b, | |
622 | const void *key) | |
96cc8ab5 DS |
623 | { |
624 | const struct pattern_entry *ee1 = | |
625 | container_of(a, struct pattern_entry, ent); | |
626 | const struct pattern_entry *ee2 = | |
627 | container_of(b, struct pattern_entry, ent); | |
628 | ||
629 | size_t min_len = ee1->patternlen <= ee2->patternlen | |
630 | ? ee1->patternlen | |
631 | : ee2->patternlen; | |
632 | ||
190a65f9 DS |
633 | if (ignore_case) |
634 | return strncasecmp(ee1->pattern, ee2->pattern, min_len); | |
96cc8ab5 DS |
635 | return strncmp(ee1->pattern, ee2->pattern, min_len); |
636 | } | |
637 | ||
638 | static void add_pattern_to_hashsets(struct pattern_list *pl, struct path_pattern *given) | |
639 | { | |
640 | struct pattern_entry *translated; | |
641 | char *truncated; | |
642 | char *data = NULL; | |
643 | ||
644 | if (!pl->use_cone_patterns) | |
645 | return; | |
646 | ||
647 | if (given->flags & PATTERN_FLAG_NEGATIVE && | |
648 | given->flags & PATTERN_FLAG_MUSTBEDIR && | |
649 | !strcmp(given->pattern, "/*")) { | |
650 | pl->full_cone = 0; | |
651 | return; | |
652 | } | |
653 | ||
654 | if (!given->flags && !strcmp(given->pattern, "/*")) { | |
655 | pl->full_cone = 1; | |
656 | return; | |
657 | } | |
658 | ||
659 | if (given->patternlen > 2 && | |
660 | !strcmp(given->pattern + given->patternlen - 2, "/*")) { | |
661 | if (!(given->flags & PATTERN_FLAG_NEGATIVE)) { | |
662 | /* Not a cone pattern. */ | |
663 | pl->use_cone_patterns = 0; | |
664 | warning(_("unrecognized pattern: '%s'"), given->pattern); | |
665 | goto clear_hashmaps; | |
666 | } | |
667 | ||
668 | truncated = xstrdup(given->pattern); | |
669 | truncated[given->patternlen - 2] = 0; | |
670 | ||
671 | translated = xmalloc(sizeof(struct pattern_entry)); | |
672 | translated->pattern = truncated; | |
673 | translated->patternlen = given->patternlen - 2; | |
674 | hashmap_entry_init(&translated->ent, | |
190a65f9 DS |
675 | ignore_case ? |
676 | strihash(translated->pattern) : | |
677 | strhash(translated->pattern)); | |
96cc8ab5 DS |
678 | |
679 | if (!hashmap_get_entry(&pl->recursive_hashmap, | |
680 | translated, ent, NULL)) { | |
681 | /* We did not see the "parent" included */ | |
682 | warning(_("unrecognized negative pattern: '%s'"), | |
683 | given->pattern); | |
684 | free(truncated); | |
685 | free(translated); | |
686 | goto clear_hashmaps; | |
687 | } | |
688 | ||
689 | hashmap_add(&pl->parent_hashmap, &translated->ent); | |
690 | hashmap_remove(&pl->recursive_hashmap, &translated->ent, &data); | |
691 | free(data); | |
692 | return; | |
693 | } | |
694 | ||
695 | if (given->flags & PATTERN_FLAG_NEGATIVE) { | |
696 | warning(_("unrecognized negative pattern: '%s'"), | |
697 | given->pattern); | |
698 | goto clear_hashmaps; | |
699 | } | |
700 | ||
701 | translated = xmalloc(sizeof(struct pattern_entry)); | |
702 | ||
703 | translated->pattern = xstrdup(given->pattern); | |
704 | translated->patternlen = given->patternlen; | |
705 | hashmap_entry_init(&translated->ent, | |
190a65f9 DS |
706 | ignore_case ? |
707 | strihash(translated->pattern) : | |
708 | strhash(translated->pattern)); | |
96cc8ab5 DS |
709 | |
710 | hashmap_add(&pl->recursive_hashmap, &translated->ent); | |
711 | ||
712 | if (hashmap_get_entry(&pl->parent_hashmap, translated, ent, NULL)) { | |
713 | /* we already included this at the parent level */ | |
714 | warning(_("your sparse-checkout file may have issues: pattern '%s' is repeated"), | |
715 | given->pattern); | |
716 | hashmap_remove(&pl->parent_hashmap, &translated->ent, &data); | |
717 | free(data); | |
718 | free(translated); | |
719 | } | |
720 | ||
721 | return; | |
722 | ||
723 | clear_hashmaps: | |
724 | warning(_("disabling cone pattern matching")); | |
725 | hashmap_free_entries(&pl->parent_hashmap, struct pattern_entry, ent); | |
726 | hashmap_free_entries(&pl->recursive_hashmap, struct pattern_entry, ent); | |
727 | pl->use_cone_patterns = 0; | |
728 | } | |
729 | ||
730 | static int hashmap_contains_path(struct hashmap *map, | |
731 | struct strbuf *pattern) | |
732 | { | |
733 | struct pattern_entry p; | |
734 | ||
735 | /* Check straight mapping */ | |
736 | p.pattern = pattern->buf; | |
737 | p.patternlen = pattern->len; | |
190a65f9 DS |
738 | hashmap_entry_init(&p.ent, |
739 | ignore_case ? | |
740 | strihash(p.pattern) : | |
741 | strhash(p.pattern)); | |
96cc8ab5 DS |
742 | return !!hashmap_get_entry(map, &p, ent, NULL); |
743 | } | |
744 | ||
745 | int hashmap_contains_parent(struct hashmap *map, | |
746 | const char *path, | |
747 | struct strbuf *buffer) | |
748 | { | |
749 | char *slash_pos; | |
750 | ||
751 | strbuf_setlen(buffer, 0); | |
752 | ||
753 | if (path[0] != '/') | |
754 | strbuf_addch(buffer, '/'); | |
755 | ||
756 | strbuf_addstr(buffer, path); | |
757 | ||
758 | slash_pos = strrchr(buffer->buf, '/'); | |
759 | ||
760 | while (slash_pos > buffer->buf) { | |
761 | strbuf_setlen(buffer, slash_pos - buffer->buf); | |
762 | ||
763 | if (hashmap_contains_path(map, buffer)) | |
764 | return 1; | |
765 | ||
766 | slash_pos = strrchr(buffer->buf, '/'); | |
767 | } | |
768 | ||
769 | return 0; | |
770 | } | |
771 | ||
65edd96a | 772 | void add_pattern(const char *string, const char *base, |
caa3d554 | 773 | int baselen, struct pattern_list *pl, int srcpos) |
453ec4bd | 774 | { |
ab8db613 | 775 | struct path_pattern *pattern; |
84460eec | 776 | int patternlen; |
f8708998 | 777 | unsigned flags; |
84460eec | 778 | int nowildcardlen; |
453ec4bd | 779 | |
65edd96a | 780 | parse_path_pattern(&string, &patternlen, &flags, &nowildcardlen); |
4ff89ee5 | 781 | if (flags & PATTERN_FLAG_MUSTBEDIR) { |
ab8db613 | 782 | FLEXPTR_ALLOC_MEM(pattern, pattern, string, patternlen); |
d6b8fc30 | 783 | } else { |
ab8db613 DS |
784 | pattern = xmalloc(sizeof(*pattern)); |
785 | pattern->pattern = string; | |
d6b8fc30 | 786 | } |
ab8db613 DS |
787 | pattern->patternlen = patternlen; |
788 | pattern->nowildcardlen = nowildcardlen; | |
789 | pattern->base = base; | |
790 | pattern->baselen = baselen; | |
791 | pattern->flags = flags; | |
792 | pattern->srcpos = srcpos; | |
caa3d554 DS |
793 | ALLOC_GROW(pl->patterns, pl->nr + 1, pl->alloc); |
794 | pl->patterns[pl->nr++] = pattern; | |
795 | pattern->pl = pl; | |
96cc8ab5 DS |
796 | |
797 | add_pattern_to_hashsets(pl, pattern); | |
453ec4bd LT |
798 | } |
799 | ||
578d81d0 JH |
800 | static int read_skip_worktree_file_from_index(const struct index_state *istate, |
801 | const char *path, | |
4b33e602 PO |
802 | size_t *size_out, char **data_out, |
803 | struct oid_stat *oid_stat) | |
c28b3d6e NTND |
804 | { |
805 | int pos, len; | |
c28b3d6e NTND |
806 | |
807 | len = strlen(path); | |
6f52b741 | 808 | pos = index_name_pos(istate, path, len); |
c28b3d6e | 809 | if (pos < 0) |
578d81d0 | 810 | return -1; |
6f52b741 | 811 | if (!ce_skip_worktree(istate->cache[pos])) |
578d81d0 JH |
812 | return -1; |
813 | ||
4b33e602 | 814 | return do_read_blob(&istate->cache[pos]->oid, oid_stat, size_out, data_out); |
c28b3d6e NTND |
815 | } |
816 | ||
f6198812 | 817 | /* |
caa3d554 DS |
818 | * Frees memory within pl which was allocated for exclude patterns and |
819 | * the file buffer. Does not free pl itself. | |
f6198812 | 820 | */ |
65edd96a | 821 | void clear_pattern_list(struct pattern_list *pl) |
0fd0e241 NTND |
822 | { |
823 | int i; | |
824 | ||
caa3d554 DS |
825 | for (i = 0; i < pl->nr; i++) |
826 | free(pl->patterns[i]); | |
827 | free(pl->patterns); | |
828 | free(pl->filebuf); | |
0fd0e241 | 829 | |
caa3d554 | 830 | memset(pl, 0, sizeof(*pl)); |
0fd0e241 NTND |
831 | } |
832 | ||
7e2e4b37 | 833 | static void trim_trailing_spaces(char *buf) |
16402b99 | 834 | { |
e61a6c1d PB |
835 | char *p, *last_space = NULL; |
836 | ||
837 | for (p = buf; *p; p++) | |
838 | switch (*p) { | |
839 | case ' ': | |
840 | if (!last_space) | |
841 | last_space = p; | |
842 | break; | |
843 | case '\\': | |
844 | p++; | |
845 | if (!*p) | |
846 | return; | |
847 | /* fallthrough */ | |
848 | default: | |
849 | last_space = NULL; | |
850 | } | |
851 | ||
852 | if (last_space) | |
853 | *last_space = '\0'; | |
16402b99 NTND |
854 | } |
855 | ||
0dcb8d7f NTND |
856 | /* |
857 | * Given a subdirectory name and "dir" of the current directory, | |
858 | * search the subdir in "dir" and return it, or create a new one if it | |
859 | * does not exist in "dir". | |
860 | * | |
861 | * If "name" has the trailing slash, it'll be excluded in the search. | |
862 | */ | |
863 | static struct untracked_cache_dir *lookup_untracked(struct untracked_cache *uc, | |
864 | struct untracked_cache_dir *dir, | |
865 | const char *name, int len) | |
866 | { | |
867 | int first, last; | |
868 | struct untracked_cache_dir *d; | |
869 | if (!dir) | |
870 | return NULL; | |
871 | if (len && name[len - 1] == '/') | |
872 | len--; | |
873 | first = 0; | |
874 | last = dir->dirs_nr; | |
875 | while (last > first) { | |
568a05c5 | 876 | int cmp, next = first + ((last - first) >> 1); |
0dcb8d7f NTND |
877 | d = dir->dirs[next]; |
878 | cmp = strncmp(name, d->name, len); | |
879 | if (!cmp && strlen(d->name) > len) | |
880 | cmp = -1; | |
881 | if (!cmp) | |
882 | return d; | |
883 | if (cmp < 0) { | |
884 | last = next; | |
885 | continue; | |
886 | } | |
887 | first = next+1; | |
888 | } | |
889 | ||
890 | uc->dir_created++; | |
96ffc06f | 891 | FLEX_ALLOC_MEM(d, name, name, len); |
0dcb8d7f NTND |
892 | |
893 | ALLOC_GROW(dir->dirs, dir->dirs_nr + 1, dir->dirs_alloc); | |
f919ffeb SG |
894 | MOVE_ARRAY(dir->dirs + first + 1, dir->dirs + first, |
895 | dir->dirs_nr - first); | |
0dcb8d7f NTND |
896 | dir->dirs_nr++; |
897 | dir->dirs[first] = d; | |
898 | return d; | |
899 | } | |
900 | ||
ccad261f NTND |
901 | static void do_invalidate_gitignore(struct untracked_cache_dir *dir) |
902 | { | |
903 | int i; | |
904 | dir->valid = 0; | |
905 | dir->untracked_nr = 0; | |
906 | for (i = 0; i < dir->dirs_nr; i++) | |
907 | do_invalidate_gitignore(dir->dirs[i]); | |
908 | } | |
909 | ||
910 | static void invalidate_gitignore(struct untracked_cache *uc, | |
911 | struct untracked_cache_dir *dir) | |
912 | { | |
913 | uc->gitignore_invalidated++; | |
914 | do_invalidate_gitignore(dir); | |
915 | } | |
916 | ||
91a2288b NTND |
917 | static void invalidate_directory(struct untracked_cache *uc, |
918 | struct untracked_cache_dir *dir) | |
919 | { | |
26cb0182 | 920 | int i; |
b6403131 NTND |
921 | |
922 | /* | |
923 | * Invalidation increment here is just roughly correct. If | |
924 | * untracked_nr or any of dirs[].recurse is non-zero, we | |
925 | * should increment dir_invalidated too. But that's more | |
926 | * expensive to do. | |
927 | */ | |
928 | if (dir->valid) | |
929 | uc->dir_invalidated++; | |
930 | ||
91a2288b NTND |
931 | dir->valid = 0; |
932 | dir->untracked_nr = 0; | |
26cb0182 NTND |
933 | for (i = 0; i < dir->dirs_nr; i++) |
934 | dir->dirs[i]->recurse = 0; | |
91a2288b NTND |
935 | } |
936 | ||
65edd96a | 937 | static int add_patterns_from_buffer(char *buf, size_t size, |
578d81d0 | 938 | const char *base, int baselen, |
caa3d554 | 939 | struct pattern_list *pl); |
578d81d0 | 940 | |
55fe6f51 NTND |
941 | /* |
942 | * Given a file with name "fname", read it (either from disk, or from | |
473e3930 | 943 | * an index if 'istate' is non-null), parse it and store the |
caa3d554 | 944 | * exclude rules in "pl". |
55fe6f51 NTND |
945 | * |
946 | * If "ss" is not NULL, compute SHA-1 of the exclude file and fill | |
65edd96a | 947 | * stat data from disk (only valid if add_patterns returns zero). If |
55fe6f51 NTND |
948 | * ss_valid is non-zero, "ss" must contain good value as input. |
949 | */ | |
65edd96a | 950 | static int add_patterns(const char *fname, const char *base, int baselen, |
caa3d554 | 951 | struct pattern_list *pl, struct index_state *istate, |
4b33e602 | 952 | struct oid_stat *oid_stat) |
453ec4bd | 953 | { |
c470701a | 954 | struct stat st; |
578d81d0 JH |
955 | int r; |
956 | int fd; | |
9d14017a | 957 | size_t size = 0; |
578d81d0 | 958 | char *buf; |
453ec4bd LT |
959 | |
960 | fd = open(fname, O_RDONLY); | |
c28b3d6e | 961 | if (fd < 0 || fstat(fd, &st) < 0) { |
11dc1fcb NTND |
962 | if (fd < 0) |
963 | warn_on_fopen_errors(fname); | |
964 | else | |
c28b3d6e | 965 | close(fd); |
578d81d0 | 966 | if (!istate) |
c28b3d6e | 967 | return -1; |
578d81d0 JH |
968 | r = read_skip_worktree_file_from_index(istate, fname, |
969 | &size, &buf, | |
4b33e602 | 970 | oid_stat); |
578d81d0 JH |
971 | if (r != 1) |
972 | return r; | |
d961baa8 | 973 | } else { |
c28b3d6e NTND |
974 | size = xsize_t(st.st_size); |
975 | if (size == 0) { | |
4b33e602 PO |
976 | if (oid_stat) { |
977 | fill_stat_data(&oid_stat->stat, &st); | |
ba2df751 | 978 | oidcpy(&oid_stat->oid, the_hash_algo->empty_blob); |
4b33e602 | 979 | oid_stat->valid = 1; |
55fe6f51 | 980 | } |
c28b3d6e NTND |
981 | close(fd); |
982 | return 0; | |
983 | } | |
3733e694 | 984 | buf = xmallocz(size); |
c28b3d6e | 985 | if (read_in_full(fd, buf, size) != size) { |
45d76f17 | 986 | free(buf); |
c28b3d6e NTND |
987 | close(fd); |
988 | return -1; | |
989 | } | |
45d76f17 | 990 | buf[size++] = '\n'; |
c28b3d6e | 991 | close(fd); |
4b33e602 | 992 | if (oid_stat) { |
55fe6f51 | 993 | int pos; |
4b33e602 PO |
994 | if (oid_stat->valid && |
995 | !match_stat_data_racy(istate, &oid_stat->stat, &st)) | |
55fe6f51 | 996 | ; /* no content change, ss->sha1 still good */ |
473e3930 BW |
997 | else if (istate && |
998 | (pos = index_name_pos(istate, fname, strlen(fname))) >= 0 && | |
999 | !ce_stage(istate->cache[pos]) && | |
1000 | ce_uptodate(istate->cache[pos]) && | |
82b474e0 | 1001 | !would_convert_to_git(istate, fname)) |
4b33e602 PO |
1002 | oidcpy(&oid_stat->oid, |
1003 | &istate->cache[pos]->oid); | |
55fe6f51 | 1004 | else |
f070facc PO |
1005 | hash_object_file(buf, size, "blob", |
1006 | &oid_stat->oid); | |
4b33e602 PO |
1007 | fill_stat_data(&oid_stat->stat, &st); |
1008 | oid_stat->valid = 1; | |
55fe6f51 | 1009 | } |
6ba78238 | 1010 | } |
453ec4bd | 1011 | |
65edd96a | 1012 | add_patterns_from_buffer(buf, size, base, baselen, pl); |
578d81d0 JH |
1013 | return 0; |
1014 | } | |
1015 | ||
65edd96a | 1016 | static int add_patterns_from_buffer(char *buf, size_t size, |
578d81d0 | 1017 | const char *base, int baselen, |
caa3d554 | 1018 | struct pattern_list *pl) |
578d81d0 JH |
1019 | { |
1020 | int i, lineno = 1; | |
1021 | char *entry; | |
1022 | ||
96cc8ab5 DS |
1023 | hashmap_init(&pl->recursive_hashmap, pl_hashmap_cmp, NULL, 0); |
1024 | hashmap_init(&pl->parent_hashmap, pl_hashmap_cmp, NULL, 0); | |
1025 | ||
caa3d554 | 1026 | pl->filebuf = buf; |
245e1c19 | 1027 | |
dde843e7 | 1028 | if (skip_utf8_bom(&buf, size)) |
caa3d554 | 1029 | size -= buf - pl->filebuf; |
dde843e7 | 1030 | |
453ec4bd | 1031 | entry = buf; |
245e1c19 | 1032 | |
45d76f17 NTND |
1033 | for (i = 0; i < size; i++) { |
1034 | if (buf[i] == '\n') { | |
453ec4bd LT |
1035 | if (entry != buf + i && entry[0] != '#') { |
1036 | buf[i - (i && buf[i-1] == '\r')] = 0; | |
7e2e4b37 | 1037 | trim_trailing_spaces(entry); |
65edd96a | 1038 | add_pattern(entry, base, baselen, pl, lineno); |
453ec4bd | 1039 | } |
c04318e4 | 1040 | lineno++; |
453ec4bd LT |
1041 | entry = buf + i + 1; |
1042 | } | |
1043 | } | |
1044 | return 0; | |
453ec4bd LT |
1045 | } |
1046 | ||
65edd96a | 1047 | int add_patterns_from_file_to_list(const char *fname, const char *base, |
caa3d554 | 1048 | int baselen, struct pattern_list *pl, |
473e3930 | 1049 | struct index_state *istate) |
55fe6f51 | 1050 | { |
65edd96a | 1051 | return add_patterns(fname, base, baselen, pl, istate, NULL); |
55fe6f51 NTND |
1052 | } |
1053 | ||
65edd96a | 1054 | int add_patterns_from_blob_to_list( |
578d81d0 JH |
1055 | struct object_id *oid, |
1056 | const char *base, int baselen, | |
caa3d554 | 1057 | struct pattern_list *pl) |
578d81d0 JH |
1058 | { |
1059 | char *buf; | |
1060 | size_t size; | |
1061 | int r; | |
1062 | ||
1063 | r = do_read_blob(oid, NULL, &size, &buf); | |
1064 | if (r != 1) | |
1065 | return r; | |
1066 | ||
65edd96a | 1067 | add_patterns_from_buffer(buf, size, base, baselen, pl); |
578d81d0 JH |
1068 | return 0; |
1069 | } | |
1070 | ||
65edd96a | 1071 | struct pattern_list *add_pattern_list(struct dir_struct *dir, |
c04318e4 | 1072 | int group_type, const char *src) |
c082df24 | 1073 | { |
caa3d554 | 1074 | struct pattern_list *pl; |
c082df24 AS |
1075 | struct exclude_list_group *group; |
1076 | ||
1077 | group = &dir->exclude_list_group[group_type]; | |
caa3d554 DS |
1078 | ALLOC_GROW(group->pl, group->nr + 1, group->alloc); |
1079 | pl = &group->pl[group->nr++]; | |
1080 | memset(pl, 0, sizeof(*pl)); | |
1081 | pl->src = src; | |
1082 | return pl; | |
c082df24 AS |
1083 | } |
1084 | ||
1085 | /* | |
1086 | * Used to set up core.excludesfile and .git/info/exclude lists. | |
1087 | */ | |
65edd96a | 1088 | static void add_patterns_from_file_1(struct dir_struct *dir, const char *fname, |
4b33e602 | 1089 | struct oid_stat *oid_stat) |
453ec4bd | 1090 | { |
caa3d554 | 1091 | struct pattern_list *pl; |
ccad261f NTND |
1092 | /* |
1093 | * catch setup_standard_excludes() that's called before | |
1094 | * dir->untracked is assigned. That function behaves | |
1095 | * differently when dir->untracked is non-NULL. | |
1096 | */ | |
1097 | if (!dir->untracked) | |
1098 | dir->unmanaged_exclude_files++; | |
65edd96a DS |
1099 | pl = add_pattern_list(dir, EXC_FILE, fname); |
1100 | if (add_patterns(fname, "", 0, pl, NULL, oid_stat) < 0) | |
a80897c1 | 1101 | die(_("cannot use %s as an exclude file"), fname); |
453ec4bd LT |
1102 | } |
1103 | ||
65edd96a | 1104 | void add_patterns_from_file(struct dir_struct *dir, const char *fname) |
0dcb8d7f | 1105 | { |
ccad261f | 1106 | dir->unmanaged_exclude_files++; /* see validate_untracked_cache() */ |
65edd96a | 1107 | add_patterns_from_file_1(dir, fname, NULL); |
0dcb8d7f NTND |
1108 | } |
1109 | ||
82dce998 NTND |
1110 | int match_basename(const char *basename, int basenamelen, |
1111 | const char *pattern, int prefix, int patternlen, | |
f8708998 | 1112 | unsigned flags) |
593cb880 NTND |
1113 | { |
1114 | if (prefix == patternlen) { | |
0b6e56df | 1115 | if (patternlen == basenamelen && |
ba0897e6 | 1116 | !fspathncmp(pattern, basename, basenamelen)) |
593cb880 | 1117 | return 1; |
4ff89ee5 | 1118 | } else if (flags & PATTERN_FLAG_ENDSWITH) { |
0b6e56df | 1119 | /* "*literal" matching against "fooliteral" */ |
593cb880 | 1120 | if (patternlen - 1 <= basenamelen && |
ba0897e6 | 1121 | !fspathncmp(pattern + 1, |
0b6e56df JH |
1122 | basename + basenamelen - (patternlen - 1), |
1123 | patternlen - 1)) | |
593cb880 NTND |
1124 | return 1; |
1125 | } else { | |
0b6e56df JH |
1126 | if (fnmatch_icase_mem(pattern, patternlen, |
1127 | basename, basenamelen, | |
1128 | 0) == 0) | |
593cb880 NTND |
1129 | return 1; |
1130 | } | |
1131 | return 0; | |
1132 | } | |
1133 | ||
82dce998 NTND |
1134 | int match_pathname(const char *pathname, int pathlen, |
1135 | const char *base, int baselen, | |
1136 | const char *pattern, int prefix, int patternlen, | |
f8708998 | 1137 | unsigned flags) |
b5592632 NTND |
1138 | { |
1139 | const char *name; | |
1140 | int namelen; | |
1141 | ||
1142 | /* | |
1143 | * match with FNM_PATHNAME; the pattern has base implicitly | |
1144 | * in front of it. | |
1145 | */ | |
1146 | if (*pattern == '/') { | |
1147 | pattern++; | |
982ac873 | 1148 | patternlen--; |
b5592632 NTND |
1149 | prefix--; |
1150 | } | |
1151 | ||
1152 | /* | |
1153 | * baselen does not count the trailing slash. base[] may or | |
1154 | * may not end with a trailing slash though. | |
1155 | */ | |
1156 | if (pathlen < baselen + 1 || | |
1157 | (baselen && pathname[baselen] != '/') || | |
ba0897e6 | 1158 | fspathncmp(pathname, base, baselen)) |
b5592632 NTND |
1159 | return 0; |
1160 | ||
1161 | namelen = baselen ? pathlen - baselen - 1 : pathlen; | |
1162 | name = pathname + pathlen - namelen; | |
1163 | ||
1164 | if (prefix) { | |
1165 | /* | |
1166 | * if the non-wildcard part is longer than the | |
1167 | * remaining pathname, surely it cannot match. | |
1168 | */ | |
1169 | if (prefix > namelen) | |
1170 | return 0; | |
1171 | ||
ba0897e6 | 1172 | if (fspathncmp(pattern, name, prefix)) |
b5592632 NTND |
1173 | return 0; |
1174 | pattern += prefix; | |
ab3aebc1 | 1175 | patternlen -= prefix; |
b5592632 NTND |
1176 | name += prefix; |
1177 | namelen -= prefix; | |
ab3aebc1 JK |
1178 | |
1179 | /* | |
1180 | * If the whole pattern did not have a wildcard, | |
1181 | * then our prefix match is all we need; we | |
1182 | * do not need to call fnmatch at all. | |
1183 | */ | |
5cee3493 | 1184 | if (!patternlen && !namelen) |
ab3aebc1 | 1185 | return 1; |
b5592632 NTND |
1186 | } |
1187 | ||
ab3aebc1 JK |
1188 | return fnmatch_icase_mem(pattern, patternlen, |
1189 | name, namelen, | |
f30366b2 | 1190 | WM_PATHNAME) == 0; |
b5592632 NTND |
1191 | } |
1192 | ||
578cd7c3 AS |
1193 | /* |
1194 | * Scan the given exclude list in reverse to see whether pathname | |
1195 | * should be ignored. The first match (i.e. the last on the list), if | |
1196 | * any, determines the fate. Returns the exclude_list element which | |
1197 | * matched, or NULL for undecided. | |
453ec4bd | 1198 | */ |
65edd96a | 1199 | static struct path_pattern *last_matching_pattern_from_list(const char *pathname, |
578cd7c3 AS |
1200 | int pathlen, |
1201 | const char *basename, | |
1202 | int *dtype, | |
caa3d554 | 1203 | struct pattern_list *pl, |
2b70e88d | 1204 | struct index_state *istate) |
453ec4bd | 1205 | { |
ab8db613 | 1206 | struct path_pattern *res = NULL; /* undecided */ |
5cee3493 | 1207 | int i; |
453ec4bd | 1208 | |
caa3d554 | 1209 | if (!pl->nr) |
578cd7c3 | 1210 | return NULL; /* undefined */ |
d6b8fc30 | 1211 | |
caa3d554 DS |
1212 | for (i = pl->nr - 1; 0 <= i; i--) { |
1213 | struct path_pattern *pattern = pl->patterns[i]; | |
ab8db613 DS |
1214 | const char *exclude = pattern->pattern; |
1215 | int prefix = pattern->nowildcardlen; | |
35a94d44 | 1216 | |
4ff89ee5 | 1217 | if (pattern->flags & PATTERN_FLAG_MUSTBEDIR) { |
35a94d44 | 1218 | if (*dtype == DT_UNKNOWN) |
2b70e88d | 1219 | *dtype = get_dtype(NULL, istate, pathname, pathlen); |
35a94d44 NTND |
1220 | if (*dtype != DT_DIR) |
1221 | continue; | |
1222 | } | |
d6b8fc30 | 1223 | |
4ff89ee5 | 1224 | if (pattern->flags & PATTERN_FLAG_NODIR) { |
593cb880 NTND |
1225 | if (match_basename(basename, |
1226 | pathlen - (basename - pathname), | |
ab8db613 DS |
1227 | exclude, prefix, pattern->patternlen, |
1228 | pattern->flags)) { | |
1229 | res = pattern; | |
e6efecc4 NTND |
1230 | break; |
1231 | } | |
35a94d44 NTND |
1232 | continue; |
1233 | } | |
1234 | ||
ab8db613 DS |
1235 | assert(pattern->baselen == 0 || |
1236 | pattern->base[pattern->baselen - 1] == '/'); | |
b5592632 | 1237 | if (match_pathname(pathname, pathlen, |
ab8db613 DS |
1238 | pattern->base, |
1239 | pattern->baselen ? pattern->baselen - 1 : 0, | |
1240 | exclude, prefix, pattern->patternlen, | |
1241 | pattern->flags)) { | |
1242 | res = pattern; | |
e6efecc4 NTND |
1243 | break; |
1244 | } | |
453ec4bd | 1245 | } |
ab8db613 | 1246 | return res; |
578cd7c3 AS |
1247 | } |
1248 | ||
1249 | /* | |
468ce99b DS |
1250 | * Scan the list of patterns to determine if the ordered list |
1251 | * of patterns matches on 'pathname'. | |
1252 | * | |
1253 | * Return 1 for a match, 0 for not matched and -1 for undecided. | |
578cd7c3 | 1254 | */ |
468ce99b DS |
1255 | enum pattern_match_result path_matches_pattern_list( |
1256 | const char *pathname, int pathlen, | |
1257 | const char *basename, int *dtype, | |
1258 | struct pattern_list *pl, | |
1259 | struct index_state *istate) | |
578cd7c3 | 1260 | { |
ab8db613 | 1261 | struct path_pattern *pattern; |
96cc8ab5 DS |
1262 | struct strbuf parent_pathname = STRBUF_INIT; |
1263 | int result = NOT_MATCHED; | |
1264 | const char *slash_pos; | |
1265 | ||
1266 | if (!pl->use_cone_patterns) { | |
1267 | pattern = last_matching_pattern_from_list(pathname, pathlen, basename, | |
1268 | dtype, pl, istate); | |
1269 | if (pattern) { | |
1270 | if (pattern->flags & PATTERN_FLAG_NEGATIVE) | |
1271 | return NOT_MATCHED; | |
1272 | else | |
1273 | return MATCHED; | |
1274 | } | |
1275 | ||
1276 | return UNDECIDED; | |
1277 | } | |
1278 | ||
1279 | if (pl->full_cone) | |
1280 | return MATCHED; | |
1281 | ||
1282 | strbuf_addch(&parent_pathname, '/'); | |
1283 | strbuf_add(&parent_pathname, pathname, pathlen); | |
1284 | ||
1285 | if (hashmap_contains_path(&pl->recursive_hashmap, | |
1286 | &parent_pathname)) { | |
eb42feca | 1287 | result = MATCHED_RECURSIVE; |
96cc8ab5 | 1288 | goto done; |
468ce99b DS |
1289 | } |
1290 | ||
96cc8ab5 DS |
1291 | slash_pos = strrchr(parent_pathname.buf, '/'); |
1292 | ||
1293 | if (slash_pos == parent_pathname.buf) { | |
1294 | /* include every file in root */ | |
1295 | result = MATCHED; | |
1296 | goto done; | |
1297 | } | |
1298 | ||
1299 | strbuf_setlen(&parent_pathname, slash_pos - parent_pathname.buf); | |
1300 | ||
1301 | if (hashmap_contains_path(&pl->parent_hashmap, &parent_pathname)) { | |
1302 | result = MATCHED; | |
1303 | goto done; | |
1304 | } | |
1305 | ||
1306 | if (hashmap_contains_parent(&pl->recursive_hashmap, | |
1307 | pathname, | |
1308 | &parent_pathname)) | |
eb42feca | 1309 | result = MATCHED_RECURSIVE; |
96cc8ab5 DS |
1310 | |
1311 | done: | |
1312 | strbuf_release(&parent_pathname); | |
1313 | return result; | |
453ec4bd LT |
1314 | } |
1315 | ||
65edd96a | 1316 | static struct path_pattern *last_matching_pattern_from_lists( |
ab8db613 DS |
1317 | struct dir_struct *dir, struct index_state *istate, |
1318 | const char *pathname, int pathlen, | |
1319 | const char *basename, int *dtype_p) | |
46aa2f95 KB |
1320 | { |
1321 | int i, j; | |
1322 | struct exclude_list_group *group; | |
ab8db613 | 1323 | struct path_pattern *pattern; |
46aa2f95 KB |
1324 | for (i = EXC_CMDL; i <= EXC_FILE; i++) { |
1325 | group = &dir->exclude_list_group[i]; | |
1326 | for (j = group->nr - 1; j >= 0; j--) { | |
65edd96a | 1327 | pattern = last_matching_pattern_from_list( |
46aa2f95 | 1328 | pathname, pathlen, basename, dtype_p, |
caa3d554 | 1329 | &group->pl[j], istate); |
ab8db613 DS |
1330 | if (pattern) |
1331 | return pattern; | |
46aa2f95 KB |
1332 | } |
1333 | } | |
1334 | return NULL; | |
1335 | } | |
1336 | ||
6cd5c582 KB |
1337 | /* |
1338 | * Loads the per-directory exclude list for the substring of base | |
1339 | * which has a char length of baselen. | |
1340 | */ | |
e799ed40 BW |
1341 | static void prep_exclude(struct dir_struct *dir, |
1342 | struct index_state *istate, | |
1343 | const char *base, int baselen) | |
6cd5c582 KB |
1344 | { |
1345 | struct exclude_list_group *group; | |
caa3d554 | 1346 | struct pattern_list *pl; |
6cd5c582 | 1347 | struct exclude_stack *stk = NULL; |
0dcb8d7f | 1348 | struct untracked_cache_dir *untracked; |
6cd5c582 KB |
1349 | int current; |
1350 | ||
6cd5c582 KB |
1351 | group = &dir->exclude_list_group[EXC_DIRS]; |
1352 | ||
d961baa8 NTND |
1353 | /* |
1354 | * Pop the exclude lists from the EXCL_DIRS exclude_list_group | |
6cd5c582 | 1355 | * which originate from directories not in the prefix of the |
d961baa8 NTND |
1356 | * path being checked. |
1357 | */ | |
6cd5c582 KB |
1358 | while ((stk = dir->exclude_stack) != NULL) { |
1359 | if (stk->baselen <= baselen && | |
aceb9429 | 1360 | !strncmp(dir->basebuf.buf, base, stk->baselen)) |
6cd5c582 | 1361 | break; |
caa3d554 | 1362 | pl = &group->pl[dir->exclude_stack->exclude_ix]; |
6cd5c582 | 1363 | dir->exclude_stack = stk->prev; |
ab8db613 | 1364 | dir->pattern = NULL; |
caa3d554 | 1365 | free((char *)pl->src); /* see strbuf_detach() below */ |
65edd96a | 1366 | clear_pattern_list(pl); |
6cd5c582 KB |
1367 | free(stk); |
1368 | group->nr--; | |
1369 | } | |
1370 | ||
95c6f271 | 1371 | /* Skip traversing into sub directories if the parent is excluded */ |
ab8db613 | 1372 | if (dir->pattern) |
95c6f271 KB |
1373 | return; |
1374 | ||
aceb9429 NTND |
1375 | /* |
1376 | * Lazy initialization. All call sites currently just | |
1377 | * memset(dir, 0, sizeof(*dir)) before use. Changing all of | |
1378 | * them seems lots of work for little benefit. | |
1379 | */ | |
1380 | if (!dir->basebuf.buf) | |
1381 | strbuf_init(&dir->basebuf, PATH_MAX); | |
1382 | ||
6cd5c582 KB |
1383 | /* Read from the parent directories and push them down. */ |
1384 | current = stk ? stk->baselen : -1; | |
aceb9429 | 1385 | strbuf_setlen(&dir->basebuf, current < 0 ? 0 : current); |
0dcb8d7f NTND |
1386 | if (dir->untracked) |
1387 | untracked = stk ? stk->ucd : dir->untracked->root; | |
1388 | else | |
1389 | untracked = NULL; | |
1390 | ||
6cd5c582 | 1391 | while (current < baselen) { |
6cd5c582 | 1392 | const char *cp; |
4b33e602 | 1393 | struct oid_stat oid_stat; |
6cd5c582 | 1394 | |
03e11a71 | 1395 | stk = xcalloc(1, sizeof(*stk)); |
6cd5c582 KB |
1396 | if (current < 0) { |
1397 | cp = base; | |
1398 | current = 0; | |
d961baa8 | 1399 | } else { |
6cd5c582 KB |
1400 | cp = strchr(base + current + 1, '/'); |
1401 | if (!cp) | |
1402 | die("oops in prep_exclude"); | |
1403 | cp++; | |
0dcb8d7f NTND |
1404 | untracked = |
1405 | lookup_untracked(dir->untracked, untracked, | |
1406 | base + current, | |
1407 | cp - base - current); | |
6cd5c582 KB |
1408 | } |
1409 | stk->prev = dir->exclude_stack; | |
1410 | stk->baselen = cp - base; | |
95c6f271 | 1411 | stk->exclude_ix = group->nr; |
0dcb8d7f | 1412 | stk->ucd = untracked; |
65edd96a | 1413 | pl = add_pattern_list(dir, EXC_DIRS, NULL); |
aceb9429 NTND |
1414 | strbuf_add(&dir->basebuf, base + current, stk->baselen - current); |
1415 | assert(stk->baselen == dir->basebuf.len); | |
95c6f271 KB |
1416 | |
1417 | /* Abort if the directory is excluded */ | |
1418 | if (stk->baselen) { | |
1419 | int dt = DT_DIR; | |
aceb9429 | 1420 | dir->basebuf.buf[stk->baselen - 1] = 0; |
65edd96a | 1421 | dir->pattern = last_matching_pattern_from_lists(dir, |
e799ed40 | 1422 | istate, |
aceb9429 NTND |
1423 | dir->basebuf.buf, stk->baselen - 1, |
1424 | dir->basebuf.buf + current, &dt); | |
1425 | dir->basebuf.buf[stk->baselen - 1] = '/'; | |
ab8db613 | 1426 | if (dir->pattern && |
4ff89ee5 | 1427 | dir->pattern->flags & PATTERN_FLAG_NEGATIVE) |
ab8db613 DS |
1428 | dir->pattern = NULL; |
1429 | if (dir->pattern) { | |
95c6f271 KB |
1430 | dir->exclude_stack = stk; |
1431 | return; | |
1432 | } | |
1433 | } | |
1434 | ||
aceb9429 | 1435 | /* Try to read per-directory file */ |
4b33e602 PO |
1436 | oidclr(&oid_stat.oid); |
1437 | oid_stat.valid = 0; | |
27b099ae NTND |
1438 | if (dir->exclude_per_dir && |
1439 | /* | |
1440 | * If we know that no files have been added in | |
1441 | * this directory (i.e. valid_cached_dir() has | |
1442 | * been executed and set untracked->valid) .. | |
1443 | */ | |
1444 | (!untracked || !untracked->valid || | |
1445 | /* | |
1446 | * .. and .gitignore does not exist before | |
70c369cd | 1447 | * (i.e. null exclude_oid). Then we can skip |
7687252f DT |
1448 | * loading .gitignore, which would result in |
1449 | * ENOENT anyway. | |
27b099ae | 1450 | */ |
70c369cd | 1451 | !is_null_oid(&untracked->exclude_oid))) { |
95c6f271 KB |
1452 | /* |
1453 | * dir->basebuf gets reused by the traversal, but we | |
1454 | * need fname to remain unchanged to ensure the src | |
ab8db613 | 1455 | * member of each struct path_pattern correctly |
95c6f271 | 1456 | * back-references its source file. Other invocations |
65edd96a | 1457 | * of add_pattern_list provide stable strings, so we |
aceb9429 | 1458 | * strbuf_detach() and free() here in the caller. |
95c6f271 | 1459 | */ |
aceb9429 NTND |
1460 | struct strbuf sb = STRBUF_INIT; |
1461 | strbuf_addbuf(&sb, &dir->basebuf); | |
1462 | strbuf_addstr(&sb, dir->exclude_per_dir); | |
caa3d554 | 1463 | pl->src = strbuf_detach(&sb, NULL); |
65edd96a | 1464 | add_patterns(pl->src, pl->src, stk->baselen, pl, istate, |
4b33e602 | 1465 | untracked ? &oid_stat : NULL); |
0dcb8d7f | 1466 | } |
5ebf79ad NTND |
1467 | /* |
1468 | * NEEDSWORK: when untracked cache is enabled, prep_exclude() | |
1469 | * will first be called in valid_cached_dir() then maybe many | |
65edd96a DS |
1470 | * times more in last_matching_pattern(). When the cache is |
1471 | * used, last_matching_pattern() will not be called and | |
5ebf79ad NTND |
1472 | * reading .gitignore content will be a waste. |
1473 | * | |
1474 | * So when it's called by valid_cached_dir() and we can get | |
1475 | * .gitignore SHA-1 from the index (i.e. .gitignore is not | |
1476 | * modified on work tree), we could delay reading the | |
1477 | * .gitignore content until we absolutely need it in | |
65edd96a | 1478 | * last_matching_pattern(). Be careful about ignore rule |
5ebf79ad NTND |
1479 | * order, though, if you do that. |
1480 | */ | |
1481 | if (untracked && | |
9001dc2a | 1482 | !oideq(&oid_stat.oid, &untracked->exclude_oid)) { |
5ebf79ad | 1483 | invalidate_gitignore(dir->untracked, untracked); |
70c369cd | 1484 | oidcpy(&untracked->exclude_oid, &oid_stat.oid); |
95c6f271 | 1485 | } |
6cd5c582 KB |
1486 | dir->exclude_stack = stk; |
1487 | current = stk->baselen; | |
1488 | } | |
aceb9429 | 1489 | strbuf_setlen(&dir->basebuf, baselen); |
6cd5c582 KB |
1490 | } |
1491 | ||
f4cd69a6 AS |
1492 | /* |
1493 | * Loads the exclude lists for the directory containing pathname, then | |
1494 | * scans all exclude lists to determine whether pathname is excluded. | |
1495 | * Returns the exclude_list element which matched, or NULL for | |
1496 | * undecided. | |
1497 | */ | |
65edd96a | 1498 | struct path_pattern *last_matching_pattern(struct dir_struct *dir, |
a0bba65b BW |
1499 | struct index_state *istate, |
1500 | const char *pathname, | |
1501 | int *dtype_p) | |
453ec4bd LT |
1502 | { |
1503 | int pathlen = strlen(pathname); | |
68492fc7 LK |
1504 | const char *basename = strrchr(pathname, '/'); |
1505 | basename = (basename) ? basename+1 : pathname; | |
453ec4bd | 1506 | |
a0bba65b | 1507 | prep_exclude(dir, istate, pathname, basename-pathname); |
c082df24 | 1508 | |
ab8db613 DS |
1509 | if (dir->pattern) |
1510 | return dir->pattern; | |
95c6f271 | 1511 | |
65edd96a | 1512 | return last_matching_pattern_from_lists(dir, istate, pathname, pathlen, |
46aa2f95 | 1513 | basename, dtype_p); |
f4cd69a6 AS |
1514 | } |
1515 | ||
1516 | /* | |
1517 | * Loads the exclude lists for the directory containing pathname, then | |
1518 | * scans all exclude lists to determine whether pathname is excluded. | |
1519 | * Returns 1 if true, otherwise 0. | |
1520 | */ | |
a0bba65b BW |
1521 | int is_excluded(struct dir_struct *dir, struct index_state *istate, |
1522 | const char *pathname, int *dtype_p) | |
f4cd69a6 | 1523 | { |
ab8db613 | 1524 | struct path_pattern *pattern = |
65edd96a | 1525 | last_matching_pattern(dir, istate, pathname, dtype_p); |
ab8db613 | 1526 | if (pattern) |
4ff89ee5 | 1527 | return pattern->flags & PATTERN_FLAG_NEGATIVE ? 0 : 1; |
453ec4bd LT |
1528 | return 0; |
1529 | } | |
1530 | ||
f3fa1838 JH |
1531 | static struct dir_entry *dir_entry_new(const char *pathname, int len) |
1532 | { | |
453ec4bd LT |
1533 | struct dir_entry *ent; |
1534 | ||
96ffc06f | 1535 | FLEX_ALLOC_MEM(ent, name, pathname, len); |
453ec4bd | 1536 | ent->len = len; |
4d06f8ac | 1537 | return ent; |
453ec4bd LT |
1538 | } |
1539 | ||
9e58beca BW |
1540 | static struct dir_entry *dir_add_name(struct dir_struct *dir, |
1541 | struct index_state *istate, | |
1542 | const char *pathname, int len) | |
6815e569 | 1543 | { |
9e58beca | 1544 | if (index_file_exists(istate, pathname, len, ignore_case)) |
6815e569 JK |
1545 | return NULL; |
1546 | ||
25fd2f7a | 1547 | ALLOC_GROW(dir->entries, dir->nr+1, dir->alloc); |
6815e569 JK |
1548 | return dir->entries[dir->nr++] = dir_entry_new(pathname, len); |
1549 | } | |
1550 | ||
9e58beca BW |
1551 | struct dir_entry *dir_add_ignored(struct dir_struct *dir, |
1552 | struct index_state *istate, | |
1553 | const char *pathname, int len) | |
2abd31b0 | 1554 | { |
9e58beca | 1555 | if (!index_name_is_other(istate, pathname, len)) |
2abd31b0 JK |
1556 | return NULL; |
1557 | ||
25fd2f7a | 1558 | ALLOC_GROW(dir->ignored, dir->ignored_nr+1, dir->ignored_alloc); |
2abd31b0 JK |
1559 | return dir->ignored[dir->ignored_nr++] = dir_entry_new(pathname, len); |
1560 | } | |
1561 | ||
09595258 LT |
1562 | enum exist_status { |
1563 | index_nonexistent = 0, | |
1564 | index_directory, | |
4b05548f | 1565 | index_gitdir |
09595258 LT |
1566 | }; |
1567 | ||
5102c617 | 1568 | /* |
71261027 | 1569 | * Do not use the alphabetically sorted index to look up |
5102c617 | 1570 | * the directory name; instead, use the case insensitive |
ebbd7439 | 1571 | * directory hash. |
5102c617 | 1572 | */ |
ae520e36 BW |
1573 | static enum exist_status directory_exists_in_index_icase(struct index_state *istate, |
1574 | const char *dirname, int len) | |
5102c617 | 1575 | { |
41284eb0 | 1576 | struct cache_entry *ce; |
5102c617 | 1577 | |
ae520e36 | 1578 | if (index_dir_exists(istate, dirname, len)) |
5102c617 JJ |
1579 | return index_directory; |
1580 | ||
ae520e36 | 1581 | ce = index_file_exists(istate, dirname, len, ignore_case); |
41284eb0 | 1582 | if (ce && S_ISGITLINK(ce->ce_mode)) |
5102c617 JJ |
1583 | return index_gitdir; |
1584 | ||
5102c617 JJ |
1585 | return index_nonexistent; |
1586 | } | |
1587 | ||
09595258 LT |
1588 | /* |
1589 | * The index sorts alphabetically by entry name, which | |
1590 | * means that a gitlink sorts as '\0' at the end, while | |
1591 | * a directory (which is defined not as an entry, but as | |
1592 | * the files it contains) will sort with the '/' at the | |
1593 | * end. | |
1594 | */ | |
ae520e36 BW |
1595 | static enum exist_status directory_exists_in_index(struct index_state *istate, |
1596 | const char *dirname, int len) | |
453ec4bd | 1597 | { |
5102c617 JJ |
1598 | int pos; |
1599 | ||
1600 | if (ignore_case) | |
ae520e36 | 1601 | return directory_exists_in_index_icase(istate, dirname, len); |
5102c617 | 1602 | |
ae520e36 | 1603 | pos = index_name_pos(istate, dirname, len); |
09595258 LT |
1604 | if (pos < 0) |
1605 | pos = -pos-1; | |
ae520e36 BW |
1606 | while (pos < istate->cache_nr) { |
1607 | const struct cache_entry *ce = istate->cache[pos++]; | |
09595258 LT |
1608 | unsigned char endchar; |
1609 | ||
1610 | if (strncmp(ce->name, dirname, len)) | |
1611 | break; | |
1612 | endchar = ce->name[len]; | |
1613 | if (endchar > '/') | |
1614 | break; | |
1615 | if (endchar == '/') | |
1616 | return index_directory; | |
7a51ed66 | 1617 | if (!endchar && S_ISGITLINK(ce->ce_mode)) |
09595258 LT |
1618 | return index_gitdir; |
1619 | } | |
1620 | return index_nonexistent; | |
1621 | } | |
1622 | ||
1623 | /* | |
1624 | * When we find a directory when traversing the filesystem, we | |
1625 | * have three distinct cases: | |
1626 | * | |
1627 | * - ignore it | |
1628 | * - see it as a directory | |
1629 | * - recurse into it | |
1630 | * | |
1631 | * and which one we choose depends on a combination of existing | |
1632 | * git index contents and the flags passed into the directory | |
1633 | * traversal routine. | |
1634 | * | |
1635 | * Case 1: If we *already* have entries in the index under that | |
5bd8e2d8 KB |
1636 | * directory name, we always recurse into the directory to see |
1637 | * all the files. | |
09595258 LT |
1638 | * |
1639 | * Case 2: If we *already* have that directory name as a gitlink, | |
1640 | * we always continue to see it as a gitlink, regardless of whether | |
1641 | * there is an actual git directory there or not (it might not | |
1642 | * be checked out as a subproject!) | |
1643 | * | |
1644 | * Case 3: if we didn't have it in the index previously, we | |
1645 | * have a few sub-cases: | |
1646 | * | |
1647 | * (a) if "show_other_directories" is true, we show it as | |
1648 | * just a directory, unless "hide_empty_directories" is | |
defd7c7b KB |
1649 | * also true, in which case we need to check if it contains any |
1650 | * untracked and / or ignored files. | |
09595258 | 1651 | * (b) if it looks like a git directory, and we don't have |
302b9282 | 1652 | * 'no_gitlinks' set we treat it as a gitlink, and show it |
09595258 LT |
1653 | * as a directory. |
1654 | * (c) otherwise, we recurse into it. | |
1655 | */ | |
defd7c7b | 1656 | static enum path_treatment treat_directory(struct dir_struct *dir, |
0ef8e169 | 1657 | struct index_state *istate, |
0dcb8d7f | 1658 | struct untracked_cache_dir *untracked, |
2e5910f2 | 1659 | const char *dirname, int len, int baselen, int exclude, |
e1b8c7bd | 1660 | const struct pathspec *pathspec) |
09595258 | 1661 | { |
c847dfaf EN |
1662 | int nested_repo = 0; |
1663 | ||
09595258 | 1664 | /* The "len-1" is to strip the final '/' */ |
0ef8e169 | 1665 | switch (directory_exists_in_index(istate, dirname, len-1)) { |
09595258 | 1666 | case index_directory: |
defd7c7b | 1667 | return path_recurse; |
09595258 LT |
1668 | |
1669 | case index_gitdir: | |
26c986e1 | 1670 | return path_none; |
09595258 LT |
1671 | |
1672 | case index_nonexistent: | |
c847dfaf EN |
1673 | if ((dir->flags & DIR_SKIP_NESTED_GIT) || |
1674 | !(dir->flags & DIR_NO_GITLINKS)) { | |
09487f2c EN |
1675 | struct strbuf sb = STRBUF_INIT; |
1676 | strbuf_addstr(&sb, dirname); | |
1677 | nested_repo = is_nonbare_repository_dir(&sb); | |
1678 | strbuf_release(&sb); | |
09487f2c | 1679 | } |
c847dfaf EN |
1680 | if (nested_repo) |
1681 | return ((dir->flags & DIR_SKIP_NESTED_GIT) ? path_none : | |
1682 | (exclude ? path_excluded : path_untracked)); | |
09487f2c | 1683 | |
7c4c97c0 | 1684 | if (dir->flags & DIR_SHOW_OTHER_DIRECTORIES) |
09595258 | 1685 | break; |
eec0f7f2 JM |
1686 | if (exclude && |
1687 | (dir->flags & DIR_SHOW_IGNORED_TOO) && | |
1688 | (dir->flags & DIR_SHOW_IGNORED_TOO_MODE_MATCHING)) { | |
1689 | ||
1690 | /* | |
1691 | * This is an excluded directory and we are | |
1692 | * showing ignored paths that match an exclude | |
1693 | * pattern. (e.g. show directory as ignored | |
1694 | * only if it matches an exclude pattern). | |
1695 | * This path will either be 'path_excluded` | |
1696 | * (if we are showing empty directories or if | |
1697 | * the directory is not empty), or will be | |
1698 | * 'path_none' (empty directory, and we are | |
1699 | * not showing empty directories). | |
1700 | */ | |
1701 | if (!(dir->flags & DIR_HIDE_EMPTY_DIRECTORIES)) | |
1702 | return path_excluded; | |
1703 | ||
1704 | if (read_directory_recursive(dir, istate, dirname, len, | |
1705 | untracked, 1, 1, pathspec) == path_excluded) | |
1706 | return path_excluded; | |
1707 | ||
1708 | return path_none; | |
1709 | } | |
defd7c7b | 1710 | return path_recurse; |
09595258 LT |
1711 | } |
1712 | ||
1713 | /* This is the "show_other_directories" case */ | |
721ac4ed | 1714 | |
184d2a8e | 1715 | if (!(dir->flags & DIR_HIDE_EMPTY_DIRECTORIES)) |
defd7c7b KB |
1716 | return exclude ? path_excluded : path_untracked; |
1717 | ||
2e5910f2 DT |
1718 | untracked = lookup_untracked(dir->untracked, untracked, |
1719 | dirname + baselen, len - baselen); | |
5aaa7fd3 JM |
1720 | |
1721 | /* | |
1722 | * If this is an excluded directory, then we only need to check if | |
1723 | * the directory contains any files. | |
1724 | */ | |
0ef8e169 | 1725 | return read_directory_recursive(dir, istate, dirname, len, |
5aaa7fd3 | 1726 | untracked, 1, exclude, pathspec); |
453ec4bd LT |
1727 | } |
1728 | ||
9fc42d60 LT |
1729 | /* |
1730 | * This is an inexact early pruning of any recursive directory | |
1731 | * reading - if the path cannot possibly be in the pathspec, | |
1732 | * return true, and we'll skip it early. | |
1733 | */ | |
e1b8c7bd BW |
1734 | static int simplify_away(const char *path, int pathlen, |
1735 | const struct pathspec *pathspec) | |
9fc42d60 | 1736 | { |
e1b8c7bd | 1737 | int i; |
9fc42d60 | 1738 | |
e1b8c7bd BW |
1739 | if (!pathspec || !pathspec->nr) |
1740 | return 0; | |
1741 | ||
1742 | GUARD_PATHSPEC(pathspec, | |
1743 | PATHSPEC_FROMTOP | | |
1744 | PATHSPEC_MAXDEPTH | | |
1745 | PATHSPEC_LITERAL | | |
1746 | PATHSPEC_GLOB | | |
1747 | PATHSPEC_ICASE | | |
b0db7046 BW |
1748 | PATHSPEC_EXCLUDE | |
1749 | PATHSPEC_ATTR); | |
e1b8c7bd BW |
1750 | |
1751 | for (i = 0; i < pathspec->nr; i++) { | |
1752 | const struct pathspec_item *item = &pathspec->items[i]; | |
1753 | int len = item->nowildcard_len; | |
1754 | ||
1755 | if (len > pathlen) | |
1756 | len = pathlen; | |
1757 | if (!ps_strncmp(item, item->match, path, len)) | |
1758 | return 0; | |
9fc42d60 | 1759 | } |
e1b8c7bd BW |
1760 | |
1761 | return 1; | |
9fc42d60 LT |
1762 | } |
1763 | ||
29209cbe JK |
1764 | /* |
1765 | * This function tells us whether an excluded path matches a | |
1766 | * list of "interesting" pathspecs. That is, whether a path matched | |
1767 | * by any of the pathspecs could possibly be ignored by excluding | |
1768 | * the specified path. This can happen if: | |
1769 | * | |
1770 | * 1. the path is mentioned explicitly in the pathspec | |
1771 | * | |
1772 | * 2. the path is a directory prefix of some element in the | |
1773 | * pathspec | |
1774 | */ | |
e1b8c7bd BW |
1775 | static int exclude_matches_pathspec(const char *path, int pathlen, |
1776 | const struct pathspec *pathspec) | |
1777 | { | |
1778 | int i; | |
1779 | ||
1780 | if (!pathspec || !pathspec->nr) | |
1781 | return 0; | |
1782 | ||
1783 | GUARD_PATHSPEC(pathspec, | |
1784 | PATHSPEC_FROMTOP | | |
1785 | PATHSPEC_MAXDEPTH | | |
1786 | PATHSPEC_LITERAL | | |
1787 | PATHSPEC_GLOB | | |
1788 | PATHSPEC_ICASE | | |
1789 | PATHSPEC_EXCLUDE); | |
1790 | ||
1791 | for (i = 0; i < pathspec->nr; i++) { | |
1792 | const struct pathspec_item *item = &pathspec->items[i]; | |
1793 | int len = item->nowildcard_len; | |
1794 | ||
1795 | if (len == pathlen && | |
1796 | !ps_strncmp(item, item->match, path, pathlen)) | |
1797 | return 1; | |
1798 | if (len > pathlen && | |
1799 | item->match[pathlen] == '/' && | |
1800 | !ps_strncmp(item, item->match, path, pathlen)) | |
1801 | return 1; | |
e96980ef JK |
1802 | } |
1803 | return 0; | |
1804 | } | |
1805 | ||
98f2a687 BW |
1806 | static int get_index_dtype(struct index_state *istate, |
1807 | const char *path, int len) | |
443e061a LT |
1808 | { |
1809 | int pos; | |
9c5e6c80 | 1810 | const struct cache_entry *ce; |
443e061a | 1811 | |
98f2a687 | 1812 | ce = index_file_exists(istate, path, len, 0); |
443e061a LT |
1813 | if (ce) { |
1814 | if (!ce_uptodate(ce)) | |
1815 | return DT_UNKNOWN; | |
1816 | if (S_ISGITLINK(ce->ce_mode)) | |
1817 | return DT_DIR; | |
1818 | /* | |
1819 | * Nobody actually cares about the | |
1820 | * difference between DT_LNK and DT_REG | |
1821 | */ | |
1822 | return DT_REG; | |
1823 | } | |
1824 | ||
1825 | /* Try to look it up as a directory */ | |
98f2a687 | 1826 | pos = index_name_pos(istate, path, len); |
443e061a LT |
1827 | if (pos >= 0) |
1828 | return DT_UNKNOWN; | |
1829 | pos = -pos-1; | |
98f2a687 BW |
1830 | while (pos < istate->cache_nr) { |
1831 | ce = istate->cache[pos++]; | |
443e061a LT |
1832 | if (strncmp(ce->name, path, len)) |
1833 | break; | |
1834 | if (ce->name[len] > '/') | |
1835 | break; | |
1836 | if (ce->name[len] < '/') | |
1837 | continue; | |
1838 | if (!ce_uptodate(ce)) | |
1839 | break; /* continue? */ | |
1840 | return DT_DIR; | |
1841 | } | |
1842 | return DT_UNKNOWN; | |
1843 | } | |
1844 | ||
98f2a687 BW |
1845 | static int get_dtype(struct dirent *de, struct index_state *istate, |
1846 | const char *path, int len) | |
07134421 | 1847 | { |
6831a88a | 1848 | int dtype = de ? DTYPE(de) : DT_UNKNOWN; |
07134421 LT |
1849 | struct stat st; |
1850 | ||
1851 | if (dtype != DT_UNKNOWN) | |
1852 | return dtype; | |
98f2a687 | 1853 | dtype = get_index_dtype(istate, path, len); |
443e061a LT |
1854 | if (dtype != DT_UNKNOWN) |
1855 | return dtype; | |
1856 | if (lstat(path, &st)) | |
07134421 LT |
1857 | return dtype; |
1858 | if (S_ISREG(st.st_mode)) | |
1859 | return DT_REG; | |
1860 | if (S_ISDIR(st.st_mode)) | |
1861 | return DT_DIR; | |
1862 | if (S_ISLNK(st.st_mode)) | |
1863 | return DT_LNK; | |
1864 | return dtype; | |
1865 | } | |
1866 | ||
16e2cfa9 | 1867 | static enum path_treatment treat_one_path(struct dir_struct *dir, |
0dcb8d7f | 1868 | struct untracked_cache_dir *untracked, |
0ef8e169 | 1869 | struct index_state *istate, |
49dc2cc2 | 1870 | struct strbuf *path, |
2e5910f2 | 1871 | int baselen, |
e1b8c7bd | 1872 | const struct pathspec *pathspec, |
16e2cfa9 | 1873 | int dtype, struct dirent *de) |
53cc5356 | 1874 | { |
8aaf8d77 | 1875 | int exclude; |
0ef8e169 | 1876 | int has_path_in_index = !!index_file_exists(istate, path->buf, path->len, ignore_case); |
07966ed1 | 1877 | enum path_treatment path_treatment; |
2eac2a4c | 1878 | |
8aaf8d77 | 1879 | if (dtype == DT_UNKNOWN) |
0ef8e169 | 1880 | dtype = get_dtype(de, istate, path->buf, path->len); |
8aaf8d77 KB |
1881 | |
1882 | /* Always exclude indexed files */ | |
2eac2a4c | 1883 | if (dtype != DT_DIR && has_path_in_index) |
defd7c7b | 1884 | return path_none; |
8aaf8d77 | 1885 | |
2eac2a4c JH |
1886 | /* |
1887 | * When we are looking at a directory P in the working tree, | |
1888 | * there are three cases: | |
1889 | * | |
1890 | * (1) P exists in the index. Everything inside the directory P in | |
1891 | * the working tree needs to go when P is checked out from the | |
1892 | * index. | |
1893 | * | |
1894 | * (2) P does not exist in the index, but there is P/Q in the index. | |
1895 | * We know P will stay a directory when we check out the contents | |
1896 | * of the index, but we do not know yet if there is a directory | |
1897 | * P/Q in the working tree to be killed, so we need to recurse. | |
1898 | * | |
1899 | * (3) P does not exist in the index, and there is no P/Q in the index | |
1900 | * to require P to be a directory, either. Only in this case, we | |
1901 | * know that everything inside P will not be killed without | |
1902 | * recursing. | |
1903 | */ | |
1904 | if ((dir->flags & DIR_COLLECT_KILLED_ONLY) && | |
1905 | (dtype == DT_DIR) && | |
de372b1b | 1906 | !has_path_in_index && |
0ef8e169 | 1907 | (directory_exists_in_index(istate, path->buf, path->len) == index_nonexistent)) |
de372b1b | 1908 | return path_none; |
8aaf8d77 | 1909 | |
0ef8e169 | 1910 | exclude = is_excluded(dir, istate, path->buf, &dtype); |
53cc5356 JH |
1911 | |
1912 | /* | |
1913 | * Excluded? If we don't explicitly want to show | |
1914 | * ignored files, ignore it | |
1915 | */ | |
0aaf62b6 | 1916 | if (exclude && !(dir->flags & (DIR_SHOW_IGNORED|DIR_SHOW_IGNORED_TOO))) |
defd7c7b | 1917 | return path_excluded; |
53cc5356 | 1918 | |
53cc5356 JH |
1919 | switch (dtype) { |
1920 | default: | |
defd7c7b | 1921 | return path_none; |
53cc5356 | 1922 | case DT_DIR: |
49dc2cc2 | 1923 | strbuf_addch(path, '/'); |
07966ed1 JM |
1924 | path_treatment = treat_directory(dir, istate, untracked, |
1925 | path->buf, path->len, | |
1926 | baselen, exclude, pathspec); | |
1927 | /* | |
1928 | * If 1) we only want to return directories that | |
1929 | * match an exclude pattern and 2) this directory does | |
1930 | * not match an exclude pattern but all of its | |
1931 | * contents are excluded, then indicate that we should | |
1932 | * recurse into this directory (instead of marking the | |
1933 | * directory itself as an ignored path). | |
1934 | */ | |
1935 | if (!exclude && | |
1936 | path_treatment == path_excluded && | |
1937 | (dir->flags & DIR_SHOW_IGNORED_TOO) && | |
1938 | (dir->flags & DIR_SHOW_IGNORED_TOO_MODE_MATCHING)) | |
1939 | return path_recurse; | |
1940 | return path_treatment; | |
53cc5356 JH |
1941 | case DT_REG: |
1942 | case DT_LNK: | |
defd7c7b | 1943 | return exclude ? path_excluded : path_untracked; |
53cc5356 | 1944 | } |
53cc5356 JH |
1945 | } |
1946 | ||
91a2288b NTND |
1947 | static enum path_treatment treat_path_fast(struct dir_struct *dir, |
1948 | struct untracked_cache_dir *untracked, | |
1949 | struct cached_dir *cdir, | |
0ef8e169 | 1950 | struct index_state *istate, |
91a2288b NTND |
1951 | struct strbuf *path, |
1952 | int baselen, | |
e1b8c7bd | 1953 | const struct pathspec *pathspec) |
91a2288b NTND |
1954 | { |
1955 | strbuf_setlen(path, baselen); | |
1956 | if (!cdir->ucd) { | |
1957 | strbuf_addstr(path, cdir->file); | |
1958 | return path_untracked; | |
1959 | } | |
1960 | strbuf_addstr(path, cdir->ucd->name); | |
1961 | /* treat_one_path() does this before it calls treat_directory() */ | |
00b6c178 | 1962 | strbuf_complete(path, '/'); |
91a2288b NTND |
1963 | if (cdir->ucd->check_only) |
1964 | /* | |
1965 | * check_only is set as a result of treat_directory() getting | |
1966 | * to its bottom. Verify again the same set of directories | |
1967 | * with check_only set. | |
1968 | */ | |
0ef8e169 | 1969 | return read_directory_recursive(dir, istate, path->buf, path->len, |
5aaa7fd3 | 1970 | cdir->ucd, 1, 0, pathspec); |
91a2288b NTND |
1971 | /* |
1972 | * We get path_recurse in the first run when | |
1973 | * directory_exists_in_index() returns index_nonexistent. We | |
1974 | * are sure that new changes in the index does not impact the | |
1975 | * outcome. Return now. | |
1976 | */ | |
1977 | return path_recurse; | |
1978 | } | |
1979 | ||
16e2cfa9 | 1980 | static enum path_treatment treat_path(struct dir_struct *dir, |
0dcb8d7f | 1981 | struct untracked_cache_dir *untracked, |
cf7c6148 | 1982 | struct cached_dir *cdir, |
0ef8e169 | 1983 | struct index_state *istate, |
49dc2cc2 | 1984 | struct strbuf *path, |
16e2cfa9 | 1985 | int baselen, |
e1b8c7bd | 1986 | const struct pathspec *pathspec) |
16e2cfa9 JH |
1987 | { |
1988 | int dtype; | |
cf7c6148 | 1989 | struct dirent *de = cdir->de; |
16e2cfa9 | 1990 | |
91a2288b | 1991 | if (!de) |
0ef8e169 | 1992 | return treat_path_fast(dir, untracked, cdir, istate, path, |
e1b8c7bd | 1993 | baselen, pathspec); |
0cacebf0 | 1994 | if (is_dot_or_dotdot(de->d_name) || !fspathcmp(de->d_name, ".git")) |
defd7c7b | 1995 | return path_none; |
49dc2cc2 RS |
1996 | strbuf_setlen(path, baselen); |
1997 | strbuf_addstr(path, de->d_name); | |
e1b8c7bd | 1998 | if (simplify_away(path->buf, path->len, pathspec)) |
defd7c7b | 1999 | return path_none; |
16e2cfa9 JH |
2000 | |
2001 | dtype = DTYPE(de); | |
0ef8e169 | 2002 | return treat_one_path(dir, untracked, istate, path, baselen, pathspec, dtype, de); |
0dcb8d7f NTND |
2003 | } |
2004 | ||
2005 | static void add_untracked(struct untracked_cache_dir *dir, const char *name) | |
2006 | { | |
2007 | if (!dir) | |
2008 | return; | |
2009 | ALLOC_GROW(dir->untracked, dir->untracked_nr + 1, | |
2010 | dir->untracked_alloc); | |
2011 | dir->untracked[dir->untracked_nr++] = xstrdup(name); | |
16e2cfa9 JH |
2012 | } |
2013 | ||
91a2288b NTND |
2014 | static int valid_cached_dir(struct dir_struct *dir, |
2015 | struct untracked_cache_dir *untracked, | |
207a06ce | 2016 | struct index_state *istate, |
91a2288b NTND |
2017 | struct strbuf *path, |
2018 | int check_only) | |
2019 | { | |
2020 | struct stat st; | |
2021 | ||
2022 | if (!untracked) | |
2023 | return 0; | |
2024 | ||
883e248b BP |
2025 | /* |
2026 | * With fsmonitor, we can trust the untracked cache's valid field. | |
2027 | */ | |
2028 | refresh_fsmonitor(istate); | |
2029 | if (!(dir->untracked->use_fsmonitor && untracked->valid)) { | |
2523c4be | 2030 | if (lstat(path->len ? path->buf : ".", &st)) { |
883e248b BP |
2031 | memset(&untracked->stat_data, 0, sizeof(untracked->stat_data)); |
2032 | return 0; | |
2033 | } | |
2034 | if (!untracked->valid || | |
2035 | match_stat_data_racy(istate, &untracked->stat_data, &st)) { | |
883e248b BP |
2036 | fill_stat_data(&untracked->stat_data, &st); |
2037 | return 0; | |
2038 | } | |
91a2288b NTND |
2039 | } |
2040 | ||
b6403131 | 2041 | if (untracked->check_only != !!check_only) |
91a2288b | 2042 | return 0; |
91a2288b NTND |
2043 | |
2044 | /* | |
2045 | * prep_exclude will be called eventually on this directory, | |
65edd96a | 2046 | * but it's called much later in last_matching_pattern(). We |
91a2288b NTND |
2047 | * need it now to determine the validity of the cache for this |
2048 | * path. The next calls will be nearly no-op, the way | |
2049 | * prep_exclude() is designed. | |
2050 | */ | |
2051 | if (path->len && path->buf[path->len - 1] != '/') { | |
2052 | strbuf_addch(path, '/'); | |
207a06ce | 2053 | prep_exclude(dir, istate, path->buf, path->len); |
91a2288b NTND |
2054 | strbuf_setlen(path, path->len - 1); |
2055 | } else | |
207a06ce | 2056 | prep_exclude(dir, istate, path->buf, path->len); |
91a2288b NTND |
2057 | |
2058 | /* hopefully prep_exclude() haven't invalidated this entry... */ | |
2059 | return untracked->valid; | |
2060 | } | |
2061 | ||
cf7c6148 NTND |
2062 | static int open_cached_dir(struct cached_dir *cdir, |
2063 | struct dir_struct *dir, | |
2064 | struct untracked_cache_dir *untracked, | |
207a06ce | 2065 | struct index_state *istate, |
cf7c6148 NTND |
2066 | struct strbuf *path, |
2067 | int check_only) | |
2068 | { | |
b6731550 NTND |
2069 | const char *c_path; |
2070 | ||
cf7c6148 NTND |
2071 | memset(cdir, 0, sizeof(*cdir)); |
2072 | cdir->untracked = untracked; | |
207a06ce | 2073 | if (valid_cached_dir(dir, untracked, istate, path, check_only)) |
91a2288b | 2074 | return 0; |
b6731550 NTND |
2075 | c_path = path->len ? path->buf : "."; |
2076 | cdir->fdir = opendir(c_path); | |
2077 | if (!cdir->fdir) | |
2078 | warning_errno(_("could not open directory '%s'"), c_path); | |
b6403131 NTND |
2079 | if (dir->untracked) { |
2080 | invalidate_directory(dir->untracked, untracked); | |
91a2288b | 2081 | dir->untracked->dir_opened++; |
b6403131 | 2082 | } |
cf7c6148 NTND |
2083 | if (!cdir->fdir) |
2084 | return -1; | |
2085 | return 0; | |
2086 | } | |
2087 | ||
2088 | static int read_cached_dir(struct cached_dir *cdir) | |
2089 | { | |
2090 | if (cdir->fdir) { | |
2091 | cdir->de = readdir(cdir->fdir); | |
2092 | if (!cdir->de) | |
2093 | return -1; | |
2094 | return 0; | |
2095 | } | |
91a2288b NTND |
2096 | while (cdir->nr_dirs < cdir->untracked->dirs_nr) { |
2097 | struct untracked_cache_dir *d = cdir->untracked->dirs[cdir->nr_dirs]; | |
26cb0182 NTND |
2098 | if (!d->recurse) { |
2099 | cdir->nr_dirs++; | |
2100 | continue; | |
2101 | } | |
91a2288b NTND |
2102 | cdir->ucd = d; |
2103 | cdir->nr_dirs++; | |
2104 | return 0; | |
2105 | } | |
2106 | cdir->ucd = NULL; | |
2107 | if (cdir->nr_files < cdir->untracked->untracked_nr) { | |
2108 | struct untracked_cache_dir *d = cdir->untracked; | |
2109 | cdir->file = d->untracked[cdir->nr_files++]; | |
2110 | return 0; | |
2111 | } | |
cf7c6148 NTND |
2112 | return -1; |
2113 | } | |
2114 | ||
2115 | static void close_cached_dir(struct cached_dir *cdir) | |
2116 | { | |
2117 | if (cdir->fdir) | |
2118 | closedir(cdir->fdir); | |
91a2288b NTND |
2119 | /* |
2120 | * We have gone through this directory and found no untracked | |
2121 | * entries. Mark it valid. | |
2122 | */ | |
26cb0182 | 2123 | if (cdir->untracked) { |
91a2288b | 2124 | cdir->untracked->valid = 1; |
26cb0182 NTND |
2125 | cdir->untracked->recurse = 1; |
2126 | } | |
16e2cfa9 JH |
2127 | } |
2128 | ||
c5c4eddd EN |
2129 | static void add_path_to_appropriate_result_list(struct dir_struct *dir, |
2130 | struct untracked_cache_dir *untracked, | |
2131 | struct cached_dir *cdir, | |
2132 | struct index_state *istate, | |
2133 | struct strbuf *path, | |
2134 | int baselen, | |
2135 | const struct pathspec *pathspec, | |
2136 | enum path_treatment state) | |
2137 | { | |
2138 | /* add the path to the appropriate result list */ | |
2139 | switch (state) { | |
2140 | case path_excluded: | |
2141 | if (dir->flags & DIR_SHOW_IGNORED) | |
2142 | dir_add_name(dir, istate, path->buf, path->len); | |
2143 | else if ((dir->flags & DIR_SHOW_IGNORED_TOO) || | |
2144 | ((dir->flags & DIR_COLLECT_IGNORED) && | |
2145 | exclude_matches_pathspec(path->buf, path->len, | |
2146 | pathspec))) | |
2147 | dir_add_ignored(dir, istate, path->buf, path->len); | |
2148 | break; | |
2149 | ||
2150 | case path_untracked: | |
2151 | if (dir->flags & DIR_SHOW_IGNORED) | |
2152 | break; | |
2153 | dir_add_name(dir, istate, path->buf, path->len); | |
2154 | if (cdir->fdir) | |
2155 | add_untracked(untracked, path->buf + baselen); | |
2156 | break; | |
2157 | ||
2158 | default: | |
2159 | break; | |
2160 | } | |
2161 | } | |
2162 | ||
453ec4bd LT |
2163 | /* |
2164 | * Read a directory tree. We currently ignore anything but | |
2165 | * directories, regular files and symlinks. That's because git | |
2166 | * doesn't handle them at all yet. Maybe that will change some | |
2167 | * day. | |
2168 | * | |
2169 | * Also, we ignore the name ".git" (even if it is not a directory). | |
2170 | * That likely will not change. | |
defd7c7b | 2171 | * |
5aaa7fd3 JM |
2172 | * If 'stop_at_first_file' is specified, 'path_excluded' is returned |
2173 | * to signal that a file was found. This is the least significant value that | |
2174 | * indicates that a file was encountered that does not depend on the order of | |
2175 | * whether an untracked or exluded path was encountered first. | |
2176 | * | |
defd7c7b | 2177 | * Returns the most significant path_treatment value encountered in the scan. |
5aaa7fd3 JM |
2178 | * If 'stop_at_first_file' is specified, `path_excluded` is the most |
2179 | * significant path_treatment value that will be returned. | |
453ec4bd | 2180 | */ |
5aaa7fd3 | 2181 | |
defd7c7b | 2182 | static enum path_treatment read_directory_recursive(struct dir_struct *dir, |
0ef8e169 BW |
2183 | struct index_state *istate, const char *base, int baselen, |
2184 | struct untracked_cache_dir *untracked, int check_only, | |
5aaa7fd3 | 2185 | int stop_at_first_file, const struct pathspec *pathspec) |
453ec4bd | 2186 | { |
777b4203 EN |
2187 | /* |
2188 | * WARNING WARNING WARNING: | |
2189 | * | |
2190 | * Any updates to the traversal logic here may need corresponding | |
2191 | * updates in treat_leading_path(). See the commit message for the | |
2192 | * commit adding this warning as well as the commit preceding it | |
2193 | * for details. | |
2194 | */ | |
2195 | ||
cf7c6148 | 2196 | struct cached_dir cdir; |
defd7c7b | 2197 | enum path_treatment state, subdir_state, dir_state = path_none; |
bef36921 | 2198 | struct strbuf path = STRBUF_INIT; |
453ec4bd | 2199 | |
bef36921 | 2200 | strbuf_add(&path, base, baselen); |
02cb6753 | 2201 | |
0ef8e169 | 2202 | if (open_cached_dir(&cdir, dir, untracked, istate, &path, check_only)) |
1528d247 RS |
2203 | goto out; |
2204 | ||
0dcb8d7f NTND |
2205 | if (untracked) |
2206 | untracked->check_only = !!check_only; | |
2207 | ||
cf7c6148 | 2208 | while (!read_cached_dir(&cdir)) { |
defd7c7b | 2209 | /* check how the file or directory should be treated */ |
0ef8e169 | 2210 | state = treat_path(dir, untracked, &cdir, istate, &path, |
e1b8c7bd | 2211 | baselen, pathspec); |
0dcb8d7f | 2212 | |
defd7c7b KB |
2213 | if (state > dir_state) |
2214 | dir_state = state; | |
2215 | ||
2216 | /* recurse into subdir if instructed by treat_path */ | |
df5bcdf8 SL |
2217 | if ((state == path_recurse) || |
2218 | ((state == path_untracked) && | |
89a1f4aa EN |
2219 | (get_dtype(cdir.de, istate, path.buf, path.len) == DT_DIR) && |
2220 | ((dir->flags & DIR_SHOW_IGNORED_TOO) || | |
69f272b9 EN |
2221 | (pathspec && |
2222 | do_match_pathspec(istate, pathspec, path.buf, path.len, | |
2223 | baselen, NULL, DO_MATCH_LEADING_PATHSPEC) == MATCHED_RECURSIVELY_LEADING_PATHSPEC)))) { | |
0dcb8d7f NTND |
2224 | struct untracked_cache_dir *ud; |
2225 | ud = lookup_untracked(dir->untracked, untracked, | |
2226 | path.buf + baselen, | |
2227 | path.len - baselen); | |
2228 | subdir_state = | |
0ef8e169 | 2229 | read_directory_recursive(dir, istate, path.buf, |
e1b8c7bd | 2230 | path.len, ud, |
5aaa7fd3 | 2231 | check_only, stop_at_first_file, pathspec); |
defd7c7b KB |
2232 | if (subdir_state > dir_state) |
2233 | dir_state = subdir_state; | |
404ebced | 2234 | |
69f272b9 EN |
2235 | if (pathspec && |
2236 | !match_pathspec(istate, pathspec, path.buf, path.len, | |
404ebced EN |
2237 | 0 /* prefix */, NULL, |
2238 | 0 /* do NOT special case dirs */)) | |
2239 | state = path_none; | |
defd7c7b KB |
2240 | } |
2241 | ||
2242 | if (check_only) { | |
5aaa7fd3 JM |
2243 | if (stop_at_first_file) { |
2244 | /* | |
2245 | * If stopping at first file, then | |
2246 | * signal that a file was found by | |
2247 | * returning `path_excluded`. This is | |
2248 | * to return a consistent value | |
2249 | * regardless of whether an ignored or | |
2250 | * excluded file happened to be | |
2251 | * encountered 1st. | |
2252 | * | |
2253 | * In current usage, the | |
2254 | * `stop_at_first_file` is passed when | |
2255 | * an ancestor directory has matched | |
2256 | * an exclude pattern, so any found | |
2257 | * files will be excluded. | |
2258 | */ | |
2259 | if (dir_state >= path_excluded) { | |
2260 | dir_state = path_excluded; | |
2261 | break; | |
2262 | } | |
2263 | } | |
2264 | ||
defd7c7b | 2265 | /* abort early if maximum state has been reached */ |
0dcb8d7f | 2266 | if (dir_state == path_untracked) { |
91a2288b | 2267 | if (cdir.fdir) |
0dcb8d7f | 2268 | add_untracked(untracked, path.buf + baselen); |
defd7c7b | 2269 | break; |
0dcb8d7f | 2270 | } |
defd7c7b | 2271 | /* skip the dir_add_* part */ |
02cb6753 | 2272 | continue; |
453ec4bd | 2273 | } |
defd7c7b | 2274 | |
c5c4eddd EN |
2275 | add_path_to_appropriate_result_list(dir, untracked, &cdir, |
2276 | istate, &path, baselen, | |
2277 | pathspec, state); | |
453ec4bd | 2278 | } |
cf7c6148 | 2279 | close_cached_dir(&cdir); |
1528d247 | 2280 | out: |
bef36921 | 2281 | strbuf_release(&path); |
453ec4bd | 2282 | |
defd7c7b | 2283 | return dir_state; |
453ec4bd LT |
2284 | } |
2285 | ||
bbf504a9 | 2286 | int cmp_dir_entry(const void *p1, const void *p2) |
453ec4bd LT |
2287 | { |
2288 | const struct dir_entry *e1 = *(const struct dir_entry **)p1; | |
2289 | const struct dir_entry *e2 = *(const struct dir_entry **)p2; | |
2290 | ||
ccdd4a0f | 2291 | return name_compare(e1->name, e1->len, e2->name, e2->len); |
453ec4bd LT |
2292 | } |
2293 | ||
fb898888 | 2294 | /* check if *out lexically strictly contains *in */ |
bbf504a9 | 2295 | int check_dir_entry_contains(const struct dir_entry *out, const struct dir_entry *in) |
fb898888 SL |
2296 | { |
2297 | return (out->len < in->len) && | |
2298 | (out->name[out->len - 1] == '/') && | |
2299 | !memcmp(out->name, in->name, out->len); | |
2300 | } | |
2301 | ||
48ffef96 | 2302 | static int treat_leading_path(struct dir_struct *dir, |
0ef8e169 | 2303 | struct index_state *istate, |
48ffef96 | 2304 | const char *path, int len, |
e1b8c7bd | 2305 | const struct pathspec *pathspec) |
48ffef96 | 2306 | { |
777b4203 EN |
2307 | /* |
2308 | * WARNING WARNING WARNING: | |
2309 | * | |
2310 | * Any updates to the traversal logic here may need corresponding | |
2311 | * updates in treat_leading_path(). See the commit message for the | |
2312 | * commit adding this warning as well as the commit preceding it | |
2313 | * for details. | |
2314 | */ | |
2315 | ||
49dc2cc2 | 2316 | struct strbuf sb = STRBUF_INIT; |
b9670c1f | 2317 | int prevlen, baselen; |
48ffef96 | 2318 | const char *cp; |
b9670c1f EN |
2319 | struct cached_dir cdir; |
2320 | struct dirent *de; | |
2321 | enum path_treatment state = path_none; | |
2322 | ||
2323 | /* | |
2324 | * For each directory component of path, we are going to check whether | |
2325 | * that path is relevant given the pathspec. For example, if path is | |
2326 | * foo/bar/baz/ | |
2327 | * then we will ask treat_path() whether we should go into foo, then | |
2328 | * whether we should go into bar, then whether baz is relevant. | |
2329 | * Checking each is important because e.g. if path is | |
2330 | * .git/info/ | |
2331 | * then we need to check .git to know we shouldn't traverse it. | |
2332 | * If the return from treat_path() is: | |
2333 | * * path_none, for any path, we return false. | |
2334 | * * path_recurse, for all path components, we return true | |
2335 | * * <anything else> for some intermediate component, we make sure | |
2336 | * to add that path to the relevant list but return false | |
2337 | * signifying that we shouldn't recurse into it. | |
2338 | */ | |
48ffef96 JH |
2339 | |
2340 | while (len && path[len - 1] == '/') | |
2341 | len--; | |
2342 | if (!len) | |
2343 | return 1; | |
b9670c1f EN |
2344 | |
2345 | /* | |
2346 | * We need a manufactured dirent with sufficient space to store a | |
2347 | * leading directory component of path in its d_name. Here, we | |
2348 | * assume that the dirent's d_name is either declared as | |
2349 | * char d_name[BIG_ENOUGH] | |
2350 | * or that it is declared at the end of the struct as | |
2351 | * char d_name[] | |
2352 | * For either case, padding with len+1 bytes at the end will ensure | |
2353 | * sufficient storage space. | |
2354 | */ | |
6836d2fe | 2355 | de = xcalloc(1, st_add3(sizeof(struct dirent), len, 1)); |
b9670c1f EN |
2356 | memset(&cdir, 0, sizeof(cdir)); |
2357 | cdir.de = de; | |
2358 | #if defined(DT_UNKNOWN) && !defined(NO_D_TYPE_IN_DIRENT) | |
2359 | de->d_type = DT_DIR; | |
2360 | #endif | |
48ffef96 | 2361 | baselen = 0; |
b9670c1f | 2362 | prevlen = 0; |
48ffef96 | 2363 | while (1) { |
b9670c1f EN |
2364 | prevlen = baselen + !!baselen; |
2365 | cp = path + prevlen; | |
48ffef96 JH |
2366 | cp = memchr(cp, '/', path + len - cp); |
2367 | if (!cp) | |
2368 | baselen = len; | |
2369 | else | |
2370 | baselen = cp - path; | |
b9670c1f | 2371 | strbuf_reset(&sb); |
49dc2cc2 RS |
2372 | strbuf_add(&sb, path, baselen); |
2373 | if (!is_directory(sb.buf)) | |
2374 | break; | |
b9670c1f EN |
2375 | strbuf_reset(&sb); |
2376 | strbuf_add(&sb, path, prevlen); | |
2377 | memcpy(de->d_name, path+prevlen, baselen-prevlen); | |
2378 | de->d_name[baselen-prevlen] = '\0'; | |
2379 | state = treat_path(dir, NULL, &cdir, istate, &sb, prevlen, | |
2380 | pathspec); | |
777b4203 EN |
2381 | if (state == path_untracked && |
2382 | get_dtype(cdir.de, istate, sb.buf, sb.len) == DT_DIR && | |
2383 | (dir->flags & DIR_SHOW_IGNORED_TOO || | |
2384 | do_match_pathspec(istate, pathspec, sb.buf, sb.len, | |
2385 | baselen, NULL, DO_MATCH_LEADING_PATHSPEC) == MATCHED_RECURSIVELY_LEADING_PATHSPEC)) { | |
2386 | add_path_to_appropriate_result_list(dir, NULL, &cdir, | |
2387 | istate, | |
2388 | &sb, baselen, | |
2389 | pathspec, state); | |
2390 | state = path_recurse; | |
2391 | } | |
2392 | ||
b9670c1f | 2393 | if (state != path_recurse) |
49dc2cc2 | 2394 | break; /* do not recurse into it */ |
b9670c1f | 2395 | if (len <= baselen) |
49dc2cc2 | 2396 | break; /* finished checking */ |
48ffef96 | 2397 | } |
b9670c1f EN |
2398 | add_path_to_appropriate_result_list(dir, NULL, &cdir, istate, |
2399 | &sb, baselen, pathspec, | |
2400 | state); | |
2401 | ||
2402 | free(de); | |
49dc2cc2 | 2403 | strbuf_release(&sb); |
b9670c1f | 2404 | return state == path_recurse; |
48ffef96 JH |
2405 | } |
2406 | ||
1e8fef60 NTND |
2407 | static const char *get_ident_string(void) |
2408 | { | |
2409 | static struct strbuf sb = STRBUF_INIT; | |
2410 | struct utsname uts; | |
2411 | ||
2412 | if (sb.len) | |
2413 | return sb.buf; | |
100e4337 | 2414 | if (uname(&uts) < 0) |
1e8fef60 | 2415 | die_errno(_("failed to get kernel name and information")); |
0e0f7618 CC |
2416 | strbuf_addf(&sb, "Location %s, system %s", get_git_work_tree(), |
2417 | uts.sysname); | |
1e8fef60 NTND |
2418 | return sb.buf; |
2419 | } | |
2420 | ||
2421 | static int ident_in_untracked(const struct untracked_cache *uc) | |
2422 | { | |
0e0f7618 CC |
2423 | /* |
2424 | * Previous git versions may have saved many NUL separated | |
2425 | * strings in the "ident" field, but it is insane to manage | |
2426 | * many locations, so just take care of the first one. | |
2427 | */ | |
1e8fef60 | 2428 | |
0e0f7618 | 2429 | return !strcmp(uc->ident.buf, get_ident_string()); |
1e8fef60 NTND |
2430 | } |
2431 | ||
0e0f7618 | 2432 | static void set_untracked_ident(struct untracked_cache *uc) |
1e8fef60 | 2433 | { |
0e0f7618 | 2434 | strbuf_reset(&uc->ident); |
1e8fef60 | 2435 | strbuf_addstr(&uc->ident, get_ident_string()); |
0e0f7618 CC |
2436 | |
2437 | /* | |
2438 | * This strbuf used to contain a list of NUL separated | |
2439 | * strings, so save NUL too for backward compatibility. | |
2440 | */ | |
1e8fef60 NTND |
2441 | strbuf_addch(&uc->ident, 0); |
2442 | } | |
2443 | ||
4a4ca479 CC |
2444 | static void new_untracked_cache(struct index_state *istate) |
2445 | { | |
2446 | struct untracked_cache *uc = xcalloc(1, sizeof(*uc)); | |
2447 | strbuf_init(&uc->ident, 100); | |
2448 | uc->exclude_per_dir = ".gitignore"; | |
2449 | /* should be the same flags used by git-status */ | |
2450 | uc->dir_flags = DIR_SHOW_OTHER_DIRECTORIES | DIR_HIDE_EMPTY_DIRECTORIES; | |
0e0f7618 | 2451 | set_untracked_ident(uc); |
4a4ca479 | 2452 | istate->untracked = uc; |
0e0f7618 | 2453 | istate->cache_changed |= UNTRACKED_CHANGED; |
4a4ca479 CC |
2454 | } |
2455 | ||
2456 | void add_untracked_cache(struct index_state *istate) | |
2457 | { | |
2458 | if (!istate->untracked) { | |
2459 | new_untracked_cache(istate); | |
0e0f7618 CC |
2460 | } else { |
2461 | if (!ident_in_untracked(istate->untracked)) { | |
2462 | free_untracked_cache(istate->untracked); | |
2463 | new_untracked_cache(istate); | |
2464 | } | |
2465 | } | |
4a4ca479 CC |
2466 | } |
2467 | ||
07b29bfd CC |
2468 | void remove_untracked_cache(struct index_state *istate) |
2469 | { | |
2470 | if (istate->untracked) { | |
2471 | free_untracked_cache(istate->untracked); | |
2472 | istate->untracked = NULL; | |
2473 | istate->cache_changed |= UNTRACKED_CHANGED; | |
2474 | } | |
2475 | } | |
2476 | ||
ccad261f NTND |
2477 | static struct untracked_cache_dir *validate_untracked_cache(struct dir_struct *dir, |
2478 | int base_len, | |
2479 | const struct pathspec *pathspec) | |
2480 | { | |
2481 | struct untracked_cache_dir *root; | |
026336cb | 2482 | static int untracked_cache_disabled = -1; |
ccad261f | 2483 | |
026336cb JH |
2484 | if (!dir->untracked) |
2485 | return NULL; | |
2486 | if (untracked_cache_disabled < 0) | |
2487 | untracked_cache_disabled = git_env_bool("GIT_DISABLE_UNTRACKED_CACHE", 0); | |
2488 | if (untracked_cache_disabled) | |
ccad261f NTND |
2489 | return NULL; |
2490 | ||
2491 | /* | |
2492 | * We only support $GIT_DIR/info/exclude and core.excludesfile | |
2493 | * as the global ignore rule files. Any other additions | |
2494 | * (e.g. from command line) invalidate the cache. This | |
2495 | * condition also catches running setup_standard_excludes() | |
2496 | * before setting dir->untracked! | |
2497 | */ | |
2498 | if (dir->unmanaged_exclude_files) | |
2499 | return NULL; | |
2500 | ||
2501 | /* | |
2502 | * Optimize for the main use case only: whole-tree git | |
2503 | * status. More work involved in treat_leading_path() if we | |
2504 | * use cache on just a subset of the worktree. pathspec | |
2505 | * support could make the matter even worse. | |
2506 | */ | |
2507 | if (base_len || (pathspec && pathspec->nr)) | |
2508 | return NULL; | |
2509 | ||
2510 | /* Different set of flags may produce different results */ | |
2511 | if (dir->flags != dir->untracked->dir_flags || | |
2512 | /* | |
2513 | * See treat_directory(), case index_nonexistent. Without | |
2514 | * this flag, we may need to also cache .git file content | |
2515 | * for the resolve_gitlink_ref() call, which we don't. | |
2516 | */ | |
2517 | !(dir->flags & DIR_SHOW_OTHER_DIRECTORIES) || | |
2518 | /* We don't support collecting ignore files */ | |
2519 | (dir->flags & (DIR_SHOW_IGNORED | DIR_SHOW_IGNORED_TOO | | |
2520 | DIR_COLLECT_IGNORED))) | |
2521 | return NULL; | |
2522 | ||
2523 | /* | |
2524 | * If we use .gitignore in the cache and now you change it to | |
2525 | * .gitexclude, everything will go wrong. | |
2526 | */ | |
2527 | if (dir->exclude_per_dir != dir->untracked->exclude_per_dir && | |
2528 | strcmp(dir->exclude_per_dir, dir->untracked->exclude_per_dir)) | |
2529 | return NULL; | |
2530 | ||
2531 | /* | |
2532 | * EXC_CMDL is not considered in the cache. If people set it, | |
2533 | * skip the cache. | |
2534 | */ | |
2535 | if (dir->exclude_list_group[EXC_CMDL].nr) | |
2536 | return NULL; | |
2537 | ||
1e8fef60 | 2538 | if (!ident_in_untracked(dir->untracked)) { |
1a07e59c | 2539 | warning(_("untracked cache is disabled on this system or location")); |
1e8fef60 NTND |
2540 | return NULL; |
2541 | } | |
2542 | ||
ccad261f NTND |
2543 | if (!dir->untracked->root) { |
2544 | const int len = sizeof(*dir->untracked->root); | |
2545 | dir->untracked->root = xmalloc(len); | |
2546 | memset(dir->untracked->root, 0, len); | |
2547 | } | |
2548 | ||
2549 | /* Validate $GIT_DIR/info/exclude and core.excludesfile */ | |
2550 | root = dir->untracked->root; | |
9001dc2a | 2551 | if (!oideq(&dir->ss_info_exclude.oid, |
4b33e602 | 2552 | &dir->untracked->ss_info_exclude.oid)) { |
ccad261f NTND |
2553 | invalidate_gitignore(dir->untracked, root); |
2554 | dir->untracked->ss_info_exclude = dir->ss_info_exclude; | |
2555 | } | |
9001dc2a | 2556 | if (!oideq(&dir->ss_excludes_file.oid, |
4b33e602 | 2557 | &dir->untracked->ss_excludes_file.oid)) { |
ccad261f NTND |
2558 | invalidate_gitignore(dir->untracked, root); |
2559 | dir->untracked->ss_excludes_file = dir->ss_excludes_file; | |
2560 | } | |
26cb0182 NTND |
2561 | |
2562 | /* Make sure this directory is not dropped out at saving phase */ | |
2563 | root->recurse = 1; | |
ccad261f NTND |
2564 | return root; |
2565 | } | |
2566 | ||
2c1eb104 BW |
2567 | int read_directory(struct dir_struct *dir, struct index_state *istate, |
2568 | const char *path, int len, const struct pathspec *pathspec) | |
9fc42d60 | 2569 | { |
ccad261f | 2570 | struct untracked_cache_dir *untracked; |
b4189aa8 | 2571 | |
c46c406a NTND |
2572 | trace_performance_enter(); |
2573 | ||
2574 | if (has_symlink_leading_path(path, len)) { | |
2575 | trace_performance_leave("read directory %.*s", len, path); | |
725b0605 | 2576 | return dir->nr; |
c46c406a | 2577 | } |
725b0605 | 2578 | |
ccad261f NTND |
2579 | untracked = validate_untracked_cache(dir, len, pathspec); |
2580 | if (!untracked) | |
2581 | /* | |
2582 | * make sure untracked cache code path is disabled, | |
2583 | * e.g. prep_exclude() | |
2584 | */ | |
2585 | dir->untracked = NULL; | |
2c1eb104 | 2586 | if (!len || treat_leading_path(dir, istate, path, len, pathspec)) |
5aaa7fd3 | 2587 | read_directory_recursive(dir, istate, path, len, untracked, 0, 0, pathspec); |
bbf504a9 SL |
2588 | QSORT(dir->entries, dir->nr, cmp_dir_entry); |
2589 | QSORT(dir->ignored, dir->ignored_nr, cmp_dir_entry); | |
fb898888 SL |
2590 | |
2591 | /* | |
2592 | * If DIR_SHOW_IGNORED_TOO is set, read_directory_recursive() will | |
2593 | * also pick up untracked contents of untracked dirs; by default | |
2594 | * we discard these, but given DIR_KEEP_UNTRACKED_CONTENTS we do not. | |
2595 | */ | |
2596 | if ((dir->flags & DIR_SHOW_IGNORED_TOO) && | |
2597 | !(dir->flags & DIR_KEEP_UNTRACKED_CONTENTS)) { | |
2598 | int i, j; | |
2599 | ||
2600 | /* remove from dir->entries untracked contents of untracked dirs */ | |
2601 | for (i = j = 0; j < dir->nr; j++) { | |
bbf504a9 SL |
2602 | if (i && |
2603 | check_dir_entry_contains(dir->entries[i - 1], dir->entries[j])) { | |
6a83d902 | 2604 | FREE_AND_NULL(dir->entries[j]); |
fb898888 SL |
2605 | } else { |
2606 | dir->entries[i++] = dir->entries[j]; | |
2607 | } | |
2608 | } | |
2609 | ||
2610 | dir->nr = i; | |
2611 | } | |
2612 | ||
c46c406a | 2613 | trace_performance_leave("read directory %.*s", len, path); |
c9ccb5d3 | 2614 | if (dir->untracked) { |
026336cb | 2615 | static int force_untracked_cache = -1; |
c9ccb5d3 | 2616 | static struct trace_key trace_untracked_stats = TRACE_KEY_INIT(UNTRACKED_STATS); |
026336cb JH |
2617 | |
2618 | if (force_untracked_cache < 0) | |
2619 | force_untracked_cache = | |
2620 | git_env_bool("GIT_FORCE_UNTRACKED_CACHE", 0); | |
c9ccb5d3 NTND |
2621 | trace_printf_key(&trace_untracked_stats, |
2622 | "node creation: %u\n" | |
2623 | "gitignore invalidation: %u\n" | |
2624 | "directory invalidation: %u\n" | |
2625 | "opendir: %u\n", | |
2626 | dir->untracked->dir_created, | |
2627 | dir->untracked->gitignore_invalidated, | |
2628 | dir->untracked->dir_invalidated, | |
2629 | dir->untracked->dir_opened); | |
026336cb | 2630 | if (force_untracked_cache && |
fc9ecbeb | 2631 | dir->untracked == istate->untracked && |
1bbb3dba NTND |
2632 | (dir->untracked->dir_opened || |
2633 | dir->untracked->gitignore_invalidated || | |
2634 | dir->untracked->dir_invalidated)) | |
2c1eb104 BW |
2635 | istate->cache_changed |= UNTRACKED_CHANGED; |
2636 | if (dir->untracked != istate->untracked) { | |
6a83d902 | 2637 | FREE_AND_NULL(dir->untracked); |
1bbb3dba | 2638 | } |
c9ccb5d3 | 2639 | } |
453ec4bd LT |
2640 | return dir->nr; |
2641 | } | |
c91f0d92 | 2642 | |
686a4a06 | 2643 | int file_exists(const char *f) |
c91f0d92 | 2644 | { |
686a4a06 | 2645 | struct stat sb; |
a50f9fc5 | 2646 | return lstat(f, &sb) == 0; |
c91f0d92 | 2647 | } |
e6636747 | 2648 | |
0488481e NTND |
2649 | int repo_file_exists(struct repository *repo, const char *path) |
2650 | { | |
2651 | if (repo != the_repository) | |
2652 | BUG("do not know how to check file existence in arbitrary repo"); | |
2653 | ||
2654 | return file_exists(path); | |
2655 | } | |
2656 | ||
63ec5e1f JS |
2657 | static int cmp_icase(char a, char b) |
2658 | { | |
2659 | if (a == b) | |
2660 | return 0; | |
2661 | if (ignore_case) | |
2662 | return toupper(a) - toupper(b); | |
2663 | return a - b; | |
2664 | } | |
2665 | ||
e6636747 | 2666 | /* |
9b125da4 NTND |
2667 | * Given two normalized paths (a trailing slash is ok), if subdir is |
2668 | * outside dir, return -1. Otherwise return the offset in subdir that | |
2669 | * can be used as relative path to dir. | |
e6636747 | 2670 | */ |
9b125da4 | 2671 | int dir_inside_of(const char *subdir, const char *dir) |
e6636747 | 2672 | { |
9b125da4 | 2673 | int offset = 0; |
e6636747 | 2674 | |
9b125da4 | 2675 | assert(dir && subdir && *dir && *subdir); |
e6636747 | 2676 | |
63ec5e1f | 2677 | while (*dir && *subdir && !cmp_icase(*dir, *subdir)) { |
e6636747 | 2678 | dir++; |
9b125da4 NTND |
2679 | subdir++; |
2680 | offset++; | |
490544b1 | 2681 | } |
9b125da4 NTND |
2682 | |
2683 | /* hel[p]/me vs hel[l]/yeah */ | |
2684 | if (*dir && *subdir) | |
2685 | return -1; | |
2686 | ||
2687 | if (!*subdir) | |
2688 | return !*dir ? offset : -1; /* same dir */ | |
2689 | ||
2690 | /* foo/[b]ar vs foo/[] */ | |
2691 | if (is_dir_sep(dir[-1])) | |
2692 | return is_dir_sep(subdir[-1]) ? offset : -1; | |
2693 | ||
2694 | /* foo[/]bar vs foo[] */ | |
2695 | return is_dir_sep(*subdir) ? offset + 1 : -1; | |
e6636747 JS |
2696 | } |
2697 | ||
2698 | int is_inside_dir(const char *dir) | |
2699 | { | |
56b9f6e7 RS |
2700 | char *cwd; |
2701 | int rc; | |
2702 | ||
b892913d NTND |
2703 | if (!dir) |
2704 | return 0; | |
56b9f6e7 RS |
2705 | |
2706 | cwd = xgetcwd(); | |
2707 | rc = (dir_inside_of(cwd, dir) >= 0); | |
2708 | free(cwd); | |
2709 | return rc; | |
e6636747 | 2710 | } |
7155b727 | 2711 | |
55892d23 AP |
2712 | int is_empty_dir(const char *path) |
2713 | { | |
2714 | DIR *dir = opendir(path); | |
2715 | struct dirent *e; | |
2716 | int ret = 1; | |
2717 | ||
2718 | if (!dir) | |
2719 | return 0; | |
2720 | ||
2721 | while ((e = readdir(dir)) != NULL) | |
2722 | if (!is_dot_or_dotdot(e->d_name)) { | |
2723 | ret = 0; | |
2724 | break; | |
2725 | } | |
2726 | ||
2727 | closedir(dir); | |
2728 | return ret; | |
2729 | } | |
2730 | ||
ae2f203e | 2731 | static int remove_dir_recurse(struct strbuf *path, int flag, int *kept_up) |
7155b727 | 2732 | { |
a0f4afbe | 2733 | DIR *dir; |
7155b727 | 2734 | struct dirent *e; |
ae2f203e | 2735 | int ret = 0, original_len = path->len, len, kept_down = 0; |
a0f4afbe | 2736 | int only_empty = (flag & REMOVE_DIR_EMPTY_ONLY); |
c844a803 | 2737 | int keep_toplevel = (flag & REMOVE_DIR_KEEP_TOPLEVEL); |
1053fe82 | 2738 | struct object_id submodule_head; |
7155b727 | 2739 | |
a0f4afbe | 2740 | if ((flag & REMOVE_DIR_KEEP_NESTED_GIT) && |
a98e6101 | 2741 | !resolve_gitlink_ref(path->buf, "HEAD", &submodule_head)) { |
a0f4afbe | 2742 | /* Do not descend and nuke a nested git work tree. */ |
ae2f203e JH |
2743 | if (kept_up) |
2744 | *kept_up = 1; | |
a0f4afbe | 2745 | return 0; |
ae2f203e | 2746 | } |
a0f4afbe | 2747 | |
ae2f203e | 2748 | flag &= ~REMOVE_DIR_KEEP_TOPLEVEL; |
a0f4afbe | 2749 | dir = opendir(path->buf); |
c844a803 | 2750 | if (!dir) { |
863808cd MH |
2751 | if (errno == ENOENT) |
2752 | return keep_toplevel ? -1 : 0; | |
2753 | else if (errno == EACCES && !keep_toplevel) | |
ecb2c282 MH |
2754 | /* |
2755 | * An empty dir could be removable even if it | |
2756 | * is unreadable: | |
2757 | */ | |
c844a803 JH |
2758 | return rmdir(path->buf); |
2759 | else | |
2760 | return -1; | |
2761 | } | |
00b6c178 | 2762 | strbuf_complete(path, '/'); |
7155b727 JS |
2763 | |
2764 | len = path->len; | |
2765 | while ((e = readdir(dir)) != NULL) { | |
2766 | struct stat st; | |
8ca12c0d AP |
2767 | if (is_dot_or_dotdot(e->d_name)) |
2768 | continue; | |
7155b727 JS |
2769 | |
2770 | strbuf_setlen(path, len); | |
2771 | strbuf_addstr(path, e->d_name); | |
863808cd MH |
2772 | if (lstat(path->buf, &st)) { |
2773 | if (errno == ENOENT) | |
2774 | /* | |
2775 | * file disappeared, which is what we | |
2776 | * wanted anyway | |
2777 | */ | |
2778 | continue; | |
15beaaa3 | 2779 | /* fall through */ |
863808cd | 2780 | } else if (S_ISDIR(st.st_mode)) { |
ae2f203e | 2781 | if (!remove_dir_recurse(path, flag, &kept_down)) |
7155b727 | 2782 | continue; /* happy */ |
863808cd MH |
2783 | } else if (!only_empty && |
2784 | (!unlink(path->buf) || errno == ENOENT)) { | |
7155b727 | 2785 | continue; /* happy, too */ |
863808cd | 2786 | } |
7155b727 JS |
2787 | |
2788 | /* path too long, stat fails, or non-directory still exists */ | |
2789 | ret = -1; | |
2790 | break; | |
2791 | } | |
2792 | closedir(dir); | |
2793 | ||
2794 | strbuf_setlen(path, original_len); | |
ae2f203e | 2795 | if (!ret && !keep_toplevel && !kept_down) |
863808cd | 2796 | ret = (!rmdir(path->buf) || errno == ENOENT) ? 0 : -1; |
ae2f203e JH |
2797 | else if (kept_up) |
2798 | /* | |
2799 | * report the uplevel that it is not an error that we | |
2800 | * did not rmdir() our directory. | |
2801 | */ | |
2802 | *kept_up = !ret; | |
7155b727 JS |
2803 | return ret; |
2804 | } | |
039bc64e | 2805 | |
ae2f203e JH |
2806 | int remove_dir_recursively(struct strbuf *path, int flag) |
2807 | { | |
2808 | return remove_dir_recurse(path, flag, NULL); | |
2809 | } | |
2810 | ||
f932729c JK |
2811 | static GIT_PATH_FUNC(git_path_info_exclude, "info/exclude") |
2812 | ||
039bc64e JH |
2813 | void setup_standard_excludes(struct dir_struct *dir) |
2814 | { | |
039bc64e | 2815 | dir->exclude_per_dir = ".gitignore"; |
099d2d86 | 2816 | |
51d18631 | 2817 | /* core.excludesfile defaulting to $XDG_CONFIG_HOME/git/ignore */ |
2845ce7f PT |
2818 | if (!excludes_file) |
2819 | excludes_file = xdg_config_home("ignore"); | |
4698c8fe | 2820 | if (excludes_file && !access_or_warn(excludes_file, R_OK, 0)) |
65edd96a | 2821 | add_patterns_from_file_1(dir, excludes_file, |
38ccaf93 | 2822 | dir->untracked ? &dir->ss_excludes_file : NULL); |
099d2d86 JH |
2823 | |
2824 | /* per repository user preference */ | |
f0056f64 JK |
2825 | if (startup_info->have_repository) { |
2826 | const char *path = git_path_info_exclude(); | |
2827 | if (!access_or_warn(path, R_OK, 0)) | |
65edd96a | 2828 | add_patterns_from_file_1(dir, path, |
f0056f64 JK |
2829 | dir->untracked ? &dir->ss_info_exclude : NULL); |
2830 | } | |
039bc64e | 2831 | } |
4a92d1bf AR |
2832 | |
2833 | int remove_path(const char *name) | |
2834 | { | |
2835 | char *slash; | |
2836 | ||
c7054209 | 2837 | if (unlink(name) && !is_missing_file_error(errno)) |
4a92d1bf AR |
2838 | return -1; |
2839 | ||
2840 | slash = strrchr(name, '/'); | |
2841 | if (slash) { | |
2842 | char *dirs = xstrdup(name); | |
2843 | slash = dirs + (slash - name); | |
2844 | do { | |
2845 | *slash = '\0'; | |
3fc0d131 | 2846 | } while (rmdir(dirs) == 0 && (slash = strrchr(dirs, '/'))); |
4a92d1bf AR |
2847 | free(dirs); |
2848 | } | |
2849 | return 0; | |
2850 | } | |
2851 | ||
270be816 AS |
2852 | /* |
2853 | * Frees memory within dir which was allocated for exclude lists and | |
2854 | * the exclude_stack. Does not free dir itself. | |
2855 | */ | |
2856 | void clear_directory(struct dir_struct *dir) | |
2857 | { | |
2858 | int i, j; | |
2859 | struct exclude_list_group *group; | |
caa3d554 | 2860 | struct pattern_list *pl; |
270be816 AS |
2861 | struct exclude_stack *stk; |
2862 | ||
2863 | for (i = EXC_CMDL; i <= EXC_FILE; i++) { | |
2864 | group = &dir->exclude_list_group[i]; | |
2865 | for (j = 0; j < group->nr; j++) { | |
caa3d554 | 2866 | pl = &group->pl[j]; |
270be816 | 2867 | if (i == EXC_DIRS) |
caa3d554 | 2868 | free((char *)pl->src); |
65edd96a | 2869 | clear_pattern_list(pl); |
270be816 | 2870 | } |
caa3d554 | 2871 | free(group->pl); |
270be816 AS |
2872 | } |
2873 | ||
2874 | stk = dir->exclude_stack; | |
2875 | while (stk) { | |
2876 | struct exclude_stack *prev = stk->prev; | |
2877 | free(stk); | |
2878 | stk = prev; | |
2879 | } | |
aceb9429 | 2880 | strbuf_release(&dir->basebuf); |
270be816 | 2881 | } |
83c094ad NTND |
2882 | |
2883 | struct ondisk_untracked_cache { | |
2884 | struct stat_data info_exclude_stat; | |
2885 | struct stat_data excludes_file_stat; | |
2886 | uint32_t dir_flags; | |
83c094ad NTND |
2887 | }; |
2888 | ||
268ba201 | 2889 | #define ouc_offset(x) offsetof(struct ondisk_untracked_cache, x) |
83c094ad NTND |
2890 | |
2891 | struct write_data { | |
2892 | int index; /* number of written untracked_cache_dir */ | |
2893 | struct ewah_bitmap *check_only; /* from untracked_cache_dir */ | |
2894 | struct ewah_bitmap *valid; /* from untracked_cache_dir */ | |
2895 | struct ewah_bitmap *sha1_valid; /* set if exclude_sha1 is not null */ | |
2896 | struct strbuf out; | |
2897 | struct strbuf sb_stat; | |
2898 | struct strbuf sb_sha1; | |
2899 | }; | |
2900 | ||
2901 | static void stat_data_to_disk(struct stat_data *to, const struct stat_data *from) | |
2902 | { | |
2903 | to->sd_ctime.sec = htonl(from->sd_ctime.sec); | |
2904 | to->sd_ctime.nsec = htonl(from->sd_ctime.nsec); | |
2905 | to->sd_mtime.sec = htonl(from->sd_mtime.sec); | |
2906 | to->sd_mtime.nsec = htonl(from->sd_mtime.nsec); | |
2907 | to->sd_dev = htonl(from->sd_dev); | |
2908 | to->sd_ino = htonl(from->sd_ino); | |
2909 | to->sd_uid = htonl(from->sd_uid); | |
2910 | to->sd_gid = htonl(from->sd_gid); | |
2911 | to->sd_size = htonl(from->sd_size); | |
2912 | } | |
2913 | ||
2914 | static void write_one_dir(struct untracked_cache_dir *untracked, | |
2915 | struct write_data *wd) | |
2916 | { | |
2917 | struct stat_data stat_data; | |
2918 | struct strbuf *out = &wd->out; | |
2919 | unsigned char intbuf[16]; | |
2920 | unsigned int intlen, value; | |
2921 | int i = wd->index++; | |
2922 | ||
2923 | /* | |
2924 | * untracked_nr should be reset whenever valid is clear, but | |
2925 | * for safety.. | |
2926 | */ | |
2927 | if (!untracked->valid) { | |
2928 | untracked->untracked_nr = 0; | |
2929 | untracked->check_only = 0; | |
2930 | } | |
2931 | ||
2932 | if (untracked->check_only) | |
2933 | ewah_set(wd->check_only, i); | |
2934 | if (untracked->valid) { | |
2935 | ewah_set(wd->valid, i); | |
2936 | stat_data_to_disk(&stat_data, &untracked->stat_data); | |
2937 | strbuf_add(&wd->sb_stat, &stat_data, sizeof(stat_data)); | |
2938 | } | |
70c369cd | 2939 | if (!is_null_oid(&untracked->exclude_oid)) { |
83c094ad | 2940 | ewah_set(wd->sha1_valid, i); |
70c369cd | 2941 | strbuf_add(&wd->sb_sha1, untracked->exclude_oid.hash, |
2942 | the_hash_algo->rawsz); | |
83c094ad NTND |
2943 | } |
2944 | ||
2945 | intlen = encode_varint(untracked->untracked_nr, intbuf); | |
2946 | strbuf_add(out, intbuf, intlen); | |
2947 | ||
2948 | /* skip non-recurse directories */ | |
2949 | for (i = 0, value = 0; i < untracked->dirs_nr; i++) | |
2950 | if (untracked->dirs[i]->recurse) | |
2951 | value++; | |
2952 | intlen = encode_varint(value, intbuf); | |
2953 | strbuf_add(out, intbuf, intlen); | |
2954 | ||
2955 | strbuf_add(out, untracked->name, strlen(untracked->name) + 1); | |
2956 | ||
2957 | for (i = 0; i < untracked->untracked_nr; i++) | |
2958 | strbuf_add(out, untracked->untracked[i], | |
2959 | strlen(untracked->untracked[i]) + 1); | |
2960 | ||
2961 | for (i = 0; i < untracked->dirs_nr; i++) | |
2962 | if (untracked->dirs[i]->recurse) | |
2963 | write_one_dir(untracked->dirs[i], wd); | |
2964 | } | |
2965 | ||
2966 | void write_untracked_extension(struct strbuf *out, struct untracked_cache *untracked) | |
2967 | { | |
2968 | struct ondisk_untracked_cache *ouc; | |
2969 | struct write_data wd; | |
2970 | unsigned char varbuf[16]; | |
e0b83735 | 2971 | int varint_len; |
3899b88b | 2972 | const unsigned hashsz = the_hash_algo->rawsz; |
e0b83735 | 2973 | |
3899b88b | 2974 | ouc = xcalloc(1, sizeof(*ouc)); |
83c094ad NTND |
2975 | stat_data_to_disk(&ouc->info_exclude_stat, &untracked->ss_info_exclude.stat); |
2976 | stat_data_to_disk(&ouc->excludes_file_stat, &untracked->ss_excludes_file.stat); | |
83c094ad | 2977 | ouc->dir_flags = htonl(untracked->dir_flags); |
1e8fef60 NTND |
2978 | |
2979 | varint_len = encode_varint(untracked->ident.len, varbuf); | |
2980 | strbuf_add(out, varbuf, varint_len); | |
8109984d | 2981 | strbuf_addbuf(out, &untracked->ident); |
1e8fef60 | 2982 | |
3899b88b | 2983 | strbuf_add(out, ouc, sizeof(*ouc)); |
2984 | strbuf_add(out, untracked->ss_info_exclude.oid.hash, hashsz); | |
2985 | strbuf_add(out, untracked->ss_excludes_file.oid.hash, hashsz); | |
2986 | strbuf_add(out, untracked->exclude_per_dir, strlen(untracked->exclude_per_dir) + 1); | |
6a83d902 | 2987 | FREE_AND_NULL(ouc); |
83c094ad NTND |
2988 | |
2989 | if (!untracked->root) { | |
2990 | varint_len = encode_varint(0, varbuf); | |
2991 | strbuf_add(out, varbuf, varint_len); | |
2992 | return; | |
2993 | } | |
2994 | ||
2995 | wd.index = 0; | |
2996 | wd.check_only = ewah_new(); | |
2997 | wd.valid = ewah_new(); | |
2998 | wd.sha1_valid = ewah_new(); | |
2999 | strbuf_init(&wd.out, 1024); | |
3000 | strbuf_init(&wd.sb_stat, 1024); | |
3001 | strbuf_init(&wd.sb_sha1, 1024); | |
3002 | write_one_dir(untracked->root, &wd); | |
3003 | ||
3004 | varint_len = encode_varint(wd.index, varbuf); | |
3005 | strbuf_add(out, varbuf, varint_len); | |
3006 | strbuf_addbuf(out, &wd.out); | |
3007 | ewah_serialize_strbuf(wd.valid, out); | |
3008 | ewah_serialize_strbuf(wd.check_only, out); | |
3009 | ewah_serialize_strbuf(wd.sha1_valid, out); | |
3010 | strbuf_addbuf(out, &wd.sb_stat); | |
3011 | strbuf_addbuf(out, &wd.sb_sha1); | |
3012 | strbuf_addch(out, '\0'); /* safe guard for string lists */ | |
3013 | ||
3014 | ewah_free(wd.valid); | |
3015 | ewah_free(wd.check_only); | |
3016 | ewah_free(wd.sha1_valid); | |
3017 | strbuf_release(&wd.out); | |
3018 | strbuf_release(&wd.sb_stat); | |
3019 | strbuf_release(&wd.sb_sha1); | |
3020 | } | |
f9e6c649 NTND |
3021 | |
3022 | static void free_untracked(struct untracked_cache_dir *ucd) | |
3023 | { | |
3024 | int i; | |
3025 | if (!ucd) | |
3026 | return; | |
3027 | for (i = 0; i < ucd->dirs_nr; i++) | |
3028 | free_untracked(ucd->dirs[i]); | |
3029 | for (i = 0; i < ucd->untracked_nr; i++) | |
3030 | free(ucd->untracked[i]); | |
3031 | free(ucd->untracked); | |
3032 | free(ucd->dirs); | |
3033 | free(ucd); | |
3034 | } | |
3035 | ||
3036 | void free_untracked_cache(struct untracked_cache *uc) | |
3037 | { | |
3038 | if (uc) | |
3039 | free_untracked(uc->root); | |
3040 | free(uc); | |
3041 | } | |
3042 | ||
3043 | struct read_data { | |
3044 | int index; | |
3045 | struct untracked_cache_dir **ucd; | |
3046 | struct ewah_bitmap *check_only; | |
3047 | struct ewah_bitmap *valid; | |
3048 | struct ewah_bitmap *sha1_valid; | |
3049 | const unsigned char *data; | |
3050 | const unsigned char *end; | |
3051 | }; | |
3052 | ||
268ba201 | 3053 | static void stat_data_from_disk(struct stat_data *to, const unsigned char *data) |
f9e6c649 | 3054 | { |
268ba201 RS |
3055 | memcpy(to, data, sizeof(*to)); |
3056 | to->sd_ctime.sec = ntohl(to->sd_ctime.sec); | |
3057 | to->sd_ctime.nsec = ntohl(to->sd_ctime.nsec); | |
3058 | to->sd_mtime.sec = ntohl(to->sd_mtime.sec); | |
3059 | to->sd_mtime.nsec = ntohl(to->sd_mtime.nsec); | |
3060 | to->sd_dev = ntohl(to->sd_dev); | |
3061 | to->sd_ino = ntohl(to->sd_ino); | |
3062 | to->sd_uid = ntohl(to->sd_uid); | |
3063 | to->sd_gid = ntohl(to->sd_gid); | |
3064 | to->sd_size = ntohl(to->sd_size); | |
f9e6c649 NTND |
3065 | } |
3066 | ||
3067 | static int read_one_dir(struct untracked_cache_dir **untracked_, | |
3068 | struct read_data *rd) | |
3069 | { | |
3070 | struct untracked_cache_dir ud, *untracked; | |
b511d6d5 | 3071 | const unsigned char *data = rd->data, *end = rd->end; |
c6909f99 | 3072 | const unsigned char *eos; |
f9e6c649 | 3073 | unsigned int value; |
08bf354d | 3074 | int i; |
f9e6c649 NTND |
3075 | |
3076 | memset(&ud, 0, sizeof(ud)); | |
3077 | ||
b511d6d5 JK |
3078 | value = decode_varint(&data); |
3079 | if (data > end) | |
f9e6c649 NTND |
3080 | return -1; |
3081 | ud.recurse = 1; | |
3082 | ud.untracked_alloc = value; | |
3083 | ud.untracked_nr = value; | |
3084 | if (ud.untracked_nr) | |
b32fa95f | 3085 | ALLOC_ARRAY(ud.untracked, ud.untracked_nr); |
f9e6c649 | 3086 | |
b511d6d5 JK |
3087 | ud.dirs_alloc = ud.dirs_nr = decode_varint(&data); |
3088 | if (data > end) | |
f9e6c649 | 3089 | return -1; |
b32fa95f | 3090 | ALLOC_ARRAY(ud.dirs, ud.dirs_nr); |
f9e6c649 | 3091 | |
c6909f99 JK |
3092 | eos = memchr(data, '\0', end - data); |
3093 | if (!eos || eos == end) | |
f9e6c649 | 3094 | return -1; |
c6909f99 | 3095 | |
08bf354d | 3096 | *untracked_ = untracked = xmalloc(st_add3(sizeof(*untracked), eos - data, 1)); |
f9e6c649 | 3097 | memcpy(untracked, &ud, sizeof(ud)); |
08bf354d | 3098 | memcpy(untracked->name, data, eos - data + 1); |
b511d6d5 | 3099 | data = eos + 1; |
f9e6c649 NTND |
3100 | |
3101 | for (i = 0; i < untracked->untracked_nr; i++) { | |
c6909f99 JK |
3102 | eos = memchr(data, '\0', end - data); |
3103 | if (!eos || eos == end) | |
f9e6c649 | 3104 | return -1; |
08bf354d | 3105 | untracked->untracked[i] = xmemdupz(data, eos - data); |
b511d6d5 | 3106 | data = eos + 1; |
f9e6c649 NTND |
3107 | } |
3108 | ||
3109 | rd->ucd[rd->index++] = untracked; | |
3110 | rd->data = data; | |
3111 | ||
3112 | for (i = 0; i < untracked->dirs_nr; i++) { | |
08bf354d | 3113 | if (read_one_dir(untracked->dirs + i, rd) < 0) |
f9e6c649 NTND |
3114 | return -1; |
3115 | } | |
3116 | return 0; | |
3117 | } | |
3118 | ||
3119 | static void set_check_only(size_t pos, void *cb) | |
3120 | { | |
3121 | struct read_data *rd = cb; | |
3122 | struct untracked_cache_dir *ud = rd->ucd[pos]; | |
3123 | ud->check_only = 1; | |
3124 | } | |
3125 | ||
3126 | static void read_stat(size_t pos, void *cb) | |
3127 | { | |
3128 | struct read_data *rd = cb; | |
3129 | struct untracked_cache_dir *ud = rd->ucd[pos]; | |
3130 | if (rd->data + sizeof(struct stat_data) > rd->end) { | |
3131 | rd->data = rd->end + 1; | |
3132 | return; | |
3133 | } | |
268ba201 | 3134 | stat_data_from_disk(&ud->stat_data, rd->data); |
f9e6c649 NTND |
3135 | rd->data += sizeof(struct stat_data); |
3136 | ud->valid = 1; | |
3137 | } | |
3138 | ||
70c369cd | 3139 | static void read_oid(size_t pos, void *cb) |
f9e6c649 NTND |
3140 | { |
3141 | struct read_data *rd = cb; | |
3142 | struct untracked_cache_dir *ud = rd->ucd[pos]; | |
70c369cd | 3143 | if (rd->data + the_hash_algo->rawsz > rd->end) { |
f9e6c649 NTND |
3144 | rd->data = rd->end + 1; |
3145 | return; | |
3146 | } | |
70c369cd | 3147 | hashcpy(ud->exclude_oid.hash, rd->data); |
3148 | rd->data += the_hash_algo->rawsz; | |
f9e6c649 NTND |
3149 | } |
3150 | ||
4b33e602 PO |
3151 | static void load_oid_stat(struct oid_stat *oid_stat, const unsigned char *data, |
3152 | const unsigned char *sha1) | |
f9e6c649 | 3153 | { |
4b33e602 PO |
3154 | stat_data_from_disk(&oid_stat->stat, data); |
3155 | hashcpy(oid_stat->oid.hash, sha1); | |
3156 | oid_stat->valid = 1; | |
f9e6c649 NTND |
3157 | } |
3158 | ||
3159 | struct untracked_cache *read_untracked_extension(const void *data, unsigned long sz) | |
3160 | { | |
f9e6c649 NTND |
3161 | struct untracked_cache *uc; |
3162 | struct read_data rd; | |
3163 | const unsigned char *next = data, *end = (const unsigned char *)data + sz; | |
1e8fef60 | 3164 | const char *ident; |
1140bf01 JK |
3165 | int ident_len; |
3166 | ssize_t len; | |
268ba201 | 3167 | const char *exclude_per_dir; |
3899b88b | 3168 | const unsigned hashsz = the_hash_algo->rawsz; |
3169 | const unsigned offset = sizeof(struct ondisk_untracked_cache); | |
3170 | const unsigned exclude_per_dir_offset = offset + 2 * hashsz; | |
f9e6c649 NTND |
3171 | |
3172 | if (sz <= 1 || end[-1] != '\0') | |
3173 | return NULL; | |
3174 | end--; | |
3175 | ||
1e8fef60 NTND |
3176 | ident_len = decode_varint(&next); |
3177 | if (next + ident_len > end) | |
3178 | return NULL; | |
3179 | ident = (const char *)next; | |
3180 | next += ident_len; | |
3181 | ||
3899b88b | 3182 | if (next + exclude_per_dir_offset + 1 > end) |
f9e6c649 NTND |
3183 | return NULL; |
3184 | ||
3185 | uc = xcalloc(1, sizeof(*uc)); | |
1e8fef60 NTND |
3186 | strbuf_init(&uc->ident, ident_len); |
3187 | strbuf_add(&uc->ident, ident, ident_len); | |
4b33e602 PO |
3188 | load_oid_stat(&uc->ss_info_exclude, |
3189 | next + ouc_offset(info_exclude_stat), | |
3899b88b | 3190 | next + offset); |
4b33e602 PO |
3191 | load_oid_stat(&uc->ss_excludes_file, |
3192 | next + ouc_offset(excludes_file_stat), | |
3899b88b | 3193 | next + offset + hashsz); |
268ba201 | 3194 | uc->dir_flags = get_be32(next + ouc_offset(dir_flags)); |
3899b88b | 3195 | exclude_per_dir = (const char *)next + exclude_per_dir_offset; |
268ba201 | 3196 | uc->exclude_per_dir = xstrdup(exclude_per_dir); |
f9e6c649 | 3197 | /* NUL after exclude_per_dir is covered by sizeof(*ouc) */ |
3899b88b | 3198 | next += exclude_per_dir_offset + strlen(exclude_per_dir) + 1; |
f9e6c649 NTND |
3199 | if (next >= end) |
3200 | goto done2; | |
3201 | ||
3202 | len = decode_varint(&next); | |
3203 | if (next > end || len == 0) | |
3204 | goto done2; | |
3205 | ||
3206 | rd.valid = ewah_new(); | |
3207 | rd.check_only = ewah_new(); | |
3208 | rd.sha1_valid = ewah_new(); | |
3209 | rd.data = next; | |
3210 | rd.end = end; | |
3211 | rd.index = 0; | |
b32fa95f | 3212 | ALLOC_ARRAY(rd.ucd, len); |
f9e6c649 NTND |
3213 | |
3214 | if (read_one_dir(&uc->root, &rd) || rd.index != len) | |
3215 | goto done; | |
3216 | ||
3217 | next = rd.data; | |
3218 | len = ewah_read_mmap(rd.valid, next, end - next); | |
3219 | if (len < 0) | |
3220 | goto done; | |
3221 | ||
3222 | next += len; | |
3223 | len = ewah_read_mmap(rd.check_only, next, end - next); | |
3224 | if (len < 0) | |
3225 | goto done; | |
3226 | ||
3227 | next += len; | |
3228 | len = ewah_read_mmap(rd.sha1_valid, next, end - next); | |
3229 | if (len < 0) | |
3230 | goto done; | |
3231 | ||
3232 | ewah_each_bit(rd.check_only, set_check_only, &rd); | |
3233 | rd.data = next + len; | |
3234 | ewah_each_bit(rd.valid, read_stat, &rd); | |
70c369cd | 3235 | ewah_each_bit(rd.sha1_valid, read_oid, &rd); |
f9e6c649 NTND |
3236 | next = rd.data; |
3237 | ||
3238 | done: | |
3239 | free(rd.ucd); | |
3240 | ewah_free(rd.valid); | |
3241 | ewah_free(rd.check_only); | |
3242 | ewah_free(rd.sha1_valid); | |
3243 | done2: | |
3244 | if (next != end) { | |
3245 | free_untracked_cache(uc); | |
3246 | uc = NULL; | |
3247 | } | |
3248 | return uc; | |
3249 | } | |
e931371a | 3250 | |
73f9145f NTND |
3251 | static void invalidate_one_directory(struct untracked_cache *uc, |
3252 | struct untracked_cache_dir *ucd) | |
3253 | { | |
3254 | uc->dir_invalidated++; | |
3255 | ucd->valid = 0; | |
3256 | ucd->untracked_nr = 0; | |
3257 | } | |
3258 | ||
3259 | /* | |
3260 | * Normally when an entry is added or removed from a directory, | |
3261 | * invalidating that directory is enough. No need to touch its | |
3262 | * ancestors. When a directory is shown as "foo/bar/" in git-status | |
3263 | * however, deleting or adding an entry may have cascading effect. | |
3264 | * | |
3265 | * Say the "foo/bar/file" has become untracked, we need to tell the | |
3266 | * untracked_cache_dir of "foo" that "bar/" is not an untracked | |
3267 | * directory any more (because "bar" is managed by foo as an untracked | |
3268 | * "file"). | |
3269 | * | |
3270 | * Similarly, if "foo/bar/file" moves from untracked to tracked and it | |
3271 | * was the last untracked entry in the entire "foo", we should show | |
3272 | * "foo/" instead. Which means we have to invalidate past "bar" up to | |
3273 | * "foo". | |
3274 | * | |
3275 | * This function traverses all directories from root to leaf. If there | |
3276 | * is a chance of one of the above cases happening, we invalidate back | |
3277 | * to root. Otherwise we just invalidate the leaf. There may be a more | |
3278 | * sophisticated way than checking for SHOW_OTHER_DIRECTORIES to | |
3279 | * detect these cases and avoid unnecessary invalidation, for example, | |
3280 | * checking for the untracked entry named "bar/" in "foo", but for now | |
3281 | * stick to something safe and simple. | |
3282 | */ | |
3283 | static int invalidate_one_component(struct untracked_cache *uc, | |
3284 | struct untracked_cache_dir *dir, | |
3285 | const char *path, int len) | |
3286 | { | |
3287 | const char *rest = strchr(path, '/'); | |
3288 | ||
3289 | if (rest) { | |
3290 | int component_len = rest - path; | |
3291 | struct untracked_cache_dir *d = | |
3292 | lookup_untracked(uc, dir, path, component_len); | |
3293 | int ret = | |
3294 | invalidate_one_component(uc, d, rest + 1, | |
3295 | len - (component_len + 1)); | |
3296 | if (ret) | |
3297 | invalidate_one_directory(uc, dir); | |
3298 | return ret; | |
3299 | } | |
3300 | ||
3301 | invalidate_one_directory(uc, dir); | |
3302 | return uc->dir_flags & DIR_SHOW_OTHER_DIRECTORIES; | |
3303 | } | |
3304 | ||
e931371a | 3305 | void untracked_cache_invalidate_path(struct index_state *istate, |
0cacebf0 | 3306 | const char *path, int safe_path) |
e931371a | 3307 | { |
e931371a NTND |
3308 | if (!istate->untracked || !istate->untracked->root) |
3309 | return; | |
68f95b26 | 3310 | if (!safe_path && !verify_path(path, 0)) |
0cacebf0 | 3311 | return; |
73f9145f NTND |
3312 | invalidate_one_component(istate->untracked, istate->untracked->root, |
3313 | path, strlen(path)); | |
e931371a NTND |
3314 | } |
3315 | ||
3316 | void untracked_cache_remove_from_index(struct index_state *istate, | |
3317 | const char *path) | |
3318 | { | |
0cacebf0 | 3319 | untracked_cache_invalidate_path(istate, path, 1); |
e931371a NTND |
3320 | } |
3321 | ||
3322 | void untracked_cache_add_to_index(struct index_state *istate, | |
3323 | const char *path) | |
3324 | { | |
0cacebf0 | 3325 | untracked_cache_invalidate_path(istate, path, 1); |
e931371a | 3326 | } |
47e83eb3 | 3327 | |
da62f786 SB |
3328 | static void connect_wt_gitdir_in_nested(const char *sub_worktree, |
3329 | const char *sub_gitdir) | |
3330 | { | |
3331 | int i; | |
3332 | struct repository subrepo; | |
3333 | struct strbuf sub_wt = STRBUF_INIT; | |
3334 | struct strbuf sub_gd = STRBUF_INIT; | |
3335 | ||
3336 | const struct submodule *sub; | |
3337 | ||
3338 | /* If the submodule has no working tree, we can ignore it. */ | |
3339 | if (repo_init(&subrepo, sub_gitdir, sub_worktree)) | |
3340 | return; | |
3341 | ||
3342 | if (repo_read_index(&subrepo) < 0) | |
a80897c1 | 3343 | die(_("index file corrupt in repo %s"), subrepo.gitdir); |
da62f786 SB |
3344 | |
3345 | for (i = 0; i < subrepo.index->cache_nr; i++) { | |
3346 | const struct cache_entry *ce = subrepo.index->cache[i]; | |
3347 | ||
3348 | if (!S_ISGITLINK(ce->ce_mode)) | |
3349 | continue; | |
3350 | ||
3351 | while (i + 1 < subrepo.index->cache_nr && | |
3352 | !strcmp(ce->name, subrepo.index->cache[i + 1]->name)) | |
3353 | /* | |
3354 | * Skip entries with the same name in different stages | |
3355 | * to make sure an entry is returned only once. | |
3356 | */ | |
3357 | i++; | |
3358 | ||
3359 | sub = submodule_from_path(&subrepo, &null_oid, ce->name); | |
3360 | if (!sub || !is_submodule_active(&subrepo, ce->name)) | |
3361 | /* .gitmodules broken or inactive sub */ | |
3362 | continue; | |
3363 | ||
3364 | strbuf_reset(&sub_wt); | |
3365 | strbuf_reset(&sub_gd); | |
3366 | strbuf_addf(&sub_wt, "%s/%s", sub_worktree, sub->path); | |
3367 | strbuf_addf(&sub_gd, "%s/modules/%s", sub_gitdir, sub->name); | |
3368 | ||
3369 | connect_work_tree_and_git_dir(sub_wt.buf, sub_gd.buf, 1); | |
3370 | } | |
3371 | strbuf_release(&sub_wt); | |
3372 | strbuf_release(&sub_gd); | |
3373 | repo_clear(&subrepo); | |
3374 | } | |
3375 | ||
3376 | void connect_work_tree_and_git_dir(const char *work_tree_, | |
3377 | const char *git_dir_, | |
3378 | int recurse_into_nested) | |
47e83eb3 | 3379 | { |
365444a6 SB |
3380 | struct strbuf gitfile_sb = STRBUF_INIT; |
3381 | struct strbuf cfg_sb = STRBUF_INIT; | |
47e83eb3 | 3382 | struct strbuf rel_path = STRBUF_INIT; |
365444a6 | 3383 | char *git_dir, *work_tree; |
47e83eb3 | 3384 | |
365444a6 SB |
3385 | /* Prepare .git file */ |
3386 | strbuf_addf(&gitfile_sb, "%s/.git", work_tree_); | |
3387 | if (safe_create_leading_directories_const(gitfile_sb.buf)) | |
3388 | die(_("could not create directories for %s"), gitfile_sb.buf); | |
3389 | ||
3390 | /* Prepare config file */ | |
3391 | strbuf_addf(&cfg_sb, "%s/config", git_dir_); | |
3392 | if (safe_create_leading_directories_const(cfg_sb.buf)) | |
3393 | die(_("could not create directories for %s"), cfg_sb.buf); | |
47e83eb3 | 3394 | |
e394fa01 JH |
3395 | git_dir = real_pathdup(git_dir_, 1); |
3396 | work_tree = real_pathdup(work_tree_, 1); | |
365444a6 SB |
3397 | |
3398 | /* Write .git file */ | |
3399 | write_file(gitfile_sb.buf, "gitdir: %s", | |
3400 | relative_path(git_dir, work_tree, &rel_path)); | |
47e83eb3 | 3401 | /* Update core.worktree setting */ |
365444a6 | 3402 | git_config_set_in_file(cfg_sb.buf, "core.worktree", |
47e83eb3 SB |
3403 | relative_path(work_tree, git_dir, &rel_path)); |
3404 | ||
365444a6 SB |
3405 | strbuf_release(&gitfile_sb); |
3406 | strbuf_release(&cfg_sb); | |
47e83eb3 | 3407 | strbuf_release(&rel_path); |
da62f786 SB |
3408 | |
3409 | if (recurse_into_nested) | |
3410 | connect_wt_gitdir_in_nested(work_tree, git_dir); | |
3411 | ||
47e83eb3 SB |
3412 | free(work_tree); |
3413 | free(git_dir); | |
3414 | } | |
f6f85861 SB |
3415 | |
3416 | /* | |
3417 | * Migrate the git directory of the given path from old_git_dir to new_git_dir. | |
3418 | */ | |
3419 | void relocate_gitdir(const char *path, const char *old_git_dir, const char *new_git_dir) | |
3420 | { | |
3421 | if (rename(old_git_dir, new_git_dir) < 0) | |
3422 | die_errno(_("could not migrate git directory from '%s' to '%s'"), | |
3423 | old_git_dir, new_git_dir); | |
3424 | ||
da62f786 | 3425 | connect_work_tree_and_git_dir(path, new_git_dir, 0); |
f6f85861 | 3426 | } |