]>
Commit | Line | Data |
---|---|---|
453ec4bd LT |
1 | /* |
2 | * This handles recursive filename detection with exclude | |
3 | * files, index knowledge etc.. | |
4 | * | |
5 | * Copyright (C) Linus Torvalds, 2005-2006 | |
6 | * Junio Hamano, 2005-2006 | |
7 | */ | |
453ec4bd | 8 | #include "cache.h" |
b2141fc1 | 9 | #include "config.h" |
453ec4bd | 10 | #include "dir.h" |
cbd53a21 | 11 | #include "object-store.h" |
b0db7046 | 12 | #include "attr.h" |
09595258 | 13 | #include "refs.h" |
237ec6e4 | 14 | #include "wildmatch.h" |
64acde94 | 15 | #include "pathspec.h" |
dde843e7 | 16 | #include "utf8.h" |
83c094ad NTND |
17 | #include "varint.h" |
18 | #include "ewah/ewok.h" | |
883e248b | 19 | #include "fsmonitor.h" |
da62f786 | 20 | #include "submodule-config.h" |
453ec4bd | 21 | |
defd7c7b KB |
22 | /* |
23 | * Tells read_directory_recursive how a file or directory should be treated. | |
24 | * Values are ordered by significance, e.g. if a directory contains both | |
25 | * excluded and untracked files, it is listed as untracked because | |
26 | * path_untracked > path_excluded. | |
27 | */ | |
28 | enum path_treatment { | |
29 | path_none = 0, | |
30 | path_recurse, | |
31 | path_excluded, | |
32 | path_untracked | |
33 | }; | |
34 | ||
cf7c6148 NTND |
35 | /* |
36 | * Support data structure for our opendir/readdir/closedir wrappers | |
37 | */ | |
38 | struct cached_dir { | |
39 | DIR *fdir; | |
40 | struct untracked_cache_dir *untracked; | |
91a2288b NTND |
41 | int nr_files; |
42 | int nr_dirs; | |
43 | ||
ad6f2157 JK |
44 | const char *d_name; |
45 | int d_type; | |
91a2288b NTND |
46 | const char *file; |
47 | struct untracked_cache_dir *ucd; | |
cf7c6148 NTND |
48 | }; |
49 | ||
defd7c7b | 50 | static enum path_treatment read_directory_recursive(struct dir_struct *dir, |
0ef8e169 BW |
51 | struct index_state *istate, const char *path, int len, |
52 | struct untracked_cache_dir *untracked, | |
5aaa7fd3 | 53 | int check_only, int stop_at_first_file, const struct pathspec *pathspec); |
ad6f2157 JK |
54 | static int resolve_dtype(int dtype, struct index_state *istate, |
55 | const char *path, int len); | |
09595258 | 56 | |
e0556a92 PC |
57 | int count_slashes(const char *s) |
58 | { | |
59 | int cnt = 0; | |
60 | while (*s) | |
61 | if (*s++ == '/') | |
62 | cnt++; | |
63 | return cnt; | |
64 | } | |
65 | ||
ba0897e6 | 66 | int fspathcmp(const char *a, const char *b) |
8cf2a84e JJ |
67 | { |
68 | return ignore_case ? strcasecmp(a, b) : strcmp(a, b); | |
69 | } | |
70 | ||
ba0897e6 | 71 | int fspathncmp(const char *a, const char *b, size_t count) |
8cf2a84e JJ |
72 | { |
73 | return ignore_case ? strncasecmp(a, b, count) : strncmp(a, b, count); | |
74 | } | |
75 | ||
1f26ce61 CB |
76 | int git_fnmatch(const struct pathspec_item *item, |
77 | const char *pattern, const char *string, | |
78 | int prefix) | |
5d74762d | 79 | { |
5d74762d | 80 | if (prefix > 0) { |
93d93537 | 81 | if (ps_strncmp(item, pattern, string, prefix)) |
eb07894f | 82 | return WM_NOMATCH; |
5d74762d NTND |
83 | pattern += prefix; |
84 | string += prefix; | |
85 | } | |
bd30c2e4 | 86 | if (item->flags & PATHSPEC_ONESTAR) { |
8c6abbcd NTND |
87 | int pattern_len = strlen(++pattern); |
88 | int string_len = strlen(string); | |
89 | return string_len < pattern_len || | |
93d93537 NTND |
90 | ps_strcmp(item, pattern, |
91 | string + string_len - pattern_len); | |
8c6abbcd | 92 | } |
bd30c2e4 | 93 | if (item->magic & PATHSPEC_GLOB) |
93d93537 NTND |
94 | return wildmatch(pattern, string, |
95 | WM_PATHNAME | | |
55d34269 | 96 | (item->magic & PATHSPEC_ICASE ? WM_CASEFOLD : 0)); |
bd30c2e4 NTND |
97 | else |
98 | /* wildmatch has not learned no FNM_PATHNAME mode yet */ | |
eb07894f | 99 | return wildmatch(pattern, string, |
55d34269 | 100 | item->magic & PATHSPEC_ICASE ? WM_CASEFOLD : 0); |
5d74762d NTND |
101 | } |
102 | ||
0b6e56df JH |
103 | static int fnmatch_icase_mem(const char *pattern, int patternlen, |
104 | const char *string, int stringlen, | |
105 | int flags) | |
106 | { | |
107 | int match_status; | |
108 | struct strbuf pat_buf = STRBUF_INIT; | |
109 | struct strbuf str_buf = STRBUF_INIT; | |
110 | const char *use_pat = pattern; | |
111 | const char *use_str = string; | |
112 | ||
113 | if (pattern[patternlen]) { | |
114 | strbuf_add(&pat_buf, pattern, patternlen); | |
115 | use_pat = pat_buf.buf; | |
116 | } | |
117 | if (string[stringlen]) { | |
118 | strbuf_add(&str_buf, string, stringlen); | |
119 | use_str = str_buf.buf; | |
120 | } | |
121 | ||
f30366b2 JH |
122 | if (ignore_case) |
123 | flags |= WM_CASEFOLD; | |
55d34269 | 124 | match_status = wildmatch(use_pat, use_str, flags); |
0b6e56df JH |
125 | |
126 | strbuf_release(&pat_buf); | |
127 | strbuf_release(&str_buf); | |
128 | ||
129 | return match_status; | |
130 | } | |
131 | ||
827f4d6c | 132 | static size_t common_prefix_len(const struct pathspec *pathspec) |
3c6a370b | 133 | { |
827f4d6c | 134 | int n; |
4a085b16 | 135 | size_t max = 0; |
3c6a370b | 136 | |
93d93537 NTND |
137 | /* |
138 | * ":(icase)path" is treated as a pathspec full of | |
139 | * wildcard. In other words, only prefix is considered common | |
140 | * prefix. If the pathspec is abc/foo abc/bar, running in | |
bbbb6b0b | 141 | * subdir xyz, the common prefix is still xyz, not xyz/abc as |
93d93537 NTND |
142 | * in non-:(icase). |
143 | */ | |
5c6933d2 NTND |
144 | GUARD_PATHSPEC(pathspec, |
145 | PATHSPEC_FROMTOP | | |
146 | PATHSPEC_MAXDEPTH | | |
bd30c2e4 | 147 | PATHSPEC_LITERAL | |
93d93537 | 148 | PATHSPEC_GLOB | |
ef79b1f8 | 149 | PATHSPEC_ICASE | |
b0db7046 BW |
150 | PATHSPEC_EXCLUDE | |
151 | PATHSPEC_ATTR); | |
4a085b16 | 152 | |
827f4d6c | 153 | for (n = 0; n < pathspec->nr; n++) { |
93d93537 | 154 | size_t i = 0, len = 0, item_len; |
ef79b1f8 NTND |
155 | if (pathspec->items[n].magic & PATHSPEC_EXCLUDE) |
156 | continue; | |
93d93537 NTND |
157 | if (pathspec->items[n].magic & PATHSPEC_ICASE) |
158 | item_len = pathspec->items[n].prefix; | |
159 | else | |
160 | item_len = pathspec->items[n].nowildcard_len; | |
161 | while (i < item_len && (n == 0 || i < max)) { | |
827f4d6c NTND |
162 | char c = pathspec->items[n].match[i]; |
163 | if (c != pathspec->items[0].match[i]) | |
4a085b16 JH |
164 | break; |
165 | if (c == '/') | |
166 | len = i + 1; | |
827f4d6c | 167 | i++; |
4a085b16 | 168 | } |
827f4d6c | 169 | if (n == 0 || len < max) { |
4a085b16 JH |
170 | max = len; |
171 | if (!max) | |
172 | break; | |
173 | } | |
3c6a370b | 174 | } |
4a085b16 | 175 | return max; |
3c6a370b LT |
176 | } |
177 | ||
f950eb95 CB |
178 | /* |
179 | * Returns a copy of the longest leading path common among all | |
180 | * pathspecs. | |
181 | */ | |
827f4d6c | 182 | char *common_prefix(const struct pathspec *pathspec) |
f950eb95 CB |
183 | { |
184 | unsigned long len = common_prefix_len(pathspec); | |
185 | ||
827f4d6c | 186 | return len ? xmemdupz(pathspec->items[0].match, len) : NULL; |
f950eb95 CB |
187 | } |
188 | ||
0d32c183 BW |
189 | int fill_directory(struct dir_struct *dir, |
190 | struct index_state *istate, | |
191 | const struct pathspec *pathspec) | |
1d8842d9 | 192 | { |
bec5ab89 | 193 | const char *prefix; |
966de302 | 194 | size_t prefix_len; |
1d8842d9 LT |
195 | |
196 | /* | |
197 | * Calculate common prefix for the pathspec, and | |
198 | * use that to optimize the directory walk | |
199 | */ | |
bec5ab89 RS |
200 | prefix_len = common_prefix_len(pathspec); |
201 | prefix = prefix_len ? pathspec->items[0].match : ""; | |
1d8842d9 LT |
202 | |
203 | /* Read the directory and prune it */ | |
0d32c183 | 204 | read_directory(dir, istate, prefix, prefix_len, pathspec); |
966de302 | 205 | |
966de302 | 206 | return prefix_len; |
1d8842d9 LT |
207 | } |
208 | ||
bc96cc87 NTND |
209 | int within_depth(const char *name, int namelen, |
210 | int depth, int max_depth) | |
211 | { | |
212 | const char *cp = name, *cpe = name + namelen; | |
213 | ||
214 | while (cp < cpe) { | |
215 | if (*cp++ != '/') | |
216 | continue; | |
217 | depth++; | |
218 | if (depth > max_depth) | |
219 | return 0; | |
220 | } | |
221 | return 1; | |
222 | } | |
223 | ||
578d81d0 JH |
224 | /* |
225 | * Read the contents of the blob with the given OID into a buffer. | |
226 | * Append a trailing LF to the end if the last line doesn't have one. | |
227 | * | |
228 | * Returns: | |
229 | * -1 when the OID is invalid or unknown or does not refer to a blob. | |
230 | * 0 when the blob is empty. | |
231 | * 1 along with { data, size } of the (possibly augmented) buffer | |
232 | * when successful. | |
233 | * | |
4b33e602 | 234 | * Optionally updates the given oid_stat with the given OID (when valid). |
578d81d0 | 235 | */ |
4b33e602 PO |
236 | static int do_read_blob(const struct object_id *oid, struct oid_stat *oid_stat, |
237 | size_t *size_out, char **data_out) | |
578d81d0 JH |
238 | { |
239 | enum object_type type; | |
240 | unsigned long sz; | |
241 | char *data; | |
242 | ||
243 | *size_out = 0; | |
244 | *data_out = NULL; | |
245 | ||
b4f5aca4 | 246 | data = read_object_file(oid, &type, &sz); |
578d81d0 JH |
247 | if (!data || type != OBJ_BLOB) { |
248 | free(data); | |
249 | return -1; | |
250 | } | |
251 | ||
4b33e602 PO |
252 | if (oid_stat) { |
253 | memset(&oid_stat->stat, 0, sizeof(oid_stat->stat)); | |
254 | oidcpy(&oid_stat->oid, oid); | |
578d81d0 JH |
255 | } |
256 | ||
257 | if (sz == 0) { | |
258 | free(data); | |
259 | return 0; | |
260 | } | |
261 | ||
262 | if (data[sz - 1] != '\n') { | |
263 | data = xrealloc(data, st_add(sz, 1)); | |
264 | data[sz++] = '\n'; | |
265 | } | |
266 | ||
267 | *size_out = xsize_t(sz); | |
268 | *data_out = data; | |
269 | ||
270 | return 1; | |
271 | } | |
272 | ||
75a6315f BW |
273 | #define DO_MATCH_EXCLUDE (1<<0) |
274 | #define DO_MATCH_DIRECTORY (1<<1) | |
a3d89d8f | 275 | #define DO_MATCH_LEADING_PATHSPEC (1<<2) |
42b0874a | 276 | |
61cf2820 | 277 | /* |
29b577b9 | 278 | * Does the given pathspec match the given name? A match is found if |
61cf2820 | 279 | * |
29b577b9 EN |
280 | * (1) the pathspec string is leading directory of 'name' ("RECURSIVELY"), or |
281 | * (2) the pathspec string has a leading part matching 'name' ("LEADING"), or | |
282 | * (3) the pathspec string is a wildcard and matches 'name' ("WILDCARD"), or | |
283 | * (4) the pathspec string is exactly the same as 'name' ("EXACT"). | |
61cf2820 | 284 | * |
29b577b9 | 285 | * Return value tells which case it was (1-4), or 0 when there is no match. |
61cf2820 | 286 | * |
29b577b9 EN |
287 | * It may be instructive to look at a small table of concrete examples |
288 | * to understand the differences between 1, 2, and 4: | |
289 | * | |
290 | * Pathspecs | |
291 | * | a/b | a/b/ | a/b/c | |
292 | * ------+-----------+-----------+------------ | |
293 | * a/b | EXACT | EXACT[1] | LEADING[2] | |
294 | * Names a/b/ | RECURSIVE | EXACT | LEADING[2] | |
295 | * a/b/c | RECURSIVE | RECURSIVE | EXACT | |
296 | * | |
297 | * [1] Only if DO_MATCH_DIRECTORY is passed; otherwise, this is NOT a match. | |
298 | * [2] Only if DO_MATCH_LEADING_PATHSPEC is passed; otherwise, not a match. | |
61cf2820 | 299 | */ |
6d2df284 NTND |
300 | static int match_pathspec_item(const struct index_state *istate, |
301 | const struct pathspec_item *item, int prefix, | |
42b0874a | 302 | const char *name, int namelen, unsigned flags) |
61cf2820 NTND |
303 | { |
304 | /* name/namelen has prefix cut off by caller */ | |
305 | const char *match = item->match + prefix; | |
306 | int matchlen = item->len - prefix; | |
307 | ||
93d93537 NTND |
308 | /* |
309 | * The normal call pattern is: | |
310 | * 1. prefix = common_prefix_len(ps); | |
311 | * 2. prune something, or fill_directory | |
854b0959 | 312 | * 3. match_pathspec() |
93d93537 NTND |
313 | * |
314 | * 'prefix' at #1 may be shorter than the command's prefix and | |
315 | * it's ok for #2 to match extra files. Those extras will be | |
316 | * trimmed at #3. | |
317 | * | |
318 | * Suppose the pathspec is 'foo' and '../bar' running from | |
319 | * subdir 'xyz'. The common prefix at #1 will be empty, thanks | |
320 | * to "../". We may have xyz/foo _and_ XYZ/foo after #2. The | |
321 | * user does not want XYZ/foo, only the "foo" part should be | |
322 | * case-insensitive. We need to filter out XYZ/foo here. In | |
323 | * other words, we do not trust the caller on comparing the | |
324 | * prefix part when :(icase) is involved. We do exact | |
325 | * comparison ourselves. | |
326 | * | |
327 | * Normally the caller (common_prefix_len() in fact) does | |
328 | * _exact_ matching on name[-prefix+1..-1] and we do not need | |
329 | * to check that part. Be defensive and check it anyway, in | |
330 | * case common_prefix_len is changed, or a new caller is | |
331 | * introduced that does not use common_prefix_len. | |
332 | * | |
333 | * If the penalty turns out too high when prefix is really | |
334 | * long, maybe change it to | |
335 | * strncmp(match, name, item->prefix - prefix) | |
336 | */ | |
337 | if (item->prefix && (item->magic & PATHSPEC_ICASE) && | |
338 | strncmp(item->match, name - prefix, item->prefix)) | |
339 | return 0; | |
340 | ||
22af33be NTND |
341 | if (item->attr_match_nr && |
342 | !match_pathspec_attrs(istate, name, namelen, item)) | |
b0db7046 BW |
343 | return 0; |
344 | ||
61cf2820 NTND |
345 | /* If the match was just the prefix, we matched */ |
346 | if (!*match) | |
347 | return MATCHED_RECURSIVELY; | |
348 | ||
93d93537 | 349 | if (matchlen <= namelen && !ps_strncmp(item, match, name, matchlen)) { |
61cf2820 NTND |
350 | if (matchlen == namelen) |
351 | return MATCHED_EXACTLY; | |
352 | ||
353 | if (match[matchlen-1] == '/' || name[matchlen] == '/') | |
354 | return MATCHED_RECURSIVELY; | |
68690fdd NTND |
355 | } else if ((flags & DO_MATCH_DIRECTORY) && |
356 | match[matchlen - 1] == '/' && | |
357 | namelen == matchlen - 1 && | |
358 | !ps_strncmp(item, match, name, namelen)) | |
359 | return MATCHED_EXACTLY; | |
61cf2820 | 360 | |
5d74762d | 361 | if (item->nowildcard_len < item->len && |
bd30c2e4 | 362 | !git_fnmatch(item, match, name, |
8c6abbcd | 363 | item->nowildcard_len - prefix)) |
61cf2820 NTND |
364 | return MATCHED_FNMATCH; |
365 | ||
29b577b9 | 366 | /* Perform checks to see if "name" is a leading string of the pathspec */ |
a3d89d8f | 367 | if (flags & DO_MATCH_LEADING_PATHSPEC) { |
75a6315f | 368 | /* name is a literal prefix of the pathspec */ |
a5e916c7 | 369 | int offset = name[namelen-1] == '/' ? 1 : 0; |
75a6315f | 370 | if ((namelen < matchlen) && |
a5e916c7 | 371 | (match[namelen-offset] == '/') && |
75a6315f | 372 | !ps_strncmp(item, match, name, namelen)) |
89a1f4aa | 373 | return MATCHED_RECURSIVELY_LEADING_PATHSPEC; |
75a6315f | 374 | |
2f5d3847 | 375 | /* name doesn't match up to the first wild character */ |
75a6315f BW |
376 | if (item->nowildcard_len < item->len && |
377 | ps_strncmp(item, match, name, | |
378 | item->nowildcard_len - prefix)) | |
379 | return 0; | |
380 | ||
072a2310 EN |
381 | /* |
382 | * name has no wildcard, and it didn't match as a leading | |
383 | * pathspec so return. | |
384 | */ | |
385 | if (item->nowildcard_len == item->len) | |
386 | return 0; | |
387 | ||
75a6315f BW |
388 | /* |
389 | * Here is where we would perform a wildmatch to check if | |
390 | * "name" can be matched as a directory (or a prefix) against | |
391 | * the pathspec. Since wildmatch doesn't have this capability | |
392 | * at the present we have to punt and say that it is a match, | |
393 | * potentially returning a false positive | |
394 | * The submodules themselves will be able to perform more | |
395 | * accurate matching to determine if the pathspec matches. | |
396 | */ | |
89a1f4aa | 397 | return MATCHED_RECURSIVELY_LEADING_PATHSPEC; |
75a6315f BW |
398 | } |
399 | ||
61cf2820 NTND |
400 | return 0; |
401 | } | |
402 | ||
403 | /* | |
52ed1894 AS |
404 | * Given a name and a list of pathspecs, returns the nature of the |
405 | * closest (i.e. most specific) match of the name to any of the | |
406 | * pathspecs. | |
407 | * | |
408 | * The caller typically calls this multiple times with the same | |
409 | * pathspec and seen[] array but with different name/namelen | |
410 | * (e.g. entries from the index) and is interested in seeing if and | |
411 | * how each pathspec matches all the names it calls this function | |
412 | * with. A mark is left in the seen[] array for each pathspec element | |
413 | * indicating the closest type of match that element achieved, so if | |
414 | * seen[n] remains zero after multiple invocations, that means the nth | |
415 | * pathspec did not match any names, which could indicate that the | |
416 | * user mistyped the nth pathspec. | |
61cf2820 | 417 | */ |
6d2df284 NTND |
418 | static int do_match_pathspec(const struct index_state *istate, |
419 | const struct pathspec *ps, | |
854b0959 NTND |
420 | const char *name, int namelen, |
421 | int prefix, char *seen, | |
42b0874a | 422 | unsigned flags) |
61cf2820 | 423 | { |
42b0874a | 424 | int i, retval = 0, exclude = flags & DO_MATCH_EXCLUDE; |
61cf2820 | 425 | |
5c6933d2 NTND |
426 | GUARD_PATHSPEC(ps, |
427 | PATHSPEC_FROMTOP | | |
428 | PATHSPEC_MAXDEPTH | | |
bd30c2e4 | 429 | PATHSPEC_LITERAL | |
93d93537 | 430 | PATHSPEC_GLOB | |
ef79b1f8 | 431 | PATHSPEC_ICASE | |
b0db7046 BW |
432 | PATHSPEC_EXCLUDE | |
433 | PATHSPEC_ATTR); | |
8f4f8f45 | 434 | |
61cf2820 | 435 | if (!ps->nr) { |
6330a171 NTND |
436 | if (!ps->recursive || |
437 | !(ps->magic & PATHSPEC_MAXDEPTH) || | |
438 | ps->max_depth == -1) | |
61cf2820 NTND |
439 | return MATCHED_RECURSIVELY; |
440 | ||
441 | if (within_depth(name, namelen, 0, ps->max_depth)) | |
442 | return MATCHED_EXACTLY; | |
443 | else | |
444 | return 0; | |
445 | } | |
446 | ||
447 | name += prefix; | |
448 | namelen -= prefix; | |
449 | ||
450 | for (i = ps->nr - 1; i >= 0; i--) { | |
451 | int how; | |
ef79b1f8 NTND |
452 | |
453 | if ((!exclude && ps->items[i].magic & PATHSPEC_EXCLUDE) || | |
454 | ( exclude && !(ps->items[i].magic & PATHSPEC_EXCLUDE))) | |
455 | continue; | |
456 | ||
61cf2820 NTND |
457 | if (seen && seen[i] == MATCHED_EXACTLY) |
458 | continue; | |
ef79b1f8 NTND |
459 | /* |
460 | * Make exclude patterns optional and never report | |
461 | * "pathspec ':(exclude)foo' matches no files" | |
462 | */ | |
463 | if (seen && ps->items[i].magic & PATHSPEC_EXCLUDE) | |
464 | seen[i] = MATCHED_FNMATCH; | |
6d2df284 | 465 | how = match_pathspec_item(istate, ps->items+i, prefix, name, |
42b0874a | 466 | namelen, flags); |
6330a171 NTND |
467 | if (ps->recursive && |
468 | (ps->magic & PATHSPEC_MAXDEPTH) && | |
469 | ps->max_depth != -1 && | |
61cf2820 NTND |
470 | how && how != MATCHED_FNMATCH) { |
471 | int len = ps->items[i].len; | |
472 | if (name[len] == '/') | |
473 | len++; | |
474 | if (within_depth(name+len, namelen-len, 0, ps->max_depth)) | |
475 | how = MATCHED_EXACTLY; | |
476 | else | |
477 | how = 0; | |
478 | } | |
479 | if (how) { | |
480 | if (retval < how) | |
481 | retval = how; | |
482 | if (seen && seen[i] < how) | |
483 | seen[i] = how; | |
484 | } | |
485 | } | |
486 | return retval; | |
487 | } | |
488 | ||
6d2df284 NTND |
489 | int match_pathspec(const struct index_state *istate, |
490 | const struct pathspec *ps, | |
854b0959 | 491 | const char *name, int namelen, |
ae8d0824 | 492 | int prefix, char *seen, int is_dir) |
ef79b1f8 NTND |
493 | { |
494 | int positive, negative; | |
ae8d0824 | 495 | unsigned flags = is_dir ? DO_MATCH_DIRECTORY : 0; |
6d2df284 | 496 | positive = do_match_pathspec(istate, ps, name, namelen, |
42b0874a | 497 | prefix, seen, flags); |
ef79b1f8 NTND |
498 | if (!(ps->magic & PATHSPEC_EXCLUDE) || !positive) |
499 | return positive; | |
6d2df284 | 500 | negative = do_match_pathspec(istate, ps, name, namelen, |
42b0874a NTND |
501 | prefix, seen, |
502 | flags | DO_MATCH_EXCLUDE); | |
ef79b1f8 NTND |
503 | return negative ? 0 : positive; |
504 | } | |
505 | ||
75a6315f BW |
506 | /** |
507 | * Check if a submodule is a superset of the pathspec | |
508 | */ | |
6d2df284 NTND |
509 | int submodule_path_match(const struct index_state *istate, |
510 | const struct pathspec *ps, | |
75a6315f BW |
511 | const char *submodule_name, |
512 | char *seen) | |
513 | { | |
6d2df284 | 514 | int matched = do_match_pathspec(istate, ps, submodule_name, |
75a6315f BW |
515 | strlen(submodule_name), |
516 | 0, seen, | |
517 | DO_MATCH_DIRECTORY | | |
a3d89d8f | 518 | DO_MATCH_LEADING_PATHSPEC); |
75a6315f BW |
519 | return matched; |
520 | } | |
521 | ||
777c55a6 | 522 | int report_path_error(const char *ps_matched, |
c5c33504 | 523 | const struct pathspec *pathspec) |
777c55a6 JH |
524 | { |
525 | /* | |
526 | * Make sure all pathspec matched; otherwise it is an error. | |
527 | */ | |
777c55a6 JH |
528 | int num, errors = 0; |
529 | for (num = 0; num < pathspec->nr; num++) { | |
530 | int other, found_dup; | |
531 | ||
532 | if (ps_matched[num]) | |
533 | continue; | |
534 | /* | |
535 | * The caller might have fed identical pathspec | |
536 | * twice. Do not barf on such a mistake. | |
537 | * FIXME: parse_pathspec should have eliminated | |
538 | * duplicate pathspec. | |
539 | */ | |
540 | for (found_dup = other = 0; | |
541 | !found_dup && other < pathspec->nr; | |
542 | other++) { | |
543 | if (other == num || !ps_matched[other]) | |
544 | continue; | |
545 | if (!strcmp(pathspec->items[other].original, | |
546 | pathspec->items[num].original)) | |
547 | /* | |
548 | * Ok, we have a match already. | |
549 | */ | |
550 | found_dup = 1; | |
551 | } | |
552 | if (found_dup) | |
553 | continue; | |
554 | ||
a80897c1 | 555 | error(_("pathspec '%s' did not match any file(s) known to git"), |
777c55a6 JH |
556 | pathspec->items[num].original); |
557 | errors++; | |
558 | } | |
777c55a6 JH |
559 | return errors; |
560 | } | |
561 | ||
fcd631ed NTND |
562 | /* |
563 | * Return the length of the "simple" part of a path match limiter. | |
564 | */ | |
87323bda | 565 | int simple_length(const char *match) |
fcd631ed NTND |
566 | { |
567 | int len = -1; | |
568 | ||
569 | for (;;) { | |
570 | unsigned char c = *match++; | |
571 | len++; | |
572 | if (c == '\0' || is_glob_special(c)) | |
573 | return len; | |
574 | } | |
575 | } | |
576 | ||
87323bda | 577 | int no_wildcard(const char *string) |
68492fc7 | 578 | { |
fcd631ed | 579 | return string[simple_length(string)] == '\0'; |
68492fc7 LK |
580 | } |
581 | ||
65edd96a | 582 | void parse_path_pattern(const char **pattern, |
82dce998 | 583 | int *patternlen, |
f8708998 | 584 | unsigned *flags, |
82dce998 | 585 | int *nowildcardlen) |
84460eec NTND |
586 | { |
587 | const char *p = *pattern; | |
588 | size_t i, len; | |
589 | ||
590 | *flags = 0; | |
591 | if (*p == '!') { | |
4ff89ee5 | 592 | *flags |= PATTERN_FLAG_NEGATIVE; |
84460eec NTND |
593 | p++; |
594 | } | |
595 | len = strlen(p); | |
596 | if (len && p[len - 1] == '/') { | |
597 | len--; | |
4ff89ee5 | 598 | *flags |= PATTERN_FLAG_MUSTBEDIR; |
84460eec NTND |
599 | } |
600 | for (i = 0; i < len; i++) { | |
601 | if (p[i] == '/') | |
602 | break; | |
603 | } | |
604 | if (i == len) | |
4ff89ee5 | 605 | *flags |= PATTERN_FLAG_NODIR; |
84460eec NTND |
606 | *nowildcardlen = simple_length(p); |
607 | /* | |
608 | * we should have excluded the trailing slash from 'p' too, | |
609 | * but that's one more allocation. Instead just make sure | |
610 | * nowildcardlen does not exceed real patternlen | |
611 | */ | |
612 | if (*nowildcardlen > len) | |
613 | *nowildcardlen = len; | |
614 | if (*p == '*' && no_wildcard(p + 1)) | |
4ff89ee5 | 615 | *flags |= PATTERN_FLAG_ENDSWITH; |
84460eec NTND |
616 | *pattern = p; |
617 | *patternlen = len; | |
618 | } | |
619 | ||
af09ce24 DS |
620 | int pl_hashmap_cmp(const void *unused_cmp_data, |
621 | const struct hashmap_entry *a, | |
622 | const struct hashmap_entry *b, | |
623 | const void *key) | |
96cc8ab5 DS |
624 | { |
625 | const struct pattern_entry *ee1 = | |
626 | container_of(a, struct pattern_entry, ent); | |
627 | const struct pattern_entry *ee2 = | |
628 | container_of(b, struct pattern_entry, ent); | |
629 | ||
630 | size_t min_len = ee1->patternlen <= ee2->patternlen | |
631 | ? ee1->patternlen | |
632 | : ee2->patternlen; | |
633 | ||
190a65f9 DS |
634 | if (ignore_case) |
635 | return strncasecmp(ee1->pattern, ee2->pattern, min_len); | |
96cc8ab5 DS |
636 | return strncmp(ee1->pattern, ee2->pattern, min_len); |
637 | } | |
638 | ||
639 | static void add_pattern_to_hashsets(struct pattern_list *pl, struct path_pattern *given) | |
640 | { | |
641 | struct pattern_entry *translated; | |
642 | char *truncated; | |
643 | char *data = NULL; | |
644 | ||
645 | if (!pl->use_cone_patterns) | |
646 | return; | |
647 | ||
648 | if (given->flags & PATTERN_FLAG_NEGATIVE && | |
649 | given->flags & PATTERN_FLAG_MUSTBEDIR && | |
650 | !strcmp(given->pattern, "/*")) { | |
651 | pl->full_cone = 0; | |
652 | return; | |
653 | } | |
654 | ||
655 | if (!given->flags && !strcmp(given->pattern, "/*")) { | |
656 | pl->full_cone = 1; | |
657 | return; | |
658 | } | |
659 | ||
660 | if (given->patternlen > 2 && | |
661 | !strcmp(given->pattern + given->patternlen - 2, "/*")) { | |
662 | if (!(given->flags & PATTERN_FLAG_NEGATIVE)) { | |
663 | /* Not a cone pattern. */ | |
664 | pl->use_cone_patterns = 0; | |
665 | warning(_("unrecognized pattern: '%s'"), given->pattern); | |
666 | goto clear_hashmaps; | |
667 | } | |
668 | ||
669 | truncated = xstrdup(given->pattern); | |
670 | truncated[given->patternlen - 2] = 0; | |
671 | ||
672 | translated = xmalloc(sizeof(struct pattern_entry)); | |
673 | translated->pattern = truncated; | |
674 | translated->patternlen = given->patternlen - 2; | |
675 | hashmap_entry_init(&translated->ent, | |
190a65f9 DS |
676 | ignore_case ? |
677 | strihash(translated->pattern) : | |
678 | strhash(translated->pattern)); | |
96cc8ab5 DS |
679 | |
680 | if (!hashmap_get_entry(&pl->recursive_hashmap, | |
681 | translated, ent, NULL)) { | |
682 | /* We did not see the "parent" included */ | |
683 | warning(_("unrecognized negative pattern: '%s'"), | |
684 | given->pattern); | |
685 | free(truncated); | |
686 | free(translated); | |
687 | goto clear_hashmaps; | |
688 | } | |
689 | ||
690 | hashmap_add(&pl->parent_hashmap, &translated->ent); | |
691 | hashmap_remove(&pl->recursive_hashmap, &translated->ent, &data); | |
692 | free(data); | |
693 | return; | |
694 | } | |
695 | ||
696 | if (given->flags & PATTERN_FLAG_NEGATIVE) { | |
697 | warning(_("unrecognized negative pattern: '%s'"), | |
698 | given->pattern); | |
699 | goto clear_hashmaps; | |
700 | } | |
701 | ||
702 | translated = xmalloc(sizeof(struct pattern_entry)); | |
703 | ||
704 | translated->pattern = xstrdup(given->pattern); | |
705 | translated->patternlen = given->patternlen; | |
706 | hashmap_entry_init(&translated->ent, | |
190a65f9 DS |
707 | ignore_case ? |
708 | strihash(translated->pattern) : | |
709 | strhash(translated->pattern)); | |
96cc8ab5 DS |
710 | |
711 | hashmap_add(&pl->recursive_hashmap, &translated->ent); | |
712 | ||
713 | if (hashmap_get_entry(&pl->parent_hashmap, translated, ent, NULL)) { | |
714 | /* we already included this at the parent level */ | |
715 | warning(_("your sparse-checkout file may have issues: pattern '%s' is repeated"), | |
716 | given->pattern); | |
717 | hashmap_remove(&pl->parent_hashmap, &translated->ent, &data); | |
718 | free(data); | |
719 | free(translated); | |
720 | } | |
721 | ||
722 | return; | |
723 | ||
724 | clear_hashmaps: | |
725 | warning(_("disabling cone pattern matching")); | |
726 | hashmap_free_entries(&pl->parent_hashmap, struct pattern_entry, ent); | |
727 | hashmap_free_entries(&pl->recursive_hashmap, struct pattern_entry, ent); | |
728 | pl->use_cone_patterns = 0; | |
729 | } | |
730 | ||
731 | static int hashmap_contains_path(struct hashmap *map, | |
732 | struct strbuf *pattern) | |
733 | { | |
734 | struct pattern_entry p; | |
735 | ||
736 | /* Check straight mapping */ | |
737 | p.pattern = pattern->buf; | |
738 | p.patternlen = pattern->len; | |
190a65f9 DS |
739 | hashmap_entry_init(&p.ent, |
740 | ignore_case ? | |
741 | strihash(p.pattern) : | |
742 | strhash(p.pattern)); | |
96cc8ab5 DS |
743 | return !!hashmap_get_entry(map, &p, ent, NULL); |
744 | } | |
745 | ||
746 | int hashmap_contains_parent(struct hashmap *map, | |
747 | const char *path, | |
748 | struct strbuf *buffer) | |
749 | { | |
750 | char *slash_pos; | |
751 | ||
752 | strbuf_setlen(buffer, 0); | |
753 | ||
754 | if (path[0] != '/') | |
755 | strbuf_addch(buffer, '/'); | |
756 | ||
757 | strbuf_addstr(buffer, path); | |
758 | ||
759 | slash_pos = strrchr(buffer->buf, '/'); | |
760 | ||
761 | while (slash_pos > buffer->buf) { | |
762 | strbuf_setlen(buffer, slash_pos - buffer->buf); | |
763 | ||
764 | if (hashmap_contains_path(map, buffer)) | |
765 | return 1; | |
766 | ||
767 | slash_pos = strrchr(buffer->buf, '/'); | |
768 | } | |
769 | ||
770 | return 0; | |
771 | } | |
772 | ||
65edd96a | 773 | void add_pattern(const char *string, const char *base, |
caa3d554 | 774 | int baselen, struct pattern_list *pl, int srcpos) |
453ec4bd | 775 | { |
ab8db613 | 776 | struct path_pattern *pattern; |
84460eec | 777 | int patternlen; |
f8708998 | 778 | unsigned flags; |
84460eec | 779 | int nowildcardlen; |
453ec4bd | 780 | |
65edd96a | 781 | parse_path_pattern(&string, &patternlen, &flags, &nowildcardlen); |
4ff89ee5 | 782 | if (flags & PATTERN_FLAG_MUSTBEDIR) { |
ab8db613 | 783 | FLEXPTR_ALLOC_MEM(pattern, pattern, string, patternlen); |
d6b8fc30 | 784 | } else { |
ab8db613 DS |
785 | pattern = xmalloc(sizeof(*pattern)); |
786 | pattern->pattern = string; | |
d6b8fc30 | 787 | } |
ab8db613 DS |
788 | pattern->patternlen = patternlen; |
789 | pattern->nowildcardlen = nowildcardlen; | |
790 | pattern->base = base; | |
791 | pattern->baselen = baselen; | |
792 | pattern->flags = flags; | |
793 | pattern->srcpos = srcpos; | |
caa3d554 DS |
794 | ALLOC_GROW(pl->patterns, pl->nr + 1, pl->alloc); |
795 | pl->patterns[pl->nr++] = pattern; | |
796 | pattern->pl = pl; | |
96cc8ab5 DS |
797 | |
798 | add_pattern_to_hashsets(pl, pattern); | |
453ec4bd LT |
799 | } |
800 | ||
578d81d0 JH |
801 | static int read_skip_worktree_file_from_index(const struct index_state *istate, |
802 | const char *path, | |
4b33e602 PO |
803 | size_t *size_out, char **data_out, |
804 | struct oid_stat *oid_stat) | |
c28b3d6e NTND |
805 | { |
806 | int pos, len; | |
c28b3d6e NTND |
807 | |
808 | len = strlen(path); | |
6f52b741 | 809 | pos = index_name_pos(istate, path, len); |
c28b3d6e | 810 | if (pos < 0) |
578d81d0 | 811 | return -1; |
6f52b741 | 812 | if (!ce_skip_worktree(istate->cache[pos])) |
578d81d0 JH |
813 | return -1; |
814 | ||
4b33e602 | 815 | return do_read_blob(&istate->cache[pos]->oid, oid_stat, size_out, data_out); |
c28b3d6e NTND |
816 | } |
817 | ||
f6198812 | 818 | /* |
caa3d554 DS |
819 | * Frees memory within pl which was allocated for exclude patterns and |
820 | * the file buffer. Does not free pl itself. | |
f6198812 | 821 | */ |
65edd96a | 822 | void clear_pattern_list(struct pattern_list *pl) |
0fd0e241 NTND |
823 | { |
824 | int i; | |
825 | ||
caa3d554 DS |
826 | for (i = 0; i < pl->nr; i++) |
827 | free(pl->patterns[i]); | |
828 | free(pl->patterns); | |
829 | free(pl->filebuf); | |
0fd0e241 | 830 | |
caa3d554 | 831 | memset(pl, 0, sizeof(*pl)); |
0fd0e241 NTND |
832 | } |
833 | ||
7e2e4b37 | 834 | static void trim_trailing_spaces(char *buf) |
16402b99 | 835 | { |
e61a6c1d PB |
836 | char *p, *last_space = NULL; |
837 | ||
838 | for (p = buf; *p; p++) | |
839 | switch (*p) { | |
840 | case ' ': | |
841 | if (!last_space) | |
842 | last_space = p; | |
843 | break; | |
844 | case '\\': | |
845 | p++; | |
846 | if (!*p) | |
847 | return; | |
848 | /* fallthrough */ | |
849 | default: | |
850 | last_space = NULL; | |
851 | } | |
852 | ||
853 | if (last_space) | |
854 | *last_space = '\0'; | |
16402b99 NTND |
855 | } |
856 | ||
0dcb8d7f NTND |
857 | /* |
858 | * Given a subdirectory name and "dir" of the current directory, | |
859 | * search the subdir in "dir" and return it, or create a new one if it | |
860 | * does not exist in "dir". | |
861 | * | |
862 | * If "name" has the trailing slash, it'll be excluded in the search. | |
863 | */ | |
864 | static struct untracked_cache_dir *lookup_untracked(struct untracked_cache *uc, | |
865 | struct untracked_cache_dir *dir, | |
866 | const char *name, int len) | |
867 | { | |
868 | int first, last; | |
869 | struct untracked_cache_dir *d; | |
870 | if (!dir) | |
871 | return NULL; | |
872 | if (len && name[len - 1] == '/') | |
873 | len--; | |
874 | first = 0; | |
875 | last = dir->dirs_nr; | |
876 | while (last > first) { | |
568a05c5 | 877 | int cmp, next = first + ((last - first) >> 1); |
0dcb8d7f NTND |
878 | d = dir->dirs[next]; |
879 | cmp = strncmp(name, d->name, len); | |
880 | if (!cmp && strlen(d->name) > len) | |
881 | cmp = -1; | |
882 | if (!cmp) | |
883 | return d; | |
884 | if (cmp < 0) { | |
885 | last = next; | |
886 | continue; | |
887 | } | |
888 | first = next+1; | |
889 | } | |
890 | ||
891 | uc->dir_created++; | |
96ffc06f | 892 | FLEX_ALLOC_MEM(d, name, name, len); |
0dcb8d7f NTND |
893 | |
894 | ALLOC_GROW(dir->dirs, dir->dirs_nr + 1, dir->dirs_alloc); | |
f919ffeb SG |
895 | MOVE_ARRAY(dir->dirs + first + 1, dir->dirs + first, |
896 | dir->dirs_nr - first); | |
0dcb8d7f NTND |
897 | dir->dirs_nr++; |
898 | dir->dirs[first] = d; | |
899 | return d; | |
900 | } | |
901 | ||
ccad261f NTND |
902 | static void do_invalidate_gitignore(struct untracked_cache_dir *dir) |
903 | { | |
904 | int i; | |
905 | dir->valid = 0; | |
906 | dir->untracked_nr = 0; | |
907 | for (i = 0; i < dir->dirs_nr; i++) | |
908 | do_invalidate_gitignore(dir->dirs[i]); | |
909 | } | |
910 | ||
911 | static void invalidate_gitignore(struct untracked_cache *uc, | |
912 | struct untracked_cache_dir *dir) | |
913 | { | |
914 | uc->gitignore_invalidated++; | |
915 | do_invalidate_gitignore(dir); | |
916 | } | |
917 | ||
91a2288b NTND |
918 | static void invalidate_directory(struct untracked_cache *uc, |
919 | struct untracked_cache_dir *dir) | |
920 | { | |
26cb0182 | 921 | int i; |
b6403131 NTND |
922 | |
923 | /* | |
924 | * Invalidation increment here is just roughly correct. If | |
925 | * untracked_nr or any of dirs[].recurse is non-zero, we | |
926 | * should increment dir_invalidated too. But that's more | |
927 | * expensive to do. | |
928 | */ | |
929 | if (dir->valid) | |
930 | uc->dir_invalidated++; | |
931 | ||
91a2288b NTND |
932 | dir->valid = 0; |
933 | dir->untracked_nr = 0; | |
26cb0182 NTND |
934 | for (i = 0; i < dir->dirs_nr; i++) |
935 | dir->dirs[i]->recurse = 0; | |
91a2288b NTND |
936 | } |
937 | ||
65edd96a | 938 | static int add_patterns_from_buffer(char *buf, size_t size, |
578d81d0 | 939 | const char *base, int baselen, |
caa3d554 | 940 | struct pattern_list *pl); |
578d81d0 | 941 | |
55fe6f51 NTND |
942 | /* |
943 | * Given a file with name "fname", read it (either from disk, or from | |
473e3930 | 944 | * an index if 'istate' is non-null), parse it and store the |
caa3d554 | 945 | * exclude rules in "pl". |
55fe6f51 NTND |
946 | * |
947 | * If "ss" is not NULL, compute SHA-1 of the exclude file and fill | |
65edd96a | 948 | * stat data from disk (only valid if add_patterns returns zero). If |
55fe6f51 NTND |
949 | * ss_valid is non-zero, "ss" must contain good value as input. |
950 | */ | |
65edd96a | 951 | static int add_patterns(const char *fname, const char *base, int baselen, |
caa3d554 | 952 | struct pattern_list *pl, struct index_state *istate, |
4b33e602 | 953 | struct oid_stat *oid_stat) |
453ec4bd | 954 | { |
c470701a | 955 | struct stat st; |
578d81d0 JH |
956 | int r; |
957 | int fd; | |
9d14017a | 958 | size_t size = 0; |
578d81d0 | 959 | char *buf; |
453ec4bd LT |
960 | |
961 | fd = open(fname, O_RDONLY); | |
c28b3d6e | 962 | if (fd < 0 || fstat(fd, &st) < 0) { |
11dc1fcb NTND |
963 | if (fd < 0) |
964 | warn_on_fopen_errors(fname); | |
965 | else | |
c28b3d6e | 966 | close(fd); |
578d81d0 | 967 | if (!istate) |
c28b3d6e | 968 | return -1; |
578d81d0 JH |
969 | r = read_skip_worktree_file_from_index(istate, fname, |
970 | &size, &buf, | |
4b33e602 | 971 | oid_stat); |
578d81d0 JH |
972 | if (r != 1) |
973 | return r; | |
d961baa8 | 974 | } else { |
c28b3d6e NTND |
975 | size = xsize_t(st.st_size); |
976 | if (size == 0) { | |
4b33e602 PO |
977 | if (oid_stat) { |
978 | fill_stat_data(&oid_stat->stat, &st); | |
ba2df751 | 979 | oidcpy(&oid_stat->oid, the_hash_algo->empty_blob); |
4b33e602 | 980 | oid_stat->valid = 1; |
55fe6f51 | 981 | } |
c28b3d6e NTND |
982 | close(fd); |
983 | return 0; | |
984 | } | |
3733e694 | 985 | buf = xmallocz(size); |
c28b3d6e | 986 | if (read_in_full(fd, buf, size) != size) { |
45d76f17 | 987 | free(buf); |
c28b3d6e NTND |
988 | close(fd); |
989 | return -1; | |
990 | } | |
45d76f17 | 991 | buf[size++] = '\n'; |
c28b3d6e | 992 | close(fd); |
4b33e602 | 993 | if (oid_stat) { |
55fe6f51 | 994 | int pos; |
4b33e602 PO |
995 | if (oid_stat->valid && |
996 | !match_stat_data_racy(istate, &oid_stat->stat, &st)) | |
55fe6f51 | 997 | ; /* no content change, ss->sha1 still good */ |
473e3930 BW |
998 | else if (istate && |
999 | (pos = index_name_pos(istate, fname, strlen(fname))) >= 0 && | |
1000 | !ce_stage(istate->cache[pos]) && | |
1001 | ce_uptodate(istate->cache[pos]) && | |
82b474e0 | 1002 | !would_convert_to_git(istate, fname)) |
4b33e602 PO |
1003 | oidcpy(&oid_stat->oid, |
1004 | &istate->cache[pos]->oid); | |
55fe6f51 | 1005 | else |
f070facc PO |
1006 | hash_object_file(buf, size, "blob", |
1007 | &oid_stat->oid); | |
4b33e602 PO |
1008 | fill_stat_data(&oid_stat->stat, &st); |
1009 | oid_stat->valid = 1; | |
55fe6f51 | 1010 | } |
6ba78238 | 1011 | } |
453ec4bd | 1012 | |
65edd96a | 1013 | add_patterns_from_buffer(buf, size, base, baselen, pl); |
578d81d0 JH |
1014 | return 0; |
1015 | } | |
1016 | ||
65edd96a | 1017 | static int add_patterns_from_buffer(char *buf, size_t size, |
578d81d0 | 1018 | const char *base, int baselen, |
caa3d554 | 1019 | struct pattern_list *pl) |
578d81d0 JH |
1020 | { |
1021 | int i, lineno = 1; | |
1022 | char *entry; | |
1023 | ||
96cc8ab5 DS |
1024 | hashmap_init(&pl->recursive_hashmap, pl_hashmap_cmp, NULL, 0); |
1025 | hashmap_init(&pl->parent_hashmap, pl_hashmap_cmp, NULL, 0); | |
1026 | ||
caa3d554 | 1027 | pl->filebuf = buf; |
245e1c19 | 1028 | |
dde843e7 | 1029 | if (skip_utf8_bom(&buf, size)) |
caa3d554 | 1030 | size -= buf - pl->filebuf; |
dde843e7 | 1031 | |
453ec4bd | 1032 | entry = buf; |
245e1c19 | 1033 | |
45d76f17 NTND |
1034 | for (i = 0; i < size; i++) { |
1035 | if (buf[i] == '\n') { | |
453ec4bd LT |
1036 | if (entry != buf + i && entry[0] != '#') { |
1037 | buf[i - (i && buf[i-1] == '\r')] = 0; | |
7e2e4b37 | 1038 | trim_trailing_spaces(entry); |
65edd96a | 1039 | add_pattern(entry, base, baselen, pl, lineno); |
453ec4bd | 1040 | } |
c04318e4 | 1041 | lineno++; |
453ec4bd LT |
1042 | entry = buf + i + 1; |
1043 | } | |
1044 | } | |
1045 | return 0; | |
453ec4bd LT |
1046 | } |
1047 | ||
65edd96a | 1048 | int add_patterns_from_file_to_list(const char *fname, const char *base, |
caa3d554 | 1049 | int baselen, struct pattern_list *pl, |
473e3930 | 1050 | struct index_state *istate) |
55fe6f51 | 1051 | { |
65edd96a | 1052 | return add_patterns(fname, base, baselen, pl, istate, NULL); |
55fe6f51 NTND |
1053 | } |
1054 | ||
65edd96a | 1055 | int add_patterns_from_blob_to_list( |
578d81d0 JH |
1056 | struct object_id *oid, |
1057 | const char *base, int baselen, | |
caa3d554 | 1058 | struct pattern_list *pl) |
578d81d0 JH |
1059 | { |
1060 | char *buf; | |
1061 | size_t size; | |
1062 | int r; | |
1063 | ||
1064 | r = do_read_blob(oid, NULL, &size, &buf); | |
1065 | if (r != 1) | |
1066 | return r; | |
1067 | ||
65edd96a | 1068 | add_patterns_from_buffer(buf, size, base, baselen, pl); |
578d81d0 JH |
1069 | return 0; |
1070 | } | |
1071 | ||
65edd96a | 1072 | struct pattern_list *add_pattern_list(struct dir_struct *dir, |
c04318e4 | 1073 | int group_type, const char *src) |
c082df24 | 1074 | { |
caa3d554 | 1075 | struct pattern_list *pl; |
c082df24 AS |
1076 | struct exclude_list_group *group; |
1077 | ||
1078 | group = &dir->exclude_list_group[group_type]; | |
caa3d554 DS |
1079 | ALLOC_GROW(group->pl, group->nr + 1, group->alloc); |
1080 | pl = &group->pl[group->nr++]; | |
1081 | memset(pl, 0, sizeof(*pl)); | |
1082 | pl->src = src; | |
1083 | return pl; | |
c082df24 AS |
1084 | } |
1085 | ||
1086 | /* | |
1087 | * Used to set up core.excludesfile and .git/info/exclude lists. | |
1088 | */ | |
65edd96a | 1089 | static void add_patterns_from_file_1(struct dir_struct *dir, const char *fname, |
4b33e602 | 1090 | struct oid_stat *oid_stat) |
453ec4bd | 1091 | { |
caa3d554 | 1092 | struct pattern_list *pl; |
ccad261f NTND |
1093 | /* |
1094 | * catch setup_standard_excludes() that's called before | |
1095 | * dir->untracked is assigned. That function behaves | |
1096 | * differently when dir->untracked is non-NULL. | |
1097 | */ | |
1098 | if (!dir->untracked) | |
1099 | dir->unmanaged_exclude_files++; | |
65edd96a DS |
1100 | pl = add_pattern_list(dir, EXC_FILE, fname); |
1101 | if (add_patterns(fname, "", 0, pl, NULL, oid_stat) < 0) | |
a80897c1 | 1102 | die(_("cannot use %s as an exclude file"), fname); |
453ec4bd LT |
1103 | } |
1104 | ||
65edd96a | 1105 | void add_patterns_from_file(struct dir_struct *dir, const char *fname) |
0dcb8d7f | 1106 | { |
ccad261f | 1107 | dir->unmanaged_exclude_files++; /* see validate_untracked_cache() */ |
65edd96a | 1108 | add_patterns_from_file_1(dir, fname, NULL); |
0dcb8d7f NTND |
1109 | } |
1110 | ||
82dce998 NTND |
1111 | int match_basename(const char *basename, int basenamelen, |
1112 | const char *pattern, int prefix, int patternlen, | |
f8708998 | 1113 | unsigned flags) |
593cb880 NTND |
1114 | { |
1115 | if (prefix == patternlen) { | |
0b6e56df | 1116 | if (patternlen == basenamelen && |
ba0897e6 | 1117 | !fspathncmp(pattern, basename, basenamelen)) |
593cb880 | 1118 | return 1; |
4ff89ee5 | 1119 | } else if (flags & PATTERN_FLAG_ENDSWITH) { |
0b6e56df | 1120 | /* "*literal" matching against "fooliteral" */ |
593cb880 | 1121 | if (patternlen - 1 <= basenamelen && |
ba0897e6 | 1122 | !fspathncmp(pattern + 1, |
0b6e56df JH |
1123 | basename + basenamelen - (patternlen - 1), |
1124 | patternlen - 1)) | |
593cb880 NTND |
1125 | return 1; |
1126 | } else { | |
0b6e56df JH |
1127 | if (fnmatch_icase_mem(pattern, patternlen, |
1128 | basename, basenamelen, | |
1129 | 0) == 0) | |
593cb880 NTND |
1130 | return 1; |
1131 | } | |
1132 | return 0; | |
1133 | } | |
1134 | ||
82dce998 NTND |
1135 | int match_pathname(const char *pathname, int pathlen, |
1136 | const char *base, int baselen, | |
1137 | const char *pattern, int prefix, int patternlen, | |
f8708998 | 1138 | unsigned flags) |
b5592632 NTND |
1139 | { |
1140 | const char *name; | |
1141 | int namelen; | |
1142 | ||
1143 | /* | |
1144 | * match with FNM_PATHNAME; the pattern has base implicitly | |
1145 | * in front of it. | |
1146 | */ | |
1147 | if (*pattern == '/') { | |
1148 | pattern++; | |
982ac873 | 1149 | patternlen--; |
b5592632 NTND |
1150 | prefix--; |
1151 | } | |
1152 | ||
1153 | /* | |
1154 | * baselen does not count the trailing slash. base[] may or | |
1155 | * may not end with a trailing slash though. | |
1156 | */ | |
1157 | if (pathlen < baselen + 1 || | |
1158 | (baselen && pathname[baselen] != '/') || | |
ba0897e6 | 1159 | fspathncmp(pathname, base, baselen)) |
b5592632 NTND |
1160 | return 0; |
1161 | ||
1162 | namelen = baselen ? pathlen - baselen - 1 : pathlen; | |
1163 | name = pathname + pathlen - namelen; | |
1164 | ||
1165 | if (prefix) { | |
1166 | /* | |
1167 | * if the non-wildcard part is longer than the | |
1168 | * remaining pathname, surely it cannot match. | |
1169 | */ | |
1170 | if (prefix > namelen) | |
1171 | return 0; | |
1172 | ||
ba0897e6 | 1173 | if (fspathncmp(pattern, name, prefix)) |
b5592632 NTND |
1174 | return 0; |
1175 | pattern += prefix; | |
ab3aebc1 | 1176 | patternlen -= prefix; |
b5592632 NTND |
1177 | name += prefix; |
1178 | namelen -= prefix; | |
ab3aebc1 JK |
1179 | |
1180 | /* | |
1181 | * If the whole pattern did not have a wildcard, | |
1182 | * then our prefix match is all we need; we | |
1183 | * do not need to call fnmatch at all. | |
1184 | */ | |
5cee3493 | 1185 | if (!patternlen && !namelen) |
ab3aebc1 | 1186 | return 1; |
b5592632 NTND |
1187 | } |
1188 | ||
ab3aebc1 JK |
1189 | return fnmatch_icase_mem(pattern, patternlen, |
1190 | name, namelen, | |
f30366b2 | 1191 | WM_PATHNAME) == 0; |
b5592632 NTND |
1192 | } |
1193 | ||
578cd7c3 AS |
1194 | /* |
1195 | * Scan the given exclude list in reverse to see whether pathname | |
1196 | * should be ignored. The first match (i.e. the last on the list), if | |
1197 | * any, determines the fate. Returns the exclude_list element which | |
1198 | * matched, or NULL for undecided. | |
453ec4bd | 1199 | */ |
65edd96a | 1200 | static struct path_pattern *last_matching_pattern_from_list(const char *pathname, |
578cd7c3 AS |
1201 | int pathlen, |
1202 | const char *basename, | |
1203 | int *dtype, | |
caa3d554 | 1204 | struct pattern_list *pl, |
2b70e88d | 1205 | struct index_state *istate) |
453ec4bd | 1206 | { |
ab8db613 | 1207 | struct path_pattern *res = NULL; /* undecided */ |
5cee3493 | 1208 | int i; |
453ec4bd | 1209 | |
caa3d554 | 1210 | if (!pl->nr) |
578cd7c3 | 1211 | return NULL; /* undefined */ |
d6b8fc30 | 1212 | |
caa3d554 DS |
1213 | for (i = pl->nr - 1; 0 <= i; i--) { |
1214 | struct path_pattern *pattern = pl->patterns[i]; | |
ab8db613 DS |
1215 | const char *exclude = pattern->pattern; |
1216 | int prefix = pattern->nowildcardlen; | |
35a94d44 | 1217 | |
4ff89ee5 | 1218 | if (pattern->flags & PATTERN_FLAG_MUSTBEDIR) { |
ad6f2157 | 1219 | *dtype = resolve_dtype(*dtype, 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 | ||
ad6f2157 JK |
1845 | static int resolve_dtype(int dtype, struct index_state *istate, |
1846 | const char *path, int len) | |
07134421 | 1847 | { |
07134421 LT |
1848 | struct stat st; |
1849 | ||
1850 | if (dtype != DT_UNKNOWN) | |
1851 | return dtype; | |
98f2a687 | 1852 | dtype = get_index_dtype(istate, path, len); |
443e061a LT |
1853 | if (dtype != DT_UNKNOWN) |
1854 | return dtype; | |
1855 | if (lstat(path, &st)) | |
07134421 LT |
1856 | return dtype; |
1857 | if (S_ISREG(st.st_mode)) | |
1858 | return DT_REG; | |
1859 | if (S_ISDIR(st.st_mode)) | |
1860 | return DT_DIR; | |
1861 | if (S_ISLNK(st.st_mode)) | |
1862 | return DT_LNK; | |
1863 | return dtype; | |
1864 | } | |
1865 | ||
16e2cfa9 | 1866 | static enum path_treatment treat_one_path(struct dir_struct *dir, |
0dcb8d7f | 1867 | struct untracked_cache_dir *untracked, |
0ef8e169 | 1868 | struct index_state *istate, |
49dc2cc2 | 1869 | struct strbuf *path, |
2e5910f2 | 1870 | int baselen, |
e1b8c7bd | 1871 | const struct pathspec *pathspec, |
ad6f2157 | 1872 | int dtype) |
53cc5356 | 1873 | { |
8aaf8d77 | 1874 | int exclude; |
0ef8e169 | 1875 | int has_path_in_index = !!index_file_exists(istate, path->buf, path->len, ignore_case); |
07966ed1 | 1876 | enum path_treatment path_treatment; |
2eac2a4c | 1877 | |
ad6f2157 | 1878 | dtype = resolve_dtype(dtype, istate, path->buf, path->len); |
8aaf8d77 KB |
1879 | |
1880 | /* Always exclude indexed files */ | |
2eac2a4c | 1881 | if (dtype != DT_DIR && has_path_in_index) |
defd7c7b | 1882 | return path_none; |
8aaf8d77 | 1883 | |
2eac2a4c JH |
1884 | /* |
1885 | * When we are looking at a directory P in the working tree, | |
1886 | * there are three cases: | |
1887 | * | |
1888 | * (1) P exists in the index. Everything inside the directory P in | |
1889 | * the working tree needs to go when P is checked out from the | |
1890 | * index. | |
1891 | * | |
1892 | * (2) P does not exist in the index, but there is P/Q in the index. | |
1893 | * We know P will stay a directory when we check out the contents | |
1894 | * of the index, but we do not know yet if there is a directory | |
1895 | * P/Q in the working tree to be killed, so we need to recurse. | |
1896 | * | |
1897 | * (3) P does not exist in the index, and there is no P/Q in the index | |
1898 | * to require P to be a directory, either. Only in this case, we | |
1899 | * know that everything inside P will not be killed without | |
1900 | * recursing. | |
1901 | */ | |
1902 | if ((dir->flags & DIR_COLLECT_KILLED_ONLY) && | |
1903 | (dtype == DT_DIR) && | |
de372b1b | 1904 | !has_path_in_index && |
0ef8e169 | 1905 | (directory_exists_in_index(istate, path->buf, path->len) == index_nonexistent)) |
de372b1b | 1906 | return path_none; |
8aaf8d77 | 1907 | |
0ef8e169 | 1908 | exclude = is_excluded(dir, istate, path->buf, &dtype); |
53cc5356 JH |
1909 | |
1910 | /* | |
1911 | * Excluded? If we don't explicitly want to show | |
1912 | * ignored files, ignore it | |
1913 | */ | |
0aaf62b6 | 1914 | if (exclude && !(dir->flags & (DIR_SHOW_IGNORED|DIR_SHOW_IGNORED_TOO))) |
defd7c7b | 1915 | return path_excluded; |
53cc5356 | 1916 | |
53cc5356 JH |
1917 | switch (dtype) { |
1918 | default: | |
defd7c7b | 1919 | return path_none; |
53cc5356 | 1920 | case DT_DIR: |
49dc2cc2 | 1921 | strbuf_addch(path, '/'); |
07966ed1 JM |
1922 | path_treatment = treat_directory(dir, istate, untracked, |
1923 | path->buf, path->len, | |
1924 | baselen, exclude, pathspec); | |
1925 | /* | |
1926 | * If 1) we only want to return directories that | |
1927 | * match an exclude pattern and 2) this directory does | |
1928 | * not match an exclude pattern but all of its | |
1929 | * contents are excluded, then indicate that we should | |
1930 | * recurse into this directory (instead of marking the | |
1931 | * directory itself as an ignored path). | |
1932 | */ | |
1933 | if (!exclude && | |
1934 | path_treatment == path_excluded && | |
1935 | (dir->flags & DIR_SHOW_IGNORED_TOO) && | |
1936 | (dir->flags & DIR_SHOW_IGNORED_TOO_MODE_MATCHING)) | |
1937 | return path_recurse; | |
1938 | return path_treatment; | |
53cc5356 JH |
1939 | case DT_REG: |
1940 | case DT_LNK: | |
defd7c7b | 1941 | return exclude ? path_excluded : path_untracked; |
53cc5356 | 1942 | } |
53cc5356 JH |
1943 | } |
1944 | ||
91a2288b NTND |
1945 | static enum path_treatment treat_path_fast(struct dir_struct *dir, |
1946 | struct untracked_cache_dir *untracked, | |
1947 | struct cached_dir *cdir, | |
0ef8e169 | 1948 | struct index_state *istate, |
91a2288b NTND |
1949 | struct strbuf *path, |
1950 | int baselen, | |
e1b8c7bd | 1951 | const struct pathspec *pathspec) |
91a2288b NTND |
1952 | { |
1953 | strbuf_setlen(path, baselen); | |
1954 | if (!cdir->ucd) { | |
1955 | strbuf_addstr(path, cdir->file); | |
1956 | return path_untracked; | |
1957 | } | |
1958 | strbuf_addstr(path, cdir->ucd->name); | |
1959 | /* treat_one_path() does this before it calls treat_directory() */ | |
00b6c178 | 1960 | strbuf_complete(path, '/'); |
91a2288b NTND |
1961 | if (cdir->ucd->check_only) |
1962 | /* | |
1963 | * check_only is set as a result of treat_directory() getting | |
1964 | * to its bottom. Verify again the same set of directories | |
1965 | * with check_only set. | |
1966 | */ | |
0ef8e169 | 1967 | return read_directory_recursive(dir, istate, path->buf, path->len, |
5aaa7fd3 | 1968 | cdir->ucd, 1, 0, pathspec); |
91a2288b NTND |
1969 | /* |
1970 | * We get path_recurse in the first run when | |
1971 | * directory_exists_in_index() returns index_nonexistent. We | |
1972 | * are sure that new changes in the index does not impact the | |
1973 | * outcome. Return now. | |
1974 | */ | |
1975 | return path_recurse; | |
1976 | } | |
1977 | ||
16e2cfa9 | 1978 | static enum path_treatment treat_path(struct dir_struct *dir, |
0dcb8d7f | 1979 | struct untracked_cache_dir *untracked, |
cf7c6148 | 1980 | struct cached_dir *cdir, |
0ef8e169 | 1981 | struct index_state *istate, |
49dc2cc2 | 1982 | struct strbuf *path, |
16e2cfa9 | 1983 | int baselen, |
e1b8c7bd | 1984 | const struct pathspec *pathspec) |
16e2cfa9 | 1985 | { |
ad6f2157 | 1986 | if (!cdir->d_name) |
0ef8e169 | 1987 | return treat_path_fast(dir, untracked, cdir, istate, path, |
e1b8c7bd | 1988 | baselen, pathspec); |
ad6f2157 | 1989 | if (is_dot_or_dotdot(cdir->d_name) || !fspathcmp(cdir->d_name, ".git")) |
defd7c7b | 1990 | return path_none; |
49dc2cc2 | 1991 | strbuf_setlen(path, baselen); |
ad6f2157 | 1992 | strbuf_addstr(path, cdir->d_name); |
e1b8c7bd | 1993 | if (simplify_away(path->buf, path->len, pathspec)) |
defd7c7b | 1994 | return path_none; |
16e2cfa9 | 1995 | |
ad6f2157 JK |
1996 | return treat_one_path(dir, untracked, istate, path, baselen, pathspec, |
1997 | cdir->d_type); | |
0dcb8d7f NTND |
1998 | } |
1999 | ||
2000 | static void add_untracked(struct untracked_cache_dir *dir, const char *name) | |
2001 | { | |
2002 | if (!dir) | |
2003 | return; | |
2004 | ALLOC_GROW(dir->untracked, dir->untracked_nr + 1, | |
2005 | dir->untracked_alloc); | |
2006 | dir->untracked[dir->untracked_nr++] = xstrdup(name); | |
16e2cfa9 JH |
2007 | } |
2008 | ||
91a2288b NTND |
2009 | static int valid_cached_dir(struct dir_struct *dir, |
2010 | struct untracked_cache_dir *untracked, | |
207a06ce | 2011 | struct index_state *istate, |
91a2288b NTND |
2012 | struct strbuf *path, |
2013 | int check_only) | |
2014 | { | |
2015 | struct stat st; | |
2016 | ||
2017 | if (!untracked) | |
2018 | return 0; | |
2019 | ||
883e248b BP |
2020 | /* |
2021 | * With fsmonitor, we can trust the untracked cache's valid field. | |
2022 | */ | |
2023 | refresh_fsmonitor(istate); | |
2024 | if (!(dir->untracked->use_fsmonitor && untracked->valid)) { | |
2523c4be | 2025 | if (lstat(path->len ? path->buf : ".", &st)) { |
883e248b BP |
2026 | memset(&untracked->stat_data, 0, sizeof(untracked->stat_data)); |
2027 | return 0; | |
2028 | } | |
2029 | if (!untracked->valid || | |
2030 | match_stat_data_racy(istate, &untracked->stat_data, &st)) { | |
883e248b BP |
2031 | fill_stat_data(&untracked->stat_data, &st); |
2032 | return 0; | |
2033 | } | |
91a2288b NTND |
2034 | } |
2035 | ||
b6403131 | 2036 | if (untracked->check_only != !!check_only) |
91a2288b | 2037 | return 0; |
91a2288b NTND |
2038 | |
2039 | /* | |
2040 | * prep_exclude will be called eventually on this directory, | |
65edd96a | 2041 | * but it's called much later in last_matching_pattern(). We |
91a2288b NTND |
2042 | * need it now to determine the validity of the cache for this |
2043 | * path. The next calls will be nearly no-op, the way | |
2044 | * prep_exclude() is designed. | |
2045 | */ | |
2046 | if (path->len && path->buf[path->len - 1] != '/') { | |
2047 | strbuf_addch(path, '/'); | |
207a06ce | 2048 | prep_exclude(dir, istate, path->buf, path->len); |
91a2288b NTND |
2049 | strbuf_setlen(path, path->len - 1); |
2050 | } else | |
207a06ce | 2051 | prep_exclude(dir, istate, path->buf, path->len); |
91a2288b NTND |
2052 | |
2053 | /* hopefully prep_exclude() haven't invalidated this entry... */ | |
2054 | return untracked->valid; | |
2055 | } | |
2056 | ||
cf7c6148 NTND |
2057 | static int open_cached_dir(struct cached_dir *cdir, |
2058 | struct dir_struct *dir, | |
2059 | struct untracked_cache_dir *untracked, | |
207a06ce | 2060 | struct index_state *istate, |
cf7c6148 NTND |
2061 | struct strbuf *path, |
2062 | int check_only) | |
2063 | { | |
b6731550 NTND |
2064 | const char *c_path; |
2065 | ||
cf7c6148 NTND |
2066 | memset(cdir, 0, sizeof(*cdir)); |
2067 | cdir->untracked = untracked; | |
207a06ce | 2068 | if (valid_cached_dir(dir, untracked, istate, path, check_only)) |
91a2288b | 2069 | return 0; |
b6731550 NTND |
2070 | c_path = path->len ? path->buf : "."; |
2071 | cdir->fdir = opendir(c_path); | |
2072 | if (!cdir->fdir) | |
2073 | warning_errno(_("could not open directory '%s'"), c_path); | |
b6403131 NTND |
2074 | if (dir->untracked) { |
2075 | invalidate_directory(dir->untracked, untracked); | |
91a2288b | 2076 | dir->untracked->dir_opened++; |
b6403131 | 2077 | } |
cf7c6148 NTND |
2078 | if (!cdir->fdir) |
2079 | return -1; | |
2080 | return 0; | |
2081 | } | |
2082 | ||
2083 | static int read_cached_dir(struct cached_dir *cdir) | |
2084 | { | |
ad6f2157 JK |
2085 | struct dirent *de; |
2086 | ||
cf7c6148 | 2087 | if (cdir->fdir) { |
ad6f2157 JK |
2088 | de = readdir(cdir->fdir); |
2089 | if (!de) { | |
2090 | cdir->d_name = NULL; | |
2091 | cdir->d_type = DT_UNKNOWN; | |
cf7c6148 | 2092 | return -1; |
ad6f2157 JK |
2093 | } |
2094 | cdir->d_name = de->d_name; | |
2095 | cdir->d_type = DTYPE(de); | |
cf7c6148 NTND |
2096 | return 0; |
2097 | } | |
91a2288b NTND |
2098 | while (cdir->nr_dirs < cdir->untracked->dirs_nr) { |
2099 | struct untracked_cache_dir *d = cdir->untracked->dirs[cdir->nr_dirs]; | |
26cb0182 NTND |
2100 | if (!d->recurse) { |
2101 | cdir->nr_dirs++; | |
2102 | continue; | |
2103 | } | |
91a2288b NTND |
2104 | cdir->ucd = d; |
2105 | cdir->nr_dirs++; | |
2106 | return 0; | |
2107 | } | |
2108 | cdir->ucd = NULL; | |
2109 | if (cdir->nr_files < cdir->untracked->untracked_nr) { | |
2110 | struct untracked_cache_dir *d = cdir->untracked; | |
2111 | cdir->file = d->untracked[cdir->nr_files++]; | |
2112 | return 0; | |
2113 | } | |
cf7c6148 NTND |
2114 | return -1; |
2115 | } | |
2116 | ||
2117 | static void close_cached_dir(struct cached_dir *cdir) | |
2118 | { | |
2119 | if (cdir->fdir) | |
2120 | closedir(cdir->fdir); | |
91a2288b NTND |
2121 | /* |
2122 | * We have gone through this directory and found no untracked | |
2123 | * entries. Mark it valid. | |
2124 | */ | |
26cb0182 | 2125 | if (cdir->untracked) { |
91a2288b | 2126 | cdir->untracked->valid = 1; |
26cb0182 NTND |
2127 | cdir->untracked->recurse = 1; |
2128 | } | |
16e2cfa9 JH |
2129 | } |
2130 | ||
c5c4eddd EN |
2131 | static void add_path_to_appropriate_result_list(struct dir_struct *dir, |
2132 | struct untracked_cache_dir *untracked, | |
2133 | struct cached_dir *cdir, | |
2134 | struct index_state *istate, | |
2135 | struct strbuf *path, | |
2136 | int baselen, | |
2137 | const struct pathspec *pathspec, | |
2138 | enum path_treatment state) | |
2139 | { | |
2140 | /* add the path to the appropriate result list */ | |
2141 | switch (state) { | |
2142 | case path_excluded: | |
2143 | if (dir->flags & DIR_SHOW_IGNORED) | |
2144 | dir_add_name(dir, istate, path->buf, path->len); | |
2145 | else if ((dir->flags & DIR_SHOW_IGNORED_TOO) || | |
2146 | ((dir->flags & DIR_COLLECT_IGNORED) && | |
2147 | exclude_matches_pathspec(path->buf, path->len, | |
2148 | pathspec))) | |
2149 | dir_add_ignored(dir, istate, path->buf, path->len); | |
2150 | break; | |
2151 | ||
2152 | case path_untracked: | |
2153 | if (dir->flags & DIR_SHOW_IGNORED) | |
2154 | break; | |
2155 | dir_add_name(dir, istate, path->buf, path->len); | |
2156 | if (cdir->fdir) | |
2157 | add_untracked(untracked, path->buf + baselen); | |
2158 | break; | |
2159 | ||
2160 | default: | |
2161 | break; | |
2162 | } | |
2163 | } | |
2164 | ||
453ec4bd LT |
2165 | /* |
2166 | * Read a directory tree. We currently ignore anything but | |
2167 | * directories, regular files and symlinks. That's because git | |
2168 | * doesn't handle them at all yet. Maybe that will change some | |
2169 | * day. | |
2170 | * | |
2171 | * Also, we ignore the name ".git" (even if it is not a directory). | |
2172 | * That likely will not change. | |
defd7c7b | 2173 | * |
5aaa7fd3 JM |
2174 | * If 'stop_at_first_file' is specified, 'path_excluded' is returned |
2175 | * to signal that a file was found. This is the least significant value that | |
2176 | * indicates that a file was encountered that does not depend on the order of | |
2177 | * whether an untracked or exluded path was encountered first. | |
2178 | * | |
defd7c7b | 2179 | * Returns the most significant path_treatment value encountered in the scan. |
5aaa7fd3 JM |
2180 | * If 'stop_at_first_file' is specified, `path_excluded` is the most |
2181 | * significant path_treatment value that will be returned. | |
453ec4bd | 2182 | */ |
5aaa7fd3 | 2183 | |
defd7c7b | 2184 | static enum path_treatment read_directory_recursive(struct dir_struct *dir, |
0ef8e169 BW |
2185 | struct index_state *istate, const char *base, int baselen, |
2186 | struct untracked_cache_dir *untracked, int check_only, | |
5aaa7fd3 | 2187 | int stop_at_first_file, const struct pathspec *pathspec) |
453ec4bd | 2188 | { |
777b4203 EN |
2189 | /* |
2190 | * WARNING WARNING WARNING: | |
2191 | * | |
2192 | * Any updates to the traversal logic here may need corresponding | |
2193 | * updates in treat_leading_path(). See the commit message for the | |
2194 | * commit adding this warning as well as the commit preceding it | |
2195 | * for details. | |
2196 | */ | |
2197 | ||
cf7c6148 | 2198 | struct cached_dir cdir; |
defd7c7b | 2199 | enum path_treatment state, subdir_state, dir_state = path_none; |
bef36921 | 2200 | struct strbuf path = STRBUF_INIT; |
453ec4bd | 2201 | |
bef36921 | 2202 | strbuf_add(&path, base, baselen); |
02cb6753 | 2203 | |
0ef8e169 | 2204 | if (open_cached_dir(&cdir, dir, untracked, istate, &path, check_only)) |
1528d247 RS |
2205 | goto out; |
2206 | ||
0dcb8d7f NTND |
2207 | if (untracked) |
2208 | untracked->check_only = !!check_only; | |
2209 | ||
cf7c6148 | 2210 | while (!read_cached_dir(&cdir)) { |
defd7c7b | 2211 | /* check how the file or directory should be treated */ |
0ef8e169 | 2212 | state = treat_path(dir, untracked, &cdir, istate, &path, |
e1b8c7bd | 2213 | baselen, pathspec); |
0dcb8d7f | 2214 | |
defd7c7b KB |
2215 | if (state > dir_state) |
2216 | dir_state = state; | |
2217 | ||
2218 | /* recurse into subdir if instructed by treat_path */ | |
df5bcdf8 SL |
2219 | if ((state == path_recurse) || |
2220 | ((state == path_untracked) && | |
ad6f2157 | 2221 | (resolve_dtype(cdir.d_type, istate, path.buf, path.len) == DT_DIR) && |
89a1f4aa | 2222 | ((dir->flags & DIR_SHOW_IGNORED_TOO) || |
69f272b9 EN |
2223 | (pathspec && |
2224 | do_match_pathspec(istate, pathspec, path.buf, path.len, | |
2225 | baselen, NULL, DO_MATCH_LEADING_PATHSPEC) == MATCHED_RECURSIVELY_LEADING_PATHSPEC)))) { | |
0dcb8d7f NTND |
2226 | struct untracked_cache_dir *ud; |
2227 | ud = lookup_untracked(dir->untracked, untracked, | |
2228 | path.buf + baselen, | |
2229 | path.len - baselen); | |
2230 | subdir_state = | |
0ef8e169 | 2231 | read_directory_recursive(dir, istate, path.buf, |
e1b8c7bd | 2232 | path.len, ud, |
5aaa7fd3 | 2233 | check_only, stop_at_first_file, pathspec); |
defd7c7b KB |
2234 | if (subdir_state > dir_state) |
2235 | dir_state = subdir_state; | |
404ebced | 2236 | |
69f272b9 EN |
2237 | if (pathspec && |
2238 | !match_pathspec(istate, pathspec, path.buf, path.len, | |
404ebced EN |
2239 | 0 /* prefix */, NULL, |
2240 | 0 /* do NOT special case dirs */)) | |
2241 | state = path_none; | |
defd7c7b KB |
2242 | } |
2243 | ||
2244 | if (check_only) { | |
5aaa7fd3 JM |
2245 | if (stop_at_first_file) { |
2246 | /* | |
2247 | * If stopping at first file, then | |
2248 | * signal that a file was found by | |
2249 | * returning `path_excluded`. This is | |
2250 | * to return a consistent value | |
2251 | * regardless of whether an ignored or | |
2252 | * excluded file happened to be | |
2253 | * encountered 1st. | |
2254 | * | |
2255 | * In current usage, the | |
2256 | * `stop_at_first_file` is passed when | |
2257 | * an ancestor directory has matched | |
2258 | * an exclude pattern, so any found | |
2259 | * files will be excluded. | |
2260 | */ | |
2261 | if (dir_state >= path_excluded) { | |
2262 | dir_state = path_excluded; | |
2263 | break; | |
2264 | } | |
2265 | } | |
2266 | ||
defd7c7b | 2267 | /* abort early if maximum state has been reached */ |
0dcb8d7f | 2268 | if (dir_state == path_untracked) { |
91a2288b | 2269 | if (cdir.fdir) |
0dcb8d7f | 2270 | add_untracked(untracked, path.buf + baselen); |
defd7c7b | 2271 | break; |
0dcb8d7f | 2272 | } |
defd7c7b | 2273 | /* skip the dir_add_* part */ |
02cb6753 | 2274 | continue; |
453ec4bd | 2275 | } |
defd7c7b | 2276 | |
c5c4eddd EN |
2277 | add_path_to_appropriate_result_list(dir, untracked, &cdir, |
2278 | istate, &path, baselen, | |
2279 | pathspec, state); | |
453ec4bd | 2280 | } |
cf7c6148 | 2281 | close_cached_dir(&cdir); |
1528d247 | 2282 | out: |
bef36921 | 2283 | strbuf_release(&path); |
453ec4bd | 2284 | |
defd7c7b | 2285 | return dir_state; |
453ec4bd LT |
2286 | } |
2287 | ||
bbf504a9 | 2288 | int cmp_dir_entry(const void *p1, const void *p2) |
453ec4bd LT |
2289 | { |
2290 | const struct dir_entry *e1 = *(const struct dir_entry **)p1; | |
2291 | const struct dir_entry *e2 = *(const struct dir_entry **)p2; | |
2292 | ||
ccdd4a0f | 2293 | return name_compare(e1->name, e1->len, e2->name, e2->len); |
453ec4bd LT |
2294 | } |
2295 | ||
fb898888 | 2296 | /* check if *out lexically strictly contains *in */ |
bbf504a9 | 2297 | int check_dir_entry_contains(const struct dir_entry *out, const struct dir_entry *in) |
fb898888 SL |
2298 | { |
2299 | return (out->len < in->len) && | |
2300 | (out->name[out->len - 1] == '/') && | |
2301 | !memcmp(out->name, in->name, out->len); | |
2302 | } | |
2303 | ||
48ffef96 | 2304 | static int treat_leading_path(struct dir_struct *dir, |
0ef8e169 | 2305 | struct index_state *istate, |
48ffef96 | 2306 | const char *path, int len, |
e1b8c7bd | 2307 | const struct pathspec *pathspec) |
48ffef96 | 2308 | { |
777b4203 EN |
2309 | /* |
2310 | * WARNING WARNING WARNING: | |
2311 | * | |
2312 | * Any updates to the traversal logic here may need corresponding | |
0cbb6057 JK |
2313 | * updates in read_directory_recursive(). See 777b420347 (dir: |
2314 | * synchronize treat_leading_path() and read_directory_recursive(), | |
2315 | * 2019-12-19) and its parent commit for details. | |
777b4203 EN |
2316 | */ |
2317 | ||
49dc2cc2 | 2318 | struct strbuf sb = STRBUF_INIT; |
ad6f2157 | 2319 | struct strbuf subdir = STRBUF_INIT; |
b9670c1f | 2320 | int prevlen, baselen; |
48ffef96 | 2321 | const char *cp; |
b9670c1f | 2322 | struct cached_dir cdir; |
b9670c1f EN |
2323 | enum path_treatment state = path_none; |
2324 | ||
2325 | /* | |
2326 | * For each directory component of path, we are going to check whether | |
2327 | * that path is relevant given the pathspec. For example, if path is | |
2328 | * foo/bar/baz/ | |
2329 | * then we will ask treat_path() whether we should go into foo, then | |
2330 | * whether we should go into bar, then whether baz is relevant. | |
2331 | * Checking each is important because e.g. if path is | |
2332 | * .git/info/ | |
2333 | * then we need to check .git to know we shouldn't traverse it. | |
2334 | * If the return from treat_path() is: | |
2335 | * * path_none, for any path, we return false. | |
2336 | * * path_recurse, for all path components, we return true | |
2337 | * * <anything else> for some intermediate component, we make sure | |
2338 | * to add that path to the relevant list but return false | |
2339 | * signifying that we shouldn't recurse into it. | |
2340 | */ | |
48ffef96 JH |
2341 | |
2342 | while (len && path[len - 1] == '/') | |
2343 | len--; | |
2344 | if (!len) | |
2345 | return 1; | |
b9670c1f | 2346 | |
b9670c1f | 2347 | memset(&cdir, 0, sizeof(cdir)); |
ad6f2157 | 2348 | cdir.d_type = DT_DIR; |
48ffef96 | 2349 | baselen = 0; |
b9670c1f | 2350 | prevlen = 0; |
48ffef96 | 2351 | while (1) { |
b9670c1f EN |
2352 | prevlen = baselen + !!baselen; |
2353 | cp = path + prevlen; | |
48ffef96 JH |
2354 | cp = memchr(cp, '/', path + len - cp); |
2355 | if (!cp) | |
2356 | baselen = len; | |
2357 | else | |
2358 | baselen = cp - path; | |
b9670c1f | 2359 | strbuf_reset(&sb); |
49dc2cc2 RS |
2360 | strbuf_add(&sb, path, baselen); |
2361 | if (!is_directory(sb.buf)) | |
2362 | break; | |
b9670c1f EN |
2363 | strbuf_reset(&sb); |
2364 | strbuf_add(&sb, path, prevlen); | |
ad6f2157 JK |
2365 | strbuf_reset(&subdir); |
2366 | strbuf_add(&subdir, path+prevlen, baselen-prevlen); | |
2367 | cdir.d_name = subdir.buf; | |
b9670c1f EN |
2368 | state = treat_path(dir, NULL, &cdir, istate, &sb, prevlen, |
2369 | pathspec); | |
777b4203 | 2370 | if (state == path_untracked && |
ad6f2157 | 2371 | resolve_dtype(cdir.d_type, istate, sb.buf, sb.len) == DT_DIR && |
777b4203 EN |
2372 | (dir->flags & DIR_SHOW_IGNORED_TOO || |
2373 | do_match_pathspec(istate, pathspec, sb.buf, sb.len, | |
2374 | baselen, NULL, DO_MATCH_LEADING_PATHSPEC) == MATCHED_RECURSIVELY_LEADING_PATHSPEC)) { | |
22705334 EN |
2375 | if (!match_pathspec(istate, pathspec, sb.buf, sb.len, |
2376 | 0 /* prefix */, NULL, | |
2377 | 0 /* do NOT special case dirs */)) | |
2378 | state = path_none; | |
777b4203 EN |
2379 | add_path_to_appropriate_result_list(dir, NULL, &cdir, |
2380 | istate, | |
2381 | &sb, baselen, | |
2382 | pathspec, state); | |
2383 | state = path_recurse; | |
2384 | } | |
2385 | ||
b9670c1f | 2386 | if (state != path_recurse) |
49dc2cc2 | 2387 | break; /* do not recurse into it */ |
b9670c1f | 2388 | if (len <= baselen) |
49dc2cc2 | 2389 | break; /* finished checking */ |
48ffef96 | 2390 | } |
b9670c1f EN |
2391 | add_path_to_appropriate_result_list(dir, NULL, &cdir, istate, |
2392 | &sb, baselen, pathspec, | |
2393 | state); | |
2394 | ||
ad6f2157 | 2395 | strbuf_release(&subdir); |
49dc2cc2 | 2396 | strbuf_release(&sb); |
b9670c1f | 2397 | return state == path_recurse; |
48ffef96 JH |
2398 | } |
2399 | ||
1e8fef60 NTND |
2400 | static const char *get_ident_string(void) |
2401 | { | |
2402 | static struct strbuf sb = STRBUF_INIT; | |
2403 | struct utsname uts; | |
2404 | ||
2405 | if (sb.len) | |
2406 | return sb.buf; | |
100e4337 | 2407 | if (uname(&uts) < 0) |
1e8fef60 | 2408 | die_errno(_("failed to get kernel name and information")); |
0e0f7618 CC |
2409 | strbuf_addf(&sb, "Location %s, system %s", get_git_work_tree(), |
2410 | uts.sysname); | |
1e8fef60 NTND |
2411 | return sb.buf; |
2412 | } | |
2413 | ||
2414 | static int ident_in_untracked(const struct untracked_cache *uc) | |
2415 | { | |
0e0f7618 CC |
2416 | /* |
2417 | * Previous git versions may have saved many NUL separated | |
2418 | * strings in the "ident" field, but it is insane to manage | |
2419 | * many locations, so just take care of the first one. | |
2420 | */ | |
1e8fef60 | 2421 | |
0e0f7618 | 2422 | return !strcmp(uc->ident.buf, get_ident_string()); |
1e8fef60 NTND |
2423 | } |
2424 | ||
0e0f7618 | 2425 | static void set_untracked_ident(struct untracked_cache *uc) |
1e8fef60 | 2426 | { |
0e0f7618 | 2427 | strbuf_reset(&uc->ident); |
1e8fef60 | 2428 | strbuf_addstr(&uc->ident, get_ident_string()); |
0e0f7618 CC |
2429 | |
2430 | /* | |
2431 | * This strbuf used to contain a list of NUL separated | |
2432 | * strings, so save NUL too for backward compatibility. | |
2433 | */ | |
1e8fef60 NTND |
2434 | strbuf_addch(&uc->ident, 0); |
2435 | } | |
2436 | ||
4a4ca479 CC |
2437 | static void new_untracked_cache(struct index_state *istate) |
2438 | { | |
2439 | struct untracked_cache *uc = xcalloc(1, sizeof(*uc)); | |
2440 | strbuf_init(&uc->ident, 100); | |
2441 | uc->exclude_per_dir = ".gitignore"; | |
2442 | /* should be the same flags used by git-status */ | |
2443 | uc->dir_flags = DIR_SHOW_OTHER_DIRECTORIES | DIR_HIDE_EMPTY_DIRECTORIES; | |
0e0f7618 | 2444 | set_untracked_ident(uc); |
4a4ca479 | 2445 | istate->untracked = uc; |
0e0f7618 | 2446 | istate->cache_changed |= UNTRACKED_CHANGED; |
4a4ca479 CC |
2447 | } |
2448 | ||
2449 | void add_untracked_cache(struct index_state *istate) | |
2450 | { | |
2451 | if (!istate->untracked) { | |
2452 | new_untracked_cache(istate); | |
0e0f7618 CC |
2453 | } else { |
2454 | if (!ident_in_untracked(istate->untracked)) { | |
2455 | free_untracked_cache(istate->untracked); | |
2456 | new_untracked_cache(istate); | |
2457 | } | |
2458 | } | |
4a4ca479 CC |
2459 | } |
2460 | ||
07b29bfd CC |
2461 | void remove_untracked_cache(struct index_state *istate) |
2462 | { | |
2463 | if (istate->untracked) { | |
2464 | free_untracked_cache(istate->untracked); | |
2465 | istate->untracked = NULL; | |
2466 | istate->cache_changed |= UNTRACKED_CHANGED; | |
2467 | } | |
2468 | } | |
2469 | ||
ccad261f NTND |
2470 | static struct untracked_cache_dir *validate_untracked_cache(struct dir_struct *dir, |
2471 | int base_len, | |
2472 | const struct pathspec *pathspec) | |
2473 | { | |
2474 | struct untracked_cache_dir *root; | |
026336cb | 2475 | static int untracked_cache_disabled = -1; |
ccad261f | 2476 | |
026336cb JH |
2477 | if (!dir->untracked) |
2478 | return NULL; | |
2479 | if (untracked_cache_disabled < 0) | |
2480 | untracked_cache_disabled = git_env_bool("GIT_DISABLE_UNTRACKED_CACHE", 0); | |
2481 | if (untracked_cache_disabled) | |
ccad261f NTND |
2482 | return NULL; |
2483 | ||
2484 | /* | |
2485 | * We only support $GIT_DIR/info/exclude and core.excludesfile | |
2486 | * as the global ignore rule files. Any other additions | |
2487 | * (e.g. from command line) invalidate the cache. This | |
2488 | * condition also catches running setup_standard_excludes() | |
2489 | * before setting dir->untracked! | |
2490 | */ | |
2491 | if (dir->unmanaged_exclude_files) | |
2492 | return NULL; | |
2493 | ||
2494 | /* | |
2495 | * Optimize for the main use case only: whole-tree git | |
2496 | * status. More work involved in treat_leading_path() if we | |
2497 | * use cache on just a subset of the worktree. pathspec | |
2498 | * support could make the matter even worse. | |
2499 | */ | |
2500 | if (base_len || (pathspec && pathspec->nr)) | |
2501 | return NULL; | |
2502 | ||
2503 | /* Different set of flags may produce different results */ | |
2504 | if (dir->flags != dir->untracked->dir_flags || | |
2505 | /* | |
2506 | * See treat_directory(), case index_nonexistent. Without | |
2507 | * this flag, we may need to also cache .git file content | |
2508 | * for the resolve_gitlink_ref() call, which we don't. | |
2509 | */ | |
2510 | !(dir->flags & DIR_SHOW_OTHER_DIRECTORIES) || | |
2511 | /* We don't support collecting ignore files */ | |
2512 | (dir->flags & (DIR_SHOW_IGNORED | DIR_SHOW_IGNORED_TOO | | |
2513 | DIR_COLLECT_IGNORED))) | |
2514 | return NULL; | |
2515 | ||
2516 | /* | |
2517 | * If we use .gitignore in the cache and now you change it to | |
2518 | * .gitexclude, everything will go wrong. | |
2519 | */ | |
2520 | if (dir->exclude_per_dir != dir->untracked->exclude_per_dir && | |
2521 | strcmp(dir->exclude_per_dir, dir->untracked->exclude_per_dir)) | |
2522 | return NULL; | |
2523 | ||
2524 | /* | |
2525 | * EXC_CMDL is not considered in the cache. If people set it, | |
2526 | * skip the cache. | |
2527 | */ | |
2528 | if (dir->exclude_list_group[EXC_CMDL].nr) | |
2529 | return NULL; | |
2530 | ||
1e8fef60 | 2531 | if (!ident_in_untracked(dir->untracked)) { |
1a07e59c | 2532 | warning(_("untracked cache is disabled on this system or location")); |
1e8fef60 NTND |
2533 | return NULL; |
2534 | } | |
2535 | ||
ccad261f NTND |
2536 | if (!dir->untracked->root) { |
2537 | const int len = sizeof(*dir->untracked->root); | |
2538 | dir->untracked->root = xmalloc(len); | |
2539 | memset(dir->untracked->root, 0, len); | |
2540 | } | |
2541 | ||
2542 | /* Validate $GIT_DIR/info/exclude and core.excludesfile */ | |
2543 | root = dir->untracked->root; | |
9001dc2a | 2544 | if (!oideq(&dir->ss_info_exclude.oid, |
4b33e602 | 2545 | &dir->untracked->ss_info_exclude.oid)) { |
ccad261f NTND |
2546 | invalidate_gitignore(dir->untracked, root); |
2547 | dir->untracked->ss_info_exclude = dir->ss_info_exclude; | |
2548 | } | |
9001dc2a | 2549 | if (!oideq(&dir->ss_excludes_file.oid, |
4b33e602 | 2550 | &dir->untracked->ss_excludes_file.oid)) { |
ccad261f NTND |
2551 | invalidate_gitignore(dir->untracked, root); |
2552 | dir->untracked->ss_excludes_file = dir->ss_excludes_file; | |
2553 | } | |
26cb0182 NTND |
2554 | |
2555 | /* Make sure this directory is not dropped out at saving phase */ | |
2556 | root->recurse = 1; | |
ccad261f NTND |
2557 | return root; |
2558 | } | |
2559 | ||
2c1eb104 BW |
2560 | int read_directory(struct dir_struct *dir, struct index_state *istate, |
2561 | const char *path, int len, const struct pathspec *pathspec) | |
9fc42d60 | 2562 | { |
ccad261f | 2563 | struct untracked_cache_dir *untracked; |
b4189aa8 | 2564 | |
c46c406a NTND |
2565 | trace_performance_enter(); |
2566 | ||
2567 | if (has_symlink_leading_path(path, len)) { | |
2568 | trace_performance_leave("read directory %.*s", len, path); | |
725b0605 | 2569 | return dir->nr; |
c46c406a | 2570 | } |
725b0605 | 2571 | |
ccad261f NTND |
2572 | untracked = validate_untracked_cache(dir, len, pathspec); |
2573 | if (!untracked) | |
2574 | /* | |
2575 | * make sure untracked cache code path is disabled, | |
2576 | * e.g. prep_exclude() | |
2577 | */ | |
2578 | dir->untracked = NULL; | |
2c1eb104 | 2579 | if (!len || treat_leading_path(dir, istate, path, len, pathspec)) |
5aaa7fd3 | 2580 | read_directory_recursive(dir, istate, path, len, untracked, 0, 0, pathspec); |
bbf504a9 SL |
2581 | QSORT(dir->entries, dir->nr, cmp_dir_entry); |
2582 | QSORT(dir->ignored, dir->ignored_nr, cmp_dir_entry); | |
fb898888 SL |
2583 | |
2584 | /* | |
2585 | * If DIR_SHOW_IGNORED_TOO is set, read_directory_recursive() will | |
2586 | * also pick up untracked contents of untracked dirs; by default | |
2587 | * we discard these, but given DIR_KEEP_UNTRACKED_CONTENTS we do not. | |
2588 | */ | |
2589 | if ((dir->flags & DIR_SHOW_IGNORED_TOO) && | |
2590 | !(dir->flags & DIR_KEEP_UNTRACKED_CONTENTS)) { | |
2591 | int i, j; | |
2592 | ||
2593 | /* remove from dir->entries untracked contents of untracked dirs */ | |
2594 | for (i = j = 0; j < dir->nr; j++) { | |
bbf504a9 SL |
2595 | if (i && |
2596 | check_dir_entry_contains(dir->entries[i - 1], dir->entries[j])) { | |
6a83d902 | 2597 | FREE_AND_NULL(dir->entries[j]); |
fb898888 SL |
2598 | } else { |
2599 | dir->entries[i++] = dir->entries[j]; | |
2600 | } | |
2601 | } | |
2602 | ||
2603 | dir->nr = i; | |
2604 | } | |
2605 | ||
c46c406a | 2606 | trace_performance_leave("read directory %.*s", len, path); |
c9ccb5d3 | 2607 | if (dir->untracked) { |
026336cb | 2608 | static int force_untracked_cache = -1; |
c9ccb5d3 | 2609 | static struct trace_key trace_untracked_stats = TRACE_KEY_INIT(UNTRACKED_STATS); |
026336cb JH |
2610 | |
2611 | if (force_untracked_cache < 0) | |
2612 | force_untracked_cache = | |
2613 | git_env_bool("GIT_FORCE_UNTRACKED_CACHE", 0); | |
c9ccb5d3 NTND |
2614 | trace_printf_key(&trace_untracked_stats, |
2615 | "node creation: %u\n" | |
2616 | "gitignore invalidation: %u\n" | |
2617 | "directory invalidation: %u\n" | |
2618 | "opendir: %u\n", | |
2619 | dir->untracked->dir_created, | |
2620 | dir->untracked->gitignore_invalidated, | |
2621 | dir->untracked->dir_invalidated, | |
2622 | dir->untracked->dir_opened); | |
026336cb | 2623 | if (force_untracked_cache && |
fc9ecbeb | 2624 | dir->untracked == istate->untracked && |
1bbb3dba NTND |
2625 | (dir->untracked->dir_opened || |
2626 | dir->untracked->gitignore_invalidated || | |
2627 | dir->untracked->dir_invalidated)) | |
2c1eb104 BW |
2628 | istate->cache_changed |= UNTRACKED_CHANGED; |
2629 | if (dir->untracked != istate->untracked) { | |
6a83d902 | 2630 | FREE_AND_NULL(dir->untracked); |
1bbb3dba | 2631 | } |
c9ccb5d3 | 2632 | } |
453ec4bd LT |
2633 | return dir->nr; |
2634 | } | |
c91f0d92 | 2635 | |
686a4a06 | 2636 | int file_exists(const char *f) |
c91f0d92 | 2637 | { |
686a4a06 | 2638 | struct stat sb; |
a50f9fc5 | 2639 | return lstat(f, &sb) == 0; |
c91f0d92 | 2640 | } |
e6636747 | 2641 | |
0488481e NTND |
2642 | int repo_file_exists(struct repository *repo, const char *path) |
2643 | { | |
2644 | if (repo != the_repository) | |
2645 | BUG("do not know how to check file existence in arbitrary repo"); | |
2646 | ||
2647 | return file_exists(path); | |
2648 | } | |
2649 | ||
63ec5e1f JS |
2650 | static int cmp_icase(char a, char b) |
2651 | { | |
2652 | if (a == b) | |
2653 | return 0; | |
2654 | if (ignore_case) | |
2655 | return toupper(a) - toupper(b); | |
2656 | return a - b; | |
2657 | } | |
2658 | ||
e6636747 | 2659 | /* |
9b125da4 NTND |
2660 | * Given two normalized paths (a trailing slash is ok), if subdir is |
2661 | * outside dir, return -1. Otherwise return the offset in subdir that | |
2662 | * can be used as relative path to dir. | |
e6636747 | 2663 | */ |
9b125da4 | 2664 | int dir_inside_of(const char *subdir, const char *dir) |
e6636747 | 2665 | { |
9b125da4 | 2666 | int offset = 0; |
e6636747 | 2667 | |
9b125da4 | 2668 | assert(dir && subdir && *dir && *subdir); |
e6636747 | 2669 | |
63ec5e1f | 2670 | while (*dir && *subdir && !cmp_icase(*dir, *subdir)) { |
e6636747 | 2671 | dir++; |
9b125da4 NTND |
2672 | subdir++; |
2673 | offset++; | |
490544b1 | 2674 | } |
9b125da4 NTND |
2675 | |
2676 | /* hel[p]/me vs hel[l]/yeah */ | |
2677 | if (*dir && *subdir) | |
2678 | return -1; | |
2679 | ||
2680 | if (!*subdir) | |
2681 | return !*dir ? offset : -1; /* same dir */ | |
2682 | ||
2683 | /* foo/[b]ar vs foo/[] */ | |
2684 | if (is_dir_sep(dir[-1])) | |
2685 | return is_dir_sep(subdir[-1]) ? offset : -1; | |
2686 | ||
2687 | /* foo[/]bar vs foo[] */ | |
2688 | return is_dir_sep(*subdir) ? offset + 1 : -1; | |
e6636747 JS |
2689 | } |
2690 | ||
2691 | int is_inside_dir(const char *dir) | |
2692 | { | |
56b9f6e7 RS |
2693 | char *cwd; |
2694 | int rc; | |
2695 | ||
b892913d NTND |
2696 | if (!dir) |
2697 | return 0; | |
56b9f6e7 RS |
2698 | |
2699 | cwd = xgetcwd(); | |
2700 | rc = (dir_inside_of(cwd, dir) >= 0); | |
2701 | free(cwd); | |
2702 | return rc; | |
e6636747 | 2703 | } |
7155b727 | 2704 | |
55892d23 AP |
2705 | int is_empty_dir(const char *path) |
2706 | { | |
2707 | DIR *dir = opendir(path); | |
2708 | struct dirent *e; | |
2709 | int ret = 1; | |
2710 | ||
2711 | if (!dir) | |
2712 | return 0; | |
2713 | ||
2714 | while ((e = readdir(dir)) != NULL) | |
2715 | if (!is_dot_or_dotdot(e->d_name)) { | |
2716 | ret = 0; | |
2717 | break; | |
2718 | } | |
2719 | ||
2720 | closedir(dir); | |
2721 | return ret; | |
2722 | } | |
2723 | ||
ae2f203e | 2724 | static int remove_dir_recurse(struct strbuf *path, int flag, int *kept_up) |
7155b727 | 2725 | { |
a0f4afbe | 2726 | DIR *dir; |
7155b727 | 2727 | struct dirent *e; |
ae2f203e | 2728 | int ret = 0, original_len = path->len, len, kept_down = 0; |
a0f4afbe | 2729 | int only_empty = (flag & REMOVE_DIR_EMPTY_ONLY); |
c844a803 | 2730 | int keep_toplevel = (flag & REMOVE_DIR_KEEP_TOPLEVEL); |
1053fe82 | 2731 | struct object_id submodule_head; |
7155b727 | 2732 | |
a0f4afbe | 2733 | if ((flag & REMOVE_DIR_KEEP_NESTED_GIT) && |
a98e6101 | 2734 | !resolve_gitlink_ref(path->buf, "HEAD", &submodule_head)) { |
a0f4afbe | 2735 | /* Do not descend and nuke a nested git work tree. */ |
ae2f203e JH |
2736 | if (kept_up) |
2737 | *kept_up = 1; | |
a0f4afbe | 2738 | return 0; |
ae2f203e | 2739 | } |
a0f4afbe | 2740 | |
ae2f203e | 2741 | flag &= ~REMOVE_DIR_KEEP_TOPLEVEL; |
a0f4afbe | 2742 | dir = opendir(path->buf); |
c844a803 | 2743 | if (!dir) { |
863808cd MH |
2744 | if (errno == ENOENT) |
2745 | return keep_toplevel ? -1 : 0; | |
2746 | else if (errno == EACCES && !keep_toplevel) | |
ecb2c282 MH |
2747 | /* |
2748 | * An empty dir could be removable even if it | |
2749 | * is unreadable: | |
2750 | */ | |
c844a803 JH |
2751 | return rmdir(path->buf); |
2752 | else | |
2753 | return -1; | |
2754 | } | |
00b6c178 | 2755 | strbuf_complete(path, '/'); |
7155b727 JS |
2756 | |
2757 | len = path->len; | |
2758 | while ((e = readdir(dir)) != NULL) { | |
2759 | struct stat st; | |
8ca12c0d AP |
2760 | if (is_dot_or_dotdot(e->d_name)) |
2761 | continue; | |
7155b727 JS |
2762 | |
2763 | strbuf_setlen(path, len); | |
2764 | strbuf_addstr(path, e->d_name); | |
863808cd MH |
2765 | if (lstat(path->buf, &st)) { |
2766 | if (errno == ENOENT) | |
2767 | /* | |
2768 | * file disappeared, which is what we | |
2769 | * wanted anyway | |
2770 | */ | |
2771 | continue; | |
15beaaa3 | 2772 | /* fall through */ |
863808cd | 2773 | } else if (S_ISDIR(st.st_mode)) { |
ae2f203e | 2774 | if (!remove_dir_recurse(path, flag, &kept_down)) |
7155b727 | 2775 | continue; /* happy */ |
863808cd MH |
2776 | } else if (!only_empty && |
2777 | (!unlink(path->buf) || errno == ENOENT)) { | |
7155b727 | 2778 | continue; /* happy, too */ |
863808cd | 2779 | } |
7155b727 JS |
2780 | |
2781 | /* path too long, stat fails, or non-directory still exists */ | |
2782 | ret = -1; | |
2783 | break; | |
2784 | } | |
2785 | closedir(dir); | |
2786 | ||
2787 | strbuf_setlen(path, original_len); | |
ae2f203e | 2788 | if (!ret && !keep_toplevel && !kept_down) |
863808cd | 2789 | ret = (!rmdir(path->buf) || errno == ENOENT) ? 0 : -1; |
ae2f203e JH |
2790 | else if (kept_up) |
2791 | /* | |
2792 | * report the uplevel that it is not an error that we | |
2793 | * did not rmdir() our directory. | |
2794 | */ | |
2795 | *kept_up = !ret; | |
7155b727 JS |
2796 | return ret; |
2797 | } | |
039bc64e | 2798 | |
ae2f203e JH |
2799 | int remove_dir_recursively(struct strbuf *path, int flag) |
2800 | { | |
2801 | return remove_dir_recurse(path, flag, NULL); | |
2802 | } | |
2803 | ||
f932729c JK |
2804 | static GIT_PATH_FUNC(git_path_info_exclude, "info/exclude") |
2805 | ||
039bc64e JH |
2806 | void setup_standard_excludes(struct dir_struct *dir) |
2807 | { | |
039bc64e | 2808 | dir->exclude_per_dir = ".gitignore"; |
099d2d86 | 2809 | |
51d18631 | 2810 | /* core.excludesfile defaulting to $XDG_CONFIG_HOME/git/ignore */ |
2845ce7f PT |
2811 | if (!excludes_file) |
2812 | excludes_file = xdg_config_home("ignore"); | |
4698c8fe | 2813 | if (excludes_file && !access_or_warn(excludes_file, R_OK, 0)) |
65edd96a | 2814 | add_patterns_from_file_1(dir, excludes_file, |
38ccaf93 | 2815 | dir->untracked ? &dir->ss_excludes_file : NULL); |
099d2d86 JH |
2816 | |
2817 | /* per repository user preference */ | |
f0056f64 JK |
2818 | if (startup_info->have_repository) { |
2819 | const char *path = git_path_info_exclude(); | |
2820 | if (!access_or_warn(path, R_OK, 0)) | |
65edd96a | 2821 | add_patterns_from_file_1(dir, path, |
f0056f64 JK |
2822 | dir->untracked ? &dir->ss_info_exclude : NULL); |
2823 | } | |
039bc64e | 2824 | } |
4a92d1bf AR |
2825 | |
2826 | int remove_path(const char *name) | |
2827 | { | |
2828 | char *slash; | |
2829 | ||
c7054209 | 2830 | if (unlink(name) && !is_missing_file_error(errno)) |
4a92d1bf AR |
2831 | return -1; |
2832 | ||
2833 | slash = strrchr(name, '/'); | |
2834 | if (slash) { | |
2835 | char *dirs = xstrdup(name); | |
2836 | slash = dirs + (slash - name); | |
2837 | do { | |
2838 | *slash = '\0'; | |
3fc0d131 | 2839 | } while (rmdir(dirs) == 0 && (slash = strrchr(dirs, '/'))); |
4a92d1bf AR |
2840 | free(dirs); |
2841 | } | |
2842 | return 0; | |
2843 | } | |
2844 | ||
270be816 AS |
2845 | /* |
2846 | * Frees memory within dir which was allocated for exclude lists and | |
2847 | * the exclude_stack. Does not free dir itself. | |
2848 | */ | |
2849 | void clear_directory(struct dir_struct *dir) | |
2850 | { | |
2851 | int i, j; | |
2852 | struct exclude_list_group *group; | |
caa3d554 | 2853 | struct pattern_list *pl; |
270be816 AS |
2854 | struct exclude_stack *stk; |
2855 | ||
2856 | for (i = EXC_CMDL; i <= EXC_FILE; i++) { | |
2857 | group = &dir->exclude_list_group[i]; | |
2858 | for (j = 0; j < group->nr; j++) { | |
caa3d554 | 2859 | pl = &group->pl[j]; |
270be816 | 2860 | if (i == EXC_DIRS) |
caa3d554 | 2861 | free((char *)pl->src); |
65edd96a | 2862 | clear_pattern_list(pl); |
270be816 | 2863 | } |
caa3d554 | 2864 | free(group->pl); |
270be816 AS |
2865 | } |
2866 | ||
2867 | stk = dir->exclude_stack; | |
2868 | while (stk) { | |
2869 | struct exclude_stack *prev = stk->prev; | |
2870 | free(stk); | |
2871 | stk = prev; | |
2872 | } | |
aceb9429 | 2873 | strbuf_release(&dir->basebuf); |
270be816 | 2874 | } |
83c094ad NTND |
2875 | |
2876 | struct ondisk_untracked_cache { | |
2877 | struct stat_data info_exclude_stat; | |
2878 | struct stat_data excludes_file_stat; | |
2879 | uint32_t dir_flags; | |
83c094ad NTND |
2880 | }; |
2881 | ||
268ba201 | 2882 | #define ouc_offset(x) offsetof(struct ondisk_untracked_cache, x) |
83c094ad NTND |
2883 | |
2884 | struct write_data { | |
2885 | int index; /* number of written untracked_cache_dir */ | |
2886 | struct ewah_bitmap *check_only; /* from untracked_cache_dir */ | |
2887 | struct ewah_bitmap *valid; /* from untracked_cache_dir */ | |
2888 | struct ewah_bitmap *sha1_valid; /* set if exclude_sha1 is not null */ | |
2889 | struct strbuf out; | |
2890 | struct strbuf sb_stat; | |
2891 | struct strbuf sb_sha1; | |
2892 | }; | |
2893 | ||
2894 | static void stat_data_to_disk(struct stat_data *to, const struct stat_data *from) | |
2895 | { | |
2896 | to->sd_ctime.sec = htonl(from->sd_ctime.sec); | |
2897 | to->sd_ctime.nsec = htonl(from->sd_ctime.nsec); | |
2898 | to->sd_mtime.sec = htonl(from->sd_mtime.sec); | |
2899 | to->sd_mtime.nsec = htonl(from->sd_mtime.nsec); | |
2900 | to->sd_dev = htonl(from->sd_dev); | |
2901 | to->sd_ino = htonl(from->sd_ino); | |
2902 | to->sd_uid = htonl(from->sd_uid); | |
2903 | to->sd_gid = htonl(from->sd_gid); | |
2904 | to->sd_size = htonl(from->sd_size); | |
2905 | } | |
2906 | ||
2907 | static void write_one_dir(struct untracked_cache_dir *untracked, | |
2908 | struct write_data *wd) | |
2909 | { | |
2910 | struct stat_data stat_data; | |
2911 | struct strbuf *out = &wd->out; | |
2912 | unsigned char intbuf[16]; | |
2913 | unsigned int intlen, value; | |
2914 | int i = wd->index++; | |
2915 | ||
2916 | /* | |
2917 | * untracked_nr should be reset whenever valid is clear, but | |
2918 | * for safety.. | |
2919 | */ | |
2920 | if (!untracked->valid) { | |
2921 | untracked->untracked_nr = 0; | |
2922 | untracked->check_only = 0; | |
2923 | } | |
2924 | ||
2925 | if (untracked->check_only) | |
2926 | ewah_set(wd->check_only, i); | |
2927 | if (untracked->valid) { | |
2928 | ewah_set(wd->valid, i); | |
2929 | stat_data_to_disk(&stat_data, &untracked->stat_data); | |
2930 | strbuf_add(&wd->sb_stat, &stat_data, sizeof(stat_data)); | |
2931 | } | |
70c369cd | 2932 | if (!is_null_oid(&untracked->exclude_oid)) { |
83c094ad | 2933 | ewah_set(wd->sha1_valid, i); |
70c369cd | 2934 | strbuf_add(&wd->sb_sha1, untracked->exclude_oid.hash, |
2935 | the_hash_algo->rawsz); | |
83c094ad NTND |
2936 | } |
2937 | ||
2938 | intlen = encode_varint(untracked->untracked_nr, intbuf); | |
2939 | strbuf_add(out, intbuf, intlen); | |
2940 | ||
2941 | /* skip non-recurse directories */ | |
2942 | for (i = 0, value = 0; i < untracked->dirs_nr; i++) | |
2943 | if (untracked->dirs[i]->recurse) | |
2944 | value++; | |
2945 | intlen = encode_varint(value, intbuf); | |
2946 | strbuf_add(out, intbuf, intlen); | |
2947 | ||
2948 | strbuf_add(out, untracked->name, strlen(untracked->name) + 1); | |
2949 | ||
2950 | for (i = 0; i < untracked->untracked_nr; i++) | |
2951 | strbuf_add(out, untracked->untracked[i], | |
2952 | strlen(untracked->untracked[i]) + 1); | |
2953 | ||
2954 | for (i = 0; i < untracked->dirs_nr; i++) | |
2955 | if (untracked->dirs[i]->recurse) | |
2956 | write_one_dir(untracked->dirs[i], wd); | |
2957 | } | |
2958 | ||
2959 | void write_untracked_extension(struct strbuf *out, struct untracked_cache *untracked) | |
2960 | { | |
2961 | struct ondisk_untracked_cache *ouc; | |
2962 | struct write_data wd; | |
2963 | unsigned char varbuf[16]; | |
e0b83735 | 2964 | int varint_len; |
3899b88b | 2965 | const unsigned hashsz = the_hash_algo->rawsz; |
e0b83735 | 2966 | |
3899b88b | 2967 | ouc = xcalloc(1, sizeof(*ouc)); |
83c094ad NTND |
2968 | stat_data_to_disk(&ouc->info_exclude_stat, &untracked->ss_info_exclude.stat); |
2969 | stat_data_to_disk(&ouc->excludes_file_stat, &untracked->ss_excludes_file.stat); | |
83c094ad | 2970 | ouc->dir_flags = htonl(untracked->dir_flags); |
1e8fef60 NTND |
2971 | |
2972 | varint_len = encode_varint(untracked->ident.len, varbuf); | |
2973 | strbuf_add(out, varbuf, varint_len); | |
8109984d | 2974 | strbuf_addbuf(out, &untracked->ident); |
1e8fef60 | 2975 | |
3899b88b | 2976 | strbuf_add(out, ouc, sizeof(*ouc)); |
2977 | strbuf_add(out, untracked->ss_info_exclude.oid.hash, hashsz); | |
2978 | strbuf_add(out, untracked->ss_excludes_file.oid.hash, hashsz); | |
2979 | strbuf_add(out, untracked->exclude_per_dir, strlen(untracked->exclude_per_dir) + 1); | |
6a83d902 | 2980 | FREE_AND_NULL(ouc); |
83c094ad NTND |
2981 | |
2982 | if (!untracked->root) { | |
2983 | varint_len = encode_varint(0, varbuf); | |
2984 | strbuf_add(out, varbuf, varint_len); | |
2985 | return; | |
2986 | } | |
2987 | ||
2988 | wd.index = 0; | |
2989 | wd.check_only = ewah_new(); | |
2990 | wd.valid = ewah_new(); | |
2991 | wd.sha1_valid = ewah_new(); | |
2992 | strbuf_init(&wd.out, 1024); | |
2993 | strbuf_init(&wd.sb_stat, 1024); | |
2994 | strbuf_init(&wd.sb_sha1, 1024); | |
2995 | write_one_dir(untracked->root, &wd); | |
2996 | ||
2997 | varint_len = encode_varint(wd.index, varbuf); | |
2998 | strbuf_add(out, varbuf, varint_len); | |
2999 | strbuf_addbuf(out, &wd.out); | |
3000 | ewah_serialize_strbuf(wd.valid, out); | |
3001 | ewah_serialize_strbuf(wd.check_only, out); | |
3002 | ewah_serialize_strbuf(wd.sha1_valid, out); | |
3003 | strbuf_addbuf(out, &wd.sb_stat); | |
3004 | strbuf_addbuf(out, &wd.sb_sha1); | |
3005 | strbuf_addch(out, '\0'); /* safe guard for string lists */ | |
3006 | ||
3007 | ewah_free(wd.valid); | |
3008 | ewah_free(wd.check_only); | |
3009 | ewah_free(wd.sha1_valid); | |
3010 | strbuf_release(&wd.out); | |
3011 | strbuf_release(&wd.sb_stat); | |
3012 | strbuf_release(&wd.sb_sha1); | |
3013 | } | |
f9e6c649 NTND |
3014 | |
3015 | static void free_untracked(struct untracked_cache_dir *ucd) | |
3016 | { | |
3017 | int i; | |
3018 | if (!ucd) | |
3019 | return; | |
3020 | for (i = 0; i < ucd->dirs_nr; i++) | |
3021 | free_untracked(ucd->dirs[i]); | |
3022 | for (i = 0; i < ucd->untracked_nr; i++) | |
3023 | free(ucd->untracked[i]); | |
3024 | free(ucd->untracked); | |
3025 | free(ucd->dirs); | |
3026 | free(ucd); | |
3027 | } | |
3028 | ||
3029 | void free_untracked_cache(struct untracked_cache *uc) | |
3030 | { | |
3031 | if (uc) | |
3032 | free_untracked(uc->root); | |
3033 | free(uc); | |
3034 | } | |
3035 | ||
3036 | struct read_data { | |
3037 | int index; | |
3038 | struct untracked_cache_dir **ucd; | |
3039 | struct ewah_bitmap *check_only; | |
3040 | struct ewah_bitmap *valid; | |
3041 | struct ewah_bitmap *sha1_valid; | |
3042 | const unsigned char *data; | |
3043 | const unsigned char *end; | |
3044 | }; | |
3045 | ||
268ba201 | 3046 | static void stat_data_from_disk(struct stat_data *to, const unsigned char *data) |
f9e6c649 | 3047 | { |
268ba201 RS |
3048 | memcpy(to, data, sizeof(*to)); |
3049 | to->sd_ctime.sec = ntohl(to->sd_ctime.sec); | |
3050 | to->sd_ctime.nsec = ntohl(to->sd_ctime.nsec); | |
3051 | to->sd_mtime.sec = ntohl(to->sd_mtime.sec); | |
3052 | to->sd_mtime.nsec = ntohl(to->sd_mtime.nsec); | |
3053 | to->sd_dev = ntohl(to->sd_dev); | |
3054 | to->sd_ino = ntohl(to->sd_ino); | |
3055 | to->sd_uid = ntohl(to->sd_uid); | |
3056 | to->sd_gid = ntohl(to->sd_gid); | |
3057 | to->sd_size = ntohl(to->sd_size); | |
f9e6c649 NTND |
3058 | } |
3059 | ||
3060 | static int read_one_dir(struct untracked_cache_dir **untracked_, | |
3061 | struct read_data *rd) | |
3062 | { | |
3063 | struct untracked_cache_dir ud, *untracked; | |
b511d6d5 | 3064 | const unsigned char *data = rd->data, *end = rd->end; |
c6909f99 | 3065 | const unsigned char *eos; |
f9e6c649 | 3066 | unsigned int value; |
08bf354d | 3067 | int i; |
f9e6c649 NTND |
3068 | |
3069 | memset(&ud, 0, sizeof(ud)); | |
3070 | ||
b511d6d5 JK |
3071 | value = decode_varint(&data); |
3072 | if (data > end) | |
f9e6c649 NTND |
3073 | return -1; |
3074 | ud.recurse = 1; | |
3075 | ud.untracked_alloc = value; | |
3076 | ud.untracked_nr = value; | |
3077 | if (ud.untracked_nr) | |
b32fa95f | 3078 | ALLOC_ARRAY(ud.untracked, ud.untracked_nr); |
f9e6c649 | 3079 | |
b511d6d5 JK |
3080 | ud.dirs_alloc = ud.dirs_nr = decode_varint(&data); |
3081 | if (data > end) | |
f9e6c649 | 3082 | return -1; |
b32fa95f | 3083 | ALLOC_ARRAY(ud.dirs, ud.dirs_nr); |
f9e6c649 | 3084 | |
c6909f99 JK |
3085 | eos = memchr(data, '\0', end - data); |
3086 | if (!eos || eos == end) | |
f9e6c649 | 3087 | return -1; |
c6909f99 | 3088 | |
08bf354d | 3089 | *untracked_ = untracked = xmalloc(st_add3(sizeof(*untracked), eos - data, 1)); |
f9e6c649 | 3090 | memcpy(untracked, &ud, sizeof(ud)); |
08bf354d | 3091 | memcpy(untracked->name, data, eos - data + 1); |
b511d6d5 | 3092 | data = eos + 1; |
f9e6c649 NTND |
3093 | |
3094 | for (i = 0; i < untracked->untracked_nr; i++) { | |
c6909f99 JK |
3095 | eos = memchr(data, '\0', end - data); |
3096 | if (!eos || eos == end) | |
f9e6c649 | 3097 | return -1; |
08bf354d | 3098 | untracked->untracked[i] = xmemdupz(data, eos - data); |
b511d6d5 | 3099 | data = eos + 1; |
f9e6c649 NTND |
3100 | } |
3101 | ||
3102 | rd->ucd[rd->index++] = untracked; | |
3103 | rd->data = data; | |
3104 | ||
3105 | for (i = 0; i < untracked->dirs_nr; i++) { | |
08bf354d | 3106 | if (read_one_dir(untracked->dirs + i, rd) < 0) |
f9e6c649 NTND |
3107 | return -1; |
3108 | } | |
3109 | return 0; | |
3110 | } | |
3111 | ||
3112 | static void set_check_only(size_t pos, void *cb) | |
3113 | { | |
3114 | struct read_data *rd = cb; | |
3115 | struct untracked_cache_dir *ud = rd->ucd[pos]; | |
3116 | ud->check_only = 1; | |
3117 | } | |
3118 | ||
3119 | static void read_stat(size_t pos, void *cb) | |
3120 | { | |
3121 | struct read_data *rd = cb; | |
3122 | struct untracked_cache_dir *ud = rd->ucd[pos]; | |
3123 | if (rd->data + sizeof(struct stat_data) > rd->end) { | |
3124 | rd->data = rd->end + 1; | |
3125 | return; | |
3126 | } | |
268ba201 | 3127 | stat_data_from_disk(&ud->stat_data, rd->data); |
f9e6c649 NTND |
3128 | rd->data += sizeof(struct stat_data); |
3129 | ud->valid = 1; | |
3130 | } | |
3131 | ||
70c369cd | 3132 | static void read_oid(size_t pos, void *cb) |
f9e6c649 NTND |
3133 | { |
3134 | struct read_data *rd = cb; | |
3135 | struct untracked_cache_dir *ud = rd->ucd[pos]; | |
70c369cd | 3136 | if (rd->data + the_hash_algo->rawsz > rd->end) { |
f9e6c649 NTND |
3137 | rd->data = rd->end + 1; |
3138 | return; | |
3139 | } | |
70c369cd | 3140 | hashcpy(ud->exclude_oid.hash, rd->data); |
3141 | rd->data += the_hash_algo->rawsz; | |
f9e6c649 NTND |
3142 | } |
3143 | ||
4b33e602 PO |
3144 | static void load_oid_stat(struct oid_stat *oid_stat, const unsigned char *data, |
3145 | const unsigned char *sha1) | |
f9e6c649 | 3146 | { |
4b33e602 PO |
3147 | stat_data_from_disk(&oid_stat->stat, data); |
3148 | hashcpy(oid_stat->oid.hash, sha1); | |
3149 | oid_stat->valid = 1; | |
f9e6c649 NTND |
3150 | } |
3151 | ||
3152 | struct untracked_cache *read_untracked_extension(const void *data, unsigned long sz) | |
3153 | { | |
f9e6c649 NTND |
3154 | struct untracked_cache *uc; |
3155 | struct read_data rd; | |
3156 | const unsigned char *next = data, *end = (const unsigned char *)data + sz; | |
1e8fef60 | 3157 | const char *ident; |
1140bf01 JK |
3158 | int ident_len; |
3159 | ssize_t len; | |
268ba201 | 3160 | const char *exclude_per_dir; |
3899b88b | 3161 | const unsigned hashsz = the_hash_algo->rawsz; |
3162 | const unsigned offset = sizeof(struct ondisk_untracked_cache); | |
3163 | const unsigned exclude_per_dir_offset = offset + 2 * hashsz; | |
f9e6c649 NTND |
3164 | |
3165 | if (sz <= 1 || end[-1] != '\0') | |
3166 | return NULL; | |
3167 | end--; | |
3168 | ||
1e8fef60 NTND |
3169 | ident_len = decode_varint(&next); |
3170 | if (next + ident_len > end) | |
3171 | return NULL; | |
3172 | ident = (const char *)next; | |
3173 | next += ident_len; | |
3174 | ||
3899b88b | 3175 | if (next + exclude_per_dir_offset + 1 > end) |
f9e6c649 NTND |
3176 | return NULL; |
3177 | ||
3178 | uc = xcalloc(1, sizeof(*uc)); | |
1e8fef60 NTND |
3179 | strbuf_init(&uc->ident, ident_len); |
3180 | strbuf_add(&uc->ident, ident, ident_len); | |
4b33e602 PO |
3181 | load_oid_stat(&uc->ss_info_exclude, |
3182 | next + ouc_offset(info_exclude_stat), | |
3899b88b | 3183 | next + offset); |
4b33e602 PO |
3184 | load_oid_stat(&uc->ss_excludes_file, |
3185 | next + ouc_offset(excludes_file_stat), | |
3899b88b | 3186 | next + offset + hashsz); |
268ba201 | 3187 | uc->dir_flags = get_be32(next + ouc_offset(dir_flags)); |
3899b88b | 3188 | exclude_per_dir = (const char *)next + exclude_per_dir_offset; |
268ba201 | 3189 | uc->exclude_per_dir = xstrdup(exclude_per_dir); |
f9e6c649 | 3190 | /* NUL after exclude_per_dir is covered by sizeof(*ouc) */ |
3899b88b | 3191 | next += exclude_per_dir_offset + strlen(exclude_per_dir) + 1; |
f9e6c649 NTND |
3192 | if (next >= end) |
3193 | goto done2; | |
3194 | ||
3195 | len = decode_varint(&next); | |
3196 | if (next > end || len == 0) | |
3197 | goto done2; | |
3198 | ||
3199 | rd.valid = ewah_new(); | |
3200 | rd.check_only = ewah_new(); | |
3201 | rd.sha1_valid = ewah_new(); | |
3202 | rd.data = next; | |
3203 | rd.end = end; | |
3204 | rd.index = 0; | |
b32fa95f | 3205 | ALLOC_ARRAY(rd.ucd, len); |
f9e6c649 NTND |
3206 | |
3207 | if (read_one_dir(&uc->root, &rd) || rd.index != len) | |
3208 | goto done; | |
3209 | ||
3210 | next = rd.data; | |
3211 | len = ewah_read_mmap(rd.valid, next, end - next); | |
3212 | if (len < 0) | |
3213 | goto done; | |
3214 | ||
3215 | next += len; | |
3216 | len = ewah_read_mmap(rd.check_only, next, end - next); | |
3217 | if (len < 0) | |
3218 | goto done; | |
3219 | ||
3220 | next += len; | |
3221 | len = ewah_read_mmap(rd.sha1_valid, next, end - next); | |
3222 | if (len < 0) | |
3223 | goto done; | |
3224 | ||
3225 | ewah_each_bit(rd.check_only, set_check_only, &rd); | |
3226 | rd.data = next + len; | |
3227 | ewah_each_bit(rd.valid, read_stat, &rd); | |
70c369cd | 3228 | ewah_each_bit(rd.sha1_valid, read_oid, &rd); |
f9e6c649 NTND |
3229 | next = rd.data; |
3230 | ||
3231 | done: | |
3232 | free(rd.ucd); | |
3233 | ewah_free(rd.valid); | |
3234 | ewah_free(rd.check_only); | |
3235 | ewah_free(rd.sha1_valid); | |
3236 | done2: | |
3237 | if (next != end) { | |
3238 | free_untracked_cache(uc); | |
3239 | uc = NULL; | |
3240 | } | |
3241 | return uc; | |
3242 | } | |
e931371a | 3243 | |
73f9145f NTND |
3244 | static void invalidate_one_directory(struct untracked_cache *uc, |
3245 | struct untracked_cache_dir *ucd) | |
3246 | { | |
3247 | uc->dir_invalidated++; | |
3248 | ucd->valid = 0; | |
3249 | ucd->untracked_nr = 0; | |
3250 | } | |
3251 | ||
3252 | /* | |
3253 | * Normally when an entry is added or removed from a directory, | |
3254 | * invalidating that directory is enough. No need to touch its | |
3255 | * ancestors. When a directory is shown as "foo/bar/" in git-status | |
3256 | * however, deleting or adding an entry may have cascading effect. | |
3257 | * | |
3258 | * Say the "foo/bar/file" has become untracked, we need to tell the | |
3259 | * untracked_cache_dir of "foo" that "bar/" is not an untracked | |
3260 | * directory any more (because "bar" is managed by foo as an untracked | |
3261 | * "file"). | |
3262 | * | |
3263 | * Similarly, if "foo/bar/file" moves from untracked to tracked and it | |
3264 | * was the last untracked entry in the entire "foo", we should show | |
3265 | * "foo/" instead. Which means we have to invalidate past "bar" up to | |
3266 | * "foo". | |
3267 | * | |
3268 | * This function traverses all directories from root to leaf. If there | |
3269 | * is a chance of one of the above cases happening, we invalidate back | |
3270 | * to root. Otherwise we just invalidate the leaf. There may be a more | |
3271 | * sophisticated way than checking for SHOW_OTHER_DIRECTORIES to | |
3272 | * detect these cases and avoid unnecessary invalidation, for example, | |
3273 | * checking for the untracked entry named "bar/" in "foo", but for now | |
3274 | * stick to something safe and simple. | |
3275 | */ | |
3276 | static int invalidate_one_component(struct untracked_cache *uc, | |
3277 | struct untracked_cache_dir *dir, | |
3278 | const char *path, int len) | |
3279 | { | |
3280 | const char *rest = strchr(path, '/'); | |
3281 | ||
3282 | if (rest) { | |
3283 | int component_len = rest - path; | |
3284 | struct untracked_cache_dir *d = | |
3285 | lookup_untracked(uc, dir, path, component_len); | |
3286 | int ret = | |
3287 | invalidate_one_component(uc, d, rest + 1, | |
3288 | len - (component_len + 1)); | |
3289 | if (ret) | |
3290 | invalidate_one_directory(uc, dir); | |
3291 | return ret; | |
3292 | } | |
3293 | ||
3294 | invalidate_one_directory(uc, dir); | |
3295 | return uc->dir_flags & DIR_SHOW_OTHER_DIRECTORIES; | |
3296 | } | |
3297 | ||
e931371a | 3298 | void untracked_cache_invalidate_path(struct index_state *istate, |
0cacebf0 | 3299 | const char *path, int safe_path) |
e931371a | 3300 | { |
e931371a NTND |
3301 | if (!istate->untracked || !istate->untracked->root) |
3302 | return; | |
68f95b26 | 3303 | if (!safe_path && !verify_path(path, 0)) |
0cacebf0 | 3304 | return; |
73f9145f NTND |
3305 | invalidate_one_component(istate->untracked, istate->untracked->root, |
3306 | path, strlen(path)); | |
e931371a NTND |
3307 | } |
3308 | ||
3309 | void untracked_cache_remove_from_index(struct index_state *istate, | |
3310 | const char *path) | |
3311 | { | |
0cacebf0 | 3312 | untracked_cache_invalidate_path(istate, path, 1); |
e931371a NTND |
3313 | } |
3314 | ||
3315 | void untracked_cache_add_to_index(struct index_state *istate, | |
3316 | const char *path) | |
3317 | { | |
0cacebf0 | 3318 | untracked_cache_invalidate_path(istate, path, 1); |
e931371a | 3319 | } |
47e83eb3 | 3320 | |
da62f786 SB |
3321 | static void connect_wt_gitdir_in_nested(const char *sub_worktree, |
3322 | const char *sub_gitdir) | |
3323 | { | |
3324 | int i; | |
3325 | struct repository subrepo; | |
3326 | struct strbuf sub_wt = STRBUF_INIT; | |
3327 | struct strbuf sub_gd = STRBUF_INIT; | |
3328 | ||
3329 | const struct submodule *sub; | |
3330 | ||
3331 | /* If the submodule has no working tree, we can ignore it. */ | |
3332 | if (repo_init(&subrepo, sub_gitdir, sub_worktree)) | |
3333 | return; | |
3334 | ||
3335 | if (repo_read_index(&subrepo) < 0) | |
a80897c1 | 3336 | die(_("index file corrupt in repo %s"), subrepo.gitdir); |
da62f786 SB |
3337 | |
3338 | for (i = 0; i < subrepo.index->cache_nr; i++) { | |
3339 | const struct cache_entry *ce = subrepo.index->cache[i]; | |
3340 | ||
3341 | if (!S_ISGITLINK(ce->ce_mode)) | |
3342 | continue; | |
3343 | ||
3344 | while (i + 1 < subrepo.index->cache_nr && | |
3345 | !strcmp(ce->name, subrepo.index->cache[i + 1]->name)) | |
3346 | /* | |
3347 | * Skip entries with the same name in different stages | |
3348 | * to make sure an entry is returned only once. | |
3349 | */ | |
3350 | i++; | |
3351 | ||
3352 | sub = submodule_from_path(&subrepo, &null_oid, ce->name); | |
3353 | if (!sub || !is_submodule_active(&subrepo, ce->name)) | |
3354 | /* .gitmodules broken or inactive sub */ | |
3355 | continue; | |
3356 | ||
3357 | strbuf_reset(&sub_wt); | |
3358 | strbuf_reset(&sub_gd); | |
3359 | strbuf_addf(&sub_wt, "%s/%s", sub_worktree, sub->path); | |
3360 | strbuf_addf(&sub_gd, "%s/modules/%s", sub_gitdir, sub->name); | |
3361 | ||
3362 | connect_work_tree_and_git_dir(sub_wt.buf, sub_gd.buf, 1); | |
3363 | } | |
3364 | strbuf_release(&sub_wt); | |
3365 | strbuf_release(&sub_gd); | |
3366 | repo_clear(&subrepo); | |
3367 | } | |
3368 | ||
3369 | void connect_work_tree_and_git_dir(const char *work_tree_, | |
3370 | const char *git_dir_, | |
3371 | int recurse_into_nested) | |
47e83eb3 | 3372 | { |
365444a6 SB |
3373 | struct strbuf gitfile_sb = STRBUF_INIT; |
3374 | struct strbuf cfg_sb = STRBUF_INIT; | |
47e83eb3 | 3375 | struct strbuf rel_path = STRBUF_INIT; |
365444a6 | 3376 | char *git_dir, *work_tree; |
47e83eb3 | 3377 | |
365444a6 SB |
3378 | /* Prepare .git file */ |
3379 | strbuf_addf(&gitfile_sb, "%s/.git", work_tree_); | |
3380 | if (safe_create_leading_directories_const(gitfile_sb.buf)) | |
3381 | die(_("could not create directories for %s"), gitfile_sb.buf); | |
3382 | ||
3383 | /* Prepare config file */ | |
3384 | strbuf_addf(&cfg_sb, "%s/config", git_dir_); | |
3385 | if (safe_create_leading_directories_const(cfg_sb.buf)) | |
3386 | die(_("could not create directories for %s"), cfg_sb.buf); | |
47e83eb3 | 3387 | |
e394fa01 JH |
3388 | git_dir = real_pathdup(git_dir_, 1); |
3389 | work_tree = real_pathdup(work_tree_, 1); | |
365444a6 SB |
3390 | |
3391 | /* Write .git file */ | |
3392 | write_file(gitfile_sb.buf, "gitdir: %s", | |
3393 | relative_path(git_dir, work_tree, &rel_path)); | |
47e83eb3 | 3394 | /* Update core.worktree setting */ |
365444a6 | 3395 | git_config_set_in_file(cfg_sb.buf, "core.worktree", |
47e83eb3 SB |
3396 | relative_path(work_tree, git_dir, &rel_path)); |
3397 | ||
365444a6 SB |
3398 | strbuf_release(&gitfile_sb); |
3399 | strbuf_release(&cfg_sb); | |
47e83eb3 | 3400 | strbuf_release(&rel_path); |
da62f786 SB |
3401 | |
3402 | if (recurse_into_nested) | |
3403 | connect_wt_gitdir_in_nested(work_tree, git_dir); | |
3404 | ||
47e83eb3 SB |
3405 | free(work_tree); |
3406 | free(git_dir); | |
3407 | } | |
f6f85861 SB |
3408 | |
3409 | /* | |
3410 | * Migrate the git directory of the given path from old_git_dir to new_git_dir. | |
3411 | */ | |
3412 | void relocate_gitdir(const char *path, const char *old_git_dir, const char *new_git_dir) | |
3413 | { | |
3414 | if (rename(old_git_dir, new_git_dir) < 0) | |
3415 | die_errno(_("could not migrate git directory from '%s' to '%s'"), | |
3416 | old_git_dir, new_git_dir); | |
3417 | ||
da62f786 | 3418 | connect_work_tree_and_git_dir(path, new_git_dir, 0); |
f6f85861 | 3419 | } |