]>
Commit | Line | Data |
---|---|---|
5010cb5f JH |
1 | /* |
2 | * Builtin "git grep" | |
3 | * | |
4 | * Copyright (c) 2006 Junio C Hamano | |
5 | */ | |
6 | #include "cache.h" | |
7 | #include "blob.h" | |
8 | #include "tree.h" | |
9 | #include "commit.h" | |
10 | #include "tag.h" | |
1362671f | 11 | #include "tree-walk.h" |
5010cb5f | 12 | #include "builtin.h" |
3e230fa1 | 13 | #include "parse-options.h" |
60ecac98 | 14 | #include "userdiff.h" |
83b5d2f5 | 15 | #include "grep.h" |
493b7a08 | 16 | #include "quote.h" |
30816237 | 17 | #include "dir.h" |
5f7c643a | 18 | |
5b594f45 FK |
19 | #ifndef NO_PTHREADS |
20 | #include "thread-utils.h" | |
21 | #include <pthread.h> | |
22 | #endif | |
23 | ||
3e230fa1 RS |
24 | static char const * const grep_usage[] = { |
25 | "git grep [options] [-e] <pattern> [<rev>...] [[--] path...]", | |
26 | NULL | |
27 | }; | |
28 | ||
5b594f45 FK |
29 | static int use_threads = 1; |
30 | ||
31 | #ifndef NO_PTHREADS | |
32 | #define THREADS 8 | |
33 | static pthread_t threads[THREADS]; | |
34 | ||
35 | static void *load_sha1(const unsigned char *sha1, unsigned long *size, | |
36 | const char *name); | |
37 | static void *load_file(const char *filename, size_t *sz); | |
38 | ||
39 | enum work_type {WORK_SHA1, WORK_FILE}; | |
40 | ||
41 | /* We use one producer thread and THREADS consumer | |
42 | * threads. The producer adds struct work_items to 'todo' and the | |
43 | * consumers pick work items from the same array. | |
44 | */ | |
45 | struct work_item | |
46 | { | |
47 | enum work_type type; | |
48 | char *name; | |
49 | ||
50 | /* if type == WORK_SHA1, then 'identifier' is a SHA1, | |
51 | * otherwise type == WORK_FILE, and 'identifier' is a NUL | |
52 | * terminated filename. | |
53 | */ | |
54 | void *identifier; | |
55 | char done; | |
56 | struct strbuf out; | |
57 | }; | |
58 | ||
59 | /* In the range [todo_done, todo_start) in 'todo' we have work_items | |
60 | * that have been or are processed by a consumer thread. We haven't | |
61 | * written the result for these to stdout yet. | |
62 | * | |
63 | * The work_items in [todo_start, todo_end) are waiting to be picked | |
64 | * up by a consumer thread. | |
65 | * | |
66 | * The ranges are modulo TODO_SIZE. | |
67 | */ | |
68 | #define TODO_SIZE 128 | |
69 | static struct work_item todo[TODO_SIZE]; | |
70 | static int todo_start; | |
71 | static int todo_end; | |
72 | static int todo_done; | |
73 | ||
74 | /* Has all work items been added? */ | |
75 | static int all_work_added; | |
76 | ||
77 | /* This lock protects all the variables above. */ | |
78 | static pthread_mutex_t grep_mutex; | |
79 | ||
80 | /* Used to serialize calls to read_sha1_file. */ | |
81 | static pthread_mutex_t read_sha1_mutex; | |
82 | ||
83 | #define grep_lock() pthread_mutex_lock(&grep_mutex) | |
84 | #define grep_unlock() pthread_mutex_unlock(&grep_mutex) | |
85 | #define read_sha1_lock() pthread_mutex_lock(&read_sha1_mutex) | |
86 | #define read_sha1_unlock() pthread_mutex_unlock(&read_sha1_mutex) | |
87 | ||
88 | /* Signalled when a new work_item is added to todo. */ | |
89 | static pthread_cond_t cond_add; | |
90 | ||
91 | /* Signalled when the result from one work_item is written to | |
92 | * stdout. | |
93 | */ | |
94 | static pthread_cond_t cond_write; | |
95 | ||
96 | /* Signalled when we are finished with everything. */ | |
97 | static pthread_cond_t cond_result; | |
98 | ||
99 | static void add_work(enum work_type type, char *name, void *id) | |
100 | { | |
101 | grep_lock(); | |
102 | ||
103 | while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) { | |
104 | pthread_cond_wait(&cond_write, &grep_mutex); | |
105 | } | |
106 | ||
107 | todo[todo_end].type = type; | |
108 | todo[todo_end].name = name; | |
109 | todo[todo_end].identifier = id; | |
110 | todo[todo_end].done = 0; | |
111 | strbuf_reset(&todo[todo_end].out); | |
112 | todo_end = (todo_end + 1) % ARRAY_SIZE(todo); | |
113 | ||
114 | pthread_cond_signal(&cond_add); | |
115 | grep_unlock(); | |
116 | } | |
117 | ||
118 | static struct work_item *get_work(void) | |
119 | { | |
120 | struct work_item *ret; | |
121 | ||
122 | grep_lock(); | |
123 | while (todo_start == todo_end && !all_work_added) { | |
124 | pthread_cond_wait(&cond_add, &grep_mutex); | |
125 | } | |
126 | ||
127 | if (todo_start == todo_end && all_work_added) { | |
128 | ret = NULL; | |
129 | } else { | |
130 | ret = &todo[todo_start]; | |
131 | todo_start = (todo_start + 1) % ARRAY_SIZE(todo); | |
132 | } | |
133 | grep_unlock(); | |
134 | return ret; | |
135 | } | |
136 | ||
137 | static void grep_sha1_async(struct grep_opt *opt, char *name, | |
138 | const unsigned char *sha1) | |
139 | { | |
140 | unsigned char *s; | |
141 | s = xmalloc(20); | |
142 | memcpy(s, sha1, 20); | |
143 | add_work(WORK_SHA1, name, s); | |
144 | } | |
145 | ||
146 | static void grep_file_async(struct grep_opt *opt, char *name, | |
147 | const char *filename) | |
148 | { | |
149 | add_work(WORK_FILE, name, xstrdup(filename)); | |
150 | } | |
151 | ||
152 | static void work_done(struct work_item *w) | |
153 | { | |
154 | int old_done; | |
155 | ||
156 | grep_lock(); | |
157 | w->done = 1; | |
158 | old_done = todo_done; | |
159 | for(; todo[todo_done].done && todo_done != todo_start; | |
160 | todo_done = (todo_done+1) % ARRAY_SIZE(todo)) { | |
161 | w = &todo[todo_done]; | |
162 | write_or_die(1, w->out.buf, w->out.len); | |
163 | free(w->name); | |
164 | free(w->identifier); | |
165 | } | |
166 | ||
167 | if (old_done != todo_done) | |
168 | pthread_cond_signal(&cond_write); | |
169 | ||
170 | if (all_work_added && todo_done == todo_end) | |
171 | pthread_cond_signal(&cond_result); | |
172 | ||
173 | grep_unlock(); | |
174 | } | |
175 | ||
176 | static void *run(void *arg) | |
177 | { | |
178 | int hit = 0; | |
179 | struct grep_opt *opt = arg; | |
180 | ||
181 | while (1) { | |
182 | struct work_item *w = get_work(); | |
183 | if (!w) | |
184 | break; | |
185 | ||
186 | opt->output_priv = w; | |
187 | if (w->type == WORK_SHA1) { | |
188 | unsigned long sz; | |
189 | void* data = load_sha1(w->identifier, &sz, w->name); | |
190 | ||
191 | if (data) { | |
192 | hit |= grep_buffer(opt, w->name, data, sz); | |
193 | free(data); | |
194 | } | |
195 | } else if (w->type == WORK_FILE) { | |
196 | size_t sz; | |
197 | void* data = load_file(w->identifier, &sz); | |
198 | if (data) { | |
199 | hit |= grep_buffer(opt, w->name, data, sz); | |
200 | free(data); | |
201 | } | |
202 | } else { | |
203 | assert(0); | |
204 | } | |
205 | ||
206 | work_done(w); | |
207 | } | |
208 | ||
209 | return (void*) (intptr_t) hit; | |
210 | } | |
211 | ||
212 | static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size) | |
213 | { | |
214 | struct work_item *w = opt->output_priv; | |
215 | strbuf_add(&w->out, buf, size); | |
216 | } | |
217 | ||
218 | static void start_threads(struct grep_opt *opt) | |
219 | { | |
220 | int i; | |
221 | ||
222 | pthread_mutex_init(&grep_mutex, NULL); | |
223 | pthread_mutex_init(&read_sha1_mutex, NULL); | |
224 | pthread_cond_init(&cond_add, NULL); | |
225 | pthread_cond_init(&cond_write, NULL); | |
226 | pthread_cond_init(&cond_result, NULL); | |
227 | ||
228 | for (i = 0; i < ARRAY_SIZE(todo); i++) { | |
229 | strbuf_init(&todo[i].out, 0); | |
230 | } | |
231 | ||
232 | for (i = 0; i < ARRAY_SIZE(threads); i++) { | |
233 | int err; | |
234 | struct grep_opt *o = grep_opt_dup(opt); | |
235 | o->output = strbuf_out; | |
236 | compile_grep_patterns(o); | |
237 | err = pthread_create(&threads[i], NULL, run, o); | |
238 | ||
239 | if (err) | |
240 | die("grep: failed to create thread: %s", | |
241 | strerror(err)); | |
242 | } | |
243 | } | |
244 | ||
245 | static int wait_all(void) | |
246 | { | |
247 | int hit = 0; | |
248 | int i; | |
249 | ||
250 | grep_lock(); | |
251 | all_work_added = 1; | |
252 | ||
253 | /* Wait until all work is done. */ | |
254 | while (todo_done != todo_end) | |
255 | pthread_cond_wait(&cond_result, &grep_mutex); | |
256 | ||
257 | /* Wake up all the consumer threads so they can see that there | |
258 | * is no more work to do. | |
259 | */ | |
260 | pthread_cond_broadcast(&cond_add); | |
261 | grep_unlock(); | |
262 | ||
263 | for (i = 0; i < ARRAY_SIZE(threads); i++) { | |
264 | void *h; | |
265 | pthread_join(threads[i], &h); | |
266 | hit |= (int) (intptr_t) h; | |
267 | } | |
268 | ||
269 | pthread_mutex_destroy(&grep_mutex); | |
270 | pthread_mutex_destroy(&read_sha1_mutex); | |
271 | pthread_cond_destroy(&cond_add); | |
272 | pthread_cond_destroy(&cond_write); | |
273 | pthread_cond_destroy(&cond_result); | |
274 | ||
275 | return hit; | |
276 | } | |
277 | #else /* !NO_PTHREADS */ | |
278 | #define read_sha1_lock() | |
279 | #define read_sha1_unlock() | |
280 | ||
281 | static int wait_all(void) | |
282 | { | |
283 | return 0; | |
284 | } | |
285 | #endif | |
286 | ||
7e8f59d5 RS |
287 | static int grep_config(const char *var, const char *value, void *cb) |
288 | { | |
289 | struct grep_opt *opt = cb; | |
290 | ||
60ecac98 RS |
291 | switch (userdiff_config(var, value)) { |
292 | case 0: break; | |
293 | case -1: return -1; | |
294 | default: return 0; | |
295 | } | |
296 | ||
fe3420b6 | 297 | if (!strcmp(var, "color.grep")) { |
7e8f59d5 RS |
298 | opt->color = git_config_colorbool(var, value, -1); |
299 | return 0; | |
300 | } | |
fe3420b6 | 301 | if (!strcmp(var, "color.grep.match")) { |
7e8f59d5 RS |
302 | if (!value) |
303 | return config_error_nonbool(var); | |
304 | color_parse(value, var, opt->color_match); | |
305 | return 0; | |
306 | } | |
307 | return git_color_default_config(var, value, cb); | |
308 | } | |
309 | ||
a91f453f MK |
310 | /* |
311 | * Return non-zero if max_depth is negative or path has no more then max_depth | |
312 | * slashes. | |
313 | */ | |
314 | static int accept_subdir(const char *path, int max_depth) | |
315 | { | |
316 | if (max_depth < 0) | |
317 | return 1; | |
318 | ||
319 | while ((path = strchr(path, '/')) != NULL) { | |
320 | max_depth--; | |
321 | if (max_depth < 0) | |
322 | return 0; | |
323 | path++; | |
324 | } | |
325 | return 1; | |
326 | } | |
327 | ||
328 | /* | |
329 | * Return non-zero if name is a subdirectory of match and is not too deep. | |
330 | */ | |
331 | static int is_subdir(const char *name, int namelen, | |
332 | const char *match, int matchlen, int max_depth) | |
333 | { | |
334 | if (matchlen > namelen || strncmp(name, match, matchlen)) | |
335 | return 0; | |
336 | ||
337 | if (name[matchlen] == '\0') /* exact match */ | |
338 | return 1; | |
339 | ||
340 | if (!matchlen || match[matchlen-1] == '/' || name[matchlen] == '/') | |
341 | return accept_subdir(name + matchlen + 1, max_depth); | |
342 | ||
343 | return 0; | |
344 | } | |
345 | ||
e0eb889f JH |
346 | /* |
347 | * git grep pathspecs are somewhat different from diff-tree pathspecs; | |
348 | * pathname wildcards are allowed. | |
349 | */ | |
a91f453f | 350 | static int pathspec_matches(const char **paths, const char *name, int max_depth) |
5010cb5f | 351 | { |
e0eb889f | 352 | int namelen, i; |
1362671f | 353 | if (!paths || !*paths) |
a91f453f | 354 | return accept_subdir(name, max_depth); |
5010cb5f | 355 | namelen = strlen(name); |
1362671f JH |
356 | for (i = 0; paths[i]; i++) { |
357 | const char *match = paths[i]; | |
358 | int matchlen = strlen(match); | |
1e3d90e0 | 359 | const char *cp, *meta; |
e0eb889f | 360 | |
a91f453f | 361 | if (is_subdir(name, namelen, match, matchlen, max_depth)) |
e0eb889f JH |
362 | return 1; |
363 | if (!fnmatch(match, name, 0)) | |
364 | return 1; | |
365 | if (name[namelen-1] != '/') | |
5010cb5f | 366 | continue; |
e0eb889f | 367 | |
1e3d90e0 | 368 | /* We are being asked if the directory ("name") is worth |
e0eb889f JH |
369 | * descending into. |
370 | * | |
371 | * Find the longest leading directory name that does | |
372 | * not have metacharacter in the pathspec; the name | |
373 | * we are looking at must overlap with that directory. | |
5010cb5f | 374 | */ |
1e3d90e0 | 375 | for (cp = match, meta = NULL; cp - match < matchlen; cp++) { |
e0eb889f | 376 | char ch = *cp; |
1e3d90e0 JH |
377 | if (ch == '*' || ch == '[' || ch == '?') { |
378 | meta = cp; | |
5010cb5f | 379 | break; |
1e3d90e0 | 380 | } |
e0eb889f | 381 | } |
1e3d90e0 JH |
382 | if (!meta) |
383 | meta = cp; /* fully literal */ | |
384 | ||
385 | if (namelen <= meta - match) { | |
e0eb889f JH |
386 | /* Looking at "Documentation/" and |
387 | * the pattern says "Documentation/howto/", or | |
1e3d90e0 JH |
388 | * "Documentation/diff*.txt". The name we |
389 | * have should match prefix. | |
e0eb889f JH |
390 | */ |
391 | if (!memcmp(match, name, namelen)) | |
392 | return 1; | |
1e3d90e0 | 393 | continue; |
e0eb889f | 394 | } |
1e3d90e0 JH |
395 | |
396 | if (meta - match < namelen) { | |
e0eb889f | 397 | /* Looking at "Documentation/howto/" and |
1e3d90e0 JH |
398 | * the pattern says "Documentation/h*"; |
399 | * match up to "Do.../h"; this avoids descending | |
400 | * into "Documentation/technical/". | |
e0eb889f | 401 | */ |
1e3d90e0 | 402 | if (!memcmp(match, name, meta - match)) |
e0eb889f | 403 | return 1; |
1e3d90e0 | 404 | continue; |
e0eb889f | 405 | } |
5010cb5f JH |
406 | } |
407 | return 0; | |
408 | } | |
409 | ||
5b594f45 FK |
410 | static void *load_sha1(const unsigned char *sha1, unsigned long *size, |
411 | const char *name) | |
5010cb5f | 412 | { |
21666f1a | 413 | enum object_type type; |
5b594f45 | 414 | char *data; |
0d042fec | 415 | |
5b594f45 FK |
416 | read_sha1_lock(); |
417 | data = read_sha1_file(sha1, &type, size); | |
418 | read_sha1_unlock(); | |
419 | ||
420 | if (!data) | |
5010cb5f | 421 | error("'%s': unable to read %s", name, sha1_to_hex(sha1)); |
5b594f45 FK |
422 | |
423 | return data; | |
424 | } | |
425 | ||
426 | static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, | |
427 | const char *filename, int tree_name_len) | |
428 | { | |
429 | struct strbuf pathbuf = STRBUF_INIT; | |
430 | char *name; | |
431 | ||
0d042fec | 432 | if (opt->relative && opt->prefix_length) { |
5b594f45 FK |
433 | quote_path_relative(filename + tree_name_len, -1, &pathbuf, |
434 | opt->prefix); | |
435 | strbuf_insert(&pathbuf, 0, filename, tree_name_len); | |
436 | } else { | |
437 | strbuf_addstr(&pathbuf, filename); | |
438 | } | |
439 | ||
440 | name = strbuf_detach(&pathbuf, NULL); | |
441 | ||
442 | #ifndef NO_PTHREADS | |
443 | if (use_threads) { | |
444 | grep_sha1_async(opt, name, sha1); | |
445 | return 0; | |
446 | } else | |
447 | #endif | |
448 | { | |
449 | int hit; | |
450 | unsigned long sz; | |
451 | void *data = load_sha1(sha1, &sz, name); | |
452 | if (!data) | |
453 | hit = 0; | |
454 | else | |
455 | hit = grep_buffer(opt, name, data, sz); | |
456 | ||
457 | free(data); | |
458 | free(name); | |
459 | return hit; | |
0d042fec | 460 | } |
5010cb5f JH |
461 | } |
462 | ||
5b594f45 | 463 | static void *load_file(const char *filename, size_t *sz) |
5010cb5f JH |
464 | { |
465 | struct stat st; | |
5010cb5f | 466 | char *data; |
5b594f45 | 467 | int i; |
dc49cd76 | 468 | |
5010cb5f JH |
469 | if (lstat(filename, &st) < 0) { |
470 | err_ret: | |
471 | if (errno != ENOENT) | |
472 | error("'%s': %s", filename, strerror(errno)); | |
473 | return 0; | |
474 | } | |
5010cb5f JH |
475 | if (!S_ISREG(st.st_mode)) |
476 | return 0; | |
5b594f45 | 477 | *sz = xsize_t(st.st_size); |
5010cb5f JH |
478 | i = open(filename, O_RDONLY); |
479 | if (i < 0) | |
480 | goto err_ret; | |
5b594f45 FK |
481 | data = xmalloc(*sz + 1); |
482 | if (st.st_size != read_in_full(i, data, *sz)) { | |
5010cb5f JH |
483 | error("'%s': short read %s", filename, strerror(errno)); |
484 | close(i); | |
485 | free(data); | |
486 | return 0; | |
487 | } | |
488 | close(i); | |
5b594f45 FK |
489 | data[*sz] = 0; |
490 | return data; | |
491 | } | |
492 | ||
493 | static int grep_file(struct grep_opt *opt, const char *filename) | |
494 | { | |
495 | struct strbuf buf = STRBUF_INIT; | |
496 | char *name; | |
497 | ||
0d042fec | 498 | if (opt->relative && opt->prefix_length) |
5b594f45 FK |
499 | quote_path_relative(filename, -1, &buf, opt->prefix); |
500 | else | |
501 | strbuf_addstr(&buf, filename); | |
502 | name = strbuf_detach(&buf, NULL); | |
503 | ||
504 | #ifndef NO_PTHREADS | |
505 | if (use_threads) { | |
506 | grep_file_async(opt, name, filename); | |
507 | return 0; | |
508 | } else | |
509 | #endif | |
510 | { | |
511 | int hit; | |
512 | size_t sz; | |
513 | void *data = load_file(filename, &sz); | |
514 | if (!data) | |
515 | hit = 0; | |
516 | else | |
517 | hit = grep_buffer(opt, name, data, sz); | |
518 | ||
519 | free(data); | |
520 | free(name); | |
521 | return hit; | |
522 | } | |
5010cb5f JH |
523 | } |
524 | ||
bbc09c22 | 525 | static int grep_cache(struct grep_opt *opt, const char **paths, int cached) |
5010cb5f JH |
526 | { |
527 | int hit = 0; | |
528 | int nr; | |
529 | read_cache(); | |
530 | ||
531 | for (nr = 0; nr < active_nr; nr++) { | |
532 | struct cache_entry *ce = active_cache[nr]; | |
7a51ed66 | 533 | if (!S_ISREG(ce->ce_mode)) |
5010cb5f | 534 | continue; |
a91f453f | 535 | if (!pathspec_matches(paths, ce->name, opt->max_depth)) |
5010cb5f | 536 | continue; |
57d43466 NTND |
537 | /* |
538 | * If CE_VALID is on, we assume worktree file and its cache entry | |
539 | * are identical, even if worktree file has been modified, so use | |
540 | * cache version instead | |
541 | */ | |
b4d1690d | 542 | if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) { |
36f2587f JH |
543 | if (ce_stage(ce)) |
544 | continue; | |
0d042fec | 545 | hit |= grep_sha1(opt, ce->sha1, ce->name, 0); |
36f2587f | 546 | } |
5010cb5f JH |
547 | else |
548 | hit |= grep_file(opt, ce->name); | |
36f2587f JH |
549 | if (ce_stage(ce)) { |
550 | do { | |
551 | nr++; | |
552 | } while (nr < active_nr && | |
553 | !strcmp(ce->name, active_cache[nr]->name)); | |
554 | nr--; /* compensate for loop control */ | |
555 | } | |
c8610a2e JH |
556 | if (hit && opt->status_only) |
557 | break; | |
5010cb5f | 558 | } |
b48fb5b6 | 559 | free_grep_patterns(opt); |
5010cb5f JH |
560 | return hit; |
561 | } | |
562 | ||
1362671f | 563 | static int grep_tree(struct grep_opt *opt, const char **paths, |
5010cb5f JH |
564 | struct tree_desc *tree, |
565 | const char *tree_name, const char *base) | |
566 | { | |
5010cb5f JH |
567 | int len; |
568 | int hit = 0; | |
4c068a98 | 569 | struct name_entry entry; |
e0eb889f | 570 | char *down; |
0d042fec | 571 | int tn_len = strlen(tree_name); |
620e2bb9 DP |
572 | struct strbuf pathbuf; |
573 | ||
574 | strbuf_init(&pathbuf, PATH_MAX + tn_len); | |
5010cb5f | 575 | |
0d042fec | 576 | if (tn_len) { |
620e2bb9 DP |
577 | strbuf_add(&pathbuf, tree_name, tn_len); |
578 | strbuf_addch(&pathbuf, ':'); | |
579 | tn_len = pathbuf.len; | |
5010cb5f | 580 | } |
620e2bb9 DP |
581 | strbuf_addstr(&pathbuf, base); |
582 | len = pathbuf.len; | |
5010cb5f | 583 | |
4c068a98 | 584 | while (tree_entry(tree, &entry)) { |
620e2bb9 DP |
585 | int te_len = tree_entry_len(entry.path, entry.sha1); |
586 | pathbuf.len = len; | |
587 | strbuf_add(&pathbuf, entry.path, te_len); | |
5010cb5f | 588 | |
4c068a98 | 589 | if (S_ISDIR(entry.mode)) |
e0eb889f JH |
590 | /* Match "abc/" against pathspec to |
591 | * decide if we want to descend into "abc" | |
592 | * directory. | |
593 | */ | |
620e2bb9 | 594 | strbuf_addch(&pathbuf, '/'); |
e0eb889f | 595 | |
620e2bb9 | 596 | down = pathbuf.buf + tn_len; |
a91f453f | 597 | if (!pathspec_matches(paths, down, opt->max_depth)) |
5010cb5f | 598 | ; |
4c068a98 | 599 | else if (S_ISREG(entry.mode)) |
620e2bb9 | 600 | hit |= grep_sha1(opt, entry.sha1, pathbuf.buf, tn_len); |
4c068a98 | 601 | else if (S_ISDIR(entry.mode)) { |
21666f1a | 602 | enum object_type type; |
5010cb5f JH |
603 | struct tree_desc sub; |
604 | void *data; | |
6fda5e51 LT |
605 | unsigned long size; |
606 | ||
5b594f45 | 607 | read_sha1_lock(); |
6fda5e51 | 608 | data = read_sha1_file(entry.sha1, &type, &size); |
5b594f45 FK |
609 | read_sha1_unlock(); |
610 | ||
5010cb5f JH |
611 | if (!data) |
612 | die("unable to read tree (%s)", | |
4c068a98 | 613 | sha1_to_hex(entry.sha1)); |
6fda5e51 | 614 | init_tree_desc(&sub, data, size); |
1362671f | 615 | hit |= grep_tree(opt, paths, &sub, tree_name, down); |
5010cb5f JH |
616 | free(data); |
617 | } | |
c8610a2e JH |
618 | if (hit && opt->status_only) |
619 | break; | |
5010cb5f | 620 | } |
620e2bb9 | 621 | strbuf_release(&pathbuf); |
5010cb5f JH |
622 | return hit; |
623 | } | |
624 | ||
1362671f | 625 | static int grep_object(struct grep_opt *opt, const char **paths, |
5010cb5f JH |
626 | struct object *obj, const char *name) |
627 | { | |
1974632c | 628 | if (obj->type == OBJ_BLOB) |
0d042fec | 629 | return grep_sha1(opt, obj->sha1, name, 0); |
1974632c | 630 | if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) { |
5010cb5f JH |
631 | struct tree_desc tree; |
632 | void *data; | |
6fda5e51 | 633 | unsigned long size; |
5010cb5f JH |
634 | int hit; |
635 | data = read_object_with_reference(obj->sha1, tree_type, | |
6fda5e51 | 636 | &size, NULL); |
5010cb5f JH |
637 | if (!data) |
638 | die("unable to read tree (%s)", sha1_to_hex(obj->sha1)); | |
6fda5e51 | 639 | init_tree_desc(&tree, data, size); |
1362671f | 640 | hit = grep_tree(opt, paths, &tree, name, ""); |
5010cb5f JH |
641 | free(data); |
642 | return hit; | |
643 | } | |
885a86ab | 644 | die("unable to grep from object of type %s", typename(obj->type)); |
5010cb5f JH |
645 | } |
646 | ||
30816237 JH |
647 | static int grep_directory(struct grep_opt *opt, const char **paths) |
648 | { | |
649 | struct dir_struct dir; | |
650 | int i, hit = 0; | |
651 | ||
652 | memset(&dir, 0, sizeof(dir)); | |
653 | setup_standard_excludes(&dir); | |
654 | ||
655 | fill_directory(&dir, paths); | |
c8610a2e | 656 | for (i = 0; i < dir.nr; i++) { |
30816237 | 657 | hit |= grep_file(opt, dir.entries[i]->name); |
c8610a2e JH |
658 | if (hit && opt->status_only) |
659 | break; | |
660 | } | |
30816237 JH |
661 | free_grep_patterns(opt); |
662 | return hit; | |
663 | } | |
664 | ||
ff3c7f9a RS |
665 | static int context_callback(const struct option *opt, const char *arg, |
666 | int unset) | |
3e230fa1 RS |
667 | { |
668 | struct grep_opt *grep_opt = opt->value; | |
669 | int value; | |
670 | const char *endp; | |
671 | ||
672 | if (unset) { | |
673 | grep_opt->pre_context = grep_opt->post_context = 0; | |
674 | return 0; | |
675 | } | |
676 | value = strtol(arg, (char **)&endp, 10); | |
677 | if (*endp) { | |
678 | return error("switch `%c' expects a numerical value", | |
679 | opt->short_name); | |
680 | } | |
681 | grep_opt->pre_context = grep_opt->post_context = value; | |
682 | return 0; | |
683 | } | |
684 | ||
ff3c7f9a | 685 | static int file_callback(const struct option *opt, const char *arg, int unset) |
3e230fa1 RS |
686 | { |
687 | struct grep_opt *grep_opt = opt->value; | |
688 | FILE *patterns; | |
689 | int lno = 0; | |
cfe370c6 | 690 | struct strbuf sb = STRBUF_INIT; |
3e230fa1 RS |
691 | |
692 | patterns = fopen(arg, "r"); | |
693 | if (!patterns) | |
d824cbba | 694 | die_errno("cannot open '%s'", arg); |
3e230fa1 RS |
695 | while (strbuf_getline(&sb, patterns, '\n') == 0) { |
696 | /* ignore empty line like grep does */ | |
697 | if (sb.len == 0) | |
698 | continue; | |
699 | append_grep_pattern(grep_opt, strbuf_detach(&sb, NULL), arg, | |
700 | ++lno, GREP_PATTERN); | |
701 | } | |
702 | fclose(patterns); | |
703 | strbuf_release(&sb); | |
704 | return 0; | |
705 | } | |
706 | ||
ff3c7f9a | 707 | static int not_callback(const struct option *opt, const char *arg, int unset) |
3e230fa1 RS |
708 | { |
709 | struct grep_opt *grep_opt = opt->value; | |
710 | append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT); | |
711 | return 0; | |
712 | } | |
713 | ||
ff3c7f9a | 714 | static int and_callback(const struct option *opt, const char *arg, int unset) |
3e230fa1 RS |
715 | { |
716 | struct grep_opt *grep_opt = opt->value; | |
717 | append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND); | |
718 | return 0; | |
719 | } | |
720 | ||
ff3c7f9a | 721 | static int open_callback(const struct option *opt, const char *arg, int unset) |
3e230fa1 RS |
722 | { |
723 | struct grep_opt *grep_opt = opt->value; | |
724 | append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN); | |
725 | return 0; | |
726 | } | |
727 | ||
ff3c7f9a | 728 | static int close_callback(const struct option *opt, const char *arg, int unset) |
3e230fa1 RS |
729 | { |
730 | struct grep_opt *grep_opt = opt->value; | |
731 | append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN); | |
732 | return 0; | |
733 | } | |
734 | ||
ff3c7f9a RS |
735 | static int pattern_callback(const struct option *opt, const char *arg, |
736 | int unset) | |
3e230fa1 RS |
737 | { |
738 | struct grep_opt *grep_opt = opt->value; | |
739 | append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN); | |
740 | return 0; | |
741 | } | |
5010cb5f | 742 | |
ff3c7f9a | 743 | static int help_callback(const struct option *opt, const char *arg, int unset) |
3e230fa1 RS |
744 | { |
745 | return -1; | |
746 | } | |
088b084b | 747 | |
a633fca0 | 748 | int cmd_grep(int argc, const char **argv, const char *prefix) |
5010cb5f | 749 | { |
5010cb5f | 750 | int hit = 0; |
5010cb5f | 751 | int cached = 0; |
5acd64ed | 752 | int seen_dashdash = 0; |
bbc09c22 | 753 | int external_grep_allowed__ignored; |
5010cb5f | 754 | struct grep_opt opt; |
1f1e895f | 755 | struct object_array list = { 0, 0, NULL }; |
1362671f | 756 | const char **paths = NULL; |
5acd64ed | 757 | int i; |
3e230fa1 | 758 | int dummy; |
7e622650 | 759 | int nongit = 0, use_index = 1; |
3e230fa1 RS |
760 | struct option options[] = { |
761 | OPT_BOOLEAN(0, "cached", &cached, | |
762 | "search in index instead of in the work tree"), | |
30816237 JH |
763 | OPT_BOOLEAN(0, "index", &use_index, |
764 | "--no-index finds in contents not managed by git"), | |
3e230fa1 RS |
765 | OPT_GROUP(""), |
766 | OPT_BOOLEAN('v', "invert-match", &opt.invert, | |
767 | "show non-matching lines"), | |
5183bf67 BC |
768 | OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case, |
769 | "case insensitive matching"), | |
3e230fa1 RS |
770 | OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp, |
771 | "match patterns only at word boundaries"), | |
772 | OPT_SET_INT('a', "text", &opt.binary, | |
773 | "process binary files as text", GREP_BINARY_TEXT), | |
774 | OPT_SET_INT('I', NULL, &opt.binary, | |
775 | "don't match patterns in binary files", | |
776 | GREP_BINARY_NOMATCH), | |
a91f453f MK |
777 | { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth", |
778 | "descend at most <depth> levels", PARSE_OPT_NONEG, | |
779 | NULL, 1 }, | |
3e230fa1 RS |
780 | OPT_GROUP(""), |
781 | OPT_BIT('E', "extended-regexp", &opt.regflags, | |
782 | "use extended POSIX regular expressions", REG_EXTENDED), | |
783 | OPT_NEGBIT('G', "basic-regexp", &opt.regflags, | |
784 | "use basic POSIX regular expressions (default)", | |
785 | REG_EXTENDED), | |
786 | OPT_BOOLEAN('F', "fixed-strings", &opt.fixed, | |
787 | "interpret patterns as fixed strings"), | |
788 | OPT_GROUP(""), | |
789 | OPT_BOOLEAN('n', NULL, &opt.linenum, "show line numbers"), | |
790 | OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1), | |
791 | OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1), | |
792 | OPT_NEGBIT(0, "full-name", &opt.relative, | |
793 | "show filenames relative to top directory", 1), | |
794 | OPT_BOOLEAN('l', "files-with-matches", &opt.name_only, | |
795 | "show only filenames instead of matching lines"), | |
796 | OPT_BOOLEAN(0, "name-only", &opt.name_only, | |
797 | "synonym for --files-with-matches"), | |
798 | OPT_BOOLEAN('L', "files-without-match", | |
799 | &opt.unmatch_name_only, | |
800 | "show only the names of files without match"), | |
801 | OPT_BOOLEAN('z', "null", &opt.null_following_name, | |
802 | "print NUL after filenames"), | |
803 | OPT_BOOLEAN('c', "count", &opt.count, | |
804 | "show the number of matches instead of matching lines"), | |
805 | OPT_SET_INT(0, "color", &opt.color, "highlight matches", 1), | |
806 | OPT_GROUP(""), | |
807 | OPT_CALLBACK('C', NULL, &opt, "n", | |
808 | "show <n> context lines before and after matches", | |
809 | context_callback), | |
810 | OPT_INTEGER('B', NULL, &opt.pre_context, | |
811 | "show <n> context lines before matches"), | |
812 | OPT_INTEGER('A', NULL, &opt.post_context, | |
813 | "show <n> context lines after matches"), | |
814 | OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM", | |
815 | context_callback), | |
2944e4e6 RS |
816 | OPT_BOOLEAN('p', "show-function", &opt.funcname, |
817 | "show a line with the function name before matches"), | |
3e230fa1 RS |
818 | OPT_GROUP(""), |
819 | OPT_CALLBACK('f', NULL, &opt, "file", | |
820 | "read patterns from file", file_callback), | |
821 | { OPTION_CALLBACK, 'e', NULL, &opt, "pattern", | |
822 | "match <pattern>", PARSE_OPT_NONEG, pattern_callback }, | |
823 | { OPTION_CALLBACK, 0, "and", &opt, NULL, | |
824 | "combine patterns specified with -e", | |
825 | PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback }, | |
826 | OPT_BOOLEAN(0, "or", &dummy, ""), | |
827 | { OPTION_CALLBACK, 0, "not", &opt, NULL, "", | |
828 | PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback }, | |
829 | { OPTION_CALLBACK, '(', NULL, &opt, NULL, "", | |
830 | PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH, | |
831 | open_callback }, | |
832 | { OPTION_CALLBACK, ')', NULL, &opt, NULL, "", | |
833 | PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH, | |
834 | close_callback }, | |
c8610a2e JH |
835 | OPT_BOOLEAN('q', "quick", &opt.status_only, |
836 | "indicate hit with exit status without output"), | |
3e230fa1 RS |
837 | OPT_BOOLEAN(0, "all-match", &opt.all_match, |
838 | "show only matches from files that match all patterns"), | |
839 | OPT_GROUP(""), | |
bbc09c22 JH |
840 | OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored, |
841 | "allow calling of grep(1) (ignored by this build)"), | |
3e230fa1 RS |
842 | { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage", |
843 | PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback }, | |
844 | OPT_END() | |
845 | }; | |
5010cb5f | 846 | |
7e622650 JH |
847 | prefix = setup_git_directory_gently(&nongit); |
848 | ||
9c855c31 JN |
849 | /* |
850 | * 'git grep -h', unlike 'git grep -h <pattern>', is a request | |
851 | * to show usage information and exit. | |
852 | */ | |
853 | if (argc == 2 && !strcmp(argv[1], "-h")) | |
854 | usage_with_options(grep_usage, options); | |
855 | ||
5010cb5f | 856 | memset(&opt, 0, sizeof(opt)); |
493b7a08 | 857 | opt.prefix = prefix; |
0d042fec JH |
858 | opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0; |
859 | opt.relative = 1; | |
7977f0ea | 860 | opt.pathname = 1; |
f9b9faf6 | 861 | opt.pattern_tail = &opt.pattern_list; |
5010cb5f | 862 | opt.regflags = REG_NEWLINE; |
a91f453f | 863 | opt.max_depth = -1; |
5010cb5f | 864 | |
7e8f59d5 RS |
865 | strcpy(opt.color_match, GIT_COLOR_RED GIT_COLOR_BOLD); |
866 | opt.color = -1; | |
867 | git_config(grep_config, &opt); | |
868 | if (opt.color == -1) | |
869 | opt.color = git_use_color_default; | |
870 | ||
5010cb5f | 871 | /* |
5acd64ed JH |
872 | * If there is no -- then the paths must exist in the working |
873 | * tree. If there is no explicit pattern specified with -e or | |
874 | * -f, we take the first unrecognized non option to be the | |
875 | * pattern, but then what follows it must be zero or more | |
876 | * valid refs up to the -- (if exists), and then existing | |
877 | * paths. If there is an explicit pattern, then the first | |
82e5a82f | 878 | * unrecognized non option is the beginning of the refs list |
5acd64ed | 879 | * that continues up to the -- (if exists), and then paths. |
5010cb5f | 880 | */ |
37782920 | 881 | argc = parse_options(argc, argv, prefix, options, grep_usage, |
3e230fa1 RS |
882 | PARSE_OPT_KEEP_DASHDASH | |
883 | PARSE_OPT_STOP_AT_NON_OPTION | | |
884 | PARSE_OPT_NO_INTERNAL_HELP); | |
885 | ||
7e622650 JH |
886 | if (use_index && nongit) |
887 | /* die the same way as if we did it at the beginning */ | |
888 | setup_git_directory(); | |
889 | ||
3e230fa1 RS |
890 | /* First unrecognized non-option token */ |
891 | if (argc > 0 && !opt.pattern_list) { | |
892 | append_grep_pattern(&opt, argv[0], "command line", 0, | |
893 | GREP_PATTERN); | |
894 | argv++; | |
895 | argc--; | |
5010cb5f | 896 | } |
5acd64ed | 897 | |
f9b9faf6 | 898 | if (!opt.pattern_list) |
5010cb5f | 899 | die("no pattern given."); |
5183bf67 BC |
900 | if (!opt.fixed && opt.ignore_case) |
901 | opt.regflags |= REG_ICASE; | |
07ea91d8 JH |
902 | if ((opt.regflags != REG_NEWLINE) && opt.fixed) |
903 | die("cannot mix --fixed-strings and regexp"); | |
5b594f45 FK |
904 | |
905 | #ifndef NO_PTHREADS | |
906 | if (online_cpus() == 1 || !grep_threads_ok(&opt)) | |
907 | use_threads = 0; | |
908 | ||
909 | if (use_threads) | |
910 | start_threads(&opt); | |
911 | #else | |
912 | use_threads = 0; | |
913 | #endif | |
914 | ||
83b5d2f5 | 915 | compile_grep_patterns(&opt); |
5acd64ed JH |
916 | |
917 | /* Check revs and then paths */ | |
3e230fa1 | 918 | for (i = 0; i < argc; i++) { |
5acd64ed | 919 | const char *arg = argv[i]; |
1362671f | 920 | unsigned char sha1[20]; |
5acd64ed JH |
921 | /* Is it a rev? */ |
922 | if (!get_sha1(arg, sha1)) { | |
923 | struct object *object = parse_object(sha1); | |
5acd64ed JH |
924 | if (!object) |
925 | die("bad object %s", arg); | |
1f1e895f | 926 | add_object_array(object, arg, &list); |
5acd64ed JH |
927 | continue; |
928 | } | |
929 | if (!strcmp(arg, "--")) { | |
930 | i++; | |
931 | seen_dashdash = 1; | |
932 | } | |
933 | break; | |
1362671f | 934 | } |
5acd64ed JH |
935 | |
936 | /* The rest are paths */ | |
937 | if (!seen_dashdash) { | |
938 | int j; | |
c39c4f47 | 939 | for (j = i; j < argc; j++) |
5acd64ed JH |
940 | verify_filename(prefix, argv[j]); |
941 | } | |
942 | ||
493b7a08 | 943 | if (i < argc) |
5acd64ed | 944 | paths = get_pathspec(prefix, argv + i); |
1362671f JH |
945 | else if (prefix) { |
946 | paths = xcalloc(2, sizeof(const char *)); | |
947 | paths[0] = prefix; | |
948 | paths[1] = NULL; | |
949 | } | |
5010cb5f | 950 | |
30816237 | 951 | if (!use_index) { |
5b594f45 | 952 | int hit; |
30816237 JH |
953 | if (cached) |
954 | die("--cached cannot be used with --no-index."); | |
955 | if (list.nr) | |
956 | die("--no-index cannot be used with revs."); | |
5b594f45 FK |
957 | hit = grep_directory(&opt, paths); |
958 | if (use_threads) | |
959 | hit |= wait_all(); | |
960 | return !hit; | |
30816237 JH |
961 | } |
962 | ||
6577f542 | 963 | if (!list.nr) { |
5b594f45 | 964 | int hit; |
6577f542 NTND |
965 | if (!cached) |
966 | setup_work_tree(); | |
5b594f45 FK |
967 | |
968 | hit = grep_cache(&opt, paths, cached); | |
969 | if (use_threads) | |
970 | hit |= wait_all(); | |
971 | return !hit; | |
6577f542 | 972 | } |
aa8c79ad | 973 | |
5010cb5f | 974 | if (cached) |
aa8c79ad | 975 | die("both --cached and trees are given."); |
5010cb5f | 976 | |
1f1e895f | 977 | for (i = 0; i < list.nr; i++) { |
5010cb5f | 978 | struct object *real_obj; |
1f1e895f | 979 | real_obj = deref_tag(list.objects[i].item, NULL, 0); |
c8610a2e | 980 | if (grep_object(&opt, paths, real_obj, list.objects[i].name)) { |
5010cb5f | 981 | hit = 1; |
c8610a2e JH |
982 | if (opt.status_only) |
983 | break; | |
984 | } | |
5010cb5f | 985 | } |
5b594f45 FK |
986 | |
987 | if (use_threads) | |
988 | hit |= wait_all(); | |
b48fb5b6 | 989 | free_grep_patterns(&opt); |
5010cb5f JH |
990 | return !hit; |
991 | } |