sha1_file: convert read_object_with_reference to object_id
[git/git.git] / builtin / grep.c
CommitLineData
5010cb5f
JH
1/*
2 * Builtin "git grep"
3 *
4 * Copyright (c) 2006 Junio C Hamano
5 */
6#include "cache.h"
627d9342 7#include "repository.h"
b2141fc1 8#include "config.h"
5010cb5f
JH
9#include "blob.h"
10#include "tree.h"
11#include "commit.h"
12#include "tag.h"
1362671f 13#include "tree-walk.h"
5010cb5f 14#include "builtin.h"
3e230fa1 15#include "parse-options.h"
678e484b
JS
16#include "string-list.h"
17#include "run-command.h"
60ecac98 18#include "userdiff.h"
83b5d2f5 19#include "grep.h"
493b7a08 20#include "quote.h"
59332d13 21#include "dir.h"
64acde94 22#include "pathspec.h"
0281e487 23#include "submodule.h"
74ed4371 24#include "submodule-config.h"
5b594f45 25
3e230fa1 26static char const * const grep_usage[] = {
9c9b4f2f 27 N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
3e230fa1
RS
28 NULL
29};
30
0281e487 31static int recurse_submodules;
0281e487 32
89f09dd3
VL
33#define GREP_NUM_THREADS_DEFAULT 8
34static int num_threads;
5b594f45
FK
35
36#ifndef NO_PTHREADS
89f09dd3 37static pthread_t *threads;
5b594f45 38
5b594f45
FK
39/* We use one producer thread and THREADS consumer
40 * threads. The producer adds struct work_items to 'todo' and the
41 * consumers pick work items from the same array.
42 */
9cba13ca 43struct work_item {
8f24a632 44 struct grep_source source;
5b594f45
FK
45 char done;
46 struct strbuf out;
47};
48
49/* In the range [todo_done, todo_start) in 'todo' we have work_items
50 * that have been or are processed by a consumer thread. We haven't
51 * written the result for these to stdout yet.
52 *
53 * The work_items in [todo_start, todo_end) are waiting to be picked
54 * up by a consumer thread.
55 *
56 * The ranges are modulo TODO_SIZE.
57 */
58#define TODO_SIZE 128
59static struct work_item todo[TODO_SIZE];
60static int todo_start;
61static int todo_end;
62static int todo_done;
63
64/* Has all work items been added? */
65static int all_work_added;
66
67/* This lock protects all the variables above. */
68static pthread_mutex_t grep_mutex;
69
1487a12b
JH
70static inline void grep_lock(void)
71{
8df4c295
ÆAB
72 assert(num_threads);
73 pthread_mutex_lock(&grep_mutex);
1487a12b
JH
74}
75
76static inline void grep_unlock(void)
77{
8df4c295
ÆAB
78 assert(num_threads);
79 pthread_mutex_unlock(&grep_mutex);
1487a12b
JH
80}
81
5b594f45
FK
82/* Signalled when a new work_item is added to todo. */
83static pthread_cond_t cond_add;
84
85/* Signalled when the result from one work_item is written to
86 * stdout.
87 */
88static pthread_cond_t cond_write;
89
90/* Signalled when we are finished with everything. */
91static pthread_cond_t cond_result;
92
08303c36 93static int skip_first_line;
431d6e7b 94
e2e05d61 95static void add_work(struct grep_opt *opt, const struct grep_source *gs)
5b594f45
FK
96{
97 grep_lock();
98
99 while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
100 pthread_cond_wait(&cond_write, &grep_mutex);
101 }
102
e2e05d61 103 todo[todo_end].source = *gs;
9dd5245c
JK
104 if (opt->binary != GREP_BINARY_TEXT)
105 grep_source_load_driver(&todo[todo_end].source);
5b594f45
FK
106 todo[todo_end].done = 0;
107 strbuf_reset(&todo[todo_end].out);
108 todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
109
110 pthread_cond_signal(&cond_add);
111 grep_unlock();
112}
113
114static struct work_item *get_work(void)
115{
116 struct work_item *ret;
117
118 grep_lock();
119 while (todo_start == todo_end && !all_work_added) {
120 pthread_cond_wait(&cond_add, &grep_mutex);
121 }
122
123 if (todo_start == todo_end && all_work_added) {
124 ret = NULL;
125 } else {
126 ret = &todo[todo_start];
127 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
128 }
129 grep_unlock();
130 return ret;
131}
132
5b594f45
FK
133static void work_done(struct work_item *w)
134{
135 int old_done;
136
137 grep_lock();
138 w->done = 1;
139 old_done = todo_done;
140 for(; todo[todo_done].done && todo_done != todo_start;
141 todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
142 w = &todo[todo_done];
431d6e7b 143 if (w->out.len) {
08303c36
RS
144 const char *p = w->out.buf;
145 size_t len = w->out.len;
146
147 /* Skip the leading hunk mark of the first file. */
148 if (skip_first_line) {
149 while (len) {
150 len--;
151 if (*p++ == '\n')
152 break;
153 }
154 skip_first_line = 0;
155 }
156
157 write_or_die(1, p, len);
431d6e7b 158 }
8f24a632 159 grep_source_clear(&w->source);
5b594f45
FK
160 }
161
162 if (old_done != todo_done)
163 pthread_cond_signal(&cond_write);
164
165 if (all_work_added && todo_done == todo_end)
166 pthread_cond_signal(&cond_result);
167
168 grep_unlock();
169}
170
171static void *run(void *arg)
172{
173 int hit = 0;
174 struct grep_opt *opt = arg;
175
176 while (1) {
177 struct work_item *w = get_work();
178 if (!w)
179 break;
180
181 opt->output_priv = w;
f9ee2fcd 182 hit |= grep_source(opt, &w->source);
8f24a632 183 grep_source_clear_data(&w->source);
5b594f45
FK
184 work_done(w);
185 }
bfac23d9
DM
186 free_grep_patterns(arg);
187 free(arg);
5b594f45
FK
188
189 return (void*) (intptr_t) hit;
190}
191
192static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
193{
194 struct work_item *w = opt->output_priv;
195 strbuf_add(&w->out, buf, size);
196}
197
198static void start_threads(struct grep_opt *opt)
199{
200 int i;
201
202 pthread_mutex_init(&grep_mutex, NULL);
b3aeb285 203 pthread_mutex_init(&grep_read_mutex, NULL);
0579f91d 204 pthread_mutex_init(&grep_attr_mutex, NULL);
5b594f45
FK
205 pthread_cond_init(&cond_add, NULL);
206 pthread_cond_init(&cond_write, NULL);
207 pthread_cond_init(&cond_result, NULL);
78db6ea9 208 grep_use_locks = 1;
5b594f45
FK
209
210 for (i = 0; i < ARRAY_SIZE(todo); i++) {
211 strbuf_init(&todo[i].out, 0);
212 }
213
89f09dd3
VL
214 threads = xcalloc(num_threads, sizeof(*threads));
215 for (i = 0; i < num_threads; i++) {
5b594f45
FK
216 int err;
217 struct grep_opt *o = grep_opt_dup(opt);
218 o->output = strbuf_out;
6d423dd5
ÆAB
219 if (i)
220 o->debug = 0;
5b594f45
FK
221 compile_grep_patterns(o);
222 err = pthread_create(&threads[i], NULL, run, o);
223
224 if (err)
2fc5f9f1 225 die(_("grep: failed to create thread: %s"),
5b594f45
FK
226 strerror(err));
227 }
228}
229
230static int wait_all(void)
231{
232 int hit = 0;
233 int i;
234
235 grep_lock();
236 all_work_added = 1;
237
238 /* Wait until all work is done. */
239 while (todo_done != todo_end)
240 pthread_cond_wait(&cond_result, &grep_mutex);
241
242 /* Wake up all the consumer threads so they can see that there
243 * is no more work to do.
244 */
245 pthread_cond_broadcast(&cond_add);
246 grep_unlock();
247
89f09dd3 248 for (i = 0; i < num_threads; i++) {
5b594f45
FK
249 void *h;
250 pthread_join(threads[i], &h);
251 hit |= (int) (intptr_t) h;
252 }
253
89f09dd3
VL
254 free(threads);
255
5b594f45 256 pthread_mutex_destroy(&grep_mutex);
b3aeb285 257 pthread_mutex_destroy(&grep_read_mutex);
0579f91d 258 pthread_mutex_destroy(&grep_attr_mutex);
5b594f45
FK
259 pthread_cond_destroy(&cond_add);
260 pthread_cond_destroy(&cond_write);
261 pthread_cond_destroy(&cond_result);
78db6ea9 262 grep_use_locks = 0;
5b594f45
FK
263
264 return hit;
265}
266#else /* !NO_PTHREADS */
5b594f45
FK
267
268static int wait_all(void)
269{
270 return 0;
271}
272#endif
273
15fabd1b
JH
274static int grep_cmd_config(const char *var, const char *value, void *cb)
275{
276 int st = grep_config(var, value, cb);
33c643bb 277 if (git_color_default_config(var, value, cb) < 0)
15fabd1b 278 st = -1;
89f09dd3
VL
279
280 if (!strcmp(var, "grep.threads")) {
281 num_threads = git_config_int(var, value);
282 if (num_threads < 0)
283 die(_("invalid number of threads specified (%d) for %s"),
284 num_threads, var);
d1edee4a
ÆAB
285#ifdef NO_PTHREADS
286 else if (num_threads && num_threads != 1) {
287 /*
288 * TRANSLATORS: %s is the configuration
289 * variable for tweaking threads, currently
290 * grep.threads
291 */
292 warning(_("no threads support, ignoring %s"), var);
293 num_threads = 0;
294 }
295#endif
89f09dd3
VL
296 }
297
9071c078
SB
298 if (!strcmp(var, "submodule.recurse"))
299 recurse_submodules = git_config_bool(var, value);
300
15fabd1b
JH
301 return st;
302}
303
1db11086 304static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
5f02d315
JH
305{
306 void *data;
307
b3aeb285 308 grep_read_lock();
1db11086 309 data = read_sha1_file(oid->hash, type, size);
b3aeb285 310 grep_read_unlock();
5b594f45
FK
311 return data;
312}
313
1db11086 314static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
55c61688
NTND
315 const char *filename, int tree_name_len,
316 const char *path)
5b594f45
FK
317{
318 struct strbuf pathbuf = STRBUF_INIT;
e2e05d61 319 struct grep_source gs;
5b594f45 320
f9ee2fcd
BW
321 if (opt->relative && opt->prefix_length) {
322 quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
323 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
5b594f45
FK
324 } else {
325 strbuf_addstr(&pathbuf, filename);
326 }
327
e2e05d61 328 grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
38ef24dc 329 strbuf_release(&pathbuf);
e2e05d61 330
5b594f45 331#ifndef NO_PTHREADS
89f09dd3 332 if (num_threads) {
e2e05d61
RV
333 /*
334 * add_work() copies gs and thus assumes ownership of
335 * its fields, so do not call grep_source_clear()
336 */
337 add_work(opt, &gs);
5b594f45
FK
338 return 0;
339 } else
340#endif
341 {
342 int hit;
5010cb5f 343
8f24a632 344 hit = grep_source(opt, &gs);
dc49cd76 345
8f24a632
JK
346 grep_source_clear(&gs);
347 return hit;
5010cb5f 348 }
5b594f45
FK
349}
350
351static int grep_file(struct grep_opt *opt, const char *filename)
352{
353 struct strbuf buf = STRBUF_INIT;
e2e05d61 354 struct grep_source gs;
5b594f45 355
f9ee2fcd
BW
356 if (opt->relative && opt->prefix_length)
357 quote_path_relative(filename, opt->prefix, &buf);
358 else
359 strbuf_addstr(&buf, filename);
5b594f45 360
e2e05d61 361 grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
38ef24dc 362 strbuf_release(&buf);
e2e05d61 363
5b594f45 364#ifndef NO_PTHREADS
89f09dd3 365 if (num_threads) {
e2e05d61
RV
366 /*
367 * add_work() copies gs and thus assumes ownership of
368 * its fields, so do not call grep_source_clear()
369 */
370 add_work(opt, &gs);
5b594f45
FK
371 return 0;
372 } else
373#endif
374 {
375 int hit;
5b594f45 376
8f24a632
JK
377 hit = grep_source(opt, &gs);
378
379 grep_source_clear(&gs);
5b594f45
FK
380 return hit;
381 }
5010cb5f
JH
382}
383
678e484b
JS
384static void append_path(struct grep_opt *opt, const void *data, size_t len)
385{
386 struct string_list *path_list = opt->output_priv;
387
388 if (len == 1 && *(const char *)data == '\0')
389 return;
0c72cead 390 string_list_append(path_list, xstrndup(data, len));
678e484b
JS
391}
392
393static void run_pager(struct grep_opt *opt, const char *prefix)
394{
395 struct string_list *path_list = opt->output_priv;
850d2fec 396 struct child_process child = CHILD_PROCESS_INIT;
678e484b
JS
397 int i, status;
398
399 for (i = 0; i < path_list->nr; i++)
850d2fec
JK
400 argv_array_push(&child.args, path_list->items[i].string);
401 child.dir = prefix;
402 child.use_shell = 1;
678e484b 403
850d2fec 404 status = run_command(&child);
678e484b
JS
405 if (status)
406 exit(status);
678e484b
JS
407}
408
f9ee2fcd
BW
409static int grep_cache(struct grep_opt *opt, struct repository *repo,
410 const struct pathspec *pathspec, int cached);
411static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
412 struct tree_desc *tree, struct strbuf *base, int tn_len,
413 int check_attr, struct repository *repo);
0281e487 414
f9ee2fcd
BW
415static int grep_submodule(struct grep_opt *opt, struct repository *superproject,
416 const struct pathspec *pathspec,
417 const struct object_id *oid,
418 const char *filename, const char *path)
0281e487 419{
f9ee2fcd
BW
420 struct repository submodule;
421 int hit;
74ed4371 422
f9ee2fcd
BW
423 if (!is_submodule_active(superproject, path))
424 return 0;
0281e487 425
f9ee2fcd
BW
426 if (repo_submodule_init(&submodule, superproject, path))
427 return 0;
be80a239 428
f9ee2fcd 429 repo_read_gitmodules(&submodule);
0281e487 430
74ed4371 431 /*
f9ee2fcd
BW
432 * NEEDSWORK: This adds the submodule's object directory to the list of
433 * alternates for the single in-memory object store. This has some bad
434 * consequences for memory (processed objects will never be freed) and
435 * performance (this increases the number of pack files git has to pay
436 * attention to, to the sum of the number of pack files in all the
437 * repositories processed so far). This can be removed once the object
438 * store is no longer global and instead is a member of the repository
439 * object.
74ed4371 440 */
9560e624 441 grep_read_lock();
f9ee2fcd 442 add_to_alternates_memory(submodule.objectdir);
9560e624 443 grep_read_unlock();
74ed4371 444
f9ee2fcd
BW
445 if (oid) {
446 struct object *object;
447 struct tree_desc tree;
448 void *data;
449 unsigned long size;
450 struct strbuf base = STRBUF_INIT;
74ed4371 451
f9ee2fcd 452 object = parse_object_or_die(oid, oid_to_hex(oid));
0281e487 453
f9ee2fcd 454 grep_read_lock();
02f0547e 455 data = read_object_with_reference(&object->oid, tree_type,
f9ee2fcd
BW
456 &size, NULL);
457 grep_read_unlock();
0281e487 458
f9ee2fcd
BW
459 if (!data)
460 die(_("unable to read tree (%s)"), oid_to_hex(&object->oid));
0281e487 461
f9ee2fcd
BW
462 strbuf_addstr(&base, filename);
463 strbuf_addch(&base, '/');
0281e487 464
f9ee2fcd
BW
465 init_tree_desc(&tree, data, size);
466 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
467 object->type == OBJ_COMMIT, &submodule);
468 strbuf_release(&base);
469 free(data);
470 } else {
471 hit = grep_cache(opt, &submodule, pathspec, 1);
e6fac7f3 472 }
0281e487 473
f9ee2fcd
BW
474 repo_clear(&submodule);
475 return hit;
0281e487
BW
476}
477
f9ee2fcd
BW
478static int grep_cache(struct grep_opt *opt, struct repository *repo,
479 const struct pathspec *pathspec, int cached)
5010cb5f
JH
480{
481 int hit = 0;
482 int nr;
0281e487
BW
483 struct strbuf name = STRBUF_INIT;
484 int name_base_len = 0;
f9ee2fcd
BW
485 if (repo->submodule_prefix) {
486 name_base_len = strlen(repo->submodule_prefix);
487 strbuf_addstr(&name, repo->submodule_prefix);
0281e487
BW
488 }
489
f9ee2fcd 490 repo_read_index(repo);
5010cb5f 491
f9ee2fcd
BW
492 for (nr = 0; nr < repo->index->cache_nr; nr++) {
493 const struct cache_entry *ce = repo->index->cache[nr];
0281e487
BW
494 strbuf_setlen(&name, name_base_len);
495 strbuf_addstr(&name, ce->name);
496
497 if (S_ISREG(ce->ce_mode) &&
498 match_pathspec(pathspec, name.buf, name.len, 0, NULL,
499 S_ISDIR(ce->ce_mode) ||
500 S_ISGITLINK(ce->ce_mode))) {
501 /*
502 * If CE_VALID is on, we assume worktree file and its
503 * cache entry are identical, even if worktree file has
504 * been modified, so use cache version instead
505 */
506 if (cached || (ce->ce_flags & CE_VALID) ||
507 ce_skip_worktree(ce)) {
508 if (ce_stage(ce) || ce_intent_to_add(ce))
509 continue;
f9ee2fcd
BW
510 hit |= grep_oid(opt, &ce->oid, name.buf,
511 0, name.buf);
0281e487 512 } else {
f9ee2fcd 513 hit |= grep_file(opt, name.buf);
0281e487
BW
514 }
515 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
516 submodule_path_match(pathspec, name.buf, NULL)) {
f9ee2fcd 517 hit |= grep_submodule(opt, repo, pathspec, NULL, ce->name, ce->name);
0281e487 518 } else {
5010cb5f 519 continue;
36f2587f 520 }
0281e487 521
36f2587f
JH
522 if (ce_stage(ce)) {
523 do {
524 nr++;
f9ee2fcd
BW
525 } while (nr < repo->index->cache_nr &&
526 !strcmp(ce->name, repo->index->cache[nr]->name));
36f2587f
JH
527 nr--; /* compensate for loop control */
528 }
c8610a2e
JH
529 if (hit && opt->status_only)
530 break;
5010cb5f 531 }
0281e487
BW
532
533 strbuf_release(&name);
5010cb5f
JH
534 return hit;
535}
536
f34bbc15 537static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
55c61688 538 struct tree_desc *tree, struct strbuf *base, int tn_len,
f9ee2fcd 539 int check_attr, struct repository *repo)
5010cb5f 540{
d688cf07
NTND
541 int hit = 0;
542 enum interesting match = entry_not_interesting;
4c068a98 543 struct name_entry entry;
e5e062b6 544 int old_baselen = base->len;
74ed4371
BW
545 struct strbuf name = STRBUF_INIT;
546 int name_base_len = 0;
f9ee2fcd
BW
547 if (repo->submodule_prefix) {
548 strbuf_addstr(&name, repo->submodule_prefix);
74ed4371
BW
549 name_base_len = name.len;
550 }
5010cb5f 551
4c068a98 552 while (tree_entry(tree, &entry)) {
0de16337 553 int te_len = tree_entry_len(&entry);
e5e062b6 554
d688cf07 555 if (match != all_entries_interesting) {
74ed4371
BW
556 strbuf_addstr(&name, base->buf + tn_len);
557 match = tree_entry_interesting(&entry, &name,
558 0, pathspec);
559 strbuf_setlen(&name, name_base_len);
560
d688cf07 561 if (match == all_entries_not_interesting)
97d0b74a 562 break;
d688cf07 563 if (match == entry_not_interesting)
1376e507
NTND
564 continue;
565 }
5010cb5f 566
1376e507 567 strbuf_add(base, entry.path, te_len);
e0eb889f 568
1376e507 569 if (S_ISREG(entry.mode)) {
1db11086 570 hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
55c61688 571 check_attr ? base->buf + tn_len : NULL);
74ed4371 572 } else if (S_ISDIR(entry.mode)) {
21666f1a 573 enum object_type type;
5010cb5f
JH
574 struct tree_desc sub;
575 void *data;
6fda5e51
LT
576 unsigned long size;
577
1db11086 578 data = lock_and_read_oid_file(entry.oid, &type, &size);
5010cb5f 579 if (!data)
2fc5f9f1 580 die(_("unable to read tree (%s)"),
7d924c91 581 oid_to_hex(entry.oid));
1376e507
NTND
582
583 strbuf_addch(base, '/');
6fda5e51 584 init_tree_desc(&sub, data, size);
55c61688 585 hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
f9ee2fcd 586 check_attr, repo);
5010cb5f 587 free(data);
74ed4371 588 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
f9ee2fcd
BW
589 hit |= grep_submodule(opt, repo, pathspec, entry.oid,
590 base->buf, base->buf + tn_len);
5010cb5f 591 }
74ed4371 592
e5e062b6
NTND
593 strbuf_setlen(base, old_baselen);
594
c8610a2e
JH
595 if (hit && opt->status_only)
596 break;
5010cb5f 597 }
74ed4371
BW
598
599 strbuf_release(&name);
5010cb5f
JH
600 return hit;
601}
602
f34bbc15 603static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
f9ee2fcd
BW
604 struct object *obj, const char *name, const char *path,
605 struct repository *repo)
5010cb5f 606{
1974632c 607 if (obj->type == OBJ_BLOB)
1db11086 608 return grep_oid(opt, &obj->oid, name, 0, path);
1974632c 609 if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
5010cb5f
JH
610 struct tree_desc tree;
611 void *data;
6fda5e51 612 unsigned long size;
e5e062b6
NTND
613 struct strbuf base;
614 int hit, len;
615
b3aeb285 616 grep_read_lock();
02f0547e 617 data = read_object_with_reference(&obj->oid, tree_type,
6fda5e51 618 &size, NULL);
b3aeb285 619 grep_read_unlock();
8cb5775b 620
5010cb5f 621 if (!data)
f2fd0760 622 die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
e5e062b6
NTND
623
624 len = name ? strlen(name) : 0;
625 strbuf_init(&base, PATH_MAX + len + 1);
626 if (len) {
627 strbuf_add(&base, name, len);
628 strbuf_addch(&base, ':');
629 }
6fda5e51 630 init_tree_desc(&tree, data, size);
55c61688 631 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
f9ee2fcd 632 obj->type == OBJ_COMMIT, repo);
e5e062b6 633 strbuf_release(&base);
5010cb5f
JH
634 free(data);
635 return hit;
636 }
debca9d2 637 die(_("unable to grep from object of type %s"), type_name(obj->type));
5010cb5f
JH
638}
639
f34bbc15 640static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
f9ee2fcd 641 struct repository *repo,
30d00c39
JN
642 const struct object_array *list)
643{
644 unsigned int i;
645 int hit = 0;
646 const unsigned int nr = list->nr;
647
648 for (i = 0; i < nr; i++) {
649 struct object *real_obj;
650 real_obj = deref_tag(list->objects[i].item, NULL, 0);
74ed4371
BW
651
652 /* load the gitmodules file for this rev */
653 if (recurse_submodules) {
654 submodule_free();
cd73de47 655 gitmodules_config_oid(&real_obj->oid);
74ed4371 656 }
f9ee2fcd
BW
657 if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path,
658 repo)) {
30d00c39
JN
659 hit = 1;
660 if (opt->status_only)
661 break;
662 }
663 }
664 return hit;
665}
666
dbfae86a 667static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
85975c0c 668 int exc_std, int use_index)
59332d13
JH
669{
670 struct dir_struct dir;
671 int i, hit = 0;
672
673 memset(&dir, 0, sizeof(dir));
85975c0c
JK
674 if (!use_index)
675 dir.flags |= DIR_NO_GITLINKS;
0a93fb8a
JH
676 if (exc_std)
677 setup_standard_excludes(&dir);
59332d13 678
0d32c183 679 fill_directory(&dir, &the_index, pathspec);
59332d13 680 for (i = 0; i < dir.nr; i++) {
ebb32893 681 if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
9d8b831b 682 continue;
59332d13
JH
683 hit |= grep_file(opt, dir.entries[i]->name);
684 if (hit && opt->status_only)
685 break;
686 }
59332d13
JH
687 return hit;
688}
689
ff3c7f9a
RS
690static int context_callback(const struct option *opt, const char *arg,
691 int unset)
3e230fa1
RS
692{
693 struct grep_opt *grep_opt = opt->value;
694 int value;
695 const char *endp;
696
697 if (unset) {
698 grep_opt->pre_context = grep_opt->post_context = 0;
699 return 0;
700 }
701 value = strtol(arg, (char **)&endp, 10);
702 if (*endp) {
2fc5f9f1 703 return error(_("switch `%c' expects a numerical value"),
3e230fa1
RS
704 opt->short_name);
705 }
706 grep_opt->pre_context = grep_opt->post_context = value;
707 return 0;
708}
709
ff3c7f9a 710static int file_callback(const struct option *opt, const char *arg, int unset)
3e230fa1
RS
711{
712 struct grep_opt *grep_opt = opt->value;
c41dd2fd 713 int from_stdin = !strcmp(arg, "-");
3e230fa1
RS
714 FILE *patterns;
715 int lno = 0;
cfe370c6 716 struct strbuf sb = STRBUF_INIT;
3e230fa1 717
c41dd2fd 718 patterns = from_stdin ? stdin : fopen(arg, "r");
3e230fa1 719 if (!patterns)
2fc5f9f1 720 die_errno(_("cannot open '%s'"), arg);
a5518431 721 while (strbuf_getline(&sb, patterns) == 0) {
3e230fa1
RS
722 /* ignore empty line like grep does */
723 if (sb.len == 0)
724 continue;
ed40a095 725
ec830611
RS
726 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
727 GREP_PATTERN);
3e230fa1 728 }
c41dd2fd
RS
729 if (!from_stdin)
730 fclose(patterns);
3e230fa1
RS
731 strbuf_release(&sb);
732 return 0;
733}
734
ff3c7f9a 735static int not_callback(const struct option *opt, const char *arg, int unset)
3e230fa1
RS
736{
737 struct grep_opt *grep_opt = opt->value;
738 append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
739 return 0;
740}
741
ff3c7f9a 742static int and_callback(const struct option *opt, const char *arg, int unset)
3e230fa1
RS
743{
744 struct grep_opt *grep_opt = opt->value;
745 append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
746 return 0;
747}
748
ff3c7f9a 749static int open_callback(const struct option *opt, const char *arg, int unset)
3e230fa1
RS
750{
751 struct grep_opt *grep_opt = opt->value;
752 append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
753 return 0;
754}
755
ff3c7f9a 756static int close_callback(const struct option *opt, const char *arg, int unset)
3e230fa1
RS
757{
758 struct grep_opt *grep_opt = opt->value;
759 append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
760 return 0;
761}
762
ff3c7f9a
RS
763static int pattern_callback(const struct option *opt, const char *arg,
764 int unset)
3e230fa1
RS
765{
766 struct grep_opt *grep_opt = opt->value;
767 append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
768 return 0;
769}
5010cb5f 770
a633fca0 771int cmd_grep(int argc, const char **argv, const char *prefix)
5010cb5f 772{
5010cb5f 773 int hit = 0;
0a93fb8a 774 int cached = 0, untracked = 0, opt_exclude = -1;
5acd64ed 775 int seen_dashdash = 0;
bbc09c22 776 int external_grep_allowed__ignored;
0af88c15 777 const char *show_in_pager = NULL, *default_pager = "dummy";
5010cb5f 778 struct grep_opt opt;
3cd47459 779 struct object_array list = OBJECT_ARRAY_INIT;
f34bbc15 780 struct pathspec pathspec;
183113a5 781 struct string_list path_list = STRING_LIST_INIT_NODUP;
5acd64ed 782 int i;
3e230fa1 783 int dummy;
ff38d1a9 784 int use_index = 1;
84befcd0 785 int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
131f3c96 786 int allow_revs;
cca2c172 787
3e230fa1 788 struct option options[] = {
d5d09d47 789 OPT_BOOL(0, "cached", &cached,
4b407bc5 790 N_("search in index instead of in the work tree")),
cbb08c2e 791 OPT_NEGBIT(0, "no-index", &use_index,
f63cf8c9 792 N_("find in contents not managed by git"), 1),
d5d09d47 793 OPT_BOOL(0, "untracked", &untracked,
4b407bc5 794 N_("search in both tracked and untracked files")),
0a93fb8a 795 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
77fdb8a8 796 N_("ignore files specified via '.gitignore'"), 1),
0281e487 797 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
4fb1c6aa 798 N_("recursively search in each submodule")),
3e230fa1 799 OPT_GROUP(""),
d5d09d47 800 OPT_BOOL('v', "invert-match", &opt.invert,
4b407bc5 801 N_("show non-matching lines")),
d5d09d47 802 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
4b407bc5 803 N_("case insensitive matching")),
d5d09d47 804 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
4b407bc5 805 N_("match patterns only at word boundaries")),
3e230fa1 806 OPT_SET_INT('a', "text", &opt.binary,
4b407bc5 807 N_("process binary files as text"), GREP_BINARY_TEXT),
3e230fa1 808 OPT_SET_INT('I', NULL, &opt.binary,
4b407bc5 809 N_("don't match patterns in binary files"),
3e230fa1 810 GREP_BINARY_NOMATCH),
335ec3bf
JK
811 OPT_BOOL(0, "textconv", &opt.allow_textconv,
812 N_("process binary files with textconv filters")),
4b407bc5
NTND
813 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
814 N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
a91f453f 815 NULL, 1 },
3e230fa1 816 OPT_GROUP(""),
84befcd0 817 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
4b407bc5 818 N_("use extended POSIX regular expressions"),
84befcd0
S
819 GREP_PATTERN_TYPE_ERE),
820 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
4b407bc5 821 N_("use basic POSIX regular expressions (default)"),
84befcd0
S
822 GREP_PATTERN_TYPE_BRE),
823 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
4b407bc5 824 N_("interpret patterns as fixed strings"),
84befcd0
S
825 GREP_PATTERN_TYPE_FIXED),
826 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
4b407bc5 827 N_("use Perl-compatible regular expressions"),
84befcd0 828 GREP_PATTERN_TYPE_PCRE),
3e230fa1 829 OPT_GROUP(""),
d5d09d47 830 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
4b407bc5
NTND
831 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
832 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
3e230fa1 833 OPT_NEGBIT(0, "full-name", &opt.relative,
4b407bc5 834 N_("show filenames relative to top directory"), 1),
d5d09d47 835 OPT_BOOL('l', "files-with-matches", &opt.name_only,
4b407bc5 836 N_("show only filenames instead of matching lines")),
d5d09d47 837 OPT_BOOL(0, "name-only", &opt.name_only,
4b407bc5 838 N_("synonym for --files-with-matches")),
d5d09d47 839 OPT_BOOL('L', "files-without-match",
3e230fa1 840 &opt.unmatch_name_only,
4b407bc5 841 N_("show only the names of files without match")),
d5d09d47 842 OPT_BOOL('z', "null", &opt.null_following_name,
4b407bc5 843 N_("print NUL after filenames")),
d5d09d47 844 OPT_BOOL('c', "count", &opt.count,
4b407bc5
NTND
845 N_("show the number of matches instead of matching lines")),
846 OPT__COLOR(&opt.color, N_("highlight matches")),
d5d09d47 847 OPT_BOOL(0, "break", &opt.file_break,
4b407bc5 848 N_("print empty line between matches from different files")),
d5d09d47 849 OPT_BOOL(0, "heading", &opt.heading,
4b407bc5 850 N_("show filename only once above matches from same file")),
3e230fa1 851 OPT_GROUP(""),
4b407bc5
NTND
852 OPT_CALLBACK('C', "context", &opt, N_("n"),
853 N_("show <n> context lines before and after matches"),
3e230fa1 854 context_callback),
317f63c2 855 OPT_INTEGER('B', "before-context", &opt.pre_context,
4b407bc5 856 N_("show <n> context lines before matches")),
317f63c2 857 OPT_INTEGER('A', "after-context", &opt.post_context,
4b407bc5 858 N_("show <n> context lines after matches")),
89f09dd3
VL
859 OPT_INTEGER(0, "threads", &num_threads,
860 N_("use <n> worker threads")),
4b407bc5 861 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
3e230fa1 862 context_callback),
d5d09d47 863 OPT_BOOL('p', "show-function", &opt.funcname,
4b407bc5 864 N_("show a line with the function name before matches")),
d5d09d47 865 OPT_BOOL('W', "function-context", &opt.funcbody,
4b407bc5 866 N_("show the surrounding function")),
3e230fa1 867 OPT_GROUP(""),
4b407bc5
NTND
868 OPT_CALLBACK('f', NULL, &opt, N_("file"),
869 N_("read patterns from file"), file_callback),
870 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
871 N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
3e230fa1 872 { OPTION_CALLBACK, 0, "and", &opt, NULL,
4b407bc5 873 N_("combine patterns specified with -e"),
3e230fa1 874 PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
d5d09d47 875 OPT_BOOL(0, "or", &dummy, ""),
3e230fa1
RS
876 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
877 PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
878 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
879 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
880 open_callback },
881 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
882 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
883 close_callback },
d52ee6e6 884 OPT__QUIET(&opt.status_only,
4b407bc5 885 N_("indicate hit with exit status without output")),
d5d09d47 886 OPT_BOOL(0, "all-match", &opt.all_match,
4b407bc5 887 N_("show only matches from files that match all patterns")),
17bf35a3 888 { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
3d7535e4 889 N_("show parse tree for grep expression"),
17bf35a3 890 PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
3e230fa1 891 OPT_GROUP(""),
0af88c15 892 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
4b407bc5 893 N_("pager"), N_("show matching files in the pager"),
0af88c15 894 PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
d5d09d47
SB
895 OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
896 N_("allow calling of grep(1) (ignored by this build)")),
3e230fa1
RS
897 OPT_END()
898 };
5010cb5f 899
15fabd1b
JH
900 init_grep_defaults();
901 git_config(grep_cmd_config, NULL);
902 grep_init(&opt, prefix);
7e8f59d5 903
5010cb5f 904 /*
5acd64ed
JH
905 * If there is no -- then the paths must exist in the working
906 * tree. If there is no explicit pattern specified with -e or
907 * -f, we take the first unrecognized non option to be the
908 * pattern, but then what follows it must be zero or more
909 * valid refs up to the -- (if exists), and then existing
910 * paths. If there is an explicit pattern, then the first
82e5a82f 911 * unrecognized non option is the beginning of the refs list
5acd64ed 912 * that continues up to the -- (if exists), and then paths.
5010cb5f 913 */
37782920 914 argc = parse_options(argc, argv, prefix, options, grep_usage,
3e230fa1 915 PARSE_OPT_KEEP_DASHDASH |
44415499 916 PARSE_OPT_STOP_AT_NON_OPTION);
c5c31d33 917 grep_commit_pattern_type(pattern_type_arg, &opt);
3e230fa1 918
ecd9ba61
TG
919 if (use_index && !startup_info->have_repository) {
920 int fallback = 0;
921 git_config_get_bool("grep.fallbacktonoindex", &fallback);
922 if (fallback)
923 use_index = 0;
924 else
925 /* die the same way as if we did it at the beginning */
926 setup_git_directory();
927 }
59332d13 928
1123c67c
JK
929 /*
930 * skip a -- separator; we know it cannot be
931 * separating revisions from pathnames if
932 * we haven't even had any patterns yet
933 */
934 if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
935 argv++;
936 argc--;
937 }
938
3e230fa1
RS
939 /* First unrecognized non-option token */
940 if (argc > 0 && !opt.pattern_list) {
941 append_grep_pattern(&opt, argv[0], "command line", 0,
942 GREP_PATTERN);
943 argv++;
944 argc--;
5010cb5f 945 }
5acd64ed 946
0af88c15
JS
947 if (show_in_pager == default_pager)
948 show_in_pager = git_pager(1);
678e484b 949 if (show_in_pager) {
e7b082a4 950 opt.color = 0;
0af88c15
JS
951 opt.name_only = 1;
952 opt.null_following_name = 1;
953 opt.output_priv = &path_list;
954 opt.output = append_path;
0c72cead 955 string_list_append(&path_list, show_in_pager);
678e484b
JS
956 }
957
f9b9faf6 958 if (!opt.pattern_list)
2fc5f9f1 959 die(_("no pattern given."));
5b594f45 960
b5b81136
JK
961 /*
962 * We have to find "--" in a separate pass, because its presence
963 * influences how we will parse arguments that come before it.
964 */
965 for (i = 0; i < argc; i++) {
966 if (!strcmp(argv[i], "--")) {
967 seen_dashdash = 1;
968 break;
969 }
970 }
971
972 /*
973 * Resolve any rev arguments. If we have a dashdash, then everything up
974 * to it must resolve as a rev. If not, then we stop at the first
975 * non-rev and assume everything else is a path.
976 */
131f3c96 977 allow_revs = use_index && !untracked;
3e230fa1 978 for (i = 0; i < argc; i++) {
5acd64ed 979 const char *arg = argv[i];
1db11086 980 struct object_id oid;
afa15f3c 981 struct object_context oc;
20d6421c
JK
982 struct object *object;
983
dca3b5f5
JT
984 if (!strcmp(arg, "--")) {
985 i++;
dca3b5f5
JT
986 break;
987 }
20d6421c 988
131f3c96 989 if (!allow_revs) {
d0ffc069 990 if (seen_dashdash)
131f3c96 991 die(_("--no-index or --untracked cannot be used with revs"));
d0ffc069
JK
992 break;
993 }
994
321c89bf 995 if (get_oid_with_context(arg, GET_OID_RECORD_PATH,
e82caf38 996 &oid, &oc)) {
b5b81136
JK
997 if (seen_dashdash)
998 die(_("unable to resolve revision: %s"), arg);
20d6421c 999 break;
b5b81136 1000 }
20d6421c 1001
c251c83d 1002 object = parse_object_or_die(&oid, arg);
20d6421c
JK
1003 if (!seen_dashdash)
1004 verify_non_filename(prefix, arg);
1005 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
dc944b65 1006 free(oc.path);
1362671f 1007 }
5acd64ed 1008
b5b81136
JK
1009 /*
1010 * Anything left over is presumed to be a path. But in the non-dashdash
1011 * "do what I mean" case, we verify and complain when that isn't true.
1012 */
a0fe2b0d
JK
1013 if (!seen_dashdash) {
1014 int j;
1015 for (j = i; j < argc; j++)
131f3c96 1016 verify_filename(prefix, argv[j], j == i && allow_revs);
a0fe2b0d
JK
1017 }
1018
1019 parse_pathspec(&pathspec, 0,
1020 PATHSPEC_PREFER_CWD |
1021 (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1022 prefix, argv + i);
1023 pathspec.max_depth = opt.max_depth;
1024 pathspec.recursive = 1;
eef3df5a 1025 pathspec.recurse_submodules = !!recurse_submodules;
a0fe2b0d 1026
53b8d931 1027#ifndef NO_PTHREADS
044b1f3c 1028 if (list.nr || cached || show_in_pager)
89f09dd3
VL
1029 num_threads = 0;
1030 else if (num_threads == 0)
1031 num_threads = GREP_NUM_THREADS_DEFAULT;
1032 else if (num_threads < 0)
1033 die(_("invalid number of threads specified (%d)"), num_threads);
9ec726a4
ÆAB
1034 if (num_threads == 1)
1035 num_threads = 0;
53b8d931 1036#else
d1edee4a
ÆAB
1037 if (num_threads)
1038 warning(_("no threads support, ignoring --threads"));
89f09dd3 1039 num_threads = 0;
53b8d931
TR
1040#endif
1041
6d423dd5
ÆAB
1042 if (!num_threads)
1043 /*
1044 * The compiled patterns on the main path are only
1045 * used when not using threading. Otherwise
1046 * start_threads() below calls compile_grep_patterns()
1047 * for each thread.
1048 */
1049 compile_grep_patterns(&opt);
1050
53b8d931 1051#ifndef NO_PTHREADS
89f09dd3 1052 if (num_threads) {
50dd0f2f
AY
1053 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1054 && (opt.pre_context || opt.post_context ||
1055 opt.file_break || opt.funcbody))
53b8d931
TR
1056 skip_first_line = 1;
1057 start_threads(&opt);
1058 }
1059#endif
1060
678e484b 1061 if (show_in_pager && (cached || list.nr))
e4fe4ba5 1062 die(_("--open-files-in-pager only works on the worktree"));
678e484b
JS
1063
1064 if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1065 const char *pager = path_list.items[0].string;
1066 int len = strlen(pager);
1067
1068 if (len > 4 && is_dir_sep(pager[len - 5]))
1069 pager += len - 4;
1070
f7febbea
JS
1071 if (opt.ignore_case && !strcmp("less", pager))
1072 string_list_append(&path_list, "-I");
1073
678e484b
JS
1074 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1075 struct strbuf buf = STRBUF_INIT;
1076 strbuf_addf(&buf, "+/%s%s",
1077 strcmp("less", pager) ? "" : "*",
1078 opt.pattern_list->pattern);
0c72cead 1079 string_list_append(&path_list, buf.buf);
678e484b
JS
1080 strbuf_detach(&buf, NULL);
1081 }
1082 }
1083
74ed4371 1084 if (recurse_submodules && (!use_index || untracked))
0281e487
BW
1085 die(_("option not supported with --recurse-submodules."));
1086
c2048f0b 1087 if (!show_in_pager && !opt.status_only)
678e484b
JS
1088 setup_pager();
1089
0a93fb8a 1090 if (!use_index && (untracked || cached))
dbfae86a 1091 die(_("--cached or --untracked cannot be used with --no-index."));
678e484b 1092
0a93fb8a 1093 if (!use_index || untracked) {
0a93fb8a 1094 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
85975c0c 1095 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
dbfae86a 1096 } else if (0 <= opt_exclude) {
9fddaf78 1097 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
685359cf 1098 } else if (!list.nr) {
6577f542
NTND
1099 if (!cached)
1100 setup_work_tree();
5b594f45 1101
f9ee2fcd 1102 hit = grep_cache(&opt, the_repository, &pathspec, cached);
685359cf
JS
1103 } else {
1104 if (cached)
2fc5f9f1 1105 die(_("both --cached and trees are given."));
f9ee2fcd
BW
1106
1107 hit = grep_objects(&opt, &pathspec, the_repository, &list);
5010cb5f 1108 }
5b594f45 1109
89f09dd3 1110 if (num_threads)
5b594f45 1111 hit |= wait_all();
678e484b
JS
1112 if (hit && show_in_pager)
1113 run_pager(&opt, prefix);
7861fa07 1114 clear_pathspec(&pathspec);
b48fb5b6 1115 free_grep_patterns(&opt);
5010cb5f
JH
1116 return !hit;
1117}