Git 2.16.2
[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
9dd5245c 95static void add_work(struct grep_opt *opt, enum grep_source_type type,
55c61688 96 const char *name, const char *path, const void *id)
5b594f45
FK
97{
98 grep_lock();
99
100 while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
101 pthread_cond_wait(&cond_write, &grep_mutex);
102 }
103
55c61688 104 grep_source_init(&todo[todo_end].source, type, name, path, id);
9dd5245c
JK
105 if (opt->binary != GREP_BINARY_TEXT)
106 grep_source_load_driver(&todo[todo_end].source);
5b594f45
FK
107 todo[todo_end].done = 0;
108 strbuf_reset(&todo[todo_end].out);
109 todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
110
111 pthread_cond_signal(&cond_add);
112 grep_unlock();
113}
114
115static struct work_item *get_work(void)
116{
117 struct work_item *ret;
118
119 grep_lock();
120 while (todo_start == todo_end && !all_work_added) {
121 pthread_cond_wait(&cond_add, &grep_mutex);
122 }
123
124 if (todo_start == todo_end && all_work_added) {
125 ret = NULL;
126 } else {
127 ret = &todo[todo_start];
128 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
129 }
130 grep_unlock();
131 return ret;
132}
133
5b594f45
FK
134static void work_done(struct work_item *w)
135{
136 int old_done;
137
138 grep_lock();
139 w->done = 1;
140 old_done = todo_done;
141 for(; todo[todo_done].done && todo_done != todo_start;
142 todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
143 w = &todo[todo_done];
431d6e7b 144 if (w->out.len) {
08303c36
RS
145 const char *p = w->out.buf;
146 size_t len = w->out.len;
147
148 /* Skip the leading hunk mark of the first file. */
149 if (skip_first_line) {
150 while (len) {
151 len--;
152 if (*p++ == '\n')
153 break;
154 }
155 skip_first_line = 0;
156 }
157
158 write_or_die(1, p, len);
431d6e7b 159 }
8f24a632 160 grep_source_clear(&w->source);
5b594f45
FK
161 }
162
163 if (old_done != todo_done)
164 pthread_cond_signal(&cond_write);
165
166 if (all_work_added && todo_done == todo_end)
167 pthread_cond_signal(&cond_result);
168
169 grep_unlock();
170}
171
172static void *run(void *arg)
173{
174 int hit = 0;
175 struct grep_opt *opt = arg;
176
177 while (1) {
178 struct work_item *w = get_work();
179 if (!w)
180 break;
181
182 opt->output_priv = w;
f9ee2fcd 183 hit |= grep_source(opt, &w->source);
8f24a632 184 grep_source_clear_data(&w->source);
5b594f45
FK
185 work_done(w);
186 }
bfac23d9
DM
187 free_grep_patterns(arg);
188 free(arg);
5b594f45
FK
189
190 return (void*) (intptr_t) hit;
191}
192
193static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
194{
195 struct work_item *w = opt->output_priv;
196 strbuf_add(&w->out, buf, size);
197}
198
199static void start_threads(struct grep_opt *opt)
200{
201 int i;
202
203 pthread_mutex_init(&grep_mutex, NULL);
b3aeb285 204 pthread_mutex_init(&grep_read_mutex, NULL);
0579f91d 205 pthread_mutex_init(&grep_attr_mutex, NULL);
5b594f45
FK
206 pthread_cond_init(&cond_add, NULL);
207 pthread_cond_init(&cond_write, NULL);
208 pthread_cond_init(&cond_result, NULL);
78db6ea9 209 grep_use_locks = 1;
5b594f45
FK
210
211 for (i = 0; i < ARRAY_SIZE(todo); i++) {
212 strbuf_init(&todo[i].out, 0);
213 }
214
89f09dd3
VL
215 threads = xcalloc(num_threads, sizeof(*threads));
216 for (i = 0; i < num_threads; i++) {
5b594f45
FK
217 int err;
218 struct grep_opt *o = grep_opt_dup(opt);
219 o->output = strbuf_out;
6d423dd5
ÆAB
220 if (i)
221 o->debug = 0;
5b594f45
FK
222 compile_grep_patterns(o);
223 err = pthread_create(&threads[i], NULL, run, o);
224
225 if (err)
2fc5f9f1 226 die(_("grep: failed to create thread: %s"),
5b594f45
FK
227 strerror(err));
228 }
229}
230
231static int wait_all(void)
232{
233 int hit = 0;
234 int i;
235
236 grep_lock();
237 all_work_added = 1;
238
239 /* Wait until all work is done. */
240 while (todo_done != todo_end)
241 pthread_cond_wait(&cond_result, &grep_mutex);
242
243 /* Wake up all the consumer threads so they can see that there
244 * is no more work to do.
245 */
246 pthread_cond_broadcast(&cond_add);
247 grep_unlock();
248
89f09dd3 249 for (i = 0; i < num_threads; i++) {
5b594f45
FK
250 void *h;
251 pthread_join(threads[i], &h);
252 hit |= (int) (intptr_t) h;
253 }
254
89f09dd3
VL
255 free(threads);
256
5b594f45 257 pthread_mutex_destroy(&grep_mutex);
b3aeb285 258 pthread_mutex_destroy(&grep_read_mutex);
0579f91d 259 pthread_mutex_destroy(&grep_attr_mutex);
5b594f45
FK
260 pthread_cond_destroy(&cond_add);
261 pthread_cond_destroy(&cond_write);
262 pthread_cond_destroy(&cond_result);
78db6ea9 263 grep_use_locks = 0;
5b594f45
FK
264
265 return hit;
266}
267#else /* !NO_PTHREADS */
5b594f45
FK
268
269static int wait_all(void)
270{
271 return 0;
272}
273#endif
274
15fabd1b
JH
275static int grep_cmd_config(const char *var, const char *value, void *cb)
276{
277 int st = grep_config(var, value, cb);
33c643bb 278 if (git_color_default_config(var, value, cb) < 0)
15fabd1b 279 st = -1;
89f09dd3
VL
280
281 if (!strcmp(var, "grep.threads")) {
282 num_threads = git_config_int(var, value);
283 if (num_threads < 0)
284 die(_("invalid number of threads specified (%d) for %s"),
285 num_threads, var);
d1edee4a
ÆAB
286#ifdef NO_PTHREADS
287 else if (num_threads && num_threads != 1) {
288 /*
289 * TRANSLATORS: %s is the configuration
290 * variable for tweaking threads, currently
291 * grep.threads
292 */
293 warning(_("no threads support, ignoring %s"), var);
294 num_threads = 0;
295 }
296#endif
89f09dd3
VL
297 }
298
9071c078
SB
299 if (!strcmp(var, "submodule.recurse"))
300 recurse_submodules = git_config_bool(var, value);
301
15fabd1b
JH
302 return st;
303}
304
1db11086 305static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
5f02d315
JH
306{
307 void *data;
308
b3aeb285 309 grep_read_lock();
1db11086 310 data = read_sha1_file(oid->hash, type, size);
b3aeb285 311 grep_read_unlock();
5b594f45
FK
312 return data;
313}
314
1db11086 315static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
55c61688
NTND
316 const char *filename, int tree_name_len,
317 const char *path)
5b594f45
FK
318{
319 struct strbuf pathbuf = STRBUF_INIT;
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
5b594f45 328#ifndef NO_PTHREADS
89f09dd3 329 if (num_threads) {
1c41c82b 330 add_work(opt, GREP_SOURCE_OID, pathbuf.buf, path, oid);
8f24a632 331 strbuf_release(&pathbuf);
5b594f45
FK
332 return 0;
333 } else
334#endif
335 {
8f24a632 336 struct grep_source gs;
5b594f45 337 int hit;
5010cb5f 338
1c41c82b 339 grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
8f24a632
JK
340 strbuf_release(&pathbuf);
341 hit = grep_source(opt, &gs);
dc49cd76 342
8f24a632
JK
343 grep_source_clear(&gs);
344 return hit;
5010cb5f 345 }
5b594f45
FK
346}
347
348static int grep_file(struct grep_opt *opt, const char *filename)
349{
350 struct strbuf buf = STRBUF_INIT;
5b594f45 351
f9ee2fcd
BW
352 if (opt->relative && opt->prefix_length)
353 quote_path_relative(filename, opt->prefix, &buf);
354 else
355 strbuf_addstr(&buf, filename);
5b594f45
FK
356
357#ifndef NO_PTHREADS
89f09dd3 358 if (num_threads) {
55c61688 359 add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
8f24a632 360 strbuf_release(&buf);
5b594f45
FK
361 return 0;
362 } else
363#endif
364 {
8f24a632 365 struct grep_source gs;
5b594f45 366 int hit;
5b594f45 367
55c61688 368 grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
8f24a632
JK
369 strbuf_release(&buf);
370 hit = grep_source(opt, &gs);
371
372 grep_source_clear(&gs);
5b594f45
FK
373 return hit;
374 }
5010cb5f
JH
375}
376
678e484b
JS
377static void append_path(struct grep_opt *opt, const void *data, size_t len)
378{
379 struct string_list *path_list = opt->output_priv;
380
381 if (len == 1 && *(const char *)data == '\0')
382 return;
0c72cead 383 string_list_append(path_list, xstrndup(data, len));
678e484b
JS
384}
385
386static void run_pager(struct grep_opt *opt, const char *prefix)
387{
388 struct string_list *path_list = opt->output_priv;
850d2fec 389 struct child_process child = CHILD_PROCESS_INIT;
678e484b
JS
390 int i, status;
391
392 for (i = 0; i < path_list->nr; i++)
850d2fec
JK
393 argv_array_push(&child.args, path_list->items[i].string);
394 child.dir = prefix;
395 child.use_shell = 1;
678e484b 396
850d2fec 397 status = run_command(&child);
678e484b
JS
398 if (status)
399 exit(status);
678e484b
JS
400}
401
f9ee2fcd
BW
402static int grep_cache(struct grep_opt *opt, struct repository *repo,
403 const struct pathspec *pathspec, int cached);
404static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
405 struct tree_desc *tree, struct strbuf *base, int tn_len,
406 int check_attr, struct repository *repo);
0281e487 407
f9ee2fcd
BW
408static int grep_submodule(struct grep_opt *opt, struct repository *superproject,
409 const struct pathspec *pathspec,
410 const struct object_id *oid,
411 const char *filename, const char *path)
0281e487 412{
f9ee2fcd
BW
413 struct repository submodule;
414 int hit;
74ed4371 415
f9ee2fcd
BW
416 if (!is_submodule_active(superproject, path))
417 return 0;
0281e487 418
f9ee2fcd
BW
419 if (repo_submodule_init(&submodule, superproject, path))
420 return 0;
be80a239 421
f9ee2fcd 422 repo_read_gitmodules(&submodule);
0281e487 423
74ed4371 424 /*
f9ee2fcd
BW
425 * NEEDSWORK: This adds the submodule's object directory to the list of
426 * alternates for the single in-memory object store. This has some bad
427 * consequences for memory (processed objects will never be freed) and
428 * performance (this increases the number of pack files git has to pay
429 * attention to, to the sum of the number of pack files in all the
430 * repositories processed so far). This can be removed once the object
431 * store is no longer global and instead is a member of the repository
432 * object.
74ed4371 433 */
9560e624 434 grep_read_lock();
f9ee2fcd 435 add_to_alternates_memory(submodule.objectdir);
9560e624 436 grep_read_unlock();
74ed4371 437
f9ee2fcd
BW
438 if (oid) {
439 struct object *object;
440 struct tree_desc tree;
441 void *data;
442 unsigned long size;
443 struct strbuf base = STRBUF_INIT;
74ed4371 444
f9ee2fcd 445 object = parse_object_or_die(oid, oid_to_hex(oid));
0281e487 446
f9ee2fcd
BW
447 grep_read_lock();
448 data = read_object_with_reference(object->oid.hash, tree_type,
449 &size, NULL);
450 grep_read_unlock();
0281e487 451
f9ee2fcd
BW
452 if (!data)
453 die(_("unable to read tree (%s)"), oid_to_hex(&object->oid));
0281e487 454
f9ee2fcd
BW
455 strbuf_addstr(&base, filename);
456 strbuf_addch(&base, '/');
0281e487 457
f9ee2fcd
BW
458 init_tree_desc(&tree, data, size);
459 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
460 object->type == OBJ_COMMIT, &submodule);
461 strbuf_release(&base);
462 free(data);
463 } else {
464 hit = grep_cache(opt, &submodule, pathspec, 1);
e6fac7f3 465 }
0281e487 466
f9ee2fcd
BW
467 repo_clear(&submodule);
468 return hit;
0281e487
BW
469}
470
f9ee2fcd
BW
471static int grep_cache(struct grep_opt *opt, struct repository *repo,
472 const struct pathspec *pathspec, int cached)
5010cb5f
JH
473{
474 int hit = 0;
475 int nr;
0281e487
BW
476 struct strbuf name = STRBUF_INIT;
477 int name_base_len = 0;
f9ee2fcd
BW
478 if (repo->submodule_prefix) {
479 name_base_len = strlen(repo->submodule_prefix);
480 strbuf_addstr(&name, repo->submodule_prefix);
0281e487
BW
481 }
482
f9ee2fcd 483 repo_read_index(repo);
5010cb5f 484
f9ee2fcd
BW
485 for (nr = 0; nr < repo->index->cache_nr; nr++) {
486 const struct cache_entry *ce = repo->index->cache[nr];
0281e487
BW
487 strbuf_setlen(&name, name_base_len);
488 strbuf_addstr(&name, ce->name);
489
490 if (S_ISREG(ce->ce_mode) &&
491 match_pathspec(pathspec, name.buf, name.len, 0, NULL,
492 S_ISDIR(ce->ce_mode) ||
493 S_ISGITLINK(ce->ce_mode))) {
494 /*
495 * If CE_VALID is on, we assume worktree file and its
496 * cache entry are identical, even if worktree file has
497 * been modified, so use cache version instead
498 */
499 if (cached || (ce->ce_flags & CE_VALID) ||
500 ce_skip_worktree(ce)) {
501 if (ce_stage(ce) || ce_intent_to_add(ce))
502 continue;
f9ee2fcd
BW
503 hit |= grep_oid(opt, &ce->oid, name.buf,
504 0, name.buf);
0281e487 505 } else {
f9ee2fcd 506 hit |= grep_file(opt, name.buf);
0281e487
BW
507 }
508 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
509 submodule_path_match(pathspec, name.buf, NULL)) {
f9ee2fcd 510 hit |= grep_submodule(opt, repo, pathspec, NULL, ce->name, ce->name);
0281e487 511 } else {
5010cb5f 512 continue;
36f2587f 513 }
0281e487 514
36f2587f
JH
515 if (ce_stage(ce)) {
516 do {
517 nr++;
f9ee2fcd
BW
518 } while (nr < repo->index->cache_nr &&
519 !strcmp(ce->name, repo->index->cache[nr]->name));
36f2587f
JH
520 nr--; /* compensate for loop control */
521 }
c8610a2e
JH
522 if (hit && opt->status_only)
523 break;
5010cb5f 524 }
0281e487
BW
525
526 strbuf_release(&name);
5010cb5f
JH
527 return hit;
528}
529
f34bbc15 530static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
55c61688 531 struct tree_desc *tree, struct strbuf *base, int tn_len,
f9ee2fcd 532 int check_attr, struct repository *repo)
5010cb5f 533{
d688cf07
NTND
534 int hit = 0;
535 enum interesting match = entry_not_interesting;
4c068a98 536 struct name_entry entry;
e5e062b6 537 int old_baselen = base->len;
74ed4371
BW
538 struct strbuf name = STRBUF_INIT;
539 int name_base_len = 0;
f9ee2fcd
BW
540 if (repo->submodule_prefix) {
541 strbuf_addstr(&name, repo->submodule_prefix);
74ed4371
BW
542 name_base_len = name.len;
543 }
5010cb5f 544
4c068a98 545 while (tree_entry(tree, &entry)) {
0de16337 546 int te_len = tree_entry_len(&entry);
e5e062b6 547
d688cf07 548 if (match != all_entries_interesting) {
74ed4371
BW
549 strbuf_addstr(&name, base->buf + tn_len);
550 match = tree_entry_interesting(&entry, &name,
551 0, pathspec);
552 strbuf_setlen(&name, name_base_len);
553
d688cf07 554 if (match == all_entries_not_interesting)
97d0b74a 555 break;
d688cf07 556 if (match == entry_not_interesting)
1376e507
NTND
557 continue;
558 }
5010cb5f 559
1376e507 560 strbuf_add(base, entry.path, te_len);
e0eb889f 561
1376e507 562 if (S_ISREG(entry.mode)) {
1db11086 563 hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
55c61688 564 check_attr ? base->buf + tn_len : NULL);
74ed4371 565 } else if (S_ISDIR(entry.mode)) {
21666f1a 566 enum object_type type;
5010cb5f
JH
567 struct tree_desc sub;
568 void *data;
6fda5e51
LT
569 unsigned long size;
570
1db11086 571 data = lock_and_read_oid_file(entry.oid, &type, &size);
5010cb5f 572 if (!data)
2fc5f9f1 573 die(_("unable to read tree (%s)"),
7d924c91 574 oid_to_hex(entry.oid));
1376e507
NTND
575
576 strbuf_addch(base, '/');
6fda5e51 577 init_tree_desc(&sub, data, size);
55c61688 578 hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
f9ee2fcd 579 check_attr, repo);
5010cb5f 580 free(data);
74ed4371 581 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
f9ee2fcd
BW
582 hit |= grep_submodule(opt, repo, pathspec, entry.oid,
583 base->buf, base->buf + tn_len);
5010cb5f 584 }
74ed4371 585
e5e062b6
NTND
586 strbuf_setlen(base, old_baselen);
587
c8610a2e
JH
588 if (hit && opt->status_only)
589 break;
5010cb5f 590 }
74ed4371
BW
591
592 strbuf_release(&name);
5010cb5f
JH
593 return hit;
594}
595
f34bbc15 596static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
f9ee2fcd
BW
597 struct object *obj, const char *name, const char *path,
598 struct repository *repo)
5010cb5f 599{
1974632c 600 if (obj->type == OBJ_BLOB)
1db11086 601 return grep_oid(opt, &obj->oid, name, 0, path);
1974632c 602 if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
5010cb5f
JH
603 struct tree_desc tree;
604 void *data;
6fda5e51 605 unsigned long size;
e5e062b6
NTND
606 struct strbuf base;
607 int hit, len;
608
b3aeb285 609 grep_read_lock();
f2fd0760 610 data = read_object_with_reference(obj->oid.hash, tree_type,
6fda5e51 611 &size, NULL);
b3aeb285 612 grep_read_unlock();
8cb5775b 613
5010cb5f 614 if (!data)
f2fd0760 615 die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
e5e062b6
NTND
616
617 len = name ? strlen(name) : 0;
618 strbuf_init(&base, PATH_MAX + len + 1);
619 if (len) {
620 strbuf_add(&base, name, len);
621 strbuf_addch(&base, ':');
622 }
6fda5e51 623 init_tree_desc(&tree, data, size);
55c61688 624 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
f9ee2fcd 625 obj->type == OBJ_COMMIT, repo);
e5e062b6 626 strbuf_release(&base);
5010cb5f
JH
627 free(data);
628 return hit;
629 }
2fc5f9f1 630 die(_("unable to grep from object of type %s"), typename(obj->type));
5010cb5f
JH
631}
632
f34bbc15 633static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
f9ee2fcd 634 struct repository *repo,
30d00c39
JN
635 const struct object_array *list)
636{
637 unsigned int i;
638 int hit = 0;
639 const unsigned int nr = list->nr;
640
641 for (i = 0; i < nr; i++) {
642 struct object *real_obj;
643 real_obj = deref_tag(list->objects[i].item, NULL, 0);
74ed4371
BW
644
645 /* load the gitmodules file for this rev */
646 if (recurse_submodules) {
647 submodule_free();
cd73de47 648 gitmodules_config_oid(&real_obj->oid);
74ed4371 649 }
f9ee2fcd
BW
650 if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path,
651 repo)) {
30d00c39
JN
652 hit = 1;
653 if (opt->status_only)
654 break;
655 }
656 }
657 return hit;
658}
659
dbfae86a 660static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
85975c0c 661 int exc_std, int use_index)
59332d13
JH
662{
663 struct dir_struct dir;
664 int i, hit = 0;
665
666 memset(&dir, 0, sizeof(dir));
85975c0c
JK
667 if (!use_index)
668 dir.flags |= DIR_NO_GITLINKS;
0a93fb8a
JH
669 if (exc_std)
670 setup_standard_excludes(&dir);
59332d13 671
0d32c183 672 fill_directory(&dir, &the_index, pathspec);
59332d13 673 for (i = 0; i < dir.nr; i++) {
ebb32893 674 if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
9d8b831b 675 continue;
59332d13
JH
676 hit |= grep_file(opt, dir.entries[i]->name);
677 if (hit && opt->status_only)
678 break;
679 }
59332d13
JH
680 return hit;
681}
682
ff3c7f9a
RS
683static int context_callback(const struct option *opt, const char *arg,
684 int unset)
3e230fa1
RS
685{
686 struct grep_opt *grep_opt = opt->value;
687 int value;
688 const char *endp;
689
690 if (unset) {
691 grep_opt->pre_context = grep_opt->post_context = 0;
692 return 0;
693 }
694 value = strtol(arg, (char **)&endp, 10);
695 if (*endp) {
2fc5f9f1 696 return error(_("switch `%c' expects a numerical value"),
3e230fa1
RS
697 opt->short_name);
698 }
699 grep_opt->pre_context = grep_opt->post_context = value;
700 return 0;
701}
702
ff3c7f9a 703static int file_callback(const struct option *opt, const char *arg, int unset)
3e230fa1
RS
704{
705 struct grep_opt *grep_opt = opt->value;
c41dd2fd 706 int from_stdin = !strcmp(arg, "-");
3e230fa1
RS
707 FILE *patterns;
708 int lno = 0;
cfe370c6 709 struct strbuf sb = STRBUF_INIT;
3e230fa1 710
c41dd2fd 711 patterns = from_stdin ? stdin : fopen(arg, "r");
3e230fa1 712 if (!patterns)
2fc5f9f1 713 die_errno(_("cannot open '%s'"), arg);
a5518431 714 while (strbuf_getline(&sb, patterns) == 0) {
3e230fa1
RS
715 /* ignore empty line like grep does */
716 if (sb.len == 0)
717 continue;
ed40a095 718
ec830611
RS
719 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
720 GREP_PATTERN);
3e230fa1 721 }
c41dd2fd
RS
722 if (!from_stdin)
723 fclose(patterns);
3e230fa1
RS
724 strbuf_release(&sb);
725 return 0;
726}
727
ff3c7f9a 728static int not_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, "--not", "command line", 0, GREP_NOT);
732 return 0;
733}
734
ff3c7f9a 735static int and_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, "--and", "command line", 0, GREP_AND);
739 return 0;
740}
741
ff3c7f9a 742static int open_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, "(", "command line", 0, GREP_OPEN_PAREN);
746 return 0;
747}
748
ff3c7f9a 749static int close_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_CLOSE_PAREN);
753 return 0;
754}
755
ff3c7f9a
RS
756static int pattern_callback(const struct option *opt, const char *arg,
757 int unset)
3e230fa1
RS
758{
759 struct grep_opt *grep_opt = opt->value;
760 append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
761 return 0;
762}
5010cb5f 763
a633fca0 764int cmd_grep(int argc, const char **argv, const char *prefix)
5010cb5f 765{
5010cb5f 766 int hit = 0;
0a93fb8a 767 int cached = 0, untracked = 0, opt_exclude = -1;
5acd64ed 768 int seen_dashdash = 0;
bbc09c22 769 int external_grep_allowed__ignored;
0af88c15 770 const char *show_in_pager = NULL, *default_pager = "dummy";
5010cb5f 771 struct grep_opt opt;
3cd47459 772 struct object_array list = OBJECT_ARRAY_INIT;
f34bbc15 773 struct pathspec pathspec;
183113a5 774 struct string_list path_list = STRING_LIST_INIT_NODUP;
5acd64ed 775 int i;
3e230fa1 776 int dummy;
ff38d1a9 777 int use_index = 1;
84befcd0 778 int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
131f3c96 779 int allow_revs;
cca2c172 780
3e230fa1 781 struct option options[] = {
d5d09d47 782 OPT_BOOL(0, "cached", &cached,
4b407bc5 783 N_("search in index instead of in the work tree")),
cbb08c2e 784 OPT_NEGBIT(0, "no-index", &use_index,
f63cf8c9 785 N_("find in contents not managed by git"), 1),
d5d09d47 786 OPT_BOOL(0, "untracked", &untracked,
4b407bc5 787 N_("search in both tracked and untracked files")),
0a93fb8a 788 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
77fdb8a8 789 N_("ignore files specified via '.gitignore'"), 1),
0281e487 790 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
4fb1c6aa 791 N_("recursively search in each submodule")),
3e230fa1 792 OPT_GROUP(""),
d5d09d47 793 OPT_BOOL('v', "invert-match", &opt.invert,
4b407bc5 794 N_("show non-matching lines")),
d5d09d47 795 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
4b407bc5 796 N_("case insensitive matching")),
d5d09d47 797 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
4b407bc5 798 N_("match patterns only at word boundaries")),
3e230fa1 799 OPT_SET_INT('a', "text", &opt.binary,
4b407bc5 800 N_("process binary files as text"), GREP_BINARY_TEXT),
3e230fa1 801 OPT_SET_INT('I', NULL, &opt.binary,
4b407bc5 802 N_("don't match patterns in binary files"),
3e230fa1 803 GREP_BINARY_NOMATCH),
335ec3bf
JK
804 OPT_BOOL(0, "textconv", &opt.allow_textconv,
805 N_("process binary files with textconv filters")),
4b407bc5
NTND
806 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
807 N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
a91f453f 808 NULL, 1 },
3e230fa1 809 OPT_GROUP(""),
84befcd0 810 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
4b407bc5 811 N_("use extended POSIX regular expressions"),
84befcd0
S
812 GREP_PATTERN_TYPE_ERE),
813 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
4b407bc5 814 N_("use basic POSIX regular expressions (default)"),
84befcd0
S
815 GREP_PATTERN_TYPE_BRE),
816 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
4b407bc5 817 N_("interpret patterns as fixed strings"),
84befcd0
S
818 GREP_PATTERN_TYPE_FIXED),
819 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
4b407bc5 820 N_("use Perl-compatible regular expressions"),
84befcd0 821 GREP_PATTERN_TYPE_PCRE),
3e230fa1 822 OPT_GROUP(""),
d5d09d47 823 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
4b407bc5
NTND
824 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
825 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
3e230fa1 826 OPT_NEGBIT(0, "full-name", &opt.relative,
4b407bc5 827 N_("show filenames relative to top directory"), 1),
d5d09d47 828 OPT_BOOL('l', "files-with-matches", &opt.name_only,
4b407bc5 829 N_("show only filenames instead of matching lines")),
d5d09d47 830 OPT_BOOL(0, "name-only", &opt.name_only,
4b407bc5 831 N_("synonym for --files-with-matches")),
d5d09d47 832 OPT_BOOL('L', "files-without-match",
3e230fa1 833 &opt.unmatch_name_only,
4b407bc5 834 N_("show only the names of files without match")),
d5d09d47 835 OPT_BOOL('z', "null", &opt.null_following_name,
4b407bc5 836 N_("print NUL after filenames")),
d5d09d47 837 OPT_BOOL('c', "count", &opt.count,
4b407bc5
NTND
838 N_("show the number of matches instead of matching lines")),
839 OPT__COLOR(&opt.color, N_("highlight matches")),
d5d09d47 840 OPT_BOOL(0, "break", &opt.file_break,
4b407bc5 841 N_("print empty line between matches from different files")),
d5d09d47 842 OPT_BOOL(0, "heading", &opt.heading,
4b407bc5 843 N_("show filename only once above matches from same file")),
3e230fa1 844 OPT_GROUP(""),
4b407bc5
NTND
845 OPT_CALLBACK('C', "context", &opt, N_("n"),
846 N_("show <n> context lines before and after matches"),
3e230fa1 847 context_callback),
317f63c2 848 OPT_INTEGER('B', "before-context", &opt.pre_context,
4b407bc5 849 N_("show <n> context lines before matches")),
317f63c2 850 OPT_INTEGER('A', "after-context", &opt.post_context,
4b407bc5 851 N_("show <n> context lines after matches")),
89f09dd3
VL
852 OPT_INTEGER(0, "threads", &num_threads,
853 N_("use <n> worker threads")),
4b407bc5 854 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
3e230fa1 855 context_callback),
d5d09d47 856 OPT_BOOL('p', "show-function", &opt.funcname,
4b407bc5 857 N_("show a line with the function name before matches")),
d5d09d47 858 OPT_BOOL('W', "function-context", &opt.funcbody,
4b407bc5 859 N_("show the surrounding function")),
3e230fa1 860 OPT_GROUP(""),
4b407bc5
NTND
861 OPT_CALLBACK('f', NULL, &opt, N_("file"),
862 N_("read patterns from file"), file_callback),
863 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
864 N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
3e230fa1 865 { OPTION_CALLBACK, 0, "and", &opt, NULL,
4b407bc5 866 N_("combine patterns specified with -e"),
3e230fa1 867 PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
d5d09d47 868 OPT_BOOL(0, "or", &dummy, ""),
3e230fa1
RS
869 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
870 PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
871 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
872 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
873 open_callback },
874 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
875 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
876 close_callback },
d52ee6e6 877 OPT__QUIET(&opt.status_only,
4b407bc5 878 N_("indicate hit with exit status without output")),
d5d09d47 879 OPT_BOOL(0, "all-match", &opt.all_match,
4b407bc5 880 N_("show only matches from files that match all patterns")),
17bf35a3 881 { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
3d7535e4 882 N_("show parse tree for grep expression"),
17bf35a3 883 PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
3e230fa1 884 OPT_GROUP(""),
0af88c15 885 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
4b407bc5 886 N_("pager"), N_("show matching files in the pager"),
0af88c15 887 PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
d5d09d47
SB
888 OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
889 N_("allow calling of grep(1) (ignored by this build)")),
3e230fa1
RS
890 OPT_END()
891 };
5010cb5f 892
15fabd1b
JH
893 init_grep_defaults();
894 git_config(grep_cmd_config, NULL);
895 grep_init(&opt, prefix);
7e8f59d5 896
5010cb5f 897 /*
5acd64ed
JH
898 * If there is no -- then the paths must exist in the working
899 * tree. If there is no explicit pattern specified with -e or
900 * -f, we take the first unrecognized non option to be the
901 * pattern, but then what follows it must be zero or more
902 * valid refs up to the -- (if exists), and then existing
903 * paths. If there is an explicit pattern, then the first
82e5a82f 904 * unrecognized non option is the beginning of the refs list
5acd64ed 905 * that continues up to the -- (if exists), and then paths.
5010cb5f 906 */
37782920 907 argc = parse_options(argc, argv, prefix, options, grep_usage,
3e230fa1 908 PARSE_OPT_KEEP_DASHDASH |
44415499 909 PARSE_OPT_STOP_AT_NON_OPTION);
c5c31d33 910 grep_commit_pattern_type(pattern_type_arg, &opt);
3e230fa1 911
ecd9ba61
TG
912 if (use_index && !startup_info->have_repository) {
913 int fallback = 0;
914 git_config_get_bool("grep.fallbacktonoindex", &fallback);
915 if (fallback)
916 use_index = 0;
917 else
918 /* die the same way as if we did it at the beginning */
919 setup_git_directory();
920 }
59332d13 921
1123c67c
JK
922 /*
923 * skip a -- separator; we know it cannot be
924 * separating revisions from pathnames if
925 * we haven't even had any patterns yet
926 */
927 if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
928 argv++;
929 argc--;
930 }
931
3e230fa1
RS
932 /* First unrecognized non-option token */
933 if (argc > 0 && !opt.pattern_list) {
934 append_grep_pattern(&opt, argv[0], "command line", 0,
935 GREP_PATTERN);
936 argv++;
937 argc--;
5010cb5f 938 }
5acd64ed 939
0af88c15
JS
940 if (show_in_pager == default_pager)
941 show_in_pager = git_pager(1);
678e484b 942 if (show_in_pager) {
e7b082a4 943 opt.color = 0;
0af88c15
JS
944 opt.name_only = 1;
945 opt.null_following_name = 1;
946 opt.output_priv = &path_list;
947 opt.output = append_path;
0c72cead 948 string_list_append(&path_list, show_in_pager);
678e484b
JS
949 }
950
f9b9faf6 951 if (!opt.pattern_list)
2fc5f9f1 952 die(_("no pattern given."));
5b594f45 953
b5b81136
JK
954 /*
955 * We have to find "--" in a separate pass, because its presence
956 * influences how we will parse arguments that come before it.
957 */
958 for (i = 0; i < argc; i++) {
959 if (!strcmp(argv[i], "--")) {
960 seen_dashdash = 1;
961 break;
962 }
963 }
964
965 /*
966 * Resolve any rev arguments. If we have a dashdash, then everything up
967 * to it must resolve as a rev. If not, then we stop at the first
968 * non-rev and assume everything else is a path.
969 */
131f3c96 970 allow_revs = use_index && !untracked;
3e230fa1 971 for (i = 0; i < argc; i++) {
5acd64ed 972 const char *arg = argv[i];
1db11086 973 struct object_id oid;
afa15f3c 974 struct object_context oc;
20d6421c
JK
975 struct object *object;
976
dca3b5f5
JT
977 if (!strcmp(arg, "--")) {
978 i++;
dca3b5f5
JT
979 break;
980 }
20d6421c 981
131f3c96 982 if (!allow_revs) {
d0ffc069 983 if (seen_dashdash)
131f3c96 984 die(_("--no-index or --untracked cannot be used with revs"));
d0ffc069
JK
985 break;
986 }
987
321c89bf 988 if (get_oid_with_context(arg, GET_OID_RECORD_PATH,
e82caf38 989 &oid, &oc)) {
b5b81136
JK
990 if (seen_dashdash)
991 die(_("unable to resolve revision: %s"), arg);
20d6421c 992 break;
b5b81136 993 }
20d6421c 994
c251c83d 995 object = parse_object_or_die(&oid, arg);
20d6421c
JK
996 if (!seen_dashdash)
997 verify_non_filename(prefix, arg);
998 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
dc944b65 999 free(oc.path);
1362671f 1000 }
5acd64ed 1001
b5b81136
JK
1002 /*
1003 * Anything left over is presumed to be a path. But in the non-dashdash
1004 * "do what I mean" case, we verify and complain when that isn't true.
1005 */
a0fe2b0d
JK
1006 if (!seen_dashdash) {
1007 int j;
1008 for (j = i; j < argc; j++)
131f3c96 1009 verify_filename(prefix, argv[j], j == i && allow_revs);
a0fe2b0d
JK
1010 }
1011
1012 parse_pathspec(&pathspec, 0,
1013 PATHSPEC_PREFER_CWD |
1014 (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1015 prefix, argv + i);
1016 pathspec.max_depth = opt.max_depth;
1017 pathspec.recursive = 1;
eef3df5a 1018 pathspec.recurse_submodules = !!recurse_submodules;
a0fe2b0d 1019
53b8d931 1020#ifndef NO_PTHREADS
044b1f3c 1021 if (list.nr || cached || show_in_pager)
89f09dd3
VL
1022 num_threads = 0;
1023 else if (num_threads == 0)
1024 num_threads = GREP_NUM_THREADS_DEFAULT;
1025 else if (num_threads < 0)
1026 die(_("invalid number of threads specified (%d)"), num_threads);
9ec726a4
ÆAB
1027 if (num_threads == 1)
1028 num_threads = 0;
53b8d931 1029#else
d1edee4a
ÆAB
1030 if (num_threads)
1031 warning(_("no threads support, ignoring --threads"));
89f09dd3 1032 num_threads = 0;
53b8d931
TR
1033#endif
1034
6d423dd5
ÆAB
1035 if (!num_threads)
1036 /*
1037 * The compiled patterns on the main path are only
1038 * used when not using threading. Otherwise
1039 * start_threads() below calls compile_grep_patterns()
1040 * for each thread.
1041 */
1042 compile_grep_patterns(&opt);
1043
53b8d931 1044#ifndef NO_PTHREADS
89f09dd3 1045 if (num_threads) {
50dd0f2f
AY
1046 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1047 && (opt.pre_context || opt.post_context ||
1048 opt.file_break || opt.funcbody))
53b8d931
TR
1049 skip_first_line = 1;
1050 start_threads(&opt);
1051 }
1052#endif
1053
678e484b 1054 if (show_in_pager && (cached || list.nr))
e4fe4ba5 1055 die(_("--open-files-in-pager only works on the worktree"));
678e484b
JS
1056
1057 if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1058 const char *pager = path_list.items[0].string;
1059 int len = strlen(pager);
1060
1061 if (len > 4 && is_dir_sep(pager[len - 5]))
1062 pager += len - 4;
1063
f7febbea
JS
1064 if (opt.ignore_case && !strcmp("less", pager))
1065 string_list_append(&path_list, "-I");
1066
678e484b
JS
1067 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1068 struct strbuf buf = STRBUF_INIT;
1069 strbuf_addf(&buf, "+/%s%s",
1070 strcmp("less", pager) ? "" : "*",
1071 opt.pattern_list->pattern);
0c72cead 1072 string_list_append(&path_list, buf.buf);
678e484b
JS
1073 strbuf_detach(&buf, NULL);
1074 }
1075 }
1076
74ed4371 1077 if (recurse_submodules && (!use_index || untracked))
0281e487
BW
1078 die(_("option not supported with --recurse-submodules."));
1079
c2048f0b 1080 if (!show_in_pager && !opt.status_only)
678e484b
JS
1081 setup_pager();
1082
0a93fb8a 1083 if (!use_index && (untracked || cached))
dbfae86a 1084 die(_("--cached or --untracked cannot be used with --no-index."));
678e484b 1085
0a93fb8a 1086 if (!use_index || untracked) {
0a93fb8a 1087 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
85975c0c 1088 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
dbfae86a 1089 } else if (0 <= opt_exclude) {
9fddaf78 1090 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
685359cf 1091 } else if (!list.nr) {
6577f542
NTND
1092 if (!cached)
1093 setup_work_tree();
5b594f45 1094
f9ee2fcd 1095 hit = grep_cache(&opt, the_repository, &pathspec, cached);
685359cf
JS
1096 } else {
1097 if (cached)
2fc5f9f1 1098 die(_("both --cached and trees are given."));
f9ee2fcd
BW
1099
1100 hit = grep_objects(&opt, &pathspec, the_repository, &list);
5010cb5f 1101 }
5b594f45 1102
89f09dd3 1103 if (num_threads)
5b594f45 1104 hit |= wait_all();
678e484b
JS
1105 if (hit && show_in_pager)
1106 run_pager(&opt, prefix);
7861fa07 1107 clear_pathspec(&pathspec);
b48fb5b6 1108 free_grep_patterns(&opt);
5010cb5f
JH
1109 return !hit;
1110}