git

1
/*
2
 * Builtin "git grep"
3
 *
4
 * Copyright (c) 2006 Junio C Hamano
5
 */
6
#include "builtin.h"
7
#include "abspath.h"
8
#include "gettext.h"
9
#include "hex.h"
10
#include "repository.h"
11
#include "config.h"
12
#include "tag.h"
13
#include "tree-walk.h"
14
#include "parse-options.h"
15
#include "string-list.h"
16
#include "run-command.h"
17
#include "grep.h"
18
#include "quote.h"
19
#include "dir.h"
20
#include "pathspec.h"
21
#include "setup.h"
22
#include "submodule.h"
23
#include "submodule-config.h"
24
#include "object-file.h"
25
#include "object-name.h"
26
#include "object-store-ll.h"
27
#include "packfile.h"
28
#include "pager.h"
29
#include "path.h"
30
#include "read-cache-ll.h"
31
#include "write-or-die.h"
32

33
static const char *grep_prefix;
34

35
static char const * const grep_usage[] = {
36
	N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
37
	NULL
38
};
39

40
static int recurse_submodules;
41

42
static int num_threads;
43

44
static pthread_t *threads;
45

46
/* We use one producer thread and THREADS consumer
47
 * threads. The producer adds struct work_items to 'todo' and the
48
 * consumers pick work items from the same array.
49
 */
50
struct work_item {
51
	struct grep_source source;
52
	char done;
53
	struct strbuf out;
54
};
55

56
/* In the range [todo_done, todo_start) in 'todo' we have work_items
57
 * that have been or are processed by a consumer thread. We haven't
58
 * written the result for these to stdout yet.
59
 *
60
 * The work_items in [todo_start, todo_end) are waiting to be picked
61
 * up by a consumer thread.
62
 *
63
 * The ranges are modulo TODO_SIZE.
64
 */
65
#define TODO_SIZE 128
66
static struct work_item todo[TODO_SIZE];
67
static int todo_start;
68
static int todo_end;
69
static int todo_done;
70

71
/* Has all work items been added? */
72
static int all_work_added;
73

74
static struct repository **repos_to_free;
75
static size_t repos_to_free_nr, repos_to_free_alloc;
76

77
/* This lock protects all the variables above. */
78
static pthread_mutex_t grep_mutex;
79

80
static inline void grep_lock(void)
81
{
82
	pthread_mutex_lock(&grep_mutex);
83
}
84

85
static inline void grep_unlock(void)
86
{
87
	pthread_mutex_unlock(&grep_mutex);
88
}
89

90
/* Signalled when a new work_item is added to todo. */
91
static pthread_cond_t cond_add;
92

93
/* Signalled when the result from one work_item is written to
94
 * stdout.
95
 */
96
static pthread_cond_t cond_write;
97

98
/* Signalled when we are finished with everything. */
99
static pthread_cond_t cond_result;
100

101
static int skip_first_line;
102

103
static void add_work(struct grep_opt *opt, struct grep_source *gs)
104
{
105
	if (opt->binary != GREP_BINARY_TEXT)
106
		grep_source_load_driver(gs, opt->repo->index);
107

108
	grep_lock();
109

110
	while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
111
		pthread_cond_wait(&cond_write, &grep_mutex);
112
	}
113

114
	todo[todo_end].source = *gs;
115
	todo[todo_end].done = 0;
116
	strbuf_reset(&todo[todo_end].out);
117
	todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
118

119
	pthread_cond_signal(&cond_add);
120
	grep_unlock();
121
}
122

123
static struct work_item *get_work(void)
124
{
125
	struct work_item *ret;
126

127
	grep_lock();
128
	while (todo_start == todo_end && !all_work_added) {
129
		pthread_cond_wait(&cond_add, &grep_mutex);
130
	}
131

132
	if (todo_start == todo_end && all_work_added) {
133
		ret = NULL;
134
	} else {
135
		ret = &todo[todo_start];
136
		todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
137
	}
138
	grep_unlock();
139
	return ret;
140
}
141

142
static void work_done(struct work_item *w)
143
{
144
	int old_done;
145

146
	grep_lock();
147
	w->done = 1;
148
	old_done = todo_done;
149
	for(; todo[todo_done].done && todo_done != todo_start;
150
	    todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
151
		w = &todo[todo_done];
152
		if (w->out.len) {
153
			const char *p = w->out.buf;
154
			size_t len = w->out.len;
155

156
			/* Skip the leading hunk mark of the first file. */
157
			if (skip_first_line) {
158
				while (len) {
159
					len--;
160
					if (*p++ == '\n')
161
						break;
162
				}
163
				skip_first_line = 0;
164
			}
165

166
			write_or_die(1, p, len);
167
		}
168
		grep_source_clear(&w->source);
169
	}
170

171
	if (old_done != todo_done)
172
		pthread_cond_signal(&cond_write);
173

174
	if (all_work_added && todo_done == todo_end)
175
		pthread_cond_signal(&cond_result);
176

177
	grep_unlock();
178
}
179

180
static void free_repos(void)
181
{
182
	int i;
183

184
	for (i = 0; i < repos_to_free_nr; i++) {
185
		repo_clear(repos_to_free[i]);
186
		free(repos_to_free[i]);
187
	}
188
	FREE_AND_NULL(repos_to_free);
189
	repos_to_free_nr = 0;
190
	repos_to_free_alloc = 0;
191
}
192

193
static void *run(void *arg)
194
{
195
	int hit = 0;
196
	struct grep_opt *opt = arg;
197

198
	while (1) {
199
		struct work_item *w = get_work();
200
		if (!w)
201
			break;
202

203
		opt->output_priv = w;
204
		hit |= grep_source(opt, &w->source);
205
		grep_source_clear_data(&w->source);
206
		work_done(w);
207
	}
208
	free_grep_patterns(opt);
209
	free(opt);
210

211
	return (void*) (intptr_t) hit;
212
}
213

214
static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
215
{
216
	struct work_item *w = opt->output_priv;
217
	strbuf_add(&w->out, buf, size);
218
}
219

220
static void start_threads(struct grep_opt *opt)
221
{
222
	int i;
223

224
	pthread_mutex_init(&grep_mutex, NULL);
225
	pthread_mutex_init(&grep_attr_mutex, NULL);
226
	pthread_cond_init(&cond_add, NULL);
227
	pthread_cond_init(&cond_write, NULL);
228
	pthread_cond_init(&cond_result, NULL);
229
	grep_use_locks = 1;
230
	enable_obj_read_lock();
231

232
	for (i = 0; i < ARRAY_SIZE(todo); i++) {
233
		strbuf_init(&todo[i].out, 0);
234
	}
235

236
	CALLOC_ARRAY(threads, num_threads);
237
	for (i = 0; i < num_threads; i++) {
238
		int err;
239
		struct grep_opt *o = grep_opt_dup(opt);
240
		o->output = strbuf_out;
241
		compile_grep_patterns(o);
242
		err = pthread_create(&threads[i], NULL, run, o);
243

244
		if (err)
245
			die(_("grep: failed to create thread: %s"),
246
			    strerror(err));
247
	}
248
}
249

250
static int wait_all(void)
251
{
252
	int hit = 0;
253
	int i;
254

255
	if (!HAVE_THREADS)
256
		BUG("Never call this function unless you have started threads");
257

258
	grep_lock();
259
	all_work_added = 1;
260

261
	/* Wait until all work is done. */
262
	while (todo_done != todo_end)
263
		pthread_cond_wait(&cond_result, &grep_mutex);
264

265
	/* Wake up all the consumer threads so they can see that there
266
	 * is no more work to do.
267
	 */
268
	pthread_cond_broadcast(&cond_add);
269
	grep_unlock();
270

271
	for (i = 0; i < num_threads; i++) {
272
		void *h;
273
		pthread_join(threads[i], &h);
274
		hit |= (int) (intptr_t) h;
275
	}
276

277
	free(threads);
278

279
	pthread_mutex_destroy(&grep_mutex);
280
	pthread_mutex_destroy(&grep_attr_mutex);
281
	pthread_cond_destroy(&cond_add);
282
	pthread_cond_destroy(&cond_write);
283
	pthread_cond_destroy(&cond_result);
284
	grep_use_locks = 0;
285
	disable_obj_read_lock();
286

287
	return hit;
288
}
289

290
static int grep_cmd_config(const char *var, const char *value,
291
			   const struct config_context *ctx, void *cb)
292
{
293
	int st = grep_config(var, value, ctx, cb);
294

295
	if (git_color_config(var, value, cb) < 0)
296
		st = -1;
297
	else if (git_default_config(var, value, ctx, cb) < 0)
298
		st = -1;
299

300
	if (!strcmp(var, "grep.threads")) {
301
		num_threads = git_config_int(var, value, ctx->kvi);
302
		if (num_threads < 0)
303
			die(_("invalid number of threads specified (%d) for %s"),
304
			    num_threads, var);
305
		else if (!HAVE_THREADS && num_threads > 1) {
306
			/*
307
			 * TRANSLATORS: %s is the configuration
308
			 * variable for tweaking threads, currently
309
			 * grep.threads
310
			 */
311
			warning(_("no threads support, ignoring %s"), var);
312
			num_threads = 1;
313
		}
314
	}
315

316
	if (!strcmp(var, "submodule.recurse"))
317
		recurse_submodules = git_config_bool(var, value);
318

319
	return st;
320
}
321

322
static void grep_source_name(struct grep_opt *opt, const char *filename,
323
			     int tree_name_len, struct strbuf *out)
324
{
325
	strbuf_reset(out);
326

327
	if (opt->null_following_name) {
328
		if (opt->relative && grep_prefix) {
329
			struct strbuf rel_buf = STRBUF_INIT;
330
			const char *rel_name =
331
				relative_path(filename + tree_name_len,
332
					      grep_prefix, &rel_buf);
333

334
			if (tree_name_len)
335
				strbuf_add(out, filename, tree_name_len);
336

337
			strbuf_addstr(out, rel_name);
338
			strbuf_release(&rel_buf);
339
		} else {
340
			strbuf_addstr(out, filename);
341
		}
342
		return;
343
	}
344

345
	if (opt->relative && grep_prefix)
346
		quote_path(filename + tree_name_len, grep_prefix, out, 0);
347
	else
348
		quote_c_style(filename + tree_name_len, out, NULL, 0);
349

350
	if (tree_name_len)
351
		strbuf_insert(out, 0, filename, tree_name_len);
352
}
353

354
static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
355
		     const char *filename, int tree_name_len,
356
		     const char *path)
357
{
358
	struct strbuf pathbuf = STRBUF_INIT;
359
	struct grep_source gs;
360

361
	grep_source_name(opt, filename, tree_name_len, &pathbuf);
362
	grep_source_init_oid(&gs, pathbuf.buf, path, oid, opt->repo);
363
	strbuf_release(&pathbuf);
364

365
	if (num_threads > 1) {
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);
371
		return 0;
372
	} else {
373
		int hit;
374

375
		hit = grep_source(opt, &gs);
376

377
		grep_source_clear(&gs);
378
		return hit;
379
	}
380
}
381

382
static int grep_file(struct grep_opt *opt, const char *filename)
383
{
384
	struct strbuf buf = STRBUF_INIT;
385
	struct grep_source gs;
386

387
	grep_source_name(opt, filename, 0, &buf);
388
	grep_source_init_file(&gs, buf.buf, filename);
389
	strbuf_release(&buf);
390

391
	if (num_threads > 1) {
392
		/*
393
		 * add_work() copies gs and thus assumes ownership of
394
		 * its fields, so do not call grep_source_clear()
395
		 */
396
		add_work(opt, &gs);
397
		return 0;
398
	} else {
399
		int hit;
400

401
		hit = grep_source(opt, &gs);
402

403
		grep_source_clear(&gs);
404
		return hit;
405
	}
406
}
407

408
static void append_path(struct grep_opt *opt, const void *data, size_t len)
409
{
410
	struct string_list *path_list = opt->output_priv;
411

412
	if (len == 1 && *(const char *)data == '\0')
413
		return;
414
	string_list_append_nodup(path_list, xstrndup(data, len));
415
}
416

417
static void run_pager(struct grep_opt *opt, const char *prefix)
418
{
419
	struct string_list *path_list = opt->output_priv;
420
	struct child_process child = CHILD_PROCESS_INIT;
421
	int i, status;
422

423
	for (i = 0; i < path_list->nr; i++)
424
		strvec_push(&child.args, path_list->items[i].string);
425
	child.dir = prefix;
426
	child.use_shell = 1;
427

428
	status = run_command(&child);
429
	if (status)
430
		exit(status);
431
}
432

433
static int grep_cache(struct grep_opt *opt,
434
		      const struct pathspec *pathspec, int cached);
435
static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
436
		     struct tree_desc *tree, struct strbuf *base, int tn_len,
437
		     int check_attr);
438

439
static int grep_submodule(struct grep_opt *opt,
440
			  const struct pathspec *pathspec,
441
			  const struct object_id *oid,
442
			  const char *filename, const char *path, int cached)
443
{
444
	struct repository *subrepo;
445
	struct repository *superproject = opt->repo;
446
	struct grep_opt subopt;
447
	int hit = 0;
448

449
	if (!is_submodule_active(superproject, path))
450
		return 0;
451

452
	subrepo = xmalloc(sizeof(*subrepo));
453
	if (repo_submodule_init(subrepo, superproject, path, null_oid())) {
454
		free(subrepo);
455
		return 0;
456
	}
457
	ALLOC_GROW(repos_to_free, repos_to_free_nr + 1, repos_to_free_alloc);
458
	repos_to_free[repos_to_free_nr++] = subrepo;
459

460
	/*
461
	 * NEEDSWORK: repo_read_gitmodules() might call
462
	 * add_to_alternates_memory() via config_from_gitmodules(). This
463
	 * operation causes a race condition with concurrent object readings
464
	 * performed by the worker threads. That's why we need obj_read_lock()
465
	 * here. It should be removed once it's no longer necessary to add the
466
	 * subrepo's odbs to the in-memory alternates list.
467
	 */
468
	obj_read_lock();
469

470
	/*
471
	 * NEEDSWORK: when reading a submodule, the sparsity settings in the
472
	 * superproject are incorrectly forgotten or misused. For example:
473
	 *
474
	 * 1. "command_requires_full_index"
475
	 * 	When this setting is turned on for `grep`, only the superproject
476
	 *	knows it. All the submodules are read with their own configs
477
	 *	and get prepare_repo_settings()'d. Therefore, these submodules
478
	 *	"forget" the sparse-index feature switch. As a result, the index
479
	 *	of these submodules are expanded unexpectedly.
480
	 *
481
	 * 2. "core_apply_sparse_checkout"
482
	 *	When running `grep` in the superproject, this setting is
483
	 *	populated using the superproject's configs. However, once
484
	 *	initialized, this config is globally accessible and is read by
485
	 *	prepare_repo_settings() for the submodules. For instance, if a
486
	 *	submodule is using a sparse-checkout, however, the superproject
487
	 *	is not, the result is that the config from the superproject will
488
	 *	dictate the behavior for the submodule, making it "forget" its
489
	 *	sparse-checkout state.
490
	 *
491
	 * 3. "core_sparse_checkout_cone"
492
	 *	ditto.
493
	 *
494
	 * Note that this list is not exhaustive.
495
	 */
496
	repo_read_gitmodules(subrepo, 0);
497

498
	/*
499
	 * All code paths tested by test code no longer need submodule ODBs to
500
	 * be added as alternates, but add it to the list just in case.
501
	 * Submodule ODBs added through add_submodule_odb_by_path() will be
502
	 * lazily registered as alternates when needed (and except in an
503
	 * unexpected code interaction, it won't be needed).
504
	 */
505
	add_submodule_odb_by_path(subrepo->objects->odb->path);
506
	obj_read_unlock();
507

508
	memcpy(&subopt, opt, sizeof(subopt));
509
	subopt.repo = subrepo;
510

511
	if (oid) {
512
		enum object_type object_type;
513
		struct tree_desc tree;
514
		void *data;
515
		unsigned long size;
516
		struct strbuf base = STRBUF_INIT;
517

518
		obj_read_lock();
519
		object_type = oid_object_info(subrepo, oid, NULL);
520
		obj_read_unlock();
521
		data = read_object_with_reference(subrepo,
522
						  oid, OBJ_TREE,
523
						  &size, NULL);
524
		if (!data)
525
			die(_("unable to read tree (%s)"), oid_to_hex(oid));
526

527
		strbuf_addstr(&base, filename);
528
		strbuf_addch(&base, '/');
529

530
		init_tree_desc(&tree, oid, data, size);
531
		hit = grep_tree(&subopt, pathspec, &tree, &base, base.len,
532
				object_type == OBJ_COMMIT);
533
		strbuf_release(&base);
534
		free(data);
535
	} else {
536
		hit = grep_cache(&subopt, pathspec, cached);
537
	}
538

539
	return hit;
540
}
541

542
static int grep_cache(struct grep_opt *opt,
543
		      const struct pathspec *pathspec, int cached)
544
{
545
	struct repository *repo = opt->repo;
546
	int hit = 0;
547
	int nr;
548
	struct strbuf name = STRBUF_INIT;
549
	int name_base_len = 0;
550
	if (repo->submodule_prefix) {
551
		name_base_len = strlen(repo->submodule_prefix);
552
		strbuf_addstr(&name, repo->submodule_prefix);
553
	}
554

555
	if (repo_read_index(repo) < 0)
556
		die(_("index file corrupt"));
557

558
	for (nr = 0; nr < repo->index->cache_nr; nr++) {
559
		const struct cache_entry *ce = repo->index->cache[nr];
560

561
		if (!cached && ce_skip_worktree(ce))
562
			continue;
563

564
		strbuf_setlen(&name, name_base_len);
565
		strbuf_addstr(&name, ce->name);
566
		if (S_ISSPARSEDIR(ce->ce_mode)) {
567
			enum object_type type;
568
			struct tree_desc tree;
569
			void *data;
570
			unsigned long size;
571

572
			data = repo_read_object_file(the_repository, &ce->oid,
573
						     &type, &size);
574
			if (!data)
575
				die(_("unable to read tree %s"), oid_to_hex(&ce->oid));
576
			init_tree_desc(&tree, &ce->oid, data, size);
577

578
			hit |= grep_tree(opt, pathspec, &tree, &name, 0, 0);
579
			strbuf_setlen(&name, name_base_len);
580
			strbuf_addstr(&name, ce->name);
581
			free(data);
582
		} else if (S_ISREG(ce->ce_mode) &&
583
		    match_pathspec(repo->index, pathspec, name.buf, name.len, 0, NULL,
584
				   S_ISDIR(ce->ce_mode) ||
585
				   S_ISGITLINK(ce->ce_mode))) {
586
			/*
587
			 * If CE_VALID is on, we assume worktree file and its
588
			 * cache entry are identical, even if worktree file has
589
			 * been modified, so use cache version instead
590
			 */
591
			if (cached || (ce->ce_flags & CE_VALID)) {
592
				if (ce_stage(ce) || ce_intent_to_add(ce))
593
					continue;
594
				hit |= grep_oid(opt, &ce->oid, name.buf,
595
						 0, name.buf);
596
			} else {
597
				hit |= grep_file(opt, name.buf);
598
			}
599
		} else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
600
			   submodule_path_match(repo->index, pathspec, name.buf, NULL)) {
601
			hit |= grep_submodule(opt, pathspec, NULL, ce->name,
602
					      ce->name, cached);
603
		} else {
604
			continue;
605
		}
606

607
		if (ce_stage(ce)) {
608
			do {
609
				nr++;
610
			} while (nr < repo->index->cache_nr &&
611
				 !strcmp(ce->name, repo->index->cache[nr]->name));
612
			nr--; /* compensate for loop control */
613
		}
614
		if (hit && opt->status_only)
615
			break;
616
	}
617

618
	strbuf_release(&name);
619
	return hit;
620
}
621

622
static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
623
		     struct tree_desc *tree, struct strbuf *base, int tn_len,
624
		     int check_attr)
625
{
626
	struct repository *repo = opt->repo;
627
	int hit = 0;
628
	enum interesting match = entry_not_interesting;
629
	struct name_entry entry;
630
	int old_baselen = base->len;
631
	struct strbuf name = STRBUF_INIT;
632
	int name_base_len = 0;
633
	if (repo->submodule_prefix) {
634
		strbuf_addstr(&name, repo->submodule_prefix);
635
		name_base_len = name.len;
636
	}
637

638
	while (tree_entry(tree, &entry)) {
639
		int te_len = tree_entry_len(&entry);
640

641
		if (match != all_entries_interesting) {
642
			strbuf_addstr(&name, base->buf + tn_len);
643
			match = tree_entry_interesting(repo->index,
644
						       &entry, &name,
645
						       pathspec);
646
			strbuf_setlen(&name, name_base_len);
647

648
			if (match == all_entries_not_interesting)
649
				break;
650
			if (match == entry_not_interesting)
651
				continue;
652
		}
653

654
		strbuf_add(base, entry.path, te_len);
655

656
		if (S_ISREG(entry.mode)) {
657
			hit |= grep_oid(opt, &entry.oid, base->buf, tn_len,
658
					 check_attr ? base->buf + tn_len : NULL);
659
		} else if (S_ISDIR(entry.mode)) {
660
			enum object_type type;
661
			struct tree_desc sub;
662
			void *data;
663
			unsigned long size;
664

665
			data = repo_read_object_file(the_repository,
666
						     &entry.oid, &type, &size);
667
			if (!data)
668
				die(_("unable to read tree (%s)"),
669
				    oid_to_hex(&entry.oid));
670

671
			strbuf_addch(base, '/');
672
			init_tree_desc(&sub, &entry.oid, data, size);
673
			hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
674
					 check_attr);
675
			free(data);
676
		} else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
677
			hit |= grep_submodule(opt, pathspec, &entry.oid,
678
					      base->buf, base->buf + tn_len,
679
					      1); /* ignored */
680
		}
681

682
		strbuf_setlen(base, old_baselen);
683

684
		if (hit && opt->status_only)
685
			break;
686
	}
687

688
	strbuf_release(&name);
689
	return hit;
690
}
691

692
static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
693
		       struct object *obj, const char *name, const char *path)
694
{
695
	if (obj->type == OBJ_BLOB)
696
		return grep_oid(opt, &obj->oid, name, 0, path);
697
	if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
698
		struct tree_desc tree;
699
		void *data;
700
		unsigned long size;
701
		struct strbuf base;
702
		int hit, len;
703

704
		data = read_object_with_reference(opt->repo,
705
						  &obj->oid, OBJ_TREE,
706
						  &size, NULL);
707
		if (!data)
708
			die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
709

710
		len = name ? strlen(name) : 0;
711
		strbuf_init(&base, PATH_MAX + len + 1);
712
		if (len) {
713
			strbuf_add(&base, name, len);
714
			strbuf_addch(&base, ':');
715
		}
716
		init_tree_desc(&tree, &obj->oid, data, size);
717
		hit = grep_tree(opt, pathspec, &tree, &base, base.len,
718
				obj->type == OBJ_COMMIT);
719
		strbuf_release(&base);
720
		free(data);
721
		return hit;
722
	}
723
	die(_("unable to grep from object of type %s"), type_name(obj->type));
724
}
725

726
static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
727
			const struct object_array *list)
728
{
729
	unsigned int i;
730
	int hit = 0;
731
	const unsigned int nr = list->nr;
732

733
	for (i = 0; i < nr; i++) {
734
		struct object *real_obj;
735

736
		obj_read_lock();
737
		real_obj = deref_tag(opt->repo, list->objects[i].item,
738
				     NULL, 0);
739
		obj_read_unlock();
740

741
		if (!real_obj) {
742
			char hex[GIT_MAX_HEXSZ + 1];
743
			const char *name = list->objects[i].name;
744

745
			if (!name) {
746
				oid_to_hex_r(hex, &list->objects[i].item->oid);
747
				name = hex;
748
			}
749
			die(_("invalid object '%s' given."), name);
750
		}
751

752
		/* load the gitmodules file for this rev */
753
		if (recurse_submodules) {
754
			submodule_free(opt->repo);
755
			obj_read_lock();
756
			gitmodules_config_oid(&real_obj->oid);
757
			obj_read_unlock();
758
		}
759
		if (grep_object(opt, pathspec, real_obj, list->objects[i].name,
760
				list->objects[i].path)) {
761
			hit = 1;
762
			if (opt->status_only)
763
				break;
764
		}
765
	}
766
	return hit;
767
}
768

769
static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
770
			  int exc_std, int use_index)
771
{
772
	struct dir_struct dir = DIR_INIT;
773
	int i, hit = 0;
774

775
	if (!use_index)
776
		dir.flags |= DIR_NO_GITLINKS;
777
	if (exc_std)
778
		setup_standard_excludes(&dir);
779

780
	fill_directory(&dir, opt->repo->index, pathspec);
781
	for (i = 0; i < dir.nr; i++) {
782
		hit |= grep_file(opt, dir.entries[i]->name);
783
		if (hit && opt->status_only)
784
			break;
785
	}
786
	dir_clear(&dir);
787
	return hit;
788
}
789

790
static int context_callback(const struct option *opt, const char *arg,
791
			    int unset)
792
{
793
	struct grep_opt *grep_opt = opt->value;
794
	int value;
795
	const char *endp;
796

797
	if (unset) {
798
		grep_opt->pre_context = grep_opt->post_context = 0;
799
		return 0;
800
	}
801
	value = strtol(arg, (char **)&endp, 10);
802
	if (*endp) {
803
		return error(_("switch `%c' expects a numerical value"),
804
			     opt->short_name);
805
	}
806
	grep_opt->pre_context = grep_opt->post_context = value;
807
	return 0;
808
}
809

810
static int file_callback(const struct option *opt, const char *arg, int unset)
811
{
812
	struct grep_opt *grep_opt = opt->value;
813
	int from_stdin;
814
	const char *filename = arg;
815
	FILE *patterns;
816
	int lno = 0;
817
	struct strbuf sb = STRBUF_INIT;
818

819
	BUG_ON_OPT_NEG(unset);
820

821
	if (!*filename)
822
		; /* leave it as-is */
823
	else
824
		filename = prefix_filename_except_for_dash(grep_prefix, filename);
825

826
	from_stdin = !strcmp(filename, "-");
827
	patterns = from_stdin ? stdin : fopen(filename, "r");
828
	if (!patterns)
829
		die_errno(_("cannot open '%s'"), arg);
830
	while (strbuf_getline(&sb, patterns) == 0) {
831
		/* ignore empty line like grep does */
832
		if (sb.len == 0)
833
			continue;
834

835
		append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
836
				GREP_PATTERN);
837
	}
838
	if (!from_stdin)
839
		fclose(patterns);
840
	strbuf_release(&sb);
841
	if (filename != arg)
842
		free((void *)filename);
843
	return 0;
844
}
845

846
static int not_callback(const struct option *opt, const char *arg, int unset)
847
{
848
	struct grep_opt *grep_opt = opt->value;
849
	BUG_ON_OPT_NEG(unset);
850
	BUG_ON_OPT_ARG(arg);
851
	append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
852
	return 0;
853
}
854

855
static int and_callback(const struct option *opt, const char *arg, int unset)
856
{
857
	struct grep_opt *grep_opt = opt->value;
858
	BUG_ON_OPT_NEG(unset);
859
	BUG_ON_OPT_ARG(arg);
860
	append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
861
	return 0;
862
}
863

864
static int open_callback(const struct option *opt, const char *arg, int unset)
865
{
866
	struct grep_opt *grep_opt = opt->value;
867
	BUG_ON_OPT_NEG(unset);
868
	BUG_ON_OPT_ARG(arg);
869
	append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
870
	return 0;
871
}
872

873
static int close_callback(const struct option *opt, const char *arg, int unset)
874
{
875
	struct grep_opt *grep_opt = opt->value;
876
	BUG_ON_OPT_NEG(unset);
877
	BUG_ON_OPT_ARG(arg);
878
	append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
879
	return 0;
880
}
881

882
static int pattern_callback(const struct option *opt, const char *arg,
883
			    int unset)
884
{
885
	struct grep_opt *grep_opt = opt->value;
886
	BUG_ON_OPT_NEG(unset);
887
	append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
888
	return 0;
889
}
890

891
int cmd_grep(int argc, const char **argv, const char *prefix)
892
{
893
	int hit = 0;
894
	int cached = 0, untracked = 0, opt_exclude = -1;
895
	int seen_dashdash = 0;
896
	int external_grep_allowed__ignored;
897
	const char *show_in_pager = NULL, *default_pager = "dummy";
898
	struct grep_opt opt;
899
	struct object_array list = OBJECT_ARRAY_INIT;
900
	struct pathspec pathspec;
901
	struct string_list path_list = STRING_LIST_INIT_DUP;
902
	int i;
903
	int dummy;
904
	int use_index = 1;
905
	int allow_revs;
906

907
	struct option options[] = {
908
		OPT_BOOL(0, "cached", &cached,
909
			N_("search in index instead of in the work tree")),
910
		OPT_NEGBIT(0, "no-index", &use_index,
911
			 N_("find in contents not managed by git"), 1),
912
		OPT_BOOL(0, "untracked", &untracked,
913
			N_("search in both tracked and untracked files")),
914
		OPT_SET_INT(0, "exclude-standard", &opt_exclude,
915
			    N_("ignore files specified via '.gitignore'"), 1),
916
		OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
917
			 N_("recursively search in each submodule")),
918
		OPT_GROUP(""),
919
		OPT_BOOL('v', "invert-match", &opt.invert,
920
			N_("show non-matching lines")),
921
		OPT_BOOL('i', "ignore-case", &opt.ignore_case,
922
			N_("case insensitive matching")),
923
		OPT_BOOL('w', "word-regexp", &opt.word_regexp,
924
			N_("match patterns only at word boundaries")),
925
		OPT_SET_INT('a', "text", &opt.binary,
926
			N_("process binary files as text"), GREP_BINARY_TEXT),
927
		OPT_SET_INT('I', NULL, &opt.binary,
928
			N_("don't match patterns in binary files"),
929
			GREP_BINARY_NOMATCH),
930
		OPT_BOOL(0, "textconv", &opt.allow_textconv,
931
			 N_("process binary files with textconv filters")),
932
		OPT_SET_INT('r', "recursive", &opt.max_depth,
933
			    N_("search in subdirectories (default)"), -1),
934
		OPT_INTEGER_F(0, "max-depth", &opt.max_depth,
935
			N_("descend at most <n> levels"), PARSE_OPT_NONEG),
936
		OPT_GROUP(""),
937
		OPT_SET_INT('E', "extended-regexp", &opt.pattern_type_option,
938
			    N_("use extended POSIX regular expressions"),
939
			    GREP_PATTERN_TYPE_ERE),
940
		OPT_SET_INT('G', "basic-regexp", &opt.pattern_type_option,
941
			    N_("use basic POSIX regular expressions (default)"),
942
			    GREP_PATTERN_TYPE_BRE),
943
		OPT_SET_INT('F', "fixed-strings", &opt.pattern_type_option,
944
			    N_("interpret patterns as fixed strings"),
945
			    GREP_PATTERN_TYPE_FIXED),
946
		OPT_SET_INT('P', "perl-regexp", &opt.pattern_type_option,
947
			    N_("use Perl-compatible regular expressions"),
948
			    GREP_PATTERN_TYPE_PCRE),
949
		OPT_GROUP(""),
950
		OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
951
		OPT_BOOL(0, "column", &opt.columnnum, N_("show column number of first match")),
952
		OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
953
		OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
954
		OPT_NEGBIT(0, "full-name", &opt.relative,
955
			N_("show filenames relative to top directory"), 1),
956
		OPT_BOOL('l', "files-with-matches", &opt.name_only,
957
			N_("show only filenames instead of matching lines")),
958
		OPT_BOOL(0, "name-only", &opt.name_only,
959
			N_("synonym for --files-with-matches")),
960
		OPT_BOOL('L', "files-without-match",
961
			&opt.unmatch_name_only,
962
			N_("show only the names of files without match")),
963
		OPT_BOOL_F('z', "null", &opt.null_following_name,
964
			   N_("print NUL after filenames"),
965
			   PARSE_OPT_NOCOMPLETE),
966
		OPT_BOOL('o', "only-matching", &opt.only_matching,
967
			N_("show only matching parts of a line")),
968
		OPT_BOOL('c', "count", &opt.count,
969
			N_("show the number of matches instead of matching lines")),
970
		OPT__COLOR(&opt.color, N_("highlight matches")),
971
		OPT_BOOL(0, "break", &opt.file_break,
972
			N_("print empty line between matches from different files")),
973
		OPT_BOOL(0, "heading", &opt.heading,
974
			N_("show filename only once above matches from same file")),
975
		OPT_GROUP(""),
976
		OPT_CALLBACK('C', "context", &opt, N_("n"),
977
			N_("show <n> context lines before and after matches"),
978
			context_callback),
979
		OPT_INTEGER('B', "before-context", &opt.pre_context,
980
			N_("show <n> context lines before matches")),
981
		OPT_INTEGER('A', "after-context", &opt.post_context,
982
			N_("show <n> context lines after matches")),
983
		OPT_INTEGER(0, "threads", &num_threads,
984
			N_("use <n> worker threads")),
985
		OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
986
			context_callback),
987
		OPT_BOOL('p', "show-function", &opt.funcname,
988
			N_("show a line with the function name before matches")),
989
		OPT_BOOL('W', "function-context", &opt.funcbody,
990
			N_("show the surrounding function")),
991
		OPT_GROUP(""),
992
		OPT_CALLBACK('f', NULL, &opt, N_("file"),
993
			N_("read patterns from file"), file_callback),
994
		OPT_CALLBACK_F('e', NULL, &opt, N_("pattern"),
995
			N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback),
996
		OPT_CALLBACK_F(0, "and", &opt, NULL,
997
			N_("combine patterns specified with -e"),
998
			PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback),
999
		OPT_BOOL_F(0, "or", &dummy, "", PARSE_OPT_NONEG),
1000
		OPT_CALLBACK_F(0, "not", &opt, NULL, "",
1001
			PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback),
1002
		OPT_CALLBACK_F('(', NULL, &opt, NULL, "",
1003
			PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1004
			open_callback),
1005
		OPT_CALLBACK_F(')', NULL, &opt, NULL, "",
1006
			PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1007
			close_callback),
1008
		OPT__QUIET(&opt.status_only,
1009
			   N_("indicate hit with exit status without output")),
1010
		OPT_BOOL(0, "all-match", &opt.all_match,
1011
			N_("show only matches from files that match all patterns")),
1012
		OPT_GROUP(""),
1013
		{ OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
1014
			N_("pager"), N_("show matching files in the pager"),
1015
			PARSE_OPT_OPTARG | PARSE_OPT_NOCOMPLETE,
1016
			NULL, (intptr_t)default_pager },
1017
		OPT_BOOL_F(0, "ext-grep", &external_grep_allowed__ignored,
1018
			   N_("allow calling of grep(1) (ignored by this build)"),
1019
			   PARSE_OPT_NOCOMPLETE),
1020
		OPT_INTEGER('m', "max-count", &opt.max_count,
1021
			N_("maximum number of results per file")),
1022
		OPT_END()
1023
	};
1024
	grep_prefix = prefix;
1025

1026
	grep_init(&opt, the_repository);
1027
	git_config(grep_cmd_config, &opt);
1028

1029
	/*
1030
	 * If there is no -- then the paths must exist in the working
1031
	 * tree.  If there is no explicit pattern specified with -e or
1032
	 * -f, we take the first unrecognized non option to be the
1033
	 * pattern, but then what follows it must be zero or more
1034
	 * valid refs up to the -- (if exists), and then existing
1035
	 * paths.  If there is an explicit pattern, then the first
1036
	 * unrecognized non option is the beginning of the refs list
1037
	 * that continues up to the -- (if exists), and then paths.
1038
	 */
1039
	argc = parse_options(argc, argv, prefix, options, grep_usage,
1040
			     PARSE_OPT_KEEP_DASHDASH |
1041
			     PARSE_OPT_STOP_AT_NON_OPTION);
1042

1043
	if (the_repository->gitdir) {
1044
		prepare_repo_settings(the_repository);
1045
		the_repository->settings.command_requires_full_index = 0;
1046
	}
1047

1048
	if (use_index && !startup_info->have_repository) {
1049
		int fallback = 0;
1050
		git_config_get_bool("grep.fallbacktonoindex", &fallback);
1051
		if (fallback)
1052
			use_index = 0;
1053
		else
1054
			/* die the same way as if we did it at the beginning */
1055
			setup_git_directory();
1056
	}
1057
	/* Ignore --recurse-submodules if --no-index is given or implied */
1058
	if (!use_index)
1059
		recurse_submodules = 0;
1060

1061
	/*
1062
	 * skip a -- separator; we know it cannot be
1063
	 * separating revisions from pathnames if
1064
	 * we haven't even had any patterns yet
1065
	 */
1066
	if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
1067
		argv++;
1068
		argc--;
1069
	}
1070

1071
	/* First unrecognized non-option token */
1072
	if (argc > 0 && !opt.pattern_list) {
1073
		append_grep_pattern(&opt, argv[0], "command line", 0,
1074
				    GREP_PATTERN);
1075
		argv++;
1076
		argc--;
1077
	}
1078

1079
	if (show_in_pager == default_pager)
1080
		show_in_pager = git_pager(1);
1081
	if (show_in_pager) {
1082
		opt.color = 0;
1083
		opt.name_only = 1;
1084
		opt.null_following_name = 1;
1085
		opt.output_priv = &path_list;
1086
		opt.output = append_path;
1087
		string_list_append(&path_list, show_in_pager);
1088
	}
1089

1090
	if (!opt.pattern_list)
1091
		die(_("no pattern given"));
1092

1093
	/* --only-matching has no effect with --invert. */
1094
	if (opt.invert)
1095
		opt.only_matching = 0;
1096

1097
	/*
1098
	 * We have to find "--" in a separate pass, because its presence
1099
	 * influences how we will parse arguments that come before it.
1100
	 */
1101
	for (i = 0; i < argc; i++) {
1102
		if (!strcmp(argv[i], "--")) {
1103
			seen_dashdash = 1;
1104
			break;
1105
		}
1106
	}
1107

1108
	/*
1109
	 * Resolve any rev arguments. If we have a dashdash, then everything up
1110
	 * to it must resolve as a rev. If not, then we stop at the first
1111
	 * non-rev and assume everything else is a path.
1112
	 */
1113
	allow_revs = use_index && !untracked;
1114
	for (i = 0; i < argc; i++) {
1115
		const char *arg = argv[i];
1116
		struct object_id oid;
1117
		struct object_context oc = {0};
1118
		struct object *object;
1119

1120
		if (!strcmp(arg, "--")) {
1121
			i++;
1122
			break;
1123
		}
1124

1125
		if (!allow_revs) {
1126
			if (seen_dashdash)
1127
				die(_("--no-index or --untracked cannot be used with revs"));
1128
			break;
1129
		}
1130

1131
		if (get_oid_with_context(the_repository, arg,
1132
					 GET_OID_RECORD_PATH,
1133
					 &oid, &oc)) {
1134
			if (seen_dashdash)
1135
				die(_("unable to resolve revision: %s"), arg);
1136
			break;
1137
		}
1138

1139
		object = parse_object_or_die(&oid, arg);
1140
		if (!seen_dashdash)
1141
			verify_non_filename(prefix, arg);
1142
		add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1143
		object_context_release(&oc);
1144
	}
1145

1146
	/*
1147
	 * Anything left over is presumed to be a path. But in the non-dashdash
1148
	 * "do what I mean" case, we verify and complain when that isn't true.
1149
	 */
1150
	if (!seen_dashdash) {
1151
		int j;
1152
		for (j = i; j < argc; j++)
1153
			verify_filename(prefix, argv[j], j == i && allow_revs);
1154
	}
1155

1156
	parse_pathspec(&pathspec, 0,
1157
		       PATHSPEC_PREFER_CWD |
1158
		       (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1159
		       prefix, argv + i);
1160
	pathspec.max_depth = opt.max_depth;
1161
	pathspec.recursive = 1;
1162
	pathspec.recurse_submodules = !!recurse_submodules;
1163

1164
	if (recurse_submodules && untracked)
1165
		die(_("--untracked not supported with --recurse-submodules"));
1166

1167
	/*
1168
	 * Optimize out the case where the amount of matches is limited to zero.
1169
	 * We do this to keep results consistent with GNU grep(1).
1170
	 */
1171
	if (opt.max_count == 0)
1172
		return 1;
1173

1174
	if (show_in_pager) {
1175
		if (num_threads > 1)
1176
			warning(_("invalid option combination, ignoring --threads"));
1177
		num_threads = 1;
1178
	} else if (!HAVE_THREADS && num_threads > 1) {
1179
		warning(_("no threads support, ignoring --threads"));
1180
		num_threads = 1;
1181
	} else if (num_threads < 0)
1182
		die(_("invalid number of threads specified (%d)"), num_threads);
1183
	else if (num_threads == 0)
1184
		num_threads = HAVE_THREADS ? online_cpus() : 1;
1185

1186
	if (num_threads > 1) {
1187
		if (!HAVE_THREADS)
1188
			BUG("Somebody got num_threads calculation wrong!");
1189
		if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1190
		    && (opt.pre_context || opt.post_context ||
1191
			opt.file_break || opt.funcbody))
1192
			skip_first_line = 1;
1193

1194
		/*
1195
		 * Pre-read gitmodules (if not read already) and force eager
1196
		 * initialization of packed_git to prevent racy lazy
1197
		 * reading/initialization once worker threads are started.
1198
		 */
1199
		if (recurse_submodules)
1200
			repo_read_gitmodules(the_repository, 1);
1201
		if (startup_info->have_repository)
1202
			(void)get_packed_git(the_repository);
1203

1204
		start_threads(&opt);
1205
	} else {
1206
		/*
1207
		 * The compiled patterns on the main path are only
1208
		 * used when not using threading. Otherwise
1209
		 * start_threads() above calls compile_grep_patterns()
1210
		 * for each thread.
1211
		 */
1212
		compile_grep_patterns(&opt);
1213
	}
1214

1215
	if (show_in_pager && (cached || list.nr))
1216
		die(_("--open-files-in-pager only works on the worktree"));
1217

1218
	if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1219
		const char *pager = path_list.items[0].string;
1220
		int len = strlen(pager);
1221

1222
		if (len > 4 && is_dir_sep(pager[len - 5]))
1223
			pager += len - 4;
1224

1225
		if (opt.ignore_case && !strcmp("less", pager))
1226
			string_list_append(&path_list, "-I");
1227

1228
		if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1229
			struct strbuf buf = STRBUF_INIT;
1230
			strbuf_addf(&buf, "+/%s%s",
1231
					strcmp("less", pager) ? "" : "*",
1232
					opt.pattern_list->pattern);
1233
			string_list_append_nodup(&path_list,
1234
						 strbuf_detach(&buf, NULL));
1235
		}
1236
	}
1237

1238
	if (!show_in_pager && !opt.status_only)
1239
		setup_pager();
1240

1241
	die_for_incompatible_opt3(!use_index, "--no-index",
1242
				  untracked, "--untracked",
1243
				  cached, "--cached");
1244

1245
	if (!use_index || untracked) {
1246
		int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1247
		hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1248
	} else if (0 <= opt_exclude) {
1249
		die(_("--[no-]exclude-standard cannot be used for tracked contents"));
1250
	} else if (!list.nr) {
1251
		if (!cached)
1252
			setup_work_tree();
1253

1254
		hit = grep_cache(&opt, &pathspec, cached);
1255
	} else {
1256
		if (cached)
1257
			die(_("both --cached and trees are given"));
1258

1259
		hit = grep_objects(&opt, &pathspec, &list);
1260
	}
1261

1262
	if (num_threads > 1)
1263
		hit |= wait_all();
1264
	if (hit && show_in_pager)
1265
		run_pager(&opt, prefix);
1266
	clear_pathspec(&pathspec);
1267
	string_list_clear(&path_list, 0);
1268
	free_grep_patterns(&opt);
1269
	object_array_clear(&list);
1270
	free_repos();
1271
	return !hit;
1272
}
1273