Merge branch 'rs/remote-mv-leakfix'
[git] / builtin / grep.c
1 /*
2  * Builtin "git grep"
3  *
4  * Copyright (c) 2006 Junio C Hamano
5  */
6 #include "cache.h"
7 #include "repository.h"
8 #include "config.h"
9 #include "blob.h"
10 #include "tree.h"
11 #include "commit.h"
12 #include "tag.h"
13 #include "tree-walk.h"
14 #include "builtin.h"
15 #include "parse-options.h"
16 #include "string-list.h"
17 #include "run-command.h"
18 #include "userdiff.h"
19 #include "grep.h"
20 #include "quote.h"
21 #include "dir.h"
22 #include "pathspec.h"
23 #include "submodule.h"
24 #include "submodule-config.h"
25 #include "object-store.h"
26
27 static char const * const grep_usage[] = {
28         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
29         NULL
30 };
31
32 static int recurse_submodules;
33
34 #define GREP_NUM_THREADS_DEFAULT 8
35 static int num_threads;
36
37 #ifndef NO_PTHREADS
38 static pthread_t *threads;
39
40 /* We use one producer thread and THREADS consumer
41  * threads. The producer adds struct work_items to 'todo' and the
42  * consumers pick work items from the same array.
43  */
44 struct work_item {
45         struct grep_source source;
46         char done;
47         struct strbuf out;
48 };
49
50 /* In the range [todo_done, todo_start) in 'todo' we have work_items
51  * that have been or are processed by a consumer thread. We haven't
52  * written the result for these to stdout yet.
53  *
54  * The work_items in [todo_start, todo_end) are waiting to be picked
55  * up by a consumer thread.
56  *
57  * The ranges are modulo TODO_SIZE.
58  */
59 #define TODO_SIZE 128
60 static struct work_item todo[TODO_SIZE];
61 static int todo_start;
62 static int todo_end;
63 static int todo_done;
64
65 /* Has all work items been added? */
66 static int all_work_added;
67
68 /* This lock protects all the variables above. */
69 static pthread_mutex_t grep_mutex;
70
71 static inline void grep_lock(void)
72 {
73         assert(num_threads);
74         pthread_mutex_lock(&grep_mutex);
75 }
76
77 static inline void grep_unlock(void)
78 {
79         assert(num_threads);
80         pthread_mutex_unlock(&grep_mutex);
81 }
82
83 /* Signalled when a new work_item is added to todo. */
84 static pthread_cond_t cond_add;
85
86 /* Signalled when the result from one work_item is written to
87  * stdout.
88  */
89 static pthread_cond_t cond_write;
90
91 /* Signalled when we are finished with everything. */
92 static pthread_cond_t cond_result;
93
94 static int skip_first_line;
95
96 static void add_work(struct grep_opt *opt, const struct grep_source *gs)
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
104         todo[todo_end].source = *gs;
105         if (opt->binary != GREP_BINARY_TEXT)
106                 grep_source_load_driver(&todo[todo_end].source);
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
115 static 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
134 static 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];
144                 if (w->out.len) {
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);
159                 }
160                 grep_source_clear(&w->source);
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
172 static 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;
183                 hit |= grep_source(opt, &w->source);
184                 grep_source_clear_data(&w->source);
185                 work_done(w);
186         }
187         free_grep_patterns(arg);
188         free(arg);
189
190         return (void*) (intptr_t) hit;
191 }
192
193 static 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
199 static void start_threads(struct grep_opt *opt)
200 {
201         int i;
202
203         pthread_mutex_init(&grep_mutex, NULL);
204         pthread_mutex_init(&grep_read_mutex, NULL);
205         pthread_mutex_init(&grep_attr_mutex, NULL);
206         pthread_cond_init(&cond_add, NULL);
207         pthread_cond_init(&cond_write, NULL);
208         pthread_cond_init(&cond_result, NULL);
209         grep_use_locks = 1;
210
211         for (i = 0; i < ARRAY_SIZE(todo); i++) {
212                 strbuf_init(&todo[i].out, 0);
213         }
214
215         threads = xcalloc(num_threads, sizeof(*threads));
216         for (i = 0; i < num_threads; i++) {
217                 int err;
218                 struct grep_opt *o = grep_opt_dup(opt);
219                 o->output = strbuf_out;
220                 if (i)
221                         o->debug = 0;
222                 compile_grep_patterns(o);
223                 err = pthread_create(&threads[i], NULL, run, o);
224
225                 if (err)
226                         die(_("grep: failed to create thread: %s"),
227                             strerror(err));
228         }
229 }
230
231 static 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
249         for (i = 0; i < num_threads; i++) {
250                 void *h;
251                 pthread_join(threads[i], &h);
252                 hit |= (int) (intptr_t) h;
253         }
254
255         free(threads);
256
257         pthread_mutex_destroy(&grep_mutex);
258         pthread_mutex_destroy(&grep_read_mutex);
259         pthread_mutex_destroy(&grep_attr_mutex);
260         pthread_cond_destroy(&cond_add);
261         pthread_cond_destroy(&cond_write);
262         pthread_cond_destroy(&cond_result);
263         grep_use_locks = 0;
264
265         return hit;
266 }
267 #else /* !NO_PTHREADS */
268
269 static int wait_all(void)
270 {
271         return 0;
272 }
273 #endif
274
275 static int grep_cmd_config(const char *var, const char *value, void *cb)
276 {
277         int st = grep_config(var, value, cb);
278         if (git_color_default_config(var, value, cb) < 0)
279                 st = -1;
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);
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
297         }
298
299         if (!strcmp(var, "submodule.recurse"))
300                 recurse_submodules = git_config_bool(var, value);
301
302         return st;
303 }
304
305 static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
306 {
307         void *data;
308
309         grep_read_lock();
310         data = read_object_file(oid, type, size);
311         grep_read_unlock();
312         return data;
313 }
314
315 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
316                      const char *filename, int tree_name_len,
317                      const char *path)
318 {
319         struct strbuf pathbuf = STRBUF_INIT;
320         struct grep_source gs;
321
322         if (opt->relative && opt->prefix_length) {
323                 quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
324                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
325         } else {
326                 strbuf_addstr(&pathbuf, filename);
327         }
328
329         grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
330         strbuf_release(&pathbuf);
331
332 #ifndef NO_PTHREADS
333         if (num_threads) {
334                 /*
335                  * add_work() copies gs and thus assumes ownership of
336                  * its fields, so do not call grep_source_clear()
337                  */
338                 add_work(opt, &gs);
339                 return 0;
340         } else
341 #endif
342         {
343                 int hit;
344
345                 hit = grep_source(opt, &gs);
346
347                 grep_source_clear(&gs);
348                 return hit;
349         }
350 }
351
352 static int grep_file(struct grep_opt *opt, const char *filename)
353 {
354         struct strbuf buf = STRBUF_INIT;
355         struct grep_source gs;
356
357         if (opt->relative && opt->prefix_length)
358                 quote_path_relative(filename, opt->prefix, &buf);
359         else
360                 strbuf_addstr(&buf, filename);
361
362         grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
363         strbuf_release(&buf);
364
365 #ifndef NO_PTHREADS
366         if (num_threads) {
367                 /*
368                  * add_work() copies gs and thus assumes ownership of
369                  * its fields, so do not call grep_source_clear()
370                  */
371                 add_work(opt, &gs);
372                 return 0;
373         } else
374 #endif
375         {
376                 int hit;
377
378                 hit = grep_source(opt, &gs);
379
380                 grep_source_clear(&gs);
381                 return hit;
382         }
383 }
384
385 static void append_path(struct grep_opt *opt, const void *data, size_t len)
386 {
387         struct string_list *path_list = opt->output_priv;
388
389         if (len == 1 && *(const char *)data == '\0')
390                 return;
391         string_list_append(path_list, xstrndup(data, len));
392 }
393
394 static void run_pager(struct grep_opt *opt, const char *prefix)
395 {
396         struct string_list *path_list = opt->output_priv;
397         struct child_process child = CHILD_PROCESS_INIT;
398         int i, status;
399
400         for (i = 0; i < path_list->nr; i++)
401                 argv_array_push(&child.args, path_list->items[i].string);
402         child.dir = prefix;
403         child.use_shell = 1;
404
405         status = run_command(&child);
406         if (status)
407                 exit(status);
408 }
409
410 static int grep_cache(struct grep_opt *opt, struct repository *repo,
411                       const struct pathspec *pathspec, int cached);
412 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
413                      struct tree_desc *tree, struct strbuf *base, int tn_len,
414                      int check_attr, struct repository *repo);
415
416 static int grep_submodule(struct grep_opt *opt, struct repository *superproject,
417                           const struct pathspec *pathspec,
418                           const struct object_id *oid,
419                           const char *filename, const char *path)
420 {
421         struct repository submodule;
422         int hit;
423
424         if (!is_submodule_active(superproject, path))
425                 return 0;
426
427         if (repo_submodule_init(&submodule, superproject, path))
428                 return 0;
429
430         repo_read_gitmodules(&submodule);
431
432         /*
433          * NEEDSWORK: This adds the submodule's object directory to the list of
434          * alternates for the single in-memory object store.  This has some bad
435          * consequences for memory (processed objects will never be freed) and
436          * performance (this increases the number of pack files git has to pay
437          * attention to, to the sum of the number of pack files in all the
438          * repositories processed so far).  This can be removed once the object
439          * store is no longer global and instead is a member of the repository
440          * object.
441          */
442         grep_read_lock();
443         add_to_alternates_memory(submodule.objects->objectdir);
444         grep_read_unlock();
445
446         if (oid) {
447                 struct object *object;
448                 struct tree_desc tree;
449                 void *data;
450                 unsigned long size;
451                 struct strbuf base = STRBUF_INIT;
452
453                 object = parse_object_or_die(oid, oid_to_hex(oid));
454
455                 grep_read_lock();
456                 data = read_object_with_reference(&object->oid, tree_type,
457                                                   &size, NULL);
458                 grep_read_unlock();
459
460                 if (!data)
461                         die(_("unable to read tree (%s)"), oid_to_hex(&object->oid));
462
463                 strbuf_addstr(&base, filename);
464                 strbuf_addch(&base, '/');
465
466                 init_tree_desc(&tree, data, size);
467                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
468                                 object->type == OBJ_COMMIT, &submodule);
469                 strbuf_release(&base);
470                 free(data);
471         } else {
472                 hit = grep_cache(opt, &submodule, pathspec, 1);
473         }
474
475         repo_clear(&submodule);
476         return hit;
477 }
478
479 static int grep_cache(struct grep_opt *opt, struct repository *repo,
480                       const struct pathspec *pathspec, int cached)
481 {
482         int hit = 0;
483         int nr;
484         struct strbuf name = STRBUF_INIT;
485         int name_base_len = 0;
486         if (repo->submodule_prefix) {
487                 name_base_len = strlen(repo->submodule_prefix);
488                 strbuf_addstr(&name, repo->submodule_prefix);
489         }
490
491         if (repo_read_index(repo) < 0)
492                 die(_("index file corrupt"));
493
494         for (nr = 0; nr < repo->index->cache_nr; nr++) {
495                 const struct cache_entry *ce = repo->index->cache[nr];
496                 strbuf_setlen(&name, name_base_len);
497                 strbuf_addstr(&name, ce->name);
498
499                 if (S_ISREG(ce->ce_mode) &&
500                     match_pathspec(pathspec, name.buf, name.len, 0, NULL,
501                                    S_ISDIR(ce->ce_mode) ||
502                                    S_ISGITLINK(ce->ce_mode))) {
503                         /*
504                          * If CE_VALID is on, we assume worktree file and its
505                          * cache entry are identical, even if worktree file has
506                          * been modified, so use cache version instead
507                          */
508                         if (cached || (ce->ce_flags & CE_VALID) ||
509                             ce_skip_worktree(ce)) {
510                                 if (ce_stage(ce) || ce_intent_to_add(ce))
511                                         continue;
512                                 hit |= grep_oid(opt, &ce->oid, name.buf,
513                                                  0, name.buf);
514                         } else {
515                                 hit |= grep_file(opt, name.buf);
516                         }
517                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
518                            submodule_path_match(pathspec, name.buf, NULL)) {
519                         hit |= grep_submodule(opt, repo, pathspec, NULL, ce->name, ce->name);
520                 } else {
521                         continue;
522                 }
523
524                 if (ce_stage(ce)) {
525                         do {
526                                 nr++;
527                         } while (nr < repo->index->cache_nr &&
528                                  !strcmp(ce->name, repo->index->cache[nr]->name));
529                         nr--; /* compensate for loop control */
530                 }
531                 if (hit && opt->status_only)
532                         break;
533         }
534
535         strbuf_release(&name);
536         return hit;
537 }
538
539 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
540                      struct tree_desc *tree, struct strbuf *base, int tn_len,
541                      int check_attr, struct repository *repo)
542 {
543         int hit = 0;
544         enum interesting match = entry_not_interesting;
545         struct name_entry entry;
546         int old_baselen = base->len;
547         struct strbuf name = STRBUF_INIT;
548         int name_base_len = 0;
549         if (repo->submodule_prefix) {
550                 strbuf_addstr(&name, repo->submodule_prefix);
551                 name_base_len = name.len;
552         }
553
554         while (tree_entry(tree, &entry)) {
555                 int te_len = tree_entry_len(&entry);
556
557                 if (match != all_entries_interesting) {
558                         strbuf_addstr(&name, base->buf + tn_len);
559                         match = tree_entry_interesting(&entry, &name,
560                                                        0, pathspec);
561                         strbuf_setlen(&name, name_base_len);
562
563                         if (match == all_entries_not_interesting)
564                                 break;
565                         if (match == entry_not_interesting)
566                                 continue;
567                 }
568
569                 strbuf_add(base, entry.path, te_len);
570
571                 if (S_ISREG(entry.mode)) {
572                         hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
573                                          check_attr ? base->buf + tn_len : NULL);
574                 } else if (S_ISDIR(entry.mode)) {
575                         enum object_type type;
576                         struct tree_desc sub;
577                         void *data;
578                         unsigned long size;
579
580                         data = lock_and_read_oid_file(entry.oid, &type, &size);
581                         if (!data)
582                                 die(_("unable to read tree (%s)"),
583                                     oid_to_hex(entry.oid));
584
585                         strbuf_addch(base, '/');
586                         init_tree_desc(&sub, data, size);
587                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
588                                          check_attr, repo);
589                         free(data);
590                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
591                         hit |= grep_submodule(opt, repo, pathspec, entry.oid,
592                                               base->buf, base->buf + tn_len);
593                 }
594
595                 strbuf_setlen(base, old_baselen);
596
597                 if (hit && opt->status_only)
598                         break;
599         }
600
601         strbuf_release(&name);
602         return hit;
603 }
604
605 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
606                        struct object *obj, const char *name, const char *path)
607 {
608         if (obj->type == OBJ_BLOB)
609                 return grep_oid(opt, &obj->oid, name, 0, path);
610         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
611                 struct tree_desc tree;
612                 void *data;
613                 unsigned long size;
614                 struct strbuf base;
615                 int hit, len;
616
617                 grep_read_lock();
618                 data = read_object_with_reference(&obj->oid, tree_type,
619                                                   &size, NULL);
620                 grep_read_unlock();
621
622                 if (!data)
623                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
624
625                 len = name ? strlen(name) : 0;
626                 strbuf_init(&base, PATH_MAX + len + 1);
627                 if (len) {
628                         strbuf_add(&base, name, len);
629                         strbuf_addch(&base, ':');
630                 }
631                 init_tree_desc(&tree, data, size);
632                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
633                                 obj->type == OBJ_COMMIT, the_repository);
634                 strbuf_release(&base);
635                 free(data);
636                 return hit;
637         }
638         die(_("unable to grep from object of type %s"), type_name(obj->type));
639 }
640
641 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
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(the_repository, list->objects[i].item,
651                                      NULL, 0);
652
653                 /* load the gitmodules file for this rev */
654                 if (recurse_submodules) {
655                         submodule_free(the_repository);
656                         gitmodules_config_oid(&real_obj->oid);
657                 }
658                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name,
659                                 list->objects[i].path)) {
660                         hit = 1;
661                         if (opt->status_only)
662                                 break;
663                 }
664         }
665         return hit;
666 }
667
668 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
669                           int exc_std, int use_index)
670 {
671         struct dir_struct dir;
672         int i, hit = 0;
673
674         memset(&dir, 0, sizeof(dir));
675         if (!use_index)
676                 dir.flags |= DIR_NO_GITLINKS;
677         if (exc_std)
678                 setup_standard_excludes(&dir);
679
680         fill_directory(&dir, &the_index, pathspec);
681         for (i = 0; i < dir.nr; i++) {
682                 if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
683                         continue;
684                 hit |= grep_file(opt, dir.entries[i]->name);
685                 if (hit && opt->status_only)
686                         break;
687         }
688         return hit;
689 }
690
691 static int context_callback(const struct option *opt, const char *arg,
692                             int unset)
693 {
694         struct grep_opt *grep_opt = opt->value;
695         int value;
696         const char *endp;
697
698         if (unset) {
699                 grep_opt->pre_context = grep_opt->post_context = 0;
700                 return 0;
701         }
702         value = strtol(arg, (char **)&endp, 10);
703         if (*endp) {
704                 return error(_("switch `%c' expects a numerical value"),
705                              opt->short_name);
706         }
707         grep_opt->pre_context = grep_opt->post_context = value;
708         return 0;
709 }
710
711 static int file_callback(const struct option *opt, const char *arg, int unset)
712 {
713         struct grep_opt *grep_opt = opt->value;
714         int from_stdin = !strcmp(arg, "-");
715         FILE *patterns;
716         int lno = 0;
717         struct strbuf sb = STRBUF_INIT;
718
719         patterns = from_stdin ? stdin : fopen(arg, "r");
720         if (!patterns)
721                 die_errno(_("cannot open '%s'"), arg);
722         while (strbuf_getline(&sb, patterns) == 0) {
723                 /* ignore empty line like grep does */
724                 if (sb.len == 0)
725                         continue;
726
727                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
728                                 GREP_PATTERN);
729         }
730         if (!from_stdin)
731                 fclose(patterns);
732         strbuf_release(&sb);
733         return 0;
734 }
735
736 static int not_callback(const struct option *opt, const char *arg, int unset)
737 {
738         struct grep_opt *grep_opt = opt->value;
739         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
740         return 0;
741 }
742
743 static int and_callback(const struct option *opt, const char *arg, int unset)
744 {
745         struct grep_opt *grep_opt = opt->value;
746         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
747         return 0;
748 }
749
750 static int open_callback(const struct option *opt, const char *arg, int unset)
751 {
752         struct grep_opt *grep_opt = opt->value;
753         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
754         return 0;
755 }
756
757 static int close_callback(const struct option *opt, const char *arg, int unset)
758 {
759         struct grep_opt *grep_opt = opt->value;
760         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
761         return 0;
762 }
763
764 static int pattern_callback(const struct option *opt, const char *arg,
765                             int unset)
766 {
767         struct grep_opt *grep_opt = opt->value;
768         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
769         return 0;
770 }
771
772 int cmd_grep(int argc, const char **argv, const char *prefix)
773 {
774         int hit = 0;
775         int cached = 0, untracked = 0, opt_exclude = -1;
776         int seen_dashdash = 0;
777         int external_grep_allowed__ignored;
778         const char *show_in_pager = NULL, *default_pager = "dummy";
779         struct grep_opt opt;
780         struct object_array list = OBJECT_ARRAY_INIT;
781         struct pathspec pathspec;
782         struct string_list path_list = STRING_LIST_INIT_NODUP;
783         int i;
784         int dummy;
785         int use_index = 1;
786         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
787         int allow_revs;
788
789         struct option options[] = {
790                 OPT_BOOL(0, "cached", &cached,
791                         N_("search in index instead of in the work tree")),
792                 OPT_NEGBIT(0, "no-index", &use_index,
793                          N_("find in contents not managed by git"), 1),
794                 OPT_BOOL(0, "untracked", &untracked,
795                         N_("search in both tracked and untracked files")),
796                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
797                             N_("ignore files specified via '.gitignore'"), 1),
798                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
799                          N_("recursively search in each submodule")),
800                 OPT_GROUP(""),
801                 OPT_BOOL('v', "invert-match", &opt.invert,
802                         N_("show non-matching lines")),
803                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
804                         N_("case insensitive matching")),
805                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
806                         N_("match patterns only at word boundaries")),
807                 OPT_SET_INT('a', "text", &opt.binary,
808                         N_("process binary files as text"), GREP_BINARY_TEXT),
809                 OPT_SET_INT('I', NULL, &opt.binary,
810                         N_("don't match patterns in binary files"),
811                         GREP_BINARY_NOMATCH),
812                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
813                          N_("process binary files with textconv filters")),
814                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
815                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
816                         NULL, 1 },
817                 OPT_GROUP(""),
818                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
819                             N_("use extended POSIX regular expressions"),
820                             GREP_PATTERN_TYPE_ERE),
821                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
822                             N_("use basic POSIX regular expressions (default)"),
823                             GREP_PATTERN_TYPE_BRE),
824                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
825                             N_("interpret patterns as fixed strings"),
826                             GREP_PATTERN_TYPE_FIXED),
827                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
828                             N_("use Perl-compatible regular expressions"),
829                             GREP_PATTERN_TYPE_PCRE),
830                 OPT_GROUP(""),
831                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
832                 OPT_BOOL(0, "column", &opt.columnnum, N_("show column number of first match")),
833                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
834                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
835                 OPT_NEGBIT(0, "full-name", &opt.relative,
836                         N_("show filenames relative to top directory"), 1),
837                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
838                         N_("show only filenames instead of matching lines")),
839                 OPT_BOOL(0, "name-only", &opt.name_only,
840                         N_("synonym for --files-with-matches")),
841                 OPT_BOOL('L', "files-without-match",
842                         &opt.unmatch_name_only,
843                         N_("show only the names of files without match")),
844                 OPT_BOOL_F('z', "null", &opt.null_following_name,
845                            N_("print NUL after filenames"),
846                            PARSE_OPT_NOCOMPLETE),
847                 OPT_BOOL('o', "only-matching", &opt.only_matching,
848                         N_("show only matching parts of a line")),
849                 OPT_BOOL('c', "count", &opt.count,
850                         N_("show the number of matches instead of matching lines")),
851                 OPT__COLOR(&opt.color, N_("highlight matches")),
852                 OPT_BOOL(0, "break", &opt.file_break,
853                         N_("print empty line between matches from different files")),
854                 OPT_BOOL(0, "heading", &opt.heading,
855                         N_("show filename only once above matches from same file")),
856                 OPT_GROUP(""),
857                 OPT_CALLBACK('C', "context", &opt, N_("n"),
858                         N_("show <n> context lines before and after matches"),
859                         context_callback),
860                 OPT_INTEGER('B', "before-context", &opt.pre_context,
861                         N_("show <n> context lines before matches")),
862                 OPT_INTEGER('A', "after-context", &opt.post_context,
863                         N_("show <n> context lines after matches")),
864                 OPT_INTEGER(0, "threads", &num_threads,
865                         N_("use <n> worker threads")),
866                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
867                         context_callback),
868                 OPT_BOOL('p', "show-function", &opt.funcname,
869                         N_("show a line with the function name before matches")),
870                 OPT_BOOL('W', "function-context", &opt.funcbody,
871                         N_("show the surrounding function")),
872                 OPT_GROUP(""),
873                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
874                         N_("read patterns from file"), file_callback),
875                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
876                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
877                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
878                   N_("combine patterns specified with -e"),
879                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
880                 OPT_BOOL(0, "or", &dummy, ""),
881                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
882                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
883                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
884                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
885                   open_callback },
886                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
887                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
888                   close_callback },
889                 OPT__QUIET(&opt.status_only,
890                            N_("indicate hit with exit status without output")),
891                 OPT_BOOL(0, "all-match", &opt.all_match,
892                         N_("show only matches from files that match all patterns")),
893                 OPT_SET_INT_F(0, "debug", &opt.debug,
894                               N_("show parse tree for grep expression"),
895                               1, PARSE_OPT_HIDDEN),
896                 OPT_GROUP(""),
897                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
898                         N_("pager"), N_("show matching files in the pager"),
899                         PARSE_OPT_OPTARG | PARSE_OPT_NOCOMPLETE,
900                         NULL, (intptr_t)default_pager },
901                 OPT_BOOL_F(0, "ext-grep", &external_grep_allowed__ignored,
902                            N_("allow calling of grep(1) (ignored by this build)"),
903                            PARSE_OPT_NOCOMPLETE),
904                 OPT_END()
905         };
906
907         init_grep_defaults();
908         git_config(grep_cmd_config, NULL);
909         grep_init(&opt, prefix);
910
911         /*
912          * If there is no -- then the paths must exist in the working
913          * tree.  If there is no explicit pattern specified with -e or
914          * -f, we take the first unrecognized non option to be the
915          * pattern, but then what follows it must be zero or more
916          * valid refs up to the -- (if exists), and then existing
917          * paths.  If there is an explicit pattern, then the first
918          * unrecognized non option is the beginning of the refs list
919          * that continues up to the -- (if exists), and then paths.
920          */
921         argc = parse_options(argc, argv, prefix, options, grep_usage,
922                              PARSE_OPT_KEEP_DASHDASH |
923                              PARSE_OPT_STOP_AT_NON_OPTION);
924         grep_commit_pattern_type(pattern_type_arg, &opt);
925
926         if (use_index && !startup_info->have_repository) {
927                 int fallback = 0;
928                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
929                 if (fallback)
930                         use_index = 0;
931                 else
932                         /* die the same way as if we did it at the beginning */
933                         setup_git_directory();
934         }
935
936         /*
937          * skip a -- separator; we know it cannot be
938          * separating revisions from pathnames if
939          * we haven't even had any patterns yet
940          */
941         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
942                 argv++;
943                 argc--;
944         }
945
946         /* First unrecognized non-option token */
947         if (argc > 0 && !opt.pattern_list) {
948                 append_grep_pattern(&opt, argv[0], "command line", 0,
949                                     GREP_PATTERN);
950                 argv++;
951                 argc--;
952         }
953
954         if (show_in_pager == default_pager)
955                 show_in_pager = git_pager(1);
956         if (show_in_pager) {
957                 opt.color = 0;
958                 opt.name_only = 1;
959                 opt.null_following_name = 1;
960                 opt.output_priv = &path_list;
961                 opt.output = append_path;
962                 string_list_append(&path_list, show_in_pager);
963         }
964
965         if (!opt.pattern_list)
966                 die(_("no pattern given"));
967
968         /* --only-matching has no effect with --invert. */
969         if (opt.invert)
970                 opt.only_matching = 0;
971
972         /*
973          * We have to find "--" in a separate pass, because its presence
974          * influences how we will parse arguments that come before it.
975          */
976         for (i = 0; i < argc; i++) {
977                 if (!strcmp(argv[i], "--")) {
978                         seen_dashdash = 1;
979                         break;
980                 }
981         }
982
983         /*
984          * Resolve any rev arguments. If we have a dashdash, then everything up
985          * to it must resolve as a rev. If not, then we stop at the first
986          * non-rev and assume everything else is a path.
987          */
988         allow_revs = use_index && !untracked;
989         for (i = 0; i < argc; i++) {
990                 const char *arg = argv[i];
991                 struct object_id oid;
992                 struct object_context oc;
993                 struct object *object;
994
995                 if (!strcmp(arg, "--")) {
996                         i++;
997                         break;
998                 }
999
1000                 if (!allow_revs) {
1001                         if (seen_dashdash)
1002                                 die(_("--no-index or --untracked cannot be used with revs"));
1003                         break;
1004                 }
1005
1006                 if (get_oid_with_context(arg, GET_OID_RECORD_PATH,
1007                                          &oid, &oc)) {
1008                         if (seen_dashdash)
1009                                 die(_("unable to resolve revision: %s"), arg);
1010                         break;
1011                 }
1012
1013                 object = parse_object_or_die(&oid, arg);
1014                 if (!seen_dashdash)
1015                         verify_non_filename(prefix, arg);
1016                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1017                 free(oc.path);
1018         }
1019
1020         /*
1021          * Anything left over is presumed to be a path. But in the non-dashdash
1022          * "do what I mean" case, we verify and complain when that isn't true.
1023          */
1024         if (!seen_dashdash) {
1025                 int j;
1026                 for (j = i; j < argc; j++)
1027                         verify_filename(prefix, argv[j], j == i && allow_revs);
1028         }
1029
1030         parse_pathspec(&pathspec, 0,
1031                        PATHSPEC_PREFER_CWD |
1032                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1033                        prefix, argv + i);
1034         pathspec.max_depth = opt.max_depth;
1035         pathspec.recursive = 1;
1036         pathspec.recurse_submodules = !!recurse_submodules;
1037
1038 #ifndef NO_PTHREADS
1039         if (list.nr || cached || show_in_pager)
1040                 num_threads = 0;
1041         else if (num_threads == 0)
1042                 num_threads = GREP_NUM_THREADS_DEFAULT;
1043         else if (num_threads < 0)
1044                 die(_("invalid number of threads specified (%d)"), num_threads);
1045         if (num_threads == 1)
1046                 num_threads = 0;
1047 #else
1048         if (num_threads)
1049                 warning(_("no threads support, ignoring --threads"));
1050         num_threads = 0;
1051 #endif
1052
1053         if (!num_threads)
1054                 /*
1055                  * The compiled patterns on the main path are only
1056                  * used when not using threading. Otherwise
1057                  * start_threads() below calls compile_grep_patterns()
1058                  * for each thread.
1059                  */
1060                 compile_grep_patterns(&opt);
1061
1062 #ifndef NO_PTHREADS
1063         if (num_threads) {
1064                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1065                     && (opt.pre_context || opt.post_context ||
1066                         opt.file_break || opt.funcbody))
1067                         skip_first_line = 1;
1068                 start_threads(&opt);
1069         }
1070 #endif
1071
1072         if (show_in_pager && (cached || list.nr))
1073                 die(_("--open-files-in-pager only works on the worktree"));
1074
1075         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1076                 const char *pager = path_list.items[0].string;
1077                 int len = strlen(pager);
1078
1079                 if (len > 4 && is_dir_sep(pager[len - 5]))
1080                         pager += len - 4;
1081
1082                 if (opt.ignore_case && !strcmp("less", pager))
1083                         string_list_append(&path_list, "-I");
1084
1085                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1086                         struct strbuf buf = STRBUF_INIT;
1087                         strbuf_addf(&buf, "+/%s%s",
1088                                         strcmp("less", pager) ? "" : "*",
1089                                         opt.pattern_list->pattern);
1090                         string_list_append(&path_list, buf.buf);
1091                         strbuf_detach(&buf, NULL);
1092                 }
1093         }
1094
1095         if (recurse_submodules && (!use_index || untracked))
1096                 die(_("option not supported with --recurse-submodules"));
1097
1098         if (!show_in_pager && !opt.status_only)
1099                 setup_pager();
1100
1101         if (!use_index && (untracked || cached))
1102                 die(_("--cached or --untracked cannot be used with --no-index"));
1103
1104         if (!use_index || untracked) {
1105                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1106                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1107         } else if (0 <= opt_exclude) {
1108                 die(_("--[no-]exclude-standard cannot be used for tracked contents"));
1109         } else if (!list.nr) {
1110                 if (!cached)
1111                         setup_work_tree();
1112
1113                 hit = grep_cache(&opt, the_repository, &pathspec, cached);
1114         } else {
1115                 if (cached)
1116                         die(_("both --cached and trees are given"));
1117
1118                 hit = grep_objects(&opt, &pathspec, &list);
1119         }
1120
1121         if (num_threads)
1122                 hit |= wait_all();
1123         if (hit && show_in_pager)
1124                 run_pager(&opt, prefix);
1125         clear_pathspec(&pathspec);
1126         free_grep_patterns(&opt);
1127         return !hit;
1128 }