Merge branch 'sg/daemon-test-signal-fix'
[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 static pthread_t *threads;
38
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  */
43 struct work_item {
44         struct grep_source source;
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
59 static struct work_item todo[TODO_SIZE];
60 static int todo_start;
61 static int todo_end;
62 static int todo_done;
63
64 /* Has all work items been added? */
65 static int all_work_added;
66
67 /* This lock protects all the variables above. */
68 static pthread_mutex_t grep_mutex;
69
70 static inline void grep_lock(void)
71 {
72         pthread_mutex_lock(&grep_mutex);
73 }
74
75 static inline void grep_unlock(void)
76 {
77         pthread_mutex_unlock(&grep_mutex);
78 }
79
80 /* Signalled when a new work_item is added to todo. */
81 static pthread_cond_t cond_add;
82
83 /* Signalled when the result from one work_item is written to
84  * stdout.
85  */
86 static pthread_cond_t cond_write;
87
88 /* Signalled when we are finished with everything. */
89 static pthread_cond_t cond_result;
90
91 static int skip_first_line;
92
93 static void add_work(struct grep_opt *opt, const struct grep_source *gs)
94 {
95         grep_lock();
96
97         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
98                 pthread_cond_wait(&cond_write, &grep_mutex);
99         }
100
101         todo[todo_end].source = *gs;
102         if (opt->binary != GREP_BINARY_TEXT)
103                 grep_source_load_driver(&todo[todo_end].source,
104                                         opt->repo->index);
105         todo[todo_end].done = 0;
106         strbuf_reset(&todo[todo_end].out);
107         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
108
109         pthread_cond_signal(&cond_add);
110         grep_unlock();
111 }
112
113 static struct work_item *get_work(void)
114 {
115         struct work_item *ret;
116
117         grep_lock();
118         while (todo_start == todo_end && !all_work_added) {
119                 pthread_cond_wait(&cond_add, &grep_mutex);
120         }
121
122         if (todo_start == todo_end && all_work_added) {
123                 ret = NULL;
124         } else {
125                 ret = &todo[todo_start];
126                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
127         }
128         grep_unlock();
129         return ret;
130 }
131
132 static void work_done(struct work_item *w)
133 {
134         int old_done;
135
136         grep_lock();
137         w->done = 1;
138         old_done = todo_done;
139         for(; todo[todo_done].done && todo_done != todo_start;
140             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
141                 w = &todo[todo_done];
142                 if (w->out.len) {
143                         const char *p = w->out.buf;
144                         size_t len = w->out.len;
145
146                         /* Skip the leading hunk mark of the first file. */
147                         if (skip_first_line) {
148                                 while (len) {
149                                         len--;
150                                         if (*p++ == '\n')
151                                                 break;
152                                 }
153                                 skip_first_line = 0;
154                         }
155
156                         write_or_die(1, p, len);
157                 }
158                 grep_source_clear(&w->source);
159         }
160
161         if (old_done != todo_done)
162                 pthread_cond_signal(&cond_write);
163
164         if (all_work_added && todo_done == todo_end)
165                 pthread_cond_signal(&cond_result);
166
167         grep_unlock();
168 }
169
170 static void *run(void *arg)
171 {
172         int hit = 0;
173         struct grep_opt *opt = arg;
174
175         while (1) {
176                 struct work_item *w = get_work();
177                 if (!w)
178                         break;
179
180                 opt->output_priv = w;
181                 hit |= grep_source(opt, &w->source);
182                 grep_source_clear_data(&w->source);
183                 work_done(w);
184         }
185         free_grep_patterns(arg);
186         free(arg);
187
188         return (void*) (intptr_t) hit;
189 }
190
191 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
192 {
193         struct work_item *w = opt->output_priv;
194         strbuf_add(&w->out, buf, size);
195 }
196
197 static void start_threads(struct grep_opt *opt)
198 {
199         int i;
200
201         pthread_mutex_init(&grep_mutex, NULL);
202         pthread_mutex_init(&grep_read_mutex, NULL);
203         pthread_mutex_init(&grep_attr_mutex, NULL);
204         pthread_cond_init(&cond_add, NULL);
205         pthread_cond_init(&cond_write, NULL);
206         pthread_cond_init(&cond_result, NULL);
207         grep_use_locks = 1;
208
209         for (i = 0; i < ARRAY_SIZE(todo); i++) {
210                 strbuf_init(&todo[i].out, 0);
211         }
212
213         threads = xcalloc(num_threads, sizeof(*threads));
214         for (i = 0; i < num_threads; i++) {
215                 int err;
216                 struct grep_opt *o = grep_opt_dup(opt);
217                 o->output = strbuf_out;
218                 if (i)
219                         o->debug = 0;
220                 compile_grep_patterns(o);
221                 err = pthread_create(&threads[i], NULL, run, o);
222
223                 if (err)
224                         die(_("grep: failed to create thread: %s"),
225                             strerror(err));
226         }
227 }
228
229 static int wait_all(void)
230 {
231         int hit = 0;
232         int i;
233
234         if (!HAVE_THREADS)
235                 BUG("Never call this function unless you have started threads");
236
237         grep_lock();
238         all_work_added = 1;
239
240         /* Wait until all work is done. */
241         while (todo_done != todo_end)
242                 pthread_cond_wait(&cond_result, &grep_mutex);
243
244         /* Wake up all the consumer threads so they can see that there
245          * is no more work to do.
246          */
247         pthread_cond_broadcast(&cond_add);
248         grep_unlock();
249
250         for (i = 0; i < num_threads; i++) {
251                 void *h;
252                 pthread_join(threads[i], &h);
253                 hit |= (int) (intptr_t) h;
254         }
255
256         free(threads);
257
258         pthread_mutex_destroy(&grep_mutex);
259         pthread_mutex_destroy(&grep_read_mutex);
260         pthread_mutex_destroy(&grep_attr_mutex);
261         pthread_cond_destroy(&cond_add);
262         pthread_cond_destroy(&cond_write);
263         pthread_cond_destroy(&cond_result);
264         grep_use_locks = 0;
265
266         return hit;
267 }
268
269 static int grep_cmd_config(const char *var, const char *value, void *cb)
270 {
271         int st = grep_config(var, value, cb);
272         if (git_color_default_config(var, value, cb) < 0)
273                 st = -1;
274
275         if (!strcmp(var, "grep.threads")) {
276                 num_threads = git_config_int(var, value);
277                 if (num_threads < 0)
278                         die(_("invalid number of threads specified (%d) for %s"),
279                             num_threads, var);
280                 else if (!HAVE_THREADS && num_threads > 1) {
281                         /*
282                          * TRANSLATORS: %s is the configuration
283                          * variable for tweaking threads, currently
284                          * grep.threads
285                          */
286                         warning(_("no threads support, ignoring %s"), var);
287                         num_threads = 1;
288                 }
289         }
290
291         if (!strcmp(var, "submodule.recurse"))
292                 recurse_submodules = git_config_bool(var, value);
293
294         return st;
295 }
296
297 static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
298 {
299         void *data;
300
301         grep_read_lock();
302         data = read_object_file(oid, type, size);
303         grep_read_unlock();
304         return data;
305 }
306
307 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
308                      const char *filename, int tree_name_len,
309                      const char *path)
310 {
311         struct strbuf pathbuf = STRBUF_INIT;
312         struct grep_source gs;
313
314         if (opt->relative && opt->prefix_length) {
315                 quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
316                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
317         } else {
318                 strbuf_addstr(&pathbuf, filename);
319         }
320
321         grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
322         strbuf_release(&pathbuf);
323
324         if (num_threads > 1) {
325                 /*
326                  * add_work() copies gs and thus assumes ownership of
327                  * its fields, so do not call grep_source_clear()
328                  */
329                 add_work(opt, &gs);
330                 return 0;
331         } else {
332                 int hit;
333
334                 hit = grep_source(opt, &gs);
335
336                 grep_source_clear(&gs);
337                 return hit;
338         }
339 }
340
341 static int grep_file(struct grep_opt *opt, const char *filename)
342 {
343         struct strbuf buf = STRBUF_INIT;
344         struct grep_source gs;
345
346         if (opt->relative && opt->prefix_length)
347                 quote_path_relative(filename, opt->prefix, &buf);
348         else
349                 strbuf_addstr(&buf, filename);
350
351         grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
352         strbuf_release(&buf);
353
354         if (num_threads > 1) {
355                 /*
356                  * add_work() copies gs and thus assumes ownership of
357                  * its fields, so do not call grep_source_clear()
358                  */
359                 add_work(opt, &gs);
360                 return 0;
361         } else {
362                 int hit;
363
364                 hit = grep_source(opt, &gs);
365
366                 grep_source_clear(&gs);
367                 return hit;
368         }
369 }
370
371 static void append_path(struct grep_opt *opt, const void *data, size_t len)
372 {
373         struct string_list *path_list = opt->output_priv;
374
375         if (len == 1 && *(const char *)data == '\0')
376                 return;
377         string_list_append(path_list, xstrndup(data, len));
378 }
379
380 static void run_pager(struct grep_opt *opt, const char *prefix)
381 {
382         struct string_list *path_list = opt->output_priv;
383         struct child_process child = CHILD_PROCESS_INIT;
384         int i, status;
385
386         for (i = 0; i < path_list->nr; i++)
387                 argv_array_push(&child.args, path_list->items[i].string);
388         child.dir = prefix;
389         child.use_shell = 1;
390
391         status = run_command(&child);
392         if (status)
393                 exit(status);
394 }
395
396 static int grep_cache(struct grep_opt *opt, struct repository *repo,
397                       const struct pathspec *pathspec, int cached);
398 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
399                      struct tree_desc *tree, struct strbuf *base, int tn_len,
400                      int check_attr, struct repository *repo);
401
402 static int grep_submodule(struct grep_opt *opt, struct repository *superproject,
403                           const struct pathspec *pathspec,
404                           const struct object_id *oid,
405                           const char *filename, const char *path)
406 {
407         struct repository submodule;
408         int hit;
409
410         /*
411          * NEEDSWORK: submodules functions need to be protected because they
412          * access the object store via config_from_gitmodules(): the latter
413          * uses get_oid() which, for now, relies on the global the_repository
414          * object.
415          */
416         grep_read_lock();
417
418         if (!is_submodule_active(superproject, path)) {
419                 grep_read_unlock();
420                 return 0;
421         }
422
423         if (repo_submodule_init(&submodule, superproject, path)) {
424                 grep_read_unlock();
425                 return 0;
426         }
427
428         repo_read_gitmodules(&submodule);
429
430         /*
431          * NEEDSWORK: This adds the submodule's object directory to the list of
432          * alternates for the single in-memory object store.  This has some bad
433          * consequences for memory (processed objects will never be freed) and
434          * performance (this increases the number of pack files git has to pay
435          * attention to, to the sum of the number of pack files in all the
436          * repositories processed so far).  This can be removed once the object
437          * store is no longer global and instead is a member of the repository
438          * object.
439          */
440         add_to_alternates_memory(submodule.objects->objectdir);
441         grep_read_unlock();
442
443         if (oid) {
444                 struct object *object;
445                 struct tree_desc tree;
446                 void *data;
447                 unsigned long size;
448                 struct strbuf base = STRBUF_INIT;
449
450                 object = parse_object_or_die(oid, oid_to_hex(oid));
451
452                 grep_read_lock();
453                 data = read_object_with_reference(&object->oid, tree_type,
454                                                   &size, NULL);
455                 grep_read_unlock();
456
457                 if (!data)
458                         die(_("unable to read tree (%s)"), oid_to_hex(&object->oid));
459
460                 strbuf_addstr(&base, filename);
461                 strbuf_addch(&base, '/');
462
463                 init_tree_desc(&tree, data, size);
464                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
465                                 object->type == OBJ_COMMIT, &submodule);
466                 strbuf_release(&base);
467                 free(data);
468         } else {
469                 hit = grep_cache(opt, &submodule, pathspec, 1);
470         }
471
472         repo_clear(&submodule);
473         return hit;
474 }
475
476 static int grep_cache(struct grep_opt *opt, struct repository *repo,
477                       const struct pathspec *pathspec, int cached)
478 {
479         int hit = 0;
480         int nr;
481         struct strbuf name = STRBUF_INIT;
482         int name_base_len = 0;
483         if (repo->submodule_prefix) {
484                 name_base_len = strlen(repo->submodule_prefix);
485                 strbuf_addstr(&name, repo->submodule_prefix);
486         }
487
488         if (repo_read_index(repo) < 0)
489                 die(_("index file corrupt"));
490
491         for (nr = 0; nr < repo->index->cache_nr; nr++) {
492                 const struct cache_entry *ce = repo->index->cache[nr];
493                 strbuf_setlen(&name, name_base_len);
494                 strbuf_addstr(&name, ce->name);
495
496                 if (S_ISREG(ce->ce_mode) &&
497                     match_pathspec(repo->index, pathspec, name.buf, name.len, 0, NULL,
498                                    S_ISDIR(ce->ce_mode) ||
499                                    S_ISGITLINK(ce->ce_mode))) {
500                         /*
501                          * If CE_VALID is on, we assume worktree file and its
502                          * cache entry are identical, even if worktree file has
503                          * been modified, so use cache version instead
504                          */
505                         if (cached || (ce->ce_flags & CE_VALID) ||
506                             ce_skip_worktree(ce)) {
507                                 if (ce_stage(ce) || ce_intent_to_add(ce))
508                                         continue;
509                                 hit |= grep_oid(opt, &ce->oid, name.buf,
510                                                  0, name.buf);
511                         } else {
512                                 hit |= grep_file(opt, name.buf);
513                         }
514                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
515                            submodule_path_match(repo->index, pathspec, name.buf, NULL)) {
516                         hit |= grep_submodule(opt, repo, pathspec, NULL, ce->name, ce->name);
517                 } else {
518                         continue;
519                 }
520
521                 if (ce_stage(ce)) {
522                         do {
523                                 nr++;
524                         } while (nr < repo->index->cache_nr &&
525                                  !strcmp(ce->name, repo->index->cache[nr]->name));
526                         nr--; /* compensate for loop control */
527                 }
528                 if (hit && opt->status_only)
529                         break;
530         }
531
532         strbuf_release(&name);
533         return hit;
534 }
535
536 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
537                      struct tree_desc *tree, struct strbuf *base, int tn_len,
538                      int check_attr, struct repository *repo)
539 {
540         int hit = 0;
541         enum interesting match = entry_not_interesting;
542         struct name_entry entry;
543         int old_baselen = base->len;
544         struct strbuf name = STRBUF_INIT;
545         int name_base_len = 0;
546         if (repo->submodule_prefix) {
547                 strbuf_addstr(&name, repo->submodule_prefix);
548                 name_base_len = name.len;
549         }
550
551         while (tree_entry(tree, &entry)) {
552                 int te_len = tree_entry_len(&entry);
553
554                 if (match != all_entries_interesting) {
555                         strbuf_addstr(&name, base->buf + tn_len);
556                         match = tree_entry_interesting(&entry, &name,
557                                                        0, pathspec);
558                         strbuf_setlen(&name, name_base_len);
559
560                         if (match == all_entries_not_interesting)
561                                 break;
562                         if (match == entry_not_interesting)
563                                 continue;
564                 }
565
566                 strbuf_add(base, entry.path, te_len);
567
568                 if (S_ISREG(entry.mode)) {
569                         hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
570                                          check_attr ? base->buf + tn_len : NULL);
571                 } else if (S_ISDIR(entry.mode)) {
572                         enum object_type type;
573                         struct tree_desc sub;
574                         void *data;
575                         unsigned long size;
576
577                         data = lock_and_read_oid_file(entry.oid, &type, &size);
578                         if (!data)
579                                 die(_("unable to read tree (%s)"),
580                                     oid_to_hex(entry.oid));
581
582                         strbuf_addch(base, '/');
583                         init_tree_desc(&sub, data, size);
584                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
585                                          check_attr, repo);
586                         free(data);
587                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
588                         hit |= grep_submodule(opt, repo, pathspec, entry.oid,
589                                               base->buf, base->buf + tn_len);
590                 }
591
592                 strbuf_setlen(base, old_baselen);
593
594                 if (hit && opt->status_only)
595                         break;
596         }
597
598         strbuf_release(&name);
599         return hit;
600 }
601
602 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
603                        struct object *obj, const char *name, const char *path)
604 {
605         if (obj->type == OBJ_BLOB)
606                 return grep_oid(opt, &obj->oid, name, 0, path);
607         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
608                 struct tree_desc tree;
609                 void *data;
610                 unsigned long size;
611                 struct strbuf base;
612                 int hit, len;
613
614                 grep_read_lock();
615                 data = read_object_with_reference(&obj->oid, tree_type,
616                                                   &size, NULL);
617                 grep_read_unlock();
618
619                 if (!data)
620                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
621
622                 len = name ? strlen(name) : 0;
623                 strbuf_init(&base, PATH_MAX + len + 1);
624                 if (len) {
625                         strbuf_add(&base, name, len);
626                         strbuf_addch(&base, ':');
627                 }
628                 init_tree_desc(&tree, data, size);
629                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
630                                 obj->type == OBJ_COMMIT, the_repository);
631                 strbuf_release(&base);
632                 free(data);
633                 return hit;
634         }
635         die(_("unable to grep from object of type %s"), type_name(obj->type));
636 }
637
638 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
639                         const struct object_array *list)
640 {
641         unsigned int i;
642         int hit = 0;
643         const unsigned int nr = list->nr;
644
645         for (i = 0; i < nr; i++) {
646                 struct object *real_obj;
647                 real_obj = deref_tag(the_repository, list->objects[i].item,
648                                      NULL, 0);
649
650                 /* load the gitmodules file for this rev */
651                 if (recurse_submodules) {
652                         submodule_free(the_repository);
653                         gitmodules_config_oid(&real_obj->oid);
654                 }
655                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name,
656                                 list->objects[i].path)) {
657                         hit = 1;
658                         if (opt->status_only)
659                                 break;
660                 }
661         }
662         return hit;
663 }
664
665 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
666                           int exc_std, int use_index)
667 {
668         struct dir_struct dir;
669         int i, hit = 0;
670
671         memset(&dir, 0, sizeof(dir));
672         if (!use_index)
673                 dir.flags |= DIR_NO_GITLINKS;
674         if (exc_std)
675                 setup_standard_excludes(&dir);
676
677         fill_directory(&dir, &the_index, pathspec);
678         for (i = 0; i < dir.nr; i++) {
679                 if (!dir_path_match(&the_index, dir.entries[i], pathspec, 0, NULL))
680                         continue;
681                 hit |= grep_file(opt, dir.entries[i]->name);
682                 if (hit && opt->status_only)
683                         break;
684         }
685         return hit;
686 }
687
688 static int context_callback(const struct option *opt, const char *arg,
689                             int unset)
690 {
691         struct grep_opt *grep_opt = opt->value;
692         int value;
693         const char *endp;
694
695         if (unset) {
696                 grep_opt->pre_context = grep_opt->post_context = 0;
697                 return 0;
698         }
699         value = strtol(arg, (char **)&endp, 10);
700         if (*endp) {
701                 return error(_("switch `%c' expects a numerical value"),
702                              opt->short_name);
703         }
704         grep_opt->pre_context = grep_opt->post_context = value;
705         return 0;
706 }
707
708 static int file_callback(const struct option *opt, const char *arg, int unset)
709 {
710         struct grep_opt *grep_opt = opt->value;
711         int from_stdin;
712         FILE *patterns;
713         int lno = 0;
714         struct strbuf sb = STRBUF_INIT;
715
716         BUG_ON_OPT_NEG(unset);
717
718         from_stdin = !strcmp(arg, "-");
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         BUG_ON_OPT_NEG(unset);
740         BUG_ON_OPT_ARG(arg);
741         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
742         return 0;
743 }
744
745 static int and_callback(const struct option *opt, const char *arg, int unset)
746 {
747         struct grep_opt *grep_opt = opt->value;
748         BUG_ON_OPT_NEG(unset);
749         BUG_ON_OPT_ARG(arg);
750         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
751         return 0;
752 }
753
754 static int open_callback(const struct option *opt, const char *arg, int unset)
755 {
756         struct grep_opt *grep_opt = opt->value;
757         BUG_ON_OPT_NEG(unset);
758         BUG_ON_OPT_ARG(arg);
759         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
760         return 0;
761 }
762
763 static int close_callback(const struct option *opt, const char *arg, int unset)
764 {
765         struct grep_opt *grep_opt = opt->value;
766         BUG_ON_OPT_NEG(unset);
767         BUG_ON_OPT_ARG(arg);
768         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
769         return 0;
770 }
771
772 static int pattern_callback(const struct option *opt, const char *arg,
773                             int unset)
774 {
775         struct grep_opt *grep_opt = opt->value;
776         BUG_ON_OPT_NEG(unset);
777         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
778         return 0;
779 }
780
781 int cmd_grep(int argc, const char **argv, const char *prefix)
782 {
783         int hit = 0;
784         int cached = 0, untracked = 0, opt_exclude = -1;
785         int seen_dashdash = 0;
786         int external_grep_allowed__ignored;
787         const char *show_in_pager = NULL, *default_pager = "dummy";
788         struct grep_opt opt;
789         struct object_array list = OBJECT_ARRAY_INIT;
790         struct pathspec pathspec;
791         struct string_list path_list = STRING_LIST_INIT_NODUP;
792         int i;
793         int dummy;
794         int use_index = 1;
795         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
796         int allow_revs;
797
798         struct option options[] = {
799                 OPT_BOOL(0, "cached", &cached,
800                         N_("search in index instead of in the work tree")),
801                 OPT_NEGBIT(0, "no-index", &use_index,
802                          N_("find in contents not managed by git"), 1),
803                 OPT_BOOL(0, "untracked", &untracked,
804                         N_("search in both tracked and untracked files")),
805                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
806                             N_("ignore files specified via '.gitignore'"), 1),
807                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
808                          N_("recursively search in each submodule")),
809                 OPT_GROUP(""),
810                 OPT_BOOL('v', "invert-match", &opt.invert,
811                         N_("show non-matching lines")),
812                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
813                         N_("case insensitive matching")),
814                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
815                         N_("match patterns only at word boundaries")),
816                 OPT_SET_INT('a', "text", &opt.binary,
817                         N_("process binary files as text"), GREP_BINARY_TEXT),
818                 OPT_SET_INT('I', NULL, &opt.binary,
819                         N_("don't match patterns in binary files"),
820                         GREP_BINARY_NOMATCH),
821                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
822                          N_("process binary files with textconv filters")),
823                 OPT_SET_INT('r', "recursive", &opt.max_depth,
824                             N_("search in subdirectories (default)"), -1),
825                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
826                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
827                         NULL, 1 },
828                 OPT_GROUP(""),
829                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
830                             N_("use extended POSIX regular expressions"),
831                             GREP_PATTERN_TYPE_ERE),
832                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
833                             N_("use basic POSIX regular expressions (default)"),
834                             GREP_PATTERN_TYPE_BRE),
835                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
836                             N_("interpret patterns as fixed strings"),
837                             GREP_PATTERN_TYPE_FIXED),
838                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
839                             N_("use Perl-compatible regular expressions"),
840                             GREP_PATTERN_TYPE_PCRE),
841                 OPT_GROUP(""),
842                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
843                 OPT_BOOL(0, "column", &opt.columnnum, N_("show column number of first match")),
844                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
845                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
846                 OPT_NEGBIT(0, "full-name", &opt.relative,
847                         N_("show filenames relative to top directory"), 1),
848                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
849                         N_("show only filenames instead of matching lines")),
850                 OPT_BOOL(0, "name-only", &opt.name_only,
851                         N_("synonym for --files-with-matches")),
852                 OPT_BOOL('L', "files-without-match",
853                         &opt.unmatch_name_only,
854                         N_("show only the names of files without match")),
855                 OPT_BOOL_F('z', "null", &opt.null_following_name,
856                            N_("print NUL after filenames"),
857                            PARSE_OPT_NOCOMPLETE),
858                 OPT_BOOL('o', "only-matching", &opt.only_matching,
859                         N_("show only matching parts of a line")),
860                 OPT_BOOL('c', "count", &opt.count,
861                         N_("show the number of matches instead of matching lines")),
862                 OPT__COLOR(&opt.color, N_("highlight matches")),
863                 OPT_BOOL(0, "break", &opt.file_break,
864                         N_("print empty line between matches from different files")),
865                 OPT_BOOL(0, "heading", &opt.heading,
866                         N_("show filename only once above matches from same file")),
867                 OPT_GROUP(""),
868                 OPT_CALLBACK('C', "context", &opt, N_("n"),
869                         N_("show <n> context lines before and after matches"),
870                         context_callback),
871                 OPT_INTEGER('B', "before-context", &opt.pre_context,
872                         N_("show <n> context lines before matches")),
873                 OPT_INTEGER('A', "after-context", &opt.post_context,
874                         N_("show <n> context lines after matches")),
875                 OPT_INTEGER(0, "threads", &num_threads,
876                         N_("use <n> worker threads")),
877                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
878                         context_callback),
879                 OPT_BOOL('p', "show-function", &opt.funcname,
880                         N_("show a line with the function name before matches")),
881                 OPT_BOOL('W', "function-context", &opt.funcbody,
882                         N_("show the surrounding function")),
883                 OPT_GROUP(""),
884                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
885                         N_("read patterns from file"), file_callback),
886                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
887                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
888                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
889                   N_("combine patterns specified with -e"),
890                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
891                 OPT_BOOL(0, "or", &dummy, ""),
892                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
893                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
894                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
895                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
896                   open_callback },
897                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
898                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
899                   close_callback },
900                 OPT__QUIET(&opt.status_only,
901                            N_("indicate hit with exit status without output")),
902                 OPT_BOOL(0, "all-match", &opt.all_match,
903                         N_("show only matches from files that match all patterns")),
904                 OPT_SET_INT_F(0, "debug", &opt.debug,
905                               N_("show parse tree for grep expression"),
906                               1, PARSE_OPT_HIDDEN),
907                 OPT_GROUP(""),
908                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
909                         N_("pager"), N_("show matching files in the pager"),
910                         PARSE_OPT_OPTARG | PARSE_OPT_NOCOMPLETE,
911                         NULL, (intptr_t)default_pager },
912                 OPT_BOOL_F(0, "ext-grep", &external_grep_allowed__ignored,
913                            N_("allow calling of grep(1) (ignored by this build)"),
914                            PARSE_OPT_NOCOMPLETE),
915                 OPT_END()
916         };
917
918         init_grep_defaults(the_repository);
919         git_config(grep_cmd_config, NULL);
920         grep_init(&opt, the_repository, prefix);
921
922         /*
923          * If there is no -- then the paths must exist in the working
924          * tree.  If there is no explicit pattern specified with -e or
925          * -f, we take the first unrecognized non option to be the
926          * pattern, but then what follows it must be zero or more
927          * valid refs up to the -- (if exists), and then existing
928          * paths.  If there is an explicit pattern, then the first
929          * unrecognized non option is the beginning of the refs list
930          * that continues up to the -- (if exists), and then paths.
931          */
932         argc = parse_options(argc, argv, prefix, options, grep_usage,
933                              PARSE_OPT_KEEP_DASHDASH |
934                              PARSE_OPT_STOP_AT_NON_OPTION);
935         grep_commit_pattern_type(pattern_type_arg, &opt);
936
937         if (use_index && !startup_info->have_repository) {
938                 int fallback = 0;
939                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
940                 if (fallback)
941                         use_index = 0;
942                 else
943                         /* die the same way as if we did it at the beginning */
944                         setup_git_directory();
945         }
946
947         /*
948          * skip a -- separator; we know it cannot be
949          * separating revisions from pathnames if
950          * we haven't even had any patterns yet
951          */
952         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
953                 argv++;
954                 argc--;
955         }
956
957         /* First unrecognized non-option token */
958         if (argc > 0 && !opt.pattern_list) {
959                 append_grep_pattern(&opt, argv[0], "command line", 0,
960                                     GREP_PATTERN);
961                 argv++;
962                 argc--;
963         }
964
965         if (show_in_pager == default_pager)
966                 show_in_pager = git_pager(1);
967         if (show_in_pager) {
968                 opt.color = 0;
969                 opt.name_only = 1;
970                 opt.null_following_name = 1;
971                 opt.output_priv = &path_list;
972                 opt.output = append_path;
973                 string_list_append(&path_list, show_in_pager);
974         }
975
976         if (!opt.pattern_list)
977                 die(_("no pattern given"));
978
979         /* --only-matching has no effect with --invert. */
980         if (opt.invert)
981                 opt.only_matching = 0;
982
983         /*
984          * We have to find "--" in a separate pass, because its presence
985          * influences how we will parse arguments that come before it.
986          */
987         for (i = 0; i < argc; i++) {
988                 if (!strcmp(argv[i], "--")) {
989                         seen_dashdash = 1;
990                         break;
991                 }
992         }
993
994         /*
995          * Resolve any rev arguments. If we have a dashdash, then everything up
996          * to it must resolve as a rev. If not, then we stop at the first
997          * non-rev and assume everything else is a path.
998          */
999         allow_revs = use_index && !untracked;
1000         for (i = 0; i < argc; i++) {
1001                 const char *arg = argv[i];
1002                 struct object_id oid;
1003                 struct object_context oc;
1004                 struct object *object;
1005
1006                 if (!strcmp(arg, "--")) {
1007                         i++;
1008                         break;
1009                 }
1010
1011                 if (!allow_revs) {
1012                         if (seen_dashdash)
1013                                 die(_("--no-index or --untracked cannot be used with revs"));
1014                         break;
1015                 }
1016
1017                 if (get_oid_with_context(arg, GET_OID_RECORD_PATH,
1018                                          &oid, &oc)) {
1019                         if (seen_dashdash)
1020                                 die(_("unable to resolve revision: %s"), arg);
1021                         break;
1022                 }
1023
1024                 object = parse_object_or_die(&oid, arg);
1025                 if (!seen_dashdash)
1026                         verify_non_filename(prefix, arg);
1027                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1028                 free(oc.path);
1029         }
1030
1031         /*
1032          * Anything left over is presumed to be a path. But in the non-dashdash
1033          * "do what I mean" case, we verify and complain when that isn't true.
1034          */
1035         if (!seen_dashdash) {
1036                 int j;
1037                 for (j = i; j < argc; j++)
1038                         verify_filename(prefix, argv[j], j == i && allow_revs);
1039         }
1040
1041         parse_pathspec(&pathspec, 0,
1042                        PATHSPEC_PREFER_CWD |
1043                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1044                        prefix, argv + i);
1045         pathspec.max_depth = opt.max_depth;
1046         pathspec.recursive = 1;
1047         pathspec.recurse_submodules = !!recurse_submodules;
1048
1049         if (list.nr || cached || show_in_pager) {
1050                 if (num_threads > 1)
1051                         warning(_("invalid option combination, ignoring --threads"));
1052                 num_threads = 1;
1053         } else if (!HAVE_THREADS && num_threads > 1) {
1054                 warning(_("no threads support, ignoring --threads"));
1055                 num_threads = 1;
1056         } else if (num_threads < 0)
1057                 die(_("invalid number of threads specified (%d)"), num_threads);
1058         else if (num_threads == 0)
1059                 num_threads = HAVE_THREADS ? GREP_NUM_THREADS_DEFAULT : 1;
1060
1061         if (num_threads > 1) {
1062                 if (!HAVE_THREADS)
1063                         BUG("Somebody got num_threads calculation wrong!");
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         } else {
1070                 /*
1071                  * The compiled patterns on the main path are only
1072                  * used when not using threading. Otherwise
1073                  * start_threads() above calls compile_grep_patterns()
1074                  * for each thread.
1075                  */
1076                 compile_grep_patterns(&opt);
1077         }
1078
1079         if (show_in_pager && (cached || list.nr))
1080                 die(_("--open-files-in-pager only works on the worktree"));
1081
1082         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1083                 const char *pager = path_list.items[0].string;
1084                 int len = strlen(pager);
1085
1086                 if (len > 4 && is_dir_sep(pager[len - 5]))
1087                         pager += len - 4;
1088
1089                 if (opt.ignore_case && !strcmp("less", pager))
1090                         string_list_append(&path_list, "-I");
1091
1092                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1093                         struct strbuf buf = STRBUF_INIT;
1094                         strbuf_addf(&buf, "+/%s%s",
1095                                         strcmp("less", pager) ? "" : "*",
1096                                         opt.pattern_list->pattern);
1097                         string_list_append(&path_list, buf.buf);
1098                         strbuf_detach(&buf, NULL);
1099                 }
1100         }
1101
1102         if (recurse_submodules && (!use_index || untracked))
1103                 die(_("option not supported with --recurse-submodules"));
1104
1105         if (!show_in_pager && !opt.status_only)
1106                 setup_pager();
1107
1108         if (!use_index && (untracked || cached))
1109                 die(_("--cached or --untracked cannot be used with --no-index"));
1110
1111         if (!use_index || untracked) {
1112                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1113                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1114         } else if (0 <= opt_exclude) {
1115                 die(_("--[no-]exclude-standard cannot be used for tracked contents"));
1116         } else if (!list.nr) {
1117                 if (!cached)
1118                         setup_work_tree();
1119
1120                 hit = grep_cache(&opt, the_repository, &pathspec, cached);
1121         } else {
1122                 if (cached)
1123                         die(_("both --cached and trees are given"));
1124
1125                 hit = grep_objects(&opt, &pathspec, &list);
1126         }
1127
1128         if (num_threads > 1)
1129                 hit |= wait_all();
1130         if (hit && show_in_pager)
1131                 run_pager(&opt, prefix);
1132         clear_pathspec(&pathspec);
1133         free_grep_patterns(&opt);
1134         return !hit;
1135 }