Merge branch 'lt/date-human'
[git] / builtin / grep.c
1 /*
2  * Builtin "git grep"
3  *
4  * Copyright (c) 2006 Junio C Hamano
5  */
6 #define USE_THE_INDEX_COMPATIBILITY_MACROS
7 #include "cache.h"
8 #include "repository.h"
9 #include "config.h"
10 #include "blob.h"
11 #include "tree.h"
12 #include "commit.h"
13 #include "tag.h"
14 #include "tree-walk.h"
15 #include "builtin.h"
16 #include "parse-options.h"
17 #include "string-list.h"
18 #include "run-command.h"
19 #include "userdiff.h"
20 #include "grep.h"
21 #include "quote.h"
22 #include "dir.h"
23 #include "pathspec.h"
24 #include "submodule.h"
25 #include "submodule-config.h"
26 #include "object-store.h"
27
28 static char const * const grep_usage[] = {
29         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
30         NULL
31 };
32
33 static int recurse_submodules;
34
35 #define GREP_NUM_THREADS_DEFAULT 8
36 static int num_threads;
37
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         pthread_mutex_lock(&grep_mutex);
74 }
75
76 static inline void grep_unlock(void)
77 {
78         pthread_mutex_unlock(&grep_mutex);
79 }
80
81 /* Signalled when a new work_item is added to todo. */
82 static pthread_cond_t cond_add;
83
84 /* Signalled when the result from one work_item is written to
85  * stdout.
86  */
87 static pthread_cond_t cond_write;
88
89 /* Signalled when we are finished with everything. */
90 static pthread_cond_t cond_result;
91
92 static int skip_first_line;
93
94 static void add_work(struct grep_opt *opt, const struct grep_source *gs)
95 {
96         grep_lock();
97
98         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
99                 pthread_cond_wait(&cond_write, &grep_mutex);
100         }
101
102         todo[todo_end].source = *gs;
103         if (opt->binary != GREP_BINARY_TEXT)
104                 grep_source_load_driver(&todo[todo_end].source,
105                                         opt->repo->index);
106         todo[todo_end].done = 0;
107         strbuf_reset(&todo[todo_end].out);
108         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
109
110         pthread_cond_signal(&cond_add);
111         grep_unlock();
112 }
113
114 static struct work_item *get_work(void)
115 {
116         struct work_item *ret;
117
118         grep_lock();
119         while (todo_start == todo_end && !all_work_added) {
120                 pthread_cond_wait(&cond_add, &grep_mutex);
121         }
122
123         if (todo_start == todo_end && all_work_added) {
124                 ret = NULL;
125         } else {
126                 ret = &todo[todo_start];
127                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
128         }
129         grep_unlock();
130         return ret;
131 }
132
133 static void work_done(struct work_item *w)
134 {
135         int old_done;
136
137         grep_lock();
138         w->done = 1;
139         old_done = todo_done;
140         for(; todo[todo_done].done && todo_done != todo_start;
141             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
142                 w = &todo[todo_done];
143                 if (w->out.len) {
144                         const char *p = w->out.buf;
145                         size_t len = w->out.len;
146
147                         /* Skip the leading hunk mark of the first file. */
148                         if (skip_first_line) {
149                                 while (len) {
150                                         len--;
151                                         if (*p++ == '\n')
152                                                 break;
153                                 }
154                                 skip_first_line = 0;
155                         }
156
157                         write_or_die(1, p, len);
158                 }
159                 grep_source_clear(&w->source);
160         }
161
162         if (old_done != todo_done)
163                 pthread_cond_signal(&cond_write);
164
165         if (all_work_added && todo_done == todo_end)
166                 pthread_cond_signal(&cond_result);
167
168         grep_unlock();
169 }
170
171 static void *run(void *arg)
172 {
173         int hit = 0;
174         struct grep_opt *opt = arg;
175
176         while (1) {
177                 struct work_item *w = get_work();
178                 if (!w)
179                         break;
180
181                 opt->output_priv = w;
182                 hit |= grep_source(opt, &w->source);
183                 grep_source_clear_data(&w->source);
184                 work_done(w);
185         }
186         free_grep_patterns(arg);
187         free(arg);
188
189         return (void*) (intptr_t) hit;
190 }
191
192 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
193 {
194         struct work_item *w = opt->output_priv;
195         strbuf_add(&w->out, buf, size);
196 }
197
198 static void start_threads(struct grep_opt *opt)
199 {
200         int i;
201
202         pthread_mutex_init(&grep_mutex, NULL);
203         pthread_mutex_init(&grep_read_mutex, NULL);
204         pthread_mutex_init(&grep_attr_mutex, NULL);
205         pthread_cond_init(&cond_add, NULL);
206         pthread_cond_init(&cond_write, NULL);
207         pthread_cond_init(&cond_result, NULL);
208         grep_use_locks = 1;
209
210         for (i = 0; i < ARRAY_SIZE(todo); i++) {
211                 strbuf_init(&todo[i].out, 0);
212         }
213
214         threads = xcalloc(num_threads, sizeof(*threads));
215         for (i = 0; i < num_threads; i++) {
216                 int err;
217                 struct grep_opt *o = grep_opt_dup(opt);
218                 o->output = strbuf_out;
219                 if (i)
220                         o->debug = 0;
221                 compile_grep_patterns(o);
222                 err = pthread_create(&threads[i], NULL, run, o);
223
224                 if (err)
225                         die(_("grep: failed to create thread: %s"),
226                             strerror(err));
227         }
228 }
229
230 static int wait_all(void)
231 {
232         int hit = 0;
233         int i;
234
235         if (!HAVE_THREADS)
236                 BUG("Never call this function unless you have started threads");
237
238         grep_lock();
239         all_work_added = 1;
240
241         /* Wait until all work is done. */
242         while (todo_done != todo_end)
243                 pthread_cond_wait(&cond_result, &grep_mutex);
244
245         /* Wake up all the consumer threads so they can see that there
246          * is no more work to do.
247          */
248         pthread_cond_broadcast(&cond_add);
249         grep_unlock();
250
251         for (i = 0; i < num_threads; i++) {
252                 void *h;
253                 pthread_join(threads[i], &h);
254                 hit |= (int) (intptr_t) h;
255         }
256
257         free(threads);
258
259         pthread_mutex_destroy(&grep_mutex);
260         pthread_mutex_destroy(&grep_read_mutex);
261         pthread_mutex_destroy(&grep_attr_mutex);
262         pthread_cond_destroy(&cond_add);
263         pthread_cond_destroy(&cond_write);
264         pthread_cond_destroy(&cond_result);
265         grep_use_locks = 0;
266
267         return hit;
268 }
269
270 static int grep_cmd_config(const char *var, const char *value, void *cb)
271 {
272         int st = grep_config(var, value, cb);
273         if (git_color_default_config(var, value, cb) < 0)
274                 st = -1;
275
276         if (!strcmp(var, "grep.threads")) {
277                 num_threads = git_config_int(var, value);
278                 if (num_threads < 0)
279                         die(_("invalid number of threads specified (%d) for %s"),
280                             num_threads, var);
281                 else if (!HAVE_THREADS && num_threads > 1) {
282                         /*
283                          * TRANSLATORS: %s is the configuration
284                          * variable for tweaking threads, currently
285                          * grep.threads
286                          */
287                         warning(_("no threads support, ignoring %s"), var);
288                         num_threads = 1;
289                 }
290         }
291
292         if (!strcmp(var, "submodule.recurse"))
293                 recurse_submodules = git_config_bool(var, value);
294
295         return st;
296 }
297
298 static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
299 {
300         void *data;
301
302         grep_read_lock();
303         data = read_object_file(oid, type, size);
304         grep_read_unlock();
305         return data;
306 }
307
308 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
309                      const char *filename, int tree_name_len,
310                      const char *path)
311 {
312         struct strbuf pathbuf = STRBUF_INIT;
313         struct grep_source gs;
314
315         if (opt->relative && opt->prefix_length) {
316                 quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
317                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
318         } else {
319                 strbuf_addstr(&pathbuf, filename);
320         }
321
322         grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
323         strbuf_release(&pathbuf);
324
325         if (num_threads > 1) {
326                 /*
327                  * add_work() copies gs and thus assumes ownership of
328                  * its fields, so do not call grep_source_clear()
329                  */
330                 add_work(opt, &gs);
331                 return 0;
332         } else {
333                 int hit;
334
335                 hit = grep_source(opt, &gs);
336
337                 grep_source_clear(&gs);
338                 return hit;
339         }
340 }
341
342 static int grep_file(struct grep_opt *opt, const char *filename)
343 {
344         struct strbuf buf = STRBUF_INIT;
345         struct grep_source gs;
346
347         if (opt->relative && opt->prefix_length)
348                 quote_path_relative(filename, opt->prefix, &buf);
349         else
350                 strbuf_addstr(&buf, filename);
351
352         grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
353         strbuf_release(&buf);
354
355         if (num_threads > 1) {
356                 /*
357                  * add_work() copies gs and thus assumes ownership of
358                  * its fields, so do not call grep_source_clear()
359                  */
360                 add_work(opt, &gs);
361                 return 0;
362         } else {
363                 int hit;
364
365                 hit = grep_source(opt, &gs);
366
367                 grep_source_clear(&gs);
368                 return hit;
369         }
370 }
371
372 static void append_path(struct grep_opt *opt, const void *data, size_t len)
373 {
374         struct string_list *path_list = opt->output_priv;
375
376         if (len == 1 && *(const char *)data == '\0')
377                 return;
378         string_list_append(path_list, xstrndup(data, len));
379 }
380
381 static void run_pager(struct grep_opt *opt, const char *prefix)
382 {
383         struct string_list *path_list = opt->output_priv;
384         struct child_process child = CHILD_PROCESS_INIT;
385         int i, status;
386
387         for (i = 0; i < path_list->nr; i++)
388                 argv_array_push(&child.args, path_list->items[i].string);
389         child.dir = prefix;
390         child.use_shell = 1;
391
392         status = run_command(&child);
393         if (status)
394                 exit(status);
395 }
396
397 static int grep_cache(struct grep_opt *opt,
398                       const struct pathspec *pathspec, int cached);
399 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
400                      struct tree_desc *tree, struct strbuf *base, int tn_len,
401                      int check_attr);
402
403 static int grep_submodule(struct grep_opt *opt,
404                           const struct pathspec *pathspec,
405                           const struct object_id *oid,
406                           const char *filename, const char *path)
407 {
408         struct repository subrepo;
409         struct repository *superproject = opt->repo;
410         const struct submodule *sub = submodule_from_path(superproject,
411                                                           &null_oid, path);
412         struct grep_opt subopt;
413         int hit;
414
415         /*
416          * NEEDSWORK: submodules functions need to be protected because they
417          * access the object store via config_from_gitmodules(): the latter
418          * uses get_oid() which, for now, relies on the global the_repository
419          * object.
420          */
421         grep_read_lock();
422
423         if (!is_submodule_active(superproject, path)) {
424                 grep_read_unlock();
425                 return 0;
426         }
427
428         if (repo_submodule_init(&subrepo, superproject, sub)) {
429                 grep_read_unlock();
430                 return 0;
431         }
432
433         repo_read_gitmodules(&subrepo);
434
435         /*
436          * NEEDSWORK: This adds the submodule's object directory to the list of
437          * alternates for the single in-memory object store.  This has some bad
438          * consequences for memory (processed objects will never be freed) and
439          * performance (this increases the number of pack files git has to pay
440          * attention to, to the sum of the number of pack files in all the
441          * repositories processed so far).  This can be removed once the object
442          * store is no longer global and instead is a member of the repository
443          * object.
444          */
445         add_to_alternates_memory(subrepo.objects->odb->path);
446         grep_read_unlock();
447
448         memcpy(&subopt, opt, sizeof(subopt));
449         subopt.repo = &subrepo;
450
451         if (oid) {
452                 struct object *object;
453                 struct tree_desc tree;
454                 void *data;
455                 unsigned long size;
456                 struct strbuf base = STRBUF_INIT;
457
458                 object = parse_object_or_die(oid, oid_to_hex(oid));
459
460                 grep_read_lock();
461                 data = read_object_with_reference(&object->oid, tree_type,
462                                                   &size, NULL);
463                 grep_read_unlock();
464
465                 if (!data)
466                         die(_("unable to read tree (%s)"), oid_to_hex(&object->oid));
467
468                 strbuf_addstr(&base, filename);
469                 strbuf_addch(&base, '/');
470
471                 init_tree_desc(&tree, data, size);
472                 hit = grep_tree(&subopt, pathspec, &tree, &base, base.len,
473                                 object->type == OBJ_COMMIT);
474                 strbuf_release(&base);
475                 free(data);
476         } else {
477                 hit = grep_cache(&subopt, pathspec, 1);
478         }
479
480         repo_clear(&subrepo);
481         return hit;
482 }
483
484 static int grep_cache(struct grep_opt *opt,
485                       const struct pathspec *pathspec, int cached)
486 {
487         struct repository *repo = opt->repo;
488         int hit = 0;
489         int nr;
490         struct strbuf name = STRBUF_INIT;
491         int name_base_len = 0;
492         if (repo->submodule_prefix) {
493                 name_base_len = strlen(repo->submodule_prefix);
494                 strbuf_addstr(&name, repo->submodule_prefix);
495         }
496
497         if (repo_read_index(repo) < 0)
498                 die(_("index file corrupt"));
499
500         for (nr = 0; nr < repo->index->cache_nr; nr++) {
501                 const struct cache_entry *ce = repo->index->cache[nr];
502                 strbuf_setlen(&name, name_base_len);
503                 strbuf_addstr(&name, ce->name);
504
505                 if (S_ISREG(ce->ce_mode) &&
506                     match_pathspec(repo->index, pathspec, name.buf, name.len, 0, NULL,
507                                    S_ISDIR(ce->ce_mode) ||
508                                    S_ISGITLINK(ce->ce_mode))) {
509                         /*
510                          * If CE_VALID is on, we assume worktree file and its
511                          * cache entry are identical, even if worktree file has
512                          * been modified, so use cache version instead
513                          */
514                         if (cached || (ce->ce_flags & CE_VALID) ||
515                             ce_skip_worktree(ce)) {
516                                 if (ce_stage(ce) || ce_intent_to_add(ce))
517                                         continue;
518                                 hit |= grep_oid(opt, &ce->oid, name.buf,
519                                                  0, name.buf);
520                         } else {
521                                 hit |= grep_file(opt, name.buf);
522                         }
523                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
524                            submodule_path_match(repo->index, pathspec, name.buf, NULL)) {
525                         hit |= grep_submodule(opt, pathspec, NULL, ce->name, ce->name);
526                 } else {
527                         continue;
528                 }
529
530                 if (ce_stage(ce)) {
531                         do {
532                                 nr++;
533                         } while (nr < repo->index->cache_nr &&
534                                  !strcmp(ce->name, repo->index->cache[nr]->name));
535                         nr--; /* compensate for loop control */
536                 }
537                 if (hit && opt->status_only)
538                         break;
539         }
540
541         strbuf_release(&name);
542         return hit;
543 }
544
545 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
546                      struct tree_desc *tree, struct strbuf *base, int tn_len,
547                      int check_attr)
548 {
549         struct repository *repo = opt->repo;
550         int hit = 0;
551         enum interesting match = entry_not_interesting;
552         struct name_entry entry;
553         int old_baselen = base->len;
554         struct strbuf name = STRBUF_INIT;
555         int name_base_len = 0;
556         if (repo->submodule_prefix) {
557                 strbuf_addstr(&name, repo->submodule_prefix);
558                 name_base_len = name.len;
559         }
560
561         while (tree_entry(tree, &entry)) {
562                 int te_len = tree_entry_len(&entry);
563
564                 if (match != all_entries_interesting) {
565                         strbuf_addstr(&name, base->buf + tn_len);
566                         match = tree_entry_interesting(repo->index,
567                                                        &entry, &name,
568                                                        0, pathspec);
569                         strbuf_setlen(&name, name_base_len);
570
571                         if (match == all_entries_not_interesting)
572                                 break;
573                         if (match == entry_not_interesting)
574                                 continue;
575                 }
576
577                 strbuf_add(base, entry.path, te_len);
578
579                 if (S_ISREG(entry.mode)) {
580                         hit |= grep_oid(opt, &entry.oid, base->buf, tn_len,
581                                          check_attr ? base->buf + tn_len : NULL);
582                 } else if (S_ISDIR(entry.mode)) {
583                         enum object_type type;
584                         struct tree_desc sub;
585                         void *data;
586                         unsigned long size;
587
588                         data = lock_and_read_oid_file(&entry.oid, &type, &size);
589                         if (!data)
590                                 die(_("unable to read tree (%s)"),
591                                     oid_to_hex(&entry.oid));
592
593                         strbuf_addch(base, '/');
594                         init_tree_desc(&sub, data, size);
595                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
596                                          check_attr);
597                         free(data);
598                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
599                         hit |= grep_submodule(opt, pathspec, &entry.oid,
600                                               base->buf, base->buf + tn_len);
601                 }
602
603                 strbuf_setlen(base, old_baselen);
604
605                 if (hit && opt->status_only)
606                         break;
607         }
608
609         strbuf_release(&name);
610         return hit;
611 }
612
613 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
614                        struct object *obj, const char *name, const char *path)
615 {
616         if (obj->type == OBJ_BLOB)
617                 return grep_oid(opt, &obj->oid, name, 0, path);
618         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
619                 struct tree_desc tree;
620                 void *data;
621                 unsigned long size;
622                 struct strbuf base;
623                 int hit, len;
624
625                 grep_read_lock();
626                 data = read_object_with_reference(&obj->oid, tree_type,
627                                                   &size, NULL);
628                 grep_read_unlock();
629
630                 if (!data)
631                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
632
633                 len = name ? strlen(name) : 0;
634                 strbuf_init(&base, PATH_MAX + len + 1);
635                 if (len) {
636                         strbuf_add(&base, name, len);
637                         strbuf_addch(&base, ':');
638                 }
639                 init_tree_desc(&tree, data, size);
640                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
641                                 obj->type == OBJ_COMMIT);
642                 strbuf_release(&base);
643                 free(data);
644                 return hit;
645         }
646         die(_("unable to grep from object of type %s"), type_name(obj->type));
647 }
648
649 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
650                         const struct object_array *list)
651 {
652         unsigned int i;
653         int hit = 0;
654         const unsigned int nr = list->nr;
655
656         for (i = 0; i < nr; i++) {
657                 struct object *real_obj;
658                 real_obj = deref_tag(opt->repo, list->objects[i].item,
659                                      NULL, 0);
660
661                 /* load the gitmodules file for this rev */
662                 if (recurse_submodules) {
663                         submodule_free(opt->repo);
664                         gitmodules_config_oid(&real_obj->oid);
665                 }
666                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name,
667                                 list->objects[i].path)) {
668                         hit = 1;
669                         if (opt->status_only)
670                                 break;
671                 }
672         }
673         return hit;
674 }
675
676 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
677                           int exc_std, int use_index)
678 {
679         struct dir_struct dir;
680         int i, hit = 0;
681
682         memset(&dir, 0, sizeof(dir));
683         if (!use_index)
684                 dir.flags |= DIR_NO_GITLINKS;
685         if (exc_std)
686                 setup_standard_excludes(&dir);
687
688         fill_directory(&dir, opt->repo->index, pathspec);
689         for (i = 0; i < dir.nr; i++) {
690                 if (!dir_path_match(opt->repo->index, dir.entries[i], pathspec, 0, NULL))
691                         continue;
692                 hit |= grep_file(opt, dir.entries[i]->name);
693                 if (hit && opt->status_only)
694                         break;
695         }
696         return hit;
697 }
698
699 static int context_callback(const struct option *opt, const char *arg,
700                             int unset)
701 {
702         struct grep_opt *grep_opt = opt->value;
703         int value;
704         const char *endp;
705
706         if (unset) {
707                 grep_opt->pre_context = grep_opt->post_context = 0;
708                 return 0;
709         }
710         value = strtol(arg, (char **)&endp, 10);
711         if (*endp) {
712                 return error(_("switch `%c' expects a numerical value"),
713                              opt->short_name);
714         }
715         grep_opt->pre_context = grep_opt->post_context = value;
716         return 0;
717 }
718
719 static int file_callback(const struct option *opt, const char *arg, int unset)
720 {
721         struct grep_opt *grep_opt = opt->value;
722         int from_stdin;
723         FILE *patterns;
724         int lno = 0;
725         struct strbuf sb = STRBUF_INIT;
726
727         BUG_ON_OPT_NEG(unset);
728
729         from_stdin = !strcmp(arg, "-");
730         patterns = from_stdin ? stdin : fopen(arg, "r");
731         if (!patterns)
732                 die_errno(_("cannot open '%s'"), arg);
733         while (strbuf_getline(&sb, patterns) == 0) {
734                 /* ignore empty line like grep does */
735                 if (sb.len == 0)
736                         continue;
737
738                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
739                                 GREP_PATTERN);
740         }
741         if (!from_stdin)
742                 fclose(patterns);
743         strbuf_release(&sb);
744         return 0;
745 }
746
747 static int not_callback(const struct option *opt, const char *arg, int unset)
748 {
749         struct grep_opt *grep_opt = opt->value;
750         BUG_ON_OPT_NEG(unset);
751         BUG_ON_OPT_ARG(arg);
752         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
753         return 0;
754 }
755
756 static int and_callback(const struct option *opt, const char *arg, int unset)
757 {
758         struct grep_opt *grep_opt = opt->value;
759         BUG_ON_OPT_NEG(unset);
760         BUG_ON_OPT_ARG(arg);
761         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
762         return 0;
763 }
764
765 static int open_callback(const struct option *opt, const char *arg, int unset)
766 {
767         struct grep_opt *grep_opt = opt->value;
768         BUG_ON_OPT_NEG(unset);
769         BUG_ON_OPT_ARG(arg);
770         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
771         return 0;
772 }
773
774 static int close_callback(const struct option *opt, const char *arg, int unset)
775 {
776         struct grep_opt *grep_opt = opt->value;
777         BUG_ON_OPT_NEG(unset);
778         BUG_ON_OPT_ARG(arg);
779         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
780         return 0;
781 }
782
783 static int pattern_callback(const struct option *opt, const char *arg,
784                             int unset)
785 {
786         struct grep_opt *grep_opt = opt->value;
787         BUG_ON_OPT_NEG(unset);
788         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
789         return 0;
790 }
791
792 int cmd_grep(int argc, const char **argv, const char *prefix)
793 {
794         int hit = 0;
795         int cached = 0, untracked = 0, opt_exclude = -1;
796         int seen_dashdash = 0;
797         int external_grep_allowed__ignored;
798         const char *show_in_pager = NULL, *default_pager = "dummy";
799         struct grep_opt opt;
800         struct object_array list = OBJECT_ARRAY_INIT;
801         struct pathspec pathspec;
802         struct string_list path_list = STRING_LIST_INIT_NODUP;
803         int i;
804         int dummy;
805         int use_index = 1;
806         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
807         int allow_revs;
808
809         struct option options[] = {
810                 OPT_BOOL(0, "cached", &cached,
811                         N_("search in index instead of in the work tree")),
812                 OPT_NEGBIT(0, "no-index", &use_index,
813                          N_("find in contents not managed by git"), 1),
814                 OPT_BOOL(0, "untracked", &untracked,
815                         N_("search in both tracked and untracked files")),
816                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
817                             N_("ignore files specified via '.gitignore'"), 1),
818                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
819                          N_("recursively search in each submodule")),
820                 OPT_GROUP(""),
821                 OPT_BOOL('v', "invert-match", &opt.invert,
822                         N_("show non-matching lines")),
823                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
824                         N_("case insensitive matching")),
825                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
826                         N_("match patterns only at word boundaries")),
827                 OPT_SET_INT('a', "text", &opt.binary,
828                         N_("process binary files as text"), GREP_BINARY_TEXT),
829                 OPT_SET_INT('I', NULL, &opt.binary,
830                         N_("don't match patterns in binary files"),
831                         GREP_BINARY_NOMATCH),
832                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
833                          N_("process binary files with textconv filters")),
834                 OPT_SET_INT('r', "recursive", &opt.max_depth,
835                             N_("search in subdirectories (default)"), -1),
836                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
837                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
838                         NULL, 1 },
839                 OPT_GROUP(""),
840                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
841                             N_("use extended POSIX regular expressions"),
842                             GREP_PATTERN_TYPE_ERE),
843                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
844                             N_("use basic POSIX regular expressions (default)"),
845                             GREP_PATTERN_TYPE_BRE),
846                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
847                             N_("interpret patterns as fixed strings"),
848                             GREP_PATTERN_TYPE_FIXED),
849                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
850                             N_("use Perl-compatible regular expressions"),
851                             GREP_PATTERN_TYPE_PCRE),
852                 OPT_GROUP(""),
853                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
854                 OPT_BOOL(0, "column", &opt.columnnum, N_("show column number of first match")),
855                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
856                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
857                 OPT_NEGBIT(0, "full-name", &opt.relative,
858                         N_("show filenames relative to top directory"), 1),
859                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
860                         N_("show only filenames instead of matching lines")),
861                 OPT_BOOL(0, "name-only", &opt.name_only,
862                         N_("synonym for --files-with-matches")),
863                 OPT_BOOL('L', "files-without-match",
864                         &opt.unmatch_name_only,
865                         N_("show only the names of files without match")),
866                 OPT_BOOL_F('z', "null", &opt.null_following_name,
867                            N_("print NUL after filenames"),
868                            PARSE_OPT_NOCOMPLETE),
869                 OPT_BOOL('o', "only-matching", &opt.only_matching,
870                         N_("show only matching parts of a line")),
871                 OPT_BOOL('c', "count", &opt.count,
872                         N_("show the number of matches instead of matching lines")),
873                 OPT__COLOR(&opt.color, N_("highlight matches")),
874                 OPT_BOOL(0, "break", &opt.file_break,
875                         N_("print empty line between matches from different files")),
876                 OPT_BOOL(0, "heading", &opt.heading,
877                         N_("show filename only once above matches from same file")),
878                 OPT_GROUP(""),
879                 OPT_CALLBACK('C', "context", &opt, N_("n"),
880                         N_("show <n> context lines before and after matches"),
881                         context_callback),
882                 OPT_INTEGER('B', "before-context", &opt.pre_context,
883                         N_("show <n> context lines before matches")),
884                 OPT_INTEGER('A', "after-context", &opt.post_context,
885                         N_("show <n> context lines after matches")),
886                 OPT_INTEGER(0, "threads", &num_threads,
887                         N_("use <n> worker threads")),
888                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
889                         context_callback),
890                 OPT_BOOL('p', "show-function", &opt.funcname,
891                         N_("show a line with the function name before matches")),
892                 OPT_BOOL('W', "function-context", &opt.funcbody,
893                         N_("show the surrounding function")),
894                 OPT_GROUP(""),
895                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
896                         N_("read patterns from file"), file_callback),
897                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
898                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
899                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
900                   N_("combine patterns specified with -e"),
901                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
902                 OPT_BOOL(0, "or", &dummy, ""),
903                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
904                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
905                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
906                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
907                   open_callback },
908                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
909                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
910                   close_callback },
911                 OPT__QUIET(&opt.status_only,
912                            N_("indicate hit with exit status without output")),
913                 OPT_BOOL(0, "all-match", &opt.all_match,
914                         N_("show only matches from files that match all patterns")),
915                 OPT_SET_INT_F(0, "debug", &opt.debug,
916                               N_("show parse tree for grep expression"),
917                               1, PARSE_OPT_HIDDEN),
918                 OPT_GROUP(""),
919                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
920                         N_("pager"), N_("show matching files in the pager"),
921                         PARSE_OPT_OPTARG | PARSE_OPT_NOCOMPLETE,
922                         NULL, (intptr_t)default_pager },
923                 OPT_BOOL_F(0, "ext-grep", &external_grep_allowed__ignored,
924                            N_("allow calling of grep(1) (ignored by this build)"),
925                            PARSE_OPT_NOCOMPLETE),
926                 OPT_END()
927         };
928
929         init_grep_defaults(the_repository);
930         git_config(grep_cmd_config, NULL);
931         grep_init(&opt, the_repository, prefix);
932
933         /*
934          * If there is no -- then the paths must exist in the working
935          * tree.  If there is no explicit pattern specified with -e or
936          * -f, we take the first unrecognized non option to be the
937          * pattern, but then what follows it must be zero or more
938          * valid refs up to the -- (if exists), and then existing
939          * paths.  If there is an explicit pattern, then the first
940          * unrecognized non option is the beginning of the refs list
941          * that continues up to the -- (if exists), and then paths.
942          */
943         argc = parse_options(argc, argv, prefix, options, grep_usage,
944                              PARSE_OPT_KEEP_DASHDASH |
945                              PARSE_OPT_STOP_AT_NON_OPTION);
946         grep_commit_pattern_type(pattern_type_arg, &opt);
947
948         if (use_index && !startup_info->have_repository) {
949                 int fallback = 0;
950                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
951                 if (fallback)
952                         use_index = 0;
953                 else
954                         /* die the same way as if we did it at the beginning */
955                         setup_git_directory();
956         }
957
958         /*
959          * skip a -- separator; we know it cannot be
960          * separating revisions from pathnames if
961          * we haven't even had any patterns yet
962          */
963         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
964                 argv++;
965                 argc--;
966         }
967
968         /* First unrecognized non-option token */
969         if (argc > 0 && !opt.pattern_list) {
970                 append_grep_pattern(&opt, argv[0], "command line", 0,
971                                     GREP_PATTERN);
972                 argv++;
973                 argc--;
974         }
975
976         if (show_in_pager == default_pager)
977                 show_in_pager = git_pager(1);
978         if (show_in_pager) {
979                 opt.color = 0;
980                 opt.name_only = 1;
981                 opt.null_following_name = 1;
982                 opt.output_priv = &path_list;
983                 opt.output = append_path;
984                 string_list_append(&path_list, show_in_pager);
985         }
986
987         if (!opt.pattern_list)
988                 die(_("no pattern given"));
989
990         /* --only-matching has no effect with --invert. */
991         if (opt.invert)
992                 opt.only_matching = 0;
993
994         /*
995          * We have to find "--" in a separate pass, because its presence
996          * influences how we will parse arguments that come before it.
997          */
998         for (i = 0; i < argc; i++) {
999                 if (!strcmp(argv[i], "--")) {
1000                         seen_dashdash = 1;
1001                         break;
1002                 }
1003         }
1004
1005         /*
1006          * Resolve any rev arguments. If we have a dashdash, then everything up
1007          * to it must resolve as a rev. If not, then we stop at the first
1008          * non-rev and assume everything else is a path.
1009          */
1010         allow_revs = use_index && !untracked;
1011         for (i = 0; i < argc; i++) {
1012                 const char *arg = argv[i];
1013                 struct object_id oid;
1014                 struct object_context oc;
1015                 struct object *object;
1016
1017                 if (!strcmp(arg, "--")) {
1018                         i++;
1019                         break;
1020                 }
1021
1022                 if (!allow_revs) {
1023                         if (seen_dashdash)
1024                                 die(_("--no-index or --untracked cannot be used with revs"));
1025                         break;
1026                 }
1027
1028                 if (get_oid_with_context(the_repository, arg,
1029                                          GET_OID_RECORD_PATH,
1030                                          &oid, &oc)) {
1031                         if (seen_dashdash)
1032                                 die(_("unable to resolve revision: %s"), arg);
1033                         break;
1034                 }
1035
1036                 object = parse_object_or_die(&oid, arg);
1037                 if (!seen_dashdash)
1038                         verify_non_filename(prefix, arg);
1039                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1040                 free(oc.path);
1041         }
1042
1043         /*
1044          * Anything left over is presumed to be a path. But in the non-dashdash
1045          * "do what I mean" case, we verify and complain when that isn't true.
1046          */
1047         if (!seen_dashdash) {
1048                 int j;
1049                 for (j = i; j < argc; j++)
1050                         verify_filename(prefix, argv[j], j == i && allow_revs);
1051         }
1052
1053         parse_pathspec(&pathspec, 0,
1054                        PATHSPEC_PREFER_CWD |
1055                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1056                        prefix, argv + i);
1057         pathspec.max_depth = opt.max_depth;
1058         pathspec.recursive = 1;
1059         pathspec.recurse_submodules = !!recurse_submodules;
1060
1061         if (list.nr || cached || show_in_pager) {
1062                 if (num_threads > 1)
1063                         warning(_("invalid option combination, ignoring --threads"));
1064                 num_threads = 1;
1065         } else if (!HAVE_THREADS && num_threads > 1) {
1066                 warning(_("no threads support, ignoring --threads"));
1067                 num_threads = 1;
1068         } else if (num_threads < 0)
1069                 die(_("invalid number of threads specified (%d)"), num_threads);
1070         else if (num_threads == 0)
1071                 num_threads = HAVE_THREADS ? GREP_NUM_THREADS_DEFAULT : 1;
1072
1073         if (num_threads > 1) {
1074                 if (!HAVE_THREADS)
1075                         BUG("Somebody got num_threads calculation wrong!");
1076                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1077                     && (opt.pre_context || opt.post_context ||
1078                         opt.file_break || opt.funcbody))
1079                         skip_first_line = 1;
1080                 start_threads(&opt);
1081         } else {
1082                 /*
1083                  * The compiled patterns on the main path are only
1084                  * used when not using threading. Otherwise
1085                  * start_threads() above calls compile_grep_patterns()
1086                  * for each thread.
1087                  */
1088                 compile_grep_patterns(&opt);
1089         }
1090
1091         if (show_in_pager && (cached || list.nr))
1092                 die(_("--open-files-in-pager only works on the worktree"));
1093
1094         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1095                 const char *pager = path_list.items[0].string;
1096                 int len = strlen(pager);
1097
1098                 if (len > 4 && is_dir_sep(pager[len - 5]))
1099                         pager += len - 4;
1100
1101                 if (opt.ignore_case && !strcmp("less", pager))
1102                         string_list_append(&path_list, "-I");
1103
1104                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1105                         struct strbuf buf = STRBUF_INIT;
1106                         strbuf_addf(&buf, "+/%s%s",
1107                                         strcmp("less", pager) ? "" : "*",
1108                                         opt.pattern_list->pattern);
1109                         string_list_append(&path_list, buf.buf);
1110                         strbuf_detach(&buf, NULL);
1111                 }
1112         }
1113
1114         if (recurse_submodules && (!use_index || untracked))
1115                 die(_("option not supported with --recurse-submodules"));
1116
1117         if (!show_in_pager && !opt.status_only)
1118                 setup_pager();
1119
1120         if (!use_index && (untracked || cached))
1121                 die(_("--cached or --untracked cannot be used with --no-index"));
1122
1123         if (!use_index || untracked) {
1124                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1125                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1126         } else if (0 <= opt_exclude) {
1127                 die(_("--[no-]exclude-standard cannot be used for tracked contents"));
1128         } else if (!list.nr) {
1129                 if (!cached)
1130                         setup_work_tree();
1131
1132                 hit = grep_cache(&opt, &pathspec, cached);
1133         } else {
1134                 if (cached)
1135                         die(_("both --cached and trees are given"));
1136
1137                 hit = grep_objects(&opt, &pathspec, &list);
1138         }
1139
1140         if (num_threads > 1)
1141                 hit |= wait_all();
1142         if (hit && show_in_pager)
1143                 run_pager(&opt, prefix);
1144         clear_pathspec(&pathspec);
1145         free_grep_patterns(&opt);
1146         return !hit;
1147 }