Merge branch 'js/ci-windows-update'
[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 #include "packfile.h"
28
29 static char const * const grep_usage[] = {
30         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
31         NULL
32 };
33
34 static int recurse_submodules;
35
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, struct grep_source *gs)
95 {
96         if (opt->binary != GREP_BINARY_TEXT)
97                 grep_source_load_driver(gs, opt->repo->index);
98
99         grep_lock();
100
101         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
102                 pthread_cond_wait(&cond_write, &grep_mutex);
103         }
104
105         todo[todo_end].source = *gs;
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_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         enable_obj_read_lock();
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_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         disable_obj_read_lock();
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 int grep_oid(struct grep_opt *opt, const struct object_id *oid,
299                      const char *filename, int tree_name_len,
300                      const char *path)
301 {
302         struct strbuf pathbuf = STRBUF_INIT;
303         struct grep_source gs;
304
305         if (opt->relative && opt->prefix_length) {
306                 quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
307                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
308         } else {
309                 strbuf_addstr(&pathbuf, filename);
310         }
311
312         grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
313         strbuf_release(&pathbuf);
314
315         if (num_threads > 1) {
316                 /*
317                  * add_work() copies gs and thus assumes ownership of
318                  * its fields, so do not call grep_source_clear()
319                  */
320                 add_work(opt, &gs);
321                 return 0;
322         } else {
323                 int hit;
324
325                 hit = grep_source(opt, &gs);
326
327                 grep_source_clear(&gs);
328                 return hit;
329         }
330 }
331
332 static int grep_file(struct grep_opt *opt, const char *filename)
333 {
334         struct strbuf buf = STRBUF_INIT;
335         struct grep_source gs;
336
337         if (opt->relative && opt->prefix_length)
338                 quote_path_relative(filename, opt->prefix, &buf);
339         else
340                 strbuf_addstr(&buf, filename);
341
342         grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
343         strbuf_release(&buf);
344
345         if (num_threads > 1) {
346                 /*
347                  * add_work() copies gs and thus assumes ownership of
348                  * its fields, so do not call grep_source_clear()
349                  */
350                 add_work(opt, &gs);
351                 return 0;
352         } else {
353                 int hit;
354
355                 hit = grep_source(opt, &gs);
356
357                 grep_source_clear(&gs);
358                 return hit;
359         }
360 }
361
362 static void append_path(struct grep_opt *opt, const void *data, size_t len)
363 {
364         struct string_list *path_list = opt->output_priv;
365
366         if (len == 1 && *(const char *)data == '\0')
367                 return;
368         string_list_append(path_list, xstrndup(data, len));
369 }
370
371 static void run_pager(struct grep_opt *opt, const char *prefix)
372 {
373         struct string_list *path_list = opt->output_priv;
374         struct child_process child = CHILD_PROCESS_INIT;
375         int i, status;
376
377         for (i = 0; i < path_list->nr; i++)
378                 argv_array_push(&child.args, path_list->items[i].string);
379         child.dir = prefix;
380         child.use_shell = 1;
381
382         status = run_command(&child);
383         if (status)
384                 exit(status);
385 }
386
387 static int grep_cache(struct grep_opt *opt,
388                       const struct pathspec *pathspec, int cached);
389 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
390                      struct tree_desc *tree, struct strbuf *base, int tn_len,
391                      int check_attr);
392
393 static int grep_submodule(struct grep_opt *opt,
394                           const struct pathspec *pathspec,
395                           const struct object_id *oid,
396                           const char *filename, const char *path, int cached)
397 {
398         struct repository subrepo;
399         struct repository *superproject = opt->repo;
400         const struct submodule *sub;
401         struct grep_opt subopt;
402         int hit;
403
404         sub = submodule_from_path(superproject, &null_oid, path);
405
406         if (!is_submodule_active(superproject, path))
407                 return 0;
408
409         if (repo_submodule_init(&subrepo, superproject, sub))
410                 return 0;
411
412         /*
413          * NEEDSWORK: repo_read_gitmodules() might call
414          * add_to_alternates_memory() via config_from_gitmodules(). This
415          * operation causes a race condition with concurrent object readings
416          * performed by the worker threads. That's why we need obj_read_lock()
417          * here. It should be removed once it's no longer necessary to add the
418          * subrepo's odbs to the in-memory alternates list.
419          */
420         obj_read_lock();
421         repo_read_gitmodules(&subrepo, 0);
422
423         /*
424          * NEEDSWORK: This adds the submodule's object directory to the list of
425          * alternates for the single in-memory object store.  This has some bad
426          * consequences for memory (processed objects will never be freed) and
427          * performance (this increases the number of pack files git has to pay
428          * attention to, to the sum of the number of pack files in all the
429          * repositories processed so far).  This can be removed once the object
430          * store is no longer global and instead is a member of the repository
431          * object.
432          */
433         add_to_alternates_memory(subrepo.objects->odb->path);
434         obj_read_unlock();
435
436         memcpy(&subopt, opt, sizeof(subopt));
437         subopt.repo = &subrepo;
438
439         if (oid) {
440                 struct object *object;
441                 struct tree_desc tree;
442                 void *data;
443                 unsigned long size;
444                 struct strbuf base = STRBUF_INIT;
445
446                 obj_read_lock();
447                 object = parse_object_or_die(oid, oid_to_hex(oid));
448                 obj_read_unlock();
449                 data = read_object_with_reference(&subrepo,
450                                                   &object->oid, tree_type,
451                                                   &size, NULL);
452                 if (!data)
453                         die(_("unable to read tree (%s)"), oid_to_hex(&object->oid));
454
455                 strbuf_addstr(&base, filename);
456                 strbuf_addch(&base, '/');
457
458                 init_tree_desc(&tree, data, size);
459                 hit = grep_tree(&subopt, pathspec, &tree, &base, base.len,
460                                 object->type == OBJ_COMMIT);
461                 strbuf_release(&base);
462                 free(data);
463         } else {
464                 hit = grep_cache(&subopt, pathspec, cached);
465         }
466
467         repo_clear(&subrepo);
468         return hit;
469 }
470
471 static int grep_cache(struct grep_opt *opt,
472                       const struct pathspec *pathspec, int cached)
473 {
474         struct repository *repo = opt->repo;
475         int hit = 0;
476         int nr;
477         struct strbuf name = STRBUF_INIT;
478         int name_base_len = 0;
479         if (repo->submodule_prefix) {
480                 name_base_len = strlen(repo->submodule_prefix);
481                 strbuf_addstr(&name, repo->submodule_prefix);
482         }
483
484         if (repo_read_index(repo) < 0)
485                 die(_("index file corrupt"));
486
487         for (nr = 0; nr < repo->index->cache_nr; nr++) {
488                 const struct cache_entry *ce = repo->index->cache[nr];
489                 strbuf_setlen(&name, name_base_len);
490                 strbuf_addstr(&name, ce->name);
491
492                 if (S_ISREG(ce->ce_mode) &&
493                     match_pathspec(repo->index, pathspec, name.buf, name.len, 0, NULL,
494                                    S_ISDIR(ce->ce_mode) ||
495                                    S_ISGITLINK(ce->ce_mode))) {
496                         /*
497                          * If CE_VALID is on, we assume worktree file and its
498                          * cache entry are identical, even if worktree file has
499                          * been modified, so use cache version instead
500                          */
501                         if (cached || (ce->ce_flags & CE_VALID) ||
502                             ce_skip_worktree(ce)) {
503                                 if (ce_stage(ce) || ce_intent_to_add(ce))
504                                         continue;
505                                 hit |= grep_oid(opt, &ce->oid, name.buf,
506                                                  0, name.buf);
507                         } else {
508                                 hit |= grep_file(opt, name.buf);
509                         }
510                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
511                            submodule_path_match(repo->index, pathspec, name.buf, NULL)) {
512                         hit |= grep_submodule(opt, pathspec, NULL, ce->name,
513                                               ce->name, cached);
514                 } else {
515                         continue;
516                 }
517
518                 if (ce_stage(ce)) {
519                         do {
520                                 nr++;
521                         } while (nr < repo->index->cache_nr &&
522                                  !strcmp(ce->name, repo->index->cache[nr]->name));
523                         nr--; /* compensate for loop control */
524                 }
525                 if (hit && opt->status_only)
526                         break;
527         }
528
529         strbuf_release(&name);
530         return hit;
531 }
532
533 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
534                      struct tree_desc *tree, struct strbuf *base, int tn_len,
535                      int check_attr)
536 {
537         struct repository *repo = opt->repo;
538         int hit = 0;
539         enum interesting match = entry_not_interesting;
540         struct name_entry entry;
541         int old_baselen = base->len;
542         struct strbuf name = STRBUF_INIT;
543         int name_base_len = 0;
544         if (repo->submodule_prefix) {
545                 strbuf_addstr(&name, repo->submodule_prefix);
546                 name_base_len = name.len;
547         }
548
549         while (tree_entry(tree, &entry)) {
550                 int te_len = tree_entry_len(&entry);
551
552                 if (match != all_entries_interesting) {
553                         strbuf_addstr(&name, base->buf + tn_len);
554                         match = tree_entry_interesting(repo->index,
555                                                        &entry, &name,
556                                                        0, pathspec);
557                         strbuf_setlen(&name, name_base_len);
558
559                         if (match == all_entries_not_interesting)
560                                 break;
561                         if (match == entry_not_interesting)
562                                 continue;
563                 }
564
565                 strbuf_add(base, entry.path, te_len);
566
567                 if (S_ISREG(entry.mode)) {
568                         hit |= grep_oid(opt, &entry.oid, base->buf, tn_len,
569                                          check_attr ? base->buf + tn_len : NULL);
570                 } else if (S_ISDIR(entry.mode)) {
571                         enum object_type type;
572                         struct tree_desc sub;
573                         void *data;
574                         unsigned long size;
575
576                         data = read_object_file(&entry.oid, &type, &size);
577                         if (!data)
578                                 die(_("unable to read tree (%s)"),
579                                     oid_to_hex(&entry.oid));
580
581                         strbuf_addch(base, '/');
582                         init_tree_desc(&sub, data, size);
583                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
584                                          check_attr);
585                         free(data);
586                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
587                         hit |= grep_submodule(opt, pathspec, &entry.oid,
588                                               base->buf, base->buf + tn_len,
589                                               1); /* ignored */
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                 data = read_object_with_reference(opt->repo,
615                                                   &obj->oid, tree_type,
616                                                   &size, NULL);
617                 if (!data)
618                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
619
620                 len = name ? strlen(name) : 0;
621                 strbuf_init(&base, PATH_MAX + len + 1);
622                 if (len) {
623                         strbuf_add(&base, name, len);
624                         strbuf_addch(&base, ':');
625                 }
626                 init_tree_desc(&tree, data, size);
627                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
628                                 obj->type == OBJ_COMMIT);
629                 strbuf_release(&base);
630                 free(data);
631                 return hit;
632         }
633         die(_("unable to grep from object of type %s"), type_name(obj->type));
634 }
635
636 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
637                         const struct object_array *list)
638 {
639         unsigned int i;
640         int hit = 0;
641         const unsigned int nr = list->nr;
642
643         for (i = 0; i < nr; i++) {
644                 struct object *real_obj;
645
646                 obj_read_lock();
647                 real_obj = deref_tag(opt->repo, list->objects[i].item,
648                                      NULL, 0);
649                 obj_read_unlock();
650
651                 /* load the gitmodules file for this rev */
652                 if (recurse_submodules) {
653                         submodule_free(opt->repo);
654                         obj_read_lock();
655                         gitmodules_config_oid(&real_obj->oid);
656                         obj_read_unlock();
657                 }
658                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name,
659                                 list->objects[i].path)) {
660                         hit = 1;
661                         if (opt->status_only)
662                                 break;
663                 }
664         }
665         return hit;
666 }
667
668 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
669                           int exc_std, int use_index)
670 {
671         struct dir_struct dir;
672         int i, hit = 0;
673
674         memset(&dir, 0, sizeof(dir));
675         if (!use_index)
676                 dir.flags |= DIR_NO_GITLINKS;
677         if (exc_std)
678                 setup_standard_excludes(&dir);
679
680         fill_directory(&dir, opt->repo->index, pathspec);
681         for (i = 0; i < dir.nr; i++) {
682                 if (!dir_path_match(opt->repo->index, dir.entries[i], pathspec, 0, NULL))
683                         continue;
684                 hit |= grep_file(opt, dir.entries[i]->name);
685                 if (hit && opt->status_only)
686                         break;
687         }
688         return hit;
689 }
690
691 static int context_callback(const struct option *opt, const char *arg,
692                             int unset)
693 {
694         struct grep_opt *grep_opt = opt->value;
695         int value;
696         const char *endp;
697
698         if (unset) {
699                 grep_opt->pre_context = grep_opt->post_context = 0;
700                 return 0;
701         }
702         value = strtol(arg, (char **)&endp, 10);
703         if (*endp) {
704                 return error(_("switch `%c' expects a numerical value"),
705                              opt->short_name);
706         }
707         grep_opt->pre_context = grep_opt->post_context = value;
708         return 0;
709 }
710
711 static int file_callback(const struct option *opt, const char *arg, int unset)
712 {
713         struct grep_opt *grep_opt = opt->value;
714         int from_stdin;
715         FILE *patterns;
716         int lno = 0;
717         struct strbuf sb = STRBUF_INIT;
718
719         BUG_ON_OPT_NEG(unset);
720
721         from_stdin = !strcmp(arg, "-");
722         patterns = from_stdin ? stdin : fopen(arg, "r");
723         if (!patterns)
724                 die_errno(_("cannot open '%s'"), arg);
725         while (strbuf_getline(&sb, patterns) == 0) {
726                 /* ignore empty line like grep does */
727                 if (sb.len == 0)
728                         continue;
729
730                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
731                                 GREP_PATTERN);
732         }
733         if (!from_stdin)
734                 fclose(patterns);
735         strbuf_release(&sb);
736         return 0;
737 }
738
739 static int not_callback(const struct option *opt, const char *arg, int unset)
740 {
741         struct grep_opt *grep_opt = opt->value;
742         BUG_ON_OPT_NEG(unset);
743         BUG_ON_OPT_ARG(arg);
744         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
745         return 0;
746 }
747
748 static int and_callback(const struct option *opt, const char *arg, int unset)
749 {
750         struct grep_opt *grep_opt = opt->value;
751         BUG_ON_OPT_NEG(unset);
752         BUG_ON_OPT_ARG(arg);
753         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
754         return 0;
755 }
756
757 static int open_callback(const struct option *opt, const char *arg, int unset)
758 {
759         struct grep_opt *grep_opt = opt->value;
760         BUG_ON_OPT_NEG(unset);
761         BUG_ON_OPT_ARG(arg);
762         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
763         return 0;
764 }
765
766 static int close_callback(const struct option *opt, const char *arg, int unset)
767 {
768         struct grep_opt *grep_opt = opt->value;
769         BUG_ON_OPT_NEG(unset);
770         BUG_ON_OPT_ARG(arg);
771         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
772         return 0;
773 }
774
775 static int pattern_callback(const struct option *opt, const char *arg,
776                             int unset)
777 {
778         struct grep_opt *grep_opt = opt->value;
779         BUG_ON_OPT_NEG(unset);
780         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
781         return 0;
782 }
783
784 int cmd_grep(int argc, const char **argv, const char *prefix)
785 {
786         int hit = 0;
787         int cached = 0, untracked = 0, opt_exclude = -1;
788         int seen_dashdash = 0;
789         int external_grep_allowed__ignored;
790         const char *show_in_pager = NULL, *default_pager = "dummy";
791         struct grep_opt opt;
792         struct object_array list = OBJECT_ARRAY_INIT;
793         struct pathspec pathspec;
794         struct string_list path_list = STRING_LIST_INIT_NODUP;
795         int i;
796         int dummy;
797         int use_index = 1;
798         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
799         int allow_revs;
800
801         struct option options[] = {
802                 OPT_BOOL(0, "cached", &cached,
803                         N_("search in index instead of in the work tree")),
804                 OPT_NEGBIT(0, "no-index", &use_index,
805                          N_("find in contents not managed by git"), 1),
806                 OPT_BOOL(0, "untracked", &untracked,
807                         N_("search in both tracked and untracked files")),
808                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
809                             N_("ignore files specified via '.gitignore'"), 1),
810                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
811                          N_("recursively search in each submodule")),
812                 OPT_GROUP(""),
813                 OPT_BOOL('v', "invert-match", &opt.invert,
814                         N_("show non-matching lines")),
815                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
816                         N_("case insensitive matching")),
817                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
818                         N_("match patterns only at word boundaries")),
819                 OPT_SET_INT('a', "text", &opt.binary,
820                         N_("process binary files as text"), GREP_BINARY_TEXT),
821                 OPT_SET_INT('I', NULL, &opt.binary,
822                         N_("don't match patterns in binary files"),
823                         GREP_BINARY_NOMATCH),
824                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
825                          N_("process binary files with textconv filters")),
826                 OPT_SET_INT('r', "recursive", &opt.max_depth,
827                             N_("search in subdirectories (default)"), -1),
828                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
829                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
830                         NULL, 1 },
831                 OPT_GROUP(""),
832                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
833                             N_("use extended POSIX regular expressions"),
834                             GREP_PATTERN_TYPE_ERE),
835                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
836                             N_("use basic POSIX regular expressions (default)"),
837                             GREP_PATTERN_TYPE_BRE),
838                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
839                             N_("interpret patterns as fixed strings"),
840                             GREP_PATTERN_TYPE_FIXED),
841                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
842                             N_("use Perl-compatible regular expressions"),
843                             GREP_PATTERN_TYPE_PCRE),
844                 OPT_GROUP(""),
845                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
846                 OPT_BOOL(0, "column", &opt.columnnum, N_("show column number of first match")),
847                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
848                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
849                 OPT_NEGBIT(0, "full-name", &opt.relative,
850                         N_("show filenames relative to top directory"), 1),
851                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
852                         N_("show only filenames instead of matching lines")),
853                 OPT_BOOL(0, "name-only", &opt.name_only,
854                         N_("synonym for --files-with-matches")),
855                 OPT_BOOL('L', "files-without-match",
856                         &opt.unmatch_name_only,
857                         N_("show only the names of files without match")),
858                 OPT_BOOL_F('z', "null", &opt.null_following_name,
859                            N_("print NUL after filenames"),
860                            PARSE_OPT_NOCOMPLETE),
861                 OPT_BOOL('o', "only-matching", &opt.only_matching,
862                         N_("show only matching parts of a line")),
863                 OPT_BOOL('c', "count", &opt.count,
864                         N_("show the number of matches instead of matching lines")),
865                 OPT__COLOR(&opt.color, N_("highlight matches")),
866                 OPT_BOOL(0, "break", &opt.file_break,
867                         N_("print empty line between matches from different files")),
868                 OPT_BOOL(0, "heading", &opt.heading,
869                         N_("show filename only once above matches from same file")),
870                 OPT_GROUP(""),
871                 OPT_CALLBACK('C', "context", &opt, N_("n"),
872                         N_("show <n> context lines before and after matches"),
873                         context_callback),
874                 OPT_INTEGER('B', "before-context", &opt.pre_context,
875                         N_("show <n> context lines before matches")),
876                 OPT_INTEGER('A', "after-context", &opt.post_context,
877                         N_("show <n> context lines after matches")),
878                 OPT_INTEGER(0, "threads", &num_threads,
879                         N_("use <n> worker threads")),
880                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
881                         context_callback),
882                 OPT_BOOL('p', "show-function", &opt.funcname,
883                         N_("show a line with the function name before matches")),
884                 OPT_BOOL('W', "function-context", &opt.funcbody,
885                         N_("show the surrounding function")),
886                 OPT_GROUP(""),
887                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
888                         N_("read patterns from file"), file_callback),
889                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
890                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
891                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
892                   N_("combine patterns specified with -e"),
893                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
894                 OPT_BOOL(0, "or", &dummy, ""),
895                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
896                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
897                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
898                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
899                   open_callback },
900                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
901                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
902                   close_callback },
903                 OPT__QUIET(&opt.status_only,
904                            N_("indicate hit with exit status without output")),
905                 OPT_BOOL(0, "all-match", &opt.all_match,
906                         N_("show only matches from files that match all patterns")),
907                 OPT_SET_INT_F(0, "debug", &opt.debug,
908                               N_("show parse tree for grep expression"),
909                               1, PARSE_OPT_HIDDEN),
910                 OPT_GROUP(""),
911                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
912                         N_("pager"), N_("show matching files in the pager"),
913                         PARSE_OPT_OPTARG | PARSE_OPT_NOCOMPLETE,
914                         NULL, (intptr_t)default_pager },
915                 OPT_BOOL_F(0, "ext-grep", &external_grep_allowed__ignored,
916                            N_("allow calling of grep(1) (ignored by this build)"),
917                            PARSE_OPT_NOCOMPLETE),
918                 OPT_END()
919         };
920
921         init_grep_defaults(the_repository);
922         git_config(grep_cmd_config, NULL);
923         grep_init(&opt, the_repository, prefix);
924
925         /*
926          * If there is no -- then the paths must exist in the working
927          * tree.  If there is no explicit pattern specified with -e or
928          * -f, we take the first unrecognized non option to be the
929          * pattern, but then what follows it must be zero or more
930          * valid refs up to the -- (if exists), and then existing
931          * paths.  If there is an explicit pattern, then the first
932          * unrecognized non option is the beginning of the refs list
933          * that continues up to the -- (if exists), and then paths.
934          */
935         argc = parse_options(argc, argv, prefix, options, grep_usage,
936                              PARSE_OPT_KEEP_DASHDASH |
937                              PARSE_OPT_STOP_AT_NON_OPTION);
938         grep_commit_pattern_type(pattern_type_arg, &opt);
939
940         if (use_index && !startup_info->have_repository) {
941                 int fallback = 0;
942                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
943                 if (fallback)
944                         use_index = 0;
945                 else
946                         /* die the same way as if we did it at the beginning */
947                         setup_git_directory();
948         }
949         /* Ignore --recurse-submodules if --no-index is given or implied */
950         if (!use_index)
951                 recurse_submodules = 0;
952
953         /*
954          * skip a -- separator; we know it cannot be
955          * separating revisions from pathnames if
956          * we haven't even had any patterns yet
957          */
958         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
959                 argv++;
960                 argc--;
961         }
962
963         /* First unrecognized non-option token */
964         if (argc > 0 && !opt.pattern_list) {
965                 append_grep_pattern(&opt, argv[0], "command line", 0,
966                                     GREP_PATTERN);
967                 argv++;
968                 argc--;
969         }
970
971         if (show_in_pager == default_pager)
972                 show_in_pager = git_pager(1);
973         if (show_in_pager) {
974                 opt.color = 0;
975                 opt.name_only = 1;
976                 opt.null_following_name = 1;
977                 opt.output_priv = &path_list;
978                 opt.output = append_path;
979                 string_list_append(&path_list, show_in_pager);
980         }
981
982         if (!opt.pattern_list)
983                 die(_("no pattern given"));
984
985         /* --only-matching has no effect with --invert. */
986         if (opt.invert)
987                 opt.only_matching = 0;
988
989         /*
990          * We have to find "--" in a separate pass, because its presence
991          * influences how we will parse arguments that come before it.
992          */
993         for (i = 0; i < argc; i++) {
994                 if (!strcmp(argv[i], "--")) {
995                         seen_dashdash = 1;
996                         break;
997                 }
998         }
999
1000         /*
1001          * Resolve any rev arguments. If we have a dashdash, then everything up
1002          * to it must resolve as a rev. If not, then we stop at the first
1003          * non-rev and assume everything else is a path.
1004          */
1005         allow_revs = use_index && !untracked;
1006         for (i = 0; i < argc; i++) {
1007                 const char *arg = argv[i];
1008                 struct object_id oid;
1009                 struct object_context oc;
1010                 struct object *object;
1011
1012                 if (!strcmp(arg, "--")) {
1013                         i++;
1014                         break;
1015                 }
1016
1017                 if (!allow_revs) {
1018                         if (seen_dashdash)
1019                                 die(_("--no-index or --untracked cannot be used with revs"));
1020                         break;
1021                 }
1022
1023                 if (get_oid_with_context(the_repository, arg,
1024                                          GET_OID_RECORD_PATH,
1025                                          &oid, &oc)) {
1026                         if (seen_dashdash)
1027                                 die(_("unable to resolve revision: %s"), arg);
1028                         break;
1029                 }
1030
1031                 object = parse_object_or_die(&oid, arg);
1032                 if (!seen_dashdash)
1033                         verify_non_filename(prefix, arg);
1034                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1035                 free(oc.path);
1036         }
1037
1038         /*
1039          * Anything left over is presumed to be a path. But in the non-dashdash
1040          * "do what I mean" case, we verify and complain when that isn't true.
1041          */
1042         if (!seen_dashdash) {
1043                 int j;
1044                 for (j = i; j < argc; j++)
1045                         verify_filename(prefix, argv[j], j == i && allow_revs);
1046         }
1047
1048         parse_pathspec(&pathspec, 0,
1049                        PATHSPEC_PREFER_CWD |
1050                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1051                        prefix, argv + i);
1052         pathspec.max_depth = opt.max_depth;
1053         pathspec.recursive = 1;
1054         pathspec.recurse_submodules = !!recurse_submodules;
1055
1056         if (recurse_submodules && untracked)
1057                 die(_("--untracked not supported with --recurse-submodules"));
1058
1059         if (show_in_pager) {
1060                 if (num_threads > 1)
1061                         warning(_("invalid option combination, ignoring --threads"));
1062                 num_threads = 1;
1063         } else if (!HAVE_THREADS && num_threads > 1) {
1064                 warning(_("no threads support, ignoring --threads"));
1065                 num_threads = 1;
1066         } else if (num_threads < 0)
1067                 die(_("invalid number of threads specified (%d)"), num_threads);
1068         else if (num_threads == 0)
1069                 num_threads = HAVE_THREADS ? online_cpus() : 1;
1070
1071         if (num_threads > 1) {
1072                 if (!HAVE_THREADS)
1073                         BUG("Somebody got num_threads calculation wrong!");
1074                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1075                     && (opt.pre_context || opt.post_context ||
1076                         opt.file_break || opt.funcbody))
1077                         skip_first_line = 1;
1078
1079                 /*
1080                  * Pre-read gitmodules (if not read already) and force eager
1081                  * initialization of packed_git to prevent racy lazy
1082                  * reading/initialization once worker threads are started.
1083                  */
1084                 if (recurse_submodules)
1085                         repo_read_gitmodules(the_repository, 1);
1086                 if (startup_info->have_repository)
1087                         (void)get_packed_git(the_repository);
1088
1089                 start_threads(&opt);
1090         } else {
1091                 /*
1092                  * The compiled patterns on the main path are only
1093                  * used when not using threading. Otherwise
1094                  * start_threads() above calls compile_grep_patterns()
1095                  * for each thread.
1096                  */
1097                 compile_grep_patterns(&opt);
1098         }
1099
1100         if (show_in_pager && (cached || list.nr))
1101                 die(_("--open-files-in-pager only works on the worktree"));
1102
1103         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1104                 const char *pager = path_list.items[0].string;
1105                 int len = strlen(pager);
1106
1107                 if (len > 4 && is_dir_sep(pager[len - 5]))
1108                         pager += len - 4;
1109
1110                 if (opt.ignore_case && !strcmp("less", pager))
1111                         string_list_append(&path_list, "-I");
1112
1113                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1114                         struct strbuf buf = STRBUF_INIT;
1115                         strbuf_addf(&buf, "+/%s%s",
1116                                         strcmp("less", pager) ? "" : "*",
1117                                         opt.pattern_list->pattern);
1118                         string_list_append(&path_list,
1119                                            strbuf_detach(&buf, NULL));
1120                 }
1121         }
1122
1123         if (!show_in_pager && !opt.status_only)
1124                 setup_pager();
1125
1126         if (!use_index && (untracked || cached))
1127                 die(_("--cached or --untracked cannot be used with --no-index"));
1128
1129         if (!use_index || untracked) {
1130                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1131                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1132         } else if (0 <= opt_exclude) {
1133                 die(_("--[no-]exclude-standard cannot be used for tracked contents"));
1134         } else if (!list.nr) {
1135                 if (!cached)
1136                         setup_work_tree();
1137
1138                 hit = grep_cache(&opt, &pathspec, cached);
1139         } else {
1140                 if (cached)
1141                         die(_("both --cached and trees are given"));
1142
1143                 hit = grep_objects(&opt, &pathspec, &list);
1144         }
1145
1146         if (num_threads > 1)
1147                 hit |= wait_all();
1148         if (hit && show_in_pager)
1149                 run_pager(&opt, prefix);
1150         clear_pathspec(&pathspec);
1151         free_grep_patterns(&opt);
1152         grep_destroy();
1153         return !hit;
1154 }