builtin/pack-objects.c: respect 'pack.writeReverseIndex'
[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 void grep_source_name(struct grep_opt *opt, const char *filename,
299                              int tree_name_len, struct strbuf *out)
300 {
301         strbuf_reset(out);
302
303         if (opt->null_following_name) {
304                 if (opt->relative && opt->prefix_length) {
305                         struct strbuf rel_buf = STRBUF_INIT;
306                         const char *rel_name =
307                                 relative_path(filename + tree_name_len,
308                                               opt->prefix, &rel_buf);
309
310                         if (tree_name_len)
311                                 strbuf_add(out, filename, tree_name_len);
312
313                         strbuf_addstr(out, rel_name);
314                         strbuf_release(&rel_buf);
315                 } else {
316                         strbuf_addstr(out, filename);
317                 }
318                 return;
319         }
320
321         if (opt->relative && opt->prefix_length)
322                 quote_path(filename + tree_name_len, opt->prefix, out, 0);
323         else
324                 quote_c_style(filename + tree_name_len, out, NULL, 0);
325
326         if (tree_name_len)
327                 strbuf_insert(out, 0, filename, tree_name_len);
328 }
329
330 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
331                      const char *filename, int tree_name_len,
332                      const char *path)
333 {
334         struct strbuf pathbuf = STRBUF_INIT;
335         struct grep_source gs;
336
337         grep_source_name(opt, filename, tree_name_len, &pathbuf);
338         grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
339         strbuf_release(&pathbuf);
340
341         if (num_threads > 1) {
342                 /*
343                  * add_work() copies gs and thus assumes ownership of
344                  * its fields, so do not call grep_source_clear()
345                  */
346                 add_work(opt, &gs);
347                 return 0;
348         } else {
349                 int hit;
350
351                 hit = grep_source(opt, &gs);
352
353                 grep_source_clear(&gs);
354                 return hit;
355         }
356 }
357
358 static int grep_file(struct grep_opt *opt, const char *filename)
359 {
360         struct strbuf buf = STRBUF_INIT;
361         struct grep_source gs;
362
363         grep_source_name(opt, filename, 0, &buf);
364         grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
365         strbuf_release(&buf);
366
367         if (num_threads > 1) {
368                 /*
369                  * add_work() copies gs and thus assumes ownership of
370                  * its fields, so do not call grep_source_clear()
371                  */
372                 add_work(opt, &gs);
373                 return 0;
374         } else {
375                 int hit;
376
377                 hit = grep_source(opt, &gs);
378
379                 grep_source_clear(&gs);
380                 return hit;
381         }
382 }
383
384 static void append_path(struct grep_opt *opt, const void *data, size_t len)
385 {
386         struct string_list *path_list = opt->output_priv;
387
388         if (len == 1 && *(const char *)data == '\0')
389                 return;
390         string_list_append(path_list, xstrndup(data, len));
391 }
392
393 static void run_pager(struct grep_opt *opt, const char *prefix)
394 {
395         struct string_list *path_list = opt->output_priv;
396         struct child_process child = CHILD_PROCESS_INIT;
397         int i, status;
398
399         for (i = 0; i < path_list->nr; i++)
400                 strvec_push(&child.args, path_list->items[i].string);
401         child.dir = prefix;
402         child.use_shell = 1;
403
404         status = run_command(&child);
405         if (status)
406                 exit(status);
407 }
408
409 static int grep_cache(struct grep_opt *opt,
410                       const struct pathspec *pathspec, int cached);
411 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
412                      struct tree_desc *tree, struct strbuf *base, int tn_len,
413                      int check_attr);
414
415 static int grep_submodule(struct grep_opt *opt,
416                           const struct pathspec *pathspec,
417                           const struct object_id *oid,
418                           const char *filename, const char *path, int cached)
419 {
420         struct repository subrepo;
421         struct repository *superproject = opt->repo;
422         const struct submodule *sub;
423         struct grep_opt subopt;
424         int hit;
425
426         sub = submodule_from_path(superproject, &null_oid, path);
427
428         if (!is_submodule_active(superproject, path))
429                 return 0;
430
431         if (repo_submodule_init(&subrepo, superproject, sub))
432                 return 0;
433
434         /*
435          * NEEDSWORK: repo_read_gitmodules() might call
436          * add_to_alternates_memory() via config_from_gitmodules(). This
437          * operation causes a race condition with concurrent object readings
438          * performed by the worker threads. That's why we need obj_read_lock()
439          * here. It should be removed once it's no longer necessary to add the
440          * subrepo's odbs to the in-memory alternates list.
441          */
442         obj_read_lock();
443         repo_read_gitmodules(&subrepo, 0);
444
445         /*
446          * NEEDSWORK: This adds the submodule's object directory to the list of
447          * alternates for the single in-memory object store.  This has some bad
448          * consequences for memory (processed objects will never be freed) and
449          * performance (this increases the number of pack files git has to pay
450          * attention to, to the sum of the number of pack files in all the
451          * repositories processed so far).  This can be removed once the object
452          * store is no longer global and instead is a member of the repository
453          * object.
454          */
455         add_to_alternates_memory(subrepo.objects->odb->path);
456         obj_read_unlock();
457
458         memcpy(&subopt, opt, sizeof(subopt));
459         subopt.repo = &subrepo;
460
461         if (oid) {
462                 struct object *object;
463                 struct tree_desc tree;
464                 void *data;
465                 unsigned long size;
466                 struct strbuf base = STRBUF_INIT;
467
468                 obj_read_lock();
469                 object = parse_object_or_die(oid, NULL);
470                 obj_read_unlock();
471                 data = read_object_with_reference(&subrepo,
472                                                   &object->oid, tree_type,
473                                                   &size, NULL);
474                 if (!data)
475                         die(_("unable to read tree (%s)"), oid_to_hex(&object->oid));
476
477                 strbuf_addstr(&base, filename);
478                 strbuf_addch(&base, '/');
479
480                 init_tree_desc(&tree, data, size);
481                 hit = grep_tree(&subopt, pathspec, &tree, &base, base.len,
482                                 object->type == OBJ_COMMIT);
483                 strbuf_release(&base);
484                 free(data);
485         } else {
486                 hit = grep_cache(&subopt, pathspec, cached);
487         }
488
489         repo_clear(&subrepo);
490         return hit;
491 }
492
493 static int grep_cache(struct grep_opt *opt,
494                       const struct pathspec *pathspec, int cached)
495 {
496         struct repository *repo = opt->repo;
497         int hit = 0;
498         int nr;
499         struct strbuf name = STRBUF_INIT;
500         int name_base_len = 0;
501         if (repo->submodule_prefix) {
502                 name_base_len = strlen(repo->submodule_prefix);
503                 strbuf_addstr(&name, repo->submodule_prefix);
504         }
505
506         if (repo_read_index(repo) < 0)
507                 die(_("index file corrupt"));
508
509         for (nr = 0; nr < repo->index->cache_nr; nr++) {
510                 const struct cache_entry *ce = repo->index->cache[nr];
511                 strbuf_setlen(&name, name_base_len);
512                 strbuf_addstr(&name, ce->name);
513
514                 if (S_ISREG(ce->ce_mode) &&
515                     match_pathspec(repo->index, pathspec, name.buf, name.len, 0, NULL,
516                                    S_ISDIR(ce->ce_mode) ||
517                                    S_ISGITLINK(ce->ce_mode))) {
518                         /*
519                          * If CE_VALID is on, we assume worktree file and its
520                          * cache entry are identical, even if worktree file has
521                          * been modified, so use cache version instead
522                          */
523                         if (cached || (ce->ce_flags & CE_VALID) ||
524                             ce_skip_worktree(ce)) {
525                                 if (ce_stage(ce) || ce_intent_to_add(ce))
526                                         continue;
527                                 hit |= grep_oid(opt, &ce->oid, name.buf,
528                                                  0, name.buf);
529                         } else {
530                                 hit |= grep_file(opt, name.buf);
531                         }
532                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
533                            submodule_path_match(repo->index, pathspec, name.buf, NULL)) {
534                         hit |= grep_submodule(opt, pathspec, NULL, ce->name,
535                                               ce->name, cached);
536                 } else {
537                         continue;
538                 }
539
540                 if (ce_stage(ce)) {
541                         do {
542                                 nr++;
543                         } while (nr < repo->index->cache_nr &&
544                                  !strcmp(ce->name, repo->index->cache[nr]->name));
545                         nr--; /* compensate for loop control */
546                 }
547                 if (hit && opt->status_only)
548                         break;
549         }
550
551         strbuf_release(&name);
552         return hit;
553 }
554
555 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
556                      struct tree_desc *tree, struct strbuf *base, int tn_len,
557                      int check_attr)
558 {
559         struct repository *repo = opt->repo;
560         int hit = 0;
561         enum interesting match = entry_not_interesting;
562         struct name_entry entry;
563         int old_baselen = base->len;
564         struct strbuf name = STRBUF_INIT;
565         int name_base_len = 0;
566         if (repo->submodule_prefix) {
567                 strbuf_addstr(&name, repo->submodule_prefix);
568                 name_base_len = name.len;
569         }
570
571         while (tree_entry(tree, &entry)) {
572                 int te_len = tree_entry_len(&entry);
573
574                 if (match != all_entries_interesting) {
575                         strbuf_addstr(&name, base->buf + tn_len);
576                         match = tree_entry_interesting(repo->index,
577                                                        &entry, &name,
578                                                        0, pathspec);
579                         strbuf_setlen(&name, name_base_len);
580
581                         if (match == all_entries_not_interesting)
582                                 break;
583                         if (match == entry_not_interesting)
584                                 continue;
585                 }
586
587                 strbuf_add(base, entry.path, te_len);
588
589                 if (S_ISREG(entry.mode)) {
590                         hit |= grep_oid(opt, &entry.oid, base->buf, tn_len,
591                                          check_attr ? base->buf + tn_len : NULL);
592                 } else if (S_ISDIR(entry.mode)) {
593                         enum object_type type;
594                         struct tree_desc sub;
595                         void *data;
596                         unsigned long size;
597
598                         data = read_object_file(&entry.oid, &type, &size);
599                         if (!data)
600                                 die(_("unable to read tree (%s)"),
601                                     oid_to_hex(&entry.oid));
602
603                         strbuf_addch(base, '/');
604                         init_tree_desc(&sub, data, size);
605                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
606                                          check_attr);
607                         free(data);
608                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
609                         hit |= grep_submodule(opt, pathspec, &entry.oid,
610                                               base->buf, base->buf + tn_len,
611                                               1); /* ignored */
612                 }
613
614                 strbuf_setlen(base, old_baselen);
615
616                 if (hit && opt->status_only)
617                         break;
618         }
619
620         strbuf_release(&name);
621         return hit;
622 }
623
624 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
625                        struct object *obj, const char *name, const char *path)
626 {
627         if (obj->type == OBJ_BLOB)
628                 return grep_oid(opt, &obj->oid, name, 0, path);
629         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
630                 struct tree_desc tree;
631                 void *data;
632                 unsigned long size;
633                 struct strbuf base;
634                 int hit, len;
635
636                 data = read_object_with_reference(opt->repo,
637                                                   &obj->oid, tree_type,
638                                                   &size, NULL);
639                 if (!data)
640                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
641
642                 len = name ? strlen(name) : 0;
643                 strbuf_init(&base, PATH_MAX + len + 1);
644                 if (len) {
645                         strbuf_add(&base, name, len);
646                         strbuf_addch(&base, ':');
647                 }
648                 init_tree_desc(&tree, data, size);
649                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
650                                 obj->type == OBJ_COMMIT);
651                 strbuf_release(&base);
652                 free(data);
653                 return hit;
654         }
655         die(_("unable to grep from object of type %s"), type_name(obj->type));
656 }
657
658 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
659                         const struct object_array *list)
660 {
661         unsigned int i;
662         int hit = 0;
663         const unsigned int nr = list->nr;
664
665         for (i = 0; i < nr; i++) {
666                 struct object *real_obj;
667
668                 obj_read_lock();
669                 real_obj = deref_tag(opt->repo, list->objects[i].item,
670                                      NULL, 0);
671                 obj_read_unlock();
672
673                 if (!real_obj) {
674                         char hex[GIT_MAX_HEXSZ + 1];
675                         const char *name = list->objects[i].name;
676
677                         if (!name) {
678                                 oid_to_hex_r(hex, &list->objects[i].item->oid);
679                                 name = hex;
680                         }
681                         die(_("invalid object '%s' given."), name);
682                 }
683
684                 /* load the gitmodules file for this rev */
685                 if (recurse_submodules) {
686                         submodule_free(opt->repo);
687                         obj_read_lock();
688                         gitmodules_config_oid(&real_obj->oid);
689                         obj_read_unlock();
690                 }
691                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name,
692                                 list->objects[i].path)) {
693                         hit = 1;
694                         if (opt->status_only)
695                                 break;
696                 }
697         }
698         return hit;
699 }
700
701 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
702                           int exc_std, int use_index)
703 {
704         struct dir_struct dir;
705         int i, hit = 0;
706
707         dir_init(&dir);
708         if (!use_index)
709                 dir.flags |= DIR_NO_GITLINKS;
710         if (exc_std)
711                 setup_standard_excludes(&dir);
712
713         fill_directory(&dir, opt->repo->index, pathspec);
714         for (i = 0; i < dir.nr; i++) {
715                 hit |= grep_file(opt, dir.entries[i]->name);
716                 if (hit && opt->status_only)
717                         break;
718         }
719         dir_clear(&dir);
720         return hit;
721 }
722
723 static int context_callback(const struct option *opt, const char *arg,
724                             int unset)
725 {
726         struct grep_opt *grep_opt = opt->value;
727         int value;
728         const char *endp;
729
730         if (unset) {
731                 grep_opt->pre_context = grep_opt->post_context = 0;
732                 return 0;
733         }
734         value = strtol(arg, (char **)&endp, 10);
735         if (*endp) {
736                 return error(_("switch `%c' expects a numerical value"),
737                              opt->short_name);
738         }
739         grep_opt->pre_context = grep_opt->post_context = value;
740         return 0;
741 }
742
743 static int file_callback(const struct option *opt, const char *arg, int unset)
744 {
745         struct grep_opt *grep_opt = opt->value;
746         int from_stdin;
747         FILE *patterns;
748         int lno = 0;
749         struct strbuf sb = STRBUF_INIT;
750
751         BUG_ON_OPT_NEG(unset);
752
753         from_stdin = !strcmp(arg, "-");
754         patterns = from_stdin ? stdin : fopen(arg, "r");
755         if (!patterns)
756                 die_errno(_("cannot open '%s'"), arg);
757         while (strbuf_getline(&sb, patterns) == 0) {
758                 /* ignore empty line like grep does */
759                 if (sb.len == 0)
760                         continue;
761
762                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
763                                 GREP_PATTERN);
764         }
765         if (!from_stdin)
766                 fclose(patterns);
767         strbuf_release(&sb);
768         return 0;
769 }
770
771 static int not_callback(const struct option *opt, const char *arg, int unset)
772 {
773         struct grep_opt *grep_opt = opt->value;
774         BUG_ON_OPT_NEG(unset);
775         BUG_ON_OPT_ARG(arg);
776         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
777         return 0;
778 }
779
780 static int and_callback(const struct option *opt, const char *arg, int unset)
781 {
782         struct grep_opt *grep_opt = opt->value;
783         BUG_ON_OPT_NEG(unset);
784         BUG_ON_OPT_ARG(arg);
785         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
786         return 0;
787 }
788
789 static int open_callback(const struct option *opt, const char *arg, int unset)
790 {
791         struct grep_opt *grep_opt = opt->value;
792         BUG_ON_OPT_NEG(unset);
793         BUG_ON_OPT_ARG(arg);
794         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
795         return 0;
796 }
797
798 static int close_callback(const struct option *opt, const char *arg, int unset)
799 {
800         struct grep_opt *grep_opt = opt->value;
801         BUG_ON_OPT_NEG(unset);
802         BUG_ON_OPT_ARG(arg);
803         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
804         return 0;
805 }
806
807 static int pattern_callback(const struct option *opt, const char *arg,
808                             int unset)
809 {
810         struct grep_opt *grep_opt = opt->value;
811         BUG_ON_OPT_NEG(unset);
812         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
813         return 0;
814 }
815
816 int cmd_grep(int argc, const char **argv, const char *prefix)
817 {
818         int hit = 0;
819         int cached = 0, untracked = 0, opt_exclude = -1;
820         int seen_dashdash = 0;
821         int external_grep_allowed__ignored;
822         const char *show_in_pager = NULL, *default_pager = "dummy";
823         struct grep_opt opt;
824         struct object_array list = OBJECT_ARRAY_INIT;
825         struct pathspec pathspec;
826         struct string_list path_list = STRING_LIST_INIT_NODUP;
827         int i;
828         int dummy;
829         int use_index = 1;
830         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
831         int allow_revs;
832
833         struct option options[] = {
834                 OPT_BOOL(0, "cached", &cached,
835                         N_("search in index instead of in the work tree")),
836                 OPT_NEGBIT(0, "no-index", &use_index,
837                          N_("find in contents not managed by git"), 1),
838                 OPT_BOOL(0, "untracked", &untracked,
839                         N_("search in both tracked and untracked files")),
840                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
841                             N_("ignore files specified via '.gitignore'"), 1),
842                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
843                          N_("recursively search in each submodule")),
844                 OPT_GROUP(""),
845                 OPT_BOOL('v', "invert-match", &opt.invert,
846                         N_("show non-matching lines")),
847                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
848                         N_("case insensitive matching")),
849                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
850                         N_("match patterns only at word boundaries")),
851                 OPT_SET_INT('a', "text", &opt.binary,
852                         N_("process binary files as text"), GREP_BINARY_TEXT),
853                 OPT_SET_INT('I', NULL, &opt.binary,
854                         N_("don't match patterns in binary files"),
855                         GREP_BINARY_NOMATCH),
856                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
857                          N_("process binary files with textconv filters")),
858                 OPT_SET_INT('r', "recursive", &opt.max_depth,
859                             N_("search in subdirectories (default)"), -1),
860                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
861                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
862                         NULL, 1 },
863                 OPT_GROUP(""),
864                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
865                             N_("use extended POSIX regular expressions"),
866                             GREP_PATTERN_TYPE_ERE),
867                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
868                             N_("use basic POSIX regular expressions (default)"),
869                             GREP_PATTERN_TYPE_BRE),
870                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
871                             N_("interpret patterns as fixed strings"),
872                             GREP_PATTERN_TYPE_FIXED),
873                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
874                             N_("use Perl-compatible regular expressions"),
875                             GREP_PATTERN_TYPE_PCRE),
876                 OPT_GROUP(""),
877                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
878                 OPT_BOOL(0, "column", &opt.columnnum, N_("show column number of first match")),
879                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
880                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
881                 OPT_NEGBIT(0, "full-name", &opt.relative,
882                         N_("show filenames relative to top directory"), 1),
883                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
884                         N_("show only filenames instead of matching lines")),
885                 OPT_BOOL(0, "name-only", &opt.name_only,
886                         N_("synonym for --files-with-matches")),
887                 OPT_BOOL('L', "files-without-match",
888                         &opt.unmatch_name_only,
889                         N_("show only the names of files without match")),
890                 OPT_BOOL_F('z', "null", &opt.null_following_name,
891                            N_("print NUL after filenames"),
892                            PARSE_OPT_NOCOMPLETE),
893                 OPT_BOOL('o', "only-matching", &opt.only_matching,
894                         N_("show only matching parts of a line")),
895                 OPT_BOOL('c', "count", &opt.count,
896                         N_("show the number of matches instead of matching lines")),
897                 OPT__COLOR(&opt.color, N_("highlight matches")),
898                 OPT_BOOL(0, "break", &opt.file_break,
899                         N_("print empty line between matches from different files")),
900                 OPT_BOOL(0, "heading", &opt.heading,
901                         N_("show filename only once above matches from same file")),
902                 OPT_GROUP(""),
903                 OPT_CALLBACK('C', "context", &opt, N_("n"),
904                         N_("show <n> context lines before and after matches"),
905                         context_callback),
906                 OPT_INTEGER('B', "before-context", &opt.pre_context,
907                         N_("show <n> context lines before matches")),
908                 OPT_INTEGER('A', "after-context", &opt.post_context,
909                         N_("show <n> context lines after matches")),
910                 OPT_INTEGER(0, "threads", &num_threads,
911                         N_("use <n> worker threads")),
912                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
913                         context_callback),
914                 OPT_BOOL('p', "show-function", &opt.funcname,
915                         N_("show a line with the function name before matches")),
916                 OPT_BOOL('W', "function-context", &opt.funcbody,
917                         N_("show the surrounding function")),
918                 OPT_GROUP(""),
919                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
920                         N_("read patterns from file"), file_callback),
921                 OPT_CALLBACK_F('e', NULL, &opt, N_("pattern"),
922                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback),
923                 OPT_CALLBACK_F(0, "and", &opt, NULL,
924                         N_("combine patterns specified with -e"),
925                         PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback),
926                 OPT_BOOL(0, "or", &dummy, ""),
927                 OPT_CALLBACK_F(0, "not", &opt, NULL, "",
928                         PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback),
929                 OPT_CALLBACK_F('(', NULL, &opt, NULL, "",
930                         PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
931                         open_callback),
932                 OPT_CALLBACK_F(')', NULL, &opt, NULL, "",
933                         PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
934                         close_callback),
935                 OPT__QUIET(&opt.status_only,
936                            N_("indicate hit with exit status without output")),
937                 OPT_BOOL(0, "all-match", &opt.all_match,
938                         N_("show only matches from files that match all patterns")),
939                 OPT_SET_INT_F(0, "debug", &opt.debug,
940                               N_("show parse tree for grep expression"),
941                               1, PARSE_OPT_HIDDEN),
942                 OPT_GROUP(""),
943                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
944                         N_("pager"), N_("show matching files in the pager"),
945                         PARSE_OPT_OPTARG | PARSE_OPT_NOCOMPLETE,
946                         NULL, (intptr_t)default_pager },
947                 OPT_BOOL_F(0, "ext-grep", &external_grep_allowed__ignored,
948                            N_("allow calling of grep(1) (ignored by this build)"),
949                            PARSE_OPT_NOCOMPLETE),
950                 OPT_END()
951         };
952
953         git_config(grep_cmd_config, NULL);
954         grep_init(&opt, the_repository, prefix);
955
956         /*
957          * If there is no -- then the paths must exist in the working
958          * tree.  If there is no explicit pattern specified with -e or
959          * -f, we take the first unrecognized non option to be the
960          * pattern, but then what follows it must be zero or more
961          * valid refs up to the -- (if exists), and then existing
962          * paths.  If there is an explicit pattern, then the first
963          * unrecognized non option is the beginning of the refs list
964          * that continues up to the -- (if exists), and then paths.
965          */
966         argc = parse_options(argc, argv, prefix, options, grep_usage,
967                              PARSE_OPT_KEEP_DASHDASH |
968                              PARSE_OPT_STOP_AT_NON_OPTION);
969         grep_commit_pattern_type(pattern_type_arg, &opt);
970
971         if (use_index && !startup_info->have_repository) {
972                 int fallback = 0;
973                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
974                 if (fallback)
975                         use_index = 0;
976                 else
977                         /* die the same way as if we did it at the beginning */
978                         setup_git_directory();
979         }
980         /* Ignore --recurse-submodules if --no-index is given or implied */
981         if (!use_index)
982                 recurse_submodules = 0;
983
984         /*
985          * skip a -- separator; we know it cannot be
986          * separating revisions from pathnames if
987          * we haven't even had any patterns yet
988          */
989         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
990                 argv++;
991                 argc--;
992         }
993
994         /* First unrecognized non-option token */
995         if (argc > 0 && !opt.pattern_list) {
996                 append_grep_pattern(&opt, argv[0], "command line", 0,
997                                     GREP_PATTERN);
998                 argv++;
999                 argc--;
1000         }
1001
1002         if (show_in_pager == default_pager)
1003                 show_in_pager = git_pager(1);
1004         if (show_in_pager) {
1005                 opt.color = 0;
1006                 opt.name_only = 1;
1007                 opt.null_following_name = 1;
1008                 opt.output_priv = &path_list;
1009                 opt.output = append_path;
1010                 string_list_append(&path_list, show_in_pager);
1011         }
1012
1013         if (!opt.pattern_list)
1014                 die(_("no pattern given"));
1015
1016         /* --only-matching has no effect with --invert. */
1017         if (opt.invert)
1018                 opt.only_matching = 0;
1019
1020         /*
1021          * We have to find "--" in a separate pass, because its presence
1022          * influences how we will parse arguments that come before it.
1023          */
1024         for (i = 0; i < argc; i++) {
1025                 if (!strcmp(argv[i], "--")) {
1026                         seen_dashdash = 1;
1027                         break;
1028                 }
1029         }
1030
1031         /*
1032          * Resolve any rev arguments. If we have a dashdash, then everything up
1033          * to it must resolve as a rev. If not, then we stop at the first
1034          * non-rev and assume everything else is a path.
1035          */
1036         allow_revs = use_index && !untracked;
1037         for (i = 0; i < argc; i++) {
1038                 const char *arg = argv[i];
1039                 struct object_id oid;
1040                 struct object_context oc;
1041                 struct object *object;
1042
1043                 if (!strcmp(arg, "--")) {
1044                         i++;
1045                         break;
1046                 }
1047
1048                 if (!allow_revs) {
1049                         if (seen_dashdash)
1050                                 die(_("--no-index or --untracked cannot be used with revs"));
1051                         break;
1052                 }
1053
1054                 if (get_oid_with_context(the_repository, arg,
1055                                          GET_OID_RECORD_PATH,
1056                                          &oid, &oc)) {
1057                         if (seen_dashdash)
1058                                 die(_("unable to resolve revision: %s"), arg);
1059                         break;
1060                 }
1061
1062                 object = parse_object_or_die(&oid, arg);
1063                 if (!seen_dashdash)
1064                         verify_non_filename(prefix, arg);
1065                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1066                 free(oc.path);
1067         }
1068
1069         /*
1070          * Anything left over is presumed to be a path. But in the non-dashdash
1071          * "do what I mean" case, we verify and complain when that isn't true.
1072          */
1073         if (!seen_dashdash) {
1074                 int j;
1075                 for (j = i; j < argc; j++)
1076                         verify_filename(prefix, argv[j], j == i && allow_revs);
1077         }
1078
1079         parse_pathspec(&pathspec, 0,
1080                        PATHSPEC_PREFER_CWD |
1081                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1082                        prefix, argv + i);
1083         pathspec.max_depth = opt.max_depth;
1084         pathspec.recursive = 1;
1085         pathspec.recurse_submodules = !!recurse_submodules;
1086
1087         if (recurse_submodules && untracked)
1088                 die(_("--untracked not supported with --recurse-submodules"));
1089
1090         if (show_in_pager) {
1091                 if (num_threads > 1)
1092                         warning(_("invalid option combination, ignoring --threads"));
1093                 num_threads = 1;
1094         } else if (!HAVE_THREADS && num_threads > 1) {
1095                 warning(_("no threads support, ignoring --threads"));
1096                 num_threads = 1;
1097         } else if (num_threads < 0)
1098                 die(_("invalid number of threads specified (%d)"), num_threads);
1099         else if (num_threads == 0)
1100                 num_threads = HAVE_THREADS ? online_cpus() : 1;
1101
1102         if (num_threads > 1) {
1103                 if (!HAVE_THREADS)
1104                         BUG("Somebody got num_threads calculation wrong!");
1105                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1106                     && (opt.pre_context || opt.post_context ||
1107                         opt.file_break || opt.funcbody))
1108                         skip_first_line = 1;
1109
1110                 /*
1111                  * Pre-read gitmodules (if not read already) and force eager
1112                  * initialization of packed_git to prevent racy lazy
1113                  * reading/initialization once worker threads are started.
1114                  */
1115                 if (recurse_submodules)
1116                         repo_read_gitmodules(the_repository, 1);
1117                 if (startup_info->have_repository)
1118                         (void)get_packed_git(the_repository);
1119
1120                 start_threads(&opt);
1121         } else {
1122                 /*
1123                  * The compiled patterns on the main path are only
1124                  * used when not using threading. Otherwise
1125                  * start_threads() above calls compile_grep_patterns()
1126                  * for each thread.
1127                  */
1128                 compile_grep_patterns(&opt);
1129         }
1130
1131         if (show_in_pager && (cached || list.nr))
1132                 die(_("--open-files-in-pager only works on the worktree"));
1133
1134         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1135                 const char *pager = path_list.items[0].string;
1136                 int len = strlen(pager);
1137
1138                 if (len > 4 && is_dir_sep(pager[len - 5]))
1139                         pager += len - 4;
1140
1141                 if (opt.ignore_case && !strcmp("less", pager))
1142                         string_list_append(&path_list, "-I");
1143
1144                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1145                         struct strbuf buf = STRBUF_INIT;
1146                         strbuf_addf(&buf, "+/%s%s",
1147                                         strcmp("less", pager) ? "" : "*",
1148                                         opt.pattern_list->pattern);
1149                         string_list_append(&path_list,
1150                                            strbuf_detach(&buf, NULL));
1151                 }
1152         }
1153
1154         if (!show_in_pager && !opt.status_only)
1155                 setup_pager();
1156
1157         if (!use_index && (untracked || cached))
1158                 die(_("--cached or --untracked cannot be used with --no-index"));
1159
1160         if (!use_index || untracked) {
1161                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1162                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1163         } else if (0 <= opt_exclude) {
1164                 die(_("--[no-]exclude-standard cannot be used for tracked contents"));
1165         } else if (!list.nr) {
1166                 if (!cached)
1167                         setup_work_tree();
1168
1169                 hit = grep_cache(&opt, &pathspec, cached);
1170         } else {
1171                 if (cached)
1172                         die(_("both --cached and trees are given"));
1173
1174                 hit = grep_objects(&opt, &pathspec, &list);
1175         }
1176
1177         if (num_threads > 1)
1178                 hit |= wait_all();
1179         if (hit && show_in_pager)
1180                 run_pager(&opt, prefix);
1181         clear_pathspec(&pathspec);
1182         free_grep_patterns(&opt);
1183         grep_destroy();
1184         return !hit;
1185 }