Merge branch 'jk/drop-sha1-entry-pos'
[git] / builtin / grep.c
1 /*
2  * Builtin "git grep"
3  *
4  * Copyright (c) 2006 Junio C Hamano
5  */
6 #include "cache.h"
7 #include "repository.h"
8 #include "config.h"
9 #include "blob.h"
10 #include "tree.h"
11 #include "commit.h"
12 #include "tag.h"
13 #include "tree-walk.h"
14 #include "builtin.h"
15 #include "parse-options.h"
16 #include "string-list.h"
17 #include "run-command.h"
18 #include "userdiff.h"
19 #include "grep.h"
20 #include "quote.h"
21 #include "dir.h"
22 #include "pathspec.h"
23 #include "submodule.h"
24 #include "submodule-config.h"
25
26 static char const * const grep_usage[] = {
27         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
28         NULL
29 };
30
31 static int recurse_submodules;
32
33 #define GREP_NUM_THREADS_DEFAULT 8
34 static int num_threads;
35
36 #ifndef NO_PTHREADS
37 static pthread_t *threads;
38
39 /* We use one producer thread and THREADS consumer
40  * threads. The producer adds struct work_items to 'todo' and the
41  * consumers pick work items from the same array.
42  */
43 struct work_item {
44         struct grep_source source;
45         char done;
46         struct strbuf out;
47 };
48
49 /* In the range [todo_done, todo_start) in 'todo' we have work_items
50  * that have been or are processed by a consumer thread. We haven't
51  * written the result for these to stdout yet.
52  *
53  * The work_items in [todo_start, todo_end) are waiting to be picked
54  * up by a consumer thread.
55  *
56  * The ranges are modulo TODO_SIZE.
57  */
58 #define TODO_SIZE 128
59 static struct work_item todo[TODO_SIZE];
60 static int todo_start;
61 static int todo_end;
62 static int todo_done;
63
64 /* Has all work items been added? */
65 static int all_work_added;
66
67 /* This lock protects all the variables above. */
68 static pthread_mutex_t grep_mutex;
69
70 static inline void grep_lock(void)
71 {
72         assert(num_threads);
73         pthread_mutex_lock(&grep_mutex);
74 }
75
76 static inline void grep_unlock(void)
77 {
78         assert(num_threads);
79         pthread_mutex_unlock(&grep_mutex);
80 }
81
82 /* Signalled when a new work_item is added to todo. */
83 static pthread_cond_t cond_add;
84
85 /* Signalled when the result from one work_item is written to
86  * stdout.
87  */
88 static pthread_cond_t cond_write;
89
90 /* Signalled when we are finished with everything. */
91 static pthread_cond_t cond_result;
92
93 static int skip_first_line;
94
95 static void add_work(struct grep_opt *opt, enum grep_source_type type,
96                      const char *name, const char *path, const void *id)
97 {
98         grep_lock();
99
100         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
101                 pthread_cond_wait(&cond_write, &grep_mutex);
102         }
103
104         grep_source_init(&todo[todo_end].source, type, name, path, id);
105         if (opt->binary != GREP_BINARY_TEXT)
106                 grep_source_load_driver(&todo[todo_end].source);
107         todo[todo_end].done = 0;
108         strbuf_reset(&todo[todo_end].out);
109         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
110
111         pthread_cond_signal(&cond_add);
112         grep_unlock();
113 }
114
115 static struct work_item *get_work(void)
116 {
117         struct work_item *ret;
118
119         grep_lock();
120         while (todo_start == todo_end && !all_work_added) {
121                 pthread_cond_wait(&cond_add, &grep_mutex);
122         }
123
124         if (todo_start == todo_end && all_work_added) {
125                 ret = NULL;
126         } else {
127                 ret = &todo[todo_start];
128                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
129         }
130         grep_unlock();
131         return ret;
132 }
133
134 static void work_done(struct work_item *w)
135 {
136         int old_done;
137
138         grep_lock();
139         w->done = 1;
140         old_done = todo_done;
141         for(; todo[todo_done].done && todo_done != todo_start;
142             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
143                 w = &todo[todo_done];
144                 if (w->out.len) {
145                         const char *p = w->out.buf;
146                         size_t len = w->out.len;
147
148                         /* Skip the leading hunk mark of the first file. */
149                         if (skip_first_line) {
150                                 while (len) {
151                                         len--;
152                                         if (*p++ == '\n')
153                                                 break;
154                                 }
155                                 skip_first_line = 0;
156                         }
157
158                         write_or_die(1, p, len);
159                 }
160                 grep_source_clear(&w->source);
161         }
162
163         if (old_done != todo_done)
164                 pthread_cond_signal(&cond_write);
165
166         if (all_work_added && todo_done == todo_end)
167                 pthread_cond_signal(&cond_result);
168
169         grep_unlock();
170 }
171
172 static void *run(void *arg)
173 {
174         int hit = 0;
175         struct grep_opt *opt = arg;
176
177         while (1) {
178                 struct work_item *w = get_work();
179                 if (!w)
180                         break;
181
182                 opt->output_priv = w;
183                 hit |= grep_source(opt, &w->source);
184                 grep_source_clear_data(&w->source);
185                 work_done(w);
186         }
187         free_grep_patterns(arg);
188         free(arg);
189
190         return (void*) (intptr_t) hit;
191 }
192
193 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
194 {
195         struct work_item *w = opt->output_priv;
196         strbuf_add(&w->out, buf, size);
197 }
198
199 static void start_threads(struct grep_opt *opt)
200 {
201         int i;
202
203         pthread_mutex_init(&grep_mutex, NULL);
204         pthread_mutex_init(&grep_read_mutex, NULL);
205         pthread_mutex_init(&grep_attr_mutex, NULL);
206         pthread_cond_init(&cond_add, NULL);
207         pthread_cond_init(&cond_write, NULL);
208         pthread_cond_init(&cond_result, NULL);
209         grep_use_locks = 1;
210
211         for (i = 0; i < ARRAY_SIZE(todo); i++) {
212                 strbuf_init(&todo[i].out, 0);
213         }
214
215         threads = xcalloc(num_threads, sizeof(*threads));
216         for (i = 0; i < num_threads; i++) {
217                 int err;
218                 struct grep_opt *o = grep_opt_dup(opt);
219                 o->output = strbuf_out;
220                 if (i)
221                         o->debug = 0;
222                 compile_grep_patterns(o);
223                 err = pthread_create(&threads[i], NULL, run, o);
224
225                 if (err)
226                         die(_("grep: failed to create thread: %s"),
227                             strerror(err));
228         }
229 }
230
231 static int wait_all(void)
232 {
233         int hit = 0;
234         int i;
235
236         grep_lock();
237         all_work_added = 1;
238
239         /* Wait until all work is done. */
240         while (todo_done != todo_end)
241                 pthread_cond_wait(&cond_result, &grep_mutex);
242
243         /* Wake up all the consumer threads so they can see that there
244          * is no more work to do.
245          */
246         pthread_cond_broadcast(&cond_add);
247         grep_unlock();
248
249         for (i = 0; i < num_threads; i++) {
250                 void *h;
251                 pthread_join(threads[i], &h);
252                 hit |= (int) (intptr_t) h;
253         }
254
255         free(threads);
256
257         pthread_mutex_destroy(&grep_mutex);
258         pthread_mutex_destroy(&grep_read_mutex);
259         pthread_mutex_destroy(&grep_attr_mutex);
260         pthread_cond_destroy(&cond_add);
261         pthread_cond_destroy(&cond_write);
262         pthread_cond_destroy(&cond_result);
263         grep_use_locks = 0;
264
265         return hit;
266 }
267 #else /* !NO_PTHREADS */
268
269 static int wait_all(void)
270 {
271         return 0;
272 }
273 #endif
274
275 static int grep_cmd_config(const char *var, const char *value, void *cb)
276 {
277         int st = grep_config(var, value, cb);
278         if (git_default_config(var, value, cb) < 0)
279                 st = -1;
280
281         if (!strcmp(var, "grep.threads")) {
282                 num_threads = git_config_int(var, value);
283                 if (num_threads < 0)
284                         die(_("invalid number of threads specified (%d) for %s"),
285                             num_threads, var);
286 #ifdef NO_PTHREADS
287                 else if (num_threads && num_threads != 1) {
288                         /*
289                          * TRANSLATORS: %s is the configuration
290                          * variable for tweaking threads, currently
291                          * grep.threads
292                          */
293                         warning(_("no threads support, ignoring %s"), var);
294                         num_threads = 0;
295                 }
296 #endif
297         }
298
299         if (!strcmp(var, "submodule.recurse"))
300                 recurse_submodules = git_config_bool(var, value);
301
302         return st;
303 }
304
305 static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
306 {
307         void *data;
308
309         grep_read_lock();
310         data = read_sha1_file(oid->hash, type, size);
311         grep_read_unlock();
312         return data;
313 }
314
315 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
316                      const char *filename, int tree_name_len,
317                      const char *path)
318 {
319         struct strbuf pathbuf = STRBUF_INIT;
320
321         if (opt->relative && opt->prefix_length) {
322                 quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
323                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
324         } else {
325                 strbuf_addstr(&pathbuf, filename);
326         }
327
328 #ifndef NO_PTHREADS
329         if (num_threads) {
330                 add_work(opt, GREP_SOURCE_OID, pathbuf.buf, path, oid);
331                 strbuf_release(&pathbuf);
332                 return 0;
333         } else
334 #endif
335         {
336                 struct grep_source gs;
337                 int hit;
338
339                 grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
340                 strbuf_release(&pathbuf);
341                 hit = grep_source(opt, &gs);
342
343                 grep_source_clear(&gs);
344                 return hit;
345         }
346 }
347
348 static int grep_file(struct grep_opt *opt, const char *filename)
349 {
350         struct strbuf buf = STRBUF_INIT;
351
352         if (opt->relative && opt->prefix_length)
353                 quote_path_relative(filename, opt->prefix, &buf);
354         else
355                 strbuf_addstr(&buf, filename);
356
357 #ifndef NO_PTHREADS
358         if (num_threads) {
359                 add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
360                 strbuf_release(&buf);
361                 return 0;
362         } else
363 #endif
364         {
365                 struct grep_source gs;
366                 int hit;
367
368                 grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
369                 strbuf_release(&buf);
370                 hit = grep_source(opt, &gs);
371
372                 grep_source_clear(&gs);
373                 return hit;
374         }
375 }
376
377 static void append_path(struct grep_opt *opt, const void *data, size_t len)
378 {
379         struct string_list *path_list = opt->output_priv;
380
381         if (len == 1 && *(const char *)data == '\0')
382                 return;
383         string_list_append(path_list, xstrndup(data, len));
384 }
385
386 static void run_pager(struct grep_opt *opt, const char *prefix)
387 {
388         struct string_list *path_list = opt->output_priv;
389         struct child_process child = CHILD_PROCESS_INIT;
390         int i, status;
391
392         for (i = 0; i < path_list->nr; i++)
393                 argv_array_push(&child.args, path_list->items[i].string);
394         child.dir = prefix;
395         child.use_shell = 1;
396
397         status = run_command(&child);
398         if (status)
399                 exit(status);
400 }
401
402 static int grep_cache(struct grep_opt *opt, struct repository *repo,
403                       const struct pathspec *pathspec, int cached);
404 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
405                      struct tree_desc *tree, struct strbuf *base, int tn_len,
406                      int check_attr, struct repository *repo);
407
408 static int grep_submodule(struct grep_opt *opt, struct repository *superproject,
409                           const struct pathspec *pathspec,
410                           const struct object_id *oid,
411                           const char *filename, const char *path)
412 {
413         struct repository submodule;
414         int hit;
415
416         if (!is_submodule_active(superproject, path))
417                 return 0;
418
419         if (repo_submodule_init(&submodule, superproject, path))
420                 return 0;
421
422         repo_read_gitmodules(&submodule);
423
424         /*
425          * NEEDSWORK: This adds the submodule's object directory to the list of
426          * alternates for the single in-memory object store.  This has some bad
427          * consequences for memory (processed objects will never be freed) and
428          * performance (this increases the number of pack files git has to pay
429          * attention to, to the sum of the number of pack files in all the
430          * repositories processed so far).  This can be removed once the object
431          * store is no longer global and instead is a member of the repository
432          * object.
433          */
434         add_to_alternates_memory(submodule.objectdir);
435
436         if (oid) {
437                 struct object *object;
438                 struct tree_desc tree;
439                 void *data;
440                 unsigned long size;
441                 struct strbuf base = STRBUF_INIT;
442
443                 object = parse_object_or_die(oid, oid_to_hex(oid));
444
445                 grep_read_lock();
446                 data = read_object_with_reference(object->oid.hash, tree_type,
447                                                   &size, NULL);
448                 grep_read_unlock();
449
450                 if (!data)
451                         die(_("unable to read tree (%s)"), oid_to_hex(&object->oid));
452
453                 strbuf_addstr(&base, filename);
454                 strbuf_addch(&base, '/');
455
456                 init_tree_desc(&tree, data, size);
457                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
458                                 object->type == OBJ_COMMIT, &submodule);
459                 strbuf_release(&base);
460                 free(data);
461         } else {
462                 hit = grep_cache(opt, &submodule, pathspec, 1);
463         }
464
465         repo_clear(&submodule);
466         return hit;
467 }
468
469 static int grep_cache(struct grep_opt *opt, struct repository *repo,
470                       const struct pathspec *pathspec, int cached)
471 {
472         int hit = 0;
473         int nr;
474         struct strbuf name = STRBUF_INIT;
475         int name_base_len = 0;
476         if (repo->submodule_prefix) {
477                 name_base_len = strlen(repo->submodule_prefix);
478                 strbuf_addstr(&name, repo->submodule_prefix);
479         }
480
481         repo_read_index(repo);
482
483         for (nr = 0; nr < repo->index->cache_nr; nr++) {
484                 const struct cache_entry *ce = repo->index->cache[nr];
485                 strbuf_setlen(&name, name_base_len);
486                 strbuf_addstr(&name, ce->name);
487
488                 if (S_ISREG(ce->ce_mode) &&
489                     match_pathspec(pathspec, name.buf, name.len, 0, NULL,
490                                    S_ISDIR(ce->ce_mode) ||
491                                    S_ISGITLINK(ce->ce_mode))) {
492                         /*
493                          * If CE_VALID is on, we assume worktree file and its
494                          * cache entry are identical, even if worktree file has
495                          * been modified, so use cache version instead
496                          */
497                         if (cached || (ce->ce_flags & CE_VALID) ||
498                             ce_skip_worktree(ce)) {
499                                 if (ce_stage(ce) || ce_intent_to_add(ce))
500                                         continue;
501                                 hit |= grep_oid(opt, &ce->oid, name.buf,
502                                                  0, name.buf);
503                         } else {
504                                 hit |= grep_file(opt, name.buf);
505                         }
506                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
507                            submodule_path_match(pathspec, name.buf, NULL)) {
508                         hit |= grep_submodule(opt, repo, pathspec, NULL, ce->name, ce->name);
509                 } else {
510                         continue;
511                 }
512
513                 if (ce_stage(ce)) {
514                         do {
515                                 nr++;
516                         } while (nr < repo->index->cache_nr &&
517                                  !strcmp(ce->name, repo->index->cache[nr]->name));
518                         nr--; /* compensate for loop control */
519                 }
520                 if (hit && opt->status_only)
521                         break;
522         }
523
524         strbuf_release(&name);
525         return hit;
526 }
527
528 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
529                      struct tree_desc *tree, struct strbuf *base, int tn_len,
530                      int check_attr, struct repository *repo)
531 {
532         int hit = 0;
533         enum interesting match = entry_not_interesting;
534         struct name_entry entry;
535         int old_baselen = base->len;
536         struct strbuf name = STRBUF_INIT;
537         int name_base_len = 0;
538         if (repo->submodule_prefix) {
539                 strbuf_addstr(&name, repo->submodule_prefix);
540                 name_base_len = name.len;
541         }
542
543         while (tree_entry(tree, &entry)) {
544                 int te_len = tree_entry_len(&entry);
545
546                 if (match != all_entries_interesting) {
547                         strbuf_addstr(&name, base->buf + tn_len);
548                         match = tree_entry_interesting(&entry, &name,
549                                                        0, pathspec);
550                         strbuf_setlen(&name, name_base_len);
551
552                         if (match == all_entries_not_interesting)
553                                 break;
554                         if (match == entry_not_interesting)
555                                 continue;
556                 }
557
558                 strbuf_add(base, entry.path, te_len);
559
560                 if (S_ISREG(entry.mode)) {
561                         hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
562                                          check_attr ? base->buf + tn_len : NULL);
563                 } else if (S_ISDIR(entry.mode)) {
564                         enum object_type type;
565                         struct tree_desc sub;
566                         void *data;
567                         unsigned long size;
568
569                         data = lock_and_read_oid_file(entry.oid, &type, &size);
570                         if (!data)
571                                 die(_("unable to read tree (%s)"),
572                                     oid_to_hex(entry.oid));
573
574                         strbuf_addch(base, '/');
575                         init_tree_desc(&sub, data, size);
576                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
577                                          check_attr, repo);
578                         free(data);
579                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
580                         hit |= grep_submodule(opt, repo, pathspec, entry.oid,
581                                               base->buf, base->buf + tn_len);
582                 }
583
584                 strbuf_setlen(base, old_baselen);
585
586                 if (hit && opt->status_only)
587                         break;
588         }
589
590         strbuf_release(&name);
591         return hit;
592 }
593
594 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
595                        struct object *obj, const char *name, const char *path,
596                        struct repository *repo)
597 {
598         if (obj->type == OBJ_BLOB)
599                 return grep_oid(opt, &obj->oid, name, 0, path);
600         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
601                 struct tree_desc tree;
602                 void *data;
603                 unsigned long size;
604                 struct strbuf base;
605                 int hit, len;
606
607                 grep_read_lock();
608                 data = read_object_with_reference(obj->oid.hash, tree_type,
609                                                   &size, NULL);
610                 grep_read_unlock();
611
612                 if (!data)
613                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
614
615                 len = name ? strlen(name) : 0;
616                 strbuf_init(&base, PATH_MAX + len + 1);
617                 if (len) {
618                         strbuf_add(&base, name, len);
619                         strbuf_addch(&base, ':');
620                 }
621                 init_tree_desc(&tree, data, size);
622                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
623                                 obj->type == OBJ_COMMIT, repo);
624                 strbuf_release(&base);
625                 free(data);
626                 return hit;
627         }
628         die(_("unable to grep from object of type %s"), typename(obj->type));
629 }
630
631 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
632                         struct repository *repo,
633                         const struct object_array *list)
634 {
635         unsigned int i;
636         int hit = 0;
637         const unsigned int nr = list->nr;
638
639         for (i = 0; i < nr; i++) {
640                 struct object *real_obj;
641                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
642
643                 /* load the gitmodules file for this rev */
644                 if (recurse_submodules) {
645                         submodule_free();
646                         gitmodules_config_oid(&real_obj->oid);
647                 }
648                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path,
649                                 repo)) {
650                         hit = 1;
651                         if (opt->status_only)
652                                 break;
653                 }
654         }
655         return hit;
656 }
657
658 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
659                           int exc_std, int use_index)
660 {
661         struct dir_struct dir;
662         int i, hit = 0;
663
664         memset(&dir, 0, sizeof(dir));
665         if (!use_index)
666                 dir.flags |= DIR_NO_GITLINKS;
667         if (exc_std)
668                 setup_standard_excludes(&dir);
669
670         fill_directory(&dir, &the_index, pathspec);
671         for (i = 0; i < dir.nr; i++) {
672                 if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
673                         continue;
674                 hit |= grep_file(opt, dir.entries[i]->name);
675                 if (hit && opt->status_only)
676                         break;
677         }
678         return hit;
679 }
680
681 static int context_callback(const struct option *opt, const char *arg,
682                             int unset)
683 {
684         struct grep_opt *grep_opt = opt->value;
685         int value;
686         const char *endp;
687
688         if (unset) {
689                 grep_opt->pre_context = grep_opt->post_context = 0;
690                 return 0;
691         }
692         value = strtol(arg, (char **)&endp, 10);
693         if (*endp) {
694                 return error(_("switch `%c' expects a numerical value"),
695                              opt->short_name);
696         }
697         grep_opt->pre_context = grep_opt->post_context = value;
698         return 0;
699 }
700
701 static int file_callback(const struct option *opt, const char *arg, int unset)
702 {
703         struct grep_opt *grep_opt = opt->value;
704         int from_stdin = !strcmp(arg, "-");
705         FILE *patterns;
706         int lno = 0;
707         struct strbuf sb = STRBUF_INIT;
708
709         patterns = from_stdin ? stdin : fopen(arg, "r");
710         if (!patterns)
711                 die_errno(_("cannot open '%s'"), arg);
712         while (strbuf_getline(&sb, patterns) == 0) {
713                 /* ignore empty line like grep does */
714                 if (sb.len == 0)
715                         continue;
716
717                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
718                                 GREP_PATTERN);
719         }
720         if (!from_stdin)
721                 fclose(patterns);
722         strbuf_release(&sb);
723         return 0;
724 }
725
726 static int not_callback(const struct option *opt, const char *arg, int unset)
727 {
728         struct grep_opt *grep_opt = opt->value;
729         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
730         return 0;
731 }
732
733 static int and_callback(const struct option *opt, const char *arg, int unset)
734 {
735         struct grep_opt *grep_opt = opt->value;
736         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
737         return 0;
738 }
739
740 static int open_callback(const struct option *opt, const char *arg, int unset)
741 {
742         struct grep_opt *grep_opt = opt->value;
743         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
744         return 0;
745 }
746
747 static int close_callback(const struct option *opt, const char *arg, int unset)
748 {
749         struct grep_opt *grep_opt = opt->value;
750         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
751         return 0;
752 }
753
754 static int pattern_callback(const struct option *opt, const char *arg,
755                             int unset)
756 {
757         struct grep_opt *grep_opt = opt->value;
758         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
759         return 0;
760 }
761
762 int cmd_grep(int argc, const char **argv, const char *prefix)
763 {
764         int hit = 0;
765         int cached = 0, untracked = 0, opt_exclude = -1;
766         int seen_dashdash = 0;
767         int external_grep_allowed__ignored;
768         const char *show_in_pager = NULL, *default_pager = "dummy";
769         struct grep_opt opt;
770         struct object_array list = OBJECT_ARRAY_INIT;
771         struct pathspec pathspec;
772         struct string_list path_list = STRING_LIST_INIT_NODUP;
773         int i;
774         int dummy;
775         int use_index = 1;
776         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
777         int allow_revs;
778
779         struct option options[] = {
780                 OPT_BOOL(0, "cached", &cached,
781                         N_("search in index instead of in the work tree")),
782                 OPT_NEGBIT(0, "no-index", &use_index,
783                          N_("find in contents not managed by git"), 1),
784                 OPT_BOOL(0, "untracked", &untracked,
785                         N_("search in both tracked and untracked files")),
786                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
787                             N_("ignore files specified via '.gitignore'"), 1),
788                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
789                          N_("recursively search in each submodule")),
790                 OPT_GROUP(""),
791                 OPT_BOOL('v', "invert-match", &opt.invert,
792                         N_("show non-matching lines")),
793                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
794                         N_("case insensitive matching")),
795                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
796                         N_("match patterns only at word boundaries")),
797                 OPT_SET_INT('a', "text", &opt.binary,
798                         N_("process binary files as text"), GREP_BINARY_TEXT),
799                 OPT_SET_INT('I', NULL, &opt.binary,
800                         N_("don't match patterns in binary files"),
801                         GREP_BINARY_NOMATCH),
802                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
803                          N_("process binary files with textconv filters")),
804                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
805                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
806                         NULL, 1 },
807                 OPT_GROUP(""),
808                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
809                             N_("use extended POSIX regular expressions"),
810                             GREP_PATTERN_TYPE_ERE),
811                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
812                             N_("use basic POSIX regular expressions (default)"),
813                             GREP_PATTERN_TYPE_BRE),
814                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
815                             N_("interpret patterns as fixed strings"),
816                             GREP_PATTERN_TYPE_FIXED),
817                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
818                             N_("use Perl-compatible regular expressions"),
819                             GREP_PATTERN_TYPE_PCRE),
820                 OPT_GROUP(""),
821                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
822                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
823                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
824                 OPT_NEGBIT(0, "full-name", &opt.relative,
825                         N_("show filenames relative to top directory"), 1),
826                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
827                         N_("show only filenames instead of matching lines")),
828                 OPT_BOOL(0, "name-only", &opt.name_only,
829                         N_("synonym for --files-with-matches")),
830                 OPT_BOOL('L', "files-without-match",
831                         &opt.unmatch_name_only,
832                         N_("show only the names of files without match")),
833                 OPT_BOOL('z', "null", &opt.null_following_name,
834                         N_("print NUL after filenames")),
835                 OPT_BOOL('c', "count", &opt.count,
836                         N_("show the number of matches instead of matching lines")),
837                 OPT__COLOR(&opt.color, N_("highlight matches")),
838                 OPT_BOOL(0, "break", &opt.file_break,
839                         N_("print empty line between matches from different files")),
840                 OPT_BOOL(0, "heading", &opt.heading,
841                         N_("show filename only once above matches from same file")),
842                 OPT_GROUP(""),
843                 OPT_CALLBACK('C', "context", &opt, N_("n"),
844                         N_("show <n> context lines before and after matches"),
845                         context_callback),
846                 OPT_INTEGER('B', "before-context", &opt.pre_context,
847                         N_("show <n> context lines before matches")),
848                 OPT_INTEGER('A', "after-context", &opt.post_context,
849                         N_("show <n> context lines after matches")),
850                 OPT_INTEGER(0, "threads", &num_threads,
851                         N_("use <n> worker threads")),
852                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
853                         context_callback),
854                 OPT_BOOL('p', "show-function", &opt.funcname,
855                         N_("show a line with the function name before matches")),
856                 OPT_BOOL('W', "function-context", &opt.funcbody,
857                         N_("show the surrounding function")),
858                 OPT_GROUP(""),
859                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
860                         N_("read patterns from file"), file_callback),
861                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
862                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
863                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
864                   N_("combine patterns specified with -e"),
865                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
866                 OPT_BOOL(0, "or", &dummy, ""),
867                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
868                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
869                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
870                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
871                   open_callback },
872                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
873                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
874                   close_callback },
875                 OPT__QUIET(&opt.status_only,
876                            N_("indicate hit with exit status without output")),
877                 OPT_BOOL(0, "all-match", &opt.all_match,
878                         N_("show only matches from files that match all patterns")),
879                 { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
880                   N_("show parse tree for grep expression"),
881                   PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
882                 OPT_GROUP(""),
883                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
884                         N_("pager"), N_("show matching files in the pager"),
885                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
886                 OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
887                          N_("allow calling of grep(1) (ignored by this build)")),
888                 OPT_END()
889         };
890
891         init_grep_defaults();
892         git_config(grep_cmd_config, NULL);
893         grep_init(&opt, prefix);
894
895         /*
896          * If there is no -- then the paths must exist in the working
897          * tree.  If there is no explicit pattern specified with -e or
898          * -f, we take the first unrecognized non option to be the
899          * pattern, but then what follows it must be zero or more
900          * valid refs up to the -- (if exists), and then existing
901          * paths.  If there is an explicit pattern, then the first
902          * unrecognized non option is the beginning of the refs list
903          * that continues up to the -- (if exists), and then paths.
904          */
905         argc = parse_options(argc, argv, prefix, options, grep_usage,
906                              PARSE_OPT_KEEP_DASHDASH |
907                              PARSE_OPT_STOP_AT_NON_OPTION);
908         grep_commit_pattern_type(pattern_type_arg, &opt);
909
910         if (use_index && !startup_info->have_repository) {
911                 int fallback = 0;
912                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
913                 if (fallback)
914                         use_index = 0;
915                 else
916                         /* die the same way as if we did it at the beginning */
917                         setup_git_directory();
918         }
919
920         /*
921          * skip a -- separator; we know it cannot be
922          * separating revisions from pathnames if
923          * we haven't even had any patterns yet
924          */
925         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
926                 argv++;
927                 argc--;
928         }
929
930         /* First unrecognized non-option token */
931         if (argc > 0 && !opt.pattern_list) {
932                 append_grep_pattern(&opt, argv[0], "command line", 0,
933                                     GREP_PATTERN);
934                 argv++;
935                 argc--;
936         }
937
938         if (show_in_pager == default_pager)
939                 show_in_pager = git_pager(1);
940         if (show_in_pager) {
941                 opt.color = 0;
942                 opt.name_only = 1;
943                 opt.null_following_name = 1;
944                 opt.output_priv = &path_list;
945                 opt.output = append_path;
946                 string_list_append(&path_list, show_in_pager);
947         }
948
949         if (!opt.pattern_list)
950                 die(_("no pattern given."));
951
952         /*
953          * We have to find "--" in a separate pass, because its presence
954          * influences how we will parse arguments that come before it.
955          */
956         for (i = 0; i < argc; i++) {
957                 if (!strcmp(argv[i], "--")) {
958                         seen_dashdash = 1;
959                         break;
960                 }
961         }
962
963         /*
964          * Resolve any rev arguments. If we have a dashdash, then everything up
965          * to it must resolve as a rev. If not, then we stop at the first
966          * non-rev and assume everything else is a path.
967          */
968         allow_revs = use_index && !untracked;
969         for (i = 0; i < argc; i++) {
970                 const char *arg = argv[i];
971                 struct object_id oid;
972                 struct object_context oc;
973                 struct object *object;
974
975                 if (!strcmp(arg, "--")) {
976                         i++;
977                         break;
978                 }
979
980                 if (!allow_revs) {
981                         if (seen_dashdash)
982                                 die(_("--no-index or --untracked cannot be used with revs"));
983                         break;
984                 }
985
986                 if (get_oid_with_context(arg, GET_OID_RECORD_PATH,
987                                          &oid, &oc)) {
988                         if (seen_dashdash)
989                                 die(_("unable to resolve revision: %s"), arg);
990                         break;
991                 }
992
993                 object = parse_object_or_die(&oid, arg);
994                 if (!seen_dashdash)
995                         verify_non_filename(prefix, arg);
996                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
997                 free(oc.path);
998         }
999
1000         /*
1001          * Anything left over is presumed to be a path. But in the non-dashdash
1002          * "do what I mean" case, we verify and complain when that isn't true.
1003          */
1004         if (!seen_dashdash) {
1005                 int j;
1006                 for (j = i; j < argc; j++)
1007                         verify_filename(prefix, argv[j], j == i && allow_revs);
1008         }
1009
1010         parse_pathspec(&pathspec, 0,
1011                        PATHSPEC_PREFER_CWD |
1012                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1013                        prefix, argv + i);
1014         pathspec.max_depth = opt.max_depth;
1015         pathspec.recursive = 1;
1016
1017 #ifndef NO_PTHREADS
1018         if (list.nr || cached || show_in_pager)
1019                 num_threads = 0;
1020         else if (num_threads == 0)
1021                 num_threads = GREP_NUM_THREADS_DEFAULT;
1022         else if (num_threads < 0)
1023                 die(_("invalid number of threads specified (%d)"), num_threads);
1024         if (num_threads == 1)
1025                 num_threads = 0;
1026 #else
1027         if (num_threads)
1028                 warning(_("no threads support, ignoring --threads"));
1029         num_threads = 0;
1030 #endif
1031
1032         if (!num_threads)
1033                 /*
1034                  * The compiled patterns on the main path are only
1035                  * used when not using threading. Otherwise
1036                  * start_threads() below calls compile_grep_patterns()
1037                  * for each thread.
1038                  */
1039                 compile_grep_patterns(&opt);
1040
1041 #ifndef NO_PTHREADS
1042         if (num_threads) {
1043                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1044                     && (opt.pre_context || opt.post_context ||
1045                         opt.file_break || opt.funcbody))
1046                         skip_first_line = 1;
1047                 start_threads(&opt);
1048         }
1049 #endif
1050
1051         if (recurse_submodules) {
1052                 gitmodules_config();
1053         }
1054
1055         if (show_in_pager && (cached || list.nr))
1056                 die(_("--open-files-in-pager only works on the worktree"));
1057
1058         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1059                 const char *pager = path_list.items[0].string;
1060                 int len = strlen(pager);
1061
1062                 if (len > 4 && is_dir_sep(pager[len - 5]))
1063                         pager += len - 4;
1064
1065                 if (opt.ignore_case && !strcmp("less", pager))
1066                         string_list_append(&path_list, "-I");
1067
1068                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1069                         struct strbuf buf = STRBUF_INIT;
1070                         strbuf_addf(&buf, "+/%s%s",
1071                                         strcmp("less", pager) ? "" : "*",
1072                                         opt.pattern_list->pattern);
1073                         string_list_append(&path_list, buf.buf);
1074                         strbuf_detach(&buf, NULL);
1075                 }
1076         }
1077
1078         if (recurse_submodules && (!use_index || untracked))
1079                 die(_("option not supported with --recurse-submodules."));
1080
1081         if (!show_in_pager && !opt.status_only)
1082                 setup_pager();
1083
1084         if (!use_index && (untracked || cached))
1085                 die(_("--cached or --untracked cannot be used with --no-index."));
1086
1087         if (!use_index || untracked) {
1088                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1089                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1090         } else if (0 <= opt_exclude) {
1091                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1092         } else if (!list.nr) {
1093                 if (!cached)
1094                         setup_work_tree();
1095
1096                 hit = grep_cache(&opt, the_repository, &pathspec, cached);
1097         } else {
1098                 if (cached)
1099                         die(_("both --cached and trees are given."));
1100
1101                 hit = grep_objects(&opt, &pathspec, the_repository, &list);
1102         }
1103
1104         if (num_threads)
1105                 hit |= wait_all();
1106         if (hit && show_in_pager)
1107                 run_pager(&opt, prefix);
1108         clear_pathspec(&pathspec);
1109         free_grep_patterns(&opt);
1110         return !hit;
1111 }