Merge branch 'jk/no-looking-at-dotgit-outside-repo-final'
[git] / builtin / grep.c
1 /*
2  * Builtin "git grep"
3  *
4  * Copyright (c) 2006 Junio C Hamano
5  */
6 #include "cache.h"
7 #include "blob.h"
8 #include "tree.h"
9 #include "commit.h"
10 #include "tag.h"
11 #include "tree-walk.h"
12 #include "builtin.h"
13 #include "parse-options.h"
14 #include "string-list.h"
15 #include "run-command.h"
16 #include "userdiff.h"
17 #include "grep.h"
18 #include "quote.h"
19 #include "dir.h"
20 #include "pathspec.h"
21 #include "submodule.h"
22 #include "submodule-config.h"
23
24 static char const * const grep_usage[] = {
25         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
26         NULL
27 };
28
29 static const char *super_prefix;
30 static int recurse_submodules;
31 static struct argv_array submodule_options = ARGV_ARRAY_INIT;
32 static const char *parent_basename;
33
34 static int grep_submodule_launch(struct grep_opt *opt,
35                                  const struct grep_source *gs);
36
37 #define GREP_NUM_THREADS_DEFAULT 8
38 static int num_threads;
39
40 #ifndef NO_PTHREADS
41 static pthread_t *threads;
42
43 /* We use one producer thread and THREADS consumer
44  * threads. The producer adds struct work_items to 'todo' and the
45  * consumers pick work items from the same array.
46  */
47 struct work_item {
48         struct grep_source source;
49         char done;
50         struct strbuf out;
51 };
52
53 /* In the range [todo_done, todo_start) in 'todo' we have work_items
54  * that have been or are processed by a consumer thread. We haven't
55  * written the result for these to stdout yet.
56  *
57  * The work_items in [todo_start, todo_end) are waiting to be picked
58  * up by a consumer thread.
59  *
60  * The ranges are modulo TODO_SIZE.
61  */
62 #define TODO_SIZE 128
63 static struct work_item todo[TODO_SIZE];
64 static int todo_start;
65 static int todo_end;
66 static int todo_done;
67
68 /* Has all work items been added? */
69 static int all_work_added;
70
71 /* This lock protects all the variables above. */
72 static pthread_mutex_t grep_mutex;
73
74 static inline void grep_lock(void)
75 {
76         if (num_threads)
77                 pthread_mutex_lock(&grep_mutex);
78 }
79
80 static inline void grep_unlock(void)
81 {
82         if (num_threads)
83                 pthread_mutex_unlock(&grep_mutex);
84 }
85
86 /* Signalled when a new work_item is added to todo. */
87 static pthread_cond_t cond_add;
88
89 /* Signalled when the result from one work_item is written to
90  * stdout.
91  */
92 static pthread_cond_t cond_write;
93
94 /* Signalled when we are finished with everything. */
95 static pthread_cond_t cond_result;
96
97 static int skip_first_line;
98
99 static void add_work(struct grep_opt *opt, enum grep_source_type type,
100                      const char *name, const char *path, const void *id)
101 {
102         grep_lock();
103
104         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
105                 pthread_cond_wait(&cond_write, &grep_mutex);
106         }
107
108         grep_source_init(&todo[todo_end].source, type, name, path, id);
109         if (opt->binary != GREP_BINARY_TEXT)
110                 grep_source_load_driver(&todo[todo_end].source);
111         todo[todo_end].done = 0;
112         strbuf_reset(&todo[todo_end].out);
113         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
114
115         pthread_cond_signal(&cond_add);
116         grep_unlock();
117 }
118
119 static struct work_item *get_work(void)
120 {
121         struct work_item *ret;
122
123         grep_lock();
124         while (todo_start == todo_end && !all_work_added) {
125                 pthread_cond_wait(&cond_add, &grep_mutex);
126         }
127
128         if (todo_start == todo_end && all_work_added) {
129                 ret = NULL;
130         } else {
131                 ret = &todo[todo_start];
132                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
133         }
134         grep_unlock();
135         return ret;
136 }
137
138 static void work_done(struct work_item *w)
139 {
140         int old_done;
141
142         grep_lock();
143         w->done = 1;
144         old_done = todo_done;
145         for(; todo[todo_done].done && todo_done != todo_start;
146             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
147                 w = &todo[todo_done];
148                 if (w->out.len) {
149                         const char *p = w->out.buf;
150                         size_t len = w->out.len;
151
152                         /* Skip the leading hunk mark of the first file. */
153                         if (skip_first_line) {
154                                 while (len) {
155                                         len--;
156                                         if (*p++ == '\n')
157                                                 break;
158                                 }
159                                 skip_first_line = 0;
160                         }
161
162                         write_or_die(1, p, len);
163                 }
164                 grep_source_clear(&w->source);
165         }
166
167         if (old_done != todo_done)
168                 pthread_cond_signal(&cond_write);
169
170         if (all_work_added && todo_done == todo_end)
171                 pthread_cond_signal(&cond_result);
172
173         grep_unlock();
174 }
175
176 static void *run(void *arg)
177 {
178         int hit = 0;
179         struct grep_opt *opt = arg;
180
181         while (1) {
182                 struct work_item *w = get_work();
183                 if (!w)
184                         break;
185
186                 opt->output_priv = w;
187                 if (w->source.type == GREP_SOURCE_SUBMODULE)
188                         hit |= grep_submodule_launch(opt, &w->source);
189                 else
190                         hit |= grep_source(opt, &w->source);
191                 grep_source_clear_data(&w->source);
192                 work_done(w);
193         }
194         free_grep_patterns(arg);
195         free(arg);
196
197         return (void*) (intptr_t) hit;
198 }
199
200 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
201 {
202         struct work_item *w = opt->output_priv;
203         strbuf_add(&w->out, buf, size);
204 }
205
206 static void start_threads(struct grep_opt *opt)
207 {
208         int i;
209
210         pthread_mutex_init(&grep_mutex, NULL);
211         pthread_mutex_init(&grep_read_mutex, NULL);
212         pthread_mutex_init(&grep_attr_mutex, NULL);
213         pthread_cond_init(&cond_add, NULL);
214         pthread_cond_init(&cond_write, NULL);
215         pthread_cond_init(&cond_result, NULL);
216         grep_use_locks = 1;
217
218         for (i = 0; i < ARRAY_SIZE(todo); i++) {
219                 strbuf_init(&todo[i].out, 0);
220         }
221
222         threads = xcalloc(num_threads, sizeof(*threads));
223         for (i = 0; i < num_threads; i++) {
224                 int err;
225                 struct grep_opt *o = grep_opt_dup(opt);
226                 o->output = strbuf_out;
227                 o->debug = 0;
228                 compile_grep_patterns(o);
229                 err = pthread_create(&threads[i], NULL, run, o);
230
231                 if (err)
232                         die(_("grep: failed to create thread: %s"),
233                             strerror(err));
234         }
235 }
236
237 static int wait_all(void)
238 {
239         int hit = 0;
240         int i;
241
242         grep_lock();
243         all_work_added = 1;
244
245         /* Wait until all work is done. */
246         while (todo_done != todo_end)
247                 pthread_cond_wait(&cond_result, &grep_mutex);
248
249         /* Wake up all the consumer threads so they can see that there
250          * is no more work to do.
251          */
252         pthread_cond_broadcast(&cond_add);
253         grep_unlock();
254
255         for (i = 0; i < num_threads; i++) {
256                 void *h;
257                 pthread_join(threads[i], &h);
258                 hit |= (int) (intptr_t) h;
259         }
260
261         free(threads);
262
263         pthread_mutex_destroy(&grep_mutex);
264         pthread_mutex_destroy(&grep_read_mutex);
265         pthread_mutex_destroy(&grep_attr_mutex);
266         pthread_cond_destroy(&cond_add);
267         pthread_cond_destroy(&cond_write);
268         pthread_cond_destroy(&cond_result);
269         grep_use_locks = 0;
270
271         return hit;
272 }
273 #else /* !NO_PTHREADS */
274
275 static int wait_all(void)
276 {
277         return 0;
278 }
279 #endif
280
281 static int grep_cmd_config(const char *var, const char *value, void *cb)
282 {
283         int st = grep_config(var, value, cb);
284         if (git_color_default_config(var, value, cb) < 0)
285                 st = -1;
286
287         if (!strcmp(var, "grep.threads")) {
288                 num_threads = git_config_int(var, value);
289                 if (num_threads < 0)
290                         die(_("invalid number of threads specified (%d) for %s"),
291                             num_threads, var);
292         }
293
294         return st;
295 }
296
297 static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
298 {
299         void *data;
300
301         grep_read_lock();
302         data = read_sha1_file(oid->hash, type, size);
303         grep_read_unlock();
304         return data;
305 }
306
307 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
308                      const char *filename, int tree_name_len,
309                      const char *path)
310 {
311         struct strbuf pathbuf = STRBUF_INIT;
312
313         if (opt->relative && opt->prefix_length) {
314                 quote_path_relative(filename + tree_name_len, opt->prefix, &pathbuf);
315                 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
316         } else if (super_prefix) {
317                 strbuf_add(&pathbuf, filename, tree_name_len);
318                 strbuf_addstr(&pathbuf, super_prefix);
319                 strbuf_addstr(&pathbuf, filename + tree_name_len);
320         } else {
321                 strbuf_addstr(&pathbuf, filename);
322         }
323
324 #ifndef NO_PTHREADS
325         if (num_threads) {
326                 add_work(opt, GREP_SOURCE_SHA1, pathbuf.buf, path, oid);
327                 strbuf_release(&pathbuf);
328                 return 0;
329         } else
330 #endif
331         {
332                 struct grep_source gs;
333                 int hit;
334
335                 grep_source_init(&gs, GREP_SOURCE_SHA1, pathbuf.buf, path, oid);
336                 strbuf_release(&pathbuf);
337                 hit = grep_source(opt, &gs);
338
339                 grep_source_clear(&gs);
340                 return hit;
341         }
342 }
343
344 static int grep_file(struct grep_opt *opt, const char *filename)
345 {
346         struct strbuf buf = STRBUF_INIT;
347
348         if (opt->relative && opt->prefix_length) {
349                 quote_path_relative(filename, opt->prefix, &buf);
350         } else {
351                 if (super_prefix)
352                         strbuf_addstr(&buf, super_prefix);
353                 strbuf_addstr(&buf, filename);
354         }
355
356 #ifndef NO_PTHREADS
357         if (num_threads) {
358                 add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
359                 strbuf_release(&buf);
360                 return 0;
361         } else
362 #endif
363         {
364                 struct grep_source gs;
365                 int hit;
366
367                 grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
368                 strbuf_release(&buf);
369                 hit = grep_source(opt, &gs);
370
371                 grep_source_clear(&gs);
372                 return hit;
373         }
374 }
375
376 static void append_path(struct grep_opt *opt, const void *data, size_t len)
377 {
378         struct string_list *path_list = opt->output_priv;
379
380         if (len == 1 && *(const char *)data == '\0')
381                 return;
382         string_list_append(path_list, xstrndup(data, len));
383 }
384
385 static void run_pager(struct grep_opt *opt, const char *prefix)
386 {
387         struct string_list *path_list = opt->output_priv;
388         struct child_process child = CHILD_PROCESS_INIT;
389         int i, status;
390
391         for (i = 0; i < path_list->nr; i++)
392                 argv_array_push(&child.args, path_list->items[i].string);
393         child.dir = prefix;
394         child.use_shell = 1;
395
396         status = run_command(&child);
397         if (status)
398                 exit(status);
399 }
400
401 static void compile_submodule_options(const struct grep_opt *opt,
402                                       const struct pathspec *pathspec,
403                                       int cached, int untracked,
404                                       int opt_exclude, int use_index,
405                                       int pattern_type_arg)
406 {
407         struct grep_pat *pattern;
408         int i;
409
410         if (recurse_submodules)
411                 argv_array_push(&submodule_options, "--recurse-submodules");
412
413         if (cached)
414                 argv_array_push(&submodule_options, "--cached");
415         if (!use_index)
416                 argv_array_push(&submodule_options, "--no-index");
417         if (untracked)
418                 argv_array_push(&submodule_options, "--untracked");
419         if (opt_exclude > 0)
420                 argv_array_push(&submodule_options, "--exclude-standard");
421
422         if (opt->invert)
423                 argv_array_push(&submodule_options, "-v");
424         if (opt->ignore_case)
425                 argv_array_push(&submodule_options, "-i");
426         if (opt->word_regexp)
427                 argv_array_push(&submodule_options, "-w");
428         switch (opt->binary) {
429         case GREP_BINARY_NOMATCH:
430                 argv_array_push(&submodule_options, "-I");
431                 break;
432         case GREP_BINARY_TEXT:
433                 argv_array_push(&submodule_options, "-a");
434                 break;
435         default:
436                 break;
437         }
438         if (opt->allow_textconv)
439                 argv_array_push(&submodule_options, "--textconv");
440         if (opt->max_depth != -1)
441                 argv_array_pushf(&submodule_options, "--max-depth=%d",
442                                  opt->max_depth);
443         if (opt->linenum)
444                 argv_array_push(&submodule_options, "-n");
445         if (!opt->pathname)
446                 argv_array_push(&submodule_options, "-h");
447         if (!opt->relative)
448                 argv_array_push(&submodule_options, "--full-name");
449         if (opt->name_only)
450                 argv_array_push(&submodule_options, "-l");
451         if (opt->unmatch_name_only)
452                 argv_array_push(&submodule_options, "-L");
453         if (opt->null_following_name)
454                 argv_array_push(&submodule_options, "-z");
455         if (opt->count)
456                 argv_array_push(&submodule_options, "-c");
457         if (opt->file_break)
458                 argv_array_push(&submodule_options, "--break");
459         if (opt->heading)
460                 argv_array_push(&submodule_options, "--heading");
461         if (opt->pre_context)
462                 argv_array_pushf(&submodule_options, "--before-context=%d",
463                                  opt->pre_context);
464         if (opt->post_context)
465                 argv_array_pushf(&submodule_options, "--after-context=%d",
466                                  opt->post_context);
467         if (opt->funcname)
468                 argv_array_push(&submodule_options, "-p");
469         if (opt->funcbody)
470                 argv_array_push(&submodule_options, "-W");
471         if (opt->all_match)
472                 argv_array_push(&submodule_options, "--all-match");
473         if (opt->debug)
474                 argv_array_push(&submodule_options, "--debug");
475         if (opt->status_only)
476                 argv_array_push(&submodule_options, "-q");
477
478         switch (pattern_type_arg) {
479         case GREP_PATTERN_TYPE_BRE:
480                 argv_array_push(&submodule_options, "-G");
481                 break;
482         case GREP_PATTERN_TYPE_ERE:
483                 argv_array_push(&submodule_options, "-E");
484                 break;
485         case GREP_PATTERN_TYPE_FIXED:
486                 argv_array_push(&submodule_options, "-F");
487                 break;
488         case GREP_PATTERN_TYPE_PCRE:
489                 argv_array_push(&submodule_options, "-P");
490                 break;
491         case GREP_PATTERN_TYPE_UNSPECIFIED:
492                 break;
493         }
494
495         for (pattern = opt->pattern_list; pattern != NULL;
496              pattern = pattern->next) {
497                 switch (pattern->token) {
498                 case GREP_PATTERN:
499                         argv_array_pushf(&submodule_options, "-e%s",
500                                          pattern->pattern);
501                         break;
502                 case GREP_AND:
503                 case GREP_OPEN_PAREN:
504                 case GREP_CLOSE_PAREN:
505                 case GREP_NOT:
506                 case GREP_OR:
507                         argv_array_push(&submodule_options, pattern->pattern);
508                         break;
509                 /* BODY and HEAD are not used by git-grep */
510                 case GREP_PATTERN_BODY:
511                 case GREP_PATTERN_HEAD:
512                         break;
513                 }
514         }
515
516         /*
517          * Limit number of threads for child process to use.
518          * This is to prevent potential fork-bomb behavior of git-grep as each
519          * submodule process has its own thread pool.
520          */
521         argv_array_pushf(&submodule_options, "--threads=%d",
522                          (num_threads + 1) / 2);
523
524         /* Add Pathspecs */
525         argv_array_push(&submodule_options, "--");
526         for (i = 0; i < pathspec->nr; i++)
527                 argv_array_push(&submodule_options,
528                                 pathspec->items[i].original);
529 }
530
531 /*
532  * Launch child process to grep contents of a submodule
533  */
534 static int grep_submodule_launch(struct grep_opt *opt,
535                                  const struct grep_source *gs)
536 {
537         struct child_process cp = CHILD_PROCESS_INIT;
538         int status, i;
539         const char *end_of_base;
540         const char *name;
541         struct strbuf child_output = STRBUF_INIT;
542
543         end_of_base = strchr(gs->name, ':');
544         if (gs->identifier && end_of_base)
545                 name = end_of_base + 1;
546         else
547                 name = gs->name;
548
549         prepare_submodule_repo_env(&cp.env_array);
550         argv_array_push(&cp.env_array, GIT_DIR_ENVIRONMENT);
551
552         /* Add super prefix */
553         argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
554                          super_prefix ? super_prefix : "",
555                          name);
556         argv_array_push(&cp.args, "grep");
557
558         /*
559          * Add basename of parent project
560          * When performing grep on a tree object the filename is prefixed
561          * with the object's name: 'tree-name:filename'.  In order to
562          * provide uniformity of output we want to pass the name of the
563          * parent project's object name to the submodule so the submodule can
564          * prefix its output with the parent's name and not its own SHA1.
565          */
566         if (gs->identifier && end_of_base)
567                 argv_array_pushf(&cp.args, "--parent-basename=%.*s",
568                                  (int) (end_of_base - gs->name),
569                                  gs->name);
570
571         /* Add options */
572         for (i = 0; i < submodule_options.argc; i++) {
573                 /*
574                  * If there is a tree identifier for the submodule, add the
575                  * rev after adding the submodule options but before the
576                  * pathspecs.  To do this we listen for the '--' and insert the
577                  * sha1 before pushing the '--' onto the child process argv
578                  * array.
579                  */
580                 if (gs->identifier &&
581                     !strcmp("--", submodule_options.argv[i])) {
582                         argv_array_push(&cp.args, sha1_to_hex(gs->identifier));
583                 }
584
585                 argv_array_push(&cp.args, submodule_options.argv[i]);
586         }
587
588         cp.git_cmd = 1;
589         cp.dir = gs->path;
590
591         /*
592          * Capture output to output buffer and check the return code from the
593          * child process.  A '0' indicates a hit, a '1' indicates no hit and
594          * anything else is an error.
595          */
596         status = capture_command(&cp, &child_output, 0);
597         if (status && (status != 1)) {
598                 /* flush the buffer */
599                 write_or_die(1, child_output.buf, child_output.len);
600                 die("process for submodule '%s' failed with exit code: %d",
601                     gs->name, status);
602         }
603
604         opt->output(opt, child_output.buf, child_output.len);
605         strbuf_release(&child_output);
606         /* invert the return code to make a hit equal to 1 */
607         return !status;
608 }
609
610 /*
611  * Prep grep structures for a submodule grep
612  * sha1: the sha1 of the submodule or NULL if using the working tree
613  * filename: name of the submodule including tree name of parent
614  * path: location of the submodule
615  */
616 static int grep_submodule(struct grep_opt *opt, const unsigned char *sha1,
617                           const char *filename, const char *path)
618 {
619         if (!is_submodule_initialized(path))
620                 return 0;
621         if (!is_submodule_populated_gently(path, NULL)) {
622                 /*
623                  * If searching history, check for the presense of the
624                  * submodule's gitdir before skipping the submodule.
625                  */
626                 if (sha1) {
627                         const struct submodule *sub =
628                                         submodule_from_path(null_sha1, path);
629                         if (sub)
630                                 path = git_path("modules/%s", sub->name);
631
632                         if (!(is_directory(path) && is_git_directory(path)))
633                                 return 0;
634                 } else {
635                         return 0;
636                 }
637         }
638
639 #ifndef NO_PTHREADS
640         if (num_threads) {
641                 add_work(opt, GREP_SOURCE_SUBMODULE, filename, path, sha1);
642                 return 0;
643         } else
644 #endif
645         {
646                 struct grep_source gs;
647                 int hit;
648
649                 grep_source_init(&gs, GREP_SOURCE_SUBMODULE,
650                                  filename, path, sha1);
651                 hit = grep_submodule_launch(opt, &gs);
652
653                 grep_source_clear(&gs);
654                 return hit;
655         }
656 }
657
658 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec,
659                       int cached)
660 {
661         int hit = 0;
662         int nr;
663         struct strbuf name = STRBUF_INIT;
664         int name_base_len = 0;
665         if (super_prefix) {
666                 name_base_len = strlen(super_prefix);
667                 strbuf_addstr(&name, super_prefix);
668         }
669
670         read_cache();
671
672         for (nr = 0; nr < active_nr; nr++) {
673                 const struct cache_entry *ce = active_cache[nr];
674                 strbuf_setlen(&name, name_base_len);
675                 strbuf_addstr(&name, ce->name);
676
677                 if (S_ISREG(ce->ce_mode) &&
678                     match_pathspec(pathspec, name.buf, name.len, 0, NULL,
679                                    S_ISDIR(ce->ce_mode) ||
680                                    S_ISGITLINK(ce->ce_mode))) {
681                         /*
682                          * If CE_VALID is on, we assume worktree file and its
683                          * cache entry are identical, even if worktree file has
684                          * been modified, so use cache version instead
685                          */
686                         if (cached || (ce->ce_flags & CE_VALID) ||
687                             ce_skip_worktree(ce)) {
688                                 if (ce_stage(ce) || ce_intent_to_add(ce))
689                                         continue;
690                                 hit |= grep_oid(opt, &ce->oid, ce->name,
691                                                  0, ce->name);
692                         } else {
693                                 hit |= grep_file(opt, ce->name);
694                         }
695                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
696                            submodule_path_match(pathspec, name.buf, NULL)) {
697                         hit |= grep_submodule(opt, NULL, ce->name, ce->name);
698                 } else {
699                         continue;
700                 }
701
702                 if (ce_stage(ce)) {
703                         do {
704                                 nr++;
705                         } while (nr < active_nr &&
706                                  !strcmp(ce->name, active_cache[nr]->name));
707                         nr--; /* compensate for loop control */
708                 }
709                 if (hit && opt->status_only)
710                         break;
711         }
712
713         strbuf_release(&name);
714         return hit;
715 }
716
717 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
718                      struct tree_desc *tree, struct strbuf *base, int tn_len,
719                      int check_attr)
720 {
721         int hit = 0;
722         enum interesting match = entry_not_interesting;
723         struct name_entry entry;
724         int old_baselen = base->len;
725         struct strbuf name = STRBUF_INIT;
726         int name_base_len = 0;
727         if (super_prefix) {
728                 strbuf_addstr(&name, super_prefix);
729                 name_base_len = name.len;
730         }
731
732         while (tree_entry(tree, &entry)) {
733                 int te_len = tree_entry_len(&entry);
734
735                 if (match != all_entries_interesting) {
736                         strbuf_addstr(&name, base->buf + tn_len);
737                         match = tree_entry_interesting(&entry, &name,
738                                                        0, pathspec);
739                         strbuf_setlen(&name, name_base_len);
740
741                         if (match == all_entries_not_interesting)
742                                 break;
743                         if (match == entry_not_interesting)
744                                 continue;
745                 }
746
747                 strbuf_add(base, entry.path, te_len);
748
749                 if (S_ISREG(entry.mode)) {
750                         hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
751                                          check_attr ? base->buf + tn_len : NULL);
752                 } else if (S_ISDIR(entry.mode)) {
753                         enum object_type type;
754                         struct tree_desc sub;
755                         void *data;
756                         unsigned long size;
757
758                         data = lock_and_read_oid_file(entry.oid, &type, &size);
759                         if (!data)
760                                 die(_("unable to read tree (%s)"),
761                                     oid_to_hex(entry.oid));
762
763                         strbuf_addch(base, '/');
764                         init_tree_desc(&sub, data, size);
765                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
766                                          check_attr);
767                         free(data);
768                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
769                         hit |= grep_submodule(opt, entry.oid->hash, base->buf,
770                                               base->buf + tn_len);
771                 }
772
773                 strbuf_setlen(base, old_baselen);
774
775                 if (hit && opt->status_only)
776                         break;
777         }
778
779         strbuf_release(&name);
780         return hit;
781 }
782
783 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
784                        struct object *obj, const char *name, const char *path)
785 {
786         if (obj->type == OBJ_BLOB)
787                 return grep_oid(opt, &obj->oid, name, 0, path);
788         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
789                 struct tree_desc tree;
790                 void *data;
791                 unsigned long size;
792                 struct strbuf base;
793                 int hit, len;
794
795                 grep_read_lock();
796                 data = read_object_with_reference(obj->oid.hash, tree_type,
797                                                   &size, NULL);
798                 grep_read_unlock();
799
800                 if (!data)
801                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
802
803                 /* Use parent's name as base when recursing submodules */
804                 if (recurse_submodules && parent_basename)
805                         name = parent_basename;
806
807                 len = name ? strlen(name) : 0;
808                 strbuf_init(&base, PATH_MAX + len + 1);
809                 if (len) {
810                         strbuf_add(&base, name, len);
811                         strbuf_addch(&base, ':');
812                 }
813                 init_tree_desc(&tree, data, size);
814                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
815                                 obj->type == OBJ_COMMIT);
816                 strbuf_release(&base);
817                 free(data);
818                 return hit;
819         }
820         die(_("unable to grep from object of type %s"), typename(obj->type));
821 }
822
823 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
824                         const struct object_array *list)
825 {
826         unsigned int i;
827         int hit = 0;
828         const unsigned int nr = list->nr;
829
830         for (i = 0; i < nr; i++) {
831                 struct object *real_obj;
832                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
833
834                 /* load the gitmodules file for this rev */
835                 if (recurse_submodules) {
836                         submodule_free();
837                         gitmodules_config_sha1(real_obj->oid.hash);
838                 }
839                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path)) {
840                         hit = 1;
841                         if (opt->status_only)
842                                 break;
843                 }
844         }
845         return hit;
846 }
847
848 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
849                           int exc_std, int use_index)
850 {
851         struct dir_struct dir;
852         int i, hit = 0;
853
854         memset(&dir, 0, sizeof(dir));
855         if (!use_index)
856                 dir.flags |= DIR_NO_GITLINKS;
857         if (exc_std)
858                 setup_standard_excludes(&dir);
859
860         fill_directory(&dir, pathspec);
861         for (i = 0; i < dir.nr; i++) {
862                 if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
863                         continue;
864                 hit |= grep_file(opt, dir.entries[i]->name);
865                 if (hit && opt->status_only)
866                         break;
867         }
868         return hit;
869 }
870
871 static int context_callback(const struct option *opt, const char *arg,
872                             int unset)
873 {
874         struct grep_opt *grep_opt = opt->value;
875         int value;
876         const char *endp;
877
878         if (unset) {
879                 grep_opt->pre_context = grep_opt->post_context = 0;
880                 return 0;
881         }
882         value = strtol(arg, (char **)&endp, 10);
883         if (*endp) {
884                 return error(_("switch `%c' expects a numerical value"),
885                              opt->short_name);
886         }
887         grep_opt->pre_context = grep_opt->post_context = value;
888         return 0;
889 }
890
891 static int file_callback(const struct option *opt, const char *arg, int unset)
892 {
893         struct grep_opt *grep_opt = opt->value;
894         int from_stdin = !strcmp(arg, "-");
895         FILE *patterns;
896         int lno = 0;
897         struct strbuf sb = STRBUF_INIT;
898
899         patterns = from_stdin ? stdin : fopen(arg, "r");
900         if (!patterns)
901                 die_errno(_("cannot open '%s'"), arg);
902         while (strbuf_getline(&sb, patterns) == 0) {
903                 /* ignore empty line like grep does */
904                 if (sb.len == 0)
905                         continue;
906
907                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
908                                 GREP_PATTERN);
909         }
910         if (!from_stdin)
911                 fclose(patterns);
912         strbuf_release(&sb);
913         return 0;
914 }
915
916 static int not_callback(const struct option *opt, const char *arg, int unset)
917 {
918         struct grep_opt *grep_opt = opt->value;
919         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
920         return 0;
921 }
922
923 static int and_callback(const struct option *opt, const char *arg, int unset)
924 {
925         struct grep_opt *grep_opt = opt->value;
926         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
927         return 0;
928 }
929
930 static int open_callback(const struct option *opt, const char *arg, int unset)
931 {
932         struct grep_opt *grep_opt = opt->value;
933         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
934         return 0;
935 }
936
937 static int close_callback(const struct option *opt, const char *arg, int unset)
938 {
939         struct grep_opt *grep_opt = opt->value;
940         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
941         return 0;
942 }
943
944 static int pattern_callback(const struct option *opt, const char *arg,
945                             int unset)
946 {
947         struct grep_opt *grep_opt = opt->value;
948         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
949         return 0;
950 }
951
952 int cmd_grep(int argc, const char **argv, const char *prefix)
953 {
954         int hit = 0;
955         int cached = 0, untracked = 0, opt_exclude = -1;
956         int seen_dashdash = 0;
957         int external_grep_allowed__ignored;
958         const char *show_in_pager = NULL, *default_pager = "dummy";
959         struct grep_opt opt;
960         struct object_array list = OBJECT_ARRAY_INIT;
961         struct pathspec pathspec;
962         struct string_list path_list = STRING_LIST_INIT_NODUP;
963         int i;
964         int dummy;
965         int use_index = 1;
966         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
967         int allow_revs;
968
969         struct option options[] = {
970                 OPT_BOOL(0, "cached", &cached,
971                         N_("search in index instead of in the work tree")),
972                 OPT_NEGBIT(0, "no-index", &use_index,
973                          N_("find in contents not managed by git"), 1),
974                 OPT_BOOL(0, "untracked", &untracked,
975                         N_("search in both tracked and untracked files")),
976                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
977                             N_("ignore files specified via '.gitignore'"), 1),
978                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
979                          N_("recursivley search in each submodule")),
980                 OPT_STRING(0, "parent-basename", &parent_basename,
981                            N_("basename"),
982                            N_("prepend parent project's basename to output")),
983                 OPT_GROUP(""),
984                 OPT_BOOL('v', "invert-match", &opt.invert,
985                         N_("show non-matching lines")),
986                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
987                         N_("case insensitive matching")),
988                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
989                         N_("match patterns only at word boundaries")),
990                 OPT_SET_INT('a', "text", &opt.binary,
991                         N_("process binary files as text"), GREP_BINARY_TEXT),
992                 OPT_SET_INT('I', NULL, &opt.binary,
993                         N_("don't match patterns in binary files"),
994                         GREP_BINARY_NOMATCH),
995                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
996                          N_("process binary files with textconv filters")),
997                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
998                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
999                         NULL, 1 },
1000                 OPT_GROUP(""),
1001                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
1002                             N_("use extended POSIX regular expressions"),
1003                             GREP_PATTERN_TYPE_ERE),
1004                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
1005                             N_("use basic POSIX regular expressions (default)"),
1006                             GREP_PATTERN_TYPE_BRE),
1007                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
1008                             N_("interpret patterns as fixed strings"),
1009                             GREP_PATTERN_TYPE_FIXED),
1010                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
1011                             N_("use Perl-compatible regular expressions"),
1012                             GREP_PATTERN_TYPE_PCRE),
1013                 OPT_GROUP(""),
1014                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
1015                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
1016                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
1017                 OPT_NEGBIT(0, "full-name", &opt.relative,
1018                         N_("show filenames relative to top directory"), 1),
1019                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
1020                         N_("show only filenames instead of matching lines")),
1021                 OPT_BOOL(0, "name-only", &opt.name_only,
1022                         N_("synonym for --files-with-matches")),
1023                 OPT_BOOL('L', "files-without-match",
1024                         &opt.unmatch_name_only,
1025                         N_("show only the names of files without match")),
1026                 OPT_BOOL('z', "null", &opt.null_following_name,
1027                         N_("print NUL after filenames")),
1028                 OPT_BOOL('c', "count", &opt.count,
1029                         N_("show the number of matches instead of matching lines")),
1030                 OPT__COLOR(&opt.color, N_("highlight matches")),
1031                 OPT_BOOL(0, "break", &opt.file_break,
1032                         N_("print empty line between matches from different files")),
1033                 OPT_BOOL(0, "heading", &opt.heading,
1034                         N_("show filename only once above matches from same file")),
1035                 OPT_GROUP(""),
1036                 OPT_CALLBACK('C', "context", &opt, N_("n"),
1037                         N_("show <n> context lines before and after matches"),
1038                         context_callback),
1039                 OPT_INTEGER('B', "before-context", &opt.pre_context,
1040                         N_("show <n> context lines before matches")),
1041                 OPT_INTEGER('A', "after-context", &opt.post_context,
1042                         N_("show <n> context lines after matches")),
1043                 OPT_INTEGER(0, "threads", &num_threads,
1044                         N_("use <n> worker threads")),
1045                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
1046                         context_callback),
1047                 OPT_BOOL('p', "show-function", &opt.funcname,
1048                         N_("show a line with the function name before matches")),
1049                 OPT_BOOL('W', "function-context", &opt.funcbody,
1050                         N_("show the surrounding function")),
1051                 OPT_GROUP(""),
1052                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
1053                         N_("read patterns from file"), file_callback),
1054                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
1055                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
1056                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
1057                   N_("combine patterns specified with -e"),
1058                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
1059                 OPT_BOOL(0, "or", &dummy, ""),
1060                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
1061                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
1062                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
1063                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1064                   open_callback },
1065                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
1066                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1067                   close_callback },
1068                 OPT__QUIET(&opt.status_only,
1069                            N_("indicate hit with exit status without output")),
1070                 OPT_BOOL(0, "all-match", &opt.all_match,
1071                         N_("show only matches from files that match all patterns")),
1072                 { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
1073                   N_("show parse tree for grep expression"),
1074                   PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
1075                 OPT_GROUP(""),
1076                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
1077                         N_("pager"), N_("show matching files in the pager"),
1078                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
1079                 OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
1080                          N_("allow calling of grep(1) (ignored by this build)")),
1081                 OPT_END()
1082         };
1083
1084         init_grep_defaults();
1085         git_config(grep_cmd_config, NULL);
1086         grep_init(&opt, prefix);
1087         super_prefix = get_super_prefix();
1088
1089         /*
1090          * If there is no -- then the paths must exist in the working
1091          * tree.  If there is no explicit pattern specified with -e or
1092          * -f, we take the first unrecognized non option to be the
1093          * pattern, but then what follows it must be zero or more
1094          * valid refs up to the -- (if exists), and then existing
1095          * paths.  If there is an explicit pattern, then the first
1096          * unrecognized non option is the beginning of the refs list
1097          * that continues up to the -- (if exists), and then paths.
1098          */
1099         argc = parse_options(argc, argv, prefix, options, grep_usage,
1100                              PARSE_OPT_KEEP_DASHDASH |
1101                              PARSE_OPT_STOP_AT_NON_OPTION);
1102         grep_commit_pattern_type(pattern_type_arg, &opt);
1103
1104         if (use_index && !startup_info->have_repository) {
1105                 int fallback = 0;
1106                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
1107                 if (fallback)
1108                         use_index = 0;
1109                 else
1110                         /* die the same way as if we did it at the beginning */
1111                         setup_git_directory();
1112         }
1113
1114         /*
1115          * skip a -- separator; we know it cannot be
1116          * separating revisions from pathnames if
1117          * we haven't even had any patterns yet
1118          */
1119         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
1120                 argv++;
1121                 argc--;
1122         }
1123
1124         /* First unrecognized non-option token */
1125         if (argc > 0 && !opt.pattern_list) {
1126                 append_grep_pattern(&opt, argv[0], "command line", 0,
1127                                     GREP_PATTERN);
1128                 argv++;
1129                 argc--;
1130         }
1131
1132         if (show_in_pager == default_pager)
1133                 show_in_pager = git_pager(1);
1134         if (show_in_pager) {
1135                 opt.color = 0;
1136                 opt.name_only = 1;
1137                 opt.null_following_name = 1;
1138                 opt.output_priv = &path_list;
1139                 opt.output = append_path;
1140                 string_list_append(&path_list, show_in_pager);
1141         }
1142
1143         if (!opt.pattern_list)
1144                 die(_("no pattern given."));
1145         if (!opt.fixed && opt.ignore_case)
1146                 opt.regflags |= REG_ICASE;
1147
1148         compile_grep_patterns(&opt);
1149
1150         /*
1151          * We have to find "--" in a separate pass, because its presence
1152          * influences how we will parse arguments that come before it.
1153          */
1154         for (i = 0; i < argc; i++) {
1155                 if (!strcmp(argv[i], "--")) {
1156                         seen_dashdash = 1;
1157                         break;
1158                 }
1159         }
1160
1161         /*
1162          * Resolve any rev arguments. If we have a dashdash, then everything up
1163          * to it must resolve as a rev. If not, then we stop at the first
1164          * non-rev and assume everything else is a path.
1165          */
1166         allow_revs = use_index && !untracked;
1167         for (i = 0; i < argc; i++) {
1168                 const char *arg = argv[i];
1169                 struct object_id oid;
1170                 struct object_context oc;
1171                 struct object *object;
1172
1173                 if (!strcmp(arg, "--")) {
1174                         i++;
1175                         break;
1176                 }
1177
1178                 if (!allow_revs) {
1179                         if (seen_dashdash)
1180                                 die(_("--no-index or --untracked cannot be used with revs"));
1181                         break;
1182                 }
1183
1184                 if (get_sha1_with_context(arg, 0, oid.hash, &oc)) {
1185                         if (seen_dashdash)
1186                                 die(_("unable to resolve revision: %s"), arg);
1187                         break;
1188                 }
1189
1190                 object = parse_object_or_die(oid.hash, arg);
1191                 if (!seen_dashdash)
1192                         verify_non_filename(prefix, arg);
1193                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1194         }
1195
1196         /*
1197          * Anything left over is presumed to be a path. But in the non-dashdash
1198          * "do what I mean" case, we verify and complain when that isn't true.
1199          */
1200         if (!seen_dashdash) {
1201                 int j;
1202                 for (j = i; j < argc; j++)
1203                         verify_filename(prefix, argv[j], j == i && allow_revs);
1204         }
1205
1206         parse_pathspec(&pathspec, 0,
1207                        PATHSPEC_PREFER_CWD |
1208                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1209                        prefix, argv + i);
1210         pathspec.max_depth = opt.max_depth;
1211         pathspec.recursive = 1;
1212
1213 #ifndef NO_PTHREADS
1214         if (list.nr || cached || show_in_pager)
1215                 num_threads = 0;
1216         else if (num_threads == 0)
1217                 num_threads = GREP_NUM_THREADS_DEFAULT;
1218         else if (num_threads < 0)
1219                 die(_("invalid number of threads specified (%d)"), num_threads);
1220 #else
1221         num_threads = 0;
1222 #endif
1223
1224 #ifndef NO_PTHREADS
1225         if (num_threads) {
1226                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1227                     && (opt.pre_context || opt.post_context ||
1228                         opt.file_break || opt.funcbody))
1229                         skip_first_line = 1;
1230                 start_threads(&opt);
1231         }
1232 #endif
1233
1234         if (recurse_submodules) {
1235                 gitmodules_config();
1236                 compile_submodule_options(&opt, &pathspec, cached, untracked,
1237                                           opt_exclude, use_index,
1238                                           pattern_type_arg);
1239         }
1240
1241         if (show_in_pager && (cached || list.nr))
1242                 die(_("--open-files-in-pager only works on the worktree"));
1243
1244         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1245                 const char *pager = path_list.items[0].string;
1246                 int len = strlen(pager);
1247
1248                 if (len > 4 && is_dir_sep(pager[len - 5]))
1249                         pager += len - 4;
1250
1251                 if (opt.ignore_case && !strcmp("less", pager))
1252                         string_list_append(&path_list, "-I");
1253
1254                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1255                         struct strbuf buf = STRBUF_INIT;
1256                         strbuf_addf(&buf, "+/%s%s",
1257                                         strcmp("less", pager) ? "" : "*",
1258                                         opt.pattern_list->pattern);
1259                         string_list_append(&path_list, buf.buf);
1260                         strbuf_detach(&buf, NULL);
1261                 }
1262         }
1263
1264         if (recurse_submodules && (!use_index || untracked))
1265                 die(_("option not supported with --recurse-submodules."));
1266
1267         if (!show_in_pager && !opt.status_only)
1268                 setup_pager();
1269
1270         if (!use_index && (untracked || cached))
1271                 die(_("--cached or --untracked cannot be used with --no-index."));
1272
1273         if (!use_index || untracked) {
1274                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1275                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1276         } else if (0 <= opt_exclude) {
1277                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1278         } else if (!list.nr) {
1279                 if (!cached)
1280                         setup_work_tree();
1281
1282                 hit = grep_cache(&opt, &pathspec, cached);
1283         } else {
1284                 if (cached)
1285                         die(_("both --cached and trees are given."));
1286                 hit = grep_objects(&opt, &pathspec, &list);
1287         }
1288
1289         if (num_threads)
1290                 hit |= wait_all();
1291         if (hit && show_in_pager)
1292                 run_pager(&opt, prefix);
1293         free_grep_patterns(&opt);
1294         return !hit;
1295 }