config: don't include config.h by default
[git] / builtin / grep.c
1 /*
2  * Builtin "git grep"
3  *
4  * Copyright (c) 2006 Junio C Hamano
5  */
6 #include "cache.h"
7 #include "config.h"
8 #include "blob.h"
9 #include "tree.h"
10 #include "commit.h"
11 #include "tag.h"
12 #include "tree-walk.h"
13 #include "builtin.h"
14 #include "parse-options.h"
15 #include "string-list.h"
16 #include "run-command.h"
17 #include "userdiff.h"
18 #include "grep.h"
19 #include "quote.h"
20 #include "dir.h"
21 #include "pathspec.h"
22 #include "submodule.h"
23 #include "submodule-config.h"
24
25 static char const * const grep_usage[] = {
26         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
27         NULL
28 };
29
30 static const char *super_prefix;
31 static int recurse_submodules;
32 static struct argv_array submodule_options = ARGV_ARRAY_INIT;
33 static const char *parent_basename;
34
35 static int grep_submodule_launch(struct grep_opt *opt,
36                                  const struct grep_source *gs);
37
38 #define GREP_NUM_THREADS_DEFAULT 8
39 static int num_threads;
40
41 #ifndef NO_PTHREADS
42 static pthread_t *threads;
43
44 /* We use one producer thread and THREADS consumer
45  * threads. The producer adds struct work_items to 'todo' and the
46  * consumers pick work items from the same array.
47  */
48 struct work_item {
49         struct grep_source source;
50         char done;
51         struct strbuf out;
52 };
53
54 /* In the range [todo_done, todo_start) in 'todo' we have work_items
55  * that have been or are processed by a consumer thread. We haven't
56  * written the result for these to stdout yet.
57  *
58  * The work_items in [todo_start, todo_end) are waiting to be picked
59  * up by a consumer thread.
60  *
61  * The ranges are modulo TODO_SIZE.
62  */
63 #define TODO_SIZE 128
64 static struct work_item todo[TODO_SIZE];
65 static int todo_start;
66 static int todo_end;
67 static int todo_done;
68
69 /* Has all work items been added? */
70 static int all_work_added;
71
72 /* This lock protects all the variables above. */
73 static pthread_mutex_t grep_mutex;
74
75 static inline void grep_lock(void)
76 {
77         if (num_threads)
78                 pthread_mutex_lock(&grep_mutex);
79 }
80
81 static inline void grep_unlock(void)
82 {
83         if (num_threads)
84                 pthread_mutex_unlock(&grep_mutex);
85 }
86
87 /* Signalled when a new work_item is added to todo. */
88 static pthread_cond_t cond_add;
89
90 /* Signalled when the result from one work_item is written to
91  * stdout.
92  */
93 static pthread_cond_t cond_write;
94
95 /* Signalled when we are finished with everything. */
96 static pthread_cond_t cond_result;
97
98 static int skip_first_line;
99
100 static void add_work(struct grep_opt *opt, enum grep_source_type type,
101                      const char *name, const char *path, const void *id)
102 {
103         grep_lock();
104
105         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
106                 pthread_cond_wait(&cond_write, &grep_mutex);
107         }
108
109         grep_source_init(&todo[todo_end].source, type, name, path, id);
110         if (opt->binary != GREP_BINARY_TEXT)
111                 grep_source_load_driver(&todo[todo_end].source);
112         todo[todo_end].done = 0;
113         strbuf_reset(&todo[todo_end].out);
114         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
115
116         pthread_cond_signal(&cond_add);
117         grep_unlock();
118 }
119
120 static struct work_item *get_work(void)
121 {
122         struct work_item *ret;
123
124         grep_lock();
125         while (todo_start == todo_end && !all_work_added) {
126                 pthread_cond_wait(&cond_add, &grep_mutex);
127         }
128
129         if (todo_start == todo_end && all_work_added) {
130                 ret = NULL;
131         } else {
132                 ret = &todo[todo_start];
133                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
134         }
135         grep_unlock();
136         return ret;
137 }
138
139 static void work_done(struct work_item *w)
140 {
141         int old_done;
142
143         grep_lock();
144         w->done = 1;
145         old_done = todo_done;
146         for(; todo[todo_done].done && todo_done != todo_start;
147             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
148                 w = &todo[todo_done];
149                 if (w->out.len) {
150                         const char *p = w->out.buf;
151                         size_t len = w->out.len;
152
153                         /* Skip the leading hunk mark of the first file. */
154                         if (skip_first_line) {
155                                 while (len) {
156                                         len--;
157                                         if (*p++ == '\n')
158                                                 break;
159                                 }
160                                 skip_first_line = 0;
161                         }
162
163                         write_or_die(1, p, len);
164                 }
165                 grep_source_clear(&w->source);
166         }
167
168         if (old_done != todo_done)
169                 pthread_cond_signal(&cond_write);
170
171         if (all_work_added && todo_done == todo_end)
172                 pthread_cond_signal(&cond_result);
173
174         grep_unlock();
175 }
176
177 static void *run(void *arg)
178 {
179         int hit = 0;
180         struct grep_opt *opt = arg;
181
182         while (1) {
183                 struct work_item *w = get_work();
184                 if (!w)
185                         break;
186
187                 opt->output_priv = w;
188                 if (w->source.type == GREP_SOURCE_SUBMODULE)
189                         hit |= grep_submodule_launch(opt, &w->source);
190                 else
191                         hit |= grep_source(opt, &w->source);
192                 grep_source_clear_data(&w->source);
193                 work_done(w);
194         }
195         free_grep_patterns(arg);
196         free(arg);
197
198         return (void*) (intptr_t) hit;
199 }
200
201 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
202 {
203         struct work_item *w = opt->output_priv;
204         strbuf_add(&w->out, buf, size);
205 }
206
207 static void start_threads(struct grep_opt *opt)
208 {
209         int i;
210
211         pthread_mutex_init(&grep_mutex, NULL);
212         pthread_mutex_init(&grep_read_mutex, NULL);
213         pthread_mutex_init(&grep_attr_mutex, NULL);
214         pthread_cond_init(&cond_add, NULL);
215         pthread_cond_init(&cond_write, NULL);
216         pthread_cond_init(&cond_result, NULL);
217         grep_use_locks = 1;
218
219         for (i = 0; i < ARRAY_SIZE(todo); i++) {
220                 strbuf_init(&todo[i].out, 0);
221         }
222
223         threads = xcalloc(num_threads, sizeof(*threads));
224         for (i = 0; i < num_threads; i++) {
225                 int err;
226                 struct grep_opt *o = grep_opt_dup(opt);
227                 o->output = strbuf_out;
228                 o->debug = 0;
229                 compile_grep_patterns(o);
230                 err = pthread_create(&threads[i], NULL, run, o);
231
232                 if (err)
233                         die(_("grep: failed to create thread: %s"),
234                             strerror(err));
235         }
236 }
237
238 static int wait_all(void)
239 {
240         int hit = 0;
241         int i;
242
243         grep_lock();
244         all_work_added = 1;
245
246         /* Wait until all work is done. */
247         while (todo_done != todo_end)
248                 pthread_cond_wait(&cond_result, &grep_mutex);
249
250         /* Wake up all the consumer threads so they can see that there
251          * is no more work to do.
252          */
253         pthread_cond_broadcast(&cond_add);
254         grep_unlock();
255
256         for (i = 0; i < num_threads; i++) {
257                 void *h;
258                 pthread_join(threads[i], &h);
259                 hit |= (int) (intptr_t) h;
260         }
261
262         free(threads);
263
264         pthread_mutex_destroy(&grep_mutex);
265         pthread_mutex_destroy(&grep_read_mutex);
266         pthread_mutex_destroy(&grep_attr_mutex);
267         pthread_cond_destroy(&cond_add);
268         pthread_cond_destroy(&cond_write);
269         pthread_cond_destroy(&cond_result);
270         grep_use_locks = 0;
271
272         return hit;
273 }
274 #else /* !NO_PTHREADS */
275
276 static int wait_all(void)
277 {
278         return 0;
279 }
280 #endif
281
282 static int grep_cmd_config(const char *var, const char *value, void *cb)
283 {
284         int st = grep_config(var, value, cb);
285         if (git_color_default_config(var, value, cb) < 0)
286                 st = -1;
287
288         if (!strcmp(var, "grep.threads")) {
289                 num_threads = git_config_int(var, value);
290                 if (num_threads < 0)
291                         die(_("invalid number of threads specified (%d) for %s"),
292                             num_threads, var);
293         }
294
295         return st;
296 }
297
298 static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
299 {
300         void *data;
301
302         grep_read_lock();
303         data = read_sha1_file(oid->hash, type, size);
304         grep_read_unlock();
305         return data;
306 }
307
308 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
309                      const char *filename, int tree_name_len,
310                      const char *path)
311 {
312         struct strbuf pathbuf = STRBUF_INIT;
313
314         if (super_prefix) {
315                 strbuf_add(&pathbuf, filename, tree_name_len);
316                 strbuf_addstr(&pathbuf, super_prefix);
317                 strbuf_addstr(&pathbuf, filename + tree_name_len);
318         } else {
319                 strbuf_addstr(&pathbuf, filename);
320         }
321
322         if (opt->relative && opt->prefix_length) {
323                 char *name = strbuf_detach(&pathbuf, NULL);
324                 quote_path_relative(name + tree_name_len, opt->prefix, &pathbuf);
325                 strbuf_insert(&pathbuf, 0, name, tree_name_len);
326                 free(name);
327         }
328
329 #ifndef NO_PTHREADS
330         if (num_threads) {
331                 add_work(opt, GREP_SOURCE_SHA1, pathbuf.buf, path, oid);
332                 strbuf_release(&pathbuf);
333                 return 0;
334         } else
335 #endif
336         {
337                 struct grep_source gs;
338                 int hit;
339
340                 grep_source_init(&gs, GREP_SOURCE_SHA1, pathbuf.buf, path, oid);
341                 strbuf_release(&pathbuf);
342                 hit = grep_source(opt, &gs);
343
344                 grep_source_clear(&gs);
345                 return hit;
346         }
347 }
348
349 static int grep_file(struct grep_opt *opt, const char *filename)
350 {
351         struct strbuf buf = STRBUF_INIT;
352
353         if (super_prefix)
354                 strbuf_addstr(&buf, super_prefix);
355         strbuf_addstr(&buf, filename);
356
357         if (opt->relative && opt->prefix_length) {
358                 char *name = strbuf_detach(&buf, NULL);
359                 quote_path_relative(name, opt->prefix, &buf);
360                 free(name);
361         }
362
363 #ifndef NO_PTHREADS
364         if (num_threads) {
365                 add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
366                 strbuf_release(&buf);
367                 return 0;
368         } else
369 #endif
370         {
371                 struct grep_source gs;
372                 int hit;
373
374                 grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
375                 strbuf_release(&buf);
376                 hit = grep_source(opt, &gs);
377
378                 grep_source_clear(&gs);
379                 return hit;
380         }
381 }
382
383 static void append_path(struct grep_opt *opt, const void *data, size_t len)
384 {
385         struct string_list *path_list = opt->output_priv;
386
387         if (len == 1 && *(const char *)data == '\0')
388                 return;
389         string_list_append(path_list, xstrndup(data, len));
390 }
391
392 static void run_pager(struct grep_opt *opt, const char *prefix)
393 {
394         struct string_list *path_list = opt->output_priv;
395         struct child_process child = CHILD_PROCESS_INIT;
396         int i, status;
397
398         for (i = 0; i < path_list->nr; i++)
399                 argv_array_push(&child.args, path_list->items[i].string);
400         child.dir = prefix;
401         child.use_shell = 1;
402
403         status = run_command(&child);
404         if (status)
405                 exit(status);
406 }
407
408 static void compile_submodule_options(const struct grep_opt *opt,
409                                       const char **argv,
410                                       int cached, int untracked,
411                                       int opt_exclude, int use_index,
412                                       int pattern_type_arg)
413 {
414         struct grep_pat *pattern;
415
416         if (recurse_submodules)
417                 argv_array_push(&submodule_options, "--recurse-submodules");
418
419         if (cached)
420                 argv_array_push(&submodule_options, "--cached");
421         if (!use_index)
422                 argv_array_push(&submodule_options, "--no-index");
423         if (untracked)
424                 argv_array_push(&submodule_options, "--untracked");
425         if (opt_exclude > 0)
426                 argv_array_push(&submodule_options, "--exclude-standard");
427
428         if (opt->invert)
429                 argv_array_push(&submodule_options, "-v");
430         if (opt->ignore_case)
431                 argv_array_push(&submodule_options, "-i");
432         if (opt->word_regexp)
433                 argv_array_push(&submodule_options, "-w");
434         switch (opt->binary) {
435         case GREP_BINARY_NOMATCH:
436                 argv_array_push(&submodule_options, "-I");
437                 break;
438         case GREP_BINARY_TEXT:
439                 argv_array_push(&submodule_options, "-a");
440                 break;
441         default:
442                 break;
443         }
444         if (opt->allow_textconv)
445                 argv_array_push(&submodule_options, "--textconv");
446         if (opt->max_depth != -1)
447                 argv_array_pushf(&submodule_options, "--max-depth=%d",
448                                  opt->max_depth);
449         if (opt->linenum)
450                 argv_array_push(&submodule_options, "-n");
451         if (!opt->pathname)
452                 argv_array_push(&submodule_options, "-h");
453         if (!opt->relative)
454                 argv_array_push(&submodule_options, "--full-name");
455         if (opt->name_only)
456                 argv_array_push(&submodule_options, "-l");
457         if (opt->unmatch_name_only)
458                 argv_array_push(&submodule_options, "-L");
459         if (opt->null_following_name)
460                 argv_array_push(&submodule_options, "-z");
461         if (opt->count)
462                 argv_array_push(&submodule_options, "-c");
463         if (opt->file_break)
464                 argv_array_push(&submodule_options, "--break");
465         if (opt->heading)
466                 argv_array_push(&submodule_options, "--heading");
467         if (opt->pre_context)
468                 argv_array_pushf(&submodule_options, "--before-context=%d",
469                                  opt->pre_context);
470         if (opt->post_context)
471                 argv_array_pushf(&submodule_options, "--after-context=%d",
472                                  opt->post_context);
473         if (opt->funcname)
474                 argv_array_push(&submodule_options, "-p");
475         if (opt->funcbody)
476                 argv_array_push(&submodule_options, "-W");
477         if (opt->all_match)
478                 argv_array_push(&submodule_options, "--all-match");
479         if (opt->debug)
480                 argv_array_push(&submodule_options, "--debug");
481         if (opt->status_only)
482                 argv_array_push(&submodule_options, "-q");
483
484         switch (pattern_type_arg) {
485         case GREP_PATTERN_TYPE_BRE:
486                 argv_array_push(&submodule_options, "-G");
487                 break;
488         case GREP_PATTERN_TYPE_ERE:
489                 argv_array_push(&submodule_options, "-E");
490                 break;
491         case GREP_PATTERN_TYPE_FIXED:
492                 argv_array_push(&submodule_options, "-F");
493                 break;
494         case GREP_PATTERN_TYPE_PCRE:
495                 argv_array_push(&submodule_options, "-P");
496                 break;
497         case GREP_PATTERN_TYPE_UNSPECIFIED:
498                 break;
499         }
500
501         for (pattern = opt->pattern_list; pattern != NULL;
502              pattern = pattern->next) {
503                 switch (pattern->token) {
504                 case GREP_PATTERN:
505                         argv_array_pushf(&submodule_options, "-e%s",
506                                          pattern->pattern);
507                         break;
508                 case GREP_AND:
509                 case GREP_OPEN_PAREN:
510                 case GREP_CLOSE_PAREN:
511                 case GREP_NOT:
512                 case GREP_OR:
513                         argv_array_push(&submodule_options, pattern->pattern);
514                         break;
515                 /* BODY and HEAD are not used by git-grep */
516                 case GREP_PATTERN_BODY:
517                 case GREP_PATTERN_HEAD:
518                         break;
519                 }
520         }
521
522         /*
523          * Limit number of threads for child process to use.
524          * This is to prevent potential fork-bomb behavior of git-grep as each
525          * submodule process has its own thread pool.
526          */
527         argv_array_pushf(&submodule_options, "--threads=%d",
528                          (num_threads + 1) / 2);
529
530         /* Add Pathspecs */
531         argv_array_push(&submodule_options, "--");
532         for (; *argv; argv++)
533                 argv_array_push(&submodule_options, *argv);
534 }
535
536 /*
537  * Launch child process to grep contents of a submodule
538  */
539 static int grep_submodule_launch(struct grep_opt *opt,
540                                  const struct grep_source *gs)
541 {
542         struct child_process cp = CHILD_PROCESS_INIT;
543         int status, i;
544         const char *end_of_base;
545         const char *name;
546         struct strbuf child_output = STRBUF_INIT;
547
548         end_of_base = strchr(gs->name, ':');
549         if (gs->identifier && end_of_base)
550                 name = end_of_base + 1;
551         else
552                 name = gs->name;
553
554         prepare_submodule_repo_env(&cp.env_array);
555         argv_array_push(&cp.env_array, GIT_DIR_ENVIRONMENT);
556
557         if (opt->relative && opt->prefix_length)
558                 argv_array_pushf(&cp.env_array, "%s=%s",
559                                  GIT_TOPLEVEL_PREFIX_ENVIRONMENT,
560                                  opt->prefix);
561
562         /* Add super prefix */
563         argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
564                          super_prefix ? super_prefix : "",
565                          name);
566         argv_array_push(&cp.args, "grep");
567
568         /*
569          * Add basename of parent project
570          * When performing grep on a tree object the filename is prefixed
571          * with the object's name: 'tree-name:filename'.  In order to
572          * provide uniformity of output we want to pass the name of the
573          * parent project's object name to the submodule so the submodule can
574          * prefix its output with the parent's name and not its own SHA1.
575          */
576         if (gs->identifier && end_of_base)
577                 argv_array_pushf(&cp.args, "--parent-basename=%.*s",
578                                  (int) (end_of_base - gs->name),
579                                  gs->name);
580
581         /* Add options */
582         for (i = 0; i < submodule_options.argc; i++) {
583                 /*
584                  * If there is a tree identifier for the submodule, add the
585                  * rev after adding the submodule options but before the
586                  * pathspecs.  To do this we listen for the '--' and insert the
587                  * sha1 before pushing the '--' onto the child process argv
588                  * array.
589                  */
590                 if (gs->identifier &&
591                     !strcmp("--", submodule_options.argv[i])) {
592                         argv_array_push(&cp.args, sha1_to_hex(gs->identifier));
593                 }
594
595                 argv_array_push(&cp.args, submodule_options.argv[i]);
596         }
597
598         cp.git_cmd = 1;
599         cp.dir = gs->path;
600
601         /*
602          * Capture output to output buffer and check the return code from the
603          * child process.  A '0' indicates a hit, a '1' indicates no hit and
604          * anything else is an error.
605          */
606         status = capture_command(&cp, &child_output, 0);
607         if (status && (status != 1)) {
608                 /* flush the buffer */
609                 write_or_die(1, child_output.buf, child_output.len);
610                 die("process for submodule '%s' failed with exit code: %d",
611                     gs->name, status);
612         }
613
614         opt->output(opt, child_output.buf, child_output.len);
615         strbuf_release(&child_output);
616         /* invert the return code to make a hit equal to 1 */
617         return !status;
618 }
619
620 /*
621  * Prep grep structures for a submodule grep
622  * sha1: the sha1 of the submodule or NULL if using the working tree
623  * filename: name of the submodule including tree name of parent
624  * path: location of the submodule
625  */
626 static int grep_submodule(struct grep_opt *opt, const unsigned char *sha1,
627                           const char *filename, const char *path)
628 {
629         if (!is_submodule_initialized(path))
630                 return 0;
631         if (!is_submodule_populated_gently(path, NULL)) {
632                 /*
633                  * If searching history, check for the presense of the
634                  * submodule's gitdir before skipping the submodule.
635                  */
636                 if (sha1) {
637                         const struct submodule *sub =
638                                         submodule_from_path(null_sha1, path);
639                         if (sub)
640                                 path = git_path("modules/%s", sub->name);
641
642                         if (!(is_directory(path) && is_git_directory(path)))
643                                 return 0;
644                 } else {
645                         return 0;
646                 }
647         }
648
649 #ifndef NO_PTHREADS
650         if (num_threads) {
651                 add_work(opt, GREP_SOURCE_SUBMODULE, filename, path, sha1);
652                 return 0;
653         } else
654 #endif
655         {
656                 struct grep_source gs;
657                 int hit;
658
659                 grep_source_init(&gs, GREP_SOURCE_SUBMODULE,
660                                  filename, path, sha1);
661                 hit = grep_submodule_launch(opt, &gs);
662
663                 grep_source_clear(&gs);
664                 return hit;
665         }
666 }
667
668 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec,
669                       int cached)
670 {
671         int hit = 0;
672         int nr;
673         struct strbuf name = STRBUF_INIT;
674         int name_base_len = 0;
675         if (super_prefix) {
676                 name_base_len = strlen(super_prefix);
677                 strbuf_addstr(&name, super_prefix);
678         }
679
680         read_cache();
681
682         for (nr = 0; nr < active_nr; nr++) {
683                 const struct cache_entry *ce = active_cache[nr];
684                 strbuf_setlen(&name, name_base_len);
685                 strbuf_addstr(&name, ce->name);
686
687                 if (S_ISREG(ce->ce_mode) &&
688                     match_pathspec(pathspec, name.buf, name.len, 0, NULL,
689                                    S_ISDIR(ce->ce_mode) ||
690                                    S_ISGITLINK(ce->ce_mode))) {
691                         /*
692                          * If CE_VALID is on, we assume worktree file and its
693                          * cache entry are identical, even if worktree file has
694                          * been modified, so use cache version instead
695                          */
696                         if (cached || (ce->ce_flags & CE_VALID) ||
697                             ce_skip_worktree(ce)) {
698                                 if (ce_stage(ce) || ce_intent_to_add(ce))
699                                         continue;
700                                 hit |= grep_oid(opt, &ce->oid, ce->name,
701                                                  0, ce->name);
702                         } else {
703                                 hit |= grep_file(opt, ce->name);
704                         }
705                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
706                            submodule_path_match(pathspec, name.buf, NULL)) {
707                         hit |= grep_submodule(opt, NULL, ce->name, ce->name);
708                 } else {
709                         continue;
710                 }
711
712                 if (ce_stage(ce)) {
713                         do {
714                                 nr++;
715                         } while (nr < active_nr &&
716                                  !strcmp(ce->name, active_cache[nr]->name));
717                         nr--; /* compensate for loop control */
718                 }
719                 if (hit && opt->status_only)
720                         break;
721         }
722
723         strbuf_release(&name);
724         return hit;
725 }
726
727 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
728                      struct tree_desc *tree, struct strbuf *base, int tn_len,
729                      int check_attr)
730 {
731         int hit = 0;
732         enum interesting match = entry_not_interesting;
733         struct name_entry entry;
734         int old_baselen = base->len;
735         struct strbuf name = STRBUF_INIT;
736         int name_base_len = 0;
737         if (super_prefix) {
738                 strbuf_addstr(&name, super_prefix);
739                 name_base_len = name.len;
740         }
741
742         while (tree_entry(tree, &entry)) {
743                 int te_len = tree_entry_len(&entry);
744
745                 if (match != all_entries_interesting) {
746                         strbuf_addstr(&name, base->buf + tn_len);
747                         match = tree_entry_interesting(&entry, &name,
748                                                        0, pathspec);
749                         strbuf_setlen(&name, name_base_len);
750
751                         if (match == all_entries_not_interesting)
752                                 break;
753                         if (match == entry_not_interesting)
754                                 continue;
755                 }
756
757                 strbuf_add(base, entry.path, te_len);
758
759                 if (S_ISREG(entry.mode)) {
760                         hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
761                                          check_attr ? base->buf + tn_len : NULL);
762                 } else if (S_ISDIR(entry.mode)) {
763                         enum object_type type;
764                         struct tree_desc sub;
765                         void *data;
766                         unsigned long size;
767
768                         data = lock_and_read_oid_file(entry.oid, &type, &size);
769                         if (!data)
770                                 die(_("unable to read tree (%s)"),
771                                     oid_to_hex(entry.oid));
772
773                         strbuf_addch(base, '/');
774                         init_tree_desc(&sub, data, size);
775                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
776                                          check_attr);
777                         free(data);
778                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
779                         hit |= grep_submodule(opt, entry.oid->hash, base->buf,
780                                               base->buf + tn_len);
781                 }
782
783                 strbuf_setlen(base, old_baselen);
784
785                 if (hit && opt->status_only)
786                         break;
787         }
788
789         strbuf_release(&name);
790         return hit;
791 }
792
793 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
794                        struct object *obj, const char *name, const char *path)
795 {
796         if (obj->type == OBJ_BLOB)
797                 return grep_oid(opt, &obj->oid, name, 0, path);
798         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
799                 struct tree_desc tree;
800                 void *data;
801                 unsigned long size;
802                 struct strbuf base;
803                 int hit, len;
804
805                 grep_read_lock();
806                 data = read_object_with_reference(obj->oid.hash, tree_type,
807                                                   &size, NULL);
808                 grep_read_unlock();
809
810                 if (!data)
811                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
812
813                 /* Use parent's name as base when recursing submodules */
814                 if (recurse_submodules && parent_basename)
815                         name = parent_basename;
816
817                 len = name ? strlen(name) : 0;
818                 strbuf_init(&base, PATH_MAX + len + 1);
819                 if (len) {
820                         strbuf_add(&base, name, len);
821                         strbuf_addch(&base, ':');
822                 }
823                 init_tree_desc(&tree, data, size);
824                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
825                                 obj->type == OBJ_COMMIT);
826                 strbuf_release(&base);
827                 free(data);
828                 return hit;
829         }
830         die(_("unable to grep from object of type %s"), typename(obj->type));
831 }
832
833 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
834                         const struct object_array *list)
835 {
836         unsigned int i;
837         int hit = 0;
838         const unsigned int nr = list->nr;
839
840         for (i = 0; i < nr; i++) {
841                 struct object *real_obj;
842                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
843
844                 /* load the gitmodules file for this rev */
845                 if (recurse_submodules) {
846                         submodule_free();
847                         gitmodules_config_sha1(real_obj->oid.hash);
848                 }
849                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path)) {
850                         hit = 1;
851                         if (opt->status_only)
852                                 break;
853                 }
854         }
855         return hit;
856 }
857
858 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
859                           int exc_std, int use_index)
860 {
861         struct dir_struct dir;
862         int i, hit = 0;
863
864         memset(&dir, 0, sizeof(dir));
865         if (!use_index)
866                 dir.flags |= DIR_NO_GITLINKS;
867         if (exc_std)
868                 setup_standard_excludes(&dir);
869
870         fill_directory(&dir, pathspec);
871         for (i = 0; i < dir.nr; i++) {
872                 if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
873                         continue;
874                 hit |= grep_file(opt, dir.entries[i]->name);
875                 if (hit && opt->status_only)
876                         break;
877         }
878         return hit;
879 }
880
881 static int context_callback(const struct option *opt, const char *arg,
882                             int unset)
883 {
884         struct grep_opt *grep_opt = opt->value;
885         int value;
886         const char *endp;
887
888         if (unset) {
889                 grep_opt->pre_context = grep_opt->post_context = 0;
890                 return 0;
891         }
892         value = strtol(arg, (char **)&endp, 10);
893         if (*endp) {
894                 return error(_("switch `%c' expects a numerical value"),
895                              opt->short_name);
896         }
897         grep_opt->pre_context = grep_opt->post_context = value;
898         return 0;
899 }
900
901 static int file_callback(const struct option *opt, const char *arg, int unset)
902 {
903         struct grep_opt *grep_opt = opt->value;
904         int from_stdin = !strcmp(arg, "-");
905         FILE *patterns;
906         int lno = 0;
907         struct strbuf sb = STRBUF_INIT;
908
909         patterns = from_stdin ? stdin : fopen(arg, "r");
910         if (!patterns)
911                 die_errno(_("cannot open '%s'"), arg);
912         while (strbuf_getline(&sb, patterns) == 0) {
913                 /* ignore empty line like grep does */
914                 if (sb.len == 0)
915                         continue;
916
917                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
918                                 GREP_PATTERN);
919         }
920         if (!from_stdin)
921                 fclose(patterns);
922         strbuf_release(&sb);
923         return 0;
924 }
925
926 static int not_callback(const struct option *opt, const char *arg, int unset)
927 {
928         struct grep_opt *grep_opt = opt->value;
929         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
930         return 0;
931 }
932
933 static int and_callback(const struct option *opt, const char *arg, int unset)
934 {
935         struct grep_opt *grep_opt = opt->value;
936         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
937         return 0;
938 }
939
940 static int open_callback(const struct option *opt, const char *arg, int unset)
941 {
942         struct grep_opt *grep_opt = opt->value;
943         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
944         return 0;
945 }
946
947 static int close_callback(const struct option *opt, const char *arg, int unset)
948 {
949         struct grep_opt *grep_opt = opt->value;
950         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
951         return 0;
952 }
953
954 static int pattern_callback(const struct option *opt, const char *arg,
955                             int unset)
956 {
957         struct grep_opt *grep_opt = opt->value;
958         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
959         return 0;
960 }
961
962 int cmd_grep(int argc, const char **argv, const char *prefix)
963 {
964         int hit = 0;
965         int cached = 0, untracked = 0, opt_exclude = -1;
966         int seen_dashdash = 0;
967         int external_grep_allowed__ignored;
968         const char *show_in_pager = NULL, *default_pager = "dummy";
969         struct grep_opt opt;
970         struct object_array list = OBJECT_ARRAY_INIT;
971         struct pathspec pathspec;
972         struct string_list path_list = STRING_LIST_INIT_NODUP;
973         int i;
974         int dummy;
975         int use_index = 1;
976         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
977         int allow_revs;
978
979         struct option options[] = {
980                 OPT_BOOL(0, "cached", &cached,
981                         N_("search in index instead of in the work tree")),
982                 OPT_NEGBIT(0, "no-index", &use_index,
983                          N_("find in contents not managed by git"), 1),
984                 OPT_BOOL(0, "untracked", &untracked,
985                         N_("search in both tracked and untracked files")),
986                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
987                             N_("ignore files specified via '.gitignore'"), 1),
988                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
989                          N_("recursively search in each submodule")),
990                 OPT_STRING(0, "parent-basename", &parent_basename,
991                            N_("basename"),
992                            N_("prepend parent project's basename to output")),
993                 OPT_GROUP(""),
994                 OPT_BOOL('v', "invert-match", &opt.invert,
995                         N_("show non-matching lines")),
996                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
997                         N_("case insensitive matching")),
998                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
999                         N_("match patterns only at word boundaries")),
1000                 OPT_SET_INT('a', "text", &opt.binary,
1001                         N_("process binary files as text"), GREP_BINARY_TEXT),
1002                 OPT_SET_INT('I', NULL, &opt.binary,
1003                         N_("don't match patterns in binary files"),
1004                         GREP_BINARY_NOMATCH),
1005                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
1006                          N_("process binary files with textconv filters")),
1007                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
1008                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
1009                         NULL, 1 },
1010                 OPT_GROUP(""),
1011                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
1012                             N_("use extended POSIX regular expressions"),
1013                             GREP_PATTERN_TYPE_ERE),
1014                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
1015                             N_("use basic POSIX regular expressions (default)"),
1016                             GREP_PATTERN_TYPE_BRE),
1017                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
1018                             N_("interpret patterns as fixed strings"),
1019                             GREP_PATTERN_TYPE_FIXED),
1020                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
1021                             N_("use Perl-compatible regular expressions"),
1022                             GREP_PATTERN_TYPE_PCRE),
1023                 OPT_GROUP(""),
1024                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
1025                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
1026                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
1027                 OPT_NEGBIT(0, "full-name", &opt.relative,
1028                         N_("show filenames relative to top directory"), 1),
1029                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
1030                         N_("show only filenames instead of matching lines")),
1031                 OPT_BOOL(0, "name-only", &opt.name_only,
1032                         N_("synonym for --files-with-matches")),
1033                 OPT_BOOL('L', "files-without-match",
1034                         &opt.unmatch_name_only,
1035                         N_("show only the names of files without match")),
1036                 OPT_BOOL('z', "null", &opt.null_following_name,
1037                         N_("print NUL after filenames")),
1038                 OPT_BOOL('c', "count", &opt.count,
1039                         N_("show the number of matches instead of matching lines")),
1040                 OPT__COLOR(&opt.color, N_("highlight matches")),
1041                 OPT_BOOL(0, "break", &opt.file_break,
1042                         N_("print empty line between matches from different files")),
1043                 OPT_BOOL(0, "heading", &opt.heading,
1044                         N_("show filename only once above matches from same file")),
1045                 OPT_GROUP(""),
1046                 OPT_CALLBACK('C', "context", &opt, N_("n"),
1047                         N_("show <n> context lines before and after matches"),
1048                         context_callback),
1049                 OPT_INTEGER('B', "before-context", &opt.pre_context,
1050                         N_("show <n> context lines before matches")),
1051                 OPT_INTEGER('A', "after-context", &opt.post_context,
1052                         N_("show <n> context lines after matches")),
1053                 OPT_INTEGER(0, "threads", &num_threads,
1054                         N_("use <n> worker threads")),
1055                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
1056                         context_callback),
1057                 OPT_BOOL('p', "show-function", &opt.funcname,
1058                         N_("show a line with the function name before matches")),
1059                 OPT_BOOL('W', "function-context", &opt.funcbody,
1060                         N_("show the surrounding function")),
1061                 OPT_GROUP(""),
1062                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
1063                         N_("read patterns from file"), file_callback),
1064                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
1065                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
1066                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
1067                   N_("combine patterns specified with -e"),
1068                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
1069                 OPT_BOOL(0, "or", &dummy, ""),
1070                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
1071                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
1072                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
1073                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1074                   open_callback },
1075                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
1076                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1077                   close_callback },
1078                 OPT__QUIET(&opt.status_only,
1079                            N_("indicate hit with exit status without output")),
1080                 OPT_BOOL(0, "all-match", &opt.all_match,
1081                         N_("show only matches from files that match all patterns")),
1082                 { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
1083                   N_("show parse tree for grep expression"),
1084                   PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
1085                 OPT_GROUP(""),
1086                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
1087                         N_("pager"), N_("show matching files in the pager"),
1088                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
1089                 OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
1090                          N_("allow calling of grep(1) (ignored by this build)")),
1091                 OPT_END()
1092         };
1093
1094         init_grep_defaults();
1095         git_config(grep_cmd_config, NULL);
1096         grep_init(&opt, prefix);
1097         super_prefix = get_super_prefix();
1098
1099         /*
1100          * If there is no -- then the paths must exist in the working
1101          * tree.  If there is no explicit pattern specified with -e or
1102          * -f, we take the first unrecognized non option to be the
1103          * pattern, but then what follows it must be zero or more
1104          * valid refs up to the -- (if exists), and then existing
1105          * paths.  If there is an explicit pattern, then the first
1106          * unrecognized non option is the beginning of the refs list
1107          * that continues up to the -- (if exists), and then paths.
1108          */
1109         argc = parse_options(argc, argv, prefix, options, grep_usage,
1110                              PARSE_OPT_KEEP_DASHDASH |
1111                              PARSE_OPT_STOP_AT_NON_OPTION);
1112         grep_commit_pattern_type(pattern_type_arg, &opt);
1113
1114         if (use_index && !startup_info->have_repository) {
1115                 int fallback = 0;
1116                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
1117                 if (fallback)
1118                         use_index = 0;
1119                 else
1120                         /* die the same way as if we did it at the beginning */
1121                         setup_git_directory();
1122         }
1123
1124         /*
1125          * skip a -- separator; we know it cannot be
1126          * separating revisions from pathnames if
1127          * we haven't even had any patterns yet
1128          */
1129         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
1130                 argv++;
1131                 argc--;
1132         }
1133
1134         /* First unrecognized non-option token */
1135         if (argc > 0 && !opt.pattern_list) {
1136                 append_grep_pattern(&opt, argv[0], "command line", 0,
1137                                     GREP_PATTERN);
1138                 argv++;
1139                 argc--;
1140         }
1141
1142         if (show_in_pager == default_pager)
1143                 show_in_pager = git_pager(1);
1144         if (show_in_pager) {
1145                 opt.color = 0;
1146                 opt.name_only = 1;
1147                 opt.null_following_name = 1;
1148                 opt.output_priv = &path_list;
1149                 opt.output = append_path;
1150                 string_list_append(&path_list, show_in_pager);
1151         }
1152
1153         if (!opt.pattern_list)
1154                 die(_("no pattern given."));
1155         if (!opt.fixed && opt.ignore_case)
1156                 opt.regflags |= REG_ICASE;
1157
1158         compile_grep_patterns(&opt);
1159
1160         /*
1161          * We have to find "--" in a separate pass, because its presence
1162          * influences how we will parse arguments that come before it.
1163          */
1164         for (i = 0; i < argc; i++) {
1165                 if (!strcmp(argv[i], "--")) {
1166                         seen_dashdash = 1;
1167                         break;
1168                 }
1169         }
1170
1171         /*
1172          * Resolve any rev arguments. If we have a dashdash, then everything up
1173          * to it must resolve as a rev. If not, then we stop at the first
1174          * non-rev and assume everything else is a path.
1175          */
1176         allow_revs = use_index && !untracked;
1177         for (i = 0; i < argc; i++) {
1178                 const char *arg = argv[i];
1179                 struct object_id oid;
1180                 struct object_context oc;
1181                 struct object *object;
1182
1183                 if (!strcmp(arg, "--")) {
1184                         i++;
1185                         break;
1186                 }
1187
1188                 if (!allow_revs) {
1189                         if (seen_dashdash)
1190                                 die(_("--no-index or --untracked cannot be used with revs"));
1191                         break;
1192                 }
1193
1194                 if (get_sha1_with_context(arg, 0, oid.hash, &oc)) {
1195                         if (seen_dashdash)
1196                                 die(_("unable to resolve revision: %s"), arg);
1197                         break;
1198                 }
1199
1200                 object = parse_object_or_die(oid.hash, arg);
1201                 if (!seen_dashdash)
1202                         verify_non_filename(prefix, arg);
1203                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1204         }
1205
1206         /*
1207          * Anything left over is presumed to be a path. But in the non-dashdash
1208          * "do what I mean" case, we verify and complain when that isn't true.
1209          */
1210         if (!seen_dashdash) {
1211                 int j;
1212                 for (j = i; j < argc; j++)
1213                         verify_filename(prefix, argv[j], j == i && allow_revs);
1214         }
1215
1216         parse_pathspec(&pathspec, 0,
1217                        PATHSPEC_PREFER_CWD |
1218                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1219                        prefix, argv + i);
1220         pathspec.max_depth = opt.max_depth;
1221         pathspec.recursive = 1;
1222
1223 #ifndef NO_PTHREADS
1224         if (list.nr || cached || show_in_pager)
1225                 num_threads = 0;
1226         else if (num_threads == 0)
1227                 num_threads = GREP_NUM_THREADS_DEFAULT;
1228         else if (num_threads < 0)
1229                 die(_("invalid number of threads specified (%d)"), num_threads);
1230 #else
1231         num_threads = 0;
1232 #endif
1233
1234 #ifndef NO_PTHREADS
1235         if (num_threads) {
1236                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1237                     && (opt.pre_context || opt.post_context ||
1238                         opt.file_break || opt.funcbody))
1239                         skip_first_line = 1;
1240                 start_threads(&opt);
1241         }
1242 #endif
1243
1244         if (recurse_submodules) {
1245                 gitmodules_config();
1246                 compile_submodule_options(&opt, argv + i, cached, untracked,
1247                                           opt_exclude, use_index,
1248                                           pattern_type_arg);
1249         }
1250
1251         if (show_in_pager && (cached || list.nr))
1252                 die(_("--open-files-in-pager only works on the worktree"));
1253
1254         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1255                 const char *pager = path_list.items[0].string;
1256                 int len = strlen(pager);
1257
1258                 if (len > 4 && is_dir_sep(pager[len - 5]))
1259                         pager += len - 4;
1260
1261                 if (opt.ignore_case && !strcmp("less", pager))
1262                         string_list_append(&path_list, "-I");
1263
1264                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1265                         struct strbuf buf = STRBUF_INIT;
1266                         strbuf_addf(&buf, "+/%s%s",
1267                                         strcmp("less", pager) ? "" : "*",
1268                                         opt.pattern_list->pattern);
1269                         string_list_append(&path_list, buf.buf);
1270                         strbuf_detach(&buf, NULL);
1271                 }
1272         }
1273
1274         if (recurse_submodules && (!use_index || untracked))
1275                 die(_("option not supported with --recurse-submodules."));
1276
1277         if (!show_in_pager && !opt.status_only)
1278                 setup_pager();
1279
1280         if (!use_index && (untracked || cached))
1281                 die(_("--cached or --untracked cannot be used with --no-index."));
1282
1283         if (!use_index || untracked) {
1284                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1285                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1286         } else if (0 <= opt_exclude) {
1287                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1288         } else if (!list.nr) {
1289                 if (!cached)
1290                         setup_work_tree();
1291
1292                 hit = grep_cache(&opt, &pathspec, cached);
1293         } else {
1294                 if (cached)
1295                         die(_("both --cached and trees are given."));
1296                 hit = grep_objects(&opt, &pathspec, &list);
1297         }
1298
1299         if (num_threads)
1300                 hit |= wait_all();
1301         if (hit && show_in_pager)
1302                 run_pager(&opt, prefix);
1303         clear_pathspec(&pathspec);
1304         free_grep_patterns(&opt);
1305         return !hit;
1306 }