Merge branch 'bw/config-h'
[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         assert(num_threads);
78         pthread_mutex_lock(&grep_mutex);
79 }
80
81 static inline void grep_unlock(void)
82 {
83         assert(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                 if (i)
229                         o->debug = 0;
230                 compile_grep_patterns(o);
231                 err = pthread_create(&threads[i], NULL, run, o);
232
233                 if (err)
234                         die(_("grep: failed to create thread: %s"),
235                             strerror(err));
236         }
237 }
238
239 static int wait_all(void)
240 {
241         int hit = 0;
242         int i;
243
244         grep_lock();
245         all_work_added = 1;
246
247         /* Wait until all work is done. */
248         while (todo_done != todo_end)
249                 pthread_cond_wait(&cond_result, &grep_mutex);
250
251         /* Wake up all the consumer threads so they can see that there
252          * is no more work to do.
253          */
254         pthread_cond_broadcast(&cond_add);
255         grep_unlock();
256
257         for (i = 0; i < num_threads; i++) {
258                 void *h;
259                 pthread_join(threads[i], &h);
260                 hit |= (int) (intptr_t) h;
261         }
262
263         free(threads);
264
265         pthread_mutex_destroy(&grep_mutex);
266         pthread_mutex_destroy(&grep_read_mutex);
267         pthread_mutex_destroy(&grep_attr_mutex);
268         pthread_cond_destroy(&cond_add);
269         pthread_cond_destroy(&cond_write);
270         pthread_cond_destroy(&cond_result);
271         grep_use_locks = 0;
272
273         return hit;
274 }
275 #else /* !NO_PTHREADS */
276
277 static int wait_all(void)
278 {
279         return 0;
280 }
281 #endif
282
283 static int grep_cmd_config(const char *var, const char *value, void *cb)
284 {
285         int st = grep_config(var, value, cb);
286         if (git_color_default_config(var, value, cb) < 0)
287                 st = -1;
288
289         if (!strcmp(var, "grep.threads")) {
290                 num_threads = git_config_int(var, value);
291                 if (num_threads < 0)
292                         die(_("invalid number of threads specified (%d) for %s"),
293                             num_threads, var);
294 #ifdef NO_PTHREADS
295                 else if (num_threads && num_threads != 1) {
296                         /*
297                          * TRANSLATORS: %s is the configuration
298                          * variable for tweaking threads, currently
299                          * grep.threads
300                          */
301                         warning(_("no threads support, ignoring %s"), var);
302                         num_threads = 0;
303                 }
304 #endif
305         }
306
307         if (!strcmp(var, "submodule.recurse"))
308                 recurse_submodules = git_config_bool(var, value);
309
310         return st;
311 }
312
313 static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
314 {
315         void *data;
316
317         grep_read_lock();
318         data = read_sha1_file(oid->hash, type, size);
319         grep_read_unlock();
320         return data;
321 }
322
323 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
324                      const char *filename, int tree_name_len,
325                      const char *path)
326 {
327         struct strbuf pathbuf = STRBUF_INIT;
328
329         if (super_prefix) {
330                 strbuf_add(&pathbuf, filename, tree_name_len);
331                 strbuf_addstr(&pathbuf, super_prefix);
332                 strbuf_addstr(&pathbuf, filename + tree_name_len);
333         } else {
334                 strbuf_addstr(&pathbuf, filename);
335         }
336
337         if (opt->relative && opt->prefix_length) {
338                 char *name = strbuf_detach(&pathbuf, NULL);
339                 quote_path_relative(name + tree_name_len, opt->prefix, &pathbuf);
340                 strbuf_insert(&pathbuf, 0, name, tree_name_len);
341                 free(name);
342         }
343
344 #ifndef NO_PTHREADS
345         if (num_threads) {
346                 add_work(opt, GREP_SOURCE_OID, pathbuf.buf, path, oid);
347                 strbuf_release(&pathbuf);
348                 return 0;
349         } else
350 #endif
351         {
352                 struct grep_source gs;
353                 int hit;
354
355                 grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
356                 strbuf_release(&pathbuf);
357                 hit = grep_source(opt, &gs);
358
359                 grep_source_clear(&gs);
360                 return hit;
361         }
362 }
363
364 static int grep_file(struct grep_opt *opt, const char *filename)
365 {
366         struct strbuf buf = STRBUF_INIT;
367
368         if (super_prefix)
369                 strbuf_addstr(&buf, super_prefix);
370         strbuf_addstr(&buf, filename);
371
372         if (opt->relative && opt->prefix_length) {
373                 char *name = strbuf_detach(&buf, NULL);
374                 quote_path_relative(name, opt->prefix, &buf);
375                 free(name);
376         }
377
378 #ifndef NO_PTHREADS
379         if (num_threads) {
380                 add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
381                 strbuf_release(&buf);
382                 return 0;
383         } else
384 #endif
385         {
386                 struct grep_source gs;
387                 int hit;
388
389                 grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
390                 strbuf_release(&buf);
391                 hit = grep_source(opt, &gs);
392
393                 grep_source_clear(&gs);
394                 return hit;
395         }
396 }
397
398 static void append_path(struct grep_opt *opt, const void *data, size_t len)
399 {
400         struct string_list *path_list = opt->output_priv;
401
402         if (len == 1 && *(const char *)data == '\0')
403                 return;
404         string_list_append(path_list, xstrndup(data, len));
405 }
406
407 static void run_pager(struct grep_opt *opt, const char *prefix)
408 {
409         struct string_list *path_list = opt->output_priv;
410         struct child_process child = CHILD_PROCESS_INIT;
411         int i, status;
412
413         for (i = 0; i < path_list->nr; i++)
414                 argv_array_push(&child.args, path_list->items[i].string);
415         child.dir = prefix;
416         child.use_shell = 1;
417
418         status = run_command(&child);
419         if (status)
420                 exit(status);
421 }
422
423 static void compile_submodule_options(const struct grep_opt *opt,
424                                       const char **argv,
425                                       int cached, int untracked,
426                                       int opt_exclude, int use_index,
427                                       int pattern_type_arg)
428 {
429         struct grep_pat *pattern;
430
431         if (recurse_submodules)
432                 argv_array_push(&submodule_options, "--recurse-submodules");
433
434         if (cached)
435                 argv_array_push(&submodule_options, "--cached");
436         if (!use_index)
437                 argv_array_push(&submodule_options, "--no-index");
438         if (untracked)
439                 argv_array_push(&submodule_options, "--untracked");
440         if (opt_exclude > 0)
441                 argv_array_push(&submodule_options, "--exclude-standard");
442
443         if (opt->invert)
444                 argv_array_push(&submodule_options, "-v");
445         if (opt->ignore_case)
446                 argv_array_push(&submodule_options, "-i");
447         if (opt->word_regexp)
448                 argv_array_push(&submodule_options, "-w");
449         switch (opt->binary) {
450         case GREP_BINARY_NOMATCH:
451                 argv_array_push(&submodule_options, "-I");
452                 break;
453         case GREP_BINARY_TEXT:
454                 argv_array_push(&submodule_options, "-a");
455                 break;
456         default:
457                 break;
458         }
459         if (opt->allow_textconv)
460                 argv_array_push(&submodule_options, "--textconv");
461         if (opt->max_depth != -1)
462                 argv_array_pushf(&submodule_options, "--max-depth=%d",
463                                  opt->max_depth);
464         if (opt->linenum)
465                 argv_array_push(&submodule_options, "-n");
466         if (!opt->pathname)
467                 argv_array_push(&submodule_options, "-h");
468         if (!opt->relative)
469                 argv_array_push(&submodule_options, "--full-name");
470         if (opt->name_only)
471                 argv_array_push(&submodule_options, "-l");
472         if (opt->unmatch_name_only)
473                 argv_array_push(&submodule_options, "-L");
474         if (opt->null_following_name)
475                 argv_array_push(&submodule_options, "-z");
476         if (opt->count)
477                 argv_array_push(&submodule_options, "-c");
478         if (opt->file_break)
479                 argv_array_push(&submodule_options, "--break");
480         if (opt->heading)
481                 argv_array_push(&submodule_options, "--heading");
482         if (opt->pre_context)
483                 argv_array_pushf(&submodule_options, "--before-context=%d",
484                                  opt->pre_context);
485         if (opt->post_context)
486                 argv_array_pushf(&submodule_options, "--after-context=%d",
487                                  opt->post_context);
488         if (opt->funcname)
489                 argv_array_push(&submodule_options, "-p");
490         if (opt->funcbody)
491                 argv_array_push(&submodule_options, "-W");
492         if (opt->all_match)
493                 argv_array_push(&submodule_options, "--all-match");
494         if (opt->debug)
495                 argv_array_push(&submodule_options, "--debug");
496         if (opt->status_only)
497                 argv_array_push(&submodule_options, "-q");
498
499         switch (pattern_type_arg) {
500         case GREP_PATTERN_TYPE_BRE:
501                 argv_array_push(&submodule_options, "-G");
502                 break;
503         case GREP_PATTERN_TYPE_ERE:
504                 argv_array_push(&submodule_options, "-E");
505                 break;
506         case GREP_PATTERN_TYPE_FIXED:
507                 argv_array_push(&submodule_options, "-F");
508                 break;
509         case GREP_PATTERN_TYPE_PCRE:
510                 argv_array_push(&submodule_options, "-P");
511                 break;
512         case GREP_PATTERN_TYPE_UNSPECIFIED:
513                 break;
514         default:
515                 die("BUG: Added a new grep pattern type without updating switch statement");
516         }
517
518         for (pattern = opt->pattern_list; pattern != NULL;
519              pattern = pattern->next) {
520                 switch (pattern->token) {
521                 case GREP_PATTERN:
522                         argv_array_pushf(&submodule_options, "-e%s",
523                                          pattern->pattern);
524                         break;
525                 case GREP_AND:
526                 case GREP_OPEN_PAREN:
527                 case GREP_CLOSE_PAREN:
528                 case GREP_NOT:
529                 case GREP_OR:
530                         argv_array_push(&submodule_options, pattern->pattern);
531                         break;
532                 /* BODY and HEAD are not used by git-grep */
533                 case GREP_PATTERN_BODY:
534                 case GREP_PATTERN_HEAD:
535                         break;
536                 }
537         }
538
539         /*
540          * Limit number of threads for child process to use.
541          * This is to prevent potential fork-bomb behavior of git-grep as each
542          * submodule process has its own thread pool.
543          */
544         argv_array_pushf(&submodule_options, "--threads=%d",
545                          (num_threads + 1) / 2);
546
547         /* Add Pathspecs */
548         argv_array_push(&submodule_options, "--");
549         for (; *argv; argv++)
550                 argv_array_push(&submodule_options, *argv);
551 }
552
553 /*
554  * Launch child process to grep contents of a submodule
555  */
556 static int grep_submodule_launch(struct grep_opt *opt,
557                                  const struct grep_source *gs)
558 {
559         struct child_process cp = CHILD_PROCESS_INIT;
560         int status, i;
561         const char *end_of_base;
562         const char *name;
563         struct strbuf child_output = STRBUF_INIT;
564
565         end_of_base = strchr(gs->name, ':');
566         if (gs->identifier && end_of_base)
567                 name = end_of_base + 1;
568         else
569                 name = gs->name;
570
571         prepare_submodule_repo_env(&cp.env_array);
572         argv_array_push(&cp.env_array, GIT_DIR_ENVIRONMENT);
573
574         if (opt->relative && opt->prefix_length)
575                 argv_array_pushf(&cp.env_array, "%s=%s",
576                                  GIT_TOPLEVEL_PREFIX_ENVIRONMENT,
577                                  opt->prefix);
578
579         /* Add super prefix */
580         argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
581                          super_prefix ? super_prefix : "",
582                          name);
583         argv_array_push(&cp.args, "grep");
584
585         /*
586          * Add basename of parent project
587          * When performing grep on a tree object the filename is prefixed
588          * with the object's name: 'tree-name:filename'.  In order to
589          * provide uniformity of output we want to pass the name of the
590          * parent project's object name to the submodule so the submodule can
591          * prefix its output with the parent's name and not its own OID.
592          */
593         if (gs->identifier && end_of_base)
594                 argv_array_pushf(&cp.args, "--parent-basename=%.*s",
595                                  (int) (end_of_base - gs->name),
596                                  gs->name);
597
598         /* Add options */
599         for (i = 0; i < submodule_options.argc; i++) {
600                 /*
601                  * If there is a tree identifier for the submodule, add the
602                  * rev after adding the submodule options but before the
603                  * pathspecs.  To do this we listen for the '--' and insert the
604                  * oid before pushing the '--' onto the child process argv
605                  * array.
606                  */
607                 if (gs->identifier &&
608                     !strcmp("--", submodule_options.argv[i])) {
609                         argv_array_push(&cp.args, oid_to_hex(gs->identifier));
610                 }
611
612                 argv_array_push(&cp.args, submodule_options.argv[i]);
613         }
614
615         cp.git_cmd = 1;
616         cp.dir = gs->path;
617
618         /*
619          * Capture output to output buffer and check the return code from the
620          * child process.  A '0' indicates a hit, a '1' indicates no hit and
621          * anything else is an error.
622          */
623         status = capture_command(&cp, &child_output, 0);
624         if (status && (status != 1)) {
625                 /* flush the buffer */
626                 write_or_die(1, child_output.buf, child_output.len);
627                 die("process for submodule '%s' failed with exit code: %d",
628                     gs->name, status);
629         }
630
631         opt->output(opt, child_output.buf, child_output.len);
632         strbuf_release(&child_output);
633         /* invert the return code to make a hit equal to 1 */
634         return !status;
635 }
636
637 /*
638  * Prep grep structures for a submodule grep
639  * oid: the oid of the submodule or NULL if using the working tree
640  * filename: name of the submodule including tree name of parent
641  * path: location of the submodule
642  */
643 static int grep_submodule(struct grep_opt *opt, const struct object_id *oid,
644                           const char *filename, const char *path)
645 {
646         if (!is_submodule_initialized(path))
647                 return 0;
648         if (!is_submodule_populated_gently(path, NULL)) {
649                 /*
650                  * If searching history, check for the presense of the
651                  * submodule's gitdir before skipping the submodule.
652                  */
653                 if (oid) {
654                         const struct submodule *sub =
655                                         submodule_from_path(null_sha1, path);
656                         if (sub)
657                                 path = git_path("modules/%s", sub->name);
658
659                         if (!(is_directory(path) && is_git_directory(path)))
660                                 return 0;
661                 } else {
662                         return 0;
663                 }
664         }
665
666 #ifndef NO_PTHREADS
667         if (num_threads) {
668                 add_work(opt, GREP_SOURCE_SUBMODULE, filename, path, oid);
669                 return 0;
670         } else
671 #endif
672         {
673                 struct grep_source gs;
674                 int hit;
675
676                 grep_source_init(&gs, GREP_SOURCE_SUBMODULE,
677                                  filename, path, oid);
678                 hit = grep_submodule_launch(opt, &gs);
679
680                 grep_source_clear(&gs);
681                 return hit;
682         }
683 }
684
685 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec,
686                       int cached)
687 {
688         int hit = 0;
689         int nr;
690         struct strbuf name = STRBUF_INIT;
691         int name_base_len = 0;
692         if (super_prefix) {
693                 name_base_len = strlen(super_prefix);
694                 strbuf_addstr(&name, super_prefix);
695         }
696
697         read_cache();
698
699         for (nr = 0; nr < active_nr; nr++) {
700                 const struct cache_entry *ce = active_cache[nr];
701                 strbuf_setlen(&name, name_base_len);
702                 strbuf_addstr(&name, ce->name);
703
704                 if (S_ISREG(ce->ce_mode) &&
705                     match_pathspec(pathspec, name.buf, name.len, 0, NULL,
706                                    S_ISDIR(ce->ce_mode) ||
707                                    S_ISGITLINK(ce->ce_mode))) {
708                         /*
709                          * If CE_VALID is on, we assume worktree file and its
710                          * cache entry are identical, even if worktree file has
711                          * been modified, so use cache version instead
712                          */
713                         if (cached || (ce->ce_flags & CE_VALID) ||
714                             ce_skip_worktree(ce)) {
715                                 if (ce_stage(ce) || ce_intent_to_add(ce))
716                                         continue;
717                                 hit |= grep_oid(opt, &ce->oid, ce->name,
718                                                  0, ce->name);
719                         } else {
720                                 hit |= grep_file(opt, ce->name);
721                         }
722                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
723                            submodule_path_match(pathspec, name.buf, NULL)) {
724                         hit |= grep_submodule(opt, NULL, ce->name, ce->name);
725                 } else {
726                         continue;
727                 }
728
729                 if (ce_stage(ce)) {
730                         do {
731                                 nr++;
732                         } while (nr < active_nr &&
733                                  !strcmp(ce->name, active_cache[nr]->name));
734                         nr--; /* compensate for loop control */
735                 }
736                 if (hit && opt->status_only)
737                         break;
738         }
739
740         strbuf_release(&name);
741         return hit;
742 }
743
744 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
745                      struct tree_desc *tree, struct strbuf *base, int tn_len,
746                      int check_attr)
747 {
748         int hit = 0;
749         enum interesting match = entry_not_interesting;
750         struct name_entry entry;
751         int old_baselen = base->len;
752         struct strbuf name = STRBUF_INIT;
753         int name_base_len = 0;
754         if (super_prefix) {
755                 strbuf_addstr(&name, super_prefix);
756                 name_base_len = name.len;
757         }
758
759         while (tree_entry(tree, &entry)) {
760                 int te_len = tree_entry_len(&entry);
761
762                 if (match != all_entries_interesting) {
763                         strbuf_addstr(&name, base->buf + tn_len);
764                         match = tree_entry_interesting(&entry, &name,
765                                                        0, pathspec);
766                         strbuf_setlen(&name, name_base_len);
767
768                         if (match == all_entries_not_interesting)
769                                 break;
770                         if (match == entry_not_interesting)
771                                 continue;
772                 }
773
774                 strbuf_add(base, entry.path, te_len);
775
776                 if (S_ISREG(entry.mode)) {
777                         hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
778                                          check_attr ? base->buf + tn_len : NULL);
779                 } else if (S_ISDIR(entry.mode)) {
780                         enum object_type type;
781                         struct tree_desc sub;
782                         void *data;
783                         unsigned long size;
784
785                         data = lock_and_read_oid_file(entry.oid, &type, &size);
786                         if (!data)
787                                 die(_("unable to read tree (%s)"),
788                                     oid_to_hex(entry.oid));
789
790                         strbuf_addch(base, '/');
791                         init_tree_desc(&sub, data, size);
792                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
793                                          check_attr);
794                         free(data);
795                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
796                         hit |= grep_submodule(opt, entry.oid, base->buf,
797                                               base->buf + tn_len);
798                 }
799
800                 strbuf_setlen(base, old_baselen);
801
802                 if (hit && opt->status_only)
803                         break;
804         }
805
806         strbuf_release(&name);
807         return hit;
808 }
809
810 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
811                        struct object *obj, const char *name, const char *path)
812 {
813         if (obj->type == OBJ_BLOB)
814                 return grep_oid(opt, &obj->oid, name, 0, path);
815         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
816                 struct tree_desc tree;
817                 void *data;
818                 unsigned long size;
819                 struct strbuf base;
820                 int hit, len;
821
822                 grep_read_lock();
823                 data = read_object_with_reference(obj->oid.hash, tree_type,
824                                                   &size, NULL);
825                 grep_read_unlock();
826
827                 if (!data)
828                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
829
830                 /* Use parent's name as base when recursing submodules */
831                 if (recurse_submodules && parent_basename)
832                         name = parent_basename;
833
834                 len = name ? strlen(name) : 0;
835                 strbuf_init(&base, PATH_MAX + len + 1);
836                 if (len) {
837                         strbuf_add(&base, name, len);
838                         strbuf_addch(&base, ':');
839                 }
840                 init_tree_desc(&tree, data, size);
841                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
842                                 obj->type == OBJ_COMMIT);
843                 strbuf_release(&base);
844                 free(data);
845                 return hit;
846         }
847         die(_("unable to grep from object of type %s"), typename(obj->type));
848 }
849
850 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
851                         const struct object_array *list)
852 {
853         unsigned int i;
854         int hit = 0;
855         const unsigned int nr = list->nr;
856
857         for (i = 0; i < nr; i++) {
858                 struct object *real_obj;
859                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
860
861                 /* load the gitmodules file for this rev */
862                 if (recurse_submodules) {
863                         submodule_free();
864                         gitmodules_config_sha1(real_obj->oid.hash);
865                 }
866                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path)) {
867                         hit = 1;
868                         if (opt->status_only)
869                                 break;
870                 }
871         }
872         return hit;
873 }
874
875 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
876                           int exc_std, int use_index)
877 {
878         struct dir_struct dir;
879         int i, hit = 0;
880
881         memset(&dir, 0, sizeof(dir));
882         if (!use_index)
883                 dir.flags |= DIR_NO_GITLINKS;
884         if (exc_std)
885                 setup_standard_excludes(&dir);
886
887         fill_directory(&dir, &the_index, pathspec);
888         for (i = 0; i < dir.nr; i++) {
889                 if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
890                         continue;
891                 hit |= grep_file(opt, dir.entries[i]->name);
892                 if (hit && opt->status_only)
893                         break;
894         }
895         return hit;
896 }
897
898 static int context_callback(const struct option *opt, const char *arg,
899                             int unset)
900 {
901         struct grep_opt *grep_opt = opt->value;
902         int value;
903         const char *endp;
904
905         if (unset) {
906                 grep_opt->pre_context = grep_opt->post_context = 0;
907                 return 0;
908         }
909         value = strtol(arg, (char **)&endp, 10);
910         if (*endp) {
911                 return error(_("switch `%c' expects a numerical value"),
912                              opt->short_name);
913         }
914         grep_opt->pre_context = grep_opt->post_context = value;
915         return 0;
916 }
917
918 static int file_callback(const struct option *opt, const char *arg, int unset)
919 {
920         struct grep_opt *grep_opt = opt->value;
921         int from_stdin = !strcmp(arg, "-");
922         FILE *patterns;
923         int lno = 0;
924         struct strbuf sb = STRBUF_INIT;
925
926         patterns = from_stdin ? stdin : fopen(arg, "r");
927         if (!patterns)
928                 die_errno(_("cannot open '%s'"), arg);
929         while (strbuf_getline(&sb, patterns) == 0) {
930                 /* ignore empty line like grep does */
931                 if (sb.len == 0)
932                         continue;
933
934                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
935                                 GREP_PATTERN);
936         }
937         if (!from_stdin)
938                 fclose(patterns);
939         strbuf_release(&sb);
940         return 0;
941 }
942
943 static int not_callback(const struct option *opt, const char *arg, int unset)
944 {
945         struct grep_opt *grep_opt = opt->value;
946         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
947         return 0;
948 }
949
950 static int and_callback(const struct option *opt, const char *arg, int unset)
951 {
952         struct grep_opt *grep_opt = opt->value;
953         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
954         return 0;
955 }
956
957 static int open_callback(const struct option *opt, const char *arg, int unset)
958 {
959         struct grep_opt *grep_opt = opt->value;
960         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
961         return 0;
962 }
963
964 static int close_callback(const struct option *opt, const char *arg, int unset)
965 {
966         struct grep_opt *grep_opt = opt->value;
967         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
968         return 0;
969 }
970
971 static int pattern_callback(const struct option *opt, const char *arg,
972                             int unset)
973 {
974         struct grep_opt *grep_opt = opt->value;
975         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
976         return 0;
977 }
978
979 int cmd_grep(int argc, const char **argv, const char *prefix)
980 {
981         int hit = 0;
982         int cached = 0, untracked = 0, opt_exclude = -1;
983         int seen_dashdash = 0;
984         int external_grep_allowed__ignored;
985         const char *show_in_pager = NULL, *default_pager = "dummy";
986         struct grep_opt opt;
987         struct object_array list = OBJECT_ARRAY_INIT;
988         struct pathspec pathspec;
989         struct string_list path_list = STRING_LIST_INIT_NODUP;
990         int i;
991         int dummy;
992         int use_index = 1;
993         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
994         int allow_revs;
995
996         struct option options[] = {
997                 OPT_BOOL(0, "cached", &cached,
998                         N_("search in index instead of in the work tree")),
999                 OPT_NEGBIT(0, "no-index", &use_index,
1000                          N_("find in contents not managed by git"), 1),
1001                 OPT_BOOL(0, "untracked", &untracked,
1002                         N_("search in both tracked and untracked files")),
1003                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
1004                             N_("ignore files specified via '.gitignore'"), 1),
1005                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
1006                          N_("recursively search in each submodule")),
1007                 OPT_STRING(0, "parent-basename", &parent_basename,
1008                            N_("basename"),
1009                            N_("prepend parent project's basename to output")),
1010                 OPT_GROUP(""),
1011                 OPT_BOOL('v', "invert-match", &opt.invert,
1012                         N_("show non-matching lines")),
1013                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
1014                         N_("case insensitive matching")),
1015                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
1016                         N_("match patterns only at word boundaries")),
1017                 OPT_SET_INT('a', "text", &opt.binary,
1018                         N_("process binary files as text"), GREP_BINARY_TEXT),
1019                 OPT_SET_INT('I', NULL, &opt.binary,
1020                         N_("don't match patterns in binary files"),
1021                         GREP_BINARY_NOMATCH),
1022                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
1023                          N_("process binary files with textconv filters")),
1024                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
1025                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
1026                         NULL, 1 },
1027                 OPT_GROUP(""),
1028                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
1029                             N_("use extended POSIX regular expressions"),
1030                             GREP_PATTERN_TYPE_ERE),
1031                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
1032                             N_("use basic POSIX regular expressions (default)"),
1033                             GREP_PATTERN_TYPE_BRE),
1034                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
1035                             N_("interpret patterns as fixed strings"),
1036                             GREP_PATTERN_TYPE_FIXED),
1037                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
1038                             N_("use Perl-compatible regular expressions"),
1039                             GREP_PATTERN_TYPE_PCRE),
1040                 OPT_GROUP(""),
1041                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
1042                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
1043                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
1044                 OPT_NEGBIT(0, "full-name", &opt.relative,
1045                         N_("show filenames relative to top directory"), 1),
1046                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
1047                         N_("show only filenames instead of matching lines")),
1048                 OPT_BOOL(0, "name-only", &opt.name_only,
1049                         N_("synonym for --files-with-matches")),
1050                 OPT_BOOL('L', "files-without-match",
1051                         &opt.unmatch_name_only,
1052                         N_("show only the names of files without match")),
1053                 OPT_BOOL('z', "null", &opt.null_following_name,
1054                         N_("print NUL after filenames")),
1055                 OPT_BOOL('c', "count", &opt.count,
1056                         N_("show the number of matches instead of matching lines")),
1057                 OPT__COLOR(&opt.color, N_("highlight matches")),
1058                 OPT_BOOL(0, "break", &opt.file_break,
1059                         N_("print empty line between matches from different files")),
1060                 OPT_BOOL(0, "heading", &opt.heading,
1061                         N_("show filename only once above matches from same file")),
1062                 OPT_GROUP(""),
1063                 OPT_CALLBACK('C', "context", &opt, N_("n"),
1064                         N_("show <n> context lines before and after matches"),
1065                         context_callback),
1066                 OPT_INTEGER('B', "before-context", &opt.pre_context,
1067                         N_("show <n> context lines before matches")),
1068                 OPT_INTEGER('A', "after-context", &opt.post_context,
1069                         N_("show <n> context lines after matches")),
1070                 OPT_INTEGER(0, "threads", &num_threads,
1071                         N_("use <n> worker threads")),
1072                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
1073                         context_callback),
1074                 OPT_BOOL('p', "show-function", &opt.funcname,
1075                         N_("show a line with the function name before matches")),
1076                 OPT_BOOL('W', "function-context", &opt.funcbody,
1077                         N_("show the surrounding function")),
1078                 OPT_GROUP(""),
1079                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
1080                         N_("read patterns from file"), file_callback),
1081                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
1082                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
1083                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
1084                   N_("combine patterns specified with -e"),
1085                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
1086                 OPT_BOOL(0, "or", &dummy, ""),
1087                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
1088                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
1089                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
1090                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1091                   open_callback },
1092                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
1093                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1094                   close_callback },
1095                 OPT__QUIET(&opt.status_only,
1096                            N_("indicate hit with exit status without output")),
1097                 OPT_BOOL(0, "all-match", &opt.all_match,
1098                         N_("show only matches from files that match all patterns")),
1099                 { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
1100                   N_("show parse tree for grep expression"),
1101                   PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
1102                 OPT_GROUP(""),
1103                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
1104                         N_("pager"), N_("show matching files in the pager"),
1105                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
1106                 OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
1107                          N_("allow calling of grep(1) (ignored by this build)")),
1108                 OPT_END()
1109         };
1110
1111         init_grep_defaults();
1112         git_config(grep_cmd_config, NULL);
1113         grep_init(&opt, prefix);
1114         super_prefix = get_super_prefix();
1115
1116         /*
1117          * If there is no -- then the paths must exist in the working
1118          * tree.  If there is no explicit pattern specified with -e or
1119          * -f, we take the first unrecognized non option to be the
1120          * pattern, but then what follows it must be zero or more
1121          * valid refs up to the -- (if exists), and then existing
1122          * paths.  If there is an explicit pattern, then the first
1123          * unrecognized non option is the beginning of the refs list
1124          * that continues up to the -- (if exists), and then paths.
1125          */
1126         argc = parse_options(argc, argv, prefix, options, grep_usage,
1127                              PARSE_OPT_KEEP_DASHDASH |
1128                              PARSE_OPT_STOP_AT_NON_OPTION);
1129         grep_commit_pattern_type(pattern_type_arg, &opt);
1130
1131         if (use_index && !startup_info->have_repository) {
1132                 int fallback = 0;
1133                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
1134                 if (fallback)
1135                         use_index = 0;
1136                 else
1137                         /* die the same way as if we did it at the beginning */
1138                         setup_git_directory();
1139         }
1140
1141         /*
1142          * skip a -- separator; we know it cannot be
1143          * separating revisions from pathnames if
1144          * we haven't even had any patterns yet
1145          */
1146         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
1147                 argv++;
1148                 argc--;
1149         }
1150
1151         /* First unrecognized non-option token */
1152         if (argc > 0 && !opt.pattern_list) {
1153                 append_grep_pattern(&opt, argv[0], "command line", 0,
1154                                     GREP_PATTERN);
1155                 argv++;
1156                 argc--;
1157         }
1158
1159         if (show_in_pager == default_pager)
1160                 show_in_pager = git_pager(1);
1161         if (show_in_pager) {
1162                 opt.color = 0;
1163                 opt.name_only = 1;
1164                 opt.null_following_name = 1;
1165                 opt.output_priv = &path_list;
1166                 opt.output = append_path;
1167                 string_list_append(&path_list, show_in_pager);
1168         }
1169
1170         if (!opt.pattern_list)
1171                 die(_("no pattern given."));
1172         if (!opt.fixed && opt.ignore_case)
1173                 opt.regflags |= REG_ICASE;
1174
1175         /*
1176          * We have to find "--" in a separate pass, because its presence
1177          * influences how we will parse arguments that come before it.
1178          */
1179         for (i = 0; i < argc; i++) {
1180                 if (!strcmp(argv[i], "--")) {
1181                         seen_dashdash = 1;
1182                         break;
1183                 }
1184         }
1185
1186         /*
1187          * Resolve any rev arguments. If we have a dashdash, then everything up
1188          * to it must resolve as a rev. If not, then we stop at the first
1189          * non-rev and assume everything else is a path.
1190          */
1191         allow_revs = use_index && !untracked;
1192         for (i = 0; i < argc; i++) {
1193                 const char *arg = argv[i];
1194                 struct object_id oid;
1195                 struct object_context oc;
1196                 struct object *object;
1197
1198                 if (!strcmp(arg, "--")) {
1199                         i++;
1200                         break;
1201                 }
1202
1203                 if (!allow_revs) {
1204                         if (seen_dashdash)
1205                                 die(_("--no-index or --untracked cannot be used with revs"));
1206                         break;
1207                 }
1208
1209                 if (get_sha1_with_context(arg, GET_SHA1_RECORD_PATH,
1210                                           oid.hash, &oc)) {
1211                         if (seen_dashdash)
1212                                 die(_("unable to resolve revision: %s"), arg);
1213                         break;
1214                 }
1215
1216                 object = parse_object_or_die(&oid, arg);
1217                 if (!seen_dashdash)
1218                         verify_non_filename(prefix, arg);
1219                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1220                 free(oc.path);
1221         }
1222
1223         /*
1224          * Anything left over is presumed to be a path. But in the non-dashdash
1225          * "do what I mean" case, we verify and complain when that isn't true.
1226          */
1227         if (!seen_dashdash) {
1228                 int j;
1229                 for (j = i; j < argc; j++)
1230                         verify_filename(prefix, argv[j], j == i && allow_revs);
1231         }
1232
1233         parse_pathspec(&pathspec, 0,
1234                        PATHSPEC_PREFER_CWD |
1235                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1236                        prefix, argv + i);
1237         pathspec.max_depth = opt.max_depth;
1238         pathspec.recursive = 1;
1239
1240 #ifndef NO_PTHREADS
1241         if (list.nr || cached || show_in_pager)
1242                 num_threads = 0;
1243         else if (num_threads == 0)
1244                 num_threads = GREP_NUM_THREADS_DEFAULT;
1245         else if (num_threads < 0)
1246                 die(_("invalid number of threads specified (%d)"), num_threads);
1247         if (num_threads == 1)
1248                 num_threads = 0;
1249 #else
1250         if (num_threads)
1251                 warning(_("no threads support, ignoring --threads"));
1252         num_threads = 0;
1253 #endif
1254
1255         if (!num_threads)
1256                 /*
1257                  * The compiled patterns on the main path are only
1258                  * used when not using threading. Otherwise
1259                  * start_threads() below calls compile_grep_patterns()
1260                  * for each thread.
1261                  */
1262                 compile_grep_patterns(&opt);
1263
1264 #ifndef NO_PTHREADS
1265         if (num_threads) {
1266                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1267                     && (opt.pre_context || opt.post_context ||
1268                         opt.file_break || opt.funcbody))
1269                         skip_first_line = 1;
1270                 start_threads(&opt);
1271         }
1272 #endif
1273
1274         if (recurse_submodules) {
1275                 gitmodules_config();
1276                 compile_submodule_options(&opt, argv + i, cached, untracked,
1277                                           opt_exclude, use_index,
1278                                           pattern_type_arg);
1279         }
1280
1281         if (show_in_pager && (cached || list.nr))
1282                 die(_("--open-files-in-pager only works on the worktree"));
1283
1284         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1285                 const char *pager = path_list.items[0].string;
1286                 int len = strlen(pager);
1287
1288                 if (len > 4 && is_dir_sep(pager[len - 5]))
1289                         pager += len - 4;
1290
1291                 if (opt.ignore_case && !strcmp("less", pager))
1292                         string_list_append(&path_list, "-I");
1293
1294                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1295                         struct strbuf buf = STRBUF_INIT;
1296                         strbuf_addf(&buf, "+/%s%s",
1297                                         strcmp("less", pager) ? "" : "*",
1298                                         opt.pattern_list->pattern);
1299                         string_list_append(&path_list, buf.buf);
1300                         strbuf_detach(&buf, NULL);
1301                 }
1302         }
1303
1304         if (recurse_submodules && (!use_index || untracked))
1305                 die(_("option not supported with --recurse-submodules."));
1306
1307         if (!show_in_pager && !opt.status_only)
1308                 setup_pager();
1309
1310         if (!use_index && (untracked || cached))
1311                 die(_("--cached or --untracked cannot be used with --no-index."));
1312
1313         if (!use_index || untracked) {
1314                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1315                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1316         } else if (0 <= opt_exclude) {
1317                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1318         } else if (!list.nr) {
1319                 if (!cached)
1320                         setup_work_tree();
1321
1322                 hit = grep_cache(&opt, &pathspec, cached);
1323         } else {
1324                 if (cached)
1325                         die(_("both --cached and trees are given."));
1326                 hit = grep_objects(&opt, &pathspec, &list);
1327         }
1328
1329         if (num_threads)
1330                 hit |= wait_all();
1331         if (hit && show_in_pager)
1332                 run_pager(&opt, prefix);
1333         clear_pathspec(&pathspec);
1334         free_grep_patterns(&opt);
1335         return !hit;
1336 }