grep: add support for the PCRE v1 JIT API
[git] / builtin / grep.c
1 /*
2  * Builtin "git grep"
3  *
4  * Copyright (c) 2006 Junio C Hamano
5  */
6 #include "cache.h"
7 #include "blob.h"
8 #include "tree.h"
9 #include "commit.h"
10 #include "tag.h"
11 #include "tree-walk.h"
12 #include "builtin.h"
13 #include "parse-options.h"
14 #include "string-list.h"
15 #include "run-command.h"
16 #include "userdiff.h"
17 #include "grep.h"
18 #include "quote.h"
19 #include "dir.h"
20 #include "pathspec.h"
21 #include "submodule.h"
22 #include "submodule-config.h"
23
24 static char const * const grep_usage[] = {
25         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
26         NULL
27 };
28
29 static const char *super_prefix;
30 static int recurse_submodules;
31 static struct argv_array submodule_options = ARGV_ARRAY_INIT;
32 static const char *parent_basename;
33
34 static int grep_submodule_launch(struct grep_opt *opt,
35                                  const struct grep_source *gs);
36
37 #define GREP_NUM_THREADS_DEFAULT 8
38 static int num_threads;
39
40 #ifndef NO_PTHREADS
41 static pthread_t *threads;
42
43 /* We use one producer thread and THREADS consumer
44  * threads. The producer adds struct work_items to 'todo' and the
45  * consumers pick work items from the same array.
46  */
47 struct work_item {
48         struct grep_source source;
49         char done;
50         struct strbuf out;
51 };
52
53 /* In the range [todo_done, todo_start) in 'todo' we have work_items
54  * that have been or are processed by a consumer thread. We haven't
55  * written the result for these to stdout yet.
56  *
57  * The work_items in [todo_start, todo_end) are waiting to be picked
58  * up by a consumer thread.
59  *
60  * The ranges are modulo TODO_SIZE.
61  */
62 #define TODO_SIZE 128
63 static struct work_item todo[TODO_SIZE];
64 static int todo_start;
65 static int todo_end;
66 static int todo_done;
67
68 /* Has all work items been added? */
69 static int all_work_added;
70
71 /* This lock protects all the variables above. */
72 static pthread_mutex_t grep_mutex;
73
74 static inline void grep_lock(void)
75 {
76         assert(num_threads);
77         pthread_mutex_lock(&grep_mutex);
78 }
79
80 static inline void grep_unlock(void)
81 {
82         assert(num_threads);
83         pthread_mutex_unlock(&grep_mutex);
84 }
85
86 /* Signalled when a new work_item is added to todo. */
87 static pthread_cond_t cond_add;
88
89 /* Signalled when the result from one work_item is written to
90  * stdout.
91  */
92 static pthread_cond_t cond_write;
93
94 /* Signalled when we are finished with everything. */
95 static pthread_cond_t cond_result;
96
97 static int skip_first_line;
98
99 static void add_work(struct grep_opt *opt, enum grep_source_type type,
100                      const char *name, const char *path, const void *id)
101 {
102         grep_lock();
103
104         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
105                 pthread_cond_wait(&cond_write, &grep_mutex);
106         }
107
108         grep_source_init(&todo[todo_end].source, type, name, path, id);
109         if (opt->binary != GREP_BINARY_TEXT)
110                 grep_source_load_driver(&todo[todo_end].source);
111         todo[todo_end].done = 0;
112         strbuf_reset(&todo[todo_end].out);
113         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
114
115         pthread_cond_signal(&cond_add);
116         grep_unlock();
117 }
118
119 static struct work_item *get_work(void)
120 {
121         struct work_item *ret;
122
123         grep_lock();
124         while (todo_start == todo_end && !all_work_added) {
125                 pthread_cond_wait(&cond_add, &grep_mutex);
126         }
127
128         if (todo_start == todo_end && all_work_added) {
129                 ret = NULL;
130         } else {
131                 ret = &todo[todo_start];
132                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
133         }
134         grep_unlock();
135         return ret;
136 }
137
138 static void work_done(struct work_item *w)
139 {
140         int old_done;
141
142         grep_lock();
143         w->done = 1;
144         old_done = todo_done;
145         for(; todo[todo_done].done && todo_done != todo_start;
146             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
147                 w = &todo[todo_done];
148                 if (w->out.len) {
149                         const char *p = w->out.buf;
150                         size_t len = w->out.len;
151
152                         /* Skip the leading hunk mark of the first file. */
153                         if (skip_first_line) {
154                                 while (len) {
155                                         len--;
156                                         if (*p++ == '\n')
157                                                 break;
158                                 }
159                                 skip_first_line = 0;
160                         }
161
162                         write_or_die(1, p, len);
163                 }
164                 grep_source_clear(&w->source);
165         }
166
167         if (old_done != todo_done)
168                 pthread_cond_signal(&cond_write);
169
170         if (all_work_added && todo_done == todo_end)
171                 pthread_cond_signal(&cond_result);
172
173         grep_unlock();
174 }
175
176 static void *run(void *arg)
177 {
178         int hit = 0;
179         struct grep_opt *opt = arg;
180
181         while (1) {
182                 struct work_item *w = get_work();
183                 if (!w)
184                         break;
185
186                 opt->output_priv = w;
187                 if (w->source.type == GREP_SOURCE_SUBMODULE)
188                         hit |= grep_submodule_launch(opt, &w->source);
189                 else
190                         hit |= grep_source(opt, &w->source);
191                 grep_source_clear_data(&w->source);
192                 work_done(w);
193         }
194         free_grep_patterns(arg);
195         free(arg);
196
197         return (void*) (intptr_t) hit;
198 }
199
200 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
201 {
202         struct work_item *w = opt->output_priv;
203         strbuf_add(&w->out, buf, size);
204 }
205
206 static void start_threads(struct grep_opt *opt)
207 {
208         int i;
209
210         pthread_mutex_init(&grep_mutex, NULL);
211         pthread_mutex_init(&grep_read_mutex, NULL);
212         pthread_mutex_init(&grep_attr_mutex, NULL);
213         pthread_cond_init(&cond_add, NULL);
214         pthread_cond_init(&cond_write, NULL);
215         pthread_cond_init(&cond_result, NULL);
216         grep_use_locks = 1;
217
218         for (i = 0; i < ARRAY_SIZE(todo); i++) {
219                 strbuf_init(&todo[i].out, 0);
220         }
221
222         threads = xcalloc(num_threads, sizeof(*threads));
223         for (i = 0; i < num_threads; i++) {
224                 int err;
225                 struct grep_opt *o = grep_opt_dup(opt);
226                 o->output = strbuf_out;
227                 if (i)
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 #ifdef NO_PTHREADS
294                 else if (num_threads && num_threads != 1) {
295                         /*
296                          * TRANSLATORS: %s is the configuration
297                          * variable for tweaking threads, currently
298                          * grep.threads
299                          */
300                         warning(_("no threads support, ignoring %s"), var);
301                         num_threads = 0;
302                 }
303 #endif
304         }
305
306         return st;
307 }
308
309 static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
310 {
311         void *data;
312
313         grep_read_lock();
314         data = read_sha1_file(oid->hash, type, size);
315         grep_read_unlock();
316         return data;
317 }
318
319 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
320                      const char *filename, int tree_name_len,
321                      const char *path)
322 {
323         struct strbuf pathbuf = STRBUF_INIT;
324
325         if (super_prefix) {
326                 strbuf_add(&pathbuf, filename, tree_name_len);
327                 strbuf_addstr(&pathbuf, super_prefix);
328                 strbuf_addstr(&pathbuf, filename + tree_name_len);
329         } else {
330                 strbuf_addstr(&pathbuf, filename);
331         }
332
333         if (opt->relative && opt->prefix_length) {
334                 char *name = strbuf_detach(&pathbuf, NULL);
335                 quote_path_relative(name + tree_name_len, opt->prefix, &pathbuf);
336                 strbuf_insert(&pathbuf, 0, name, tree_name_len);
337                 free(name);
338         }
339
340 #ifndef NO_PTHREADS
341         if (num_threads) {
342                 add_work(opt, GREP_SOURCE_SHA1, pathbuf.buf, path, oid);
343                 strbuf_release(&pathbuf);
344                 return 0;
345         } else
346 #endif
347         {
348                 struct grep_source gs;
349                 int hit;
350
351                 grep_source_init(&gs, GREP_SOURCE_SHA1, pathbuf.buf, path, oid);
352                 strbuf_release(&pathbuf);
353                 hit = grep_source(opt, &gs);
354
355                 grep_source_clear(&gs);
356                 return hit;
357         }
358 }
359
360 static int grep_file(struct grep_opt *opt, const char *filename)
361 {
362         struct strbuf buf = STRBUF_INIT;
363
364         if (super_prefix)
365                 strbuf_addstr(&buf, super_prefix);
366         strbuf_addstr(&buf, filename);
367
368         if (opt->relative && opt->prefix_length) {
369                 char *name = strbuf_detach(&buf, NULL);
370                 quote_path_relative(name, opt->prefix, &buf);
371                 free(name);
372         }
373
374 #ifndef NO_PTHREADS
375         if (num_threads) {
376                 add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
377                 strbuf_release(&buf);
378                 return 0;
379         } else
380 #endif
381         {
382                 struct grep_source gs;
383                 int hit;
384
385                 grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
386                 strbuf_release(&buf);
387                 hit = grep_source(opt, &gs);
388
389                 grep_source_clear(&gs);
390                 return hit;
391         }
392 }
393
394 static void append_path(struct grep_opt *opt, const void *data, size_t len)
395 {
396         struct string_list *path_list = opt->output_priv;
397
398         if (len == 1 && *(const char *)data == '\0')
399                 return;
400         string_list_append(path_list, xstrndup(data, len));
401 }
402
403 static void run_pager(struct grep_opt *opt, const char *prefix)
404 {
405         struct string_list *path_list = opt->output_priv;
406         struct child_process child = CHILD_PROCESS_INIT;
407         int i, status;
408
409         for (i = 0; i < path_list->nr; i++)
410                 argv_array_push(&child.args, path_list->items[i].string);
411         child.dir = prefix;
412         child.use_shell = 1;
413
414         status = run_command(&child);
415         if (status)
416                 exit(status);
417 }
418
419 static void compile_submodule_options(const struct grep_opt *opt,
420                                       const char **argv,
421                                       int cached, int untracked,
422                                       int opt_exclude, int use_index,
423                                       int pattern_type_arg)
424 {
425         struct grep_pat *pattern;
426
427         if (recurse_submodules)
428                 argv_array_push(&submodule_options, "--recurse-submodules");
429
430         if (cached)
431                 argv_array_push(&submodule_options, "--cached");
432         if (!use_index)
433                 argv_array_push(&submodule_options, "--no-index");
434         if (untracked)
435                 argv_array_push(&submodule_options, "--untracked");
436         if (opt_exclude > 0)
437                 argv_array_push(&submodule_options, "--exclude-standard");
438
439         if (opt->invert)
440                 argv_array_push(&submodule_options, "-v");
441         if (opt->ignore_case)
442                 argv_array_push(&submodule_options, "-i");
443         if (opt->word_regexp)
444                 argv_array_push(&submodule_options, "-w");
445         switch (opt->binary) {
446         case GREP_BINARY_NOMATCH:
447                 argv_array_push(&submodule_options, "-I");
448                 break;
449         case GREP_BINARY_TEXT:
450                 argv_array_push(&submodule_options, "-a");
451                 break;
452         default:
453                 break;
454         }
455         if (opt->allow_textconv)
456                 argv_array_push(&submodule_options, "--textconv");
457         if (opt->max_depth != -1)
458                 argv_array_pushf(&submodule_options, "--max-depth=%d",
459                                  opt->max_depth);
460         if (opt->linenum)
461                 argv_array_push(&submodule_options, "-n");
462         if (!opt->pathname)
463                 argv_array_push(&submodule_options, "-h");
464         if (!opt->relative)
465                 argv_array_push(&submodule_options, "--full-name");
466         if (opt->name_only)
467                 argv_array_push(&submodule_options, "-l");
468         if (opt->unmatch_name_only)
469                 argv_array_push(&submodule_options, "-L");
470         if (opt->null_following_name)
471                 argv_array_push(&submodule_options, "-z");
472         if (opt->count)
473                 argv_array_push(&submodule_options, "-c");
474         if (opt->file_break)
475                 argv_array_push(&submodule_options, "--break");
476         if (opt->heading)
477                 argv_array_push(&submodule_options, "--heading");
478         if (opt->pre_context)
479                 argv_array_pushf(&submodule_options, "--before-context=%d",
480                                  opt->pre_context);
481         if (opt->post_context)
482                 argv_array_pushf(&submodule_options, "--after-context=%d",
483                                  opt->post_context);
484         if (opt->funcname)
485                 argv_array_push(&submodule_options, "-p");
486         if (opt->funcbody)
487                 argv_array_push(&submodule_options, "-W");
488         if (opt->all_match)
489                 argv_array_push(&submodule_options, "--all-match");
490         if (opt->debug)
491                 argv_array_push(&submodule_options, "--debug");
492         if (opt->status_only)
493                 argv_array_push(&submodule_options, "-q");
494
495         switch (pattern_type_arg) {
496         case GREP_PATTERN_TYPE_BRE:
497                 argv_array_push(&submodule_options, "-G");
498                 break;
499         case GREP_PATTERN_TYPE_ERE:
500                 argv_array_push(&submodule_options, "-E");
501                 break;
502         case GREP_PATTERN_TYPE_FIXED:
503                 argv_array_push(&submodule_options, "-F");
504                 break;
505         case GREP_PATTERN_TYPE_PCRE:
506                 argv_array_push(&submodule_options, "-P");
507                 break;
508         case GREP_PATTERN_TYPE_UNSPECIFIED:
509                 break;
510         default:
511                 die("BUG: Added a new grep pattern type without updating switch statement");
512         }
513
514         for (pattern = opt->pattern_list; pattern != NULL;
515              pattern = pattern->next) {
516                 switch (pattern->token) {
517                 case GREP_PATTERN:
518                         argv_array_pushf(&submodule_options, "-e%s",
519                                          pattern->pattern);
520                         break;
521                 case GREP_AND:
522                 case GREP_OPEN_PAREN:
523                 case GREP_CLOSE_PAREN:
524                 case GREP_NOT:
525                 case GREP_OR:
526                         argv_array_push(&submodule_options, pattern->pattern);
527                         break;
528                 /* BODY and HEAD are not used by git-grep */
529                 case GREP_PATTERN_BODY:
530                 case GREP_PATTERN_HEAD:
531                         break;
532                 }
533         }
534
535         /*
536          * Limit number of threads for child process to use.
537          * This is to prevent potential fork-bomb behavior of git-grep as each
538          * submodule process has its own thread pool.
539          */
540         argv_array_pushf(&submodule_options, "--threads=%d",
541                          (num_threads + 1) / 2);
542
543         /* Add Pathspecs */
544         argv_array_push(&submodule_options, "--");
545         for (; *argv; argv++)
546                 argv_array_push(&submodule_options, *argv);
547 }
548
549 /*
550  * Launch child process to grep contents of a submodule
551  */
552 static int grep_submodule_launch(struct grep_opt *opt,
553                                  const struct grep_source *gs)
554 {
555         struct child_process cp = CHILD_PROCESS_INIT;
556         int status, i;
557         const char *end_of_base;
558         const char *name;
559         struct strbuf child_output = STRBUF_INIT;
560
561         end_of_base = strchr(gs->name, ':');
562         if (gs->identifier && end_of_base)
563                 name = end_of_base + 1;
564         else
565                 name = gs->name;
566
567         prepare_submodule_repo_env(&cp.env_array);
568         argv_array_push(&cp.env_array, GIT_DIR_ENVIRONMENT);
569
570         if (opt->relative && opt->prefix_length)
571                 argv_array_pushf(&cp.env_array, "%s=%s",
572                                  GIT_TOPLEVEL_PREFIX_ENVIRONMENT,
573                                  opt->prefix);
574
575         /* Add super prefix */
576         argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
577                          super_prefix ? super_prefix : "",
578                          name);
579         argv_array_push(&cp.args, "grep");
580
581         /*
582          * Add basename of parent project
583          * When performing grep on a tree object the filename is prefixed
584          * with the object's name: 'tree-name:filename'.  In order to
585          * provide uniformity of output we want to pass the name of the
586          * parent project's object name to the submodule so the submodule can
587          * prefix its output with the parent's name and not its own SHA1.
588          */
589         if (gs->identifier && end_of_base)
590                 argv_array_pushf(&cp.args, "--parent-basename=%.*s",
591                                  (int) (end_of_base - gs->name),
592                                  gs->name);
593
594         /* Add options */
595         for (i = 0; i < submodule_options.argc; i++) {
596                 /*
597                  * If there is a tree identifier for the submodule, add the
598                  * rev after adding the submodule options but before the
599                  * pathspecs.  To do this we listen for the '--' and insert the
600                  * sha1 before pushing the '--' onto the child process argv
601                  * array.
602                  */
603                 if (gs->identifier &&
604                     !strcmp("--", submodule_options.argv[i])) {
605                         argv_array_push(&cp.args, sha1_to_hex(gs->identifier));
606                 }
607
608                 argv_array_push(&cp.args, submodule_options.argv[i]);
609         }
610
611         cp.git_cmd = 1;
612         cp.dir = gs->path;
613
614         /*
615          * Capture output to output buffer and check the return code from the
616          * child process.  A '0' indicates a hit, a '1' indicates no hit and
617          * anything else is an error.
618          */
619         status = capture_command(&cp, &child_output, 0);
620         if (status && (status != 1)) {
621                 /* flush the buffer */
622                 write_or_die(1, child_output.buf, child_output.len);
623                 die("process for submodule '%s' failed with exit code: %d",
624                     gs->name, status);
625         }
626
627         opt->output(opt, child_output.buf, child_output.len);
628         strbuf_release(&child_output);
629         /* invert the return code to make a hit equal to 1 */
630         return !status;
631 }
632
633 /*
634  * Prep grep structures for a submodule grep
635  * sha1: the sha1 of the submodule or NULL if using the working tree
636  * filename: name of the submodule including tree name of parent
637  * path: location of the submodule
638  */
639 static int grep_submodule(struct grep_opt *opt, const unsigned char *sha1,
640                           const char *filename, const char *path)
641 {
642         if (!is_submodule_initialized(path))
643                 return 0;
644         if (!is_submodule_populated_gently(path, NULL)) {
645                 /*
646                  * If searching history, check for the presense of the
647                  * submodule's gitdir before skipping the submodule.
648                  */
649                 if (sha1) {
650                         const struct submodule *sub =
651                                         submodule_from_path(null_sha1, path);
652                         if (sub)
653                                 path = git_path("modules/%s", sub->name);
654
655                         if (!(is_directory(path) && is_git_directory(path)))
656                                 return 0;
657                 } else {
658                         return 0;
659                 }
660         }
661
662 #ifndef NO_PTHREADS
663         if (num_threads) {
664                 add_work(opt, GREP_SOURCE_SUBMODULE, filename, path, sha1);
665                 return 0;
666         } else
667 #endif
668         {
669                 struct grep_source gs;
670                 int hit;
671
672                 grep_source_init(&gs, GREP_SOURCE_SUBMODULE,
673                                  filename, path, sha1);
674                 hit = grep_submodule_launch(opt, &gs);
675
676                 grep_source_clear(&gs);
677                 return hit;
678         }
679 }
680
681 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec,
682                       int cached)
683 {
684         int hit = 0;
685         int nr;
686         struct strbuf name = STRBUF_INIT;
687         int name_base_len = 0;
688         if (super_prefix) {
689                 name_base_len = strlen(super_prefix);
690                 strbuf_addstr(&name, super_prefix);
691         }
692
693         read_cache();
694
695         for (nr = 0; nr < active_nr; nr++) {
696                 const struct cache_entry *ce = active_cache[nr];
697                 strbuf_setlen(&name, name_base_len);
698                 strbuf_addstr(&name, ce->name);
699
700                 if (S_ISREG(ce->ce_mode) &&
701                     match_pathspec(pathspec, name.buf, name.len, 0, NULL,
702                                    S_ISDIR(ce->ce_mode) ||
703                                    S_ISGITLINK(ce->ce_mode))) {
704                         /*
705                          * If CE_VALID is on, we assume worktree file and its
706                          * cache entry are identical, even if worktree file has
707                          * been modified, so use cache version instead
708                          */
709                         if (cached || (ce->ce_flags & CE_VALID) ||
710                             ce_skip_worktree(ce)) {
711                                 if (ce_stage(ce) || ce_intent_to_add(ce))
712                                         continue;
713                                 hit |= grep_oid(opt, &ce->oid, ce->name,
714                                                  0, ce->name);
715                         } else {
716                                 hit |= grep_file(opt, ce->name);
717                         }
718                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
719                            submodule_path_match(pathspec, name.buf, NULL)) {
720                         hit |= grep_submodule(opt, NULL, ce->name, ce->name);
721                 } else {
722                         continue;
723                 }
724
725                 if (ce_stage(ce)) {
726                         do {
727                                 nr++;
728                         } while (nr < active_nr &&
729                                  !strcmp(ce->name, active_cache[nr]->name));
730                         nr--; /* compensate for loop control */
731                 }
732                 if (hit && opt->status_only)
733                         break;
734         }
735
736         strbuf_release(&name);
737         return hit;
738 }
739
740 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
741                      struct tree_desc *tree, struct strbuf *base, int tn_len,
742                      int check_attr)
743 {
744         int hit = 0;
745         enum interesting match = entry_not_interesting;
746         struct name_entry entry;
747         int old_baselen = base->len;
748         struct strbuf name = STRBUF_INIT;
749         int name_base_len = 0;
750         if (super_prefix) {
751                 strbuf_addstr(&name, super_prefix);
752                 name_base_len = name.len;
753         }
754
755         while (tree_entry(tree, &entry)) {
756                 int te_len = tree_entry_len(&entry);
757
758                 if (match != all_entries_interesting) {
759                         strbuf_addstr(&name, base->buf + tn_len);
760                         match = tree_entry_interesting(&entry, &name,
761                                                        0, pathspec);
762                         strbuf_setlen(&name, name_base_len);
763
764                         if (match == all_entries_not_interesting)
765                                 break;
766                         if (match == entry_not_interesting)
767                                 continue;
768                 }
769
770                 strbuf_add(base, entry.path, te_len);
771
772                 if (S_ISREG(entry.mode)) {
773                         hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
774                                          check_attr ? base->buf + tn_len : NULL);
775                 } else if (S_ISDIR(entry.mode)) {
776                         enum object_type type;
777                         struct tree_desc sub;
778                         void *data;
779                         unsigned long size;
780
781                         data = lock_and_read_oid_file(entry.oid, &type, &size);
782                         if (!data)
783                                 die(_("unable to read tree (%s)"),
784                                     oid_to_hex(entry.oid));
785
786                         strbuf_addch(base, '/');
787                         init_tree_desc(&sub, data, size);
788                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
789                                          check_attr);
790                         free(data);
791                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
792                         hit |= grep_submodule(opt, entry.oid->hash, base->buf,
793                                               base->buf + tn_len);
794                 }
795
796                 strbuf_setlen(base, old_baselen);
797
798                 if (hit && opt->status_only)
799                         break;
800         }
801
802         strbuf_release(&name);
803         return hit;
804 }
805
806 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
807                        struct object *obj, const char *name, const char *path)
808 {
809         if (obj->type == OBJ_BLOB)
810                 return grep_oid(opt, &obj->oid, name, 0, path);
811         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
812                 struct tree_desc tree;
813                 void *data;
814                 unsigned long size;
815                 struct strbuf base;
816                 int hit, len;
817
818                 grep_read_lock();
819                 data = read_object_with_reference(obj->oid.hash, tree_type,
820                                                   &size, NULL);
821                 grep_read_unlock();
822
823                 if (!data)
824                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
825
826                 /* Use parent's name as base when recursing submodules */
827                 if (recurse_submodules && parent_basename)
828                         name = parent_basename;
829
830                 len = name ? strlen(name) : 0;
831                 strbuf_init(&base, PATH_MAX + len + 1);
832                 if (len) {
833                         strbuf_add(&base, name, len);
834                         strbuf_addch(&base, ':');
835                 }
836                 init_tree_desc(&tree, data, size);
837                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
838                                 obj->type == OBJ_COMMIT);
839                 strbuf_release(&base);
840                 free(data);
841                 return hit;
842         }
843         die(_("unable to grep from object of type %s"), typename(obj->type));
844 }
845
846 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
847                         const struct object_array *list)
848 {
849         unsigned int i;
850         int hit = 0;
851         const unsigned int nr = list->nr;
852
853         for (i = 0; i < nr; i++) {
854                 struct object *real_obj;
855                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
856
857                 /* load the gitmodules file for this rev */
858                 if (recurse_submodules) {
859                         submodule_free();
860                         gitmodules_config_sha1(real_obj->oid.hash);
861                 }
862                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path)) {
863                         hit = 1;
864                         if (opt->status_only)
865                                 break;
866                 }
867         }
868         return hit;
869 }
870
871 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
872                           int exc_std, int use_index)
873 {
874         struct dir_struct dir;
875         int i, hit = 0;
876
877         memset(&dir, 0, sizeof(dir));
878         if (!use_index)
879                 dir.flags |= DIR_NO_GITLINKS;
880         if (exc_std)
881                 setup_standard_excludes(&dir);
882
883         fill_directory(&dir, pathspec);
884         for (i = 0; i < dir.nr; i++) {
885                 if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
886                         continue;
887                 hit |= grep_file(opt, dir.entries[i]->name);
888                 if (hit && opt->status_only)
889                         break;
890         }
891         return hit;
892 }
893
894 static int context_callback(const struct option *opt, const char *arg,
895                             int unset)
896 {
897         struct grep_opt *grep_opt = opt->value;
898         int value;
899         const char *endp;
900
901         if (unset) {
902                 grep_opt->pre_context = grep_opt->post_context = 0;
903                 return 0;
904         }
905         value = strtol(arg, (char **)&endp, 10);
906         if (*endp) {
907                 return error(_("switch `%c' expects a numerical value"),
908                              opt->short_name);
909         }
910         grep_opt->pre_context = grep_opt->post_context = value;
911         return 0;
912 }
913
914 static int file_callback(const struct option *opt, const char *arg, int unset)
915 {
916         struct grep_opt *grep_opt = opt->value;
917         int from_stdin = !strcmp(arg, "-");
918         FILE *patterns;
919         int lno = 0;
920         struct strbuf sb = STRBUF_INIT;
921
922         patterns = from_stdin ? stdin : fopen(arg, "r");
923         if (!patterns)
924                 die_errno(_("cannot open '%s'"), arg);
925         while (strbuf_getline(&sb, patterns) == 0) {
926                 /* ignore empty line like grep does */
927                 if (sb.len == 0)
928                         continue;
929
930                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
931                                 GREP_PATTERN);
932         }
933         if (!from_stdin)
934                 fclose(patterns);
935         strbuf_release(&sb);
936         return 0;
937 }
938
939 static int not_callback(const struct option *opt, const char *arg, int unset)
940 {
941         struct grep_opt *grep_opt = opt->value;
942         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
943         return 0;
944 }
945
946 static int and_callback(const struct option *opt, const char *arg, int unset)
947 {
948         struct grep_opt *grep_opt = opt->value;
949         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
950         return 0;
951 }
952
953 static int open_callback(const struct option *opt, const char *arg, int unset)
954 {
955         struct grep_opt *grep_opt = opt->value;
956         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
957         return 0;
958 }
959
960 static int close_callback(const struct option *opt, const char *arg, int unset)
961 {
962         struct grep_opt *grep_opt = opt->value;
963         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
964         return 0;
965 }
966
967 static int pattern_callback(const struct option *opt, const char *arg,
968                             int unset)
969 {
970         struct grep_opt *grep_opt = opt->value;
971         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
972         return 0;
973 }
974
975 int cmd_grep(int argc, const char **argv, const char *prefix)
976 {
977         int hit = 0;
978         int cached = 0, untracked = 0, opt_exclude = -1;
979         int seen_dashdash = 0;
980         int external_grep_allowed__ignored;
981         const char *show_in_pager = NULL, *default_pager = "dummy";
982         struct grep_opt opt;
983         struct object_array list = OBJECT_ARRAY_INIT;
984         struct pathspec pathspec;
985         struct string_list path_list = STRING_LIST_INIT_NODUP;
986         int i;
987         int dummy;
988         int use_index = 1;
989         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
990         int allow_revs;
991
992         struct option options[] = {
993                 OPT_BOOL(0, "cached", &cached,
994                         N_("search in index instead of in the work tree")),
995                 OPT_NEGBIT(0, "no-index", &use_index,
996                          N_("find in contents not managed by git"), 1),
997                 OPT_BOOL(0, "untracked", &untracked,
998                         N_("search in both tracked and untracked files")),
999                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
1000                             N_("ignore files specified via '.gitignore'"), 1),
1001                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
1002                          N_("recursively search in each submodule")),
1003                 OPT_STRING(0, "parent-basename", &parent_basename,
1004                            N_("basename"),
1005                            N_("prepend parent project's basename to output")),
1006                 OPT_GROUP(""),
1007                 OPT_BOOL('v', "invert-match", &opt.invert,
1008                         N_("show non-matching lines")),
1009                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
1010                         N_("case insensitive matching")),
1011                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
1012                         N_("match patterns only at word boundaries")),
1013                 OPT_SET_INT('a', "text", &opt.binary,
1014                         N_("process binary files as text"), GREP_BINARY_TEXT),
1015                 OPT_SET_INT('I', NULL, &opt.binary,
1016                         N_("don't match patterns in binary files"),
1017                         GREP_BINARY_NOMATCH),
1018                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
1019                          N_("process binary files with textconv filters")),
1020                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
1021                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
1022                         NULL, 1 },
1023                 OPT_GROUP(""),
1024                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
1025                             N_("use extended POSIX regular expressions"),
1026                             GREP_PATTERN_TYPE_ERE),
1027                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
1028                             N_("use basic POSIX regular expressions (default)"),
1029                             GREP_PATTERN_TYPE_BRE),
1030                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
1031                             N_("interpret patterns as fixed strings"),
1032                             GREP_PATTERN_TYPE_FIXED),
1033                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
1034                             N_("use Perl-compatible regular expressions"),
1035                             GREP_PATTERN_TYPE_PCRE),
1036                 OPT_GROUP(""),
1037                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
1038                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
1039                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
1040                 OPT_NEGBIT(0, "full-name", &opt.relative,
1041                         N_("show filenames relative to top directory"), 1),
1042                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
1043                         N_("show only filenames instead of matching lines")),
1044                 OPT_BOOL(0, "name-only", &opt.name_only,
1045                         N_("synonym for --files-with-matches")),
1046                 OPT_BOOL('L', "files-without-match",
1047                         &opt.unmatch_name_only,
1048                         N_("show only the names of files without match")),
1049                 OPT_BOOL('z', "null", &opt.null_following_name,
1050                         N_("print NUL after filenames")),
1051                 OPT_BOOL('c', "count", &opt.count,
1052                         N_("show the number of matches instead of matching lines")),
1053                 OPT__COLOR(&opt.color, N_("highlight matches")),
1054                 OPT_BOOL(0, "break", &opt.file_break,
1055                         N_("print empty line between matches from different files")),
1056                 OPT_BOOL(0, "heading", &opt.heading,
1057                         N_("show filename only once above matches from same file")),
1058                 OPT_GROUP(""),
1059                 OPT_CALLBACK('C', "context", &opt, N_("n"),
1060                         N_("show <n> context lines before and after matches"),
1061                         context_callback),
1062                 OPT_INTEGER('B', "before-context", &opt.pre_context,
1063                         N_("show <n> context lines before matches")),
1064                 OPT_INTEGER('A', "after-context", &opt.post_context,
1065                         N_("show <n> context lines after matches")),
1066                 OPT_INTEGER(0, "threads", &num_threads,
1067                         N_("use <n> worker threads")),
1068                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
1069                         context_callback),
1070                 OPT_BOOL('p', "show-function", &opt.funcname,
1071                         N_("show a line with the function name before matches")),
1072                 OPT_BOOL('W', "function-context", &opt.funcbody,
1073                         N_("show the surrounding function")),
1074                 OPT_GROUP(""),
1075                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
1076                         N_("read patterns from file"), file_callback),
1077                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
1078                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
1079                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
1080                   N_("combine patterns specified with -e"),
1081                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
1082                 OPT_BOOL(0, "or", &dummy, ""),
1083                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
1084                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
1085                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
1086                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1087                   open_callback },
1088                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
1089                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1090                   close_callback },
1091                 OPT__QUIET(&opt.status_only,
1092                            N_("indicate hit with exit status without output")),
1093                 OPT_BOOL(0, "all-match", &opt.all_match,
1094                         N_("show only matches from files that match all patterns")),
1095                 { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
1096                   N_("show parse tree for grep expression"),
1097                   PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
1098                 OPT_GROUP(""),
1099                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
1100                         N_("pager"), N_("show matching files in the pager"),
1101                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
1102                 OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
1103                          N_("allow calling of grep(1) (ignored by this build)")),
1104                 OPT_END()
1105         };
1106
1107         init_grep_defaults();
1108         git_config(grep_cmd_config, NULL);
1109         grep_init(&opt, prefix);
1110         super_prefix = get_super_prefix();
1111
1112         /*
1113          * If there is no -- then the paths must exist in the working
1114          * tree.  If there is no explicit pattern specified with -e or
1115          * -f, we take the first unrecognized non option to be the
1116          * pattern, but then what follows it must be zero or more
1117          * valid refs up to the -- (if exists), and then existing
1118          * paths.  If there is an explicit pattern, then the first
1119          * unrecognized non option is the beginning of the refs list
1120          * that continues up to the -- (if exists), and then paths.
1121          */
1122         argc = parse_options(argc, argv, prefix, options, grep_usage,
1123                              PARSE_OPT_KEEP_DASHDASH |
1124                              PARSE_OPT_STOP_AT_NON_OPTION);
1125         grep_commit_pattern_type(pattern_type_arg, &opt);
1126
1127         if (use_index && !startup_info->have_repository) {
1128                 int fallback = 0;
1129                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
1130                 if (fallback)
1131                         use_index = 0;
1132                 else
1133                         /* die the same way as if we did it at the beginning */
1134                         setup_git_directory();
1135         }
1136
1137         /*
1138          * skip a -- separator; we know it cannot be
1139          * separating revisions from pathnames if
1140          * we haven't even had any patterns yet
1141          */
1142         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
1143                 argv++;
1144                 argc--;
1145         }
1146
1147         /* First unrecognized non-option token */
1148         if (argc > 0 && !opt.pattern_list) {
1149                 append_grep_pattern(&opt, argv[0], "command line", 0,
1150                                     GREP_PATTERN);
1151                 argv++;
1152                 argc--;
1153         }
1154
1155         if (show_in_pager == default_pager)
1156                 show_in_pager = git_pager(1);
1157         if (show_in_pager) {
1158                 opt.color = 0;
1159                 opt.name_only = 1;
1160                 opt.null_following_name = 1;
1161                 opt.output_priv = &path_list;
1162                 opt.output = append_path;
1163                 string_list_append(&path_list, show_in_pager);
1164         }
1165
1166         if (!opt.pattern_list)
1167                 die(_("no pattern given."));
1168         if (!opt.fixed && opt.ignore_case)
1169                 opt.regflags |= REG_ICASE;
1170
1171         /*
1172          * We have to find "--" in a separate pass, because its presence
1173          * influences how we will parse arguments that come before it.
1174          */
1175         for (i = 0; i < argc; i++) {
1176                 if (!strcmp(argv[i], "--")) {
1177                         seen_dashdash = 1;
1178                         break;
1179                 }
1180         }
1181
1182         /*
1183          * Resolve any rev arguments. If we have a dashdash, then everything up
1184          * to it must resolve as a rev. If not, then we stop at the first
1185          * non-rev and assume everything else is a path.
1186          */
1187         allow_revs = use_index && !untracked;
1188         for (i = 0; i < argc; i++) {
1189                 const char *arg = argv[i];
1190                 struct object_id oid;
1191                 struct object_context oc;
1192                 struct object *object;
1193
1194                 if (!strcmp(arg, "--")) {
1195                         i++;
1196                         break;
1197                 }
1198
1199                 if (!allow_revs) {
1200                         if (seen_dashdash)
1201                                 die(_("--no-index or --untracked cannot be used with revs"));
1202                         break;
1203                 }
1204
1205                 if (get_sha1_with_context(arg, 0, oid.hash, &oc)) {
1206                         if (seen_dashdash)
1207                                 die(_("unable to resolve revision: %s"), arg);
1208                         break;
1209                 }
1210
1211                 object = parse_object_or_die(oid.hash, arg);
1212                 if (!seen_dashdash)
1213                         verify_non_filename(prefix, arg);
1214                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1215         }
1216
1217         /*
1218          * Anything left over is presumed to be a path. But in the non-dashdash
1219          * "do what I mean" case, we verify and complain when that isn't true.
1220          */
1221         if (!seen_dashdash) {
1222                 int j;
1223                 for (j = i; j < argc; j++)
1224                         verify_filename(prefix, argv[j], j == i && allow_revs);
1225         }
1226
1227         parse_pathspec(&pathspec, 0,
1228                        PATHSPEC_PREFER_CWD |
1229                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1230                        prefix, argv + i);
1231         pathspec.max_depth = opt.max_depth;
1232         pathspec.recursive = 1;
1233
1234 #ifndef NO_PTHREADS
1235         if (list.nr || cached || show_in_pager)
1236                 num_threads = 0;
1237         else if (num_threads == 0)
1238                 num_threads = GREP_NUM_THREADS_DEFAULT;
1239         else if (num_threads < 0)
1240                 die(_("invalid number of threads specified (%d)"), num_threads);
1241         if (num_threads == 1)
1242                 num_threads = 0;
1243 #else
1244         if (num_threads)
1245                 warning(_("no threads support, ignoring --threads"));
1246         num_threads = 0;
1247 #endif
1248
1249         if (!num_threads)
1250                 /*
1251                  * The compiled patterns on the main path are only
1252                  * used when not using threading. Otherwise
1253                  * start_threads() below calls compile_grep_patterns()
1254                  * for each thread.
1255                  */
1256                 compile_grep_patterns(&opt);
1257
1258 #ifndef NO_PTHREADS
1259         if (num_threads) {
1260                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1261                     && (opt.pre_context || opt.post_context ||
1262                         opt.file_break || opt.funcbody))
1263                         skip_first_line = 1;
1264                 start_threads(&opt);
1265         }
1266 #endif
1267
1268         if (recurse_submodules) {
1269                 gitmodules_config();
1270                 compile_submodule_options(&opt, argv + i, cached, untracked,
1271                                           opt_exclude, use_index,
1272                                           pattern_type_arg);
1273         }
1274
1275         if (show_in_pager && (cached || list.nr))
1276                 die(_("--open-files-in-pager only works on the worktree"));
1277
1278         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1279                 const char *pager = path_list.items[0].string;
1280                 int len = strlen(pager);
1281
1282                 if (len > 4 && is_dir_sep(pager[len - 5]))
1283                         pager += len - 4;
1284
1285                 if (opt.ignore_case && !strcmp("less", pager))
1286                         string_list_append(&path_list, "-I");
1287
1288                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1289                         struct strbuf buf = STRBUF_INIT;
1290                         strbuf_addf(&buf, "+/%s%s",
1291                                         strcmp("less", pager) ? "" : "*",
1292                                         opt.pattern_list->pattern);
1293                         string_list_append(&path_list, buf.buf);
1294                         strbuf_detach(&buf, NULL);
1295                 }
1296         }
1297
1298         if (recurse_submodules && (!use_index || untracked))
1299                 die(_("option not supported with --recurse-submodules."));
1300
1301         if (!show_in_pager && !opt.status_only)
1302                 setup_pager();
1303
1304         if (!use_index && (untracked || cached))
1305                 die(_("--cached or --untracked cannot be used with --no-index."));
1306
1307         if (!use_index || untracked) {
1308                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1309                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1310         } else if (0 <= opt_exclude) {
1311                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1312         } else if (!list.nr) {
1313                 if (!cached)
1314                         setup_work_tree();
1315
1316                 hit = grep_cache(&opt, &pathspec, cached);
1317         } else {
1318                 if (cached)
1319                         die(_("both --cached and trees are given."));
1320                 hit = grep_objects(&opt, &pathspec, &list);
1321         }
1322
1323         if (num_threads)
1324                 hit |= wait_all();
1325         if (hit && show_in_pager)
1326                 run_pager(&opt, prefix);
1327         clear_pathspec(&pathspec);
1328         free_grep_patterns(&opt);
1329         return !hit;
1330 }