Merge branch 'mf/no-dashed-subcommands' into maint
[git] / builtin / grep.c
1 /*
2  * Builtin "git grep"
3  *
4  * Copyright (c) 2006 Junio C Hamano
5  */
6 #include "cache.h"
7 #include "repository.h"
8 #include "config.h"
9 #include "blob.h"
10 #include "tree.h"
11 #include "commit.h"
12 #include "tag.h"
13 #include "tree-walk.h"
14 #include "builtin.h"
15 #include "parse-options.h"
16 #include "string-list.h"
17 #include "run-command.h"
18 #include "userdiff.h"
19 #include "grep.h"
20 #include "quote.h"
21 #include "dir.h"
22 #include "pathspec.h"
23 #include "submodule.h"
24 #include "submodule-config.h"
25
26 static char const * const grep_usage[] = {
27         N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"),
28         NULL
29 };
30
31 static const char *super_prefix;
32 static int recurse_submodules;
33 static struct argv_array submodule_options = ARGV_ARRAY_INIT;
34 static const char *parent_basename;
35
36 static int grep_submodule_launch(struct grep_opt *opt,
37                                  const struct grep_source *gs);
38
39 #define GREP_NUM_THREADS_DEFAULT 8
40 static int num_threads;
41
42 #ifndef NO_PTHREADS
43 static pthread_t *threads;
44
45 /* We use one producer thread and THREADS consumer
46  * threads. The producer adds struct work_items to 'todo' and the
47  * consumers pick work items from the same array.
48  */
49 struct work_item {
50         struct grep_source source;
51         char done;
52         struct strbuf out;
53 };
54
55 /* In the range [todo_done, todo_start) in 'todo' we have work_items
56  * that have been or are processed by a consumer thread. We haven't
57  * written the result for these to stdout yet.
58  *
59  * The work_items in [todo_start, todo_end) are waiting to be picked
60  * up by a consumer thread.
61  *
62  * The ranges are modulo TODO_SIZE.
63  */
64 #define TODO_SIZE 128
65 static struct work_item todo[TODO_SIZE];
66 static int todo_start;
67 static int todo_end;
68 static int todo_done;
69
70 /* Has all work items been added? */
71 static int all_work_added;
72
73 /* This lock protects all the variables above. */
74 static pthread_mutex_t grep_mutex;
75
76 static inline void grep_lock(void)
77 {
78         assert(num_threads);
79         pthread_mutex_lock(&grep_mutex);
80 }
81
82 static inline void grep_unlock(void)
83 {
84         assert(num_threads);
85         pthread_mutex_unlock(&grep_mutex);
86 }
87
88 /* Signalled when a new work_item is added to todo. */
89 static pthread_cond_t cond_add;
90
91 /* Signalled when the result from one work_item is written to
92  * stdout.
93  */
94 static pthread_cond_t cond_write;
95
96 /* Signalled when we are finished with everything. */
97 static pthread_cond_t cond_result;
98
99 static int skip_first_line;
100
101 static void add_work(struct grep_opt *opt, enum grep_source_type type,
102                      const char *name, const char *path, const void *id)
103 {
104         grep_lock();
105
106         while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
107                 pthread_cond_wait(&cond_write, &grep_mutex);
108         }
109
110         grep_source_init(&todo[todo_end].source, type, name, path, id);
111         if (opt->binary != GREP_BINARY_TEXT)
112                 grep_source_load_driver(&todo[todo_end].source);
113         todo[todo_end].done = 0;
114         strbuf_reset(&todo[todo_end].out);
115         todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
116
117         pthread_cond_signal(&cond_add);
118         grep_unlock();
119 }
120
121 static struct work_item *get_work(void)
122 {
123         struct work_item *ret;
124
125         grep_lock();
126         while (todo_start == todo_end && !all_work_added) {
127                 pthread_cond_wait(&cond_add, &grep_mutex);
128         }
129
130         if (todo_start == todo_end && all_work_added) {
131                 ret = NULL;
132         } else {
133                 ret = &todo[todo_start];
134                 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
135         }
136         grep_unlock();
137         return ret;
138 }
139
140 static void work_done(struct work_item *w)
141 {
142         int old_done;
143
144         grep_lock();
145         w->done = 1;
146         old_done = todo_done;
147         for(; todo[todo_done].done && todo_done != todo_start;
148             todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
149                 w = &todo[todo_done];
150                 if (w->out.len) {
151                         const char *p = w->out.buf;
152                         size_t len = w->out.len;
153
154                         /* Skip the leading hunk mark of the first file. */
155                         if (skip_first_line) {
156                                 while (len) {
157                                         len--;
158                                         if (*p++ == '\n')
159                                                 break;
160                                 }
161                                 skip_first_line = 0;
162                         }
163
164                         write_or_die(1, p, len);
165                 }
166                 grep_source_clear(&w->source);
167         }
168
169         if (old_done != todo_done)
170                 pthread_cond_signal(&cond_write);
171
172         if (all_work_added && todo_done == todo_end)
173                 pthread_cond_signal(&cond_result);
174
175         grep_unlock();
176 }
177
178 static void *run(void *arg)
179 {
180         int hit = 0;
181         struct grep_opt *opt = arg;
182
183         while (1) {
184                 struct work_item *w = get_work();
185                 if (!w)
186                         break;
187
188                 opt->output_priv = w;
189                 if (w->source.type == GREP_SOURCE_SUBMODULE)
190                         hit |= grep_submodule_launch(opt, &w->source);
191                 else
192                         hit |= grep_source(opt, &w->source);
193                 grep_source_clear_data(&w->source);
194                 work_done(w);
195         }
196         free_grep_patterns(arg);
197         free(arg);
198
199         return (void*) (intptr_t) hit;
200 }
201
202 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
203 {
204         struct work_item *w = opt->output_priv;
205         strbuf_add(&w->out, buf, size);
206 }
207
208 static void start_threads(struct grep_opt *opt)
209 {
210         int i;
211
212         pthread_mutex_init(&grep_mutex, NULL);
213         pthread_mutex_init(&grep_read_mutex, NULL);
214         pthread_mutex_init(&grep_attr_mutex, NULL);
215         pthread_cond_init(&cond_add, NULL);
216         pthread_cond_init(&cond_write, NULL);
217         pthread_cond_init(&cond_result, NULL);
218         grep_use_locks = 1;
219
220         for (i = 0; i < ARRAY_SIZE(todo); i++) {
221                 strbuf_init(&todo[i].out, 0);
222         }
223
224         threads = xcalloc(num_threads, sizeof(*threads));
225         for (i = 0; i < num_threads; i++) {
226                 int err;
227                 struct grep_opt *o = grep_opt_dup(opt);
228                 o->output = strbuf_out;
229                 if (i)
230                         o->debug = 0;
231                 compile_grep_patterns(o);
232                 err = pthread_create(&threads[i], NULL, run, o);
233
234                 if (err)
235                         die(_("grep: failed to create thread: %s"),
236                             strerror(err));
237         }
238 }
239
240 static int wait_all(void)
241 {
242         int hit = 0;
243         int i;
244
245         grep_lock();
246         all_work_added = 1;
247
248         /* Wait until all work is done. */
249         while (todo_done != todo_end)
250                 pthread_cond_wait(&cond_result, &grep_mutex);
251
252         /* Wake up all the consumer threads so they can see that there
253          * is no more work to do.
254          */
255         pthread_cond_broadcast(&cond_add);
256         grep_unlock();
257
258         for (i = 0; i < num_threads; i++) {
259                 void *h;
260                 pthread_join(threads[i], &h);
261                 hit |= (int) (intptr_t) h;
262         }
263
264         free(threads);
265
266         pthread_mutex_destroy(&grep_mutex);
267         pthread_mutex_destroy(&grep_read_mutex);
268         pthread_mutex_destroy(&grep_attr_mutex);
269         pthread_cond_destroy(&cond_add);
270         pthread_cond_destroy(&cond_write);
271         pthread_cond_destroy(&cond_result);
272         grep_use_locks = 0;
273
274         return hit;
275 }
276 #else /* !NO_PTHREADS */
277
278 static int wait_all(void)
279 {
280         return 0;
281 }
282 #endif
283
284 static int grep_cmd_config(const char *var, const char *value, void *cb)
285 {
286         int st = grep_config(var, value, cb);
287         if (git_default_config(var, value, cb) < 0)
288                 st = -1;
289
290         if (!strcmp(var, "grep.threads")) {
291                 num_threads = git_config_int(var, value);
292                 if (num_threads < 0)
293                         die(_("invalid number of threads specified (%d) for %s"),
294                             num_threads, var);
295 #ifdef NO_PTHREADS
296                 else if (num_threads && num_threads != 1) {
297                         /*
298                          * TRANSLATORS: %s is the configuration
299                          * variable for tweaking threads, currently
300                          * grep.threads
301                          */
302                         warning(_("no threads support, ignoring %s"), var);
303                         num_threads = 0;
304                 }
305 #endif
306         }
307
308         if (!strcmp(var, "submodule.recurse"))
309                 recurse_submodules = git_config_bool(var, value);
310
311         return st;
312 }
313
314 static void *lock_and_read_oid_file(const struct object_id *oid, enum object_type *type, unsigned long *size)
315 {
316         void *data;
317
318         grep_read_lock();
319         data = read_sha1_file(oid->hash, type, size);
320         grep_read_unlock();
321         return data;
322 }
323
324 static int grep_oid(struct grep_opt *opt, const struct object_id *oid,
325                      const char *filename, int tree_name_len,
326                      const char *path)
327 {
328         struct strbuf pathbuf = STRBUF_INIT;
329
330         if (super_prefix) {
331                 strbuf_add(&pathbuf, filename, tree_name_len);
332                 strbuf_addstr(&pathbuf, super_prefix);
333                 strbuf_addstr(&pathbuf, filename + tree_name_len);
334         } else {
335                 strbuf_addstr(&pathbuf, filename);
336         }
337
338         if (opt->relative && opt->prefix_length) {
339                 char *name = strbuf_detach(&pathbuf, NULL);
340                 quote_path_relative(name + tree_name_len, opt->prefix, &pathbuf);
341                 strbuf_insert(&pathbuf, 0, name, tree_name_len);
342                 free(name);
343         }
344
345 #ifndef NO_PTHREADS
346         if (num_threads) {
347                 add_work(opt, GREP_SOURCE_OID, pathbuf.buf, path, oid);
348                 strbuf_release(&pathbuf);
349                 return 0;
350         } else
351 #endif
352         {
353                 struct grep_source gs;
354                 int hit;
355
356                 grep_source_init(&gs, GREP_SOURCE_OID, pathbuf.buf, path, oid);
357                 strbuf_release(&pathbuf);
358                 hit = grep_source(opt, &gs);
359
360                 grep_source_clear(&gs);
361                 return hit;
362         }
363 }
364
365 static int grep_file(struct grep_opt *opt, const char *filename)
366 {
367         struct strbuf buf = STRBUF_INIT;
368
369         if (super_prefix)
370                 strbuf_addstr(&buf, super_prefix);
371         strbuf_addstr(&buf, filename);
372
373         if (opt->relative && opt->prefix_length) {
374                 char *name = strbuf_detach(&buf, NULL);
375                 quote_path_relative(name, opt->prefix, &buf);
376                 free(name);
377         }
378
379 #ifndef NO_PTHREADS
380         if (num_threads) {
381                 add_work(opt, GREP_SOURCE_FILE, buf.buf, filename, filename);
382                 strbuf_release(&buf);
383                 return 0;
384         } else
385 #endif
386         {
387                 struct grep_source gs;
388                 int hit;
389
390                 grep_source_init(&gs, GREP_SOURCE_FILE, buf.buf, filename, filename);
391                 strbuf_release(&buf);
392                 hit = grep_source(opt, &gs);
393
394                 grep_source_clear(&gs);
395                 return hit;
396         }
397 }
398
399 static void append_path(struct grep_opt *opt, const void *data, size_t len)
400 {
401         struct string_list *path_list = opt->output_priv;
402
403         if (len == 1 && *(const char *)data == '\0')
404                 return;
405         string_list_append(path_list, xstrndup(data, len));
406 }
407
408 static void run_pager(struct grep_opt *opt, const char *prefix)
409 {
410         struct string_list *path_list = opt->output_priv;
411         struct child_process child = CHILD_PROCESS_INIT;
412         int i, status;
413
414         for (i = 0; i < path_list->nr; i++)
415                 argv_array_push(&child.args, path_list->items[i].string);
416         child.dir = prefix;
417         child.use_shell = 1;
418
419         status = run_command(&child);
420         if (status)
421                 exit(status);
422 }
423
424 static void compile_submodule_options(const struct grep_opt *opt,
425                                       const char **argv,
426                                       int cached, int untracked,
427                                       int opt_exclude, int use_index,
428                                       int pattern_type_arg)
429 {
430         struct grep_pat *pattern;
431
432         if (recurse_submodules)
433                 argv_array_push(&submodule_options, "--recurse-submodules");
434
435         if (cached)
436                 argv_array_push(&submodule_options, "--cached");
437         if (!use_index)
438                 argv_array_push(&submodule_options, "--no-index");
439         if (untracked)
440                 argv_array_push(&submodule_options, "--untracked");
441         if (opt_exclude > 0)
442                 argv_array_push(&submodule_options, "--exclude-standard");
443
444         if (opt->invert)
445                 argv_array_push(&submodule_options, "-v");
446         if (opt->ignore_case)
447                 argv_array_push(&submodule_options, "-i");
448         if (opt->word_regexp)
449                 argv_array_push(&submodule_options, "-w");
450         switch (opt->binary) {
451         case GREP_BINARY_NOMATCH:
452                 argv_array_push(&submodule_options, "-I");
453                 break;
454         case GREP_BINARY_TEXT:
455                 argv_array_push(&submodule_options, "-a");
456                 break;
457         default:
458                 break;
459         }
460         if (opt->allow_textconv)
461                 argv_array_push(&submodule_options, "--textconv");
462         if (opt->max_depth != -1)
463                 argv_array_pushf(&submodule_options, "--max-depth=%d",
464                                  opt->max_depth);
465         if (opt->linenum)
466                 argv_array_push(&submodule_options, "-n");
467         if (!opt->pathname)
468                 argv_array_push(&submodule_options, "-h");
469         if (!opt->relative)
470                 argv_array_push(&submodule_options, "--full-name");
471         if (opt->name_only)
472                 argv_array_push(&submodule_options, "-l");
473         if (opt->unmatch_name_only)
474                 argv_array_push(&submodule_options, "-L");
475         if (opt->null_following_name)
476                 argv_array_push(&submodule_options, "-z");
477         if (opt->count)
478                 argv_array_push(&submodule_options, "-c");
479         if (opt->file_break)
480                 argv_array_push(&submodule_options, "--break");
481         if (opt->heading)
482                 argv_array_push(&submodule_options, "--heading");
483         if (opt->pre_context)
484                 argv_array_pushf(&submodule_options, "--before-context=%d",
485                                  opt->pre_context);
486         if (opt->post_context)
487                 argv_array_pushf(&submodule_options, "--after-context=%d",
488                                  opt->post_context);
489         if (opt->funcname)
490                 argv_array_push(&submodule_options, "-p");
491         if (opt->funcbody)
492                 argv_array_push(&submodule_options, "-W");
493         if (opt->all_match)
494                 argv_array_push(&submodule_options, "--all-match");
495         if (opt->debug)
496                 argv_array_push(&submodule_options, "--debug");
497         if (opt->status_only)
498                 argv_array_push(&submodule_options, "-q");
499
500         switch (pattern_type_arg) {
501         case GREP_PATTERN_TYPE_BRE:
502                 argv_array_push(&submodule_options, "-G");
503                 break;
504         case GREP_PATTERN_TYPE_ERE:
505                 argv_array_push(&submodule_options, "-E");
506                 break;
507         case GREP_PATTERN_TYPE_FIXED:
508                 argv_array_push(&submodule_options, "-F");
509                 break;
510         case GREP_PATTERN_TYPE_PCRE:
511                 argv_array_push(&submodule_options, "-P");
512                 break;
513         case GREP_PATTERN_TYPE_UNSPECIFIED:
514                 break;
515         default:
516                 die("BUG: Added a new grep pattern type without updating switch statement");
517         }
518
519         for (pattern = opt->pattern_list; pattern != NULL;
520              pattern = pattern->next) {
521                 switch (pattern->token) {
522                 case GREP_PATTERN:
523                         argv_array_pushf(&submodule_options, "-e%s",
524                                          pattern->pattern);
525                         break;
526                 case GREP_AND:
527                 case GREP_OPEN_PAREN:
528                 case GREP_CLOSE_PAREN:
529                 case GREP_NOT:
530                 case GREP_OR:
531                         argv_array_push(&submodule_options, pattern->pattern);
532                         break;
533                 /* BODY and HEAD are not used by git-grep */
534                 case GREP_PATTERN_BODY:
535                 case GREP_PATTERN_HEAD:
536                         break;
537                 }
538         }
539
540         /*
541          * Limit number of threads for child process to use.
542          * This is to prevent potential fork-bomb behavior of git-grep as each
543          * submodule process has its own thread pool.
544          */
545         argv_array_pushf(&submodule_options, "--threads=%d",
546                          DIV_ROUND_UP(num_threads, 2));
547
548         /* Add Pathspecs */
549         argv_array_push(&submodule_options, "--");
550         for (; *argv; argv++)
551                 argv_array_push(&submodule_options, *argv);
552 }
553
554 /*
555  * Launch child process to grep contents of a submodule
556  */
557 static int grep_submodule_launch(struct grep_opt *opt,
558                                  const struct grep_source *gs)
559 {
560         struct child_process cp = CHILD_PROCESS_INIT;
561         int status, i;
562         const char *end_of_base;
563         const char *name;
564         struct strbuf child_output = STRBUF_INIT;
565
566         end_of_base = strchr(gs->name, ':');
567         if (gs->identifier && end_of_base)
568                 name = end_of_base + 1;
569         else
570                 name = gs->name;
571
572         prepare_submodule_repo_env(&cp.env_array);
573         argv_array_push(&cp.env_array, GIT_DIR_ENVIRONMENT);
574
575         if (opt->relative && opt->prefix_length)
576                 argv_array_pushf(&cp.env_array, "%s=%s",
577                                  GIT_TOPLEVEL_PREFIX_ENVIRONMENT,
578                                  opt->prefix);
579
580         /* Add super prefix */
581         argv_array_pushf(&cp.args, "--super-prefix=%s%s/",
582                          super_prefix ? super_prefix : "",
583                          name);
584         argv_array_push(&cp.args, "grep");
585
586         /*
587          * Add basename of parent project
588          * When performing grep on a tree object the filename is prefixed
589          * with the object's name: 'tree-name:filename'.  In order to
590          * provide uniformity of output we want to pass the name of the
591          * parent project's object name to the submodule so the submodule can
592          * prefix its output with the parent's name and not its own OID.
593          */
594         if (gs->identifier && end_of_base)
595                 argv_array_pushf(&cp.args, "--parent-basename=%.*s",
596                                  (int) (end_of_base - gs->name),
597                                  gs->name);
598
599         /* Add options */
600         for (i = 0; i < submodule_options.argc; i++) {
601                 /*
602                  * If there is a tree identifier for the submodule, add the
603                  * rev after adding the submodule options but before the
604                  * pathspecs.  To do this we listen for the '--' and insert the
605                  * oid before pushing the '--' onto the child process argv
606                  * array.
607                  */
608                 if (gs->identifier &&
609                     !strcmp("--", submodule_options.argv[i])) {
610                         argv_array_push(&cp.args, oid_to_hex(gs->identifier));
611                 }
612
613                 argv_array_push(&cp.args, submodule_options.argv[i]);
614         }
615
616         cp.git_cmd = 1;
617         cp.dir = gs->path;
618
619         /*
620          * Capture output to output buffer and check the return code from the
621          * child process.  A '0' indicates a hit, a '1' indicates no hit and
622          * anything else is an error.
623          */
624         status = capture_command(&cp, &child_output, 0);
625         if (status && (status != 1)) {
626                 /* flush the buffer */
627                 write_or_die(1, child_output.buf, child_output.len);
628                 die("process for submodule '%s' failed with exit code: %d",
629                     gs->name, status);
630         }
631
632         opt->output(opt, child_output.buf, child_output.len);
633         strbuf_release(&child_output);
634         /* invert the return code to make a hit equal to 1 */
635         return !status;
636 }
637
638 /*
639  * Prep grep structures for a submodule grep
640  * oid: the oid of the submodule or NULL if using the working tree
641  * filename: name of the submodule including tree name of parent
642  * path: location of the submodule
643  */
644 static int grep_submodule(struct grep_opt *opt, const struct object_id *oid,
645                           const char *filename, const char *path)
646 {
647         if (!is_submodule_active(the_repository, path))
648                 return 0;
649         if (!is_submodule_populated_gently(path, NULL)) {
650                 /*
651                  * If searching history, check for the presence of the
652                  * submodule's gitdir before skipping the submodule.
653                  */
654                 if (oid) {
655                         const struct submodule *sub =
656                                         submodule_from_path(null_sha1, path);
657                         if (sub)
658                                 path = git_path("modules/%s", sub->name);
659
660                         if (!(is_directory(path) && is_git_directory(path)))
661                                 return 0;
662                 } else {
663                         return 0;
664                 }
665         }
666
667 #ifndef NO_PTHREADS
668         if (num_threads) {
669                 add_work(opt, GREP_SOURCE_SUBMODULE, filename, path, oid);
670                 return 0;
671         } else
672 #endif
673         {
674                 struct grep_source gs;
675                 int hit;
676
677                 grep_source_init(&gs, GREP_SOURCE_SUBMODULE,
678                                  filename, path, oid);
679                 hit = grep_submodule_launch(opt, &gs);
680
681                 grep_source_clear(&gs);
682                 return hit;
683         }
684 }
685
686 static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec,
687                       int cached)
688 {
689         int hit = 0;
690         int nr;
691         struct strbuf name = STRBUF_INIT;
692         int name_base_len = 0;
693         if (super_prefix) {
694                 name_base_len = strlen(super_prefix);
695                 strbuf_addstr(&name, super_prefix);
696         }
697
698         read_cache();
699
700         for (nr = 0; nr < active_nr; nr++) {
701                 const struct cache_entry *ce = active_cache[nr];
702                 strbuf_setlen(&name, name_base_len);
703                 strbuf_addstr(&name, ce->name);
704
705                 if (S_ISREG(ce->ce_mode) &&
706                     match_pathspec(pathspec, name.buf, name.len, 0, NULL,
707                                    S_ISDIR(ce->ce_mode) ||
708                                    S_ISGITLINK(ce->ce_mode))) {
709                         /*
710                          * If CE_VALID is on, we assume worktree file and its
711                          * cache entry are identical, even if worktree file has
712                          * been modified, so use cache version instead
713                          */
714                         if (cached || (ce->ce_flags & CE_VALID) ||
715                             ce_skip_worktree(ce)) {
716                                 if (ce_stage(ce) || ce_intent_to_add(ce))
717                                         continue;
718                                 hit |= grep_oid(opt, &ce->oid, ce->name,
719                                                  0, ce->name);
720                         } else {
721                                 hit |= grep_file(opt, ce->name);
722                         }
723                 } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) &&
724                            submodule_path_match(pathspec, name.buf, NULL)) {
725                         hit |= grep_submodule(opt, NULL, ce->name, ce->name);
726                 } else {
727                         continue;
728                 }
729
730                 if (ce_stage(ce)) {
731                         do {
732                                 nr++;
733                         } while (nr < active_nr &&
734                                  !strcmp(ce->name, active_cache[nr]->name));
735                         nr--; /* compensate for loop control */
736                 }
737                 if (hit && opt->status_only)
738                         break;
739         }
740
741         strbuf_release(&name);
742         return hit;
743 }
744
745 static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
746                      struct tree_desc *tree, struct strbuf *base, int tn_len,
747                      int check_attr)
748 {
749         int hit = 0;
750         enum interesting match = entry_not_interesting;
751         struct name_entry entry;
752         int old_baselen = base->len;
753         struct strbuf name = STRBUF_INIT;
754         int name_base_len = 0;
755         if (super_prefix) {
756                 strbuf_addstr(&name, super_prefix);
757                 name_base_len = name.len;
758         }
759
760         while (tree_entry(tree, &entry)) {
761                 int te_len = tree_entry_len(&entry);
762
763                 if (match != all_entries_interesting) {
764                         strbuf_addstr(&name, base->buf + tn_len);
765                         match = tree_entry_interesting(&entry, &name,
766                                                        0, pathspec);
767                         strbuf_setlen(&name, name_base_len);
768
769                         if (match == all_entries_not_interesting)
770                                 break;
771                         if (match == entry_not_interesting)
772                                 continue;
773                 }
774
775                 strbuf_add(base, entry.path, te_len);
776
777                 if (S_ISREG(entry.mode)) {
778                         hit |= grep_oid(opt, entry.oid, base->buf, tn_len,
779                                          check_attr ? base->buf + tn_len : NULL);
780                 } else if (S_ISDIR(entry.mode)) {
781                         enum object_type type;
782                         struct tree_desc sub;
783                         void *data;
784                         unsigned long size;
785
786                         data = lock_and_read_oid_file(entry.oid, &type, &size);
787                         if (!data)
788                                 die(_("unable to read tree (%s)"),
789                                     oid_to_hex(entry.oid));
790
791                         strbuf_addch(base, '/');
792                         init_tree_desc(&sub, data, size);
793                         hit |= grep_tree(opt, pathspec, &sub, base, tn_len,
794                                          check_attr);
795                         free(data);
796                 } else if (recurse_submodules && S_ISGITLINK(entry.mode)) {
797                         hit |= grep_submodule(opt, entry.oid, base->buf,
798                                               base->buf + tn_len);
799                 }
800
801                 strbuf_setlen(base, old_baselen);
802
803                 if (hit && opt->status_only)
804                         break;
805         }
806
807         strbuf_release(&name);
808         return hit;
809 }
810
811 static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
812                        struct object *obj, const char *name, const char *path)
813 {
814         if (obj->type == OBJ_BLOB)
815                 return grep_oid(opt, &obj->oid, name, 0, path);
816         if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
817                 struct tree_desc tree;
818                 void *data;
819                 unsigned long size;
820                 struct strbuf base;
821                 int hit, len;
822
823                 grep_read_lock();
824                 data = read_object_with_reference(obj->oid.hash, tree_type,
825                                                   &size, NULL);
826                 grep_read_unlock();
827
828                 if (!data)
829                         die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid));
830
831                 /* Use parent's name as base when recursing submodules */
832                 if (recurse_submodules && parent_basename)
833                         name = parent_basename;
834
835                 len = name ? strlen(name) : 0;
836                 strbuf_init(&base, PATH_MAX + len + 1);
837                 if (len) {
838                         strbuf_add(&base, name, len);
839                         strbuf_addch(&base, ':');
840                 }
841                 init_tree_desc(&tree, data, size);
842                 hit = grep_tree(opt, pathspec, &tree, &base, base.len,
843                                 obj->type == OBJ_COMMIT);
844                 strbuf_release(&base);
845                 free(data);
846                 return hit;
847         }
848         die(_("unable to grep from object of type %s"), typename(obj->type));
849 }
850
851 static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
852                         const struct object_array *list)
853 {
854         unsigned int i;
855         int hit = 0;
856         const unsigned int nr = list->nr;
857
858         for (i = 0; i < nr; i++) {
859                 struct object *real_obj;
860                 real_obj = deref_tag(list->objects[i].item, NULL, 0);
861
862                 /* load the gitmodules file for this rev */
863                 if (recurse_submodules) {
864                         submodule_free();
865                         gitmodules_config_sha1(real_obj->oid.hash);
866                 }
867                 if (grep_object(opt, pathspec, real_obj, list->objects[i].name, list->objects[i].path)) {
868                         hit = 1;
869                         if (opt->status_only)
870                                 break;
871                 }
872         }
873         return hit;
874 }
875
876 static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
877                           int exc_std, int use_index)
878 {
879         struct dir_struct dir;
880         int i, hit = 0;
881
882         memset(&dir, 0, sizeof(dir));
883         if (!use_index)
884                 dir.flags |= DIR_NO_GITLINKS;
885         if (exc_std)
886                 setup_standard_excludes(&dir);
887
888         fill_directory(&dir, &the_index, pathspec);
889         for (i = 0; i < dir.nr; i++) {
890                 if (!dir_path_match(dir.entries[i], pathspec, 0, NULL))
891                         continue;
892                 hit |= grep_file(opt, dir.entries[i]->name);
893                 if (hit && opt->status_only)
894                         break;
895         }
896         return hit;
897 }
898
899 static int context_callback(const struct option *opt, const char *arg,
900                             int unset)
901 {
902         struct grep_opt *grep_opt = opt->value;
903         int value;
904         const char *endp;
905
906         if (unset) {
907                 grep_opt->pre_context = grep_opt->post_context = 0;
908                 return 0;
909         }
910         value = strtol(arg, (char **)&endp, 10);
911         if (*endp) {
912                 return error(_("switch `%c' expects a numerical value"),
913                              opt->short_name);
914         }
915         grep_opt->pre_context = grep_opt->post_context = value;
916         return 0;
917 }
918
919 static int file_callback(const struct option *opt, const char *arg, int unset)
920 {
921         struct grep_opt *grep_opt = opt->value;
922         int from_stdin = !strcmp(arg, "-");
923         FILE *patterns;
924         int lno = 0;
925         struct strbuf sb = STRBUF_INIT;
926
927         patterns = from_stdin ? stdin : fopen(arg, "r");
928         if (!patterns)
929                 die_errno(_("cannot open '%s'"), arg);
930         while (strbuf_getline(&sb, patterns) == 0) {
931                 /* ignore empty line like grep does */
932                 if (sb.len == 0)
933                         continue;
934
935                 append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno,
936                                 GREP_PATTERN);
937         }
938         if (!from_stdin)
939                 fclose(patterns);
940         strbuf_release(&sb);
941         return 0;
942 }
943
944 static int not_callback(const struct option *opt, const char *arg, int unset)
945 {
946         struct grep_opt *grep_opt = opt->value;
947         append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
948         return 0;
949 }
950
951 static int and_callback(const struct option *opt, const char *arg, int unset)
952 {
953         struct grep_opt *grep_opt = opt->value;
954         append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
955         return 0;
956 }
957
958 static int open_callback(const struct option *opt, const char *arg, int unset)
959 {
960         struct grep_opt *grep_opt = opt->value;
961         append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
962         return 0;
963 }
964
965 static int close_callback(const struct option *opt, const char *arg, int unset)
966 {
967         struct grep_opt *grep_opt = opt->value;
968         append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
969         return 0;
970 }
971
972 static int pattern_callback(const struct option *opt, const char *arg,
973                             int unset)
974 {
975         struct grep_opt *grep_opt = opt->value;
976         append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
977         return 0;
978 }
979
980 int cmd_grep(int argc, const char **argv, const char *prefix)
981 {
982         int hit = 0;
983         int cached = 0, untracked = 0, opt_exclude = -1;
984         int seen_dashdash = 0;
985         int external_grep_allowed__ignored;
986         const char *show_in_pager = NULL, *default_pager = "dummy";
987         struct grep_opt opt;
988         struct object_array list = OBJECT_ARRAY_INIT;
989         struct pathspec pathspec;
990         struct string_list path_list = STRING_LIST_INIT_NODUP;
991         int i;
992         int dummy;
993         int use_index = 1;
994         int pattern_type_arg = GREP_PATTERN_TYPE_UNSPECIFIED;
995         int allow_revs;
996
997         struct option options[] = {
998                 OPT_BOOL(0, "cached", &cached,
999                         N_("search in index instead of in the work tree")),
1000                 OPT_NEGBIT(0, "no-index", &use_index,
1001                          N_("find in contents not managed by git"), 1),
1002                 OPT_BOOL(0, "untracked", &untracked,
1003                         N_("search in both tracked and untracked files")),
1004                 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
1005                             N_("ignore files specified via '.gitignore'"), 1),
1006                 OPT_BOOL(0, "recurse-submodules", &recurse_submodules,
1007                          N_("recursively search in each submodule")),
1008                 OPT_STRING(0, "parent-basename", &parent_basename,
1009                            N_("basename"),
1010                            N_("prepend parent project's basename to output")),
1011                 OPT_GROUP(""),
1012                 OPT_BOOL('v', "invert-match", &opt.invert,
1013                         N_("show non-matching lines")),
1014                 OPT_BOOL('i', "ignore-case", &opt.ignore_case,
1015                         N_("case insensitive matching")),
1016                 OPT_BOOL('w', "word-regexp", &opt.word_regexp,
1017                         N_("match patterns only at word boundaries")),
1018                 OPT_SET_INT('a', "text", &opt.binary,
1019                         N_("process binary files as text"), GREP_BINARY_TEXT),
1020                 OPT_SET_INT('I', NULL, &opt.binary,
1021                         N_("don't match patterns in binary files"),
1022                         GREP_BINARY_NOMATCH),
1023                 OPT_BOOL(0, "textconv", &opt.allow_textconv,
1024                          N_("process binary files with textconv filters")),
1025                 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, N_("depth"),
1026                         N_("descend at most <depth> levels"), PARSE_OPT_NONEG,
1027                         NULL, 1 },
1028                 OPT_GROUP(""),
1029                 OPT_SET_INT('E', "extended-regexp", &pattern_type_arg,
1030                             N_("use extended POSIX regular expressions"),
1031                             GREP_PATTERN_TYPE_ERE),
1032                 OPT_SET_INT('G', "basic-regexp", &pattern_type_arg,
1033                             N_("use basic POSIX regular expressions (default)"),
1034                             GREP_PATTERN_TYPE_BRE),
1035                 OPT_SET_INT('F', "fixed-strings", &pattern_type_arg,
1036                             N_("interpret patterns as fixed strings"),
1037                             GREP_PATTERN_TYPE_FIXED),
1038                 OPT_SET_INT('P', "perl-regexp", &pattern_type_arg,
1039                             N_("use Perl-compatible regular expressions"),
1040                             GREP_PATTERN_TYPE_PCRE),
1041                 OPT_GROUP(""),
1042                 OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")),
1043                 OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1),
1044                 OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1),
1045                 OPT_NEGBIT(0, "full-name", &opt.relative,
1046                         N_("show filenames relative to top directory"), 1),
1047                 OPT_BOOL('l', "files-with-matches", &opt.name_only,
1048                         N_("show only filenames instead of matching lines")),
1049                 OPT_BOOL(0, "name-only", &opt.name_only,
1050                         N_("synonym for --files-with-matches")),
1051                 OPT_BOOL('L', "files-without-match",
1052                         &opt.unmatch_name_only,
1053                         N_("show only the names of files without match")),
1054                 OPT_BOOL('z', "null", &opt.null_following_name,
1055                         N_("print NUL after filenames")),
1056                 OPT_BOOL('c', "count", &opt.count,
1057                         N_("show the number of matches instead of matching lines")),
1058                 OPT__COLOR(&opt.color, N_("highlight matches")),
1059                 OPT_BOOL(0, "break", &opt.file_break,
1060                         N_("print empty line between matches from different files")),
1061                 OPT_BOOL(0, "heading", &opt.heading,
1062                         N_("show filename only once above matches from same file")),
1063                 OPT_GROUP(""),
1064                 OPT_CALLBACK('C', "context", &opt, N_("n"),
1065                         N_("show <n> context lines before and after matches"),
1066                         context_callback),
1067                 OPT_INTEGER('B', "before-context", &opt.pre_context,
1068                         N_("show <n> context lines before matches")),
1069                 OPT_INTEGER('A', "after-context", &opt.post_context,
1070                         N_("show <n> context lines after matches")),
1071                 OPT_INTEGER(0, "threads", &num_threads,
1072                         N_("use <n> worker threads")),
1073                 OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"),
1074                         context_callback),
1075                 OPT_BOOL('p', "show-function", &opt.funcname,
1076                         N_("show a line with the function name before matches")),
1077                 OPT_BOOL('W', "function-context", &opt.funcbody,
1078                         N_("show the surrounding function")),
1079                 OPT_GROUP(""),
1080                 OPT_CALLBACK('f', NULL, &opt, N_("file"),
1081                         N_("read patterns from file"), file_callback),
1082                 { OPTION_CALLBACK, 'e', NULL, &opt, N_("pattern"),
1083                         N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback },
1084                 { OPTION_CALLBACK, 0, "and", &opt, NULL,
1085                   N_("combine patterns specified with -e"),
1086                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
1087                 OPT_BOOL(0, "or", &dummy, ""),
1088                 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
1089                   PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
1090                 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
1091                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1092                   open_callback },
1093                 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
1094                   PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
1095                   close_callback },
1096                 OPT__QUIET(&opt.status_only,
1097                            N_("indicate hit with exit status without output")),
1098                 OPT_BOOL(0, "all-match", &opt.all_match,
1099                         N_("show only matches from files that match all patterns")),
1100                 { OPTION_SET_INT, 0, "debug", &opt.debug, NULL,
1101                   N_("show parse tree for grep expression"),
1102                   PARSE_OPT_NOARG | PARSE_OPT_HIDDEN, NULL, 1 },
1103                 OPT_GROUP(""),
1104                 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
1105                         N_("pager"), N_("show matching files in the pager"),
1106                         PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
1107                 OPT_BOOL(0, "ext-grep", &external_grep_allowed__ignored,
1108                          N_("allow calling of grep(1) (ignored by this build)")),
1109                 OPT_END()
1110         };
1111
1112         init_grep_defaults();
1113         git_config(grep_cmd_config, NULL);
1114         grep_init(&opt, prefix);
1115         super_prefix = get_super_prefix();
1116
1117         /*
1118          * If there is no -- then the paths must exist in the working
1119          * tree.  If there is no explicit pattern specified with -e or
1120          * -f, we take the first unrecognized non option to be the
1121          * pattern, but then what follows it must be zero or more
1122          * valid refs up to the -- (if exists), and then existing
1123          * paths.  If there is an explicit pattern, then the first
1124          * unrecognized non option is the beginning of the refs list
1125          * that continues up to the -- (if exists), and then paths.
1126          */
1127         argc = parse_options(argc, argv, prefix, options, grep_usage,
1128                              PARSE_OPT_KEEP_DASHDASH |
1129                              PARSE_OPT_STOP_AT_NON_OPTION);
1130         grep_commit_pattern_type(pattern_type_arg, &opt);
1131
1132         if (use_index && !startup_info->have_repository) {
1133                 int fallback = 0;
1134                 git_config_get_bool("grep.fallbacktonoindex", &fallback);
1135                 if (fallback)
1136                         use_index = 0;
1137                 else
1138                         /* die the same way as if we did it at the beginning */
1139                         setup_git_directory();
1140         }
1141
1142         /*
1143          * skip a -- separator; we know it cannot be
1144          * separating revisions from pathnames if
1145          * we haven't even had any patterns yet
1146          */
1147         if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
1148                 argv++;
1149                 argc--;
1150         }
1151
1152         /* First unrecognized non-option token */
1153         if (argc > 0 && !opt.pattern_list) {
1154                 append_grep_pattern(&opt, argv[0], "command line", 0,
1155                                     GREP_PATTERN);
1156                 argv++;
1157                 argc--;
1158         }
1159
1160         if (show_in_pager == default_pager)
1161                 show_in_pager = git_pager(1);
1162         if (show_in_pager) {
1163                 opt.color = 0;
1164                 opt.name_only = 1;
1165                 opt.null_following_name = 1;
1166                 opt.output_priv = &path_list;
1167                 opt.output = append_path;
1168                 string_list_append(&path_list, show_in_pager);
1169         }
1170
1171         if (!opt.pattern_list)
1172                 die(_("no pattern given."));
1173
1174         /*
1175          * We have to find "--" in a separate pass, because its presence
1176          * influences how we will parse arguments that come before it.
1177          */
1178         for (i = 0; i < argc; i++) {
1179                 if (!strcmp(argv[i], "--")) {
1180                         seen_dashdash = 1;
1181                         break;
1182                 }
1183         }
1184
1185         /*
1186          * Resolve any rev arguments. If we have a dashdash, then everything up
1187          * to it must resolve as a rev. If not, then we stop at the first
1188          * non-rev and assume everything else is a path.
1189          */
1190         allow_revs = use_index && !untracked;
1191         for (i = 0; i < argc; i++) {
1192                 const char *arg = argv[i];
1193                 struct object_id oid;
1194                 struct object_context oc;
1195                 struct object *object;
1196
1197                 if (!strcmp(arg, "--")) {
1198                         i++;
1199                         break;
1200                 }
1201
1202                 if (!allow_revs) {
1203                         if (seen_dashdash)
1204                                 die(_("--no-index or --untracked cannot be used with revs"));
1205                         break;
1206                 }
1207
1208                 if (get_sha1_with_context(arg, GET_SHA1_RECORD_PATH,
1209                                           oid.hash, &oc)) {
1210                         if (seen_dashdash)
1211                                 die(_("unable to resolve revision: %s"), arg);
1212                         break;
1213                 }
1214
1215                 object = parse_object_or_die(&oid, arg);
1216                 if (!seen_dashdash)
1217                         verify_non_filename(prefix, arg);
1218                 add_object_array_with_path(object, arg, &list, oc.mode, oc.path);
1219                 free(oc.path);
1220         }
1221
1222         /*
1223          * Anything left over is presumed to be a path. But in the non-dashdash
1224          * "do what I mean" case, we verify and complain when that isn't true.
1225          */
1226         if (!seen_dashdash) {
1227                 int j;
1228                 for (j = i; j < argc; j++)
1229                         verify_filename(prefix, argv[j], j == i && allow_revs);
1230         }
1231
1232         parse_pathspec(&pathspec, 0,
1233                        PATHSPEC_PREFER_CWD |
1234                        (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0),
1235                        prefix, argv + i);
1236         pathspec.max_depth = opt.max_depth;
1237         pathspec.recursive = 1;
1238
1239 #ifndef NO_PTHREADS
1240         if (list.nr || cached || show_in_pager)
1241                 num_threads = 0;
1242         else if (num_threads == 0)
1243                 num_threads = GREP_NUM_THREADS_DEFAULT;
1244         else if (num_threads < 0)
1245                 die(_("invalid number of threads specified (%d)"), num_threads);
1246         if (num_threads == 1)
1247                 num_threads = 0;
1248 #else
1249         if (num_threads)
1250                 warning(_("no threads support, ignoring --threads"));
1251         num_threads = 0;
1252 #endif
1253
1254         if (!num_threads)
1255                 /*
1256                  * The compiled patterns on the main path are only
1257                  * used when not using threading. Otherwise
1258                  * start_threads() below calls compile_grep_patterns()
1259                  * for each thread.
1260                  */
1261                 compile_grep_patterns(&opt);
1262
1263 #ifndef NO_PTHREADS
1264         if (num_threads) {
1265                 if (!(opt.name_only || opt.unmatch_name_only || opt.count)
1266                     && (opt.pre_context || opt.post_context ||
1267                         opt.file_break || opt.funcbody))
1268                         skip_first_line = 1;
1269                 start_threads(&opt);
1270         }
1271 #endif
1272
1273         if (recurse_submodules) {
1274                 gitmodules_config();
1275                 compile_submodule_options(&opt, argv + i, cached, untracked,
1276                                           opt_exclude, use_index,
1277                                           pattern_type_arg);
1278         }
1279
1280         if (show_in_pager && (cached || list.nr))
1281                 die(_("--open-files-in-pager only works on the worktree"));
1282
1283         if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1284                 const char *pager = path_list.items[0].string;
1285                 int len = strlen(pager);
1286
1287                 if (len > 4 && is_dir_sep(pager[len - 5]))
1288                         pager += len - 4;
1289
1290                 if (opt.ignore_case && !strcmp("less", pager))
1291                         string_list_append(&path_list, "-I");
1292
1293                 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1294                         struct strbuf buf = STRBUF_INIT;
1295                         strbuf_addf(&buf, "+/%s%s",
1296                                         strcmp("less", pager) ? "" : "*",
1297                                         opt.pattern_list->pattern);
1298                         string_list_append(&path_list, buf.buf);
1299                         strbuf_detach(&buf, NULL);
1300                 }
1301         }
1302
1303         if (recurse_submodules && (!use_index || untracked))
1304                 die(_("option not supported with --recurse-submodules."));
1305
1306         if (!show_in_pager && !opt.status_only)
1307                 setup_pager();
1308
1309         if (!use_index && (untracked || cached))
1310                 die(_("--cached or --untracked cannot be used with --no-index."));
1311
1312         if (!use_index || untracked) {
1313                 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
1314                 hit = grep_directory(&opt, &pathspec, use_exclude, use_index);
1315         } else if (0 <= opt_exclude) {
1316                 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
1317         } else if (!list.nr) {
1318                 if (!cached)
1319                         setup_work_tree();
1320
1321                 hit = grep_cache(&opt, &pathspec, cached);
1322         } else {
1323                 if (cached)
1324                         die(_("both --cached and trees are given."));
1325                 hit = grep_objects(&opt, &pathspec, &list);
1326         }
1327
1328         if (num_threads)
1329                 hit |= wait_all();
1330         if (hit && show_in_pager)
1331                 run_pager(&opt, prefix);
1332         clear_pathspec(&pathspec);
1333         free_grep_patterns(&opt);
1334         return !hit;
1335 }