grep: refactor the concept of "grep source" into an object
[git/git.git] / builtin / grep.c
CommitLineData
5010cb5f
JH
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"
1362671f 11#include "tree-walk.h"
5010cb5f 12#include "builtin.h"
3e230fa1 13#include "parse-options.h"
678e484b
JS
14#include "string-list.h"
15#include "run-command.h"
60ecac98 16#include "userdiff.h"
83b5d2f5 17#include "grep.h"
493b7a08 18#include "quote.h"
59332d13 19#include "dir.h"
5b594f45 20
3e230fa1 21static char const * const grep_usage[] = {
62b4698e 22 "git grep [options] [-e] <pattern> [<rev>...] [[--] <path>...]",
3e230fa1
RS
23 NULL
24};
25
5b594f45
FK
26static int use_threads = 1;
27
28#ifndef NO_PTHREADS
29#define THREADS 8
30static pthread_t threads[THREADS];
31
32static void *load_sha1(const unsigned char *sha1, unsigned long *size,
33 const char *name);
34static void *load_file(const char *filename, size_t *sz);
35
36enum work_type {WORK_SHA1, WORK_FILE};
37
38/* We use one producer thread and THREADS consumer
39 * threads. The producer adds struct work_items to 'todo' and the
40 * consumers pick work items from the same array.
41 */
9cba13ca 42struct work_item {
5b594f45
FK
43 enum work_type type;
44 char *name;
45
46 /* if type == WORK_SHA1, then 'identifier' is a SHA1,
47 * otherwise type == WORK_FILE, and 'identifier' is a NUL
48 * terminated filename.
49 */
50 void *identifier;
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
65static struct work_item todo[TODO_SIZE];
66static int todo_start;
67static int todo_end;
68static int todo_done;
69
70/* Has all work items been added? */
71static int all_work_added;
72
73/* This lock protects all the variables above. */
74static pthread_mutex_t grep_mutex;
75
1487a12b
JH
76static inline void grep_lock(void)
77{
78 if (use_threads)
79 pthread_mutex_lock(&grep_mutex);
80}
81
82static inline void grep_unlock(void)
83{
84 if (use_threads)
85 pthread_mutex_unlock(&grep_mutex);
86}
87
5b594f45
FK
88/* Signalled when a new work_item is added to todo. */
89static pthread_cond_t cond_add;
90
91/* Signalled when the result from one work_item is written to
92 * stdout.
93 */
94static pthread_cond_t cond_write;
95
96/* Signalled when we are finished with everything. */
97static pthread_cond_t cond_result;
98
08303c36 99static int skip_first_line;
431d6e7b 100
5b594f45
FK
101static void add_work(enum work_type type, char *name, void *id)
102{
103 grep_lock();
104
105 while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
106 pthread_cond_wait(&cond_write, &grep_mutex);
107 }
108
109 todo[todo_end].type = type;
110 todo[todo_end].name = name;
111 todo[todo_end].identifier = id;
112 todo[todo_end].done = 0;
113 strbuf_reset(&todo[todo_end].out);
114 todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
115
116 pthread_cond_signal(&cond_add);
117 grep_unlock();
118}
119
120static struct work_item *get_work(void)
121{
122 struct work_item *ret;
123
124 grep_lock();
125 while (todo_start == todo_end && !all_work_added) {
126 pthread_cond_wait(&cond_add, &grep_mutex);
127 }
128
129 if (todo_start == todo_end && all_work_added) {
130 ret = NULL;
131 } else {
132 ret = &todo[todo_start];
133 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
134 }
135 grep_unlock();
136 return ret;
137}
138
139static void grep_sha1_async(struct grep_opt *opt, char *name,
140 const unsigned char *sha1)
141{
142 unsigned char *s;
143 s = xmalloc(20);
144 memcpy(s, sha1, 20);
145 add_work(WORK_SHA1, name, s);
146}
147
148static void grep_file_async(struct grep_opt *opt, char *name,
149 const char *filename)
150{
151 add_work(WORK_FILE, name, xstrdup(filename));
152}
153
154static void work_done(struct work_item *w)
155{
156 int old_done;
157
158 grep_lock();
159 w->done = 1;
160 old_done = todo_done;
161 for(; todo[todo_done].done && todo_done != todo_start;
162 todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
163 w = &todo[todo_done];
431d6e7b 164 if (w->out.len) {
08303c36
RS
165 const char *p = w->out.buf;
166 size_t len = w->out.len;
167
168 /* Skip the leading hunk mark of the first file. */
169 if (skip_first_line) {
170 while (len) {
171 len--;
172 if (*p++ == '\n')
173 break;
174 }
175 skip_first_line = 0;
176 }
177
178 write_or_die(1, p, len);
431d6e7b 179 }
5b594f45
FK
180 free(w->name);
181 free(w->identifier);
182 }
183
184 if (old_done != todo_done)
185 pthread_cond_signal(&cond_write);
186
187 if (all_work_added && todo_done == todo_end)
188 pthread_cond_signal(&cond_result);
189
190 grep_unlock();
191}
192
193static void *run(void *arg)
194{
195 int hit = 0;
196 struct grep_opt *opt = arg;
197
198 while (1) {
199 struct work_item *w = get_work();
200 if (!w)
201 break;
202
203 opt->output_priv = w;
204 if (w->type == WORK_SHA1) {
205 unsigned long sz;
206 void* data = load_sha1(w->identifier, &sz, w->name);
207
208 if (data) {
209 hit |= grep_buffer(opt, w->name, data, sz);
210 free(data);
211 }
212 } else if (w->type == WORK_FILE) {
213 size_t sz;
214 void* data = load_file(w->identifier, &sz);
215 if (data) {
216 hit |= grep_buffer(opt, w->name, data, sz);
217 free(data);
218 }
219 } else {
220 assert(0);
221 }
222
223 work_done(w);
224 }
bfac23d9
DM
225 free_grep_patterns(arg);
226 free(arg);
5b594f45
FK
227
228 return (void*) (intptr_t) hit;
229}
230
231static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
232{
233 struct work_item *w = opt->output_priv;
234 strbuf_add(&w->out, buf, size);
235}
236
237static void start_threads(struct grep_opt *opt)
238{
239 int i;
240
241 pthread_mutex_init(&grep_mutex, NULL);
b3aeb285 242 pthread_mutex_init(&grep_read_mutex, NULL);
0579f91d 243 pthread_mutex_init(&grep_attr_mutex, NULL);
5b594f45
FK
244 pthread_cond_init(&cond_add, NULL);
245 pthread_cond_init(&cond_write, NULL);
246 pthread_cond_init(&cond_result, NULL);
78db6ea9 247 grep_use_locks = 1;
5b594f45
FK
248
249 for (i = 0; i < ARRAY_SIZE(todo); i++) {
250 strbuf_init(&todo[i].out, 0);
251 }
252
253 for (i = 0; i < ARRAY_SIZE(threads); i++) {
254 int err;
255 struct grep_opt *o = grep_opt_dup(opt);
256 o->output = strbuf_out;
257 compile_grep_patterns(o);
258 err = pthread_create(&threads[i], NULL, run, o);
259
260 if (err)
2fc5f9f1 261 die(_("grep: failed to create thread: %s"),
5b594f45
FK
262 strerror(err));
263 }
264}
265
266static int wait_all(void)
267{
268 int hit = 0;
269 int i;
270
271 grep_lock();
272 all_work_added = 1;
273
274 /* Wait until all work is done. */
275 while (todo_done != todo_end)
276 pthread_cond_wait(&cond_result, &grep_mutex);
277
278 /* Wake up all the consumer threads so they can see that there
279 * is no more work to do.
280 */
281 pthread_cond_broadcast(&cond_add);
282 grep_unlock();
283
284 for (i = 0; i < ARRAY_SIZE(threads); i++) {
285 void *h;
286 pthread_join(threads[i], &h);
287 hit |= (int) (intptr_t) h;
288 }
289
290 pthread_mutex_destroy(&grep_mutex);
b3aeb285 291 pthread_mutex_destroy(&grep_read_mutex);
0579f91d 292 pthread_mutex_destroy(&grep_attr_mutex);
5b594f45
FK
293 pthread_cond_destroy(&cond_add);
294 pthread_cond_destroy(&cond_write);
295 pthread_cond_destroy(&cond_result);
78db6ea9 296 grep_use_locks = 0;
5b594f45
FK
297
298 return hit;
299}
300#else /* !NO_PTHREADS */
5b594f45
FK
301
302static int wait_all(void)
303{
304 return 0;
305}
306#endif
307
7e8f59d5
RS
308static int grep_config(const char *var, const char *value, void *cb)
309{
310 struct grep_opt *opt = cb;
55f638bd 311 char *color = NULL;
7e8f59d5 312
60ecac98
RS
313 switch (userdiff_config(var, value)) {
314 case 0: break;
315 case -1: return -1;
316 default: return 0;
317 }
318
b22520a3
JR
319 if (!strcmp(var, "grep.extendedregexp")) {
320 if (git_config_bool(var, value))
321 opt->regflags |= REG_EXTENDED;
322 else
323 opt->regflags &= ~REG_EXTENDED;
324 return 0;
325 }
326
327 if (!strcmp(var, "grep.linenumber")) {
328 opt->linenum = git_config_bool(var, value);
329 return 0;
330 }
331
55f638bd 332 if (!strcmp(var, "color.grep"))
e269eb79 333 opt->color = git_config_colorbool(var, value);
00588bb5
ML
334 else if (!strcmp(var, "color.grep.context"))
335 color = opt->color_context;
55f638bd
ML
336 else if (!strcmp(var, "color.grep.filename"))
337 color = opt->color_filename;
00588bb5
ML
338 else if (!strcmp(var, "color.grep.function"))
339 color = opt->color_function;
55f638bd
ML
340 else if (!strcmp(var, "color.grep.linenumber"))
341 color = opt->color_lineno;
342 else if (!strcmp(var, "color.grep.match"))
343 color = opt->color_match;
00588bb5
ML
344 else if (!strcmp(var, "color.grep.selected"))
345 color = opt->color_selected;
55f638bd
ML
346 else if (!strcmp(var, "color.grep.separator"))
347 color = opt->color_sep;
348 else
349 return git_color_default_config(var, value, cb);
350 if (color) {
7e8f59d5
RS
351 if (!value)
352 return config_error_nonbool(var);
55f638bd 353 color_parse(value, var, color);
7e8f59d5 354 }
55f638bd 355 return 0;
7e8f59d5
RS
356}
357
5f02d315
JH
358static void *lock_and_read_sha1_file(const unsigned char *sha1, enum object_type *type, unsigned long *size)
359{
360 void *data;
361
b3aeb285 362 grep_read_lock();
76416139 363 data = read_sha1_file(sha1, type, size);
b3aeb285 364 grep_read_unlock();
5f02d315
JH
365 return data;
366}
367
5b594f45
FK
368static void *load_sha1(const unsigned char *sha1, unsigned long *size,
369 const char *name)
5010cb5f 370{
21666f1a 371 enum object_type type;
5f02d315 372 void *data = lock_and_read_sha1_file(sha1, &type, size);
5b594f45
FK
373
374 if (!data)
2fc5f9f1 375 error(_("'%s': unable to read %s"), name, sha1_to_hex(sha1));
5b594f45
FK
376
377 return data;
378}
379
380static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
381 const char *filename, int tree_name_len)
382{
383 struct strbuf pathbuf = STRBUF_INIT;
384 char *name;
385
0d042fec 386 if (opt->relative && opt->prefix_length) {
5b594f45
FK
387 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
388 opt->prefix);
389 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
390 } else {
391 strbuf_addstr(&pathbuf, filename);
392 }
393
394 name = strbuf_detach(&pathbuf, NULL);
395
396#ifndef NO_PTHREADS
397 if (use_threads) {
398 grep_sha1_async(opt, name, sha1);
399 return 0;
400 } else
401#endif
402 {
403 int hit;
404 unsigned long sz;
405 void *data = load_sha1(sha1, &sz, name);
406 if (!data)
407 hit = 0;
408 else
409 hit = grep_buffer(opt, name, data, sz);
410
411 free(data);
412 free(name);
413 return hit;
0d042fec 414 }
5010cb5f
JH
415}
416
5b594f45 417static void *load_file(const char *filename, size_t *sz)
5010cb5f
JH
418{
419 struct stat st;
5010cb5f 420 char *data;
5b594f45 421 int i;
dc49cd76 422
5010cb5f
JH
423 if (lstat(filename, &st) < 0) {
424 err_ret:
425 if (errno != ENOENT)
2fc5f9f1 426 error(_("'%s': %s"), filename, strerror(errno));
1e4cd68c 427 return NULL;
5010cb5f 428 }
5010cb5f 429 if (!S_ISREG(st.st_mode))
1e4cd68c 430 return NULL;
5b594f45 431 *sz = xsize_t(st.st_size);
5010cb5f
JH
432 i = open(filename, O_RDONLY);
433 if (i < 0)
434 goto err_ret;
5b594f45
FK
435 data = xmalloc(*sz + 1);
436 if (st.st_size != read_in_full(i, data, *sz)) {
2fc5f9f1 437 error(_("'%s': short read %s"), filename, strerror(errno));
5010cb5f
JH
438 close(i);
439 free(data);
1e4cd68c 440 return NULL;
5010cb5f
JH
441 }
442 close(i);
5b594f45
FK
443 data[*sz] = 0;
444 return data;
445}
446
447static int grep_file(struct grep_opt *opt, const char *filename)
448{
449 struct strbuf buf = STRBUF_INIT;
450 char *name;
451
0d042fec 452 if (opt->relative && opt->prefix_length)
5b594f45
FK
453 quote_path_relative(filename, -1, &buf, opt->prefix);
454 else
455 strbuf_addstr(&buf, filename);
456 name = strbuf_detach(&buf, NULL);
457
458#ifndef NO_PTHREADS
459 if (use_threads) {
460 grep_file_async(opt, name, filename);
461 return 0;
462 } else
463#endif
464 {
465 int hit;
466 size_t sz;
467 void *data = load_file(filename, &sz);
468 if (!data)
469 hit = 0;
470 else
471 hit = grep_buffer(opt, name, data, sz);
472
473 free(data);
474 free(name);
475 return hit;
476 }
5010cb5f
JH
477}
478
678e484b
JS
479static void append_path(struct grep_opt *opt, const void *data, size_t len)
480{
481 struct string_list *path_list = opt->output_priv;
482
483 if (len == 1 && *(const char *)data == '\0')
484 return;
0c72cead 485 string_list_append(path_list, xstrndup(data, len));
678e484b
JS
486}
487
488static void run_pager(struct grep_opt *opt, const char *prefix)
489{
490 struct string_list *path_list = opt->output_priv;
491 const char **argv = xmalloc(sizeof(const char *) * (path_list->nr + 1));
492 int i, status;
493
494 for (i = 0; i < path_list->nr; i++)
495 argv[i] = path_list->items[i].string;
496 argv[path_list->nr] = NULL;
497
498 if (prefix && chdir(prefix))
2fc5f9f1 499 die(_("Failed to chdir: %s"), prefix);
678e484b
JS
500 status = run_command_v_opt(argv, RUN_USING_SHELL);
501 if (status)
502 exit(status);
503 free(argv);
504}
505
f34bbc15 506static int grep_cache(struct grep_opt *opt, const struct pathspec *pathspec, int cached)
5010cb5f
JH
507{
508 int hit = 0;
509 int nr;
510 read_cache();
511
512 for (nr = 0; nr < active_nr; nr++) {
513 struct cache_entry *ce = active_cache[nr];
7a51ed66 514 if (!S_ISREG(ce->ce_mode))
5010cb5f 515 continue;
2ed2437a 516 if (!match_pathspec_depth(pathspec, ce->name, ce_namelen(ce), 0, NULL))
5010cb5f 517 continue;
57d43466
NTND
518 /*
519 * If CE_VALID is on, we assume worktree file and its cache entry
520 * are identical, even if worktree file has been modified, so use
521 * cache version instead
522 */
b4d1690d 523 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
36f2587f
JH
524 if (ce_stage(ce))
525 continue;
0d042fec 526 hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
36f2587f 527 }
5010cb5f
JH
528 else
529 hit |= grep_file(opt, ce->name);
36f2587f
JH
530 if (ce_stage(ce)) {
531 do {
532 nr++;
533 } while (nr < active_nr &&
534 !strcmp(ce->name, active_cache[nr]->name));
535 nr--; /* compensate for loop control */
536 }
c8610a2e
JH
537 if (hit && opt->status_only)
538 break;
5010cb5f
JH
539 }
540 return hit;
541}
542
f34bbc15 543static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec,
e5e062b6 544 struct tree_desc *tree, struct strbuf *base, int tn_len)
5010cb5f 545{
d688cf07
NTND
546 int hit = 0;
547 enum interesting match = entry_not_interesting;
4c068a98 548 struct name_entry entry;
e5e062b6 549 int old_baselen = base->len;
5010cb5f 550
4c068a98 551 while (tree_entry(tree, &entry)) {
0de16337 552 int te_len = tree_entry_len(&entry);
e5e062b6 553
d688cf07 554 if (match != all_entries_interesting) {
97d0b74a 555 match = tree_entry_interesting(&entry, base, tn_len, pathspec);
d688cf07 556 if (match == all_entries_not_interesting)
97d0b74a 557 break;
d688cf07 558 if (match == entry_not_interesting)
1376e507
NTND
559 continue;
560 }
5010cb5f 561
1376e507 562 strbuf_add(base, entry.path, te_len);
e0eb889f 563
1376e507 564 if (S_ISREG(entry.mode)) {
e5e062b6
NTND
565 hit |= grep_sha1(opt, entry.sha1, base->buf, tn_len);
566 }
4c068a98 567 else if (S_ISDIR(entry.mode)) {
21666f1a 568 enum object_type type;
5010cb5f
JH
569 struct tree_desc sub;
570 void *data;
6fda5e51
LT
571 unsigned long size;
572
5f02d315 573 data = lock_and_read_sha1_file(entry.sha1, &type, &size);
5010cb5f 574 if (!data)
2fc5f9f1 575 die(_("unable to read tree (%s)"),
4c068a98 576 sha1_to_hex(entry.sha1));
1376e507
NTND
577
578 strbuf_addch(base, '/');
6fda5e51 579 init_tree_desc(&sub, data, size);
e5e062b6 580 hit |= grep_tree(opt, pathspec, &sub, base, tn_len);
5010cb5f
JH
581 free(data);
582 }
e5e062b6
NTND
583 strbuf_setlen(base, old_baselen);
584
c8610a2e
JH
585 if (hit && opt->status_only)
586 break;
5010cb5f
JH
587 }
588 return hit;
589}
590
f34bbc15 591static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec,
5010cb5f
JH
592 struct object *obj, const char *name)
593{
1974632c 594 if (obj->type == OBJ_BLOB)
0d042fec 595 return grep_sha1(opt, obj->sha1, name, 0);
1974632c 596 if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
5010cb5f
JH
597 struct tree_desc tree;
598 void *data;
6fda5e51 599 unsigned long size;
e5e062b6
NTND
600 struct strbuf base;
601 int hit, len;
602
b3aeb285 603 grep_read_lock();
5010cb5f 604 data = read_object_with_reference(obj->sha1, tree_type,
6fda5e51 605 &size, NULL);
b3aeb285 606 grep_read_unlock();
8cb5775b 607
5010cb5f 608 if (!data)
2fc5f9f1 609 die(_("unable to read tree (%s)"), sha1_to_hex(obj->sha1));
e5e062b6
NTND
610
611 len = name ? strlen(name) : 0;
612 strbuf_init(&base, PATH_MAX + len + 1);
613 if (len) {
614 strbuf_add(&base, name, len);
615 strbuf_addch(&base, ':');
616 }
6fda5e51 617 init_tree_desc(&tree, data, size);
e5e062b6
NTND
618 hit = grep_tree(opt, pathspec, &tree, &base, base.len);
619 strbuf_release(&base);
5010cb5f
JH
620 free(data);
621 return hit;
622 }
2fc5f9f1 623 die(_("unable to grep from object of type %s"), typename(obj->type));
5010cb5f
JH
624}
625
f34bbc15 626static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec,
30d00c39
JN
627 const struct object_array *list)
628{
629 unsigned int i;
630 int hit = 0;
631 const unsigned int nr = list->nr;
632
633 for (i = 0; i < nr; i++) {
634 struct object *real_obj;
635 real_obj = deref_tag(list->objects[i].item, NULL, 0);
f34bbc15 636 if (grep_object(opt, pathspec, real_obj, list->objects[i].name)) {
30d00c39
JN
637 hit = 1;
638 if (opt->status_only)
639 break;
640 }
641 }
642 return hit;
643}
644
dbfae86a
JH
645static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec,
646 int exc_std)
59332d13
JH
647{
648 struct dir_struct dir;
649 int i, hit = 0;
650
651 memset(&dir, 0, sizeof(dir));
0a93fb8a
JH
652 if (exc_std)
653 setup_standard_excludes(&dir);
59332d13 654
f34bbc15 655 fill_directory(&dir, pathspec->raw);
59332d13 656 for (i = 0; i < dir.nr; i++) {
9d8b831b
JH
657 const char *name = dir.entries[i]->name;
658 int namelen = strlen(name);
659 if (!match_pathspec_depth(pathspec, name, namelen, 0, NULL))
660 continue;
59332d13
JH
661 hit |= grep_file(opt, dir.entries[i]->name);
662 if (hit && opt->status_only)
663 break;
664 }
59332d13
JH
665 return hit;
666}
667
ff3c7f9a
RS
668static int context_callback(const struct option *opt, const char *arg,
669 int unset)
3e230fa1
RS
670{
671 struct grep_opt *grep_opt = opt->value;
672 int value;
673 const char *endp;
674
675 if (unset) {
676 grep_opt->pre_context = grep_opt->post_context = 0;
677 return 0;
678 }
679 value = strtol(arg, (char **)&endp, 10);
680 if (*endp) {
2fc5f9f1 681 return error(_("switch `%c' expects a numerical value"),
3e230fa1
RS
682 opt->short_name);
683 }
684 grep_opt->pre_context = grep_opt->post_context = value;
685 return 0;
686}
687
ff3c7f9a 688static int file_callback(const struct option *opt, const char *arg, int unset)
3e230fa1
RS
689{
690 struct grep_opt *grep_opt = opt->value;
c41dd2fd 691 int from_stdin = !strcmp(arg, "-");
3e230fa1
RS
692 FILE *patterns;
693 int lno = 0;
cfe370c6 694 struct strbuf sb = STRBUF_INIT;
3e230fa1 695
c41dd2fd 696 patterns = from_stdin ? stdin : fopen(arg, "r");
3e230fa1 697 if (!patterns)
2fc5f9f1 698 die_errno(_("cannot open '%s'"), arg);
3e230fa1 699 while (strbuf_getline(&sb, patterns, '\n') == 0) {
ed40a095
RS
700 char *s;
701 size_t len;
702
3e230fa1
RS
703 /* ignore empty line like grep does */
704 if (sb.len == 0)
705 continue;
ed40a095
RS
706
707 s = strbuf_detach(&sb, &len);
708 append_grep_pat(grep_opt, s, len, arg, ++lno, GREP_PATTERN);
3e230fa1 709 }
c41dd2fd
RS
710 if (!from_stdin)
711 fclose(patterns);
3e230fa1
RS
712 strbuf_release(&sb);
713 return 0;
714}
715
ff3c7f9a 716static int not_callback(const struct option *opt, const char *arg, int unset)
3e230fa1
RS
717{
718 struct grep_opt *grep_opt = opt->value;
719 append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
720 return 0;
721}
722
ff3c7f9a 723static int and_callback(const struct option *opt, const char *arg, int unset)
3e230fa1
RS
724{
725 struct grep_opt *grep_opt = opt->value;
726 append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
727 return 0;
728}
729
ff3c7f9a 730static int open_callback(const struct option *opt, const char *arg, int unset)
3e230fa1
RS
731{
732 struct grep_opt *grep_opt = opt->value;
733 append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
734 return 0;
735}
736
ff3c7f9a 737static int close_callback(const struct option *opt, const char *arg, int unset)
3e230fa1
RS
738{
739 struct grep_opt *grep_opt = opt->value;
740 append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
741 return 0;
742}
743
ff3c7f9a
RS
744static int pattern_callback(const struct option *opt, const char *arg,
745 int unset)
3e230fa1
RS
746{
747 struct grep_opt *grep_opt = opt->value;
748 append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
749 return 0;
750}
5010cb5f 751
ff3c7f9a 752static int help_callback(const struct option *opt, const char *arg, int unset)
3e230fa1
RS
753{
754 return -1;
755}
088b084b 756
a633fca0 757int cmd_grep(int argc, const char **argv, const char *prefix)
5010cb5f 758{
5010cb5f 759 int hit = 0;
0a93fb8a 760 int cached = 0, untracked = 0, opt_exclude = -1;
5acd64ed 761 int seen_dashdash = 0;
bbc09c22 762 int external_grep_allowed__ignored;
0af88c15 763 const char *show_in_pager = NULL, *default_pager = "dummy";
5010cb5f 764 struct grep_opt opt;
3cd47459 765 struct object_array list = OBJECT_ARRAY_INIT;
1362671f 766 const char **paths = NULL;
f34bbc15 767 struct pathspec pathspec;
183113a5 768 struct string_list path_list = STRING_LIST_INIT_NODUP;
5acd64ed 769 int i;
3e230fa1 770 int dummy;
ff38d1a9 771 int use_index = 1;
cca2c172
JH
772 enum {
773 pattern_type_unspecified = 0,
774 pattern_type_bre,
775 pattern_type_ere,
776 pattern_type_fixed,
777 pattern_type_pcre,
778 };
779 int pattern_type = pattern_type_unspecified;
780
3e230fa1
RS
781 struct option options[] = {
782 OPT_BOOLEAN(0, "cached", &cached,
783 "search in index instead of in the work tree"),
ac1d33dd
BW
784 { OPTION_BOOLEAN, 0, "index", &use_index, NULL,
785 "finds in contents not managed by git",
786 PARSE_OPT_NOARG | PARSE_OPT_NEGHELP },
0a93fb8a
JH
787 OPT_BOOLEAN(0, "untracked", &untracked,
788 "search in both tracked and untracked files"),
789 OPT_SET_INT(0, "exclude-standard", &opt_exclude,
790 "search also in ignored files", 1),
3e230fa1
RS
791 OPT_GROUP(""),
792 OPT_BOOLEAN('v', "invert-match", &opt.invert,
793 "show non-matching lines"),
5183bf67
BC
794 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
795 "case insensitive matching"),
3e230fa1
RS
796 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
797 "match patterns only at word boundaries"),
798 OPT_SET_INT('a', "text", &opt.binary,
799 "process binary files as text", GREP_BINARY_TEXT),
800 OPT_SET_INT('I', NULL, &opt.binary,
801 "don't match patterns in binary files",
802 GREP_BINARY_NOMATCH),
a91f453f
MK
803 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
804 "descend at most <depth> levels", PARSE_OPT_NONEG,
805 NULL, 1 },
3e230fa1 806 OPT_GROUP(""),
cca2c172
JH
807 OPT_SET_INT('E', "extended-regexp", &pattern_type,
808 "use extended POSIX regular expressions",
809 pattern_type_ere),
810 OPT_SET_INT('G', "basic-regexp", &pattern_type,
811 "use basic POSIX regular expressions (default)",
812 pattern_type_bre),
813 OPT_SET_INT('F', "fixed-strings", &pattern_type,
814 "interpret patterns as fixed strings",
815 pattern_type_fixed),
816 OPT_SET_INT('P', "perl-regexp", &pattern_type,
817 "use Perl-compatible regular expressions",
818 pattern_type_pcre),
3e230fa1 819 OPT_GROUP(""),
7d6cb10b 820 OPT_BOOLEAN('n', "line-number", &opt.linenum, "show line numbers"),
3e230fa1
RS
821 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
822 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
823 OPT_NEGBIT(0, "full-name", &opt.relative,
824 "show filenames relative to top directory", 1),
825 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
826 "show only filenames instead of matching lines"),
827 OPT_BOOLEAN(0, "name-only", &opt.name_only,
828 "synonym for --files-with-matches"),
829 OPT_BOOLEAN('L', "files-without-match",
830 &opt.unmatch_name_only,
831 "show only the names of files without match"),
832 OPT_BOOLEAN('z', "null", &opt.null_following_name,
833 "print NUL after filenames"),
834 OPT_BOOLEAN('c', "count", &opt.count,
835 "show the number of matches instead of matching lines"),
73e9da01 836 OPT__COLOR(&opt.color, "highlight matches"),
a8f0e764
RS
837 OPT_BOOLEAN(0, "break", &opt.file_break,
838 "print empty line between matches from different files"),
1d84f72e
RS
839 OPT_BOOLEAN(0, "heading", &opt.heading,
840 "show filename only once above matches from same file"),
3e230fa1 841 OPT_GROUP(""),
317f63c2 842 OPT_CALLBACK('C', "context", &opt, "n",
3e230fa1
RS
843 "show <n> context lines before and after matches",
844 context_callback),
317f63c2 845 OPT_INTEGER('B', "before-context", &opt.pre_context,
3e230fa1 846 "show <n> context lines before matches"),
317f63c2 847 OPT_INTEGER('A', "after-context", &opt.post_context,
3e230fa1
RS
848 "show <n> context lines after matches"),
849 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
850 context_callback),
2944e4e6
RS
851 OPT_BOOLEAN('p', "show-function", &opt.funcname,
852 "show a line with the function name before matches"),
317f63c2 853 OPT_BOOLEAN('W', "function-context", &opt.funcbody,
ba8ea749 854 "show the surrounding function"),
3e230fa1
RS
855 OPT_GROUP(""),
856 OPT_CALLBACK('f', NULL, &opt, "file",
857 "read patterns from file", file_callback),
858 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
859 "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
860 { OPTION_CALLBACK, 0, "and", &opt, NULL,
861 "combine patterns specified with -e",
862 PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
863 OPT_BOOLEAN(0, "or", &dummy, ""),
864 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
865 PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
866 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
867 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
868 open_callback },
869 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
870 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
871 close_callback },
d52ee6e6
RS
872 OPT__QUIET(&opt.status_only,
873 "indicate hit with exit status without output"),
3e230fa1
RS
874 OPT_BOOLEAN(0, "all-match", &opt.all_match,
875 "show only matches from files that match all patterns"),
876 OPT_GROUP(""),
0af88c15
JS
877 { OPTION_STRING, 'O', "open-files-in-pager", &show_in_pager,
878 "pager", "show matching files in the pager",
879 PARSE_OPT_OPTARG, NULL, (intptr_t)default_pager },
bbc09c22
JH
880 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
881 "allow calling of grep(1) (ignored by this build)"),
3e230fa1
RS
882 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
883 PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
884 OPT_END()
885 };
5010cb5f 886
9c855c31
JN
887 /*
888 * 'git grep -h', unlike 'git grep -h <pattern>', is a request
889 * to show usage information and exit.
890 */
891 if (argc == 2 && !strcmp(argv[1], "-h"))
892 usage_with_options(grep_usage, options);
893
5010cb5f 894 memset(&opt, 0, sizeof(opt));
493b7a08 895 opt.prefix = prefix;
0d042fec
JH
896 opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
897 opt.relative = 1;
7977f0ea 898 opt.pathname = 1;
f9b9faf6 899 opt.pattern_tail = &opt.pattern_list;
80235ba7 900 opt.header_tail = &opt.header_list;
5010cb5f 901 opt.regflags = REG_NEWLINE;
a91f453f 902 opt.max_depth = -1;
5010cb5f 903
00588bb5 904 strcpy(opt.color_context, "");
55f638bd 905 strcpy(opt.color_filename, "");
00588bb5 906 strcpy(opt.color_function, "");
55f638bd 907 strcpy(opt.color_lineno, "");
758df17a 908 strcpy(opt.color_match, GIT_COLOR_BOLD_RED);
00588bb5 909 strcpy(opt.color_selected, "");
55f638bd 910 strcpy(opt.color_sep, GIT_COLOR_CYAN);
7e8f59d5
RS
911 opt.color = -1;
912 git_config(grep_config, &opt);
7e8f59d5 913
5010cb5f 914 /*
5acd64ed
JH
915 * If there is no -- then the paths must exist in the working
916 * tree. If there is no explicit pattern specified with -e or
917 * -f, we take the first unrecognized non option to be the
918 * pattern, but then what follows it must be zero or more
919 * valid refs up to the -- (if exists), and then existing
920 * paths. If there is an explicit pattern, then the first
82e5a82f 921 * unrecognized non option is the beginning of the refs list
5acd64ed 922 * that continues up to the -- (if exists), and then paths.
5010cb5f 923 */
37782920 924 argc = parse_options(argc, argv, prefix, options, grep_usage,
3e230fa1
RS
925 PARSE_OPT_KEEP_DASHDASH |
926 PARSE_OPT_STOP_AT_NON_OPTION |
927 PARSE_OPT_NO_INTERNAL_HELP);
cca2c172
JH
928 switch (pattern_type) {
929 case pattern_type_fixed:
930 opt.fixed = 1;
931 opt.pcre = 0;
932 break;
933 case pattern_type_bre:
934 opt.fixed = 0;
935 opt.pcre = 0;
936 opt.regflags &= ~REG_EXTENDED;
937 break;
938 case pattern_type_ere:
939 opt.fixed = 0;
940 opt.pcre = 0;
941 opt.regflags |= REG_EXTENDED;
942 break;
943 case pattern_type_pcre:
944 opt.fixed = 0;
945 opt.pcre = 1;
946 break;
947 default:
948 break; /* nothing */
949 }
3e230fa1 950
ff38d1a9 951 if (use_index && !startup_info->have_repository)
59332d13
JH
952 /* die the same way as if we did it at the beginning */
953 setup_git_directory();
954
1123c67c
JK
955 /*
956 * skip a -- separator; we know it cannot be
957 * separating revisions from pathnames if
958 * we haven't even had any patterns yet
959 */
960 if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) {
961 argv++;
962 argc--;
963 }
964
3e230fa1
RS
965 /* First unrecognized non-option token */
966 if (argc > 0 && !opt.pattern_list) {
967 append_grep_pattern(&opt, argv[0], "command line", 0,
968 GREP_PATTERN);
969 argv++;
970 argc--;
5010cb5f 971 }
5acd64ed 972
0af88c15
JS
973 if (show_in_pager == default_pager)
974 show_in_pager = git_pager(1);
678e484b 975 if (show_in_pager) {
e7b082a4 976 opt.color = 0;
0af88c15
JS
977 opt.name_only = 1;
978 opt.null_following_name = 1;
979 opt.output_priv = &path_list;
980 opt.output = append_path;
0c72cead 981 string_list_append(&path_list, show_in_pager);
0af88c15 982 use_threads = 0;
678e484b
JS
983 }
984
f9b9faf6 985 if (!opt.pattern_list)
2fc5f9f1 986 die(_("no pattern given."));
5183bf67
BC
987 if (!opt.fixed && opt.ignore_case)
988 opt.regflags |= REG_ICASE;
5b594f45 989
83b5d2f5 990 compile_grep_patterns(&opt);
5acd64ed
JH
991
992 /* Check revs and then paths */
3e230fa1 993 for (i = 0; i < argc; i++) {
5acd64ed 994 const char *arg = argv[i];
1362671f 995 unsigned char sha1[20];
5acd64ed
JH
996 /* Is it a rev? */
997 if (!get_sha1(arg, sha1)) {
998 struct object *object = parse_object(sha1);
5acd64ed 999 if (!object)
2fc5f9f1 1000 die(_("bad object %s"), arg);
1f1e895f 1001 add_object_array(object, arg, &list);
5acd64ed
JH
1002 continue;
1003 }
1004 if (!strcmp(arg, "--")) {
1005 i++;
1006 seen_dashdash = 1;
1007 }
1008 break;
1362671f 1009 }
5acd64ed 1010
53b8d931
TR
1011#ifndef NO_PTHREADS
1012 if (list.nr || cached || online_cpus() == 1)
1013 use_threads = 0;
1014#else
1015 use_threads = 0;
1016#endif
1017
53b8d931
TR
1018#ifndef NO_PTHREADS
1019 if (use_threads) {
1020 if (opt.pre_context || opt.post_context || opt.file_break ||
1021 opt.funcbody)
1022 skip_first_line = 1;
1023 start_threads(&opt);
1024 }
1025#endif
1026
5acd64ed
JH
1027 /* The rest are paths */
1028 if (!seen_dashdash) {
1029 int j;
c39c4f47 1030 for (j = i; j < argc; j++)
5acd64ed
JH
1031 verify_filename(prefix, argv[j]);
1032 }
1033
7c5f3cc4 1034 paths = get_pathspec(prefix, argv + i);
f34bbc15 1035 init_pathspec(&pathspec, paths);
1376e507
NTND
1036 pathspec.max_depth = opt.max_depth;
1037 pathspec.recursive = 1;
5010cb5f 1038
678e484b 1039 if (show_in_pager && (cached || list.nr))
e4fe4ba5 1040 die(_("--open-files-in-pager only works on the worktree"));
678e484b
JS
1041
1042 if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) {
1043 const char *pager = path_list.items[0].string;
1044 int len = strlen(pager);
1045
1046 if (len > 4 && is_dir_sep(pager[len - 5]))
1047 pager += len - 4;
1048
1049 if (!strcmp("less", pager) || !strcmp("vi", pager)) {
1050 struct strbuf buf = STRBUF_INIT;
1051 strbuf_addf(&buf, "+/%s%s",
1052 strcmp("less", pager) ? "" : "*",
1053 opt.pattern_list->pattern);
0c72cead 1054 string_list_append(&path_list, buf.buf);
678e484b
JS
1055 strbuf_detach(&buf, NULL);
1056 }
1057 }
1058
1059 if (!show_in_pager)
1060 setup_pager();
1061
0a93fb8a 1062 if (!use_index && (untracked || cached))
dbfae86a 1063 die(_("--cached or --untracked cannot be used with --no-index."));
678e484b 1064
0a93fb8a 1065 if (!use_index || untracked) {
0a93fb8a 1066 int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude;
59332d13 1067 if (list.nr)
dbfae86a
JH
1068 die(_("--no-index or --untracked cannot be used with revs."));
1069 hit = grep_directory(&opt, &pathspec, use_exclude);
1070 } else if (0 <= opt_exclude) {
9fddaf78 1071 die(_("--[no-]exclude-standard cannot be used for tracked contents."));
685359cf 1072 } else if (!list.nr) {
6577f542
NTND
1073 if (!cached)
1074 setup_work_tree();
5b594f45 1075
f34bbc15 1076 hit = grep_cache(&opt, &pathspec, cached);
685359cf
JS
1077 } else {
1078 if (cached)
2fc5f9f1 1079 die(_("both --cached and trees are given."));
f34bbc15 1080 hit = grep_objects(&opt, &pathspec, &list);
5010cb5f 1081 }
5b594f45
FK
1082
1083 if (use_threads)
1084 hit |= wait_all();
678e484b
JS
1085 if (hit && show_in_pager)
1086 run_pager(&opt, prefix);
b48fb5b6 1087 free_grep_patterns(&opt);
5010cb5f
JH
1088 return !hit;
1089}