Merge branch 'jc/grep-lookahead'
[git/git.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 "userdiff.h"
15 #include "grep.h"
16 #include "quote.h"
17 #include "dir.h"
18
19 static char const * const grep_usage[] = {
20 "git grep [options] [-e] <pattern> [<rev>...] [[--] path...]",
21 NULL
22 };
23
24 static int grep_config(const char *var, const char *value, void *cb)
25 {
26 struct grep_opt *opt = cb;
27
28 switch (userdiff_config(var, value)) {
29 case 0: break;
30 case -1: return -1;
31 default: return 0;
32 }
33
34 if (!strcmp(var, "color.grep")) {
35 opt->color = git_config_colorbool(var, value, -1);
36 return 0;
37 }
38 if (!strcmp(var, "color.grep.match")) {
39 if (!value)
40 return config_error_nonbool(var);
41 color_parse(value, var, opt->color_match);
42 return 0;
43 }
44 return git_color_default_config(var, value, cb);
45 }
46
47 /*
48 * Return non-zero if max_depth is negative or path has no more then max_depth
49 * slashes.
50 */
51 static int accept_subdir(const char *path, int max_depth)
52 {
53 if (max_depth < 0)
54 return 1;
55
56 while ((path = strchr(path, '/')) != NULL) {
57 max_depth--;
58 if (max_depth < 0)
59 return 0;
60 path++;
61 }
62 return 1;
63 }
64
65 /*
66 * Return non-zero if name is a subdirectory of match and is not too deep.
67 */
68 static int is_subdir(const char *name, int namelen,
69 const char *match, int matchlen, int max_depth)
70 {
71 if (matchlen > namelen || strncmp(name, match, matchlen))
72 return 0;
73
74 if (name[matchlen] == '\0') /* exact match */
75 return 1;
76
77 if (!matchlen || match[matchlen-1] == '/' || name[matchlen] == '/')
78 return accept_subdir(name + matchlen + 1, max_depth);
79
80 return 0;
81 }
82
83 /*
84 * git grep pathspecs are somewhat different from diff-tree pathspecs;
85 * pathname wildcards are allowed.
86 */
87 static int pathspec_matches(const char **paths, const char *name, int max_depth)
88 {
89 int namelen, i;
90 if (!paths || !*paths)
91 return accept_subdir(name, max_depth);
92 namelen = strlen(name);
93 for (i = 0; paths[i]; i++) {
94 const char *match = paths[i];
95 int matchlen = strlen(match);
96 const char *cp, *meta;
97
98 if (is_subdir(name, namelen, match, matchlen, max_depth))
99 return 1;
100 if (!fnmatch(match, name, 0))
101 return 1;
102 if (name[namelen-1] != '/')
103 continue;
104
105 /* We are being asked if the directory ("name") is worth
106 * descending into.
107 *
108 * Find the longest leading directory name that does
109 * not have metacharacter in the pathspec; the name
110 * we are looking at must overlap with that directory.
111 */
112 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
113 char ch = *cp;
114 if (ch == '*' || ch == '[' || ch == '?') {
115 meta = cp;
116 break;
117 }
118 }
119 if (!meta)
120 meta = cp; /* fully literal */
121
122 if (namelen <= meta - match) {
123 /* Looking at "Documentation/" and
124 * the pattern says "Documentation/howto/", or
125 * "Documentation/diff*.txt". The name we
126 * have should match prefix.
127 */
128 if (!memcmp(match, name, namelen))
129 return 1;
130 continue;
131 }
132
133 if (meta - match < namelen) {
134 /* Looking at "Documentation/howto/" and
135 * the pattern says "Documentation/h*";
136 * match up to "Do.../h"; this avoids descending
137 * into "Documentation/technical/".
138 */
139 if (!memcmp(match, name, meta - match))
140 return 1;
141 continue;
142 }
143 }
144 return 0;
145 }
146
147 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name, int tree_name_len)
148 {
149 unsigned long size;
150 char *data;
151 enum object_type type;
152 int hit;
153 struct strbuf pathbuf = STRBUF_INIT;
154
155 data = read_sha1_file(sha1, &type, &size);
156 if (!data) {
157 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
158 return 0;
159 }
160 if (opt->relative && opt->prefix_length) {
161 quote_path_relative(name + tree_name_len, -1, &pathbuf, opt->prefix);
162 strbuf_insert(&pathbuf, 0, name, tree_name_len);
163 name = pathbuf.buf;
164 }
165 hit = grep_buffer(opt, name, data, size);
166 strbuf_release(&pathbuf);
167 free(data);
168 return hit;
169 }
170
171 static int grep_file(struct grep_opt *opt, const char *filename)
172 {
173 struct stat st;
174 int i;
175 char *data;
176 size_t sz;
177 struct strbuf buf = STRBUF_INIT;
178
179 if (lstat(filename, &st) < 0) {
180 err_ret:
181 if (errno != ENOENT)
182 error("'%s': %s", filename, strerror(errno));
183 return 0;
184 }
185 if (!S_ISREG(st.st_mode))
186 return 0;
187 sz = xsize_t(st.st_size);
188 i = open(filename, O_RDONLY);
189 if (i < 0)
190 goto err_ret;
191 data = xmalloc(sz + 1);
192 if (st.st_size != read_in_full(i, data, sz)) {
193 error("'%s': short read %s", filename, strerror(errno));
194 close(i);
195 free(data);
196 return 0;
197 }
198 close(i);
199 data[sz] = 0;
200 if (opt->relative && opt->prefix_length)
201 filename = quote_path_relative(filename, -1, &buf, opt->prefix);
202 i = grep_buffer(opt, filename, data, sz);
203 strbuf_release(&buf);
204 free(data);
205 return i;
206 }
207
208 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
209 {
210 int hit = 0;
211 int nr;
212 read_cache();
213
214 for (nr = 0; nr < active_nr; nr++) {
215 struct cache_entry *ce = active_cache[nr];
216 if (!S_ISREG(ce->ce_mode))
217 continue;
218 if (!pathspec_matches(paths, ce->name, opt->max_depth))
219 continue;
220 /*
221 * If CE_VALID is on, we assume worktree file and its cache entry
222 * are identical, even if worktree file has been modified, so use
223 * cache version instead
224 */
225 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
226 if (ce_stage(ce))
227 continue;
228 hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
229 }
230 else
231 hit |= grep_file(opt, ce->name);
232 if (ce_stage(ce)) {
233 do {
234 nr++;
235 } while (nr < active_nr &&
236 !strcmp(ce->name, active_cache[nr]->name));
237 nr--; /* compensate for loop control */
238 }
239 }
240 free_grep_patterns(opt);
241 return hit;
242 }
243
244 static int grep_tree(struct grep_opt *opt, const char **paths,
245 struct tree_desc *tree,
246 const char *tree_name, const char *base)
247 {
248 int len;
249 int hit = 0;
250 struct name_entry entry;
251 char *down;
252 int tn_len = strlen(tree_name);
253 struct strbuf pathbuf;
254
255 strbuf_init(&pathbuf, PATH_MAX + tn_len);
256
257 if (tn_len) {
258 strbuf_add(&pathbuf, tree_name, tn_len);
259 strbuf_addch(&pathbuf, ':');
260 tn_len = pathbuf.len;
261 }
262 strbuf_addstr(&pathbuf, base);
263 len = pathbuf.len;
264
265 while (tree_entry(tree, &entry)) {
266 int te_len = tree_entry_len(entry.path, entry.sha1);
267 pathbuf.len = len;
268 strbuf_add(&pathbuf, entry.path, te_len);
269
270 if (S_ISDIR(entry.mode))
271 /* Match "abc/" against pathspec to
272 * decide if we want to descend into "abc"
273 * directory.
274 */
275 strbuf_addch(&pathbuf, '/');
276
277 down = pathbuf.buf + tn_len;
278 if (!pathspec_matches(paths, down, opt->max_depth))
279 ;
280 else if (S_ISREG(entry.mode))
281 hit |= grep_sha1(opt, entry.sha1, pathbuf.buf, tn_len);
282 else if (S_ISDIR(entry.mode)) {
283 enum object_type type;
284 struct tree_desc sub;
285 void *data;
286 unsigned long size;
287
288 data = read_sha1_file(entry.sha1, &type, &size);
289 if (!data)
290 die("unable to read tree (%s)",
291 sha1_to_hex(entry.sha1));
292 init_tree_desc(&sub, data, size);
293 hit |= grep_tree(opt, paths, &sub, tree_name, down);
294 free(data);
295 }
296 }
297 strbuf_release(&pathbuf);
298 return hit;
299 }
300
301 static int grep_object(struct grep_opt *opt, const char **paths,
302 struct object *obj, const char *name)
303 {
304 if (obj->type == OBJ_BLOB)
305 return grep_sha1(opt, obj->sha1, name, 0);
306 if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
307 struct tree_desc tree;
308 void *data;
309 unsigned long size;
310 int hit;
311 data = read_object_with_reference(obj->sha1, tree_type,
312 &size, NULL);
313 if (!data)
314 die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
315 init_tree_desc(&tree, data, size);
316 hit = grep_tree(opt, paths, &tree, name, "");
317 free(data);
318 return hit;
319 }
320 die("unable to grep from object of type %s", typename(obj->type));
321 }
322
323 static int grep_directory(struct grep_opt *opt, const char **paths)
324 {
325 struct dir_struct dir;
326 int i, hit = 0;
327
328 memset(&dir, 0, sizeof(dir));
329 setup_standard_excludes(&dir);
330
331 fill_directory(&dir, paths);
332 for (i = 0; i < dir.nr; i++)
333 hit |= grep_file(opt, dir.entries[i]->name);
334 free_grep_patterns(opt);
335 return hit;
336 }
337
338 static int context_callback(const struct option *opt, const char *arg,
339 int unset)
340 {
341 struct grep_opt *grep_opt = opt->value;
342 int value;
343 const char *endp;
344
345 if (unset) {
346 grep_opt->pre_context = grep_opt->post_context = 0;
347 return 0;
348 }
349 value = strtol(arg, (char **)&endp, 10);
350 if (*endp) {
351 return error("switch `%c' expects a numerical value",
352 opt->short_name);
353 }
354 grep_opt->pre_context = grep_opt->post_context = value;
355 return 0;
356 }
357
358 static int file_callback(const struct option *opt, const char *arg, int unset)
359 {
360 struct grep_opt *grep_opt = opt->value;
361 FILE *patterns;
362 int lno = 0;
363 struct strbuf sb = STRBUF_INIT;
364
365 patterns = fopen(arg, "r");
366 if (!patterns)
367 die_errno("cannot open '%s'", arg);
368 while (strbuf_getline(&sb, patterns, '\n') == 0) {
369 /* ignore empty line like grep does */
370 if (sb.len == 0)
371 continue;
372 append_grep_pattern(grep_opt, strbuf_detach(&sb, NULL), arg,
373 ++lno, GREP_PATTERN);
374 }
375 fclose(patterns);
376 strbuf_release(&sb);
377 return 0;
378 }
379
380 static int not_callback(const struct option *opt, const char *arg, int unset)
381 {
382 struct grep_opt *grep_opt = opt->value;
383 append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
384 return 0;
385 }
386
387 static int and_callback(const struct option *opt, const char *arg, int unset)
388 {
389 struct grep_opt *grep_opt = opt->value;
390 append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
391 return 0;
392 }
393
394 static int open_callback(const struct option *opt, const char *arg, int unset)
395 {
396 struct grep_opt *grep_opt = opt->value;
397 append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
398 return 0;
399 }
400
401 static int close_callback(const struct option *opt, const char *arg, int unset)
402 {
403 struct grep_opt *grep_opt = opt->value;
404 append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
405 return 0;
406 }
407
408 static int pattern_callback(const struct option *opt, const char *arg,
409 int unset)
410 {
411 struct grep_opt *grep_opt = opt->value;
412 append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
413 return 0;
414 }
415
416 static int help_callback(const struct option *opt, const char *arg, int unset)
417 {
418 return -1;
419 }
420
421 int cmd_grep(int argc, const char **argv, const char *prefix)
422 {
423 int hit = 0;
424 int cached = 0;
425 int seen_dashdash = 0;
426 int external_grep_allowed__ignored;
427 struct grep_opt opt;
428 struct object_array list = { 0, 0, NULL };
429 const char **paths = NULL;
430 int i;
431 int dummy;
432 int nongit = 0, use_index = 1;
433 struct option options[] = {
434 OPT_BOOLEAN(0, "cached", &cached,
435 "search in index instead of in the work tree"),
436 OPT_BOOLEAN(0, "index", &use_index,
437 "--no-index finds in contents not managed by git"),
438 OPT_GROUP(""),
439 OPT_BOOLEAN('v', "invert-match", &opt.invert,
440 "show non-matching lines"),
441 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
442 "case insensitive matching"),
443 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
444 "match patterns only at word boundaries"),
445 OPT_SET_INT('a', "text", &opt.binary,
446 "process binary files as text", GREP_BINARY_TEXT),
447 OPT_SET_INT('I', NULL, &opt.binary,
448 "don't match patterns in binary files",
449 GREP_BINARY_NOMATCH),
450 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
451 "descend at most <depth> levels", PARSE_OPT_NONEG,
452 NULL, 1 },
453 OPT_GROUP(""),
454 OPT_BIT('E', "extended-regexp", &opt.regflags,
455 "use extended POSIX regular expressions", REG_EXTENDED),
456 OPT_NEGBIT('G', "basic-regexp", &opt.regflags,
457 "use basic POSIX regular expressions (default)",
458 REG_EXTENDED),
459 OPT_BOOLEAN('F', "fixed-strings", &opt.fixed,
460 "interpret patterns as fixed strings"),
461 OPT_GROUP(""),
462 OPT_BOOLEAN('n', NULL, &opt.linenum, "show line numbers"),
463 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
464 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
465 OPT_NEGBIT(0, "full-name", &opt.relative,
466 "show filenames relative to top directory", 1),
467 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
468 "show only filenames instead of matching lines"),
469 OPT_BOOLEAN(0, "name-only", &opt.name_only,
470 "synonym for --files-with-matches"),
471 OPT_BOOLEAN('L', "files-without-match",
472 &opt.unmatch_name_only,
473 "show only the names of files without match"),
474 OPT_BOOLEAN('z', "null", &opt.null_following_name,
475 "print NUL after filenames"),
476 OPT_BOOLEAN('c', "count", &opt.count,
477 "show the number of matches instead of matching lines"),
478 OPT_SET_INT(0, "color", &opt.color, "highlight matches", 1),
479 OPT_GROUP(""),
480 OPT_CALLBACK('C', NULL, &opt, "n",
481 "show <n> context lines before and after matches",
482 context_callback),
483 OPT_INTEGER('B', NULL, &opt.pre_context,
484 "show <n> context lines before matches"),
485 OPT_INTEGER('A', NULL, &opt.post_context,
486 "show <n> context lines after matches"),
487 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
488 context_callback),
489 OPT_BOOLEAN('p', "show-function", &opt.funcname,
490 "show a line with the function name before matches"),
491 OPT_GROUP(""),
492 OPT_CALLBACK('f', NULL, &opt, "file",
493 "read patterns from file", file_callback),
494 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
495 "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
496 { OPTION_CALLBACK, 0, "and", &opt, NULL,
497 "combine patterns specified with -e",
498 PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
499 OPT_BOOLEAN(0, "or", &dummy, ""),
500 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
501 PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
502 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
503 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
504 open_callback },
505 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
506 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
507 close_callback },
508 OPT_BOOLEAN(0, "all-match", &opt.all_match,
509 "show only matches from files that match all patterns"),
510 OPT_GROUP(""),
511 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
512 "allow calling of grep(1) (ignored by this build)"),
513 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
514 PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
515 OPT_END()
516 };
517
518 prefix = setup_git_directory_gently(&nongit);
519
520 /*
521 * 'git grep -h', unlike 'git grep -h <pattern>', is a request
522 * to show usage information and exit.
523 */
524 if (argc == 2 && !strcmp(argv[1], "-h"))
525 usage_with_options(grep_usage, options);
526
527 memset(&opt, 0, sizeof(opt));
528 opt.prefix = prefix;
529 opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
530 opt.relative = 1;
531 opt.pathname = 1;
532 opt.pattern_tail = &opt.pattern_list;
533 opt.regflags = REG_NEWLINE;
534 opt.max_depth = -1;
535
536 strcpy(opt.color_match, GIT_COLOR_RED GIT_COLOR_BOLD);
537 opt.color = -1;
538 git_config(grep_config, &opt);
539 if (opt.color == -1)
540 opt.color = git_use_color_default;
541
542 /*
543 * If there is no -- then the paths must exist in the working
544 * tree. If there is no explicit pattern specified with -e or
545 * -f, we take the first unrecognized non option to be the
546 * pattern, but then what follows it must be zero or more
547 * valid refs up to the -- (if exists), and then existing
548 * paths. If there is an explicit pattern, then the first
549 * unrecognized non option is the beginning of the refs list
550 * that continues up to the -- (if exists), and then paths.
551 */
552 argc = parse_options(argc, argv, prefix, options, grep_usage,
553 PARSE_OPT_KEEP_DASHDASH |
554 PARSE_OPT_STOP_AT_NON_OPTION |
555 PARSE_OPT_NO_INTERNAL_HELP);
556
557 if (use_index && nongit)
558 /* die the same way as if we did it at the beginning */
559 setup_git_directory();
560
561 /* First unrecognized non-option token */
562 if (argc > 0 && !opt.pattern_list) {
563 append_grep_pattern(&opt, argv[0], "command line", 0,
564 GREP_PATTERN);
565 argv++;
566 argc--;
567 }
568
569 if (!opt.pattern_list)
570 die("no pattern given.");
571 if (!opt.fixed && opt.ignore_case)
572 opt.regflags |= REG_ICASE;
573 if ((opt.regflags != REG_NEWLINE) && opt.fixed)
574 die("cannot mix --fixed-strings and regexp");
575 compile_grep_patterns(&opt);
576
577 /* Check revs and then paths */
578 for (i = 0; i < argc; i++) {
579 const char *arg = argv[i];
580 unsigned char sha1[20];
581 /* Is it a rev? */
582 if (!get_sha1(arg, sha1)) {
583 struct object *object = parse_object(sha1);
584 if (!object)
585 die("bad object %s", arg);
586 add_object_array(object, arg, &list);
587 continue;
588 }
589 if (!strcmp(arg, "--")) {
590 i++;
591 seen_dashdash = 1;
592 }
593 break;
594 }
595
596 /* The rest are paths */
597 if (!seen_dashdash) {
598 int j;
599 for (j = i; j < argc; j++)
600 verify_filename(prefix, argv[j]);
601 }
602
603 if (i < argc)
604 paths = get_pathspec(prefix, argv + i);
605 else if (prefix) {
606 paths = xcalloc(2, sizeof(const char *));
607 paths[0] = prefix;
608 paths[1] = NULL;
609 }
610
611 if (!use_index) {
612 if (cached)
613 die("--cached cannot be used with --no-index.");
614 if (list.nr)
615 die("--no-index cannot be used with revs.");
616 return !grep_directory(&opt, paths);
617 }
618
619 if (!list.nr) {
620 if (!cached)
621 setup_work_tree();
622 return !grep_cache(&opt, paths, cached);
623 }
624
625 if (cached)
626 die("both --cached and trees are given.");
627
628 for (i = 0; i < list.nr; i++) {
629 struct object *real_obj;
630 real_obj = deref_tag(list.objects[i].item, NULL, 0);
631 if (grep_object(&opt, paths, real_obj, list.objects[i].name))
632 hit = 1;
633 }
634 free_grep_patterns(&opt);
635 return !hit;
636 }