grep: expose "status-only" feature via -q
[git/jrn.git] / builtin-grep.c
blob64cdfefdd1f701cdb1dbb7921e82cf16f2440668
1 /*
2 * Builtin "git grep"
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"
19 static char const * const grep_usage[] = {
20 "git grep [options] [-e] <pattern> [<rev>...] [[--] path...]",
21 NULL
24 static int grep_config(const char *var, const char *value, void *cb)
26 struct grep_opt *opt = cb;
28 switch (userdiff_config(var, value)) {
29 case 0: break;
30 case -1: return -1;
31 default: return 0;
34 if (!strcmp(var, "color.grep")) {
35 opt->color = git_config_colorbool(var, value, -1);
36 return 0;
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;
44 return git_color_default_config(var, value, cb);
48 * Return non-zero if max_depth is negative or path has no more then max_depth
49 * slashes.
51 static int accept_subdir(const char *path, int max_depth)
53 if (max_depth < 0)
54 return 1;
56 while ((path = strchr(path, '/')) != NULL) {
57 max_depth--;
58 if (max_depth < 0)
59 return 0;
60 path++;
62 return 1;
66 * Return non-zero if name is a subdirectory of match and is not too deep.
68 static int is_subdir(const char *name, int namelen,
69 const char *match, int matchlen, int max_depth)
71 if (matchlen > namelen || strncmp(name, match, matchlen))
72 return 0;
74 if (name[matchlen] == '\0') /* exact match */
75 return 1;
77 if (!matchlen || match[matchlen-1] == '/' || name[matchlen] == '/')
78 return accept_subdir(name + matchlen + 1, max_depth);
80 return 0;
84 * git grep pathspecs are somewhat different from diff-tree pathspecs;
85 * pathname wildcards are allowed.
87 static int pathspec_matches(const char **paths, const char *name, int max_depth)
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;
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;
105 /* We are being asked if the directory ("name") is worth
106 * descending into.
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.
112 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
113 char ch = *cp;
114 if (ch == '*' || ch == '[' || ch == '?') {
115 meta = cp;
116 break;
119 if (!meta)
120 meta = cp; /* fully literal */
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.
128 if (!memcmp(match, name, namelen))
129 return 1;
130 continue;
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/".
139 if (!memcmp(match, name, meta - match))
140 return 1;
141 continue;
144 return 0;
147 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name, int tree_name_len)
149 unsigned long size;
150 char *data;
151 enum object_type type;
152 int hit;
153 struct strbuf pathbuf = STRBUF_INIT;
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;
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;
165 hit = grep_buffer(opt, name, data, size);
166 strbuf_release(&pathbuf);
167 free(data);
168 return hit;
171 static int grep_file(struct grep_opt *opt, const char *filename)
173 struct stat st;
174 int i;
175 char *data;
176 size_t sz;
177 struct strbuf buf = STRBUF_INIT;
179 if (lstat(filename, &st) < 0) {
180 err_ret:
181 if (errno != ENOENT)
182 error("'%s': %s", filename, strerror(errno));
183 return 0;
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;
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;
208 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
210 int hit = 0;
211 int nr;
212 read_cache();
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;
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
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);
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 */
239 if (hit && opt->status_only)
240 break;
242 free_grep_patterns(opt);
243 return hit;
246 static int grep_tree(struct grep_opt *opt, const char **paths,
247 struct tree_desc *tree,
248 const char *tree_name, const char *base)
250 int len;
251 int hit = 0;
252 struct name_entry entry;
253 char *down;
254 int tn_len = strlen(tree_name);
255 struct strbuf pathbuf;
257 strbuf_init(&pathbuf, PATH_MAX + tn_len);
259 if (tn_len) {
260 strbuf_add(&pathbuf, tree_name, tn_len);
261 strbuf_addch(&pathbuf, ':');
262 tn_len = pathbuf.len;
264 strbuf_addstr(&pathbuf, base);
265 len = pathbuf.len;
267 while (tree_entry(tree, &entry)) {
268 int te_len = tree_entry_len(entry.path, entry.sha1);
269 pathbuf.len = len;
270 strbuf_add(&pathbuf, entry.path, te_len);
272 if (S_ISDIR(entry.mode))
273 /* Match "abc/" against pathspec to
274 * decide if we want to descend into "abc"
275 * directory.
277 strbuf_addch(&pathbuf, '/');
279 down = pathbuf.buf + tn_len;
280 if (!pathspec_matches(paths, down, opt->max_depth))
282 else if (S_ISREG(entry.mode))
283 hit |= grep_sha1(opt, entry.sha1, pathbuf.buf, tn_len);
284 else if (S_ISDIR(entry.mode)) {
285 enum object_type type;
286 struct tree_desc sub;
287 void *data;
288 unsigned long size;
290 data = read_sha1_file(entry.sha1, &type, &size);
291 if (!data)
292 die("unable to read tree (%s)",
293 sha1_to_hex(entry.sha1));
294 init_tree_desc(&sub, data, size);
295 hit |= grep_tree(opt, paths, &sub, tree_name, down);
296 free(data);
298 if (hit && opt->status_only)
299 break;
301 strbuf_release(&pathbuf);
302 return hit;
305 static int grep_object(struct grep_opt *opt, const char **paths,
306 struct object *obj, const char *name)
308 if (obj->type == OBJ_BLOB)
309 return grep_sha1(opt, obj->sha1, name, 0);
310 if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
311 struct tree_desc tree;
312 void *data;
313 unsigned long size;
314 int hit;
315 data = read_object_with_reference(obj->sha1, tree_type,
316 &size, NULL);
317 if (!data)
318 die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
319 init_tree_desc(&tree, data, size);
320 hit = grep_tree(opt, paths, &tree, name, "");
321 free(data);
322 return hit;
324 die("unable to grep from object of type %s", typename(obj->type));
327 static int grep_directory(struct grep_opt *opt, const char **paths)
329 struct dir_struct dir;
330 int i, hit = 0;
332 memset(&dir, 0, sizeof(dir));
333 setup_standard_excludes(&dir);
335 fill_directory(&dir, paths);
336 for (i = 0; i < dir.nr; i++) {
337 hit |= grep_file(opt, dir.entries[i]->name);
338 if (hit && opt->status_only)
339 break;
341 free_grep_patterns(opt);
342 return hit;
345 static int context_callback(const struct option *opt, const char *arg,
346 int unset)
348 struct grep_opt *grep_opt = opt->value;
349 int value;
350 const char *endp;
352 if (unset) {
353 grep_opt->pre_context = grep_opt->post_context = 0;
354 return 0;
356 value = strtol(arg, (char **)&endp, 10);
357 if (*endp) {
358 return error("switch `%c' expects a numerical value",
359 opt->short_name);
361 grep_opt->pre_context = grep_opt->post_context = value;
362 return 0;
365 static int file_callback(const struct option *opt, const char *arg, int unset)
367 struct grep_opt *grep_opt = opt->value;
368 FILE *patterns;
369 int lno = 0;
370 struct strbuf sb = STRBUF_INIT;
372 patterns = fopen(arg, "r");
373 if (!patterns)
374 die_errno("cannot open '%s'", arg);
375 while (strbuf_getline(&sb, patterns, '\n') == 0) {
376 /* ignore empty line like grep does */
377 if (sb.len == 0)
378 continue;
379 append_grep_pattern(grep_opt, strbuf_detach(&sb, NULL), arg,
380 ++lno, GREP_PATTERN);
382 fclose(patterns);
383 strbuf_release(&sb);
384 return 0;
387 static int not_callback(const struct option *opt, const char *arg, int unset)
389 struct grep_opt *grep_opt = opt->value;
390 append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
391 return 0;
394 static int and_callback(const struct option *opt, const char *arg, int unset)
396 struct grep_opt *grep_opt = opt->value;
397 append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
398 return 0;
401 static int open_callback(const struct option *opt, const char *arg, int unset)
403 struct grep_opt *grep_opt = opt->value;
404 append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
405 return 0;
408 static int close_callback(const struct option *opt, const char *arg, int unset)
410 struct grep_opt *grep_opt = opt->value;
411 append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
412 return 0;
415 static int pattern_callback(const struct option *opt, const char *arg,
416 int unset)
418 struct grep_opt *grep_opt = opt->value;
419 append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
420 return 0;
423 static int help_callback(const struct option *opt, const char *arg, int unset)
425 return -1;
428 int cmd_grep(int argc, const char **argv, const char *prefix)
430 int hit = 0;
431 int cached = 0;
432 int seen_dashdash = 0;
433 int external_grep_allowed__ignored;
434 struct grep_opt opt;
435 struct object_array list = { 0, 0, NULL };
436 const char **paths = NULL;
437 int i;
438 int dummy;
439 int nongit = 0, use_index = 1;
440 struct option options[] = {
441 OPT_BOOLEAN(0, "cached", &cached,
442 "search in index instead of in the work tree"),
443 OPT_BOOLEAN(0, "index", &use_index,
444 "--no-index finds in contents not managed by git"),
445 OPT_GROUP(""),
446 OPT_BOOLEAN('v', "invert-match", &opt.invert,
447 "show non-matching lines"),
448 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
449 "case insensitive matching"),
450 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
451 "match patterns only at word boundaries"),
452 OPT_SET_INT('a', "text", &opt.binary,
453 "process binary files as text", GREP_BINARY_TEXT),
454 OPT_SET_INT('I', NULL, &opt.binary,
455 "don't match patterns in binary files",
456 GREP_BINARY_NOMATCH),
457 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
458 "descend at most <depth> levels", PARSE_OPT_NONEG,
459 NULL, 1 },
460 OPT_GROUP(""),
461 OPT_BIT('E', "extended-regexp", &opt.regflags,
462 "use extended POSIX regular expressions", REG_EXTENDED),
463 OPT_NEGBIT('G', "basic-regexp", &opt.regflags,
464 "use basic POSIX regular expressions (default)",
465 REG_EXTENDED),
466 OPT_BOOLEAN('F', "fixed-strings", &opt.fixed,
467 "interpret patterns as fixed strings"),
468 OPT_GROUP(""),
469 OPT_BOOLEAN('n', NULL, &opt.linenum, "show line numbers"),
470 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
471 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
472 OPT_NEGBIT(0, "full-name", &opt.relative,
473 "show filenames relative to top directory", 1),
474 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
475 "show only filenames instead of matching lines"),
476 OPT_BOOLEAN(0, "name-only", &opt.name_only,
477 "synonym for --files-with-matches"),
478 OPT_BOOLEAN('L', "files-without-match",
479 &opt.unmatch_name_only,
480 "show only the names of files without match"),
481 OPT_BOOLEAN('z', "null", &opt.null_following_name,
482 "print NUL after filenames"),
483 OPT_BOOLEAN('c', "count", &opt.count,
484 "show the number of matches instead of matching lines"),
485 OPT_SET_INT(0, "color", &opt.color, "highlight matches", 1),
486 OPT_GROUP(""),
487 OPT_CALLBACK('C', NULL, &opt, "n",
488 "show <n> context lines before and after matches",
489 context_callback),
490 OPT_INTEGER('B', NULL, &opt.pre_context,
491 "show <n> context lines before matches"),
492 OPT_INTEGER('A', NULL, &opt.post_context,
493 "show <n> context lines after matches"),
494 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
495 context_callback),
496 OPT_BOOLEAN('p', "show-function", &opt.funcname,
497 "show a line with the function name before matches"),
498 OPT_GROUP(""),
499 OPT_CALLBACK('f', NULL, &opt, "file",
500 "read patterns from file", file_callback),
501 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
502 "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
503 { OPTION_CALLBACK, 0, "and", &opt, NULL,
504 "combine patterns specified with -e",
505 PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
506 OPT_BOOLEAN(0, "or", &dummy, ""),
507 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
508 PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
509 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
510 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
511 open_callback },
512 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
513 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
514 close_callback },
515 OPT_BOOLEAN('q', "quick", &opt.status_only,
516 "indicate hit with exit status without output"),
517 OPT_BOOLEAN(0, "all-match", &opt.all_match,
518 "show only matches from files that match all patterns"),
519 OPT_GROUP(""),
520 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
521 "allow calling of grep(1) (ignored by this build)"),
522 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
523 PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
524 OPT_END()
527 prefix = setup_git_directory_gently(&nongit);
530 * 'git grep -h', unlike 'git grep -h <pattern>', is a request
531 * to show usage information and exit.
533 if (argc == 2 && !strcmp(argv[1], "-h"))
534 usage_with_options(grep_usage, options);
536 memset(&opt, 0, sizeof(opt));
537 opt.prefix = prefix;
538 opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
539 opt.relative = 1;
540 opt.pathname = 1;
541 opt.pattern_tail = &opt.pattern_list;
542 opt.regflags = REG_NEWLINE;
543 opt.max_depth = -1;
545 strcpy(opt.color_match, GIT_COLOR_RED GIT_COLOR_BOLD);
546 opt.color = -1;
547 git_config(grep_config, &opt);
548 if (opt.color == -1)
549 opt.color = git_use_color_default;
552 * If there is no -- then the paths must exist in the working
553 * tree. If there is no explicit pattern specified with -e or
554 * -f, we take the first unrecognized non option to be the
555 * pattern, but then what follows it must be zero or more
556 * valid refs up to the -- (if exists), and then existing
557 * paths. If there is an explicit pattern, then the first
558 * unrecognized non option is the beginning of the refs list
559 * that continues up to the -- (if exists), and then paths.
561 argc = parse_options(argc, argv, prefix, options, grep_usage,
562 PARSE_OPT_KEEP_DASHDASH |
563 PARSE_OPT_STOP_AT_NON_OPTION |
564 PARSE_OPT_NO_INTERNAL_HELP);
566 if (use_index && nongit)
567 /* die the same way as if we did it at the beginning */
568 setup_git_directory();
570 /* First unrecognized non-option token */
571 if (argc > 0 && !opt.pattern_list) {
572 append_grep_pattern(&opt, argv[0], "command line", 0,
573 GREP_PATTERN);
574 argv++;
575 argc--;
578 if (!opt.pattern_list)
579 die("no pattern given.");
580 if (!opt.fixed && opt.ignore_case)
581 opt.regflags |= REG_ICASE;
582 if ((opt.regflags != REG_NEWLINE) && opt.fixed)
583 die("cannot mix --fixed-strings and regexp");
584 compile_grep_patterns(&opt);
586 /* Check revs and then paths */
587 for (i = 0; i < argc; i++) {
588 const char *arg = argv[i];
589 unsigned char sha1[20];
590 /* Is it a rev? */
591 if (!get_sha1(arg, sha1)) {
592 struct object *object = parse_object(sha1);
593 if (!object)
594 die("bad object %s", arg);
595 add_object_array(object, arg, &list);
596 continue;
598 if (!strcmp(arg, "--")) {
599 i++;
600 seen_dashdash = 1;
602 break;
605 /* The rest are paths */
606 if (!seen_dashdash) {
607 int j;
608 for (j = i; j < argc; j++)
609 verify_filename(prefix, argv[j]);
612 if (i < argc)
613 paths = get_pathspec(prefix, argv + i);
614 else if (prefix) {
615 paths = xcalloc(2, sizeof(const char *));
616 paths[0] = prefix;
617 paths[1] = NULL;
620 if (!use_index) {
621 if (cached)
622 die("--cached cannot be used with --no-index.");
623 if (list.nr)
624 die("--no-index cannot be used with revs.");
625 return !grep_directory(&opt, paths);
628 if (!list.nr) {
629 if (!cached)
630 setup_work_tree();
631 return !grep_cache(&opt, paths, cached);
634 if (cached)
635 die("both --cached and trees are given.");
637 for (i = 0; i < list.nr; i++) {
638 struct object *real_obj;
639 real_obj = deref_tag(list.objects[i].item, NULL, 0);
640 if (grep_object(&opt, paths, real_obj, list.objects[i].name)) {
641 hit = 1;
642 if (opt.status_only)
643 break;
646 free_grep_patterns(&opt);
647 return !hit;