4 * Copyright (c) 2006 Junio C Hamano
11 #include "tree-walk.h"
16 * git grep pathspecs are somewhat different from diff-tree pathspecs;
17 * pathname wildcards are allowed.
19 static int pathspec_matches(const char **paths, const char *name)
22 if (!paths || !*paths)
24 namelen = strlen(name);
25 for (i = 0; paths[i]; i++) {
26 const char *match = paths[i];
27 int matchlen = strlen(match);
28 const char *cp, *meta;
31 ((matchlen <= namelen) &&
32 !strncmp(name, match, matchlen) &&
33 (match[matchlen-1] == '/' ||
34 name[matchlen] == '\0' || name[matchlen] == '/')))
36 if (!fnmatch(match, name, 0))
38 if (name[namelen-1] != '/')
41 /* We are being asked if the directory ("name") is worth
44 * Find the longest leading directory name that does
45 * not have metacharacter in the pathspec; the name
46 * we are looking at must overlap with that directory.
48 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
50 if (ch == '*' || ch == '[' || ch == '?') {
56 meta = cp; /* fully literal */
58 if (namelen <= meta - match) {
59 /* Looking at "Documentation/" and
60 * the pattern says "Documentation/howto/", or
61 * "Documentation/diff*.txt". The name we
62 * have should match prefix.
64 if (!memcmp(match, name, namelen))
69 if (meta - match < namelen) {
70 /* Looking at "Documentation/howto/" and
71 * the pattern says "Documentation/h*";
72 * match up to "Do.../h"; this avoids descending
73 * into "Documentation/technical/".
75 if (!memcmp(match, name, meta - match))
83 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name, int tree_name_len)
91 data = read_sha1_file(sha1, type, &size);
93 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
96 if (opt->relative && opt->prefix_length) {
97 static char name_buf[PATH_MAX];
99 int name_len = strlen(name) - opt->prefix_length + 1;
102 name += opt->prefix_length;
104 if (ARRAY_SIZE(name_buf) <= name_len)
105 cp = to_free = xmalloc(name_len);
108 memcpy(cp, name, tree_name_len);
109 strcpy(cp + tree_name_len,
110 name + tree_name_len + opt->prefix_length);
114 hit = grep_buffer(opt, name, data, size);
120 static int grep_file(struct grep_opt *opt, const char *filename)
125 if (lstat(filename, &st) < 0) {
128 error("'%s': %s", filename, strerror(errno));
132 return 0; /* empty file -- no grep hit */
133 if (!S_ISREG(st.st_mode))
135 i = open(filename, O_RDONLY);
138 data = xmalloc(st.st_size + 1);
139 if (st.st_size != read_in_full(i, data, st.st_size)) {
140 error("'%s': short read %s", filename, strerror(errno));
146 if (opt->relative && opt->prefix_length)
147 filename += opt->prefix_length;
148 i = grep_buffer(opt, filename, data, st.st_size);
153 static int exec_grep(int argc, const char **argv)
163 execvp("grep", (char **) argv);
166 while (waitpid(pid, &status, 0) < 0) {
171 if (WIFEXITED(status)) {
172 if (!WEXITSTATUS(status))
181 #define push_arg(a) do { \
182 if (nr < MAXARGS) argv[nr++] = (a); \
183 else die("maximum number of args exceeded"); \
186 static int external_grep(struct grep_opt *opt, const char **paths, int cached)
188 int i, nr, argc, hit, len, status;
189 const char *argv[MAXARGS+1];
190 char randarg[ARGBUF];
191 char *argptr = randarg;
194 if (opt->extended || (opt->relative && opt->prefix_length))
204 if (opt->regflags & REG_EXTENDED)
206 if (opt->regflags & REG_ICASE)
208 if (opt->word_regexp)
212 if (opt->unmatch_name_only)
216 if (opt->post_context || opt->pre_context) {
217 if (opt->post_context != opt->pre_context) {
218 if (opt->pre_context) {
220 len += snprintf(argptr, sizeof(randarg)-len,
221 "%u", opt->pre_context);
222 if (sizeof(randarg) <= len)
223 die("maximum length of args exceeded");
227 if (opt->post_context) {
229 len += snprintf(argptr, sizeof(randarg)-len,
230 "%u", opt->post_context);
231 if (sizeof(randarg) <= len)
232 die("maximum length of args exceeded");
239 len += snprintf(argptr, sizeof(randarg)-len,
240 "%u", opt->post_context);
241 if (sizeof(randarg) <= len)
242 die("maximum length of args exceeded");
247 for (p = opt->pattern_list; p; p = p->next) {
249 push_arg(p->pattern);
253 * To make sure we get the header printed out when we want it,
254 * add /dev/null to the paths to grep. This is unnecessary
255 * (and wrong) with "-l" or "-L", which always print out the
258 * GNU grep has "-H", but this is portable.
260 if (!opt->name_only && !opt->unmatch_name_only)
261 push_arg("/dev/null");
265 for (i = 0; i < active_nr; i++) {
266 struct cache_entry *ce = active_cache[i];
268 if (!S_ISREG(ntohl(ce->ce_mode)))
270 if (!pathspec_matches(paths, ce->name))
273 if (name[0] == '-') {
274 int len = ce_namelen(ce);
275 name = xmalloc(len + 3);
276 memcpy(name, "./", 2);
277 memcpy(name + 2, ce->name, len + 1);
280 if (argc < MAXARGS && !ce_stage(ce))
282 status = exec_grep(argc, argv);
289 } while (i < active_nr &&
290 !strcmp(ce->name, active_cache[i]->name));
291 i--; /* compensate for loop control */
295 status = exec_grep(argc, argv);
302 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
310 * Use the external "grep" command for the case where
311 * we grep through the checked-out files. It tends to
312 * be a lot more optimized
315 hit = external_grep(opt, paths, cached);
321 for (nr = 0; nr < active_nr; nr++) {
322 struct cache_entry *ce = active_cache[nr];
323 if (!S_ISREG(ntohl(ce->ce_mode)))
325 if (!pathspec_matches(paths, ce->name))
330 hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
333 hit |= grep_file(opt, ce->name);
337 } while (nr < active_nr &&
338 !strcmp(ce->name, active_cache[nr]->name));
339 nr--; /* compensate for loop control */
342 free_grep_patterns(opt);
346 static int grep_tree(struct grep_opt *opt, const char **paths,
347 struct tree_desc *tree,
348 const char *tree_name, const char *base)
352 struct name_entry entry;
354 int tn_len = strlen(tree_name);
355 char *path_buf = xmalloc(PATH_MAX + tn_len + 100);
358 tn_len = sprintf(path_buf, "%s:", tree_name);
359 down = path_buf + tn_len;
366 len = strlen(path_buf);
368 while (tree_entry(tree, &entry)) {
369 strcpy(path_buf + len, entry.path);
371 if (S_ISDIR(entry.mode))
372 /* Match "abc/" against pathspec to
373 * decide if we want to descend into "abc"
376 strcpy(path_buf + len + entry.pathlen, "/");
378 if (!pathspec_matches(paths, down))
380 else if (S_ISREG(entry.mode))
381 hit |= grep_sha1(opt, entry.sha1, path_buf, tn_len);
382 else if (S_ISDIR(entry.mode)) {
384 struct tree_desc sub;
386 data = read_sha1_file(entry.sha1, type, &sub.size);
388 die("unable to read tree (%s)",
389 sha1_to_hex(entry.sha1));
391 hit |= grep_tree(opt, paths, &sub, tree_name, down);
398 static int grep_object(struct grep_opt *opt, const char **paths,
399 struct object *obj, const char *name)
401 if (obj->type == OBJ_BLOB)
402 return grep_sha1(opt, obj->sha1, name, 0);
403 if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
404 struct tree_desc tree;
407 data = read_object_with_reference(obj->sha1, tree_type,
410 die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
412 hit = grep_tree(opt, paths, &tree, name, "");
416 die("unable to grep from object of type %s", typename(obj->type));
419 static const char builtin_grep_usage[] =
420 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
422 static const char emsg_invalid_context_len[] =
423 "%s: invalid context length argument";
424 static const char emsg_missing_context_len[] =
425 "missing context length argument";
426 static const char emsg_missing_argument[] =
427 "option requires an argument -%s";
429 int cmd_grep(int argc, const char **argv, const char *prefix)
433 int seen_dashdash = 0;
435 struct object_array list = { 0, 0, NULL };
436 const char **paths = NULL;
439 memset(&opt, 0, sizeof(opt));
440 opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
443 opt.pattern_tail = &opt.pattern_list;
444 opt.regflags = REG_NEWLINE;
447 * If there is no -- then the paths must exist in the working
448 * tree. If there is no explicit pattern specified with -e or
449 * -f, we take the first unrecognized non option to be the
450 * pattern, but then what follows it must be zero or more
451 * valid refs up to the -- (if exists), and then existing
452 * paths. If there is an explicit pattern, then the first
453 * unrecognized non option is the beginning of the refs list
454 * that continues up to the -- (if exists), and then paths.
458 const char *arg = argv[1];
460 if (!strcmp("--cached", arg)) {
464 if (!strcmp("-a", arg) ||
465 !strcmp("--text", arg)) {
466 opt.binary = GREP_BINARY_TEXT;
469 if (!strcmp("-i", arg) ||
470 !strcmp("--ignore-case", arg)) {
471 opt.regflags |= REG_ICASE;
474 if (!strcmp("-I", arg)) {
475 opt.binary = GREP_BINARY_NOMATCH;
478 if (!strcmp("-v", arg) ||
479 !strcmp("--invert-match", arg)) {
483 if (!strcmp("-E", arg) ||
484 !strcmp("--extended-regexp", arg)) {
485 opt.regflags |= REG_EXTENDED;
488 if (!strcmp("-F", arg) ||
489 !strcmp("--fixed-strings", arg)) {
493 if (!strcmp("-G", arg) ||
494 !strcmp("--basic-regexp", arg)) {
495 opt.regflags &= ~REG_EXTENDED;
498 if (!strcmp("-n", arg)) {
502 if (!strcmp("-h", arg)) {
506 if (!strcmp("-H", arg)) {
510 if (!strcmp("-l", arg) ||
511 !strcmp("--files-with-matches", arg)) {
515 if (!strcmp("-L", arg) ||
516 !strcmp("--files-without-match", arg)) {
517 opt.unmatch_name_only = 1;
520 if (!strcmp("-c", arg) ||
521 !strcmp("--count", arg)) {
525 if (!strcmp("-w", arg) ||
526 !strcmp("--word-regexp", arg)) {
530 if (!strncmp("-A", arg, 2) ||
531 !strncmp("-B", arg, 2) ||
532 !strncmp("-C", arg, 2) ||
533 (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
537 case 'A': case 'B': case 'C':
540 die(emsg_missing_context_len);
551 if (sscanf(scan, "%u", &num) != 1)
552 die(emsg_invalid_context_len, scan);
555 opt.post_context = num;
559 opt.post_context = num;
561 opt.pre_context = num;
566 if (!strcmp("-f", arg)) {
571 die(emsg_missing_argument, arg);
572 patterns = fopen(argv[1], "r");
574 die("'%s': %s", argv[1], strerror(errno));
575 while (fgets(buf, sizeof(buf), patterns)) {
576 int len = strlen(buf);
577 if (buf[len-1] == '\n')
579 /* ignore empty line like grep does */
582 append_grep_pattern(&opt, xstrdup(buf),
591 if (!strcmp("--not", arg)) {
592 append_grep_pattern(&opt, arg, "command line", 0,
596 if (!strcmp("--and", arg)) {
597 append_grep_pattern(&opt, arg, "command line", 0,
601 if (!strcmp("--or", arg))
602 continue; /* no-op */
603 if (!strcmp("(", arg)) {
604 append_grep_pattern(&opt, arg, "command line", 0,
608 if (!strcmp(")", arg)) {
609 append_grep_pattern(&opt, arg, "command line", 0,
613 if (!strcmp("--all-match", arg)) {
617 if (!strcmp("-e", arg)) {
619 append_grep_pattern(&opt, argv[1],
626 die(emsg_missing_argument, arg);
628 if (!strcmp("--full-name", arg)) {
632 if (!strcmp("--", arg)) {
633 /* later processing wants to have this at argv[1] */
639 usage(builtin_grep_usage);
641 /* First unrecognized non-option token */
642 if (!opt.pattern_list) {
643 append_grep_pattern(&opt, arg, "command line", 0,
648 /* We are looking at the first path or rev;
649 * it is found at argv[1] after leaving the
657 if (!opt.pattern_list)
658 die("no pattern given.");
659 if ((opt.regflags != REG_NEWLINE) && opt.fixed)
660 die("cannot mix --fixed-strings and regexp");
661 compile_grep_patterns(&opt);
663 /* Check revs and then paths */
664 for (i = 1; i < argc; i++) {
665 const char *arg = argv[i];
666 unsigned char sha1[20];
668 if (!get_sha1(arg, sha1)) {
669 struct object *object = parse_object(sha1);
671 die("bad object %s", arg);
672 add_object_array(object, arg, &list);
675 if (!strcmp(arg, "--")) {
682 /* The rest are paths */
683 if (!seen_dashdash) {
685 for (j = i; j < argc; j++)
686 verify_filename(prefix, argv[j]);
690 paths = get_pathspec(prefix, argv + i);
691 if (opt.prefix_length && opt.relative) {
692 /* Make sure we do not get outside of paths */
693 for (i = 0; paths[i]; i++)
694 if (strncmp(prefix, paths[i], opt.prefix_length))
695 die("git-grep: cannot generate relative filenames containing '..'");
699 paths = xcalloc(2, sizeof(const char *));
705 return !grep_cache(&opt, paths, cached);
708 die("both --cached and trees are given.");
710 for (i = 0; i < list.nr; i++) {
711 struct object *real_obj;
712 real_obj = deref_tag(list.objects[i].item, NULL, 0);
713 if (grep_object(&opt, paths, real_obj, list.objects[i].name))
716 free_grep_patterns(&opt);