4 * Copyright (c) 2006 Junio C Hamano
11 #include "tree-walk.h"
18 * git grep pathspecs are somewhat different from diff-tree pathspecs;
19 * pathname wildcards are allowed.
21 static int pathspec_matches(const char **paths, const char *name)
24 if (!paths || !*paths)
26 namelen = strlen(name);
27 for (i = 0; paths[i]; i++) {
28 const char *match = paths[i];
29 int matchlen = strlen(match);
30 const char *cp, *meta;
32 if ((matchlen <= namelen) &&
33 !strncmp(name, match, matchlen) &&
34 (match[matchlen-1] == '/' ||
35 name[matchlen] == '\0' || name[matchlen] == '/'))
37 if (!fnmatch(match, name, 0))
39 if (name[namelen-1] != '/')
42 /* We are being asked if the directory ("name") is worth
45 * Find the longest leading directory name that does
46 * not have metacharacter in the pathspec; the name
47 * we are looking at must overlap with that directory.
49 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
51 if (ch == '*' || ch == '[' || ch == '?') {
57 meta = cp; /* fully literal */
59 if (namelen <= meta - match) {
60 /* Looking at "Documentation/" and
61 * the pattern says "Documentation/howto/", or
62 * "Documentation/diff*.txt". The name we
63 * have should match prefix.
65 if (!memcmp(match, name, namelen))
70 if (meta - match < namelen) {
71 /* Looking at "Documentation/howto/" and
72 * the pattern says "Documentation/h*";
73 * match up to "Do.../h"; this avoids descending
74 * into "Documentation/technical/".
76 if (!memcmp(match, name, meta - match))
85 struct grep_pat *next;
93 struct grep_pat *pattern_list;
94 struct grep_pat **pattern_tail;
99 unsigned unmatch_name_only:1;
101 unsigned word_regexp:1;
103 #define GREP_BINARY_DEFAULT 0
104 #define GREP_BINARY_NOMATCH 1
105 #define GREP_BINARY_TEXT 2
108 unsigned pre_context;
109 unsigned post_context;
112 static void add_pattern(struct grep_opt *opt, const char *pat,
113 const char *origin, int no)
115 struct grep_pat *p = xcalloc(1, sizeof(*p));
119 *opt->pattern_tail = p;
120 opt->pattern_tail = &p->next;
124 static void compile_patterns(struct grep_opt *opt)
127 for (p = opt->pattern_list; p; p = p->next) {
128 int err = regcomp(&p->regexp, p->pattern, opt->regflags);
133 sprintf(where, "In '%s' at %d, ",
136 sprintf(where, "%s, ", p->origin);
139 regerror(err, &p->regexp, errbuf, 1024);
141 die("%s'%s': %s", where, p->pattern, errbuf);
146 static char *end_of_line(char *cp, unsigned long *left)
148 unsigned long l = *left;
149 while (l && *cp != '\n') {
157 static int word_char(char ch)
159 return isalnum(ch) || ch == '_';
162 static void show_line(struct grep_opt *opt, const char *bol, const char *eol,
163 const char *name, unsigned lno, char sign)
165 printf("%s%c", name, sign);
167 printf("%d%c", lno, sign);
168 printf("%.*s\n", (int)(eol-bol), bol);
172 * NEEDSWORK: share code with diff.c
174 #define FIRST_FEW_BYTES 8000
175 static int buffer_is_binary(const char *ptr, unsigned long size)
177 if (FIRST_FEW_BYTES < size)
178 size = FIRST_FEW_BYTES;
179 if (memchr(ptr, 0, size))
184 static int fixmatch(const char *pattern, char *line, regmatch_t *match)
186 char *hit = strstr(line, pattern);
188 match->rm_so = match->rm_eo = -1;
192 match->rm_so = hit - line;
193 match->rm_eo = match->rm_so + strlen(pattern);
198 static int grep_buffer(struct grep_opt *opt, const char *name,
199 char *buf, unsigned long size)
202 unsigned long left = size;
204 struct pre_context_line {
207 } *prev = NULL, *pcl;
208 unsigned last_hit = 0;
209 unsigned last_shown = 0;
210 int binary_match_only = 0;
211 const char *hunk_mark = "";
214 if (buffer_is_binary(buf, size)) {
215 switch (opt->binary) {
216 case GREP_BINARY_DEFAULT:
217 binary_match_only = 1;
219 case GREP_BINARY_NOMATCH:
220 return 0; /* Assume unmatch */
227 if (opt->pre_context)
228 prev = xcalloc(opt->pre_context, sizeof(*prev));
229 if (opt->pre_context || opt->post_context)
233 regmatch_t pmatch[10];
238 eol = end_of_line(bol, &left);
242 for (p = opt->pattern_list; p; p = p->next) {
244 regex_t *exp = &p->regexp;
245 hit = !regexec(exp, bol, ARRAY_SIZE(pmatch),
249 hit = !fixmatch(p->pattern, bol, pmatch);
252 if (hit && opt->word_regexp) {
253 /* Match beginning must be either
254 * beginning of the line, or at word
255 * boundary (i.e. the last char must
256 * not be alnum or underscore).
258 if ((pmatch[0].rm_so < 0) ||
259 (eol - bol) <= pmatch[0].rm_so ||
260 (pmatch[0].rm_eo < 0) ||
261 (eol - bol) < pmatch[0].rm_eo)
262 die("regexp returned nonsense");
263 if (pmatch[0].rm_so != 0 &&
264 word_char(bol[pmatch[0].rm_so-1]))
266 if (pmatch[0].rm_eo != (eol-bol) &&
267 word_char(bol[pmatch[0].rm_eo]))
273 /* "grep -v -e foo -e bla" should list lines
274 * that do not have either, so inversion should
279 if (opt->unmatch_name_only) {
286 if (binary_match_only) {
287 printf("Binary file %s matches\n", name);
290 if (opt->name_only) {
291 printf("%s\n", name);
294 /* Hit at this line. If we haven't shown the
295 * pre-context lines, we would need to show them.
296 * When asked to do "count", this still show
297 * the context which is nonsense, but the user
298 * deserves to get that ;-).
300 if (opt->pre_context) {
302 if (opt->pre_context < lno)
303 from = lno - opt->pre_context;
306 if (from <= last_shown)
307 from = last_shown + 1;
308 if (last_shown && from != last_shown + 1)
311 pcl = &prev[lno-from-1];
312 show_line(opt, pcl->bol, pcl->eol,
318 if (last_shown && lno != last_shown + 1)
321 show_line(opt, bol, eol, name, lno, ':');
322 last_shown = last_hit = lno;
325 lno <= last_hit + opt->post_context) {
326 /* If the last hit is within the post context,
327 * we need to show this line.
329 if (last_shown && lno != last_shown + 1)
331 show_line(opt, bol, eol, name, lno, '-');
334 if (opt->pre_context) {
335 memmove(prev+1, prev,
336 (opt->pre_context-1) * sizeof(*prev));
350 if (opt->unmatch_name_only) {
351 /* We did not see any hit, so we want to show this */
352 printf("%s\n", name);
357 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
358 * which feels mostly useless but sometimes useful. Maybe
359 * make it another option? For now suppress them.
361 if (opt->count && count)
362 printf("%s:%u\n", name, count);
366 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name)
372 data = read_sha1_file(sha1, type, &size);
374 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
377 hit = grep_buffer(opt, name, data, size);
382 static int grep_file(struct grep_opt *opt, const char *filename)
387 if (lstat(filename, &st) < 0) {
390 error("'%s': %s", filename, strerror(errno));
394 return 0; /* empty file -- no grep hit */
395 if (!S_ISREG(st.st_mode))
397 i = open(filename, O_RDONLY);
400 data = xmalloc(st.st_size + 1);
401 if (st.st_size != xread(i, data, st.st_size)) {
402 error("'%s': short read %s", filename, strerror(errno));
408 i = grep_buffer(opt, filename, data, st.st_size);
413 static int exec_grep(int argc, const char **argv)
423 execvp("grep", (char **) argv);
426 while (waitpid(pid, &status, 0) < 0) {
431 if (WIFEXITED(status)) {
432 if (!WEXITSTATUS(status))
441 #define push_arg(a) do { \
442 if (nr < MAXARGS) argv[nr++] = (a); \
443 else die("maximum number of args exceeded"); \
446 static int external_grep(struct grep_opt *opt, const char **paths, int cached)
448 int i, nr, argc, hit, len;
449 const char *argv[MAXARGS+1];
450 char randarg[ARGBUF];
451 char *argptr = randarg;
461 if (opt->regflags & REG_EXTENDED)
463 if (opt->word_regexp)
467 if (opt->unmatch_name_only)
471 if (opt->post_context || opt->pre_context) {
472 if (opt->post_context != opt->pre_context) {
473 if (opt->pre_context) {
475 len += snprintf(argptr, sizeof(randarg)-len,
476 "%u", opt->pre_context);
477 if (sizeof(randarg) <= len)
478 die("maximum length of args exceeded");
482 if (opt->post_context) {
484 len += snprintf(argptr, sizeof(randarg)-len,
485 "%u", opt->post_context);
486 if (sizeof(randarg) <= len)
487 die("maximum length of args exceeded");
494 len += snprintf(argptr, sizeof(randarg)-len,
495 "%u", opt->post_context);
496 if (sizeof(randarg) <= len)
497 die("maximum length of args exceeded");
502 for (p = opt->pattern_list; p; p = p->next) {
504 push_arg(p->pattern);
510 for (i = 0; i < active_nr; i++) {
511 struct cache_entry *ce = active_cache[i];
512 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
514 if (!pathspec_matches(paths, ce->name))
516 argv[argc++] = ce->name;
519 hit += exec_grep(argc, argv);
523 hit += exec_grep(argc, argv);
527 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
535 * Use the external "grep" command for the case where
536 * we grep through the checked-out files. It tends to
537 * be a lot more optimized
540 hit = external_grep(opt, paths, cached);
546 for (nr = 0; nr < active_nr; nr++) {
547 struct cache_entry *ce = active_cache[nr];
548 if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
550 if (!pathspec_matches(paths, ce->name))
553 hit |= grep_sha1(opt, ce->sha1, ce->name);
555 hit |= grep_file(opt, ce->name);
560 static int grep_tree(struct grep_opt *opt, const char **paths,
561 struct tree_desc *tree,
562 const char *tree_name, const char *base)
568 const unsigned char *sha1;
570 char *path_buf = xmalloc(PATH_MAX + strlen(tree_name) + 100);
573 int offset = sprintf(path_buf, "%s:", tree_name);
574 down = path_buf + offset;
581 len = strlen(path_buf);
585 sha1 = tree_entry_extract(tree, &path, &mode);
586 pathlen = strlen(path);
587 strcpy(path_buf + len, path);
590 /* Match "abc/" against pathspec to
591 * decide if we want to descend into "abc"
594 strcpy(path_buf + len + pathlen, "/");
596 if (!pathspec_matches(paths, down))
598 else if (S_ISREG(mode))
599 hit |= grep_sha1(opt, sha1, path_buf);
600 else if (S_ISDIR(mode)) {
602 struct tree_desc sub;
604 data = read_sha1_file(sha1, type, &sub.size);
606 die("unable to read tree (%s)",
609 hit |= grep_tree(opt, paths, &sub, tree_name, down);
612 update_tree_entry(tree);
617 static int grep_object(struct grep_opt *opt, const char **paths,
618 struct object *obj, const char *name)
620 if (!strcmp(obj->type, blob_type))
621 return grep_sha1(opt, obj->sha1, name);
622 if (!strcmp(obj->type, commit_type) ||
623 !strcmp(obj->type, tree_type)) {
624 struct tree_desc tree;
627 data = read_object_with_reference(obj->sha1, tree_type,
630 die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
632 hit = grep_tree(opt, paths, &tree, name, "");
636 die("unable to grep from object of type %s", obj->type);
639 static const char builtin_grep_usage[] =
640 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
642 int cmd_grep(int argc, const char **argv, char **envp)
646 int seen_dashdash = 0;
648 struct object_list *list, **tail, *object_list = NULL;
649 const char *prefix = setup_git_directory();
650 const char **paths = NULL;
653 memset(&opt, 0, sizeof(opt));
654 opt.pattern_tail = &opt.pattern_list;
655 opt.regflags = REG_NEWLINE;
658 * If there is no -- then the paths must exist in the working
659 * tree. If there is no explicit pattern specified with -e or
660 * -f, we take the first unrecognized non option to be the
661 * pattern, but then what follows it must be zero or more
662 * valid refs up to the -- (if exists), and then existing
663 * paths. If there is an explicit pattern, then the first
664 * unrecocnized non option is the beginning of the refs list
665 * that continues up to the -- (if exists), and then paths.
670 const char *arg = argv[1];
672 if (!strcmp("--cached", arg)) {
676 if (!strcmp("-a", arg) ||
677 !strcmp("--text", arg)) {
678 opt.binary = GREP_BINARY_TEXT;
681 if (!strcmp("-i", arg) ||
682 !strcmp("--ignore-case", arg)) {
683 opt.regflags |= REG_ICASE;
686 if (!strcmp("-I", arg)) {
687 opt.binary = GREP_BINARY_NOMATCH;
690 if (!strcmp("-v", arg) ||
691 !strcmp("--invert-match", arg)) {
695 if (!strcmp("-E", arg) ||
696 !strcmp("--extended-regexp", arg)) {
697 opt.regflags |= REG_EXTENDED;
700 if (!strcmp("-F", arg) ||
701 !strcmp("--fixed-strings", arg)) {
705 if (!strcmp("-G", arg) ||
706 !strcmp("--basic-regexp", arg)) {
707 opt.regflags &= ~REG_EXTENDED;
710 if (!strcmp("-n", arg)) {
714 if (!strcmp("-H", arg)) {
715 /* We always show the pathname, so this
720 if (!strcmp("-l", arg) ||
721 !strcmp("--files-with-matches", arg)) {
725 if (!strcmp("-L", arg) ||
726 !strcmp("--files-without-match", arg)) {
727 opt.unmatch_name_only = 1;
730 if (!strcmp("-c", arg) ||
731 !strcmp("--count", arg)) {
735 if (!strcmp("-w", arg) ||
736 !strcmp("--word-regexp", arg)) {
740 if (!strncmp("-A", arg, 2) ||
741 !strncmp("-B", arg, 2) ||
742 !strncmp("-C", arg, 2) ||
743 (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
747 case 'A': case 'B': case 'C':
750 usage(builtin_grep_usage);
761 if (sscanf(scan, "%u", &num) != 1)
762 usage(builtin_grep_usage);
765 opt.post_context = num;
769 opt.post_context = num;
771 opt.pre_context = num;
776 if (!strcmp("-f", arg)) {
781 usage(builtin_grep_usage);
782 patterns = fopen(argv[1], "r");
784 die("'%s': %s", argv[1], strerror(errno));
785 while (fgets(buf, sizeof(buf), patterns)) {
786 int len = strlen(buf);
787 if (buf[len-1] == '\n')
789 /* ignore empty line like grep does */
792 add_pattern(&opt, strdup(buf), argv[1], ++lno);
799 if (!strcmp("-e", arg)) {
801 add_pattern(&opt, argv[1], "-e option", 0);
806 usage(builtin_grep_usage);
808 if (!strcmp("--", arg))
811 usage(builtin_grep_usage);
813 /* First unrecognized non-option token */
814 if (!opt.pattern_list) {
815 add_pattern(&opt, arg, "command line", 0);
819 /* We are looking at the first path or rev;
820 * it is found at argv[1] after leaving the
828 if (!opt.pattern_list)
829 die("no pattern given.");
830 if ((opt.regflags != REG_NEWLINE) && opt.fixed)
831 die("cannot mix --fixed-strings and regexp");
833 compile_patterns(&opt);
835 /* Check revs and then paths */
836 for (i = 1; i < argc; i++) {
837 const char *arg = argv[i];
838 unsigned char sha1[20];
840 if (!get_sha1(arg, sha1)) {
841 struct object *object = parse_object(sha1);
842 struct object_list *elem;
844 die("bad object %s", arg);
845 elem = object_list_insert(object, tail);
850 if (!strcmp(arg, "--")) {
857 /* The rest are paths */
858 if (!seen_dashdash) {
860 for (j = i; j < argc; j++)
861 verify_filename(prefix, argv[j]);
865 paths = get_pathspec(prefix, argv + i);
867 paths = xcalloc(2, sizeof(const char *));
873 return !grep_cache(&opt, paths, cached);
876 die("both --cached and trees are given.");
878 for (list = object_list; list; list = list->next) {
879 struct object *real_obj;
880 real_obj = deref_tag(list->item, NULL, 0);
881 if (grep_object(&opt, paths, real_obj, list->name))