4 * Copyright (c) 2006 Junio C Hamano
11 #include "tree-walk.h"
13 #include "parse-options.h"
19 #include "thread-utils.h"
23 static char const * const grep_usage[] = {
24 "git grep [options] [-e] <pattern> [<rev>...] [[--] path...]",
28 static int use_threads = 1;
32 static pthread_t threads[THREADS];
34 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
36 static void *load_file(const char *filename, size_t *sz);
38 enum work_type {WORK_SHA1, WORK_FILE};
40 /* We use one producer thread and THREADS consumer
41 * threads. The producer adds struct work_items to 'todo' and the
42 * consumers pick work items from the same array.
49 /* if type == WORK_SHA1, then 'identifier' is a SHA1,
50 * otherwise type == WORK_FILE, and 'identifier' is a NUL
51 * terminated filename.
58 /* In the range [todo_done, todo_start) in 'todo' we have work_items
59 * that have been or are processed by a consumer thread. We haven't
60 * written the result for these to stdout yet.
62 * The work_items in [todo_start, todo_end) are waiting to be picked
63 * up by a consumer thread.
65 * The ranges are modulo TODO_SIZE.
68 static struct work_item todo[TODO_SIZE];
69 static int todo_start;
73 /* Has all work items been added? */
74 static int all_work_added;
76 /* This lock protects all the variables above. */
77 static pthread_mutex_t grep_mutex;
79 /* Used to serialize calls to read_sha1_file. */
80 static pthread_mutex_t read_sha1_mutex;
82 #define grep_lock() pthread_mutex_lock(&grep_mutex)
83 #define grep_unlock() pthread_mutex_unlock(&grep_mutex)
84 #define read_sha1_lock() pthread_mutex_lock(&read_sha1_mutex)
85 #define read_sha1_unlock() pthread_mutex_unlock(&read_sha1_mutex)
87 /* Signalled when a new work_item is added to todo. */
88 static pthread_cond_t cond_add;
90 /* Signalled when the result from one work_item is written to
93 static pthread_cond_t cond_write;
95 /* Signalled when we are finished with everything. */
96 static pthread_cond_t cond_result;
98 static void add_work(enum work_type type, char *name, void *id)
102 while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) {
103 pthread_cond_wait(&cond_write, &grep_mutex);
106 todo[todo_end].type = type;
107 todo[todo_end].name = name;
108 todo[todo_end].identifier = id;
109 todo[todo_end].done = 0;
110 strbuf_reset(&todo[todo_end].out);
111 todo_end = (todo_end + 1) % ARRAY_SIZE(todo);
113 pthread_cond_signal(&cond_add);
117 static struct work_item *get_work(void)
119 struct work_item *ret;
122 while (todo_start == todo_end && !all_work_added) {
123 pthread_cond_wait(&cond_add, &grep_mutex);
126 if (todo_start == todo_end && all_work_added) {
129 ret = &todo[todo_start];
130 todo_start = (todo_start + 1) % ARRAY_SIZE(todo);
136 static void grep_sha1_async(struct grep_opt *opt, char *name,
137 const unsigned char *sha1)
142 add_work(WORK_SHA1, name, s);
145 static void grep_file_async(struct grep_opt *opt, char *name,
146 const char *filename)
148 add_work(WORK_FILE, name, xstrdup(filename));
151 static void work_done(struct work_item *w)
157 old_done = todo_done;
158 for(; todo[todo_done].done && todo_done != todo_start;
159 todo_done = (todo_done+1) % ARRAY_SIZE(todo)) {
160 w = &todo[todo_done];
161 write_or_die(1, w->out.buf, w->out.len);
166 if (old_done != todo_done)
167 pthread_cond_signal(&cond_write);
169 if (all_work_added && todo_done == todo_end)
170 pthread_cond_signal(&cond_result);
175 static void *run(void *arg)
178 struct grep_opt *opt = arg;
181 struct work_item *w = get_work();
185 opt->output_priv = w;
186 if (w->type == WORK_SHA1) {
188 void* data = load_sha1(w->identifier, &sz, w->name);
191 hit |= grep_buffer(opt, w->name, data, sz);
194 } else if (w->type == WORK_FILE) {
196 void* data = load_file(w->identifier, &sz);
198 hit |= grep_buffer(opt, w->name, data, sz);
207 free_grep_patterns(arg);
210 return (void*) (intptr_t) hit;
213 static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size)
215 struct work_item *w = opt->output_priv;
216 strbuf_add(&w->out, buf, size);
219 static void start_threads(struct grep_opt *opt)
223 pthread_mutex_init(&grep_mutex, NULL);
224 pthread_mutex_init(&read_sha1_mutex, NULL);
225 pthread_cond_init(&cond_add, NULL);
226 pthread_cond_init(&cond_write, NULL);
227 pthread_cond_init(&cond_result, NULL);
229 for (i = 0; i < ARRAY_SIZE(todo); i++) {
230 strbuf_init(&todo[i].out, 0);
233 for (i = 0; i < ARRAY_SIZE(threads); i++) {
235 struct grep_opt *o = grep_opt_dup(opt);
236 o->output = strbuf_out;
237 compile_grep_patterns(o);
238 err = pthread_create(&threads[i], NULL, run, o);
241 die("grep: failed to create thread: %s",
246 static int wait_all(void)
254 /* Wait until all work is done. */
255 while (todo_done != todo_end)
256 pthread_cond_wait(&cond_result, &grep_mutex);
258 /* Wake up all the consumer threads so they can see that there
259 * is no more work to do.
261 pthread_cond_broadcast(&cond_add);
264 for (i = 0; i < ARRAY_SIZE(threads); i++) {
266 pthread_join(threads[i], &h);
267 hit |= (int) (intptr_t) h;
270 pthread_mutex_destroy(&grep_mutex);
271 pthread_mutex_destroy(&read_sha1_mutex);
272 pthread_cond_destroy(&cond_add);
273 pthread_cond_destroy(&cond_write);
274 pthread_cond_destroy(&cond_result);
278 #else /* !NO_PTHREADS */
279 #define read_sha1_lock()
280 #define read_sha1_unlock()
282 static int wait_all(void)
288 static int grep_config(const char *var, const char *value, void *cb)
290 struct grep_opt *opt = cb;
292 switch (userdiff_config(var, value)) {
298 if (!strcmp(var, "color.grep")) {
299 opt->color = git_config_colorbool(var, value, -1);
302 if (!strcmp(var, "color.grep.match")) {
304 return config_error_nonbool(var);
305 color_parse(value, var, opt->color_match);
308 return git_color_default_config(var, value, cb);
312 * Return non-zero if max_depth is negative or path has no more then max_depth
315 static int accept_subdir(const char *path, int max_depth)
320 while ((path = strchr(path, '/')) != NULL) {
330 * Return non-zero if name is a subdirectory of match and is not too deep.
332 static int is_subdir(const char *name, int namelen,
333 const char *match, int matchlen, int max_depth)
335 if (matchlen > namelen || strncmp(name, match, matchlen))
338 if (name[matchlen] == '\0') /* exact match */
341 if (!matchlen || match[matchlen-1] == '/' || name[matchlen] == '/')
342 return accept_subdir(name + matchlen + 1, max_depth);
348 * git grep pathspecs are somewhat different from diff-tree pathspecs;
349 * pathname wildcards are allowed.
351 static int pathspec_matches(const char **paths, const char *name, int max_depth)
354 if (!paths || !*paths)
355 return accept_subdir(name, max_depth);
356 namelen = strlen(name);
357 for (i = 0; paths[i]; i++) {
358 const char *match = paths[i];
359 int matchlen = strlen(match);
360 const char *cp, *meta;
362 if (is_subdir(name, namelen, match, matchlen, max_depth))
364 if (!fnmatch(match, name, 0))
366 if (name[namelen-1] != '/')
369 /* We are being asked if the directory ("name") is worth
372 * Find the longest leading directory name that does
373 * not have metacharacter in the pathspec; the name
374 * we are looking at must overlap with that directory.
376 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
378 if (ch == '*' || ch == '[' || ch == '?') {
384 meta = cp; /* fully literal */
386 if (namelen <= meta - match) {
387 /* Looking at "Documentation/" and
388 * the pattern says "Documentation/howto/", or
389 * "Documentation/diff*.txt". The name we
390 * have should match prefix.
392 if (!memcmp(match, name, namelen))
397 if (meta - match < namelen) {
398 /* Looking at "Documentation/howto/" and
399 * the pattern says "Documentation/h*";
400 * match up to "Do.../h"; this avoids descending
401 * into "Documentation/technical/".
403 if (!memcmp(match, name, meta - match))
411 static void *load_sha1(const unsigned char *sha1, unsigned long *size,
414 enum object_type type;
418 data = read_sha1_file(sha1, &type, size);
422 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
427 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1,
428 const char *filename, int tree_name_len)
430 struct strbuf pathbuf = STRBUF_INIT;
433 if (opt->relative && opt->prefix_length) {
434 quote_path_relative(filename + tree_name_len, -1, &pathbuf,
436 strbuf_insert(&pathbuf, 0, filename, tree_name_len);
438 strbuf_addstr(&pathbuf, filename);
441 name = strbuf_detach(&pathbuf, NULL);
445 grep_sha1_async(opt, name, sha1);
452 void *data = load_sha1(sha1, &sz, name);
456 hit = grep_buffer(opt, name, data, sz);
464 static void *load_file(const char *filename, size_t *sz)
470 if (lstat(filename, &st) < 0) {
473 error("'%s': %s", filename, strerror(errno));
476 if (!S_ISREG(st.st_mode))
478 *sz = xsize_t(st.st_size);
479 i = open(filename, O_RDONLY);
482 data = xmalloc(*sz + 1);
483 if (st.st_size != read_in_full(i, data, *sz)) {
484 error("'%s': short read %s", filename, strerror(errno));
494 static int grep_file(struct grep_opt *opt, const char *filename)
496 struct strbuf buf = STRBUF_INIT;
499 if (opt->relative && opt->prefix_length)
500 quote_path_relative(filename, -1, &buf, opt->prefix);
502 strbuf_addstr(&buf, filename);
503 name = strbuf_detach(&buf, NULL);
507 grep_file_async(opt, name, filename);
514 void *data = load_file(filename, &sz);
518 hit = grep_buffer(opt, name, data, sz);
526 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
532 for (nr = 0; nr < active_nr; nr++) {
533 struct cache_entry *ce = active_cache[nr];
534 if (!S_ISREG(ce->ce_mode))
536 if (!pathspec_matches(paths, ce->name, opt->max_depth))
539 * If CE_VALID is on, we assume worktree file and its cache entry
540 * are identical, even if worktree file has been modified, so use
541 * cache version instead
543 if (cached || (ce->ce_flags & CE_VALID) || ce_skip_worktree(ce)) {
546 hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
549 hit |= grep_file(opt, ce->name);
553 } while (nr < active_nr &&
554 !strcmp(ce->name, active_cache[nr]->name));
555 nr--; /* compensate for loop control */
557 if (hit && opt->status_only)
560 free_grep_patterns(opt);
564 static int grep_tree(struct grep_opt *opt, const char **paths,
565 struct tree_desc *tree,
566 const char *tree_name, const char *base)
570 struct name_entry entry;
572 int tn_len = strlen(tree_name);
573 struct strbuf pathbuf;
575 strbuf_init(&pathbuf, PATH_MAX + tn_len);
578 strbuf_add(&pathbuf, tree_name, tn_len);
579 strbuf_addch(&pathbuf, ':');
580 tn_len = pathbuf.len;
582 strbuf_addstr(&pathbuf, base);
585 while (tree_entry(tree, &entry)) {
586 int te_len = tree_entry_len(entry.path, entry.sha1);
588 strbuf_add(&pathbuf, entry.path, te_len);
590 if (S_ISDIR(entry.mode))
591 /* Match "abc/" against pathspec to
592 * decide if we want to descend into "abc"
595 strbuf_addch(&pathbuf, '/');
597 down = pathbuf.buf + tn_len;
598 if (!pathspec_matches(paths, down, opt->max_depth))
600 else if (S_ISREG(entry.mode))
601 hit |= grep_sha1(opt, entry.sha1, pathbuf.buf, tn_len);
602 else if (S_ISDIR(entry.mode)) {
603 enum object_type type;
604 struct tree_desc sub;
609 data = read_sha1_file(entry.sha1, &type, &size);
613 die("unable to read tree (%s)",
614 sha1_to_hex(entry.sha1));
615 init_tree_desc(&sub, data, size);
616 hit |= grep_tree(opt, paths, &sub, tree_name, down);
619 if (hit && opt->status_only)
622 strbuf_release(&pathbuf);
626 static int grep_object(struct grep_opt *opt, const char **paths,
627 struct object *obj, const char *name)
629 if (obj->type == OBJ_BLOB)
630 return grep_sha1(opt, obj->sha1, name, 0);
631 if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
632 struct tree_desc tree;
636 data = read_object_with_reference(obj->sha1, tree_type,
639 die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
640 init_tree_desc(&tree, data, size);
641 hit = grep_tree(opt, paths, &tree, name, "");
645 die("unable to grep from object of type %s", typename(obj->type));
648 static int context_callback(const struct option *opt, const char *arg,
651 struct grep_opt *grep_opt = opt->value;
656 grep_opt->pre_context = grep_opt->post_context = 0;
659 value = strtol(arg, (char **)&endp, 10);
661 return error("switch `%c' expects a numerical value",
664 grep_opt->pre_context = grep_opt->post_context = value;
668 static int file_callback(const struct option *opt, const char *arg, int unset)
670 struct grep_opt *grep_opt = opt->value;
673 struct strbuf sb = STRBUF_INIT;
675 patterns = fopen(arg, "r");
677 die_errno("cannot open '%s'", arg);
678 while (strbuf_getline(&sb, patterns, '\n') == 0) {
679 /* ignore empty line like grep does */
682 append_grep_pattern(grep_opt, strbuf_detach(&sb, NULL), arg,
683 ++lno, GREP_PATTERN);
690 static int not_callback(const struct option *opt, const char *arg, int unset)
692 struct grep_opt *grep_opt = opt->value;
693 append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
697 static int and_callback(const struct option *opt, const char *arg, int unset)
699 struct grep_opt *grep_opt = opt->value;
700 append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
704 static int open_callback(const struct option *opt, const char *arg, int unset)
706 struct grep_opt *grep_opt = opt->value;
707 append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
711 static int close_callback(const struct option *opt, const char *arg, int unset)
713 struct grep_opt *grep_opt = opt->value;
714 append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
718 static int pattern_callback(const struct option *opt, const char *arg,
721 struct grep_opt *grep_opt = opt->value;
722 append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
726 static int help_callback(const struct option *opt, const char *arg, int unset)
731 int cmd_grep(int argc, const char **argv, const char *prefix)
735 int seen_dashdash = 0;
736 int external_grep_allowed__ignored;
738 struct object_array list = { 0, 0, NULL };
739 const char **paths = NULL;
742 struct option options[] = {
743 OPT_BOOLEAN(0, "cached", &cached,
744 "search in index instead of in the work tree"),
746 OPT_BOOLEAN('v', "invert-match", &opt.invert,
747 "show non-matching lines"),
748 OPT_BOOLEAN('i', "ignore-case", &opt.ignore_case,
749 "case insensitive matching"),
750 OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
751 "match patterns only at word boundaries"),
752 OPT_SET_INT('a', "text", &opt.binary,
753 "process binary files as text", GREP_BINARY_TEXT),
754 OPT_SET_INT('I', NULL, &opt.binary,
755 "don't match patterns in binary files",
756 GREP_BINARY_NOMATCH),
757 { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
758 "descend at most <depth> levels", PARSE_OPT_NONEG,
761 OPT_BIT('E', "extended-regexp", &opt.regflags,
762 "use extended POSIX regular expressions", REG_EXTENDED),
763 OPT_NEGBIT('G', "basic-regexp", &opt.regflags,
764 "use basic POSIX regular expressions (default)",
766 OPT_BOOLEAN('F', "fixed-strings", &opt.fixed,
767 "interpret patterns as fixed strings"),
769 OPT_BOOLEAN('n', NULL, &opt.linenum, "show line numbers"),
770 OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
771 OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
772 OPT_NEGBIT(0, "full-name", &opt.relative,
773 "show filenames relative to top directory", 1),
774 OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
775 "show only filenames instead of matching lines"),
776 OPT_BOOLEAN(0, "name-only", &opt.name_only,
777 "synonym for --files-with-matches"),
778 OPT_BOOLEAN('L', "files-without-match",
779 &opt.unmatch_name_only,
780 "show only the names of files without match"),
781 OPT_BOOLEAN('z', "null", &opt.null_following_name,
782 "print NUL after filenames"),
783 OPT_BOOLEAN('c', "count", &opt.count,
784 "show the number of matches instead of matching lines"),
785 OPT_SET_INT(0, "color", &opt.color, "highlight matches", 1),
787 OPT_CALLBACK('C', NULL, &opt, "n",
788 "show <n> context lines before and after matches",
790 OPT_INTEGER('B', NULL, &opt.pre_context,
791 "show <n> context lines before matches"),
792 OPT_INTEGER('A', NULL, &opt.post_context,
793 "show <n> context lines after matches"),
794 OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
796 OPT_BOOLEAN('p', "show-function", &opt.funcname,
797 "show a line with the function name before matches"),
799 OPT_CALLBACK('f', NULL, &opt, "file",
800 "read patterns from file", file_callback),
801 { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
802 "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
803 { OPTION_CALLBACK, 0, "and", &opt, NULL,
804 "combine patterns specified with -e",
805 PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
806 OPT_BOOLEAN(0, "or", &dummy, ""),
807 { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
808 PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
809 { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
810 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
812 { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
813 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
815 OPT_BOOLEAN('q', "quiet", &opt.status_only,
816 "indicate hit with exit status without output"),
817 OPT_BOOLEAN(0, "all-match", &opt.all_match,
818 "show only matches from files that match all patterns"),
820 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
821 "allow calling of grep(1) (ignored by this build)"),
822 { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
823 PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
828 * 'git grep -h', unlike 'git grep -h <pattern>', is a request
829 * to show usage information and exit.
831 if (argc == 2 && !strcmp(argv[1], "-h"))
832 usage_with_options(grep_usage, options);
834 memset(&opt, 0, sizeof(opt));
836 opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
839 opt.pattern_tail = &opt.pattern_list;
840 opt.regflags = REG_NEWLINE;
843 strcpy(opt.color_match, GIT_COLOR_RED GIT_COLOR_BOLD);
845 git_config(grep_config, &opt);
847 opt.color = git_use_color_default;
850 * If there is no -- then the paths must exist in the working
851 * tree. If there is no explicit pattern specified with -e or
852 * -f, we take the first unrecognized non option to be the
853 * pattern, but then what follows it must be zero or more
854 * valid refs up to the -- (if exists), and then existing
855 * paths. If there is an explicit pattern, then the first
856 * unrecognized non option is the beginning of the refs list
857 * that continues up to the -- (if exists), and then paths.
859 argc = parse_options(argc, argv, prefix, options, grep_usage,
860 PARSE_OPT_KEEP_DASHDASH |
861 PARSE_OPT_STOP_AT_NON_OPTION |
862 PARSE_OPT_NO_INTERNAL_HELP);
864 /* First unrecognized non-option token */
865 if (argc > 0 && !opt.pattern_list) {
866 append_grep_pattern(&opt, argv[0], "command line", 0,
872 if (!opt.pattern_list)
873 die("no pattern given.");
874 if (!opt.fixed && opt.ignore_case)
875 opt.regflags |= REG_ICASE;
876 if ((opt.regflags != REG_NEWLINE) && opt.fixed)
877 die("cannot mix --fixed-strings and regexp");
880 if (online_cpus() == 1 || !grep_threads_ok(&opt))
889 compile_grep_patterns(&opt);
891 /* Check revs and then paths */
892 for (i = 0; i < argc; i++) {
893 const char *arg = argv[i];
894 unsigned char sha1[20];
896 if (!get_sha1(arg, sha1)) {
897 struct object *object = parse_object(sha1);
899 die("bad object %s", arg);
900 add_object_array(object, arg, &list);
903 if (!strcmp(arg, "--")) {
910 /* The rest are paths */
911 if (!seen_dashdash) {
913 for (j = i; j < argc; j++)
914 verify_filename(prefix, argv[j]);
918 paths = get_pathspec(prefix, argv + i);
920 paths = xcalloc(2, sizeof(const char *));
930 hit = grep_cache(&opt, paths, cached);
937 die("both --cached and trees are given.");
939 for (i = 0; i < list.nr; i++) {
940 struct object *real_obj;
941 real_obj = deref_tag(list.objects[i].item, NULL, 0);
942 if (grep_object(&opt, paths, real_obj, list.objects[i].name)) {
951 free_grep_patterns(&opt);