4 * Copyright (c) 2006 Junio C Hamano
11 #include "tree-walk.h"
13 #include "parse-options.h"
18 #ifndef NO_EXTERNAL_GREP
20 #define NO_EXTERNAL_GREP 0
22 #define NO_EXTERNAL_GREP 1
26 static char const * const grep_usage
[] = {
27 "git grep [options] [-e] <pattern> [<rev>...] [[--] path...]",
31 static int grep_config(const char *var
, const char *value
, void *cb
)
33 struct grep_opt
*opt
= cb
;
35 switch (userdiff_config(var
, value
)) {
41 if (!strcmp(var
, "color.grep")) {
42 opt
->color
= git_config_colorbool(var
, value
, -1);
45 if (!strcmp(var
, "color.grep.external"))
46 return git_config_string(&(opt
->color_external
), var
, value
);
47 if (!strcmp(var
, "color.grep.match")) {
49 return config_error_nonbool(var
);
50 color_parse(value
, var
, opt
->color_match
);
53 return git_color_default_config(var
, value
, cb
);
57 * git grep pathspecs are somewhat different from diff-tree pathspecs;
58 * pathname wildcards are allowed.
60 static int pathspec_matches(const char **paths
, const char *name
)
63 if (!paths
|| !*paths
)
65 namelen
= strlen(name
);
66 for (i
= 0; paths
[i
]; i
++) {
67 const char *match
= paths
[i
];
68 int matchlen
= strlen(match
);
69 const char *cp
, *meta
;
72 ((matchlen
<= namelen
) &&
73 !strncmp(name
, match
, matchlen
) &&
74 (match
[matchlen
-1] == '/' ||
75 name
[matchlen
] == '\0' || name
[matchlen
] == '/')))
77 if (!fnmatch(match
, name
, 0))
79 if (name
[namelen
-1] != '/')
82 /* We are being asked if the directory ("name") is worth
85 * Find the longest leading directory name that does
86 * not have metacharacter in the pathspec; the name
87 * we are looking at must overlap with that directory.
89 for (cp
= match
, meta
= NULL
; cp
- match
< matchlen
; cp
++) {
91 if (ch
== '*' || ch
== '[' || ch
== '?') {
97 meta
= cp
; /* fully literal */
99 if (namelen
<= meta
- match
) {
100 /* Looking at "Documentation/" and
101 * the pattern says "Documentation/howto/", or
102 * "Documentation/diff*.txt". The name we
103 * have should match prefix.
105 if (!memcmp(match
, name
, namelen
))
110 if (meta
- match
< namelen
) {
111 /* Looking at "Documentation/howto/" and
112 * the pattern says "Documentation/h*";
113 * match up to "Do.../h"; this avoids descending
114 * into "Documentation/technical/".
116 if (!memcmp(match
, name
, meta
- match
))
124 static int grep_sha1(struct grep_opt
*opt
, const unsigned char *sha1
, const char *name
, int tree_name_len
)
128 enum object_type type
;
130 struct strbuf pathbuf
= STRBUF_INIT
;
132 data
= read_sha1_file(sha1
, &type
, &size
);
134 error("'%s': unable to read %s", name
, sha1_to_hex(sha1
));
137 if (opt
->relative
&& opt
->prefix_length
) {
138 quote_path_relative(name
+ tree_name_len
, -1, &pathbuf
, opt
->prefix
);
139 strbuf_insert(&pathbuf
, 0, name
, tree_name_len
);
142 hit
= grep_buffer(opt
, name
, data
, size
);
143 strbuf_release(&pathbuf
);
148 static int grep_file(struct grep_opt
*opt
, const char *filename
)
154 struct strbuf buf
= STRBUF_INIT
;
156 if (lstat(filename
, &st
) < 0) {
159 error("'%s': %s", filename
, strerror(errno
));
163 return 0; /* empty file -- no grep hit */
164 if (!S_ISREG(st
.st_mode
))
166 sz
= xsize_t(st
.st_size
);
167 i
= open(filename
, O_RDONLY
);
170 data
= xmalloc(sz
+ 1);
171 if (st
.st_size
!= read_in_full(i
, data
, sz
)) {
172 error("'%s': short read %s", filename
, strerror(errno
));
178 if (opt
->relative
&& opt
->prefix_length
)
179 filename
= quote_path_relative(filename
, -1, &buf
, opt
->prefix
);
180 i
= grep_buffer(opt
, filename
, data
, sz
);
181 strbuf_release(&buf
);
186 #if !NO_EXTERNAL_GREP
187 static int exec_grep(int argc
, const char **argv
)
197 execvp("grep", (char **) argv
);
200 while (waitpid(pid
, &status
, 0) < 0) {
205 if (WIFEXITED(status
)) {
206 if (!WEXITSTATUS(status
))
215 #define push_arg(a) do { \
216 if (nr < MAXARGS) argv[nr++] = (a); \
217 else die("maximum number of args exceeded"); \
221 * If you send a singleton filename to grep, it does not give
222 * the name of the file. GNU grep has "-H" but we would want
223 * that behaviour in a portable way.
225 * So we keep two pathnames in argv buffer unsent to grep in
226 * the main loop if we need to do more than one grep.
228 static int flush_grep(struct grep_opt
*opt
,
229 int argc
, int arg0
, const char **argv
, int *kept
)
232 int count
= argc
- arg0
;
233 const char *kept_0
= NULL
;
237 * Because we keep at least 2 paths in the call from
238 * the main loop (i.e. kept != NULL), and MAXARGS is
239 * far greater than 2, this usually is a call to
240 * conclude the grep. However, the user could attempt
241 * to overflow the argv buffer by giving too many
242 * options to leave very small number of real
243 * arguments even for the call in the main loop.
246 die("insanely many options to grep");
249 * If we have two or more paths, we do not have to do
250 * anything special, but we need to push /dev/null to
251 * get "-H" behaviour of GNU grep portably but when we
252 * are not doing "-l" nor "-L" nor "-c".
256 !opt
->unmatch_name_only
&&
258 argv
[argc
++] = "/dev/null";
265 * Called because we found many paths and haven't finished
266 * iterating over the cache yet. We keep two paths
267 * for the concluding call. argv[argc-2] and argv[argc-1]
268 * has the last two paths, so save the first one away,
269 * replace it with NULL while sending the list to grep,
270 * and recover them after we are done.
273 kept_0
= argv
[argc
-2];
278 if (opt
->pre_context
|| opt
->post_context
) {
280 * grep handles hunk marks between files, but we need to
281 * do that ourselves between multiple calls.
283 if (opt
->show_hunk_mark
)
284 write_or_die(1, "--\n", 3);
286 opt
->show_hunk_mark
= 1;
289 status
= exec_grep(argc
, argv
);
293 * Then recover them. Now the last arg is beyond the
294 * terminating NULL which is at argc, and the second
295 * from the last is what we saved away in kept_0
297 argv
[arg0
++] = kept_0
;
298 argv
[arg0
] = argv
[argc
+1];
303 static void grep_add_color(struct strbuf
*sb
, const char *escape_seq
)
305 size_t orig_len
= sb
->len
;
307 while (*escape_seq
) {
308 if (*escape_seq
== 'm')
309 strbuf_addch(sb
, ';');
310 else if (*escape_seq
!= '\033' && *escape_seq
!= '[')
311 strbuf_addch(sb
, *escape_seq
);
314 if (sb
->len
> orig_len
&& sb
->buf
[sb
->len
- 1] == ';')
315 strbuf_setlen(sb
, sb
->len
- 1);
318 static int external_grep(struct grep_opt
*opt
, const char **paths
, int cached
)
320 int i
, nr
, argc
, hit
, len
, status
;
321 const char *argv
[MAXARGS
+1];
322 char randarg
[ARGBUF
];
323 char *argptr
= randarg
;
326 if (opt
->extended
|| (opt
->relative
&& opt
->prefix_length
))
336 if (opt
->regflags
& REG_EXTENDED
)
338 if (opt
->regflags
& REG_ICASE
)
340 if (opt
->binary
== GREP_BINARY_NOMATCH
)
342 if (opt
->word_regexp
)
346 if (opt
->unmatch_name_only
)
348 if (opt
->null_following_name
)
349 /* in GNU grep git's "-z" translates to "-Z" */
353 if (opt
->post_context
|| opt
->pre_context
) {
354 if (opt
->post_context
!= opt
->pre_context
) {
355 if (opt
->pre_context
) {
357 len
+= snprintf(argptr
, sizeof(randarg
)-len
,
358 "%u", opt
->pre_context
) + 1;
359 if (sizeof(randarg
) <= len
)
360 die("maximum length of args exceeded");
364 if (opt
->post_context
) {
366 len
+= snprintf(argptr
, sizeof(randarg
)-len
,
367 "%u", opt
->post_context
) + 1;
368 if (sizeof(randarg
) <= len
)
369 die("maximum length of args exceeded");
376 len
+= snprintf(argptr
, sizeof(randarg
)-len
,
377 "%u", opt
->post_context
) + 1;
378 if (sizeof(randarg
) <= len
)
379 die("maximum length of args exceeded");
384 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
386 push_arg(p
->pattern
);
389 struct strbuf sb
= STRBUF_INIT
;
391 grep_add_color(&sb
, opt
->color_match
);
392 setenv("GREP_COLOR", sb
.buf
, 1);
395 strbuf_addstr(&sb
, "mt=");
396 grep_add_color(&sb
, opt
->color_match
);
397 strbuf_addstr(&sb
, ":sl=:cx=:fn=:ln=:bn=:se=");
398 setenv("GREP_COLORS", sb
.buf
, 1);
402 if (opt
->color_external
&& strlen(opt
->color_external
) > 0)
403 push_arg(opt
->color_external
);
408 for (i
= 0; i
< active_nr
; i
++) {
409 struct cache_entry
*ce
= active_cache
[i
];
412 if (!S_ISREG(ce
->ce_mode
))
414 if (!pathspec_matches(paths
, ce
->name
))
417 if (name
[0] == '-') {
418 int len
= ce_namelen(ce
);
419 name
= xmalloc(len
+ 3);
420 memcpy(name
, "./", 2);
421 memcpy(name
+ 2, ce
->name
, len
+ 1);
424 if (MAXARGS
<= argc
) {
425 status
= flush_grep(opt
, argc
, nr
, argv
, &kept
);
433 } while (i
< active_nr
&&
434 !strcmp(ce
->name
, active_cache
[i
]->name
));
435 i
--; /* compensate for loop control */
439 status
= flush_grep(opt
, argc
, nr
, argv
, NULL
);
447 static int grep_cache(struct grep_opt
*opt
, const char **paths
, int cached
,
448 int external_grep_allowed
)
454 #if !NO_EXTERNAL_GREP
456 * Use the external "grep" command for the case where
457 * we grep through the checked-out files. It tends to
458 * be a lot more optimized
460 if (!cached
&& external_grep_allowed
) {
461 hit
= external_grep(opt
, paths
, cached
);
468 for (nr
= 0; nr
< active_nr
; nr
++) {
469 struct cache_entry
*ce
= active_cache
[nr
];
470 if (!S_ISREG(ce
->ce_mode
))
472 if (!pathspec_matches(paths
, ce
->name
))
475 * If CE_VALID is on, we assume worktree file and its cache entry
476 * are identical, even if worktree file has been modified, so use
477 * cache version instead
479 if (cached
|| (ce
->ce_flags
& CE_VALID
)) {
482 hit
|= grep_sha1(opt
, ce
->sha1
, ce
->name
, 0);
485 hit
|= grep_file(opt
, ce
->name
);
489 } while (nr
< active_nr
&&
490 !strcmp(ce
->name
, active_cache
[nr
]->name
));
491 nr
--; /* compensate for loop control */
494 free_grep_patterns(opt
);
498 static int grep_tree(struct grep_opt
*opt
, const char **paths
,
499 struct tree_desc
*tree
,
500 const char *tree_name
, const char *base
)
504 struct name_entry entry
;
506 int tn_len
= strlen(tree_name
);
507 struct strbuf pathbuf
;
509 strbuf_init(&pathbuf
, PATH_MAX
+ tn_len
);
512 strbuf_add(&pathbuf
, tree_name
, tn_len
);
513 strbuf_addch(&pathbuf
, ':');
514 tn_len
= pathbuf
.len
;
516 strbuf_addstr(&pathbuf
, base
);
519 while (tree_entry(tree
, &entry
)) {
520 int te_len
= tree_entry_len(entry
.path
, entry
.sha1
);
522 strbuf_add(&pathbuf
, entry
.path
, te_len
);
524 if (S_ISDIR(entry
.mode
))
525 /* Match "abc/" against pathspec to
526 * decide if we want to descend into "abc"
529 strbuf_addch(&pathbuf
, '/');
531 down
= pathbuf
.buf
+ tn_len
;
532 if (!pathspec_matches(paths
, down
))
534 else if (S_ISREG(entry
.mode
))
535 hit
|= grep_sha1(opt
, entry
.sha1
, pathbuf
.buf
, tn_len
);
536 else if (S_ISDIR(entry
.mode
)) {
537 enum object_type type
;
538 struct tree_desc sub
;
542 data
= read_sha1_file(entry
.sha1
, &type
, &size
);
544 die("unable to read tree (%s)",
545 sha1_to_hex(entry
.sha1
));
546 init_tree_desc(&sub
, data
, size
);
547 hit
|= grep_tree(opt
, paths
, &sub
, tree_name
, down
);
551 strbuf_release(&pathbuf
);
555 static int grep_object(struct grep_opt
*opt
, const char **paths
,
556 struct object
*obj
, const char *name
)
558 if (obj
->type
== OBJ_BLOB
)
559 return grep_sha1(opt
, obj
->sha1
, name
, 0);
560 if (obj
->type
== OBJ_COMMIT
|| obj
->type
== OBJ_TREE
) {
561 struct tree_desc tree
;
565 data
= read_object_with_reference(obj
->sha1
, tree_type
,
568 die("unable to read tree (%s)", sha1_to_hex(obj
->sha1
));
569 init_tree_desc(&tree
, data
, size
);
570 hit
= grep_tree(opt
, paths
, &tree
, name
, "");
574 die("unable to grep from object of type %s", typename(obj
->type
));
577 static int context_callback(const struct option
*opt
, const char *arg
,
580 struct grep_opt
*grep_opt
= opt
->value
;
585 grep_opt
->pre_context
= grep_opt
->post_context
= 0;
588 value
= strtol(arg
, (char **)&endp
, 10);
590 return error("switch `%c' expects a numerical value",
593 grep_opt
->pre_context
= grep_opt
->post_context
= value
;
597 static int file_callback(const struct option
*opt
, const char *arg
, int unset
)
599 struct grep_opt
*grep_opt
= opt
->value
;
602 struct strbuf sb
= STRBUF_INIT
;
604 patterns
= fopen(arg
, "r");
606 die_errno("cannot open '%s'", arg
);
607 while (strbuf_getline(&sb
, patterns
, '\n') == 0) {
608 /* ignore empty line like grep does */
611 append_grep_pattern(grep_opt
, strbuf_detach(&sb
, NULL
), arg
,
612 ++lno
, GREP_PATTERN
);
619 static int not_callback(const struct option
*opt
, const char *arg
, int unset
)
621 struct grep_opt
*grep_opt
= opt
->value
;
622 append_grep_pattern(grep_opt
, "--not", "command line", 0, GREP_NOT
);
626 static int and_callback(const struct option
*opt
, const char *arg
, int unset
)
628 struct grep_opt
*grep_opt
= opt
->value
;
629 append_grep_pattern(grep_opt
, "--and", "command line", 0, GREP_AND
);
633 static int open_callback(const struct option
*opt
, const char *arg
, int unset
)
635 struct grep_opt
*grep_opt
= opt
->value
;
636 append_grep_pattern(grep_opt
, "(", "command line", 0, GREP_OPEN_PAREN
);
640 static int close_callback(const struct option
*opt
, const char *arg
, int unset
)
642 struct grep_opt
*grep_opt
= opt
->value
;
643 append_grep_pattern(grep_opt
, ")", "command line", 0, GREP_CLOSE_PAREN
);
647 static int pattern_callback(const struct option
*opt
, const char *arg
,
650 struct grep_opt
*grep_opt
= opt
->value
;
651 append_grep_pattern(grep_opt
, arg
, "-e option", 0, GREP_PATTERN
);
655 static int help_callback(const struct option
*opt
, const char *arg
, int unset
)
660 int cmd_grep(int argc
, const char **argv
, const char *prefix
)
664 int external_grep_allowed
= 1;
665 int seen_dashdash
= 0;
667 struct object_array list
= { 0, 0, NULL
};
668 const char **paths
= NULL
;
671 struct option options
[] = {
672 OPT_BOOLEAN(0, "cached", &cached
,
673 "search in index instead of in the work tree"),
675 OPT_BOOLEAN('v', "invert-match", &opt
.invert
,
676 "show non-matching lines"),
677 OPT_BIT('i', "ignore-case", &opt
.regflags
,
678 "case insensitive matching", REG_ICASE
),
679 OPT_BOOLEAN('w', "word-regexp", &opt
.word_regexp
,
680 "match patterns only at word boundaries"),
681 OPT_SET_INT('a', "text", &opt
.binary
,
682 "process binary files as text", GREP_BINARY_TEXT
),
683 OPT_SET_INT('I', NULL
, &opt
.binary
,
684 "don't match patterns in binary files",
685 GREP_BINARY_NOMATCH
),
687 OPT_BIT('E', "extended-regexp", &opt
.regflags
,
688 "use extended POSIX regular expressions", REG_EXTENDED
),
689 OPT_NEGBIT('G', "basic-regexp", &opt
.regflags
,
690 "use basic POSIX regular expressions (default)",
692 OPT_BOOLEAN('F', "fixed-strings", &opt
.fixed
,
693 "interpret patterns as fixed strings"),
695 OPT_BOOLEAN('n', NULL
, &opt
.linenum
, "show line numbers"),
696 OPT_NEGBIT('h', NULL
, &opt
.pathname
, "don't show filenames", 1),
697 OPT_BIT('H', NULL
, &opt
.pathname
, "show filenames", 1),
698 OPT_NEGBIT(0, "full-name", &opt
.relative
,
699 "show filenames relative to top directory", 1),
700 OPT_BOOLEAN('l', "files-with-matches", &opt
.name_only
,
701 "show only filenames instead of matching lines"),
702 OPT_BOOLEAN(0, "name-only", &opt
.name_only
,
703 "synonym for --files-with-matches"),
704 OPT_BOOLEAN('L', "files-without-match",
705 &opt
.unmatch_name_only
,
706 "show only the names of files without match"),
707 OPT_BOOLEAN('z', "null", &opt
.null_following_name
,
708 "print NUL after filenames"),
709 OPT_BOOLEAN('c', "count", &opt
.count
,
710 "show the number of matches instead of matching lines"),
711 OPT_SET_INT(0, "color", &opt
.color
, "highlight matches", 1),
713 OPT_CALLBACK('C', NULL
, &opt
, "n",
714 "show <n> context lines before and after matches",
716 OPT_INTEGER('B', NULL
, &opt
.pre_context
,
717 "show <n> context lines before matches"),
718 OPT_INTEGER('A', NULL
, &opt
.post_context
,
719 "show <n> context lines after matches"),
720 OPT_NUMBER_CALLBACK(&opt
, "shortcut for -C NUM",
722 OPT_BOOLEAN('p', "show-function", &opt
.funcname
,
723 "show a line with the function name before matches"),
725 OPT_CALLBACK('f', NULL
, &opt
, "file",
726 "read patterns from file", file_callback
),
727 { OPTION_CALLBACK
, 'e', NULL
, &opt
, "pattern",
728 "match <pattern>", PARSE_OPT_NONEG
, pattern_callback
},
729 { OPTION_CALLBACK
, 0, "and", &opt
, NULL
,
730 "combine patterns specified with -e",
731 PARSE_OPT_NOARG
| PARSE_OPT_NONEG
, and_callback
},
732 OPT_BOOLEAN(0, "or", &dummy
, ""),
733 { OPTION_CALLBACK
, 0, "not", &opt
, NULL
, "",
734 PARSE_OPT_NOARG
| PARSE_OPT_NONEG
, not_callback
},
735 { OPTION_CALLBACK
, '(', NULL
, &opt
, NULL
, "",
736 PARSE_OPT_NOARG
| PARSE_OPT_NONEG
| PARSE_OPT_NODASH
,
738 { OPTION_CALLBACK
, ')', NULL
, &opt
, NULL
, "",
739 PARSE_OPT_NOARG
| PARSE_OPT_NONEG
| PARSE_OPT_NODASH
,
741 OPT_BOOLEAN(0, "all-match", &opt
.all_match
,
742 "show only matches from files that match all patterns"),
745 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed
,
746 "allow calling of grep(1) (ignored by this build)"),
748 OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed
,
749 "allow calling of grep(1) (default)"),
751 { OPTION_CALLBACK
, 0, "help-all", &options
, NULL
, "show usage",
752 PARSE_OPT_HIDDEN
| PARSE_OPT_NOARG
, help_callback
},
756 memset(&opt
, 0, sizeof(opt
));
758 opt
.prefix_length
= (prefix
&& *prefix
) ? strlen(prefix
) : 0;
761 opt
.pattern_tail
= &opt
.pattern_list
;
762 opt
.regflags
= REG_NEWLINE
;
764 strcpy(opt
.color_match
, GIT_COLOR_RED GIT_COLOR_BOLD
);
766 git_config(grep_config
, &opt
);
768 opt
.color
= git_use_color_default
;
771 * If there is no -- then the paths must exist in the working
772 * tree. If there is no explicit pattern specified with -e or
773 * -f, we take the first unrecognized non option to be the
774 * pattern, but then what follows it must be zero or more
775 * valid refs up to the -- (if exists), and then existing
776 * paths. If there is an explicit pattern, then the first
777 * unrecognized non option is the beginning of the refs list
778 * that continues up to the -- (if exists), and then paths.
780 argc
= parse_options(argc
, argv
, prefix
, options
, grep_usage
,
781 PARSE_OPT_KEEP_DASHDASH
|
782 PARSE_OPT_STOP_AT_NON_OPTION
|
783 PARSE_OPT_NO_INTERNAL_HELP
);
785 /* First unrecognized non-option token */
786 if (argc
> 0 && !opt
.pattern_list
) {
787 append_grep_pattern(&opt
, argv
[0], "command line", 0,
793 if ((opt
.color
&& !opt
.color_external
) || opt
.funcname
)
794 external_grep_allowed
= 0;
795 if (!opt
.pattern_list
)
796 die("no pattern given.");
797 if ((opt
.regflags
!= REG_NEWLINE
) && opt
.fixed
)
798 die("cannot mix --fixed-strings and regexp");
799 compile_grep_patterns(&opt
);
801 /* Check revs and then paths */
802 for (i
= 0; i
< argc
; i
++) {
803 const char *arg
= argv
[i
];
804 unsigned char sha1
[20];
806 if (!get_sha1(arg
, sha1
)) {
807 struct object
*object
= parse_object(sha1
);
809 die("bad object %s", arg
);
810 add_object_array(object
, arg
, &list
);
813 if (!strcmp(arg
, "--")) {
820 /* The rest are paths */
821 if (!seen_dashdash
) {
823 for (j
= i
; j
< argc
; j
++)
824 verify_filename(prefix
, argv
[j
]);
828 paths
= get_pathspec(prefix
, argv
+ i
);
830 paths
= xcalloc(2, sizeof(const char *));
838 return !grep_cache(&opt
, paths
, cached
, external_grep_allowed
);
842 die("both --cached and trees are given.");
844 for (i
= 0; i
< list
.nr
; i
++) {
845 struct object
*real_obj
;
846 real_obj
= deref_tag(list
.objects
[i
].item
, NULL
, 0);
847 if (grep_object(&opt
, paths
, real_obj
, list
.objects
[i
].name
))
850 free_grep_patterns(&opt
);