4 * Copyright (c) 2006 Junio C Hamano
11 #include "tree-walk.h"
17 * git grep pathspecs are somewhat different from diff-tree pathspecs;
18 * pathname wildcards are allowed.
20 static int pathspec_matches(const char **paths
, const char *name
)
23 if (!paths
|| !*paths
)
25 namelen
= strlen(name
);
26 for (i
= 0; paths
[i
]; i
++) {
27 const char *match
= paths
[i
];
28 int matchlen
= strlen(match
);
29 const char *cp
, *meta
;
31 if ((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
))
84 struct grep_pat
*next
;
92 struct grep_pat
*pattern_list
;
93 struct grep_pat
**pattern_tail
;
98 unsigned unmatch_name_only
:1;
100 unsigned word_regexp
:1;
102 #define GREP_BINARY_DEFAULT 0
103 #define GREP_BINARY_NOMATCH 1
104 #define GREP_BINARY_TEXT 2
107 unsigned pre_context
;
108 unsigned post_context
;
111 static void add_pattern(struct grep_opt
*opt
, const char *pat
,
112 const char *origin
, int no
)
114 struct grep_pat
*p
= xcalloc(1, sizeof(*p
));
118 *opt
->pattern_tail
= p
;
119 opt
->pattern_tail
= &p
->next
;
123 static void compile_patterns(struct grep_opt
*opt
)
126 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
127 int err
= regcomp(&p
->regexp
, p
->pattern
, opt
->regflags
);
132 sprintf(where
, "In '%s' at %d, ",
135 sprintf(where
, "%s, ", p
->origin
);
138 regerror(err
, &p
->regexp
, errbuf
, 1024);
140 die("%s'%s': %s", where
, p
->pattern
, errbuf
);
145 static char *end_of_line(char *cp
, unsigned long *left
)
147 unsigned long l
= *left
;
148 while (l
&& *cp
!= '\n') {
156 static int word_char(char ch
)
158 return isalnum(ch
) || ch
== '_';
161 static void show_line(struct grep_opt
*opt
, const char *bol
, const char *eol
,
162 const char *name
, unsigned lno
, char sign
)
164 printf("%s%c", name
, sign
);
166 printf("%d%c", lno
, sign
);
167 printf("%.*s\n", (int)(eol
-bol
), bol
);
171 * NEEDSWORK: share code with diff.c
173 #define FIRST_FEW_BYTES 8000
174 static int buffer_is_binary(const char *ptr
, unsigned long size
)
176 if (FIRST_FEW_BYTES
< size
)
177 size
= FIRST_FEW_BYTES
;
178 if (memchr(ptr
, 0, size
))
183 static int fixmatch(const char *pattern
, char *line
, regmatch_t
*match
)
185 char *hit
= strstr(line
, pattern
);
187 match
->rm_so
= match
->rm_eo
= -1;
191 match
->rm_so
= hit
- line
;
192 match
->rm_eo
= match
->rm_so
+ strlen(pattern
);
197 static int grep_buffer(struct grep_opt
*opt
, const char *name
,
198 char *buf
, unsigned long size
)
201 unsigned long left
= size
;
203 struct pre_context_line
{
206 } *prev
= NULL
, *pcl
;
207 unsigned last_hit
= 0;
208 unsigned last_shown
= 0;
209 int binary_match_only
= 0;
210 const char *hunk_mark
= "";
213 if (buffer_is_binary(buf
, size
)) {
214 switch (opt
->binary
) {
215 case GREP_BINARY_DEFAULT
:
216 binary_match_only
= 1;
218 case GREP_BINARY_NOMATCH
:
219 return 0; /* Assume unmatch */
226 if (opt
->pre_context
)
227 prev
= xcalloc(opt
->pre_context
, sizeof(*prev
));
228 if (opt
->pre_context
|| opt
->post_context
)
232 regmatch_t pmatch
[10];
237 eol
= end_of_line(bol
, &left
);
241 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
243 regex_t
*exp
= &p
->regexp
;
244 hit
= !regexec(exp
, bol
, ARRAY_SIZE(pmatch
),
248 hit
= !fixmatch(p
->pattern
, bol
, pmatch
);
251 if (hit
&& opt
->word_regexp
) {
252 /* Match beginning must be either
253 * beginning of the line, or at word
254 * boundary (i.e. the last char must
255 * not be alnum or underscore).
257 if ((pmatch
[0].rm_so
< 0) ||
258 (eol
- bol
) <= pmatch
[0].rm_so
||
259 (pmatch
[0].rm_eo
< 0) ||
260 (eol
- bol
) < pmatch
[0].rm_eo
)
261 die("regexp returned nonsense");
262 if (pmatch
[0].rm_so
!= 0 &&
263 word_char(bol
[pmatch
[0].rm_so
-1]))
265 if (pmatch
[0].rm_eo
!= (eol
-bol
) &&
266 word_char(bol
[pmatch
[0].rm_eo
]))
272 /* "grep -v -e foo -e bla" should list lines
273 * that do not have either, so inversion should
278 if (opt
->unmatch_name_only
) {
285 if (binary_match_only
) {
286 printf("Binary file %s matches\n", name
);
289 if (opt
->name_only
) {
290 printf("%s\n", name
);
293 /* Hit at this line. If we haven't shown the
294 * pre-context lines, we would need to show them.
295 * When asked to do "count", this still show
296 * the context which is nonsense, but the user
297 * deserves to get that ;-).
299 if (opt
->pre_context
) {
301 if (opt
->pre_context
< lno
)
302 from
= lno
- opt
->pre_context
;
305 if (from
<= last_shown
)
306 from
= last_shown
+ 1;
307 if (last_shown
&& from
!= last_shown
+ 1)
310 pcl
= &prev
[lno
-from
-1];
311 show_line(opt
, pcl
->bol
, pcl
->eol
,
317 if (last_shown
&& lno
!= last_shown
+ 1)
320 show_line(opt
, bol
, eol
, name
, lno
, ':');
321 last_shown
= last_hit
= lno
;
324 lno
<= last_hit
+ opt
->post_context
) {
325 /* If the last hit is within the post context,
326 * we need to show this line.
328 if (last_shown
&& lno
!= last_shown
+ 1)
330 show_line(opt
, bol
, eol
, name
, lno
, '-');
333 if (opt
->pre_context
) {
334 memmove(prev
+1, prev
,
335 (opt
->pre_context
-1) * sizeof(*prev
));
349 if (opt
->unmatch_name_only
) {
350 /* We did not see any hit, so we want to show this */
351 printf("%s\n", name
);
356 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
357 * which feels mostly useless but sometimes useful. Maybe
358 * make it another option? For now suppress them.
360 if (opt
->count
&& count
)
361 printf("%s:%u\n", name
, count
);
365 static int grep_sha1(struct grep_opt
*opt
, const unsigned char *sha1
, const char *name
)
371 data
= read_sha1_file(sha1
, type
, &size
);
373 error("'%s': unable to read %s", name
, sha1_to_hex(sha1
));
376 hit
= grep_buffer(opt
, name
, data
, size
);
381 static int grep_file(struct grep_opt
*opt
, const char *filename
)
386 if (lstat(filename
, &st
) < 0) {
389 error("'%s': %s", filename
, strerror(errno
));
393 return 0; /* empty file -- no grep hit */
394 if (!S_ISREG(st
.st_mode
))
396 i
= open(filename
, O_RDONLY
);
399 data
= xmalloc(st
.st_size
+ 1);
400 if (st
.st_size
!= xread(i
, data
, st
.st_size
)) {
401 error("'%s': short read %s", filename
, strerror(errno
));
407 i
= grep_buffer(opt
, filename
, data
, st
.st_size
);
412 static int grep_cache(struct grep_opt
*opt
, const char **paths
, int cached
)
418 for (nr
= 0; nr
< active_nr
; nr
++) {
419 struct cache_entry
*ce
= active_cache
[nr
];
420 if (ce_stage(ce
) || !S_ISREG(ntohl(ce
->ce_mode
)))
422 if (!pathspec_matches(paths
, ce
->name
))
425 hit
|= grep_sha1(opt
, ce
->sha1
, ce
->name
);
427 hit
|= grep_file(opt
, ce
->name
);
432 static int grep_tree(struct grep_opt
*opt
, const char **paths
,
433 struct tree_desc
*tree
,
434 const char *tree_name
, const char *base
)
440 const unsigned char *sha1
;
442 char *path_buf
= xmalloc(PATH_MAX
+ strlen(tree_name
) + 100);
445 int offset
= sprintf(path_buf
, "%s:", tree_name
);
446 down
= path_buf
+ offset
;
453 len
= strlen(path_buf
);
457 sha1
= tree_entry_extract(tree
, &path
, &mode
);
458 pathlen
= strlen(path
);
459 strcpy(path_buf
+ len
, path
);
462 /* Match "abc/" against pathspec to
463 * decide if we want to descend into "abc"
466 strcpy(path_buf
+ len
+ pathlen
, "/");
468 if (!pathspec_matches(paths
, down
))
470 else if (S_ISREG(mode
))
471 hit
|= grep_sha1(opt
, sha1
, path_buf
);
472 else if (S_ISDIR(mode
)) {
474 struct tree_desc sub
;
476 data
= read_sha1_file(sha1
, type
, &sub
.size
);
478 die("unable to read tree (%s)",
481 hit
|= grep_tree(opt
, paths
, &sub
, tree_name
, down
);
484 update_tree_entry(tree
);
489 static int grep_object(struct grep_opt
*opt
, const char **paths
,
490 struct object
*obj
, const char *name
)
492 if (!strcmp(obj
->type
, blob_type
))
493 return grep_sha1(opt
, obj
->sha1
, name
);
494 if (!strcmp(obj
->type
, commit_type
) ||
495 !strcmp(obj
->type
, tree_type
)) {
496 struct tree_desc tree
;
499 data
= read_object_with_reference(obj
->sha1
, tree_type
,
502 die("unable to read tree (%s)", sha1_to_hex(obj
->sha1
));
504 hit
= grep_tree(opt
, paths
, &tree
, name
, "");
508 die("unable to grep from object of type %s", obj
->type
);
511 static const char builtin_grep_usage
[] =
512 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
514 int cmd_grep(int argc
, const char **argv
, char **envp
)
518 int seen_dashdash
= 0;
520 struct object_list
*list
, **tail
, *object_list
= NULL
;
521 const char *prefix
= setup_git_directory();
522 const char **paths
= NULL
;
525 memset(&opt
, 0, sizeof(opt
));
526 opt
.pattern_tail
= &opt
.pattern_list
;
527 opt
.regflags
= REG_NEWLINE
;
530 * If there is no -- then the paths must exist in the working
531 * tree. If there is no explicit pattern specified with -e or
532 * -f, we take the first unrecognized non option to be the
533 * pattern, but then what follows it must be zero or more
534 * valid refs up to the -- (if exists), and then existing
535 * paths. If there is an explicit pattern, then the first
536 * unrecocnized non option is the beginning of the refs list
537 * that continues up to the -- (if exists), and then paths.
542 const char *arg
= argv
[1];
544 if (!strcmp("--cached", arg
)) {
548 if (!strcmp("-a", arg
) ||
549 !strcmp("--text", arg
)) {
550 opt
.binary
= GREP_BINARY_TEXT
;
553 if (!strcmp("-i", arg
) ||
554 !strcmp("--ignore-case", arg
)) {
555 opt
.regflags
|= REG_ICASE
;
558 if (!strcmp("-I", arg
)) {
559 opt
.binary
= GREP_BINARY_NOMATCH
;
562 if (!strcmp("-v", arg
) ||
563 !strcmp("--invert-match", arg
)) {
567 if (!strcmp("-E", arg
) ||
568 !strcmp("--extended-regexp", arg
)) {
569 opt
.regflags
|= REG_EXTENDED
;
572 if (!strcmp("-F", arg
) ||
573 !strcmp("--fixed-strings", arg
)) {
577 if (!strcmp("-G", arg
) ||
578 !strcmp("--basic-regexp", arg
)) {
579 opt
.regflags
&= ~REG_EXTENDED
;
582 if (!strcmp("-n", arg
)) {
586 if (!strcmp("-H", arg
)) {
587 /* We always show the pathname, so this
592 if (!strcmp("-l", arg
) ||
593 !strcmp("--files-with-matches", arg
)) {
597 if (!strcmp("-L", arg
) ||
598 !strcmp("--files-without-match", arg
)) {
599 opt
.unmatch_name_only
= 1;
602 if (!strcmp("-c", arg
) ||
603 !strcmp("--count", arg
)) {
607 if (!strcmp("-w", arg
) ||
608 !strcmp("--word-regexp", arg
)) {
612 if (!strncmp("-A", arg
, 2) ||
613 !strncmp("-B", arg
, 2) ||
614 !strncmp("-C", arg
, 2) ||
615 (arg
[0] == '-' && '1' <= arg
[1] && arg
[1] <= '9')) {
619 case 'A': case 'B': case 'C':
622 usage(builtin_grep_usage
);
633 if (sscanf(scan
, "%u", &num
) != 1)
634 usage(builtin_grep_usage
);
637 opt
.post_context
= num
;
641 opt
.post_context
= num
;
643 opt
.pre_context
= num
;
648 if (!strcmp("-f", arg
)) {
653 usage(builtin_grep_usage
);
654 patterns
= fopen(argv
[1], "r");
656 die("'%s': %s", argv
[1], strerror(errno
));
657 while (fgets(buf
, sizeof(buf
), patterns
)) {
658 int len
= strlen(buf
);
659 if (buf
[len
-1] == '\n')
661 /* ignore empty line like grep does */
664 add_pattern(&opt
, strdup(buf
), argv
[1], ++lno
);
671 if (!strcmp("-e", arg
)) {
673 add_pattern(&opt
, argv
[1], "-e option", 0);
678 usage(builtin_grep_usage
);
680 if (!strcmp("--", arg
))
683 usage(builtin_grep_usage
);
685 /* First unrecognized non-option token */
686 if (!opt
.pattern_list
) {
687 add_pattern(&opt
, arg
, "command line", 0);
691 /* We are looking at the first path or rev;
692 * it is found at argv[1] after leaving the
700 if (!opt
.pattern_list
)
701 die("no pattern given.");
702 if ((opt
.regflags
!= REG_NEWLINE
) && opt
.fixed
)
703 die("cannot mix --fixed-strings and regexp");
705 compile_patterns(&opt
);
707 /* Check revs and then paths */
708 for (i
= 1; i
< argc
; i
++) {
709 const char *arg
= argv
[i
];
710 unsigned char sha1
[20];
712 if (!get_sha1(arg
, sha1
)) {
713 struct object
*object
= parse_object(sha1
);
714 struct object_list
*elem
;
716 die("bad object %s", arg
);
717 elem
= object_list_insert(object
, tail
);
722 if (!strcmp(arg
, "--")) {
729 /* The rest are paths */
730 if (!seen_dashdash
) {
732 for (j
= i
; j
< argc
; j
++)
733 verify_filename(prefix
, argv
[j
]);
737 paths
= get_pathspec(prefix
, argv
+ i
);
739 paths
= xcalloc(2, sizeof(const char *));
745 return !grep_cache(&opt
, paths
, cached
);
748 die("both --cached and trees are given.");
750 for (list
= object_list
; list
; list
= list
->next
) {
751 struct object
*real_obj
;
752 real_obj
= deref_tag(list
->item
, NULL
, 0);
753 if (grep_object(&opt
, paths
, real_obj
, list
->name
))