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
)
87 enum object_type type
;
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
);
154 static int exec_grep(int argc
, const char **argv
)
164 execvp("grep", (char **) argv
);
167 while (waitpid(pid
, &status
, 0) < 0) {
172 if (WIFEXITED(status
)) {
173 if (!WEXITSTATUS(status
))
182 #define push_arg(a) do { \
183 if (nr < MAXARGS) argv[nr++] = (a); \
184 else die("maximum number of args exceeded"); \
187 static int external_grep(struct grep_opt
*opt
, const char **paths
, int cached
)
189 int i
, nr
, argc
, hit
, len
, status
;
190 const char *argv
[MAXARGS
+1];
191 char randarg
[ARGBUF
];
192 char *argptr
= randarg
;
195 if (opt
->extended
|| (opt
->relative
&& opt
->prefix_length
))
205 if (opt
->regflags
& REG_EXTENDED
)
207 if (opt
->regflags
& REG_ICASE
)
209 if (opt
->word_regexp
)
213 if (opt
->unmatch_name_only
)
217 if (opt
->post_context
|| opt
->pre_context
) {
218 if (opt
->post_context
!= opt
->pre_context
) {
219 if (opt
->pre_context
) {
221 len
+= snprintf(argptr
, sizeof(randarg
)-len
,
222 "%u", opt
->pre_context
);
223 if (sizeof(randarg
) <= len
)
224 die("maximum length of args exceeded");
228 if (opt
->post_context
) {
230 len
+= snprintf(argptr
, sizeof(randarg
)-len
,
231 "%u", opt
->post_context
);
232 if (sizeof(randarg
) <= len
)
233 die("maximum length of args exceeded");
240 len
+= snprintf(argptr
, sizeof(randarg
)-len
,
241 "%u", opt
->post_context
);
242 if (sizeof(randarg
) <= len
)
243 die("maximum length of args exceeded");
248 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
250 push_arg(p
->pattern
);
254 * To make sure we get the header printed out when we want it,
255 * add /dev/null to the paths to grep. This is unnecessary
256 * (and wrong) with "-l" or "-L", which always print out the
259 * GNU grep has "-H", but this is portable.
261 if (!opt
->name_only
&& !opt
->unmatch_name_only
)
262 push_arg("/dev/null");
266 for (i
= 0; i
< active_nr
; i
++) {
267 struct cache_entry
*ce
= active_cache
[i
];
269 if (!S_ISREG(ntohl(ce
->ce_mode
)))
271 if (!pathspec_matches(paths
, ce
->name
))
274 if (name
[0] == '-') {
275 int len
= ce_namelen(ce
);
276 name
= xmalloc(len
+ 3);
277 memcpy(name
, "./", 2);
278 memcpy(name
+ 2, ce
->name
, len
+ 1);
281 if (argc
< MAXARGS
&& !ce_stage(ce
))
283 status
= exec_grep(argc
, argv
);
290 } while (i
< active_nr
&&
291 !strcmp(ce
->name
, active_cache
[i
]->name
));
292 i
--; /* compensate for loop control */
296 status
= exec_grep(argc
, argv
);
304 static int grep_cache(struct grep_opt
*opt
, const char **paths
, int cached
)
312 * Use the external "grep" command for the case where
313 * we grep through the checked-out files. It tends to
314 * be a lot more optimized
317 hit
= external_grep(opt
, paths
, cached
);
323 for (nr
= 0; nr
< active_nr
; nr
++) {
324 struct cache_entry
*ce
= active_cache
[nr
];
325 if (!S_ISREG(ntohl(ce
->ce_mode
)))
327 if (!pathspec_matches(paths
, ce
->name
))
332 hit
|= grep_sha1(opt
, ce
->sha1
, ce
->name
, 0);
335 hit
|= grep_file(opt
, ce
->name
);
339 } while (nr
< active_nr
&&
340 !strcmp(ce
->name
, active_cache
[nr
]->name
));
341 nr
--; /* compensate for loop control */
344 free_grep_patterns(opt
);
348 static int grep_tree(struct grep_opt
*opt
, const char **paths
,
349 struct tree_desc
*tree
,
350 const char *tree_name
, const char *base
)
354 struct name_entry entry
;
356 int tn_len
= strlen(tree_name
);
357 char *path_buf
= xmalloc(PATH_MAX
+ tn_len
+ 100);
360 tn_len
= sprintf(path_buf
, "%s:", tree_name
);
361 down
= path_buf
+ tn_len
;
368 len
= strlen(path_buf
);
370 while (tree_entry(tree
, &entry
)) {
371 strcpy(path_buf
+ len
, entry
.path
);
373 if (S_ISDIR(entry
.mode
))
374 /* Match "abc/" against pathspec to
375 * decide if we want to descend into "abc"
378 strcpy(path_buf
+ len
+ entry
.pathlen
, "/");
380 if (!pathspec_matches(paths
, down
))
382 else if (S_ISREG(entry
.mode
))
383 hit
|= grep_sha1(opt
, entry
.sha1
, path_buf
, tn_len
);
384 else if (S_ISDIR(entry
.mode
)) {
385 enum object_type type
;
386 struct tree_desc sub
;
388 data
= read_sha1_file(entry
.sha1
, &type
, &sub
.size
);
390 die("unable to read tree (%s)",
391 sha1_to_hex(entry
.sha1
));
393 hit
|= grep_tree(opt
, paths
, &sub
, tree_name
, down
);
400 static int grep_object(struct grep_opt
*opt
, const char **paths
,
401 struct object
*obj
, const char *name
)
403 if (obj
->type
== OBJ_BLOB
)
404 return grep_sha1(opt
, obj
->sha1
, name
, 0);
405 if (obj
->type
== OBJ_COMMIT
|| obj
->type
== OBJ_TREE
) {
406 struct tree_desc tree
;
409 data
= read_object_with_reference(obj
->sha1
, tree_type
,
412 die("unable to read tree (%s)", sha1_to_hex(obj
->sha1
));
414 hit
= grep_tree(opt
, paths
, &tree
, name
, "");
418 die("unable to grep from object of type %s", typename(obj
->type
));
421 static const char builtin_grep_usage
[] =
422 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
424 static const char emsg_invalid_context_len
[] =
425 "%s: invalid context length argument";
426 static const char emsg_missing_context_len
[] =
427 "missing context length argument";
428 static const char emsg_missing_argument
[] =
429 "option requires an argument -%s";
431 int cmd_grep(int argc
, const char **argv
, const char *prefix
)
435 int seen_dashdash
= 0;
437 struct object_array list
= { 0, 0, NULL
};
438 const char **paths
= NULL
;
441 memset(&opt
, 0, sizeof(opt
));
442 opt
.prefix_length
= (prefix
&& *prefix
) ? strlen(prefix
) : 0;
445 opt
.pattern_tail
= &opt
.pattern_list
;
446 opt
.regflags
= REG_NEWLINE
;
449 * If there is no -- then the paths must exist in the working
450 * tree. If there is no explicit pattern specified with -e or
451 * -f, we take the first unrecognized non option to be the
452 * pattern, but then what follows it must be zero or more
453 * valid refs up to the -- (if exists), and then existing
454 * paths. If there is an explicit pattern, then the first
455 * unrecognized non option is the beginning of the refs list
456 * that continues up to the -- (if exists), and then paths.
460 const char *arg
= argv
[1];
462 if (!strcmp("--cached", arg
)) {
466 if (!strcmp("-a", arg
) ||
467 !strcmp("--text", arg
)) {
468 opt
.binary
= GREP_BINARY_TEXT
;
471 if (!strcmp("-i", arg
) ||
472 !strcmp("--ignore-case", arg
)) {
473 opt
.regflags
|= REG_ICASE
;
476 if (!strcmp("-I", arg
)) {
477 opt
.binary
= GREP_BINARY_NOMATCH
;
480 if (!strcmp("-v", arg
) ||
481 !strcmp("--invert-match", arg
)) {
485 if (!strcmp("-E", arg
) ||
486 !strcmp("--extended-regexp", arg
)) {
487 opt
.regflags
|= REG_EXTENDED
;
490 if (!strcmp("-F", arg
) ||
491 !strcmp("--fixed-strings", arg
)) {
495 if (!strcmp("-G", arg
) ||
496 !strcmp("--basic-regexp", arg
)) {
497 opt
.regflags
&= ~REG_EXTENDED
;
500 if (!strcmp("-n", arg
)) {
504 if (!strcmp("-h", arg
)) {
508 if (!strcmp("-H", arg
)) {
512 if (!strcmp("-l", arg
) ||
513 !strcmp("--files-with-matches", arg
)) {
517 if (!strcmp("-L", arg
) ||
518 !strcmp("--files-without-match", arg
)) {
519 opt
.unmatch_name_only
= 1;
522 if (!strcmp("-c", arg
) ||
523 !strcmp("--count", arg
)) {
527 if (!strcmp("-w", arg
) ||
528 !strcmp("--word-regexp", arg
)) {
532 if (!prefixcmp(arg
, "-A") ||
533 !prefixcmp(arg
, "-B") ||
534 !prefixcmp(arg
, "-C") ||
535 (arg
[0] == '-' && '1' <= arg
[1] && arg
[1] <= '9')) {
539 case 'A': case 'B': case 'C':
542 die(emsg_missing_context_len
);
553 if (sscanf(scan
, "%u", &num
) != 1)
554 die(emsg_invalid_context_len
, scan
);
557 opt
.post_context
= num
;
561 opt
.post_context
= num
;
563 opt
.pre_context
= num
;
568 if (!strcmp("-f", arg
)) {
573 die(emsg_missing_argument
, arg
);
574 patterns
= fopen(argv
[1], "r");
576 die("'%s': %s", argv
[1], strerror(errno
));
577 while (fgets(buf
, sizeof(buf
), patterns
)) {
578 int len
= strlen(buf
);
579 if (buf
[len
-1] == '\n')
581 /* ignore empty line like grep does */
584 append_grep_pattern(&opt
, xstrdup(buf
),
593 if (!strcmp("--not", arg
)) {
594 append_grep_pattern(&opt
, arg
, "command line", 0,
598 if (!strcmp("--and", arg
)) {
599 append_grep_pattern(&opt
, arg
, "command line", 0,
603 if (!strcmp("--or", arg
))
604 continue; /* no-op */
605 if (!strcmp("(", arg
)) {
606 append_grep_pattern(&opt
, arg
, "command line", 0,
610 if (!strcmp(")", arg
)) {
611 append_grep_pattern(&opt
, arg
, "command line", 0,
615 if (!strcmp("--all-match", arg
)) {
619 if (!strcmp("-e", arg
)) {
621 append_grep_pattern(&opt
, argv
[1],
628 die(emsg_missing_argument
, arg
);
630 if (!strcmp("--full-name", arg
)) {
634 if (!strcmp("--", arg
)) {
635 /* later processing wants to have this at argv[1] */
641 usage(builtin_grep_usage
);
643 /* First unrecognized non-option token */
644 if (!opt
.pattern_list
) {
645 append_grep_pattern(&opt
, arg
, "command line", 0,
650 /* We are looking at the first path or rev;
651 * it is found at argv[1] after leaving the
659 if (!opt
.pattern_list
)
660 die("no pattern given.");
661 if ((opt
.regflags
!= REG_NEWLINE
) && opt
.fixed
)
662 die("cannot mix --fixed-strings and regexp");
663 compile_grep_patterns(&opt
);
665 /* Check revs and then paths */
666 for (i
= 1; i
< argc
; i
++) {
667 const char *arg
= argv
[i
];
668 unsigned char sha1
[20];
670 if (!get_sha1(arg
, sha1
)) {
671 struct object
*object
= parse_object(sha1
);
673 die("bad object %s", arg
);
674 add_object_array(object
, arg
, &list
);
677 if (!strcmp(arg
, "--")) {
684 /* The rest are paths */
685 if (!seen_dashdash
) {
687 for (j
= i
; j
< argc
; j
++)
688 verify_filename(prefix
, argv
[j
]);
692 paths
= get_pathspec(prefix
, argv
+ i
);
693 if (opt
.prefix_length
&& opt
.relative
) {
694 /* Make sure we do not get outside of paths */
695 for (i
= 0; paths
[i
]; i
++)
696 if (strncmp(prefix
, paths
[i
], opt
.prefix_length
))
697 die("git-grep: cannot generate relative filenames containing '..'");
701 paths
= xcalloc(2, sizeof(const char *));
707 return !grep_cache(&opt
, paths
, cached
);
710 die("both --cached and trees are given.");
712 for (i
= 0; i
< list
.nr
; i
++) {
713 struct object
*real_obj
;
714 real_obj
= deref_tag(list
.objects
[i
].item
, NULL
, 0);
715 if (grep_object(&opt
, paths
, real_obj
, list
.objects
[i
].name
))
718 free_grep_patterns(&opt
);