4 * Copyright (c) 2006 Junio C Hamano
18 * git grep pathspecs are somewhat different from diff-tree pathspecs;
19 * pathname wildcards are allowed.
21 static int pathspec_matches(struct diff_options
*opt
, const char *name
)
26 namelen
= strlen(name
);
27 for (i
= 0; i
< opt
->nr_paths
; i
++) {
28 const char *match
= opt
->paths
[i
];
29 int matchlen
= opt
->pathlens
[i
];
30 const char *slash
, *cp
;
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 name directory 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
, slash
= NULL
; cp
- match
< matchlen
; cp
++) {
53 if (ch
== '*' || ch
== '[')
57 slash
= match
; /* toplevel */
60 if (namelen
<= slash
- match
) {
61 /* Looking at "Documentation/" and
62 * the pattern says "Documentation/howto/", or
63 * "Documentation/diff*.txt".
65 if (!memcmp(match
, name
, namelen
))
69 /* Looking at "Documentation/howto/" and
70 * the pattern says "Documentation/h*".
72 if (!memcmp(match
, name
, slash
- match
))
87 unsigned post_context
;
90 static char *end_of_line(char *cp
, unsigned long *left
)
92 unsigned long l
= *left
;
93 while (l
&& *cp
!= '\n') {
101 static void show_line(struct grep_opt
*opt
, const char *bol
, const char *eol
,
102 const char *name
, unsigned lno
, char sign
)
104 printf("%s%c", name
, sign
);
106 printf("%d%c", lno
, sign
);
107 printf("%.*s\n", eol
-bol
, bol
);
110 static int grep_buffer(struct grep_opt
*opt
, const char *name
,
111 char *buf
, unsigned long size
)
114 unsigned long left
= size
;
116 struct pre_context_line
{
119 } *prev
= NULL
, *pcl
;
120 unsigned last_hit
= 0;
121 unsigned last_shown
= 0;
122 const char *hunk_mark
= "";
124 if (opt
->pre_context
)
125 prev
= xcalloc(opt
->pre_context
, sizeof(*prev
));
126 if (opt
->pre_context
|| opt
->post_context
)
130 regmatch_t pmatch
[10];
134 eol
= end_of_line(bol
, &left
);
138 hit
= !regexec(&opt
->regexp
, bol
, ARRAY_SIZE(pmatch
),
143 if (opt
->name_only
) {
144 printf("%s\n", name
);
147 /* Hit at this line. If we haven't shown the
148 * pre-context lines, we would need to show them.
150 if (opt
->pre_context
) {
152 if (opt
->pre_context
< lno
)
153 from
= lno
- opt
->pre_context
;
156 if (from
<= last_shown
)
157 from
= last_shown
+ 1;
158 if (last_shown
&& from
!= last_shown
+ 1)
161 pcl
= &prev
[lno
-from
-1];
162 show_line(opt
, pcl
->bol
, pcl
->eol
,
168 if (last_shown
&& lno
!= last_shown
+ 1)
170 show_line(opt
, bol
, eol
, name
, lno
, ':');
171 last_shown
= last_hit
= lno
;
174 lno
<= last_hit
+ opt
->post_context
) {
175 /* If the last hit is within the post context,
176 * we need to show this line.
178 if (last_shown
&& lno
!= last_shown
+ 1)
180 show_line(opt
, bol
, eol
, name
, lno
, '-');
183 if (opt
->pre_context
) {
184 memmove(prev
+1, prev
,
185 (opt
->pre_context
-1) * sizeof(*prev
));
197 static int grep_sha1(struct grep_opt
*opt
, const unsigned char *sha1
, const char *name
)
203 data
= read_sha1_file(sha1
, type
, &size
);
205 error("'%s': unable to read %s", name
, sha1_to_hex(sha1
));
208 hit
= grep_buffer(opt
, name
, data
, size
);
213 static int grep_file(struct grep_opt
*opt
, const char *filename
)
218 if (lstat(filename
, &st
) < 0) {
221 error("'%s': %s", filename
, strerror(errno
));
225 return 0; /* empty file -- no grep hit */
226 if (!S_ISREG(st
.st_mode
))
228 i
= open(filename
, O_RDONLY
);
231 data
= xmalloc(st
.st_size
+ 1);
232 if (st
.st_size
!= xread(i
, data
, st
.st_size
)) {
233 error("'%s': short read %s", filename
, strerror(errno
));
239 i
= grep_buffer(opt
, filename
, data
, st
.st_size
);
244 static int grep_cache(struct grep_opt
*opt
, struct rev_info
*revs
, int cached
)
250 for (nr
= 0; nr
< active_nr
; nr
++) {
251 struct cache_entry
*ce
= active_cache
[nr
];
252 if (ce_stage(ce
) || !S_ISREG(ntohl(ce
->ce_mode
)))
254 if (!pathspec_matches(&revs
->diffopt
, ce
->name
))
257 hit
|= grep_sha1(opt
, ce
->sha1
, ce
->name
);
259 hit
|= grep_file(opt
, ce
->name
);
264 static int grep_tree(struct grep_opt
*opt
, struct rev_info
*revs
,
265 struct tree_desc
*tree
,
266 const char *tree_name
, const char *base
)
272 const unsigned char *sha1
;
274 char *path_buf
= xmalloc(PATH_MAX
+ strlen(tree_name
) + 100);
277 int offset
= sprintf(path_buf
, "%s:", tree_name
);
278 down
= path_buf
+ offset
;
285 len
= strlen(path_buf
);
289 sha1
= tree_entry_extract(tree
, &path
, &mode
);
290 pathlen
= strlen(path
);
291 strcpy(path_buf
+ len
, path
);
294 /* Match "abc/" against pathspec to
295 * decide if we want to descend into "abc"
298 strcpy(path_buf
+ len
+ pathlen
, "/");
300 if (!pathspec_matches(&revs
->diffopt
, down
))
302 else if (S_ISREG(mode
))
303 hit
|= grep_sha1(opt
, sha1
, path_buf
);
304 else if (S_ISDIR(mode
)) {
306 struct tree_desc sub
;
308 data
= read_sha1_file(sha1
, type
, &sub
.size
);
310 die("unable to read tree (%s)",
313 hit
|= grep_tree(opt
, revs
, &sub
, tree_name
, down
);
316 update_tree_entry(tree
);
321 static int grep_object(struct grep_opt
*opt
, struct rev_info
*revs
,
322 struct object
*obj
, const char *name
)
324 if (!strcmp(obj
->type
, blob_type
))
325 return grep_sha1(opt
, obj
->sha1
, name
);
326 if (!strcmp(obj
->type
, commit_type
) ||
327 !strcmp(obj
->type
, tree_type
)) {
328 struct tree_desc tree
;
331 data
= read_object_with_reference(obj
->sha1
, tree_type
,
334 die("unable to read tree (%s)", sha1_to_hex(obj
->sha1
));
336 hit
= grep_tree(opt
, revs
, &tree
, name
, "");
340 die("unable to grep from object of type %s", obj
->type
);
343 static const char builtin_grep_usage
[] =
344 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
346 int cmd_grep(int argc
, const char **argv
, char **envp
)
349 const char **dst
, **src
;
354 int seen_noncommit
= 0;
357 struct object_list
*list
;
359 memset(&opt
, 0, sizeof(opt
));
360 opt
.regflags
= REG_NEWLINE
;
363 * Interpret and remove the grep options upfront. Sigh...
365 for (dst
= src
= &argv
[1]; src
< argc
+ argv
; ) {
366 const char *arg
= *src
++;
368 if (!strcmp("--", arg
)) {
373 if (!strcmp("--cached", arg
)) {
377 if (!strcmp("-i", arg
) ||
378 !strcmp("--ignore-case", arg
)) {
379 opt
.regflags
|= REG_ICASE
;
382 if (!strcmp("-v", arg
) ||
383 !strcmp("--invert-match", arg
)) {
387 if (!strcmp("-E", arg
) ||
388 !strcmp("--extended-regexp", arg
)) {
389 opt
.regflags
|= REG_EXTENDED
;
392 if (!strcmp("-G", arg
) ||
393 !strcmp("--basic-regexp", arg
)) {
394 opt
.regflags
&= ~REG_EXTENDED
;
397 if (!strcmp("-e", arg
)) {
398 if (src
< argc
+ argv
) {
399 opt
.pattern
= *src
++;
402 usage(builtin_grep_usage
);
404 if (!strcmp("-n", arg
)) {
408 if (!strcmp("-H", arg
)) {
409 /* We always show the pathname, so this
414 if (!strcmp("-l", arg
) ||
415 !strcmp("--files-with-matches", arg
)) {
419 if (!strcmp("-A", arg
) ||
420 !strcmp("-B", arg
) ||
421 !strcmp("-C", arg
)) {
423 if ((argc
+ argv
<= src
) ||
424 sscanf(*src
++, "%u", &num
) != 1)
425 usage(builtin_grep_usage
);
428 opt
.post_context
= num
;
431 opt
.post_context
= num
;
433 opt
.pre_context
= num
;
442 die("no pattern given.");
444 err
= regcomp(&opt
.regexp
, opt
.pattern
, opt
.regflags
);
447 regerror(err
, &opt
.regexp
, errbuf
, 1024);
448 regfree(&opt
.regexp
);
449 die("'%s': %s", opt
.pattern
, errbuf
);
452 init_revisions(&rev
);
454 argc
= setup_revisions(dst
- argv
, argv
, &rev
, NULL
);
457 * Do not walk "grep -e foo master next pu -- Documentation/"
458 * but do walk "grep -e foo master..next -- Documentation/".
459 * Ranged request mixed with a blob or tree object, like
460 * "grep -e foo v1.0.0:Documentation/ master..next"
461 * so detect that and complain.
463 for (list
= rev
.pending_objects
; list
; list
= list
->next
) {
464 struct object
*real_obj
;
465 if (list
->item
->flags
& UNINTERESTING
)
467 real_obj
= deref_tag(list
->item
, NULL
, 0);
468 if (strcmp(real_obj
->type
, commit_type
))
471 if (!rev
.pending_objects
)
472 return !grep_cache(&opt
, &rev
, cached
);
474 die("both --cached and revisions given.");
476 if (seen_range
&& seen_noncommit
)
477 die("both A..B and non commit are given.");
479 struct commit
*commit
;
480 prepare_revision_walk(&rev
);
481 while ((commit
= get_revision(&rev
)) != NULL
) {
482 unsigned char *sha1
= commit
->object
.sha1
;
483 const char *n
= find_unique_abbrev(sha1
, rev
.abbrev
);
486 if (grep_object(&opt
, &rev
, &commit
->object
, rev_name
))
488 commit
->buffer
= NULL
;
493 /* all of them are non-commit; do not walk, and
494 * do not lose their names.
496 for (list
= rev
.pending_objects
; list
; list
= list
->next
) {
497 struct object
*real_obj
;
498 real_obj
= deref_tag(list
->item
, NULL
, 0);
499 if (grep_object(&opt
, &rev
, real_obj
, list
->name
))