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 *slash
, *cp
;
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 name directory 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
, slash
= NULL
; cp
- match
< matchlen
; cp
++) {
52 if (ch
== '*' || ch
== '[')
56 slash
= match
; /* toplevel */
59 if (namelen
<= slash
- match
) {
60 /* Looking at "Documentation/" and
61 * the pattern says "Documentation/howto/", or
62 * "Documentation/diff*.txt".
64 if (!memcmp(match
, name
, namelen
))
68 /* Looking at "Documentation/howto/" and
69 * the pattern says "Documentation/h*".
71 if (!memcmp(match
, name
, slash
- match
))
86 unsigned post_context
;
89 static char *end_of_line(char *cp
, unsigned long *left
)
91 unsigned long l
= *left
;
92 while (l
&& *cp
!= '\n') {
100 static void show_line(struct grep_opt
*opt
, const char *bol
, const char *eol
,
101 const char *name
, unsigned lno
, char sign
)
103 printf("%s%c", name
, sign
);
105 printf("%d%c", lno
, sign
);
106 printf("%.*s\n", (int)(eol
-bol
), bol
);
109 static int grep_buffer(struct grep_opt
*opt
, const char *name
,
110 char *buf
, unsigned long size
)
113 unsigned long left
= size
;
115 struct pre_context_line
{
118 } *prev
= NULL
, *pcl
;
119 unsigned last_hit
= 0;
120 unsigned last_shown
= 0;
121 const char *hunk_mark
= "";
123 if (opt
->pre_context
)
124 prev
= xcalloc(opt
->pre_context
, sizeof(*prev
));
125 if (opt
->pre_context
|| opt
->post_context
)
129 regmatch_t pmatch
[10];
133 eol
= end_of_line(bol
, &left
);
137 hit
= !regexec(&opt
->regexp
, bol
, ARRAY_SIZE(pmatch
),
142 if (opt
->name_only
) {
143 printf("%s\n", name
);
146 /* Hit at this line. If we haven't shown the
147 * pre-context lines, we would need to show them.
149 if (opt
->pre_context
) {
151 if (opt
->pre_context
< lno
)
152 from
= lno
- opt
->pre_context
;
155 if (from
<= last_shown
)
156 from
= last_shown
+ 1;
157 if (last_shown
&& from
!= last_shown
+ 1)
160 pcl
= &prev
[lno
-from
-1];
161 show_line(opt
, pcl
->bol
, pcl
->eol
,
167 if (last_shown
&& lno
!= last_shown
+ 1)
169 show_line(opt
, bol
, eol
, name
, lno
, ':');
170 last_shown
= last_hit
= lno
;
173 lno
<= last_hit
+ opt
->post_context
) {
174 /* If the last hit is within the post context,
175 * we need to show this line.
177 if (last_shown
&& lno
!= last_shown
+ 1)
179 show_line(opt
, bol
, eol
, name
, lno
, '-');
182 if (opt
->pre_context
) {
183 memmove(prev
+1, prev
,
184 (opt
->pre_context
-1) * sizeof(*prev
));
196 static int grep_sha1(struct grep_opt
*opt
, const unsigned char *sha1
, const char *name
)
202 data
= read_sha1_file(sha1
, type
, &size
);
204 error("'%s': unable to read %s", name
, sha1_to_hex(sha1
));
207 hit
= grep_buffer(opt
, name
, data
, size
);
212 static int grep_file(struct grep_opt
*opt
, const char *filename
)
217 if (lstat(filename
, &st
) < 0) {
220 error("'%s': %s", filename
, strerror(errno
));
224 return 0; /* empty file -- no grep hit */
225 if (!S_ISREG(st
.st_mode
))
227 i
= open(filename
, O_RDONLY
);
230 data
= xmalloc(st
.st_size
+ 1);
231 if (st
.st_size
!= xread(i
, data
, st
.st_size
)) {
232 error("'%s': short read %s", filename
, strerror(errno
));
238 i
= grep_buffer(opt
, filename
, data
, st
.st_size
);
243 static int grep_cache(struct grep_opt
*opt
, const char **paths
, int cached
)
249 for (nr
= 0; nr
< active_nr
; nr
++) {
250 struct cache_entry
*ce
= active_cache
[nr
];
251 if (ce_stage(ce
) || !S_ISREG(ntohl(ce
->ce_mode
)))
253 if (!pathspec_matches(paths
, ce
->name
))
256 hit
|= grep_sha1(opt
, ce
->sha1
, ce
->name
);
258 hit
|= grep_file(opt
, ce
->name
);
263 static int grep_tree(struct grep_opt
*opt
, const char **paths
,
264 struct tree_desc
*tree
,
265 const char *tree_name
, const char *base
)
271 const unsigned char *sha1
;
273 char *path_buf
= xmalloc(PATH_MAX
+ strlen(tree_name
) + 100);
276 int offset
= sprintf(path_buf
, "%s:", tree_name
);
277 down
= path_buf
+ offset
;
284 len
= strlen(path_buf
);
288 sha1
= tree_entry_extract(tree
, &path
, &mode
);
289 pathlen
= strlen(path
);
290 strcpy(path_buf
+ len
, path
);
293 /* Match "abc/" against pathspec to
294 * decide if we want to descend into "abc"
297 strcpy(path_buf
+ len
+ pathlen
, "/");
299 if (!pathspec_matches(paths
, down
))
301 else if (S_ISREG(mode
))
302 hit
|= grep_sha1(opt
, sha1
, path_buf
);
303 else if (S_ISDIR(mode
)) {
305 struct tree_desc sub
;
307 data
= read_sha1_file(sha1
, type
, &sub
.size
);
309 die("unable to read tree (%s)",
312 hit
|= grep_tree(opt
, paths
, &sub
, tree_name
, down
);
315 update_tree_entry(tree
);
320 static int grep_object(struct grep_opt
*opt
, const char **paths
,
321 struct object
*obj
, const char *name
)
323 if (!strcmp(obj
->type
, blob_type
))
324 return grep_sha1(opt
, obj
->sha1
, name
);
325 if (!strcmp(obj
->type
, commit_type
) ||
326 !strcmp(obj
->type
, tree_type
)) {
327 struct tree_desc tree
;
330 data
= read_object_with_reference(obj
->sha1
, tree_type
,
333 die("unable to read tree (%s)", sha1_to_hex(obj
->sha1
));
335 hit
= grep_tree(opt
, paths
, &tree
, name
, "");
339 die("unable to grep from object of type %s", obj
->type
);
342 static const char builtin_grep_usage
[] =
343 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
345 int cmd_grep(int argc
, const char **argv
, char **envp
)
349 int no_more_flags
= 0;
350 int seen_noncommit
= 0;
353 struct object_list
*list
, **tail
, *object_list
= NULL
;
354 const char *prefix
= setup_git_directory();
355 const char **paths
= NULL
;
357 memset(&opt
, 0, sizeof(opt
));
358 opt
.regflags
= REG_NEWLINE
;
361 * No point using rev_info, really.
364 const char *arg
= argv
[1];
366 if (!strcmp("--cached", arg
)) {
370 if (!strcmp("-i", arg
) ||
371 !strcmp("--ignore-case", arg
)) {
372 opt
.regflags
|= REG_ICASE
;
375 if (!strcmp("-v", arg
) ||
376 !strcmp("--invert-match", arg
)) {
380 if (!strcmp("-E", arg
) ||
381 !strcmp("--extended-regexp", arg
)) {
382 opt
.regflags
|= REG_EXTENDED
;
385 if (!strcmp("-G", arg
) ||
386 !strcmp("--basic-regexp", arg
)) {
387 opt
.regflags
&= ~REG_EXTENDED
;
390 if (!strcmp("-n", arg
)) {
394 if (!strcmp("-H", arg
)) {
395 /* We always show the pathname, so this
400 if (!strcmp("-l", arg
) ||
401 !strcmp("--files-with-matches", arg
)) {
405 if (!strcmp("-A", arg
) ||
406 !strcmp("-B", arg
) ||
407 !strcmp("-C", arg
)) {
410 sscanf(*++argv
, "%u", &num
) != 1)
411 usage(builtin_grep_usage
);
415 opt
.post_context
= num
;
418 opt
.post_context
= num
;
420 opt
.pre_context
= num
;
425 if (!strcmp("-e", arg
)) {
427 /* We probably would want to do
428 * -e pat1 -e pat2 as well later...
431 die("more than one pattern?");
432 opt
.pattern
= *++argv
;
436 usage(builtin_grep_usage
);
438 if (!strcmp("--", arg
)) {
442 /* Either unrecognized option or a single pattern */
443 if (!no_more_flags
&& *arg
== '-')
444 usage(builtin_grep_usage
);
450 /* We are looking at the first path or rev;
451 * it is found at argv[0] after leaving the
459 die("no pattern given.");
460 err
= regcomp(&opt
.regexp
, opt
.pattern
, opt
.regflags
);
463 regerror(err
, &opt
.regexp
, errbuf
, 1024);
464 regfree(&opt
.regexp
);
465 die("'%s': %s", opt
.pattern
, errbuf
);
469 struct object
*object
;
470 struct object_list
*elem
;
471 const char *arg
= argv
[1];
472 unsigned char sha1
[20];
473 if (get_sha1(arg
, sha1
) < 0)
475 object
= parse_object(sha1
);
477 die("bad object %s", arg
);
478 elem
= object_list_insert(object
, tail
);
484 paths
= get_pathspec(prefix
, argv
+ 1);
486 paths
= xcalloc(2, sizeof(const char *));
492 return !grep_cache(&opt
, paths
, cached
);
494 * Do not walk "grep -e foo master next pu -- Documentation/"
495 * but do walk "grep -e foo master..next -- Documentation/".
496 * Ranged request mixed with a blob or tree object, like
497 * "grep -e foo v1.0.0:Documentation/ master..next"
498 * so detect that and complain.
500 for (list
= object_list
; list
; list
= list
->next
) {
501 struct object
*real_obj
;
502 real_obj
= deref_tag(list
->item
, NULL
, 0);
503 if (strcmp(real_obj
->type
, commit_type
))
507 die("both --cached and revisions given.");
509 for (list
= object_list
; list
; list
= list
->next
) {
510 struct object
*real_obj
;
511 real_obj
= deref_tag(list
->item
, NULL
, 0);
512 if (grep_object(&opt
, paths
, real_obj
, list
->name
))