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
))
79 struct grep_pat
*next
;
85 struct grep_pat
*pattern_list
;
86 struct grep_pat
**pattern_tail
;
93 unsigned post_context
;
96 static void add_pattern(struct grep_opt
*opt
, const char *pat
)
98 struct grep_pat
*p
= xcalloc(1, sizeof(*p
));
100 *opt
->pattern_tail
= p
;
101 opt
->pattern_tail
= &p
->next
;
105 static void compile_patterns(struct grep_opt
*opt
)
108 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
109 int err
= regcomp(&p
->regexp
, p
->pattern
, opt
->regflags
);
112 regerror(err
, &p
->regexp
, errbuf
, 1024);
114 die("'%s': %s", p
->pattern
, errbuf
);
119 static char *end_of_line(char *cp
, unsigned long *left
)
121 unsigned long l
= *left
;
122 while (l
&& *cp
!= '\n') {
130 static void show_line(struct grep_opt
*opt
, const char *bol
, const char *eol
,
131 const char *name
, unsigned lno
, char sign
)
133 printf("%s%c", name
, sign
);
135 printf("%d%c", lno
, sign
);
136 printf("%.*s\n", (int)(eol
-bol
), bol
);
139 static int grep_buffer(struct grep_opt
*opt
, const char *name
,
140 char *buf
, unsigned long size
)
143 unsigned long left
= size
;
145 struct pre_context_line
{
148 } *prev
= NULL
, *pcl
;
149 unsigned last_hit
= 0;
150 unsigned last_shown
= 0;
151 const char *hunk_mark
= "";
153 if (opt
->pre_context
)
154 prev
= xcalloc(opt
->pre_context
, sizeof(*prev
));
155 if (opt
->pre_context
|| opt
->post_context
)
159 regmatch_t pmatch
[10];
164 eol
= end_of_line(bol
, &left
);
168 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
169 regex_t
*exp
= &p
->regexp
;
170 hit
= !regexec(exp
, bol
, ARRAY_SIZE(pmatch
),
175 /* "grep -v -e foo -e bla" should list lines
176 * that do not have either, so inversion should
182 if (opt
->name_only
) {
183 printf("%s\n", name
);
186 /* Hit at this line. If we haven't shown the
187 * pre-context lines, we would need to show them.
189 if (opt
->pre_context
) {
191 if (opt
->pre_context
< lno
)
192 from
= lno
- opt
->pre_context
;
195 if (from
<= last_shown
)
196 from
= last_shown
+ 1;
197 if (last_shown
&& from
!= last_shown
+ 1)
200 pcl
= &prev
[lno
-from
-1];
201 show_line(opt
, pcl
->bol
, pcl
->eol
,
207 if (last_shown
&& lno
!= last_shown
+ 1)
209 show_line(opt
, bol
, eol
, name
, lno
, ':');
210 last_shown
= last_hit
= lno
;
213 lno
<= last_hit
+ opt
->post_context
) {
214 /* If the last hit is within the post context,
215 * we need to show this line.
217 if (last_shown
&& lno
!= last_shown
+ 1)
219 show_line(opt
, bol
, eol
, name
, lno
, '-');
222 if (opt
->pre_context
) {
223 memmove(prev
+1, prev
,
224 (opt
->pre_context
-1) * sizeof(*prev
));
236 static int grep_sha1(struct grep_opt
*opt
, const unsigned char *sha1
, const char *name
)
242 data
= read_sha1_file(sha1
, type
, &size
);
244 error("'%s': unable to read %s", name
, sha1_to_hex(sha1
));
247 hit
= grep_buffer(opt
, name
, data
, size
);
252 static int grep_file(struct grep_opt
*opt
, const char *filename
)
257 if (lstat(filename
, &st
) < 0) {
260 error("'%s': %s", filename
, strerror(errno
));
264 return 0; /* empty file -- no grep hit */
265 if (!S_ISREG(st
.st_mode
))
267 i
= open(filename
, O_RDONLY
);
270 data
= xmalloc(st
.st_size
+ 1);
271 if (st
.st_size
!= xread(i
, data
, st
.st_size
)) {
272 error("'%s': short read %s", filename
, strerror(errno
));
278 i
= grep_buffer(opt
, filename
, data
, st
.st_size
);
283 static int grep_cache(struct grep_opt
*opt
, const char **paths
, int cached
)
289 for (nr
= 0; nr
< active_nr
; nr
++) {
290 struct cache_entry
*ce
= active_cache
[nr
];
291 if (ce_stage(ce
) || !S_ISREG(ntohl(ce
->ce_mode
)))
293 if (!pathspec_matches(paths
, ce
->name
))
296 hit
|= grep_sha1(opt
, ce
->sha1
, ce
->name
);
298 hit
|= grep_file(opt
, ce
->name
);
303 static int grep_tree(struct grep_opt
*opt
, const char **paths
,
304 struct tree_desc
*tree
,
305 const char *tree_name
, const char *base
)
311 const unsigned char *sha1
;
313 char *path_buf
= xmalloc(PATH_MAX
+ strlen(tree_name
) + 100);
316 int offset
= sprintf(path_buf
, "%s:", tree_name
);
317 down
= path_buf
+ offset
;
324 len
= strlen(path_buf
);
328 sha1
= tree_entry_extract(tree
, &path
, &mode
);
329 pathlen
= strlen(path
);
330 strcpy(path_buf
+ len
, path
);
333 /* Match "abc/" against pathspec to
334 * decide if we want to descend into "abc"
337 strcpy(path_buf
+ len
+ pathlen
, "/");
339 if (!pathspec_matches(paths
, down
))
341 else if (S_ISREG(mode
))
342 hit
|= grep_sha1(opt
, sha1
, path_buf
);
343 else if (S_ISDIR(mode
)) {
345 struct tree_desc sub
;
347 data
= read_sha1_file(sha1
, type
, &sub
.size
);
349 die("unable to read tree (%s)",
352 hit
|= grep_tree(opt
, paths
, &sub
, tree_name
, down
);
355 update_tree_entry(tree
);
360 static int grep_object(struct grep_opt
*opt
, const char **paths
,
361 struct object
*obj
, const char *name
)
363 if (!strcmp(obj
->type
, blob_type
))
364 return grep_sha1(opt
, obj
->sha1
, name
);
365 if (!strcmp(obj
->type
, commit_type
) ||
366 !strcmp(obj
->type
, tree_type
)) {
367 struct tree_desc tree
;
370 data
= read_object_with_reference(obj
->sha1
, tree_type
,
373 die("unable to read tree (%s)", sha1_to_hex(obj
->sha1
));
375 hit
= grep_tree(opt
, paths
, &tree
, name
, "");
379 die("unable to grep from object of type %s", obj
->type
);
382 static const char builtin_grep_usage
[] =
383 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
385 int cmd_grep(int argc
, const char **argv
, char **envp
)
388 int no_more_flags
= 0;
389 int seen_noncommit
= 0;
392 struct object_list
*list
, **tail
, *object_list
= NULL
;
393 const char *prefix
= setup_git_directory();
394 const char **paths
= NULL
;
396 memset(&opt
, 0, sizeof(opt
));
397 opt
.pattern_tail
= &opt
.pattern_list
;
398 opt
.regflags
= REG_NEWLINE
;
401 * No point using rev_info, really.
404 const char *arg
= argv
[1];
406 if (!strcmp("--cached", arg
)) {
410 if (!strcmp("-i", arg
) ||
411 !strcmp("--ignore-case", arg
)) {
412 opt
.regflags
|= REG_ICASE
;
415 if (!strcmp("-v", arg
) ||
416 !strcmp("--invert-match", arg
)) {
420 if (!strcmp("-E", arg
) ||
421 !strcmp("--extended-regexp", arg
)) {
422 opt
.regflags
|= REG_EXTENDED
;
425 if (!strcmp("-G", arg
) ||
426 !strcmp("--basic-regexp", arg
)) {
427 opt
.regflags
&= ~REG_EXTENDED
;
430 if (!strcmp("-n", arg
)) {
434 if (!strcmp("-H", arg
)) {
435 /* We always show the pathname, so this
440 if (!strcmp("-l", arg
) ||
441 !strcmp("--files-with-matches", arg
)) {
445 if (!strncmp("-A", arg
, 2) ||
446 !strncmp("-B", arg
, 2) ||
447 !strncmp("-C", arg
, 2) ||
448 (arg
[0] == '-' && '1' <= arg
[1] && arg
[1] <= '9')) {
452 case 'A': case 'B': case 'C':
455 usage(builtin_grep_usage
);
466 if (sscanf(scan
, "%u", &num
) != 1)
467 usage(builtin_grep_usage
);
470 opt
.post_context
= num
;
474 opt
.post_context
= num
;
476 opt
.pre_context
= num
;
481 if (!strcmp("-e", arg
)) {
483 add_pattern(&opt
, argv
[1]);
488 usage(builtin_grep_usage
);
490 if (!strcmp("--", arg
)) {
494 /* Either unrecognized option or a single pattern */
495 if (!no_more_flags
&& *arg
== '-')
496 usage(builtin_grep_usage
);
497 if (!opt
.pattern_list
) {
498 add_pattern(&opt
, arg
);
502 /* We are looking at the first path or rev;
503 * it is found at argv[0] after leaving the
510 if (!opt
.pattern_list
)
511 die("no pattern given.");
512 compile_patterns(&opt
);
515 struct object
*object
;
516 struct object_list
*elem
;
517 const char *arg
= argv
[1];
518 unsigned char sha1
[20];
519 if (get_sha1(arg
, sha1
) < 0)
521 object
= parse_object(sha1
);
523 die("bad object %s", arg
);
524 elem
= object_list_insert(object
, tail
);
530 paths
= get_pathspec(prefix
, argv
+ 1);
532 paths
= xcalloc(2, sizeof(const char *));
538 return !grep_cache(&opt
, paths
, cached
);
540 * Do not walk "grep -e foo master next pu -- Documentation/"
541 * but do walk "grep -e foo master..next -- Documentation/".
542 * Ranged request mixed with a blob or tree object, like
543 * "grep -e foo v1.0.0:Documentation/ master..next"
544 * so detect that and complain.
546 for (list
= object_list
; list
; list
= list
->next
) {
547 struct object
*real_obj
;
548 real_obj
= deref_tag(list
->item
, NULL
, 0);
549 if (strcmp(real_obj
->type
, commit_type
))
553 die("both --cached and revisions given.");
555 for (list
= object_list
; list
; list
= list
->next
) {
556 struct object
*real_obj
;
557 real_obj
= deref_tag(list
->item
, NULL
, 0);
558 if (grep_object(&opt
, paths
, real_obj
, list
->name
))