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
;
92 unsigned word_regexp
:1;
95 unsigned post_context
;
98 static void add_pattern(struct grep_opt
*opt
, const char *pat
)
100 struct grep_pat
*p
= xcalloc(1, sizeof(*p
));
102 *opt
->pattern_tail
= p
;
103 opt
->pattern_tail
= &p
->next
;
107 static void compile_patterns(struct grep_opt
*opt
)
110 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
111 int err
= regcomp(&p
->regexp
, p
->pattern
, opt
->regflags
);
114 regerror(err
, &p
->regexp
, errbuf
, 1024);
116 die("'%s': %s", p
->pattern
, errbuf
);
121 static char *end_of_line(char *cp
, unsigned long *left
)
123 unsigned long l
= *left
;
124 while (l
&& *cp
!= '\n') {
132 static int word_char(char ch
)
134 return isalnum(ch
) || ch
== '_';
137 static void show_line(struct grep_opt
*opt
, const char *bol
, const char *eol
,
138 const char *name
, unsigned lno
, char sign
)
140 printf("%s%c", name
, sign
);
142 printf("%d%c", lno
, sign
);
143 printf("%.*s\n", (int)(eol
-bol
), bol
);
146 static int grep_buffer(struct grep_opt
*opt
, const char *name
,
147 char *buf
, unsigned long size
)
150 unsigned long left
= size
;
152 struct pre_context_line
{
155 } *prev
= NULL
, *pcl
;
156 unsigned last_hit
= 0;
157 unsigned last_shown
= 0;
158 const char *hunk_mark
= "";
161 if (opt
->pre_context
)
162 prev
= xcalloc(opt
->pre_context
, sizeof(*prev
));
163 if (opt
->pre_context
|| opt
->post_context
)
167 regmatch_t pmatch
[10];
172 eol
= end_of_line(bol
, &left
);
176 for (p
= opt
->pattern_list
; p
; p
= p
->next
) {
177 regex_t
*exp
= &p
->regexp
;
178 hit
= !regexec(exp
, bol
, ARRAY_SIZE(pmatch
),
181 if (hit
&& opt
->word_regexp
) {
182 /* Match beginning must be either
183 * beginning of the line, or at word
184 * boundary (i.e. the last char must
185 * not be alnum or underscore).
187 if ((pmatch
[0].rm_so
< 0) ||
188 (eol
- bol
) <= pmatch
[0].rm_so
||
189 (pmatch
[0].rm_eo
< 0) ||
190 (eol
- bol
) < pmatch
[0].rm_eo
)
191 die("regexp returned nonsense");
192 if (pmatch
[0].rm_so
!= 0 &&
193 word_char(bol
[pmatch
[0].rm_so
-1]))
194 continue; /* not a word boundary */
195 if ((eol
-bol
) < pmatch
[0].rm_eo
&&
196 word_char(bol
[pmatch
[0].rm_eo
]))
197 continue; /* not a word boundary */
202 /* "grep -v -e foo -e bla" should list lines
203 * that do not have either, so inversion should
210 if (opt
->name_only
) {
211 printf("%s\n", name
);
214 /* Hit at this line. If we haven't shown the
215 * pre-context lines, we would need to show them.
216 * When asked to do "count", this still show
217 * the context which is nonsense, but the user
218 * deserves to get that ;-).
220 if (opt
->pre_context
) {
222 if (opt
->pre_context
< lno
)
223 from
= lno
- opt
->pre_context
;
226 if (from
<= last_shown
)
227 from
= last_shown
+ 1;
228 if (last_shown
&& from
!= last_shown
+ 1)
231 pcl
= &prev
[lno
-from
-1];
232 show_line(opt
, pcl
->bol
, pcl
->eol
,
238 if (last_shown
&& lno
!= last_shown
+ 1)
241 show_line(opt
, bol
, eol
, name
, lno
, ':');
242 last_shown
= last_hit
= lno
;
245 lno
<= last_hit
+ opt
->post_context
) {
246 /* If the last hit is within the post context,
247 * we need to show this line.
249 if (last_shown
&& lno
!= last_shown
+ 1)
251 show_line(opt
, bol
, eol
, name
, lno
, '-');
254 if (opt
->pre_context
) {
255 memmove(prev
+1, prev
,
256 (opt
->pre_context
-1) * sizeof(*prev
));
266 * The real "grep -c foo *.c" gives many "bar.c:0" lines,
267 * which feels mostly useless but sometimes useful. Maybe
268 * make it another option? For now suppress them.
270 if (opt
->count
&& count
)
271 printf("%s:%u\n", name
, count
);
275 static int grep_sha1(struct grep_opt
*opt
, const unsigned char *sha1
, const char *name
)
281 data
= read_sha1_file(sha1
, type
, &size
);
283 error("'%s': unable to read %s", name
, sha1_to_hex(sha1
));
286 hit
= grep_buffer(opt
, name
, data
, size
);
291 static int grep_file(struct grep_opt
*opt
, const char *filename
)
296 if (lstat(filename
, &st
) < 0) {
299 error("'%s': %s", filename
, strerror(errno
));
303 return 0; /* empty file -- no grep hit */
304 if (!S_ISREG(st
.st_mode
))
306 i
= open(filename
, O_RDONLY
);
309 data
= xmalloc(st
.st_size
+ 1);
310 if (st
.st_size
!= xread(i
, data
, st
.st_size
)) {
311 error("'%s': short read %s", filename
, strerror(errno
));
317 i
= grep_buffer(opt
, filename
, data
, st
.st_size
);
322 static int grep_cache(struct grep_opt
*opt
, const char **paths
, int cached
)
328 for (nr
= 0; nr
< active_nr
; nr
++) {
329 struct cache_entry
*ce
= active_cache
[nr
];
330 if (ce_stage(ce
) || !S_ISREG(ntohl(ce
->ce_mode
)))
332 if (!pathspec_matches(paths
, ce
->name
))
335 hit
|= grep_sha1(opt
, ce
->sha1
, ce
->name
);
337 hit
|= grep_file(opt
, ce
->name
);
342 static int grep_tree(struct grep_opt
*opt
, const char **paths
,
343 struct tree_desc
*tree
,
344 const char *tree_name
, const char *base
)
350 const unsigned char *sha1
;
352 char *path_buf
= xmalloc(PATH_MAX
+ strlen(tree_name
) + 100);
355 int offset
= sprintf(path_buf
, "%s:", tree_name
);
356 down
= path_buf
+ offset
;
363 len
= strlen(path_buf
);
367 sha1
= tree_entry_extract(tree
, &path
, &mode
);
368 pathlen
= strlen(path
);
369 strcpy(path_buf
+ len
, path
);
372 /* Match "abc/" against pathspec to
373 * decide if we want to descend into "abc"
376 strcpy(path_buf
+ len
+ pathlen
, "/");
378 if (!pathspec_matches(paths
, down
))
380 else if (S_ISREG(mode
))
381 hit
|= grep_sha1(opt
, sha1
, path_buf
);
382 else if (S_ISDIR(mode
)) {
384 struct tree_desc sub
;
386 data
= read_sha1_file(sha1
, type
, &sub
.size
);
388 die("unable to read tree (%s)",
391 hit
|= grep_tree(opt
, paths
, &sub
, tree_name
, down
);
394 update_tree_entry(tree
);
399 static int grep_object(struct grep_opt
*opt
, const char **paths
,
400 struct object
*obj
, const char *name
)
402 if (!strcmp(obj
->type
, blob_type
))
403 return grep_sha1(opt
, obj
->sha1
, name
);
404 if (!strcmp(obj
->type
, commit_type
) ||
405 !strcmp(obj
->type
, tree_type
)) {
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", obj
->type
);
421 static const char builtin_grep_usage
[] =
422 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
424 int cmd_grep(int argc
, const char **argv
, char **envp
)
427 int no_more_flags
= 0;
428 int seen_noncommit
= 0;
431 struct object_list
*list
, **tail
, *object_list
= NULL
;
432 const char *prefix
= setup_git_directory();
433 const char **paths
= NULL
;
435 memset(&opt
, 0, sizeof(opt
));
436 opt
.pattern_tail
= &opt
.pattern_list
;
437 opt
.regflags
= REG_NEWLINE
;
440 * No point using rev_info, really.
443 const char *arg
= argv
[1];
445 if (!strcmp("--cached", arg
)) {
449 if (!strcmp("-i", arg
) ||
450 !strcmp("--ignore-case", arg
)) {
451 opt
.regflags
|= REG_ICASE
;
454 if (!strcmp("-v", arg
) ||
455 !strcmp("--invert-match", arg
)) {
459 if (!strcmp("-E", arg
) ||
460 !strcmp("--extended-regexp", arg
)) {
461 opt
.regflags
|= REG_EXTENDED
;
464 if (!strcmp("-G", arg
) ||
465 !strcmp("--basic-regexp", arg
)) {
466 opt
.regflags
&= ~REG_EXTENDED
;
469 if (!strcmp("-n", arg
)) {
473 if (!strcmp("-H", arg
)) {
474 /* We always show the pathname, so this
479 if (!strcmp("-l", arg
) ||
480 !strcmp("--files-with-matches", arg
)) {
484 if (!strcmp("-c", arg
) ||
485 !strcmp("--count", arg
)) {
489 if (!strcmp("-w", arg
) ||
490 !strcmp("--word-regexp", arg
)) {
494 if (!strncmp("-A", arg
, 2) ||
495 !strncmp("-B", arg
, 2) ||
496 !strncmp("-C", arg
, 2) ||
497 (arg
[0] == '-' && '1' <= arg
[1] && arg
[1] <= '9')) {
501 case 'A': case 'B': case 'C':
504 usage(builtin_grep_usage
);
515 if (sscanf(scan
, "%u", &num
) != 1)
516 usage(builtin_grep_usage
);
519 opt
.post_context
= num
;
523 opt
.post_context
= num
;
525 opt
.pre_context
= num
;
530 if (!strcmp("-e", arg
)) {
532 add_pattern(&opt
, argv
[1]);
537 usage(builtin_grep_usage
);
539 if (!strcmp("--", arg
)) {
543 /* Either unrecognized option or a single pattern */
544 if (!no_more_flags
&& *arg
== '-')
545 usage(builtin_grep_usage
);
546 if (!opt
.pattern_list
) {
547 add_pattern(&opt
, arg
);
551 /* We are looking at the first path or rev;
552 * it is found at argv[0] after leaving the
559 if (!opt
.pattern_list
)
560 die("no pattern given.");
561 compile_patterns(&opt
);
564 struct object
*object
;
565 struct object_list
*elem
;
566 const char *arg
= argv
[1];
567 unsigned char sha1
[20];
568 if (get_sha1(arg
, sha1
) < 0)
570 object
= parse_object(sha1
);
572 die("bad object %s", arg
);
573 elem
= object_list_insert(object
, tail
);
579 paths
= get_pathspec(prefix
, argv
+ 1);
581 paths
= xcalloc(2, sizeof(const char *));
587 return !grep_cache(&opt
, paths
, cached
);
589 * Do not walk "grep -e foo master next pu -- Documentation/"
590 * but do walk "grep -e foo master..next -- Documentation/".
591 * Ranged request mixed with a blob or tree object, like
592 * "grep -e foo v1.0.0:Documentation/ master..next"
593 * so detect that and complain.
595 for (list
= object_list
; list
; list
= list
->next
) {
596 struct object
*real_obj
;
597 real_obj
= deref_tag(list
->item
, NULL
, 0);
598 if (strcmp(real_obj
->type
, commit_type
))
602 die("both --cached and revisions given.");
604 for (list
= object_list
; list
; list
= list
->next
) {
605 struct object
*real_obj
;
606 real_obj
= deref_tag(list
->item
, NULL
, 0);
607 if (grep_object(&opt
, paths
, real_obj
, list
->name
))