2 * This handles recursive filename detection with exclude
3 * files, index knowledge etc..
5 * Copyright (C) Linus Torvalds, 2005-2006
6 * Junio Hamano, 2005-2006
11 #include "wildmatch.h"
13 struct path_simplify
{
18 static int read_directory_recursive(struct dir_struct
*dir
, const char *path
, int len
,
19 int check_only
, const struct path_simplify
*simplify
);
20 static int get_dtype(struct dirent
*de
, const char *path
, int len
);
22 /* helper string functions with support for the ignore_case flag */
23 int strcmp_icase(const char *a
, const char *b
)
25 return ignore_case
? strcasecmp(a
, b
) : strcmp(a
, b
);
28 int strncmp_icase(const char *a
, const char *b
, size_t count
)
30 return ignore_case
? strncasecmp(a
, b
, count
) : strncmp(a
, b
, count
);
33 int fnmatch_icase(const char *pattern
, const char *string
, int flags
)
35 return fnmatch(pattern
, string
, flags
| (ignore_case
? FNM_CASEFOLD
: 0));
38 static size_t common_prefix_len(const char **pathspec
)
40 const char *n
, *first
;
47 while ((n
= *pathspec
++)) {
49 for (i
= 0; first
== n
|| i
< max
; i
++) {
51 if (!c
|| c
!= first
[i
] || is_glob_special(c
))
56 if (first
== n
|| len
< max
) {
66 * Returns a copy of the longest leading path common among all
69 char *common_prefix(const char **pathspec
)
71 unsigned long len
= common_prefix_len(pathspec
);
73 return len
? xmemdupz(*pathspec
, len
) : NULL
;
76 int fill_directory(struct dir_struct
*dir
, const char **pathspec
)
81 * Calculate common prefix for the pathspec, and
82 * use that to optimize the directory walk
84 len
= common_prefix_len(pathspec
);
86 /* Read the directory and prune it */
87 read_directory(dir
, pathspec
? *pathspec
: "", len
, pathspec
);
91 int within_depth(const char *name
, int namelen
,
92 int depth
, int max_depth
)
94 const char *cp
= name
, *cpe
= name
+ namelen
;
100 if (depth
> max_depth
)
107 * Does 'match' match the given name?
108 * A match is found if
110 * (1) the 'match' string is leading directory of 'name', or
111 * (2) the 'match' string is a wildcard and matches 'name', or
112 * (3) the 'match' string is exactly the same as 'name'.
114 * and the return value tells which case it was.
116 * It returns 0 when there is no match.
118 static int match_one(const char *match
, const char *name
, int namelen
)
122 /* If the match was just the prefix, we matched */
124 return MATCHED_RECURSIVELY
;
128 unsigned char c1
= tolower(*match
);
129 unsigned char c2
= tolower(*name
);
130 if (c1
== '\0' || is_glob_special(c1
))
140 unsigned char c1
= *match
;
141 unsigned char c2
= *name
;
142 if (c1
== '\0' || is_glob_special(c1
))
154 * If we don't match the matchstring exactly,
155 * we need to match by fnmatch
157 matchlen
= strlen(match
);
158 if (strncmp_icase(match
, name
, matchlen
))
159 return !fnmatch_icase(match
, name
, 0) ? MATCHED_FNMATCH
: 0;
161 if (namelen
== matchlen
)
162 return MATCHED_EXACTLY
;
163 if (match
[matchlen
-1] == '/' || name
[matchlen
] == '/')
164 return MATCHED_RECURSIVELY
;
169 * Given a name and a list of pathspecs, see if the name matches
170 * any of the pathspecs. The caller is also interested in seeing
171 * all pathspec matches some names it calls this function with
172 * (otherwise the user could have mistyped the unmatched pathspec),
173 * and a mark is left in seen[] array for pathspec element that
174 * actually matched anything.
176 int match_pathspec(const char **pathspec
, const char *name
, int namelen
,
177 int prefix
, char *seen
)
187 for (i
= 0; pathspec
[i
] != NULL
; i
++) {
189 const char *match
= pathspec
[i
] + prefix
;
190 if (seen
&& seen
[i
] == MATCHED_EXACTLY
)
192 how
= match_one(match
, name
, namelen
);
196 if (seen
&& seen
[i
] < how
)
204 * Does 'match' match the given name?
205 * A match is found if
207 * (1) the 'match' string is leading directory of 'name', or
208 * (2) the 'match' string is a wildcard and matches 'name', or
209 * (3) the 'match' string is exactly the same as 'name'.
211 * and the return value tells which case it was.
213 * It returns 0 when there is no match.
215 static int match_pathspec_item(const struct pathspec_item
*item
, int prefix
,
216 const char *name
, int namelen
)
218 /* name/namelen has prefix cut off by caller */
219 const char *match
= item
->match
+ prefix
;
220 int matchlen
= item
->len
- prefix
;
222 /* If the match was just the prefix, we matched */
224 return MATCHED_RECURSIVELY
;
226 if (matchlen
<= namelen
&& !strncmp(match
, name
, matchlen
)) {
227 if (matchlen
== namelen
)
228 return MATCHED_EXACTLY
;
230 if (match
[matchlen
-1] == '/' || name
[matchlen
] == '/')
231 return MATCHED_RECURSIVELY
;
234 if (item
->use_wildcard
&& !fnmatch(match
, name
, 0))
235 return MATCHED_FNMATCH
;
241 * Given a name and a list of pathspecs, see if the name matches
242 * any of the pathspecs. The caller is also interested in seeing
243 * all pathspec matches some names it calls this function with
244 * (otherwise the user could have mistyped the unmatched pathspec),
245 * and a mark is left in seen[] array for pathspec element that
246 * actually matched anything.
248 int match_pathspec_depth(const struct pathspec
*ps
,
249 const char *name
, int namelen
,
250 int prefix
, char *seen
)
255 if (!ps
->recursive
|| ps
->max_depth
== -1)
256 return MATCHED_RECURSIVELY
;
258 if (within_depth(name
, namelen
, 0, ps
->max_depth
))
259 return MATCHED_EXACTLY
;
267 for (i
= ps
->nr
- 1; i
>= 0; i
--) {
269 if (seen
&& seen
[i
] == MATCHED_EXACTLY
)
271 how
= match_pathspec_item(ps
->items
+i
, prefix
, name
, namelen
);
272 if (ps
->recursive
&& ps
->max_depth
!= -1 &&
273 how
&& how
!= MATCHED_FNMATCH
) {
274 int len
= ps
->items
[i
].len
;
275 if (name
[len
] == '/')
277 if (within_depth(name
+len
, namelen
-len
, 0, ps
->max_depth
))
278 how
= MATCHED_EXACTLY
;
285 if (seen
&& seen
[i
] < how
)
293 * Return the length of the "simple" part of a path match limiter.
295 static int simple_length(const char *match
)
300 unsigned char c
= *match
++;
302 if (c
== '\0' || is_glob_special(c
))
307 static int no_wildcard(const char *string
)
309 return string
[simple_length(string
)] == '\0';
312 void parse_exclude_pattern(const char **pattern
,
317 const char *p
= *pattern
;
322 *flags
|= EXC_FLAG_NEGATIVE
;
326 if (len
&& p
[len
- 1] == '/') {
328 *flags
|= EXC_FLAG_MUSTBEDIR
;
330 for (i
= 0; i
< len
; i
++) {
335 *flags
|= EXC_FLAG_NODIR
;
336 *nowildcardlen
= simple_length(p
);
338 * we should have excluded the trailing slash from 'p' too,
339 * but that's one more allocation. Instead just make sure
340 * nowildcardlen does not exceed real patternlen
342 if (*nowildcardlen
> len
)
343 *nowildcardlen
= len
;
344 if (*p
== '*' && no_wildcard(p
+ 1))
345 *flags
|= EXC_FLAG_ENDSWITH
;
350 void add_exclude(const char *string
, const char *base
,
351 int baselen
, struct exclude_list
*which
)
358 parse_exclude_pattern(&string
, &patternlen
, &flags
, &nowildcardlen
);
359 if (flags
& EXC_FLAG_MUSTBEDIR
) {
361 x
= xmalloc(sizeof(*x
) + patternlen
+ 1);
363 memcpy(s
, string
, patternlen
);
364 s
[patternlen
] = '\0';
367 x
= xmalloc(sizeof(*x
));
370 x
->patternlen
= patternlen
;
371 x
->nowildcardlen
= nowildcardlen
;
373 x
->baselen
= baselen
;
375 ALLOC_GROW(which
->excludes
, which
->nr
+ 1, which
->alloc
);
376 which
->excludes
[which
->nr
++] = x
;
379 static void *read_skip_worktree_file_from_index(const char *path
, size_t *size
)
383 enum object_type type
;
385 struct index_state
*istate
= &the_index
;
388 pos
= index_name_pos(istate
, path
, len
);
391 if (!ce_skip_worktree(istate
->cache
[pos
]))
393 data
= read_sha1_file(istate
->cache
[pos
]->sha1
, &type
, &sz
);
394 if (!data
|| type
!= OBJ_BLOB
) {
402 void free_excludes(struct exclude_list
*el
)
406 for (i
= 0; i
< el
->nr
; i
++)
407 free(el
->excludes
[i
]);
414 int add_excludes_from_file_to_list(const char *fname
,
418 struct exclude_list
*which
,
426 fd
= open(fname
, O_RDONLY
);
427 if (fd
< 0 || fstat(fd
, &st
) < 0) {
431 (buf
= read_skip_worktree_file_from_index(fname
, &size
)) == NULL
)
437 if (buf
[size
-1] != '\n') {
438 buf
= xrealloc(buf
, size
+1);
443 size
= xsize_t(st
.st_size
);
448 buf
= xmalloc(size
+1);
449 if (read_in_full(fd
, buf
, size
) != size
) {
461 for (i
= 0; i
< size
; i
++) {
462 if (buf
[i
] == '\n') {
463 if (entry
!= buf
+ i
&& entry
[0] != '#') {
464 buf
[i
- (i
&& buf
[i
-1] == '\r')] = 0;
465 add_exclude(entry
, base
, baselen
, which
);
473 void add_excludes_from_file(struct dir_struct
*dir
, const char *fname
)
475 if (add_excludes_from_file_to_list(fname
, "", 0, NULL
,
476 &dir
->exclude_list
[EXC_FILE
], 0) < 0)
477 die("cannot use %s as an exclude file", fname
);
480 static void prep_exclude(struct dir_struct
*dir
, const char *base
, int baselen
)
482 struct exclude_list
*el
;
483 struct exclude_stack
*stk
= NULL
;
486 if ((!dir
->exclude_per_dir
) ||
487 (baselen
+ strlen(dir
->exclude_per_dir
) >= PATH_MAX
))
488 return; /* too long a path -- ignore */
490 /* Pop the ones that are not the prefix of the path being checked. */
491 el
= &dir
->exclude_list
[EXC_DIRS
];
492 while ((stk
= dir
->exclude_stack
) != NULL
) {
493 if (stk
->baselen
<= baselen
&&
494 !strncmp(dir
->basebuf
, base
, stk
->baselen
))
496 dir
->exclude_stack
= stk
->prev
;
497 while (stk
->exclude_ix
< el
->nr
)
498 free(el
->excludes
[--el
->nr
]);
503 /* Read from the parent directories and push them down. */
504 current
= stk
? stk
->baselen
: -1;
505 while (current
< baselen
) {
506 struct exclude_stack
*stk
= xcalloc(1, sizeof(*stk
));
514 cp
= strchr(base
+ current
+ 1, '/');
516 die("oops in prep_exclude");
519 stk
->prev
= dir
->exclude_stack
;
520 stk
->baselen
= cp
- base
;
521 stk
->exclude_ix
= el
->nr
;
522 memcpy(dir
->basebuf
+ current
, base
+ current
,
523 stk
->baselen
- current
);
524 strcpy(dir
->basebuf
+ stk
->baselen
, dir
->exclude_per_dir
);
525 add_excludes_from_file_to_list(dir
->basebuf
,
526 dir
->basebuf
, stk
->baselen
,
527 &stk
->filebuf
, el
, 1);
528 dir
->exclude_stack
= stk
;
529 current
= stk
->baselen
;
531 dir
->basebuf
[baselen
] = '\0';
534 int match_basename(const char *basename
, int basenamelen
,
535 const char *pattern
, int prefix
, int patternlen
,
538 if (prefix
== patternlen
) {
539 if (!strcmp_icase(pattern
, basename
))
541 } else if (flags
& EXC_FLAG_ENDSWITH
) {
542 if (patternlen
- 1 <= basenamelen
&&
543 !strcmp_icase(pattern
+ 1,
544 basename
+ basenamelen
- patternlen
+ 1))
547 if (fnmatch_icase(pattern
, basename
, 0) == 0)
553 int match_pathname(const char *pathname
, int pathlen
,
554 const char *base
, int baselen
,
555 const char *pattern
, int prefix
, int patternlen
,
562 * match with FNM_PATHNAME; the pattern has base implicitly
565 if (*pattern
== '/') {
571 * baselen does not count the trailing slash. base[] may or
572 * may not end with a trailing slash though.
574 if (pathlen
< baselen
+ 1 ||
575 (baselen
&& pathname
[baselen
] != '/') ||
576 strncmp_icase(pathname
, base
, baselen
))
579 namelen
= baselen
? pathlen
- baselen
- 1 : pathlen
;
580 name
= pathname
+ pathlen
- namelen
;
584 * if the non-wildcard part is longer than the
585 * remaining pathname, surely it cannot match.
587 if (prefix
> namelen
)
590 if (strncmp_icase(pattern
, name
, prefix
))
597 return wildmatch(pattern
, name
,
598 ignore_case
? WM_CASEFOLD
: 0,
602 /* Scan the list and let the last match determine the fate.
603 * Return 1 for exclude, 0 for include and -1 for undecided.
605 int excluded_from_list(const char *pathname
,
606 int pathlen
, const char *basename
, int *dtype
,
607 struct exclude_list
*el
)
612 return -1; /* undefined */
614 for (i
= el
->nr
- 1; 0 <= i
; i
--) {
615 struct exclude
*x
= el
->excludes
[i
];
616 const char *exclude
= x
->pattern
;
617 int to_exclude
= x
->flags
& EXC_FLAG_NEGATIVE
? 0 : 1;
618 int prefix
= x
->nowildcardlen
;
620 if (x
->flags
& EXC_FLAG_MUSTBEDIR
) {
621 if (*dtype
== DT_UNKNOWN
)
622 *dtype
= get_dtype(NULL
, pathname
, pathlen
);
623 if (*dtype
!= DT_DIR
)
627 if (x
->flags
& EXC_FLAG_NODIR
) {
628 if (match_basename(basename
,
629 pathlen
- (basename
- pathname
),
630 exclude
, prefix
, x
->patternlen
,
636 assert(x
->baselen
== 0 || x
->base
[x
->baselen
- 1] == '/');
637 if (match_pathname(pathname
, pathlen
,
638 x
->base
, x
->baselen
? x
->baselen
- 1 : 0,
639 exclude
, prefix
, x
->patternlen
, x
->flags
))
642 return -1; /* undecided */
645 static int excluded(struct dir_struct
*dir
, const char *pathname
, int *dtype_p
)
647 int pathlen
= strlen(pathname
);
649 const char *basename
= strrchr(pathname
, '/');
650 basename
= (basename
) ? basename
+1 : pathname
;
652 prep_exclude(dir
, pathname
, basename
-pathname
);
653 for (st
= EXC_CMDL
; st
<= EXC_FILE
; st
++) {
654 switch (excluded_from_list(pathname
, pathlen
, basename
,
655 dtype_p
, &dir
->exclude_list
[st
])) {
665 void path_exclude_check_init(struct path_exclude_check
*check
,
666 struct dir_struct
*dir
)
669 strbuf_init(&check
->path
, 256);
672 void path_exclude_check_clear(struct path_exclude_check
*check
)
674 strbuf_release(&check
->path
);
678 * Is this name excluded? This is for a caller like show_files() that
679 * do not honor directory hierarchy and iterate through paths that are
680 * possibly in an ignored directory.
682 * A path to a directory known to be excluded is left in check->path to
683 * optimize for repeated checks for files in the same excluded directory.
685 int path_excluded(struct path_exclude_check
*check
,
686 const char *name
, int namelen
, int *dtype
)
689 struct strbuf
*path
= &check
->path
;
692 * we allow the caller to pass namelen as an optimization; it
693 * must match the length of the name, as we eventually call
694 * excluded() on the whole name string.
697 namelen
= strlen(name
);
700 path
->len
<= namelen
&&
701 !memcmp(name
, path
->buf
, path
->len
) &&
702 (!name
[path
->len
] || name
[path
->len
] == '/'))
705 strbuf_setlen(path
, 0);
706 for (i
= 0; name
[i
]; i
++) {
711 if (excluded(check
->dir
, path
->buf
, &dt
))
714 strbuf_addch(path
, ch
);
717 /* An entry in the index; cannot be a directory with subentries */
718 strbuf_setlen(path
, 0);
720 return excluded(check
->dir
, name
, dtype
);
723 static struct dir_entry
*dir_entry_new(const char *pathname
, int len
)
725 struct dir_entry
*ent
;
727 ent
= xmalloc(sizeof(*ent
) + len
+ 1);
729 memcpy(ent
->name
, pathname
, len
);
734 static struct dir_entry
*dir_add_name(struct dir_struct
*dir
, const char *pathname
, int len
)
736 if (cache_name_exists(pathname
, len
, ignore_case
))
739 ALLOC_GROW(dir
->entries
, dir
->nr
+1, dir
->alloc
);
740 return dir
->entries
[dir
->nr
++] = dir_entry_new(pathname
, len
);
743 struct dir_entry
*dir_add_ignored(struct dir_struct
*dir
, const char *pathname
, int len
)
745 if (!cache_name_is_other(pathname
, len
))
748 ALLOC_GROW(dir
->ignored
, dir
->ignored_nr
+1, dir
->ignored_alloc
);
749 return dir
->ignored
[dir
->ignored_nr
++] = dir_entry_new(pathname
, len
);
753 index_nonexistent
= 0,
759 * Do not use the alphabetically stored index to look up
760 * the directory name; instead, use the case insensitive
763 static enum exist_status
directory_exists_in_index_icase(const char *dirname
, int len
)
765 struct cache_entry
*ce
= index_name_exists(&the_index
, dirname
, len
+ 1, ignore_case
);
766 unsigned char endchar
;
769 return index_nonexistent
;
770 endchar
= ce
->name
[len
];
773 * The cache_entry structure returned will contain this dirname
774 * and possibly additional path components.
777 return index_directory
;
780 * If there are no additional path components, then this cache_entry
781 * represents a submodule. Submodules, despite being directories,
782 * are stored in the cache without a closing slash.
784 if (!endchar
&& S_ISGITLINK(ce
->ce_mode
))
787 /* This should never be hit, but it exists just in case. */
788 return index_nonexistent
;
792 * The index sorts alphabetically by entry name, which
793 * means that a gitlink sorts as '\0' at the end, while
794 * a directory (which is defined not as an entry, but as
795 * the files it contains) will sort with the '/' at the
798 static enum exist_status
directory_exists_in_index(const char *dirname
, int len
)
803 return directory_exists_in_index_icase(dirname
, len
);
805 pos
= cache_name_pos(dirname
, len
);
808 while (pos
< active_nr
) {
809 struct cache_entry
*ce
= active_cache
[pos
++];
810 unsigned char endchar
;
812 if (strncmp(ce
->name
, dirname
, len
))
814 endchar
= ce
->name
[len
];
818 return index_directory
;
819 if (!endchar
&& S_ISGITLINK(ce
->ce_mode
))
822 return index_nonexistent
;
826 * When we find a directory when traversing the filesystem, we
827 * have three distinct cases:
830 * - see it as a directory
833 * and which one we choose depends on a combination of existing
834 * git index contents and the flags passed into the directory
837 * Case 1: If we *already* have entries in the index under that
838 * directory name, we always recurse into the directory to see
841 * Case 2: If we *already* have that directory name as a gitlink,
842 * we always continue to see it as a gitlink, regardless of whether
843 * there is an actual git directory there or not (it might not
844 * be checked out as a subproject!)
846 * Case 3: if we didn't have it in the index previously, we
847 * have a few sub-cases:
849 * (a) if "show_other_directories" is true, we show it as
850 * just a directory, unless "hide_empty_directories" is
851 * also true and the directory is empty, in which case
852 * we just ignore it entirely.
853 * (b) if it looks like a git directory, and we don't have
854 * 'no_gitlinks' set we treat it as a gitlink, and show it
856 * (c) otherwise, we recurse into it.
858 enum directory_treatment
{
861 recurse_into_directory
864 static enum directory_treatment
treat_directory(struct dir_struct
*dir
,
865 const char *dirname
, int len
,
866 const struct path_simplify
*simplify
)
868 /* The "len-1" is to strip the final '/' */
869 switch (directory_exists_in_index(dirname
, len
-1)) {
870 case index_directory
:
871 return recurse_into_directory
;
874 if (dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
)
875 return ignore_directory
;
876 return show_directory
;
878 case index_nonexistent
:
879 if (dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
)
881 if (!(dir
->flags
& DIR_NO_GITLINKS
)) {
882 unsigned char sha1
[20];
883 if (resolve_gitlink_ref(dirname
, "HEAD", sha1
) == 0)
884 return show_directory
;
886 return recurse_into_directory
;
889 /* This is the "show_other_directories" case */
890 if (!(dir
->flags
& DIR_HIDE_EMPTY_DIRECTORIES
))
891 return show_directory
;
892 if (!read_directory_recursive(dir
, dirname
, len
, 1, simplify
))
893 return ignore_directory
;
894 return show_directory
;
898 * This is an inexact early pruning of any recursive directory
899 * reading - if the path cannot possibly be in the pathspec,
900 * return true, and we'll skip it early.
902 static int simplify_away(const char *path
, int pathlen
, const struct path_simplify
*simplify
)
906 const char *match
= simplify
->path
;
907 int len
= simplify
->len
;
913 if (!memcmp(path
, match
, len
))
923 * This function tells us whether an excluded path matches a
924 * list of "interesting" pathspecs. That is, whether a path matched
925 * by any of the pathspecs could possibly be ignored by excluding
926 * the specified path. This can happen if:
928 * 1. the path is mentioned explicitly in the pathspec
930 * 2. the path is a directory prefix of some element in the
933 static int exclude_matches_pathspec(const char *path
, int len
,
934 const struct path_simplify
*simplify
)
937 for (; simplify
->path
; simplify
++) {
938 if (len
== simplify
->len
939 && !memcmp(path
, simplify
->path
, len
))
941 if (len
< simplify
->len
942 && simplify
->path
[len
] == '/'
943 && !memcmp(path
, simplify
->path
, len
))
950 static int get_index_dtype(const char *path
, int len
)
953 struct cache_entry
*ce
;
955 ce
= cache_name_exists(path
, len
, 0);
957 if (!ce_uptodate(ce
))
959 if (S_ISGITLINK(ce
->ce_mode
))
962 * Nobody actually cares about the
963 * difference between DT_LNK and DT_REG
968 /* Try to look it up as a directory */
969 pos
= cache_name_pos(path
, len
);
973 while (pos
< active_nr
) {
974 ce
= active_cache
[pos
++];
975 if (strncmp(ce
->name
, path
, len
))
977 if (ce
->name
[len
] > '/')
979 if (ce
->name
[len
] < '/')
981 if (!ce_uptodate(ce
))
982 break; /* continue? */
988 static int get_dtype(struct dirent
*de
, const char *path
, int len
)
990 int dtype
= de
? DTYPE(de
) : DT_UNKNOWN
;
993 if (dtype
!= DT_UNKNOWN
)
995 dtype
= get_index_dtype(path
, len
);
996 if (dtype
!= DT_UNKNOWN
)
998 if (lstat(path
, &st
))
1000 if (S_ISREG(st
.st_mode
))
1002 if (S_ISDIR(st
.st_mode
))
1004 if (S_ISLNK(st
.st_mode
))
1009 enum path_treatment
{
1015 static enum path_treatment
treat_one_path(struct dir_struct
*dir
,
1016 struct strbuf
*path
,
1017 const struct path_simplify
*simplify
,
1018 int dtype
, struct dirent
*de
)
1020 int exclude
= excluded(dir
, path
->buf
, &dtype
);
1021 if (exclude
&& (dir
->flags
& DIR_COLLECT_IGNORED
)
1022 && exclude_matches_pathspec(path
->buf
, path
->len
, simplify
))
1023 dir_add_ignored(dir
, path
->buf
, path
->len
);
1026 * Excluded? If we don't explicitly want to show
1027 * ignored files, ignore it
1029 if (exclude
&& !(dir
->flags
& DIR_SHOW_IGNORED
))
1030 return path_ignored
;
1032 if (dtype
== DT_UNKNOWN
)
1033 dtype
= get_dtype(de
, path
->buf
, path
->len
);
1036 * Do we want to see just the ignored files?
1037 * We still need to recurse into directories,
1038 * even if we don't ignore them, since the
1039 * directory may contain files that we do..
1041 if (!exclude
&& (dir
->flags
& DIR_SHOW_IGNORED
)) {
1042 if (dtype
!= DT_DIR
)
1043 return path_ignored
;
1048 return path_ignored
;
1050 strbuf_addch(path
, '/');
1051 switch (treat_directory(dir
, path
->buf
, path
->len
, simplify
)) {
1052 case show_directory
:
1053 if (exclude
!= !!(dir
->flags
1054 & DIR_SHOW_IGNORED
))
1055 return path_ignored
;
1057 case recurse_into_directory
:
1058 return path_recurse
;
1059 case ignore_directory
:
1060 return path_ignored
;
1067 return path_handled
;
1070 static enum path_treatment
treat_path(struct dir_struct
*dir
,
1072 struct strbuf
*path
,
1074 const struct path_simplify
*simplify
)
1078 if (is_dot_or_dotdot(de
->d_name
) || !strcmp(de
->d_name
, ".git"))
1079 return path_ignored
;
1080 strbuf_setlen(path
, baselen
);
1081 strbuf_addstr(path
, de
->d_name
);
1082 if (simplify_away(path
->buf
, path
->len
, simplify
))
1083 return path_ignored
;
1086 return treat_one_path(dir
, path
, simplify
, dtype
, de
);
1090 * Read a directory tree. We currently ignore anything but
1091 * directories, regular files and symlinks. That's because git
1092 * doesn't handle them at all yet. Maybe that will change some
1095 * Also, we ignore the name ".git" (even if it is not a directory).
1096 * That likely will not change.
1098 static int read_directory_recursive(struct dir_struct
*dir
,
1099 const char *base
, int baselen
,
1101 const struct path_simplify
*simplify
)
1106 struct strbuf path
= STRBUF_INIT
;
1108 strbuf_add(&path
, base
, baselen
);
1110 fdir
= opendir(path
.len
? path
.buf
: ".");
1114 while ((de
= readdir(fdir
)) != NULL
) {
1115 switch (treat_path(dir
, de
, &path
, baselen
, simplify
)) {
1117 contents
+= read_directory_recursive(dir
, path
.buf
,
1129 dir_add_name(dir
, path
.buf
, path
.len
);
1133 strbuf_release(&path
);
1138 static int cmp_name(const void *p1
, const void *p2
)
1140 const struct dir_entry
*e1
= *(const struct dir_entry
**)p1
;
1141 const struct dir_entry
*e2
= *(const struct dir_entry
**)p2
;
1143 return cache_name_compare(e1
->name
, e1
->len
,
1147 static struct path_simplify
*create_simplify(const char **pathspec
)
1150 struct path_simplify
*simplify
= NULL
;
1155 for (nr
= 0 ; ; nr
++) {
1158 alloc
= alloc_nr(alloc
);
1159 simplify
= xrealloc(simplify
, alloc
* sizeof(*simplify
));
1161 match
= *pathspec
++;
1164 simplify
[nr
].path
= match
;
1165 simplify
[nr
].len
= simple_length(match
);
1167 simplify
[nr
].path
= NULL
;
1168 simplify
[nr
].len
= 0;
1172 static void free_simplify(struct path_simplify
*simplify
)
1177 static int treat_leading_path(struct dir_struct
*dir
,
1178 const char *path
, int len
,
1179 const struct path_simplify
*simplify
)
1181 struct strbuf sb
= STRBUF_INIT
;
1182 int baselen
, rc
= 0;
1185 while (len
&& path
[len
- 1] == '/')
1191 cp
= path
+ baselen
+ !!baselen
;
1192 cp
= memchr(cp
, '/', path
+ len
- cp
);
1196 baselen
= cp
- path
;
1197 strbuf_setlen(&sb
, 0);
1198 strbuf_add(&sb
, path
, baselen
);
1199 if (!is_directory(sb
.buf
))
1201 if (simplify_away(sb
.buf
, sb
.len
, simplify
))
1203 if (treat_one_path(dir
, &sb
, simplify
,
1204 DT_DIR
, NULL
) == path_ignored
)
1205 break; /* do not recurse into it */
1206 if (len
<= baselen
) {
1208 break; /* finished checking */
1211 strbuf_release(&sb
);
1215 int read_directory(struct dir_struct
*dir
, const char *path
, int len
, const char **pathspec
)
1217 struct path_simplify
*simplify
;
1219 if (has_symlink_leading_path(path
, len
))
1222 simplify
= create_simplify(pathspec
);
1223 if (!len
|| treat_leading_path(dir
, path
, len
, simplify
))
1224 read_directory_recursive(dir
, path
, len
, 0, simplify
);
1225 free_simplify(simplify
);
1226 qsort(dir
->entries
, dir
->nr
, sizeof(struct dir_entry
*), cmp_name
);
1227 qsort(dir
->ignored
, dir
->ignored_nr
, sizeof(struct dir_entry
*), cmp_name
);
1231 int file_exists(const char *f
)
1234 return lstat(f
, &sb
) == 0;
1238 * Given two normalized paths (a trailing slash is ok), if subdir is
1239 * outside dir, return -1. Otherwise return the offset in subdir that
1240 * can be used as relative path to dir.
1242 int dir_inside_of(const char *subdir
, const char *dir
)
1246 assert(dir
&& subdir
&& *dir
&& *subdir
);
1248 while (*dir
&& *subdir
&& *dir
== *subdir
) {
1254 /* hel[p]/me vs hel[l]/yeah */
1255 if (*dir
&& *subdir
)
1259 return !*dir
? offset
: -1; /* same dir */
1261 /* foo/[b]ar vs foo/[] */
1262 if (is_dir_sep(dir
[-1]))
1263 return is_dir_sep(subdir
[-1]) ? offset
: -1;
1265 /* foo[/]bar vs foo[] */
1266 return is_dir_sep(*subdir
) ? offset
+ 1 : -1;
1269 int is_inside_dir(const char *dir
)
1274 if (!getcwd(cwd
, sizeof(cwd
)))
1275 die_errno("can't find the current directory");
1276 return dir_inside_of(cwd
, dir
) >= 0;
1279 int is_empty_dir(const char *path
)
1281 DIR *dir
= opendir(path
);
1288 while ((e
= readdir(dir
)) != NULL
)
1289 if (!is_dot_or_dotdot(e
->d_name
)) {
1298 static int remove_dir_recurse(struct strbuf
*path
, int flag
, int *kept_up
)
1302 int ret
= 0, original_len
= path
->len
, len
, kept_down
= 0;
1303 int only_empty
= (flag
& REMOVE_DIR_EMPTY_ONLY
);
1304 int keep_toplevel
= (flag
& REMOVE_DIR_KEEP_TOPLEVEL
);
1305 unsigned char submodule_head
[20];
1307 if ((flag
& REMOVE_DIR_KEEP_NESTED_GIT
) &&
1308 !resolve_gitlink_ref(path
->buf
, "HEAD", submodule_head
)) {
1309 /* Do not descend and nuke a nested git work tree. */
1315 flag
&= ~REMOVE_DIR_KEEP_TOPLEVEL
;
1316 dir
= opendir(path
->buf
);
1318 /* an empty dir could be removed even if it is unreadble */
1320 return rmdir(path
->buf
);
1324 if (path
->buf
[original_len
- 1] != '/')
1325 strbuf_addch(path
, '/');
1328 while ((e
= readdir(dir
)) != NULL
) {
1330 if (is_dot_or_dotdot(e
->d_name
))
1333 strbuf_setlen(path
, len
);
1334 strbuf_addstr(path
, e
->d_name
);
1335 if (lstat(path
->buf
, &st
))
1337 else if (S_ISDIR(st
.st_mode
)) {
1338 if (!remove_dir_recurse(path
, flag
, &kept_down
))
1339 continue; /* happy */
1340 } else if (!only_empty
&& !unlink(path
->buf
))
1341 continue; /* happy, too */
1343 /* path too long, stat fails, or non-directory still exists */
1349 strbuf_setlen(path
, original_len
);
1350 if (!ret
&& !keep_toplevel
&& !kept_down
)
1351 ret
= rmdir(path
->buf
);
1354 * report the uplevel that it is not an error that we
1355 * did not rmdir() our directory.
1361 int remove_dir_recursively(struct strbuf
*path
, int flag
)
1363 return remove_dir_recurse(path
, flag
, NULL
);
1366 void setup_standard_excludes(struct dir_struct
*dir
)
1370 dir
->exclude_per_dir
= ".gitignore";
1371 path
= git_path("info/exclude");
1372 if (!access(path
, R_OK
))
1373 add_excludes_from_file(dir
, path
);
1374 if (excludes_file
&& !access(excludes_file
, R_OK
))
1375 add_excludes_from_file(dir
, excludes_file
);
1378 int remove_path(const char *name
)
1382 if (unlink(name
) && errno
!= ENOENT
)
1385 slash
= strrchr(name
, '/');
1387 char *dirs
= xstrdup(name
);
1388 slash
= dirs
+ (slash
- name
);
1391 } while (rmdir(dirs
) == 0 && (slash
= strrchr(dirs
, '/')));
1397 static int pathspec_item_cmp(const void *a_
, const void *b_
)
1399 struct pathspec_item
*a
, *b
;
1401 a
= (struct pathspec_item
*)a_
;
1402 b
= (struct pathspec_item
*)b_
;
1403 return strcmp(a
->match
, b
->match
);
1406 int init_pathspec(struct pathspec
*pathspec
, const char **paths
)
1408 const char **p
= paths
;
1411 memset(pathspec
, 0, sizeof(*pathspec
));
1416 pathspec
->raw
= paths
;
1417 pathspec
->nr
= p
- paths
;
1421 pathspec
->items
= xmalloc(sizeof(struct pathspec_item
)*pathspec
->nr
);
1422 for (i
= 0; i
< pathspec
->nr
; i
++) {
1423 struct pathspec_item
*item
= pathspec
->items
+i
;
1424 const char *path
= paths
[i
];
1427 item
->len
= strlen(path
);
1428 item
->use_wildcard
= !no_wildcard(path
);
1429 if (item
->use_wildcard
)
1430 pathspec
->has_wildcard
= 1;
1433 qsort(pathspec
->items
, pathspec
->nr
,
1434 sizeof(struct pathspec_item
), pathspec_item_cmp
);
1439 void free_pathspec(struct pathspec
*pathspec
)
1441 free(pathspec
->items
);
1442 pathspec
->items
= NULL
;