2 * This handles recursive filename detection with exclude
3 * files, index knowledge etc..
5 * See Documentation/technical/api-directory-listing.txt
7 * Copyright (C) Linus Torvalds, 2005-2006
8 * Junio Hamano, 2005-2006
13 #include "wildmatch.h"
16 struct path_simplify
{
22 * Tells read_directory_recursive how a file or directory should be treated.
23 * Values are ordered by significance, e.g. if a directory contains both
24 * excluded and untracked files, it is listed as untracked because
25 * path_untracked > path_excluded.
34 static enum path_treatment
read_directory_recursive(struct dir_struct
*dir
,
35 const char *path
, int len
,
36 int check_only
, const struct path_simplify
*simplify
);
37 static int get_dtype(struct dirent
*de
, const char *path
, int len
);
39 /* helper string functions with support for the ignore_case flag */
40 int strcmp_icase(const char *a
, const char *b
)
42 return ignore_case
? strcasecmp(a
, b
) : strcmp(a
, b
);
45 int strncmp_icase(const char *a
, const char *b
, size_t count
)
47 return ignore_case
? strncasecmp(a
, b
, count
) : strncmp(a
, b
, count
);
50 int fnmatch_icase(const char *pattern
, const char *string
, int flags
)
52 return fnmatch(pattern
, string
, flags
| (ignore_case
? FNM_CASEFOLD
: 0));
55 inline int git_fnmatch(const struct pathspec_item
*item
,
56 const char *pattern
, const char *string
,
60 if (ps_strncmp(item
, pattern
, string
, prefix
))
65 if (item
->flags
& PATHSPEC_ONESTAR
) {
66 int pattern_len
= strlen(++pattern
);
67 int string_len
= strlen(string
);
68 return string_len
< pattern_len
||
69 ps_strcmp(item
, pattern
,
70 string
+ string_len
- pattern_len
);
72 if (item
->magic
& PATHSPEC_GLOB
)
73 return wildmatch(pattern
, string
,
75 (item
->magic
& PATHSPEC_ICASE
? WM_CASEFOLD
: 0),
78 /* wildmatch has not learned no FNM_PATHNAME mode yet */
79 return fnmatch(pattern
, string
,
80 item
->magic
& PATHSPEC_ICASE
? FNM_CASEFOLD
: 0);
83 static int fnmatch_icase_mem(const char *pattern
, int patternlen
,
84 const char *string
, int stringlen
,
88 struct strbuf pat_buf
= STRBUF_INIT
;
89 struct strbuf str_buf
= STRBUF_INIT
;
90 const char *use_pat
= pattern
;
91 const char *use_str
= string
;
93 if (pattern
[patternlen
]) {
94 strbuf_add(&pat_buf
, pattern
, patternlen
);
95 use_pat
= pat_buf
.buf
;
97 if (string
[stringlen
]) {
98 strbuf_add(&str_buf
, string
, stringlen
);
99 use_str
= str_buf
.buf
;
103 flags
|= WM_CASEFOLD
;
104 match_status
= wildmatch(use_pat
, use_str
, flags
, NULL
);
106 strbuf_release(&pat_buf
);
107 strbuf_release(&str_buf
);
112 static size_t common_prefix_len(const struct pathspec
*pathspec
)
118 * ":(icase)path" is treated as a pathspec full of
119 * wildcard. In other words, only prefix is considered common
120 * prefix. If the pathspec is abc/foo abc/bar, running in
121 * subdir xyz, the common prefix is still xyz, not xuz/abc as
124 GUARD_PATHSPEC(pathspec
,
131 for (n
= 0; n
< pathspec
->nr
; n
++) {
132 size_t i
= 0, len
= 0, item_len
;
133 if (pathspec
->items
[n
].magic
& PATHSPEC_ICASE
)
134 item_len
= pathspec
->items
[n
].prefix
;
136 item_len
= pathspec
->items
[n
].nowildcard_len
;
137 while (i
< item_len
&& (n
== 0 || i
< max
)) {
138 char c
= pathspec
->items
[n
].match
[i
];
139 if (c
!= pathspec
->items
[0].match
[i
])
145 if (n
== 0 || len
< max
) {
155 * Returns a copy of the longest leading path common among all
158 char *common_prefix(const struct pathspec
*pathspec
)
160 unsigned long len
= common_prefix_len(pathspec
);
162 return len
? xmemdupz(pathspec
->items
[0].match
, len
) : NULL
;
165 int fill_directory(struct dir_struct
*dir
, const struct pathspec
*pathspec
)
170 * Calculate common prefix for the pathspec, and
171 * use that to optimize the directory walk
173 len
= common_prefix_len(pathspec
);
175 /* Read the directory and prune it */
176 read_directory(dir
, pathspec
->nr
? pathspec
->_raw
[0] : "", len
, pathspec
);
180 int within_depth(const char *name
, int namelen
,
181 int depth
, int max_depth
)
183 const char *cp
= name
, *cpe
= name
+ namelen
;
189 if (depth
> max_depth
)
196 * Does 'match' match the given name?
197 * A match is found if
199 * (1) the 'match' string is leading directory of 'name', or
200 * (2) the 'match' string is a wildcard and matches 'name', or
201 * (3) the 'match' string is exactly the same as 'name'.
203 * and the return value tells which case it was.
205 * It returns 0 when there is no match.
207 static int match_pathspec_item(const struct pathspec_item
*item
, int prefix
,
208 const char *name
, int namelen
)
210 /* name/namelen has prefix cut off by caller */
211 const char *match
= item
->match
+ prefix
;
212 int matchlen
= item
->len
- prefix
;
215 * The normal call pattern is:
216 * 1. prefix = common_prefix_len(ps);
217 * 2. prune something, or fill_directory
218 * 3. match_pathspec_depth()
220 * 'prefix' at #1 may be shorter than the command's prefix and
221 * it's ok for #2 to match extra files. Those extras will be
224 * Suppose the pathspec is 'foo' and '../bar' running from
225 * subdir 'xyz'. The common prefix at #1 will be empty, thanks
226 * to "../". We may have xyz/foo _and_ XYZ/foo after #2. The
227 * user does not want XYZ/foo, only the "foo" part should be
228 * case-insensitive. We need to filter out XYZ/foo here. In
229 * other words, we do not trust the caller on comparing the
230 * prefix part when :(icase) is involved. We do exact
231 * comparison ourselves.
233 * Normally the caller (common_prefix_len() in fact) does
234 * _exact_ matching on name[-prefix+1..-1] and we do not need
235 * to check that part. Be defensive and check it anyway, in
236 * case common_prefix_len is changed, or a new caller is
237 * introduced that does not use common_prefix_len.
239 * If the penalty turns out too high when prefix is really
240 * long, maybe change it to
241 * strncmp(match, name, item->prefix - prefix)
243 if (item
->prefix
&& (item
->magic
& PATHSPEC_ICASE
) &&
244 strncmp(item
->match
, name
- prefix
, item
->prefix
))
247 /* If the match was just the prefix, we matched */
249 return MATCHED_RECURSIVELY
;
251 if (matchlen
<= namelen
&& !ps_strncmp(item
, match
, name
, matchlen
)) {
252 if (matchlen
== namelen
)
253 return MATCHED_EXACTLY
;
255 if (match
[matchlen
-1] == '/' || name
[matchlen
] == '/')
256 return MATCHED_RECURSIVELY
;
259 if (item
->nowildcard_len
< item
->len
&&
260 !git_fnmatch(item
, match
, name
,
261 item
->nowildcard_len
- prefix
))
262 return MATCHED_FNMATCH
;
268 * Given a name and a list of pathspecs, returns the nature of the
269 * closest (i.e. most specific) match of the name to any of the
272 * The caller typically calls this multiple times with the same
273 * pathspec and seen[] array but with different name/namelen
274 * (e.g. entries from the index) and is interested in seeing if and
275 * how each pathspec matches all the names it calls this function
276 * with. A mark is left in the seen[] array for each pathspec element
277 * indicating the closest type of match that element achieved, so if
278 * seen[n] remains zero after multiple invocations, that means the nth
279 * pathspec did not match any names, which could indicate that the
280 * user mistyped the nth pathspec.
282 int match_pathspec_depth(const struct pathspec
*ps
,
283 const char *name
, int namelen
,
284 int prefix
, char *seen
)
296 if (!ps
->recursive
||
297 !(ps
->magic
& PATHSPEC_MAXDEPTH
) ||
299 return MATCHED_RECURSIVELY
;
301 if (within_depth(name
, namelen
, 0, ps
->max_depth
))
302 return MATCHED_EXACTLY
;
310 for (i
= ps
->nr
- 1; i
>= 0; i
--) {
312 if (seen
&& seen
[i
] == MATCHED_EXACTLY
)
314 how
= match_pathspec_item(ps
->items
+i
, prefix
, name
, namelen
);
316 (ps
->magic
& PATHSPEC_MAXDEPTH
) &&
317 ps
->max_depth
!= -1 &&
318 how
&& how
!= MATCHED_FNMATCH
) {
319 int len
= ps
->items
[i
].len
;
320 if (name
[len
] == '/')
322 if (within_depth(name
+len
, namelen
-len
, 0, ps
->max_depth
))
323 how
= MATCHED_EXACTLY
;
330 if (seen
&& seen
[i
] < how
)
338 * Return the length of the "simple" part of a path match limiter.
340 int simple_length(const char *match
)
345 unsigned char c
= *match
++;
347 if (c
== '\0' || is_glob_special(c
))
352 int no_wildcard(const char *string
)
354 return string
[simple_length(string
)] == '\0';
357 void parse_exclude_pattern(const char **pattern
,
362 const char *p
= *pattern
;
367 *flags
|= EXC_FLAG_NEGATIVE
;
371 if (len
&& p
[len
- 1] == '/') {
373 *flags
|= EXC_FLAG_MUSTBEDIR
;
375 for (i
= 0; i
< len
; i
++) {
380 *flags
|= EXC_FLAG_NODIR
;
381 *nowildcardlen
= simple_length(p
);
383 * we should have excluded the trailing slash from 'p' too,
384 * but that's one more allocation. Instead just make sure
385 * nowildcardlen does not exceed real patternlen
387 if (*nowildcardlen
> len
)
388 *nowildcardlen
= len
;
389 if (*p
== '*' && no_wildcard(p
+ 1))
390 *flags
|= EXC_FLAG_ENDSWITH
;
395 void add_exclude(const char *string
, const char *base
,
396 int baselen
, struct exclude_list
*el
, int srcpos
)
403 parse_exclude_pattern(&string
, &patternlen
, &flags
, &nowildcardlen
);
404 if (flags
& EXC_FLAG_MUSTBEDIR
) {
406 x
= xmalloc(sizeof(*x
) + patternlen
+ 1);
408 memcpy(s
, string
, patternlen
);
409 s
[patternlen
] = '\0';
412 x
= xmalloc(sizeof(*x
));
415 x
->patternlen
= patternlen
;
416 x
->nowildcardlen
= nowildcardlen
;
418 x
->baselen
= baselen
;
421 ALLOC_GROW(el
->excludes
, el
->nr
+ 1, el
->alloc
);
422 el
->excludes
[el
->nr
++] = x
;
426 static void *read_skip_worktree_file_from_index(const char *path
, size_t *size
)
430 enum object_type type
;
434 pos
= cache_name_pos(path
, len
);
437 if (!ce_skip_worktree(active_cache
[pos
]))
439 data
= read_sha1_file(active_cache
[pos
]->sha1
, &type
, &sz
);
440 if (!data
|| type
!= OBJ_BLOB
) {
449 * Frees memory within el which was allocated for exclude patterns and
450 * the file buffer. Does not free el itself.
452 void clear_exclude_list(struct exclude_list
*el
)
456 for (i
= 0; i
< el
->nr
; i
++)
457 free(el
->excludes
[i
]);
466 static void trim_trailing_spaces(char *buf
)
468 int i
, last_space
= -1, nr_spaces
, len
= strlen(buf
);
469 for (i
= 0; i
< len
; i
++)
472 else if (buf
[i
] == ' ') {
473 if (last_space
== -1) {
481 if (last_space
!= -1 && last_space
+ nr_spaces
== len
)
482 buf
[last_space
] = '\0';
485 int add_excludes_from_file_to_list(const char *fname
,
488 struct exclude_list
*el
,
492 int fd
, i
, lineno
= 1;
496 fd
= open(fname
, O_RDONLY
);
497 if (fd
< 0 || fstat(fd
, &st
) < 0) {
499 warn_on_inaccessible(fname
);
503 (buf
= read_skip_worktree_file_from_index(fname
, &size
)) == NULL
)
509 if (buf
[size
-1] != '\n') {
510 buf
= xrealloc(buf
, size
+1);
515 size
= xsize_t(st
.st_size
);
520 buf
= xmalloc(size
+1);
521 if (read_in_full(fd
, buf
, size
) != size
) {
532 for (i
= 0; i
< size
; i
++) {
533 if (buf
[i
] == '\n') {
534 if (entry
!= buf
+ i
&& entry
[0] != '#') {
535 buf
[i
- (i
&& buf
[i
-1] == '\r')] = 0;
536 trim_trailing_spaces(entry
);
537 add_exclude(entry
, base
, baselen
, el
, lineno
);
546 struct exclude_list
*add_exclude_list(struct dir_struct
*dir
,
547 int group_type
, const char *src
)
549 struct exclude_list
*el
;
550 struct exclude_list_group
*group
;
552 group
= &dir
->exclude_list_group
[group_type
];
553 ALLOC_GROW(group
->el
, group
->nr
+ 1, group
->alloc
);
554 el
= &group
->el
[group
->nr
++];
555 memset(el
, 0, sizeof(*el
));
561 * Used to set up core.excludesfile and .git/info/exclude lists.
563 void add_excludes_from_file(struct dir_struct
*dir
, const char *fname
)
565 struct exclude_list
*el
;
566 el
= add_exclude_list(dir
, EXC_FILE
, fname
);
567 if (add_excludes_from_file_to_list(fname
, "", 0, el
, 0) < 0)
568 die("cannot use %s as an exclude file", fname
);
571 int match_basename(const char *basename
, int basenamelen
,
572 const char *pattern
, int prefix
, int patternlen
,
575 if (prefix
== patternlen
) {
576 if (patternlen
== basenamelen
&&
577 !strncmp_icase(pattern
, basename
, basenamelen
))
579 } else if (flags
& EXC_FLAG_ENDSWITH
) {
580 /* "*literal" matching against "fooliteral" */
581 if (patternlen
- 1 <= basenamelen
&&
582 !strncmp_icase(pattern
+ 1,
583 basename
+ basenamelen
- (patternlen
- 1),
587 if (fnmatch_icase_mem(pattern
, patternlen
,
588 basename
, basenamelen
,
595 int match_pathname(const char *pathname
, int pathlen
,
596 const char *base
, int baselen
,
597 const char *pattern
, int prefix
, int patternlen
,
604 * match with FNM_PATHNAME; the pattern has base implicitly
607 if (*pattern
== '/') {
614 * baselen does not count the trailing slash. base[] may or
615 * may not end with a trailing slash though.
617 if (pathlen
< baselen
+ 1 ||
618 (baselen
&& pathname
[baselen
] != '/') ||
619 strncmp_icase(pathname
, base
, baselen
))
622 namelen
= baselen
? pathlen
- baselen
- 1 : pathlen
;
623 name
= pathname
+ pathlen
- namelen
;
627 * if the non-wildcard part is longer than the
628 * remaining pathname, surely it cannot match.
630 if (prefix
> namelen
)
633 if (strncmp_icase(pattern
, name
, prefix
))
636 patternlen
-= prefix
;
641 * If the whole pattern did not have a wildcard,
642 * then our prefix match is all we need; we
643 * do not need to call fnmatch at all.
645 if (!patternlen
&& !namelen
)
649 return fnmatch_icase_mem(pattern
, patternlen
,
655 * Scan the given exclude list in reverse to see whether pathname
656 * should be ignored. The first match (i.e. the last on the list), if
657 * any, determines the fate. Returns the exclude_list element which
658 * matched, or NULL for undecided.
660 static struct exclude
*last_exclude_matching_from_list(const char *pathname
,
662 const char *basename
,
664 struct exclude_list
*el
)
669 return NULL
; /* undefined */
671 for (i
= el
->nr
- 1; 0 <= i
; i
--) {
672 struct exclude
*x
= el
->excludes
[i
];
673 const char *exclude
= x
->pattern
;
674 int prefix
= x
->nowildcardlen
;
676 if (x
->flags
& EXC_FLAG_MUSTBEDIR
) {
677 if (*dtype
== DT_UNKNOWN
)
678 *dtype
= get_dtype(NULL
, pathname
, pathlen
);
679 if (*dtype
!= DT_DIR
)
683 if (x
->flags
& EXC_FLAG_NODIR
) {
684 if (match_basename(basename
,
685 pathlen
- (basename
- pathname
),
686 exclude
, prefix
, x
->patternlen
,
692 assert(x
->baselen
== 0 || x
->base
[x
->baselen
- 1] == '/');
693 if (match_pathname(pathname
, pathlen
,
694 x
->base
, x
->baselen
? x
->baselen
- 1 : 0,
695 exclude
, prefix
, x
->patternlen
, x
->flags
))
698 return NULL
; /* undecided */
702 * Scan the list and let the last match determine the fate.
703 * Return 1 for exclude, 0 for include and -1 for undecided.
705 int is_excluded_from_list(const char *pathname
,
706 int pathlen
, const char *basename
, int *dtype
,
707 struct exclude_list
*el
)
709 struct exclude
*exclude
;
710 exclude
= last_exclude_matching_from_list(pathname
, pathlen
, basename
, dtype
, el
);
712 return exclude
->flags
& EXC_FLAG_NEGATIVE
? 0 : 1;
713 return -1; /* undecided */
716 static struct exclude
*last_exclude_matching_from_lists(struct dir_struct
*dir
,
717 const char *pathname
, int pathlen
, const char *basename
,
721 struct exclude_list_group
*group
;
722 struct exclude
*exclude
;
723 for (i
= EXC_CMDL
; i
<= EXC_FILE
; i
++) {
724 group
= &dir
->exclude_list_group
[i
];
725 for (j
= group
->nr
- 1; j
>= 0; j
--) {
726 exclude
= last_exclude_matching_from_list(
727 pathname
, pathlen
, basename
, dtype_p
,
737 * Loads the per-directory exclude list for the substring of base
738 * which has a char length of baselen.
740 static void prep_exclude(struct dir_struct
*dir
, const char *base
, int baselen
)
742 struct exclude_list_group
*group
;
743 struct exclude_list
*el
;
744 struct exclude_stack
*stk
= NULL
;
747 group
= &dir
->exclude_list_group
[EXC_DIRS
];
749 /* Pop the exclude lists from the EXCL_DIRS exclude_list_group
750 * which originate from directories not in the prefix of the
751 * path being checked. */
752 while ((stk
= dir
->exclude_stack
) != NULL
) {
753 if (stk
->baselen
<= baselen
&&
754 !strncmp(dir
->basebuf
, base
, stk
->baselen
))
756 el
= &group
->el
[dir
->exclude_stack
->exclude_ix
];
757 dir
->exclude_stack
= stk
->prev
;
759 free((char *)el
->src
); /* see strdup() below */
760 clear_exclude_list(el
);
765 /* Skip traversing into sub directories if the parent is excluded */
769 /* Read from the parent directories and push them down. */
770 current
= stk
? stk
->baselen
: -1;
771 while (current
< baselen
) {
772 struct exclude_stack
*stk
= xcalloc(1, sizeof(*stk
));
780 cp
= strchr(base
+ current
+ 1, '/');
782 die("oops in prep_exclude");
785 stk
->prev
= dir
->exclude_stack
;
786 stk
->baselen
= cp
- base
;
787 stk
->exclude_ix
= group
->nr
;
788 el
= add_exclude_list(dir
, EXC_DIRS
, NULL
);
789 memcpy(dir
->basebuf
+ current
, base
+ current
,
790 stk
->baselen
- current
);
792 /* Abort if the directory is excluded */
795 dir
->basebuf
[stk
->baselen
- 1] = 0;
796 dir
->exclude
= last_exclude_matching_from_lists(dir
,
797 dir
->basebuf
, stk
->baselen
- 1,
798 dir
->basebuf
+ current
, &dt
);
799 dir
->basebuf
[stk
->baselen
- 1] = '/';
801 dir
->exclude
->flags
& EXC_FLAG_NEGATIVE
)
804 dir
->basebuf
[stk
->baselen
] = 0;
805 dir
->exclude_stack
= stk
;
810 /* Try to read per-directory file unless path is too long */
811 if (dir
->exclude_per_dir
&&
812 stk
->baselen
+ strlen(dir
->exclude_per_dir
) < PATH_MAX
) {
813 strcpy(dir
->basebuf
+ stk
->baselen
,
814 dir
->exclude_per_dir
);
816 * dir->basebuf gets reused by the traversal, but we
817 * need fname to remain unchanged to ensure the src
818 * member of each struct exclude correctly
819 * back-references its source file. Other invocations
820 * of add_exclude_list provide stable strings, so we
821 * strdup() and free() here in the caller.
823 el
->src
= strdup(dir
->basebuf
);
824 add_excludes_from_file_to_list(dir
->basebuf
,
825 dir
->basebuf
, stk
->baselen
, el
, 1);
827 dir
->exclude_stack
= stk
;
828 current
= stk
->baselen
;
830 dir
->basebuf
[baselen
] = '\0';
834 * Loads the exclude lists for the directory containing pathname, then
835 * scans all exclude lists to determine whether pathname is excluded.
836 * Returns the exclude_list element which matched, or NULL for
839 struct exclude
*last_exclude_matching(struct dir_struct
*dir
,
840 const char *pathname
,
843 int pathlen
= strlen(pathname
);
844 const char *basename
= strrchr(pathname
, '/');
845 basename
= (basename
) ? basename
+1 : pathname
;
847 prep_exclude(dir
, pathname
, basename
-pathname
);
852 return last_exclude_matching_from_lists(dir
, pathname
, pathlen
,
857 * Loads the exclude lists for the directory containing pathname, then
858 * scans all exclude lists to determine whether pathname is excluded.
859 * Returns 1 if true, otherwise 0.
861 int is_excluded(struct dir_struct
*dir
, const char *pathname
, int *dtype_p
)
863 struct exclude
*exclude
=
864 last_exclude_matching(dir
, pathname
, dtype_p
);
866 return exclude
->flags
& EXC_FLAG_NEGATIVE
? 0 : 1;
870 static struct dir_entry
*dir_entry_new(const char *pathname
, int len
)
872 struct dir_entry
*ent
;
874 ent
= xmalloc(sizeof(*ent
) + len
+ 1);
876 memcpy(ent
->name
, pathname
, len
);
881 static struct dir_entry
*dir_add_name(struct dir_struct
*dir
, const char *pathname
, int len
)
883 if (cache_file_exists(pathname
, len
, ignore_case
))
886 ALLOC_GROW(dir
->entries
, dir
->nr
+1, dir
->alloc
);
887 return dir
->entries
[dir
->nr
++] = dir_entry_new(pathname
, len
);
890 struct dir_entry
*dir_add_ignored(struct dir_struct
*dir
, const char *pathname
, int len
)
892 if (!cache_name_is_other(pathname
, len
))
895 ALLOC_GROW(dir
->ignored
, dir
->ignored_nr
+1, dir
->ignored_alloc
);
896 return dir
->ignored
[dir
->ignored_nr
++] = dir_entry_new(pathname
, len
);
900 index_nonexistent
= 0,
906 * Do not use the alphabetically sorted index to look up
907 * the directory name; instead, use the case insensitive
910 static enum exist_status
directory_exists_in_index_icase(const char *dirname
, int len
)
912 const struct cache_entry
*ce
= cache_dir_exists(dirname
, len
);
913 unsigned char endchar
;
916 return index_nonexistent
;
917 endchar
= ce
->name
[len
];
920 * The cache_entry structure returned will contain this dirname
921 * and possibly additional path components.
924 return index_directory
;
927 * If there are no additional path components, then this cache_entry
928 * represents a submodule. Submodules, despite being directories,
929 * are stored in the cache without a closing slash.
931 if (!endchar
&& S_ISGITLINK(ce
->ce_mode
))
934 /* This should never be hit, but it exists just in case. */
935 return index_nonexistent
;
939 * The index sorts alphabetically by entry name, which
940 * means that a gitlink sorts as '\0' at the end, while
941 * a directory (which is defined not as an entry, but as
942 * the files it contains) will sort with the '/' at the
945 static enum exist_status
directory_exists_in_index(const char *dirname
, int len
)
950 return directory_exists_in_index_icase(dirname
, len
);
952 pos
= cache_name_pos(dirname
, len
);
955 while (pos
< active_nr
) {
956 const struct cache_entry
*ce
= active_cache
[pos
++];
957 unsigned char endchar
;
959 if (strncmp(ce
->name
, dirname
, len
))
961 endchar
= ce
->name
[len
];
965 return index_directory
;
966 if (!endchar
&& S_ISGITLINK(ce
->ce_mode
))
969 return index_nonexistent
;
973 * When we find a directory when traversing the filesystem, we
974 * have three distinct cases:
977 * - see it as a directory
980 * and which one we choose depends on a combination of existing
981 * git index contents and the flags passed into the directory
984 * Case 1: If we *already* have entries in the index under that
985 * directory name, we always recurse into the directory to see
988 * Case 2: If we *already* have that directory name as a gitlink,
989 * we always continue to see it as a gitlink, regardless of whether
990 * there is an actual git directory there or not (it might not
991 * be checked out as a subproject!)
993 * Case 3: if we didn't have it in the index previously, we
994 * have a few sub-cases:
996 * (a) if "show_other_directories" is true, we show it as
997 * just a directory, unless "hide_empty_directories" is
998 * also true, in which case we need to check if it contains any
999 * untracked and / or ignored files.
1000 * (b) if it looks like a git directory, and we don't have
1001 * 'no_gitlinks' set we treat it as a gitlink, and show it
1003 * (c) otherwise, we recurse into it.
1005 static enum path_treatment
treat_directory(struct dir_struct
*dir
,
1006 const char *dirname
, int len
, int exclude
,
1007 const struct path_simplify
*simplify
)
1009 /* The "len-1" is to strip the final '/' */
1010 switch (directory_exists_in_index(dirname
, len
-1)) {
1011 case index_directory
:
1012 return path_recurse
;
1017 case index_nonexistent
:
1018 if (dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
)
1020 if (!(dir
->flags
& DIR_NO_GITLINKS
)) {
1021 unsigned char sha1
[20];
1022 if (resolve_gitlink_ref(dirname
, "HEAD", sha1
) == 0)
1023 return path_untracked
;
1025 return path_recurse
;
1028 /* This is the "show_other_directories" case */
1030 if (!(dir
->flags
& DIR_HIDE_EMPTY_DIRECTORIES
))
1031 return exclude
? path_excluded
: path_untracked
;
1033 return read_directory_recursive(dir
, dirname
, len
, 1, simplify
);
1037 * This is an inexact early pruning of any recursive directory
1038 * reading - if the path cannot possibly be in the pathspec,
1039 * return true, and we'll skip it early.
1041 static int simplify_away(const char *path
, int pathlen
, const struct path_simplify
*simplify
)
1045 const char *match
= simplify
->path
;
1046 int len
= simplify
->len
;
1052 if (!memcmp(path
, match
, len
))
1062 * This function tells us whether an excluded path matches a
1063 * list of "interesting" pathspecs. That is, whether a path matched
1064 * by any of the pathspecs could possibly be ignored by excluding
1065 * the specified path. This can happen if:
1067 * 1. the path is mentioned explicitly in the pathspec
1069 * 2. the path is a directory prefix of some element in the
1072 static int exclude_matches_pathspec(const char *path
, int len
,
1073 const struct path_simplify
*simplify
)
1076 for (; simplify
->path
; simplify
++) {
1077 if (len
== simplify
->len
1078 && !memcmp(path
, simplify
->path
, len
))
1080 if (len
< simplify
->len
1081 && simplify
->path
[len
] == '/'
1082 && !memcmp(path
, simplify
->path
, len
))
1089 static int get_index_dtype(const char *path
, int len
)
1092 const struct cache_entry
*ce
;
1094 ce
= cache_file_exists(path
, len
, 0);
1096 if (!ce_uptodate(ce
))
1098 if (S_ISGITLINK(ce
->ce_mode
))
1101 * Nobody actually cares about the
1102 * difference between DT_LNK and DT_REG
1107 /* Try to look it up as a directory */
1108 pos
= cache_name_pos(path
, len
);
1112 while (pos
< active_nr
) {
1113 ce
= active_cache
[pos
++];
1114 if (strncmp(ce
->name
, path
, len
))
1116 if (ce
->name
[len
] > '/')
1118 if (ce
->name
[len
] < '/')
1120 if (!ce_uptodate(ce
))
1121 break; /* continue? */
1127 static int get_dtype(struct dirent
*de
, const char *path
, int len
)
1129 int dtype
= de
? DTYPE(de
) : DT_UNKNOWN
;
1132 if (dtype
!= DT_UNKNOWN
)
1134 dtype
= get_index_dtype(path
, len
);
1135 if (dtype
!= DT_UNKNOWN
)
1137 if (lstat(path
, &st
))
1139 if (S_ISREG(st
.st_mode
))
1141 if (S_ISDIR(st
.st_mode
))
1143 if (S_ISLNK(st
.st_mode
))
1148 static enum path_treatment
treat_one_path(struct dir_struct
*dir
,
1149 struct strbuf
*path
,
1150 const struct path_simplify
*simplify
,
1151 int dtype
, struct dirent
*de
)
1154 int has_path_in_index
= !!cache_file_exists(path
->buf
, path
->len
, ignore_case
);
1156 if (dtype
== DT_UNKNOWN
)
1157 dtype
= get_dtype(de
, path
->buf
, path
->len
);
1159 /* Always exclude indexed files */
1160 if (dtype
!= DT_DIR
&& has_path_in_index
)
1164 * When we are looking at a directory P in the working tree,
1165 * there are three cases:
1167 * (1) P exists in the index. Everything inside the directory P in
1168 * the working tree needs to go when P is checked out from the
1171 * (2) P does not exist in the index, but there is P/Q in the index.
1172 * We know P will stay a directory when we check out the contents
1173 * of the index, but we do not know yet if there is a directory
1174 * P/Q in the working tree to be killed, so we need to recurse.
1176 * (3) P does not exist in the index, and there is no P/Q in the index
1177 * to require P to be a directory, either. Only in this case, we
1178 * know that everything inside P will not be killed without
1181 if ((dir
->flags
& DIR_COLLECT_KILLED_ONLY
) &&
1182 (dtype
== DT_DIR
) &&
1183 !has_path_in_index
&&
1184 (directory_exists_in_index(path
->buf
, path
->len
) == index_nonexistent
))
1187 exclude
= is_excluded(dir
, path
->buf
, &dtype
);
1190 * Excluded? If we don't explicitly want to show
1191 * ignored files, ignore it
1193 if (exclude
&& !(dir
->flags
& (DIR_SHOW_IGNORED
|DIR_SHOW_IGNORED_TOO
)))
1194 return path_excluded
;
1200 strbuf_addch(path
, '/');
1201 return treat_directory(dir
, path
->buf
, path
->len
, exclude
,
1205 return exclude
? path_excluded
: path_untracked
;
1209 static enum path_treatment
treat_path(struct dir_struct
*dir
,
1211 struct strbuf
*path
,
1213 const struct path_simplify
*simplify
)
1217 if (is_dot_or_dotdot(de
->d_name
) || !strcmp(de
->d_name
, ".git"))
1219 strbuf_setlen(path
, baselen
);
1220 strbuf_addstr(path
, de
->d_name
);
1221 if (simplify_away(path
->buf
, path
->len
, simplify
))
1225 return treat_one_path(dir
, path
, simplify
, dtype
, de
);
1229 * Read a directory tree. We currently ignore anything but
1230 * directories, regular files and symlinks. That's because git
1231 * doesn't handle them at all yet. Maybe that will change some
1234 * Also, we ignore the name ".git" (even if it is not a directory).
1235 * That likely will not change.
1237 * Returns the most significant path_treatment value encountered in the scan.
1239 static enum path_treatment
read_directory_recursive(struct dir_struct
*dir
,
1240 const char *base
, int baselen
,
1242 const struct path_simplify
*simplify
)
1245 enum path_treatment state
, subdir_state
, dir_state
= path_none
;
1247 struct strbuf path
= STRBUF_INIT
;
1249 strbuf_add(&path
, base
, baselen
);
1251 fdir
= opendir(path
.len
? path
.buf
: ".");
1255 while ((de
= readdir(fdir
)) != NULL
) {
1256 /* check how the file or directory should be treated */
1257 state
= treat_path(dir
, de
, &path
, baselen
, simplify
);
1258 if (state
> dir_state
)
1261 /* recurse into subdir if instructed by treat_path */
1262 if (state
== path_recurse
) {
1263 subdir_state
= read_directory_recursive(dir
, path
.buf
,
1264 path
.len
, check_only
, simplify
);
1265 if (subdir_state
> dir_state
)
1266 dir_state
= subdir_state
;
1270 /* abort early if maximum state has been reached */
1271 if (dir_state
== path_untracked
)
1273 /* skip the dir_add_* part */
1277 /* add the path to the appropriate result list */
1280 if (dir
->flags
& DIR_SHOW_IGNORED
)
1281 dir_add_name(dir
, path
.buf
, path
.len
);
1282 else if ((dir
->flags
& DIR_SHOW_IGNORED_TOO
) ||
1283 ((dir
->flags
& DIR_COLLECT_IGNORED
) &&
1284 exclude_matches_pathspec(path
.buf
, path
.len
,
1286 dir_add_ignored(dir
, path
.buf
, path
.len
);
1289 case path_untracked
:
1290 if (!(dir
->flags
& DIR_SHOW_IGNORED
))
1291 dir_add_name(dir
, path
.buf
, path
.len
);
1300 strbuf_release(&path
);
1305 static int cmp_name(const void *p1
, const void *p2
)
1307 const struct dir_entry
*e1
= *(const struct dir_entry
**)p1
;
1308 const struct dir_entry
*e2
= *(const struct dir_entry
**)p2
;
1310 return cache_name_compare(e1
->name
, e1
->len
,
1314 static struct path_simplify
*create_simplify(const char **pathspec
)
1317 struct path_simplify
*simplify
= NULL
;
1322 for (nr
= 0 ; ; nr
++) {
1325 alloc
= alloc_nr(alloc
);
1326 simplify
= xrealloc(simplify
, alloc
* sizeof(*simplify
));
1328 match
= *pathspec
++;
1331 simplify
[nr
].path
= match
;
1332 simplify
[nr
].len
= simple_length(match
);
1334 simplify
[nr
].path
= NULL
;
1335 simplify
[nr
].len
= 0;
1339 static void free_simplify(struct path_simplify
*simplify
)
1344 static int treat_leading_path(struct dir_struct
*dir
,
1345 const char *path
, int len
,
1346 const struct path_simplify
*simplify
)
1348 struct strbuf sb
= STRBUF_INIT
;
1349 int baselen
, rc
= 0;
1351 int old_flags
= dir
->flags
;
1353 while (len
&& path
[len
- 1] == '/')
1358 dir
->flags
&= ~DIR_SHOW_OTHER_DIRECTORIES
;
1360 cp
= path
+ baselen
+ !!baselen
;
1361 cp
= memchr(cp
, '/', path
+ len
- cp
);
1365 baselen
= cp
- path
;
1366 strbuf_setlen(&sb
, 0);
1367 strbuf_add(&sb
, path
, baselen
);
1368 if (!is_directory(sb
.buf
))
1370 if (simplify_away(sb
.buf
, sb
.len
, simplify
))
1372 if (treat_one_path(dir
, &sb
, simplify
,
1373 DT_DIR
, NULL
) == path_none
)
1374 break; /* do not recurse into it */
1375 if (len
<= baselen
) {
1377 break; /* finished checking */
1380 strbuf_release(&sb
);
1381 dir
->flags
= old_flags
;
1385 int read_directory(struct dir_struct
*dir
, const char *path
, int len
, const struct pathspec
*pathspec
)
1387 struct path_simplify
*simplify
;
1390 * Check out create_simplify()
1393 GUARD_PATHSPEC(pathspec
,
1400 if (has_symlink_leading_path(path
, len
))
1403 simplify
= create_simplify(pathspec
? pathspec
->_raw
: NULL
);
1404 if (!len
|| treat_leading_path(dir
, path
, len
, simplify
))
1405 read_directory_recursive(dir
, path
, len
, 0, simplify
);
1406 free_simplify(simplify
);
1407 qsort(dir
->entries
, dir
->nr
, sizeof(struct dir_entry
*), cmp_name
);
1408 qsort(dir
->ignored
, dir
->ignored_nr
, sizeof(struct dir_entry
*), cmp_name
);
1412 int file_exists(const char *f
)
1415 return lstat(f
, &sb
) == 0;
1419 * Given two normalized paths (a trailing slash is ok), if subdir is
1420 * outside dir, return -1. Otherwise return the offset in subdir that
1421 * can be used as relative path to dir.
1423 int dir_inside_of(const char *subdir
, const char *dir
)
1427 assert(dir
&& subdir
&& *dir
&& *subdir
);
1429 while (*dir
&& *subdir
&& *dir
== *subdir
) {
1435 /* hel[p]/me vs hel[l]/yeah */
1436 if (*dir
&& *subdir
)
1440 return !*dir
? offset
: -1; /* same dir */
1442 /* foo/[b]ar vs foo/[] */
1443 if (is_dir_sep(dir
[-1]))
1444 return is_dir_sep(subdir
[-1]) ? offset
: -1;
1446 /* foo[/]bar vs foo[] */
1447 return is_dir_sep(*subdir
) ? offset
+ 1 : -1;
1450 int is_inside_dir(const char *dir
)
1455 if (!getcwd(cwd
, sizeof(cwd
)))
1456 die_errno("can't find the current directory");
1457 return dir_inside_of(cwd
, dir
) >= 0;
1460 int is_empty_dir(const char *path
)
1462 DIR *dir
= opendir(path
);
1469 while ((e
= readdir(dir
)) != NULL
)
1470 if (!is_dot_or_dotdot(e
->d_name
)) {
1479 static int remove_dir_recurse(struct strbuf
*path
, int flag
, int *kept_up
)
1483 int ret
= 0, original_len
= path
->len
, len
, kept_down
= 0;
1484 int only_empty
= (flag
& REMOVE_DIR_EMPTY_ONLY
);
1485 int keep_toplevel
= (flag
& REMOVE_DIR_KEEP_TOPLEVEL
);
1486 unsigned char submodule_head
[20];
1488 if ((flag
& REMOVE_DIR_KEEP_NESTED_GIT
) &&
1489 !resolve_gitlink_ref(path
->buf
, "HEAD", submodule_head
)) {
1490 /* Do not descend and nuke a nested git work tree. */
1496 flag
&= ~REMOVE_DIR_KEEP_TOPLEVEL
;
1497 dir
= opendir(path
->buf
);
1499 /* an empty dir could be removed even if it is unreadble */
1501 return rmdir(path
->buf
);
1505 if (path
->buf
[original_len
- 1] != '/')
1506 strbuf_addch(path
, '/');
1509 while ((e
= readdir(dir
)) != NULL
) {
1511 if (is_dot_or_dotdot(e
->d_name
))
1514 strbuf_setlen(path
, len
);
1515 strbuf_addstr(path
, e
->d_name
);
1516 if (lstat(path
->buf
, &st
))
1518 else if (S_ISDIR(st
.st_mode
)) {
1519 if (!remove_dir_recurse(path
, flag
, &kept_down
))
1520 continue; /* happy */
1521 } else if (!only_empty
&& !unlink(path
->buf
))
1522 continue; /* happy, too */
1524 /* path too long, stat fails, or non-directory still exists */
1530 strbuf_setlen(path
, original_len
);
1531 if (!ret
&& !keep_toplevel
&& !kept_down
)
1532 ret
= rmdir(path
->buf
);
1535 * report the uplevel that it is not an error that we
1536 * did not rmdir() our directory.
1542 int remove_dir_recursively(struct strbuf
*path
, int flag
)
1544 return remove_dir_recurse(path
, flag
, NULL
);
1547 void setup_standard_excludes(struct dir_struct
*dir
)
1552 dir
->exclude_per_dir
= ".gitignore";
1553 path
= git_path("info/exclude");
1554 if (!excludes_file
) {
1555 home_config_paths(NULL
, &xdg_path
, "ignore");
1556 excludes_file
= xdg_path
;
1558 if (!access_or_warn(path
, R_OK
, 0))
1559 add_excludes_from_file(dir
, path
);
1560 if (excludes_file
&& !access_or_warn(excludes_file
, R_OK
, 0))
1561 add_excludes_from_file(dir
, excludes_file
);
1564 int remove_path(const char *name
)
1568 if (unlink(name
) && errno
!= ENOENT
&& errno
!= ENOTDIR
)
1571 slash
= strrchr(name
, '/');
1573 char *dirs
= xstrdup(name
);
1574 slash
= dirs
+ (slash
- name
);
1577 } while (rmdir(dirs
) == 0 && (slash
= strrchr(dirs
, '/')));
1584 * Frees memory within dir which was allocated for exclude lists and
1585 * the exclude_stack. Does not free dir itself.
1587 void clear_directory(struct dir_struct
*dir
)
1590 struct exclude_list_group
*group
;
1591 struct exclude_list
*el
;
1592 struct exclude_stack
*stk
;
1594 for (i
= EXC_CMDL
; i
<= EXC_FILE
; i
++) {
1595 group
= &dir
->exclude_list_group
[i
];
1596 for (j
= 0; j
< group
->nr
; j
++) {
1599 free((char *)el
->src
);
1600 clear_exclude_list(el
);
1605 stk
= dir
->exclude_stack
;
1607 struct exclude_stack
*prev
= stk
->prev
;