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"
15 struct path_simplify
{
20 static int read_directory_recursive(struct dir_struct
*dir
, const char *path
, int len
,
21 int check_only
, const struct path_simplify
*simplify
);
22 static int get_dtype(struct dirent
*de
, const char *path
, int len
);
24 /* helper string functions with support for the ignore_case flag */
25 int strcmp_icase(const char *a
, const char *b
)
27 return ignore_case
? strcasecmp(a
, b
) : strcmp(a
, b
);
30 int strncmp_icase(const char *a
, const char *b
, size_t count
)
32 return ignore_case
? strncasecmp(a
, b
, count
) : strncmp(a
, b
, count
);
35 int fnmatch_icase(const char *pattern
, const char *string
, int flags
)
37 return fnmatch(pattern
, string
, flags
| (ignore_case
? FNM_CASEFOLD
: 0));
40 inline int git_fnmatch(const char *pattern
, const char *string
,
41 int flags
, int prefix
)
44 if (flags
& GFNM_PATHNAME
)
45 fnm_flags
|= FNM_PATHNAME
;
47 if (strncmp(pattern
, string
, prefix
))
52 if (flags
& GFNM_ONESTAR
) {
53 int pattern_len
= strlen(++pattern
);
54 int string_len
= strlen(string
);
55 return string_len
< pattern_len
||
57 string
+ string_len
- pattern_len
);
59 return fnmatch(pattern
, string
, fnm_flags
);
62 static size_t common_prefix_len(const char **pathspec
)
64 const char *n
, *first
;
66 int literal
= limit_pathspec_to_literal();
72 while ((n
= *pathspec
++)) {
74 for (i
= 0; first
== n
|| i
< max
; i
++) {
76 if (!c
|| c
!= first
[i
] || (!literal
&& is_glob_special(c
)))
81 if (first
== n
|| len
< max
) {
91 * Returns a copy of the longest leading path common among all
94 char *common_prefix(const char **pathspec
)
96 unsigned long len
= common_prefix_len(pathspec
);
98 return len
? xmemdupz(*pathspec
, len
) : NULL
;
101 int fill_directory(struct dir_struct
*dir
, const char **pathspec
)
106 * Calculate common prefix for the pathspec, and
107 * use that to optimize the directory walk
109 len
= common_prefix_len(pathspec
);
111 /* Read the directory and prune it */
112 read_directory(dir
, pathspec
? *pathspec
: "", len
, pathspec
);
116 int within_depth(const char *name
, int namelen
,
117 int depth
, int max_depth
)
119 const char *cp
= name
, *cpe
= name
+ namelen
;
125 if (depth
> max_depth
)
132 * Does 'match' match the given name?
133 * A match is found if
135 * (1) the 'match' string is leading directory of 'name', or
136 * (2) the 'match' string is a wildcard and matches 'name', or
137 * (3) the 'match' string is exactly the same as 'name'.
139 * and the return value tells which case it was.
141 * It returns 0 when there is no match.
143 static int match_one(const char *match
, const char *name
, int namelen
)
146 int literal
= limit_pathspec_to_literal();
148 /* If the match was just the prefix, we matched */
150 return MATCHED_RECURSIVELY
;
154 unsigned char c1
= tolower(*match
);
155 unsigned char c2
= tolower(*name
);
156 if (c1
== '\0' || (!literal
&& is_glob_special(c1
)))
166 unsigned char c1
= *match
;
167 unsigned char c2
= *name
;
168 if (c1
== '\0' || (!literal
&& is_glob_special(c1
)))
179 * If we don't match the matchstring exactly,
180 * we need to match by fnmatch
182 matchlen
= strlen(match
);
183 if (strncmp_icase(match
, name
, matchlen
)) {
186 return !fnmatch_icase(match
, name
, 0) ? MATCHED_FNMATCH
: 0;
189 if (namelen
== matchlen
)
190 return MATCHED_EXACTLY
;
191 if (match
[matchlen
-1] == '/' || name
[matchlen
] == '/')
192 return MATCHED_RECURSIVELY
;
197 * Given a name and a list of pathspecs, returns the nature of the
198 * closest (i.e. most specific) match of the name to any of the
201 * The caller typically calls this multiple times with the same
202 * pathspec and seen[] array but with different name/namelen
203 * (e.g. entries from the index) and is interested in seeing if and
204 * how each pathspec matches all the names it calls this function
205 * with. A mark is left in the seen[] array for each pathspec element
206 * indicating the closest type of match that element achieved, so if
207 * seen[n] remains zero after multiple invocations, that means the nth
208 * pathspec did not match any names, which could indicate that the
209 * user mistyped the nth pathspec.
211 int match_pathspec(const char **pathspec
, const char *name
, int namelen
,
212 int prefix
, char *seen
)
222 for (i
= 0; pathspec
[i
] != NULL
; i
++) {
224 const char *match
= pathspec
[i
] + prefix
;
225 if (seen
&& seen
[i
] == MATCHED_EXACTLY
)
227 how
= match_one(match
, name
, namelen
);
231 if (seen
&& seen
[i
] < how
)
239 * Does 'match' match the given name?
240 * A match is found if
242 * (1) the 'match' string is leading directory of 'name', or
243 * (2) the 'match' string is a wildcard and matches 'name', or
244 * (3) the 'match' string is exactly the same as 'name'.
246 * and the return value tells which case it was.
248 * It returns 0 when there is no match.
250 static int match_pathspec_item(const struct pathspec_item
*item
, int prefix
,
251 const char *name
, int namelen
)
253 /* name/namelen has prefix cut off by caller */
254 const char *match
= item
->match
+ prefix
;
255 int matchlen
= item
->len
- prefix
;
257 /* If the match was just the prefix, we matched */
259 return MATCHED_RECURSIVELY
;
261 if (matchlen
<= namelen
&& !strncmp(match
, name
, matchlen
)) {
262 if (matchlen
== namelen
)
263 return MATCHED_EXACTLY
;
265 if (match
[matchlen
-1] == '/' || name
[matchlen
] == '/')
266 return MATCHED_RECURSIVELY
;
269 if (item
->nowildcard_len
< item
->len
&&
270 !git_fnmatch(match
, name
,
271 item
->flags
& PATHSPEC_ONESTAR
? GFNM_ONESTAR
: 0,
272 item
->nowildcard_len
- prefix
))
273 return MATCHED_FNMATCH
;
279 * Given a name and a list of pathspecs, returns the nature of the
280 * closest (i.e. most specific) match of the name to any of the
283 * The caller typically calls this multiple times with the same
284 * pathspec and seen[] array but with different name/namelen
285 * (e.g. entries from the index) and is interested in seeing if and
286 * how each pathspec matches all the names it calls this function
287 * with. A mark is left in the seen[] array for each pathspec element
288 * indicating the closest type of match that element achieved, so if
289 * seen[n] remains zero after multiple invocations, that means the nth
290 * pathspec did not match any names, which could indicate that the
291 * user mistyped the nth pathspec.
293 int match_pathspec_depth(const struct pathspec
*ps
,
294 const char *name
, int namelen
,
295 int prefix
, char *seen
)
300 if (!ps
->recursive
|| ps
->max_depth
== -1)
301 return MATCHED_RECURSIVELY
;
303 if (within_depth(name
, namelen
, 0, ps
->max_depth
))
304 return MATCHED_EXACTLY
;
312 for (i
= ps
->nr
- 1; i
>= 0; i
--) {
314 if (seen
&& seen
[i
] == MATCHED_EXACTLY
)
316 how
= match_pathspec_item(ps
->items
+i
, prefix
, name
, namelen
);
317 if (ps
->recursive
&& 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 static int simple_length(const char *match
)
345 unsigned char c
= *match
++;
347 if (c
== '\0' || is_glob_special(c
))
352 static 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
;
432 struct index_state
*istate
= &the_index
;
435 pos
= index_name_pos(istate
, path
, len
);
438 if (!ce_skip_worktree(istate
->cache
[pos
]))
440 data
= read_sha1_file(istate
->cache
[pos
]->sha1
, &type
, &sz
);
441 if (!data
|| type
!= OBJ_BLOB
) {
450 * Frees memory within el which was allocated for exclude patterns and
451 * the file buffer. Does not free el itself.
453 void clear_exclude_list(struct exclude_list
*el
)
457 for (i
= 0; i
< el
->nr
; i
++)
458 free(el
->excludes
[i
]);
467 int add_excludes_from_file_to_list(const char *fname
,
470 struct exclude_list
*el
,
474 int fd
, i
, lineno
= 1;
478 fd
= open(fname
, O_RDONLY
);
479 if (fd
< 0 || fstat(fd
, &st
) < 0) {
481 warn_on_inaccessible(fname
);
485 (buf
= read_skip_worktree_file_from_index(fname
, &size
)) == NULL
)
491 if (buf
[size
-1] != '\n') {
492 buf
= xrealloc(buf
, size
+1);
497 size
= xsize_t(st
.st_size
);
502 buf
= xmalloc(size
+1);
503 if (read_in_full(fd
, buf
, size
) != size
) {
514 for (i
= 0; i
< size
; i
++) {
515 if (buf
[i
] == '\n') {
516 if (entry
!= buf
+ i
&& entry
[0] != '#') {
517 buf
[i
- (i
&& buf
[i
-1] == '\r')] = 0;
518 add_exclude(entry
, base
, baselen
, el
, lineno
);
527 struct exclude_list
*add_exclude_list(struct dir_struct
*dir
,
528 int group_type
, const char *src
)
530 struct exclude_list
*el
;
531 struct exclude_list_group
*group
;
533 group
= &dir
->exclude_list_group
[group_type
];
534 ALLOC_GROW(group
->el
, group
->nr
+ 1, group
->alloc
);
535 el
= &group
->el
[group
->nr
++];
536 memset(el
, 0, sizeof(*el
));
542 * Used to set up core.excludesfile and .git/info/exclude lists.
544 void add_excludes_from_file(struct dir_struct
*dir
, const char *fname
)
546 struct exclude_list
*el
;
547 el
= add_exclude_list(dir
, EXC_FILE
, fname
);
548 if (add_excludes_from_file_to_list(fname
, "", 0, el
, 0) < 0)
549 die("cannot use %s as an exclude file", fname
);
553 * Loads the per-directory exclude list for the substring of base
554 * which has a char length of baselen.
556 static void prep_exclude(struct dir_struct
*dir
, const char *base
, int baselen
)
558 struct exclude_list_group
*group
;
559 struct exclude_list
*el
;
560 struct exclude_stack
*stk
= NULL
;
563 if ((!dir
->exclude_per_dir
) ||
564 (baselen
+ strlen(dir
->exclude_per_dir
) >= PATH_MAX
))
565 return; /* too long a path -- ignore */
567 group
= &dir
->exclude_list_group
[EXC_DIRS
];
569 /* Pop the exclude lists from the EXCL_DIRS exclude_list_group
570 * which originate from directories not in the prefix of the
571 * path being checked. */
572 while ((stk
= dir
->exclude_stack
) != NULL
) {
573 if (stk
->baselen
<= baselen
&&
574 !strncmp(dir
->basebuf
, base
, stk
->baselen
))
576 el
= &group
->el
[dir
->exclude_stack
->exclude_ix
];
577 dir
->exclude_stack
= stk
->prev
;
578 free((char *)el
->src
); /* see strdup() below */
579 clear_exclude_list(el
);
584 /* Read from the parent directories and push them down. */
585 current
= stk
? stk
->baselen
: -1;
586 while (current
< baselen
) {
587 struct exclude_stack
*stk
= xcalloc(1, sizeof(*stk
));
595 cp
= strchr(base
+ current
+ 1, '/');
597 die("oops in prep_exclude");
600 stk
->prev
= dir
->exclude_stack
;
601 stk
->baselen
= cp
- base
;
602 memcpy(dir
->basebuf
+ current
, base
+ current
,
603 stk
->baselen
- current
);
604 strcpy(dir
->basebuf
+ stk
->baselen
, dir
->exclude_per_dir
);
606 * dir->basebuf gets reused by the traversal, but we
607 * need fname to remain unchanged to ensure the src
608 * member of each struct exclude correctly
609 * back-references its source file. Other invocations
610 * of add_exclude_list provide stable strings, so we
611 * strdup() and free() here in the caller.
613 el
= add_exclude_list(dir
, EXC_DIRS
, strdup(dir
->basebuf
));
614 stk
->exclude_ix
= group
->nr
- 1;
615 add_excludes_from_file_to_list(dir
->basebuf
,
616 dir
->basebuf
, stk
->baselen
,
618 dir
->exclude_stack
= stk
;
619 current
= stk
->baselen
;
621 dir
->basebuf
[baselen
] = '\0';
624 int match_basename(const char *basename
, int basenamelen
,
625 const char *pattern
, int prefix
, int patternlen
,
628 if (prefix
== patternlen
) {
629 if (!strcmp_icase(pattern
, basename
))
631 } else if (flags
& EXC_FLAG_ENDSWITH
) {
632 if (patternlen
- 1 <= basenamelen
&&
633 !strcmp_icase(pattern
+ 1,
634 basename
+ basenamelen
- patternlen
+ 1))
637 if (fnmatch_icase(pattern
, basename
, 0) == 0)
643 int match_pathname(const char *pathname
, int pathlen
,
644 const char *base
, int baselen
,
645 const char *pattern
, int prefix
, int patternlen
,
652 * match with FNM_PATHNAME; the pattern has base implicitly
655 if (*pattern
== '/') {
661 * baselen does not count the trailing slash. base[] may or
662 * may not end with a trailing slash though.
664 if (pathlen
< baselen
+ 1 ||
665 (baselen
&& pathname
[baselen
] != '/') ||
666 strncmp_icase(pathname
, base
, baselen
))
669 namelen
= baselen
? pathlen
- baselen
- 1 : pathlen
;
670 name
= pathname
+ pathlen
- namelen
;
674 * if the non-wildcard part is longer than the
675 * remaining pathname, surely it cannot match.
677 if (prefix
> namelen
)
680 if (strncmp_icase(pattern
, name
, prefix
))
687 return wildmatch(pattern
, name
,
688 ignore_case
? FNM_CASEFOLD
: 0) == 0;
692 * Scan the given exclude list in reverse to see whether pathname
693 * should be ignored. The first match (i.e. the last on the list), if
694 * any, determines the fate. Returns the exclude_list element which
695 * matched, or NULL for undecided.
697 static struct exclude
*last_exclude_matching_from_list(const char *pathname
,
699 const char *basename
,
701 struct exclude_list
*el
)
706 return NULL
; /* undefined */
708 for (i
= el
->nr
- 1; 0 <= i
; i
--) {
709 struct exclude
*x
= el
->excludes
[i
];
710 const char *exclude
= x
->pattern
;
711 int prefix
= x
->nowildcardlen
;
713 if (x
->flags
& EXC_FLAG_MUSTBEDIR
) {
714 if (*dtype
== DT_UNKNOWN
)
715 *dtype
= get_dtype(NULL
, pathname
, pathlen
);
716 if (*dtype
!= DT_DIR
)
720 if (x
->flags
& EXC_FLAG_NODIR
) {
721 if (match_basename(basename
,
722 pathlen
- (basename
- pathname
),
723 exclude
, prefix
, x
->patternlen
,
729 assert(x
->baselen
== 0 || x
->base
[x
->baselen
- 1] == '/');
730 if (match_pathname(pathname
, pathlen
,
731 x
->base
, x
->baselen
? x
->baselen
- 1 : 0,
732 exclude
, prefix
, x
->patternlen
, x
->flags
))
735 return NULL
; /* undecided */
739 * Scan the list and let the last match determine the fate.
740 * Return 1 for exclude, 0 for include and -1 for undecided.
742 int is_excluded_from_list(const char *pathname
,
743 int pathlen
, const char *basename
, int *dtype
,
744 struct exclude_list
*el
)
746 struct exclude
*exclude
;
747 exclude
= last_exclude_matching_from_list(pathname
, pathlen
, basename
, dtype
, el
);
749 return exclude
->flags
& EXC_FLAG_NEGATIVE
? 0 : 1;
750 return -1; /* undecided */
754 * Loads the exclude lists for the directory containing pathname, then
755 * scans all exclude lists to determine whether pathname is excluded.
756 * Returns the exclude_list element which matched, or NULL for
759 static struct exclude
*last_exclude_matching(struct dir_struct
*dir
,
760 const char *pathname
,
763 int pathlen
= strlen(pathname
);
765 struct exclude_list_group
*group
;
766 struct exclude
*exclude
;
767 const char *basename
= strrchr(pathname
, '/');
768 basename
= (basename
) ? basename
+1 : pathname
;
770 prep_exclude(dir
, pathname
, basename
-pathname
);
772 for (i
= EXC_CMDL
; i
<= EXC_FILE
; i
++) {
773 group
= &dir
->exclude_list_group
[i
];
774 for (j
= group
->nr
- 1; j
>= 0; j
--) {
775 exclude
= last_exclude_matching_from_list(
776 pathname
, pathlen
, basename
, dtype_p
,
786 * Loads the exclude lists for the directory containing pathname, then
787 * scans all exclude lists to determine whether pathname is excluded.
788 * Returns 1 if true, otherwise 0.
790 static int is_excluded(struct dir_struct
*dir
, const char *pathname
, int *dtype_p
)
792 struct exclude
*exclude
=
793 last_exclude_matching(dir
, pathname
, dtype_p
);
795 return exclude
->flags
& EXC_FLAG_NEGATIVE
? 0 : 1;
799 void path_exclude_check_init(struct path_exclude_check
*check
,
800 struct dir_struct
*dir
)
803 check
->exclude
= NULL
;
804 strbuf_init(&check
->path
, 256);
807 void path_exclude_check_clear(struct path_exclude_check
*check
)
809 strbuf_release(&check
->path
);
813 * For each subdirectory in name, starting with the top-most, checks
814 * to see if that subdirectory is excluded, and if so, returns the
815 * corresponding exclude structure. Otherwise, checks whether name
816 * itself (which is presumably a file) is excluded.
818 * A path to a directory known to be excluded is left in check->path to
819 * optimize for repeated checks for files in the same excluded directory.
821 struct exclude
*last_exclude_matching_path(struct path_exclude_check
*check
,
822 const char *name
, int namelen
,
826 struct strbuf
*path
= &check
->path
;
827 struct exclude
*exclude
;
830 * we allow the caller to pass namelen as an optimization; it
831 * must match the length of the name, as we eventually call
832 * is_excluded() on the whole name string.
835 namelen
= strlen(name
);
838 * If path is non-empty, and name is equal to path or a
839 * subdirectory of path, name should be excluded, because
840 * it's inside a directory which is already known to be
841 * excluded and was previously left in check->path.
844 path
->len
<= namelen
&&
845 !memcmp(name
, path
->buf
, path
->len
) &&
846 (!name
[path
->len
] || name
[path
->len
] == '/'))
847 return check
->exclude
;
849 strbuf_setlen(path
, 0);
850 for (i
= 0; name
[i
]; i
++) {
855 exclude
= last_exclude_matching(check
->dir
,
858 check
->exclude
= exclude
;
862 strbuf_addch(path
, ch
);
865 /* An entry in the index; cannot be a directory with subentries */
866 strbuf_setlen(path
, 0);
868 return last_exclude_matching(check
->dir
, name
, dtype
);
872 * Is this name excluded? This is for a caller like show_files() that
873 * do not honor directory hierarchy and iterate through paths that are
874 * possibly in an ignored directory.
876 int is_path_excluded(struct path_exclude_check
*check
,
877 const char *name
, int namelen
, int *dtype
)
879 struct exclude
*exclude
=
880 last_exclude_matching_path(check
, name
, namelen
, dtype
);
882 return exclude
->flags
& EXC_FLAG_NEGATIVE
? 0 : 1;
886 static struct dir_entry
*dir_entry_new(const char *pathname
, int len
)
888 struct dir_entry
*ent
;
890 ent
= xmalloc(sizeof(*ent
) + len
+ 1);
892 memcpy(ent
->name
, pathname
, len
);
897 static struct dir_entry
*dir_add_name(struct dir_struct
*dir
, const char *pathname
, int len
)
899 if (!(dir
->flags
& DIR_SHOW_IGNORED
) &&
900 cache_name_exists(pathname
, len
, ignore_case
))
903 ALLOC_GROW(dir
->entries
, dir
->nr
+1, dir
->alloc
);
904 return dir
->entries
[dir
->nr
++] = dir_entry_new(pathname
, len
);
907 struct dir_entry
*dir_add_ignored(struct dir_struct
*dir
, const char *pathname
, int len
)
909 if (!cache_name_is_other(pathname
, len
))
912 ALLOC_GROW(dir
->ignored
, dir
->ignored_nr
+1, dir
->ignored_alloc
);
913 return dir
->ignored
[dir
->ignored_nr
++] = dir_entry_new(pathname
, len
);
917 index_nonexistent
= 0,
923 * Do not use the alphabetically stored index to look up
924 * the directory name; instead, use the case insensitive
927 static enum exist_status
directory_exists_in_index_icase(const char *dirname
, int len
)
929 struct cache_entry
*ce
= index_name_exists(&the_index
, dirname
, len
+ 1, ignore_case
);
930 unsigned char endchar
;
933 return index_nonexistent
;
934 endchar
= ce
->name
[len
];
937 * The cache_entry structure returned will contain this dirname
938 * and possibly additional path components.
941 return index_directory
;
944 * If there are no additional path components, then this cache_entry
945 * represents a submodule. Submodules, despite being directories,
946 * are stored in the cache without a closing slash.
948 if (!endchar
&& S_ISGITLINK(ce
->ce_mode
))
951 /* This should never be hit, but it exists just in case. */
952 return index_nonexistent
;
956 * The index sorts alphabetically by entry name, which
957 * means that a gitlink sorts as '\0' at the end, while
958 * a directory (which is defined not as an entry, but as
959 * the files it contains) will sort with the '/' at the
962 static enum exist_status
directory_exists_in_index(const char *dirname
, int len
)
967 return directory_exists_in_index_icase(dirname
, len
);
969 pos
= cache_name_pos(dirname
, len
);
972 while (pos
< active_nr
) {
973 struct cache_entry
*ce
= active_cache
[pos
++];
974 unsigned char endchar
;
976 if (strncmp(ce
->name
, dirname
, len
))
978 endchar
= ce
->name
[len
];
982 return index_directory
;
983 if (!endchar
&& S_ISGITLINK(ce
->ce_mode
))
986 return index_nonexistent
;
990 * When we find a directory when traversing the filesystem, we
991 * have three distinct cases:
994 * - see it as a directory
997 * and which one we choose depends on a combination of existing
998 * git index contents and the flags passed into the directory
1001 * Case 1: If we *already* have entries in the index under that
1002 * directory name, we recurse into the directory to see all the files,
1003 * unless the directory is excluded and we want to show ignored
1006 * Case 2: If we *already* have that directory name as a gitlink,
1007 * we always continue to see it as a gitlink, regardless of whether
1008 * there is an actual git directory there or not (it might not
1009 * be checked out as a subproject!)
1011 * Case 3: if we didn't have it in the index previously, we
1012 * have a few sub-cases:
1014 * (a) if "show_other_directories" is true, we show it as
1015 * just a directory, unless "hide_empty_directories" is
1016 * also true and the directory is empty, in which case
1017 * we just ignore it entirely.
1018 * if we are looking for ignored directories, look if it
1019 * contains only ignored files to decide if it must be shown as
1021 * (b) if it looks like a git directory, and we don't have
1022 * 'no_gitlinks' set we treat it as a gitlink, and show it
1024 * (c) otherwise, we recurse into it.
1026 enum directory_treatment
{
1029 recurse_into_directory
1032 static enum directory_treatment
treat_directory(struct dir_struct
*dir
,
1033 const char *dirname
, int len
, int exclude
,
1034 const struct path_simplify
*simplify
)
1036 /* The "len-1" is to strip the final '/' */
1037 switch (directory_exists_in_index(dirname
, len
-1)) {
1038 case index_directory
:
1039 if ((dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
) && exclude
)
1042 return recurse_into_directory
;
1045 if (dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
)
1046 return ignore_directory
;
1047 return show_directory
;
1049 case index_nonexistent
:
1050 if (dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
)
1052 if (!(dir
->flags
& DIR_NO_GITLINKS
)) {
1053 unsigned char sha1
[20];
1054 if (resolve_gitlink_ref(dirname
, "HEAD", sha1
) == 0)
1055 return show_directory
;
1057 return recurse_into_directory
;
1060 /* This is the "show_other_directories" case */
1063 * We are looking for ignored files and our directory is not ignored,
1064 * check if it contains only ignored files
1066 if ((dir
->flags
& DIR_SHOW_IGNORED
) && !exclude
) {
1068 dir
->flags
&= ~DIR_SHOW_IGNORED
;
1069 dir
->flags
|= DIR_HIDE_EMPTY_DIRECTORIES
;
1070 ignored
= read_directory_recursive(dir
, dirname
, len
, 1, simplify
);
1071 dir
->flags
&= ~DIR_HIDE_EMPTY_DIRECTORIES
;
1072 dir
->flags
|= DIR_SHOW_IGNORED
;
1074 return ignored
? ignore_directory
: show_directory
;
1076 if (!(dir
->flags
& DIR_SHOW_IGNORED
) &&
1077 !(dir
->flags
& DIR_HIDE_EMPTY_DIRECTORIES
))
1078 return show_directory
;
1079 if (!read_directory_recursive(dir
, dirname
, len
, 1, simplify
))
1080 return ignore_directory
;
1081 return show_directory
;
1085 * Decide what to do when we find a file while traversing the
1086 * filesystem. Mostly two cases:
1088 * 1. We are looking for ignored files
1089 * (a) File is ignored, include it
1090 * (b) File is in ignored path, include it
1091 * (c) File is not ignored, exclude it
1093 * 2. Other scenarios, include the file if not excluded
1095 * Return 1 for exclude, 0 for include.
1097 static int treat_file(struct dir_struct
*dir
, struct strbuf
*path
, int exclude
, int *dtype
)
1099 struct path_exclude_check check
;
1100 int exclude_file
= 0;
1103 exclude_file
= !(dir
->flags
& DIR_SHOW_IGNORED
);
1104 else if (dir
->flags
& DIR_SHOW_IGNORED
) {
1105 /* Always exclude indexed files */
1106 struct cache_entry
*ce
= index_name_exists(&the_index
,
1107 path
->buf
, path
->len
, ignore_case
);
1112 path_exclude_check_init(&check
, dir
);
1114 if (!is_path_excluded(&check
, path
->buf
, path
->len
, dtype
))
1117 path_exclude_check_clear(&check
);
1120 return exclude_file
;
1124 * This is an inexact early pruning of any recursive directory
1125 * reading - if the path cannot possibly be in the pathspec,
1126 * return true, and we'll skip it early.
1128 static int simplify_away(const char *path
, int pathlen
, const struct path_simplify
*simplify
)
1132 const char *match
= simplify
->path
;
1133 int len
= simplify
->len
;
1139 if (!memcmp(path
, match
, len
))
1149 * This function tells us whether an excluded path matches a
1150 * list of "interesting" pathspecs. That is, whether a path matched
1151 * by any of the pathspecs could possibly be ignored by excluding
1152 * the specified path. This can happen if:
1154 * 1. the path is mentioned explicitly in the pathspec
1156 * 2. the path is a directory prefix of some element in the
1159 static int exclude_matches_pathspec(const char *path
, int len
,
1160 const struct path_simplify
*simplify
)
1163 for (; simplify
->path
; simplify
++) {
1164 if (len
== simplify
->len
1165 && !memcmp(path
, simplify
->path
, len
))
1167 if (len
< simplify
->len
1168 && simplify
->path
[len
] == '/'
1169 && !memcmp(path
, simplify
->path
, len
))
1176 static int get_index_dtype(const char *path
, int len
)
1179 struct cache_entry
*ce
;
1181 ce
= cache_name_exists(path
, len
, 0);
1183 if (!ce_uptodate(ce
))
1185 if (S_ISGITLINK(ce
->ce_mode
))
1188 * Nobody actually cares about the
1189 * difference between DT_LNK and DT_REG
1194 /* Try to look it up as a directory */
1195 pos
= cache_name_pos(path
, len
);
1199 while (pos
< active_nr
) {
1200 ce
= active_cache
[pos
++];
1201 if (strncmp(ce
->name
, path
, len
))
1203 if (ce
->name
[len
] > '/')
1205 if (ce
->name
[len
] < '/')
1207 if (!ce_uptodate(ce
))
1208 break; /* continue? */
1214 static int get_dtype(struct dirent
*de
, const char *path
, int len
)
1216 int dtype
= de
? DTYPE(de
) : DT_UNKNOWN
;
1219 if (dtype
!= DT_UNKNOWN
)
1221 dtype
= get_index_dtype(path
, len
);
1222 if (dtype
!= DT_UNKNOWN
)
1224 if (lstat(path
, &st
))
1226 if (S_ISREG(st
.st_mode
))
1228 if (S_ISDIR(st
.st_mode
))
1230 if (S_ISLNK(st
.st_mode
))
1235 enum path_treatment
{
1241 static enum path_treatment
treat_one_path(struct dir_struct
*dir
,
1242 struct strbuf
*path
,
1243 const struct path_simplify
*simplify
,
1244 int dtype
, struct dirent
*de
)
1246 int exclude
= is_excluded(dir
, path
->buf
, &dtype
);
1247 if (exclude
&& (dir
->flags
& DIR_COLLECT_IGNORED
)
1248 && exclude_matches_pathspec(path
->buf
, path
->len
, simplify
))
1249 dir_add_ignored(dir
, path
->buf
, path
->len
);
1252 * Excluded? If we don't explicitly want to show
1253 * ignored files, ignore it
1255 if (exclude
&& !(dir
->flags
& DIR_SHOW_IGNORED
))
1256 return path_ignored
;
1258 if (dtype
== DT_UNKNOWN
)
1259 dtype
= get_dtype(de
, path
->buf
, path
->len
);
1263 return path_ignored
;
1265 strbuf_addch(path
, '/');
1267 switch (treat_directory(dir
, path
->buf
, path
->len
, exclude
, simplify
)) {
1268 case show_directory
:
1270 case recurse_into_directory
:
1271 return path_recurse
;
1272 case ignore_directory
:
1273 return path_ignored
;
1278 switch (treat_file(dir
, path
, exclude
, &dtype
)) {
1280 return path_ignored
;
1285 return path_handled
;
1288 static enum path_treatment
treat_path(struct dir_struct
*dir
,
1290 struct strbuf
*path
,
1292 const struct path_simplify
*simplify
)
1296 if (is_dot_or_dotdot(de
->d_name
) || !strcmp(de
->d_name
, ".git"))
1297 return path_ignored
;
1298 strbuf_setlen(path
, baselen
);
1299 strbuf_addstr(path
, de
->d_name
);
1300 if (simplify_away(path
->buf
, path
->len
, simplify
))
1301 return path_ignored
;
1304 return treat_one_path(dir
, path
, simplify
, dtype
, de
);
1308 * Read a directory tree. We currently ignore anything but
1309 * directories, regular files and symlinks. That's because git
1310 * doesn't handle them at all yet. Maybe that will change some
1313 * Also, we ignore the name ".git" (even if it is not a directory).
1314 * That likely will not change.
1316 static int read_directory_recursive(struct dir_struct
*dir
,
1317 const char *base
, int baselen
,
1319 const struct path_simplify
*simplify
)
1324 struct strbuf path
= STRBUF_INIT
;
1326 strbuf_add(&path
, base
, baselen
);
1328 fdir
= opendir(path
.len
? path
.buf
: ".");
1332 while ((de
= readdir(fdir
)) != NULL
) {
1333 switch (treat_path(dir
, de
, &path
, baselen
, simplify
)) {
1335 contents
+= read_directory_recursive(dir
, path
.buf
,
1347 dir_add_name(dir
, path
.buf
, path
.len
);
1351 strbuf_release(&path
);
1356 static int cmp_name(const void *p1
, const void *p2
)
1358 const struct dir_entry
*e1
= *(const struct dir_entry
**)p1
;
1359 const struct dir_entry
*e2
= *(const struct dir_entry
**)p2
;
1361 return cache_name_compare(e1
->name
, e1
->len
,
1365 static struct path_simplify
*create_simplify(const char **pathspec
)
1368 struct path_simplify
*simplify
= NULL
;
1373 for (nr
= 0 ; ; nr
++) {
1376 alloc
= alloc_nr(alloc
);
1377 simplify
= xrealloc(simplify
, alloc
* sizeof(*simplify
));
1379 match
= *pathspec
++;
1382 simplify
[nr
].path
= match
;
1383 simplify
[nr
].len
= simple_length(match
);
1385 simplify
[nr
].path
= NULL
;
1386 simplify
[nr
].len
= 0;
1390 static void free_simplify(struct path_simplify
*simplify
)
1395 static int treat_leading_path(struct dir_struct
*dir
,
1396 const char *path
, int len
,
1397 const struct path_simplify
*simplify
)
1399 struct strbuf sb
= STRBUF_INIT
;
1400 int baselen
, rc
= 0;
1403 while (len
&& path
[len
- 1] == '/')
1409 cp
= path
+ baselen
+ !!baselen
;
1410 cp
= memchr(cp
, '/', path
+ len
- cp
);
1414 baselen
= cp
- path
;
1415 strbuf_setlen(&sb
, 0);
1416 strbuf_add(&sb
, path
, baselen
);
1417 if (!is_directory(sb
.buf
))
1419 if (simplify_away(sb
.buf
, sb
.len
, simplify
))
1421 if (treat_one_path(dir
, &sb
, simplify
,
1422 DT_DIR
, NULL
) == path_ignored
)
1423 break; /* do not recurse into it */
1424 if (len
<= baselen
) {
1426 break; /* finished checking */
1429 strbuf_release(&sb
);
1433 int read_directory(struct dir_struct
*dir
, const char *path
, int len
, const char **pathspec
)
1435 struct path_simplify
*simplify
;
1437 if (has_symlink_leading_path(path
, len
))
1440 simplify
= create_simplify(pathspec
);
1441 if (!len
|| treat_leading_path(dir
, path
, len
, simplify
))
1442 read_directory_recursive(dir
, path
, len
, 0, simplify
);
1443 free_simplify(simplify
);
1444 qsort(dir
->entries
, dir
->nr
, sizeof(struct dir_entry
*), cmp_name
);
1445 qsort(dir
->ignored
, dir
->ignored_nr
, sizeof(struct dir_entry
*), cmp_name
);
1449 int file_exists(const char *f
)
1452 return lstat(f
, &sb
) == 0;
1456 * Given two normalized paths (a trailing slash is ok), if subdir is
1457 * outside dir, return -1. Otherwise return the offset in subdir that
1458 * can be used as relative path to dir.
1460 int dir_inside_of(const char *subdir
, const char *dir
)
1464 assert(dir
&& subdir
&& *dir
&& *subdir
);
1466 while (*dir
&& *subdir
&& *dir
== *subdir
) {
1472 /* hel[p]/me vs hel[l]/yeah */
1473 if (*dir
&& *subdir
)
1477 return !*dir
? offset
: -1; /* same dir */
1479 /* foo/[b]ar vs foo/[] */
1480 if (is_dir_sep(dir
[-1]))
1481 return is_dir_sep(subdir
[-1]) ? offset
: -1;
1483 /* foo[/]bar vs foo[] */
1484 return is_dir_sep(*subdir
) ? offset
+ 1 : -1;
1487 int is_inside_dir(const char *dir
)
1492 if (!getcwd(cwd
, sizeof(cwd
)))
1493 die_errno("can't find the current directory");
1494 return dir_inside_of(cwd
, dir
) >= 0;
1497 int is_empty_dir(const char *path
)
1499 DIR *dir
= opendir(path
);
1506 while ((e
= readdir(dir
)) != NULL
)
1507 if (!is_dot_or_dotdot(e
->d_name
)) {
1516 static int remove_dir_recurse(struct strbuf
*path
, int flag
, int *kept_up
)
1520 int ret
= 0, original_len
= path
->len
, len
, kept_down
= 0;
1521 int only_empty
= (flag
& REMOVE_DIR_EMPTY_ONLY
);
1522 int keep_toplevel
= (flag
& REMOVE_DIR_KEEP_TOPLEVEL
);
1523 unsigned char submodule_head
[20];
1525 if ((flag
& REMOVE_DIR_KEEP_NESTED_GIT
) &&
1526 !resolve_gitlink_ref(path
->buf
, "HEAD", submodule_head
)) {
1527 /* Do not descend and nuke a nested git work tree. */
1533 flag
&= ~REMOVE_DIR_KEEP_TOPLEVEL
;
1534 dir
= opendir(path
->buf
);
1536 /* an empty dir could be removed even if it is unreadble */
1538 return rmdir(path
->buf
);
1542 if (path
->buf
[original_len
- 1] != '/')
1543 strbuf_addch(path
, '/');
1546 while ((e
= readdir(dir
)) != NULL
) {
1548 if (is_dot_or_dotdot(e
->d_name
))
1551 strbuf_setlen(path
, len
);
1552 strbuf_addstr(path
, e
->d_name
);
1553 if (lstat(path
->buf
, &st
))
1555 else if (S_ISDIR(st
.st_mode
)) {
1556 if (!remove_dir_recurse(path
, flag
, &kept_down
))
1557 continue; /* happy */
1558 } else if (!only_empty
&& !unlink(path
->buf
))
1559 continue; /* happy, too */
1561 /* path too long, stat fails, or non-directory still exists */
1567 strbuf_setlen(path
, original_len
);
1568 if (!ret
&& !keep_toplevel
&& !kept_down
)
1569 ret
= rmdir(path
->buf
);
1572 * report the uplevel that it is not an error that we
1573 * did not rmdir() our directory.
1579 int remove_dir_recursively(struct strbuf
*path
, int flag
)
1581 return remove_dir_recurse(path
, flag
, NULL
);
1584 void setup_standard_excludes(struct dir_struct
*dir
)
1589 dir
->exclude_per_dir
= ".gitignore";
1590 path
= git_path("info/exclude");
1591 if (!excludes_file
) {
1592 home_config_paths(NULL
, &xdg_path
, "ignore");
1593 excludes_file
= xdg_path
;
1595 if (!access_or_warn(path
, R_OK
))
1596 add_excludes_from_file(dir
, path
);
1597 if (excludes_file
&& !access_or_warn(excludes_file
, R_OK
))
1598 add_excludes_from_file(dir
, excludes_file
);
1601 int remove_path(const char *name
)
1605 if (unlink(name
) && errno
!= ENOENT
)
1608 slash
= strrchr(name
, '/');
1610 char *dirs
= xstrdup(name
);
1611 slash
= dirs
+ (slash
- name
);
1614 } while (rmdir(dirs
) == 0 && (slash
= strrchr(dirs
, '/')));
1620 static int pathspec_item_cmp(const void *a_
, const void *b_
)
1622 struct pathspec_item
*a
, *b
;
1624 a
= (struct pathspec_item
*)a_
;
1625 b
= (struct pathspec_item
*)b_
;
1626 return strcmp(a
->match
, b
->match
);
1629 int init_pathspec(struct pathspec
*pathspec
, const char **paths
)
1631 const char **p
= paths
;
1634 memset(pathspec
, 0, sizeof(*pathspec
));
1639 pathspec
->raw
= paths
;
1640 pathspec
->nr
= p
- paths
;
1644 pathspec
->items
= xmalloc(sizeof(struct pathspec_item
)*pathspec
->nr
);
1645 for (i
= 0; i
< pathspec
->nr
; i
++) {
1646 struct pathspec_item
*item
= pathspec
->items
+i
;
1647 const char *path
= paths
[i
];
1650 item
->len
= strlen(path
);
1652 if (limit_pathspec_to_literal()) {
1653 item
->nowildcard_len
= item
->len
;
1655 item
->nowildcard_len
= simple_length(path
);
1656 if (item
->nowildcard_len
< item
->len
) {
1657 pathspec
->has_wildcard
= 1;
1658 if (path
[item
->nowildcard_len
] == '*' &&
1659 no_wildcard(path
+ item
->nowildcard_len
+ 1))
1660 item
->flags
|= PATHSPEC_ONESTAR
;
1665 qsort(pathspec
->items
, pathspec
->nr
,
1666 sizeof(struct pathspec_item
), pathspec_item_cmp
);
1671 void free_pathspec(struct pathspec
*pathspec
)
1673 free(pathspec
->items
);
1674 pathspec
->items
= NULL
;
1677 int limit_pathspec_to_literal(void)
1679 static int flag
= -1;
1681 flag
= git_env_bool(GIT_LITERAL_PATHSPECS_ENVIRONMENT
, 0);
1686 * Frees memory within dir which was allocated for exclude lists and
1687 * the exclude_stack. Does not free dir itself.
1689 void clear_directory(struct dir_struct
*dir
)
1692 struct exclude_list_group
*group
;
1693 struct exclude_list
*el
;
1694 struct exclude_stack
*stk
;
1696 for (i
= EXC_CMDL
; i
<= EXC_FILE
; i
++) {
1697 group
= &dir
->exclude_list_group
[i
];
1698 for (j
= 0; j
< group
->nr
; j
++) {
1701 free((char *)el
->src
);
1702 clear_exclude_list(el
);
1707 stk
= dir
->exclude_stack
;
1709 struct exclude_stack
*prev
= stk
->prev
;