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
14 struct path_simplify
{
19 static int read_directory_recursive(struct dir_struct
*dir
, const char *path
, int len
,
20 int check_only
, const struct path_simplify
*simplify
);
21 static int get_dtype(struct dirent
*de
, const char *path
, int len
);
23 /* helper string functions with support for the ignore_case flag */
24 int strcmp_icase(const char *a
, const char *b
)
26 return ignore_case
? strcasecmp(a
, b
) : strcmp(a
, b
);
29 int strncmp_icase(const char *a
, const char *b
, size_t count
)
31 return ignore_case
? strncasecmp(a
, b
, count
) : strncmp(a
, b
, count
);
34 int fnmatch_icase(const char *pattern
, const char *string
, int flags
)
36 return fnmatch(pattern
, string
, flags
| (ignore_case
? FNM_CASEFOLD
: 0));
39 static size_t common_prefix_len(const char **pathspec
)
41 const char *n
, *first
;
48 while ((n
= *pathspec
++)) {
50 for (i
= 0; first
== n
|| i
< max
; i
++) {
52 if (!c
|| c
!= first
[i
] || is_glob_special(c
))
57 if (first
== n
|| len
< max
) {
67 * Returns a copy of the longest leading path common among all
70 char *common_prefix(const char **pathspec
)
72 unsigned long len
= common_prefix_len(pathspec
);
74 return len
? xmemdupz(*pathspec
, len
) : NULL
;
77 int fill_directory(struct dir_struct
*dir
, const char **pathspec
)
82 * Calculate common prefix for the pathspec, and
83 * use that to optimize the directory walk
85 len
= common_prefix_len(pathspec
);
87 /* Read the directory and prune it */
88 read_directory(dir
, pathspec
? *pathspec
: "", len
, pathspec
);
92 int within_depth(const char *name
, int namelen
,
93 int depth
, int max_depth
)
95 const char *cp
= name
, *cpe
= name
+ namelen
;
101 if (depth
> max_depth
)
108 * Does 'match' match the given name?
109 * A match is found if
111 * (1) the 'match' string is leading directory of 'name', or
112 * (2) the 'match' string is a wildcard and matches 'name', or
113 * (3) the 'match' string is exactly the same as 'name'.
115 * and the return value tells which case it was.
117 * It returns 0 when there is no match.
119 static int match_one(const char *match
, const char *name
, int namelen
)
123 /* If the match was just the prefix, we matched */
125 return MATCHED_RECURSIVELY
;
129 unsigned char c1
= tolower(*match
);
130 unsigned char c2
= tolower(*name
);
131 if (c1
== '\0' || is_glob_special(c1
))
141 unsigned char c1
= *match
;
142 unsigned char c2
= *name
;
143 if (c1
== '\0' || is_glob_special(c1
))
155 * If we don't match the matchstring exactly,
156 * we need to match by fnmatch
158 matchlen
= strlen(match
);
159 if (strncmp_icase(match
, name
, matchlen
))
160 return !fnmatch_icase(match
, name
, 0) ? MATCHED_FNMATCH
: 0;
162 if (namelen
== matchlen
)
163 return MATCHED_EXACTLY
;
164 if (match
[matchlen
-1] == '/' || name
[matchlen
] == '/')
165 return MATCHED_RECURSIVELY
;
170 * Given a name and a list of pathspecs, see if the name matches
171 * any of the pathspecs. The caller is also interested in seeing
172 * all pathspec matches some names it calls this function with
173 * (otherwise the user could have mistyped the unmatched pathspec),
174 * and a mark is left in seen[] array for pathspec element that
175 * actually matched anything.
177 int match_pathspec(const char **pathspec
, const char *name
, int namelen
,
178 int prefix
, char *seen
)
188 for (i
= 0; pathspec
[i
] != NULL
; i
++) {
190 const char *match
= pathspec
[i
] + prefix
;
191 if (seen
&& seen
[i
] == MATCHED_EXACTLY
)
193 how
= match_one(match
, name
, namelen
);
197 if (seen
&& seen
[i
] < how
)
205 * Does 'match' match the given name?
206 * A match is found if
208 * (1) the 'match' string is leading directory of 'name', or
209 * (2) the 'match' string is a wildcard and matches 'name', or
210 * (3) the 'match' string is exactly the same as 'name'.
212 * and the return value tells which case it was.
214 * It returns 0 when there is no match.
216 static int match_pathspec_item(const struct pathspec_item
*item
, int prefix
,
217 const char *name
, int namelen
)
219 /* name/namelen has prefix cut off by caller */
220 const char *match
= item
->match
+ prefix
;
221 int matchlen
= item
->len
- prefix
;
223 /* If the match was just the prefix, we matched */
225 return MATCHED_RECURSIVELY
;
227 if (matchlen
<= namelen
&& !strncmp(match
, name
, matchlen
)) {
228 if (matchlen
== namelen
)
229 return MATCHED_EXACTLY
;
231 if (match
[matchlen
-1] == '/' || name
[matchlen
] == '/')
232 return MATCHED_RECURSIVELY
;
235 if (item
->use_wildcard
&& !fnmatch(match
, name
, 0))
236 return MATCHED_FNMATCH
;
242 * Given a name and a list of pathspecs, see if the name matches
243 * any of the pathspecs. The caller is also interested in seeing
244 * all pathspec matches some names it calls this function with
245 * (otherwise the user could have mistyped the unmatched pathspec),
246 * and a mark is left in seen[] array for pathspec element that
247 * actually matched anything.
249 int match_pathspec_depth(const struct pathspec
*ps
,
250 const char *name
, int namelen
,
251 int prefix
, char *seen
)
256 if (!ps
->recursive
|| ps
->max_depth
== -1)
257 return MATCHED_RECURSIVELY
;
259 if (within_depth(name
, namelen
, 0, ps
->max_depth
))
260 return MATCHED_EXACTLY
;
268 for (i
= ps
->nr
- 1; i
>= 0; i
--) {
270 if (seen
&& seen
[i
] == MATCHED_EXACTLY
)
272 how
= match_pathspec_item(ps
->items
+i
, prefix
, name
, namelen
);
273 if (ps
->recursive
&& ps
->max_depth
!= -1 &&
274 how
&& how
!= MATCHED_FNMATCH
) {
275 int len
= ps
->items
[i
].len
;
276 if (name
[len
] == '/')
278 if (within_depth(name
+len
, namelen
-len
, 0, ps
->max_depth
))
279 how
= MATCHED_EXACTLY
;
286 if (seen
&& seen
[i
] < how
)
294 * Return the length of the "simple" part of a path match limiter.
296 static int simple_length(const char *match
)
301 unsigned char c
= *match
++;
303 if (c
== '\0' || is_glob_special(c
))
308 static int no_wildcard(const char *string
)
310 return string
[simple_length(string
)] == '\0';
313 void parse_exclude_pattern(const char **pattern
,
318 const char *p
= *pattern
;
323 *flags
|= EXC_FLAG_NEGATIVE
;
327 if (len
&& p
[len
- 1] == '/') {
329 *flags
|= EXC_FLAG_MUSTBEDIR
;
331 for (i
= 0; i
< len
; i
++) {
336 *flags
|= EXC_FLAG_NODIR
;
337 *nowildcardlen
= simple_length(p
);
339 * we should have excluded the trailing slash from 'p' too,
340 * but that's one more allocation. Instead just make sure
341 * nowildcardlen does not exceed real patternlen
343 if (*nowildcardlen
> len
)
344 *nowildcardlen
= len
;
345 if (*p
== '*' && no_wildcard(p
+ 1))
346 *flags
|= EXC_FLAG_ENDSWITH
;
351 void add_exclude(const char *string
, const char *base
,
352 int baselen
, struct exclude_list
*el
, int srcpos
)
359 parse_exclude_pattern(&string
, &patternlen
, &flags
, &nowildcardlen
);
360 if (flags
& EXC_FLAG_MUSTBEDIR
) {
362 x
= xmalloc(sizeof(*x
) + patternlen
+ 1);
364 memcpy(s
, string
, patternlen
);
365 s
[patternlen
] = '\0';
368 x
= xmalloc(sizeof(*x
));
371 x
->patternlen
= patternlen
;
372 x
->nowildcardlen
= nowildcardlen
;
374 x
->baselen
= baselen
;
377 ALLOC_GROW(el
->excludes
, el
->nr
+ 1, el
->alloc
);
378 el
->excludes
[el
->nr
++] = x
;
382 static void *read_skip_worktree_file_from_index(const char *path
, size_t *size
)
386 enum object_type type
;
388 struct index_state
*istate
= &the_index
;
391 pos
= index_name_pos(istate
, path
, len
);
394 if (!ce_skip_worktree(istate
->cache
[pos
]))
396 data
= read_sha1_file(istate
->cache
[pos
]->sha1
, &type
, &sz
);
397 if (!data
|| type
!= OBJ_BLOB
) {
406 * Frees memory within el which was allocated for exclude patterns and
407 * the file buffer. Does not free el itself.
409 void clear_exclude_list(struct exclude_list
*el
)
413 for (i
= 0; i
< el
->nr
; i
++)
414 free(el
->excludes
[i
]);
423 int add_excludes_from_file_to_list(const char *fname
,
426 struct exclude_list
*el
,
430 int fd
, i
, lineno
= 1;
434 fd
= open(fname
, O_RDONLY
);
435 if (fd
< 0 || fstat(fd
, &st
) < 0) {
439 (buf
= read_skip_worktree_file_from_index(fname
, &size
)) == NULL
)
445 if (buf
[size
-1] != '\n') {
446 buf
= xrealloc(buf
, size
+1);
451 size
= xsize_t(st
.st_size
);
456 buf
= xmalloc(size
+1);
457 if (read_in_full(fd
, buf
, size
) != size
) {
468 for (i
= 0; i
< size
; i
++) {
469 if (buf
[i
] == '\n') {
470 if (entry
!= buf
+ i
&& entry
[0] != '#') {
471 buf
[i
- (i
&& buf
[i
-1] == '\r')] = 0;
472 add_exclude(entry
, base
, baselen
, el
, lineno
);
481 struct exclude_list
*add_exclude_list(struct dir_struct
*dir
,
482 int group_type
, const char *src
)
484 struct exclude_list
*el
;
485 struct exclude_list_group
*group
;
487 group
= &dir
->exclude_list_group
[group_type
];
488 ALLOC_GROW(group
->el
, group
->nr
+ 1, group
->alloc
);
489 el
= &group
->el
[group
->nr
++];
490 memset(el
, 0, sizeof(*el
));
496 * Used to set up core.excludesfile and .git/info/exclude lists.
498 void add_excludes_from_file(struct dir_struct
*dir
, const char *fname
)
500 struct exclude_list
*el
;
501 el
= add_exclude_list(dir
, EXC_FILE
, fname
);
502 if (add_excludes_from_file_to_list(fname
, "", 0, el
, 0) < 0)
503 die("cannot use %s as an exclude file", fname
);
507 * Loads the per-directory exclude list for the substring of base
508 * which has a char length of baselen.
510 static void prep_exclude(struct dir_struct
*dir
, const char *base
, int baselen
)
512 struct exclude_list_group
*group
;
513 struct exclude_list
*el
;
514 struct exclude_stack
*stk
= NULL
;
517 if ((!dir
->exclude_per_dir
) ||
518 (baselen
+ strlen(dir
->exclude_per_dir
) >= PATH_MAX
))
519 return; /* too long a path -- ignore */
521 group
= &dir
->exclude_list_group
[EXC_DIRS
];
523 /* Pop the exclude lists from the EXCL_DIRS exclude_list_group
524 * which originate from directories not in the prefix of the
525 * path being checked. */
526 while ((stk
= dir
->exclude_stack
) != NULL
) {
527 if (stk
->baselen
<= baselen
&&
528 !strncmp(dir
->basebuf
, base
, stk
->baselen
))
530 el
= &group
->el
[dir
->exclude_stack
->exclude_ix
];
531 dir
->exclude_stack
= stk
->prev
;
532 free((char *)el
->src
); /* see strdup() below */
533 clear_exclude_list(el
);
538 /* Read from the parent directories and push them down. */
539 current
= stk
? stk
->baselen
: -1;
540 while (current
< baselen
) {
541 struct exclude_stack
*stk
= xcalloc(1, sizeof(*stk
));
549 cp
= strchr(base
+ current
+ 1, '/');
551 die("oops in prep_exclude");
554 stk
->prev
= dir
->exclude_stack
;
555 stk
->baselen
= cp
- base
;
556 memcpy(dir
->basebuf
+ current
, base
+ current
,
557 stk
->baselen
- current
);
558 strcpy(dir
->basebuf
+ stk
->baselen
, dir
->exclude_per_dir
);
560 * dir->basebuf gets reused by the traversal, but we
561 * need fname to remain unchanged to ensure the src
562 * member of each struct exclude correctly
563 * back-references its source file. Other invocations
564 * of add_exclude_list provide stable strings, so we
565 * strdup() and free() here in the caller.
567 el
= add_exclude_list(dir
, EXC_DIRS
, strdup(dir
->basebuf
));
568 stk
->exclude_ix
= group
->nr
- 1;
569 add_excludes_from_file_to_list(dir
->basebuf
,
570 dir
->basebuf
, stk
->baselen
,
572 dir
->exclude_stack
= stk
;
573 current
= stk
->baselen
;
575 dir
->basebuf
[baselen
] = '\0';
578 int match_basename(const char *basename
, int basenamelen
,
579 const char *pattern
, int prefix
, int patternlen
,
582 if (prefix
== patternlen
) {
583 if (!strcmp_icase(pattern
, basename
))
585 } else if (flags
& EXC_FLAG_ENDSWITH
) {
586 if (patternlen
- 1 <= basenamelen
&&
587 !strcmp_icase(pattern
+ 1,
588 basename
+ basenamelen
- patternlen
+ 1))
591 if (fnmatch_icase(pattern
, basename
, 0) == 0)
597 int match_pathname(const char *pathname
, int pathlen
,
598 const char *base
, int baselen
,
599 const char *pattern
, int prefix
, int patternlen
,
606 * match with FNM_PATHNAME; the pattern has base implicitly
609 if (*pattern
== '/') {
615 * baselen does not count the trailing slash. base[] may or
616 * may not end with a trailing slash though.
618 if (pathlen
< baselen
+ 1 ||
619 (baselen
&& pathname
[baselen
] != '/') ||
620 strncmp_icase(pathname
, base
, baselen
))
623 namelen
= baselen
? pathlen
- baselen
- 1 : pathlen
;
624 name
= pathname
+ pathlen
- namelen
;
628 * if the non-wildcard part is longer than the
629 * remaining pathname, surely it cannot match.
631 if (prefix
> namelen
)
634 if (strncmp_icase(pattern
, name
, prefix
))
641 return fnmatch_icase(pattern
, name
, FNM_PATHNAME
) == 0;
645 * Scan the given exclude list in reverse to see whether pathname
646 * should be ignored. The first match (i.e. the last on the list), if
647 * any, determines the fate. Returns the exclude_list element which
648 * matched, or NULL for undecided.
650 static struct exclude
*last_exclude_matching_from_list(const char *pathname
,
652 const char *basename
,
654 struct exclude_list
*el
)
659 return NULL
; /* undefined */
661 for (i
= el
->nr
- 1; 0 <= i
; i
--) {
662 struct exclude
*x
= el
->excludes
[i
];
663 const char *exclude
= x
->pattern
;
664 int prefix
= x
->nowildcardlen
;
666 if (x
->flags
& EXC_FLAG_MUSTBEDIR
) {
667 if (*dtype
== DT_UNKNOWN
)
668 *dtype
= get_dtype(NULL
, pathname
, pathlen
);
669 if (*dtype
!= DT_DIR
)
673 if (x
->flags
& EXC_FLAG_NODIR
) {
674 if (match_basename(basename
,
675 pathlen
- (basename
- pathname
),
676 exclude
, prefix
, x
->patternlen
,
682 assert(x
->baselen
== 0 || x
->base
[x
->baselen
- 1] == '/');
683 if (match_pathname(pathname
, pathlen
,
684 x
->base
, x
->baselen
? x
->baselen
- 1 : 0,
685 exclude
, prefix
, x
->patternlen
, x
->flags
))
688 return NULL
; /* undecided */
692 * Scan the list and let the last match determine the fate.
693 * Return 1 for exclude, 0 for include and -1 for undecided.
695 int is_excluded_from_list(const char *pathname
,
696 int pathlen
, const char *basename
, int *dtype
,
697 struct exclude_list
*el
)
699 struct exclude
*exclude
;
700 exclude
= last_exclude_matching_from_list(pathname
, pathlen
, basename
, dtype
, el
);
702 return exclude
->flags
& EXC_FLAG_NEGATIVE
? 0 : 1;
703 return -1; /* undecided */
707 * Loads the exclude lists for the directory containing pathname, then
708 * scans all exclude lists to determine whether pathname is excluded.
709 * Returns the exclude_list element which matched, or NULL for
712 static struct exclude
*last_exclude_matching(struct dir_struct
*dir
,
713 const char *pathname
,
716 int pathlen
= strlen(pathname
);
718 struct exclude_list_group
*group
;
719 struct exclude
*exclude
;
720 const char *basename
= strrchr(pathname
, '/');
721 basename
= (basename
) ? basename
+1 : pathname
;
723 prep_exclude(dir
, pathname
, basename
-pathname
);
725 for (i
= EXC_CMDL
; i
<= EXC_FILE
; i
++) {
726 group
= &dir
->exclude_list_group
[i
];
727 for (j
= group
->nr
- 1; j
>= 0; j
--) {
728 exclude
= last_exclude_matching_from_list(
729 pathname
, pathlen
, basename
, dtype_p
,
739 * Loads the exclude lists for the directory containing pathname, then
740 * scans all exclude lists to determine whether pathname is excluded.
741 * Returns 1 if true, otherwise 0.
743 static int is_excluded(struct dir_struct
*dir
, const char *pathname
, int *dtype_p
)
745 struct exclude
*exclude
=
746 last_exclude_matching(dir
, pathname
, dtype_p
);
748 return exclude
->flags
& EXC_FLAG_NEGATIVE
? 0 : 1;
752 void path_exclude_check_init(struct path_exclude_check
*check
,
753 struct dir_struct
*dir
)
756 check
->exclude
= NULL
;
757 strbuf_init(&check
->path
, 256);
760 void path_exclude_check_clear(struct path_exclude_check
*check
)
762 strbuf_release(&check
->path
);
766 * For each subdirectory in name, starting with the top-most, checks
767 * to see if that subdirectory is excluded, and if so, returns the
768 * corresponding exclude structure. Otherwise, checks whether name
769 * itself (which is presumably a file) is excluded.
771 * A path to a directory known to be excluded is left in check->path to
772 * optimize for repeated checks for files in the same excluded directory.
774 struct exclude
*last_exclude_matching_path(struct path_exclude_check
*check
,
775 const char *name
, int namelen
,
779 struct strbuf
*path
= &check
->path
;
780 struct exclude
*exclude
;
783 * we allow the caller to pass namelen as an optimization; it
784 * must match the length of the name, as we eventually call
785 * is_excluded() on the whole name string.
788 namelen
= strlen(name
);
791 * If path is non-empty, and name is equal to path or a
792 * subdirectory of path, name should be excluded, because
793 * it's inside a directory which is already known to be
794 * excluded and was previously left in check->path.
797 path
->len
<= namelen
&&
798 !memcmp(name
, path
->buf
, path
->len
) &&
799 (!name
[path
->len
] || name
[path
->len
] == '/'))
800 return check
->exclude
;
802 strbuf_setlen(path
, 0);
803 for (i
= 0; name
[i
]; i
++) {
808 exclude
= last_exclude_matching(check
->dir
,
811 check
->exclude
= exclude
;
815 strbuf_addch(path
, ch
);
818 /* An entry in the index; cannot be a directory with subentries */
819 strbuf_setlen(path
, 0);
821 return last_exclude_matching(check
->dir
, name
, dtype
);
825 * Is this name excluded? This is for a caller like show_files() that
826 * do not honor directory hierarchy and iterate through paths that are
827 * possibly in an ignored directory.
829 int is_path_excluded(struct path_exclude_check
*check
,
830 const char *name
, int namelen
, int *dtype
)
832 struct exclude
*exclude
=
833 last_exclude_matching_path(check
, name
, namelen
, dtype
);
835 return exclude
->flags
& EXC_FLAG_NEGATIVE
? 0 : 1;
839 static struct dir_entry
*dir_entry_new(const char *pathname
, int len
)
841 struct dir_entry
*ent
;
843 ent
= xmalloc(sizeof(*ent
) + len
+ 1);
845 memcpy(ent
->name
, pathname
, len
);
850 static struct dir_entry
*dir_add_name(struct dir_struct
*dir
, const char *pathname
, int len
)
852 if (cache_name_exists(pathname
, len
, ignore_case
))
855 ALLOC_GROW(dir
->entries
, dir
->nr
+1, dir
->alloc
);
856 return dir
->entries
[dir
->nr
++] = dir_entry_new(pathname
, len
);
859 struct dir_entry
*dir_add_ignored(struct dir_struct
*dir
, const char *pathname
, int len
)
861 if (!cache_name_is_other(pathname
, len
))
864 ALLOC_GROW(dir
->ignored
, dir
->ignored_nr
+1, dir
->ignored_alloc
);
865 return dir
->ignored
[dir
->ignored_nr
++] = dir_entry_new(pathname
, len
);
869 index_nonexistent
= 0,
875 * Do not use the alphabetically stored index to look up
876 * the directory name; instead, use the case insensitive
879 static enum exist_status
directory_exists_in_index_icase(const char *dirname
, int len
)
881 struct cache_entry
*ce
= index_name_exists(&the_index
, dirname
, len
+ 1, ignore_case
);
882 unsigned char endchar
;
885 return index_nonexistent
;
886 endchar
= ce
->name
[len
];
889 * The cache_entry structure returned will contain this dirname
890 * and possibly additional path components.
893 return index_directory
;
896 * If there are no additional path components, then this cache_entry
897 * represents a submodule. Submodules, despite being directories,
898 * are stored in the cache without a closing slash.
900 if (!endchar
&& S_ISGITLINK(ce
->ce_mode
))
903 /* This should never be hit, but it exists just in case. */
904 return index_nonexistent
;
908 * The index sorts alphabetically by entry name, which
909 * means that a gitlink sorts as '\0' at the end, while
910 * a directory (which is defined not as an entry, but as
911 * the files it contains) will sort with the '/' at the
914 static enum exist_status
directory_exists_in_index(const char *dirname
, int len
)
919 return directory_exists_in_index_icase(dirname
, len
);
921 pos
= cache_name_pos(dirname
, len
);
924 while (pos
< active_nr
) {
925 struct cache_entry
*ce
= active_cache
[pos
++];
926 unsigned char endchar
;
928 if (strncmp(ce
->name
, dirname
, len
))
930 endchar
= ce
->name
[len
];
934 return index_directory
;
935 if (!endchar
&& S_ISGITLINK(ce
->ce_mode
))
938 return index_nonexistent
;
942 * When we find a directory when traversing the filesystem, we
943 * have three distinct cases:
946 * - see it as a directory
949 * and which one we choose depends on a combination of existing
950 * git index contents and the flags passed into the directory
953 * Case 1: If we *already* have entries in the index under that
954 * directory name, we always recurse into the directory to see
957 * Case 2: If we *already* have that directory name as a gitlink,
958 * we always continue to see it as a gitlink, regardless of whether
959 * there is an actual git directory there or not (it might not
960 * be checked out as a subproject!)
962 * Case 3: if we didn't have it in the index previously, we
963 * have a few sub-cases:
965 * (a) if "show_other_directories" is true, we show it as
966 * just a directory, unless "hide_empty_directories" is
967 * also true and the directory is empty, in which case
968 * we just ignore it entirely.
969 * (b) if it looks like a git directory, and we don't have
970 * 'no_gitlinks' set we treat it as a gitlink, and show it
972 * (c) otherwise, we recurse into it.
974 enum directory_treatment
{
977 recurse_into_directory
980 static enum directory_treatment
treat_directory(struct dir_struct
*dir
,
981 const char *dirname
, int len
,
982 const struct path_simplify
*simplify
)
984 /* The "len-1" is to strip the final '/' */
985 switch (directory_exists_in_index(dirname
, len
-1)) {
986 case index_directory
:
987 return recurse_into_directory
;
990 if (dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
)
991 return ignore_directory
;
992 return show_directory
;
994 case index_nonexistent
:
995 if (dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
)
997 if (!(dir
->flags
& DIR_NO_GITLINKS
)) {
998 unsigned char sha1
[20];
999 if (resolve_gitlink_ref(dirname
, "HEAD", sha1
) == 0)
1000 return show_directory
;
1002 return recurse_into_directory
;
1005 /* This is the "show_other_directories" case */
1006 if (!(dir
->flags
& DIR_HIDE_EMPTY_DIRECTORIES
))
1007 return show_directory
;
1008 if (!read_directory_recursive(dir
, dirname
, len
, 1, simplify
))
1009 return ignore_directory
;
1010 return show_directory
;
1014 * This is an inexact early pruning of any recursive directory
1015 * reading - if the path cannot possibly be in the pathspec,
1016 * return true, and we'll skip it early.
1018 static int simplify_away(const char *path
, int pathlen
, const struct path_simplify
*simplify
)
1022 const char *match
= simplify
->path
;
1023 int len
= simplify
->len
;
1029 if (!memcmp(path
, match
, len
))
1039 * This function tells us whether an excluded path matches a
1040 * list of "interesting" pathspecs. That is, whether a path matched
1041 * by any of the pathspecs could possibly be ignored by excluding
1042 * the specified path. This can happen if:
1044 * 1. the path is mentioned explicitly in the pathspec
1046 * 2. the path is a directory prefix of some element in the
1049 static int exclude_matches_pathspec(const char *path
, int len
,
1050 const struct path_simplify
*simplify
)
1053 for (; simplify
->path
; simplify
++) {
1054 if (len
== simplify
->len
1055 && !memcmp(path
, simplify
->path
, len
))
1057 if (len
< simplify
->len
1058 && simplify
->path
[len
] == '/'
1059 && !memcmp(path
, simplify
->path
, len
))
1066 static int get_index_dtype(const char *path
, int len
)
1069 struct cache_entry
*ce
;
1071 ce
= cache_name_exists(path
, len
, 0);
1073 if (!ce_uptodate(ce
))
1075 if (S_ISGITLINK(ce
->ce_mode
))
1078 * Nobody actually cares about the
1079 * difference between DT_LNK and DT_REG
1084 /* Try to look it up as a directory */
1085 pos
= cache_name_pos(path
, len
);
1089 while (pos
< active_nr
) {
1090 ce
= active_cache
[pos
++];
1091 if (strncmp(ce
->name
, path
, len
))
1093 if (ce
->name
[len
] > '/')
1095 if (ce
->name
[len
] < '/')
1097 if (!ce_uptodate(ce
))
1098 break; /* continue? */
1104 static int get_dtype(struct dirent
*de
, const char *path
, int len
)
1106 int dtype
= de
? DTYPE(de
) : DT_UNKNOWN
;
1109 if (dtype
!= DT_UNKNOWN
)
1111 dtype
= get_index_dtype(path
, len
);
1112 if (dtype
!= DT_UNKNOWN
)
1114 if (lstat(path
, &st
))
1116 if (S_ISREG(st
.st_mode
))
1118 if (S_ISDIR(st
.st_mode
))
1120 if (S_ISLNK(st
.st_mode
))
1125 enum path_treatment
{
1131 static enum path_treatment
treat_one_path(struct dir_struct
*dir
,
1132 struct strbuf
*path
,
1133 const struct path_simplify
*simplify
,
1134 int dtype
, struct dirent
*de
)
1136 int exclude
= is_excluded(dir
, path
->buf
, &dtype
);
1137 if (exclude
&& (dir
->flags
& DIR_COLLECT_IGNORED
)
1138 && exclude_matches_pathspec(path
->buf
, path
->len
, simplify
))
1139 dir_add_ignored(dir
, path
->buf
, path
->len
);
1142 * Excluded? If we don't explicitly want to show
1143 * ignored files, ignore it
1145 if (exclude
&& !(dir
->flags
& DIR_SHOW_IGNORED
))
1146 return path_ignored
;
1148 if (dtype
== DT_UNKNOWN
)
1149 dtype
= get_dtype(de
, path
->buf
, path
->len
);
1152 * Do we want to see just the ignored files?
1153 * We still need to recurse into directories,
1154 * even if we don't ignore them, since the
1155 * directory may contain files that we do..
1157 if (!exclude
&& (dir
->flags
& DIR_SHOW_IGNORED
)) {
1158 if (dtype
!= DT_DIR
)
1159 return path_ignored
;
1164 return path_ignored
;
1166 strbuf_addch(path
, '/');
1167 switch (treat_directory(dir
, path
->buf
, path
->len
, simplify
)) {
1168 case show_directory
:
1169 if (exclude
!= !!(dir
->flags
1170 & DIR_SHOW_IGNORED
))
1171 return path_ignored
;
1173 case recurse_into_directory
:
1174 return path_recurse
;
1175 case ignore_directory
:
1176 return path_ignored
;
1183 return path_handled
;
1186 static enum path_treatment
treat_path(struct dir_struct
*dir
,
1188 struct strbuf
*path
,
1190 const struct path_simplify
*simplify
)
1194 if (is_dot_or_dotdot(de
->d_name
) || !strcmp(de
->d_name
, ".git"))
1195 return path_ignored
;
1196 strbuf_setlen(path
, baselen
);
1197 strbuf_addstr(path
, de
->d_name
);
1198 if (simplify_away(path
->buf
, path
->len
, simplify
))
1199 return path_ignored
;
1202 return treat_one_path(dir
, path
, simplify
, dtype
, de
);
1206 * Read a directory tree. We currently ignore anything but
1207 * directories, regular files and symlinks. That's because git
1208 * doesn't handle them at all yet. Maybe that will change some
1211 * Also, we ignore the name ".git" (even if it is not a directory).
1212 * That likely will not change.
1214 static int read_directory_recursive(struct dir_struct
*dir
,
1215 const char *base
, int baselen
,
1217 const struct path_simplify
*simplify
)
1222 struct strbuf path
= STRBUF_INIT
;
1224 strbuf_add(&path
, base
, baselen
);
1226 fdir
= opendir(path
.len
? path
.buf
: ".");
1230 while ((de
= readdir(fdir
)) != NULL
) {
1231 switch (treat_path(dir
, de
, &path
, baselen
, simplify
)) {
1233 contents
+= read_directory_recursive(dir
, path
.buf
,
1245 dir_add_name(dir
, path
.buf
, path
.len
);
1249 strbuf_release(&path
);
1254 static int cmp_name(const void *p1
, const void *p2
)
1256 const struct dir_entry
*e1
= *(const struct dir_entry
**)p1
;
1257 const struct dir_entry
*e2
= *(const struct dir_entry
**)p2
;
1259 return cache_name_compare(e1
->name
, e1
->len
,
1263 static struct path_simplify
*create_simplify(const char **pathspec
)
1266 struct path_simplify
*simplify
= NULL
;
1271 for (nr
= 0 ; ; nr
++) {
1274 alloc
= alloc_nr(alloc
);
1275 simplify
= xrealloc(simplify
, alloc
* sizeof(*simplify
));
1277 match
= *pathspec
++;
1280 simplify
[nr
].path
= match
;
1281 simplify
[nr
].len
= simple_length(match
);
1283 simplify
[nr
].path
= NULL
;
1284 simplify
[nr
].len
= 0;
1288 static void free_simplify(struct path_simplify
*simplify
)
1293 static int treat_leading_path(struct dir_struct
*dir
,
1294 const char *path
, int len
,
1295 const struct path_simplify
*simplify
)
1297 struct strbuf sb
= STRBUF_INIT
;
1298 int baselen
, rc
= 0;
1301 while (len
&& path
[len
- 1] == '/')
1307 cp
= path
+ baselen
+ !!baselen
;
1308 cp
= memchr(cp
, '/', path
+ len
- cp
);
1312 baselen
= cp
- path
;
1313 strbuf_setlen(&sb
, 0);
1314 strbuf_add(&sb
, path
, baselen
);
1315 if (!is_directory(sb
.buf
))
1317 if (simplify_away(sb
.buf
, sb
.len
, simplify
))
1319 if (treat_one_path(dir
, &sb
, simplify
,
1320 DT_DIR
, NULL
) == path_ignored
)
1321 break; /* do not recurse into it */
1322 if (len
<= baselen
) {
1324 break; /* finished checking */
1327 strbuf_release(&sb
);
1331 int read_directory(struct dir_struct
*dir
, const char *path
, int len
, const char **pathspec
)
1333 struct path_simplify
*simplify
;
1335 if (has_symlink_leading_path(path
, len
))
1338 simplify
= create_simplify(pathspec
);
1339 if (!len
|| treat_leading_path(dir
, path
, len
, simplify
))
1340 read_directory_recursive(dir
, path
, len
, 0, simplify
);
1341 free_simplify(simplify
);
1342 qsort(dir
->entries
, dir
->nr
, sizeof(struct dir_entry
*), cmp_name
);
1343 qsort(dir
->ignored
, dir
->ignored_nr
, sizeof(struct dir_entry
*), cmp_name
);
1347 int file_exists(const char *f
)
1350 return lstat(f
, &sb
) == 0;
1354 * Given two normalized paths (a trailing slash is ok), if subdir is
1355 * outside dir, return -1. Otherwise return the offset in subdir that
1356 * can be used as relative path to dir.
1358 int dir_inside_of(const char *subdir
, const char *dir
)
1362 assert(dir
&& subdir
&& *dir
&& *subdir
);
1364 while (*dir
&& *subdir
&& *dir
== *subdir
) {
1370 /* hel[p]/me vs hel[l]/yeah */
1371 if (*dir
&& *subdir
)
1375 return !*dir
? offset
: -1; /* same dir */
1377 /* foo/[b]ar vs foo/[] */
1378 if (is_dir_sep(dir
[-1]))
1379 return is_dir_sep(subdir
[-1]) ? offset
: -1;
1381 /* foo[/]bar vs foo[] */
1382 return is_dir_sep(*subdir
) ? offset
+ 1 : -1;
1385 int is_inside_dir(const char *dir
)
1390 if (!getcwd(cwd
, sizeof(cwd
)))
1391 die_errno("can't find the current directory");
1392 return dir_inside_of(cwd
, dir
) >= 0;
1395 int is_empty_dir(const char *path
)
1397 DIR *dir
= opendir(path
);
1404 while ((e
= readdir(dir
)) != NULL
)
1405 if (!is_dot_or_dotdot(e
->d_name
)) {
1414 static int remove_dir_recurse(struct strbuf
*path
, int flag
, int *kept_up
)
1418 int ret
= 0, original_len
= path
->len
, len
, kept_down
= 0;
1419 int only_empty
= (flag
& REMOVE_DIR_EMPTY_ONLY
);
1420 int keep_toplevel
= (flag
& REMOVE_DIR_KEEP_TOPLEVEL
);
1421 unsigned char submodule_head
[20];
1423 if ((flag
& REMOVE_DIR_KEEP_NESTED_GIT
) &&
1424 !resolve_gitlink_ref(path
->buf
, "HEAD", submodule_head
)) {
1425 /* Do not descend and nuke a nested git work tree. */
1431 flag
&= ~REMOVE_DIR_KEEP_TOPLEVEL
;
1432 dir
= opendir(path
->buf
);
1434 /* an empty dir could be removed even if it is unreadble */
1436 return rmdir(path
->buf
);
1440 if (path
->buf
[original_len
- 1] != '/')
1441 strbuf_addch(path
, '/');
1444 while ((e
= readdir(dir
)) != NULL
) {
1446 if (is_dot_or_dotdot(e
->d_name
))
1449 strbuf_setlen(path
, len
);
1450 strbuf_addstr(path
, e
->d_name
);
1451 if (lstat(path
->buf
, &st
))
1453 else if (S_ISDIR(st
.st_mode
)) {
1454 if (!remove_dir_recurse(path
, flag
, &kept_down
))
1455 continue; /* happy */
1456 } else if (!only_empty
&& !unlink(path
->buf
))
1457 continue; /* happy, too */
1459 /* path too long, stat fails, or non-directory still exists */
1465 strbuf_setlen(path
, original_len
);
1466 if (!ret
&& !keep_toplevel
&& !kept_down
)
1467 ret
= rmdir(path
->buf
);
1470 * report the uplevel that it is not an error that we
1471 * did not rmdir() our directory.
1477 int remove_dir_recursively(struct strbuf
*path
, int flag
)
1479 return remove_dir_recurse(path
, flag
, NULL
);
1482 void setup_standard_excludes(struct dir_struct
*dir
)
1486 dir
->exclude_per_dir
= ".gitignore";
1487 path
= git_path("info/exclude");
1488 if (!access(path
, R_OK
))
1489 add_excludes_from_file(dir
, path
);
1490 if (excludes_file
&& !access(excludes_file
, R_OK
))
1491 add_excludes_from_file(dir
, excludes_file
);
1494 int remove_path(const char *name
)
1498 if (unlink(name
) && errno
!= ENOENT
)
1501 slash
= strrchr(name
, '/');
1503 char *dirs
= xstrdup(name
);
1504 slash
= dirs
+ (slash
- name
);
1507 } while (rmdir(dirs
) == 0 && (slash
= strrchr(dirs
, '/')));
1513 static int pathspec_item_cmp(const void *a_
, const void *b_
)
1515 struct pathspec_item
*a
, *b
;
1517 a
= (struct pathspec_item
*)a_
;
1518 b
= (struct pathspec_item
*)b_
;
1519 return strcmp(a
->match
, b
->match
);
1522 int init_pathspec(struct pathspec
*pathspec
, const char **paths
)
1524 const char **p
= paths
;
1527 memset(pathspec
, 0, sizeof(*pathspec
));
1532 pathspec
->raw
= paths
;
1533 pathspec
->nr
= p
- paths
;
1537 pathspec
->items
= xmalloc(sizeof(struct pathspec_item
)*pathspec
->nr
);
1538 for (i
= 0; i
< pathspec
->nr
; i
++) {
1539 struct pathspec_item
*item
= pathspec
->items
+i
;
1540 const char *path
= paths
[i
];
1543 item
->len
= strlen(path
);
1544 item
->use_wildcard
= !no_wildcard(path
);
1545 if (item
->use_wildcard
)
1546 pathspec
->has_wildcard
= 1;
1549 qsort(pathspec
->items
, pathspec
->nr
,
1550 sizeof(struct pathspec_item
), pathspec_item_cmp
);
1555 void free_pathspec(struct pathspec
*pathspec
)
1557 free(pathspec
->items
);
1558 pathspec
->items
= NULL
;
1562 * Frees memory within dir which was allocated for exclude lists and
1563 * the exclude_stack. Does not free dir itself.
1565 void clear_directory(struct dir_struct
*dir
)
1568 struct exclude_list_group
*group
;
1569 struct exclude_list
*el
;
1570 struct exclude_stack
*stk
;
1572 for (i
= EXC_CMDL
; i
<= EXC_FILE
; i
++) {
1573 group
= &dir
->exclude_list_group
[i
];
1574 for (j
= 0; j
< group
->nr
; j
++) {
1577 free((char *)el
->src
);
1578 clear_exclude_list(el
);
1583 stk
= dir
->exclude_stack
;
1585 struct exclude_stack
*prev
= stk
->prev
;