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
)
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
;
376 ALLOC_GROW(el
->excludes
, el
->nr
+ 1, el
->alloc
);
377 el
->excludes
[el
->nr
++] = x
;
380 static void *read_skip_worktree_file_from_index(const char *path
, size_t *size
)
384 enum object_type type
;
386 struct index_state
*istate
= &the_index
;
389 pos
= index_name_pos(istate
, path
, len
);
392 if (!ce_skip_worktree(istate
->cache
[pos
]))
394 data
= read_sha1_file(istate
->cache
[pos
]->sha1
, &type
, &sz
);
395 if (!data
|| type
!= OBJ_BLOB
) {
403 void free_excludes(struct exclude_list
*el
)
407 for (i
= 0; i
< el
->nr
; i
++)
408 free(el
->excludes
[i
]);
415 int add_excludes_from_file_to_list(const char *fname
,
419 struct exclude_list
*el
,
427 fd
= open(fname
, O_RDONLY
);
428 if (fd
< 0 || fstat(fd
, &st
) < 0) {
432 (buf
= read_skip_worktree_file_from_index(fname
, &size
)) == NULL
)
438 if (buf
[size
-1] != '\n') {
439 buf
= xrealloc(buf
, size
+1);
444 size
= xsize_t(st
.st_size
);
449 buf
= xmalloc(size
+1);
450 if (read_in_full(fd
, buf
, size
) != size
) {
462 for (i
= 0; i
< size
; i
++) {
463 if (buf
[i
] == '\n') {
464 if (entry
!= buf
+ i
&& entry
[0] != '#') {
465 buf
[i
- (i
&& buf
[i
-1] == '\r')] = 0;
466 add_exclude(entry
, base
, baselen
, el
);
474 void add_excludes_from_file(struct dir_struct
*dir
, const char *fname
)
476 if (add_excludes_from_file_to_list(fname
, "", 0, NULL
,
477 &dir
->exclude_list
[EXC_FILE
], 0) < 0)
478 die("cannot use %s as an exclude file", fname
);
482 * Loads the per-directory exclude list for the substring of base
483 * which has a char length of baselen.
485 static void prep_exclude(struct dir_struct
*dir
, const char *base
, int baselen
)
487 struct exclude_list
*el
;
488 struct exclude_stack
*stk
= NULL
;
491 if ((!dir
->exclude_per_dir
) ||
492 (baselen
+ strlen(dir
->exclude_per_dir
) >= PATH_MAX
))
493 return; /* too long a path -- ignore */
495 /* Pop the directories that are not the prefix of the path being checked. */
496 el
= &dir
->exclude_list
[EXC_DIRS
];
497 while ((stk
= dir
->exclude_stack
) != NULL
) {
498 if (stk
->baselen
<= baselen
&&
499 !strncmp(dir
->basebuf
, base
, stk
->baselen
))
501 dir
->exclude_stack
= stk
->prev
;
502 while (stk
->exclude_ix
< el
->nr
)
503 free(el
->excludes
[--el
->nr
]);
508 /* Read from the parent directories and push them down. */
509 current
= stk
? stk
->baselen
: -1;
510 while (current
< baselen
) {
511 struct exclude_stack
*stk
= xcalloc(1, sizeof(*stk
));
519 cp
= strchr(base
+ current
+ 1, '/');
521 die("oops in prep_exclude");
524 stk
->prev
= dir
->exclude_stack
;
525 stk
->baselen
= cp
- base
;
526 stk
->exclude_ix
= el
->nr
;
527 memcpy(dir
->basebuf
+ current
, base
+ current
,
528 stk
->baselen
- current
);
529 strcpy(dir
->basebuf
+ stk
->baselen
, dir
->exclude_per_dir
);
530 add_excludes_from_file_to_list(dir
->basebuf
,
531 dir
->basebuf
, stk
->baselen
,
532 &stk
->filebuf
, el
, 1);
533 dir
->exclude_stack
= stk
;
534 current
= stk
->baselen
;
536 dir
->basebuf
[baselen
] = '\0';
539 int match_basename(const char *basename
, int basenamelen
,
540 const char *pattern
, int prefix
, int patternlen
,
543 if (prefix
== patternlen
) {
544 if (!strcmp_icase(pattern
, basename
))
546 } else if (flags
& EXC_FLAG_ENDSWITH
) {
547 if (patternlen
- 1 <= basenamelen
&&
548 !strcmp_icase(pattern
+ 1,
549 basename
+ basenamelen
- patternlen
+ 1))
552 if (fnmatch_icase(pattern
, basename
, 0) == 0)
558 int match_pathname(const char *pathname
, int pathlen
,
559 const char *base
, int baselen
,
560 const char *pattern
, int prefix
, int patternlen
,
567 * match with FNM_PATHNAME; the pattern has base implicitly
570 if (*pattern
== '/') {
576 * baselen does not count the trailing slash. base[] may or
577 * may not end with a trailing slash though.
579 if (pathlen
< baselen
+ 1 ||
580 (baselen
&& pathname
[baselen
] != '/') ||
581 strncmp_icase(pathname
, base
, baselen
))
584 namelen
= baselen
? pathlen
- baselen
- 1 : pathlen
;
585 name
= pathname
+ pathlen
- namelen
;
589 * if the non-wildcard part is longer than the
590 * remaining pathname, surely it cannot match.
592 if (prefix
> namelen
)
595 if (strncmp_icase(pattern
, name
, prefix
))
602 return fnmatch_icase(pattern
, name
, FNM_PATHNAME
) == 0;
606 * Scan the given exclude list in reverse to see whether pathname
607 * should be ignored. The first match (i.e. the last on the list), if
608 * any, determines the fate. Returns the exclude_list element which
609 * matched, or NULL for undecided.
611 static struct exclude
*last_exclude_matching_from_list(const char *pathname
,
613 const char *basename
,
615 struct exclude_list
*el
)
620 return NULL
; /* undefined */
622 for (i
= el
->nr
- 1; 0 <= i
; i
--) {
623 struct exclude
*x
= el
->excludes
[i
];
624 const char *exclude
= x
->pattern
;
625 int prefix
= x
->nowildcardlen
;
627 if (x
->flags
& EXC_FLAG_MUSTBEDIR
) {
628 if (*dtype
== DT_UNKNOWN
)
629 *dtype
= get_dtype(NULL
, pathname
, pathlen
);
630 if (*dtype
!= DT_DIR
)
634 if (x
->flags
& EXC_FLAG_NODIR
) {
635 if (match_basename(basename
,
636 pathlen
- (basename
- pathname
),
637 exclude
, prefix
, x
->patternlen
,
643 assert(x
->baselen
== 0 || x
->base
[x
->baselen
- 1] == '/');
644 if (match_pathname(pathname
, pathlen
,
645 x
->base
, x
->baselen
? x
->baselen
- 1 : 0,
646 exclude
, prefix
, x
->patternlen
, x
->flags
))
649 return NULL
; /* undecided */
653 * Scan the list and let the last match determine the fate.
654 * Return 1 for exclude, 0 for include and -1 for undecided.
656 int is_excluded_from_list(const char *pathname
,
657 int pathlen
, const char *basename
, int *dtype
,
658 struct exclude_list
*el
)
660 struct exclude
*exclude
;
661 exclude
= last_exclude_matching_from_list(pathname
, pathlen
, basename
, dtype
, el
);
663 return exclude
->flags
& EXC_FLAG_NEGATIVE
? 0 : 1;
664 return -1; /* undecided */
667 static int is_excluded(struct dir_struct
*dir
, const char *pathname
, int *dtype_p
)
669 int pathlen
= strlen(pathname
);
671 const char *basename
= strrchr(pathname
, '/');
672 basename
= (basename
) ? basename
+1 : pathname
;
674 prep_exclude(dir
, pathname
, basename
-pathname
);
675 for (st
= EXC_CMDL
; st
<= EXC_FILE
; st
++) {
676 switch (is_excluded_from_list(pathname
, pathlen
,
678 &dir
->exclude_list
[st
])) {
688 void path_exclude_check_init(struct path_exclude_check
*check
,
689 struct dir_struct
*dir
)
692 strbuf_init(&check
->path
, 256);
695 void path_exclude_check_clear(struct path_exclude_check
*check
)
697 strbuf_release(&check
->path
);
701 * Is this name excluded? This is for a caller like show_files() that
702 * do not honor directory hierarchy and iterate through paths that are
703 * possibly in an ignored directory.
705 * A path to a directory known to be excluded is left in check->path to
706 * optimize for repeated checks for files in the same excluded directory.
708 int is_path_excluded(struct path_exclude_check
*check
,
709 const char *name
, int namelen
, int *dtype
)
712 struct strbuf
*path
= &check
->path
;
715 * we allow the caller to pass namelen as an optimization; it
716 * must match the length of the name, as we eventually call
717 * is_excluded() on the whole name string.
720 namelen
= strlen(name
);
723 path
->len
<= namelen
&&
724 !memcmp(name
, path
->buf
, path
->len
) &&
725 (!name
[path
->len
] || name
[path
->len
] == '/'))
728 strbuf_setlen(path
, 0);
729 for (i
= 0; name
[i
]; i
++) {
734 if (is_excluded(check
->dir
, path
->buf
, &dt
))
737 strbuf_addch(path
, ch
);
740 /* An entry in the index; cannot be a directory with subentries */
741 strbuf_setlen(path
, 0);
743 return is_excluded(check
->dir
, name
, dtype
);
746 static struct dir_entry
*dir_entry_new(const char *pathname
, int len
)
748 struct dir_entry
*ent
;
750 ent
= xmalloc(sizeof(*ent
) + len
+ 1);
752 memcpy(ent
->name
, pathname
, len
);
757 static struct dir_entry
*dir_add_name(struct dir_struct
*dir
, const char *pathname
, int len
)
759 if (cache_name_exists(pathname
, len
, ignore_case
))
762 ALLOC_GROW(dir
->entries
, dir
->nr
+1, dir
->alloc
);
763 return dir
->entries
[dir
->nr
++] = dir_entry_new(pathname
, len
);
766 struct dir_entry
*dir_add_ignored(struct dir_struct
*dir
, const char *pathname
, int len
)
768 if (!cache_name_is_other(pathname
, len
))
771 ALLOC_GROW(dir
->ignored
, dir
->ignored_nr
+1, dir
->ignored_alloc
);
772 return dir
->ignored
[dir
->ignored_nr
++] = dir_entry_new(pathname
, len
);
776 index_nonexistent
= 0,
782 * Do not use the alphabetically stored index to look up
783 * the directory name; instead, use the case insensitive
786 static enum exist_status
directory_exists_in_index_icase(const char *dirname
, int len
)
788 struct cache_entry
*ce
= index_name_exists(&the_index
, dirname
, len
+ 1, ignore_case
);
789 unsigned char endchar
;
792 return index_nonexistent
;
793 endchar
= ce
->name
[len
];
796 * The cache_entry structure returned will contain this dirname
797 * and possibly additional path components.
800 return index_directory
;
803 * If there are no additional path components, then this cache_entry
804 * represents a submodule. Submodules, despite being directories,
805 * are stored in the cache without a closing slash.
807 if (!endchar
&& S_ISGITLINK(ce
->ce_mode
))
810 /* This should never be hit, but it exists just in case. */
811 return index_nonexistent
;
815 * The index sorts alphabetically by entry name, which
816 * means that a gitlink sorts as '\0' at the end, while
817 * a directory (which is defined not as an entry, but as
818 * the files it contains) will sort with the '/' at the
821 static enum exist_status
directory_exists_in_index(const char *dirname
, int len
)
826 return directory_exists_in_index_icase(dirname
, len
);
828 pos
= cache_name_pos(dirname
, len
);
831 while (pos
< active_nr
) {
832 struct cache_entry
*ce
= active_cache
[pos
++];
833 unsigned char endchar
;
835 if (strncmp(ce
->name
, dirname
, len
))
837 endchar
= ce
->name
[len
];
841 return index_directory
;
842 if (!endchar
&& S_ISGITLINK(ce
->ce_mode
))
845 return index_nonexistent
;
849 * When we find a directory when traversing the filesystem, we
850 * have three distinct cases:
853 * - see it as a directory
856 * and which one we choose depends on a combination of existing
857 * git index contents and the flags passed into the directory
860 * Case 1: If we *already* have entries in the index under that
861 * directory name, we always recurse into the directory to see
864 * Case 2: If we *already* have that directory name as a gitlink,
865 * we always continue to see it as a gitlink, regardless of whether
866 * there is an actual git directory there or not (it might not
867 * be checked out as a subproject!)
869 * Case 3: if we didn't have it in the index previously, we
870 * have a few sub-cases:
872 * (a) if "show_other_directories" is true, we show it as
873 * just a directory, unless "hide_empty_directories" is
874 * also true and the directory is empty, in which case
875 * we just ignore it entirely.
876 * (b) if it looks like a git directory, and we don't have
877 * 'no_gitlinks' set we treat it as a gitlink, and show it
879 * (c) otherwise, we recurse into it.
881 enum directory_treatment
{
884 recurse_into_directory
887 static enum directory_treatment
treat_directory(struct dir_struct
*dir
,
888 const char *dirname
, int len
,
889 const struct path_simplify
*simplify
)
891 /* The "len-1" is to strip the final '/' */
892 switch (directory_exists_in_index(dirname
, len
-1)) {
893 case index_directory
:
894 return recurse_into_directory
;
897 if (dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
)
898 return ignore_directory
;
899 return show_directory
;
901 case index_nonexistent
:
902 if (dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
)
904 if (!(dir
->flags
& DIR_NO_GITLINKS
)) {
905 unsigned char sha1
[20];
906 if (resolve_gitlink_ref(dirname
, "HEAD", sha1
) == 0)
907 return show_directory
;
909 return recurse_into_directory
;
912 /* This is the "show_other_directories" case */
913 if (!(dir
->flags
& DIR_HIDE_EMPTY_DIRECTORIES
))
914 return show_directory
;
915 if (!read_directory_recursive(dir
, dirname
, len
, 1, simplify
))
916 return ignore_directory
;
917 return show_directory
;
921 * This is an inexact early pruning of any recursive directory
922 * reading - if the path cannot possibly be in the pathspec,
923 * return true, and we'll skip it early.
925 static int simplify_away(const char *path
, int pathlen
, const struct path_simplify
*simplify
)
929 const char *match
= simplify
->path
;
930 int len
= simplify
->len
;
936 if (!memcmp(path
, match
, len
))
946 * This function tells us whether an excluded path matches a
947 * list of "interesting" pathspecs. That is, whether a path matched
948 * by any of the pathspecs could possibly be ignored by excluding
949 * the specified path. This can happen if:
951 * 1. the path is mentioned explicitly in the pathspec
953 * 2. the path is a directory prefix of some element in the
956 static int exclude_matches_pathspec(const char *path
, int len
,
957 const struct path_simplify
*simplify
)
960 for (; simplify
->path
; simplify
++) {
961 if (len
== simplify
->len
962 && !memcmp(path
, simplify
->path
, len
))
964 if (len
< simplify
->len
965 && simplify
->path
[len
] == '/'
966 && !memcmp(path
, simplify
->path
, len
))
973 static int get_index_dtype(const char *path
, int len
)
976 struct cache_entry
*ce
;
978 ce
= cache_name_exists(path
, len
, 0);
980 if (!ce_uptodate(ce
))
982 if (S_ISGITLINK(ce
->ce_mode
))
985 * Nobody actually cares about the
986 * difference between DT_LNK and DT_REG
991 /* Try to look it up as a directory */
992 pos
= cache_name_pos(path
, len
);
996 while (pos
< active_nr
) {
997 ce
= active_cache
[pos
++];
998 if (strncmp(ce
->name
, path
, len
))
1000 if (ce
->name
[len
] > '/')
1002 if (ce
->name
[len
] < '/')
1004 if (!ce_uptodate(ce
))
1005 break; /* continue? */
1011 static int get_dtype(struct dirent
*de
, const char *path
, int len
)
1013 int dtype
= de
? DTYPE(de
) : DT_UNKNOWN
;
1016 if (dtype
!= DT_UNKNOWN
)
1018 dtype
= get_index_dtype(path
, len
);
1019 if (dtype
!= DT_UNKNOWN
)
1021 if (lstat(path
, &st
))
1023 if (S_ISREG(st
.st_mode
))
1025 if (S_ISDIR(st
.st_mode
))
1027 if (S_ISLNK(st
.st_mode
))
1032 enum path_treatment
{
1038 static enum path_treatment
treat_one_path(struct dir_struct
*dir
,
1039 struct strbuf
*path
,
1040 const struct path_simplify
*simplify
,
1041 int dtype
, struct dirent
*de
)
1043 int exclude
= is_excluded(dir
, path
->buf
, &dtype
);
1044 if (exclude
&& (dir
->flags
& DIR_COLLECT_IGNORED
)
1045 && exclude_matches_pathspec(path
->buf
, path
->len
, simplify
))
1046 dir_add_ignored(dir
, path
->buf
, path
->len
);
1049 * Excluded? If we don't explicitly want to show
1050 * ignored files, ignore it
1052 if (exclude
&& !(dir
->flags
& DIR_SHOW_IGNORED
))
1053 return path_ignored
;
1055 if (dtype
== DT_UNKNOWN
)
1056 dtype
= get_dtype(de
, path
->buf
, path
->len
);
1059 * Do we want to see just the ignored files?
1060 * We still need to recurse into directories,
1061 * even if we don't ignore them, since the
1062 * directory may contain files that we do..
1064 if (!exclude
&& (dir
->flags
& DIR_SHOW_IGNORED
)) {
1065 if (dtype
!= DT_DIR
)
1066 return path_ignored
;
1071 return path_ignored
;
1073 strbuf_addch(path
, '/');
1074 switch (treat_directory(dir
, path
->buf
, path
->len
, simplify
)) {
1075 case show_directory
:
1076 if (exclude
!= !!(dir
->flags
1077 & DIR_SHOW_IGNORED
))
1078 return path_ignored
;
1080 case recurse_into_directory
:
1081 return path_recurse
;
1082 case ignore_directory
:
1083 return path_ignored
;
1090 return path_handled
;
1093 static enum path_treatment
treat_path(struct dir_struct
*dir
,
1095 struct strbuf
*path
,
1097 const struct path_simplify
*simplify
)
1101 if (is_dot_or_dotdot(de
->d_name
) || !strcmp(de
->d_name
, ".git"))
1102 return path_ignored
;
1103 strbuf_setlen(path
, baselen
);
1104 strbuf_addstr(path
, de
->d_name
);
1105 if (simplify_away(path
->buf
, path
->len
, simplify
))
1106 return path_ignored
;
1109 return treat_one_path(dir
, path
, simplify
, dtype
, de
);
1113 * Read a directory tree. We currently ignore anything but
1114 * directories, regular files and symlinks. That's because git
1115 * doesn't handle them at all yet. Maybe that will change some
1118 * Also, we ignore the name ".git" (even if it is not a directory).
1119 * That likely will not change.
1121 static int read_directory_recursive(struct dir_struct
*dir
,
1122 const char *base
, int baselen
,
1124 const struct path_simplify
*simplify
)
1129 struct strbuf path
= STRBUF_INIT
;
1131 strbuf_add(&path
, base
, baselen
);
1133 fdir
= opendir(path
.len
? path
.buf
: ".");
1137 while ((de
= readdir(fdir
)) != NULL
) {
1138 switch (treat_path(dir
, de
, &path
, baselen
, simplify
)) {
1140 contents
+= read_directory_recursive(dir
, path
.buf
,
1152 dir_add_name(dir
, path
.buf
, path
.len
);
1156 strbuf_release(&path
);
1161 static int cmp_name(const void *p1
, const void *p2
)
1163 const struct dir_entry
*e1
= *(const struct dir_entry
**)p1
;
1164 const struct dir_entry
*e2
= *(const struct dir_entry
**)p2
;
1166 return cache_name_compare(e1
->name
, e1
->len
,
1170 static struct path_simplify
*create_simplify(const char **pathspec
)
1173 struct path_simplify
*simplify
= NULL
;
1178 for (nr
= 0 ; ; nr
++) {
1181 alloc
= alloc_nr(alloc
);
1182 simplify
= xrealloc(simplify
, alloc
* sizeof(*simplify
));
1184 match
= *pathspec
++;
1187 simplify
[nr
].path
= match
;
1188 simplify
[nr
].len
= simple_length(match
);
1190 simplify
[nr
].path
= NULL
;
1191 simplify
[nr
].len
= 0;
1195 static void free_simplify(struct path_simplify
*simplify
)
1200 static int treat_leading_path(struct dir_struct
*dir
,
1201 const char *path
, int len
,
1202 const struct path_simplify
*simplify
)
1204 struct strbuf sb
= STRBUF_INIT
;
1205 int baselen
, rc
= 0;
1208 while (len
&& path
[len
- 1] == '/')
1214 cp
= path
+ baselen
+ !!baselen
;
1215 cp
= memchr(cp
, '/', path
+ len
- cp
);
1219 baselen
= cp
- path
;
1220 strbuf_setlen(&sb
, 0);
1221 strbuf_add(&sb
, path
, baselen
);
1222 if (!is_directory(sb
.buf
))
1224 if (simplify_away(sb
.buf
, sb
.len
, simplify
))
1226 if (treat_one_path(dir
, &sb
, simplify
,
1227 DT_DIR
, NULL
) == path_ignored
)
1228 break; /* do not recurse into it */
1229 if (len
<= baselen
) {
1231 break; /* finished checking */
1234 strbuf_release(&sb
);
1238 int read_directory(struct dir_struct
*dir
, const char *path
, int len
, const char **pathspec
)
1240 struct path_simplify
*simplify
;
1242 if (has_symlink_leading_path(path
, len
))
1245 simplify
= create_simplify(pathspec
);
1246 if (!len
|| treat_leading_path(dir
, path
, len
, simplify
))
1247 read_directory_recursive(dir
, path
, len
, 0, simplify
);
1248 free_simplify(simplify
);
1249 qsort(dir
->entries
, dir
->nr
, sizeof(struct dir_entry
*), cmp_name
);
1250 qsort(dir
->ignored
, dir
->ignored_nr
, sizeof(struct dir_entry
*), cmp_name
);
1254 int file_exists(const char *f
)
1257 return lstat(f
, &sb
) == 0;
1261 * Given two normalized paths (a trailing slash is ok), if subdir is
1262 * outside dir, return -1. Otherwise return the offset in subdir that
1263 * can be used as relative path to dir.
1265 int dir_inside_of(const char *subdir
, const char *dir
)
1269 assert(dir
&& subdir
&& *dir
&& *subdir
);
1271 while (*dir
&& *subdir
&& *dir
== *subdir
) {
1277 /* hel[p]/me vs hel[l]/yeah */
1278 if (*dir
&& *subdir
)
1282 return !*dir
? offset
: -1; /* same dir */
1284 /* foo/[b]ar vs foo/[] */
1285 if (is_dir_sep(dir
[-1]))
1286 return is_dir_sep(subdir
[-1]) ? offset
: -1;
1288 /* foo[/]bar vs foo[] */
1289 return is_dir_sep(*subdir
) ? offset
+ 1 : -1;
1292 int is_inside_dir(const char *dir
)
1297 if (!getcwd(cwd
, sizeof(cwd
)))
1298 die_errno("can't find the current directory");
1299 return dir_inside_of(cwd
, dir
) >= 0;
1302 int is_empty_dir(const char *path
)
1304 DIR *dir
= opendir(path
);
1311 while ((e
= readdir(dir
)) != NULL
)
1312 if (!is_dot_or_dotdot(e
->d_name
)) {
1321 static int remove_dir_recurse(struct strbuf
*path
, int flag
, int *kept_up
)
1325 int ret
= 0, original_len
= path
->len
, len
, kept_down
= 0;
1326 int only_empty
= (flag
& REMOVE_DIR_EMPTY_ONLY
);
1327 int keep_toplevel
= (flag
& REMOVE_DIR_KEEP_TOPLEVEL
);
1328 unsigned char submodule_head
[20];
1330 if ((flag
& REMOVE_DIR_KEEP_NESTED_GIT
) &&
1331 !resolve_gitlink_ref(path
->buf
, "HEAD", submodule_head
)) {
1332 /* Do not descend and nuke a nested git work tree. */
1338 flag
&= ~REMOVE_DIR_KEEP_TOPLEVEL
;
1339 dir
= opendir(path
->buf
);
1341 /* an empty dir could be removed even if it is unreadble */
1343 return rmdir(path
->buf
);
1347 if (path
->buf
[original_len
- 1] != '/')
1348 strbuf_addch(path
, '/');
1351 while ((e
= readdir(dir
)) != NULL
) {
1353 if (is_dot_or_dotdot(e
->d_name
))
1356 strbuf_setlen(path
, len
);
1357 strbuf_addstr(path
, e
->d_name
);
1358 if (lstat(path
->buf
, &st
))
1360 else if (S_ISDIR(st
.st_mode
)) {
1361 if (!remove_dir_recurse(path
, flag
, &kept_down
))
1362 continue; /* happy */
1363 } else if (!only_empty
&& !unlink(path
->buf
))
1364 continue; /* happy, too */
1366 /* path too long, stat fails, or non-directory still exists */
1372 strbuf_setlen(path
, original_len
);
1373 if (!ret
&& !keep_toplevel
&& !kept_down
)
1374 ret
= rmdir(path
->buf
);
1377 * report the uplevel that it is not an error that we
1378 * did not rmdir() our directory.
1384 int remove_dir_recursively(struct strbuf
*path
, int flag
)
1386 return remove_dir_recurse(path
, flag
, NULL
);
1389 void setup_standard_excludes(struct dir_struct
*dir
)
1393 dir
->exclude_per_dir
= ".gitignore";
1394 path
= git_path("info/exclude");
1395 if (!access(path
, R_OK
))
1396 add_excludes_from_file(dir
, path
);
1397 if (excludes_file
&& !access(excludes_file
, R_OK
))
1398 add_excludes_from_file(dir
, excludes_file
);
1401 int remove_path(const char *name
)
1405 if (unlink(name
) && errno
!= ENOENT
)
1408 slash
= strrchr(name
, '/');
1410 char *dirs
= xstrdup(name
);
1411 slash
= dirs
+ (slash
- name
);
1414 } while (rmdir(dirs
) == 0 && (slash
= strrchr(dirs
, '/')));
1420 static int pathspec_item_cmp(const void *a_
, const void *b_
)
1422 struct pathspec_item
*a
, *b
;
1424 a
= (struct pathspec_item
*)a_
;
1425 b
= (struct pathspec_item
*)b_
;
1426 return strcmp(a
->match
, b
->match
);
1429 int init_pathspec(struct pathspec
*pathspec
, const char **paths
)
1431 const char **p
= paths
;
1434 memset(pathspec
, 0, sizeof(*pathspec
));
1439 pathspec
->raw
= paths
;
1440 pathspec
->nr
= p
- paths
;
1444 pathspec
->items
= xmalloc(sizeof(struct pathspec_item
)*pathspec
->nr
);
1445 for (i
= 0; i
< pathspec
->nr
; i
++) {
1446 struct pathspec_item
*item
= pathspec
->items
+i
;
1447 const char *path
= paths
[i
];
1450 item
->len
= strlen(path
);
1451 item
->use_wildcard
= !no_wildcard(path
);
1452 if (item
->use_wildcard
)
1453 pathspec
->has_wildcard
= 1;
1456 qsort(pathspec
->items
, pathspec
->nr
,
1457 sizeof(struct pathspec_item
), pathspec_item_cmp
);
1462 void free_pathspec(struct pathspec
*pathspec
)
1464 free(pathspec
->items
);
1465 pathspec
->items
= NULL
;