2 * This handles recursive filename detection with exclude
3 * files, index knowledge etc..
5 * Copyright (C) Linus Torvalds, 2005-2006
6 * Junio Hamano, 2005-2006
12 struct path_simplify
{
17 static int read_directory_recursive(struct dir_struct
*dir
, const char *path
, int len
,
18 int check_only
, const struct path_simplify
*simplify
);
19 static int get_dtype(struct dirent
*de
, const char *path
, int len
);
21 /* helper string functions with support for the ignore_case flag */
22 int strcmp_icase(const char *a
, const char *b
)
24 return ignore_case
? strcasecmp(a
, b
) : strcmp(a
, b
);
27 int strncmp_icase(const char *a
, const char *b
, size_t count
)
29 return ignore_case
? strncasecmp(a
, b
, count
) : strncmp(a
, b
, count
);
32 int fnmatch_icase(const char *pattern
, const char *string
, int flags
)
34 return fnmatch(pattern
, string
, flags
| (ignore_case
? FNM_CASEFOLD
: 0));
37 static int fnmatch_icase_mem(const char *pattern
, int patternlen
,
38 const char *string
, int stringlen
,
42 struct strbuf pat_buf
= STRBUF_INIT
;
43 struct strbuf str_buf
= STRBUF_INIT
;
44 const char *use_pat
= pattern
;
45 const char *use_str
= string
;
47 if (pattern
[patternlen
]) {
48 strbuf_add(&pat_buf
, pattern
, patternlen
);
49 use_pat
= pat_buf
.buf
;
51 if (string
[stringlen
]) {
52 strbuf_add(&str_buf
, string
, stringlen
);
53 use_str
= str_buf
.buf
;
56 match_status
= fnmatch_icase(use_pat
, use_str
, flags
);
58 strbuf_release(&pat_buf
);
59 strbuf_release(&str_buf
);
64 static size_t common_prefix_len(const char **pathspec
)
66 const char *n
, *first
;
73 while ((n
= *pathspec
++)) {
75 for (i
= 0; first
== n
|| i
< max
; i
++) {
77 if (!c
|| c
!= first
[i
] || is_glob_special(c
))
82 if (first
== n
|| len
< max
) {
92 * Returns a copy of the longest leading path common among all
95 char *common_prefix(const char **pathspec
)
97 unsigned long len
= common_prefix_len(pathspec
);
99 return len
? xmemdupz(*pathspec
, len
) : NULL
;
102 int fill_directory(struct dir_struct
*dir
, const char **pathspec
)
107 * Calculate common prefix for the pathspec, and
108 * use that to optimize the directory walk
110 len
= common_prefix_len(pathspec
);
112 /* Read the directory and prune it */
113 read_directory(dir
, pathspec
? *pathspec
: "", len
, pathspec
);
117 int within_depth(const char *name
, int namelen
,
118 int depth
, int max_depth
)
120 const char *cp
= name
, *cpe
= name
+ namelen
;
126 if (depth
> max_depth
)
133 * Does 'match' match the given name?
134 * A match is found if
136 * (1) the 'match' string is leading directory of 'name', or
137 * (2) the 'match' string is a wildcard and matches 'name', or
138 * (3) the 'match' string is exactly the same as 'name'.
140 * and the return value tells which case it was.
142 * It returns 0 when there is no match.
144 static int match_one(const char *match
, const char *name
, int namelen
)
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' || is_glob_special(c1
))
166 unsigned char c1
= *match
;
167 unsigned char c2
= *name
;
168 if (c1
== '\0' || is_glob_special(c1
))
180 * If we don't match the matchstring exactly,
181 * we need to match by fnmatch
183 matchlen
= strlen(match
);
184 if (strncmp_icase(match
, name
, matchlen
))
185 return !fnmatch_icase(match
, name
, 0) ? MATCHED_FNMATCH
: 0;
187 if (namelen
== matchlen
)
188 return MATCHED_EXACTLY
;
189 if (match
[matchlen
-1] == '/' || name
[matchlen
] == '/')
190 return MATCHED_RECURSIVELY
;
195 * Given a name and a list of pathspecs, see if the name matches
196 * any of the pathspecs. The caller is also interested in seeing
197 * all pathspec matches some names it calls this function with
198 * (otherwise the user could have mistyped the unmatched pathspec),
199 * and a mark is left in seen[] array for pathspec element that
200 * actually matched anything.
202 int match_pathspec(const char **pathspec
, const char *name
, int namelen
,
203 int prefix
, char *seen
)
213 for (i
= 0; pathspec
[i
] != NULL
; i
++) {
215 const char *match
= pathspec
[i
] + prefix
;
216 if (seen
&& seen
[i
] == MATCHED_EXACTLY
)
218 how
= match_one(match
, name
, namelen
);
222 if (seen
&& seen
[i
] < how
)
230 * Does 'match' match the given name?
231 * A match is found if
233 * (1) the 'match' string is leading directory of 'name', or
234 * (2) the 'match' string is a wildcard and matches 'name', or
235 * (3) the 'match' string is exactly the same as 'name'.
237 * and the return value tells which case it was.
239 * It returns 0 when there is no match.
241 static int match_pathspec_item(const struct pathspec_item
*item
, int prefix
,
242 const char *name
, int namelen
)
244 /* name/namelen has prefix cut off by caller */
245 const char *match
= item
->match
+ prefix
;
246 int matchlen
= item
->len
- prefix
;
248 /* If the match was just the prefix, we matched */
250 return MATCHED_RECURSIVELY
;
252 if (matchlen
<= namelen
&& !strncmp(match
, name
, matchlen
)) {
253 if (matchlen
== namelen
)
254 return MATCHED_EXACTLY
;
256 if (match
[matchlen
-1] == '/' || name
[matchlen
] == '/')
257 return MATCHED_RECURSIVELY
;
260 if (item
->use_wildcard
&& !fnmatch(match
, name
, 0))
261 return MATCHED_FNMATCH
;
267 * Given a name and a list of pathspecs, see if the name matches
268 * any of the pathspecs. The caller is also interested in seeing
269 * all pathspec matches some names it calls this function with
270 * (otherwise the user could have mistyped the unmatched pathspec),
271 * and a mark is left in seen[] array for pathspec element that
272 * actually matched anything.
274 int match_pathspec_depth(const struct pathspec
*ps
,
275 const char *name
, int namelen
,
276 int prefix
, char *seen
)
281 if (!ps
->recursive
|| ps
->max_depth
== -1)
282 return MATCHED_RECURSIVELY
;
284 if (within_depth(name
, namelen
, 0, ps
->max_depth
))
285 return MATCHED_EXACTLY
;
293 for (i
= ps
->nr
- 1; i
>= 0; i
--) {
295 if (seen
&& seen
[i
] == MATCHED_EXACTLY
)
297 how
= match_pathspec_item(ps
->items
+i
, prefix
, name
, namelen
);
298 if (ps
->recursive
&& ps
->max_depth
!= -1 &&
299 how
&& how
!= MATCHED_FNMATCH
) {
300 int len
= ps
->items
[i
].len
;
301 if (name
[len
] == '/')
303 if (within_depth(name
+len
, namelen
-len
, 0, ps
->max_depth
))
304 how
= MATCHED_EXACTLY
;
311 if (seen
&& seen
[i
] < how
)
319 * Return the length of the "simple" part of a path match limiter.
321 static int simple_length(const char *match
)
326 unsigned char c
= *match
++;
328 if (c
== '\0' || is_glob_special(c
))
333 static int no_wildcard(const char *string
)
335 return string
[simple_length(string
)] == '\0';
338 void parse_exclude_pattern(const char **pattern
,
343 const char *p
= *pattern
;
348 *flags
|= EXC_FLAG_NEGATIVE
;
352 if (len
&& p
[len
- 1] == '/') {
354 *flags
|= EXC_FLAG_MUSTBEDIR
;
356 for (i
= 0; i
< len
; i
++) {
361 *flags
|= EXC_FLAG_NODIR
;
362 *nowildcardlen
= simple_length(p
);
364 * we should have excluded the trailing slash from 'p' too,
365 * but that's one more allocation. Instead just make sure
366 * nowildcardlen does not exceed real patternlen
368 if (*nowildcardlen
> len
)
369 *nowildcardlen
= len
;
370 if (*p
== '*' && no_wildcard(p
+ 1))
371 *flags
|= EXC_FLAG_ENDSWITH
;
376 void add_exclude(const char *string
, const char *base
,
377 int baselen
, struct exclude_list
*which
)
384 parse_exclude_pattern(&string
, &patternlen
, &flags
, &nowildcardlen
);
385 if (flags
& EXC_FLAG_MUSTBEDIR
) {
387 x
= xmalloc(sizeof(*x
) + patternlen
+ 1);
389 memcpy(s
, string
, patternlen
);
390 s
[patternlen
] = '\0';
393 x
= xmalloc(sizeof(*x
));
396 x
->patternlen
= patternlen
;
397 x
->nowildcardlen
= nowildcardlen
;
399 x
->baselen
= baselen
;
401 ALLOC_GROW(which
->excludes
, which
->nr
+ 1, which
->alloc
);
402 which
->excludes
[which
->nr
++] = x
;
405 static void *read_skip_worktree_file_from_index(const char *path
, size_t *size
)
409 enum object_type type
;
411 struct index_state
*istate
= &the_index
;
414 pos
= index_name_pos(istate
, path
, len
);
417 if (!ce_skip_worktree(istate
->cache
[pos
]))
419 data
= read_sha1_file(istate
->cache
[pos
]->sha1
, &type
, &sz
);
420 if (!data
|| type
!= OBJ_BLOB
) {
428 void free_excludes(struct exclude_list
*el
)
432 for (i
= 0; i
< el
->nr
; i
++)
433 free(el
->excludes
[i
]);
440 int add_excludes_from_file_to_list(const char *fname
,
444 struct exclude_list
*which
,
452 fd
= open(fname
, O_RDONLY
);
453 if (fd
< 0 || fstat(fd
, &st
) < 0) {
455 warn_on_inaccessible(fname
);
459 (buf
= read_skip_worktree_file_from_index(fname
, &size
)) == NULL
)
465 if (buf
[size
-1] != '\n') {
466 buf
= xrealloc(buf
, size
+1);
471 size
= xsize_t(st
.st_size
);
476 buf
= xmalloc(size
+1);
477 if (read_in_full(fd
, buf
, size
) != size
) {
489 for (i
= 0; i
< size
; i
++) {
490 if (buf
[i
] == '\n') {
491 if (entry
!= buf
+ i
&& entry
[0] != '#') {
492 buf
[i
- (i
&& buf
[i
-1] == '\r')] = 0;
493 add_exclude(entry
, base
, baselen
, which
);
501 void add_excludes_from_file(struct dir_struct
*dir
, const char *fname
)
503 if (add_excludes_from_file_to_list(fname
, "", 0, NULL
,
504 &dir
->exclude_list
[EXC_FILE
], 0) < 0)
505 die("cannot use %s as an exclude file", fname
);
508 static void prep_exclude(struct dir_struct
*dir
, const char *base
, int baselen
)
510 struct exclude_list
*el
;
511 struct exclude_stack
*stk
= NULL
;
514 if ((!dir
->exclude_per_dir
) ||
515 (baselen
+ strlen(dir
->exclude_per_dir
) >= PATH_MAX
))
516 return; /* too long a path -- ignore */
518 /* Pop the ones that are not the prefix of the path being checked. */
519 el
= &dir
->exclude_list
[EXC_DIRS
];
520 while ((stk
= dir
->exclude_stack
) != NULL
) {
521 if (stk
->baselen
<= baselen
&&
522 !strncmp(dir
->basebuf
, base
, stk
->baselen
))
524 dir
->exclude_stack
= stk
->prev
;
525 while (stk
->exclude_ix
< el
->nr
)
526 free(el
->excludes
[--el
->nr
]);
531 /* Read from the parent directories and push them down. */
532 current
= stk
? stk
->baselen
: -1;
533 while (current
< baselen
) {
534 struct exclude_stack
*stk
= xcalloc(1, sizeof(*stk
));
542 cp
= strchr(base
+ current
+ 1, '/');
544 die("oops in prep_exclude");
547 stk
->prev
= dir
->exclude_stack
;
548 stk
->baselen
= cp
- base
;
549 stk
->exclude_ix
= el
->nr
;
550 memcpy(dir
->basebuf
+ current
, base
+ current
,
551 stk
->baselen
- current
);
552 strcpy(dir
->basebuf
+ stk
->baselen
, dir
->exclude_per_dir
);
553 add_excludes_from_file_to_list(dir
->basebuf
,
554 dir
->basebuf
, stk
->baselen
,
555 &stk
->filebuf
, el
, 1);
556 dir
->exclude_stack
= stk
;
557 current
= stk
->baselen
;
559 dir
->basebuf
[baselen
] = '\0';
562 int match_basename(const char *basename
, int basenamelen
,
563 const char *pattern
, int prefix
, int patternlen
,
566 if (prefix
== patternlen
) {
567 if (patternlen
== basenamelen
&&
568 !strncmp_icase(pattern
, basename
, basenamelen
))
570 } else if (flags
& EXC_FLAG_ENDSWITH
) {
571 /* "*literal" matching against "fooliteral" */
572 if (patternlen
- 1 <= basenamelen
&&
573 !strncmp_icase(pattern
+ 1,
574 basename
+ basenamelen
- (patternlen
- 1),
578 if (fnmatch_icase_mem(pattern
, patternlen
,
579 basename
, basenamelen
,
586 int match_pathname(const char *pathname
, int pathlen
,
587 const char *base
, int baselen
,
588 const char *pattern
, int prefix
, int patternlen
,
595 * match with FNM_PATHNAME; the pattern has base implicitly
598 if (*pattern
== '/') {
604 * baselen does not count the trailing slash. base[] may or
605 * may not end with a trailing slash though.
607 if (pathlen
< baselen
+ 1 ||
608 (baselen
&& pathname
[baselen
] != '/') ||
609 strncmp_icase(pathname
, base
, baselen
))
612 namelen
= baselen
? pathlen
- baselen
- 1 : pathlen
;
613 name
= pathname
+ pathlen
- namelen
;
617 * if the non-wildcard part is longer than the
618 * remaining pathname, surely it cannot match.
620 if (prefix
> namelen
)
623 if (strncmp_icase(pattern
, name
, prefix
))
630 return fnmatch_icase(pattern
, name
, FNM_PATHNAME
) == 0;
633 /* Scan the list and let the last match determine the fate.
634 * Return 1 for exclude, 0 for include and -1 for undecided.
636 int excluded_from_list(const char *pathname
,
637 int pathlen
, const char *basename
, int *dtype
,
638 struct exclude_list
*el
)
643 return -1; /* undefined */
645 for (i
= el
->nr
- 1; 0 <= i
; i
--) {
646 struct exclude
*x
= el
->excludes
[i
];
647 const char *exclude
= x
->pattern
;
648 int to_exclude
= x
->flags
& EXC_FLAG_NEGATIVE
? 0 : 1;
649 int prefix
= x
->nowildcardlen
;
651 if (x
->flags
& EXC_FLAG_MUSTBEDIR
) {
652 if (*dtype
== DT_UNKNOWN
)
653 *dtype
= get_dtype(NULL
, pathname
, pathlen
);
654 if (*dtype
!= DT_DIR
)
658 if (x
->flags
& EXC_FLAG_NODIR
) {
659 if (match_basename(basename
,
660 pathlen
- (basename
- pathname
),
661 exclude
, prefix
, x
->patternlen
,
667 assert(x
->baselen
== 0 || x
->base
[x
->baselen
- 1] == '/');
668 if (match_pathname(pathname
, pathlen
,
669 x
->base
, x
->baselen
? x
->baselen
- 1 : 0,
670 exclude
, prefix
, x
->patternlen
, x
->flags
))
673 return -1; /* undecided */
676 static int excluded(struct dir_struct
*dir
, const char *pathname
, int *dtype_p
)
678 int pathlen
= strlen(pathname
);
680 const char *basename
= strrchr(pathname
, '/');
681 basename
= (basename
) ? basename
+1 : pathname
;
683 prep_exclude(dir
, pathname
, basename
-pathname
);
684 for (st
= EXC_CMDL
; st
<= EXC_FILE
; st
++) {
685 switch (excluded_from_list(pathname
, pathlen
, basename
,
686 dtype_p
, &dir
->exclude_list
[st
])) {
696 void path_exclude_check_init(struct path_exclude_check
*check
,
697 struct dir_struct
*dir
)
700 strbuf_init(&check
->path
, 256);
703 void path_exclude_check_clear(struct path_exclude_check
*check
)
705 strbuf_release(&check
->path
);
709 * Is this name excluded? This is for a caller like show_files() that
710 * do not honor directory hierarchy and iterate through paths that are
711 * possibly in an ignored directory.
713 * A path to a directory known to be excluded is left in check->path to
714 * optimize for repeated checks for files in the same excluded directory.
716 int path_excluded(struct path_exclude_check
*check
,
717 const char *name
, int namelen
, int *dtype
)
720 struct strbuf
*path
= &check
->path
;
723 * we allow the caller to pass namelen as an optimization; it
724 * must match the length of the name, as we eventually call
725 * excluded() on the whole name string.
728 namelen
= strlen(name
);
731 path
->len
<= namelen
&&
732 !memcmp(name
, path
->buf
, path
->len
) &&
733 (!name
[path
->len
] || name
[path
->len
] == '/'))
736 strbuf_setlen(path
, 0);
737 for (i
= 0; name
[i
]; i
++) {
742 if (excluded(check
->dir
, path
->buf
, &dt
))
745 strbuf_addch(path
, ch
);
748 /* An entry in the index; cannot be a directory with subentries */
749 strbuf_setlen(path
, 0);
751 return excluded(check
->dir
, name
, dtype
);
754 static struct dir_entry
*dir_entry_new(const char *pathname
, int len
)
756 struct dir_entry
*ent
;
758 ent
= xmalloc(sizeof(*ent
) + len
+ 1);
760 memcpy(ent
->name
, pathname
, len
);
765 static struct dir_entry
*dir_add_name(struct dir_struct
*dir
, const char *pathname
, int len
)
767 if (cache_name_exists(pathname
, len
, ignore_case
))
770 ALLOC_GROW(dir
->entries
, dir
->nr
+1, dir
->alloc
);
771 return dir
->entries
[dir
->nr
++] = dir_entry_new(pathname
, len
);
774 struct dir_entry
*dir_add_ignored(struct dir_struct
*dir
, const char *pathname
, int len
)
776 if (!cache_name_is_other(pathname
, len
))
779 ALLOC_GROW(dir
->ignored
, dir
->ignored_nr
+1, dir
->ignored_alloc
);
780 return dir
->ignored
[dir
->ignored_nr
++] = dir_entry_new(pathname
, len
);
784 index_nonexistent
= 0,
790 * Do not use the alphabetically stored index to look up
791 * the directory name; instead, use the case insensitive
794 static enum exist_status
directory_exists_in_index_icase(const char *dirname
, int len
)
796 struct cache_entry
*ce
= index_name_exists(&the_index
, dirname
, len
+ 1, ignore_case
);
797 unsigned char endchar
;
800 return index_nonexistent
;
801 endchar
= ce
->name
[len
];
804 * The cache_entry structure returned will contain this dirname
805 * and possibly additional path components.
808 return index_directory
;
811 * If there are no additional path components, then this cache_entry
812 * represents a submodule. Submodules, despite being directories,
813 * are stored in the cache without a closing slash.
815 if (!endchar
&& S_ISGITLINK(ce
->ce_mode
))
818 /* This should never be hit, but it exists just in case. */
819 return index_nonexistent
;
823 * The index sorts alphabetically by entry name, which
824 * means that a gitlink sorts as '\0' at the end, while
825 * a directory (which is defined not as an entry, but as
826 * the files it contains) will sort with the '/' at the
829 static enum exist_status
directory_exists_in_index(const char *dirname
, int len
)
834 return directory_exists_in_index_icase(dirname
, len
);
836 pos
= cache_name_pos(dirname
, len
);
839 while (pos
< active_nr
) {
840 struct cache_entry
*ce
= active_cache
[pos
++];
841 unsigned char endchar
;
843 if (strncmp(ce
->name
, dirname
, len
))
845 endchar
= ce
->name
[len
];
849 return index_directory
;
850 if (!endchar
&& S_ISGITLINK(ce
->ce_mode
))
853 return index_nonexistent
;
857 * When we find a directory when traversing the filesystem, we
858 * have three distinct cases:
861 * - see it as a directory
864 * and which one we choose depends on a combination of existing
865 * git index contents and the flags passed into the directory
868 * Case 1: If we *already* have entries in the index under that
869 * directory name, we always recurse into the directory to see
872 * Case 2: If we *already* have that directory name as a gitlink,
873 * we always continue to see it as a gitlink, regardless of whether
874 * there is an actual git directory there or not (it might not
875 * be checked out as a subproject!)
877 * Case 3: if we didn't have it in the index previously, we
878 * have a few sub-cases:
880 * (a) if "show_other_directories" is true, we show it as
881 * just a directory, unless "hide_empty_directories" is
882 * also true and the directory is empty, in which case
883 * we just ignore it entirely.
884 * (b) if it looks like a git directory, and we don't have
885 * 'no_gitlinks' set we treat it as a gitlink, and show it
887 * (c) otherwise, we recurse into it.
889 enum directory_treatment
{
892 recurse_into_directory
895 static enum directory_treatment
treat_directory(struct dir_struct
*dir
,
896 const char *dirname
, int len
,
897 const struct path_simplify
*simplify
)
899 /* The "len-1" is to strip the final '/' */
900 switch (directory_exists_in_index(dirname
, len
-1)) {
901 case index_directory
:
902 return recurse_into_directory
;
905 if (dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
)
906 return ignore_directory
;
907 return show_directory
;
909 case index_nonexistent
:
910 if (dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
)
912 if (!(dir
->flags
& DIR_NO_GITLINKS
)) {
913 unsigned char sha1
[20];
914 if (resolve_gitlink_ref(dirname
, "HEAD", sha1
) == 0)
915 return show_directory
;
917 return recurse_into_directory
;
920 /* This is the "show_other_directories" case */
921 if (!(dir
->flags
& DIR_HIDE_EMPTY_DIRECTORIES
))
922 return show_directory
;
923 if (!read_directory_recursive(dir
, dirname
, len
, 1, simplify
))
924 return ignore_directory
;
925 return show_directory
;
929 * This is an inexact early pruning of any recursive directory
930 * reading - if the path cannot possibly be in the pathspec,
931 * return true, and we'll skip it early.
933 static int simplify_away(const char *path
, int pathlen
, const struct path_simplify
*simplify
)
937 const char *match
= simplify
->path
;
938 int len
= simplify
->len
;
944 if (!memcmp(path
, match
, len
))
954 * This function tells us whether an excluded path matches a
955 * list of "interesting" pathspecs. That is, whether a path matched
956 * by any of the pathspecs could possibly be ignored by excluding
957 * the specified path. This can happen if:
959 * 1. the path is mentioned explicitly in the pathspec
961 * 2. the path is a directory prefix of some element in the
964 static int exclude_matches_pathspec(const char *path
, int len
,
965 const struct path_simplify
*simplify
)
968 for (; simplify
->path
; simplify
++) {
969 if (len
== simplify
->len
970 && !memcmp(path
, simplify
->path
, len
))
972 if (len
< simplify
->len
973 && simplify
->path
[len
] == '/'
974 && !memcmp(path
, simplify
->path
, len
))
981 static int get_index_dtype(const char *path
, int len
)
984 struct cache_entry
*ce
;
986 ce
= cache_name_exists(path
, len
, 0);
988 if (!ce_uptodate(ce
))
990 if (S_ISGITLINK(ce
->ce_mode
))
993 * Nobody actually cares about the
994 * difference between DT_LNK and DT_REG
999 /* Try to look it up as a directory */
1000 pos
= cache_name_pos(path
, len
);
1004 while (pos
< active_nr
) {
1005 ce
= active_cache
[pos
++];
1006 if (strncmp(ce
->name
, path
, len
))
1008 if (ce
->name
[len
] > '/')
1010 if (ce
->name
[len
] < '/')
1012 if (!ce_uptodate(ce
))
1013 break; /* continue? */
1019 static int get_dtype(struct dirent
*de
, const char *path
, int len
)
1021 int dtype
= de
? DTYPE(de
) : DT_UNKNOWN
;
1024 if (dtype
!= DT_UNKNOWN
)
1026 dtype
= get_index_dtype(path
, len
);
1027 if (dtype
!= DT_UNKNOWN
)
1029 if (lstat(path
, &st
))
1031 if (S_ISREG(st
.st_mode
))
1033 if (S_ISDIR(st
.st_mode
))
1035 if (S_ISLNK(st
.st_mode
))
1040 enum path_treatment
{
1046 static enum path_treatment
treat_one_path(struct dir_struct
*dir
,
1047 struct strbuf
*path
,
1048 const struct path_simplify
*simplify
,
1049 int dtype
, struct dirent
*de
)
1051 int exclude
= excluded(dir
, path
->buf
, &dtype
);
1052 if (exclude
&& (dir
->flags
& DIR_COLLECT_IGNORED
)
1053 && exclude_matches_pathspec(path
->buf
, path
->len
, simplify
))
1054 dir_add_ignored(dir
, path
->buf
, path
->len
);
1057 * Excluded? If we don't explicitly want to show
1058 * ignored files, ignore it
1060 if (exclude
&& !(dir
->flags
& DIR_SHOW_IGNORED
))
1061 return path_ignored
;
1063 if (dtype
== DT_UNKNOWN
)
1064 dtype
= get_dtype(de
, path
->buf
, path
->len
);
1067 * Do we want to see just the ignored files?
1068 * We still need to recurse into directories,
1069 * even if we don't ignore them, since the
1070 * directory may contain files that we do..
1072 if (!exclude
&& (dir
->flags
& DIR_SHOW_IGNORED
)) {
1073 if (dtype
!= DT_DIR
)
1074 return path_ignored
;
1079 return path_ignored
;
1081 strbuf_addch(path
, '/');
1082 switch (treat_directory(dir
, path
->buf
, path
->len
, simplify
)) {
1083 case show_directory
:
1084 if (exclude
!= !!(dir
->flags
1085 & DIR_SHOW_IGNORED
))
1086 return path_ignored
;
1088 case recurse_into_directory
:
1089 return path_recurse
;
1090 case ignore_directory
:
1091 return path_ignored
;
1098 return path_handled
;
1101 static enum path_treatment
treat_path(struct dir_struct
*dir
,
1103 struct strbuf
*path
,
1105 const struct path_simplify
*simplify
)
1109 if (is_dot_or_dotdot(de
->d_name
) || !strcmp(de
->d_name
, ".git"))
1110 return path_ignored
;
1111 strbuf_setlen(path
, baselen
);
1112 strbuf_addstr(path
, de
->d_name
);
1113 if (simplify_away(path
->buf
, path
->len
, simplify
))
1114 return path_ignored
;
1117 return treat_one_path(dir
, path
, simplify
, dtype
, de
);
1121 * Read a directory tree. We currently ignore anything but
1122 * directories, regular files and symlinks. That's because git
1123 * doesn't handle them at all yet. Maybe that will change some
1126 * Also, we ignore the name ".git" (even if it is not a directory).
1127 * That likely will not change.
1129 static int read_directory_recursive(struct dir_struct
*dir
,
1130 const char *base
, int baselen
,
1132 const struct path_simplify
*simplify
)
1137 struct strbuf path
= STRBUF_INIT
;
1139 strbuf_add(&path
, base
, baselen
);
1141 fdir
= opendir(path
.len
? path
.buf
: ".");
1145 while ((de
= readdir(fdir
)) != NULL
) {
1146 switch (treat_path(dir
, de
, &path
, baselen
, simplify
)) {
1148 contents
+= read_directory_recursive(dir
, path
.buf
,
1160 dir_add_name(dir
, path
.buf
, path
.len
);
1164 strbuf_release(&path
);
1169 static int cmp_name(const void *p1
, const void *p2
)
1171 const struct dir_entry
*e1
= *(const struct dir_entry
**)p1
;
1172 const struct dir_entry
*e2
= *(const struct dir_entry
**)p2
;
1174 return cache_name_compare(e1
->name
, e1
->len
,
1178 static struct path_simplify
*create_simplify(const char **pathspec
)
1181 struct path_simplify
*simplify
= NULL
;
1186 for (nr
= 0 ; ; nr
++) {
1189 alloc
= alloc_nr(alloc
);
1190 simplify
= xrealloc(simplify
, alloc
* sizeof(*simplify
));
1192 match
= *pathspec
++;
1195 simplify
[nr
].path
= match
;
1196 simplify
[nr
].len
= simple_length(match
);
1198 simplify
[nr
].path
= NULL
;
1199 simplify
[nr
].len
= 0;
1203 static void free_simplify(struct path_simplify
*simplify
)
1208 static int treat_leading_path(struct dir_struct
*dir
,
1209 const char *path
, int len
,
1210 const struct path_simplify
*simplify
)
1212 struct strbuf sb
= STRBUF_INIT
;
1213 int baselen
, rc
= 0;
1216 while (len
&& path
[len
- 1] == '/')
1222 cp
= path
+ baselen
+ !!baselen
;
1223 cp
= memchr(cp
, '/', path
+ len
- cp
);
1227 baselen
= cp
- path
;
1228 strbuf_setlen(&sb
, 0);
1229 strbuf_add(&sb
, path
, baselen
);
1230 if (!is_directory(sb
.buf
))
1232 if (simplify_away(sb
.buf
, sb
.len
, simplify
))
1234 if (treat_one_path(dir
, &sb
, simplify
,
1235 DT_DIR
, NULL
) == path_ignored
)
1236 break; /* do not recurse into it */
1237 if (len
<= baselen
) {
1239 break; /* finished checking */
1242 strbuf_release(&sb
);
1246 int read_directory(struct dir_struct
*dir
, const char *path
, int len
, const char **pathspec
)
1248 struct path_simplify
*simplify
;
1250 if (has_symlink_leading_path(path
, len
))
1253 simplify
= create_simplify(pathspec
);
1254 if (!len
|| treat_leading_path(dir
, path
, len
, simplify
))
1255 read_directory_recursive(dir
, path
, len
, 0, simplify
);
1256 free_simplify(simplify
);
1257 qsort(dir
->entries
, dir
->nr
, sizeof(struct dir_entry
*), cmp_name
);
1258 qsort(dir
->ignored
, dir
->ignored_nr
, sizeof(struct dir_entry
*), cmp_name
);
1262 int file_exists(const char *f
)
1265 return lstat(f
, &sb
) == 0;
1269 * Given two normalized paths (a trailing slash is ok), if subdir is
1270 * outside dir, return -1. Otherwise return the offset in subdir that
1271 * can be used as relative path to dir.
1273 int dir_inside_of(const char *subdir
, const char *dir
)
1277 assert(dir
&& subdir
&& *dir
&& *subdir
);
1279 while (*dir
&& *subdir
&& *dir
== *subdir
) {
1285 /* hel[p]/me vs hel[l]/yeah */
1286 if (*dir
&& *subdir
)
1290 return !*dir
? offset
: -1; /* same dir */
1292 /* foo/[b]ar vs foo/[] */
1293 if (is_dir_sep(dir
[-1]))
1294 return is_dir_sep(subdir
[-1]) ? offset
: -1;
1296 /* foo[/]bar vs foo[] */
1297 return is_dir_sep(*subdir
) ? offset
+ 1 : -1;
1300 int is_inside_dir(const char *dir
)
1305 if (!getcwd(cwd
, sizeof(cwd
)))
1306 die_errno("can't find the current directory");
1307 return dir_inside_of(cwd
, dir
) >= 0;
1310 int is_empty_dir(const char *path
)
1312 DIR *dir
= opendir(path
);
1319 while ((e
= readdir(dir
)) != NULL
)
1320 if (!is_dot_or_dotdot(e
->d_name
)) {
1329 static int remove_dir_recurse(struct strbuf
*path
, int flag
, int *kept_up
)
1333 int ret
= 0, original_len
= path
->len
, len
, kept_down
= 0;
1334 int only_empty
= (flag
& REMOVE_DIR_EMPTY_ONLY
);
1335 int keep_toplevel
= (flag
& REMOVE_DIR_KEEP_TOPLEVEL
);
1336 unsigned char submodule_head
[20];
1338 if ((flag
& REMOVE_DIR_KEEP_NESTED_GIT
) &&
1339 !resolve_gitlink_ref(path
->buf
, "HEAD", submodule_head
)) {
1340 /* Do not descend and nuke a nested git work tree. */
1346 flag
&= ~REMOVE_DIR_KEEP_TOPLEVEL
;
1347 dir
= opendir(path
->buf
);
1349 /* an empty dir could be removed even if it is unreadble */
1351 return rmdir(path
->buf
);
1355 if (path
->buf
[original_len
- 1] != '/')
1356 strbuf_addch(path
, '/');
1359 while ((e
= readdir(dir
)) != NULL
) {
1361 if (is_dot_or_dotdot(e
->d_name
))
1364 strbuf_setlen(path
, len
);
1365 strbuf_addstr(path
, e
->d_name
);
1366 if (lstat(path
->buf
, &st
))
1368 else if (S_ISDIR(st
.st_mode
)) {
1369 if (!remove_dir_recurse(path
, flag
, &kept_down
))
1370 continue; /* happy */
1371 } else if (!only_empty
&& !unlink(path
->buf
))
1372 continue; /* happy, too */
1374 /* path too long, stat fails, or non-directory still exists */
1380 strbuf_setlen(path
, original_len
);
1381 if (!ret
&& !keep_toplevel
&& !kept_down
)
1382 ret
= rmdir(path
->buf
);
1385 * report the uplevel that it is not an error that we
1386 * did not rmdir() our directory.
1392 int remove_dir_recursively(struct strbuf
*path
, int flag
)
1394 return remove_dir_recurse(path
, flag
, NULL
);
1397 void setup_standard_excludes(struct dir_struct
*dir
)
1402 dir
->exclude_per_dir
= ".gitignore";
1403 path
= git_path("info/exclude");
1404 if (!excludes_file
) {
1405 home_config_paths(NULL
, &xdg_path
, "ignore");
1406 excludes_file
= xdg_path
;
1408 if (!access_or_warn(path
, R_OK
))
1409 add_excludes_from_file(dir
, path
);
1410 if (excludes_file
&& !access_or_warn(excludes_file
, R_OK
))
1411 add_excludes_from_file(dir
, excludes_file
);
1414 int remove_path(const char *name
)
1418 if (unlink(name
) && errno
!= ENOENT
)
1421 slash
= strrchr(name
, '/');
1423 char *dirs
= xstrdup(name
);
1424 slash
= dirs
+ (slash
- name
);
1427 } while (rmdir(dirs
) == 0 && (slash
= strrchr(dirs
, '/')));
1433 static int pathspec_item_cmp(const void *a_
, const void *b_
)
1435 struct pathspec_item
*a
, *b
;
1437 a
= (struct pathspec_item
*)a_
;
1438 b
= (struct pathspec_item
*)b_
;
1439 return strcmp(a
->match
, b
->match
);
1442 int init_pathspec(struct pathspec
*pathspec
, const char **paths
)
1444 const char **p
= paths
;
1447 memset(pathspec
, 0, sizeof(*pathspec
));
1452 pathspec
->raw
= paths
;
1453 pathspec
->nr
= p
- paths
;
1457 pathspec
->items
= xmalloc(sizeof(struct pathspec_item
)*pathspec
->nr
);
1458 for (i
= 0; i
< pathspec
->nr
; i
++) {
1459 struct pathspec_item
*item
= pathspec
->items
+i
;
1460 const char *path
= paths
[i
];
1463 item
->len
= strlen(path
);
1464 item
->use_wildcard
= !no_wildcard(path
);
1465 if (item
->use_wildcard
)
1466 pathspec
->has_wildcard
= 1;
1469 qsort(pathspec
->items
, pathspec
->nr
,
1470 sizeof(struct pathspec_item
), pathspec_item_cmp
);
1475 void free_pathspec(struct pathspec
*pathspec
)
1477 free(pathspec
->items
);
1478 pathspec
->items
= NULL
;