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
11 #include "wildmatch.h"
13 struct path_simplify
{
18 static int read_directory_recursive(struct dir_struct
*dir
, const char *path
, int len
,
19 int check_only
, const struct path_simplify
*simplify
);
20 static int get_dtype(struct dirent
*de
, const char *path
, int len
);
22 /* helper string functions with support for the ignore_case flag */
23 int strcmp_icase(const char *a
, const char *b
)
25 return ignore_case
? strcasecmp(a
, b
) : strcmp(a
, b
);
28 int strncmp_icase(const char *a
, const char *b
, size_t count
)
30 return ignore_case
? strncasecmp(a
, b
, count
) : strncmp(a
, b
, count
);
33 int fnmatch_icase(const char *pattern
, const char *string
, int flags
)
35 return fnmatch(pattern
, string
, flags
| (ignore_case
? FNM_CASEFOLD
: 0));
38 inline int git_fnmatch(const char *pattern
, const char *string
,
39 int flags
, int prefix
)
42 if (flags
& GFNM_PATHNAME
)
43 fnm_flags
|= FNM_PATHNAME
;
45 if (strncmp(pattern
, string
, prefix
))
50 if (flags
& GFNM_ONESTAR
) {
51 int pattern_len
= strlen(++pattern
);
52 int string_len
= strlen(string
);
53 return string_len
< pattern_len
||
55 string
+ string_len
- pattern_len
);
57 return fnmatch(pattern
, string
, fnm_flags
);
60 static size_t common_prefix_len(const char **pathspec
)
62 const char *n
, *first
;
64 int literal
= limit_pathspec_to_literal();
70 while ((n
= *pathspec
++)) {
72 for (i
= 0; first
== n
|| i
< max
; i
++) {
74 if (!c
|| c
!= first
[i
] || (!literal
&& is_glob_special(c
)))
79 if (first
== n
|| len
< max
) {
89 * Returns a copy of the longest leading path common among all
92 char *common_prefix(const char **pathspec
)
94 unsigned long len
= common_prefix_len(pathspec
);
96 return len
? xmemdupz(*pathspec
, len
) : NULL
;
99 int fill_directory(struct dir_struct
*dir
, const char **pathspec
)
104 * Calculate common prefix for the pathspec, and
105 * use that to optimize the directory walk
107 len
= common_prefix_len(pathspec
);
109 /* Read the directory and prune it */
110 read_directory(dir
, pathspec
? *pathspec
: "", len
, pathspec
);
114 int within_depth(const char *name
, int namelen
,
115 int depth
, int max_depth
)
117 const char *cp
= name
, *cpe
= name
+ namelen
;
123 if (depth
> max_depth
)
130 * Does 'match' match the given name?
131 * A match is found if
133 * (1) the 'match' string is leading directory of 'name', or
134 * (2) the 'match' string is a wildcard and matches 'name', or
135 * (3) the 'match' string is exactly the same as 'name'.
137 * and the return value tells which case it was.
139 * It returns 0 when there is no match.
141 static int match_one(const char *match
, const char *name
, int namelen
)
144 int literal
= limit_pathspec_to_literal();
146 /* If the match was just the prefix, we matched */
148 return MATCHED_RECURSIVELY
;
152 unsigned char c1
= tolower(*match
);
153 unsigned char c2
= tolower(*name
);
154 if (c1
== '\0' || (!literal
&& is_glob_special(c1
)))
164 unsigned char c1
= *match
;
165 unsigned char c2
= *name
;
166 if (c1
== '\0' || (!literal
&& is_glob_special(c1
)))
177 * If we don't match the matchstring exactly,
178 * we need to match by fnmatch
180 matchlen
= strlen(match
);
181 if (strncmp_icase(match
, name
, matchlen
)) {
184 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
->nowildcard_len
< item
->len
&&
261 !git_fnmatch(match
, name
,
262 item
->flags
& PATHSPEC_ONESTAR
? GFNM_ONESTAR
: 0,
263 item
->nowildcard_len
- prefix
))
264 return MATCHED_FNMATCH
;
270 * Given a name and a list of pathspecs, see if the name matches
271 * any of the pathspecs. The caller is also interested in seeing
272 * all pathspec matches some names it calls this function with
273 * (otherwise the user could have mistyped the unmatched pathspec),
274 * and a mark is left in seen[] array for pathspec element that
275 * actually matched anything.
277 int match_pathspec_depth(const struct pathspec
*ps
,
278 const char *name
, int namelen
,
279 int prefix
, char *seen
)
284 if (!ps
->recursive
|| ps
->max_depth
== -1)
285 return MATCHED_RECURSIVELY
;
287 if (within_depth(name
, namelen
, 0, ps
->max_depth
))
288 return MATCHED_EXACTLY
;
296 for (i
= ps
->nr
- 1; i
>= 0; i
--) {
298 if (seen
&& seen
[i
] == MATCHED_EXACTLY
)
300 how
= match_pathspec_item(ps
->items
+i
, prefix
, name
, namelen
);
301 if (ps
->recursive
&& ps
->max_depth
!= -1 &&
302 how
&& how
!= MATCHED_FNMATCH
) {
303 int len
= ps
->items
[i
].len
;
304 if (name
[len
] == '/')
306 if (within_depth(name
+len
, namelen
-len
, 0, ps
->max_depth
))
307 how
= MATCHED_EXACTLY
;
314 if (seen
&& seen
[i
] < how
)
322 * Return the length of the "simple" part of a path match limiter.
324 static int simple_length(const char *match
)
329 unsigned char c
= *match
++;
331 if (c
== '\0' || is_glob_special(c
))
336 static int no_wildcard(const char *string
)
338 return string
[simple_length(string
)] == '\0';
341 void parse_exclude_pattern(const char **pattern
,
346 const char *p
= *pattern
;
351 *flags
|= EXC_FLAG_NEGATIVE
;
355 if (len
&& p
[len
- 1] == '/') {
357 *flags
|= EXC_FLAG_MUSTBEDIR
;
359 for (i
= 0; i
< len
; i
++) {
364 *flags
|= EXC_FLAG_NODIR
;
365 *nowildcardlen
= simple_length(p
);
367 * we should have excluded the trailing slash from 'p' too,
368 * but that's one more allocation. Instead just make sure
369 * nowildcardlen does not exceed real patternlen
371 if (*nowildcardlen
> len
)
372 *nowildcardlen
= len
;
373 if (*p
== '*' && no_wildcard(p
+ 1))
374 *flags
|= EXC_FLAG_ENDSWITH
;
379 void add_exclude(const char *string
, const char *base
,
380 int baselen
, struct exclude_list
*which
)
387 parse_exclude_pattern(&string
, &patternlen
, &flags
, &nowildcardlen
);
388 if (flags
& EXC_FLAG_MUSTBEDIR
) {
390 x
= xmalloc(sizeof(*x
) + patternlen
+ 1);
392 memcpy(s
, string
, patternlen
);
393 s
[patternlen
] = '\0';
396 x
= xmalloc(sizeof(*x
));
399 x
->patternlen
= patternlen
;
400 x
->nowildcardlen
= nowildcardlen
;
402 x
->baselen
= baselen
;
404 ALLOC_GROW(which
->excludes
, which
->nr
+ 1, which
->alloc
);
405 which
->excludes
[which
->nr
++] = x
;
408 static void *read_skip_worktree_file_from_index(const char *path
, size_t *size
)
412 enum object_type type
;
414 struct index_state
*istate
= &the_index
;
417 pos
= index_name_pos(istate
, path
, len
);
420 if (!ce_skip_worktree(istate
->cache
[pos
]))
422 data
= read_sha1_file(istate
->cache
[pos
]->sha1
, &type
, &sz
);
423 if (!data
|| type
!= OBJ_BLOB
) {
431 void free_excludes(struct exclude_list
*el
)
435 for (i
= 0; i
< el
->nr
; i
++)
436 free(el
->excludes
[i
]);
443 int add_excludes_from_file_to_list(const char *fname
,
447 struct exclude_list
*which
,
455 fd
= open(fname
, O_RDONLY
);
456 if (fd
< 0 || fstat(fd
, &st
) < 0) {
458 warn_on_inaccessible(fname
);
462 (buf
= read_skip_worktree_file_from_index(fname
, &size
)) == NULL
)
468 if (buf
[size
-1] != '\n') {
469 buf
= xrealloc(buf
, size
+1);
474 size
= xsize_t(st
.st_size
);
479 buf
= xmalloc(size
+1);
480 if (read_in_full(fd
, buf
, size
) != size
) {
492 for (i
= 0; i
< size
; i
++) {
493 if (buf
[i
] == '\n') {
494 if (entry
!= buf
+ i
&& entry
[0] != '#') {
495 buf
[i
- (i
&& buf
[i
-1] == '\r')] = 0;
496 add_exclude(entry
, base
, baselen
, which
);
504 void add_excludes_from_file(struct dir_struct
*dir
, const char *fname
)
506 if (add_excludes_from_file_to_list(fname
, "", 0, NULL
,
507 &dir
->exclude_list
[EXC_FILE
], 0) < 0)
508 die("cannot use %s as an exclude file", fname
);
511 static void prep_exclude(struct dir_struct
*dir
, const char *base
, int baselen
)
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 /* Pop the ones that are not the prefix of the path being checked. */
522 el
= &dir
->exclude_list
[EXC_DIRS
];
523 while ((stk
= dir
->exclude_stack
) != NULL
) {
524 if (stk
->baselen
<= baselen
&&
525 !strncmp(dir
->basebuf
, base
, stk
->baselen
))
527 dir
->exclude_stack
= stk
->prev
;
528 while (stk
->exclude_ix
< el
->nr
)
529 free(el
->excludes
[--el
->nr
]);
534 /* Read from the parent directories and push them down. */
535 current
= stk
? stk
->baselen
: -1;
536 while (current
< baselen
) {
537 struct exclude_stack
*stk
= xcalloc(1, sizeof(*stk
));
545 cp
= strchr(base
+ current
+ 1, '/');
547 die("oops in prep_exclude");
550 stk
->prev
= dir
->exclude_stack
;
551 stk
->baselen
= cp
- base
;
552 stk
->exclude_ix
= el
->nr
;
553 memcpy(dir
->basebuf
+ current
, base
+ current
,
554 stk
->baselen
- current
);
555 strcpy(dir
->basebuf
+ stk
->baselen
, dir
->exclude_per_dir
);
556 add_excludes_from_file_to_list(dir
->basebuf
,
557 dir
->basebuf
, stk
->baselen
,
558 &stk
->filebuf
, el
, 1);
559 dir
->exclude_stack
= stk
;
560 current
= stk
->baselen
;
562 dir
->basebuf
[baselen
] = '\0';
565 int match_basename(const char *basename
, int basenamelen
,
566 const char *pattern
, int prefix
, int patternlen
,
569 if (prefix
== patternlen
) {
570 if (!strcmp_icase(pattern
, basename
))
572 } else if (flags
& EXC_FLAG_ENDSWITH
) {
573 if (patternlen
- 1 <= basenamelen
&&
574 !strcmp_icase(pattern
+ 1,
575 basename
+ basenamelen
- patternlen
+ 1))
578 if (fnmatch_icase(pattern
, basename
, 0) == 0)
584 int match_pathname(const char *pathname
, int pathlen
,
585 const char *base
, int baselen
,
586 const char *pattern
, int prefix
, int patternlen
,
593 * match with FNM_PATHNAME; the pattern has base implicitly
596 if (*pattern
== '/') {
602 * baselen does not count the trailing slash. base[] may or
603 * may not end with a trailing slash though.
605 if (pathlen
< baselen
+ 1 ||
606 (baselen
&& pathname
[baselen
] != '/') ||
607 strncmp_icase(pathname
, base
, baselen
))
610 namelen
= baselen
? pathlen
- baselen
- 1 : pathlen
;
611 name
= pathname
+ pathlen
- namelen
;
615 * if the non-wildcard part is longer than the
616 * remaining pathname, surely it cannot match.
618 if (prefix
> namelen
)
621 if (strncmp_icase(pattern
, name
, prefix
))
628 return wildmatch(pattern
, name
,
629 WM_PATHNAME
| (ignore_case
? WM_CASEFOLD
: 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 recurse into the directory to see all the files,
870 * unless the directory is excluded and we want to show ignored
873 * Case 2: If we *already* have that directory name as a gitlink,
874 * we always continue to see it as a gitlink, regardless of whether
875 * there is an actual git directory there or not (it might not
876 * be checked out as a subproject!)
878 * Case 3: if we didn't have it in the index previously, we
879 * have a few sub-cases:
881 * (a) if "show_other_directories" is true, we show it as
882 * just a directory, unless "hide_empty_directories" is
883 * also true and the directory is empty, in which case
884 * we just ignore it entirely.
885 * if we are looking for ignored directories, look if it
886 * contains only ignored files to decide if it must be shown as
888 * (b) if it looks like a git directory, and we don't have
889 * 'no_gitlinks' set we treat it as a gitlink, and show it
891 * (c) otherwise, we recurse into it.
893 enum directory_treatment
{
896 recurse_into_directory
899 static enum directory_treatment
treat_directory(struct dir_struct
*dir
,
900 const char *dirname
, int len
, int exclude
,
901 const struct path_simplify
*simplify
)
903 /* The "len-1" is to strip the final '/' */
904 switch (directory_exists_in_index(dirname
, len
-1)) {
905 case index_directory
:
906 if ((dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
) && exclude
)
909 return recurse_into_directory
;
912 if (dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
)
913 return ignore_directory
;
914 return show_directory
;
916 case index_nonexistent
:
917 if (dir
->flags
& DIR_SHOW_OTHER_DIRECTORIES
)
919 if (!(dir
->flags
& DIR_NO_GITLINKS
)) {
920 unsigned char sha1
[20];
921 if (resolve_gitlink_ref(dirname
, "HEAD", sha1
) == 0)
922 return show_directory
;
924 return recurse_into_directory
;
927 /* This is the "show_other_directories" case */
930 * We are looking for ignored files and our directory is not ignored,
931 * check if it contains only ignored files
933 if ((dir
->flags
& DIR_SHOW_IGNORED
) && !exclude
) {
935 dir
->flags
&= ~DIR_SHOW_IGNORED
;
936 dir
->flags
|= DIR_HIDE_EMPTY_DIRECTORIES
;
937 ignored
= read_directory_recursive(dir
, dirname
, len
, 1, simplify
);
938 dir
->flags
&= ~DIR_HIDE_EMPTY_DIRECTORIES
;
939 dir
->flags
|= DIR_SHOW_IGNORED
;
941 return ignored
? ignore_directory
: show_directory
;
943 if (!(dir
->flags
& DIR_SHOW_IGNORED
) &&
944 !(dir
->flags
& DIR_HIDE_EMPTY_DIRECTORIES
))
945 return show_directory
;
946 if (!read_directory_recursive(dir
, dirname
, len
, 1, simplify
))
947 return ignore_directory
;
948 return show_directory
;
952 * Decide what to do when we find a file while traversing the
953 * filesystem. Mostly two cases:
955 * 1. We are looking for ignored files
956 * (a) File is ignored, include it
957 * (b) File is in ignored path, include it
958 * (c) File is not ignored, exclude it
960 * 2. Other scenarios, include the file if not excluded
962 * Return 1 for exclude, 0 for include.
964 static int treat_file(struct dir_struct
*dir
, struct strbuf
*path
, int exclude
, int *dtype
)
966 struct path_exclude_check check
;
967 int exclude_file
= 0;
970 exclude_file
= !(dir
->flags
& DIR_SHOW_IGNORED
);
971 else if (dir
->flags
& DIR_SHOW_IGNORED
) {
974 * Don't spend time on indexed files, they won't be
975 * added to the list anyway
977 struct cache_entry
*ce
= index_name_exists(&the_index
,
978 path
->buf
, path
->len
, ignore_case
);
983 path_exclude_check_init(&check
, dir
);
985 if (!path_excluded(&check
, path
->buf
, path
->len
, dtype
))
988 path_exclude_check_clear(&check
);
995 * This is an inexact early pruning of any recursive directory
996 * reading - if the path cannot possibly be in the pathspec,
997 * return true, and we'll skip it early.
999 static int simplify_away(const char *path
, int pathlen
, const struct path_simplify
*simplify
)
1003 const char *match
= simplify
->path
;
1004 int len
= simplify
->len
;
1010 if (!memcmp(path
, match
, len
))
1020 * This function tells us whether an excluded path matches a
1021 * list of "interesting" pathspecs. That is, whether a path matched
1022 * by any of the pathspecs could possibly be ignored by excluding
1023 * the specified path. This can happen if:
1025 * 1. the path is mentioned explicitly in the pathspec
1027 * 2. the path is a directory prefix of some element in the
1030 static int exclude_matches_pathspec(const char *path
, int len
,
1031 const struct path_simplify
*simplify
)
1034 for (; simplify
->path
; simplify
++) {
1035 if (len
== simplify
->len
1036 && !memcmp(path
, simplify
->path
, len
))
1038 if (len
< simplify
->len
1039 && simplify
->path
[len
] == '/'
1040 && !memcmp(path
, simplify
->path
, len
))
1047 static int get_index_dtype(const char *path
, int len
)
1050 struct cache_entry
*ce
;
1052 ce
= cache_name_exists(path
, len
, 0);
1054 if (!ce_uptodate(ce
))
1056 if (S_ISGITLINK(ce
->ce_mode
))
1059 * Nobody actually cares about the
1060 * difference between DT_LNK and DT_REG
1065 /* Try to look it up as a directory */
1066 pos
= cache_name_pos(path
, len
);
1070 while (pos
< active_nr
) {
1071 ce
= active_cache
[pos
++];
1072 if (strncmp(ce
->name
, path
, len
))
1074 if (ce
->name
[len
] > '/')
1076 if (ce
->name
[len
] < '/')
1078 if (!ce_uptodate(ce
))
1079 break; /* continue? */
1085 static int get_dtype(struct dirent
*de
, const char *path
, int len
)
1087 int dtype
= de
? DTYPE(de
) : DT_UNKNOWN
;
1090 if (dtype
!= DT_UNKNOWN
)
1092 dtype
= get_index_dtype(path
, len
);
1093 if (dtype
!= DT_UNKNOWN
)
1095 if (lstat(path
, &st
))
1097 if (S_ISREG(st
.st_mode
))
1099 if (S_ISDIR(st
.st_mode
))
1101 if (S_ISLNK(st
.st_mode
))
1106 enum path_treatment
{
1112 static enum path_treatment
treat_one_path(struct dir_struct
*dir
,
1113 struct strbuf
*path
,
1114 const struct path_simplify
*simplify
,
1115 int dtype
, struct dirent
*de
)
1117 int exclude
= excluded(dir
, path
->buf
, &dtype
);
1118 if (exclude
&& (dir
->flags
& DIR_COLLECT_IGNORED
)
1119 && exclude_matches_pathspec(path
->buf
, path
->len
, simplify
))
1120 dir_add_ignored(dir
, path
->buf
, path
->len
);
1123 * Excluded? If we don't explicitly want to show
1124 * ignored files, ignore it
1126 if (exclude
&& !(dir
->flags
& DIR_SHOW_IGNORED
))
1127 return path_ignored
;
1129 if (dtype
== DT_UNKNOWN
)
1130 dtype
= get_dtype(de
, path
->buf
, path
->len
);
1134 return path_ignored
;
1136 strbuf_addch(path
, '/');
1138 switch (treat_directory(dir
, path
->buf
, path
->len
, exclude
, simplify
)) {
1139 case show_directory
:
1141 case recurse_into_directory
:
1142 return path_recurse
;
1143 case ignore_directory
:
1144 return path_ignored
;
1149 switch (treat_file(dir
, path
, exclude
, &dtype
)) {
1151 return path_ignored
;
1156 return path_handled
;
1159 static enum path_treatment
treat_path(struct dir_struct
*dir
,
1161 struct strbuf
*path
,
1163 const struct path_simplify
*simplify
)
1167 if (is_dot_or_dotdot(de
->d_name
) || !strcmp(de
->d_name
, ".git"))
1168 return path_ignored
;
1169 strbuf_setlen(path
, baselen
);
1170 strbuf_addstr(path
, de
->d_name
);
1171 if (simplify_away(path
->buf
, path
->len
, simplify
))
1172 return path_ignored
;
1175 return treat_one_path(dir
, path
, simplify
, dtype
, de
);
1179 * Read a directory tree. We currently ignore anything but
1180 * directories, regular files and symlinks. That's because git
1181 * doesn't handle them at all yet. Maybe that will change some
1184 * Also, we ignore the name ".git" (even if it is not a directory).
1185 * That likely will not change.
1187 static int read_directory_recursive(struct dir_struct
*dir
,
1188 const char *base
, int baselen
,
1190 const struct path_simplify
*simplify
)
1195 struct strbuf path
= STRBUF_INIT
;
1197 strbuf_add(&path
, base
, baselen
);
1199 fdir
= opendir(path
.len
? path
.buf
: ".");
1203 while ((de
= readdir(fdir
)) != NULL
) {
1204 switch (treat_path(dir
, de
, &path
, baselen
, simplify
)) {
1206 contents
+= read_directory_recursive(dir
, path
.buf
,
1218 dir_add_name(dir
, path
.buf
, path
.len
);
1222 strbuf_release(&path
);
1227 static int cmp_name(const void *p1
, const void *p2
)
1229 const struct dir_entry
*e1
= *(const struct dir_entry
**)p1
;
1230 const struct dir_entry
*e2
= *(const struct dir_entry
**)p2
;
1232 return cache_name_compare(e1
->name
, e1
->len
,
1236 static struct path_simplify
*create_simplify(const char **pathspec
)
1239 struct path_simplify
*simplify
= NULL
;
1244 for (nr
= 0 ; ; nr
++) {
1247 alloc
= alloc_nr(alloc
);
1248 simplify
= xrealloc(simplify
, alloc
* sizeof(*simplify
));
1250 match
= *pathspec
++;
1253 simplify
[nr
].path
= match
;
1254 simplify
[nr
].len
= simple_length(match
);
1256 simplify
[nr
].path
= NULL
;
1257 simplify
[nr
].len
= 0;
1261 static void free_simplify(struct path_simplify
*simplify
)
1266 static int treat_leading_path(struct dir_struct
*dir
,
1267 const char *path
, int len
,
1268 const struct path_simplify
*simplify
)
1270 struct strbuf sb
= STRBUF_INIT
;
1271 int baselen
, rc
= 0;
1274 while (len
&& path
[len
- 1] == '/')
1280 cp
= path
+ baselen
+ !!baselen
;
1281 cp
= memchr(cp
, '/', path
+ len
- cp
);
1285 baselen
= cp
- path
;
1286 strbuf_setlen(&sb
, 0);
1287 strbuf_add(&sb
, path
, baselen
);
1288 if (!is_directory(sb
.buf
))
1290 if (simplify_away(sb
.buf
, sb
.len
, simplify
))
1292 if (treat_one_path(dir
, &sb
, simplify
,
1293 DT_DIR
, NULL
) == path_ignored
)
1294 break; /* do not recurse into it */
1295 if (len
<= baselen
) {
1297 break; /* finished checking */
1300 strbuf_release(&sb
);
1304 int read_directory(struct dir_struct
*dir
, const char *path
, int len
, const char **pathspec
)
1306 struct path_simplify
*simplify
;
1308 if (has_symlink_leading_path(path
, len
))
1311 simplify
= create_simplify(pathspec
);
1312 if (!len
|| treat_leading_path(dir
, path
, len
, simplify
))
1313 read_directory_recursive(dir
, path
, len
, 0, simplify
);
1314 free_simplify(simplify
);
1315 qsort(dir
->entries
, dir
->nr
, sizeof(struct dir_entry
*), cmp_name
);
1316 qsort(dir
->ignored
, dir
->ignored_nr
, sizeof(struct dir_entry
*), cmp_name
);
1320 int file_exists(const char *f
)
1323 return lstat(f
, &sb
) == 0;
1327 * Given two normalized paths (a trailing slash is ok), if subdir is
1328 * outside dir, return -1. Otherwise return the offset in subdir that
1329 * can be used as relative path to dir.
1331 int dir_inside_of(const char *subdir
, const char *dir
)
1335 assert(dir
&& subdir
&& *dir
&& *subdir
);
1337 while (*dir
&& *subdir
&& *dir
== *subdir
) {
1343 /* hel[p]/me vs hel[l]/yeah */
1344 if (*dir
&& *subdir
)
1348 return !*dir
? offset
: -1; /* same dir */
1350 /* foo/[b]ar vs foo/[] */
1351 if (is_dir_sep(dir
[-1]))
1352 return is_dir_sep(subdir
[-1]) ? offset
: -1;
1354 /* foo[/]bar vs foo[] */
1355 return is_dir_sep(*subdir
) ? offset
+ 1 : -1;
1358 int is_inside_dir(const char *dir
)
1363 if (!getcwd(cwd
, sizeof(cwd
)))
1364 die_errno("can't find the current directory");
1365 return dir_inside_of(cwd
, dir
) >= 0;
1368 int is_empty_dir(const char *path
)
1370 DIR *dir
= opendir(path
);
1377 while ((e
= readdir(dir
)) != NULL
)
1378 if (!is_dot_or_dotdot(e
->d_name
)) {
1387 static int remove_dir_recurse(struct strbuf
*path
, int flag
, int *kept_up
)
1391 int ret
= 0, original_len
= path
->len
, len
, kept_down
= 0;
1392 int only_empty
= (flag
& REMOVE_DIR_EMPTY_ONLY
);
1393 int keep_toplevel
= (flag
& REMOVE_DIR_KEEP_TOPLEVEL
);
1394 unsigned char submodule_head
[20];
1396 if ((flag
& REMOVE_DIR_KEEP_NESTED_GIT
) &&
1397 !resolve_gitlink_ref(path
->buf
, "HEAD", submodule_head
)) {
1398 /* Do not descend and nuke a nested git work tree. */
1404 flag
&= ~REMOVE_DIR_KEEP_TOPLEVEL
;
1405 dir
= opendir(path
->buf
);
1407 /* an empty dir could be removed even if it is unreadble */
1409 return rmdir(path
->buf
);
1413 if (path
->buf
[original_len
- 1] != '/')
1414 strbuf_addch(path
, '/');
1417 while ((e
= readdir(dir
)) != NULL
) {
1419 if (is_dot_or_dotdot(e
->d_name
))
1422 strbuf_setlen(path
, len
);
1423 strbuf_addstr(path
, e
->d_name
);
1424 if (lstat(path
->buf
, &st
))
1426 else if (S_ISDIR(st
.st_mode
)) {
1427 if (!remove_dir_recurse(path
, flag
, &kept_down
))
1428 continue; /* happy */
1429 } else if (!only_empty
&& !unlink(path
->buf
))
1430 continue; /* happy, too */
1432 /* path too long, stat fails, or non-directory still exists */
1438 strbuf_setlen(path
, original_len
);
1439 if (!ret
&& !keep_toplevel
&& !kept_down
)
1440 ret
= rmdir(path
->buf
);
1443 * report the uplevel that it is not an error that we
1444 * did not rmdir() our directory.
1450 int remove_dir_recursively(struct strbuf
*path
, int flag
)
1452 return remove_dir_recurse(path
, flag
, NULL
);
1455 void setup_standard_excludes(struct dir_struct
*dir
)
1460 dir
->exclude_per_dir
= ".gitignore";
1461 path
= git_path("info/exclude");
1462 if (!excludes_file
) {
1463 home_config_paths(NULL
, &xdg_path
, "ignore");
1464 excludes_file
= xdg_path
;
1466 if (!access_or_warn(path
, R_OK
))
1467 add_excludes_from_file(dir
, path
);
1468 if (excludes_file
&& !access_or_warn(excludes_file
, R_OK
))
1469 add_excludes_from_file(dir
, excludes_file
);
1472 int remove_path(const char *name
)
1476 if (unlink(name
) && errno
!= ENOENT
)
1479 slash
= strrchr(name
, '/');
1481 char *dirs
= xstrdup(name
);
1482 slash
= dirs
+ (slash
- name
);
1485 } while (rmdir(dirs
) == 0 && (slash
= strrchr(dirs
, '/')));
1491 static int pathspec_item_cmp(const void *a_
, const void *b_
)
1493 struct pathspec_item
*a
, *b
;
1495 a
= (struct pathspec_item
*)a_
;
1496 b
= (struct pathspec_item
*)b_
;
1497 return strcmp(a
->match
, b
->match
);
1500 int init_pathspec(struct pathspec
*pathspec
, const char **paths
)
1502 const char **p
= paths
;
1505 memset(pathspec
, 0, sizeof(*pathspec
));
1510 pathspec
->raw
= paths
;
1511 pathspec
->nr
= p
- paths
;
1515 pathspec
->items
= xmalloc(sizeof(struct pathspec_item
)*pathspec
->nr
);
1516 for (i
= 0; i
< pathspec
->nr
; i
++) {
1517 struct pathspec_item
*item
= pathspec
->items
+i
;
1518 const char *path
= paths
[i
];
1521 item
->len
= strlen(path
);
1523 if (limit_pathspec_to_literal()) {
1524 item
->nowildcard_len
= item
->len
;
1526 item
->nowildcard_len
= simple_length(path
);
1527 if (item
->nowildcard_len
< item
->len
) {
1528 pathspec
->has_wildcard
= 1;
1529 if (path
[item
->nowildcard_len
] == '*' &&
1530 no_wildcard(path
+ item
->nowildcard_len
+ 1))
1531 item
->flags
|= PATHSPEC_ONESTAR
;
1536 qsort(pathspec
->items
, pathspec
->nr
,
1537 sizeof(struct pathspec_item
), pathspec_item_cmp
);
1542 void free_pathspec(struct pathspec
*pathspec
)
1544 free(pathspec
->items
);
1545 pathspec
->items
= NULL
;
1548 int limit_pathspec_to_literal(void)
1550 static int flag
= -1;
1552 flag
= git_env_bool(GIT_LITERAL_PATHSPECS_ENVIRONMENT
, 0);