1 /**********************************************************************
6 created at: Wed Jan 5 09:51:01 JST 1994
8 Copyright (C) 1993-2007 Yukihiro Matsumoto
9 Copyright (C) 2000 Network Applied Communication Laboratory, Inc.
10 Copyright (C) 2000 Information-technology Promotion Agency, Japan
12 **********************************************************************/
14 #include "ruby/ruby.h"
15 #include "ruby/encoding.h"
17 #include <sys/types.h>
24 #if defined HAVE_DIRENT_H && !defined _WIN32
26 # define NAMLEN(dirent) strlen((dirent)->d_name)
27 #elif defined HAVE_DIRECT_H && !defined _WIN32
29 # define NAMLEN(dirent) strlen((dirent)->d_name)
31 # define dirent direct
32 # if !defined __NeXT__
33 # define NAMLEN(dirent) (dirent)->d_namlen
35 # /* On some versions of NextStep, d_namlen is always zero, so avoid it. */
36 # define NAMLEN(dirent) strlen((dirent)->d_name)
39 # include <sys/ndir.h>
48 # include "win32/dir.h"
59 char *strchr(char*,char);
64 #include "ruby/util.h"
66 #if !defined HAVE_LSTAT && !defined lstat
70 #define FNM_NOESCAPE 0x01
71 #define FNM_PATHNAME 0x02
72 #define FNM_DOTMATCH 0x04
73 #define FNM_CASEFOLD 0x08
74 #if CASEFOLD_FILESYSTEM
75 #define FNM_SYSCASE FNM_CASEFOLD
83 # define Next(p, e, enc) (p + rb_enc_mbclen(p, e, enc))
84 # define Inc(p, e, enc) ((p) = Next(p, e, enc))
87 char_casecmp(const char *p1
, const char *p2
, rb_encoding
*enc
, const int nocase
)
89 const char *p1end
, *p2end
;
92 if (!*p1
|| !*p2
) return !!*p1
- !!*p2
;
93 p1end
= p1
+ strlen(p1
);
94 p2end
= p2
+ strlen(p2
);
95 c1
= rb_enc_codepoint(p1
, p1end
, enc
);
96 c2
= rb_enc_codepoint(p2
, p2end
, enc
);
98 if (c1
== c2
) return 0;
100 c1
= rb_enc_toupper(c1
, enc
);
101 c2
= rb_enc_toupper(c2
, enc
);
108 const char *p
, /* pattern (next to '[') */
109 const char *s
, /* string */
113 const char *pend
= p
+ strlen(p
);
114 const int nocase
= flags
& FNM_CASEFOLD
;
115 const int escape
= !(flags
& FNM_NOESCAPE
);
119 if (*p
== '!' || *p
== '^') {
126 if (escape
&& *t1
== '\\')
130 p
= Next(t1
, pend
, enc
);
131 if (p
[0] == '-' && p
[1] != ']') {
132 const char *t2
= p
+ 1;
133 if (escape
&& *t2
== '\\')
137 p
= Next(t2
, pend
, enc
);
138 if (!ok
&& char_casecmp(t1
, s
, enc
, nocase
) <= 0 && char_casecmp(s
, t2
, enc
, nocase
) <= 0)
142 if (!ok
&& char_casecmp(t1
, s
, enc
, nocase
) == 0)
146 return ok
== not ? NULL
: (char *)p
+ 1;
149 /* If FNM_PATHNAME is set, only path element will be matched. (upto '/' or '\0')
150 Otherwise, entire string will be matched.
151 End marker itself won't be compared.
152 And if function succeeds, *pcur reaches end marker.
154 #define UNESCAPE(p) (escape && *(p) == '\\' ? (p) + 1 : (p))
155 #define ISEND(p) (!*(p) || (pathname && *(p) == '/'))
156 #define RETURN(val) return *pcur = p, *scur = s, (val);
160 const char **pcur
, /* pattern */
161 const char **scur
, /* string */
165 const int period
= !(flags
& FNM_DOTMATCH
);
166 const int pathname
= flags
& FNM_PATHNAME
;
167 const int escape
= !(flags
& FNM_NOESCAPE
);
168 const int nocase
= flags
& FNM_CASEFOLD
;
170 const char *ptmp
= 0;
171 const char *stmp
= 0;
173 const char *p
= *pcur
;
174 const char *pend
= p
+ strlen(p
);
175 const char *s
= *scur
;
176 const char *send
= s
+ strlen(s
);
178 if (period
&& *s
== '.' && *UNESCAPE(p
) != '.') /* leading period */
184 do { p
++; } while (*p
== '*');
185 if (ISEND(UNESCAPE(p
))) {
206 if ((t
= bracket(p
+ 1, s
, flags
, enc
)) != 0) {
218 RETURN(ISEND(p
) ? 0 : FNM_NOMATCH
);
221 if (char_casecmp(p
, s
, enc
, nocase
) != 0)
227 failed
: /* try next '*' position */
230 Inc(stmp
, send
, enc
); /* !ISEND(*stmp) */
245 const char *p
= pattern
;
246 const char *s
= string
;
247 const char *send
= s
+ strlen(string
);
248 const int period
= !(flags
& FNM_DOTMATCH
);
249 const int pathname
= flags
& FNM_PATHNAME
;
251 const char *ptmp
= 0;
252 const char *stmp
= 0;
256 if (p
[0] == '*' && p
[1] == '*' && p
[2] == '/') {
257 do { p
+= 3; } while (p
[0] == '*' && p
[1] == '*' && p
[2] == '/');
261 if (fnmatch_helper(&p
, &s
, flags
, enc
) == 0) {
262 while (*s
&& *s
!= '/') Inc(s
, send
, enc
);
271 /* failed : try next recursion */
272 if (ptmp
&& stmp
&& !(period
&& *stmp
== '.')) {
273 while (*stmp
&& *stmp
!= '/') Inc(stmp
, send
, enc
);
285 return fnmatch_helper(&p
, &s
, flags
, enc
);
297 mark_dir(struct dir_data
*dir
)
299 rb_gc_mark(dir
->path
);
303 free_dir(struct dir_data
*dir
)
306 if (dir
->dir
) closedir(dir
->dir
);
311 static VALUE
dir_close(VALUE
);
314 dir_s_alloc(VALUE klass
)
316 struct dir_data
*dirp
;
317 VALUE obj
= Data_Make_Struct(klass
, struct dir_data
, mark_dir
, free_dir
, dirp
);
328 * Dir.new( string ) -> aDir
330 * Returns a new directory object for the named directory.
333 dir_initialize(int argc
, VALUE
*argv
, VALUE dir
)
336 static rb_encoding
*fs_encoding
;
337 rb_encoding
*extencoding
;
339 static VALUE sym_extenc
;
342 sym_extenc
= ID2SYM(rb_intern("external_encoding"));
343 fs_encoding
= rb_filesystem_encoding();
346 extencoding
= fs_encoding
;
347 rb_scan_args(argc
, argv
, "11", &dirname
, &opt
);
350 VALUE v
, extenc
=Qnil
;
351 opt
= rb_convert_type(opt
, T_HASH
, "Hash", "to_hash");
353 v
= rb_hash_aref(opt
, sym_extenc
);
354 if (!NIL_P(v
)) extenc
= v
;
356 if (!NIL_P(extenc
)) {
357 extencoding
= rb_to_encoding(extenc
);
361 FilePathValue(dirname
);
363 Data_Get_Struct(dir
, struct dir_data
, dp
);
364 if (dp
->dir
) closedir(dp
->dir
);
367 dp
->extenc
= extencoding
;
368 dp
->dir
= opendir(RSTRING_PTR(dirname
));
369 if (dp
->dir
== NULL
) {
370 if (errno
== EMFILE
|| errno
== ENFILE
) {
372 dp
->dir
= opendir(RSTRING_PTR(dirname
));
374 if (dp
->dir
== NULL
) {
375 rb_sys_fail(RSTRING_PTR(dirname
));
378 dp
->path
= rb_str_dup_frozen(dirname
);
385 * Dir.open( string ) => aDir
386 * Dir.open( string ) {| aDir | block } => anObject
388 * With no block, <code>open</code> is a synonym for
389 * <code>Dir::new</code>. If a block is present, it is passed
390 * <i>aDir</i> as a parameter. The directory is closed at the end of
391 * the block, and <code>Dir::open</code> returns the value of the
395 dir_s_open(int argc
, VALUE
*argv
, VALUE klass
)
398 VALUE dir
= Data_Make_Struct(klass
, struct dir_data
, mark_dir
, free_dir
, dp
);
400 dir_initialize(argc
, argv
, dir
);
401 if (rb_block_given_p()) {
402 return rb_ensure(rb_yield
, dir
, dir_close
, dir
);
411 rb_raise(rb_eIOError
, "closed directory");
417 if (!OBJ_TAINTED(dir
) && rb_safe_level() >= 4)
418 rb_raise(rb_eSecurityError
, "Insecure: operation on untainted Dir");
419 rb_check_frozen(dir
);
422 #define GetDIR(obj, dirp) do {\
424 Data_Get_Struct(obj, struct dir_data, dirp);\
425 if (dirp->dir == NULL) dir_closed();\
429 dir_enc_str(VALUE str
, struct dir_data
*dirp
)
431 rb_enc_associate(str
, dirp
->extenc
);
437 * dir.inspect => string
439 * Return a string describing this Dir object.
442 dir_inspect(VALUE dir
)
444 struct dir_data
*dirp
;
446 Data_Get_Struct(dir
, struct dir_data
, dirp
);
447 if (!NIL_P(dirp
->path
)) {
448 const char *c
= rb_obj_classname(dir
);
449 return rb_sprintf("#<%s:%s>", c
, RSTRING_PTR(dirp
->path
));
451 return rb_funcall(dir
, rb_intern("to_s"), 0, 0);
456 * dir.path => string or nil
458 * Returns the path parameter passed to <em>dir</em>'s constructor.
466 struct dir_data
*dirp
;
468 Data_Get_Struct(dir
, struct dir_data
, dirp
);
469 if (NIL_P(dirp
->path
)) return Qnil
;
470 return rb_str_dup(dirp
->path
);
475 * dir.read => string or nil
477 * Reads the next entry from <em>dir</em> and returns it as a string.
478 * Returns <code>nil</code> at the end of the stream.
480 * d = Dir.new("testdir")
483 * d.read #=> "config.h"
488 struct dir_data
*dirp
;
493 dp
= readdir(dirp
->dir
);
495 return dir_enc_str(rb_tainted_str_new(dp
->d_name
, NAMLEN(dp
)), dirp
);
497 else if (errno
== 0) { /* end of stream */
503 return Qnil
; /* not reached */
508 * dir.each { |filename| block } => dir
510 * Calls the block once for each entry in this directory, passing the
511 * filename of each entry as a parameter to the block.
513 * d = Dir.new("testdir")
514 * d.each {|x| puts "Got #{x}" }
526 struct dir_data
*dirp
;
529 RETURN_ENUMERATOR(dir
, 0, 0);
531 rewinddir(dirp
->dir
);
532 for (dp
= readdir(dirp
->dir
); dp
!= NULL
; dp
= readdir(dirp
->dir
)) {
533 rb_yield(dir_enc_str(rb_tainted_str_new(dp
->d_name
, NAMLEN(dp
)), dirp
));
534 if (dirp
->dir
== NULL
) dir_closed();
542 * dir.tell => integer
544 * Returns the current position in <em>dir</em>. See also
545 * <code>Dir#seek</code>.
547 * d = Dir.new("testdir")
556 struct dir_data
*dirp
;
560 pos
= telldir(dirp
->dir
);
561 return rb_int2inum(pos
);
569 * dir.seek( integer ) => dir
571 * Seeks to a particular location in <em>dir</em>. <i>integer</i>
572 * must be a value returned by <code>Dir#tell</code>.
574 * d = Dir.new("testdir") #=> #<Dir:0x401b3c40>
578 * d.seek(i) #=> #<Dir:0x401b3c40>
582 dir_seek(VALUE dir
, VALUE pos
)
584 struct dir_data
*dirp
;
585 off_t p
= NUM2OFFT(pos
);
589 seekdir(dirp
->dir
, p
);
598 * dir.pos( integer ) => integer
600 * Synonym for <code>Dir#seek</code>, but returns the position
603 * d = Dir.new("testdir") #=> #<Dir:0x401b3c40>
611 dir_set_pos(VALUE dir
, VALUE pos
)
621 * Repositions <em>dir</em> to the first entry.
623 * d = Dir.new("testdir")
625 * d.rewind #=> #<Dir:0x401b3fb0>
629 dir_rewind(VALUE dir
)
631 struct dir_data
*dirp
;
633 if (rb_safe_level() >= 4 && !OBJ_TAINTED(dir
)) {
634 rb_raise(rb_eSecurityError
, "Insecure: can't close");
637 rewinddir(dirp
->dir
);
645 * Closes the directory stream. Any further attempts to access
646 * <em>dir</em> will raise an <code>IOError</code>.
648 * d = Dir.new("testdir")
654 struct dir_data
*dirp
;
664 dir_chdir(VALUE path
)
666 if (chdir(RSTRING_PTR(path
)) < 0)
667 rb_sys_fail(RSTRING_PTR(path
));
670 static int chdir_blocking
= 0;
671 static VALUE chdir_thread
= Qnil
;
674 VALUE old_path
, new_path
;
679 chdir_yield(struct chdir_data
*args
)
681 dir_chdir(args
->new_path
);
684 if (chdir_thread
== Qnil
)
685 chdir_thread
= rb_thread_current();
686 return rb_yield(args
->new_path
);
690 chdir_restore(struct chdir_data
*args
)
694 if (chdir_blocking
== 0)
696 dir_chdir(args
->old_path
);
703 * Dir.chdir( [ string] ) => 0
704 * Dir.chdir( [ string] ) {| path | block } => anObject
706 * Changes the current working directory of the process to the given
707 * string. When called without an argument, changes the directory to
708 * the value of the environment variable <code>HOME</code>, or
709 * <code>LOGDIR</code>. <code>SystemCallError</code> (probably
710 * <code>Errno::ENOENT</code>) if the target directory does not exist.
712 * If a block is given, it is passed the name of the new current
713 * directory, and the block is executed with that as the current
714 * directory. The original working directory is restored when the block
715 * exits. The return value of <code>chdir</code> is the value of the
716 * block. <code>chdir</code> blocks can be nested, but in a
717 * multi-threaded program an error will be raised if a thread attempts
718 * to open a <code>chdir</code> block while another thread has one
721 * Dir.chdir("/var/spool/mail")
723 * Dir.chdir("/tmp") do
725 * Dir.chdir("/usr") do
741 dir_s_chdir(int argc
, VALUE
*argv
, VALUE obj
)
746 if (rb_scan_args(argc
, argv
, "01", &path
) == 1) {
750 const char *dist
= getenv("HOME");
752 dist
= getenv("LOGDIR");
753 if (!dist
) rb_raise(rb_eArgError
, "HOME/LOGDIR not set");
755 path
= rb_str_new2(dist
);
758 if (chdir_blocking
> 0) {
759 if (!rb_block_given_p() || rb_thread_current() != chdir_thread
)
760 rb_warn("conflicting chdir during another chdir block");
763 if (rb_block_given_p()) {
764 struct chdir_data args
;
765 char *cwd
= my_getcwd();
767 args
.old_path
= rb_tainted_str_new2(cwd
); xfree(cwd
);
768 args
.new_path
= path
;
770 return rb_ensure(chdir_yield
, (VALUE
)&args
, chdir_restore
, (VALUE
)&args
);
779 * Dir.getwd => string
782 * Returns the path to the current working directory of this process as
785 * Dir.chdir("/tmp") #=> 0
786 * Dir.getwd #=> "/tmp"
789 dir_s_getwd(VALUE dir
)
796 cwd
= rb_tainted_str_new2(path
);
803 check_dirname(volatile VALUE
*dir
)
809 path
= RSTRING_PTR(*dir
);
810 if (path
&& *(pend
= rb_path_end(rb_path_skip_prefix(path
)))) {
811 *dir
= rb_str_new(path
, pend
- path
);
817 * Dir.chroot( string ) => 0
819 * Changes this process's idea of the file system root. Only a
820 * privileged process may make this call. Not available on all
821 * platforms. On Unix systems, see <code>chroot(2)</code> for more
825 dir_s_chroot(VALUE dir
, VALUE path
)
827 #if defined(HAVE_CHROOT) && !defined(__CHECKER__)
828 check_dirname(&path
);
830 if (chroot(RSTRING_PTR(path
)) == -1)
831 rb_sys_fail(RSTRING_PTR(path
));
836 return Qnil
; /* not reached */
842 * Dir.mkdir( string [, integer] ) => 0
844 * Makes a new directory named by <i>string</i>, with permissions
845 * specified by the optional parameter <i>anInteger</i>. The
846 * permissions may be modified by the value of
847 * <code>File::umask</code>, and are ignored on NT. Raises a
848 * <code>SystemCallError</code> if the directory cannot be created. See
849 * also the discussion of permissions in the class documentation for
854 dir_s_mkdir(int argc
, VALUE
*argv
, VALUE obj
)
859 if (rb_scan_args(argc
, argv
, "11", &path
, &vmode
) == 2) {
860 mode
= NUM2INT(vmode
);
866 check_dirname(&path
);
867 if (mkdir(RSTRING_PTR(path
), mode
) == -1)
868 rb_sys_fail(RSTRING_PTR(path
));
875 * Dir.delete( string ) => 0
876 * Dir.rmdir( string ) => 0
877 * Dir.unlink( string ) => 0
879 * Deletes the named directory. Raises a subclass of
880 * <code>SystemCallError</code> if the directory isn't empty.
883 dir_s_rmdir(VALUE obj
, VALUE dir
)
886 if (rmdir(RSTRING_PTR(dir
)) < 0)
887 rb_sys_fail(RSTRING_PTR(dir
));
893 sys_warning_1(const char* mesg
)
895 rb_sys_warning("%s", mesg
);
898 #define GLOB_VERBOSE (1UL << (sizeof(int) * CHAR_BIT - 1))
899 #define sys_warning(val) \
900 (void)((flags & GLOB_VERBOSE) && rb_protect((VALUE (*)(VALUE))sys_warning_1, (VALUE)(val), 0))
902 #define GLOB_ALLOC(type) (type *)malloc(sizeof(type))
903 #define GLOB_ALLOC_N(type, n) (type *)malloc(sizeof(type) * (n))
904 #define GLOB_FREE(ptr) free(ptr)
905 #define GLOB_JUMP_TAG(status) ((status == -1) ? rb_memerror() : rb_jump_tag(status))
908 * ENOTDIR can be returned by stat(2) if a non-leaf element of the path
909 * is not a directory.
911 #define to_be_ignored(e) ((e) == ENOENT || (e) == ENOTDIR)
913 /* System call with warning */
915 do_stat(const char *path
, struct stat
*pst
, int flags
)
918 int ret
= stat(path
, pst
);
919 if (ret
< 0 && !to_be_ignored(errno
))
926 do_lstat(const char *path
, struct stat
*pst
, int flags
)
928 int ret
= lstat(path
, pst
);
929 if (ret
< 0 && !to_be_ignored(errno
))
936 do_opendir(const char *path
, int flags
)
938 DIR *dirp
= opendir(path
);
939 if (dirp
== NULL
&& !to_be_ignored(errno
))
945 /* Return nonzero if S has any special globbing chars in it. */
947 has_magic(const char *s
, int flags
, rb_encoding
*enc
)
949 const int escape
= !(flags
& FNM_NOESCAPE
);
950 const int nocase
= flags
& FNM_CASEFOLD
;
952 register const char *p
= s
;
953 register const char *pend
= p
+ strlen(p
);
956 while ((c
= *p
++) != 0) {
964 if (escape
&& !(c
= *p
++))
969 if (!FNM_SYSCASE
&& ISALPHA(c
) && nocase
)
973 p
= Next(p
-1, pend
, enc
);
979 /* Find separator in globbing pattern. */
981 find_dirsep(const char *s
, int flags
, rb_encoding
*enc
)
983 const int escape
= !(flags
& FNM_NOESCAPE
);
985 register const char *p
= s
;
986 register const char *pend
= p
+ strlen(p
);
990 while ((c
= *p
++) != 0) {
1005 if (escape
&& !(c
= *p
++))
1010 p
= Next(p
-1, pend
, enc
);
1016 /* Remove escaping backslashes */
1018 remove_backslashes(char *p
, rb_encoding
*enc
)
1020 register const char *pend
= p
+ strlen(p
);
1027 memmove(t
, s
, p
- s
);
1038 memmove(t
, s
, p
- s
); /* move '\0' too */
1041 /* Globing pattern */
1042 enum glob_pattern_type
{ PLAIN
, MAGICAL
, RECURSIVE
, MATCH_ALL
, MATCH_DIR
};
1044 struct glob_pattern
{
1046 enum glob_pattern_type type
;
1047 struct glob_pattern
*next
;
1050 static void glob_free_pattern(struct glob_pattern
*list
);
1052 static struct glob_pattern
*
1053 glob_make_pattern(const char *p
, int flags
, rb_encoding
*enc
)
1055 struct glob_pattern
*list
, *tmp
, **tail
= &list
;
1056 int dirsep
= 0; /* pattern is terminated with '/' */
1059 tmp
= GLOB_ALLOC(struct glob_pattern
);
1060 if (!tmp
) goto error
;
1061 if (p
[0] == '*' && p
[1] == '*' && p
[2] == '/') {
1062 /* fold continuous RECURSIVEs (needed in glob_helper) */
1063 do { p
+= 3; } while (p
[0] == '*' && p
[1] == '*' && p
[2] == '/');
1064 tmp
->type
= RECURSIVE
;
1069 const char *m
= find_dirsep(p
, flags
, enc
);
1070 char *buf
= GLOB_ALLOC_N(char, m
-p
+1);
1075 memcpy(buf
, p
, m
-p
);
1077 tmp
->type
= has_magic(buf
, flags
, enc
) ? MAGICAL
: PLAIN
;
1092 tmp
= GLOB_ALLOC(struct glob_pattern
);
1096 glob_free_pattern(list
);
1099 tmp
->type
= dirsep
? MATCH_DIR
: MATCH_ALL
;
1108 glob_free_pattern(struct glob_pattern
*list
)
1111 struct glob_pattern
*tmp
= list
;
1114 GLOB_FREE(tmp
->str
);
1120 join_path(const char *path
, int dirsep
, const char *name
)
1122 long len
= strlen(path
);
1123 char *buf
= GLOB_ALLOC_N(char, len
+strlen(name
)+(dirsep
?1:0)+1);
1126 memcpy(buf
, path
, len
);
1128 strcpy(buf
+len
, "/");
1131 strcpy(buf
+len
, name
);
1135 enum answer
{ YES
, NO
, UNKNOWN
};
1138 # define S_ISDIR(m) ((m & S_IFMT) == S_IFDIR)
1143 # define S_ISLNK(m) (0)
1145 # define S_ISLNK(m) ((m & S_IFMT) == S_IFLNK)
1150 void (*func
)(const char *, VALUE
, void *);
1157 glob_func_caller(VALUE val
)
1159 struct glob_args
*args
= (struct glob_args
*)val
;
1161 (*args
->func
)(args
->path
, args
->value
, args
->enc
);
1165 #define glob_call_func(func, path, arg, enc) (*func)(path, arg, enc)
1170 int dirsep
, /* '/' should be placed before appending child entry's name to 'path'. */
1171 enum answer exist
, /* Does 'path' indicate an existing entry? */
1172 enum answer isdir
, /* Does 'path' indicate a directory or a symlink to a directory? */
1173 struct glob_pattern
**beg
,
1174 struct glob_pattern
**end
,
1176 ruby_glob_func
*func
,
1182 struct glob_pattern
**cur
, **new_beg
, **new_end
;
1183 int plain
= 0, magical
= 0, recursive
= 0, match_all
= 0, match_dir
= 0;
1184 int escape
= !(flags
& FNM_NOESCAPE
);
1186 for (cur
= beg
; cur
< end
; ++cur
) {
1187 struct glob_pattern
*p
= *cur
;
1188 if (p
->type
== RECURSIVE
) {
1206 rb_bug("continuous RECURSIVEs");
1211 if (match_all
&& exist
== UNKNOWN
) {
1212 if (do_lstat(path
, &st
, flags
) == 0) {
1214 isdir
= S_ISDIR(st
.st_mode
) ? YES
: S_ISLNK(st
.st_mode
) ? UNKNOWN
: NO
;
1221 if (match_dir
&& isdir
== UNKNOWN
) {
1222 if (do_stat(path
, &st
, flags
) == 0) {
1224 isdir
= S_ISDIR(st
.st_mode
) ? YES
: NO
;
1231 if (match_all
&& exist
== YES
) {
1232 status
= glob_call_func(func
, path
, arg
, enc
);
1233 if (status
) return status
;
1235 if (match_dir
&& isdir
== YES
) {
1236 char *tmp
= join_path(path
, dirsep
, "");
1237 if (!tmp
) return -1;
1238 status
= glob_call_func(func
, tmp
, arg
, enc
);
1240 if (status
) return status
;
1244 if (exist
== NO
|| isdir
== NO
) return 0;
1246 if (magical
|| recursive
) {
1248 DIR *dirp
= do_opendir(*path
? path
: ".", flags
);
1249 if (dirp
== NULL
) return 0;
1251 for (dp
= readdir(dirp
); dp
!= NULL
; dp
= readdir(dirp
)) {
1252 char *buf
= join_path(path
, dirsep
, dp
->d_name
);
1253 enum answer new_isdir
= UNKNOWN
;
1259 if (recursive
&& strcmp(dp
->d_name
, ".") != 0 && strcmp(dp
->d_name
, "..") != 0
1260 && fnmatch("*", rb_usascii_encoding(), dp
->d_name
, flags
) == 0) {
1262 if (do_lstat(buf
, &st
, flags
) == 0)
1263 new_isdir
= S_ISDIR(st
.st_mode
) ? YES
: S_ISLNK(st
.st_mode
) ? UNKNOWN
: NO
;
1267 new_isdir
= dp
->d_isdir
? (!dp
->d_isrep
? YES
: UNKNOWN
) : NO
;
1271 new_beg
= new_end
= GLOB_ALLOC_N(struct glob_pattern
*, (end
- beg
) * 2);
1278 for (cur
= beg
; cur
< end
; ++cur
) {
1279 struct glob_pattern
*p
= *cur
;
1280 if (p
->type
== RECURSIVE
) {
1281 if (new_isdir
== YES
) /* not symlink but real directory */
1282 *new_end
++ = p
; /* append recursive pattern */
1283 p
= p
->next
; /* 0 times recursion */
1285 if (p
->type
== PLAIN
|| p
->type
== MAGICAL
) {
1286 if (fnmatch(p
->str
, enc
, dp
->d_name
, flags
) == 0)
1287 *new_end
++ = p
->next
;
1291 status
= glob_helper(buf
, 1, YES
, new_isdir
, new_beg
, new_end
,
1292 flags
, func
, arg
, enc
);
1301 struct glob_pattern
**copy_beg
, **copy_end
, **cur2
;
1303 copy_beg
= copy_end
= GLOB_ALLOC_N(struct glob_pattern
*, end
- beg
);
1304 if (!copy_beg
) return -1;
1305 for (cur
= beg
; cur
< end
; ++cur
)
1306 *copy_end
++ = (*cur
)->type
== PLAIN
? *cur
: 0;
1308 for (cur
= copy_beg
; cur
< copy_end
; ++cur
) {
1312 name
= GLOB_ALLOC_N(char, strlen((*cur
)->str
) + 1);
1317 strcpy(name
, (*cur
)->str
);
1318 if (escape
) remove_backslashes(name
, enc
);
1320 new_beg
= new_end
= GLOB_ALLOC_N(struct glob_pattern
*, end
- beg
);
1326 *new_end
++ = (*cur
)->next
;
1327 for (cur2
= cur
+ 1; cur2
< copy_end
; ++cur2
) {
1328 if (*cur2
&& fnmatch((*cur2
)->str
, enc
, name
, flags
) == 0) {
1329 *new_end
++ = (*cur2
)->next
;
1334 buf
= join_path(path
, dirsep
, name
);
1341 status
= glob_helper(buf
, 1, UNKNOWN
, UNKNOWN
, new_beg
,
1342 new_end
, flags
, func
, arg
, enc
);
1349 GLOB_FREE(copy_beg
);
1356 ruby_glob0(const char *path
, int flags
, ruby_glob_func
*func
, VALUE arg
, rb_encoding
*enc
)
1358 struct glob_pattern
*list
;
1359 const char *root
, *start
;
1364 start
= root
= path
;
1365 flags
|= FNM_SYSCASE
;
1367 root
= rb_path_skip_prefix(root
);
1370 if (root
&& *root
== '/') root
++;
1373 buf
= GLOB_ALLOC_N(char, n
+ 1);
1374 if (!buf
) return -1;
1375 MEMCPY(buf
, start
, char, n
);
1378 list
= glob_make_pattern(root
, flags
, enc
);
1383 status
= glob_helper(buf
, 0, UNKNOWN
, UNKNOWN
, &list
, &list
+ 1, flags
, func
, arg
, enc
);
1384 glob_free_pattern(list
);
1391 ruby_glob(const char *path
, int flags
, ruby_glob_func
*func
, VALUE arg
)
1393 return ruby_glob0(path
, flags
& ~GLOB_VERBOSE
, func
, arg
,
1394 rb_ascii8bit_encoding());
1398 rb_glob_caller(const char *path
, VALUE a
, void *enc
)
1401 struct glob_args
*args
= (struct glob_args
*)a
;
1404 rb_protect(glob_func_caller
, a
, &status
);
1409 rb_glob2(const char *path
, int flags
,
1410 void (*func
)(const char *, VALUE
, void *), VALUE arg
,
1413 struct glob_args args
;
1419 if (flags
& FNM_SYSCASE
) {
1420 rb_warning("Dir.glob() ignores File::FNM_CASEFOLD");
1423 return ruby_glob0(path
, flags
| GLOB_VERBOSE
, rb_glob_caller
, (VALUE
)&args
,
1428 rb_glob(const char *path
, void (*func
)(const char *, VALUE
, void *), VALUE arg
)
1430 int status
= rb_glob2(path
, 0, func
, arg
, rb_ascii8bit_encoding());
1431 if (status
) GLOB_JUMP_TAG(status
);
1435 push_pattern(const char *path
, VALUE ary
, void *enc
)
1437 VALUE vpath
= rb_tainted_str_new2(path
);
1438 rb_enc_associate(vpath
, enc
);
1439 rb_ary_push(ary
, vpath
);
1443 ruby_brace_expand(const char *str
, int flags
, ruby_glob_func
*func
, VALUE arg
,
1446 const int escape
= !(flags
& FNM_NOESCAPE
);
1447 const char *p
= str
;
1448 const char *pend
= p
+ strlen(p
);
1450 const char *lbrace
= 0, *rbrace
= 0;
1451 int nest
= 0, status
= 0;
1454 if (*p
== '{' && nest
++ == 0) {
1457 if (*p
== '}' && --nest
<= 0) {
1461 if (*p
== '\\' && escape
) {
1467 if (lbrace
&& rbrace
) {
1468 char *buf
= GLOB_ALLOC_N(char, strlen(s
) + 1);
1471 if (!buf
) return -1;
1472 memcpy(buf
, s
, lbrace
-s
);
1475 while (p
< rbrace
) {
1476 const char *t
= ++p
;
1478 while (p
< rbrace
&& !(*p
== ',' && nest
== 0)) {
1479 if (*p
== '{') nest
++;
1480 if (*p
== '}') nest
--;
1481 if (*p
== '\\' && escape
) {
1482 if (++p
== rbrace
) break;
1486 memcpy(buf
+shift
, t
, p
-t
);
1487 strcpy(buf
+shift
+(p
-t
), rbrace
+1);
1488 status
= ruby_brace_expand(buf
, flags
, func
, arg
, enc
);
1493 else if (!lbrace
&& !rbrace
) {
1494 status
= (*func
)(s
, arg
, enc
);
1501 ruby_glob_func
*func
;
1507 glob_brace(const char *path
, VALUE val
, void *enc
)
1509 struct brace_args
*arg
= (struct brace_args
*)val
;
1511 return ruby_glob0(path
, arg
->flags
, arg
->func
, arg
->value
, enc
);
1515 ruby_brace_glob0(const char *str
, int flags
, ruby_glob_func
*func
, VALUE arg
,
1518 struct brace_args args
;
1523 return ruby_brace_expand(str
, flags
, glob_brace
, (VALUE
)&args
, enc
);
1527 ruby_brace_glob(const char *str
, int flags
, ruby_glob_func
*func
, VALUE arg
)
1529 return ruby_brace_glob0(str
, flags
& ~GLOB_VERBOSE
, func
, arg
,
1530 rb_ascii8bit_encoding());
1534 push_glob(VALUE ary
, VALUE str
, int flags
)
1536 struct glob_args args
;
1537 rb_encoding
*enc
= rb_enc_get(str
);
1539 args
.func
= push_pattern
;
1543 return ruby_brace_glob0(RSTRING_PTR(str
), flags
| GLOB_VERBOSE
,
1544 rb_glob_caller
, (VALUE
)&args
, enc
);
1548 rb_push_glob(VALUE str
, int flags
) /* '\0' is delimiter */
1556 while (offset
< RSTRING_LEN(str
)) {
1559 p
= RSTRING_PTR(str
) + offset
;
1560 status
= push_glob(ary
, rb_enc_str_new(p
, strlen(p
), rb_enc_get(str
)),
1562 if (status
) GLOB_JUMP_TAG(status
);
1563 if (offset
>= RSTRING_LEN(str
)) break;
1565 pend
= RSTRING_PTR(str
) + RSTRING_LEN(str
);
1566 while (p
< pend
&& !*p
)
1568 offset
= p
- RSTRING_PTR(str
);
1575 dir_globs(long argc
, VALUE
*argv
, int flags
)
1577 VALUE ary
= rb_ary_new();
1580 for (i
= 0; i
< argc
; ++i
) {
1582 VALUE str
= argv
[i
];
1584 status
= push_glob(ary
, str
, flags
);
1585 if (status
) GLOB_JUMP_TAG(status
);
1593 * Dir[ array ] => array
1594 * Dir[ string [, string ...] ] => array
1596 * Equivalent to calling
1597 * <code>Dir.glob(</code><i>array,</i><code>0)</code> and
1598 * <code>Dir.glob([</code><i>string,...</i><code>],0)</code>.
1602 dir_s_aref(int argc
, VALUE
*argv
, VALUE obj
)
1605 return rb_push_glob(argv
[0], 0);
1607 return dir_globs(argc
, argv
, 0);
1612 * Dir.glob( pattern, [flags] ) => array
1613 * Dir.glob( pattern, [flags] ) {| filename | block } => nil
1615 * Returns the filenames found by expanding <i>pattern</i> which is
1616 * an +Array+ of the patterns or the pattern +String+, either as an
1617 * <i>array</i> or as parameters to the block. Note that this pattern
1618 * is not a regexp (it's closer to a shell glob). See
1619 * <code>File::fnmatch</code> for the meaning of the <i>flags</i>
1620 * parameter. Note that case sensitivity depends on your system (so
1621 * <code>File::FNM_CASEFOLD</code> is ignored)
1623 * <code>*</code>:: Matches any file. Can be restricted by
1624 * other values in the glob. <code>*</code>
1625 * will match all files; <code>c*</code> will
1626 * match all files beginning with
1627 * <code>c</code>; <code>*c</code> will match
1628 * all files ending with <code>c</code>; and
1629 * <code>*c*</code> will match all files that
1630 * have <code>c</code> in them (including at
1631 * the beginning or end). Equivalent to
1632 * <code>/ .* /x</code> in regexp.
1633 * <code>**</code>:: Matches directories recursively.
1634 * <code>?</code>:: Matches any one character. Equivalent to
1635 * <code>/.{1}/</code> in regexp.
1636 * <code>[set]</code>:: Matches any one character in +set+.
1637 * Behaves exactly like character sets in
1638 * Regexp, including set negation
1639 * (<code>[^a-z]</code>).
1640 * <code>{p,q}</code>:: Matches either literal <code>p</code> or
1641 * literal <code>q</code>. Matching literals
1642 * may be more than one character in length.
1643 * More than two literals may be specified.
1644 * Equivalent to pattern alternation in
1646 * <code>\</code>:: Escapes the next metacharacter.
1648 * Dir["config.?"] #=> ["config.h"]
1649 * Dir.glob("config.?") #=> ["config.h"]
1650 * Dir.glob("*.[a-z][a-z]") #=> ["main.rb"]
1651 * Dir.glob("*.[^r]*") #=> ["config.h"]
1652 * Dir.glob("*.{rb,h}") #=> ["main.rb", "config.h"]
1653 * Dir.glob("*") #=> ["config.h", "main.rb"]
1654 * Dir.glob("*", File::FNM_DOTMATCH) #=> [".", "..", "config.h", "main.rb"]
1656 * rbfiles = File.join("**", "*.rb")
1657 * Dir.glob(rbfiles) #=> ["main.rb",
1659 * # "lib/song/karaoke.rb"]
1660 * libdirs = File.join("**", "lib")
1661 * Dir.glob(libdirs) #=> ["lib"]
1663 * librbfiles = File.join("**", "lib", "**", "*.rb")
1664 * Dir.glob(librbfiles) #=> ["lib/song.rb",
1665 * # "lib/song/karaoke.rb"]
1667 * librbfiles = File.join("**", "lib", "*.rb")
1668 * Dir.glob(librbfiles) #=> ["lib/song.rb"]
1671 dir_s_glob(int argc
, VALUE
*argv
, VALUE obj
)
1673 VALUE str
, rflags
, ary
;
1676 if (rb_scan_args(argc
, argv
, "11", &str
, &rflags
) == 2)
1677 flags
= NUM2INT(rflags
);
1681 ary
= rb_check_array_type(str
);
1683 ary
= rb_push_glob(str
, flags
);
1686 volatile VALUE v
= ary
;
1687 ary
= dir_globs(RARRAY_LEN(v
), RARRAY_PTR(v
), flags
);
1690 if (rb_block_given_p()) {
1698 dir_open_dir(int argc
, VALUE
*argv
)
1700 VALUE dir
= rb_funcall2(rb_cDir
, rb_intern("open"), argc
, argv
);
1702 if (TYPE(dir
) != T_DATA
||
1703 RDATA(dir
)->dfree
!= (RUBY_DATA_FUNC
)free_dir
) {
1704 rb_raise(rb_eTypeError
, "wrong argument type %s (expected Dir)",
1705 rb_obj_classname(dir
));
1713 * Dir.foreach( dirname ) {| filename | block } => nil
1715 * Calls the block once for each entry in the named directory, passing
1716 * the filename of each entry as a parameter to the block.
1718 * Dir.foreach("testdir") {|x| puts "Got #{x}" }
1720 * <em>produces:</em>
1729 dir_foreach(int argc
, VALUE
*argv
, VALUE io
)
1733 RETURN_ENUMERATOR(io
, argc
, argv
);
1734 dir
= dir_open_dir(argc
, argv
);
1735 rb_ensure(dir_each
, dir
, dir_close
, dir
);
1741 * Dir.entries( dirname ) => array
1743 * Returns an array containing all of the filenames in the given
1744 * directory. Will raise a <code>SystemCallError</code> if the named
1745 * directory doesn't exist.
1747 * Dir.entries("testdir") #=> [".", "..", "config.h", "main.rb"]
1751 dir_entries(int argc
, VALUE
*argv
, VALUE io
)
1755 dir
= dir_open_dir(argc
, argv
);
1756 return rb_ensure(rb_Array
, dir
, dir_close
, dir
);
1761 * File.fnmatch( pattern, path, [flags] ) => (true or false)
1762 * File.fnmatch?( pattern, path, [flags] ) => (true or false)
1764 * Returns true if <i>path</i> matches against <i>pattern</i> The
1765 * pattern is not a regular expression; instead it follows rules
1766 * similar to shell filename globbing. It may contain the following
1769 * <code>*</code>:: Matches any file. Can be restricted by
1770 * other values in the glob. <code>*</code>
1771 * will match all files; <code>c*</code> will
1772 * match all files beginning with
1773 * <code>c</code>; <code>*c</code> will match
1774 * all files ending with <code>c</code>; and
1775 * <code>*c*</code> will match all files that
1776 * have <code>c</code> in them (including at
1777 * the beginning or end). Equivalent to
1778 * <code>/ .* /x</code> in regexp.
1779 * <code>**</code>:: Matches directories recursively or files
1781 * <code>?</code>:: Matches any one character. Equivalent to
1782 * <code>/.{1}/</code> in regexp.
1783 * <code>[set]</code>:: Matches any one character in +set+.
1784 * Behaves exactly like character sets in
1785 * Regexp, including set negation
1786 * (<code>[^a-z]</code>).
1787 * <code>\</code>:: Escapes the next metacharacter.
1789 * <i>flags</i> is a bitwise OR of the <code>FNM_xxx</code>
1790 * parameters. The same glob pattern and flags are used by
1791 * <code>Dir::glob</code>.
1793 * File.fnmatch('cat', 'cat') #=> true # match entire string
1794 * File.fnmatch('cat', 'category') #=> false # only match partial string
1795 * File.fnmatch('c{at,ub}s', 'cats') #=> false # { } isn't supported
1797 * File.fnmatch('c?t', 'cat') #=> true # '?' match only 1 character
1798 * File.fnmatch('c??t', 'cat') #=> false # ditto
1799 * File.fnmatch('c*', 'cats') #=> true # '*' match 0 or more characters
1800 * File.fnmatch('c*t', 'c/a/b/t') #=> true # ditto
1801 * File.fnmatch('ca[a-z]', 'cat') #=> true # inclusive bracket expression
1802 * File.fnmatch('ca[^t]', 'cat') #=> false # exclusive bracket expression ('^' or '!')
1804 * File.fnmatch('cat', 'CAT') #=> false # case sensitive
1805 * File.fnmatch('cat', 'CAT', File::FNM_CASEFOLD) #=> true # case insensitive
1807 * File.fnmatch('?', '/', File::FNM_PATHNAME) #=> false # wildcard doesn't match '/' on FNM_PATHNAME
1808 * File.fnmatch('*', '/', File::FNM_PATHNAME) #=> false # ditto
1809 * File.fnmatch('[/]', '/', File::FNM_PATHNAME) #=> false # ditto
1811 * File.fnmatch('\?', '?') #=> true # escaped wildcard becomes ordinary
1812 * File.fnmatch('\a', 'a') #=> true # escaped ordinary remains ordinary
1813 * File.fnmatch('\a', '\a', File::FNM_NOESCAPE) #=> true # FNM_NOESACPE makes '\' ordinary
1814 * File.fnmatch('[\?]', '?') #=> true # can escape inside bracket expression
1816 * File.fnmatch('*', '.profile') #=> false # wildcard doesn't match leading
1817 * File.fnmatch('*', '.profile', File::FNM_DOTMATCH) #=> true # period by default.
1818 * File.fnmatch('.*', '.profile') #=> true
1820 * rbfiles = '**' '/' '*.rb' # you don't have to do like this. just write in single string.
1821 * File.fnmatch(rbfiles, 'main.rb') #=> false
1822 * File.fnmatch(rbfiles, './main.rb') #=> false
1823 * File.fnmatch(rbfiles, 'lib/song.rb') #=> true
1824 * File.fnmatch('**.rb', 'main.rb') #=> true
1825 * File.fnmatch('**.rb', './main.rb') #=> false
1826 * File.fnmatch('**.rb', 'lib/song.rb') #=> true
1827 * File.fnmatch('*', 'dave/.profile') #=> true
1829 * pattern = '*' '/' '*'
1830 * File.fnmatch(pattern, 'dave/.profile', File::FNM_PATHNAME) #=> false
1831 * File.fnmatch(pattern, 'dave/.profile', File::FNM_PATHNAME | File::FNM_DOTMATCH) #=> true
1833 * pattern = '**' '/' 'foo'
1834 * File.fnmatch(pattern, 'a/b/c/foo', File::FNM_PATHNAME) #=> true
1835 * File.fnmatch(pattern, '/a/b/c/foo', File::FNM_PATHNAME) #=> true
1836 * File.fnmatch(pattern, 'c:/a/b/c/foo', File::FNM_PATHNAME) #=> true
1837 * File.fnmatch(pattern, 'a/.b/c/foo', File::FNM_PATHNAME) #=> false
1838 * File.fnmatch(pattern, 'a/.b/c/foo', File::FNM_PATHNAME | File::FNM_DOTMATCH) #=> true
1841 file_s_fnmatch(int argc
, VALUE
*argv
, VALUE obj
)
1843 VALUE pattern
, path
;
1847 if (rb_scan_args(argc
, argv
, "21", &pattern
, &path
, &rflags
) == 3)
1848 flags
= NUM2INT(rflags
);
1852 StringValue(pattern
);
1853 FilePathStringValue(path
);
1855 if (fnmatch(RSTRING_PTR(pattern
), rb_enc_get(pattern
), RSTRING_PTR(path
),
1863 * Objects of class <code>Dir</code> are directory streams representing
1864 * directories in the underlying file system. They provide a variety of
1865 * ways to list directories and their contents. See also
1866 * <code>File</code>.
1868 * The directory used in these examples contains the two regular files
1869 * (<code>config.h</code> and <code>main.rb</code>), the parent
1870 * directory (<code>..</code>), and the directory itself
1876 rb_cDir
= rb_define_class("Dir", rb_cObject
);
1878 rb_include_module(rb_cDir
, rb_mEnumerable
);
1880 rb_define_alloc_func(rb_cDir
, dir_s_alloc
);
1881 rb_define_singleton_method(rb_cDir
, "open", dir_s_open
, -1);
1882 rb_define_singleton_method(rb_cDir
, "foreach", dir_foreach
, -1);
1883 rb_define_singleton_method(rb_cDir
, "entries", dir_entries
, -1);
1885 rb_define_method(rb_cDir
,"initialize", dir_initialize
, -1);
1886 rb_define_method(rb_cDir
,"path", dir_path
, 0);
1887 rb_define_method(rb_cDir
,"inspect", dir_inspect
, 0);
1888 rb_define_method(rb_cDir
,"read", dir_read
, 0);
1889 rb_define_method(rb_cDir
,"each", dir_each
, 0);
1890 rb_define_method(rb_cDir
,"rewind", dir_rewind
, 0);
1891 rb_define_method(rb_cDir
,"tell", dir_tell
, 0);
1892 rb_define_method(rb_cDir
,"seek", dir_seek
, 1);
1893 rb_define_method(rb_cDir
,"pos", dir_tell
, 0);
1894 rb_define_method(rb_cDir
,"pos=", dir_set_pos
, 1);
1895 rb_define_method(rb_cDir
,"close", dir_close
, 0);
1897 rb_define_singleton_method(rb_cDir
,"chdir", dir_s_chdir
, -1);
1898 rb_define_singleton_method(rb_cDir
,"getwd", dir_s_getwd
, 0);
1899 rb_define_singleton_method(rb_cDir
,"pwd", dir_s_getwd
, 0);
1900 rb_define_singleton_method(rb_cDir
,"chroot", dir_s_chroot
, 1);
1901 rb_define_singleton_method(rb_cDir
,"mkdir", dir_s_mkdir
, -1);
1902 rb_define_singleton_method(rb_cDir
,"rmdir", dir_s_rmdir
, 1);
1903 rb_define_singleton_method(rb_cDir
,"delete", dir_s_rmdir
, 1);
1904 rb_define_singleton_method(rb_cDir
,"unlink", dir_s_rmdir
, 1);
1906 rb_define_singleton_method(rb_cDir
,"glob", dir_s_glob
, -1);
1907 rb_define_singleton_method(rb_cDir
,"[]", dir_s_aref
, -1);
1908 rb_define_singleton_method(rb_cDir
,"exist?", rb_file_directory_p
, 1); /* in file.c */
1909 rb_define_singleton_method(rb_cDir
,"exists?", rb_file_directory_p
, 1); /* in file.c */
1911 rb_define_singleton_method(rb_cFile
,"fnmatch", file_s_fnmatch
, -1);
1912 rb_define_singleton_method(rb_cFile
,"fnmatch?", file_s_fnmatch
, -1);
1914 rb_file_const("FNM_NOESCAPE", INT2FIX(FNM_NOESCAPE
));
1915 rb_file_const("FNM_PATHNAME", INT2FIX(FNM_PATHNAME
));
1916 rb_file_const("FNM_DOTMATCH", INT2FIX(FNM_DOTMATCH
));
1917 rb_file_const("FNM_CASEFOLD", INT2FIX(FNM_CASEFOLD
));
1918 rb_file_const("FNM_SYSCASE", INT2FIX(FNM_SYSCASE
));