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"
16 #include <sys/types.h>
23 #if defined HAVE_DIRENT_H && !defined _WIN32
25 # define NAMLEN(dirent) strlen((dirent)->d_name)
26 #elif defined HAVE_DIRECT_H && !defined _WIN32
28 # define NAMLEN(dirent) strlen((dirent)->d_name)
30 # define dirent direct
31 # if !defined __NeXT__
32 # define NAMLEN(dirent) (dirent)->d_namlen
34 # /* On some versions of NextStep, d_namlen is always zero, so avoid it. */
35 # define NAMLEN(dirent) strlen((dirent)->d_name)
38 # include <sys/ndir.h>
47 # include "win32/dir.h"
58 char *strchr(char*,char);
63 #include "ruby/util.h"
65 #if !defined HAVE_LSTAT && !defined lstat
69 #ifndef CASEFOLD_FILESYSTEM
70 # if defined DOSISH || defined __VMS
71 # define CASEFOLD_FILESYSTEM 1
73 # define CASEFOLD_FILESYSTEM 0
77 #define FNM_NOESCAPE 0x01
78 #define FNM_PATHNAME 0x02
79 #define FNM_DOTMATCH 0x04
80 #define FNM_CASEFOLD 0x08
81 #if CASEFOLD_FILESYSTEM
82 #define FNM_SYSCASE FNM_CASEFOLD
90 #define downcase(c) (nocase && ISUPPER(c) ? TOLOWER(c) : (c))
91 #define compare(c1, c2) (((unsigned char)(c1)) - ((unsigned char)(c2)))
93 /* caution: in case *p == '\0'
94 Next(p) == p + 1 in single byte environment
95 Next(p) == p in multi byte environment
98 # define Next(p) CharNext(p)
100 # define Next(p) ((p) + mblen(p, RUBY_MBCHAR_MAXSIZE))
101 #elif defined(__EMX__)
102 # define Next(p) ((p) + emx_mblen(p))
104 emx_mblen(const char *p
)
106 int n
= mblen(p
, RUBY_MBCHAR_MAXSIZE
);
107 return (n
< 0) ? 1 : n
;
111 #ifndef Next /* single byte environment */
112 # define Next(p) ((p) + 1)
113 # define Inc(p) (++(p))
114 # define Compare(p1, p2) (compare(downcase(*(p1)), downcase(*(p2))))
115 #else /* multi byte environment */
116 # define Inc(p) ((p) = Next(p))
117 # define Compare(p1, p2) (CompareImpl(p1, p2, nocase))
119 CompareImpl(const char *p1
, const char *p2
, int nocase
)
121 const int len1
= Next(p1
) - p1
;
122 const int len2
= Next(p2
) - p2
;
124 char buf1
[10], buf2
[10]; /* large enough? */
127 if (len1
< 0 || len2
< 0) {
128 rb_fatal("CompareImpl: negative len");
131 if (len1
== 0) return len2
;
132 if (len2
== 0) return -len1
;
135 if (nocase
&& rb_w32_iswinnt()) {
137 if (len1
>= sizeof(buf1
)) {
138 rb_fatal("CompareImpl: too large len");
140 memcpy(buf1
, p1
, len1
);
143 p1
= buf1
; /* trick */
146 if (len2
>= sizeof(buf2
)) {
147 rb_fatal("CompareImpl: too large len");
149 memcpy(buf2
, p2
, len2
);
152 p2
= buf2
; /* trick */
158 return compare(downcase(*p1
), downcase(*p2
));
160 const int ret
= compare(downcase(*p1
), *p2
);
161 return ret
? ret
: -1;
165 const int ret
= compare(*p1
, downcase(*p2
));
166 return ret
? ret
: 1;
169 const int ret
= memcmp(p1
, p2
, len1
< len2
? len1
: len2
);
170 return ret
? ret
: len1
- len2
;
173 #endif /* environment */
177 const char *p
, /* pattern (next to '[') */
178 const char *s
, /* string */
181 const int nocase
= flags
& FNM_CASEFOLD
;
182 const int escape
= !(flags
& FNM_NOESCAPE
);
186 if (*p
== '!' || *p
== '^') {
193 if (escape
&& *t1
== '\\')
198 if (p
[0] == '-' && p
[1] != ']') {
199 const char *t2
= p
+ 1;
200 if (escape
&& *t2
== '\\')
205 if (!ok
&& Compare(t1
, s
) <= 0 && Compare(s
, t2
) <= 0)
209 if (!ok
&& Compare(t1
, s
) == 0)
213 return ok
== not ? NULL
: (char *)p
+ 1;
216 /* If FNM_PATHNAME is set, only path element will be matched. (upto '/' or '\0')
217 Otherwise, entire string will be matched.
218 End marker itself won't be compared.
219 And if function succeeds, *pcur reaches end marker.
221 #define UNESCAPE(p) (escape && *(p) == '\\' ? (p) + 1 : (p))
222 #define ISEND(p) (!*(p) || (pathname && *(p) == '/'))
223 #define RETURN(val) return *pcur = p, *scur = s, (val);
227 const char **pcur
, /* pattern */
228 const char **scur
, /* string */
231 const int period
= !(flags
& FNM_DOTMATCH
);
232 const int pathname
= flags
& FNM_PATHNAME
;
233 const int escape
= !(flags
& FNM_NOESCAPE
);
234 const int nocase
= flags
& FNM_CASEFOLD
;
236 const char *ptmp
= 0;
237 const char *stmp
= 0;
239 const char *p
= *pcur
;
240 const char *s
= *scur
;
242 if (period
&& *s
== '.' && *UNESCAPE(p
) != '.') /* leading period */
248 do { p
++; } while (*p
== '*');
249 if (ISEND(UNESCAPE(p
))) {
270 if ((t
= bracket(p
+ 1, s
, flags
)) != 0) {
282 RETURN(ISEND(p
) ? 0 : FNM_NOMATCH
);
285 if (Compare(p
, s
) != 0)
291 failed
: /* try next '*' position */
294 Inc(stmp
); /* !ISEND(*stmp) */
304 const char *p
, /* pattern */
305 const char *s
, /* string */
308 const int period
= !(flags
& FNM_DOTMATCH
);
309 const int pathname
= flags
& FNM_PATHNAME
;
311 const char *ptmp
= 0;
312 const char *stmp
= 0;
316 if (p
[0] == '*' && p
[1] == '*' && p
[2] == '/') {
317 do { p
+= 3; } while (p
[0] == '*' && p
[1] == '*' && p
[2] == '/');
321 if (fnmatch_helper(&p
, &s
, flags
) == 0) {
322 while (*s
&& *s
!= '/') Inc(s
);
331 /* failed : try next recursion */
332 if (ptmp
&& stmp
&& !(period
&& *stmp
== '.')) {
333 while (*stmp
&& *stmp
!= '/') Inc(stmp
);
345 return fnmatch_helper(&p
, &s
, flags
);
356 free_dir(struct dir_data
*dir
)
359 if (dir
->dir
) closedir(dir
->dir
);
360 if (dir
->path
) free(dir
->path
);
365 static VALUE
dir_close(VALUE
);
368 dir_s_alloc(VALUE klass
)
370 struct dir_data
*dirp
;
371 VALUE obj
= Data_Make_Struct(klass
, struct dir_data
, 0, free_dir
, dirp
);
381 * Dir.new( string ) -> aDir
383 * Returns a new directory object for the named directory.
386 dir_initialize(VALUE dir
, VALUE dirname
)
390 FilePathValue(dirname
);
391 Data_Get_Struct(dir
, struct dir_data
, dp
);
392 if (dp
->dir
) closedir(dp
->dir
);
393 if (dp
->path
) free(dp
->path
);
396 dp
->dir
= opendir(RSTRING_PTR(dirname
));
397 if (dp
->dir
== NULL
) {
398 if (errno
== EMFILE
|| errno
== ENFILE
) {
400 dp
->dir
= opendir(RSTRING_PTR(dirname
));
402 if (dp
->dir
== NULL
) {
403 rb_sys_fail(RSTRING_PTR(dirname
));
406 dp
->path
= strdup(RSTRING_PTR(dirname
));
413 * Dir.open( string ) => aDir
414 * Dir.open( string ) {| aDir | block } => anObject
416 * With no block, <code>open</code> is a synonym for
417 * <code>Dir::new</code>. If a block is present, it is passed
418 * <i>aDir</i> as a parameter. The directory is closed at the end of
419 * the block, and <code>Dir::open</code> returns the value of the
423 dir_s_open(VALUE klass
, VALUE dirname
)
426 VALUE dir
= Data_Make_Struct(klass
, struct dir_data
, 0, free_dir
, dp
);
428 dir_initialize(dir
, dirname
);
429 if (rb_block_given_p()) {
430 return rb_ensure(rb_yield
, dir
, dir_close
, dir
);
439 rb_raise(rb_eIOError
, "closed directory");
445 if (!OBJ_TAINTED(dir
) && rb_safe_level() >= 4)
446 rb_raise(rb_eSecurityError
, "Insecure: operation on untainted Dir");
447 rb_check_frozen(dir
);
450 #define GetDIR(obj, dirp) do {\
452 Data_Get_Struct(obj, struct dir_data, dirp);\
453 if (dirp->dir == NULL) dir_closed();\
458 * dir.inspect => string
460 * Return a string describing this Dir object.
463 dir_inspect(VALUE dir
)
465 struct dir_data
*dirp
;
467 Data_Get_Struct(dir
, struct dir_data
, dirp
);
469 char *c
= rb_obj_classname(dir
);
470 int len
= strlen(c
) + strlen(dirp
->path
) + 4;
471 VALUE s
= rb_str_new(0, len
);
472 snprintf(RSTRING_PTR(s
), len
+1, "#<%s:%s>", c
, dirp
->path
);
475 return rb_funcall(dir
, rb_intern("to_s"), 0, 0);
480 * dir.path => string or nil
482 * Returns the path parameter passed to <em>dir</em>'s constructor.
490 struct dir_data
*dirp
;
492 Data_Get_Struct(dir
, struct dir_data
, dirp
);
493 if (!dirp
->path
) return Qnil
;
494 return rb_str_new2(dirp
->path
);
499 * dir.read => string or nil
501 * Reads the next entry from <em>dir</em> and returns it as a string.
502 * Returns <code>nil</code> at the end of the stream.
504 * d = Dir.new("testdir")
507 * d.read #=> "config.h"
512 struct dir_data
*dirp
;
517 dp
= readdir(dirp
->dir
);
519 return rb_tainted_str_new(dp
->d_name
, NAMLEN(dp
));
521 else if (errno
== 0) { /* end of stream */
527 return Qnil
; /* not reached */
532 * dir.each { |filename| block } => dir
534 * Calls the block once for each entry in this directory, passing the
535 * filename of each entry as a parameter to the block.
537 * d = Dir.new("testdir")
538 * d.each {|x| puts "Got #{x}" }
550 struct dir_data
*dirp
;
553 RETURN_ENUMERATOR(dir
, 0, 0);
555 rewinddir(dirp
->dir
);
556 for (dp
= readdir(dirp
->dir
); dp
!= NULL
; dp
= readdir(dirp
->dir
)) {
557 rb_yield(rb_tainted_str_new(dp
->d_name
, NAMLEN(dp
)));
558 if (dirp
->dir
== NULL
) dir_closed();
566 * dir.tell => integer
568 * Returns the current position in <em>dir</em>. See also
569 * <code>Dir#seek</code>.
571 * d = Dir.new("testdir")
580 struct dir_data
*dirp
;
584 pos
= telldir(dirp
->dir
);
585 return rb_int2inum(pos
);
593 * dir.seek( integer ) => dir
595 * Seeks to a particular location in <em>dir</em>. <i>integer</i>
596 * must be a value returned by <code>Dir#tell</code>.
598 * d = Dir.new("testdir") #=> #<Dir:0x401b3c40>
602 * d.seek(i) #=> #<Dir:0x401b3c40>
606 dir_seek(VALUE dir
, VALUE pos
)
608 struct dir_data
*dirp
;
609 off_t p
= NUM2OFFT(pos
);
613 seekdir(dirp
->dir
, p
);
622 * dir.pos( integer ) => integer
624 * Synonym for <code>Dir#seek</code>, but returns the position
627 * d = Dir.new("testdir") #=> #<Dir:0x401b3c40>
635 dir_set_pos(VALUE dir
, VALUE pos
)
645 * Repositions <em>dir</em> to the first entry.
647 * d = Dir.new("testdir")
649 * d.rewind #=> #<Dir:0x401b3fb0>
653 dir_rewind(VALUE dir
)
655 struct dir_data
*dirp
;
657 if (rb_safe_level() >= 4 && !OBJ_TAINTED(dir
)) {
658 rb_raise(rb_eSecurityError
, "Insecure: can't close");
661 rewinddir(dirp
->dir
);
669 * Closes the directory stream. Any further attempts to access
670 * <em>dir</em> will raise an <code>IOError</code>.
672 * d = Dir.new("testdir")
678 struct dir_data
*dirp
;
688 dir_chdir(VALUE path
)
690 if (chdir(RSTRING_PTR(path
)) < 0)
691 rb_sys_fail(RSTRING_PTR(path
));
694 static int chdir_blocking
= 0;
695 static VALUE chdir_thread
= Qnil
;
698 VALUE old_path
, new_path
;
703 chdir_yield(struct chdir_data
*args
)
705 dir_chdir(args
->new_path
);
708 if (chdir_thread
== Qnil
)
709 chdir_thread
= rb_thread_current();
710 return rb_yield(args
->new_path
);
714 chdir_restore(struct chdir_data
*args
)
718 if (chdir_blocking
== 0)
720 dir_chdir(args
->old_path
);
727 * Dir.chdir( [ string] ) => 0
728 * Dir.chdir( [ string] ) {| path | block } => anObject
730 * Changes the current working directory of the process to the given
731 * string. When called without an argument, changes the directory to
732 * the value of the environment variable <code>HOME</code>, or
733 * <code>LOGDIR</code>. <code>SystemCallError</code> (probably
734 * <code>Errno::ENOENT</code>) if the target directory does not exist.
736 * If a block is given, it is passed the name of the new current
737 * directory, and the block is executed with that as the current
738 * directory. The original working directory is restored when the block
739 * exits. The return value of <code>chdir</code> is the value of the
740 * block. <code>chdir</code> blocks can be nested, but in a
741 * multi-threaded program an error will be raised if a thread attempts
742 * to open a <code>chdir</code> block while another thread has one
745 * Dir.chdir("/var/spool/mail")
747 * Dir.chdir("/tmp") do
749 * Dir.chdir("/usr") do
765 dir_s_chdir(int argc
, VALUE
*argv
, VALUE obj
)
770 if (rb_scan_args(argc
, argv
, "01", &path
) == 1) {
774 const char *dist
= getenv("HOME");
776 dist
= getenv("LOGDIR");
777 if (!dist
) rb_raise(rb_eArgError
, "HOME/LOGDIR not set");
779 path
= rb_str_new2(dist
);
782 if (chdir_blocking
> 0) {
783 if (!rb_block_given_p() || rb_thread_current() != chdir_thread
)
784 rb_warn("conflicting chdir during another chdir block");
787 if (rb_block_given_p()) {
788 struct chdir_data args
;
789 char *cwd
= my_getcwd();
791 args
.old_path
= rb_tainted_str_new2(cwd
); free(cwd
);
792 args
.new_path
= path
;
794 return rb_ensure(chdir_yield
, (VALUE
)&args
, chdir_restore
, (VALUE
)&args
);
803 * Dir.getwd => string
806 * Returns the path to the current working directory of this process as
809 * Dir.chdir("/tmp") #=> 0
810 * Dir.getwd #=> "/tmp"
813 dir_s_getwd(VALUE dir
)
820 cwd
= rb_tainted_str_new2(path
);
827 check_dirname(volatile VALUE
*dir
)
833 path
= RSTRING_PTR(*dir
);
834 if (path
&& *(pend
= rb_path_end(rb_path_skip_prefix(path
)))) {
835 *dir
= rb_str_new(path
, pend
- path
);
841 * Dir.chroot( string ) => 0
843 * Changes this process's idea of the file system root. Only a
844 * privileged process may make this call. Not available on all
845 * platforms. On Unix systems, see <code>chroot(2)</code> for more
849 dir_s_chroot(VALUE dir
, VALUE path
)
851 #if defined(HAVE_CHROOT) && !defined(__CHECKER__)
852 check_dirname(&path
);
854 if (chroot(RSTRING_PTR(path
)) == -1)
855 rb_sys_fail(RSTRING_PTR(path
));
860 return Qnil
; /* not reached */
866 * Dir.mkdir( string [, integer] ) => 0
868 * Makes a new directory named by <i>string</i>, with permissions
869 * specified by the optional parameter <i>anInteger</i>. The
870 * permissions may be modified by the value of
871 * <code>File::umask</code>, and are ignored on NT. Raises a
872 * <code>SystemCallError</code> if the directory cannot be created. See
873 * also the discussion of permissions in the class documentation for
878 dir_s_mkdir(int argc
, VALUE
*argv
, VALUE obj
)
883 if (rb_scan_args(argc
, argv
, "11", &path
, &vmode
) == 2) {
884 mode
= NUM2INT(vmode
);
890 check_dirname(&path
);
891 if (mkdir(RSTRING_PTR(path
), mode
) == -1)
892 rb_sys_fail(RSTRING_PTR(path
));
899 * Dir.delete( string ) => 0
900 * Dir.rmdir( string ) => 0
901 * Dir.unlink( string ) => 0
903 * Deletes the named directory. Raises a subclass of
904 * <code>SystemCallError</code> if the directory isn't empty.
907 dir_s_rmdir(VALUE obj
, VALUE dir
)
910 if (rmdir(RSTRING_PTR(dir
)) < 0)
911 rb_sys_fail(RSTRING_PTR(dir
));
917 sys_warning_1(const char* mesg
)
919 rb_sys_warning("%s", mesg
);
922 #define GLOB_VERBOSE (1UL << (sizeof(int) * CHAR_BIT - 1))
923 #define sys_warning(val) \
924 (void)((flags & GLOB_VERBOSE) && rb_protect((VALUE (*)(VALUE))sys_warning_1, (VALUE)(val), 0))
926 #define GLOB_ALLOC(type) (type *)malloc(sizeof(type))
927 #define GLOB_ALLOC_N(type, n) (type *)malloc(sizeof(type) * (n))
928 #define GLOB_JUMP_TAG(status) ((status == -1) ? rb_memerror() : rb_jump_tag(status))
931 * ENOTDIR can be returned by stat(2) if a non-leaf element of the path
932 * is not a directory.
934 #define to_be_ignored(e) ((e) == ENOENT || (e) == ENOTDIR)
936 /* System call with warning */
938 do_stat(const char *path
, struct stat
*pst
, int flags
)
941 int ret
= stat(path
, pst
);
942 if (ret
< 0 && !to_be_ignored(errno
))
949 do_lstat(const char *path
, struct stat
*pst
, int flags
)
951 int ret
= lstat(path
, pst
);
952 if (ret
< 0 && !to_be_ignored(errno
))
959 do_opendir(const char *path
, int flags
)
961 DIR *dirp
= opendir(path
);
962 if (dirp
== NULL
&& !to_be_ignored(errno
))
968 /* Return nonzero if S has any special globbing chars in it. */
970 has_magic(const char *s
, int flags
)
972 const int escape
= !(flags
& FNM_NOESCAPE
);
973 const int nocase
= flags
& FNM_CASEFOLD
;
975 register const char *p
= s
;
978 while ((c
= *p
++) != 0) {
986 if (escape
&& !(c
= *p
++))
991 if (!FNM_SYSCASE
&& ISALPHA(c
) && nocase
)
1001 /* Find separator in globbing pattern. */
1003 find_dirsep(const char *s
, int flags
)
1005 const int escape
= !(flags
& FNM_NOESCAPE
);
1007 register const char *p
= s
;
1011 while ((c
= *p
++) != 0) {
1026 if (escape
&& !(c
= *p
++))
1037 /* Remove escaping backslashes */
1039 remove_backslashes(char *p
)
1047 memmove(t
, s
, p
- s
);
1058 memmove(t
, s
, p
- s
); /* move '\0' too */
1061 /* Globing pattern */
1062 enum glob_pattern_type
{ PLAIN
, MAGICAL
, RECURSIVE
, MATCH_ALL
, MATCH_DIR
};
1064 struct glob_pattern
{
1066 enum glob_pattern_type type
;
1067 struct glob_pattern
*next
;
1070 static void glob_free_pattern(struct glob_pattern
*list
);
1072 static struct glob_pattern
*
1073 glob_make_pattern(const char *p
, int flags
)
1075 struct glob_pattern
*list
, *tmp
, **tail
= &list
;
1076 int dirsep
= 0; /* pattern is terminated with '/' */
1079 tmp
= GLOB_ALLOC(struct glob_pattern
);
1080 if (!tmp
) goto error
;
1081 if (p
[0] == '*' && p
[1] == '*' && p
[2] == '/') {
1082 /* fold continuous RECURSIVEs (needed in glob_helper) */
1083 do { p
+= 3; } while (p
[0] == '*' && p
[1] == '*' && p
[2] == '/');
1084 tmp
->type
= RECURSIVE
;
1089 const char *m
= find_dirsep(p
, flags
);
1090 char *buf
= GLOB_ALLOC_N(char, m
-p
+1);
1095 memcpy(buf
, p
, m
-p
);
1097 tmp
->type
= has_magic(buf
, flags
) ? MAGICAL
: PLAIN
;
1112 tmp
= GLOB_ALLOC(struct glob_pattern
);
1116 glob_free_pattern(list
);
1119 tmp
->type
= dirsep
? MATCH_DIR
: MATCH_ALL
;
1128 glob_free_pattern(struct glob_pattern
*list
)
1131 struct glob_pattern
*tmp
= list
;
1140 join_path(const char *path
, int dirsep
, const char *name
)
1142 long len
= strlen(path
);
1143 char *buf
= GLOB_ALLOC_N(char, len
+strlen(name
)+(dirsep
?1:0)+1);
1146 memcpy(buf
, path
, len
);
1148 strcpy(buf
+len
, "/");
1151 strcpy(buf
+len
, name
);
1155 enum answer
{ YES
, NO
, UNKNOWN
};
1158 # define S_ISDIR(m) ((m & S_IFMT) == S_IFDIR)
1163 # define S_ISLNK(m) (0)
1165 # define S_ISLNK(m) ((m & S_IFMT) == S_IFLNK)
1170 void (*func
)(const char *, VALUE
);
1176 glob_func_caller(VALUE val
)
1178 struct glob_args
*args
= (struct glob_args
*)val
;
1180 (*args
->func
)(args
->path
, args
->value
);
1184 #define glob_call_func(func, path, arg) (*func)(path, arg)
1189 int dirsep
, /* '/' should be placed before appending child entry's name to 'path'. */
1190 enum answer exist
, /* Does 'path' indicate an existing entry? */
1191 enum answer isdir
, /* Does 'path' indicate a directory or a symlink to a directory? */
1192 struct glob_pattern
**beg
,
1193 struct glob_pattern
**end
,
1195 ruby_glob_func
*func
,
1200 struct glob_pattern
**cur
, **new_beg
, **new_end
;
1201 int plain
= 0, magical
= 0, recursive
= 0, match_all
= 0, match_dir
= 0;
1202 int escape
= !(flags
& FNM_NOESCAPE
);
1204 for (cur
= beg
; cur
< end
; ++cur
) {
1205 struct glob_pattern
*p
= *cur
;
1206 if (p
->type
== RECURSIVE
) {
1224 rb_bug("continuous RECURSIVEs");
1229 if (match_all
&& exist
== UNKNOWN
) {
1230 if (do_lstat(path
, &st
, flags
) == 0) {
1232 isdir
= S_ISDIR(st
.st_mode
) ? YES
: S_ISLNK(st
.st_mode
) ? UNKNOWN
: NO
;
1239 if (match_dir
&& isdir
== UNKNOWN
) {
1240 if (do_stat(path
, &st
, flags
) == 0) {
1242 isdir
= S_ISDIR(st
.st_mode
) ? YES
: NO
;
1249 if (match_all
&& exist
== YES
) {
1250 status
= glob_call_func(func
, path
, arg
);
1251 if (status
) return status
;
1253 if (match_dir
&& isdir
== YES
) {
1254 char *tmp
= join_path(path
, dirsep
, "");
1255 if (!tmp
) return -1;
1256 status
= glob_call_func(func
, tmp
, arg
);
1258 if (status
) return status
;
1262 if (exist
== NO
|| isdir
== NO
) return 0;
1264 if (magical
|| recursive
) {
1266 DIR *dirp
= do_opendir(*path
? path
: ".", flags
);
1267 if (dirp
== NULL
) return 0;
1269 for (dp
= readdir(dirp
); dp
!= NULL
; dp
= readdir(dirp
)) {
1270 char *buf
= join_path(path
, dirsep
, dp
->d_name
);
1271 enum answer new_isdir
= UNKNOWN
;
1277 if (recursive
&& strcmp(dp
->d_name
, ".") != 0 && strcmp(dp
->d_name
, "..") != 0
1278 && fnmatch("*", dp
->d_name
, flags
) == 0) {
1280 if (do_lstat(buf
, &st
, flags
) == 0)
1281 new_isdir
= S_ISDIR(st
.st_mode
) ? YES
: S_ISLNK(st
.st_mode
) ? UNKNOWN
: NO
;
1285 new_isdir
= dp
->d_isdir
? (!dp
->d_isrep
? YES
: UNKNOWN
) : NO
;
1289 new_beg
= new_end
= GLOB_ALLOC_N(struct glob_pattern
*, (end
- beg
) * 2);
1295 for (cur
= beg
; cur
< end
; ++cur
) {
1296 struct glob_pattern
*p
= *cur
;
1297 if (p
->type
== RECURSIVE
) {
1298 if (new_isdir
== YES
) /* not symlink but real directory */
1299 *new_end
++ = p
; /* append recursive pattern */
1300 p
= p
->next
; /* 0 times recursion */
1302 if (p
->type
== PLAIN
|| p
->type
== MAGICAL
) {
1303 if (fnmatch(p
->str
, dp
->d_name
, flags
) == 0)
1304 *new_end
++ = p
->next
;
1308 status
= glob_helper(buf
, 1, YES
, new_isdir
, new_beg
, new_end
, flags
, func
, arg
);
1317 struct glob_pattern
**copy_beg
, **copy_end
, **cur2
;
1319 copy_beg
= copy_end
= GLOB_ALLOC_N(struct glob_pattern
*, end
- beg
);
1320 if (!copy_beg
) return -1;
1321 for (cur
= beg
; cur
< end
; ++cur
)
1322 *copy_end
++ = (*cur
)->type
== PLAIN
? *cur
: 0;
1324 for (cur
= copy_beg
; cur
< copy_end
; ++cur
) {
1328 name
= GLOB_ALLOC_N(char, strlen((*cur
)->str
) + 1);
1333 strcpy(name
, (*cur
)->str
);
1334 if (escape
) remove_backslashes(name
);
1336 new_beg
= new_end
= GLOB_ALLOC_N(struct glob_pattern
*, end
- beg
);
1342 *new_end
++ = (*cur
)->next
;
1343 for (cur2
= cur
+ 1; cur2
< copy_end
; ++cur2
) {
1344 if (*cur2
&& fnmatch((*cur2
)->str
, name
, flags
) == 0) {
1345 *new_end
++ = (*cur2
)->next
;
1350 buf
= join_path(path
, dirsep
, name
);
1357 status
= glob_helper(buf
, 1, UNKNOWN
, UNKNOWN
, new_beg
, new_end
, flags
, func
, arg
);
1371 ruby_glob0(const char *path
, int flags
, ruby_glob_func
*func
, VALUE arg
)
1373 struct glob_pattern
*list
;
1374 const char *root
, *start
;
1379 start
= root
= path
;
1380 flags
|= FNM_SYSCASE
;
1382 root
= rb_path_skip_prefix(root
);
1385 if (root
&& *root
== '/') root
++;
1388 buf
= GLOB_ALLOC_N(char, n
+ 1);
1389 if (!buf
) return -1;
1390 MEMCPY(buf
, start
, char, n
);
1393 list
= glob_make_pattern(root
, flags
);
1398 status
= glob_helper(buf
, 0, UNKNOWN
, UNKNOWN
, &list
, &list
+ 1, flags
, func
, arg
);
1399 glob_free_pattern(list
);
1406 ruby_glob(const char *path
, int flags
, ruby_glob_func
*func
, VALUE arg
)
1408 return ruby_glob0(path
, flags
& ~GLOB_VERBOSE
, func
, arg
);
1412 rb_glob_caller(const char *path
, VALUE a
)
1415 struct glob_args
*args
= (struct glob_args
*)a
;
1418 rb_protect(glob_func_caller
, a
, &status
);
1423 rb_glob2(const char *path
, int flags
, void (*func
)(const char *, VALUE
), VALUE arg
)
1425 struct glob_args args
;
1430 if (flags
& FNM_SYSCASE
) {
1431 rb_warning("Dir.glob() ignores File::FNM_CASEFOLD");
1434 return ruby_glob0(path
, flags
| GLOB_VERBOSE
, rb_glob_caller
, (VALUE
)&args
);
1438 rb_glob(const char *path
, void (*func
)(const char *, VALUE
), VALUE arg
)
1440 int status
= rb_glob2(path
, 0, func
, arg
);
1441 if (status
) GLOB_JUMP_TAG(status
);
1445 push_pattern(const char *path
, VALUE ary
)
1447 rb_ary_push(ary
, rb_tainted_str_new2(path
));
1451 ruby_brace_expand(const char *str
, int flags
, ruby_glob_func
*func
, VALUE arg
)
1453 const int escape
= !(flags
& FNM_NOESCAPE
);
1454 const char *p
= str
;
1456 const char *lbrace
= 0, *rbrace
= 0;
1457 int nest
= 0, status
= 0;
1460 if (*p
== '{' && nest
++ == 0) {
1463 if (*p
== '}' && --nest
<= 0) {
1467 if (*p
== '\\' && escape
) {
1473 if (lbrace
&& rbrace
) {
1474 char *buf
= GLOB_ALLOC_N(char, strlen(s
) + 1);
1477 if (!buf
) return -1;
1478 memcpy(buf
, s
, lbrace
-s
);
1481 while (p
< rbrace
) {
1482 const char *t
= ++p
;
1484 while (p
< rbrace
&& !(*p
== ',' && nest
== 0)) {
1485 if (*p
== '{') nest
++;
1486 if (*p
== '}') nest
--;
1487 if (*p
== '\\' && escape
) {
1488 if (++p
== rbrace
) break;
1492 memcpy(buf
+shift
, t
, p
-t
);
1493 strcpy(buf
+shift
+(p
-t
), rbrace
+1);
1494 status
= ruby_brace_expand(buf
, flags
, func
, arg
);
1499 else if (!lbrace
&& !rbrace
) {
1500 status
= (*func
)(s
, arg
);
1507 ruby_glob_func
*func
;
1513 glob_brace(const char *path
, VALUE val
)
1515 struct brace_args
*arg
= (struct brace_args
*)val
;
1517 return ruby_glob0(path
, arg
->flags
, arg
->func
, arg
->value
);
1521 ruby_brace_glob0(const char *str
, int flags
, ruby_glob_func
*func
, VALUE arg
)
1523 struct brace_args args
;
1528 return ruby_brace_expand(str
, flags
, glob_brace
, (VALUE
)&args
);
1532 ruby_brace_glob(const char *str
, int flags
, ruby_glob_func
*func
, VALUE arg
)
1534 return ruby_brace_glob0(str
, flags
& ~GLOB_VERBOSE
, func
, arg
);
1538 push_glob(VALUE ary
, const char *str
, int flags
)
1540 struct glob_args args
;
1542 args
.func
= push_pattern
;
1544 return ruby_brace_glob0(str
, flags
| GLOB_VERBOSE
, rb_glob_caller
, (VALUE
)&args
);
1548 rb_push_glob(VALUE str
, int flags
) /* '\0' is delimiter */
1556 while (offset
< RSTRING_LEN(str
)) {
1557 int status
= push_glob(ary
, RSTRING_PTR(str
) + offset
, flags
);
1559 if (status
) GLOB_JUMP_TAG(status
);
1560 if (offset
>= RSTRING_LEN(str
)) break;
1561 p
= RSTRING_PTR(str
) + offset
;
1563 pend
= RSTRING_PTR(str
) + RSTRING_LEN(str
);
1564 while (p
< pend
&& !*p
)
1566 offset
= p
- RSTRING_PTR(str
);
1573 dir_globs(long argc
, VALUE
*argv
, int flags
)
1575 VALUE ary
= rb_ary_new();
1578 for (i
= 0; i
< argc
; ++i
) {
1580 VALUE str
= argv
[i
];
1582 status
= push_glob(ary
, RSTRING_PTR(str
), flags
);
1583 if (status
) GLOB_JUMP_TAG(status
);
1591 * Dir[ array ] => array
1592 * Dir[ string [, string ...] ] => array
1594 * Equivalent to calling
1595 * <code>Dir.glob(</code><i>array,</i><code>0)</code> and
1596 * <code>Dir.glob([</code><i>string,...</i><code>],0)</code>.
1600 dir_s_aref(int argc
, VALUE
*argv
, VALUE obj
)
1603 return rb_push_glob(argv
[0], 0);
1605 return dir_globs(argc
, argv
, 0);
1610 * Dir.glob( pattern, [flags] ) => array
1611 * Dir.glob( pattern, [flags] ) {| filename | block } => nil
1613 * Returns the filenames found by expanding <i>pattern</i> which is
1614 * an +Array+ of the patterns or the pattern +String+, either as an
1615 * <i>array</i> or as parameters to the block. Note that this pattern
1616 * is not a regexp (it's closer to a shell glob). See
1617 * <code>File::fnmatch</code> for the meaning of the <i>flags</i>
1618 * parameter. Note that case sensitivity depends on your system (so
1619 * <code>File::FNM_CASEFOLD</code> is ignored)
1621 * <code>*</code>:: Matches any file. Can be restricted by
1622 * other values in the glob. <code>*</code>
1623 * will match all files; <code>c*</code> will
1624 * match all files beginning with
1625 * <code>c</code>; <code>*c</code> will match
1626 * all files ending with <code>c</code>; and
1627 * <code>*c*</code> will match all files that
1628 * have <code>c</code> in them (including at
1629 * the beginning or end). Equivalent to
1630 * <code>/ .* /x</code> in regexp.
1631 * <code>**</code>:: Matches directories recursively.
1632 * <code>?</code>:: Matches any one character. Equivalent to
1633 * <code>/.{1}/</code> in regexp.
1634 * <code>[set]</code>:: Matches any one character in +set+.
1635 * Behaves exactly like character sets in
1636 * Regexp, including set negation
1637 * (<code>[^a-z]</code>).
1638 * <code>{p,q}</code>:: Matches either literal <code>p</code> or
1639 * literal <code>q</code>. Matching literals
1640 * may be more than one character in length.
1641 * More than two literals may be specified.
1642 * Equivalent to pattern alternation in
1644 * <code>\</code>:: Escapes the next metacharacter.
1646 * Dir["config.?"] #=> ["config.h"]
1647 * Dir.glob("config.?") #=> ["config.h"]
1648 * Dir.glob("*.[a-z][a-z]") #=> ["main.rb"]
1649 * Dir.glob("*.[^r]*") #=> ["config.h"]
1650 * Dir.glob("*.{rb,h}") #=> ["main.rb", "config.h"]
1651 * Dir.glob("*") #=> ["config.h", "main.rb"]
1652 * Dir.glob("*", File::FNM_DOTMATCH) #=> [".", "..", "config.h", "main.rb"]
1654 * rbfiles = File.join("**", "*.rb")
1655 * Dir.glob(rbfiles) #=> ["main.rb",
1657 * # "lib/song/karaoke.rb"]
1658 * libdirs = File.join("**", "lib")
1659 * Dir.glob(libdirs) #=> ["lib"]
1661 * librbfiles = File.join("**", "lib", "**", "*.rb")
1662 * Dir.glob(librbfiles) #=> ["lib/song.rb",
1663 * # "lib/song/karaoke.rb"]
1665 * librbfiles = File.join("**", "lib", "*.rb")
1666 * Dir.glob(librbfiles) #=> ["lib/song.rb"]
1669 dir_s_glob(int argc
, VALUE
*argv
, VALUE obj
)
1671 VALUE str
, rflags
, ary
;
1674 if (rb_scan_args(argc
, argv
, "11", &str
, &rflags
) == 2)
1675 flags
= NUM2INT(rflags
);
1679 ary
= rb_check_array_type(str
);
1681 ary
= rb_push_glob(str
, flags
);
1684 volatile VALUE v
= ary
;
1685 ary
= dir_globs(RARRAY_LEN(v
), RARRAY_PTR(v
), flags
);
1688 if (rb_block_given_p()) {
1696 dir_open_dir(VALUE path
)
1698 VALUE dir
= rb_funcall(rb_cDir
, rb_intern("open"), 1, path
);
1700 if (TYPE(dir
) != T_DATA
||
1701 RDATA(dir
)->dfree
!= (RUBY_DATA_FUNC
)free_dir
) {
1702 rb_raise(rb_eTypeError
, "wrong argument type %s (expected Dir)",
1703 rb_obj_classname(dir
));
1711 * Dir.foreach( dirname ) {| filename | block } => nil
1713 * Calls the block once for each entry in the named directory, passing
1714 * the filename of each entry as a parameter to the block.
1716 * Dir.foreach("testdir") {|x| puts "Got #{x}" }
1718 * <em>produces:</em>
1727 dir_foreach(VALUE io
, VALUE dirname
)
1731 RETURN_ENUMERATOR(io
, 1, &dirname
);
1732 dir
= dir_open_dir(dirname
);
1733 rb_ensure(dir_each
, dir
, dir_close
, dir
);
1739 * Dir.entries( dirname ) => array
1741 * Returns an array containing all of the filenames in the given
1742 * directory. Will raise a <code>SystemCallError</code> if the named
1743 * directory doesn't exist.
1745 * Dir.entries("testdir") #=> [".", "..", "config.h", "main.rb"]
1749 dir_entries(VALUE io
, VALUE dirname
)
1753 dir
= dir_open_dir(dirname
);
1754 return rb_ensure(rb_Array
, dir
, dir_close
, dir
);
1759 * File.fnmatch( pattern, path, [flags] ) => (true or false)
1760 * File.fnmatch?( pattern, path, [flags] ) => (true or false)
1762 * Returns true if <i>path</i> matches against <i>pattern</i> The
1763 * pattern is not a regular expression; instead it follows rules
1764 * similar to shell filename globbing. It may contain the following
1767 * <code>*</code>:: Matches any file. Can be restricted by
1768 * other values in the glob. <code>*</code>
1769 * will match all files; <code>c*</code> will
1770 * match all files beginning with
1771 * <code>c</code>; <code>*c</code> will match
1772 * all files ending with <code>c</code>; and
1773 * <code>*c*</code> will match all files that
1774 * have <code>c</code> in them (including at
1775 * the beginning or end). Equivalent to
1776 * <code>/ .* /x</code> in regexp.
1777 * <code>**</code>:: Matches directories recursively or files
1779 * <code>?</code>:: Matches any one character. Equivalent to
1780 * <code>/.{1}/</code> in regexp.
1781 * <code>[set]</code>:: Matches any one character in +set+.
1782 * Behaves exactly like character sets in
1783 * Regexp, including set negation
1784 * (<code>[^a-z]</code>).
1785 * <code>\</code>:: Escapes the next metacharacter.
1787 * <i>flags</i> is a bitwise OR of the <code>FNM_xxx</code>
1788 * parameters. The same glob pattern and flags are used by
1789 * <code>Dir::glob</code>.
1791 * File.fnmatch('cat', 'cat') #=> true # match entire string
1792 * File.fnmatch('cat', 'category') #=> false # only match partial string
1793 * File.fnmatch('c{at,ub}s', 'cats') #=> false # { } isn't supported
1795 * File.fnmatch('c?t', 'cat') #=> true # '?' match only 1 character
1796 * File.fnmatch('c??t', 'cat') #=> false # ditto
1797 * File.fnmatch('c*', 'cats') #=> true # '*' match 0 or more characters
1798 * File.fnmatch('c*t', 'c/a/b/t') #=> true # ditto
1799 * File.fnmatch('ca[a-z]', 'cat') #=> true # inclusive bracket expression
1800 * File.fnmatch('ca[^t]', 'cat') #=> false # exclusive bracket expression ('^' or '!')
1802 * File.fnmatch('cat', 'CAT') #=> false # case sensitive
1803 * File.fnmatch('cat', 'CAT', File::FNM_CASEFOLD) #=> true # case insensitive
1805 * File.fnmatch('?', '/', File::FNM_PATHNAME) #=> false # wildcard doesn't match '/' on FNM_PATHNAME
1806 * File.fnmatch('*', '/', File::FNM_PATHNAME) #=> false # ditto
1807 * File.fnmatch('[/]', '/', File::FNM_PATHNAME) #=> false # ditto
1809 * File.fnmatch('\?', '?') #=> true # escaped wildcard becomes ordinary
1810 * File.fnmatch('\a', 'a') #=> true # escaped ordinary remains ordinary
1811 * File.fnmatch('\a', '\a', File::FNM_NOESCAPE) #=> true # FNM_NOESACPE makes '\' ordinary
1812 * File.fnmatch('[\?]', '?') #=> true # can escape inside bracket expression
1814 * File.fnmatch('*', '.profile') #=> false # wildcard doesn't match leading
1815 * File.fnmatch('*', '.profile', File::FNM_DOTMATCH) #=> true # period by default.
1816 * File.fnmatch('.*', '.profile') #=> true
1818 * rbfiles = '**' '/' '*.rb' # you don't have to do like this. just write in single string.
1819 * File.fnmatch(rbfiles, 'main.rb') #=> false
1820 * File.fnmatch(rbfiles, './main.rb') #=> false
1821 * File.fnmatch(rbfiles, 'lib/song.rb') #=> true
1822 * File.fnmatch('**.rb', 'main.rb') #=> true
1823 * File.fnmatch('**.rb', './main.rb') #=> false
1824 * File.fnmatch('**.rb', 'lib/song.rb') #=> true
1825 * File.fnmatch('*', 'dave/.profile') #=> true
1827 * pattern = '*' '/' '*'
1828 * File.fnmatch(pattern, 'dave/.profile', File::FNM_PATHNAME) #=> false
1829 * File.fnmatch(pattern, 'dave/.profile', File::FNM_PATHNAME | File::FNM_DOTMATCH) #=> true
1831 * pattern = '**' '/' 'foo'
1832 * File.fnmatch(pattern, 'a/b/c/foo', File::FNM_PATHNAME) #=> true
1833 * File.fnmatch(pattern, '/a/b/c/foo', File::FNM_PATHNAME) #=> true
1834 * File.fnmatch(pattern, 'c:/a/b/c/foo', File::FNM_PATHNAME) #=> true
1835 * File.fnmatch(pattern, 'a/.b/c/foo', File::FNM_PATHNAME) #=> false
1836 * File.fnmatch(pattern, 'a/.b/c/foo', File::FNM_PATHNAME | File::FNM_DOTMATCH) #=> true
1839 file_s_fnmatch(int argc
, VALUE
*argv
, VALUE obj
)
1841 VALUE pattern
, path
;
1845 if (rb_scan_args(argc
, argv
, "21", &pattern
, &path
, &rflags
) == 3)
1846 flags
= NUM2INT(rflags
);
1850 StringValue(pattern
);
1851 FilePathStringValue(path
);
1853 if (fnmatch(RSTRING_PTR(pattern
), RSTRING_PTR(path
), flags
) == 0)
1860 * Objects of class <code>Dir</code> are directory streams representing
1861 * directories in the underlying file system. They provide a variety of
1862 * ways to list directories and their contents. See also
1863 * <code>File</code>.
1865 * The directory used in these examples contains the two regular files
1866 * (<code>config.h</code> and <code>main.rb</code>), the parent
1867 * directory (<code>..</code>), and the directory itself
1873 rb_cDir
= rb_define_class("Dir", rb_cObject
);
1875 rb_include_module(rb_cDir
, rb_mEnumerable
);
1877 rb_define_alloc_func(rb_cDir
, dir_s_alloc
);
1878 rb_define_singleton_method(rb_cDir
, "open", dir_s_open
, 1);
1879 rb_define_singleton_method(rb_cDir
, "foreach", dir_foreach
, 1);
1880 rb_define_singleton_method(rb_cDir
, "entries", dir_entries
, 1);
1882 rb_define_method(rb_cDir
,"initialize", dir_initialize
, 1);
1883 rb_define_method(rb_cDir
,"path", dir_path
, 0);
1884 rb_define_method(rb_cDir
,"inspect", dir_inspect
, 0);
1885 rb_define_method(rb_cDir
,"read", dir_read
, 0);
1886 rb_define_method(rb_cDir
,"each", dir_each
, 0);
1887 rb_define_method(rb_cDir
,"rewind", dir_rewind
, 0);
1888 rb_define_method(rb_cDir
,"tell", dir_tell
, 0);
1889 rb_define_method(rb_cDir
,"seek", dir_seek
, 1);
1890 rb_define_method(rb_cDir
,"pos", dir_tell
, 0);
1891 rb_define_method(rb_cDir
,"pos=", dir_set_pos
, 1);
1892 rb_define_method(rb_cDir
,"close", dir_close
, 0);
1894 rb_define_singleton_method(rb_cDir
,"chdir", dir_s_chdir
, -1);
1895 rb_define_singleton_method(rb_cDir
,"getwd", dir_s_getwd
, 0);
1896 rb_define_singleton_method(rb_cDir
,"pwd", dir_s_getwd
, 0);
1897 rb_define_singleton_method(rb_cDir
,"chroot", dir_s_chroot
, 1);
1898 rb_define_singleton_method(rb_cDir
,"mkdir", dir_s_mkdir
, -1);
1899 rb_define_singleton_method(rb_cDir
,"rmdir", dir_s_rmdir
, 1);
1900 rb_define_singleton_method(rb_cDir
,"delete", dir_s_rmdir
, 1);
1901 rb_define_singleton_method(rb_cDir
,"unlink", dir_s_rmdir
, 1);
1903 rb_define_singleton_method(rb_cDir
,"glob", dir_s_glob
, -1);
1904 rb_define_singleton_method(rb_cDir
,"[]", dir_s_aref
, -1);
1905 rb_define_singleton_method(rb_cDir
,"exist?", rb_file_directory_p
, 1); /* in file.c */
1906 rb_define_singleton_method(rb_cDir
,"exists?", rb_file_directory_p
, 1); /* in file.c */
1908 rb_define_singleton_method(rb_cFile
,"fnmatch", file_s_fnmatch
, -1);
1909 rb_define_singleton_method(rb_cFile
,"fnmatch?", file_s_fnmatch
, -1);
1911 rb_file_const("FNM_NOESCAPE", INT2FIX(FNM_NOESCAPE
));
1912 rb_file_const("FNM_PATHNAME", INT2FIX(FNM_PATHNAME
));
1913 rb_file_const("FNM_DOTMATCH", INT2FIX(FNM_DOTMATCH
));
1914 rb_file_const("FNM_CASEFOLD", INT2FIX(FNM_CASEFOLD
));
1915 rb_file_const("FNM_SYSCASE", INT2FIX(FNM_SYSCASE
));