[ruby-dev:23010] Re: File.fnmatchとDir.glob

山本です。

>こんにちは、なかむら(う)です。
>
>ぜんぜん中身と関係ない話で恐縮ですが、
>
>できれば -p オプションもつけていただけないでしょうか?
>あると格段に追いやすさが違いますので。

つけてみました。


cvs diff -u -wb -p dir.c (in directory E:\ruby-cvs\ruby\)
Index: dir.c
===================================================================
RCS file: /ruby/ruby/dir.c,v
retrieving revision 1.107
diff -u -w -b -p -r1.107 dir.c
--- dir.c	16 Feb 2004 01:33:12 -0000	1.107
+++ dir.c	23 Feb 2004 05:38:42 -0000
@@ -68,7 +68,7 @@ char *strchr _((char*,char));
 #endif
 
 #define FNM_NOESCAPE	0x01
-#define FNM_PATHNAME	0x02
+#define FNM_SEPMATCH	0x02
 #define FNM_DOTMATCH	0x04
 #define FNM_CASEFOLD	0x08
 
@@ -164,143 +164,151 @@ CompareImpl(p1, p2, nocase)
 }
 #endif /* environment */
 
-#if defined DOSISH
-#define isdirsep(c) ((c) == '/' || (c) == '\\')
-#else
-#define isdirsep(c) ((c) == '/')
-#endif
-
-static char *
-range(pat, test, flags)
-    char *pat;
-    char *test;
-    int flags;
+static int
+bracket(
+    const char *p,
+    const char *pend,
+    const char *test,
+    int flags)
 {
-    int not, ok = 0;
+    int ok = 1;
     int nocase = flags & FNM_CASEFOLD;
-    int escape = !(flags & FNM_NOESCAPE);
 
-    not = *pat == '!' || *pat == '^';
-    if (not)
-	pat++;
-
-    while (*pat) {
-	char *pstart, *pend;
-	pstart = pend = pat;
-	if (*pstart == ']')
-	    return ok == not ? 0 : ++pat;
-	else if (escape && *pstart == '\\')
-	    pstart = pend = ++pat;
-	Inc(pat);
-	if (*pat == '-' && pat[1] != ']') {
-	    if (escape && pat[1] == '\\')
-		pat++;
-	    pend = pat+1;
-	    if (!*pend)
-		return 0;
-	    pat = Next(pend);
+    if (p < pend && (*p == '!' || *p == '^')) {
+	ok = 0;
+	p++;
 	}
-	if (Compare(pstart, test) <= 0 && Compare(test, pend) <= 0)
-	    ok = 1;
+
+    while (p < pend) {
+	const char *t1 = p;
+	Inc(p);
+	if (p < pend && *p == '-') {
+	    const char *t2 = p + 1;
+	    if (t2 < pend) /* range */
+		if (Compare(t1, test) <= 0 && Compare(test, t2) <= 0)
+		    return ok;
+		else {
+		    p = Next(t2);
+		    if (p < pend && *p == '-') /* behavoir is undefined in POSIX.2 */
+			p = t2;
+		    continue;
     }
-    return 0;
+	}
+	if (Compare(t1, test) == 0)
+	    return ok;
+    }
+    return !ok;
 }
 
-#define ISDIRSEP(c) (pathname && isdirsep(c))
-#define PERIOD_S() (period && *s == '.' && \
-    (!s_prev || ISDIRSEP(*s_prev)))
-#define INC_S() (s = Next(s_prev = s))
 static int
-fnmatch(pat, string, flags)
-    const char *pat;
-    const char *string;
-    int flags;
+fnmatch_helper(
+    const char *p, /* pattern */
+    const char *s, /* string */
+    int leading, /* 's' is leading. */
+    int flags)
 {
-    int c;
-    const char *test;
-    const char *s = string, *s_prev = 0;
     int escape = !(flags & FNM_NOESCAPE);
-    int pathname = flags & FNM_PATHNAME;
     int period = !(flags & FNM_DOTMATCH);
     int nocase = flags & FNM_CASEFOLD;
+    int pathname = !(flags & FNM_SEPMATCH);
 
-    while (c = *pat) {
-	switch (c) {
+    while (*p) {
+	switch (*p) {
 	  case '?':
-	    if (!*s || ISDIRSEP(*s) || PERIOD_S())
+	    if (period && (leading && *s == '.')) /* leading period */
 		return FNM_NOMATCH;
-	    INC_S();
-	    ++pat;
-	    break;
+	    if (!*s || (pathname && *s == '/'))
+		return FNM_NOMATCH;
+	    p++;
+	    leading = 0;
+	    Inc(s);
+	    continue;
 
 	  case '*':
-	    while ((c = *++pat) == '*')
-		;
-
-	    if (PERIOD_S())
+	    if (period && (leading && *s == '.')) /* leading period */
 		return FNM_NOMATCH;
 
-	    if (!c) {
-		if (pathname && *rb_path_next(s))
-		    return FNM_NOMATCH;
+	    do { p++; } while (*p == '*');
+
+	    if (!*p) {
+		if (pathname) {
+		    while (*s && *s != '/') Inc(s);
+		    return !*s ? 0 : FNM_NOMATCH;
+		}
 		else
 		    return 0;
 	    }
-	    else if (ISDIRSEP(c)) {
-		s = rb_path_next(s);
-		if (*s) {
-		    INC_S();
-		    break;
+
+	    if (pathname && *p == '/') {
+		while (*s && *s != '/') Inc(s);
+		if (*s == '/') {
+		    p++;
+		    leading = 1;
+		    s++;
+		    continue;
                 }
+		else
 		return FNM_NOMATCH;
 	    }
 
-	    test = escape && c == '\\' ? pat+1 : pat;
-	    while (*s) {
-		if ((c == '?' || c == '[' || Compare(s, test) == 0) &&
-		    !fnmatch(pat, s, flags | FNM_DOTMATCH))
+	    while (*s && !(pathname && *s == '/')) {
+		if (0 == fnmatch_helper(p, s, leading, flags))
 		    return 0;
-		else if (ISDIRSEP(*s))
-		    break;
-		INC_S();
+		leading = 0;
+		Inc(s);
 	    }
 	    return FNM_NOMATCH;
 
-	  case '[':
-	    if (!*s || ISDIRSEP(*s) || PERIOD_S())
+	  case '[': {
+	    const char *pend = ++p;
+	    if (period && (leading && *s == '.')) /* leading period */
 		return FNM_NOMATCH;
-	    pat = range(pat+1, s, flags);
-	    if (!pat)
+	    if (!*s || (pathname && *s == '/'))
 		return FNM_NOMATCH;
-	    INC_S();
-	    break;
+	    if (*pend == ']') /* bracket expression includes ']' */
+		pend++;
+	    while (*pend && !(pathname && *pend == '/') && *pend != ']')
+		Inc(pend);
+	    if (*pend == ']') { /* bracket expression */
+		if (bracket(p, pend, s, flags)) {
+		    p = pend + 1;
+		    leading = 0;
+		    Inc(s);
+		    continue;
+		}
+		else
+		    return FNM_NOMATCH;
+	    }
+	    p--; /* treat '[' as ordinary character */
+	    break; /* goto ordinary */
+	  }
 
 	  case '\\':
-	    if (escape && pat[1]
-#if defined DOSISH
-		&& strchr("*?[]\\", pat[1])
-#endif
-		) {
-		c = *++pat;
+	    if (escape && p[1])
+		p++;
+	    break; /* goto ordinary */
 	    }
-	    /* FALLTHROUGH */
 
-	  default:
-#if defined DOSISH
-	    if (ISDIRSEP(c) && isdirsep(*s))
-		;
-	    else
-#endif
-	    if (Compare(pat, s) != 0)
+	/* ordinary */
+
+	if (Compare(p, s) != 0)
 		return FNM_NOMATCH;
-	    INC_S();
-	    Inc(pat);
-	    break;
-	}
+	Inc(p);
+	leading = pathname && *s == '/';
+	Inc(s);
     }
     return !*s ? 0 : FNM_NOMATCH;
 }
 
+static int
+fnmatch(
+    const char *p, /* pattern */
+    const char *s, /* string */
+    int flags)
+{
+    return fnmatch_helper(p, s, 1, flags);
+}
+
 VALUE rb_cDir;
 
 struct dir_data {
@@ -919,15 +927,18 @@ has_magic(s, flags)
 	    return 1;
 
 	  case '[':	/* Only accept an open brace if there is a close */
-	    open++;	/* brace to match it.  Bracket expressions must be */
-	    continue;	/* complete, according to Posix.2 */
+	    open = 1;		/* brace to match it.  Bracket expressions must be */
+	    if (*p == ']')	/* complete, according to Posix.2 */
+		p++;
+	    continue;
+
 	  case ']':
 	    if (open)
 		return 1;
 	    continue;
 
 	  case '\\':
-	    if (escape && !(c = *p++))
+	    if (!open && escape && !(c = *p++))	/* escape is disabled inside bracket */
 		return 0;
 	    continue;
 	}
@@ -1272,17 +1283,15 @@ rb_glob2(path, flags, func, arg)
     int n;
     int status;
 
-    if (flags & FNM_CASEFOLD) {
-	rb_warn("Dir.glob() ignores File::FNM_CASEFOLD");
-    }
+    if (flags & FNM_CASEFOLD)
+	rb_raise(rb_eArgError, "File::FNM_CASEFOLD cannot be used. case sensitivities depends on system.");
+    if (flags & FNM_SEPMATCH)
+	rb_raise(rb_eArgError, "File::FNM_SEPMATCH cannot be used.");
 
 #if defined DOSISH
     flags |= FNM_CASEFOLD;
     root = rb_path_skip_prefix(root);
-#else
-    flags &= ~FNM_CASEFOLD;
 #endif
-
     if (*root == '/') root++;
 
     n = root - path;
@@ -1652,7 +1661,7 @@ Init_Dir()
     rb_define_singleton_method(rb_cFile,"fnmatch?", file_s_fnmatch, -1);
 
     rb_file_const("FNM_NOESCAPE", INT2FIX(FNM_NOESCAPE));
-    rb_file_const("FNM_PATHNAME", INT2FIX(FNM_PATHNAME));
+    rb_file_const("FNM_SEPMATCH", INT2FIX(FNM_SEPMATCH));
     rb_file_const("FNM_DOTMATCH", INT2FIX(FNM_DOTMATCH));
     rb_file_const("FNM_CASEFOLD", INT2FIX(FNM_CASEFOLD));
 }