naruse	2009-10-12 03:49:41 +0900 (Mon, 12 Oct 2009)

  New Revision: 25303

  http://svn.ruby-lang.org/cgi-bin/viewvc.cgi?view=rev&revision=25303

  Log:
    * marshal.c (has_encoding): added for check the regexp
      is dumped by 1.8 or 1.9.
    
    * marshal.c (r_object0): use has_encoding.

  Modified files:
    trunk/ChangeLog
    trunk/marshal.c
    trunk/test/ruby/test_marshal.rb

Index: ChangeLog
===================================================================
--- ChangeLog	(revision 25302)
+++ ChangeLog	(revision 25303)
@@ -1,3 +1,10 @@
+Mon Oct 12 03:47:42 2009  NARUSE, Yui  <naruse@r...>
+
+	* marshal.c (has_encoding): added for check the regexp
+	  is dumped by 1.8 or 1.9.
+
+	* marshal.c (r_object0): use has_encoding.
+
 Sun Oct 11 15:54:52 2009  Nobuyoshi Nakada  <nobu@r...>
 
 	* ext/gdbm/gdbm.c (fgdbm_select): fixed rdoc.  a patch from Justin
Index: marshal.c
===================================================================
--- marshal.c	(revision 25302)
+++ marshal.c	(revision 25303)
@@ -1304,6 +1304,35 @@
 
 #define div0(x) ruby_div0(x)
 
+static int
+has_encoding(struct load_arg *arg)
+{
+    int res = FALSE;
+    long offset = arg->offset;
+    r_long(arg);
+    switch (r_byte(arg)) {
+      case ':':
+	switch (r_byte(arg)) {
+	  case 6:
+	    if (r_byte(arg) == 'E') res = TRUE;
+	    break;
+	  case 13:
+	    if (r_byte(arg) == 'e') res = TRUE;
+	    break;
+	}
+	break;
+      case ';':
+	{
+	    ID id = r_symlink(arg);
+	    if (id == rb_intern("E") || id == rb_id_encoding())
+		res = TRUE;
+	}
+	break;
+    }
+    arg->offset = offset;
+    return res;
+}
+
 static VALUE
 r_object0(struct load_arg *arg, int *ivp, VALUE extmod)
 {
@@ -1474,21 +1503,17 @@
 	{
 	    volatile VALUE str = r_bytes(arg);
 	    int options = r_byte(arg);
-	    VALUE args[2];
-	    args[0] = str;
-	    args[1] = INT2FIX(options);
-	    v = r_entry(rb_obj_alloc(rb_cRegexp), arg);
-	    if (ivp) {
-		r_ivar(v, arg);
-		*ivp = FALSE;
+
+	    if (!ivp || !has_encoding(arg)) {
+		VALUE pat;
+		VALUE dst;
+		char *rsrc =
+		    "(?<!\\\\)((?:\\\\\\\\)*)\\\\([ghijklmopquyEFHIJKLNOPQRSTUVXY])";
+		pat = rb_reg_new(rsrc, strlen(rsrc), 0);
+		dst = rb_usascii_str_new_cstr("\\1\\2");
+		rb_funcall(str, rb_intern("gsub!"), 2, pat, dst);
 	    }
-	    if (rb_enc_get_index(v) != rb_usascii_encindex())
-		rb_enc_copy(str, v);
-	    if (rb_enc_get_index(str) != rb_utf8_encindex()) {
-#define f_gsub_bang(x,y,z) rb_funcall(x, rb_intern("gsub!"), 2, y, z)
-		f_gsub_bang(str, rb_reg_new("\\\\u", 3, 0), rb_usascii_str_new_cstr("u"));
-	    }
-	    rb_obj_call_init(v, 2, args);
+	    v = r_entry(rb_reg_new_str(str, options), arg);
 	    v = r_leave(v, arg);
 	}
 	break;
Index: test/ruby/test_marshal.rb
===================================================================
--- test/ruby/test_marshal.rb	(revision 25302)
+++ test/ruby/test_marshal.rb	(revision 25303)
@@ -206,11 +206,15 @@
   end
 
   def test_regexp
+    assert_equal(/\\u/, Marshal.load("\004\b/\b\\\\u\000"))
+    assert_equal(/u/, Marshal.load("\004\b/\a\\u\000"))
+    assert_equal(/u/, Marshal.load("\004\bI/\a\\u\000\006:\016@encoding\"\vEUC-JP"))
+
     bug2109 = '[ruby-core:25625]'
     a = "\x82\xa0".force_encoding(Encoding::Windows_31J)
     b = "\x82\xa2".force_encoding(Encoding::Windows_31J)
     c = [/#{a}/, /#{b}/]
-    assert_equal(c, Marshal.load(Marshal.dump(c)))
+    assert_equal(c, Marshal.load(Marshal.dump(c)), bug2109)
   end
 
   class DumpTest

--
ML: ruby-changes@q...
Info: http://www.atdot.net/~ko1/quickml/