rdoc: non-binary multibyte

* lib/rdoc/parser.rb (RDoc.binary?): binary read data may have
  incomplete multibyte sequence.  [ruby-core:44798][Bug #6393]

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@35531 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

ChangeLog

@@ -1,3 +1,8 @@
+Fri May  4 01:31:19 2012  Nobuyoshi Nakada  <nobu@ruby-lang.org>
+
+	* lib/rdoc/parser.rb (RDoc.binary?): binary read data may have
+	  incomplete multibyte sequence.  [ruby-core:44798][Bug #6393]
+
 Wed May  2 23:55:51 2012  Nobuyoshi Nakada  <nobu@ruby-lang.org>
 
 	* lib/test/unit.rb (Test::Unit::RequireFiles#non_options): expand

lib/rdoc/parser.rb

@@ -80,14 +80,14 @@ class RDoc::Parser
 
     have_encoding = s.respond_to? :encoding
 
-    if have_encoding then
-      return false if s.encoding != Encoding::ASCII_8BIT and s.valid_encoding?
-    end
-
     return true if s[0, 2] == Marshal.dump('')[0, 2] or s.index("\x00")
 
     if have_encoding then
-      s.force_encoding Encoding.default_external
+      mode = "r"
+      s.sub!(/\A#!.*\n/, '')     # assume shebang line isn't longer than 1024.
+      encoding = s[/^\s*\#\s*(?:-\*-\s*)?(?:en)?coding:\s*\([^\s;]+?\)(?:-\*-)?/, 1]
+      mode = "r:#{encoding}" if encoding
+      s = File.open(file, mode) {|f| f.gets(nil, 1024)}
 
       not s.valid_encoding?
     else

test/rdoc/test_rdoc_parser.rb

@@ -29,7 +29,7 @@ class TestRDocParser < MiniTest::Unit::TestCase
   end
 
   def test_class_binary_large_japanese_rdoc
-    file_name = File.expand_path '../test.ja.large.rdoc', __FILE__
+    file_name = File.expand_path '../test.ja.largedoc', __FILE__
     assert !@RP.binary?(file_name)
   end