* enc/trans/gbk.trans, gbk-tbl.rb:

new Chinese GBK transcoding (from Yoshihiro Kambayashi)

* test/ruby/test_transcode.rb: added tests for the above
  (from Yoshihiro Kambayashi)


git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@21315 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

ChangeLog

@@ -1,4 +1,12 @@
-Sun Jan  4 17:39:39 2009  Martin Duerst  <duerst@it.aoyama.ac.jp>
+Sun Jan  4 18:10:10 2009  Martin Duerst  <duerst@it.aoyama.ac.jp>
+
+	* enc/trans/gbk.trans, gbk-tbl.rb:
+	  new Chinese GBK transcoding (from Yoshihiro Kambayashi)
+
+	* test/ruby/test_transcode.rb: added tests for the above
+	  (from Yoshihiro Kambayashi)
+
+Sun Jan  4 17:55:55 2009  Martin Duerst  <duerst@it.aoyama.ac.jp>
 
 	* test/ruby/test_transcode.rb: added tests for GB2312
 	  (from Yoshihiro Kambayashi)

enc/trans/gbk.trans

@@ -0,0 +1,16 @@
+#include "transcode_data.h"
+
+<%
+  require "gbk-tbl"
+
+  transcode_tblgen "GBK", "UTF-8", [["{00-7f}", :nomap], *GBK_TO_UCS_TBL]
+  transcode_tblgen "UTF-8", "GBK", [["{00-7f}", :nomap], *GBK_TO_UCS_TBL.map {|a,b| [b,a] }]
+%>
+
+<%= transcode_generated_code %>
+
+void
+Init_gbk(void)
+{
+<%= transcode_register_code %>
+}

test/ruby/test_transcode.rb

@@ -1448,6 +1448,120 @@ class TestTranscode < Test::Unit::TestCase
     check_both_ways("\u9752\u5C71\u5B66\u9662\u5927\u5B66", "\xC7\xE0\xC9\xBD\xD1\xA7\xD4\xBA\xB4\xF3\xD1\xA7", 'GB2312') # 青山学院大学
   end
   
+  def test_gbk
+    check_both_ways("\u4E02", "\x81\x40", 'GBK') # 丂
+    check_both_ways("\u4E8A", "\x81\x7E", 'GBK') # 亊
+    check_both_ways("\u4E90", "\x81\x80", 'GBK') # 亐
+    check_both_ways("\u4FA2", "\x81\xFE", 'GBK') # 侢
+    check_both_ways("\u5EC6", "\x8F\x40", 'GBK') # 廆
+    check_both_ways("\u5F24", "\x8F\x7E", 'GBK') # 弤
+    check_both_ways("\u5F28", "\x8F\x80", 'GBK') # 弨
+    check_both_ways("\u6007", "\x8F\xFE", 'GBK') # 怇
+    check_both_ways("\u6008", "\x90\x40", 'GBK') # 怈
+    check_both_ways("\u6080", "\x90\x7E", 'GBK') # 悀
+    check_both_ways("\u6081", "\x90\x80", 'GBK') # 悁
+    check_both_ways("\u6146", "\x90\xFE", 'GBK') # 慆
+    check_both_ways("\u70DC", "\x9F\x40", 'GBK') # 烜
+    check_both_ways("\u7134", "\x9F\x7E", 'GBK') # 焴
+    check_both_ways("\u7135", "\x9F\x80", 'GBK') # 焵
+    check_both_ways("\u71D3", "\x9F\xFE", 'GBK') # 燓
+    check_both_ways("\u71D6", "\xA0\x40", 'GBK') # 燖
+    check_both_ways("\u721A", "\xA0\x7E", 'GBK') # 爚
+    check_both_ways("\u721B", "\xA0\x80", 'GBK') # 爛
+    check_both_ways("\u72DB", "\xA0\xFE", 'GBK') # 狛
+    check_both_ways("\u3000", "\xA1\xA1", 'GBK') # full-width space
+    check_both_ways("\u3001", "\xA1\xA2", 'GBK') # 、
+    check_both_ways("\u3013", "\xA1\xFE", 'GBK') # 〓
+    assert_raise(Encoding::UndefinedConversionError) { "\xA2\xA0".encode("utf-8", 'GBK') }
+    check_both_ways("\u2170", "\xA2\xA1", 'GBK') # ⅰ
+    assert_raise(Encoding::UndefinedConversionError) { "\xA2\xB0".encode("utf-8", 'GBK') }
+    check_both_ways("\u2488", "\xA2\xB1", 'GBK') # ⒈
+    assert_raise(Encoding::UndefinedConversionError) { "\xA2\xE4".encode("utf-8", 'GBK') }
+    check_both_ways("\u3220", "\xA2\xE5", 'GBK') # ㈠
+    assert_raise(Encoding::UndefinedConversionError) { "\xA2\xF0".encode("utf-8", 'GBK') }
+    check_both_ways("\u2160", "\xA2\xF1", 'GBK') # Ⅰ
+    assert_raise(Encoding::UndefinedConversionError) { "\xA3\xA0".encode("utf-8", 'GBK') }
+    check_both_ways("\uFF01", "\xA3\xA1", 'GBK') # ！
+    check_both_ways("\uFFE3", "\xA3\xFE", 'GBK') # ￣
+    assert_raise(Encoding::UndefinedConversionError) { "\xA4\xA0".encode("utf-8", 'GBK') }
+    check_both_ways("\u3041", "\xA4\xA1", 'GBK') # ぁ
+    assert_raise(Encoding::UndefinedConversionError) { "\xA5\xA0".encode("utf-8", 'GBK') }
+    check_both_ways("\u30A1", "\xA5\xA1", 'GBK') # ァ
+    check_both_ways("\u0391", "\xA6\xA1", 'GBK') # Α
+    check_both_ways("\u03B1", "\xA6\xC1", 'GBK') # α
+    assert_raise(Encoding::UndefinedConversionError) { "\xA6\xED".encode("utf-8", 'GBK') }
+    check_both_ways("\uFE3B", "\xA6\xEE", 'GBK') # ︻
+    check_both_ways("\u0410", "\xA7\xA1", 'GBK') # А
+    check_both_ways("\u0430", "\xA7\xD1", 'GBK') # а
+    check_both_ways("\u02CA", "\xA8\x40", 'GBK') # ˊ
+    check_both_ways("\u2587", "\xA8\x7E", 'GBK') # ▇
+    assert_raise(Encoding::UndefinedConversionError) { "\xA8\x96".encode("utf-8", 'GBK') }
+    check_both_ways("\u0101", "\xA8\xA1", 'GBK') # ā
+    assert_raise(Encoding::UndefinedConversionError) { "\xA8\xBC".encode("utf-8", 'GBK') }
+    assert_raise(Encoding::UndefinedConversionError) { "\xA8\xBF".encode("utf-8", 'GBK') }
+    assert_raise(Encoding::UndefinedConversionError) { "\xA8\xC4".encode("utf-8", 'GBK') }
+    check_both_ways("\u3105", "\xA8\xC5", 'GBK') # ㄅ
+    check_both_ways("\u3021", "\xA9\x40", 'GBK') # 〡
+    assert_raise(Encoding::UndefinedConversionError) { "\xA9\x58".encode("utf-8", 'GBK') }
+    assert_raise(Encoding::UndefinedConversionError) { "\xA9\x5B".encode("utf-8", 'GBK') }
+    assert_raise(Encoding::UndefinedConversionError) { "\xA9\x5D".encode("utf-8", 'GBK') }
+    check_both_ways("\u3007", "\xA9\x96", 'GBK') # 〇
+    assert_raise(Encoding::UndefinedConversionError) { "\xA9\xA3".encode("utf-8", 'GBK') }
+    check_both_ways("\u2500", "\xA9\xA4", 'GBK') # ─
+    assert_raise(Encoding::UndefinedConversionError) { "\xA9\xF0".encode("utf-8", 'GBK') }
+    check_both_ways("\u7588", "\xAF\x40", 'GBK') # 疈
+    check_both_ways("\u7607", "\xAF\x7E", 'GBK') # 瘇
+    check_both_ways("\u7608", "\xAF\x80", 'GBK') # 瘈
+    check_both_ways("\u7644", "\xAF\xA0", 'GBK') # 癄
+    assert_raise(Encoding::UndefinedConversionError) { "\xAF\xA1".encode("utf-8", 'GBK') }
+    check_both_ways("\u7645", "\xB0\x40", 'GBK') # 癅
+    check_both_ways("\u769B", "\xB0\x7E", 'GBK') # 皛
+    check_both_ways("\u769C", "\xB0\x80", 'GBK') # 皜
+    check_both_ways("\u5265", "\xB0\xFE", 'GBK') # 剥
+    check_both_ways("\u7DFB", "\xBF\x40", 'GBK') # 緻
+    check_both_ways("\u7E39", "\xBF\x7E", 'GBK') # 縹
+    check_both_ways("\u7E3A", "\xBF\x80", 'GBK') # 縺
+    check_both_ways("\u5080", "\xBF\xFE", 'GBK') # 傀
+    check_both_ways("\u7E5E", "\xC0\x40", 'GBK') # 繞
+    check_both_ways("\u7E9E", "\xC0\x7E", 'GBK') # 纞
+    check_both_ways("\u7EAE", "\xC0\x80", 'GBK') # 纮
+    check_both_ways("\u4FD0", "\xC0\xFE", 'GBK') # 俐
+    check_both_ways("\u87A5", "\xCF\x40", 'GBK') # 螥
+    check_both_ways("\u87F8", "\xCF\x7E", 'GBK') # 蟸
+    check_both_ways("\u87FA", "\xCF\x80", 'GBK') # 蟺
+    check_both_ways("\u6653", "\xCF\xFE", 'GBK') # 晓
+    check_both_ways("\u8824", "\xD0\x40", 'GBK') # 蠤
+    check_both_ways("\u887A", "\xD0\x7E", 'GBK') # 衺
+    check_both_ways("\u887B", "\xD0\x80", 'GBK') # 衻
+    check_both_ways("\u7384", "\xD0\xFE", 'GBK') # 玄
+    check_both_ways("\u9019", "\xDF\x40", 'GBK') # 這
+    check_both_ways("\u9081", "\xDF\x7E", 'GBK') # 邁
+    check_both_ways("\u9084", "\xDF\x80", 'GBK') # 還
+    check_both_ways("\u553C", "\xDF\xFE", 'GBK') # 唼
+    check_both_ways("\u90C2", "\xE0\x40", 'GBK') # 郂
+    check_both_ways("\u911C", "\xE0\x7E", 'GBK') # 鄜
+    check_both_ways("\u911D", "\xE0\x80", 'GBK') # 鄝
+    check_both_ways("\u5E3C", "\xE0\xFE", 'GBK') # 帼
+    check_both_ways("\u986F", "\xEF\x40", 'GBK') # 顯
+    check_both_ways("\u98E4", "\xEF\x7E", 'GBK') # 飤
+    check_both_ways("\u98E5", "\xEF\x80", 'GBK') # 飥
+    check_both_ways("\u7A14", "\xEF\xFE", 'GBK') # 稔
+    check_both_ways("\u9908", "\xF0\x40", 'GBK') # 餈
+    check_both_ways("\u9949", "\xF0\x7E", 'GBK') # 饉
+    check_both_ways("\u994A", "\xF0\x80", 'GBK') # 饊
+    check_both_ways("\u7619", "\xF0\xFE", 'GBK') # 瘙
+    check_both_ways("\u9F32", "\xFD\x40", 'GBK') # 鼲
+    check_both_ways("\u9F78", "\xFD\x7E", 'GBK') # 齸
+    check_both_ways("\u9F79", "\xFD\x80", 'GBK') # 齹
+    check_both_ways("\uF9F1", "\xFD\xA0", 'GBK') # 隣
+    assert_raise(Encoding::UndefinedConversionError) { "\xFD\xA1".encode("utf-8", 'GBK') }
+    check_both_ways("\uFA0C", "\xFE\x40", 'GBK') # 兀
+    check_both_ways("\uFA29", "\xFE\x4F", 'GBK') # 﨩
+    assert_raise(Encoding::UndefinedConversionError) { "\xFE\x50".encode("utf-8", 'GBK') }
+    check_both_ways("\u9752\u5C71\u5B66\u9662\u5927\u5B66", "\xC7\xE0\xC9\xBD\xD1\xA7\xD4\xBA\xB4\xF3\xD1\xA7", 'GBK') # 青山学院大学
+    check_both_ways("\u795E\u6797\u7FA9\u535A", "\xC9\xF1\xC1\xD6\xC1\x78\xB2\xA9", 'GBK') # 神林義博
+  end
+
   def test_Big5
     check_both_ways("\u3000", "\xA1\x40", 'Big5') # full-width space
     check_both_ways("\uFE5A", "\xA1\x7E", 'Big5') # ﹚