ICU-10299 Fix CjkBreakEngine fSet to include 30FC,FF70; fix broken test data (ICU4C)

X-SVN-Rev: 34118

diff --git a/icu4c/source/common/dictbe.cpp b/icu4c/source/common/dictbe.cpp
index 15df9fb8dce42654471a28dc08d21240d6ae3a9b..bbedf14694d036854e66eebe7e5fc932a7d65ceb 100644 (file)
--- a/icu4c/source/common/dictbe.cpp
+++ b/icu4c/source/common/dictbe.cpp
@@ -1,6 +1,6 @@
 /**
  *******************************************************************************
- * Copyright (C) 2006-2012, International Business Machines Corporation
+ * Copyright (C) 2006-2013, International Business Machines Corporation
  * and others. All Rights Reserved.
  *******************************************************************************
  */
@@ -667,7 +667,8 @@ CjkBreakEngine::CjkBreakEngine(DictionaryMatcher *adoptDictionary, LanguageType
             cjSet.addAll(fHanWordSet);
             cjSet.addAll(fKatakanaWordSet);
             cjSet.addAll(fHiraganaWordSet);
-            cjSet.add(UNICODE_STRING_SIMPLE("\\uff70\\u30fc"));
+            cjSet.add(0xFF70); // HALFWIDTH KATAKANA-HIRAGANA PROLONGED SOUND MARK
+            cjSet.add(0x30FC); // KATAKANA-HIRAGANA PROLONGED SOUND MARK
             setCharacters(cjSet);
         }
     }

diff --git a/icu4c/source/test/testdata/rbbitst.txt b/icu4c/source/test/testdata/rbbitst.txt
index aef6b939a2c5e01e56fd50cf56bede994a60785a..3301893101c9a6dcea82879dbe395e47e3ca0167 100644 (file)
--- a/icu4c/source/test/testdata/rbbitst.txt
+++ b/icu4c/source/test/testdata/rbbitst.txt
@@ -757,11 +757,11 @@ Bangkok)•</data>
 
 <locale ja>
 <word>
-<data>•私<400>達<400>に<400>一<400>〇<400>〇〇<400>の<400>コンピュ<400>ー<400>タ<400>が<400>ある<400>。<0>奈々<400>は<400>ワ<400>ー<400>ド<400>で<400>ある<400>。•</data>
+<data>•私<400>達<400>に<400>一<400>〇<400>〇〇<400>の<400>コンピュータ<400>が<400>ある<400>。<0>奈々<400>は<400>ワード<400>で<400>ある<400>。•</data>
 
 <locale root>
 <word>
-<data>•私<400>達<400>に<400>一<400>〇<400>〇〇<400>の<400>コンピュ<400>ー<400>タ<400>が<400>ある<400>。<0>奈々<400>は<400>ワ<400>ー<400>ド<400>で<400>ある<400>。•</data>
+<data>•私<400>達<400>に<400>一<400>〇<400>〇〇<400>の<400>コンピュータ<400>が<400>ある<400>。<0>奈々<400>は<400>ワード<400>で<400>ある<400>。•</data>
 
 # UBreakIteratorType UBRK_SENTENCE, Locale "el"
 # Add break after Greek question mark (cldrbug #2069).