[Libreoffice-commits] core.git: i18npool/qa i18npool/source

[Nathan Wells]

 i18npool/qa/cppunit/test_breakiterator.cxx              |   13 ++++++-------
 i18npool/source/breakiterator/breakiterator_unicode.cxx |    4 ++--
 2 files changed, 8 insertions(+), 9 deletions(-)

New commits:
commit 10199478b841a87e6436996bde221e424d1df708
Author: Nathan Wells <sungkhum at gmail.com>
Date:   Thu Sep 3 12:21:05 2015 +0100

    Related: tdf#52020 Disable ICU Breakiterator for Khmer
    
    Enabling the ICU Breakiterator causes Khmer spelling checkers to be worthless
    as well as wrecks havoc on various minority languages that use the Khmer
    script. See this https://bugs.documentfoundation.org/show_bug.cgi?id=52020
    this: https://bugs.freedesktop.org/show_bug.cgi?id=59448 and this:
    https://bugs.documentfoundation.org/show_bug.cgi?id=59447&redirected_from=fdo
    
    Change-Id: I84529db9cca72ffe1108d2e12c1885f63a48f89d

diff --git a/i18npool/qa/cppunit/test_breakiterator.cxx b/i18npool/qa/cppunit/test_breakiterator.cxx
index 22aa0d7..693d19c 100644
--- a/i18npool/qa/cppunit/test_breakiterator.cxx
+++ b/i18npool/qa/cppunit/test_breakiterator.cxx
@@ -40,11 +40,11 @@ public:
     void testThai();
 #if (U_ICU_VERSION_MAJOR_NUM > 51)
     void testLao();
-#endif
 #ifdef TODO
     void testNorthernThai();
-#endif
     void testKhmer();
+#endif
+#endif
     void testJapanese();
     void testChinese();
 
@@ -57,12 +57,10 @@ public:
     CPPUNIT_TEST(testThai);
 #if (U_ICU_VERSION_MAJOR_NUM > 51)
     CPPUNIT_TEST(testLao);
-#endif
 #ifdef TODO
+    CPPUNIT_TEST(testKhmer);
     CPPUNIT_TEST(testNorthernThai);
 #endif
-#if (U_ICU_VERSION_MAJOR_NUM > 4)
-    CPPUNIT_TEST(testKhmer);
 #endif
     CPPUNIT_TEST(testJapanese);
     CPPUNIT_TEST(testChinese);
@@ -889,12 +887,13 @@ void TestBreakIterator::testNorthernThai()
     CPPUNIT_ASSERT_MESSAGE("Should skip full word",
         aBounds.startPos == 0 && aBounds.endPos == aTest.getLength());
 }
-#endif
 
-#if (U_ICU_VERSION_MAJOR_NUM > 4)
 // Not sure if any version earlier than 49 did have Khmer word boundary
 // dictionaries, 4.6 does not.
 
+// As of icu 54, word boundary detection for Khmer is still considered
+// insufficient, so icu khmer stuff is disabled
+
 //A test to ensure that our khmer word boundary detection is useful
 //https://bugs.libreoffice.org/show_bug.cgi?id=52020
 void TestBreakIterator::testKhmer()
diff --git a/i18npool/source/breakiterator/breakiterator_unicode.cxx b/i18npool/source/breakiterator/breakiterator_unicode.cxx
index 53c1916..0f7512b 100644
--- a/i18npool/source/breakiterator/breakiterator_unicode.cxx
+++ b/i18npool/source/breakiterator/breakiterator_unicode.cxx
@@ -123,8 +123,8 @@ void SAL_CALL BreakIterator_Unicode::loadICUBreakIterator(const com::sun::star::
                 rbi = new OOoRuleBasedBreakIterator(udata_open("OpenOffice", "brk",
                     OUStringToOString(breakRules[breakType], RTL_TEXTENCODING_ASCII_US).getStr(), &status), status);
             }
-            //use icu's breakiterator for Thai, Khmer, Tibetan and Dzongkha
-            else if (rLocale.Language != "th" && rLocale.Language != "lo" && rLocale.Language != "km" && rLocale.Language != "bo" && rLocale.Language != "dz")
+            //use icu's breakiterator for Thai, Tibetan and Dzongkha
+            else if (rLocale.Language != "th" && rLocale.Language != "lo" && rLocale.Language != "bo" && rLocale.Language != "dz")
             {
                 status = U_ZERO_ERROR;
                 OStringBuffer aUDName(64);