summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
-rw-r--r--i18npool/qa/cppunit/test_breakiterator.cxx24
-rw-r--r--i18npool/source/breakiterator/breakiterator_unicode.cxx2
2 files changed, 25 insertions, 1 deletions
diff --git a/i18npool/qa/cppunit/test_breakiterator.cxx b/i18npool/qa/cppunit/test_breakiterator.cxx
index ffd590c0c05a..a8a050aacdae 100644
--- a/i18npool/qa/cppunit/test_breakiterator.cxx
+++ b/i18npool/qa/cppunit/test_breakiterator.cxx
@@ -62,6 +62,7 @@ public:
#if TODO
void testNorthernThai();
#endif
+ void testKhmer();
CPPUNIT_TEST_SUITE(TestBreakIterator);
CPPUNIT_TEST(testLineBreaking);
@@ -72,6 +73,7 @@ public:
#if TODO
CPPUNIT_TEST(testNorthernThai);
#endif
+ CPPUNIT_TEST(testKhmer);
CPPUNIT_TEST_SUITE_END();
private:
uno::Reference<i18n::XBreakIterator> m_xBreak;
@@ -344,6 +346,28 @@ void TestBreakIterator::testNorthernThai()
}
#endif
+//A test to ensure that our khmer word boundary detection is useful
+//https://bugs.freedesktop.org/show_bug.cgi?id=52020
+void TestBreakIterator::testKhmer()
+{
+ lang::Locale aLocale;
+ aLocale.Language = ::rtl::OUString(RTL_CONSTASCII_USTRINGPARAM("km"));
+ aLocale.Country = ::rtl::OUString(RTL_CONSTASCII_USTRINGPARAM("KH"));
+
+ const sal_Unicode KHMER1[] = { 0x17B2, 0x17D2, 0x1799, 0x1782, 0x17C1 };
+
+ ::rtl::OUString aTest(KHMER1, SAL_N_ELEMENTS(KHMER1));
+ i18n::Boundary aBounds = m_xBreak->getWordBoundary(aTest, 0, aLocale,
+ i18n::WordType::DICTIONARY_WORD, true);
+
+ CPPUNIT_ASSERT(aBounds.startPos == 0 && aBounds.endPos == 3);
+
+ aBounds = m_xBreak->getWordBoundary(aTest, aBounds.endPos, aLocale,
+ i18n::WordType::DICTIONARY_WORD, true);
+
+ CPPUNIT_ASSERT(aBounds.startPos == 3 && aBounds.endPos == 5);
+}
+
void TestBreakIterator::setUp()
{
BootstrapFixtureBase::setUp();
diff --git a/i18npool/source/breakiterator/breakiterator_unicode.cxx b/i18npool/source/breakiterator/breakiterator_unicode.cxx
index 34e6918f5182..e90c888e6816 100644
--- a/i18npool/source/breakiterator/breakiterator_unicode.cxx
+++ b/i18npool/source/breakiterator/breakiterator_unicode.cxx
@@ -135,7 +135,7 @@ 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);
}
- else if ( rLocale.Language != "th" ) //use icu's breakiterator for Thai
+ else if (rLocale.Language != "th" && rLocale.Language != "km") //use icu's breakiterator for Thai and Khmer
{
status = U_ZERO_ERROR;
OStringBuffer aUDName(64);