AuroraMiddleware/scuffed-code
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

ICU-3818 Fix for Thai Dictionary Break Iterator following(1) failure.

Browse Source 
X-SVN-Rev: 15815
This commit is contained in:
Andy Heninger 2004-06-09 18:12:01 +00:00
parent 5612811355
commit 53e521de42
3 changed files with 52 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
icu4c/source/common/dbbi.cpp
View File

@ -17,6 +17,7 @@
#include "dbbi_tbl.h"
#include "uvector.h"
#include "cmemory.h"
#include "uassert.h"
U_NAMESPACE_BEGIN
@ -142,7 +143,14 @@ DictionaryBasedBreakIterator::previous()
reset();
int32_t result = RuleBasedBreakIterator::previous();
if (cachedBreakPositions != NULL) {
positionInCache = numCachedBreakPositions - 2;
for (positionInCache=0;
cachedBreakPositions[positionInCache] != result;
positionInCache++);
U_ASSERT(positionInCache < numCachedBreakPositions);
if (positionInCache >= numCachedBreakPositions) {
// Something has gone wrong. Dump the cache.
reset();
}
}
return result;
}
@ -257,8 +265,13 @@ DictionaryBasedBreakIterator::handleNext()
// for the new range
if (fDictionaryCharCount > 1 && result - startPos > 1) {
divideUpDictionaryRange(startPos, result, status);
U_ASSERT(U_SUCCESS(status));
if (U_FAILURE(status)) {
return -9999; // SHOULD NEVER GET HERE!
// Something went badly wrong, an internal error.
// We have no way from here to report it to caller.
// Treat as if this is if the dictionary did not apply to range.
reset();
return result;
}
}

41
icu4c/source/test/intltest/rbbitst.cpp
View File

@ -546,6 +546,34 @@ void RBBITest::TestThaiWordBreak() {
}
void RBBITest::TestBug3818() {
UErrorCode status = U_ZERO_ERROR;
// Four Thai words...
static const UChar thaiWordData[] = { 0x0E43,0x0E2B,0x0E0D,0x0E48, 0x0E43,0x0E2B,0x0E0D,0x0E48,
0x0E43,0x0E2B,0x0E0D,0x0E48, 0x0E43,0x0E2B,0x0E0D,0x0E48, 0 };
UnicodeString thaiStr(thaiWordData);
RuleBasedBreakIterator* bi =
(RuleBasedBreakIterator *)BreakIterator::createWordInstance(Locale("th"), status);
if (U_FAILURE(status) || bi == NULL) {
errln("Fail at file %s, line %d, status = %s", __FILE__, __LINE__, u_errorName(status));
return;
}
bi->setText(thaiStr);
int32_t startOfSecondWord = bi->following(1);
if (startOfSecondWord != 4) {
errln("Fail at file %s, line %d expected start of word at 4, got %d",
__FILE__, __LINE__, startOfSecondWord);
}
startOfSecondWord = bi->following(0);
if (startOfSecondWord != 4) {
errln("Fail at file %s, line %d expected start of word at 4, got %d",
__FILE__, __LINE__, startOfSecondWord);
}
}
//---------------------------------------------
// runIndexedTest
//---------------------------------------------
@ -584,13 +612,13 @@ void RBBITest::runIndexedTest( int32_t index, UBool exec, const char* &name, cha
case 11: name = "TestMaiyamok";
if(exec) TestMaiyamok(); break;
case 12: name = "TestWordBreaks";
if(exec) TestWordBreaks(); break;
if(exec) TestWordBreaks(); break;
case 13: name = "TestWordBoundary";
if(exec) TestWordBoundary(); break;
if(exec) TestWordBoundary(); break;
case 14: name = "TestLineBreaks";
if(exec) TestLineBreaks(); break;
if(exec) TestLineBreaks(); break;
case 15: name = "TestSentBreaks";
if(exec) TestSentBreaks(); break;
if(exec) TestSentBreaks(); break;
case 16: name = "TestExtended";
if(exec) TestExtended(); break;
case 17: name = "TestMonkey";
@ -601,7 +629,10 @@ void RBBITest::runIndexedTest( int32_t index, UBool exec, const char* &name, cha
logln("skipping TestMonkey (UCONFIG_NO_REGULAR_EXPRESSIONS)");
#endif
}
break;
break;
case 18: name = "TestBug3818";
if(exec) TestBug3818(); break;
default: name = ""; break; //needed to end loop
}
}

1
icu4c/source/test/intltest/rbbitst.h
View File

@ -74,6 +74,7 @@ public:
void TestWordBoundary();
void TestLineBreaks();
void TestSentBreaks();
void TestBug3818();
/***********************/