ICU-4157 Word Break, fix problem with CR <combining> sequences
X-SVN-Rev: 17427
This commit is contained in:
parent
3fdcb27040
commit
2b714f2bce
@ -52,7 +52,7 @@ $ExtendNumLetEx = $ExtendNumLet $Extend*;
|
|||||||
# see character breaks.
|
# see character breaks.
|
||||||
|
|
||||||
$CR $LF;
|
$CR $LF;
|
||||||
[^$Control] $Extend+;
|
[^$Control $CR $LF] $Extend+;
|
||||||
$NumericEx $Extend* {100};
|
$NumericEx $Extend* {100};
|
||||||
$ALetterEx $Extend* {200};
|
$ALetterEx $Extend* {200};
|
||||||
|
|
||||||
@ -114,7 +114,7 @@ $LF $CR;
|
|||||||
|
|
||||||
# see character breaks
|
# see character breaks
|
||||||
|
|
||||||
$Extend* [^$Control];
|
$Extend* [^$Control $CR $LF];
|
||||||
|
|
||||||
# rule 5
|
# rule 5
|
||||||
|
|
||||||
|
@ -2139,6 +2139,8 @@ RBBIWordMonkey::RBBIWordMonkey() : fGCFMatcher(0),
|
|||||||
fOtherSet->removeAll(*fMidNumSet);
|
fOtherSet->removeAll(*fMidNumSet);
|
||||||
fOtherSet->removeAll(*fNumericSet);
|
fOtherSet->removeAll(*fNumericSet);
|
||||||
fOtherSet->removeAll(*fExtendNumLetSet);
|
fOtherSet->removeAll(*fExtendNumLetSet);
|
||||||
|
fOtherSet->removeAll(*fFormatSet);
|
||||||
|
fOtherSet->removeAll(*fExtendSet);
|
||||||
|
|
||||||
fSets->addElement(fALetterSet, status);
|
fSets->addElement(fALetterSet, status);
|
||||||
fSets->addElement(fKatakanaSet, status);
|
fSets->addElement(fKatakanaSet, status);
|
||||||
@ -2146,6 +2148,7 @@ RBBIWordMonkey::RBBIWordMonkey() : fGCFMatcher(0),
|
|||||||
fSets->addElement(fMidNumSet, status);
|
fSets->addElement(fMidNumSet, status);
|
||||||
fSets->addElement(fNumericSet, status);
|
fSets->addElement(fNumericSet, status);
|
||||||
fSets->addElement(fFormatSet, status);
|
fSets->addElement(fFormatSet, status);
|
||||||
|
fSets->addElement(fExtendSet, status);
|
||||||
fSets->addElement(fOtherSet, status);
|
fSets->addElement(fOtherSet, status);
|
||||||
fSets->addElement(fExtendNumLetSet, status);
|
fSets->addElement(fExtendNumLetSet, status);
|
||||||
|
|
||||||
|
Loading…
Reference in New Issue
Block a user