Resolves: rhbz#654200 revert icu#5431

This commit is contained in:
Caolán McNamara 2010-11-25 15:11:55 +00:00
parent b35314b5bd
commit e000c20502
2 changed files with 141 additions and 7 deletions

View File

@ -0,0 +1,129 @@
Index: icu/trunk/source/layout/IndicReordering.cpp
===================================================================
--- icu/trunk/source/layout/IndicReordering.cpp (revision 25772)
+++ icu/trunk/source/layout/IndicReordering.cpp (revision 26090)
@@ -126,4 +126,8 @@
FeatureMask fSMFeatures;
+ LEUnicode fPreBaseConsonant;
+ LEUnicode fPreBaseVirama;
+ le_int32 fPBCIndex;
+ FeatureMask fPBCFeatures;
void saveMatra(LEUnicode matra, le_int32 matraIndex, IndicClassTable::CharClass matraClass)
@@ -172,5 +176,6 @@
fMatraFeatures(0), fMPreOutIndex(-1), fMPreFixups(mpreFixups),
fVMabove(0), fVMpost(0), fVMIndex(0), fVMFeatures(0),
- fSMabove(0), fSMbelow(0), fSMIndex(0), fSMFeatures(0)
+ fSMabove(0), fSMbelow(0), fSMIndex(0), fSMFeatures(0),
+ fPreBaseConsonant(0), fPreBaseVirama(0), fPBCIndex(0), fPBCFeatures(0)
{
// nothing else to do...
@@ -191,4 +196,6 @@
fVMabove = fVMpost = 0;
fSMabove = fSMbelow = 0;
+
+ fPreBaseConsonant = fPreBaseVirama = 0;
}
@@ -386,4 +393,12 @@
}
+ void notePreBaseConsonant(le_uint32 index,LEUnicode PBConsonant, LEUnicode PBVirama, FeatureMask features)
+ {
+ fPBCIndex = index;
+ fPreBaseConsonant = PBConsonant;
+ fPreBaseVirama = PBVirama;
+ fPBCFeatures = features;
+ }
+
void noteBaseConsonant()
{
@@ -465,4 +480,20 @@
}
+ void writePreBaseConsonant()
+ {
+ // The TDIL spec says that consonant + virama + RRA should produce a rakar in Malayalam. However,
+ // it seems that almost none of the fonts for Malayalam are set up to handle this.
+ // So, we're going to force the issue here by using the rakar as defined with RA in most fonts.
+
+ if (fPreBaseConsonant == 0x0d31) { // RRA
+ fPreBaseConsonant = 0x0d30; // RA
+ }
+
+ if (fPreBaseConsonant != 0) {
+ writeChar(fPreBaseConsonant, fPBCIndex, fPBCFeatures);
+ writeChar(fPreBaseVirama,fPBCIndex-1,fPBCFeatures);
+ }
+ }
+
le_int32 getOutputIndex()
{
@@ -723,4 +754,5 @@
}
+
IndicClassTable::CharClass charClass = CC_RESERVED;
IndicClassTable::CharClass nextClass = CC_RESERVED;
@@ -730,7 +762,9 @@
le_bool seenVattu = FALSE;
le_bool seenBelowBaseForm = FALSE;
+ le_bool seenPreBaseForm = FALSE;
le_bool hasNukta = FALSE;
le_bool hasBelowBaseForm = FALSE;
le_bool hasPostBaseForm = FALSE;
+ le_bool hasPreBaseForm = FALSE;
if (postBase < markStart && classTable->isNukta(chars[postBase])) {
@@ -746,12 +780,20 @@
hasBelowBaseForm = IndicClassTable::hasBelowBaseForm(charClass) && !hasNukta;
hasPostBaseForm = IndicClassTable::hasPostBaseForm(charClass) && !hasNukta;
+ hasPreBaseForm = IndicClassTable::hasPreBaseForm(charClass) && !hasNukta;
if (IndicClassTable::isConsonant(charClass)) {
if (postBaseLimit == 0 || seenVattu ||
(baseConsonant > baseLimit && !classTable->isVirama(chars[baseConsonant - 1])) ||
- !(hasBelowBaseForm || hasPostBaseForm)) {
+ !(hasBelowBaseForm || hasPostBaseForm || hasPreBaseForm)) {
break;
}
+ // Note any pre-base consonants
+ if ( baseConsonant == lastConsonant && lastConsonant > 0 &&
+ hasPreBaseForm && classTable->isVirama(chars[baseConsonant - 1])) {
+ output.notePreBaseConsonant(lastConsonant,chars[lastConsonant],chars[lastConsonant-1],tagArray2);
+ seenPreBaseForm = TRUE;
+
+ }
// consonants with nuktas are never vattus
seenVattu = IndicClassTable::isVattu(charClass) && !hasNukta;
@@ -786,10 +828,12 @@
// write any pre-base consonants
+ output.writePreBaseConsonant();
+
le_bool supressVattu = TRUE;
for (i = baseLimit; i < baseConsonant; i += 1) {
LEUnicode ch = chars[i];
- // Don't put 'blwf' on first consonant.
- FeatureMask features = (i == baseLimit? tagArray2 : tagArray1);
+ // Don't put 'pstf' or 'blwf' on anything before the base consonant.
+ FeatureMask features = tagArray1 & ~( pstfFeatureMask | blwfFeatureMask );
charClass = classTable->getCharClass(ch);
@@ -842,5 +886,5 @@
// write below-base consonants
- if (baseConsonant != lastConsonant) {
+ if (baseConsonant != lastConsonant && !seenPreBaseForm) {
for (i = bcSpan + 1; i < postBase; i += 1) {
output.writeChar(chars[i], i, tagArray1);
@@ -872,5 +916,5 @@
// write post-base consonants
// FIXME: does this put the right tags on post-base consonants?
- if (baseConsonant != lastConsonant) {
+ if (baseConsonant != lastConsonant && !seenPreBaseForm) {
if (postBase <= lastConsonant) {
for (i = postBase; i <= lastConsonant; i += 1) {

View File

@ -1,6 +1,6 @@
Name: icu
Version: 4.4.2
Release: 1%{?dist}
Release: 2%{?dist}
Summary: International Components for Unicode
Group: Development/Tools
License: MIT and UCD and Public Domain
@ -11,12 +11,13 @@ BuildRoot: %{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n)
BuildRequires: doxygen, autoconf
Requires: lib%{name} = %{version}-%{release}
Patch1: icu-3.4-multiarchdevel.patch
Patch2: icu.6995.kannada.patch
Patch3: icu.7971.buildfix.patch
Patch4: icu.7972.buildfix.patch
Patch5: icu.7932.doublecompare.patch
Patch6: icu.8011.buildfix.patch
Patch1: icu-3.4-multiarchdevel.patch
Patch2: icu.6995.kannada.patch
Patch3: icu.7971.buildfix.patch
Patch4: icu.7972.buildfix.patch
Patch5: icu.7932.doublecompare.patch
Patch6: icu.8011.buildfix.patch
Patch7: icu.8198.revert.icu5431.patch
%description
Tools and utilities for developing with icu.
@ -62,6 +63,7 @@ BuildArch: noarch
%patch4 -p1 -b .icu7972.buildfix.patch
%patch5 -p1 -b .icu7932.doublecompare.patch
%patch6 -p1 -b .icu8011.buildfix.patch
%patch7 -p2 -R -b .icu8198.revert.icu5431.patch
%build
cd source
@ -143,6 +145,9 @@ rm -rf $RPM_BUILD_ROOT
%doc source/__docs/%{name}/html/*
%changelog
* Thu Nov 25 2010 Caolán McNamara <caolanm@redhat.com> - 4.4.2-2
- Resolves: rhbz#654200 revert icu#5431
* Mon Oct 04 2010 Caolán McNamara <caolanm@redhat.com> - 4.4.2-1
- latest version