sl@0: /*
sl@0: *
sl@0: * (C) Copyright IBM Corp. 1998-2005 - All Rights Reserved
sl@0: *
sl@0: */
sl@0:
sl@0: #include "LETypes.h"
sl@0: #include "OpenTypeTables.h"
sl@0: #include "OpenTypeUtilities.h"
sl@0: #include "IndicReordering.h"
sl@0: #include "LEGlyphStorage.h"
sl@0: #include "MPreFixups.h"
sl@0:
sl@0: U_NAMESPACE_BEGIN
sl@0:
sl@0: const LEUnicode KBearURA = 0x0A73;
sl@0: const LEUnicode KDepVowelU = 0x0A41;
sl@0: const LEUnicode KDepVowelUU = 0x0A42;
sl@0: const LEUnicode KDepVowelOO = 0x0A4B;
sl@0: const LEUnicode KVowelU = 0x0A09;
sl@0: const LEUnicode KVowelUU = 0x0A0A;
sl@0: const LEUnicode KVowelOO = 0x0A13;
sl@0:
sl@0: class ReorderingOutput : public UMemory {
sl@0: private:
sl@0: le_int32 fOutIndex;
sl@0: LEUnicode *fOutChars;
sl@0:
sl@0: LEGlyphStorage &fGlyphStorage;
sl@0:
sl@0: LEUnicode fMpre;
sl@0: le_int32 fMpreIndex;
sl@0:
sl@0: LEUnicode fMbelow;
sl@0: le_int32 fMbelowIndex;
sl@0:
sl@0: LEUnicode fMabove;
sl@0: le_int32 fMaboveIndex;
sl@0:
sl@0: LEUnicode fMpost;
sl@0: le_int32 fMpostIndex;
sl@0:
sl@0: LEUnicode fLengthMark;
sl@0: le_int32 fLengthMarkIndex;
sl@0:
sl@0: const LETag *fMatraTags;
sl@0:
sl@0: le_int32 fMPreOutIndex;
sl@0: MPreFixups *fMPreFixups;
sl@0:
sl@0: LEUnicode fVMabove;
sl@0: LEUnicode fVMpost;
sl@0: le_int32 fVMIndex;
sl@0: const LETag *fVMTags;
sl@0:
sl@0: LEUnicode fSMabove;
sl@0: LEUnicode fSMbelow;
sl@0: le_int32 fSMIndex;
sl@0: const LETag *fSMTags;
sl@0:
sl@0: void saveMatra(LEUnicode matra, le_int32 matraIndex, IndicClassTable::CharClass matraClass)
sl@0: {
sl@0: // FIXME: check if already set, or if not a matra...
sl@0: if (IndicClassTable::isLengthMark(matraClass)) {
sl@0: fLengthMark = matra;
sl@0: fLengthMarkIndex = matraIndex;
sl@0: } else {
sl@0: switch (matraClass & CF_POS_MASK) {
sl@0: case CF_POS_BEFORE:
sl@0: fMpre = matra;
sl@0: fMpreIndex = matraIndex;
sl@0: break;
sl@0:
sl@0: case CF_POS_BELOW:
sl@0: fMbelow = matra;
sl@0: fMbelowIndex = matraIndex;
sl@0: break;
sl@0:
sl@0: case CF_POS_ABOVE:
sl@0: fMabove = matra;
sl@0: fMaboveIndex = matraIndex;
sl@0: break;
sl@0:
sl@0: case CF_POS_AFTER:
sl@0: fMpost = matra;
sl@0: fMpostIndex = matraIndex;
sl@0: break;
sl@0:
sl@0: default:
sl@0: // can't get here...
sl@0: break;
sl@0: }
sl@0: }
sl@0:
sl@0: // 1922 mlyl : only the right part of matra AU should be displayed -->
sl@0: if (fMpre == CC_DEPENDENT_VOWEL_SIGN_MLYL_E && fLengthMark != 0)
sl@0: {
sl@0: fMpre = 0;
sl@0: fMpreIndex = -1;
sl@0: }
sl@0: // <-- 1922 mlyl
sl@0: }
sl@0:
sl@0: public:
sl@0: ReorderingOutput(LEUnicode *outChars, LEGlyphStorage &glyphStorage, MPreFixups *mpreFixups)
sl@0: : fOutIndex(0), fOutChars(outChars), fGlyphStorage(glyphStorage),
sl@0: fMpre(0), fMpreIndex(0), fMbelow(0), fMbelowIndex(0), fMabove(0), fMaboveIndex(0),
sl@0: fMpost(0), fMpostIndex(0), fLengthMark(0), fLengthMarkIndex(0), fMatraTags(NULL),
sl@0: fMPreOutIndex(-1), fMPreFixups(mpreFixups),
sl@0: fVMabove(0), fVMpost(0), fVMIndex(0), fVMTags(NULL),
sl@0: fSMabove(0), fSMbelow(0), fSMIndex(0), fSMTags(NULL)
sl@0: {
sl@0: // nothing else to do...
sl@0: }
sl@0:
sl@0: ~ReorderingOutput()
sl@0: {
sl@0: // nothing to do here...
sl@0: }
sl@0:
sl@0: void reset()
sl@0: {
sl@0: fMpre = fMbelow = fMabove = fMpost = fLengthMark = 0;
sl@0: fMPreOutIndex = -1;
sl@0:
sl@0: fVMabove = fVMpost = 0;
sl@0: fSMabove = fSMbelow = 0;
sl@0: }
sl@0:
sl@0: void writeChar(LEUnicode ch, le_uint32 charIndex, const LETag *charTags)
sl@0: {
sl@0: LEErrorCode success = LE_NO_ERROR;
sl@0:
sl@0: fOutChars[fOutIndex] = ch;
sl@0:
sl@0: fGlyphStorage.setCharIndex(fOutIndex, charIndex, success);
sl@0: fGlyphStorage.setAuxData(fOutIndex, (void *) charTags, success);
sl@0:
sl@0: fOutIndex += 1;
sl@0: }
sl@0:
sl@0: le_bool noteMatra(const IndicClassTable *classTable, LEUnicode matra, le_uint32 matraIndex, const LETag *matraTags)
sl@0: {
sl@0: IndicClassTable::CharClass matraClass = classTable->getCharClass(matra);
sl@0:
sl@0: fMatraTags = matraTags;
sl@0:
sl@0: if (IndicClassTable::isMatra(matraClass)) {
sl@0: if (IndicClassTable::isSplitMatra(matraClass)) {
sl@0: const SplitMatra *splitMatra = classTable->getSplitMatra(matraClass);
sl@0: int i;
sl@0:
sl@0: for (i = 0; i < 3 && (*splitMatra)[i] != 0; i += 1) {
sl@0: LEUnicode piece = (*splitMatra)[i];
sl@0: IndicClassTable::CharClass pieceClass = classTable->getCharClass(piece);
sl@0:
sl@0: saveMatra(piece, matraIndex, pieceClass);
sl@0: }
sl@0: } else {
sl@0: saveMatra(matra, matraIndex, matraClass);
sl@0: }
sl@0:
sl@0: return TRUE;
sl@0: }
sl@0:
sl@0: return FALSE;
sl@0: }
sl@0:
sl@0: void noteVowelModifier(const IndicClassTable *classTable, LEUnicode vowelModifier, le_uint32 vowelModifierIndex, const LETag *vowelModifierTags)
sl@0: {
sl@0: IndicClassTable::CharClass vmClass = classTable->getCharClass(vowelModifier);
sl@0:
sl@0: fVMIndex = vowelModifierIndex;
sl@0: fVMTags = vowelModifierTags;
sl@0:
sl@0: if (IndicClassTable::isVowelModifier(vmClass)) {
sl@0: switch (vmClass & CF_POS_MASK) {
sl@0: case CF_POS_ABOVE:
sl@0: fVMabove = vowelModifier;
sl@0: break;
sl@0:
sl@0: case CF_POS_AFTER:
sl@0: fVMpost = vowelModifier;
sl@0: break;
sl@0:
sl@0: default:
sl@0: // FIXME: this is an error...
sl@0: break;
sl@0: }
sl@0: }
sl@0: }
sl@0:
sl@0: void noteStressMark(const IndicClassTable *classTable, LEUnicode stressMark, le_uint32 stressMarkIndex, const LETag *stressMarkTags)
sl@0: {
sl@0: IndicClassTable::CharClass smClass = classTable->getCharClass(stressMark);
sl@0:
sl@0: fSMIndex = stressMarkIndex;
sl@0: fSMTags = stressMarkTags;
sl@0:
sl@0: if (IndicClassTable::isStressMark(smClass)) {
sl@0: switch (smClass & CF_POS_MASK) {
sl@0: case CF_POS_ABOVE:
sl@0: fSMabove = stressMark;
sl@0: break;
sl@0:
sl@0: case CF_POS_BELOW:
sl@0: fSMbelow = stressMark;
sl@0: break;
sl@0:
sl@0: default:
sl@0: // FIXME: this is an error...
sl@0: break;
sl@0: }
sl@0: }
sl@0: }
sl@0:
sl@0: void noteBaseConsonant()
sl@0: {
sl@0: if (fMPreFixups != NULL && fMPreOutIndex >= 0) {
sl@0: fMPreFixups->add(fOutIndex, fMPreOutIndex);
sl@0: }
sl@0: }
sl@0:
sl@0: void writeMpre()
sl@0: {
sl@0: if (fMpre != 0) {
sl@0: fMPreOutIndex = fOutIndex;
sl@0: writeChar(fMpre, fMpreIndex, fMatraTags);
sl@0: }
sl@0: }
sl@0:
sl@0: void writeMbelow()
sl@0: {
sl@0: if (fMbelow != 0) {
sl@0: writeChar(fMbelow, fMbelowIndex, fMatraTags);
sl@0: }
sl@0: }
sl@0:
sl@0: void writeMabove()
sl@0: {
sl@0: if (fMabove != 0) {
sl@0: writeChar(fMabove, fMaboveIndex, fMatraTags);
sl@0: }
sl@0: }
sl@0:
sl@0: void writeMpost()
sl@0: {
sl@0: if (fMpost != 0) {
sl@0: writeChar(fMpost, fMpostIndex, fMatraTags);
sl@0: }
sl@0: }
sl@0:
sl@0: void writeLengthMark()
sl@0: {
sl@0: if (fLengthMark != 0) {
sl@0: writeChar(fLengthMark, fLengthMarkIndex, fMatraTags);
sl@0: }
sl@0: }
sl@0:
sl@0: void writeVMabove()
sl@0: {
sl@0: if (fVMabove != 0) {
sl@0: writeChar(fVMabove, fVMIndex, fVMTags);
sl@0: }
sl@0: }
sl@0:
sl@0: void writeVMpost()
sl@0: {
sl@0: if (fVMpost != 0) {
sl@0: writeChar(fVMpost, fVMIndex, fVMTags);
sl@0: }
sl@0: }
sl@0:
sl@0: void writeSMabove()
sl@0: {
sl@0: if (fSMabove != 0) {
sl@0: writeChar(fSMabove, fSMIndex, fSMTags);
sl@0: }
sl@0: }
sl@0:
sl@0: void writeSMbelow()
sl@0: {
sl@0: if (fSMbelow != 0) {
sl@0: writeChar(fSMbelow, fSMIndex, fSMTags);
sl@0: }
sl@0: }
sl@0:
sl@0: le_int32 getOutputIndex()
sl@0: {
sl@0: return fOutIndex;
sl@0: }
sl@0: };
sl@0:
sl@0: enum
sl@0: {
sl@0: C_DOTTED_CIRCLE = 0x25CC
sl@0: };
sl@0:
sl@0: static const LETag emptyTag = 0x00000000; // ''
sl@0:
sl@0: static const LETag nuktFeatureTag = LE_NUKT_FEATURE_TAG;
sl@0: static const LETag akhnFeatureTag = LE_AKHN_FEATURE_TAG;
sl@0: static const LETag rphfFeatureTag = LE_RPHF_FEATURE_TAG;
sl@0: static const LETag blwfFeatureTag = LE_BLWF_FEATURE_TAG;
sl@0: static const LETag halfFeatureTag = LE_HALF_FEATURE_TAG;
sl@0: static const LETag pstfFeatureTag = LE_PSTF_FEATURE_TAG;
sl@0: static const LETag vatuFeatureTag = LE_VATU_FEATURE_TAG;
sl@0: static const LETag presFeatureTag = LE_PRES_FEATURE_TAG;
sl@0: static const LETag blwsFeatureTag = LE_BLWS_FEATURE_TAG;
sl@0: static const LETag abvsFeatureTag = LE_ABVS_FEATURE_TAG;
sl@0: static const LETag pstsFeatureTag = LE_PSTS_FEATURE_TAG;
sl@0: static const LETag halnFeatureTag = LE_HALN_FEATURE_TAG;
sl@0:
sl@0: static const LETag blwmFeatureTag = LE_BLWM_FEATURE_TAG;
sl@0: static const LETag abvmFeatureTag = LE_ABVM_FEATURE_TAG;
sl@0: static const LETag distFeatureTag = LE_DIST_FEATURE_TAG;
sl@0: static const LETag loclFeatureTag = LE_LOCL_FEATURE_TAG;
sl@0:
sl@0: // These are in the order in which the features need to be applied
sl@0: // for correct processing
sl@0: static const LETag featureOrder[] =
sl@0: {
sl@0: loclFeatureTag,
sl@0: nuktFeatureTag, akhnFeatureTag, rphfFeatureTag, blwfFeatureTag, halfFeatureTag, pstfFeatureTag,
sl@0: vatuFeatureTag, presFeatureTag, blwsFeatureTag, abvsFeatureTag, pstsFeatureTag, halnFeatureTag,
sl@0: blwmFeatureTag, abvmFeatureTag, distFeatureTag, emptyTag
sl@0: };
sl@0:
sl@0: // The order of these is determined so that the tag array of each glyph can start
sl@0: // at an offset into this array
sl@0: // FIXME: do we want a seperate tag array for each kind of character??
sl@0: // FIXME: are there cases where this ordering causes glyphs to get tags
sl@0: // that they shouldn't?
sl@0: static const LETag tagArray[] =
sl@0: {
sl@0: rphfFeatureTag, blwfFeatureTag, halfFeatureTag, pstfFeatureTag, nuktFeatureTag, akhnFeatureTag,
sl@0: vatuFeatureTag, presFeatureTag, blwsFeatureTag, abvsFeatureTag, pstsFeatureTag, halnFeatureTag,
sl@0: blwmFeatureTag, abvmFeatureTag, distFeatureTag, loclFeatureTag, emptyTag
sl@0: };
sl@0:
sl@0:
sl@0: // Nokia: New state responses ia and de, and new states for Devanagari Independent Vowel A
sl@0: // added by Nokia to allow the special Devanagari CANDRA A formation using A + CANDRA E, and
sl@0: // a new zero width joiner state specially for ZWJ's that come after the Independent Vowel A
sl@0: // Also, changed the consonant response to ZWJ by allowing it to have a ZWJ straight after it.
sl@0: // This allows Kannada RA to take a ZWJ after it and not make it a REPHA
sl@0: // Nokia: New states for Gurmukhi bearers to support decompositions of the independent vowels
sl@0: // into a sequence of a vowel bearer and a depending vowel sign.
sl@0: static const le_int8 stateTable[][CC_COUNT] =
sl@0: {
sl@0: // xx vm sm iv i2 ct cn nu dv s1 s2 s3 vr zw ia de cr to ba bi bu ga gi gu
sl@0: { 1, 1, 1, 5, 8, 3, 2, 1, 1, 9, 5, 1, 1, 1, 20, 1, 25, 26, 27, 28, 29, 1, 1, 1}, // 0 - ground state
sl@0: {-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1}, // 1 - exit state
sl@0: {-1, 6, 1, -1, -1, -1, -1, -1, 5, 9, 5, 5, 4, -1, -1, 5, -1, 26, -1, -1, -1, 5, 5, 5}, // 2 - consonant with nukta
sl@0: {-1, 6, 1, -1, -1, -1, -1, 2, 5, 9, 5, 5, 4, -1, -1, 5, -1, 26, -1, -1, -1, 5, 5, 5}, // 3 - consonant
sl@0: {-1, -1, -1, -1, -1, 12, 11, -1, -1, -1, -1, -1, -1, 22, -1, -1, 12, -1, -1, -1, -1, -1, -1, -1}, // 4 - ct vr
sl@0: {-1, 6, 1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1}, // 5 - dependent vowels
sl@0: {-1, -1, 1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1}, // 6 - vowel mark
sl@0: {-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 4, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1}, // 7 - ZWJ, ZWNJ seen after a consonant: Modified by Nokia
sl@0: {-1, 6, 1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 4, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1}, // 8 - independent vowels that can take a virama
sl@0: {-1, 6, 1, -1, -1, -1, -1, -1, -1, -1, 10, 5, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1}, // 9 - first part of split vowel
sl@0: {-1, 6, 1, -1, -1, -1, -1, -1, -1, -1, -1, 5, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1}, // 10 - second part of split vowel
sl@0: {-1, 6, 1, -1, -1, -1, -1, -1, 5, 9, 5, 5, 13, -1, -1, 5, -1, 26, -1, -1, -1, 5, 5, 5}, // 11 - ct vr ct nu
sl@0: {-1, 6, 1, -1, -1, -1, -1, 11, 5, 9, 5, 5, 13, -1, -1, 5, -1, 26, -1, -1, -1, 5, 5, 5}, // 12 - ct vr ct
sl@0: {-1, -1, -1, -1, -1, 15, 14, -1, -1, -1, -1, -1, -1, 23, -1, -1, 15, -1, -1, -1, -1, -1, -1, -1}, // 13 - ct vr ct vr
sl@0: {-1, 6, 1, -1, -1, -1, -1, -1, 5, 9, 5, 5, 16, -1, -1, 5, -1, 26, -1, -1, -1, 5, 5, 5}, // 14 - ct vr ct vr ct nu
sl@0: {-1, 6, 1, -1, -1, -1, -1, 14, 5, 9, 5, 5, 16, -1, -1, 5, -1, 26, -1, -1, -1, 5, 5, 5}, // 15 - ct vr ct vr ct
sl@0: {-1, -1, -1, -1, -1, 18, 17, -1, -1, -1, -1, -1, -1, 24, -1, -1, 18, -1, -1, -1, -1, -1, -1, -1}, // 16 - ct vr ct vr ct vr
sl@0: {-1, 6, 1, -1, -1, -1, -1, -1, 5, 9, 5, 5, 19, -1, -1, 5, -1, 26, -1, -1, -1, 5, 5, 5}, // 17 - ct vr ct vr ct vr ct nu
sl@0: {-1, 6, 1, -1, -1, -1, -1, 17, 5, 9, 5, 5, 19, -1, -1, 5, -1, 26, -1, -1, -1, 5, 5, 5}, // 18 - ct vr ct vr ct vr ct
sl@0: {-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1}, // 19 - ct vr ct vr ct vr ct vr
sl@0: {-1, 6, 1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 21, -1, 5, -1, -1, -1, -1, -1, -1, -1, -1}, // 20 - devanagari independent vowel A that can take a modifier - CANDRA E: Added by Nokia
sl@0: {-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 5, -1, -1, -1, -1, -1, -1, -1, -1}, // 21 - ZWJ after independent vowel A: Added by Nokia
sl@0: {-1, 6, -1, -1, -1, 12, 11, -1, -1, -1, -1, -1, -1, -1, -1, -1, 12, -1, -1, -1, -1, -1, -1, -1}, // 22 - ZWJ after a virama: Added by Nokia
sl@0: {-1, 6, -1, -1, -1, 15, 14, -1, -1, -1, -1, -1, -1, -1, -1, -1, 15, -1, -1, -1, -1, -1, -1, -1}, // 23 - ZWJ after a ct vr ct vr: Added by Nokia
sl@0: {-1, 6, -1, -1, -1, 18, 17, -1, -1, -1, -1, -1, -1, -1, -1, -1, 18, -1, -1, -1, -1, -1, -1, -1}, // 24 - ZWJ after a ct vr ct vr ct vr: Added by Nokia
sl@0: {-1, 6, 1, -1, -1, -1, -1, 2, 5, 9, 5, 5, 4, 7, -1, 5, -1, 26, -1, -1, -1, 5, 5, 5}, // 25 - Kannada consonant Ra: Added by Nokia
sl@0: {-1, 6, 1, -1, -1, -1, -1, -1, -1, -1, 10, 5, 4, 21, -1, 5, -1, -1, -1, -1, -1, -1, -1, -1}, // 26 - devanagari independent vowel A that can take a modifier - CANDRA E: Added by Nokia
sl@0: {-1, 6, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 5, -1, -1}, // 27 - Gurmukhi bearer ARA and independent A
sl@0: {-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 5, -1}, // 28 - Gurmukhi bearer IRI
sl@0: {-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 5}, // 29 - Gurmukhi bearer URA
sl@0: };
sl@0:
sl@0:
sl@0: const LETag *IndicReordering::getFeatureOrder()
sl@0: {
sl@0: return featureOrder;
sl@0: }
sl@0:
sl@0: le_int32 IndicReordering::findSyllable(const IndicClassTable *classTable, const LEUnicode *chars, le_int32 prev, le_int32 charCount)
sl@0: {
sl@0: le_int32 cursor = prev;
sl@0: le_int8 state = 0;
sl@0:
sl@0: while (cursor < charCount) {
sl@0: IndicClassTable::CharClass charClass = classTable->getCharClass(chars[cursor]);
sl@0:
sl@0: state = stateTable[state][charClass & CF_CLASS_MASK];
sl@0:
sl@0: if (state < 0) {
sl@0: break;
sl@0: }
sl@0:
sl@0: cursor += 1;
sl@0: }
sl@0:
sl@0: return cursor;
sl@0: }
sl@0:
sl@0: le_int32 IndicReordering::reorder(const LEUnicode *chars, le_int32 charCount, le_int32 scriptCode,
sl@0: LEUnicode *outChars, LEGlyphStorage &glyphStorage,
sl@0: MPreFixups **outMPreFixups, LEErrorCode& success)
sl@0: {
sl@0: MPreFixups *mpreFixups = NULL;
sl@0: const IndicClassTable *classTable = IndicClassTable::getScriptClassTable(scriptCode);
sl@0:
sl@0: if (classTable->scriptFlags & SF_MPRE_FIXUP) {
sl@0: mpreFixups = new MPreFixups(charCount);
sl@0: if (!mpreFixups) {
sl@0: success = LE_MEMORY_ALLOCATION_ERROR;
sl@0: return 0;
sl@0: }
sl@0: }
sl@0:
sl@0: ReorderingOutput output(outChars, glyphStorage, mpreFixups);
sl@0: le_int32 i, prev = 0;
sl@0:
sl@0: while (prev < charCount) {
sl@0: le_int32 syllable = findSyllable(classTable, chars, prev, charCount);
sl@0: le_int32 matra, markStart = syllable;
sl@0:
sl@0: output.reset();
sl@0:
sl@0: if (classTable->isStressMark(chars[markStart - 1])) {
sl@0: markStart -= 1;
sl@0: output.noteStressMark(classTable, chars[markStart], markStart, &tagArray[1]);
sl@0: }
sl@0:
sl@0: if (prev != markStart && classTable->isVowelModifier(chars[markStart - 1])) {
sl@0: markStart -= 1;
sl@0: output.noteVowelModifier(classTable, chars[markStart], markStart, &tagArray[1]);
sl@0: }
sl@0:
sl@0: matra = markStart - 1;
sl@0:
sl@0: // Added by Nokia. Workaround ligature substitution: Bearer URA + Dependent Vowel -> Independent Vowel
sl@0: // For these sequences, no need to write matra since it will be replaced.
sl@0: if (!(chars[prev] == KBearURA && (prev + 1 < syllable) &&
sl@0: (chars[prev+1] == KDepVowelU || chars[prev+1] == KDepVowelUU || chars[prev+1] == KDepVowelOO ))) {
sl@0:
sl@0: while (prev <= matra && output.noteMatra(classTable, chars[matra], matra, &tagArray[1])) {
sl@0: matra -= 1;
sl@0: }
sl@0: }
sl@0:
sl@0:
sl@0: switch (classTable->getCharClass(chars[prev]) & CF_CLASS_MASK) {
sl@0: case CC_RESERVED:
sl@0: case CC_INDEPENDENT_VOWEL:
sl@0: case CC_ZERO_WIDTH_MARK:
sl@0: case CC_INDEPENDENT_VOWEL_A: // Added by Nokia
sl@0: case CC_INDEPENDENT_VOWEL_TAMIL_O: // Added by Nokia
sl@0: case CC_GUR_BEARER_U: //Added by Nokia
sl@0:
sl@0: // Added by Nokia. Workaround ligature substitution: Bearer URA + Dependent Vowel -> Independent Vowel
sl@0: if (chars[prev] == KBearURA && (prev + 1 < markStart)){
sl@0: if (chars[prev+1] == KDepVowelU){
sl@0: output.writeChar(KVowelU, prev, &tagArray[1]);
sl@0: prev = prev+2;
sl@0: }
sl@0: else if (chars[prev+1] == KDepVowelUU){
sl@0: output.writeChar(KVowelUU, prev, &tagArray[1]);
sl@0: prev = prev+2;
sl@0: }
sl@0: else if (chars[prev+1] == KDepVowelOO){
sl@0: output.writeChar(KVowelOO, prev, &tagArray[1]);
sl@0: prev = prev+2;
sl@0: }
sl@0: }
sl@0:
sl@0: for (i = prev; i < syllable; i += 1) {
sl@0: output.writeChar(chars[i], i, &tagArray[1]);
sl@0: }
sl@0:
sl@0: break;
sl@0:
sl@0: case CC_NUKTA:
sl@0: case CC_VIRAMA:
sl@0: output.writeChar(C_DOTTED_CIRCLE, prev, &tagArray[1]);
sl@0: output.writeChar(chars[prev], prev, &tagArray[1]);
sl@0: break;
sl@0:
sl@0: case CC_DEPENDENT_VOWEL:
sl@0: case CC_DEPENDENT_VOWEL_CANDRA_E: // Added by Nokia
sl@0: case CC_SPLIT_VOWEL_PIECE_1:
sl@0: case CC_SPLIT_VOWEL_PIECE_2:
sl@0: case CC_SPLIT_VOWEL_PIECE_3:
sl@0: case CC_VOWEL_MODIFIER:
sl@0: case CC_STRESS_MARK:
sl@0: case CC_GUR_DEPENDENT_VOWEL_A:
sl@0: case CC_GUR_DEPENDENT_VOWEL_I:
sl@0: case CC_GUR_DEPENDENT_VOWEL_U:
sl@0: output.writeMpre();
sl@0:
sl@0: output.writeChar(C_DOTTED_CIRCLE, prev, &tagArray[1]);
sl@0:
sl@0: output.writeMbelow();
sl@0: output.writeSMbelow();
sl@0: output.writeMabove();
sl@0:
sl@0: if ((classTable->scriptFlags & SF_MATRAS_AFTER_BASE) != 0) {
sl@0: output.writeMpost();
sl@0: }
sl@0:
sl@0: if ((classTable->scriptFlags & SF_REPH_AFTER_BELOW) != 0) {
sl@0: output.writeVMabove();
sl@0: output.writeSMabove(); // FIXME: there are no SM's in these scripts...
sl@0: }
sl@0:
sl@0: if ((classTable->scriptFlags & SF_MATRAS_AFTER_BASE) == 0) {
sl@0: output.writeMpost();
sl@0: }
sl@0:
sl@0: output.writeLengthMark();
sl@0:
sl@0: if ((classTable->scriptFlags & SF_REPH_AFTER_BELOW) == 0) {
sl@0: output.writeVMabove();
sl@0: output.writeSMabove();
sl@0: }
sl@0:
sl@0: output.writeVMpost();
sl@0: break;
sl@0:
sl@0: case CC_INDEPENDENT_VOWEL_2:
sl@0: case CC_CONSONANT:
sl@0: case CC_CONSONANT_WITH_NUKTA:
sl@0: case CC_CONSONANT_KANNADA_BENGALI_RA:
sl@0: case CC_GUR_BEARER_A: // Added by Nokia
sl@0: case CC_GUR_BEARER_I: // Added by Nokia
sl@0: {
sl@0: le_uint32 length = markStart - prev;
sl@0: le_int32 lastConsonant = markStart - 1;
sl@0: le_int32 baseLimit = prev;
sl@0:
sl@0: // Check for REPH at front of syllable
sl@0: if (length > 2 && classTable->isReph(chars[prev]) && classTable->isVirama(chars[prev + 1])) {
sl@0: baseLimit += 2;
sl@0:
sl@0: // Check for eyelash RA, if the script supports it
sl@0: if ((classTable->scriptFlags & SF_EYELASH_RA) != 0 &&
sl@0: chars[baseLimit] == C_SIGN_ZWJ) {
sl@0: if (length > 3) {
sl@0: baseLimit += 1;
sl@0: } else {
sl@0: baseLimit -= 2;
sl@0: }
sl@0: }
sl@0: }
sl@0:
sl@0: while (lastConsonant > baseLimit && !classTable->isConsonant(chars[lastConsonant])) {
sl@0: lastConsonant -= 1;
sl@0: }
sl@0:
sl@0: le_int32 baseConsonant = lastConsonant;
sl@0: le_int32 postBase = lastConsonant + 1;
sl@0: le_int32 postBaseLimit = classTable->scriptFlags & SF_POST_BASE_LIMIT_MASK;
sl@0: le_bool seenVattu = FALSE;
sl@0: le_bool seenBelowBaseForm = FALSE;
sl@0:
sl@0: if (postBase < markStart && classTable->isNukta(chars[postBase])) {
sl@0: postBase += 1;
sl@0: }
sl@0:
sl@0: while (baseConsonant > baseLimit) {
sl@0: IndicClassTable::CharClass charClass = classTable->getCharClass(chars[baseConsonant]);
sl@0:
sl@0: if (IndicClassTable::isConsonant(charClass)) {
sl@0: IndicClassTable::CharClass nextCharClass
sl@0: = classTable->getCharClass(chars[baseConsonant + 1]);
sl@0: le_bool hasNukta = IndicClassTable::isNukta(nextCharClass);
sl@0: // Consonants with nuktas never have post-base or below-base forms
sl@0: le_bool hasPostBaseForm = !hasNukta
sl@0: && IndicClassTable::hasPostBaseForm(charClass);
sl@0: le_bool hasBelowBaseForm = !hasNukta
sl@0: && IndicClassTable::hasBelowBaseForm(charClass);
sl@0:
sl@0: // Added by Nokia -- A Bengali Ya should only have a post base form if it is the last component in a consonant cluster.
sl@0: if (IndicClassTable::isBengaliYa(charClass) && (baseConsonant != lastConsonant))
sl@0: hasPostBaseForm = FALSE;
sl@0:
sl@0: // Added by Nokia 1922 mlyl: YAKAR/VAKAR/RAKAR are not formed after LLA/LLLA/RRA, but when LLA/LLLA/RRA is one of combination ligature?-->
sl@0: if ((chars[baseConsonant] == CC_CONSONANT_MLYL_YA || chars[baseConsonant] == CC_CONSONANT_MLYL_VA || chars[baseConsonant] == CC_CONSONANT_MLYL_RA)
sl@0: && baseConsonant >=2
sl@0: && (chars[baseConsonant - 2] == CC_CONSONANT_MLYL_LLA || chars[baseConsonant - 2] == CC_CONSONANT_MLYL_LLLA || chars[baseConsonant - 2] == CC_CONSONANT_MLYL_RRA))
sl@0: hasPostBaseForm = FALSE;
sl@0:
sl@0: // D15 + D4D + D2F/D35 + D4D + D30: YAKAR/VAKAR not formed
sl@0: if ((chars[baseConsonant] == CC_CONSONANT_MLYL_YA || chars[baseConsonant] == CC_CONSONANT_MLYL_VA)
sl@0: && lastConsonant >= 4
sl@0: && chars[lastConsonant] == CC_CONSONANT_MLYL_RA)
sl@0: hasPostBaseForm = FALSE;
sl@0: // <-- 1922 mlyl
sl@0:
sl@0: if (postBaseLimit == 0 || seenVattu ||
sl@0: (baseConsonant > baseLimit && !classTable->isVirama(chars[baseConsonant - 1])) ||
sl@0: (!hasPostBaseForm && !hasBelowBaseForm)) {
sl@0: break;
sl@0: }
sl@0:
sl@0: // Consonants with nuktas are never vattus
sl@0: seenVattu = !hasNukta && IndicClassTable::isVattu(charClass);
sl@0:
sl@0: if (hasPostBaseForm) {
sl@0: if (seenBelowBaseForm) {
sl@0: break;
sl@0: }
sl@0:
sl@0: postBase = baseConsonant;
sl@0: } else if (hasBelowBaseForm) {
sl@0: seenBelowBaseForm = TRUE;
sl@0: }
sl@0:
sl@0: postBaseLimit -= 1;
sl@0: }
sl@0:
sl@0: baseConsonant -= 1;
sl@0: }
sl@0:
sl@0: // Write Mpre
sl@0: output.writeMpre();
sl@0:
sl@0: // Write eyelash RA
sl@0: // NOTE: baseLimit == prev + 3 iff eyelash RA present...
sl@0: if (baseLimit == prev + 3) {
sl@0: output.writeChar(chars[prev], prev, &tagArray[2]);
sl@0: output.writeChar(chars[prev + 1], prev + 1, &tagArray[2]);
sl@0: output.writeChar(chars[prev + 2], prev + 2, &tagArray[2]);
sl@0: }
sl@0:
sl@0: // write any pre-base consonants
sl@0: le_bool supressVattu = TRUE;
sl@0:
sl@0: for (i = baseLimit; i < baseConsonant; i += 1) {
sl@0: LEUnicode ch = chars[i];
sl@0: // Don't put 'blwf' on first consonant.
sl@0: const LETag *tag = (i == baseLimit? &tagArray[2] : &tagArray[1]);
sl@0: IndicClassTable::CharClass charClass = classTable->getCharClass(ch);
sl@0: IndicClassTable::CharClass nextCharClass =
sl@0: classTable->getCharClass(chars[i + 1]);
sl@0:
sl@0: // Nokia: Added a check to see if the character that is two characters after
sl@0: // the RA is a ZWJ, which then doesn't allow the RA to become a VATTU and allows
sl@0: // the formation of an EYELASH RA
sl@0: le_bool isGenuineVattu = IndicClassTable::isVattu(charClass)
sl@0: && !IndicClassTable::isNukta(nextCharClass)
sl@0: && chars[i + 2] != C_SIGN_ZWJ;
sl@0:
sl@0: if (IndicClassTable::isConsonant(charClass)) {
sl@0: if (isGenuineVattu && supressVattu) {
sl@0: tag = &tagArray[4];
sl@0: }
sl@0:
sl@0: supressVattu = isGenuineVattu;
sl@0: } else if (IndicClassTable::isVirama(charClass) && chars[i + 1] == C_SIGN_ZWNJ)
sl@0: {
sl@0: tag = &tagArray[4];
sl@0: }
sl@0:
sl@0: output.writeChar(ch, i, tag);
sl@0: }
sl@0:
sl@0: le_int32 bcSpan = baseConsonant + 1;
sl@0:
sl@0: if (bcSpan < markStart && classTable->isNukta(chars[bcSpan])) {
sl@0: bcSpan += 1;
sl@0: }
sl@0:
sl@0: if (baseConsonant == lastConsonant && bcSpan < markStart && classTable->isVirama(chars[bcSpan])) {
sl@0: bcSpan += 1;
sl@0:
sl@0: if (bcSpan < markStart && chars[bcSpan] == C_SIGN_ZWNJ) {
sl@0: bcSpan += 1;
sl@0: }
sl@0: }
sl@0:
sl@0: // note the base consonant for post-GSUB fixups
sl@0: output.noteBaseConsonant();
sl@0:
sl@0: // write base consonant
sl@0: for (i = baseConsonant; i < bcSpan; i += 1) {
sl@0: // 1922 mlyl -->
sl@0: if (i >= 2 && chars[i] == C_SIGN_ZWJ && classTable->isMlylChillu(chars[i-2]))
sl@0: output.writeChar(chars[i], i, &tagArray[2]);
sl@0: else
sl@0: // <-- 1922 mlyl
sl@0: output.writeChar(chars[i], i, &tagArray[4]);
sl@0: }
sl@0:
sl@0: //write Mbelow, SMbelow, Mabove, Mpost after base consonant with flag SF_MATRAS_AFTER_BASE setting
sl@0: if ((classTable->scriptFlags & SF_MATRAS_AFTER_BASE) != 0) {
sl@0: output.writeMbelow();
sl@0: output.writeSMbelow(); // FIXME: there are no SMs in these scripts...
sl@0: output.writeMabove();
sl@0: output.writeMpost();
sl@0: }
sl@0:
sl@0: // write below-base consonants
sl@0: if (baseConsonant != lastConsonant) {
sl@0: for (i = bcSpan + 1; i < postBase; i += 1) {
sl@0: output.writeChar(chars[i], i, &tagArray[1]);
sl@0: }
sl@0:
sl@0: if (postBase > lastConsonant) {
sl@0: // write halant that was after base consonant
sl@0: output.writeChar(chars[bcSpan], bcSpan, &tagArray[1]);
sl@0: }
sl@0: }
sl@0:
sl@0: // Added by Nokia: write Mbelow, SMbelow, Mabove after below-base consonants without flag SF_MATRAS_AFTER_BASE or SF_MATRAS_AFTER_POSTBASE
sl@0: if (((classTable->scriptFlags & SF_MATRAS_AFTER_BASE) == 0) &&
sl@0: ((classTable->scriptFlags & SF_MATRAS_AFTER_POSTBASE) == 0)) {
sl@0: output.writeMbelow();
sl@0: output.writeSMbelow();
sl@0: output.writeMabove();
sl@0: }
sl@0:
sl@0: if ((classTable->scriptFlags & SF_REPH_AFTER_BELOW) != 0) {
sl@0: if (baseLimit == prev + 2) {
sl@0: output.writeChar(chars[prev], prev, &tagArray[0]);
sl@0: output.writeChar(chars[prev + 1], prev + 1, &tagArray[0]);
sl@0: }
sl@0:
sl@0: output.writeVMabove();
sl@0: output.writeSMabove(); // FIXME: there are no SM's in these scripts...
sl@0: }
sl@0:
sl@0: // write post-base consonants
sl@0: // FIXME: does this put the right tags on post-base consonants?
sl@0: if (baseConsonant != lastConsonant) {
sl@0: if (postBase <= lastConsonant) {
sl@0: for (i = postBase; i <= lastConsonant; i += 1) {
sl@0: output.writeChar(chars[i], i, &tagArray[3]);
sl@0: }
sl@0:
sl@0: // write halant that was after base consonant
sl@0: output.writeChar(chars[bcSpan], bcSpan, &tagArray[1]);
sl@0: }
sl@0:
sl@0: // write the training halant, if there is one
sl@0: if (lastConsonant < matra && classTable->isVirama(chars[matra])) {
sl@0: output.writeChar(chars[matra], matra, &tagArray[4]);
sl@0: }
sl@0: }
sl@0:
sl@0: // Added by Nokia: write Mbelow, SMbelow, Mabove and Mpost after post-base consonants with flag SF_MATRAS_AFTER_POSTBASE
sl@0: if ((classTable->scriptFlags & SF_MATRAS_AFTER_POSTBASE) != 0) {
sl@0: output.writeMbelow();
sl@0: output.writeSMbelow();
sl@0: output.writeMabove();
sl@0: output.writeMpost();
sl@0: }
sl@0:
sl@0: //Added by Nokia: write Mpost finally
sl@0: if (((classTable->scriptFlags & SF_MATRAS_AFTER_BASE) == 0) &&
sl@0: ((classTable->scriptFlags & SF_MATRAS_AFTER_POSTBASE) == 0)) {
sl@0: output.writeMpost();
sl@0: }
sl@0:
sl@0: output.writeLengthMark();
sl@0:
sl@0: // write reph
sl@0: if ((classTable->scriptFlags & SF_REPH_AFTER_BELOW) == 0) {
sl@0: if (baseLimit == prev + 2) {
sl@0: output.writeChar(chars[prev], prev, &tagArray[0]);
sl@0: output.writeChar(chars[prev + 1], prev + 1, &tagArray[0]);
sl@0: }
sl@0:
sl@0: output.writeVMabove();
sl@0: output.writeSMabove();
sl@0: }
sl@0:
sl@0: output.writeVMpost();
sl@0:
sl@0: break;
sl@0: }
sl@0:
sl@0: default:
sl@0: break;
sl@0: }
sl@0:
sl@0: prev = syllable;
sl@0: }
sl@0:
sl@0: *outMPreFixups = mpreFixups;
sl@0:
sl@0: return output.getOutputIndex();
sl@0: }
sl@0:
sl@0: void IndicReordering::adjustMPres(MPreFixups *mpreFixups,
sl@0: LEGlyphStorage &glyphStorage, LEErrorCode& success)
sl@0: {
sl@0: if (LE_SUCCESS(success) && mpreFixups != NULL) {
sl@0: mpreFixups->apply(glyphStorage, success);
sl@0:
sl@0: delete mpreFixups;
sl@0: }
sl@0: }
sl@0:
sl@0: U_NAMESPACE_END