ICU-3274 Initial checkin of Hangul OpenType code.
X-SVN-Rev: 19227
This commit is contained in:
Eric Mader
parent
422d7dbe64
commit
10b15d2a44
338
icu4c/source/layout/HangulLayoutEngine.cpp
Normal file
338
icu4c/source/layout/HangulLayoutEngine.cpp
Normal file
@ -0,0 +1,338 @@
|
||||
/*
|
||||
* HangulLayoutEngine.cpp: OpenType processing for Han fonts.
|
||||
*
|
||||
* (C) Copyright IBM Corp. 1998-2006 - All Rights Reserved.
|
||||
*/
|
||||
|
||||
#include "LETypes.h"
|
||||
#include "LEScripts.h"
|
||||
#include "LELanguages.h"
|
||||
|
||||
#include "LayoutEngine.h"
|
||||
#include "OpenTypeLayoutEngine.h"
|
||||
#include "HangulLayoutEngine.h"
|
||||
#include "ScriptAndLanguageTags.h"
|
||||
#include "LEGlyphStorage.h"
|
||||
#include "OpenTypeTables.h"
|
||||
|
||||
U_NAMESPACE_BEGIN
|
||||
|
||||
UOBJECT_DEFINE_RTTI_IMPLEMENTATION(HangulOpenTypeLayoutEngine)
|
||||
|
||||
|
||||
#define FEATURE_MAP(name) {name ## FeatureTag, name ## FeatureMask}
|
||||
|
||||
#define LJMO_FIRST 0x1100
|
||||
#define LJMO_LAST 0x1159
|
||||
#define LJMO_FILL 0x115F
|
||||
#define LJMO_COUNT 19
|
||||
|
||||
#define VJMO_FIRST 0x1161
|
||||
#define VJMO_LAST 0x11A2
|
||||
#define VJMO_FILL 0x1160
|
||||
#define VJMO_COUNT 21
|
||||
|
||||
#define TJMO_FIRST 0x11A7
|
||||
#define TJMO_LAST 0x11F9
|
||||
#define TJMO_COUNT 28
|
||||
|
||||
#define HSYL_FIRST 0xAC00
|
||||
#define HSYL_COUNT 11172
|
||||
#define HSYL_LVCNT (VJMO_COUNT * TJMO_COUNT)
|
||||
|
||||
// Character classes
|
||||
enum
|
||||
{
|
||||
CC_L = 0,
|
||||
CC_V,
|
||||
CC_T,
|
||||
CC_LV,
|
||||
CC_LVT,
|
||||
CC_X,
|
||||
CC_COUNT
|
||||
};
|
||||
|
||||
// Action flags
|
||||
#define AF_L 1
|
||||
#define AF_V 2
|
||||
#define AF_T 4
|
||||
|
||||
// Actions
|
||||
#define a_N 0
|
||||
#define a_L (AF_L)
|
||||
#define a_V (AF_V)
|
||||
#define a_T (AF_T)
|
||||
#define a_VT (AF_V | AF_T)
|
||||
#define a_LV (AF_L | AF_V)
|
||||
#define a_LVT (AF_L | AF_V | AF_T)
|
||||
|
||||
typedef struct
|
||||
{
|
||||
int32_t newState;
|
||||
int32_t actionFlags;
|
||||
} StateTransition;
|
||||
|
||||
static const StateTransition stateTable[][CC_COUNT] =
|
||||
{
|
||||
// L V T LV LVT X
|
||||
{ {1, a_L}, {2, a_LV}, {3, a_LVT}, {2, a_LV}, {3, a_LVT}, {4, a_T}}, // 0 - start
|
||||
{ {1, a_L}, {2, a_V}, {3, a_VT}, {2, a_LV}, {3, a_LVT}, {-1, a_V}}, // 1 - L+
|
||||
{{-1, a_N}, {2, a_V}, {3, a_T}, {-1, a_N}, {-1, a_N}, {-1, a_N}}, // 2 - L+V+
|
||||
{{-1, a_N}, {-1, a_N}, {3, a_T}, {-1, a_N}, {-1, a_N}, {-1, a_N}}, // 3 - L+V+T*
|
||||
{{-1, a_N}, {-1, a_N}, {-1, a_N}, {-1, a_N}, {-1, a_N}, {4, a_T}} // 4 - X+
|
||||
};
|
||||
|
||||
|
||||
#define ccmpFeatureTag LE_CCMP_FEATURE_TAG
|
||||
#define ljmoFeatureTag LE_LJMO_FEATURE_TAG
|
||||
#define vjmoFeatureTag LE_VJMO_FEATURE_TAG
|
||||
#define tjmoFeatureTag LE_TJMO_FEATURE_TAG
|
||||
|
||||
#define ccmpFeatureMask 0x80000000UL
|
||||
#define ljmoFeatureMask 0x40000000UL
|
||||
#define vjmoFeatureMask 0x20000000UL
|
||||
#define tjmoFeatureMask 0x10000000UL
|
||||
|
||||
static const FeatureMap featureMap[] =
|
||||
{
|
||||
{ccmpFeatureTag, ccmpFeatureMask},
|
||||
{ljmoFeatureTag, ljmoFeatureMask},
|
||||
{vjmoFeatureTag, vjmoFeatureMask},
|
||||
{tjmoFeatureTag, tjmoFeatureMask}
|
||||
};
|
||||
|
||||
static const le_int32 featureMapCount = LE_ARRAY_SIZE(featureMap);
|
||||
|
||||
#define nullFeatures 0
|
||||
#define ljmoFeatures (ccmpFeatureMask | ljmoFeatureMask)
|
||||
#define vjmoFeatures (ccmpFeatureMask | vjmoFeatureMask | ljmoFeatureMask | tjmoFeatureMask)
|
||||
#define tjmoFeatures (ccmpFeatureMask | tjmoFeatureMask | ljmoFeatureMask | vjmoFeatureMask)
|
||||
|
||||
static le_int32 compose(LEUnicode lead, LEUnicode vowel, LEUnicode trail, LEUnicode &syllable)
|
||||
{
|
||||
le_int32 lIndex = lead - LJMO_FIRST;
|
||||
le_int32 vIndex = vowel - VJMO_FIRST;
|
||||
le_int32 tIndex = trail - TJMO_FIRST;
|
||||
le_int32 result = 3;
|
||||
|
||||
if ((lIndex < 0 || lIndex >= LJMO_COUNT ) || (vIndex < 0 || vIndex >= VJMO_COUNT)) {
|
||||
return 0;
|
||||
}
|
||||
|
||||
if (tIndex <= 0 || tIndex >= TJMO_COUNT) {
|
||||
tIndex = 0;
|
||||
result = 2;
|
||||
}
|
||||
|
||||
syllable = (LEUnicode) ((lIndex * VJMO_COUNT + vIndex) * TJMO_COUNT + tIndex + HSYL_FIRST);
|
||||
|
||||
return result;
|
||||
}
|
||||
|
||||
static le_int32 decompose(LEUnicode syllable, LEUnicode &lead, LEUnicode &vowel, LEUnicode &trail)
|
||||
{
|
||||
le_int32 sIndex = syllable - HSYL_FIRST;
|
||||
|
||||
if (sIndex < 0 || sIndex >= HSYL_COUNT) {
|
||||
return 0;
|
||||
}
|
||||
|
||||
lead = LJMO_FIRST + (sIndex / HSYL_LVCNT);
|
||||
vowel = VJMO_FIRST + (sIndex % HSYL_LVCNT) / TJMO_COUNT;
|
||||
trail = TJMO_FIRST + (sIndex % TJMO_COUNT);
|
||||
|
||||
if (trail == TJMO_FIRST) {
|
||||
return 2;
|
||||
}
|
||||
|
||||
return 3;
|
||||
}
|
||||
|
||||
static le_int32 getCharClass(LEUnicode ch, LEUnicode &lead, LEUnicode &vowel, LEUnicode &trail)
|
||||
{
|
||||
lead = LJMO_FILL;
|
||||
vowel = VJMO_FILL;
|
||||
trail = TJMO_FIRST;
|
||||
|
||||
if (ch >= LJMO_FIRST && ch <= LJMO_LAST) {
|
||||
lead = ch;
|
||||
return CC_L;
|
||||
}
|
||||
|
||||
if (ch >= VJMO_FIRST && ch <= VJMO_LAST) {
|
||||
vowel = ch;
|
||||
return CC_V;
|
||||
}
|
||||
|
||||
if (ch > TJMO_FIRST && ch <= TJMO_LAST) {
|
||||
trail = ch;
|
||||
return CC_T;
|
||||
}
|
||||
|
||||
le_int32 c = decompose(ch, lead, vowel, trail);
|
||||
|
||||
if (c == 2) {
|
||||
return CC_LV;
|
||||
}
|
||||
|
||||
if (c == 3) {
|
||||
return CC_LVT;
|
||||
}
|
||||
|
||||
trail = ch;
|
||||
return CC_X;
|
||||
}
|
||||
|
||||
HangulOpenTypeLayoutEngine::HangulOpenTypeLayoutEngine(const LEFontInstance *fontInstance, le_int32 scriptCode, le_int32 languageCode,
|
||||
le_int32 typoFlags, const GlyphSubstitutionTableHeader *gsubTable)
|
||||
: OpenTypeLayoutEngine(fontInstance, scriptCode, korLanguageCode, typoFlags, gsubTable)
|
||||
{
|
||||
fFeatureMap = featureMap;
|
||||
fFeatureMapCount = featureMapCount;
|
||||
fFeatureOrder = TRUE;
|
||||
}
|
||||
|
||||
HangulOpenTypeLayoutEngine::HangulOpenTypeLayoutEngine(const LEFontInstance *fontInstance, le_int32 scriptCode, le_int32 languageCode,
|
||||
le_int32 typoFlags)
|
||||
: OpenTypeLayoutEngine(fontInstance, scriptCode, korLanguageCode, typoFlags)
|
||||
{
|
||||
fFeatureMap = featureMap;
|
||||
fFeatureMapCount = featureMapCount;
|
||||
fFeatureOrder = TRUE;
|
||||
}
|
||||
|
||||
HangulOpenTypeLayoutEngine::~HangulOpenTypeLayoutEngine()
|
||||
{
|
||||
// nothing to do
|
||||
}
|
||||
|
||||
le_int32 HangulOpenTypeLayoutEngine::characterProcessing(const LEUnicode chars[], le_int32 offset, le_int32 count, le_int32 max, le_bool rightToLeft,
|
||||
LEUnicode *&outChars, LEGlyphStorage &glyphStorage, LEErrorCode &success)
|
||||
{
|
||||
if (LE_FAILURE(success)) {
|
||||
return 0;
|
||||
}
|
||||
|
||||
if (chars == NULL || offset < 0 || count < 0 || max < 0 || offset >= max || offset + count > max) {
|
||||
success = LE_ILLEGAL_ARGUMENT_ERROR;
|
||||
return 0;
|
||||
}
|
||||
|
||||
le_int32 worstCase = count * 3;
|
||||
|
||||
outChars = LE_NEW_ARRAY(LEUnicode, worstCase);
|
||||
|
||||
if (outChars == NULL) {
|
||||
success = LE_MEMORY_ALLOCATION_ERROR;
|
||||
return 0;
|
||||
}
|
||||
|
||||
glyphStorage.allocateGlyphArray(worstCase, rightToLeft, success);
|
||||
glyphStorage.allocateAuxData(success);
|
||||
|
||||
if (LE_FAILURE(success)) {
|
||||
LE_DELETE_ARRAY(outChars);
|
||||
return 0;
|
||||
}
|
||||
|
||||
le_int32 outCharCount = 0;
|
||||
le_int32 limit = offset + count;
|
||||
le_int32 i = offset;
|
||||
|
||||
while (i < limit) {
|
||||
le_int32 state = 0;
|
||||
le_int32 inStart = i;
|
||||
le_int32 outStart = outCharCount;
|
||||
|
||||
while( i < limit) {
|
||||
LEUnicode lead = 0;
|
||||
LEUnicode vowel = 0;
|
||||
LEUnicode trail = 0;
|
||||
int32_t chClass = getCharClass(chars[i], lead, vowel, trail);
|
||||
const StateTransition transition = stateTable[state][chClass];
|
||||
|
||||
if (chClass == CC_X) {
|
||||
/* Any character of type X will be stored as a trail jamo */
|
||||
if ((transition.actionFlags & AF_T) != 0) {
|
||||
outChars[outCharCount] = trail;
|
||||
glyphStorage.setCharIndex(outCharCount, i-offset, success);
|
||||
glyphStorage.setAuxData(outCharCount++, nullFeatures, success);
|
||||
}
|
||||
} else {
|
||||
/* Any Hangul will be fully decomposed. Output the decomposed characters. */
|
||||
if ((transition.actionFlags & AF_L) != 0) {
|
||||
outChars[outCharCount] = lead;
|
||||
glyphStorage.setCharIndex(outCharCount, i-offset, success);
|
||||
glyphStorage.setAuxData(outCharCount++, ljmoFeatures, success);
|
||||
}
|
||||
|
||||
if ((transition.actionFlags & AF_V) != 0) {
|
||||
outChars[outCharCount] = vowel;
|
||||
glyphStorage.setCharIndex(outCharCount, i-offset, success);
|
||||
glyphStorage.setAuxData(outCharCount++, vjmoFeatures, success);
|
||||
}
|
||||
|
||||
if ((transition.actionFlags & AF_T) != 0) {
|
||||
outChars[outCharCount] = trail;
|
||||
glyphStorage.setCharIndex(outCharCount, i-offset, success);
|
||||
glyphStorage.setAuxData(outCharCount++, tjmoFeatures, success);
|
||||
}
|
||||
}
|
||||
|
||||
state = transition.newState;
|
||||
|
||||
/* Negative next state means stop. */
|
||||
if (state < 0) {
|
||||
break;
|
||||
}
|
||||
|
||||
i += 1;
|
||||
}
|
||||
|
||||
le_int32 inLength = i - inStart;
|
||||
le_int32 outLength = outCharCount - outStart;
|
||||
|
||||
/*
|
||||
* See if the syllable can be composed into a single character. There are 5
|
||||
* possible cases:
|
||||
*
|
||||
* Input Decomposed to Compose to
|
||||
* LV L, V LV
|
||||
* LVT L, V, T LVT
|
||||
* L, V L, V LV, DEL
|
||||
* LV, T L, V, T LVT, DEL
|
||||
* L, V, T L, V, T LVT, DEL, DEL
|
||||
*/
|
||||
if ((inLength >= 1 && inLength <= 3) && (outLength == 2 || outLength == 3)) {
|
||||
LEUnicode syllable = 0x0000;
|
||||
LEUnicode lead = outChars[outStart];
|
||||
LEUnicode vowel = outChars[outStart + 1];
|
||||
LEUnicode trail = outLength == 3? outChars[outStart + 2] : TJMO_FIRST;
|
||||
|
||||
/*
|
||||
* If the composition consumes the whole decomposed syllable,
|
||||
* we can use it.
|
||||
*/
|
||||
if (compose(lead, vowel, trail, syllable) == outLength) {
|
||||
outCharCount = outStart;
|
||||
outChars[outCharCount] = syllable;
|
||||
glyphStorage.setCharIndex(outCharCount, inStart-offset, success);
|
||||
glyphStorage.setAuxData(outCharCount++, nullFeatures, success);
|
||||
|
||||
/*
|
||||
* Replace the rest of the input characters with DEL.
|
||||
*/
|
||||
for(le_int32 d = inStart + 1; d < i; d += 1) {
|
||||
outChars[outCharCount] = 0xFFFF;
|
||||
glyphStorage.setCharIndex(outCharCount, d - offset, success);
|
||||
glyphStorage.setAuxData(outCharCount++, nullFeatures, success);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
glyphStorage.adoptGlyphCount(outCharCount);
|
||||
return outCharCount;
|
||||
}
|
||||
|
||||
U_NAMESPACE_END
|
||||
127
icu4c/source/layout/HangulLayoutEngine.h
Normal file
127
icu4c/source/layout/HangulLayoutEngine.h
Normal file
@ -0,0 +1,127 @@
|
||||
|
||||
/*
|
||||
*
|
||||
* (C) Copyright IBM Corp. 1998-2006 - All Rights Reserved
|
||||
*
|
||||
*/
|
||||
|
||||
#ifndef __HANGULAYOUTENGINE_H
|
||||
#define __HANGULAYOUTENGINE_H
|
||||
|
||||
#include "LETypes.h"
|
||||
#include "LEFontInstance.h"
|
||||
#include "LEGlyphFilter.h"
|
||||
#include "LayoutEngine.h"
|
||||
#include "OpenTypeLayoutEngine.h"
|
||||
|
||||
#include "GlyphSubstitutionTables.h"
|
||||
#include "GlyphDefinitionTables.h"
|
||||
#include "GlyphPositioningTables.h"
|
||||
|
||||
U_NAMESPACE_BEGIN
|
||||
|
||||
class MPreFixups;
|
||||
class LEGlyphStorage;
|
||||
|
||||
/**
|
||||
* This class implements OpenType layout for Indic OpenType fonts, as
|
||||
* specified by Microsoft in "Creating and Supporting OpenType Fonts for
|
||||
* Indic Scripts" (http://www.microsoft.com/typography/otspec/indicot/default.htm)
|
||||
*
|
||||
* This class overrides the characterProcessing method to do Indic character processing
|
||||
* and reordering, and the glyphProcessing method to implement post-GSUB processing for
|
||||
* left matras. (See the MS spec. for more details)
|
||||
*
|
||||
* @internal
|
||||
*/
|
||||
class HangulOpenTypeLayoutEngine : public OpenTypeLayoutEngine
|
||||
{
|
||||
public:
|
||||
/**
|
||||
* This is the main constructor. It constructs an instance of IndicOpenTypeLayoutEngine for
|
||||
* a particular font, script and language. It takes the GSUB table as a parameter since
|
||||
* LayoutEngine::layoutEngineFactory has to read the GSUB table to know that it has an
|
||||
* Indic OpenType font.
|
||||
*
|
||||
* @param fontInstance - the font
|
||||
* @param scriptCode - the script
|
||||
* @param langaugeCode - the language
|
||||
* @param gsubTable - the GSUB table
|
||||
*
|
||||
* @see LayoutEngine::layoutEngineFactory
|
||||
* @see OpenTypeLayoutEngine
|
||||
* @see ScriptAndLangaugeTags.h for script and language codes
|
||||
*
|
||||
* @internal
|
||||
*/
|
||||
HangulOpenTypeLayoutEngine(const LEFontInstance *fontInstance, le_int32 scriptCode, le_int32 languageCode,
|
||||
le_int32 typoFlags, const GlyphSubstitutionTableHeader *gsubTable);
|
||||
|
||||
/**
|
||||
* This constructor is used when the font requires a "canned" GSUB table which can't be known
|
||||
* until after this constructor has been invoked.
|
||||
*
|
||||
* @param fontInstance - the font
|
||||
* @param scriptCode - the script
|
||||
* @param langaugeCode - the language
|
||||
*
|
||||
* @see OpenTypeLayoutEngine
|
||||
* @see ScriptAndLangaugeTags.h for script and language codes
|
||||
*
|
||||
* @internal
|
||||
*/
|
||||
HangulOpenTypeLayoutEngine(const LEFontInstance *fontInstance, le_int32 scriptCode, le_int32 languageCode,
|
||||
le_int32 typoFlags);
|
||||
|
||||
/**
|
||||
* The destructor, virtual for correct polymorphic invocation.
|
||||
*
|
||||
* @internal
|
||||
*/
|
||||
virtual ~HangulOpenTypeLayoutEngine();
|
||||
|
||||
/**
|
||||
* ICU "poor man's RTTI", returns a UClassID for the actual class.
|
||||
*
|
||||
* @stable ICU 2.8
|
||||
*/
|
||||
virtual UClassID getDynamicClassID() const;
|
||||
|
||||
/**
|
||||
* ICU "poor man's RTTI", returns a UClassID for this class.
|
||||
*
|
||||
* @stable ICU 2.8
|
||||
*/
|
||||
static UClassID getStaticClassID();
|
||||
|
||||
protected:
|
||||
|
||||
/**
|
||||
* This method does Hangul OpenType character processing. It assigns the OpenType feature
|
||||
* tags to the characters, and may generate output characters which have been reordered. For
|
||||
* some Indic scripts, it may also split some vowels, resulting in more output characters
|
||||
* than input characters.
|
||||
*
|
||||
* Input parameters:
|
||||
* @param chars - the input character context
|
||||
* @param offset - the index of the first character to process
|
||||
* @param count - the number of characters to process
|
||||
* @param max - the number of characters in the input context
|
||||
* @param rightToLeft - <code>TRUE</code> if the characters are in a right to left directional run
|
||||
* @param glyphStorage - the glyph storage object. The glyph and character index arrays will be set.
|
||||
* the auxillary data array will be set to the feature tags.
|
||||
*
|
||||
* Output parameters:
|
||||
* @param success - set to an error code if the operation fails
|
||||
*
|
||||
* @return the output character count
|
||||
*
|
||||
* @internal
|
||||
*/
|
||||
virtual le_int32 characterProcessing(const LEUnicode chars[], le_int32 offset, le_int32 count, le_int32 max, le_bool rightToLeft,
|
||||
LEUnicode *&outChars, LEGlyphStorage &glyphStorage, LEErrorCode &success);
|
||||
};
|
||||
|
||||
U_NAMESPACE_END
|
||||
#endif
|
||||
|
||||
@ -1,7 +1,7 @@
|
||||
|
||||
/*
|
||||
*
|
||||
* (C) Copyright IBM Corp. 1998-2005 - All Rights Reserved
|
||||
* (C) Copyright IBM Corp. 1998-2006 - All Rights Reserved
|
||||
*
|
||||
*/
|
||||
|
||||
@ -13,6 +13,7 @@
|
||||
#include "ArabicLayoutEngine.h"
|
||||
#include "CanonShaping.h"
|
||||
#include "HanLayoutEngine.h"
|
||||
#include "HangulLayoutEngine.h"
|
||||
#include "IndicLayoutEngine.h"
|
||||
#include "KhmerLayoutEngine.h"
|
||||
#include "ThaiLayoutEngine.h"
|
||||
@ -463,6 +464,10 @@ LayoutEngine *LayoutEngine::layoutEngineFactory(const LEFontInstance *fontInstan
|
||||
result = new ArabicOpenTypeLayoutEngine(fontInstance, scriptCode, languageCode, typoFlags, gsubTable);
|
||||
break;
|
||||
|
||||
case hangScriptCode:
|
||||
result = new HangulOpenTypeLayoutEngine(fontInstance, scriptCode, languageCode, typoFlags, gsubTable);
|
||||
break;
|
||||
|
||||
case haniScriptCode:
|
||||
languageTag = OpenTypeLayoutEngine::getLangSysTag(languageCode);
|
||||
|
||||
|
||||
@ -1,6 +1,6 @@
|
||||
#******************************************************************************
|
||||
#
|
||||
# Copyright (C) 1999-2005, International Business Machines
|
||||
# Copyright (C) 1999-2006, International Business Machines
|
||||
# Corporation and others. All Rights Reserved.
|
||||
#
|
||||
#******************************************************************************
|
||||
@ -129,6 +129,7 @@ KhmerLayoutEngine.o \
|
||||
KhmerReordering.o \
|
||||
TibetanLayoutEngine.o \
|
||||
TibetanReordering.o \
|
||||
HangulLayoutEngine.o \
|
||||
KernTable.o
|
||||
|
||||
## Header files to install
|
||||
|
||||
@ -222,6 +222,9 @@
|
||||
<File
|
||||
RelativePath=".\GXLayoutEngine.cpp">
|
||||
</File>
|
||||
<File
|
||||
RelativePath=".\HangulLayoutEngine.cpp">
|
||||
</File>
|
||||
<File
|
||||
RelativePath=".\HanLayoutEngine.cpp">
|
||||
</File>
|
||||
@ -454,6 +457,9 @@
|
||||
<File
|
||||
RelativePath=".\GXLayoutEngine.h">
|
||||
</File>
|
||||
<File
|
||||
RelativePath=".\HangulLayoutEngine.h">
|
||||
</File>
|
||||
<File
|
||||
RelativePath=".\HanLayoutEngine.h">
|
||||
</File>
|
||||
|
||||
Loading…
Reference in New Issue
Block a user