icuSources/layout/HangulLayoutEngine.cpp

   1 /*
   2  * HangulLayoutEngine.cpp: OpenType processing for Han fonts.
   3  *
   4  * (C) Copyright IBM Corp. 1998-2008 - All Rights Reserved.
   5  */
   6
   7 #include "LETypes.h"
   8 #include "LEScripts.h"
   9 #include "LELanguages.h"
  10
  11 #include "LayoutEngine.h"
  12 #include "OpenTypeLayoutEngine.h"
  13 #include "HangulLayoutEngine.h"
  14 #include "ScriptAndLanguageTags.h"
  15 #include "LEGlyphStorage.h"
  16 #include "OpenTypeTables.h"
  17
  18 U_NAMESPACE_BEGIN
  19
  20 UOBJECT_DEFINE_RTTI_IMPLEMENTATION(HangulOpenTypeLayoutEngine)
  21
  22
  23 #define FEATURE_MAP(name) {name ## FeatureTag, name ## FeatureMask}
  24
  25 #define LJMO_FIRST 0x1100
  26 #define LJMO_LAST  0x1159
  27 #define LJMO_FILL  0x115F
  28 #define LJMO_COUNT 19
  29
  30 #define VJMO_FIRST 0x1161
  31 #define VJMO_LAST  0x11A2
  32 #define VJMO_FILL  0x1160
  33 #define VJMO_COUNT 21
  34
  35 #define TJMO_FIRST 0x11A7
  36 #define TJMO_LAST  0x11F9
  37 #define TJMO_COUNT 28
  38
  39 #define HSYL_FIRST 0xAC00
  40 #define HSYL_COUNT 11172
  41 #define HSYL_LVCNT (VJMO_COUNT * TJMO_COUNT)
  42
  43 // Character classes
  44 enum
  45 {
  46     CC_L = 0,
  47     CC_V,
  48     CC_T,
  49     CC_LV,
  50     CC_LVT,
  51     CC_X,
  52     CC_COUNT
  53 };
  54
  55 // Action flags
  56 #define AF_L 1
  57 #define AF_V 2
  58 #define AF_T 4
  59
  60 // Actions
  61 #define a_N   0
  62 #define a_L   (AF_L)
  63 #define a_V   (AF_V)
  64 #define a_T   (AF_T)
  65 #define a_VT  (AF_V | AF_T)
  66 #define a_LV  (AF_L | AF_V)
  67 #define a_LVT (AF_L | AF_V | AF_T)
  68
  69 typedef struct
  70 {
  71     int32_t newState;
  72     int32_t actionFlags;
  73 } StateTransition;
  74
  75 static const StateTransition stateTable[][CC_COUNT] =
  76 {
  77 //       L          V          T          LV         LVT           X
  78     { {1, a_L},  {2, a_LV}, {3, a_LVT}, {2, a_LV}, {3, a_LVT},  {4, a_T}}, // 0 - start
  79     { {1, a_L},  {2, a_V},  {3, a_VT},  {2, a_LV}, {3, a_LVT}, {-1, a_V}}, // 1 - L+
  80     {{-1, a_N},  {2, a_V},  {3, a_T},  {-1, a_N}, {-1, a_N},   {-1, a_N}}, // 2 - L+V+
  81     {{-1, a_N}, {-1, a_N},  {3, a_T},  {-1, a_N}, {-1, a_N},   {-1, a_N}}, // 3 - L+V+T*
  82     {{-1, a_N}, {-1, a_N}, {-1, a_N},  {-1, a_N}, {-1, a_N},    {4, a_T}}  // 4 - X+
  83 };
  84
  85
  86 #define ccmpFeatureTag LE_CCMP_FEATURE_TAG
  87 #define ljmoFeatureTag LE_LJMO_FEATURE_TAG
  88 #define vjmoFeatureTag LE_VJMO_FEATURE_TAG
  89 #define tjmoFeatureTag LE_TJMO_FEATURE_TAG
  90
  91 #define ccmpFeatureMask 0x80000000UL
  92 #define ljmoFeatureMask 0x40000000UL
  93 #define vjmoFeatureMask 0x20000000UL
  94 #define tjmoFeatureMask 0x10000000UL
  95
  96 static const FeatureMap featureMap[] =
  97 {
  98     {ccmpFeatureTag, ccmpFeatureMask},
  99     {ljmoFeatureTag, ljmoFeatureMask},
 100     {vjmoFeatureTag, vjmoFeatureMask},
 101     {tjmoFeatureTag, tjmoFeatureMask}
 102 };
 103
 104 static const le_int32 featureMapCount = LE_ARRAY_SIZE(featureMap);
 105
 106 #define nullFeatures 0
 107 #define ljmoFeatures (ccmpFeatureMask | ljmoFeatureMask)
 108 #define vjmoFeatures (ccmpFeatureMask | vjmoFeatureMask | ljmoFeatureMask | tjmoFeatureMask)
 109 #define tjmoFeatures (ccmpFeatureMask | tjmoFeatureMask | ljmoFeatureMask | vjmoFeatureMask)
 110
 111 static le_int32 compose(LEUnicode lead, LEUnicode vowel, LEUnicode trail, LEUnicode &syllable)
 112 {
 113     le_int32 lIndex = lead  - LJMO_FIRST;
 114     le_int32 vIndex = vowel - VJMO_FIRST;
 115     le_int32 tIndex = trail - TJMO_FIRST;
 116     le_int32 result = 3;
 117
 118     if ((lIndex < 0 || lIndex >= LJMO_COUNT ) || (vIndex < 0 || vIndex >= VJMO_COUNT)) {
 119         return 0;
 120     }
 121
 122     if (tIndex <= 0 || tIndex >= TJMO_COUNT) {
 123         tIndex = 0;
 124         result = 2;
 125     }
 126
 127     syllable = (LEUnicode) ((lIndex * VJMO_COUNT + vIndex) * TJMO_COUNT + tIndex + HSYL_FIRST);
 128
 129     return result;
 130 }
 131
 132 static le_int32 decompose(LEUnicode syllable, LEUnicode &lead, LEUnicode &vowel, LEUnicode &trail)
 133 {
 134     le_int32 sIndex = syllable - HSYL_FIRST;
 135
 136     if (sIndex < 0 || sIndex >= HSYL_COUNT) {
 137         return 0;
 138     }
 139
 140     lead  = LJMO_FIRST + (sIndex / HSYL_LVCNT);
 141     vowel = VJMO_FIRST + (sIndex % HSYL_LVCNT) / TJMO_COUNT;
 142     trail = TJMO_FIRST + (sIndex % TJMO_COUNT);
 143
 144     if (trail == TJMO_FIRST) {
 145         return 2;
 146     }
 147
 148     return 3;
 149 }
 150
 151 static le_int32 getCharClass(LEUnicode ch, LEUnicode &lead, LEUnicode &vowel, LEUnicode &trail)
 152 {
 153     lead  = LJMO_FILL;
 154     vowel = VJMO_FILL;
 155     trail = TJMO_FIRST;
 156
 157     if (ch >= LJMO_FIRST && ch <= LJMO_LAST) {
 158         lead  = ch;
 159         return CC_L;
 160     }
 161
 162     if (ch >= VJMO_FIRST && ch <= VJMO_LAST) {
 163         vowel = ch;
 164         return CC_V;
 165     }
 166
 167     if (ch > TJMO_FIRST && ch <= TJMO_LAST) {
 168         trail = ch;
 169         return CC_T;
 170     }
 171
 172     le_int32 c = decompose(ch, lead, vowel, trail);
 173
 174     if (c == 2) {
 175         return CC_LV;
 176     }
 177
 178     if (c == 3) {
 179         return CC_LVT;
 180     }
 181
 182     trail = ch;
 183     return CC_X;
 184 }
 185
 186 HangulOpenTypeLayoutEngine::HangulOpenTypeLayoutEngine(const LEFontInstance *fontInstance, le_int32 scriptCode, le_int32 /*languageCode*/,
 187                                        le_int32 typoFlags, const GlyphSubstitutionTableHeader *gsubTable, LEErrorCode &success)
 188     : OpenTypeLayoutEngine(fontInstance, scriptCode, korLanguageCode, typoFlags, gsubTable, success)
 189 {
 190     fFeatureMap = featureMap;
 191     fFeatureMapCount = featureMapCount;
 192     fFeatureOrder = TRUE;
 193 }
 194
 195 HangulOpenTypeLayoutEngine::HangulOpenTypeLayoutEngine(const LEFontInstance *fontInstance, le_int32 scriptCode, le_int32 /*languageCode*/,
 196                                                            le_int32 typoFlags, LEErrorCode &success)
 197     : OpenTypeLayoutEngine(fontInstance, scriptCode, korLanguageCode, typoFlags, success)
 198 {
 199     fFeatureMap = featureMap;
 200     fFeatureMapCount = featureMapCount;
 201     fFeatureOrder = TRUE;
 202 }
 203
 204 HangulOpenTypeLayoutEngine::~HangulOpenTypeLayoutEngine()
 205 {
 206     // nothing to do
 207 }
 208
 209 le_int32 HangulOpenTypeLayoutEngine::characterProcessing(const LEUnicode chars[], le_int32 offset, le_int32 count, le_int32 max, le_bool rightToLeft,
 210         LEUnicode *&outChars, LEGlyphStorage &glyphStorage, LEErrorCode &success)
 211 {
 212     if (LE_FAILURE(success)) {
 213         return 0;
 214     }
 215
 216     if (chars == NULL || offset < 0 || count < 0 || max < 0 || offset >= max || offset + count > max) {
 217         success = LE_ILLEGAL_ARGUMENT_ERROR;
 218         return 0;
 219     }
 220
 221     le_int32 worstCase = count * 3;
 222
 223     outChars = LE_NEW_ARRAY(LEUnicode, worstCase);
 224
 225     if (outChars == NULL) {
 226         success = LE_MEMORY_ALLOCATION_ERROR;
 227         return 0;
 228     }
 229
 230     glyphStorage.allocateGlyphArray(worstCase, rightToLeft, success);
 231     glyphStorage.allocateAuxData(success);
 232
 233     if (LE_FAILURE(success)) {
 234         LE_DELETE_ARRAY(outChars);
 235         return 0;
 236     }
 237
 238     le_int32 outCharCount = 0;
 239     le_int32 limit = offset + count;
 240     le_int32 i = offset;
 241
 242     while (i < limit) {
 243         le_int32 state    = 0;
 244         le_int32 inStart  = i;
 245         le_int32 outStart = outCharCount;
 246
 247         while( i < limit) {
 248             LEUnicode lead  = 0;
 249             LEUnicode vowel = 0;
 250             LEUnicode trail = 0;
 251             int32_t chClass = getCharClass(chars[i], lead, vowel, trail);
 252             const StateTransition transition = stateTable[state][chClass];
 253
 254             if (chClass == CC_X) {
 255                 /* Any character of type X will be stored as a trail jamo */
 256                 if ((transition.actionFlags & AF_T) != 0) {
 257                     outChars[outCharCount] = trail;
 258                     glyphStorage.setCharIndex(outCharCount, i-offset, success);
 259                     glyphStorage.setAuxData(outCharCount++, nullFeatures, success);
 260                 }
 261             } else {
 262                 /* Any Hangul will be fully decomposed. Output the decomposed characters. */
 263                 if ((transition.actionFlags & AF_L) != 0) {
 264                     outChars[outCharCount] = lead;
 265                     glyphStorage.setCharIndex(outCharCount, i-offset, success);
 266                     glyphStorage.setAuxData(outCharCount++, ljmoFeatures, success);
 267                 }
 268
 269                 if ((transition.actionFlags & AF_V) != 0) {
 270                     outChars[outCharCount] = vowel;
 271                     glyphStorage.setCharIndex(outCharCount, i-offset, success);
 272                     glyphStorage.setAuxData(outCharCount++, vjmoFeatures, success);
 273                 }
 274
 275                 if ((transition.actionFlags & AF_T) != 0) {
 276                     outChars[outCharCount] = trail;
 277                     glyphStorage.setCharIndex(outCharCount, i-offset, success);
 278                     glyphStorage.setAuxData(outCharCount++, tjmoFeatures, success);
 279                 }
 280             }
 281
 282             state = transition.newState;
 283
 284             /* Negative next state means stop. */
 285             if (state < 0) {
 286                 break;
 287             }
 288
 289             i += 1;
 290         }
 291
 292         le_int32 inLength  = i - inStart;
 293         le_int32 outLength = outCharCount - outStart;
 294
 295         /*
 296          * See if the syllable can be composed into a single character. There are 5
 297          * possible cases:
 298          *
 299          *   Input     Decomposed to    Compose to
 300          *   LV        L, V             LV
 301          *   LVT       L, V, T          LVT
 302          *   L, V      L, V             LV, DEL
 303          *   LV, T     L, V, T          LVT, DEL
 304          *   L, V, T   L, V, T          LVT, DEL, DEL
 305          */
 306         if ((inLength >= 1 && inLength <= 3) && (outLength == 2 || outLength == 3)) {
 307             LEUnicode syllable = 0x0000;
 308             LEUnicode lead  = outChars[outStart];
 309             LEUnicode vowel = outChars[outStart + 1];
 310             LEUnicode trail = outLength == 3? outChars[outStart + 2] : TJMO_FIRST;
 311
 312             /*
 313              * If the composition consumes the whole decomposed syllable,
 314              * we can use it.
 315              */
 316             if (compose(lead, vowel, trail, syllable) == outLength) {
 317                 outCharCount = outStart;
 318                 outChars[outCharCount] = syllable;
 319                 glyphStorage.setCharIndex(outCharCount, inStart-offset, success);
 320                 glyphStorage.setAuxData(outCharCount++, nullFeatures, success);
 321
 322                 /*
 323                  * Replace the rest of the input characters with DEL.
 324                  */
 325                 for(le_int32 d = inStart + 1; d < i; d += 1) {
 326                     outChars[outCharCount] = 0xFFFF;
 327                     glyphStorage.setCharIndex(outCharCount, d - offset, success);
 328                     glyphStorage.setAuxData(outCharCount++, nullFeatures, success);
 329                 }
 330             }
 331         }
 332     }
 333
 334     glyphStorage.adoptGlyphCount(outCharCount);
 335     return outCharCount;
 336 }
 337
 338 U_NAMESPACE_END