icuSources/layout/IndicReordering.h

   1 /*
   2  *
   3  * (C) Copyright IBM Corp. 1998-2008 - All Rights Reserved
   4  *
   5  */
   6
   7 #ifndef __INDICREORDERING_H
   8 #define __INDICREORDERING_H
   9
  10 /**
  11  * \file
  12  * \internal
  13  */
  14
  15 #include "LETypes.h"
  16 #include "OpenTypeTables.h"
  17
  18 U_NAMESPACE_BEGIN
  19
  20 // Characters that get refered to by name...
  21 #define C_SIGN_ZWNJ           0x200C
  22 #define C_SIGN_ZWJ            0x200D
  23
  24 // Character class values
  25 #define CC_RESERVED               0U
  26 #define CC_VOWEL_MODIFIER         1U
  27 #define CC_STRESS_MARK            2U
  28 #define CC_INDEPENDENT_VOWEL      3U
  29 #define CC_INDEPENDENT_VOWEL_2    4U
  30 #define CC_INDEPENDENT_VOWEL_3    5U
  31 #define CC_CONSONANT              6U
  32 #define CC_CONSONANT_WITH_NUKTA   7U
  33 #define CC_NUKTA                  8U
  34 #define CC_DEPENDENT_VOWEL        9U
  35 #define CC_SPLIT_VOWEL_PIECE_1   10U
  36 #define CC_SPLIT_VOWEL_PIECE_2   11U
  37 #define CC_SPLIT_VOWEL_PIECE_3   12U
  38 #define CC_VIRAMA                13U
  39 #define CC_ZERO_WIDTH_MARK       14U
  40 #define CC_AL_LAKUNA             15U
  41 #define CC_COUNT                 16U
  42
  43 // Character class flags
  44 #define CF_CLASS_MASK    0x0000FFFFU
  45
  46 #define CF_CONSONANT     0x80000000U
  47
  48 #define CF_REPH          0x40000000U
  49 #define CF_VATTU         0x20000000U
  50 #define CF_BELOW_BASE    0x10000000U
  51 #define CF_POST_BASE     0x08000000U
  52 #define CF_LENGTH_MARK   0x04000000U
  53
  54 #define CF_POS_BEFORE    0x00300000U
  55 #define CF_POS_BELOW     0x00200000U
  56 #define CF_POS_ABOVE     0x00100000U
  57 #define CF_POS_AFTER     0x00000000U
  58 #define CF_POS_MASK      0x00300000U
  59
  60 #define CF_INDEX_MASK    0x000F0000U
  61 #define CF_INDEX_SHIFT   16
  62
  63 // Script flag bits
  64 #define SF_MATRAS_AFTER_BASE     0x80000000U
  65 #define SF_REPH_AFTER_BELOW      0x40000000U
  66 #define SF_EYELASH_RA            0x20000000U
  67 #define SF_MPRE_FIXUP            0x10000000U
  68 #define SF_FILTER_ZERO_WIDTH     0x08000000U
  69
  70 #define SF_POST_BASE_LIMIT_MASK  0x0000FFFFU
  71 #define SF_NO_POST_BASE_LIMIT    0x00007FFFU
  72
  73 typedef LEUnicode SplitMatra[3];
  74
  75 class MPreFixups;
  76 class LEGlyphStorage;
  77
  78 struct IndicClassTable
  79 {
  80     typedef le_uint32 CharClass;
  81     typedef le_uint32 ScriptFlags;
  82
  83     LEUnicode firstChar;
  84     LEUnicode lastChar;
  85     le_int32 worstCaseExpansion;
  86     ScriptFlags scriptFlags;
  87     const CharClass *classTable;
  88     const SplitMatra *splitMatraTable;
  89
  90     inline le_int32 getWorstCaseExpansion() const;
  91     inline le_bool getFilterZeroWidth() const;
  92
  93     CharClass getCharClass(LEUnicode ch) const;
  94
  95     inline const SplitMatra *getSplitMatra(CharClass charClass) const;
  96
  97     inline le_bool isVowelModifier(LEUnicode ch) const;
  98     inline le_bool isStressMark(LEUnicode ch) const;
  99     inline le_bool isConsonant(LEUnicode ch) const;
 100     inline le_bool isReph(LEUnicode ch) const;
 101     inline le_bool isVirama(LEUnicode ch) const;
 102     inline le_bool isAlLakuna(LEUnicode ch) const;
 103     inline le_bool isNukta(LEUnicode ch) const;
 104     inline le_bool isVattu(LEUnicode ch) const;
 105     inline le_bool isMatra(LEUnicode ch) const;
 106     inline le_bool isSplitMatra(LEUnicode ch) const;
 107     inline le_bool isLengthMark(LEUnicode ch) const;
 108     inline le_bool hasPostOrBelowBaseForm(LEUnicode ch) const;
 109     inline le_bool hasPostBaseForm(LEUnicode ch) const;
 110     inline le_bool hasBelowBaseForm(LEUnicode ch) const;
 111
 112     inline static le_bool isVowelModifier(CharClass charClass);
 113     inline static le_bool isStressMark(CharClass charClass);
 114     inline static le_bool isConsonant(CharClass charClass);
 115     inline static le_bool isReph(CharClass charClass);
 116     inline static le_bool isVirama(CharClass charClass);
 117     inline static le_bool isAlLakuna(CharClass charClass);
 118     inline static le_bool isNukta(CharClass charClass);
 119     inline static le_bool isVattu(CharClass charClass);
 120     inline static le_bool isMatra(CharClass charClass);
 121     inline static le_bool isSplitMatra(CharClass charClass);
 122     inline static le_bool isLengthMark(CharClass charClass);
 123     inline static le_bool hasPostOrBelowBaseForm(CharClass charClass);
 124     inline static le_bool hasPostBaseForm(CharClass charClass);
 125     inline static le_bool hasBelowBaseForm(CharClass charClass);
 126
 127     static const IndicClassTable *getScriptClassTable(le_int32 scriptCode);
 128 };
 129
 130 class IndicReordering /* not : public UObject because all methods are static */ {
 131 public:
 132     static le_int32 getWorstCaseExpansion(le_int32 scriptCode);
 133
 134     static le_bool getFilterZeroWidth(le_int32 scriptCode);
 135
 136     static le_int32 reorder(const LEUnicode *theChars, le_int32 charCount, le_int32 scriptCode,
 137         LEUnicode *outChars, LEGlyphStorage &glyphStorage,
 138         MPreFixups **outMPreFixups);
 139
 140     static void adjustMPres(MPreFixups *mpreFixups, LEGlyphStorage &glyphStorage);
 141
 142     static const FeatureMap *getFeatureMap(le_int32 &count);
 143
 144 private:
 145     // do not instantiate
 146     IndicReordering();
 147
 148     static le_int32 findSyllable(const IndicClassTable *classTable, const LEUnicode *chars, le_int32 prev, le_int32 charCount);
 149
 150 };
 151
 152 inline le_int32 IndicClassTable::getWorstCaseExpansion() const
 153 {
 154     return worstCaseExpansion;
 155 }
 156
 157 inline le_bool IndicClassTable::getFilterZeroWidth() const
 158 {
 159     return (scriptFlags & SF_FILTER_ZERO_WIDTH) != 0;
 160 }
 161
 162 inline const SplitMatra *IndicClassTable::getSplitMatra(CharClass charClass) const
 163 {
 164     le_int32 index = (charClass & CF_INDEX_MASK) >> CF_INDEX_SHIFT;
 165
 166     return &splitMatraTable[index - 1];
 167 }
 168
 169 inline le_bool IndicClassTable::isVowelModifier(CharClass charClass)
 170 {
 171     return (charClass & CF_CLASS_MASK) == CC_VOWEL_MODIFIER;
 172 }
 173
 174 inline le_bool IndicClassTable::isStressMark(CharClass charClass)
 175 {
 176     return (charClass & CF_CLASS_MASK) == CC_STRESS_MARK;
 177 }
 178
 179 inline le_bool IndicClassTable::isConsonant(CharClass charClass)
 180 {
 181     return (charClass & CF_CONSONANT) != 0;
 182 }
 183
 184 inline le_bool IndicClassTable::isReph(CharClass charClass)
 185 {
 186     return (charClass & CF_REPH) != 0;
 187 }
 188
 189 inline le_bool IndicClassTable::isNukta(CharClass charClass)
 190 {
 191     return (charClass & CF_CLASS_MASK) == CC_NUKTA;
 192 }
 193
 194 inline le_bool IndicClassTable::isVirama(CharClass charClass)
 195 {
 196     return (charClass & CF_CLASS_MASK) == CC_VIRAMA;
 197 }
 198
 199 inline le_bool IndicClassTable::isAlLakuna(CharClass charClass)
 200 {
 201     return (charClass & CF_CLASS_MASK) == CC_AL_LAKUNA;
 202 }
 203
 204 inline le_bool IndicClassTable::isVattu(CharClass charClass)
 205 {
 206     return (charClass & CF_VATTU) != 0;
 207 }
 208
 209 inline le_bool IndicClassTable::isMatra(CharClass charClass)
 210 {
 211     charClass &= CF_CLASS_MASK;
 212
 213     return charClass >= CC_DEPENDENT_VOWEL && charClass <= CC_SPLIT_VOWEL_PIECE_3;
 214 }
 215
 216 inline le_bool IndicClassTable::isSplitMatra(CharClass charClass)
 217 {
 218     return (charClass & CF_INDEX_MASK) != 0;
 219 }
 220
 221 inline le_bool IndicClassTable::isLengthMark(CharClass charClass)
 222 {
 223     return (charClass & CF_LENGTH_MARK) != 0;
 224 }
 225
 226 inline le_bool IndicClassTable::hasPostOrBelowBaseForm(CharClass charClass)
 227 {
 228     return (charClass & (CF_POST_BASE | CF_BELOW_BASE)) != 0;
 229 }
 230
 231 inline le_bool IndicClassTable::hasPostBaseForm(CharClass charClass)
 232 {
 233     return (charClass & CF_POST_BASE) != 0;
 234 }
 235
 236 inline le_bool IndicClassTable::hasBelowBaseForm(CharClass charClass)
 237 {
 238     return (charClass & CF_BELOW_BASE) != 0;
 239 }
 240
 241 inline le_bool IndicClassTable::isVowelModifier(LEUnicode ch) const
 242 {
 243     return isVowelModifier(getCharClass(ch));
 244 }
 245
 246 inline le_bool IndicClassTable::isStressMark(LEUnicode ch) const
 247 {
 248     return isStressMark(getCharClass(ch));
 249 }
 250
 251 inline le_bool IndicClassTable::isConsonant(LEUnicode ch) const
 252 {
 253     return isConsonant(getCharClass(ch));
 254 }
 255
 256 inline le_bool IndicClassTable::isReph(LEUnicode ch) const
 257 {
 258     return isReph(getCharClass(ch));
 259 }
 260
 261 inline le_bool IndicClassTable::isVirama(LEUnicode ch) const
 262 {
 263     return isVirama(getCharClass(ch));
 264 }
 265
 266 inline le_bool IndicClassTable::isAlLakuna(LEUnicode ch) const
 267 {
 268     return isAlLakuna(getCharClass(ch));
 269 }
 270
 271 inline le_bool IndicClassTable::isNukta(LEUnicode ch) const
 272 {
 273     return isNukta(getCharClass(ch));
 274 }
 275
 276 inline le_bool IndicClassTable::isVattu(LEUnicode ch) const
 277 {
 278     return isVattu(getCharClass(ch));
 279 }
 280
 281 inline le_bool IndicClassTable::isMatra(LEUnicode ch) const
 282 {
 283     return isMatra(getCharClass(ch));
 284 }
 285
 286 inline le_bool IndicClassTable::isSplitMatra(LEUnicode ch) const
 287 {
 288     return isSplitMatra(getCharClass(ch));
 289 }
 290
 291 inline le_bool IndicClassTable::isLengthMark(LEUnicode ch) const
 292 {
 293     return isLengthMark(getCharClass(ch));
 294 }
 295
 296 inline le_bool IndicClassTable::hasPostOrBelowBaseForm(LEUnicode ch) const
 297 {
 298     return hasPostOrBelowBaseForm(getCharClass(ch));
 299 }
 300
 301 inline le_bool IndicClassTable::hasPostBaseForm(LEUnicode ch) const
 302 {
 303     return hasPostBaseForm(getCharClass(ch));
 304 }
 305
 306 inline le_bool IndicClassTable::hasBelowBaseForm(LEUnicode ch) const
 307 {
 308     return hasBelowBaseForm(getCharClass(ch));
 309 }
 310
 311 U_NAMESPACE_END
 312 #endif