1 /* 2 * 3 * (C) Copyright IBM Corp. 1998-2009 - All Rights Reserved 4 * 5 */ 6 7 #include "LETypes.h" 8 #include "LEScripts.h" 9 #include "OpenTypeTables.h" 10 #include "OpenTypeUtilities.h" 11 #include "IndicReordering.h" 12 13 U_NAMESPACE_BEGIN 14 15 // Split matra table indices 16 #define _x1 (1 << CF_INDEX_SHIFT) 17 #define _x2 (2 << CF_INDEX_SHIFT) 18 #define _x3 (3 << CF_INDEX_SHIFT) 19 #define _x4 (4 << CF_INDEX_SHIFT) 20 #define _x5 (5 << CF_INDEX_SHIFT) 21 #define _x6 (6 << CF_INDEX_SHIFT) 22 #define _x7 (7 << CF_INDEX_SHIFT) 23 #define _x8 (8 << CF_INDEX_SHIFT) 24 #define _x9 (9 << CF_INDEX_SHIFT) 25 26 // simple classes 27 #define _xx (CC_RESERVED) 28 #define _ma (CC_VOWEL_MODIFIER | CF_POS_ABOVE) 29 #define _mp (CC_VOWEL_MODIFIER | CF_POS_AFTER) 30 #define _sa (CC_STRESS_MARK | CF_POS_ABOVE) 31 #define _sb (CC_STRESS_MARK | CF_POS_BELOW) 32 #define _iv (CC_INDEPENDENT_VOWEL) 33 #define _i2 (CC_INDEPENDENT_VOWEL_2) 34 #define _i3 (CC_INDEPENDENT_VOWEL_3) 35 #define _ct (CC_CONSONANT | CF_CONSONANT) 36 #define _cn (CC_CONSONANT_WITH_NUKTA | CF_CONSONANT) 37 #define _nu (CC_NUKTA) 38 #define _dv (CC_DEPENDENT_VOWEL) 39 #define _dl (_dv | CF_POS_BEFORE) 40 #define _db (_dv | CF_POS_BELOW) 41 #define _da (_dv | CF_POS_ABOVE) 42 #define _dr (_dv | CF_POS_AFTER) 43 #define _lm (_dv | CF_LENGTH_MARK) 44 #define _l1 (CC_SPLIT_VOWEL_PIECE_1 | CF_POS_BEFORE) 45 #define _a1 (CC_SPLIT_VOWEL_PIECE_1 | CF_POS_ABOVE) 46 #define _b2 (CC_SPLIT_VOWEL_PIECE_2 | CF_POS_BELOW) 47 #define _r2 (CC_SPLIT_VOWEL_PIECE_2 | CF_POS_AFTER) 48 #define _m2 (CC_SPLIT_VOWEL_PIECE_2 | CF_LENGTH_MARK) 49 #define _m3 (CC_SPLIT_VOWEL_PIECE_3 | CF_LENGTH_MARK) 50 #define _vr (CC_VIRAMA) 51 #define _al (CC_AL_LAKUNA) 52 53 // split matras 54 #define _s1 (_dv | _x1) 55 #define _s2 (_dv | _x2) 56 #define _s3 (_dv | _x3) 57 #define _s4 (_dv | _x4) 58 #define _s5 (_dv | _x5) 59 #define _s6 (_dv | _x6) 60 #define _s7 (_dv | _x7) 61 #define _s8 (_dv | _x8) 62 #define _s9 (_dv | _x9) 63 64 // consonants with special forms 65 // NOTE: this assumes that no consonants with nukta have 66 // special forms... (Bengali RA?) 67 #define _bb (_ct | CF_BELOW_BASE) 68 #define _pb (_ct | CF_POST_BASE) 69 #define _vt (_bb | CF_VATTU) 70 #define _rv (_vt | CF_REPH) 71 #define _rp (_pb | CF_REPH) 72 #define _rb (_bb | CF_REPH) 73 74 // 75 // Character class tables 76 // 77 static const IndicClassTable::CharClass devaCharClasses[] = 78 { 79 _xx, _ma, _ma, _mp, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, // 0900 - 090F 80 _iv, _iv, _iv, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0910 - 091F 81 _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _cn, _ct, _ct, _ct, _ct, _ct, _ct, // 0920 - 092F 82 _rv, _cn, _ct, _ct, _cn, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _dr, _dl, // 0930 - 093F 83 _dr, _db, _db, _db, _db, _da, _da, _da, _da, _dr, _dr, _dr, _dr, _vr, _xx, _xx, // 0940 - 094F 84 _xx, _sa, _sb, _sa, _sa, _xx, _xx, _xx, _cn, _cn, _cn, _cn, _cn, _cn, _cn, _cn, // 0950 - 095F 85 _iv, _iv, _db, _db, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0960 - 096F 86 _xx // 0970 87 }; 88 89 static const IndicClassTable::CharClass bengCharClasses[] = 90 { 91 _xx, _ma, _mp, _mp, _xx, _i2, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _i2, // 0980 - 098F 92 _iv, _xx, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0990 - 099F 93 _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _bb, _ct, _ct, _pb, // 09A0 - 09AF 94 _rv, _xx, _ct, _xx, _xx, _xx, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _r2, _dl, // 09B0 - 09BF 95 _dr, _db, _db, _db, _db, _xx, _xx, _l1, _dl, _xx, _xx, _s1, _s2, _vr, _xx, _xx, // 09C0 - 09CF 96 _xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _cn, _cn, _xx, _cn, // 09D0 - 09DF 97 _iv, _iv, _dv, _dv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 09E0 - 09EF 98 _rv, _ct, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx // 09F0 - 09FA 99 }; 100 101 static const IndicClassTable::CharClass punjCharClasses[] = 102 { 103 _xx, _ma, _ma, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _xx, _iv, // 0A00 - 0A0F 104 _iv, _xx, _xx, _i3, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0A10 - 0A1F 105 _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _bb, // 0A20 - 0A2F 106 _vt, _xx, _ct, _cn, _xx, _bb, _cn, _xx, _ct, _bb, _xx, _xx, _nu, _xx, _dr, _dl, // 0A30 - 0A3F 107 _dr, _b2, _db, _xx, _xx, _xx, _xx, _da, _da, _xx, _xx, _a1, _da, _vr, _xx, _xx, // 0A40 - 0A4F 108 _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _cn, _cn, _cn, _ct, _xx, _cn, _xx, // 0A50 - 0A5F 109 _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0A60 - 0A6F 110 _ma, _ma, _xx, _xx, _xx // 0A70 - 0A74 111 }; 112 113 static const IndicClassTable::CharClass gujrCharClasses[] = 114 { 115 _xx, _ma, _ma, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _xx, _iv, // 0A80 - 0A8F 116 _iv, _iv, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0A90 - 0A9F 117 _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _ct, // 0AA0 - 0AAF 118 _rv, _xx, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _dr, _dl, // 0AB0 - 0ABF 119 _dr, _db, _db, _db, _db, _da, _xx, _da, _da, _dr, _xx, _dr, _dr, _vr, _xx, _xx, // 0AC0 - 0ACF 120 _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0AD0 - 0ADF 121 _iv, _iv, _db, _db, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx // 0AE0 - 0AEF 122 }; 123 124 #if 1 125 static const IndicClassTable::CharClass oryaCharClasses[] = 126 { 127 _xx, _ma, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _iv, /* 0B00 - 0B0F */ 128 _iv, _xx, _xx, _iv, _iv, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _ct, _bb, /* 0B10 - 0B1F */ 129 _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _pb, /* 0B20 - 0B2F */ 130 _rb, _xx, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _xx, _xx, _nu, _xx, _dr, _da, /* 0B30 - 0B3F */ 131 _dr, _db, _db, _db, _xx, _xx, _xx, _dl, _s1, _xx, _xx, _s2, _s3, _vr, _xx, _xx, /* 0B40 - 0B4F */ 132 _xx, _xx, _xx, _xx, _xx, _xx, _da, _dr, _xx, _xx, _xx, _xx, _cn, _cn, _xx, _pb, /* 0B50 - 0B5F */ 133 _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, /* 0B60 - 0B6F */ 134 _xx, _bb /* 0B70 - 0B71 */ 135 }; 136 #else 137 static const IndicClassTable::CharClass oryaCharClasses[] = 138 { 139 _xx, _ma, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _iv, // 0B00 - 0B0F 140 _iv, _xx, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0B10 - 0B1F 141 _ct, _ct, _ct, _ct, _bb, _ct, _ct, _ct, _bb, _xx, _ct, _ct, _bb, _bb, _bb, _pb, // 0B20 - 0B2F 142 _rb, _xx, _bb, _bb, _xx, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _r2, _da, // 0B30 - 0B3F 143 _dr, _db, _db, _db, _xx, _xx, _xx, _l1, _s1, _xx, _xx, _s2, _s3, _vr, _xx, _xx, // 0B40 - 0B4F 144 _xx, _xx, _xx, _xx, _xx, _xx, _m2, _m2, _xx, _xx, _xx, _xx, _cn, _cn, _xx, _cn, // 0B50 - 0B5F 145 _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0B60 - 0B6F 146 _xx, _ct // 0B70 - 0B71 147 }; 148 #endif 149 150 static const IndicClassTable::CharClass tamlCharClasses[] = 151 { 152 _xx, _xx, _ma, _xx, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _iv, _iv, // 0B80 - 0B8F 153 _iv, _xx, _iv, _iv, _iv, _ct, _xx, _xx, _xx, _ct, _ct, _xx, _ct, _xx, _ct, _ct, // 0B90 - 0B9F 154 _xx, _xx, _xx, _ct, _ct, _xx, _xx, _xx, _ct, _ct, _ct, _xx, _xx, _xx, _ct, _ct, // 0BA0 - 0BAF 155 _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _xx, _xx, _r2, _dr, // 0BB0 - 0BBF 156 _da, _dr, _dr, _xx, _xx, _xx, _l1, _l1, _dl, _xx, _s1, _s2, _s3, _vr, _xx, _xx, // 0BC0 - 0BCF 157 _xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0BD0 - 0BDF 158 _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0BE0 - 0BEF 159 _xx, _xx, _xx // 0BF0 - 0BF2 160 }; 161 162 // FIXME: Should some of the bb's be pb's? (KA, NA, MA, YA, VA, etc. (approx 13)) 163 static const IndicClassTable::CharClass teluCharClasses[] = 164 { 165 _xx, _mp, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _iv, // 0C00 - 0C0F 166 _iv, _xx, _iv, _iv, _iv, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, // 0C10 - 0C1F 167 _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _bb, // 0C20 - 0C2F 168 _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _xx, _xx, _xx, _xx, _da, _da, // 0C30 - 0C3F 169 _da, _dr, _dr, _dr, _dr, _xx, _a1, _da, _s1, _xx, _da, _da, _da, _vr, _xx, _xx, // 0C40 - 0C4F 170 _xx, _xx, _xx, _xx, _xx, _da, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0C50 - 0C5F 171 _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx // 0C60 - 0C6F 172 }; 173 174 // U+CC3 and U+CC4 are _lm here not _dr since the Kannada rendering 175 // rules want them below and to the right of the entire cluster 176 // 177 // There's some information about this in: 178 // 179 // http://brahmi.sourceforge.net/docs/KannadaComputing.html 180 static const IndicClassTable::CharClass kndaCharClasses[] = 181 { 182 _xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, // 0C80 - 0C8F 183 _iv, _xx, _iv, _iv, _iv, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, // 0C90 - 0C9F 184 _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _bb, // 0CA0 - 0CAF 185 _rb, _ct, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _xx, _xx, _xx, _xx, _dr, _da, // 0CB0 - 0CBF 186 _s1, _dr, _r2, _lm, _lm, _xx, _a1, _s2, _s3, _xx, _s4, _s5, _da, _vr, _xx, _xx, // 0CC0 - 0CCF 187 _xx, _xx, _xx, _xx, _xx, _m3, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _ct, _xx, // 0CD0 - 0CDF 188 _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx // 0CE0 - 0CEF 189 }; 190 191 // FIXME: this is correct for old-style Malayalam (MAL) but not for reformed Malayalam (MLR) 192 // FIXME: should there be a REPH for old-style Malayalam? 193 static const IndicClassTable::CharClass mlymCharClasses[] = 194 { 195 _xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _iv, // 0D00 - 0D0F 196 _iv, _xx, _iv, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0D10 - 0D1F 197 _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _pb, // 0D20 - 0D2F 198 _pb, _cn, _bb, _ct, _ct, _pb, _ct, _ct, _ct, _ct, _xx, _xx, _xx, _xx, _r2, _dr, // 0D30 - 0D3F 199 _dr, _dr, _dr, _dr, _xx, _xx, _l1, _l1, _dl, _xx, _s1, _s2, _s3, _vr, _xx, _xx, // 0D40 - 0D4F 200 _xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0D50 - 0D5F 201 _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx // 0D60 - 0D6F 202 }; 203 204 static const IndicClassTable::CharClass sinhCharClasses[] = 205 { 206 _xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, // 0D80 - 0D8F 207 _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _ct, _ct, _ct, _ct, _ct, _ct, // 0D90 - 0D9F 208 _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0DA0 - 0DAF 209 _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _xx, _xx, // 0DB0 - 0DBF 210 _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _xx, _al, _xx, _xx, _xx, _xx, _dr, // 0DC0 - 0DCF 211 _dr, _dr, _da, _da, _db, _xx, _db, _xx, _dr, _dl, _s1, _dl, _s2, _s3, _s4, _dr, // 0DD0 - 0DDF 212 _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0DE0 - 0DEF 213 _xx, _xx, _dr, _dr, _xx // 0DF0 - 0DF4 214 }; 215 216 // 217 // Split matra tables 218 // 219 static const SplitMatra bengSplitTable[] = {{0x09C7, 0x09BE}, {0x09C7, 0x09D7}}; 220 221 static const SplitMatra oryaSplitTable[] = {{0x0B47, 0x0B56}, {0x0B47, 0x0B3E}, {0x0B47, 0x0B57}}; 222 223 static const SplitMatra tamlSplitTable[] = {{0x0BC6, 0x0BBE}, {0x0BC7, 0x0BBE}, {0x0BC6, 0x0BD7}}; 224 225 static const SplitMatra teluSplitTable[] = {{0x0C46, 0x0C56}}; 226 227 static const SplitMatra kndaSplitTable[] = {{0x0CBF, 0x0CD5}, {0x0CC6, 0x0CD5}, {0x0CC6, 0x0CD6}, {0x0CC6, 0x0CC2}, 228 {0x0CC6, 0x0CC2, 0x0CD5}}; 229 230 static const SplitMatra mlymSplitTable[] = {{0x0D46, 0x0D3E}, {0x0D47, 0x0D3E}, {0x0D46, 0x0D57}}; 231 232 233 static const SplitMatra sinhSplitTable[] = {{0x0DD9, 0x0DCA}, {0x0DD9, 0x0DCF}, {0x0DD9, 0x0DCF, 0x0DCA}, 234 {0x0DD9, 0x0DDF}}; 235 // 236 // Script Flags 237 // 238 239 // FIXME: post 'GSUB' reordering of MATRA_PRE's for Malayalam and Tamil 240 // FIXME: reformed Malayalam needs to reorder VATTU to before base glyph... 241 // FIXME: not sure passing ZWJ/ZWNJ is best way to render Malayalam Cillu... 242 // FIXME: eyelash RA only for Devanagari?? 243 #define DEVA_SCRIPT_FLAGS (SF_EYELASH_RA | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH) 244 #define BENG_SCRIPT_FLAGS (SF_REPH_AFTER_BELOW | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH) 245 #define PUNJ_SCRIPT_FLAGS (SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH) 246 #define GUJR_SCRIPT_FLAGS (SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH) 247 #define ORYA_SCRIPT_FLAGS (SF_REPH_AFTER_BELOW | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH) 248 #define TAML_SCRIPT_FLAGS (SF_MPRE_FIXUP | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH) 249 #define TELU_SCRIPT_FLAGS (SF_MATRAS_AFTER_BASE | SF_FILTER_ZERO_WIDTH | 3) 250 #define KNDA_SCRIPT_FLAGS (SF_MATRAS_AFTER_BASE | SF_FILTER_ZERO_WIDTH | 3) 251 #define MLYM_SCRIPT_FLAGS (SF_MPRE_FIXUP | SF_NO_POST_BASE_LIMIT /*| SF_FILTER_ZERO_WIDTH*/) 252 #define SINH_SCRIPT_FLAGS (SF_NO_POST_BASE_LIMIT) 253 254 // 255 // Indic Class Tables 256 // 257 static const IndicClassTable devaClassTable = {0x0900, 0x0970, 2, DEVA_SCRIPT_FLAGS, devaCharClasses, NULL}; 258 259 static const IndicClassTable bengClassTable = {0x0980, 0x09FA, 3, BENG_SCRIPT_FLAGS, bengCharClasses, bengSplitTable}; 260 261 static const IndicClassTable punjClassTable = {0x0A00, 0x0A74, 2, PUNJ_SCRIPT_FLAGS, punjCharClasses, NULL}; 262 263 static const IndicClassTable gujrClassTable = {0x0A80, 0x0AEF, 2, GUJR_SCRIPT_FLAGS, gujrCharClasses, NULL}; 264 265 static const IndicClassTable oryaClassTable = {0x0B00, 0x0B71, 3, ORYA_SCRIPT_FLAGS, oryaCharClasses, oryaSplitTable}; 266 267 static const IndicClassTable tamlClassTable = {0x0B80, 0x0BF2, 3, TAML_SCRIPT_FLAGS, tamlCharClasses, tamlSplitTable}; 268 269 static const IndicClassTable teluClassTable = {0x0C00, 0x0C6F, 3, TELU_SCRIPT_FLAGS, teluCharClasses, teluSplitTable}; 270 271 static const IndicClassTable kndaClassTable = {0x0C80, 0x0CEF, 4, KNDA_SCRIPT_FLAGS, kndaCharClasses, kndaSplitTable}; 272 273 static const IndicClassTable mlymClassTable = {0x0D00, 0x0D6F, 3, MLYM_SCRIPT_FLAGS, mlymCharClasses, mlymSplitTable}; 274 275 static const IndicClassTable sinhClassTable = {0x0D80, 0x0DF4, 4, SINH_SCRIPT_FLAGS, sinhCharClasses, sinhSplitTable}; 276 277 // 278 // IndicClassTable addresses 279 // 280 static const IndicClassTable * const indicClassTables[scriptCodeCount] = { 281 NULL, /* 'zyyy' (COMMON) */ 282 NULL, /* 'qaai' (INHERITED) */ 283 NULL, /* 'arab' (ARABIC) */ 284 NULL, /* 'armn' (ARMENIAN) */ 285 &bengClassTable, /* 'beng' (BENGALI) */ 286 NULL, /* 'bopo' (BOPOMOFO) */ 287 NULL, /* 'cher' (CHEROKEE) */ 288 NULL, /* 'copt' (COPTIC) */ 289 NULL, /* 'cyrl' (CYRILLIC) */ 290 NULL, /* 'dsrt' (DESERET) */ 291 &devaClassTable, /* 'deva' (DEVANAGARI) */ 292 NULL, /* 'ethi' (ETHIOPIC) */ 293 NULL, /* 'geor' (GEORGIAN) */ 294 NULL, /* 'goth' (GOTHIC) */ 295 NULL, /* 'grek' (GREEK) */ 296 &gujrClassTable, /* 'gujr' (GUJARATI) */ 297 &punjClassTable, /* 'guru' (GURMUKHI) */ 298 NULL, /* 'hani' (HAN) */ 299 NULL, /* 'hang' (HANGUL) */ 300 NULL, /* 'hebr' (HEBREW) */ 301 NULL, /* 'hira' (HIRAGANA) */ 302 &kndaClassTable, /* 'knda' (KANNADA) */ 303 NULL, /* 'kata' (KATAKANA) */ 304 NULL, /* 'khmr' (KHMER) */ 305 NULL, /* 'laoo' (LAO) */ 306 NULL, /* 'latn' (LATIN) */ 307 &mlymClassTable, /* 'mlym' (MALAYALAM) */ 308 NULL, /* 'mong' (MONGOLIAN) */ 309 NULL, /* 'mymr' (MYANMAR) */ 310 NULL, /* 'ogam' (OGHAM) */ 311 NULL, /* 'ital' (OLD-ITALIC) */ 312 &oryaClassTable, /* 'orya' (ORIYA) */ 313 NULL, /* 'runr' (RUNIC) */ 314 &sinhClassTable, /* 'sinh' (SINHALA) */ 315 NULL, /* 'syrc' (SYRIAC) */ 316 &tamlClassTable, /* 'taml' (TAMIL) */ 317 &teluClassTable, /* 'telu' (TELUGU) */ 318 NULL, /* 'thaa' (THAANA) */ 319 NULL, /* 'thai' (THAI) */ 320 NULL, /* 'tibt' (TIBETAN) */ 321 NULL, /* 'cans' (CANADIAN-ABORIGINAL) */ 322 NULL, /* 'yiii' (YI) */ 323 NULL, /* 'tglg' (TAGALOG) */ 324 NULL, /* 'hano' (HANUNOO) */ 325 NULL, /* 'buhd' (BUHID) */ 326 NULL, /* 'tagb' (TAGBANWA) */ 327 NULL, /* 'brai' (BRAILLE) */ 328 NULL, /* 'cprt' (CYPRIOT) */ 329 NULL, /* 'limb' (LIMBU) */ 330 NULL, /* 'linb' (LINEAR_B) */ 331 NULL, /* 'osma' (OSMANYA) */ 332 NULL, /* 'shaw' (SHAVIAN) */ 333 NULL, /* 'tale' (TAI_LE) */ 334 NULL, /* 'ugar' (UGARITIC) */ 335 NULL, /* 'hrkt' (KATAKANA_OR_HIRAGANA) */ 336 NULL, /* 'bugi' (BUGINESE) */ 337 NULL, /* 'glag' (GLAGOLITIC) */ 338 NULL, /* 'khar' (KHAROSHTHI) */ 339 NULL, /* 'sylo' (SYLOTI_NAGRI) */ 340 NULL, /* 'talu' (NEW_TAI_LUE) */ 341 NULL, /* 'tfng' (TIFINAGH) */ 342 NULL, /* 'xpeo' (OLD_PERSIAN) */ 343 NULL, /* 'bali' (BALINESE) */ 344 NULL, /* 'batk' (BATK) */ 345 NULL, /* 'blis' (BLIS) */ 346 NULL, /* 'brah' (BRAH) */ 347 NULL, /* 'cham' (CHAM) */ 348 NULL, /* 'cirt' (CIRT) */ 349 NULL, /* 'cyrs' (CYRS) */ 350 NULL, /* 'egyd' (EGYD) */ 351 NULL, /* 'egyh' (EGYH) */ 352 NULL, /* 'egyp' (EGYP) */ 353 NULL, /* 'geok' (GEOK) */ 354 NULL, /* 'hans' (HANS) */ 355 NULL, /* 'hant' (HANT) */ 356 NULL, /* 'hmng' (HMNG) */ 357 NULL, /* 'hung' (HUNG) */ 358 NULL, /* 'inds' (INDS) */ 359 NULL, /* 'java' (JAVA) */ 360 NULL, /* 'kali' (KALI) */ 361 NULL, /* 'latf' (LATF) */ 362 NULL, /* 'latg' (LATG) */ 363 NULL, /* 'lepc' (LEPC) */ 364 NULL, /* 'lina' (LINA) */ 365 NULL, /* 'mand' (MAND) */ 366 NULL, /* 'maya' (MAYA) */ 367 NULL, /* 'mero' (MERO) */ 368 NULL, /* 'nko ' (NKO) */ 369 NULL, /* 'orkh' (ORKH) */ 370 NULL, /* 'perm' (PERM) */ 371 NULL, /* 'phag' (PHAGS_PA) */ 372 NULL, /* 'phnx' (PHOENICIAN) */ 373 NULL, /* 'plrd' (PLRD) */ 374 NULL, /* 'roro' (RORO) */ 375 NULL, /* 'sara' (SARA) */ 376 NULL, /* 'syre' (SYRE) */ 377 NULL, /* 'syrj' (SYRJ) */ 378 NULL, /* 'syrn' (SYRN) */ 379 NULL, /* 'teng' (TENG) */ 380 NULL, /* 'vai ' (VAII) */ 381 NULL, /* 'visp' (VISP) */ 382 NULL, /* 'xsux' (CUNEIFORM) */ 383 NULL, /* 'zxxx' (ZXXX) */ 384 NULL, /* 'zzzz' (UNKNOWN) */ 385 NULL, /* 'cari' (CARI) */ 386 NULL, /* 'jpan' (JPAN) */ 387 NULL, /* 'lana' (LANA) */ 388 NULL, /* 'lyci' (LYCI) */ 389 NULL, /* 'lydi' (LYDI) */ 390 NULL, /* 'olck' (OLCK) */ 391 NULL, /* 'rjng' (RJNG) */ 392 NULL, /* 'saur' (SAUR) */ 393 NULL, /* 'sgnw' (SGNW) */ 394 NULL, /* 'sund' (SUND) */ 395 NULL, /* 'moon' (MOON) */ 396 NULL, /* 'mtei' (MTEI) */ 397 NULL, /* 'armi' (ARMI) */ 398 NULL, /* 'avst' (AVST) */ 399 NULL, /* 'cakm' (CAKM) */ 400 NULL, /* 'kore' (KORE) */ 401 NULL, /* 'kthi' (KTHI) */ 402 NULL, /* 'mani' (MANI) */ 403 NULL, /* 'phli' (PHLI) */ 404 NULL, /* 'phlp' (PHLP) */ 405 NULL, /* 'phlv' (PHLV) */ 406 NULL, /* 'prti' (PRTI) */ 407 NULL, /* 'samr' (SAMR) */ 408 NULL, /* 'tavt' (TAVT) */ 409 NULL, /* 'zmth' (ZMTH) */ 410 NULL /* 'zsym' (ZSYM) */ 411 }; 412 413 IndicClassTable::CharClass IndicClassTable::getCharClass(LEUnicode ch) const 414 { 415 if (ch == C_SIGN_ZWJ) { 416 return CF_CONSONANT | CC_ZERO_WIDTH_MARK; 417 } 418 419 if (ch == C_SIGN_ZWNJ) { 420 return CC_ZERO_WIDTH_MARK; 421 } 422 423 if (ch < firstChar || ch > lastChar) { 424 return CC_RESERVED; 425 } 426 427 return classTable[ch - firstChar]; 428 } 429 430 const IndicClassTable *IndicClassTable::getScriptClassTable(le_int32 scriptCode) 431 { 432 if (scriptCode < 0 || scriptCode >= scriptCodeCount) { 433 return NULL; 434 } 435 436 return indicClassTables[scriptCode]; 437 } 438 439 le_int32 IndicReordering::getWorstCaseExpansion(le_int32 scriptCode) 440 { 441 const IndicClassTable *classTable = IndicClassTable::getScriptClassTable(scriptCode); 442 443 if (classTable == NULL) { 444 return 1; 445 } 446 447 return classTable->getWorstCaseExpansion(); 448 } 449 450 le_bool IndicReordering::getFilterZeroWidth(le_int32 scriptCode) 451 { 452 const IndicClassTable *classTable = IndicClassTable::getScriptClassTable(scriptCode); 453 454 if (classTable == NULL) { 455 return TRUE; 456 } 457 458 return classTable->getFilterZeroWidth(); 459 } 460 461 U_NAMESPACE_END 462
