icuSources/layout/IndicClassTables.cpp

   1 /*
   2  *
   3  * (C) Copyright IBM Corp. 1998-2014 - 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 _fb  (_ct | CF_PRE_BASE)
  70 #define _vt  (_bb | CF_VATTU)
  71 #define _rv  (_vt | CF_REPH)
  72 #define _rp  (_pb | CF_REPH)
  73 #define _rb  (_bb | CF_REPH)
  74
  75 //
  76 // Character class tables
  77 //
  78 static const IndicClassTable::CharClass devaCharClasses[] =
  79 {
  80     _xx, _ma, _ma, _mp, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, // 0900 - 090F
  81     _iv, _iv, _iv, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0910 - 091F
  82     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _cn, _ct, _ct, _ct, _ct, _ct, _ct, // 0920 - 092F
  83     _rv, _cn, _ct, _ct, _cn, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _dr, _dl, // 0930 - 093F
  84     _dr, _db, _db, _db, _db, _da, _da, _da, _da, _dr, _dr, _dr, _dr, _vr, _xx, _xx, // 0940 - 094F
  85     _xx, _sa, _sb, _sa, _sa, _xx, _xx, _xx, _cn, _cn, _cn, _cn, _cn, _cn, _cn, _cn, // 0950 - 095F
  86     _iv, _iv, _db, _db, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0960 - 096F
  87     _xx                                                                             // 0970
  88 };
  89
  90 static const IndicClassTable::CharClass bengCharClasses[] =
  91 {
  92     _xx, _ma, _mp, _mp, _xx, _i2, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _i2, // 0980 - 098F
  93     _iv, _xx, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0990 - 099F
  94     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _bb, _ct, _ct, _pb, // 09A0 - 09AF
  95     _rv, _xx, _ct, _xx, _xx, _xx, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _r2, _dl, // 09B0 - 09BF
  96     _dr, _db, _db, _db, _db, _xx, _xx, _l1, _dl, _xx, _xx, _s1, _s2, _vr, _xx, _xx, // 09C0 - 09CF
  97     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _cn, _cn, _xx, _cn, // 09D0 - 09DF
  98     _iv, _iv, _dv, _dv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 09E0 - 09EF
  99     _rv, _ct, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx                           // 09F0 - 09FA
 100 };
 101
 102 static const IndicClassTable::CharClass punjCharClasses[] =
 103 {
 104     _xx, _ma, _ma, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _xx, _iv, // 0A00 - 0A0F
 105     _iv, _xx, _xx, _i3, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0A10 - 0A1F
 106     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _bb, // 0A20 - 0A2F
 107     _vt, _xx, _ct, _cn, _xx, _bb, _cn, _xx, _ct, _bb, _xx, _xx, _nu, _xx, _dr, _dl, // 0A30 - 0A3F
 108     _dr, _b2, _db, _xx, _xx, _xx, _xx, _da, _da, _xx, _xx, _a1, _da, _vr, _xx, _xx, // 0A40 - 0A4F
 109     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _cn, _cn, _cn, _ct, _xx, _cn, _xx, // 0A50 - 0A5F
 110     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0A60 - 0A6F
 111     _ma, _ma, _xx, _xx, _xx                                                         // 0A70 - 0A74
 112 };
 113
 114 static const IndicClassTable::CharClass gujrCharClasses[] =
 115 {
 116     _xx, _ma, _ma, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _xx, _iv, // 0A80 - 0A8F
 117     _iv, _iv, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0A90 - 0A9F
 118     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _ct, // 0AA0 - 0AAF
 119     _rv, _xx, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _dr, _dl, // 0AB0 - 0ABF
 120     _dr, _db, _db, _db, _db, _da, _xx, _da, _da, _dr, _xx, _dr, _dr, _vr, _xx, _xx, // 0AC0 - 0ACF
 121     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0AD0 - 0ADF
 122     _iv, _iv, _db, _db, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx  // 0AE0 - 0AEF
 123 };
 124
 125 #if 1
 126 static const IndicClassTable::CharClass oryaCharClasses[] =
 127 {
 128     _xx, _ma, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _iv, /* 0B00 - 0B0F */
 129     _iv, _xx, _xx, _iv, _iv, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _ct, _bb, /* 0B10 - 0B1F */
 130     _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _pb, /* 0B20 - 0B2F */
 131     _rb, _xx, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _xx, _xx, _nu, _xx, _dr, _da, /* 0B30 - 0B3F */
 132     _dr, _db, _db, _db, _xx, _xx, _xx, _dl, _s1, _xx, _xx, _s2, _s3, _vr, _xx, _xx, /* 0B40 - 0B4F */
 133     _xx, _xx, _xx, _xx, _xx, _xx, _da, _dr, _xx, _xx, _xx, _xx, _cn, _cn, _xx, _pb, /* 0B50 - 0B5F */
 134     _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, /* 0B60 - 0B6F */
 135     _xx, _bb                                                                        /* 0B70 - 0B71 */
 136 };
 137 #else
 138 static const IndicClassTable::CharClass oryaCharClasses[] =
 139 {
 140     _xx, _ma, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _iv, // 0B00 - 0B0F
 141     _iv, _xx, _xx, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0B10 - 0B1F
 142     _ct, _ct, _ct, _ct, _bb, _ct, _ct, _ct, _bb, _xx, _ct, _ct, _bb, _bb, _bb, _pb, // 0B20 - 0B2F
 143     _rb, _xx, _bb, _bb, _xx, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _nu, _xx, _r2, _da, // 0B30 - 0B3F
 144     _dr, _db, _db, _db, _xx, _xx, _xx, _l1, _s1, _xx, _xx, _s2, _s3, _vr, _xx, _xx, // 0B40 - 0B4F
 145     _xx, _xx, _xx, _xx, _xx, _xx, _m2, _m2, _xx, _xx, _xx, _xx, _cn, _cn, _xx, _cn, // 0B50 - 0B5F
 146     _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0B60 - 0B6F
 147     _xx, _ct                                                                        // 0B70 - 0B71
 148 };
 149 #endif
 150
 151 static const IndicClassTable::CharClass tamlCharClasses[] =
 152 {
 153     _xx, _xx, _ma, _xx, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _iv, _iv, // 0B80 - 0B8F
 154     _iv, _xx, _iv, _iv, _iv, _ct, _xx, _xx, _xx, _ct, _ct, _xx, _ct, _xx, _ct, _ct, // 0B90 - 0B9F
 155     _xx, _xx, _xx, _ct, _ct, _xx, _xx, _xx, _ct, _ct, _ct, _xx, _xx, _xx, _ct, _ct, // 0BA0 - 0BAF
 156     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _xx, _xx, _r2, _dr, // 0BB0 - 0BBF
 157     _da, _dr, _dr, _xx, _xx, _xx, _l1, _l1, _dl, _xx, _s1, _s2, _s3, _vr, _xx, _xx, // 0BC0 - 0BCF
 158     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0BD0 - 0BDF
 159     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0BE0 - 0BEF
 160     _xx, _xx, _xx                                                                   // 0BF0 - 0BF2
 161 };
 162
 163 // FIXME: Should some of the bb's be pb's? (KA, NA, MA, YA, VA, etc. (approx 13))
 164 // U+C43 and U+C44 are _lm here not _dr.  Similar to the situation with U+CC3 and
 165 // U+CC4 in Kannada below.
 166 static const IndicClassTable::CharClass teluCharClasses[] =
 167 {
 168     _xx, _mp, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _iv, // 0C00 - 0C0F
 169     _iv, _xx, _iv, _iv, _iv, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, // 0C10 - 0C1F
 170     _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _bb, // 0C20 - 0C2F
 171     _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _xx, _xx, _xx, _xx, _da, _da, // 0C30 - 0C3F
 172     _da, _dr, _dr, _lm, _lm, _xx, _a1, _da, _s1, _xx, _da, _da, _da, _vr, _xx, _xx, // 0C40 - 0C4F
 173     _xx, _xx, _xx, _xx, _xx, _da, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0C50 - 0C5F
 174     _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx  // 0C60 - 0C6F
 175 };
 176
 177 // U+CC3 and U+CC4 are _lm here not _dr since the Kannada rendering
 178 // rules want them below and to the right of the entire cluster
 179 //
 180 // There's some information about this in:
 181 //
 182 //  http://brahmi.sourceforge.net/docs/KannadaComputing.html
 183 static const IndicClassTable::CharClass kndaCharClasses[] =
 184 {
 185     _xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _iv, // 0C80 - 0C8F
 186     _iv, _xx, _iv, _iv, _iv, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, // 0C90 - 0C9F
 187     _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _bb, // 0CA0 - 0CAF
 188     _rb, _ct, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _xx, _xx, _xx, _xx, _dr, _da, // 0CB0 - 0CBF
 189     _s1, _dr, _r2, _lm, _lm, _xx, _a1, _s2, _s3, _xx, _s4, _s5, _da, _vr, _xx, _xx, // 0CC0 - 0CCF
 190     _xx, _xx, _xx, _xx, _xx, _m3, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _ct, _xx, // 0CD0 - 0CDF
 191     _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx  // 0CE0 - 0CEF
 192 };
 193
 194 // FIXME: this is correct for old-style Malayalam (MAL) but not for reformed Malayalam (MLR)
 195 // FIXME: should there be a REPH for old-style Malayalam?
 196 static const IndicClassTable::CharClass mlymCharClasses[] =
 197 {
 198     _xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _iv, // 0D00 - 0D0F
 199     _iv, _xx, _iv, _iv, _iv, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0D10 - 0D1F
 200     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _pb, // 0D20 - 0D2F
 201     _fb, _fb, _bb, _ct, _ct, _pb, _ct, _ct, _ct, _ct, _xx, _xx, _xx, _xx, _r2, _dr, // 0D30 - 0D3F
 202     _dr, _dr, _dr, _dr, _xx, _xx, _l1, _l1, _dl, _xx, _s1, _s2, _s3, _vr, _xx, _xx, // 0D40 - 0D4F
 203     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _m2, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0D50 - 0D5F
 204     _iv, _iv, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx  // 0D60 - 0D6F
 205 };
 206
 207 static const IndicClassTable::CharClass sinhCharClasses[] =
 208 {
 209     _xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, // 0D80 - 0D8F
 210     _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _xx, _xx, _ct, _ct, _ct, _ct, _ct, _ct, // 0D90 - 0D9F
 211     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, // 0DA0 - 0DAF
 212     _ct, _ct, _xx, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _ct, _xx, _xx, // 0DB0 - 0DBF
 213     _ct, _ct, _ct, _ct, _ct, _ct, _ct, _xx, _xx, _xx, _al, _xx, _xx, _xx, _xx, _dr, // 0DC0 - 0DCF
 214     _dr, _dr, _da, _da, _db, _xx, _db, _xx, _dr, _dl, _s1, _dl, _s2, _s3, _s4, _dr, // 0DD0 - 0DDF
 215     _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, _xx, // 0DE0 - 0DEF
 216     _xx, _xx, _dr, _dr, _xx                                                         // 0DF0 - 0DF4
 217 };
 218
 219 //
 220 // Split matra tables
 221 //
 222 static const SplitMatra bengSplitTable[] = {{0x09C7, 0x09BE}, {0x09C7, 0x09D7}};
 223
 224 static const SplitMatra oryaSplitTable[] = {{0x0B47, 0x0B56}, {0x0B47, 0x0B3E}, {0x0B47, 0x0B57}};
 225
 226 static const SplitMatra tamlSplitTable[] = {{0x0BC6, 0x0BBE}, {0x0BC7, 0x0BBE}, {0x0BC6, 0x0BD7}};
 227
 228 static const SplitMatra teluSplitTable[] = {{0x0C46, 0x0C56}};
 229
 230 static const SplitMatra kndaSplitTable[] = {{0x0CBF, 0x0CD5}, {0x0CC6, 0x0CD5}, {0x0CC6, 0x0CD6}, {0x0CC6, 0x0CC2},
 231                                             {0x0CC6, 0x0CC2, 0x0CD5}};
 232
 233 static const SplitMatra mlymSplitTable[] = {{0x0D46, 0x0D3E}, {0x0D47, 0x0D3E}, {0x0D46, 0x0D57}};
 234
 235
 236 static const SplitMatra sinhSplitTable[] = {{0x0DD9, 0x0DCA}, {0x0DD9, 0x0DCF}, {0x0DD9, 0x0DCF, 0x0DCA},
 237                                             {0x0DD9, 0x0DDF}};
 238 //
 239 // Script Flags
 240 //
 241
 242 // FIXME: post 'GSUB' reordering of MATRA_PRE's for Malayalam and Tamil
 243 // FIXME: reformed Malayalam needs to reorder VATTU to before base glyph...
 244 // FIXME: not sure passing ZWJ/ZWNJ is best way to render Malayalam Cillu...
 245 // FIXME: eyelash RA only for Devanagari??
 246 #define DEVA_SCRIPT_FLAGS (SF_EYELASH_RA | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
 247 #define BENG_SCRIPT_FLAGS (SF_REPH_AFTER_BELOW | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
 248 #define PUNJ_SCRIPT_FLAGS (SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
 249 #define GUJR_SCRIPT_FLAGS (SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
 250 #define ORYA_SCRIPT_FLAGS (SF_REPH_AFTER_BELOW | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
 251 #define TAML_SCRIPT_FLAGS (SF_MPRE_FIXUP | SF_NO_POST_BASE_LIMIT | SF_FILTER_ZERO_WIDTH)
 252 #define TELU_SCRIPT_FLAGS (SF_MATRAS_AFTER_BASE | SF_FILTER_ZERO_WIDTH | 3)
 253 #define KNDA_SCRIPT_FLAGS (SF_MATRAS_AFTER_BASE | SF_FILTER_ZERO_WIDTH | 3)
 254 #define MLYM_SCRIPT_FLAGS (SF_MPRE_FIXUP | SF_NO_POST_BASE_LIMIT /*| SF_FILTER_ZERO_WIDTH*/)
 255 #define SINH_SCRIPT_FLAGS (SF_NO_POST_BASE_LIMIT)
 256
 257 //
 258 // Indic Class Tables
 259 //
 260 static const IndicClassTable devaClassTable = {0x0900, 0x0970, 2, DEVA_SCRIPT_FLAGS, devaCharClasses, NULL};
 261
 262 static const IndicClassTable bengClassTable = {0x0980, 0x09FA, 3, BENG_SCRIPT_FLAGS, bengCharClasses, bengSplitTable};
 263
 264 static const IndicClassTable punjClassTable = {0x0A00, 0x0A74, 2, PUNJ_SCRIPT_FLAGS, punjCharClasses, NULL};
 265
 266 static const IndicClassTable gujrClassTable = {0x0A80, 0x0AEF, 2, GUJR_SCRIPT_FLAGS, gujrCharClasses, NULL};
 267
 268 static const IndicClassTable oryaClassTable = {0x0B00, 0x0B71, 3, ORYA_SCRIPT_FLAGS, oryaCharClasses, oryaSplitTable};
 269
 270 static const IndicClassTable tamlClassTable = {0x0B80, 0x0BF2, 3, TAML_SCRIPT_FLAGS, tamlCharClasses, tamlSplitTable};
 271
 272 static const IndicClassTable teluClassTable = {0x0C00, 0x0C6F, 3, TELU_SCRIPT_FLAGS, teluCharClasses, teluSplitTable};
 273
 274 static const IndicClassTable kndaClassTable = {0x0C80, 0x0CEF, 4, KNDA_SCRIPT_FLAGS, kndaCharClasses, kndaSplitTable};
 275
 276 static const IndicClassTable mlymClassTable = {0x0D00, 0x0D6F, 4, MLYM_SCRIPT_FLAGS, mlymCharClasses, mlymSplitTable};
 277
 278 static const IndicClassTable sinhClassTable = {0x0D80, 0x0DF4, 4, SINH_SCRIPT_FLAGS, sinhCharClasses, sinhSplitTable};
 279
 280 //
 281 // IndicClassTable addresses
 282 //
 283 static const IndicClassTable * const indicClassTables[scriptCodeCount] = {
 284     NULL,            /* 'zyyy' (COMMON) */
 285     NULL,            /* 'qaai' (INHERITED) */
 286     NULL,            /* 'arab' (ARABIC) */
 287     NULL,            /* 'armn' (ARMENIAN) */
 288     &bengClassTable, /* 'beng' (BENGALI) */
 289     NULL,            /* 'bopo' (BOPOMOFO) */
 290     NULL,            /* 'cher' (CHEROKEE) */
 291     NULL,            /* 'copt' (COPTIC) */
 292     NULL,            /* 'cyrl' (CYRILLIC) */
 293     NULL,            /* 'dsrt' (DESERET) */
 294     &devaClassTable, /* 'deva' (DEVANAGARI) */
 295     NULL,            /* 'ethi' (ETHIOPIC) */
 296     NULL,            /* 'geor' (GEORGIAN) */
 297     NULL,            /* 'goth' (GOTHIC) */
 298     NULL,            /* 'grek' (GREEK) */
 299     &gujrClassTable, /* 'gujr' (GUJARATI) */
 300     &punjClassTable, /* 'guru' (GURMUKHI) */
 301     NULL,            /* 'hani' (HAN) */
 302     NULL,            /* 'hang' (HANGUL) */
 303     NULL,            /* 'hebr' (HEBREW) */
 304     NULL,            /* 'hira' (HIRAGANA) */
 305     &kndaClassTable, /* 'knda' (KANNADA) */
 306     NULL,            /* 'kata' (KATAKANA) */
 307     NULL,            /* 'khmr' (KHMER) */
 308     NULL,            /* 'laoo' (LAO) */
 309     NULL,            /* 'latn' (LATIN) */
 310     &mlymClassTable, /* 'mlym' (MALAYALAM) */
 311     NULL,            /* 'mong' (MONGOLIAN) */
 312     NULL,            /* 'mymr' (MYANMAR) */
 313     NULL,            /* 'ogam' (OGHAM) */
 314     NULL,            /* 'ital' (OLD-ITALIC) */
 315     &oryaClassTable, /* 'orya' (ORIYA) */
 316     NULL,            /* 'runr' (RUNIC) */
 317     &sinhClassTable, /* 'sinh' (SINHALA) */
 318     NULL,            /* 'syrc' (SYRIAC) */
 319     &tamlClassTable, /* 'taml' (TAMIL) */
 320     &teluClassTable, /* 'telu' (TELUGU) */
 321     NULL,            /* 'thaa' (THAANA) */
 322     NULL,            /* 'thai' (THAI) */
 323     NULL,            /* 'tibt' (TIBETAN) */
 324     NULL,            /* 'cans' (CANADIAN-ABORIGINAL) */
 325     NULL,            /* 'yiii' (YI) */
 326     NULL,            /* 'tglg' (TAGALOG) */
 327     NULL,            /* 'hano' (HANUNOO) */
 328     NULL,            /* 'buhd' (BUHID) */
 329     NULL,            /* 'tagb' (TAGBANWA) */
 330     NULL,            /* 'brai' (BRAILLE) */
 331     NULL,            /* 'cprt' (CYPRIOT) */
 332     NULL,            /* 'limb' (LIMBU) */
 333     NULL,            /* 'linb' (LINEAR_B) */
 334     NULL,            /* 'osma' (OSMANYA) */
 335     NULL,            /* 'shaw' (SHAVIAN) */
 336     NULL,            /* 'tale' (TAI_LE) */
 337     NULL,            /* 'ugar' (UGARITIC) */
 338     NULL,            /* 'hrkt' (KATAKANA_OR_HIRAGANA) */
 339     NULL,            /* 'bugi' (BUGINESE) */
 340     NULL,            /* 'glag' (GLAGOLITIC) */
 341     NULL,            /* 'khar' (KHAROSHTHI) */
 342     NULL,            /* 'sylo' (SYLOTI_NAGRI) */
 343     NULL,            /* 'talu' (NEW_TAI_LUE) */
 344     NULL,            /* 'tfng' (TIFINAGH) */
 345     NULL,            /* 'xpeo' (OLD_PERSIAN) */
 346     NULL,            /* 'bali' (BALINESE) */
 347     NULL,            /* 'batk' (BATK) */
 348     NULL,            /* 'blis' (BLIS) */
 349     NULL,            /* 'brah' (BRAH) */
 350     NULL,            /* 'cham' (CHAM) */
 351     NULL,            /* 'cirt' (CIRT) */
 352     NULL,            /* 'cyrs' (CYRS) */
 353     NULL,            /* 'egyd' (EGYD) */
 354     NULL,            /* 'egyh' (EGYH) */
 355     NULL,            /* 'egyp' (EGYP) */
 356     NULL,            /* 'geok' (GEOK) */
 357     NULL,            /* 'hans' (HANS) */
 358     NULL,            /* 'hant' (HANT) */
 359     NULL,            /* 'hmng' (HMNG) */
 360     NULL,            /* 'hung' (HUNG) */
 361     NULL,            /* 'inds' (INDS) */
 362     NULL,            /* 'java' (JAVA) */
 363     NULL,            /* 'kali' (KALI) */
 364     NULL,            /* 'latf' (LATF) */
 365     NULL,            /* 'latg' (LATG) */
 366     NULL,            /* 'lepc' (LEPC) */
 367     NULL,            /* 'lina' (LINA) */
 368     NULL,            /* 'mand' (MAND) */
 369     NULL,            /* 'maya' (MAYA) */
 370     NULL,            /* 'mero' (MERO) */
 371     NULL,            /* 'nko ' (NKO) */
 372     NULL,            /* 'orkh' (ORKH) */
 373     NULL,            /* 'perm' (PERM) */
 374     NULL,            /* 'phag' (PHAGS_PA) */
 375     NULL,            /* 'phnx' (PHOENICIAN) */
 376     NULL,            /* 'plrd' (PLRD) */
 377     NULL,            /* 'roro' (RORO) */
 378     NULL,            /* 'sara' (SARA) */
 379     NULL,            /* 'syre' (SYRE) */
 380     NULL,            /* 'syrj' (SYRJ) */
 381     NULL,            /* 'syrn' (SYRN) */
 382     NULL,            /* 'teng' (TENG) */
 383     NULL,            /* 'vai ' (VAII) */
 384     NULL,            /* 'visp' (VISP) */
 385     NULL,            /* 'xsux' (CUNEIFORM) */
 386     NULL,            /* 'zxxx' (ZXXX) */
 387     NULL,            /* 'zzzz' (UNKNOWN) */
 388     NULL,            /* 'cari' (CARI) */
 389     NULL,            /* 'jpan' (JPAN) */
 390     NULL,            /* 'lana' (LANA) */
 391     NULL,            /* 'lyci' (LYCI) */
 392     NULL,            /* 'lydi' (LYDI) */
 393     NULL,            /* 'olck' (OLCK) */
 394     NULL,            /* 'rjng' (RJNG) */
 395     NULL,            /* 'saur' (SAUR) */
 396     NULL,            /* 'sgnw' (SGNW) */
 397     NULL,            /* 'sund' (SUND) */
 398     NULL,            /* 'moon' (MOON) */
 399     NULL,            /* 'mtei' (MTEI) */
 400     NULL,            /* 'armi' (ARMI) */
 401     NULL,            /* 'avst' (AVST) */
 402     NULL,            /* 'cakm' (CAKM) */
 403     NULL,            /* 'kore' (KORE) */
 404     NULL,            /* 'kthi' (KTHI) */
 405     NULL,            /* 'mani' (MANI) */
 406     NULL,            /* 'phli' (PHLI) */
 407     NULL,            /* 'phlp' (PHLP) */
 408     NULL,            /* 'phlv' (PHLV) */
 409     NULL,            /* 'prti' (PRTI) */
 410     NULL,            /* 'samr' (SAMR) */
 411     NULL,            /* 'tavt' (TAVT) */
 412     NULL,            /* 'zmth' (ZMTH) */
 413     NULL,            /* 'zsym' (ZSYM) */
 414     NULL,            /* 'bamu' (BAMUM) */
 415     NULL,            /* 'lisu' (LISU) */
 416     NULL,            /* 'nkgb' (NKGB) */
 417     NULL             /* 'sarb' (OLD_SOUTH_ARABIAN) */
 418 };
 419
 420 IndicClassTable::CharClass IndicClassTable::getCharClass(LEUnicode ch) const
 421 {
 422     if (ch == C_SIGN_ZWJ) {
 423         return CF_CONSONANT | CC_ZERO_WIDTH_MARK;
 424     }
 425
 426     if (ch == C_SIGN_ZWNJ) {
 427         return CC_ZERO_WIDTH_MARK;
 428     }
 429
 430     if (ch < firstChar || ch > lastChar) {
 431         return CC_RESERVED;
 432     }
 433
 434     return classTable[ch - firstChar];
 435 }
 436
 437 const IndicClassTable *IndicClassTable::getScriptClassTable(le_int32 scriptCode)
 438 {
 439     if (scriptCode < 0 || scriptCode >= scriptCodeCount) {
 440         return NULL;
 441     }
 442
 443     return indicClassTables[scriptCode];
 444 }
 445
 446 le_int32 IndicReordering::getWorstCaseExpansion(le_int32 scriptCode)
 447 {
 448     const IndicClassTable *classTable = IndicClassTable::getScriptClassTable(scriptCode);
 449
 450     if (classTable == NULL) {
 451         return 1;
 452     }
 453
 454     return classTable->getWorstCaseExpansion();
 455 }
 456
 457 le_bool IndicReordering::getFilterZeroWidth(le_int32 scriptCode)
 458 {
 459     const IndicClassTable *classTable = IndicClassTable::getScriptClassTable(scriptCode);
 460
 461     if (classTable == NULL) {
 462         return TRUE;
 463     }
 464
 465     return classTable->getFilterZeroWidth();
 466 }
 467
 468 U_NAMESPACE_END