lib/taranges.c

   1 /* taranges.c */
   2
   3 /*
   4  * Copyright (C) 2014-2020 by Werner Lemberg.
   5  *
   6  * This file is part of the ttfautohint library, and may only be used,
   7  * modified, and distributed under the terms given in `COPYING'.  By
   8  * continuing to use, modify, or distribute this file you indicate that you
   9  * have read `COPYING' and understand and accept it fully.
  10  *
  11  * The file `COPYING' mentioned in the previous paragraph is distributed
  12  * with the ttfautohint library.
  13  */
  14
  15
  16 /* originally file `afranges.c' (2014-Jan-11) from FreeType */
  17
  18 /* heavily modified 2014 by Werner Lemberg <wl@gnu.org> */
  19
  20 #include "taranges.h"
  21
  22 /*
  23  * This file gets also processed with the `taranges.sed' script to produce
  24  * documentation of character ranges.  To make this simple approach work,
  25  * don't change the formatting in this file!
  26  *
  27  *   - Everything before the first `const' keyword (starting a line) gets
  28  *     removed.
  29  *
  30  *   - The line containing `none_uniranges' and everything after it gets
  31  *     removed, too.
  32  *
  33  *   - Comments after a `TA_UNIRANGE_REC' entry are used for character range
  34  *     documentation within a table.
  35  *
  36  *   - Comments indented by two spaces are also used within a table.
  37  *
  38  *   - Other comments are inserted into the documentation as-is (after
  39  *     stripping off the comment characters).
  40  */
  41
  42
  43 /*
  44  * The algorithm for assigning properties and styles to the `glyph_styles'
  45  * array is as follows (cf. the implementation in
  46  * `ta_face_globals_compute_style_coverage').
  47  *
  48  *   Walk over all scripts (as listed in `tascript.h').
  49  *
  50  *   For a given script, walk over all styles (as listed in `tastyles.h').
  51  *   The order of styles is important and should be as follows.
  52  *
  53  *   - First come styles based on OpenType features (small caps, for
  54  *     example).  Since features rely on glyph indices, thus completely
  55  *     bypassing character codes, no properties are assigned.
  56  *
  57  *   - Next comes the default style, using the character ranges as defined
  58  *     below.  This also assigns properties.
  59  *
  60  *   Note that there also exist fallback scripts, mainly covering
  61  *   superscript and subscript glyphs of a script that are not present as
  62  *   OpenType features.  Fallback scripts are defined below, also
  63  *   assigning properties; they are applied after the corresponding
  64  *   script.
  65  *
  66  */
  67
  68
  69 /* XXX Check base character ranges again: */
  70 /*     Right now, they are quickly derived by visual inspection. */
  71 /*     I can imagine that fine-tuning is necessary. */
  72
  73 /* for the auto-hinter, a `non-base character' is something that should */
  74 /* not be affected by blue zones, regardless of whether this is a */
  75 /* spacing or no-spacing glyph */
  76
  77 /* the `ta_xxxx_nonbase_uniranges' ranges must be strict subsets */
  78 /* of the corresponding `ta_xxxx_uniranges' ranges */
  79
  80
  81 const TA_Script_UniRangeRec ta_adlm_uniranges[] =
  82 {
  83   TA_UNIRANGE_REC(0x1E900, 0x1E95F), /* Adlam */
  84   TA_UNIRANGE_REC(      0,       0)
  85 };
  86
  87 const TA_Script_UniRangeRec ta_adlm_nonbase_uniranges[] =
  88 {
  89   TA_UNIRANGE_REC(0x1D944, 0x1E94A),
  90   TA_UNIRANGE_REC(      0,       0)
  91 };
  92
  93
  94 const TA_Script_UniRangeRec ta_arab_uniranges[] =
  95 {
  96   TA_UNIRANGE_REC( 0x0600,  0x06FF), /* Arabic */
  97   TA_UNIRANGE_REC( 0x0750,  0x07FF), /* Arabic Supplement */
  98   TA_UNIRANGE_REC( 0x08A0,  0x08FF), /* Arabic Extended-A */
  99   TA_UNIRANGE_REC( 0xFB50,  0xFDFF), /* Arabic Presentation Forms-A */
 100   TA_UNIRANGE_REC( 0xFE70,  0xFEFF), /* Arabic Presentation Forms-B */
 101   TA_UNIRANGE_REC(0x1EE00, 0x1EEFF), /* Arabic Mathematical Alphabetic Symbols */
 102   TA_UNIRANGE_REC(      0,       0)
 103 };
 104
 105 const TA_Script_UniRangeRec ta_arab_nonbase_uniranges[] =
 106 {
 107   TA_UNIRANGE_REC(0x0600, 0x0605),
 108   TA_UNIRANGE_REC(0x0610, 0x061A),
 109   TA_UNIRANGE_REC(0x064B, 0x065F),
 110   TA_UNIRANGE_REC(0x0670, 0x0670),
 111   TA_UNIRANGE_REC(0x06D6, 0x06DC),
 112   TA_UNIRANGE_REC(0x06DF, 0x06E4),
 113   TA_UNIRANGE_REC(0x06E7, 0x06E8),
 114   TA_UNIRANGE_REC(0x06EA, 0x06ED),
 115   TA_UNIRANGE_REC(0x08D3, 0x08FF),
 116   TA_UNIRANGE_REC(0xFBB2, 0xFBC1),
 117   TA_UNIRANGE_REC(0xFE70, 0xFE70),
 118   TA_UNIRANGE_REC(0xFE72, 0xFE72),
 119   TA_UNIRANGE_REC(0xFE74, 0xFE74),
 120   TA_UNIRANGE_REC(0xFE76, 0xFE76),
 121   TA_UNIRANGE_REC(0xFE78, 0xFE78),
 122   TA_UNIRANGE_REC(0xFE7A, 0xFE7A),
 123   TA_UNIRANGE_REC(0xFE7C, 0xFE7C),
 124   TA_UNIRANGE_REC(0xFE7E, 0xFE7E),
 125   TA_UNIRANGE_REC(     0,      0)
 126 };
 127
 128
 129 const TA_Script_UniRangeRec ta_armn_uniranges[] =
 130 {
 131   TA_UNIRANGE_REC(0x0530, 0x058F), /* Armenian */
 132   TA_UNIRANGE_REC(0xFB13, 0xFB17), /* Alphab. Present. Forms (Armenian) */
 133   TA_UNIRANGE_REC(     0,      0)
 134 };
 135
 136 const TA_Script_UniRangeRec ta_armn_nonbase_uniranges[] =
 137 {
 138   TA_UNIRANGE_REC(0x0559, 0x055F),
 139   TA_UNIRANGE_REC(     0,      0)
 140 };
 141
 142
 143 const TA_Script_UniRangeRec ta_avst_uniranges[] =
 144 {
 145   TA_UNIRANGE_REC(0x10B00, 0x10B3F), /* Avestan */
 146   TA_UNIRANGE_REC(      0,       0)
 147 };
 148
 149 const TA_Script_UniRangeRec ta_avst_nonbase_uniranges[] =
 150 {
 151   TA_UNIRANGE_REC(0x10B39, 0x10B3F),
 152   TA_UNIRANGE_REC(      0,       0)
 153 };
 154
 155
 156 const TA_Script_UniRangeRec ta_bamu_uniranges[] =
 157 {
 158   TA_UNIRANGE_REC( 0xA6A0,  0xA6FF), /* Bamum */
 159 #if 0
 160   /* The characters in the Bamum supplement are pictograms, */
 161   /* not (directly) related to the syllabic Bamum script */
 162   TA_UNIRANGE_REC(0x16800, 0x16A3F), /* Bamum Supplement */
 163 #endif
 164   TA_UNIRANGE_REC(      0,       0)
 165 };
 166
 167 const TA_Script_UniRangeRec ta_bamu_nonbase_uniranges[] =
 168 {
 169   TA_UNIRANGE_REC(0xA6F0,  0xA6F1),
 170   TA_UNIRANGE_REC(     0,       0)
 171 };
 172
 173
 174 const TA_Script_UniRangeRec ta_beng_uniranges[] =
 175 {
 176   TA_UNIRANGE_REC(0x0980, 0x09FF), /* Bengali */
 177   TA_UNIRANGE_REC(     0,      0)
 178 };
 179
 180 const TA_Script_UniRangeRec ta_beng_nonbase_uniranges[] =
 181 {
 182   TA_UNIRANGE_REC(0x0981, 0x0981),
 183   TA_UNIRANGE_REC(0x09BC, 0x09BC),
 184   TA_UNIRANGE_REC(0x09C1, 0x09C4),
 185   TA_UNIRANGE_REC(0x09CD, 0x09CD),
 186   TA_UNIRANGE_REC(0x09E2, 0x09E3),
 187   TA_UNIRANGE_REC(0x09FE, 0x09FE),
 188   TA_UNIRANGE_REC(     0,      0)
 189 };
 190
 191
 192 const TA_Script_UniRangeRec ta_buhd_uniranges[] =
 193 {
 194   TA_UNIRANGE_REC(0x1740,  0x175F), /* Buhid */
 195   TA_UNIRANGE_REC(     0,       0)
 196 };
 197
 198 const TA_Script_UniRangeRec ta_buhd_nonbase_uniranges[] =
 199 {
 200   TA_UNIRANGE_REC(0x1752,  0x1753),
 201   TA_UNIRANGE_REC(     0,       0)
 202 };
 203
 204
 205 const TA_Script_UniRangeRec ta_cakm_uniranges[] =
 206 {
 207   TA_UNIRANGE_REC(0x11100, 0x1114F), /* Chakma */
 208   TA_UNIRANGE_REC(      0,       0)
 209 };
 210
 211 const TA_Script_UniRangeRec ta_cakm_nonbase_uniranges[] =
 212 {
 213   TA_UNIRANGE_REC(0x11100, 0x11102),
 214   TA_UNIRANGE_REC(0x11127, 0x11134),
 215   TA_UNIRANGE_REC(0x11146, 0x11146),
 216   TA_UNIRANGE_REC(      0,       0)
 217 };
 218
 219
 220 const TA_Script_UniRangeRec ta_cans_uniranges[] =
 221 {
 222   TA_UNIRANGE_REC(0x1400,  0x167F), /* Unified Canadian Aboriginal Syllabics */
 223   TA_UNIRANGE_REC(0x18B0,  0x18FF), /* Unified Canadian Aboriginal Syllabics Extended */
 224   TA_UNIRANGE_REC(     0,       0)
 225 };
 226
 227 const TA_Script_UniRangeRec ta_cans_nonbase_uniranges[] =
 228 {
 229   TA_UNIRANGE_REC(0, 0)
 230 };
 231
 232
 233 const TA_Script_UniRangeRec ta_cari_uniranges[] =
 234 {
 235   TA_UNIRANGE_REC(0x102A0, 0x102DF), /* Carian */
 236   TA_UNIRANGE_REC(      0,       0)
 237 };
 238
 239 const TA_Script_UniRangeRec ta_cari_nonbase_uniranges[] =
 240 {
 241   TA_UNIRANGE_REC(0, 0)
 242 };
 243
 244
 245 const TA_Script_UniRangeRec ta_cher_uniranges[] =
 246 {
 247   TA_UNIRANGE_REC(0x13A0, 0x13FF), /* Cherokee */
 248   TA_UNIRANGE_REC(0xAB70, 0xABBF), /* Cherokee Supplement */
 249   TA_UNIRANGE_REC(     0,      0)
 250 };
 251
 252 const TA_Script_UniRangeRec ta_cher_nonbase_uniranges[] =
 253 {
 254   TA_UNIRANGE_REC(0, 0)
 255 };
 256
 257
 258 const TA_Script_UniRangeRec ta_copt_uniranges[] =
 259 {
 260   TA_UNIRANGE_REC(0x2C80, 0x2CFF), /* Coptic */
 261   TA_UNIRANGE_REC(     0,      0)
 262 };
 263
 264 const TA_Script_UniRangeRec ta_copt_nonbase_uniranges[] =
 265 {
 266   TA_UNIRANGE_REC(0x2CEF, 0x2CF1),
 267   TA_UNIRANGE_REC(     0,      0)
 268 };
 269
 270
 271 const TA_Script_UniRangeRec ta_cprt_uniranges[] =
 272 {
 273   TA_UNIRANGE_REC(0x10800, 0x1083F), /* Cypriot */
 274   TA_UNIRANGE_REC(      0,       0)
 275 };
 276
 277 const TA_Script_UniRangeRec ta_cprt_nonbase_uniranges[] =
 278 {
 279   TA_UNIRANGE_REC(0, 0)
 280 };
 281
 282
 283 const TA_Script_UniRangeRec ta_cyrl_uniranges[] =
 284 {
 285   TA_UNIRANGE_REC(0x0400, 0x04FF), /* Cyrillic */
 286   TA_UNIRANGE_REC(0x0500, 0x052F), /* Cyrillic Supplement */
 287   TA_UNIRANGE_REC(0x2DE0, 0x2DFF), /* Cyrillic Extended-A */
 288   TA_UNIRANGE_REC(0xA640, 0xA69F), /* Cyrillic Extended-B */
 289   TA_UNIRANGE_REC(     0,      0)
 290 };
 291
 292 const TA_Script_UniRangeRec ta_cyrl_nonbase_uniranges[] =
 293 {
 294   TA_UNIRANGE_REC(0x0483, 0x0489),
 295   TA_UNIRANGE_REC(0x2DE0, 0x2DFF),
 296   TA_UNIRANGE_REC(0xA66F, 0xA67F),
 297   TA_UNIRANGE_REC(0xA69E, 0xA69F),
 298   TA_UNIRANGE_REC(     0,      0)
 299 };
 300
 301
 302 /* There are some characters in the Devanagari Unicode block that are */
 303 /* generic to Indic scripts; we omit them so that their presence doesn't */
 304 /* trigger Devanagari. */
 305
 306 const TA_Script_UniRangeRec ta_deva_uniranges[] =
 307 {
 308   TA_UNIRANGE_REC(0x0900, 0x093B), /* Devanagari */
 309   /* omitting U+093C nukta */
 310   TA_UNIRANGE_REC(0x093D, 0x0950), /* ... continued */
 311   /* omitting U+0951 udatta, U+0952 anudatta */
 312   TA_UNIRANGE_REC(0x0953, 0x0963), /* ... continued */
 313   /* omitting U+0964 danda, U+0965 double danda */
 314   TA_UNIRANGE_REC(0x0966, 0x097F), /* ... continued */
 315   TA_UNIRANGE_REC(0x20B9, 0x20B9), /* (new) Rupee sign */
 316   TA_UNIRANGE_REC(0xA8E0, 0xA8FF), /* Devanagari Extended */
 317   TA_UNIRANGE_REC(     0,      0)
 318 };
 319
 320 const TA_Script_UniRangeRec ta_deva_nonbase_uniranges[] =
 321 {
 322   TA_UNIRANGE_REC(0x0900, 0x0902),
 323   TA_UNIRANGE_REC(0x093A, 0x093A),
 324   TA_UNIRANGE_REC(0x0941, 0x0948),
 325   TA_UNIRANGE_REC(0x094D, 0x094D),
 326   TA_UNIRANGE_REC(0x0953, 0x0957),
 327   TA_UNIRANGE_REC(0x0962, 0x0963),
 328   TA_UNIRANGE_REC(0xA8E0, 0xA8F1),
 329   TA_UNIRANGE_REC(0xA8FF, 0xA8FF),
 330   TA_UNIRANGE_REC(     0,      0)
 331 };
 332
 333
 334 const TA_Script_UniRangeRec ta_dsrt_uniranges[] =
 335 {
 336   TA_UNIRANGE_REC(0x10400, 0x1044F), /* Deseret */
 337   TA_UNIRANGE_REC(      0,       0)
 338 };
 339
 340 const TA_Script_UniRangeRec ta_dsrt_nonbase_uniranges[] =
 341 {
 342   TA_UNIRANGE_REC(0, 0)
 343 };
 344
 345
 346 const TA_Script_UniRangeRec ta_ethi_uniranges[] =
 347 {
 348   TA_UNIRANGE_REC(0x1200, 0x137F), /* Ethiopic */
 349   TA_UNIRANGE_REC(0x1380, 0x139F), /* Ethiopic Supplement */
 350   TA_UNIRANGE_REC(0x2D80, 0x2DDF), /* Ethiopic Extended */
 351   TA_UNIRANGE_REC(0xAB00, 0xAB2F), /* Ethiopic Extended-A */
 352   TA_UNIRANGE_REC(     0,      0)
 353 };
 354
 355 const TA_Script_UniRangeRec ta_ethi_nonbase_uniranges[] =
 356 {
 357   TA_UNIRANGE_REC(0x135D, 0x135F),
 358   TA_UNIRANGE_REC(     0,      0)
 359 };
 360
 361
 362 const TA_Script_UniRangeRec ta_geor_uniranges[] =
 363 {
 364   TA_UNIRANGE_REC(0x10D0, 0x10FF), /* Georgian (Mkhedruli) */
 365   TA_UNIRANGE_REC(0x1C90, 0x1CBF), /* Georgian Extended (Mtavruli) */
 366   TA_UNIRANGE_REC(     0,      0)
 367 };
 368
 369 const TA_Script_UniRangeRec ta_geor_nonbase_uniranges[] =
 370 {
 371   TA_UNIRANGE_REC(0, 0)
 372 };
 373
 374
 375 const TA_Script_UniRangeRec ta_geok_uniranges[] =
 376 {
 377   /* Khutsuri */
 378   TA_UNIRANGE_REC(0x10A0, 0x10CD), /* Georgian (Asomtavruli) */
 379   TA_UNIRANGE_REC(0x2D00, 0x2D2D), /* Georgian Supplement (Nuskhuri) */
 380   TA_UNIRANGE_REC(     0,      0)
 381 };
 382
 383 const TA_Script_UniRangeRec ta_geok_nonbase_uniranges[] =
 384 {
 385   TA_UNIRANGE_REC(0, 0)
 386 };
 387
 388
 389 const TA_Script_UniRangeRec ta_glag_uniranges[] =
 390 {
 391   TA_UNIRANGE_REC( 0x2C00,  0x2C5F), /* Glagolitic */
 392   TA_UNIRANGE_REC(0x1E000, 0x1E02F), /* Glagolitic Supplement */
 393   TA_UNIRANGE_REC(      0,       0)
 394 };
 395
 396 const TA_Script_UniRangeRec ta_glag_nonbase_uniranges[] =
 397 {
 398   TA_UNIRANGE_REC(0x1E000, 0x1E02F),
 399   TA_UNIRANGE_REC(      0,       0)
 400 };
 401
 402
 403 const TA_Script_UniRangeRec ta_goth_uniranges[] =
 404 {
 405   TA_UNIRANGE_REC(0x10330, 0x1034F), /* Gothic */
 406   TA_UNIRANGE_REC(      0,       0)
 407 };
 408
 409 const TA_Script_UniRangeRec ta_goth_nonbase_uniranges[] =
 410 {
 411   TA_UNIRANGE_REC(0, 0)
 412 };
 413
 414
 415 const TA_Script_UniRangeRec ta_grek_uniranges[] =
 416 {
 417   TA_UNIRANGE_REC(0x0370, 0x03FF), /* Greek and Coptic */
 418   TA_UNIRANGE_REC(0x1F00, 0x1FFF), /* Greek Extended */
 419   TA_UNIRANGE_REC(     0,      0)
 420 };
 421
 422 const TA_Script_UniRangeRec ta_grek_nonbase_uniranges[] =
 423 {
 424   TA_UNIRANGE_REC(0x037A, 0x037A),
 425   TA_UNIRANGE_REC(0x0384, 0x0385),
 426   TA_UNIRANGE_REC(0x1FBD, 0x1FC1),
 427   TA_UNIRANGE_REC(0x1FCD, 0x1FCF),
 428   TA_UNIRANGE_REC(0x1FDD, 0x1FDF),
 429   TA_UNIRANGE_REC(0x1FED, 0x1FEF),
 430   TA_UNIRANGE_REC(0x1FFD, 0x1FFE),
 431   TA_UNIRANGE_REC(     0,      0)
 432 };
 433
 434
 435 const TA_Script_UniRangeRec ta_gujr_uniranges[] =
 436 {
 437   TA_UNIRANGE_REC(0x0A80, 0x0AFF), /* Gujarati */
 438   TA_UNIRANGE_REC(     0,      0)
 439 };
 440
 441 const TA_Script_UniRangeRec ta_gujr_nonbase_uniranges[] =
 442 {
 443   TA_UNIRANGE_REC(0x0A81, 0x0A82),
 444   TA_UNIRANGE_REC(0x0ABC, 0x0ABC),
 445   TA_UNIRANGE_REC(0x0AC1, 0x0AC8),
 446   TA_UNIRANGE_REC(0x0ACD, 0x0ACD),
 447   TA_UNIRANGE_REC(0x0AE2, 0x0AE3),
 448   TA_UNIRANGE_REC(0x0AFA, 0x0AFF),
 449   TA_UNIRANGE_REC(     0,      0)
 450 };
 451
 452
 453 const TA_Script_UniRangeRec ta_guru_uniranges[] =
 454 {
 455   TA_UNIRANGE_REC(0x0A00, 0x0A7F), /* Gurmukhi */
 456   TA_UNIRANGE_REC(     0,      0)
 457 };
 458
 459 const TA_Script_UniRangeRec ta_guru_nonbase_uniranges[] =
 460 {
 461   TA_UNIRANGE_REC(0x0A01, 0x0A02),
 462   TA_UNIRANGE_REC(0x0A3C, 0x0A3C),
 463   TA_UNIRANGE_REC(0x0A41, 0x0A51),
 464   TA_UNIRANGE_REC(0x0A70, 0x0A71),
 465   TA_UNIRANGE_REC(0x0A75, 0x0A75),
 466   TA_UNIRANGE_REC(     0,      0)
 467 };
 468
 469
 470 const TA_Script_UniRangeRec ta_hebr_uniranges[] =
 471 {
 472   TA_UNIRANGE_REC(0x0590, 0x05FF), /* Hebrew */
 473   TA_UNIRANGE_REC(0xFB1D, 0xFB4F), /* Alphab. Present. Forms (Hebrew) */
 474   TA_UNIRANGE_REC(     0,      0)
 475 };
 476
 477 const TA_Script_UniRangeRec ta_hebr_nonbase_uniranges[] =
 478 {
 479   TA_UNIRANGE_REC(0x0591, 0x05BF),
 480   TA_UNIRANGE_REC(0x05C1, 0x05C2),
 481   TA_UNIRANGE_REC(0x05C4, 0x05C5),
 482   TA_UNIRANGE_REC(0x05C7, 0x05C7),
 483   TA_UNIRANGE_REC(0xFB1E, 0xFB1E),
 484   TA_UNIRANGE_REC(     0,      0)
 485 };
 486
 487
 488 const TA_Script_UniRangeRec ta_kali_uniranges[] =
 489 {
 490   TA_UNIRANGE_REC(0xA900, 0xA92F), /* Kayah Li */
 491   TA_UNIRANGE_REC(     0,      0)
 492 };
 493
 494 const TA_Script_UniRangeRec ta_kali_nonbase_uniranges[] =
 495 {
 496   TA_UNIRANGE_REC(0xA926, 0xA92D),
 497   TA_UNIRANGE_REC(     0,      0)
 498 };
 499
 500
 501 const TA_Script_UniRangeRec ta_knda_uniranges[] =
 502 {
 503   TA_UNIRANGE_REC(0x0C80, 0x0CFF), /* Kannada */
 504   TA_UNIRANGE_REC(     0,      0)
 505 };
 506
 507 const TA_Script_UniRangeRec ta_knda_nonbase_uniranges[] =
 508 {
 509   TA_UNIRANGE_REC(0x0C81, 0x0C81),
 510   TA_UNIRANGE_REC(0x0CBC, 0x0CBC),
 511   TA_UNIRANGE_REC(0x0CBF, 0x0CBF),
 512   TA_UNIRANGE_REC(0x0CC6, 0x0CC6),
 513   TA_UNIRANGE_REC(0x0CCC, 0x0CCD),
 514   TA_UNIRANGE_REC(0x0CE2, 0x0CE3),
 515   TA_UNIRANGE_REC(     0,      0)
 516 };
 517
 518
 519 const TA_Script_UniRangeRec ta_khmr_uniranges[] =
 520 {
 521   TA_UNIRANGE_REC(0x1780, 0x17FF), /* Khmer */
 522   TA_UNIRANGE_REC(     0,      0)
 523 };
 524
 525 const TA_Script_UniRangeRec ta_khmr_nonbase_uniranges[] =
 526 {
 527   TA_UNIRANGE_REC(0x17B7, 0x17BD),
 528   TA_UNIRANGE_REC(0x17C6, 0x17C6),
 529   TA_UNIRANGE_REC(0x17C9, 0x17D3),
 530   TA_UNIRANGE_REC(0x17DD, 0x17DD),
 531   TA_UNIRANGE_REC(     0,      0)
 532 };
 533
 534
 535 const TA_Script_UniRangeRec ta_khms_uniranges[] =
 536 {
 537   TA_UNIRANGE_REC(0x19E0, 0x19FF), /* Khmer Symbols */
 538   TA_UNIRANGE_REC(     0,      0)
 539 };
 540
 541 const TA_Script_UniRangeRec ta_khms_nonbase_uniranges[] =
 542 {
 543   TA_UNIRANGE_REC(0, 0)
 544 };
 545
 546
 547 const TA_Script_UniRangeRec ta_lao_uniranges[] =
 548 {
 549   TA_UNIRANGE_REC(0x0E80, 0x0EFF), /* Lao */
 550   TA_UNIRANGE_REC(     0,      0)
 551 };
 552
 553 const TA_Script_UniRangeRec ta_lao_nonbase_uniranges[] =
 554 {
 555   TA_UNIRANGE_REC(0x0EB1, 0x0EB1),
 556   TA_UNIRANGE_REC(0x0EB4, 0x0EBC),
 557   TA_UNIRANGE_REC(0x0EC8, 0x0ECD),
 558   TA_UNIRANGE_REC(     0,      0)
 559 };
 560
 561
 562 const TA_Script_UniRangeRec ta_latn_uniranges[] =
 563 {
 564   TA_UNIRANGE_REC( 0x0020,  0x007F), /* Basic Latin (no control chars) */
 565   TA_UNIRANGE_REC( 0x00A0,  0x00A9), /* Latin-1 Supplement (no control chars) */
 566   TA_UNIRANGE_REC( 0x00AB,  0x00B1), /* ... continued */
 567   TA_UNIRANGE_REC( 0x00B4,  0x00B8), /* ... continued */
 568   TA_UNIRANGE_REC( 0x00BB,  0x00FF), /* ... continued */
 569   TA_UNIRANGE_REC( 0x0100,  0x017F), /* Latin Extended-A */
 570   TA_UNIRANGE_REC( 0x0180,  0x024F), /* Latin Extended-B */
 571   TA_UNIRANGE_REC( 0x0250,  0x02AF), /* IPA Extensions */
 572   TA_UNIRANGE_REC( 0x02B9,  0x02DF), /* Spacing Modifier Letters */
 573   TA_UNIRANGE_REC( 0x02E5,  0x02FF), /* ... continued */
 574   TA_UNIRANGE_REC( 0x0300,  0x036F), /* Combining Diacritical Marks */
 575   TA_UNIRANGE_REC( 0x1AB0,  0x1ABE), /* Combining Diacritical Marks Extended */
 576   TA_UNIRANGE_REC( 0x1D00,  0x1D2B), /* Phonetic Extensions */
 577   TA_UNIRANGE_REC( 0x1D6B,  0x1D77), /* ... continued */
 578   TA_UNIRANGE_REC( 0x1D79,  0x1D7F), /* ... continued */
 579   TA_UNIRANGE_REC( 0x1D80,  0x1D9A), /* Phonetic Extensions Supplement */
 580   TA_UNIRANGE_REC( 0x1DC0,  0x1DFF), /* Combining Diacritical Marks Supplement */
 581   TA_UNIRANGE_REC( 0x1E00,  0x1EFF), /* Latin Extended Additional */
 582   TA_UNIRANGE_REC( 0x2000,  0x206F), /* General Punctuation */
 583   TA_UNIRANGE_REC( 0x20A0,  0x20B8), /* Currency Symbols ... */
 584   TA_UNIRANGE_REC( 0x20BA,  0x20CF), /* ... except new Rupee sign */
 585   TA_UNIRANGE_REC( 0x2150,  0x218F), /* Number Forms */
 586   TA_UNIRANGE_REC( 0x2C60,  0x2C7B), /* Latin Extended-C */
 587   TA_UNIRANGE_REC( 0x2C7E,  0x2C7F), /* ... continued */
 588   TA_UNIRANGE_REC( 0x2E00,  0x2E7F), /* Supplemental Punctuation */
 589   TA_UNIRANGE_REC( 0xA720,  0xA76F), /* Latin Extended-D */
 590   TA_UNIRANGE_REC( 0xA771,  0xA7F7), /* ... continued */
 591   TA_UNIRANGE_REC( 0xA7FA,  0xA7FF), /* ... continued */
 592   TA_UNIRANGE_REC( 0xAB30,  0xAB5B), /* Latin Extended-E */
 593   TA_UNIRANGE_REC( 0xAB60,  0xAB6F), /* ... continued */
 594   TA_UNIRANGE_REC( 0xFB00,  0xFB06), /* Alphab. Present. Forms (Latin Ligs) */
 595   TA_UNIRANGE_REC(0x1D400, 0x1D7FF), /* Mathematical Alphanumeric Symbols */
 596   TA_UNIRANGE_REC(      0,       0)
 597 };
 598
 599 const TA_Script_UniRangeRec ta_latn_nonbase_uniranges[] =
 600 {
 601   TA_UNIRANGE_REC(0x005E, 0x0060),
 602   TA_UNIRANGE_REC(0x007E, 0x007E),
 603   TA_UNIRANGE_REC(0x00A8, 0x00A9),
 604   TA_UNIRANGE_REC(0x00AE, 0x00B0),
 605   TA_UNIRANGE_REC(0x00B4, 0x00B4),
 606   TA_UNIRANGE_REC(0x00B8, 0x00B8),
 607   TA_UNIRANGE_REC(0x00BC, 0x00BE),
 608   TA_UNIRANGE_REC(0x02B9, 0x02DF),
 609   TA_UNIRANGE_REC(0x02E5, 0x02FF),
 610   TA_UNIRANGE_REC(0x0300, 0x036F),
 611   TA_UNIRANGE_REC(0x1AB0, 0x1ABE),
 612   TA_UNIRANGE_REC(0x1DC0, 0x1DFF),
 613   TA_UNIRANGE_REC(0x2017, 0x2017),
 614   TA_UNIRANGE_REC(0x203E, 0x203E),
 615   TA_UNIRANGE_REC(0xA788, 0xA788),
 616   TA_UNIRANGE_REC(0xA7F8, 0xA7FA),
 617   TA_UNIRANGE_REC(     0,      0)
 618 };
 619
 620
 621 const TA_Script_UniRangeRec ta_latb_uniranges[] =
 622 {
 623   TA_UNIRANGE_REC(0x1D62, 0x1D6A), /* some small subscript letters */
 624   TA_UNIRANGE_REC(0x2080, 0x209C), /* subscript digits and letters */
 625   TA_UNIRANGE_REC(0x2C7C, 0x2C7C), /* latin subscript small letter j */
 626   TA_UNIRANGE_REC(     0,      0)
 627 };
 628
 629 const TA_Script_UniRangeRec ta_latb_nonbase_uniranges[] =
 630 {
 631   TA_UNIRANGE_REC(0, 0)
 632 };
 633
 634
 635 const TA_Script_UniRangeRec ta_latp_uniranges[] =
 636 {
 637   TA_UNIRANGE_REC(0x00AA, 0x00AA), /* feminine ordinal indicator */
 638   TA_UNIRANGE_REC(0x00B2, 0x00B3), /* superscript two and three */
 639   TA_UNIRANGE_REC(0x00B9, 0x00BA), /* superscript one, masc. ord. indic. */
 640   TA_UNIRANGE_REC(0x02B0, 0x02B8), /* some latin superscript mod. letters */
 641   TA_UNIRANGE_REC(0x02E0, 0x02E4), /* some IPA modifier letters */
 642   TA_UNIRANGE_REC(0x1D2C, 0x1D61), /* latin superscript modifier letters */
 643   TA_UNIRANGE_REC(0x1D78, 0x1D78), /* modifier letter cyrillic en */
 644   TA_UNIRANGE_REC(0x1D9B, 0x1DBF), /* more modifier letters */
 645   TA_UNIRANGE_REC(0x2070, 0x207F), /* superscript digits and letters */
 646   TA_UNIRANGE_REC(0x2C7D, 0x2C7D), /* modifier letter capital v */
 647   TA_UNIRANGE_REC(0xA770, 0xA770), /* modifier letter us */
 648   TA_UNIRANGE_REC(0xA7F8, 0xA7F9), /* more modifier letters */
 649   TA_UNIRANGE_REC(0xAB5C, 0xAB5F), /* more modifier letters */
 650   TA_UNIRANGE_REC(     0,      0)
 651 };
 652
 653 const TA_Script_UniRangeRec ta_latp_nonbase_uniranges[] =
 654 {
 655   TA_UNIRANGE_REC(0, 0)
 656 };
 657
 658
 659 const TA_Script_UniRangeRec ta_lisu_uniranges[] =
 660 {
 661   TA_UNIRANGE_REC(0xA4D0, 0xA4FF), /* Lisu */
 662   TA_UNIRANGE_REC(     0,      0)
 663 };
 664
 665 const TA_Script_UniRangeRec ta_lisu_nonbase_uniranges[] =
 666 {
 667   TA_UNIRANGE_REC(0, 0)
 668 };
 669
 670
 671 const TA_Script_UniRangeRec ta_mlym_uniranges[] =
 672 {
 673   TA_UNIRANGE_REC(0x0D00, 0x0D7F), /* Malayalam */
 674   TA_UNIRANGE_REC(     0,      0)
 675 };
 676
 677 const TA_Script_UniRangeRec ta_mlym_nonbase_uniranges[] =
 678 {
 679   TA_UNIRANGE_REC(0x0D00, 0x0D01),
 680   TA_UNIRANGE_REC(0x0D3B, 0x0D3C),
 681   TA_UNIRANGE_REC(0x0D4D, 0x0D4E),
 682   TA_UNIRANGE_REC(0x0D62, 0x0D63),
 683   TA_UNIRANGE_REC(     0,      0)
 684 };
 685
 686
 687 const TA_Script_UniRangeRec ta_mong_uniranges[] =
 688 {
 689   TA_UNIRANGE_REC( 0x1800,  0x18AF), /* Mongolian */
 690   TA_UNIRANGE_REC(0x11660, 0x1167F), /* Mongolian Supplement */
 691   TA_UNIRANGE_REC(      0,       0)
 692 };
 693
 694 const TA_Script_UniRangeRec ta_mong_nonbase_uniranges[] =
 695 {
 696   TA_UNIRANGE_REC(0x1885, 0x1886),
 697   TA_UNIRANGE_REC(0x18A9, 0x18A9),
 698   TA_UNIRANGE_REC(     0,      0)
 699 };
 700
 701
 702 const TA_Script_UniRangeRec ta_mymr_uniranges[] =
 703 {
 704   TA_UNIRANGE_REC(0x1000, 0x109F), /* Myanmar */
 705   TA_UNIRANGE_REC(0xA9E0, 0xA9FF), /* Myanmar Extended-B */
 706   TA_UNIRANGE_REC(0xAA60, 0xAA7F), /* Myanmar Extended-A */
 707   TA_UNIRANGE_REC(     0,      0)
 708 };
 709
 710 const TA_Script_UniRangeRec ta_mymr_nonbase_uniranges[] =
 711 {
 712   TA_UNIRANGE_REC(0x102D, 0x1030),
 713   TA_UNIRANGE_REC(0x1032, 0x1037),
 714   TA_UNIRANGE_REC(0x103A, 0x103A),
 715   TA_UNIRANGE_REC(0x103D, 0x103E),
 716   TA_UNIRANGE_REC(0x1058, 0x1059),
 717   TA_UNIRANGE_REC(0x105E, 0x1060),
 718   TA_UNIRANGE_REC(0x1071, 0x1074),
 719   TA_UNIRANGE_REC(0x1082, 0x1082),
 720   TA_UNIRANGE_REC(0x1085, 0x1086),
 721   TA_UNIRANGE_REC(0x108D, 0x108D),
 722   TA_UNIRANGE_REC(0xA9E5, 0xA9E5),
 723   TA_UNIRANGE_REC(0xAA7C, 0xAA7C),
 724   TA_UNIRANGE_REC(     0,      0)
 725 };
 726
 727
 728 const TA_Script_UniRangeRec ta_nkoo_uniranges[] =
 729 {
 730   TA_UNIRANGE_REC(0x07C0, 0x07FF), /* N'Ko */
 731   TA_UNIRANGE_REC(     0,      0)
 732 };
 733
 734 const TA_Script_UniRangeRec ta_nkoo_nonbase_uniranges[] =
 735 {
 736   TA_UNIRANGE_REC(0x07EB, 0x07F5),
 737   TA_UNIRANGE_REC(0x07FD, 0x07FD),
 738   TA_UNIRANGE_REC(     0,      0)
 739 };
 740
 741
 742 const TA_Script_UniRangeRec ta_olck_uniranges[] =
 743 {
 744   TA_UNIRANGE_REC(0x1C50, 0x1C7F), /* Ol Chiki */
 745   TA_UNIRANGE_REC(     0,      0)
 746 };
 747
 748 const TA_Script_UniRangeRec ta_olck_nonbase_uniranges[] =
 749 {
 750   TA_UNIRANGE_REC(0, 0)
 751 };
 752
 753
 754 const TA_Script_UniRangeRec ta_orkh_uniranges[] =
 755 {
 756   TA_UNIRANGE_REC(0x10C00, 0x10C4F), /* Old Turkic */
 757   TA_UNIRANGE_REC(      0,       0)
 758 };
 759
 760 const TA_Script_UniRangeRec ta_orkh_nonbase_uniranges[] =
 761 {
 762   TA_UNIRANGE_REC(0, 0)
 763 };
 764
 765
 766 const TA_Script_UniRangeRec ta_osge_uniranges[] =
 767 {
 768   TA_UNIRANGE_REC(0x104B0, 0x104FF), /* Osage */
 769   TA_UNIRANGE_REC(      0,       0)
 770 };
 771
 772 const TA_Script_UniRangeRec ta_osge_nonbase_uniranges[] =
 773 {
 774   TA_UNIRANGE_REC(0, 0)
 775 };
 776
 777
 778 const TA_Script_UniRangeRec ta_osma_uniranges[] =
 779 {
 780   TA_UNIRANGE_REC(0x10480, 0x104AF), /* Osmanya */
 781   TA_UNIRANGE_REC(      0,       0)
 782 };
 783
 784 const TA_Script_UniRangeRec ta_osma_nonbase_uniranges[] =
 785 {
 786   TA_UNIRANGE_REC(0, 0)
 787 };
 788
 789
 790 const TA_Script_UniRangeRec ta_saur_uniranges[] =
 791 {
 792   TA_UNIRANGE_REC(0xA880, 0xA8DF), /* Saurashtra */
 793   TA_UNIRANGE_REC(     0,      0)
 794 };
 795
 796 const TA_Script_UniRangeRec ta_saur_nonbase_uniranges[] =
 797 {
 798   TA_UNIRANGE_REC(0xA880, 0xA881),
 799   TA_UNIRANGE_REC(0xA8B4, 0xA8C5),
 800   TA_UNIRANGE_REC(     0,      0)
 801 };
 802
 803
 804 const TA_Script_UniRangeRec ta_shaw_uniranges[] =
 805 {
 806   TA_UNIRANGE_REC(0x10450, 0x1047F), /* Shavian */
 807   TA_UNIRANGE_REC(      0,       0)
 808 };
 809
 810 const TA_Script_UniRangeRec ta_shaw_nonbase_uniranges[] =
 811 {
 812   TA_UNIRANGE_REC(0, 0)
 813 };
 814
 815
 816 const TA_Script_UniRangeRec ta_sinh_uniranges[] =
 817 {
 818   TA_UNIRANGE_REC(0x0D80, 0x0DFF), /* Sinhala */
 819   TA_UNIRANGE_REC(     0,      0)
 820 };
 821
 822 const TA_Script_UniRangeRec ta_sinh_nonbase_uniranges[] =
 823 {
 824   TA_UNIRANGE_REC(0x0DCA, 0x0DCA),
 825   TA_UNIRANGE_REC(0x0DD2, 0x0DD6),
 826   TA_UNIRANGE_REC(     0,      0)
 827 };
 828
 829
 830 const TA_Script_UniRangeRec ta_sund_uniranges[] =
 831 {
 832   TA_UNIRANGE_REC(0x1B80, 0x1BBF), /* Sundanese */
 833   TA_UNIRANGE_REC(0x1CC0, 0x1CCF), /* Sundanese Supplement */
 834   TA_UNIRANGE_REC(     0,      0)
 835 };
 836
 837 const TA_Script_UniRangeRec ta_sund_nonbase_uniranges[] =
 838 {
 839   TA_UNIRANGE_REC(0x1B80, 0x1B82),
 840   TA_UNIRANGE_REC(0x1BA1, 0x1BAD),
 841   TA_UNIRANGE_REC(     0,      0)
 842 };
 843
 844
 845 const TA_Script_UniRangeRec ta_taml_uniranges[] =
 846 {
 847   TA_UNIRANGE_REC(0x0B80, 0x0BFF), /* Tamil */
 848   TA_UNIRANGE_REC(     0,      0)
 849 };
 850
 851 const TA_Script_UniRangeRec ta_taml_nonbase_uniranges[] =
 852 {
 853   TA_UNIRANGE_REC(0x0B82, 0x0B82),
 854   TA_UNIRANGE_REC(0x0BC0, 0x0BC2),
 855   TA_UNIRANGE_REC(0x0BCD, 0x0BCD),
 856   TA_UNIRANGE_REC(     0,      0)
 857 };
 858
 859
 860 const TA_Script_UniRangeRec ta_tavt_uniranges[] =
 861 {
 862   TA_UNIRANGE_REC(0xAA80, 0xAADF), /* Tai Viet */
 863   TA_UNIRANGE_REC(     0,      0)
 864 };
 865
 866 const TA_Script_UniRangeRec ta_tavt_nonbase_uniranges[] =
 867 {
 868   TA_UNIRANGE_REC(0xAAB0, 0xAAB0),
 869   TA_UNIRANGE_REC(0xAAB2, 0xAAB4),
 870   TA_UNIRANGE_REC(0xAAB7, 0xAAB8),
 871   TA_UNIRANGE_REC(0xAABE, 0xAABF),
 872   TA_UNIRANGE_REC(0xAAC1, 0xAAC1),
 873   TA_UNIRANGE_REC(     0,      0)
 874 };
 875
 876
 877 const TA_Script_UniRangeRec ta_telu_uniranges[] =
 878 {
 879   TA_UNIRANGE_REC(0x0C00, 0x0C7F), /* Telugu */
 880   TA_UNIRANGE_REC(     0,      0)
 881 };
 882
 883 const TA_Script_UniRangeRec ta_telu_nonbase_uniranges[] =
 884 {
 885   TA_UNIRANGE_REC(0x0C00, 0x0C00),
 886   TA_UNIRANGE_REC(0x0C04, 0x0C04),
 887   TA_UNIRANGE_REC(0x0C3E, 0x0C40),
 888   TA_UNIRANGE_REC(0x0C46, 0x0C56),
 889   TA_UNIRANGE_REC(0x0C62, 0x0C63),
 890   TA_UNIRANGE_REC(     0,      0)
 891 };
 892
 893
 894 const TA_Script_UniRangeRec ta_thai_uniranges[] =
 895 {
 896   TA_UNIRANGE_REC(0x0E00, 0x0E7F), /* Thai */
 897   TA_UNIRANGE_REC(     0,      0)
 898 };
 899
 900 const TA_Script_UniRangeRec ta_thai_nonbase_uniranges[] =
 901 {
 902   TA_UNIRANGE_REC(0x0E31, 0x0E31),
 903   TA_UNIRANGE_REC(0x0E34, 0x0E3A),
 904   TA_UNIRANGE_REC(0x0E47, 0x0E4E),
 905   TA_UNIRANGE_REC(     0,      0)
 906 };
 907
 908
 909 const TA_Script_UniRangeRec ta_tfng_uniranges[] =
 910 {
 911   TA_UNIRANGE_REC(0x2D30, 0x2D7F), /* Tifinagh */
 912   TA_UNIRANGE_REC(     0,      0)
 913 };
 914
 915 const TA_Script_UniRangeRec ta_tfng_nonbase_uniranges[] =
 916 {
 917   TA_UNIRANGE_REC(0, 0)
 918 };
 919
 920
 921 const TA_Script_UniRangeRec ta_vaii_uniranges[] =
 922 {
 923   TA_UNIRANGE_REC(0xA500, 0xA63F), /* Vai */
 924   TA_UNIRANGE_REC(     0,      0)
 925 };
 926
 927 const TA_Script_UniRangeRec ta_vaii_nonbase_uniranges[] =
 928 {
 929   TA_UNIRANGE_REC(0, 0)
 930 };
 931
 932
 933 const TA_Script_UniRangeRec ta_none_uniranges[] =
 934 {
 935   TA_UNIRANGE_REC(0, 0)
 936 };
 937
 938 const TA_Script_UniRangeRec ta_none_nonbase_uniranges[] =
 939 {
 940   TA_UNIRANGE_REC(0, 0)
 941 };
 942
 943 /* end of taranges.c */