i18nutil/source/utility/unicode.cxx

   1 /*************************************************************************
   2  *
   3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   4  *
   5  * Copyright 2008 by Sun Microsystems, Inc.
   6  *
   7  * OpenOffice.org - a multi-platform office productivity suite
   8  *
   9  * $RCSfile: unicode.cxx,v $
  10  * $Revision: 1.6 $
  11  *
  12  * This file is part of OpenOffice.org.
  13  *
  14  * OpenOffice.org is free software: you can redistribute it and/or modify
  15  * it under the terms of the GNU Lesser General Public License version 3
  16  * only, as published by the Free Software Foundation.
  17  *
  18  * OpenOffice.org is distributed in the hope that it will be useful,
  19  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  20  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  21  * GNU Lesser General Public License version 3 for more details
  22  * (a copy is included in the LICENSE file that accompanied this code).
  23  *
  24  * You should have received a copy of the GNU Lesser General Public License
  25  * version 3 along with OpenOffice.org.  If not, see
  26  * <http://www.openoffice.org/license.html>
  27  * for a copy of the LGPLv3 License.
  28  *
  29  ************************************************************************/
  30
  31 #include <com/sun/star/i18n/UnicodeType.hpp>
  32 #include <com/sun/star/i18n/KCharacterType.hpp>
  33 #include <i18nutil/unicode.hxx>
  34 #include "unicode_data.h"
  35
  36 // Workaround for glibc braindamage:
  37 // glibc 2.4's langinfo.h does "#define CURRENCY_SYMBOL __CURRENCY_SYMBOL"
  38 // which (obviously) breaks UnicodeType::CURRENCY_SYMBOL
  39 #undef CURRENCY_SYMBOL
  40
  41 using namespace ::com::sun::star::i18n;
  42
  43 static ScriptTypeList defaultTypeList[] = {
  44     { UnicodeScript_kBasicLatin,
  45       UnicodeScript_kBasicLatin,
  46       UnicodeScript_kBasicLatin },      // 0,
  47     { UnicodeScript_kLatin1Supplement,
  48       UnicodeScript_kLatin1Supplement,
  49       UnicodeScript_kLatin1Supplement },// 1,
  50     { UnicodeScript_kLatinExtendedA,
  51       UnicodeScript_kLatinExtendedA,
  52       UnicodeScript_kLatinExtendedA }, // 2,
  53     { UnicodeScript_kLatinExtendedB,
  54       UnicodeScript_kLatinExtendedB,
  55       UnicodeScript_kLatinExtendedB }, // 3,
  56     { UnicodeScript_kIPAExtension,
  57       UnicodeScript_kIPAExtension,
  58       UnicodeScript_kIPAExtension }, // 4,
  59     { UnicodeScript_kSpacingModifier,
  60       UnicodeScript_kSpacingModifier,
  61       UnicodeScript_kSpacingModifier }, // 5,
  62     { UnicodeScript_kCombiningDiacritical,
  63       UnicodeScript_kCombiningDiacritical,
  64       UnicodeScript_kCombiningDiacritical }, // 6,
  65     { UnicodeScript_kGreek,
  66       UnicodeScript_kGreek,
  67       UnicodeScript_kGreek }, // 7,
  68     { UnicodeScript_kCyrillic,
  69       UnicodeScript_kCyrillic,
  70       UnicodeScript_kCyrillic }, // 8,
  71     { UnicodeScript_kArmenian,
  72       UnicodeScript_kArmenian,
  73       UnicodeScript_kArmenian }, // 9,
  74     { UnicodeScript_kHebrew,
  75       UnicodeScript_kHebrew,
  76       UnicodeScript_kHebrew }, // 10,
  77     { UnicodeScript_kArabic,
  78       UnicodeScript_kArabic,
  79       UnicodeScript_kArabic }, // 11,
  80     { UnicodeScript_kSyriac,
  81       UnicodeScript_kSyriac,
  82       UnicodeScript_kSyriac }, // 12,
  83     { UnicodeScript_kThaana,
  84       UnicodeScript_kThaana,
  85       UnicodeScript_kThaana }, // 13,
  86     { UnicodeScript_kDevanagari,
  87       UnicodeScript_kDevanagari,
  88       UnicodeScript_kDevanagari }, // 14,
  89     { UnicodeScript_kBengali,
  90       UnicodeScript_kBengali,
  91       UnicodeScript_kBengali }, // 15,
  92     { UnicodeScript_kGurmukhi,
  93       UnicodeScript_kGurmukhi,
  94       UnicodeScript_kGurmukhi }, // 16,
  95     { UnicodeScript_kGujarati,
  96       UnicodeScript_kGujarati,
  97       UnicodeScript_kGujarati }, // 17,
  98     { UnicodeScript_kOriya,
  99       UnicodeScript_kOriya,
 100       UnicodeScript_kOriya }, // 18,
 101     { UnicodeScript_kTamil,
 102       UnicodeScript_kTamil,
 103       UnicodeScript_kTamil }, // 19,
 104     { UnicodeScript_kTelugu,
 105       UnicodeScript_kTelugu,
 106       UnicodeScript_kTelugu }, // 20,
 107     { UnicodeScript_kKannada,
 108       UnicodeScript_kKannada,
 109       UnicodeScript_kKannada }, // 21,
 110     { UnicodeScript_kMalayalam,
 111       UnicodeScript_kMalayalam,
 112       UnicodeScript_kMalayalam }, // 22,
 113     { UnicodeScript_kSinhala,
 114       UnicodeScript_kSinhala,
 115       UnicodeScript_kSinhala }, // 23,
 116     { UnicodeScript_kThai,
 117       UnicodeScript_kThai,
 118       UnicodeScript_kThai }, // 24,
 119     { UnicodeScript_kLao,
 120       UnicodeScript_kLao,
 121       UnicodeScript_kLao }, // 25,
 122     { UnicodeScript_kTibetan,
 123       UnicodeScript_kTibetan,
 124       UnicodeScript_kTibetan }, // 26,
 125     { UnicodeScript_kMyanmar,
 126       UnicodeScript_kMyanmar,
 127       UnicodeScript_kMyanmar }, // 27,
 128     { UnicodeScript_kGeorgian,
 129       UnicodeScript_kGeorgian,
 130       UnicodeScript_kGeorgian }, // 28,
 131     { UnicodeScript_kHangulJamo,
 132       UnicodeScript_kHangulJamo,
 133       UnicodeScript_kHangulJamo }, // 29,
 134     { UnicodeScript_kEthiopic,
 135       UnicodeScript_kEthiopic,
 136       UnicodeScript_kEthiopic }, // 30,
 137     { UnicodeScript_kCherokee,
 138       UnicodeScript_kCherokee,
 139       UnicodeScript_kCherokee }, // 31,
 140     { UnicodeScript_kUnifiedCanadianAboriginalSyllabics,
 141       UnicodeScript_kUnifiedCanadianAboriginalSyllabics,
 142       UnicodeScript_kUnifiedCanadianAboriginalSyllabics }, // 32,
 143     { UnicodeScript_kOgham,
 144       UnicodeScript_kOgham,
 145       UnicodeScript_kOgham }, // 33,
 146     { UnicodeScript_kRunic,
 147       UnicodeScript_kRunic,
 148       UnicodeScript_kRunic }, // 34,
 149     { UnicodeScript_kKhmer,
 150       UnicodeScript_kKhmer,
 151       UnicodeScript_kKhmer }, // 35,
 152     { UnicodeScript_kMongolian,
 153       UnicodeScript_kMongolian,
 154       UnicodeScript_kMongolian }, // 36,
 155     { UnicodeScript_kLatinExtendedAdditional,
 156       UnicodeScript_kLatinExtendedAdditional,
 157       UnicodeScript_kLatinExtendedAdditional }, // 37,
 158     { UnicodeScript_kGreekExtended,
 159       UnicodeScript_kGreekExtended,
 160       UnicodeScript_kGreekExtended }, // 38,
 161     { UnicodeScript_kGeneralPunctuation,
 162       UnicodeScript_kGeneralPunctuation,
 163       UnicodeScript_kGeneralPunctuation }, // 39,
 164     { UnicodeScript_kSuperSubScript,
 165       UnicodeScript_kSuperSubScript,
 166       UnicodeScript_kSuperSubScript }, // 40,
 167     { UnicodeScript_kCurrencySymbolScript,
 168       UnicodeScript_kCurrencySymbolScript,
 169       UnicodeScript_kCurrencySymbolScript }, // 41,
 170     { UnicodeScript_kSymbolCombiningMark,
 171       UnicodeScript_kSymbolCombiningMark,
 172       UnicodeScript_kSymbolCombiningMark }, // 42,
 173     { UnicodeScript_kLetterlikeSymbol,
 174       UnicodeScript_kLetterlikeSymbol,
 175       UnicodeScript_kLetterlikeSymbol }, // 43,
 176     { UnicodeScript_kNumberForm,
 177       UnicodeScript_kNumberForm,
 178       UnicodeScript_kNumberForm }, // 44,
 179     { UnicodeScript_kArrow,
 180       UnicodeScript_kArrow,
 181       UnicodeScript_kArrow }, // 45,
 182     { UnicodeScript_kMathOperator,
 183       UnicodeScript_kMathOperator,
 184       UnicodeScript_kMathOperator }, // 46,
 185     { UnicodeScript_kMiscTechnical,
 186       UnicodeScript_kMiscTechnical,
 187       UnicodeScript_kMiscTechnical }, // 47,
 188     { UnicodeScript_kControlPicture,
 189       UnicodeScript_kControlPicture,
 190       UnicodeScript_kControlPicture }, // 48,
 191     { UnicodeScript_kOpticalCharacter,
 192       UnicodeScript_kOpticalCharacter,
 193       UnicodeScript_kOpticalCharacter }, // 49,
 194     { UnicodeScript_kEnclosedAlphanumeric,
 195       UnicodeScript_kEnclosedAlphanumeric,
 196       UnicodeScript_kEnclosedAlphanumeric }, // 50,
 197     { UnicodeScript_kBoxDrawing,
 198       UnicodeScript_kBoxDrawing,
 199       UnicodeScript_kBoxDrawing }, // 51,
 200     { UnicodeScript_kBlockElement,
 201       UnicodeScript_kBlockElement,
 202       UnicodeScript_kBlockElement }, // 52,
 203     { UnicodeScript_kGeometricShape,
 204       UnicodeScript_kGeometricShape,
 205       UnicodeScript_kGeometricShape }, // 53,
 206     { UnicodeScript_kMiscSymbol,
 207       UnicodeScript_kMiscSymbol,
 208       UnicodeScript_kMiscSymbol }, // 54,
 209     { UnicodeScript_kDingbat,
 210       UnicodeScript_kDingbat,
 211       UnicodeScript_kDingbat }, // 55,
 212     { UnicodeScript_kBraillePatterns,
 213       UnicodeScript_kBraillePatterns,
 214       UnicodeScript_kBraillePatterns }, // 56,
 215     { UnicodeScript_kCJKRadicalsSupplement,
 216       UnicodeScript_kCJKRadicalsSupplement,
 217       UnicodeScript_kCJKRadicalsSupplement }, // 57,
 218     { UnicodeScript_kKangxiRadicals,
 219       UnicodeScript_kKangxiRadicals,
 220       UnicodeScript_kKangxiRadicals }, // 58,
 221     { UnicodeScript_kIdeographicDescriptionCharacters,
 222       UnicodeScript_kIdeographicDescriptionCharacters,
 223       UnicodeScript_kIdeographicDescriptionCharacters }, // 59,
 224     { UnicodeScript_kCJKSymbolPunctuation,
 225       UnicodeScript_kCJKSymbolPunctuation,
 226       UnicodeScript_kCJKSymbolPunctuation }, // 60,
 227     { UnicodeScript_kHiragana,
 228       UnicodeScript_kHiragana,
 229       UnicodeScript_kHiragana }, // 61,
 230     { UnicodeScript_kKatakana,
 231       UnicodeScript_kKatakana,
 232       UnicodeScript_kKatakana }, // 62,
 233     { UnicodeScript_kBopomofo,
 234       UnicodeScript_kBopomofo,
 235       UnicodeScript_kBopomofo }, // 63,
 236     { UnicodeScript_kHangulCompatibilityJamo,
 237       UnicodeScript_kHangulCompatibilityJamo,
 238       UnicodeScript_kHangulCompatibilityJamo }, // 64,
 239     { UnicodeScript_kKanbun,
 240       UnicodeScript_kKanbun,
 241       UnicodeScript_kKanbun }, // 65,
 242     { UnicodeScript_kBopomofoExtended,
 243       UnicodeScript_kBopomofoExtended,
 244       UnicodeScript_kBopomofoExtended }, // 66,
 245     { UnicodeScript_kEnclosedCJKLetterMonth,
 246       UnicodeScript_kEnclosedCJKLetterMonth,
 247       UnicodeScript_kEnclosedCJKLetterMonth }, // 67,
 248     { UnicodeScript_kCJKCompatibility,
 249       UnicodeScript_kCJKCompatibility,
 250       UnicodeScript_kCJKCompatibility }, // 68,
 251     { UnicodeScript_k_CJKUnifiedIdeographsExtensionA,
 252       UnicodeScript_k_CJKUnifiedIdeographsExtensionA,
 253       UnicodeScript_k_CJKUnifiedIdeographsExtensionA }, // 69,
 254     { UnicodeScript_kCJKUnifiedIdeograph,
 255       UnicodeScript_kCJKUnifiedIdeograph,
 256       UnicodeScript_kCJKUnifiedIdeograph }, // 70,
 257     { UnicodeScript_kYiSyllables,
 258       UnicodeScript_kYiSyllables,
 259       UnicodeScript_kYiSyllables }, // 71,
 260     { UnicodeScript_kYiRadicals,
 261       UnicodeScript_kYiRadicals,
 262       UnicodeScript_kYiRadicals }, // 72,
 263     { UnicodeScript_kHangulSyllable,
 264       UnicodeScript_kHangulSyllable,
 265       UnicodeScript_kHangulSyllable }, // 73,
 266     { UnicodeScript_kHighSurrogate,
 267       UnicodeScript_kHighSurrogate,
 268       UnicodeScript_kHighSurrogate }, // 74,
 269     { UnicodeScript_kHighPrivateUseSurrogate,
 270       UnicodeScript_kHighPrivateUseSurrogate,
 271       UnicodeScript_kHighPrivateUseSurrogate }, // 75,
 272     { UnicodeScript_kLowSurrogate,
 273       UnicodeScript_kLowSurrogate,
 274       UnicodeScript_kLowSurrogate }, // 76,
 275     { UnicodeScript_kPrivateUse,
 276       UnicodeScript_kPrivateUse,
 277       UnicodeScript_kPrivateUse }, // 77,
 278     { UnicodeScript_kCJKCompatibilityIdeograph,
 279       UnicodeScript_kCJKCompatibilityIdeograph,
 280       UnicodeScript_kCJKCompatibilityIdeograph }, // 78,
 281     { UnicodeScript_kAlphabeticPresentation,
 282       UnicodeScript_kAlphabeticPresentation,
 283       UnicodeScript_kAlphabeticPresentation }, // 79,
 284     { UnicodeScript_kArabicPresentationA,
 285       UnicodeScript_kArabicPresentationA,
 286       UnicodeScript_kArabicPresentationA }, // 80,
 287     { UnicodeScript_kCombiningHalfMark,
 288       UnicodeScript_kCombiningHalfMark,
 289       UnicodeScript_kCombiningHalfMark }, // 81,
 290     { UnicodeScript_kCJKCompatibilityForm,
 291       UnicodeScript_kCJKCompatibilityForm,
 292       UnicodeScript_kCJKCompatibilityForm }, // 82,
 293     { UnicodeScript_kSmallFormVariant,
 294       UnicodeScript_kSmallFormVariant,
 295       UnicodeScript_kSmallFormVariant }, // 83,
 296     { UnicodeScript_kArabicPresentationB,
 297       UnicodeScript_kArabicPresentationB,
 298       UnicodeScript_kArabicPresentationB }, // 84,
 299     { UnicodeScript_kNoScript,
 300       UnicodeScript_kNoScript,
 301       UnicodeScript_kNoScript }, // 85,
 302     { UnicodeScript_kHalfwidthFullwidthForm,
 303       UnicodeScript_kHalfwidthFullwidthForm,
 304       UnicodeScript_kHalfwidthFullwidthForm }, // 86,
 305     { UnicodeScript_kScriptCount,
 306       UnicodeScript_kScriptCount,
 307       UnicodeScript_kNoScript } // 87,
 308 };
 309
 310 sal_Int16 SAL_CALL
 311 unicode::getUnicodeScriptType( const sal_Unicode ch, ScriptTypeList* typeList, sal_Int16 unknownType ) {
 312
 313     if (!typeList) {
 314         typeList = defaultTypeList;
 315         unknownType = UnicodeScript_kNoScript;
 316     }
 317
 318     sal_Int16 i = 0, type = typeList[0].to;
 319     while (type < UnicodeScript_kScriptCount && ch > UnicodeScriptType[type][UnicodeScriptTypeTo]) {
 320         type = typeList[++i].to;
 321     }
 322
 323     return (type < UnicodeScript_kScriptCount &&
 324             ch >= UnicodeScriptType[typeList[i].from][UnicodeScriptTypeFrom]) ?
 325             typeList[i].value : unknownType;
 326 }
 327
 328 sal_Bool SAL_CALL
 329 unicode::isUnicodeScriptType( const sal_Unicode ch, sal_Int16 type) {
 330     return ch >= UnicodeScriptType[type][UnicodeScriptTypeFrom] &&
 331         ch <= UnicodeScriptType[type][UnicodeScriptTypeTo];
 332 }
 333
 334 sal_Unicode SAL_CALL
 335 unicode::getUnicodeScriptStart( UnicodeScript type) {
 336     return UnicodeScriptType[type][UnicodeScriptTypeFrom];
 337 }
 338
 339 sal_Unicode SAL_CALL
 340 unicode::getUnicodeScriptEnd( UnicodeScript type) {
 341     return UnicodeScriptType[type][UnicodeScriptTypeTo];
 342 }
 343
 344 sal_Int16 SAL_CALL
 345 unicode::getUnicodeType( const sal_Unicode ch ) {
 346     static sal_Unicode c = 0x00;
 347     static sal_Int16 r = 0x00;
 348
 349     if (ch == c) return r;
 350     else c = ch;
 351
 352     sal_Int16 address = UnicodeTypeIndex[ch >> 8];
 353     return r = (sal_Int16)((address < UnicodeTypeNumberBlock) ? UnicodeTypeBlockValue[address] :
 354         UnicodeTypeValue[((address - UnicodeTypeNumberBlock) << 8) + (ch & 0xff)]);
 355 }
 356
 357 sal_uInt8 SAL_CALL
 358 unicode::getUnicodeDirection( const sal_Unicode ch ) {
 359     static sal_Unicode c = 0x00;
 360     static sal_uInt8 r = 0x00;
 361
 362     if (ch == c) return r;
 363     else c = ch;
 364
 365     sal_Int16 address = UnicodeDirectionIndex[ch >> 8];
 366     return r = ((address < UnicodeDirectionNumberBlock) ? UnicodeDirectionBlockValue[address] :
 367         UnicodeDirectionValue[((address - UnicodeDirectionNumberBlock) << 8) + (ch & 0xff)]);
 368
 369 }
 370
 371 #define bit(name)   (1 << name)
 372
 373 #define UPPERMASK   bit(UnicodeType::UPPERCASE_LETTER)
 374
 375 #define LOWERMASK   bit(UnicodeType::LOWERCASE_LETTER)
 376
 377 #define TITLEMASK   bit(UnicodeType::TITLECASE_LETTER)
 378
 379 #define DIGITMASK   bit(UnicodeType::DECIMAL_DIGIT_NUMBER)|\
 380             bit(UnicodeType::LETTER_NUMBER)|\
 381             bit(UnicodeType::OTHER_NUMBER)
 382
 383 #define ALPHAMASK   UPPERMASK|LOWERMASK|TITLEMASK|\
 384             bit(UnicodeType::MODIFIER_LETTER)|\
 385             bit(UnicodeType::OTHER_LETTER)
 386
 387 #define BASEMASK    DIGITMASK|ALPHAMASK|\
 388             bit(UnicodeType::NON_SPACING_MARK)|\
 389             bit(UnicodeType::ENCLOSING_MARK)|\
 390             bit(UnicodeType::COMBINING_SPACING_MARK)
 391
 392 #define SPACEMASK   bit(UnicodeType::SPACE_SEPARATOR)|\
 393             bit(UnicodeType::LINE_SEPARATOR)|\
 394             bit(UnicodeType::PARAGRAPH_SEPARATOR)
 395
 396 #define PUNCTUATIONMASK bit(UnicodeType::DASH_PUNCTUATION)|\
 397             bit(UnicodeType::INITIAL_PUNCTUATION)|\
 398             bit(UnicodeType::FINAL_PUNCTUATION)|\
 399             bit(UnicodeType::CONNECTOR_PUNCTUATION)|\
 400             bit(UnicodeType::OTHER_PUNCTUATION)
 401
 402 #define SYMBOLMASK  bit(UnicodeType::MATH_SYMBOL)|\
 403             bit(UnicodeType::CURRENCY_SYMBOL)|\
 404             bit(UnicodeType::MODIFIER_SYMBOL)|\
 405             bit(UnicodeType::OTHER_SYMBOL)
 406
 407 #define PRINTMASK   BASEMASK|SPACEMASK|PUNCTUATIONMASK|SYMBOLMASK
 408
 409 #define CONTROLMASK bit(UnicodeType::CONTROL)|\
 410             bit(UnicodeType::FORMAT)|\
 411             bit(UnicodeType::LINE_SEPARATOR)|\
 412             bit(UnicodeType::PARAGRAPH_SEPARATOR)
 413
 414 #define IsType(func, mask)  \
 415 sal_Bool SAL_CALL func( const sal_Unicode ch) {\
 416     return (bit(getUnicodeType(ch)) & (mask)) != 0;\
 417 }
 418
 419 IsType(unicode::isUpper, UPPERMASK)
 420 IsType(unicode::isLower, LOWERMASK)
 421 IsType(unicode::isTitle, DIGITMASK)
 422 IsType(unicode::isControl, CONTROLMASK)
 423 IsType(unicode::isPrint, PRINTMASK)
 424 IsType(unicode::isAlpha, ALPHAMASK)
 425 IsType(unicode::isDigit, DIGITMASK)
 426 IsType(unicode::isAlphaDigit, ALPHAMASK|DIGITMASK)
 427 IsType(unicode::isSpace, SPACEMASK)
 428 IsType(unicode::isBase, BASEMASK)
 429 IsType(unicode::isPunctuation, PUNCTUATIONMASK)
 430
 431 #define CONTROLSPACE    bit(0x09)|bit(0x0a)|bit(0x0b)|bit(0x0c)|bit(0x0d)|\
 432             bit(0x1c)|bit(0x1d)|bit(0x1e)|bit(0x1f)
 433
 434 sal_Bool SAL_CALL unicode::isWhiteSpace( const sal_Unicode ch) {
 435     return (ch != 0xa0 && isSpace(ch)) || (ch <= 0x1F && (bit(ch) & (CONTROLSPACE)));
 436 }
 437
 438 sal_Int32 SAL_CALL unicode::getCharType( const sal_Unicode ch )
 439 {
 440     using namespace ::com::sun::star::i18n::KCharacterType;
 441
 442     switch ( getUnicodeType( ch ) ) {
 443     // Upper
 444     case UnicodeType::UPPERCASE_LETTER :
 445         return UPPER|LETTER|PRINTABLE|BASE_FORM;
 446
 447     // Lower
 448     case UnicodeType::LOWERCASE_LETTER :
 449         return LOWER|LETTER|PRINTABLE|BASE_FORM;
 450
 451     // Title
 452     case UnicodeType::TITLECASE_LETTER :
 453         return TITLE_CASE|LETTER|PRINTABLE|BASE_FORM;
 454
 455     // Letter
 456     case UnicodeType::MODIFIER_LETTER :
 457     case UnicodeType::OTHER_LETTER :
 458         return LETTER|PRINTABLE|BASE_FORM;
 459
 460     // Digit
 461     case UnicodeType::DECIMAL_DIGIT_NUMBER:
 462     case UnicodeType::LETTER_NUMBER:
 463     case UnicodeType::OTHER_NUMBER:
 464         return DIGIT|PRINTABLE|BASE_FORM;
 465
 466     // Base
 467     case UnicodeType::NON_SPACING_MARK:
 468     case UnicodeType::ENCLOSING_MARK:
 469     case UnicodeType::COMBINING_SPACING_MARK:
 470         return BASE_FORM|PRINTABLE;
 471
 472     // Print
 473     case UnicodeType::SPACE_SEPARATOR:
 474
 475     case UnicodeType::DASH_PUNCTUATION:
 476     case UnicodeType::INITIAL_PUNCTUATION:
 477     case UnicodeType::FINAL_PUNCTUATION:
 478     case UnicodeType::CONNECTOR_PUNCTUATION:
 479     case UnicodeType::OTHER_PUNCTUATION:
 480
 481     case UnicodeType::MATH_SYMBOL:
 482     case UnicodeType::CURRENCY_SYMBOL:
 483     case UnicodeType::MODIFIER_SYMBOL:
 484     case UnicodeType::OTHER_SYMBOL:
 485         return PRINTABLE;
 486
 487     // Control
 488     case UnicodeType::CONTROL:
 489     case UnicodeType::FORMAT:
 490         return CONTROL;
 491
 492     case UnicodeType::LINE_SEPARATOR:
 493     case UnicodeType::PARAGRAPH_SEPARATOR:
 494         return CONTROL|PRINTABLE;
 495
 496     // for all others
 497     default:
 498         return 0;
 499     }
 500 }
 501
 502