dali/devel-api/text-abstraction/script.cpp

   1 /*
   2  * Copyright (c) 2021 Samsung Electronics Co., Ltd.
   3  *
   4  * Licensed under the Apache License, Version 2.0 (the "License");
   5  * you may not use this file except in compliance with the License.
   6  * You may obtain a copy of the License at
   7  *
   8  * http://www.apache.org/licenses/LICENSE-2.0
   9  *
  10  * Unless required by applicable law or agreed to in writing, software
  11  * distributed under the License is distributed on an "AS IS" BASIS,
  12  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13  * See the License for the specific language governing permissions and
  14  * limitations under the License.
  15  *
  16  */
  17
  18 // FILE HEADER
  19 #include <dali/devel-api/text-abstraction/script.h>
  20
  21 namespace Dali
  22 {
  23 namespace TextAbstraction
  24 {
  25 namespace
  26 {
  27 //TODO: Move the below defined characters to "defined-characters.h"
  28 constexpr unsigned int WHITE_SPACE_THRESHOLD = 0x21;   ///< All characters below 0x21 are considered white spaces.
  29 constexpr unsigned int CHAR_SPACE            = 0x20;   ///< Space.
  30 constexpr unsigned int CHAR_LF               = 0x000A; ///< NL Line feed, new line.
  31 constexpr unsigned int CHAR_VT               = 0x000B; ///< Vertical tab.
  32 constexpr unsigned int CHAR_FF               = 0x000C; ///< NP Form feed, new page.
  33 constexpr unsigned int CHAR_CR               = 0x000D; ///< Carriage return, new line.
  34 constexpr unsigned int CHAR_NEL              = 0x0085; ///< Next line.
  35 constexpr unsigned int CHAR_LS               = 0x2028; ///< Line separator.
  36 constexpr unsigned int CHAR_PS               = 0x2029; ///< Paragraph separator
  37
  38 constexpr unsigned int CHAR_ZWNJ = 0x200C; ///< Zero width non joiner.
  39 constexpr unsigned int CHAR_ZWJ  = 0x200D; ///< Zero width joiner.
  40 constexpr unsigned int CHAR_LTRM = 0x200E; ///< Left to Right Mark.
  41 constexpr unsigned int CHAR_RTLM = 0x200F; ///< Right to Left Mark.
  42 constexpr unsigned int CHAR_TS   = 0x2009; ///< Thin Space.
  43
  44 // Latin script:   It contains punctuation characters and symbols which are not part of the latin script. https://en.wikipedia.org/wiki/Latin_script_in_Unicode
  45 // 0x0000 - 0x007f C0 Controls and Basic Latin
  46 //
  47 //                 ASCII digits (not part of LATIN script):
  48 //                 0x0030 - 0x0039
  49 //
  50 //                 ASCII punctuation and symbols (not part of LATIN script):
  51 //                 0x0020 - 0x002F
  52 //                 0x003A - 0x0040
  53 //                 0x005B - 0x0060
  54 //                 0x007B - 0x007E
  55 //
  56 //                 Controls (not part of LATIN script):
  57 //                 0x007F
  58 //
  59 // 0x0080 - 0x00ff C1 Controls and Latin-1 Supplement
  60 //
  61 //                 Controls (not part of LATIN script):
  62 //                 0x0080 - 0x009F
  63 //
  64 //                 Punctuations and symbols (not part of LATIN script):
  65 //                 0x00A0 - 0x00BF
  66 //
  67 //                 Mathematical operators (not part of LATIN script):
  68 //                 0x00D7
  69 //                 0x00F7
  70 //
  71 // 0x0100 - 0x017f Latin Extended-A
  72 // 0x0180 - 0x024f Latin Extended-B
  73 // 0x0250 - 0x02af IPA Extensions
  74 // 0x02b0 - 0x02ff Spacing Modifier Letters
  75 //
  76 //                 Punctuation (not part of LATIN script):
  77 //                 0x02B9 - 0x02BF
  78 //
  79 // 0x1d00 - 0x1d7f Phonetic Extensions
  80 //
  81 //                 Uralic Phonetic (not part of LATIN script):
  82 //                 0x1D26 - 0x1D2B
  83 //
  84 //                 Subscripts and superscripts
  85 //                 0x1D5D - 0x1D61
  86 //                 0x1D66 - 0x1D6A
  87 //                 0x1D78
  88 //
  89 // 0x1d80 - 0x1dbf Phonetic Extensions Supplement
  90 //
  91 //                 0x1DBF (subscript or superscript. Not part of LATIN script )
  92 //
  93 // 0x1e00 - 0x1eff Latin Extended Additional
  94 // 0x2070 - 0x209f Superscripts and Subscripts
  95 //
  96 //                 0x2070          (not part of LATIN script)
  97 //                 0x2074 - 0x207E (not part of LATIN script)
  98 //
  99 // 0x2100 - 0x214f Letterlike symbols (not part of LATIN script)
 100 //
 101 //                 0x212A - 0x212B (are part of LATIN script)
 102 //                 0x2132          (are part of LATIN script)
 103 //                 0x214E          (are part of LATIN script)
 104 //
 105 // 0x2150 - 0x2189 Number Forms
 106 //
 107 //                 0x2150 - 0x215F Fractions (not part of LATIN script)
 108 //                 0x2189          Fractions (not part of LATIN script)
 109 //
 110 // 0x2c60 - 0x2c7f Latin Extended-C
 111 // 0xa720 - 0xa7ff Latin Extended-D
 112 //
 113 //                 0xA720 - 0xA721 Uralic Phonetic (not part of LATIN script)
 114 //                 0xA788          (not part of LATIN script)
 115 //                 0xA789 - 0xA78A Budu (not part of LATIN script)
 116 //
 117 // 0xab30 - 0xab6f Latin Extended-E
 118 //
 119 // 0xfb00 - 0xfb06 Latin Alphabetic Presentation Forms
 120 // 0xff00 - 0xffef Halfwidth and Fullwidth Forms
 121 //
 122 //                 0xFF00 - 0xFF20 HWFW Symbols (not part of LATIN script)
 123 //                 0xFF3B - 0xFF40 HWFW Symbols (not part of LATIN script)
 124 //                 0xFF5B - 0xFFEF HWFW Symbols (not part of LATIN script)
 125
 126 // Brahmic scripts:
 127 // 0x0900 - 0x097f Devanagari
 128 // 0x0980 - 0x09ff Bengali
 129 // 0x0a00 - 0x0a7f Gurmukhi
 130 // 0x0a80 - 0x0aff Gujarati
 131 // 0x0b00 - 0x0b7f Oriya
 132 // 0x0b80 - 0x0bff Tamil
 133 // 0x0c00 - 0x0c7f Telugu
 134 // 0x0c80 - 0x0cff Kannada
 135 // 0x0d00 - 0x0d7f Malayalam
 136
 137 // Sinhala script.
 138 // 0x0d80 - 0x0dff Sinhala
 139
 140 // Arabic script.
 141 // 0x0600 - 0x06ff Arabic
 142 // 0x0750 - 0x077f Arabic Supplement
 143 // 0x08A0 - 0x08ff Arabic Extended-A
 144 // 0xfb50 - 0xfdff Arabic Presentation Forms-A
 145 // 0xfe70 - 0xfeff Arabic Presentation Forms-B
 146 // 0x1ee00 - 0x1eeff Arabic Mathematical Alphabetic Symbols
 147
 148 // CJK (Chinese, Japanese and Korean) and Vietnamese script.
 149 // 0x2e80 - 0x2eff CJK Radicals Supplement
 150 // 0x2f00 - 0x2fdf Kangxi Radicals
 151 // 0x3000 - 0x303f CJK Symbols and Punctuation
 152 // 0x3200 - 0x32ff Enclosed CJK Letters and Months
 153 // 0x3400 - 0x4dbf CJK Unified Ideographs Extension A
 154 // 0x4e00 - 0x62ff CJK Unified Ideographs
 155 // 0x6300 - 0x77ff CJK Unified Ideographs
 156 // 0x7800 - 0x8cff CJK Unified Ideographs
 157 // 0x8d00 - 0x9fff CJK Unified Ideographs
 158 // 0x20000 - 0x215ff CJK Unified Ideographs Extension B
 159 // 0x21600 - 0x230ff CJK Unified Ideographs Extension B
 160 // 0x23100 - 0x245ff CJK Unified Ideographs Extension B
 161 // 0x24600 - 0x260ff CJK Unified Ideographs Extension B
 162 // 0x26100 - 0x275ff CJK Unified Ideographs Extension B
 163 // 0x27600 - 0x290ff CJK Unified Ideographs Extension B
 164 // 0x29100 - 0x2a6df CJK Unified Ideographs Extension B
 165 // 0x2a700 - 0x2b73f CJK Unified Ideographs Extension C
 166 // 0x2b740 - 0x2b81f CJK Unified Ideographs Extension D
 167
 168 // Japanese scripts.
 169 // 0x3040 - 0x309f Hiragana
 170 // 0x30a0 - 0x30ff Katakana
 171
 172 // Hangul script
 173 // 0x1100 - 0x11ff Hangul jamo
 174 // 0x3130 - 0x318f Hangul Compatibility Jamo
 175 // 0xa960 - 0xa97f Hangul Jamo Extended-A
 176 // 0xac00 - 0xd7af Hangul Syllables
 177 // 0xd7b0 - 0xd7ff Hangul Jamo Extended-B
 178
 179 // Bopomofo script
 180 // 0x3100 - 0x312f Bopomofo
 181 // 0x31a0 - 0x31bf Bopomofo Extended
 182
 183 // Khmer script
 184 // 0x1780 - 0x17ff Khmer
 185 // 0x19e0 - 0x19ff Khmer Symbols
 186
 187 // Lao script
 188 // 0x0e80 - 0x0eff Lao
 189
 190 // Thai script
 191 // 0x0e00 - 0x0e7f Thai
 192
 193 // Burmese script
 194 // 0x1000 - 0x109f Myanmar
 195
 196 // Hebrew script
 197 // 0x0591 - 0x05f4 Hebrew
 198 // 0xfb1d - 0xfb4f Hebrew subset of Alphabetic Presentation Forms
 199
 200 // Cyrillic script
 201 // 0x0400 - 0x04ff Cyrillic
 202 // 0x0500 - 0x052f Cyrillic suplement
 203 // 0x2de0 - 0x2dff Cyrillic Extended-A
 204 // 0xa640 - 0xa69f Cyrillic Extended-B
 205
 206 // Georgian script
 207 // 0x10a0 - 0x10ff Georgian
 208 // 0x2d00 - 0x2d2f Georgian suplement
 209
 210 // Greek script
 211 // 0x0370 - 0x03ff Greek & Coptic
 212 // 0x1f00 - 0x1fff Greek Extended
 213
 214 // Armenian script
 215 // 0x0530 - 0x058f Armenian
 216 // 0xfb13 - 0xfb17 Armenian subset of Alphabetic prefentation forms
 217
 218 // Javanese script
 219 // 0xa980 - 0xa9fd Javanese
 220
 221 // Sundanese script
 222 // 0x1b80 - 0x1bbf Sundanese
 223 // 0x1cc0 - 0x1ccf Sundanese supplement
 224
 225 // Ge'ez script (Ethiopic)
 226 // 0x1200 - 0x137f Ethiopic
 227 // 0x1380 - 0x139f Ethiopic supplement
 228 // 0x2d80 - 0x2ddf Ethiopic Extended
 229 // 0xab00 - 0xab2f Ethiopic Extended-A
 230
 231 // Baybayin Script
 232 // 0x1700 - 0x171f Baybayin
 233
 234 // Ol Chiki Script
 235 // 0x1c50 - 0x1c7f Ol Chiki
 236
 237 // Meitei Script
 238 // 0xabc0 - 0xabff Meetei Mayek
 239 // 0xaae0 - 0xaaff Meetei Mayek Extensions
 240
 241 // The Emoji which map to standardized Unicode characters
 242 // 1. Emoticons ( 1F601 - 1F64F )
 243 // 2. Dingbats ( 2700 - 27BF )
 244 // 3. Transport and map symbols ( 1F680 - 1F6C0 )
 245 // 4. Enclosed characters ( 24C2 - 1F251 )
 246 // 5. Uncategorized :-S
 247 // 6. Additional Emoticons ( 1F600 - 1F636 )
 248 // 6b. Additional transport and map symbols ( 1F680 - 1F6FF ): http://unicode.org/charts/PDF/U1F680.pdf
 249 // 6c. Other additional symbols ( 1F30D - 1F567 )
 250 // 7. Supplemental Symbols and Pictographs ( 1F900–1F9FF ): http://unicode.org/charts/PDF/U1F900.pdf
 251
 252 // Symbols. Work around for these symbols.
 253 // 0x25cb
 254 // 0x25cf
 255 // 0x25a1
 256 // 0x25a0
 257 // 0x2664
 258 // 0x2661
 259 // 0x2662
 260 // 0x2667
 261 // 0x2606
 262 // 0x25aa
 263 // 0x262a
 264
 265 /// character <= 0x077f
 266 inline Script GetScriptTillArabicSupplement(Character character)
 267 {
 268   Script script = UNKNOWN;
 269
 270   if((0x0030 <= character) && (character <= 0x0039))
 271   {
 272     script = ASCII_DIGITS;
 273   }
 274   else if(character <= 0x007E)
 275   {
 276     if((0x0020 <= character) && (character <= 0x002F))
 277     {
 278       script = ASCII_PS;
 279     }
 280     else if((0x003A <= character) && (character <= 0x0040))
 281     {
 282       script = ASCII_PS;
 283     }
 284     else if((0x005B <= character) && (character <= 0x0060))
 285     {
 286       script = ASCII_PS;
 287     }
 288     else if((0x007B <= character) && (character <= 0x007E))
 289     {
 290       script = ASCII_PS;
 291     }
 292     else
 293     {
 294       script = LATIN;
 295     }
 296   }
 297   else if((0x007F <= character) && (character <= 0x009F))
 298   {
 299     // 0x007F is actually part of C0 Controls and Basic Latin. However, is the last and only control character of its block
 300     // and the following characters of the next block are consecutive.
 301     script = C1_CONTROLS;
 302   }
 303   else if((0x00A0 <= character) && (character <= 0x00BF))
 304   {
 305     if(character == 0x00A9)
 306     {
 307       script = EMOJI; // 5. Uncategorized: copyright sign
 308     }
 309     else if(character == 0x00AE)
 310     {
 311       script = EMOJI; // 5. Uncategorized: registered sign
 312     }
 313     else
 314     {
 315       script = C1_PS;
 316     }
 317   }
 318   else if(character == 0x00D7)
 319   {
 320     script = C1_MATH;
 321   }
 322   else if(character == 0x00F7)
 323   {
 324     script = C1_MATH;
 325   }
 326   else if((0x00C0 <= character) && (character <= 0x02ff))
 327   {
 328     if((0x02B9 <= character) && (character <= 0x02BF))
 329     {
 330       script = SML_P;
 331     }
 332     else
 333     {
 334       script = LATIN;
 335     }
 336   }
 337   else if((0x0370 <= character) && (character <= 0x03ff))
 338   {
 339     script = GREEK;
 340   }
 341   else if((0x0400 <= character) && (character <= 0x04ff))
 342   {
 343     script = CYRILLIC;
 344   }
 345   else if((0x0500 <= character) && (character <= 0x052f))
 346   {
 347     script = CYRILLIC;
 348   }
 349   else if((0x0530 <= character) && (character <= 0x058f))
 350   {
 351     script = ARMENIAN;
 352   }
 353   else if((0x0591 <= character) && (character <= 0x05f4))
 354   {
 355     script = HEBREW;
 356   }
 357   else if((0x0600 <= character) && (character <= 0x06ff))
 358   {
 359     script = ARABIC;
 360   }
 361   else if((0x0750 <= character) && (character <= 0x077f))
 362   {
 363     script = ARABIC;
 364   }
 365
 366   return script;
 367 }
 368
 369 /// character <= 0x09ff
 370 inline Script GetScriptTillBengali(Character character)
 371 {
 372   Script script = UNKNOWN;
 373
 374   if(character <= 0x077f)
 375   {
 376     script = GetScriptTillArabicSupplement(character);
 377   }
 378   else // > 0x077f
 379   {
 380     if((0x08A0 <= character) && (character <= 0x08ff))
 381     {
 382       script = ARABIC;
 383     }
 384     else if((0x0900 <= character) && (character <= 0x097f))
 385     {
 386       script = DEVANAGARI;
 387     }
 388     else if((0x0980 <= character) && (character <= 0x09ff))
 389     {
 390       script = BENGALI;
 391     }
 392   }
 393
 394   return script;
 395 }
 396
 397 /// 0x09ff < character <= 0x0cff
 398 inline Script GetScriptBetweenBengaliAndKannada(Character character)
 399 {
 400   Script script = UNKNOWN;
 401
 402   if(character <= 0x0b7f)
 403   {
 404     if((0x0a00 <= character) && (character <= 0x0a7f))
 405     {
 406       script = GURMUKHI;
 407     }
 408     else if((0x0a80 <= character) && (character <= 0x0aff))
 409     {
 410       script = GUJARATI;
 411     }
 412     else if((0x0b00 <= character) && (character <= 0x0b7f))
 413     {
 414       script = ORIYA;
 415     }
 416   }
 417   else // > 0x0b7f
 418   {
 419     if((0x0b80 <= character) && (character <= 0x0bff))
 420     {
 421       script = TAMIL;
 422     }
 423     else if((0x0c00 <= character) && (character <= 0x0c7f))
 424     {
 425       script = TELUGU;
 426     }
 427     else if((0x0c80 <= character) && (character <= 0x0cff))
 428     {
 429       script = KANNADA;
 430     }
 431   }
 432
 433   return script;
 434 }
 435
 436 /// 0x0cff < character <= 0x1eff
 437 inline Script GetScriptBetweenKannadaAndLatinExtendedAdditional(Character character)
 438 {
 439   Script script = UNKNOWN;
 440
 441   if((0x0d00 <= character) && (character <= 0x0d7f))
 442   {
 443     script = MALAYALAM;
 444   }
 445   else if((0x0d80 <= character) && (character <= 0x0dff))
 446   {
 447     script = SINHALA;
 448   }
 449   else if((0x0e00 <= character) && (character <= 0x0e7f))
 450   {
 451     script = THAI;
 452   }
 453   else if((0x0e80 <= character) && (character <= 0x0eff))
 454   {
 455     script = LAO;
 456   }
 457   else if((0x1000 <= character) && (character <= 0x109f))
 458   {
 459     script = BURMESE;
 460   }
 461   else if((0x10a0 <= character) && (character <= 0x10ff))
 462   {
 463     script = GEORGIAN;
 464   }
 465   else if((0x1100 <= character) && (character <= 0x11ff))
 466   {
 467     script = HANGUL;
 468   }
 469   else if((0x1200 <= character) && (character <= 0x137f))
 470   {
 471     script = GEEZ;
 472   }
 473   else if((0x1380 <= character) && (character <= 0x139f))
 474   {
 475     script = GEEZ;
 476   }
 477   else if((0x1700 <= character) && (character <= 0x171f))
 478   {
 479     script = BAYBAYIN;
 480   }
 481   else if((0x1780 <= character) && (character <= 0x17ff))
 482   {
 483     script = KHMER;
 484   }
 485   else if((0x19e0 <= character) && (character <= 0x19ff))
 486   {
 487     script = KHMER;
 488   }
 489   else if((0x1b80 <= character) && (character <= 0x1bbf))
 490   {
 491     script = SUNDANESE;
 492   }
 493   else if((0x1c50 <= character) && (character <= 0x1c7f))
 494   {
 495     script = OL_CHIKI;
 496   }
 497   else if((0x1cc0 <= character) && (character <= 0x1ccf))
 498   {
 499     script = SUNDANESE;
 500   }
 501   else if((0x1d00 <= character) && (character <= 0x1eff))
 502   {
 503     if((0x1D26 <= character) && (character <= 0x1D2B))
 504     {
 505       script = PHONETIC_U;
 506     }
 507     else if((0x1D5D <= character) && (character <= 0x1D61))
 508     {
 509       script = PHONETIC_SS;
 510     }
 511     else if((0x1D66 <= character) && (character <= 0x1D6A))
 512     {
 513       script = PHONETIC_SS;
 514     }
 515     else if(character == 0x1D78)
 516     {
 517       script = PHONETIC_SS;
 518     }
 519     else if(character == 0x1DBF)
 520     {
 521       script = PHONETIC_SS;
 522     }
 523     else
 524     {
 525       script = LATIN;
 526     }
 527   }
 528
 529   return script;
 530 }
 531
 532 /// 0x1eff < character <= 0x2c7f
 533 inline Script GetScriptBetweenLatinExtendedAdditionalAndLatinExtendedC(Character character)
 534 {
 535   Script script = UNKNOWN;
 536
 537   if((0x1f00 <= character) && (character <= 0x1fff))
 538   {
 539     script = GREEK;
 540   }
 541   else if(character == 0x203c)
 542   {
 543     script = EMOJI; // 5. Uncategorized: double exclamation mark
 544   }
 545   else if(character == 0x2049)
 546   {
 547     script = EMOJI; // 5. Uncategorized: exclamation question mark
 548   }
 549   else if((0x2070 <= character) && (character <= 0x209f))
 550   {
 551     if(character == 0x2070)
 552     {
 553       script = NUMERIC_SS;
 554     }
 555     else if((0x2074 <= character) && (character <= 0x207E))
 556     {
 557       script = NUMERIC_SS;
 558     }
 559     else
 560     {
 561       script = LATIN;
 562     }
 563   }
 564   else if(character == 0x20e3)
 565   {
 566     script = EMOJI; // 5. Uncategorized: combining enclosing keycap
 567   }
 568   else if(character == 0x2122)
 569   {
 570     script = EMOJI; // 5. Uncategorized: trade mark sign
 571   }
 572   else if(character == 0x2139)
 573   {
 574     script = EMOJI; // 5. Uncategorized: information source
 575   }
 576   else if((0x2100 <= character) && (character <= 0x2189))
 577   {
 578     if((0x2100 <= character) && (character <= 0x214f))
 579     {
 580       if((0x212A <= character) && (character <= 0x212B))
 581       {
 582         script = LATIN;
 583       }
 584       else if(character == 0x2132)
 585       {
 586         script = LATIN;
 587       }
 588       else if(character == 0x214E)
 589       {
 590         script = LATIN;
 591       }
 592       else
 593       {
 594         script = LETTER_LIKE;
 595       }
 596     }
 597     else if((0x2150 <= character) && (character <= 0x215F))
 598     {
 599       script = FRACTIONS_NF;
 600     }
 601     else if(character == 0x2189)
 602     {
 603       script = FRACTIONS_NF;
 604     }
 605     else
 606     {
 607       script = LATIN;
 608     }
 609   }
 610   // Symbols
 611   else if((0x25cb == character) ||
 612           (0x25cf == character) ||
 613           (0x25a1 == character))
 614   {
 615     script = SYMBOLS1;
 616   }
 617   else if(0x25a0 == character)
 618   {
 619     script = SYMBOLS2;
 620   }
 621   else if((0x2664 == character) ||
 622           (0x2661 == character) ||
 623           (0x2662 == character) ||
 624           (0x2667 == character))
 625   {
 626     script = SYMBOLS3;
 627   }
 628   else if((0x2606 == character) ||
 629           (0x25aa == character))
 630   {
 631     script = SYMBOLS4;
 632   }
 633   else if(0x262a == character)
 634   {
 635     script = SYMBOLS5;
 636   }
 637   // U+2194 5. Uncategorized: left right arrow
 638   // U+2B55 5. Uncategorized: heavy large circle
 639   else if((0x2194 <= character) && (character <= 0x2B55))
 640   {
 641     script = EMOJI;
 642   }
 643   else if((0x2c60 <= character) && (character <= 0x2c7f))
 644   {
 645     script = LATIN;
 646   }
 647
 648   return script;
 649 }
 650
 651 /// 0x0cff < character <= 0x2c7f
 652 inline Script GetScriptBetweenKannadaAndLatinExtendedC(Character character)
 653 {
 654   Script script = UNKNOWN;
 655
 656   if(character <= 0x1eff)
 657   {
 658     script = GetScriptBetweenKannadaAndLatinExtendedAdditional(character);
 659   }
 660   else // > 0x1eff
 661   {
 662     script = GetScriptBetweenLatinExtendedAdditionalAndLatinExtendedC(character);
 663   }
 664
 665   return script;
 666 }
 667
 668 /// 0x2c7f < character <= 0xa7ff
 669 inline Script GetScriptBetweenLatinExtendedCAndLatinExtendedD(Character character)
 670 {
 671   Script script = UNKNOWN;
 672
 673   if((0x2d00 <= character) && (character <= 0x2d2f))
 674   {
 675     script = GEORGIAN;
 676   }
 677   else if((0x2d80 <= character) && (character <= 0x2ddf))
 678   {
 679     script = GEEZ;
 680   }
 681   else if((0x2de0 <= character) && (character <= 0x2dff))
 682   {
 683     script = CYRILLIC;
 684   }
 685   else if((0x2e80 <= character) && (character <= 0x2eff))
 686   {
 687     script = CJK;
 688   }
 689   else if((0x2f00 <= character) && (character <= 0x2fdf))
 690   {
 691     script = CJK;
 692   }
 693   else if((0x3000 <= character) && (character <= 0x303f))
 694   {
 695     script = CJK;
 696   }
 697   else if((0x3040 <= character) && (character <= 0x309f))
 698   {
 699     script = HIRAGANA;
 700   }
 701   else if((0x30a0 <= character) && (character <= 0x30ff))
 702   {
 703     script = KATAKANA;
 704   }
 705   else if((0x3100 <= character) && (character <= 0x312f))
 706   {
 707     script = BOPOMOFO;
 708   }
 709   else if((0x3130 <= character) && (character <= 0x318f))
 710   {
 711     script = HANGUL;
 712   }
 713   else if((0x31a0 <= character) && (character <= 0x31bf))
 714   {
 715     script = BOPOMOFO;
 716   }
 717   else if((0x3200 <= character) && (character <= 0x32ff))
 718   {
 719     script = CJK;
 720   }
 721   else if((0x3400 <= character) && (character <= 0x4dbf))
 722   {
 723     script = CJK;
 724   }
 725   else if((0x4e00 <= character) && (character <= 0x62ff))
 726   {
 727     script = CJK;
 728   }
 729   else if((0x6300 <= character) && (character <= 0x77ff))
 730   {
 731     script = CJK;
 732   }
 733   else if((0x7800 <= character) && (character <= 0x8cff))
 734   {
 735     script = CJK;
 736   }
 737   else if((0x8d00 <= character) && (character <= 0x9fff))
 738   {
 739     script = CJK;
 740   }
 741   else if((0xa640 <= character) && (character <= 0xa69f))
 742   {
 743     script = CYRILLIC;
 744   }
 745   else if((0xa720 <= character) && (character <= 0xa7ff))
 746   {
 747     if(character == 0xA720)
 748     {
 749       script = PHONETIC_U;
 750     }
 751     else if(character == 0xA721)
 752     {
 753       script = PHONETIC_U;
 754     }
 755     else if(character == 0xA788)
 756     {
 757       script = NON_LATIN_LED;
 758     }
 759     else if(character == 0xA789)
 760     {
 761       script = NON_LATIN_LED;
 762     }
 763     else if(character == 0xA78A)
 764     {
 765       script = NON_LATIN_LED;
 766     }
 767     else
 768     {
 769       script = LATIN;
 770     }
 771   }
 772
 773   return script;
 774 }
 775
 776 /// 0x2c7f < character <= 0xfdff
 777 inline Script GetScriptBetweenLatinExtendedCAndArabicPresentationFormsA(Character character)
 778 {
 779   Script script = GetScriptBetweenLatinExtendedCAndLatinExtendedD(character);
 780
 781   if((0xa960 <= character) && (character <= 0xa97f))
 782   {
 783     script = HANGUL;
 784   }
 785   else if((0xa980 <= character) && (character <= 0xa9fd))
 786   {
 787     script = JAVANESE;
 788   }
 789   else if((0xab00 <= character) && (character <= 0xab2f))
 790   {
 791     script = GEEZ;
 792   }
 793   else if((0xab30 <= character) && (character <= 0xab6f))
 794   {
 795     script = LATIN;
 796   }
 797   else if((0xaae0 <= character) && (character <= 0xaaff))
 798   {
 799     script = MEITEI;
 800   }
 801   else if((0xabc0 <= character) && (character <= 0xabff))
 802   {
 803     script = MEITEI;
 804   }
 805   else if((0xac00 <= character) && (character <= 0xd7af))
 806   {
 807     script = HANGUL;
 808   }
 809   else if((0xd7b0 <= character) && (character <= 0xd7ff))
 810   {
 811     script = HANGUL;
 812   }
 813   else if((0xfb00 <= character) && (character <= 0xfb06))
 814   {
 815     script = LATIN;
 816   }
 817   else if((0xfb13 <= character) && (character <= 0xfb17))
 818   {
 819     script = ARMENIAN;
 820   }
 821   else if((0xfb1d <= character) && (character <= 0xfb4f))
 822   {
 823     script = HEBREW;
 824   }
 825   else if((0xfb50 <= character) && (character <= 0xfdff))
 826   {
 827     script = ARABIC;
 828   }
 829
 830   return script;
 831 }
 832
 833 /// character > 0xfdff
 834 inline Script GetScriptAboveArabicPresentationFormsA(Character character)
 835 {
 836   Script script = UNKNOWN;
 837
 838   if((0xfe70 <= character) && (character <= 0xfeff))
 839   {
 840     script = ARABIC;
 841   }
 842   else if((0xff00 <= character) && (character <= 0xffef))
 843   {
 844     if((0xFF00 <= character) && (character <= 0xFF20))
 845     {
 846       script = HWFW_S;
 847     }
 848     else if((0xFF3B <= character) && (character <= 0xFF40))
 849     {
 850       script = HWFW_S;
 851     }
 852     else if((0xFF5B <= character) && (character <= 0xFFEF))
 853     {
 854       script = HWFW_S;
 855     }
 856     else
 857     {
 858       script = LATIN;
 859     }
 860   }
 861   else if((0x1ee00 <= character) && (character <= 0x1eeff))
 862   {
 863     script = ARABIC;
 864   }
 865   // U+1f170 4. Enclosed characters: negative squared latin capital letter A
 866   // U+1f6ff 6b. Additional transport and map symbols
 867   // Exclude U+1f170 ~ U+1f189. They are SYMBOLS_NSLCL (negative squared latin capital letter)
 868   else if((0x1f170 <= character) && (character <= 0x1f6ff))
 869   {
 870     script = EMOJI;
 871   }
 872   // 7. Supplemental Symbols and Pictographs
 873   else if((0x1f900 <= character) && (character <= 0x1f9ff))
 874   {
 875     script = EMOJI;
 876   }
 877   else if((0x20000 <= character) && (character <= 0x215ff))
 878   {
 879     script = CJK;
 880   }
 881   else if((0x21600 <= character) && (character <= 0x230ff))
 882   {
 883     script = CJK;
 884   }
 885   else if((0x23100 <= character) && (character <= 0x245ff))
 886   {
 887     script = CJK;
 888   }
 889   else if((0x24600 <= character) && (character <= 0x260ff))
 890   {
 891     script = CJK;
 892   }
 893   else if((0x26100 <= character) && (character <= 0x275ff))
 894   {
 895     script = CJK;
 896   }
 897   else if((0x27600 <= character) && (character <= 0x290ff))
 898   {
 899     script = CJK;
 900   }
 901   else if((0x29100 <= character) && (character <= 0x2a6df))
 902   {
 903     script = CJK;
 904   }
 905   else if((0x2a700 <= character) && (character <= 0x2b73f))
 906   {
 907     script = CJK;
 908   }
 909   else if((0x2b740 <= character) && (character <= 0x2b81f))
 910   {
 911     script = CJK;
 912   }
 913
 914   return script;
 915 }
 916
 917 /// character > 0x2c7f
 918 inline Script GetScriptAboveLatinExtendedC(Character character)
 919 {
 920   Script script = UNKNOWN;
 921
 922   if(character <= 0xfdff)
 923   {
 924     script = GetScriptBetweenLatinExtendedCAndArabicPresentationFormsA(character);
 925   }
 926   else // > 0xfdff
 927   {
 928     script = GetScriptAboveArabicPresentationFormsA(character);
 929   }
 930
 931   return script;
 932 }
 933
 934 } // namespace
 935
 936 bool IsRightToLeftScript(Script script)
 937 {
 938   return ((ARABIC == script) ||
 939           (HEBREW == script));
 940 }
 941
 942 Script GetCharacterScript(Character character)
 943 {
 944   Script script = UNKNOWN;
 945
 946   if(IsTextPresentationSelector(character))
 947   {
 948     script = EMOJI_TEXT;
 949   }
 950   else if(IsEmojiPresentationSelector(character))
 951   {
 952     script = EMOJI_COLOR;
 953   }
 954   else if(IsEmojiItem(character))
 955   {
 956     script = EMOJI;
 957   }
 958   else if(IsNegativeSquaredLatinCapitalLetter(character))
 959   {
 960     script = SYMBOLS_NSLCL;
 961   }
 962   else if(IsCommonScript(character))
 963   {
 964     script = COMMON;
 965   }
 966   else if(character <= 0x0cff)
 967   {
 968     if(character <= 0x09ff)
 969     {
 970       script = GetScriptTillBengali(character);
 971     }
 972     else // > 0x09ff
 973     {
 974       script = GetScriptBetweenBengaliAndKannada(character);
 975     }
 976   }
 977   else // > 0x0cff
 978   {
 979     if(character <= 0x2c7f)
 980     {
 981       script = GetScriptBetweenKannadaAndLatinExtendedC(character);
 982     }
 983     else // > 0x2c7f
 984     {
 985       script = GetScriptAboveLatinExtendedC(character);
 986     }
 987   }
 988
 989   return script;
 990 }
 991
 992 bool IsWhiteSpace(Character character)
 993 {
 994   return character < WHITE_SPACE_THRESHOLD;
 995 }
 996
 997 bool IsSpace(Character character)
 998 {
 999   return CHAR_SPACE == character;
1000 }
1001
1002 bool IsNewParagraph(Character character)
1003 {
1004   return ((CHAR_LF == character) ||
1005           (CHAR_VT == character) ||
1006           (CHAR_FF == character) ||
1007           (CHAR_CR == character) ||
1008           (CHAR_NEL == character) ||
1009           (CHAR_LS == character) ||
1010           (CHAR_PS == character));
1011 }
1012
1013 bool IsZeroWidthNonJoiner(Character character)
1014 {
1015   return CHAR_ZWNJ == character;
1016 }
1017
1018 bool IsZeroWidthJoiner(Character character)
1019 {
1020   return CHAR_ZWJ == character;
1021 }
1022
1023 bool IsZeroWidthSpace(Character character)
1024 {
1025   return CHAR_ZWS == character;
1026 }
1027
1028 bool IsLeftToRightMark(Character character)
1029 {
1030   return CHAR_LTRM == character;
1031 }
1032
1033 bool IsRightToLeftMark(Character character)
1034 {
1035   return CHAR_RTLM == character;
1036 }
1037
1038 bool IsThinSpace(Character character)
1039 {
1040   return CHAR_TS == character;
1041 }
1042
1043 bool IsCommonScript(Character character)
1044 {
1045   return (IsWhiteSpace(character) ||
1046           IsZeroWidthNonJoiner(character) ||
1047           IsZeroWidthJoiner(character) ||
1048           IsZeroWidthSpace(character) ||
1049           IsLeftToRightMark(character) ||
1050           IsRightToLeftMark(character) ||
1051           IsThinSpace(character) ||
1052           IsNewParagraph(character));
1053 }
1054
1055 bool HasLigatureMustBreak(Script script)
1056 {
1057   return ((LATIN == script) ||
1058           (ARABIC == script));
1059 }
1060
1061 Length GetNumberOfScripts()
1062 {
1063   return SYMBOLS_NSLCL + 1;
1064 }
1065
1066 } // namespace TextAbstraction
1067
1068 } // namespace Dali