1 // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: 2 // 3 // ucd-generate word-break ucd-15.0.0 --chars 4 // 5 // Unicode version: 15.0.0. 6 // 7 // ucd-generate 0.2.14 is available on crates.io. 8 9 pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ 10 ("ALetter", ALETTER), 11 ("CR", CR), 12 ("Double_Quote", DOUBLE_QUOTE), 13 ("Extend", EXTEND), 14 ("ExtendNumLet", EXTENDNUMLET), 15 ("Format", FORMAT), 16 ("Hebrew_Letter", HEBREW_LETTER), 17 ("Katakana", KATAKANA), 18 ("LF", LF), 19 ("MidLetter", MIDLETTER), 20 ("MidNum", MIDNUM), 21 ("MidNumLet", MIDNUMLET), 22 ("Newline", NEWLINE), 23 ("Numeric", NUMERIC), 24 ("Regional_Indicator", REGIONAL_INDICATOR), 25 ("Single_Quote", SINGLE_QUOTE), 26 ("WSegSpace", WSEGSPACE), 27 ("ZWJ", ZWJ), 28 ]; 29 30 pub const ALETTER: &'static [(char, char)] = &[ 31 ('A', 'Z'), 32 ('a', 'z'), 33 ('ª', 'ª'), 34 ('µ', 'µ'), 35 ('º', 'º'), 36 ('À', 'Ö'), 37 ('Ø', 'ö'), 38 ('ø', '˗'), 39 ('˞', '˿'), 40 ('Ͱ', 'ʹ'), 41 ('Ͷ', 'ͷ'), 42 ('ͺ', 'ͽ'), 43 ('Ϳ', 'Ϳ'), 44 ('Ά', 'Ά'), 45 ('Έ', 'Ί'), 46 ('Ό', 'Ό'), 47 ('Ύ', 'Ρ'), 48 ('Σ', 'ϵ'), 49 ('Ϸ', 'ҁ'), 50 ('Ҋ', 'ԯ'), 51 ('Ա', 'Ֆ'), 52 ('ՙ', '՜'), 53 ('՞', '՞'), 54 ('ՠ', 'ֈ'), 55 ('֊', '֊'), 56 ('׳', '׳'), 57 ('ؠ', 'ي'), 58 ('ٮ', 'ٯ'), 59 ('ٱ', 'ۓ'), 60 ('ە', 'ە'), 61 ('ۥ', 'ۦ'), 62 ('ۮ', 'ۯ'), 63 ('ۺ', 'ۼ'), 64 ('ۿ', 'ۿ'), 65 ('ܐ', 'ܐ'), 66 ('ܒ', 'ܯ'), 67 ('ݍ', 'ޥ'), 68 ('ޱ', 'ޱ'), 69 ('ߊ', 'ߪ'), 70 ('ߴ', 'ߵ'), 71 ('ߺ', 'ߺ'), 72 ('ࠀ', 'ࠕ'), 73 ('ࠚ', 'ࠚ'), 74 ('ࠤ', 'ࠤ'), 75 ('ࠨ', 'ࠨ'), 76 ('ࡀ', 'ࡘ'), 77 ('ࡠ', 'ࡪ'), 78 ('ࡰ', 'ࢇ'), 79 ('ࢉ', 'ࢎ'), 80 ('ࢠ', 'ࣉ'), 81 ('ऄ', 'ह'), 82 ('ऽ', 'ऽ'), 83 ('ॐ', 'ॐ'), 84 ('क़', 'ॡ'), 85 ('ॱ', 'ঀ'), 86 ('অ', 'ঌ'), 87 ('এ', 'ঐ'), 88 ('ও', 'ন'), 89 ('প', 'র'), 90 ('ল', 'ল'), 91 ('শ', 'হ'), 92 ('ঽ', 'ঽ'), 93 ('ৎ', 'ৎ'), 94 ('ড়', 'ঢ়'), 95 ('য়', 'ৡ'), 96 ('ৰ', 'ৱ'), 97 ('ৼ', 'ৼ'), 98 ('ਅ', 'ਊ'), 99 ('ਏ', 'ਐ'), 100 ('ਓ', 'ਨ'), 101 ('ਪ', 'ਰ'), 102 ('ਲ', 'ਲ਼'), 103 ('ਵ', 'ਸ਼'), 104 ('ਸ', 'ਹ'), 105 ('ਖ਼', 'ੜ'), 106 ('ਫ਼', 'ਫ਼'), 107 ('ੲ', 'ੴ'), 108 ('અ', 'ઍ'), 109 ('એ', 'ઑ'), 110 ('ઓ', 'ન'), 111 ('પ', 'ર'), 112 ('લ', 'ળ'), 113 ('વ', 'હ'), 114 ('ઽ', 'ઽ'), 115 ('ૐ', 'ૐ'), 116 ('ૠ', 'ૡ'), 117 ('ૹ', 'ૹ'), 118 ('ଅ', 'ଌ'), 119 ('ଏ', 'ଐ'), 120 ('ଓ', 'ନ'), 121 ('ପ', 'ର'), 122 ('ଲ', 'ଳ'), 123 ('ଵ', 'ହ'), 124 ('ଽ', 'ଽ'), 125 ('ଡ଼', 'ଢ଼'), 126 ('ୟ', 'ୡ'), 127 ('ୱ', 'ୱ'), 128 ('ஃ', 'ஃ'), 129 ('அ', 'ஊ'), 130 ('எ', 'ஐ'), 131 ('ஒ', 'க'), 132 ('ங', 'ச'), 133 ('ஜ', 'ஜ'), 134 ('ஞ', 'ட'), 135 ('ண', 'த'), 136 ('ந', 'ப'), 137 ('ம', 'ஹ'), 138 ('ௐ', 'ௐ'), 139 ('అ', 'ఌ'), 140 ('ఎ', 'ఐ'), 141 ('ఒ', 'న'), 142 ('ప', 'హ'), 143 ('ఽ', 'ఽ'), 144 ('ౘ', 'ౚ'), 145 ('ౝ', 'ౝ'), 146 ('ౠ', 'ౡ'), 147 ('ಀ', 'ಀ'), 148 ('ಅ', 'ಌ'), 149 ('ಎ', 'ಐ'), 150 ('ಒ', 'ನ'), 151 ('ಪ', 'ಳ'), 152 ('ವ', 'ಹ'), 153 ('ಽ', 'ಽ'), 154 ('ೝ', 'ೞ'), 155 ('ೠ', 'ೡ'), 156 ('ೱ', 'ೲ'), 157 ('ഄ', 'ഌ'), 158 ('എ', 'ഐ'), 159 ('ഒ', 'ഺ'), 160 ('ഽ', 'ഽ'), 161 ('ൎ', 'ൎ'), 162 ('ൔ', 'ൖ'), 163 ('ൟ', 'ൡ'), 164 ('ൺ', 'ൿ'), 165 ('අ', 'ඖ'), 166 ('ක', 'න'), 167 ('ඳ', 'ර'), 168 ('ල', 'ල'), 169 ('ව', 'ෆ'), 170 ('ༀ', 'ༀ'), 171 ('ཀ', 'ཇ'), 172 ('ཉ', 'ཬ'), 173 ('ྈ', 'ྌ'), 174 ('Ⴀ', 'Ⴥ'), 175 ('Ⴧ', 'Ⴧ'), 176 ('Ⴭ', 'Ⴭ'), 177 ('ა', 'ჺ'), 178 ('ჼ', 'ቈ'), 179 ('ቊ', 'ቍ'), 180 ('ቐ', 'ቖ'), 181 ('ቘ', 'ቘ'), 182 ('ቚ', 'ቝ'), 183 ('በ', 'ኈ'), 184 ('ኊ', 'ኍ'), 185 ('ነ', 'ኰ'), 186 ('ኲ', 'ኵ'), 187 ('ኸ', 'ኾ'), 188 ('ዀ', 'ዀ'), 189 ('ዂ', 'ዅ'), 190 ('ወ', 'ዖ'), 191 ('ዘ', 'ጐ'), 192 ('ጒ', 'ጕ'), 193 ('ጘ', 'ፚ'), 194 ('ᎀ', 'ᎏ'), 195 ('Ꭰ', 'Ᏽ'), 196 ('ᏸ', 'ᏽ'), 197 ('ᐁ', 'ᙬ'), 198 ('ᙯ', 'ᙿ'), 199 ('ᚁ', 'ᚚ'), 200 ('ᚠ', 'ᛪ'), 201 ('ᛮ', 'ᛸ'), 202 ('ᜀ', 'ᜑ'), 203 ('ᜟ', 'ᜱ'), 204 ('ᝀ', 'ᝑ'), 205 ('ᝠ', 'ᝬ'), 206 ('ᝮ', 'ᝰ'), 207 ('ᠠ', 'ᡸ'), 208 ('ᢀ', 'ᢄ'), 209 ('ᢇ', 'ᢨ'), 210 ('ᢪ', 'ᢪ'), 211 ('ᢰ', 'ᣵ'), 212 ('ᤀ', 'ᤞ'), 213 ('ᨀ', 'ᨖ'), 214 ('ᬅ', 'ᬳ'), 215 ('ᭅ', 'ᭌ'), 216 ('ᮃ', 'ᮠ'), 217 ('ᮮ', 'ᮯ'), 218 ('ᮺ', 'ᯥ'), 219 ('ᰀ', 'ᰣ'), 220 ('ᱍ', 'ᱏ'), 221 ('ᱚ', 'ᱽ'), 222 ('ᲀ', 'ᲈ'), 223 ('Ა', 'Ჺ'), 224 ('Ჽ', 'Ჿ'), 225 ('ᳩ', 'ᳬ'), 226 ('ᳮ', 'ᳳ'), 227 ('ᳵ', 'ᳶ'), 228 ('ᳺ', 'ᳺ'), 229 ('ᴀ', 'ᶿ'), 230 ('Ḁ', 'ἕ'), 231 ('Ἐ', 'Ἕ'), 232 ('ἠ', 'ὅ'), 233 ('Ὀ', 'Ὅ'), 234 ('ὐ', 'ὗ'), 235 ('Ὑ', 'Ὑ'), 236 ('Ὓ', 'Ὓ'), 237 ('Ὕ', 'Ὕ'), 238 ('Ὗ', 'ώ'), 239 ('ᾀ', 'ᾴ'), 240 ('ᾶ', 'ᾼ'), 241 ('ι', 'ι'), 242 ('ῂ', 'ῄ'), 243 ('ῆ', 'ῌ'), 244 ('ῐ', 'ΐ'), 245 ('ῖ', 'Ί'), 246 ('ῠ', 'Ῥ'), 247 ('ῲ', 'ῴ'), 248 ('ῶ', 'ῼ'), 249 ('ⁱ', 'ⁱ'), 250 ('ⁿ', 'ⁿ'), 251 ('ₐ', 'ₜ'), 252 ('ℂ', 'ℂ'), 253 ('ℇ', 'ℇ'), 254 ('ℊ', 'ℓ'), 255 ('ℕ', 'ℕ'), 256 ('ℙ', 'ℝ'), 257 ('ℤ', 'ℤ'), 258 ('Ω', 'Ω'), 259 ('ℨ', 'ℨ'), 260 ('K', 'ℭ'), 261 ('ℯ', 'ℹ'), 262 ('ℼ', 'ℿ'), 263 ('ⅅ', 'ⅉ'), 264 ('ⅎ', 'ⅎ'), 265 ('Ⅰ', 'ↈ'), 266 ('Ⓐ', 'ⓩ'), 267 ('Ⰰ', 'ⳤ'), 268 ('Ⳬ', 'ⳮ'), 269 ('Ⳳ', 'ⳳ'), 270 ('ⴀ', 'ⴥ'), 271 ('ⴧ', 'ⴧ'), 272 ('ⴭ', 'ⴭ'), 273 ('ⴰ', 'ⵧ'), 274 ('ⵯ', 'ⵯ'), 275 ('ⶀ', 'ⶖ'), 276 ('ⶠ', 'ⶦ'), 277 ('ⶨ', 'ⶮ'), 278 ('ⶰ', 'ⶶ'), 279 ('ⶸ', 'ⶾ'), 280 ('ⷀ', 'ⷆ'), 281 ('ⷈ', 'ⷎ'), 282 ('ⷐ', 'ⷖ'), 283 ('ⷘ', 'ⷞ'), 284 ('ⸯ', 'ⸯ'), 285 ('々', '々'), 286 ('〻', '〼'), 287 ('ㄅ', 'ㄯ'), 288 ('ㄱ', 'ㆎ'), 289 ('ㆠ', 'ㆿ'), 290 ('ꀀ', 'ꒌ'), 291 ('ꓐ', 'ꓽ'), 292 ('ꔀ', 'ꘌ'), 293 ('ꘐ', 'ꘟ'), 294 ('ꘪ', 'ꘫ'), 295 ('Ꙁ', 'ꙮ'), 296 ('ꙿ', 'ꚝ'), 297 ('ꚠ', 'ꛯ'), 298 ('꜈', 'ꟊ'), 299 ('Ꟑ', 'ꟑ'), 300 ('ꟓ', 'ꟓ'), 301 ('ꟕ', 'ꟙ'), 302 ('ꟲ', 'ꠁ'), 303 ('ꠃ', 'ꠅ'), 304 ('ꠇ', 'ꠊ'), 305 ('ꠌ', 'ꠢ'), 306 ('ꡀ', 'ꡳ'), 307 ('ꢂ', 'ꢳ'), 308 ('ꣲ', 'ꣷ'), 309 ('ꣻ', 'ꣻ'), 310 ('ꣽ', 'ꣾ'), 311 ('ꤊ', 'ꤥ'), 312 ('ꤰ', 'ꥆ'), 313 ('ꥠ', 'ꥼ'), 314 ('ꦄ', 'ꦲ'), 315 ('ꧏ', 'ꧏ'), 316 ('ꨀ', 'ꨨ'), 317 ('ꩀ', 'ꩂ'), 318 ('ꩄ', 'ꩋ'), 319 ('ꫠ', 'ꫪ'), 320 ('ꫲ', 'ꫴ'), 321 ('ꬁ', 'ꬆ'), 322 ('ꬉ', 'ꬎ'), 323 ('ꬑ', 'ꬖ'), 324 ('ꬠ', 'ꬦ'), 325 ('ꬨ', 'ꬮ'), 326 ('ꬰ', 'ꭩ'), 327 ('ꭰ', 'ꯢ'), 328 ('가', '힣'), 329 ('ힰ', 'ퟆ'), 330 ('ퟋ', 'ퟻ'), 331 ('ff', 'st'), 332 ('ﬓ', 'ﬗ'), 333 ('ﭐ', 'ﮱ'), 334 ('ﯓ', 'ﴽ'), 335 ('ﵐ', 'ﶏ'), 336 ('ﶒ', 'ﷇ'), 337 ('ﷰ', 'ﷻ'), 338 ('ﹰ', 'ﹴ'), 339 ('ﹶ', 'ﻼ'), 340 ('Ａ', 'Ｚ'), 341 ('ａ', 'ｚ'), 342 ('ﾠ', 'ﾾ'), 343 ('ￂ', 'ￇ'), 344 ('ￊ', 'ￏ'), 345 ('ￒ', 'ￗ'), 346 ('ￚ', 'ￜ'), 347 ('��', '��'), 348 ('��', '��'), 349 ('��', '��'), 350 ('��', '��'), 351 ('��', '��'), 352 ('��', '��'), 353 ('��', '��'), 354 ('��', '��'), 355 ('��', '��'), 356 ('��', '��'), 357 ('��', '��'), 358 ('��', '��'), 359 ('��', '��'), 360 ('��', '��'), 361 ('��', '��'), 362 ('��', '��'), 363 ('��', '��'), 364 ('��', '��'), 365 ('��', '��'), 366 ('��', '��'), 367 ('��', '��'), 368 ('��', '��'), 369 ('��', '��'), 370 ('��', '��'), 371 ('��', '��'), 372 ('��', '��'), 373 ('��', '��'), 374 ('��', '��'), 375 ('��', '��'), 376 ('��', '��'), 377 ('��', '��'), 378 ('��', '��'), 379 ('��', '��'), 380 ('��', '��'), 381 ('��', '��'), 382 ('��', '��'), 383 ('��', '��'), 384 ('��', '��'), 385 ('��', '��'), 386 ('��', '��'), 387 ('��', '��'), 388 ('��', '��'), 389 ('��', '��'), 390 ('��', '��'), 391 ('��', '��'), 392 ('��', '��'), 393 ('��', '��'), 394 ('��', '��'), 395 ('��', '��'), 396 ('��', '��'), 397 ('��', '��'), 398 ('��', '��'), 399 ('��', '��'), 400 ('��', '��'), 401 ('��', '��'), 402 ('��', '��'), 403 ('��', '��'), 404 ('��', '��'), 405 ('��', '��'), 406 ('��', '��'), 407 ('��', '��'), 408 ('��', '��'), 409 ('��', '��'), 410 ('��', '��'), 411 ('��', '��'), 412 ('��', '��'), 413 ('��', '��'), 414 ('��', '��'), 415 ('��', '��'), 416 ('��', '��'), 417 ('��', '��'), 418 ('��', '��'), 419 ('��', '��'), 420 ('��', '��'), 421 ('��', '��'), 422 ('��', '��'), 423 ('��', '��'), 424 ('��', '��'), 425 ('��', '��'), 426 ('��', '��'), 427 ('��', '��'), 428 ('��', '��'), 429 ('��', '��'), 430 ('��', '��'), 431 ('��', '��'), 432 ('��', '��'), 433 ('��', '��'), 434 ('��', '��'), 435 ('��', '��'), 436 ('��', '��'), 437 ('��', '��'), 438 ('��', '��'), 439 ('��', '��'), 440 ('��', '��'), 441 ('��', '��'), 442 ('��', '��'), 443 ('��', '��'), 444 ('��', '��'), 445 ('��', '��'), 446 ('��', '��'), 447 ('��', '��'), 448 ('��', '��'), 449 ('��', '��'), 450 ('��', '��'), 451 ('��', '��'), 452 ('��', '��'), 453 ('��', '��'), 454 ('��', '��'), 455 ('��', '��'), 456 ('��', '��'), 457 ('��', '��'), 458 ('��', '��'), 459 ('��', '��'), 460 ('��', '��'), 461 ('��', '��'), 462 ('��', '��'), 463 ('��', '��'), 464 ('��', '��'), 465 ('��', '��'), 466 ('��', '��'), 467 ('��', '��'), 468 ('��', '��'), 469 ('��', '��'), 470 ('��', '��'), 471 ('��', '��'), 472 ('��', '��'), 473 ('��', '��'), 474 ('��', '��'), 475 ('��', '��'), 476 ('��', '��'), 477 ('��', '��'), 478 ('��', '��'), 479 ('��', '��'), 480 ('��', '��'), 481 ('��', '��'), 482 ('��', '��'), 483 ('��', '��'), 484 ('��', '��'), 485 ('��', '��'), 486 ('��', '��'), 487 ('��', '��'), 488 ('��', '��'), 489 ('��', '��'), 490 ('��', '��'), 491 ('��', '��'), 492 ('��', '��'), 493 ('��', '��'), 494 ('��', '��'), 495 ('��', '��'), 496 ('��', '��'), 497 ('��', '��'), 498 ('��', '��'), 499 ('��', '��'), 500 ('��', '��'), 501 ('��', '��'), 502 ('��', '��'), 503 ('��', '��'), 504 ('��', '��'), 505 ('��', '��'), 506 ('��', '��'), 507 ('��', '��'), 508 ('��', '��'), 509 ('��', '��'), 510 ('��', '��'), 511 ('��', '��'), 512 ('��', '��'), 513 ('��', '��'), 514 ('��', '��'), 515 ('��', '��'), 516 ('��', '��'), 517 ('��', '��'), 518 ('��', '��'), 519 ('��', '��'), 520 ('��', '��'), 521 ('��', '��'), 522 ('��', '��'), 523 ('��', '��'), 524 ('��', '��'), 525 ('��', '��'), 526 ('��', '��'), 527 ('��', '��'), 528 ('��', '��'), 529 ('��', '��'), 530 ('��', '��'), 531 ('��', '��'), 532 ('��', '��'), 533 ('��', '��'), 534 ('��', '��'), 535 ('��', '��'), 536 ('��', '��'), 537 ('��', '��'), 538 ('��', '��'), 539 ('��', '��'), 540 ('��', '��'), 541 ('��', '��'), 542 ('��', '��'), 543 ('��', '��'), 544 ('��', '��'), 545 ('��', '��'), 546 ('��', '��'), 547 ('��', '��'), 548 ('��', '��'), 549 ('��', '��'), 550 ('��', '��'), 551 ('��', '��'), 552 ('��', '��'), 553 ('��', '��'), 554 ('��', '��'), 555 ('��', '��'), 556 ('��', '��'), 557 ('��', '��'), 558 ('��', '��'), 559 ('��', '��'), 560 ('��', '��'), 561 ('��', '��'), 562 ('��', '��'), 563 ('��', '��'), 564 ('��', '��'), 565 ('��', '��'), 566 ('��', '��'), 567 ('��', '��'), 568 ('��', '��'), 569 ('��', '��'), 570 ('��', '��'), 571 ('��', '��'), 572 ('��', '��'), 573 ('��', '��'), 574 ('��', '��'), 575 ('��', '��'), 576 ('��', '��'), 577 ('��', '��'), 578 ('��', '��'), 579 ('��', '��'), 580 ('��', '��'), 581 ('��', '��'), 582 ('��', '��'), 583 ('��', '��'), 584 ('��', '��'), 585 ('��', '��'), 586 ('��', '��'), 587 ('��', '��'), 588 ('��', '��'), 589 ('��', '��'), 590 ('��', '��'), 591 ('��', '��'), 592 ('��', '��'), 593 ('��', '��'), 594 ('��', '��'), 595 ('��', '��'), 596 ('��', '��'), 597 ('��', '��'), 598 ('��', '��'), 599 ('��', '��'), 600 ('��', '��'), 601 ('��', '��'), 602 ('��', '��'), 603 ('��', '��'), 604 ('��', '��'), 605 ('��', '��'), 606 ('��', '��'), 607 ('��', '��'), 608 ('��', '��'), 609 ]; 610 611 pub const CR: &'static [(char, char)] = &[('\r', '\r')]; 612 613 pub const DOUBLE_QUOTE: &'static [(char, char)] = &[('"', '"')]; 614 615 pub const EXTEND: &'static [(char, char)] = &[ 616 ('\u{300}', '\u{36f}'), 617 ('\u{483}', '\u{489}'), 618 ('\u{591}', '\u{5bd}'), 619 ('\u{5bf}', '\u{5bf}'), 620 ('\u{5c1}', '\u{5c2}'), 621 ('\u{5c4}', '\u{5c5}'), 622 ('\u{5c7}', '\u{5c7}'), 623 ('\u{610}', '\u{61a}'), 624 ('\u{64b}', '\u{65f}'), 625 ('\u{670}', '\u{670}'), 626 ('\u{6d6}', '\u{6dc}'), 627 ('\u{6df}', '\u{6e4}'), 628 ('\u{6e7}', '\u{6e8}'), 629 ('\u{6ea}', '\u{6ed}'), 630 ('\u{711}', '\u{711}'), 631 ('\u{730}', '\u{74a}'), 632 ('\u{7a6}', '\u{7b0}'), 633 ('\u{7eb}', '\u{7f3}'), 634 ('\u{7fd}', '\u{7fd}'), 635 ('\u{816}', '\u{819}'), 636 ('\u{81b}', '\u{823}'), 637 ('\u{825}', '\u{827}'), 638 ('\u{829}', '\u{82d}'), 639 ('\u{859}', '\u{85b}'), 640 ('\u{898}', '\u{89f}'), 641 ('\u{8ca}', '\u{8e1}'), 642 ('\u{8e3}', 'ः'), 643 ('\u{93a}', '\u{93c}'), 644 ('ा', 'ॏ'), 645 ('\u{951}', '\u{957}'), 646 ('\u{962}', '\u{963}'), 647 ('\u{981}', 'ঃ'), 648 ('\u{9bc}', '\u{9bc}'), 649 ('\u{9be}', '\u{9c4}'), 650 ('ে', 'ৈ'), 651 ('ো', '\u{9cd}'), 652 ('\u{9d7}', '\u{9d7}'), 653 ('\u{9e2}', '\u{9e3}'), 654 ('\u{9fe}', '\u{9fe}'), 655 ('\u{a01}', 'ਃ'), 656 ('\u{a3c}', '\u{a3c}'), 657 ('ਾ', '\u{a42}'), 658 ('\u{a47}', '\u{a48}'), 659 ('\u{a4b}', '\u{a4d}'), 660 ('\u{a51}', '\u{a51}'), 661 ('\u{a70}', '\u{a71}'), 662 ('\u{a75}', '\u{a75}'), 663 ('\u{a81}', 'ઃ'), 664 ('\u{abc}', '\u{abc}'), 665 ('ા', '\u{ac5}'), 666 ('\u{ac7}', 'ૉ'), 667 ('ો', '\u{acd}'), 668 ('\u{ae2}', '\u{ae3}'), 669 ('\u{afa}', '\u{aff}'), 670 ('\u{b01}', 'ଃ'), 671 ('\u{b3c}', '\u{b3c}'), 672 ('\u{b3e}', '\u{b44}'), 673 ('େ', 'ୈ'), 674 ('ୋ', '\u{b4d}'), 675 ('\u{b55}', '\u{b57}'), 676 ('\u{b62}', '\u{b63}'), 677 ('\u{b82}', '\u{b82}'), 678 ('\u{bbe}', 'ூ'), 679 ('ெ', 'ை'), 680 ('ொ', '\u{bcd}'), 681 ('\u{bd7}', '\u{bd7}'), 682 ('\u{c00}', '\u{c04}'), 683 ('\u{c3c}', '\u{c3c}'), 684 ('\u{c3e}', 'ౄ'), 685 ('\u{c46}', '\u{c48}'), 686 ('\u{c4a}', '\u{c4d}'), 687 ('\u{c55}', '\u{c56}'), 688 ('\u{c62}', '\u{c63}'), 689 ('\u{c81}', 'ಃ'), 690 ('\u{cbc}', '\u{cbc}'), 691 ('ಾ', 'ೄ'), 692 ('\u{cc6}', 'ೈ'), 693 ('ೊ', '\u{ccd}'), 694 ('\u{cd5}', '\u{cd6}'), 695 ('\u{ce2}', '\u{ce3}'), 696 ('ೳ', 'ೳ'), 697 ('\u{d00}', 'ഃ'), 698 ('\u{d3b}', '\u{d3c}'), 699 ('\u{d3e}', '\u{d44}'), 700 ('െ', 'ൈ'), 701 ('ൊ', '\u{d4d}'), 702 ('\u{d57}', '\u{d57}'), 703 ('\u{d62}', '\u{d63}'), 704 ('\u{d81}', 'ඃ'), 705 ('\u{dca}', '\u{dca}'), 706 ('\u{dcf}', '\u{dd4}'), 707 ('\u{dd6}', '\u{dd6}'), 708 ('ෘ', '\u{ddf}'), 709 ('ෲ', 'ෳ'), 710 ('\u{e31}', '\u{e31}'), 711 ('\u{e34}', '\u{e3a}'), 712 ('\u{e47}', '\u{e4e}'), 713 ('\u{eb1}', '\u{eb1}'), 714 ('\u{eb4}', '\u{ebc}'), 715 ('\u{ec8}', '\u{ece}'), 716 ('\u{f18}', '\u{f19}'), 717 ('\u{f35}', '\u{f35}'), 718 ('\u{f37}', '\u{f37}'), 719 ('\u{f39}', '\u{f39}'), 720 ('༾', '༿'), 721 ('\u{f71}', '\u{f84}'), 722 ('\u{f86}', '\u{f87}'), 723 ('\u{f8d}', '\u{f97}'), 724 ('\u{f99}', '\u{fbc}'), 725 ('\u{fc6}', '\u{fc6}'), 726 ('ါ', '\u{103e}'), 727 ('ၖ', '\u{1059}'), 728 ('\u{105e}', '\u{1060}'), 729 ('ၢ', 'ၤ'), 730 ('ၧ', 'ၭ'), 731 ('\u{1071}', '\u{1074}'), 732 ('\u{1082}', '\u{108d}'), 733 ('ႏ', 'ႏ'), 734 ('ႚ', '\u{109d}'), 735 ('\u{135d}', '\u{135f}'), 736 ('\u{1712}', '᜕'), 737 ('\u{1732}', '᜴'), 738 ('\u{1752}', '\u{1753}'), 739 ('\u{1772}', '\u{1773}'), 740 ('\u{17b4}', '\u{17d3}'), 741 ('\u{17dd}', '\u{17dd}'), 742 ('\u{180b}', '\u{180d}'), 743 ('\u{180f}', '\u{180f}'), 744 ('\u{1885}', '\u{1886}'), 745 ('\u{18a9}', '\u{18a9}'), 746 ('\u{1920}', 'ᤫ'), 747 ('ᤰ', '\u{193b}'), 748 ('\u{1a17}', '\u{1a1b}'), 749 ('ᩕ', '\u{1a5e}'), 750 ('\u{1a60}', '\u{1a7c}'), 751 ('\u{1a7f}', '\u{1a7f}'), 752 ('\u{1ab0}', '\u{1ace}'), 753 ('\u{1b00}', 'ᬄ'), 754 ('\u{1b34}', '᭄'), 755 ('\u{1b6b}', '\u{1b73}'), 756 ('\u{1b80}', 'ᮂ'), 757 ('ᮡ', '\u{1bad}'), 758 ('\u{1be6}', '᯳'), 759 ('ᰤ', '\u{1c37}'), 760 ('\u{1cd0}', '\u{1cd2}'), 761 ('\u{1cd4}', '\u{1ce8}'), 762 ('\u{1ced}', '\u{1ced}'), 763 ('\u{1cf4}', '\u{1cf4}'), 764 ('᳷', '\u{1cf9}'), 765 ('\u{1dc0}', '\u{1dff}'), 766 ('\u{200c}', '\u{200c}'), 767 ('\u{20d0}', '\u{20f0}'), 768 ('\u{2cef}', '\u{2cf1}'), 769 ('\u{2d7f}', '\u{2d7f}'), 770 ('\u{2de0}', '\u{2dff}'), 771 ('\u{302a}', '\u{302f}'), 772 ('\u{3099}', '\u{309a}'), 773 ('\u{a66f}', '\u{a672}'), 774 ('\u{a674}', '\u{a67d}'), 775 ('\u{a69e}', '\u{a69f}'), 776 ('\u{a6f0}', '\u{a6f1}'), 777 ('\u{a802}', '\u{a802}'), 778 ('\u{a806}', '\u{a806}'), 779 ('\u{a80b}', '\u{a80b}'), 780 ('ꠣ', 'ꠧ'), 781 ('\u{a82c}', '\u{a82c}'), 782 ('ꢀ', 'ꢁ'), 783 ('ꢴ', '\u{a8c5}'), 784 ('\u{a8e0}', '\u{a8f1}'), 785 ('\u{a8ff}', '\u{a8ff}'), 786 ('\u{a926}', '\u{a92d}'), 787 ('\u{a947}', '꥓'), 788 ('\u{a980}', 'ꦃ'), 789 ('\u{a9b3}', '꧀'), 790 ('\u{a9e5}', '\u{a9e5}'), 791 ('\u{aa29}', '\u{aa36}'), 792 ('\u{aa43}', '\u{aa43}'), 793 ('\u{aa4c}', 'ꩍ'), 794 ('ꩻ', 'ꩽ'), 795 ('\u{aab0}', '\u{aab0}'), 796 ('\u{aab2}', '\u{aab4}'), 797 ('\u{aab7}', '\u{aab8}'), 798 ('\u{aabe}', '\u{aabf}'), 799 ('\u{aac1}', '\u{aac1}'), 800 ('ꫫ', 'ꫯ'), 801 ('ꫵ', '\u{aaf6}'), 802 ('ꯣ', 'ꯪ'), 803 ('꯬', '\u{abed}'), 804 ('\u{fb1e}', '\u{fb1e}'), 805 ('\u{fe00}', '\u{fe0f}'), 806 ('\u{fe20}', '\u{fe2f}'), 807 ('\u{ff9e}', '\u{ff9f}'), 808 ('\u{101fd}', '\u{101fd}'), 809 ('\u{102e0}', '\u{102e0}'), 810 ('\u{10376}', '\u{1037a}'), 811 ('\u{10a01}', '\u{10a03}'), 812 ('\u{10a05}', '\u{10a06}'), 813 ('\u{10a0c}', '\u{10a0f}'), 814 ('\u{10a38}', '\u{10a3a}'), 815 ('\u{10a3f}', '\u{10a3f}'), 816 ('\u{10ae5}', '\u{10ae6}'), 817 ('\u{10d24}', '\u{10d27}'), 818 ('\u{10eab}', '\u{10eac}'), 819 ('\u{10efd}', '\u{10eff}'), 820 ('\u{10f46}', '\u{10f50}'), 821 ('\u{10f82}', '\u{10f85}'), 822 ('��', '��'), 823 ('\u{11038}', '\u{11046}'), 824 ('\u{11070}', '\u{11070}'), 825 ('\u{11073}', '\u{11074}'), 826 ('\u{1107f}', '��'), 827 ('��', '\u{110ba}'), 828 ('\u{110c2}', '\u{110c2}'), 829 ('\u{11100}', '\u{11102}'), 830 ('\u{11127}', '\u{11134}'), 831 ('��', '��'), 832 ('\u{11173}', '\u{11173}'), 833 ('\u{11180}', '��'), 834 ('��', '��'), 835 ('\u{111c9}', '\u{111cc}'), 836 ('��', '\u{111cf}'), 837 ('��', '\u{11237}'), 838 ('\u{1123e}', '\u{1123e}'), 839 ('\u{11241}', '\u{11241}'), 840 ('\u{112df}', '\u{112ea}'), 841 ('\u{11300}', '��'), 842 ('\u{1133b}', '\u{1133c}'), 843 ('\u{1133e}', '��'), 844 ('��', '��'), 845 ('��', '��'), 846 ('\u{11357}', '\u{11357}'), 847 ('��', '��'), 848 ('\u{11366}', '\u{1136c}'), 849 ('\u{11370}', '\u{11374}'), 850 ('��', '\u{11446}'), 851 ('\u{1145e}', '\u{1145e}'), 852 ('\u{114b0}', '\u{114c3}'), 853 ('\u{115af}', '\u{115b5}'), 854 ('��', '\u{115c0}'), 855 ('\u{115dc}', '\u{115dd}'), 856 ('��', '\u{11640}'), 857 ('\u{116ab}', '\u{116b7}'), 858 ('\u{1171d}', '\u{1172b}'), 859 ('��', '\u{1183a}'), 860 ('\u{11930}', '��'), 861 ('��', '��'), 862 ('\u{1193b}', '\u{1193e}'), 863 ('��', '��'), 864 ('��', '\u{11943}'), 865 ('��', '\u{119d7}'), 866 ('\u{119da}', '\u{119e0}'), 867 ('��', '��'), 868 ('\u{11a01}', '\u{11a0a}'), 869 ('\u{11a33}', '��'), 870 ('\u{11a3b}', '\u{11a3e}'), 871 ('\u{11a47}', '\u{11a47}'), 872 ('\u{11a51}', '\u{11a5b}'), 873 ('\u{11a8a}', '\u{11a99}'), 874 ('��', '\u{11c36}'), 875 ('\u{11c38}', '\u{11c3f}'), 876 ('\u{11c92}', '\u{11ca7}'), 877 ('��', '\u{11cb6}'), 878 ('\u{11d31}', '\u{11d36}'), 879 ('\u{11d3a}', '\u{11d3a}'), 880 ('\u{11d3c}', '\u{11d3d}'), 881 ('\u{11d3f}', '\u{11d45}'), 882 ('\u{11d47}', '\u{11d47}'), 883 ('��', '��'), 884 ('\u{11d90}', '\u{11d91}'), 885 ('��', '\u{11d97}'), 886 ('\u{11ef3}', '��'), 887 ('\u{11f00}', '\u{11f01}'), 888 ('��', '��'), 889 ('��', '\u{11f3a}'), 890 ('��', '\u{11f42}'), 891 ('\u{13440}', '\u{13440}'), 892 ('\u{13447}', '\u{13455}'), 893 ('\u{16af0}', '\u{16af4}'), 894 ('\u{16b30}', '\u{16b36}'), 895 ('\u{16f4f}', '\u{16f4f}'), 896 ('��', '��'), 897 ('\u{16f8f}', '\u{16f92}'), 898 ('\u{16fe4}', '\u{16fe4}'), 899 ('��', '��'), 900 ('\u{1bc9d}', '\u{1bc9e}'), 901 ('\u{1cf00}', '\u{1cf2d}'), 902 ('\u{1cf30}', '\u{1cf46}'), 903 ('\u{1d165}', '\u{1d169}'), 904 ('��', '\u{1d172}'), 905 ('\u{1d17b}', '\u{1d182}'), 906 ('\u{1d185}', '\u{1d18b}'), 907 ('\u{1d1aa}', '\u{1d1ad}'), 908 ('\u{1d242}', '\u{1d244}'), 909 ('\u{1da00}', '\u{1da36}'), 910 ('\u{1da3b}', '\u{1da6c}'), 911 ('\u{1da75}', '\u{1da75}'), 912 ('\u{1da84}', '\u{1da84}'), 913 ('\u{1da9b}', '\u{1da9f}'), 914 ('\u{1daa1}', '\u{1daaf}'), 915 ('\u{1e000}', '\u{1e006}'), 916 ('\u{1e008}', '\u{1e018}'), 917 ('\u{1e01b}', '\u{1e021}'), 918 ('\u{1e023}', '\u{1e024}'), 919 ('\u{1e026}', '\u{1e02a}'), 920 ('\u{1e08f}', '\u{1e08f}'), 921 ('\u{1e130}', '\u{1e136}'), 922 ('\u{1e2ae}', '\u{1e2ae}'), 923 ('\u{1e2ec}', '\u{1e2ef}'), 924 ('\u{1e4ec}', '\u{1e4ef}'), 925 ('\u{1e8d0}', '\u{1e8d6}'), 926 ('\u{1e944}', '\u{1e94a}'), 927 ('��', '��'), 928 ('\u{e0020}', '\u{e007f}'), 929 ('\u{e0100}', '\u{e01ef}'), 930 ]; 931 932 pub const EXTENDNUMLET: &'static [(char, char)] = &[ 933 ('_', '_'), 934 ('\u{202f}', '\u{202f}'), 935 ('‿', '⁀'), 936 ('⁔', '⁔'), 937 ('︳', '︴'), 938 ('﹍', '﹏'), 939 ('＿', '＿'), 940 ]; 941 942 pub const FORMAT: &'static [(char, char)] = &[ 943 ('\u{ad}', '\u{ad}'), 944 ('\u{600}', '\u{605}'), 945 ('\u{61c}', '\u{61c}'), 946 ('\u{6dd}', '\u{6dd}'), 947 ('\u{70f}', '\u{70f}'), 948 ('\u{890}', '\u{891}'), 949 ('\u{8e2}', '\u{8e2}'), 950 ('\u{180e}', '\u{180e}'), 951 ('\u{200e}', '\u{200f}'), 952 ('\u{202a}', '\u{202e}'), 953 ('\u{2060}', '\u{2064}'), 954 ('\u{2066}', '\u{206f}'), 955 ('\u{feff}', '\u{feff}'), 956 ('\u{fff9}', '\u{fffb}'), 957 ('\u{110bd}', '\u{110bd}'), 958 ('\u{110cd}', '\u{110cd}'), 959 ('\u{13430}', '\u{1343f}'), 960 ('\u{1bca0}', '\u{1bca3}'), 961 ('\u{1d173}', '\u{1d17a}'), 962 ('\u{e0001}', '\u{e0001}'), 963 ]; 964 965 pub const HEBREW_LETTER: &'static [(char, char)] = &[ 966 ('א', 'ת'), 967 ('ׯ', 'ײ'), 968 ('יִ', 'יִ'), 969 ('ײַ', 'ﬨ'), 970 ('שׁ', 'זּ'), 971 ('טּ', 'לּ'), 972 ('מּ', 'מּ'), 973 ('נּ', 'סּ'), 974 ('ףּ', 'פּ'), 975 ('צּ', 'ﭏ'), 976 ]; 977 978 pub const KATAKANA: &'static [(char, char)] = &[ 979 ('〱', '〵'), 980 ('゛', '゜'), 981 ('゠', 'ヺ'), 982 ('ー', 'ヿ'), 983 ('ㇰ', 'ㇿ'), 984 ('㋐', '㋾'), 985 ('㌀', '㍗'), 986 ('ｦ', 'ﾝ'), 987 ('��', '��'), 988 ('��', '��'), 989 ('��', '��'), 990 ('��', '��'), 991 ('��', '��'), 992 ('��', '��'), 993 ('��', '��'), 994 ]; 995 996 pub const LF: &'static [(char, char)] = &[('\n', '\n')]; 997 998 pub const MIDLETTER: &'static [(char, char)] = &[ 999 (':', ':'), 1000 ('·', '·'), 1001 ('·', '·'), 1002 ('՟', '՟'), 1003 ('״', '״'), 1004 ('‧', '‧'), 1005 ('︓', '︓'), 1006 ('﹕', '﹕'), 1007 ('：', '：'), 1008 ]; 1009 1010 pub const MIDNUM: &'static [(char, char)] = &[ 1011 (',', ','), 1012 (';', ';'), 1013 (';', ';'), 1014 ('։', '։'), 1015 ('،', '؍'), 1016 ('٬', '٬'), 1017 ('߸', '߸'), 1018 ('⁄', '⁄'), 1019 ('︐', '︐'), 1020 ('︔', '︔'), 1021 ('﹐', '﹐'), 1022 ('﹔', '﹔'), 1023 ('，', '，'), 1024 ('；', '；'), 1025 ]; 1026 1027 pub const MIDNUMLET: &'static [(char, char)] = &[ 1028 ('.', '.'), 1029 ('‘', '’'), 1030 ('․', '․'), 1031 ('﹒', '﹒'), 1032 ('＇', '＇'), 1033 ('．', '．'), 1034 ]; 1035 1036 pub const NEWLINE: &'static [(char, char)] = 1037 &[('\u{b}', '\u{c}'), ('\u{85}', '\u{85}'), ('\u{2028}', '\u{2029}')]; 1038 1039 pub const NUMERIC: &'static [(char, char)] = &[ 1040 ('0', '9'), 1041 ('٠', '٩'), 1042 ('٫', '٫'), 1043 ('۰', '۹'), 1044 ('߀', '߉'), 1045 ('०', '९'), 1046 ('০', '৯'), 1047 ('੦', '੯'), 1048 ('૦', '૯'), 1049 ('୦', '୯'), 1050 ('௦', '௯'), 1051 ('౦', '౯'), 1052 ('೦', '೯'), 1053 ('൦', '൯'), 1054 ('෦', '෯'), 1055 ('๐', '๙'), 1056 ('໐', '໙'), 1057 ('༠', '༩'), 1058 ('၀', '၉'), 1059 ('႐', '႙'), 1060 ('០', '៩'), 1061 ('᠐', '᠙'), 1062 ('᥆', '᥏'), 1063 ('᧐', '᧙'), 1064 ('᪀', '᪉'), 1065 ('᪐', '᪙'), 1066 ('᭐', '᭙'), 1067 ('᮰', '᮹'), 1068 ('᱀', '᱉'), 1069 ('᱐', '᱙'), 1070 ('꘠', '꘩'), 1071 ('꣐', '꣙'), 1072 ('꤀', '꤉'), 1073 ('꧐', '꧙'), 1074 ('꧰', '꧹'), 1075 ('꩐', '꩙'), 1076 ('꯰', '꯹'), 1077 ('０', '９'), 1078 ('��', '��'), 1079 ('��', '��'), 1080 ('��', '��'), 1081 ('��', '��'), 1082 ('��', '��'), 1083 ('��', '��'), 1084 ('��', '��'), 1085 ('��', '��'), 1086 ('��', '��'), 1087 ('��', '��'), 1088 ('��', '��'), 1089 ('��', '��'), 1090 ('��', '��'), 1091 ('��', '��'), 1092 ('��', '��'), 1093 ('��', '��'), 1094 ('��', '��'), 1095 ('��', '��'), 1096 ('��', '��'), 1097 ('��', '��'), 1098 ('��', '��'), 1099 ('��', '��'), 1100 ('��', '��'), 1101 ('��', '��'), 1102 ('��', '��'), 1103 ('��', '��'), 1104 ('��', '��'), 1105 ]; 1106 1107 pub const REGIONAL_INDICATOR: &'static [(char, char)] = &[('��', '��')]; 1108 1109 pub const SINGLE_QUOTE: &'static [(char, char)] = &[('\'', '\'')]; 1110 1111 pub const WSEGSPACE: &'static [(char, char)] = &[ 1112 (' ', ' '), 1113 ('\u{1680}', '\u{1680}'), 1114 ('\u{2000}', '\u{2006}'), 1115 ('\u{2008}', '\u{200a}'), 1116 ('\u{205f}', '\u{205f}'), 1117 ('\u{3000}', '\u{3000}'), 1118 ]; 1119 1120 pub const ZWJ: &'static [(char, char)] = &[('\u{200d}', '\u{200d}')]; 1121