00001 /* pr29.h --- Detect strings that are non-idempotent under NFKC in Unicode 3.2. 00002 * Copyright (C) 2004, 2005 Simon Josefsson. 00003 * 00004 * This file is part of GNU Libidn. 00005 * 00006 * GNU Libidn is free software; you can redistribute it and/or 00007 * modify it under the terms of the GNU Lesser General Public 00008 * License as published by the Free Software Foundation; either 00009 * version 2.1 of the License, or (at your option) any later version. 00010 * 00011 * GNU Libidn is distributed in the hope that it will be useful, 00012 * but WITHOUT ANY WARRANTY; without even the implied warranty of 00013 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 00014 * Lesser General Public License for more details. 00015 * 00016 * You should have received a copy of the GNU Lesser General Public 00017 * License along with GNU Libidn; if not, write to the Free Software 00018 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA 00019 * 00020 */ 00021 00022 #include "pr29.h" 00023 00024 /* Get stringprep_utf8_to_ucs4. */ 00025 #include <stringprep.h> 00026 00027 /* 00028 * The tables used in this file was extracted by Simon Josefsson from 00029 * pr-29.html and DerivedCombiningClass-3.2.0.txt, as published by 00030 * Unicode Inc., for the GNU Libidn project. 00031 * 00032 */ 00033 00034 /* These are the characters with non-zero combination class, extracted 00035 from DerivedCombiningClass-3.2.0.txt. */ 00036 static uint32_t nzcc[] = { 00037 /* 1 # Mn [5] COMBINING TILDE OVERLAY.. 00038 * ..COMBINING LONG SOLIDUS OVERLAY */ 00039 0x0334, 00040 0x0335, 00041 0x0336, 00042 0x0337, 00043 0x0338, 00044 /* 1 # Mn [2] COMBINING LONG VERTICAL LINE OVERLAY.. 00045 * ..COMBINING SHORT VERTICAL LINE OVERLAY */ 00046 0x20D2, 00047 0x20D3, 00048 /* 1 # Mn [3] COMBINING RING OVERLAY.. 00049 * ..COMBINING ANTICLOCKWISE RING OVERLAY */ 00050 0x20D8, 00051 0x20D9, 00052 0x20DA, 00053 /* 1 # Mn [2] COMBINING REVERSE SOLIDUS OVERLAY.. 00054 * ..COMBINING DOUBLE VERTICAL STROKE OVERLAY */ 00055 0x20E5, 00056 0x20E6, 00057 /* 1 # Mn COMBINING LEFTWARDS ARROW OVERLAY */ 00058 0x20EA, 00059 /* 1 # Mn [3] MUSICAL SYMBOL COMBINING TREMOLO-1.. 00060 * ..MUSICAL SYMBOL COMBINING TREMOLO-3 */ 00061 0x1D167, 00062 0x1D168, 00063 0x1D169, 00064 /* 7 # Mn DEVANAGARI SIGN NUKTA */ 00065 0x093C, 00066 /* 7 # Mn BENGALI SIGN NUKTA */ 00067 0x09BC, 00068 /* 7 # Mn GURMUKHI SIGN NUKTA */ 00069 0x0A3C, 00070 /* 7 # Mn GUJARATI SIGN NUKTA */ 00071 0x0ABC, 00072 /* 7 # Mn ORIYA SIGN NUKTA */ 00073 0x0B3C, 00074 /* 7 # Mn MYANMAR SIGN DOT BELOW */ 00075 0x1037, 00076 /* 8 # Mn [2] COMBINING KATAKANA-HIRAGANA VOICED SOUND MARK.. 00077 * ..COMBINING KATAKANA-HIRAGANA SEMI-VOICED SOUND MARK */ 00078 0x3099, 00079 0x309A, 00080 /* 9 # Mn DEVANAGARI SIGN VIRAMA */ 00081 0x094D, 00082 /* 9 # Mn BENGALI SIGN VIRAMA */ 00083 0x09CD, 00084 /* 9 # Mn GURMUKHI SIGN VIRAMA */ 00085 0x0A4D, 00086 /* 9 # Mn GUJARATI SIGN VIRAMA */ 00087 0x0ACD, 00088 /* 9 # Mn ORIYA SIGN VIRAMA */ 00089 0x0B4D, 00090 /* 9 # Mn TAMIL SIGN VIRAMA */ 00091 0x0BCD, 00092 /* 9 # Mn TELUGU SIGN VIRAMA */ 00093 0x0C4D, 00094 /* 9 # Mn KANNADA SIGN VIRAMA */ 00095 0x0CCD, 00096 /* 9 # Mn MALAYALAM SIGN VIRAMA */ 00097 0x0D4D, 00098 /* 9 # Mn SINHALA SIGN AL-LAKUNA */ 00099 0x0DCA, 00100 /* 9 # Mn THAI CHARACTER PHINTHU */ 00101 0x0E3A, 00102 /* 9 # Mn TIBETAN MARK HALANTA */ 00103 0x0F84, 00104 /* 9 # Mn MYANMAR SIGN VIRAMA */ 00105 0x1039, 00106 /* 9 # Mn TAGALOG SIGN VIRAMA */ 00107 0x1714, 00108 /* 9 # Mn HANUNOO SIGN PAMUDPOD */ 00109 0x1734, 00110 /* 9 # Mn KHMER SIGN COENG */ 00111 0x17D2, 00112 /* 10 # Mn HEBREW POINT SHEVA */ 00113 0x05B0, 00114 /* 11 # Mn HEBREW POINT HATAF SEGOL */ 00115 0x05B1, 00116 /* 12 # Mn HEBREW POINT HATAF PATAH */ 00117 0x05B2, 00118 /* 13 # Mn HEBREW POINT HATAF QAMATS */ 00119 0x05B3, 00120 /* 14 # Mn HEBREW POINT HIRIQ */ 00121 0x05B4, 00122 /* 15 # Mn HEBREW POINT TSERE */ 00123 0x05B5, 00124 /* 16 # Mn HEBREW POINT SEGOL */ 00125 0x05B6, 00126 /* 17 # Mn HEBREW POINT PATAH */ 00127 0x05B7, 00128 /* 18 # Mn HEBREW POINT QAMATS */ 00129 0x05B8, 00130 /* 19 # Mn HEBREW POINT HOLAM */ 00131 0x05B9, 00132 /* 20 # Mn HEBREW POINT QUBUTS */ 00133 0x05BB, 00134 /* 21 # Mn HEBREW POINT DAGESH OR MAPIQ */ 00135 0x05BC, 00136 /* 22 # Mn HEBREW POINT METEG */ 00137 0x05BD, 00138 /* 23 # Mn HEBREW POINT RAFE */ 00139 0x05BF, 00140 /* 24 # Mn HEBREW POINT SHIN DOT */ 00141 0x05C1, 00142 /* 25 # Mn HEBREW POINT SIN DOT */ 00143 0x05C2, 00144 /* 26 # Mn HEBREW POINT JUDEO-SPANISH VARIKA */ 00145 0xFB1E, 00146 /* 27 # Mn ARABIC FATHATAN */ 00147 0x064B, 00148 /* 28 # Mn ARABIC DAMMATAN */ 00149 0x064C, 00150 /* 29 # Mn ARABIC KASRATAN */ 00151 0x064D, 00152 /* 30 # Mn ARABIC FATHA */ 00153 0x064E, 00154 /* 31 # Mn ARABIC DAMMA */ 00155 0x064F, 00156 /* 32 # Mn ARABIC KASRA */ 00157 0x0650, 00158 /* 33 # Mn ARABIC SHADDA */ 00159 0x0651, 00160 /* 34 # Mn ARABIC SUKUN */ 00161 0x0652, 00162 /* 35 # Mn ARABIC LETTER SUPERSCRIPT ALEF */ 00163 0x0670, 00164 /* 36 # Mn SYRIAC LETTER SUPERSCRIPT ALAPH */ 00165 0x0711, 00166 /* 84 # Mn TELUGU LENGTH MARK */ 00167 0x0C55, 00168 /* 91 # Mn TELUGU AI LENGTH MARK */ 00169 0x0C56, 00170 /* 103 # Mn [2] THAI CHARACTER SARA U.. 00171 * ..THAI CHARACTER SARA UU */ 00172 0x0E38, 00173 0x0E39, 00174 /* 107 # Mn [4] THAI CHARACTER MAI EK.. 00175 * ..THAI CHARACTER MAI CHATTAWA */ 00176 0x0E48, 00177 0x0E49, 00178 0x0E4A, 00179 0x04EB, 00180 /* 118 # Mn [2] LAO VOWEL SIGN U.. 00181 * ..LAO VOWEL SIGN UU */ 00182 0x0EB8, 00183 0x0EB9, 00184 /* 122 # Mn [4] LAO TONE MAI EK.. 00185 * ..LAO TONE MAI CATAWA */ 00186 0x0EC8, 00187 0x0EC9, 00188 0x0ECA, 00189 0x0ECB, 00190 /* 129 # Mn TIBETAN VOWEL SIGN AA */ 00191 0x0F71, 00192 /* 130 # Mn TIBETAN VOWEL SIGN I */ 00193 0x0F72, 00194 /* 130 # Mn [4] TIBETAN VOWEL SIGN E.. 00195 * ..TIBETAN VOWEL SIGN OO */ 00196 0x0F7A, 00197 0x0F7B, 00198 0x0F7C, 00199 0x0F7D, 00200 /* 130 # Mn TIBETAN VOWEL SIGN REVERSED I */ 00201 0x0F80, 00202 /* 132 # Mn TIBETAN VOWEL SIGN U */ 00203 0x0F74, 00204 /* 202 # Mn [2] COMBINING PALATALIZED HOOK BELOW.. 00205 * ..COMBINING RETROFLEX HOOK BELOW */ 00206 0x0321, 00207 0x0322, 00208 /* 202 # Mn [2] COMBINING CEDILLA.. 00209 * ..COMBINING OGONEK */ 00210 0x0327, 00211 0x0328, 00212 /* 216 # Mn COMBINING HORN */ 00213 0x031B, 00214 /* 216 # Mn TIBETAN MARK TSA -PHRU */ 00215 0x0F39, 00216 /* 216 # Mc [2] MUSICAL SYMBOL COMBINING STEM.. 00217 * ..MUSICAL SYMBOL COMBINING SPRECHGESANG STEM */ 00218 0x1D165, 00219 0x1D166, 00220 /* 216 # Mc [5] MUSICAL SYMBOL COMBINING FLAG-1.. 00221 * ..MUSICAL SYMBOL COMBINING FLAG-5 */ 00222 0x1D16E, 00223 0x1D16F, 00224 0x1D170, 00225 0x1D171, 00226 0x1D172, 00227 /* 218 # Mn IDEOGRAPHIC LEVEL TONE MARK */ 00228 0x302A, 00229 /* 220 # Mn [4] COMBINING GRAVE ACCENT BELOW.. 00230 * ..COMBINING RIGHT TACK BELOW */ 00231 0x0316, 00232 0x0317, 00233 0x0318, 00234 0x0319, 00235 /* 220 # Mn [5] COMBINING LEFT HALF RING BELOW.. 00236 * ..COMBINING MINUS SIGN BELOW */ 00237 0x031C, 00238 0x031D, 00239 0x031E, 00240 0x031F, 00241 0x0320, 00242 /* 220 # Mn [4] COMBINING DOT BELOW.. 00243 * ..COMBINING COMMA BELOW */ 00244 0x0323, 00245 0x0324, 00246 0x0325, 00247 0x0326, 00248 /* 220 # Mn [11] COMBINING VERTICAL LINE BELOW.. 00249 * ..COMBINING DOUBLE LOW LINE */ 00250 0x0329, 00251 0x032A, 00252 0x032B, 00253 0x032C, 00254 0x032D, 00255 0x032E, 00256 0x032F, 00257 0x0330, 00258 0x0331, 00259 0x0332, 00260 0x0333, 00261 /* 220 # Mn [4] COMBINING RIGHT HALF RING BELOW.. 00262 * ..COMBINING SEAGULL BELOW */ 00263 0x0339, 00264 0x033A, 00265 0x033B, 00266 0x033C, 00267 /* 220 # Mn [3] COMBINING EQUALS SIGN BELOW.. 00268 * ..COMBINING LEFT ANGLE BELOW */ 00269 0x0347, 00270 0x0348, 00271 0x0349, 00272 /* 220 # Mn [2] COMBINING LEFT RIGHT ARROW BELOW.. 00273 * ..COMBINING UPWARDS ARROW BELOW */ 00274 0x034D, 00275 0x034E, 00276 /* 220 # Mn HEBREW ACCENT ETNAHTA */ 00277 0x0591, 00278 /* 220 # Mn HEBREW ACCENT TIPEHA */ 00279 0x0596, 00280 /* 220 # Mn HEBREW ACCENT TEVIR */ 00281 0x059B, 00282 /* 220 # Mn [5] HEBREW ACCENT MUNAH.. 00283 * ..HEBREW ACCENT DARGA */ 00284 0x05A3, 00285 0x05A4, 00286 0x05A5, 00287 0x05A6, 00288 0x05A7, 00289 /* 220 # Mn HEBREW ACCENT YERAH BEN YOMO */ 00290 0x05AA, 00291 /* 220 # Mn ARABIC HAMZA BELOW */ 00292 0x0655, 00293 /* 220 # Mn ARABIC SMALL LOW SEEN */ 00294 0x06E3, 00295 /* 220 # Mn ARABIC EMPTY CENTRE LOW STOP */ 00296 0x06EA, 00297 /* 220 # Mn ARABIC SMALL LOW MEEM */ 00298 0x06ED, 00299 /* 220 # Mn SYRIAC PTHAHA BELOW */ 00300 0x0731, 00301 /* 220 # Mn SYRIAC ZQAPHA BELOW */ 00302 0x0734, 00303 /* 220 # Mn [3] SYRIAC RBASA BELOW.. 00304 * ..SYRIAC DOTTED ZLAMA ANGULAR */ 00305 0x0737, 00306 0x0738, 00307 0x0739, 00308 /* 220 # Mn [2] SYRIAC HBASA BELOW.. 00309 * ..SYRIAC HBASA-ESASA DOTTED */ 00310 0x073B, 00311 0x073C, 00312 /* 220 # Mn SYRIAC ESASA BELOW */ 00313 0x073E, 00314 /* 220 # Mn SYRIAC RUKKAKHA */ 00315 0x0742, 00316 /* 220 # Mn SYRIAC TWO VERTICAL DOTS BELOW */ 00317 0x0744, 00318 /* 220 # Mn SYRIAC THREE DOTS BELOW */ 00319 0x0746, 00320 /* 220 # Mn SYRIAC OBLIQUE LINE BELOW */ 00321 0x0748, 00322 /* 220 # Mn DEVANAGARI STRESS SIGN ANUDATTA */ 00323 0x0952, 00324 /* 220 # Mn [2] TIBETAN ASTROLOGICAL SIGN -KHYUD PA.. 00325 * ..TIBETAN ASTROLOGICAL SIGN SDONG TSHUGS */ 00326 0x0F18, 00327 0x0F19, 00328 /* 220 # Mn TIBETAN MARK NGAS BZUNG NYI ZLA */ 00329 0x0F35, 00330 /* 220 # Mn TIBETAN MARK NGAS BZUNG SGOR RTAGS */ 00331 0x0F37, 00332 /* 220 # Mn TIBETAN SYMBOL PADMA GDAN */ 00333 0x0FC6, 00334 /* 220 # Mn COMBINING TRIPLE UNDERDOT */ 00335 0x20E8, 00336 /* 220 # Mn [8] MUSICAL SYMBOL COMBINING ACCENT.. 00337 * ..MUSICAL SYMBOL COMBINING LOURE */ 00338 0x1D17B, 00339 0x1D17C, 00340 0x1D17D, 00341 0x1D17E, 00342 0x1D17F, 00343 0x1D180, 00344 0x1D181, 00345 0x1D182, 00346 /* 220 # Mn [2] MUSICAL SYMBOL COMBINING DOUBLE TONGUE.. 00347 * ..MUSICAL SYMBOL COMBINING TRIPLE TONGUE */ 00348 0x1D18A, 00349 0x1D18B, 00350 /* 222 # Mn HEBREW ACCENT YETIV */ 00351 0x059A, 00352 /* 222 # Mn HEBREW ACCENT DEHI */ 00353 0x05AD, 00354 /* 222 # Mn IDEOGRAPHIC ENTERING TONE MARK */ 00355 0x302D, 00356 /* 224 # Mn [2] HANGUL SINGLE DOT TONE MARK.. 00357 * ..HANGUL DOUBLE DOT TONE MARK */ 00358 0x302E, 00359 0x302F, 00360 /* 226 # Mc MUSICAL SYMBOL COMBINING AUGMENTATION DOT */ 00361 0x1D16D, 00362 /* 228 # Mn HEBREW ACCENT ZINOR */ 00363 0x05AE, 00364 /* 228 # Mn MONGOLIAN LETTER ALI GALI DAGALGA */ 00365 0x18A9, 00366 /* 228 # Mn IDEOGRAPHIC RISING TONE MARK */ 00367 0x302B, 00368 /* 230 # Mn [21] COMBINING GRAVE ACCENT.. 00369 * ..COMBINING REVERSED COMMA ABOVE */ 00370 0x0300, 00371 0x0301, 00372 0x0302, 00373 0x0303, 00374 0x0304, 00375 0x0305, 00376 0x0306, 00377 0x0307, 00378 0x0308, 00379 0x0309, 00380 0x030A, 00381 0x030B, 00382 0x030C, 00383 0x030D, 00384 0x030E, 00385 0x030F, 00386 0x0310, 00387 0x0311, 00388 0x0312, 00389 0x0313, 00390 0x0314, 00391 /* 230 # Mn [8] COMBINING X ABOVE.. 00392 * ..COMBINING GREEK DIALYTIKA TONOS */ 00393 0x033D, 00394 0x033E, 00395 0x033F, 00396 0x0340, 00397 0x0341, 00398 0x0342, 00399 0x0343, 00400 0x0344, 00401 /* 230 # Mn COMBINING BRIDGE ABOVE */ 00402 0x0346, 00403 /* 230 # Mn [3] COMBINING NOT TILDE ABOVE.. 00404 * ..COMBINING ALMOST EQUAL TO ABOVE */ 00405 0x034A, 00406 0x034B, 00407 0x034C, 00408 /* 230 # Mn [13] COMBINING LATIN SMALL LETTER A.. 00409 * ..COMBINING LATIN SMALL LETTER X */ 00410 0x0363, 00411 0x0364, 00412 0x0365, 00413 0x0366, 00414 0x0367, 00415 0x0368, 00416 0x0369, 00417 0x036A, 00418 0x036B, 00419 0x036C, 00420 0x036D, 00421 0x036E, 00422 0x036F, 00423 /* 230 # Mn [4] COMBINING CYRILLIC TITLO.. 00424 * ..COMBINING CYRILLIC PSILI PNEUMATA */ 00425 0x0483, 00426 0x0484, 00427 0x0485, 00428 0x0486, 00429 /* 230 # Mn [4] HEBREW ACCENT SEGOL.. 00430 * ..HEBREW ACCENT ZAQEF GADOL */ 00431 0x0592, 00432 0x0593, 00433 0x0594, 00434 0x0595, 00435 /* 230 # Mn [3] HEBREW ACCENT REVIA.. 00436 * ..HEBREW ACCENT PASHTA */ 00437 0x0597, 00438 0x0598, 00439 0x0599, 00440 /* 230 # Mn [6] HEBREW ACCENT GERESH.. 00441 * ..HEBREW ACCENT PAZER */ 00442 0x059C, 00443 0x059D, 00444 0x059E, 00445 0x059F, 00446 0x05A0, 00447 0x05A1, 00448 /* 230 # Mn [2] HEBREW ACCENT QADMA.. 00449 * ..HEBREW ACCENT TELISHA QETANA */ 00450 0x05A8, 00451 0x05A9, 00452 /* 230 # Mn [2] HEBREW ACCENT OLE.. 00453 * ..HEBREW ACCENT ILUY */ 00454 0x05AB, 00455 0x05AC, 00456 /* 230 # Mn HEBREW MARK MASORA CIRCLE */ 00457 0x05AF, 00458 /* 230 # Mn HEBREW MARK UPPER DOT */ 00459 0x05C4, 00460 /* 230 # Mn [2] ARABIC MADDAH ABOVE.. 00461 * ..ARABIC HAMZA ABOVE */ 00462 0x0653, 00463 0x0654, 00464 /* 230 # Mn [7] ARABIC SMALL HIGH LIGATURE SAD WITH LAM WITH ALEF MAKSURA.. 00465 * ..ARABIC SMALL HIGH SEEN */ 00466 0x06D6, 00467 0x06D7, 00468 0x06D8, 00469 0x06D9, 00470 0x06DA, 00471 0x06DB, 00472 0x06DC, 00473 /* 230 # Mn [4] ARABIC SMALL HIGH ROUNDED ZERO.. 00474 * ..ARABIC SMALL HIGH MEEM ISOLATED FORM */ 00475 0x06DF, 00476 0x06E0, 00477 0x06E1, 00478 0x06E2, 00479 /* 230 # Mn ARABIC SMALL HIGH MADDA */ 00480 0x06E4, 00481 /* 230 # Mn [2] ARABIC SMALL HIGH YEH.. 00482 * ..ARABIC SMALL HIGH NOON */ 00483 0x06E7, 00484 0x06E8, 00485 /* 230 # Mn [2] ARABIC EMPTY CENTRE HIGH STOP.. 00486 * ..ARABIC ROUNDED HIGH STOP WITH FILLED CENTRE */ 00487 0x06EB, 00488 0x06EC, 00489 /* 230 # Mn SYRIAC PTHAHA ABOVE */ 00490 0x0730, 00491 /* 230 # Mn [2] SYRIAC PTHAHA DOTTED.. 00492 * ..SYRIAC ZQAPHA ABOVE */ 00493 0x0732, 00494 0x0733, 00495 /* 230 # Mn [2] SYRIAC ZQAPHA DOTTED.. 00496 * ..SYRIAC RBASA ABOVE */ 00497 0x0735, 00498 0x0736, 00499 /* 230 # Mn SYRIAC HBASA ABOVE */ 00500 0x073A, 00501 /* 230 # Mn SYRIAC ESASA ABOVE */ 00502 0x073D, 00503 /* 230 # Mn [3] SYRIAC RWAHA.. 00504 * ..SYRIAC QUSHSHAYA */ 00505 0x073F, 00506 0x0740, 00507 0x0741, 00508 /* 230 # Mn SYRIAC TWO VERTICAL DOTS ABOVE */ 00509 0x0743, 00510 /* 230 # Mn SYRIAC THREE DOTS ABOVE */ 00511 0x0745, 00512 /* 230 # Mn SYRIAC OBLIQUE LINE ABOVE */ 00513 0x0747, 00514 /* 230 # Mn [2] SYRIAC MUSIC.. 00515 * ..SYRIAC BARREKH */ 00516 0x0749, 00517 0x074A, 00518 /* 230 # Mn DEVANAGARI STRESS SIGN UDATTA */ 00519 0x0951, 00520 /* 230 # Mn [2] DEVANAGARI GRAVE ACCENT.. 00521 * ..DEVANAGARI ACUTE ACCENT */ 00522 0x0953, 00523 0x0954, 00524 /* 230 # Mn [2] TIBETAN SIGN NYI ZLA NAA DA.. 00525 * ..TIBETAN SIGN SNA LDAN */ 00526 0x0F82, 00527 0x0F83, 00528 /* 230 # Mn [2] TIBETAN SIGN LCI RTAGS.. 00529 * ..TIBETAN SIGN YANG RTAGS */ 00530 0x0F86, 00531 0x0F87, 00532 /* 230 # Mn [2] COMBINING LEFT HARPOON ABOVE.. 00533 * ..COMBINING RIGHT HARPOON ABOVE */ 00534 0x20D0, 00535 0x20D1, 00536 /* 230 # Mn [4] COMBINING ANTICLOCKWISE ARROW ABOVE.. 00537 * ..COMBINING RIGHT ARROW ABOVE */ 00538 0x20D4, 00539 0x20D5, 00540 0x20D6, 00541 0x20D7, 00542 /* 230 # Mn [2] COMBINING THREE DOTS ABOVE.. 00543 * ..COMBINING FOUR DOTS ABOVE */ 00544 0x20DB, 00545 0x20DC, 00546 /* 230 # Mn COMBINING LEFT RIGHT ARROW ABOVE */ 00547 0x20E1, 00548 /* 230 # Mn COMBINING ANNUITY SYMBOL */ 00549 0x20E7, 00550 /* 230 # Mn COMBINING WIDE BRIDGE ABOVE */ 00551 0x20E9, 00552 /* 230 # Mn [4] COMBINING LIGATURE LEFT HALF.. 00553 * ..COMBINING DOUBLE TILDE RIGHT HALF */ 00554 0xFE20, 00555 0xFE21, 00556 0xFE22, 00557 0xFE23, 00558 /* 230 # Mn [5] MUSICAL SYMBOL COMBINING DOIT.. 00559 * ..MUSICAL SYMBOL COMBINING BEND */ 00560 0x1D185, 00561 0x1D186, 00562 0x1D187, 00563 0x1D188, 00564 0x1D189, 00565 /* 230 # Mn [4] MUSICAL SYMBOL COMBINING DOWN BOW.. 00566 * ..MUSICAL SYMBOL COMBINING SNAP PIZZICATO */ 00567 0x1D1AA, 00568 0x1D1AB, 00569 0x1D1AC, 00570 0x1D1AD, 00571 /* 232 # Mn COMBINING COMMA ABOVE RIGHT */ 00572 0x0315, 00573 /* 232 # Mn COMBINING LEFT ANGLE ABOVE */ 00574 0x031A, 00575 /* 232 # Mn IDEOGRAPHIC DEPARTING TONE MARK */ 00576 0x302C, 00577 /* 233 # Mn COMBINING DOUBLE RIGHTWARDS ARROW BELOW */ 00578 0x0362, 00579 /* 234 # Mn [2] COMBINING DOUBLE TILDE.. 00580 * ..COMBINING DOUBLE INVERTED BREVE */ 00581 0x0360, 00582 0x0361, 00583 /* 240 # Mn COMBINING GREEK YPOGEGRAMMENI */ 00584 0x0345, 00585 0 00586 }; 00587 00588 /* 00589 * 09C7 BENGALI VOWEL SIGN E 09BE BENGALI VOWEL SIGN AA or 00590 * 09D7 BENGALI AU LENGTH MARK 00591 */ 00592 00593 static const uint32_t pr29_1_1[] = { 00594 0x09C7, 0 00595 }; 00596 00597 static const uint32_t pr29_1_2[] = { 00598 0x09BE, 0x09D7, 0 00599 }; 00600 00601 /* 00602 * 0B47 ORIYA VOWEL SIGN E 0B3E ORIYA VOWEL SIGN AA or 00603 * 0B56 ORIYA AI LENGTH MARK or 00604 * 0B57 ORIYA AU LENGTH MARK 00605 */ 00606 00607 static const uint32_t pr29_2_1[] = { 00608 0x0B47, 0 00609 }; 00610 00611 static const uint32_t pr29_2_2[] = { 00612 0x0B3E, 0x0B56, 0x0B57, 0 00613 }; 00614 00615 /* 00616 * 0BC6 TAMIL VOWEL SIGN E 0BBE TAMIL VOWEL SIGN AA or 00617 * 0BD7 TAMIL AU LENGTH MARK 00618 */ 00619 00620 static const uint32_t pr29_3_1[] = { 00621 0x0BC6, 0 00622 }; 00623 00624 static const uint32_t pr29_3_2[] = { 00625 0x0BBE, 0x0BD7, 0 00626 }; 00627 00628 /* 00629 * 0BC7 TAMIL VOWEL SIGN EE 0BBE TAMIL VOWEL SIGN AA 00630 */ 00631 00632 static const uint32_t pr29_4_1[] = { 00633 0x0BC7, 0 00634 }; 00635 00636 static const uint32_t pr29_4_2[] = { 00637 0x0BBE, 0 00638 }; 00639 00640 /* 00641 * 0B92 TAMIL LETTER O 0BD7 TAMIL AU LENGTH MARK 00642 */ 00643 00644 static const uint32_t pr29_5_1[] = { 00645 0x0B92, 0 00646 }; 00647 00648 static const uint32_t pr29_5_2[] = { 00649 0x0BD7, 0 00650 }; 00651 00652 /* 00653 * 0CC6 KANNADA VOWEL SIGN E 0CC2 KANNADA VOWEL SIGN UU or 00654 * 0CD5 KANNADA LENGTH MARK or 00655 * 0CD6 KANNADA AI LENGTH MARK 00656 */ 00657 00658 static const uint32_t pr29_6_1[] = { 00659 0x0CC6, 0 00660 }; 00661 00662 static const uint32_t pr29_6_2[] = { 00663 0x0CC2, 0xCD5, 0xCD6, 0 00664 }; 00665 00666 /* 00667 * 0CBF KANNADA VOWEL SIGN I or 00668 * 0CCA KANNADA VOWEL SIGN O 0CD5 KANNADA LENGTH MARK 00669 */ 00670 00671 static const uint32_t pr29_7_1[] = { 00672 0x0CBF, 0xCCA, 0 00673 }; 00674 00675 static const uint32_t pr29_7_2[] = { 00676 0x0CD5, 0 00677 }; 00678 00679 /* 00680 * 0D47 MALAYALAM VOWEL SIGN EE 0D3E MALAYALAM VOWEL SIGN AA 00681 */ 00682 00683 static const uint32_t pr29_8_1[] = { 00684 0x0D47, 0 00685 }; 00686 00687 static const uint32_t pr29_8_2[] = { 00688 0x0D3E, 0 00689 }; 00690 00691 /* 00692 * 0D46 MALAYALAM VOWEL SIGN E 0D3E MALAYALAM VOWEL SIGN AA or 00693 * 0D57 MALAYALAM AU LENGTH MARK 00694 */ 00695 00696 static const uint32_t pr29_9_1[] = { 00697 0x0D46, 0 00698 }; 00699 00700 static const uint32_t pr29_9_2[] = { 00701 0x0D3E, 0x0D57, 0 00702 }; 00703 00704 /* 00705 * 1025 MYANMAR LETTER U 102E MYANMAR VOWEL SIGN II 00706 */ 00707 00708 static const uint32_t pr29_10_1[] = { 00709 0x1025, 0 00710 }; 00711 00712 static const uint32_t pr29_10_2[] = { 00713 0x102E, 0 00714 }; 00715 00716 /* 00717 * 0DD9 SINHALA VOWEL SIGN KOMBUVA 0DCF SINHALA VOWEL SIGN AELA-PILLA or 00718 * 0DDF SINHALA VOWEL SIGN GAYANUKITTA 00719 */ 00720 00721 static const uint32_t pr29_11_1[] = { 00722 0x0DD9, 0 00723 }; 00724 00725 static const uint32_t pr29_11_2[] = { 00726 0x0DCF, 0x0DDF, 0 00727 }; 00728 00729 /* 00730 * 1100..1112 HANGUL CHOSEONG KIYEOK..HIEUH [19 instances] 00731 * 1161..1175 HANGUL JUNGSEONG A..I [21 instances] 00732 */ 00733 00734 static const uint32_t pr29_12_1[] = { 00735 0x1100, 0x1101, 0x1102, 0x1103, 0x1104, 0x1105, 0x1106, 0x1107, 00736 0x1108, 0x1109, 0x110A, 0x110B, 0x110C, 0x110D, 0x110E, 0x110F, 00737 0x1110, 0x1111, 0x1112, 0 00738 }; 00739 00740 static const uint32_t pr29_12_2[] = { 00741 0x1161, 0x1162, 0x1163, 0x1164, 0x1165, 0x1166, 0x1167, 0x1168, 00742 0x1169, 0x116A, 0x116B, 0x116C, 0x116D, 0x116E, 0x116F, 0x1170, 00743 0x1171, 0x1172, 0x1173, 0x1174, 0x1175, 0 00744 }; 00745 00746 00747 /* 00748 * [:HangulSyllableType=LV:] 00749 * 11A8..11C2 HANGUL JONGSEONG KIYEOK..HIEUH [27 instances] 00750 */ 00751 00752 static const uint32_t pr29_13_1[] = { 00753 0xAC00, /* LV # Lo HANGUL SYLLABLE GA */ 00754 0xAC1C, /* LV # Lo HANGUL SYLLABLE GAE */ 00755 0xAC38, /* LV # Lo HANGUL SYLLABLE GYA */ 00756 0xAC54, /* LV # Lo HANGUL SYLLABLE GYAE */ 00757 0xAC70, /* LV # Lo HANGUL SYLLABLE GEO */ 00758 0xAC8C, /* LV # Lo HANGUL SYLLABLE GE */ 00759 0xACA8, /* LV # Lo HANGUL SYLLABLE GYEO */ 00760 0xACC4, /* LV # Lo HANGUL SYLLABLE GYE */ 00761 0xACE0, /* LV # Lo HANGUL SYLLABLE GO */ 00762 0xACFC, /* LV # Lo HANGUL SYLLABLE GWA */ 00763 0xAD18, /* LV # Lo HANGUL SYLLABLE GWAE */ 00764 0xAD34, /* LV # Lo HANGUL SYLLABLE GOE */ 00765 0xAD50, /* LV # Lo HANGUL SYLLABLE GYO */ 00766 0xAD6C, /* LV # Lo HANGUL SYLLABLE GU */ 00767 0xAD88, /* LV # Lo HANGUL SYLLABLE GWEO */ 00768 0xADA4, /* LV # Lo HANGUL SYLLABLE GWE */ 00769 0xADC0, /* LV # Lo HANGUL SYLLABLE GWI */ 00770 0xADDC, /* LV # Lo HANGUL SYLLABLE GYU */ 00771 0xADF8, /* LV # Lo HANGUL SYLLABLE GEU */ 00772 0xAE14, /* LV # Lo HANGUL SYLLABLE GYI */ 00773 0xAE30, /* LV # Lo HANGUL SYLLABLE GI */ 00774 0xAE4C, /* LV # Lo HANGUL SYLLABLE GGA */ 00775 0xAE68, /* LV # Lo HANGUL SYLLABLE GGAE */ 00776 0xAE84, /* LV # Lo HANGUL SYLLABLE GGYA */ 00777 0xAEA0, /* LV # Lo HANGUL SYLLABLE GGYAE */ 00778 0xAEBC, /* LV # Lo HANGUL SYLLABLE GGEO */ 00779 0xAED8, /* LV # Lo HANGUL SYLLABLE GGE */ 00780 0xAEF4, /* LV # Lo HANGUL SYLLABLE GGYEO */ 00781 0xAF10, /* LV # Lo HANGUL SYLLABLE GGYE */ 00782 0xAF2C, /* LV # Lo HANGUL SYLLABLE GGO */ 00783 0xAF48, /* LV # Lo HANGUL SYLLABLE GGWA */ 00784 0xAF64, /* LV # Lo HANGUL SYLLABLE GGWAE */ 00785 0xAF80, /* LV # Lo HANGUL SYLLABLE GGOE */ 00786 0xAF9C, /* LV # Lo HANGUL SYLLABLE GGYO */ 00787 0xAFB8, /* LV # Lo HANGUL SYLLABLE GGU */ 00788 0xAFD4, /* LV # Lo HANGUL SYLLABLE GGWEO */ 00789 0xAFF0, /* LV # Lo HANGUL SYLLABLE GGWE */ 00790 0xB00C, /* LV # Lo HANGUL SYLLABLE GGWI */ 00791 0xB028, /* LV # Lo HANGUL SYLLABLE GGYU */ 00792 0xB044, /* LV # Lo HANGUL SYLLABLE GGEU */ 00793 0xB060, /* LV # Lo HANGUL SYLLABLE GGYI */ 00794 0xB07C, /* LV # Lo HANGUL SYLLABLE GGI */ 00795 0xB098, /* LV # Lo HANGUL SYLLABLE NA */ 00796 0xB0B4, /* LV # Lo HANGUL SYLLABLE NAE */ 00797 0xB0D0, /* LV # Lo HANGUL SYLLABLE NYA */ 00798 0xB0EC, /* LV # Lo HANGUL SYLLABLE NYAE */ 00799 0xB108, /* LV # Lo HANGUL SYLLABLE NEO */ 00800 0xB124, /* LV # Lo HANGUL SYLLABLE NE */ 00801 0xB140, /* LV # Lo HANGUL SYLLABLE NYEO */ 00802 0xB15C, /* LV # Lo HANGUL SYLLABLE NYE */ 00803 0xB178, /* LV # Lo HANGUL SYLLABLE NO */ 00804 0xB194, /* LV # Lo HANGUL SYLLABLE NWA */ 00805 0xB1B0, /* LV # Lo HANGUL SYLLABLE NWAE */ 00806 0xB1CC, /* LV # Lo HANGUL SYLLABLE NOE */ 00807 0xB1E8, /* LV # Lo HANGUL SYLLABLE NYO */ 00808 0xB204, /* LV # Lo HANGUL SYLLABLE NU */ 00809 0xB220, /* LV # Lo HANGUL SYLLABLE NWEO */ 00810 0xB23C, /* LV # Lo HANGUL SYLLABLE NWE */ 00811 0xB258, /* LV # Lo HANGUL SYLLABLE NWI */ 00812 0xB274, /* LV # Lo HANGUL SYLLABLE NYU */ 00813 0xB290, /* LV # Lo HANGUL SYLLABLE NEU */ 00814 0xB2AC, /* LV # Lo HANGUL SYLLABLE NYI */ 00815 0xB2C8, /* LV # Lo HANGUL SYLLABLE NI */ 00816 0xB2E4, /* LV # Lo HANGUL SYLLABLE DA */ 00817 0xB300, /* LV # Lo HANGUL SYLLABLE DAE */ 00818 0xB31C, /* LV # Lo HANGUL SYLLABLE DYA */ 00819 0xB338, /* LV # Lo HANGUL SYLLABLE DYAE */ 00820 0xB354, /* LV # Lo HANGUL SYLLABLE DEO */ 00821 0xB370, /* LV # Lo HANGUL SYLLABLE DE */ 00822 0xB38C, /* LV # Lo HANGUL SYLLABLE DYEO */ 00823 0xB3A8, /* LV # Lo HANGUL SYLLABLE DYE */ 00824 0xB3C4, /* LV # Lo HANGUL SYLLABLE DO */ 00825 0xB3E0, /* LV # Lo HANGUL SYLLABLE DWA */ 00826 0xB3FC, /* LV # Lo HANGUL SYLLABLE DWAE */ 00827 0xB418, /* LV # Lo HANGUL SYLLABLE DOE */ 00828 0xB434, /* LV # Lo HANGUL SYLLABLE DYO */ 00829 0xB450, /* LV # Lo HANGUL SYLLABLE DU */ 00830 0xB46C, /* LV # Lo HANGUL SYLLABLE DWEO */ 00831 0xB488, /* LV # Lo HANGUL SYLLABLE DWE */ 00832 0xB4A4, /* LV # Lo HANGUL SYLLABLE DWI */ 00833 0xB4C0, /* LV # Lo HANGUL SYLLABLE DYU */ 00834 0xB4DC, /* LV # Lo HANGUL SYLLABLE DEU */ 00835 0xB4F8, /* LV # Lo HANGUL SYLLABLE DYI */ 00836 0xB514, /* LV # Lo HANGUL SYLLABLE DI */ 00837 0xB530, /* LV # Lo HANGUL SYLLABLE DDA */ 00838 0xB54C, /* LV # Lo HANGUL SYLLABLE DDAE */ 00839 0xB568, /* LV # Lo HANGUL SYLLABLE DDYA */ 00840 0xB584, /* LV # Lo HANGUL SYLLABLE DDYAE */ 00841 0xB5A0, /* LV # Lo HANGUL SYLLABLE DDEO */ 00842 0xB5BC, /* LV # Lo HANGUL SYLLABLE DDE */ 00843 0xB5D8, /* LV # Lo HANGUL SYLLABLE DDYEO */ 00844 0xB5F4, /* LV # Lo HANGUL SYLLABLE DDYE */ 00845 0xB610, /* LV # Lo HANGUL SYLLABLE DDO */ 00846 0xB62C, /* LV # Lo HANGUL SYLLABLE DDWA */ 00847 0xB648, /* LV # Lo HANGUL SYLLABLE DDWAE */ 00848 0xB664, /* LV # Lo HANGUL SYLLABLE DDOE */ 00849 0xB680, /* LV # Lo HANGUL SYLLABLE DDYO */ 00850 0xB69C, /* LV # Lo HANGUL SYLLABLE DDU */ 00851 0xB6B8, /* LV # Lo HANGUL SYLLABLE DDWEO */ 00852 0xB6D4, /* LV # Lo HANGUL SYLLABLE DDWE */ 00853 0xB6F0, /* LV # Lo HANGUL SYLLABLE DDWI */ 00854 0xB70C, /* LV # Lo HANGUL SYLLABLE DDYU */ 00855 0xB728, /* LV # Lo HANGUL SYLLABLE DDEU */ 00856 0xB744, /* LV # Lo HANGUL SYLLABLE DDYI */ 00857 0xB760, /* LV # Lo HANGUL SYLLABLE DDI */ 00858 0xB77C, /* LV # Lo HANGUL SYLLABLE RA */ 00859 0xB798, /* LV # Lo HANGUL SYLLABLE RAE */ 00860 0xB7B4, /* LV # Lo HANGUL SYLLABLE RYA */ 00861 0xB7D0, /* LV # Lo HANGUL SYLLABLE RYAE */ 00862 0xB7EC, /* LV # Lo HANGUL SYLLABLE REO */ 00863 0xB808, /* LV # Lo HANGUL SYLLABLE RE */ 00864 0xB824, /* LV # Lo HANGUL SYLLABLE RYEO */ 00865 0xB840, /* LV # Lo HANGUL SYLLABLE RYE */ 00866 0xB85C, /* LV # Lo HANGUL SYLLABLE RO */ 00867 0xB878, /* LV # Lo HANGUL SYLLABLE RWA */ 00868 0xB894, /* LV # Lo HANGUL SYLLABLE RWAE */ 00869 0xB8B0, /* LV # Lo HANGUL SYLLABLE ROE */ 00870 0xB8CC, /* LV # Lo HANGUL SYLLABLE RYO */ 00871 0xB8E8, /* LV # Lo HANGUL SYLLABLE RU */ 00872 0xB904, /* LV # Lo HANGUL SYLLABLE RWEO */ 00873 0xB920, /* LV # Lo HANGUL SYLLABLE RWE */ 00874 0xB93C, /* LV # Lo HANGUL SYLLABLE RWI */ 00875 0xB958, /* LV # Lo HANGUL SYLLABLE RYU */ 00876 0xB974, /* LV # Lo HANGUL SYLLABLE REU */ 00877 0xB990, /* LV # Lo HANGUL SYLLABLE RYI */ 00878 0xB9AC, /* LV # Lo HANGUL SYLLABLE RI */ 00879 0xB9C8, /* LV # Lo HANGUL SYLLABLE MA */ 00880 0xB9E4, /* LV # Lo HANGUL SYLLABLE MAE */ 00881 0xBA00, /* LV # Lo HANGUL SYLLABLE MYA */ 00882 0xBA1C, /* LV # Lo HANGUL SYLLABLE MYAE */ 00883 0xBA38, /* LV # Lo HANGUL SYLLABLE MEO */ 00884 0xBA54, /* LV # Lo HANGUL SYLLABLE ME */ 00885 0xBA70, /* LV # Lo HANGUL SYLLABLE MYEO */ 00886 0xBA8C, /* LV # Lo HANGUL SYLLABLE MYE */ 00887 0xBAA8, /* LV # Lo HANGUL SYLLABLE MO */ 00888 0xBAC4, /* LV # Lo HANGUL SYLLABLE MWA */ 00889 0xBAE0, /* LV # Lo HANGUL SYLLABLE MWAE */ 00890 0xBAFC, /* LV # Lo HANGUL SYLLABLE MOE */ 00891 0xBB18, /* LV # Lo HANGUL SYLLABLE MYO */ 00892 0xBB34, /* LV # Lo HANGUL SYLLABLE MU */ 00893 0xBB50, /* LV # Lo HANGUL SYLLABLE MWEO */ 00894 0xBB6C, /* LV # Lo HANGUL SYLLABLE MWE */ 00895 0xBB88, /* LV # Lo HANGUL SYLLABLE MWI */ 00896 0xBBA4, /* LV # Lo HANGUL SYLLABLE MYU */ 00897 0xBBC0, /* LV # Lo HANGUL SYLLABLE MEU */ 00898 0xBBDC, /* LV # Lo HANGUL SYLLABLE MYI */ 00899 0xBBF8, /* LV # Lo HANGUL SYLLABLE MI */ 00900 0xBC14, /* LV # Lo HANGUL SYLLABLE BA */ 00901 0xBC30, /* LV # Lo HANGUL SYLLABLE BAE */ 00902 0xBC4C, /* LV # Lo HANGUL SYLLABLE BYA */ 00903 0xBC68, /* LV # Lo HANGUL SYLLABLE BYAE */ 00904 0xBC84, /* LV # Lo HANGUL SYLLABLE BEO */ 00905 0xBCA0, /* LV # Lo HANGUL SYLLABLE BE */ 00906 0xBCBC, /* LV # Lo HANGUL SYLLABLE BYEO */ 00907 0xBCD8, /* LV # Lo HANGUL SYLLABLE BYE */ 00908 0xBCF4, /* LV # Lo HANGUL SYLLABLE BO */ 00909 0xBD10, /* LV # Lo HANGUL SYLLABLE BWA */ 00910 0xBD2C, /* LV # Lo HANGUL SYLLABLE BWAE */ 00911 0xBD48, /* LV # Lo HANGUL SYLLABLE BOE */ 00912 0xBD64, /* LV # Lo HANGUL SYLLABLE BYO */ 00913 0xBD80, /* LV # Lo HANGUL SYLLABLE BU */ 00914 0xBD9C, /* LV # Lo HANGUL SYLLABLE BWEO */ 00915 0xBDB8, /* LV # Lo HANGUL SYLLABLE BWE */ 00916 0xBDD4, /* LV # Lo HANGUL SYLLABLE BWI */ 00917 0xBDF0, /* LV # Lo HANGUL SYLLABLE BYU */ 00918 0xBE0C, /* LV # Lo HANGUL SYLLABLE BEU */ 00919 0xBE28, /* LV # Lo HANGUL SYLLABLE BYI */ 00920 0xBE44, /* LV # Lo HANGUL SYLLABLE BI */ 00921 0xBE60, /* LV # Lo HANGUL SYLLABLE BBA */ 00922 0xBE7C, /* LV # Lo HANGUL SYLLABLE BBAE */ 00923 0xBE98, /* LV # Lo HANGUL SYLLABLE BBYA */ 00924 0xBEB4, /* LV # Lo HANGUL SYLLABLE BBYAE */ 00925 0xBED0, /* LV # Lo HANGUL SYLLABLE BBEO */ 00926 0xBEEC, /* LV # Lo HANGUL SYLLABLE BBE */ 00927 0xBF08, /* LV # Lo HANGUL SYLLABLE BBYEO */ 00928 0xBF24, /* LV # Lo HANGUL SYLLABLE BBYE */ 00929 0xBF40, /* LV # Lo HANGUL SYLLABLE BBO */ 00930 0xBF5C, /* LV # Lo HANGUL SYLLABLE BBWA */ 00931 0xBF78, /* LV # Lo HANGUL SYLLABLE BBWAE */ 00932 0xBF94, /* LV # Lo HANGUL SYLLABLE BBOE */ 00933 0xBFB0, /* LV # Lo HANGUL SYLLABLE BBYO */ 00934 0xBFCC, /* LV # Lo HANGUL SYLLABLE BBU */ 00935 0xBFE8, /* LV # Lo HANGUL SYLLABLE BBWEO */ 00936 0xC004, /* LV # Lo HANGUL SYLLABLE BBWE */ 00937 0xC020, /* LV # Lo HANGUL SYLLABLE BBWI */ 00938 0xC03C, /* LV # Lo HANGUL SYLLABLE BBYU */ 00939 0xC058, /* LV # Lo HANGUL SYLLABLE BBEU */ 00940 0xC074, /* LV # Lo HANGUL SYLLABLE BBYI */ 00941 0xC090, /* LV # Lo HANGUL SYLLABLE BBI */ 00942 0xC0AC, /* LV # Lo HANGUL SYLLABLE SA */ 00943 0xC0C8, /* LV # Lo HANGUL SYLLABLE SAE */ 00944 0xC0E4, /* LV # Lo HANGUL SYLLABLE SYA */ 00945 0xC100, /* LV # Lo HANGUL SYLLABLE SYAE */ 00946 0xC11C, /* LV # Lo HANGUL SYLLABLE SEO */ 00947 0xC138, /* LV # Lo HANGUL SYLLABLE SE */ 00948 0xC154, /* LV # Lo HANGUL SYLLABLE SYEO */ 00949 0xC170, /* LV # Lo HANGUL SYLLABLE SYE */ 00950 0xC18C, /* LV # Lo HANGUL SYLLABLE SO */ 00951 0xC1A8, /* LV # Lo HANGUL SYLLABLE SWA */ 00952 0xC1C4, /* LV # Lo HANGUL SYLLABLE SWAE */ 00953 0xC1E0, /* LV # Lo HANGUL SYLLABLE SOE */ 00954 0xC1FC, /* LV # Lo HANGUL SYLLABLE SYO */ 00955 0xC218, /* LV # Lo HANGUL SYLLABLE SU */ 00956 0xC234, /* LV # Lo HANGUL SYLLABLE SWEO */ 00957 0xC250, /* LV # Lo HANGUL SYLLABLE SWE */ 00958 0xC26C, /* LV # Lo HANGUL SYLLABLE SWI */ 00959 0xC288, /* LV # Lo HANGUL SYLLABLE SYU */ 00960 0xC2A4, /* LV # Lo HANGUL SYLLABLE SEU */ 00961 0xC2C0, /* LV # Lo HANGUL SYLLABLE SYI */ 00962 0xC2DC, /* LV # Lo HANGUL SYLLABLE SI */ 00963 0xC2F8, /* LV # Lo HANGUL SYLLABLE SSA */ 00964 0xC314, /* LV # Lo HANGUL SYLLABLE SSAE */ 00965 0xC330, /* LV # Lo HANGUL SYLLABLE SSYA */ 00966 0xC34C, /* LV # Lo HANGUL SYLLABLE SSYAE */ 00967 0xC368, /* LV # Lo HANGUL SYLLABLE SSEO */ 00968 0xC384, /* LV # Lo HANGUL SYLLABLE SSE */ 00969 0xC3A0, /* LV # Lo HANGUL SYLLABLE SSYEO */ 00970 0xC3BC, /* LV # Lo HANGUL SYLLABLE SSYE */ 00971 0xC3D8, /* LV # Lo HANGUL SYLLABLE SSO */ 00972 0xC3F4, /* LV # Lo HANGUL SYLLABLE SSWA */ 00973 0xC410, /* LV # Lo HANGUL SYLLABLE SSWAE */ 00974 0xC42C, /* LV # Lo HANGUL SYLLABLE SSOE */ 00975 0xC448, /* LV # Lo HANGUL SYLLABLE SSYO */ 00976 0xC464, /* LV # Lo HANGUL SYLLABLE SSU */ 00977 0xC480, /* LV # Lo HANGUL SYLLABLE SSWEO */ 00978 0xC49C, /* LV # Lo HANGUL SYLLABLE SSWE */ 00979 0xC4B8, /* LV # Lo HANGUL SYLLABLE SSWI */ 00980 0xC4D4, /* LV # Lo HANGUL SYLLABLE SSYU */ 00981 0xC4F0, /* LV # Lo HANGUL SYLLABLE SSEU */ 00982 0xC50C, /* LV # Lo HANGUL SYLLABLE SSYI */ 00983 0xC528, /* LV # Lo HANGUL SYLLABLE SSI */ 00984 0xC544, /* LV # Lo HANGUL SYLLABLE A */ 00985 0xC560, /* LV # Lo HANGUL SYLLABLE AE */ 00986 0xC57C, /* LV # Lo HANGUL SYLLABLE YA */ 00987 0xC598, /* LV # Lo HANGUL SYLLABLE YAE */ 00988 0xC5B4, /* LV # Lo HANGUL SYLLABLE EO */ 00989 0xC5D0, /* LV # Lo HANGUL SYLLABLE E */ 00990 0xC5EC, /* LV # Lo HANGUL SYLLABLE YEO */ 00991 0xC608, /* LV # Lo HANGUL SYLLABLE YE */ 00992 0xC624, /* LV # Lo HANGUL SYLLABLE O */ 00993 0xC640, /* LV # Lo HANGUL SYLLABLE WA */ 00994 0xC65C, /* LV # Lo HANGUL SYLLABLE WAE */ 00995 0xC678, /* LV # Lo HANGUL SYLLABLE OE */ 00996 0xC694, /* LV # Lo HANGUL SYLLABLE YO */ 00997 0xC6B0, /* LV # Lo HANGUL SYLLABLE U */ 00998 0xC6CC, /* LV # Lo HANGUL SYLLABLE WEO */ 00999 0xC6E8, /* LV # Lo HANGUL SYLLABLE WE */ 01000 0xC704, /* LV # Lo HANGUL SYLLABLE WI */ 01001 0xC720, /* LV # Lo HANGUL SYLLABLE YU */ 01002 0xC73C, /* LV # Lo HANGUL SYLLABLE EU */ 01003 0xC758, /* LV # Lo HANGUL SYLLABLE YI */ 01004 0xC774, /* LV # Lo HANGUL SYLLABLE I */ 01005 0xC790, /* LV # Lo HANGUL SYLLABLE JA */ 01006 0xC7AC, /* LV # Lo HANGUL SYLLABLE JAE */ 01007 0xC7C8, /* LV # Lo HANGUL SYLLABLE JYA */ 01008 0xC7E4, /* LV # Lo HANGUL SYLLABLE JYAE */ 01009 0xC800, /* LV # Lo HANGUL SYLLABLE JEO */ 01010 0xC81C, /* LV # Lo HANGUL SYLLABLE JE */ 01011 0xC838, /* LV # Lo HANGUL SYLLABLE JYEO */ 01012 0xC854, /* LV # Lo HANGUL SYLLABLE JYE */ 01013 0xC870, /* LV # Lo HANGUL SYLLABLE JO */ 01014 0xC88C, /* LV # Lo HANGUL SYLLABLE JWA */ 01015 0xC8A8, /* LV # Lo HANGUL SYLLABLE JWAE */ 01016 0xC8C4, /* LV # Lo HANGUL SYLLABLE JOE */ 01017 0xC8E0, /* LV # Lo HANGUL SYLLABLE JYO */ 01018 0xC8FC, /* LV # Lo HANGUL SYLLABLE JU */ 01019 0xC918, /* LV # Lo HANGUL SYLLABLE JWEO */ 01020 0xC934, /* LV # Lo HANGUL SYLLABLE JWE */ 01021 0xC950, /* LV # Lo HANGUL SYLLABLE JWI */ 01022 0xC96C, /* LV # Lo HANGUL SYLLABLE JYU */ 01023 0xC988, /* LV # Lo HANGUL SYLLABLE JEU */ 01024 0xC9A4, /* LV # Lo HANGUL SYLLABLE JYI */ 01025 0xC9C0, /* LV # Lo HANGUL SYLLABLE JI */ 01026 0xC9DC, /* LV # Lo HANGUL SYLLABLE JJA */ 01027 0xC9F8, /* LV # Lo HANGUL SYLLABLE JJAE */ 01028 0xCA14, /* LV # Lo HANGUL SYLLABLE JJYA */ 01029 0xCA30, /* LV # Lo HANGUL SYLLABLE JJYAE */ 01030 0xCA4C, /* LV # Lo HANGUL SYLLABLE JJEO */ 01031 0xCA68, /* LV # Lo HANGUL SYLLABLE JJE */ 01032 0xCA84, /* LV # Lo HANGUL SYLLABLE JJYEO */ 01033 0xCAA0, /* LV # Lo HANGUL SYLLABLE JJYE */ 01034 0xCABC, /* LV # Lo HANGUL SYLLABLE JJO */ 01035 0xCAD8, /* LV # Lo HANGUL SYLLABLE JJWA */ 01036 0xCAF4, /* LV # Lo HANGUL SYLLABLE JJWAE */ 01037 0xCB10, /* LV # Lo HANGUL SYLLABLE JJOE */ 01038 0xCB2C, /* LV # Lo HANGUL SYLLABLE JJYO */ 01039 0xCB48, /* LV # Lo HANGUL SYLLABLE JJU */ 01040 0xCB64, /* LV # Lo HANGUL SYLLABLE JJWEO */ 01041 0xCB80, /* LV # Lo HANGUL SYLLABLE JJWE */ 01042 0xCB9C, /* LV # Lo HANGUL SYLLABLE JJWI */ 01043 0xCBB8, /* LV # Lo HANGUL SYLLABLE JJYU */ 01044 0xCBD4, /* LV # Lo HANGUL SYLLABLE JJEU */ 01045 0xCBF0, /* LV # Lo HANGUL SYLLABLE JJYI */ 01046 0xCC0C, /* LV # Lo HANGUL SYLLABLE JJI */ 01047 0xCC28, /* LV # Lo HANGUL SYLLABLE CA */ 01048 0xCC44, /* LV # Lo HANGUL SYLLABLE CAE */ 01049 0xCC60, /* LV # Lo HANGUL SYLLABLE CYA */ 01050 0xCC7C, /* LV # Lo HANGUL SYLLABLE CYAE */ 01051 0xCC98, /* LV # Lo HANGUL SYLLABLE CEO */ 01052 0xCCB4, /* LV # Lo HANGUL SYLLABLE CE */ 01053 0xCCD0, /* LV # Lo HANGUL SYLLABLE CYEO */ 01054 0xCCEC, /* LV # Lo HANGUL SYLLABLE CYE */ 01055 0xCD08, /* LV # Lo HANGUL SYLLABLE CO */ 01056 0xCD24, /* LV # Lo HANGUL SYLLABLE CWA */ 01057 0xCD40, /* LV # Lo HANGUL SYLLABLE CWAE */ 01058 0xCD5C, /* LV # Lo HANGUL SYLLABLE COE */ 01059 0xCD78, /* LV # Lo HANGUL SYLLABLE CYO */ 01060 0xCD94, /* LV # Lo HANGUL SYLLABLE CU */ 01061 0xCDB0, /* LV # Lo HANGUL SYLLABLE CWEO */ 01062 0xCDCC, /* LV # Lo HANGUL SYLLABLE CWE */ 01063 0xCDE8, /* LV # Lo HANGUL SYLLABLE CWI */ 01064 0xCE04, /* LV # Lo HANGUL SYLLABLE CYU */ 01065 0xCE20, /* LV # Lo HANGUL SYLLABLE CEU */ 01066 0xCE3C, /* LV # Lo HANGUL SYLLABLE CYI */ 01067 0xCE58, /* LV # Lo HANGUL SYLLABLE CI */ 01068 0xCE74, /* LV # Lo HANGUL SYLLABLE KA */ 01069 0xCE90, /* LV # Lo HANGUL SYLLABLE KAE */ 01070 0xCEAC, /* LV # Lo HANGUL SYLLABLE KYA */ 01071 0xCEC8, /* LV # Lo HANGUL SYLLABLE KYAE */ 01072 0xCEE4, /* LV # Lo HANGUL SYLLABLE KEO */ 01073 0xCF00, /* LV # Lo HANGUL SYLLABLE KE */ 01074 0xCF1C, /* LV # Lo HANGUL SYLLABLE KYEO */ 01075 0xCF38, /* LV # Lo HANGUL SYLLABLE KYE */ 01076 0xCF54, /* LV # Lo HANGUL SYLLABLE KO */ 01077 0xCF70, /* LV # Lo HANGUL SYLLABLE KWA */ 01078 0xCF8C, /* LV # Lo HANGUL SYLLABLE KWAE */ 01079 0xCFA8, /* LV # Lo HANGUL SYLLABLE KOE */ 01080 0xCFC4, /* LV # Lo HANGUL SYLLABLE KYO */ 01081 0xCFE0, /* LV # Lo HANGUL SYLLABLE KU */ 01082 0xCFFC, /* LV # Lo HANGUL SYLLABLE KWEO */ 01083 0xD018, /* LV # Lo HANGUL SYLLABLE KWE */ 01084 0xD034, /* LV # Lo HANGUL SYLLABLE KWI */ 01085 0xD050, /* LV # Lo HANGUL SYLLABLE KYU */ 01086 0xD06C, /* LV # Lo HANGUL SYLLABLE KEU */ 01087 0xD088, /* LV # Lo HANGUL SYLLABLE KYI */ 01088 0xD0A4, /* LV # Lo HANGUL SYLLABLE KI */ 01089 0xD0C0, /* LV # Lo HANGUL SYLLABLE TA */ 01090 0xD0DC, /* LV # Lo HANGUL SYLLABLE TAE */ 01091 0xD0F8, /* LV # Lo HANGUL SYLLABLE TYA */ 01092 0xD114, /* LV # Lo HANGUL SYLLABLE TYAE */ 01093 0xD130, /* LV # Lo HANGUL SYLLABLE TEO */ 01094 0xD14C, /* LV # Lo HANGUL SYLLABLE TE */ 01095 0xD168, /* LV # Lo HANGUL SYLLABLE TYEO */ 01096 0xD184, /* LV # Lo HANGUL SYLLABLE TYE */ 01097 0xD1A0, /* LV # Lo HANGUL SYLLABLE TO */ 01098 0xD1BC, /* LV # Lo HANGUL SYLLABLE TWA */ 01099 0xD1D8, /* LV # Lo HANGUL SYLLABLE TWAE */ 01100 0xD1F4, /* LV # Lo HANGUL SYLLABLE TOE */ 01101 0xD210, /* LV # Lo HANGUL SYLLABLE TYO */ 01102 0xD22C, /* LV # Lo HANGUL SYLLABLE TU */ 01103 0xD248, /* LV # Lo HANGUL SYLLABLE TWEO */ 01104 0xD264, /* LV # Lo HANGUL SYLLABLE TWE */ 01105 0xD280, /* LV # Lo HANGUL SYLLABLE TWI */ 01106 0xD29C, /* LV # Lo HANGUL SYLLABLE TYU */ 01107 0xD2B8, /* LV # Lo HANGUL SYLLABLE TEU */ 01108 0xD2D4, /* LV # Lo HANGUL SYLLABLE TYI */ 01109 0xD2F0, /* LV # Lo HANGUL SYLLABLE TI */ 01110 0xD30C, /* LV # Lo HANGUL SYLLABLE PA */ 01111 0xD328, /* LV # Lo HANGUL SYLLABLE PAE */ 01112 0xD344, /* LV # Lo HANGUL SYLLABLE PYA */ 01113 0xD360, /* LV # Lo HANGUL SYLLABLE PYAE */ 01114 0xD37C, /* LV # Lo HANGUL SYLLABLE PEO */ 01115 0xD398, /* LV # Lo HANGUL SYLLABLE PE */ 01116 0xD3B4, /* LV # Lo HANGUL SYLLABLE PYEO */ 01117 0xD3D0, /* LV # Lo HANGUL SYLLABLE PYE */ 01118 0xD3EC, /* LV # Lo HANGUL SYLLABLE PO */ 01119 0xD408, /* LV # Lo HANGUL SYLLABLE PWA */ 01120 0xD424, /* LV # Lo HANGUL SYLLABLE PWAE */ 01121 0xD440, /* LV # Lo HANGUL SYLLABLE POE */ 01122 0xD45C, /* LV # Lo HANGUL SYLLABLE PYO */ 01123 0xD478, /* LV # Lo HANGUL SYLLABLE PU */ 01124 0xD494, /* LV # Lo HANGUL SYLLABLE PWEO */ 01125 0xD4B0, /* LV # Lo HANGUL SYLLABLE PWE */ 01126 0xD4CC, /* LV # Lo HANGUL SYLLABLE PWI */ 01127 0xD4E8, /* LV # Lo HANGUL SYLLABLE PYU */ 01128 0xD504, /* LV # Lo HANGUL SYLLABLE PEU */ 01129 0xD520, /* LV # Lo HANGUL SYLLABLE PYI */ 01130 0xD53C, /* LV # Lo HANGUL SYLLABLE PI */ 01131 0xD558, /* LV # Lo HANGUL SYLLABLE HA */ 01132 0xD574, /* LV # Lo HANGUL SYLLABLE HAE */ 01133 0xD590, /* LV # Lo HANGUL SYLLABLE HYA */ 01134 0xD5AC, /* LV # Lo HANGUL SYLLABLE HYAE */ 01135 0xD5C8, /* LV # Lo HANGUL SYLLABLE HEO */ 01136 0xD5E4, /* LV # Lo HANGUL SYLLABLE HE */ 01137 0xD600, /* LV # Lo HANGUL SYLLABLE HYEO */ 01138 0xD61C, /* LV # Lo HANGUL SYLLABLE HYE */ 01139 0xD638, /* LV # Lo HANGUL SYLLABLE HO */ 01140 0xD654, /* LV # Lo HANGUL SYLLABLE HWA */ 01141 0xD670, /* LV # Lo HANGUL SYLLABLE HWAE */ 01142 0xD68C, /* LV # Lo HANGUL SYLLABLE HOE */ 01143 0xD6A8, /* LV # Lo HANGUL SYLLABLE HYO */ 01144 0xD6C4, /* LV # Lo HANGUL SYLLABLE HU */ 01145 0xD6E0, /* LV # Lo HANGUL SYLLABLE HWEO */ 01146 0xD6FC, /* LV # Lo HANGUL SYLLABLE HWE */ 01147 0xD718, /* LV # Lo HANGUL SYLLABLE HWI */ 01148 0xD734, /* LV # Lo HANGUL SYLLABLE HYU */ 01149 0xD750, /* LV # Lo HANGUL SYLLABLE HEU */ 01150 0xD76C, /* LV # Lo HANGUL SYLLABLE HYI */ 01151 0xD788, /* LV # Lo HANGUL SYLLABLE HI */ 01152 0 01153 }; 01154 01155 static const uint32_t pr29_13_2[] = { 01156 0x11A8, 0x11A9, 0x11AA, 0x11AB, 0x11AC, 0x11AD, 0x11AE, 0x11AF, 01157 0x11B0, 0x11B1, 0x11B2, 0x11B3, 0x11B4, 0x11B5, 0x11B6, 0x11B7, 01158 0x11B8, 0x11B9, 0x11BA, 0x11BB, 0x11BC, 0x11BD, 0x11BE, 0x11BF, 01159 0x11C0, 0x11C1, 0x11C2, 0 01160 }; 01161 01162 typedef struct 01163 { 01164 const uint32_t *first; 01165 const uint32_t *last; 01166 } Pr29; 01167 01168 static const Pr29 pr29[] = { 01169 {&pr29_1_1[0], &pr29_1_2[0]}, 01170 {&pr29_2_1[0], &pr29_2_2[0]}, 01171 {&pr29_3_1[0], &pr29_3_2[0]}, 01172 {&pr29_4_1[0], &pr29_4_2[0]}, 01173 {&pr29_5_1[0], &pr29_5_2[0]}, 01174 {&pr29_6_1[0], &pr29_6_2[0]}, 01175 {&pr29_7_1[0], &pr29_7_2[0]}, 01176 {&pr29_8_1[0], &pr29_8_2[0]}, 01177 {&pr29_9_1[0], &pr29_9_2[0]}, 01178 {&pr29_10_1[0], &pr29_10_2[0]}, 01179 {&pr29_11_1[0], &pr29_11_2[0]}, 01180 {&pr29_12_1[0], &pr29_12_2[0]}, 01181 {&pr29_13_1[0], &pr29_13_2[0]}, 01182 {NULL, NULL} 01183 }; 01184 01185 static size_t 01186 first_column (uint32_t c) 01187 { 01188 size_t i, j; 01189 01190 for (i = 0; pr29[i].first; i++) 01191 for (j = 0; pr29[i].first[j]; j++) 01192 if (c == pr29[i].first[j]) 01193 return i + 1; 01194 01195 return 0; 01196 } 01197 01198 static int 01199 in_last_column_row (uint32_t c, size_t row) 01200 { 01201 size_t i; 01202 01203 for (i = 0; pr29[row - 1].last[i]; i++) 01204 if (c == pr29[row - 1].last[i]) 01205 return 1; 01206 01207 return 0; 01208 } 01209 01210 static size_t 01211 combinationclass (uint32_t c) 01212 { 01213 size_t i; 01214 01215 for (i = 0; nzcc[i]; i++) 01216 if (c == nzcc[i]) 01217 return i + 1; 01218 01219 return 0; 01220 } 01221 01236 int 01237 pr29_4 (const uint32_t * in, size_t len) 01238 { 01239 size_t i, j, k, row; 01240 01241 /* 01242 * The problem sequence are of the form: 01243 * 01244 * first_character intervening_character+ last_character 01245 * 01246 * where the first_character and last_character come from the same 01247 * row in the following table, and there is at least one 01248 * intervening_character with non-zero Canonical Combining 01249 * Class. (The '+' above means one or more occurrences.) 01250 * 01251 */ 01252 01253 for (i = 0; i < len; i++) 01254 if ((row = first_column (in[i])) > 0) 01255 for (j = i + 1; j < len; j++) 01256 if (combinationclass (in[j])) 01257 for (k = j + 1; k < len; j++) 01258 if (in_last_column_row (in[k], row)) 01259 return PR29_PROBLEM; 01260 01261 return PR29_SUCCESS; 01262 } 01263 01277 int 01278 pr29_4z (const uint32_t * in) 01279 { 01280 size_t len; 01281 01282 for (len = 0; in[len]; len++) 01283 ; 01284 01285 return pr29_4 (in, len); 01286 } 01287 01302 int 01303 pr29_8z (const char *in) 01304 { 01305 uint32_t *p; 01306 int rc; 01307 01308 p = stringprep_utf8_to_ucs4 (in, -1, NULL); 01309 if (!p) 01310 return PR29_STRINGPREP_ERROR; 01311 01312 rc = pr29_4z (p); 01313 01314 free (p); 01315 01316 return rc; 01317 } 01318
1.4.7