Return to html_entity_decode_win1251.phpt CVS log

Up to [ELWIX - Embedded LightWeight unIX -] / embedaddon / php / ext / standard / tests / strings

Annotation of embedaddon/php/ext/standard/tests/strings/html_entity_decode_win1251.phpt, revision 1.1.1.1

1.1       misho       1: --TEST--
                      2: Translation of HTML entities for encoding WIN-1251
                      3: --FILE--
                      4: <?php
                      5: $arr = array(
                      6: 0x0402 => array(0x80, "CYRILLIC CAPITAL LETTER DJE"),
                      7: 0x0403 => array(0x81, "CYRILLIC CAPITAL LETTER GJE"),
                      8: 0x201A => array(0x82, "SINGLE LOW-9 QUOTATION MARK"),
                      9: 0x0453 => array(0x83, "CYRILLIC SMALL LETTER GJE"),
                     10: 0x201E => array(0x84, "DOUBLE LOW-9 QUOTATION MARK"),
                     11: 0x2026 => array(0x85, "HORIZONTAL ELLIPSIS"),
                     12: 0x2020 => array(0x86, "DAGGER"),
                     13: 0x2021 => array(0x87, "DOUBLE DAGGER"),
                     14: 0x20AC => array(0x88, "EURO SIGN"),
                     15: 0x2030 => array(0x89, "PER MILLE SIGN"),
                     16: 0x0409 => array(0x8A, "CYRILLIC CAPITAL LETTER LJE"),
                     17: 0x2039 => array(0x8B, "SINGLE LEFT-POINTING ANGLE QUOTATION MARK"),
                     18: 0x040A => array(0x8C, "CYRILLIC CAPITAL LETTER NJE"),
                     19: 0x040C => array(0x8D, "CYRILLIC CAPITAL LETTER KJE"),
                     20: 0x040B => array(0x8E, "CYRILLIC CAPITAL LETTER TSHE"),
                     21: 0x040F => array(0x8F, "CYRILLIC CAPITAL LETTER DZHE"),
                     22: 0x0452 => array(0x90, "CYRILLIC SMALL LETTER DJE"),
                     23: 0x2018 => array(0x91, "LEFT SINGLE QUOTATION MARK"),
                     24: 0x2019 => array(0x92, "RIGHT SINGLE QUOTATION MARK"),
                     25: 0x201C => array(0x93, "LEFT DOUBLE QUOTATION MARK"),
                     26: 0x201D => array(0x94, "RIGHT DOUBLE QUOTATION MARK"),
                     27: 0x2022 => array(0x95, "BULLET"),
                     28: 0x2013 => array(0x96, "EN DASH"),
                     29: 0x2014 => array(0x97, "EM DASH"),
                     30: //0x98         #UNDEFINED
                     31: 0x2122 => array(0x99, "TRADE MARK SIGN"),
                     32: 0x0459 => array(0x9A, "CYRILLIC SMALL LETTER LJE"),
                     33: 0x203A => array(0x9B, "SINGLE RIGHT-POINTING ANGLE QUOTATION MARK"),
                     34: 0x045A => array(0x9C, "CYRILLIC SMALL LETTER NJE"),
                     35: 0x045C => array(0x9D, "CYRILLIC SMALL LETTER KJE"),
                     36: 0x045B => array(0x9E, "CYRILLIC SMALL LETTER TSHE"),
                     37: 0x045F => array(0x9F, "CYRILLIC SMALL LETTER DZHE"),
                     38: 0x00A0 => array(0xA0, "NO-BREAK SPACE"),
                     39: 0x040E => array(0xA1, "CYRILLIC CAPITAL LETTER SHORT U"),
                     40: 0x045E => array(0xA2, "CYRILLIC SMALL LETTER SHORT U"),
                     41: 0x0408 => array(0xA3, "CYRILLIC CAPITAL LETTER JE"),
                     42: 0x00A4 => array(0xA4, "CURRENCY SIGN"),
                     43: 0x0490 => array(0xA5, "CYRILLIC CAPITAL LETTER GHE WITH UPTURN"),
                     44: 0x00A6 => array(0xA6, "BROKEN BAR"),
                     45: 0x00A7 => array(0xA7, "SECTION SIGN"),
                     46: 0x0401 => array(0xA8, "CYRILLIC CAPITAL LETTER IO"),
                     47: 0x00A9 => array(0xA9, "COPYRIGHT SIGN"),
                     48: 0x0404 => array(0xAA, "CYRILLIC CAPITAL LETTER UKRAINIAN IE"),
                     49: 0x00AB => array(0xAB, "LEFT-POINTING DOUBLE ANGLE QUOTATION MARK"),
                     50: 0x00AC => array(0xAC, "NOT SIGN"),
                     51: 0x00AD => array(0xAD, "SOFT HYPHEN"),
                     52: 0x00AE => array(0xAE, "REGISTERED SIGN"),
                     53: 0x0407 => array(0xAF, "CYRILLIC CAPITAL LETTER YI"),
                     54: 0x00B0 => array(0xB0, "DEGREE SIGN"),
                     55: 0x00B1 => array(0xB1, "PLUS-MINUS SIGN"),
                     56: 0x0406 => array(0xB2, "CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I"),
                     57: 0x0456 => array(0xB3, "CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I"),
                     58: 0x0491 => array(0xB4, "CYRILLIC SMALL LETTER GHE WITH UPTURN"),
                     59: 0x00B5 => array(0xB5, "MICRO SIGN"),
                     60: 0x00B6 => array(0xB6, "PILCROW SIGN"),
                     61: 0x00B7 => array(0xB7, "MIDDLE DOT"),
                     62: 0x0451 => array(0xB8, "CYRILLIC SMALL LETTER IO"),
                     63: 0x2116 => array(0xB9, "NUMERO SIGN"),
                     64: 0x0454 => array(0xBA, "CYRILLIC SMALL LETTER UKRAINIAN IE"),
                     65: 0x00BB => array(0xBB, "RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK"),
                     66: 0x0458 => array(0xBC, "CYRILLIC SMALL LETTER JE"),
                     67: 0x0405 => array(0xBD, "CYRILLIC CAPITAL LETTER DZE"),
                     68: 0x0455 => array(0xBE, "CYRILLIC SMALL LETTER DZE"),
                     69: 0x0457 => array(0xBF, "CYRILLIC SMALL LETTER YI"),
                     70: 0x0410 => array(0xC0, "CYRILLIC CAPITAL LETTER A"),
                     71: 0x0411 => array(0xC1, "CYRILLIC CAPITAL LETTER BE"),
                     72: 0x0412 => array(0xC2, "CYRILLIC CAPITAL LETTER VE"),
                     73: 0x0413 => array(0xC3, "CYRILLIC CAPITAL LETTER GHE"),
                     74: 0x0414 => array(0xC4, "CYRILLIC CAPITAL LETTER DE"),
                     75: 0x0415 => array(0xC5, "CYRILLIC CAPITAL LETTER IE"),
                     76: 0x0416 => array(0xC6, "CYRILLIC CAPITAL LETTER ZHE"),
                     77: 0x0417 => array(0xC7, "CYRILLIC CAPITAL LETTER ZE"),
                     78: 0x0418 => array(0xC8, "CYRILLIC CAPITAL LETTER I"),
                     79: 0x0419 => array(0xC9, "CYRILLIC CAPITAL LETTER SHORT I"),
                     80: 0x041A => array(0xCA, "CYRILLIC CAPITAL LETTER KA"),
                     81: 0x041B => array(0xCB, "CYRILLIC CAPITAL LETTER EL"),
                     82: 0x041C => array(0xCC, "CYRILLIC CAPITAL LETTER EM"),
                     83: 0x041D => array(0xCD, "CYRILLIC CAPITAL LETTER EN"),
                     84: 0x041E => array(0xCE, "CYRILLIC CAPITAL LETTER O"),
                     85: 0x041F => array(0xCF, "CYRILLIC CAPITAL LETTER PE"),
                     86: 0x0420 => array(0xD0, "CYRILLIC CAPITAL LETTER ER"),
                     87: 0x0421 => array(0xD1, "CYRILLIC CAPITAL LETTER ES"),
                     88: 0x0422 => array(0xD2, "CYRILLIC CAPITAL LETTER TE"),
                     89: 0x0423 => array(0xD3, "CYRILLIC CAPITAL LETTER U"),
                     90: 0x0424 => array(0xD4, "CYRILLIC CAPITAL LETTER EF"),
                     91: 0x0425 => array(0xD5, "CYRILLIC CAPITAL LETTER HA"),
                     92: 0x0426 => array(0xD6, "CYRILLIC CAPITAL LETTER TSE"),
                     93: 0x0427 => array(0xD7, "CYRILLIC CAPITAL LETTER CHE"),
                     94: 0x0428 => array(0xD8, "CYRILLIC CAPITAL LETTER SHA"),
                     95: 0x0429 => array(0xD9, "CYRILLIC CAPITAL LETTER SHCHA"),
                     96: 0x042A => array(0xDA, "CYRILLIC CAPITAL LETTER HARD SIGN"),
                     97: 0x042B => array(0xDB, "CYRILLIC CAPITAL LETTER YERU"),
                     98: 0x042C => array(0xDC, "CYRILLIC CAPITAL LETTER SOFT SIGN"),
                     99: 0x042D => array(0xDD, "CYRILLIC CAPITAL LETTER E"),
                    100: 0x042E => array(0xDE, "CYRILLIC CAPITAL LETTER YU"),
                    101: 0x042F => array(0xDF, "CYRILLIC CAPITAL LETTER YA"),
                    102: 0x0430 => array(0xE0, "CYRILLIC SMALL LETTER A"),
                    103: 0x0431 => array(0xE1, "CYRILLIC SMALL LETTER BE"),
                    104: 0x0432 => array(0xE2, "CYRILLIC SMALL LETTER VE"),
                    105: 0x0433 => array(0xE3, "CYRILLIC SMALL LETTER GHE"),
                    106: 0x0434 => array(0xE4, "CYRILLIC SMALL LETTER DE"),
                    107: 0x0435 => array(0xE5, "CYRILLIC SMALL LETTER IE"),
                    108: 0x0436 => array(0xE6, "CYRILLIC SMALL LETTER ZHE"),
                    109: 0x0437 => array(0xE7, "CYRILLIC SMALL LETTER ZE"),
                    110: 0x0438 => array(0xE8, "CYRILLIC SMALL LETTER I"),
                    111: 0x0439 => array(0xE9, "CYRILLIC SMALL LETTER SHORT I"),
                    112: 0x043A => array(0xEA, "CYRILLIC SMALL LETTER KA"),
                    113: 0x043B => array(0xEB, "CYRILLIC SMALL LETTER EL"),
                    114: 0x043C => array(0xEC, "CYRILLIC SMALL LETTER EM"),
                    115: 0x043D => array(0xED, "CYRILLIC SMALL LETTER EN"),
                    116: 0x043E => array(0xEE, "CYRILLIC SMALL LETTER O"),
                    117: 0x043F => array(0xEF, "CYRILLIC SMALL LETTER PE"),
                    118: 0x0440 => array(0xF0, "CYRILLIC SMALL LETTER ER"),
                    119: 0x0441 => array(0xF1, "CYRILLIC SMALL LETTER ES"),
                    120: 0x0442 => array(0xF2, "CYRILLIC SMALL LETTER TE"),
                    121: 0x0443 => array(0xF3, "CYRILLIC SMALL LETTER U"),
                    122: 0x0444 => array(0xF4, "CYRILLIC SMALL LETTER EF"),
                    123: 0x0445 => array(0xF5, "CYRILLIC SMALL LETTER HA"),
                    124: 0x0446 => array(0xF6, "CYRILLIC SMALL LETTER TSE"),
                    125: 0x0447 => array(0xF7, "CYRILLIC SMALL LETTER CHE"),
                    126: 0x0448 => array(0xF8, "CYRILLIC SMALL LETTER SHA"),
                    127: 0x0449 => array(0xF9, "CYRILLIC SMALL LETTER SHCHA"),
                    128: 0x044A => array(0xFA, "CYRILLIC SMALL LETTER HARD SIGN"),
                    129: 0x044B => array(0xFB, "CYRILLIC SMALL LETTER YERU"),
                    130: 0x044C => array(0xFC, "CYRILLIC SMALL LETTER SOFT SIGN"),
                    131: 0x044D => array(0xFD, "CYRILLIC SMALL LETTER E"),
                    132: 0x044E => array(0xFE, "CYRILLIC SMALL LETTER YU"),
                    133: 0x044F => array(0xFF, "CYRILLIC SMALL LETTER YA"),
                    134: );
                    135: 
                    136: $res = html_entity_decode("&#x98;", ENT_QUOTES, 'WINDOWS-1251');
                    137: echo "Special test for &#x98; (shouldn't decode):\n";
                    138: echo $res,"\n\n";
                    139: 
                    140: foreach ($arr as $u => $v) {
                    141:     $ent = sprintf("&#x%X;", $u);
                    142:     $res = html_entity_decode($ent, ENT_QUOTES, 'WINDOWS-1251');
                    143:     $d = unpack("H*", $res);
                    144:     echo sprintf("%s: %s => %s\n", $v[1], $ent, $d[1]);
                    145:     
                    146:     $ent = sprintf("&#x%X;", $v[0]);
                    147:     $res = html_entity_decode($ent, ENT_QUOTES, 'WINDOWS-1251');
                    148:     if ($res[0] != "&" || $res[1] != "#")
                    149:         $res = unpack("H*", $res)[1];
                    150:     echo sprintf("%s => %s\n\n", $ent, $res);
                    151: }
                    152: --EXPECT--
                    153: Special test for &#x98; (shouldn't decode):
                    154: &#x98;
                    155: 
                    156: CYRILLIC CAPITAL LETTER DJE: &#x402; => 80
                    157: &#x80; => &#x80;
                    158: 
                    159: CYRILLIC CAPITAL LETTER GJE: &#x403; => 81
                    160: &#x81; => &#x81;
                    161: 
                    162: SINGLE LOW-9 QUOTATION MARK: &#x201A; => 82
                    163: &#x82; => &#x82;
                    164: 
                    165: CYRILLIC SMALL LETTER GJE: &#x453; => 83
                    166: &#x83; => &#x83;
                    167: 
                    168: DOUBLE LOW-9 QUOTATION MARK: &#x201E; => 84
                    169: &#x84; => &#x84;
                    170: 
                    171: HORIZONTAL ELLIPSIS: &#x2026; => 85
                    172: &#x85; => &#x85;
                    173: 
                    174: DAGGER: &#x2020; => 86
                    175: &#x86; => &#x86;
                    176: 
                    177: DOUBLE DAGGER: &#x2021; => 87
                    178: &#x87; => &#x87;
                    179: 
                    180: EURO SIGN: &#x20AC; => 88
                    181: &#x88; => &#x88;
                    182: 
                    183: PER MILLE SIGN: &#x2030; => 89
                    184: &#x89; => &#x89;
                    185: 
                    186: CYRILLIC CAPITAL LETTER LJE: &#x409; => 8a
                    187: &#x8A; => &#x8A;
                    188: 
                    189: SINGLE LEFT-POINTING ANGLE QUOTATION MARK: &#x2039; => 8b
                    190: &#x8B; => &#x8B;
                    191: 
                    192: CYRILLIC CAPITAL LETTER NJE: &#x40A; => 8c
                    193: &#x8C; => &#x8C;
                    194: 
                    195: CYRILLIC CAPITAL LETTER KJE: &#x40C; => 8d
                    196: &#x8D; => &#x8D;
                    197: 
                    198: CYRILLIC CAPITAL LETTER TSHE: &#x40B; => 8e
                    199: &#x8E; => &#x8E;
                    200: 
                    201: CYRILLIC CAPITAL LETTER DZHE: &#x40F; => 8f
                    202: &#x8F; => &#x8F;
                    203: 
                    204: CYRILLIC SMALL LETTER DJE: &#x452; => 90
                    205: &#x90; => &#x90;
                    206: 
                    207: LEFT SINGLE QUOTATION MARK: &#x2018; => 91
                    208: &#x91; => &#x91;
                    209: 
                    210: RIGHT SINGLE QUOTATION MARK: &#x2019; => 92
                    211: &#x92; => &#x92;
                    212: 
                    213: LEFT DOUBLE QUOTATION MARK: &#x201C; => 93
                    214: &#x93; => &#x93;
                    215: 
                    216: RIGHT DOUBLE QUOTATION MARK: &#x201D; => 94
                    217: &#x94; => &#x94;
                    218: 
                    219: BULLET: &#x2022; => 95
                    220: &#x95; => &#x95;
                    221: 
                    222: EN DASH: &#x2013; => 96
                    223: &#x96; => &#x96;
                    224: 
                    225: EM DASH: &#x2014; => 97
                    226: &#x97; => &#x97;
                    227: 
                    228: TRADE MARK SIGN: &#x2122; => 99
                    229: &#x99; => &#x99;
                    230: 
                    231: CYRILLIC SMALL LETTER LJE: &#x459; => 9a
                    232: &#x9A; => &#x9A;
                    233: 
                    234: SINGLE RIGHT-POINTING ANGLE QUOTATION MARK: &#x203A; => 9b
                    235: &#x9B; => &#x9B;
                    236: 
                    237: CYRILLIC SMALL LETTER NJE: &#x45A; => 9c
                    238: &#x9C; => &#x9C;
                    239: 
                    240: CYRILLIC SMALL LETTER KJE: &#x45C; => 9d
                    241: &#x9D; => &#x9D;
                    242: 
                    243: CYRILLIC SMALL LETTER TSHE: &#x45B; => 9e
                    244: &#x9E; => &#x9E;
                    245: 
                    246: CYRILLIC SMALL LETTER DZHE: &#x45F; => 9f
                    247: &#x9F; => &#x9F;
                    248: 
                    249: NO-BREAK SPACE: &#xA0; => a0
                    250: &#xA0; => a0
                    251: 
                    252: CYRILLIC CAPITAL LETTER SHORT U: &#x40E; => a1
                    253: &#xA1; => &#xA1;
                    254: 
                    255: CYRILLIC SMALL LETTER SHORT U: &#x45E; => a2
                    256: &#xA2; => &#xA2;
                    257: 
                    258: CYRILLIC CAPITAL LETTER JE: &#x408; => a3
                    259: &#xA3; => &#xA3;
                    260: 
                    261: CURRENCY SIGN: &#xA4; => a4
                    262: &#xA4; => a4
                    263: 
                    264: CYRILLIC CAPITAL LETTER GHE WITH UPTURN: &#x490; => a5
                    265: &#xA5; => &#xA5;
                    266: 
                    267: BROKEN BAR: &#xA6; => a6
                    268: &#xA6; => a6
                    269: 
                    270: SECTION SIGN: &#xA7; => a7
                    271: &#xA7; => a7
                    272: 
                    273: CYRILLIC CAPITAL LETTER IO: &#x401; => a8
                    274: &#xA8; => &#xA8;
                    275: 
                    276: COPYRIGHT SIGN: &#xA9; => a9
                    277: &#xA9; => a9
                    278: 
                    279: CYRILLIC CAPITAL LETTER UKRAINIAN IE: &#x404; => aa
                    280: &#xAA; => &#xAA;
                    281: 
                    282: LEFT-POINTING DOUBLE ANGLE QUOTATION MARK: &#xAB; => ab
                    283: &#xAB; => ab
                    284: 
                    285: NOT SIGN: &#xAC; => ac
                    286: &#xAC; => ac
                    287: 
                    288: SOFT HYPHEN: &#xAD; => ad
                    289: &#xAD; => ad
                    290: 
                    291: REGISTERED SIGN: &#xAE; => ae
                    292: &#xAE; => ae
                    293: 
                    294: CYRILLIC CAPITAL LETTER YI: &#x407; => af
                    295: &#xAF; => &#xAF;
                    296: 
                    297: DEGREE SIGN: &#xB0; => b0
                    298: &#xB0; => b0
                    299: 
                    300: PLUS-MINUS SIGN: &#xB1; => b1
                    301: &#xB1; => b1
                    302: 
                    303: CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I: &#x406; => b2
                    304: &#xB2; => &#xB2;
                    305: 
                    306: CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I: &#x456; => b3
                    307: &#xB3; => &#xB3;
                    308: 
                    309: CYRILLIC SMALL LETTER GHE WITH UPTURN: &#x491; => b4
                    310: &#xB4; => &#xB4;
                    311: 
                    312: MICRO SIGN: &#xB5; => b5
                    313: &#xB5; => b5
                    314: 
                    315: PILCROW SIGN: &#xB6; => b6
                    316: &#xB6; => b6
                    317: 
                    318: MIDDLE DOT: &#xB7; => b7
                    319: &#xB7; => b7
                    320: 
                    321: CYRILLIC SMALL LETTER IO: &#x451; => b8
                    322: &#xB8; => &#xB8;
                    323: 
                    324: NUMERO SIGN: &#x2116; => b9
                    325: &#xB9; => &#xB9;
                    326: 
                    327: CYRILLIC SMALL LETTER UKRAINIAN IE: &#x454; => ba
                    328: &#xBA; => &#xBA;
                    329: 
                    330: RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK: &#xBB; => bb
                    331: &#xBB; => bb
                    332: 
                    333: CYRILLIC SMALL LETTER JE: &#x458; => bc
                    334: &#xBC; => &#xBC;
                    335: 
                    336: CYRILLIC CAPITAL LETTER DZE: &#x405; => bd
                    337: &#xBD; => &#xBD;
                    338: 
                    339: CYRILLIC SMALL LETTER DZE: &#x455; => be
                    340: &#xBE; => &#xBE;
                    341: 
                    342: CYRILLIC SMALL LETTER YI: &#x457; => bf
                    343: &#xBF; => &#xBF;
                    344: 
                    345: CYRILLIC CAPITAL LETTER A: &#x410; => c0
                    346: &#xC0; => &#xC0;
                    347: 
                    348: CYRILLIC CAPITAL LETTER BE: &#x411; => c1
                    349: &#xC1; => &#xC1;
                    350: 
                    351: CYRILLIC CAPITAL LETTER VE: &#x412; => c2
                    352: &#xC2; => &#xC2;
                    353: 
                    354: CYRILLIC CAPITAL LETTER GHE: &#x413; => c3
                    355: &#xC3; => &#xC3;
                    356: 
                    357: CYRILLIC CAPITAL LETTER DE: &#x414; => c4
                    358: &#xC4; => &#xC4;
                    359: 
                    360: CYRILLIC CAPITAL LETTER IE: &#x415; => c5
                    361: &#xC5; => &#xC5;
                    362: 
                    363: CYRILLIC CAPITAL LETTER ZHE: &#x416; => c6
                    364: &#xC6; => &#xC6;
                    365: 
                    366: CYRILLIC CAPITAL LETTER ZE: &#x417; => c7
                    367: &#xC7; => &#xC7;
                    368: 
                    369: CYRILLIC CAPITAL LETTER I: &#x418; => c8
                    370: &#xC8; => &#xC8;
                    371: 
                    372: CYRILLIC CAPITAL LETTER SHORT I: &#x419; => c9
                    373: &#xC9; => &#xC9;
                    374: 
                    375: CYRILLIC CAPITAL LETTER KA: &#x41A; => ca
                    376: &#xCA; => &#xCA;
                    377: 
                    378: CYRILLIC CAPITAL LETTER EL: &#x41B; => cb
                    379: &#xCB; => &#xCB;
                    380: 
                    381: CYRILLIC CAPITAL LETTER EM: &#x41C; => cc
                    382: &#xCC; => &#xCC;
                    383: 
                    384: CYRILLIC CAPITAL LETTER EN: &#x41D; => cd
                    385: &#xCD; => &#xCD;
                    386: 
                    387: CYRILLIC CAPITAL LETTER O: &#x41E; => ce
                    388: &#xCE; => &#xCE;
                    389: 
                    390: CYRILLIC CAPITAL LETTER PE: &#x41F; => cf
                    391: &#xCF; => &#xCF;
                    392: 
                    393: CYRILLIC CAPITAL LETTER ER: &#x420; => d0
                    394: &#xD0; => &#xD0;
                    395: 
                    396: CYRILLIC CAPITAL LETTER ES: &#x421; => d1
                    397: &#xD1; => &#xD1;
                    398: 
                    399: CYRILLIC CAPITAL LETTER TE: &#x422; => d2
                    400: &#xD2; => &#xD2;
                    401: 
                    402: CYRILLIC CAPITAL LETTER U: &#x423; => d3
                    403: &#xD3; => &#xD3;
                    404: 
                    405: CYRILLIC CAPITAL LETTER EF: &#x424; => d4
                    406: &#xD4; => &#xD4;
                    407: 
                    408: CYRILLIC CAPITAL LETTER HA: &#x425; => d5
                    409: &#xD5; => &#xD5;
                    410: 
                    411: CYRILLIC CAPITAL LETTER TSE: &#x426; => d6
                    412: &#xD6; => &#xD6;
                    413: 
                    414: CYRILLIC CAPITAL LETTER CHE: &#x427; => d7
                    415: &#xD7; => &#xD7;
                    416: 
                    417: CYRILLIC CAPITAL LETTER SHA: &#x428; => d8
                    418: &#xD8; => &#xD8;
                    419: 
                    420: CYRILLIC CAPITAL LETTER SHCHA: &#x429; => d9
                    421: &#xD9; => &#xD9;
                    422: 
                    423: CYRILLIC CAPITAL LETTER HARD SIGN: &#x42A; => da
                    424: &#xDA; => &#xDA;
                    425: 
                    426: CYRILLIC CAPITAL LETTER YERU: &#x42B; => db
                    427: &#xDB; => &#xDB;
                    428: 
                    429: CYRILLIC CAPITAL LETTER SOFT SIGN: &#x42C; => dc
                    430: &#xDC; => &#xDC;
                    431: 
                    432: CYRILLIC CAPITAL LETTER E: &#x42D; => dd
                    433: &#xDD; => &#xDD;
                    434: 
                    435: CYRILLIC CAPITAL LETTER YU: &#x42E; => de
                    436: &#xDE; => &#xDE;
                    437: 
                    438: CYRILLIC CAPITAL LETTER YA: &#x42F; => df
                    439: &#xDF; => &#xDF;
                    440: 
                    441: CYRILLIC SMALL LETTER A: &#x430; => e0
                    442: &#xE0; => &#xE0;
                    443: 
                    444: CYRILLIC SMALL LETTER BE: &#x431; => e1
                    445: &#xE1; => &#xE1;
                    446: 
                    447: CYRILLIC SMALL LETTER VE: &#x432; => e2
                    448: &#xE2; => &#xE2;
                    449: 
                    450: CYRILLIC SMALL LETTER GHE: &#x433; => e3
                    451: &#xE3; => &#xE3;
                    452: 
                    453: CYRILLIC SMALL LETTER DE: &#x434; => e4
                    454: &#xE4; => &#xE4;
                    455: 
                    456: CYRILLIC SMALL LETTER IE: &#x435; => e5
                    457: &#xE5; => &#xE5;
                    458: 
                    459: CYRILLIC SMALL LETTER ZHE: &#x436; => e6
                    460: &#xE6; => &#xE6;
                    461: 
                    462: CYRILLIC SMALL LETTER ZE: &#x437; => e7
                    463: &#xE7; => &#xE7;
                    464: 
                    465: CYRILLIC SMALL LETTER I: &#x438; => e8
                    466: &#xE8; => &#xE8;
                    467: 
                    468: CYRILLIC SMALL LETTER SHORT I: &#x439; => e9
                    469: &#xE9; => &#xE9;
                    470: 
                    471: CYRILLIC SMALL LETTER KA: &#x43A; => ea
                    472: &#xEA; => &#xEA;
                    473: 
                    474: CYRILLIC SMALL LETTER EL: &#x43B; => eb
                    475: &#xEB; => &#xEB;
                    476: 
                    477: CYRILLIC SMALL LETTER EM: &#x43C; => ec
                    478: &#xEC; => &#xEC;
                    479: 
                    480: CYRILLIC SMALL LETTER EN: &#x43D; => ed
                    481: &#xED; => &#xED;
                    482: 
                    483: CYRILLIC SMALL LETTER O: &#x43E; => ee
                    484: &#xEE; => &#xEE;
                    485: 
                    486: CYRILLIC SMALL LETTER PE: &#x43F; => ef
                    487: &#xEF; => &#xEF;
                    488: 
                    489: CYRILLIC SMALL LETTER ER: &#x440; => f0
                    490: &#xF0; => &#xF0;
                    491: 
                    492: CYRILLIC SMALL LETTER ES: &#x441; => f1
                    493: &#xF1; => &#xF1;
                    494: 
                    495: CYRILLIC SMALL LETTER TE: &#x442; => f2
                    496: &#xF2; => &#xF2;
                    497: 
                    498: CYRILLIC SMALL LETTER U: &#x443; => f3
                    499: &#xF3; => &#xF3;
                    500: 
                    501: CYRILLIC SMALL LETTER EF: &#x444; => f4
                    502: &#xF4; => &#xF4;
                    503: 
                    504: CYRILLIC SMALL LETTER HA: &#x445; => f5
                    505: &#xF5; => &#xF5;
                    506: 
                    507: CYRILLIC SMALL LETTER TSE: &#x446; => f6
                    508: &#xF6; => &#xF6;
                    509: 
                    510: CYRILLIC SMALL LETTER CHE: &#x447; => f7
                    511: &#xF7; => &#xF7;
                    512: 
                    513: CYRILLIC SMALL LETTER SHA: &#x448; => f8
                    514: &#xF8; => &#xF8;
                    515: 
                    516: CYRILLIC SMALL LETTER SHCHA: &#x449; => f9
                    517: &#xF9; => &#xF9;
                    518: 
                    519: CYRILLIC SMALL LETTER HARD SIGN: &#x44A; => fa
                    520: &#xFA; => &#xFA;
                    521: 
                    522: CYRILLIC SMALL LETTER YERU: &#x44B; => fb
                    523: &#xFB; => &#xFB;
                    524: 
                    525: CYRILLIC SMALL LETTER SOFT SIGN: &#x44C; => fc
                    526: &#xFC; => &#xFC;
                    527: 
                    528: CYRILLIC SMALL LETTER E: &#x44D; => fd
                    529: &#xFD; => &#xFD;
                    530: 
                    531: CYRILLIC SMALL LETTER YU: &#x44E; => fe
                    532: &#xFE; => &#xFE;
                    533: 
                    534: CYRILLIC SMALL LETTER YA: &#x44F; => ff
                    535: &#xFF; => &#xFF;
                    536: 
                    537: