jpayne@7: # -*- coding: utf-8 -*-
jpayne@7: from codecs import BOM_UTF8, BOM_UTF16_BE, BOM_UTF16_LE, BOM_UTF32_BE, BOM_UTF32_LE
jpayne@7: from encodings.aliases import aliases
jpayne@7: from re import IGNORECASE, compile as re_compile
jpayne@7: from typing import Dict, List, Set, Union
jpayne@7: 
jpayne@7: # Contain for each eligible encoding a list of/item bytes SIG/BOM
jpayne@7: ENCODING_MARKS: Dict[str, Union[bytes, List[bytes]]] = {
jpayne@7:     "utf_8": BOM_UTF8,
jpayne@7:     "utf_7": [
jpayne@7:         b"\x2b\x2f\x76\x38",
jpayne@7:         b"\x2b\x2f\x76\x39",
jpayne@7:         b"\x2b\x2f\x76\x2b",
jpayne@7:         b"\x2b\x2f\x76\x2f",
jpayne@7:         b"\x2b\x2f\x76\x38\x2d",
jpayne@7:     ],
jpayne@7:     "gb18030": b"\x84\x31\x95\x33",
jpayne@7:     "utf_32": [BOM_UTF32_BE, BOM_UTF32_LE],
jpayne@7:     "utf_16": [BOM_UTF16_BE, BOM_UTF16_LE],
jpayne@7: }
jpayne@7: 
jpayne@7: TOO_SMALL_SEQUENCE: int = 32
jpayne@7: TOO_BIG_SEQUENCE: int = int(10e6)
jpayne@7: 
jpayne@7: UTF8_MAXIMAL_ALLOCATION: int = 1_112_064
jpayne@7: 
jpayne@7: # Up-to-date Unicode ucd/15.0.0
jpayne@7: UNICODE_RANGES_COMBINED: Dict[str, range] = {
jpayne@7:     "Control character": range(32),
jpayne@7:     "Basic Latin": range(32, 128),
jpayne@7:     "Latin-1 Supplement": range(128, 256),
jpayne@7:     "Latin Extended-A": range(256, 384),
jpayne@7:     "Latin Extended-B": range(384, 592),
jpayne@7:     "IPA Extensions": range(592, 688),
jpayne@7:     "Spacing Modifier Letters": range(688, 768),
jpayne@7:     "Combining Diacritical Marks": range(768, 880),
jpayne@7:     "Greek and Coptic": range(880, 1024),
jpayne@7:     "Cyrillic": range(1024, 1280),
jpayne@7:     "Cyrillic Supplement": range(1280, 1328),
jpayne@7:     "Armenian": range(1328, 1424),
jpayne@7:     "Hebrew": range(1424, 1536),
jpayne@7:     "Arabic": range(1536, 1792),
jpayne@7:     "Syriac": range(1792, 1872),
jpayne@7:     "Arabic Supplement": range(1872, 1920),
jpayne@7:     "Thaana": range(1920, 1984),
jpayne@7:     "NKo": range(1984, 2048),
jpayne@7:     "Samaritan": range(2048, 2112),
jpayne@7:     "Mandaic": range(2112, 2144),
jpayne@7:     "Syriac Supplement": range(2144, 2160),
jpayne@7:     "Arabic Extended-B": range(2160, 2208),
jpayne@7:     "Arabic Extended-A": range(2208, 2304),
jpayne@7:     "Devanagari": range(2304, 2432),
jpayne@7:     "Bengali": range(2432, 2560),
jpayne@7:     "Gurmukhi": range(2560, 2688),
jpayne@7:     "Gujarati": range(2688, 2816),
jpayne@7:     "Oriya": range(2816, 2944),
jpayne@7:     "Tamil": range(2944, 3072),
jpayne@7:     "Telugu": range(3072, 3200),
jpayne@7:     "Kannada": range(3200, 3328),
jpayne@7:     "Malayalam": range(3328, 3456),
jpayne@7:     "Sinhala": range(3456, 3584),
jpayne@7:     "Thai": range(3584, 3712),
jpayne@7:     "Lao": range(3712, 3840),
jpayne@7:     "Tibetan": range(3840, 4096),
jpayne@7:     "Myanmar": range(4096, 4256),
jpayne@7:     "Georgian": range(4256, 4352),
jpayne@7:     "Hangul Jamo": range(4352, 4608),
jpayne@7:     "Ethiopic": range(4608, 4992),
jpayne@7:     "Ethiopic Supplement": range(4992, 5024),
jpayne@7:     "Cherokee": range(5024, 5120),
jpayne@7:     "Unified Canadian Aboriginal Syllabics": range(5120, 5760),
jpayne@7:     "Ogham": range(5760, 5792),
jpayne@7:     "Runic": range(5792, 5888),
jpayne@7:     "Tagalog": range(5888, 5920),
jpayne@7:     "Hanunoo": range(5920, 5952),
jpayne@7:     "Buhid": range(5952, 5984),
jpayne@7:     "Tagbanwa": range(5984, 6016),
jpayne@7:     "Khmer": range(6016, 6144),
jpayne@7:     "Mongolian": range(6144, 6320),
jpayne@7:     "Unified Canadian Aboriginal Syllabics Extended": range(6320, 6400),
jpayne@7:     "Limbu": range(6400, 6480),
jpayne@7:     "Tai Le": range(6480, 6528),
jpayne@7:     "New Tai Lue": range(6528, 6624),
jpayne@7:     "Khmer Symbols": range(6624, 6656),
jpayne@7:     "Buginese": range(6656, 6688),
jpayne@7:     "Tai Tham": range(6688, 6832),
jpayne@7:     "Combining Diacritical Marks Extended": range(6832, 6912),
jpayne@7:     "Balinese": range(6912, 7040),
jpayne@7:     "Sundanese": range(7040, 7104),
jpayne@7:     "Batak": range(7104, 7168),
jpayne@7:     "Lepcha": range(7168, 7248),
jpayne@7:     "Ol Chiki": range(7248, 7296),
jpayne@7:     "Cyrillic Extended-C": range(7296, 7312),
jpayne@7:     "Georgian Extended": range(7312, 7360),
jpayne@7:     "Sundanese Supplement": range(7360, 7376),
jpayne@7:     "Vedic Extensions": range(7376, 7424),
jpayne@7:     "Phonetic Extensions": range(7424, 7552),
jpayne@7:     "Phonetic Extensions Supplement": range(7552, 7616),
jpayne@7:     "Combining Diacritical Marks Supplement": range(7616, 7680),
jpayne@7:     "Latin Extended Additional": range(7680, 7936),
jpayne@7:     "Greek Extended": range(7936, 8192),
jpayne@7:     "General Punctuation": range(8192, 8304),
jpayne@7:     "Superscripts and Subscripts": range(8304, 8352),
jpayne@7:     "Currency Symbols": range(8352, 8400),
jpayne@7:     "Combining Diacritical Marks for Symbols": range(8400, 8448),
jpayne@7:     "Letterlike Symbols": range(8448, 8528),
jpayne@7:     "Number Forms": range(8528, 8592),
jpayne@7:     "Arrows": range(8592, 8704),
jpayne@7:     "Mathematical Operators": range(8704, 8960),
jpayne@7:     "Miscellaneous Technical": range(8960, 9216),
jpayne@7:     "Control Pictures": range(9216, 9280),
jpayne@7:     "Optical Character Recognition": range(9280, 9312),
jpayne@7:     "Enclosed Alphanumerics": range(9312, 9472),
jpayne@7:     "Box Drawing": range(9472, 9600),
jpayne@7:     "Block Elements": range(9600, 9632),
jpayne@7:     "Geometric Shapes": range(9632, 9728),
jpayne@7:     "Miscellaneous Symbols": range(9728, 9984),
jpayne@7:     "Dingbats": range(9984, 10176),
jpayne@7:     "Miscellaneous Mathematical Symbols-A": range(10176, 10224),
jpayne@7:     "Supplemental Arrows-A": range(10224, 10240),
jpayne@7:     "Braille Patterns": range(10240, 10496),
jpayne@7:     "Supplemental Arrows-B": range(10496, 10624),
jpayne@7:     "Miscellaneous Mathematical Symbols-B": range(10624, 10752),
jpayne@7:     "Supplemental Mathematical Operators": range(10752, 11008),
jpayne@7:     "Miscellaneous Symbols and Arrows": range(11008, 11264),
jpayne@7:     "Glagolitic": range(11264, 11360),
jpayne@7:     "Latin Extended-C": range(11360, 11392),
jpayne@7:     "Coptic": range(11392, 11520),
jpayne@7:     "Georgian Supplement": range(11520, 11568),
jpayne@7:     "Tifinagh": range(11568, 11648),
jpayne@7:     "Ethiopic Extended": range(11648, 11744),
jpayne@7:     "Cyrillic Extended-A": range(11744, 11776),
jpayne@7:     "Supplemental Punctuation": range(11776, 11904),
jpayne@7:     "CJK Radicals Supplement": range(11904, 12032),
jpayne@7:     "Kangxi Radicals": range(12032, 12256),
jpayne@7:     "Ideographic Description Characters": range(12272, 12288),
jpayne@7:     "CJK Symbols and Punctuation": range(12288, 12352),
jpayne@7:     "Hiragana": range(12352, 12448),
jpayne@7:     "Katakana": range(12448, 12544),
jpayne@7:     "Bopomofo": range(12544, 12592),
jpayne@7:     "Hangul Compatibility Jamo": range(12592, 12688),
jpayne@7:     "Kanbun": range(12688, 12704),
jpayne@7:     "Bopomofo Extended": range(12704, 12736),
jpayne@7:     "CJK Strokes": range(12736, 12784),
jpayne@7:     "Katakana Phonetic Extensions": range(12784, 12800),
jpayne@7:     "Enclosed CJK Letters and Months": range(12800, 13056),
jpayne@7:     "CJK Compatibility": range(13056, 13312),
jpayne@7:     "CJK Unified Ideographs Extension A": range(13312, 19904),
jpayne@7:     "Yijing Hexagram Symbols": range(19904, 19968),
jpayne@7:     "CJK Unified Ideographs": range(19968, 40960),
jpayne@7:     "Yi Syllables": range(40960, 42128),
jpayne@7:     "Yi Radicals": range(42128, 42192),
jpayne@7:     "Lisu": range(42192, 42240),
jpayne@7:     "Vai": range(42240, 42560),
jpayne@7:     "Cyrillic Extended-B": range(42560, 42656),
jpayne@7:     "Bamum": range(42656, 42752),
jpayne@7:     "Modifier Tone Letters": range(42752, 42784),
jpayne@7:     "Latin Extended-D": range(42784, 43008),
jpayne@7:     "Syloti Nagri": range(43008, 43056),
jpayne@7:     "Common Indic Number Forms": range(43056, 43072),
jpayne@7:     "Phags-pa": range(43072, 43136),
jpayne@7:     "Saurashtra": range(43136, 43232),
jpayne@7:     "Devanagari Extended": range(43232, 43264),
jpayne@7:     "Kayah Li": range(43264, 43312),
jpayne@7:     "Rejang": range(43312, 43360),
jpayne@7:     "Hangul Jamo Extended-A": range(43360, 43392),
jpayne@7:     "Javanese": range(43392, 43488),
jpayne@7:     "Myanmar Extended-B": range(43488, 43520),
jpayne@7:     "Cham": range(43520, 43616),
jpayne@7:     "Myanmar Extended-A": range(43616, 43648),
jpayne@7:     "Tai Viet": range(43648, 43744),
jpayne@7:     "Meetei Mayek Extensions": range(43744, 43776),
jpayne@7:     "Ethiopic Extended-A": range(43776, 43824),
jpayne@7:     "Latin Extended-E": range(43824, 43888),
jpayne@7:     "Cherokee Supplement": range(43888, 43968),
jpayne@7:     "Meetei Mayek": range(43968, 44032),
jpayne@7:     "Hangul Syllables": range(44032, 55216),
jpayne@7:     "Hangul Jamo Extended-B": range(55216, 55296),
jpayne@7:     "High Surrogates": range(55296, 56192),
jpayne@7:     "High Private Use Surrogates": range(56192, 56320),
jpayne@7:     "Low Surrogates": range(56320, 57344),
jpayne@7:     "Private Use Area": range(57344, 63744),
jpayne@7:     "CJK Compatibility Ideographs": range(63744, 64256),
jpayne@7:     "Alphabetic Presentation Forms": range(64256, 64336),
jpayne@7:     "Arabic Presentation Forms-A": range(64336, 65024),
jpayne@7:     "Variation Selectors": range(65024, 65040),
jpayne@7:     "Vertical Forms": range(65040, 65056),
jpayne@7:     "Combining Half Marks": range(65056, 65072),
jpayne@7:     "CJK Compatibility Forms": range(65072, 65104),
jpayne@7:     "Small Form Variants": range(65104, 65136),
jpayne@7:     "Arabic Presentation Forms-B": range(65136, 65280),
jpayne@7:     "Halfwidth and Fullwidth Forms": range(65280, 65520),
jpayne@7:     "Specials": range(65520, 65536),
jpayne@7:     "Linear B Syllabary": range(65536, 65664),
jpayne@7:     "Linear B Ideograms": range(65664, 65792),
jpayne@7:     "Aegean Numbers": range(65792, 65856),
jpayne@7:     "Ancient Greek Numbers": range(65856, 65936),
jpayne@7:     "Ancient Symbols": range(65936, 66000),
jpayne@7:     "Phaistos Disc": range(66000, 66048),
jpayne@7:     "Lycian": range(66176, 66208),
jpayne@7:     "Carian": range(66208, 66272),
jpayne@7:     "Coptic Epact Numbers": range(66272, 66304),
jpayne@7:     "Old Italic": range(66304, 66352),
jpayne@7:     "Gothic": range(66352, 66384),
jpayne@7:     "Old Permic": range(66384, 66432),
jpayne@7:     "Ugaritic": range(66432, 66464),
jpayne@7:     "Old Persian": range(66464, 66528),
jpayne@7:     "Deseret": range(66560, 66640),
jpayne@7:     "Shavian": range(66640, 66688),
jpayne@7:     "Osmanya": range(66688, 66736),
jpayne@7:     "Osage": range(66736, 66816),
jpayne@7:     "Elbasan": range(66816, 66864),
jpayne@7:     "Caucasian Albanian": range(66864, 66928),
jpayne@7:     "Vithkuqi": range(66928, 67008),
jpayne@7:     "Linear A": range(67072, 67456),
jpayne@7:     "Latin Extended-F": range(67456, 67520),
jpayne@7:     "Cypriot Syllabary": range(67584, 67648),
jpayne@7:     "Imperial Aramaic": range(67648, 67680),
jpayne@7:     "Palmyrene": range(67680, 67712),
jpayne@7:     "Nabataean": range(67712, 67760),
jpayne@7:     "Hatran": range(67808, 67840),
jpayne@7:     "Phoenician": range(67840, 67872),
jpayne@7:     "Lydian": range(67872, 67904),
jpayne@7:     "Meroitic Hieroglyphs": range(67968, 68000),
jpayne@7:     "Meroitic Cursive": range(68000, 68096),
jpayne@7:     "Kharoshthi": range(68096, 68192),
jpayne@7:     "Old South Arabian": range(68192, 68224),
jpayne@7:     "Old North Arabian": range(68224, 68256),
jpayne@7:     "Manichaean": range(68288, 68352),
jpayne@7:     "Avestan": range(68352, 68416),
jpayne@7:     "Inscriptional Parthian": range(68416, 68448),
jpayne@7:     "Inscriptional Pahlavi": range(68448, 68480),
jpayne@7:     "Psalter Pahlavi": range(68480, 68528),
jpayne@7:     "Old Turkic": range(68608, 68688),
jpayne@7:     "Old Hungarian": range(68736, 68864),
jpayne@7:     "Hanifi Rohingya": range(68864, 68928),
jpayne@7:     "Rumi Numeral Symbols": range(69216, 69248),
jpayne@7:     "Yezidi": range(69248, 69312),
jpayne@7:     "Arabic Extended-C": range(69312, 69376),
jpayne@7:     "Old Sogdian": range(69376, 69424),
jpayne@7:     "Sogdian": range(69424, 69488),
jpayne@7:     "Old Uyghur": range(69488, 69552),
jpayne@7:     "Chorasmian": range(69552, 69600),
jpayne@7:     "Elymaic": range(69600, 69632),
jpayne@7:     "Brahmi": range(69632, 69760),
jpayne@7:     "Kaithi": range(69760, 69840),
jpayne@7:     "Sora Sompeng": range(69840, 69888),
jpayne@7:     "Chakma": range(69888, 69968),
jpayne@7:     "Mahajani": range(69968, 70016),
jpayne@7:     "Sharada": range(70016, 70112),
jpayne@7:     "Sinhala Archaic Numbers": range(70112, 70144),
jpayne@7:     "Khojki": range(70144, 70224),
jpayne@7:     "Multani": range(70272, 70320),
jpayne@7:     "Khudawadi": range(70320, 70400),
jpayne@7:     "Grantha": range(70400, 70528),
jpayne@7:     "Newa": range(70656, 70784),
jpayne@7:     "Tirhuta": range(70784, 70880),
jpayne@7:     "Siddham": range(71040, 71168),
jpayne@7:     "Modi": range(71168, 71264),
jpayne@7:     "Mongolian Supplement": range(71264, 71296),
jpayne@7:     "Takri": range(71296, 71376),
jpayne@7:     "Ahom": range(71424, 71504),
jpayne@7:     "Dogra": range(71680, 71760),
jpayne@7:     "Warang Citi": range(71840, 71936),
jpayne@7:     "Dives Akuru": range(71936, 72032),
jpayne@7:     "Nandinagari": range(72096, 72192),
jpayne@7:     "Zanabazar Square": range(72192, 72272),
jpayne@7:     "Soyombo": range(72272, 72368),
jpayne@7:     "Unified Canadian Aboriginal Syllabics Extended-A": range(72368, 72384),
jpayne@7:     "Pau Cin Hau": range(72384, 72448),
jpayne@7:     "Devanagari Extended-A": range(72448, 72544),
jpayne@7:     "Bhaiksuki": range(72704, 72816),
jpayne@7:     "Marchen": range(72816, 72896),
jpayne@7:     "Masaram Gondi": range(72960, 73056),
jpayne@7:     "Gunjala Gondi": range(73056, 73136),
jpayne@7:     "Makasar": range(73440, 73472),
jpayne@7:     "Kawi": range(73472, 73568),
jpayne@7:     "Lisu Supplement": range(73648, 73664),
jpayne@7:     "Tamil Supplement": range(73664, 73728),
jpayne@7:     "Cuneiform": range(73728, 74752),
jpayne@7:     "Cuneiform Numbers and Punctuation": range(74752, 74880),
jpayne@7:     "Early Dynastic Cuneiform": range(74880, 75088),
jpayne@7:     "Cypro-Minoan": range(77712, 77824),
jpayne@7:     "Egyptian Hieroglyphs": range(77824, 78896),
jpayne@7:     "Egyptian Hieroglyph Format Controls": range(78896, 78944),
jpayne@7:     "Anatolian Hieroglyphs": range(82944, 83584),
jpayne@7:     "Bamum Supplement": range(92160, 92736),
jpayne@7:     "Mro": range(92736, 92784),
jpayne@7:     "Tangsa": range(92784, 92880),
jpayne@7:     "Bassa Vah": range(92880, 92928),
jpayne@7:     "Pahawh Hmong": range(92928, 93072),
jpayne@7:     "Medefaidrin": range(93760, 93856),
jpayne@7:     "Miao": range(93952, 94112),
jpayne@7:     "Ideographic Symbols and Punctuation": range(94176, 94208),
jpayne@7:     "Tangut": range(94208, 100352),
jpayne@7:     "Tangut Components": range(100352, 101120),
jpayne@7:     "Khitan Small Script": range(101120, 101632),
jpayne@7:     "Tangut Supplement": range(101632, 101760),
jpayne@7:     "Kana Extended-B": range(110576, 110592),
jpayne@7:     "Kana Supplement": range(110592, 110848),
jpayne@7:     "Kana Extended-A": range(110848, 110896),
jpayne@7:     "Small Kana Extension": range(110896, 110960),
jpayne@7:     "Nushu": range(110960, 111360),
jpayne@7:     "Duployan": range(113664, 113824),
jpayne@7:     "Shorthand Format Controls": range(113824, 113840),
jpayne@7:     "Znamenny Musical Notation": range(118528, 118736),
jpayne@7:     "Byzantine Musical Symbols": range(118784, 119040),
jpayne@7:     "Musical Symbols": range(119040, 119296),
jpayne@7:     "Ancient Greek Musical Notation": range(119296, 119376),
jpayne@7:     "Kaktovik Numerals": range(119488, 119520),
jpayne@7:     "Mayan Numerals": range(119520, 119552),
jpayne@7:     "Tai Xuan Jing Symbols": range(119552, 119648),
jpayne@7:     "Counting Rod Numerals": range(119648, 119680),
jpayne@7:     "Mathematical Alphanumeric Symbols": range(119808, 120832),
jpayne@7:     "Sutton SignWriting": range(120832, 121520),
jpayne@7:     "Latin Extended-G": range(122624, 122880),
jpayne@7:     "Glagolitic Supplement": range(122880, 122928),
jpayne@7:     "Cyrillic Extended-D": range(122928, 123024),
jpayne@7:     "Nyiakeng Puachue Hmong": range(123136, 123216),
jpayne@7:     "Toto": range(123536, 123584),
jpayne@7:     "Wancho": range(123584, 123648),
jpayne@7:     "Nag Mundari": range(124112, 124160),
jpayne@7:     "Ethiopic Extended-B": range(124896, 124928),
jpayne@7:     "Mende Kikakui": range(124928, 125152),
jpayne@7:     "Adlam": range(125184, 125280),
jpayne@7:     "Indic Siyaq Numbers": range(126064, 126144),
jpayne@7:     "Ottoman Siyaq Numbers": range(126208, 126288),
jpayne@7:     "Arabic Mathematical Alphabetic Symbols": range(126464, 126720),
jpayne@7:     "Mahjong Tiles": range(126976, 127024),
jpayne@7:     "Domino Tiles": range(127024, 127136),
jpayne@7:     "Playing Cards": range(127136, 127232),
jpayne@7:     "Enclosed Alphanumeric Supplement": range(127232, 127488),
jpayne@7:     "Enclosed Ideographic Supplement": range(127488, 127744),
jpayne@7:     "Miscellaneous Symbols and Pictographs": range(127744, 128512),
jpayne@7:     "Emoticons range(Emoji)": range(128512, 128592),
jpayne@7:     "Ornamental Dingbats": range(128592, 128640),
jpayne@7:     "Transport and Map Symbols": range(128640, 128768),
jpayne@7:     "Alchemical Symbols": range(128768, 128896),
jpayne@7:     "Geometric Shapes Extended": range(128896, 129024),
jpayne@7:     "Supplemental Arrows-C": range(129024, 129280),
jpayne@7:     "Supplemental Symbols and Pictographs": range(129280, 129536),
jpayne@7:     "Chess Symbols": range(129536, 129648),
jpayne@7:     "Symbols and Pictographs Extended-A": range(129648, 129792),
jpayne@7:     "Symbols for Legacy Computing": range(129792, 130048),
jpayne@7:     "CJK Unified Ideographs Extension B": range(131072, 173792),
jpayne@7:     "CJK Unified Ideographs Extension C": range(173824, 177984),
jpayne@7:     "CJK Unified Ideographs Extension D": range(177984, 178208),
jpayne@7:     "CJK Unified Ideographs Extension E": range(178208, 183984),
jpayne@7:     "CJK Unified Ideographs Extension F": range(183984, 191472),
jpayne@7:     "CJK Compatibility Ideographs Supplement": range(194560, 195104),
jpayne@7:     "CJK Unified Ideographs Extension G": range(196608, 201552),
jpayne@7:     "CJK Unified Ideographs Extension H": range(201552, 205744),
jpayne@7:     "Tags": range(917504, 917632),
jpayne@7:     "Variation Selectors Supplement": range(917760, 918000),
jpayne@7:     "Supplementary Private Use Area-A": range(983040, 1048576),
jpayne@7:     "Supplementary Private Use Area-B": range(1048576, 1114112),
jpayne@7: }
jpayne@7: 
jpayne@7: 
jpayne@7: UNICODE_SECONDARY_RANGE_KEYWORD: List[str] = [
jpayne@7:     "Supplement",
jpayne@7:     "Extended",
jpayne@7:     "Extensions",
jpayne@7:     "Modifier",
jpayne@7:     "Marks",
jpayne@7:     "Punctuation",
jpayne@7:     "Symbols",
jpayne@7:     "Forms",
jpayne@7:     "Operators",
jpayne@7:     "Miscellaneous",
jpayne@7:     "Drawing",
jpayne@7:     "Block",
jpayne@7:     "Shapes",
jpayne@7:     "Supplemental",
jpayne@7:     "Tags",
jpayne@7: ]
jpayne@7: 
jpayne@7: RE_POSSIBLE_ENCODING_INDICATION = re_compile(
jpayne@7:     r"(?:(?:encoding)|(?:charset)|(?:coding))(?:[\:= ]{1,10})(?:[\"\']?)([a-zA-Z0-9\-_]+)(?:[\"\']?)",
jpayne@7:     IGNORECASE,
jpayne@7: )
jpayne@7: 
jpayne@7: IANA_NO_ALIASES = [
jpayne@7:     "cp720",
jpayne@7:     "cp737",
jpayne@7:     "cp856",
jpayne@7:     "cp874",
jpayne@7:     "cp875",
jpayne@7:     "cp1006",
jpayne@7:     "koi8_r",
jpayne@7:     "koi8_t",
jpayne@7:     "koi8_u",
jpayne@7: ]
jpayne@7: 
jpayne@7: IANA_SUPPORTED: List[str] = sorted(
jpayne@7:     filter(
jpayne@7:         lambda x: x.endswith("_codec") is False
jpayne@7:         and x not in {"rot_13", "tactis", "mbcs"},
jpayne@7:         list(set(aliases.values())) + IANA_NO_ALIASES,
jpayne@7:     )
jpayne@7: )
jpayne@7: 
jpayne@7: IANA_SUPPORTED_COUNT: int = len(IANA_SUPPORTED)
jpayne@7: 
jpayne@7: # pre-computed code page that are similar using the function cp_similarity.
jpayne@7: IANA_SUPPORTED_SIMILAR: Dict[str, List[str]] = {
jpayne@7:     "cp037": ["cp1026", "cp1140", "cp273", "cp500"],
jpayne@7:     "cp1026": ["cp037", "cp1140", "cp273", "cp500"],
jpayne@7:     "cp1125": ["cp866"],
jpayne@7:     "cp1140": ["cp037", "cp1026", "cp273", "cp500"],
jpayne@7:     "cp1250": ["iso8859_2"],
jpayne@7:     "cp1251": ["kz1048", "ptcp154"],
jpayne@7:     "cp1252": ["iso8859_15", "iso8859_9", "latin_1"],
jpayne@7:     "cp1253": ["iso8859_7"],
jpayne@7:     "cp1254": ["iso8859_15", "iso8859_9", "latin_1"],
jpayne@7:     "cp1257": ["iso8859_13"],
jpayne@7:     "cp273": ["cp037", "cp1026", "cp1140", "cp500"],
jpayne@7:     "cp437": ["cp850", "cp858", "cp860", "cp861", "cp862", "cp863", "cp865"],
jpayne@7:     "cp500": ["cp037", "cp1026", "cp1140", "cp273"],
jpayne@7:     "cp850": ["cp437", "cp857", "cp858", "cp865"],
jpayne@7:     "cp857": ["cp850", "cp858", "cp865"],
jpayne@7:     "cp858": ["cp437", "cp850", "cp857", "cp865"],
jpayne@7:     "cp860": ["cp437", "cp861", "cp862", "cp863", "cp865"],
jpayne@7:     "cp861": ["cp437", "cp860", "cp862", "cp863", "cp865"],
jpayne@7:     "cp862": ["cp437", "cp860", "cp861", "cp863", "cp865"],
jpayne@7:     "cp863": ["cp437", "cp860", "cp861", "cp862", "cp865"],
jpayne@7:     "cp865": ["cp437", "cp850", "cp857", "cp858", "cp860", "cp861", "cp862", "cp863"],
jpayne@7:     "cp866": ["cp1125"],
jpayne@7:     "iso8859_10": ["iso8859_14", "iso8859_15", "iso8859_4", "iso8859_9", "latin_1"],
jpayne@7:     "iso8859_11": ["tis_620"],
jpayne@7:     "iso8859_13": ["cp1257"],
jpayne@7:     "iso8859_14": [
jpayne@7:         "iso8859_10",
jpayne@7:         "iso8859_15",
jpayne@7:         "iso8859_16",
jpayne@7:         "iso8859_3",
jpayne@7:         "iso8859_9",
jpayne@7:         "latin_1",
jpayne@7:     ],
jpayne@7:     "iso8859_15": [
jpayne@7:         "cp1252",
jpayne@7:         "cp1254",
jpayne@7:         "iso8859_10",
jpayne@7:         "iso8859_14",
jpayne@7:         "iso8859_16",
jpayne@7:         "iso8859_3",
jpayne@7:         "iso8859_9",
jpayne@7:         "latin_1",
jpayne@7:     ],
jpayne@7:     "iso8859_16": [
jpayne@7:         "iso8859_14",
jpayne@7:         "iso8859_15",
jpayne@7:         "iso8859_2",
jpayne@7:         "iso8859_3",
jpayne@7:         "iso8859_9",
jpayne@7:         "latin_1",
jpayne@7:     ],
jpayne@7:     "iso8859_2": ["cp1250", "iso8859_16", "iso8859_4"],
jpayne@7:     "iso8859_3": ["iso8859_14", "iso8859_15", "iso8859_16", "iso8859_9", "latin_1"],
jpayne@7:     "iso8859_4": ["iso8859_10", "iso8859_2", "iso8859_9", "latin_1"],
jpayne@7:     "iso8859_7": ["cp1253"],
jpayne@7:     "iso8859_9": [
jpayne@7:         "cp1252",
jpayne@7:         "cp1254",
jpayne@7:         "cp1258",
jpayne@7:         "iso8859_10",
jpayne@7:         "iso8859_14",
jpayne@7:         "iso8859_15",
jpayne@7:         "iso8859_16",
jpayne@7:         "iso8859_3",
jpayne@7:         "iso8859_4",
jpayne@7:         "latin_1",
jpayne@7:     ],
jpayne@7:     "kz1048": ["cp1251", "ptcp154"],
jpayne@7:     "latin_1": [
jpayne@7:         "cp1252",
jpayne@7:         "cp1254",
jpayne@7:         "cp1258",
jpayne@7:         "iso8859_10",
jpayne@7:         "iso8859_14",
jpayne@7:         "iso8859_15",
jpayne@7:         "iso8859_16",
jpayne@7:         "iso8859_3",
jpayne@7:         "iso8859_4",
jpayne@7:         "iso8859_9",
jpayne@7:     ],
jpayne@7:     "mac_iceland": ["mac_roman", "mac_turkish"],
jpayne@7:     "mac_roman": ["mac_iceland", "mac_turkish"],
jpayne@7:     "mac_turkish": ["mac_iceland", "mac_roman"],
jpayne@7:     "ptcp154": ["cp1251", "kz1048"],
jpayne@7:     "tis_620": ["iso8859_11"],
jpayne@7: }
jpayne@7: 
jpayne@7: 
jpayne@7: CHARDET_CORRESPONDENCE: Dict[str, str] = {
jpayne@7:     "iso2022_kr": "ISO-2022-KR",
jpayne@7:     "iso2022_jp": "ISO-2022-JP",
jpayne@7:     "euc_kr": "EUC-KR",
jpayne@7:     "tis_620": "TIS-620",
jpayne@7:     "utf_32": "UTF-32",
jpayne@7:     "euc_jp": "EUC-JP",
jpayne@7:     "koi8_r": "KOI8-R",
jpayne@7:     "iso8859_1": "ISO-8859-1",
jpayne@7:     "iso8859_2": "ISO-8859-2",
jpayne@7:     "iso8859_5": "ISO-8859-5",
jpayne@7:     "iso8859_6": "ISO-8859-6",
jpayne@7:     "iso8859_7": "ISO-8859-7",
jpayne@7:     "iso8859_8": "ISO-8859-8",
jpayne@7:     "utf_16": "UTF-16",
jpayne@7:     "cp855": "IBM855",
jpayne@7:     "mac_cyrillic": "MacCyrillic",
jpayne@7:     "gb2312": "GB2312",
jpayne@7:     "gb18030": "GB18030",
jpayne@7:     "cp932": "CP932",
jpayne@7:     "cp866": "IBM866",
jpayne@7:     "utf_8": "utf-8",
jpayne@7:     "utf_8_sig": "UTF-8-SIG",
jpayne@7:     "shift_jis": "SHIFT_JIS",
jpayne@7:     "big5": "Big5",
jpayne@7:     "cp1250": "windows-1250",
jpayne@7:     "cp1251": "windows-1251",
jpayne@7:     "cp1252": "Windows-1252",
jpayne@7:     "cp1253": "windows-1253",
jpayne@7:     "cp1255": "windows-1255",
jpayne@7:     "cp1256": "windows-1256",
jpayne@7:     "cp1254": "Windows-1254",
jpayne@7:     "cp949": "CP949",
jpayne@7: }
jpayne@7: 
jpayne@7: 
jpayne@7: COMMON_SAFE_ASCII_CHARACTERS: Set[str] = {
jpayne@7:     "<",
jpayne@7:     ">",
jpayne@7:     "=",
jpayne@7:     ":",
jpayne@7:     "/",
jpayne@7:     "&",
jpayne@7:     ";",
jpayne@7:     "{",
jpayne@7:     "}",
jpayne@7:     "[",
jpayne@7:     "]",
jpayne@7:     ",",
jpayne@7:     "|",
jpayne@7:     '"',
jpayne@7:     "-",
jpayne@7: }
jpayne@7: 
jpayne@7: 
jpayne@7: KO_NAMES: Set[str] = {"johab", "cp949", "euc_kr"}
jpayne@7: ZH_NAMES: Set[str] = {"big5", "cp950", "big5hkscs", "hz"}
jpayne@7: 
jpayne@7: # Logging LEVEL below DEBUG
jpayne@7: TRACE: int = 5
jpayne@7: 
jpayne@7: 
jpayne@7: # Language label that contain the em dash "—"
jpayne@7: # character are to be considered alternative seq to origin
jpayne@7: FREQUENCIES: Dict[str, List[str]] = {
jpayne@7:     "English": [
jpayne@7:         "e",
jpayne@7:         "a",
jpayne@7:         "t",
jpayne@7:         "i",
jpayne@7:         "o",
jpayne@7:         "n",
jpayne@7:         "s",
jpayne@7:         "r",
jpayne@7:         "h",
jpayne@7:         "l",
jpayne@7:         "d",
jpayne@7:         "c",
jpayne@7:         "u",
jpayne@7:         "m",
jpayne@7:         "f",
jpayne@7:         "p",
jpayne@7:         "g",
jpayne@7:         "w",
jpayne@7:         "y",
jpayne@7:         "b",
jpayne@7:         "v",
jpayne@7:         "k",
jpayne@7:         "x",
jpayne@7:         "j",
jpayne@7:         "z",
jpayne@7:         "q",
jpayne@7:     ],
jpayne@7:     "English—": [
jpayne@7:         "e",
jpayne@7:         "a",
jpayne@7:         "t",
jpayne@7:         "i",
jpayne@7:         "o",
jpayne@7:         "n",
jpayne@7:         "s",
jpayne@7:         "r",
jpayne@7:         "h",
jpayne@7:         "l",
jpayne@7:         "d",
jpayne@7:         "c",
jpayne@7:         "m",
jpayne@7:         "u",
jpayne@7:         "f",
jpayne@7:         "p",
jpayne@7:         "g",
jpayne@7:         "w",
jpayne@7:         "b",
jpayne@7:         "y",
jpayne@7:         "v",
jpayne@7:         "k",
jpayne@7:         "j",
jpayne@7:         "x",
jpayne@7:         "z",
jpayne@7:         "q",
jpayne@7:     ],
jpayne@7:     "German": [
jpayne@7:         "e",
jpayne@7:         "n",
jpayne@7:         "i",
jpayne@7:         "r",
jpayne@7:         "s",
jpayne@7:         "t",
jpayne@7:         "a",
jpayne@7:         "d",
jpayne@7:         "h",
jpayne@7:         "u",
jpayne@7:         "l",
jpayne@7:         "g",
jpayne@7:         "o",
jpayne@7:         "c",
jpayne@7:         "m",
jpayne@7:         "b",
jpayne@7:         "f",
jpayne@7:         "k",
jpayne@7:         "w",
jpayne@7:         "z",
jpayne@7:         "p",
jpayne@7:         "v",
jpayne@7:         "ü",
jpayne@7:         "ä",
jpayne@7:         "ö",
jpayne@7:         "j",
jpayne@7:     ],
jpayne@7:     "French": [
jpayne@7:         "e",
jpayne@7:         "a",
jpayne@7:         "s",
jpayne@7:         "n",
jpayne@7:         "i",
jpayne@7:         "t",
jpayne@7:         "r",
jpayne@7:         "l",
jpayne@7:         "u",
jpayne@7:         "o",
jpayne@7:         "d",
jpayne@7:         "c",
jpayne@7:         "p",
jpayne@7:         "m",
jpayne@7:         "é",
jpayne@7:         "v",
jpayne@7:         "g",
jpayne@7:         "f",
jpayne@7:         "b",
jpayne@7:         "h",
jpayne@7:         "q",
jpayne@7:         "à",
jpayne@7:         "x",
jpayne@7:         "è",
jpayne@7:         "y",
jpayne@7:         "j",
jpayne@7:     ],
jpayne@7:     "Dutch": [
jpayne@7:         "e",
jpayne@7:         "n",
jpayne@7:         "a",
jpayne@7:         "i",
jpayne@7:         "r",
jpayne@7:         "t",
jpayne@7:         "o",
jpayne@7:         "d",
jpayne@7:         "s",
jpayne@7:         "l",
jpayne@7:         "g",
jpayne@7:         "h",
jpayne@7:         "v",
jpayne@7:         "m",
jpayne@7:         "u",
jpayne@7:         "k",
jpayne@7:         "c",
jpayne@7:         "p",
jpayne@7:         "b",
jpayne@7:         "w",
jpayne@7:         "j",
jpayne@7:         "z",
jpayne@7:         "f",
jpayne@7:         "y",
jpayne@7:         "x",
jpayne@7:         "ë",
jpayne@7:     ],
jpayne@7:     "Italian": [
jpayne@7:         "e",
jpayne@7:         "i",
jpayne@7:         "a",
jpayne@7:         "o",
jpayne@7:         "n",
jpayne@7:         "l",
jpayne@7:         "t",
jpayne@7:         "r",
jpayne@7:         "s",
jpayne@7:         "c",
jpayne@7:         "d",
jpayne@7:         "u",
jpayne@7:         "p",
jpayne@7:         "m",
jpayne@7:         "g",
jpayne@7:         "v",
jpayne@7:         "f",
jpayne@7:         "b",
jpayne@7:         "z",
jpayne@7:         "h",
jpayne@7:         "q",
jpayne@7:         "è",
jpayne@7:         "à",
jpayne@7:         "k",
jpayne@7:         "y",
jpayne@7:         "ò",
jpayne@7:     ],
jpayne@7:     "Polish": [
jpayne@7:         "a",
jpayne@7:         "i",
jpayne@7:         "o",
jpayne@7:         "e",
jpayne@7:         "n",
jpayne@7:         "r",
jpayne@7:         "z",
jpayne@7:         "w",
jpayne@7:         "s",
jpayne@7:         "c",
jpayne@7:         "t",
jpayne@7:         "k",
jpayne@7:         "y",
jpayne@7:         "d",
jpayne@7:         "p",
jpayne@7:         "m",
jpayne@7:         "u",
jpayne@7:         "l",
jpayne@7:         "j",
jpayne@7:         "ł",
jpayne@7:         "g",
jpayne@7:         "b",
jpayne@7:         "h",
jpayne@7:         "ą",
jpayne@7:         "ę",
jpayne@7:         "ó",
jpayne@7:     ],
jpayne@7:     "Spanish": [
jpayne@7:         "e",
jpayne@7:         "a",
jpayne@7:         "o",
jpayne@7:         "n",
jpayne@7:         "s",
jpayne@7:         "r",
jpayne@7:         "i",
jpayne@7:         "l",
jpayne@7:         "d",
jpayne@7:         "t",
jpayne@7:         "c",
jpayne@7:         "u",
jpayne@7:         "m",
jpayne@7:         "p",
jpayne@7:         "b",
jpayne@7:         "g",
jpayne@7:         "v",
jpayne@7:         "f",
jpayne@7:         "y",
jpayne@7:         "ó",
jpayne@7:         "h",
jpayne@7:         "q",
jpayne@7:         "í",
jpayne@7:         "j",
jpayne@7:         "z",
jpayne@7:         "á",
jpayne@7:     ],
jpayne@7:     "Russian": [
jpayne@7:         "о",
jpayne@7:         "а",
jpayne@7:         "е",
jpayne@7:         "и",
jpayne@7:         "н",
jpayne@7:         "с",
jpayne@7:         "т",
jpayne@7:         "р",
jpayne@7:         "в",
jpayne@7:         "л",
jpayne@7:         "к",
jpayne@7:         "м",
jpayne@7:         "д",
jpayne@7:         "п",
jpayne@7:         "у",
jpayne@7:         "г",
jpayne@7:         "я",
jpayne@7:         "ы",
jpayne@7:         "з",
jpayne@7:         "б",
jpayne@7:         "й",
jpayne@7:         "ь",
jpayne@7:         "ч",
jpayne@7:         "х",
jpayne@7:         "ж",
jpayne@7:         "ц",
jpayne@7:     ],
jpayne@7:     # Jap-Kanji
jpayne@7:     "Japanese": [
jpayne@7:         "人",
jpayne@7:         "一",
jpayne@7:         "大",
jpayne@7:         "亅",
jpayne@7:         "丁",
jpayne@7:         "丨",
jpayne@7:         "竹",
jpayne@7:         "笑",
jpayne@7:         "口",
jpayne@7:         "日",
jpayne@7:         "今",
jpayne@7:         "二",
jpayne@7:         "彳",
jpayne@7:         "行",
jpayne@7:         "十",
jpayne@7:         "土",
jpayne@7:         "丶",
jpayne@7:         "寸",
jpayne@7:         "寺",
jpayne@7:         "時",
jpayne@7:         "乙",
jpayne@7:         "丿",
jpayne@7:         "乂",
jpayne@7:         "气",
jpayne@7:         "気",
jpayne@7:         "冂",
jpayne@7:         "巾",
jpayne@7:         "亠",
jpayne@7:         "市",
jpayne@7:         "目",
jpayne@7:         "儿",
jpayne@7:         "見",
jpayne@7:         "八",
jpayne@7:         "小",
jpayne@7:         "凵",
jpayne@7:         "県",
jpayne@7:         "月",
jpayne@7:         "彐",
jpayne@7:         "門",
jpayne@7:         "間",
jpayne@7:         "木",
jpayne@7:         "東",
jpayne@7:         "山",
jpayne@7:         "出",
jpayne@7:         "本",
jpayne@7:         "中",
jpayne@7:         "刀",
jpayne@7:         "分",
jpayne@7:         "耳",
jpayne@7:         "又",
jpayne@7:         "取",
jpayne@7:         "最",
jpayne@7:         "言",
jpayne@7:         "田",
jpayne@7:         "心",
jpayne@7:         "思",
jpayne@7:         "刂",
jpayne@7:         "前",
jpayne@7:         "京",
jpayne@7:         "尹",
jpayne@7:         "事",
jpayne@7:         "生",
jpayne@7:         "厶",
jpayne@7:         "云",
jpayne@7:         "会",
jpayne@7:         "未",
jpayne@7:         "来",
jpayne@7:         "白",
jpayne@7:         "冫",
jpayne@7:         "楽",
jpayne@7:         "灬",
jpayne@7:         "馬",
jpayne@7:         "尸",
jpayne@7:         "尺",
jpayne@7:         "駅",
jpayne@7:         "明",
jpayne@7:         "耂",
jpayne@7:         "者",
jpayne@7:         "了",
jpayne@7:         "阝",
jpayne@7:         "都",
jpayne@7:         "高",
jpayne@7:         "卜",
jpayne@7:         "占",
jpayne@7:         "厂",
jpayne@7:         "广",
jpayne@7:         "店",
jpayne@7:         "子",
jpayne@7:         "申",
jpayne@7:         "奄",
jpayne@7:         "亻",
jpayne@7:         "俺",
jpayne@7:         "上",
jpayne@7:         "方",
jpayne@7:         "冖",
jpayne@7:         "学",
jpayne@7:         "衣",
jpayne@7:         "艮",
jpayne@7:         "食",
jpayne@7:         "自",
jpayne@7:     ],
jpayne@7:     # Jap-Katakana
jpayne@7:     "Japanese—": [
jpayne@7:         "ー",
jpayne@7:         "ン",
jpayne@7:         "ス",
jpayne@7:         "・",
jpayne@7:         "ル",
jpayne@7:         "ト",
jpayne@7:         "リ",
jpayne@7:         "イ",
jpayne@7:         "ア",
jpayne@7:         "ラ",
jpayne@7:         "ッ",
jpayne@7:         "ク",
jpayne@7:         "ド",
jpayne@7:         "シ",
jpayne@7:         "レ",
jpayne@7:         "ジ",
jpayne@7:         "タ",
jpayne@7:         "フ",
jpayne@7:         "ロ",
jpayne@7:         "カ",
jpayne@7:         "テ",
jpayne@7:         "マ",
jpayne@7:         "ィ",
jpayne@7:         "グ",
jpayne@7:         "バ",
jpayne@7:         "ム",
jpayne@7:         "プ",
jpayne@7:         "オ",
jpayne@7:         "コ",
jpayne@7:         "デ",
jpayne@7:         "ニ",
jpayne@7:         "ウ",
jpayne@7:         "メ",
jpayne@7:         "サ",
jpayne@7:         "ビ",
jpayne@7:         "ナ",
jpayne@7:         "ブ",
jpayne@7:         "ャ",
jpayne@7:         "エ",
jpayne@7:         "ュ",
jpayne@7:         "チ",
jpayne@7:         "キ",
jpayne@7:         "ズ",
jpayne@7:         "ダ",
jpayne@7:         "パ",
jpayne@7:         "ミ",
jpayne@7:         "ェ",
jpayne@7:         "ョ",
jpayne@7:         "ハ",
jpayne@7:         "セ",
jpayne@7:         "ベ",
jpayne@7:         "ガ",
jpayne@7:         "モ",
jpayne@7:         "ツ",
jpayne@7:         "ネ",
jpayne@7:         "ボ",
jpayne@7:         "ソ",
jpayne@7:         "ノ",
jpayne@7:         "ァ",
jpayne@7:         "ヴ",
jpayne@7:         "ワ",
jpayne@7:         "ポ",
jpayne@7:         "ペ",
jpayne@7:         "ピ",
jpayne@7:         "ケ",
jpayne@7:         "ゴ",
jpayne@7:         "ギ",
jpayne@7:         "ザ",
jpayne@7:         "ホ",
jpayne@7:         "ゲ",
jpayne@7:         "ォ",
jpayne@7:         "ヤ",
jpayne@7:         "ヒ",
jpayne@7:         "ユ",
jpayne@7:         "ヨ",
jpayne@7:         "ヘ",
jpayne@7:         "ゼ",
jpayne@7:         "ヌ",
jpayne@7:         "ゥ",
jpayne@7:         "ゾ",
jpayne@7:         "ヶ",
jpayne@7:         "ヂ",
jpayne@7:         "ヲ",
jpayne@7:         "ヅ",
jpayne@7:         "ヵ",
jpayne@7:         "ヱ",
jpayne@7:         "ヰ",
jpayne@7:         "ヮ",
jpayne@7:         "ヽ",
jpayne@7:         "゠",
jpayne@7:         "ヾ",
jpayne@7:         "ヷ",
jpayne@7:         "ヿ",
jpayne@7:         "ヸ",
jpayne@7:         "ヹ",
jpayne@7:         "ヺ",
jpayne@7:     ],
jpayne@7:     # Jap-Hiragana
jpayne@7:     "Japanese——": [
jpayne@7:         "の",
jpayne@7:         "に",
jpayne@7:         "る",
jpayne@7:         "た",
jpayne@7:         "と",
jpayne@7:         "は",
jpayne@7:         "し",
jpayne@7:         "い",
jpayne@7:         "を",
jpayne@7:         "で",
jpayne@7:         "て",
jpayne@7:         "が",
jpayne@7:         "な",
jpayne@7:         "れ",
jpayne@7:         "か",
jpayne@7:         "ら",
jpayne@7:         "さ",
jpayne@7:         "っ",
jpayne@7:         "り",
jpayne@7:         "す",
jpayne@7:         "あ",
jpayne@7:         "も",
jpayne@7:         "こ",
jpayne@7:         "ま",
jpayne@7:         "う",
jpayne@7:         "く",
jpayne@7:         "よ",
jpayne@7:         "き",
jpayne@7:         "ん",
jpayne@7:         "め",
jpayne@7:         "お",
jpayne@7:         "け",
jpayne@7:         "そ",
jpayne@7:         "つ",
jpayne@7:         "だ",
jpayne@7:         "や",
jpayne@7:         "え",
jpayne@7:         "ど",
jpayne@7:         "わ",
jpayne@7:         "ち",
jpayne@7:         "み",
jpayne@7:         "せ",
jpayne@7:         "じ",
jpayne@7:         "ば",
jpayne@7:         "へ",
jpayne@7:         "び",
jpayne@7:         "ず",
jpayne@7:         "ろ",
jpayne@7:         "ほ",
jpayne@7:         "げ",
jpayne@7:         "む",
jpayne@7:         "べ",
jpayne@7:         "ひ",
jpayne@7:         "ょ",
jpayne@7:         "ゆ",
jpayne@7:         "ぶ",
jpayne@7:         "ご",
jpayne@7:         "ゃ",
jpayne@7:         "ね",
jpayne@7:         "ふ",
jpayne@7:         "ぐ",
jpayne@7:         "ぎ",
jpayne@7:         "ぼ",
jpayne@7:         "ゅ",
jpayne@7:         "づ",
jpayne@7:         "ざ",
jpayne@7:         "ぞ",
jpayne@7:         "ぬ",
jpayne@7:         "ぜ",
jpayne@7:         "ぱ",
jpayne@7:         "ぽ",
jpayne@7:         "ぷ",
jpayne@7:         "ぴ",
jpayne@7:         "ぃ",
jpayne@7:         "ぁ",
jpayne@7:         "ぇ",
jpayne@7:         "ぺ",
jpayne@7:         "ゞ",
jpayne@7:         "ぢ",
jpayne@7:         "ぉ",
jpayne@7:         "ぅ",
jpayne@7:         "ゐ",
jpayne@7:         "ゝ",
jpayne@7:         "ゑ",
jpayne@7:         "゛",
jpayne@7:         "゜",
jpayne@7:         "ゎ",
jpayne@7:         "ゔ",
jpayne@7:         "゚",
jpayne@7:         "ゟ",
jpayne@7:         "゙",
jpayne@7:         "ゕ",
jpayne@7:         "ゖ",
jpayne@7:     ],
jpayne@7:     "Portuguese": [
jpayne@7:         "a",
jpayne@7:         "e",
jpayne@7:         "o",
jpayne@7:         "s",
jpayne@7:         "i",
jpayne@7:         "r",
jpayne@7:         "d",
jpayne@7:         "n",
jpayne@7:         "t",
jpayne@7:         "m",
jpayne@7:         "u",
jpayne@7:         "c",
jpayne@7:         "l",
jpayne@7:         "p",
jpayne@7:         "g",
jpayne@7:         "v",
jpayne@7:         "b",
jpayne@7:         "f",
jpayne@7:         "h",
jpayne@7:         "ã",
jpayne@7:         "q",
jpayne@7:         "é",
jpayne@7:         "ç",
jpayne@7:         "á",
jpayne@7:         "z",
jpayne@7:         "í",
jpayne@7:     ],
jpayne@7:     "Swedish": [
jpayne@7:         "e",
jpayne@7:         "a",
jpayne@7:         "n",
jpayne@7:         "r",
jpayne@7:         "t",
jpayne@7:         "s",
jpayne@7:         "i",
jpayne@7:         "l",
jpayne@7:         "d",
jpayne@7:         "o",
jpayne@7:         "m",
jpayne@7:         "k",
jpayne@7:         "g",
jpayne@7:         "v",
jpayne@7:         "h",
jpayne@7:         "f",
jpayne@7:         "u",
jpayne@7:         "p",
jpayne@7:         "ä",
jpayne@7:         "c",
jpayne@7:         "b",
jpayne@7:         "ö",
jpayne@7:         "å",
jpayne@7:         "y",
jpayne@7:         "j",
jpayne@7:         "x",
jpayne@7:     ],
jpayne@7:     "Chinese": [
jpayne@7:         "的",
jpayne@7:         "一",
jpayne@7:         "是",
jpayne@7:         "不",
jpayne@7:         "了",
jpayne@7:         "在",
jpayne@7:         "人",
jpayne@7:         "有",
jpayne@7:         "我",
jpayne@7:         "他",
jpayne@7:         "这",
jpayne@7:         "个",
jpayne@7:         "们",
jpayne@7:         "中",
jpayne@7:         "来",
jpayne@7:         "上",
jpayne@7:         "大",
jpayne@7:         "为",
jpayne@7:         "和",
jpayne@7:         "国",
jpayne@7:         "地",
jpayne@7:         "到",
jpayne@7:         "以",
jpayne@7:         "说",
jpayne@7:         "时",
jpayne@7:         "要",
jpayne@7:         "就",
jpayne@7:         "出",
jpayne@7:         "会",
jpayne@7:         "可",
jpayne@7:         "也",
jpayne@7:         "你",
jpayne@7:         "对",
jpayne@7:         "生",
jpayne@7:         "能",
jpayne@7:         "而",
jpayne@7:         "子",
jpayne@7:         "那",
jpayne@7:         "得",
jpayne@7:         "于",
jpayne@7:         "着",
jpayne@7:         "下",
jpayne@7:         "自",
jpayne@7:         "之",
jpayne@7:         "年",
jpayne@7:         "过",
jpayne@7:         "发",
jpayne@7:         "后",
jpayne@7:         "作",
jpayne@7:         "里",
jpayne@7:         "用",
jpayne@7:         "道",
jpayne@7:         "行",
jpayne@7:         "所",
jpayne@7:         "然",
jpayne@7:         "家",
jpayne@7:         "种",
jpayne@7:         "事",
jpayne@7:         "成",
jpayne@7:         "方",
jpayne@7:         "多",
jpayne@7:         "经",
jpayne@7:         "么",
jpayne@7:         "去",
jpayne@7:         "法",
jpayne@7:         "学",
jpayne@7:         "如",
jpayne@7:         "都",
jpayne@7:         "同",
jpayne@7:         "现",
jpayne@7:         "当",
jpayne@7:         "没",
jpayne@7:         "动",
jpayne@7:         "面",
jpayne@7:         "起",
jpayne@7:         "看",
jpayne@7:         "定",
jpayne@7:         "天",
jpayne@7:         "分",
jpayne@7:         "还",
jpayne@7:         "进",
jpayne@7:         "好",
jpayne@7:         "小",
jpayne@7:         "部",
jpayne@7:         "其",
jpayne@7:         "些",
jpayne@7:         "主",
jpayne@7:         "样",
jpayne@7:         "理",
jpayne@7:         "心",
jpayne@7:         "她",
jpayne@7:         "本",
jpayne@7:         "前",
jpayne@7:         "开",
jpayne@7:         "但",
jpayne@7:         "因",
jpayne@7:         "只",
jpayne@7:         "从",
jpayne@7:         "想",
jpayne@7:         "实",
jpayne@7:     ],
jpayne@7:     "Ukrainian": [
jpayne@7:         "о",
jpayne@7:         "а",
jpayne@7:         "н",
jpayne@7:         "і",
jpayne@7:         "и",
jpayne@7:         "р",
jpayne@7:         "в",
jpayne@7:         "т",
jpayne@7:         "е",
jpayne@7:         "с",
jpayne@7:         "к",
jpayne@7:         "л",
jpayne@7:         "у",
jpayne@7:         "д",
jpayne@7:         "м",
jpayne@7:         "п",
jpayne@7:         "з",
jpayne@7:         "я",
jpayne@7:         "ь",
jpayne@7:         "б",
jpayne@7:         "г",
jpayne@7:         "й",
jpayne@7:         "ч",
jpayne@7:         "х",
jpayne@7:         "ц",
jpayne@7:         "ї",
jpayne@7:     ],
jpayne@7:     "Norwegian": [
jpayne@7:         "e",
jpayne@7:         "r",
jpayne@7:         "n",
jpayne@7:         "t",
jpayne@7:         "a",
jpayne@7:         "s",
jpayne@7:         "i",
jpayne@7:         "o",
jpayne@7:         "l",
jpayne@7:         "d",
jpayne@7:         "g",
jpayne@7:         "k",
jpayne@7:         "m",
jpayne@7:         "v",
jpayne@7:         "f",
jpayne@7:         "p",
jpayne@7:         "u",
jpayne@7:         "b",
jpayne@7:         "h",
jpayne@7:         "å",
jpayne@7:         "y",
jpayne@7:         "j",
jpayne@7:         "ø",
jpayne@7:         "c",
jpayne@7:         "æ",
jpayne@7:         "w",
jpayne@7:     ],
jpayne@7:     "Finnish": [
jpayne@7:         "a",
jpayne@7:         "i",
jpayne@7:         "n",
jpayne@7:         "t",
jpayne@7:         "e",
jpayne@7:         "s",
jpayne@7:         "l",
jpayne@7:         "o",
jpayne@7:         "u",
jpayne@7:         "k",
jpayne@7:         "ä",
jpayne@7:         "m",
jpayne@7:         "r",
jpayne@7:         "v",
jpayne@7:         "j",
jpayne@7:         "h",
jpayne@7:         "p",
jpayne@7:         "y",
jpayne@7:         "d",
jpayne@7:         "ö",
jpayne@7:         "g",
jpayne@7:         "c",
jpayne@7:         "b",
jpayne@7:         "f",
jpayne@7:         "w",
jpayne@7:         "z",
jpayne@7:     ],
jpayne@7:     "Vietnamese": [
jpayne@7:         "n",
jpayne@7:         "h",
jpayne@7:         "t",
jpayne@7:         "i",
jpayne@7:         "c",
jpayne@7:         "g",
jpayne@7:         "a",
jpayne@7:         "o",
jpayne@7:         "u",
jpayne@7:         "m",
jpayne@7:         "l",
jpayne@7:         "r",
jpayne@7:         "à",
jpayne@7:         "đ",
jpayne@7:         "s",
jpayne@7:         "e",
jpayne@7:         "v",
jpayne@7:         "p",
jpayne@7:         "b",
jpayne@7:         "y",
jpayne@7:         "ư",
jpayne@7:         "d",
jpayne@7:         "á",
jpayne@7:         "k",
jpayne@7:         "ộ",
jpayne@7:         "ế",
jpayne@7:     ],
jpayne@7:     "Czech": [
jpayne@7:         "o",
jpayne@7:         "e",
jpayne@7:         "a",
jpayne@7:         "n",
jpayne@7:         "t",
jpayne@7:         "s",
jpayne@7:         "i",
jpayne@7:         "l",
jpayne@7:         "v",
jpayne@7:         "r",
jpayne@7:         "k",
jpayne@7:         "d",
jpayne@7:         "u",
jpayne@7:         "m",
jpayne@7:         "p",
jpayne@7:         "í",
jpayne@7:         "c",
jpayne@7:         "h",
jpayne@7:         "z",
jpayne@7:         "á",
jpayne@7:         "y",
jpayne@7:         "j",
jpayne@7:         "b",
jpayne@7:         "ě",
jpayne@7:         "é",
jpayne@7:         "ř",
jpayne@7:     ],
jpayne@7:     "Hungarian": [
jpayne@7:         "e",
jpayne@7:         "a",
jpayne@7:         "t",
jpayne@7:         "l",
jpayne@7:         "s",
jpayne@7:         "n",
jpayne@7:         "k",
jpayne@7:         "r",
jpayne@7:         "i",
jpayne@7:         "o",
jpayne@7:         "z",
jpayne@7:         "á",
jpayne@7:         "é",
jpayne@7:         "g",
jpayne@7:         "m",
jpayne@7:         "b",
jpayne@7:         "y",
jpayne@7:         "v",
jpayne@7:         "d",
jpayne@7:         "h",
jpayne@7:         "u",
jpayne@7:         "p",
jpayne@7:         "j",
jpayne@7:         "ö",
jpayne@7:         "f",
jpayne@7:         "c",
jpayne@7:     ],
jpayne@7:     "Korean": [
jpayne@7:         "이",
jpayne@7:         "다",
jpayne@7:         "에",
jpayne@7:         "의",
jpayne@7:         "는",
jpayne@7:         "로",
jpayne@7:         "하",
jpayne@7:         "을",
jpayne@7:         "가",
jpayne@7:         "고",
jpayne@7:         "지",
jpayne@7:         "서",
jpayne@7:         "한",
jpayne@7:         "은",
jpayne@7:         "기",
jpayne@7:         "으",
jpayne@7:         "년",
jpayne@7:         "대",
jpayne@7:         "사",
jpayne@7:         "시",
jpayne@7:         "를",
jpayne@7:         "리",
jpayne@7:         "도",
jpayne@7:         "인",
jpayne@7:         "스",
jpayne@7:         "일",
jpayne@7:     ],
jpayne@7:     "Indonesian": [
jpayne@7:         "a",
jpayne@7:         "n",
jpayne@7:         "e",
jpayne@7:         "i",
jpayne@7:         "r",
jpayne@7:         "t",
jpayne@7:         "u",
jpayne@7:         "s",
jpayne@7:         "d",
jpayne@7:         "k",
jpayne@7:         "m",
jpayne@7:         "l",
jpayne@7:         "g",
jpayne@7:         "p",
jpayne@7:         "b",
jpayne@7:         "o",
jpayne@7:         "h",
jpayne@7:         "y",
jpayne@7:         "j",
jpayne@7:         "c",
jpayne@7:         "w",
jpayne@7:         "f",
jpayne@7:         "v",
jpayne@7:         "z",
jpayne@7:         "x",
jpayne@7:         "q",
jpayne@7:     ],
jpayne@7:     "Turkish": [
jpayne@7:         "a",
jpayne@7:         "e",
jpayne@7:         "i",
jpayne@7:         "n",
jpayne@7:         "r",
jpayne@7:         "l",
jpayne@7:         "ı",
jpayne@7:         "k",
jpayne@7:         "d",
jpayne@7:         "t",
jpayne@7:         "s",
jpayne@7:         "m",
jpayne@7:         "y",
jpayne@7:         "u",
jpayne@7:         "o",
jpayne@7:         "b",
jpayne@7:         "ü",
jpayne@7:         "ş",
jpayne@7:         "v",
jpayne@7:         "g",
jpayne@7:         "z",
jpayne@7:         "h",
jpayne@7:         "c",
jpayne@7:         "p",
jpayne@7:         "ç",
jpayne@7:         "ğ",
jpayne@7:     ],
jpayne@7:     "Romanian": [
jpayne@7:         "e",
jpayne@7:         "i",
jpayne@7:         "a",
jpayne@7:         "r",
jpayne@7:         "n",
jpayne@7:         "t",
jpayne@7:         "u",
jpayne@7:         "l",
jpayne@7:         "o",
jpayne@7:         "c",
jpayne@7:         "s",
jpayne@7:         "d",
jpayne@7:         "p",
jpayne@7:         "m",
jpayne@7:         "ă",
jpayne@7:         "f",
jpayne@7:         "v",
jpayne@7:         "î",
jpayne@7:         "g",
jpayne@7:         "b",
jpayne@7:         "ș",
jpayne@7:         "ț",
jpayne@7:         "z",
jpayne@7:         "h",
jpayne@7:         "â",
jpayne@7:         "j",
jpayne@7:     ],
jpayne@7:     "Farsi": [
jpayne@7:         "ا",
jpayne@7:         "ی",
jpayne@7:         "ر",
jpayne@7:         "د",
jpayne@7:         "ن",
jpayne@7:         "ه",
jpayne@7:         "و",
jpayne@7:         "م",
jpayne@7:         "ت",
jpayne@7:         "ب",
jpayne@7:         "س",
jpayne@7:         "ل",
jpayne@7:         "ک",
jpayne@7:         "ش",
jpayne@7:         "ز",
jpayne@7:         "ف",
jpayne@7:         "گ",
jpayne@7:         "ع",
jpayne@7:         "خ",
jpayne@7:         "ق",
jpayne@7:         "ج",
jpayne@7:         "آ",
jpayne@7:         "پ",
jpayne@7:         "ح",
jpayne@7:         "ط",
jpayne@7:         "ص",
jpayne@7:     ],
jpayne@7:     "Arabic": [
jpayne@7:         "ا",
jpayne@7:         "ل",
jpayne@7:         "ي",
jpayne@7:         "م",
jpayne@7:         "و",
jpayne@7:         "ن",
jpayne@7:         "ر",
jpayne@7:         "ت",
jpayne@7:         "ب",
jpayne@7:         "ة",
jpayne@7:         "ع",
jpayne@7:         "د",
jpayne@7:         "س",
jpayne@7:         "ف",
jpayne@7:         "ه",
jpayne@7:         "ك",
jpayne@7:         "ق",
jpayne@7:         "أ",
jpayne@7:         "ح",
jpayne@7:         "ج",
jpayne@7:         "ش",
jpayne@7:         "ط",
jpayne@7:         "ص",
jpayne@7:         "ى",
jpayne@7:         "خ",
jpayne@7:         "إ",
jpayne@7:     ],
jpayne@7:     "Danish": [
jpayne@7:         "e",
jpayne@7:         "r",
jpayne@7:         "n",
jpayne@7:         "t",
jpayne@7:         "a",
jpayne@7:         "i",
jpayne@7:         "s",
jpayne@7:         "d",
jpayne@7:         "l",
jpayne@7:         "o",
jpayne@7:         "g",
jpayne@7:         "m",
jpayne@7:         "k",
jpayne@7:         "f",
jpayne@7:         "v",
jpayne@7:         "u",
jpayne@7:         "b",
jpayne@7:         "h",
jpayne@7:         "p",
jpayne@7:         "å",
jpayne@7:         "y",
jpayne@7:         "ø",
jpayne@7:         "æ",
jpayne@7:         "c",
jpayne@7:         "j",
jpayne@7:         "w",
jpayne@7:     ],
jpayne@7:     "Serbian": [
jpayne@7:         "а",
jpayne@7:         "и",
jpayne@7:         "о",
jpayne@7:         "е",
jpayne@7:         "н",
jpayne@7:         "р",
jpayne@7:         "с",
jpayne@7:         "у",
jpayne@7:         "т",
jpayne@7:         "к",
jpayne@7:         "ј",
jpayne@7:         "в",
jpayne@7:         "д",
jpayne@7:         "м",
jpayne@7:         "п",
jpayne@7:         "л",
jpayne@7:         "г",
jpayne@7:         "з",
jpayne@7:         "б",
jpayne@7:         "a",
jpayne@7:         "i",
jpayne@7:         "e",
jpayne@7:         "o",
jpayne@7:         "n",
jpayne@7:         "ц",
jpayne@7:         "ш",
jpayne@7:     ],
jpayne@7:     "Lithuanian": [
jpayne@7:         "i",
jpayne@7:         "a",
jpayne@7:         "s",
jpayne@7:         "o",
jpayne@7:         "r",
jpayne@7:         "e",
jpayne@7:         "t",
jpayne@7:         "n",
jpayne@7:         "u",
jpayne@7:         "k",
jpayne@7:         "m",
jpayne@7:         "l",
jpayne@7:         "p",
jpayne@7:         "v",
jpayne@7:         "d",
jpayne@7:         "j",
jpayne@7:         "g",
jpayne@7:         "ė",
jpayne@7:         "b",
jpayne@7:         "y",
jpayne@7:         "ų",
jpayne@7:         "š",
jpayne@7:         "ž",
jpayne@7:         "c",
jpayne@7:         "ą",
jpayne@7:         "į",
jpayne@7:     ],
jpayne@7:     "Slovene": [
jpayne@7:         "e",
jpayne@7:         "a",
jpayne@7:         "i",
jpayne@7:         "o",
jpayne@7:         "n",
jpayne@7:         "r",
jpayne@7:         "s",
jpayne@7:         "l",
jpayne@7:         "t",
jpayne@7:         "j",
jpayne@7:         "v",
jpayne@7:         "k",
jpayne@7:         "d",
jpayne@7:         "p",
jpayne@7:         "m",
jpayne@7:         "u",
jpayne@7:         "z",
jpayne@7:         "b",
jpayne@7:         "g",
jpayne@7:         "h",
jpayne@7:         "č",
jpayne@7:         "c",
jpayne@7:         "š",
jpayne@7:         "ž",
jpayne@7:         "f",
jpayne@7:         "y",
jpayne@7:     ],
jpayne@7:     "Slovak": [
jpayne@7:         "o",
jpayne@7:         "a",
jpayne@7:         "e",
jpayne@7:         "n",
jpayne@7:         "i",
jpayne@7:         "r",
jpayne@7:         "v",
jpayne@7:         "t",
jpayne@7:         "s",
jpayne@7:         "l",
jpayne@7:         "k",
jpayne@7:         "d",
jpayne@7:         "m",
jpayne@7:         "p",
jpayne@7:         "u",
jpayne@7:         "c",
jpayne@7:         "h",
jpayne@7:         "j",
jpayne@7:         "b",
jpayne@7:         "z",
jpayne@7:         "á",
jpayne@7:         "y",
jpayne@7:         "ý",
jpayne@7:         "í",
jpayne@7:         "č",
jpayne@7:         "é",
jpayne@7:     ],
jpayne@7:     "Hebrew": [
jpayne@7:         "י",
jpayne@7:         "ו",
jpayne@7:         "ה",
jpayne@7:         "ל",
jpayne@7:         "ר",
jpayne@7:         "ב",
jpayne@7:         "ת",
jpayne@7:         "מ",
jpayne@7:         "א",
jpayne@7:         "ש",
jpayne@7:         "נ",
jpayne@7:         "ע",
jpayne@7:         "ם",
jpayne@7:         "ד",
jpayne@7:         "ק",
jpayne@7:         "ח",
jpayne@7:         "פ",
jpayne@7:         "ס",
jpayne@7:         "כ",
jpayne@7:         "ג",
jpayne@7:         "ט",
jpayne@7:         "צ",
jpayne@7:         "ן",
jpayne@7:         "ז",
jpayne@7:         "ך",
jpayne@7:     ],
jpayne@7:     "Bulgarian": [
jpayne@7:         "а",
jpayne@7:         "и",
jpayne@7:         "о",
jpayne@7:         "е",
jpayne@7:         "н",
jpayne@7:         "т",
jpayne@7:         "р",
jpayne@7:         "с",
jpayne@7:         "в",
jpayne@7:         "л",
jpayne@7:         "к",
jpayne@7:         "д",
jpayne@7:         "п",
jpayne@7:         "м",
jpayne@7:         "з",
jpayne@7:         "г",
jpayne@7:         "я",
jpayne@7:         "ъ",
jpayne@7:         "у",
jpayne@7:         "б",
jpayne@7:         "ч",
jpayne@7:         "ц",
jpayne@7:         "й",
jpayne@7:         "ж",
jpayne@7:         "щ",
jpayne@7:         "х",
jpayne@7:     ],
jpayne@7:     "Croatian": [
jpayne@7:         "a",
jpayne@7:         "i",
jpayne@7:         "o",
jpayne@7:         "e",
jpayne@7:         "n",
jpayne@7:         "r",
jpayne@7:         "j",
jpayne@7:         "s",
jpayne@7:         "t",
jpayne@7:         "u",
jpayne@7:         "k",
jpayne@7:         "l",
jpayne@7:         "v",
jpayne@7:         "d",
jpayne@7:         "m",
jpayne@7:         "p",
jpayne@7:         "g",
jpayne@7:         "z",
jpayne@7:         "b",
jpayne@7:         "c",
jpayne@7:         "č",
jpayne@7:         "h",
jpayne@7:         "š",
jpayne@7:         "ž",
jpayne@7:         "ć",
jpayne@7:         "f",
jpayne@7:     ],
jpayne@7:     "Hindi": [
jpayne@7:         "क",
jpayne@7:         "र",
jpayne@7:         "स",
jpayne@7:         "न",
jpayne@7:         "त",
jpayne@7:         "म",
jpayne@7:         "ह",
jpayne@7:         "प",
jpayne@7:         "य",
jpayne@7:         "ल",
jpayne@7:         "व",
jpayne@7:         "ज",
jpayne@7:         "द",
jpayne@7:         "ग",
jpayne@7:         "ब",
jpayne@7:         "श",
jpayne@7:         "ट",
jpayne@7:         "अ",
jpayne@7:         "ए",
jpayne@7:         "थ",
jpayne@7:         "भ",
jpayne@7:         "ड",
jpayne@7:         "च",
jpayne@7:         "ध",
jpayne@7:         "ष",
jpayne@7:         "इ",
jpayne@7:     ],
jpayne@7:     "Estonian": [
jpayne@7:         "a",
jpayne@7:         "i",
jpayne@7:         "e",
jpayne@7:         "s",
jpayne@7:         "t",
jpayne@7:         "l",
jpayne@7:         "u",
jpayne@7:         "n",
jpayne@7:         "o",
jpayne@7:         "k",
jpayne@7:         "r",
jpayne@7:         "d",
jpayne@7:         "m",
jpayne@7:         "v",
jpayne@7:         "g",
jpayne@7:         "p",
jpayne@7:         "j",
jpayne@7:         "h",
jpayne@7:         "ä",
jpayne@7:         "b",
jpayne@7:         "õ",
jpayne@7:         "ü",
jpayne@7:         "f",
jpayne@7:         "c",
jpayne@7:         "ö",
jpayne@7:         "y",
jpayne@7:     ],
jpayne@7:     "Thai": [
jpayne@7:         "า",
jpayne@7:         "น",
jpayne@7:         "ร",
jpayne@7:         "อ",
jpayne@7:         "ก",
jpayne@7:         "เ",
jpayne@7:         "ง",
jpayne@7:         "ม",
jpayne@7:         "ย",
jpayne@7:         "ล",
jpayne@7:         "ว",
jpayne@7:         "ด",
jpayne@7:         "ท",
jpayne@7:         "ส",
jpayne@7:         "ต",
jpayne@7:         "ะ",
jpayne@7:         "ป",
jpayne@7:         "บ",
jpayne@7:         "ค",
jpayne@7:         "ห",
jpayne@7:         "แ",
jpayne@7:         "จ",
jpayne@7:         "พ",
jpayne@7:         "ช",
jpayne@7:         "ข",
jpayne@7:         "ใ",
jpayne@7:     ],
jpayne@7:     "Greek": [
jpayne@7:         "α",
jpayne@7:         "τ",
jpayne@7:         "ο",
jpayne@7:         "ι",
jpayne@7:         "ε",
jpayne@7:         "ν",
jpayne@7:         "ρ",
jpayne@7:         "σ",
jpayne@7:         "κ",
jpayne@7:         "η",
jpayne@7:         "π",
jpayne@7:         "ς",
jpayne@7:         "υ",
jpayne@7:         "μ",
jpayne@7:         "λ",
jpayne@7:         "ί",
jpayne@7:         "ό",
jpayne@7:         "ά",
jpayne@7:         "γ",
jpayne@7:         "έ",
jpayne@7:         "δ",
jpayne@7:         "ή",
jpayne@7:         "ω",
jpayne@7:         "χ",
jpayne@7:         "θ",
jpayne@7:         "ύ",
jpayne@7:     ],
jpayne@7:     "Tamil": [
jpayne@7:         "க",
jpayne@7:         "த",
jpayne@7:         "ப",
jpayne@7:         "ட",
jpayne@7:         "ர",
jpayne@7:         "ம",
jpayne@7:         "ல",
jpayne@7:         "ன",
jpayne@7:         "வ",
jpayne@7:         "ற",
jpayne@7:         "ய",
jpayne@7:         "ள",
jpayne@7:         "ச",
jpayne@7:         "ந",
jpayne@7:         "இ",
jpayne@7:         "ண",
jpayne@7:         "அ",
jpayne@7:         "ஆ",
jpayne@7:         "ழ",
jpayne@7:         "ங",
jpayne@7:         "எ",
jpayne@7:         "உ",
jpayne@7:         "ஒ",
jpayne@7:         "ஸ",
jpayne@7:     ],
jpayne@7:     "Kazakh": [
jpayne@7:         "а",
jpayne@7:         "ы",
jpayne@7:         "е",
jpayne@7:         "н",
jpayne@7:         "т",
jpayne@7:         "р",
jpayne@7:         "л",
jpayne@7:         "і",
jpayne@7:         "д",
jpayne@7:         "с",
jpayne@7:         "м",
jpayne@7:         "қ",
jpayne@7:         "к",
jpayne@7:         "о",
jpayne@7:         "б",
jpayne@7:         "и",
jpayne@7:         "у",
jpayne@7:         "ғ",
jpayne@7:         "ж",
jpayne@7:         "ң",
jpayne@7:         "з",
jpayne@7:         "ш",
jpayne@7:         "й",
jpayne@7:         "п",
jpayne@7:         "г",
jpayne@7:         "ө",
jpayne@7:     ],
jpayne@7: }
jpayne@7: 
jpayne@7: LANGUAGE_SUPPORTED_COUNT: int = len(FREQUENCIES)