Modulu:scripts/data
Documentation for this module may be created at Modulu:scripts/data/dok
-- When adding new scripts to this file, please don't forget to add -- style definitons for the script in [[MediaWiki:Common.css]]. local u = mw.ustring.char local m = {} m["Adlm"] = { canonicalName = "Adlam", characters = "𞤀-𞥟", direction = "rtl", } m["Afak"] = { canonicalName = "Afaka", } m["Aghb"] = { canonicalName = "Caucasian Albanian", characters = "𐔰-𐕣𐕯", } m["Ahom"] = { canonicalName = "Ahom", characters = "𑜀-𑜿", systems = {"abugida"}, } m["Arab"] = { canonicalName = "Arabic", otherNames = {"Jawi"}, characters = "-ۿݐ-ݿࢠ-ࣿﭐ-﷽ﹰ-ﻼ", direction = "rtl", } m["fa-Arab"] = { canonicalName = "Arabic", otherNames = {"Perso-Arabic"}, characters = m["Arab"].characters, direction = "rtl", parent = "Arab", } m["kk-Arab"] = { canonicalName = "Arabic", characters = m["Arab"].characters, direction = "rtl", parent = "Arab", } m["ks-Arab"] = { canonicalName = "Arabic", characters = m["Arab"].characters, direction = "rtl", parent = "Arab", } m["ku-Arab"] = { canonicalName = "Arabic", characters = m["Arab"].characters, direction = "rtl", parent = "Arab", } m["mzn-Arab"] = { canonicalName = "Arabic", characters = m["Arab"].characters, direction = "rtl", parent = "Arab", } m["ota-Arab"] = { canonicalName = "Arabic", characters = m["Arab"].characters, direction = "rtl", parent = "Arab", } m["pa-Arab"] = { canonicalName = "Arabic", otherNames = {"Shahmukhi"}, characters = m["Arab"].characters, direction = "rtl", parent = "Arab", } m["ps-Arab"] = { canonicalName = "Arabic", characters = m["Arab"].characters, direction = "rtl", parent = "Arab", } m["sd-Arab"] = { canonicalName = "Arabic", characters = m["Arab"].characters, direction = "rtl", parent = "Arab", } m["tt-Arab"] = { canonicalName = "Arabic", characters = m["Arab"].characters, direction = "rtl", parent = "Arab", } m["ug-Arab"] = { canonicalName = "Arabic", characters = m["Arab"].characters, direction = "rtl", parent = "Arab", } m["ur-Arab"] = { canonicalName = "Arabic", characters = m["Arab"].characters, direction = "rtl", parent = "Arab", } --[[ m["Aran"] = { canonicalName = "Nastaliq", otherNames = {"Nastaleeq"}, characters = m["Arab"].characters, direction = "rtl", } ]] m["Armi"] = { canonicalName = "Imperial Aramaic", characters = "𐡀-𐡟", direction = "rtl", } m["Armn"] = { canonicalName = "Armenian", characters = "Ա-֏ﬓ-ﬗ", } m["Avst"] = { canonicalName = "Avestan", characters = "𐬀-𐬿", direction = "rtl", } m["Bali"] = { canonicalName = "Balinese", characters = "ᬀ-᭼", systems = {"abugida"}, } m["Bamu"] = { canonicalName = "Bamum", characters = "ꚠ-꛷𖠀-𖨸", } m["Bass"] = { canonicalName = "Bassa", otherNames = {"Bassa Vah", "Vah"}, characters = "𖫐-𖫵", } m["Batk"] = { canonicalName = "Batak", characters = "ᯀ-᯿", systems = {"abugida"}, } m["Beng"] = { canonicalName = "Bengali", characters = "ঁ-৺", systems = {"abugida"}, } m["Bhks"] = { canonicalName = "Bhaiksuki", characters = "𑰀-𑱬", systems = {"abugida"}, } m["Bopo"] = { canonicalName = "Zhuyin", otherNames = {"Zhuyin Fuhao", "Bopomofo"}, characters = "ㄅ-ㄭㆠ-ㆺ", } m["Brah"] = { canonicalName = "Brahmi", characters = "𑀀-𑁯", systems = {"abugida"}, } m["Brai"] = { canonicalName = "Braille", characters = "⠀-⣿", } m["Bugi"] = { canonicalName = "Buginese", otherNames = {"Lontara"}, characters = "ᨀ-᨟", systems = {"abugida"}, } m["Buhd"] = { canonicalName = "Buhid", characters = "ᝀ-ᝓ", } m["Cakm"] = { canonicalName = "Chakma", characters = "𑄀-𑅃", systems = {"abugida"}, } m["Cans"] = { canonicalName = "Canadian syllabics", characters = "᐀-ᙿ", systems = {"abugida"}, } m["Cari"] = { canonicalName = "Carian", characters = "𐊠-𐋐", } m["Cham"] = { canonicalName = "Cham", characters = "ꨀ-꩟", systems = {"abugida"}, } m["Cher"] = { canonicalName = "Cherokee", characters = "Ꭰ-Ᏼꭰ-ꮿ", } m["Copt"] = { canonicalName = "Coptic", characters = "Ϣ-ϯⲀ-⳿𐋡-𐋻", -- this is mostly "Coptic", not unified "Greek and Coptic" } m["Cprt"] = { canonicalName = "Cypriot", characters = "𐠀-𐠿", direction = "rtl", } m["Cyrl"] = { canonicalName = "Cyrillic", characters = "Ѐ-џѢѣѪѫѬѭѲѳѴѵҊ-ԧꚀ-ꚗ", } m["Cyrs"] = { canonicalName = "Old Cyrillic", characters = "Ѐ-ԧꙀ-ꚗ", } m["Deva"] = { canonicalName = "Devanagari", characters = "ऀ-ॿ꣠-ꣻ", systems = {"abugida"}, } m["Dsrt"] = { canonicalName = "Deseret", characters = "𐐀-𐑏", } m["Dupl"] = { canonicalName = "Duployan", characters = "𛰀-𛲟", } m["Egyd"] = { canonicalName = "Demotic", } m["Egyp"] = { canonicalName = "Egyptian hieroglyphic", characters = "𓀀-𓐮", } m["Elba"] = { canonicalName = "Elbasan", characters = "𐔀-𐔧", } m["Ethi"] = { canonicalName = "Ethiopic", otherNames = {"Ge'ez"}, characters = "ሀ-᎙ⶀ-ⷞꬁ-ꬮ", systems = {"abugida"}, } m["Geok"] = { canonicalName = "Khutsuri", otherNames = {"Nuskhuri", "Asomtavruli"}, characters = "Ⴀ-Ⴭⴀ-ⴭ", -- Ⴀ-Ⴭ is Asomtavruli, ⴀ-ⴭ is Nuskhuri } m["Geor"] = { canonicalName = "Georgian", otherNames = {"Mkhedruli"}, characters = "Ⴀ-ჼ", -- technically only the range [ა-ჿ] is Mkhedruli } m["Glag"] = { canonicalName = "Glagolitic", characters = "Ⰰ-ⱞ𞀀-𞀪", } m["Gonm"] = { canonicalName = "Masaram Gondi", characters = u(0x11D00).."-"..u(0x11D59), } m["Goth"] = { canonicalName = "Gothic", characters = "𐌰-𐍊", } m["Gran"] = { canonicalName = "Grantha", characters = "𑌁-𑍴", systems = {"abugida"}, } m["Grek"] = { canonicalName = "Greek", characters = "Ͱ-Ͽ", } m["polytonic"] = { canonicalName = "Greek", characters = "ἀ-῾" .. m["Grek"].characters, parent = "Grek", } m["Gujr"] = { canonicalName = "Gujarati", characters = "ઁ-૱", systems = {"abugida"}, } m["Guru"] = { canonicalName = "Gurmukhi", characters = "ਁ-ੵ", systems = {"abugida"}, } m["Hang"] = { canonicalName = "Hangul", characters = "ᄀ-ᇿ가-힣ㄱ-ㆎ", } m["Hani"] = { canonicalName = "Han", otherNames = {"Hanzi", "Chu Nom"}, characters = "一-鿿㐀-䶵𠀀-"..u(0x2EBE0).."﨎﨏﨑﨓﨔﨟﨡﨣﨤﨧﨨﨩⺀-㇀-㇣ -〿㍻-㍿", } m["Hans"] = { canonicalName = "Simplified Han", characters = m["Hani"].characters, } m["Hant"] = { canonicalName = "Traditional Han", characters = m["Hani"].characters, } m["Hatr"] = { canonicalName = "Hatran", characters = "𐣠-𐣿", } m["Hira"] = { canonicalName = "Hiragana", characters = "ぁ-ゟ𛀁", } m["Hluw"] = { canonicalName = "Anatolian Hieroglyphs", characters = "𔐀-𔙆", } m["Hung"] = { canonicalName = "Hungarian runes", characters = "𐲀-𐲲", } m["Kana"] = { canonicalName = "Katakana", characters = "゠-ヿㇰ-ㇿ𛀀㌀-㍗", } -- These should be defined after the scripts they are composed of m["Jpan"] = { canonicalName = "Japanese", -- Add hentaigana either characters = m["Hira"].characters .. m["Kana"].characters .. m["Hani"].characters .. u(0x1B002).."-"..u(0x1B11E), } m["Kore"] = { canonicalName = "Korean", characters = m["Hang"].characters .. m["Hani"].characters .. "！-￮", } m["CGK"] = { canonicalName = "Korean", } m["Hano"] = { canonicalName = "Hanunoo", characters = "ᜠ-᜴", systems = {"abugida"}, } m["Hebr"] = { canonicalName = "Hebrew", characters = u(0x0590) .. "-" .. u(0x05FF) .. u(0xFB1D) .. "-" .. u(0xFB4F), direction = "rtl", } m["Hmng"] = { canonicalName = "Hmong", otherNames = {"Pahawh Hmong"}, characters = "𖬀-𖮏", } m["Ibrn"] = { canonicalName = "Iberian", } m["Imag"] = { -- To be used to avoid any formatting or link processing canonicalName = "Image-rendered", -- This should not have any characters listed character_category = false, } m["Inds"] = { canonicalName = "Indus", otherNames = {"Harappan", "Indus Valley"}, } m["IPAchar"] = { canonicalName = "International Phonetic Alphabet", } m["Ital"] = { canonicalName = "Old Italic", characters = "𐌀-𐌣", } m["Java"] = { canonicalName = "Javanese", characters = "ꦀ-꧟", systems = {"abugida"}, } m["Jurc"] = { canonicalName = "Jurchen", } m["Kali"] = { canonicalName = "Kayah Li", characters = "꤀-꤯", } m["Khar"] = { canonicalName = "Kharoshthi", characters = "𐨀-𐩘", systems = {"abugida"}, direction = "rtl", } m["Khmr"] = { canonicalName = "Khmer", characters = "ក-៹᧠-᧿", systems = {"abugida"}, } m["Khoj"] = { canonicalName = "Khojki", characters = "𑈀-𑈽", systems = {"abugida"}, } m["Kitl"] = { canonicalName = "Khitan Large", } m["Kits"] = { canonicalName = "Khitan Small", } m["Knda"] = { canonicalName = "Kannada", characters = "ಂ-ೲ", systems = {"abugida"}, } m["Kthi"] = { canonicalName = "Kaithi", characters = "𑂀-𑃁", systems = {"abugida"}, } m["Lana"] = { canonicalName = "Tai Tham", otherNames = {"Tham", "Tua Mueang", "Lanna"}, characters = "ᨠ-᪭", } m["Laoo"] = { canonicalName = "Lao", characters = "ກ-ໟ", systems = {"abugida"}, } m["Latn"] = { canonicalName = "Latin", otherNames = {"Roman", "Rumi", "Romaji", "Rōmaji"}, characters = "A-Za-zÀ-ÖØ-öø-ɏḀ-ỿ", } m["Latf"] = { canonicalName = "Fraktur", otherNames = {"Blackletter"}, characters = m["Latn"].characters, } m["Latinx"] = { canonicalName = "Latin", characters = m["Latn"].characters .. "Ⱡ-Ɀ꜠-ꟿꬰ-ꭥ", parent = "Latn", } m["nv-Latn"] = { canonicalName = "Latin", characters = m["Latn"].characters, parent = "Latn", } m["pjt-Latn"] = { canonicalName = "Latin", characters = m["Latn"].characters, parent = "Latn", } m["Leke"] = { canonicalName = "Leke", systems = {"abugida"}, } m["Lepc"] = { canonicalName = "Lepcha", characters = "ᰀ-ᱏ", systems = {"abugida"}, } m["Limb"] = { canonicalName = "Limbu", characters = "ᤀ-᥏", systems = {"abugida"}, } m["Lina"] = { canonicalName = "Linear A", characters = "𐘀-𐝧", } m["Linb"] = { canonicalName = "Linear B", characters = "𐀀-𐃺", } m["Lisu"] = { canonicalName = "Lisu", otherNames = {"Fraser"}, characters = "ꓐ-꓿", } m["Lyci"] = { canonicalName = "Lycian", characters = "𐊀-𐊜", } m["Lydi"] = { canonicalName = "Lydian", characters = "𐤠-𐤿", } m["Mahj"] = { canonicalName = "Mahajani", characters = "𑅶-𑅐", systems = {"abugida"}, } m["Mand"] = { canonicalName = "Mandaic", characters = "ࡀ-࡞", direction = "rtl", } m["Mani"] = { canonicalName = "Manichaean", characters = "𐫀-𐫶", direction = "rtl", } m["Maya"] = { canonicalName = "Maya", otherNames = {"Maya hieroglyphic", "Mayan", "Mayan hieroglyphic"}, } m["Mend"] = { canonicalName = "Mende", otherNames = {"Mende Kikakui"}, characters = "𞠀-𞣖", direction = "rtl", } m["Merc"] = { canonicalName = "Meroitic cursive", characters = "𐦠-𐦿", systems = {"abugida"}, } m["Mero"] = { canonicalName = "Meroitic hieroglyphic", characters = "𐦀-𐦟", systems = {"abugida"}, } m["Mlym"] = { canonicalName = "Malayalam", characters = "ം-ൿ", systems = {"abugida"}, } m["Modi"] = { canonicalName = "Modi", characters = "𑘀-𑙙", systems = {"abugida"}, } m["Mong"] = { canonicalName = "Mongolian", characters = "᠀-ᢪ", direction = "down", } m["Morse"] = { canonicalName = "Morse code", } m["Mroo"] = { canonicalName = "Mro", characters = "𖩀-𖩯", } m["Mtei"] = { canonicalName = "Meitei Mayek", characters = "ꯀ-ꫠ-", } m["Mult"] = { canonicalName = "Multani", characters = "𑊀-𑊩", systems = {"abugida"}, } m["musical"] = { canonicalName = "Musical notation", characters = "𝄀-𝇝", } m["Mymr"] = { canonicalName = "Burmese", otherNames = {"Myanmar"}, characters = "က-႟ꩠ-ꩿꧠ-ꧾ", systems = {"abugida"}, } m["Narb"] = { canonicalName = "Old North Arabian", characters = "𐪀-𐪟", } m["Nbat"] = { canonicalName = "Nabataean", otherNames = {"Nabatean"}, characters = "𐢀-𐢯", direction = "rtl", } m["Newa"] = { canonicalName = "Newa", otherNames = {"Newar", "Newari"}, characters = "𑐀-𑑝", systems = {"abugida"}, } m["Nkoo"] = { canonicalName = "N'Ko", characters = "߀-ߺ", direction = "rtl", } m["None"] = { canonicalName = "Unspecified", -- "The word 'script' is automatically added, so this renders as 'unspecified script'." -- This should not have any characters listed character_category = false, -- none } m["Nshu"] = { canonicalName = "Nushu", otherNames = {"Nüshu"}, characters = u(0x1B170).."-"..u(0x1B2FB), } m["Ogam"] = { canonicalName = "Ogham", characters = " -᚜", } m["Olck"] = { canonicalName = "Ol Chiki", characters = "᱐-᱿", } m["Orkh"] = { canonicalName = "Orkhon runes", characters = "𐰀-𐱈", direction = "rtl", } m["Orya"] = { canonicalName = "Oriya", characters = "ଁ-୷", } m["Osge"] = { canonicalName = "Osage", characters = "𐒰-𐓻", } m["Osma"] = { canonicalName = "Osmanya", characters = "𐒀-𐒩", } m["Palm"] = { canonicalName = "Palmyrene", characters = "𐡠-𐡿", } m["Pauc"] = { canonicalName = "Pau Cin Hau", characters = "𑫀-𑫸", } m["Perm"] = { canonicalName = "Old Permic", characters = "𐍐-𐍺", } m["Phag"] = { canonicalName = "Phags-pa", characters = "ꡀ-꡷", systems = {"abugida"}, } m["Marc"] = { canonicalName = "Marchen", characters = "𑱰-𑲶", systems = {"abugida"}, } m["Phli"] = { canonicalName = "Inscriptional Pahlavi", characters = "𐭠-𐭿", direction = "rtl", } m["Phlp"] = { canonicalName = "Psalter Pahlavi", characters = "𐮀-𐮯", direction = "rtl", } m["Phlv"] = { canonicalName = "Book Pahlavi", direction = "rtl", -- Not in Unicode } m["Phnx"] = { canonicalName = "Phoenician", characters = "𐤀-𐤟", direction = "rtl", } m["Plrd"] = { canonicalName = "Pollard", characters = "𖼀-𖾟", systems = {"abugida"}, } m["Prti"] = { canonicalName = "Parthian", characters = "𐭀-𐭟", direction = "rtl", } m["Rjng"] = { canonicalName = "Rejang", characters = "ꤰ-꥟", systems = {"abugida"}, } m["Ruminumerals"] = { canonicalName = "Rumi numerals", characters = "𐹠-𐹾", character_category = "Rumi numerals", } m["Runr"] = { canonicalName = "Runic", characters = "ᚠ-ᛰ", } m["Samr"] = { canonicalName = "Samaritan", characters = "ࠀ-࠾", direction = "rtl", } m["Sarb"] = { canonicalName = "Old South Arabian", characters = "𐩠-𐩿", direction = "rtl", } m["Saur"] = { canonicalName = "Saurashtra", characters = "ꢀ-꣙", } m["Semap"] = { canonicalName = "flag semaphore", } m["Sgnw"] = { canonicalName = "SignWriting", characters = "𝠀-𝪯", } m["Shaw"] = { canonicalName = "Shavian", characters = "𐑐-𐑿", } m["Shrd"] = { canonicalName = "Sharada", characters = "𑆀-𑇙", systems = {"abugida"}, } m["Sidd"] = { canonicalName = "Siddham", characters = "𑖀-𑗝", systems = {"abugida"}, } m["Sind"] = { canonicalName = "Khudawadi", characters = "𑊰-𑋹", systems = {"abugida"}, } m["Sinh"] = { canonicalName = "Sinhalese", characters = "ං-෴", systems = {"abugida"}, } m["Sora"] = { canonicalName = "Sorang Sompeng", otherNames = {"Sora Sompeng"}, characters = "𑃐-𑃹", } m["Soyo"] = { canonicalName = "Soyombo", characters = u(0x11A50).."-"..u(0x11AA2) } m["Sund"] = { canonicalName = "Sundanese", characters = "ᮀ-ᮿ", systems = {"abugida"}, } m["Sylo"] = { canonicalName = "Syloti Nagri", otherNames = {"Sylheti Nagari"}, systems = {"abugida"}, } m["Syrc"] = { canonicalName = "Syriac", characters = "܀-ݏ"..u(0x0860).."-"..u(0x086A), direction = "rtl", } m["Tagb"] = { canonicalName = "Tagbanwa", characters = "ᝠ-ᝳ", systems = {"abugida"}, } m["Takr"] = { canonicalName = "Takri", characters = "𑚀-𑛉", systems = {"abugida"}, } m["Tale"] = { canonicalName = "Tai Nüa", otherNames = {"Tai Nuea", "New Tai Nüa", "New Tai Nuea", "Dehong Dai", "Tai Dehong", "Tai Le"}, characters = "ᥐ-ᥴ", systems = {"abugida"}, } m["Talu"] = { canonicalName = "New Tai Lue", characters = "ᦀ-᧟", systems = {"abugida"}, } m["Taml"] = { canonicalName = "Tamil", characters = "ஂ-௺", systems = {"abugida"}, } m["Tang"] = { canonicalName = "Tangut", characters = "𖿠𗀀-𘟬𘠀-𘫲", } m["Tavt"] = { canonicalName = "Tai Viet", characters = "ꪀ-꫟", systems = {"abugida"}, } m["Telu"] = { canonicalName = "Telugu", characters = "ఁ-౿", systems = {"abugida"}, } m["Teng"] = { canonicalName = "Tengwar", } m["Tfng"] = { canonicalName = "Tifinagh", otherNames = {"Libyco-Berber", "Berber"}, characters = "ⴰ-⵿", } m["Tglg"] = { canonicalName = "Tagalog", characters = "ᜀ-᜔", } m["Thaa"] = { canonicalName = "Thaana", characters = "ހ-ޱ", systems = {"abugida"}, direction = "rtl", } m["Thai"] = { canonicalName = "Thai", characters = "ก-๛", systems = {"abugida"}, } m["Tibt"] = { canonicalName = "Tibetan", characters = "ༀ-࿚", systems = {"abugida"}, } m["Tirh"] = { canonicalName = "Tirhuta", characters = "𑒀-𑓙", systems = {"abugida"}, } m["xzh-Tibt"] = { canonicalName = "Zhang-Zhung", systems = {"abugida"}, } m["Ugar"] = { canonicalName = "Ugaritic", characters = "𐎀-𐎟", } m["Vaii"] = { canonicalName = "Vai", characters = "ꔀ-ꘫ", } m["Wara"] = { canonicalName = "Varang Kshiti", characters = "𑢠-𑣿", } m["Xpeo"] = { canonicalName = "Old Persian", characters = "𐎠-𐏕", } m["Xsux"] = { canonicalName = "Cuneiform", otherNames = {"Sumero-Akkadian Cuneiform"}, characters = "𒀀-𒍮𒐀-𒑳", } m["Yiii"] = { canonicalName = "Yi", characters = "ꀀ-꓆", } m["Zanb"] = { canonicalName = "Zanabazar Square", characters = u(0x11A00).."-"..u(0x11A47), } m["Zmth"] = { canonicalName = "mathematical notation", characters = "ℵ∀-⋿⟀-⟯⦀-⧿⨀-⫿𝐀-𝟿", character_category = "Mathematical notation symbols", -- ? } m["Zsym"] = { canonicalName = "symbol", characters = "─-➿←⇿⌀-⏳🌀-🛅", character_category = false, -- none } m["Zyyy"] = { canonicalName = "undetermined", -- This should not have any characters listed, probably character_category = false, -- none characters = m["Latn"].characters, } m["Zzzz"] = { canonicalName = "uncoded", -- This should not have any characters listed character_category = false, -- none } return m