module Cadmium::Tokenizer::Diacritics
Direct including types
Defined in:
cadmium/tokenizer/diacritics.cr
Constant Summary
-
DIACRITICS =
[{base: "A", letters: ::Regex.new("AⒶＡÀÁÂẦẤẪẨÃĀĂẰẮẴẲȦǠÄǞẢÅǺǍȀȂẠẬẶḀĄȺⱯ]/")}, {base: "AA", letters: ::Regex.new("Ꜳ]/")}, {base: "AE", letters: ::Regex.new("ÆǼǢ]/")}, {base: "AO", letters: ::Regex.new("Ꜵ]/")}, {base: "AU", letters: ::Regex.new("Ꜷ]/")}, {base: "AV", letters: ::Regex.new("ꜸꜺ]/")}, {base: "AY", letters: ::Regex.new("Ꜽ]/")}, {base: "B", letters: ::Regex.new("BⒷＢḂḄḆɃƂƁ]/")}, {base: "C", letters: ::Regex.new("CⒸＣĆĈĊČÇḈƇȻꜾ]/")}, {base: "D", letters: ::Regex.new("DⒹＤḊĎḌḐḒḎĐƋƊƉꝹ]/")}, {base: "DZ", letters: ::Regex.new("DZDŽ]/")}, {base: "Dz", letters: ::Regex.new("DzDž]/")}, {base: "E", letters: ::Regex.new("EⒺＥÈÉÊỀẾỄỂẼĒḔḖĔĖËẺĚȄȆẸỆȨḜĘḘḚƐƎ]/")}, {base: "F", letters: ::Regex.new("FⒻＦḞƑꝻ]/")}, {base: "G", letters: ::Regex.new("GⒼＧǴĜḠĞĠǦĢǤƓꞠꝽꝾ]/")}, {base: "H", letters: ::Regex.new("HⒽＨĤḢḦȞḤḨḪĦⱧⱵꞍ]/")}, {base: "I", letters: ::Regex.new("IⒾＩÌÍÎĨĪĬİÏḮỈǏȈȊỊĮḬƗ]/")}, {base: "J", letters: ::Regex.new("JⒿＪĴɈ]/")}, {base: "K", letters: ::Regex.new("KⓀＫḰǨḲĶḴƘⱩꝀꝂꝄꞢ]/")}, {base: "L", letters: ::Regex.new("LⓁＬĿĹĽḶḸĻḼḺŁȽⱢⱠꝈꝆꞀ]/")}, {base: "LJ", letters: ::Regex.new("LJ]/")}, {base: "Lj", letters: ::Regex.new("Lj]/")}, {base: "M", letters: ::Regex.new("MⓂＭḾṀṂⱮƜ]/")}, {base: "N", letters: ::Regex.new("NⓃＮǸŃÑṄŇṆŅṊṈȠƝꞐꞤ]/")}, {base: "NJ", letters: ::Regex.new("NJ]/")}, {base: "Nj", letters: ::Regex.new("Nj]/")}, {base: "O", letters: ::Regex.new("OⓄＯÒÓÔỒỐỖỔÕṌȬṎŌṐṒŎȮȰÖȪỎŐǑȌȎƠỜỚỠỞỢỌỘǪǬØǾƆƟꝊꝌ]/")}, {base: "OE", letters: ::Regex.new("Œ]/")}, {base: "OI", letters: ::Regex.new("Ƣ]/")}, {base: "OO", letters: ::Regex.new("Ꝏ]/")}, {base: "OU", letters: ::Regex.new("Ȣ]/")}, {base: "P", letters: ::Regex.new("PⓅＰṔṖƤⱣꝐꝒꝔ]/")}, {base: "Q", letters: ::Regex.new("QⓆＱꝖꝘɊ]/")}, {base: "R", letters: ::Regex.new("RⓇＲŔṘŘȐȒṚṜŖṞɌⱤꝚꞦꞂ]/")}, {base: "S", letters: ::Regex.new("SⓈＳẞŚṤŜṠŠṦṢṨȘŞⱾꞨꞄ]/")}, {base: "T", letters: ::Regex.new("TⓉＴṪŤṬȚŢṰṮŦƬƮȾꞆ]/")}, {base: "TZ", letters: ::Regex.new("Ꜩ]/")}, {base: "U", letters: ::Regex.new("UⓊＵÙÚÛŨṸŪṺŬÜǛǗǕǙỦŮŰǓȔȖƯỪỨỮỬỰỤṲŲṶṴɄ]/")}, {base: "V", letters: ::Regex.new("VⓋＶṼṾƲꝞɅ]/")}, {base: "VY", letters: ::Regex.new("Ꝡ]/")}, {base: "W", letters: ::Regex.new("WⓌＷẀẂŴẆẄẈⱲ]/")}, {base: "X", letters: ::Regex.new("XⓍＸẊẌ]/")}, {base: "Y", letters: ::Regex.new("YⓎＹỲÝŶỸȲẎŸỶỴƳɎỾ]/")}, {base: "Z", letters: ::Regex.new("ZⓏＺŹẐŻŽẒẔƵȤⱿⱫꝢ]/")}, {base: "a", letters: ::Regex.new("aⓐａẚàáâầấẫẩãāăằắẵẳȧǡäǟảåǻǎȁȃạậặḁąⱥɐ]/")}, {base: "aa", letters: ::Regex.new("ꜳ]/")}, {base: "ae", letters: ::Regex.new("æǽǣ]/")}, {base: "ao", letters: ::Regex.new("ꜵ]/")}, {base: "au", letters: ::Regex.new("ꜷ]/")}, {base: "av", letters: ::Regex.new("ꜹꜻ]/")}, {base: "ay", letters: ::Regex.new("ꜽ]/")}, {base: "b", letters: ::Regex.new("bⓑｂḃḅḇƀƃɓ]/")}, {base: "c", letters: ::Regex.new("cⓒｃćĉċčçḉƈȼꜿↄ]/")}, {base: "d", letters: ::Regex.new("dⓓｄḋďḍḑḓḏđƌɖɗꝺ]/")}, {base: "dz", letters: ::Regex.new("dzdž]/")}, {base: "e", letters: ::Regex.new("eⓔｅèéêềếễểẽēḕḗĕėëẻěȅȇẹệȩḝęḙḛɇɛǝ]/")}, {base: "f", letters: ::Regex.new("fⓕｆḟƒꝼ]/")}, {base: "g", letters: ::Regex.new("gⓖｇǵĝḡğġǧģǥɠꞡᵹꝿ]/")}, {base: "h", letters: ::Regex.new("hⓗｈĥḣḧȟḥḩḫẖħⱨⱶɥ]/")}, {base: "hv", letters: ::Regex.new("ƕ]/")}, {base: "i", letters: ::Regex.new("iⓘｉìíîĩīĭïḯỉǐȉȋịįḭɨı]/")}, {base: "j", letters: ::Regex.new("jⓙｊĵǰɉ]/")}, {base: "k", letters: ::Regex.new("kⓚｋḱǩḳķḵƙⱪꝁꝃꝅꞣ]/")}, {base: "l", letters: ::Regex.new("lⓛｌŀĺľḷḹļḽḻſłƚɫⱡꝉꞁꝇ]/")}, {base: "lj", letters: ::Regex.new("lj]/")}, {base: "m", letters: ::Regex.new("mⓜｍḿṁṃɱɯ]/")}, {base: "n", letters: ::Regex.new("nⓝｎǹńñṅňṇņṋṉƞɲʼnꞑꞥ]/")}, {base: "nj", letters: ::Regex.new("nj]/")}, {base: "o", letters: ::Regex.new("oⓞｏòóôồốỗổõṍȭṏōṑṓŏȯȱöȫỏőǒȍȏơờớỡởợọộǫǭøǿɔꝋꝍɵ]/")}, {base: "oe", letters: ::Regex.new("œ]/")}, {base: "oi", letters: ::Regex.new("ƣ]/")}, {base: "ou", letters: ::Regex.new("ȣ]/")}, {base: "oo", letters: ::Regex.new("ꝏ]/")}, {base: "p", letters: ::Regex.new("pⓟｐṕṗƥᵽꝑꝓꝕ]/")}, {base: "q", letters: ::Regex.new("qⓠｑɋꝗꝙ]/")}, {base: "r", letters: ::Regex.new("rⓡｒŕṙřȑȓṛṝŗṟɍɽꝛꞧꞃ]/")}, {base: "s", letters: ::Regex.new("sⓢｓßśṥŝṡšṧṣṩșşȿꞩꞅẛ]/")}, {base: "t", letters: ::Regex.new("tⓣｔṫẗťṭțţṱṯŧƭʈⱦꞇ]/")}, {base: "tz", letters: ::Regex.new("ꜩ]/")}, {base: "u", letters: ::Regex.new("uⓤｕùúûũṹūṻŭüǜǘǖǚủůűǔȕȗưừứữửựụṳųṷṵʉ]/")}, {base: "v", letters: ::Regex.new("vⓥｖṽṿʋꝟʌ]/")}, {base: "vy", letters: ::Regex.new("ꝡ]/")}, {base: "w", letters: ::Regex.new("wⓦｗẁẃŵẇẅẘẉⱳ]/")}, {base: "x", letters: ::Regex.new("xⓧｘẋẍ]/")}, {base: "y", letters: ::Regex.new("yⓨｙỳýŷỹȳẏÿỷẙỵƴɏỿ]/")}, {base: "z", letters: ::Regex.new("zⓩｚźẑżžẓẕƶȥɀⱬꝣ]/")}]
Instance Method Summary
Instance Method Detail
def
remove_diacritics(str : String)
#