Module:xchc-translit: Difference between revisions

← Older edit

Module:xchc-translit (view source)

Revision as of 16:02, 26 November 2023

175 bytes added , 26 November 2023

no edit summary

Sware

Bureaucrats, Administrators

45,362

edits

@@ Line 1: / Line 1: @@
 local u = mw.ustring.char
-local M = u(0x0304)
+local M = u(0x0304) -- COMBINING MACRON
-local B = u(0x0306)
+local B = u(0x0306) -- COMBINING BREVE
-local D = u(0x0308)
+local D = u(0x0308) -- COMBINING DIAERESIS
-local MB = M .. B
+local A = u(0x0301) -- COMBINING ACUTE
+local DA = u(0x030B) -- COMBINING DOUBLE ACUTE
 local export = {}
-local tt = { -- Unofficial transliteration loosely based on ISO 9
+local MorA = "[" .. M .. A .. "]"
-	["І"]='J', ["і"]='j',
-	["Э"]='E', ["э"]='e', ["Ē" .. B]='Ḗ', ["ē" .. B]='ḗ', ["Ĕ"]='É', ["ĕ"]='é',
+local tt_lepik = {
-	["О"]='O', ["о"]='o', ["Ō" .. B]='Ṓ', ["ō" .. B]='ṓ', ["Ŏ"]='Ó', ["ŏ"]='ó',
+	["Е" .. M .. D]='Jo' .. M, ["е" .. M .. D]='jo' .. M,
+	["Е" .. D]='Jo', ["е" .. D]='jo',
+	["А"]='A', ["а"]='a',
+	["Я"]='Ja', ["я"]='ja',
-	["Е" .. M .. D]='Jō', ["е" .. M .. D]='jō', ["Ё"]='Jo', ["ё"]='jo',
-	["А" .. M]='Ā', ["а" .. M]='ā', ["А"]='A', ["а"]='a',
 	["Б"]='B', ["б"]='b',
 	["В"]='V', ["в"]='v',
 	["Г"]='G', ["г"]='g',
-	["Ғ"]='Ǧ', ["ғ"]='ǧ',
+	["Ғ"]='Gh', ["ғ"]='gh',
-	["Д"]='D', ["д"]='d',
-	["Е" .. M]='Jē', ["е" .. M]='jē', ["Е"]='Je', ["е"]='je',
+	["Д"]='D', ["д"]='d', ["З"]='Z', ["з"]='z', ["Ж"]='Ž', ["ж"]='ž',
-	["Ж"]='Ž', ["ж"]='ž',
-	["З"]='Z', ["з"]='z',
+	["Ы"]='I', ["ы"]='i',
-	["Ӣ"]='Jī', ["ӣ"]='jī', ["И"]='Ji', ["и"]='ji',
+	["И"]='Ji', ["и"]='ji',
 	["К"]='K', ["к"]='k',
 	["Л"]='L', ["л"]='l',
 	["М"]='M', ["м"]='m',
 	["Н"]='N', ["н"]='n',
-	["Ӈ"]='Ŋ', ["ӈ"]='ŋ',
+	["Ӈ"]='Ng', ["ӈ"]='ng',
-	["О" .. M]='Ō', ["о" .. M]='ō', ["О"]='O', ["о"]='o',
-	["Ө" .. M]='Ȫ', ["ө" .. M]='ȫ', ["Ө"]='Ö', ["ө"]='ö',
+	["[ОO]" .. B]='A' .. D, ["[оo]" .. B]='a' .. D,
+	["О" .. M .. B]='A' .. DA, ["о" .. M .. B]='a' .. DA,
+	["О"]='O', ["о"]='o',
+	["Ө" .. M]='O' .. DA, ["ө" .. M]='o' .. DA,
+	["Ө"]='O' .. D, ["ө"]='o' .. D,
 	["П"]='P', ["п"]='p',
 	["Р"]='R', ["р"]='r',
 	["С"]='S', ["с"]='s',
 	["Т"]='T', ["т"]='t',
-	["Ӯ"]='Ū', ["ӯ"]='ū', ["У"]='U', ["у"]='u',
-	["Ү" .. M]='Ȳ', ["ү" .. M]='ȳ', ["Ү"]='Y', ["ү"]='y',
+	["У"]='U', ["у"]='u',
+	["Ю"]='Ju', ["ю"]='ju',
+	["Ү" .. M]='U' .. DA, ["ү" .. M]='u' .. DA,
+	["Ү"]='U' .. D, ["ү"]='u' .. D,
 	["Ф"]='F', ["ф"]='f',
-	["ХӀ"]='H', ["хӀ"]='h', ["Х"]='X', ["х"]='x',
-	["Ц"]='Ts', ["ц"]='ts',
+	["ХӀ"]='H', ["хӀ"]='h',
-	["Ч"]='Č', ["ч"]='č',
+	["Х"]='Ch', ["х"]='ch',
+	["Ц"]='C', ["ц"]='c',
+	["Ч"]='Cs', ["ч"]='cs',
 	["Ш"]='Š', ["ш"]='š',
-	["Ы" .. M]='Ī', ["ы" .. M]='ī', ["Ы"]='I', ["ы"]='i',
-	["Ю" .. M]='Jū', ["ю" .. M]='jū', ["Ю"]='Ju', ["ю"]='ju',
-	["Я" .. M]='Jā', ["я" .. M]='jā', ["Я"]='Ja', ["я"]='ja',
-	["L[ьj]"] = "Ľ", ["l[ьj]"] = "ľ",
+	["[ЭE]" .. M .. B]='E' .. DA, ["[эe]" .. M .. B]='e' .. DA,
-	["N[ьj]"]='Ň', ["n[ьj]"]='ň',
+	["Э" .. B]='E' .. D, ["э" .. B]='e' .. D,
-	["[ьЬ]"]="'", ["[Ъъ]"]='"',
+	["Э"]='E', ["э"]='e',
-	["L['j]"]='Ľ', ["l['j]"]='ľ', ["N'"]='Ň', ["n'"]='ň', ["lja"]='lja',
+	["І"]='J', ["і"]='j',
+	["Ь"]='Y', ["ь"]='y',
-	["Й"]='J', ["й"]='j',
+	["[Ъъ]"]="'",
+	["([LNln])j([aeiouy])"]='%1y%2'
 };
 function export.tr(text, lang, sc)
-	for pat, repl in pairs(tt) do
+	text = text:gsub("Й", "J"):gsub("й", "j")
+	text = mw.ustring.toNFD(text)
+	for pat, repl in pairs(tt_lepik) do
 		text = mw.ustring.gsub(text, pat, repl)
 	end
-	return text
+	text = mw.ustring.toNFD(text):gsub("([LNln])j([aeiouy])", "%1y%2")
+								 :gsub("O" .. B, "A" .. D):gsub("o" .. B, "a" .. D)
+								 :gsub(M,A)
+								 :gsub(A .. D, DA):gsub(D .. A, DA)
+	return mw.ustring.toNFC(text):gsub("([Dd])z","%1c"):gsub("([Dd])ž","%1z")
 end
 return export