Modul:translit/be

-- Denna modul används för att transkribera belarusiska/vitryska.
-- 
-- För att transkribera bulgariska, ukrainska och
-- ryska kan ett liknande system användas. Det hanteras dock av
-- särskilda moduler. Se:
-- 
-- Modul:translit/ru
-- Modul:translit/bg
-- Modul:translit/uk
-- 
-- Systemet för translitterering följer det som anges i
-- "Svenska skrivregler" (2017), fjärde upplagan, red. Ola Karlsson
-- En kopia av avsnittet finns i skrivande stund här:
-- https://www.isof.se/download/18.648bef4b18093ee2f03ee80/1652364347434/Kyrillisk%20tabell%20SS2017.pdf
-- 
-- Denna modul används av följande mall:
-- {{tr}}
--
-- Modulen är författad av (och underhålls av) ]

local export = {}
local Q = require("Modul:queue")

local multibyte_char_pattern = ".*"

latin_by_cyrillic = {
	="A", ="a", ="B", ="b", ="V", ="v",
	="H", ="h", ="D", ="d", ="E", ="e",
	="Jo", ="jo", ="Zj", ="zj", ="Z", ="z",
	="I", ="i", ="I", ="i", ="J", ="j",
	="K", ="k", ="L", ="l", ="M", ="m",
	="N", ="n", ="O", ="o", ="P", ="p",
	="R", ="r", ="S", ="s", ="T", ="t",
	="U", ="u", ="Ŭ", ="ŭ", ="F", ="f",
	="Ch", ="ch", ="Ts", ="ts", ="Tj", ="tj",
	="Sj", ="sj", ="Sjtj", ="sjtj", ="", ="",
	="Y", ="y", ="J", ="j", ="E", ="e",
	="Ju", ="ju", ="Ja", ="ja"
}

-- Själva transkriberingsfunktionen för vitryska/belarusiska
function export.tr( text )
	cyrillic_q = Q()
	latin_q    = Q()

	for c in string.gmatch(text, multibyte_char_pattern) do
		Q.enqueue(cyrillic_q, c)
	end

	repeat

		local x = Q.dequeue(cyrillic_q)
		local y = Q.peekFirst(cyrillic_q)
		local z = Q.peekSecond(cyrillic_q)
		local u = Q.peekLast(latin_q)
		local accent = "́"
    
		if not latin_by_cyrillic then -- icke-kyrilliskt tecken
			Q.enqueue(latin_q, x)
		elseif (Q.isEmpty(latin_q) or u == " ") and x == "Е" then
			Q.enqueue(latin_q, "Je")
		elseif (Q.isEmpty(latin_q) or u == " ") and x == "е" then
			Q.enqueue(latin_q, "je")
		elseif x == "ь" then
			if y == "и" then
				Q.enqueue(latin_q, "ji")
				Q.dequeue(cyrillic_q)
			end
		elseif x == "Ь" then
			if y == "и" then
				Q.enqueue(latin_q, "Ji")
				Q.dequeue(cyrillic_q)
			elseif y == "И" then
				Q.enqueue(latin_q, "JI")
				Q.dequeue(cyrillic_q)
			end
		elseif mw.ustring.find(x, "") then
			Q.enqueue(latin_q, latin_by_cyrillic)
			if y == "ь" then
				if z == "е" then
					Q.enqueue(latin_q, "ie")
					Q.dequeue(cyrillic_q)
					Q.dequeue(cyrillic_q)
				elseif z == "ё" then
					Q.enqueue(latin_q, "io")
					Q.dequeue(cyrillic_q)
					Q.dequeue(cyrillic_q)
				elseif z == "ю" then
					Q.enqueue(latin_q, "iu")
					Q.dequeue(cyrillic_q)
					Q.dequeue(cyrillic_q)
				elseif z == "я" then
					Q.enqueue(latin_q, "ia")
					Q.dequeue(cyrillic_q)
					Q.dequeue(cyrillic_q)
				end
			elseif y == "ё" then
				Q.enqueue(latin_q, "io")
				Q.dequeue(cyrillic_q)
			elseif y == "ю" then
				Q.enqueue(latin_q, "iu")
				Q.dequeue(cyrillic_q)
			elseif y == "я" then
				Q.enqueue(latin_q, "ia")
				Q.dequeue(cyrillic_q)
			end
		elseif mw.ustring.find(x, "") then
			Q.enqueue(latin_q, latin_by_cyrillic)
			if y == "ё" then
				Q.enqueue(latin_q, "o")
				Q.dequeue(cyrillic_q)
			end
    	elseif x == "е" then
			if mw.ustring.find(u, "") or u == accent then
				Q.enqueue(latin_q, "je")
			else
				Q.enqueue(latin_q, "e")
			end
		else
			Q.enqueue(latin_q, latin_by_cyrillic)
		end
	until Q.isEmpty(cyrillic_q)

	local tbl = {}

	repeat
    	local x = Q.dequeue(latin_q)
    	local y = Q.peekFirst(latin_q) or ""
		local accent = "́"

		if y == accent then
			x = mw.ustring.toNFC(x .. y)
			Q.dequeue(latin_q)
		end

		table.insert(tbl, x)

	until Q.isEmpty(latin_q)

	return table.concat(tbl, "")
end

return export