Modül:ks-Arab-alfabeçeviri

-- TODO: long í support
-- TODO: sort out short e vs palatalisation
-- TODO: add rule for CẹC = CyaC

local u = mw.ustring.char
local gsub = mw.ustring.gsub

local export = {}

local vav = u(0x0648)
local ye = u(0x06CC)
local alif = u(0x0627)
local he = 'ه'

local vw_s_cfu = u(0x0650) -- pesh (i)
local vw_s_ccu = u(0x0655) -- hamza below
local vw_s_cbr = u(0x064F) -- zer (u)
local vw_s_mcu = u(0x0654) -- hamza above
local vw_s_ocu = u(0x064E) -- zabar (a)

local vw_l_cbr = u(0x0657) -- inverted zer
local vw_l_cfu = u(0x0656) -- subscript alif

local hat = u(0x065A)
local inverted_hat = u(0x065B)
local hats = hat .. inverted_hat

local short_vowels_list = vw_s_cfu .. vw_s_ccu .. vw_s_cbr .. vw_s_mcu .. vw_s_ocu 

-- carrier + diacritic combos
local long_u = vav .. vw_l_cbr
local short_o = vav .. inverted_hat
local long_i = ye .. vw_l_cfu
local short_e = ye .. inverted_hat

local vocalised_carrier = long_u .. short_o .. long_i .. short_e .. 'ۆ' .. 'ێ' .. 'ۆ' .. 'اٟ'

local palatalisers = "ۍؠ"

local consonants = "بپتٹجچدڈرڑزژسشفکگلمنهھےثحخذصضطظعغقۍۄیٲآ"

local conv = {
	-- consonants
	 = 'b',  = 'p',  = 't',  = 'ṭ',  = 's',
	 = 'j',  = 'c',  = 'h',  = 'x',
	 = 'd',  = 'ḍ',  = 'z',
	 = 'r',  = 'ḍ',  = 'z',  = 'ć',
	 = 's',  = 'ś',  = 's',  = 'z',
	 = 't',  = 'z',
	 = 'ʿ',  = 'ġ',
	 = 'f',  = 'q',
	 = 'k',  = 'g',
	 = 'l',  = 'm',  = 'n',
	 = 'h',  = 'h',
	
	-- always word-final
	 = 'y',
	
	-- incorrect palatalisation marker
	 = '\'',

	-- broken/open vowels
	 = 'ọ',  = 'ẹ', -- optionally ẹ = ya
	
	-- a carries long vowels
	 = 'ạ̄',  = 'ā',

     = 'ī',  = 'ụ̄',

    -- vowels
     = 'o',  = 'o',  = 'e',  = 'ạ',
	
	-- numerals
	 = '0',  = '1',  = '2',  = '3',  = '4',  = '5',  = '6',  = '7',  = '8',  = '9',
}

local short_vowels = {
	-- high vowels
	 = 'i',  = 'ụ',  = 'u',  = 'ụ̄',
	
	-- central vowels
	 = 'ạ',
	
	-- low vowels
	 = 'a',
}

local alif = 'ا'
local waw = 'و'
local ye = 'ی'

function export.tr(text, lang, sc)
    text = gsub(text, '(%f)', '\'')

    -- interconsonantal vav is a long ō sound
    text = gsub(text,
        '(ھ?)' .. vav .. '()',
        "%1ō%2")

    -- intervocalic alif is a long a sound
	text = gsub(text, '()' .. alif .. '()', "%1ā%2")

    -- final he + short vowel disregards the he and transliterates the vowel
    text = gsub(text, 'ہ()', short_vowels)

    -- word-initial alif + vowelled carrier drops the alif
    text = gsub(text, '^' .. alif .. '()', "%1")

    -- word-initial alif + short vowel diacritic drops the alif
    text = gsub(text, '^' .. alif .. '()', "%1")
	
    -- long /u:/ and /i:/
    text = gsub(text, vav .. vw_s_cbr .. vav .. "()", vav .. "ū%1")
    text = gsub(text, "()" .. vw_s_cfu .. ye .. "()", "%1ī%2")

    -- vav with hat = short o
    text = gsub(text, vav .. "", "o")

    -- vav with short vowel
    text = gsub(text,
        vav .. "()",
        function(c)
            return "v" .. short_vowels
        end)

    -- nun or re with hat
    -- TODO: add support for re
    text = gsub(text, "ن" .. "", "n")

    -- ye with hat = short e
    text = gsub(text, ye .. "", "e")

    -- vav with inverted pish = long u
    text = gsub(text, long_u, "ū")

    -- long i
    text = gsub(text, ye .. vw_l_cfu, 'ī') 

    -- intervocalic ye is a long a sound
    text = gsub(text, '()' .. ye .. '()', "%1ē%2")

    -- word-final alif and ye
    text = gsub(text, '()' .. ye .. '$', "%1ī")
    text = gsub(text, '()' .. alif .. '$', "%1ā")

    -- regard the consonant + short vowel combinations throughout
	text = gsub(text, '.', short_vowels)

	text = gsub(text, '', conv)
	
	-- normal consonants left over
	text = gsub(text, vav, 'v')
	text = gsub(text, 'ہ', 'h')
    text = gsub(text, "ی", "y")
	
	return text
end

return export