Modül:lo-alfabeçeviri

local export = {}
local gsub = mw.ustring.gsub
local len = mw.ustring.len
local match = mw.ustring.match
local sub = mw.ustring.sub

local initial_conv = {
	 = "k",  = "kh",  = "kh",  = "ng", 
	 = "ch",  = "s",  = "s",  = "ny", 
	 = "d",  = "t",  = "th",  = "th",  = "n", 
	 = "b",  = "p",  = "ph",  = "f",  = "ph",  = "f",  = "m", 
	 = "y",  = "r",  = "l",  = "w",
	 = "h",  = "’",  = "h",
	
	 = "ng",
	 = "ny",
	 = "n",  = "n", 
	 = "m",  = "m",
	 = "r",  = "r",
	 = "l",  = "l",
	 = "w"
}

local vowel_conv = {
	 = "a",  = "a", 
	 = "i", 
	 = "ư",  = "u", 
	 = "e",  = "e", 
	 = "æ",  = "æ", 
	 = "o",  = "o", 
	 = "ǫ",  = "ǫ", 
	 = "œ", 
	 = "ia",  = "ia", 
	 = "ưa", 
	 = "ua",  = "ua",  = "ua",
	 = "ai",  = "ai",  = "ai", 
	 = "ao", 
	 = "uau",
	 = "am",  = "am", 
	 = "uam",
	
	 = "ā", 
	 = "āo", 
	 = "ī", 
	 = "ư̄", 
	 = "ū", 
	 = "ē", 
	 = "ǣ", 
	 = "ō", 
	 = "ōi",  = "ōi",
	 = "ǭ",  = "ǭ", 
	 = "ǭi",  = "ǭi",
	 = "œ̄", 
	 = "œ̄i",  = "œ̄i",
	 = "īa",  = "īa",  = "īa", 
	 = "ư̄a", 
	 = "ūa",  = "ūa", 
	 = "uāi",  = "uāi",  = "uīan",
	 = "āi",  = "āi",
	 = "uā",
	 = "uāi",  = "uāi",
	 = "ǣu", -- ແ_ວ can bo both ǣu and uǣ. The first is more common.
	 = "īu",  = "iu",
	 = "iāu",
	 = "uīu",
}

local coda_conv = {
	 = "k",  = "k",  = "k", 
	 = "ng", 
	 = "t",  = "t",  = "t", 
	 = "ny", 
	 = "t",  = "t",  = "t",  = "t", 
	 = "n", 
	 = "p",  = "p",  = "p",  = "p",  = "p",  = "p", 
	 = "m", 
	 = "y", 
	 = "n",  = "n", 
	 = "w",
	 = "",
}

local sp_symbols = {
	 = "〃",  = "〃",
	 = "",
	 = "0",  = "1",  = "2",  = "3",  = "4",
	 = "5",  = "6",  = "7",  = "8",  = "9"
}

local char_type = {
	 = 'coda',	 = 'coda',	 = 'coda',	 = 'coda',
	 = 'coda',	 = 'coda',	 = 'ambig',
	 = 'coda',	 = 'coda',	 = 'coda',	 = 'coda',	 = 'coda',
	 = 'coda',	 = 'coda',	 = 'coda',	 = 'coda',	 = 'coda',	 = 'coda',	 = 'coda',
	 = 'coda',	 = 'coda',	 = 'coda',	 = 'ambig',
	 = 'coda',	 = 'cons',	 = 'ambig',  = 'cons',
	 = 'iter_symbol',
	 = 'vowel_let',  = 'suf_vowel',  = 'vowel_let',  = 'suf_vowel',
	 = 'suf_vowel',  = 'suf_vowel',  = 'suf_vowel',  = 'suf_vowel',
	 = 'suf_vowel',  = 'suf_vowel',  = 'suf_vowel',
	 = 'cons',
	 = 'vowel_let',
	 = 'pref_vowel',  = 'pref_vowel',
	 = 'pref_vowel',  = 'pref_vowel',  = 'pref_vowel',
	 = 'iter_symbol',
	 = 'tone',  = 'tone',  = 'tone',  = 'tone',
	 = 'canc_symbol',  = 'suf_vowel',
	 = 'number',  = 'number',  = 'number',  = 'number',  = 'number',
	 = 'number',  = 'number',  = 'number',  = 'number',  = 'number',
	 = 'cons',  = 'cons'
}

function export.tr(text, lang, sc, debug_mode)
	text = gsub(text, '', '')

	for lao_text in mw.ustring.gmatch(text, '+') do
		local word, c, chartype, output = {}, {}, {}, {}
		local curr_word, curr_initial, curr_vowel, curr_coda = {}, {}, {}, {}
		local i = 1
		local original_text = lao_text
		
		for i = 1, len(lao_text) do
			c = sub(lao_text, i, i)
			chartype = char_type] or table.insert(word, c)
		end
	
		for i = 1, #c + 1 do
			if chartype == 'pref_vowel' or i == #c + 1 then
				if #curr_word ~= 0 then
					table.insert(word, table.concat(curr_word))
					curr_word, curr_initial, curr_vowel, curr_coda = {c}, {}, {c}, {}
				else
					table.insert(curr_vowel, c)
					table.insert(curr_word, c)
				end
				
			elseif chartype == 'suf_vowel' then
				table.insert(curr_vowel, c)
				table.insert(curr_word, c)
				
			elseif chartype == 'ambig' then
				if #curr_initial ~= 0 and vowel_conv] and (chartype ~= 'suf_vowel' or match(c, '')) and #curr_coda == 0 then
					table.insert(curr_vowel, c)
					table.insert(curr_word, c)
				elseif (#curr_initial == 0 and char_type == 'pref_vowel') or (#curr_initial ~= 0 and initial_conv]) then
					table.insert(curr_initial, c)
					table.insert(curr_word, c)
				else
					if #curr_word ~= 0 then
						table.insert(word, table.concat(curr_word))
					end
					curr_word, curr_initial, curr_vowel, curr_coda = {c}, {c}, {}, {}
				end
			
			elseif chartype == 'vowel_let' then
				table.insert(curr_vowel, c)
				table.insert(curr_word, c)
		
			elseif chartype == 'coda' and #curr_coda == 0 and #curr_initial ~= 0 and chartype ~= 'suf_vowel' and chartype ~= 'vowel_let' and not (chartype == 'ambig' and match(chartype or '', 'co')) and table.concat(curr_vowel) ~= "ວີວ" then
				table.insert(curr_coda, c)
				table.insert(curr_word, c)

			elseif chartype == 'cons' or chartype == 'coda' then
				if #curr_coda == 0 and initial_conv] and (#curr_vowel == 0 or char_type == 'pref_vowel') then
					table.insert(curr_initial, c)
					table.insert(curr_word, c)
				else
					table.insert(word, table.concat(curr_word))
					curr_word, curr_initial, curr_vowel, curr_coda = {c}, {c}, {}, {}
				end
			
			elseif chartype == 'iter_symbol' then
				if #curr_word ~= 0 then
					table.insert(word, table.concat(curr_word))
				end
				curr_word, curr_initial, curr_vowel, curr_coda = {c}, {}, {}, {}
		
			elseif chartype == 'canc_symbol' then
				table.insert(curr_word, c)
		
			elseif chartype == 'number' then
				table.insert(curr_word, sp_symbols])
			end
		end
		for i = 1, #word do
			word = gsub(word, '^(?)(ຫ?ຼ?)(*)(໌?)$', function(a, b, c, e)
				if match(sub(c, -1, -1), '') then
					d = sub(c, -1, -1)
					c = sub(c, 1, -2)
				else
					d = ''
				end
				if a..c == '' then
					c = 'ະ'
				end
				vowel = vowel_conv or (vowel_conv or a .. c) .. (coda_conv or d)
				if match(vowel, '') then
					vowel = gsub(vowel, '^(.*)()(.*)$', function(x, y, z)
						return (vowel_conv or x) .. ' ' .. (initial_conv or y) .. (vowel_conv or z) end)
				end
				return (initial_conv or b) .. vowel .. e end)
			
			if char_type] == 'iter_symbol' and i >= 2 then
				word = '<small><u>' .. word .. '</u></small>'
			end
		
			if match(word, '໌') and len(word) > 1 then
				word = gsub(word, '(.)໌', '<small><del>%1</del></small>')
			end
			
			table.insert(output, word)
		end
		lao_text = table.concat(output, " ")
		lao_text = gsub(lao_text, '.', sp_symbols)
	
		text = gsub(text, original_text, lao_text)
	end

	if match(text, '') and not debug_mode then
		return nil
	else
		return text
	end
end

return export