Module:mnw-pron

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

local system_list = {
	{ 1,  = "phonetic",		 = "IPA" },
	{ 2,  = "orthographic",	 = "MLCTS" },
	{ 3,  = "orthographic",	 = "ALA-LC" },
	{ 4,  = "phonetic",		 = "BGN/PCGN" },
	{ 5,  = "phonetic",		 = "Okell" },
}

local initial_table = {
		=	{ "k", "k", "k", "k", "k" },
		=	{ "kʰ", "hk", "kh", "hk", "hk" },
		=	{ "kʰ", "hk", "kh", "hk", "hk" },
		=	{ "khac", "khac", "khac", "khac", "khac" },
		=	{ "kɛk", "kɛk", "kɛk", "kɛk", "kɛk" },
		=                { "kac", "kac", "kac", "kac", "kac" },
		=	{ "koc", "koc", "koc", "koc", "koc" },
		=	{ "kɤk", "kɤk", "kɤk", "kɤk", "kɤk" },
		=	{ "kɔc", "kɔc", "kɔc", "kɔc", "kɔc" },
		=	{ "kòk", "kòk", "kòk", "kòk", "kòk" },
		=	{ "kɔk", "kɔk", "kɛ̀k", "kɛ̀k", "kɛ̀k" },
		=	{ "kɔˀ", "kɔˀ", "kɔˀ", "kɔˀ", "kɔˀ" },
		=	{ "keˀ", "keˀ", "keˀ", "keˀ", "keˀ" },
		=	{ "kɒˀ", "kɒˀ", "kɒˀ", "kɒˀ", "kɒˀ" },
		=	{ "kɜˀ", "kɜˀ", "kɜˀ", "kɜˀ", "kɜˀ" },

}


local final_table = {
		=	{ "a̰", "a.", "a", "a.", "á" },
		=	{ "ɛk", "ɛk", "ɛk‘", "ɛk", "ɛk" },

}

local nucleus_table = {
		=	{ "à", "a", "a", "a", "a" }, 
		=	{ "ò", "ò", "ò", "ò", "ò" },
}

local indep_letter_table = {
		=	{ "ɛ̀ə", "ɛ̀ə.", "ɛ̀ə", "ɛ̀ə.", "ɛ̀ə" },
		=	{ "ḭ", "i.", "i", "i.", "í" },
		=	{ "ɒə", "ɒə", "ɒə", "ɒə", "ɒə" },
		=	{ "ṵ", "u.", "u", "u.", "aoˀ" },
		=	{ "ù", "u", "ū", "u", "u" },
		=	{ "ey", "ɛ", "ɛ", "ɛ", "èy" },
		=	{ "ɔə", "uə", "ɔ̀ə", "ɔə.", "ùə" },
		=	{ "ò", "èə:", "o", "èə:", "ò" },
		=	{ "ɤ̀", "ao", "o‘", "ao", "ò" },
		=	{ "ɔm", "ɔ̀m.", "ɔˀ", "ɔ̀m", "ɔ̀m" },
		=	{ "ɛ̀h", "ah", "ɛ̀h", "ɛ̀h.", "ɛ̀h" },
}

local tone_table = {
	 = { "́", ":", "″", ":", "̀" },
	 = { "̰", ".", "′", ".", "́" },
}

local repl_string = "(**)(?)"
function syllabify(text)
	text = gsub(text, "('?)(*)", function(a, b)
		if a .. b ~= "" then return a .. " " .. b end
		end)
	
	text = gsub(text, "(?)(.?)(.?)", function(a, b, c)
		return (c == "္" and " "..a..b.." "..c or (c == "်" and " "..a..b..c or " "..a.." "..b..c))
		end) .. " "
	
	text = gsub(text, "(်း?'?)", "%1 ")
	text = gsub(text, "()(်)", "%2%1")
	
	while match(text, repl_string) do
		text = gsub(text, repl_string, "%1 %2")
	end
	
	text = gsub(text, "္", " , ")
	text = gsub(text, " +", " ")
	text = gsub(text, "^ ?(.*) ?$", "%1")
	text = gsub(text, " , ", " ")
	text = gsub(text, " ()", "%1")
	return text
end

function initial_by_char(initial_string, system_index, ref_table)
	local initial_set = {}
	for character in mw.text.gsplit(initial_string, "") do
		local temp_initial = ref_table or error("Initial data not found.")
		table.insert(initial_set, temp_initial or temp_initial)
	end
	return table.concat(initial_set)
end

function generate_respelling(text)
	text = gsub(text, " ", " 　 ")
	text = gsub(text, "ါ", "ာ")
	if match(text, "") then return text end
	text = gsub(text, "(%+?)(*)(/?)(+)", function(voicing_mark, latin_initial, opt_sep, latin_final)
		return 
			voicing_mark .. 
			(reverse_table or initial_by_char(latin_initial, nil, reverse_table)) .. 
			opt_sep ..
			reverse_table
	end)
	return text
end

function process(initial, final, tone, schwa, system, system_index)
	if match(initial .. final, "ွှ?") and system == "phonetic" then
		initial = gsub(initial, "", "")
		final = "ွ" .. final
	else
		initial = gsub(initial, "/", "")
	end
	
	initial_new = system == "phonetic" and gsub(initial, "%+.", initial_voicing) or initial
	
	if indep_letter_table then
		initial_new = match(initial_new, "") and "-" or ""
		final = initial .. final
	end

	initial_data = 
		initial_table
		or initial_table", "")]
		or (system == "orthographic" 
			and initial_by_char(initial_new, system_index, initial_table)
			or error("Initial data not found."))
		
	initial_value = initial_data or initial_data

	if match(initial, "^%+") and system_index == 5 then
		initial_value = initial_table
		initial_value = gsub(initial_value, "^(+)", "<u>%1</u>")
	end

	final_data =
		final_table .. schwa == "phonetic'" and schwa or final]
		or (system == "phonetic" 
			and (final_table or indep_letter_table) 
			or indep_letter_table)
		or gsub(final, "^(*)()(်?)$", function(first, second, third) 
			first_data = nucleus_table or final_table or indep_letter_table or first
			second_data = initial_table or second
			first = first_data ~= first and first_data or first
			second = second_data ~= second
				and second_data .. ((system_index == 3 and third ~= "") and "‘" or "")
				or second
			return (gsub(first .. second, "()(.*)", "%2"))
			end)
		
	final_value = type(final_data) == "table" and final_data or final_data
	final_value = mw.ustring.toNFD(final_value)
	if tone == "" then
		tone_value = ""
	else
		if system_index ~= 4 then final_value = gsub(final_value, "̀", "") end
		final_value = gsub(final_value, "", "")
		if system .. schwa == "phonetic'" then
			tone_value = ""
		else
			tone_data = tone_table or error("Tone data not found.")
			tone_value = tone_data
		end
	end

	if system_index == 1 then
		final_value = gsub(final_value, "^()", "%1" .. tone_value)
	elseif system_index == 5 then
		final_value = gsub(final_value, "()(*)$", "%1" .. tone_value .. "%2")
	else
		final_value = final_value .. tone_value
	end
	
	return mw.ustring.toNFC(initial_value .. final_value)
end

function remove_wide_space(text)
	return (gsub(text, "　", ""))
end

function concatenate(set, system_index)
	if system_index == 1 then return remove_wide_space(table.concat(set)) end
	result_text = remove_wide_space(table.concat(set, " "))
	
	for count = 1, 3 do
		result_text = gsub(result_text, "(.) (.)(?)",
			function(previous, next, after_next)
				if ambig_intersyl 
				or ((system_index == 2 or system_index == 4)
					and (match(previous .. " " .. next, " ")
					or (match(previous .. next .. after_next, "") and not match(after_next, "")))) then
						return previous .. "-" .. next .. after_next
				else
					return previous .. next .. after_next
				end
			end)
	end
		
	return result_text
end

function export.get_romanisation(word, pronunciations, system, system_index, mode)
	local sentences = {}
	word = gsub(word, " ", "|")
	word = syllabify(word)
	word = gsub(word, "ါ", "ာ")
	if system == "phonetic" then
		word = gsub(word, "ဝ()", "ဝွ%1")
	end
	for phrase in mw.text.gsplit(word, "|", true) do
		local temp = {}
		local syllable = mw.text.split(phrase, " ", true)
		for syllable_index = 1, #syllable do
			syllable = gsub(syllable, "()(်)", "%2%1")
			temp = gsub(
				syllable, 
				"^(*?ွ?ှ?/?)(*)(?)('?)$",
				function(initial, final, tone, schwa)
					return process(initial, final, tone, schwa, system, system_index)
				end)
		end
		table.insert(sentences, concatenate(temp, system_index))
	end
	if mode == "translit_module" then return table.concat(sentences, " ") end
	table.insert(pronunciations, table.concat(sentences, " "))
	return pronunciations
end

function respelling_format(phonetic, page_title)
	local page_title_set = mw.text.split(syllabify(page_title), " ")
	local new_respellings = {}
	for _, respelling in ipairs(phonetic) do
		local respelling_set = mw.text.split(syllabify(respelling), " ")
		if gsub(table.concat(respelling_set), "", "") == (gsub(table.concat(page_title_set), "ါ", "ာ")) then
			for index, element in ipairs(respelling_set) do
				if element ~= page_title_set then
					respelling_set = '<span style="font-size:110%; color:#A32214; font-weight: bold">' .. element .. '</span>'
				end
			end
		end
		table.insert(new_respellings, table.concat(respelling_set))
	end
	text = table.concat(new_respellings, ", ")
	text = remove_wide_space(text)
	text = gsub(text, ".", initial_voicing)
	text = gsub(text, "(ေ?)ာ", "%1ါ")
	return text
end

function export.generate_tests(word, respelling)
	respelling, word = generate_respelling(respelling), generate_respelling(word)
	local pronunciations = {
		 = {},
		 = {},
		 = {},
		 = {},
		 = {},
	}
	local p, result = {  = word,  = respelling or word }, {}
	
	table.sort(system_list, function(first, second) return first < second end)
	for system_index, system in ipairs(system_list) do
		pronunciations = export.get_romanisation(p], pronunciations, system, system_index)
	end
	for system_index = 1, 5 do
		table.insert(result, table.concat(pronunciations))
	end
	return (gsub(gsub(table.concat(result, " | "), "<u>", "("), "</u>", ")"))
end

function export.make(frame)
	local args = frame:getParent().args
	local page_title = mw.title.getCurrentTitle().text
	local title = generate_respelling(args or page_title)
	
	local p, result = {  = { title },  = {} }, {}
	local pronunciations = {
		 = {},
		 = {},
		 = {},
		 = {},
		 = {},
	}

	if not args then args = { title } end
	for index, item in ipairs(args) do
		table.insert(p, (item ~= "") and generate_respelling(item) or nil)
	end
	
	table.sort(system_list, function(first, second) return first < second end)
	for system_index, system in ipairs(system_list) do
		for _, word in ipairs(p]) do
		 	pronunciations = export.get_romanisation(word, pronunciations, system, system_index)
		end
	end
	
	if title ~= table.concat(args) then
		table.insert(result, 
			"* phonetic" .. (#p > 1 and "s" or "") .. ": " ..
			tostring( mw.html.create( "span" )
				:attr( "lang", "mnw" )
				:attr( "class", "mnw" )
				:wikitext( respelling_format( p, page_title ))) .. "\n" )
	end

	table.insert(result,
		'* ]' ..
		'<sup>(])</sup>: ' ..
		
		(tostring( mw.html.create( "span" )
			:attr( "class", "IPA" )
			:wikitext( "/" .. gsub(table.concat(pronunciations, "/, /"), "ʔʔ", "ʔ.ʔ") .. "/" ))) ..
		
		'\n* ] ')
		
	for system_index = 2, 5 do
		table.insert(result, 
			(system_index ~= 2 and " • " or "") ..
			"''" .. system_list .. ":'' " .. 
			table.concat(pronunciations, "/"))
	end
	
	return table.concat(result)
end

return export