Module:luh-pron

local export = {}
local m_string_utils = require("Module:string utilities")

local find = m_string_utils.find
local match = m_string_utils.match
local split = mw.text.split
local gsplit = mw.text.gsplit
local gsub = m_string_utils.gsub

local initial_ipa = {
	 = "p",  = "pʰ",  = "m",  = "b",
	 = "t",  = "tʰ",  = "n",  = "l",
	 = "k",  = "kʰ",  = "ŋ",  = "h",
	 = "t͡s",  = "t͡sʰ",  = "s",  = "z",
	 = "t͡s",  = "t͡sʰ",  = "s",
	 = ""
}

local final_ipa = {
	 = "i",  = "u",
	 = "a",  = "ia",  = "ua",
	 = "ɔ",  = "iɔ",
	 = "ɛ",  = "iɛ",  = "uɛ",
	 = "ai",  = "uai",
	 = "au",  = "iau",
	 = "ɛu",  = "iu",
	 = "ɔi",  = "ui",
	 = "am",  = "iam",  = "m̩",
	 = "em",  = "im",
	 = "aŋ",  = "iaŋ",  = "uaŋ",
	 = "eŋ",  = "ieŋ",  = "ŋ̍",
	 = "ɔŋ",  = "iɔŋ",
	 = "iŋ",  = "uŋ",
	 = "ap̚",  = "iap̚",
	 = "ep̚",  = "ip̚",
	 = "ak̚",  = "iak̚",  = "uak̚",
	 = "ek̚",  = "iek̚",  = "uek̚",
	 = "ik̚",  = "uk̚",
	 = "ɔk̚",  = "iɔk̚",
}

local tone_chao = {
	 = "³⁵",  = "³¹",  = "²¹",  = "⁵",
	 = "²²",  = "³³",  = "⁵⁵",  = "²"
}

-- find the sandhied tone of the first syllable in a two-syllable word
-- returns nil if the tone of the first syllable does not change
local function tone_sandhi(tone1, tone2)
	if tone1 == "1" then
		return "6"
	elseif (tone1 == "2" and match(tone2, "")) or (tone1 == "3" and match(tone2, "")) then
		return "7"
	end
end

-- convert Leizhounese Pinyin to IPA
function export.ipa(text)
	if type(text) == "table" then
		text = text.args
	end
	
	local result = {}
	
	for word in gsplit(text, "/") do
		local syllables = split(word, " ")
		local initial, final, tone, block, sandhi, ipa = {}, {}, {}, {}, {}, {}
		for i, syllable in ipairs(syllables) do
			initial, final, tone, block = match(syllable, "^(??)(??g?)()(#?)$")
		
			-- check z/c/s vs. j/q/x
			if (find(initial, "^$") and find(final, "^i")) or (find(initial, "^$") and find(final, "^")) then
				error("Invalid Leizhou pinyin input \"" .. syllable .. "\": initial " .. initial .. " cannot go with final " .. final .. ".")
			end	
			
			-- check checked tones
			if (find(final, "$") and find(tone, "^$")) or ((find(final, "$") or find(final, "ng$")) and find(tone, "^$")) then
				error("Invalid Leizhou pinyin input \"" .. syllable .. "\": final " .. final .. " cannot go with tone " .. tone .. ".")
			end
		end
		
		for i=1,#syllables,1 do
			if i+1 <= #syllables and block ~= "#" then
				sandhi = tone_sandhi(tone, tone)
			end
		end
		
		for i=1,#syllables,1 do
			actual_tone = tone_chao] .. (sandhi and "⁻" .. tone_chao] or "")
			ipa = initial_ipa] .. final_ipa] .. actual_tone
		end
		table.insert(result, table.concat(ipa, " "))
	end
	
	return "/" .. table.concat(result, "/, /") .. "/"
end

function export.rom(text)
	text = gsub(text, "/", " / ")
	text = gsub(text, '(+)', '<sup>%1</sup>')
	text = gsub(text, '#', '')
	return text
end

return export