Module:mnp-pron

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

local gsub = m_string_utils.gsub
local gsplit = mw.text.gsplit
local match = m_string_utils.match
local lower = m_string_utils.lower
local toNFC = mw.ustring.toNFC
local toNFD = mw.ustring.toNFD

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

local final_ipa = {
	 = "i",  = "u",  = "y",
	 = "a",  = "ia",  = "ua",
	 = "ɛ",  = "iɛ",  = "uɛ",  = "yɛ",
	 = "ɔ",  = "iɔ",
	 = "e",  = "œ",  = "o",
	 = "ai",  = "uai",
	 = "au",  = "iau",
	 = "iu",
	 = "iŋ",  = "uiŋ",  = "yiŋ",
	 = "aŋ",  = "iaŋ",  = "uaŋ",
	 = "ɔŋ",  = "ɔŋ",  = "iɔŋ",  = "iɔŋ",  = "uɔŋ",
	 = "eiŋ", --missing /ieiŋ/
	 = "œyŋ",
	 = "aiŋ",  = "uaiŋ",
}

local tone_marks = ""

local tone_from_mark = {
	 = 1, --陰平 (acute)
	 = 3, --陽平 (circumflex)
	 = 2, --上 (caron)
	 = 3, --陰去 (double overline)
	 = 4, --陽去 (macron)
	 = 5, --陰入 (breve)
	 = 6, --陽入 (grave)
}

local tone_value = {
	 = "⁵⁴", --陰平
	 = "²¹", --上
	 = "³³", --陰去
	 = "⁵⁵", --陽去
	 = "²⁴", --陰入
	 = "⁴²", --陽入
}

--[=[
	Given a syllable, return tone value of the syllable's tone and the syllable without the tone mark
]=]
local function get_tone_value(syllable)
	if match(syllable, "̄̄") then -- check for double macron
		local corrected = gsub(syllable, "̄̄", "̿")
		error("Please change " .. syllable .. " to " .. corrected .. ".")
	end
	local tone = match(syllable, tone_marks)
	if tone then
		return tone_value], gsub(syllable, tone, "")
	else
		error("No tone detected.")
	end
end

function export.rom(text)
	return (text
		:gsub("/", " / ")
		:gsub(">(+)", "<sup>→%1</sup>"))
end

function export.ipa(text)
	text = toNFD(lower(text))
	local words_ipa = {}
	for word in gsplit(text, "/") do
		word = gsub(word, " ", "-")
		local syllables_ipa = {}
		for syllable in gsplit(word, "-") do
			syllable = match(syllable, ">(.*)$") or syllable
			local initial, final, tone_value
			tone_value, syllable = get_tone_value(syllable)
			initial, final = match(syllable, "^(??)(+)$")
			initial, final = toNFC(initial), toNFC(final)
			initial, final = initial_ipa, final_ipa
			table.insert(syllables_ipa, initial..final..tone_value)
		end
		table.insert(words_ipa, table.concat(syllables_ipa, " "))
	end
	return "/" .. table.concat(words_ipa, "/, /") .. "/"
end

return export