Module:zhx-sic-pron

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

local find = m_string_utils.find
local gsub = m_string_utils.gsub
local match = m_string_utils.match
local gmatch = m_string_utils.gmatch
local gsplit = mw.text.gsplit
local lower = m_string_utils.lower
local upper = m_string_utils.upper

local initialConv = {
	 = "p",  = "t",  = "k",
	 = "pʰ",  = "tʰ",  = "kʰ",
	 = "t͡s",  = "t͡ɕ",
	 = "t͡sʰ",  = "t͡ɕʰ",
	 = "m",  = "n",  = "nʲ",  = "ŋ",
	 = "f",  = "s",  = "ɕ",  = "x",
	 = "v",  = "z",
	 = "",
}

-- note that 'ir' is for internal use by the code and not used in actual sichuanese pinyin
local finalConv = {
	 = "z̩",  = "ɚ",

	 = "a",  = "o",  = "ɛ",
	 = "ai",  = "ei",  = "au",  = "əu",
	 = "an",  = "ən",  = "aŋ",  = "oŋ",

	 = "i",  = "ia",  = "iɛ",
	 = "iɛi",  = "iau",  = "iəu",
	 = "iɛn",  = "in",  = "iaŋ",

	 = "u",  = "ua",  = "uɛ",
	 = "uai",  = "uei",
	 = "uan",  = "uən",  = "uaŋ",

	 = "y",  = "yo",  = "ye",
	 = "yan",  = "yn",  = "yoŋ",
}

local toneConv = {
	 = "⁵⁵",  = "²¹",  = "⁵³",  = "²¹³",  = "⁻",
}

local initialConv_swz = {
	 = "g",  = "k",  = "l",  = "n",  = "x",  = "",  = "rh",
}
local finalConv_swz = {
	 = "",  = "r",  = "ung",  = "uong",  = "y",  = "iuo",  = "ye",  = "uan",  = "un",  = "yng",
}

local function fix(initial, final)
	-- ju /tɕy/
	if find(initial, '^$') and find(final, '^u') then
		final = gsub(final, '^u', 'ü')
	end
	
	if initial == 'y' then
		initial = ''
		if final == 'ou' then
			final = 'iu'
		elseif not find(final, '^') then -- yin /in/, yuan /yan/, ya /ia/
			final = 'i' .. final
		end
	end
	
	-- wei /uei/ (/-uei/ is usually spelled <-ui> but /uei/ is not <wui>)
	-- wu /vu/
	-- wai /uai/
	-- wen /uən/
	if initial == 'w' then
		initial = (final == 'u') and 'w' or ''
		if final == 'ei' then
			final = 'ui'
		elseif final == 'en' then
			final = 'un'
		elseif final ~= 'u' then
			final = 'u' .. final
		end
	end

	-- distinguish the two 'i's
	if find(initial, '^$') and final == 'i' then
		final = 'ir'
	end

	return initial, final
end

local function warn(initial, final, tone)
	if initial == "" and find(final, "^") then
		error("Syllables in Sichuanese Pinyin do not begin with i-/u-. Add y-/w-.")
	end
	
	if not initialConv and initial ~= "y" then
		error("Invalid initial: " .. initial)
	end

	if not finalConv and final ~= "uo" then
		error("Invalid final: " .. final)
	end

	if tone == "5" then
		error("Chengdu does not have the fifth tone anymore. Use 2.")
	end
end

function export.convert(text, scheme)
	if type(text) == "table" then
		text, scheme = text.args, text.args
	end

	local result = {}
	for word in gsplit(text, '/') do
		local converted = {}

		local extra2 = match(word, '^*')
		for syllable in gmatch(word, '++*') do
			local initial, final, erhua, tone, extra = match(syllable, '^(??)(+)(r?)(+)(*)$')
			local caps = false

			if find(initial .. final, '') then
				caps = true
				initial, final = lower(initial), lower(final)
			end

			warn(initial, final, tone)

			initial, final = fix(initial, final)
			if final == 'e' and erhua == 'r' then
				final, erhua = 'er', ''
			end

			if scheme == 'IPA' then
				initial = initialConv
				final = finalConv
				tone = gsub(tone, '.', function(char) return toneConv end)

				if erhua == 'r' then
					if find(final, '^y') then -- 撮口呼
						final = 'yɚ'
					elseif find(final, '^i') then -- 齊齒呼
						final = 'iɚ'
					elseif find(final, '^u') then -- 合口呼
						final = 'uɚ'
					elseif (final == 'o' or final == 'oŋ') and find(initial, '^') then
						final = 'ɚ'
					elseif final == 'o' or final == 'oŋ' then
						final = 'uɚ'
					else -- 開口呼
						final = 'ɚ'
					end
				end

				syllable = initial .. final .. tone

				table.insert(converted, syllable)
			elseif scheme == 'SWZ' then
				initial = initialConv_swz or initial
				final = finalConv_swz or final

				tone = gsub(tone, '(%d)%-(%d)', '%2')

				-- XXX: what happens with erhua? (disabled output for now)
				-- cf the given example 貓(mer)
				if erhua == 'r' then return false end

				if tone == '3' and (final == 'a' or final == 'ai') then
					final = 'a' .. final
				end

				syllable = initial .. final

				if caps then syllable = gsub(syllable, '^.', upper) end

				table.insert(converted, '@' .. syllable .. extra)
			else
				error('Convert to what representation?')
			end
		end

		if scheme == 'IPA' then
			local text = '/' .. table.concat(converted, ' ') .. '/'
			table.insert(result, text)
		elseif scheme == 'SWZ' then
			local text = table.concat(converted, '')
			text = gsub(text, '()@(u)', '%1w')
			text = gsub(text, '()@(i)', '%1j')
			text = gsub(text, '()@(y)', '%1j')
			text = gsub(text, '@un', 'wen')
			text = gsub(text, '@', '')
			table.insert(result, extra2 .. text)
		end
	end

	if scheme == 'IPA' then
		return table.concat(result, ', ')
	else
		return table.concat(result, ' / ')
	end
end

return export