Module:cmn-pron-Jianghuai

The following documentation is located at Module:cmn-pron-Jianghuai/documentation. ^[edit]

Useful links: subpage list • links • transclusions • testcases • sandbox

This module does romanisation conversion, IPA conversion, etc. for Jianghuai Mandarin. See {{zh-pron}}.

---

local export = {}
local find = mw.ustring.find
local gsub = mw.ustring.gsub
local match = mw.ustring.match
local gmatch = mw.ustring.gmatch
local gsplit = mw.text.gsplit
local lower = mw.ustring.lower
local upper = mw.ustring.upper

local initialConv = {
	["b"] = "p", ["d"] = "t", ["g"] = "k",
	["p"] = "pʰ", ["t"] = "tʰ", ["k"] = "kʰ",
	["z"] = "t͡s", ["j"] = "t͡ɕ",
	["c"] = "t͡sʰ", ["q"] = "t͡ɕʰ",
	["m"] = "m", ["l"] = "l", 
	["f"] = "f", ["s"] = "s", ["x"] = "ɕ", ["h"] = "x",
	['r'] = 'ʐ', ['zh'] = 'ʈ͡ʂ',
	['ch'] = 'ʈ͡ʂʰ', ['sh'] = 'ʂ',
	[""] = "",
}

-- note that 'ir' is for internal use by the code and not used in actual sichuanese pinyin
local finalConv = {
	["y"] = "ɿ", ["r"] = "ʅ", ["er"] = "ɚ",
	["a"] = "a", ["o"] = "o", ["e"] = "e",
	["ä"] = "ɛ", ["ei"] = "əɪ", ["ao"] = "ɔ", ["ou"] = "əɯ",
	["en"] = "ən", ["än"] = "ẽ", ["ang"] = "ã", ["ong"] = "oŋ",

	["i"] = "i", ["ia"] = "ia", ["ie"] = "ie",
	["iä"] = "iɛ", ["iao"] = "iau", ["iou"] = "iəɯ",
	["iän"] = "iẽ", ["in"] = "in", ["iang"] = "iã", ["iong"] = "ioŋ",

	["u"] = "u", ["ua"] = "ua", ["uä"] = "uɛ",
	["uei"] = "uəɪ", ["uen"] = "uən", ["uang"] = "uã",

	["ü"] = "y", ["üe"] = "ye",
	["üän"] = "yẽ", ["üin"] = "yin", 
}

local toneConv = {
	["1"] = "³¹", ["2"] = "¹³", ["3"] = "²¹²", ["4"] = "⁴⁴", ["5"] = "⁵",   ["-"] = "⁻",
}

local function fix(initial, final)

	return initial, final
end

local function warn(initial, final, tone)
    if (initial == "j" or initial == "q" or initial == "x") and final == "u" then
		error("Syllables in Nankinese Pinyin do not include ju, qu, xu, but include jü, qü, xü.")
	end
	if (initial == "j" or initial == "q" or initial == "x") and ( final == "iao"  or final == "io"  or final == "iang"  or final == "iä"  or final == "iän"  or final == "ia" or final == "ie"  or final == "iong" ) then
		error("Syllables in Nankinese Pinyin do not include ji+vowel, qi+vowel, xi+vowel. You can try j-, q-, x- instead ji-, qi-, xi- ")
	end
	if (initial == "j" or initial == "q" or initial == "x") and ( final == "iou"  or final == "iu") then
		error("Nankinese Pinyin does not use jiou, qiou ,xiou ,jiu, qiu ,xiu, but uses jou, qou, xou instead.")
	end
	if (initial == "zh" or initial == "ch" or initial == "sh" or initial == "r") and (final == "i") then
		error("Nankinese Pinyin does not use zhi, chi, shi ,ri, but uses zhr, chr, shr, r instead.")
	end
	if  (final == "uo") then
		error("Nankinese Pinyin does not use -uo, but uses -o instead.")
	end
end

function export.convert(text, scheme)
	if type(text) == "table" then
		text, scheme = text.args[1], text.args[2]
	end

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

		local extra2 = match(word, '^[^A-ZÄÜa-zäü]*')
		for syllable in gmatch(word, '[A-ZÄÜa-zäü]+[%d%-]+[^A-ZÄÜa-zäü]*') do
			local initial, final, erhua, tone, extra = match(syllable, '^([BDGPTKZJCQLMNFSXHVRWUIÜbdgptkzjlcqmnfsxhvrw]?h?)([AEOÄÜRIUYaiueoäüyrng]+)(r?)([%d%-]+)([^A-ZÄÜa-zäü]*)$')
			local caps = false

			if find(initial .. final, '[A-Z]') 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[initial]
				final = finalConv[final]
				tone = gsub(tone, '.', function(char) return toneConv[char] end)

				if erhua == 'r' then
					if find(final, '^ü') then -- 撮口呼
						final = 'uɚ'
					elseif find(final, '^i') then -- 齊齒呼
						final = 'ɚ'
					elseif find(final, '^u') then -- 合口呼
						final = 'uɚ'
					elseif final == 'o' or final == 'ong' then
						final = 'uɚ'
					else -- 開口呼
						final = 'ɚ'
					end
				end

				syllable = initial .. final .. tone

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

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

				-- idk what happens with erhua, so disable output
				if erhua == 'r' then return false 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
			table.insert(result, table.concat(converted, ' '))
		elseif scheme == 'NKG' then
			local text = table.concat(converted, '')
			text = gsub(text, '([a-z])@(u)', '%1w')
			text = gsub(text, '([a-z])@(i)', '%1j')
			text = gsub(text, '([ng])@(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