Jump to content

Module:nan-pron-Leizhou

Wiktionary වෙතින්

This module page is experimental.
The details of its operation have not yet been fully decided upon. Do not deploy widely until the module page is finished.

This module does romanisation (雷州话拼音方案 devised by 蔡叶青) to IPA conversion for Leizhou Min. See {{zh-pron}}.


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

-- local sub = m_string_utils.sub
-- local len = m_string_utils.len
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 = {
	["b"] = "p", ["p"] = "pʰ", ["m"] = "m", ["bh"] = "b",
	["d"] = "t", ["t"] = "tʰ", ["n"] = "n", ["l"] = "l",
	["g"] = "k", ["k"] = "kʰ", ["ng"] = "ŋ", ["h"] = "h",
	["j"] = "t͡ɕ", ["q"] = "t͡ɕʰ", ["x"] = "ɕ", ["y"] = "z", ["ʑ"] = "ʑ",
	["z"] = "t͡s", ["c"] = "t͡sʰ", ["s"] = "s",
	[""] = ""
}

local final_ipa = {
	["i"] = "i", ["u"] = "u",
	["a"] = "a", ["ia"] = "ia", ["ua"] = "ua",
	["o"] = "ɔ", ["io"] = "iɔ",
	["e"] = "ɛ", ["ie"] = "iɛ", ["ue"] = "uɛ",
	["ai"] = "ai", ["uai"] = "uai",
	["ao"] = "au", ["iao"] = "iau",
	["eu"] = "ɛu", ["iu"] = "iu",
	["oi"] = "ɔi", ["ui"] = "ui",
	["am"] = "am", ["iam"] = "iam", ["m"] = "m̩",
	["em"] = "em", ["im"] = "im",
	["ang"] = "aŋ", ["iang"] = "iaŋ", ["uang"] = "uaŋ",
	["eng"] = "eŋ", ["ieng"] = "ieŋ", ["ng"] = "ŋ̍",
	["ong"] = "ɔŋ", ["iong"] = "iɔŋ",
	["ing"] = "iŋ", ["ung"] = "uŋ",
	["ab"] = "ap̚", ["iab"] = "iap̚",
	["eb"] = "ep̚", ["ib"] = "ip̚",
	["ag"] = "ak̚", ["iag"] = "iak̚", ["uag"] = "uak̚",
	["eg"] = "ek̚", ["ieg"] = "iek̚", ["ueg"] = "uek̚",
	["ig"] = "ik̚", ["ug"] = "uk̚",
	["og"] = "ɔk̚", ["iog"] = "iɔk̚",
}

local tone_chao = {
	["1"] = "³⁵", ["2"] = "³¹", ["3"] = "²¹", ["4"] = "⁵",
	["5"] = "²²", ["6"] = "³³", ["7"] = "⁵⁵", ["8"] = "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, "[1256]")) or (tone1 == "3" and match(tone2, "[123568]")) then
		return "7"
	end
end

-- convert Leizhounese Pinyin to IPA
function export.ipa(text)
	if type(text) == "table" then
		text = text.args[1]
	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[i], final[i], tone[i], block[i] = match(syllable, "^([bpmdtnlgkhjqxyzcs]?[hg]?)([aeiou][aeioumnbg]?[iomnbg]?g?)([1-8])(#?)$")
		
			-- check z/c/s vs. j/q/x
			if (find(initial[i], "^[zcs]$") and find(final[i], "^i")) or (find(initial[i], "^[jqx]$") and find(final[i], "^[aeou]")) then
				error("Invalid Leizhou pinyin input \"" .. syllable .. "\": initial " .. initial[i] .. " cannot go with final " .. final[i] .. ".")
			end	
	
			-- check yi
			if (find(initial[i], "^y$") and find(final[i], "^i")) then
				initial[i] = "ʑ"
			end
			
			-- check checked tones
			if (find(final[i], "[^n][bdg]$") and find(tone[i], "^[^48]$")) or ((find(final[i], "[^bdg]$") or find(final[i], "ng$")) and find(tone[i], "^[48]$")) then
				error("Invalid Leizhou pinyin input \"" .. syllable .. "\": final " .. final[i] .. " cannot go with tone " .. tone[i] .. ".")
			end
		end
		
		for i=1,#syllables,1 do
			if i+1 <= #syllables and block[i] ~= "#" then
				sandhi[i] = tone_sandhi(tone[i], tone[i+1])
			end
		end
		
		for i=1,#syllables,1 do
			actual_tone = tone_chao[tone[i]] .. (sandhi[i] and "⁻" .. tone_chao[sandhi[i]] or "")
			ipa[i] = initial_ipa[initial[i]] .. final_ipa[final[i]] .. 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, '([1-9-]+)', '<sup>%1</sup>')
	return text
end

return export
"https://si.wiktionary.org/w/index.php?title=Module:nan-pron-Leizhou&oldid=82060" වෙතින් සම්ප්‍රවේශනය කෙරිණි