Module:wuu-pron

Documentation for this module may be created at Module:wuu-pron/doc

local export = {}

local initial = {
	["p"] = 'p', ["ph"] = 'pʰ', ["b"] = 'b̻', ["m"] = 'm', ["'m"] = 'ʔm', ["f"] = 'f', ["v"] = 'v̻',
	["t"] = 't', ["th"] = 'tʰ', ["d"] = 'd̻', ["n"] = 'n', ["'n"] = 'ʔn', ["l"] = 'l', ["'l"] = 'ʔl',
	["ts"] = 't͡s', ["tsh"] = 't͡sʰ', ["s"] = 's', ["z"] = 'z̻', ["j"] = 't͡ɕ', ["q"] = 't͡ɕʰ',
	["jj"] = 'd̥͡ʑ', ["ny"] = 'ȵ', ["'ny"] = 'ʔȵ', ["x"] = 'ɕ', ["xx"] = 'ʑ̻', ["k"] = 'k',
	["kh"] = 'kʰ', ["g"] = 'g̊', ["ng"] = 'ŋ', ["'ng"] = 'ʔŋ', ["h"] = 'h', ["'"] = 'ʔ', ["hh"] = 'ɦ', [""] = ''
}

local final = {
	["a"] = 'ᴀ', ["o"] = 'o', ["au"] = 'ɔ', ["eu"] = 'ɜ', ["e"] = 'ᴇ', ["oe"] = 'ø', ["i"] = 'i',
	["ia"] = 'iᴀ', ["ie"] = 'iᴇ', ["io"] = 'io', ["iau"] = 'iɔ', ["ieu"] = 'iɜ', ["u"] = 'ɯ', ["ua"] = 'ɯᴀ',
	["ue"] = 'ɯᴇ', ["uoe"] = 'ɯø', ["y"] = 'y', ["yoe"] = 'yø', ["an"] = 'ã', ["aan"] = 'ɑ̃',
	["en"] = 'əɲ', ["on"] = 'ʊŋ', ["aq"] = 'ᴀʔ', ["oq"] = 'ʊʔ', ["eq"] = 'əʔ', ["ian"] = 'iã',
	["iaan"] = 'iɑ̃', ["in"] = 'ɪɲ', ["ion"] = 'i̯ʊŋ', ["iaq"] = 'iᴀʔ', ["ioq"] = 'i̯ʊʔ', ["iq"] = 'i̯ɪʔ',
	["uan"] = 'ɯã', ["uaan"] = 'ɯɑ̃', ["un"] = 'ɯə̯ɲ', ["uaq"] = 'ɯᴀʔ', ["ueq"] = 'ɯə̯ʔ', ["yn"] = 'ʏɲ',
	["yq"] = 'ɥ̯ɪʔ', ["er"] = 'əɻ', ["r"] = 'z̩'
}

local tone_contours = {
	['1-0'] = '',
	['1-1'] = '⁵³', ['1-2'] = '³⁴', ['1-3'] = '²³', ['1-4'] = '⁵⁵', ['1-5'] = '¹²',
	['2-1'] = '⁵⁵ ²¹', ['2-2'] = '³³ ⁴⁴', ['2-3'] = '²² ⁴⁴', ['2-4'] = '³³ ⁴⁴', ['2-5'] = '¹¹ ²³',
	['3-1'] = '⁵⁵ ³³ ²¹', ['3-2'] = '³³ ⁵⁵ ²¹', ['3-3'] = '²² ⁵⁵ ²¹', ['3-4'] = '³³ ⁵⁵ ²¹', ['3-5'] = '¹¹ ²² ²³',
	['4-1'] = '⁵⁵ ³³ ³³ ²¹', ['4-2'] = '³³ ⁵⁵ ³³ ²¹', ['4-3'] = '²² ⁵⁵ ³³ ²¹', ['4-4'] = '³³ ⁵⁵ ³³ ²¹', ['4-5'] = '²² ⁵⁵ ³³ ²¹'
}

local syllabic = {
	['mm'] = 'm̩', ['ngg'] = 'ŋ̍'
}

function export.ipa_conv(original_text)
	if type(original_text) == 'table' then original_text = original_text.args[1] end
	original_text = mw.ustring.lower(original_text)
	local text, conv_text = '',''
	local reading = mw.text.split(original_text, ",", true)
	local syllable = {}
	local syl_tone = {}
	for i = 1, #reading, 1 do
		parts = mw.text.split(reading[i], '+', true)
		for i = 1, #parts do
			text = parts[i]
			wordlen = string.len(text:gsub('[^ ]', '')) + 1
			if mw.ustring.match(mw.ustring.sub(text,1,1), '[0-9]') then
				tone = wordlen .. '-' .. text:sub(1,1)
				tone = tone_contours[tone] or tone
				text = text:sub(2,-1)
			else
				tone = tostring(text:gsub('%l', ''))
				tone = mw.ustring.gsub(tone, '[1-5]', {[1]='¹', [2]='²', [3]='³', [4]='⁴', [5]='⁵'})
				text = text:gsub('%d', '')
			end
			syllable = mw.text.split(text, ' ', true)
			syl_tone = mw.text.split(tone, ' ', true)
			for i = 1, wordlen, 1 do
				syllable[i] = export.ipa_syl_conv(syllable[i])
				syllable[i] = syllable[i] .. syl_tone[i]
			end
			parts[i] = table.concat(syllable, " ")
		end
		reading[i] = table.concat(parts, "  ")
	end
	return table.concat(reading, "/, /")
end

function export.ipa_syl_conv(text)
	if text:match('^h?h?[mn][mg]?g?$') then
		text = mw.ustring.gsub(text, "^(h?h?)([mn][mg]?g?)$", function(a, b) return (initial[a] or '') .. (syllabic[b] or error(("Invalid syllable: \"%s\""):format(text))) end)
	elseif text:match("^%'?ny") then
		text = mw.ustring.gsub(text, "^([\']?ny)([aeiouyr][aeou]?[aeu]?[nqr]?)$",
			function(a, b) return initial[a] .. (final[b] or error(("Unrecognised final: \"%s\""):format(b))) end)
	elseif text:match("^[\']?[pbmfvtdnlszjqxkghr%']?[sjgx]?h?[aeiouyr][aeou]?[aeu]?[nqr]?$") then
		text = mw.ustring.gsub(text, "^([\']?[pbmfvtdnlszjqxkghr%']?[sjgx]?[h]?)([aeiouyr][aeou]?[aeu]?[nqr]?)$",
			function(a, b) return
				(initial[a] or error(("Unrecognised initial: \"%s\""):format(a))) ..
				(final[b] or error(("Unrecognised final: \"%s\""):format(b))) end)
	else
		return error(("Invalid syllable: \"%s\""):format(text))
	end
	return text
end

function export.rom(text)
	if type(text) == 'table' then text = text.args[1] end
	text = mw.text.split(text, ",", true)
	for i = 1, #text, 1 do
		parts = mw.text.split(text[i], '+', true)
		for i = 1, #parts do
			if string.match(parts[i]:sub(1,1), '[0-9]') then
				parts[i] = parts[i]:sub(2,-1) .. ' (T' .. parts[i]:sub(1,1) .. ')'
			end
		end
		text[i] = table.concat(parts, ' + ')
	end
	return table.concat(text, "; ")
end

return export