Module:pcc-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 len = mw.ustring.len

-- https://en.wikipedia.orghttps://dictious.com/en/Bouyei_language

local initialConv = {
	   = 'p',
		= 'pʰ',
	  = 'ɓ',
	   = 'm',
	   = 'f',
	   = 'v',
	   = 'w', -- to support /v/ ~  in some cases

	   = 't',
		= 'tʰ',
	  = 'ɗ',
	   = 'n',
		= 'ɬ',
	   = 'l',

	   = 'k',
	   = 'kʰ',
	  = 'ŋ',
	   = 'x',
		= 'ɣ',

		= 't͡ɕ',
		= 't͡ɕʰ',
	  = 'ɲ',
	   = 'ɕ',
	   = 'j',
	
		= 't͡s',
		= 't͡sʰ',
	   = 's',
	   = 'z',
	
	  = 'pʲ',
	  = 'mʲ',
		= 'ˀj',
	
	  = 'kʷ',
	 = 'ŋʷ',
	  = 'ˀv',

	    = 'ʔ',
}

local rimeConv = {
			= 'a',
			= 'o',
			= 'ɔ',
			= 'e',
			= 'ɛ', -- e in Chinese loanwords
			= 'i',
			= 'z̩',
			= 'u',
			= 'ɯ',
	
			= 'aːi',
			= 'ɐi',
			= 'oːi',
			= 'ɯi',
	
			= 'aːu',
			= 'ɐu',
			= 'eːu',
			= 'iu',
	
			= 'ɐɯ',
			= 'iə',
			= 'uə',
			= 'ɯə',
	
			= 'aːm',
			= 'ɐm',
			= 'oːm',
			= 'ɔm',
			= 'eːm',
			= 'iəm',
			= 'im',
			= 'uəm',
			= 'um',
			= 'ɯəm',
	
			= 'aːn',
			= 'ɐn',
			= 'oːn',
			= 'ɔn',
			= 'eːn',
			= 'iən',
			= 'in',
			= 'uən',
			= 'un',
			= 'ɯən',
			= 'ɯn',
	
		= 'aːŋ',
			= 'ɐŋ',
		= 'oːŋ',
			= 'ɔŋ',
		= 'eːŋ',
		= 'iəŋ',
			= 'iŋ',
		= 'uəŋ',
			= 'uŋ',
		= 'ɯəŋ',
			= 'ɯŋ',
	
			= 'aːp̚',
			= 'ɐp̚',
			= 'oːp̚',
			= 'ɔp̚',
			= 'eːp̚',
			= 'iəp̚',
			= 'ip̚',
			= 'uəp̚',
			= 'up̚',
			= 'ɯəp̚',
			= 'ɯp̚',
	
			= 'aːt̚',
			= 'ɐt̚',
			= 'oːt̚',
			= 'ɔt̚',
			= 'eːt̚',
			= 'iət̚',
			= 'it̚',
			= 'uət̚',
			= 'ut̚',
			= 'ɯət̚',
			= 'ɯt̚',
	
			= 'ɐk̚',
			= 'ɔk̚',
			= 'ek̚',
			= 'ik̚',
			= 'uk̚',
			= 'ɯk̚',
	
			= 'ia',
			= 'io',
			= 'iɐu',
			= 'ua',
			= 'ui',
			= 'uɐi',
			= 'aːu',
			= 'əu',
			= 'ɚ',
}

local toneConv = {
		= '˨˦',
		= '˩',
		= '˥˧',
		= '˧˩',
		= '˧˥',
		= '˧',
		= '˧˥',
		= '˧',
	
		= '˧',
		= '˧˩',
		= '˥˧',
		= '˨˦',
}

local function get_tone(syllable)
	local toneless, tone = syllable, ""
	if find(syllable, "$") then
		toneless, tone = match(syllable, "(+)()$")
	end
	mw.log(toneless, tone)
	return toneless, tone
end

local function syllabify(text)
	text = gsub(text, "'", " ")
	text = gsub(text, "()()", "%1 %2")
	--text = gsub(text, "()()", "%1 %2")
	return mw.text.gsplit(text, "")
end

function export.ipa(text)
	text = string.lower(text)
	local syllables = {}
	for syllable in syllabify(text) do
		local initial, rime, tone
		syllable, tone = get_tone(syllable)
		initial, rime = match(syllable, "^(??v?)(??g?)$")
		if not initial or not rime then
			error(syllable .. " cannot be recognized")
		end
		if find(tone, "^$") then
			if initial == "e" then
				initial = "ê"
			elseif initial == "o" then
				initial = "ô"
			elseif initial == "i" and find(final, "^$") then
				initial = "î"
			end
		end
		local initial_ipa, rime_ipa, tone_value = initialConv, rimeConv, toneConv
		if not initial_ipa then
			error(initial .. " is not a valid initial")
		elseif not rime_ipa then
			error(rime .. " is not a valid rime")
		end
		table.insert(syllables, initial_ipa .. rime_ipa .. tone_value)
	end
	return "/" .. table.concat(syllables, ".") .. "/"
end

function export.show(frame)
	local params = {
		 = { },
	}
	local args = require("Module:parameters").process(frame:getParent().args, params)
	local text = args
	if not text then text = mw.title.getCurrentTitle().text end
	
	local display = string.format("* %s",
		require("Module:IPA").format_IPA_full {
			lang = require("Module:languages").getByCode("pcc"),
			items = {{ pron = export.ipa(text) }},
		}
	)
	
	return display
end

return export