Module:lo-pron

local export = {}
local PAGENAME = mw.title.getCurrentTitle().text
local find = mw.ustring.find
local sub = mw.ustring.sub
local gsub = mw.ustring.gsub
local rhyme

local lang = require("Module:languages").getByCode("lo")

-- for easier processing
local pre = {
	 = { "ຫງ", "໠" }, -- faux hng
	 = { "ຫ", "໡" }, -- faux hny
	 = { "ຫນ", "ໜ" },
	 = { "ຫມ", "ໝ" },
	 = { "ຫ", "໢" }, -- faux hl
	 = { "ຫວ", "໣" }, -- faux hw
	 = {"()(?)", "%2%1"},
	 = {"(.*)()(.*)", "%1%3%2"},
}

local initials = {
	 = "k",  = "kʰ",  = "kʰ",  = "ŋ",  = "ŋ",
	 = "t͡ɕ",  = "s",  = "s",  = "ɲ",  = "ɲ",
	 = "d",  = "t",  = "tʰ",  = "tʰ",  = "n",  = "n",
	 = "b",  = "p",  = "pʰ",  = "pʰ",
	 = "f",  = "f",  = "m",  = "m",
	 = "j",  = "l",  = "l",  = "l",
	 = "ʋ",  = "ʋ",  = "h",  = "h",  = "ʔ",
	 = "kʷ",  = "kʷʰ",  = "kʷʰ",  = "ŋʷ",
	 = "t͡ɕʷ",  = "sʷ",  = "sʷ",
	 = "tʷʰ",  = "tʷʰ",  = "lʷ",  = "ʔʷ",  = "hʷ",
}

local vowels = {
	 = "#",
	 = "aʔ",  = "a",  = "aː",
	 = "iʔ",  = "i",  = "iː",
	 = "ɯʔ",  = "ɯ",  = "ɯː",
	 = "uʔ",  = "u",  = "uː",
	 = "eʔ",  = "e",  = "eː",
	 = "ɛʔ",  = "ɛ",  = "ɛː",
	 = "oʔ",  = "o",  = "oː",
	 = "ɔʔ",  = "ɔ",  = "ɔː",  = "ɔː",
	 = "ɤʔ",  = "ɤ",  = "ɤː",
	 = "uə̯ʔ",  = "uːə̯",  = "uːə̯",  = "uːə̯",
	 = "iə̯ʔ",  = "iə̯ʔ",  = "iə̯",
	 = "iːə̯",  = "iːə̯",  = "iːə̯",  = "iːə̯",  = "iːə̯",
	 = "ɯə̯ʔ",  = "ɯə̯",  = "ɯːə̯",
	 = "am",  = "aɰ",  = "aj",  = "aw",
}

local finals = {
	 = "",  = "k̚",  = "t̚",  = "p̚",
	 = "ŋ",  = "n",  = "m",
	 = "j",  = "j",  = "w",  = "s",
	--force final
	 = "k̚",  = "t̚",  = "p̚",
	 = "ŋ",  = "n",  = "m",
	 = "j",  = "w",  = "s",
}


-- Vientiane
local vt_unmarked_tones = { -- M,L,H
	live       = { "˩(˧)", "˧˥", "˩(˧)" },
	dead_short = { "˧˥", "˧", "˧˥" },
	dead_long  = { "˧˩", "˥˨", "˧˩" },
}

local vt_marked_tones = { -- M,L,H
	 = { "˧", "˧", "˧" },
	 = { "˥˨", "˥˨", "˧˩" },
	 = { "˧˥", "˧˥", "˧˥" },
	 = { "˩˧", "˩˧", "˩˧" },
}

-- Luang Prabang
local lp_unmarked_tones = { -- M,L,H
	live       = { "˩˨", "˩˨", "˥˧˧" },
	dead_short = { "˩˨", "˧˨", "˩˨" },
	dead_long  = { "˥˥˨", "˧˦", "˥˥˨" },
}

local lp_marked_tones = { -- M,L,H
	 = { "˧˨", "˧˨", "˧˨" },
	 = { "˧˦", "˧˦", "˥˥˨" },
	 = { "˧˦", "˧˦", "˧˦" },
	 = { "˩˨", "˩˨", "˩˨" },
}


function export.show(frame)

	local accents = {"Vientiane", "Luang Prabang"}
	local output = {}

	for _,accent in ipairs(accents) do
		table.insert(output, toIPA(frame, accent))
	end

	-- reduced hyphenation (FIXME: Use the standard module)
	table.insert(output, "* Hyphenation: <span class='Laoo lo-reading' lang='lo'>" .. (frame:getParent().args or PAGENAME) .. "</span>")

	-- global var trick (FIXME: YUCK, clean this up)
	local rhymes
	if rhyme == "aɰ" then
		rhymes = {{rhyme = "aj"}, {rhyme = rhyme}}
	elseif rhyme and not find(rhyme, "^#") then
		rhymes = {{rhyme = rhyme}}
	end
	table.insert(output, "* " .. require("Module:rhymes").format_rhymes {
		lang = lang,
		rhymes = rhymes,
	})

	return table.concat(output, "\n")
end

function toIPA(frame, accent)

	local args = frame:getParent().args
	local text = args or PAGENAME -- supports only one pronunciation
	local debug = args or nil

	local unmarked_tones, marked_tones
	if accent == "Vientiane" then
		unmarked_tones, marked_tones = vt_unmarked_tones, vt_marked_tones
		text = gsub(text, "ໃ", "ໄ")
	elseif accent == "Luang Prabang" then
		unmarked_tones, marked_tones = lp_unmarked_tones, lp_marked_tones
	else
		error("The accent " .. accent .. " is not accepted.")
	end

	local ipa = {}

	for syl in mw.text.gsplit(text, "") do
		for _, v in ipairs(pre) do
			syl = mw.ustring.gsub(syl, v, v)
		end

		local i, v, f, t = mw.ustring.match(syl, "^(?)('?*)(%+??)(?)$")

		if find(i, ".$") then
			error("The -ຣ-, -ລ- medials are either unpronounced or disyllabic. Please respell.")
		end

		if find(i, "^ວ?$") then -- middle class
			tn = 1
		elseif find(i, "^ວ?$") then -- low class
			tn = 2
		else --high class
			tn = 3
		end

		if find(i, ".ວ$") then
			if v == "" then
				i = sub(i, 1, -2)
				v = "ວ"
			elseif v == "ຍ" then
				i = sub(i, 1, -2)
				v = "ວ"
				f = "ຍ"
			elseif v == "າ" then -- Cwaa > Cuua
				i = sub(i, 1, -2)
				v = "ວ"
			elseif v == "າຍ" then -- Cwaa > Cuua
				i = sub(i, 1, -2)
				v = "ວ"
				f = "ຍ"
			end
		end

		if find(v, "ວ$") and v ~= "ົວ" and v ~= "ວ" and v ~= "'ວ" then
			v = sub(v, 1, -2)
			f = "ວ"
		end

		if find(v, "ຍ$") and v ~= "ເຍ" and v ~= "ເັຍ" and v ~= "'ຍ" then
			v = sub(v, 1, -2)
			f = "ຍ"
		end

		if find(v, "ຽ$") and v ~= "ັຽ" and v ~= "ເັຽ" and v ~= "ຽ" and v ~= "'ຽ" then
			v = sub(v, 1, -2)
			f = "ຍ"
		end

		if f == "" and (v == "ິ" or v == "ຶ" or v == "ຸ" or v == "ເິ" or v == "ເຶອ") then
			v = v .. "*" -- added ʔ
		end

		if t ~= "" then
			tm = marked_tones
		else
			dead = (find(f, "%+?") or find(vowels, "ʔ$"))
			long = find(vowels, "ː")
			tm = unmarked_tones
		end

		if debug then
			table.insert(ipa, "")
		else
			table.insert(ipa, initials .. vowels .. finals .. tm)
			rhyme = vowels .. finals -- It becomes rhyme after the last iteration
		end
	end

	local ipa_final = table.concat(ipa, ".")
	ipa_final = gsub(ipa_final, "a(ʔ+)%.", "a(%1).")
	ipa_final = ""

	return "* " .. require("Module:IPA").format_IPA_full {
		lang = lang,
		items = {{pron = ipa_final}},
		a = {accent},
	} .. (#ipa > 0 and require("Module:utilities").format_categories("Lao " .. #ipa .. "-syllable words", lang) or "")
end

return export