Module:kpv-IPA

local export = {}

local langcode = "kpv"

local vowel = ""
local vowel_soft = ""
local vowel_prej = ""
local vowel_accent = "́"

local vowel_phonemic = {
	 = "a",  = "e",  = "o",  = "i",
	 = "i",  = "o",  = "ɘ",  = "u",
	 = "ɯ",  = "e",  = "u",  = "a"
}

local vowel_phonetic = {
	 = "ä",  = "o̞",  = "o̞",  = "ɯ̈",  = "ä"
}

local consonant = ""

local consonant_voiced = ""
local consonant_unvoiced = ""

local consonant_devoice = {
	 = "п",
	 = "т",
}

-- normal, soft
local consonant_phonemic = {
	 =  { "b",	"b" },
	 =  { "ʋ",	"ʋ" },
	 =  { "ɡ",	"ɡ" },
	 =  { "d",	"ɟ" },
	 = { "d͡ʑ", 	"d͡ʑ" },
	 = { "d͡ʒ",	"d͡ʒ" },
	 =  { "ʒ",	"ʒ" },
	 =  { "z",	"ʑ" },
	 =  { "j",	"j" },
	 =  { "k",	"k" },
	 =  { "l",	"ʎ" },
	 =  { "m",	"m" },
	 =  { "n",	"ɲ" },
	 =  { "p",	"p" },
	 =  { "r",	"r" },
	 =  { "s",	"ɕ" },
	 =  { "t",	"c" },
	 = { "t͡ʃ",	"t͡ʃ" },
	 =  { "f",	"f" },
	 =  { "x",	"x" },
	 =  { "t͡s",	"t͡s" },
	 =  { "t͡ɕ",	"t͡ɕ" },
	 =  { "ʃ",	"ʃ" },
	 =  { "ʃː",	"ʃː" },
}

-- normal, soft, voiced, voiced+soft
local consonant_phonetic = {
	 = { "d͡ʒ̺",	"d͡ʒ̺",	"d͡ʒ̺",	"d͡ʒ̺" },
	 =  { "ʒ̺",	"ʒ̺",	"ʒ̺",	"ʒ̺" },
	 =  { "k",	"k",	"ɡ",	"ɡ" },
	 =  { "ɫ",	"ʎ",	"ɫ",	"ʎ" },
	 =  { "p",	"p",	"b",	"b" },
	 =  { "s",	"ɕ",	"z",	"ʑ" },
	 =  { "t",	"c",	"d",	"ɟ" },
	 = { "t͡ʃ̺",	"t͡ʃ̺",	"d͡ʒ̺",	"d͡ʒ̺" },
	 =  { "t͡s",	"t͡s",	"d͡z",	"d͡z" },
	 =  { "t͡ɕ",	"t͡ɕ",	"d͡ʑ",	"d͡ʑ" },
	 =  { "ʃ̺",	"ʃ̺",	"ʃ̺",	"ʃ̺" },
	 =  { "ʃ̺ː",	"ʃ̺ː",	"ʃ̺ː",	"ʃ̺ː" },
}

local vowl = ""
local vowld = ""
local cons = ""
local consd = ""
local affricate1 = ""
local affricate2b = ""
local affricate2 = "͡" .. affricate2b
local postalveolar = {
	 = "ʃ̺",
	 = "ʒ̺",
	 = "ʒ̺",
	 = "ʒ̺",
}

local function convert_vowel(c, is_phonetic)
	if is_phonetic and vowel_phonetic then
		return vowel_phonetic
	end
	return vowel_phonemic
end

local function convert_consonant(c, is_phonetic, is_soft, is_voiced)
	local num = is_soft and 2 or 1
	if is_phonetic and consonant_phonetic then
		if is_voiced then
			num = num + 2
		end
		return consonant_phonetic
	end
	return consonant_phonemic
end

local digraphs = {
	 = "зж",
	 = "ш"
}

local function split_phonemes(s)
	local phonemes = { }
	local index = 1
	local len = mw.ustring.len(s)
	local stress_at = 1
	if mw.ustring.match(s, "^-") then
		table.insert(phonemes, { nil, "-" })
		stress_at = nil
		index = 2
	end
	while index <= len do
		if mw.ustring.match(s, "^" .. vowel .. vowel_accent .. "?", index) then
			local v = mw.ustring.match(s, "^(" .. vowel .. vowel_accent .. "?)", index)
			table.insert(phonemes, { "v", v })
			if mw.ustring.find(v, vowel_accent) then
				stress_at = nil
			end
			index = index + mw.ustring.len(v)
		elseif mw.ustring.match(s, "^" .. consonant, index) then
			local c = mw.ustring.match(s, "^" .. consonant, index)
			
			-- digraphs
			if digraphs then
				local cc = mw.ustring.match(s, "^" .. c .. " .. "]", index)
				if cc then c = cc end
			end

			table.insert(phonemes, { "c", c })
			index = index + mw.ustring.len(c)
		elseif mw.ustring.match(s, "^ь", index) then
			table.insert(phonemes, { "s", nil })
			index = index + 1
		elseif mw.ustring.match(s, "^ъ", index) then
			table.insert(phonemes, { "h", nil})
			index = index + 1
		elseif mw.ustring.match(s, "^-", index) then
			if index < len then
				table.insert(phonemes, { nil, "." })
			else
				table.insert(phonemes, { nil, "-" })
			end
			index = index + 1
		elseif mw.ustring.match(s, "^%s", index) then
			if stress_at then
				table.insert(phonemes, stress_at, { nil, "ˈ" })
			end
			table.insert(phonemes, { "w", mw.ustring.sub(s, index, index) })
			stress_at = #phonemes + 1
			index = index + 1
		else -- something else...
			table.insert(phonemes, { nil, mw.ustring.sub(s, index, index) })
			index = index + 1
		end
	end
	if stress_at then
		table.insert(phonemes, stress_at, { nil, "ˈ" })
	end
	return phonemes
end

local function is_next_consonant(phonemes, i, consonant)
	i = i + 1
	while i <= #phonemes do
		local type, x = phonemes, phonemes
		if type == "v" then return false end
		if type == "c" then return mw.ustring.match(x, consonant) end
		i = i + 1
	end
	return false
end

local function has_next_consonant(phonemes, i, consonant)
	i = i + 1
	while i <= #phonemes do
		local type, x = phonemes, phonemes
		if type == "v" then return false end
		if type == "c" then
			if mw.ustring.match(x, consonant) then
				return true
			end
		end
		i = i + 1
	end
	return false
end

local function handle_long_digraphs(s)
	s = mw.ustring.gsub(s, "()%1()", "%1ь%1%2")
	for k, v in pairs(digraphs) do
		s = mw.ustring.gsub(s, k .. k .. "()", k .. "%1" .. k .. "%1")
	end
	return s
end

function export.convert(s, is_phonetic)
	local phonemes = split_phonemes(handle_long_digraphs(s))
	local result = ""
	local stress_preinserted = false
	for i, phoneme in ipairs(phonemes) do
		local type, x = phoneme, phoneme
		if type == "v" then
			-- vowel
			local v, s = mw.ustring.match(x, "(" .. vowel .. ")(" .. vowel_accent .. "?)", index)
			if #s > 0 and not stress_preinserted then
				result = result .. "ˈ"
			end
			stress_preinserted = false
			if (i < 2 or phonemes ~= "c") and mw.ustring.match(v, vowel_prej) then
				result = result .. "j"
			end
			result = result .. convert_vowel(v, is_phonetic)
		elseif type == "c" then
			-- consonant
			local c = x
			local soft = false
			local voiced = false
			if i < #phonemes then
				soft = (phonemes == "v" and mw.ustring.match(phonemes, "^" .. vowel_soft)) or phonemes == "s"
				if is_phonetic then
					if has_next_consonant(phonemes, i, "^" .. consonant_voiced) then
						voiced = true
					elseif consonant_devoice and is_next_consonant(phonemes, i, "^" .. consonant_unvoiced) then
						c = consonant_devoice
					end
				end
				if phonemes == "v" and mw.ustring.find(phonemes, vowel_accent) then
					result = result .. "ˈ"
					stress_preinserted = true
				end
			end
			result = result .. convert_consonant(c, is_phonetic, soft, voiced)
		elseif type == "s" then
			-- soft sign, ignore here
		elseif type == "h" then
			-- hard sign, ignore here
		elseif type == "w" then
			-- whitespace
			result = result .. x
		elseif not type then
			-- anything else
			result = result .. x
		end
	end

	if is_phonetic then
		-- assimilation of /s/ and /z/ to following /ʃ/, /ʒ/ or affricate with either
		result = mw.ustring.gsub(result, "(ʲ?)(*)(" .. consd .. "*ʲ?)",
			function (orig, space, palatal)
				return postalveolar .. space .. palatal
			end)
		result = mw.ustring.gsub(result, "(ʲ?)(*)(" .. affricate1 .. "͡)(" .. affricate2b .. consd .. "*ʲ?)",
			function (orig, space, plosive, palatal)
				return postalveolar .. space .. plosive .. palatal
			end)
	end

	-- long consonants
	result = mw.ustring.gsub(result, "(" .. affricate1 .. affricate2 .. consd .. "*ʲ?)%1", "%1ː")
	if is_phonetic then
		result = mw.ustring.gsub(result, "(" .. affricate1 .. ")%1(" .. affricate2 .. consd .. "*ʲ?)", "%1%2ː")
	end
	result = mw.ustring.gsub(result, "(" .. cons .. ")%1()", "%1ː%2")
	result = mw.ustring.gsub(result, "(" .. cons .. ")%1$", "%1ː")
	result = mw.ustring.gsub(result, "(" .. vowl .. vowld .. "*)%1", "%1ː")

	return result
end

function export.main(frame)
	local title = mw.title.getCurrentTitle().text
	if type(frame) == "table" then
		title = frame:getParent().args or title
	end
	
	local phonemic = export.convert(mw.ustring.lower(title), false)
	local phonetic = export.convert(mw.ustring.lower(title), true)
	
	return require("Module:IPA").format_IPA_full {
		lang = require("Module:languages").getByCode(langcode),
		items = {
			{pron = "/" .. phonemic .. "/"},
			{pron = ""}
		},
	}
end

return export