Module:lt-pron

local export = {}

local U = require("Module:string/char")
local udecomp = mw.ustring.toNFD
local rsubn = mw.ustring.gsub
local sub = mw.ustring.sub
local rmatch = mw.ustring.match
local rfind = mw.ustring.find
local ugmatch = mw.ustring.gmatch
local ulen = mw.ustring.len
local ulower = mw.ustring.lower

-- version of rsubn() that discards all but the first return value
local function rsub(term, foo, bar, n)
	local retval = rsubn(term, foo, bar, n)
	return retval
end

-- unicode codepoints
local ACUTE = U(0x0301) -- acute =  ́
local GRAVE = U(0x0300) -- grave =  ̀
local CIRC = U(0x0302) -- circumflex =  ̂
local TILDE = U(0x0303) -- tilde =  ̃
local MACRON = U(0x0304) -- macron =  ̄
local BREVE = U(0x0306) -- breve =  ̆
local DOT = U(0x0307) -- dot above = ̇
local CARON = U(0x030C) -- caron =  ̌
local OGONEK = U(0x0328) -- ogonek =  ̨

-- character classes
local accents = ACUTE .. GRAVE .. CIRC .. TILDE .. MACRON .. BREVE
local diacritics = accents .. DOT .. CARON .. OGONEK
local vowels = ""
local consonants = "" -- does not include j

-- various substitutions
local subs = {
	glyphs = {
		 = "ç",
		 = "ć",
		 = "ð",
		 = "þ",
		 = "ɔ"
	},
	ipa_c = {
		 = "t͡ʃ",
		 = "t͡s",
		 = "t͡s",
		 = "x",
		 = "d͡ʒ",
		 = "d͡z",
		 = "ɡ",
		 = "ɣ",
		 = "kʋ",
		 = "k",
		 = "ʃ",
		 = "ʋ",
		 = "ʒ"
	},
	ipa_v = {
		 = "ɐ",
		 = "ɛ",
		 = "ɪ",
		 = "ʊ"
	}
}

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

function export.link(term)
	return require("Module:links").full_link{ term = term, lang = lang }
end

--[=[
Takes the orthographic representation to make it closer
to the phonological output by respelling and adding missing segments
]=]
function export.respell(term)
	-- decompose accents from term
	term = udecomp(term)

	-- replace digraph consonants with temporary placeholders
	term = rsub(term, ".?", subs.glyphs)

	-- add missing /j/: ievà > jievà
	if rfind(term, "^i?e" .. TILDE .. "?") then
		term = "j" .. term
	end
	-- pãieškos > pãjieškos
	term = rsub(
		term,
		"(" .. vowels .. "*)" .. "(i?e" .. TILDE .. "?)",
		"%1j%2"
	)

	-- show palatalisation
	term = rsub(term, "i(*)(.?)",
		function(vow, next_char)
			if next_char == "u" then
				return "i" .. vow .. next_char
			else
				return "ʲ" .. vow .. next_char
			end
		end
	)
	term = rsub(term, "(" .. consonants .. CARON .. "?)()", "%1ʲ%2")
	term = rsub(term, "(" .. consonants .. "+)(" .. consonants .. "ʲ)",
		function(cons, soft)
			local out = ""
			for c in ugmatch(cons, ".") do
				-- k does not become palatalised: krienas, apniūkti do not have kʲ
				if c == "k" then
					out = out .. c
				else
					out = out .. c .. "ʲ"
				end
			end
			return out .. soft
		end
	)

	return term
end

--[=[
Splits the term into its syllable boundaries
]=]
function export.syllabify(term, table)
	term = rsub(term,
		"([aeioɔuy" .. diacritics ..
		"]*-)(?" .. CARON .. "?ʲ??" .. CARON .. "?ʲ??ʲ?)",
		"%1.%2"
	)
	term = rsub(term, "^%.", "")
	term = rsub(term, "%.ʲ", "ʲ.")
	term = rsub(term, "%.(ʲ?)()", "%1.%2")

	return table and ugmatch(term, ".") or term
end

--[=[
Render pronunciation
]=]
local function pron(term)
	-- replace consonants with their ipa equivalents
	term = rsub(term, ".?", subs.ipa_c)

	-- consonant allophones
	term = rsub(term, "ʃʲ", "ɕ")
	term = rsub(term, "ʒʲ", "ʑ")
	term = rsub(term, "l()", "ɫ%1")
	term = rsub(term, "n(ʲ?%.?)", "ŋ%1")

	-- vowels
	term = rsub(term,
		"^(-)",
		"ˈ%1"
	)
	term = rsub(term,
		"%.(-)",
		"ˈ%1"
	)
	-- diphtongs
	term = rsub(term, "ia" .. ACUTE .. "u", "æ" .. CIRC .. "ʊ")
	term = rsub(term, "iau" .. TILDE .. "?", "ɛʊ")
	term = rsub(term, "au" .. TILDE .. "?", "ɐʊ̯")
	term = rsub(term, "a" .. ACUTE .. "i", "ɐ" .. CIRC .. "ɪ")
	term = rsub(term, "ai" .. TILDE .. "?", "ɐɪ")
	term = rsub(term, "e" .. ACUTE .. "i", "ɛ" .. CIRC .. "ɪ")
	term = rsub(term, "ei" .. TILDE .. "?", "ɛɪ")
	term = rsub(term, "a" .. ACUTE .. "u", "a" .. CIRC .. "ʊ")
	term = rsub(term, "au" .. TILDE .. "?", "ɒʊ")
	term = rsub(term, "e" .. ACUTE .. "u", "ɛ" .. CIRC .. "ʊ")
	term = rsub(term, "eu" .. TILDE .. "?", "ɛʊ")
	term = rsub(term, "i" .. ACUTE .. "e", "i" .. CIRC .. "ə")
	term = rsub(term, "ie" .. TILDE .. "?", "iə")
	term = rsub(term, "u" .. ACUTE .. "o", "u" .. CIRC .. "ə")
	term = rsub(term, "uo" .. TILDE .. "?", "uə")
	term = rsub(term, "u" .. ACUTE .. "i", "ʊ" .. CIRC .. "ɪ")
	term = rsub(term, "ui" .. TILDE .. "?", "ʊɪ")
	term = rsub(term, "o" .. ACUTE .. "u", "ɔ" .. CIRC .. "ɪ")
	term = rsub(term, "u" .. ACUTE .. "u", "ɔ" .. CIRC .. "ʊ")

	term = rsub(term, "a", "ɐ")
	term = rsub(term, "ɐ" .. ACUTE, "âː")
	term = rsub(term, "ɐ" .. TILDE, "aː")

	term = rsub(term, "e", "ɛ")
	term = rsub(term, "ɛ" .. DOT, "eː")
	term = rsub(term, "ɛ" .. OGONEK, "æː")

	term = rsub(term, "i", "ɪ")
	term = rsub(term, "ɪ" .. OGONEK, "iː")

	term = rsub(term, "y", "iː")

	term = rsub(term, "u", "ʊ")
	term = rsub(term, "ʊ" .. MACRON, "uː")
	term = rsub(term, "ʊ" .. OGONEK, "uː")

	term = rsub(term, "o" .. TILDE .. "?", "oː")
	term = rsub(term, "ʲaː", "ʲæː")
	term = rsub(term, "ʲɐ", "ʲɛ")

	term = rsub(term, "ː" .. ACUTE, CIRC .. "ː") -- acutes = stressed + circumflex tone
	term = rsub(term, TILDE, "") -- tilde = stressed + long
	term = rsub(term, GRAVE, "") -- grave = stressed + short

	term = rsub(term, "()a(" .. OGONEK .. "?)", "%1e%2")
	return term
end

--[=[
	Returns the respelled term and its pronunciation
]=]
function export.test_respell(frame)
	local args = require("Module:parameters").process(frame:getParent().args, { = {default = ""}})
	local syll = export.syllabify(export.respell(args))

	return syll .. " → "
end

--[=[
Converts the term to IPA
]=]
function export.toIPA(text)
	return pron(ulower(export.syllabify(export.respell(text))))
end

--[=[
	Displays the IPA of the term
]=]
function export.show(frame)
	local params = {
		 = {default = mw.title.getCurrentTitle().text}
	}
	local args = require("Module:parameters").process(frame:getParent().args, params)

	return args
end

return export