Module:eo-pron

local export = {}

local concat = table.concat
local insert = table.insert
local max = math.max
local remove = table.remove
local ulen = mw.ustring.len
local ulower = mw.ustring.lower
local usub = mw.ustring.sub

local lang = require("Module:languages").getByCode("eo")
local dz = mw.ustring.char(0xF000)

local consonants = {
	 = "b",
	 = "t͡s",
	 = "t͡ʃ",
	 = "d",
	 = "d͡z",
	 = "f",
	 = "ɡ",
	 = "d͡ʒ",
	 = "h",
	 = "x",
	 = "j",
	 = "ʒ",
	 = "k",
	 = "l",
	 = "m",
	 = "n",
	 = "p",
	 = "r",
	 = "s",
	 = "ʃ",
	 = "t",
	 = "v",
	 = "z",
	 = "w"}

local vowels = {
	 = "a",
	 = "e",
	 = "i",
	 = "o",
	 = "u",

}

local phonemes = {}

-- combine into single table
for k, v in pairs(vowels) do phonemes = v end
for k, v in pairs(consonants) do phonemes = v end

local function has_vowel(term)
	return ulower(term):find("") ~= nil
end

local function letters_to_syllables(letters)
	if not letters then
		return { = letters}
	end
	local l_r_exceptions = { = true,  = true,  = true,  = true}
	
	local i = 1
	while true do
		local letter = letters
		if not letter then
			break
		elseif ulower(letter) == "d" then
			local letter1 = letters
			if not letter1 then
				break
			end
			if ulower(letter1) == "z" then
				letter = letter .. letter1
				letters = letter
				remove(letters, i + 1)
			end
		end
		i = i + 1
	end

	local result = { = ""}
	local j = 1
	for i = 1, #letters - 2 do
		result = result .. letters
		local letter = ulower(letters)
		local letter1 = ulower(letters)
		local letter2 = ulower(letters)
		
		if phonemes then
			if consonants and vowels then
				if vowels or ulen(result) ~= 1 then
					-- single consonant goes with following vowel
					if has_vowel(result) and (letter1 ~= 'ŭ') then
						j = j + 1
						result = ""
					end
				end
			elseif consonants and not l_r_exceptions and (letter2 == 'l' or letter2 == 'r') and (letter1 ~= 'l' and letter1 ~= 'r') then
				-- consonant followed by l or r goes with l or r
				if has_vowel(result) then
					j = j + 1
					result = ""
				end
			elseif vowels then
				-- two vowels
				if has_vowel(result) then
					j = j + 1
					result = ""
				end
			end
		end
	end
	
	-- add last two letters
	if letters then
		local c1 = letters
		local c2 = letters
			
		if c1 ~= 'ŭ' then
            if vowels and vowels then
                result = result .. c1
                j = j + 1
                result = c2
            elseif has_vowel(result) and has_vowel(c1 .. c2) then
            	j = j + 1
            	result = c1 .. c2
        	else
        		result = result .. c1 .. c2
    		end
        	
        else
            if vowels] and vowels then
            	result = result .. c1
                j = j + 1
                result = c2
            elseif has_vowel(result) and has_vowel(c1 .. c2) then
            	j = j + 1
            	result = c1 .. c2
        	else
        		result = result .. c1 .. c2
    		end
        end
	end
	
	local result2 = {}
	for i, j in ipairs(result) do
		if j and j ~= "" then
			insert(result2, j)
		end
	end
	return result2
end

local function string_to_letters(term)
	return mw.text.split(term, "")
end

local function string_to_syllables(term)
	-- split if given artificial syllable breaks
	local split_input = mw.text.split(term, '‧', true)
	local result = {}
	for _, split in ipairs(split_input) do
		for j, syllable in ipairs(letters_to_syllables(string_to_letters(split))) do
			insert(result, syllable)
		end
	end
	
	return result
end

local function letter_to_ipa(letter)
	return letter == dz and phonemes.dz or phonemes or ""
end

local function string_to_ipa(syllable)
	return (syllable:gsub("", dz)
		:gsub(".*", letter_to_ipa)
		:gsub("()j", "%1i̯")
		:gsub("()w", "%1u̯")
	)
end

local function strip_initial_consonants(term)
	for i, letter in term:gmatch("()(.*)") do
		if vowels then
			return term:sub(i)
		end
	end
	return term
end

local function term_to_words(term)
	-- split by spaces, hyphens, or periods
	return mw.text.split(term, '')
end

local function term_to_IPA_and_rhyme(term)
	local words = term_to_words(term)
	local result, rhyme = {}

	for i, word in ipairs(words) do
		if word ~= "" then
			-- Add /o/ if the word is a single character and a consonant.
			if ulen(word) == 1 then
				if consonants then
					word = word .. "o"
				end
			end

			-- Break into syllables, and make each into IPA.
			local hyphenated, word_result = string_to_syllables(word), {}
			for j, syllable in ipairs(hyphenated) do
				word_result = string_to_ipa(syllable)
			end

			-- Add stress to the penultimate syllable.
			if word_result then
				insert(word_result, #word_result - 1, "ˈ")
			end
			-- Set rhyme to last two syllables, or the last syllable if this is
			-- a one-syllable word.
			rhyme = concat(hyphenated, nil, max(#hyphenated - 1, 1))
			rhyme = string_to_syllables(strip_initial_consonants(rhyme))

			result = concat(word_result)
		end
	end

	-- Rhyme to IPA.
	if rhyme then
		for i, syllable in ipairs(rhyme) do
			rhyme = string_to_ipa(syllable)
		end
		rhyme = concat(rhyme)
	end
	
	return { = result,  = rhyme}
end

local function parse_input(input)
	-- no input -> use page title
	return input or mw.title.getCurrentTitle().text
end

function export.IPA(IPA_input)
	IPA_input = ulower(parse_input(IPA_input))
	return  "/" .. concat(term_to_IPA_and_rhyme(IPA_input).IPA, " ") .. "/"
end

function export.rhyme(rhyme_input)
	rhyme_input = parse_input(rhyme_input)
	return term_to_IPA_and_rhyme(rhyme_input).rhyme
end

function export.hyphenation(hyphenation_input)
	hyphenation_input = parse_input(hyphenation_input)
	
	local words = term_to_words(hyphenation_input)
	local result = {}
	local hyphenated
	
	for i, word in ipairs(words) do
		hyphenated = string_to_syllables(word)
		insert(result, concat(hyphenated, "‧"))
	end
	
	return concat(result, ' ')
end

function export.letters(letters_input)
	letters_input = parse_input(letters_input)
	return concat(string_to_letters(letters_input), '-')
end

local function get_artificial_breaks(frame)
	-- override for syllable breaks
	local args = frame:getParent().args
	if not args then
		return nil
	end
	
	local param = 1
	local result = {}
	while true do
		if not args then
			if not result then
				return nil
			end
			
			return concat(result, "‧")

		end
		insert(result, args)
		param = param + 1
	end
	
end

function export.pronunciation_section(frame)
	local args = frame:getParent().args
	local artificial_breaks = get_artificial_breaks(frame)
	
	local IPA_override = args
	if not IPA_override then
		IPA_override = export.IPA(artificial_breaks)
	end

	local hyphenation_override = args
	if not hyphenation_override then
		hyphenation_override = export.hyphenation(artificial_breaks)
	end
	
	local rhyme_override = args
	if not rhyme_override then
		rhyme_override = export.rhyme(artificial_breaks)
	end
		
	local audio = args

	if rhyme_override then
		rhyme_override = require("Module:rhymes").format_rhymes { lang = lang, rhymes = {{ rhyme = rhyme_override }} }
	end
	IPA_override = require("Module:IPA").format_IPA_full { lang = lang, items = {{pron = IPA_override }} }

	local result = "<ul>"
	result = result .. "<li>" .. IPA_override .. "</li>"
	
	if audio then
		audio = require("Module:audio").format_audio { lang = lang, file = audio }
		result = result .. "<li>" .. audio .. "</li>"
	end
	
	if rhyme_override then
		result = result .. "<li>" .. rhyme_override .. "</li>"
	end
	
	result = result .. "<li>Hyphenation: " .. hyphenation_override .. "</li>"
	result = result .. "</ul>"
	
	return result
end

return export