Module:pl-pronunciation

local export = {}

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

local m_IPA = require("Module:IPA")
local m_pl_IPA = require("Module:pl-IPA")

local vowels = "aeɘiouyąęó"
local vowel = ""
local consonants = "bcćdfghjklłmnńpqrsśtuvwxyzźż"
local consonant = ""
-- vowel digraphs, not necessarily actual phonetic diphthongs
local diphthong_i_v2 = ""
local diphthongs = {
	 = "u",
	 = "u",
	 = diphthong_i_v2
}
-- consonant digraphs (key = first letter, value = possible second letters)
local digraphs = {
	 = "",
	 = "",
	 = "u",
	 = "z",
	 = "z",
}

local past_tense_suffixes = {
	"liśmy", "liście", "łyśmy", "łyście",
}

local latin_borrowing_suffixes = {
	"ika", "yka",
	"iki", "yki",
	"ika", "yka",
	"ice", "yce",
	"ikom", "ykom",
	"ikę", "ykę",
	"iką", "yką",
	"ice", "yce",
	"ikach", "ykach",
	"iko", "yko",
}

-- if this is changed, the next two functions also need to be
local function is_respelling_close_enough(respelling, word)
	word = mw.ustring.gsub(word, "j(" .. diphthong_i_v2 .. ")", "i%1")
	respelling = mw.ustring.gsub(respelling, "", "")
	respelling = mw.ustring.gsub(respelling, "j(" .. diphthong_i_v2 .. ")", "i%1")
	return word == respelling
end

local function partition(word, oword)
	local parts = {}
	local lenword = mw.ustring.len(word)
	local pos = 1
	local offset = 0
	word = mw.ustring.gsub(word, "", ".")
	while pos <= lenword do
		if mw.ustring.find(mw.ustring.lower(word), "^" .. vowel, pos) then
			local initial = mw.ustring.sub(mw.ustring.lower(word), pos, pos)
			local seq = 1
			if diphthongs and mw.ustring.find(mw.ustring.lower(word), "^" .. initial .. diphthongs, pos) then
				seq = 2
			end
			table.insert(parts, { "v", mw.ustring.sub(oword, pos - offset, pos - offset + seq - 1) })
			pos = pos + seq
		elseif mw.ustring.find(mw.ustring.lower(word), "^" .. consonant, pos) then
			local initial = mw.ustring.sub(mw.ustring.lower(word), pos, pos)
			local seq = 1
			if digraphs and mw.ustring.find(mw.ustring.lower(word), "^" .. initial .. digraphs, pos) then
				seq = 2
			end
			table.insert(parts, { "c", mw.ustring.sub(oword, pos - offset, pos - offset + seq - 1) })
			pos = pos + seq
		elseif mw.ustring.find(word, "^% ", pos) then
			-- multiword, do not hyphenate
			return nil
		elseif mw.ustring.find(word, "^%.", pos) then
			-- syllable break
			if not mw.ustring.find(oword, "^", pos - offset) then
				offset = offset + 1
			end
			table.insert(parts, { "b", nil })
			pos = pos + 1
		else
			-- unrecognized symbol
			return nil
		end
	end
	return parts
end

local function get_word_suffix(word)
	word = word:gsub("()", "")
	local word_suffix = 0
	for i,v in ipairs(past_tense_suffixes) do
		if word:sub(-string.len(v)) == v 
		then
			word_suffix = 1
		end
	end
	for i,v in ipairs(latin_borrowing_suffixes) do
		if word:sub(-string.len(v)) == v 
		then
			word_suffix = 2
		end
	end
	
	return word_suffix
end

function export.generate_hyphenation(word, otitle)
	local syllables = {}
	local cursyl = ""
	local nucleus = false
	local coda = nil
	local pos = 1
	local parts = partition(word, otitle)
	if not parts then return nil end
	for pos, p in ipairs(parts) do
		local kind, part = unpack(p)
		if kind == "v" then
			if coda then
				cursyl = cursyl .. mw.ustring.sub(syllables, -coda)
				syllables = mw.ustring.sub(syllables, 1, -coda - 1)
				coda = nil
			end
			if nucleus then
				table.insert(syllables, cursyl)
				cursyl = ""
			end
			nucleus = true
			coda = nil
			cursyl = cursyl .. part
		elseif kind == "c" then
			cursyl = cursyl .. part
			if nucleus then
				table.insert(syllables, cursyl)
				cursyl = ""
				nucleus = false
				coda = mw.ustring.len(part)
			else
				coda = nil
			end
		elseif kind == "b" then
			-- implicit syllable break
			if #cursyl > 0 then
				if nucleus or #syllables < 1 then
					table.insert(syllables, cursyl)
				else
					syllables = syllables .. cursyl
				end
			end
			cursyl = ""
			nucleus = false
			coda = nil
		else
			-- unrecognized kind
			return nil
		end
	end
	if #cursyl > 0 then
		if nucleus or #syllables < 1 then
			table.insert(syllables, cursyl)
		else
			syllables = syllables .. cursyl
		end
	end
	return syllables
end

local ipavowel = ""
function export.generate_rhyme(ipa)
	local vowels_at = { }
	local pos = 1
	while true do
		local posnext = mw.ustring.find(ipa, ipavowel, pos)
		if not posnext then break end
		table.insert(vowels_at, posnext)
		pos = posnext + 1
	end
	local vend
	if #vowels_at < 1 then return nil end
	if #vowels_at > 1 then
		vend = vowels_at
	else
		vend = vowels_at
	end
	local snippet = mw.ustring.sub(ipa, vend)
	snippet = mw.ustring.gsub(snippet, "", "")
	if mw.ustring.find(snippet, " ") then
		return nil -- copout, something must be wrong
	end
	return snippet
end

function export.show(frame)
	local args = require("Module:parameters").process(frame:getParent().args, {
		 = { list = true },
		
		 = { list = true, default = nil, allow_holes = true },
		 = { list = true, allow_holes = true },
		 = { list = true, allow_holes = true },
		 = { list = true, allow_holes = true },  = {},
		 = { list = true, allow_holes = true },  = {},
		 = { list = true, default = nil },  = {},
		 = { list = true, default = nil },  = {},
		 = { default = "" },  = {},
		 = { list = true, allow_holes = true },
		
		 = { list = true, default = nil, allow_holes = true },
		 = { list = true, default = nil, allow_holes = true },
		 = { list = true, default = nil, allow_holes = true },
		 = { list = true, default = nil, allow_holes = true },
			
		 = { type = "boolean", default = false },
		 = { type = "boolean", default = false },
		 = { type = "boolean" },  = {},
		
		 = { default = nil }, -- for debugging or demonstration only
	})

	local words, transcriptions, transcriptions_raw
	local lines = {}
	local categories = {}
	local actual = args or mw.title.getCurrentTitle().text
	if next(args) ~= nil then
		words = args
	else
		words = { actual }
	end
	local multiword = mw.ustring.find(words, " ")
	local hyphenations = args
	local rhymes = args
	
	local ipa = args
	if #ipa < 1 then ipa = nil end
	
	local qualifiers = args
	if not qualifiers or qualifiers.maxindex < 1 then qualifiers = args end
	
	local mp = args
	
	local hyphlabels = args
	local rhymlabels = args
	local nohyphen = args
	local norhymes = args
	local fixstress = args
	if args then fixstress = args end
	local homophones = mw.text.split(args, ",")
	local homophonelabels = args
	if #homophones == 1 and homophones == "" then homophones = {} end
	local audio = {}
	local audios = args
	local captions = args
	
	local word_suffix = 0
	if not ipa and #words == 1 then
		-- 0 - normal word
		-- 1 - past tense verb stressed antepenultimately
		-- 2 - Latin borrowing stressed antepenultimately
		word_suffix = get_word_suffix(words)
	end
	if not (fixstress or (fixstress == nil and word_suffix == 1)) then
		word_suffix = 0
	end
		
	if args then hyphenation = args end
	if args then rhymes = args end
	if args then homophones = mw.text.split(args, ",") end
	if args then audios = args end
	if args then captions = args end
	local respelling_ok = true
	for i, w in ipairs(words) do
		if not is_respelling_close_enough(w, actual) then
			respelling_ok = false
			break
		end
	end

	for i, audiofile in ipairs(audios) do
		if audiofile then
			table.insert(audio, {file = audiofile, caption = captions})
		end
	end
	if #hyphenations == 1 and hyphenations == "-" then
		nohyphen = true
	end
	if #rhymes == 1 and rhymes == "-" then
		norhymes = true
	end

	if word_suffix == 0 then
		if ipa then
			transcriptions = {}
			transcriptions_raw = {}
			for i = 1, #ipa do
				local qual = qualifiers
				table.insert(transcriptions, {
					pron = ipa,
					qualifiers = qual and { qual } or nil,
					note = args.n
				})
			end
		else
			transcriptions = {}
			transcriptions_raw = {}
			for i = 1, #words do
				local qual = qualifiers
				local ipaconv = m_pl_IPA.convert_to_IPA(words)
				table.insert(transcriptions_raw, ipaconv)
				table.insert(transcriptions, {
					pron = "/" .. ipaconv .. "/",
					qualifiers = qual and { qual } or nil,
					note = args.n
				})
			end
		end
		table.insert(lines, "* " .. m_IPA.format_IPA_full { lang = lang, items = transcriptions })
	else
		transcriptions = {}
		transcriptions2 = {}
		transcriptions_raw = {}
		if word_suffix == 1 then
			qualifier1 = {"prescriptive standard; rarely used"}
			qualifier2 = {"colloquial; overall more common"}
		elseif word_suffix == 2 then
			qualifier1 = {"standard"}
			qualifier2 = {"colloquial; common in casual speech"}
		end
		local ipaconv = m_pl_IPA.convert_to_IPA(words)
		table.insert(transcriptions_raw, ipaconv)
		ipaconv_syllables = mw.text.split(ipaconv, "()")
		for j, syl in ipairs(ipaconv_syllables) do
			if j == (#ipaconv_syllables - 2) then
				ipaconv_syllables = "ˈ" .. syl
			elseif j ~= 1 then
				ipaconv_syllables = "." .. syl
			end
		end
		ipacov_fixed_stress = table.concat(ipaconv_syllables)
		table.insert(transcriptions, {
			pron = "/" .. ipacov_fixed_stress .. "/",
			qualifiers = qualifier1,
		})
		table.insert(lines, "* " .. m_IPA.format_IPA_full { lang = lang, items = transcriptions })
		table.insert(transcriptions2, {
			pron = "/" .. ipaconv .. "/",
			qualifiers = qualifier2,
		})
		table.insert(lines, "* " .. m_IPA.format_IPA_full { lang = lang, items = transcriptions2 })
	end
	
	if #mp > 0 then
		transcriptions = {}
		for i = 1, #mp do
			if mp == "+" then
				mp = actual
			end
			local ipaconv = require("Module:zlw-mpl-IPA").convert_to_IPA_tables({mp});
			for _, v  in ipairs(ipaconv) do
				table.insert(transcriptions, v);
			end
		end
		table.insert(lines, "* " .. require("Module:accent qualifier").format_qualifiers(lang, {"Middle Polish"}) .. " " .. m_IPA.format_IPA_full { lang = lang, items = transcriptions })
	end
	
	for i, a in ipairs(audio) do
		table.insert(lines, "* " .. require("Module:audio").format_audio { lang = lang, file = a, caption = a })
	end

	if not ipa and #hyphenations < 1 and respelling_ok and not multiword then
		local autohyph = export.generate_hyphenation(words, actual)
		if autohyph then
			table.insert(hyphenations, autohyph)
		end
	elseif #hyphenations >= 1 then
		local newhyphenations = {}
		for i, h in ipairs(hyphenations) do
			local t = {}
			for x in mw.text.gsplit(h, "") do
				table.insert(t, x)	
			end
			newhyphenations = t
		end
		hyphenations = newhyphenations
	end

	if not norhymes then
		if not ipa and #rhymes < 1 and #transcriptions_raw > 0 then
			local autorhyme = export.generate_rhyme(transcriptions_raw)
			if autorhyme then
				table.insert(rhymes, autorhyme)
			end
		end
	
		if #rhymes > 0 then
			-- merge rhymes if they have identical labels
			local last_label = false
			local new_rhymes = {}
			local new_labels = {}
			local current_list = {}
			
			for i, r in ipairs(rhymes) do
				local label = rhymlabels
				if last_label == label then
					table.insert(current_list, r)
				else
					if #current_list > 0 then
						table.insert(new_rhymes, current_list)
					end
					if last_label ~= false then
						table.insert(new_labels, last_label)
					end
					current_list = { r }
					last_label = label
				end
			end
			
			table.insert(new_rhymes, current_list)
			table.insert(new_labels, last_label)
			rhymes = new_rhymes
			rhymlabels = new_labels
		end
		
		for i, r in ipairs(rhymes) do
			local label = ""
			if rhymlabels then
				label = " " .. require("Module:qualifier").format_qualifier(rhymlabels)
			end
			if #r >= 1 then
				local sylkeys = {}
				local sylcounts = {}
				-- get all possible syllable counts from syllabifications
				for i, h in ipairs(hyphenations) do
					local hl = #h
					if hl > 0 and not sylkeys then
						table.insert(sylcounts, hl)
						sylkeys = true
					end
				end
				local rhymeobjs = {}
				for _, rhyme in ipairs(r) do
					table.insert(rhymeobjs, {rhyme = rhyme})
				end
				table.insert(lines, "* " .. require("Module:rhymes").format_rhymes(
					{ lang = lang, rhymes = rhymeobjs, num_syl = sylcounts }) .. label)
			end
		end
	end

	if not nohyphen then
		if #transcriptions > 0 and #hyphenations > 0 then
			local syl_IPA = require("Module:syllables").getVowels(transcriptions.pron, lang)
			local syl_hyph = #hyphenations
			if syl_IPA and syl_IPA ~= syl_hyph then
				table.insert(categories, "pl-pronunciation syllable count mismatch")
			end
		end
		
		if not actual:find("") and #hyphenations < 1 then
			table.insert(categories, "pl-pronunciation without hyphenation")
		end
		
		for i, h in ipairs(hyphenations) do
			local label = ""
			if hyphlabels then
				label = " " .. require("Module:qualifier").format_qualifier(hyphlabels)
			end
			table.insert(lines, "* Syllabification: " .. require("Module:links").full_link({lang = lang, alt = table.concat(h, "‧"), tr = "-"}) .. label)
		end
	end

	if #homophones > 0 then
		local homophone_objs = {}
		for i, h in ipairs(homophones) do
			table.insert(homophone_objs, {term = h, qq = homophonelabels and {homophonelabels} or nil})
		end
		table.insert(lines, "* " .. require("Module:homophones").format_homophones { lang = lang, homophones = homophone_objs })
	end
	
	return table.concat(lines, "\n") .. require("Module:utilities").format_categories(categories, lang)
end

return export