Module:eu-pron

-- Based on ] by: Benwing
-- Adapted by Santi2222

local export = {}

local m_IPA = require("Module:IPA")
local m_str_utils = require("Module:string utilities")

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

local u = m_str_utils.char
local rsubn = m_str_utils.gsub
local rsplit = m_str_utils.split
local toNFC = mw.ustring.toNFC
local toNFD = mw.ustring.toNFD
local ulower = m_str_utils.lower
local ulen = m_str_utils.len

local TILDE = u(0x0303) -- tilde =  ̃

local vowel = "aeiou" -- vowel;
local V = ""
local W = "" -- glide
local separator = "# ."
local separator_c = ""
local C = "" -- consonant

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

-- version of rsubn() that returns a 2nd argument boolean indicating whether
-- a substitution was made.
local function rsubb(term, foo, bar)
	local retval, nsubs = rsubn(term, foo, bar)
	return retval, nsubs > 0
end

-- apply rsub() repeatedly until no change
local function rsub_repeatedly(term, foo, bar)
	while true do
		local new_term = rsub(term, foo, bar)
		if new_term == term then
			return term
		end
		term = new_term
	end
end

-- style == one of the following:
-- "northern_style": lack of palatalization, /h/, <j> = /j/
-- "southern_style": palatalization, lack of /h/, <j> = /x/

function export.IPA(text, style, phonetic)

	local northern = style == "northern_style"
	local n_diff = false

	text = ulower(text or mw.title.getCurrentTitle().text)
	-- decompose everything but ñ
	text = toNFD(text)
	text = rsub(text, ".", {
		 = "ñ",
	})
	-- convert commas and en/en dashes to IPA foot boundaries
	text = rsub(text, "%s*%s*", " | ")
	-- question mark or exclamation point in the middle of a sentence -> IPA foot boundary
	text = rsub(text, "()%s*%s*()", "%1 | %2")

	-- canonicalize multiple spaces and remove leading and trailing spaces
	local function canon_spaces(text)
		text = rsub(text, "%s+", " ")
		text = rsub(text, "^ ", "")
		text = rsub(text, " $", "")
		return text
	end

	text = canon_spaces(text)

	-- Convert hyphens to spaces
	text = rsub(text, "%-", " ")
	-- canonicalize multiple spaces again, which may have been introduced by hyphens
	text = canon_spaces(text)
	-- now eliminate punctuation
	text = rsub(text, "", "")
	-- put # at word beginning and end and double ## at text/foot boundary beginning/end
	text = rsub(text, " | ", "# | #")
	text = "##" .. rsub(text, " ", "# #") .. "##"

	--cases in which <j> is always /x/ (requires respelling)
	text = rsubb(text, "kh", "X")

	--determining whether "h" denotes a phoneme, or the lack of palatalization, palatalization when needed
	text = rsub(text, "nh", "N")
	text = rsub(text, "lh", "L")

	text = rsub(text, "()il(*)", (northern and "%1ił%2" or "%1ĺ%2"))
	text = rsub(text, "()in(*)", (northern and "%1iň%2" or "%1ń%2"))

	text = rsub(text, "uin(*)", (northern and "uiň%1" or "uiń%1"))
	text = rsub(text, "uil(*)", (northern and "uił%1" or "uiĺ%1"))

	text = rsub(text, "il(*)", (northern and "ił%1" or "iĺ%1"))
	text = rsub(text, "in(*)", (northern and "iň%1" or "iń%1"))

	if text:find("") then
		n_diff = true
	end

	text = rsub(text, "ĺ", "ʎ")
	text = rsub(text, "ń", "ɲ")
	text = rsub(text, "N", "n")
	text = rsub(text, "L", "l")
	text = rsub(text, "ň", "n")
	text = rsub(text, "ł", "l")

	--c, g, q
	text = rsub(text, "ng()", "n%1") -- ], ], etc.
	text = rsub(text, "q", "k") -- ], ], ], ], etc.
	text = rsub(text, "c", "k") -- ], etc.

	--alphabet-to-phoneme
	text = rsub(text, "tx", "C") --not the real sound
	text = rsub(text, "tz", "Ś") --not the real sound
	text = rsub(text, "ts", "S") --not the real sound
	text = rsubb(text, "ll", "ʎ")
	text = rsubb(text, "dd", "ɟ")
	text = rsubb(text, "tt", "c")
	text = rsub(text, "x", "ʃ")
	text = rsub(text, "#p()", "#%1") -- ], ]
	text = rsub(text, "",
			{  = "ɡ",  = "ɲ",  = "b",  = "ś" })

	--/x/ sound
	text = rsubb(text, "j", (northern and "J" or "x"))
	if text:find("") then
		n_diff = true
	end
	text = rsub(text, "J", "j")

	--cases in which <j> is always /j/ (requires respelling)
	text = rsubb(text, "y", "j")

	-- trilled r (in all cases except between vowels)
	text = rsub(text, "()r()", "%1ɾ%2")
	text = rsub(text, "()r()", "%1ɾ%2") -- it has to be applied twice in order to handle VrVrV sequences
	text = rsub(text, "rr", "r")

	text = rsub(text, "n(*)", "m%1")
	
	-- lack of /h/ in Southern dialects
	text = rsub(text, "aha", (northern and "aha" or "a"))
	text = rsub(text, "ehe", (northern and "ehe" or "e"))
	text = rsub(text, "ihi", (northern and "ihi" or "i"))
	text = rsub(text, "oho", (northern and "oho" or "o"))
	text = rsub(text, "uhu", (northern and "uhu" or "u"))
	text = rsub(text, "h", (northern and "h" or ""))
	
	----------------------
	--syllable division
	
	-- i and u between vowels -> i. and u.
	text = rsub_repeatedly(text, "ui", "U") --this dipthong requires special treatment
	
	local vowel_to_glide = {  = "i.",  = "u." }
	text = rsub_repeatedly(text, "(" .. V .. "*)()(" .. V .. ")",
			function(v1, iu, v2)
				return v1 .. vowel_to_glide .. v2
			end
	)
	text = rsub_repeatedly(text, "U", "ui")
	
	text = rsub_repeatedly(text, "(" .. V .. ")(" .. C .. W .. "?" .. V .. ")", "%1.%2")
	text = rsub_repeatedly(text, "(" .. V .. C .. ")(" .. C .. V .. ")", "%1.%2")
	text = rsub_repeatedly(text, "(" .. V .. C .. ")(" .. C .. C .. V .. ")", "%1.%2")
	text = rsub_repeatedly(text, "(" .. V .. C .. C .. ")(" .. C .. C .. V .. ")", "%1.%2")
	text = rsub(text, "()%.()", ".%1%2")
	text = rsub_repeatedly(text, "(" .. C .. ")%.s(" .. C .. ")", "%1s.%2")
	-- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
	text = rsub_repeatedly(text, "(" .. ")ui", "%1u.i")
	text = rsub_repeatedly(text, "(" .. ")()", "%1.%2")
	text = rsub_repeatedly(text, "ii", "i.i")
	text = rsub_repeatedly(text, "(" .. ")u", "%1.u")

	local words = rsplit(text, " ")
	for j, word in ipairs(words) do

		local syllables = rsplit(word, "%.")

		-- Vowels are nasalized if followed by nasal in same syllable.
		if phonetic then
			for i = 1, #syllables do
				-- first check for two vowels (veinte)
				syllables = rsub(syllables, "(" .. V .. ")(" .. V .. ")()",
						"%1" .. TILDE .. "%2" .. TILDE .. "%3")
				-- then for one vowel
				syllables = rsub(syllables, "(" .. V .. ")()", "%1" .. TILDE .. "%2")
			end
		end

		-- Reconstruct the word.
		words = table.concat(syllables, phonetic and "." or "")
	end

	text = table.concat(words, " ")

	--phonetic transcription
	if phonetic then
		-- s, z, f before voiced consonants
		local voiced = "mnɲbdɟɡʎl"
		local r = "ɾr"
		local tovoiced = {
			 = "z̺",
			 = "z̻",
		}
		local function voice(sound, following)
			return tovoiced .. following
		end
		text = rsub(text, "()(" .. separator_c .. "*)", voice)

		-- fricative vs. stop allophones; first convert stops to fricatives, then back to stops
		-- after nasals and sometimes after l
		local stop_to_fricative = {  = "β",  = "ð",  = "ɣ" }
		local fricative_to_stop = {  = "b",  = "d",  = "ɡ" }
		text = rsub(text, "", stop_to_fricative)
		text = rsub(text, "(" .. separator_c .. "*)()",
				function(nasal, fricative)
					return nasal .. fricative_to_stop
				end
		)
		text = rsub(text, "(" .. separator_c .. "*)()",
				function(nasal_l, fricative)
					return nasal_l .. fricative_to_stop
				end
		)
		text = rsub(text, "##β", "##b")
		text = rsub(text, "##ð", "##d")
		text = rsub(text, "##ɣ", "##ɡ")

		text = rsub(text, "", {  = "t̪",  = "d̪" })

		-- nasal assimilation before consonants
		local labiodental, dentialveolar, alveolopalatal, palatal, velar = "ɱ", "n̪", "nʲ", "ɲ", "ŋ"
		local nasal_assimilation = {
			 = labiodental,
			 = dentialveolar,  = dentialveolar,
			 = alveolopalatal,
			 = alveolopalatal,
			 = palatal,  = palatal,  = palatal,  = palatal,
			 = velar,  = velar,  = velar,
		}
		text = rsub(text, "n(" .. separator_c .. "*)(.)",
				function(stress, following)
					return (nasal_assimilation or "n") .. stress .. following
				end
		)

		-- lateral assimilation before consonants
		text = rsub(text, "l(" .. separator_c .. "*)(.)",
				function(stress, following)
					local l = "l"
					if following == "t" or following == "d" then
						-- dentialveolar
						l = "l̪"
					elseif following == "C" or following == "ʃ" then
						-- alveolopalatal
						l = "lʲ"
					elseif following == "ɟ" or following == "c" or following == "j" then
						-- alveolopalatal
						l = "ʎ"
					end
					return l .. stress .. following
				end)

		-- voiced fricatives are actually approximants
		text = rsub(text, "()", "%1̞")

		-- northern /h/ is usually 
		text = rsub(text, "h", "ɦ")
	end

	--semivowels
	text = rsub(text, "()" .. TILDE .. "()", "%1" .. TILDE .. "%2̯")
	text = rsub(text, "()" .. TILDE .. "()", "%1" .. TILDE .. "%2̯")
	text = rsub(text, "()", "%1̯")
	text = rsub(text, "()", "%1̯")
	text = rsub(text, "u̯i̯", "u̯i") --correct the mistake appeareing in words with aui, eui
	
	if phonetic then
		-- /e/ and /o/ are actually lowered
		text = rsub(text, "", {  = "e̞",  = "o̞" })
	end
	
	-- convert fake symbols to real ones
	local final_conversions = {
		 = "t͡s̺",
		 = "t͡s̻",
		 = "s̺",
		 = "s̻",
		 = "t͡ʃ",
		 = "x"
	}
	text = rsub(text, "", final_conversions)

	-- remove # symbols at word and text boundaries
	text = rsub(text, "#", "")
	text = toNFC(text)

	local ret = {
		text = text,
		northern_different = n_diff,
	}

	return ret
end

-- %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
function export.show(frame)
	local params = {
		 = {},
		 = {},
		 = {},
		 = {},
		 = {},
		 = { type = "number", default = 1 },
	}
	local parargs = frame:getParent().args
	local args = require("Module:parameters").process(parargs, params)
	local phonemic = {}
	local phonetic = {}
	local expressed_styles = {}
	local text = args or mw.title.getCurrentTitle().text
	local function dostyle(style)
		phonemic = export.IPA(text, style, false)
		phonetic = export.IPA(text, style, true)
	end
	local function express_style(hidden_tag, tag, style)
		if not phonemic then
			dostyle(style)
		end
		local new_style = {
			tag = tag,
			phonemic = phonemic,
			phonetic = phonetic,
		}
		for _, hidden_tag_style in ipairs(expressed_styles) do
			if hidden_tag_style.tag == hidden_tag then
				table.insert(hidden_tag_style.styles, new_style)
				return
			end
		end
		table.insert(expressed_styles, {
			tag = hidden_tag,
			styles = { new_style },
		})
	end
	dostyle("northern_style")
	local northern_different = phonemic.northern_different
	if not northern_different then
		express_style(false, false, "southern_style")
	else
		express_style("Southern", "Southern", "southern_style")
		express_style("Northern", "Northern", "northern_style")
	end

	-- If only one style group, don't indicate the style.
	if #expressed_styles == 1 then
		expressed_styles.tag = false
		if #expressed_styles.styles == 1 then
			expressed_styles.styles.tag = false
		end
	end

	local lines = {}

	local function format_style(tag, expressed_style, is_first)
		local pronunciations = {}
		table.insert(pronunciations, {
			pron = "/" .. expressed_style.phonemic.text .. "/",
			qualifiers = tag and { tag } or nil,
		})
		table.insert(pronunciations, {
			pron = "",
		})
		local bullet = string.rep("*", args.bullets) .. " "
		local pre = is_first and args.pre and args.pre .. " " or ""
		local post = is_first and (args.ref or "") .. (args.post and " " .. args.post or "") or ""
		local formatted = bullet .. pre .. m_IPA.format_IPA_full { lang = lang, items = pronunciations } .. post
		local formatted_for_len = bullet .. pre .. "IPA(key): " .. (tag and "(" .. tag .. ") " or "") ..
				"/" .. expressed_style.phonemic.text .. "/, " .. post
		return formatted, formatted_for_len
	end

	for i, style_group in ipairs(expressed_styles) do
		if #style_group.styles == 1 then
			style_group.formatted, style_group.formatted_for_len = format_style(style_group.styles.tag, style_group.styles, i == 1)
		else
			style_group.formatted, style_group.formatted_for_len = format_style(style_group.tag, style_group.styles, i == 1)
			for j, style in ipairs(style_group.styles) do
				style.formatted, style.formatted_for_len = format_style(style.tag, style, i == 1 and j == 1)
			end
		end
	end

	local maxlen = 0
	for _, style_group in ipairs(expressed_styles) do
		local this_len = ulen(style_group.formatted_for_len)
		if #style_group.styles > 1 then
			for _, style in ipairs(style_group.styles) do
				this_len = math.max(this_len, ulen(style.formatted_for_len))
			end
		end
		maxlen = math.max(maxlen, this_len)
	end

	for _, style_group in ipairs(expressed_styles) do
		if #style_group.styles == 1 then
			table.insert(lines, style_group.formatted)
		else
			local inline = '\n<div class="vsShow" style="display:none">\n' .. style_group.formatted .. "</div>"
			local full_prons = {}
			for _, style in ipairs(style_group.styles) do
				table.insert(full_prons, style.formatted)
			end
			local full = '\n<div class="vsHide">\n' .. table.concat(full_prons, "\n") .. "</div>"
			local em_length = math.floor(maxlen * 0.68) -- from ]
			table.insert(lines, '<div class="vsSwitcher" data-toggle-category="pronunciations" style="width: ' .. em_length .. 'em; max-width:100%;"><span class="vsToggleElement" style="float: right;">&nbsp;</span>' .. inline .. full .. "</div>")
		end
	end

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

return export