Module:ast-IPA

-- Based on [[Module:es-pronunc]] by: Benwing
-- Adapted by Santi2222

local export = {}

local m_IPA = require("Module:IPA")

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

local u = require("Module:string/char")
local rfind = mw.ustring.find
local rsubn = mw.ustring.gsub
local rsplit = mw.text.split
local ulower = mw.ustring.lower

local AC = u(0x0301) -- acute =  ́
local GR = u(0x0300) -- grave =  ̀
local CFLEX = u(0x0302) -- circumflex =  ̂
local TILDE = u(0x0303) -- tilde =  ̃
local DIA = u(0x0308) -- diaeresis =  ̈

local vowel = "aeiouy" -- vowel; include y so we get single-word y correct
local V = "[" .. vowel .. "]"
local W = "[jw]" -- glide
local accent = AC .. GR .. CFLEX
local accent_c = "[" .. accent .. "]"
local stress_c = "[" .. AC .. GR .. "]"
local ipa_stress = "ˈˌ"
local ipa_stress_c = "[" .. ipa_stress .. "]"
local separator = accent .. ipa_stress .. "# ."
local separator_c = "[" .. separator .. "]"
local C = "[^" .. vowel .. separator .. "]" -- consonant
local T = "[^" .. vowel .. "lrɾjw" .. separator .. "]" -- obstruent or nasal

local unstressed_words = require("Module:table").listToSet({ --feel free to add more unstressed words
	"el", "la", "les", "lo", "los", -- definite articles
	"que", "quien", "onde", "au", "le", "nos", "os", -- relative pronouns
	"mos", "nos", "so", "te", -- unstressed pronouns
	"y", "ya", "o", -- conjunctions
	"con", "a", "de", "en", "pa", "per", "por", "sin", "tres" --prepositions
})

-- 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

-- ɟ and ĉ are used internally to represent [ʝ⁓ɟ͡ʝ] and [t͡ʃ]
--
function export.IPA(text, phonetic)
	local debug = {}
	------------------------------------

	text = ulower(text or mw.title.getCurrentTitle().text)
	-- decompose everything but ñ and ü
	text = mw.ustring.toNFD(text)
	text = rsub(text, ".[" .. TILDE .. DIA .. "]", {
		["n" .. TILDE] = "ñ",
		["u" .. DIA] = "ü",
	})
	-- 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*[¡!¿?]%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)

	-- Make prefixes unstressed unless they have an explicit stress marker; also make certain
	-- monosyllabic words (e.g. [[el]], [[la]], [[de]], [[en]], etc.) without stress marks be
	-- unstressed.
	local words = rsplit(text, " ")
	for i, word in ipairs(words) do
		if rfind(word, "%-$") and not rfind(word, accent_c) or unstressed_words[word] then
			-- add CFLEX to the last vowel not the first one, or we will mess up 'que' by
			-- adding the CFLEX after the 'u'
			words[i] = rsub(word, "^(.*" .. V .. ")", "%1" .. CFLEX)
		end
	end
	text = table.concat(words, " ")
	-- Convert hyphens to spaces
	text = rsub(text, "-y", "y") --intercept the pronoun "-y"
	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, " ", "# #") .. "##"

	table.insert(debug, text)

	--determining whether "y" is a consonant or a vowel
	text = rsub(text, "y(" .. V .. ")", "ɟ%1") -- not the real sound
	text = rsub(text, "uy#", "uY#") -- a temporary symbol; replaced with i below
	text = rsub(text, "y", "i")

	--x
	text = rsub(text, "x", "ʃ")

	--c, g, q
	text = rsub(text, "c([ie])", "θ%1")
	text = rsub(text, "gü([ie])", "ɡw%1")
	text = rsub(text, "gu([ie])", "ɡ%1") -- special IPA g, not normal g
	text = rsub(text, "g([ie])", "x%1") -- must happen after handling of x above
	-- following must happen before stress assignment; [[branding]] has initial stress like 'brandin'
	text = rsub(text, "ng([^aeiouüwhlr])", "n%1") -- [[Bangkok]], [[ángstrom]], [[branding]]
	text = rsub(text, "qu([ie])", "k%1")
	text = rsub(text, "ü", "u") -- [[Düsseldorf]], [[hübnerita]], obsolete [[freqüentemente]], etc.
	text = rsub(text, "q", "k") -- [[quark]], [[Qatar]], [[burqa]], [[Iraq]], etc.
	text = rsub(text, "z", "θ")

	table.insert(debug, text)

	--alphabet-to-phoneme
	text = rsub(text, "ch", "ĉ") --not the real sound
	text = rsub(text, "#p([st])", "#%1") -- [[psicología]], [[pterodáctilo]]
	text = rsub(text, "[cgjñrvy]",
	--["g"]="ɡ":  U+0067 LATIN SMALL LETTER G → U+0261 LATIN SMALL LETTER SCRIPT G
			{ ["c"] = "k", ["g"] = "ɡ", ["j"] = "x", ["ñ"] = "ɲ", ["r"] = "ɾ", ["v"] = "b" })

	-- double l
	text = rsubb(text, "ll", "ʎ")

	-- trill in #r, lr, nr, sr, rr
	text = rsub(text, "ɾɾ", "r")
	-- FIXME: does this also apply to /θr/ (e.g. [[Azrael]], [[cruzrojista]])?
	text = rsub(text, "([#lnsz])ɾ", "%1r")

	-- reduce any remaining double consonants ([[Addis Abeba]], [[cappa]], [[descender]] in Latin America ...);
	-- do this before handling of -nm- e.g. in [[inmigración]], which generates a double consonant, and do this
	-- before voicing stops before obstruents, to avoid problems with [[cappa]] and [[crackear]]
	text = rsub(text, "(" .. C .. ")%1", "%1")

	-- voiceless stop to voiced before obstruent or nasal; but intercept -ts-, -tz-
	local voice_stop = { ["p"] = "b", ["t"] = "d", ["k"] = "ɡ" }
	text = rsub(text, "t(" .. separator_c .. "*[szθ])", "!%1") -- temporary symbol
	text = rsub(text, "([ptk])(" .. separator_c .. "*" .. T .. ")",
			function(stop, after)
				return voice_stop[stop] .. after
			end)
	text = rsub(text, "!", "t")

	text = rsub(text, "n([# .]*[bpm])", "m%1")

	table.insert(debug, text)

	--syllable division
	local vowel_to_glide = { ["i"] = "j", ["u"] = "w" }
	-- i and u between vowels -> j and w ([[paranoia]], [[baiano]], [[abreuense]], [[alauita]], [[Malaui]], etc.)
	text = rsub_repeatedly(text, "(" .. V .. accent_c .. "*)([iu])(" .. V .. ")",
			function(v1, iu, v2)
				return v1 .. vowel_to_glide[iu] .. v2
			end
	)
	text = rsub_repeatedly(text, "(" .. V .. accent_c .. "*)(" .. C .. W .. "?" .. V .. ")", "%1.%2")
	text = rsub_repeatedly(text, "(" .. V .. accent_c .. "*" .. C .. ")(" .. C .. V .. ")", "%1.%2")
	text = rsub_repeatedly(text, "(" .. V .. accent_c .. "*" .. C .. "+)(" .. C .. C .. V .. ")", "%1.%2")
	text = rsub(text, "([pbktdɡ])%.([lɾ])", ".%1%2")
	text = rsub_repeatedly(text, "(" .. C .. ")%.s(" .. C .. ")", "%1s.%2")
	-- Any aeo, or stressed iu, should be syllabically divided from a following aeo or stressed iu.
	text = rsub_repeatedly(text, "([aeo]" .. accent_c .. "*)([aeo])", "%1.%2")
	text = rsub_repeatedly(text, "([aeo]" .. accent_c .. "*)(" .. V .. stress_c .. ")", "%1.%2")
	text = rsub(text, "([iu]" .. stress_c .. ")([aeo])", "%1.%2")
	text = rsub_repeatedly(text, "([iu]" .. stress_c .. ")(" .. V .. stress_c .. ")", "%1.%2")
	text = rsub_repeatedly(text, "i(" .. accent_c .. "*)i", "i%1.i")
	text = rsub_repeatedly(text, "u(" .. accent_c .. "*)u", "u%1.u")

	table.insert(debug, text)

	--diphthongs
	text = rsub(text, "ih?([aeou])", "j%1")
	text = rsub(text, "uh?([aeio])", "w%1")

	table.insert(debug, text)

	local accent_to_stress_mark = { [AC] = "ˈ", [GR] = "ˌ", [CFLEX] = "" }

	local function accent_word(word, syllables)
		-- Now stress the word. If any accent exists in the word (including ^ indicating an unaccented word),
		-- put the stress mark(s) at the beginning of the indicated syllable(s). Otherwise, apply the default
		-- stress rule.
		if rfind(word, accent_c) then
			for i = 1, #syllables do
				syllables[i] = rsub(syllables[i], "^(.*)(" .. accent_c .. ")(.*)$",
						function(pre, accent, post)
							return accent_to_stress_mark[accent] .. pre .. post
						end
				)
			end
		else
			-- Default stress rule. Words without vowels (e.g. IPA foot boundaries) don't get stress.
			if #syllables > 1 and rfind(word, "[^aeiouns#]#") or #syllables == 1 and rfind(word, "[aeiou]") then
				syllables[#syllables] = "ˈ" .. syllables[#syllables]
			elseif #syllables > 1 then
				syllables[#syllables - 1] = "ˈ" .. syllables[#syllables - 1]
			end
		end
	end

	local words = rsplit(text, " ")
	for j, word in ipairs(words) do
		-- accentuation
		local syllables = rsplit(word, "%.")

		if rfind(word, "men%.te#") then
			local mente_syllables
			-- Words ends in -mente (converted above to ménte); add a stress to the preceding portion
			-- (e.g. [[agriamente]] -> 'ágriaménte') unless already stressed (e.g. [[rápidamente]]).
			-- It will be converted to secondary stress further below. Essentially, we rip the word apart
			-- into two words ('mente' and the preceding portion) and stress each one independently.
			mente_syllables = {}
			mente_syllables[2] = table.remove(syllables)
			mente_syllables[1] = table.remove(syllables)
			accent_word(table.concat(syllables, "."), syllables)
			accent_word(table.concat(mente_syllables, "."), mente_syllables)
			table.insert(syllables, mente_syllables[1])
			table.insert(syllables, mente_syllables[2])
		else
			accent_word(word, syllables)
		end

		-- 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[i] = rsub(syllables[i], "(" .. V .. ")(" .. V .. ")([mnɲ])",
						"%1" .. TILDE .. "%2" .. TILDE .. "%3")
				-- then for one vowel
				syllables[i] = rsub(syllables[i], "(" .. V .. ")([mnɲ])", "%1" .. TILDE .. "%2")
			end
		end

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

	text = table.concat(words, " ")

	text = rsub(text, "Y", "i") --final -uy
	-- suppress syllable mark before IPA stress indicator
	text = rsub(text, "%.(" .. ipa_stress_c .. ")", "%1")
	--make all primary stresses but the last one be secondary
	text = rsub_repeatedly(text, "ˈ(.+)ˈ", "ˌ%1ˈ")

	--phonetic transcription
	if phonetic then
		-- θ, s, f before voiced consonants
		local voiced = "mnɲbdɟɡʎl"
		local r = "ɾr"
		local tovoiced = {
			["θ"] = "θ̬",
			["s"] = "z",
			["ʃ"] = "ʒ",
		}
		local function voice(sound, following)
			return tovoiced[sound] .. following
		end
		text = rsub(text, "([θsʃ])(" .. separator_c .. "*[" .. voiced .. r .. "])", voice)

		-- fricative vs. stop allophones; first convert stops to fricatives, then back to stops
		-- after nasals and sometimes after l
		local stop_to_fricative = { ["b"] = "β", ["d"] = "ð", ["ɟ"] = "ʝ", ["ɡ"] = "ɣ" }
		local fricative_to_stop = { ["β"] = "b", ["ð"] = "d", ["ʝ"] = "ɟ", ["ɣ"] = "ɡ" }
		text = rsub(text, "[bdɟɡ]", stop_to_fricative)
		text = rsub(text, "([mnɲ]" .. separator_c .. "*)([βɣ])",
				function(nasal, fricative)
					return nasal .. fricative_to_stop[fricative]
				end
		)
		text = rsub(text, "([lʎmnɲ]" .. separator_c .. "*)([ð])",
				function(nasal_l, fricative)
					return nasal_l .. fricative_to_stop[fricative]
				end
		)
		text = rsub(text, "([mnɲ]" .. separator_c .. "*)([ʝ])",
				function(nasal_l, fricative)
					return nasal_l .. fricative_to_stop[fricative]
				end
		)
		text = rsub(text, "(##" .. ipa_stress_c .. "*)([βɣðʝ])",
				function(stress, fricative)
					return stress .. fricative_to_stop[fricative]
				end
		)
		text = rsub(text, "[td]", { ["t"] = "t̪", ["d"] = "d̪" })

		--dental s/z
		local dental = "dtθð"
		local todental = {
			["s"] = "s̪",
			["z"] = "z̪",
		}
		local function dent(sound, following)
			return todental[sound] .. following
		end
		text = rsub(text, "([sz])(" .. separator_c .. "*[" .. dental .. "])", dent)

		-- nasal assimilation before consonants
		local labiodental, dentialveolar, dental, palatal, velar, alveolar = "ɱ", "n̪", "n̟", "ɲ", "ŋ", "n"
		local nasal_assimilation = {
			["f"] = labiodental,
			["t"] = dentialveolar, ["d"] = dentialveolar,
			["θ"] = dental,
			["ĉ"] = palatal,
			["ɲ"] = palatal,
			["ʃ"] = palatal,
			["ʒ"] = palatal,
			["ɟ"] = palatal, ["ʎ"] = palatal,
			["k"] = velar, ["x"] = velar, ["ɡ"] = velar,
			["s"] = alveolar, ["z"] = alveolar, ["l"] = alveolar, ["r"] = alveolar, ["ɾ"] = alveolar, ["n"] = alveolar,
		}
		text = rsub(text, "n(" .. separator_c .. "*)(.)",
				function(stress, following)
					return (nasal_assimilation[following] or "n") .. stress .. following
				end
		)

		--velar ŋ in final position before vowels
		text = rsub(text, "n#", "ŋ#")
		text = rsub(text, "ŋ(" .. separator_c .. "*)(.)",
				function(stress, following)
					return (nasal_assimilation[following] or "ŋ") .. 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 == "θ" then
						-- dental
						l = "l̟"
					elseif following == "ĉ" or following == "ɲ" or following == "ɟ" or following == "ʎ" or following == "ʃ" then
						-- palatal
						l = "ʎ"
					end
					return l .. stress .. following
				end)

		--semivowels
		text = rsub(text, "eu", "øu")
		text = rsub(text, "([aeouãẽõũ][iĩ])", "%1̯")
		text = rsub(text, "([aeioãẽĩõø][uũ])", "%1̯")

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

	end

	table.insert(debug, text)

	-- remove silent "h" and convert fake symbols to real ones
	local final_conversions = {
		["h"] = "", -- silent "h"
		["ĉ"] = "t͡ʃ", -- fake "ch" to real "ch"
		["ɟ"] = phonetic and "ɟ͡ʝ" or "ʝ", -- fake "y" to real "y"
	}
	local final_conversions_phonetic = { -- post-alveolars are actually palatal or pre-palatal
		["ʃ"] = "ɕ",
		["ʒ"] = "ʑ",
		["ĉ"] = "t͡ɕ",
	}

	if phonetic then
		text = rsub(text, "[ʃʒĉ]", final_conversions_phonetic)
	end
	text = rsub(text, "[hĉɟ]", final_conversions)

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

	return mw.ustring.toNFC(text)
end

function export.show(frame)
	local params = {
		[1] = {},
		["pre"] = {},
	}

	local parargs = frame:getParent().args
	local args = require("Module:parameters").process(parargs, params)

	local results = {}

	local text = args[1] or mw.title.getCurrentTitle().text

	table.insert(results, { pron = "/" .. export.IPA(text, false) .. "/" })
	table.insert(results, { pron = "[" .. export.IPA(text, true) .. "]" })

	local pre = args.pre and args.pre .. " " or ""

	return "* " .. pre .. m_IPA.format_IPA_full { lang = lang, items = results }
end

return export