Module:Hv-pronunc

Documentation for this module may be created at Module:Hv-pronunc/documentation

-- Based on https://linguifex.com/wiki/Module:qhv-pron by https://linguifex.com/wiki/User:Sware with permission. 

local sub = mw.ustring.sub
local find = mw.ustring.find
local gmatch = mw.ustring.gmatch
local gsub = mw.ustring.gsub
local match = mw.ustring.match
local u = mw.ustring.char
local split = mw.text.split
local gsplit = mw.text.gsplit

local export = {}

local palatal = "ɲʎɟj"
local velar = "kɡxɣ"
local uvular = "q"
local consonants = "[mpbvwθntdszlrɾɲŋɴʎhɥjɟṛʃ" .. velar .. uvular .. "]"
local NONSYLLABIC = u(0x032F) -- non-syllabic, combining inverted breve below
local DIPHTHONG = u(0x035C) -- double articulation, combining double breve below
local vowels = "[aeiouyáéíóúýàèìòùỳː" .. NONSYLLABIC .. DIPHTHONG .. "]"
local vowelsonly = "[aeiouyáéíóúýàèìòùỳ" .. NONSYLLABIC .. DIPHTHONG .. "]"

local rules = {
	{"rh", "ṛ"}, {"mh", "m"}, {"th", "θ"}, {"lj", "ʎ"}, {"ñ", "ɲ"}, {"kh", "x"}, {"[gɡ]h", "ɣ"}, {"g", "ɡ"}, {"j", "ɟ"}, {"sh", "ʃ"},
	
	{"([aāáà][eo])", "%1" .. NONSYLLABIC}, {"([iíuú])([aāeēoōáéóàèò])", "%1" .. DIPHTHONG .. "%2"}, 
	{"ā", "aː"}, {"ē", "eː"}, {"ī", "iː"}, {"ō", "oː"}, {"ū", "uː"}, {"ȳ", "yː"},
	{"à", "áː"}, {"è", "éː"}, {"ì", "íː"}, {"ò", "óː"}, {"ù", "úː"}, {"ỳ", "ýː"},
	
	{"n([" .. palatal .. "])", "ɲ%1"}, {"n([" .. velar .. "])", "ŋ%1"}, {"n([" .. uvular .. "])", "ɴ%1"}, 
	{"ɲi", "ni"}, {"ʎi", "li"}, 
}

local deacuter = {
	["á"] = "a", ["é"] = "e", ["í"] = "i", ["ó"] = "o", ["ú"] = "u", ["ý"] = "y",
}

function export.write_stress(term, from_module)
	local pattern = "(" .. consonants .. "*".. vowels .. "*" .. consonants .. "-)"
	local weight = {}
	term = gsub(term, pattern, "·%1")
	term = gsub(term, "^·", "")
	term = gsub(term, "·$", "")
	term = gsub(term, "·(" .. consonants .. ")·", "·%1")
	term = gsub(term, "·(" .. consonants .. ")$", "%1")
	term = gsub(term, "·(" .. consonants .. ")(" .. consonants .. ")", "%1·%2")  
	term = gsub(term, "([ptkbdɡ])·([rlṛsz])", "·%1%2")
	term = gsub(term, "·(" .. consonants .. "?" .. consonants .. ")$", "%1")
	
	local syllables = split(term, "·")
	
	if from_module then return #syllables end -- allow other modules to know the number of syllables
	
	if #syllables == 1 then return table.concat(syllables) end -- account for monosyllables
	for i, syllable in ipairs(syllables) do
		if match(syllable, "[áéíóúý]") then -- if the user inputted manual stress, ignore all the rest
			table.insert(syllables, i, "ˈ")
			return table.concat(syllables)
		end	
		if match(syllable, consonants .. "$") or match(syllable, "ː$") then weight[i] = "h"
		elseif match(syllable, NONSYLLABIC .. "$") then weight[i] = "h"
		else weight[i] = "l"
		end
	end
	
	local a, p = weight[#weight-2], weight[#weight-1]
	if p == nil then table.insert(syllables, #syllables, "ˈ")
	elseif a == nil then table.insert(syllables, #syllables-1, "ˈ")
	elseif p == "l" and a == "h" then table.insert(syllables, #syllables-2, "ˈ")
	elseif (p == "l" and a == "l") or (p == "h") then table.insert(syllables, #syllables-1, "ˈ")
	else table.insert(syllables, #syllables-1, "ˈ") end
	
	return table.concat(syllables)
end

function export.crux(term)
	term = mw.ustring.lower(term)
	
	for _, rule in ipairs(rules) do
		term = gsub(term, rule[1], rule[2])
	end
	
	return export.write_stress(term, false)
end	

function export.postclassical_crux(term)
	local postclassicalrules = {
	{"sri(".. vowelsonly .. ")", "ɟ%1"}, {"sˈri(".. vowelsonly .. ")", "ˈɟ%1"}, {"sr", "ɟ"}, {"sˈr", "ˈɟ"}, 
	}
	
	term = export.crux(term)
	
	for _, postclassicalrule in ipairs(postclassicalrules) do
		term = gsub(term, postclassicalrule[1], postclassicalrule[2])	
	end
	
	return term
end

local function IPA_span(items)
	local bits = {}
	for _, item in ipairs(items) do
		local bit = item.pron
		table.insert(bits, bit)
	end
	return table.concat(bits)
end

local function format_IPA(items)
	return "[[Wiktionary:International Phonetic Alphabet|IPA]]<sup>([[Appendix:High_Valyrian_pronunciation|key]])</sup>:&#32;" .. IPA_span(items)
end

local function format_IPA_bare(items)
	return IPA_span(items)
end

function line_format(pronunciation, register)
	local full_pronunciations = {}
	local IPA_args = {{pron = '/' .. pronunciation .. '/'}}
	table.insert(full_pronunciations, format_IPA(IPA_args))
	return "(''" .. table.concat(register, ", ") .. "'')" .. ' ' .. table.concat(full_pronunciations, ' or ')
end

function line_format_bare(pronunciation)
	local full_pronunciations = {}
	local IPA_args = {{pron = '/' .. pronunciation .. '/'}}
	table.insert(full_pronunciations, format_IPA_bare(IPA_args))
	return table.concat(full_pronunciations, ' or ')
end

function separate_word(term, b)
	local result = {}
	
	for word in gsplit(term, " ") do
		if b then table.insert(result, export.postclassical_crux(word))
		else table.insert(result, export.crux(word)) end
	end
	
	return table.concat(result, " ")
end

function export.show(frame)
	local params = {
		[1] = { default = mw.title.getCurrentTitle().nsText == 'Template' and "kostōba" or mw.title.getCurrentTitle().text },
		nocat = { default = 0 },
	}
	local args = require("Module:parameters").process(frame:getParent().args, params)
	local term = args[1]
	
	local ipa = ""
	ipa = ipa .. line_format(separate_word(term, false), {'Classical'})
	if export.crux(term) ~= export.postclassical_crux(term) then
		ipa = ipa .. "<br/>" .. line_format(separate_word(term, true), {'Post-Classical'})
	end	
	
	ipa = gsub(ipa, "ṛ", "r̥")
	ipa = gsub(ipa, u(0x032F), "")
	ipa = gsub(ipa, u(0x035C), "")
	ipa = gsub(ipa, "([áéíóúý])", function(v) return deacuter[v] end)
	
	return ipa
end	

function export.bare(frame)
	local params = {
		[1] = { default = mw.title.getCurrentTitle().nsText == 'Template' and "kostōba" or mw.title.getCurrentTitle().text },
		nocat = { default = 0 },
	}
	local args = require("Module:parameters").process(frame:getParent().args, params)
	local term = args[1]
	
	local ipa = ""
	ipa = ipa .. line_format_bare(separate_word(term, false))
	
	ipa = gsub(ipa, "ṛ", "r̥")
	ipa = gsub(ipa, u(0x032F), "")
	ipa = gsub(ipa, u(0x035C), "")
	ipa = gsub(ipa, "([áéíóúý])", function(v) return deacuter[v] end)
	
	return ipa
end

return export