Module:haw-IPA

local export = {}

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

local decomp = mw.ustring.toNFD
local find = m_str_utils.find
local gmatch = m_str_utils.gmatch
local gsub = m_str_utils.gsub
local len = m_str_utils.len
local lower = m_str_utils.lower
local sub = m_str_utils.sub

local lang = require("Module:languages").getByCode("haw")
local ipa = require("Module:IPA")

local U = require("Module:string/char")
local breve = U(0x032F)
local macron = U(0x0304)

local cons = "bcdfghklmnpqrstvwzʻ"
local vowels = "aeiouāēīōū"
local diphs = {
	"āi", "āe", "āo", "āu", "ēi", "ōu",
	"ae", "ai", "ao", "au", "ei", "eu", "iu", "ou", "oi"
}
local respell = {
	 = "k",
	 = "p",
	 = "l",
	 = "w"
}

local stressed = "ˈˌ"
local boundary = "%.ˈˌ%+-"

-- syllabify word
local function syllabify(word)
	-- mark diphthongs with brackets
	for _, diph in ipairs(diphs) do
		word = gsub(word, diph, "{" .. diph .. "}")
	end

	-- split between vowels and consonants
	word = gsub(word, "()()", "%1.%2")

	-- remove diphthong brackets and any dots inside them
	word = gsub(word, "{*}", function(a) return gsub(a, "", "") end)

	-- syllables that end in consonants are invalid
	if find(word, "$") then
		error("One of the entries has an invalid Hawaiian syllable structure.")
	end

	return word
end

local function stress(word)
	-- decompose macrons from vowels
	word = decomp(word)

	-- stress the last vowel if it has a macron, second last otherwise
	local stress_vowel_pos = len(gsub(word, "", "")) + (find(word, macron .. "$") and 1 or 0)

	-- add stressed vowel at given position
	word = gsub(word, "()", function(v)
		stress_vowel_pos = stress_vowel_pos - 1
		return (stress_vowel_pos == 1) and 'ˈ' .. v or v
	end)

	-- shift stress mark before space or syllable boundary
	return gsub(word, "(*)ˈ", "ˈ%1")
end

-- generate pronunciation of words
function export.toIPA(entry, is_phonetic)
	if type(entry) == "table" then
		entry = table.concat(entry)
	end

	-- make entry lowercase
	entry = lower(entry)

	-- validate characters
	if not find(entry, "") then
		error("Invalid characters. Only valid Hawaiian letters, spaces and symbols '+' and '-' are accepted.")
	end

	-- syllabify entry
	entry = syllabify(entry)

	-- syllabify and stress entry
	local result = ""
	for segment in gmatch(syllabify(entry), "+?") do
		-- apply stress to each segment
		segment = stress(segment)

		-- convert to secondary stress if part of compound
		if find(segment, "") then
			segment = gsub(segment, "ˈ", "ˌ"):gsub("", ".")
		end

		result = result .. segment
	end

	-- remove syllable boundaries before stress
	result = gsub(result, "%.()", "%1")

	-- mark long vowels
	result = gsub(result, macron, "ː")

	-- mark diphthongised vowels
	result = gsub(result, "()iu", "%1i" .. breve .. "u"):gsub("()oi", "%1o" .. breve .. "i")
	result = gsub(result, "(ː?)", "%1" .. breve)

	-- replace respellt consonants
	for k, v in pairs(respell) do
		result = gsub(result, "", v)
	end

	-- real value of glottal stop
	result = gsub(result, "ʻ", "ʔ")

	-- phonetic transcriptions
	if is_phonetic then
		-- allophony of /k/ before /i/
		result = gsub(result, "ki", "ti")

		-- real pronunciation of /w/
		result = gsub(result, "w", "ʋ")

		-- allophony of glides /j/, /w/
		result = gsub(result, "(" .. breve .. "?)()", "%1j%2")
		result = gsub(result, "(" .. breve .. "?)()", "%1w%2")

		-- actual pronunciation of diphthongs
		result = gsub(result, "i" .. breve, "j")
		result = gsub(result, "u" .. breve, "w")

		-- allophony of /a/
		result = gsub(result, "(?)a", "%1ɐ")
		result = gsub(result, "a", "ə")
		result = gsub(result, "ː", "aː")

		-- allophony of /e/
		result = gsub(result, "(?)e", "%1ɛ")
		result = gsub(result, "e(*)ɛ", "ɛ%1ɛ")
		result = gsub(result, "ɛ(*)e", "ɛ%1ɛ")
		result = gsub(result, "ɛː", "eː")
	end

	return result
end

-- rapid pronunciation of diphthongs
function export.rapid(entry)
	return gsub(entry, "ɐj", "ɛj"):gsub("ɐw", "ɔw")
end

-- display pronunciation
function export.show(frame)
	local args = frame:getParent().args
	local pagetitle = mw.title.getCurrentTitle().text

	local p, results = {}, {}

	if args then
		for _, v in ipairs(args) do
			table.insert(p, (v ~= "") and v or nil)
		end
	else
		p = { pagetitle }
	end

	for _, word in ipairs(p) do
		local phonemic = export.toIPA(word, false)
		local phonetic = export.toIPA(word, true)
		local rapid = export.rapid(phonetic)

		table.insert(results, { pron = "/" .. phonemic .. "/" })
		if phonemic ~= phonetic then table.insert(results, { pron = "" }) end
		if phonetic ~= rapid then table.insert(results, { pron = "", qq = {"rapid speech"} }) end
	end

	return ipa.format_IPA_full { lang = lang, items = results }
end

return export