Modul:hi-IPA

local export = {}

local lang = require("Module:languages").getByCode("hi")
local sc = require("Module:scripts").getByCode("Deva")
local m_IPA = require("Module:IPA")

local gsub = mw.ustring.gsub
local gmatch = mw.ustring.gmatch
local find = mw.ustring.find

local correspondences = {
	 = "ŋ",  = "ɡ", 
	 = "t͡ʃ",  = "d͡ʒ",  = "ɲ",
	 = "ʈ",  = "ɖ",  = "n",
	 = "t̪",  = "d̪",
	 = "j",  = "ɾ",  = "ʋ",  = "l̪",
	 = "ʃ",  = "ʃ",  = "ɦ",
	 = "ɽ",  = "ʒ",  = "ɭ",  = "ɣ",  = "q",  = "x",  = "n",  = "ɹ",

	 = "ə",  = "ɑː",  = "ɪ",
	 = "iː",  = "oː",  = "eː",
	 = "ʊ",  = "uː",  = "ɔː",  = "æː",

	 = "ʊ̃",  = "õː",  = "ə̃",  = "ɑ̃ː", 

	 = "oːm",  = "ʰ",  = "(ʔ)",
}

local perso_arabic = {
	 = "kh",  = "g",  = "k",  = "z",  = "",
}

local lengthen = {
	 = "ā",  = "ī",  = "ū",
}

local vowels = "aāiīuūoŏĕʊɪɔɔ̃ɛeæãā̃ẽĩī̃õũū̃ː"
local vowel = "ː?"
local weak_h = "()h"
local aspirate = "()"
local syllabify_pattern = "(̃?)(+)(̃?)"

local function find_consonants(text)
	local current = ""
	local cons = {}
	for cc in mw.ustring.gcodepoint(text .. " ") do
		local ch = mw.ustring.char(cc)
		if find(current .. ch, "^$") or find(current .. ch, "^h$") then
			current = current .. ch
		else
			table.insert(cons, current)
			current = ch
		end
	end
	return cons
end

local function syllabify(text)
	for count = 1, 2 do
		text = gsub(text, syllabify_pattern, function(a, b, c)
			b_set = find_consonants(b)
			table.insert(b_set, #b_set > 1 and #b_set or 1, ".")
			return a .. table.concat(b_set) .. c
			end)
		text = gsub(text, "(" .. vowel .. ")(?=" .. vowel .. ")", "%1.")
	end
	-- text = gsub(text, "(" .. vowel .. ")(" .. vowel .. ")", "%1.%2")
	return text
end

local identical = "knlsfzθ"
for character in gmatch(identical, ".") do
	correspondences = character
end

local function transliterate(text)
	return lang:transliterate(text)
end

function export.link(term)
	return require("Module:links").full_link{ term = term, lang = lang, sc = sc }
end

function export.toIPA(text, style)
	local translit = transliterate(text)
	if not translit then
		error('The term "' .. Hindi .. '" could not be transliterated.')
	end
	
	if style == "standard" then
		translit = gsub(translit, "", perso_arabic)
	end
	
	-- force final schwa
	translit = gsub(translit, "a~$", "ə")

	-- remove final schwa (Pandey, 2014)
	translit = gsub(translit, "a$", "")
	translit = gsub(translit, "(...)a ", "%1 ")
	translit = gsub(translit, "(...)a%-", "%1-")
	
	-- vowels
	translit = gsub(translit, "͠", "̃")
	translit = gsub(translit, "a()()", function(a, b)
		return "a" .. lengthen .. b
	end)
	translit = gsub(translit, 'a(̃?)i', 'ɛ%1ː')
	translit = gsub(translit, 'a(̃?)u', 'ɔ%1ː')
	translit = gsub(translit, "%-$", "")
	translit = gsub(translit, "^%-", "")
	translit = gsub(translit, "%-", ".")
	translit = gsub(translit, "ŕ$", "r")
	translit = gsub(translit, "ŕ", "ri")
	translit = gsub(translit, ",", "")
	translit = gsub(translit, " ", "..")
	
	translit = syllabify(translit)
	
	-- gy
	translit = gsub(translit, 'jñ', 'gy')
	
	-- aspiration rules
	translit = gsub(translit, 'ah%.(h?)', 'ɛːʱ.%1')
	translit = gsub(translit, 'a%.ha', 'ɛːʱ')
	translit = gsub(translit, aspirate .. "h", '%1ʰ')
	translit = gsub(translit, weak_h, '%1ʱ')
	
	translit = gsub(translit, "%.ː", "ː.")
	
	local result = gsub(translit, ".", correspondences)

	result = gsub(result, "%.%.", "‿")
	
	-- formatting
	result = gsub(result, "ː̃", "̃ː")
	result = gsub(result, "ː.̃", "̃ː.")
	result = gsub(result, "%.$", "")

	-- i and u lengthening
	result = gsub(result, "ʊ(̃?)(ʱ?)$", "u%1ː%2")
	result = gsub(result, "ɪ(̃?)(ʱ?)$", "i%1ː%2")
	result = gsub(result, "ɪ%.j", "iː.j")
	
	return result
end

function export.make(frame)
	local args = frame:getParent().args
	local pagetitle = mw.title.getCurrentTitle().text
	
	local p, results = {}, {}, {}
	
	if args then
		for index, item in ipairs(args) do
			table.insert(p, (item ~= "") and item or nil)
		end
	else
		p = { pagetitle }
	end
	
	for _, Hindi in ipairs(p) do
		table.insert(results, { pron = "/" .. export.toIPA(Hindi, "standard") .. "/" })
		if export.toIPA(Hindi, "standard") ~= export.toIPA(Hindi, "persianized") then
			table.insert(results, { pron = "/" .. export.toIPA(Hindi, "persianized") .. "/" })
		end
	end
	
	return  m_IPA.format_IPA_full(lang, results)
end

function export.make_ur(frame)
	local args = frame:getParent().args
	local pagetitle = mw.title.getCurrentTitle().text
	local lang = require("Module:languages").getByCode("ur")
	local sc = require("Module:scripts").getByCode("ur-Arab")
	
	local p, results = {}, {}, {}
	
	if args then
		for index, item in ipairs(args) do
			table.insert(p, (item ~= "") and item or nil)
		end
	else
		error("No transliterations given.")
	end
	
	for _, Urdu in ipairs(p) do
		table.insert(results, { pron = "/" .. export.toIPA(Urdu, "persianized") .. "/" })
	end
	
	return  m_IPA.format_IPA_full(lang, results)
end

return export