Module:ta-IPA

From Wiktionary, the free dictionary
Jump to navigation Jump to search


local export = {}

local consonants = {
	['க']='k', ['ங']='ŋ', ['ச']='t͡ɕ', ['ஞ']='ɲ', ['ட']='ʈ', ['ண']='ɳ', ['த']='t̪',
	['ந']='n̪', ['ப']='p', ['ம']='m', ['ய']='j', ['ர']='ɾ', ['ல']='l', ['வ']='ʋ',
	['ழ']='ɻ', ['ள']='ɭ', ['ற']='r', ['ன']='n', ['ஶ']='ɕ', ['ஜ']='d͡ʑ', ['ஷ']='ʂ',
	['ஸ']='s', ['ஹ']='h', ['ஃப']='f', ['ஃஜ']='z', ['ஃக']='x',
	['ஃ']='',
    ['க3']='ɡ', ['ட3']='ɖ', ['ப3']='b', ['த3']='d̪', ['ச1']='t͡ɕ', ['ச5']='s'
}

local numbers_sa = {
	['1']='1', ['2']='2', ['3']='3', ['4']='4', 
	['¹']='1', ['²']='2', ['³']='3', ['⁴']='4',
	['₁']='1', ['₂']='2', ['₃']='3', ['₄']='4'
}

local diacritics = {
	['ா']= 'aː', ['ி']='ɪ', ['ீ']='iː', ['ு']='ʊ', ['ூ']='uː',  ['ெ']='ɛ',
	['ே']='eː', ['ை']='ɐɪ̯', ['ொ']='ɔ', ['ோ']='oː', ['ௌ']='ɐʊ̯',
	['்']='',	--halant, supresses the inherent vowel "a"
	-- no diacritic
	[''] = 'ɐ'
}

local nonconsonants = {
	-- independent vowels
	['அ']='ɐ', ['ஆ']='aː', ['இ']='ɪ', ['ஈ']='iː', ['உ']='ʊ', ['ஊ']='uː',
	['எ']='ɛ', ['ஏ']='eː', ['ஐ']='ɐɪ̯', ['ஒ']='ɔ', ['ஓ']='oː', ['ஔ']='ɐʊ̯',
	-- other symbols
	['ஃ']='',
}

local adjust1 = {
	['ŋk']='ŋɡ', ['ɳʈ']='ɳɖ', ['n̪t̪']='n̪d̪', ['mp']='mb',	['ɲt͡ɕ']='ɲd͡ʑ',
	['([aɐeiouɛɪɔʊɯi̯u̯jɾlɭɻnɳ]ː?)k([aɐeioɛɪɔʊuɯi̯u̯])']='%1ɡ%2',
	['([aɐeiouɛɪɔʊɯi̯u̯jɾlɭɻnɳ]ː?)ʈ([aɐeioɛɪɔʊɯui̯u̯])']='%1ɖ%2',
	['([aɐeiouɛɪɔʊɯi̯u̯jɾlɭɻnɳ]ː?)t̪([aɐeioɛɪɔʊɯui̯u̯])']='%1d̪%2',
	['([aɐeiouɛɪɔʊɯi̯u̯jɾlɭɻnɳ]ː?)p([aɐeioɛɪɔʊɯui̯u̯])']='%1b%2'
}

local adjust2 = {
	['t͡ɕ'] = 's', ['sː'] = 't͡ɕː',
	['ʊ([%s%p])']='ɯ%1', ['ʊ$']='ɯ', ['ɪ([%s%p])']='i%1', ['ɪ$']='i', ['$t͡ɕ']='s', ['([%s%p])t͡ɕ([aɐeiouɯi̯u̯])']='%1s',
	['ɲs']='ɲd͡ʑ', ['rr']='tr', ['rː']='tr', ['nr']='ndr',
	['t͡ɕ([aɐeioɯuɛɪɔʊi̯u̯])']='s%1',
	['([aɐeioɯuɛɪɔʊi̯u̯]ː?)t͡ɕ([aɐeiouɛɪɔʊɯi̯u̯])']='%1s%2',
}

function export.to_IPA(text)

	text = mw.ustring.gsub(
		text,
		'(ஃ?)([க-ஹ])([ா-்]?)([1234¹²³⁴₁₂₃₄]?)',
		function(h, c, d, n)
			n = numbers_sa[n] or n
			return ((consonants[h..c..n] or consonants[h..c]) or (consonants[h] .. (consonants[c..n] or consonant[c] or c))) .. diacritics[d]
		end)

	text = mw.ustring.gsub(text, '[அ-ஔ]', nonconsonants)

	for k, v in pairs(adjust1) do
		local count = 1
		while count <= 2 do --twice
		text = mw.ustring.gsub(text, k, v)
		count = count + 1
		end
	end

	--convert consonant gemination to triangular colon
	text = mw.ustring.gsub(text, "([kŋɲʈɳpmjʋɻɭrnɕʂshfzx])%1", "%1ː")
	text = mw.ustring.gsub(text, "([tnɾl]̪)%1", "%1ː")
	text = mw.ustring.gsub(text, "([td]͡[ɕʑ])%1", "%1ː")

	--if an independent vowel is after another vowel, assume diphthong
	text = mw.ustring.gsub(text, "([aeiouɛɪɔʊ]ː?)•", "%1")

	text2 = text --phonetic

	for k, v in pairs(adjust2) do
		text2 = mw.ustring.gsub(text2, k, v)
	end

	return (text == text2 and { text } or { text, text2 })

end

function export.show(frame)

	local args = frame:getParent().args
	local page_title = mw.title.getCurrentTitle().text
	local text = args[1] or page_title
	local qualifier = args["q"] or nil

	local transcriptions = export.to_IPA(text)
	local IPA_text
	if not transcriptions[2] then
		IPA_text = require("Module:IPA").format_IPA_full {
			lang = require("Module:languages").getByCode("ta"),
			items = {{ pron = "/" .. transcriptions[1] .. "/" }},
		}
	else
		IPA_text = require("Module:IPA").format_IPA_full {
			lang = require("Module:languages").getByCode("ta"),
			items = {{ pron = "/" .. transcriptions[1] .. "/" }, { pron = "[" .. transcriptions[2] .. "]" }},
		}
	end

	return "* " .. (qualifier and require("Module:qualifier").format_qualifier{qualifier} .. " " or "")
		.. IPA_text

end

return export