Module:bfq-IPA

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

local export = {}

local consonants = {
	['க​']='k', ['க​']='kʰ', ['க​'] = 'ɡ', ['க​'] ='ɡʱ', ['ங்']='ŋ',
	['ச​']='t͡ʃ', ['ச​']='t͡ʃʰ', ['ச​']='d͡ʒ', ['ச​']='d͡ʒʱ', ['ஞ​']='ɲ',
	['ட']='ʈ', ['ட']='ʈʰ', ['ட']='ɖ', ['ட']='ɖʱ', ['ண​']='ɳ',
	['த']='t̪', ['த']='t̪ʰ', ['த']='d̪', ['த']='d̪ʱ', ['ந​']='n̪', 
	['ப​']='p', ['ப​']='pʰ', ['ப​']='b', ['ப​']='bʱ',  ['ம']='m',
	['ஜ​']='j', ['ர​']='ɾ', ['ல​']='l', ['வ​']='ʋ', 
	['ஶ']='ʃ', ['ஷ']='ʂ', ['ஸ​']='s', ['ஹ​']='h', 
	['ள​']='ɭ', ['ழ']='ɻ', ['ற']='r', 
}

local vowel_diacritics = {
	['ா']= 'aː', ['ி']='i', ['ீ']='iː', ['ு']='u', ['ூ']='uː', 
	['ெ']='e', ['ே']='eː', ['ை']='ɐi̯', ['ொ']='o', ['ோ']='oː', ['ௌ']='ɐu̯',
	['்']='',	-- Virama - suppresses the inherent vowel "a"
	[''] = 'ɐ'	-- No diacritic; inherent vowel	
}

local other = {
	-- independent vowels
	['அ']='ɐ', ['ஆ']='aː', ['இ']='i', ['ஈ']='iː', ['உ']='u', ['ஊ']='uː',
	['எ']='e', ['ஏ']='eː', ['ஐ']='ɐi̯', ['ஒ']='o', ['ஓ']='oː', ['ஔ']='ɐu̯',
	-- Other symbols
	['ஃ']='',

}

local adjust1 = {
	-- Assimilate the anusvara
	['m̃([kɡŋ])']='ŋ%1',
	['m̃([td]͡[ʃʒ])']='ɲ%1', ['m̃(ɲ)']='ɲ%1',
	['m̃([ʈɖɳ])']='ɳ%1',
	['m̃([td]̪)']='n̪%1', ['m̃([td]͡[sz])']='n̪%1', ['m̃(n̪)']='n̪%1',
	['m̃([pbmjɾlʋʃʂshɭ])']='m%1',
	['m̃([%s%p])']='m%1', ['m̃$']='m'
}

local adjust2 = {
	-- Account for differences in phonemes vs. phones
}

function export.to_IPA(text)

	text = mw.ustring.gsub(
		text,
		'([ಕ-ಹ])(಼?)([ಾ-್]?)',
		function(c, n, d)
			return ((consonants[c..n] or consonants[c]) or c) .. vowel_diacritics[d]
		end)

	text = mw.ustring.gsub(text, '[ಂ-ೡ]', other)

	for k, v in pairs(adjust1) do
		text = mw.ustring.gsub(text, k, v)
	end

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

	-- Phonetic transcription
	text2 = text
	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(
			require('Module:languages').getByCode('bfq'),
			{ { pron = '/' .. transcriptions[1] .. '/' } })
	else
		IPA_text = require('Module:IPA').format_IPA_full(
			require('Module:languages').getByCode('bfq'),
			{ { pron = '/' .. transcriptions[1] .. '/' }, { pron = '[' .. transcriptions[2] .. ']' } })
	end

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

end

return export