local export = {}
local consonants = {
['ಕ']='k', ['ಖ']='kʰ', ['ಗ'] = 'ɡ', ['ಘ'] ='ɡʱ', ['ಙ']='ŋ',
['ಚ']='t͡ʃ', ['ಛ']='t͡ʃʰ', ['ಜ']='d͡ʒ', ['ಝ']='d͡ʒʱ', ['ಞ']='ɲ',
['ಟ']='ʈ', ['ಠ']='ʈʰ', ['ಡ']='ɖ', ['ಢ']='ɖʱ', ['ಣ']='ɳ',
['ತ']='t̪', ['ಥ']='t̪ʰ', ['ದ']='d̪', ['ಧ']='d̪ʱ', ['ನ']='n̪',
['ಪ']='p', ['ಫ']='pʰ', ['ಬ']='b', ['ಭ']='bʱ', ['ಮ']='m',
['ಯ']='j', ['ರ']='ɾ', ['ಲ']='l', ['ವ']='ʋ',
['ಶ']='ʃ', ['ಷ']='ʂ', ['ಸ']='s', ['ಹ']='h',
['ಳ']='ɭ', ['ೞ']='ɻ', ['ಱ']='r',
}
local vowel_diacritics = {
['ಾ']= 'aː', ['ಿ']='i', ['ೀ']='iː', ['ು']='u', ['ೂ']='uː', ['ೃ'] = 'ɾi', ['ೄ'] = 'ɾiː', ['ೢ'] = 'li', ['ೣ'] = 'liː',
['ೆ']='e', ['ೇ']='eː', ['ೈ']='ɐi̯', ['ೊ']='o', ['ೋ']='oː', ['ೌ']='ɐu̯',
['್']='', -- Virama - suppresses the inherent vowel "a"
[''] = 'ɐ' -- No diacritic; inherent vowel
}
local other = {
-- independent vowels
['ಅ']='ɐ', ['ಆ']='aː', ['ಇ']='i', ['ಈ']='iː', ['ಉ']='u', ['ಊ']='uː', ['ಋ'] = 'ɾi', ['ೠ'] = 'ɾiː', ['ಌ'] = 'li', ['ೡ'] = 'liː',
['ಎ']='e', ['ಏ']='eː', ['ಐ']='ɐi̯', ['ಒ']='o', ['ಓ']='oː', ['ಔ']='ɐu̯',
-- Other symbols
['ಂ']='m̃', ['ಃ']='h',
['ೝ']='n',
['ಽ']='',
}
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$']='m',
['e2']='ɛ',
}
local adjust2 = {
-- Account for differences in phonemes vs. phones
}
local function shift_to_codas(syllables)
-- shift codas to previous syllable using the Weerasinghe-Wasala-Gamage method
local to_move = 0
for i, syll in ipairs(syllables) do
if i == 1 then
-- no need to shift to coda if in the first syllable
elseif #syll < 3 then
-- coda movement only needed for onset clusters of 2 or more
elseif #syll == 3 then
-- V.CCV => VC.CV
to_move = 1
elseif #syll == 4 then
if syll[#syll - 1] == "ɾ" or syll[#syll - 1] == "j" or (stop_list[syll[1]] and stop_list[syll[2]]) then
-- V.CCrV or V.CCyV => VC.CrV or VC.CyV
-- if the first two consonants are stops, VC.CCV
to_move = 1
else
-- V.CCCV => VCC.CV
to_move = 2
end
else
-- 4 consonants or more
if syll[#syll - 1] == "ɾ" or syll[#syll - 1] == "j" then
to_move = #syll - 3
else
-- find index of consonant of least sonority
to_move = #syll - 1
local min_son = consonant_sonority[syll[#syll - 1]]
for i = (#syll - 1), 1, -1 do
if consonant_sonority[syll[i]] < min_son then
to_move = i
min_son = consonant_sonority[syll[i]]
end
end
end
end
while to_move > 0 do
table.insert(syllables[i - 1], table.remove(syllables[i], 1))
to_move = to_move - 1
end
end
return syllables
end
local function syllabify(remainder, accent)
local syllables = {}
local syll = {}
while #remainder > 0 do
local phoneme = table.remove(remainder, 1)
if vowel_list[phoneme] then
table.insert(syll, phoneme)
table.insert(syllables, syll)
syll = {}
else
table.insert(syll, phoneme)
end
end
-- store whatever consonants remain
local final_cons = syll
-- Vedic pitch accent
if accent ~= nil and accent <= #syllables then
syll = syllables[accent]
syllables[accent][#syll] = accent_vowel[syll[#syll]]
end
syllables = shift_to_codas(syllables)
local short_vowel_patt = "^[ɐiurl]" .. SYLLABIC .. "?" .. ACUTE .. "?$"
-- Classic stress accent
local num_sylls = #syllables
if num_sylls == 2 then
table.insert(syllables[1], 1, 'ˈ')
elseif num_sylls == 3 then
-- if the final segment of the second syllable is not a short vowel, stress the second syllable
if mw.ustring.match(syllables[2][#syllables[2]], short_vowel_patt) == nil then
table.insert(syllables[2], 1, 'ˈ')
-- else stress the third
else
table.insert(syllables[1], 1, 'ˈ')
end
elseif num_sylls >= 4 then
if mw.ustring.match(syllables[num_sylls - 1][#syllables[num_sylls - 1]], short_vowel_patt) == nil then
table.insert(syllables[num_sylls - 1], 1, 'ˈ')
elseif mw.ustring.match(syllables[num_sylls - 2][#syllables[num_sylls - 2]], short_vowel_patt) == nil then
table.insert(syllables[num_sylls - 2], 1, 'ˈ')
else
table.insert(syllables[num_sylls - 3], 1, 'ˈ')
end
end
-- If there are phonemes left, then the word ends in a consonant
-- Add them to the last syllable
for _, phoneme in ipairs(final_cons) do
table.insert(syllables[#syllables], phoneme)
end
for i, _ in ipairs(syllables) do
syllables[i] = table.concat(syllables[i], "")
end
return table.concat(syllables, ".")
end
function export.to_IPA(text)
local VIRAMA = '್' -- final virama rules
text = mw.ustring.gsub(text, VIRAMA .. "([%,%.%!%?%:%;]?)$", VIRAMA .. "ɯ%1")
text = mw.ustring.gsub(text, VIRAMA .. "([%,%.%!%?%:%;]?) ", VIRAMA .. "ɯ%1 ")
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('tcy'),
{ { pron = '/' .. transcriptions[1] .. '/' } })
else
IPA_text = require('Module:IPA').format_IPA_full(
require('Module:languages').getByCode('tcy'),
{ { pron = '/' .. transcriptions[1] .. '/' }, { pron = '[' .. transcriptions[2] .. ']' } })
end
return '* ' .. (qualifier and require("Module:qualifier").format_qualifier{qualifier} .. ' ' or '')
.. IPA_text
end
return export
