local gsub = mw.ustring.gsub
local lower = mw.ustring.lower
local toNFC = mw.ustring.toNFC
local toNFD = mw.ustring.toNFD
local u = require("Module:string/char")
local export = {}
local m_IPA = require("Module:IPA")
local lang = require("Module:languages").getByCode("yo")
-- apply rsub() repeatedly until no change
local function gsub_repeatedly(term, foo, bar)
while true do
local new_term = gsub(term, foo, bar)
if new_term == term then
return term
end
term = new_term
end
end
local MACRON = u(0x0304) -- macron = ̄
local DOT = u(0x0323) -- dot = ̣
local NASAL = u(0x0303) -- nasal = ̃
local consonants = {
["b"] = "b",
["d"] = "d",
["f"] = "f",
["g"] = "ɡ",
["gb"] = "ɡ͡b",
["h"] = "h",
["j"] = "d͡ʒ",
["k"] = "k",
["l"] = "l",
["m"] = "m",
["n"] = "n",
["p"] = "k͡p",
["r"] = "ɾ",
["s"] = "s",
["ṣ"] = "ʃ",
["t"] = "t",
["w"] = "w",
["y"] = "j",
["gh"] = "ɣ",
["gw"] = "ɡʷ",
}
local vowels = {
["a"] = "a",
["e"] = "e",
["ẹ"] = "ɛ",
["i"] = "i",
["o"] = "o",
["ọ"] = "ɔ",
["u"] = "u",
["ị"] = "ɪ",
["ụ"] = "ʊ",
}
local nasals = {
["an"] = "ã",
["ẹn"] = "ɛ̃",
["in"] = "ĩ",
["ọn"] = "ɔ̃",
["un"] = "ũ",
["m"] = "m",
["n"] = "ŋ",
}
function export.IPA(text, loanword, downstep)
text = lower(text)
text = gsub(text, "%-", "z")
text = toNFD(text)
text = gsub(text, ".[" .. DOT .. "]", {
["e" .. DOT] = "ẹ",
["o" .. DOT] = "ọ",
["s" .. DOT] = "ṣ",
["i" .. DOT] = "ị",
["u" .. DOT] = "ụ",
})
text = text .. "x"
text = gsub(text, "%s+", " ")
text = gsub(text, "^ ", "")
text = gsub(text, " $", "")
text = gsub(text, "\n", "")
text = gsub(text, "'", "")
text = gsub(text, "ẹ", "ɛ")
text = gsub(text, "gb", "ɡ͡b")
text = gsub(text, "gh", "ɣ")
text = gsub(text, "gw", "ɡʷ")
text = gsub(text, "g", "ɡ")
text = gsub(text, "ị", "ɪ")
text = gsub(text, "j", "d͡ʒ")
text = gsub(text, "ọ", "ɔ")
text = gsub(text, "p", "k͡p")
text = gsub(text, "r", "ɾ")
text = gsub(text, "ṣ", "ʃ")
text = gsub(text, "ụ", "ʊ")
text = gsub(text, "y", "j")
text = gsub(text, "m([́̀̄])", "ŋ" .. "%1")
text = gsub(text, "n([́̀̄])", "ŋ" .. "%1")
text = gsub_repeatedly(text, "([aeɛiɪoɔuʊ])([aeɛiɪoɔuʊnbdfɡhjklmnŋɾsʃtwxzɣ ])", "%1" .. MACRON .. "%2")
text = gsub(text, "([aɛiɪɔuʊ])([́̀̄])n([bdfɡhjklmnŋɾsʃtwxzɣ ])", "%1" .. NASAL .. "%2%3")
--TODO: FIX LOANWORDS
if not loanword then
text = gsub(text, "n([aiɪuʊ])([́̀̄])", "n%1" .. NASAL .. "%2")
end
if not loanword then
text = gsub(text, "m([iɪɔuʊ])([́̀̄])", "m%1" .. NASAL .. "%2")
end
text = gsub_repeatedly(text, "([aɛiɪɔuʊ])([̃])([́̀̄])([aɛiɔu])([́̀̄])", "%1%2%3%4" .. NASAL .. "%5")
text = gsub(text, "([aeɛiɪoɔuʊ])([́̀̄])", "%1" .. "%2")
text = gsub(text, "a([̃])([́̀̄])za","a" .. "%1" .. "%2a")
text = gsub(text, "ɛ([̃])([́̀̄])zɛ","ɛ" .. "%1" .. "%2ɛ")
text = gsub(text, "i([̃])([́̀̄])zi","i" .. "%1" .. "%2i")
text = gsub(text, "ɪ([̃])([́̀̄])zɪ","ɪ" .. "%1" .. "%2ɪ")
text = gsub(text, "ɔ([̃])([́̀̄])zɔ","ɔ" .. "%1" .. "%2ɔ")
text = gsub(text, "u([̃])([́̀̄])zu","u" .. "%1" .. "%2u")
text = gsub(text, "ʊ([̃])([́̀̄])zʊ","ʊ" .. "%1" .. "%2ʊ")
text = gsub(text, "z", ".")
-- Split into syllables
text = gsub(text, "a([́̀̄])([eɛiɪoɔuʊbdfɡhjklmnŋɾsʃtwɣ])","a" .. "%1.%2")
text = gsub(text, "a([̃])([́̀̄])([eɛiɪoɔuʊbdfɡhjklmnŋɾsʃtwɣ])","a" .. "%1" .. "%2.%3")
text = gsub(text, "e([́̀̄])([aɛiɪoɔuʊbdfɡhjklmnŋɾsʃtwɣ])","e" .. "%1.%2")
text = gsub(text, "ɛ([́̀̄])([aeiɪoɔuʊbdfɡhjklmnŋɾsʃtwɣ])","ɛ" .. "%1.%2")
text = gsub(text, "ɛ([̃])([́̀̄])([aeiɪoɔuʊbdfɡhjklmnŋɾsʃtwɣ])","ɛ" .. "%1" .. "%2.%3")
text = gsub(text, "i([́̀̄])([aeɛɪɔuʊbdfɡhjklmnŋɾsʃtwɣ])","i" .. "%1.%2")
text = gsub(text, "i([̃])([́̀̄])([aeɛɪoɔuʊbdfɡhjklmnŋɾsʃtwɣ])","i" .. "%1" .. "%2.%3")
text = gsub(text, "ɪ([́̀̄])([aeɛioɔuʊbdfɡhjklmnŋɾsʃtwɣ])","ɪ" .. "%1.%2")
text = gsub(text, "ɪ([̃])([́̀̄])([aeɛioɔuʊbdfɡhjklmnŋɾsʃtwɣ])","ɪ" .. "%1" .. "%2.%3")
text = gsub(text, "o([́̀̄])([aeɛiɪɔuʊbdfɡhjklmnŋɾsʃtwɣ])","o" .. "%1.%2")
text = gsub(text, "ɔ([́̀̄])([aeɛiɪouʊbdfɡhjklmnŋɾsʃtwɣ])","ɔ" .. "%1.%2")
text = gsub(text, "ɔ([̃])([́̀̄])([aeɛiɪouʊbdfɡhjklmnŋɾsʃtwɣ])","ɔ" .. "%1" .. "%2.%3")
text = gsub(text, "u([́̀̄])([aeɛiɪɔoʊbdfɡhjklmnŋɾsʃtwɣ])","u" .. "%1.%2")
text = gsub(text, "u([̃])([́̀̄])([aeɛiɪɔoʊbdfɡhjklmnŋɾsʃtwɣ])","u" .. "%1" .. "%2.%3")
text = gsub(text, "ʊ([́̀̄])([aeɛiɪɔobdfɡhjklmnŋɾsʃtwɣ])","ʊ" .. "%1.%2")
text = gsub(text, "ʊ([̃])([́̀̄])([aeɛiɪoɔubdfɡhjklmnŋɾsʃtwɣ])","ʊ" .. "%1" .. "%2.%3")
text = gsub_repeatedly(text, "m([́̀̄])([aeɛiɪoɔuʊbdfɡhjklnŋɾsʃtwɣ])","m" .. "%1.%2")
text = gsub_repeatedly(text, "ŋ([́̀̄])([aeɛiɪoɔuʊbdfɡhjklmnɾsʃtwɣ])","ŋ" .. "%1.%2")
text = gsub(text, "x", "")
-- Adds downstep ꜜ before provided syllable
if downstep then
downstep = tonumber(downstep)
local i = 0
local result = (text):gsub("[^.]+", function(syllable)
i = i + 1
if i == downstep then
return "ꜜ" .. syllable
end -- else leave it alone
end)
text = result
end
return toNFC(text)
end
function export.show(frame)
local args = frame:getParent().args
local p, results = {}, {}
if args[1] then
for _, item in ipairs(args) do
table.insert(p, (item ~= "") and item or nil)
end
else
error("Please provide a tone marked term.")
end
for _, text in ipairs(p) do
if args["loan"] == "1" then
table.insert(results, {pron = "/" .. export.IPA(text, true, args["downstep"]) .. "/", note = nil})
else
table.insert(results, {pron = "/" .. export.IPA(text, false, args["downstep"]) .. "/", note = nil})
end
end
return m_IPA.format_IPA_full { lang = lang, items = results }
end
return export
