Module:lfv-IPA
Jump to navigation
Jump to search
local export = {}
local getArgs = require('Module:Arguments').getArgs
-- DATA --
local data = {
["a"] = "a",
["e"] = "e",
["ê"] = "ɛ",
["i"] = "i",
["o"] = "o",
["ô"] = "ɔ",
["ø"] = "ø",
["u"] = "u",
["y"] = "y",
["b"] = "b",
["d"] = "d",
["ð"] = "ð",
["f"] = "f",
["g"] = "ɡ",
["ǧ"] = "ɣ",
["h"] = "x",
["j"] = "j",
["k"] = "k",
["l"] = "l",
["lj"] = "ʎ",
["m"] = "m",
["n"] = "n",
["nj"] = "ɲ",
["ñ"] = "ŋ",
["p"] = "p",
["r"] = "r",
["s"] = "s",
["š"] = "ʃ",
["t"] = "t",
["þ"] = "θ",
["v"] = "v",
["w"] = "w",
["ŵ"] = "ɥ",
["z"] = "z",
["ž"] = "ʒ",
["hl"] = "l̥",
["hlj"] = "ʎ̥",
["hm"] = "m̥",
["hn"] = "n̥",
["hnj"] = "ɲ̊",
["hñ"] = "ŋ̊",
["hr"] = "r̥",
["llj"] = "ʎʎ",
["nnj"] = "ɲɲ",
["lhl"] = "l̥l̥",
["lhlj"] = "ʎʎ",
["mhm"] = "mm",
["nhn"] = "nn",
["nhnj"] = "ɲɲ",
["ñhñ"] = "ŋŋ",
["rhr"] = "rr",
[" "] = " ",
["."] = "|",
}
data[","] = data["."]
local vowel = {
["a"] = true,
["e"] = true,
["ɛ"] = true,
["i"] = true,
["o"] = true,
["ɔ"] = true,
["ø"] = true,
["u"] = "u",
["y"] = true,
}
local consonant = {
["b"] = true,
["d"] = true,
["ð"] = true,
["f"] = true,
["ɡ"] = true,
["ɣ"] = true,
["x"] = true,
["j"] = true,
["k"] = true,
["l"] = true,
["ʎ"] = true,
["m"] = true,
["n"] = true,
["ɲ"] = true,
["ŋ"] = true,
["p"] = true,
["r"] = true,
["s"] = true,
["ʃ"] = true,
["t"] = true,
["θ"] = true,
["v"] = true,
["w"] = true,
["ɥ"] = true,
["z"] = true,
["ʒ"] = true,
}
local boundary = {
[" "] = true,
["|"] = true,
}
local intervocalic_voicing = {
["ʎ̥"] = "ʎʎ",
["m̥"] = "mm",
["n̥"] = "nn",
["ɲ̊"] = "ɲɲ",
["ŋ̊"] = "ŋŋ",
["r̥"] = "rr",
}
local devoice = {
["d"] = "d̥",
["ð"] = "ð̥",
["g"] = "g̊",
["ǧ"] = "ɣ̊",
["l"] = "l̥",
["lj"] = "ʎ̥",
["r"] = "r̥",
["z"] = "z̥",
["ž"] = "ʒ̊",
}
local voiceless = {
["f"] = true,
["x"] = true,
["k"] = true,
["p"] = true,
["s"] = true,
["ʃ"] = true,
["t"] = true,
["θ"] = true,
}
function generate_IPA(word, phon, col)
local working_IPA = {}
-- base generation --
while #word > 0 do
if data[mw.ustring.sub(word, 1, 4)] then
table.insert(working_IPA, data[mw.ustring.sub(word, 1, 4)])
word = mw.ustring.sub(word, 5)
elseif data[mw.ustring.sub(word, 1, 3)] then
table.insert(working_IPA, data[mw.ustring.sub(word, 1, 3)])
word = mw.ustring.sub(word, 4)
elseif data[mw.ustring.sub(word, 1, 2)] then
table.insert(working_IPA, data[mw.ustring.sub(word, 1, 2)])
word = mw.ustring.sub(word, 3)
elseif data[mw.ustring.sub(word, 1, 1)] then
table.insert(working_IPA, data[mw.ustring.sub(word, 1, 1)])
word = mw.ustring.sub(word, 2)
elseif data[mw.ustring.sub(word, 1, 1)] == "·" then
word = mw.ustring.sub(word, 2)
else
error("The character '" .. mw.ustring.sub(word, 1, 1) .. "' is not recognised.")
end
end
-- resolve consonants --
local i = 1
while working_IPA[i] do
local p_prev = working_IPA[i-1]
local p_current = working_IPA[i]
local p_next = working_IPA[i+1]
local function p_resolve(phone)
working_IPA[i] = phone
p_prev = working_IPA[i-1]
p_current = working_IPA[i]
p_next = working_IPA[i+1]
end
local function remove_next(check)
if check then
table.remove(working_IPA, i+1)
p_next = working_IPA[i+1]
end
end
-- intervocalic voicing and gemination of voiceless sonorants --
if vowel[p_prev] and intervocalic_voicing[p_current] ~= nil and vowel[p_next] then
p_resolve(intervocalic_voicing[p_current])
end
-- colloquial only --
if col then
if p_current == "r" and p_next == "ʃ" then
p_resolve("ʃ")
end
if p_current == "l" and p_next == "l" and working_IPA[i+2] == "i" and vowel[working_IPA[i+3]] then
p_resolve("ʎʎ")
remove_next()
remove_next()
end
if p_current == "n" and p_next == "n" and working_IPA[i+2] == "i" and vowel[working_IPA[i+3]] then
p_resolve("ɲɲ")
remove_next()
remove_next()
end
if p_current == "x" and p_next == "d" then
if vowel[working_IPA[i-2]] and (p_prev == "i" or p_prev == "u") then
table.remove(working_IPA, i)
working_IPA[i-2] = working_IPA[i-2] .. "ː"
p_current = working_IPA[i]
p_next = working_IPA[i+1]
elseif vowel[p_prev] then
p_resolve("ː")
end
end
end
-- phonetic only --
if phon then
-- l̥ → ɬ --
if p_current == "l̥" then
p_resolve("ɬ")
elseif p_current == "l̥l̥" then
p_resolve("ɬɬ")
end
-- terminal devoicing --
if devoice[p_current] and (boundary[p_next] or p_next == nil) then
p_resolve(devoice[p_current])
end
-- regressive devoicing --
if p_current == "r" and voiceless[p_next] then
p_resolve("r̥")
elseif p_current == "l" and voiceless[p_next] then
p_resolve("l̥")
end
-- /v/ --
if p_current == "v" then
if consonant[p_prev] and p_prev ~= "u̯" then
p_resolve("ʋ")
elseif consonant[p_next] or boundary[p_next] or p_next == nil then
p_resolve("u̯")
end
end
-- /w/ --
if p_current == "w" and p_prev ~= "u̯" then
p_resolve("u̯")
end
-- /ɥ/ --
if p_current == "ɥ" then
p_resolve("y̯")
end
-- /j/ --
if p_current == "j" and p_prev ~= "i̯" then
p_resolve("i̯")
end
-- /xj/ --
if p_current == "x" and p_next == "j" then
p_resolve("ç")
end
-- V{i u} --
if vowel[p_prev] and p_current == "i" then
p_resolve("i̯")
elseif vowel[p_prev] and p_current == "u" then
p_resolve("u̯")
end
end
i = i + 1
end
working_IPA = table.concat(working_IPA)
-- /xd/ --
if mw.ustring.find(working_IPA, "(xd)") then
working_IPA = mw.ustring.gsub(working_IPA, "(xd)", "hd") .. "~" .. mw.ustring.gsub(working_IPA, "(xd)", "ɦd")
end
return working_IPA
end
function export.generate(frame)
local args = getArgs(frame)
local outputIPA = args[1]
local parameters = {}
local p = 2
-- mw.log("——— Parameters ———")
while args[p] do
parameters[args[p]] = true
-- mw.log(args[p] .. " = true")
p = p + 1
end
outputIPA = mw.ustring.gsub(outputIPA, "(% %;)", " ")
outputIPA = generate_IPA(outputIPA, parameters["phon"] or false, parameters["colloquial"] or parameters["col"] or false)
local nolarge = ""
if parameters["nolarge"] then
nolarge = " nolarge"
end
if parameters["phon"] and parameters["format"] then
outputIPA = "<span class=\"IPA" .. nolarge .. "\">[" .. outputIPA .. "]</span>"
elseif parameters["format"] then
outputIPA = "<span class=\"IPA" .. nolarge .. "\">/" .. outputIPA .. "/</span>"
end
-- mw.log(outputIPA)
return outputIPA
end
return export
--[[
Debug console test string:
=p.generate(mw.getCurrentFrame():newChild{title="whatever",args={"alhlen"}})
]]