local export = {}
local ugsub = mw.ustring.gsub
local split = mw.text.split
local entering_tones = {
["1"] = "7", ["3"] = "8", ["6"] = "9"
}
local ipa_preprocess = {
[1] = {"a", "ă"}, [2] = {"yu", "y"}, [3] = {"ăă", "a"}, [4] = {"uk", "ŭk"}, [5] = {"ik", "ĭk"},
[6] = {"ou", "ŏu"}, [7] = {"eoi", "eoy"}, [8] = {"ung", "ŭng"}, [9] = {"ing", "ĭng"}, [10] = {"ei", "ĕi"},
[11] = {"oi", "oy"}, [12] = {"ui", "uy"}
}
local ipa_initial = {
["b"] = "p", ["p"] = "pʰ", ["m"] = "m", ["f"] = "f",
["d"] = "t", ["t"] = "tʰ", ["n"] = "n", ["l"] = "l", ["sl"] = "ɬ", ["nj"] = "ɲ̟",
["g"] = "k", ["k"] = "kʰ", ["ng"] = "ŋ", ["gw"] = "kʷ", ["kw"] = "kʷʰ",
["zh"] = "t͡ʃ", ["ch"] = "t͡ʃʰ", ["sh"] = "ʃ",
["z"] = "t͡s", ["c"] = "t͡sʰ", ["s"] = "s",
["h"] = "h", ["w"] = "w", ["j"] = "j",
[""] = ""
}
local ipa_nucleus = {
["a"] = "ä", ["ă"] = "ɐ",
["e"] = "ɛ", ["ĕ"] = "e",
["i"] = "i", ["ĭ"] = "ɪ",
["o"] = "ɔ", ["ŏ"] = "o",
["oe"] = "œ", ["eo"] = "ɵ",
["u"] = "u", ["ŭ"] = "ʊ",
["yu"] = "yː"
}
local ipa_coda = {
["i"] = "i̯", ["u"] = "u̯", ["y"] = "y̯",
["m"] = "m", ["n"] = "n", ["ng"] = "ŋ",
["p"] = "p̚", ["t"] = "t̚̚",
[""] = "",
}
local ipa_tone = {
["1"] = "⁴⁵",
["2"] = "²⁴",
["3"] = "³³",
["4"] = "<span style=\"cursor:help\" title=\"or 11\">²²</span>"
}
local ipa_tone_sandhi = {
["*"] = "⁻", ["-"] = "⁻", [""] = ""
}
local ipa_syllabic = {
["m"] = "m̩", ["ng"] = "ŋ̍"
}
local acute_accents = {
["a"] = "á", ["e"] = "é", ["i"] = "í", ["o"] = "ó", ["u"] = "ú"
}
local grave_accents = {
["a"] = "à", ["e"] = "è", ["i"] = "ì", ["o"] = "ò", ["u"] = "ù"
}
local macrons = {
["a"] = "ā", ["e"] = "ē", ["i"] = "ī", ["o"] = "ō", ["u"] = "ū"
}
local yale_final = {
["aa"] = "a", ["eu"] = "沒有耶魯拼音", ["em"] = "沒有耶魯拼音", ["ep"] = "沒有耶魯拼音", ["et"] = "沒有耶魯拼音",
["oe"] = "eu", ["oeng"] = "eung", ["oek"] = "euk", ["eoi"] = "eui", ["eon"] = "eun", ["eot"] = "eut",
[""] = ""
}
local canton_pinyin_regex = {
["yu"] = "y", ["eo"] = "oe", ["eoi"] = "oey", ["z(h?)"] = "dz%1", ["c(h?)"] = "ts%1",
["([ptk])([136])"]=function(a,b) return a .. entering_tones[b] end,
["([ptk][1-6][%-%*])([136])$"]=function(a,b) return a .. entering_tones[b] end
}
local function yale_tone(final,b)
if b == "4" or b == "5" or b == "6" then
final = final:gsub("([ptkmn]?g?)$", "h%1", 1)
end
if b == "1" then final = final:gsub("[aeiou]", macrons, 1) end
if b == "4" then final = final:gsub("[aeiou]", grave_accents, 1) end
if b == "2" or b == "5" then final = final:gsub("[aeiou]", acute_accents, 1) end
return final
end
function export.jyutping_to_ipa(text)
if type(text) == "table" then text = text.args[1] end
if text:find("[7-9]") then error("Invalid tone in Jyutping.") end
if text:find("[A-Z]") then error("Please do not capitalize the Jyutping.") end
if text:find("%-[A-Za-z]") then error("Please do not hyphenate the Jyutping.") end
if text:find("[0-9][a-z]") then error("Error in the Jyutping transcription.") end
text = text:lower()
text = text:gsub("jy([^u])", "j%1")
text = text:gsub("%.%.%.", " ")
text = text:gsub(",", "隔"):gsub("隔 ", ", ")
local reading = split(text, "隔")
local function retrieve_ipa1(a,b,c,d,e)
return ipa_initial[a] .. ipa_syllabic[b] .. ipa_tone[c] .. ipa_tone_sandhi[d] .. ipa_tone[e]
end
local function retrieve_ipa2(a,b,c,d,e,f)
return (ipa_initial[a] or error(("Unrecognised initial: \"%s\""):format(a))) ..
(ipa_nucleus[b] or error(("Unrecognised nucleus: \"%s\""):format(b))) ..
(ipa_coda[c] or error(("Unrecognised coda: \"%s\""):format(c))) ..
(ipa_tone[d] or error(("Unrecognised tone: \"%s\""):format(d))) ..
ipa_tone_sandhi[e] ..
(ipa_tone[f] or error(("Unrecognised tone: \"%s\""):format(f)))
end
local function get_entering_tone(a,b) return a .. entering_tones[b] end
for i = 1,#reading do
reading[i] = reading[i]:gsub("([1-6])[a-z]", "%1 "):gsub("[^a-z1-6%-%* ]", "")
local syllable = split(reading[i]:lower(), " ")
for i = 1,#syllable do
syllable[i] = syllable[i]:gsub("([zcs])yu", "%1hyu")
syllable[i] = syllable[i]:gsub("([zc])oe", "%1hoe")
syllable[i] = syllable[i]:gsub("([zc])eo", "%1heo")
if (not syllable[i]:find("^[bcdfghjklmnpqrstvwxyz]?[bcdfghjklmnpqrstvwxyz]?[aeiouy]+[mnptk]?g?[1-9][%-%*]?[1-9]?$") and not syllable[i]:find("^h?[mn]g?[1-9][%-%*]?[1-9]?$")) and syllable[i]:find("[a-z1-9]") then
error("Incorrect Jyutping format. Please check!")
end
syllable[i] = syllable[i]:gsub("^(h?)([mn]g?)([1-6])([%-%*]?)([1-6]?)$", retrieve_ipa1)
syllable[i] = syllable[i]:gsub("([ptk])([136])", get_entering_tone)
syllable[i] = syllable[i]:gsub("([ptk][1-9][%-%*])([136])$", get_entering_tone)
for regex_idx,regex_pair in ipairs(ipa_preprocess) do
syllable[i] = ugsub(syllable[i],regex_pair[1],regex_pair[2])
end
syllable[i] = ugsub(syllable[i], "spăn", "span")
syllable[i] = ugsub(syllable[i],
"([bcdfghjklmnpqrstvwxyz]?[bcdfghjklmnpqrstvwxyz]?)([aăeĕiĭoŏuŭy][eo]?)([iuymngptk]?g?)([1-9])([%-%*]?)([1-9]?)",
retrieve_ipa2)
end
reading[i] = table.concat(syllable, " ")
end
return table.concat(reading, "/, /")
end
function export.jyutping_to_yale(text)
if type(text) == "table" then text = text.args[1] end
text = text:gsub("jy([^u])", "j%1")
text = text:gsub(",", "隔"):gsub("隔 ", ", ")
local reading = split(text, "隔")
local function yale_tone2(a, b) return yale_tone((yale_final[a] or a), b) end
for i = 1,#reading do
reading[i] = reading[i]:gsub("[1-6]%-", "")
reading[i] = reading[i]:gsub("([1-9])[a-z]", "%1 ")
local syllables = split(mw.ustring.lower(reading[i]), " ",true)
for j = 1,#syllables do
local text = syllables[j]
if text:find("jy[^u]") then error("wrong usage of 'jy' in Jyutping") end
text = text:gsub("[789]",{["7"] = "1", ["8"] = "3", ["9"] = "6"})
text = text:gsub("^m[1-6]$",{["m1"] = "{{懸停|或m̀|m̄}}", ["m2"] = "ḿ", ["m3"] = "m", ["m4"] = "m̀h", ["m5"] = "ḿh", ["m6"] = "mh"})
text = text:gsub("^ng[1-6]$",{["ng1"] = "{{懸停|或ǹg|n̄g}}", ["ng2"] = "ńg", ["ng3"] = "ng", ["ng4"] = "ǹgh", ["ng5"] = "ńgh", ["ng6"] = "ngh"})
text = text:gsub("jy?", "y")
text = text:gsub("[cz]",{["z"] = "j", ["c"] = "ch"})
text = text:gsub("([aeiou][aeiou]?[iumngptk]?[g]?)([1-6])", yale_tone2)
text = text:gsub("(yu[tn]?)([1-6])", yale_tone)
text = ugsub(text, "[āēīōū]",{["ā"] = "{{懸停|或à|ā}}", ["ē"] = "{{懸停|或è|ē}}", ["ī"] = "{{懸停|或ì|ī}}", ["ō"] = "{{懸停|或ò|ō}}", ["ū"] = "{{懸停|或ù|ū}}"})
text = text:gsub("{{懸停|或([^|]+)|([^}]+)}}", '<span style="cursor:help" title="or %1">%2</span>')
syllables[j] = text
end
reading[i] = table.concat(syllables, " ")
if reading[i]:find("沒有耶魯拼音") then
reading[i] = "colloquial sounds not defined"
end
end
return table.concat(reading, ", ")
end
function export.jyutping_to_cantonese_pinyin(text)
if type(text) == "table" then text = text.args[1] end
text = text:gsub("jy([^u])", "j%1")
text = text:gsub(",", "隔"):gsub("隔 ", ", ")
local reading = split(text, "隔")
for i=1,#reading do
for regex, replace in pairs(canton_pinyin_regex) do
reading[i] = ugsub(reading[i], regex, replace)
end
end
return table.concat(reading, ", ")
end
function export.jyutping_to_guangdong(text)
if type(text) == 'table' then text = text.args[1] end
local palatal = { ['z']='j', ['c']='q', ['s']='x' }
local function palatalize(a,b) return palatal[a] .. b end
local final = { ['p']='b', ['k']='g', ['t']='d' }
local function get_final(a,b) return final[a] .. b end
text = text:gsub(",", "隔"):gsub("隔 ", ", ")
local reading = split(text, "隔")
for i, item in ipairs(reading) do
item = item:gsub('yu', 'ü')
item = item:gsub('eoi', 'êü')
item = item:gsub('j', 'y')
item = item:gsub('[oe][oe]', 'ê')
item = item:gsub('e', 'é')
item = ugsub(item, '([zcs])([iü])', palatalize)
item = ugsub(item, '([jqxyê])ü', '%1u')
item = item:gsub('a+', { ['aa']='a', ['a']='e' } )
item = item:gsub('([kg])w', '%1u')
item = item:gsub('([ae])u', '%1o')
item = item:gsub('([pkt])(%d)', get_final)
item = item:gsub('%d%-(%d)', '%1')
if item:find('é[umb]') then
item = 'colloquial sounds not defined'
end
reading[i] = item
end
return table.concat(reading, ", ")
end
function export.jyutping_format(text)
if type(text) == "table" then text = text.args[1] end
if text:find("%[%[") then
return "just a lengthy text to ensure it works"
end
text = split(text:gsub(",", "隔"):gsub("隔 ", ", "), "隔")
for i, to_be_processed in ipairs(text) do
text[i] = to_be_processed:gsub("[1-6]%-([1-6])", "%1")
end
return text
end
function export.hoisanva_to_ipa(text)
local hsv_initial = {
["b"] = "p", ["p"] = "pʰ", ["m"] = "ᵐb", ["f"] = "f", ["v"] = "v",
["d"] = "t", ["t"] = "tʰ", ["n"] = "ⁿd", ["l"] = "l",
["g"] = "k", ["k"] = "kʰ", ["ng"] = "ᵑɡ",
["z"] = "t͡s", ["c"] = "t͡sʰ",
["y"] = "ʒ", ["s"] = "s", ["h"] = "h", [""] = ""
}
local hsv_final = {
["a"] = "a", ["ai"] = "ai", ["au"] = "au", ["am"] = "am",
["an"] = "an", ["ang"] = "aŋ", ["ap"] = "ap̚", ["at"] = "at̚",
["ak"] = "ak̚",
["i"] = "i", ["iu"] = "iu", ["im"] = "im", ["in"] = "in",
["ip"] = "ip̚", ["it"] = "it̚",
["ie"] = "iɛ", ["iau"] = "iau", ["iam"] = "iam", ["iang"] = "iaŋ",
["iap"] = "iap̚", ["iak"] = "iak̚",
["u"] = "u", ["ui"] = "ui", ["un"] = "un", ["ut"] = "ut̚",
["ei"] = "ei", ["eu"] = "eu", ["em"] = "em", ["en"] = "en",
["uung"] = "ɵŋ", ["ep"] = "ep̚", ["et"] = "et̚", ["uuk"] = "ɵk̚", ["uut"] = "ɵt̚",
["o"] = "ᵘɔ", ["oi"] = "ᵘɔi", ["on"] = "ᵘɔn", ["ong"] = "ɔŋ",
["ot"] = "ᵘɔt̚", ["ok"] = "ɔk̚",
["m"] = "m̩"
}
local hsv_tone = { "³³", "⁵⁵", "²²", "²¹", "³²" }
local gsplit = mw.text.gsplit
local result = {}
for word in gsplit(text, ",") do
local initial, final, tone, tone_ch, word_result = "", "", "", "", {}
for syllable in gsplit(word, " ") do
initial, final, tone, tone_ch = syllable:match("^([^aeiou]*)([^1-5]*)([1-5])([%*%-]?[1-5]?%*?)$")
if final == "" then final, initial = initial, "" end
if not hsv_initial[initial] or not hsv_final[final] or not hsv_tone[tonumber(tone)] or (tone == "2" and tone_ch == "*") then
error("Syllable '" .. syllable .. "' is not a valid syllable for IPA conversion.")
end
table.insert(word_result,
hsv_initial[initial] .. hsv_final[final] .. hsv_tone[tonumber(tone)] ..
(tone_ch ~= "" and "⁻" or "") ..
(tone_ch == "*" and hsv_tone[tonumber(tone)] or (tone_ch:find("^%-[1-5]") and hsv_tone[tonumber(tone_ch:sub(2, 2))] or "")) ..
(tone_ch:sub(-1, -1) == "*" and "⁵" or ""))
end
table.insert(result, table.concat(word_result, " "))
end
return "/" .. table.concat(result, "/, /") .. "/"
end
return export
