local export = {}
local lang = "siwa"
local m_IPA = require("Module:IPA")
local m_su = require("Module:string utilities")
local m_table = require("Module:table")
local m_sm = mw.loadData("Module:siwa-pron/data")
local sub = mw.ustring.sub
local find = mw.ustring.find
local gmatch = mw.ustring.gmatch
local gsub = mw.ustring.gsub
local u = mw.ustring.char
local split = mw.text.split
local UNRELEASED = u(0x031A) -- COMBINING LEFT ANGLE ABOVE. ̚
local NASALIZED = u(0x0303) -- COMBINING TILDE. ̃
--obsolete ligatures and L with stroke used to remove two-character hassle. will replace later
local consonant = "[mnɲŋpbtdcɟkɡʔvðsɕxɣhʨʥrlɬłʣjwʦʔƀꝺđɣꬶɉ]" .. UNRELEASED .. "?"
local front_vowel = "iɪyeøɛœæa"
local back_vowel = "uɔ" .. NASALIZED .. "?ɑʊɤ"
local vowel = "[" .. front_vowel .. back_vowel .. "ɨ]"
local spat1 = "(·?ˈ[mnɲŋpbtdcɟkɡʔvðsɕxɣhʨʥrlɬłʣjwʦ⁽ʰ⁾ʔːƀꝺđɣꬶɉ]*"..UNRELEASED.."?"..vowel..")(h)([^ː])"
local spat2 = "(·?ˈ[mnɲŋpbtdcɟkɡʔvðsɕxɣhʨʥrlɬłʣjwʦ⁽ʰ⁾ʔːƀꝺđɣꬶɉ]*"..UNRELEASED.."?"..vowel..")(ꬶ)([^ː])"
local spat3 = "(·?ˈ[mnɲŋpbtdcɟkɡʔvðsɕxɣhʨʥrlɬłʣjwʦ⁽ʰ⁾ʔːƀꝺđɣꬶɉ]*"..UNRELEASED.."?"..vowel..")(ƀ)([^ː])"
local spat4 = "(·?ˈ[mnɲŋpbtdcɟkɡʔvðsɕxɣhʨʥrlɬłʣjwʦ⁽ʰ⁾ʔːƀꝺđɣꬶɉ]*"..UNRELEASED.."?"..vowel..")(đ)([^ː])"
local spat5 = "(·?ˈ[mnɲŋpbtdcɟkɡʔvðsɕxɣhʨʥrlɬłʣjwʦ⁽ʰ⁾ʔːƀꝺđɣꬶɉ]*"..UNRELEASED.."?"..vowel..")(ꝺ)([^ː])"
local function ncategories(categories)
local out_categories = {}
for key, cat in ipairs(categories) do
out_categories[key] = "[[Category:" .. cat .. "]]"
end
return table.concat(out_categories, "")
end
local function open_to_closed(v, a)
local otc = {}
local switch = {["ɑ"] = "a", ["e"] = "ɛ", ["i"] = "ɪ", ["ɔ"] = "ɔ", ["u"] = "ʊ", ["y"] = "œ",
["ø"] = function(s1)
if(a=="w") then s1="ɤ"
else s1="œ" end
return s1 end,}
for vc in gmatch(v, ".") do
vc = gsub(vc, vc, switch[vc])
table.insert(otc, vc)
end
return table.concat(otc)
end
local rules = {
{ --ligatures and ł
["t[ṡɕ]"] = "ʨ", ["ṡ"] = "ɕ", ["į"] = "j", ["dj"] = "ʥ",
["ḍ"] = "ð", ["dl"] = "ł", ["kj"] = "c", ["ḥ"] = "ʔ",
["nj"] = "ɲ", ["ġ"] = "x", ["ts"] = "ʦ", ["g"] = "ɡ", -- IPA g
},
{ --long consonants
["mm"] = "mː", ["bb"] = "pː", ["vv"] = "wː", ["nn"] = "nː",
["dʥ"] = "ʥː", ["dd"] = "tː", ["ðð"] = "ðː", ["ss"] = "sː",
["ɕɕ"] = "ɕː", ["rr"] = "rː", ["ll"] = "lː", ["ɡɡ"] = "kː",
["xx"] = "xː", ["nɡ"] = "ŋː", ["hh"] = "hː", ["ʔʔ"] = "ʔː",
["nɲ"] = "ɲː", ["hl"] = "ɬː",
},
{ --[[default all consonants to unstressed. [] with stroke and ꬶ (U+AB36)
to tell apart natural and stress-borne]]
["p"] = "ƀ", ["d"] = "ꝺ", ["t"] = "đ",
["ɡ"] = "ɣ", ["k"] = "ꬶ",
},
{ --default all vowels as open (open-closed distinctions are computed later)
["a"] = "ɑ", ["ả"] = "æː",
["ę"] = "æ",
["ẻ"] = "eː",
["ỉ"] = "iː",
["o"] = "ɔ", ["ỏ"] = "ʊː",
["ủ"] = "uː",
["ỷ"] = "yː",
["ů"] = "ø", ["ẻu"] = "øː",
["õ"] = "ɔ̃", ["õu"] = "ɔ̃ː̃",
},
{
["^(ˈ)ꬶ([" .. front_vowel .. "])"] = "%1c%2", --word-initial [k] palatalizes before front-vowels
["^(ˈ[ƀđꬶc])"] = "%1ʰ", --voiceless stops word-initially become aspirated
["^(.*·ˈ[ƀđꬶc])"] = "%1⁽ʰ⁾",
["^(ˈ)ɣj([" .. front_vowel .. "])"] = "%1ʣ%2", --<gį> word-initially and before front vowels is pronounced [d͡z]
["^(ˈ)ɣj([" .. back_vowel .. "])"] = "%1ɟ%2",
["^(ˈ)ɣ([" .. front_vowel .. "])"] = "%1ɟ%2",
},
{
["ˈƀ"] = "ˈp", ["ˈđ"] = "ˈt", ["ˈꬶ"] = "ˈk", ["ˈꝺ"] = "ˈd",
["đi"] = "ʨi", ["ꝺi"] = "ʥi", ["ɣi"] = "ɉi", ["ɣj"] = "jː", ["ɣjː"] = "ɟː",
},
{
[spat1] = "%1ʔ%3", [spat2] = "%1k%3", [spat3] = "%1p%3", [spat4] = "%1t%3", [spat5] = "%1ð%3"
},
{
["ƀƀ"] = "ʔp", ["pƀ"] = "ʔp",
["đđ"] = "ʔt", ["tđ"] = "ʔt",
["ꬶꬶ"] = "ʔk", ["kꬶ"] = "ʔk",
["bm"] = "ʔp̚m", ["ꝺn"] = "ʔt̚n", ["ꬶn"] = "ʔk̚ŋ",
["mn"] = "mnː", ["mʔk"] = "mkː",
["(p[msɕ])"] = "%1ː", ["pr"] = "px",
["b([sɕ])"] = "p%1",
["nꬶ"] = "ŋk", ["([ðđʦłɕꬶ])v"] = "%1wː",
["đn"] = "tnː", ["đr"] = "tx", ["đꬶv"] = "tkwː",
["(ʦ[đlmn])"] = "%1ː", ["ʦꬶv"] = "ʦkwː",
["ʦxv"] = "ʦxw", ["đx"] = "tːx",
["(ð[mꬶ])"] = "%1ː",
["ꝺx"] = "ðx",
["(sk[l])"] = "%1ː", ["sʔk"] = "skː", ["sxv"] = "sxwː",
["([lr])ʔ([ptk])"] = "%1%2ː", ["rv"] = "rwː",
["lʦ(x?)v"] = "ɬʦ%1w", ["lʦx"] = "ɬʦx",
["(ł[mnꬶ])"] = "%1ː",
["(ꬶ[msɕl])"] = "%1ː",
["ꬶsꬶ"] = "kskː", ["ꬶsl"] = "ksł",
["ɣ([mn])"] = "ŋ%1ː", ["ɣ([vsl])"] = "kv",
},
{
["(" .. vowel .. "*)(" .. consonant .. consonant .. ")"] = function(s1, s2) return open_to_closed(s1) .. s2 end,
["(" .. vowel .. "*)(" .. consonant .. ")$"] = function(s1, s2) return open_to_closed(s1) .. s2 end,
["ɑ$"] = "a",
},
{
["į"] = "j", ["l(ʦx)v"] = "ɬ%1w",
["(" .. vowel .. ")đ$"] = "%1ʔ%1", -- -Vt becomes -VʔV (or -Vht, not considered)
},
{ --undo ligatures
["ʨ"] = "t͡ɕ", ["ʥ"] = "d͡ʑ", ["ł"] = "tɬ", ["ʣ"] = "d͡z", ["ʦ"] = "t͡s",
["ƀ"] = "p", ["ꝺ"] = "d", ["đ"] = "t", ["ꬶ"] = "ɡ", ["ɉ"] = "ɟ"
},
{
["·"] = "", --remove morpheme separator
},
}
local wrules = {
{ --ligatures and ł
["t[ṡɕ]"] = "ʨ", ["ṡ"] = "ɕ", ["į"] = "j", ["dj"] = "ʥ",
["ḍ"] = "ð", ["dl"] = "ł", ["kj"] = "c", ["ḥ"] = "ʔ",
["nj"] = "ɲ", ["ġ"] = "x", ["ts"] = "ʦ", ["g"] = "ɡ", -- IPA g
},
{ --long consonants
["mm"] = "mː", ["bb"] = "pː", ["vv"] = "wː", ["nn"] = "nː",
["dʥ"] = "ʥː", ["dd"] = "tː", ["ðð"] = "ðː", ["ss"] = "sː",
["ɕɕ"] = "ɕː", ["rr"] = "rː", ["ll"] = "lː", ["ɡɡ"] = "kː",
["xx"] = "xː", ["nɡ"] = "ŋː", ["hh"] = "hː", ["ʔʔ"] = "ʔː",
["nɲ"] = "ɲː", ["hl"] = "ɬː",
},
{ --[[default all consonants to unstressed. [] with stroke and ꬶ (U+AB36)
to tell apart natural and stress-borne]]
["p"] = "ƀ", ["d"] = "ꝺ", ["t"] = "đ",
["ɡ"] = "ɣ", ["k"] = "ꬶ",
},
{ --default all vowels as open (open-closed distinctions are computed later)
["a"] = "ɑ", ["ả"] = "æː",
["ę"] = "æ",
["ẻ"] = "eː",
["ỉ"] = "iː",
["o"] = "ɔ", ["ỏ"] = "ʊː",
["ủ"] = "uː",
["ỷ"] = "yː",
["ů"] = "ø", ["ẻu"] = "øː",
["õ"] = "ɔ̃", ["õu"] = "ɔ̃ː̃",
},
{
["^(ˈ)ꬶ([" .. front_vowel .. "])"] = "%1c%2", --word-initial [k] palatalizes before front-vowels
["^(ˈ[ƀđꬶc])"] = "%1ʰ", --voiceless stops word-initially become aspirated
["^(.*·ˈ[ƀđꬶc])"] = "%1⁽ʰ⁾",
["^(ˈ)ɣj([" .. front_vowel .. "])"] = "%1ʣ%2", --<gį> word-initially and before front vowels is pronounced [d͡z]
["^(ˈ)ɣj([" .. back_vowel .. "])"] = "%1ɟ%2",
["^(ˈ)ɣ([" .. front_vowel .. "])"] = "%1ɟ%2",
},
{
["ˈƀ"] = "ˈp", ["ˈđ"] = "ˈt", ["ˈꬶ"] = "ˈk", ["ˈꝺ"] = "ˈd",
["đi"] = "ʨi", ["ꝺi"] = "ʥi", ["ɣi"] = "ɉi", ["ɣj"] = "jː", ["ɣjː"] = "ɟː",
},
{
[spat1] = "%1ʔ%3", [spat2] = "%1k%3", [spat3] = "%1p%3", [spat4] = "%1t%3", [spat5] = "%1ð%3"
},
{
["ƀƀ"] = "ʔp", ["pƀ"] = "ʔp",
["đđ"] = "ʔt", ["tđ"] = "ʔt",
["ꬶꬶ"] = "ʔk", ["kꬶ"] = "ʔk",
["bm"] = "ʔp̚m", ["ꝺn"] = "ʔt̚n", ["ꬶn"] = "ʔk̚ŋ",
["mn"] = "mnː", ["mʔk"] = "mkː",
["(p[msɕ])"] = "%1ː", ["pr"] = "px",
["b([sɕ])"] = "p%1",
["nꬶ"] = "ŋk", ["([ðđʦłɕꬶ])v"] = "%1wː",
["đn"] = "tnː", ["đr"] = "tx", ["đꬶv"] = "tkwː",
["(ʦ[đlmn])"] = "%1ː", ["ʦꬶv"] = "ʦkwː",
["ʦxv"] = "ʦxw", ["đx"] = "tːx",
["(ð[mꬶ])"] = "%1ː",
["ꝺx"] = "ðx",
["(sk[l])"] = "%1ː", ["sʔk"] = "skː", ["sxv"] = "sxwː",
["([lr])ʔ([ptk])"] = "%1%2ː", ["rv"] = "rwː",
["lʦ(x?)v"] = "ɬʦ%1w", ["lʦx"] = "ɬʦx",
["(ł[mnꬶ])"] = "%1ː",
["(ꬶ[msɕl])"] = "%1ː",
["ꬶsꬶ"] = "kskː", ["ꬶsl"] = "ksł",
["ɣ([mn])"] = "ŋ%1ː", ["ɣ([vsl])"] = "kv",
},
{
["(" .. vowel .. "*)(" .. consonant .. consonant .. ")"] = function(s1, s2) return open_to_closed(s1,"w") .. s2 end,
["(" .. vowel .. "*)(" .. consonant .. ")$"] = function(s1, s2) return open_to_closed(s1,"w") .. s2 end,
["ɑ$"] = "a",
},
{
["į"] = "j", ["l(ʦx)v"] = "ɬ%1w",
["(" .. vowel .. ")đ$"] = "%1ʔ%1", -- -Vt becomes -VʔV (or -Vht, not considered)
},
{ --undo ligatures
["ʨ"] = "t͡ɕ", ["ʥ"] = "d͡ʑ", ["ł"] = "tɬ", ["ʣ"] = "d͡z", ["ʦ"] = "t͡s",
["ƀ"] = "p", ["ꝺ"] = "d", ["đ"] = "t", ["ꬶ"] = "ɡ", ["ɉ"] = "ɟ"
},
{
["·"] = "", --remove morpheme separator
},
}
function export.morphemes(word)
local pss = {}
if gmatch(word,"·") then
pss = split(word,"·")
end
for i, m in ipairs(pss) do
if m_sm.suffix[m] and gmatch(table.concat(pss),"[ˈˌ]") then
elseif m_sm.prefix[m] then
pss[i] = "ˌ" .. pss[i]
else
pss[i] = "ˈ" .. pss[i]
end
end
return table.concat(pss,"·")
end
function export.crux(term)
local IPA = {}
term=mw.ustring.lower(term)
term=export.morphemes(term)
for _, rule in ipairs(rules) do
for regex, replacement in pairs(rule) do
term = gsub(term, regex, replacement)
end
end
table.insert(IPA, term)
return table.concat(IPA)
end
--[[function format_dialect(replaced, dialect)
local function IPA_format(items)
local bits = {}]]
-- local prefix = "[[w:IPA chart|IPA]]<sup>([[IPA for Siwa|key]])</sup>: "
--[[ local span = "<span style=\"font-size:110%;font-family:Gentium,'DejaVu Sans','Segoe UI',sans-serif>"
return prefix .. span .. items.pron .. "</span>"
end
local prons = {}
for _, done in ipairs(replaced) do
local phonetic = done[1]
local IPA_args = {{pron = '[' .. phonetic .. ']'}}
table.insert(prons, m_IPA.format_IPA_full(lang, IPA_args))
end
return "(''"..dialect.."'') " .. table.concat(prons, ' or ')
end]]
function make_row(phoneticizations, dials)
local full_pronuns = {}
for _, phoneticization in ipairs(phoneticizations) do
local phonetic = phoneticization[1]
local IPA_args = {{pron = '[' .. phonemic .. ']'}}
table.insert(full_pronuns, m_IPA.format_IPA_full(lang, IPA_args))
end
return "(''"..dials.."'')" .. ' ' .. table.concat(full_pronuns, ' or ')
end
function export.show(frame)
local params = {
[1] = { default = mw.title.getCurrentTitle().nsText == 'Template' and "uįo·sauṡṡi" or mw.title.getCurrentTitle().text },
a = {type = 'boolean', default = true},
e = {type = 'boolean', default = true},
w = {type = 'boolean', default = true},
acc = {list = true},
}
local args = require("Module:parameters").process(frame:getParent().args, params)
local term = args[1]
local accent = args.acc
local categories = {}
local ipa = "* "
if args.a then
ipa = ipa .. make_row(export.crux(term, false, false), accent and #accent > 0 or {'Aingo'})
end
if args.e then
if args.a then
ipa = ipa .. "\n* "
end
out = out .. make_row(export.crux(term, true, false), accent and #accent > 0 or {'Eastern'})
table.insert(categories, "Siwa terms with Eastern IPA pronunciation")
end
if args.w then
if args.a or args.e then
ipa = ipa .. "\n* "
end
ipa = ipa .. make_row(export.crux(term, false, true), accent and #accent > 0 or {'Western'})
table.insert(categories, "Siwa terms with Western IPA pronunciation")
end
return ipa .. ncategories(categories)
end
return export