Module:zm-pron: Difference between revisions

From Linguifex
Jump to navigation Jump to search
(Undo revision 270553 by Aleisi Galan (talk))
Tag: Undo
No edit summary
 
(15 intermediate revisions by the same user not shown)
Line 7: Line 7:
local split = mw.text.split
local split = mw.text.split
local gsplit = mw.text.gsplit
local gsplit = mw.text.gsplit


-- To avoid weird annoying cursor behavior
-- To avoid weird annoying cursor behavior
Line 20: Line 21:
local acuted = "[áéíóúýÉÓ]"
local acuted = "[áéíóúýÉÓ]"
local oral = "áéíóúýÉÓaouɔieɛ"
local oral = "áéíóúýÉÓaouɔieɛ"
local palatal = "[ʨjʎʃʒʥʤʧʦʣ]"
local palatal = "[ʨjʎʃʒʥʤʧʦʣɕʑ]"
local sonorant = "[rl]"
local sonorant = "[rl]"
local plosives = "[ptkbdɡ]"
local plosives = "[ptkbdg]"
local velar = "[ɡkx]"
local velar = "[ɡkx]"
local consonants = "[mnptkbdɡfvszxrljɲŋʋʎʨjʎʃʒʥʤʧʤʣʦ]"
local consonants = "[mnptkbdɡfvszxrljɲŋʋʎʨjʎʃʒʥʤʧʤʣʦɕʑ]"


local export = {}
local export = {}
Line 31: Line 32:
{"ai", "ɛ"}, {"aí", "É"}, {"au", "ɔ"}, {"aú", "Ó"}, {"è", "ɛ"}, {"ò", "ɔ"}, {"y", "ɪ"},
{"ai", "ɛ"}, {"aí", "É"}, {"au", "ɔ"}, {"aú", "Ó"}, {"è", "ɛ"}, {"ò", "ɔ"}, {"y", "ɪ"},
{"dź", "ʥ"}, {"dž", "ʤ"}, {"c", "ʦ"}, {"ć", "ʨ"}, {"nj", "ɲ"}, {"č", "ʧ"}, {"ž", "ʒ"},  {"š", "ʃ"},  {"lj", "ʎ"},
{"dź", "ʥ"}, {"dž", "ʤ"}, {"c", "ʦ"}, {"ć", "ʨ"}, {"nj", "ɲ"}, {"č", "ʧ"}, {"ž", "ʒ"},  {"š", "ʃ"},  {"lj", "ʎ"},
{"h", "x"}, {"g", "ɡ"}, {"ǫ", "o"}, {"ō", "o"},
{"h", "x"}, {"g", "ɡ"}, {"ǫ", "o"}, {"ō", "o"}, {"sč", "ʃt͡ʃ"}, {"ou", "u"}, {"ś", "ɕ"}, {"ź", "ʑ"},
{"(" .. vowels .. ")[lv]$", "%1ʋ"}, {"(" .. acuted .. ")[lv]$", "%1ʋ"},
{"(" .. vowels .. ")[lv]$", "%1ʋ"}, {"(" .. acuted .. ")[lv]$", "%1ʋ"},
{plosives .. "(" .. plosives .. ")$", "%1"},
{"(" .. plosives .. ")[ptkbdg]$", "[ptkbdg]"},
{"[ptk]([ptk])", "%1"}, {"[bdɡ]([bdɡ])", "%1"},
{"pt" , "t"}, {"pk" , "k"}, {"tp" , "p"}, {"tk" , "k"}, {"kp" , "p"}, {"kt" , "t"},
{"bd" , "d"}, {"bg" , "g"}, {"db" , "b"}, {"dg" , "g"}, {"gb" , "b"}, {"gd" , "d"},
{"(" .. consonants ..")([rŕ])(" .. consonants ..")", "%1(ə)%2" .. SYLL .. "%3"}, {"(" .. consonants ..")r$", "%1ər"},
{"(" .. consonants ..")([rŕ])(" .. consonants ..")", "%%2" .. SYLL .. "%3"}, {"(" .. consonants ..")r$", "%1ər"},
{"n(" .. velar .. ")", "ŋ%1"},
{"n(" .. velar .. ")", "ŋ%1"},
{"([ptk])(" .. vowels .. ")", "%%2"}, {"([ptk])(" .. acuted .. ")", "%%2"},
{"p(" .. vowels .. ")", "pʰ%1"}, {"t(" .. vowels .. ")", "%1"}, {"k(" .. vowels .. ")", "kʰ%1"},
{"p(" .. acuted .. ")", "pʰ%1"}, {"t(" .. acuted .. ")", "%1"}, {"k(" .. acuted .. ")", "kʰ%1"},
{"(" .. palatal .. ")e", "%1ɛ"}, {"(" .. palatal .. ")é", "%1É"},
{"(" .. palatal .. ")e", "%1ɛ"}, {"(" .. palatal .. ")é", "%1É"},
Line 47: Line 50:
{"É", "ɛ" .. ACUTE}, {"Ó", "ɔ" .. ACUTE}, {"ý", "ɪ" .. ACUTE},
{"É", "ɛ" .. ACUTE}, {"Ó", "ɔ" .. ACUTE}, {"ý", "ɪ" .. ACUTE},
}
local c_rules = {
{"ʨ", "ʧ"}, {"ʥ", "ʤ"},
{"[ae]j", "ɛː"}, {"a", "ɐ"}, {"[ei]", "ɪ"}, {"o", "ɔ"}, {"u", "ʊ"}
}
local s_rules = {
{"ʨ", "ʦ"}, {"ʥ", "ʣ"},
{"[ae]j", "ɛː"}, {"a", "ɐ"}, {"e", "ɛ"}, {"i", "e"}, {"o", "ɔ"}, {"u", "o"}, {"ɪ", "i"},
{"[aeiouːɛɔ]b", "[aeiouːɛɔ]β"},
}
local n_rules = {
{"[ae]", "ə"}, {"ɪ", "ɨ"}, {"i", "ɪ"}, {"o", "ɔ"}, {"u", "ʊ"},
}
}


Line 67: Line 55:
{"ʥ", "d͡ʑ"}, {"ʤ", "d͡ʒ"}, {"ʦ", "t͡s"}, {"ʨ", "t͡ɕ"}, {"ʧ", "t͡ʃ"}, {"ʣ", "d͡z"},
{"ʥ", "d͡ʑ"}, {"ʤ", "d͡ʒ"}, {"ʦ", "t͡s"}, {"ʨ", "t͡ɕ"}, {"ʧ", "t͡ʃ"}, {"ʣ", "d͡z"},
}
}


function export.crux(term, d)
function export.crux(term, d)
term = mw.ustring.lower(term)
term = mw.ustring.lower(term)
-- General
for _, rule in ipairs(rules) do
for _, rule in ipairs(rules) do
term = gsub(term, rule[1], rule[2])
term = gsub(term, rule[1], rule[2])
end
end
-- Dialects
if d == "c" then term = gsub(term, "ʨ", "ʧ"); term = gsub(term, "ʥ", "ʤ") end
for _, dia_rule in ipairs(d == "c" and c_rules or d == "n" and n_rules or d == "s" and s_rules or nil) do
if d == "s" then term = gsub(term, "ʨ", "ʦ"); term = gsub(term, "ʥ", "ʣ"); term = gsub(term, "ʰ", "") end
term = gsub(term, dia_rule[1], dia_rule[2])
if d == "s" then term = gsub(term, "aj", "ɛː"); term =gsub(term, "ej", "ɛː") end
end
if d == "c" then term = gsub(term, "aj", "ɛː"); term =gsub(term, "ej", "ɛː") end
if d == "c" then term = gsub(term, "a", "ɐ"); term = gsub(term, "e", "ɪ"); term = gsub(term, "i", "ɪ"); term = gsub(term, "o", "ɔ"); term = gsub(term, "u", "ʊ")  end
if d == "s" then term = gsub(term, "a", "ɐ"); term = gsub(term, "e", "ɛ"); term = gsub(term, "i", "e"); term = gsub(term, "o", "ɔ"); term = gsub(term, "u", "o"); term = gsub(term, "ɪ", "i"); term = gsub(term, "ɪ́", "í"); term = gsub(term, "ų", "u") end
if d == "n" then term = gsub(term, "a", "ə"); term = gsub(term, "e", "ə"); term = gsub(term, "i", "y"); term = gsub(term, "o", "ɔ"); term = gsub(term, "u", "ʊ"); term = gsub(term, "ɪ", "ɨ"); term = gsub(term, "ɪ́", "ɨ́"); term = gsub(term, "y", "ɪ")  end
 
-- Affricates
for _, aff in ipairs(affricates) do
for _, aff in ipairs(affricates) do
term = gsub(term, aff[1], aff[2])
term = gsub(term, aff[1], aff[2])
Line 131: Line 120:
ipa = ipa .. line_format(separate_word(term), {'Standard'})
ipa = ipa .. line_format(separate_word(term), {'Standard'})
if export.crux(term, "n") ~= export.crux(term) then
if export.crux(term, "c") ~= export.crux(term) then
ipa = ipa .. "\n* "
ipa = ipa .. "\n* "
ipa = ipa .. line_format(separate_word(term, "n"), {'Northern'})
ipa = ipa .. line_format(separate_word(term, "n"), {'Northern'})
end
end
if export.crux(term, "c") ~= export.crux(term) then
if export.crux(term, "s") ~= export.crux(term) then
ipa = ipa .. "\n* "
ipa = ipa .. "\n* "
ipa = ipa .. line_format(separate_word(term, "c"), {'Central'})
ipa = ipa .. line_format(separate_word(term, "c"), {'Central'})
end
end
if export.crux(term, "s") ~= export.crux(term) then
if export.crux(term, "n") ~= export.crux(term) then
ipa = ipa .. "\n* "
ipa = ipa .. "\n* "
ipa = ipa .. line_format(separate_word(term, "s"), {'Southern'})
ipa = ipa .. line_format(separate_word(term, "s"), {'Southern'})

Latest revision as of 02:48, 20 July 2022



local sub = mw.ustring.sub
local find = mw.ustring.find
local match = mw.ustring.match
local gmatch = mw.ustring.gmatch
local gsub = mw.ustring.gsub
local u = mw.ustring.char
local split = mw.text.split
local gsplit = mw.text.gsplit


-- To avoid weird annoying cursor behavior
local GRAVE = u(0x0300) -- COMBINING GRAVE ACCENT ̀◌
local ACUTE = u(0x0301) -- COMBINING ACUTE ACCENT
local SYLL = u(0x0329) -- COMBINING VERTICAL LINE BELOW ̩◌
local VL = u(0x0325) -- COMBINING RING BELOW ̥◌
local AFFR = u(0x0361) -- COMBINING DOUBLE INVERTED BREVE ͡

local back_vowel = "aouɔǫō"
local front_vowel = "ieɛɪ"
local vowels = "[" .. back_vowel .. front_vowel .. "]"
local acuted = "[áéíóúýÉÓ]"
local oral = "áéíóúýÉÓaouɔieɛ"
local palatal = "[ʨjʎʃʒʥʤʧʦʣɕʑ]"
local sonorant = "[rl]"
local plosives = "[ptkbdg]"
local velar = "[ɡkx]"
local consonants = "[mnptkbdɡfvszxrljɲŋʋʎʨjʎʃʒʥʤʧʤʣʦɕʑ]"

local export = {}

local rules = {
	{"ai", "ɛ"}, {"aí", "É"}, {"au", "ɔ"}, {"aú", "Ó"}, {"è", "ɛ"}, {"ò", "ɔ"}, {"y", "ɪ"},
	{"dź", "ʥ"}, {"dž", "ʤ"}, {"c", "ʦ"}, {"ć", "ʨ"}, {"nj", "ɲ"}, {"č", "ʧ"}, {"ž", "ʒ"},  {"š", "ʃ"},  {"lj", "ʎ"},
	{"h", "x"}, {"g", "ɡ"}, {"ǫ", "o"}, {"ō", "o"}, {"sč", "ʃt͡ʃ"}, {"ou", "u"}, {"ś", "ɕ"}, {"ź", "ʑ"},
	
	{"(" .. vowels .. ")[lv]$", "%1ʋ"}, {"(" .. acuted .. ")[lv]$", "%1ʋ"},
	
	{"(" .. plosives .. ")[ptkbdg]$", "[ptkbdg]"},
	
	{"pt" , "t"}, {"pk" , "k"}, {"tp" , "p"}, {"tk" , "k"}, {"kp" , "p"}, {"kt" , "t"},
	{"bd" , "d"}, {"bg" , "g"}, {"db" , "b"}, {"dg" , "g"}, {"gb" , "b"}, {"gd" , "d"},
	
	{"(" .. consonants ..")([rŕ])(" .. consonants ..")", "%1ə%2" .. SYLL .. "%3"}, {"(" .. consonants ..")r$", "%1ər"},
	{"n(" .. velar .. ")", "ŋ%1"},
	{"p(" .. vowels .. ")", "pʰ%1"}, {"t(" .. vowels .. ")", "tʰ%1"}, {"k(" .. vowels .. ")", "kʰ%1"},
	{"p(" .. acuted .. ")", "pʰ%1"}, {"t(" .. acuted .. ")", "tʰ%1"}, {"k(" .. acuted .. ")", "kʰ%1"},
	
	{"(" .. palatal .. ")e", "%1ɛ"}, {"(" .. palatal .. ")é", "%1É"},
	{"(" .. palatal .. ")o", "%1ɔ"}, {"(" .. palatal .. ")ó", "%1Ó"},
	
	{"É", "ɛ" .. ACUTE}, {"Ó", "ɔ" .. ACUTE}, {"ý", "ɪ" .. ACUTE},
}

local affricates = {
	{"ʥ", "d͡ʑ"}, {"ʤ", "d͡ʒ"}, {"ʦ", "t͡s"}, {"ʨ", "t͡ɕ"}, {"ʧ", "t͡ʃ"}, {"ʣ", "d͡z"},
}

function export.crux(term, d)
	term = mw.ustring.lower(term)
	
	for _, rule in ipairs(rules) do
		term = gsub(term, rule[1], rule[2])
	end
	
	if d == "c" then term = gsub(term, "ʨ", "ʧ"); term = gsub(term, "ʥ", "ʤ") end
	if d == "s" then term = gsub(term, "ʨ", "ʦ"); term = gsub(term, "ʥ", "ʣ"); term = gsub(term, "ʰ", "") end
	if d == "s" then term = gsub(term, "aj", "ɛː"); term =gsub(term, "ej", "ɛː") end
	if d == "c" then term = gsub(term, "aj", "ɛː"); term =gsub(term, "ej", "ɛː") end
	if d == "c" then term = gsub(term, "a", "ɐ"); term = gsub(term, "e", "ɪ"); term = gsub(term, "i", "ɪ"); term = gsub(term, "o", "ɔ"); term = gsub(term, "u", "ʊ")  end
	if d == "s" then term = gsub(term, "a", "ɐ"); term = gsub(term, "e", "ɛ"); term = gsub(term, "i", "e"); term = gsub(term, "o", "ɔ"); term = gsub(term, "u", "o"); term = gsub(term, "ɪ", "i"); term = gsub(term, "ɪ́", "í"); term = gsub(term, "ų", "u")  end
	if d == "n" then term = gsub(term, "a", "ə"); term = gsub(term, "e", "ə"); term = gsub(term, "i", "y"); term = gsub(term, "o", "ɔ"); term = gsub(term, "u", "ʊ"); term = gsub(term, "ɪ", "ɨ"); term = gsub(term, "ɪ́", "ɨ́"); term = gsub(term, "y", "ɪ")  end

	
	for _, aff in ipairs(affricates) do
		term = gsub(term, aff[1], aff[2])
	end
	
	return term
end


function IPA_span(items)
	local bits = {}
	for _, item in ipairs(items) do
		local bit = "<span style=\"font-size:110%;font-family:'Gentium','DejaVu Sans','Segoe UI',sans-serif>" .. item.pron .. "</span>"
		table.insert(bits, bit)
	end
	return table.concat(bits)
end

function format_IPA(items)
	return "[[w:IPA chart|IPA]]<sup>([[IPA for Zemljask|key]])</sup>:&#32;" .. IPA_span(items)
end

function line_format(pronunciation, dialect)
	local full_pronunciations = {}
	local IPA_args = {{pron = '[' .. pronunciation .. ']'}}
	table.insert(full_pronunciations, format_IPA(IPA_args))
	return "(''" .. table.concat(dialect, ", ") .. "'')" .. ' ' .. table.concat(full_pronunciations, ' or ')
end

function separate_word(term, d)
	local result = {}
	
	for word in gsplit(term, " ") do
		table.insert(result, export.crux(word, d))
	end
	
	return table.concat(result, " ")
end

function export.show(frame)
	local params = {
		[1] = { default = mw.title.getCurrentTitle().nsText == 'Template' and "dźélo" or mw.title.getCurrentTitle().text },	
	}
	local args = require("Module:parameters").process(frame:getParent().args, params)
	local term = args[1]
	
	local ipa = "* "
	ipa = ipa .. line_format(separate_word(term), {'Standard'})
	
	if export.crux(term, "c") ~= export.crux(term) then
		ipa = ipa .. "\n* "
		ipa = ipa .. line_format(separate_word(term, "n"), {'Northern'})
	end
	
	if export.crux(term, "s") ~= export.crux(term) then
		ipa = ipa .. "\n* "
		ipa = ipa .. line_format(separate_word(term, "c"), {'Central'})
	end
		
	if export.crux(term, "n") ~= export.crux(term) then
		ipa = ipa .. "\n* "
		ipa = ipa .. line_format(separate_word(term, "s"), {'Southern'})
	end
	
	return ipa
end

return export