Mòdul:it-general

Documentació del mòdul

A continuació es mostra la documentació transclosa de la subpàgina /ús. [salta a la caixa de codi]

---

-- Funcions generals per italià. Inacabat, en proves.

local export = {}

-- Clau d'ordenació per a categories en italià
local senseDiacritics = {
    ["À"] = "A", ["à"] = "a",
    ["È"] = "E", ["è"] = "e", ["É"] = "E", ["é"] = "e",
    ["Ì"] = "I", ["ì"] = "i",
    ["Ò"] = "O", ["ò"] = "o", ["Ó"] = "O", ["ó"] = "o",
    ["Ù"] = "U", ["ù"] = "u"}
local senseSignes = {
    ["/"] = "",
    ["-"] = "",
    ["."] = "",
    ["'"] = " "}
local ordreAccents = {  -- ordre eéè simulat amb ordre Unicode eèéê
    ["è"] = "ê",
    ["ò"] = "ô"}

function export.ordena(mot)
    if type(mot) == "table" then mot = mot.args[1] end -- des de plantilles via invoke o des de mòduls via require
    if mot == "" or mot == nil then
        mot = mw.title.getCurrentTitle().text
    end
    local clau           -- format complet: "clau1!clau2!clau3!mot"
    local clau1          -- clau primària: en minúscules i sense diacrítics ni signes
    local clau2          -- clau secundària: prioritat dels diacrítics eéè
    local clau3          -- clau terciària: prioritat minúscules > majúscules
                         -- clau quaternària: prioritat sense signes > amb signes
    local motSenseSignes = mw.ustring.gsub(mot, ".", senseSignes)
    local motLower = mw.ustring.lower(motSenseSignes)
    clau1 = mw.ustring.gsub(motLower, ".", senseDiacritics)
    if clau1 == mot then
        return clau1
    end
    clau2 = mw.ustring.gsub(motLower, ".", ordreAccents)
    clau3 = mw.ustring.gsub(motSenseSignes, ".", senseDiacritics)
    if clau1 == motLower then
        clau = clau1 .. "!"
    else
        clau = clau1 .. "!" .. clau2
    end
    if clau3 ~= clau1 then
        clau = clau .. "!" .. clau3
    elseif motSenseSignes ~= mot then
        clau = clau .. "!"
    end
    if motSenseSignes ~= mot then
        clau = clau .."!" .. mot
    end
    return clau
end

function export._ordena(mot)
    -- Funció que es pot usar en altres mòduls via require
    -- Pendent d'eliminar una vegada substituïda
    return export.ordena(mot)
end

--[[ 
Sil·labificació

    marcatge intern: vocals 0, obertures 1, codes 2
    síl·laba: ·(1*)0(2*)·
]]
function export.sil(mot, accent)
	if type(mot) == "table" then
		accent = mot.args[2]
		mot = mot.args[1]
	end
	if mot == "" or mot == nil then
		mot = mw.title.getCurrentTitle().text
	end
	-- Si no hi ha cap accent el posem
	if accent ~= "" and accent ~= nil then
		local from = mw.ustring.gsub(accent, ".", senseDiacritics)
		mot = mw.ustring.gsub(mot, from, accent, 1)
	end
	if not mw.ustring.find(mot, "[àèéìòóù]") then
		mot = export.accent(mot)
	end
	local sil = mw.ustring.lower(mot)
	sil = string.gsub(sil, "qu", "11")
	-- triftongs
	sil = mw.ustring.gsub(sil, "iu[aàeèéoòó]", "022")
	sil = mw.ustring.gsub(sil, "[iu][aàeèéoòó]i$", "022")
	-- diftongs creixents
	sil = mw.ustring.gsub(sil, "i[aàeèéoòóuù]", "02")
	sil = mw.ustring.gsub(sil, "u[aàeèéiìoòó]", "02")
	-- diftongs decreixents
	sil = mw.ustring.gsub(sil, "[aàeèé][iu]", "02")
	sil = mw.ustring.gsub(sil, "[oòóù]i", "02")
	-- Nuclis vocàlics
	sil = mw.ustring.gsub(sil, "[aàeèéiìoòóuù]", "0")
	-- Codes
	sil = mw.ustring.gsub(sil, "[lmnr]([^0])", "2%1")
	-- Obertures inseparables, digrammi
	sil = mw.ustring.gsub(sil, "[cg]h", "11")
	sil = mw.ustring.gsub(sil, "[bcfgġpst]l", "11")
	sil = mw.ustring.gsub(sil, "[st]m", "11")
	sil = mw.ustring.gsub(sil, "[gp]n", "11")
	sil = mw.ustring.gsub(sil, "[bcdfgpstv]r", "11")
	sil = mw.ustring.gsub(sil, "ps", "11")
	-- Obertures s impura
	sil = mw.ustring.gsub(sil, "s1", "11")
	sil = mw.ustring.gsub(sil, "ss", "21")
	sil = mw.ustring.gsub(sil, "s[^0]", "11")
	-- Obertures simples
	sil = mw.ustring.gsub(sil, "%l0", "10")
	-- Codes finals
	sil = mw.ustring.gsub(sil, "%l$", "2")
	sil = mw.ustring.gsub(sil, "%l2$", "22")
	sil = mw.ustring.gsub(sil, "%l22$", "222")
	-- Codes interiors
	sil = mw.ustring.gsub(sil, "%l([12])", "2%1")
	-- Separació de síl·labes
	local anterior = ""
	local motSil = {}
	for i = 1, mw.ustring.len(mot) do
		actual = mw.ustring.sub(sil,i,i)
		if (actual == "0" or actual == "1") and (anterior == "0" or anterior == "2") then
			table.insert(motSil, "·")
		end
		table.insert(motSil, mw.ustring.sub(mot,i,i))
		anterior = actual
	end
	return table.concat(motSil)
end

function export._sil(mot)
	-- Funció per usar en altres mòduls via require
    -- Pendent d'eliminar una vegada substituïda
    return export.sil(mot)
end

function export.accent(mot)
    if type(mot) == "table" then mot = mot.args[1] end -- des de plantilles via invoke o des de mòduls via require
    if mot == "" or mot == nil then
        mot = mw.title.getCurrentTitle().text
    end
    -- Casos é generals
    mot = mw.ustring.gsub(mot, "esero$", "ésero")
    mot = mw.ustring.gsub(mot, "erono$", "érono")
    mot = mw.ustring.gsub(mot, "essero$", "éssero")
    mot = mw.ustring.gsub(mot, "ecc(h?)i([ao])$", "écc%1i%2")
    mot = mw.ustring.gsub(mot, "ecc(h?[ei])$", "écc%1")
    mot = mw.ustring.gsub(mot, "ecc([ao])$", "écc%1")
    mot = mw.ustring.gsub(mot, "emplic([ei])$", "émplic%1")
    mot = mw.ustring.gsub(mot, "eggi([ao])$", "éggi%1")
    mot = mw.ustring.gsub(mot, "egg([ei])$", "égg%1")
    mot = mw.ustring.gsub(mot, "egn([aeio])$", "égn%1")
    mot = mw.ustring.gsub(mot, "egr([aeio])$", "égr%1")
    mot = mw.ustring.gsub(mot, "es([aeio])$", "és%1")
    mot = mw.ustring.gsub(mot, "esc(h?[aeio])$", "ésc%1")
    mot = mw.ustring.gsub(mot, "et([aeio])$", "ét%1")
    mot = mw.ustring.gsub(mot, "ett([aeio])$", "étt%1")
    mot = mw.ustring.gsub(mot, "evol([ei])$", "évol%1")
    mot = mw.ustring.gsub(mot, "evr([aeio])$", "évr%1")
    mot = mw.ustring.gsub(mot, "ezz([aeio])$", "ézz%1")
    mot = mw.ustring.gsub(mot, "ment([aeio])$", "mént%1")
    -- Casos ó generals
    mot = mw.ustring.gsub(mot, "oc([ei])$", "óc%1")
    mot = mw.ustring.gsub(mot, "ogn([aeio])$", "ógn%1")
    mot = mw.ustring.gsub(mot, "oi([aeio])$", "ói%1")
    mot = mw.ustring.gsub(mot, "omplic([aeio])$", "ómplic%1")
    mot = mw.ustring.gsub(mot, "osero$", "ósero")
    mot = mw.ustring.gsub(mot, "([^fu])on([aei])$", "%1ón%2")
    mot = mw.ustring.gsub(mot, "([^aeio])fon([aei])$", "%1fón%2")
    mot = mw.ustring.gsub(mot, "onci([aeio])$", "ónci%1")
    mot = mw.ustring.gsub(mot, "ont([aeio])$", "ónt%1")
    mot = mw.ustring.gsub(mot, "onz([aeio])$", "ónz%1")
    mot = mw.ustring.gsub(mot, "onzol([aeio])$", "ónzol%1")
    mot = mw.ustring.gsub(mot, "orci([ao])$", "órci%1")
    mot = mw.ustring.gsub(mot, "orc([ei])$", "órc%1")
    mot = mw.ustring.gsub(mot, "orn([aeio])$", "órn%1")
    mot = mw.ustring.gsub(mot, "or([ei])$", "ór%1")
    mot = mw.ustring.gsub(mot, "os([aeio])$", "ós%1")
    mot = mw.ustring.gsub(mot, "ozz([io])$", "ózz%1")
    mot = mw.ustring.gsub(mot, "ond([aeio])$", "ónd%1")
    -- Casos de estrúixoles generals
    mot = mw.ustring.gsub(mot, "a([bcdfglmnprsṣtvzẓ][bcdfghmnprstvzẓ]?l?)i([cdlmt]h?)([aeio])$", "à%1i%2%3")
    mot = mw.ustring.gsub(mot, "e([bcdfglmnpqrsṣtvzẓ][bcdfghmnprstvzẓ]?l?)i([cdlmt]h?)([aeio])$", "è%1i%2%3")
    mot = mw.ustring.gsub(mot, "i([bcdfglmnpqrsṣtvzẓ][bcdfghmnprstvzẓ]?l?)i([cdlmt]h?)([aeio])$", "ì%1i%2%3")
    mot = mw.ustring.gsub(mot, "o([bcdfglmnpqrsṣtvzẓ][bcdfghmnprstvzẓ]?l?)i([cdlmt]h?)([aeio])$", "ò%1i%2%3")
    mot = mw.ustring.gsub(mot, "u([bcdfglmnpqrsṣtvzẓ][bcdfghmnprstvzẓ]?l?)i([cdlmt]h?)([aeio])$", "ù%1i%2%3")
    mot = mw.ustring.gsub(mot, "acciol([aeio])$", "àcciol%1")
    mot = mw.ustring.gsub(mot, "a([sc]?)cher([aeio])$", "à%1cher%2")
    mot = mw.ustring.gsub(mot, "afo([bn][aeio])$", "àfo%1")
    mot = mw.ustring.gsub(mot, "ag(g?)in([ei])$", "àg%1in%2")
    mot = mw.ustring.gsub(mot, "arono$", "àrono")
    mot = mw.ustring.gsub(mot, "a([smc]?[bcgmnpt][pbndt]?)ol([aeio])$", "à%1ol%2")
    mot = mw.ustring.gsub(mot, "asser([aeio])$", "àsser%1")
    mot = mw.ustring.gsub(mot, "avol([aeio])$", "àvol%1")
    mot = mw.ustring.gsub(mot, "e([smc]?[bcgmnpt][pbndt]?)ol([aeio])$", "è%1ol%2")
    mot = mw.ustring.gsub(mot, "efon([io])$", "èfon%1")
    mot = mw.ustring.gsub(mot, "evol([ao])$", "èvol%1")
    mot = mw.ustring.gsub(mot, "i([sc]?)cher([aeio])$", "ì%1cher%2")
    mot = mw.ustring.gsub(mot, "icciol([aeio])$", "ìcciol%1")
    mot = mw.ustring.gsub(mot, "ie([dt])ere$", "iè%1ere")
    mot = mw.ustring.gsub(mot, "ifo([nr][aeio])$", "ìfo%1")
    mot = mw.ustring.gsub(mot, "i([fn]?)fer([aeio])$", "ì%1fer%2")
    mot = mw.ustring.gsub(mot, "i([ng]?)gere$", "ì%1gere")
    mot = mw.ustring.gsub(mot, "ig(g?)in([ei])$", "ìg%1in%2")
    mot = mw.ustring.gsub(mot, "irono$", "ìrono")
    mot = mw.ustring.gsub(mot, "i([smc]?[bcgmnpt][pbndt]?)ol([aeio])$", "ì%1ol%2")
    mot = mw.ustring.gsub(mot, "issero$", "ìssero")
    mot = mw.ustring.gsub(mot, "ivol([aeio])$", "ìvol%1")
    mot = mw.ustring.gsub(mot, "o([smc]?[bcgmnpt][pbndt]?)ol([aeio])$", "ò%1ol%2")
    mot = mw.ustring.gsub(mot, "o([sc]?)cher([aeio])$", "ò%1cher%2")
    mot = mw.ustring.gsub(mot, "or([dg])ere$", "òr%1ere")
    mot = mw.ustring.gsub(mot, "ossero$", "òssero")
    mot = mw.ustring.gsub(mot, "o(s?)fo([bnr][aeio])$", "ò%1fo%2")
    mot = mw.ustring.gsub(mot, "ovol([aeio])$", "òvol%1")
    mot = mw.ustring.gsub(mot, "u([sc]?)cher([aeio])$", "ù%1cher%2")
    mot = mw.ustring.gsub(mot, "ucciol([aeio])$", "ùcciol%1")
    mot = mw.ustring.gsub(mot, "udin([ei])$", "ùdin%1")
    mot = mw.ustring.gsub(mot, "ug(g?)in([ei])$", "ùg%1in%2")
    mot = mw.ustring.gsub(mot, "uotere$", "uòtere")
    mot = mw.ustring.gsub(mot, "u([smc]?[bcgmnpt][pbndt]?)ol([aeio])$", "ù%1ol%2")
    mot = mw.ustring.gsub(mot, "urono$", "ùrono")
    mot = mw.ustring.gsub(mot, "ussero$", "ùssero")
    mot = mw.ustring.gsub(mot, "uvol([aeio])$", "ùvol%1")
    mot = mw.ustring.gsub(mot, "uzzol([aeio])$", "ùzzol%1")
    if mw.ustring.find(mot, "[àèéìòóù]") then
        return mot
    end
    -- Altrament, oberta
    local accentuar = {["a"]="à", ["e"]="è", ["i"]="ì", ["o"]="ò", ["u"]="ù"}
    -- Vocal a accentuar, des del final
    local accent = 2
    if string.find(mot, "^[^aeiou]+[aeiou]+$") then -- monosíl·ab
    	if not (string.find(mot, "[ae][iu]$") or string.find(mot, "[ou]i$")) then -- ... sense diftong creixent
    		accent = 1
    	end
    elseif mw.ustring.find(mot, "i[ao]$") or mw.ustring.find(mot, "qu[aeio]$") then
        accent = 3
    elseif string.find(mot, "[ae][iu][^aeoiu]+[aeoiu]") or string.find(mot, "[ou]i[^aeoiu]+[aeoiu]") then -- diftong decreixent paroxíton
    	accent = 3
    elseif not mw.ustring.find(mot, "[aeiou]$") then
        accent = 1
    end
    local vocals = 0
    for i = mw.ustring.len(mot), 1, -1 do
        local lletra = mw.ustring.sub(mot, i, i)
        if mw.ustring.find(lletra, "[aeiou]") then
            vocals = vocals + 1
            if vocals == accent then
                local accentuat = mw.ustring.gsub(lletra, ".", accentuar) .. mw.ustring.sub(mot, i+1)
                if i>1 then
                    accentuat = mw.ustring.sub(mot, 1, i-1) .. accentuat
                end
                return accentuat
            end
        end
    end
    if not mw.ustring.find(mot, "[àèéìòóù]") then
        mot = mw.ustring.gsub(mot, "[aeiou]", accentuar, 1)
    end
    return mot
end

function export._accent(mot)
    -- Pendent d'eliminar una vegada substituïda
    return export.accent(mot)
end

-- Pronunciació
function export.pron(mot)
	if type(mot) == "table" then mot = mot.args[1] or mw.title.getCurrentTitle().text end
    mot = export._sil(mw.ustring.lower(mot))

    -- dígrafs
    mot = mw.ustring.gsub(mot, "ch", "k")
    mot = mw.ustring.gsub(mot, "qu", "kw")
    mot = mw.ustring.gsub(mot, "gn", "ɲ")
    -- consonants
    mot = mw.ustring.gsub(mot, "y", "j")
    mot = mw.ustring.gsub(mot, "q", "k")
    mot = mw.ustring.gsub(mot, "sci([aàoòóuù)])", "ʃ%1")
    mot = mw.ustring.gsub(mot, "sc([eèéiì])", "ʃ%1")
    mot = mw.ustring.gsub(mot, "ci([aàoòóuù])", "tʃ%1")
    mot = mw.ustring.gsub(mot, "c([eèéiì])", "tʃ%1")
    mot = mw.ustring.gsub(mot, "c·tʃ", "t·tʃ")
    mot = mw.ustring.gsub(mot, "c", "k")
    mot = mw.ustring.gsub(mot, "gi([aàoòóuù])", "dʒ%1")
    mot = mw.ustring.gsub(mot, "g([eèéiì])", "dʒ%1")
    mot = mw.ustring.gsub(mot, "g·dʒ", "d·dʒ")
    mot = mw.ustring.gsub(mot, "g([aàoòóuù])", "g%1")
    mot = mw.ustring.gsub(mot, "gh", "g")
    mot = mw.ustring.gsub(mot, "ġl", "ɡl")
    mot = mw.ustring.gsub(mot, "([^n])·gli([aàeèéoòóuù])", "%1·ʎ%2")
    mot = mw.ustring.gsub(mot, "([^n])·gl([iì])", "%1·ʎ%2")
    mot = mw.ustring.gsub(mot, "^gli([aàeèéoòóuù])", "ʎ%1")
    mot = mw.ustring.gsub(mot, "^gl([iì])", "ʎ%1")
    mot = mw.ustring.gsub(mot, "([aàeèéiìoòóuù])·ʎ", "%1ʎ·ʎ")
    mot = mw.ustring.gsub(mot, "n·gl([iì])", "ŋ·gl%1")
    mot = mw.ustring.gsub(mot, "([aàeèéiìoòóuù])·ɲ", "%1ɲ·ɲ")
    mot = mw.ustring.gsub(mot, "n·([kg])", "ŋ·%1")
    mot = mw.ustring.gsub(mot, "n([kg])", "ŋ%1")
    mot = mw.ustring.gsub(mot, "h", "")
    mot = mw.ustring.gsub(mot, "([mn])·([fv])", "ɱ·%2")
    mot = mw.ustring.gsub(mot, "([aàeèéiìoòóuù])·ʃ", "%1ʃ·ʃ")
    mot = mw.ustring.gsub(mot, "z·z", "t·ts")
    mot = mw.ustring.gsub(mot, "([aàeèéiìoòóuù])·z", "%1t·ts")
    mot = mw.ustring.gsub(mot, "z", "ts")
    mot = mw.ustring.gsub(mot, "ẓ·ẓ", "d·dz")
    mot = mw.ustring.gsub(mot, "([aàeèéiìoòóuù])·ẓ", "%1d·dz")
    mot = mw.ustring.gsub(mot, "ẓ", "dz")
    -- essa sonora o sorda
    mot = mw.ustring.gsub(mot, "s([bdglmnrv])", "z%1")
    mot = mw.ustring.gsub(mot, "ṣ", "z")
    mot = mw.ustring.gsub(mot, "x", "ks")
    -- vocals
    mot = mw.ustring.gsub(mot, "i([aàeèéoòóuù])", "j%1")
    mot = mw.ustring.gsub(mot, "i.([aàeèéoòóuù])", "i·%1")
    mot = mw.ustring.gsub(mot, "ɲj", "ɲ")
    mot = mw.ustring.gsub(mot, "u([aàeèéiìoòó])", "w%1")
    mot = mw.ustring.gsub(mot, "u.([aàeèéiìoòó])", "u·%1")
    -- accent
    if mw.ustring.find(mot, "[àèéìóòù]") then
        mot = mw.ustring.gsub(mot, "·(%l-[àèéìòóù])", "ˈ%1")
    else
        mot = mw.ustring.gsub(mot, "·(%l-)·(%l-)$", "ˈ%1·%2")
    end
    if not mw.ustring.find(mot, "ˈ") then -- monosíl·lab o accent en la primera
        mot = "ˈ" .. mot
    end
    mot = mw.ustring.gsub(mot, "è", "ɛ")
    mot = mw.ustring.gsub(mot, "ò", "ɔ")
    mot = mw.ustring.gsub(mot, ".", senseDiacritics)
    mot = mw.ustring.gsub(mot, "tʃ", "t͡ʃ")
    mot = mw.ustring.gsub(mot, "dʒ", "d͡ʒ")
    mot = mw.ustring.gsub(mot, "g", "ɡ")
    mot = mw.ustring.gsub(mot, "ts", "t͡s")
    mot = mw.ustring.gsub(mot, "dz", "d͡z")
    return '/' .. mw.ustring.gsub(mot, "·", ".") .. '/'
end

return export