Bước tới nội dung

Mô đun:Latin

Bách khoa toàn thư mở Wikipedia
local p = {}

function p.removeaccents(frame)
   local rv= mw.ustring.toNFC (frame.args[1])

--   if (true) then return mw.ustring.isutf8 (rv) end

   rv = mw.ustring.gsub(rv,"[ÁÀÂÄǍĂĀÃÅĄẢẠẮẰẲẴẶẤẦẨẪẬ]","A")
   rv = mw.ustring.gsub(rv,"[ÆǢǼ]","Ae")
   rv = mw.ustring.gsub(rv,"[ĆĊĈČÇ]","C")
   rv = mw.ustring.gsub(rv,"[ĎĐḌÐḐ]","D")
   rv = mw.ustring.gsub(rv,"[ÉÈĖÊËĚĔĒẼĘẸƐƎƏẺỀỂỄẾỆ]","E")
   rv = mw.ustring.gsub(rv,"[ĠĜĞĢ]","G")
   rv = mw.ustring.gsub(rv,"[ĤĦḤ]","H")
   rv = mw.ustring.gsub(rv,"[İÍÌÎÏǏĬĪĨĮỊỈ]","I")
   rv = mw.ustring.gsub(rv,"[Ĵ]","J")
   rv = mw.ustring.gsub(rv,"[Ķ]","K")
   rv = mw.ustring.gsub(rv,"[ĹĿĽĻŁḶḸ]","L")
   rv = mw.ustring.gsub(rv,"[Ṃ]","M")
   rv = mw.ustring.gsub(rv,"[ŃŇÑŅṆŊ]","N")
   rv = mw.ustring.gsub(rv,"[ÓÒÔÖǑŎŌÕǪỌŐØƆƠỎỐỒỔỖỘỚỜỞỠỢ]","O")
   rv = mw.ustring.gsub(rv,"[Œ]","Oe")
   rv = mw.ustring.gsub(rv,"[ŔŘŖṚṜ]","R")
   rv = mw.ustring.gsub(rv,"[ŚŜŠŞȘṢ]","S")
   rv = mw.ustring.gsub(rv,"[ŤŢȚṬ]","T")
   rv = mw.ustring.gsub(rv,"[Þ]","Th")
   rv = mw.ustring.gsub(rv,"[ÚÙÛÜǓŬŪŨŮŲỤŰǗǛǙǕƯỦỨỪỬỮỰ]","U")
   rv = mw.ustring.gsub(rv,"[Ŵ]","W")
   rv = mw.ustring.gsub(rv,"[ÝŶŸỸȲ]","Y")
   rv = mw.ustring.gsub(rv,"[ŹŻŽ]","Z")
   rv = mw.ustring.gsub(rv,"[áàâäǎăāãåąảạắằẳẵặấầẩẫậ]","a")
   rv = mw.ustring.gsub(rv,"[æǣǽ]","ae")
   rv = mw.ustring.gsub(rv,"[ćċĉčç]","c")
   rv = mw.ustring.gsub(rv,"[ďđḍðḑ]","d")
   rv = mw.ustring.gsub(rv,"[éèėêëěĕēẽęẹɛǝẻếềểễệ]","e")
   rv = mw.ustring.gsub(rv,"[ġĝğģ]","g")
   rv = mw.ustring.gsub(rv,"[ĥħḥḩ]","h")
   rv = mw.ustring.gsub(rv,"[ıíìîïǐĭīĩįịỉ]","i")
   rv = mw.ustring.gsub(rv,"[ĵ]","j")
   rv = mw.ustring.gsub(rv,"[ķ]","k")
   rv = mw.ustring.gsub(rv,"[ĺŀľļłḷḹ]","l")
   rv = mw.ustring.gsub(rv,"[ṃ]","m")
   rv = mw.ustring.gsub(rv,"[ńňñņṇŋ]","n")
   rv = mw.ustring.gsub(rv,"[óòôöǒŏōõǫọőøɔơỏốồổỗộớờởỡợ]","o")
   rv = mw.ustring.gsub(rv,"[œ]","oe")
   rv = mw.ustring.gsub(rv,"[ŕřŗṛṝ]","r")
   rv = mw.ustring.gsub(rv,"[śŝšşșṣ]","s")
   rv = mw.ustring.gsub(rv,"[ß]","ss")
   rv = mw.ustring.gsub(rv,"[ťţțṭ]","t")
   rv = mw.ustring.gsub(rv,"[þ]","th")
   rv = mw.ustring.gsub(rv,"[úùûüǔŭūũůųụűǘǜǚǖưủứừửữự]","u")
   rv = mw.ustring.gsub(rv,"[ŵ]","w")
   rv = mw.ustring.gsub(rv,"[ýŷÿỹȳ]","y")
   rv = mw.ustring.gsub(rv,"[źżž]","z")

   return rv
end

--[[

The next function returns a % encoding concomitant with ISO/IEC 8859-1. It encodes % and any non-reserved and non-unreserved
characters.

Space is currently encoded as an underscore.

Reserved characters are currently not encoded.

]]

function p.urlencodeISO88591(frame)
   local rv= mw.ustring.toNFC (frame.args[1])

-- % first to avoid double encoding
   rv = mw.ustring.gsub(rv,"%%","%%%%")

-- space to underscore, then punctuation which is not reserved
   rv = mw.ustring.gsub(rv," ","_")
   rv = mw.ustring.gsub(rv,"\034;","%%22") -- quote mark
   rv = mw.ustring.gsub(rv,"<","%%3C")
   rv = mw.ustring.gsub(rv,">","%%3E")
   rv = mw.ustring.gsub(rv,"\\","%%5C") -- backslash does not work with "\092", "%\" or "\"
   rv = mw.ustring.gsub(rv,"%^","%%5E")
   rv = mw.ustring.gsub(rv,"`","%%60")
   rv = mw.ustring.gsub(rv,"{","%%7B")
   rv = mw.ustring.gsub(rv,"|","%%7C")
   rv = mw.ustring.gsub(rv,"}","%%7D")

-- all the rest of the codepoints that are printable
   rv = mw.ustring.gsub(rv,"\194\160","%%A0")
   rv = mw.ustring.gsub(rv,"\194\161","%%A1")
   rv = mw.ustring.gsub(rv,"\194\162","%%A2")
   rv = mw.ustring.gsub(rv,"\194\163","%%A3")
   rv = mw.ustring.gsub(rv,"\194\164","%%A4")
   rv = mw.ustring.gsub(rv,"\194\165","%%A5")
   rv = mw.ustring.gsub(rv,"\194\166","%%A6")
   rv = mw.ustring.gsub(rv,"\194\167","%%A7")
   rv = mw.ustring.gsub(rv,"\194\168","%%A8")
   rv = mw.ustring.gsub(rv,"\194\169","%%A9")
   rv = mw.ustring.gsub(rv,"\194\170","%%AA")
   rv = mw.ustring.gsub(rv,"\194\171","%%AB")
   rv = mw.ustring.gsub(rv,"\194\172","%%AC")
   rv = mw.ustring.gsub(rv,"\194\173","%%AD")
   rv = mw.ustring.gsub(rv,"\194\174","%%AE")
   rv = mw.ustring.gsub(rv,"\194\175","%%AF")
   rv = mw.ustring.gsub(rv,"\194\176","%%B0")
   rv = mw.ustring.gsub(rv,"\194\177","%%B1")
   rv = mw.ustring.gsub(rv,"\194\178","%%B2")
   rv = mw.ustring.gsub(rv,"\194\179","%%B3")
   rv = mw.ustring.gsub(rv,"\194\180","%%B4")
   rv = mw.ustring.gsub(rv,"\194\181","%%B5")
   rv = mw.ustring.gsub(rv,"\194\182","%%B6")
   rv = mw.ustring.gsub(rv,"\194\183","%%B7")
   rv = mw.ustring.gsub(rv,"\194\184","%%B8")
   rv = mw.ustring.gsub(rv,"\194\185","%%B9")
   rv = mw.ustring.gsub(rv,"\194\186","%%BA")
   rv = mw.ustring.gsub(rv,"\194\187","%%BB")
   rv = mw.ustring.gsub(rv,"\194\188","%%BC")
   rv = mw.ustring.gsub(rv,"\194\189","%%BD")
   rv = mw.ustring.gsub(rv,"\194\190","%%BE")
   rv = mw.ustring.gsub(rv,"\194\191","%%BF")
   rv = mw.ustring.gsub(rv,"\195\128","%%C0")
   rv = mw.ustring.gsub(rv,"\195\129","%%C1")
   rv = mw.ustring.gsub(rv,"\195\130","%%C2")
   rv = mw.ustring.gsub(rv,"\195\131","%%C3")
   rv = mw.ustring.gsub(rv,"\195\132","%%C4")
   rv = mw.ustring.gsub(rv,"\195\133","%%C5")
   rv = mw.ustring.gsub(rv,"\195\134","%%C6")
   rv = mw.ustring.gsub(rv,"\195\135","%%C7")
   rv = mw.ustring.gsub(rv,"\195\136","%%C8")
   rv = mw.ustring.gsub(rv,"\195\137","%%C9")
   rv = mw.ustring.gsub(rv,"\195\138","%%CA")
   rv = mw.ustring.gsub(rv,"\195\139","%%CB")
   rv = mw.ustring.gsub(rv,"\195\140","%%CC")
   rv = mw.ustring.gsub(rv,"\195\141","%%CD")
   rv = mw.ustring.gsub(rv,"\195\142","%%CE")
   rv = mw.ustring.gsub(rv,"\195\143","%%CF")
   rv = mw.ustring.gsub(rv,"\195\144","%%D0")
   rv = mw.ustring.gsub(rv,"\195\145","%%D1")
   rv = mw.ustring.gsub(rv,"\195\146","%%D2")
   rv = mw.ustring.gsub(rv,"\195\147","%%D3")
   rv = mw.ustring.gsub(rv,"\195\148","%%D4")
   rv = mw.ustring.gsub(rv,"\195\149","%%D5")
   rv = mw.ustring.gsub(rv,"\195\150","%%D6")
   rv = mw.ustring.gsub(rv,"\195\151","%%D7")
   rv = mw.ustring.gsub(rv,"\195\152","%%D8")
   rv = mw.ustring.gsub(rv,"\195\153","%%D9")
   rv = mw.ustring.gsub(rv,"\195\154","%%DA")
   rv = mw.ustring.gsub(rv,"\195\155","%%DB")
   rv = mw.ustring.gsub(rv,"\195\156","%%DC")
   rv = mw.ustring.gsub(rv,"\195\157","%%DD")
   rv = mw.ustring.gsub(rv,"\195\158","%%DE")
   rv = mw.ustring.gsub(rv,"\195\159","%%DF")
   rv = mw.ustring.gsub(rv,"\195\160","%%E0")
   rv = mw.ustring.gsub(rv,"\195\161","%%E1")
   rv = mw.ustring.gsub(rv,"\195\162","%%E2")
   rv = mw.ustring.gsub(rv,"\195\163","%%E3")
   rv = mw.ustring.gsub(rv,"\195\164","%%E4")
   rv = mw.ustring.gsub(rv,"\195\165","%%E5")
   rv = mw.ustring.gsub(rv,"\195\166","%%E6")
   rv = mw.ustring.gsub(rv,"\195\167","%%E7")
   rv = mw.ustring.gsub(rv,"\195\168","%%E8")
   rv = mw.ustring.gsub(rv,"\195\169","%%E9")
   rv = mw.ustring.gsub(rv,"\195\170","%%EA")
   rv = mw.ustring.gsub(rv,"\195\171","%%EB")
   rv = mw.ustring.gsub(rv,"\195\172","%%EC")
   rv = mw.ustring.gsub(rv,"\195\173","%%ED")
   rv = mw.ustring.gsub(rv,"\195\174","%%EE")
   rv = mw.ustring.gsub(rv,"\195\175","%%EF")
   rv = mw.ustring.gsub(rv,"\195\176","%%F0")
   rv = mw.ustring.gsub(rv,"\195\177","%%F1")
   rv = mw.ustring.gsub(rv,"\195\178","%%F2")
   rv = mw.ustring.gsub(rv,"\195\179","%%F3")
   rv = mw.ustring.gsub(rv,"\195\180","%%F4")
   rv = mw.ustring.gsub(rv,"\195\181","%%F5")
   rv = mw.ustring.gsub(rv,"\195\182","%%F6")
   rv = mw.ustring.gsub(rv,"\195\183","%%F7")
   rv = mw.ustring.gsub(rv,"\195\184","%%F8")
   rv = mw.ustring.gsub(rv,"\195\185","%%F9")
   rv = mw.ustring.gsub(rv,"\195\186","%%FA")
   rv = mw.ustring.gsub(rv,"\195\187","%%FB")
   rv = mw.ustring.gsub(rv,"\195\188","%%FC")
   rv = mw.ustring.gsub(rv,"\195\189","%%FD")
   rv = mw.ustring.gsub(rv,"\195\190","%%FE")
   rv = mw.ustring.gsub(rv,"\195\191","%%FF")
   return rv
end

return p