imported>Ahecht: expand accent replacement

2025-10-28T17:55:28Z

expand accent replacement

New page

-- to enable us to replicate the current functioning of Country extract, we need to deal with:
-- 2 {{<name>}} DONE!
-- 3 [[<name>]] DONE!
-- 4 [[<name>|<junk>]] DONE!
-- 5 [[image:flag of <country>.[svg|gif|png|jpg]|\d+px]] DONE!

local p = {}
local getArgs = require("Module:Arguments").getArgs
local data = mw.loadData("Module:ISO 3166/data/National")

--[[----------C A T S----------]]-- -- Assign error categories
local function cats(cat, args)
if not (args.nocat and args.nocat == 'true') and mw.title.getCurrentTitle().namespace == 0 then
if cat == 'nocountry' then
return '[[Category:Wikipedia articles with obscure country]]'
elseif cat == 'nosubdivision' then
return '[[Category:Wikipedia articles with obscure subdivision]]'
end
end
return ''
end

--[[----------F I N D N A M E----------]]-- -- Finds the name in the database

local function findname(code,cdata,qry)
local sqry = p.strip(qry)
if cdata["name"] and sqry==p.strip(cdata["name"])
or cdata["isoname"] and sqry==p.strip(cdata["isoname"])
or not cdata["nocode"] and sqry==code
or sqry==cdata["alpha3"] or sqry==cdata["numeric"]
then
return true
end
for _,tname in pairs(cdata["isonames"] or {}) do
if sqry==p.strip(tname) then
return true
end
end
for _,tname in pairs(cdata["altnames"] or {}) do
if sqry==p.strip(tname) then
return true
end
end
return false
end

--[[----------I S O N A M E----------]]-- -- Find the ISO name of a country/region

local function isoname(data,code,lang)
if data[code]["isonames"] then
local name = data[code]["isodisplaynames"] and data[code]["isodisplaynames"][lang]
or data[code]["isonames"][lang]
or data[code]["isodisplaynames"] and data[code]["isodisplaynames"][data[code]["defaultlang"] or data["defaultlang"]]
or data[code]["isonames"][data[code]["defaultlang"] or data["defaultlang"]]
or data[code]["isodisplaynames"] and data[code]["isodisplaynames"]["en"]
or data[code]["isonames"]["en"]
if name then return name end
for _,iname in pairs(data[code]["isonames"]) do return iname end
return data[code]["isodisplayname"] or data[code]["isoname"]
else
return data[code]["isodisplayname"] or data[code]["isoname"]
end
end

--[[----------S T R I P----------]]-- -- Removes junk from the input

function p.strip(text)
if not text then return nil end

local accents = {["À"]="A",["Á"]="A",["Â"]="A",["Ã"]="A",["Ä"]="A",
["Å"]="A",["Æ"]="AE",["Ç"]="C",["È"]="E",["É"]="E",["Ê"]="E",["Ë"]="E",
["Ì"]="I",["Í"]="I",["Î"]="I",["Ï"]="I",["Ð"]="D",["Ñ"]="N",["Ò"]="O",
["Ó"]="O",["Ô"]="O",["Õ"]="O",["Ö"]="O",["Ø"]="O",["Ù"]="U",["Ú"]="U",
["Û"]="U",["Ü"]="U",["Ý"]="Y",["Þ"]="TH",["Ā"]="A",["Ă"]="A",["Ą"]="A",
["Ć"]="C",["Ĉ"]="C",["Ċ"]="C",["Č"]="C",["Ď"]="D",["Đ"]="D",["Ē"]="E",
["Ĕ"]="E",["Ė"]="E",["Ę"]="E",["Ě"]="E",["Ĝ"]="G",["Ğ"]="G",["Ġ"]="G",
["Ģ"]="G",["Ĥ"]="H",["Ħ"]="H",["Ĩ"]="I",["Ī"]="I",["Ĭ"]="I",["Į"]="I",
["İ"]="I",["Ĳ"]="IJ",["Ĵ"]="J",["Ķ"]="K",["Ĺ"]="L",["Ļ"]="L",["Ľ"]="L",
["Ŀ"]="L",["Ł"]="L",["Ń"]="N",["Ņ"]="N",["Ň"]="N",["Ŋ"]="N",["Ō"]="O",
["Ŏ"]="O",["Ő"]="O",["Œ"]="OE",["Ŕ"]="R",["Ŗ"]="R",["Ř"]="R",["Ś"]="S",
["Ŝ"]="S",["Ş"]="S",["Š"]="S",["Ţ"]="T",["Ť"]="T",["Ŧ"]="T",["Ũ"]="U",
["Ū"]="U",["Ŭ"]="U",["Ů"]="U",["Ű"]="U",["Ų"]="U",["Ŵ"]="W",["Ŷ"]="Y",
["Ÿ"]="Y",["Ź"]="Z",["Ż"]="Z",["Ž"]="Z",["Ɓ"]="B",["Ƃ"]="B",["Ɔ"]="C",
["Ƈ"]="C",["Ɖ"]="D",["Ɗ"]="D",["Ƌ"]="D",["Ǝ"]="E",["Ə"]="E",["Ɛ"]="E",
["Ƒ"]="F",["Ɠ"]="G",["Ɣ"]="G",["Ɩ"]="I",["Ɨ"]="I",["Ƙ"]="K",["Ɯ"]="M",
["Ɲ"]="N",["Ɵ"]="O",["Ơ"]="O",["Ƣ"]="OI",["Ƥ"]="P",["Ʃ"]="SH",["Ƭ"]="T",
["Ʈ"]="T",["Ư"]="U",["Ʊ"]="U",["Ʋ"]="V",["Ƴ"]="Y",["Ƶ"]="Z",["Ʒ"]="Z",
["Ƹ"]="Z",["Ǆ"]="DZ",["Ǉ"]="LJ",["Ǌ"]="NJ",["Ǎ"]="A",["Ǐ"]="I",
["Ǒ"]="O",["Ǔ"]="U",["Ǖ"]="U",["Ǘ"]="U",["Ǚ"]="U",["Ǜ"]="U",["Ǟ"]="A",
["Ǡ"]="A",["Ǣ"]="AE",["Ǥ"]="G",["Ǧ"]="G",["Ǩ"]="K",["Ǫ"]="O",["Ǭ"]="O",
["Ǯ"]="Z",["Ǳ"]="DZ",["Ǵ"]="G",["Ƕ"]="HV",["Ƿ"]="W",["Ǹ"]="N",["Ǻ"]="A",
["Ǽ"]="AE",["Ǿ"]="O",["Ȁ"]="A",["Ȃ"]="A",["Ȅ"]="E",["Ȇ"]="E",["Ȉ"]="I",
["Ȋ"]="I",["Ȍ"]="O",["Ȏ"]="O",["Ȑ"]="R",["Ȓ"]="R",["Ȕ"]="U",["Ȗ"]="U",
["Ș"]="S",["Ț"]="T",["Ȝ"]="Z",["Ȟ"]="H",["Ƞ"]="N",["Ȣ"]="OU",["Ȥ"]="Z",
["Ȧ"]="A",["Ȩ"]="E",["Ȫ"]="O",["Ȭ"]="O",["Ȯ"]="O",["Ȱ"]="O",["Ȳ"]="Y",
["Ⱥ"]="A",["Ȼ"]="C",["Ƚ"]="L",["Ⱦ"]="T",["Ƀ"]="B",["Ʉ"]="U",["Ʌ"]="V",
["Ɇ"]="E",["Ɉ"]="J",["Ɋ"]="Q",["Ɍ"]="R",["Ɏ"]="Y",
}
local remove = {"NATION OF","COUNTRY OF","TERRITORY OF", -- text to be removed list
"FLAG OF","FLAG","KINGDOM OF","STATE OF"," STATE ",
"PROVINCE OF","PROVINCE","TERRITORY","COUNTY","GOVERNORATE"
}
local patterns = {"(%a)%.(%a)%.",".+:","|.+","%(.-%)", -- patterns to follow (order may matter)
"%.%S.*","^THE ","%_","%-","%d%d?%d?PX"
}
local replacements = {["(%a)%.(%a)%."]="%1%2", -- Replacements for patterns above (order does not matter)
[".+:"]="",["|.+"]="",["%(.-%)"]="",
["%.%S.*"]="",["^THE "]="",["%_"]=" ",["%-"]=" ",
["%d%d?%d?PX"]="",
}

text = mw.text.decode(text) -- Decode HTML entities (passed from wikidata?)
text = mw.text.unstripNoWiki(text) -- Remove <nowiki> tags
text = mw.ustring.upper(text) -- Case insensitivity
text = mw.ustring.gsub(text,"[À-Ɏ]",accents) -- Deaccent

for _,pattern in ipairs(patterns) do -- Follow patterns
local value = replacements[pattern]
text = mw.ustring.gsub(text,pattern,value)
end

for _,words in pairs(remove) do -- Remove unneeded words
text = mw.ustring.gsub(text,words,"")
end

text = mw.ustring.gsub(text,"%W","") -- Remove non alpha-numeric

return text

end

--[[----------P . C A L L S T R I P ---------]]-- -- Calls P.strip but using Module:Arguments

function p.callstrip(frame)

local args = getArgs(frame)

return p.strip(args[1]) or ""

end

--[[----------P . L U A C O D E---------]]-- -- Makes the ISO code of a country

function p.luacode(args)
args[1], args[2] = args[1] or '', args[2] or ''
local code1 = p.strip(args[1])
local code2 = p.strip(args[2])

if args["codetype"]=="3" then args["codetype"]="alpha3" end

local eot = args.error or ""

if code1 == "" then
mw.addWarning((args[1] ~= "" and ('Invalid parameter "' .. args[1] .. '"') or 'No parameter') ..
' for the country given ([[Module:ISO 3166]]).')
return cats('nocountry', args), '<span style="font-size:100%" class="error">"No parameter for the country given"</span>'
end

local notRecognized ='%s ("%s") is not a recognized %s in [[ISO 3166-%s]] ([[Module:ISO 3166]]).'

if code2 == "" then --3166-1 code
for alpha2,cdata in pairs(data) do
if findname(alpha2,cdata,code1) then
if args["codetype"]=="numeric" or args["codetype"]=="alpha3" then
return cdata[args["codetype"]]
else
return alpha2
end
end
end
if string.find(args[1],"%-") then
args[1], args[2] = string.match(args[1],"^([^%-]*)%-(.*)$")
if args[2] then return p.luacode(args) end
end
mw.addWarning(string.format(notRecognized, code1, args[1], 'country', '1'))
return cats('nocountry', args)
else --3166-2 code
for alpha2,cdata in pairs(data) do
if findname(alpha2,cdata,code1) then
if mw.ustring.match(alpha2,"GB") then -- For England, Wales etc.
alpha2 = "GB"
end
local sdata = mw.loadData("Module:ISO 3166/data/"..alpha2)
local empty = true
for scode,scdata in pairs(sdata) do
if type(scdata)=="table" then
empty = false
if findname(scode,scdata,code2) then
return alpha2.."-"..scode
end
end
end
if string.find(args[2] or '',"%-") then
code1, args[2] = string.match(args[2] or '',"^([^%-]*)%-(.*)$")
args[1] = args[1].."-"..code1
return p.luacode(args)
end
mw.addWarning(string.format(notRecognized, code2, args[2], 'subdivision', '2:' .. alpha2))
return cats('nosubdivision', args)
end
end
if string.find(args[1],"%-") then
args[1] = string.match(args[1],"^([^%-]*)%-(.*)$")
return p.luacode(args)
end
mw.addWarning(string.format(notRecognized, code1, args[1], 'country', '1'))
return cats('nocountry', args)
end
mw.addWarning(string.format(notRecognized, code1, args[1], 'country', '1'))
return cats('nocountry', args)
end

--[[----------P . C O D E---------]]-- -- Calls P.Luacode but using Module:Arguments

function p.code(frame)

return p.luacode(getArgs(frame)) or ""

end

--[[----------P . N U M E R I C---------]]-- -- Calls P.Luacode but using Module:Arguments and setting it to output a numeric value

function p.numeric(frame)

local args = getArgs(frame)

args["codetype"]="numeric"

return p.luacode(args) or ""

end

--[[----------P . L U A N A M E---------]]-- -- Makes the ISO/common name of a country

function p.luaname(args)

local code1 = p.luacode(args)
local code2 = ''

if string.find(code1,"%-") then
code1, code2 = string.match(code1,"^([^%-]*)%-(.*)$")
end

if string.find(code1,"^%u%u$") then
if code2=="" then --3166-1 alpha-2 code
if data[code1] then
return (args.isoname or args.lang) and isoname(data,code1,args.lang)
or (data[code1]["displayname"] or data[code1]["name"])
else
return cats('nocountry', args)
end
else --3166-2 code
local sdata
if data[code1] then
sdata = mw.loadData("Module:ISO 3166/data/"..code1)
else
return cats('nocountry', args)
end
if sdata[code2] then
return (args.isoname or args.lang) and isoname(sdata,code2,args.lang)
or (sdata[code2]["displayname"] or sdata[code2]["name"])
else
return cats('nocountry', args)
end
end
elseif args.notfound then
return args.notfound
end

end

--[[----------P . N A M E---------]]-- -- Calls P.Luaname but using Module:Arguments

function p.name(frame)

return p.luaname(getArgs(frame)) or ""

end

--[[----------P . G E O C O O R D I N S E R T---------]]-- -- Wrapper for Module:Coordinates.coordinsert
function p.geocoordinsert(frame)
-- {{#invoke:ISO 3166|geocoordinsert|{{coord|...}}
-- |country=..|subdivision1=...|subdivision2=...
-- |type=...|scale=...|dim=...|source=...|globe=...
-- }}
local args = frame.args
local subdivisionqueried = false
local tracking = ''
local targs = {}
targs[1] = args[1] or ''
for i, v in pairs(args) do
if i == 'country' and not mw.ustring.find(targs[1], 'region:') then
local country = v
local k, region = 1, ''
-- look for a valid subdivision
while region == '' and k < 3 do
local subdivision = args['subdivision' .. k] or ''
if subdivision ~= '' then
region = p.luacode({country, subdivision, nocat = 'true'})
subdivisionqueried = true
end
k = k + 1
end
-- subdivision lookup failed or never attempted, try country only
if region == '' then
region = p.luacode({country, nocat = 'true'})
if region == '' then
tracking = tracking .. cats('nocountry', args)
elseif subdivisionqueried == true then
tracking = tracking .. cats('nosubdivision', args)
end
end
-- something worked, add it to the targs
if region ~= '' then
targs[#targs + 1] = 'region:' .. region
end
elseif i == 'type' or i == 'scale' or i == 'dim'
or i == 'source' or i == 'globe' then
targs[#targs + 1] = i .. ':' .. v
end
end
-- call Module:Coordinates.coordinsert if there is something to insert
if #targs > 1 then
local coordinsert = require('Module:Coordinates').coordinsert
return coordinsert({args = targs}) .. tracking
end
-- otherwise, just return the coordinates
return targs[1] .. tracking
end

return p

Module:ISO 3166 - Revision history

imported>Ahecht: expand accent replacement