Content deleted Content added
fix replace order, add accented characters |
sync |
||
Line 80:
["Ğ"]="G",["Ģ"]="G",["Ĝ"]="G",["Ġ"]="G",["Ĥ"]="H",
["Ħ"]="H",["Ĭ"]="I",["İ"]="I",["Ī"]="I",["Į"]="I",
["Ĩ"]="I",["IJ"]="
["Ľ"]="L",["Ļ"]="L",["Ŀ"]="L",["Ł"]="L",["Ń"]="N",
["Ň"]="N",["Ņ"]="N",["Ŏ"]="O",["Ő"]="O",["Ō"]="O",
["Œ"]="
["Š"]="S",["Ş"]="S",["Ŝ"]="S",["Ť"]="T",["Ţ"]="T",
["Ŧ"]="T",["Ŭ"]="U",["Ű"]="U",["Ū"]="U",["Ų"]="U",
Line 91:
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
}
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"]="",
}
Line 118:
text = mw.ustring.gsub(text,"%W","") -- Remove non alpha-numeric
return
end
|