#!/bin/rc
unicodedata=UnicodeData.txt

for(i)
	unicodedata=$i

fn Unicode {
	grep $rune < $unicodedata | tr -d '\015'
}

Unicode | awk -F^';' '
function s(x){
	if(x == "001C")
		return "file seperator"
	if(x == "001D")
		return "group separator"
	if(x == "001E")
		return "record separator"
	if(x == "001F")
		return "unit separator"
	return ""
}

{
	a = s(x)
	if(length(a) > 0){
	}
	else
	if(!s($1) && length($2) > 0 && match($2, "<") == 0)
		a = $2
	else
	if(length($11) > 0)
		a = $11
	b=""
	if(length($13) > 0)
		b = ";" $13
	if(length(a)>0)
		print $1 "	" a b
}' |tr 'A-Z' 'a-z'  >unicode