#!/usr/bin/perl

# Badly named script. Converts non-ascii UTF-8 text into equivalent decimal
# entities for use in HTML and XML.

# Warning: not sure which versions of perl work. This GIVES WRONG ANSWERS on
# some computers I have access to, but does on others. I'd go with the python or
# C versions, they're both faster anyway.

use encoding 'utf8';

while (<>) {
	s/([^[:ascii:]])/sprintf("&#%d;",ord($1))/eg;
	print $_;
}