#!/usr/bin/python

"""Badly named script. Converts non-ascii UTF-8 text into equivalent decimal
entities for use in HTML and XML. I have C and perl versions of this too.

Requires Python >=2.0. Now works with Python 3.x."""

# Iain Murray 2005, 2019

import fileinput, sys

if sys.version_info[0] >= 3:
    for txt in fileinput.input():
        sys.stdout.buffer.write(txt.encode('ascii','xmlcharrefreplace'))
elif (sys.version_info[0] > 2) or ((sys.version_info[0] == 2) and (sys.version_info[1] >= 3)):
    # From python 2.3 xml conversion is built in.
    for txt in fileinput.input():
        txt = unicode(txt, 'utf-8')
        sys.stdout.write(txt.encode('ascii', 'xmlcharrefreplace'))
else:
    # Horrible version for pythons 2.0--2.2 (not tested recently, might be broken)
    import re
    def xmlescape(x):
        return ''.join(('&#', str(ord(x.group(0))), ';'))
    nonascii=re.compile('[^\x00-\x7F]')
    for txt in fileinput.input():
        txt = unicode(txt, 'utf-8')
        sys.stdout.write(nonascii.sub(xmlescape,txt).encode('ascii'))