Command line programs for working with HTML5 character entities.

ent2utf8.py

#!/usr/bin/env python3

import html
import sys

if sys.stdin.isatty():
    try:
        while 1:
            sys.stdout.write(html.unescape(input())+'\n')
    except EOFError: pass
else:
    for line in sys.stdin:
        sys.stdout.write(html.unescape(line))

utf82ent.py

#!/usr/bin/env python3

import html.entities as he
import sys

html5_r = dict()

for k, v in he.html5.items():
    if not v in html5_r:
        html5_r[v] = list()
    html5_r[v].append('&'+k)

def find(uni):
    if not uni in html5_r:
        return ["&#x{:X};".format(ord(uni))]
    else:
        return html5_r[uni]

if sys.stdin.isatty():
    try:
        while 1:
            out = [' or '.join(find(c)) for c in input()]
            out = ' followed by '.join(out)
            sys.stdout.write(out+'\n')
    except EOFError: pass
else:
    for line in sys.stdin:
        sys.stdout.write(html.unescape(line))