Definitely a python2/3 issue. While python2 uses byte strings by default, strings are Unicode strings in python3 and byte strings need to be decoded first.
Surely, you know that and the question is: which part of the website
creation is not fully ported to python3.
python2 [2.7.17]
>>> 'abc', u'abc', b'abc', 'abc'.encode()
('abc', u'abc', 'abc', 'abc')
python3 [3.6.9]
>>> 'abc', u'abc', b'abc', 'abc'.encode()
('abc', 'abc', b'abc', b'abc')
Best,
Joram
