HTMLParser html entities preprocessing is undocumented

[Kees Cook]

The attached script does not emit HTML entities on Xenial, but works correctly on Trusty. I don't see any difference in the html2lib package, so I assume something must have changed in Python 3.

Trusty (correct):

$ ./test.py
Start tag: body
Named ent: uuml
End tag : body

Xenial (incorrect):

$ ./test.py
Start tag: body
Data : ü
End tag : body

[Kees Cook (kees)]

Nevermind, it's not html5lib, it's libpython3.5-stdlib, and the issue is the new undocumented "convert_charrefs" argument to _init__.