HTML Parsing is not robust enough

[ericdbarry]

I am pointing to a login page that has some oddities in it, exposing some assumptions about the pages being processed.

Two examples:

1. escape function does not like None being passed in
2. name tag is required, yet login_form_html_node.findall(".//input")) does not filter

Both of these are valid markup - the first being a boolean attribute being present, the second being an input tag without a name attribute.