html parser raises IndexError on iterlinks()

[Sardar]

lxml/html/__init__.py

Sometimes HtmlMixin.iter() returns an element with el.tag == ''
The HtmlMixin.iterlinks() on line 322 uses _nons(el.tag), which directly accesses tag[0], which raises IndexError.
I know the HTML is invalid, but the parser is expected to be robust.

Fix:

if el.tag == '':
    continue

thus skip broken tag, it is not a link anyway.

Note: this is purely python code problem, not with lxml libxml2/libxslt library.

[scoder (scoder)]

Would you have an example of a page that shows this behaviour? I wonder how it is possible that the parser returns an empty tag name in the first place.

[scoder (scoder)]

Cannot reproduce.

[scoder (scoder)]

Closing as outdated and missing a way to reproduce the problem.