Some email addresses are messed up with parse_email=False

[ThiefMaster]

>>> linkify('doesntmatter@foo-bar.de')
u'doesntmatter@foo<a href="http://-bar.de" rel="nofollow">-bar.de</a>'

[jsocol]

This only occurs with parse_email=False. Hmm. Unfortunately I'm not sure this is fixable, though I can tweaked the URL regex not to match domains that start with -, because I'm fairly sure that's not allowed.

[ThiefMaster]

Sounds like a good workaround.

However, why not always parse email but simply do not modify them if parse_email=False. Or replace them with a placeholder hash and restore them later so the URL regex doesn't catch them.

[jsocol]

That's an interesting idea but I really want to get 1.2 out, so I'm going to bump to 1.3 and see if that's a good path to follow.

[jmulmer]

I have a similar problem when setting parse_email=False. Any email with following the pattern something.tlds@somedomain.something where tlds is bleach.__init__.TLDS displays a broken link.

For example...

john.coop@domain.com

becomes...

<a href="john.coop">john.coop</a>@domain.com

But...

john.coops@domain.com

is fine, since "coops" isn't in bleach.__init__.TLDS

[willkg]

I'm going to drop this from the milestone. I think it needs more thinking and I'm not going to get to it for v2.0.

[willkg]

I spent some time fiddling with this today. If we always parse out email addresses, but have parse_email dictate whether we turn them into a url or not, then we break urls that have usernames and passwords in them like this test case:

==================================================== FAILURES ====================================================
_________________________________ tests/test_linkify.py::test_link_http_complete _________________________________

    def test_link_http_complete():
>       assert (
            linkify('https://user:pass@ftp.mozilla.org/x/y.exe?a=b&c=d&e#f') ==
            '<a href="https://user:pass@ftp.mozilla.org/x/y.exe?a=b&amp;c=d&amp;e#f" rel="nofollow">'
            'https://user:pass@ftp.mozilla.org/x/y.exe?a=b&amp;c=d&amp;e#f</a>'
        )
E       assert 'https://user...p;c=d&amp;e#f' == '<a href="http...d&amp;e#f</a>'
E         - https://user:pass@ftp.mozilla.org/x/y.exe?a=b&amp;c=d&amp;e#f
E         + <a href="https://user:pass@ftp.mozilla.org/x/y.exe?a=b&amp;c=d&amp;e#f" rel="nofollow">https://user:pass@ftp.mozilla.org/x/y.exe?a=b&amp;c=d&amp;e#f</a>

tests/test_linkify.py:329: AssertionError
====================================== 1 failed, 299 passed in 0.84 seconds ======================================

I don't think we can get both cases to work without a rewrite of some kind, but I'm not sure what that would look like.

I don't know how to fix this and I'm not that interested in spending more time on it. If someone else wants, they should take a stab at it and let us know what they find. Otherwise, I'll probably close it out because it's been open for 6 years and no one seems interested enough to look into it further.

[willkg]

This is a problem, but this isn't something we're going to fix. Sorry!