702290 - Wiki parser generates invalid HTML when MEDIA_URL includes the host

Status: RESOLVED WONTFIX

(support.mozilla.org :: Knowledge Base Software, task, P5)

Description

[James Socol [:jsocol, :james]]

Setting MEDIA_URL to a full or protocol-relative URL, including the hostname, causes 3 tests in sumo.tests.test_parser to fail (at least for me). It's related to our image lazy-loading: the src attribute of the <img> inside the <noscript> gets turned into a link, meaning it looks like 

<noscript><img src="<a href="//path/to/image">...</a>"></noscript>

Ricky can't reproduce it, which is further interesting. But it we want to push more content onto the CDN we will have to make sure this is just my VM and not a real problem.

Comment 1

[James Socol [:jsocol, :james]]

So, we narrowed this down further (details in regex):

MEDIA_URL = '(https?:)?//' + 

'localhost(:8000)?' -> works
'localhost.localdomain(:8000)?' -> works
'example.com(:8000)?' -> FAILS

+ '/media/'

That leads me to suspect it's from pywikimarkup's naive URL linkification.

Comment 2

[James Socol [:jsocol, :james]]

Attachment: a failing test

Here's a patch that demos a failing test because of this bug.

Comment 3

[Ricky Rosario [:rrosario, :r1cky]]

Cleaning up 2011Q4

Comment 4

[Roland Tanglao needinfo please :rolandtanglao, :mohnkuchen, :adobo, :sinigang, :roland]

hey we have a reproducible failing test, sounds like something for good first bug