adding a test for html doctype embeded in a <pre><code>

[cmerot]

When an item contains:

<pre><code>
#!html
&lt;!DOCTYPE html>
</code></pre>

with the relativize_paths filter, the result is messed up by the following code:

# FIXME cleanup because it is ugly
# # Because using the `Nokogiri::XML::DocumentFragment` class DOCTYPE 
# pseudonodes becomes even more creepy than usual.
result.sub!(/(!DOCTYPE.+?)(&gt;)/, '<\1>')

Is that still usefull?

[denisdefreyne]

I don’t think this is necessary anymore. From what I can tell, it was necessary before because the relativize_paths filter used DocumentFragment, but it does no longer.

I’ll clean this up!

[denisdefreyne]

See #296, which not only includes the test but also a fix.