Re: [xml] Bug in HTML parser output?

[Matt Sergeant <matt sergeant org>]

On Fri, 26 Apr 2002, Daniel Veillard wrote:

> On Fri, Apr 26, 2002 at 11:16:51AM +0100, Matt Sergeant wrote:
> > In using libxml2's HTML parser to create valid XML, I noticed a "bug"...
> >
> > xmllint --html --format http://www.messagelabs.com/VirusEye/ | xmllint -
> >
> > Croaks on the bad ---> comment in the HTML.
> >
> > Is there any way to make this just "work"?
>
>   hum, right this seems a loophole, the HTML parser is overly flexible to
> be able to parse what's found on the net, but doesn't take corrective measures
> to cleanup things like HTML comments
>
> > (yeah I know I should get them to fix thier nasty HTML too)
>
>   I wonder what's the best approach:
>     - fix the HTML importer
>     - fix the XML serializer
>
> the second case sounds quite more generic, I would be tempted to go that
> way.

Yeah, unfortunately there's no way in XML to allow through that dash, even
by turning it into two comments. But frankly I wouldn't care if it got
stripped completely ;-)

> How urgent is this ?

It's not. I just added in an additional "| perl -pe 's/--->/-->/g'" into
the pipeline to fix it for now. Plus I work for MessageLabs, so I can kick
some butt fairly easily ;-)

-- 
<!-- Matt -->
<:->Get a smart net</:->