Re: Handling invalid headers without sending status code 400

[Dan Winship <danw gnome org>]

On 02/19/2014 11:59 AM, Jon Deon wrote:
> I'm trying to use libsoup to create a REST server as an AirPlay client.  (The REST interface for AirPlay is 
> "documented" here - http://nto.github.io/AirPlay.html)
>
> The problem that I'm having is that the soup server appears to choke on the requests that the iPad's 
> AirPlay client is sending.  I grabbed a tcpdump of the exchange and opened it in WireShark, a request looks 
> like:
>
> GET /server-info HTTP/1.1
> X-Apple-Client-Name: iPad
> X-Apple-VV: 16777984
> Content-Length: 0
> X-Apple-Device-ID: (Hex value here)
> X-Apple-Session-ID: (UUID here)
> User-Agent: AirPlay/190.9 (Photos)
>
> (The linebreaks are \r\n, FWIW)
>
>
> My SoupServer gets this request, and immediately returns 
>
> HTTP/1.1 400 Bad Request
> Content-Length: 0

There's nothing obviously wrong with the request as you pasted it... can
you send me the actual tcpdump output?

> Is there some way to handle these bad requests via libsoup that I'm not aware of?  The signals don't seem 
> to give me access to any of the raw data, and the library is choking on the request.  I could always switch 
> to some other way of handling HTTP for this, but I've already done some other implementation that is 
> libsoup-based so I'd like to handle this in my code if possible.  

No... we've started implementing stackable I/O filters
(soup-content-processor.c), which can be used to rewrite the data as
it's read in (eg, to implement Content-Encoding). However, it's
currently not yet part of the public API, and it's not yet used on the
server side anyway. So there's no way you can get at or fix the data
before libsoup has tried to parse it.

-- Dan