[OGo-Developer] multipart/form-data and text encoding

[Wolfgang Sourdeau]

On 2007-08-23 14:01:38 -0500 Helge Hess <helge.hess@opengroupware.org> 
wrote:

> On 23.08.2007, at 00:09, Wolfgang Sourdeau wrote:
>> I think there was mention of modifying the parsers instead
> 
> I have not found the time to look at the changes yet, but changing  
> default 
> encodings for decoding from ASCII to UTF-8 should almost  always make 
> sense 
> (since every ASCII char is a valid UTF-8 char).

My idea is actually that the parser or the parser delegate should 
return the charset for NGMime objects which have none specified. Hence 
the change to return nil instead of returning a default, in 
NGMimeBodyPart.

> Anyways, the more important thing is that we should not work on 
> NGHttpRequest and the NGMime based HTTP parser! Quite a while ago I  
> started 
> that "SimpleHTTPParser" which can be enabled using the 
> "WOHttpTransactionUseSimpleParser" default. (and in fact its much  
> more 
> reliable and efficient for WebDAV usage!, so its recommended for 
> ZideStore/xmlrpcd)
> 
> The NGMime stuff is *quite* complex (and I'm relunctant to change a  
> lot 
> there, because it works quite well for b0rked messages). This is  
> necessary 
> for mail, but completely unnecessary for HTTP which is  rather clean 
> and well 
> implemented in 'da real world.

I could do this and then rely on the WOMessageUseUTF8 to parse the 
results. We might as well change
WOMessageUseUTF8 to WOMessageCharset instead, for the sake people 
using yet another different charset.

> There is only one issue which is why the SimpleParser is not enabled  
> by 
> default: it doesn't implement multipart/formdata=>WORequest  decoding 
> (which 
> is not that hard, but some work).

I already did such a parser in perl a few years ago, I think I could 
write one very quickly.


Wolfgang