Gajim should be fetching the avatar from the transport (this is the
only place it can fetch them from).
If you take a look in yahoo/avatar.py you'll find that the avatars are
being fetched from
http://img1.avatar.vip.dcn.yahoo.com/users/<hash>.medium.png (asking
google about this host shows that _only_ the yahoo transports are
using it)
If you can find similar code in Gaim, and find out where there are
looking for the data, that would be great! (I was looking an
yahoo_picture.c, but I couldn't see any url's)
It's very possible that we need to announce a different version to the
yahoo servers to get the new data.
On 10/4/06, Jean Verger <[EMAIL PROTECTED]> wrote:
Hi,
new updates on Yahoo Transport and Avatars ...
Well, the thing is that any modification on the Yahoo Transport
regarding the avatars was not affecting Gajim ... mmmm why? well, I
dived in the code of Gajim and notice that Gajim implements its own
functions to get the avatar ... not using the ones offered by the transport.
To the question of:
what happens if you alter:
if pay[each][213] == '1':
to
if pay[each][213] == '1' or pay[each][213] == '2':
Apparently, the messages/answers were the same ... I mean, when dealing
with pay[each][213] == '1' or pay[each][213] == '2' the functions
replies "OK" ... mmmm it's weird because i didn't see the new avatar,
although the return value was as OK.
The new approach is to dive in the code of "Gaim" (not Gajim) to see how
Gaim download any kind of avatar. Gajim only gets to download Buddy
Icons ... Gaim gets any kind of download ...
So far, this is the only code I got from Gaim ... apparently it does not
make any distinction from any kind of avatars ... it should work the
same :(((((
case 197: /* Avatars */
{
char *decoded, *tmp;
guint len;
if (pair->value) {
gaim_base64_decode(pair->value, &decoded, &len);
if (len) {
tmp = gaim_str_binary_to_ascii(decoded, len);
gaim_debug_info("yahoo", "Got key 197, value =
%s\n", tmp);
g_free(tmp);
}
g_free(decoded);
}
break;
}
I'll keep posted if i get any updates,
Jean
Norman Rasmussen wrote:
> On 1/22/07, Jean Verger <[EMAIL PROTECTED]> wrote:
>> - Any reason for this behavior?
>
> Not really, I haven't installed the Yahoo client, so I didn't know
> there was an Option B. Option A is easy to test because you can do
> everything via the web-site.
>
>> - I've been trying to modify the code. I thought that the key was in
>> ylib.py:
>>
>> if pay[each][213] == '1':
>> b = avatar.getavatar(pay[each][197], self.dumpProtocol)
>> if b != None and self.handlers.has_key('avatar'):
>>
>> self.handlers['avatar'](self.fromjid,pay[each][7],b)
>>
>> Analyzing the packets, I noticed that avatars type B have pay[each][213]
>> equals to 2.
>> However, the surprising/frustrating/funny part of all this is that when
>> I comment out any part of the code that makes a reference to the avatar,
>> the avatar is still been displayed ...
>
> I'm pretty sure Gajim caches the avatar, so you'd have to clear out
> the Gajim cache too to really notice that the avatar is gone.
>
> what happens if you alter:
>> if pay[each][213] == '1':
> to:
>> if pay[each][213] == '1' or pay[each][213] == '2':
> ?
>
>> Any hint on how Yahoo Transport is dealing with avatars? Any
>> document/experience?
> no docs, best thing to do is make sure dumpProtocol is enabled, then
> you can see the raw data.
>
> At the moment what happens is that the yahoo server send a hash of the
> image data, which is then fetched via http from the yahoo image
> server. If the Option B images are stored on the same server then
> adding the above code should make things 'just work'. If there's a
> different url for those images then something like ethereal is useful
> to sniff it out.
>
> If you get it working, please post the patch so that I can apply it :-)
>
--
- Norman Rasmussen
- Email: [EMAIL PROTECTED]
- Home page: http://norman.rasmussen.co.za/
_______________________________________________
py-transports mailing list
py-transports@blathersource.org
http://www.modevia.com/cgi-bin/mailman/listinfo/py-transports
