[PATCH 04/13] Update to match semantics of new imaplib2

[Ethan Glasser-Camp]

On 02/07/2011 07:12 AM, Sebastian Spaeth wrote:
> On Sun,  6 Feb 2011 11:58:58 -0500, Ethan Glasser-Camp<ethan at betacantrips.com>  wrote:
>> The biggest change here is that imapobj.untagged_responses is no
>> longer a dictionary, but a list. To access it, I use the semi-private
>> _get_untagged_response method.
>> diff --git a/offlineimap/folder/IMAP.py b/offlineimap/folder/IMAP.py
>> index db1bdae..2c6ced7 100644
> ...
>> @@ -85,14 +85,14 @@ class IMAPFolder(BaseFolder):
>>           imapobj = self.imapserver.acquireconnection()
>>           try:
>>               # Primes untagged_responses
>> -            imapobj.select(self.getfullname(), readonly = 1, force = 1)
>> +            imaptype, imapdata = imapobj.select(self.getfullname(), readonly = 1, force = 1)
>>               try:
>>                   # 1. Some mail servers do not return an EXISTS response
>>                   # if the folder is empty.  2. ZIMBRA servers can return
>>                   # multiple EXISTS replies in the form 500, 1000, 1500,
>>                   # 1623 so check for potentially multiple replies.
>>                   maxmsgid = 0
>> -                for msgid in imapobj.untagged_responses['EXISTS']:
>> +                for msgid in imapdata:
>>                       maxmsgid = max(long(msgid), maxmsgid)
>>               except KeyError:
>>                   return True
>
> Careful here, are we really sure that we are doing the right thing here
> (and analogue in the quickchanged function)? There are 3 possible return
> scenarioes here (ignoring the "BAD" reply for a moment which we should
> have protected against anyway):
>
> 1) Return exactly one "EXISTS" reply. Everything is fine.
> 2) Don't return any EXISTS when the folder is empty (as some servers are
>     doing per comment). Previously the KeyError exception would have
>     handled that. Now, we can never trigger the KeyError case, so that
>     one has become useless. (but it seems we would set maxmsgid to 0, so
>     we should be fine with regard to what we return. Do away with the
>     KeyError exception here?
> 3) Return multiple EXISTS messages, will the imaplib2 code return all of
>     them or only one? imaplib2 does:
>          typ, dat = self._untagged_response(typ, [None], 'EXISTS')
>     in select(). Will that return a list of multiple EXISTS values if we
>     get them? Perhaps it is, but I am not sure.

Hmm, good question. It looks to me like _untagged_response returns a 
list no matter what. If there's something in self.untagged_responses for 
the command, we return that. If not, we return [None].

untagged_responses is generally populated by _append_untagged(), which 
either inserts an element of [name, [result]] if none exists, or adds a 
result to an existing list otherwise. So if we get multiple EXISTS, I'm 
pretty sure we'll get them as a list.

But in the [None] case, I think we'll crash, since long(None) throws an 
exception. So I guess I better remove the KeyError handling and replace 
it with a check for None.

Ethan