[ mailman-Patches-1189243 ] Unhandle exception in Message.get_content_charset()

[SourceForge.net]

Patches item #1189243, was opened at 2005-04-25 00:16
Message generated for change (Comment added) made by bwarsaw
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=300103&aid=1189243&group_id=103

Category: None
Group: Mailman 2.1
Status: Closed
Resolution: Accepted
Priority: 5
Submitted By: Jon Parise (jparise)
Assigned to: Nobody/Anonymous (nobody)
Summary: Unhandle exception in Message.get_content_charset()

Initial Comment:
Message.get_content_charset() can encounter an
unhandled exception "(TypeError: unicode() argument 2
must be string, not None") when charset is None.  This
can can only occur when the call to
email.Message.Message.get_content_charset(self,
failobj) itself returns None, but, not being acquainted
with the email package, I don't understand the
circumstances which can cause this situation.

Regardless, when charset is None, the call to unicode()
will throw the TypeError exception.  The attached patch
recognizes this as a valid failure case such that
failobj will be returned.

----------------------------------------------------------------------

>Comment By: Barry A. Warsaw (bwarsaw)
Date: 2005-04-29 09:30

Message:
Logged In: YES 
user_id=12800

Tokio, I don't like having duplicate code, so I think it's
best to fix the problem in one place and not keep the
workaround in Mailman's copy.  I've installed email 2.5.6
now and will release Mailman 2.1.6rc3 in a little while.  At
my own site, all the shunted messages that were broken
because of this problem have now been fixed and properly
unshunted.  I will be updating python.org hopefully this
weekend.

Let's give rc3 a couple of weeks to shake out.

----------------------------------------------------------------------

Comment By: Tokio Kikuchi (tkikuchi)
Date: 2005-04-28 20:02

Message:
Logged In: YES 
user_id=67709


Barry, TypeError is already avoided in email package by this
statement:
    pcharset = charset[0] or 'us-ascii'

I want to keep the latter try/except in Mailman/Message.py
to assure the charset is known to mailman so that it can
serve as security precaution to future mailman code changes,
even after the email package is updated to 2.5.6.




----------------------------------------------------------------------

Comment By: Barry A. Warsaw (bwarsaw)
Date: 2005-04-25 13:02

Message:
Logged In: YES 
user_id=12800

If it's not too much trouble.  I'd love to have a test case
in the email package for the TypeError exception.  If you
can't, no worries.

----------------------------------------------------------------------

Comment By: Jon Parise (jparise)
Date: 2005-04-25 11:46

Message:
Logged In: YES 
user_id=485579

Sorry, I don't.  Those messages were unshunted as a result
of my fix.  I could try to instrument a collection mechanism
by trapping that exception and storing a copy of the
offensive message, however.  Would that still be useful?

----------------------------------------------------------------------

Comment By: Barry A. Warsaw (bwarsaw)
Date: 2005-04-25 10:32

Message:
Logged In: YES 
user_id=12800

Is there any possibility that you have a sample message that
exhibits this problem?  I would /really/ like to add a test
case for this to the email package, which is where the patch
really belongs.


----------------------------------------------------------------------

Comment By: Tokio Kikuchi (tkikuchi)
Date: 2005-04-25 00:50

Message:
Logged In: YES 
user_id=67709

Thank you!  I just fixed in CVS.
(I'd rather make it bare except ...)


----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=300103&aid=1189243&group_id=103