[spambayes-bugs] [ spambayes-Bugs-936821 ] sb_mboxtrain.py requires writeaccess to mbox

[SourceForge.net]

Bugs item #936821, was opened at 2004-04-17 08:45
Message generated for change (Tracker Item Submitted) made by Item Submitter
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=498103&aid=936821&group_id=61702

Category: None
Group: Source code - CVS
Status: Open
Resolution: None
Priority: 5
Submitted By: PieterB (pieterb)
Assigned to: Nobody/Anonymous (nobody)
Summary: sb_mboxtrain.py requires writeaccess to mbox 

Initial Comment:
This is tested with spambayes1.0b1.1, which isn't listed 
in the Group pulldownbox on sourceforge (yet).

At our system we use a spamtrap which catches a lot of 
spam. I want to use that spammailbox for improving 
spambayes. When I try to learn from this read-only 
mbox I get the following traceback:

Training spam (/var/mail/spamtrap):
  Reading as Unix mbox
Traceback (most recent call last):
  File "/usr/local/bin/sb_mboxtrain.py", line 350, in ?
    main()
  File "/usr/local/bin/sb_mboxtrain.py", line 341, in main
    train(h, s, True, force, trainnew, removetrained)
  File "/usr/local/bin/sb_mboxtrain.py", line 265, in train
    mbox_train(h, path, is_spam, force)
  File "/usr/local/bin/sb_mboxtrain.py", line 172, in 
mbox_train
    f = file(path, "r+b")
IOError: [Errno 13] Permission 
denied: '/var/mail/spamtrap'

It can be fixed by first trying to open the file readonly, 
and when that fails open the file for 'r+b'.
In the method mbox_train from sb_mboxtrain.py,
use the following code.

    # Open and lock the mailbox.  Some systems require 
it be opened for
    # writes in order to assert an exclusive lock.
    try: # opening the file read only
       f = file(path, "rb")
       fcntl.flock(f, fcntl.LOCK_EX)
    except: # opening file fails
       f = file(path, "r+b")
       fcntl.flock(f, fcntl.LOCK_EX)

This makes spambayes start learning from the mbox.
Unfortuately, spambayes really wants to write to the 
mailbox and it fails:

Training spam (/var/mail/spamtrap):
  Reading as Unix mbox
  3960Problem truncating mbox--nothing written
Traceback (most recent call last):
  File "/usr/local/bin/sb_mboxtrain.py", line 355, in ?
    main()
  File "/usr/local/bin/sb_mboxtrain.py", line 346, in main
    train(h, s, True, force, trainnew, removetrained)
  File "/usr/local/bin/sb_mboxtrain.py", line 270, in train
    mbox_train(h, path, is_spam, force)
  File "/usr/local/bin/sb_mboxtrain.py", line 202, in 
mbox_train
    os.ftruncate(f.fileno(), 0)
IOError: [Errno 22] Invalid argument

This can be fixed by setting Headers:include_trained to 
false, e.g.

/usr/local/bin/sb_mboxtrain.py -o 
Headers:include_trained:False -
p /home/foobar/var/spambayes.db -s /var/mail/spamtrap
   
I would expect sb_mboxtrain.py not to open files for 
writing if it's not necessary/requested.

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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=498103&aid=936821&group_id=61702