[Spambayes] bug: long lines trashed w/ mboxtrain.py

[Sam Shah]

Consider a simple mbox like (excuse long to line):

    From test at umich.edu Fri Jan 24 17:18:05 2003
    From: "Test" <test at umich.edu>
    To: "Someone Test #A" <someone at eecs.umich.edu>,<someone at eecs.umich.edu>,"Someone Test #B" <someone at umich.edu>, "Someone Test #C" <someone at eecs.umich.edu>, "Someone Test #D" <someone at eecs.umich.edu>
    Subject: testing
    Date: Fri, 24 Jan 2003 17:21:35 -0500
    Status: RO
    Content-Length: 5
    Lines: 1

    Test

We run mboxtrain.py on it, twice:

    [sligo tmp/spambayes-1.0a2 ]$ ./mboxtrain.py -d test -g msg
    Training ham (msg):
      Reading as Unix mbox
      Trained 1 out of 1 messages                
    [sligo tmp/spambayes-1.0a2 ]$ ./mboxtrain.py -d test -g msg
    Training ham (msg):
      Reading as Unix mbox
      Trained 0 out of 1 messages                

We get the following. Notice the To line is somehow munged.

    From test at umich.edu Fri Jan 24 17:18:05 2003
    From: "Test" <test at umich.edu>
    To: "Someone Test #A"
            <someone at eecs.umich.edu>,<someone at eecs.umich.edu>,"Someone Test #B"
            "Someone Test	#D" <someone at eecs.umich.edu>
    Subject: testing
    Date: Fri, 24 Jan 2003 17:21:35 -0500
    Status: RO
    Content-Length: 5
    Lines: 1
    X-Spambayes-Trained: ham

    Test

I tried looking at the code for mboxtrain, but I couldn't find an
obvious problem. This occurs with Spambayes v1.0a2 running on
Python 2.2.2.

Thanks,
Sam