Andrea Bencini wrote:
I perform a dump of the Bayes database (sa-learn --dump data)
0.049 0 1 1114618731 7bc32206de
0.026 0 2 1114618851 b214d213fd
0.958 1 0 1114607858 ea2d4af4f7
0.026 0 2 1114618998 5bb6204bec
0.005 0 10 1114619097 3236284507
0.049 0 1 1114618779 7af61f2550
it is not in human readable format, but in hex (7bc32206de).
How can I do?
In sa 3.0.0 or higher the tokens are stored as SHA1 hashes, there is no
practical way to reverse them back to text strings.
The conversion to hashes speeds up SA's database maintenance, as the tokens are
now fixed-size.
Unfortunately you can't browse your bayes db tokens to see what's been learned
as what. This does however have the side-benefit of improving privacy on
multi-user systems.
One thing you CAN still do is run a message through spamassassin -D and look at
the bayes ratings of the tokens that message hit. Since SA has the message in
text form, it will be able present the tokens in text format not hash format.
Usually if you're questioning the contents of the bayes DB, it's in relation to
a specific message, so this should cover most situations.
