Good morning!
Seeing some messages come through with large amounts of bayes poison text
inserted between style /style tags.
Short of using a 'rawbody' test, is there some other characteristic that
we could catch?
For example, and another question:
Is there any mechanism in SpamAssassin to count the number of times a rule
is matched within a message? For example, a single line that contains
ten or more unpunctuated words is probable, but having them make up more
than half the overall lines in the mail is quite improbable and most
likely spam. How could we count that?
Can a flag be set (or can we create one) on a rule so that it counts a
small score for *each* match rather than just score it once? I realize
these tests would have to be used with considerable care, like a
'rawbody', because they would always have to scan the whole body, rather
than stop at the first match.
- Charles
