0

Consider these mail headers:

X-Spam-Level: ***
X-Spam-Status: Yes, score=3.7 required=3.0 tests= (remaining truncated for brevity)

I simply want to match the headers

"X-Spam-Level: ***" (at least with more "*" okay)

or

"X-Spam-Status: Yes, score=3.7" (or a higher score)

I have this, which matches the X-Spam-Level header, but I'm seeing stuff slip through with a score of 3.7 or higher, making me believe the alternation is being skipped:

:0 H
    * ^X-Spam-Status: Yes
    * ^(X-Spam-Level: \*\*\*\*)|(^X-Spam-Status: Yes, score=3\.[7-9])
    { 
        do stuff
}

Procmail's regex implementation is just not quite the same as what I'm used to and can't get this to work, despite matching on sites like regex101.com

I'm perfectly fine with a solution just matching the score of 3.7 or higher and removing the X-Spam-Level condition completely.

Note: I am aware that the minimum score I've set in Spamassassin is 3.0 to flag as spam, however, I deliver stuff to the user's mailbox between 3 and 3.6 and move anything higher to a server spam mailbox. Presently, it's only getting delivered to the server's spam mailbox if 4.1 or higher, as denoted by the asterisks in X-Spam-Level.

Montclair
  • 81
  • 1
  • 10
  • Your example works fine. Can you provide an actual sample message which fails your conditions? (Granted, the grouping of your regex is slightly odd, but it works for this particular sample.) – tripleee Apr 15 '20 at 08:47
  • They all fail and all use the format I specified in the example. Not sure why the regex didn't work. – Montclair Apr 17 '20 at 14:06

1 Answers1

0

I never could get this working. Instead, I modified the Spamassassin local.cf config to add a header line:

add_header all Score _SCORE_

This creates a new mail header of X-Spam-Score: n.n where n are digits.

In the procmailrc file, I added this line to the top:

# Change this value to be whatever threshold
# you want
# Spam scored between 3 and 3.6 will be delivered.
# Spam scored over 3.6 won't.
SPAMTHRESHOLD=3.6

Later in the file, after spamc is called, I declare a variable to capture the score then check against it:

# Create SCORE variable which contains the header I created in Spamassassin
:0 # H flag is implicit
    * ^X-Spam-Score: \/.+
    { SCORE=$MATCH }

# Check for Spam-Status: Yes, and 
# $SCORE > $SPAMTHRESHOLD
:0 H
    * ^X-Spam-Status: Yes
    * $ -${SPAMTHRESHOLD}^0
    * $ ${SCORE}^0
    { 
        # do stuff. in this case move the mail to 
        # /var/mail/spam
        :0:spam.lock
        /var/mail/spam
}
Montclair
  • 81
  • 1
  • 10