Re: Question regarding meta's
Matt Kettler writes: Dan wrote: Thanks Matt, That certainly would explain my problem. The entry is listed near the bottom of this page: http://bmrc.berkeley.edu/resources/how_to/email/Mail_SpamAssassin_Conf.html Checking Google, its the only page in the world saying meta SYMBOLIC_TEST_NAME regular expression. Its not clear which version this relates to. Back to the drawing board... Yeah, I just checked the latest SVN snapshot, it doesn't have this either. Looking closer at that page, it also mentions required_hits, but does not mention it as deprecated. required_hits became deprecated in SA 3.0.0. It also doesn't mention report_safe, which was present in the release of SA 2.60 However it does mention the use of arithmetic meta's, which were new in SA 2.50. My only guess is that page is built from some kind of SVN build of a pre-release version of SA 2.50. They may have been intending to add regex support, and then backed it out when it caused problems. Yeah, we had that in a *long* time ago. I think we took it out for 2.60; nobody was using it, and it was tricky to support. --j.
Re: Question regarding meta's
I have an similar meta question, I'm trying to include all tests conforming to a wildcard (*) meta entry so I don't have to type out each member. This is the configuration I have so far but as you can see, RemoveBBB is not scoring. Is what I'm attempting possible and am I doing it correctly? There is no SCORE line, so it should default to 1.0: CONTENT Disable This Rubbish Discontinue Correspondence Discontinue Emails Like These Discontinue This Kind of Message TESTS body REMOVE_B16 /Disable This Rubbish/i body REMOVE_B17 /Discontinue Correspondence/i body REMOVE_B18 /Discontinue Emails Like These/i body REMOVE_B19 /Discontinue This Kind of Message/i META meta RemoveBBB ( (REMOVE_B.* +) 1) RESULTS X-SpamAssassin: score=4.0 tests=REMOVE_B16,REMOVE_B17,REMOVE_B18,REMOVE_B19 This is the section of the manual I'm following: meta SYMBOLIC_TEST_NAME regular expression Finally, parts of a meta rule may be defined by a regexp followed by an operator. All rules matching the regular expression will be strung together, with the given operator between each expression. The rule compiler will add ``^'' and ``$'' before and after each regexp, so the regexp must match the entire rule; a rule name begining with a ``.'' or a ``['' will be treated as a regexp. As an example: meta TOO_MANY_UA ( (USER_AGENT.* +) 1) Thanks! Dan
Re: Question regarding meta's
Dan wrote: This is the section of the manual I'm following: meta SYMBOLIC_TEST_NAME regular expression Finally, parts of a meta rule may be defined by a regexp followed by an operator. All rules matching the regular expression will be strung together, with the given operator between each expression. The rule compiler will add ``^'' and ``$'' before and after each regexp, so the regexp must match the entire rule; a rule name begining with a ``.'' or a ``['' will be treated as a regexp. As an example: meta TOO_MANY_UA ( (USER_AGENT.* +) 1) Erm, what version of the man page contains that? This syntax is not in SA 3.0.x or 3.1.x.
Re: Question regarding meta's
Thanks Matt, That certainly would explain my problem. The entry is listed near the bottom of this page: http://bmrc.berkeley.edu/resources/how_to/email/ Mail_SpamAssassin_Conf.html Checking Google, its the only page in the world saying meta SYMBOLIC_TEST_NAME regular expression. Its not clear which version this relates to. Back to the drawing board... Dan On Apr 20, 2006, at 17:13, Matt Kettler wrote: Dan wrote: This is the section of the manual I'm following: meta SYMBOLIC_TEST_NAME regular expression Finally, parts of a meta rule may be defined by a regexp followed by an operator. All rules matching the regular expression will be strung together, with the given operator between each expression. The rule compiler will add ``^'' and ``$'' before and after each regexp, so the regexp must match the entire rule; a rule name begining with a ``.'' or a ``['' will be treated as a regexp. As an example: meta TOO_MANY_UA ( (USER_AGENT.* +) 1) Erm, what version of the man page contains that? This syntax is not in SA 3.0.x or 3.1.x.
Re: Question regarding meta's
Dan wrote: Thanks Matt, That certainly would explain my problem. The entry is listed near the bottom of this page: http://bmrc.berkeley.edu/resources/how_to/email/Mail_SpamAssassin_Conf.html Checking Google, its the only page in the world saying meta SYMBOLIC_TEST_NAME regular expression. Its not clear which version this relates to. Back to the drawing board... Yeah, I just checked the latest SVN snapshot, it doesn't have this either. Looking closer at that page, it also mentions required_hits, but does not mention it as deprecated. required_hits became deprecated in SA 3.0.0. It also doesn't mention report_safe, which was present in the release of SA 2.60 However it does mention the use of arithmetic meta's, which were new in SA 2.50. My only guess is that page is built from some kind of SVN build of a pre-release version of SA 2.50. They may have been intending to add regex support, and then backed it out when it caused problems.
Re: Question regarding meta's
Yeah, I just checked the latest SVN snapshot, it doesn't have this either. Looking closer at that page, it also mentions required_hits, but does not mention it as deprecated. required_hits became deprecated in SA 3.0.0. It also doesn't mention report_safe, which was present in the release of SA 2.60 However it does mention the use of arithmetic meta's, which were new in SA 2.50. My only guess is that page is built from some kind of SVN build of a pre-release version of SA 2.50. They may have been intending to add regex support, and then backed it out when it caused problems. Superb info, thanks!
Re: Question regarding meta's
On Thu, Apr 13, 2006 at 08:40:30PM +0200, Ruben Cardenal wrote: header __ID1 /regexp1/ header __ID2 /regexp2/ header __ID3 /regexp3/ meta MYID ((__ID1 + __ID2 + __ID3) 1) When a message triggers MYID, is there any way in the X-Spam-Report of showing which individual parts of the meta the message matched? As far as I know, you can't do that without a plugin. You could write a small plugin such that _SUBTESTS_ or something would be rewritten to the list of subtests (starts with __) that hit, and then include that in the report. -- Randomly Generated Tagline: It's a question of consistency. With a Republican president, I think you should just expect a certain amount of corruption -- And with a Democratic president, you should expect a [ bleep ] in the oval office. - Dave Foley on Politically Incorrect, 2001.12.07 pgpQsFQAHB14A.pgp Description: PGP signature
Re: Question regarding meta's
Ruben Cardenal wrote: Hi, Let's say I have: header __ID1 /regexp1/ header __ID2 /regexp2/ header __ID3 /regexp3/ meta MYID ((__ID1 + __ID2 + __ID3) 1) score MYID 1 When a message triggers MYID, is there any way in the X-Spam-Report of showing which individual parts of the meta the message matched? No, but you can do something like this: header ID1 /regexp1/ score ID1 0.0001 header ID2 /regexp2/ score ID2 0.0001 header ID3 /regexp3/ score ID3 0.0001 meta MYID ((ID1 + ID2 + ID3) 1) score MYID 1 This will force ID1-3 to be evaluated as normal rules and show up in the hit list, but will give them an insignificant score. (You can't make the score 0, that will disable them)