I am wanting to write a framework with the features below for my corporation
as an upgrade to our current spam blocking solution. I would suggest them
as ideas for SpamAssassin v3.
Create a framework where the message comes into the daemon and is translated
to 'feeds' by 'feed modules' aka preprocessors.
The message is passed to each feed module, and each module returns a result
('feed') that is a scalar, a hash, or an array, and is known as a feed.
New feed modules can be added.
Feed modules can build on each other.
Examples of feeds:
A message2text feed module would do its best to convert the message to a
single set of text, ignoring mime parts, etc.
A tokeniser might the results of the message2text feed and convert it to a
hash of single and two word tokens in the message and how many times they
were counted.
A scrubber feed module might remove all 'lsdkjf laskdjf lasdj lkasdfj
ljjasdf' type garbage strings from the results of the message2text feed or
even the message feed. For instance spammers are boosting the byte count of
messages with these random strings, but the scrubber would remove them.
Create a framework where the message or message feeds are then scored by
scoring modules
New scoring modules can be added to act upon feeds or environment variables
passed, such as sender ip address.
Modules can depend upon the scores of one another.
Modules can depend on certain feeds and not run if they don't exist.
Modules can be grouped into categories.
For instance, you have the whitelist category of module
If the message is from a certain ip address that is determined trusted, you
add points to the category score.
If the message is from a domain name determined trusted, you add points to
the category score.
You have a system where the score of a category of modules will determine if
another category of modules is even checked.
For instance, if the whitelist category score is over 3.00, you don't run
the Bayesian module, the RBL module, the Razor module, etc.
-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk
