On 4/10/2011 4:43 PM, John Hardin wrote: > On Sun, 10 Apr 2011, rokdominko wrote: > >> The problem is, we want to tell our client exactly which words, >> phrases or sentences are problematic, so we need Spamassassin to >> return the list of these words, phrases or sentences, so that we can >> tell our client what exactly is wrong with their message. >> >> We have written a PHP script, which connects to spamd process on our >> server >> (on port 783) and it checks the message with no problems and if it's >> spam it >> doesn't allow sending it. > > That level of detail isn't available via spamd. You'd have to run > spamassassin in debug mode with rules tracing and then parse the results. > > Take a look at the output from: > > spamassassin -t --debug area=rules < your_message_file >
Keep in mind that when using the '-t' flag, spamassassin will always claim the mail is spam. You will need to ignore this and focus on the score instead. -- Bowie
