-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Matt Kettler writes: > Craig Jackson wrote: > > Regarding the report that is made part of the header, instead of a > > summary of the email (which makes downloading tiresomely long even on a > > high speed connection), I would like to see only the text that tripped > > each particular test. Users often want to know what exactly in the email > > caused it to be tagged as spam. It would be nice if I could show them in > > the header at the workstation, rather than run a test from my > > workstation and print it out. If this can already be done, let me know > > and I'll figure out how to do it. > > Currently, that's pretty much impossible to do in spamassassin, and it would > be > difficult to do without increasing memory usage. > > Right now, SA just maintains a simple hit list when it scans a message. It > doesn't extract the text which matched at all, much less save it where it can > be > referenced later. > > Quoting the part of the email that caused the rule to trip is particularly > difficult for eval tests, as their code could look at anything or everything. > > It's also a little tricky for meta tests which wind up triggering as a cascade > of other rules trip off. (ie: DRUGS_MANYKINDS). > > I'd also question who would find such a feature more useful, end users or > spammers. > > Certainly it has some valid uses for end users and non-spammers, but it's > considerably more useful to a spammer. For end users it answers the "why did > this happen" questions, for spammers it provides useful spam-tuning feedback > that makes their spam more effective at evading detection by SA. actually, we do already support it ;) if you run SpamAssassin 3.1.0 with - -D, it'll log what parts of the message hit for some of the rules (just the simple regexp rules, not eval tests). - --j. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.5 (GNU/Linux) Comment: Exmh CVS iD8DBQFCtxlVMJF5cimLx9ARAuUPAJkBtnZG+REM8mVUISHMumdxKmNLlACfbt77 2MhOg5NLA4GBLS9HEaQsxTo= =y3Gn -----END PGP SIGNATURE-----
