All patches and comments are welcome. Please squash your changes to logical
commits before using git-format-patch and git-send-email to
patches@git.madduck.net.
If you'd read over the Git project's submission guidelines and adhered to them,
I'd be especially grateful.
3 #TODO: rewrite to use SPAM variable, and do not autotrain spam here, only ham
5 PMDIR=${PMDIR:-$HOME/.etc/mailfilter/procmail}
10 # PMVAR is not defined, so we are being called as filter
11 # thus source the standard defines
12 INCLUDERC=$PMDIR/defines
13 # prevent feeding back to procmail and delete the leading From line
15 # and tell the fucking procmail piece-of-shit to continue to be a filter
21 # no need to reprocess messages that went into a spamtrap
22 # UPDATE: retrain them only if diagnosed as non-spam, see below
23 # Note: add E flag to next recipe when uncommenting
27 # LOG="spamfilter: skipping checks for spamtrapped message$NL"
29 # |$FORMAIL -I"X-Spam: spamtrapped"
32 # check whether this message is being resubmitted
39 LOG="spamfilter: skipping already trained $TRAINED_AS$NL"
42 { IS_SPAM=already-trained }
46 { LOG="spamfilter: skipping resubmitted message$NL" }
49 # do not run spamfilters if the message destination is already set
53 LOG="spamfilter: message already routed to '$DEST'$NL"
54 SPAM_UNKNOWN=already-destined
57 # let earlier parts of the mailfilter cause bypassing the checks
59 * SKIP_SPAMCHECKS ?? .
61 LOG="spamfilter: skipping checks as requested: $SKIP_SPAMCHECKS$NL"
63 |$FORMAIL -I"X-Spam: unknown (skip requested)"
64 SPAM_UNKNOWN=skip-requested
67 # honour skip-spamchecks to exclude certain messages from spam checks
70 * ? $EGREP -qif $CONF/skip-spamchecks
72 LOG="spamfilter: skipping checks as per skip-spamchecks$NL"
74 |$FORMAIL -I"X-Spam: unknown (check skipped)"
75 SPAM_UNKNOWN=skip-match
79 # sanity check on message size
81 * > $SPAMCHECK_MAX_MESSAGE_SIZE
83 LOG="spamfilter: skipping check because message size exceeds $SPAMCHECK_MAX_MESSAGE_SIZE bytes$NL"
85 |$FORMAIL -I"X-Spam: unknown (message larger than $SPAMCHECK_MAX_MESSAGE_SIZE bytes)"
86 SPAM_UNKNOWN=too-large
89 # now run the spamfilters
92 INCLUDERC=$PMDIR/spamtraps
93 INCLUDERC=$PMDIR/spammers
94 INCLUDERC=$PMDIR/pre-spam-cleanup
102 #TODO: somehow filter out headers we added
107 * ^X-CRM114-Status: \/[A-Z]+
111 * ^X-CRM114-Status: .+\([ ]*\/-?[.0-9]+
114 LOG="crm114: $CRM_SPAM/$CRM_SCORE$NL"
128 * ^X-Spam-Status: \/[A-Za-z]+
132 * ^X-Spam-Status: .+score=\/-?[.0-9]+
136 * ^X-Spam-Status: .+tests=\/[^ ]+
139 LOG="SA: $SA_SPAM/$SA_SCORE/$SA_TESTS$NL"
142 ## CASE 0: crm114 is unsure/untrained
146 # retrain spamtrapped message
150 LOG="spamfilter: scheduling retraining with SPAM due to spamtrap$NL"
152 |$FORMAIL -A "X-CRM114-Autotrain: spam, due to spamtrap"
158 * ? perl -e "$SA_SCORE <= $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM || exit 1"
160 LOG="spamfilter: scheduling retraining with HAM (score $SA_SCORE <= $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM)$NL"
162 |$FORMAIL -A "X-CRM114-Autotrain: ham, according to SA (score $SA_SCORE <= $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM)"
168 * 1^0 ? perl -e "$SA_SCORE > $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM || exit 1"
170 LOG="spamfilter: scheduling retraining with SPAM (score $SA_SCORE > $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM)$NL"
172 |$FORMAIL -A "X-CRM114-Autotrain: spam, according to SA (score $SA_SCORE > $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM)"
176 # skip retraining if SA is not convinced
179 LOG="spamfilter: will not autotrain crm114 because SA is not convinced ($CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM <= $SA_SCORE < $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM)$NL"
181 |$FORMAIL -A "X-CRM114-Autotrain: SA is unsure ($CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM <= $SA_SCORE < $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM)"
182 SPAM_UNSURE=sa-unsure
186 ## CASE 1: disagreement, SA sees ham
191 # message was spamtrapped anyway
195 LOG="spamfilter: resolving crm114/SA disagreement due to spamtrap ($CRM_SCORE/$SA_SCORE)$NL"
198 |$FORMAIL -A "X-CRM114-Retrain: spam, due to spamtrap"
201 # SA is convincing, so retrain crm114
203 * ? perl -e "$SA_SCORE <= $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_HAM || exit 1"
205 LOG="spamfilter: crm114 found spam ($CRM_SCORE), but SA is more convincing ($SA_SCORE <= $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_HAM)$NL"
208 |$FORMAIL -A "X-CRM114-Retrain: ham, according to SA (score $SA_SCORE <= $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_HAM)"
211 # SA is not convincing, mark as disagreement
214 LOG="spamfilter: crm114 found spam ($CRM_SCORE), but SA thinks it's ham ($SA_SCORE)$NL"
217 |$FORMAIL -I "X-Spam: disagree (crm114:spam/$CRM_SCORE SA:ham/$SA_SCORE)"
221 ## CASE 1: disagreement, SA sees spam
226 # message was spamtrapped anyway
230 LOG="spamfilter: resolving crm114/SA disagreement due to spamtrap ($CRM_SCORE/$SA_SCORE)$NL"
233 |$FORMAIL -A "X-CRM114-Retrain: spam, due to spamtrap"
236 # SA is convincing, so retrain crm114
238 * ? perl -e "$SA_SCORE > $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_SPAM || exit 1"
240 LOG="spamfilter: crm114 found ham ($CRM_SCORE), but SA is more convincing ($SA_SCORE > $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_SPAM)$NL"
243 |$FORMAIL -A "X-CRM114-Retrain: spam, according to SA (score $SA_SCORE > $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_SPAM)"
246 # SA is not convincing, mark as disagreement
249 LOG="spamfilter: crm114 found ham ($CRM_SCORE), but SA thinks it's spam ($SA_SCORE)$NL"
250 SPAM_DISAGREE=sa-spam
252 |$FORMAIL -I "X-Spam: disagree (crm114:ham/$CRM_SCORE SA:spam/$SA_SCORE)"
262 |$FORMAIL -I"X-Spam: yes (crm114:$CRM_SCORE SA:$SA_SCORE)"
266 |$FORMAIL -I"X-Spam: no (crm114:$CRM_SCORE SA:$SA_SCORE)"
269 # schedule spamtrapped ham for retraining as spam
272 * ! SKIP_SPAMCHECKS ?? .
275 LOG="spamfilter: found spamtrapped ham, retraining...$NL"
277 |$FORMAIL -I"X-Spam: spamtrapped ham"
278 IS_SPAM=spamtrapped-ham