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 # let earlier parts of the mailfilter cause bypassing the checks
51 * SKIP_SPAMCHECKS ?? .
53 LOG="spamfilter: skipping checks as requested: $SKIP_SPAMCHECKS$NL"
55 |$FORMAIL -I"X-Spam: unknown (skip requested)"
56 SPAM_UNKNOWN=skip-requested
59 # honour skip-spamchecks to exclude certain messages from spam checks
62 * ? $EGREP -qif $CONF/skip-spamchecks
64 LOG="spamfilter: skipping checks as per skip-spamchecks$NL"
66 |$FORMAIL -I"X-Spam: unknown (check skipped)"
67 SPAM_UNKNOWN=skip-match
71 # sanity check on message size
73 * > $SPAMCHECK_MAX_MESSAGE_SIZE
75 LOG="spamfilter: skipping check because message size exceeds $SPAMCHECK_MAX_MESSAGE_SIZE bytes$NL"
77 |$FORMAIL -I"X-Spam: unknown (message larger than $SPAMCHECK_MAX_MESSAGE_SIZE bytes)"
78 SPAM_UNKNOWN=too-large
81 # now run the spamfilters
84 INCLUDERC=$PMDIR/pre-spam-cleanup
96 * ^X-CRM114-Status: \/[A-Z]+
100 * ^X-CRM114-Status: .+\([ ]*\/-?[.0-9]+
103 LOG="crm114: $CRM_SPAM/$CRM_SCORE$NL"
117 * ^X-Spam-Status: \/[A-Za-z]+
121 * ^X-Spam-Status: .+score=\/-?[.0-9]+
125 * ^X-Spam-Status: .+tests=\/[^ ]+
128 LOG="SA: $SA_SPAM/$SA_SCORE/$SA_TESTS$NL"
131 ## CASE 0: crm114 is unsure/untrained
135 # retrain spamtrapped message
139 LOG="spamfilter: scheduling retraining with SPAM due to spamtrap$NL"
141 |$FORMAIL -A "X-CRM114-Autotrain: spam, due to spamtrap"
147 * ? perl -e "$SA_SCORE <= $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM || exit 1"
149 LOG="spamfilter: scheduling retraining with HAM (score $SA_SCORE <= $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM)$NL"
151 |$FORMAIL -A "X-CRM114-Autotrain: ham, according to SA (score $SA_SCORE <= $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM)"
157 * 1^0 ? perl -e "$SA_SCORE > $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM || exit 1"
159 LOG="spamfilter: scheduling retraining with SPAM (score $SA_SCORE > $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM)$NL"
161 |$FORMAIL -A "X-CRM114-Autotrain: spam, according to SA (score $SA_SCORE > $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM)"
165 # skip retraining if SA is not convinced
168 LOG="spamfilter: will not autotrain crm114 because SA is not convinced ($CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM <= $SA_SCORE < $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM)$NL"
170 |$FORMAIL -A "X-CRM114-Autotrain: SA is unsure ($CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM <= $SA_SCORE < $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM)"
171 SPAM_UNSURE=sa-unsure
175 ## CASE 1: disagreement, SA sees ham
180 # message was spamtrapped anyway
184 LOG="spamfilter: resolving crm114/SA disagreement due to spamtrap ($CRM_SCORE/$SA_SCORE)$NL"
187 |$FORMAIL -A "X-CRM114-Retrain: spam, due to spamtrap"
190 # SA is convincing, so retrain crm114
192 * ? perl -e "$SA_SCORE <= $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_HAM || exit 1"
194 LOG="spamfilter: crm114 found spam ($CRM_SCORE), but SA is more convincing ($SA_SCORE <= $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_HAM)$NL"
197 |$FORMAIL -A "X-CRM114-Retrain: ham, according to SA (score $SA_SCORE <= $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_HAM)"
200 # SA is not convincing, mark as disagreement
203 LOG="spamfilter: crm114 found spam ($CRM_SCORE), but SA thinks it's ham ($SA_SCORE)$NL"
206 |$FORMAIL -I "X-Spam: disagree (crm114:spam/$CRM_SCORE SA:ham/$SA_SCORE)"
210 ## CASE 1: disagreement, SA sees spam
215 # message was spamtrapped anyway
219 LOG="spamfilter: resolving crm114/SA disagreement due to spamtrap ($CRM_SCORE/$SA_SCORE)$NL"
222 |$FORMAIL -A "X-CRM114-Retrain: spam, due to spamtrap"
225 # SA is convincing, so retrain crm114
227 * ? perl -e "$SA_SCORE > $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_SPAM || exit 1"
229 LOG="spamfilter: crm114 found ham ($CRM_SCORE), but SA is more convincing ($SA_SCORE > $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_SPAM)$NL"
232 |$FORMAIL -A "X-CRM114-Retrain: spam, according to SA (score $SA_SCORE > $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_SPAM)"
235 # SA is not convincing, mark as disagreement
238 LOG="spamfilter: crm114 found ham ($CRM_SCORE), but SA thinks it's spam ($SA_SCORE)$NL"
239 SPAM_DISAGREE=sa-spam
241 |$FORMAIL -I "X-Spam: disagree (crm114:ham/$CRM_SCORE SA:spam/$SA_SCORE)"
251 |$FORMAIL -I"X-Spam: yes (crm114:$CRM_SCORE SA:$SA_SCORE)"
255 |$FORMAIL -I"X-Spam: no (crm114:$CRM_SCORE SA:$SA_SCORE)"
258 # schedule spamtrapped ham for retraining as spam
261 * ! SKIP_SPAMCHECKS ?? .
264 LOG="spamfilter: found spamtrapped ham, retraining...$NL"
266 |$FORMAIL -I"X-Spam: spamtrapped ham"
267 IS_SPAM=spamtrapped-ham