rename marknew to dejavu

[etc/mailfilter.git] / procmail / spamfilter

diff --git a/procmail/spamfilter b/procmail/spamfilter
index a882087f3a732d052e0501f46452d00530bed70e..1b0d0d1b83a5e339190e98d1a3d39d84a033887e 100755 (executable)
--- a/procmail/spamfilter
+++ b/procmail/spamfilter
@@ -18,15 +18,27 @@ PMDIR=${PMDIR:-$HOME/.etc/mailfilter/procmail}
 
 #VERBOSE=yes
 
-INCLUDERC=$PMDIR/pre-spam-cleanup
-
 # no need to reprocess messages that went into a spamtrap
+# UPDATE: retrain them only if diagnosed as non-spam, see below
+# Note: add E flag to next recipe when uncommenting
+#:0
+#* SPAMTRAPPED ?? .
+#{ 
+#  LOG="spamfilter:  skipping checks for spamtrapped message$NL"
+#  :0 fw
+#  |$FORMAIL -I"X-Spam: spamtrapped"
+#}
+
+# check whether this message is being reinjected
+TRAINED_AS
 :0
-* SPAMTRAPPED ?? .
-{ 
-  LOG="spamfilter:  skipping checks for spamtrapped message$NL"
-  :0 fw
-  |$FORMAIL -I"X-Spam: spamtrapped"
+*$ $MSG_DEJAVU
+* ^X-Trained-As: \/(h|sp)am
+{
+  LOG="spamfilter:  skipping already trained $MATCH$NL"
+  :0
+  * MATCH ?? spam
+  { IS_SPAM=already-trained }
 }
 
 # let earlier parts of the mailfilter cause bypassing the checks
@@ -48,6 +60,7 @@ INCLUDERC=$PMDIR/pre-spam-cleanup
   :0 fw
   |$FORMAIL -I"X-Spam: unknown (check skipped)"
   SPAM_UNKNOWN=skip-match
+  SKIP_SPAMCHECKS=match
 }
 
 # sanity check on message size
@@ -63,6 +76,8 @@ INCLUDERC=$PMDIR/pre-spam-cleanup
 # now run the spamfilters
 :0 E
 {
+  INCLUDERC=$PMDIR/pre-spam-cleanup
+
   # crm114
   CRM_SPAM=UNKNOWN
   CRM_SCORE=0
@@ -112,24 +127,34 @@ INCLUDERC=$PMDIR/pre-spam-cleanup
   :0
   * CRM_SPAM ?? UNSURE
   {
-    # retrain as ham
+    # retrain spamtrapped message
     :0
+    * SPAMTRAPPED ?? .
+    {
+      LOG="spamfilter:  scheduling retraining with SPAM due to spamtrap$NL"
+      :0 fw
+      |$FORMAIL -A "X-CRM114-Autotrain: spam, due to spamtrap"
+      RETRAIN=spam
+    }
+
+    # retrain as ham
+    :0 E
     * ? perl -e "$SA_SCORE <= $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM || exit 1"
     {
-      LOG="spamfilter:  scheduling crm114 retraining with HAM (score $SA_SCORE <= $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM)$NL"
+      LOG="spamfilter:  scheduling retraining with HAM (score $SA_SCORE <= $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM)$NL"
       :0 fw
       |$FORMAIL -A "X-CRM114-Autotrain: ham, according to SA (score $SA_SCORE <= $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM)"
-      CRM_RETRAIN=ham
+      RETRAIN=ham
     }
 
     # retrain as spam
     :0 E
-    * ? perl -e "$SA_SCORE > $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM || exit 1"
+    * 1^0 ? perl -e "$SA_SCORE > $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM || exit 1"
     {
-      LOG="spamfilter:  scheduling crm114 retraining with SPAM (score $SA_SCORE > $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM)$NL"
+      LOG="spamfilter:  scheduling retraining with SPAM (score $SA_SCORE > $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM)$NL"
       :0 fw
       |$FORMAIL -A "X-CRM114-Autotrain: spam, according to SA (score $SA_SCORE > $CRM_UNSURE_SA_AUTOTRAIN_LIMIT_SPAM)"
-      CRM_RETRAIN=spam
+      RETRAIN=spam
     }
 
     # skip retraining if SA is not convinced
@@ -147,12 +172,22 @@ INCLUDERC=$PMDIR/pre-spam-cleanup
   * CRM_SPAM ?? SPAM
   * SA_SPAM ?? No
   {
-    # SA is convincing, so retrain crm114
+    # message was spamtrapped anyway
     :0
+    * SPAMTRAPPED ?? .
+    {
+      LOG="spamfilter:  resolving crm114/SA disagreement due to spamtrap ($CRM_SCORE/$SA_SCORE)$NL"
+      RETRAIN=spam
+      :0 fw
+      |$FORMAIL -A "X-CRM114-Retrain: spam, due to spamtrap"
+    }
+
+    # SA is convincing, so retrain crm114
+    :0 E
     * ? perl -e "$SA_SCORE <= $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_HAM || exit 1"
     {
       LOG="spamfilter:  crm114 found spam ($CRM_SCORE), but SA is more convincing ($SA_SCORE <= $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_HAM)$NL"
-      CRM_RETRAIN=ham
+      RETRAIN=ham
       :0 fw
       |$FORMAIL -A "X-CRM114-Retrain: ham, according to SA (score $SA_SCORE <= $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_HAM)"
     }
@@ -172,12 +207,22 @@ INCLUDERC=$PMDIR/pre-spam-cleanup
   * CRM_SPAM ?? GOOD
   * SA_SPAM ?? Yes
   {
+    # message was spamtrapped anyway
+    :0
+    * SPAMTRAPPED ?? .
+    {
+      LOG="spamfilter:  resolving crm114/SA disagreement due to spamtrap ($CRM_SCORE/$SA_SCORE)$NL"
+      RETRAIN=spam
+      :0 fw
+      |$FORMAIL -A "X-CRM114-Retrain: spam, due to spamtrap"
+    }
+
     # SA is convincing, so retrain crm114
     :0
     * ? perl -e "$SA_SCORE > $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_SPAM || exit 1"
     {
       LOG="spamfilter:  crm114 found ham ($CRM_SCORE), but SA is more convincing ($SA_SCORE > $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_SPAM)$NL"
-      CRM_RETRAIN=spam
+      RETRAIN=spam
       :0 fw
       |$FORMAIL -A "X-CRM114-Retrain: spam, according to SA (score $SA_SCORE > $CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_SPAM)"
     }
@@ -205,5 +250,18 @@ INCLUDERC=$PMDIR/pre-spam-cleanup
   |$FORMAIL -I"X-Spam: no (crm114:$CRM_SCORE SA:$SA_SCORE)"
 }
 
+# schedule spamtrapped ham for retraining as spam
+:0
+* SPAMTRAPPED ?? .
+* ! SKIP_SPAMCHECKS ?? .
+* ! IS_SPAM ?? .
+{ 
+  LOG="spamfilter:  found spamtrapped ham, retraining...$NL"
+  :0 fw
+  |$FORMAIL -I"X-Spam: spamtrapped ham"
+  IS_SPAM=spamtrapped-ham
+  RETRAIN=spam
+}
+
 INCLUDERC=$PMDIR/handlespam
 #VERBOSE=no