lower SA limit for autotraining ham to unsure crm

[etc/mailfilter.git] / procmail / defines

diff --git a/procmail/defines b/procmail/defines
index 511e614baa14dc1707865155bed2011fcda819ca..43945cdb9c89492f13ec6c1c85def19e27383982 100644 (file)
--- a/procmail/defines
+++ b/procmail/defines
@@ -17,12 +17,15 @@ REPLIES=$MAILFILT/autoreplies
 LOGFILE=${LOGFILE:-$PMVAR/log}
 LOGABSTRACT=no
 
+LC_CTYPE=de_CH.UTF-8
+
 #NICE='/usr/bin/nice -20'
 
 PROCMAIL="$NICE /usr/bin/procmail -p $PMDIR/procmailrc"
 FORMAIL="$NICE /usr/bin/formail -f"
 EGREP="$NICE /bin/egrep"
 SED="$NICE /bin/sed"
+BIN_DATE="/bin/date"
 DELIVER="$NICE /usr/lib/dovecot/deliver"
 
 CRM114="$NICE /usr/share/crm114/mailreaver.crm -u $MAILFILT/crm114/"
@@ -30,10 +33,9 @@ SA_PREFS="$MAILFILT/spamassassin/user_prefs"
 SPAMASSASSIN="$NICE /usr/bin/spamassassin --prefs-file=$SA_PREFS"
 SPAMC="$NICE /usr/bin/spamc --log-to-stderr --no-safe-fallback"
 #SPAMC="$SPAMASSASSIN"
-TRAINER="$MAILFILT/bin/train"
+TRAINER="$NICE $MAILFILT/bin/train"
 
-OURDATE=`date -R`
-OURDATE_SHORT=`date +%Y.%m.%d.%H.%M.%N`
+SQLITE="$NICE /usr/bin/sqlite3"
 
 BASE=$HOME/.maildir
 
@@ -42,11 +44,11 @@ DEFAULT=$BASE/
 ORGMAIL=$HOME/BOUNCED-MAIL
 
 # maximum message size for spam checking
-SPAMCHECK_MAX_MESSAGE_SIZE=2500000
+SPAMCHECK_MAX_MESSAGE_SIZE=512000
 
 # if crm114 is unsure and SA returns a score less-than-or-equal to this,
 # autotrain crm114 with ham
-CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM=2.0
+CRM_UNSURE_SA_AUTOTRAIN_LIMIT_HAM=0.0
 # if crm114 classifies a message as spam but SA returns a score
 # less-than-or-equal to this, retrain crm114
 CRM_MISCLASSIFY_SA_AUTOTRAIN_LIMIT_HAM=-1.0
@@ -62,17 +64,36 @@ NL="
 "
 RE_MYDOMAIN="(.+\.)*madduck\.net"
 RE_MAILRELAYS="(seamus|clegg)\.madduck\.net"
-RE_SPACE_NEWLINE="(^|[         ])"
+RE_SPACE="[    ]"
+RE_NOT_SPACE="[^       ]"
+RE_SPACE_NEWLINE="(^|$RE_SPACE)"
 RE_FIRSTNAME="martin($RE_SPACE_NEWLINE+f(\.?|elix))?"
 RE_LASTNAME="kraff?t"
-RE_EXTRACT_HEADER_VALUE="[     ]*\/[^  ].*"
+RE_EXTRACT_HEADER_VALUE="$RE_SPACE*\/$RE_NOT_SPACE.*"
 
 DEJAVU_HEADER=X-Deja-Vu
 
 NULL=/dev/null
+DELAYED_QUEUE=$BASE/.delayed/
+TICKLER_QUEUE=$BASE/.store/
 DISCARD=$BASE/.discard/
+SPAM=$BASE/.spam/
 #DISCARD="'|$DELIVER -m BASE.discard'"
 
+DELAY_NEXT_WEEKEND='next sunday 28 hours ago' # fri night, 20:00
+DELAY_TONIGHT='tomorrow 00:00 4 hours ago' # tonight at 20:00
+
+OURDATE="`$BIN_DATE +'%s %Y.%m.%d.%H.%M.%N %a, %d %b %Y %T %z'`"
+:0
+*$ OURDATE ?? ^\/${RE_NOT_SPACE}+
+{ OURDATE_TS="$MATCH" }
+:0
+*$ OURDATE ?? ^[0-9]+${RE_SPACE}+\/${RE_NOT_SPACE}+
+{ OURDATE_SHORT="$MATCH" }
+:0
+*$ OURDATE ?? ^[0-9]+${RE_SPACE}+[0-9.]+${RE_SPACE}+\/.+
+{ OURDATE="$MATCH" }
+
 ### variables from the message
 
 ### local recipient data
@@ -109,14 +130,35 @@ INCLUDERC=$PMDIR/get-msgid
 
 :0
 *$ ^Subject:$RE_EXTRACT_HEADER_VALUE
-{ SUBJECT="$MATCH" }
+{
+  SUBJECT=$MATCH
+
+  # mimedecode.c: * Disclaimer: We only handle charset of iso-8859-1
+  :0
+  * SUBJECT ?? ^=\?iso-8859-1\?[QB]\?.+\?=$
+  {
+    DECODED="`echo Subject: $SUBJECT | mimedecode | iconv -f latin1 -t utf-8`"
+    :0
+    *$ DECODED ?? ^Subject:$RE_EXTRACT_HEADER_VALUE
+    { SUBJECT=$MATCH }
+  }
+}
 
 :0
 *$ ^X-Original-To:$RE_EXTRACT_HEADER_VALUE
 { ORIGINAL_TO="$MATCH" }
 :0 E
+* ^Received:
 { LOG="NO ORIGINAL_TO: $MSGID" }
 
+:0
+*$ ^X-Trained-As:$RE_EXTRACT_HEADER_VALUE
+{ TRAINED_AS="$MATCH" }
+
+:0
+*$ ^X-Postponed:$RE_EXTRACT_HEADER_VALUE
+{ POSTPONED="$MATCH" }
+
 # fix variable values for special cases
 INCLUDERC=$PMDIR/normalise
 
@@ -136,6 +178,11 @@ INCLUDERC=$PMDIR/normalise
 # to the X-Been-There header).
 MSG_DEJAVU=!
 
+# DEST
+# set this to a folder if the message should be delivered elsewhere than the
+# default
+DEST
+
 # SKIP_SPAMCHECKS
 # if set, cuases spamchecks to be skipped, value lists reason
 SKIP_SPAMCHECKS
@@ -172,3 +219,7 @@ RETRAIN
 # JUSTME
 # if set, contains reason why justme message was passed
 JUSTME
+
+# DISABLE_DELAYS
+# if set, disables delaying messages
+DISABLE_DELAYS