IMAP Learn

imap_learn.sh

#!/bin/bash
#
# IMAP Learn
#
# Copyright 2012 Colin Mollenhour (http://colin.mollenhour.com)
#
###############################################################################
#
# This script allows you to use an imap account to train your spamassassin server.
# Example script:
# 
# #!/bin/bash
# AUTH="--server imap.example.com:993 --user user@example.com --password SECRET"
# SA_ARGS="--dbpath /etc/spamassassin/.spamassassin"
# imap_learn.sh $AUTH --folder "Junk E-mail" --type spam --delete-remote -- $SA_ARGS
# imap_learn.sh $AUTH --folder "Ham" --type ham -- $SA_ARGS
# sa-learn $SA_ARGS --sync
#
# Notes:
#  - SSL support is required, port must be specified in --server option
#  - Spamassassin already tracks duplicates so duplicate learning is not a concern
#  - After running imap_learn, you should run "sa-learn --sync"
#
###############################################################################

LEARN=1
KEEP_LOCAL=0
KEEP_REMOTE=1
SINCE=$(date +%d-%b-%Y -d '-3 days')
SLEEP_TIME=1
while true; do
  [ -n "$1" ] || break
  arg="$1"; shift
  case "$arg" in
    --server)
      IMAP_SERVER="$1"
      ;;
    --user)
      IMAP_USER="$1"
      ;;
    --password)
      IMAP_PW="$1"
      ;;
    --folder)
      IMAP_FOLDER="$1"
      ;;
    --since)
      SINCE="$1"
      ;;
    --type)
      TYPE="$1"
      ;;
    --dry-run)
      LEARN=0
      continue
      ;;
    --keep-local)
      KEEP_LOCAL=1
      continue
      ;;
    --delete-remote)
      KEEP_REMOTE=0
      continue
      ;;
    --)
      break
      ;;
    *)
      echo "Invlaid argument: $arg"
      exit 1
      ;;
  esac
  shift
done

if [ -z "$IMAP_SERVER" -o -z "$IMAP_USER" -o -z "$IMAP_PW" -o -z "$IMAP_FOLDER" -o -z "$SINCE" -o -z "$TYPE" ]; then
  echo "You did not specify all required arguments."
  exit 1
fi

connect_imap="openssl s_client -crlf -connect $IMAP_SERVER"
pause="sleep $SLEEP_TIME"

IDS="$( 
$connect_imap < <( 
        echo "A01 LOGIN $IMAP_USER $IMAP_PW"; 
$pause; echo "A02 SELECT \"$IMAP_FOLDER\"";
$pause; echo "A03 SEARCH SEEN SINCE $SINCE";
$pause; echo "A04 LOGOUT";
$pause; echo ""; 
) 2>/dev/null | tr -d "\r" | grep -E "^[*] SEARCH" | sed "s/^[*] SEARCH \([0-9 ]*\)$/\1/g";
)";
if [ "$IDS" = "" ]; then
  echo "No messages for $IMAP_USER"
  exit
fi

tmpfile=$(mktemp /tmp/$TYPE.XXXX)
$connect_imap < <(
        echo "A01 LOGIN $IMAP_USER $IMAP_PW"; 
$pause; echo "A02 SELECT \"$IMAP_FOLDER\"";
for id in $IDS; do $pause; echo "A03 FETCH $id BODY[]"; done
$pause; echo "A04 LOGOUT";
) 2>/dev/null > $tmpfile

if [ $KEEP_REMOTE -eq 0 ]; then
  $connect_imap < <(
          echo "A01 LOGIN $IMAP_USER $IMAP_PW"; 
  $pause; echo "A02 SELECT \"$IMAP_FOLDER\"";
  $pause; echo "A03 STORE ${IDS// /,} +FLAGS (\Deleted)";
  $pause; echo "A04 EXPUNGE";
  $pause; echo "A05 LOGOUT";
  ) 2>&1 > /dev/null
fi

echo Fetched Messages: ${IDS// /,} into $tmpfile

tmpdir=$tmpfile-emails/
mkdir -p $tmpdir || { echo Could not make $tmpdir; exit 1; }
awk "
BEGIN { p=0; i=0; }
/^\\)\\r$/ { p=0; }
{ if (p) print \$0 >> \"$tmpdir\"i\".eml\"; }
/[*] [0-9]+ FETCH/ { if (!p) { p=1; i++; } }
END { print \"Found \"i\" emails\"; }
" $tmpfile

[ $LEARN -eq 1 ] && sa-learn --$TYPE --no-sync "$@" $tmpdir
[ $KEEP_LOCAL -eq 1 ] || rm -rf $tmpfile $tmpdir