PATH=/bin:/usr/bin:/usr/local/bin:$HOME/bin
MAILDIR=$HOME/Maildir
DEFAULT=$MAILDIR/
PMDIR=$HOME/.procmail


# Logging to keep track of my emails.
LOGFILE=$PMDIR/log
LOGABSTRACT=all
LINEBUF=4096
#VERBOSE=yes

# Rewrite PGP emails a little so they come out in a format that mutt will
# understand automatically.
#:0
#* !^Content-Type: message/
#* !^Content-Type: multipart/
#* !^Content-Type: application/pgp
#{
#    :0 fBw
#    * ^-----BEGIN PGP MESSAGE-----
#    * ^-----END PGP MESSAGE-----
#    | formail \
#        -i "Content-Type: application/pgp; format=text; x-action=encrypt"
#
#    :0 fBw
#    * ^-----BEGIN PGP SIGNED MESSAGE-----
#    * ^-----BEGIN PGP SIGNATURE-----
#    * ^-----END PGP SIGNATURE-----
#    | formail \
#        -i "Content-Type: application/pgp-signed; format=text; x-action=sign"
#}
#

# Count the number of lines, if necessary.  (This is rather pointless for
# mbox-delievered messages, but it's here just in case I need it.)
:0
* ! ^Lines:
{
    # Count number of lines
    :0B
    * 1^1 ^.*$
    { }
    LINES = $=

    # Add Lines: header
    :0 fhw
    | formail -a "Lines: $LINES"
}


# Silly people trying to set their own email status.
:0 fhw
* ^(X-)?Status:
| formail -I Status -I X-Status

# Likewise for SpamAssassin.  I'll do my own checks, thanks.
:0 fw
* ^X-Spam-(Status|Level|Score|Report):
| spamassassin -d

# In a similar vein, I made this header up, but might as well have a
# recipe to remove it for paranoia's sake.
:0 fhw
* ^X-List-Classify:
| formail -I X-List-Classify

# Useless use of Reply-To:
:0 fhw
*   ^Reply-To: \/.+
* $ ^From:.*$MATCH
| formail -I Reply-To


# No filtering on wedding email, just in case.
:0
* ^TO_wedding@aperiodic.net
.wedding/


# And now the mailing list rules.

# Per-list tidying.

:0 fhw
* ^List-Id: tinyfugue
| sed -e "s/^\\(Subject:.*\\)\\[TF\\]\\( \\)*/\\1/I"

:0 fhw
* ^List-Id: PDF Rangers
| sed -e "s/^\\(Subject:.*\\)\\[PDF Rangers\\]\\( \\)*/\\1/I"

# Dropping list mail into mailboxes.

# To avoid duplication of recipes, I have the various types of mailing
# lists set an X-List-Classify header and then have a single ruleset that
# triggers on that header.

# Note that this allows the creation or alteration of near-arbitrary files on
# your computer.  Take care with the data.  I disallow dot (.) and forward
# slash (/) from the list names for this reason.

# most lists - RFC2919 and some variants.
:0 fhw
* ^((List-Id|X-(Mailing-)?List):(.*[<]\/[^>@\.\/]*))
| formail -I "X-List-Classify: $MATCH"

# Majordomo
:0 fhw
* ^Sender: owner-[^@]+@[^@\+]+
* ^Sender: owner-\/[^@\+\.\/]+
| formail -I "X-List-Classify: $MATCH"

# yahoogroups
:0 fhw
* ^Mailing-List: list \/[^@\.\/]+
| formail -I "X-List-Classify: $MATCH"

# ezmlm
:0
* ^Mailing-List: contact \/[^@\+\.\/]+
{
    LISTID=`echo $MATCH | sed -e 's/-help$//'`

    :0 fhw
    | formail -I "X-List-Classify: $LISTID"
}


# Lists that aren't caught by the generic rules.

# Technically, the moderator address isn't a list, but I like to treat it as
# one.
:0 fhw
* ^X-eGroups-Moderators: freecyclebaltimore
| formail -I "X-List-Classify: owner-freecyclebaltimore"

:0 fhw
* ^TO_burningass-owner@playadelfuego.org
| formail -I "X-List-Classify: owner-burningass"

# Grumble, grumble, Mailman 2.1.4, not RFC2919 compliant...
:0 fhw
* ^List-Id: void.slab.org
| formail -I "X-List-Classify: void"

# calefaction, too.
:0 fhw
* ^List-id: calefaction\.
| formail -I "X-List-Classify: calefaction"

# stumpwm-devel, too.
:0 fhw
* ^List-Id: stumpwm-devel\.
| formail -I "X-List-Classify: stumpwm-devel"

:0 fhw
* ^Subject: (Best of )?Internet Oracularities
| formail -I "X-List-Classify: oracle"

:0 fhw
* ^Sender: risks-owner
| formail -I "X-List-Classify: risks"

:0 fhw
* ^Subject: Physics News Update
| formail -I "X-List-Classify: physnews"

:0 fhw
* ^From: OK Go Newsletter
| formail -I "X-List-Classify: okgo"

:0 fhw
* ^Subject: THIS is TRUE
| formail -I "X-List-Classify: thisistrue

:0 fhw
* ^To: fire-conclave-announce
| formail -I "X-List-Classify: fire-conclave"

:0 fhw
* ^Sender: midrange-nontech-bounces
| formail -I "X-List-Classify: midrange-nontech"

:0 fhw
* ^Sender: pasdoc-main-admin
| formail -I "X-List-Classify: pasdoc-main"

:0 fhw
* ^reply-to: otakon-staff-discussion@otakon.org
| formail -I "X-List-Classify: otakon-staff"

:0 fhw
* ^List-Id: <[a-zA-Z0-9_].alerts.google.com>
| formail -I "X-List-Classify: google-alerts"

:0 fhw
* ^X-Mailing-List: ota-staff-info
| formail -I "X-List-Classify: ota-staff-info"

:0 fhw
* ^List-Id:.*[<]talk.openstreetmap.org[>]
| formail -I "X-List-Classify: osm-talk"

:0 fhw
* ^List-Id:.*[<]announce.openstreetmap.org[>]
| formail -I "X-List-Classify: osm-announce"

:0 fhw
* ^List-Id:.*[<]newbies.openstreetmap.org[>]
| formail -I "X-List-Classify: osm-newbies"

:0 fhw
* ^List-Id:.*[<]talk-us.openstreetmap.org[>]
| formail -I "X-List-Classify: osm-talk-us"

:0 fhw
* ^List-Id:.*[<]tagging.openstreetmap.org[>]
| formail -I "X-List-Classify: osm-tagging"

:0 fhw
* ^List-Id:.*[<]dev.openstreetmap.org[>]
| formail -I "X-List-Classify: osm-dev"

:0 fhw
* ^List-Id:.*[<]talk-transit.openstreetmap.org[>]
| formail -I "X-List-Classify: osm-talk-transit"


# Fix some list names.

:0 fhw
* ^X-List-Classify: baltolug_baltinet
| formail -I "X-List-Classify: baltolug"

:0 fhw
* ^X-List-Classify: baltwash-burning-playadelfuego
| formail -I "X-List-Classify: baltwash-burning"

:0 fhw
* ^X-List-Classify: burningass-playadelfuego
| formail -I "X-List-Classify: burningass"

:0 fhw
* ^X-List-Classify: greeters-playadelfuego
| formail -I "X-List-Classify: greeters"

:0 fhw
* ^X-List-Classify: rangers-playadelfuego
| formail -I "X-List-Classify: rangers"

:0 fhw
* ^X-List-Classify: technical-playadelfuego
| formail -I "X-List-Classify: technical"


:0 fhw
* ^X-List-Classify: otakon-staff
| sed -e "s/^\\(Subject:.*\\)\\[Otakon Staff\\] */\\1/I"


# Now actually do something with the email.

:0
* ^X-List-Classify: \/[0-9a-z-]+
{
    LIST=$MATCH

    # First off, junk any duplicate emails, but only if they've been seen
    # before.
    :0 Wh: $MAILDIR/msgids/$LIST.lock
    | formail -D 8192 $MAILDIR/msgids/$LIST.cache

    :0 fhw
    | formail -I X-List-Classify

    # List management emails should go in my inbox.
    :0
    * ^Message-ID: <mailman
    $DEFAULT

    # Likewise, but for ezmlm
    :0
    * ^Message-ID: <[0-9.]+.ezmlm-warn@
    $DEFAULT

    # These are going in separate mailboxes by list, so I don't need [listname]
    # in the subject.  Also, I use mutt's list-reply feature, so I don't need
    # munged Reply-To fields.  (But only remove munged fields.)
    :0 fhw
    | sed -e "s/^\\(Subject:.*\\)\\[$LIST\\]\\( \\)*/\\1/I" -e "/^Reply-To:.*$LIST@/ d" -e "s/^Old-Reply-To:/Reply-To:/"

    # Some lists append a message number to the [listname] in the Subject,
    # which causes the above replacement to fail.  This takes care of those.
    :0
    * $ ^Subject:.*\[$LIST:\/[0-9]+
    {
        :0 fhw
        | formail -I "X-Message-Num: $MATCH"

        :0 fhw
        | sed -re "s/^(Subject:.*)\\[${LIST}:[0-9]+\\] */\\1/I"
    }

    # Now that any list-added Subject: stuff is gone, compress any duplicate
    # Re:s
    :0 fhw
    * ^Subject\: *Re\: *Re\:
    | sed -re "s/^(Subject: *)(Re: *)(Re: *)+/\\1\\2/I"

    # For reading.
    :0 c
    .lists.$LIST/

    # Prepare for the archive.
    :0 fhw
    | formail -I "Received" -I "Delivered-To"

    # Archive with lurker
    :0 wc
    | lurker-index -l $LIST -m

    # Archive locally.
    :0
    .archives.$LIST/
}


:0
* ^Sender: CronDaemon
.cronjobs/

:0
* ^X-Cron-Env
.cronjobs/

:0
* ^Subject: (aragorn|dalejr) [0-9]+[/-][0-9]+[/-][0-9]+ [0-9]+:[0-9]+ (system check|System Events|Security Violations|Security Events)
.logcheck/

:0
* ^Reply-To: reply@friendster.com
.friendster/

:0
* ^From: .*Tribe.net>
.tribe/

:0
* ^From: .*member@orkut.com>
* ^X-Mailer: aspNetEmail
.orkut/

:0
* ^From: MTA <mtainfo@mdot.state.md.us>
.mta/

:0
* ^From: .*ay2ewx5x@facebookmail.com>
.facebook/


# Get rid of pesky viruses.

# Some rules here courtesy of http://www.xs4all.nl/~rsmith/spamblock.html

# Filter the W32/KLEZ,W32/Yaha, W32/Sobig.F viruses.
:0 HB
* ^Content-Type: multipart/(mixed|alternative)
* ^Content-Type:.*(audio/x-|application|x-rasmol)
* name=.*\.(scr|com|bat|pif|lnk|exe)
{
  LOG="(virus) [KLEZ/YAHA/Sobig.F] "

  :0
  /dev/null
}

:0
* ^X-MailScanner: Found to be clean
{
  LOG="(virus) [Sobig.F] "

  :0
  /dev/null
}

# WIN32 programs. This should catch them all, since it detects the
# bas64 encoded executable header.
:0 B
* ^Content-Transfer-Encoding:.*base64
* ^TVqQAAMAAAAEAAAA//8AALg
* 4fug4AtAnNIbg
{
  LOG="(virus) [win32 exe] "

  :0
  /dev/null
}


# SpamAssassin stuff.

# First, run the email through SpamAssassin
:0 fw
| spamc

# Note the spam score
:0 hc
| process_spam_score

# Save it if it's really highly scored.
# Current best: 82.2 with v3.1.4
:0
* ^X-Spam-Status: Yes, score=\/[0-9-]+
* ? test $MATCH -gt 82
.notable-spam/

# Drop all other spam over score 10.
:0
* ^X-Spam-Status: Yes, score=\/[0-9-]+
* ? test $MATCH -gt 10
/dev/null

# Keep the rest for double-checking.
:0
* ^X-Spam-Status: Yes
.spam/


# And run the rest through lbdb to capture email addresses.
:0 hc
| lbdb-fetchaddr -a