#!/bin/sh

#  mbox-compress - reduces the file size of mbox mail folder by
#  removing some headers.
#
#  formail(1), part of procmail(1), is used to keep only certain
#  headers, removing all other headers.
#
#  I have hard coded the directoties here to avoid typing errors.
#  This can be easily adjusted for flexible use.  find(1) is used to
#  find the files (mbox format is assumed) in given directoties.


#  Directories in which to find mbox files.
dirs="${HOME}/mail/In*"
dirs="$dirs ${HOME}/mail/Ignore"
dirs="$dirs ${HOME}/mail/Live-elements"

case $# in
  0 )
  ;;

  * )
    echo "No input is accepted; will use files in '$dirs'." 1>&2
    set -
  ;;
esac

botched_copy()
{
  echo "Error occured $@" 1>&2
  echo "$mbox remains intact, botched copy is ${mbox}.new" 1>&2
}

make_list()
{
  flag=$1
  shift
  var=

  case $# in
    0 )
    ;;

    * )
      for v in $@ ; do var="$var $flag $v" ; done
    ;;
  esac

  printf "$var"
}

#  List of headers to keep.
Keep=$( make_list -X \
          Return-Path: Sender: \
          Delivered-To: X-Envelope-To: \
          \
          Date: \
          Subject: \
          From: Reply-To: To: CC: Bcc: \
          Message-ID: In-Reply-To: References: \
          \
          Newsgroups: Followup-to: \
          Mail-Followup-to: \
          \
          X-Bogosity: \
          \
          Mailing-List: List-ID: \
          List-Subscribe: List-Unsubscribe: \
          \
          X-FreeBSD-CVS-Branch: \
          \
          Status: X-Status: \
          Lines: \
          MIME-Version: \
          Content-Length:  Content-Type:  \
          Content-Disposition: \
          Content-Transfer-Encoding: \
          \
          Fax:  X-Fax: Phone:  X-Phone: URL:  X-URL: \
          \
          X-Corp: X-Corp-Fax: X-Corp-Phone: X-Corp-URL: \
          X-Work-Start: X-Work-location: \
          \
          X-pair-Authenticated: \
      )

#  List of headers to remove specifically.
Remove=$( make_list -I X-Face: )

#  Find the files in $dirs, and operate on them.
for mbox in $( find $dirs -type f -not -name 'xx.*' -print )
do
  #  Skip "special" files.
  case $mbox in
    *[Cc][Aa][Cc][Hh][Ee]* | *.db | *.tar | *.bz2 | *.gz | *.tbz2 | *.tgz )
      continue
    ;;
  esac

  [ ! -s "$mbox" -o ! -f "$mbox" -o ! -w "$mbox" ] \
  && { echo "Skipped $mbox ...." 1>&2 ; continue; }

  #echo "Processing $mbox ..."
  [ -e "${mbox}.new" ]  && rm -f "${mbox}.new"

  #  Do the real work, finally.
  if { formail -z -c -U Delivered-To: \
        -k -X 'From ' \
        ${Keep} ${Remove} \
        -s < "$mbox"  >> "${mbox}.new"
     }
  then
    #  touch(1) restores the access & modfication times so that MUA
    #  like mutt(1) don't list the updates files to contain new
    #  messages.
    touch -r "$mbox" "${mbox}.new" \
    && mv -f "${mbox}.new" "$mbox" \
    || botched_copy "while moving ${mbox}.new to $mbox"
  else
    botched_copy "while processing"
  fi

done

echo '... done'