gma.sh

   1 #!/bin/bash
   2
   3 #########################################################################
   4 # This program is free software: you can redistribute it and/or modify  #
   5 # it under the terms of the version 3 of the GNU General Public License #
   6 # as published by the Free Software Foundation.                         #
   7 #                                                                       #
   8 # This program is distributed in the hope that it will be useful, but   #
   9 # WITHOUT ANY WARRANTY; without even the implied warranty of            #
  10 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU      #
  11 # General Public License for more details.                              #
  12 #                                                                       #
  13 # You should have received a copy of the GNU General Public License     #
  14 # along with this program. If not, see <http://www.gnu.org/licenses/>.  #
  15 #                                                                       #
  16 # Written by and Copyright (C) Francois Fleuret                         #
  17 # Contact <francois.fleuret@idiap.ch> for comments & bug reports        #
  18 #########################################################################
  19
  20 set -e
  21
  22 function usage () {
  23     echo "gma.sh [--body] [--keepspam] [--subject|--from|--to|--fromto] <pattern> [<year>]"
  24 }
  25
  26 RESULT_MBOX="/tmp/gma"
  27
  28 if [[ -z "${MAIL_ARCHIVE_DIR}" ]]; then
  29     echo "Undefined \$MAIL_ARCHIVE_DIR" >&2
  30     exit 1
  31 fi
  32
  33 if [[ $(which ionice) ]]; then
  34     IONICE="ionice -c3"
  35 else
  36     IONICE=""
  37 fi
  38
  39 HEADER_ONLY="-H"
  40
  41 while [[ $1 ]]; do
  42
  43     case $1 in
  44
  45         "--help"|"-h")
  46             usage
  47             exit 0
  48             ;;
  49
  50         "--body")
  51             HEADER_ONLY=""
  52             ;;
  53
  54         "--intense")
  55             IONICE=""
  56             ;;
  57
  58         "--keepspam")
  59             KEEP_SPAM=1
  60             ;;
  61
  62         "--from")
  63             RE_PREFIX="^From:.*"
  64             ;;
  65
  66         "--to")
  67             RE_PREFIX="^To:.*"
  68             ;;
  69
  70         "--fromto")
  71             RE_PREFIX="^(From|To):.*"
  72             ;;
  73
  74         "--subject")
  75             RE_PREFIX="^Subject:.*"
  76             ;;
  77
  78         *)
  79             if [[ ${PATTERN} ]]; then
  80                 # If we already have the pattern to match in the
  81                 # message, get the new argument as a pattern for the
  82                 # filename
  83                 if [[ ${FILE_PATTERN} ]]; then
  84                     usage
  85                     exit 1
  86                 else
  87                     FILE_PATTERN="${MAIL_ARCHIVE_DIR}/${1}/*/*"
  88                     SIZE_FILE_PATTERN="${MAIL_ARCHIVE_DIR}/${1}"
  89                 fi
  90             else
  91                 PATTERN=$1
  92             fi
  93             ;;
  94     esac
  95     shift
  96 done
  97
  98 if [[ -z "${PATTERN}" ]]; then
  99     usage
 100     exit 1
 101 fi
 102
 103 [[ ${FILE_PATTERN} ]] || FILE_PATTERN="${MAIL_ARCHIVE_DIR}/*/*/*"
 104 [[ ${SIZE_FILE_PATTERN} ]] || SIZE_FILE_PATTERN="${MAIL_ARCHIVE_DIR}"
 105
 106 SIZE=$(\dus ${SIZE_FILE_PATTERN} | cut -f 1 -d" ")
 107
 108 echo "Looking for ${RE_PREFIX}${PATTERN}"
 109 echo "        in ${FILE_PATTERN}"
 110 echo "        writing result in ${RESULT_MBOX}"
 111
 112 if [[ ${KEEP_SPAM} ]]; then
 113
 114     ${IONICE} cat ${FILE_PATTERN} \
 115         | pv -s ${SIZE} -p -t -e - \
 116         | mboxgrep ${HEADER_ONLY} -i "${RE_PREFIX}${PATTERN}" > ${RESULT_MBOX}
 117
 118 else
 119
 120     ${IONICE} cat ${FILE_PATTERN} \
 121         | pv -s ${SIZE} -p -t -e - \
 122         | mboxgrep ${HEADER_ONLY} -i "${RE_PREFIX}${PATTERN}" \
 123         | mboxgrep ${HEADER_ONLY} -v '^Subject:.*SPAM' > ${RESULT_MBOX}
 124
 125 fi
 126
 127 echo "Found "$(grep ^"From " ${RESULT_MBOX} | wc -l)" messages."