script to convert mdir to mbox format


#! /bin/sh
#
# Get a directory name as input and convert all mail files inside
# to mbox format
#
# NOTE: processing of subdirectories not yet implemented correctly:
# all mails in subfolders are put into the same mbox
# (it would be better if an mbox file will be generated for
# each subfolder)
# NOTE: calculation of message date in case of 'From:' doesn't recognise
# time zones
#
# History:
# Feb 06 2001 Joerg Reinhardt
# - first edition
# Feb 07 2001 Joerg Reinhardt
# - added usage output
# Feb 12 2001 Joerg Reinhardt
# - mails not containing a 'From:' field but an 'X-From-Line:' or a
# 'Reply-To:' field are now recognised and also processed (e.g. put into
# the mbox file); this works fine for all my mails
# - added progress information
# - warning about corrupt files is now written to stderr

# check for argument or help argument respectively
if [[ ($1 == "") ||
  ($1 == "-h") ||
  ($1 == "--help") ||
  ($1 == "-help") ]]; then
  echo "Usage: "$0" <Xfmail-mail-directory>";
fi;

# check if parameter is a directory
if [[ -d $1 ]]; then
# set target filename
  dirname=`echo $1 | awk '{while(substr($0,length($0),1)=="/"){$0=substr($0,1,length($0)-1);}print $0;}'`;
  mboxfile=$dirname'.mbox';

# check if directory is empty
  if [[ `find $dirname -type f` == "" ]]; then
 echo $dirname": directory empty."
 exit 1;
  fi;

# prevent automatic overwriting of target
  if [[ -e $mboxfile ]]; then \
 dialogtext="Write file "$mboxfile"?";
 if dialog --yesno "$dialogtext" 10 60; then
  clear;
  rm -vf $mboxfile;
 else
  clear; exit 1;
 fi;
  fi;

  echo "writing xfmail mail directory '$1' to '$mboxfile'.";


# collect files inside Xfmail mail-directory and produce MBOX format
# target file
  for i in `find $1/* -type f`; do
# output progress information
 echo -n -e \\r" "
 echo -n -e \\rprocessing $i
# look for senders email address in the order
# 'From:'
# 'X-From-Line:'
# 'Reply-To:'
 shortfromflag='true';
 fromline=`grep 'From:' $i`;
# parse 'From:' field
 from=`echo $fromline | awk 'BEGIN{FS="<";}{if($0~/</) {pos=index($2,">");if(pos!=0) {print substr($2,1,pos-1);}} else {pos=index($0,":");print substr($0,pos+1);}}'`;
 if [[ $from == "" ]]; then
  shortfromflag='false';
  fromline=`grep 'X-From-Line:' $i`;
  from=`echo $fromline | awk 'BEGIN{FS="Line:";}{print $2;}'`;
  if [[ $from == "" ]]; then
  shortfromflag='true';
  fromline=`grep 'Reply-To:' $i`;
# parse 'Reply-To:' field
  from=`echo $fromline | awk 'BEGIN{FS="<";}{if($0~/</) {pos=index($2,">");if(pos!=0) {print substr($2,1,pos-1);}} else {pos=index($0,":");print substr($0,pos+1);}}'`;
  if [[ $from == "" ]]; then
  echo;
  echo "WARNING: "$i": no 'From:' nor 'X-From-Line:' nor 'Reply-To:' field found." >&2;
  continue;
  fi;
  fi;
 fi;
 if [[ $shortfromflag == "true" ]]; then
# parse date field
  dateline=`grep 'Date:' $i`;
  if [[ $dateline == "" ]]; then
# set dummy date if no date field found
  dateline="Date: Thu, 01 Jan 1970 00:00:00 +0000 (GMT)";
  fi;
  weekday=`echo $dateline | awk '{gsub(/,/,"",$2);print $2;}'`;
  day=`echo $dateline | awk '{print $3;}'`;
  month=`echo $dateline | awk '{print $4;}'`;
  year=`echo $dateline | awk '{print $5;}'`;
  time=`echo $dateline | awk '{print $6;}'`;
  diffGMT=`echo $dateline | awk '{print $7;}'`;
  timezone=`echo $dateline | awk '{print $8;}'`;

# output MBOX mail header
  echo "From " $from $weekday $month $day $time $year >> $mboxfile;
 else
# output long MBOX mail header found in 'X-From-Line:' field
  echo $from >> $mboxfile;
 fi;

# output mail itself
 cat $i >> $mboxfile;
  done;
  echo;
else
  echo $1": not a directory.";
fi;

No comments:

Other Articles

Enter your email address: