mbox2sympa.pl

#!/usr/bin/perl
# Archive converter from standard mbox format to Sympa's archive format,
# creating the Sympa compatible log.* files in the same directory as the
# input files.  When done here, copy the log.* files to the "archives"
# subdirectory of the list's directory under ~sympa/expl, and then run
# the Sympa distribution script ~sympa/bin/arc2webarc.pl.
#
# Note: Should work on listproc or majordomo archives too, maybe others.
#
# Usage:
#        mbox2sympa.pl <input>
#
#        <input> can be either a single mbox file or the name of a
#        directory containing files in mbox format, with arbitrary names.
#
# Written by Adam Bernstein (adam@amberbug.org), 11/08/2002
#   (pardon my beginner's Perl syntax -- I'm really a C/C-shell programmer)
# Expanded from mjarc2sympa.pl by Petr Prazak <prazak@grisoft.cz>
 
use File::Find;
 
my %months = ('Jan',1,'Feb',2,'Mar',3,'Apr',4,'May',5,'Jun',6,'Jul',7,'Aug',8,'Sep',9,'Oct',10,'Nov',11,'Dec',12);
 
my $separator = "\n------- CUT --- CUT\n\n";
my $header = "\n------- THIS IS A RFC934 COMPLIANT DIGEST\n\n";
 
my $outdir = ".";
my $total = 0;      #total converted messages
 
# Start by assuming the most standard From line date format, which seems
# to be like "From <...> Wed Jan 01 00:00:00 2000" (different than the
# standard format in the Date: line):
my $month_index = 3;
my $year_index = 6;
my $num_fields = 7;
 
my $arg = $ARGV[0];
 
if (-d $arg) {
   print "Processing the directory $arg\n";
   $outdir = $arg;
   my $pth = $arg . "/*";
   @filelist = glob($pth);
   set_date_format($filelist[0]);
   foreach $file  ( @filelist )  {
      next if ($file =~ /^\./);
      process_file($file) if ( -f $file && -r $file);
   }
   print "\nConverted $total messages in total.\n";
} elsif (-r $arg) {
    set_date_format($arg);
    process_file($arg);
} else {
    print STDERR "Bad argument $arg, not a file or directory\n";
}
 
sub process_file {
   my $mj_file = shift;
   my $year, $month, $list;
 
   open FH, "<$mj_file" || die "Cannot open $mj_file: $!";
 
   print "Converting mbox archive for file $mj_file\n";
   my $prev_month = 0;
   my $prev_year = 0;
   my $mailcount = 0;
   my $subtotal = 0;
 
   while (<FH>) {
# Look for "From " lines, doing a basic check against the number of fields
# in the line to avoid false hits (ie. message text starting with "From "):
#       if ((/^From /) && (split==$num_fields)){   ----- modification CB
       if ((/^From -/) && (split==$num_fields)){
         @array = split;
         $month = $months{$array[$month_index]};
         die "Sorry, month not ready correctly -- probably a date format problem.\nRun again and set date format interactively.\n" if ( $month == NULL );
         $year = $array[$year_index];
         $outfile = sprintf "%s/log.%d%02d",$outdir,$year,$month;
 
# Reset things and make a new output file when a new month starts:
         if ( $month != $prev_month ) {
             if ( $prev_month != 0 ) {
                 print OUT $separator if ($mailcount > 0);
                 close OUT;
                 print "Archived $mailcount messages for $prev_month/$prev_year.\n";
             }
             $mailcount = 0;
             $newfile = 1;
             $newfile = 0 if (-e $outfile);
             open OUT, ">>$outfile" || die "Cannot open output file $outfile: $!";
             if ( $newfile ) {
                 print "print header\n";
                 print OUT $header;
             }
             $first = 1;
         }
         if ($first) {
             $first = 0;
         }
         else {
             print OUT $separator;
         }
         ++$mailcount;
         ++$subtotal;
         $prev_month = $month;
         $prev_year = $year;
     }
     print OUT $_;
 }
   print OUT $separator;
   close FH;
   close OUT;
   $total += $subtotal;
   print "Archived $mailcount messages for $month/$year,\n $subtotal messages total\n";
}
 
sub set_date_format {
    my $file = shift;
 
    print "Do you want to interactively set the date format? (y/n) [n]: ";
    if (<STDIN> =~ /y/) {
        open FH, "<$file" || die "***Cannot open $file: $!\n";
        while (<FH>) {
            last if (/^From /);
        }
        close FH;
        print "Here's the first From line in the file:\n";
        print "\n    $_\n";
        print "Does this line contain a date? (y/n) [y]: ";
        if ( <STDIN> =~ /n/) {
            die "***Can't proceed without dates in From lines\n";
            }
        $num_fields=split;
        print "\n     @_[0] xxxxx @_[2] @_[3] @_[4] @_[5] @_[6]\n\n";
        print "Counting the first field as 0, which field in this line gives the month? ";
        $month_index=<STDIN>;
        while ( $months{@_[$month_index]} == NULL) {
            print "\nThat doesn't seem right -- month must be \"Jan\", \"Feb\", etc.\n";
            print "Try again, counting the first field as 0: ";
            $month_index=<STDIN>;
        }
        print "Which field in this line gives the year? ";
        $year_index = <STDIN>;
    }
}