mbox-split.pl

#!/usr/bin/perl

use strict;
use warnings;

use Getopt::Long;
use Mail::Mbox::MessageParser;

use MIME::Parser;
use Date::Parse;

my $input_file = undef;
my $out_dir    = undef;
my $start_time = undef;
my $end_time   = undef;

GetOptions(
   "mbox=s"   => \$input_file,
   "out=s"    => \$out_dir,
   "start=s"  => sub { $start_time = str2time($_[1]); },
   "end=s"    => sub { $end_time   = str2time($_[1]); },
);

die "Input mailbox not specified" unless $input_file;
die "Output directory not specified" unless $out_dir;

mkdir $out_dir;

my $mbox_fh = new FileHandle($input_file);
my $mbox    = new Mail::Mbox::MessageParser({
   file_name => $input_file,
   file_handle => $mbox_fh,
   enable_cache => 0, # super buggy
   enable_grep => 1,
});

my $parser  = new MIME::Parser;
$parser->decode_headers(1);

while (!$mbox->end_of_file()) {
   my $mime_raw = $mbox->read_next_email();
   my $entity   = $parser->parse_data($$mime_raw);
   my $msg_id   = $entity->head()->get('Message-ID', 0);
   my $date     = $entity->head()->get('Date', 0);
   my $udate    = str2time($date);

   chomp $msg_id;

   my $keep     = 1;

   if ($start_time && $end_time) {
      $keep = 0;
      $keep = ($udate >= $start_time) && ($udate <= $end_time);
   } else {
      $keep = 0;
      $keep ||= $udate >= $start_time if $start_time;
      $keep ||= $udate <= $end_time   if $end_time;
   }

   if ($keep || !$date) {
      print "Writing $msg_id to $out_dir/$msg_id.eml\n";
      open my $out_file, ">$out_dir/$msg_id.eml";
      print $out_file $$mime_raw;
      close $out_file;
   }
}