mws/trunk/mbox2plucene.pl

#!/usr/bin/perl -w

use MWS;
use Data::Dumper;

my $mws = MWS->new('global.conf');

my $debug = 1;

foreach my $mbox ($mws->{config}->Parameters('folders')) {
        my $mbox_path = $mws->{config}->val('folders', $mbox);

        print STDERR "working on $mbox [$mbox_path]\n" if ($debug);

        my $folder = $mws->{mgr}->open($mbox_path, extract => 'LAZY');
        foreach my $message ($folder->messages) {

                my $id = $message->messageId;
                print "## $id\n";

                my $document = {
                        id => $id,
                        folder => $mbox,
                };


                foreach my $direction (qw(to from cc bcc)) {
                        foreach my $f ($message->$direction) {
                                print "$direction: ";
                                foreach my $part (qw(phrase address comment)) {
                                        next if (! $f->$part);
                                        print "[$part]",$f->$part," ";
                                        $document->{$direction.'_'.$part} .= $f->$part."\n";
                                }
                                print "\n";
        #                       push @{$document->$d.'_phrase'}, $f->phrase,
        #                       push @{$document->$d.'_address'}, $f->address,
        ##                      push @{$document->$d.'_comment'}, $f->comment,
                        }
                }

                $document->{'subject'} = $message->get('Subject') || 'no subject';

#               print Dumper($document);
                $mws->{index}->add("$mbox $id" => $document);

        }

        $folder->close;
}

$mws->{index}->optimize;
1	dpavlin	1	#!/usr/bin/perl -w
2
3	dpavlin	4	use MWS;
4	dpavlin	1	use Data::Dumper;
5
6	dpavlin	4	my $mws = MWS->new('global.conf');
7	dpavlin	1
8	dpavlin	4	my $debug = 1;
9	dpavlin	1
10	dpavlin	4	foreach my $mbox ($mws->{config}->Parameters('folders')) {
11			my $mbox_path = $mws->{config}->val('folders', $mbox);
12	dpavlin	1
13	dpavlin	4	print STDERR "working on $mbox [$mbox_path]\n" if ($debug);
14	dpavlin	1
15	dpavlin	4	my $folder = $mws->{mgr}->open($mbox_path, extract => 'LAZY');
16			foreach my $message ($folder->messages) {
17	dpavlin	1
18	dpavlin	4	my $id = $message->messageId;
19			print "## $id\n";
20	dpavlin	1
21	dpavlin	4	my $document = {
22			id => $id,
23			folder => $mbox,
24			};
25
26
27
28			foreach my $direction (qw(to from cc bcc)) {
29			foreach my $f ($message->$direction) {
30			print "$direction: ";
31			foreach my $part (qw(phrase address comment)) {
32			next if (! $f->$part);
33			print "[$part]",$f->$part," ";
34			$document->{$direction.'_'.$part} .= $f->$part."\n";
35			}
36			print "\n";
37			# push @{$document->$d.'_phrase'}, $f->phrase,
38			# push @{$document->$d.'_address'}, $f->address,
39			## push @{$document->$d.'_comment'}, $f->comment,
40	dpavlin	1	}
41			}
42
43	dpavlin	4	$document->{'subject'} = $message->get('Subject') \|\| 'no subject';
44	dpavlin	1
45	dpavlin	4	# print Dumper($document);
46			$mws->{index}->add("$mbox $id" => $document);
47	dpavlin	1
48	dpavlin	4	}
49
50			$folder->close;
51	dpavlin	1	}
52
53	dpavlin	4	$mws->{index}->optimize;