mws/trunk/mbox2index.pl

#!/usr/bin/perl -w

use Mail::Box::Manager;
use Plucene::Simple;
use Data::Dumper;

my $mbox="test";
my $index = Plucene::Simple->open( "index" ) || die "can't open index: $!";

my $mgr    = Mail::Box::Manager->new;
my $folder = $mgr->open($mbox, 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);
        $index->add("$mbox $id" => $document);

}

$folder->close;
$index->optimize;
1	#!/usr/bin/perl -w
2
3	use Mail::Box::Manager;
4	use Plucene::Simple;
5	use Data::Dumper;
6
7	my $mbox="test";
8	my $index = Plucene::Simple->open( "index" ) \|\| die "can't open index: $!";
9
10	my $mgr = Mail::Box::Manager->new;
11	my $folder = $mgr->open($mbox, extract => 'LAZY');
12	foreach my $message ($folder->messages) {
13
14	my $id = $message->messageId;
15	print "## $id\n";
16
17	my $document = {
18	id => $id,
19	folder => $mbox,
20	};
21
22
23
24	foreach my $direction (qw(to from cc bcc)) {
25	foreach my $f ($message->$direction) {
26	print "$direction: ";
27	foreach my $part (qw(phrase address comment)) {
28	next if (! $f->$part);
29	print "[$part]",$f->$part," ";
30	$document->{$direction.'_'.$part} .= $f->$part."\n";
31	}
32	print "\n";
33	# push @{$document->$d.'_phrase'}, $f->phrase,
34	# push @{$document->$d.'_address'}, $f->address,
35	## push @{$document->$d.'_comment'}, $f->comment,
36	}
37	}
38
39	$document->{'subject'} = $message->get('Subject') \|\| 'no subject';
40
41	print Dumper($document);
42	$index->add("$mbox $id" => $document);
43
44	}
45
46	$folder->close;
47	$index->optimize;