mws/trunk/mbox2index.pl

#!/usr/bin/perl -w

use MWS;
use Data::Dumper;
use Date::Parse;
use POSIX qw(strftime);

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 = $mws->open_folder($mbox);

        print STDERR $folder->size," bytes\n" if ($debug);

        foreach my $message ($folder->messages) {

                my $id = $message->messageId;

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

                foreach my $direction (qw(to from cc bcc)) {
                        foreach my $part (qw(phrase address comment)) {
                                $document->{$direction.'_'.$part} = join("##", $mws->unroll($message,$direction,$part));
                        }
                }

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

                $document->{'body'} = $mws->plain_text_body($message);

                my $utime = str2time($message->date);

                $document->{'date_utime'} = $utime;
                $document->{'date'} = strftime("%Y-%m-%d %H:%M:%S", localtime($utime));

#               print Dumper($document);
                $mws->add_index("$mbox $id" => $document);

                # clear internal MWS cache to keep memory usage down
                # (this should be replaced by garbage collector in MWS,
                # but without it this is the best solution to keep machine
                # alive while indexing)
                $mws->{cache} = {};

                undef $message;

        }

        $mws->close_folder($mbox);
        #$folder->close;
}

$mws->close_index;
1	dpavlin	12	#!/usr/bin/perl -w
2
3			use MWS;
4			use Data::Dumper;
5	dpavlin	13	use Date::Parse;
6			use POSIX qw(strftime);
7	dpavlin	12
8			my $mws = MWS->new('global.conf');
9
10			my $debug = 1;
11
12			foreach my $mbox ($mws->{config}->Parameters('folders')) {
13			my $mbox_path = $mws->{config}->val('folders', $mbox);
14
15			print STDERR "working on $mbox [$mbox_path]\n" if ($debug);
16
17			my $folder = $mws->open_folder($mbox);
18
19	dpavlin	14	print STDERR $folder->size," bytes\n" if ($debug);
20
21	dpavlin	12	foreach my $message ($folder->messages) {
22
23			my $id = $message->messageId;
24
25			my $document = {
26			id => $id,
27			folder => $mbox,
28			};
29
30			foreach my $direction (qw(to from cc bcc)) {
31	dpavlin	13	foreach my $part (qw(phrase address comment)) {
32	dpavlin	14	$document->{$direction.'_'.$part} = join("##", $mws->unroll($message,$direction,$part));
33	dpavlin	12	}
34			}
35
36			$document->{'subject'} = $message->get('Subject') \|\| 'no subject';
37
38	dpavlin	14	$document->{'body'} = $mws->plain_text_body($message);
39	dpavlin	13
40			my $utime = str2time($message->date);
41
42			$document->{'date_utime'} = $utime;
43			$document->{'date'} = strftime("%Y-%m-%d %H:%M:%S", localtime($utime));
44
45	dpavlin	12	# print Dumper($document);
46	dpavlin	14	$mws->add_index("$mbox $id" => $document);
47	dpavlin	12
48	dpavlin	14	# clear internal MWS cache to keep memory usage down
49			# (this should be replaced by garbage collector in MWS,
50			# but without it this is the best solution to keep machine
51			# alive while indexing)
52			$mws->{cache} = {};
53	dpavlin	12
54	dpavlin	14	undef $message;
55	dpavlin	12
56			}
57
58	dpavlin	14	$mws->close_folder($mbox);
59			#$folder->close;
60	dpavlin	12	}
61
62			$mws->close_index;