mws/trunk/mbox2index.pl

#!/usr/bin/perl -w

use MWS;
use Data::Dumper;
use XML::Simple;
use Text::Iconv;
use Date::Parse;
use POSIX qw(strftime);

my $iso2utf = Text::Iconv->new('ISO-8859-2','UTF-8');

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);

        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)) {
                                my @mail_fullnames = $mws->unroll($message,$direction,$part);
                                $document->{$direction.'_'.$part} = join("##",@mail_fullnames);
                        }
                }

                $document->{'subject'} = $message->get('Subject') || 'no subject';
                my $body = $mws->plain_text_body($message);
                $body =~ s,[\n\r],<br/>,gs;

                $document->{'body'} = $body;

                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);

#               my $xml = XMLout($document);

                my $xml = qq{<message>};
                foreach my $tag (keys %$document) {
                        $xml .= "<$tag><![CDATA[".$document->{$tag}."]]></$tag>\n";
                }
                $xml .= qq{</message>};

                $xml = $iso2utf->convert($xml);
                use bytes;      # as opposed to chars
                print "Path-Name: $mbox $id\n";
                print "Content-Length: ".(length($xml)+1)."\n";
                print "Document-Type: XML\n\n$xml\n";

        }

        $folder->close;
}

$mws->close_index;
1	dpavlin	12	#!/usr/bin/perl -w
2
3			use MWS;
4			use Data::Dumper;
5			use XML::Simple;
6			use Text::Iconv;
7	dpavlin	13	use Date::Parse;
8			use POSIX qw(strftime);
9	dpavlin	12
10			my $iso2utf = Text::Iconv->new('ISO-8859-2','UTF-8');
11
12			my $mws = MWS->new('global.conf');
13
14			my $debug = 1;
15
16			foreach my $mbox ($mws->{config}->Parameters('folders')) {
17			my $mbox_path = $mws->{config}->val('folders', $mbox);
18
19			print STDERR "working on $mbox [$mbox_path]\n" if ($debug);
20
21			my $folder = $mws->open_folder($mbox);
22
23			foreach my $message ($folder->messages) {
24
25			my $id = $message->messageId;
26
27			my $document = {
28			id => $id,
29			folder => $mbox,
30			};
31
32			foreach my $direction (qw(to from cc bcc)) {
33	dpavlin	13	foreach my $part (qw(phrase address comment)) {
34			my @mail_fullnames = $mws->unroll($message,$direction,$part);
35			$document->{$direction.'_'.$part} = join("##",@mail_fullnames);
36	dpavlin	12	}
37			}
38
39			$document->{'subject'} = $message->get('Subject') \|\| 'no subject';
40	dpavlin	13	my $body = $mws->plain_text_body($message);
41			$body =~ s,[\n\r],<br/>,gs;
42	dpavlin	12
43	dpavlin	13	$document->{'body'} = $body;
44
45			my $utime = str2time($message->date);
46
47			$document->{'date_utime'} = $utime;
48			$document->{'date'} = strftime("%Y-%m-%d %H:%M:%S", localtime($utime));
49
50	dpavlin	12	# print Dumper($document);
51			# $mws->add_index("$mbox $id" => $document);
52
53			# my $xml = XMLout($document);
54
55			my $xml = qq{<message>};
56			foreach my $tag (keys %$document) {
57			$xml .= "<$tag><![CDATA[".$document->{$tag}."]]></$tag>\n";
58			}
59			$xml .= qq{</message>};
60
61			$xml = $iso2utf->convert($xml);
62			use bytes; # as opposed to chars
63			print "Path-Name: $mbox $id\n";
64			print "Content-Length: ".(length($xml)+1)."\n";
65			print "Document-Type: XML\n\n$xml\n";
66
67			}
68
69			$folder->close;
70			}
71
72			$mws->close_index;