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 |
27 |
use Getopt::Long; |
8 |
dpavlin |
12 |
|
9 |
dpavlin |
27 |
# are we called from this script? |
10 |
|
|
my $recursive = 0; |
11 |
dpavlin |
12 |
|
12 |
dpavlin |
27 |
my $r = GetOptions("recursive" => \$recursive); |
13 |
|
|
|
14 |
|
|
my $config_file = shift @ARGV || 'global.conf'; |
15 |
|
|
|
16 |
|
|
if (! -f $config_file) { |
17 |
|
|
print qq{Usage: $0 [/path/to/local.conf] |
18 |
|
|
|
19 |
|
|
If local.conf is not specified, global.conf in current directory will |
20 |
|
|
be used. |
21 |
|
|
}; |
22 |
|
|
exit 1; |
23 |
|
|
} |
24 |
|
|
|
25 |
|
|
my $mws = MWS->new($config_file); |
26 |
|
|
|
27 |
|
|
$mws->create_index if (! $recursive); |
28 |
|
|
|
29 |
|
|
print STDERR "starting indexing..."; |
30 |
|
|
|
31 |
dpavlin |
12 |
my $debug = 1; |
32 |
|
|
|
33 |
|
|
foreach my $mbox ($mws->{config}->Parameters('folders')) { |
34 |
|
|
my $mbox_path = $mws->{config}->val('folders', $mbox); |
35 |
|
|
|
36 |
|
|
print STDERR "working on $mbox [$mbox_path]\n" if ($debug); |
37 |
|
|
|
38 |
|
|
my $folder = $mws->open_folder($mbox); |
39 |
|
|
|
40 |
dpavlin |
14 |
print STDERR $folder->size," bytes\n" if ($debug); |
41 |
|
|
|
42 |
dpavlin |
12 |
foreach my $message ($folder->messages) { |
43 |
|
|
|
44 |
|
|
my $id = $message->messageId; |
45 |
|
|
|
46 |
|
|
my $document = { |
47 |
|
|
id => $id, |
48 |
|
|
folder => $mbox, |
49 |
|
|
}; |
50 |
|
|
|
51 |
|
|
foreach my $direction (qw(to from cc bcc)) { |
52 |
dpavlin |
13 |
foreach my $part (qw(phrase address comment)) { |
53 |
dpavlin |
14 |
$document->{$direction.'_'.$part} = join("##", $mws->unroll($message,$direction,$part)); |
54 |
dpavlin |
12 |
} |
55 |
|
|
} |
56 |
|
|
|
57 |
dpavlin |
17 |
$document->{'subject'} = $mws->decode_qp($message->subject) || 'no subject'; |
58 |
dpavlin |
12 |
|
59 |
dpavlin |
14 |
$document->{'body'} = $mws->plain_text_body($message); |
60 |
dpavlin |
13 |
|
61 |
|
|
my $utime = str2time($message->date); |
62 |
|
|
|
63 |
|
|
$document->{'date_utime'} = $utime; |
64 |
|
|
$document->{'date'} = strftime("%Y-%m-%d %H:%M:%S", localtime($utime)); |
65 |
|
|
|
66 |
dpavlin |
12 |
# print Dumper($document); |
67 |
dpavlin |
14 |
$mws->add_index("$mbox $id" => $document); |
68 |
dpavlin |
12 |
|
69 |
dpavlin |
14 |
# clear internal MWS cache to keep memory usage down |
70 |
|
|
# (this should be replaced by garbage collector in MWS, |
71 |
|
|
# but without it this is the best solution to keep machine |
72 |
|
|
# alive while indexing) |
73 |
|
|
$mws->{cache} = {}; |
74 |
dpavlin |
12 |
|
75 |
dpavlin |
15 |
# this is not complete solution. see mailbox-destruct.diff |
76 |
|
|
$message->destruct(); |
77 |
dpavlin |
12 |
|
78 |
|
|
} |
79 |
|
|
|
80 |
dpavlin |
14 |
$mws->close_folder($mbox); |
81 |
dpavlin |
15 |
|
82 |
dpavlin |
12 |
} |
83 |
|
|
|
84 |
|
|
$mws->close_index; |