| 1 |
#!/usr/bin/perl -w |
| 2 |
|
| 3 |
BEGIN { |
| 4 |
my $basedir = readlink($0) || $0; $basedir =~ s#/[^/]+$#/lib#; |
| 5 |
unshift(@INC, $basedir); |
| 6 |
} |
| 7 |
|
| 8 |
=head1 NAME |
| 9 |
|
| 10 |
mbox2index.pl - indexing script for Mail::Box Web Search |
| 11 |
|
| 12 |
=head1 SYNOPSYS |
| 13 |
|
| 14 |
mbox2index.pl [local.conf] |
| 15 |
|
| 16 |
=head1 DESCRIPTION |
| 17 |
|
| 18 |
This script will index mailboxes defined in C<global.conf> or local |
| 19 |
configuration file supplied at command line. |
| 20 |
|
| 21 |
In normal operation, using MWS::SWISH it will exec swish-e which will in |
| 22 |
turn again call this script, but this time with C<--recursive> option. |
| 23 |
|
| 24 |
=head1 SEE ALSO |
| 25 |
|
| 26 |
C<MWS> perl modules which are part of this package |
| 27 |
|
| 28 |
=cut |
| 29 |
|
| 30 |
#use MWS::SWISH; |
| 31 |
use MWS::Plucene; |
| 32 |
use Data::Dumper; |
| 33 |
use Date::Parse; |
| 34 |
use POSIX qw(strftime); |
| 35 |
use Getopt::Long; |
| 36 |
|
| 37 |
# are we called from this script? |
| 38 |
my $recursive = 0; |
| 39 |
|
| 40 |
my $r = GetOptions("recursive" => \$recursive); |
| 41 |
|
| 42 |
my $config_file = shift @ARGV || 'global.conf'; |
| 43 |
|
| 44 |
if (! -f $config_file) { |
| 45 |
print qq{Usage: $0 [/path/to/local.conf] |
| 46 |
|
| 47 |
If local.conf is not specified, global.conf in current directory will |
| 48 |
be used. |
| 49 |
}; |
| 50 |
exit 1; |
| 51 |
} |
| 52 |
|
| 53 |
#my $mws = MWS::SWISH->new(config_file => $config_file); |
| 54 |
my $mws = MWS::Indexer->new(config_file => $config_file); |
| 55 |
|
| 56 |
$mws->create_index if (! $recursive); |
| 57 |
|
| 58 |
print STDERR "starting indexing...\n"; |
| 59 |
|
| 60 |
my $debug = 1; |
| 61 |
|
| 62 |
foreach my $mbox ($mws->{config}->Parameters('folders')) { |
| 63 |
my $mbox_path = $mws->{config}->val('folders', $mbox); |
| 64 |
|
| 65 |
print STDERR "working on $mbox [$mbox_path]\n" if ($debug); |
| 66 |
|
| 67 |
my $folder = $mws->open_folder($mbox); |
| 68 |
|
| 69 |
my $total = scalar $folder->messageIds; |
| 70 |
|
| 71 |
print STDERR "$total messages\n" if ($debug); |
| 72 |
|
| 73 |
my $count = 0; |
| 74 |
|
| 75 |
foreach my $message ($folder->messages) { |
| 76 |
|
| 77 |
my $id = $message->messageId; |
| 78 |
|
| 79 |
my $document = { |
| 80 |
id => $id, |
| 81 |
folder => $mbox, |
| 82 |
}; |
| 83 |
|
| 84 |
foreach my $direction (qw(to from cc bcc)) { |
| 85 |
foreach my $part (qw(phrase address comment)) { |
| 86 |
my @data = $mws->unroll($message,$direction,$part); |
| 87 |
if (@data) { |
| 88 |
$document->{$direction.'_'.$part} = join("##", @data); |
| 89 |
$document->{$direction.'_'.$part} =~ s/\s*\(e\s*-\s*mail\)\s*//gi; |
| 90 |
} |
| 91 |
} |
| 92 |
} |
| 93 |
|
| 94 |
$document->{'subject'} = $mws->decode_qp($message->subject) || 'no subject'; |
| 95 |
|
| 96 |
$document->{'body'} = $mws->plain_text_body($message); |
| 97 |
|
| 98 |
my $utime = $message->timestamp; |
| 99 |
|
| 100 |
$document->{'date_utime'} = $utime; |
| 101 |
$document->{'date'} = strftime("%Y-%m-%d %H:%M:%S", localtime($utime)) if ($utime); |
| 102 |
|
| 103 |
# print Dumper($document); |
| 104 |
$mws->add_index("$mbox $id" => $document); |
| 105 |
|
| 106 |
# clear internal MWS cache to keep memory usage down |
| 107 |
# (this should be replaced by garbage collector in MWS, |
| 108 |
# but without it this is the best solution to keep machine |
| 109 |
# alive while indexing) |
| 110 |
$mws->{cache} = {}; |
| 111 |
|
| 112 |
# this is not complete solution. see mailbox-destruct.diff |
| 113 |
$message->destruct(); |
| 114 |
|
| 115 |
$count++; |
| 116 |
printf STDERR "%d messages in $mbox done [%d %%]\n",$count,($count * 100/$total) if ($count % 100 == 0); |
| 117 |
|
| 118 |
} |
| 119 |
|
| 120 |
$mws->close_folder($mbox); |
| 121 |
|
| 122 |
} |
| 123 |
|
| 124 |
$mws->close_index; |
Parent Directory
| 
Revision Log