--- trunk/mbox2index.pl	2004/05/11 15:36:51	43
+++ trunk/mbox2index.pl	2004/05/25 18:55:46	52
@@ -1,10 +1,32 @@
 #!/usr/bin/perl -w
 
 BEGIN {
-        my $basedir = readlink($0) || $0; $basedir =~ s#/[^/]+$##;
+        my $basedir = readlink($0) || $0; $basedir =~ s#/[^/]+$#lib#;
 	unshift(@INC, $basedir);
 }
 
+=head1 NAME
+
+mbox2index.pl - indexing script for Mail::Box Web Search
+
+=head1 SYNOPSYS
+
+ mbox2index.pl [local.conf]
+
+=head1 DESCRIPTION
+
+This script will index mailboxes defined in C<global.conf> or local
+configuration file supplied at command line.
+
+In normal operation, using MWS::SWISH it will exec swish-e which will in
+turn again call this script, but this time with C<--recursive> option.
+
+=head1 SEE ALSO
+
+C<MWS> perl modules which are part of this package
+
+=cut
+
 use MWS::SWISH;
 #use MWS::Plucene;
 use Data::Dumper;
@@ -44,7 +66,11 @@
 
 	my $folder = $mws->open_folder($mbox);
 
-	print STDERR $folder->size," bytes\n" if ($debug);
+	my $total = scalar $folder->messageIds;
+	
+	print STDERR "$total messages\n" if ($debug);
+
+	my $count = 0;
 
 	foreach my $message ($folder->messages) {
 
@@ -58,7 +84,10 @@
 		foreach my $direction (qw(to from cc bcc)) {
 			foreach my $part (qw(phrase address comment)) {
 				my @data = $mws->unroll($message,$direction,$part);
-				$document->{$direction.'_'.$part} = join("##", @data) if (@data);
+				if (@data) {
+					$document->{$direction.'_'.$part} = join("##", @data);
+					$document->{$direction.'_'.$part} =~ s/\s*\(e\s*-\s*mail\)\s*//gi;
+				}
 			}
 		}
 
@@ -69,7 +98,7 @@
 		my $utime = str2time($message->date);
 
 		$document->{'date_utime'} = $utime;
-		$document->{'date'} = strftime("%Y-%m-%d %H:%M:%S", localtime($utime));
+		$document->{'date'} = strftime("%Y-%m-%d %H:%M:%S", localtime($utime)) if ($utime);
 
 #		print Dumper($document);
 		$mws->add_index("$mbox $id" => $document);
@@ -83,6 +112,9 @@
 		# this is not complete solution. see mailbox-destruct.diff
 		$message->destruct();
 
+		$count++;
+		printf STDERR "%d messages in $mbox done [%d %%]\n",$count,($count * 100/$total) if ($count % 100 == 0);
+
 	}
 
 	$mws->close_folder($mbox);