--- trunk/MWS.pm	2004/05/06 19:46:58	14
+++ trunk/MWS.pm	2004/05/07 20:52:34	19
@@ -10,6 +10,8 @@
 use Config::IniFiles;
 use POSIX qw(strftime);
 use Text::Autoformat;
+use Text::Iconv;
+use Text::Unaccent;
 
 #use MWS_plucene;
 use MWS_swish;
@@ -46,10 +48,56 @@
 	$self->{folder} = {};
 
 	$self->{wrap_margin} = $self->{config}->val('global', 'wrap_margin');
+	$self->{max_results} = $self->{config}->val('global', 'max_results') || 100;
+	$self->reset_counters;
 
 	return $self;
 }
 
+sub normalize_string {
+	my $self = shift;
+
+	my $v = shift || return;
+
+	$v = unac_string('ISO-8859-2', $v);
+	$v = join('',sort split(/\s+/,$v));
+	$v =~ s/\W+//g;
+
+	return $v;
+}
+
+# reset tables for search results
+sub reset_counters {
+	my $self = shift;
+
+	$self->{counter} = {};
+
+#	foreach my $c (qw(thread from to cc bcc lists links att)) {
+#		$self->{counter}->{$c} = {};
+#	}
+
+}
+
+sub add_counter($$) {
+	my $self = shift;
+
+	my ($c,$v) = @_;
+	my $k = $self->normalize_string($v);
+
+	$self->{counter}->{$c}->{$k}->{name} = $v;
+	return $self->{counter}->{$c}->{$k}->{usage}++;
+}
+
+sub counter {
+	my $self = shift;
+
+	my $c = shift || return;
+
+	return if (! $self->{counter}->{$c});
+
+	return $self->{counter}->{$c};
+}
+
 sub mbox_name2path {
 	my $self = shift;
 
@@ -73,6 +121,8 @@
 		print STDERR "open_folder($mbox) ok\n" if ($debug);
 	}
 
+	$self->{fetch_count} = 0;
+
 	return $self->{folder}->{$mbox};
 
 }
@@ -82,7 +132,11 @@
 
 	my $mbox = shift || croak "open_folder needs mbox name";
 
-	return $self->{folder}->{$mbox}->close(write => 'NEVER');
+	$self->{folder}->{$mbox}->close(write => 'NEVER') || croak "can't close folder $mbox";
+
+	# XXX this is rather agressive!!!
+	$self->{folder} = {};
+	return 
 }
 
 sub fetch_message {
@@ -93,8 +147,19 @@
 
 	# return message with ID
 	print STDERR "fetch $id from $mbox\n" if ($debug);
-	return $self->open_folder($mbox)->find($id) ||
+
+	if ($self->{fetch_count}++ > 100) {
+		$self->close_folder($mbox);
+		print STDERR "close_folder($mbox) forced on ",$self->{fetch_count},"iteration\n";
+	}
+	
+	my $msg = $self->open_folder($mbox)->find($id);
+	if ($msg) {
+		return $msg;
+	} else {
 		print STDERR "can't find message $id in $mbox. Time to re-index?\n";
+		return;
+	}
 }
 
 
@@ -103,13 +168,17 @@
 
 	my $s = shift || carp "search called without argument!";
 
+	$self->reset_counters;
+
 	print STDERR "search_index($s)\n" if ($debug == 2);
 	my @index_ids = $self->search_index($s);
 
 	$self->{'index_ids'} = \@index_ids;
 
-	my $results = $#index_ids + 1;
-	$self->{'results'} = $results;
+	#my $results = $#index_ids + 1;
+	#$self->{'results'} = $results;
+	
+	my $results = $self->{'total_hits'} || ($#index_ids + 1);
 
 	$self->{'curr_result'} = 0;
 
@@ -118,6 +187,26 @@
 	return $results || 'error';
 }
 
+sub decode_qp($) {
+	my $self = shift;
+
+	my $tmp = shift || return;
+
+	sub decode($$) {
+		my ($cp,$qp) = @_;
+		my $iconv = Text::Iconv->new($cp,'ISO-8859-2');
+	print STDERR "decode($cp,$qp) -> " if ($debug == 2);
+		$qp =~ s/=([a-f0-9][a-f0-9])/chr(hex($1))/ieg;
+		$qp =~ s/_/ /g;
+	print STDERR "$qp\n" if ($debug == 2);
+		return $iconv->convert($qp) || $qp;
+	}
+
+	$tmp =~ s/=\?([^\?]+)\?Q\?(.+?)\?=/decode($1,$2)/ex;
+	$tmp =~ s/^\s*["']+(.*?)["']+\s*$/$1/g;
+	return $tmp;
+}
+
 sub unroll($$$) {
 	my $self = shift;
 
@@ -125,12 +214,13 @@
 
 	my @arr;
 
+	return if (! $message->$part);
+
 	foreach my $from ($message->$part) {
-		my $tmp = $from->$sub;
-		if ($tmp) {
-			$tmp =~ s/^\s*["'](.*)["']\s*$/$1/;
-			push @arr, $tmp;
-		}
+		my $tmp = $from->$sub || next;
+
+		$tmp = $self->decode_qp($tmp);
+		push @arr, $tmp;
 	}
 
 	return @arr;
@@ -149,6 +239,7 @@
 		push @arr, $self->fetch_result_by_id($id);
 	}
 
+
 	return @arr;
 }
 
@@ -188,7 +279,8 @@
 	# reformat with Text::Autoformat
 	my $wrap = $self->{wrap_margin};
 	if ($wrap && $body && $body =~ m/^.{$wrap}..*$/m) {
-		$body = autoformat $body;
+		$body =~ s/[\r\n]/\n/gs;
+		$body = autoformat($body, {right=>$wrap});
 		$body .="\n[reformated using autoformat, margin at $wrap]" if ($debug == 2);
 	}
 
@@ -207,13 +299,17 @@
 
 		print STDERR "fetch_result_by_id($id) not in cache, hitting disk\n" if ($debug == 2);
 
-		my $message = $self->fetch_message($id) || print STDERR "can't fetch message '$id'";
+		my $message = $self->fetch_message($id) || return;
 
 		$row->{'id'} = $id;
-		@{$row->{'from'}} = $self->unroll($message,'from','phrase');
-		@{$row->{'to'}} = $self->unroll($message,'to','phrase');
-		@{$row->{'cc'}} = $self->unroll($message,'cc','phrase');
-		$row->{'subject'} = $message->subject;
+
+		foreach my $p (qw(from to cc bcc)) {
+			foreach my $v ($self->unroll($message,'from','phrase')) {
+				push @{$row->{$p}},$v;
+				$self->add_counter($p,$v);
+			}
+		}
+		$row->{'subject'} = $self->decode_qp($message->subject);
 		$row->{'body'} = $self->plain_text_body($message);
 		$row->{'date'} = $message->date;