--- trunk/bin/BackupPC_updatedb	2005/08/26 17:18:27	74
+++ trunk/bin/BackupPC_updatedb	2005/08/28 18:02:58	92
@@ -40,26 +40,192 @@
 
 my $dsn = $Conf{SearchDSN} || die "Need SearchDSN in config.pl\n";
 my $user = $Conf{SearchUser} || '';
+my $index_path = $Conf{HyperEstraierIndex};
+$index_path = $TopDir . '/' . $index_path;
+$index_path =~ s#//#/#g;
+
 
 my $dbh = DBI->connect($dsn, $user, "", { RaiseError => 1, AutoCommit => 0 });
 
 my %opt;
 
-if ( !getopts("cdm:v:", \%opt ) ) {
+if ( !getopts("cdm:v:i", \%opt ) ) {
 	print STDERR <<EOF;
-usage: $0 [-c|-d] [-m num] [-v|-v level]
+usage: $0 [-c|-d] [-m num] [-v|-v level] [-i]
 
 Options:
 	-c	create database on first use
 	-d	delete database before import
 	-m num	import just num	increments for one host
 	-v num	set verbosity (debug) level (default $debug)
+	-i	update HyperEstraier full text index
 EOF
 	exit 1;
 }
 
-###################################create tables############################3
+if ($opt{v}) {
+	print "Debug level at $opt{v}\n";
+	$debug = $opt{v};
+}
+
+#---- subs ----
+
+sub fmt_time {
+	my $t = shift || return;
+	my $out = "";
+	my ($ss,$mm,$hh) = gmtime($t);
+	$out .= "${hh}h" if ($hh);
+	$out .= sprintf("%02d:%02d", $mm,$ss);
+	return $out;
+}
+
+sub curr_time {
+	return strftime($t_fmt,localtime());
+}
+
+my $hest_db;
+
+sub hest_update {
+
+	my ($host_id, $share_id, $num) = @_;
+
+	print curr_time," updating HyperEstraier: select files";
+
+	my $t = time();
+
+	my $where = '';
+	if ($host_id && $share_id && $num) {
+		$where = qq{
+		WHERE
+			hosts.id = ? AND
+			shares.id = ? AND
+			files.backupnum = ?
+		};
+	}
+
+	my $sth = $dbh->prepare(qq{
+		SELECT
+			files.id			AS fid,
+			hosts.name			AS hname,
+			shares.name			AS sname,
+			-- shares.share			AS sharename,
+			files.backupnum			AS backupnum,
+			-- files.name			AS filename,
+			files.path			AS filepath,
+			files.date			AS date,
+			files.type			AS type,
+			files.size			AS size,
+			files.shareid			AS shareid,
+			backups.date			AS backup_date
+		FROM files 
+			INNER JOIN shares	ON files.shareID=shares.ID
+			INNER JOIN hosts	ON hosts.ID = shares.hostID
+			INNER JOIN backups	ON backups.num = files.backupNum and backups.hostID = hosts.ID AND backups.shareID = shares.ID
+		$where
+	});
+
+	$sth->execute(@_);
+	my $results = $sth->rows;
+
+	if ($results == 0) {
+		print " - no files, skipping\n";
+		return;
+	}
+
+	my $dot = int($results / 15) || 1;
+
+	print " $results ($dot/#)";
+
+	sub fmt_date {
+		my $t = shift || return;
+		my $iso = BackupPC::Lib::timeStamp($t);
+		$iso =~ s/\s/T/;
+		return $iso;
+	}
+
+	my $max = int($results / $dot);
+
+	print ", opening index $index_path...";
+	use HyperEstraier;
+	my $db = HyperEstraier::Database->new();
+
+#	unless ($hest_db) {
+#		print " open reader";
+#		$hest_db = HyperEstraier::Database->new();
+#
+#	}
+
+
+	$db->open($index_path, $HyperEstraier::Database::DBWRITER | $HyperEstraier::Database::DBCREAT);
+
+	my $added = 0;
+
+	while (my $row = $sth->fetchrow_hashref()) {
+
+		my $fid = $row->{'fid'} || die "no fid?";
+		my $uri = 'file:///' . $fid;
+
+		my $id = $db->uri_to_id($uri);
+		next unless ($id == -1);
+
+		# create a document object 
+		my $doc = HyperEstraier::Document->new;
+
+		# add attributes to the document object 
+		$doc->add_attr('@uri', $uri);
+
+		foreach my $c (@{ $sth->{NAME} }) {
+			$doc->add_attr($c, $row->{$c}) if ($row->{$c});
+		}
+
+		#$doc->add_attr('@cdate', fmt_date($row->{'date'}));
+
+		# add the body text to the document object 
+		my $path = $row->{'filepath'};
+		$doc->add_text($path);
+		$path =~ s/(.)/$1 /g;
+		$doc->add_hidden_text($path);
+
+		print STDERR $doc->dump_draft,"\n" if ($debug > 1);
+
+		# register the document object to the database
+		$db->put_doc($doc, $HyperEstraier::Database::PDCLEAN);
+
+		$added++;
+		if ($added % $dot == 0) {
+			print "$max ";
+			$max--;
+		}
+
+	}
+
+	print "sync $added new files";
+	$db->sync();
+	print ", close";
+	$db->close();
+
+	my $dur = (time() - $t) || 1;
+	printf(" [%.2f/s new %.2f/s dur: %s]\n",
+		( $results / $dur ),
+		( $added / $dur ),
+		fmt_time($dur)
+	);
+}
+
+#---- /subs ----
 
+
+## update index ##
+if ($opt{i} || ($index_path && ! -e $index_path)) {
+	# update all
+	print "force update of HyperEstraier index ";
+	print "importing existing data" unless (-e $index_path);
+	print "by -i flag" if ($opt{i});
+	print "\n";
+	hest_update();
+}
+
+## create tables ##
 if ($opt{c}) {
 	sub do_index {
 		my $index = shift || return;
@@ -149,6 +315,7 @@
 
 }
 
+## delete data before inseting ##
 if ($opt{d}) {
 	print "deleting ";
 	foreach my $table (qw(files dvds backups shares hosts)) {
@@ -160,12 +327,7 @@
 	$dbh->commit;
 }
 
-if ($opt{v}) {
-	print "Debug level at $opt{v}\n";
-	$debug = $opt{v};
-}
-
-#################################INSERT VALUES#############################
+## insert new values ##
 
 # get hosts
 $hosts = $bpc->HostInfoRead();
@@ -199,15 +361,6 @@
 	VALUES (?,?,?,?,?,?,?)
 });
 
-sub fmt_time {
-	my $t = shift || return;
-	my $out = "";
-	my ($ss,$mm,$hh) = gmtime($t);
-	$out .= "${hh}h" if ($hh);
-	$out .= sprintf("%02d:%02d", $mm,$ss);
-	return $out;
-}
-
 foreach my $host_key (keys %{$hosts}) {
 
 	my $hostname = $hosts->{$host_key}->{'host'} || die "can't find host for $host_key";
@@ -263,7 +416,7 @@
 			next if ($count > 0);
 
 			# dump some log
-			print strftime($t_fmt,localtime())," ", $share;
+			print curr_time," ", $share;
 
 			my ($f, $nf, $d, $nd, $size) = recurseDir($bpc, $hostname, $files, $backupNum, $share, "", $shareID);
 
@@ -286,6 +439,8 @@
 				( ($f+$d) / $dur ),
 				fmt_time($dur)
 			);
+
+			hest_update($hostID, $shareID, $backupNum);
 		}
 
 	}