less output

[BackupPC.git] / lib / BackupPC / Search / KinoSearch.pm

diff --git a/lib/BackupPC/Search/KinoSearch.pm b/lib/BackupPC/Search/KinoSearch.pm
index 30cc073..d0e7713 100644 (file)
--- a/lib/BackupPC/Search/KinoSearch.pm
+++ b/lib/BackupPC/Search/KinoSearch.pm
@@ -11,6 +11,8 @@ use Data::Dump qw(dump);
 
 # my $tokenizer = KinoSearch::Analysis::Tokenizer->new( pattern => '\\w' );
 
+# numeric_padding values are used in range search, and muse be sortable
+
 sub new {
        my $class = shift @_;
        my %Conf = @_;
@@ -20,8 +22,13 @@ sub new {
        my $self = bless {
                index => $index_path,
                first_time_indexing => ! -d $index_path,
+               numeric_padding => [ qw(
+                       backup_date
+                       date
+               ) ],
+
        }, $class;
-       warn "# ",dump($self);
+       #warn "# ",dump($self);
        return $self;
 }
 
@@ -48,7 +55,6 @@ sub indexer {
 
        # numeric
        $schema->spec_field( name => $_, type => $string_type ) foreach ( qw/
-               backup_date
                fid
                shareid
                type
@@ -69,6 +75,7 @@ sub indexer {
        # sortable numeric
        $schema->spec_field( name => $_, type => $sort_type ) foreach (qw/
                backupnum
+               backup_date
                date
                size
        /);
@@ -84,7 +91,7 @@ sub indexer {
                create => 1,
        );
 
-       warn "# created indexer";
+       #warn "# created indexer";
 
        return $self->{_indexer} = $indexer;
 
@@ -112,6 +119,8 @@ sub exists {
        return $hits->total_hits;
 }
 
+sub _numeric_padding { sprintf "%010d", $_[0] } # pad up to 32bit number (timestamp)
+
 sub add_doc {
        my ($self,$row) = @_;
 
@@ -120,10 +129,14 @@ sub add_doc {
        $path =~ s/(.)/$1 /g; # XXX our tokenize
        $row->{_file_path_split} = $path;
 
-       warn "XXX ",dump($row) if $ENV{DEBUG};
-
        $self->{stats}->{add_doc}++;
 
+       foreach my $col ( @{ $self->{numeric_padding} } ) {
+               $row->{$col} = _numeric_padding $row->{$col};
+       }
+
+       warn "XXX ",dump($row) if $ENV{DEBUG};
+
        $self->indexer->add_doc( $row );
 
 }
@@ -131,22 +144,26 @@ sub add_doc {
 sub commit {
        my $self = shift;
        $self->indexer->commit;
-       warn "# commit index ", dump($self->{stats});
+       print STDERR "[commit]";
 }
 
 sub _field_lower_upper_term {
-       my ( $field, $l, $u ) = @_;
+       my ( $self, $field, $l, $u ) = @_;
+       my $numeric_padding = grep { /^$field$/ } @{ $self->{numeric_padding} };
        my $range;
        if ( $l ) {
-               $range->{lower_term} = $l;
+               $range->{lower_term} = $numeric_padding ? _numeric_padding $l : $l;
                $range->{include_lower} = 1;
        }
        if ( $u ) {
-               $range->{upper_term} = $u;
+               $range->{upper_term} = $numeric_padding ? _numeric_padding $u : $u;
                $range->{include_upper} = 1;
        }
-       $range->{field} = $field if $range;
-       warn "# $field $l - $u ",dump($range);
+       if ( $range ) {
+               $range->{field} = $field;
+
+               #warn "# $field $l - $u numeric_padding:$numeric_padding ",dump($range);
+       }
        return $range;
 }
 
@@ -180,10 +197,10 @@ sub search {
                push @and_query, KinoSearch::Search::TermQuery->new( field => 'shareid', term => $shareid );
        }
 
-       if ( my $range = _field_lower_upper_term( 'backup_date', $backup_from, $backup_to ) ) {
+       if ( my $range = $self->_field_lower_upper_term( 'backup_date', $backup_from, $backup_to ) ) {
                push @and_query, KinoSearch::Search::RangeQuery->new( %$range );
        }
-       if ( my $range = _field_lower_upper_term( 'date', $files_from, $files_to ) ) {
+       if ( my $range = $self->_field_lower_upper_term( 'date', $files_from, $files_to ) ) {
                push @and_query, KinoSearch::Search::RangeQuery->new( %$range );
        }