my $index_path = $Conf{KinoPath} || die "no KinoPath";
- my $self = bless { index => $index_path }, $class;
+ my $self = bless {
+ index => $index_path,
+ first_time_indexing => ! -d $index_path,
+ numeric_padding => [ qw(
+ backup_date
+ date
+ size
+ ) ],
+
+ }, $class;
warn "# ",dump($self);
return $self;
}
);
my $blob_type = KinoSearch::Plan::BlobType->new( stored => 1 );
my $string_type = KinoSearch::Plan::StringType->new; # non-tokenized
- my $num_type = KinoSearch::Plan::Int64Type->new;
+ my $num_type = KinoSearch::Plan::Int64Type->new( sortable => 1 );
+ my $sort_type = KinoSearch::Plan::StringType->new( sortable => 1 ); # non-tokenized
+ # numeric
$schema->spec_field( name => $_, type => $string_type ) foreach ( qw/
backup_date
- backupnum
- date
fid
shareid
- size
type
/ );
+ # non-tokenized strings
$schema->spec_field( name => $_, type => $string_type ) foreach ( qw/
- _uri _file_path_split filepath hname sname
+ _uri
+ hname
+ /);
+
+ # sortable
+ $schema->spec_field( name => $_, type => $sort_type ) foreach (qw/
+ sname
+ filepath
+ /);
+
+ # sortable numeric
+ $schema->spec_field( name => $_, type => $sort_type ) foreach (qw/
+ backupnum
+ date
+ size
/);
+ # tokenized magic columns for infix search
+ $schema->spec_field( name => '_file_path_split', type => $ft_type );
+
# $schema->spec_field( name => '_doc', type => $blob_type );
my $indexer = KinoSearch::Index::Indexer->new(
create => 1,
);
- $indexer->commit; # make sure that index exists
-
warn "# created indexer";
return $self->{_indexer} = $indexer;
sub exists {
my ($self,$row) = @_;
+ return 0 if $self->{first_time_indexing};
+
my $uri = $row->{hname} . ':' . $row->{sname} . '#' . $row->{backupnum} . ' ' . $row->{filepath};
my $hits = $self->searcher->hits( query => "_uri:$uri" );
+
+
+ $self->{stat}->{exists}->{ $hits->total_hits }++;
+
return $hits->total_hits;
}
+sub _numeric_padding { sprintf "%011d", $_[0] }
+
sub add_doc {
my ($self,$row) = @_;
$path =~ s/(.)/$1 /g; # XXX our tokenize
$row->{_file_path_split} = $path;
+ $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 );
sub commit {
my $self = shift;
$self->indexer->commit;
- warn "# commit index";
+ warn "# commit index ", dump($self->{stats});
+}
+
+sub _field_lower_upper_term {
+ my ( $self, $field, $l, $u ) = @_;
+ my $numeric_padding = grep { /^$field$/ } @{ $self->{numeric_padding} };
+ my $range;
+ if ( $l ) {
+ $range->{lower_term} = $numeric_padding ? _numeric_padding $l : $l;
+ $range->{include_lower} = 1;
+ }
+ if ( $u ) {
+ $range->{upper_term} = $numeric_padding ? _numeric_padding $u : $u;
+ $range->{include_upper} = 1;
+ }
+ if ( $range ) {
+ $range->{field} = $field;
+
+ warn "# $field $l - $u numeric_padding:$numeric_padding ",dump($range);
+ }
+ return $range;
}
sub search {
my $sort_spec = KinoSearch::Search::SortSpec->new( rules => $rules );
-# $q =~ s/(.)/$1 /g;
+ my $split = $q;
+ $split =~ s/(.)/$1 /g; # _file_path_split
+ my $split_query = KinoSearch::Search::TermQuery->new( field => '_file_path_split', term => $split );
+#warn "XXX ",dump($split_query);
+
+
+ my $query_parser = KinoSearch::Search::QueryParser->new(
+ schema => $self->searcher->get_schema,
+ fields => ['_file_path_split'],
+ );
+ my $query = $query_parser->parse( '"' . $split . '"' );
+
+ my @and_query;
+
+ if ( $shareid ) {
+ push @and_query, KinoSearch::Search::TermQuery->new( field => 'shareid', term => $shareid );
+ }
+
+ 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 = $self->_field_lower_upper_term( 'date', $files_from, $files_to ) ) {
+ push @and_query, KinoSearch::Search::RangeQuery->new( %$range );
+ }
+
+ if ( @and_query ) {
+ push @and_query, $query;
+ $query = KinoSearch::Search::ANDQuery->new( children => [ @and_query ] );
+ }
+
my $hits = $self->searcher->hits(
- query => $q,
+ query => m/:/ ? $q : $query,
+ offset => $offset,
+ num_wanted => $on_page,
sort_spec => $sort_spec,
);
my $results;
while ( my $hit = $hits->next ) {
-warn "XXX ",dump($hit);
+ warn "## hit = ",dump($hit) if $ENV{DEBUG};
push @$results, $hit;
}