--- trunk/lib/WebPAC/Normalize.pm	2005/11/12 21:21:50	38
+++ trunk/lib/WebPAC/Normalize.pm	2005/12/05 17:48:08	219
@@ -11,11 +11,11 @@
 
 =head1 VERSION
 
-Version 0.01
+Version 0.04
 
 =cut
 
-our $VERSION = '0.01';
+our $VERSION = '0.04';
 
 =head1 SYNOPSIS
 
@@ -82,11 +82,15 @@
 	db => $db_obj,
 	lookup_regex => $lookup->regex,
 	lookup => $lookup_obj,
+	prefix => 'foobar',
   );
 
 Parametar C<filter> defines user supplied snippets of perl code which can
 be use with C<filter{...}> notation.
 
+C<prefix> is used to form filename for database record (to support multiple
+source files which are joined in one database).
+
 Recommended parametar C<lookup_regex> is used to enable parsing of lookups
 in structures. If you pass this parametar, you must also pass C<lookup>
 which is C<WebPAC::Lookup> object.
@@ -111,6 +115,8 @@
 
 	$log->logdie("lookup must be WebPAC::Lookup object") if ($self->{'lookup'} && ! $self->{'lookup'}->isa('WebPAC::Lookup'));
 
+	$log->warn("no prefix defined. please check that!") unless ($self->{'prefix'});
+
 	$self ? return $self : return undef;
 }
 
@@ -122,13 +128,7 @@
 
 This structures are used to produce output.
 
- my @ds = $webpac->data_structure($rec);
-
-B<Note: historical oddity follows>
-
-This method will also set C<< $webpac->{'currnet_filename'} >> if there is
-C<< <filename> >> tag and C<< $webpac->{'headline'} >> if there is
-C<< <headline> >> tag.
+ my $ds = $webpac->data_structure($rec);
 
 =cut
 
@@ -140,12 +140,18 @@
 	my $rec = shift;
 	$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o);
 
+	$log->debug("data_structure rec = ", sub { Dumper($rec) });
+
+	$log->logdie("need unique ID (mfn) in field 000 of record ", sub { Dumper($rec) } ) unless (defined($rec->{'000'}));
+
+	my $id = $rec->{'000'}->[0] || $log->logdie("field 000 isn't array!");
+
 	my $cache_file;
 
 	if ($self->{'db'}) {
-		my @ds = $self->{'db'}->load_ds($rec);
-		$log->debug("load_ds( rec = ", sub { Dumper($rec) }, ") = ", sub { Dumper(@ds) });
-		return @ds if ($#ds > 0);
+		my $ds = $self->{'db'}->load_ds( id => $id, prefix => $self->{prefix} );
+		$log->debug("load_ds( rec = ", sub { Dumper($rec) }, ") = ", sub { Dumper($ds) });
+		return $ds if ($ds);
 		$log->debug("cache miss, creating");
 	}
 
@@ -160,7 +166,7 @@
 		$self->{tags_by_order} = \@sorted_tags;
 	}
 
-	my @ds;
+	my $ds;
 
 	$log->debug("tags: ",sub { join(", ",@sorted_tags) });
 
@@ -195,22 +201,13 @@
 				@v = map { $self->apply_format($tag->{'format_name'},$tag->{'format_delimiter'},$_) } @v;
 			}
 
-			if ($field eq 'filename') {
-				$self->{'current_filename'} = join('',@v);
-				$log->debug("filename: ",$self->{'current_filename'});
-			} elsif ($field eq 'headline') {
-				$self->{'headline'} .= join('',@v);
-				$log->debug("headline: ",$self->{'headline'});
-				next; # don't return headline in data_structure!
-			}
-
 			# delimiter will join repeatable fields
 			if ($tag->{'delimiter'}) {
 				@v = ( join($tag->{'delimiter'}, @v) );
 			}
 
 			# default types 
-			my @types = qw(display swish);
+			my @types = qw(display search);
 			# override by type attribute
 			@types = ( $tag->{'type'} ) if ($tag->{'type'});
 
@@ -243,14 +240,14 @@
 
 			# TODO: name_sigular, name_plural
 			my $name = $self->{'import_xml'}->{'indexer'}->{$field}->{'name'};
-			$row->{'name'} = $name ? $self->_x($name) : $field;
+			my $row_name = $name ? $self->_x($name) : $field;
 
 			# post-sort all values in field
 			if ($self->{'import_xml'}->{'indexer'}->{$field}->{'sort'}) {
 				$log->warn("sort at field tag not implemented");
 			}
 
-			push @ds, $row;
+			$ds->{$row_name} = $row;
 
 			$log->debug("row $field: ",sub { Dumper($row) });
 		}
@@ -258,14 +255,16 @@
 	}
 
 	$self->{'db'}->save_ds(
-		ds => \@ds,
-		current_filename => $self->{'current_filename'},
-		headline => $self->{'headline'},
+		id => $id,
+		ds => $ds,
+		prefix => $self->{prefix},
 	) if ($self->{'db'});
 
-	$log->debug("ds: ", sub { Dumper(@ds) });
+	$log->debug("ds: ", sub { Dumper($ds) });
 
-	return @ds;
+	$log->logconfess("data structure returned is not array any more!") if wantarray;
+
+	return $ds;
 
 }
 
@@ -538,7 +537,7 @@
 		if ($sf && $$rec->{$f}->[$i]->{$sf}) {
 			$$found++ if (defined($$found));
 			return $$rec->{$f}->[$i]->{$sf};
-		} elsif ($$rec->{$f}->[$i]) {
+		} elsif (! $sf && $$rec->{$f}->[$i]) {
 			$$found++ if (defined($$found));
 			# it still might have subfield, just
 			# not specified, so we'll dump all
@@ -551,6 +550,8 @@
 			} else {
 				return $$rec->{$f}->[$i];
 			}
+		} else {
+			return '';
 		}
 	} else {
 		return '';
@@ -674,4 +675,4 @@
 
 =cut
 
-1; # End of WebPAC::DB
+1; # End of WebPAC::Normalize