--- trunk/lib/WebPAC/Normalize.pm	2005/07/24 11:17:44	29
+++ trunk/lib/WebPAC/Normalize.pm	2005/12/05 17:47:51	217
@@ -11,11 +11,11 @@
 
 =head1 VERSION
 
-Version 0.01
+Version 0.03
 
 =cut
 
-our $VERSION = '0.01';
+our $VERSION = '0.03';
 
 =head1 SYNOPSIS
 
@@ -81,13 +81,15 @@
 	},
 	db => $db_obj,
 	lookup_regex => $lookup->regex,
+	lookup => $lookup_obj,
   );
 
 Parametar C<filter> defines user supplied snippets of perl code which can
 be use with C<filter{...}> notation.
 
 Recommended parametar C<lookup_regex> is used to enable parsing of lookups
-in structures.
+in structures. If you pass this parametar, you must also pass C<lookup>
+which is C<WebPAC::Lookup> object.
 
 =cut
 
@@ -96,6 +98,19 @@
         my $self = {@_};
         bless($self, $class);
 
+	my $r = $self->{'lookup_regex'} ? 1 : 0;
+	my $l = $self->{'lookup'} ? 1 : 0;
+
+	my $log = $self->_get_logger();
+
+	# those two must be in pair
+	if ( ($r & $l) != ($r || $l) ) {
+		my $log = $self->_get_logger();
+		$log->logdie("lookup_regex and lookup must be in pair");
+	}
+
+	$log->logdie("lookup must be WebPAC::Lookup object") if ($self->{'lookup'} && ! $self->{'lookup'}->isa('WebPAC::Lookup'));
+
 	$self ? return $self : return undef;
 }
 
@@ -107,13 +122,7 @@
 
 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
 
@@ -125,12 +134,18 @@
 	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 $mfn = $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 => $mfn );
+		$log->debug("load_ds( rec = ", sub { Dumper($rec) }, ") = ", sub { Dumper($ds) });
+		return $ds if ($ds);
 		$log->debug("cache miss, creating");
 	}
 
@@ -145,7 +160,7 @@
 		$self->{tags_by_order} = \@sorted_tags;
 	}
 
-	my @ds;
+	my $ds;
 
 	$log->debug("tags: ",sub { join(", ",@sorted_tags) });
 
@@ -156,7 +171,10 @@
 #print "field $field [",$self->{'tag'},"] = ",Dumper($self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}});
 
 		foreach my $tag (@{$self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}}}) {
-			my $format = $tag->{'value'} || $tag->{'content'};
+			my $format;
+
+			$log->logdie("expected tag HASH and got $tag") unless (ref($tag) eq 'HASH');
+			$format = $tag->{'value'} || $tag->{'content'};
 
 			$log->debug("format: $format");
 
@@ -177,22 +195,13 @@
 				@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'});
 
@@ -225,14 +234,14 @@
 
 			# 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) });
 		}
@@ -240,14 +249,15 @@
 	}
 
 	$self->{'db'}->save_ds(
-		ds => \@ds,
-		current_filename => $self->{'current_filename'},
-		headline => $self->{'headline'},
+		id => $mfn,
+		ds => $ds,
 	) if ($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;
 
 }
 
@@ -443,7 +453,11 @@
 		}
 		# do we have lookups?
 		if ($self->{'lookup'}) {
-			return $self->lookup($format);
+			if ($self->{'lookup'}->can('lookup')) {
+				return $self->{'lookup'}->lookup($format);
+			} else {
+				$log->warn("Have lookup object but can't invoke lookup method");
+			}
 		} else {
 			return $format;
 		}
@@ -516,7 +530,7 @@
 		if ($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
@@ -529,6 +543,8 @@
 			} else {
 				return $$rec->{$f}->[$i];
 			}
+		} else {
+			return '';
 		}
 	} else {
 		return '';
@@ -569,7 +585,7 @@
 	$log->debug("using format $name [$format] on $data to produce: $out");
 
 	if ($self->{'lookup_regex'} && $out =~ $self->{'lookup_regex'}) {
-		return $self->lookup($out);
+		return $self->{'lookup'}->lookup($out);
 	} else {
 		return $out;
 	}
@@ -652,4 +668,4 @@
 
 =cut
 
-1; # End of WebPAC::DB
+1; # End of WebPAC::Normalize