--- trunk/lib/WebPAC/Normalize/MARC.pm	2007/11/10 11:13:58	1023
+++ trunk/lib/WebPAC/Normalize/MARC.pm	2007/11/12 10:16:16	1035
@@ -12,6 +12,8 @@
 
 use WebPAC::Normalize;
 
+my $debug = 0;
+
 =head1 NAME
 
 WebPAC::Normalize::MARC - create MARC/ISO2709 records
@@ -42,14 +44,16 @@
 		],
 	);
 
+Returns number of records produced.
+
 =cut
 
 sub marc_template {
 	my $args = {@_};
-	warn "## marc_template(",dump($args),")";
+	warn "## marc_template(",dump($args),")",$/ if $debug;
 
 	foreach ( qw/subfields_rename isis_template marc_template/ ) {
-#		warn "ref($_) = ",ref($args->{$_});
+#		warn "ref($_) = ",ref($args->{$_})  if $debug;
 		die "$_ not ARRAY" if defined($args->{$_}) && ref($args->{$_}) ne 'ARRAY';
 	}
 
@@ -61,12 +65,13 @@
 	die "record field ", $args->{from}, " isn't array ",dump( $rec ) unless (ref($r) eq 'ARRAY');
 
 	my @subfields_rename = @{ $args->{subfields_rename} };
-#	warn "### subfields_rename [$#subfields_rename] = ",dump( @subfields_rename );
+#	warn "### subfields_rename [$#subfields_rename] = ",dump( @subfields_rename )  if $debug;
 
 	confess "need mapping in pairs for subfields_rename"
 		if $#subfields_rename % 2 != 1;
 	
-	my ( $subfields_rename, $from_subfields, $to_subfields );
+	my ( $subfields_rename, $from_subfields );
+	our $to_subfields = {};
 	while ( my ( $from, $to ) = splice(@subfields_rename, 0, 2) ) {
 		my ( $f, $t ) = (
 			$from_subfields->{ $from }++,
@@ -74,41 +79,44 @@
 		);
 		$subfields_rename->{ $from }->[ $f ] = [ $to => $t ];
 	}
-	warn "### subfields_rename = ",dump( $subfields_rename ),$/;
-	warn "### from_subfields = ", dump( $from_subfields ),$/;
-	warn "### to_subfields = ", dump( $to_subfields ),$/;
+	warn "### subfields_rename = ",dump( $subfields_rename ),$/ if $debug;
+	warn "### from_subfields = ", dump( $from_subfields ),$/ if $debug;
+	warn "### to_subfields = ", dump( $to_subfields ),$/ if $debug;
 
-	my $fields_re = join('|', keys %$to_subfields );
+	our $_template;
 
-	my $fill_in;
-	my @marc_out;
+	$_template->{isis}->{fields_re} = join('|', keys %$from_subfields );
+	$_template->{marc}->{fields_re} = join('|', keys %$to_subfields );
 
-	my $marc_template;
-	_parse_template( $args->{marc_template}, $marc_template );
+	my @marc_out;
 
 	sub _parse_template {
-		my ( $templates, $marc_template_order ) = @_;
+		my ( $name, $templates ) = @_;
+
+		my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} );
 
 		foreach my $template ( @{ $templates } ) {
 			our $count = {};
-			our @marc_order = ();
+			our @order = ();
 			sub my_count {
 				my $sf = shift;
 				my $nr = $count->{$sf}++;
-				push @marc_order, [ $sf, $nr ];
+				push @order, [ $sf, $nr ];
 				return $sf . $nr;
 			}
 			my $pos_template = $template;
 			$pos_template =~ s/($fields_re)/my_count($1)/ge;
 			my $count_key = dump( $count );
-			warn "### template: |$template| -> |$pos_template| count = $count_key marc_order = ",dump( @marc_order ),$/;
-			$marc_template->{pos}->{ $count_key } = $pos_template;
-			$marc_template->{order}->{ $pos_template } = [ @marc_order ];
+			warn "### template: |$template| -> |$pos_template| count = $count_key order = ",dump( @order ),$/ if $debug;
+			$_template->{$name}->{pos}->{ $count_key } = $pos_template;
+			$_template->{$name}->{order}->{ $pos_template } = [ @order ];
 		}
-		warn "### from ",dump( $templates ), " created ", dump( $marc_template );
+		warn "### from ",dump( $templates ), " using $fields_re created ", dump( $_template ),$/ if $debug;
 	}
 
-	warn "### marc_template = ",dump( $marc_template );
+	_parse_template( 'marc', $args->{marc_template} );
+	_parse_template( 'isis', $args->{isis_template} );
+	warn "### _template = ",dump( $_template ),$/ if $debug;
 
 	my $m;
 
@@ -118,80 +126,117 @@
 		my $i2 = $r->{i2} || ' ';
 		$m = [ $args->{to}, $i1, $i2 ];
 
-		warn "### r = ",dump( $r );
+		warn "### r = ",dump( $r ),$/ if $debug;
 
-		my ( $new_r, $from_count, $to_count );
-		foreach my $sf ( keys %{$r} ) {
+		my ( $from_mapping, $to_mapping, $from_count, $to_count );
+		foreach my $from_sf ( keys %{$r} ) {
 			# skip everything which isn't one char subfield (e.g. 'subfields')
-			next unless $sf =~ m/^\w$/;
-			my $nr = $from_count->{$sf}++;
-			my $rename_to = $subfields_rename->{ $sf } ||
-				die "can't find subfield rename for $sf/$nr in ", dump( $subfields_rename );
-			warn "### rename $sf/$nr to ", dump( $rename_to->[$nr] ), $/;
-			my ( $to_sf, $to_nr ) = @{ $rename_to->[$nr] };
-			$new_r->{ $to_sf }->[ $to_nr ] = [ $sf => $nr ];
-
-			$to_count->{ $to_sf }++;
-		}
-
-		warn "### new_r = ",dump( $new_r );
-
-		my $from_count_key = dump( $to_count );
-
-		warn "### from_count = ",dump( $from_count ), $/;
-		warn "### to_count   = ",dump( $to_count ), $/;
-
-		my $template = $marc_template->{pos}->{ $from_count_key } ||
-			die "I don't have template for:\n$from_count_key\n## available templates\n", dump( $marc_template->{pos} );
+			next unless $from_sf =~ m/^\w$/;
+			my $from_nr = $from_count->{$from_sf}++;
+			my $rename_to = $subfields_rename->{ $from_sf } ||
+				die "can't find subfield rename for $from_sf/$from_nr in ", dump( $subfields_rename );
+			my ( $to_sf, $to_nr ) = @{ $rename_to->[$from_nr] };
+			$to_mapping->{ $to_sf }->[ $to_nr ] = [ $from_sf => $from_nr ];
 
-		warn "### selected template: |$template|\n";
+			my $to_nr2 = $to_count->{ $to_sf }++;
+			$from_mapping->{ $from_sf }->[ $from_nr ] = [ $to_sf => $to_nr2 ];
 
-		$fill_in = {};
+			warn "### from $from_sf/$from_nr -> $to_sf/$to_nr\tto $from_sf/$from_nr -> $to_sf/$to_nr2\n" if $debug;
+		}
 
-		my @templates = split(/\|/, $template );
-		@templates = ( $template );
+		warn "### from_mapping = ",dump( $from_mapping ), "\n### to_mapping = ",dump( $to_mapping ),$/ if $debug;
 
-		foreach my $sf ( @templates ) {
-			sub fill_in {
-				my ( $r, $sf, $nr ) = @_;
-				my ( $from_sf, $from_nr ) = @{ $new_r->{ $sf }->[ $nr ] };
-				my $v = $r->{ $from_sf }; # || die "no $from_sf/$from_nr";
-				warn "#### fill_in( $sf, $nr ) = $from_sf/$from_nr >>>> ",dump( $v ), $/;
-				if ( ref( $v ) eq 'ARRAY' ) {
-					$fill_in->{$sf}->[$nr] = $v->[$from_nr];
-					return $v->[$from_nr];
-				} elsif ( $from_nr == 0 ) {
+		my $count_key = {
+			from => dump( $from_count ),
+			to   => dump( $to_count),
+		};
+
+		warn "### count_key = ",dump( $count_key ),$/ if $debug;
+
+		my $processed_templates = 0;
+
+		# this defines order of traversal
+		foreach ( qw/isis:from marc:to/ ) {
+			my ($name,$count_name) = split(/:/);
+
+			my $ckey = $count_key->{$count_name} || die "can't find count_key $count_name in ",dump( $count_key );
+
+			my $template = $_template->{$name}->{pos}->{ $ckey } || next;
+			$processed_templates++;
+
+			warn "### traverse $name $count_name selected template: |$template|\n" if $debug;
+
+			our $fill_in = {};
+
+			my @templates = split(/\|/, $template );
+			@templates = ( $template ) unless @templates;
+
+			warn "### templates = ",dump( @templates ),$/ if $debug;
+
+			foreach my $sf ( @templates ) {
+				sub fill_in {
+					my ( $name, $r, $pre, $sf, $nr, $post ) = @_;
+					warn "#### fill_in( $name, r, '$pre', $sf, $nr, '$post' )\n" if $debug;
+					my ( $from_sf, $from_nr );
+					if ( $name eq 'marc' ) {
+						die "no $sf/$nr in to_mapping: ",dump( $to_mapping ), " form record ",dump( $r ) unless defined $to_mapping->{$sf}->[$nr];
+						( $from_sf, $from_nr ) = @{ $to_mapping->{$sf}->[$nr] };
+					} else {
+						( $from_sf, $from_nr ) = ( $sf, $nr );
+					}
+					my $v = $r->{ $from_sf }; # || die "no $from_sf/$from_nr";
+					if ( ref( $v ) eq 'ARRAY' ) {
+						$v = $pre . $v->[$from_nr] . $post;
+					} elsif ( $from_nr == 0 ) {
+						$v = $pre . $v . $post;
+					} else {
+						die "requested subfield $from_sf/$from_nr but it's ",dump( $v );
+					}
+					warn "#### fill_in( $sf, $nr ) = $from_sf/$from_nr >>>> ",dump( $v ),$/ if $debug;
 					$fill_in->{$sf}->[$nr] = $v;
 					return $v;
-				} else {
-					die "requested subfield $from_sf/$from_nr but it's ",dump( $v );
 				}
+				my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} );
+				warn "#### $sf <<<< $fields_re\n" if $debug;
+				$sf =~ s/^(.*?)($fields_re)(\d+)(.*?)$/fill_in($name,$r,$1,$2,$3,$4)/ge;
+				warn "#### >>>> $sf with fill_in = ",dump( $fill_in ),$/ if $debug;
 			}
-			warn "#### $sf <<<< $fields_re\n";
-			$sf =~ s/($fields_re)(\d+)/fill_in($r,$1,$2)/ge;
-			warn "#### >>>> $sf with fill_in = ",dump( $fill_in ),$/;
-		}
 
-		warn "## template: |$template|\n## marc_template = ",dump( $marc_template );
+			warn "## template: |$template|\n## _template->$name = ",dump( $_template->{$name} ),$/ if $debug;
 
-		foreach my $sf ( @{ $marc_template->{order}->{$template} } ) {
-			my ( $sf, $nr ) = @$sf;
-			my $v = $fill_in->{$sf}->[$nr] || die "can't find fill_in $sf/$nr";
-			warn "++ $sf/$nr |$v|\n";
-			push @$m, ( $sf, $v );
-		}
+			foreach my $sf ( @{ $_template->{$name}->{order}->{$template} } ) {
+				my ( $sf, $nr ) = @$sf;
+				my $v = $fill_in->{$sf}->[$nr] || die "can't find fill_in $sf/$nr";
+				if ( $name eq 'isis') {
+					( $sf, $nr ) = @{ $from_mapping->{$sf}->[$nr] };
+				}
+				warn "++ $sf/$nr |$v|\n" if $debug;
+				push @$m, ( $sf, $v );
+			}
+
+			warn "#### >>>> created MARC record: ", dump( $m ),$/ if $debug;
 
-		warn "#### >>>> created marc: ", dump( $m );
+			push @marc_out, $m;
 
-		push @marc_out, $m;
+			last;
+		}
+	
+		die "I don't have template for fields ",dump( $count_key ), "\n## available templates\n", dump( $_template ) unless $processed_templates;
+		warn ">>> $processed_templates templates applied to data\n",$/ if $debug;
 	}
 
-	warn "### marc_template produced: ",dump( @marc_out );
+
+	my $recs = 0;
 
 	foreach my $marc ( @marc_out ) {
-		warn "+++ ",dump( $marc );
+		warn "+++ ",dump( $marc ),$/ if $debug;
 		WebPAC::Normalize::_marc_push( $marc );
+		$recs++;
 	}
+
+	warn "### marc_template produced $recs MARC records: ",dump( @marc_out ),$/ if $debug;
+
+	return $recs;
 }
 
 1;