--- trunk/lib/WebPAC/Normalize.pm	2007/09/06 19:12:15	889
+++ trunk/lib/WebPAC/Normalize.pm	2007/11/10 00:05:36	1019
@@ -1,19 +1,22 @@
 package WebPAC::Normalize;
 use Exporter 'import';
-@EXPORT = qw/
-	_set_rec _set_lookup
+our @EXPORT = qw/
+	_set_ds _set_lookup
 	_set_load_row
 	_get_ds _clean_ds
 	_debug
 	_pack_subfields_hash
 
-	tag search display
+	search_display search display sorted
+
 	marc marc_indicators marc_repeatable_subfield
 	marc_compose marc_leader marc_fixed
 	marc_duplicate marc_remove marc_count
 	marc_original_order
+	marc_template
 
 	rec1 rec2 rec
+	frec frec_eq frec_ne
 	regex prefix suffix surround
 	first lookup join_with
 	save_into_lookup
@@ -22,6 +25,7 @@
 
 	get set
 	count
+
 /;
 
 use warnings;
@@ -35,18 +39,16 @@
 # debugging warn(s)
 my $debug = 0;
 
+use WebPAC::Normalize::ISBN;
+push @EXPORT, ( 'isbn_10', 'isbn_13' );
 
 =head1 NAME
 
 WebPAC::Normalize - describe normalisaton rules using sets
 
-=head1 VERSION
-
-Version 0.29
-
 =cut
 
-our $VERSION = '0.29';
+our $VERSION = '0.34';
 
 =head1 SYNOPSIS
 
@@ -59,7 +61,7 @@
 C<perl -c normalize.pl>.
 
 Normalisation can generate multiple output normalized data. For now, supported output
-types (on the left side of definition) are: C<tag>, C<display>, C<search> and
+types (on the left side of definition) are: C<search_display>, C<display>, C<search> and
 C<marc>.
 
 =head1 FUNCTIONS
@@ -78,7 +80,7 @@
 	marc_encoding => 'utf-8',
 	config => $config,
 	load_row_coderef => sub {
-		my ($database,$input,$mfn) = shift;
+		my ($database,$input,$mfn) = @_;
 		$store->load_row( database => $database, input => $input, id => $mfn );
 	},
   );
@@ -104,30 +106,31 @@
 	die "need row argument" unless ($arg->{row});
 	die "need normalisation argument" unless ($arg->{rules});
 
-	no strict 'subs';
 	_set_lookup( $arg->{lookup} ) if defined($arg->{lookup});
-	_set_rec( $arg->{row} );
+	_set_ds( $arg->{row} );
 	_set_config( $arg->{config} ) if defined($arg->{config});
 	_clean_ds( %{ $arg } );
 	$load_row_coderef = $arg->{load_row_coderef};
 
-	eval "$arg->{rules}";
+	no strict 'subs';
+	no warnings 'redefine';
+	eval "$arg->{rules};";
 	die "error evaling $arg->{rules}: $@\n" if ($@);
 
 	return _get_ds();
 }
 
-=head2 _set_rec
+=head2 _set_ds
 
 Set current record hash
 
-  _set_rec( $rec );
+  _set_ds( $rec );
 
 =cut
 
 my $rec;
 
-sub _set_rec {
+sub _set_ds {
 	$rec = shift or die "no record hash";
 }
 
@@ -171,6 +174,7 @@
 my ($marc_record_offset, $marc_fetch_offset) = (0, 0);
 
 sub _get_ds {
+#warn "## out = ",dump($out);
 	return $out;
 }
 
@@ -442,40 +446,51 @@
 
 Those functions generally have to first in your normalization file.
 
-=head2 tag
+=head2 search_display
 
-Define new tag for I<search> and I<display>.
+Define output for L<search> and L<display> at the same time
 
-  tag('Title', rec('200','a') );
+  search_display('Title', rec('200','a') );
 
 
 =cut
 
-sub tag {
-	my $name = shift or die "tag needs name as first argument";
+sub search_display {
+	my $name = shift or die "search_display needs name as first argument";
 	my @o = grep { defined($_) && $_ ne '' } @_;
 	return unless (@o);
-	$out->{$name}->{tag} = $name;
 	$out->{$name}->{search} = \@o;
 	$out->{$name}->{display} = \@o;
 }
 
+=head2 tag
+
+Old name for L<search_display>, but supported
+
+=cut
+
+sub tag {
+	search_display( @_ );
+}
+
 =head2 display
 
-Define tag just for I<display>
+Define output just for I<display>
 
   @v = display('Title', rec('200','a') );
 
 =cut
 
-sub display {
-	my $name = shift or die "display needs name as first argument";
+sub _field {
+	my $type = shift or confess "need type -- BUG?";
+	my $name = shift or confess "needs name as first argument";
 	my @o = grep { defined($_) && $_ ne '' } @_;
 	return unless (@o);
-	$out->{$name}->{tag} = $name;
-	$out->{$name}->{display} = \@o;
+	$out->{$name}->{$type} = \@o;
 }
 
+sub display { _field( 'display', @_ ) }
+
 =head2 search
 
 Prepare values just for I<search>
@@ -484,13 +499,18 @@
 
 =cut
 
-sub search {
-	my $name = shift or die "search needs name as first argument";
-	my @o = grep { defined($_) && $_ ne '' } @_;
-	return unless (@o);
-	$out->{$name}->{tag} = $name;
-	$out->{$name}->{search} = \@o;
-}
+sub search { _field( 'search', @_ ) }
+
+=head2 sorted
+
+Insert into lists which will be automatically sorted
+
+ sorted('Title', rec('200','a') );
+
+=cut
+
+sub sorted { _field( 'sorted', @_ ) }
+
 
 =head2 marc_leader
 
@@ -795,7 +815,7 @@
 	return unless defined($rec->{$from});
 
 	my $r = $rec->{$from};
-	die "record field $from isn't array\n" unless (ref($r) eq 'ARRAY');
+	die "record field $from isn't array ",dump( $rec ) unless (ref($r) eq 'ARRAY');
 
 	my ($i1,$i2) = defined($marc_indicators->{$to}) ? @{ $marc_indicators->{$to} } : (' ',' ');
 	warn "## marc_original_order($to,$from) source = ", dump( $r ),$/ if ($debug > 1);
@@ -840,6 +860,151 @@
 	warn "## marc_record = ", dump( $marc_record ),$/ if ($debug > 1);
 }
 
+=head2 marc_template
+
+=cut
+
+sub marc_template {
+	my $args = {@_};
+	warn "## marc_template(",dump($args),")";
+
+	foreach ( qw/subfields_rename marc_template/ ) {
+#		warn "ref($_) = ",ref($args->{$_});
+		die "$_ not ARRAY" if ref($args->{$_}) ne 'ARRAY';
+	}
+
+	my $r = $rec->{ $args->{from} } || return;
+	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 );
+
+	confess "need mapping in pairs for subfields_rename"
+		if $#subfields_rename % 2 != 1;
+	
+	my ( $subfields_rename, $from_subfields, $to_subfields );
+	while ( my ( $from, $to ) = splice(@subfields_rename, 0, 2) ) {
+		my ( $f, $t ) = (
+			$from_subfields->{ $from }++,
+			$to_subfields->{ $to }++
+		);
+		$subfields_rename->{ $from }->[ $f ] = [ $to => $t ];
+	}
+	warn "### subfields_rename = ",dump( $subfields_rename ),$/;
+	warn "### from_subfields = ", dump( $from_subfields ),$/;
+	warn "### to_subfields = ", dump( $to_subfields ),$/;
+
+	my $fields_re = join('|', keys %$to_subfields );
+
+	my $pos_templates;
+	my $count;
+	my @marc_order;
+	my $marc_template_order;
+	my $fill_in;
+	my @marc_out;
+
+	foreach my $template ( @{ $args->{marc_template} } ) {
+		$count = {};
+		@marc_order = ();
+		sub my_count {
+			my $sf = shift;
+			my $nr = $count->{$sf}++;
+			push @marc_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 = $count_key marc_order = ",dump( @marc_order ),$/;
+		$pos_templates->{ $count_key } = $pos_template;
+		$marc_template_order->{ $pos_template } = [ @marc_order ];
+	}
+	warn "### from ",dump( $args->{marc_template} ), " created ", dump( $pos_templates ), " and ", dump( $marc_template_order );
+
+	my $m;
+
+	foreach my $r ( @{ $rec->{ $args->{from} } } ) {
+
+		my $i1 = $r->{i1} || ' ';
+		my $i2 = $r->{i2} || ' ';
+		$m = [ $args->{to}, $i1, $i2 ];
+
+		warn "### r = ",dump( $r );
+
+		my ( $new_r, $from_count, $to_count );
+		foreach my $sf ( keys %{$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 = $pos_templates->{ $from_count_key } ||
+			die "I don't have template for:\n$from_count_key\n## available templates\n", dump( $pos_templates );
+
+		warn "### selected template: |$template|\n";
+
+		$fill_in = {};
+
+		my @templates = split(/\|/, $template );
+		@templates = ( $template );
+
+		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 ) {
+					$fill_in->{$sf}->[$nr] = $v;
+					return $v;
+				} else {
+					die "requested subfield $from_sf/$from_nr but it's ",dump( $v );
+				}
+			}
+			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_order = ",dump( $marc_template_order );
+
+		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 );
+		}
+
+		warn "#### >>>> created marc: ", dump( $m );
+
+		push @marc_out, $m;
+	}
+
+	warn "### marc_template produced: ",dump( @marc_out );
+
+	foreach my $marc ( @marc_out ) {
+		warn "+++ ",dump( $marc );
+		push @{ $marc_record->[ $marc_record_offset ] }, $marc;
+	}
+}
+
 =head2 marc_count
 
 Return number of MARC records created using L</marc_duplicate>.
@@ -997,6 +1162,54 @@
 	}
 }
 
+=head2 frec
+
+Returns first value from field
+
+  $v = frec('200');
+  $v = frec('200','a');
+
+=cut
+
+sub frec {
+	my @out = rec(@_);
+	warn "rec(",dump(@_),") has more than one return value, ignoring\n" if $#out > 0;
+	return shift @out;
+}
+
+=head2 frec_eq
+
+=head2 frec_ne
+
+Check if first values from two fields are same or different
+
+  if ( frec_eq( 900 => 'a', 910 => 'c' ) ) {
+  	# values are same
+  } else {
+    # values are different
+  }
+
+Strictly speaking C<frec_eq> and C<frec_ne> wouldn't be needed if you
+could write something like:
+
+  if ( frec( '900','a' ) eq frec( '910','c' ) ) {
+  	# yada tada
+  }
+
+but you can't since our parser L<WebPAC::Parser> will remove all whitespaces
+in order to parse text and create invalid function C<eqfrec>.
+
+=cut
+
+sub frec_eq {
+	my ( $f1,$sf1, $f2, $sf2 ) = @_;
+	return (rec( $f1, $sf1 ))[0] eq (rec( $f2, $sf2 ))[0];
+}
+
+sub frec_ne {
+	return ! frec_eq( @_ );
+}
+
 =head2 regex
 
 Apply regex to some or all values
@@ -1232,7 +1445,6 @@
   $database_code = config();	# use _ from hash
   $database_name = config('name');
   $database_input_name = config('input name');
-  $tag = config('input normalize tag');
 
 Up to three levels are supported.