--- trunk/lib/WebPAC/Normalize.pm	2006/07/03 21:08:07	574
+++ trunk/lib/WebPAC/Normalize.pm	2006/10/08 18:21:26	752
@@ -2,17 +2,21 @@
 use Exporter 'import';
 @EXPORT = qw/
 	_set_rec _set_lookup
+	_set_load_row
 	_get_ds _clean_ds
 	_debug
+	_pack_subfields_hash
 
 	tag search display
 	marc marc_indicators marc_repeatable_subfield
 	marc_compose marc_leader
 	marc_duplicate marc_remove
+	marc_original_order
 
 	rec1 rec2 rec
 	regex prefix suffix surround
 	first lookup join_with
+	save_into_lookup
 
 	split_rec_on
 /;
@@ -22,8 +26,8 @@
 
 #use base qw/WebPAC::Common/;
 use Data::Dump qw/dump/;
-use Encode qw/from_to/;
 use Storable qw/dclone/;
+use Carp qw/confess/;
 
 # debugging warn(s)
 my $debug = 0;
@@ -35,11 +39,11 @@
 
 =head1 VERSION
 
-Version 0.10
+Version 0.23
 
 =cut
 
-our $VERSION = '0.10';
+our $VERSION = '0.23';
 
 =head1 SYNOPSIS
 
@@ -65,15 +69,23 @@
 Return data structure
 
   my $ds = WebPAC::Normalize::data_structure(
-  	lookup => $lookup->lookup_hash,
+	lookup => $lookup_hash,
 	row => $row,
 	rules => $normalize_pl_config,
 	marc_encoding => 'utf-8',
+	config => $config,
+	load_row_coderef => sub {
+		my ($database,$input,$mfn) = shift;
+		$store->load_row( database => $database, input => $input, id => $mfn );
+	},
   );
 
-Options C<lookup>, C<row>, C<rules> and C<log> are mandatory while all
+Options C<row>, C<rules> and C<log> are mandatory while all
 other are optional.
 
+C<load_row_coderef> is closure only used when executing lookups, so they will
+die if it's not defined.
+
 This function will B<die> if normalizastion can't be evaled.
 
 Since this function isn't exported you have to call it with 
@@ -81,6 +93,8 @@
 
 =cut
 
+my $load_row_coderef;
+
 sub data_structure {
 	my $arg = {@_};
 
@@ -88,9 +102,12 @@
 	die "need normalisation argument" unless ($arg->{rules});
 
 	no strict 'subs';
-	_set_lookup( $arg->{lookup} );
+	_set_lookup( $arg->{lookup} ) if defined($arg->{lookup});
 	_set_rec( $arg->{row} );
+	_set_config( $arg->{config} ) if defined($arg->{config});
 	_clean_ds( %{ $arg } );
+	$load_row_coderef = $arg->{load_row_coderef};
+
 	eval "$arg->{rules}";
 	die "error evaling $arg->{rules}: $@\n" if ($@);
 
@@ -111,6 +128,34 @@
 	$rec = shift or die "no record hash";
 }
 
+=head2 _set_config
+
+Set current config hash
+
+  _set_config( $config );
+
+Magic keys are:
+
+=over 4
+
+=item _
+
+Code of current database
+
+=item _mfn
+
+Current MFN
+
+=back
+
+=cut
+
+my $config;
+
+sub _set_config {
+	$config = shift;
+}
+
 =head2 _get_ds
 
 Return hash formatted as data structure
@@ -119,7 +164,7 @@
 
 =cut
 
-my ($out, $marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators);
+my ($out, $marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators, $leader);
 my ($marc_record_offset, $marc_fetch_offset) = (0, 0);
 
 sub _get_ds {
@@ -136,7 +181,7 @@
 
 sub _clean_ds {
 	my $a = {@_};
-	($out,$marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators) = ();
+	($out,$marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators, $leader) = ();
 	($marc_record_offset, $marc_fetch_offset) = (0,0);
 	$marc_encoding = $a->{marc_encoding};
 }
@@ -155,6 +200,37 @@
 	$lookup = shift;
 }
 
+=head2 _get_lookup
+
+Get current lookup hash
+
+  my $lookup = _get_lookup();
+
+=cut
+
+sub _get_lookup {
+	return $lookup;
+}
+
+=head2 _set_load_row
+
+Setup code reference which will return L<data_structure> from
+L<WebPAC::Store>
+
+  _set_load_row(sub {
+		my ($database,$input,$mfn) = @_;
+		$store->load_row( database => $database, input => $input, id => $mfn );
+  });
+
+=cut
+
+sub _set_load_row {
+	my $coderef = shift;
+	confess "argument isn't CODE" unless ref($coderef) eq 'CODE';
+
+	$load_row_coderef = $coderef;
+}
+
 =head2 _get_marc_fields
 
 Get all fields defined by calls to C<marc>
@@ -326,7 +402,7 @@
 		warn "## saved/3 ", dump( $field ),$/ if ($debug);
 	}
 
-	return @m;
+	return \@m;
 }
 
 =head2 _debug
@@ -411,9 +487,9 @@
 	my ($offset,$value) = @_;
 
 	if ($offset) {
-		$out->{' leader'}->{ $offset } = $value;
+		$leader->{ $offset } = $value;
 	} else {
-		return $out->{' leader'};
+		return $leader;
 	}
 }
 
@@ -438,7 +514,6 @@
 	foreach (@_) {
 		my $v = $_;		# make var read-write for Encode
 		next unless (defined($v) && $v !~ /^\s*$/);
-		from_to($v, 'iso-8859-2', $marc_encoding) if ($marc_encoding);
 		my ($i1,$i2) = defined($marc_indicators->{$f}) ? @{ $marc_indicators->{$f} } : (' ',' ');
 		if (defined $sf) {
 			push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $i1, $i2, $sf => $v ];
@@ -495,6 +570,9 @@
   	'c', rec('200','c')
   );
 
+If you specify C<+> for subfield, value will be appended
+to previous defined subfield.
+
 =cut
 
 sub marc_compose {
@@ -504,17 +582,26 @@
 	my ($i1,$i2) = defined($marc_indicators->{$f}) ? @{ $marc_indicators->{$f} } : (' ',' ');
 	my $m = [ $f, $i1, $i2 ];
 
+	warn "### marc_compose input subfields = ", dump(@_),$/ if ($debug > 2);
+
+	if ($#_ % 2 != 1) {
+		die "ERROR: marc_compose",dump($f,@_)," not valid (must be even).\nDo you need to add first() or join() around some argument?\n";
+	}
+
 	while (@_) {
-		my $sf = shift or die "marc_compose $f needs subfield";
+		my $sf = shift;
 		my $v = shift;
 
 		next unless (defined($v) && $v !~ /^\s*$/);
-		from_to($v, 'iso-8859-2', $marc_encoding) if ($marc_encoding);
-		push @$m, ( $sf, $v );
 		warn "## ++ marc_compose($f,$sf,$v) ", dump( $m ),$/ if ($debug > 1);
+		if ($sf ne '+') {
+			push @$m, ( $sf, $v );
+		} else {
+			$m->[ $#$m ] .= $v;
+		}
 	}
 
-	warn "## marc_compose(d) ", dump( $m ),$/ if ($debug > 1);
+	warn "## marc_compose current marc = ", dump( $m ),$/ if ($debug > 1);
 
 	push @{ $marc_record->[ $marc_record_offset ] }, $m if ($#{$m} > 2);
 }
@@ -565,7 +652,8 @@
 
 	warn "### marc_remove before = ", dump( $marc ), $/ if ($debug > 2);
 
-	foreach my $i ( 0 .. $#{ $marc } ) {
+	my $i = 0;
+	foreach ( 0 .. $#{ $marc } ) {
 		last unless (defined $marc->[$i]);
 		warn "#### working on ",dump( @{ $marc->[$i] }), $/ if ($debug > 3);
 		if ($marc->[$i]->[0] eq $f) {
@@ -591,6 +679,7 @@
 				}
 			}
 		}
+		$i++;
 	}
 
 	warn "### marc_remove($f", $sf ? ",$sf" : "", ") after = ", dump( $marc ), $/ if ($debug > 2);
@@ -600,11 +689,138 @@
 	warn "## full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 1);
 }
 
+=head2 marc_original_order
+
+Copy all subfields preserving original order to marc field.
+
+  marc_original_order( marc_field_number, original_input_field_number );
+
+Please note that field numbers are consistent with other commands (marc
+field number first), but somewhat counter-intuitive (destination and then
+source).
+
+You might want to use this command if you are just renaming subfields or
+using pre-processing modify_record in C<config.yml> and don't need any
+post-processing or want to preserve order of original subfields.
+
+
+=cut
+
+sub marc_original_order {
+
+	my ($to, $from) = @_;
+	die "marc_original_order needs from and to fields\n" unless ($from && $to);
+
+	return unless defined($rec->{$from});
+
+	my $r = $rec->{$from};
+	die "record field $from isn't array\n" unless (ref($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);
+
+	foreach my $d (@$r) {
+
+		if (! defined($d->{subfields}) && ref($d->{subfields}) ne 'ARRAY') {
+			warn "# marc_original_order($to,$from): field $from doesn't have subfields specification\n";
+			next;
+		}
+	
+		my @sfs = @{ $d->{subfields} };
+
+		die "field $from doesn't have even number of subfields specifications\n" unless($#sfs % 2 == 1);
+
+		warn "#--> d: ",dump($d), "\n#--> sfs: ",dump(@sfs),$/ if ($debug > 2);
+
+		my $m = [ $to, $i1, $i2 ];
+
+		while (my $sf = shift @sfs) {
+
+			warn "#--> sf: ",dump($sf), $/ if ($debug > 2);
+			my $offset = shift @sfs;
+			die "corrupted sufields specification for field $from\n" unless defined($offset);
+
+			my $v;
+			if (ref($d->{$sf}) eq 'ARRAY') {
+				$v = $d->{$sf}->[$offset] if (defined($d->{$sf}->[$offset]));
+			} elsif ($offset == 0) {
+				$v = $d->{$sf};
+			} else {
+				die "field $from subfield '$sf' need occurence $offset which doesn't exist", dump($d->{$sf});
+			}
+			push @$m, ( $sf, $v ) if (defined($v));
+		}
+
+		if ($#{$m} > 2) {
+			push @{ $marc_record->[ $marc_record_offset ] }, $m;
+		}
+	}
+
+	warn "## marc_record = ", dump( $marc_record ),$/ if ($debug > 1);
+}
+
+
 =head1 Functions to extract data from input
 
 This function should be used inside functions to create C<data_structure> described
 above.
 
+=head2 _pack_subfields_hash
+
+ @subfields = _pack_subfields_hash( $h );
+ $subfields = _pack_subfields_hash( $h, 1 );
+
+Return each subfield value in array or pack them all together and return scalar
+with subfields (denoted by C<^>) and values.
+
+=cut
+
+sub _pack_subfields_hash {
+
+	warn "## _pack_subfields_hash( ",dump(@_), " )\n" if ($debug > 1);
+
+	my ($h,$include_subfields) = @_;
+
+	if ( defined($h->{subfields}) ) {
+		my $sfs = delete $h->{subfields} || die "no subfields?";
+		my @out;
+		while (@$sfs) {
+			my $sf = shift @$sfs;
+			push @out, '^' . $sf if ($include_subfields);
+			my $o = shift @$sfs;
+			if ($o == 0 && ref( $h->{$sf} ) ne 'ARRAY' ) {
+				# single element subfields are not arrays
+#warn "====> $sf $o / $#$sfs ", dump( $sfs, $h->{$sf} ), "\n";
+
+				push @out, $h->{$sf};
+			} else {
+#warn "====> $sf $o / $#$sfs ", dump( $sfs, $h->{$sf} ), "\n";
+				push @out, $h->{$sf}->[$o];
+			}
+		}
+		if ($include_subfields) {
+			return join('', @out);
+		} else {
+			return @out;
+		}
+	} else {
+		if ($include_subfields) {
+			my $out = '';
+			foreach my $sf (sort keys %$h) {
+				if (ref($h->{$sf}) eq 'ARRAY') {
+					$out .= '^' . $sf . join('^' . $sf, @{ $h->{$sf} });
+				} else {
+					$out .= '^' . $sf . $h->{$sf};
+				}
+			}
+			return $out;
+		} else {
+			# FIXME this should probably be in alphabetical order instead of hash order
+			values %{$h};
+		}
+	}
+}
+
 =head2 rec1
 
 Return all values in some field
@@ -621,13 +837,15 @@
 	return unless (defined($rec) && defined($rec->{$f}));
 	warn "rec1($f) = ", dump( $rec->{$f} ), $/ if ($debug > 1);
 	if (ref($rec->{$f}) eq 'ARRAY') {
-		return map { 
-			if (ref($_) eq 'HASH') {
-				values %{$_};
+		my @out;
+		foreach my $h ( @{ $rec->{$f} } ) {
+			if (ref($h) eq 'HASH') {
+				push @out, ( _pack_subfields_hash( $h ) );
 			} else {
-				$_;
+				push @out, $h;
 			}
-		} @{ $rec->{$f} };
+		}
+		return @out;
 	} elsif( defined($rec->{$f}) ) {
 		return $rec->{$f};
 	}
@@ -645,7 +863,14 @@
 	my $f = shift;
 	return unless (defined($rec && $rec->{$f}));
 	my $sf = shift;
-	return map { $_->{$sf} } grep { ref($_) eq 'HASH' && $_->{$sf} } @{ $rec->{$f} };
+	warn "rec2($f,$sf) = ", dump( $rec->{$f} ), $/ if ($debug > 1);
+	return map {
+		if (ref($_->{$sf}) eq 'ARRAY') {
+			@{ $_->{$sf} };
+		} else {
+			$_->{$sf};
+		}
+	} grep { ref($_) eq 'HASH' && $_->{$sf} } @{ $rec->{$f} };
 }
 
 =head2 rec
@@ -655,13 +880,24 @@
   @v = rec('200')
   @v = rec('200','a')
 
+If rec() returns just single value, it will
+return scalar, not array.
+
 =cut
 
 sub rec {
+	my @out;
 	if ($#_ == 0) {
-		return rec1(@_);
+		@out = rec1(@_);
 	} elsif ($#_ == 1) {
-		return rec2(@_);
+		@out = rec2(@_);
+	}
+	if ($#out == 0 && ! wantarray) {
+		return $out[0];
+	} elsif (@out) {
+		return @out;
+	} else {
+		return '';
 	}
 }
 
@@ -694,7 +930,7 @@
 =cut
 
 sub prefix {
-	my $p = shift or die "prefix needs string as first argument";
+	my $p = shift or return;
 	return map { $p . $_ } grep { defined($_) } @_;
 }
 
@@ -742,21 +978,224 @@
 
 Consult lookup hashes for some value
 
-  @v = lookup( $v );
-  @v = lookup( @v );
+  @v = lookup(
+  	sub {
+		'ffkk/peri/mfn'.rec('000')
+	},
+	'ffkk','peri','200-a-200-e',
+	sub {
+		first(rec(200,'a')).' '.first(rec('200','e'))
+	}
+  );
+
+Code like above will be B<automatically generated> using L<WebPAC::Parse> from
+normal lookup definition in C<conf/lookup/something.pl> which looks like:
+
+  lookup(
+	# which results to return from record recorded in lookup
+	sub { 'ffkk/peri/mfn' . rec('000') },
+	# from which database and input
+	'ffkk','peri',
+	# such that following values match
+	sub { first(rec(200,'a')) . ' ' . first(rec('200','e')) },
+	# if this part is missing, we will try to match same fields
+	# from lookup record and current one, or you can override
+	# which records to use from current record using
+	sub { rec('900','x') . ' ' . rec('900','y') },
+  )
+
+You can think about this lookup as SQL (if that helps):
+
+  select
+  	sub { what }
+  from
+  	database, input
+  where
+    sub { filter from lookuped record }
+  having
+    sub { optional filter on current record }
+
+Easy as pie, right?
 
 =cut
 
 sub lookup {
-	my $k = shift or return;
-	return unless (defined($lookup->{$k}));
-	if (ref($lookup->{$k}) eq 'ARRAY') {
-		return @{ $lookup->{$k} };
+	my ($what, $database, $input, $key, $having) = @_;
+
+	confess "lookup needs 5 arguments: what, database, input, key, having" unless ($#_ == 4);
+
+	warn "## lookup ($database, $input, $key)", $/ if ($debug > 1);
+	return unless (defined($lookup->{$database}->{$input}->{$key}));
+
+	confess "lookup really need load_row_coderef added to data_structure\n" unless ($load_row_coderef);
+
+	my $mfns;
+	my @having = $having->();
+
+	warn "## having = ", dump( @having ) if ($debug > 2);
+
+	foreach my $h ( @having ) {
+		if (defined($lookup->{$database}->{$input}->{$key}->{$h})) {
+			warn "lookup for $database/$input/$key/$h return ",dump($lookup->{$database}->{$input}->{$key}->{$h}),"\n" if ($debug);
+			$mfns->{$_}++ foreach keys %{ $lookup->{$database}->{$input}->{$key}->{$h} };
+		}
+	}
+
+	return unless ($mfns);
+
+	my @mfns = sort keys %$mfns;
+
+	warn "# lookup loading $database/$input/$key mfn ", join(",",@mfns)," having ",dump(@having),"\n" if ($debug);
+
+	my $old_rec = $rec;
+	my @out;
+
+	foreach my $mfn (@mfns) {
+		$rec = $load_row_coderef->( $database, $input, $mfn );
+
+		warn "got $database/$input/$mfn = ", dump($rec), $/ if ($debug);
+
+		my @vals = $what->();
+
+		push @out, ( @vals );
+
+		warn "lookup for mfn $mfn returned ", dump(@vals), $/ if ($debug);
+	}
+
+#	if (ref($lookup->{$k}) eq 'ARRAY') {
+#		return @{ $lookup->{$k} };
+#	} else {
+#		return $lookup->{$k};
+#	}
+
+	$rec = $old_rec;
+
+	warn "## lookup returns = ", dump(@out), $/ if ($debug);
+
+	if ($#out == 0) {
+		return $out[0];
 	} else {
-		return $lookup->{$k};
+		return @out;
 	}
 }
 
+=head2 save_into_lookup
+
+Save value into lookup. It associates current database, input
+and specific keys with one or more values which will be
+associated over MFN.
+
+MFN will be extracted from first occurence current of field 000
+in current record, or if it doesn't exist from L<_set_config> C<_mfn>.
+
+  my $nr = save_into_lookup($database,$input,$key,sub {
+ 	# code which produce one or more values 
+  });
+
+It returns number of items saved.
+
+This function shouldn't be called directly, it's called from code created by
+L<WebPAC::Parser>. 
+
+=cut
+
+sub save_into_lookup {
+	my ($database,$input,$key,$coderef) = @_;
+	die "save_into_lookup needs database" unless defined($database);
+	die "save_into_lookup needs input" unless defined($input);
+	die "save_into_lookup needs key" unless defined($key);
+	die "save_into_lookup needs CODE" unless ( defined($coderef) && ref($coderef) eq 'CODE' );
+
+	warn "## save_into_lookup rec = ", dump($rec), " config = ", dump($config), $/ if ($debug > 2);
+
+	my $mfn = 
+		defined($rec->{'000'}->[0])	?	$rec->{'000'}->[0]	:
+		defined($config->{_mfn})	?	$config->{_mfn}		:
+										die "mfn not defined or zero";
+
+	my $nr = 0;
+
+	foreach my $v ( $coderef->() ) {
+		$lookup->{$database}->{$input}->{$key}->{$v}->{$mfn}++;
+		warn "# saved lookup $database/$input/$key [$v] $mfn\n" if ($debug > 1);
+		$nr++;
+	}
+
+	return $nr;
+}
+
+=head2 config
+
+Consult config values stored in C<config.yml>
+
+  # return database code (key under databases in yaml)
+  $database_code = config();	# use _ from hash
+  $database_name = config('name');
+  $database_input_name = config('input name');
+  $tag = config('input normalize tag');
+
+Up to three levels are supported.
+
+=cut
+
+sub config {
+	return unless ($config);
+
+	my $p = shift;
+
+	$p ||= '';
+
+	my $v;
+
+	warn "### getting config($p)\n" if ($debug > 1);
+
+	my @p = split(/\s+/,$p);
+	if ($#p < 0) {
+		$v = $config->{ '_' };	# special, database code
+	} else {
+
+		my $c = dclone( $config );
+
+		foreach my $k (@p) {
+			warn "### k: $k c = ",dump($c),$/ if ($debug > 1);
+			if (ref($c) eq 'ARRAY') {
+				$c = shift @$c;
+				warn "config($p) taking first occurence of '$k', probably not what you wanted!\n";
+				last;
+			}
+
+			if (! defined($c->{$k}) ) {
+				$c = undef;
+				last;
+			} else {
+				$c = $c->{$k};
+			}
+		}
+		$v = $c if ($c);
+
+	}
+
+	warn "## config( '$p' ) = ",dump( $v ),$/ if ($v && $debug);
+	warn "config( '$p' ) is empty\n" if (! $v);
+
+	return $v;
+}
+
+=head2 id
+
+Returns unique id of this record
+
+  $id = id();
+
+Returns C<42/2> for 2nd occurence of MFN 42.
+
+=cut
+
+sub id {
+	my $mfn = $config->{_mfn} || die "no _mfn in config data";
+	return $mfn . $#{$marc_record} ? $#{$marc_record} + 1 : '';
+}
+
 =head2 join_with
 
 Joins walues with some delimiter
@@ -767,7 +1206,10 @@
 
 sub join_with {
 	my $d = shift;
-	return join($d, grep { defined($_) && $_ ne '' } @_);
+	warn "### join_with('$d',",dump(@_),")\n" if ($debug > 2);
+	my $v = join($d, grep { defined($_) && $_ ne '' } @_);
+	return '' unless defined($v);
+	return $v;
 }
 
 =head2 split_rec_on
@@ -795,7 +1237,7 @@
 	my $v = shift @r;
 	warn "### first rec($fld,$sf) = ",dump($v),$/ if ($debug > 2);
 
-	return '' if( ! defined($v) || $v =~ /^\s*$/);
+	return '' if ( ! defined($v) || $v =~ /^\s*$/);
 
 	my @s = split( $regex, $v );
 	warn "## split_rec_on($fld,$sf,$regex,$part) = ",dump(@s),$/ if ($debug > 1);