--- trunk/lib/WebPAC/Normalize.pm 2007/11/10 00:05:36 1019 +++ trunk/lib/WebPAC/Normalize.pm 2007/11/10 11:11:16 1021 @@ -39,16 +39,20 @@ # debugging warn(s) my $debug = 0; +# FIXME use WebPAC::Normalize::ISBN; push @EXPORT, ( 'isbn_10', 'isbn_13' ); +use WebPAC::Normalize::MARC; +push @EXPORT, ( 'marc_template' ); + =head1 NAME WebPAC::Normalize - describe normalisaton rules using sets =cut -our $VERSION = '0.34'; +our $VERSION = '0.35'; =head1 SYNOPSIS @@ -134,6 +138,14 @@ $rec = shift or die "no record hash"; } +=head2 + + my $rec = _get_rec(); + +=cut + +sub _get_rec { $rec }; + =head2 _set_config Set current config hash @@ -860,150 +872,6 @@ 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 @@ -1017,6 +885,17 @@ return $#{ $marc_record }; } +=head2 _marc_push + + _marc_push( $marc ); + +=cut + +sub _marc_push { + my $marc = shift || die "no marc?"; + push @{ $marc_record->[ $marc_record_offset ] }, $marc; +} + =head1 Functions to extract data from input