/[webpac2]/trunk/lib/WebPAC/Normalize.pm
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Diff of /trunk/lib/WebPAC/Normalize.pm

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 817 by dpavlin, Thu Apr 5 21:50:14 2007 UTC revision 990 by dpavlin, Sun Nov 4 13:27:12 2007 UTC
# Line 1  Line 1 
1  package WebPAC::Normalize;  package WebPAC::Normalize;
2  use Exporter 'import';  use Exporter 'import';
3  @EXPORT = qw/  our @EXPORT = qw/
4          _set_rec _set_lookup          _set_ds _set_lookup
5          _set_load_row          _set_load_row
6          _get_ds _clean_ds          _get_ds _clean_ds
7          _debug          _debug
8          _pack_subfields_hash          _pack_subfields_hash
9    
10          tag search display          search_display search display sorted
11    
12          marc marc_indicators marc_repeatable_subfield          marc marc_indicators marc_repeatable_subfield
13          marc_compose marc_leader marc_fixed          marc_compose marc_leader marc_fixed
14          marc_duplicate marc_remove marc_count          marc_duplicate marc_remove marc_count
15          marc_original_order          marc_original_order
16    
17          rec1 rec2 rec          rec1 rec2 rec
18            frec
19          regex prefix suffix surround          regex prefix suffix surround
20          first lookup join_with          first lookup join_with
21          save_into_lookup          save_into_lookup
# Line 22  use Exporter 'import'; Line 24  use Exporter 'import';
24    
25          get set          get set
26          count          count
27    
28  /;  /;
29    
30  use warnings;  use warnings;
# Line 35  use Carp qw/confess/; Line 38  use Carp qw/confess/;
38  # debugging warn(s)  # debugging warn(s)
39  my $debug = 0;  my $debug = 0;
40    
41    use WebPAC::Normalize::ISBN;
42    push @EXPORT, ( 'isbn_10', 'isbn_13' );
43    
44  =head1 NAME  =head1 NAME
45    
46  WebPAC::Normalize - describe normalisaton rules using sets  WebPAC::Normalize - describe normalisaton rules using sets
47    
 =head1 VERSION  
   
 Version 0.28  
   
48  =cut  =cut
49    
50  our $VERSION = '0.28';  our $VERSION = '0.32';
51    
52  =head1 SYNOPSIS  =head1 SYNOPSIS
53    
# Line 59  means that you check it's validity befor Line 60  means that you check it's validity befor
60  C<perl -c normalize.pl>.  C<perl -c normalize.pl>.
61    
62  Normalisation can generate multiple output normalized data. For now, supported output  Normalisation can generate multiple output normalized data. For now, supported output
63  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
64  C<marc>.  C<marc>.
65    
66  =head1 FUNCTIONS  =head1 FUNCTIONS
# Line 78  Return data structure Line 79  Return data structure
79          marc_encoding => 'utf-8',          marc_encoding => 'utf-8',
80          config => $config,          config => $config,
81          load_row_coderef => sub {          load_row_coderef => sub {
82                  my ($database,$input,$mfn) = shift;                  my ($database,$input,$mfn) = @_;
83                  $store->load_row( database => $database, input => $input, id => $mfn );                  $store->load_row( database => $database, input => $input, id => $mfn );
84          },          },
85    );    );
# Line 106  sub data_structure { Line 107  sub data_structure {
107    
108          no strict 'subs';          no strict 'subs';
109          _set_lookup( $arg->{lookup} ) if defined($arg->{lookup});          _set_lookup( $arg->{lookup} ) if defined($arg->{lookup});
110          _set_rec( $arg->{row} );          _set_ds( $arg->{row} );
111          _set_config( $arg->{config} ) if defined($arg->{config});          _set_config( $arg->{config} ) if defined($arg->{config});
112          _clean_ds( %{ $arg } );          _clean_ds( %{ $arg } );
113          $load_row_coderef = $arg->{load_row_coderef};          $load_row_coderef = $arg->{load_row_coderef};
# Line 117  sub data_structure { Line 118  sub data_structure {
118          return _get_ds();          return _get_ds();
119  }  }
120    
121  =head2 _set_rec  =head2 _set_ds
122    
123  Set current record hash  Set current record hash
124    
125    _set_rec( $rec );    _set_ds( $rec );
126    
127  =cut  =cut
128    
129  my $rec;  my $rec;
130    
131  sub _set_rec {  sub _set_ds {
132          $rec = shift or die "no record hash";          $rec = shift or die "no record hash";
133  }  }
134    
# Line 171  my ($out, $marc_record, $marc_encoding, Line 172  my ($out, $marc_record, $marc_encoding,
172  my ($marc_record_offset, $marc_fetch_offset) = (0, 0);  my ($marc_record_offset, $marc_fetch_offset) = (0, 0);
173    
174  sub _get_ds {  sub _get_ds {
175    #warn "## out = ",dump($out);
176          return $out;          return $out;
177  }  }
178    
# Line 442  sub _debug { Line 444  sub _debug {
444    
445  Those functions generally have to first in your normalization file.  Those functions generally have to first in your normalization file.
446    
447  =head2 tag  =head2 search_display
448    
449  Define new tag for I<search> and I<display>.  Define output for L<search> and L<display> at the same time
450    
451    tag('Title', rec('200','a') );    search_display('Title', rec('200','a') );
452    
453    
454  =cut  =cut
455    
456  sub tag {  sub search_display {
457          my $name = shift or die "tag needs name as first argument";          my $name = shift or die "search_display needs name as first argument";
458          my @o = grep { defined($_) && $_ ne '' } @_;          my @o = grep { defined($_) && $_ ne '' } @_;
459          return unless (@o);          return unless (@o);
         $out->{$name}->{tag} = $name;  
460          $out->{$name}->{search} = \@o;          $out->{$name}->{search} = \@o;
461          $out->{$name}->{display} = \@o;          $out->{$name}->{display} = \@o;
462  }  }
463    
464    =head2 tag
465    
466    Old name for L<search_display>, but supported
467    
468    =cut
469    
470    sub tag {
471            search_display( @_ );
472    }
473    
474  =head2 display  =head2 display
475    
476  Define tag just for I<display>  Define output just for I<display>
477    
478    @v = display('Title', rec('200','a') );    @v = display('Title', rec('200','a') );
479    
480  =cut  =cut
481    
482  sub display {  sub _field {
483          my $name = shift or die "display needs name as first argument";          my $type = shift or confess "need type -- BUG?";
484            my $name = shift or confess "needs name as first argument";
485          my @o = grep { defined($_) && $_ ne '' } @_;          my @o = grep { defined($_) && $_ ne '' } @_;
486          return unless (@o);          return unless (@o);
487          $out->{$name}->{tag} = $name;          $out->{$name}->{$type} = \@o;
         $out->{$name}->{display} = \@o;  
488  }  }
489    
490    sub display { _field( 'display', @_ ) }
491    
492  =head2 search  =head2 search
493    
494  Prepare values just for I<search>  Prepare values just for I<search>
# Line 484  Prepare values just for I<search> Line 497  Prepare values just for I<search>
497    
498  =cut  =cut
499    
500  sub search {  sub search { _field( 'search', @_ ) }
501          my $name = shift or die "search needs name as first argument";  
502          my @o = grep { defined($_) && $_ ne '' } @_;  =head2 sorted
503          return unless (@o);  
504          $out->{$name}->{tag} = $name;  Insert into lists which will be automatically sorted
505          $out->{$name}->{search} = \@o;  
506  }   sorted('Title', rec('200','a') );
507    
508    =cut
509    
510    sub sorted { _field( 'sorted', @_ ) }
511    
512    
513  =head2 marc_leader  =head2 marc_leader
514    
# Line 535  sub marc_fixed { Line 553  sub marc_fixed {
553          my ($f, $pos, $val) = @_;          my ($f, $pos, $val) = @_;
554          die "need marc(field, position, value)" unless defined($f) && defined($pos);          die "need marc(field, position, value)" unless defined($f) && defined($pos);
555    
556            confess "need val" unless defined $val;
557    
558          my $update = 0;          my $update = 0;
559    
560          map {          map {
561                  if ($_->[0] eq $f) {                  if ($_->[0] eq $f) {
562                          my $old = $_->[1];                          my $old = $_->[1];
563                          if (length($old) < $pos) {                          if (length($old) <= $pos) {
564                                  $_->[1] .= ' ' x ( $pos - length($old) ) . $val;                                  $_->[1] .= ' ' x ( $pos - length($old) ) . $val;
565                                  warn "## marc_fixed($f,$pos,'$val') append '$old' -> '$_->[1]'\n" if ($debug > 1);                                  warn "## marc_fixed($f,$pos,'$val') append '$old' -> '$_->[1]'\n" if ($debug > 1);
566                          } else {                          } else {
# Line 872  sub _pack_subfields_hash { Line 892  sub _pack_subfields_hash {
892    
893          my ($h,$include_subfields) = @_;          my ($h,$include_subfields) = @_;
894    
895            # sanity and ease of use
896            return $h if (ref($h) ne 'HASH');
897    
898          if ( defined($h->{subfields}) ) {          if ( defined($h->{subfields}) ) {
899                  my $sfs = delete $h->{subfields} || die "no subfields?";                  my $sfs = delete $h->{subfields} || die "no subfields?";
900                  my @out;                  my @out;
# Line 976  return scalar, not array. Line 999  return scalar, not array.
999    
1000  =cut  =cut
1001    
1002    sub frec {
1003            my @out = rec(@_);
1004            warn "rec(",dump(@_),") has more than one return value, ignoring\n" if $#out > 0;
1005            return shift @out;
1006    }
1007    
1008  sub rec {  sub rec {
1009          my @out;          my @out;
1010          if ($#_ == 0) {          if ($#_ == 0) {
# Line 1021  Prefix all values with a string Line 1050  Prefix all values with a string
1050  =cut  =cut
1051    
1052  sub prefix {  sub prefix {
1053          my $p = shift or return;          my $p = shift;
1054            return @_ unless defined( $p );
1055          return map { $p . $_ } grep { defined($_) } @_;          return map { $p . $_ } grep { defined($_) } @_;
1056  }  }
1057    
# Line 1034  suffix all values with a string Line 1064  suffix all values with a string
1064  =cut  =cut
1065    
1066  sub suffix {  sub suffix {
1067          my $s = shift or die "suffix needs string as first argument";          my $s = shift;
1068            return @_ unless defined( $s );
1069          return map { $_ . $s } grep { defined($_) } @_;          return map { $_ . $s } grep { defined($_) } @_;
1070  }  }
1071    
# Line 1047  surround all values with a two strings Line 1078  surround all values with a two strings
1078  =cut  =cut
1079    
1080  sub surround {  sub surround {
1081          my $p = shift or die "surround need prefix as first argument";          my $p = shift;
1082          my $s = shift or die "surround needs suffix as second argument";          my $s = shift;
1083            $p = '' unless defined( $p );
1084            $s = '' unless defined( $s );
1085          return map { $p . $_ . $s } grep { defined($_) } @_;          return map { $p . $_ . $s } grep { defined($_) } @_;
1086  }  }
1087    
# Line 1223  Consult config values stored in C<config Line 1256  Consult config values stored in C<config
1256    $database_code = config();    # use _ from hash    $database_code = config();    # use _ from hash
1257    $database_name = config('name');    $database_name = config('name');
1258    $database_input_name = config('input name');    $database_input_name = config('input name');
   $tag = config('input normalize tag');  
1259    
1260  Up to three levels are supported.  Up to three levels are supported.
1261    

Legend:
Removed from v.817  
changed lines
  Added in v.990

  ViewVC Help
Powered by ViewVC 1.1.26