/[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 788 by dpavlin, Sun Dec 10 12:56:59 2006 UTC revision 979 by dpavlin, Sat Nov 3 12:37:43 2007 UTC
# Line 7  use Exporter 'import'; Line 7  use Exporter 'import';
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_compose marc_leader marc_fixed
14          marc_duplicate marc_remove          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 21  use Exporter 'import'; Line 23  use Exporter 'import';
23          split_rec_on          split_rec_on
24    
25          get set          get set
26            count
27  /;  /;
28    
29  use warnings;  use warnings;
# Line 39  my $debug = 0; Line 42  my $debug = 0;
42    
43  WebPAC::Normalize - describe normalisaton rules using sets  WebPAC::Normalize - describe normalisaton rules using sets
44    
 =head1 VERSION  
   
 Version 0.25  
   
45  =cut  =cut
46    
47  our $VERSION = '0.25';  our $VERSION = '0.31';
48    
49  =head1 SYNOPSIS  =head1 SYNOPSIS
50    
# Line 58  means that you check it's validity befor Line 57  means that you check it's validity befor
57  C<perl -c normalize.pl>.  C<perl -c normalize.pl>.
58    
59  Normalisation can generate multiple output normalized data. For now, supported output  Normalisation can generate multiple output normalized data. For now, supported output
60  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
61  C<marc>.  C<marc>.
62    
63  =head1 FUNCTIONS  =head1 FUNCTIONS
# Line 77  Return data structure Line 76  Return data structure
76          marc_encoding => 'utf-8',          marc_encoding => 'utf-8',
77          config => $config,          config => $config,
78          load_row_coderef => sub {          load_row_coderef => sub {
79                  my ($database,$input,$mfn) = shift;                  my ($database,$input,$mfn) = @_;
80                  $store->load_row( database => $database, input => $input, id => $mfn );                  $store->load_row( database => $database, input => $input, id => $mfn );
81          },          },
82    );    );
# Line 103  sub data_structure { Line 102  sub data_structure {
102          die "need row argument" unless ($arg->{row});          die "need row argument" unless ($arg->{row});
103          die "need normalisation argument" unless ($arg->{rules});          die "need normalisation argument" unless ($arg->{rules});
104    
105            # FIXME load this conditionally
106    #       use WebPAC::Normalize::ISBN;
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_rec( $arg->{row} );
# Line 166  Return hash formatted as data structure Line 168  Return hash formatted as data structure
168    
169  =cut  =cut
170    
171  my ($out, $marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators, $leader);  my ($out, $marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators, $marc_leader);
172  my ($marc_record_offset, $marc_fetch_offset) = (0, 0);  my ($marc_record_offset, $marc_fetch_offset) = (0, 0);
173    
174  sub _get_ds {  sub _get_ds {
175    
176    warn "## out = ",dump($out);
177    
178          return $out;          return $out;
179  }  }
180    
# Line 183  Clean data structure hash for next recor Line 188  Clean data structure hash for next recor
188    
189  sub _clean_ds {  sub _clean_ds {
190          my $a = {@_};          my $a = {@_};
191          ($out,$marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators, $leader) = ();          ($out,$marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators, $marc_leader) = ();
192          ($marc_record_offset, $marc_fetch_offset) = (0,0);          ($marc_record_offset, $marc_fetch_offset) = (0,0);
193          $marc_encoding = $a->{marc_encoding};          $marc_encoding = $a->{marc_encoding};
194  }  }
# Line 288  will return 42th copy record (if it exis Line 293  will return 42th copy record (if it exis
293    
294  =cut  =cut
295    
296    my $fetch_pos;
297    
298  sub _get_marc_fields {  sub _get_marc_fields {
299    
300          my $arg = {@_};          my $arg = {@_};
301          warn "### _get_marc_fields arg: ", dump($arg), $/ if ($debug > 2);          warn "### _get_marc_fields arg: ", dump($arg), $/ if ($debug > 2);
302          my $offset = $marc_fetch_offset;          $fetch_pos = $marc_fetch_offset;
303          if ($arg->{offset}) {          if ($arg->{offset}) {
304                  $offset = $arg->{offset};                  $fetch_pos = $arg->{offset};
305          } elsif($arg->{fetch_next}) {          } elsif($arg->{fetch_next}) {
306                  $marc_fetch_offset++;                  $marc_fetch_offset++;
307          }          }
# Line 303  sub _get_marc_fields { Line 310  sub _get_marc_fields {
310    
311          warn "### full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 2);          warn "### full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 2);
312    
313          my $marc_rec = $marc_record->[ $offset ];          my $marc_rec = $marc_record->[ $fetch_pos ];
314    
315          warn "## _get_marc_fields (at offset: $offset) -- marc_record = ", dump( @$marc_rec ), $/ if ($debug > 1);          warn "## _get_marc_fields (at offset: $fetch_pos) -- marc_record = ", dump( @$marc_rec ), $/ if ($debug > 1);
316    
317          return if (! $marc_rec || ref($marc_rec) ne 'ARRAY' || $#{ $marc_rec } < 0);          return if (! $marc_rec || ref($marc_rec) ne 'ARRAY' || $#{ $marc_rec } < 0);
318    
# Line 326  sub _get_marc_fields { Line 333  sub _get_marc_fields {
333    
334          if ($debug) {          if ($debug) {
335                  warn "## marc_repeatable_subfield = ", dump( $marc_repeatable_subfield ), $/ if ( $marc_repeatable_subfield );                  warn "## marc_repeatable_subfield = ", dump( $marc_repeatable_subfield ), $/ if ( $marc_repeatable_subfield );
336                  warn "## marc_record[$offset] = ", dump( $marc_rec ), $/;                  warn "## marc_record[$fetch_pos] = ", dump( $marc_rec ), $/;
337                  warn "## sorted_marc_record = ", dump( \@sorted_marc_record ), $/;                  warn "## sorted_marc_record = ", dump( \@sorted_marc_record ), $/;
338                  warn "## subfield count = ", dump( $u ), $/;                  warn "## subfield count = ", dump( $u ), $/;
339          }          }
# Line 407  sub _get_marc_fields { Line 414  sub _get_marc_fields {
414          return \@m;          return \@m;
415  }  }
416    
417    =head2 _get_marc_leader
418    
419    Return leader from currently fetched record by L</_get_marc_fields>
420    
421      print WebPAC::Normalize::_get_marc_leader();
422    
423    =cut
424    
425    sub _get_marc_leader {
426            die "no fetch_pos, did you called _get_marc_fields first?" unless ( defined( $fetch_pos ) );
427            return $marc_leader->[ $fetch_pos ];
428    }
429    
430  =head2 _debug  =head2 _debug
431    
432  Change level of debug warnings  Change level of debug warnings
# Line 426  sub _debug { Line 446  sub _debug {
446    
447  Those functions generally have to first in your normalization file.  Those functions generally have to first in your normalization file.
448    
449  =head2 tag  =head2 search_display
450    
451  Define new tag for I<search> and I<display>.  Define output for L<search> and L<display> at the same time
452    
453    tag('Title', rec('200','a') );    search_display('Title', rec('200','a') );
454    
455    
456  =cut  =cut
457    
458  sub tag {  sub search_display {
459          my $name = shift or die "tag needs name as first argument";          my $name = shift or die "search_display needs name as first argument";
460          my @o = grep { defined($_) && $_ ne '' } @_;          my @o = grep { defined($_) && $_ ne '' } @_;
461          return unless (@o);          return unless (@o);
         $out->{$name}->{tag} = $name;  
462          $out->{$name}->{search} = \@o;          $out->{$name}->{search} = \@o;
463          $out->{$name}->{display} = \@o;          $out->{$name}->{display} = \@o;
464  }  }
465    
466    =head2 tag
467    
468    Old name for L<search_display>, but supported
469    
470    =cut
471    
472    sub tag {
473            search_display( @_ );
474    }
475    
476  =head2 display  =head2 display
477    
478  Define tag just for I<display>  Define output just for I<display>
479    
480    @v = display('Title', rec('200','a') );    @v = display('Title', rec('200','a') );
481    
482  =cut  =cut
483    
484  sub display {  sub _field {
485          my $name = shift or die "display needs name as first argument";          my $type = shift or confess "need type -- BUG?";
486            my $name = shift or confess "needs name as first argument";
487          my @o = grep { defined($_) && $_ ne '' } @_;          my @o = grep { defined($_) && $_ ne '' } @_;
488          return unless (@o);          return unless (@o);
489          $out->{$name}->{tag} = $name;          $out->{$name}->{$type} = \@o;
         $out->{$name}->{display} = \@o;  
490  }  }
491    
492    sub display { _field( 'display', @_ ) }
493    
494  =head2 search  =head2 search
495    
496  Prepare values just for I<search>  Prepare values just for I<search>
# Line 468  Prepare values just for I<search> Line 499  Prepare values just for I<search>
499    
500  =cut  =cut
501    
502  sub search {  sub search { _field( 'search', @_ ) }
503          my $name = shift or die "search needs name as first argument";  
504          my @o = grep { defined($_) && $_ ne '' } @_;  =head2 sorted
505          return unless (@o);  
506          $out->{$name}->{tag} = $name;  Insert into lists which will be automatically sorted
507          $out->{$name}->{search} = \@o;  
508  }   sorted('Title', rec('200','a') );
509    
510    =cut
511    
512    sub sorted { _field( 'sorted', @_ ) }
513    
514    
515  =head2 marc_leader  =head2 marc_leader
516    
# Line 489  sub marc_leader { Line 525  sub marc_leader {
525          my ($offset,$value) = @_;          my ($offset,$value) = @_;
526    
527          if ($offset) {          if ($offset) {
528                  $leader->{ $offset } = $value;                  $marc_leader->[ $marc_record_offset ]->{ $offset } = $value;
529          } else {          } else {
530                  return $leader;                  
531                    if (defined($marc_leader)) {
532                            die "marc_leader not array = ", dump( $marc_leader ) unless (ref($marc_leader) eq 'ARRAY');
533                            return $marc_leader->[ $marc_record_offset ];
534                    } else {
535                            return;
536                    }
537            }
538    }
539    
540    =head2 marc_fixed
541    
542    Create control/indentifier fields with values in fixed positions
543    
544      marc_fixed('008', 00, '070402');
545      marc_fixed('008', 39, '|');
546    
547    Positions not specified will be filled with spaces (C<0x20>).
548    
549    There will be no effort to extend last specified value to full length of
550    field in standard.
551    
552    =cut
553    
554    sub marc_fixed {
555            my ($f, $pos, $val) = @_;
556            die "need marc(field, position, value)" unless defined($f) && defined($pos);
557    
558            confess "need val" unless defined $val;
559    
560            my $update = 0;
561    
562            map {
563                    if ($_->[0] eq $f) {
564                            my $old = $_->[1];
565                            if (length($old) <= $pos) {
566                                    $_->[1] .= ' ' x ( $pos - length($old) ) . $val;
567                                    warn "## marc_fixed($f,$pos,'$val') append '$old' -> '$_->[1]'\n" if ($debug > 1);
568                            } else {
569                                    $_->[1] = substr($old, 0, $pos) . $val . substr($old, $pos + length($val));
570                                    warn "## marc_fixed($f,$pos,'$val') update '$old' -> '$_->[1]'\n" if ($debug > 1);
571                            }
572                            $update++;
573                    }
574            } @{ $marc_record->[ $marc_record_offset ] };
575    
576            if (! $update) {
577                    my $v = ' ' x $pos . $val;
578                    push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $v ];
579                    warn "## marc_fixed($f,$pos,'val') created '$v'\n" if ($debug > 1);
580          }          }
581  }  }
582    
# Line 623  sub marc_duplicate { Line 708  sub marc_duplicate {
708           my $m = $marc_record->[ -1 ];           my $m = $marc_record->[ -1 ];
709           die "can't duplicate record which isn't defined" unless ($m);           die "can't duplicate record which isn't defined" unless ($m);
710           push @{ $marc_record }, dclone( $m );           push @{ $marc_record }, dclone( $m );
711           warn "## marc_duplicate = ", dump(@$marc_record), $/ if ($debug > 1);           push @{ $marc_leader }, dclone( marc_leader() );
712             warn "## marc_duplicate = ", dump(@$marc_leader, @$marc_record), $/ if ($debug > 1);
713           $marc_record_offset = $#{ $marc_record };           $marc_record_offset = $#{ $marc_record };
714           warn "## marc_record_offset = $marc_record_offset", $/ if ($debug > 1);           warn "## marc_record_offset = $marc_record_offset", $/ if ($debug > 1);
715    
716  }  }
717    
718  =head2 marc_remove  =head2 marc_remove
# Line 773  sub marc_original_order { Line 860  sub marc_original_order {
860          warn "## marc_record = ", dump( $marc_record ),$/ if ($debug > 1);          warn "## marc_record = ", dump( $marc_record ),$/ if ($debug > 1);
861  }  }
862    
863    =head2 marc_count
864    
865    Return number of MARC records created using L</marc_duplicate>.
866    
867      print "created ", marc_count(), " records";
868    
869    =cut
870    
871    sub marc_count {
872            return $#{ $marc_record };
873    }
874    
875    
876  =head1 Functions to extract data from input  =head1 Functions to extract data from input
877    
# Line 795  sub _pack_subfields_hash { Line 894  sub _pack_subfields_hash {
894    
895          my ($h,$include_subfields) = @_;          my ($h,$include_subfields) = @_;
896    
897            # sanity and ease of use
898            return $h if (ref($h) ne 'HASH');
899    
900          if ( defined($h->{subfields}) ) {          if ( defined($h->{subfields}) ) {
901                  my $sfs = delete $h->{subfields} || die "no subfields?";                  my $sfs = delete $h->{subfields} || die "no subfields?";
902                  my @out;                  my @out;
# Line 899  return scalar, not array. Line 1001  return scalar, not array.
1001    
1002  =cut  =cut
1003    
1004    sub frec {
1005            my @out = rec(@_);
1006            warn "rec(",dump(@_),") has more than one return value, ignoring\n" if $#out > 0;
1007            return shift @out;
1008    }
1009    
1010  sub rec {  sub rec {
1011          my @out;          my @out;
1012          if ($#_ == 0) {          if ($#_ == 0) {
# Line 944  Prefix all values with a string Line 1052  Prefix all values with a string
1052  =cut  =cut
1053    
1054  sub prefix {  sub prefix {
1055          my $p = shift or return;          my $p = shift;
1056            return @_ unless defined( $p );
1057          return map { $p . $_ } grep { defined($_) } @_;          return map { $p . $_ } grep { defined($_) } @_;
1058  }  }
1059    
# Line 957  suffix all values with a string Line 1066  suffix all values with a string
1066  =cut  =cut
1067    
1068  sub suffix {  sub suffix {
1069          my $s = shift or die "suffix needs string as first argument";          my $s = shift;
1070            return @_ unless defined( $s );
1071          return map { $_ . $s } grep { defined($_) } @_;          return map { $_ . $s } grep { defined($_) } @_;
1072  }  }
1073    
# Line 970  surround all values with a two strings Line 1080  surround all values with a two strings
1080  =cut  =cut
1081    
1082  sub surround {  sub surround {
1083          my $p = shift or die "surround need prefix as first argument";          my $p = shift;
1084          my $s = shift or die "surround needs suffix as second argument";          my $s = shift;
1085            $p = '' unless defined( $p );
1086            $s = '' unless defined( $s );
1087          return map { $p . $_ . $s } grep { defined($_) } @_;          return map { $p . $_ . $s } grep { defined($_) } @_;
1088  }  }
1089    
# Line 1146  Consult config values stored in C<config Line 1258  Consult config values stored in C<config
1258    $database_code = config();    # use _ from hash    $database_code = config();    # use _ from hash
1259    $database_name = config('name');    $database_name = config('name');
1260    $database_input_name = config('input name');    $database_input_name = config('input name');
   $tag = config('input normalize tag');  
1261    
1262  Up to three levels are supported.  Up to three levels are supported.
1263    
# Line 1272  my $hash; Line 1383  my $hash;
1383    
1384  sub set {  sub set {
1385          my ($k,$v) = @_;          my ($k,$v) = @_;
1386          warn "## set ( $k => ", dump($v), " )", $/;          warn "## set ( $k => ", dump($v), " )", $/ if ( $debug );
1387          $hash->{$k} = $v;          $hash->{$k} = $v;
1388  };  };
1389    
# Line 1285  sub set { Line 1396  sub set {
1396  sub get {  sub get {
1397          my $k = shift || return;          my $k = shift || return;
1398          my $v = $hash->{$k};          my $v = $hash->{$k};
1399          warn "## get $k = ", dump( $v ), $/;          warn "## get $k = ", dump( $v ), $/ if ( $debug );
1400          return $v;          return $v;
1401  }  }
1402    
1403    =head2 count
1404    
1405      if ( count( @result ) == 1 ) {
1406            # do something if only 1 result is there
1407      }
1408    
1409    =cut
1410    
1411    sub count {
1412            warn "## count ",dump(@_),$/ if ( $debug );
1413            return @_ . '';
1414    }
1415    
1416  # END  # END
1417  1;  1;

Legend:
Removed from v.788  
changed lines
  Added in v.979

  ViewVC Help
Powered by ViewVC 1.1.26