/[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

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

Parent Directory Parent Directory | Revision Log Revision Log


Revision 912 - (hide annotations)
Tue Oct 30 17:40:13 2007 UTC (16 years, 5 months ago) by dpavlin
File size: 30446 byte(s)
 r1370@llin:  dpavlin | 2007-10-30 18:40:02 +0100
 rename tag to search_display which makes it more clean what does it do

1 dpavlin 10 package WebPAC::Normalize;
2 dpavlin 536 use Exporter 'import';
3     @EXPORT = qw/
4 dpavlin 538 _set_rec _set_lookup
5 dpavlin 736 _set_load_row
6 dpavlin 538 _get_ds _clean_ds
7 dpavlin 554 _debug
8 dpavlin 641 _pack_subfields_hash
9 dpavlin 538
10 dpavlin 912 search_display search display
11    
12 dpavlin 547 marc marc_indicators marc_repeatable_subfield
13 dpavlin 815 marc_compose marc_leader marc_fixed
14 dpavlin 813 marc_duplicate marc_remove marc_count
15 dpavlin 604 marc_original_order
16 dpavlin 540
17 dpavlin 536 rec1 rec2 rec
18     regex prefix suffix surround
19     first lookup join_with
20 dpavlin 707 save_into_lookup
21 dpavlin 562
22     split_rec_on
23 dpavlin 785
24     get set
25 dpavlin 791 count
26 dpavlin 536 /;
27 dpavlin 10
28     use warnings;
29     use strict;
30 dpavlin 536
31     #use base qw/WebPAC::Common/;
32 dpavlin 550 use Data::Dump qw/dump/;
33 dpavlin 574 use Storable qw/dclone/;
34 dpavlin 725 use Carp qw/confess/;
35 dpavlin 10
36 dpavlin 550 # debugging warn(s)
37     my $debug = 0;
38    
39    
40 dpavlin 10 =head1 NAME
41    
42 dpavlin 536 WebPAC::Normalize - describe normalisaton rules using sets
43 dpavlin 10
44     =head1 VERSION
45    
46 dpavlin 912 Version 0.30
47 dpavlin 10
48     =cut
49    
50 dpavlin 912 our $VERSION = '0.30';
51 dpavlin 10
52     =head1 SYNOPSIS
53    
54 dpavlin 536 This module uses C<conf/normalize/*.pl> files to perform normalisation
55     from input records using perl functions which are specialized for set
56     processing.
57 dpavlin 10
58 dpavlin 536 Sets are implemented as arrays, and normalisation file is valid perl, which
59     means that you check it's validity before running WebPAC using
60     C<perl -c normalize.pl>.
61 dpavlin 15
62 dpavlin 536 Normalisation can generate multiple output normalized data. For now, supported output
63 dpavlin 912 types (on the left side of definition) are: C<search_display>, C<display>, C<search> and
64 dpavlin 547 C<marc>.
65 dpavlin 15
66 dpavlin 10 =head1 FUNCTIONS
67    
68 dpavlin 538 Functions which start with C<_> are private and used by WebPAC internally.
69     All other functions are available for use within normalisation rules.
70    
71 dpavlin 536 =head2 data_structure
72 dpavlin 10
73 dpavlin 536 Return data structure
74 dpavlin 13
75 dpavlin 538 my $ds = WebPAC::Normalize::data_structure(
76 dpavlin 725 lookup => $lookup_hash,
77 dpavlin 536 row => $row,
78     rules => $normalize_pl_config,
79 dpavlin 541 marc_encoding => 'utf-8',
80 dpavlin 595 config => $config,
81 dpavlin 736 load_row_coderef => sub {
82 dpavlin 725 my ($database,$input,$mfn) = shift;
83 dpavlin 736 $store->load_row( database => $database, input => $input, id => $mfn );
84 dpavlin 725 },
85 dpavlin 13 );
86    
87 dpavlin 707 Options C<row>, C<rules> and C<log> are mandatory while all
88 dpavlin 540 other are optional.
89    
90 dpavlin 736 C<load_row_coderef> is closure only used when executing lookups, so they will
91 dpavlin 725 die if it's not defined.
92    
93 dpavlin 536 This function will B<die> if normalizastion can't be evaled.
94 dpavlin 15
95 dpavlin 538 Since this function isn't exported you have to call it with
96     C<WebPAC::Normalize::data_structure>.
97    
98 dpavlin 10 =cut
99    
100 dpavlin 736 my $load_row_coderef;
101 dpavlin 725
102 dpavlin 536 sub data_structure {
103     my $arg = {@_};
104 dpavlin 13
105 dpavlin 536 die "need row argument" unless ($arg->{row});
106     die "need normalisation argument" unless ($arg->{rules});
107 dpavlin 31
108 dpavlin 536 no strict 'subs';
109 dpavlin 730 _set_lookup( $arg->{lookup} ) if defined($arg->{lookup});
110 dpavlin 538 _set_rec( $arg->{row} );
111 dpavlin 730 _set_config( $arg->{config} ) if defined($arg->{config});
112 dpavlin 541 _clean_ds( %{ $arg } );
113 dpavlin 736 $load_row_coderef = $arg->{load_row_coderef};
114 dpavlin 725
115 dpavlin 536 eval "$arg->{rules}";
116     die "error evaling $arg->{rules}: $@\n" if ($@);
117 dpavlin 540
118 dpavlin 538 return _get_ds();
119 dpavlin 10 }
120    
121 dpavlin 538 =head2 _set_rec
122 dpavlin 13
123 dpavlin 536 Set current record hash
124 dpavlin 433
125 dpavlin 538 _set_rec( $rec );
126 dpavlin 433
127     =cut
128    
129 dpavlin 536 my $rec;
130 dpavlin 433
131 dpavlin 538 sub _set_rec {
132 dpavlin 536 $rec = shift or die "no record hash";
133 dpavlin 433 }
134    
135 dpavlin 595 =head2 _set_config
136    
137     Set current config hash
138    
139     _set_config( $config );
140    
141     Magic keys are:
142    
143     =over 4
144    
145     =item _
146    
147     Code of current database
148    
149     =item _mfn
150    
151     Current MFN
152    
153     =back
154    
155     =cut
156    
157     my $config;
158    
159     sub _set_config {
160     $config = shift;
161     }
162    
163 dpavlin 538 =head2 _get_ds
164    
165     Return hash formatted as data structure
166    
167     my $ds = _get_ds();
168    
169     =cut
170    
171 dpavlin 812 my ($out, $marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators, $marc_leader);
172 dpavlin 574 my ($marc_record_offset, $marc_fetch_offset) = (0, 0);
173 dpavlin 538
174     sub _get_ds {
175     return $out;
176     }
177    
178     =head2 _clean_ds
179    
180     Clean data structure hash for next record
181    
182     _clean_ds();
183    
184     =cut
185    
186     sub _clean_ds {
187 dpavlin 541 my $a = {@_};
188 dpavlin 812 ($out,$marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators, $marc_leader) = ();
189 dpavlin 574 ($marc_record_offset, $marc_fetch_offset) = (0,0);
190 dpavlin 541 $marc_encoding = $a->{marc_encoding};
191 dpavlin 538 }
192    
193     =head2 _set_lookup
194    
195     Set current lookup hash
196    
197     _set_lookup( $lookup );
198    
199     =cut
200    
201     my $lookup;
202    
203     sub _set_lookup {
204     $lookup = shift;
205     }
206    
207 dpavlin 707 =head2 _get_lookup
208    
209     Get current lookup hash
210    
211     my $lookup = _get_lookup();
212    
213     =cut
214    
215     sub _get_lookup {
216     return $lookup;
217     }
218    
219 dpavlin 736 =head2 _set_load_row
220 dpavlin 725
221     Setup code reference which will return L<data_structure> from
222     L<WebPAC::Store>
223    
224 dpavlin 736 _set_load_row(sub {
225 dpavlin 725 my ($database,$input,$mfn) = @_;
226 dpavlin 736 $store->load_row( database => $database, input => $input, id => $mfn );
227 dpavlin 725 });
228    
229     =cut
230    
231 dpavlin 736 sub _set_load_row {
232 dpavlin 725 my $coderef = shift;
233     confess "argument isn't CODE" unless ref($coderef) eq 'CODE';
234    
235 dpavlin 736 $load_row_coderef = $coderef;
236 dpavlin 725 }
237    
238 dpavlin 547 =head2 _get_marc_fields
239 dpavlin 540
240 dpavlin 547 Get all fields defined by calls to C<marc>
241 dpavlin 540
242 dpavlin 547 $marc->add_fields( WebPAC::Normalize:_get_marc_fields() );
243 dpavlin 540
244 dpavlin 543 We are using I<magic> which detect repeatable fields only from
245     sequence of field/subfield data generated by normalization.
246    
247 dpavlin 554 Repeatable field is created when there is second occurence of same subfield or
248     if any of indicators are different.
249 dpavlin 543
250 dpavlin 554 This is sane for most cases. Something like:
251 dpavlin 543
252 dpavlin 554 900a-1 900b-1 900c-1
253     900a-2 900b-2
254     900a-3
255    
256     will be created from any combination of:
257    
258     900a-1 900a-2 900a-3 900b-1 900b-2 900c-1
259    
260     and following rules:
261    
262     marc('900','a', rec('200','a') );
263     marc('900','b', rec('200','b') );
264     marc('900','c', rec('200','c') );
265    
266     which might not be what you have in mind. If you need repeatable subfield,
267     define it using C<marc_repeatable_subfield> like this:
268    
269 dpavlin 574 marc_repeatable_subfield('900','a');
270     marc('900','a', rec('200','a') );
271     marc('900','b', rec('200','b') );
272     marc('900','c', rec('200','c') );
273 dpavlin 554
274 dpavlin 574 will create:
275    
276     900a-1 900a-2 900a-3 900b-1 900c-1
277     900b-2
278    
279     There is also support for returning next or specific using:
280    
281     while (my $mf = WebPAC::Normalize:_get_marc_fields( fetch_next => 1 ) ) {
282     # do something with $mf
283     }
284    
285     will always return fields from next MARC record or
286    
287     my $mf = WebPAC::Normalize::_get_marc_fields( offset => 42 );
288    
289     will return 42th copy record (if it exists).
290    
291 dpavlin 540 =cut
292    
293 dpavlin 813 my $fetch_pos;
294    
295 dpavlin 547 sub _get_marc_fields {
296 dpavlin 550
297 dpavlin 574 my $arg = {@_};
298     warn "### _get_marc_fields arg: ", dump($arg), $/ if ($debug > 2);
299 dpavlin 813 $fetch_pos = $marc_fetch_offset;
300 dpavlin 574 if ($arg->{offset}) {
301 dpavlin 813 $fetch_pos = $arg->{offset};
302 dpavlin 574 } elsif($arg->{fetch_next}) {
303     $marc_fetch_offset++;
304     }
305 dpavlin 550
306 dpavlin 574 return if (! $marc_record || ref($marc_record) ne 'ARRAY');
307    
308     warn "### full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 2);
309    
310 dpavlin 813 my $marc_rec = $marc_record->[ $fetch_pos ];
311 dpavlin 574
312 dpavlin 813 warn "## _get_marc_fields (at offset: $fetch_pos) -- marc_record = ", dump( @$marc_rec ), $/ if ($debug > 1);
313 dpavlin 574
314     return if (! $marc_rec || ref($marc_rec) ne 'ARRAY' || $#{ $marc_rec } < 0);
315    
316 dpavlin 550 # first, sort all existing fields
317     # XXX might not be needed, but modern perl might randomize elements in hash
318     my @sorted_marc_record = sort {
319 dpavlin 572 $a->[0] . ( $a->[3] || '' ) cmp $b->[0] . ( $b->[3] || '')
320 dpavlin 574 } @{ $marc_rec };
321 dpavlin 550
322 dpavlin 574 @sorted_marc_record = @{ $marc_rec }; ### FIXME disable sorting
323 dpavlin 562
324 dpavlin 550 # output marc fields
325 dpavlin 542 my @m;
326 dpavlin 550
327     # count unique field-subfields (used for offset when walking to next subfield)
328     my $u;
329 dpavlin 572 map { $u->{ $_->[0] . ( $_->[3] || '') }++ } @sorted_marc_record;
330 dpavlin 550
331     if ($debug) {
332 dpavlin 574 warn "## marc_repeatable_subfield = ", dump( $marc_repeatable_subfield ), $/ if ( $marc_repeatable_subfield );
333 dpavlin 813 warn "## marc_record[$fetch_pos] = ", dump( $marc_rec ), $/;
334 dpavlin 574 warn "## sorted_marc_record = ", dump( \@sorted_marc_record ), $/;
335     warn "## subfield count = ", dump( $u ), $/;
336 dpavlin 550 }
337    
338     my $len = $#sorted_marc_record;
339     my $visited;
340     my $i = 0;
341     my $field;
342    
343     foreach ( 0 .. $len ) {
344    
345     # find next element which isn't visited
346     while ($visited->{$i}) {
347     $i = ($i + 1) % ($len + 1);
348 dpavlin 542 }
349    
350 dpavlin 550 # mark it visited
351     $visited->{$i}++;
352    
353 dpavlin 574 my $row = dclone( $sorted_marc_record[$i] );
354 dpavlin 550
355     # field and subfield which is key for
356     # marc_repeatable_subfield and u
357 dpavlin 572 my $fsf = $row->[0] . ( $row->[3] || '' );
358 dpavlin 550
359     if ($debug > 1) {
360    
361     print "### field so far [", $#$field, "] : ", dump( $field ), " ", $field ? 'T' : 'F', $/;
362     print "### this [$i]: ", dump( $row ),$/;
363     print "### sf: ", $row->[3], " vs ", $field->[3],
364     $marc_repeatable_subfield->{ $row->[0] . $row->[3] } ? ' (repeatable)' : '', $/,
365     if ($#$field >= 0);
366    
367     }
368    
369     # if field exists
370     if ( $#$field >= 0 ) {
371     if (
372     $row->[0] ne $field->[0] || # field
373     $row->[1] ne $field->[1] || # i1
374     $row->[2] ne $field->[2] # i2
375     ) {
376     push @m, $field;
377     warn "## saved/1 ", dump( $field ),$/ if ($debug);
378     $field = $row;
379    
380     } elsif (
381     ( $row->[3] lt $field->[-2] ) # subfield which is not next (e.g. a after c)
382     ||
383     ( $row->[3] eq $field->[-2] && # same subfield, but not repeatable
384     ! $marc_repeatable_subfield->{ $fsf }
385     )
386     ) {
387     push @m, $field;
388     warn "## saved/2 ", dump( $field ),$/ if ($debug);
389     $field = $row;
390    
391     } else {
392     # append new subfields to existing field
393     push @$field, ( $row->[3], $row->[4] );
394     }
395     } else {
396     # insert first field
397     $field = $row;
398     }
399    
400     if (! $marc_repeatable_subfield->{ $fsf }) {
401     # make step to next subfield
402     $i = ($i + $u->{ $fsf } ) % ($len + 1);
403     }
404 dpavlin 542 }
405    
406 dpavlin 550 if ($#$field >= 0) {
407     push @m, $field;
408     warn "## saved/3 ", dump( $field ),$/ if ($debug);
409     }
410 dpavlin 542
411 dpavlin 579 return \@m;
412 dpavlin 540 }
413    
414 dpavlin 813 =head2 _get_marc_leader
415    
416     Return leader from currently fetched record by L</_get_marc_fields>
417    
418     print WebPAC::Normalize::_get_marc_leader();
419    
420     =cut
421    
422     sub _get_marc_leader {
423     die "no fetch_pos, did you called _get_marc_fields first?" unless ( defined( $fetch_pos ) );
424     return $marc_leader->[ $fetch_pos ];
425     }
426    
427 dpavlin 554 =head2 _debug
428    
429     Change level of debug warnings
430    
431     _debug( 2 );
432    
433     =cut
434    
435     sub _debug {
436     my $l = shift;
437     return $debug unless defined($l);
438 dpavlin 568 warn "debug level $l",$/ if ($l > 0);
439 dpavlin 554 $debug = $l;
440     }
441    
442 dpavlin 540 =head1 Functions to create C<data_structure>
443    
444     Those functions generally have to first in your normalization file.
445    
446 dpavlin 912 =head2 search_display
447 dpavlin 433
448 dpavlin 912 Define output for L<search> and L<display> at the same time
449 dpavlin 433
450 dpavlin 912 search_display('Title', rec('200','a') );
451 dpavlin 13
452    
453     =cut
454    
455 dpavlin 912 sub search_display {
456     my $name = shift or die "search_display needs name as first argument";
457 dpavlin 536 my @o = grep { defined($_) && $_ ne '' } @_;
458     return unless (@o);
459     $out->{$name}->{search} = \@o;
460     $out->{$name}->{display} = \@o;
461     }
462 dpavlin 13
463 dpavlin 536 =head2 display
464 dpavlin 13
465 dpavlin 912 Define output just for I<display>
466 dpavlin 125
467 dpavlin 536 @v = display('Title', rec('200','a') );
468 dpavlin 125
469 dpavlin 536 =cut
470 dpavlin 125
471 dpavlin 536 sub display {
472     my $name = shift or die "display needs name as first argument";
473     my @o = grep { defined($_) && $_ ne '' } @_;
474     return unless (@o);
475     $out->{$name}->{display} = \@o;
476     }
477 dpavlin 13
478 dpavlin 536 =head2 search
479 dpavlin 13
480 dpavlin 536 Prepare values just for I<search>
481 dpavlin 13
482 dpavlin 536 @v = search('Title', rec('200','a') );
483 dpavlin 433
484 dpavlin 536 =cut
485 dpavlin 13
486 dpavlin 536 sub search {
487     my $name = shift or die "search needs name as first argument";
488     my @o = grep { defined($_) && $_ ne '' } @_;
489     return unless (@o);
490     $out->{$name}->{search} = \@o;
491 dpavlin 13 }
492    
493 dpavlin 564 =head2 marc_leader
494    
495     Setup fields within MARC leader or get leader
496    
497     marc_leader('05','c');
498     my $leader = marc_leader();
499    
500     =cut
501    
502     sub marc_leader {
503     my ($offset,$value) = @_;
504    
505     if ($offset) {
506 dpavlin 813 $marc_leader->[ $marc_record_offset ]->{ $offset } = $value;
507 dpavlin 564 } else {
508 dpavlin 813
509     if (defined($marc_leader)) {
510     die "marc_leader not array = ", dump( $marc_leader ) unless (ref($marc_leader) eq 'ARRAY');
511     return $marc_leader->[ $marc_record_offset ];
512     } else {
513     return;
514     }
515 dpavlin 564 }
516     }
517    
518 dpavlin 815 =head2 marc_fixed
519    
520     Create control/indentifier fields with values in fixed positions
521    
522     marc_fixed('008', 00, '070402');
523     marc_fixed('008', 39, '|');
524    
525     Positions not specified will be filled with spaces (C<0x20>).
526    
527     There will be no effort to extend last specified value to full length of
528     field in standard.
529    
530     =cut
531    
532     sub marc_fixed {
533     my ($f, $pos, $val) = @_;
534     die "need marc(field, position, value)" unless defined($f) && defined($pos);
535    
536 dpavlin 889 confess "need val" unless defined $val;
537    
538 dpavlin 815 my $update = 0;
539    
540     map {
541     if ($_->[0] eq $f) {
542     my $old = $_->[1];
543 dpavlin 889 if (length($old) <= $pos) {
544 dpavlin 815 $_->[1] .= ' ' x ( $pos - length($old) ) . $val;
545 dpavlin 817 warn "## marc_fixed($f,$pos,'$val') append '$old' -> '$_->[1]'\n" if ($debug > 1);
546 dpavlin 815 } else {
547     $_->[1] = substr($old, 0, $pos) . $val . substr($old, $pos + length($val));
548 dpavlin 817 warn "## marc_fixed($f,$pos,'$val') update '$old' -> '$_->[1]'\n" if ($debug > 1);
549 dpavlin 815 }
550     $update++;
551     }
552     } @{ $marc_record->[ $marc_record_offset ] };
553    
554     if (! $update) {
555     my $v = ' ' x $pos . $val;
556     push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $v ];
557 dpavlin 817 warn "## marc_fixed($f,$pos,'val') created '$v'\n" if ($debug > 1);
558 dpavlin 815 }
559     }
560    
561 dpavlin 547 =head2 marc
562 dpavlin 540
563     Save value for MARC field
564    
565 dpavlin 547 marc('900','a', rec('200','a') );
566 dpavlin 571 marc('001', rec('000') );
567 dpavlin 540
568     =cut
569    
570 dpavlin 547 sub marc {
571     my $f = shift or die "marc needs field";
572     die "marc field must be numer" unless ($f =~ /^\d+$/);
573 dpavlin 540
574 dpavlin 571 my $sf;
575     if ($f >= 10) {
576     $sf = shift or die "marc needs subfield";
577     }
578 dpavlin 540
579 dpavlin 541 foreach (@_) {
580     my $v = $_; # make var read-write for Encode
581 dpavlin 543 next unless (defined($v) && $v !~ /^\s*$/);
582 dpavlin 548 my ($i1,$i2) = defined($marc_indicators->{$f}) ? @{ $marc_indicators->{$f} } : (' ',' ');
583 dpavlin 571 if (defined $sf) {
584 dpavlin 574 push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $i1, $i2, $sf => $v ];
585 dpavlin 571 } else {
586 dpavlin 574 push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $v ];
587 dpavlin 571 }
588 dpavlin 540 }
589     }
590    
591 dpavlin 547 =head2 marc_repeatable_subfield
592    
593     Save values for MARC repetable subfield
594    
595     marc_repeatable_subfield('910', 'z', rec('909') );
596    
597     =cut
598    
599     sub marc_repeatable_subfield {
600 dpavlin 550 my ($f,$sf) = @_;
601     die "marc_repeatable_subfield need field and subfield!\n" unless ($f && $sf);
602     $marc_repeatable_subfield->{ $f . $sf }++;
603 dpavlin 547 marc(@_);
604     }
605    
606     =head2 marc_indicators
607    
608     Set both indicators for MARC field
609    
610     marc_indicators('900', ' ', 1);
611    
612     Any indicator value other than C<0-9> will be treated as undefined.
613    
614     =cut
615    
616     sub marc_indicators {
617     my $f = shift || die "marc_indicators need field!\n";
618     my ($i1,$i2) = @_;
619     die "marc_indicators($f, ...) need i1!\n" unless(defined($i1));
620     die "marc_indicators($f, $i1, ...) need i2!\n" unless(defined($i2));
621    
622     $i1 = ' ' if ($i1 !~ /^\d$/);
623     $i2 = ' ' if ($i2 !~ /^\d$/);
624 dpavlin 548 @{ $marc_indicators->{$f} } = ($i1,$i2);
625 dpavlin 547 }
626    
627 dpavlin 562 =head2 marc_compose
628 dpavlin 547
629 dpavlin 562 Save values for each MARC subfield explicitly
630    
631     marc_compose('900',
632     'a', rec('200','a')
633     'b', rec('201','a')
634     'a', rec('200','b')
635     'c', rec('200','c')
636     );
637    
638 dpavlin 603 If you specify C<+> for subfield, value will be appended
639     to previous defined subfield.
640    
641 dpavlin 562 =cut
642    
643     sub marc_compose {
644     my $f = shift or die "marc_compose needs field";
645     die "marc_compose field must be numer" unless ($f =~ /^\d+$/);
646    
647     my ($i1,$i2) = defined($marc_indicators->{$f}) ? @{ $marc_indicators->{$f} } : (' ',' ');
648     my $m = [ $f, $i1, $i2 ];
649    
650 dpavlin 583 warn "### marc_compose input subfields = ", dump(@_),$/ if ($debug > 2);
651    
652 dpavlin 619 if ($#_ % 2 != 1) {
653     die "ERROR: marc_compose",dump($f,@_)," not valid (must be even).\nDo you need to add first() or join() around some argument?\n";
654     }
655    
656 dpavlin 562 while (@_) {
657 dpavlin 619 my $sf = shift;
658 dpavlin 565 my $v = shift;
659 dpavlin 562
660     next unless (defined($v) && $v !~ /^\s*$/);
661 dpavlin 568 warn "## ++ marc_compose($f,$sf,$v) ", dump( $m ),$/ if ($debug > 1);
662 dpavlin 603 if ($sf ne '+') {
663     push @$m, ( $sf, $v );
664     } else {
665     $m->[ $#$m ] .= $v;
666     }
667 dpavlin 562 }
668    
669 dpavlin 586 warn "## marc_compose current marc = ", dump( $m ),$/ if ($debug > 1);
670 dpavlin 562
671 dpavlin 574 push @{ $marc_record->[ $marc_record_offset ] }, $m if ($#{$m} > 2);
672 dpavlin 562 }
673    
674 dpavlin 574 =head2 marc_duplicate
675 dpavlin 562
676 dpavlin 574 Generate copy of current MARC record and continue working on copy
677    
678     marc_duplicate();
679    
680     Copies can be accessed using C<< _get_marc_fields( fetch_next => 1 ) >> or
681     C<< _get_marc_fields( offset => 42 ) >>.
682    
683     =cut
684    
685     sub marc_duplicate {
686     my $m = $marc_record->[ -1 ];
687     die "can't duplicate record which isn't defined" unless ($m);
688     push @{ $marc_record }, dclone( $m );
689 dpavlin 813 push @{ $marc_leader }, dclone( marc_leader() );
690     warn "## marc_duplicate = ", dump(@$marc_leader, @$marc_record), $/ if ($debug > 1);
691 dpavlin 574 $marc_record_offset = $#{ $marc_record };
692     warn "## marc_record_offset = $marc_record_offset", $/ if ($debug > 1);
693 dpavlin 813
694 dpavlin 574 }
695    
696     =head2 marc_remove
697    
698     Remove some field or subfield from MARC record.
699    
700     marc_remove('200');
701     marc_remove('200','a');
702    
703     This will erase field C<200> or C<200^a> from current MARC record.
704    
705 dpavlin 786 marc_remove('*');
706    
707     Will remove all fields in current MARC record.
708    
709 dpavlin 574 This is useful after calling C<marc_duplicate> or on it's own (but, you
710     should probably just remove that subfield definition if you are not
711     using C<marc_duplicate>).
712    
713     FIXME: support fields < 10.
714    
715     =cut
716    
717     sub marc_remove {
718     my ($f, $sf) = @_;
719    
720     die "marc_remove needs record number" unless defined($f);
721    
722     my $marc = $marc_record->[ $marc_record_offset ];
723    
724     warn "### marc_remove before = ", dump( $marc ), $/ if ($debug > 2);
725    
726 dpavlin 786 if ($f eq '*') {
727    
728     delete( $marc_record->[ $marc_record_offset ] );
729 dpavlin 788 warn "## full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 1);
730 dpavlin 786
731     } else {
732    
733     my $i = 0;
734     foreach ( 0 .. $#{ $marc } ) {
735     last unless (defined $marc->[$i]);
736     warn "#### working on ",dump( @{ $marc->[$i] }), $/ if ($debug > 3);
737     if ($marc->[$i]->[0] eq $f) {
738     if (! defined $sf) {
739     # remove whole field
740     splice @$marc, $i, 1;
741     warn "#### slice \@\$marc, $i, 1 = ",dump( @{ $marc }), $/ if ($debug > 3);
742     $i--;
743     } else {
744     foreach my $j ( 0 .. (( $#{ $marc->[$i] } - 3 ) / 2) ) {
745     my $o = ($j * 2) + 3;
746     if ($marc->[$i]->[$o] eq $sf) {
747     # remove subfield
748     splice @{$marc->[$i]}, $o, 2;
749     warn "#### slice \@{\$marc->[$i]}, $o, 2 = ", dump( @{ $marc }), $/ if ($debug > 3);
750     # is record now empty?
751     if ($#{ $marc->[$i] } == 2) {
752     splice @$marc, $i, 1;
753     warn "#### slice \@\$marc, $i, 1 = ", dump( @{ $marc }), $/ if ($debug > 3);
754     $i--;
755     };
756     }
757 dpavlin 574 }
758     }
759     }
760 dpavlin 786 $i++;
761 dpavlin 574 }
762 dpavlin 786
763     warn "### marc_remove($f", $sf ? ",$sf" : "", ") after = ", dump( $marc ), $/ if ($debug > 2);
764    
765     $marc_record->[ $marc_record_offset ] = $marc;
766 dpavlin 574 }
767    
768     warn "## full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 1);
769     }
770    
771 dpavlin 604 =head2 marc_original_order
772    
773     Copy all subfields preserving original order to marc field.
774    
775 dpavlin 616 marc_original_order( marc_field_number, original_input_field_number );
776 dpavlin 604
777 dpavlin 616 Please note that field numbers are consistent with other commands (marc
778     field number first), but somewhat counter-intuitive (destination and then
779     source).
780    
781 dpavlin 604 You might want to use this command if you are just renaming subfields or
782     using pre-processing modify_record in C<config.yml> and don't need any
783     post-processing or want to preserve order of original subfields.
784    
785 dpavlin 616
786 dpavlin 604 =cut
787    
788     sub marc_original_order {
789    
790 dpavlin 616 my ($to, $from) = @_;
791 dpavlin 604 die "marc_original_order needs from and to fields\n" unless ($from && $to);
792    
793 dpavlin 616 return unless defined($rec->{$from});
794    
795     my $r = $rec->{$from};
796 dpavlin 604 die "record field $from isn't array\n" unless (ref($r) eq 'ARRAY');
797    
798     my ($i1,$i2) = defined($marc_indicators->{$to}) ? @{ $marc_indicators->{$to} } : (' ',' ');
799 dpavlin 616 warn "## marc_original_order($to,$from) source = ", dump( $r ),$/ if ($debug > 1);
800 dpavlin 604
801     foreach my $d (@$r) {
802    
803 dpavlin 605 if (! defined($d->{subfields}) && ref($d->{subfields}) ne 'ARRAY') {
804 dpavlin 616 warn "# marc_original_order($to,$from): field $from doesn't have subfields specification\n";
805 dpavlin 605 next;
806     }
807    
808 dpavlin 604 my @sfs = @{ $d->{subfields} };
809    
810     die "field $from doesn't have even number of subfields specifications\n" unless($#sfs % 2 == 1);
811    
812 dpavlin 618 warn "#--> d: ",dump($d), "\n#--> sfs: ",dump(@sfs),$/ if ($debug > 2);
813 dpavlin 604
814     my $m = [ $to, $i1, $i2 ];
815    
816     while (my $sf = shift @sfs) {
817 dpavlin 618
818     warn "#--> sf: ",dump($sf), $/ if ($debug > 2);
819 dpavlin 604 my $offset = shift @sfs;
820     die "corrupted sufields specification for field $from\n" unless defined($offset);
821    
822     my $v;
823     if (ref($d->{$sf}) eq 'ARRAY') {
824     $v = $d->{$sf}->[$offset] if (defined($d->{$sf}->[$offset]));
825     } elsif ($offset == 0) {
826     $v = $d->{$sf};
827     } else {
828     die "field $from subfield '$sf' need occurence $offset which doesn't exist", dump($d->{$sf});
829     }
830     push @$m, ( $sf, $v ) if (defined($v));
831     }
832    
833     if ($#{$m} > 2) {
834     push @{ $marc_record->[ $marc_record_offset ] }, $m;
835     }
836     }
837    
838     warn "## marc_record = ", dump( $marc_record ),$/ if ($debug > 1);
839     }
840    
841 dpavlin 813 =head2 marc_count
842 dpavlin 604
843 dpavlin 813 Return number of MARC records created using L</marc_duplicate>.
844    
845     print "created ", marc_count(), " records";
846    
847     =cut
848    
849     sub marc_count {
850     return $#{ $marc_record };
851     }
852    
853    
854 dpavlin 540 =head1 Functions to extract data from input
855    
856     This function should be used inside functions to create C<data_structure> described
857     above.
858    
859 dpavlin 641 =head2 _pack_subfields_hash
860    
861 dpavlin 669 @subfields = _pack_subfields_hash( $h );
862     $subfields = _pack_subfields_hash( $h, 1 );
863 dpavlin 641
864 dpavlin 669 Return each subfield value in array or pack them all together and return scalar
865     with subfields (denoted by C<^>) and values.
866    
867 dpavlin 641 =cut
868    
869     sub _pack_subfields_hash {
870    
871 dpavlin 642 warn "## _pack_subfields_hash( ",dump(@_), " )\n" if ($debug > 1);
872 dpavlin 641
873     my ($h,$include_subfields) = @_;
874    
875 dpavlin 831 # sanity and ease of use
876     return $h if (ref($h) ne 'HASH');
877    
878 dpavlin 641 if ( defined($h->{subfields}) ) {
879     my $sfs = delete $h->{subfields} || die "no subfields?";
880     my @out;
881     while (@$sfs) {
882     my $sf = shift @$sfs;
883     push @out, '^' . $sf if ($include_subfields);
884     my $o = shift @$sfs;
885     if ($o == 0 && ref( $h->{$sf} ) ne 'ARRAY' ) {
886     # single element subfields are not arrays
887 dpavlin 669 #warn "====> $sf $o / $#$sfs ", dump( $sfs, $h->{$sf} ), "\n";
888 dpavlin 667
889 dpavlin 641 push @out, $h->{$sf};
890     } else {
891 dpavlin 669 #warn "====> $sf $o / $#$sfs ", dump( $sfs, $h->{$sf} ), "\n";
892 dpavlin 641 push @out, $h->{$sf}->[$o];
893     }
894     }
895 dpavlin 667 if ($include_subfields) {
896     return join('', @out);
897     } else {
898     return @out;
899     }
900 dpavlin 641 } else {
901 dpavlin 667 if ($include_subfields) {
902     my $out = '';
903 dpavlin 668 foreach my $sf (sort keys %$h) {
904 dpavlin 667 if (ref($h->{$sf}) eq 'ARRAY') {
905     $out .= '^' . $sf . join('^' . $sf, @{ $h->{$sf} });
906     } else {
907     $out .= '^' . $sf . $h->{$sf};
908     }
909     }
910     return $out;
911     } else {
912     # FIXME this should probably be in alphabetical order instead of hash order
913     values %{$h};
914     }
915 dpavlin 641 }
916     }
917    
918 dpavlin 536 =head2 rec1
919 dpavlin 371
920 dpavlin 536 Return all values in some field
921 dpavlin 371
922 dpavlin 536 @v = rec1('200')
923 dpavlin 15
924 dpavlin 536 TODO: order of values is probably same as in source data, need to investigate that
925 dpavlin 15
926 dpavlin 536 =cut
927 dpavlin 15
928 dpavlin 536 sub rec1 {
929     my $f = shift;
930 dpavlin 571 warn "rec1($f) = ", dump( $rec->{$f} ), $/ if ($debug > 1);
931 dpavlin 536 return unless (defined($rec) && defined($rec->{$f}));
932 dpavlin 571 warn "rec1($f) = ", dump( $rec->{$f} ), $/ if ($debug > 1);
933 dpavlin 536 if (ref($rec->{$f}) eq 'ARRAY') {
934 dpavlin 641 my @out;
935     foreach my $h ( @{ $rec->{$f} } ) {
936     if (ref($h) eq 'HASH') {
937     push @out, ( _pack_subfields_hash( $h ) );
938 dpavlin 31 } else {
939 dpavlin 641 push @out, $h;
940 dpavlin 31 }
941 dpavlin 641 }
942     return @out;
943 dpavlin 536 } elsif( defined($rec->{$f}) ) {
944     return $rec->{$f};
945 dpavlin 15 }
946     }
947    
948 dpavlin 536 =head2 rec2
949 dpavlin 15
950 dpavlin 536 Return all values in specific field and subfield
951 dpavlin 13
952 dpavlin 536 @v = rec2('200','a')
953 dpavlin 13
954     =cut
955    
956 dpavlin 536 sub rec2 {
957     my $f = shift;
958     return unless (defined($rec && $rec->{$f}));
959     my $sf = shift;
960 dpavlin 601 warn "rec2($f,$sf) = ", dump( $rec->{$f} ), $/ if ($debug > 1);
961 dpavlin 589 return map {
962     if (ref($_->{$sf}) eq 'ARRAY') {
963     @{ $_->{$sf} };
964     } else {
965     $_->{$sf};
966     }
967     } grep { ref($_) eq 'HASH' && $_->{$sf} } @{ $rec->{$f} };
968 dpavlin 536 }
969 dpavlin 13
970 dpavlin 536 =head2 rec
971 dpavlin 13
972 dpavlin 536 syntaxtic sugar for
973 dpavlin 13
974 dpavlin 536 @v = rec('200')
975     @v = rec('200','a')
976 dpavlin 13
977 dpavlin 750 If rec() returns just single value, it will
978     return scalar, not array.
979    
980 dpavlin 536 =cut
981 dpavlin 373
982 dpavlin 536 sub rec {
983 dpavlin 583 my @out;
984 dpavlin 536 if ($#_ == 0) {
985 dpavlin 583 @out = rec1(@_);
986 dpavlin 536 } elsif ($#_ == 1) {
987 dpavlin 583 @out = rec2(@_);
988 dpavlin 13 }
989 dpavlin 750 if ($#out == 0 && ! wantarray) {
990     return $out[0];
991     } elsif (@out) {
992 dpavlin 583 return @out;
993     } else {
994     return '';
995     }
996 dpavlin 13 }
997    
998 dpavlin 536 =head2 regex
999 dpavlin 15
1000 dpavlin 536 Apply regex to some or all values
1001 dpavlin 15
1002 dpavlin 536 @v = regex( 's/foo/bar/g', @v );
1003 dpavlin 15
1004     =cut
1005    
1006 dpavlin 536 sub regex {
1007     my $r = shift;
1008     my @out;
1009 dpavlin 550 #warn "r: $r\n", dump(\@_);
1010 dpavlin 536 foreach my $t (@_) {
1011     next unless ($t);
1012     eval "\$t =~ $r";
1013     push @out, $t if ($t && $t ne '');
1014 dpavlin 368 }
1015 dpavlin 536 return @out;
1016 dpavlin 15 }
1017    
1018 dpavlin 536 =head2 prefix
1019 dpavlin 15
1020 dpavlin 536 Prefix all values with a string
1021 dpavlin 15
1022 dpavlin 536 @v = prefix( 'my_', @v );
1023 dpavlin 15
1024     =cut
1025    
1026 dpavlin 536 sub prefix {
1027 dpavlin 819 my $p = shift;
1028     return @_ unless defined( $p );
1029 dpavlin 536 return map { $p . $_ } grep { defined($_) } @_;
1030     }
1031 dpavlin 15
1032 dpavlin 536 =head2 suffix
1033 dpavlin 15
1034 dpavlin 536 suffix all values with a string
1035 dpavlin 15
1036 dpavlin 536 @v = suffix( '_my', @v );
1037 dpavlin 15
1038 dpavlin 536 =cut
1039 dpavlin 15
1040 dpavlin 536 sub suffix {
1041 dpavlin 819 my $s = shift;
1042     return @_ unless defined( $s );
1043 dpavlin 536 return map { $_ . $s } grep { defined($_) } @_;
1044 dpavlin 15 }
1045    
1046 dpavlin 536 =head2 surround
1047 dpavlin 13
1048 dpavlin 536 surround all values with a two strings
1049 dpavlin 13
1050 dpavlin 536 @v = surround( 'prefix_', '_suffix', @v );
1051 dpavlin 13
1052     =cut
1053    
1054 dpavlin 536 sub surround {
1055 dpavlin 819 my $p = shift;
1056     my $s = shift;
1057     $p = '' unless defined( $p );
1058     $s = '' unless defined( $s );
1059 dpavlin 536 return map { $p . $_ . $s } grep { defined($_) } @_;
1060 dpavlin 13 }
1061    
1062 dpavlin 536 =head2 first
1063 dpavlin 13
1064 dpavlin 536 Return first element
1065 dpavlin 15
1066 dpavlin 536 $v = first( @v );
1067 dpavlin 13
1068     =cut
1069    
1070 dpavlin 536 sub first {
1071     my $r = shift;
1072     return $r;
1073 dpavlin 13 }
1074    
1075 dpavlin 536 =head2 lookup
1076 dpavlin 13
1077 dpavlin 536 Consult lookup hashes for some value
1078 dpavlin 13
1079 dpavlin 725 @v = lookup(
1080     sub {
1081     'ffkk/peri/mfn'.rec('000')
1082     },
1083     'ffkk','peri','200-a-200-e',
1084     sub {
1085     first(rec(200,'a')).' '.first(rec('200','e'))
1086     }
1087     );
1088 dpavlin 13
1089 dpavlin 725 Code like above will be B<automatically generated> using L<WebPAC::Parse> from
1090     normal lookup definition in C<conf/lookup/something.pl> which looks like:
1091 dpavlin 707
1092 dpavlin 725 lookup(
1093     # which results to return from record recorded in lookup
1094     sub { 'ffkk/peri/mfn' . rec('000') },
1095     # from which database and input
1096     'ffkk','peri',
1097     # such that following values match
1098     sub { first(rec(200,'a')) . ' ' . first(rec('200','e')) },
1099     # if this part is missing, we will try to match same fields
1100     # from lookup record and current one, or you can override
1101     # which records to use from current record using
1102     sub { rec('900','x') . ' ' . rec('900','y') },
1103     )
1104    
1105     You can think about this lookup as SQL (if that helps):
1106    
1107     select
1108     sub { what }
1109     from
1110     database, input
1111     where
1112     sub { filter from lookuped record }
1113     having
1114     sub { optional filter on current record }
1115    
1116     Easy as pie, right?
1117    
1118 dpavlin 13 =cut
1119    
1120 dpavlin 536 sub lookup {
1121 dpavlin 725 my ($what, $database, $input, $key, $having) = @_;
1122    
1123 dpavlin 766 confess "lookup needs 5 arguments: what, database, input, key, having\n" unless ($#_ == 4);
1124 dpavlin 725
1125 dpavlin 729 warn "## lookup ($database, $input, $key)", $/ if ($debug > 1);
1126 dpavlin 725 return unless (defined($lookup->{$database}->{$input}->{$key}));
1127    
1128 dpavlin 736 confess "lookup really need load_row_coderef added to data_structure\n" unless ($load_row_coderef);
1129 dpavlin 725
1130     my $mfns;
1131     my @having = $having->();
1132    
1133 dpavlin 729 warn "## having = ", dump( @having ) if ($debug > 2);
1134 dpavlin 725
1135     foreach my $h ( @having ) {
1136     if (defined($lookup->{$database}->{$input}->{$key}->{$h})) {
1137 dpavlin 752 warn "lookup for $database/$input/$key/$h return ",dump($lookup->{$database}->{$input}->{$key}->{$h}),"\n" if ($debug);
1138 dpavlin 725 $mfns->{$_}++ foreach keys %{ $lookup->{$database}->{$input}->{$key}->{$h} };
1139     }
1140 dpavlin 536 }
1141 dpavlin 725
1142     return unless ($mfns);
1143    
1144     my @mfns = sort keys %$mfns;
1145    
1146 dpavlin 750 warn "# lookup loading $database/$input/$key mfn ", join(",",@mfns)," having ",dump(@having),"\n" if ($debug);
1147 dpavlin 725
1148     my $old_rec = $rec;
1149     my @out;
1150    
1151     foreach my $mfn (@mfns) {
1152 dpavlin 736 $rec = $load_row_coderef->( $database, $input, $mfn );
1153 dpavlin 725
1154 dpavlin 752 warn "got $database/$input/$mfn = ", dump($rec), $/ if ($debug);
1155 dpavlin 725
1156     my @vals = $what->();
1157    
1158     push @out, ( @vals );
1159    
1160 dpavlin 752 warn "lookup for mfn $mfn returned ", dump(@vals), $/ if ($debug);
1161 dpavlin 725 }
1162    
1163     # if (ref($lookup->{$k}) eq 'ARRAY') {
1164     # return @{ $lookup->{$k} };
1165     # } else {
1166     # return $lookup->{$k};
1167     # }
1168    
1169     $rec = $old_rec;
1170    
1171 dpavlin 750 warn "## lookup returns = ", dump(@out), $/ if ($debug);
1172 dpavlin 725
1173 dpavlin 740 if ($#out == 0) {
1174     return $out[0];
1175     } else {
1176     return @out;
1177     }
1178 dpavlin 13 }
1179    
1180 dpavlin 707 =head2 save_into_lookup
1181    
1182 dpavlin 725 Save value into lookup. It associates current database, input
1183     and specific keys with one or more values which will be
1184     associated over MFN.
1185 dpavlin 707
1186 dpavlin 725 MFN will be extracted from first occurence current of field 000
1187     in current record, or if it doesn't exist from L<_set_config> C<_mfn>.
1188    
1189     my $nr = save_into_lookup($database,$input,$key,sub {
1190 dpavlin 707 # code which produce one or more values
1191     });
1192    
1193 dpavlin 725 It returns number of items saved.
1194 dpavlin 707
1195 dpavlin 725 This function shouldn't be called directly, it's called from code created by
1196     L<WebPAC::Parser>.
1197    
1198 dpavlin 707 =cut
1199    
1200     sub save_into_lookup {
1201 dpavlin 712 my ($database,$input,$key,$coderef) = @_;
1202     die "save_into_lookup needs database" unless defined($database);
1203     die "save_into_lookup needs input" unless defined($input);
1204     die "save_into_lookup needs key" unless defined($key);
1205 dpavlin 707 die "save_into_lookup needs CODE" unless ( defined($coderef) && ref($coderef) eq 'CODE' );
1206 dpavlin 725
1207 dpavlin 729 warn "## save_into_lookup rec = ", dump($rec), " config = ", dump($config), $/ if ($debug > 2);
1208 dpavlin 725
1209     my $mfn =
1210     defined($rec->{'000'}->[0]) ? $rec->{'000'}->[0] :
1211     defined($config->{_mfn}) ? $config->{_mfn} :
1212     die "mfn not defined or zero";
1213    
1214     my $nr = 0;
1215    
1216 dpavlin 707 foreach my $v ( $coderef->() ) {
1217 dpavlin 712 $lookup->{$database}->{$input}->{$key}->{$v}->{$mfn}++;
1218 dpavlin 721 warn "# saved lookup $database/$input/$key [$v] $mfn\n" if ($debug > 1);
1219 dpavlin 725 $nr++;
1220 dpavlin 707 }
1221 dpavlin 725
1222     return $nr;
1223 dpavlin 707 }
1224    
1225 dpavlin 595 =head2 config
1226    
1227     Consult config values stored in C<config.yml>
1228    
1229     # return database code (key under databases in yaml)
1230     $database_code = config(); # use _ from hash
1231     $database_name = config('name');
1232     $database_input_name = config('input name');
1233    
1234     Up to three levels are supported.
1235    
1236     =cut
1237    
1238     sub config {
1239     return unless ($config);
1240    
1241     my $p = shift;
1242    
1243     $p ||= '';
1244    
1245     my $v;
1246    
1247     warn "### getting config($p)\n" if ($debug > 1);
1248    
1249     my @p = split(/\s+/,$p);
1250     if ($#p < 0) {
1251     $v = $config->{ '_' }; # special, database code
1252     } else {
1253    
1254     my $c = dclone( $config );
1255    
1256     foreach my $k (@p) {
1257     warn "### k: $k c = ",dump($c),$/ if ($debug > 1);
1258     if (ref($c) eq 'ARRAY') {
1259     $c = shift @$c;
1260     warn "config($p) taking first occurence of '$k', probably not what you wanted!\n";
1261     last;
1262     }
1263    
1264     if (! defined($c->{$k}) ) {
1265     $c = undef;
1266     last;
1267     } else {
1268     $c = $c->{$k};
1269     }
1270     }
1271     $v = $c if ($c);
1272    
1273     }
1274    
1275     warn "## config( '$p' ) = ",dump( $v ),$/ if ($v && $debug);
1276     warn "config( '$p' ) is empty\n" if (! $v);
1277    
1278     return $v;
1279     }
1280    
1281     =head2 id
1282    
1283     Returns unique id of this record
1284    
1285     $id = id();
1286    
1287     Returns C<42/2> for 2nd occurence of MFN 42.
1288    
1289     =cut
1290    
1291     sub id {
1292     my $mfn = $config->{_mfn} || die "no _mfn in config data";
1293     return $mfn . $#{$marc_record} ? $#{$marc_record} + 1 : '';
1294     }
1295    
1296 dpavlin 536 =head2 join_with
1297 dpavlin 13
1298 dpavlin 536 Joins walues with some delimiter
1299 dpavlin 10
1300 dpavlin 536 $v = join_with(", ", @v);
1301 dpavlin 10
1302 dpavlin 536 =cut
1303 dpavlin 10
1304 dpavlin 536 sub join_with {
1305     my $d = shift;
1306 dpavlin 586 warn "### join_with('$d',",dump(@_),")\n" if ($debug > 2);
1307 dpavlin 583 my $v = join($d, grep { defined($_) && $_ ne '' } @_);
1308     return '' unless defined($v);
1309     return $v;
1310 dpavlin 536 }
1311 dpavlin 10
1312 dpavlin 562 =head2 split_rec_on
1313    
1314     Split record subfield on some regex and take one of parts out
1315    
1316     $a_before_semi_column =
1317     split_rec_on('200','a', /\s*;\s*/, $part);
1318    
1319     C<$part> is optional number of element. First element is
1320     B<1>, not 0!
1321    
1322     If there is no C<$part> parameter or C<$part> is 0, this function will
1323     return all values produced by splitting.
1324    
1325     =cut
1326    
1327     sub split_rec_on {
1328     die "split_rec_on need (fld,sf,regex[,part]" if ($#_ < 2);
1329    
1330     my ($fld, $sf, $regex, $part) = @_;
1331 dpavlin 568 warn "### regex ", ref($regex), $regex, $/ if ($debug > 2);
1332 dpavlin 562
1333     my @r = rec( $fld, $sf );
1334     my $v = shift @r;
1335 dpavlin 568 warn "### first rec($fld,$sf) = ",dump($v),$/ if ($debug > 2);
1336 dpavlin 562
1337 dpavlin 604 return '' if ( ! defined($v) || $v =~ /^\s*$/);
1338 dpavlin 566
1339 dpavlin 562 my @s = split( $regex, $v );
1340 dpavlin 568 warn "## split_rec_on($fld,$sf,$regex,$part) = ",dump(@s),$/ if ($debug > 1);
1341 dpavlin 566 if ($part && $part > 0) {
1342 dpavlin 562 return $s[ $part - 1 ];
1343     } else {
1344 dpavlin 571 return @s;
1345 dpavlin 562 }
1346     }
1347    
1348 dpavlin 785 my $hash;
1349    
1350     =head2 set
1351    
1352     set( key => 'value' );
1353    
1354     =cut
1355    
1356     sub set {
1357     my ($k,$v) = @_;
1358 dpavlin 810 warn "## set ( $k => ", dump($v), " )", $/ if ( $debug );
1359 dpavlin 785 $hash->{$k} = $v;
1360     };
1361    
1362     =head2 get
1363    
1364     get( 'key' );
1365    
1366     =cut
1367    
1368     sub get {
1369     my $k = shift || return;
1370     my $v = $hash->{$k};
1371 dpavlin 810 warn "## get $k = ", dump( $v ), $/ if ( $debug );
1372 dpavlin 785 return $v;
1373     }
1374    
1375 dpavlin 791 =head2 count
1376 dpavlin 785
1377 dpavlin 791 if ( count( @result ) == 1 ) {
1378     # do something if only 1 result is there
1379     }
1380    
1381     =cut
1382    
1383     sub count {
1384 dpavlin 810 warn "## count ",dump(@_),$/ if ( $debug );
1385 dpavlin 791 return @_ . '';
1386     }
1387    
1388 dpavlin 536 # END
1389     1;

  ViewVC Help
Powered by ViewVC 1.1.26