6 use DateTime::Format::Strptime;
9 # for inherited methods to work properly, we need to force a
10 # MARC::Record version greater than 2.0.0
11 use MARC::Record "2.0.1";
12 use base 'MARC::Record';
14 use OpenILS::Utils::MFHD::Caption;
15 use OpenILS::Utils::MFHD::Holding;
19 my $class = ref($proto) || $proto;
22 $self->{_strp_date} = new DateTime::Format::Strptime(pattern => '%F');
24 $self->{_mfhd_CAPTIONS} = {};
25 $self->{_mfhd_COMPRESSIBLE} = (substr($self->leader, 17, 1) =~ /[45]/);
27 foreach my $field ('853', '854', '855') {
29 foreach my $caption ($self->field($field)) {
32 $cap_id = $caption->subfield('8') || '0';
34 if (exists $captions->{$cap_id}) {
35 carp "Multiple MFHD captions with label '$cap_id'";
38 $captions->{$cap_id} = new MFHD::Caption($caption);
39 if ($self->{_mfhd_COMPRESSIBLE}) {
40 $self->{_mfhd_COMPRESSIBLE} &&=
41 $captions->{$cap_id}->compressible;
44 $self->{_mfhd_CAPTIONS}->{$field} = $captions;
47 foreach my $field ('863', '864', '865') {
51 ($cap_field = $field) =~ s/6/5/;
53 foreach my $hfield ($self->field($field)) {
54 my ($linkage, $link_id, $seqno);
57 $linkage = $hfield->subfield('8');
58 ($link_id, $seqno) = split(/\./, $linkage);
60 if (!exists $holdings->{$link_id}) {
61 $holdings->{$link_id} = {};
64 new MFHD::Holding($seqno, $hfield,
65 $self->{_mfhd_CAPTIONS}->{$cap_field}->{$link_id});
66 $holdings->{$link_id}->{$seqno} = $holding;
68 if ($self->{_mfhd_COMPRESSIBLE}) {
69 $self->{_mfhd_COMPRESSIBLE} &&= $holding->validate;
72 $self->{_mfhd_HOLDINGS}->{$field} = $holdings;
82 return $self->{_mfhd_COMPRESSIBLE};
85 sub caption_link_ids {
89 return sort keys %{$self->{_mfhd_CAPTIONS}->{$field}};
92 # optional argument to get back a 'hashref' or an 'array' (default)
96 my $return_type = shift;
98 # TODO: add support for caption types as argument? (base, index, supplement)
99 my @sorted_ids = $self->caption_link_ids($tag);
101 if (defined($return_type) and $return_type eq 'hashref') {
103 foreach my $link_id (@sorted_ids) {
104 $captions{$link_id} = $self->{_mfhd_CAPTIONS}{$tag}{$link_id};
109 foreach my $link_id (@sorted_ids) {
110 push(@captions, $self->{_mfhd_CAPTIONS}{$tag}{$link_id});
119 my $field_count = $self->SUPER::append_fields(@_);
121 foreach my $field (@_) {
122 $self->_avoid_link_collision($field);
123 my $field_type = ref $field;
124 if ($field_type eq 'MFHD::Holding') {
125 $self->{_mfhd_HOLDINGS}{$field->tag}{$field->caption->link_id}{$field->seqno} = $field;
126 } elsif ($field_type eq 'MFHD::Caption') {
127 $self->{_mfhd_CAPTIONS}{$field->tag}{$field->link_id} = $field;
140 my $field_count = $self->SUPER::delete_field($field);
142 my $field_type = ref($field);
143 if ($field_type eq 'MFHD::Holding') {
144 delete($self->{_mfhd_HOLDINGS}{$field->tag}{$field->caption->link_id}{$field->seqno});
145 } elsif ($field_type eq 'MFHD::Caption') {
146 delete($self->{_mfhd_CAPTIONS}{$field->tag}{$field->link_id});
154 sub insert_fields_before {
158 my $field_count = $self->SUPER::insert_fields_before($before, @_);
160 foreach my $field (@_) {
161 $self->_avoid_link_collision($field);
162 my $field_type = ref $field;
163 if ($field_type eq 'MFHD::Holding') {
164 $self->{_mfhd_HOLDINGS}{$field->tag}{$field->caption->link_id}{$field->seqno} = $field;
165 } elsif ($field_type eq 'MFHD::Caption') {
166 $self->{_mfhd_CAPTIONS}{$field->tag}{$field->link_id} = $field;
175 sub insert_fields_after {
179 my $field_count = $self->SUPER::insert_fields_after($after, @_);
181 foreach my $field (@_) {
182 $self->_avoid_link_collision($field);
183 my $field_type = ref $field;
184 if ($field_type eq 'MFHD::Holding') {
185 $self->{_mfhd_HOLDINGS}{$field->tag}{$field->caption->link_id}{$field->seqno} = $field;
186 } elsif ($field_type eq 'MFHD::Caption') {
187 $self->{_mfhd_CAPTIONS}{$field->tag}{$field->link_id} = $field;
196 sub _avoid_link_collision {
200 my $fieldref = ref($field);
201 if ($fieldref eq 'MFHD::Holding') {
202 my $seqno = $field->seqno;
203 my $changed_seqno = 0;
204 if (exists($self->{_mfhd_HOLDINGS}{$field->tag}{$field->caption->link_id}{$seqno})) {
208 } while (exists($self->{_mfhd_HOLDINGS}{$field->tag}{$field->caption->link_id}{$seqno}));
210 $field->seqno($seqno) if $changed_seqno;
211 } elsif ($fieldref eq 'MFHD::Caption') {
212 my $link_id = $field->link_id;
213 my $changed_link_id = 0;
214 if (exists($self->{_mfhd_CAPTIONS}{$field->tag}{$link_id})) {
216 $changed_link_id = 1;
219 } while (exists($self->{_mfhd_CAPTIONS}{$field->tag}{$link_id}));
221 $field->link_id($link_id) if $changed_link_id;
225 sub active_captions {
229 # TODO: add support for caption types as argument? (basic, index, supplement)
233 @captions = $self->captions($tag);
235 # TODO: for now, we will assume the last 85X field is active
236 # and the rest are historical. The standard is hazy about
237 # how multiple active patterns of the same 85X type should be
238 # handled. We will, however, return as an array for future
240 push(@active_captions, $captions[-1]);
242 return @active_captions;
251 sort { $a->seqno <=> $b->seqno }
252 values %{$self->{_mfhd_HOLDINGS}->{$field}->{$capid}};
259 return $self->{_strp_date}->parse_datetime($holding->chron_to_date);
263 # generate_predictions()
264 # Accepts a hash ref of options initially defined as:
265 # base_holding : reference to the holding field to predict from
266 # num_to_predict : the number of issues you wish to predict
268 # end_holding : holding field ref, keep predicting until you meet or exceed it
270 # end_date : keep predicting until you exceed this
272 # The basic method is to first convert to a single holding if compressed, then
273 # increment the holding and save the resulting values to @predictions.
275 # returns @predictions, an array of holding field refs (including end_holding
276 # if applicable but NOT base_holding)
278 sub generate_predictions {
279 my ($self, $options) = @_;
281 my $base_holding = $options->{base_holding};
282 my $num_to_predict = $options->{num_to_predict};
283 my $end_holding = $options->{end_holding};
284 my $end_date = $options->{end_date};
285 my $max_to_predict = $options->{max_to_predict} || 10000; # fail-safe
287 if (!defined($base_holding)) {
288 carp("Base holding not defined in generate_predictions, returning empty set");
291 if ($base_holding->is_compressed) {
292 carp("Ambiguous compressed base holding in generate_predictions, returning empty set");
295 my $curr_holding = $base_holding->clone; # prevent side-effects
299 if ($num_to_predict) {
300 for (my $i = 0; $i < $num_to_predict; $i++) {
301 push(@predictions, $curr_holding->increment->clone);
303 } elsif (defined($end_holding)) {
304 $end_holding = $end_holding->clone; # prevent side-effects
305 my $next_holding = $curr_holding->increment->clone;
306 my $num_predicted = 0;
307 while ($next_holding le $end_holding) {
308 push(@predictions, $next_holding);
310 if ($num_predicted >= $max_to_predict) {
311 carp("Maximum prediction count exceeded");
314 $next_holding = $curr_holding->increment->clone;
316 } elsif (defined($end_date)) {
317 my $next_holding = $curr_holding->increment->clone;
318 my $num_predicted = 0;
319 while ($self->_holding_date($next_holding) <= $end_date) {
320 push(@predictions, $next_holding);
322 if ($num_predicted >= $max_to_predict) {
323 carp("Maximum prediction count exceeded");
326 $next_holding = $curr_holding->increment->clone;
334 # create an array of compressed holdings from all holdings for a given caption,
335 # compressing as needed
337 # Optionally you can skip sorting, but the resulting compression will be compromised
338 # if the current holdings are out of order
340 # TODO: gap marking, gap preservation
342 # TODO: some of this could be moved to the Caption object to allow for
343 # decompression in the absense of an overarching MFHD object
345 sub get_compressed_holdings {
349 my $skip_sort = $opts->{'skip_sort'};
351 # make sure none are compressed
354 @decomp_holdings = $self->get_decompressed_holdings($caption, {'skip_sort' => 1});
356 # sort for best algorithm
357 @decomp_holdings = $self->get_decompressed_holdings($caption, {'dedupe' => 1});
360 return () if !@decomp_holdings;
362 my $runner = $decomp_holdings[0]->clone->increment;
363 my $curr_holding = shift(@decomp_holdings);
364 $curr_holding = $curr_holding->clone;
366 $curr_holding->seqno($seqno);
369 foreach my $holding (@decomp_holdings) {
370 if ($runner eq $holding) {
371 $curr_holding->extend;
373 # } elsif ($holding eq $last_holding) {
374 # carp("Found duplicate holding in compression set, skipping");
375 } elsif ($runner gt $holding) { # should not happen unless holding is not in series
376 carp("Found unexpected holding, skipping");
378 push(@comp_holdings, $curr_holding);
379 while ($runner le $holding) {
382 $curr_holding = $holding->clone;
384 $curr_holding->seqno($seqno);
386 # $last_holding = $holding;
388 push(@comp_holdings, $curr_holding);
390 return @comp_holdings;
394 # create an array of single holdings from all holdings for a given caption,
395 # decompressing as needed
397 # resulting array is returned as they come in the record, unsorted
399 # optional argument will reorder and renumber the holdings before returning
401 # TODO: some of this could be moved to the Caption (and/or Holding) object to
402 # allow for decompression in the absense of an overarching MFHD object
404 sub get_decompressed_holdings {
408 my $skip_sort = $opts->{'skip_sort'};
409 my $dedupe = $opts->{'dedupe'};
411 if ($dedupe and $skip_sort) {
412 carp("Attempted deduplication without sorting, failure likely");
415 my $htag = $caption->tag;
416 my $link_id = $caption->link_id;
418 my @holdings = $self->holdings($htag, $link_id);
420 return () if !@holdings;
424 foreach my $holding (@holdings) {
425 if (!$holding->is_compressed) {
426 push(@decomp_holdings, $holding->clone);
428 my $base_holding = $holding->clone->compressed_to_first;
429 my @new_holdings = $self->generate_predictions(
430 {'base_holding' => $base_holding,
431 'end_holding' => $holding->clone->compressed_to_last});
432 push(@decomp_holdings, $base_holding, @new_holdings);
436 unless ($skip_sort) {
437 my @temp_holdings = sort {$a cmp $b} @decomp_holdings;
438 @decomp_holdings = @temp_holdings;
441 my @return_holdings = (shift(@decomp_holdings));
442 $return_holdings[0]->seqno(1);
444 foreach my $holding (@decomp_holdings) { # renumber sequence
445 if ($holding eq $return_holdings[-1] and $dedupe) {
446 carp("Found duplicate holding in decompression set, discarding");
449 $holding->seqno($seqno);
451 push(@return_holdings, $holding);
454 return @return_holdings;
458 # format_holdings(): Generate textual display of all holdings in record
459 # for given type of caption (853--855) taking into account all the
460 # captions, holdings statements, and textual
463 # returns string formatted holdings as one very long line.
464 # Caller must provide any label (such as "library has:" and insert
465 # line breaks as appropriate.
467 # Translate caption field labels to the corresponding textual holdings
468 # statement labels. That is, convert 853 "Basic bib unit" caption to
469 # 866 "basic bib unit" text holdings label.
477 sub format_holdings {
483 my $holdings_stmt = '';
486 # convert caption field id to holdings field id
487 ($holdings_field = $field) =~ s/5/6/;
489 # Textual holdings statements complicate the basic algorithm for
490 # formatting the holdings: If there's a textual holdings statement
491 # with the subfield "$80", then that overrides ALL the MFHD holdings
492 # information and is all that is displayed. Otherwise, the textual
493 # holdings statements will either replace some of the MFHD holdings
494 # information, or supplement it, depending on the value of the
495 # $8 linkage subfield.
497 if (defined $self->field($cap_to_txt{$field})) {
498 @txt_holdings = $self->field($cap_to_txt{$field});
500 foreach my $txt (@txt_holdings) {
502 # if there's a $80 subfield, then we're done, it's
503 # all the formatted holdings
504 if ($txt->subfield('8') eq '0') {
505 # textual holdings statement that completely
506 # replaces MFHD holdings in 853/863, etc.
507 $holdings_stmt = $txt->subfield('a');
509 if (defined $txt->subfield('z')) {
510 $holdings_stmt .= ' -- ' . $txt->subfield('z');
513 printf("# format_holdings() returning %s txt holdings\n",
514 $cap_to_txt{$field});
515 return $holdings_stmt;
518 # If there are non-$80 subfields in the textual holdings
519 # then we need to keep track of the subfields, so we can
520 # intersperse the textual holdings in with the the calculated
521 # holdings from the 853/863 fields.
522 foreach my $linkid ($txt->subfield('8')) {
523 $txt_link_ids{$linkid} = $txt;
528 # Now loop through all the captions, finding the corresponding
529 # holdings statements (either MFHD or textual), and build up the
530 # complete formatted holdings statement. The textual holdings statements
531 # have either the same link id field as a caption, which means that
532 # the text holdings win, or they have ids that are interfiled with
533 # the captions, which mean they go into the middle.
535 my @ids = sort($self->caption_link_ids($field), keys %txt_link_ids);
536 foreach my $cap_id (@ids) {
537 my $last_txt = undef;
539 if (exists $txt_link_ids{$cap_id}) {
540 # there's a textual holding statement with this caption ID,
541 # so just use that. This covers both the "replaces" and
542 # the "supplements" holdings information options.
544 # a single textual holdings statement can replace multiple
545 # captions. If the _last_ caption we saw had a textual
546 # holdings statement, and this caption has the same one, then
547 # we don't add the holdings again.
548 if (!defined $last_txt || ($last_txt != $txt_link_ids{$cap_id})) {
549 my $txt = $txt_link_ids{$cap_id};
550 $holdings_stmt .= ',' if $holdings_stmt;
551 $holdings_stmt .= $txt->subfield('a');
552 if (defined $txt->subfield('z')) {
553 $holdings_stmt .= ' -- ' . $txt->subfield('z');
561 # We found a caption that doesn't have a corresponding textual
562 # holdings statement, so reset $last_txt to undef.
565 my @holdings = $self->holdings($holdings_field, $cap_id);
567 next unless scalar @holdings;
569 # XXX Need to format compressed holdings. see code in test.pl
570 # for example. Try to do it without indexing?
571 $holdings_stmt .= ',' if $holdings_stmt;
573 if ($self->compressible) {
574 $start = $l = shift @holdings;
575 $holdings_stmt .= $l->format;
577 while (my $h = shift @holdings) {
578 if (!$h->matches($l->next)) {
579 # this item is not part of the current run,
580 # close out the run and record this item
582 $holdings_stmt .= '-' . $l->format;
585 $holdings_stmt .= ',' . $h->format;
587 } elsif (!scalar(@holdings) || defined($h->subfield('z'))) {
588 # This is the end of the holdings for this caption
589 # or this item has a public note that we want
591 $holdings_stmt .= '-' . $h->format;
594 if (defined $h->subfield('z')) {
595 $holdings_stmt .= ' -- ' . $h->subfield('z');
601 $holdings_stmt .= ',' if $holdings_stmt;
602 $holdings_stmt .= (shift @holdings)->format;
603 foreach my $h (@holdings) {
604 $holdings_stmt .= ',' . $h->format;
605 if (defined $h->subfield('z')) {
606 $holdings_stmt .= ' -- ' . $h->subfield('z');
612 return $holdings_stmt;