3 # Copyright (C) 2007 LibLime
5 # This file is part of Koha.
7 # Koha is free software; you can redistribute it and/or modify it under the
8 # terms of the GNU General Public License as published by the Free Software
9 # Foundation; either version 2 of the License, or (at your option) any later
12 # Koha is distributed in the hope that it will be useful, but WITHOUT ANY
13 # WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
14 # A PARTICULAR PURPOSE. See the GNU General Public License for more details.
16 # You should have received a copy of the GNU General Public License along with
17 # Koha; if not, write to the Free Software Foundation, Inc., 59 Temple Place,
18 # Suite 330, Boston, MA 02111-1307 USA
26 use vars qw($VERSION);
28 # set the version for version checking
33 C4::Matcher - find MARC records matching another one
39 my @matchers = C4::Matcher::GetMatcherList();
41 my $matcher = C4::Matcher->new($record_type);
42 $matcher->threshold($threshold);
43 $matcher->code($code);
44 $matcher->description($description);
46 $matcher->add_simple_matchpoint('isbn', 1000, '020', 'a', -1, 0, '');
47 $matcher->add_simple_matchpoint('Date', 1000, '008', '', 7, 4, '');
48 $matcher->add_matchpoint('isbn', 1000, [ { tag => '020', subfields => 'a', norms => [] } ]);
50 $matcher->add_simple_required_check('245', 'a', -1, 0, '', '245', 'a', -1, 0, '');
51 $matcher->add_required_check([ { tag => '245', subfields => 'a', norms => [] } ],
52 [ { tag => '245', subfields => 'a', norms => [] } ]);
54 my @matches = $matcher->get_matches($marc_record, $max_matches);
56 foreach $match (@matches) {
58 # matches already sorted in order of
60 print "record ID: $match->{'record_id'};
61 print "score: $match->{'score'};
75 my @matchers = C4::Matcher::GetMatcherList();
79 Returns an array of hashrefs list all matchers
80 present in the database. Each hashref includes:
89 my $dbh = C4::Context->dbh;
91 my $sth = $dbh->prepare_cached("SELECT matcher_id, code, description FROM marc_matchers ORDER BY matcher_id");
94 while (my $row = $sth->fetchrow_hashref) {
108 my $matcher = C4::Matcher->new($record_type, $threshold);
112 Creates a new Matcher. C<$record_type> indicates which search
113 database to use, e.g., 'biblio' or 'authority' and defaults to
114 'biblio', while C<$threshold> is the minimum score required for a match
115 and defaults to 1000.
123 $self->{'id'} = undef;
126 $self->{'record_type'} = shift;
128 $self->{'record_type'} = 'biblio';
132 $self->{'threshold'} = shift;
134 $self->{'threshold'} = 1000;
137 $self->{'code'} = '';
138 $self->{'description'} = '';
140 $self->{'matchpoints'} = [];
141 $self->{'required_checks'} = [];
151 my $matcher = C4::Matcher->fetch($id);
155 Creates a matcher object from the version stored
156 in the database. If a matcher with the given
157 id does not exist, returns undef.
164 my $dbh = C4::Context->dbh();
166 my $sth = $dbh->prepare_cached("SELECT * FROM marc_matchers WHERE matcher_id = ?");
168 my $row = $sth->fetchrow_hashref;
170 return undef unless defined $row;
173 $self->{'id'} = $row->{'matcher_id'};
174 $self->{'record_type'} = $row->{'record_type'};
175 $self->{'code'} = $row->{'code'};
176 $self->{'description'} = $row->{'description'};
177 $self->{'threshold'} = int($row->{'threshold'});
181 $self->{'matchpoints'} = [];
182 $sth = $dbh->prepare_cached("SELECT * FROM matcher_matchpoints WHERE matcher_id = ? ORDER BY matchpoint_id");
183 $sth->execute($self->{'id'});
184 while (my $row = $sth->fetchrow_hashref) {
185 my $matchpoint = $self->_fetch_matchpoint($row->{'matchpoint_id'});
186 push @{ $self->{'matchpoints'} }, $matchpoint;
190 $self->{'required_checks'} = [];
191 $sth = $dbh->prepare_cached("SELECT * FROM matchchecks WHERE matcher_id = ? ORDER BY matchcheck_id");
192 $sth->execute($self->{'id'});
193 while (my $row = $sth->fetchrow_hashref) {
194 my $source_matchpoint = $self->_fetch_matchpoint($row->{'source_matchpoint_id'});
195 my $target_matchpoint = $self->_fetch_matchpoint($row->{'target_matchpoint_id'});
197 $matchcheck->{'source_matchpoint'} = $source_matchpoint;
198 $matchcheck->{'target_matchpoint'} = $target_matchpoint;
199 push @{ $self->{'required_checks'} }, $matchcheck;
205 sub _fetch_matchpoint {
207 my $matchpoint_id = shift;
209 my $dbh = C4::Context->dbh;
210 my $sth = $dbh->prepare_cached("SELECT * FROM matchpoints WHERE matchpoint_id = ?");
211 $sth->execute($matchpoint_id);
212 my $row = $sth->fetchrow_hashref;
214 $matchpoint->{'index'} = $row->{'search_index'};
215 $matchpoint->{'score'} = int($row->{'score'});
218 $matchpoint->{'components'} = [];
219 $sth = $dbh->prepare_cached("SELECT * FROM matchpoint_components WHERE matchpoint_id = ? ORDER BY sequence");
220 $sth->execute($matchpoint_id);
221 while ($row = $sth->fetchrow_hashref) {
223 $component->{'tag'} = $row->{'tag'};
224 $component->{'subfields'} = { map { $_ => 1 } split(//, $row->{'subfields'}) };
225 $component->{'offset'} = int($row->{'offset'});
226 $component->{'length'} = int($row->{'length'});
227 $component->{'norms'} = [];
228 my $sth2 = $dbh->prepare_cached("SELECT *
229 FROM matchpoint_component_norms
230 WHERE matchpoint_component_id = ? ORDER BY sequence");
231 $sth2->execute($row->{'matchpoint_component_id'});
232 while (my $row2 = $sth2->fetchrow_hashref) {
233 push @{ $component->{'norms'} }, $row2->{'norm_routine'};
235 push @{ $matchpoint->{'components'} }, $component;
244 my $id = $matcher->store();
248 Stores matcher in database. The return value is the ID
249 of the marc_matchers row. If the matcher was
250 previously retrieved from the database via the fetch()
251 method, the DB representation of the matcher
259 if (defined $self->{'id'}) {
261 $self->_del_matcher_components();
262 $self->_update_marc_matchers();
265 $self->_new_marc_matchers();
267 $self->_store_matcher_components();
268 return $self->{'id'};
271 sub _del_matcher_components {
274 my $dbh = C4::Context->dbh();
275 my $sth = $dbh->prepare_cached("DELETE FROM matchpoints WHERE matcher_id = ?");
276 $sth->execute($self->{'id'});
277 $sth = $dbh->prepare_cached("DELETE FROM matchchecks WHERE matcher_id = ?");
278 $sth->execute($self->{'id'});
279 # foreign key delete cascades take care of deleting relevant rows
280 # from matcher_matchpoints, matchpoint_components, and
281 # matchpoint_component_norms
284 sub _update_marc_matchers {
287 my $dbh = C4::Context->dbh();
288 my $sth = $dbh->prepare_cached("UPDATE marc_matchers
293 WHERE matcher_id = ?");
294 $sth->execute($self->{'code'}, $self->{'description'}, $self->{'record_type'}, $self->{'threshold'}, $self->{'id'});
297 sub _new_marc_matchers {
300 my $dbh = C4::Context->dbh();
301 my $sth = $dbh->prepare_cached("INSERT INTO marc_matchers
302 (code, description, record_type, threshold)
303 VALUES (?, ?, ?, ?)");
304 $sth->execute($self->{'code'}, $self->{'description'}, $self->{'record_type'}, $self->{'threshold'});
305 $self->{'id'} = $dbh->{'mysql_insertid'};
308 sub _store_matcher_components {
311 my $dbh = C4::Context->dbh();
313 my $matcher_id = $self->{'id'};
314 foreach my $matchpoint (@{ $self->{'matchpoints'}}) {
315 my $matchpoint_id = $self->_store_matchpoint($matchpoint);
316 $sth = $dbh->prepare_cached("INSERT INTO matcher_matchpoints (matcher_id, matchpoint_id)
318 $sth->execute($matcher_id, $matchpoint_id);
320 foreach my $matchcheck (@{ $self->{'required_checks'} }) {
321 my $source_matchpoint_id = $self->_store_matchpoint($matchcheck->{'source_matchpoint'});
322 my $target_matchpoint_id = $self->_store_matchpoint($matchcheck->{'target_matchpoint'});
323 $sth = $dbh->prepare_cached("INSERT INTO matchchecks
324 (matcher_id, source_matchpoint_id, target_matchpoint_id)
326 $sth->execute($matcher_id, $source_matchpoint_id, $target_matchpoint_id);
331 sub _store_matchpoint {
333 my $matchpoint = shift;
335 my $dbh = C4::Context->dbh();
337 my $matcher_id = $self->{'id'};
338 $sth = $dbh->prepare_cached("INSERT INTO matchpoints (matcher_id, search_index, score)
340 $sth->execute($matcher_id, $matchpoint->{'index'}, $matchpoint->{'score'});
341 my $matchpoint_id = $dbh->{'mysql_insertid'};
343 foreach my $component (@{ $matchpoint->{'components'} }) {
345 $sth = $dbh->prepare_cached("INSERT INTO matchpoint_components
346 (matchpoint_id, sequence, tag, subfields, offset, length)
347 VALUES (?, ?, ?, ?, ?, ?)");
348 $sth->bind_param(1, $matchpoint_id);
349 $sth->bind_param(2, $seqnum);
350 $sth->bind_param(3, $component->{'tag'});
351 $sth->bind_param(4, join "", sort keys %{ $component->{'subfields'} });
352 $sth->bind_param(5, $component->{'offset'});
353 $sth->bind_param(6, $component->{'length'});
355 my $matchpoint_component_id = $dbh->{'mysql_insertid'};
357 foreach my $norm (@{ $component->{'norms'} }) {
359 $sth = $dbh->prepare_cached("INSERT INTO matchpoint_component_norms
360 (matchpoint_component_id, sequence, norm_routine)
362 $sth->execute($matchpoint_component_id, $normseq, $norm);
365 return $matchpoint_id;
372 $matcher->threshold(1000);
373 my $threshold = $matcher->threshold();
383 @_ ? $self->{'threshold'} = shift : $self->{'threshold'};
390 $matcher->code('ISBN');
391 my $code = $matcher->code();
401 @_ ? $self->{'code'} = shift : $self->{'code'};
408 $matcher->description('match on ISBN');
409 my $description = $matcher->description();
419 @_ ? $self->{'description'} = shift : $self->{'description'};
422 =head2 add_matchpoint
426 $matcher->add_matchpoint($index, $score, $matchcomponents);
430 Adds a matchpoint that may include multiple components. The $index
431 parameter identifies the index that will be searched, while $score
432 is the weight that will be added if a match is found.
434 $matchcomponents should be a reference to an array of matchpoint
435 compoents, each of which should be a hash containing the following
443 The normalization_rules value should in turn be a reference to an
444 array, each element of which should be a reference to a
445 normalization subroutine (under C4::Normalize) to be applied
446 to the source string.
452 my ($index, $score, $matchcomponents) = @_;
455 $matchpoint->{'index'} = $index;
456 $matchpoint->{'score'} = $score;
457 $matchpoint->{'components'} = [];
458 foreach my $input_component (@{ $matchcomponents }) {
459 push @{ $matchpoint->{'components'} }, _parse_match_component($input_component);
461 push @{ $self->{'matchpoints'} }, $matchpoint;
464 =head2 add_simple_matchpoint
468 $matcher->add_simple_matchpoint($index, $score, $source_tag, $source_subfields,
469 $source_offset, $source_length,
474 Adds a simple matchpoint rule -- after composing a key based on the source tag and subfields,
475 normalized per the normalization fuction, search the index. All records retrieved
476 will receive the assigned score.
480 sub add_simple_matchpoint {
482 my ($index, $score, $source_tag, $source_subfields, $source_offset, $source_length, $source_normalizer) = @_;
484 $self->add_matchpoint($index, $score, [
485 { tag => $source_tag, subfields => $source_subfields,
486 offset => $source_offset, length => $source_length,
487 norms => [ $source_normalizer ]
492 =head2 add_required_check
496 $match->add_required_check($source_matchpoint, $target_matchpoint);
500 Adds a required check definition. A required check means that in
501 order for a match to be considered valid, the key derived from the
502 source (incoming) record must match the key derived from the target
503 (already in DB) record.
505 Unlike a regular matchpoint, only the first repeat of each tag
506 in the source and target match criteria are considered.
508 A typical example of a required check would be verifying that the
509 titles and publication dates match.
511 $source_matchpoint and $target_matchpoint are each a reference to
512 an array of hashes, where each hash follows the same definition
513 as the matchpoint component specification in add_matchpoint, i.e.,
521 The normalization_rules value should in turn be a reference to an
522 array, each element of which should be a reference to a
523 normalization subroutine (under C4::Normalize) to be applied
524 to the source string.
528 sub add_required_check {
530 my ($source_matchpoint, $target_matchpoint) = @_;
533 $matchcheck->{'source_matchpoint'}->{'index'} = '';
534 $matchcheck->{'source_matchpoint'}->{'score'} = 0;
535 $matchcheck->{'source_matchpoint'}->{'components'} = [];
536 $matchcheck->{'target_matchpoint'}->{'index'} = '';
537 $matchcheck->{'target_matchpoint'}->{'score'} = 0;
538 $matchcheck->{'target_matchpoint'}->{'components'} = [];
539 foreach my $input_component (@{ $source_matchpoint }) {
540 push @{ $matchcheck->{'source_matchpoint'}->{'components'} }, _parse_match_component($input_component);
542 foreach my $input_component (@{ $target_matchpoint }) {
543 push @{ $matchcheck->{'target_matchpoint'}->{'components'} }, _parse_match_component($input_component);
545 push @{ $self->{'required_checks'} }, $matchcheck;
548 =head2 add_simple_required_check
550 $matcher->add_simple_required_check($source_tag, $source_subfields, $source_offset, $source_length, $source_normalizer,
551 $target_tag, $target_subfields, $target_offset, $target_length, $target_normalizer);
555 Adds a required check, which requires that the normalized keys made from the source and targets
556 must match for a match to be considered valid.
562 sub add_simple_required_check {
564 my ($source_tag, $source_subfields, $source_offset, $source_length, $source_normalizer,
565 $target_tag, $target_subfields, $target_offset, $target_length, $target_normalizer) = @_;
567 $self->add_required_check(
568 [ { tag => $source_tag, subfields => $source_subfields, offset => $source_offset, length => $source_length,
569 norms => [ $source_normalizer ] } ],
570 [ { tag => $target_tag, subfields => $target_subfields, offset => $target_offset, length => $target_length,
571 norms => [ $target_normalizer ] } ]
579 my @matches = $matcher->get_matches($marc_record, $max_matches);
580 foreach $match (@matches) {
581 # matches already sorted in order of
583 print "record ID: $match->{'record_id'};
584 print "score: $match->{'score'};
589 Identifies all of the records matching the given MARC record. For a record already
590 in the database to be considered a match, it must meet the following criteria:
594 =item 1. Total score from its matching field must exceed the supplied threshold.
596 =item 2. It must pass all required checks.
600 Only the top $max_matches matches are returned. The returned array is sorted
601 in order of decreasing score, i.e., the best match is first.
607 my ($source_record, $max_matches) = @_;
611 foreach my $matchpoint (@{ $self->{'matchpoints'} }) {
612 my @source_keys = _get_match_keys($source_record, $matchpoint);
613 next if scalar(@source_keys) == 0;
615 my $query = join(" or ", map { "$matchpoint->{'index'}=$_" } @source_keys);
616 # FIXME only searching biblio index at the moment
617 my ($error, $searchresults) = SimpleSearch($query);
619 warn "search failed ($query) $error" if $error;
620 foreach my $matched (@$searchresults) {
621 $matches{$matched} += $matchpoint->{'score'};
625 # get rid of any that don't meet the threshold
626 %matches = map { ($matches{$_} >= $self->{'threshold'}) ? ($_ => $matches{$_}) : () } keys %matches;
628 # get rid of any that don't meet the required checks
629 %matches = map { _passes_required_checks($source_record, $_, $self->{'required_checks'}) ? ($_ => $matches{$_}) : () }
633 foreach my $marcblob (keys %matches) {
634 my $target_record = MARC::Record->new_from_usmarc($marcblob);
635 my $result = TransformMarcToKoha(C4::Context->dbh, $target_record, '');
636 # FIXME - again, bibliospecific
637 # also, can search engine be induced to give just the number in the first place?
638 my $record_number = $result->{'biblionumber'};
639 push @results, { 'record_id' => $record_number, 'score' => $matches{$marcblob} };
641 @results = sort { $b->{'score'} cmp $a->{'score'} } @results;
642 if (scalar(@results) > $max_matches) {
643 @results = @results[0..$max_matches-1];
649 sub _passes_required_checks {
650 my ($source_record, $target_blob, $matchchecks) = @_;
651 my $target_record = MARC::Record->new_from_usmarc($target_blob); # FIXME -- need to avoid parsing record twice
653 # no checks supplied == automatic pass
654 return 1 if $#{ $matchchecks } == -1;
656 foreach my $matchcheck (@{ $matchchecks }) {
657 my $source_key = join "", _get_match_keys($source_record, $matchcheck->{'source_matchpoint'});
658 my $target_key = join "", _get_match_keys($target_record, $matchcheck->{'target_matchpoint'});
659 return 0 unless $source_key eq $target_key;
664 sub _get_match_keys {
665 my $source_record = shift;
666 my $matchpoint = shift;
667 my $check_only_first_repeat = @_ ? shift : 0;
669 # If there is more than one component to the matchpoint (e.g.,
670 # matchpoint includes both 003 and 001), any repeats
671 # of the first component's tag are identified; repeats
672 # of the subsequent components' tags are appended to
673 # each parallel key dervied from the first component,
674 # up to the number of repeats of the first component's tag.
676 # For example, if the record has one 003 and two 001s, only
677 # one key is retrieved because there is only one 003. The key
678 # will consist of the contents of the first 003 and first 001.
680 # If there are two 003s and two 001s, there will be two keys:
681 # first 003 + first 001
682 # second 003 + second 001
685 for (my $i = 0; $i <= $#{ $matchpoint->{'components'} }; $i++) {
686 my $component = $matchpoint->{'components'}->[$i];
688 FIELD: foreach my $field ($source_record->field($component->{'tag'})) {
690 last FIELD if $j > 0 and $check_only_first_repeat;
691 last FIELD if $i > 0 and $j > $#keys;
693 if ($field->is_control_field()) {
694 if ($component->{'length'}) {
695 $key = _normalize(substr($field->data(), $component->{'offset'}, $component->{'length'}))
696 # FIXME normalize, substr
698 $key = _normalize($field->data());
701 foreach my $subfield ($field->subfields()) {
702 if (exists $component->{'subfields'}->{$subfield->[0]}) {
703 $key .= " " . $subfield->[1];
706 $key = _normalize($key);
709 push @keys, $key if $key;
711 $keys[$j] .= " $key" if $key;
720 sub _parse_match_component {
721 my $input_component = shift;
724 $component->{'tag'} = $input_component->{'tag'};
725 $component->{'subfields'} = { map { $_ => 1 } split(//, $input_component->{'subfields'}) };
726 $component->{'offset'} = exists($input_component->{'offset'}) ? $input_component->{'offset'} : -1;
727 $component->{'length'} = $input_component->{'length'} ? $input_component->{'length'} : 0;
728 $component->{'norms'} = $input_component->{'norms'} ? $input_component->{'norms'} : [];
733 # FIXME - default normalizer
735 my $value = uc shift;
739 $value =~ s/[.;,\]\[\)\(\/"']//g;
747 Koha Development Team <info@koha.org>
749 Galen Charlton <galen.charlton@liblime.com>