3 # Copyright 2000-2002 Katipo Communications
4 # Parts Copyright 2013 Prosentient Systems
6 # This file is part of Koha.
8 # Koha is free software; you can redistribute it and/or modify it
9 # under the terms of the GNU General Public License as published by
10 # the Free Software Foundation; either version 3 of the License, or
11 # (at your option) any later version.
13 # Koha is distributed in the hope that it will be useful, but
14 # WITHOUT ANY WARRANTY; without even the implied warranty of
15 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 # GNU General Public License for more details.
18 # You should have received a copy of the GNU General Public License
19 # along with Koha; if not, see <http://www.gnu.org/licenses>.
25 use C4::Koha qw( GetNormalizedISBN );
26 use C4::Charset qw( MarcToUTF8Record SetUTF8Flag );
27 use MARC::File::USMARC;
29 use C4::ImportBatch qw( GetZ3950BatchId AddBiblioToBatch AddAuthToBatch );
30 use C4::AuthoritiesMarc; #GuessAuthTypeCode, FindDuplicateAuthority
35 our (@ISA, @EXPORT_OK);
39 @EXPORT_OK = qw(BreedingSearch Z3950Search Z3950SearchAuth);
44 C4::Breeding : module to add biblios to import_records via
45 the breeding/reservoir API.
49 Z3950Search($pars, $template);
50 ($count, @results) = &BreedingSearch($title,$isbn);
54 This module contains routines related to Koha's Z39.50 search into
55 cataloguing reservoir features.
59 ($count, @results) = &BreedingSearch($title,$isbn);
60 C<$title> contains the title,
61 C<$isbn> contains isbn or issn,
63 C<$count> is the number of items in C<@results>. C<@results> is an
64 array of references-to-hash; the keys are the items from the C<import_records> and
65 C<import_biblios> tables of the Koha database.
70 my ($search,$isbn) = @_;
71 my $dbh = C4::Context->dbh;
77 # normalise ISBN like at import
78 $isbn = C4::Koha::GetNormalizedISBN($isbn);
80 $query = "SELECT import_record_id, file_name, isbn, title, author
82 JOIN import_records USING (import_record_id)
83 JOIN import_batches USING (import_batch_id)
86 if (defined($search) && length($search)>0) {
87 $search =~ s/(\s+)/\%/g;
88 $query .= "title like ? OR author like ?";
89 push(@bind,"%$search%", "%$search%");
91 if ($#bind!=-1 && defined($isbn) && length($isbn)>0) {
94 if (defined($isbn) && length($isbn)>0) {
95 $query .= "isbn like ?";
98 $sth = $dbh->prepare($query);
100 while (my $data = $sth->fetchrow_hashref) {
101 $results[$count] = $data;
102 # FIXME - hack to reflect difference in name
103 # of columns in old marc_breeding and import_records
104 # There needs to be more separation between column names and
105 # field names used in the templates </soapbox>
106 $data->{'file'} = $data->{'file_name'};
107 $data->{'id'} = $data->{'import_record_id'};
112 return($count, @results);
113 } # sub breedingsearch
118 Z3950Search($pars, $template);
120 Parameters for Z3950 search are all passed via the $pars hash. It may contain isbn, title, author, dewey, subject, lccall, controlnumber, stdid, srchany.
121 Also it should contain an arrayref id that points to a list of id's of the z3950 targets to be queried (see z3950servers table).
122 This code is used in acqui/z3950_search and cataloging/z3950_search.
123 The second parameter $template is a Template object. The routine uses this parameter to store the found values into the template.
128 my ($pars, $template)= @_;
130 my @id= @{$pars->{id}};
131 my $page= $pars->{page};
132 my $biblionumber= $pars->{biblionumber};
137 my @breeding_loop = ();
144 my ( $zquery, $squery ) = _bib_build_query( $pars );
146 my $schema = Koha::Database->new()->schema();
147 my $rs = $schema->resultset('Z3950server')->search(
149 { result_class => 'DBIx::Class::ResultClass::HashRefInflator' },
151 my @servers = $rs->all;
152 foreach my $server ( @servers ) {
153 my $server_zquery = $zquery;
154 if(my $attributes = $server->{attributes}){
155 $server_zquery = "$attributes $zquery";
157 $oConnection[$s] = _create_connection( $server );
159 $server->{servertype} eq 'zed'?
160 $oConnection[$s]->search_pqf( $server_zquery ):
161 $oConnection[$s]->search(ZOOM::Query::CQL->new(
162 _translate_query( $server, $squery )));
165 my $xslh = Koha::XSLT::Base->new;
168 while ( $nremaining-- ) {
171 while ( ( $k = ZOOM::event( \@oConnection ) ) != 0 ) {
172 $event = $oConnection[ $k - 1 ]->last_event();
173 last if $event == ZOOM::Event::ZEND;
178 my ($error)= $oConnection[$k]->error_x(); #ignores errmsg, addinfo, diagset
180 if ($error =~ m/^(10000|10007)$/ ) {
181 push(@errconn, { server => $servers[$k]->{host}, error => $error } );
185 my $numresults = $oResult[$k]->size();
188 if ( $numresults > 0 and $numresults >= (($page-1)*20)) {
189 $show_next = 1 if $numresults >= ($page*20);
190 $total_pages = int($numresults/20)+1 if $total_pages < ($numresults/20);
191 for ($i = ($page-1)*20; $i < (($numresults < ($page*20)) ? $numresults : ($page*20)); $i++) {
192 if ( $oResult[$k]->record($i) ) {
194 ( $res, $error ) = _handle_one_result( $oResult[$k]->record($i), $servers[$k], ++$imported, $biblionumber, $xslh ); #ignores error in sequence numbering
195 push @breeding_loop, $res if $res;
196 push @errconn, { server => $servers[$k]->{servername}, error => $error, seq => $i+1 } if $error;
199 push @errconn, { 'server' => $servers[$k]->{servername}, error => ( ( $oConnection[$k]->error_x() )[0] ), seq => $i+1 };
207 numberpending => $nremaining,
208 current_page => $page,
209 total_pages => $total_pages,
210 show_nextbutton => $show_next?1:0,
211 show_prevbutton => $page!=1,
215 #close result sets and connections
217 $oResult[$_]->destroy();
218 $oConnection[$_]->destroy();
222 breeding_loop => \@breeding_loop,
223 servers => \@servers,
228 sub _auth_build_query {
232 nameany => '@attr 1=1002 "#term" ',
233 authorany => '@attr 1=1003 "#term" ',
234 authorcorp => '@attr 1=2 "#term" ',
235 authorpersonal => '@attr 1=1 "#term" ',
236 authormeetingcon => '@attr 1=3 "#term" ',
237 subject => '@attr 1=21 "#term" ',
238 subjectsubdiv => '@attr 1=47 "#term" ',
239 title => '@attr 1=4 "#term" ',
240 uniformtitle => '@attr 1=6 "#term" ',
241 srchany => '@attr 1=1016 "#term" ',
242 controlnumber => '@attr 1=12 "#term" ',
245 return _build_query( $pars, $qry_build );
248 sub _bib_build_query {
253 isbn => '@attr 1=7 @attr 5=1 "#term" ',
254 issn => '@attr 1=8 @attr 5=1 "#term" ',
255 title => '@attr 1=4 "#term" ',
256 author => '@attr 1=1003 "#term" ',
257 dewey => '@attr 1=16 "#term" ',
258 subject => '@attr 1=21 "#term" ',
259 lccall => '@attr 1=16 @attr 2=3 @attr 3=1 @attr 4=1 @attr 5=1 '.
260 '@attr 6=1 "#term" ',
261 controlnumber => '@attr 1=12 "#term" ',
262 srchany => '@attr 1=1016 "#term" ',
263 stdid => '@attr 1=1007 "#term" ',
264 publicationyear => '@attr 1=31 "#term" '
267 return _build_query( $pars, $qry_build );
272 my ( $pars, $qry_build ) = @_;
277 foreach my $k ( sort keys %$pars ) {
278 #note that the sort keys forces an identical result under Perl 5.18
279 #one of the unit tests is based on that assumption
280 if( ( my $val=$pars->{$k} ) && $qry_build->{$k} ) {
281 $qry_build->{$k} =~ s/#term/$val/g;
282 $zquery .= $qry_build->{$k};
283 $squery .= "[$k]=\"$val\" and ";
287 $zquery = "\@and " . $zquery for 2..$nterms;
288 $squery =~ s/ and $//;
289 return ( $zquery, $squery );
292 sub _handle_one_result {
293 my ( $zoomrec, $servhref, $seq, $bib, $xslh )= @_;
295 my $raw= $zoomrec->raw();
297 if( $servhref->{servertype} eq 'sru' ) {
298 $marcrecord= MARC::Record->new_from_xml( $raw, 'UTF-8',
299 $servhref->{syntax} );
300 $marcrecord->encoding('UTF-8');
302 ($marcrecord) = MarcToUTF8Record($raw, C4::Context->preference('marcflavour'), $servhref->{encoding} // "iso-5426" ); #ignores charset return values
304 SetUTF8Flag($marcrecord);
306 ( $marcrecord, $error ) = _do_xslt_proc($marcrecord, $servhref, $xslh);
308 my $batch_id = GetZ3950BatchId($servhref->{servername});
309 my $breedingid = AddBiblioToBatch($batch_id, $seq, $marcrecord, 'UTF-8', 0);
310 #Last zero indicates: no update for batch record counts
313 #call to TransformMarcToKoha replaced by next call
314 #we only need six fields from the marc record
318 biblionumber => $bib,
319 server => $servhref->{servername},
320 breedingid => $breedingid,
321 }, $marcrecord) if $breedingid;
322 return ( $row, $error );
326 my ( $marc, $server, $xslh ) = @_;
327 return $marc if !$server->{add_xslt};
329 my $htdocs = C4::Context->config('intrahtdocs');
330 my $theme = C4::Context->preference("template"); #staff
331 my $lang = C4::Languages::getlanguage() || 'en';
333 my @files= split ',', $server->{add_xslt};
334 my $xml = $marc->as_xml;
335 foreach my $f ( @files ) {
336 $f =~ s/^\s+//; $f =~ s/\s+$//; next if !$f;
337 $f = C4::XSLT::_get_best_default_xslt_filename(
338 $htdocs, $theme, $lang, $f ) unless $f =~ /^\//;
339 $xml = $xslh->transform( $xml, $f );
340 last if $xslh->err; #skip other files
343 return MARC::Record->new_from_xml($xml, 'UTF-8');
345 return ( $marc, $xslh->err ); #original record in case of errors
350 my ($row, $record)=@_;
352 title => 'biblio.title',
353 author => 'biblio.author',
354 isbn =>'biblioitems.isbn',
355 lccn =>'biblioitems.lccn', #LC control number (not call number)
356 edition =>'biblioitems.editionstatement'
358 $fetch{date} = C4::Context->preference('marcflavour') eq "MARC21" ? 'biblio.copyrightdate' : 'biblioitems.publicationyear';
360 foreach my $k (keys %fetch) {
361 $row->{$k} = C4::Biblio::TransformMarcToKohaOneField( $fetch{$k}, $record );
363 $row->{date}//= $row->{date2};
364 $row->{isbn}=_isbn_replace($row->{isbn});
366 $row = _add_custom_field_rowdata($row, $record);
371 sub _add_custom_field_rowdata
373 my ( $row, $record ) = @_;
374 my $pref_newtags = C4::Context->preference('AdditionalFieldsInZ3950ResultSearch');
375 my $pref_flavour = C4::Context->preference('MarcFlavour');
377 $pref_newtags =~ s/^\s+|\s+$//g;
378 $pref_newtags =~ s/\h+/ /g;
382 foreach my $field (split /\,/, $pref_newtags) {
383 $field =~ s/^\s+|\s+$//g ; # trim whitespace
384 my ($tag, $subtags) = split(/\$/, $field);
386 if ( $record->field($tag) ) {
389 for my $marcfield ($record->field($tag)) {
392 for my $code (split //, $subtags) {
393 if ( $marcfield->subfield($code) ) {
394 $str .= $marcfield->subfield($code) . ' ';
397 if ( not $str eq '') {
400 } elsif ( $tag == 10 ) {
401 push @content, ( $pref_flavour eq "MARC21" ? $marcfield->data : $marcfield->as_string );
402 } elsif ( $tag < 10 ) {
403 push @content, $marcfield->data();
405 push @content, $marcfield->as_string();
410 $row->{$field} = \@content;
411 push( @addnumberfields, $field );
416 $row->{'addnumberfields'} = \@addnumberfields;
423 return unless defined $isbn;
424 $isbn =~ s/ |-|\.//g;
425 $isbn =~ s/\|/ \| /g;
430 sub _create_connection {
432 my $option1= ZOOM::Options->new();
433 $option1->option( 'async' => 1 );
434 $option1->option( 'elementSetName', 'F' );
435 $option1->option( 'preferredRecordSyntax', $server->{syntax} );
436 $option1->option( 'timeout', $server->{timeout} ) if $server->{timeout};
438 if( $server->{servertype} eq 'sru' ) {
439 foreach( split ',', $server->{sru_options}//'' ) {
440 #first remove surrounding spaces at comma and equals-sign
442 my @temp= split '=', $_, 2;
443 @temp= map { my $c=$_; $c=~s/^\s+|\s+$//g; $c; } @temp;
444 $option1->option( $temp[0] => $temp[1] ) if @temp;
446 } elsif( $server->{servertype} eq 'zed' ) {
447 $option1->option( 'databaseName', $server->{db} );
448 $option1->option( 'user', $server->{userid} ) if $server->{userid};
449 $option1->option( 'password', $server->{password} ) if $server->{password};
451 my $obj= ZOOM::Connection->create($option1);
452 if( $server->{servertype} eq 'sru' ) {
453 my $host= $server->{host};
454 if( $host !~ /^https?:\/\// ) {
455 #Normally, host will not be prefixed by protocol.
456 #In that case we can (safely) assume http.
457 #In case someone prefixed with https, give it a try..
458 $host = 'http://' . $host;
460 $obj->connect( $host.':'.$server->{port}.'/'.$server->{db} );
462 $obj->connect( $server->{host}, $server->{port} );
467 sub _translate_query { #SRU query adjusted per server cf. srufields column
468 my ($server, $query) = @_;
470 #sru_fields is in format title=field,isbn=field,...
471 #if a field doesn't exist, try anywhere or remove [field]=
472 my @parts= split(',', $server->{sru_fields} );
473 my %trans= map { if( /=/ ) { ( $`,$' ) } else { () } } @parts;
474 my $any= $trans{srchany}?$trans{srchany}.'=':'';
477 foreach my $key (keys %trans) {
482 $q=~s/\[$key\]=/$any/g;
485 $q=~s/\[\w+\]=/$any/g; # remove remaining fields (not found in field list)
489 =head2 ImportBreedingAuth
491 ImportBreedingAuth( $marcrecord, $filename, $encoding, $heading );
493 ImportBreedingAuth imports MARC records in the reservoir (import_records table) or returns their id if they already exist.
497 sub ImportBreedingAuth {
498 my ( $marcrecord, $filename, $encoding, $heading ) = @_;
499 my $dbh = C4::Context->dbh;
501 my $batch_id = GetZ3950BatchId($filename);
502 my $searchbreeding = $dbh->prepare("select import_record_id from import_auths where control_number=? and authorized_heading=?");
504 my $controlnumber = $marcrecord->field('001')->data;
506 # Normalize the record so it doesn't have separated diacritics
507 SetUTF8Flag($marcrecord);
509 $searchbreeding->execute($controlnumber,$heading);
510 my ($breedingid) = $searchbreeding->fetchrow;
512 return $breedingid if $breedingid;
513 $breedingid = AddAuthToBatch($batch_id, 0, $marcrecord, $encoding);
517 =head2 Z3950SearchAuth
519 Z3950SearchAuth($pars, $template);
521 Parameters for Z3950 search are all passed via the $pars hash. It may contain nameany, namepersonal, namecorp, namemeetingcon,
522 title, uniform title, subject, subjectsubdiv, srchany.
523 Also it should contain an arrayref id that points to a list of IDs of the z3950 targets to be queried (see z3950servers table).
524 This code is used in cataloging/z3950_auth_search.
525 The second parameter $template is a Template object. The routine uses this parameter to store the found values into the template.
529 sub Z3950SearchAuth {
530 my ($pars, $template)= @_;
532 my $dbh = C4::Context->dbh;
533 my @id= @{$pars->{id}};
534 my $page= $pars->{page};
542 my @breeding_loop = ();
551 my $marcflavour = C4::Context->preference('marcflavour');
552 my $marc_type = $marcflavour eq 'UNIMARC' ? 'UNIMARCAUTH' : $marcflavour;
553 my $authid= $pars->{authid};
554 my ( $zquery, $squery ) = _auth_build_query( $pars );
555 foreach my $servid (@id) {
556 my $sth = $dbh->prepare("select * from z3950servers where id=?");
557 $sth->execute($servid);
558 while ( my $server = $sth->fetchrow_hashref ) {
559 $oConnection[$s] = _create_connection( $server );
561 if ( $server->{servertype} eq 'zed' ) {
562 my $server_zquery = $zquery;
563 if ( my $attributes = $server->{attributes} ) {
564 $server_zquery = "$attributes $zquery";
566 $oResult[$s] = $oConnection[$s]->search_pqf( $server_zquery );
569 $oResult[$s] = $oConnection[$s]->search(
570 ZOOM::Query::CQL->new(_translate_query( $server, $squery ))
573 $encoding[$s] = ($server->{encoding}?$server->{encoding}:"iso-5426");
574 $servers[$s] = $server;
580 while ( $nremaining-- ) {
583 while ( ( $k = ZOOM::event( \@oConnection ) ) != 0 ) {
584 $event = $oConnection[ $k - 1 ]->last_event();
585 last if $event == ZOOM::Event::ZEND;
590 my ($error )= $oConnection[$k]->error_x(); #ignores errmsg, addinfo, diagset
592 if ($error =~ m/^(10000|10007)$/ ) {
593 push(@errconn, {'server' => $serverhost[$k]});
597 my $numresults = $oResult[$k]->size();
600 if ( $numresults > 0 and $numresults >= (($page-1)*20)) {
601 $show_next = 1 if $numresults >= ($page*20);
602 $total_pages = int($numresults/20)+1 if $total_pages < ($numresults/20);
603 for ($i = ($page-1)*20; $i < (($numresults < ($page*20)) ? $numresults : ($page*20)); $i++) {
604 my $rec = $oResult[$k]->record($i);
608 $marcdata = $rec->raw();
610 my ($charset_result, $charset_errors);
611 if( $servers[$k]->{servertype} eq 'sru' ) {
612 $marcrecord = MARC::Record->new_from_xml( $marcdata, 'UTF-8', $servers[$k]->{syntax} );
613 $marcrecord->encoding('UTF-8');
615 ( $marcrecord, $charset_result, $charset_errors ) = MarcToUTF8Record( $marcdata, $marc_type, $encoding[$k] );
618 my $heading_authtype_code;
619 $heading_authtype_code = GuessAuthTypeCode($marcrecord);
620 next if ( not defined $heading_authtype_code ) ;
622 $heading = C4::AuthoritiesMarc::GetAuthorizedHeading({ record => $marcrecord });
624 my $breedingid = ImportBreedingAuth( $marcrecord, $serverhost[$k], $encoding[$k], $heading );
626 $row_data{server} = $servers[$k]->{'servername'};
627 $row_data{breedingid} = $breedingid;
628 $row_data{heading} = $heading;
629 $row_data{authid} = $authid;
630 $row_data{heading_code} = $heading_authtype_code;
631 push( @breeding_loop, \%row_data );
634 push(@breeding_loop,{'server'=>$servers[$k]->{'servername'},'title'=>join(': ',$oConnection[$k]->error_x()),'breedingid'=>-1,'authid'=>-1});
642 numberpending => $nremaining,
643 current_page => $page,
644 total_pages => $total_pages,
645 show_nextbutton => $show_next?1:0,
646 show_prevbutton => $page!=1,
650 #close result sets and connections
652 $oResult[$_]->destroy();
653 $oConnection[$_]->destroy();
657 foreach my $id (@id) {
658 push @servers, {id => $id};
661 breeding_loop => \@breeding_loop,
662 servers => \@servers,