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>.
27 use MARC::File::USMARC;
29 use C4::AuthoritiesMarc; #GuessAuthTypeCode, FindDuplicateAuthority
32 use Koha::XSLT_Handler;
34 use vars qw(@ISA @EXPORT @EXPORT_OK %EXPORT_TAGS);
37 # set the version for version checking
40 @EXPORT = qw(&BreedingSearch &Z3950Search &Z3950SearchAuth);
45 C4::Breeding : module to add biblios to import_records via
46 the breeding/reservoir API.
50 Z3950Search($pars, $template);
51 ($count, @results) = &BreedingSearch($title,$isbn,$random);
55 This module contains routines related to Koha's Z39.50 search into
56 cataloguing reservoir features.
60 ($count, @results) = &BreedingSearch($title,$isbn,$random);
61 C<$title> contains the title,
62 C<$isbn> contains isbn or issn,
63 C<$random> contains the random seed from a z3950 search.
65 C<$count> is the number of items in C<@results>. C<@results> is an
66 array of references-to-hash; the keys are the items from the C<import_records> and
67 C<import_biblios> tables of the Koha database.
72 my ($search,$isbn,$z3950random) = @_;
73 my $dbh = C4::Context->dbh;
79 # normalise ISBN like at import
80 $isbn = C4::Koha::GetNormalizedISBN($isbn);
82 $query = "SELECT import_record_id, file_name, isbn, title, author
84 JOIN import_records USING (import_record_id)
85 JOIN import_batches USING (import_batch_id)
88 $query .= "z3950random = ?";
92 if (defined($search) && length($search)>0) {
93 $search =~ s/(\s+)/\%/g;
94 $query .= "title like ? OR author like ?";
95 push(@bind,"%$search%", "%$search%");
97 if ($#bind!=-1 && defined($isbn) && length($isbn)>0) {
100 if (defined($isbn) && length($isbn)>0) {
101 $query .= "isbn like ?";
102 push(@bind,"$isbn%");
105 $sth = $dbh->prepare($query);
106 $sth->execute(@bind);
107 while (my $data = $sth->fetchrow_hashref) {
108 $results[$count] = $data;
109 # FIXME - hack to reflect difference in name
110 # of columns in old marc_breeding and import_records
111 # There needs to be more separation between column names and
112 # field names used in the templates </soapbox>
113 $data->{'file'} = $data->{'file_name'};
114 $data->{'id'} = $data->{'import_record_id'};
119 return($count, @results);
120 } # sub breedingsearch
125 Z3950Search($pars, $template);
127 Parameters for Z3950 search are all passed via the $pars hash. It may contain isbn, title, author, dewey, subject, lccall, controlnumber, stdid, srchany.
128 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).
129 This code is used in acqui/z3950_search and cataloging/z3950_search.
130 The second parameter $template is a Template object. The routine uses this parameter to store the found values into the template.
135 my ($pars, $template)= @_;
137 my @id= @{$pars->{id}};
138 my $page= $pars->{page};
139 my $biblionumber= $pars->{biblionumber};
144 my @breeding_loop = ();
151 my ( $zquery, $squery ) = _build_query( $pars );
153 my $schema = Koha::Database->new()->schema();
154 my $rs = $schema->resultset('Z3950server')->search(
156 { result_class => 'DBIx::Class::ResultClass::HashRefInflator' },
158 my @servers = $rs->all;
159 foreach my $server ( @servers ) {
160 $oConnection[$s] = _create_connection( $server );
162 $server->{servertype} eq 'zed'?
163 $oConnection[$s]->search_pqf( $zquery ):
164 $oConnection[$s]->search(new ZOOM::Query::CQL(
165 _translate_query( $server, $squery )));
168 my $xslh = Koha::XSLT_Handler->new;
171 while ( $nremaining-- ) {
174 while ( ( $k = ZOOM::event( \@oConnection ) ) != 0 ) {
175 $event = $oConnection[ $k - 1 ]->last_event();
176 last if $event == ZOOM::Event::ZEND;
181 my ($error)= $oConnection[$k]->error_x(); #ignores errmsg, addinfo, diagset
183 if ($error =~ m/^(10000|10007)$/ ) {
184 push(@errconn, { server => $servers[$k]->{host}, error => $error } );
188 my $numresults = $oResult[$k]->size();
191 if ( $numresults > 0 and $numresults >= (($page-1)*20)) {
192 $show_next = 1 if $numresults >= ($page*20);
193 $total_pages = int($numresults/20)+1 if $total_pages < ($numresults/20);
194 for ($i = ($page-1)*20; $i < (($numresults < ($page*20)) ? $numresults : ($page*20)); $i++) {
195 if ( $oResult[$k]->record($i) ) {
197 ( $res, $error ) = _handle_one_result( $oResult[$k]->record($i), $servers[$k], ++$imported, $biblionumber, $xslh ); #ignores error in sequence numbering
198 push @breeding_loop, $res if $res;
199 push @errconn, { server => $servers[$k]->{servername}, error => $error, seq => $i+1 } if $error;
202 push @errconn, { 'server' => $servers[$k]->{servername}, error => ( ( $oConnection[$k]->error_x() )[0] ), seq => $i+1 };
210 numberpending => $nremaining,
211 current_page => $page,
212 total_pages => $total_pages,
213 show_nextbutton => $show_next?1:0,
214 show_prevbutton => $page!=1,
218 #close result sets and connections
220 $oResult[$_]->destroy();
221 $oConnection[$_]->destroy();
225 breeding_loop => \@breeding_loop,
226 servers => \@servers,
235 isbn => '@attr 1=7 @attr 5=1 "#term" ',
236 issn => '@attr 1=8 @attr 5=1 "#term" ',
237 title => '@attr 1=4 "#term" ',
238 author => '@attr 1=1003 "#term" ',
239 dewey => '@attr 1=16 "#term" ',
240 subject => '@attr 1=21 "#term" ',
241 lccall => '@attr 1=16 @attr 2=3 @attr 3=1 @attr 4=1 @attr 5=1 '.
242 '@attr 6=1 "#term" ',
243 controlnumber => '@attr 1=12 "#term" ',
244 srchany => '@attr 1=1016 "#term" ',
245 stdid => '@attr 1=1007 "#term" ',
251 foreach my $k ( sort keys %$pars ) {
252 #note that the sort keys forces an identical result under Perl 5.18
253 #one of the unit tests is based on that assumption
254 if( ( my $val=$pars->{$k} ) && $qry_build->{$k} ) {
255 $qry_build->{$k} =~ s/#term/$val/g;
256 $zquery .= $qry_build->{$k};
257 $squery .= "[$k]=\"$val\" and ";
261 $zquery = "\@and " . $zquery for 2..$nterms;
262 $squery =~ s/ and $//;
263 return ( $zquery, $squery );
266 sub _handle_one_result {
267 my ( $zoomrec, $servhref, $seq, $bib, $xslh )= @_;
269 my $raw= $zoomrec->raw();
271 if( $servhref->{servertype} eq 'sru' ) {
272 $marcrecord= MARC::Record->new_from_xml( $raw, 'UTF-8',
273 $servhref->{syntax} );
275 ($marcrecord) = MarcToUTF8Record($raw, C4::Context->preference('marcflavour'), $servhref->{encoding} // "iso-5426" ); #ignores charset return values
277 SetUTF8Flag($marcrecord);
279 ( $marcrecord, $error ) = _do_xslt_proc($marcrecord, $servhref, $xslh);
281 my $batch_id = GetZ3950BatchId($servhref->{servername});
282 my $breedingid = AddBiblioToBatch($batch_id, $seq, $marcrecord, 'UTF-8', 0, 0);
283 #FIXME passing 0 for z3950random
284 #Will eliminate this unused field in a followup report
285 #Last zero indicates: no update for batch record counts
288 #call to TransformMarcToKoha replaced by next call
289 #we only need six fields from the marc record
293 biblionumber => $bib,
294 server => $servhref->{servername},
295 breedingid => $breedingid,
296 }, $marcrecord) if $breedingid;
297 return ( $row, $error );
301 my ( $marc, $server, $xslh ) = @_;
302 return $marc if !$server->{add_xslt};
304 my $htdocs = C4::Context->config('intrahtdocs');
305 my $theme = C4::Context->preference("template"); #staff
306 my $lang = C4::Languages::getlanguage() || 'en';
308 my @files= split ',', $server->{add_xslt};
309 my $xml = $marc->as_xml;
310 foreach my $f ( @files ) {
311 $f =~ s/^\s+//; $f =~ s/\s+$//; next if !$f;
312 $f = C4::XSLT::_get_best_default_xslt_filename(
313 $htdocs, $theme, $lang, $f ) unless $f =~ /^\//;
314 $xml = $xslh->transform( $xml, $f );
315 last if $xslh->err; #skip other files
318 return MARC::Record->new_from_xml($xml, 'UTF-8');
320 return ( $marc, 'xslt_err' ); #original record in case of errors
325 my ($row, $record)=@_;
327 title => 'biblio.title',
328 author => 'biblio.author',
329 isbn =>'biblioitems.isbn',
330 lccn =>'biblioitems.lccn', #LC control number (not call number)
331 edition =>'biblioitems.editionstatement',
332 date => 'biblio.copyrightdate', #MARC21
333 date2 => 'biblioitems.publicationyear', #UNIMARC
335 foreach my $k (keys %fetch) {
336 my ($t, $f)= split '\.', $fetch{$k};
337 $row= C4::Biblio::TransformMarcToKohaOneField($t, $f, $record, $row);
338 $row->{$k}= $row->{$f} if $k ne $f;
340 $row->{date}//= $row->{date2};
341 $row->{isbn}=_isbn_replace($row->{isbn});
347 return unless defined $isbn;
348 $isbn =~ s/ |-|\.//g;
349 $isbn =~ s/\|/ \| /g;
354 sub _create_connection {
356 my $option1= new ZOOM::Options();
357 $option1->option( 'async' => 1 );
358 $option1->option( 'elementSetName', 'F' );
359 $option1->option( 'preferredRecordSyntax', $server->{syntax} );
360 $option1->option( 'timeout', $server->{timeout} ) if $server->{timeout};
362 if( $server->{servertype} eq 'sru' ) {
363 foreach( split ',', $server->{sru_options}//'' ) {
364 #first remove surrounding spaces at comma and equals-sign
366 my @temp= split '=', $_, 2;
367 @temp= map { my $c=$_; $c=~s/^\s+|\s+$//g; $c; } @temp;
368 $option1->option( $temp[0] => $temp[1] ) if @temp;
370 } elsif( $server->{servertype} eq 'zed' ) {
371 $option1->option( 'databaseName', $server->{db} );
372 $option1->option( 'user', $server->{userid} ) if $server->{userid};
373 $option1->option( 'password', $server->{password} ) if $server->{password};
376 my $obj= ZOOM::Connection->create($option1);
377 if( $server->{servertype} eq 'sru' ) {
378 my $host= $server->{host};
379 if( $host !~ /^https?:\/\// ) {
380 #Normally, host will not be prefixed by protocol.
381 #In that case we can (safely) assume http.
382 #In case someone prefixed with https, give it a try..
383 $host = 'http://' . $host;
385 $obj->connect( $host.':'.$server->{port}.'/'.$server->{db} );
387 $obj->connect( $server->{host}, $server->{port} );
392 sub _translate_query { #SRU query adjusted per server cf. srufields column
393 my ($server, $query) = @_;
395 #sru_fields is in format title=field,isbn=field,...
396 #if a field doesn't exist, try anywhere or remove [field]=
397 my @parts= split(',', $server->{sru_fields} );
398 my %trans= map { if( /=/ ) { ( $`,$' ) } else { () } } @parts;
399 my $any= $trans{srchany}?$trans{srchany}.'=':'';
402 foreach my $key (keys %trans) {
407 $q=~s/\[$key\]=/$any/g;
410 $q=~s/\[\w+\]=/$any/g; # remove remaining fields (not found in field list)
414 =head2 ImportBreedingAuth
416 ImportBreedingAuth($marcrecords,$overwrite_auth,$filename,$encoding,$z3950random,$batch_type);
418 ImportBreedingAuth imports MARC records in the reservoir (import_records table).
419 ImportBreedingAuth is based on the ImportBreeding subroutine.
423 sub ImportBreedingAuth {
424 my ($marcrecords,$overwrite_auth,$filename,$encoding,$z3950random,$batch_type) = @_;
425 my @marcarray = split /\x1D/, $marcrecords;
427 my $dbh = C4::Context->dbh;
429 my $batch_id = GetZ3950BatchId($filename);
430 my $searchbreeding = $dbh->prepare("select import_record_id from import_auths where control_number=? and authorized_heading=?");
432 my $marcflavour = C4::Context->preference('marcflavour');
433 my $marc_type = $marcflavour eq 'UNIMARC' ? 'UNIMARCAUTH' : $marcflavour;
435 # fields used for import results
438 my $alreadyinfarm = 0;
439 my $notmarcrecord = 0;
441 for (my $i=0;$i<=$#marcarray;$i++) {
442 my ($marcrecord, $charset_result, $charset_errors);
443 ($marcrecord, $charset_result, $charset_errors) =
444 MarcToUTF8Record($marcarray[$i]."\x1D", $marc_type, $encoding);
446 # Normalize the record so it doesn't have separated diacritics
447 SetUTF8Flag($marcrecord);
449 if (scalar($marcrecord->fields()) == 0) {
453 $heading = C4::AuthoritiesMarc::GetAuthorizedHeading({ record => $marcrecord });
455 my $heading_authtype_code;
456 $heading_authtype_code = GuessAuthTypeCode($marcrecord);
459 $controlnumber = $marcrecord->field('001')->data;
461 #Check if the authority record already exists in the database...
462 my ($duplicateauthid,$duplicateauthvalue);
463 if ($marcrecord && $heading_authtype_code) {
464 ($duplicateauthid,$duplicateauthvalue) = FindDuplicateAuthority( $marcrecord, $heading_authtype_code);
467 if ($duplicateauthid && $overwrite_auth ne 2) {
468 #If the authority record exists and $overwrite_auth doesn't equal 2, then mark it as already in the DB
471 if ($controlnumber && $heading) {
472 $searchbreeding->execute($controlnumber,$heading);
473 ($breedingid) = $searchbreeding->fetchrow;
475 if ($breedingid && $overwrite_auth eq '0') {
478 if ($breedingid && $overwrite_auth eq '1') {
479 ModAuthorityInBatch($breedingid, $marcrecord);
481 my $import_id = AddAuthToBatch($batch_id, $imported, $marcrecord, $encoding, $z3950random);
482 $breedingid = $import_id;
489 return ($notmarcrecord,$alreadyindb,$alreadyinfarm,$imported,$breedingid);
492 =head2 Z3950SearchAuth
494 Z3950SearchAuth($pars, $template);
496 Parameters for Z3950 search are all passed via the $pars hash. It may contain nameany, namepersonal, namecorp, namemeetingcon,
497 title, uniform title, subject, subjectsubdiv, srchany.
498 Also it should contain an arrayref id that points to a list of IDs of the z3950 targets to be queried (see z3950servers table).
499 This code is used in cataloging/z3950_auth_search.
500 The second parameter $template is a Template object. The routine uses this parameter to store the found values into the template.
504 sub Z3950SearchAuth {
505 my ($pars, $template)= @_;
507 my $dbh = C4::Context->dbh;
508 my @id= @{$pars->{id}};
509 my $random= $pars->{random};
510 my $page= $pars->{page};
512 my $nameany= $pars->{nameany};
513 my $authorany= $pars->{authorany};
514 my $authorpersonal= $pars->{authorpersonal};
515 my $authorcorp= $pars->{authorcorp};
516 my $authormeetingcon= $pars->{authormeetingcon};
517 my $title= $pars->{title};
518 my $uniformtitle= $pars->{uniformtitle};
519 my $subject= $pars->{subject};
520 my $subjectsubdiv= $pars->{subjectsubdiv};
521 my $srchany= $pars->{srchany};
522 my $authid= $pars->{authid};
538 my @breeding_loop = ();
547 my $marcflavour = C4::Context->preference('marcflavour');
548 my $marc_type = $marcflavour eq 'UNIMARC' ? 'UNIMARCAUTH' : $marcflavour;
551 $query .= " \@attr 1=1002 \"$nameany\" "; #Any name (this includes personal, corporate, meeting/conference authors, and author names in subject headings)
552 #This attribute is supported by both the Library of Congress and Libraries Australia 08/05/2013
557 $query .= " \@attr 1=1003 \"$authorany\" "; #Author-name (this includes personal, corporate, meeting/conference authors, but not author names in subject headings)
558 #This attribute is not supported by the Library of Congress, but is supported by Libraries Australia 08/05/2013
563 $query .= " \@attr 1=2 \"$authorcorp\" "; #1005 is another valid corporate author attribute...
567 if ($authorpersonal) {
568 $query .= " \@attr 1=1 \"$authorpersonal\" "; #1004 is another valid personal name attribute...
572 if ($authormeetingcon) {
573 $query .= " \@attr 1=3 \"$authormeetingcon\" "; #1006 is another valid meeting/conference name attribute...
578 $query .= " \@attr 1=21 \"$subject\" ";
582 if ($subjectsubdiv) {
583 $query .= " \@attr 1=47 \"$subjectsubdiv\" ";
588 $query .= " \@attr 1=4 \"$title\" "; #This is a regular title search. 1=6 will give just uniform titles
593 $query .= " \@attr 1=6 \"$uniformtitle\" "; #This is the uniform title search
598 $query .= " \@attr 1=1016 \"$srchany\" ";
602 for my $i (1..$nterms-1) {
603 $query = "\@and " . $query;
606 foreach my $servid (@id) {
607 my $sth = $dbh->prepare("select * from z3950servers where id=?");
608 $sth->execute($servid);
609 while ( $server = $sth->fetchrow_hashref ) {
610 my $option1 = new ZOOM::Options();
611 $option1->option( 'async' => 1 );
612 $option1->option( 'elementSetName', 'F' );
613 $option1->option( 'databaseName', $server->{db} );
614 $option1->option( 'user', $server->{userid} ) if $server->{userid};
615 $option1->option( 'password', $server->{password} ) if $server->{password};
616 $option1->option( 'preferredRecordSyntax', $server->{syntax} );
617 $option1->option( 'timeout', $server->{timeout} ) if $server->{timeout};
618 $oConnection[$s] = create ZOOM::Connection($option1);
619 $oConnection[$s]->connect( $server->{host}, $server->{port} );
620 $serverhost[$s] = $server->{host};
621 $servername[$s] = $server->{servername};
622 $encoding[$s] = ($server->{encoding}?$server->{encoding}:"iso-5426");
628 for ( my $z = 0 ; $z < $s ; $z++ ) {
629 $oResult[$z] = $oConnection[$z]->search_pqf($query);
632 while ( $nremaining-- ) {
635 while ( ( $k = ZOOM::event( \@oConnection ) ) != 0 ) {
636 $event = $oConnection[ $k - 1 ]->last_event();
637 last if $event == ZOOM::Event::ZEND;
642 my ($error, $errmsg, $addinfo, $diagset)= $oConnection[$k]->error_x();
644 if ($error =~ m/^(10000|10007)$/ ) {
645 push(@errconn, {'server' => $serverhost[$k]});
649 my $numresults = $oResult[$k]->size();
652 if ( $numresults > 0 and $numresults >= (($page-1)*20)) {
653 $show_next = 1 if $numresults >= ($page*20);
654 $total_pages = int($numresults/20)+1 if $total_pages < ($numresults/20);
655 for ($i = ($page-1)*20; $i < (($numresults < ($page*20)) ? $numresults : ($page*20)); $i++) {
656 my $rec = $oResult[$k]->record($i);
660 $marcdata = $rec->raw();
662 my ($charset_result, $charset_errors);
663 ($marcrecord, $charset_result, $charset_errors)= MarcToUTF8Record($marcdata, $marc_type, $encoding[$k]);
666 my $heading_authtype_code;
667 $heading_authtype_code = GuessAuthTypeCode($marcrecord);
668 $heading = C4::AuthoritiesMarc::GetAuthorizedHeading({ record => $marcrecord });
670 my ($notmarcrecord, $alreadyindb, $alreadyinfarm, $imported, $breedingid)= ImportBreedingAuth( $marcdata, 2, $serverhost[$k], $encoding[$k], $random, 'z3950' );
672 $row_data{server} = $servername[$k];
673 $row_data{breedingid} = $breedingid;
674 $row_data{heading} = $heading;
675 $row_data{authid} = $authid;
676 $row_data{heading_code} = $heading_authtype_code;
677 push( @breeding_loop, \%row_data );
680 push(@breeding_loop,{'server'=>$servername[$k],'title'=>join(': ',$oConnection[$k]->error_x()),'breedingid'=>-1,'authid'=>-1});
688 numberpending => $nremaining,
689 current_page => $page,
690 total_pages => $total_pages,
691 show_nextbutton => $show_next?1:0,
692 show_prevbutton => $page!=1,
696 #close result sets and connections
698 $oResult[$_]->destroy();
699 $oConnection[$_]->destroy();
703 foreach my $id (@id) {
704 push @servers, {id => $id};
707 breeding_loop => \@breeding_loop,
708 servers => \@servers,