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($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS);
37 # set the version for version checking
38 $VERSION = 3.07.00.049;
41 @EXPORT = qw(&BreedingSearch &Z3950Search &Z3950SearchAuth);
46 C4::Breeding : module to add biblios to import_records via
47 the breeding/reservoir API.
51 Z3950Search($pars, $template);
52 ($count, @results) = &BreedingSearch($title,$isbn,$random);
56 This module contains routines related to Koha's Z39.50 search into
57 cataloguing reservoir features.
61 ($count, @results) = &BreedingSearch($title,$isbn,$random);
62 C<$title> contains the title,
63 C<$isbn> contains isbn or issn,
64 C<$random> contains the random seed from a z3950 search.
66 C<$count> is the number of items in C<@results>. C<@results> is an
67 array of references-to-hash; the keys are the items from the C<import_records> and
68 C<import_biblios> tables of the Koha database.
73 my ($search,$isbn,$z3950random) = @_;
74 my $dbh = C4
::Context
->dbh;
80 # normalise ISBN like at import
81 $isbn = C4
::Koha
::GetNormalizedISBN
($isbn);
83 $query = "SELECT import_record_id, file_name, isbn, title, author
85 JOIN import_records USING (import_record_id)
86 JOIN import_batches USING (import_batch_id)
89 $query .= "z3950random = ?";
93 if (defined($search) && length($search)>0) {
94 $search =~ s/(\s+)/\%/g;
95 $query .= "title like ? OR author like ?";
96 push(@bind,"%$search%", "%$search%");
98 if ($#bind!=-1 && defined($isbn) && length($isbn)>0) {
101 if (defined($isbn) && length($isbn)>0) {
102 $query .= "isbn like ?";
103 push(@bind,"$isbn%");
106 $sth = $dbh->prepare($query);
107 $sth->execute(@bind);
108 while (my $data = $sth->fetchrow_hashref) {
109 $results[$count] = $data;
110 # FIXME - hack to reflect difference in name
111 # of columns in old marc_breeding and import_records
112 # There needs to be more separation between column names and
113 # field names used in the templates </soapbox>
114 $data->{'file'} = $data->{'file_name'};
115 $data->{'id'} = $data->{'import_record_id'};
120 return($count, @results);
121 } # sub breedingsearch
126 Z3950Search($pars, $template);
128 Parameters for Z3950 search are all passed via the $pars hash. It may contain isbn, title, author, dewey, subject, lccall, controlnumber, stdid, srchany.
129 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).
130 This code is used in acqui/z3950_search and cataloging/z3950_search.
131 The second parameter $template is a Template object. The routine uses this parameter to store the found values into the template.
136 my ($pars, $template)= @_;
138 my @id= @
{$pars->{id
}};
139 my $page= $pars->{page
};
140 my $biblionumber= $pars->{biblionumber
};
145 my @breeding_loop = ();
152 my ( $zquery, $squery ) = _build_query
( $pars );
154 my $schema = Koha
::Database
->new()->schema();
155 my $rs = $schema->resultset('Z3950server')->search(
157 { result_class
=> 'DBIx::Class::ResultClass::HashRefInflator' },
159 my @servers = $rs->all;
160 foreach my $server ( @servers ) {
161 $oConnection[$s] = _create_connection
( $server );
163 $server->{servertype
} eq 'zed'?
164 $oConnection[$s]->search_pqf( $zquery ):
165 $oConnection[$s]->search(new ZOOM
::Query
::CQL
(
166 _translate_query
( $server, $squery )));
169 my $xslh = Koha
::XSLT_Handler
->new;
172 while ( $nremaining-- ) {
175 while ( ( $k = ZOOM
::event
( \
@oConnection ) ) != 0 ) {
176 $event = $oConnection[ $k - 1 ]->last_event();
177 last if $event == ZOOM
::Event
::ZEND
;
182 my ($error)= $oConnection[$k]->error_x(); #ignores errmsg, addinfo, diagset
184 if ($error =~ m/^(10000|10007)$/ ) {
185 push(@errconn, { server
=> $servers[$k]->{host
}, error
=> $error } );
189 my $numresults = $oResult[$k]->size();
192 if ( $numresults > 0 and $numresults >= (($page-1)*20)) {
193 $show_next = 1 if $numresults >= ($page*20);
194 $total_pages = int($numresults/20)+1 if $total_pages < ($numresults/20);
195 for ($i = ($page-1)*20; $i < (($numresults < ($page*20)) ?
$numresults : ($page*20)); $i++) {
196 if ( $oResult[$k]->record($i) ) {
198 ( $res, $error ) = _handle_one_result
( $oResult[$k]->record($i), $servers[$k], ++$imported, $biblionumber, $xslh ); #ignores error in sequence numbering
199 push @breeding_loop, $res if $res;
200 push @errconn, { server
=> $servers[$k]->{servername
}, error
=> $error, seq
=> $i+1 } if $error;
203 push @errconn, { 'server' => $servers[$k]->{servername
}, error
=> ( ( $oConnection[$k]->error_x() )[0] ), seq
=> $i+1 };
211 numberpending
=> $nremaining,
212 current_page
=> $page,
213 total_pages
=> $total_pages,
214 show_nextbutton
=> $show_next?
1:0,
215 show_prevbutton
=> $page!=1,
219 #close result sets and connections
221 $oResult[$_]->destroy();
222 $oConnection[$_]->destroy();
226 breeding_loop
=> \
@breeding_loop,
227 servers
=> \
@servers,
236 isbn
=> '@attr 1=7 @attr 5=1 "#term" ',
237 issn
=> '@attr 1=8 @attr 5=1 "#term" ',
238 title
=> '@attr 1=4 "#term" ',
239 author
=> '@attr 1=1003 "#term" ',
240 dewey
=> '@attr 1=16 "#term" ',
241 subject
=> '@attr 1=21 "#term" ',
242 lccall
=> '@attr 1=16 @attr 2=3 @attr 3=1 @attr 4=1 @attr 5=1 '.
243 '@attr 6=1 "#term" ',
244 controlnumber
=> '@attr 1=12 "#term" ',
245 srchany
=> '@attr 1=1016 "#term" ',
246 stdid
=> '@attr 1=1007 "#term" ',
252 foreach my $k ( sort keys %$pars ) {
253 #note that the sort keys forces an identical result under Perl 5.18
254 #one of the unit tests is based on that assumption
255 if( ( my $val=$pars->{$k} ) && $qry_build->{$k} ) {
256 $qry_build->{$k} =~ s/#term/$val/g;
257 $zquery .= $qry_build->{$k};
258 $squery .= "[$k]=\"$val\" and ";
262 $zquery = "\@and " . $zquery for 2..$nterms;
263 $squery =~ s/ and $//;
264 return ( $zquery, $squery );
267 sub _handle_one_result
{
268 my ( $zoomrec, $servhref, $seq, $bib, $xslh )= @_;
270 my $raw= $zoomrec->raw();
272 if( $servhref->{servertype
} eq 'sru' ) {
273 $marcrecord= MARC
::Record
->new_from_xml( $raw, 'UTF-8',
274 $servhref->{syntax
} );
276 ($marcrecord) = MarcToUTF8Record
($raw, C4
::Context
->preference('marcflavour'), $servhref->{encoding
} // "iso-5426" ); #ignores charset return values
278 SetUTF8Flag
($marcrecord);
280 ( $marcrecord, $error ) = _do_xslt_proc
($marcrecord, $servhref, $xslh);
282 my $batch_id = GetZ3950BatchId
($servhref->{servername
});
283 my $breedingid = AddBiblioToBatch
($batch_id, $seq, $marcrecord, 'UTF-8', 0, 0);
284 #FIXME passing 0 for z3950random
285 #Will eliminate this unused field in a followup report
286 #Last zero indicates: no update for batch record counts
289 #call to TransformMarcToKoha replaced by next call
290 #we only need six fields from the marc record
294 biblionumber
=> $bib,
295 server
=> $servhref->{servername
},
296 breedingid
=> $breedingid,
297 }, $marcrecord) if $breedingid;
298 return ( $row, $error );
302 my ( $marc, $server, $xslh ) = @_;
303 return $marc if !$server->{add_xslt
};
305 my $htdocs = C4
::Context
->config('intrahtdocs');
306 my $theme = C4
::Context
->preference("template"); #staff
307 my $lang = C4
::Languages
::getlanguage
() || 'en';
309 my @files= split ',', $server->{add_xslt
};
310 my $xml = $marc->as_xml;
311 foreach my $f ( @files ) {
312 $f =~ s/^\s+//; $f =~ s/\s+$//; next if !$f;
313 $f = C4
::XSLT
::_get_best_default_xslt_filename
(
314 $htdocs, $theme, $lang, $f ) unless $f =~ /^\//;
315 $xml = $xslh->transform( $xml, $f );
316 last if $xslh->err; #skip other files
319 return MARC
::Record
->new_from_xml($xml, 'UTF-8');
321 return ( $marc, 'xslt_err' ); #original record in case of errors
326 my ($row, $record)=@_;
328 title
=> 'biblio.title',
329 author
=> 'biblio.author',
330 isbn
=>'biblioitems.isbn',
331 lccn
=>'biblioitems.lccn', #LC control number (not call number)
332 edition
=>'biblioitems.editionstatement',
333 date
=> 'biblio.copyrightdate', #MARC21
334 date2
=> 'biblioitems.publicationyear', #UNIMARC
336 foreach my $k (keys %fetch) {
337 my ($t, $f)= split '\.', $fetch{$k};
338 $row= C4
::Biblio
::TransformMarcToKohaOneField
($t, $f, $record, $row);
339 $row->{$k}= $row->{$f} if $k ne $f;
341 $row->{date
}//= $row->{date2
};
342 $row->{isbn
}=_isbn_replace
($row->{isbn
});
348 return unless defined $isbn;
349 $isbn =~ s/ |-|\.//g;
350 $isbn =~ s/\|/ \| /g;
355 sub _create_connection
{
357 my $option1= new ZOOM
::Options
();
358 $option1->option( 'async' => 1 );
359 $option1->option( 'elementSetName', 'F' );
360 $option1->option( 'preferredRecordSyntax', $server->{syntax
} );
361 $option1->option( 'timeout', $server->{timeout
} ) if $server->{timeout
};
363 if( $server->{servertype
} eq 'sru' ) {
364 foreach( split ',', $server->{sru_options
}//'' ) {
365 #first remove surrounding spaces at comma and equals-sign
367 my @temp= split '=', $_, 2;
368 @temp= map { my $c=$_; $c=~s/^\s+|\s+$//g; $c; } @temp;
369 $option1->option( $temp[0] => $temp[1] ) if @temp;
371 } elsif( $server->{servertype
} eq 'zed' ) {
372 $option1->option( 'databaseName', $server->{db
} );
373 $option1->option( 'user', $server->{userid
} ) if $server->{userid
};
374 $option1->option( 'password', $server->{password
} ) if $server->{password
};
377 my $obj= ZOOM
::Connection
->create($option1);
378 if( $server->{servertype
} eq 'sru' ) {
379 my $host= $server->{host
};
380 if( $host !~ /^https?:\/\
// ) {
381 #Normally, host will not be prefixed by protocol.
382 #In that case we can (safely) assume http.
383 #In case someone prefixed with https, give it a try..
384 $host = 'http://' . $host;
386 $obj->connect( $host.':'.$server->{port
}.'/'.$server->{db
} );
388 $obj->connect( $server->{host
}, $server->{port
} );
393 sub _translate_query
{ #SRU query adjusted per server cf. srufields column
394 my ($server, $query) = @_;
396 #sru_fields is in format title=field,isbn=field,...
397 #if a field doesn't exist, try anywhere or remove [field]=
398 my @parts= split(',', $server->{sru_fields
} );
399 my %trans= map { if( /=/ ) { ( $`,$' ) } else { () } } @parts;
400 my $any= $trans{srchany}?$trans{srchany}.'=':'';
403 foreach my $key (keys %trans) {
408 $q=~s/\[$key\]=/$any/g;
411 $q=~s/\[\w+\]=/$any/g; # remove remaining fields (not found in field list)
415 =head2 ImportBreedingAuth
417 ImportBreedingAuth($marcrecords,$overwrite_auth,$filename,$encoding,$z3950random,$batch_type);
419 ImportBreedingAuth imports MARC records in the reservoir (import_records table).
420 ImportBreedingAuth is based on the ImportBreeding subroutine.
424 sub ImportBreedingAuth {
425 my ($marcrecords,$overwrite_auth,$filename,$encoding,$z3950random,$batch_type) = @_;
426 my @marcarray = split /\x1D/, $marcrecords;
428 my $dbh = C4::Context->dbh;
430 my $batch_id = GetZ3950BatchId($filename);
431 my $searchbreeding = $dbh->prepare("select import_record_id from import_auths where control_number=? and authorized_heading=?");
433 my $marcflavour = C4::Context->preference('marcflavour');
434 my $marc_type = $marcflavour eq 'UNIMARC' ? 'UNIMARCAUTH' : $marcflavour;
436 # fields used for import results
439 my $alreadyinfarm = 0;
440 my $notmarcrecord = 0;
442 for (my $i=0;$i<=$#marcarray;$i++) {
443 my ($marcrecord, $charset_result, $charset_errors);
444 ($marcrecord, $charset_result, $charset_errors) =
445 MarcToUTF8Record($marcarray[$i]."\x1D", $marc_type, $encoding);
447 # Normalize the record so it doesn't have separated diacritics
448 SetUTF8Flag($marcrecord);
450 if (scalar($marcrecord->fields()) == 0) {
454 $heading = C4::AuthoritiesMarc::GetAuthorizedHeading({ record => $marcrecord });
456 my $heading_authtype_code;
457 $heading_authtype_code = GuessAuthTypeCode($marcrecord);
460 $controlnumber = $marcrecord->field('001')->data;
462 #Check if the authority record already exists in the database...
463 my ($duplicateauthid,$duplicateauthvalue);
464 if ($marcrecord && $heading_authtype_code) {
465 ($duplicateauthid,$duplicateauthvalue) = FindDuplicateAuthority( $marcrecord, $heading_authtype_code);
468 if ($duplicateauthid && $overwrite_auth ne 2) {
469 #If the authority record exists and $overwrite_auth doesn't equal 2, then mark it as already in the DB
472 if ($controlnumber && $heading) {
473 $searchbreeding->execute($controlnumber,$heading);
474 ($breedingid) = $searchbreeding->fetchrow;
476 if ($breedingid && $overwrite_auth eq '0') {
479 if ($breedingid && $overwrite_auth eq '1') {
480 ModAuthorityInBatch($breedingid, $marcrecord);
482 my $import_id = AddAuthToBatch($batch_id, $imported, $marcrecord, $encoding, $z3950random);
483 $breedingid = $import_id;
490 return ($notmarcrecord,$alreadyindb,$alreadyinfarm,$imported,$breedingid);
493 =head2 Z3950SearchAuth
495 Z3950SearchAuth($pars, $template);
497 Parameters for Z3950 search are all passed via the $pars hash. It may contain nameany, namepersonal, namecorp, namemeetingcon,
498 title, uniform title, subject, subjectsubdiv, srchany.
499 Also it should contain an arrayref id that points to a list of IDs of the z3950 targets to be queried (see z3950servers table).
500 This code is used in cataloging/z3950_auth_search.
501 The second parameter $template is a Template object. The routine uses this parameter to store the found values into the template.
505 sub Z3950SearchAuth {
506 my ($pars, $template)= @_;
508 my $dbh = C4::Context->dbh;
509 my @id= @{$pars->{id}};
510 my $random= $pars->{random};
511 my $page= $pars->{page};
513 my $nameany= $pars->{nameany};
514 my $authorany= $pars->{authorany};
515 my $authorpersonal= $pars->{authorpersonal};
516 my $authorcorp= $pars->{authorcorp};
517 my $authormeetingcon= $pars->{authormeetingcon};
518 my $title= $pars->{title};
519 my $uniformtitle= $pars->{uniformtitle};
520 my $subject= $pars->{subject};
521 my $subjectsubdiv= $pars->{subjectsubdiv};
522 my $srchany= $pars->{srchany};
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->{name};
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{heading_code} = $heading_authtype_code;
676 push( @breeding_loop, \%row_data );
679 push(@breeding_loop,{'server'=>$servername[$k],'title'=>join(': ',$oConnection[$k]->error_x()),'breedingid'=>-1});
687 numberpending => $nremaining,
688 current_page => $page,
689 total_pages => $total_pages,
690 show_nextbutton => $show_next?1:0,
691 show_prevbutton => $page!=1,
695 #close result sets and connections
697 $oResult[$_]->destroy();
698 $oConnection[$_]->destroy();
702 foreach my $id (@id) {
703 push @servers, {id => $id};
706 breeding_loop => \@breeding_loop,
707 servers => \@servers,