1 package C4::SearchMarc;
3 # Copyright 2000-2002 Katipo Communications
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
28 use vars qw($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS);
30 # set the version for version checking
35 C4::Search - Functions for searching the Koha MARC catalog
39 This module provides the searching facilities for the Koha MARC catalog
44 @EXPORT = qw(&catalogsearch &findseealso &findsuggestion &getMARCnotes &getMARCsubjects);
46 =head1 findsuggestion($dbh,$values);
48 =head2 $dbh is a link to the DB handler.
51 my $dbh =C4::Context->dbh;
53 =head2 $values is a word
55 Searches words with the same soundex, ordered by frequency of use.
56 Useful to suggest other searches to the users.
61 my ($dbh,$values) = @_;
62 my $sth = $dbh->prepare("SELECT count( * ) AS total, word FROM marc_word WHERE sndx_word = soundex( ? ) AND word <> ? GROUP BY word ORDER BY total DESC");
64 for(my $i = 0 ; $i <= $#{$values} ; $i++) {
65 if (length(@$values[$i]) >=5) {
66 $sth->execute(@$values[$i],@$values[$i]);
69 while ((my ($count,$word) = $sth->fetchrow) and $resfound <=10) {
70 push @results, "@$values[$i]|$word|$count";
71 # $results{@$values[$i]} = \@resline;
79 =head1 findseealso($dbh,$fields);
81 =head2 $dbh is a link to the DB handler.
84 my $dbh =C4::Context->dbh;
86 =head2 $fields is a reference to the fields array
88 This function modify the @$fields array and add related fields to search on.
93 my ($dbh, $fields) = @_;
94 my $tagslib = MARCgettagslib ($dbh,1);
95 for (my $i=0;$i<=$#{$fields};$i++) {
96 my ($tag) =substr(@$fields[$i],1,3);
97 my ($subfield) =substr(@$fields[$i],4,1);
98 @$fields[$i].=','.$tagslib->{$tag}->{$subfield}->{seealso} if ($tagslib->{$tag}->{$subfield}->{seealso});
102 =head1 my ($count, @results) = catalogsearch($dbh, $tags, $and_or, $excluding, $operator, $value, $offset,$length,$orderby);
104 =head2 $dbh is a link to the DB handler.
107 my $dbh =C4::Context->dbh;
109 $tags,$and_or, $excluding, $operator, $value are references to array
113 contains the list of tags+subfields (for example : $@tags[0] = '200a')
114 A field can be a list of fields : '200f','700a','700b','701a','701b'
120 contains a list of strings containing and or or. The 1st value is useless.
124 contains 0 or 1. If 1, then the request is negated.
128 contains contains,=,start,>,>=,<,<= the = and start work on the complete subfield. The contains operator works on every word in the subfield.
131 contains home, search home anywhere.
132 = home, search a string being home.
136 contains the value to search
137 If it contains a * or a %, then the search is partial.
139 =head2 $offset and $length
141 returns $length results, beginning at $offset
145 define the field used to order the request. Any field in the biblio/biblioitem tables can be used. DESC is possible too
147 (for example title, title DESC,...)
151 returns an array containing hashes. The hash contains all biblio & biblioitems fields and a reference to an item hash. The "item hash contains one line for each callnumber & the number of items related to the callnumber.
155 =head2 my $marcnotesarray = &getMARCnotes($dbh,$bibid,$marcflavour);
157 Returns a reference to an array containing all the notes stored in the MARC database for the given bibid.
158 $marcflavour ("MARC21" or "UNIMARC") determines which tags are used for retrieving subjects.
160 =head2 my $marcsubjctsarray = &getMARCsubjects($dbh,$bibid,$marcflavour);
162 Returns a reference to an array containing all the subjects stored in the MARC database for the given bibid.
163 $marcflavour ("MARC21" or "UNIMARC") determines which tags are used for retrieving subjects.
168 my ($dbh, $tags, $and_or, $excluding, $operator, $value, $offset,$length,$orderby,$desc_or_asc) = @_;
169 # build the sql request. She will look like :
171 # from marc_subfield_table as m1, marc_subfield_table as m2
172 # where m1.bibid=m2.bibid and
173 # (m1.subfieldvalue like "Des%" and m2.subfieldvalue like "27%")
175 # last minute stripping out of stuff
176 # doesn't work @$value =~ s/\'/ /;
177 # @$value = map { $_ =~ s/\'/ /g } @$value;
179 # "Normal" statements
180 my @normal_tags = ();
181 my @normal_and_or = ();
182 my @normal_operator = ();
183 my @normal_value = ();
184 # Extracts the NOT statements from the list of statements
187 my @not_operator = ();
190 $orderby = "biblio.title" unless $orderby;
191 $desc_or_asc = "ASC" unless $desc_or_asc;
192 #last minute stripping out of ' and ,
193 # paul : quoting, it's done a few lines lated.
194 # foreach $_ (@$value) {
199 # the item.notforloan contains an integer. Every value <>0 means "book unavailable for loan".
200 # but each library can have it's own table of meaning for each value. Get them
201 # 1st search if there is a list of authorised values connected to items.notforloan
202 my $sth = $dbh->prepare('select authorised_value from marc_subfield_structure where kohafield="items.notforloan"');
204 my %notforloanstatus;
205 my ($authorised_valuecode) = $sth->fetchrow;
206 if ($authorised_valuecode) {
207 $sth = $dbh->prepare("select authorised_value,lib from authorised_values where category=?");
208 $sth->execute($authorised_valuecode);
209 while (my ($authorised_value,$lib) = $sth->fetchrow) {
210 $notforloanstatus{$authorised_value} = $lib?$lib:$authorised_value;
213 for(my $i = 0 ; $i <= $#{$value} ; $i++)
216 @$value[$i] =~ s/\*/%/g;
217 # remove % at the beginning
218 @$value[$i] =~ s/^%//g;
219 @$value[$i] =~ s/(\.|\?|\:|\!|\'|,|\-|\"|\(|\)|\[|\]|\{|\}|\/)/ /g if @$operator[$i] eq "contains";
220 if(@$excluding[$i]) # NOT statements
223 if(@$operator[$i] eq "contains")
225 foreach my $word (split(/ /, @$value[$i])) # if operator is contains, splits the words in separate requests
227 # remove the "%" for small word (3 letters. (note : the >4 is due to the % at the end)
228 # warn "word : $word";
229 $word =~ s/%//g unless length($word)>4;
230 unless (C4::Context->stopwords->{uc($word)} or length($word)==1) { #it's NOT a stopword => use it. Otherwise, ignore
231 push @not_tags, @$tags[$i];
232 push @not_and_or, "or"; # as request is negated, finds "foo" or "bar" if final request is NOT "foo" and "bar"
233 push @not_operator, @$operator[$i];
234 push @not_value, $word;
240 push @not_tags, @$tags[$i];
241 push @not_and_or, "or"; # as request is negated, finds "foo" or "bar" if final request is NOT "foo" and "bar"
242 push @not_operator, @$operator[$i];
243 push @not_value, @$value[$i];
246 else # NORMAL statements
248 if(@$operator[$i] eq "contains") # if operator is contains, splits the words in separate requests
250 foreach my $word (split(/ /, @$value[$i]))
252 # remove the "%" for small word (3 letters. (note : the >4 is due to the % at the end)
253 # warn "word : $word";
254 $word =~ s/%//g unless length($word)>4;
255 unless (C4::Context->stopwords->{uc($word)} or length($word)==1) { #it's NOT a stopword => use it. Otherwise, ignore
256 push @normal_tags, @$tags[$i];
257 push @normal_and_or, "and"; # assumes "foo" and "bar" if "foo bar" is entered
258 push @normal_operator, @$operator[$i];
259 push @normal_value, $word;
265 push @normal_tags, @$tags[$i];
266 push @normal_and_or, @$and_or[$i];
267 push @normal_operator, @$operator[$i];
268 push @normal_value, @$value[$i];
273 # Finds the basic results without the NOT requests
274 my ($sql_tables, $sql_where1, $sql_where2) = create_request($dbh,\@normal_tags, \@normal_and_or, \@normal_operator, \@normal_value);
276 $sql_where1 .= "and TO_DAYS( NOW( ) ) - TO_DAYS( biblio.timestamp ) <30" if $orderby =~ "biblio.timestamp";
279 $sth = $dbh->prepare("select distinct m1.bibid from biblio,biblioitems,marc_biblio,$sql_tables where biblio.biblionumber=marc_biblio.biblionumber and biblio.biblionumber=biblioitems.biblionumber and m1.bibid=marc_biblio.bibid and $sql_where2 and ($sql_where1) order by $orderby $desc_or_asc");
280 warn "Q2 : select distinct m1.bibid from biblio,biblioitems,marc_biblio,$sql_tables where biblio.biblionumber=marc_biblio.biblionumber and biblio.biblionumber=biblioitems.biblionumber and m1.bibid=marc_biblio.bibid and $sql_where2 and ($sql_where1) order by $orderby $desc_or_asc term is @$value";
282 $sth = $dbh->prepare("select distinct m1.bibid from biblio,biblioitems,marc_biblio,$sql_tables where biblio.biblionumber=marc_biblio.biblionumber and biblio.biblionumber=biblioitems.biblionumber and m1.bibid=marc_biblio.bibid and $sql_where1 order by $orderby $desc_or_asc");
283 warn "Q : select distinct m1.bibid from biblio,biblioitems,marc_biblio,$sql_tables where biblio.biblionumber=marc_biblio.biblionumber and biblio.biblionumber=biblioitems.biblionumber and m1.bibid=marc_biblio.bibid and $sql_where1 order by $orderby $desc_or_asc";
287 my $subtitle; # Added by JF for Subtitles
289 # Processes the NOT if any and there are results
290 my ($not_sql_tables, $not_sql_where1, $not_sql_where2);
292 if( ($sth->rows) && $any_not ) # some results to tune up and some NOT statements
294 ($not_sql_tables, $not_sql_where1, $not_sql_where2) = create_request($dbh,\@not_tags, \@not_and_or, \@not_operator, \@not_value);
298 while (my ($bibid) = $sth->fetchrow) {
299 push @tmpresult,$bibid;
302 warn "NOT : select distinct m1.bibid from $not_sql_tables where $not_sql_where2 and ($not_sql_where1)";
303 if ($not_sql_where2) {
304 $sth_not = $dbh->prepare("select distinct m1.bibid from $not_sql_tables where $not_sql_where2 and ($not_sql_where1)");
306 $sth_not = $dbh->prepare("select distinct m1.bibid from $not_sql_tables where $not_sql_where1");
313 while(my $bibid = $sth_not->fetchrow()) {
314 $not_bibids{$bibid} = 1; # populates the hashtable with the bibids matching the NOT statement
317 foreach my $bibid (@tmpresult)
319 if(!$not_bibids{$bibid})
321 push @result, $bibid;
327 else # no NOT statements
329 while (my ($bibid) = $sth->fetchrow) {
334 # we have bibid list. Now, loads title and author from [offset] to [offset]+[length]
335 my $counter = $offset;
336 # HINT : biblionumber as bn is important. The hash is fills biblionumber with items.biblionumber.
337 # so if you dont' has an item, you get a not nice empty value.
338 $sth = $dbh->prepare("SELECT biblio.biblionumber as bn,biblio.*, biblioitems.*,marc_biblio.bibid,itemtypes.notforloan,itemtypes.description
339 FROM biblio, marc_biblio
340 LEFT JOIN biblioitems on biblio.biblionumber = biblioitems.biblionumber
341 LEFT JOIN itemtypes on itemtypes.itemtype=biblioitems.itemtype
342 WHERE biblio.biblionumber = marc_biblio.biblionumber AND bibid = ?");
343 my $sth_subtitle = $dbh->prepare("SELECT subtitle FROM bibliosubtitle WHERE biblionumber=?"); # Added BY JF for Subtitles
344 my @finalresult = ();
348 my ($oldbibid, $oldauthor, $oldtitle);
349 my $sth_itemCN = $dbh->prepare("select items.* from items where biblionumber=?");
350 my $sth_issue = $dbh->prepare("select date_due,returndate from issues where itemnumber=?");
351 # parse all biblios between start & end.
352 while (($counter <= $#result) && ($counter <= ($offset + $length))) {
353 # search & parse all items & note itemcallnumber
354 $sth->execute($result[$counter]);
356 my $line = $sth->fetchrow_hashref;
357 my $biblionumber=$line->{bn};
358 # Return subtitles first ADDED BY JF
359 $sth_subtitle->execute($biblionumber);
360 my $subtitle_here.= $sth_subtitle->fetchrow." ";
362 $subtitle = $subtitle_here;
363 # warn "Here's the Biblionumber ".$biblionumber;
364 # warn "and here's the subtitle: ".$subtitle_here;
368 # $continue=0 unless $line->{bn};
369 # my $lastitemnumber;
370 $sth_itemCN->execute($biblionumber);
372 my $notforloan=1; # to see if there is at least 1 item that can be issued
373 while (my $item = $sth_itemCN->fetchrow_hashref) {
374 # parse the result, putting holdingbranch & itemcallnumber in separate array
375 # then all other fields in the main array
377 # search if item is on loan
379 $sth_issue->execute($item->{itemnumber});
380 while (my $loan = $sth_issue->fetchrow_hashref) {
381 if ($loan->{date_due} and !$loan->{returndate}) {
382 $date_due = $loan->{date_due};
387 $lineCN{holdingbranch} = $item->{holdingbranch};
388 $lineCN{itemcallnumber} = $item->{itemcallnumber};
389 $lineCN{location} = $item->{location};
390 $lineCN{date_due} = format_date($date_due);
391 $lineCN{notforloan} = $notforloanstatus{$line->{notforloan}} if ($line->{notforloan}); # setting not forloan if itemtype is not for loan
392 $lineCN{notforloan} = $notforloanstatus{$item->{notforloan}} if ($item->{notforloan}); # setting not forloan it this item is not for loan
393 $notforloan=0 unless ($item->{notforloan} or $item->{wthdrawn} or $item->{itemlost});
394 push @CNresults,\%lineCN;
397 # save the biblio in the final array, with item and item issue status
400 $newline{totitem} = $totalitems;
401 # if $totalitems == 0, check if it's being ordered.
402 if ($totalitems == 0) {
403 my $sth = $dbh->prepare("select count(*) from aqorders where biblionumber=? and datecancellationprinted is NULL");
404 $sth->execute($biblionumber);
405 my ($ordered) = $sth->fetchrow;
406 $newline{onorder} = 1 if $ordered;
408 $newline{biblionumber} = $biblionumber;
409 $newline{norequests} = 0;
410 $newline{norequests} = 1 if ($line->{notforloan}); # itemtype not issuable
411 $newline{norequests} = 1 if (!$line->{notforloan} && $notforloan); # itemtype issuable but all items not issuable for instance
412 $newline{subtitle} = $subtitle; # put the subtitle in ADDED BY JF
414 my @CNresults2= @CNresults;
415 $newline{CN} = \@CNresults2;
416 $newline{'even'} = 1 if $#finalresult % 2 == 0;
417 $newline{'odd'} = 1 if $#finalresult % 2 == 1;
418 $newline{'timestamp'} = format_date($newline{timestamp});
420 push @finalresult, \%newline;
424 my $nbresults = $#result+1;
425 return (\@finalresult, $nbresults);
428 # Creates the SQL Request
431 my ($dbh,$tags, $and_or, $operator, $value) = @_;
433 my $sql_tables; # will contain marc_subfield_table as m1,...
434 my $sql_where1; # will contain the "true" where
435 my $sql_where2 = "("; # will contain m1.bibid=m2.bibid
436 my $nb_active=0; # will contain the number of "active" entries. an entry is active if a value is provided.
437 my $nb_table=1; # will contain the number of table. ++ on each entry EXCEPT when an OR is provided.
439 my $maxloop=8; # the maximum number of words to avoid a too complex search.
440 $maxloop = @$value if @$value<$maxloop;
442 for(my $i=0; $i<=$maxloop;$i++) {
446 if (@$operator[$i] eq "start") {
447 $sql_tables .= "marc_subfield_table as m$nb_table,";
448 $sql_where1 .= "(m1.subfieldvalue like ".$dbh->quote("@$value[$i]%");
450 $sql_where1 .=" and concat(m1.tag,m1.subfieldcode) in (@$tags[$i])";
453 } elsif (@$operator[$i] eq "contains") {
454 $sql_tables .= "marc_word as m$nb_table,";
455 $sql_where1 .= "(m1.word like ".$dbh->quote("@$value[$i]");
457 $sql_where1 .=" and m1.tagsubfield in (@$tags[$i])";
461 $sql_tables .= "marc_subfield_table as m$nb_table,";
462 $sql_where1 .= "(m1.subfieldvalue @$operator[$i] ".$dbh->quote("@$value[$i]");
464 $sql_where1 .=" and concat(m1.tag,m1.subfieldcode) in (@$tags[$i])";
469 if (@$operator[$i] eq "start") {
471 $sql_tables .= "marc_subfield_table as m$nb_table,";
472 $sql_where1 .= "@$and_or[$i] (m$nb_table.subfieldvalue like ".$dbh->quote("@$value[$i]%");
474 $sql_where1 .=" and concat(m$nb_table.tag,m$nb_table.subfieldcode) in (@$tags[$i])";
477 $sql_where2 .= "m1.bibid=m$nb_table.bibid and ";
478 } elsif (@$operator[$i] eq "contains") {
479 if (@$and_or[$i] eq 'and') {
481 $sql_tables .= "marc_word as m$nb_table,";
482 $sql_where1 .= "@$and_or[$i] (m$nb_table.word like ".$dbh->quote("@$value[$i]");
484 $sql_where1 .=" and m$nb_table.tagsubfield in(@$tags[$i])";
487 $sql_where2 .= "m1.bibid=m$nb_table.bibid and ";
489 $sql_where1 .= "@$and_or[$i] (m$nb_table.word like ".$dbh->quote("@$value[$i]");
491 $sql_where1 .=" and m$nb_table.tagsubfield in (@$tags[$i])";
494 $sql_where2 .= "m1.bibid=m$nb_table.bibid and ";
498 $sql_tables .= "marc_subfield_table as m$nb_table,";
499 $sql_where1 .= "@$and_or[$i] (m$nb_table.subfieldvalue @$operator[$i] ".$dbh->quote(@$value[$i]);
501 $sql_where1 .=" and concat(m$nb_table.tag,m$nb_table.subfieldcode) in (@$tags[$i])";
503 $sql_where2 .= "m1.bibid=m$nb_table.bibid and ";
510 if($sql_where2 ne "(") # some datas added to sql_where2, processing
512 $sql_where2 = substr($sql_where2, 0, (length($sql_where2)-5)); # deletes the trailing ' and '
515 else # no sql_where2 statement, deleting '('
519 chop $sql_tables; # deletes the trailing ','
520 return ($sql_tables, $sql_where1, $sql_where2);
524 my ($dbh, $bibid, $marcflavour) = @_;
525 my ($mintag, $maxtag);
526 if ($marcflavour eq "MARC21") {
529 } else { # assume unimarc if not marc21
534 my $sth=$dbh->prepare("SELECT subfieldvalue,tag FROM marc_subfield_table WHERE bibid=? AND tag BETWEEN ? AND ? ORDER BY tagorder");
536 $sth->execute($bibid,$mintag,$maxtag);
543 while (my $data=$sth->fetchrow_arrayref) {
544 my $value=$data->[0];
545 my $thistag=$data->[1];
549 if ($thistag ne $tag && $note ne "") {
550 $marcnote = {marcnote => $note,};
551 push @marcnotes, $marcnote;
555 if ($note ne $value) {
556 $note = $note." ".$value;
561 $marcnote = {marcnote => $note};
562 push @marcnotes, $marcnote; #load last tag into array
568 my $marcnotesarray=\@marcnotes;
569 return $marcnotesarray;
573 sub getMARCsubjects {
574 my ($dbh, $bibid, $marcflavour) = @_;
575 my ($mintag, $maxtag);
576 if ($marcflavour eq "MARC21") {
579 } else { # assume unimarc if not marc21
583 my $sth=$dbh->prepare("SELECT subfieldvalue,subfieldcode FROM marc_subfield_table WHERE bibid=? AND tag BETWEEN ? AND ? ORDER BY tagorder");
585 $sth->execute($bibid,$mintag,$maxtag);
592 while (my $data=$sth->fetchrow_arrayref) {
593 my $value = $data->[0];
594 my $subfield = $data->[1];
595 if ($subfield eq "a" && $value ne $subjct) {
596 $marcsubjct = {MARCSUBJCT => $value,};
597 push @marcsubjcts, $marcsubjct;
605 my $marcsubjctsarray=\@marcsubjcts;
606 return $marcsubjctsarray;
607 } #end getMARCsubjects
609 END { } # module clean-up code here (global destructor)
618 Koha Developement team <info@koha.org>