3 # Tool for importing bulk marc records
7 # Do not use this script on a production system, it is still in development
19 print "USAGE: ./bulkmarcimport.pl filename\n";
38 my $userid=$ENV{'REMOTE_USER'};
40 '001' => 'Control number',
41 '003' => 'Control number identifier',
42 '005' => 'Date and time of latest transaction',
43 '006' => 'Fixed-length data elements -- additional material characteristics',
44 '007' => 'Physical description fixed field',
45 '008' => 'Fixed length data elements',
50 '037' => 'Source of acquisition',
51 '040' => 'Cataloging source',
52 '041' => 'Language code',
53 '043' => 'Geographic area code',
54 '050' => 'Library of Congress call number',
55 '060' => 'National Library of Medicine call number',
56 '082' => 'Dewey decimal call number',
57 '100' => 'Main entry -- Personal name',
58 '110' => 'Main entry -- Corporate name',
59 '130' => 'Main entry -- Uniform title',
60 '240' => 'Uniform title',
61 '245' => 'Title statement',
62 '246' => 'Varying form of title',
63 '250' => 'Edition statement',
64 '256' => 'Computer file characteristics',
65 '260' => 'Publication, distribution, etc.',
66 '263' => 'Projected publication date',
67 '300' => 'Physical description',
68 '306' => 'Playing time',
69 '440' => 'Series statement / Added entry -- Title',
70 '490' => 'Series statement',
71 '500' => 'General note',
72 '504' => 'Bibliography, etc. note',
73 '505' => 'Formatted contents note',
74 '508' => 'Creation/production credits note',
75 '510' => 'Citation/references note',
76 '511' => 'Participant or performer note',
77 '520' => 'Summary, etc. note',
78 '521' => 'Target audience note (ie age)',
79 '530' => 'Additional physical form available note',
80 '538' => 'System details note',
81 '586' => 'Awards note',
82 '600' => 'Subject added entry -- Personal name',
83 '610' => 'Subject added entry -- Corporate name',
84 '650' => 'Subject added entry -- Topical term',
85 '651' => 'Subject added entry -- Geographic name',
86 '656' => 'Index term -- Occupation',
87 '700' => 'Added entry -- Personal name',
88 '710' => 'Added entry -- Corporate name',
89 '730' => 'Added entry -- Uniform title',
90 '740' => 'Added entry -- Uncontrolled related/analytical title',
91 '800' => 'Series added entry -- Personal name',
92 '830' => 'Series added entry -- Uniform title',
94 '856' => 'Electronic location and access',
106 # Cycle through all of the records in the file
110 foreach $record (split(/$splitchar/, $data)) {
111 $leader=substr($record,0,24);
112 print "\n\n---------------------------------------------------------------------------\n";
113 print "Leader: $leader\n";
114 $record=substr($record,24);
121 foreach $field (split(/$splitchar2/, $record)) {
123 unless ($directory) {
124 # Parse the MARC directory and store the cotents in the %tag hash
128 while ($item=substr($directory,0,12)) {
129 $tag=substr($directory,0,3);
130 $length=substr($directory,3,4);
131 $start=substr($directory,7,6);
132 $directory=substr($directory,12);
139 $tag=$tag{$tagcounter};
142 printf "%4s %-40s ",$tag, $tagtext{$tag};
144 my @subfields=split(/$splitchar3/, $field);
145 $indicator=$subfields[0];
146 $field{'indicator'}=$indicator;
148 if ($#subfields==0) {
149 print "$indicator\n";
153 for ($i=1; $i<=$#subfields; $i++) {
154 my $text=$subfields[$i];
155 my $subfieldcode=substr($text,0,1);
156 my $subfield=substr($text,1);
157 print " $subfieldcode $subfield\n";
158 if ($subfields{$subfieldcode}) {
159 my $subfieldlist=$subfields{$subfieldcode};
160 my @subfieldlist=@$subfieldlist;
161 if ($#subfieldlist>=0) {
162 push (@subfieldlist, $subfield);
164 @subfieldlist=($subfields{$subfieldcode}, $subfield);
166 $subfields{$subfieldcode}=\@subfieldlist;
168 $subfields{$subfieldcode}=$subfield;
171 $field{'subfields'}=\%subfields;
174 my $fieldlist=$record{$tag};
175 if ($fieldlist->{'tag'}) {
176 @fieldlist=($fieldlist, \%field);
177 $fieldlist=\@fieldlist;
179 push (@$fieldlist,\%field);
181 $record{$tag}=$fieldlist;
183 $record{$tag}=[\%field];
185 push (@record, \%field);
189 my ($lccn, $isbn, $issn, $dewey, $author, $title, $place, $publisher, $publicationyear, $volume, $number, @subjects, $note, $additionalauthors, $illustrator, $copyrightdate, $barcode, $itemtype, $seriestitle, @barcodes);
191 foreach $field (sort {$a->{'tag'} cmp $b->{'tag'}} @$rec) {
192 # LCCN is stored in field 010 a
193 if ($field->{'tag'} eq '010') {
194 $lccn=$field->{'subfields'}->{'a'};
198 ($lccn) = (split(/\s+/, $lccn))[0];
200 # LCCN is stored in field 015 a
201 if ($field->{'tag'} eq '015') {
202 $lccn=$field->{'subfields'}->{'a'};
205 ($lccn) = (split(/\s+/, $lccn))[0];
207 # ISBN is stored in field 020 a
208 if ($field->{'tag'} eq '020') {
209 $isbn=$field->{'subfields'}->{'a'};
211 ($isbn) = (split(/\s+/, $isbn))[0];
213 # ISSN is stored in field 022 a
214 if ($field->{'tag'} eq '022') {
215 $issn=$field->{'subfields'}->{'a'};
217 ($issn) = (split(/\s+/, $issn))[0];
219 # Dewey number stored in field 082 a
220 # If there is more than one dewey number (more than one 'a'
221 # subfield) I just take the first one
222 if ($field->{'tag'} eq '082') {
223 $dewey=$field->{'subfields'}->{'a'};
229 # Author is stored in field 100 a
230 if ($field->{'tag'} eq '100') {
231 $author=$field->{'subfields'}->{'a'};
233 # Title is stored in field 245 a
234 # Subtitle in field 245 b
235 # Illustrator in field 245 c
236 if ($field->{'tag'} eq '245') {
237 $title=$field->{'subfields'}->{'a'};
239 $subtitle=$field->{'subfields'}->{'b'};
241 my $name=$field->{'subfields'}->{'c'};
242 if ($name=~/illustrated by]*\s+(.*)/) {
246 # Publisher Info in field 260
249 # c = publication date
250 # (also store as copyright date if date starts with a 'c' as in c1995)
251 if ($field->{'tag'} eq '260') {
252 $place=$field->{'subfields'}->{'a'};
257 $publisher=$field->{'subfields'}->{'b'};
259 $publisher=$$publisher[0];
261 $publisher=~s/\s*:$//g;
262 $publicationyear=$field->{'subfields'}->{'c'};
263 if ($publicationyear=~/c(\d\d\d\d)/) {
266 if ($publicationyear=~/[^c](\d\d\d\d)/) {
268 } elsif ($copyrightdate) {
269 $publicationyear=$copyrightdate;
271 $publicationyear=~/(\d\d\d\d)/;
275 # Physical Dimensions in field 300
278 if ($field->{'tag'} eq '300') {
279 $pages=$field->{'subfields'}->{'a'};
281 $size=$field->{'subfields'}->{'c'};
285 # Vol/No in field 362 a
286 if ($field->{'tag'} eq '362') {
287 if ($field->{'subfields'}->{'a'}=~/(\d+).*(\d+)/) {
292 # Series Title in field 440 a
293 # Vol/No in field 440 v
294 if ($field->{'tag'} eq '440') {
295 $seriestitle=$field->{'subfields'}->{'a'};
296 if ($field->{'subfields'}->{'v'}=~/(\d+).*(\d+)/) {
302 # 852 p stores barcodes
303 # 852 h stores dewey field
304 # 852 9 stores replacement price
305 # I check for an itemtype identifier in 852h as well... pb or pbk means PBK
306 # also if $dewey is > 0, then I assign JNF, otherwise JF.
307 # Note that my libraries are school libraries, so I assume Junior.
308 if ($field->{'tag'} eq '852') {
309 $barcode=$field->{'subfields'}->{'p'};
310 push (@barcodes, $barcode);
311 my $q_barcode=$dbh->quote($barcode);
312 my $deweyfield=$field->{'subfields'}->{'h'};
313 $deweyfield=~/^([\d\.]*)/;
315 if (($deweyfield=~/pbk/) || ($deweyfield=~/pb$/)) {
323 $replacementprice=$field->{'subfields'}->{'9'};
325 # 700 a stores additional authors / illustrator info
326 # 700 c will contain 'ill' if it's an illustrator
327 if ($field->{'tag'} eq '700') {
328 my $name=$field->{'subfields'}->{'a'};
329 if ($field->{'subfields'}->{'c'}=~/ill/) {
332 $additionalauthors.="$name\n";
335 # I concatenate all 5XX a entries as notes
336 if ($field->{'tag'} =~/^5/) {
337 $note.="$field->{'subfields'}->{'a'}\n";
339 # 6XX entries are subject entries
340 # Not sure why I'm skipping 691 tags
341 # 691 a contains the subject.
342 # I take subfield a, and append entries from subfield x (general
343 # subdivision) y (Chronological subdivision) and z (geographic
345 if ($field->{'tag'} =~/6\d\d/) {
346 (next) if ($field->{'tag'} eq '691');
347 my $subject=$field->{'subfields'}->{'a'};
348 print "SUBJECT: $subject\n";
350 if ($gensubdivision=$field->{'subfields'}->{'x'}) {
351 my @sub=@$gensubdivision;
358 $gensubdivision=~s/\.$//;
359 $subject.=" -- $gensubdivision";
362 if ($chronsubdivision=$field->{'subfields'}->{'y'}) {
363 my @sub=@$chronsubdivision;
370 $chronsubdivision=~s/\.$//;
371 $subject.=" -- $chronsubdivision";
374 if ($geosubdivision=$field->{'subfields'}->{'z'}) {
375 my @sub=@$geosubdivision;
382 $geosubdivision=~s/\.$//;
383 $subject.=" -- $geosubdivision";
386 push @subjects, $subject;
390 my $q_isbn=$dbh->quote($isbn);
391 my $q_issn=$dbh->quote($issn);
392 my $q_lccn=$dbh->quote($lccn);
393 my $sth=$dbh->prepare("select biblionumber,biblioitemnumber from biblioitems where issn=$q_issn or isbn=$q_isbn or lccn=$q_lccn");
396 my $biblioitemnumber=0;
398 ($biblionumber, $biblioitemnumber) = $sth->fetchrow;
400 #title already in the database
402 my $q_title=$dbh->quote("$title");
403 my $q_subtitle=$dbh->quote("$subtitle");
404 my $q_author=$dbh->quote($author);
405 my $q_copyrightdate=$dbh->quote($copyrightdate);
406 my $q_seriestitle=$dbh->quote($seriestitle);
407 $sth=$dbh->prepare("select biblionumber from biblio where title=$q_title and author=$q_author and copyrightdate=$q_copyrightdate and seriestitle=$q_seriestitle");
410 ($biblionumber) = $sth->fetchrow;
411 #title already in the database
413 $sth=$dbh->prepare("select max(biblionumber) from biblio");
415 ($biblionumber) = $sth->fetchrow;
417 my $q_notes=$dbh->quote($note);
418 $sth=$dbh->prepare("insert into biblio (biblionumber, title, author, copyrightdate, seriestitle, notes) values ($biblionumber, $q_title, $q_author, $q_copyrightdate, $q_seriestitle, $q_notes)");
420 $sth=$dbh->prepare("insert into bibliosubtitle values ($q_subtitle, $biblionumber)");
423 $sth=$dbh->prepare("select max(biblioitemnumber) from biblioitems");
425 ($biblioitemnumber) = $sth->fetchrow;
427 my $q_isbn=$dbh->quote($isbn);
428 my $q_issn=$dbh->quote($issn);
429 my $q_lccn=$dbh->quote($lccn);
430 my $q_volume=$dbh->quote($volume);
431 my $q_number=$dbh->quote($number);
432 my $q_itemtype=$dbh->quote($itemtype);
433 my $q_dewey=$dbh->quote($dewey);
434 $cleanauthor=$author;
435 $cleanauthor=~s/[^A-Za-z]//g;
436 $subclass=uc(substr($cleanauthor,0,3));
437 my $q_subclass=$dbh->quote($subclass);
438 my $q_publicationyear=$dbh->quote($publicationyear);
439 my $q_publishercode=$dbh->quote($publishercode);
440 my $q_volumedate=$dbh->quote($volumedate);
441 my $q_volumeddesc=$dbh->quote($volumeddesc);
442 my $q_illus=$dbh->quote($illustrator);
443 my $q_pages=$dbh->quote($pages);
444 my $q_notes=$dbh->quote($note);
445 ($q_notes) || ($q_notes="''");
446 my $q_size=$dbh->quote($size);
447 my $q_place=$dbh->quote($place);
448 my $q_marc=$dbh->quote($marc);
450 $sth=$dbh->prepare("insert into biblioitems (biblioitemnumber, biblionumber, volume, number, itemtype, isbn, issn, dewey, subclass, publicationyear, publishercode, volumedate, volumeddesc, illus, pages, size, place, lccn, marc) values ($biblioitemnumber, $biblionumber, $q_volume, $q_number, $q_itemtype, $q_isbn, $q_issn, $q_dewey, $q_subclass, $q_publicationyear, $q_publishercode, $q_volumedate, $q_volumeddesc, $q_illus, $q_pages, $q_size, $q_place, $q_lccn, $q_marc)");
453 foreach $subjectheading (@subjects) {
454 # convert to upper case
455 $subjectheading=uc($subjectheading);
457 my $q_subjectheading=$dbh->quote($subjectheading);
458 $sth=$dbh->prepare("insert into bibliosubject (biblionumber,subject)
459 values ($biblionumber, $q_subjectheading)");
462 my @additionalauthors=split(/\n/,$additionalauthors);
463 my $additionalauthor;
464 foreach $additionalauthor (@additionalauthors) {
465 # remove any line ending characters (Ctrl-L or Ctrl-M)
466 $additionalauthor=~s/\013//g;
467 $additionalauthor=~s/\010//g;
468 # convert to upper case
469 $additionalauthor=uc($additionalauthor);
471 my $q_additionalauthor=$dbh->quote($additionalauthor);
472 $sth=$dbh->prepare("insert into additionalauthors (biblionumber,author) values ($biblionumber, $q_additionalauthor)");
476 my $q_barcode=$dbh->quote($barcode);
477 my $q_homebranch="'$branchname'";
479 #my $replacementprice=0;
480 my $sth=$dbh->prepare("select max(itemnumber) from items");
482 my ($itemnumber) = $sth->fetchrow;
484 my @datearr=localtime(time);
485 my $date=(1900+$datearr[5])."-".($datearr[4]+1)."-".$datearr[3];
487 foreach $barcode (@barcodes) {
488 my $q_barcode=$dbh->quote($barcode);
489 my $sti=$dbh->prepare("select barcode from items where barcode=$q_barcode");
492 print "Skipping $barcode\n";
495 $replacementprice=~s/^p//;
496 ($replacementprice) || ($replacementprice=0);
497 $replacementprice=~s/\$//;
498 $task="insert into items (itemnumber, biblionumber, biblioitemnumber, barcode, itemnotes, homebranch, holdingbranch, dateaccessioned, replacementprice) values ($itemnumber, $biblionumber, $biblioitemnumber, $q_barcode, $q_notes, $q_homebranch, '$branchname', '$date', $replacementprice)";
499 $sth=$dbh->prepare($task);