acqui.simple/bulkmarcimport.pl

   1 #!/usr/bin/perl
   2 #
   3 # Tool for importing bulk marc records
   4 #
   5 # WARNING!!
   6 #
   7 # Do not use this script on a production system, it is still in development
   8 #
   9 #
  10
  11
  12
  13
  14 $file=$ARGV[0];
  15
  16 unless ($file) {
  17     print "USAGE: ./bulkmarcimport.pl filename\n";
  18     exit;
  19 }
  20
  21
  22
  23
  24 my $lc1='#dddddd';
  25 my $lc2='#ddaaaa';
  26
  27
  28 use C4::Database;
  29 use CGI;
  30 use DBI;
  31 #use strict;
  32 use C4::Acquisitions;
  33 use C4::Output;
  34 my $dbh=C4Connect;
  35 my $userid=$ENV{'REMOTE_USER'};
  36 %tagtext = (
  37     '001' => 'Control number',
  38     '003' => 'Control number identifier',
  39     '005' => 'Date and time of latest transaction',
  40     '006' => 'Fixed-length data elements -- additional material characteristics',
  41     '007' => 'Physical description fixed field',
  42     '008' => 'Fixed length data elements',
  43     '010' => 'LCCN',
  44     '015' => 'LCCN Cdn',
  45     '020' => 'ISBN',
  46     '022' => 'ISSN',
  47     '037' => 'Source of acquisition',
  48     '040' => 'Cataloging source',
  49     '041' => 'Language code',
  50     '043' => 'Geographic area code',
  51     '050' => 'Library of Congress call number',
  52     '060' => 'National Library of Medicine call number',
  53     '082' => 'Dewey decimal call number',
  54     '100' => 'Main entry -- Personal name',
  55     '110' => 'Main entry -- Corporate name',
  56     '130' => 'Main entry -- Uniform title',
  57     '240' => 'Uniform title',
  58     '245' => 'Title statement',
  59     '246' => 'Varying form of title',
  60     '250' => 'Edition statement',
  61     '256' => 'Computer file characteristics',
  62     '260' => 'Publication, distribution, etc.',
  63     '263' => 'Projected publication date',
  64     '300' => 'Physical description',
  65     '306' => 'Playing time',
  66     '440' => 'Series statement / Added entry -- Title',
  67     '490' => 'Series statement',
  68     '500' => 'General note',
  69     '504' => 'Bibliography, etc. note',
  70     '505' => 'Formatted contents note',
  71     '508' => 'Creation/production credits note',
  72     '510' => 'Citation/references note',
  73     '511' => 'Participant or performer note',
  74     '520' => 'Summary, etc. note',
  75     '521' => 'Target audience note (ie age)',
  76     '530' => 'Additional physical form available note',
  77     '538' => 'System details note',
  78     '586' => 'Awards note',
  79     '600' => 'Subject added entry -- Personal name',
  80     '610' => 'Subject added entry -- Corporate name',
  81     '650' => 'Subject added entry -- Topical term',
  82     '651' => 'Subject added entry -- Geographic name',
  83     '656' => 'Index term -- Occupation',
  84     '700' => 'Added entry -- Personal name',
  85     '710' => 'Added entry -- Corporate name',
  86     '730' => 'Added entry -- Uniform title',
  87     '740' => 'Added entry -- Uncontrolled related/analytical title',
  88     '800' => 'Series added entry -- Personal name',
  89     '830' => 'Series added entry -- Uniform title',
  90     '852' => 'Location',
  91     '856' => 'Electronic location and access',
  92 );
  93
  94
  95 my $dbh=C4Connect;
  96 if ($file) {
  97     open (F, "$file");
  98     my $data=<F>;
  99     close F;
 100     $splitchar=chr(29);
 101
 102
 103 # Cycle through all of the records in the file
 104
 105
 106 RECORD:
 107     foreach $record (split(/$splitchar/, $data)) {
 108         $leader=substr($record,0,24);
 109         print "\n\n---------------------------------------------------------------------------\n";
 110         print "Leader: $leader\n";
 111         $record=substr($record,24);
 112         $splitchar2=chr(30);
 113         my $directory=0;
 114         my $tagcounter=0;
 115         my %tag;
 116         my @record;
 117         my %record;
 118         foreach $field (split(/$splitchar2/, $record)) {
 119             my %field;
 120             unless ($directory) {
 121                 # Parse the MARC directory and store the cotents in the %tag hash
 122                 $directory=$field;
 123                 my $itemcounter=1;
 124                 $counter=0;
 125                 while ($item=substr($directory,0,12)) {
 126                     $tag=substr($directory,0,3);
 127                     $length=substr($directory,3,4);
 128                     $start=substr($directory,7,6);
 129                     $directory=substr($directory,12);
 130                     $tag{$counter}=$tag;
 131                     $counter++;
 132                 }
 133                 $directory=1;
 134                 next;
 135             }
 136             $tag=$tag{$tagcounter};
 137             $tagcounter++;
 138             $field{'tag'}=$tag;
 139             printf "%4s %-40s ",$tag, $tagtext{$tag};
 140             $splitchar3=chr(31);
 141             my @subfields=split(/$splitchar3/, $field);
 142             $indicator=$subfields[0];
 143             $field{'indicator'}=$indicator;
 144             my $firstline=1;
 145             if ($#subfields==0) {
 146                 print "$indicator\n";
 147             } else {
 148                 print "\n";
 149                 my %subfields;
 150                 for ($i=1; $i<=$#subfields; $i++) {
 151                     my $text=$subfields[$i];
 152                     my $subfieldcode=substr($text,0,1);
 153                     my $subfield=substr($text,1);
 154                     print "   $subfieldcode $subfield\n";
 155                     if ($subfields{$subfieldcode}) {
 156                         my $subfieldlist=$subfields{$subfieldcode};
 157                         my @subfieldlist=@$subfieldlist;
 158                         if ($#subfieldlist>=0) {
 159                             push (@subfieldlist, $subfield);
 160                         } else {
 161                             @subfieldlist=($subfields{$subfieldcode}, $subfield);
 162                         }
 163                         $subfields{$subfieldcode}=\@subfieldlist;
 164                     } else {
 165                         $subfields{$subfieldcode}=$subfield;
 166                     }
 167                 }
 168                 $field{'subfields'}=\%subfields;
 169             }
 170             if ($record{$tag}) {
 171                 my $fieldlist=$record{$tag};
 172                 if ($fieldlist->{'tag'}) {
 173                     @fieldlist=($fieldlist, \%field);
 174                     $fieldlist=\@fieldlist;
 175                 } else {
 176                     push (@$fieldlist,\%field);
 177                 }
 178                 $record{$tag}=$fieldlist;
 179             } else {
 180                 $record{$tag}=[\%field];
 181             }
 182             push (@record, \%field);
 183         }
 184         $rec=\@record;
 185         $counter++;
 186         my ($lccn, $isbn, $issn, $dewey, $author, $title, $place, $publisher, $publicationyear, $volume, $number, @subjects, $note, $additionalauthors, $illustrator, $copyrightdate, $barcode, $itemtype, $seriestitle, @barcodes);
 187         my $marc=$record;
 188         foreach $field (sort {$a->{'tag'} cmp $b->{'tag'}} @$rec) {
 189             # LCCN is stored in field 010 a
 190             if ($field->{'tag'} eq '010') {
 191                 $lccn=$field->{'subfields'}->{'a'};
 192                 $lccn=~s/^\s*//;
 193                 $lccn=~s/cn //;
 194                 $lccn=~s/^\s*//;
 195                 ($lccn) = (split(/\s+/, $lccn))[0];
 196             }
 197             # LCCN is stored in field 015 a
 198             if ($field->{'tag'} eq '015') {
 199                 $lccn=$field->{'subfields'}->{'a'};
 200                 $lccn=~s/^\s*//;
 201                 $lccn=~s/^C//;
 202                 ($lccn) = (split(/\s+/, $lccn))[0];
 203             }
 204             # ISBN is stored in field 020 a
 205             if ($field->{'tag'} eq '020') {
 206                 $isbn=$field->{'subfields'}->{'a'};
 207                 $isbn=~s/^\s*//;
 208                 ($isbn) = (split(/\s+/, $isbn))[0];
 209             }
 210             # ISSN is stored in field 022 a
 211             if ($field->{'tag'} eq '022') {
 212                 $issn=$field->{'subfields'}->{'a'};
 213                 $issn=~s/^\s*//;
 214                 ($issn) = (split(/\s+/, $issn))[0];
 215             }
 216             # Dewey number stored in field 082 a
 217             # If there is more than one dewey number (more than one 'a'
 218             # subfield) I just take the first one
 219             if ($field->{'tag'} eq '082') {
 220                 $dewey=$field->{'subfields'}->{'a'};
 221                 $dewey=~s/\///g;
 222                 if (@$dewey) {
 223                     $dewey=$$dewey[0];
 224                 }
 225             }
 226             # Author is stored in field 100 a
 227             if ($field->{'tag'} eq '100') {
 228                 $author=$field->{'subfields'}->{'a'};
 229             }
 230             # Title is stored in field 245 a
 231             # Subtitle in field 245 b
 232             # Illustrator in field 245 c
 233             if ($field->{'tag'} eq '245') {
 234                 $title=$field->{'subfields'}->{'a'};
 235                 $title=~s/ \/$//;
 236                 $subtitle=$field->{'subfields'}->{'b'};
 237                 $subtitle=~s/ \/$//;
 238                 my $name=$field->{'subfields'}->{'c'};
 239                 if ($name=~/illustrated by]*\s+(.*)/) {
 240                     $illustrator=$1;
 241                 }
 242             }
 243             # Publisher Info in field 260
 244             #   a = place
 245             #   b = publisher
 246             #   c = publication date
 247             #     (also store as copyright date if date starts with a 'c' as in c1995)
 248             if ($field->{'tag'} eq '260') {
 249                 $place=$field->{'subfields'}->{'a'};
 250                 if (@$place) {
 251                     $place=$$place[0];
 252                 }
 253                 $place=~s/\s*:$//g;
 254                 $publisher=$field->{'subfields'}->{'b'};
 255                 if (@$publisher) {
 256                     $publisher=$$publisher[0];
 257                 }
 258                 $publisher=~s/\s*:$//g;
 259                 $publicationyear=$field->{'subfields'}->{'c'};
 260                 if ($publicationyear=~/c(\d\d\d\d)/) {
 261                     $copyrightdate=$1;
 262                 }
 263                 if ($publicationyear=~/[^c](\d\d\d\d)/) {
 264                     $publicationyear=$1;
 265                 } elsif ($copyrightdate) {
 266                     $publicationyear=$copyrightdate;
 267                 } else {
 268                     $publicationyear=~/(\d\d\d\d)/;
 269                     $publicationyear=$1;
 270                 }
 271             }
 272             # Physical Dimensions in field 300
 273             #   a = pages
 274             #   c = size
 275             if ($field->{'tag'} eq '300') {
 276                 $pages=$field->{'subfields'}->{'a'};
 277                 $pages=~s/ \;$//;
 278                 $size=$field->{'subfields'}->{'c'};
 279                 $pages=~s/\s*:$//g;
 280                 $size=~s/\s*:$//g;
 281             }
 282             # Vol/No in field 362 a
 283             if ($field->{'tag'} eq '362') {
 284                 if ($field->{'subfields'}->{'a'}=~/(\d+).*(\d+)/) {
 285                     $volume=$1;
 286                     $number=$2;
 287                 }
 288             }
 289             # Series Title in field 440 a
 290             # Vol/No in field 440 v
 291             if ($field->{'tag'} eq '440') {
 292                 $seriestitle=$field->{'subfields'}->{'a'};
 293                 if ($field->{'subfields'}->{'v'}=~/(\d+).*(\d+)/) {
 294                     $volume=$1;
 295                     $number=$2;
 296                 }
 297             }
 298             # BARCODES!!!
 299             # 852 p stores barcodes
 300             # 852 h stores dewey field
 301             # 852 9 stores replacement price
 302             #   I check for an itemtype identifier in 852h as well... pb or pbk means PBK
 303             #   also if $dewey is > 0, then I assign JNF, otherwise JF.
 304             #   Note that my libraries are school libraries, so I assume Junior.
 305             if ($field->{'tag'} eq '852') {
 306                 $barcode=$field->{'subfields'}->{'p'};
 307                 push (@barcodes, $barcode);
 308                 my $q_barcode=$dbh->quote($barcode);
 309                 my $deweyfield=$field->{'subfields'}->{'h'};
 310                 $deweyfield=~/^([\d\.]*)/;
 311                 $dewey=$1;
 312                 if (($deweyfield=~/pbk/) || ($deweyfield=~/pb$/)) {
 313                     $itemtype='PBK';
 314                 } elsif ($dewey) {
 315                     $itemtype='JNF';
 316                 } else {
 317                     $itemtype='JF';
 318                 }
 319
 320                 $replacementprice=$field->{'subfields'}->{'9'};
 321             }
 322             # 700 a stores additional authors / illustrator info
 323             # 700 c will contain 'ill' if it's an illustrator
 324             if ($field->{'tag'} eq '700') {
 325                 my $name=$field->{'subfields'}->{'a'};
 326                 if ($field->{'subfields'}->{'c'}=~/ill/) {
 327                     $illustrator=$name;
 328                 } else {
 329                     $additionalauthors.="$name\n";
 330                 }
 331             }
 332             # I concatenate all 5XX a entries as notes
 333             if ($field->{'tag'} =~/^5/) {
 334                 $note.="$field->{'subfields'}->{'a'}\n";
 335             }
 336             # 6XX entries are subject entries
 337             #   Not sure why I'm skipping 691 tags
 338             #   691 a contains the subject.
 339             # I take subfield a, and append entries from subfield x (general
 340             # subdivision) y (Chronological subdivision) and z (geographic
 341             # subdivision)
 342             if ($field->{'tag'} =~/6\d\d/) {
 343                 (next) if ($field->{'tag'} eq '691');
 344                 my $subject=$field->{'subfields'}->{'a'};
 345                 print "SUBJECT: $subject\n";
 346                 $subject=~s/\.$//;
 347                 if ($gensubdivision=$field->{'subfields'}->{'x'}) {
 348                     my @sub=@$gensubdivision;
 349                     if ($#sub>=0) {
 350                         foreach $s (@sub) {
 351                             $s=~s/\.$//;
 352                             $subject.=" -- $s";
 353                         }
 354                     } else {
 355                         $gensubdivision=~s/\.$//;
 356                         $subject.=" -- $gensubdivision";
 357                     }
 358                 }
 359                 if ($chronsubdivision=$field->{'subfields'}->{'y'}) {
 360                     my @sub=@$chronsubdivision;
 361                     if ($#sub>=0) {
 362                         foreach $s (@sub) {
 363                             $s=~s/\.$//;
 364                             $subject.=" -- $s";
 365                         }
 366                     } else {
 367                         $chronsubdivision=~s/\.$//;
 368                         $subject.=" -- $chronsubdivision";
 369                     }
 370                 }
 371                 if ($geosubdivision=$field->{'subfields'}->{'z'}) {
 372                     my @sub=@$geosubdivision;
 373                     if ($#sub>=0) {
 374                         foreach $s (@sub) {
 375                             $s=~s/\.$//;
 376                             $subject.=" -- $s";
 377                         }
 378                     } else {
 379                         $geosubdivision=~s/\.$//;
 380                         $subject.=" -- $geosubdivision";
 381                     }
 382                 }
 383                 push @subjects, $subject;
 384             }
 385         }
 386
 387         my $q_isbn=$dbh->quote($isbn);
 388         my $q_issn=$dbh->quote($issn);
 389         my $q_lccn=$dbh->quote($lccn);
 390         my $sth=$dbh->prepare("select biblionumber,biblioitemnumber from biblioitems where issn=$q_issn or isbn=$q_isbn or lccn=$q_lccn");
 391         $sth->execute;
 392         my $biblionumber=0;
 393         my $biblioitemnumber=0;
 394         if ($sth->rows) {
 395             ($biblionumber, $biblioitemnumber) = $sth->fetchrow;
 396             my $title=$title;
 397 #title already in the database
 398         } else {
 399             my $q_title=$dbh->quote("$title");
 400             my $q_subtitle=$dbh->quote("$subtitle");
 401             my $q_author=$dbh->quote($author);
 402             my $q_copyrightdate=$dbh->quote($copyrightdate);
 403             my $q_seriestitle=$dbh->quote($seriestitle);
 404             $sth=$dbh->prepare("select biblionumber from biblio where title=$q_title and author=$q_author and copyrightdate=$q_copyrightdate and seriestitle=$q_seriestitle");
 405             $sth->execute;
 406             if ($sth->rows) {
 407                 ($biblionumber) = $sth->fetchrow;
 408 #title already in the database
 409             } else {
 410                 $sth=$dbh->prepare("select max(biblionumber) from biblio");
 411                 $sth->execute;
 412                 ($biblionumber) = $sth->fetchrow;
 413                 $biblionumber++;
 414                 my $q_notes=$dbh->quote($note);
 415                 $sth=$dbh->prepare("insert into biblio (biblionumber, title, author, copyrightdate, seriestitle, notes) values ($biblionumber, $q_title, $q_author, $q_copyrightdate, $q_seriestitle, $q_notes)");
 416                 $sth->execute;
 417                 $sth=$dbh->prepare("insert into bibliosubtitle values ($q_subtitle, $biblionumber)");
 418                 $sth->execute;
 419             }
 420             $sth=$dbh->prepare("select max(biblioitemnumber) from biblioitems");
 421             $sth->execute;
 422             ($biblioitemnumber) = $sth->fetchrow;
 423             $biblioitemnumber++;
 424             my $q_isbn=$dbh->quote($isbn);
 425             my $q_issn=$dbh->quote($issn);
 426             my $q_lccn=$dbh->quote($lccn);
 427             my $q_volume=$dbh->quote($volume);
 428             my $q_number=$dbh->quote($number);
 429             my $q_itemtype=$dbh->quote($itemtype);
 430             my $q_dewey=$dbh->quote($dewey);
 431             $cleanauthor=$author;
 432             $cleanauthor=~s/[^A-Za-z]//g;
 433             $subclass=uc(substr($cleanauthor,0,3));
 434             my $q_subclass=$dbh->quote($subclass);
 435             my $q_publicationyear=$dbh->quote($publicationyear);
 436             my $q_publishercode=$dbh->quote($publishercode);
 437             my $q_volumedate=$dbh->quote($volumedate);
 438             my $q_volumeddesc=$dbh->quote($volumeddesc);
 439             my $q_illus=$dbh->quote($illustrator);
 440             my $q_pages=$dbh->quote($pages);
 441             my $q_notes=$dbh->quote($note);
 442             ($q_notes) || ($q_notes="''");
 443             my $q_size=$dbh->quote($size);
 444             my $q_place=$dbh->quote($place);
 445             my $q_marc=$dbh->quote($marc);
 446
 447             $sth=$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)");
 448             $sth->execute;
 449             my $subjectheading;
 450             foreach $subjectheading (@subjects) {
 451                 # convert to upper case
 452                 $subjectheading=uc($subjectheading);
 453                 # quote value
 454                 my $q_subjectheading=$dbh->quote($subjectheading);
 455                 $sth=$dbh->prepare("insert into bibliosubject (biblionumber,subject)
 456                     values ($biblionumber, $q_subjectheading)");
 457                 $sth->execute;
 458             }
 459             my @additionalauthors=split(/\n/,$additionalauthors);
 460             my $additionalauthor;
 461             foreach $additionalauthor (@additionalauthors) {
 462                 # remove any line ending characters (Ctrl-L or Ctrl-M)
 463                 $additionalauthor=~s/\013//g;
 464                 $additionalauthor=~s/\010//g;
 465                 # convert to upper case
 466                 $additionalauthor=uc($additionalauthor);
 467                 # quote value
 468                 my $q_additionalauthor=$dbh->quote($additionalauthor);
 469                 $sth=$dbh->prepare("insert into additionalauthors (biblionumber,author) values ($biblionumber, $q_additionalauthor)");
 470                 $sth->execute;
 471             }
 472         }
 473         my $q_barcode=$dbh->quote($barcode);
 474         my $q_homebranch="'MAIN'";
 475         my $q_notes="''";
 476         #my $replacementprice=0;
 477         my $sth=$dbh->prepare("select max(itemnumber) from items");
 478         $sth->execute;
 479         my ($itemnumber) = $sth->fetchrow;
 480         $itemnumber++;
 481         my @datearr=localtime(time);
 482         my $date=(1900+$datearr[5])."-".($datearr[4]+1)."-".$datearr[3];
 483 BARCODE:
 484         foreach $barcode (@barcodes) {
 485             my $q_barcode=$dbh->quote($barcode);
 486             my $sti=$dbh->prepare("select barcode from items where barcode=$q_barcode");
 487             $sti->execute;
 488             if ($sti->rows) {
 489                 print "Skipping $barcode\n";
 490                 next BARCODE;
 491             }
 492             $replacementprice=~s/^p//;
 493             ($replacementprice) || ($replacementprice=0);
 494             $replacementprice=~s/\$//;
 495             $task="insert into items (itemnumber, biblionumber, biblioitemnumber, barcode, itemnotes, homebranch, holdingbranch, dateaccessioned, replacementprice) values ($itemnumber, $biblionumber, $biblioitemnumber, $q_barcode, $q_notes, $q_homebranch, 'MAIN', '$date', $replacementprice)";
 496             $sth=$dbh->prepare($task);
 497             print "$task\n";
 498             $sth->execute;
 499         }
 500     }
 501 }
 502 $dbh->disconnect;