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 $branchname='MAIN';
  17
  18 unless ($file) {
  19     print "USAGE: ./bulkmarcimport.pl filename\n";
  20     exit;
  21 }
  22
  23
  24
  25
  26 my $lc1='#dddddd';
  27 my $lc2='#ddaaaa';
  28
  29
  30 use C4::Database;
  31 use CGI;
  32 use DBI;
  33 #use strict;
  34 use C4::Acquisitions;
  35 use C4::Biblio;
  36 use C4::Output;
  37 my $dbh=C4Connect;
  38 my $userid=$ENV{'REMOTE_USER'};
  39 %tagtext = (
  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',
  46     '010' => 'LCCN',
  47     '015' => 'LCCN Cdn',
  48     '020' => 'ISBN',
  49     '022' => 'ISSN',
  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',
  93     '852' => 'Location',
  94     '856' => 'Electronic location and access',
  95 );
  96
  97
  98 my $dbh=C4Connect;
  99 if ($file) {
 100     open (F, "$file");
 101     my $data=<F>;
 102     close F;
 103     $splitchar=chr(29);
 104
 105
 106 # Cycle through all of the records in the file
 107
 108
 109 RECORD:
 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);
 115         $splitchar2=chr(30);
 116         my $directory=0;
 117         my $tagcounter=0;
 118         my %tag;
 119         my @record;
 120         my %record;
 121         foreach $field (split(/$splitchar2/, $record)) {
 122             my %field;
 123             unless ($directory) {
 124                 # Parse the MARC directory and store the cotents in the %tag hash
 125                 $directory=$field;
 126                 my $itemcounter=1;
 127                 $counter=0;
 128                 while ($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);
 133                     $tag{$counter}=$tag;
 134                     $counter++;
 135                 }
 136                 $directory=1;
 137                 next;
 138             }
 139             $tag=$tag{$tagcounter};
 140             $tagcounter++;
 141             $field{'tag'}=$tag;
 142             printf "%4s %-40s ",$tag, $tagtext{$tag};
 143             $splitchar3=chr(31);
 144             my @subfields=split(/$splitchar3/, $field);
 145             $indicator=$subfields[0];
 146             $field{'indicator'}=$indicator;
 147             my $firstline=1;
 148             if ($#subfields==0) {
 149                 print "$indicator\n";
 150             } else {
 151                 print "\n";
 152                 my %subfields;
 153                 for ($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, $subfield);
 163                         } else {
 164                             @subfieldlist=($subfields{$subfieldcode}, $subfield);
 165                         }
 166                         $subfields{$subfieldcode}=\@subfieldlist;
 167                     } else {
 168                         $subfields{$subfieldcode}=$subfield;
 169                     }
 170                 }
 171                 $field{'subfields'}=\%subfields;
 172             }
 173             if ($record{$tag}) {
 174                 my $fieldlist=$record{$tag};
 175                 if ($fieldlist->{'tag'}) {
 176                     @fieldlist=($fieldlist, \%field);
 177                     $fieldlist=\@fieldlist;
 178                 } else {
 179                     push (@$fieldlist,\%field);
 180                 }
 181                 $record{$tag}=$fieldlist;
 182             } else {
 183                 $record{$tag}=[\%field];
 184             }
 185             push (@record, \%field);
 186         }
 187         $rec=\@record;
 188         $counter++;
 189         my ($lccn, $isbn, $issn, $dewey, $author, $title, $place, $publisher, $publicationyear, $volume, $number, @subjects, $note, $additionalauthors, $illustrator, $copyrightdate, $barcode, $itemtype, $seriestitle, @barcodes);
 190         my $marc=$record;
 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'};
 195                 $lccn=~s/^\s*//;
 196                 $lccn=~s/cn //;
 197                 $lccn=~s/^\s*//;
 198                 ($lccn) = (split(/\s+/, $lccn))[0];
 199             }
 200             # LCCN is stored in field 015 a
 201             if ($field->{'tag'} eq '015') {
 202                 $lccn=$field->{'subfields'}->{'a'};
 203                 $lccn=~s/^\s*//;
 204                 $lccn=~s/^C//;
 205                 ($lccn) = (split(/\s+/, $lccn))[0];
 206             }
 207             # ISBN is stored in field 020 a
 208             if ($field->{'tag'} eq '020') {
 209                 $isbn=$field->{'subfields'}->{'a'};
 210                 $isbn=~s/^\s*//;
 211                 ($isbn) = (split(/\s+/, $isbn))[0];
 212             }
 213             # ISSN is stored in field 022 a
 214             if ($field->{'tag'} eq '022') {
 215                 $issn=$field->{'subfields'}->{'a'};
 216                 $issn=~s/^\s*//;
 217                 ($issn) = (split(/\s+/, $issn))[0];
 218             }
 219             # Dewey number stored in field 082 a
 220             # If there is more than one dewey number (more than one 'a'
 221             # subfield) I just take the first one
 222             if ($field->{'tag'} eq '082') {
 223                 $dewey=$field->{'subfields'}->{'a'};
 224                 $dewey=~s/\///g;
 225                 if (@$dewey) {
 226                     $dewey=$$dewey[0];
 227                 }
 228             }
 229             # Author is stored in field 100 a
 230             if ($field->{'tag'} eq '100') {
 231                 $author=$field->{'subfields'}->{'a'};
 232             }
 233             # Title is stored in field 245 a
 234             # Subtitle in field 245 b
 235             # Illustrator in field 245 c
 236             if ($field->{'tag'} eq '245') {
 237                 $title=$field->{'subfields'}->{'a'};
 238                 $title=~s/ \/$//;
 239                 $subtitle=$field->{'subfields'}->{'b'};
 240                 $subtitle=~s/ \/$//;
 241                 my $name=$field->{'subfields'}->{'c'};
 242                 if ($name=~/illustrated by]*\s+(.*)/) {
 243                     $illustrator=$1;
 244                 }
 245             }
 246             # Publisher Info in field 260
 247             #   a = place
 248             #   b = publisher
 249             #   c = publication date
 250             #     (also store as copyright date if date starts with a 'c' as in c1995)
 251             if ($field->{'tag'} eq '260') {
 252                 $place=$field->{'subfields'}->{'a'};
 253                 if (@$place) {
 254                     $place=$$place[0];
 255                 }
 256                 $place=~s/\s*:$//g;
 257                 $publisher=$field->{'subfields'}->{'b'};
 258                 if (@$publisher) {
 259                     $publisher=$$publisher[0];
 260                 }
 261                 $publisher=~s/\s*:$//g;
 262                 $publicationyear=$field->{'subfields'}->{'c'};
 263                 if ($publicationyear=~/c(\d\d\d\d)/) {
 264                     $copyrightdate=$1;
 265                 }
 266                 if ($publicationyear=~/[^c](\d\d\d\d)/) {
 267                     $publicationyear=$1;
 268                 } elsif ($copyrightdate) {
 269                     $publicationyear=$copyrightdate;
 270                 } else {
 271                     $publicationyear=~/(\d\d\d\d)/;
 272                     $publicationyear=$1;
 273                 }
 274             }
 275             # Physical Dimensions in field 300
 276             #   a = pages
 277             #   c = size
 278             if ($field->{'tag'} eq '300') {
 279                 $pages=$field->{'subfields'}->{'a'};
 280                 $pages=~s/ \;$//;
 281                 $size=$field->{'subfields'}->{'c'};
 282                 $pages=~s/\s*:$//g;
 283                 $size=~s/\s*:$//g;
 284             }
 285             # Vol/No in field 362 a
 286             if ($field->{'tag'} eq '362') {
 287                 if ($field->{'subfields'}->{'a'}=~/(\d+).*(\d+)/) {
 288                     $volume=$1;
 289                     $number=$2;
 290                 }
 291             }
 292             # Series Title in field 440 a
 293             # Vol/No in field 440 v
 294             if ($field->{'tag'} eq '440') {
 295                 $seriestitle=$field->{'subfields'}->{'a'};
 296                 if ($field->{'subfields'}->{'v'}=~/(\d+).*(\d+)/) {
 297                     $volume=$1;
 298                     $number=$2;
 299                 }
 300             }
 301             # BARCODES!!!
 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\.]*)/;
 314                 $dewey=$1;
 315                 if (($deweyfield=~/pbk/) || ($deweyfield=~/pb$/)) {
 316                     $itemtype='PBK';
 317                 } elsif ($dewey) {
 318                     $itemtype='JNF';
 319                 } else {
 320                     $itemtype='JF';
 321                 }
 322
 323                 $replacementprice=$field->{'subfields'}->{'9'};
 324             }
 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/) {
 330                     $illustrator=$name;
 331                 } else {
 332                     $additionalauthors.="$name\n";
 333                 }
 334             }
 335             # I concatenate all 5XX a entries as notes
 336             if ($field->{'tag'} =~/^5/) {
 337                 $note.="$field->{'subfields'}->{'a'}\n";
 338             }
 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
 344             # subdivision)
 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";
 349                 $subject=~s/\.$//;
 350                 if ($gensubdivision=$field->{'subfields'}->{'x'}) {
 351                     my @sub=@$gensubdivision;
 352                     if ($#sub>=0) {
 353                         foreach $s (@sub) {
 354                             $s=~s/\.$//;
 355                             $subject.=" -- $s";
 356                         }
 357                     } else {
 358                         $gensubdivision=~s/\.$//;
 359                         $subject.=" -- $gensubdivision";
 360                     }
 361                 }
 362                 if ($chronsubdivision=$field->{'subfields'}->{'y'}) {
 363                     my @sub=@$chronsubdivision;
 364                     if ($#sub>=0) {
 365                         foreach $s (@sub) {
 366                             $s=~s/\.$//;
 367                             $subject.=" -- $s";
 368                         }
 369                     } else {
 370                         $chronsubdivision=~s/\.$//;
 371                         $subject.=" -- $chronsubdivision";
 372                     }
 373                 }
 374                 if ($geosubdivision=$field->{'subfields'}->{'z'}) {
 375                     my @sub=@$geosubdivision;
 376                     if ($#sub>=0) {
 377                         foreach $s (@sub) {
 378                             $s=~s/\.$//;
 379                             $subject.=" -- $s";
 380                         }
 381                     } else {
 382                         $geosubdivision=~s/\.$//;
 383                         $subject.=" -- $geosubdivision";
 384                     }
 385                 }
 386                 push @subjects, $subject;
 387             }
 388         }
 389
 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");
 394         $sth->execute;
 395         my $biblionumber=0;
 396         my $biblioitemnumber=0;
 397         if ($sth->rows) {
 398             ($biblionumber, $biblioitemnumber) = $sth->fetchrow;
 399             my $title=$title;
 400 #title already in the database
 401         } else {
 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");
 408             $sth->execute;
 409             if ($sth->rows) {
 410                 ($biblionumber) = $sth->fetchrow;
 411 #title already in the database
 412             } else {
 413                 $sth=$dbh->prepare("select max(biblionumber) from biblio");
 414                 $sth->execute;
 415                 ($biblionumber) = $sth->fetchrow;
 416                 $biblionumber++;
 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)");
 419                 $sth->execute;
 420                 $sth=$dbh->prepare("insert into bibliosubtitle values ($q_subtitle, $biblionumber)");
 421                 $sth->execute;
 422             }
 423             $sth=$dbh->prepare("select max(biblioitemnumber) from biblioitems");
 424             $sth->execute;
 425             ($biblioitemnumber) = $sth->fetchrow;
 426             $biblioitemnumber++;
 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);
 449
 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)");
 451             $sth->execute;
 452             my $subjectheading;
 453             foreach $subjectheading (@subjects) {
 454                 # convert to upper case
 455                 $subjectheading=uc($subjectheading);
 456                 # quote value
 457                 my $q_subjectheading=$dbh->quote($subjectheading);
 458                 $sth=$dbh->prepare("insert into bibliosubject (biblionumber,subject)
 459                     values ($biblionumber, $q_subjectheading)");
 460                 $sth->execute;
 461             }
 462             my @additionalauthors=split(/\n/,$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);
 470                 # quote value
 471                 my $q_additionalauthor=$dbh->quote($additionalauthor);
 472                 $sth=$dbh->prepare("insert into additionalauthors (biblionumber,author) values ($biblionumber, $q_additionalauthor)");
 473                 $sth->execute;
 474             }
 475         }
 476         my $q_barcode=$dbh->quote($barcode);
 477         my $q_homebranch="'$branchname'";
 478         my $q_notes="''";
 479         #my $replacementprice=0;
 480         my $sth=$dbh->prepare("select max(itemnumber) from items");
 481         $sth->execute;
 482         my ($itemnumber) = $sth->fetchrow;
 483         $itemnumber++;
 484         my @datearr=localtime(time);
 485         my $date=(1900+$datearr[5])."-".($datearr[4]+1)."-".$datearr[3];
 486 BARCODE:
 487         foreach $barcode (@barcodes) {
 488             my $q_barcode=$dbh->quote($barcode);
 489             my $sti=$dbh->prepare("select barcode from items where barcode=$q_barcode");
 490             $sti->execute;
 491             if ($sti->rows) {
 492                 print "Skipping $barcode\n";
 493                 next BARCODE;
 494             }
 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($task);
 500             print "$task\n";
 501             $sth->execute;
 502         }
 503     }
 504 }
 505 $dbh->disconnect;