misc/cleanmarcdb.pl

   1 #!/usr/bin/perl
   2 # small script that rebuilds the non-MARC DB
   3
   4 use strict;
   5
   6 # Koha modules used
   7 # use MARC::File::USMARC;
   8 use MARC::Record;
   9 use MARC::Batch;
  10 use C4::Context;
  11 use C4::Biblio;
  12 use Time::HiRes qw(gettimeofday);
  13
  14 use Getopt::Long;
  15 my ( $input_marc_file, $number) = ('',0);
  16 my ($version, $confirm,$test_parameter);
  17 GetOptions(
  18         'c' => \$confirm,
  19         'h' => \$version,
  20         't' => \$test_parameter,
  21 );
  22
  23 if ($version || (!$confirm)) {
  24         print <<EOF
  25 This script cleans unused subfields in the MARC DB.
  26 If you alter the MARC parameters and remove a subfield (ie : move it to ignore (10) tab), existing subfields are NOT removed.
  27 It's not a bug, it prevents deleting useful values in case of erroneous move.
  28 This script definetly remove unused subfields in the MARC DB.
  29 syntax :
  30 \t./cleanmarcdb.pl -h (or without arguments => shows this screen)
  31 \t./cleanmarcdb.pl -c (c like confirm => cleans the marc DB (may be long)
  32 \t-t => test only, change nothing in DB
  33 EOF
  34 ;#'
  35 die;
  36 }
  37
  38 my $dbh = C4::Context->dbh;
  39 my $i=0;
  40 my $starttime = gettimeofday;
  41 my $cleansubfield = $dbh->prepare("delete from marc_subfield_table where tag=? and subfieldcode=?");
  42 my $cleanword = $dbh->prepare("delete from marc_word where tag=? and subfieldid=?");
  43
  44 # get tags structure
  45 my $tags = GetMarcStructure(1);
  46 foreach my $tag (sort keys(%{$tags})) {
  47         foreach my $subfield (sort keys(%{$tags->{$tag}})) {
  48                 next if $subfield eq "lib";
  49                 next if $subfield eq "mandatory";
  50                 next if $subfield eq "tab";
  51                 # DO NOT drop biblionumber, biblioitemnumber and itemnumber.
  52                 # they are stored internally, and are mapped to tab -1. This script must keep them or it will completly break Koha DB !!!
  53                 next if ($tags->{$tag}->{$subfield}->{kohafield} eq "biblio.biblionumber");
  54                 next if ($tags->{$tag}->{$subfield}->{kohafield} eq "biblioitems.biblioitemnumber");
  55                 next if ($tags->{$tag}->{$subfield}->{kohafield} eq "items.itemnumber");
  56                 # now, test => if field is ignored (in tab -1 or '') => drop everything in the MARC table !
  57                 if ($tags->{$tag}->{$subfield}->{tab} eq -1 || $tags->{$tag}->{$subfield}->{tab} eq '') {
  58                         print "dropping $tag \$ $subfield\n";
  59                         $cleansubfield->execute($tag,$subfield) unless $test_parameter;
  60                         $cleanword->execute($tag,$subfield) unless $test_parameter;
  61                         print "TEST " if $test_parameter;
  62                         print "done\n";
  63                 }
  64         }
  65 }
  66 my $timeneeded = gettimeofday - $starttime;
  67 print "done in $timeneeded seconds\n";