misc/batchDeleteUnusedSubfields.pl

   1 #!/usr/bin/perl
   2 # small script that rebuilds the non-MARC DB
   3
   4 use strict;
   5 BEGIN {
   6     # find Koha's Perl modules
   7     # test carefully before changing this
   8     use FindBin;
   9     eval { require "$FindBin::Bin/kohalib.pl" };
  10 }
  11
  12 # Koha modules used
  13 use MARC::Record;
  14 use C4::Context;
  15 use C4::Biblio;
  16 use Time::HiRes qw(gettimeofday);
  17
  18 use Getopt::Long;
  19 my ( $input_marc_file, $number) = ('',0);
  20 my ($version, $confirm,$test_parameter);
  21 GetOptions(
  22         'c' => \$confirm,
  23         'h' => \$version,
  24         't' => \$test_parameter,
  25 );
  26
  27 if ($version || (!$confirm)) {
  28         print <<EOF
  29 This script cleans unused subfields in the MARC DB.
  30 If you alter the MARC parameters and remove a subfield (ie : move it to ignore (10) tab), existing subfields are NOT removed.
  31 It's not a bug, it prevents deleting useful values in case of erroneous move.
  32 This script definetly remove unused subfields in the MARC DB.
  33 syntax :
  34 \t./cleanmarcdb.pl -h (or without arguments => shows this screen)
  35 \t./cleanmarcdb.pl -c (c like confirm => cleans the marc DB (may be long)
  36 \t-t => test only, change nothing in DB
  37 EOF
  38 ;#'
  39 die;
  40 }
  41
  42 my $dbh = C4::Context->dbh;
  43 my $i=0;
  44 my $starttime = gettimeofday;
  45 my $cleansubfield = $dbh->prepare("delete from marc_subfield_table where tag=? and subfieldcode=?");
  46 my $cleanword = $dbh->prepare("delete from marc_word where tag=? and subfieldid=?");
  47
  48 # get tags structure
  49 my $tags = GetMarcStructure(1);
  50 foreach my $tag (sort keys(%{$tags})) {
  51         foreach my $subfield (sort keys(%{$tags->{$tag}})) {
  52                 next if $subfield eq "lib";
  53                 next if $subfield eq "mandatory";
  54                 next if $subfield eq "tab";
  55                 # DO NOT drop biblionumber, biblioitemnumber and itemnumber.
  56                 # they are stored internally, and are mapped to tab -1. This script must keep them or it will completly break Koha DB !!!
  57                 next if ($tags->{$tag}->{$subfield}->{kohafield} eq "biblio.biblionumber");
  58                 next if ($tags->{$tag}->{$subfield}->{kohafield} eq "biblioitems.biblioitemnumber");
  59                 next if ($tags->{$tag}->{$subfield}->{kohafield} eq "items.itemnumber");
  60                 # now, test => if field is ignored (in tab -1 or '') => drop everything in the MARC table !
  61                 if ($tags->{$tag}->{$subfield}->{tab} eq -1 || $tags->{$tag}->{$subfield}->{tab} eq '') {
  62                         print "dropping $tag \$ $subfield\n";
  63                         $cleansubfield->execute($tag,$subfield) unless $test_parameter;
  64                         $cleanword->execute($tag,$subfield) unless $test_parameter;
  65                         print "TEST " if $test_parameter;
  66                         print "done\n";
  67                 }
  68         }
  69 }
  70 my $timeneeded = gettimeofday - $starttime;
  71 print "done in $timeneeded seconds\n";