misc/batchDeleteUnusedSubfields.pl

   1 #!/usr/bin/perl
   2 # small script that rebuilds the non-MARC DB
   3
   4 use strict;
   5 #use warnings; FIXME - Bug 2505
   6 BEGIN {
   7     # find Koha's Perl modules
   8     # test carefully before changing this
   9     use FindBin;
  10     eval { require "$FindBin::Bin/kohalib.pl" };
  11 }
  12
  13 # Koha modules used
  14 use Koha::Script;
  15 use MARC::Record;
  16 use C4::Context;
  17 use C4::Biblio;
  18 use Time::HiRes qw(gettimeofday);
  19
  20 use Getopt::Long;
  21 my ( $input_marc_file, $number) = ('',0);
  22 my ($version, $confirm,$test_parameter);
  23 GetOptions(
  24         'c' => \$confirm,
  25         'h' => \$version,
  26         't' => \$test_parameter,
  27 );
  28
  29 if ($version || (!$confirm)) {
  30         print <<EOF
  31 This script cleans unused subfields in the MARC DB.
  32 If you alter the MARC parameters and remove a subfield (ie : move it to ignore (10) tab), existing subfields are NOT removed.
  33 It's not a bug, it prevents deleting useful values in case of erroneous move.
  34 This script definitely remove unused subfields in the MARC DB.
  35 syntax :
  36 \t./cleanmarcdb.pl -h (or without arguments => shows this screen)
  37 \t./cleanmarcdb.pl -c (c like confirm => cleans the marc DB (may be long)
  38 \t-t => test only, change nothing in DB
  39 EOF
  40 ;#'
  41 die;
  42 }
  43
  44 my $dbh = C4::Context->dbh;
  45 my $i=0;
  46 my $starttime = gettimeofday;
  47 my $cleansubfield = $dbh->prepare("delete from marc_subfield_table where tag=? and subfieldcode=?");
  48 my $cleanword = $dbh->prepare("delete from marc_word where tag=? and subfieldid=?");
  49
  50 # get tags structure
  51 my $tags = GetMarcStructure(1);
  52 foreach my $tag (sort keys(%{$tags})) {
  53         foreach my $subfield (sort keys(%{$tags->{$tag}})) {
  54                 next if $subfield eq "lib";
  55                 next if $subfield eq "mandatory";
  56                 next if $subfield eq "tab";
  57                 # DO NOT drop biblionumber, biblioitemnumber and itemnumber.
  58                 # they are stored internally, and are mapped to tab -1. This script must keep them or it will completly break Koha DB !!!
  59                 next if ($tags->{$tag}->{$subfield}->{kohafield} eq "biblio.biblionumber");
  60                 next if ($tags->{$tag}->{$subfield}->{kohafield} eq "biblioitems.biblioitemnumber");
  61                 next if ($tags->{$tag}->{$subfield}->{kohafield} eq "items.itemnumber");
  62                 # now, test => if field is ignored (in tab -1 or '') => drop everything in the MARC table !
  63                 if ($tags->{$tag}->{$subfield}->{tab} eq -1 || $tags->{$tag}->{$subfield}->{tab} eq '') {
  64                         print "dropping $tag \$ $subfield\n";
  65                         $cleansubfield->execute($tag,$subfield) unless $test_parameter;
  66                         $cleanword->execute($tag,$subfield) unless $test_parameter;
  67                         print "TEST " if $test_parameter;
  68                         print "done\n";
  69                 }
  70         }
  71 }
  72 my $timeneeded = gettimeofday - $starttime;
  73 print "done in $timeneeded seconds\n";