misc/batchCompareMARCvsFrameworks.pl

   1 #!/usr/bin/perl
   2 # small script that dumps an iso2709 file.
   3
   4
   5 use strict;
   6 #use warnings; FIXME - Bug 2505
   7 BEGIN {
   8     # find Koha's Perl modules
   9     # test carefully before changing this
  10     use FindBin;
  11     eval { require "$FindBin::Bin/kohalib.pl" };
  12 }
  13
  14 # Koha modules used
  15 use Koha::Script;
  16 use C4::Context;
  17 use MARC::File::USMARC;
  18 use MARC::Record;
  19 use MARC::Batch;
  20
  21 use Getopt::Long;
  22 use IO::File;
  23
  24 my ( $input_marc_file,$number,$nowarning,$frameworkcode) = ('',0);
  25 my $version;
  26 GetOptions(
  27     'file:s'    => \$input_marc_file,
  28     'n:s' => \$number,
  29     'v' => \$version,
  30     'w' => \$nowarning,
  31     'c' => \$frameworkcode,
  32 );
  33
  34 $frameworkcode="" unless $frameworkcode;
  35
  36 if ($version || ($input_marc_file eq '')) {
  37     print <<EOF
  38 This script compares an iso2709 file and Koha's MARC frameworks
  39 It will show the marc fields/subfields used in Koha, and that
  40 are not in the iso2709 file and which fields/subfields that are
  41 used in the iso2709 file and not in Koha.
  42
  43 parameters :
  44 \tv : this version/help screen
  45 \tfile /path/to/file/to/dump : the file to dump
  46 \tw : warning and strict off. If your dump fails, try -w option. It it works, then, the file is iso2709, but a buggy one !
  47 \tc : the frameworkcode. If omitted, set to ""
  48
  49 SAMPLE : ./compare_iso_and_marc_parameters.pl -file /home/paul/koha.dev/local/npl -n 1
  50
  51 EOF
  52 ;
  53 die;
  54 }#/
  55
  56 my $fh = IO::File->new($input_marc_file); # don't let MARC::Batch open the file, as it applies the ':utf8' IO layer
  57 my $batch = MARC::Batch->new( 'USMARC', $fh );
  58 $batch->warnings_off() unless $nowarning;
  59 $batch->strict_off() unless $nowarning;
  60 my $dbh=C4::Context->dbh;
  61 my $sth = $dbh->prepare("select tagfield,tagsubfield,tab from marc_subfield_structure where frameworkcode=?");
  62 $sth->execute($frameworkcode);
  63
  64 my %hash_unused;
  65 my %hash_used;
  66 while (my ($tagfield,$tagsubfield,$tab) = $sth->fetchrow) {
  67     $hash_unused{"$tagfield$tagsubfield"} = 1 if ($tab eq -1);
  68     $hash_used{"$tagfield$tagsubfield"} = 1 if ($tab ne -1);
  69 }
  70 my $i=0;
  71 while ( my $record = $batch->next() ) {
  72     $i++;
  73     foreach my $MARCfield ($record->fields()) {
  74     next if $MARCfield->is_control_field(); # tag num < 10
  75         if ($MARCfield) {
  76             foreach my $fields ($MARCfield->subfields()) {
  77                 if ($fields) {
  78                     if ($hash_unused{$MARCfield->tag().@$fields[0]}>=1) {
  79                         $hash_unused{$MARCfield->tag().@$fields[0]}++;
  80                     }
  81                     if ($hash_used{$MARCfield->tag().@$fields[0]}>=1) {
  82                         $hash_used{$MARCfield->tag().@$fields[0]}++;
  83                     }
  84                 }
  85     #           foreach my $field (@$fields) {
  86     #               warn "==>".$MARCfield->tag().@$fields[0];
  87     #           }
  88             }
  89         }
  90     }
  91 }
  92 print "Undeclared tag/subfields that exists in the file\n";
  93 print "================================================\n";
  94 foreach my $key (sort keys %hash_unused) {
  95     print "$key => ".($hash_unused{$key}-1)."\n" unless ($hash_unused{$key}==1);
  96 }
  97
  98 print "Declared tag/subfields unused in the iso2709 file\n";
  99 print "=================================================\n";
 100 foreach my $key (sort keys %hash_used) {
 101     print "$key => ".($hash_used{$key}-1)."\n" if ($hash_used{$key}==1);
 102 }
 103
 104 # foreach my $x (sort keys %resB) {
 105 #   print "$x => ".$resB{$x}."\n";
 106 # }
 107 print "\n==================\n$i record parsed\n";