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