misc/batchCompareMARCvsFrameworks.pl

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