misc/stage_biblios_file.pl

   1 #!/usr/bin/perl
   2
   3 use strict;
   4 #use warnings; FIXME - Bug 2505
   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 use C4::Context;
  13 use C4::ImportBatch;
  14 use C4::Matcher;
  15 use Getopt::Long;
  16
  17 $| = 1;
  18
  19 # command-line parameters
  20 my $match_bibs = 0;
  21 my $add_items = 0;
  22 my $input_file = "";
  23 my $batch_comment = "";
  24 my $want_help = 0;
  25 my $no_replace ;
  26
  27 my $result = GetOptions(
  28     'file:s'        => \$input_file,
  29     'match-bibs:s'    => \$match_bibs,
  30     'add-items'     => \$add_items,
  31     'no-replace'    => \$no_replace,
  32     'comment:s'     => \$batch_comment,
  33     'h|help'        => \$want_help
  34 );
  35
  36 if (not $result or $input_file eq "" or $want_help) {
  37     print_usage();
  38     exit 0;
  39 }
  40
  41 unless (-r $input_file) {
  42     die "$0: cannot open input file $input_file: $!\n";
  43 }
  44
  45 my $dbh = C4::Context->dbh;
  46 $dbh->{AutoCommit} = 0;
  47 process_batch($input_file, $match_bibs, $add_items, $batch_comment);
  48 $dbh->commit();
  49
  50 exit 0;
  51
  52 sub process_batch {
  53     my ($input_file, $match_bibs, $add_items, $batch_comment) = @_;
  54
  55     open IN, "<$input_file" or die "$0: cannot open input file $input_file: $!\n";
  56     my $marc_records = "";
  57     $/ = "\035";
  58     my $num_input_records = 0;
  59     while (<IN>) {
  60         s/^\s+//;
  61         s/\s+$//;
  62         next unless $_; # skip if record has only whitespace, as might occur
  63                         # if file includes newlines between each MARC record
  64         $marc_records .= $_; # FIXME - this sort of string concatenation
  65                              # is probably rather inefficient
  66         $num_input_records++;
  67     }
  68     close IN;
  69
  70     my $marc_flavor = C4::Context->preference('marcflavour');
  71
  72     print "... staging MARC records -- please wait\n";
  73     my ($batch_id, $num_valid, $num_items, @import_errors) =
  74         BatchStageMarcRecords($marc_flavor, $marc_records, $input_file, $batch_comment, '', $add_items, 0,
  75                               100, \&print_progress_and_commit);
  76     print "... finished staging MARC records\n";
  77
  78     my $num_with_matches = 0;
  79     if ($match_bibs) {
  80         my $matcher = C4::Matcher->fetch($match_bibs) ;
  81         if (! defined $matcher) {
  82             $matcher = C4::Matcher->new('biblio');
  83             $matcher->add_simple_matchpoint('isbn', 1000, '020', 'a', -1, 0, '');
  84             $matcher->add_simple_required_check('245', 'a', -1, 0, '',
  85                                             '245', 'a', -1, 0, '');
  86         } else {
  87             SetImportBatchMatcher($batch_id, $match_bibs);
  88         }
  89         # set default record overlay behavior
  90         SetImportBatchOverlayAction($batch_id, ($no_replace) ? 'ignore' : 'replace');
  91         SetImportBatchNoMatchAction($batch_id, 'create_new');
  92         SetImportBatchItemAction($batch_id, 'always_add');
  93         print "... looking for matches with records already in database\n";
  94         $num_with_matches = BatchFindBibDuplicates($batch_id, $matcher, 10, 100, \&print_progress_and_commit);
  95         print "... finished looking for matches\n";
  96     }
  97
  98     my $num_invalid_bibs = scalar(@import_errors);
  99     print <<_SUMMARY_;
 100
 101 MARC record staging report
 102 ------------------------------------
 103 Input file:              $input_file
 104 Number of input bibs:    $num_input_records
 105 Number of valid bibs:    $num_valid
 106 Number of invalid bibs:  $num_invalid_bibs
 107 _SUMMARY_
 108     if ($match_bibs) {
 109         print "Number of bibs matched:  $num_with_matches\n";
 110     } else {
 111         print "Incoming bibs not matched against existing bibs (--match-bibs option not supplied)\n";
 112     }
 113     if ($add_items) {
 114         print "Number of items parsed:  $num_items\n";
 115     } else {
 116         print "No items parsed (--add-items option not supplied)\n";
 117     }
 118
 119     print "\n";
 120     print "Batch number assigned:  $batch_id\n";
 121     print "\n";
 122 }
 123
 124 sub print_progress_and_commit {
 125     my $recs = shift;
 126     $dbh->commit();
 127     print "... processed $recs records\n";
 128 }
 129
 130 sub print_usage {
 131     print <<_USAGE_;
 132 $0: stage MARC bib file into reservoir.
 133
 134 Use this batch job to load a file of MARC bibliographic records
 135 (with optional item information) into the Koha reservoir.
 136
 137 After running this program to stage your file, you can use
 138 either the batch job commit_biblios_file.pl or the Koha
 139 Tools option "Manage Staged MARC Records" to load the
 140 records into the main Koha database.
 141
 142 Parameters:
 143     --file <file_name>      name of input MARC bib file
 144     --match-bibs <match_id> use this option to match bibs
 145                             in the file with bibs already in
 146                             the database for future overlay.
 147                             If <match_id> isn't defined, a default
 148                             MARC21 ISBN & title match rule will be applied.
 149     --add-items             use this option to specify that
 150                             item data is embedded in the MARC
 151                             bibs and should be parsed.
 152     --no-replace            overlay action for bib record: default is to
 153                             replace extant bib with the imported record.
 154     --comment <comment>     optional comment to describe
 155                             the record batch; if the comment
 156                             has spaces in it, surround the
 157                             comment with quotation marks.
 158     --help or -h            show this message.
 159 _USAGE_
 160 }