misc/stage_biblios_file.pl

   1 #!/usr/bin/perl
   2
   3 use strict;
   4 BEGIN {
   5     # find Koha's Perl modules
   6     # test carefully before changing this
   7     use FindBin;
   8     eval { require "$FindBin::Bin/kohalib.pl" };
   9 }
  10
  11 use C4::Context;
  12 use C4::ImportBatch;
  13 use C4::Matcher;
  14 use Getopt::Long;
  15
  16 $| = 1;
  17
  18 # command-line parameters
  19 my $match_bibs = 0;
  20 my $add_items = 0;
  21 my $input_file = "";
  22 my $batch_comment = "";
  23 my $want_help = 0;
  24 my $no_replace ;
  25
  26 my $result = GetOptions(
  27     'file:s'        => \$input_file,
  28     'match-bibs:s'    => \$match_bibs,
  29     'add-items'     => \$add_items,
  30     'no-replace'    => \$no_replace,
  31     'comment:s'     => \$batch_comment,
  32     'h|help'        => \$want_help
  33 );
  34
  35 if (not $result or $input_file eq "" or $want_help) {
  36     print_usage();
  37     exit 0;
  38 }
  39
  40 unless (-r $input_file) {
  41     die "$0: cannot open input file $input_file: $!\n";
  42 }
  43
  44 my $dbh = C4::Context->dbh;
  45 $dbh->{AutoCommit} = 0;
  46 process_batch($input_file, $match_bibs, $add_items, $batch_comment);
  47 $dbh->commit();
  48
  49 exit 0;
  50
  51 sub process_batch {
  52     my ($input_file, $match_bibs, $add_items, $batch_comment) = @_;
  53
  54     open IN, "<$input_file" or die "$0: cannot open input file $input_file: $!\n";
  55     my $marc_records = "";
  56     $/ = "\035";
  57     my $num_input_records = 0;
  58     while (<IN>) {
  59         s/^\s+//;
  60         s/\s+$//;
  61         next unless $_; # skip if record has only whitespace, as might occur
  62                         # if file includes newlines between each MARC record
  63         $marc_records .= $_; # FIXME - this sort of string concatenation
  64                              # is probably rather inefficient
  65         $num_input_records++;
  66     }
  67     close IN;
  68
  69     my $marc_flavor = C4::Context->preference('marcflavour');
  70
  71     print "... staging MARC records -- please wait\n";
  72     my ($batch_id, $num_valid, $num_items, @import_errors) =
  73         BatchStageMarcRecords($marc_flavor, $marc_records, $input_file, $batch_comment, '', $add_items, 0,
  74                               100, \&print_progress_and_commit);
  75     print "... finished staging MARC records\n";
  76
  77     my $num_with_matches = 0;
  78     if ($match_bibs) {
  79         my $matcher = C4::Matcher->fetch($match_bibs) ;
  80         if (! defined $matcher) {
  81             $matcher = C4::Matcher->new('biblio');
  82             $matcher->add_simple_matchpoint('isbn', 1000, '020', 'a', -1, 0, '');
  83             $matcher->add_simple_required_check('245', 'a', -1, 0, '',
  84                                             '245', 'a', -1, 0, '');
  85         } else {
  86             SetImportBatchMatcher($batch_id, $match_bibs);
  87         }
  88         # set default record overlay behavior
  89         SetImportBatchOverlayAction($batch_id, ($no_replace) ? 'ignore' : 'replace');
  90         SetImportBatchNoMatchAction($batch_id, 'create_new');
  91         SetImportBatchItemAction($batch_id, 'always_add');
  92         print "... looking for matches with records already in database\n";
  93         $num_with_matches = BatchFindBibDuplicates($batch_id, $matcher, 10, 100, \&print_progress_and_commit);
  94         print "... finished looking for matches\n";
  95     }
  96
  97     my $num_invalid_bibs = scalar(@import_errors);
  98     print <<_SUMMARY_;
  99
 100 MARC record staging report
 101 ------------------------------------
 102 Input file:              $input_file
 103 Number of input bibs:    $num_input_records
 104 Number of valid bibs:    $num_valid
 105 Number of invalid bibs:  $num_invalid_bibs
 106 _SUMMARY_
 107     if ($match_bibs) {
 108         print "Number of bibs matched:  $num_with_matches\n";
 109     } else {
 110         print "Incoming bibs not matched against existing bibs (--match-bibs option not supplied)\n";
 111     }
 112     if ($add_items) {
 113         print "Number of items parsed:  $num_items\n";
 114     } else {
 115         print "No items parsed (--add-items option not supplied)\n";
 116     }
 117
 118     print "\n";
 119     print "Batch number assigned:  $batch_id\n";
 120     print "\n";
 121 }
 122
 123 sub print_progress_and_commit {
 124     my $recs = shift;
 125     $dbh->commit();
 126     print "... processed $recs records\n";
 127 }
 128
 129 sub print_usage {
 130     print <<_USAGE_;
 131 $0: stage MARC bib file into reservoir.
 132
 133 Use this batch job to load a file of MARC bibliographic records
 134 (with optional item information) into the Koha reservoir.
 135
 136 After running this program to stage your file, you can use
 137 either the batch job commit_biblios_file.pl or the Koha
 138 Tools option "Manage Staged MARC Records" to load the
 139 records into the main Koha database.
 140
 141 Parameters:
 142     --file <file_name>      name of input MARC bib file
 143     --match-bibs <match_id> use this option to match bibs
 144                             in the file with bibs already in
 145                             the database for future overlay.
 146                             If <match_id> isn't defined, a default
 147                             MARC21 ISBN & title match rule will be applied.
 148     --add-items             use this option to specify that
 149                             item data is embedded in the MARC
 150                             bibs and should be parsed.
 151     --no-replace            overlay action for bib record: default is to
 152                             replace extant bib with the imported record.
 153     --comment <comment>     optional comment to describe
 154                             the record batch; if the comment
 155                             has spaces in it, surround the
 156                             comment with quotation marks.
 157     --help or -h            show this message.
 158 _USAGE_
 159 }