[open-ils-commits] r341 - conifer/trunk/tools/migration-scripts (dbs)

[svn at svn.open-ils.org]

Author: dbs
Date: 2009-04-16 15:12:35 -0400 (Thu, 16 Apr 2009)
New Revision: 341

Modified:
   conifer/trunk/tools/migration-scripts/fixURIs.pl
Log:
miker says MARC::Batch would be more efficient; okay.
Baby steps towards pipeline processing.


Modified: conifer/trunk/tools/migration-scripts/fixURIs.pl
===================================================================
--- conifer/trunk/tools/migration-scripts/fixURIs.pl	2009-04-16 18:33:42 UTC (rev 340)
+++ conifer/trunk/tools/migration-scripts/fixURIs.pl	2009-04-16 19:12:35 UTC (rev 341)
@@ -1,7 +1,11 @@
 #!/usr/bin/perl -w
 use strict;
+use warnings;
+
+use Error qw/:try/;
 use Getopt::Long;
 use MARC::File::XML( BinaryEncoding => 'utf8', RecordFormat => 'USMARC' );
+use MARC::Batch;
 
 # Clean up URIs from MARCXML records prior to batch ingest
 #   * If we detect a proxy URL:
@@ -23,11 +27,16 @@
     exit();
 }
 
-my $input = MARC::File::XML->in( $input_file );
+my $input = new MARC::Batch( 'XML', $input_file );
+$input->strict_off();
 my $output = MARC::File::XML->out( $output_file );
 
 my ($touched, $url_cnt, $ind1_cnt, $ind2_cnt, $sub9_cnt) = (0, 0, 0, 0, 0);
-while (my $marc = $input->next()) {
+my $marc;
+while ( try { $marc = $input->next() } otherwise { $marc = -1 } ) {
+    # Skip the record if we couldn't even decode it
+    next if ($marc == -1);
+
     my $edited = 0;
     my @uri_fields = $marc->field('856');
     foreach my $uri (@uri_fields) {