X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=blobdiff_plain;f=fingerprinter;h=f388174a447f8bc0554d384aa58b9389b71a4eb2;hp=a42efc1bc67b8313a5aec71419ca12f10d1f37f8;hb=af66e5408e31e25bc6fed62b99cc39be95d480ef;hpb=55223e093c954354c1f28dca9c14de5a9d0d88da

diff --git a/fingerprinter b/fingerprinter
index a42efc1..f388174 100755
--- a/fingerprinter
+++ b/fingerprinter
@@ -25,6 +25,7 @@ use MARC::Batch;
 use Unicode::Normalize;
 use MARC::File::XML ( BinaryEncoding => 'utf-8' );
 use Equinox::Migration::SubfieldMapper;
+use Equinox::Migration::Utils qw/normalize_oclc_number/;
 
 my $conf  = {}; # configuration hashref
 my $count = 0; my $scount = 0;
@@ -34,7 +35,7 @@ $| = 1;
 initialize($conf);
 
 open OF, '>', $conf->{output} or die "$0: cannot open output file $conf->{output}: $!\n";
-open XF, '>', $conf->{exception} or die "$0: cannot open exception file $conf->{output}: $!\n";
+open XF, '>', $conf->{exception} or die "$0: cannot open exception file $conf->{exception}: $!\n";
 
 for my $file (@ARGV) {
     print XF "Processing $file\n";
@@ -148,13 +149,19 @@ sub populate_marc {
 
     # oclc
     $marc{oclc} = [];
-    push @{ $marc{oclc} }, $record->field('001')->as_string()
-      if ($record->field('001') and $record->field('003') and
-          $record->field('003')->as_string() =~ /OCo{0,1}LC/);
+    if ($record->field('001') &&
+        $record->field('003') &&
+        $record->field('003')->as_string() =~ /OCo{0,1}LC/ &&
+        defined normalize_oclc_number($record->field('001')->as_string())) {
+        push @{ $marc{oclc} }, normalize_oclc_number($record->field('001')->as_string());
+    }
     for ($record->field('035')) {
         my $oclc = $_->subfield('a');
-        push @{ $marc{oclc} }, $oclc
-          if (defined $oclc and $oclc =~ /\(OCoLC\)/ and $oclc =~/([0-9]+)/);
+        if (defined $oclc &&
+            ($oclc =~ /\(OCoLC\)/ || $oclc =~ /(ocm|ocl7|ocn|on)/) &&
+            defined normalize_oclc_number($oclc)) {
+            push @{ $marc{oclc} }, normalize_oclc_number($oclc);
+        }
     }
 
     if ($record->field('999')) {
@@ -373,11 +380,11 @@ sub dump_fingerprints {
         }
     }
 
-    if ($conf->{fingerprints}{edition} and $marc->{edition}) {
+    if ($conf->{fingerprints}{edition} and $marc->{edition} and $marc->{author}) {
         print OF join("\t", $marc->{score}, $marc->{id}, "edition",
                       $marc->{item_form}, $marc->{date1},
                       $marc->{record_type}, $marc->{bib_lvl},
-                      $marc->{title}, $marc->{edition}), "\n";
+                      $marc->{title}, $marc->{author}, $marc->{edition}), "\n";
     }
 
     if ($conf->{fingerprints}{issn} and $marc->{issn}) {