X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=blobdiff_plain;f=fingerprinter;h=dcf2e29e63f4bdb8611a0419926fa8b6f39d40f8;hp=6e04e0ba89cf4eb08983896c79f2ff01755dd0ee;hb=26d762670d7610ed9ccb1c36a8dbab3169e4e2a3;hpb=4ee76d3146c0b0f2b9c42b04dac2487c777e9165

diff --git a/fingerprinter b/fingerprinter
index 6e04e0b..dcf2e29 100755
--- a/fingerprinter
+++ b/fingerprinter
@@ -27,7 +27,15 @@ for my $file (@ARGV) {
     $batch->strict_off();
     $batch->warnings_off();
 
-    while ( $record = $batch->next ) {
+    my $record;
+    while ( 1 ) {
+        eval { $record = $batch->next; };
+        if ($@) {
+            import MARC::File::XML;
+            print "bad record\n";
+            next;
+        }
+        last unless $record;
         $count++; progress_ticker();
         my $marc = undef;
         unless ( defined $record )
@@ -88,11 +96,14 @@ sub populate_marc {
     }
     unless ($marc{date1} and $marc{date1} =~ /\d{4}/) {
         my $my_260 = $record->field('260');
-        my $date1 = $my_260->subfield('c') if $my_260;
-        if (defined $date1 and $date1 =~ /\d{4}/) {
-            $marc{date1} = $date1;
-            $marc{fudgedate} = 1;
-            print XF ">> using 260c as date1 at rec $count\n";
+        if ($my_260 and $my_260->subfield('c')) {
+            my $date1 = $my_260->subfield('c');
+            $date1 =~ s/\D//g;
+            if (defined $date1 and $date1 =~ /\d{4}/) {
+                $marc{date1} = $date1;
+                $marc{fudgedate} = 1;
+                print XF ">> using 260c as date1 at rec $count\n";
+            }
         }
     }
 
@@ -130,6 +141,11 @@ sub populate_marc {
           if (defined $oclc and $oclc =~ /\(OCoLC\)/ and $oclc =~/([0-9]+)/);
     }
 
+    if ($record->field('999')) {
+        my $koha_bib_id = $record->field('999')->subfield('c');
+        $marc{koha_bib_id} = $koha_bib_id if defined $koha_bib_id and $koha_bib_id =~ /^\d+$/;
+    }
+
     # "Accompanying material" and check for "copy" (300)
     if ($record->field('300')) {
         $marc{accomp} = $record->field('300')->subfield('e');
@@ -317,6 +333,14 @@ sub dump_fingerprints {
         }
     }
 
+    if ($conf->{fingerprints}{koha_bib_id} and exists $marc->{koha_bib_id}) {
+        print OF join("\t", $marc->{score}, $marc->{id}, "z_koha_bib_id",
+                      $marc->{item_form}, $marc->{date1},
+                      $marc->{record_type},
+                      $marc->{bib_lvl}, $marc->{title},
+                      $marc->{koha_bib_id}), "\n";
+    }
+
     if ($conf->{fingerprints}{isbn}) {
         if ((scalar @{ $marc->{isbns} } > 0) and $marc->{pages}) {
             foreach my $isbn ( @{ $marc->{isbns}} ) {
@@ -477,6 +501,7 @@ sub initialize {
         my %valid_fps = ( oclc => 1, isbn => 1, issn => 1, lccn => 1,
                           edition => 1, accomp => 1, authpub => 1,
                           baseline => 1, crap => 1,
+                          koha_bib_id => 1,
                         );
         for (split /,/, $c->{fingerprints}) {
             die "Invalid fingerprint '$_'\n" unless $valid_fps{$_};
@@ -565,7 +590,7 @@ Options
 
   --fingerprints=LIST  Fingerprints to generate, comma separated
                        Default: oclc,isbn,edition,issn,lccn,accomp,authpub
-                       Others:  baseline
+                       Others:  baseline,koha_bib_id
   --excludelist=FILE   Name of fingerprints exclusions file
 
   --scores=LIST  Scores to calculate, comma separated