folding alternative date searching into mainline fingerprinter

[migration-tools.git] / fingerprinter

diff --git a/fingerprinter b/fingerprinter
index feb6c19..06f62f7 100755 (executable)
--- a/fingerprinter
+++ b/fingerprinter
@@ -9,8 +9,7 @@ use Unicode::Normalize;
 use MARC::File::XML ( BinaryEncoding => 'utf-8' );
 
 my $conf  = {}; # configuration hashref
-my $count = 0;
-my $failcount = 0;
+my $count = 0; my $scount = 0;
 my $start = time;
 $| = 1;
 
@@ -23,16 +22,15 @@ binmode(XF, ':utf8');
 
 for my $file (@ARGV) {
     print XF "Processing $file\n";
-    open my $records, '<:utf8', $file;
+    #open my $records, '<:utf8', $file; # This hack dosn't play nice with the use MARC::File::XML ( BinaryEncoding => 'utf-8' ); hack
     my $batch = undef; my $record = undef;
 
-    $batch = MARC::Batch->new('XML', $records);
+    #$batch = MARC::Batch->new('XML', $records); # The other part of the hack
+    $batch = MARC::Batch->new($conf->{marctype}, $file);
     $batch->strict_off();
     $batch->warnings_off();
 
-    while ( eval { $record = $batch->next } ) {
-        if ($@)
-          { print XF "MARC::Batch error $@"; $failcount++; next }
+    while ( $record = $batch->next ) {
         $count++; progress_ticker();
         my $marc = undef;
         unless ( defined $record )
@@ -50,15 +48,11 @@ for my $file (@ARGV) {
         unless (marc_isvalid($marc))
           { dump_exception($marc); next; }
         dump_fingerprints($marc);
+        $scount++; progress_ticker();
     }
 }
 
-unless ($conf->{quiet}) {
-    print "\nSuccessfully processed:\t$count\n";
-    print "Failed to process:\t$failcount\n";
-    print "Total: ", ($count + $failcount), "\n";
-}
-
+print "\nSuccessfully processed:\t$count\n" unless $conf->{quiet};
 
 =head2 populate_marc
 
@@ -86,6 +80,15 @@ sub populate_marc {
     $marc{date1} = substr($my_008,7,4) if ($my_008);
     $marc{date2} = substr($my_008,11,4) if ($my_008); # UNUSED
 
+    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;
+            print XF ">> using 260c as date1 on $id\n";
+        }
+    }
+
     # item_form
     if ( $marc{record_type} =~ /[gkroef]/ ) { # MAP, VIS
         $marc{item_form} = substr($my_008,29,1) if ($my_008);
@@ -297,6 +300,7 @@ sub initialize {
                          'incoming',
                          'incumbent',
                          'exception|x=s',
+                         'marctype|m=s',
                          'output|o=s',
                          'runtype|r=s',
                          'subfield|s=s',
@@ -311,13 +315,17 @@ sub initialize {
     if ($c->{incoming}) {
         $c->{tag} = 903 unless defined $c->{tag};
         $c->{subfield} = 'a' unless defined $c->{subfield};
+        $c->{marctype} = 'XML' unless defined $c->{marctype};
         $c->{output} = 'incoming.fp' unless defined $c->{output};
         $c->{exception} = 'incoming.ex' unless defined $c->{exception};
+        $c->{runtype} = 'full' unless defined $c->{runtype};
     } elsif ($c->{incumbent}) {
         $c->{tag} = 901 unless defined $c->{tag};
         $c->{subfield} = 'c' unless defined $c->{subfield};
+        $c->{marctype} = 'XML' unless defined $c->{marctype};
         $c->{output} = 'incumbent.fp' unless defined $c->{output};
         $c->{exception} = 'incumbent.ex' unless defined $c->{exception};
+        $c->{runtype} = 'full' unless defined $c->{runtype};
     }
 
     my @keys = keys %{$c};
@@ -337,8 +345,9 @@ sub initialize {
 
 sub progress_ticker {
     return if $conf->{quiet};
-    printf("> %d (%d/s)\r", $count, ($count / (time - $start + 1)))
-      if ($count % 100 == 0);
+    printf("\r> %d recs seen; %d processed", $count, $scount);
+    printf(" (%d/s)", ($count / (time - $start + 1)))
+      if ($count % 500 == 0);
 }
 
 =head2 show_help