added marctype option with XML default

[migration-tools.git] / fingerprinter

diff --git a/fingerprinter b/fingerprinter
index cbcc9f9..71376b9 100755 (executable)
--- a/fingerprinter
+++ b/fingerprinter
@@ -9,7 +9,7 @@ use Unicode::Normalize;
 use MARC::File::XML ( BinaryEncoding => 'utf-8' );
 
 my $conf  = {}; # configuration hashref
-my $count = 0;
+my $count = 0; my $scount = 0;
 my $start = time;
 $| = 1;
 
@@ -48,6 +48,7 @@ for my $file (@ARGV) {
         unless (marc_isvalid($marc))
           { dump_exception($marc); next; }
         dump_fingerprints($marc);
+        $scount++; progress_ticker();
     }
 }
 
@@ -290,6 +291,7 @@ sub initialize {
                          'incoming',
                          'incumbent',
                          'exception|x=s',
+                         'marctype|m=s',
                          'output|o=s',
                          'runtype|r=s',
                          'subfield|s=s',
@@ -304,12 +306,14 @@ 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};
@@ -332,8 +336,9 @@ 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