gah

[migration-tools.git] / fingerprinter

diff --git a/fingerprinter b/fingerprinter
index c0a8741..e228c6e 100755 (executable)
--- a/fingerprinter
+++ b/fingerprinter
@@ -74,7 +74,7 @@ sub populate_marc {
 
     # date1, date2
     my $my_008 = $record->field('008');
-    $marc{tag008} = $my_008->as_string();# if ($my_008);
+    $marc{tag008} = $my_008->as_string() if ($my_008);
     if (defined $marc{tag008}) {
         unless (length $marc{tag008} == 40) {
             $marc{tag008} = $marc{tag008} . ('|' x (40 - length($marc{tag008})));
@@ -226,6 +226,8 @@ sub score_marc {
     # static criteria scoring
     #----------------------------------
     $marc->{misc_score} = 999;
+    $marc->{age_score}  = 999999999999;
+
     # -1 if 008 has been padded, -2 if it doesn't exist
     if ($marc->{tag008})
       { $marc->{misc_score}-- if ($marc->{tag008} =~ /\|$/) }
@@ -237,6 +239,19 @@ sub score_marc {
     $marc->{misc_score}--
       if (defined $marc->{tag300a} and $marc->{tag300a} =~ /copy/i);
 
+    # subtract record id if we want older records to win
+    #$marc->{age_score} -= $marc->{id} unless ($conf->{newwins});
+    # handle arbitrary adjustments
+    $marc->{age_score} = 1;
+    if ($conf->{'arbitrarily-lose-above'}) {
+        $marc->{age_score} = 0
+          if ($marc->{id} >= $conf->{'arbitrarily-lose-above'});
+    }
+    if ($conf->{'arbitrarily-lose-below'}) {
+        $marc->{age_score} = 0
+          if ($marc->{id} <= $conf->{'arbitrarily-lose-below'});
+    }
+
     #----------------------------------
     # dynamic calculated scoring
     #----------------------------------
@@ -273,7 +288,7 @@ sub score_marc {
     }
     $json .= 'misc:' . $marc->{misc_score} . '}';
 
-    my $compact = join('', $marc->{misc_score}, @score);
+    my $compact = join('-', $marc->{age_score}, $marc->{misc_score}, @score);
     $marc->{score} = "$compact\t$json";
 }
 
@@ -364,16 +379,16 @@ sub dump_exception {
         print XF "Undefined record at line $count; likely bad XML\n";
         return;
     }
-    print XF "Record ", $marc->{id}, " did not make the cut: ";
-    print XF "Missing item_form. " unless ($marc->{item_form});
+    print XF "Record ", $marc->{id}, " excluded: ";
+    print XF "missing item_form; " unless ($marc->{item_form});
     unless (defined $marc->{date1})
-      { print XF "Missing date1. " }
+      { print XF "missing date1; " }
     else
-      { print XF "Invalid date1: ", $marc->{date1}, " "
+      { print XF "invalid date1: '", $marc->{date1}, "'; "
           unless ($marc->{date1} =~ /\d{4}/); }
-    print XF "Missing record_type. " unless ($marc->{record_type});
-    print XF "Missing bib_lvl. " unless ($marc->{bib_lvl});
-    print XF "Missing title. " unless ($marc->{title});
+    print XF "missing record_type; " unless ($marc->{record_type});
+    print XF "missing bib_lvl; " unless ($marc->{bib_lvl});
+    print XF "missing title " unless ($marc->{title});
     print XF "\n";
 }
 
@@ -401,6 +416,9 @@ sub initialize {
                          'tag|t=s',
                          'fingerprints=s',
                          'scores=s',
+                         'arbitrarily-lose-above=i',
+                         'arbitrarily-lose-below=i',
+                         'newwins',
                          'quiet|q',
                          'help|h',
                        );
@@ -499,6 +517,12 @@ Options
 
   --scores=LIST  Scores to calculate, comma separated
                  Default: oclc,dlc,num_650,num_tags,enc_level
+  --newwins      New record IDs score higher (default is old wins)
+  --arbitrarily-lose-above
+  --arbitrarily-lose-below
+  --arbitrarily-decrease-score-by
+      Modify fingerprint scoring of records whose EG id is above or below a
+      given value, inclusive (so 5 is <= 5 or >= 5) such that they lose.
 
   --marctype=TYPE Defaults to 'XML'
 HELP