removing miker-filter series; superceded by filter_record_ids, which has far less...

[migration-tools.git] / fingerprinter

diff --git a/fingerprinter b/fingerprinter
index 681ca46..1fa1f73 100755 (executable)
--- a/fingerprinter
+++ b/fingerprinter
@@ -16,9 +16,7 @@ $| = 1;
 initialize($conf);
 
 open OF, '>', $conf->{output};
-binmode(OF, ':utf8');
 open XF, '>', $conf->{exception};
-binmode(XF, ':utf8');
 
 for my $file (@ARGV) {
     print XF "Processing $file\n";
@@ -210,6 +208,7 @@ sub marc_isvalid {
 
 sub dump_fingerprints {
     my ($marc) = @_;
+    my $good_fp = 0;
 
     if ($conf->{runtype} eq "primary") {
         print OF join("\t",$marc->{id}, $marc->{item_form},
@@ -225,9 +224,9 @@ sub dump_fingerprints {
                                   $marc->{bib_lvl}, $marc->{title},
                                   $isbn, $marc->{pages}), "\n";
             }
+            $good_fp = 1;
         }
 
-
 =pod
 
 Looks like case b isn't a terribly good idea most of the time -- it
@@ -239,6 +238,7 @@ for now, simply turning it off.
                               $marc->{item_form}, $marc->{date1},
                               $marc->{record_type}, $marc->{bib_lvl},
                               $marc->{title}, $marc->{edition}), "\n";
+            $good_fp = 1;
         }
 
 =cut
@@ -248,6 +248,7 @@ for now, simply turning it off.
                               $marc->{item_form}, $marc->{date1},
                               $marc->{record_type}, $marc->{bib_lvl},
                               $marc->{title}, $marc->{issn}), "\n";
+            $good_fp = 1;
         }
 
         if ($marc->{lccn}) { # case d : lccn
@@ -255,6 +256,7 @@ for now, simply turning it off.
                               $marc->{item_form}, $marc->{date1},
                               $marc->{record_type}, $marc->{bib_lvl},
                               $marc->{title}, $marc->{lccn}) ,"\n";
+            $good_fp = 1;
         }
 
         if ($marc->{accomp}) { # case e : accomp
@@ -262,6 +264,7 @@ for now, simply turning it off.
                               $marc->{item_form}, $marc->{date1},
                               $marc->{record_type}, $marc->{bib_lvl},
                               $marc->{title}, $marc->{accomp}) ,"\n";
+            $good_fp = 1;
         }
 
         # case o: oclc
@@ -272,6 +275,7 @@ for now, simply turning it off.
                               $marc->{record_type}, $marc->{bib_lvl},
                               $marc->{title}, $_, "\n");
             }
+            $good_fp = 1;
         }
 
         # case z : author, publisher, pubyear, pages
@@ -283,6 +287,14 @@ for now, simply turning it off.
                               $marc->{title}, $marc->{author},
                               $marc->{publisher}, $marc->{pubyear},
                               $marc->{pages}), "\n";
+            $good_fp = 1;
+        }
+
+        # case poo : nothing good; dump a primary and move on
+        unless ($good_fp) {
+            print OF join("\t", $marc->{id}, "case poo", $marc->{item_form},
+                          $marc->{date1}, $marc->{record_type},
+                          $marc->{bib_lvl}, $marc->{title}), "\n";
         }
     }
 }