if but not if defined?

[migration-tools.git] / fingerprinter

diff --git a/fingerprinter b/fingerprinter
index 7cadd53..a08efea 100755 (executable)
--- a/fingerprinter
+++ b/fingerprinter
@@ -11,6 +11,7 @@ use Unicode::Normalize;
 
 my $conf  = {}; # configuration hashref
 my $count = 0;
+my $start = time;
 $| = 1;
 
 initialize($conf);
@@ -74,7 +75,7 @@ sub populate_marc {
     # date1, date2
     my $my_008 = $record->field('008');
     $my_008 = $my_008->as_string() if ($my_008);
-    unless (length $my_008 == 40)
+    unless (defined $my_008 and length $my_008 == 40)
       { print XF ">> Bad 008 field length in rec. $id\n"; return \%marc }
     $marc{date1} = substr($my_008,7,4) if ($my_008);
     $marc{date2} = substr($my_008,11,4) if ($my_008); # UNUSED
@@ -102,7 +103,9 @@ sub populate_marc {
         }
     }
 
-    
+    # "Accompanying material" (300e)
+    $marc{accomp} = $record->field('300')->subfield('e')
+      if $record->field('300');
 
     # issn, lccn, title, desc, pages, pub, pubyear, edition
     $marc{lccn} = $record->field('010')->subfield('a') if $record->field('010');
@@ -110,7 +113,7 @@ sub populate_marc {
     $marc{desc} = $record->field('300')->subfield('a') if $record->field('300');
     $marc{pages} = $1 if (defined $marc{desc} and $marc{desc} =~ /(\d+)/);
     $marc{title} = $record->field('245')->subfield('a')
-      if defined $record->field('245');
+      if $record->field('245');
     $marc{edition} = $record->field('250')->subfield('a')
       if $record->field('250');
     if ($record->field('260')) {
@@ -222,7 +225,14 @@ sub dump_fingerprints {
                               $marc->{title}, $marc->{lccn}) ,"\n";
         }
 
-        # case e : author, publisher, pubyear, pages
+        if ($marc->{accomp}) { # case e : accomp
+            print OF join("\t", $marc->{id}, "case d",
+                              $marc->{item_form}, $marc->{date1},
+                              $marc->{record_type}, $marc->{bib_lvl},
+                              $marc->{title}, $marc->{accomp}) ,"\n";
+        }
+
+        # case z : author, publisher, pubyear, pages
         if ($marc->{author} and $marc->{publisher} and $marc->{pubyear}
             and $marc->{pages}) {
             print OF join("\t", $marc->{id}, "case e",
@@ -251,9 +261,11 @@ sub dump_exception {
     }
     print XF "Record ", $marc->{id}, " did not make the cut: ";
     print XF "Missing item_form. " unless ($marc->{item_form});
-    print XF "Missing date1. " unless (defined $marc->{date1});
-    print XF "Invalid date1: ", $marc->{date1}
-      unless ($marc->{date1} =~ /\d{4}/);
+    unless (defined $marc->{date1})
+      { print XF "Missing date1. " }
+    else
+      { 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});
@@ -306,13 +318,8 @@ sub initialize {
 
 sub progress_ticker {
     return if $conf->{quiet};
-
-    if ($count % 100 == 0) {
-        print '|';
-        print " $count \n" unless ($count % 1400);
-    } elsif ($count % 20 == 0) {
-        print '.';
-    }
+    printf("> %d (%d/s)\r", $count, ($count / (time - $start + 1)))
+      if ($count % 100 == 0);
 }
 
 =head2 show_help