track failed (evel fail) records

[migration-tools.git] / spit_csv.pl

diff --git a/spit_csv.pl b/spit_csv.pl
index b0da245..fbf29e9 100755 (executable)
--- a/spit_csv.pl
+++ b/spit_csv.pl
@@ -7,6 +7,7 @@ use Unicode::Normalize;
 
 my @desired_tags_subfields = ();
 foreach my $argnum ( 1 .. $#ARGV) {
+    print STDERR $ARGV[$argnum] . "\n";
     push @desired_tags_subfields, $ARGV[$argnum];
 }
 
@@ -19,7 +20,10 @@ foreach my $argnum ( 0 .. 0 ) {
 
        print STDERR "Processing " . $ARGV[$argnum] . "\n";
 
-       my $batch = MARC::Batch->new('XML',$ARGV[$argnum]);
+    my $M;
+    open $M, '<:utf8', $ARGV[$argnum];
+    my $batch = MARC::Batch->new('XML',$M);
+
        $batch->strict_off();
        $batch->warnings_off();
 
@@ -29,16 +33,31 @@ foreach my $argnum ( 0 .. 0 ) {
 
                print STDERR "WARNINGS: Record $count : " .  join(":",@warnings) . " : continuing...\n" if ( @warnings );
 
+        my $first = 1;
         for (my $i = 0; $i < scalar(@desired_tags_subfields); $i+=2) {
-                   my @tags; if ($record->field($desired_tags_subfields[$i])) { @tags = $record->field($desired_tags_subfields[$i]); }
+                   my @tags = (); 
+            if ($record->field($desired_tags_subfields[$i])) { 
+                @tags = $record->field($desired_tags_subfields[$i]); 
+            }
+            if (scalar(@tags)>1) { 
+                die "Multiple $desired_tags_subfields[$i]\n"; 
+            } elsif (scalar(@tags)==0) { 
+                print STDERR "Record $count missing $desired_tags_subfields[$i]\n"; 
+                goto END_OF_WHILE; 
+            }
             foreach my $f ( @tags ) { 
                 if ($f->subfield($desired_tags_subfields[$i+1])) { 
-                    print STDOUT $f->subfield($desired_tags_subfields[$i+1]) . "\t";
+                    if ($first) {
+                        $first = 0;
+                    } else {
+                        print STDOUT "\t";
+                    }
+                    print STDOUT $f->subfield($desired_tags_subfields[$i+1]);
                 } 
             }
         }
         print STDOUT "\n";
-
+        END_OF_WHILE:
        }
        print STDERR "Processed $count records\n";
 }