fixes for multi subfields

[migration-tools.git] / Equinox-Migration / lib / Equinox / Migration / MapDrivenMARCXMLProc.pm

diff --git a/Equinox-Migration/lib/Equinox/Migration/MapDrivenMARCXMLProc.pm b/Equinox-Migration/lib/Equinox/Migration/MapDrivenMARCXMLProc.pm
index ee37e78..8db9e7e 100644 (file)
--- a/Equinox-Migration/lib/Equinox/Migration/MapDrivenMARCXMLProc.pm
+++ b/Equinox-Migration/lib/Equinox/Migration/MapDrivenMARCXMLProc.pm
@@ -23,6 +23,10 @@ our $VERSION = '1.002';
 my $dstore;
 my $sfmap;
 my @mods = qw( multi bib required );
+my $multis = {};
+my $reccount;
+my $verbose = 0;
+
 
 =head1 SYNOPSIS
 
@@ -48,7 +52,9 @@ and C<marcfile> (the MARC data to be processed).
 sub new {
     my ($class, %args) = @_;
 
-    my $self = bless { 
+    $verbose = 1 if $args{verbose};
+
+    my $self = bless { multis => \$multis,
                      }, $class;
 
     # initialize map and taglist
@@ -57,9 +63,13 @@ sub new {
                                                       mods => \@mods );
 
     # initialize datastore
+    die "Datastore file 'EMMXSSTORAGE.dbmd' already exists. Exiting.\n"
+      if (-e "EMMXSSTORAGE.dbmd");
     $dstore = DBM::Deep->new( file => "EMMXSSTORAGE.dbmd",
-                              data_sector_size => 256 );
-    $dstore->{rptr} = 0;            # next record ptr
+                              data_sector_size => 256,
+                              autoflush => 0,
+                            );
+    $reccount = 0;            # next record ptr
     $dstore->{tags} = $sfmap->tags; # list of all tags
     $self->{data} = $dstore;
 
@@ -93,11 +103,14 @@ sub parse_record {
 
     # cleanup memory and increment pointer
     $record->purge;
-    $dstore->{rptr}++;
+    $reccount++;
 
     # check for required fields
     check_required();
     push @{ $dstore->{recs} }, $crec;
+
+    print STDERR "$reccount\n"
+      if ($verbose and !($reccount % 1000));
 }
 
 sub process_field {
@@ -124,8 +137,9 @@ sub process_field {
           { process_subs($tag, $sub, $crec) }
 
         # check map to ensure all declared tags and subs have a value
-        my $mods = $sfmap->mods($field);
         for my $mappedsub ( @{ $sfmap->subfields($tag) } ) {
+            my $fieldname = $sfmap->field($tag, $mappedsub);
+            my $mods = $sfmap->mods($fieldname);
             next if $mods->{multi};
             $crec->{tags}[-1]{uni}{$mappedsub} = ''
               unless defined $crec->{tags}[-1]{uni}{$mappedsub};
@@ -144,26 +158,27 @@ sub process_subs {
     # handle unmapped tag/subs
     return unless ($sfmap->has($tag, $code));
 
-    # fetch our datafield struct and fieldname
+    # fetch our datafield struct and fiel
     my $dataf = $crec->{tags}[-1];
     my $field = $sfmap->field($tag, $code);
-    $crec->{names}{$tag}{$code} = $field;
 
     # test filters
     for my $filter ( @{$sfmap->filters($field)} ) {
         return if ($sub->text =~ /$filter/i);
     }
+
     # handle multi modifier
     if (my $mods = $sfmap->mods($field)) {
         if ($mods->{multi}) {
             push @{$dataf->{multi}{$code}}, $sub->text;
+            $multis->{$tag}{$code} = 1;
             return;
         }
     }
 
     # if this were a multi field, it would be handled already. make sure its a singleton
     die "Multiple occurances of a non-multi field: $tag$code at rec ",
-      ($dstore->{rptr} + 1),"\n" if (defined $dataf->{uni}{$code});
+      ($reccount + 1),"\n" if (defined $dataf->{uni}{$code});
 
     # everything seems okay
     $dataf->{uni}{$code} = $sub->text;
@@ -183,7 +198,7 @@ sub check_required {
                 $found = 1 if ($tag->{uni}{$code});
             }
 
-            die "Required mapping $tag_id$code not found in rec ",$dstore->{rptr},"\n"
+            die "Required mapping $tag_id$code not found in rec ",$reccount,"\n"
               unless ($found);
         }
     }
@@ -196,17 +211,28 @@ Returns current record number (starting from zero)
 
 =cut
 
-sub recno { my ($self) = @_; return $self->{data}{rptr} }
+sub recno { my ($self) = @_; return $self->{data}{rcnt} }
 
 =head2 name
 
-Returns mapped fieldname when passed a record number, tag, and code
+Returns mapped fieldname when passed a tag, and code
+
+    my $name = $m->name(999,'a');
 
-    my $name = $m->name(3,999,'a');
+=cut
+
+sub name { my ($self, $t, $c) = @_; return $sfmap->field($t, $c) }
+
+=head2 get_multis
+
+Returns hashref of C<{tag}{code}> for all mapped multi fields
 
 =cut
 
-sub name { my ($self, $r, $t, $c) = @_; return $dstore->{recs}[$r]{names}{$t}{$c} };
+sub get_multis {
+    my ($self) = @_;
+    return $multis;
+}
 
 =head1 MODIFIERS