more filtering and field transcribing

lib/Univie/EoD/CrossReference.pm

@@ -4,6 +4,7 @@ use strict;
 
 use Data::Dumper;
 
+use JSON;
 use IRMA::db;
 use Util::ts;
 
@@ -100,15 +101,24 @@ sub find_books
 {
   my $self= shift;
   my $search= shift;
-  print __LINE__, " search: ", join(', ', %$search), "\n";
+  # print __LINE__, " search: ", join(', ', %$search), "\n"; ... can contain {$in:[...]} clauses!
+  # print __LINE__, " search: ", Dumper ($search);
+  print __LINE__, " search: ", JSON::encode_json ($search), "\n";
 
   my $foxml_col= $self->get_db_col('inventory_database', 'foxml.data');
   my $cur= $foxml_col->find( $search );
   # print __LINE__, ' cur: ', Dumper($cur);
 
-  while (my $foxml_rec= $cur->next())
+  FOXML_REC: while (my $foxml_rec= $cur->next())
   {
-    # print __LINE__, " foxml_rec=[$foxml_rec]\n";
+    # print __LINE__, " foxml_rec: foxml_rec=[$foxml_rec] found:", Dumper($foxml_rec);
+    if (exists ($self->{agent_cnf}->{prefiltered_pids}->{$foxml_rec->{pid}}))
+    {
+      print __LINE__, " NOTE: prefiltered pid; foxml_rec: ", Dumper($foxml_rec);
+      next FOXML_REC;
+    }
+
+    # print __LINE__, " foxml_rec=[$foxml_rec] added: ", Dumper($foxml_rec);
     $self->add_book($foxml_rec);
   }
 }
@@ -176,10 +186,10 @@ sub check_book
   return 'filtered_pids'       if (exists ($self->{agent_cnf}->{filtered_pids}->{$pid}));
 
   # check tickets
-  my $ticket_col= $self->get_db_col('irma_database', 'eod.tickets');
+  my $ticket_col= $self->get_db_col('stage_database', 'eod.tickets');
   my @tickets= $ticket_col->find({ ac_number => $ac_number })->all();
   print __LINE__, " ac_number=[$ac_number] tickets: ", scalar @tickets, "\n";
-  # print __LINE__, Dumper(\@tickets);
+  # print __LINE__, ' ', Dumper(\@tickets);
   if (@tickets  > 1)
   {
     $book->{ticket}= join(',', map { $_->{ticket} } @tickets);
@@ -190,11 +200,13 @@ sub check_book
     my $t0= @tickets[0];
     $book->{ticket}= $t0->{ticket};
     $book->{ticket_status}= $t0->{status};
-    $book->{vt}= $t0->{custom_fields}->{vt}->[1];
+    $book->{ticket_pid}= $t0->{custom_fields}->{ph_pid}->[1];
+    $book->{ticket_url}= $t0->{custom_fields}->{ph_url}->[1];
+    $book->{vt}=         $t0->{custom_fields}->{vt}->[1];
   }
 
   # check IRMA for registered identifiers
-  my $irma_col= $self->get_db_col('irma_database', 'irma.map');
+  my $irma_col= $self->get_db_col('stage_database', 'irma.map');
   my @irma_records= $irma_col->find({ ac_number => $ac_number })->all();
   # my @irma_records= $irma_col->find({ pid => $pid })->all(); # NOTE: there is no field named "pid" in the irma record!
   print __LINE__, " ac_number=[$ac_number] irma_records: ", scalar @irma_records, "\n";
@@ -214,7 +226,7 @@ sub check_book
     $self->{counters}->{missing_irma_record}++;
   }
 
-  my $marc_col= $self->get_db_col('marc_database', 'alma.marc');
+  my $marc_col= $self->get_db_col('stage_database', 'alma.marc');
   my $marc= $marc_col->find_one({ ac_number => $ac_number });
   # print __LINE__, " marc: ", Dumper($marc);
   return 'missing_marc_record' unless (defined ($marc));
@@ -229,7 +241,7 @@ sub check_book
   if (exists($book->{mex}))
   {
     my $mex= $book->{mex};
-    print __LINE__, " ac_number=[$ac_number] mex: ", Dumper($mex);
+    # print __LINE__, " ac_number=[$ac_number] mex: ", Dumper($mex);
     foreach my $mf (keys %$mex)
     {
       my $mfa= $mex->{$mf};
@@ -264,7 +276,7 @@ sub check_book
           if ($val =~ m#^11353/10\.(\d+)$#)
           {
             $copy= 1; # only transcribe those URLs that really look like some
-            if ($val ne $book->{hdl})
+            if ($book->{hdl} && $val ne $book->{hdl})
             { # TODO:
               $book->{update_hdl}= $book->{hdl};
               $self->{counters}->{update_hdl}++;
@@ -337,6 +349,13 @@ sub check_book
     push(@alma_notes, 'set_urn');
   }
 
+  if (!exists($book->{val_hdl}) && exists ($book->{hdl}))
+  {
+    $book->{update_hdl}= $book->{hdl};
+    $self->{counters}->{set_hdl}++;
+    push(@alma_notes, 'set_hdl');
+  }
+
   unless (exists($book->{df_phaidra}))
   { # TODO: if there is no phaidra_url at all, set it...
     $self->{counters}->{set_phaidra_url}++;