added extractor for EoD catalog checker

lib/Alma/MARC_Extractor.pm

@@ -11,6 +11,7 @@ sub new
   my $self=
   {
     fields => $fields,
+    mex_ot2ut => 1,
   };
 
   bless ($self, $class);
@@ -35,7 +36,9 @@ sub extract_identifiers
   $rec->{marc_record}= 'marc_data_found';
 
   # print __LINE__, " mrd: ", Dumper($mrd);
-  my ($df_urn, $val_urn, $df_doi, $val_doi, $df_othes, $val_othes, $df_utheses, $val_utheses);
+  my ($df_urn, $val_urn, $df_doi, $val_doi, $df_hdl, $val_hdl,
+      $df_othes, $val_othes, $df_utheses, $val_utheses,
+      $df_phaidra, $val_phaidra);
 
   foreach my $cf (@{$mrd->{controlfield}})
   {
@@ -49,7 +52,7 @@ sub extract_identifiers
   {
     foreach my $sf (@{$df->{subfield}})
     {
-      if ($sf->{content} =~ m#urn:nbn#)
+      if ($sf->{content} =~ m#urn:nbn:at:ubw#)
       {
         $df_urn= join(':', (map { $df->{$_} } qw(tag ind1 ind2)), $sf->{code});
         $val_urn= $sf->{content};
@@ -59,6 +62,11 @@ sub extract_identifiers
         $df_doi= join(':', (map { $df->{$_} } qw(tag ind1 ind2)), $sf->{code});
         $val_doi= $sf->{content};
       }
+      elsif ($sf->{content} =~ m#11353/#)
+      {
+        $df_hdl= join(':', (map { $df->{$_} } qw(tag ind1 ind2)), $sf->{code});
+        $val_hdl= $sf->{content};
+      }
       elsif ($sf->{content} =~ m#othes\.univie#)
       {
         $df_othes= join(':', (map { $df->{$_} } qw(tag ind1 ind2)), $sf->{code});
@@ -69,16 +77,85 @@ sub extract_identifiers
         $df_utheses= join(':', (map { $df->{$_} } qw(tag ind1 ind2)), $sf->{code});
         $val_utheses= $sf->{content};
       }
+      elsif ($sf->{content} =~ m#phaidra\.univie#)
+      {
+        $df_phaidra= join(':', (map { $df->{$_} } qw(tag ind1 ind2)), $sf->{code});
+        $val_phaidra= $sf->{content};
+      }
     }
   }
   $rec->{ df_urn}=      $df_urn;
   $rec->{val_urn}=     $val_urn;
   $rec->{ df_doi}=      $df_doi;
   $rec->{val_doi}=     $val_doi;
-  $rec->{ df_othes}=    $df_othes;
-  $rec->{val_othes}=   $val_othes;
-  $rec->{ df_utheses}=  $df_utheses;
-  $rec->{val_utheses}= $val_utheses;
+  $rec->{ df_hdl}=      $df_hdl;
+  $rec->{val_hdl}=     $val_hdl;
+
+  if ($self->{mex_ot2ut})
+  {
+    $rec->{ df_othes}=    $df_othes;
+    $rec->{val_othes}=   $val_othes;
+    $rec->{ df_utheses}=  $df_utheses;
+    $rec->{val_utheses}= $val_utheses;
+  }
+
+  if ($self->{mex_phaidra})
+  {
+    $rec->{ df_phaidra_url}=  $df_phaidra;
+    $rec->{val_phaidra_url}= $val_phaidra;
+  }
+
+  $rec;
+}
+
+sub extract_identifiers2
+{
+  my $self= shift;
+
+  my $marc= shift; # MARC record to select data from
+  my $rec= shift;  # record to fill extracted data into
+
+  foreach my $f (@{$self->{fields}}) { $rec->{$f}= $marc->{$f}; }
+
+  my $mrd= $marc->{xmlref}->{records}->[0]->{record}->[0]->{recordData}->[0]->{record}->[0];
+  unless (defined ($mrd))
+  {
+    $rec->{marc_record}= 'marc_data_not_found';
+    next MARC;
+  }
+  $rec->{marc_record}= 'marc_data_found';
+
+  foreach my $cf (@{$mrd->{controlfield}})
+  {
+    if ($cf->{tag} eq '005')
+    {
+      $rec->{ts_marc}= $cf->{content};
+    }
+  }
+
+  foreach my $df (@{$mrd->{datafield}})
+  {
+    foreach my $sf (@{$df->{subfield}})
+    {
+      my $match;
+
+         if ($sf->{content} =~ m#urn:nbn:at:ubw#)  { $match= 'urn'; }
+      elsif ($sf->{content} =~ m#10\.25365#)       { $match= 'doi'; }
+      elsif ($sf->{content} =~ m#11353/#)          { $match= 'hdl'; }
+      elsif ($sf->{content} =~ m#othes\.univie#)   { $match= 'othes' }
+      elsif ($sf->{content} =~ m#utheses\.univie#) { $match= 'utheses' }
+      elsif ($sf->{content} =~ m#phaidra\.univie#) { $match= 'phaidra' }
+
+      if (defined ($match))
+      {
+        push (@{$rec->{mex}->{$match}},
+        [
+          join(':', (map { $df->{$_} } qw(tag ind1 ind2)), $sf->{code}),
+          $sf->{content}
+        ]);
+      }
+    }
+  }
 
   $rec;
 }