update how journal information and issn are handled

[reference.git] / lib / Reference / Retrieve / PubMed.pm

diff --git a/lib/Reference/Retrieve/PubMed.pm b/lib/Reference/Retrieve/PubMed.pm
index 0ecc842ee1d068839004117d5a0336e09811a77b..78cf93533641c6b190f04c039e6f1ca77b303ba9 100644 (file)
--- a/lib/Reference/Retrieve/PubMed.pm
+++ b/lib/Reference/Retrieve/PubMed.pm
@@ -201,8 +201,8 @@ sub _create_reference_from_xml($$){
                             title      => [_fix_medline_title($ref->{MedlineCitation}->{Article}->{ArticleTitle})],
                             abstract   => [_fix_medline_abstract($ref->{MedlineCitation}->{Article}->{Abstract}->{AbstractText})],
                             journal    => [_fix_medline_journal($ref->{MedlineCitation}->{Article}->{Journal},
-                                                                $ref->{MedlineCitation}->{Article}->{MedlineJournalInfo},
-                                                                $ua,
+                                                     $ref->{MedlineCitation}->{MedlineJournalInfo},
+                                                     $ua,
                                                                 #@_, # configuration
                                                                )],
                             _fix_ids($ref),
@@ -332,12 +332,17 @@ sub _fix_medline_journal($$$;){
      #         </TranslationStack>
      # </eSearchResult>
 
-     my $ISSN = $journal->{ISSN};
+     my $ISSN = length($journal->{ISSN})? $journal->{ISSN} : $medline_journal->{ISSNLinking};
      if (ref $ISSN) {
          $ISSN = $ISSN->{content};
+  }
+     if (not length($ISSN)) {
+         use Data::Printer;
+         p $journal;
+         die "No ISSN";
      }
      my $url = qq(http://www.ncbi.nlm.nih.gov/nlmcatalog/?term=${ISSN}[ISSN]&format=text&report=xml);
-     print STDERR "url: $url" if $DEBUG;
+     print STDERR "url: $url\n" if $DEBUG;
      my $request = HTTP::Request->new('GET', $url);
      my $response = $ua->request($request);
      $response = $response->content;
@@ -345,10 +350,17 @@ sub _fix_medline_journal($$$;){
      $response =~ s/\&lt;/</gso;
      $response =~ s/^<\/?pre>//;
 
-     print STDERR "response: $response" if $DEBUG;
+     print STDERR "response: $response\n" if $DEBUG;
      my $xa = new XML::Simple;
      my $ref_struct = $xa->XMLin($response,ForceArray=>['NCBICatalogRecord'],KeyAttr=>[]);
-     my $ref = $ref_struct->{NCBICatalogRecord}[0];
+     my $ref;
+     for (0..@{$ref_struct->{NCBICatalogRecord}}) {
+         if (exists $ref_struct->{NCBICatalogRecord}[$_]{JrXml}) {
+             $ref = $ref_struct->{NCBICatalogRecord}[$_];
+             last;
+         }
+     }
+     $ref = $ref_struct->{NCBICatalogRecord}[0] if not defined $ref;
      print STDERR Dumper($ref) if $DEBUG;
 
      my %journal;
@@ -362,6 +374,11 @@ sub _fix_medline_journal($$$;){
               $journal{lc($2)} = $3;
          }
   }
+     print STDERR Dumper($ref) if $DEBUG;
+     if (ref $ref->{JrXml}{Serial}{ISSN} ne 'ARRAY') {
+         $ref->{JrXml}{Serial}{ISSN} = [$ref->{JrXml}{Serial}{ISSN}];
+     }
+     print STDERR Dumper($ref->{JrXml}) if $DEBUG;
      my $print_issn = $ref->{JrXml}{Serial}{ISSN}[0]{IssnType} eq 'Print' ? 0 : 1;
      my @journal_entry =
          (title       => $ref->{JrXml}{Serial}{Title},
@@ -420,7 +437,7 @@ sub _fix_medline_pages($){
      if (not defined $start) {
          ($start) = $pagination =~ /(\d+)/
      }
-     if ($start > $stop and defined $stop) {
+     if ($start > $stop and defined $stop and length($stop)) {
          # this must be a reduced page listing; fix it up
          $stop+=$start - $start % 10 ** (int(log($stop)/log(10))+1);
      }