X-Git-Url: https://git.donarmstrong.com/?p=function2gene.git;a=blobdiff_plain;f=bin%2Fget_ncbi_results;h=e126d5f687fc984107aafcb1263986158ddd2086;hp=7a869014bc15525ab80f36c0fc02ad50ac94f4eb;hb=50e9739109c91b53ab620b462255f71a0b870f95;hpb=8e1fe04cb712a2231b9a7112b7a701a67f9ddd31

diff --git a/bin/get_ncbi_results b/bin/get_ncbi_results
index 7a86901..e126d5f 100755
--- a/bin/get_ncbi_results
+++ b/bin/get_ncbi_results
@@ -87,9 +87,9 @@ my %options = (debug    => 0,
 	       name     => 'ncbi_${search}_results.$format',
 	       terms    => '-',
 	       orgn     => 'homo',
-	       pubmed_site => 'http://www.ncbi.nlm.nih.gov',
-	       pubmed_search_url  => '/entrez/query.fcgi?cmd=search&doptcmdl=Brief&dispmax=1000',
-	       pubmed_get_url     => '/entrez/query.fcgi?cmd=Text',
+	       ncbi_site => 'http://www.ncbi.nlm.nih.gov',
+	       ncbi_search_url  => '/entrez/query.fcgi?cmd=search&doptcmdl=Brief&dispmax=1000',
+	       ncbi_get_url     => '/entrez/query.fcgi?cmd=Text',
 	      );
 
 GetOptions(\%options,'format|f=s','database|b=s','name|n=s',
@@ -121,7 +121,7 @@ while (<$terms>) {
      chomp;
      my $search = $_;
      my $format = $options{format};
-     my $uri = URI->new($options{pubmed_site}.$options{pubmed_search_url});
+     my $uri = URI->new($options{ncbi_site}.$options{ncbi_search_url});
      $uri->query_form($uri->query_form(),
 		      term => $search.' AND '.$options{orgn}.'[Orgn]',
 		      db   => $options{database},
@@ -143,8 +143,9 @@ while (<$terms>) {
 
      # Get XML file
      my @current_ids;
+     print {$xml_file} "<opt>\n";
      while (@current_ids = splice(@gene_ids,0,5)) {
-	  $uri = URI->new($options{pubmed_site}.$options{pubmed_get_url});
+	  $uri = URI->new($options{ncbi_site}.$options{ncbi_get_url});
 	  $uri->query_form($uri->query_form(),
 			   dopt => uc($options{format}),
 			   db   => $options{database},
@@ -154,6 +155,15 @@ while (<$terms>) {
 	  print STDERR "url: $url\n";
 	  $mech->get($url);
 	  my $response = $mech->content;
+	  my $retry_count=5;
+	  while ($response =~ /Error reading from remote server/ and $retry_count > 0) {
+	       $mech->get($url);
+	       $response = $mech->content;
+	       $retry_count--;
+	  }
+	  if ($retry_count <= 0) {
+	       die 'Unable to retreive ids ['.join(',',@current_ids).'] because of a remote server error';
+	  }
 	  # For some dumb reason, they send us xml with html
 	  # entities. Ditch them.
 	  #$response = decode_entities($response);
@@ -167,10 +177,10 @@ while (<$terms>) {
 	  $response =~ s/^\s*<pre>//gso;
 	  $response =~ s#</pre>\s*$##gso;
 
-
 	  print {$xml_file} $response;
 	  sleep 10;
      }
+     print {$xml_file} "</opt>\n";
      undef $xml_file;
 }