From 76def23cfaf543f2a55dcb2a57b40cecae71687a Mon Sep 17 00:00:00 2001
From: Don Armstrong <don@donarmstrong.com>
Date: Wed, 12 Dec 2007 00:28:41 +0000
Subject: [PATCH]  * Handle when NCBI has remote server errors "gracefully"  *
 Fix pubmed/ncbi typo

git-svn-id: file:///srv/svn/function2gene/trunk@19 a0738b58-4706-0410-8799-fb830574a030
---
 bin/get_ncbi_results | 19 ++++++++++++++-----
 1 file changed, 14 insertions(+), 5 deletions(-)

diff --git a/bin/get_ncbi_results b/bin/get_ncbi_results
index 3e1a625..313ecad 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   => $options{database},
@@ -145,7 +145,7 @@ while (<$terms>) {
      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   => $options{database},
@@ -155,6 +155,15 @@ while (<$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;
+	       $retry_count--;
+	  }
+	  if ($rety_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);
-- 
2.39.2