X-Git-Url: https://git.donarmstrong.com/?p=function2gene.git;a=blobdiff_plain;f=bin%2Fget_genecard_results;h=ac38e29d6767e8cb13040eb2b591258f07edf5b2;hp=017f1e1b2b76957fbb7c67e90fd7f0d075811ede;hb=af4fd770f221db1cec02393df378e079c0b9a8fc;hpb=6d24067f20698257dc1103d5c21e8a7f6a32b97b

diff --git a/bin/get_genecard_results b/bin/get_genecard_results
index 017f1e1..ac38e29 100755
--- a/bin/get_genecard_results
+++ b/bin/get_genecard_results
@@ -70,6 +70,7 @@ BEGIN{
 use IO::File;
 use URI;
 use WWW::Mechanize;
+use Time::HiRes qw(usleep);
 
 # XXX parse config file
 
@@ -127,7 +128,22 @@ while (<$terms>) {
      # Get XML file
      my @current_urls;
      while (@current_urls = map{$options{genecard_site}.$_} splice(@result_urls,0,30)) {
-	  system(q(wget),'-nd','-nH','-w','2','--random-wait','-P',qq($options{dir}/$dir_name),@current_urls) == 0 or warn "$!";
+	  for my $url (@current_urls) {
+	       # sleep for around 2 seconds
+	       usleep((0.5+rand)*2*1000000);
+	       $mech->get($url);
+	       my $cleaned_url = $url;
+	       $cleaned_url =~ s{http://}{}g;
+	       $cleaned_url =~ s/[^\w]//g;
+	       eval {
+		    $mech->save_content($options{dir}.'/'.$dir_name.'/'.$cleaned_url);
+		    print "retreived $url\n";
+	       };
+	       if ($@) {
+		    warn $@;
+	       }
+	  }
+	  #system(q(wget),'-nd','-nH','-w','2','--random-wait','-P',qq($options{dir}/$dir_name),@current_urls) == 0 or warn "$!";
      }
 }