]> git.donarmstrong.com Git - function2gene.git/blobdiff - bin/get_harvester_results
* Stop requiring wget
[function2gene.git] / bin / get_harvester_results
index 7446e088c245d9d58adcafef59dd04f769edb6a8..4d3f5932481349f61bd213508271ac2bf00728ce 100755 (executable)
@@ -72,6 +72,7 @@ use IO::File;
 use URI;
 use WWW::Mechanize;
 use Thread::Queue;
+use Time::HiRes qw(usleep);
 
 # XXX parse config file
 
@@ -172,7 +173,24 @@ sub get_url{
 sub wget_urls{
      my ($dir,@urls) = @_;
      return unless @urls;
-     system(q(wget),'-nd','-nH','-w','2','--random-wait','-P',$dir,@urls) == 0 or warn "$!";
+     # replacing wget with WWW::Mechanize
+     my $mech = WWW::Mechanize->new(agent => "DA_get_harvester_results/$REVISION");
+     for my $url (@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 {
+              $mech->save_content($dir.'/'.$cleaned_url);
+              print "retreived $url\n";
+         };
+         if ($@) {
+              warn $@;
+         }
+     }
+     #system(q(wget),'-nd','-nH','-w','2','--random-wait','-P',$dir,@urls) == 0 or warn "$!";
 }
 
 __END__