Changeset 26930 for gs2-extensions

Timestamp:

2013-02-26T09:49:36+13:00 (11 years ago)

Author:

jmt12

Message:

Randomized order of files, and added the ability to specify a maximum number of documents in the manifest

File:

• gs2-extensions/parallel-building/trunk/src/bin/script/manifestinator.pl (modified) (4 diffs)

gs2-extensions/parallel-building/trunk/src/bin/script/manifestinator.pl

@@ -2 +2 @@
 
 use strict;
+use warnings;
+
+use List::Util 'shuffle';
 
 if (!defined $ARGV[0] || !-d $ARGV[0])
 {
-  print "usage: manifestinator.pl <import directory>\n";
+  print "usage: manifestinator.pl <import directory> [<max number of documents>]\n";
   exit(0);
 }
 
-open(XMLOUT, '>:utf8', 'manifest.xml');
+my $max_docs = 0;
+if (defined $ARGV[1] && $ARGV[1] =~ /^\d+$/)
+{
+  $max_docs = $ARGV[1];
+}
+
+my $manifest_filename = 'manifest';
+if ($max_docs > 0)
+{
+  $manifest_filename .= '-' . $max_docs;
+}
+open(XMLOUT, '>:utf8', $manifest_filename . '.xml');
 print XMLOUT "<Manifest>\n";
 print XMLOUT "  <Index>\n";
 
-&manifestify($ARGV[0], '');
+&manifestify($ARGV[0], '', $max_docs, 0);
 
 print XMLOUT "  </Index>\n";
@@ -23 +37 @@
 sub manifestify
 {
-  my ($dir, $prefix) = @_;
+  my ($dir, $prefix, $max_docs, $current_count) = @_;
+  if ($max_docs > 0 && $current_count >= $max_docs)
+  {
+    return $current_count;
+  }
   if (!opendir(DH, $dir))
   {
@@ -30 +48 @@
   my @files = readdir(DH);
   closedir(DH);
-  foreach my $file (sort @files)
+  foreach my $file (shuffle @files)
   {
     if ($file =~ /^\./)
@@ -40 +58 @@
     {
       my $new_prefix = $prefix . $file . '/';
-      &manifestify($path, $new_prefix);
+      $current_count = &manifestify($path, $new_prefix, $max_docs, $current_count);
     }
     else
     {
       print XMLOUT "    <Filename>" . $prefix . $file . "</Filename>\n";
+      $current_count++;
+    }
+    if ($max_docs > 0 && $current_count >= $max_docs)
+    {
+      return $current_count;
     }
   }
+  return $current_count;
 }