one-offs/migrate_images.pl

   1
   2 =head1 NAME
   3
   4 migrate_images.pl - image migration script
   5
   6 =head1 DESCRIPTION
   7
   8 migrates the images from a id-based directory structure to a 2 byte stemmed directory structure based on md5sums.
   9
  10 Requires the ImageAddMD5sum.pm db patch (available in cxgn/sgn/db).
  11
  12 Options:
  13
  14 =over 5
  15
  16 =item -D
  17
  18 database name (default sandbox)
  19
  20 =item -H
  21
  22 host name (default localhost)
  23
  24 =item -o
  25
  26 old image location directory (default /data/prod/public/images/image_files)
  27
  28 =item -n
  29
  30 new image location directory (default /data/prod/public/images/new_image_files)
  31
  32 =back
  33
  34 =head1 AUTHOR
  35
  36 Lukas Mueller
  37
  38 =cut
  39
  40 use strict;
  41
  42 use Getopt::Std;
  43
  44 use CXGN::DB::InsertDBH;
  45 use CXGN::Image;
  46
  47 our($opt_o, $opt_n, $opt_H, $opt_D);
  48
  49 getopts('o:n:H:D:');
  50
  51 my $dbh = CXGN::DB::InsertDBH->new( {
  52     dbname => $opt_D || 'sandbox',
  53     dbhost => $opt_H || 'localhost',
  54                                     });
  55
  56 my $old_image_dir = $opt_o || '/data/prod/public/images/image_files';
  57 my $new_image_dir = $opt_n || '/data/prod/public/images/new_image_files';
  58
  59 my $sql = "SELECT image_id FROM metadata.md_image where obsolete !='t' order by image_id";
  60 my $sth = $dbh->prepare($sql);
  61 $sth->execute();
  62
  63 my %md5sums;
  64 my $processed=0;
  65 my @not_found = ();
  66
  67 while (my ($image_id) = $sth->fetchrow_array()) {
  68
  69     my $old_image = CXGN::Image->new(dbh=>$dbh, image_id=>$image_id, image_dir=> $old_image_dir);
  70
  71     my $i = CXGN::Image->new(dbh=>$dbh, image_id=>$image_id, image_dir=>$new_image_dir);
  72
  73
  74     if (! -d $old_image_dir."/".$image_id) {
  75         push @not_found, "$image_id ".$i->get_name()."\n";
  76         print STDERR "\nNot found image: $image_id\n";
  77         next();
  78     }
  79
  80     my $filepath =  $old_image_dir."/".$image_id."/".$old_image->get_original_filename().$old_image->get_file_ext();
  81     if (! -e $filepath) {
  82         print STDERR "WARNING! $filepath does not exist!\n";
  83         next();
  84     }
  85     my $md5sum = $i -> calculate_md5sum($filepath);
  86
  87     $md5sums{$md5sum}++;
  88
  89
  90     $i->set_md5sum($md5sum);
  91     $i->store();
  92     $dbh->commit();
  93     $i->make_dirs();
  94
  95
  96     $i->copy_location($old_image_dir."/".$image_id);
  97
  98     $processed++;
  99     print STDERR "Processing image $image_id        \r";
 100 }
 101
 102 foreach my $nf (@not_found) {
 103     print STDERR $nf;
 104 }
 105
 106
 107 print STDERR "\nProcessed $processed images, generating ".scalar(keys(%md5sums))." unique md5 keys.\n";
 108
 109