bin/loading_scripts/solcap/load_solcap_images.pl

   1 #!/usr/bin/perl
   2
   3 =head1 NAME
   4
   5 load_solcap_images.pl
   6
   7 =head1 SYNOPSYS
   8
   9 load_solcap_images.pl -D [ sandbox | cxgn | trial ] -H hostname -i dirname
  10
  11 =head1 DESCRIPTION
  12
  13 Loads  images  into the SGN database, using the SGN::Image framework.
  14 Then link the loaded image with the relevant stock  (see phenome.stock_image table)
  15
  16 Requires the following parameters:
  17
  18 =over 8
  19
  20 =item -D
  21
  22 a database parameter, which can either be "cxgn", "sandbox", or "trial". "cxgn" and "sandbox" will cause the script to connect to the respective databases; "trial" will connect to sandbox, but not perform any of the database modifications.
  23
  24 =item -H
  25
  26 host name
  27
  28 =item -i
  29
  30 a dirname that contains image filenames or subdirectories named after database accessions, containing one or more images (see option -d) .
  31
  32 =item -u
  33
  34 user_name - from sgn_people.sp_person.
  35
  36 =item -l
  37
  38 location, as stored in stockprop (when plots are loaded with load_solcap_plots.pl)
  39
  40 =item -d
  41
  42 files are stored in sub directories named after database accessions
  43
  44 =item -e
  45
  46 image file extension . Defaults to 'jpg'
  47
  48
  49 =item -t
  50
  51 trial mode . Nothing will be stored.
  52
  53
  54 =back
  55
  56 The script will generate an error file, named like the filename supplied, with the extension .err.
  57
  58 =head1 AUTHOR(S)
  59
  60 Naama Menda (nm249@cornell.edu) September 2011.
  61
  62 =cut
  63
  64 use strict;
  65
  66 use CXGN::Metadata::Schema;
  67 use CXGN::Metadata::Metadbdata;
  68 use CXGN::DB::InsertDBH;
  69 use SGN::Image;
  70 use Bio::Chado::Schema;
  71 use CXGN::People::Person;
  72 use Carp qw /croak/;
  73
  74 use File::Basename;
  75 use SGN::Context;
  76 use Getopt::Std;
  77
  78
  79 our ($opt_H, $opt_D, $opt_t, $opt_i, $opt_u, $opt_d, $opt_e, $opt_f, $opt_l, $opt_y);
  80 getopts('H:D:u:i:e:f:tdl:y:');
  81
  82 my $dbhost    = $opt_H;
  83 my $dbname    = $opt_D;
  84 my $dirname   = $opt_i;
  85 my $sp_person = $opt_u;
  86 my $location  = $opt_l;
  87 my $year      = $opt_y;
  88
  89 my $ext = $opt_e || 'jpg';
  90
  91 if (!$dbhost && !$dbname) {
  92     print "dbhost = $dbhost , dbname = $dbname\n";
  93     print "opt_t = $opt_t, opt_u = $opt_u, opt_d= $opt_d opt_i = $dirname\n";
  94     usage();
  95 }
  96
  97 if (!$dirname) { print "dirname = $dirname\n" ; usage(); }
  98
  99 my $dbh = CXGN::DB::InsertDBH->new( { dbhost=>$dbhost,
 100                                       dbname=>$dbname,
 101                                     } );
 102
 103 my $schema= Bio::Chado::Schema->connect(  sub { $dbh->get_actual_dbh() } ,  { on_connect_do => ['SET search_path TO  public;'] }
 104     );
 105 my $metadata_schema = CXGN::Metadata::Schema->connect(
 106     sub { $dbh->get_actual_dbh },
 107     { on_connect_do => ['SET search_path TO metadata;'] },
 108     );
 109 my $sp_person_id= CXGN::People::Person->get_person_by_username($dbh, $sp_person);
 110 my %name2id = ();
 111
 112
 113 my $ch = SGN::Context->new();
 114 my $image_dir =  $opt_f || $ch->get_conf("image_dir");
 115
 116 print "PLEASE VERIFY:\n";
 117 print "Using dbhost: $dbhost. DB name: $dbname. \n";
 118 print "Path to image is: $image_dir\n";
 119 print "CONTINUE? ";
 120 my $a = (<STDIN>);
 121 if ($a !~ /[yY]/) { exit(); }
 122
 123 if (($dbname eq "sandbox") && ($image_dir !~ /sandbox/)) {
 124     die "The image directory needs to be set to image_files_sandbox if running on rubisco/sandbox. Please change the image_dir parameter in SGN.conf\n\n";
 125                                           }
 126 if (($dbname eq "cxgn") && ($image_dir =~ /sandbox/)) {
 127     warn "The image directory needs to be set to image_files when the script is running on the production database. Please change the image_dir parameter in SGN.conf\n\n";
 128 }
 129
 130 my %image_hash = ();  # used to retrieve images that are already loaded
 131 my %connections = (); # keep track of object -- image connections that have already been made.
 132
 133 my $object_rs = $schema->resultset("Stock::Stock")->search( { } ) ;
 134 while (my $object = $object_rs->next ) {
 135     my $id = $object->stock_id;
 136     my $name = $object->name;
 137     $name2id{lc($name)} = $id;
 138 }
 139
 140 # cache image chado object - image links to prevent reloading of the
 141 # same data
 142 #
 143 print "Caching image-stock links...\n";
 144
 145 my $q = "SELECT * FROM phenome.stock_image";
 146 my $sth = $dbh->prepare($q);
 147 $sth->execute();
 148 while ( my $hashref = $sth->fetchrow_hashref() ) {
 149     my $image_id = $hashref->{image_id};
 150     my $chado_table_id = $hashref->{stock_id};  ##### table specific
 151     my $i = SGN::Image->new($dbh, $image_id);
 152     my $original_filename = $i->get_original_filename();
 153     $image_hash{$original_filename} = $i; # this doesn't have the file extension
 154     $connections{$image_id."-".$chado_table_id}++;
 155 }
 156
 157 open (ERR, ">load_solcap_images.err") || die "Can't open error file\n";
 158
 159 my @files = glob "$dirname/*.$ext";
 160 @files = glob "$dirname/*" if $opt_d ;
 161 my @sub_files;
 162
 163 my $new_image_count = 0;
 164
 165
 166 foreach my $file (@files) {
 167     eval {
 168         chomp($file);
 169         @sub_files = ($file);
 170         @sub_files =  glob "$file/*.$ext" if $opt_d;
 171
 172         my $object_name = basename($file, ".$ext" );
 173         print  "object_name = '".$object_name."' \n";
 174         #$individual_name =~s/(W\d{3,4}).*\.JPG/$1/i if $individual_name =~m/^W\d{3}/;
 175         #2009_oh_8902_fruit-t
 176         # solcap images:
 177         #my ($year, $place, $plot, undef) = split /_/ , $object_name;
 178         #2009_ca_2138_fruit_c.jpg
 179         my ($year, $place, $name, undef,undef) = split /_/ , $object_name;
 180
 181         #my $stock = $schema->resultset("Stock::Stock")->find( {
 182         #    stock_id => $name2id{ lc($name) }  } );
 183         my ($stock) = $schema->resultset("Stock::Stock")->search(
 184             {
 185                 name => $name,
 186                 uniquename => { 'ilike' => $name . '%' . $year . '%' . $location . '%'}
 187             } );
 188
 189         if (!$stock) {
 190             warn "no stock found for plot # $name ! Skipping !!\n\n";
 191             next();
 192         }
 193
 194         foreach my $filename (@sub_files) {
 195             chomp $filename;
 196             my $file_basename =  basename($filename, ".$ext");
 197             print STDOUT "Processing file $file...\n";
 198             message( "Loading $name, image $filename\n");
 199             my $image_id; # this will be set later, depending if the image is new or not
 200             if (! -e $filename) {
 201                 warn "The specified file $filename does not exist! Skipping...\n";
 202                 next();
 203             }
 204
 205             if (!exists($name2id{lc($name)})) {
 206                 message ("$name does not exist in the database...\n");
 207             }
 208
 209             else {
 210                 message( "Adding $filename...\n");
 211                 if (exists($image_hash{$file_basename})) {
 212                     message("$filename is already loaded into the database...\n");
 213                     $image_id = $image_hash{$file_basename}->get_image_id();
 214                     print "name = $name , stock id = " . $name2id{lc($name)} . "\n\n";
 215                     $connections{$image_id."-".$name2id{lc($name)}}++;
 216                     if ($connections{$image_id."-".$name2id{lc($name)}} > 1) {
 217                         message("The connection between $name and image $filename has already been made. Skipping...\n");
 218                     }
 219                     elsif ($image_hash{$file_basename}) {
 220                         message("Associating stock " . $name2id{lc($name)} . "with already loaded image $filename...\n") ;
 221                         ################################
 222                     }
 223                 }
 224                 else {
 225                     message(" Generating new image object for image $filename and associating it with stock $name, id " . $name2id{lc($name) } ." ...\n");
 226                     my $caption = $name;
 227
 228                     if ($opt_t)  {
 229                         print STDOUT qq { Would associate file $filename to stock $name, id $name2id{lc($name)}\n };
 230                         $new_image_count++;
 231                     }
 232                     else {
 233                         my $image = SGN::Image->new($dbh);
 234                         $image_hash{$file_basename}=$image;
 235
 236                         $image->process_image("$filename", undef, undef);
 237                         $image->set_description("$caption");
 238                         $image->set_name(basename($filename , ".$ext"));
 239                         $image->set_sp_person_id($sp_person_id);
 240                         $image->set_obsolete("f");
 241                         $image_id = $image->store();
 242                         $new_image_count++;
 243                     }
 244                     my $metadata = CXGN::Metadata::Metadbdata->new($metadata_schema, $sp_person);
 245                     my $metadata_id = $metadata->store()->get_metadata_id();
 246                     die "NO METADATA ID FOUND\n" if !$metadata_id;
 247 #store the image_id - stock_id link
 248                     my $q = "INSERT INTO phenome.stock_image (stock_id, image_id, metadata_id) VALUES (?,?,?)";
 249                     my $sth  = $dbh->prepare($q);
 250                     $sth->execute($stock->stock_id, $image_id, $metadata_id);
 251                 }
 252             }
 253         }
 254     };
 255     if ($@) {
 256         print STDOUT "ERROR OCCURRED WHILE SAVING NEW INFORMATION. $@\n";
 257         $dbh->rollback();
 258     }
 259     else {
 260         $dbh->commit();
 261     }
 262 }
 263
 264
 265 close(ERR);
 266 close(F);
 267
 268
 269 print STDOUT "Inserted  $new_image_count images.\n";
 270 print STDOUT "Done. \n";
 271
 272 sub usage {
 273     print "Usage: load_images.pl -D dbname [ cxgn | sandbox ]  -H dbhost -t [trial mode ] -i input dir -l location of the plots (as loaded into stockprop). \n Images will be stored in phenome.stock_image \n";
 274     exit();
 275 }
 276
 277 sub message {
 278     my $message=shift;
 279     print STDOUT $message;
 280     print ERR $message;
 281 }