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 use 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);
  80 getopts('H:D:u:i:e:f:tdl:');
  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
  88 my $ext = $opt_e || 'jpg';
  89
  90 if (!$dbhost && !$dbname) {
  91     print "dbhost = $dbhost , dbname = $dbname\n";
  92     print "opt_t = $opt_t, opt_u = $opt_u, opt_d= $opt_d opt_i = $dirname\n";
  93     usage();
  94 }
  95
  96 if (!$dirname) { print "dirname = $dirname\n" ; usage(); }
  97
  98 my $dbh = CXGN::DB::InsertDBH->new( { dbhost=>$dbhost,
  99                                       dbname=>$dbname,
 100                                     } );
 101
 102 my $schema= Bio::Chado::Schema->connect(  sub { $dbh->get_actual_dbh() } ,  { on_connect_do => ['SET search_path TO  public;'] }
 103     );
 104 my $metadata_schema = CXGN::Metadata::Schema->connect(
 105     sub { $dbh->get_actual_dbh },
 106     { on_connect_do => ['SET search_path TO metadata;'] },
 107     );
 108 my $sp_person_id= CXGN::People::Person->get_person_by_username($dbh, $sp_person);
 109 my %name2id = ();
 110
 111
 112 my $ch = SGN::Context->new();
 113 my $image_dir =  $opt_f || $ch->get_conf("image_dir");
 114
 115 print "PLEASE VERIFY:\n";
 116 print "Using dbhost: $dbhost. DB name: $dbname. \n";
 117 print "Path to image is: $image_dir\n";
 118 print "CONTINUE? ";
 119 my $a = (<STDIN>);
 120 if ($a !~ /[yY]/) { exit(); }
 121
 122 if (($dbname eq "sandbox") && ($image_dir !~ /sandbox/)) {
 123     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";
 124                                           }
 125 if (($dbname eq "cxgn") && ($image_dir =~ /sandbox/)) {
 126     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";
 127 }
 128
 129 my %image_hash = ();  # used to retrieve images that are already loaded
 130 my %connections = (); # keep track of object -- image connections that have already been made.
 131
 132 my $object_rs = $schema->resultset("Stock::Stock")->search( { } ) ;
 133 while (my $object = $object_rs->next ) {
 134     my $id = $object->stock_id;
 135     my $name = $object->name;
 136     $name2id{lc($name)} = $id;
 137 }
 138
 139 # cache image chado object - image links to prevent reloading of the
 140 # same data
 141 #
 142 print "Caching image-stock links...\n";
 143
 144 my $q = "SELECT * FROM phenome.stock_image";
 145 my $sth = $dbh->prepare($q);
 146 $sth->execute();
 147 while ( my $hashref = $sth->fetchrow_hashref() ) {
 148     my $image_id = $hashref->{image_id};
 149     my $chado_table_id = $hashref->{stock_id};  ##### table specific
 150     my $i = SGN::Image->new($dbh, $image_id);
 151     my $original_filename = $i->get_original_filename();
 152     $image_hash{$original_filename} = $i; # this doesn't have the file extension
 153     $connections{$image_id."-".$chado_table_id}++;
 154 }
 155
 156 open (ERR, ">load_solcap_images.err") || die "Can't open error file\n";
 157
 158 my @files = glob "$dirname/*.$ext";
 159 @files = glob "$dirname/*" if $opt_d ;
 160 my @sub_files;
 161
 162 my $new_image_count = 0;
 163
 164
 165 foreach my $file (@files) {
 166     eval {
 167         chomp($file);
 168         @sub_files = ($file);
 169         @sub_files =  glob "$file/*.$ext" if $opt_d;
 170
 171         my $object_name = basename($file, ".$ext" );
 172         print  "object_name = '".$object_name."' \n";
 173         #$individual_name =~s/(W\d{3,4}).*\.JPG/$1/i if $individual_name =~m/^W\d{3}/;
 174         #2009_oh_8902_fruit-t
 175         # solcap images:
 176         #my ($year, $place, $plot, undef) = split /_/ , $object_name;
 177         #2009_ca_2138_fruit_c.jpg
 178         my ($year, $place, $name, undef,undef) = split /_/ , $object_name;
 179
 180         #my $stock = $schema->resultset("Stock::Stock")->find( {
 181         #    stock_id => $name2id{ lc($name) }  } );
 182         my ($stock) = $schema->resultset("Stock::Stockprop")->search(
 183             { value => $location, })->search_related('stock', { name=> $name } );
 184
 185         if (!$stock) {
 186             warn "no stock found for plot # $name ! Skipping !!\n\n";
 187             next();
 188         }
 189
 190         foreach my $filename (@sub_files) {
 191             chomp $filename;
 192             print STDOUT "Processing file $file...\n";
 193             print STDOUT "Loading $name, image $filename\n";
 194             print ERR "Loading $name, image $filename\n";
 195             my $image_id; # this will be set later, depending if the image is new or not
 196             if (! -e $filename) {
 197                 warn "The specified file $filename does not exist! Skipping...\n";
 198                 next();
 199             }
 200
 201             if (!exists($name2id{lc($name)})) {
 202                 message ("$name does not exist in the database...\n");
 203             }
 204
 205             else {
 206                 print ERR "Adding $filename...\n";
 207                 if (exists($image_hash{$filename})) {
 208                     print ERR "$filename is already loaded into the database...\n";
 209                     $image_id = $image_hash{$filename}->get_image_id();
 210                     $connections{$image_id."-".$name2id{lc($name)}}++;
 211                     if ($connections{$image_id."-".$name2id{lc($name)}} > 1) {
 212                         print ERR "The connection between $name and image $filename has already been made. Skipping...\n";
 213                     }
 214                     elsif ($image_hash{$filename}) {
 215                         print ERR qq  { Associating stock $name2id{lc($name)} with already loaded image $filename...\n };
 216                         ################################
 217                     }
 218                 }
 219                 else {
 220                     print ERR qq { Generating new image object for image $filename and associating it with stock $name, id $name2id{lc($name) } ...\n };
 221                     my $caption = $name;
 222
 223                     if ($opt_t)  {
 224                         print STDOUT qq { Would associate file $filename to stock $name, id $name2id{lc($name)}\n };
 225                         $new_image_count++;
 226                     }
 227                     else {
 228                         my $image = SGN::Image->new($dbh);
 229                         $image_hash{$filename}=$image;
 230
 231                         $image->process_image("$filename", undef, undef);
 232                         $image->set_description("$caption");
 233                         $image->set_name(basename($filename , ".$ext"));
 234                         $image->set_sp_person_id($sp_person_id);
 235                         $image->set_obsolete("f");
 236                         $image_id = $image->store();
 237                         $new_image_count++;
 238                     }
 239                 }
 240             }
 241             my $metadata = CXGN::Metadata::Metadbdata->new($metadata_schema, $sp_person);
 242             my $metadata_id = $metadata->store()->get_metadata_id();
 243             die "NO METADATA ID FOUND\n" if !$metadata_id;
 244 #store the image_id - stock_id link
 245             my $q = "INSERT INTO phenome.stock_image (stock_id, image_id, metadata_id) VALUES (?,?,?)";
 246             my $sth  = $dbh->prepare($q);
 247             $sth->execute($stock->stock_id, $image_id, $metadata_id);
 248         }
 249     };
 250     if ($@) {
 251         print STDOUT "ERROR OCCURRED WHILE SAVING NEW INFORMATION. $@\n";
 252         $dbh->rollback();
 253     }
 254     else {
 255         $dbh->commit();
 256     }
 257 }
 258
 259
 260 close(ERR);
 261 close(F);
 262
 263
 264 print STDOUT "Inserted  $new_image_count images.\n";
 265 print STDOUT "Done. \n";
 266
 267 sub usage {
 268     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";
 269     exit();
 270 }
 271
 272 sub message {
 273     my $message=shift;
 274     print STDOUT $message;
 275     print ERR $message;
 276 }