#!/usr/bin/perl
use Image::Size;
@concepts = ("airplane", "airport", "animal", "apple", "architecture", "art", "autumn", "baby", "ball", "bar", "barcelona", "beach", "bear", "beauty", "beer", "berlin", "bicycle", "bird", "birthday", "blue", "boat", "book", "boy", "bridge", "brown", "building", "bus", "butterfly", "cake", "california", "camera", "candle", "car", "castle", "cat", "cathedral", "chair", "child", "chocolate", "christmas", "church", "city", "clock",  "clouds", "coffee", "concert", "couple", "cow", "cross", "cute",  "dance", "dinner", "dog", "doll", "door", "dress", "duck", "easter", "england", "europe", "eye", "feet", "field", "film", "fire", "fishing", "flower", "food", "football", "forest", "fountain", "france", "funny", "garden", "germany", "girl", "graffiti", "guitar", "hair", "halloween", "hand", "happy", "hat", "heart", "holiday", "horse", "hot", "house", "ice", "island", "italy", "kiss", "kitchen", "lake", "lamp",  "leaves", "lighthouse", "london", "love", "man", "mirror", "model", "monument", "moon", "mountain", "museum", "newyork", "night",  "painting", "paris", "park", "party", "pet",   "pink", "pool", "portrait", "poster", "pub", "red", "river", "road", "rose", "sculpture", "sea", "selfportrait", "shoes", "shopping", "silhouette", "smile", "spain", "spider", "sport", "spring", "square", "statue", "storm", "street", "summer", "sunrise", "sunset", "table", "tower", "town", "train", "usa", "waterfall", "wedding", "window", "winter", "woman");
#@concepts = ("museum", "hot", "football", "statue", "lighthouse", "cake", "chocolate");
#@concepts = ("museum");
$ext_disk = "/media/disk/gender";
$loco = "flickr_gender";
open AW, "weights/all_unique";
@all = <AW>;
close AW;
$excluded = " nb noiretblanc hdr bw blackandwhite blanc noir closeup ";
foreach $con (@concepts)
{
 $timo = localtime();
 print "$con $timo\n";
 $out_con = "$loco/$con";
 if(!(-e $out_con)){mkdir $out_con;}
 $list_flickr = "flickr_metadata/$con";
 $out_fem = "$out_con/female_rank";
 $close_fem = "$out_con/close_female";
 $close_mal = "$out_con/close_male";
 if(!(-e $out_fem))
 {
  $list_concepts = "$ext_disk/concepts/$con/female_weights_all";
  $run = &rankGender($list_concepts, $list_flickr, $out_fem, $close_fem);
 }
 $out_mal = "$out_con/male_rank";
 if(!(-e $out_mal))
 {
  $list_concepts = "$ext_disk/concepts/$con/male_weights_all";
  $run = &rankGender($list_concepts, $list_flickr, $out_mal, $close_mal);
 }
 $orig_res = "$out_con/original_rank";
 if(!(-e $orig_res))
 {
  open OUTO, ">$orig_res";
  open MT, $list_flickr or die "cant open $meta_file\n";
  @metas = <MT>; 
  close MT;
  @toRerank = ();
  $initRank = 0;
  foreach $linio (@metas)
  {
   if($linio =~ m/photo id/)
   {
    $initRank++;
    $owner = substr($linio, index($linio, 'owner="')+7, index($linio, '"', index($linio, 'owner="')+7) - (index($linio, 'owner="')+7));
    $id = substr($linio, index($linio, 'o id="')+6, index($linio, '"', index($linio, 'o id="')+6) - (index($linio, 'o id="')+6)); 
    $tags = substr($linio, index($linio, 'tags="')+6, index($linio, '"', index($linio, 'tags="')+6) - (index($linio, 'tags="')+6));    
    $secret = substr($linio, index($linio, 'secret="')+8, index($linio, '"', index($linio, 'secret="')+8) - (index($linio, 'secret="')+8));
    $server = substr($linio, index($linio, 'server="')+8, index($linio, '"', index($linio, 'server="')+8) - (index($linio, 'server="')+8));
    $farm = substr($linio, index($linio, 'farm="')+6, index($linio, '"', index($linio, 'farm="')+6) - (index($linio, 'farm="')+6)); 
    $url = "http://farm$farm.static.flickr.com/$server/".$id."_$secret".".jpg"; 
    print OUTO "$initRank\t$id\t$owner\txyz\t$tags\t$url\n";
   }
  }
  close OUTO;
 }
 $html_file = "$out_con/$con".'.html';
 #print "$html_file\n";
 if(!(-e $html_file) || 1 == 1)
 {
  $run = &makeHTML($orig_res, $out_fem, $out_mal, $html_file, $con); 
  #print "$con in html $html_file\n";
 }
}

#routine for ranking with gender
sub rankGender
{
 $lst = $_[0];
 $flickr = $_[1];
 $out = $_[2];
 $close_out = $_[3];
 #print "$close_out\n";
 if(!(-e $close_out))
 {
  open LS, $lst or die "cant open $lst\n"; 
  @close1 = <LS>;
  close LS;
  @close_unranked = ();
  for($c1 = 0; $c1 <= 199; $c1++)
  {
   @cut1 = split(/\t/, @close1[$c1]);
   $cu = @cut1[2];
   if(index($excluded, " $cu ") == -1)
   {
    $local_score = 0;
    $local_score2 = 0;
    $global_score = 0;
    $got = 0;  
    $tr = 0;
    $splits = 2048;
    $min = 0;
    $max = 0;
    $search = 0;
    while($search <= $splits && $min == 0 && $max == 0)
    {
     $int = int(($search/$splits) * ($#all+1));
     chomp(@all[$int]);
     @cut_tst = split(/\t/, @all[$int]); 
     if(@cut_tst[0] gt $cu)
     {
      $min = int((($search-1)/$splits) * ($#all+1));
      $max = $int;
      if($min < 0){$min = 0;}
     }
     $search++;
    }
    while($min <= $max && $got == 0)
    {
     chomp(@all[$min]);
     @cut_now = split(/  /, @all[$min]);
     if($cu eq @cut_now[0])
     { 
      $global_score = @cut_now[1];
     }   
     $min++;
    }
    $local_score = @cut1[0];
    $local_score2 = @cut1[1];
    $local_score =~ s/^0*//;
    $local_score2 =~ s/^0*//;
    $norm_local2 = &normalizer($local_score2);
    $log1 = &log2($local_score);
    $fraction = $log1*($local_score/$global_score);
    $fraction = $fraction/10;
    if(index($fraction, 'e-0')  == -1){push(@close_unranked, "$fraction\t$norm_local2\t$cu\t$global_score");}
    #print "$fraction\t$norm_local2\t$cu\t$global_score\n";
   }
  }  
  @closos = reverse sort @close_unranked;
  $max = $#closos;
  #print "@closos\n";
  if($max > 99){$max = 99;}
  open OUTC, ">$close_out" or die "cant open out $close_out\n";
  for($cc = 0; $cc <= $max; $cc++)
  {
   print OUTC "@closos[$cc]\n";
  }
  close OUTC;
 }
 else{print "exists $close_out\n";}
 open CLO, $close_out;
 @close = <CLO>;
 close CLO; 
 #print "@close[0]@close[1]@close[2]\n";
 open FL, $flickr or die "cant open fl $flickr\n";
 @photos = <FL>;
 close FL;
 @toRank = 0;
 $initRank = 0;
 $page = 1;
 for($pp = 0; $pp <= $#photos; $pp++)
 {
  if(index(@photos[$pp], 'page="') != -1)
  {
   $page = substr(@photos[$pp], index(@photos[$pp], 'page="')+6, index(@photos[$pp], '"', index(@photos[$pp], 'page="')+6) - (index(@photos[$pp], 'page="')+6));
   #print "$page\n";
   $counter = 0; 
   @gots = ();
  }
  if(@photos[$pp] =~ m/photo id/)
  {
   $prop_all = 0;
   $initRank++;
   $tags = substr(@photos[$pp], index(@photos[$pp], 'tags="')+6, index(@photos[$pp], '"', index(@photos[$pp], 'tags="')+6) - (index(@photos[$pp], 'tags="')+6));
   $owner = substr(@photos[$pp], index(@photos[$pp], 'owner="')+7, index(@photos[$pp], '"', index(@photos[$pp], 'owner="')+7) - (index(@photos[$pp], 'owner="')+7));
   $id = substr(@photos[$pp], index(@photos[$pp], 'o id="')+6, index(@photos[$pp], '"', index(@photos[$pp], 'o id="')+6) - (index(@photos[$pp], 'o id="')+6)); 
   $secret = substr(@photos[$pp], index(@photos[$pp], 'secret="')+8, index(@photos[$pp], '"', index(@photos[$pp], 'secret="')+8) - (index(@photos[$pp], 'secret="')+8));
   $server = substr(@photos[$pp], index(@photos[$pp], 'server="')+8, index(@photos[$pp], '"', index(@photos[$pp], 'server="')+8) - (index(@photos[$pp], 'server="')+8));
   $farm = substr(@photos[$pp], index(@photos[$pp], 'farm="')+6, index(@photos[$pp], '"', index(@photos[$pp], 'farm="')+6) - (index(@photos[$pp], 'farm="')+6));  
   $url = "http://farm$farm.static.flickr.com/$server/".$id."_$secret".".jpg";
   $views = substr(@photos[$pp], index(@photos[$pp], 'views="')+7, index(@photos[$pp], '"', index(@photos[$pp], 'views="')+7) - (index(@photos[$pp], 'views="')+7));  
   $counter++;
   $pos = $page * 500 + $counter;
   $inv_pos = 99999999999 - $pos;
   @cut_tags = split(/ /, $tags);
   @ord_tags = sort @cut_tags;
   $tagos = "";
   $found = 0;
   #make sure there is at least two tags - flickr recommendation...
   if($#cut_tags >= 1)
   {
    $text = " $tags ";
    $rank = 4;
    for($mm = 0; $mm <= $#close; $mm++)
    {
     chomp(@close[$mm]);  
     @cut_close = split(/\t/, @close[$mm]);
     if($mm < 4){$rank = 4;}
     else{$rank = $mm;}
     if(index($text, " @cut_close[2] ") != -1 && $found < 3 && @gots[$mm] == 0)
     {
      @gots[$mm]++; 
      $tagos = $tagos.' '.@cut_close[2];
      $found++;
      @cut_close[0] =~ s/^1//; 
      @cut_close[0] =~ s/^0*//;
      if(@cut_close[3] > 0)  
      {
       $rank_contrib = 1/&log2($rank);
       $prop_all = $prop_all + $rank_contrib;
      }
     }
    }
    $prop_all = $prop_all;
    $score = $prop_all;
    $inv_tags = $views;
    if(index("@toRank", "$url") == -1 && length($tagos) > 0)
    {
     push(@toRank, "$score $inv_pos\t$id\t$owner\t$initRank\t$tagos\t$url");
    }
   }
  }
 } 
 @sorted = reverse sort @toRank;
 @usorted = ();
 open OUTP, ">$out";
 for($ss = 0; $ss <= $#sorted; $ss++)
 {
  @cut_ss = split(/\t/, @sorted[$ss]);
  @cut_ss[4] =~ s/^ *//;
  @cut_ss[4] =~ s/ *$//;
  @cut_sorted = split(/ /, @cut_ss[4]);
  $new = 0;
  foreach $cs (@cut_sorted)
  {
   if(index(" @usorted ", " $cs ") == -1)
   {
    push(@usorted, $cs);
    $new++;
   }
  }  
  if($new > 0)
  {
   #print "@sorted[$ss]\n";
   print OUTP "@sorted[$ss]\n";
  }
 }
 close OUTP;}



sub normalizer
{
 $ret = $_[0];
 if(length($ret) == 8){$ret = '0'.$ret;}
 elsif(length($ret) == 7){$ret = '00'.$ret;}
 elsif(length($ret) == 6){$ret = '000'.$ret;}
 elsif(length($ret) == 5){$ret = '0000'.$ret;}
 elsif(length($ret) == 4){$ret = '00000'.$ret;}
 elsif(length($ret) == 3){$ret = '000000'.$ret;}
 elsif(length($ret) == 2){$ret = '0000000'.$ret;}
 elsif(length($ret) == 1){$ret = '00000000'.$ret;}
 return $ret;
}


#routine for creating HTML files with results
sub makeHTML
{
 $orig = $_[0];
 $fem = $_[1];
 $mal = $_[2];
 $html_file = $_[3];
 $titolo = $_[4];
 $img_dir = "/mnt/user_survey/gender_concepts/flickr_images/$titolo";
 open OUTHTML, ">$html_file" or die "cannot open $html_file\n";
 print OUTHTML '<html>';
 print OUTHTML '<head>';
 print OUTHTML "<title>Answers for $titolo</title>";
 print OUTHTML '</head>';
 print OUTHTML "<body>\n";
 print OUTHTML "<h3>Query: $titolo<br><h3>\n";
 print OUTHTML "<table CELLPADDING=\"3\">\n";
 print OUTHTML "<tr><td>";
 print OUTHTML "</td></tr>";
 print OUTHTML "<tr>";
 print OUTHTML "<td><table>";
 print OUTHTML "<tr><td><table>";
 print OUTHTML "<tr><td>";
 #print the images for original results 
 print OUTHTML "<table>";
 print OUTHTML "<tr><td>original</td></tr>";
 open OR, $orig;
 @images_raw = <OR>;
 close OR;
 $ret = 0;
 @images = ();
 $already = '  ';
 $tot = 0;
 while($ret < 20 && $tot <= $#images_raw)
 {
  chomp(@images_raw[$tot]);
  @cut_image = split(/\t/, @images_raw[$tot]);
  if(index($already, "  @cut_image[4]  ") == -1)
  {
   push(@images, @images_raw[$tot]);
   $already = $already."  @cut_image[4]  ";
   $ret++;
  }
  $tot++;
 }
 for($ll = 0; $ll < 5; $ll++)
 {
  $offset = $ll * 2;
  print OUTHTML "<tr>";
  for($cols = 0; $cols < 2; $cols++)
  {
   $now = $offset+$cols;
   chomp(@images[$now]);
   @spl_answer = split(/\t/, @images[$now]);
   #print "@images[$now]\n";
   $title = "$img_dir/@spl_answer[1]".'_'.@spl_answer[2].'.jpg';
   @infos_image = imgsize($title);
   if ($infos_image[0] > $infos_image[1]){ $taille =  'width="160" ';}
   else{ $taille =  'height="160" ';}
   print OUTHTML "<td height =\"170\" width =\"170\"><a href=\"$title\"><img ".$taille." src =\"$title\" title=\"tags: @spl_answer[4]\"></a></td>";
  }
  print OUTHTML "</tr>";
 }
 print OUTHTML "</table>";
 print OUTHTML "</td>";
 #print the images for female results 
 print OUTHTML "<td width=\"50\"><p>&nbsp;</p></td>";
 print OUTHTML "<td><table>";
 print OUTHTML "<tr><td>female</td></tr>";
 open OR, $fem;
 @images_raw = <OR>;
 close OR;
 $ret = 0;
 $tot = 0;
 @images = ();
 $already = '  ';
 while($ret < 20 && $tot <= $#images_raw)
 {
  chomp(@images_raw[$tot]);
  @cut_image = split(/\t/, @images_raw[$tot]);
  if(index($already, "  @cut_image[4]  ") == -1)
  {
   push(@images, @images_raw[$tot]);
   $already = $already."  @cut_image[4]  ";
   $ret++;
  }
  $tot++;
 }
 for($ll = 0; $ll < 5; $ll++)
 {
  $offset = $ll * 2;
  print OUTHTML "<tr>";
  for($cols = 0; $cols < 2; $cols++)
  {
   $now = $offset+$cols;
   chomp(@images[$now]);
   @spl_answer = split(/\t/, @images[$now]);
   #print "@spl_answer[4]\n";
   $title = "$img_dir/@spl_answer[1]".'_'.@spl_answer[2].'.jpg';
   @infos_image = imgsize($title);
   if ($infos_image[0] > $infos_image[1]){ $taille =  'width="160" ';}
   else{ $taille =  'height="160" ';}
   print OUTHTML "<td height =\"170\" width =\"170\"><a href=\"$title\"><img ".$taille." src =\"$title\" title=\"tags: @spl_answer[4]\"></a></td>";
  }
  print OUTHTML "</tr>";
 }
 print OUTHTML "</table>";
 print OUTHTML "</td>";
 print OUTHTML "<td width=\"50\"><p>&nbsp;</p></td>";
 #print the images for male results 
 print OUTHTML "<td><table>";
 print OUTHTML "<tr><td>male</td></tr>";
 open OR, $mal;
 @images_raw = <OR>;
 close OR;
 $ret = 0;
 $tot = 0;
 @images = ();
 $already = '  ';
 while($ret < 20 && $tot <= $#images_raw)
 {
  chomp(@images_raw[$tot]);
  @cut_image = split(/\t/, @images_raw[$tot]);
  if(index($already, "  @cut_image[4]  ") == -1)
  {
   push(@images, @images_raw[$tot]);
   $already = $already."  @cut_image[4]  ";
   $ret++;
  }
  $tot++;
 }
 for($ll = 0; $ll < 5; $ll++)
 {
  $offset = $ll * 2;
  print OUTHTML "<tr>";
  for($cols = 0; $cols < 2; $cols++)
  {
   $now = $offset+$cols;
   chomp(@images[$now]);
   @spl_answer = split(/\t/, @images[$now]);
   $title = "$img_dir/@spl_answer[1]".'_'.@spl_answer[2].'.jpg';
   @infos_image = imgsize($title);
   if ($infos_image[0] > $infos_image[1]){ $taille =  'width="160" ';}
   else{ $taille =  'height="160" ';}
   print OUTHTML "<td height =\"170\" width =\"170\"><a href=\"$title\"><img ".$taille." src =\"$title\" title=\"tags: @spl_answer[4]\"></a></td>";
  }
  print OUTHTML "</tr>";
 }
 print OUTHTML "</table>";
 print OUTHTML "</td></tr>";
 print OUTHTML "</table></td></tr>";
 print OUTHTML "</table></td>";
 print OUTHTML "</tr>";
 print OUTHTML "</table>\n";
 print OUTHTML '</body>';
 print OUTHTML '</html>';
 close OUTHTML;
}

#routine for log file
sub log2 
{
 my $n = shift;
 return (log($n)/log(2));
}