function showSelect($videos, $select)
{
print '<option value=""></option>';
$maxsim = 0;
foreach ($videos as $vid) {
$thissim = 0;
$sim = similarity($select, $vid[title]);
if ($sim > $thissim) {
$thissim = $sim;
}
$sim = similarity($select, $vid[subtitle]);
if ($sim > $thissim) {
$thissim = $sim;
}
$sim = similarity($select, $vid[filename]);
if ($sim > $thissim) {
$thissim = $sim;
}
if ($thissim > $maxsim) {
$SEL = "selected";
$maxsim = $thissim;
} else {
$SEL = "";
}
print '<option value="' . $vid[id] . '" ' . $SEL . '>';
print $vid[title];
print ' - ';
print $vid[subtitle];
print ' (' . $thissim . '%)';
print '</option>';
}
}
if (isset($_GET["smiles"])) {
$variable = $_GET["smiles"];
} elseif (isset($_GET["smarts"])) {
$variable = convertSMARTS($_GET["smarts"]);
}
//Second section depending of the optional inputs
if (isset($variable) && isset($_GET["fingerprint"]) && isset($_GET["method"])) {
$value = similarity($variable, $_GET["fingerprint"], $_GET["method"]);
} else {
if (isset($variable) && isset($_GET["fingerprint"])) {
$value = similarity($variable, $_GET["fingerprint"], "Tanimoto");
} elseif (isset($variable) && isset($_GET["method"])) {
$value = similarity($variable, "Morgan", $_GET["method"]);
} else {
if (isset($variable)) {
$value = similarity($variable, "Morgan", "Tanimoto");
} else {
$value = "Missing argument";
}
}
}
break;
case "properties":
if (isset($_GET["smiles"])) {
$variable = $_GET["smiles"];
$value = properties($variable);
} elseif (isset($_GET["smarts"])) {
$variable = convertSMARTS($_GET["smarts"]);
$value = properties($variable);
} else {
$value = "Missing argument";
function group_list($branch_list, $depth, $min_similarity)
{
$min_similarity = 2;
$depth = 2;
// $branch_list est un array de labels composés de ngrammes séparés par des virgules
/// retour un array contenant
// ['grouped_indexes']: un array de groupes d'indices mentionnant les labels similaires
// ['Ngram_arrays']: un array contenant pour chaque groupe d'indices, un array pour labelliser le groupe dont les clés sont des NGram pris jusqu'à une profondeur $depth et les valeurs leur nombre d'occurrences dans les labels
// calcul des similarités entre labels de branches
$label_rows_groups = array();
$label_groups = array();
$nb_labels_treated = 0;
$label_rows_remaining_to_process = array();
// index des labels non encore catégorisés
// initialisation de ce tableau
for ($i = 0; $i < count($branch_list); $i++) {
array_push($label_rows_remaining_to_process, $i);
}
while (count($label_rows_remaining_to_process) > 0) {
$target_row = array_pop($label_rows_remaining_to_process);
$target_label_raw = array();
// liste des rangs groupés avec target_row
$target_labels = array();
// liste des labels des rangs associés
array_push($target_labels, $branch_list[$target_row][label]);
array_push($target_label_raw, $target_row);
$label_rows_remaining_to_process_second_pass = $label_rows_remaining_to_process;
while (count($label_rows_remaining_to_process_second_pass) > 0) {
$candidate_row = array_pop($label_rows_remaining_to_process_second_pass);
$candidate_label = remove_popo($branch_list[$candidate_row][label]);
$exit_here = 0;
$j = 0;
while ($j < count($target_labels) && exit_here == 0) {
$p = similarity($candidate_label, remove_popo($target_labels[$j]));
// echo 'candidate labels'.$candidate_label.'<br/>';
// echo 'target labels'.$target_labels[$j].'<br/>';
// echo $p.'<br/>';
//similar_text ( $candidate_label,$target_labels[$j],$p );
if ($p > $min_similarity) {
$exit_here = 1;
// echo 'similarity: '.$p.'<br/>';
// echo 'candidate labels '.$candidate_label.'<br/>';
// echo 'target labels' .$target_labels[$j].'<br/><br/> ';
}
$j++;
}
if ($exit_here == 1) {
//echo $branch_list[$candidate_row][label].'<br/>';
array_push($target_labels, $branch_list[$candidate_row][label]);
array_push($target_label_raw, $candidate_row);
$label_rows_remaining_to_process = array_diff($label_rows_remaining_to_process, $target_label_raw);
array_push($target_branches, $candidate_branch);
array_push($target_label_raw, $candidate_row);
}
}
array_push($label_rows_groups, $target_label_raw);
/////////////////
// on reconstruit l'array des labels_id de branche
$target_branches_label_ids = array();
for ($i = 0; $i < count($target_labels); $i++) {
array_push($target_branches_label_ids, $target_labels[$i]['label_ids']);
}
if (count($target_branches_label_ids) > 3) {
array_push($label_groups, groups_labels($target_branches_label_ids, 4, $target_label_raw, $branch_list));
} else {
array_push($label_groups, groups_labels($target_branches_label_ids, $depth, $target_label_raw, $branch_list));
}
/////////////////
array_push($label_groups, label_label_group($target_labels, $depth));
$resultat = array();
}
/// on réordonne par nombre de sujets groupés
$groups_sizes = array();
for ($i = 0; $i < count($label_rows_groups); $i++) {
array_push($groups_sizes, count($label_rows_groups[$i]));
}
uasort($groups_sizes, 'compare');
$new_order = array_keys($groups_sizes);
//
$label_rows_groups_temp = array();
for ($i = 0; $i < count($label_rows_groups); $i++) {
array_push($label_rows_groups_temp, $label_rows_groups[$new_order[$i]]);
}
$label_rows_groups = $label_rows_groups_temp;
//
$label_groups_temp = array();
for ($i = 0; $i < count($label_groups); $i++) {
array_push($label_groups_temp, $label_groups[$new_order[$i]]);
}
$label_groups = $label_groups_temp;
//
$resultat['grouped_indexes'] = $label_rows_groups;
$resultat['Ngram_arrays'] = $label_groups;
return $resultat;
}
