function doSearch(array $words, array $searchWords, $limit, $searchFunction)
{
foreach ($searchWords as $searchWord) {
$words = array_merge($words, $searchFunction($searchWord, $limit));
$words = removeDuplicates($words);
if (count($words) >= $limit) {
array_splice($words, $limit);
}
}
return $words;
}
listSynonyms($synonyms);
} else {
//IF REWRITE IS NOT REQUESTED, PROCESS QUERY:
//CALL FUNCTIONS TO GET RESULTS FROM SEARCH ENGINES
getBlekkoResults($blekkoResultsOrig);
//get Blekko
getGoogleResults($googleResultsOrig);
//get Google
getBingResults($bingResultsOrig);
//get Bing
//**************************************************************************
//**************************************************************************
//CALL FUNCTIONS TO REMOVE DUPLICATE URLS AND RE-SCORE DOCUMENTS
removeDuplicates($blekkoResultsOrig, $blekkoResults);
removeDuplicates($googleResultsOrig, $googleResults);
removeDuplicates($bingResultsOrig, $bingResults);
//**************************************************************************
//**************************************************************************
//(IF NON-AGGREGATED CHOSEN THEN DISPLAY RESULTS:)
if ($_POST['result_type'] === 'Non-Aggregated') {
echo "<div class=\"row-fluid\">" . "<div class=\"span4\">";
echo "<h2 class=\"text-center\">Bing</h2>";
displayResults($bingResults);
//display Bing
echo "</div>";
echo "<div class=\"span4\">";
echo "<h2 class=\"text-center\">Google</h2>";
displayResults($googleResults);
//display Google
echo "</div>";
echo "<div class=\"span4\">";
* findDuplicates([1,2, 3, "jablko", 'banan']) => [1,2, 3, "jablko", 'banan']
*/
/**
* @param $myArray
* @return array
*/
function removeDuplicates($myArray)
{
$removeDuplicates = array_unique($myArray);
$j = 0;
$results = array();
foreach ($removeDuplicates as $value) {
$results[$j] = $value;
$j++;
}
return $results;
}
/*
* Kod popniżej służy wygenerowaniu testów i strony poglądaowej - nie modyfikujcie go!
*/
$testCases = [[[1, 2, "jan", "kowlaski", 2, "jan"], [1, 2, "jan", "kowlaski"]], [[3.4, "jablko", 'banan', 'banan', "gruszka", 5], [3.4, "jablko", 'banan', "gruszka", 5]], [[1, 2, 3, "jablko", 'banan'], [1, 2, 3, "jablko", 'banan']], [[], []], [[1, 2, 3, 4], [1, 2, 3, 4]]];
$results = "";
foreach ($testCases as $case) {
if (($funcValue = removeDuplicates($case[0])) === $case[1]) {
$results .= "<tr class='success'><td> Ok </td><td>[" . implode($case[0], ",") . "] </td><td>[" . implode($case[1], ",") . "]</td><td>[" . implode($funcValue, ",") . "]</td></tr>";
} else {
var_dump($funcValue);
$results .= "<tr class='danger'><td> Fail </td><td> [" . implode($case[0], ",") . "] </td><td>[" . implode($case[1], ",") . "]</td><td>[" . implode($funcValue, ",") . "]</td></tr>";
}
}
echo "\n<!DOCTYPE html>\n<html lang='pl'>\n <head>\n <meta charset='utf-8'>\n <title>Usuń duplikaty</title>\n <link href='./css/bootstrap.min.css' rel='stylesheet'>\n\n </head>\n\n <body>\n\n <div class='container'>\n <div class='jumbotron'>\n <h1>Usuń duplikaty</h1>\n <p><a class='btn btn-lg btn-success' href='javascript:window.location.reload();' role='button'>Odświerz</a></p>\n </div>\n\n <div class='row'>\n <h1>Testy:</h1>\n <table class='table'>\n <thead>\n <tr>\n <th>Stan testu:</th>\n <th>Dane:</th>\n <th>Wartość spodziewana:</th>\n <th>Wartość otrzymana:</th>\n </tr>\n </thead>\n <tbody>\n {$results}\n </tbody>\n </table>\n </div>\n </div>\n </body>\n</html>\n";
function getBooks($dbpid)
{
$format = 'json';
$query = 'PREFIX dbp: <http://dbpedia.org/resource/>
PREFIX dbp2: <http://dbpedia.org/ontology/>
SELECT ?book,?author,?name,?thumbnail,?published,?released,?wikilink,?abstract
WHERE {
?book a dbpedia-owl:Book.
?book dbpedia-owl:author ?author.
?book foaf:name ?name.
?book prov:wasDerivedFrom ?wikilink.
OPTIONAL{
?book dbpprop:published ?published.
}
OPTIONAL{
?book dbpprop:releaseDate ?released.
}
OPTIONAL{
?book dbpedia-owl:thumbnail ?thumbnail.
}
OPTIONAL{
?book dbpedia-owl:abstract ?abstract
FILTER(LANG(?abstract)="de")
}
FILTER(?author=<' . $dbpid . '>)
}
LIMIT 100';
$searchUrl = 'http://dbpedia.org/sparql?' . 'query=' . urlencode($query) . '&format=' . $format;
$dbPediaResult = DDPediaRequest($searchUrl);
$jsonResult = json_decode($dbPediaResult);
if (property_exists($jsonResult, "results") && property_exists($jsonResult->results, "bindings") && count($jsonResult->results->bindings) > 0) {
return removeDuplicates($jsonResult->results->bindings, "book");
} else {
return array();
}
}
if ($tempArray[$i]['drug_code'] === $resultArray[$j]['drug_code']) {
array_push($resultArray[$j]['ingredients'], array('ingredient' => $tempArray[$i]['ingredient'], 'ai_code' => $tempArray[$i]['active_ingredient_code'], 'strength' => formatStrength($tempArray[$i]['strength']), 'strength_unit' => $tempArray[$i]['strength_unit'], 'dosage_form' => $tempArray[$i]['pharmaceutical_form']));
$match = true;
}
}
if ($match === false) {
array_push($resultArray, array('drug_code' => $tempArray[$i]['drug_code'], 'ingredients' => array(array('ingredient' => $tempArray[$i]['ingredient'], 'ai_code' => $tempArray[$i]['active_ingredient_code'], 'strength' => formatStrength($tempArray[$i]['strength']), 'strength_unit' => $tempArray[$i]['strength_unit'], 'dosage_form' => $tempArray[$i]['pharmaceutical_form'])), 'match' => 0));
}
}
}
// Sorts the ingredient array by generic_name
for ($i = 0; $i < count($resultArray); $i++) {
usort($resultArray[$i]['ingredients'], "sortAICode");
}
// Removes entries with duplicate ingredients
$resultArray = removeDuplicates($resultArray);
}
/****************************************************************************
* FINDING DRUG MATCH */
/****************************************************************************
* Uses the result array returned above to see if there is a good match *
* A good match means that all the strengths for a drug code match the $q *
* provided and only one such entry meets this requirement *
* *
* Runs only if $resultArray has at least one entry *
* *
* If a match is found, returns an array of matches *
****************************************************************************/
if (count($resultArray) > 0) {
$errorReport .= "FINDING DRUG MATCH\n";
// Cycles through all the resultArray entries and determine match status
function makeSlugs($string = '', $maxlen = 0)
{
$newStringTab = array();
$string = strtolower(noDiacritics($string));
if (function_exists('str_split')) {
$stringTab = str_split($string);
} else {
$stringTab = my_str_split($string);
}
$numbers = array("0", "1", "2", "3", "4", "5", "6", "7", "8", "9", "-");
//$numbers=array("0","1","2","3","4","5","6","7","8","9");
foreach ($stringTab as $letter) {
if (in_array($letter, range("a", "z")) || in_array($letter, $numbers)) {
$newStringTab[] = $letter;
//print($letter);
} elseif ($letter == " ") {
$newStringTab[] = "-";
}
}
if (count($newStringTab)) {
$newString = implode($newStringTab);
if ($maxlen > 0) {
$newString = substr($newString, 0, $maxlen);
}
$newString = removeDuplicates('--', '-', $newString);
} else {
$newString = '';
}
return $newString;
}
$outname = substr($file, $rdLen);
$zip->addFile($file, $outname);
}
}
}
}
foreach ($fileList['files'] as $file) {
$outname = substr($file, $rdLen);
$zip->addFile($file, $outname);
}
$res = $zip->close();
return $destination;
}
// Output the directory listing as JSON
$fileList = cleanseFileList();
$fileList = removeDuplicates($fileList);
if (count($fileList['folders']) || count($fileList['files'])) {
$fullName = zipTheFiles($fileList);
if ($fullName) {
$docRoot = $_SERVER['DOCUMENT_ROOT'];
$output = substr($fullName, strlen($docRoot . '/tmp/'));
}
} else {
$output = '';
}
$info = "\n{$fullName}";
$info .= ' ' . filesize($fullName);
header("Pragma: public");
header("Expires: 0");
header("Cache-Control: must-revalidate, post-check=0, pre-check=0");
header("Cache-Control: public");
//take first on and replace in towns
if (count($ids) > 1) {
print '<br>' . $name . ' ' . count($ids);
}
removeDuplicates($ids);
$name = $row[1];
$gov_id = $row[2];
//if(count($ids)>1) exit();
$ids = array();
}
array_push($ids, $row[0]);
}
if (count($ids) > 1) {
print '<br>' . $name . ' ' . count($ids);
}
removeDuplicates($ids);
print '<br>Removed: ' . $removed;
//
//
//
function removeDuplicates($ids)
{
global $mysqli, $removed;
if (count($ids) > 1) {
//take first on and replace in towns
$rec_id = array_shift($ids);
$ids_todel = implode(',', $ids);
$query = 'update recDetails set dtl_Value=' . $rec_id . ' where dtl_Value in (' . $ids_todel . ') and dtl_DetailTypeID=170';
//print $query; return;
$res = $mysqli->query($query);
$mysqli->query('SET foreign_key_checks = 0');
$sqlCheckTableExists = "SELECT COUNT(*) FROM information_schema.tables WHERE table_schema = 'international_cities_rrr' AND table_name = '" . $country['short_name'] . "';";
$resultsTableExistsCheck = mysql_query($sqlCheckTableExists) or die('died trying to see if the table exists: ' . mysql_error());
if (mysql_num_rows($resultsTableExistsCheck) > 0) {
// echo $country['short_name']." ";
$tableExists = mysql_fetch_assoc($resultsTableExistsCheck);
// echo $tableExists['COUNT(*)'];
// echo '<br />';
if ($tableExists['COUNT(*)'] > 0) {
//see if the country table exists in the international_cities db
mysql_select_db('international_cities_rrr', $dbconnectINTCities);
//get all duplicate city entries for this country
$sqlGetDuplicates = "SELECT cc1, COUNT(name), name FROM `" . $country['short_name'] . "` GROUP BY name HAVING(COUNT(name) > 1);";
$resultDuplicates = mysql_query($sqlGetDuplicates) or die('died trying to get duplicates: ' . mysql_error());
while ($city = mysql_fetch_assoc($resultDuplicates)) {
echo "Duplicate Found: " . $city['name'] . '<br />';
removeDuplicates($city['name'], $city['cc1']);
}
}
}
}
function removeDuplicates($cityName = null, $countryCode = null)
{
$duplicates = array();
//get all id's for city duplicates
$sql = "SELECT id FROM `{$countryCode}` WHERE name = '" . addslashes($cityName) . "'";
$results = mysql_query($sql) or die('died getting all this city\'s duplicate entries: ' . mysql_error());
while ($city = mysql_fetch_assoc($results)) {
$duplicates[] = $city['id'];
}
//go through and delete all the duplicate cities (leaving the last city record only)
for ($i = 0; $i < count($duplicates) - 1; $i++) {
