public function testEqualTexts()
{
$text1 = file_get_contents(__DIR__ . '/../../../resources/text/file1.txt');
$text2 = file_get_contents(__DIR__ . '/../../../resources/text/file1.txt');
$simhash = new SimHash();
$extractor = new SimpleTextExtractor();
$comparator = new GaussianComparator();
$fp1 = $simhash->hash($extractor->extract($text1));
$fp2 = $simhash->hash($extractor->extract($text2));
self::assertEquals(1, $comparator->compare($fp1, $fp2));
}
/**
* Compute the similarity hash on the given text
*
* @param string $text
* @param int $size
* @return Fingerprint
*/
public function hashText($text, $size = self::SIMHASH_64)
{
$extractor = new SimpleTextExtractor();
return $this->hash($extractor->extract($text), $size);
}
/**
* GoCompare!
* @since Version 3.10.0
* @return \Railpage\Users\SockpuppetManager
*/
public function compare()
{
/**
* Load our reference data first
*/
$ref = array("ips" => json_encode($this->ReferenceUser->getIPs(new DateTime("6 months ago"))));
/**
* Start our SimHash stuff
*/
$SimHash = new SimHash();
$Extractor = new SimpleTextExtractor();
$Comparator = new GaussianComparator(3);
foreach ($ref as $key => $lookup) {
$this->hashes[$key]['reference'] = $SimHash->hash($Extractor->extract($lookup), SimHash::SIMHASH_64);
}
foreach ($this->suspects as $Suspect) {
$suspectData = array("ips" => json_encode($Suspect->getIPs(new DateTime("6 months ago"))));
foreach ($suspectData as $key => $lookup) {
$this->hashes[$key]['suspect'] = $SimHash->hash($Extractor->extract($lookup), SimHash::SIMHASH_64);
}
foreach ($this->hashes as $key => $users) {
$this->results[$Suspect->id][$key] = $Comparator->compare($users['reference'], $users['suspect']);
}
}
}