public function testFilter()
{
$filter = new Hash();
$this->assertEquals('0beec7b5ea3f0fdbc95d0dd47f3c5bc275da8a33', $filter->apply('foo'));
$filter = new Hash('sha256');
$this->assertEquals('2c26b46b68ffc68ff99b453c1d30413413422d706483bfa0f98a5e886266e7ae', $filter->apply('foo'));
}
/**
* Returns similar documents ids from shingles
*
* @return array
*/
protected function _getSimilarDocsIds()
{
$similarDocsIds = array();
$shingles = $this->_getShigles();
foreach ($shingles as $docId => $shingles1Doc) {
$shinglesList = Hash::apply($shingles, '{n}.{n}', 'array_count_values');
$shingles1Doc = array_flip($shingles1Doc);
$shinglesEquals = array_intersect_key($shinglesList, $shingles1Doc);
$countSimilarShingles = array_sum($shinglesEquals) - count($shingles1Doc);
$countShingles1Doc = count($shingles1Doc);
$countShingles = count($shingles) - 1;
if ($countSimilarShingles <= 0 || $countShingles1Doc <= 0 || $countShingles <= 0) {
continue;
}
$similarity = round($countSimilarShingles / $countShingles1Doc * 100 / $countShingles, 2);
if ($similarity > $this->_params['allowSimilarity']) {
$similarDocsIds[$docId] = true;
unset($shingles[$docId]);
}
}
return $similarDocsIds;
}
/**
* testApply
*
* @return void
*/
public function testApply()
{
$data = static::articleData();
$result = Hash::apply($data, '{n}.Article.id', 'array_sum');
$this->assertEquals(15, $result);
}