/**
* @since 2.5
*
* @param SQLStore $store
*
* @return SearchTable
*/
public function newTextSanitizer()
{
$settings = ApplicationFactory::getInstance()->getSettings();
$textSanitizer = new TextSanitizer(new SanitizerFactory());
$textSanitizer->setLanguageDetection($settings->get('smwgFulltextLanguageDetection'));
$textSanitizer->setMinTokenSize($settings->get('smwgFulltextSearchMinTokenSize'));
return $textSanitizer;
}
/**
* @dataProvider textOnMockProvider
*/
public function testSanitizs($text, $expected)
{
$sanitizer = $this->getMockBuilder('\\Onoi\\Tesa\\Sanitizer')->disableOriginalConstructor()->getMock();
$sanitizer->expects($this->atLeastOnce())->method('sanitizeWith')->will($this->returnValue($text));
$stopwordAnalyzer = $this->getMockBuilder('\\Onoi\\Tesa\\StopwordAnalyzer\\StopwordAnalyzer')->disableOriginalConstructor()->getMock();
$synonymizer = $this->getMockBuilder('\\Onoi\\Tesa\\Synonymizer\\Synonymizer')->disableOriginalConstructor()->getMock();
$tokenizer = $this->getMockBuilder('\\Onoi\\Tesa\\Tokenizer\\Tokenizer')->disableOriginalConstructor()->getMock();
$this->sanitizerFactory->expects($this->atLeastOnce())->method('newSanitizer')->will($this->returnValue($sanitizer));
$this->sanitizerFactory->expects($this->atLeastOnce())->method('newPreferredTokenizerByLanguage')->will($this->returnValue($tokenizer));
$this->sanitizerFactory->expects($this->atLeastOnce())->method('newStopwordAnalyzerByLanguage')->will($this->returnValue($stopwordAnalyzer));
$this->sanitizerFactory->expects($this->atLeastOnce())->method('newSynonymizerByLanguage')->will($this->returnValue($synonymizer));
$instance = new TextSanitizer($this->sanitizerFactory);
$this->assertEquals($expected, $instance->sanitize($text));
}
private function doAggregateFromFieldChangeOp($type, $fieldChangeOp, &$aggregate)
{
$searchTable = $this->searchTableUpdater->getSearchTable();
// Exempted property -> out
if (!$fieldChangeOp->has('p_id') || $searchTable->isExemptedPropertyById($fieldChangeOp->get('p_id'))) {
return;
}
// Only text components
if (!$fieldChangeOp->has('o_blob') && !$fieldChangeOp->has('o_hash') && !$fieldChangeOp->has('o_serialized')) {
return;
}
// Re-map (url type)
if ($fieldChangeOp->has('o_serialized')) {
$fieldChangeOp->set('o_blob', $fieldChangeOp->get('o_serialized'));
}
// Build a temporary stable key for the diff match
$key = $fieldChangeOp->get('s_id') . ':' . $fieldChangeOp->get('p_id');
// If the blob value is empty then the DIHandler has put any text < 72
// into the hash field
$text = $fieldChangeOp->get('o_blob');
if ($text === null || $text === '') {
$text = $fieldChangeOp->get('o_hash');
}
if (!isset($aggregate[$key])) {
$aggregate[$key] = $type === TableChangeOp::OP_DELETE ? array() : '';
}
// Concatenate the inserts but keep the deletes separate to allow
// for them to be removed individually
if ($type === TableChangeOp::OP_INSERT) {
$aggregate[$key] = trim($aggregate[$key] . ' ' . trim($text));
} elseif ($type === TableChangeOp::OP_DELETE) {
$aggregate[$key][] = $this->textSanitizer->sanitize($text);
}
}
/**
* @since 2.5
*
* @param integer $sid
* @param integer $pid
* @param string $text
*/
public function update($sid, $pid, $text)
{
if (trim($text) === '') {
return $this->delete($sid, $pid);
}
$this->connection->update($this->searchTable->getTableName(), array('o_text' => $this->textSanitizer->sanitize($text), 'o_sort' => mb_substr($text, 0, 32)), array('s_id' => (int) $sid, 'p_id' => (int) $pid), __METHOD__);
}
