/**
* @return \Sengin\Extraction\Result
* @throws Exception\Exception
*/
public function extract()
{
$data = $this->_dataSource->getData();
$document = $this->getDocument();
/*
* If disable libxml errors is set to true then we see no more errors like that:
* Warning: DOMDocument::loadHTML(): htmlParseEntityRef: expecting ';' in Entity
*/
$previos = libxml_use_internal_errors($this->getDisableLibXmlErrors());
$isLoaded = $document->loadHTML($data);
libxml_use_internal_errors($previos);
if (!$isLoaded) {
$message = "Can't load html data from given source";
throw new Exception\Exception($message);
}
$result = new Result();
$this->extractSearchResults($result);
$this->extractSponsoredResults($result);
$this->extractSuggstions($result);
return $result;
}
public function getData()
{
if (!$this->_options->isCacheEnabled() || !$this->_dataSource instanceof Cachable) {
return $this->_dataSource->getData();
}
$cacheDir = $this->_options->getCacheDir();
$cacheKey = $this->_dataSource->getCacheKey();
$cacheFile = $cacheDir . '/' . $cacheKey . '.sengincache';
if (is_file($cacheFile)) {
$expirationTime = $this->_options->getExpirationTime();
$lastModificationTime = time() - filemtime($cacheFile);
if ($lastModificationTime <= $expirationTime) {
// from cache
return file_get_contents($cacheFile);
}
// cache expired, remove cache data
unlink($cacheFile);
}
$data = $this->_dataSource->getData();
// save cache data
file_put_contents($cacheFile, $data);
return $data;
}