/**
* Get the parser.
*
* @return ParserInterface
*/
public function getParser()
{
$ruleLoader = new RuleLoader($this->config);
$rules = $ruleLoader->getRules($this->url);
if (!empty($rules['grabber'])) {
Logger::setMessage(get_called_class() . ': Parse content with rules');
foreach ($rules['grabber'] as $pattern => $rule) {
$url = new Url($this->url);
$sub_url = $url->getFullPath();
if (preg_match($pattern, $sub_url)) {
Logger::setMessage(get_called_class() . ': Matched url ' . $sub_url);
return new RuleParser($this->html, $rule);
}
}
} elseif ($this->enableCandidateParser) {
Logger::setMessage(get_called_class() . ': Parse content with candidates');
return new CandidateParser($this->html);
}
return;
}
/**
* Called after XML parsing.
*
* @param string $content the content that should be filtered
*/
public function filterRules($content)
{
// the constructor should require a config, then this if can be removed
if ($this->config === null) {
$config = new Config();
} else {
$config = $this->config;
}
$loader = new RuleLoader($config);
$rules = $loader->getRules($this->website);
$url = new Url($this->website);
$sub_url = $url->getFullPath();
if (isset($rules['filter'])) {
foreach ($rules['filter'] as $pattern => $rule) {
if (preg_match($pattern, $sub_url)) {
foreach ($rule as $search => $replace) {
$content = preg_replace($search, $replace, $content);
}
}
}
}
return $content;
}
