/**
* {@inheritdoc}
*/
public function parse()
{
if (!$this->dom) {
return $this->meta;
}
foreach ($this->meta as $key => $value) {
$mutator = 'get' . studly_case($key);
if (method_exists($this, $mutator)) {
// We will use a mutator to process an attribute value if
// we found one, otherwise we'll do nothing.
$this->meta[$key] = $this->{$mutator}();
}
}
if ($this->config->get('strip_tags') && $this->meta['description']) {
$this->meta['description'] = strip_tags($this->meta['description']);
}
return $this->meta;
}
/**
* Parse an url address, and fill up the basic metadata.
*
* @param string $url
* @param \Buzz\Client\ClientInterface $client
* @return \Yoozi\Miner\Extractor
*/
public function fromUrl($url, HttpClientInterface $client = null)
{
$browser = new Browser($client);
$response = $browser->get($url, $this->config->get('headers'));
$request = $browser->getLastRequest();
$this->metadata = array();
foreach (array('url', 'host', 'domain', 'favicon') as $key) {
$this->metadata[$key] = $this->{'get' . studly_case($key)}($request);
}
if ($response->isSuccessful()) {
$this->document = $response->getContent();
$this->charset = $response->getHeaderAttribute('Content-Type', 'charset');
}
return $this;
}
public function testBasicMethods()
{
$items = array('parser' => 'hybrid', 'hybrid' => array('primary' => 'readability', 'secendary' => 'meta'), 'headers' => array('User-Agent' => 'Mozilla/5.0'), 'strip_tags' => true);
$config = new Config($items);
$this->assertEquals($config->toArray(), $items);
$this->assertEquals($config->toJson(), json_encode($items));
$this->assertEquals($config->get('parser'), $items['parser']);
$config->set('parser', 'Readability');
$this->assertEquals($config->get('parser'), 'Readability');
}
