public function handle(Spizer_Document $document)
{
//get! encoding_to_utf-8
$body = Diggin_Http_Response_Encoding::encode($document->getBody(), $document->getHeader('content-type'));
if (preg_match($this->_config['match'], $body, $m, PREG_OFFSET_CAPTURE)) {
$this->engine->log('RegexMatch', array('message' => 'Document body matched lookup expression', 'regex' => $this->_config['match'], 'match' => $m[0][0], 'offset' => $m[0][1]));
}
}
public function handle(Spizer_Document $document)
{
//check document is image
$content_type = $document->getHeader('content-type');
if (!preg_match('#image/.*#i', $content_type)) {
return;
}
$filepath = $this->_config['save_dir'] . DIRECTORY_SEPARATOR . rawurlencode($document->getUrl());
file_put_contents($filepath, $document->getBody(), FILE_BINARY);
$this->addHaveFiles($document->getUrl());
}
/**
* Check if the handler actually needs to be called (according to it's
* content type and status code), and if so call ::handle()
*
* @param Spizer_Document $document
*/
public function call(Spizer_Document $document)
{
$status = $document->getStatus();
$type = $document->getHeader('content-type');
$call = true;
if ($this->_config['status']) {
if (is_array($this->_config['status'])) {
if (!in_array($status, $this->_config['status'])) {
$call = false;
}
} elseif ($this->_config['status'] != $status) {
$call = false;
}
}
if ($this->_config['content-type']) {
if (is_array($this->_config['content-type'])) {
if (!in_array($type, $this->_config['content-type'])) {
$call = false;
}
} elseif ($this->_config['content-type'] != $type) {
$call = false;
}
}
if ($call) {
$this->handle($document);
}
}
