/**
* Called when the crawler has crawled the given url.
*
* @param \SimZal\Crawler\Url $url
* @param \Psr\Http\Message\ResponseInterface|null $response
*/
public function hasBeenCrawled(Url $url, $response)
{
$webpage = Webpage::where('url', (string) $url)->first();
if ($webpage) {
$webpage->crawlcount++;
$webpage->save();
}
SaveSnapshot::create($url, $response);
// $html = (string)$response->getBody( true );
//$response->getStatusCode()
// https://guzzle3.readthedocs.org/http-client/response.html
}
public static function create(Url $url, $response)
{
$webpage = Webpage::where('Url', (string) $url)->first();
if (!$webpage) {
return;
}
// dd( $webpage );
$statusCode = $response ? $response->getStatusCode() : self::UNRESPONSIVE_HOST;
$html = '';
if (self::isHtml($response)) {
$html = $response->getBody()->getContents();
$html = FilterContent::make($html);
}
Snapshot::create(['html' => $html, 'status_code' => $statusCode, 'webpage_id' => $webpage->id, 'hash' => md5($html)]);
}
/**
* Display a listing of the resource.
*
* @return \Illuminate\Http\Response
*/
public function index(Request $request)
{
$webpages = Webpage::where('site_id', $request->input('site_id'))->get();
return view('webpage.index', compact('webpages'));
}
