/**
* Request the page from IMDb
* @param $url
* @return string Page html. Empty string on failure
*/
protected function requestPage($url)
{
$this->logger->info("[Page] Requesting [{$url}]");
$req = new MDB_Request($url, $this->config);
if (!$req->sendRequest()) {
$this->logger->error("[Page] Failed to connect to server when requesting url [{$url}]");
return '';
}
if (200 == $req->getStatus()) {
return $req->getResponseBody();
} elseif ($redirectUrl = $req->getRedirect()) {
$this->logger->debug("[Page] Following redirect from [{$url}] to [{$redirectUrl}]");
return $this->requestPage($redirectUrl);
} else {
$this->logger->error("[Page] Failed to retrieve url [{url}]. Response headers:{headers}", array('url' => $url, 'headers' => $req->getLastResponseHeaders()));
return '';
}
}
/** Convert IMDB redirect-URLs of external sites to real URLs
* @method convertIMDBtoRealURL
* @param string url redirect-url
* @return string url real-url
*/
protected function convertIMDBtoRealURL($url)
{
if (preg_match('/^http:\\/\\//', $url)) {
return $url;
}
$req = new MDB_Request("");
$req->setURL("http://" . $this->imdbsite . $url);
if ($req->sendRequest() !== FALSE) {
$head = $req->getLastResponseHeaders();
foreach ($head as $header) {
if (preg_match('/:/', $header)) {
list($type, $value) = explode(':', $header, 2);
if ($type == 'Location') {
return preg_replace('/\\s/', '', $value);
}
}
}
}
return false;
}
/** Obtain page from web server
* @method protected getWebPage
* @param string wt internal name of the page
* @param string url URL to open
*/
protected function getWebPage($wt, $url)
{
$req = new MDB_Request("");
$req->setURL($url);
if ($req->sendRequest() !== FALSE) {
$head = $req->getLastResponseHeaders();
} else {
$head[0] = "HTTP/1.1 000";
}
$response = explode(" ", $head[0]);
$this->lastServerResponse = $response[1];
switch (substr($head[0], 0, 12)) {
case "HTTP/1.1 000":
$this->page[$wt] = "cannot open page";
$this->debug_scalar("cannot open page (could not connect to host): {$url}");
return false;
break;
case "HTTP/1.1 404":
$this->page[$wt] = "cannot open page";
$this->debug_scalar("cannot open page (error 404): {$url}");
$this->debug_object($response);
return false;
break;
case "HTTP/1.1 301":
// permanent redirect
// permanent redirect
case "HTTP/1.1 302":
// found
// found
case "HTTP/1.1 303":
// see other
// see other
case "HTTP/1.1 307":
// temporary redirect
// in all these cases, the correct URL is to be found in the 'Location:' header
foreach ($head as $headline) {
if (strpos(trim(strtolower($headline)), 'location') !== 0) {
continue;
}
$aline = explode(': ', $headline);
$target = trim($aline[1]);
$this->getWebPage($wt, $target);
return;
}
// echo "<pre>";print_r($head);echo "</pre>\n";
// $this->debug_object($response);
// echo "<pre>";print_r($head);echo "</pre>\n";
// $this->debug_object($response);
case "HTTP/1.1 200":
break;
default:
$this->debug_scalar("HTTP response code not handled explicitly: '" . $head[0] . "'");
break;
}
$this->page[$wt] = $req->getResponseBody();
if (strpos(get_class($this), 'imdb') !== FALSE && $this->imdb_utf8recode && function_exists('mb_detect_encoding')) {
$cur_encoding = mb_detect_encoding($this->page[$wt]);
if (!($cur_encoding == "UTF-8" && mb_check_encoding($this->page[$wt], "UTF-8"))) {
$this->page[$wt] = utf8_encode($this->page[$wt]);
}
}
}
/** Obtain page from web server
* @method private getWebPage
* @param string wt internal name of the page
* @param string url URL to open
*/
function getWebPage($wt, $url)
{
$req = new MDB_Request("");
$req->setURL($url);
if ($req->sendRequest() !== FALSE) {
$head = $req->getLastResponseHeaders();
} else {
$head[0] = "HTTP/1.1 000";
}
$response = explode(" ", $head[0]);
$this->lastServerResponse = $response[1];
switch (substr($head[0], 0, 12)) {
case "HTTP/1.1 000":
$this->page[$wt] = "cannot open page";
$this->debug_scalar("cannot open page (could not connect to host): {$url}");
return false;
break;
case "HTTP/1.1 404":
$this->page[$wt] = "cannot open page";
$this->debug_scalar("cannot open page (error 404): {$url}");
return false;
break;
case "HTTP/1.1 301":
case "HTTP/1.1 302":
case "HTTP/1.1 200":
break;
default:
$this->debug_scalar("HTTP response code not handled explicitly: '" . $head[0] . "'");
break;
}
$this->page[$wt] = $req->getResponseBody();
if (strpos(get_class($this), 'imdb') !== FALSE && $this->imdb_utf8recode && function_exists('mb_detect_encoding')) {
$cur_encoding = mb_detect_encoding($this->page[$wt]);
if (!($cur_encoding == "UTF-8" && mb_check_encoding($this->page[$wt], "UTF-8"))) {
$this->page[$wt] = utf8_encode($this->page[$wt]);
}
}
}
