public static function makeRequest(HttpRequest $request)
{
$requestUri = $request->getRequestUri()->getUri();
// if the request is towards a file URL, return the response constructed
// from file
if (0 === strpos($requestUri, "file:///")) {
return HttpResponse::fromFile($requestUri);
}
$httpResponse = new HttpResponse();
$curlChannel = curl_init();
curl_setopt($curlChannel, CURLOPT_URL, $requestUri);
curl_setopt($curlChannel, CURLOPT_FOLLOWLOCATION, 1);
curl_setopt($curlChannel, CURLOPT_RETURNTRANSFER, 1);
curl_setopt($curlChannel, CURLOPT_TIMEOUT, 10);
if ($request->getRequestMethod() === "POST") {
curl_setopt($curlChannel, CURLOPT_POST, 1);
curl_setopt($curlChannel, CURLOPT_POSTFIELDS, $request->getContent());
}
$basicAuthUser = $request->getBasicAuthUser();
$basicAuthPass = $request->getBasicAuthPass();
if (NULL !== $basicAuthUser) {
$request->setHeader("Authorization", "Basic " . base64_encode($basicAuthUser . ":" . $basicAuthPass));
}
// Including HTTP headers in request
$headers = $request->getHeaders(TRUE);
if (!empty($headers)) {
curl_setopt($curlChannel, CURLOPT_HTTPHEADER, $headers);
}
// Connect to SSL/TLS server, validate certificate and host
if ($request->getRequestUri()->getScheme() === "https") {
curl_setopt($curlChannel, CURLOPT_SSL_VERIFYPEER, 1);
curl_setopt($curlChannel, CURLOPT_SSL_VERIFYHOST, 2);
}
// Callback to extract all the HTTP headers from the response...
// In order to really correctly parse HTTP headers one would have to look at RFC 2616...
curl_setopt($curlChannel, CURLOPT_HEADERFUNCTION, function ($curlChannel, $header) use($httpResponse) {
// Ignore Status-Line (RFC 2616, section 6.1)
if (0 === preg_match('|^HTTP/\\d+.\\d+ [1-5]\\d\\d|', $header)) {
// Only deal with header lines that contain a colon
if (strpos($header, ":") !== FALSE) {
// Only deal with header lines that contain a colon
list($key, $value) = explode(":", trim($header));
$httpResponse->setHeader(trim($key), trim($value));
}
}
return strlen($header);
});
$output = curl_exec($curlChannel);
if ($errorNumber = curl_errno($curlChannel)) {
throw new OutgoingHttpRequestException(curl_error($curlChannel));
}
$httpResponse->setStatusCode(curl_getinfo($curlChannel, CURLINFO_HTTP_CODE));
$httpResponse->setContent($output);
curl_close($curlChannel);
return $httpResponse;
}
/**
* @return Retrieved HTTP response and elasticsearch document
*/
public function fetch($url, $actions, $force = false)
{
$esDoc = $this->es->get($url);
if (isset($esDoc->status->location) && $esDoc->status->location != '') {
//TODO: what if location redirects change?
$url = $esDoc->status->location;
$esDoc = $this->es->get($url);
}
$types = array();
foreach ($actions as $action) {
$types = array_merge($action::$supportedTypes);
}
$types = array_unique($types);
$req = new HttpRequest($url);
$req->setHeader('accept', implode(',', $types));
if (!$force && $esDoc && isset($esDoc->status->processed) && $esDoc->status->processed != '') {
$nCrawlTime = strtotime($esDoc->status->processed);
$req->setHeader('If-Modified-Since: ' . gmdate('r', $nCrawlTime));
}
$res = $req->send();
if ($res->getStatus() === 304) {
//not modified since last time, so don't crawl again
Log::info("Not modified since last fetch");
return false;
} else {
if ($res->getStatus() !== 200) {
throw new \Exception("Response code is not 200 but " . $res->getStatus() . ", stopping");
}
}
$effUrl = Helper::removeAnchor($res->getEffectiveUrl());
if ($effUrl != $url) {
$this->storeRedirect($url, $effUrl);
$url = $effUrl;
$esDoc = $this->es->get($url);
}
//FIXME: etag, hash on content
$retrieved = new Retrieved();
$retrieved->httpRes = $res;
$retrieved->esDoc = $esDoc;
$retrieved->url = $url;
return $retrieved;
}
/**
* Sign the given request; ie. add an Authorization: Digest header
* and increase the internal nonce counter.
*
* @param peer.http.HttpRequest $request
*/
public function sign(HttpRequest $request)
{
$url = $request->target;
$params = array();
if (is_array($request->parameters)) {
$params = array_merge($params, $request->parameters);
}
if ($request->getUrl()->hasParams()) {
$params = array_merge($params, $request->getUrl()->getParams());
}
if (sizeof($params)) {
$url .= '?';
foreach ($params as $k => $v) {
$url .= $k . '=' . $v . '&';
}
$url = substr($url, 0, -1);
}
$request->setHeader('Authorization', new Header('Authorization', $this->getValueRepresentation($request->method, $url)));
// Increase internal counter
$this->counter++;
}
public function duplicateHeader()
{
$r = new HttpRequest(new URL('http://example.com/'));
$r->setHeader('X-Binford', 6100);
$r->setHeader('X-Binford', 61000);
$this->assertEquals("GET / HTTP/1.1\r\nConnection: close\r\nHost: example.com\r\nX-Binford: 61000\r\n\r\n", $r->getRequestString());
}
/**
* @param HttpRequest $req
*/
public function applyCookie($req)
{
// fetch cookies
$host = $req->getHeader('host');
$path = $req->getUrlParam('path');
$cookies = $this->fetchCookieToSend($host, $path);
if ($this !== $req) {
$cookies = array_merge($cookies, $req->fetchCookieToSend($host, $path));
}
// add to header
$req->setHeader('cookie', null);
foreach (array_chunk(array_values($cookies), 3) as $chunk) {
$req->addHeader('cookie', implode('; ', $chunk));
}
}
/**
* Sign HTTP request
*
* @param peer.http.HttpRequest $request
*/
public function sign(HttpRequest $request)
{
$request->setHeader('Authorization', new Header('Authorization', $this->getValueRepresentation()));
}
