Class HttpClient
Synopsis
#include <lib/inc/drogon/HttpClient.h>
class HttpClient : public trantor::NonCopyable
Description
Asynchronous http client.
HttpClient implementation object uses the HttpAppFramework's event loop by default, so you should call app().run() to make the client work. Each HttpClient object establishes a persistent connection with the server. If the connection is broken, the client attempts to reconnect when calling the sendRequest method.
Using the static mathod newHttpClient(...) to get shared_ptr of the object implementing the class, the shared_ptr is retained in the framework until all response callbacks are invoked without fear of accidental deconstruction.
TODO:SSL server verification
Inheritance
Ancestors: NonCopyable
Methods
| HttpClient | ||
| ~HttpClient | ||
| addCookie overload | Add a cookie to the client. | |
| bytesReceived | ||
| bytesSent | Get the number of bytes sent or received. | |
| enableCookies | Enable cookies for the client. | |
| getLoop | Get the event loop of the client;. | |
| newHttpClient overload | Creaet a new HTTP client which use ip and port to connect to server. | |
| newHttpClient overload | Create a Http client using the hostString to connect to server. | |
| sendRequest overload | Send a request asynchronously to the server. | |
| sendRequest overload | Send a request synchronously to the server and return the response. | |
| setPipeliningDepth | Set the pipelining depth, which is the number of requests that are not responding |
Source
Lines 71-268 in lib/inc/drogon/HttpClient.h.
class HttpClient : public trantor::NonCopyable
{
public:
/**
* @brief Send a request asynchronously to the server
*
* @param req The request sent to the server.
* @param callback The callback is called when the response is received from
* the server.
* @param timeout In seconds. If the response is not received within the
* timeout, the callback is called with `ReqResult::Timeout` and an empty
* response. The zero value by default disables the timeout.
*
* @note
* The request object is altered(some headers are added to it) before it is
* sent, so calling this method with a same request object in different
* thread is dangerous.
* Please be careful when using timeout on an non-idempotent request.
*/
virtual void sendRequest(const HttpRequestPtr &req,
const HttpReqCallback &callback,
double timeout = 0) = 0;
/**
* @brief Send a request asynchronously to the server
*
* @param req The request sent to the server.
* @param callback The callback is called when the response is received from
* the server.
* @param timeout In seconds. If the response is not received within
* the timeout, the callback is called with `ReqResult::Timeout` and an
* empty response. The zero value by default disables the timeout.
*
* @note
* The request object is altered(some headers are added to it) before it is
* sent, so calling this method with a same request object in different
* thread is dangerous.
* Please be careful when using timeout on an non-idempotent request.
*/
virtual void sendRequest(const HttpRequestPtr &req,
HttpReqCallback &&callback,
double timeout = 0) = 0;
/**
* @brief Send a request synchronously to the server and return the
* response.
*
* @param req
* @param timeout In seconds. If the response is not received within the
* timeout, the `ReqResult::Timeout` and an empty response is returned. The
* zero value by default disables the timeout.
*
* @return std::pair<ReqResult, HttpResponsePtr>
* @note Never call this function in the event loop thread of the
* client (partially in the callback function of the asynchronous
* sendRequest method), otherwise the thread will be blocked forever.
* Please be careful when using timeout on an non-idempotent request.
*/
std::pair<ReqResult, HttpResponsePtr> sendRequest(const HttpRequestPtr &req,
double timeout = 0)
{
std::promise<std::pair<ReqResult, HttpResponsePtr>> prom;
auto f = prom.get_future();
sendRequest(
req,
[&prom](ReqResult r, const HttpResponsePtr &resp) {
prom.set_value({r, resp});
},
timeout);
return f.get();
}
#ifdef __cpp_impl_coroutine
/**
* @brief Send a request via coroutines to the server and return the
* response.
*
* @param req
* @param timeout In seconds. If the response is not received within the
* timeout, the `ReqResult::Timeout` and an empty response is returned. The
* zero value by default disables the timeout.
*
* @return task<HttpResponsePtr>
*/
Task<HttpResponsePtr> sendRequestCoro(HttpRequestPtr req,
double timeout = 0)
{
co_return co_await internal::HttpRespAwaiter(this,
std::move(req),
timeout);
}
#endif
/// Set the pipelining depth, which is the number of requests that are not
/// responding.
/**
* If this method is not called, the default depth value is 0 which means
* the pipelining is disabled. For details about pipelining, see
* rfc2616-8.1.2.2
*/
virtual void setPipeliningDepth(size_t depth) = 0;
/// Enable cookies for the client
/**
* @param flag if the parameter is true, all requests sent by the client
* carry the cookies set by the server side. Cookies are disabled by
* default.
*/
virtual void enableCookies(bool flag = true) = 0;
/// Add a cookie to the client
/**
* @note
* These methods are independent of the enableCookies() method. Whether the
* enableCookies() is called with true or false, the cookies added by these
* methods will be sent to the server.
*/
virtual void addCookie(const std::string &key,
const std::string &value) = 0;
/// Add a cookie to the client
/**
* @note
* These methods are independent of the enableCookies() method. Whether the
* enableCookies() is called with true or false, the cookies added by these
* methods will be sent to the server.
*/
virtual void addCookie(const Cookie &cookie) = 0;
/**
* @brief Creaet a new HTTP client which use ip and port to connect to
* server
*
* @param ip The ip address of the HTTP server
* @param port The port of the HTTP server
* @param useSSL if the parameter is set to true, the client connects to the
* server using HTTPS.
* @param loop If the loop parameter is set to nullptr, the client uses the
* HttpAppFramework's event loop, otherwise it runs in the loop identified
* by the parameter.
* @param useOldTLS If the parameter is set to true, the TLS1.0/1.1 are
* eanbled for HTTPS.
* @return HttpClientPtr The smart pointer to the new client object.
* @note: The ip parameter support for both ipv4 and ipv6 address
*/
static HttpClientPtr newHttpClient(const std::string &ip,
uint16_t port,
bool useSSL = false,
trantor::EventLoop *loop = nullptr,
bool useOldTLS = false,
bool validateCert = true);
/// Get the event loop of the client;
virtual trantor::EventLoop *getLoop() = 0;
/// Get the number of bytes sent or received
virtual size_t bytesSent() const = 0;
virtual size_t bytesReceived() const = 0;
/// Create a Http client using the hostString to connect to server
/**
*
* @param hostString this parameter must be prefixed by 'http://' or
* 'https://'.
*
* Examples for hostString:
* @code
https://www.baidu.com
http://www.baidu.com
https://127.0.0.1:8080/
http://127.0.0.1
http://[::1]:8080/ //IPv6 address must be enclosed in [], rfc2732
@endcode
*
* @param loop If the loop parameter is set to nullptr, the client uses the
* HttpAppFramework's event loop, otherwise it runs in the loop identified
* by the parameter.
*
* @param useOldTLS If the parameter is set to true, the TLS1.0/1.1 are
* enabled for HTTPS.
* @note
* Don't add path and parameters in hostString, the request path and
* parameters should be set in HttpRequestPtr when calling the sendRequest()
* method.
*
*/
static HttpClientPtr newHttpClient(const std::string &hostString,
trantor::EventLoop *loop = nullptr,
bool useOldTLS = false,
bool validateCert = true);
virtual ~HttpClient()
{
}
protected:
HttpClient() = default;
};