|
WebSocket++ 0.8.3-dev
C++ websocket client/server library
|
Handlers allow WebSocket++ programs to receive notifications about events that happen in relation to their connections. Some handlers also behave as hooks that give the program a chance to modify state or adjust settings before the connection continues.
Handlers are registered by calling the appropriate set_*_handler method on either an endpoint or connection. The * refers to the name of the handler (as specified in the signature field below). For example, to set the open handler, call set_open_handler(...).
Setting handlers on an endpoint will result in them being copied as the default handler to all new connections created by that endpoint. Changing an endpoint's handlers will not affect connections that are already in progress. This includes connections that are in the listening state. As such, it is important to set any endpoint handlers before you call endpoint::start_accept or else the handlers will not be attached to your first connection.
Setting handlers on a connection will result in the handler being changed for that connection only, starting at the next time that handler is called. This can be used to change the handler during a connection.
These handlers will be called at most once per connection in the order specified below.
| Event | Signature | Availability |
|---|---|---|
| Socket initialization | socket_init(connection_hdl, asio::ip::tcp::socket&) | 0.3.0 Asio Transport |
This hook is triggered after the socket has been initialized but before a connection is established. It allows setting arbitrary socket options before connections are sent/recieved.
| Event | Signature | Availability |
|---|---|---|
| TCP established, no data sent | tcp_pre_init(connection_hdl) | 0.3.0 Asio Transport |
This hook is triggered after the TCP connection is established, but before any pre-WebSocket-handshake operations have been run. Common pre-handshake operations include TLS handshakes and proxy connections.
| Event | Signature | Availability |
|---|---|---|
| TCP established, pre-handshake operations complete | tcp_post_init(connection_hdl) | 0.3.0 Asio Transport |
This hook is triggered after all pre-WebSocket handshake operations have been run. Common pre-handshake operations include TLS handshakes and proxy connections.
| Event | Signature | Availability |
|---|---|---|
| Request for TLS context | tls_context_ptr tls_init(connection_hdl) | 0.3.0 Asio Transport with TLS |
This hook is triggered before the TLS handshake to request the TLS context to use. The user program must return a pointer to a configured TLS context to continue. This provides the opportuinity to set up TLS settings, certificates, etc.
| Event | Signature | Availability |
|---|---|---|
| Hook to accept or reject a connection | bool validate(connection_hdl) | 0.3.0 Core, Server role only |
This hook is triggered for servers during the opening handshake after the request has been processed but before the response has been sent. It gives a program the opportunity to inspect headers and other connection details and either accept or reject the connection. Validate happens before the open or fail handler.
Return true to accept the connection, false to reject. If no validate handler is registered, all connections will be accepted.
| Event | Signature | Availability |
|---|---|---|
| Successful new connection | open(connection_hdl) | 0.3.0 Core |
Either open or fail will be called for each connection. Never both. All connections that begin with an open handler call will also have a matching close handler call when the connection ends.
| Event | Signature | Availability |
|---|---|---|
| Connection failed (before opening) | fail(connection_hdl) | 0.3.0 Core |
Either open or fail will be called for each connection. Never both. Connections that fail will never have a close handler called.
| Event | Signature | Availability |
|---|---|---|
| Connection closed (after opening) | close(connection_hdl) | 0.3.0 Core |
Close will be called exactly once for every connection that open was called for. Close is not called for failed connections.
| Event | Signature | Availability |
|---|---|---|
| Listen Prebind | lib::error_code tcp_pre_bind(lib::shared_ptr<lib::asio::ip::tcp::acceptor>) | 0.8.0 Asio Transport |
This hook is triggered during the call to endpoint::listen after the acceptor is initialized and open but before bind or listen are called. It provides an opportunity to make application specific configuation to the acceptor, most commonly setting socket options on the listening socket, such as SO_REUSEPORT or IPV6_ONLY.
The return value of the callback will be used to determine whether to proceed with listening. Return an empty/0 error code to proceed, or another error code to fail. In the fail case, the error code will be reported back to the caller of endpoint::listen.
| Event | Signature | Availability |
|---|---|---|
| Async Accept Loop end (after opening) | accept_loop(void(lib::error_code const &, lib::error_code const &)) | 0.9.0 Core, Server role only |
This handler is passed to the endpoint::start_accept() function, which initiates an asyncronous loop that accepts new connections, and is invoked when that loop ends. The handler will be called exactly once for each call of start_accept no matter when or how it exits (it might exit immediately).
The handler may be called inside the call to start_accept if that function has enough information to determine that accepting connections will be impossible. Otherwise, it may be invoked at a later time from a different asyncronous handler.
The handler returns two error codes. The first from the core library and the second a more detailed code passed through from the underlying transport.
These handers are called in response to incoming messages or message like events. They only will be called while the connection is in the open state.
| Event | Signature | Availability |
|---|---|---|
| Data message recieved | message(connection_hdl, message_ptr) | 0.3.0 Core |
Applies to all non-control messages, including both text and binary opcodes. The message_ptr type and its API depends on your endpoint type and its config.
| Event | Signature | Availability |
|---|---|---|
| Ping recieved | bool ping(connection_hdl, std::string) | 0.3.0 Core |
Second (string) argument is the binary ping payload. Handler return value indicates whether or not to respond to the ping with a pong. If no ping handler is set, WebSocket++ will respond with a pong containing the same binary data as the ping (Per requirements in RFC6455).
| Event | Signature | Availability |
|---|---|---|
| Pong recieved | pong(connection_hdl, std::string) | 0.3.0 Core |
Second (string) argument is the binary pong payload.
| Event | Signature | Availability |
|---|---|---|
| Timed out while waiting for a pong | pong_timeout(connection_hdl, std::string) | 0.3.0 Core, transport with timer support |
Triggered if there is no response to a ping after the configured duration. The second (string) argument is the binary payload of the unanswered ping.
| Event | Signature | Availability |
|---|---|---|
| HTTP request recieved | http(connection_hdl | 0.3.0 Core, Server role only |
Called when HTTP requests that are not WebSocket handshake upgrade requests are received. Allows responding to regular HTTP requests. If no handler is registered a 426/Upgrade Required error is returned.
| Event | Signature | Availability |
|---|---|---|
| Connection was manually interrupted | interrupt(connection_hdl) | 0.3.0 Core |
Interrupt events can be triggered by calling endpoint::interrupt or connection::interrupt. Interrupt is similar to a timer event with duration zero but with lower overhead. It is useful for single threaded programs to allow breaking up a very long handler into multiple parts and for multi threaded programs as a way for worker threads to signal to the main/network thread that an event is ready.