websocket protocol support #73
Conversation
|
Copyright header is missing in all the files. |
|
You might want to add yourself to |
|
|
||
| public void handshake(Request request) throws IOException { | ||
| final String version = request.getHeader(WebSocketHeaders.VERSION); | ||
| if (!"13".equals(version)) { |
| server.handleMessage(session, (BinaryMessage) message); | ||
| } else if (message instanceof CloseMessage) { | ||
| server.handleMessage(session, (CloseMessage) message); | ||
| } |
There was a problem hiding this comment.
Maybe IllegalArgumentException in case of unexpected message?
| @Override | ||
| public void close() { | ||
| try { | ||
| extensions.forEach(Extension::close); |
There was a problem hiding this comment.
Is it possible to get some exception from Extension.close and stay with some unclosed ones?
| return response; | ||
| } catch (VersionException e) { | ||
| Response response = new Response("426 Upgrade Required", Response.EMPTY); | ||
| response.addHeader(WebSocketHeaders.createVersionHeader(13)); |
There was a problem hiding this comment.
Magic constant once again.
| /** | ||
| * @author <a href="mailto:vadim.yelisseyev@gmail.com">Vadim Yelisseyev</a> | ||
| */ | ||
| public class HandshakeException extends RuntimeException { |
There was a problem hiding this comment.
Is it possible to extend WebSocketException here as siblings do?
| } | ||
|
|
||
| public Map<String, String> getParameters() { | ||
| return parameters; |
There was a problem hiding this comment.
Maybe wrap to unmodifiableMap?
| */ | ||
| public class ExtensionRequestParser { | ||
|
|
||
| public static List<ExtensionRequest> parse(String header) { |
There was a problem hiding this comment.
It seems unit test for this parsing would be valuable.
| if (serverContextTakeover) { | ||
| serverContextTakeover = false; | ||
| } else { | ||
| throw new HandshakeException("Duplicate definition of the server_no_context_takeover extension parameter"); |
There was a problem hiding this comment.
Is it really a HandshakeException?
|
|
||
| @Override | ||
| public String toString() { | ||
| return getClass().getSimpleName() + "<" + Hex.toHex(payload) + ">"; |
There was a problem hiding this comment.
How long this payload can be? Is it useful to create such a string for a long payload?
| public static final short TLS_HANDSHAKE_FAILURE = 1015; | ||
|
|
||
| public CloseMessage(byte[] payload) { | ||
| this(payload.length == 0 ? null : Short.reverseBytes(unsafe.getShort(payload, byteArrayOffset))); |
There was a problem hiding this comment.
It's quite hard to perceive unsafe due to static import.
JavaInternals.unsafe.getShort seems to be more understandable.
|
|
||
| @Override | ||
| public String toString() { | ||
| return "CloseMessage<" + payload + ">"; |
There was a problem hiding this comment.
Oh, sorry, it's short here.
| response.addHeader("Connection: Upgrade"); | ||
| response.addHeader(WebSocketHeaders.createAcceptHeader(request)); | ||
| return response; | ||
| } catch (VersionException e) { |
| } | ||
|
|
||
| protected void handleMessage(WebSocketSession session, Message<?> message) throws IOException { | ||
| if (message instanceof PingMessage) { |
There was a problem hiding this comment.
Is it possible to use message.opcode to replace numerous instanceofs with faster switch on enum?
|
|
||
| @Override | ||
| public void handleException(Throwable e) { | ||
| log.error(e); |
There was a problem hiding this comment.
What do you think about moving logging inside if? super.handleException() deals with logging already.
|
|
||
| @Override | ||
| public String toString() { | ||
| return getClass().getSimpleName() + "<" + Hex.toHex(payload) + ">"; |
There was a problem hiding this comment.
SimpleName.of() is generally faster and more predictable.
| private final boolean fin; | ||
| private final Opcode opcode; | ||
| private int rsv; | ||
| private int payloadLength; |
There was a problem hiding this comment.
Do we really need this field if we have payload.length()?
| private static final String SERVER_NO_CONTEXT_TAKEOVER = "server_no_context_takeover"; | ||
| private static final String CLIENT_NO_CONTEXT_TAKEOVER = "client_no_context_takeover"; | ||
|
|
||
| private static final int RSV_BITMASK = 0b100; |
| private static final String CLIENT_NO_CONTEXT_TAKEOVER = "client_no_context_takeover"; | ||
|
|
||
| private static final int RSV_BITMASK = 0b100; | ||
| private static final byte[] EOM_BYTES = new byte[] {0, 0, -1, -1}; |
| break; | ||
| } | ||
| } else { | ||
| if (fin && !clientContextTakeover) { |
There was a problem hiding this comment.
Unit tests especially for context takeover and framing would be appreciated.
|
|
||
| int compressedLength; | ||
| do { | ||
| compressedLength = deflater.deflate(outputBuffer, 0, outputBuffer.length, Deflater.SYNC_FLUSH); |
There was a problem hiding this comment.
JavaDoc says:
In the case of FULL_FLUSH or SYNC_FLUSH, if the return value is len, the space available in output buffer b, this method should be invoked again with the same flush parameter and more output space.
My concern is about a case when compressed payload doesn't fit outputBuffer.
| if (protocols != null) { | ||
| for (String protocol : protocols.split(",")) { | ||
| if (config.isSupportedProtocol(protocol)) { | ||
| response.addHeader(WebSocketHeaders.PROTOCOL + protocol); |
There was a problem hiding this comment.
PROTOCOL header value from request is transformed into multiple headers in response -- is that the desired behavior?
WebSocket protocol support