Java tutorial
/* * Copyright 2014 Netflix, Inc. * * Licensed under the Apache License, Version 2.0 (the "License"); * you may not use this file except in compliance with the License. * You may obtain a copy of the License at * * http://www.apache.org/licenses/LICENSE-2.0 * * Unless required by applicable law or agreed to in writing, software * distributed under the License is distributed on an "AS IS" BASIS, * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. * See the License for the specific language governing permissions and * limitations under the License. */ package io.reactivex.netty.protocol.http.sse; import io.netty.channel.ChannelHandler; import io.netty.channel.ChannelHandlerContext; import io.netty.channel.ChannelPipeline; import io.netty.channel.SimpleChannelInboundHandler; import io.netty.handler.codec.http.HttpContent; import io.netty.handler.codec.http.HttpHeaders; import io.netty.handler.codec.http.HttpResponse; import io.netty.handler.codec.http.LastHttpContent; import io.reactivex.netty.protocol.http.client.ClientRequestResponseConverter; /** * A handler to insert {@link ServerSentEventDecoder} at a proper position in the pipeline according to the protocol. <br/> * There are the following cases, this handles: * * <h1>Http response with chunked encoding</h1> * In this case, the {@link ServerSentEventDecoder} is inserted after this {@link SseChannelHandler} * * <h1>Http response with no chunking</h1> * In this case, the {@link ServerSentEventDecoder} is inserted as the first handler in the pipeline. This makes the * {@link io.netty.buffer.ByteBuf} at the origin to be converted to {@link ServerSentEvent} and hence any other handler * will not look at this message unless it is really interested. * * * <h3>Caveat</h3> * In some cases where any message is buffered before this pipeline change is made by this handler (i.e. adding * {@link ServerSentEventDecoder} as the first handler), the {@link ServerSentEventDecoder} will not be applied to those * messages. For this reason we also add {@link ServerSentEventDecoder} after this handler. In cases, where the first * {@link ServerSentEventDecoder} is applied on the incoming data, the next instance of {@link ServerSentEventDecoder} * will be redundant. * */ @ChannelHandler.Sharable public class SseChannelHandler extends SimpleChannelInboundHandler<Object> { public static final String NAME = "sse-inbound-handler"; public static final String SSE_DECODER_HANDLER_NAME = "rx-sse-decoder"; public static final String SSE_DECODER_POST_INBOUND_HANDLER = "rx-sse-decoder-post-inbound"; public SseChannelHandler() { super(false); // Never auto-release, management of buffer is done via {@link ObservableAdapter} } @Override protected void channelRead0(ChannelHandlerContext ctx, Object msg) throws Exception { if (msg instanceof HttpResponse) { /** * Since SSE is an endless stream, we can never reuse a connection and hence as soon as SSE traffic is * received, the connection is marked as discardable on close. */ ctx.channel().attr(ClientRequestResponseConverter.DISCARD_CONNECTION).set(true); // SSE traffic should always discard connection on close. ChannelPipeline pipeline = ctx.channel().pipeline(); if (!HttpHeaders.isTransferEncodingChunked((HttpResponse) msg)) { pipeline.addFirst(SSE_DECODER_HANDLER_NAME, new ServerSentEventDecoder()); /* * If there are buffered messages in the previous handler at the time this message is read, we would * not be able to convert the content into an SseEvent. For this reason, we also add the decoder after * this handler, so that we can handle the buffered messages. * See the class level javadoc for more details. */ pipeline.addAfter(NAME, SSE_DECODER_POST_INBOUND_HANDLER, new ServerSentEventDecoder()); } else { pipeline.addAfter(NAME, SSE_DECODER_HANDLER_NAME, new ServerSentEventDecoder()); } ctx.fireChannelRead(msg); } else if (msg instanceof LastHttpContent) { LastHttpContent lastHttpContent = (LastHttpContent) msg; /** * The entire pipeline is set based on the assumption that LastHttpContent signals the end of the stream. * Since, here we are only passing the content to the rest of the pipeline, it becomes imperative to * also pass LastHttpContent as such. * For this reason, we send the LastHttpContent again in the pipeline. For this event sent, the content * buffer will already be read and hence will not be read again. This message serves as only containing * the trailing headers. * However, we need to increment the ref count of the content so that the assumptions down the line of the * ByteBuf always being released by the last pipeline handler will not break (as ServerSentEventDecoder releases * the ByteBuf after read). */ lastHttpContent.content().retain(); // pseudo retain so that the last handler of the pipeline can release it. if (lastHttpContent.content().isReadable()) { ctx.fireChannelRead(lastHttpContent.content()); } ctx.fireChannelRead(msg); // Since the content is already consumed above (by the SSEDecoder), this is just // as sending just trailing headers. This is critical to mark the end of stream. } else if (msg instanceof HttpContent) { ctx.fireChannelRead(((HttpContent) msg).content()); } else { ctx.fireChannelRead(msg); } } }