netty里面FrameDecoder的理解

Question

[color=blue][size=medium]在netty里面，messageReceived的实现如下：
[/size][/color][code="java"]
Object m = e.getMessage();
if (!(m instanceof ChannelBuffer)) {
ctx.sendUpstream(e);
return;
}

    ChannelBuffer input = (ChannelBuffer) m;
    if (!input.readable()) {
        return;
    }

    ChannelBuffer cumulation = cumulation(ctx);
    if (cumulation.readable()) {
        cumulation.discardReadBytes();
        cumulation.writeBytes(input);
        callDecode(ctx, e.getChannel(), cumulation, e.getRemoteAddress());
    } else {
        callDecode(ctx, e.getChannel(), input, e.getRemoteAddress());
        if (input.readable()) {
            cumulation.writeBytes(input);
        }
    }

[/code]

[color=blue][size=medium]我们看到比较关键的是cumulation函数，函数如下：
[/size][/color][code="java"]
private ChannelBuffer cumulation(ChannelHandlerContext ctx) {
ChannelBuffer c = cumulation;
if (c == null) {
c = ChannelBuffers.dynamicBuffer(
ctx.getChannel().getConfig().getBufferFactory());
cumulation = c;
}
return c;
}
[/code]
[color=blue][size=medium]在这个里面为什么要这么实现呢，什么时候cumulation不为空呢？
我们再看看上面的实现：[/size][/color]
[code="java"]
if (cumulation.readable()) {
cumulation.discardReadBytes();
cumulation.writeBytes(input);
callDecode(ctx, e.getChannel(), cumulation, e.getRemoteAddress());
} else {
callDecode(ctx, e.getChannel(), input, e.getRemoteAddress());
if (input.readable()) {
cumulation.writeBytes(input);
}
}
[/code]
[color=blue][size=medium]这个if和else里面的逻辑应该怎么理解。
[/size][/color]

Answer 1

1、什么时候cumulation不为空呢？
初次运行，cumulation == null，所以ChannelBuffer cumulation = cumulation(ctx);创建了一个空的dynamicBuffer；

cumulation.readable()是false，所以执行的是：
else {

callDecode(ctx, e.getChannel(), input, e.getRemoteAddress());

if (input.readable()) {

cumulation.writeBytes(input);

}

}

callDecode过程中，可能只decode了input的一部分，剩下的字节就写到cumulation，等到下次messageReceived之后，就看到cumulation不为空。（可能messageReceived一次接受了1.5个frame，callDecode只decode了一个frame，剩下的一半写入cumulation）

2、
接着上面，这时候cumulation.readable()是true，所以执行的是
if (cumulation.readable()) {

cumulation.discardReadBytes();

cumulation.writeBytes(input);

callDecode(ctx, e.getChannel(), cumulation, e.getRemoteAddress());
}

这里将新接收到的input写入cumulation，然后继续decode。（又接收到了另一半，与先前接收到的合并到一起组成一个完整的frame）

3、源代码文档上说的很清楚了：
在像 TCP/IP这种基于流的传输过程中，即使在局域网环境中，报文也可能被分割和重组。FrameDecoder帮助我们将接受到的报文重新整理成有意义的、容易被应用程序逻辑理解的frames。

In a stream-based transport such as TCP/IP, packets can be fragmented and

• reassembled during transmission even in a LAN environment. For example,
• let us assume you have received three packets:
• +-----+-----+-----+
• | ABC | DEF | GHI |
• +-----+-----+-----+
• because of the packet fragmentation, a server can receive them like the
• following:
• +----+-------+---+---+
• | AB | CDEFG | H | I |
• +----+-------+---+---+ {@link FrameDecoder} helps you defrag the received packets into one or more
• meaningful frames that could be easily understood by the
• application logic. In case of the example above, your {@link FrameDecoder}
• implementation could defrag the received packets like the following:
• +-----+-----+-----+
• | ABC | DEF | GHI |
• +-----+-----+-----+