Netty LengthFieldBasedFrameDecoder

先看看LengthFieldBasedFrameDecoder的官方API
http://docs.jboss.org/netty/3.1/api/org/jboss/netty/handler/codec/frame/LengthFieldBasedFrameDecoder.html

API举例说明了LengthFieldBasedFrameDecoder的解析机制，如下：

实际消息内容是“HELLO, WORLD”，长度是12 bytes（注意逗号后面有一个空格）

实例1
lengthFieldLength = 2表示“Length”的长度，而“Length”的值
就是“Actual Content”的长度（0x000C，也就是12）：

Java代码

lengthFieldOffset = 0
lengthFieldLength = 2
lengthAdjustment = 0
initialBytesToStrip = 0 (= do not strip header)
BEFORE DECODE (14 bytes) AFTER DECODE (14 bytes)
+--------+----------------+ +--------+----------------+
| Length | Actual Content |----->| Length | Actual Content |
| 0x000C | "HELLO, WORLD" | | 0x000C | "HELLO, WORLD" |
+--------+----------------+ +--------+----------------+

实例2
initialBytesToStrip = 2 表示在decode时，要去掉多少个字节
在这个例子，表示要去掉“Length”（2个字节）
可以看到，AFTER DECODE后，“Length”没有了，只剩下“Actual Content”：

Java代码

lengthFieldOffset = 0
lengthFieldLength = 2
lengthAdjustment = 0
initialBytesToStrip = 2 (= the length of the Length field)
BEFORE DECODE (14 bytes) AFTER DECODE (12 bytes)
+--------+----------------+ +----------------+
| Length | Actual Content |----->| Actual Content |
| 0x000C | "HELLO, WORLD" | | "HELLO, WORLD" |
+--------+----------------+ +----------------+

实例3
与实例1不同，这里“Length”的值不是“Actual Content”的长度，而是
整个消息的长度（0x000E，14 = 2 + 12）。用lengthAdjustment=-2来表示
“Actual Content”的长度要减2：
wholeLength = valueOf(Length) = 14
actualContentLength = wholeLength + lengthAdjustment = 14 + (-2)=12

Java代码

lengthFieldOffset = 0
lengthFieldLength = 2
lengthAdjustment = -2 (= the length of the Length field)
initialBytesToStrip = 0
BEFORE DECODE (14 bytes) AFTER DECODE (14 bytes)
+--------+----------------+ +--------+----------------+
| Length | Actual Content |----->| Length | Actual Content |
| 0x000E | "HELLO, WORLD" | | 0x000E | "HELLO, WORLD" |
+--------+----------------+ +--------+----------------+

实例4
这个例子多了一个“Header 1”（oxCAFE，长度为2 ）
用 lengthFieldOffset = 2表示“Length”从第3个字节开始
“Length”的值仍然是“Actual Content ”的长度，不过“Length”自身的
长度是3 （值是0x00000C，而不是上面的0x000C）

Java代码

lengthFieldOffset = 2 (= the length of Header 1)
lengthFieldLength = 3
lengthAdjustment = 0
initialBytesToStrip = 0
BEFORE DECODE (17 bytes) AFTER DECODE (17 bytes)
+----------+----------+----------------+ +----------+----------+----------------+
| Header 1 | Length | Actual Content |----->| Header 1 | Length | Actual Content |
| 0xCAFE | 0x00000C | "HELLO, WORLD" | | 0xCAFE | 0x00000C | "HELLO, WORLD" |
+----------+----------+----------------+ +----------+----------+----------------+

实例5
与实例4不同的地方是，“Length”和“Header 1”的位置调换了
wholeLength = valueOf(Length) = 14
actualContentLength = wholeLength + lengthAdjustment = 14 + 2=16
因此在decode时，会认为“Header 1”也是“Actual Content”的一部分

Java代码

lengthFieldOffset = 0
lengthFieldLength = 3
lengthAdjustment = 2 (= the length of Header 1)
initialBytesToStrip = 0
BEFORE DECODE (17 bytes) AFTER DECODE (17 bytes)
+----------+----------+----------------+ +----------+----------+----------------+
| Length | Header 1 | Actual Content |----->| Length | Header 1 | Actual Content |
| 0x00000C | 0xCAFE | "HELLO, WORLD" | | 0x00000C | 0xCAFE | "HELLO, WORLD" |
+----------+----------+----------------+ +----------+----------+----------------+

实例6
看起来要比之前的例子复杂一些，但其实是上面例子的组合
lengthFieldOffset = 1表示“Length”从第2个字节开始
lengthFieldLength = 2表示“Length”的长度是2
lengthAdjustment = 1表示“HDR2”的长度是1：
wholeLength = valueOf(Length) = 0x000C = 12
actualContentLength = wholeLength + lengthAdjustment = 12 + 1=13
decode时，会认为“Length”后面的13个字节都是“Actual Content”，
因此会认为“HDR2”也是“Actual Content”的一部分
initialBytesToStrip = 3 表示decode时，去掉3个字节

Java代码

lengthFieldOffset = 1 (= the length of HDR1)
lengthFieldLength = 2
lengthAdjustment = 1 (= the length of HDR2)
initialBytesToStrip = 3 (= the length of HDR1 + LEN)
BEFORE DECODE (16 bytes) AFTER DECODE (13 bytes)
+------+--------+------+----------------+ +------+----------------+
| HDR1 | Length | HDR2 | Actual Content |----->| HDR2 | Actual Content |
| 0xCA | 0x000C | 0xFE | "HELLO, WORLD" | | 0xFE | "HELLO, WORLD" |
+------+--------+------+----------------+ +------+----------------+

实例7
与实例6不同的是，lengthAdjustment = -3，是负数
因此，
wholeLength = valueOf(Length) = 0x0010 = 16
actualContentLength = wholeLength + lengthAdjustment = 16 + (-3)=13
decode时，会认为“Length”后面的13个字节都是“Actual Content”，
最终效果与实例6一样

Java代码

lengthFieldOffset = 1
lengthFieldLength = 2
lengthAdjustment = -3 (= the length of HDR1 + LEN, negative)
initialBytesToStrip = 3
BEFORE DECODE (16 bytes) AFTER DECODE (13 bytes)
+------+--------+------+----------------+ +------+----------------+
| HDR1 | Length | HDR2 | Actual Content |----->| HDR2 | Actual Content |
| 0xCA | 0x0010 | 0xFE | "HELLO, WORLD" | | 0xFE | "HELLO, WORLD" |
+------+--------+------+----------------+ +------+----------------+

API看完，我们来看看源码（只保留关键代码）：

Java代码

public class LengthFieldBasedFrameDecoder extends FrameDecoder {
private final int maxFrameLength; //超出此长度的Frame将被丢弃
private final int lengthFieldOffset;
private final int lengthFieldLength;
private final int lengthFieldEndOffset; //这个值等于lengthFieldOffset + lengthFieldLength
private final int lengthAdjustment;
private final int initialBytesToStrip;
protected Object decode(
ChannelHandlerContext ctx, Channel channel, ChannelBuffer buffer) throws Exception {
//数据未完整，先不处理
if (buffer.readableBytes() < lengthFieldEndOffset) {
return null;
}
/*
先读取“Length”的值
在LengthFieldBasedFrameDecoder的构造函数中，限定了“Length”的长度：
“lengthFieldLength must be either 1, 2, 3, 4, or 8”
单位是bytes。这个限定不知从何而来，先不管
由于接收到的数据的类型是ChannelBuffer，也就是byte[]，那么在读取时，
就应该根据长度来分割数据
例如，lengthFieldLength=3，说明读取前3个字节就得到“Length”的值
读取时，用到了位操作，ByteOrder是BIG_ENDIAN，因此高位在前，要左移位：
public int getUnsignedMedium(int index) {
return (array[index] & 0xff) << 16 |
(array[index + 1] & 0xff) << 8 |
(array[index + 2] & 0xff) << 0;
}
注意到，这个方法不会改变readerIndex
下面代码的frameLength，其实就是上面API分析时提到的valueOf(Length)
*/
int actualLengthFieldOffset = buffer.readerIndex() + lengthFieldOffset;
long frameLength;
switch (lengthFieldLength) {
case 1:
frameLength = buffer.getUnsignedByte(actualLengthFieldOffset);
break;
case 2:
frameLength = buffer.getUnsignedShort(actualLengthFieldOffset);
break;
case 3:
frameLength = buffer.getUnsignedMedium(actualLengthFieldOffset);
break;
case 4:
frameLength = buffer.getUnsignedInt(actualLengthFieldOffset);
break;
case 8:
frameLength = buffer.getLong(actualLengthFieldOffset);
break;
default:
throw new Error("should not reach here");
}
//如分析API时所说，要加上lengthAdjustment
//那为什么还要加上lengthFieldEndOffset？
//加上之后，frameLength就代表整个frame的长度了，包括前缀和“Actual Content”
frameLength += lengthAdjustment + lengthFieldEndOffset;
int frameLengthInt = (int) frameLength;
//数据未完整，先不处理
if (buffer.readableBytes() < frameLengthInt) {
return null;
}
//readerIndex往后移，跳过指定的字节，不读取
buffer.skipBytes(initialBytesToStrip);
// extract frame ，读取消息的内容
int readerIndex = buffer.readerIndex();
int actualFrameLength = frameLengthInt - initialBytesToStrip;
ChannelBuffer frame = extractFrame(buffer, readerIndex, actualFrameLength);
//extractFrame方法不改变buffer的readerIndex，因此要手动设置
buffer.readerIndex(readerIndex + actualFrameLength);
return frame;
}
//这个方法创建了新的ChannelBuffer，不影响原buffer
protected ChannelBuffer extractFrame(ChannelBuffer buffer, int index, int length) {
ChannelBuffer frame = buffer.factory().getBuffer(length);
frame.writeBytes(buffer, index, length);
return frame;
}
}