netty 数据分包、组包、粘包处理机制

转载自 断鸿零雁 blog.163.com/linfenliang@126

1.            frame包整体功能描述

此包主要作用于对TCP/IP数据包的分包和包重组,常用于数据的流传输,是扩展的解码器。

包目录结构如下:

 

 

2.            包中各类功能详解

(1)  FrameDecoder

抽象类,将ChannelBuffers中的二进制数据转换成有意义的数据帧(frame)对象,一般不直接调用,提供给此包中的FixedLengthFrameDecoder类、DelimiterBasedFrameDecoder类和LengthFieldBasedFrameDecoder类使用,也可以提供给其他类使用(暂不探讨);

 

在数据传输中,我们发送的数据包如下所示

 

+-----+-----+-----+

 | ABC | DEF | GHI |

 +-----+-----+-----+

 

而实际接收的包的格式为:

 

+----+-------+---+---+
 | AB | CDEFG | H | I |
 +----+-------+---+---+

 

 

产生的原因为:数据在传输过程中,产生数据包碎片(TCP/IP数据传输时大数据包无法一次传输,被拆分成小数据包,小数据包即为数据包碎片),这就造成了实际接收的数据包和发送的数据包不一致的情况。

 

而通过FrameDecoder即可实现对上述接收到的数据包的整理,重新还原成如下格式:

 

+-----+-----+-----+
 | ABC | DEF | GHI |
 +-----+-----+-----+

 

如下是一个自定义的Decoder类

 

public class MyFrameDecoder extends FrameDecoder {

 

         @Override

   protected Object decode(ChannelHandlerContext ctx,

                           channel,

                           ChannelBuffer buf) throws Exception {

 

     // Make sure if the length field was received.

     if (buf.readableBytes() < 4) {

        // The length field was not received yet - return null.

        // This method will be invoked again when more packets are

        // received and appended to the buffer.

        return null;

     }

 

     // The length field is in the buffer.

 

     // Mark the current buffer position before reading the length field

     // because the whole frame might not be in the buffer yet.

     // We will reset the buffer position to the marked position if

     // there's not enough bytes in the buffer.

     buf.markReaderIndex();
 

     // Read the length field.

     int length = buf.readInt();
 

     // Make sure if there's enough bytes in the buffer.

     if (buf.readableBytes() < length) {

        // The whole bytes were not received yet - return null.

        // This method will be invoked again when more packets are

        // received and appended to the buffer.

        // Reset to the marked position to read the length field again

        // next time.

        buf.resetReaderIndex();
 
        return null;

     }

 

     // There's enough bytes in the buffer. Read it.

     ChannelBuffer frame = buf.readBytes(length);
 

     // Successfully decoded a frame.  Return the decoded frame.

     return frame;

   }

 }

 

此时,我们无需关注数据包是如何重组的,只需要做简单的验证(按照一个包验证)就可以了,FrameDecoder内部实现了组包的机制,不过,此时,需在数据的最前面封装整个数据的长度,示例中数据长度占了四个字节,即前四个字节是数据长度,后面的才是真实的数据。

(2)  FixedLengthFrameDecoder

FixedLengthFrameDecoder主要是将诸如

+----+-------+---+---+
 | AB | CDEFG | H | I |
 +----+-------+---+---+

 

此类的数据包按照指定的frame长度重新组包,比如确定长度为3,则组包为

 

+-----+-----+-----+
 | ABC | DEF | GHI |
 +-----+-----+-----+

 

构造方法为:new FixedLengthFrameDecoder(int frameLength);

 

frameLength即修正后的帧长度

 

另一个构造方法为new FixedLengthFrameDecoder(int frameLength, boolean allocateFullBuffer);

 

allocateFullBuffer如果为真,则表示初始化的ChannelBuffer大小为frameLength。

 

 

(3)  Delimiters

分隔符类,DelimiterBasedFrameDecoder类的辅助类。

 

对Flash XML的socket通信采用nulDelimiter()方法,对于一般的文本采用lineDelimiter()方法

 

(4)  DelimiterBasedFrameDecoder

对接收到的ChannelBuffers按照指定的分隔符Delimiter分隔,分隔符可以是一个或者多个

 

如将以下数据包按照“\n”分隔:

 
+--------------+
 | ABC\nDEF\r\n |
 +--------------+

 

即为:

 

+-----+-----+
 | ABC | DEF |
 +-----+-----+
 

而如果按照“\r\n”分隔,则为:

 

+----------+
 | ABC\nDEF |
 +----------+

 

对于DelimiterBasedFrameDecoder中的构造方法,其中一些参数说明:

 

maxFrameLength:解码的帧的最大长度

stripDelimiter:解码时是否去掉分隔符

failFast:为true,当frame长度超过maxFrameLength时立即报TooLongFrameException异常,为false,读取完整个帧再报异常

delimiter:分隔符


(5)  LengthFieldBasedFrameDecoder

常用的处理大数据分包传输问题的解决类,先对构造方法LengthFieldBasedFrameDecoder中的参数做以下解释说明“

maxFrameLength:解码的帧的最大长度

lengthFieldOffset :长度属性的起始位(偏移位),包中存放有整个大数据包长度的字节,这段字节的其实位置

lengthFieldLength:长度属性的长度,即存放整个大数据包长度的字节所占的长度

lengthAdjustmen:长度调节值,在总长被定义为包含包头长度时,修正信息长度。initialBytesToStrip:跳过的字节数,根据需要我们跳过lengthFieldLength个字节,以便接收端直接接受到不含长度属性的内容

failFast :为true,当frame长度超过maxFrameLength时立即报TooLongFrameException异常,为false,读取完整个帧再报异常

 

下面对各种情况分别描述:

 

1. 2 bytes length field at offset 0, do not strip header

 

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" |

 +--------+----------------+      +--------+----------------+

 

此时数据格式不做任何改变(没有跳过任何字节)

 

2. 2 bytes length field at offset 0, strip header

 

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" |
 +--------+----------------+      +----------------+

 

此时帧长度为14个字节,但由于lengthFieldOffset = 0两个lengthFieldLength = 2)字节是表示帧长度的字节,不计入数据,故真实的数据长度为12个字节。

 

3. 2 bytes length field at offset 0, do not strip header, the length field represents the length of the whole message

 

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" |
 +--------+----------------+     +--------+----------------+
 

此处定义的Length0x000E共占了两个字节,表示的帧长度为14个字节,lengthFieldOffset = 0两个lengthFieldLength = 2)字节为Length,由于设置的lengthAdjustment    = -2 (= the length of the Length field),故修正的信息实际长度补2,即解码时往前推2个字节,解码后还是14个字节长度(此种情况是把整个长度封装,一般来讲,我们只封装数据长度)

 

4. 3 bytes length field at the end of 5 bytes header, do not strip header

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"|
 +---------+---------+--------------+    +----------+--------+-----------+

 

此处lengthFieldOffset   = 2,从第3个字节开始表示数据长度,长度占3个字节,真实数据长度为0x00000C 12个字节,而lengthAdjustment=0initialBytesToStrip = 0,故解码后的数据与解码前的数据相同。

4. 3 bytes length field at the beginning of 5 bytes header, do not strip header

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" |
 +----------+----------+----------------+      +----------+----------+----------------+

此处由于修正的字节数是2,且initialBytesToStrip = 0,故整个数据的解码数据保持不变

总字节数是17,开始的三个字节表示字节长度:12,修正的字节是2,(即从第三个字节开始,再加两个开始是真正的数据,其中跳过的字节数是0

 

5. 2 bytes length field at offset 1 in the middle of 4 bytes header, strip the first header field and the length field

 

 

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" |

 +------+--------+------+----------------+      +------+----------------+

 

从第2个字节开始解码,取两个字节作为帧长度,为12个字节,然后,修正一个字节,从第5个字节到最后表示帧数据,解码时,由于initialBytesToStrip=3,表示跳过前三个字节(去掉),故从第四个字节开始解析,解析出来后,如右图所示。

 

6. 2 bytes length field at offset 1 in the middle of 4 bytes header, strip the first header field and the length field, the length field represents the length of the whole message

 

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" |

 +------+--------+------+----------------+      +------+----------------+

 

从第二个字节开始,取两个字节作为帧长度,为16个字节,然后补3个字节,故往前找三个字节,从HDP1开始解码,而又因为initialBytesToStrip=3,解码时忽略掉前三个字节,故从第四个字节开始解析,解析结果如右图所示。

 

总结:一般来讲,当lengthAdjustment 为负数时,Length表示的是整个帧的长度,当lengthAdjustment为正数或0时,表示真实数据长度。

(6)  LengthFieldPrepender

编码类,自动将

+----------------+

| "HELLO, WORLD" |

+----------------+

格式的数据转换成

+--------+----------------+

 + 0x000C | "HELLO, WORLD" |

 +--------+----------------+

格式的数据,

如果lengthIncludesLengthFieldLength设置为true,则编码为

+--------+----------------+

+ 0x000E | "HELLO, WORLD" |

+--------+----------------+

格式的数据

 

应用场景:自定义pipelineFactoryMyPipelineFactory implements ChannelPipelineFactory

pipeline.addLast("frameEncode", new LengthFieldPrepender(4, false));

(7)  TooLongFrameException

定义的数据包超过预定义大小异常类

(8)  CorruptedFrameException

定义的数据包损坏异常类

3.            frame包应用demo

解决分包问题,通常配置MyPipelineFactory中设置,示例如下:

 

public class MyPipelineFactory implements ChannelPipelineFactory {

 

    @Override

    public ChannelPipeline getPipeline() throws Exception {

       ChannelPipeline pipeline = Channels.pipeline();

       pipeline.addLast("decoder", new LengthFieldBasedFrameDecoder(Integer.MAX_VALUE, 0, 4, 0, 4)); 

        pipeline.addLast("encoder", new LengthFieldPrepender(4, false));

       pipeline.addLast("handler", new MyHandler());

       return pipeline;

    }

 

}

 

 

在客户端设置pipeline.addLast("encoder", new LengthFieldPrepender(4, false));

       pipeline.addLast("handler", new MyHandler());

 

前四个字节表示真实的发送的数据长度Length,编码时会自动加上;

 

在服务器端设置pipeline.addLast("decoder", new LengthFieldBasedFrameDecoder(Integer.MAX_VALUE, 0, 4, 0, 4));

真实数据最大字节数为Integer.MAX_VALUE,解码时自动去掉前面四个字节


你可能感兴趣的:(服务器,网络)