早之前就用過Google的Protobuf做資料編碼，一直沒有深入理解其中的原理，最近做了一次通訊抓包，發現其中很多Protobuf編碼的資料包，於是決定分析一下其中的資料包及其編碼。

一、Protobuf的使用

首先來簡單介紹一下Protobuf的使用，這裡以windows下java開發為例，幾個步驟：編寫*.proto ->使用google提供的protoc.exe生成*.java->專案中匯入protobuf的.jar包進行開發即可。先看這裡的*.proto檔案：

package com;

message CMsg
{
    required string msghead = 1;
    required string msgbody = 2;
}

message CMsgHead
{
    required int32 msglen = 1;
    required int32 msgtype = 2;
    required int32 msgseq = 3;
    required int32 termversion = 4;
    required int32 msgres = 5;
    required string termid = 6;
}

message CMsgReg
{
    optional int32 area = 1;
    optional int32 region = 2;
    optional int32 shop = 3;
    optional int32 ret = 4;
    optional string termid = 5;
}
 

使用protoc.exe生成java檔案，命令如下：

將生成的Msg.java及protobuf-java-2.3.0.jar匯入專案中進行開發，這裡寫一個伺服器端ProtobufServer及客戶端ProtobufClient

package com;

import java.io.DataInputStream;
import java.io.DataOutputStream;
import java.io.IOException;
import java.io.InputStream;
import java.net.ServerSocket;
import java.net.Socket;

import com.Msg.CMsg;
import com.Msg.CMsgHead;
import com.Msg.CMsgReg;

public class ProtoServer implements Runnable {

	@Override
	public void run() {
		try {
			System.out.println("beign:");
			ServerSocket serverSocket = new ServerSocket(12345);
			while (true) {
				System.out.println("等待接收使用者連線：");
				// 接受客戶端請求
				Socket client = serverSocket.accept();

				DataOutputStream dataOutputStream;
				DataInputStream dataInputStream;

				try {
					InputStream inputstream = client.getInputStream();
					dataOutputStream = new DataOutputStream(
							client.getOutputStream());

					byte len[] = new byte[1024];
					int count = inputstream.read(len);
					byte[] temp = new byte[count];
					for (int i = 0; i < count; i++) {
						temp[i] = len[i];
					}

					CMsg msg = CMsg.parseFrom(temp);

					CMsgHead head = CMsgHead.parseFrom(msg.getMsghead()
							.getBytes());
					System.out.println("==len===" + head.getMsglen());
					System.out.println("==res===" + head.getMsgres());
					System.out.println("==seq===" + head.getMsgseq());
					System.out.println("==type===" + head.getMsgtype());
					System.out.println("==Termid===" + head.getTermid());
					System.out.println("==Termversion==="
							+ head.getTermversion());

					CMsgReg body = CMsgReg.parseFrom(msg.getMsgbody()
							.getBytes());
					System.out.println("==area==" + body.getArea());
					System.out.println("==Region==" + body.getRegion());
					System.out.println("==shop==" + body.getShop());

					sendProtoBufBack(dataOutputStream);
					inputstream.close();

				} catch (Exception ex) {
					System.out.println(ex.getMessage());
					ex.printStackTrace();
				} finally {
					client.close();
					System.out.println("close");
				}
			}
			
		} catch (IOException e) {
			System.out.println(e.getMessage());
		}
	}

	private byte[] getProtoBufBack() {

		// head
		CMsgHead head = CMsgHead.newBuilder().setMsglen(10).setMsgtype(21)
				.setMsgseq(32).setTermversion(43).setMsgres(54)
				.setTermid("Server:head").build();

		// body
		CMsgReg body = CMsgReg.newBuilder().setArea(11).setRegion(22)
				.setShop(33).setRet(44).setTermid("Server:body").build();

		// Msg
		CMsg msg = CMsg.newBuilder()
				.setMsghead(head.toByteString().toStringUtf8())
				.setMsgbody(body.toByteString().toStringUtf8()).build();

		return msg.toByteArray();
	}

	private void sendProtoBufBack(DataOutputStream dataOutputStream) {

		byte[] backBytes = getProtoBufBack();
		
		// Integer len2 = backBytes.length;
		
		// byte[] cmdHead2 = BytesUtil.IntToBytes4(len2);

		try {
			// dataOutputStream.write(cmdHead2, 0, cmdHead2.length);
			dataOutputStream.write(backBytes, 0, backBytes.length);
			dataOutputStream.flush();
		} catch (IOException e) {
			e.printStackTrace();
		}
	}

	public static void main(String[] args) {
		Thread desktopServerThread = new Thread(new ProtoServer());
		desktopServerThread.start();
	}

}
 

package com;

import java.io.InputStream;
import java.net.Socket;

import com.Msg.CMsg;
import com.Msg.CMsgHead;
import com.Msg.CMsgReg;

public class ProtoClient {

	public static void main(String[] args) {
		ProtoClient pc=new ProtoClient();
		System.out.println("beign:");
		pc.runget();
	}

	public void runget() {
		Socket socket = null;
		try {
			//socket = new Socket("localhost", 12345);
			socket = new Socket("192.168.85.152", 12345);
			// head
			CMsgHead head = CMsgHead.newBuilder().setMsglen(5).setMsgtype(1)
					.setMsgseq(3).setTermversion(41).setMsgres(5)
					.setTermid("Client:head").build();

			// body
			CMsgReg body = CMsgReg.newBuilder().setArea(11).setRegion(22)
					.setShop(33).setRet(44).setTermid("Clent:body").build();
			
			// Msg
			CMsg msg = CMsg.newBuilder()
					.setMsghead(head.toByteString().toStringUtf8())
					.setMsgbody(body.toByteString().toStringUtf8()).build();

			// 向伺服器傳送資訊
			System.out.println("sendMsg...");
			msg.writeTo(socket.getOutputStream());

			// 接受伺服器的資訊
			InputStream input = socket.getInputStream();

			System.out.println("recvMsg:");
			byte[] by = recvMsg(input);
			printMsg(CMsg.parseFrom(by));

			input.close();
			socket.close();
		} catch (Exception e) {
			System.out.println(e.toString());
		}
	}

	public void printMsg(CMsg g) {

		try {
			CMsgHead h = CMsgHead.parseFrom(g.getMsghead().getBytes());
			StringBuffer sb = new StringBuffer();
			if (h.hasMsglen())
				sb.append("==msglen===" + h.getMsglen() + "\n");
			if (h.hasMsgres())
				sb.append("==msgres===" + h.getMsgres() + "\n");
			if (h.hasMsgseq())
				sb.append("==msgseq===" + h.getMsgseq() + "\n");
			if (h.hasMsgtype())
				sb.append("==msgtype===" + h.getMsgtype() + "\n");
			if (h.hasTermid())
				sb.append("==termid===" + h.getTermid() + "\n");
			if (h.hasTermversion())
				sb.append("==termversion===" + h.getTermversion() + "\n");

			CMsgReg bo = CMsgReg.parseFrom(g.getMsgbody().getBytes());
			if (bo.hasArea())
				sb.append("==area==" + bo.getArea() + "\n");
			if (bo.hasRegion())
				sb.append("==region==" + bo.getRegion() + "\n");
			if (bo.hasShop())
				sb.append("==shop==" + bo.getShop() + "\n");
			if (bo.hasRet())
				sb.append("==ret==" + bo.getRet() + "\n");
			if (bo.hasTermid())
				sb.append("==termid==" + bo.getTermid() + "\n");

			System.out.println(sb.toString());

		} catch (Exception e) {
			e.printStackTrace();
		}

	}

	public byte[] recvMsg(InputStream inpustream) {
		byte[] temp = null;
		try {

			byte len[] = new byte[1024];
			int count = inpustream.read(len);

			temp = new byte[count];
			for (int i = 0; i < count; i++) {
				temp[i] = len[i];
			}
			return temp;
		} catch (Exception e) {
			System.out.println(e.toString());
			return temp;
		}
	}
}
 

執行結果：

二、抓包分析

在上面socket通訊過程中我使用了wireshark對其進行抓包，結果分析如下圖

由上圖我們可以很清楚的看到，protobuf編碼其實類似tlv（tag length value）編碼，其內部就是（tag, length, value)的組合，其中tag由(field_number<<3)|wire_type計算得出，field_number由我們在proto檔案中定義，wire_type由protobuf根據proto中定義的欄位型別決定，length長度採用一種叫做Varint 的數字表示方法，它是一種緊湊的表示數字的方法，用一個或多個位元組來表示一個數字，值越小的數字使用越少的位元組數，具體細節可以谷歌Varint。總之Protobuf 序列化後所生成的二進位制訊息非常緊湊，這得益於 Protobuf 採用了上面的 Encoding 方法。