首页 > 代码库 > Android自动化性能收集

Android自动化性能收集

Android自动化性能收集

Android 功能测试自动化框架较多,UIAutomator,Robotium,Appium等。Case执行过程中,可能希望收集手机的性能指标,包括内存、cpu、流量等。使用java+shell+bat简单实现了android手机性能收集。

简述

过程很简单:

  • 在启动自动化case前,执行收集信息的命令。
    其实就是一些adb shell命令,如下:

          adb shell top -n 1| awk ‘{print $3" "$10}‘ >> cpu.dat  
          adb shell ps | awk ‘{print $5" "$9}‘ >> mem.dat  
    
          //android sdk level 大于16  
          adb shell cat /proc/uid_stat/$uid/tcp_rcv >> $uid"_recv.dat"  
          adb shell cat /proc/uid_stat/$uid/tcp_snd >> $uid"_snd.dat"  
    
          //android sdk level 小于16   
          adb shell cat /proc/$pid/net/dev | grep wlan | awk ‘{print $2" "$10}‘ >> $pid"_net.dat"  
    
  • 执行自动化case,可能需要很久。这个过程中,上面的命令在不断地执行

  • case执行结束,kill掉第一步中的命令
  • 分析收集到的dat文件,使用jscharts绘出走势图

脚本

cpu

使用top命令不断查看各进程的cpu占用

linux

get-android-cpu.sh

    #!/bin/sh

    #path--target/android-info/mem/dat
    cd ../../../
    mkdir -p target/android-info/cpu/dat
    mkdir -p target/android-info/cpu/html

    cp src/main/resources/jscharts.js target/android-info/cpu/html
    cd target/android-info/cpu/dat

    while true
    do
        adb shell top -n 1| awk ‘{print $3" "$10}‘ >> cpu.dat
        sleep 15
    done

windows

get-android-cpu.bat

    cd ..\..\jenkins\workspace\android-info-end3
    mkdir target\android-info\cpu\dat
    mkdir target\android-info\cpu\html

    copy src\main\resources\jscharts.js target\android-info\cpu\html
    cd target\android-info\cpu\dat

    :run
        adb shell top -n 1 | awk "{print $3\" \"$10}" >> cpu.dat
        ping 127.0.0.1 -n 15 > null
    goto run

memory

使用ps命令不断查看各进程的内存占用

linux

get-android-mem.sh

    #!/bin/sh

    #path--target/android-info/mem/dat
    cd ../../../
    mkdir -p target/android-info/mem/dat
    mkdir -p target/android-info/mem/html
    cp src/main/resources/jscharts.js target/android-info/mem/html
    cd target/android-info/mem/dat

    while true
    do
        adb shell ps | awk ‘{print $5" "$9}‘ >> mem.dat
        sleep 15
    done

windows

    cd ..\..\jenkins\workspace\android-info-end3
    mkdir target\android-info\mem\dat
    mkdir target\android-info\mem\html
    copy src\main\resources\jscharts.js target\android-info\mem\html

    cd target\android-info\mem\dat

    :run
        adb shell ps | awk "{print $5\" \"$9}" >> mem.dat
        ping 127.0.0.1 -n 15 > null
    goto run

流量

android 4.0以上版本可以用/proc/uid_stat/$uid/tcp_rcv/proc/uid_stat/$uid/tcp_snd来获取某个程序的上下行流量;而4.0以下版本要用cat /proc/$pid/net/dev来查看上下行流量。uid和pid的关系,可以从/data/system/packages.list这个文件中获取。

linux

判断android api level:

    #!/bin/sh
    #android 4.0以上和4.0以下方法不同
    #get android sdk level
    apileveltemp=`adb shell getprop | grep ro.build.version.sdk`
    apilevel=${apileveltemp:25:2}
    chmod +x *.sh

    echo "android api level:"$apilevel
    if [ $apilevel -gt 14 ]
    then
        ./get-android-net-gt-4.0.sh
    elif [ $apilevel -lt 14 ]
    then
        ./get-android-net-lt-4.0.sh
    fi

get-android-net-gt-4.0.sh

    #!/bin/sh

    #android api level great than 14(android 4.0)
    #path--target/android-info/net/dat
    cd ../../../
    mkdir -p target/android-info/net/dat
    mkdir -p target/android-info/net/html
    cp src/main/resources/jscharts.js target/android-info/net/html
    cd target/android-info/net/dat

    cd ..
    echo "adb pull/data/system/packages.list--start"
    adb pull /data/system/packages.list
    cd dat

    while true
    do

        echo "get net info from /proc/uid-stat/$uid"
        for i in `adb shell ls /proc/uid_stat`
        do
            #delete the Enter character
            uid=`echo $i | tr -d ["\r\n"]`
            adb shell cat /proc/uid_stat/$uid/tcp_rcv >> $uid"_recv.dat"
            adb shell cat /proc/uid_stat/$uid/tcp_snd >> $uid"_snd.dat"
        done
        sleep 15
    done

get-android-net-lt-4.0.sh

    #!/bin/sh


    #path--target/android-info/net/dat
    cd ../../../
    mkdir -p target/android-info/net/dat
    mkdir -p target/android-info/net/html
    cp src/main/resources/jscharts.js target/android-info/net/html
    cd target/android-info/net/dat

    #get /proc/$pid/net/dev
    while true
    do
        #get pid
        for i in `adb shell ps | awk ‘{print $2}‘`
        do
            pid=`echo $i | tr -d ["\r\n"]`
            echo $pid
            adb shell cat /proc/$pid/net/dev | grep wlan | awk ‘{print $2" "$10}‘ >> $pid"_net.dat"
        done

        sleep 15
    done

windows

bat命令不熟,不知道如何在windows下实现linux下地反转义,这里用java代码实现。

    import java.io.*;

    /**
     * Created by Xuemeng Wang on 14-9-15.
     * api Level > 16
     */
    public class GetNetInfo {
        public static void main(String[] args) {

            String uidString = execCmd("adb shell ls /proc/uid_stat");
            String[] uidArray = uidString.split("\n");
            int length = uidArray.length-1;
            for(int i=0;i<=length-1;i++)
            {
                String contentRcv = execCmd("adb shell cat /proc/uid_stat/"+uidArray[i]+"/tcp_rcv");
                String contentSnd = execCmd("adb shell cat /proc/uid_stat/"+uidArray[i]+"/tcp_snd");
                System.out.println(System.getProperty("user.dir"));
                try {
                    File file = new File(uidArray[i]+"_recv.dat");
                    if(!file.exists()) file.createNewFile();

                    File file2 = new File(uidArray[i]+"_snd.dat");
                    if(!file2.exists()) file2.createNewFile();

                    FileWriter fileWriter1 = new FileWriter(file, true);
                    FileWriter fileWriter2 = new FileWriter(file2, true);
                    fileWriter1.write(contentRcv);
                    fileWriter2.write(contentSnd);
                    fileWriter1.close();
                    fileWriter2.close();
                } catch (IOException e) {
                    e.printStackTrace();
                }
            }
        }
        public static String execCmd(String command)
        {
            BufferedReader br = null;
            StringBuffer stringBuffer = new StringBuffer();
            try {
                Process p = Runtime.getRuntime().exec(command);
                br = new BufferedReader(new InputStreamReader(p.getInputStream()));
                String line = null;
                while ((line = br.readLine()) != null) {
                if("".equals(line.trim())) continue;

                    stringBuffer.append(line+"\n");
                }
            } catch (Exception e) {
                e.printStackTrace();
            } finally {

                if (br != null) {
                    try {
                        br.close();
                    } catch (Exception e) {
                        e.printStackTrace();
                    }
                }
            }
            return stringBuffer.toString();
        }
    }

api level小于16的情况,我没用到,略掉:)。

杀掉进程

linux: ps -ef | grep get-android- | grep -v grep | awk ‘{print $2}‘ | xargs kill -9
windows(后台运行,其实是cmd进程,注意下面会杀掉所有的cmd进程):

    taskkill /F /IM cmd.exe
    taskkill /F /IM adb.exe

结果收集

使用java分析dat,给出分析cpu的代码,工程已上传到github,https://github.com/yeetrack/android-performance。

CpuInfo.java

    package com.meilishuo.android.performance;

    import org.slf4j.Logger;
    import org.slf4j.LoggerFactory;
    import org.testng.annotations.Test;

    import java.io.*;
    import java.util.ArrayList;
    import java.util.Iterator;
    import java.util.List;
    import java.util.TreeMap;

    /**
     * Created by victor on 14-9-9.
    */
    public class CpuInfo {
        private static final Logger logger = LoggerFactory.getLogger(MemInfo.class);
        private static final String JSCHARTPATH = "target/android-info/cpu/html/";
        private static final String MEMFILEPATH = "target/android-info/cpu/dat/cpu.dat";

        TreeMap<String, List<String>> cpuMap = new TreeMap<String, List<String>>();

        /**
        * 解析cpu.dat存储到map中
        */
        public void parseCpuFile()
        {
            File file = new File(MEMFILEPATH);
            if(null==file) return;

            try {
                BufferedReader bufferedReader = new BufferedReader(new InputStreamReader(new FileInputStream(file)));
                String line = null;

                while((line=bufferedReader.readLine())!=null)
                {
                    String[] array = line.trim().split(" "); //空格分隔
                    if(null==array || array.length!=2 || !array[0].contains("%") || "".equals(array[1]) || "Name".equals(array[1]))
                        continue;
                    if(cpuMap.size()==0 || !cpuMap.containsKey(array[1]))
                    {
                        List<String> memList = new ArrayList<String>();
                        memList.add(array[0].substring(0, array[0].indexOf("%")));
                        cpuMap.put(array[1], memList);
                    }
                    else
                    {
                        cpuMap.get(array[1]).add(array[0].substring(0, array[0].indexOf("%")));
                    }
                }
                System.out.println(cpuMap.size());
            } catch (FileNotFoundException e) {
                e.printStackTrace();
            } catch (IOException e) {
                e.printStackTrace();
            }
        }

        /**
        * 将map中的数据写到到xml中
        */
        public void writeXmlFromMap()
        {
            if(null==cpuMap || cpuMap.size()==0)
                return;
            Iterator<String> it = cpuMap.keySet().iterator();
            while(it.hasNext())
            {
                String key = it.next();
                List<String> value = http://www.mamicode.com/cpuMap.get(key);>

继续阅读->

Android自动化性能收集