【我分享】 Android应用性能测试
对于Web网页来说，页面的访问、加载速度对于用户体验来说是很重要的，而如果把Android中的每个Activity都看成是一个页面的话，Activity的启动速度凭主观的话是较难精确衡量的，因此如果可以测试每个Activity的启动速度或者获得其它基本指标并进行日常监测那就更好了。一、编写继承于Instrumentation类的LaunchPerformanceBase类[*]/**
[*]* Base class for all launch performance Instrumentation classes.
[*]*/
[*]public class LaunchPerformanceBase extends Instrumentation {
[*] public static final String TAG = "LaunchPerformanceBase";
[*] protected Bundle mResults;
[*] protected Intent mIntent;
[*]
[*] /**
[*] * Constructor.
[*] */
[*] public LaunchPerformanceBase() {
[*] mResults = new Bundle();
[*] mIntent = new Intent(Intent.ACTION_MAIN);
[*] mIntent.setFlags(Intent.FLAG_ACTIVITY_NEW_TASK);
[*] setAutomaticPerformanceSnapshots();
[*] }
[*] /**
[*] * Launches intent {@link #mIntent}, and waits for idle before
[*] * returning.
[*] */
[*] protected void LaunchApp() {
[*] startActivitySync(mIntent);
[*] waitForIdleSync();
[*] }
[*]
[*] @Override
[*] public void finish(int resultCode, Bundle results) {
[*] super.finish(resultCode, results);
[*] }
[*]}
在这个类的构造函数中setAutomaticPerformanceSnapshots()为Instrumentation设置为开启性能快照功能。
LaunchApp()方法中用于启动相应的Activity, waitForIdleSync()方法则为等待Activity空闲时，即等待Activity启动完毕。
二、编写ActivityLaunchPerformanceTest类
view plaincopyhttps://code.csdn.net/assets/CODE_ico.pnghttps://code.csdn.net/assets/ico_fork.svg
[*]public class ActivityLaunchPerformanceTest extends LaunchPerformanceBase {
[*] /**
[*] * Outfile argument name.
[*] * This argument can be passed to the instrumentation using
[*] <code>-e</code>.
[*] */
[*] private static final String launchActivityName = "launch_activity";
[*] /**
[*] * Output file name.
[*] */
[*] private String classNameForIntent;
[*] private String DEFAULT_ACTIVITY = "SplashActivity";
[*]
[*] /**
[*] * Constructor.
[*] */
[*] public ActivityLaunchPerformanceTest() {
[*] super();
[*] }
[*]
[*] @Override
[*] public void onCreate(Bundle arguments) {
[*] if ( arguments != null ) {
[*] classNameForIntent = arguments.getString(launchActivityName);
[*] }
[*] if ( classNameForIntent == null ) {
[*] classNameForIntent = DEFAULT_ACTIVITY ;
[*] }
[*] super.onCreate(arguments);
[*] mIntent.setClassName("com.company.example",
[*] "com.company.example.ui.activity." + classNameForIntent);
[*] start();
[*] }
[*]
[*] /**
[*] * Calls LaunchApp and finish.
[*] */
[*] @Override
[*] public void onStart() {
[*] super.onStart();
[*] LaunchApp();
[*] finish(Activity.RESULT_OK, mResults);
[*] }
[*]}
这个类中onCreate()方法中传入要测试的Activity的名字。当开始测试时，Instrumentation启动，onStart方法执行时，运行LaunchApp()方法启动被测试的Activity.运行完成后finish。
Instrumentation退出，测试结束。
三、修改AndroidManifest.xml文件
在Manifest文件中增加Instrumentation申明
[*]<instrumentation
[*] android:label="Activity Launch Performance"
[*] android:name="com.company.example.test.performance.ActivityLaunchPerformanceTest"
[*] android:targetPackage="com.company.example" />
四、运行Activity启动性能的测试用例
[*]adb shell am instrument -e launch_activity HomeActivity -w com.company.example.test/.performance.ActivityLaunchPerformanceTest
-e launch_activity参数后指定要测试的Activity名.测试结果大致如下：
[*]INSTRUMENTATION_RESULT: other_pss=7437
[*]INSTRUMENTATION_RESULT: java_allocated=4839
[*]INSTRUMENTATION_RESULT: global_freed_size=2583696
[*]INSTRUMENTATION_RESULT: native_private_dirty=1684
[*]INSTRUMENTATION_RESULT: native_free=81
[*]INSTRUMENTATION_RESULT: global_alloc_count=51608
[*]INSTRUMENTATION_RESULT: other_private_dirty=5468
[*]INSTRUMENTATION_RESULT: global_freed_count=18818
[*]INSTRUMENTATION_RESULT: sent_transactions=-1
[*]INSTRUMENTATION_RESULT: java_free=2784
[*]INSTRUMENTATION_RESULT: received_transactions=-1
[*]INSTRUMENTATION_RESULT: pre_sent_transactions=-1
[*]INSTRUMENTATION_RESULT: other_shared_dirty=7300
[*]INSTRUMENTATION_RESULT: pre_received_transactions=-1
[*]INSTRUMENTATION_RESULT: execution_time=749
[*]INSTRUMENTATION_RESULT: native_size=4772
[*]INSTRUMENTATION_RESULT: native_shared_dirty=620
[*]INSTRUMENTATION_RESULT: cpu_time=678
[*]INSTRUMENTATION_RESULT: java_private_dirty=320
[*]INSTRUMENTATION_RESULT: native_allocated=4690
[*]INSTRUMENTATION_RESULT: gc_invocation_count=5
[*]INSTRUMENTATION_RESULT: java_shared_dirty=1972
[*]INSTRUMENTATION_RESULT: global_alloc_size=3883815
[*]INSTRUMENTATION_RESULT: java_pss=2618
[*]INSTRUMENTATION_RESULT: java_size=7623
[*]INSTRUMENTATION_RESULT: native_pss=1699
[*]INSTRUMENTATION_CODE: -1
其中意义较大的结果有如cpu_time，execution_time分别代表Activity启动过程中进程占用的cpu时间与实际执行时间，单位毫秒。其它参数有兴趣的话可参照Instrumentation源码。
http://img.blog.csdn.net/20140407210238250?watermark/2/text/aHR0cDovL2Jsb2cuY3Nkbi5uZXQvaHVudGVybm80/font/5a6L5L2T/fontsize/400/fill/I0JBQkFCMA==/dissolve/70/gravity/SouthEast
五、对测试结果进行文本处理
1.进行格式化处理
view plaincopy在CODE上查看代码片派生到我的代码片
adbshell am instrument -e launch_activity HomeActivity -w com.company.example.test/.performance.ActivityLaunchPerformanceTest | sed 's/=/:/' | sed 's/ //' | sed 's/\r//'
对测试结果进行=替换为：去除空格等格式化处理
2.编写gawk脚本,名字为txt_to_xml.gawk
view plaincopy在CODE上查看代码片派生到我的代码片
#!/bin/bash
BEGIN{
print "<?xml version=\"1.0\" encoding=\"UTF-8\"?>"
print "<testsuite>"
FS=":"
}
$2 ~ /execution_time|cpu_time/{
print "<testcase name=\"" $2 "\" time=\"" $NF*0.001
print "\" />"
}
END{
print "</testsuite>"
}
这里只提取需要的cpu_time，execution_time两个字段的值，并将结果最终生成单元测试格式的xml文件
最终执行测试用例的命令如下：
view plaincopy在CODE上查看代码片派生到我的代码片
adb shell am instrument -e launch_activity HomeActivity -w com.company.example.test/.performance.ActivityLaunchPerformanceTest | sed 's/=/:/' | sed 's/ //' | sed 's/\r//' | gawk -f txt_to_xml.gawk > TEST-HomeActivity.xml
得到的xml结果如下：
view plaincopy在CODE上查看代码片派生到我的代码片
<?xml version="1.0" encoding="UTF-8"?>
<testsuite>
<testcase name="execution_time" time="0.939
" />
<testcase name="cpu_time" time="0.85
" />
</testsuite>
六、Jenkins结果展示
测试用例可以使用命令行方式执行，因此也就可以使用Jenkins完成自动化测试，且对于生成的单元测试格式的xml报告，可以使用Jenkins的Performance Plugin插件进行图表化展示： good，不错 :lol:lol:lol:lol:lol:lol
页:
[1]
