本人在做手机APP性能数据的过程中，又重新看了一些Android的内存相关知识，对之前写过的一篇APP性能的线程类的方法做了优化，总得来说，就是增加了PSS数据和增加了数据获取之后的数据整理工作。
获取PSS的方法原理是通过adb shell dumpsys命令获取到的，之前放弃了这个方法，因为内存数据太细分了，后来发现细分的更准确。这里没有统计Native Heap和Dalvik Heap，感觉统计数据的话并没有多大的必要。对这块也不是非常了解如果有不对的地方，还请指正。
方法如下：
/**
* 获取应用信息 利用Android系统dumpsys命令获取
* 命令能统计到java虚拟的堆内存和栈内存的使用情况
*
* @return 返回内存占用
*/
public int getMemResult() {
String cmd1 = Common.ADB_PATH + "adb shell dumpsys meminfo " + package_name;
int mem_result = 0;
try {
Process p = Runtime.getRuntime().exec(cmd1);// 通过runtime类执行cmd命令
// 正确输出流
InputStream input = p.getInputStream();// 创建并实例化输入字节流
BufferedReader reader = new BufferedReader(new InputStreamReader(input));// 先通过inputstreamreader进行流转化，在实例化bufferedreader，接收内容
String line = "";
while ((line = reader.readLine()) != null) {// 循环读取
if (line.startsWith(" TOTAL")) {
mem_result = getMemInfo(line);
}
}
reader.close();// 此处reader依赖于input，应先关闭
input.close();
// 错误输出流
InputStream errorInput = p.getErrorStream();// 创建并实例化输入字节流
BufferedReader errorReader = new BufferedReader(new InputStreamReader(errorInput));// 先通过inputstreamreader进行流转化，在实例化bufferedreader，接收内容
String eline = "";
while ((eline = errorReader.readLine()) != null) {// 循环读取
System.out.println(eline);// 输出
}
errorReader.close();// 此处有依赖关系，先关闭errorReader
errorInput.close();
} catch (IOException e) {
output("执行" + cmd1 + "失败！", e);
}
return mem_result;
}
下面是增加的统计方法，主要是在每次新建进程的时候都会记录一个mark，统计方法写在结束线程的方法里：
public void stopRecord() {
AppLocalMySql.getInstance().ClearUpPerformaceData(mark);//整理数据
PerformanceThread.key = false;//结束线程
}
/**
* 整理一次性能数据收集
*
* @param mark
* 统计mark
*/
public void ClearUpPerformaceData(int mark) {
getConnection();
int cpu = 0, pss = 0, rss = 0, vss = 0, total = 0;
String device = null, packages = null, test_name = null;
try {
if (!connection.isClosed()) {
outputSuccess();
Statement statement = connection.createStatement();
String sql = "SELECT AVG(cpu),AVG(pss),AVG(rss),AVG(vss),COUNT(*),device,package,test_name FROM app_result WHERE mark = "
+ mark;
ResultSet resultSet = statement.executeQuery(sql);
while (resultSet.next()) {
cpu = resultSet.getInt(1);
pss = resultSet.getInt(2);
rss = resultSet.getInt(3);
vss = resultSet.getInt(4);
total = resultSet.getInt(5);
device = resultSet.getString("device");
packages = resultSet.getString("package");
test_name = resultSet.getString("test_name");
}
String sql2 = "INSERT INTO app_report (mark,test_name,package,device,cpu,pss,rss,vss,total) VALUES ("
+ mark + ",'" + test_name + "','" + packages + "','" + device + "'," + cpu + "," + pss + ","
+ rss + "," + vss + "," + total + ")";
statement.executeUpdate(sql2);
statement.close();
connection.close();
}
} catch (SQLException e) {
output("统计方法出错！", e);
}
}
多线程类实现：
package source;import java.io.BufferedReader;import java.io.IOException;import java.io.InputStream;import java.io.InputStreamReader;import java.util.regex.Matcher;import java.util.regex.Pattern;/*** @author ··-·尘* @E-mail:Fhaohaizi@163.com* @version 创建时间：2017年8月23日 下午2:56:22* @alter 修改时间：2017年8月24日 10:02:19 类说明：测试内存和cpu信息*/public class PerformanceThread extends Thread {public int mark = Common.getInstance().getMark();public static boolean key = true;public static String device_id;public static String package_name;public static String test_name;// 创建实例private static PerformanceThread performanceTest;@Overridepublic void run() {while (key) {// 循环执行该方法try {sleep(15 * 1000);// 线程休眠} catch (InterruptedException e) {e.printStackTrace();// 打印异常}getCpuAndMemDate();// 执行手机数据的方法}}public static PerformanceThread getInstance() {if (performanceTest == null) {performanceTest = new PerformanceThread();}return performanceTest;}public PerformanceThread() {getDevicesId();PerformanceThread.package_name = Common.PAGEKAGE;output("欢迎使用手机性能监控线程！");}public PerformanceThread(String package_name) {getDevicesId();PerformanceThread.package_name = package_name;output("欢迎使用手机性能监控线程！");}public PerformanceThread(String testname, String package_name) {getDevicesId();PerformanceThread.package_name = package_name;PerformanceThread.test_name = testname;output("欢迎使用手机性能监控线程！");}/*** 结束线程方法*/public void stopRecord() {AppLocalMySql.getInstance().ClearUpPerformaceData(mark);PerformanceThread.key = false;}/*** 获取devicesID*/public void getDevicesId() {String cmd = "adb devices";String OSname = System.getProperty("os.name");try {Process p = null;if (OSname.contains("Mac")) {p = Runtime.getRuntime().exec(Common.ADB_PATH + cmd);} else {p = Runtime.getRuntime().exec("cmd /c " + cmd);}// 正确输出流InputStream input = p.getInputStream();// 创建并实例化输入字节流BufferedReader reader = new BufferedReader(new InputStreamReader(input));// 先通过inputstreamreader进行流转化，在实例化bufferedreader，接收内容String line = "";while ((line = reader.readLine()) != null) {// 循环读取if (line.contains("device")) {PerformanceThread.device_id = line.replaceAll("device", "");}// System.out.println(line);// 输出}reader.close();// 此处reader依赖于input，应先关闭input.close();} catch (IOException e) {output("执行" + cmd + "失败！");e.printStackTrace();}}/*** 获取cpu和mem数据，此方法用的是top命令*/public void getCpuAndMemDate() {int pss = getMemResult();// 获取内存数据int[] result = getPerformanceResult();// 获取数据int cpu_result = result[0];// 获取cpuint vss = result[1];// 获取虚拟内存数据int rss = result[2];// 获取物理内存数据output("CPU:" + cpu_result, "VSS:" + vss, "RSS:" + rss, "PSS:" + pss);AppLocalMySql.getInstance().saveMemAndCpuResult(mark, package_name, device_id, cpu_result, vss, rss, pss,test_name);// 写入数据库}/*** 获取统计结果 利用Linux系统top命令 -m表示个数，-n表示获取次数此处写为1** @return 返回int数组，包含rss，vss，cpu百分比*/public int[] getPerformanceResult() {String cmd = Common.ADB_PATH + "adb shell top -m 8 -n 1";int cpu_result = 0, vss = 0, rss = 0;try {Process p = Runtime.getRuntime().exec(cmd);// 通过runtime类执行cmd命令// 正确输出流InputStream input = p.getInputStream();// 创建并实例化输入字节流BufferedReader reader = new BufferedReader(new InputStreamReader(input));// 先通过inputstreamreader进行流转化，在实例化bufferedreader，接收内容String line = "";while ((line = reader.readLine()) != null) {// 循环读取if (line.contains(package_name)) {// 获取行cpu_result = getCpuDate(line);vss = getVss(line);rss = getRss(line);}}reader.close();// 此处reader依赖于input，应先关闭input.close();// 关闭流// 错误输出流InputStream errorInput = p.getErrorStream();// 创建并实例化输入字节流BufferedReader errorReader = new BufferedReader(new InputStreamReader(errorInput));// 先通过inputstreamreader进行流转化，在实例化bufferedreader，接收内容String eline = "";while ((eline = errorReader.readLine()) != null) {// 循环读取System.out.println(eline);// 输出}errorReader.close();// 此处有依赖关系，先关闭errorReadererrorInput.close();// 关闭流} catch (IOException e) {output("执行" + cmd + "失败！");e.printStackTrace();}int[] result = { cpu_result, vss, rss };// 新建数组保存数据return result;// 返回数组}/*** 获取应用信息 利用Android系统dumpsys命令获取 命令能统计到java虚拟的堆内存和栈内存的使用情况** @return 返回内存占用*/public int getMemResult() {String cmd1 = Common.ADB_PATH + "adb shell dumpsys meminfo " + package_name;int mem_result = 0;try {Process p = Runtime.getRuntime().exec(cmd1);// 通过runtime类执行cmd命令// 正确输出流InputStream input = p.getInputStream();// 创建并实例化输入字节流BufferedReader reader = new BufferedReader(new InputStreamReader(input));// 先通过inputstreamreader进行流转化，在实例化bufferedreader，接收内容String line = "";while ((line = reader.readLine()) != null) {// 循环读取if (line.startsWith(" TOTAL")) {mem_result = getMemInfo(line);}}reader.close();// 此处reader依赖于input，应先关闭input.close();// 错误输出流InputStream errorInput = p.getErrorStream();// 创建并实例化输入字节流BufferedReader errorReader = new BufferedReader(new InputStreamReader(errorInput));// 先通过inputstreamreader进行流转化，在实例化bufferedreader，接收内容String eline = "";while ((eline = errorReader.readLine()) != null) {// 循环读取System.out.println(eline);// 输出}errorReader.close();// 此处有依赖关系，先关闭errorReadererrorInput.close();} catch (IOException e) {output("执行" + cmd1 + "失败！", e);}return mem_result;}/*** 获取内存信息** @param info* 截取到的APP信息* @return 返回内存蚕蛹*/public int getMemInfo(String info) {int result = 0;Pattern r = Pattern.compile(" [0-9]+ ");Matcher m = r.matcher(info);if (m.find()) {// output(m.group());result = changeStringToInt(m.group().trim());}return result;}/*** 获取cpu运行信息** @param info* 截取到的APP信息* @return 返回CPU占用率，采取double数据格式*/public double getCpuInfo(String info) {String percent = info.substring(0, info.indexOf("%"));double result = changeStringToDouble(percent.trim());return result;}/*** 获取cpu运行信息** @param info* 截取到的APP信息* @return 返回CPU占用率，采取int数据格式*/public int getCpuDate(String info) {Pattern pattern = Pattern.compile(" [0-9]*%");Matcher matcher = pattern.matcher(info);matcher.find();String date = matcher.group().trim();return changeStringToInt(date.substring(0, date.length() - 1));}public int getVss(String info) {Pattern pattern = Pattern.compile(" [0-9]+K");Matcher matcher = pattern.matcher(info);matcher.find();String date = matcher.group().trim();return changeStringToInt(date.substring(0, date.length() - 1));}public int getRss(String info) {Pattern pattern = Pattern.compile("K +[0-9]+K");Matcher matcher = pattern.matcher(info);matcher.find();String date = matcher.group().substring(1, matcher.group().length() - 1);return changeStringToInt(date.trim());}// 把string类型转化为doublepublic Double changeStringToDouble(String text) {// return Integer.parseInt(text);return Double.valueOf(text);}// 把string类型转化为intpublic int changeStringToInt(String text) {// return Integer.parseInt(text);return Integer.valueOf(text);}public void output(String text) {// 明显输出System.out.println(text);}public void output(Object... object) {if (object.length == 1) {output(object[0].toString());return;}for (int i = 0; i < object.length; i++) {System.out.println("第" + (i + 1) + "个：" + object[i]);}}}
往期文章精选
公众号地图
原文作者：FunTester；如有侵权，请联系管理员删除。
下一篇：
公众号地图
请登录后发表评论
登录
全部评论
