网页主动探测中的NIO优化是怎样的
本篇文章给大家分享的是有关网页主动探测中的NIO优化是怎样的,小编觉得挺实用的,因此分享给大家学习,希望大家阅读完这篇文章后可以有所收获,话不多说,跟着小编一起来看看吧。
建表语句:
CREATE SEQUENCE seq_probe_id INCREMENT BY 1 START WITH 1 NOMAXvalue NOCYCLE CACHE 2000;
create table probe(
host varchar(40) not null,
state int not null,
type varchar(10) not null,
) ;
使用NIO优化这个程序,进一步压榨资源使用率,已经想了好长时间了
无奈NIO+多线程,网上例子都不是很靠谱.自己学的也非常头疼,一拖就是一年多.
新的程序,采用三段过程
首先 使用一个线程池不断的发送连接请求,但是不处理接收.仅仅注册一个SelectionKey.OP_READ的键
另外的一个单线程 程序,不断select符合条件的通道,然后分配给另外一个线程池,用于接收数据,解析数据.(接收和解析的过程合并了)
最后,使用一个单线程的程序,不断的把结果通过批量的方式刷入数据库.这块也算一个优化.由单条Insert改为批量入库.这块至少节约了一个CPU核的处理能力.
持久化过程和解析过程 基本复用了原来的代码
<ol start="1" class="dp-j" white-space:normal;"="">
import java.io.IOException;
import java.net.InetSocketAddress;
import java.net.SocketAddress;
import java.nio.ByteBuffer;
import java.nio.channels.SelectionKey;
import java.nio.channels.Selector;
import java.nio.channels.SocketChannel;
import java.nio.charset.Charset;
import java.sql.Connection;
import java.sql.DriverManager;
import java.sql.PreparedStatement;
import java.sql.SQLException;
import java.util.ArrayList;
import java.util.HashSet;
import java.util.Iterator;
import java.util.List;
import java.util.Set;
import java.util.concurrent.BlockingQueue;
import java.util.concurrent.CopyOnWriteArrayList;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.LinkedBlockingQueue;
import java.util.concurrent.atomic.AtomicInteger;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
public class Probe {
private static final int REQUESTTHREADCOUNT = 10;
private static final BlockingQueue CONNECTLIST = new LinkedBlockingQueue();
private static final BlockingQueue PERSISTENCELIST = new LinkedBlockingQueue();
private static ExecutorService REQUESTTHREADPOOL;
private static ExecutorService RESPONSETHREADPOOL;
private static ExecutorService PERSISTENCETHREADPOOL;
private static final List DOMAINLIST = new CopyOnWriteArrayList<>();
private static Selector SELECTOR;
static {
REQUESTTHREADPOOL = Executors.newFixedThreadPool(REQUESTTHREADCOUNT);
RESPONSETHREADPOOL = Executors.newFixedThreadPool(3);
PERSISTENCETHREADPOOL = Executors.newFixedThreadPool(1);
DOMAINLIST.add("news.163.com");
try {
SELECTOR = Selector.open();
} catch (IOException e) {
e.printStackTrace();
}
}
public static void main(String[] args) throws IOException, InterruptedException {
long start = System.currentTimeMillis();
CONNECTLIST.put(new Task("news.163.com", 80, "/index.html"));
for (int i = 0; i < REQUESTTHREADCOUNT; i++) {
REQUESTTHREADPOOL.submit(new RequestHandler(CONNECTLIST, SELECTOR));
}
RESPONSETHREADPOOL
.submit(new ResponseHandler(SELECTOR, CONNECTLIST, PERSISTENCELIST, DOMAINLIST, RESPONSETHREADPOOL));
PERSISTENCETHREADPOOL.submit(new PersistenceHandler(PERSISTENCELIST));
while (true) {
Thread.sleep(1000);
long end = System.currentTimeMillis();
float interval = ((end - start) / 1000);
int connectTotal = ResponseHandler.GETCOUNT();
int persistenceTotal = PersistenceHandler.GETCOUNT();
int connectps = Math.round(connectTotal / interval);
int persistenceps = Math.round(persistenceTotal / interval);
System.out.print(
"\r连接总数:" + connectTotal + " \t每秒连接:" + connectps + "\t连接队列剩余:" + CONNECTLIST.size() + " \t持久化总数:"
+ persistenceTotal + " \t每秒持久化:" + persistenceps + "\t持久化队列剩余:" + PERSISTENCELIST.size());
}
}
}
class RequestHandler implements Runnable {
BlockingQueue connectlist;
Selector selector;
public RequestHandler(BlockingQueue connectlist, Selector selector) {
this.connectlist = connectlist;
this.selector = selector;
}
@Override
public void run() {
while (true) {
try {
Task task = (Task) connectlist.take();
SocketAddress addr = new InetSocketAddress(task.getHost(), 80);
SocketChannel socketChannel = SocketChannel.open(addr);
socketChannel.configureBlocking(false);
ByteBuffer byteBuffer = ByteBuffer.allocate(2400);
byteBuffer.put(("GET " + task.getCurrentPath() + " HTTP/1.0\r\n").getBytes("utf8"));
byteBuffer.put(("HOST:" + task.getHost() + "\r\n").getBytes("utf8"));
byteBuffer.put(("Accept:*
private void createNewTask(Task fatherTask, String path) throws Exception {
Task newTask = new Task();
FilterChain filterchain = new FilterChain();
filterchain.doFilter(fatherTask, newTask, path, filterchain);
if (newTask.isValid()) {
synchronized (SET) {
if (SET.contains(newTask.getHost() + newTask.getCurrentPath())) {
return;
}
SET.add(newTask.getHost() + newTask.getCurrentPath());
}
connectlist.put(newTask);
}
}
private String getPrefix(String s, int count) {
String prefix = s;
while (count > 0) {
prefix = prefix.substring(0, prefix.lastIndexOf("/"));
count--;
}
return "".equals(prefix) ? "/" : prefix;
}
}
class Task {
public Task() {
}
public void init(String host, int port, String path) {
this.setCurrentPath(path);
this.host = host;
this.port = port;
}
public Task(String host, int port, String path) {
init(host, port, path);
}
private String host;
private int port;
private String currentPath;
private long starttime;
private long endtime;
public long getStarttime() {
return starttime;
}
public void setStarttime(long starttime) {
this.starttime = starttime;
}
public long getEndtime() {
return endtime;
}
public void setEndtime(long endtime) {
this.endtime = endtime;
}
private long taskTime;
private String type;
private StringBuilder content = new StringBuilder(2400);
private int state;
private boolean isValid = true;
public boolean isValid() {
return isValid;
}
public void setValid(boolean isValid) {
this.isValid = isValid;
}
public int getState() {
return state;
}
public void setState(int state) {
this.state = state;
}
public String getCurrentPath() {
return currentPath;
}
public void setCurrentPath(String currentPath) {
this.currentPath = currentPath;
int i = 0;
if (currentPath.indexOf("?") != -1) {
i = currentPath.indexOf("?");
} else {
if (currentPath.indexOf("#") != -1) {
i = currentPath.indexOf("#");
} else {
i = currentPath.length();
}
}
this.type = currentPath.substring(currentPath.indexOf(".") + 1, i);
}
public long getTaskTime() {
return getEndtime() - getStarttime();
}
public String getType() {
return type;
}
public void setType(String type) {
this.type = type;
}
public String getHost() {
return host;
}
public int getPort() {
return port;
}
public String getContent() {
return content.toString();
}
public void dropContent() {
this.content = null;
}
public void appendContent(String content) {
this.content.append(content);
}
}
class PersistenceHandler implements Runnable {
static {
try {
Class.forName("oracle.jdbc.OracleDriver");
} catch (ClassNotFoundException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
public static int GETCOUNT() {
return COUNT.get();
}
private static final AtomicInteger COUNT = new AtomicInteger();
private BlockingQueue persistencelist;
public PersistenceHandler(BlockingQueue persistencelist) {
this.persistencelist = persistencelist;
try {
conn = DriverManager.getConnection("jdbc:oracle:thin:127.0.0.1:1521:orcl", "edmond", "edmond");
ps = conn.prepareStatement(
"insert into probe(id,host,path,state,tasktime,type) values(seq_probe_id.nextval,?,?,?,?,?)");
} catch (SQLException e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
}
private Connection conn;
private PreparedStatement ps;
@Override
public void run() {
while (true) {
this.handler();
COUNT.addAndGet(1);
}
}
private void handler() {
try {
Task task = (Task) persistencelist.take();
ps.setString(1, task.getHost());
ps.setString(2, task.getCurrentPath());
ps.setInt(3, task.getState());
ps.setLong(4, task.getTaskTime());
ps.setString(5, task.getType());
ps.addBatch();
if (GETCOUNT() % 500 == 0) {
ps.executeBatch();
conn.commit();
}
} catch (InterruptedException e) {
e.printStackTrace();
} catch (SQLException e) {
e.printStackTrace();
}
}
}
每秒可以爬170-200左右的网页
因为这个速度受制于公司带宽.
CPU也基本上跑满了
这个程序还有优化的空间,主要是以下代码的阻塞和唤醒关系,还是没有搞明白.
socketChannel.register(selector, SelectionKey.OP_READ, task);
int n = selector.select();
key.selector().wakeup();
以上就是网页主动探测中的NIO优化是怎样的,小编相信有部分知识点可能是我们日常工作会见到或用到的。希望你能通过这篇文章学到更多知识。更多详情敬请关注编程网行业资讯频道。
免责声明:
① 本站未注明“稿件来源”的信息均来自网络整理。其文字、图片和音视频稿件的所属权归原作者所有。本站收集整理出于非商业性的教育和科研之目的,并不意味着本站赞同其观点或证实其内容的真实性。仅作为临时的测试数据,供内部测试之用。本站并未授权任何人以任何方式主动获取本站任何信息。
② 本站未注明“稿件来源”的临时测试数据将在测试完成后最终做删除处理。有问题或投稿请发送至: 邮箱/279061341@qq.com QQ/279061341