《HowTomcatWorks》读书笔记（四）TomcatDefaultConnector

上节《How Tomcat Works》读书笔记&＃xff08;三&＃xff09;&＃xff0c;只是讲了一个简单的connector实现。
本章我们来看看tomcat1.4中默认Connector真正的实现&＃xff08;虽然现在已经不再使用&＃xff0c;但对我们学习原理还是有很大帮助的&＃xff09;。

我们来看一下tomcat的一个基本的connector需要做什么&＃xff1a;

• 实现org.apache.catalina.Connector接口
• 创建实现了org.apache.catalina.Request接口的request对象
• 创建实现了org.apache.catalina.Response接口的response对象

再看下它都做了哪些优化&＃xff1a;

• 为一些大对象创建pool,减少了创建大对象的时间开销&＃xff0c;典型的以空间换时间方法
• 很多地方使用char[]替代String
  

private Stack processors &＃61; new Stack(); 还包含 Processor的最大值maxProcessors&＃xff0c;最小值minProcessors&＃xff0c;当前值curProcessors。

protected int minProcessors &＃61; 5;
private int maxProcessors &＃61; 20;
private int curProcessors &＃61; 0;

在初始化过程中会新建minProcessors个HttpProcessor

while (curProcessors 0) && (curProcessors >&＃61; maxProcessors))break;HttpProcessor processor &＃61; newProcessor(this);recycle(processor);
} 依次放入processors栈中。

public void recycle(HttpProcessor processor) {processors.push(processor);
} 当请求到达&＃xff08;即accept到一个socket时&＃xff09;&＃xff0c; Connector从栈中取出一个Processor来进行处理。
如果不够&＃xff0c;就新建一个Processor。
如果curProcessors已经等于maxProcessors了&＃xff0c;那么再来请求时&＃xff0c;如果发现无可用的Processor&＃xff0c;那么这个请求将被忽略&＃xff0c;直接socket.close()了。

我们来看看HttpConnector的关键代码&＃xff1a;

public class HttpConnector implements Connector, Runnable{public void run(){while (!stopped) {Socket socket &＃61; null;try {socket &＃61; serversocket.accept();}catch(Exception e) {continue;}//从pool中拿一个HttpProcessorHttpProcessor processor &＃61; createProcessor();if(processor &＃61;&＃61; null){socket.close();continue;}//把请求交给HttpProcessor处理processor.assign(socket);}}
}    

Processor也是一个线程&＃xff0c;在诞生之初&＃xff0c;会被启动&＃xff0c;然后就一直等啊等啊等....
&＃xff08;虽然Processor很讨厌等人&＃xff0c;但是没办法&＃xff0c;循环刚执行第一次&＃xff0c;还没执行完&＃xff0c;就得等在那里了。&＃xff09;

private synchronized Socket await() {// 最初available &＃61; falsewhile (!available) {try {wait();} catch (InterruptedException e){;}}// Notify the Connector that we have received thisSocket socket &＃61; this.socket;available &＃61; false;notifyAll();return socket;
}

直到Connector哪一天翻中了它的牌子。

synchronized void assign(Socket socket) {// Connector翻牌子的时候 available还是false 所以不用进入循环去等待while (available) {try {wait();} catch (InterruptedException e) {;}}// Store the newly available Socket and notify our threadthis.socket &＃61; socket;available &＃61; true;//Connector牌子都翻好了&＃xff0c;你得给我办"事"吧,就把还在等啊等啊等...的Processor叫醒啦notifyAll();...
} 看看HttpProcessor被叫醒后做了什么呢&＃xff1a;

public class HttpProcessor implements Runnable{public void run() {while (!stopped) {//等啊等啊等....Socket socket &＃61; await();//被唤醒if (socket &＃61;&＃61; null)continue;try {//处理请求&＃xff0c;解析requestLine、Headers、COOKIE&＃xff0c;生成Request、Responseprocess(socket);} catch (Throwable t) {log("process.invoke", t);}//入栈connector.recycle(this);...}}private void process(Socket socket){...//处理完之后,调用connector的invoke方法,为什么这么做一会会提到connector.invoke(request,response);}
}

Bootstrap启动类

public final class Bootstrap {public static void main(string[] args) {HttpConnector connector &＃61; new HttpConnector();//一个简单的容器&＃xff0c;用来承载ConnectorSimpleContainer container &＃61; new SimpleContainer();connector.setContainer(container);try {connector.initialize();//初始化连接器connector.start();//启动连接器System.in.read();} catch (Exception e) {e.printStackTrace();}}
}

你已经看到了&＃xff0c;我忘了讲SimpleContainer了。
"人"如其名&＃xff1a;简单容器&＃xff0c;每一个容器对应一个Connector。每一个Connector持有一个SimpleContainer的引用。HttpProcessor在process方法的最后会调用Connector中的invoke方法&＃xff0c;Connector又调用SimpleContainer的invoke方法&＃xff0c;看看invoke方法都做了什么&＃xff1a;

public void invoke(Request request,Response response)throws IoException,ServletException {string servletName &＃61; ((Httpservletrequest)request).getRequestURI();servletName &＃61; servletName.substring(servletName.lastIndexof("/") &＃43; 1);URLClassLoader loader &＃61; null;try {URL[] urls &＃61; new URL[1];URLStreamHandler streamHandler &＃61; null;File classpath &＃61; new File(WEB_ROOT);String repository &＃61; (new URL("file",null, classpath.getCanonicalpath() &＃43; File.separator)).toString(); urls[0] &＃61; new URL(null, repository, streamHandler);loader &＃61; new URLClassLoader(urls);} catch (IOException e) {System.out.println(e.toString() );}Class myClass &＃61; null;try {myClass &＃61; loader.loadclass(servletName);} catch (classNotFoundException e) {System.out.println(e.toString());}servlet servlet &＃61; null;try {servlet &＃61; (Servlet) myClass.newInstance();servlet.service((HttpServletRequest)request,(HttpServletResponse)response);     } catch (Exception e) {System.out.println(e.toString());} catch (Throwable e) {System.out.println(e.toString());}
} 看出来了吧&＃xff1f;容器现在负责加载我们请求的servlet&＃xff0c;并调用service方法处理请求啦。所以我们不需要上一节中的ServletProcessor啦。