浅析Tomcat之CoyoteAdapter

luoson1

Adapter连接了Tomcat连接器Connector和容器Container.它的实现类是CoyoteAdapter主要负责的是对请求进行封装,构造Request和Response对象.并将请求转发给Container也就是Servlet容器.
首先它实现了4个主要方法,以接口的形式给予了定义.

public interface Adapter  
{  
public void service(Request req, Response res)throws Exception;  
public boolean event(Request req, Response res, SocketStatus status)throws Exception;  
public boolean asyncDispatch(Request req,Response res, SocketStatus status)throws Exception;  
public void log(Request req, Response res, long time);  
public String getDomain();  
}
 CoyoteAdapter对其一一进行实现,以其service为例

public void service(org.apache.coyote.Request req,  
org.apache.coyote.Response res)  
throws Exception {  
Request request = (Request) req.getNote(ADAPTER_NOTES);  
Response response = (Response) res.getNote(ADAPTER_NOTES);  
if (request == null) {  
// Create objects  
request = connector.createRequest();  
request.setCoyoteRequest(req);  
response = connector.createResponse();  
response.setCoyoteResponse(res);  
// Link objects  
request.setResponse(response);  
response.setRequest(request);  
// Set as notes  
req.setNote(ADAPTER_NOTES, request);  
res.setNote(ADAPTER_NOTES, response);  
// Set query string encoding  
req.getParameters().setQueryStringEncoding  
(connector.getURIEncoding());  
}  
.....  
try {  
// Parse and set Catalina and configuration specific  
// request parameters  
req.getRequestProcessor().setWorkerThreadName(Thread.currentThread().getName());  
boolean postParseSuccess = postParseRequest(req, request, res, response);  
if (postParseSuccess) {  
//check valves if we support async  
request.setAsyncSupported(connector.getService().getContainer().getPipeline().isAsyncSupported());  
// Calling the container  
connector.getService().getContainer().getPipeline().getFirst().invoke(request, response);  
.....  
}  
AsyncContextImpl asyncConImpl = (AsyncContextImpl)request.getAsyncContext();  
if (asyncConImpl != null) {  
async = true;  
} else if (!comet) {  
request.finishRequest();  
response.finishResponse();  
if (postParseSuccess &&  
request.getMappingData().context != null) {  
// Log only if processing was invoked.  
// If postParseRequest() failed, it has already logged it.  
// If context is null this was the start of a comet request  
// that failed and has already been logged.  
((Context) request.getMappingData().context).logAccess(  
request, response,  
System.currentTimeMillis() - req.getStartTime(),  
false);  
}  
req.action(ActionCode.POST_REQUEST , null);  
}  
} catch (IOException e) {  
// Ignore  
} finally {  
req.getRequestProcessor().setWorkerThreadName(null);  
// Recycle the wrapper request and response  
if (!comet && !async) {  
request.recycle();  
response.recycle();  
} else {  
// Clear converters so that the minimum amount of memory  
// is used by this processor  
request.clearEncoders();  
response.clearEncoders();  
}  
}  
}  
    service的代码流程首先是从coyote的request和response得到connector的request和response并且他们之间互相链接.接下来是通过postParseRequest来解析请求的参数,查找context,解析CookieID,session等.然后用connector.getService().getContainer().getPipeline().getFirst().invoke(request, response)调用容器的方法.从调用的方式可以知道,这个Container已经内置在StandardService中.也就是server.xml的Engine节点.最后是结束请求.调用connector的回调方法.
对于log等方法也是调用了容器的方法.然后回调连接器进行返回.具体实现有所差异而已.由此可见连接适配器CoyoteAdapter的作用是连接器Connector和容器Container的桥梁.
 
 
首发于泛泛之辈 - http://www.lihongkun.com/archives/133