stripes 的Intercept Execution

上面的我们看到stripes通过配置文件把框架的大部分功能组件组合在一起，组合在一起，那么怎么有序运行呢？同时不能失去框架的扩展性，不能就顺序执行组件吧，现在的流行的框架都是采用IOC容器来组合这些组件，像structs2采用xwork IOC容器，tapestry5现在采用spring IOC容器。IOC（控制反转）实际上就是拦截器。在servlet Filter中，我们应该知道什么是拦截器吧。Stripes是框架，不能没有扩展性，但stripes又不想弄得太复杂。没有像别的框架借助于别的IOC容器。Stripes实现了一个小的拦截器子系统，6个LifecycleStage在拦截器子系统中执行，同时，stripes又给用户实现拦截器的接口，通过配置，可以在自己的系统中实现功能强大的拦截功能。

Interceptor.java

 public interface Interceptor {
    Resolution intercept(ExecutionContext context) throws Exception;}

 

上面的Interceptor接口的代码，我们可以看出它使用了ExecutionContext，什么是ExecutionContext就是整个stripes执行时上下文，也就是说在stripes框架在执行时，只要调用这个上下文，就可以得整个生命周斯所需要的东西。

ExecutionContext.java
public class ExecutionContext {
    private static final Log log = Log.getInstance(ExecutionContext.class);

    private Collection<Interceptor> interceptors;
    private Iterator<Interceptor> iterator;
    private Interceptor target;
    private ActionBeanContext actionBeanContext;
    private ActionBean actionBean;
    private Method handler;
    private Resolution resolution;
    private LifecycleStage lifecycleStage;

    。。。。。
    public Resolution wrap(Interceptor target) throws Exception {
        this.target = target;
        this.iterator = null;
        return proceed();
    }    
    public Resolution proceed() throws Exception {
        if (this.iterator == null) {
            log.debug("Transitioning to lifecycle stage ", lifecycleStage);
            this.iterator = this.interceptors.iterator();
        }
        if (this.iterator.hasNext()) {
            return this.iterator.next().intercept(this);
        }
        else {
            return this.target.intercept(this);
        }
    }
}

 

省去了set,get方法，ExecutionContext就只有两个功能方法。proceed()方法就像servlet filter中proceed()filterChain.doFilter(request, response);运行接下来的过滤器一样，运行下一个拦截器。，ExecutionContext有8个属性。Interceptors是指本生命周期阶段的所有要执行的Interceptor，iterator是指Interceptors集合中某一个，因为采用iterator，所以Interceptor集合中的Interceptor执行是不保证顺序的。也就是说同一个阶段的拦截器不能形成依赖关系。Target是目标Interceptor，每一个阶段都一定执行的Interceptor。Resolution是解析器，就像structs中ActionForward一样。lifecycleStage就是前面讲的6个生命阶段。 我们可以找一个例子：

DispatcherHelper.java
public static Resolution resolveHandler (final ExecutionContext ctx) throws Exception {
        final Configuration config = StripesFilter.getConfiguration();
        ctx.setLifecycleStage(LifecycleStage.HandlerResolution);
        ctx.setInterceptors(config.getInterceptors(LifecycleStage.HandlerResolution));

        return ctx.wrap( new Interceptor() {
            public Resolution intercept(ExecutionContext ctx) throws Exception {
                ActionBean bean = ctx.getActionBean();
                ActionBeanContext context = ctx.getActionBeanContext();
                ActionResolver resolver = config.getActionResolver();

                // Then lookup the event name and handler method etc.
                String eventName = resolver.getEventName(bean.getClass(), context);
                context.setEventName(eventName);

                final Method handler;
                if (eventName != null) {
                    handler = resolver.getHandler(bean.getClass(), eventName);
                }
                else {
                    handler = resolver.getDefaultHandler(bean.getClass());
                    if (handler != null) {
                        context.setEventName(resolver.getHandledEvent(handler));
                    }
                }

                // Insist that we have a handler
                if (handler == null) {
                    throw new StripesServletException(
                            "No handler method found for request with  ActionBean [" +
                            bean.getClass().getName() + "] and eventName [ " + eventName + "]");
                }

                log.debug("Resolved event: ", context.getEventName(), "; will invoke: ",
                          bean.getClass().getSimpleName(), ".", handler.getName(), "()");

                ctx.setHandler(handler);
                return null;
            }
        });
    }

 

这个方面是在生命周期的第二个阶段执行。首先给ExecutionContext设定了所处的阶段，接着去找这个阶段的Interceptors.怎么找呢？它调用了DefaultConfiguration.java中

 

public Collection<Interceptor> getInterceptors(LifecycleStage stage) {
        Collection<Interceptor> interceptors = this.interceptors.get(stage);
        if (interceptors == null) {
            interceptors = Collections.emptyList();
        }
        return interceptors;
}

 

方法。在这个方法从本实现的所有interceptors取到属性本阶段的。Interceptors是在DefaultConfiguration的init()方法中：

this.interceptors = initInterceptors();
            if (this.interceptors == null) {
                this.interceptors = new HashMap<LifecycleStage, Collection<Interceptor>>();
                Class<? extends Interceptor> bam = BeforeAfterMethodInterceptor.class;
                BeforeAfterMethodInterceptor interceptor = new BeforeAfterMethodInterceptor();
                for (LifecycleStage stage : bam.getAnnotation(Intercepts.class).value()) {
                    Collection<Interceptor> instances = new ArrayList<Interceptor>();
                    instances.add(interceptor);
                    this.interceptors.put(stage, instances);
                }

 

还记得前面讲过DefaultConfiguration是模版类吗？initInterceptors()在本类是默认的return null实现，推迟到子类实现。看看RuntimeConfiguration.java中@Override 

protected Map<LifecycleStage, Collection<Interceptor>> initInterceptors() {
        String classList = getBootstrapPropertyResolver().getProperty(INTERCEPTOR_LIST);
        if (classList == null) {
            return null;
        }
        else {
            String[] classNames = StringUtil.standardSplit(classList);
            Map<LifecycleStage, Collection<Interceptor>> map =
                    new HashMap<LifecycleStage, Collection<Interceptor>>();

            for (String className : classNames) {
                try {
                    Class<? extends Interceptor> type = ReflectUtil.findClass(className.trim());
                    Intercepts intercepts = type.getAnnotation(Intercepts.class);
                    if (intercepts == null) {
                        log.error("An interceptor of type ", type.getName(), " was configured ",
                                  "but was not marked with an @Intercepts annotation. As a ",
                                  "result it is not possible to determine at which ",
                                  "lifecycle stages the interceprot should be applied. This ",
                                  "interceptor will be ignored.");
                    }
                    else {
                        log.debug("Configuring interceptor '", type.getSimpleName(),
                                  "', for lifecycle stages: ", intercepts.value());
                    }

                    // Instantiate it and optionally call init() if the interceptor
                    // implements ConfigurableComponent
                    Interceptor interceptor = type.newInstance();
                    if (interceptor instanceof ConfigurableComponent) {
                        ((ConfigurableComponent) interceptor).init(this);
                    }

                    for (LifecycleStage stage : intercepts.value()) {
                        Collection<Interceptor> stack = map.get(stage);
                        if (stack == null) {
                            stack = new LinkedList<Interceptor>();
                            map.put(stage, stack);
                        }

                        stack.add(interceptor);
                    }
                }
                catch (Exception e) {
                    throw new StripesRuntimeException(
                            "Could not instantiate one or more configured Interceptors. The " +
                            "property '" + INTERCEPTOR_LIST + "' contained [" + classList +
                            "]. This value must contain fully qualified class names separated " +
                            "by commas.", e);
                }
            }

            return map;
        }
}

 

这个方法取到配置中的拦截器s,并实例化它们，之后按生命周期分成6类放在renturn中的Map<LifecycleStage, Collection<Interceptor>>中。现在回到resolveHandler方法，这个方法接下来就是return ctx.wrap( new Interceptor() {..});这个实现了一个匿名的Interceptor，之后执行ExcuteContext.wrap(Interceptor)方法。还记得wrap（interceptor）,它是一个包扎方法，把这个匿名Interceptor保存ExcuteContext   的target对象。this.target = target;接下的这句 this.iterator = null;仅仅是代码的实现技巧，这个是要在下面的 return proceed();做判断处理。

 

public Resolution proceed() throws Exception {
        if (this.iterator == null) {
         log.debug("Transitioning to lifecycle stage ", lifecycleStage);
            this.iterator = this.interceptors.iterator();        }
        if (this.iterator.hasNext()) {
            return this.iterator.next().intercept(this);        }
        else {      return this.target.intercept(this);       }    }
 

 

this.iterator = null的作用仅仅是为了找到这一阶段的interceptors。如果找到了就是执行这一阶段的interceptors中的iterator。在这里看到，每一个阶段只能执行一个interceptors的第一个。那剩下的呢怎么办？剩下就是用户实现的拦截器的传递了。如果用户的实现的拦截器没用传递。那之后就不执行，连本阶段的匿名Interceptor都不会执行就直接返回。返回的是Resolution不为null,那个整个流程就结束，剩下的执行Resolution的跳起页面功能（像structs的 ActionForward,比它的功能强）。用户实现的拦截如下：

@Intercepts({LifecycleStage.ActionBeanResolution})
public class HibernateInterceptor implements Interceptor
{	private static Log log = Log.getInstance(HibernateInterceptor.class);private boolean initializedHibernateProvider = false;
public Resolution intercept(ExecutionContext context) throws Exception
	{log.debug("HibernateInterceptor called");
if (!initializedHibernateProvider && (HibernateProvider.getInstance() == null) && (HibernateFilter.getCurrentInstance() != null))
		{log.info("Initializing HibernateProvider");
HibernateProvider.setInstance(HibernateFilter.getCurrentInstance());
			initializedHibernateProvider = true;		}
		return context.proceed();	}
看到最后一句的：return context.proceed();从现在开始iterator都不为空。那么就找下一个Interceptor,没有的话，就执行本阶段的目标this.target.intercept(this);这个是在本阶段最后执行的。
如果用户采用如下的话：
@Intercepts({LifecycleStage.HandlerResolution})
public class NoisyInterceptor implements Interceptor {
public Resolution intercept(ExecutionContext ctx) throws Exception {
System.out.println("Before " + ctx.getLifecycleStage());
Resolution resolution = ctx.proceed();
System.out.println("After " + ctx.getLifecycleStage());
return resolution}}

 

那就是一个嵌套的执行，最后还会返回来执行Resolution resolution = ctx.proceed();下面的语句。ctx.proceed()上面的语句是在目标Interceptor之前执行。下面的语句是在目标Interceptor之后执行。Stripes有一个默认的@before和@after就是一个在前面一个后面的实现。分析完了，是不是自己也想写一个interceptor。写interceptor很简单，只要实现implements Interceptor，也就是实现了public Resolution intercept(ExecutionContext ctx) throws Exception就可以。还有一点不能忘记，这个interceptor是在那个阶段执行的，如@Intercepts({LifecycleStage.HandlerResolution})就是在HandlerResolution执行。 

分享到：

stripes actionresolve (一)

• 2008-02-18 17:30
• 浏览 1817
• 评论(0)
• 论坛回复 / 浏览 (0 / 2355)
• 查看更多