网上对于struts2请求处理流程的讲解还是比较多的,有的还是非常详细的,所以这里我就简单地将大概流程总结下,有了个大概印象后理解起源码就会有一定的思路了:
struts2的请求处理过程实际上是在初始化中加载的配置及容器的基础上,通过请求的url分析出命名空间、action名称、方法名称,在利用命名空间检索出该命名空间下配置的所有antion,在通过action名称找到具体的action,生成action实例,如果该action上配置了拦截器则依次调用拦截器,之后调用action中方法名称指定的方法,最后通过配置中的result返回相应视图。
版本:struts2-2.1.6 xwork-2.1.2
下面就通过源码进行分析下:
struts2中处理请求是通过过滤器org.apache.struts2.dispatcher.FilterDispatcher的doFilter()方法实现的,如下:
public void doFilter(ServletRequest req, ServletResponse res, FilterChain chain) throws IOException, ServletException { HttpServletRequest request = (HttpServletRequest) req; HttpServletResponse response = (HttpServletResponse) res; ServletContext servletContext = getServletContext(); String timerKey = "FilterDispatcher_doFilter: "; try { // FIXME: this should be refactored better to not duplicate work with the action invocation ValueStack stack = dispatcher.getContainer().getInstance(ValueStackFactory.class).createValueStack(); ActionContext ctx = new ActionContext(stack.getContext()); ActionContext.setContext(ctx); UtilTimerStack.push(timerKey); request = prepareDispatcherAndWrapRequest(request, response); ActionMapping mapping; try { mapping = actionMapper.getMapping(request, dispatcher.getConfigurationManager()); } catch (Exception ex) { log.error("error getting ActionMapping", ex); dispatcher.sendError(request, response, servletContext, HttpServletResponse.SC_INTERNAL_SERVER_ERROR, ex); return; } if (mapping == null) { // there is no action in this request, should we look for a static resource? String resourcePath = RequestUtils.getServletPath(request); if ("".equals(resourcePath) && null != request.getPathInfo()) { resourcePath = request.getPathInfo(); } if (staticResourceLoader.canHandle(resourcePath)) { staticResourceLoader.findStaticResource(resourcePath, request, response); } else { // this is a normal request, let it pass through chain.doFilter(request, response); } // The framework did its job here return; } dispatcher.serviceAction(request, response, servletContext, mapping); } finally { try { ActionContextCleanUp.cleanUp(req); } finally { UtilTimerStack.pop(timerKey); } } }
前2句比较简单,向上转换req、res为标准接口HttpServletRequest、HttpServletResponse形式。req是org.apache.catalina.connector.RequestFacade的实例,RequestFacade实现了接口HttpServletRequest,而HttpServletRequest继承自ServletRequest接口。第3句,获得servletContext即servlet上下文,通过上下文对象可以访问web.xml描述文件的初始化参数。第4句定义timerKey变量,值是将当前过滤器的类名和当前方法名拼接起来,看下下面的代码发现timerKey用于UtilTimerStack.push(timerKey)和UtilTimerStack.pop(timerKey)俩句,其实这俩句并不属于处理流程的功能代码,而是性能代码,主要是监测下doFilter()方法的执行时间,然后通过日志打印出来,所以这俩句可以不用理会。ValueStack stack = dispatcher.getContainer().getInstance(ValueStackFactory.class).createValueStack()句,首先通过dispatcher.getContainer().getInstance(ValueStackFactory.class)从容器中获得type(类型)为ValueStackFactory.class的bean,然后调用该bean的createValueStack()创建一个ValueStack实例。ValueStackFactory.class默认情况下是使用struts-default.xml配置中的 <bean type="com.opensymphony.xwork2.util.ValueStackFactory" name="struts" class="com.opensymphony.xwork2.ognl.OgnlValueStackFactory" />,即通过上面的容器返回的是个com.opensymphony.xwork2.ognl.OgnlValueStackFactory实例,从名字看出其是个ValueStack工厂,通过调用该工厂的createValueStack()方法返回ValueStack实例。
ValueStack就是通常所说的"值栈",系统每次请求时都会创建个新的ValueStack,其中会保存着本次请求处理的所有中间数据,如:请求的action实例、各种servlet内置对象(request、response、session、application等)、请求参数等。F5看下定义:
public interface ValueStack { public static final String VALUE_STACK = "com.opensymphony.xwork2.util.ValueStack.ValueStack"; public static final String REPORT_ERRORS_ON_NO_PROP = "com.opensymphony.xwork2.util.ValueStack.ReportErrorsOnNoProp"; public abstract Map<String, Object> getContext(); public abstract void setDefaultType(Class defaultType); public abstract void setExprOverrides(Map<Object, Object> overrides); public abstract Map<Object, Object> getExprOverrides(); public abstract CompoundRoot getRoot(); public abstract void setValue(String expr, Object value); public abstract void setValue(String expr, Object value, boolean throwExceptionOnFailure); public abstract String findString(String expr); public abstract Object findValue(String expr); public abstract Object findValue(String expr, Class asType); public abstract Object peek(); public abstract Object pop(); public abstract void push(Object o); public abstract void set(String key, Object o); public abstract int size(); }
是个接口,因为ValueStack本身是堆栈,所以我们看到peek()、pop()、push()等堆栈方法都是有的,其中有个最重要的方法getContext()和getRoot(),getContext()返回的是个Map<String, Object>类型,一般请求中的参数、servlet各种内置对象都是存放在这个Map中。getRoot()返回的就是堆栈数据实际存放的地方,peek()、pop()、push()都是基于其操作的。CompoundRoot是个堆栈类,它继承了java.util.ArrayList,并以此为基础实现了peek()、pop()、push()方法,从而可以独立的作为堆栈使用。这样的话ValueStack就不必在单独实现堆栈功能,只需要在内部创建个CompoundRoot实例就可,其peek()、pop()、push()方法直接调用CompoundRoot的相应方法即可,实际上struts2中的ValueStack默认实现类就是这样做的。另外在这个堆栈中保存的最典型的数据就是action实例。有ognl知识的朋友知道,ognl中基于搜索的有俩个重要对象:上下文和根对象,实际上struts2在利用ognl时,是将getContext()获得的对象作为ognl的上下文、getRoot()获得的作为根对象。这就是为什么我们在通过struts2标签访问action中属性数据时不需要加"#",而访问request、response、session中数据时需要加"#"的原因了。因为ognl中访问根对象是不需要加"#"的,而访问上下文是需要加"#"的。为了更好的理解这块,建议大家先学习下ognl的使用方法及特性。为了验证上面的说法,我举个例子,看下struts2的<s:property value=""/>标签是如何使用ValueStack的,以及最后如何调用ognl的。
这里还要事先交代下,在处理流程的后面处理中会将ValueStack(引用)存于俩个地:一个是ActionContext,另一个是通过request.setAttribute(ServletActionContext.STRUTS_VALUESTACK_KEY, valuesStack)将ValueStack存于request中。<s:property value=""/>标签中使用的ValueStack是通过request获得的。下面看下<s:property value=""/>标签的源码,从struts2的标签定义文件struts-tags.tld中查到,该标签对应的类是org.apache.struts2.views.jsp.PropertyTag,如下:
public class PropertyTag extends ComponentTagSupport { private static final long serialVersionUID = 435308349113743852L; private String defaultValue; private String value; private boolean escape = true; private boolean escapeJavaScript = false; public Component getBean(ValueStack stack, HttpServletRequest req, HttpServletResponse res) { return new Property(stack); } protected void populateParams() { super.populateParams(); Property tag = (Property) component; tag.setDefault(defaultValue); tag.setValue(value); tag.setEscape(escape); tag.setEscapeJavaScript(escapeJavaScript); } public void setDefault(String defaultValue) { this.defaultValue = defaultValue; } public void setEscape(boolean escape) { this.escape = escape; } public void setEscapeJavaScript(boolean escapeJavaScript) { this.escapeJavaScript = escapeJavaScript; } public void setValue(String value) { this.value = value; } }
我们要找下标签的doStartTag(),这个方法会在标签被处理前调用。关于标签这块,建议大家先学下jsp自定义标签的使用。
去它的父类ComponentTagSupport中找下:
public abstract class ComponentTagSupport extends StrutsBodyTagSupport { protected Component component; public abstract Component getBean(ValueStack stack, HttpServletRequest req, HttpServletResponse res); public int doEndTag() throws JspException { component.end(pageContext.getOut(), getBody()); component = null; return EVAL_PAGE; } public int doStartTag() throws JspException { component = getBean(getStack(), (HttpServletRequest) pageContext.getRequest(), (HttpServletResponse) pageContext.getResponse()); Container container = Dispatcher.getInstance().getContainer(); container.inject(component); populateParams(); boolean evalBody = component.start(pageContext.getOut()); if (evalBody) { return component.usesBody() ? EVAL_BODY_BUFFERED : EVAL_BODY_INCLUDE; } else { return SKIP_BODY; } } protected void populateParams() { } public Component getComponent() { return component; } }
找到了。doStartTag()方法第1句,调用getBean(),其中有个参数调用了getStack(),实际上这个方法返回的就是我上面说的ValueStack。F5进入:
public class StrutsBodyTagSupport extends BodyTagSupport { private static final long serialVersionUID = -1201668454354226175L; protected ValueStack getStack() { return TagUtils.getStack(pageContext); } . . . //省略 }
该方法是在ComponentTagSupport的父类StrutsBodyTagSupport中。其中只有一句TagUtils.getStack(pageContext),pageContext就是servlet的页面上下文内置对象,F5进入TagUtils.getStack(pageContext):
public static ValueStack getStack(PageContext pageContext) { HttpServletRequest req = (HttpServletRequest) pageContext.getRequest(); ValueStack stack = (ValueStack) req.getAttribute(ServletActionContext.STRUTS_VALUESTACK_KEY); if (stack == null) { HttpServletResponse res = (HttpServletResponse) pageContext.getResponse(); Dispatcher du = Dispatcher.getInstance(); if (du == null) { throw new ConfigurationException("The Struts dispatcher cannot be found. This is usually caused by "+ "using Struts tags without the associated filter. Struts tags are only usable when the request "+ "has passed through its servlet filter, which initializes the Struts dispatcher needed for this tag."); } stack = du.getContainer().getInstance(ValueStackFactory.class).createValueStack(); Map<String, Object> extraContext = du.createContextMap(new RequestMap(req), req.getParameterMap(), new SessionMap(req), new ApplicationMap(pageContext.getServletContext()), req, res, pageContext.getServletContext()); extraContext.put(ServletActionContext.PAGE_CONTEXT, pageContext); stack.getContext().putAll(extraContext); req.setAttribute(ServletActionContext.STRUTS_VALUESTACK_KEY, stack); // also tie this stack/context to the ThreadLocal ActionContext.setContext(new ActionContext(stack.getContext())); } else { // let's make sure that the current page context is in the action context Map<String, Object> context = stack.getContext(); context.put(ServletActionContext.PAGE_CONTEXT, pageContext); AttributeMap attrMap = new AttributeMap(context); context.put("attr", attrMap); } return stack; }
方法第1句,通过页面上下文对象pageContext获得本次请求的request对象,第2句在通过req.getAttribute(ServletActionContext.STRUTS_VALUESTACK_KEY)获得ValueStack,因为事先struts2已经通过request.setAttribute(ServletActionContext.STRUTS_VALUESTACK_KEY, valuesStack)将ValueStack存于request的ServletActionContext.STRUTS_VALUESTACK_KEY属性中,所以通过第2句就可以获得ValueStack,直接return stack返回到ComponentTagSupport类的doStartTag()方法,如下(我在把代码粘下):
public int doStartTag() throws JspException { component = getBean(getStack(), (HttpServletRequest) pageContext.getRequest(), (HttpServletResponse) pageContext.getResponse()); Container container = Dispatcher.getInstance().getContainer(); container.inject(component); populateParams(); boolean evalBody = component.start(pageContext.getOut()); if (evalBody) { return component.usesBody() ? EVAL_BODY_BUFFERED : EVAL_BODY_INCLUDE; } else { return SKIP_BODY; } }
执行完getStack()接着看getBean()方法,在ComponentTagSupport类中可以找到,getBean()方法被定义为抽象方法,所以具体的实现要在其子类PropertyTag中找,F5进入:
public Component getBean(ValueStack stack, HttpServletRequest req, HttpServletResponse res) { return new Property(stack); }
方法只有一句通过new 生成了Property实例,同时将stack作为构造参数传进去,这个stack就是我们上面通过getStack()得到的ValueStack,执行完后程序会重新回到ComponentTagSupport类的doStartTag()方法,我把代码在粘下:
public int doStartTag() throws JspException { component = getBean(getStack(), (HttpServletRequest) pageContext.getRequest(), (HttpServletResponse) pageContext.getResponse()); Container container = Dispatcher.getInstance().getContainer(); container.inject(component); populateParams(); boolean evalBody = component.start(pageContext.getOut()); if (evalBody) { return component.usesBody() ? EVAL_BODY_BUFFERED : EVAL_BODY_INCLUDE; } else { return SKIP_BODY; } }
这样component中实际引用的就是Property实例,第2、3句,调用容器,通过container.inject(component)将Property中需要注入的属性赋值(带有@Inject标注的)。populateParams()句将标签属性添加到component相应属性中,如value、escape。核心在下面这句component.start(pageContext.getOut()),通过ognl访问ValueStack从而获得标签的value值,F5进入Property的start():
public boolean start(Writer writer) { boolean result = super.start(writer); String actualValue = null; if (value == null) { value = "top"; } else { value = stripExpressionIfAltSyntax(value); } // exception: don't call findString(), since we don't want the // expression parsed in this one case. it really // doesn't make sense, in fact. actualValue = (String) getStack().findValue(value, String.class); try { if (actualValue != null) { writer.write(prepare(actualValue)); } else if (defaultValue != null) { writer.write(prepare(defaultValue)); } } catch (IOException e) { LOG.info("Could not print out value '" + value + "'", e); } return result; }
直接看(String) getStack().findValue(value, String.class) 调用getStack()获得ValueStack,这个ValueStack实在Property实例生成时通过构造方法传入的。之后调用ValueStack的findValue()方法, 其中参数value就是<s:property value=""/>标签的value属性值。我们在上面说过ValueStack的实现类默认使用的是com.opensymphony.xwork2.ognl.OgnlValueStack类,F5进入其findValue():
public Object findValue(String expr, Class asType) { try { if (expr == null) { return null; } if ((overrides != null) && overrides.containsKey(expr)) { expr = (String) overrides.get(expr); } Object value = ognlUtil.getValue(expr, context, root, asType); if (value != null) { return value; } else { return findInContext(expr); } } catch (OgnlException e) { return findInContext(expr); } catch (Exception e) { logLookupFailure(expr, e); return findInContext(expr); } finally { ReflectionContextState.clear(context); } }
直接看ognlUtil.getValue(expr, context, root, asType)句,可以看到content、root就是我们上面说的ValueStack中getContext()和getRoot()方法中对应的属性值,我们说过它分别对应ognl中的上下文和根对象。ognlUtil是com.opensymphony.xwork2.ognl.OgnlUtil类的实例,是struts2中用于操作ognl而单独封装的管理类, F5进入ognlUtil.getValue()方法:
public Object getValue(String name, Map<String, Object> context, Object root, Class resultType) throws OgnlException { return Ognl.getValue(compile(name), context, root, resultType); }
实际上,上面的方法只是将name做处理后直接调用ognl的getValue()方法。context作为ognl的上下文,root作为ognl的根对象,name是属性名。此时的root中就存放着当前action的实例。return返回的值就是<s:property value=""/>标签最终所获得的值。说道这熟悉ognl的应该已经明白了。