Spring源码入门——AnnotationBeanNameGenerator解析
---恢复内容开始---
接上篇,上篇解析了DefaultBeanGenerator生成bean name的过程(http://www.cnblogs.com/jason0529/p/5272265.html ), 本篇我们继续解析另一类bean name生成方式。
spring定义bean有两种模式,配置文件(xml,properties)和注解。注:jpa的声明接口生成bean应该可以算第三种模式,这里不讨论。
对两种bean定义方式,spring提供了两种不同的bean name实现方式去实现不同的模式。AnnotationBeanNameGenerator能够处理 Component,Respository,Service,Controller这四个常用的注解,解析为bean name注给他们对应的value属性。另外jee的javax.annotation.ManagedBean和javax.inject.Named也可以支持。
当Component,Respository,Service,Controller注解的value树形没有自定义时,会根据类的名称生成一个短的bean name。例如: com.xyz.FooServiceImpl -> fooServiceImpl
入口肯定是BeanNameGenerator接口声明的generateBeanName(BeanDefinition, BeanDefinitionRegistry) 方法,该方法做了一个分类判断,处理AnnotationBeanDefinition和default两种方式的。
public String generateBeanName(BeanDefinition definition, BeanDefinitionRegistry registry) { //判断是否是否是AnnotatedBeanDefinition的子类, AnnotatedBeanDefinition是BeanDefinition的一个子类 //如果是AnnotatedBeanDefinition , 按照注解生成模式生成信息,否则生成默认的bean name if (definition instanceof AnnotatedBeanDefinition) { String beanName = determineBeanNameFromAnnotation((AnnotatedBeanDefinition) definition); //保证生成的bean name 非空 if (StringUtils.hasText(beanName)) { // Explicit bean name found. return beanName; } } // Fallback: generate a unique default bean name. return buildDefaultBeanName(definition, registry); }
先从相对简单的default看起,这段代码的疑点是生成的bean name并没有和DefaultBeanNameGenerator一样做唯一性校验,可能导致不同包下面存在相同的类名时,会产生两个name一样的bean,引发spring 异常。
/** * Derive a default bean name from the given bean definition. * <p>The default implementation delegates to {@link #buildDefaultBeanName(BeanDefinition)}. * @param definition the bean definition to build a bean name for * @param registry the registry that the given bean definition is being registered with * @return the default bean name (never {@code null}) */ protected String buildDefaultBeanName(BeanDefinition definition, BeanDefinitionRegistry registry) {
//没有做 name 唯一性校验。 return buildDefaultBeanName(definition); } /** * Derive a default bean name from the given bean definition. * <p>The default implementation simply builds a decapitalized version * of the short class name: e.g. "mypackage.MyJdbcDao" -> "myJdbcDao". * <p>Note that inner classes will thus have names of the form * "outerClassName.innerClassName", which because of the period in the * name may be an issue if you are autowiring by name. * @param definition the bean definition to build a bean name for * @return the default bean name (never {@code null}) */ protected String buildDefaultBeanName(BeanDefinition definition) {
//具体类名获取和格式化先不做具体讨论 String shortClassName = ClassUtils.getShortName(definition.getBeanClassName()); return Introspector.decapitalize(shortClassName); }
---恢复内容结束---
接上篇,上篇解析了DefaultBeanGenerator生成bean name的过程(http://www.cnblogs.com/jason0529/p/5272265.html ), 本篇我们继续解析另一类bean name生成方式。
spring定义bean有两种模式,配置文件(xml,properties)和注解。注:jpa的声明接口生成bean应该可以算第三种模式,这里不讨论。
对两种bean定义方式,spring提供了两种不同的bean name实现方式去实现不同的模式。AnnotationBeanNameGenerator能够处理 Component,Respository,Service,Controller这四个常用的注解,解析为bean name注给他们对应的value属性。另外jee的javax.annotation.ManagedBean和javax.inject.Named也可以支持。
当Component,Respository,Service,Controller注解的value树形没有自定义时,会根据类的名称生成一个短的bean name。例如: com.xyz.FooServiceImpl -> fooServiceImpl
入口肯定是BeanNameGenerator接口声明的generateBeanName(BeanDefinition, BeanDefinitionRegistry) 方法,该方法做了一个分类判断,处理AnnotationBeanDefinition和default两种方式的。
public String generateBeanName(BeanDefinition definition, BeanDefinitionRegistry registry) { //判断是否是否是AnnotatedBeanDefinition的子类, AnnotatedBeanDefinition是BeanDefinition的一个子类 //如果是AnnotatedBeanDefinition , 按照注解生成模式生成信息,否则生成默认的bean name if (definition instanceof AnnotatedBeanDefinition) { String beanName = determineBeanNameFromAnnotation((AnnotatedBeanDefinition) definition); //保证生成的bean name 非空 if (StringUtils.hasText(beanName)) { // Explicit bean name found. return beanName; } } // Fallback: generate a unique default bean name. return buildDefaultBeanName(definition, registry); }
先从相对简单的default看起,这段代码的疑点是生成的bean name并没有和DefaultBeanNameGenerator一样做唯一性校验,可能导致不同包下面存在相同的类名时,会产生两个name一样的bean,引发spring 异常。
/** * Derive a default bean name from the given bean definition. * <p>The default implementation delegates to {@link #buildDefaultBeanName(BeanDefinition)}. * @param definition the bean definition to build a bean name for * @param registry the registry that the given bean definition is being registered with * @return the default bean name (never {@code null}) */ protected String buildDefaultBeanName(BeanDefinition definition, BeanDefinitionRegistry registry) {
//没有做 name 唯一性校验。 return buildDefaultBeanName(definition); } /** * Derive a default bean name from the given bean definition. * <p>The default implementation simply builds a decapitalized version * of the short class name: e.g. "mypackage.MyJdbcDao" -> "myJdbcDao". * <p>Note that inner classes will thus have names of the form * "outerClassName.innerClassName", which because of the period in the * name may be an issue if you are autowiring by name. * @param definition the bean definition to build a bean name for * @return the default bean name (never {@code null}) */ protected String buildDefaultBeanName(BeanDefinition definition) {
//具体类名获取和格式化先不做具体讨论 String shortClassName = ClassUtils.getShortName(definition.getBeanClassName()); return Introspector.decapitalize(shortClassName); }
其实ClassUtils.getShortName也很简单,根据传入字符串获取一个具体类名称,不含包路径,考虑cglib代理的类,做了一个特殊处理。
/** * Get the class name without the qualified package name. * @param className the className to get the short name for * @return the class name of the class without the package name * @throws IllegalArgumentException if the className is empty */ public static String getShortName(String className) { Assert.hasLength(className, "Class name must not be empty"); int lastDotIndex = className.lastIndexOf(PACKAGE_SEPARATOR); int nameEndIndex = className.indexOf(CGLIB_CLASS_SEPARATOR); if (nameEndIndex == -1) { nameEndIndex = className.length(); } String shortName = className.substring(lastDotIndex + 1, nameEndIndex); shortName = shortName.replace(INNER_CLASS_SEPARATOR, PACKAGE_SEPARATOR); return shortName; }
接下来看第二条线,AnnotationBeanDefinition的beanName如何生成,具体处理委托给了determineBeanNameFromAnnotation(AnnotatedBeanDefinition)方法完成,该方法对该类的所有注解进行了遍历,满足三个条件:注解名称在待解析列表,存在value属性且非空,当且只有一个注解正确配置value属性。
/** * Derive a bean name from one of the annotations on the class. * 从类的注解中包含value属性的注解生成一个bean name * @param annotatedDef the annotation-aware bean definition * @return the bean name, or {@code null} if none is found */ protected String determineBeanNameFromAnnotation(AnnotatedBeanDefinition annotatedDef) { //获取注解类元信息 AnnotationMetadata amd = annotatedDef.getMetadata(); //一个类存在多个注解,故类型为集合 Set<String> types = amd.getAnnotationTypes(); String beanName = null; for (String type : types) { //获取该类型对应的属性 AnnotationAttributes attributes = MetadataUtils.attributesFor(amd, type); //判断注解类型是否包含value属性 if (isStereotypeWithNameValue(type, amd.getMetaAnnotationTypes(type), attributes)) { String value = (String) attributes.get("value"); if (StringUtils.hasLength(value)) { //不多于1个注解配置了value属性且非空,比如无法在一个类上面同时使用Component和Sevice注解同时配置beanName值 if (beanName != null && !value.equals(beanName)) { throw new IllegalStateException("Stereotype annotations suggest inconsistent " + "component names: '" + beanName + "' versus '" + value + "'"); } beanName = value; } } } return beanName; }
这个方法里面有两个关键的处理流程,第一步,读取对应annotationType对应的所有属性。
public static AnnotationAttributes attributesFor(AnnotationMetadata metadata, String annoClassName) { return AnnotationAttributes.fromMap(metadata.getAnnotationAttributes(annoClassName, false)); }
//AnnotationMetadata.java /** * Retrieve the attributes of the annotation of the given type, * if any (i.e. if defined on the underlying class, as direct * annotation or as meta-annotation). * @param annotationType the annotation type to look for * @param classValuesAsString whether to convert class references to String * class names for exposure as values in the returned Map, instead of Class * references which might potentially have to be loaded first * @return a Map of attributes, with the attribute name as key (e.g. "value") * and the defined attribute value as Map value. This return value will be * {@code null} if no matching annotation is defined. */ Map<String, Object> getAnnotationAttributes(String annotationType, boolean classValuesAsString);
第二步,判断annotationType是否具有value属性,但是metaAnnotationTypes.contains(COMPONENT_ANNOTATION_CLASSNAME)这部分并不是很懂?注解无法继承,为啥要多判断一次?
private static final String COMPONENT_ANNOTATION_CLASSNAME = "org.springframework.stereotype.Component"; /** * Check whether the given annotation is a stereotype that is allowed * to suggest a component name through its annotation {@code value()}. * @param annotationType the name of the annotation class to check * @param metaAnnotationTypes the names of meta-annotations on the given annotation * @param attributes the map of attributes for the given annotation * @return whether the annotation qualifies as a stereotype with component name */ protected boolean isStereotypeWithNameValue(String annotationType,Set<String> metaAnnotationTypes, Map<String, Object> attributes) { boolean isStereotype = annotationType.equals(COMPONENT_ANNOTATION_CLASSNAME) || (metaAnnotationTypes != null && metaAnnotationTypes.contains(COMPONENT_ANNOTATION_CLASSNAME)) || annotationType.equals("javax.annotation.ManagedBean") || annotationType.equals("javax.inject.Named"); return (isStereotype && attributes != null && attributes.containsKey("value")); }
整理下思路:
生成bean name有两条处理线,使用AnnotationBeanDefinition注解和不使用的。
不使用AnnotationBeanDefinition注解的,直接将类名(不含包名)改为驼峰形式作为bean name。
使用AnnotationBeanDefinition注解的:
1,读取所有注解类型
2,便利所有注解类型,找到所有为Component、Service,Respository,Controller含有非空value属性的注解
3,不多于一个个有效配置时生效,大于一个会抛出异常。(spring无法明确具体哪个生效)