关于后端:Spring-源码解析四Bean-的构造装载属性值解析注解扫描

33次阅读

共计 23268 个字符，预计需要花费 59 分钟才能阅读完成。

Spring 源码解析四：Bean 的结构装载、属性值解析、注解扫描

上一篇次要介绍了 Bean 的注册、解析、实例化，但留下一些点待解析：

  • ConstructorResolver.autowireConstructor 如何进行结构装载并实例化的
  • CglibSubclassingInstantiationStrategy.instantiate 如何动静实例化 bean 的
  • BeanDefinitionValueResolver.resolveValueIfNecessary 如何解析属性值的
  • MergedAnnotations.from 如何扫描注解的

这一节，就来解析这几个点

1. ConstructorResolver.autowireConstructor

ConstructorResolver
的次要性能是在结构时载入依赖的 bean 和用 factory-method 实例化 bean，这里只讲第一个：在结构时载入依赖的 bean ConstructorResolver.autowireConstructor

class ConstructorResolver {
    public BeanWrapper autowireConstructor(String beanName, RootBeanDefinition mbd,
            @Nullable Constructor<?>[] chosenCtors, @Nullable Object[] explicitArgs) {
        // 实例化一个封装对象
        BeanWrapperImpl bw = new BeanWrapperImpl();

        // 须要应用的构造方法
        Constructor<?> constructorToUse = null;
        // 须要应用的结构参数
        Object[] argsToUse = null;

        // 如果指定了参数，就应用指定的参数
        if (explicitArgs != null) {argsToUse = explicitArgs;}
        else {Object[] argsToResolve = null;
            // 并发解决
            synchronized (mbd.constructorArgumentLock) {
                // 获取定义中已解析的构造方法
                constructorToUse = (Constructor<?>) mbd.resolvedConstructorOrFactoryMethod;
                if (constructorToUse != null && mbd.constructorArgumentsResolved) {
                    // 获取定义中已解析的参数，首先是 resolvedConstructorArguments，其次是 preparedConstructorArguments
                    argsToUse = mbd.resolvedConstructorArguments;
                    if (argsToUse == null) {argsToResolve = mbd.preparedConstructorArguments;}
                }
            }
            if (argsToResolve != null) {
                // 把参数解析成理论要用的数据，如 bean 拆卸、属性载入
                argsToUse = resolvePreparedArguments(beanName, mbd, bw, constructorToUse, argsToResolve);
            }
        }

        // 如果没有指定的构造方法或参数
        if (constructorToUse == null || argsToUse == null) {Constructor<?>[] candidates = chosenCtors;
            if (candidates == null) {
                // 如果传入的 chosenCtors 是空，则获取 beanClass 的构造方法
                Class<?> beanClass = mbd.getBeanClass();
                try {candidates = (mbd.isNonPublicAccessAllowed() ?
                            beanClass.getDeclaredConstructors() : beanClass.getConstructors());
                }
                catch (Throwable ex) {// ... 代码省略}
            }

            // 如果只有一个构造方法，并且没有指定 explicitArgs 和 constructorArgumentValues
            if (candidates.length == 1 && explicitArgs == null && !mbd.hasConstructorArgumentValues()) {Constructor<?> uniqueCandidate = candidates[0];
                // 如果构造方法没有参数须要拆卸，间接实例化返回
                if (uniqueCandidate.getParameterCount() == 0) {synchronized (mbd.constructorArgumentLock) {
                        mbd.resolvedConstructorOrFactoryMethod = uniqueCandidate;
                        mbd.constructorArgumentsResolved = true;
                        mbd.resolvedConstructorArguments = EMPTY_ARGS;
                    }
                    // 实例化 bean
                    bw.setBeanInstance(instantiate(beanName, mbd, uniqueCandidate, EMPTY_ARGS));
                    return bw;
                }
            }

            // 有构造方法，且结构模式为 AUTOWIRE_CONSTRUCTOR，则须要主动拆卸 bean
            boolean autowiring = (chosenCtors != null ||
                    mbd.getResolvedAutowireMode() == AutowireCapableBeanFactory.AUTOWIRE_CONSTRUCTOR);
            ConstructorArgumentValues resolvedValues = null;

            // ... 代码省略


            // 抉择一个参数数量、参数类型、参数程序与须要实例化参数对象最靠近的一个构造方法
            // 这里的计算比较复杂，有趣味的敌人能够自行摸索一下
            for (Constructor<?> candidate : candidates) {// ... 代码省略}

            // ... 代码省略
        }

        // 实例化 bean
        bw.setBeanInstance(instantiate(beanName, mbd, constructorToUse, argsToUse));
        return bw;
    }

    // 实例化 bean，默认应用 CglibSubclassingInstantiationStrategy.instantiate 来实例化
    private Object instantiate(String beanName, RootBeanDefinition mbd, Constructor<?> constructorToUse, Object[] argsToUse) {

        try {InstantiationStrategy strategy = this.beanFactory.getInstantiationStrategy();
            if (System.getSecurityManager() != null) {return AccessController.doPrivileged((PrivilegedAction<Object>) () ->
                        strategy.instantiate(mbd, beanName, this.beanFactory, constructorToUse, argsToUse),
                        this.beanFactory.getAccessControlContext());
            }
            else {return strategy.instantiate(mbd, beanName, this.beanFactory, constructorToUse, argsToUse);
            }
        }
        catch (Throwable ex) {// ... 代码省略}
    }
}

最终的实例化依然须要 CglibSubclassingInstantiationStrategy.instantiate 来实现

2. CglibSubclassingInstantiationStrategy.instantiate

CglibSubclassingInstantiationStrategy
的次要性能是应用 cglib 的形式实例化 bean

public class CglibSubclassingInstantiationStrategy extends SimpleInstantiationStrategy {}

咱们先来看看 SimpleInstantiationStrategy

public class SimpleInstantiationStrategy implements InstantiationStrategy {
    // 用于调用 factory-method
    private static final ThreadLocal<Method> currentlyInvokedFactoryMethod = new ThreadLocal<>();

    @Override
    public Object instantiate(RootBeanDefinition bd, @Nullable String beanName, BeanFactory owner) {
        // 没有办法须要重载
        if (!bd.hasMethodOverrides()) {
            Constructor<?> constructorToUse;
            synchronized (bd.constructorArgumentLock) {
                // 首先获取已解析的构造方法
                constructorToUse = (Constructor<?>) bd.resolvedConstructorOrFactoryMethod;
                if (constructorToUse == null) {
                    // 没有已解析的构造方法，则用 beanClass 获取
                    final Class<?> clazz = bd.getBeanClass();
                    // 接口不能实例化
                    if (clazz.isInterface()) {throw new BeanInstantiationException(clazz, "Specified class is an interface");
                    }
                    try {if (System.getSecurityManager() != null) {
                            constructorToUse = AccessController.doPrivileged((PrivilegedExceptionAction<Constructor<?>>) clazz::getDeclaredConstructor);
                        }
                        else {
                            // 通过 Class 获取构造方法
                            constructorToUse = clazz.getDeclaredConstructor();}
                        // 设为已解析的构造方法
                        bd.resolvedConstructorOrFactoryMethod = constructorToUse;
                    }
                    catch (Throwable ex) {// 如果 Class 没有构造方法，报错}
                }
            }
            // 调用 Constructor..newInstance 实例化
            return BeanUtils.instantiateClass(constructorToUse);
        }
        else {
            // 有办法须要重载，则须要用 cglib 生成一个新的子类
            return instantiateWithMethodInjection(bd, beanName, owner);
        }
    }

    // 办法注入实例化，留给子类实现
    protected Object instantiateWithMethodInjection(RootBeanDefinition bd, @Nullable String beanName, BeanFactory owner) {throw new UnsupportedOperationException("Method Injection not supported in SimpleInstantiationStrategy");
    }

    @Override
    public Object instantiate(RootBeanDefinition bd, @Nullable String beanName, BeanFactory owner,
            final Constructor<?> ctor, Object... args) {
        // 没有办法须要重载
        if (!bd.hasMethodOverrides()) {
            // ... 代码省略

            // 调用 Constructor.newInstance 实例化
            return BeanUtils.instantiateClass(ctor, args);
        }
        else {
            // 有办法须要重载，则须要用 cglib 生成一个新的子类
            return instantiateWithMethodInjection(bd, beanName, owner, ctor, args);
        }
    }

    // 办法注入实例化，留给子类实现
    protected Object instantiateWithMethodInjection(RootBeanDefinition bd, @Nullable String beanName,
            BeanFactory owner, @Nullable Constructor<?> ctor, Object... args) {throw new UnsupportedOperationException("Method Injection not supported in SimpleInstantiationStrategy");
    }

    // 通过 factory-method 初始化
    @Override
    public Object instantiate(RootBeanDefinition bd, @Nullable String beanName, BeanFactory owner,
            @Nullable Object factoryBean, final Method factoryMethod, Object... args) {

        try {
            // ... 代码省略

            // 间接调用 Method.invoke 实例化
            Object result = factoryMethod.invoke(factoryBean, args);
            // 如果为 null，实例化一个 NullBean
            if (result == null) {result = new NullBean();
            }
            return result;
        }
        catch (IllegalArgumentException ex) {// ... 代码省略}
        catch (IllegalAccessException ex) {// ... 代码省略}
        catch (InvocationTargetException ex) {// ... 代码省略}
    }
}

咱们先来看看 CglibSubclassingInstantiationStrategy

public class CglibSubclassingInstantiationStrategy extends SimpleInstantiationStrategy {
    @Override
    protected Object instantiateWithMethodInjection(RootBeanDefinition bd, @Nullable String beanName, BeanFactory owner) {return instantiateWithMethodInjection(bd, beanName, owner, null);
    }

    @Override
    protected Object instantiateWithMethodInjection(RootBeanDefinition bd, @Nullable String beanName, BeanFactory owner,
            @Nullable Constructor<?> ctor, Object... args) {
        // 生成一个 CglibSubclassCreator 来实例化
        return new CglibSubclassCreator(bd, owner).instantiate(ctor, args);
    }
}

CglibSubclassCreator是 CglibSubclassingInstantiationStrategy 的外部类

private static class CglibSubclassCreator {
    // 实例化一个构造函数
    public Object instantiate(@Nullable Constructor<?> ctor, Object... args) {
        // 应用 cglib Enhancer 创立一个增强子类
        Class<?> subclass = createEnhancedSubclass(this.beanDefinition);
        Object instance;
        if (ctor == null) {
            // 如果没有指定构造函数，调用 Constructor.newInstance 实例化
            instance = BeanUtils.instantiateClass(subclass);
        }
        else {
            try {
                // 获取与 ctor 参数匹配的子类构造函数，再调用 Constructor.newInstance 实例化
                Constructor<?> enhancedSubclassConstructor = subclass.getConstructor(ctor.getParameterTypes());
                instance = enhancedSubclassConstructor.newInstance(args);
            }
            catch (Exception ex) {// ... 代码省略}
        }

        // ... 代码省略

        return instance;
    }

    // 应用 cglib Enhancer 创立一个增强子类
    private Class<?> createEnhancedSubclass(RootBeanDefinition beanDefinition) {Enhancer enhancer = new Enhancer();
        enhancer.setSuperclass(beanDefinition.getBeanClass());
        enhancer.setNamingPolicy(SpringNamingPolicy.INSTANCE);
        if (this.owner instanceof ConfigurableBeanFactory) {ClassLoader cl = ((ConfigurableBeanFactory) this.owner).getBeanClassLoader();
            enhancer.setStrategy(new ClassLoaderAwareGeneratorStrategy(cl));
        }
        enhancer.setCallbackFilter(new MethodOverrideCallbackFilter(beanDefinition));
        enhancer.setCallbackTypes(CALLBACK_TYPES);
        return enhancer.createClass();}
}

3. BeanDefinitionValueResolver.resolveValueIfNecessary

BeanDefinitionValueResolver
的次要性能是将配置的属性值装载到 bean 中

class BeanDefinitionValueResolver {public Object resolveValueIfNecessary(Object argName, @Nullable Object value) {
        // 在运行时对其余 bean 的援用
        if (value instanceof RuntimeBeanReference) {RuntimeBeanReference ref = (RuntimeBeanReference) value;
            // 解析援用 bean
            return resolveReference(argName, ref);
        }
        // 在运行时对其余 bean name 的援用
        else if (value instanceof RuntimeBeanNameReference) {String refName = ((RuntimeBeanNameReference) value).getBeanName();
            // 应用 doEvaluate 解析 bean 名字，执行 SpEL 表达式
            refName = String.valueOf(doEvaluate(refName));
            if (!this.beanFactory.containsBean(refName)) {// 没注册过 bean name，报错}
            return refName;
        }
        // 蕴含别名的 bean 定义
        else if (value instanceof BeanDefinitionHolder) {BeanDefinitionHolder bdHolder = (BeanDefinitionHolder) value;
            // 解析外部 bean
            return resolveInnerBean(argName, bdHolder.getBeanName(), bdHolder.getBeanDefinition());
        }
        // bean 定义
        else if (value instanceof BeanDefinition) {BeanDefinition bd = (BeanDefinition) value;
            // 外部 bean 名字
            String innerBeanName = "(inner bean)" + BeanFactoryUtils.GENERATED_BEAN_NAME_SEPARATOR +
                    ObjectUtils.getIdentityHexString(bd);
            // 解析外部 bean
            return resolveInnerBean(argName, innerBeanName, bd);
        }
        // 注入依赖形容(类、办法、属性)
        else if (value instanceof DependencyDescriptor) {Set<String> autowiredBeanNames = new LinkedHashSet<>(4);
            // 解析依赖
            Object result = this.beanFactory.resolveDependency((DependencyDescriptor) value, this.beanName, autowiredBeanNames, this.typeConverter);
            for (String autowiredBeanName : autowiredBeanNames) {if (this.beanFactory.containsBean(autowiredBeanName)) {
                    // 注册依赖
                    this.beanFactory.registerDependentBean(autowiredBeanName, this.beanName);
                }
            }
            return result;
        }

        // ... 代码省略

        // 是字符类型的值
        else if (value instanceof TypedStringValue) {TypedStringValue typedStringValue = (TypedStringValue) value;
            // 应用 evaluate 解析 bean 名字，执行 SpEL 表达式
            Object valueObject = evaluate(typedStringValue);
            try {
                // 解析类型
                Class<?> resolvedTargetType = resolveTargetType(typedStringValue);
                // 如果有类型，把 valueObject 转换成这个类型
                if (resolvedTargetType != null) {return this.typeConverter.convertIfNecessary(valueObject, resolvedTargetType);
                }
                // 没有指定类型，则返回本来的
                else {return valueObject;}
            }
            catch (Throwable ex) {// ... 代码省略}
        }
        // 如果是 NullBean，返回 null
        else if (value instanceof NullBean) {return null;}
        // 默认当成 SpEL 表达式解析了
        else {return evaluate(value);
        }
    }
}

BeanDefinitionValueResolver.resolveReference
解析援用 bean

class BeanDefinitionValueResolver {private Object resolveReference(Object argName, RuntimeBeanReference ref) {
        try {
            Object bean;
            // 获取 bean 类型
            Class<?> beanType = ref.getBeanType();
            if (ref.isToParent()) {
                // 指明从父 BeanFactory 中获取
                BeanFactory parent = this.beanFactory.getParentBeanFactory();

                if (beanType != null) {
                    // 如果有 beanType，间接通过类型获取 bean
                    bean = parent.getBean(beanType);
                }
                else {
                    // 应用 doEvaluate 解析 bean 名字，执行 SpEL 表达式，而后依据名字查找 bean
                    bean = parent.getBean(String.valueOf(doEvaluate(ref.getBeanName())));
                }
            }
            else {
                String resolvedName;
                if (beanType != null) {
                    // 如果有 beanType，间接通过类型获取 bean
                    NamedBeanHolder<?> namedBean = this.beanFactory.resolveNamedBean(beanType);
                    bean = namedBean.getBeanInstance();
                    resolvedName = namedBean.getBeanName();}
                else {
                    // 应用 doEvaluate 解析 bean 名字，执行 SpEL 表达式，而后依据名字查找 bean
                    resolvedName = String.valueOf(doEvaluate(ref.getBeanName()));
                    bean = this.beanFactory.getBean(resolvedName);
                }
                // 注册依赖
                this.beanFactory.registerDependentBean(resolvedName, this.beanName);
            }
            // 如果是 NullBean，返回 null
            if (bean instanceof NullBean) {bean = null;}
            return bean;
        }
        catch (BeansException ex) {// ... 代码省略}
    }

    // 调用 beanFactory 来解析 bean 名字，执行 SpEL 表达式
    private Object doEvaluate(String value) {return this.beanFactory.evaluateBeanDefinitionString(value, this.beanDefinition);
    }
}

BeanDefinitionValueResolver.resolveInnerBean
解析外部 bean

class BeanDefinitionValueResolver {private Object resolveInnerBean(Object argName, String innerBeanName, BeanDefinition innerBd) {
        RootBeanDefinition mbd = null;
        try {
            // 获取合并父 bean 定义的 bean 定义
            mbd = this.beanFactory.getMergedBeanDefinition(innerBeanName, innerBd, this.beanDefinition);

            String actualInnerBeanName = innerBeanName;
            if (mbd.isSingleton()) {
                // 如果同类型的 bean 被实例化屡次，则应用后缀 #1,#2... 来命名之后的 bean
                actualInnerBeanName = adaptInnerBeanName(innerBeanName);
            }
            // 把新生成的名字注册到 beanFactory
            this.beanFactory.registerContainedBean(actualInnerBeanName, this.beanName);
            // 获取依赖
            String[] dependsOn = mbd.getDependsOn();
            if (dependsOn != null) {for (String dependsOnBean : dependsOn) {
                    // 注册依赖
                    this.beanFactory.registerDependentBean(dependsOnBean, actualInnerBeanName);
                    // 确保加载
                    this.beanFactory.getBean(dependsOnBean);
                }
            }
            // 创立 bean
            Object innerBean = this.beanFactory.createBean(actualInnerBeanName, mbd, null);
            // 如果 bean 实例是 FactoryBean，则调用 factory-method 获取真正的 bean
            if (innerBean instanceof FactoryBean) {boolean synthetic = mbd.isSynthetic();
                innerBean = this.beanFactory.getObjectFromFactoryBean((FactoryBean<?>) innerBean, actualInnerBeanName, !synthetic);
            }
            // 如果是 NullBean，返回 null
            if (innerBean instanceof NullBean) {innerBean = null;}
            return innerBean;
        }
        catch (BeansException ex) {// ... 代码省略}
    }

    // 如果同类型的 bean 被实例化屡次，则应用后缀 #1,#2... 来命名之后的 bean
    private String adaptInnerBeanName(String innerBeanName) {
        String actualInnerBeanName = innerBeanName;
        int counter = 0;
        String prefix = innerBeanName + BeanFactoryUtils.GENERATED_BEAN_NAME_SEPARATOR;
        while (this.beanFactory.isBeanNameInUse(actualInnerBeanName)) {
            counter++;
            actualInnerBeanName = prefix + counter;
        }
        return actualInnerBeanName;
    }
}

4. MergedAnnotations.from

MergedAnnotations.from
的底层实现其实是 TypeMappedAnnotations.from

TypeMappedAnnotations的次要性能就是搜寻、操作注解

final class TypeMappedAnnotations implements MergedAnnotations {
    static MergedAnnotations from(AnnotatedElement element, SearchStrategy searchStrategy,
            RepeatableContainers repeatableContainers, AnnotationFilter annotationFilter) {
        // 没有可用的注解，间接返回 NONE
        if (AnnotationsScanner.isKnownEmpty(element, searchStrategy)) {return NONE;}
        return new TypeMappedAnnotations(element, searchStrategy, repeatableContainers, annotationFilter);
    }

    static MergedAnnotations from(@Nullable Object source, Annotation[] annotations,
            RepeatableContainers repeatableContainers, AnnotationFilter annotationFilter) {
        // 没有要查找的注解，间接返回 NONE
        if (annotations.length == 0) {return NONE;}
        return new TypeMappedAnnotations(source, annotations, repeatableContainers, annotationFilter);
    }
}
final class TypeMappedAnnotations implements MergedAnnotations {
    // 查看是否有 annotationType
    @Override
    public <A extends Annotation> boolean isPresent(Class<A> annotationType) {
        // 过滤器没通过，返回 false
        if (this.annotationFilter.matches(annotationType)) {return false;}
        // 扫描指标元素(类、办法、属性)
        return Boolean.TRUE.equals(scan(annotationType,
                IsPresent.get(this.repeatableContainers, this.annotationFilter, false)));
    }

    // 查看是否有 annotationType
    @Override
    public boolean isPresent(String annotationType) {
        // 过滤器没通过，返回 false
        if (this.annotationFilter.matches(annotationType)) {return false;}
        // 扫描指标元素(类、办法、属性)
        return Boolean.TRUE.equals(scan(annotationType,
                IsPresent.get(this.repeatableContainers, this.annotationFilter, false)));
    }

    // 获取指标元素的 annotationType 注解对象
    @Override
    public <A extends Annotation> MergedAnnotation<A> get(Class<A> annotationType,
            @Nullable Predicate<? super MergedAnnotation<A>> predicate,
            @Nullable MergedAnnotationSelector<A> selector) {
        // 过滤器没通过，返回空
        if (this.annotationFilter.matches(annotationType)) {return MergedAnnotation.missing();
        }
        // 扫描指标元素(类、办法、属性)
        MergedAnnotation<A> result = scan(annotationType,
                new MergedAnnotationFinder<>(annotationType, predicate, selector));
        return (result != null ? result : MergedAnnotation.missing());
    }

    // 获取指标元素的 annotationType 注解对象
    @Override
    public <A extends Annotation> MergedAnnotation<A> get(String annotationType,
            @Nullable Predicate<? super MergedAnnotation<A>> predicate,
            @Nullable MergedAnnotationSelector<A> selector) {
        // 过滤器没通过，返回空
        if (this.annotationFilter.matches(annotationType)) {return MergedAnnotation.missing();
        }
        // 扫描指标元素(类、办法、属性)
        MergedAnnotation<A> result = scan(annotationType,
                new MergedAnnotationFinder<>(annotationType, predicate, selector));
        return (result != null ? result : MergedAnnotation.missing());
    }

    // 扫描指标元素(类、办法、属性)
    private <C, R> R scan(C criteria, AnnotationsProcessor<C, R> processor) {
        // 如果有传入的注解对象，就不扫描指标元素了
        if (this.annotations != null) {R result = processor.doWithAnnotations(criteria, 0, this.source, this.annotations);
            return processor.finish(result);
        }
        // 有指标元素和扫描策略，扫描指标元素
        if (this.element != null && this.searchStrategy != null) {return AnnotationsScanner.scan(criteria, this.element, this.searchStrategy, processor);
        }
        return null;
    }
}

TypeMappedAnnotations.scan 的底层实现是 AnnotationsScanner.scan

abstract class AnnotationsScanner {
    // 扫描注解
    static <C, R> R scan(C context, AnnotatedElement source, SearchStrategy searchStrategy,
            AnnotationsProcessor<C, R> processor) {
        // 理论扫描解决
        R result = process(context, source, searchStrategy, processor);
        return processor.finish(result);
    }

    // 理论扫描解决
    private static <C, R> R process(C context, AnnotatedElement source,
            SearchStrategy searchStrategy, AnnotationsProcessor<C, R> processor) {
        // 如果指标元素是类，按类解决
        if (source instanceof Class) {return processClass(context, (Class<?>) source, searchStrategy, processor);
        }
        // 如果指标元素是办法，按办法解决
        if (source instanceof Method) {return processMethod(context, (Method) source, searchStrategy, processor);
        }
        // 不然按一般解决
        return processElement(context, source, processor);
    }
}

4.1. 按类解决

abstract class AnnotationsScanner {
    private static <C, R> R processClass(C context, Class<?> source,
            SearchStrategy searchStrategy, AnnotationsProcessor<C, R> processor) {switch (searchStrategy) {
            // 只搜寻本身的注解，按一般解决
            case DIRECT:
                return processElement(context, source, processor);
            // 搜寻本身与标记为 @Inherited 父类的注解
            case INHERITED_ANNOTATIONS:
                return processClassInheritedAnnotations(context, source, searchStrategy, processor);
            // 搜寻本身与父类的注解
            case SUPERCLASS:
                return processClassHierarchy(context, source, processor, false, false);
            // 搜寻本身、父类、接口的注解
            case TYPE_HIERARCHY:
                return processClassHierarchy(context, source, processor, true, false);
            // 搜寻本身、父类、接口、getEnclosingClass()的注解
            case TYPE_HIERARCHY_AND_ENCLOSING_CLASSES:
                return processClassHierarchy(context, source, processor, true, true);
        }
        throw new IllegalStateException("Unsupported search strategy" + searchStrategy);
    }

    // 搜寻本身与标记为 @Inherited 父类的注解
    private static <C, R> R processClassInheritedAnnotations(C context, Class<?> source,
            SearchStrategy searchStrategy, AnnotationsProcessor<C, R> processor) {
        try {// 如果除了本身以外，没有父类、接口、getEnclosingClass()的注解，按一般解决
            if (isWithoutHierarchy(source, searchStrategy)) {return processElement(context, source, processor);
            }
            Annotation[] relevant = null;
            int remaining = Integer.MAX_VALUE;
            int aggregateIndex = 0;
            Class<?> root = source;

            // source 对象还能够持续向父类上溯
            // 非 Object，非 Ordered，非 "java." 结尾
            while (source != null && source != Object.class && remaining > 0 &&
                    !hasPlainJavaAnnotationsOnly(source)) {

                // 聚合解决，如果有了，间接返回（默认未实现）R result = processor.doWithAggregate(context, aggregateIndex);
                if (result != null) {return result;}

                // 获取所有 public 的注解
                Annotation[] declaredAnnotations = getDeclaredAnnotations(source, true);

                // ... 代码省略

                // 处理器自定义解决
                result = processor.doWithAnnotations(context, aggregateIndex, source, declaredAnnotations);
                // 如果有了，返回
                if (result != null) {return result;}
                // 上溯父类，持续扫描
                source = source.getSuperclass();
                aggregateIndex++;
            }
        }
        catch (Throwable ex) {// ... 代码省略}
        return null;
    }

    // 搜寻本身、父类、接口、getEnclosingClass()的注解
    private static <C, R> R processClassHierarchy(C context, int[] aggregateIndex, Class<?> source,
            AnnotationsProcessor<C, R> processor, boolean includeInterfaces, boolean includeEnclosing) {
        try {
            // 聚合解决，如果有了，间接返回（默认未实现）R result = processor.doWithAggregate(context, aggregateIndex[0]);
            if (result != null) {return result;}
            // 如果是 Ordered 或 "java." 结尾，返回
            if (hasPlainJavaAnnotationsOnly(source)) {return null;}
            // 获取所有 public 的注解
            Annotation[] annotations = getDeclaredAnnotations(source, false);
            // 处理器自定义解决
            result = processor.doWithAnnotations(context, aggregateIndex[0], source, annotations);
            // 如果有了，返回
            if (result != null) {return result;}
            aggregateIndex[0]++;

            // 如果包含接口，持续搜寻接口的
            if (includeInterfaces) {for (Class<?> interfaceType : source.getInterfaces()) {
                    R interfacesResult = processClassHierarchy(context, aggregateIndex,
                        interfaceType, processor, true, includeEnclosing);
                    if (interfacesResult != null) {return interfacesResult;}
                }
            }

            // 如果有父类，持续搜寻父类的
            Class<?> superclass = source.getSuperclass();
            if (superclass != Object.class && superclass != null) {
                R superclassResult = processClassHierarchy(context, aggregateIndex,
                    superclass, processor, includeInterfaces, includeEnclosing);
                if (superclassResult != null) {return superclassResult;}
            }

            // 如果包含 getEnclosingClass()，持续搜寻 getEnclosingClass()的
            if (includeEnclosing) {
                try {Class<?> enclosingClass = source.getEnclosingClass();
                    if (enclosingClass != null) {
                        R enclosingResult = processClassHierarchy(context, aggregateIndex,
                            enclosingClass, processor, includeInterfaces, true);
                        if (enclosingResult != null) {return enclosingResult;}
                    }
                }
                catch (Throwable ex) {// ... 代码省略}
            }
        }
        catch (Throwable ex) {// ... 代码省略}
        return null;
    }
}

4.2. 按办法解决

abstract class AnnotationsScanner {
    private static <C, R> R processMethod(C context, Method source,
            SearchStrategy searchStrategy, AnnotationsProcessor<C, R> processor) {switch (searchStrategy) {
            // 只搜寻本身的注解
            case DIRECT:
            // 搜寻本身与标记为 @Inherited 父类的注解
            case INHERITED_ANNOTATIONS:
                return processMethodInheritedAnnotations(context, source, processor);
            // 搜寻本身与父类的注解
            case SUPERCLASS:
                return processMethodHierarchy(context, new int[] {0}, source.getDeclaringClass(),
                        processor, source, false);
            // 搜寻本身、父类、接口的注解
            case TYPE_HIERARCHY:
            // 搜寻本身、父类、接口、getEnclosingClass()的注解
            case TYPE_HIERARCHY_AND_ENCLOSING_CLASSES:
                return processMethodHierarchy(context, new int[] {0}, source.getDeclaringClass(),
                        processor, source, true);
        }
        throw new IllegalStateException("Unsupported search strategy" + searchStrategy);
    }

    // 搜寻本身与标记为 @Inherited 父类的注解
    private static <C, R> R processMethodInheritedAnnotations(C context, Method source,
            AnnotationsProcessor<C, R> processor) {

        try {
            // 聚合解决，如果有了，间接返回（默认未实现）// 不然，用 processMethodAnnotations 解决
            R result = processor.doWithAggregate(context, 0);
            return (result != null ? result :
                processMethodAnnotations(context, 0, source, processor));
        }
        catch (Throwable ex) {// ... 代码省略}
        return null;
    }

    // 解决办法的注解
    private static <C, R> R processMethodAnnotations(C context, int aggregateIndex, Method source,
            AnnotationsProcessor<C, R> processor) {

      // 获取所有 public 的注解
        Annotation[] annotations = getDeclaredAnnotations(source, false);
        // 处理器自定义解决
        R result = processor.doWithAnnotations(context, aggregateIndex, source, annotations);
        // 如果有了，返回
        if (result != null) {return result;}

        // ... 代码省略，桥接办法的解决

        return null;
    }

    // 搜寻本身、父类、接口、getEnclosingClass()的注解
    private static <C, R> R processMethodHierarchy(C context, int[] aggregateIndex,
            Class<?> sourceClass, AnnotationsProcessor<C, R> processor, Method rootMethod,
            boolean includeInterfaces) {
        try {
            // 聚合解决，如果有了，间接返回（默认未实现）R result = processor.doWithAggregate(context, aggregateIndex[0]);
            if (result != null) {return result;}
            // 如果是 Ordered 或 "java." 结尾，返回
            if (hasPlainJavaAnnotationsOnly(sourceClass)) {return null;}

            boolean calledProcessor = false;
            // 如果 rootMethod 是 sourceClass 的，间接解决
            if (sourceClass == rootMethod.getDeclaringClass()) {result = processMethodAnnotations(context, aggregateIndex[0],
                    rootMethod, processor);
                calledProcessor = true;
                if (result != null) {return result;}
            }
            else {
                // 获取 sourceClass 的 public 办法
                for (Method candidateMethod : getBaseTypeMethods(context, sourceClass)) {
                    // 如果办法不是 private，并且参数与 rootMethod 始终，间接解决
                    if (candidateMethod != null && isOverride(rootMethod, candidateMethod)) {result = processMethodAnnotations(context, aggregateIndex[0],
                            candidateMethod, processor);
                        calledProcessor = true;
                        if (result != null) {return result;}
                    }
                }
            }
            // 如果 rootMethod 是 private，返回 null
            if (Modifier.isPrivate(rootMethod.getModifiers())) {return null;}

            // 如果包含接口，持续搜寻接口的
            if (includeInterfaces) {for (Class<?> interfaceType : sourceClass.getInterfaces()) {
                    R interfacesResult = processMethodHierarchy(context, aggregateIndex,
                        interfaceType, processor, rootMethod, true);
                    if (interfacesResult != null) {return interfacesResult;}
                }
            }

            // 如果有父类，持续搜寻父类的
            Class<?> superclass = sourceClass.getSuperclass();
            if (superclass != Object.class && superclass != null) {
                R superclassResult = processMethodHierarchy(context, aggregateIndex,
                    superclass, processor, rootMethod, includeInterfaces);
                if (superclassResult != null) {return superclassResult;}
            }
        }
        catch (Throwable ex) {// ... 代码省略}
        return null;
    }
}

4.3 按一般解决

abstract class AnnotationsScanner {
    private static <C, R> R processElement(C context, AnnotatedElement source,
            AnnotationsProcessor<C, R> processor) {

        try {
            // 聚合解决，如果有了，间接返回（默认未实现）// 不然，调用 getDeclaredAnnotations 获取注解来解决
            R result = processor.doWithAggregate(context, 0);
            return (result != null ? result : processor.doWithAnnotations(context, 0, source, getDeclaredAnnotations(source, false)));
        }
        catch (Throwable ex) {// ... 代码省略}
        return null;
    }
}

后续

更多博客，查看 https://github.com/senntyou/blogs

作者：深予之 (@senntyou)

版权申明：自在转载 - 非商用 - 非衍生 - 放弃署名（创意共享 3.0 许可证）

正文完
后端
发表至： 后端
2021-10-26
 0
关于后端:三端一体计算方案Unify-SQL-Engine
关于后端:snpEFF和bedtools基因注释有何异同
关于后端:老杜带你学Ajax轻松掌握ajax底层实现原理
关于后端:基于-Log-的通用增量-Checkpoint-在美团的进展