SpringBoot @ConditionalOnBean、@ConditionalOnMissingBean注解源码分析与示例
前言:
Spring4推出了@Conditional注解,方便程序根据当前环境或者容器情况来动态注入bean,对@Conditional注解不熟悉的朋友可移步至 Spring @Conditional注解 详细讲解及示例 这篇博客进行学习。
继@Conditional注解后,又基于此注解推出了很多派生注解,比如@ConditionalOnBean、@ConditionalOnMissingBean、@ConditionalOnExpression、@ConditionalOnClass......动态注入bean变得更方便了。本篇将讲解@ConditionalOnBean注解。
配置类中有两个Computer类的bean,一个是笔记本电脑,一个是备用电脑。如果当前容器中已经有电脑bean了,就不注入备用电脑,如果没有,则注入备用电脑,这里需要使用到@ConditionalOnMissingBean。
-
@Configuration
-
public class BeanConfig {
-
-
@Bean(name = "notebookPC")
-
public Computer computer1(){
-
return new Computer("笔记本电脑");
-
}
-
-
@ConditionalOnMissingBean(Computer.class)
-
@Bean("reservePC")
-
public Computer computer2(){
-
return new Computer("备用电脑");
-
}
-
}
这个注解就实现了功能,这个@ConditionalOnMissingBean为我们做了什么呢?我们来一探究竟.。
一探究竟:
首先,来看@ConditionalOnMissingBean的声明:
-
//可以标注在类和方法上
-
@Target({ElementType.TYPE, ElementType.METHOD})
-
@Retention(RetentionPolicy.RUNTIME)
-
@Documented
-
//使用了@Conditional注解,条件类是OnBeanCondition
-
@Conditional({OnBeanCondition.class})
-
public @interface ConditionalOnMissingBean {
-
Class<?>[] value() default {};
-
-
String[] type() default {};
-
-
Class<?>[] ignored() default {};
-
-
String[] ignoredType() default {};
-
-
Class<? extends Annotation>[] annotation() default {};
-
-
String[] name() default {};
-
-
SearchStrategy search() default SearchStrategy.ALL;
-
}
这时候,我们就看到了我们熟悉的@Conditional注解,OnBeanCondition作为条件类。
OnBeanCondition类的声明:
-
//定义带注释的组件的排序顺序,2147483647即为默认值
-
@Order(2147483647)
-
class OnBeanCondition extends SpringBootCondition implements ConfigurationCondition {
它继承了SpringBootCondition类,OnBeanCondition类中没有matches方法,而SpringBootCondition类中有实现matches方法。OnBeanCondition还实现了ConfigurationCondition,ConfigurationCondition接口不熟悉的读者可以到Spring ConfigurationCondition接口详解 了解接口。OnBeanCondition类重写了getConfigurationPhase()方法,表示在注册bean的时候注解生效:
-
public ConfigurationPhase getConfigurationPhase() {
-
return ConfigurationPhase.REGISTER_BEAN;
-
}
就从matches方法开始:
-
//SpringBootCondition类中的matches方法
-
public final boolean matches(ConditionContext context, AnnotatedTypeMetadata metadata) {
-
//获取当前的类名或者方法名(由标注的位置决定)
-
String classOrMethodName = getClassOrMethodName(metadata);
-
-
try {
-
//关键代码:这里就会判断出结果
-
ConditionOutcome outcome = this.getMatchOutcome(context, metadata);
-
//存入日志
-
this.logOutcome(classOrMethodName, outcome);
-
//存入记录
-
this.recordEvaluation(context, classOrMethodName, outcome);
-
//最后返回ConditionOutcome的isMatch就是返回boolean类型结果
-
return outcome.isMatch();
-
} catch (NoClassDefFoundError var5) {
-
throw new IllegalStateException("Could not evaluate condition on " + classOrMethodName + " due to " + var5.getMessage() + " not found. Make sure your own configuration does not rely on that class. This can also happen if you are @ComponentScanning a springframework package (e.g. if you put a @ComponentScan in the default package by mistake)", var5);
-
} catch (RuntimeException var6) {
-
throw new IllegalStateException("Error processing condition on " + this.getName(metadata), var6);
-
}
-
}
关键代码在OnBeanCondition的getMatchOutcome方法上:
-
/**
-
* 获得判断结果的方法,ConditionOutcome类中存着boolean类型的结果
-
*/
-
public ConditionOutcome getMatchOutcome(ConditionContext context, AnnotatedTypeMetadata metadata) {
-
//返回一个新的ConditionMessage
-
ConditionMessage matchMessage = ConditionMessage.empty();
-
OnBeanCondition.BeanSearchSpec spec;
-
List matching;
-
//这是metadata会调用isAnnotated方法判断当前标注的注解是不是ConditionalOnMissingBean
-
//其实@ConditionalOnBean、@ConditionalOnMissingBean和@ConditionalOnSingleCandidate都是使用这个条件类,所以这里做判断
-
if (metadata.isAnnotated(ConditionalOnBean.class.getName())) {
-
spec = new OnBeanCondition.BeanSearchSpec(context, metadata, ConditionalOnBean.class);
-
matching = this.getMatchingBeans(context, spec);
-
if (matching.isEmpty()) {
-
return ConditionOutcome.noMatch(ConditionMessage.forCondition(ConditionalOnBean.class, new Object[]{spec}).didNotFind("any beans").atAll());
-
}
-
-
matchMessage = matchMessage.andCondition(ConditionalOnBean.class, new Object[]{spec}).found("bean", "beans").items(Style.QUOTE, matching);
-
}
-
-
if (metadata.isAnnotated(ConditionalOnSingleCandidate.class.getName())) {
-
OnBeanCondition.BeanSearchSpec spec = new OnBeanCondition.SingleCandidateBeanSearchSpec(context, metadata, ConditionalOnSingleCandidate.class);
-
matching = this.getMatchingBeans(context, spec);
-
if (matching.isEmpty()) {
-
return ConditionOutcome.noMatch(ConditionMessage.forCondition(ConditionalOnSingleCandidate.class, new Object[]{spec}).didNotFind("any beans").atAll());
-
}
-
-
if (!this.hasSingleAutowireCandidate(context.getBeanFactory(), matching, spec.getStrategy() == SearchStrategy.ALL)) {
-
return ConditionOutcome.noMatch(ConditionMessage.forCondition(ConditionalOnSingleCandidate.class, new Object[]{spec}).didNotFind("a primary bean from beans").items(Style.QUOTE, matching));
-
}
-
-
matchMessage = matchMessage.andCondition(ConditionalOnSingleCandidate.class, new Object[]{spec}).found("a primary bean from beans").items(Style.QUOTE, matching);
-
}
-
-
//如果当前注入的bean是@ConditionalOnMissingBean
-
if (metadata.isAnnotated(ConditionalOnMissingBean.class.getName())) {
-
//返回一个spec(说明),这里的spec规定了搜索的内容,比如搜索策略、需要搜索的类名......
-
spec = new OnBeanCondition.BeanSearchSpec(context, metadata, ConditionalOnMissingBean.class);
-
//主要的搜索实现在这个方法里,最后返回一个list
-
matching = this.getMatchingBeans(context, spec);
-
//判断搜索出来的结果
-
if (!matching.isEmpty()) {
-
return ConditionOutcome.noMatch(ConditionMessage.forCondition(ConditionalOnMissingBean.class, new Object[]{spec}).found("bean", "beans").items(Style.QUOTE, matching));
-
}
-
-
matchMessage = matchMessage.andCondition(ConditionalOnMissingBean.class, new Object[]{spec}).didNotFind("any beans").atAll();
-
}
-
-
return ConditionOutcome.match(matchMessage);
-
}
spec = new OnBeanCondition.BeanSearchSpec(context, metadata, ConditionalOnBean.class);
这句中,相当于从内部类中将标注@ConditionalOnMissingBean注解时的属性都取出来:
-
BeanSearchSpec(ConditionContext context, AnnotatedTypeMetadata metadata, Class<?> annotationType) {
-
this.annotationType = annotationType;
-
MultiValueMap<String, Object> attributes = metadata.getAllAnnotationAttributes(annotationType.getName(), true);
-
//将attributes这个map中的数据放到对应的list成员变量中
-
this.collect(attributes, "name", this.names);
-
this.collect(attributes, "value", this.types);
-
this.collect(attributes, "type", this.types);
-
this.collect(attributes, "annotation", this.annotations);
-
this.collect(attributes, "ignored", this.ignoredTypes);
-
this.collect(attributes, "ignoredType", this.ignoredTypes);
-
this.strategy = (SearchStrategy)metadata.getAnnotationAttributes(annotationType.getName()).get("search");
-
OnBeanCondition.BeanTypeDeductionException deductionException = null;
-
-
try {
-
if (this.types.isEmpty() && this.names.isEmpty()) {
-
this.addDeducedBeanType(context, metadata, this.types);
-
}
-
} catch (OnBeanCondition.BeanTypeDeductionException var7) {
-
deductionException = var7;
-
}
-
-
this.validate(deductionException);
-
}
-
-
//验证的方法
-
protected void validate(OnBeanCondition.BeanTypeDeductionException ex) {
-
if (!this.hasAtLeastOne(this.types, this.names, this.annotations)) {
-
String message = this.annotationName() + " did not specify a bean using type, name or annotation";
-
if (ex == null) {
-
throw new IllegalStateException(message);
-
} else {
-
throw new IllegalStateException(message + " and the attempt to deduce the bean's type failed", ex);
-
}
-
}
-
}
看一下OnBeanCondition类中的getMatchingBeans方法,里面有用到搜索策略,详见搜索策略介绍
-
private List<String> getMatchingBeans(ConditionContext context, OnBeanCondition.BeanSearchSpec beans) {
-
//获得当前bean工厂
-
ConfigurableListableBeanFactory beanFactory = context.getBeanFactory();
-
//判断当前的搜索策略是否是PARENTS或者ANCESTORS,默认是ALL
-
if (beans.getStrategy() == SearchStrategy.PARENTS || beans.getStrategy() == SearchStrategy.ANCESTORS) {
-
BeanFactory parent = beanFactory.getParentBeanFactory();
-
Assert.isInstanceOf(ConfigurableListableBeanFactory.class, parent, "Unable to use SearchStrategy.PARENTS");
-
//如果是PARENTS或者ANCESTORS,当前bean工厂就用父工厂
-
beanFactory = (ConfigurableListableBeanFactory)parent;
-
}
-
-
if (beanFactory == null) {
-
return Collections.emptyList();
-
} else {
-
List<String> beanNames = new ArrayList();
-
//如果当前搜索策略等于CURRENT,为true
-
boolean considerHierarchy = beans.getStrategy() != SearchStrategy.CURRENT;
-
//这里的type就是需要查找的bean的类型
-
//下面,会从属性中找bean
-
Iterator var6 = beans.getTypes().iterator();
-
-
String beanName;
-
while(var6.hasNext()) {
-
beanName = (String)var6.next();
-
//如果找到了类型,接下来就是根据类型找bean的实例名,找示例名的方法在下方,实际上就是一个getNamesForType
-
beanNames.addAll(this.getBeanNamesForType(beanFactory, beanName, context.getClassLoader(), considerHierarchy));
-
}
-
-
var6 = beans.getIgnoredTypes().iterator();
-
-
while(var6.hasNext()) {
-
beanName = (String)var6.next();
-
beanNames.removeAll(this.getBeanNamesForType(beanFactory, beanName, context.getClassLoader(), considerHierarchy));
-
}
-
-
var6 = beans.getAnnotations().iterator();
-
-
while(var6.hasNext()) {
-
beanName = (String)var6.next();
-
beanNames.addAll(Arrays.asList(this.getBeanNamesForAnnotation(beanFactory, beanName, context.getClassLoader(), considerHierarchy)));
-
}
-
-
var6 = beans.getNames().iterator();
-
-
while(var6.hasNext()) {
-
beanName = (String)var6.next();
-
if (this.containsBean(beanFactory, beanName, considerHierarchy)) {
-
beanNames.add(beanName);
-
}
-
}
-
//将存放bean实例名的list返回
-
return beanNames;
-
}
-
}
-
-
-
-
//根据类型获取bean的name
-
private Collection<String> getBeanNamesForType(ListableBeanFactory beanFactory, String type, ClassLoader classLoader, boolean considerHierarchy) throws LinkageError {
-
try {
-
Set<String> result = new LinkedHashSet();
-
this.collectBeanNamesForType(result, beanFactory, ClassUtils.forName(type, classLoader), considerHierarchy);
-
return result;
-
} catch (ClassNotFoundException var6) {
-
return Collections.emptySet();
-
} catch (NoClassDefFoundError var7) {
-
return Collections.emptySet();
-
}
-
}
-
-
private void collectBeanNamesForType(Set<String> result, ListableBeanFactory beanFactory, Class<?> type, boolean considerHierarchy) {
-
result.addAll(BeanTypeRegistry.get(beanFactory).getNamesForType(type));
-
if (considerHierarchy && beanFactory instanceof HierarchicalBeanFactory) {
-
BeanFactory parent = ((HierarchicalBeanFactory)beanFactory).getParentBeanFactory();
-
if (parent instanceof ListableBeanFactory) {
-
this.collectBeanNamesForType(result, (ListableBeanFactory)parent, type, considerHierarchy);
-
}
-
}
-
-
}
找完bean了之后,回到刚才的代码里:
-
//如果当前注入的bean是@ConditionalOnMissingBean
-
if (metadata.isAnnotated(ConditionalOnMissingBean.class.getName())) {
-
//返回一个spec(说明),这里的spec规定了搜索的内容,比如搜索策略、需要搜索的类名......
-
spec = new OnBeanCondition.BeanSearchSpec(context, metadata, ConditionalOnMissingBean.class);
-
matching = this.getMatchingBeans(context, spec);
-
if (!matching.isEmpty()) {
-
return ConditionOutcome.noMatch(ConditionMessage.forCondition(ConditionalOnMissingBean.class, new Object[]{spec}).found("bean", "beans").items(Style.QUOTE, matching));
-
}
-
-
matchMessage = matchMessage.andCondition(ConditionalOnMissingBean.class, new Object[]{spec}).didNotFind("any beans").atAll();
-
}
如果第5行返回的list不是空的,就会返回ConditionOutcome对象noMatch方法,表示不匹配。ConditionOutcome类用于存放过滤结果,只有两个变量:
-
/**
-
* 过滤结果类
-
*/
-
public class ConditionOutcome {
-
/**
-
* 匹配结果 true or false
-
*/
-
private final boolean match;
-
/**
-
* 匹配结果信息
-
*/
-
private final ConditionMessage message;
两者区别:
@ConditionOnBean在判断list的时候,如果list没有值,返回false,否则返回true
@ConditionOnMissingBean在判断list的时候,如果list没有值,返回true,否则返回false,其他逻辑都一样
例子:
- @ConditionalOnBean(javax.sql.DataSource.class)
Spring容器或者所有父容器中需要存在至少一个javax.sql.DataSource类的实例
SpringBoot @ConditionalOnBean、@ConditionalOnMissingBean注解源码分析与示例的更多相关文章
- SpringBoot的条件注解源码解析
SpringBoot的条件注解源码解析 @ConditionalOnBean.@ConditionalOnMissingBean 启动项目 会在ConfigurationClassBeanDefini ...
- Springboot 加载配置文件源码分析
Springboot 加载配置文件源码分析 本文的分析是基于springboot 2.2.0.RELEASE. 本篇文章的相关源码位置:https://github.com/wbo112/blogde ...
- Spring笔记(5) - 声明式事务@EnableTransactionManagement注解源码分析
一.背景 前面详解了实现Spring事务的两种方式的不同实现:编程式事务和声明式事务,对于配置都使用到了xml配置,今天介绍Spring事务的注解开发,例如下面例子: 配置类:注册数据源.JDBC模板 ...
- java io系列02之 ByteArrayInputStream的简介,源码分析和示例(包括InputStream)
我们以ByteArrayInputStream,拉开对字节类型的“输入流”的学习序幕.本章,我们会先对ByteArrayInputStream进行介绍,然后深入了解一下它的源码,最后通过示例来掌握它的 ...
- java io系列03之 ByteArrayOutputStream的简介,源码分析和示例(包括OutputStream)
前面学习ByteArrayInputStream,了解了“输入流”.接下来,我们学习与ByteArrayInputStream相对应的输出流,即ByteArrayOutputStream.本章,我们会 ...
- java io系列04之 管道(PipedOutputStream和PipedInputStream)的简介,源码分析和示例
本章,我们对java 管道进行学习. 转载请注明出处:http://www.cnblogs.com/skywang12345/p/io_04.html java 管道介绍 在java中,PipedOu ...
- springBoot从入门到源码分析
先分享一个springBoot搭建学习项目,和springboot多数据源项目的传送门:https://github.com/1057234721/springBoot 1. SpringBoot快速 ...
- spring注解源码分析--how does autowired works?
1. 背景 注解可以减少代码的开发量,spring提供了丰富的注解功能.我们可能会被问到,spring的注解到底是什么触发的呢?今天以spring最常使用的一个注解autowired来跟踪代码,进行d ...
- SpringBoot拦截器及源码分析
1.拦截器是什么 java里的拦截器(Interceptor)是动态拦截Action调用的对象,它提供了一种机制可以使开发者在一个Action执行的前后执行一段代码,也可以在一个Action执行前阻止 ...
随机推荐
- C#结构体+结构体与类的区别
C# 结构(Struct) 在 C# 中,结构是值类型数据结构.它使得一个单一变量可以存储各种数据类型的相关数据.struct 关键字用于创建结构. C# 结构的特点 您已经用了一个简单的名为 Boo ...
- spring的依赖注入如何降低了耦合
依赖注入:程序运行过程中,如需另一个对象协作(调用它的方法.访问他的属性时),无须在代码中创建被调用者,而是依赖于外部容器的注入 看过一些比较好的回答 1.一个人(Java实例,调用者)需要一把斧子( ...
- 常用JavaScript代码库(又名:WFang.js)
1.根据公司项目封装ajax请求,结合layer框架一起使用 /*提取接口公共部分*/ var ApiConf = { server:"http://localhost:8080/Batte ...
- 网站开发综合技术 第二部分 CSS样式表
第2部分 CSS样式表 CSS(Cascading Style Sheets,层叠样式表),作用是美化HTML网页. /*注释*/ 注释语法 2.1.样式表的基本概念 2.1.1.样式表分类 1 ...
- 专题七:UDP编程补充——UDP广播程序的实现
一.程序实现 UDP广播程序的实现代码: using System; using System.Net; using System.Net.Sockets; using System.Text; us ...
- Spring(二) -- 春风拂面之 核心 AOP
”万物皆对象“是面向对象编程思想OOP(Object Oriented Programming) 的最高境界.在面向对象中,我一直将自己(开发者)放在一个至高无上的位置上,可以操纵万物(对象),犹如一 ...
- Python之双色球选购和三级菜单问题
1:双色球选购# 1 双色球(假设一共八个球,6个红球,球号1-32.2个蓝球,球号1-16)# 2 确保用户不能重复选择,不能超出范围# 3 用户输入有误时有相应的错误提示# 4 最后展示用户选择的 ...
- .Net实战之反射相关类之间的人体经络关系
--1.类的部分组成成员 --2.巧记成员之间的关系 [MyTable("T_UserInfo")] public class UserInfo : Person, UserSer ...
- leetcode486 Predict the Winner
思路: 博弈. 实现: class Solution { public: bool PredictTheWinner(vector<int>& nums) { ][]; int n ...
- 前端面试题HTML
浏览器页面有哪三层构成,分别是什么,作用是什么?