spring data elasticsearch 使用及源码分析

spring integrate es.png

最近一直在用elasticsearch做一些数据的存储和查询 ，开始为了方便就直接使用了spring data elasticsearch ，期间也经历了很多坑，比如findByName(String name)这种接口默认只返回10条数据，自己也对源码做个跟踪，也研究了一下为什么elasticsearch要这样做。接下来我们通过分析spring data es的两个接口，来看一下是如果实现对es进行操作的。

1.findByName接口分析

• spring data es repository 接口

public interface UserRepository extends ElasticsearchRepository {
    /**
     * 默认只返回10条
     */
    List findByName(String name);


    List deleteByName(String name);
}

• spring 启动分析构造UserRepository代理接口
  启动的时候ElasticsearchRepositoryFactoryBean的方法会进行对应代理类的生成,并注入到spring容器

    @Override
    public void afterPropertiesSet() {
        super.afterPropertiesSet();
        Assert.notNull(operations, "ElasticsearchOperations must be configured!");
    }

这里的super指是RepositoryFactoryBeanSupport，spring data jpa启动整体代理接口注入到spring容器中也是通过这个类做的，其中spring data jpa 对应的类是JpaRepositoryFactoryBean。

#其中RepositoryFactoryBeanSupport类中的下面这段代码比较核心
public void afterPropertiesSet() {
        #getRepositoryMetadata是获取这个接口的元数据，就是这个接口的一些基本的属
        this.repositoryMetadata = this.factory.getRepositoryMetadata(repositoryInterface);
        #这个方法是生成这个接口的实现类
        this.repository = Lazy.of(() -> this.factory.getRepository(repositoryInterface, repositoryFragmentsToUse));
        #如果不是懒加载立即初始化
        if (!lazyInit) {
            this.repository.get();
        }
    }

接着我们分析上面的核心代码getRepository这个方法

public  T getRepository(Class repositoryInterface, RepositoryFragments fragments) {

        Assert.notNull(repositoryInterface, "Repository interface must not be null!");
        Assert.notNull(fragments, "RepositoryFragments must not be null!");

        RepositoryMetadata metadata = getRepositoryMetadata(repositoryInterface);
        RepositoryComposition composition = getRepositoryComposition(metadata, fragments);
        RepositoryInformation information = getRepositoryInformation(metadata, composition);

        validate(information, composition);

        //这里是构造一个对象，然后填充一些这个接口对应的entity信息，比如索引名，类型名等
        Object target = getTargetRepository(information);

        // Create proxy 接下来通过代理工厂创建代理
        ProxyFactory result = new ProxyFactory();
        result.setTarget(target);
        result.setInterfaces(repositoryInterface, Repository.class, TransactionalProxy.class);

        result.addAdvice(SurroundingTransactionDetectorMethodInterceptor.INSTANCE);
        result.addAdvisor(ExposeInvocationInterceptor.ADVISOR);

        postProcessors.forEach(processor -> processor.postProcess(result, information));
    
        result.addAdvice(new DefaultMethodInvokingMethodInterceptor());
        //这个DefaultMethodInvokingMethodInterceptor
        //会为所有我们在repository接口中自定义的方法加上切面
        result.addAdvice(new QueryExecutorMethodInterceptor(information));

        composition = composition.append(RepositoryFragment.implemented(target));
        result.addAdvice(new ImplementationMethodExecutionInterceptor(composition));
        //生产代理类返回注入到spring容器
        return (T) result.getProxy(classLoader);
    }

接下来我们看看QueryExecutorMethodInterceptor类，主要是resolveQuery这个方法

this.queries = lookupStrategy.map(it -> {

                SpelAwareProxyProjectionFactory factory = new SpelAwareProxyProjectionFactory();
                factory.setBeanClassLoader(classLoader);
                factory.setBeanFactory(beanFactory);

                return repositoryInformation.getQueryMethods().stream()//
                        .map(method -> Pair.of(method, it.resolveQuery(method, repositoryInformation, factory, namedQueries)))//
                        .peek(pair -> invokeListeners(pair.getSecond()))//
                        .collect(Pair.toMap());

            }).orElse(Collections.emptyMap());

resolveQuery 中会为每一个方法创建一个ElasticsearchPartQuery,其中ElasticsearchPartQuery是自定义操作实现的核心类,我们看一下代码

    //构建类的时候会创建一个PartTree主要用来描述这个方法是什么方法，有没有分页等到
    private final PartTree tree;
    private final MappingContext mappingContext;

    public ElasticsearchPartQuery(ElasticsearchQueryMethod method, ElasticsearchOperations elasticsearchOperations) {
        super(method, elasticsearchOperations);
        this.tree = new PartTree(method.getName(), method.getEntityInformation().getJavaType());
        this.mappingContext = elasticsearchOperations.getElasticsearchConverter().getMappingContext();
    }

    //这个是在方法运行时调用时通过切面到这里然后转换成elasticsearch的查询接口
    @Override
    public Object execute(Object[] parameters) {
        ParametersParameterAccessor accessor = new ParametersParameterAccessor(queryMethod.getParameters(), parameters);
        CriteriaQuery query = createQuery(accessor);
        if(tree.isDelete()) {
            Object result = countOrGetDocumentsForDelete(query, accessor);
            elasticsearchOperations.delete(query, queryMethod.getEntityInformation().getJavaType());
            return result;
        } else if (queryMethod.isPageQuery()) {
            query.setPageable(accessor.getPageable());
            return elasticsearchOperations.queryForPage(query, queryMethod.getEntityInformation().getJavaType());
        } else if (queryMethod.isStreamQuery()) {
            Class entityType = queryMethod.getEntityInformation().getJavaType();
            if (query.getPageable().isUnpaged()) {
                int itemCount = (int) elasticsearchOperations.count(query, queryMethod.getEntityInformation().getJavaType());
                query.setPageable(PageRequest.of(0, Math.max(1, itemCount)));
            }

            return StreamUtils.createStreamFromIterator((CloseableIterator