spring批处理JpaPagingItemReader在测试中没有EntityManager(NPE)

Question

我有一个使用JPapagingItemReader的小型spring批处理作业。当我从命令行启动时，该作业运行良好，但当我要测试@stepscope组件JPapagingItemReader时，我得到一个NullPointerException。我不明白为什么这个工作能很好地工作，但是我不能根据文档测试它。

我只有一个简单的employee类，带有基本的JPA注释。

以下是我的作业配置类：

@Slf4j
@Configuration
@EnableBatchProcessing
public class PaySalaryJobConfiguration {

    private final JobBuilderFactory jobBuilderFactory;
    private final StepBuilderFactory stepBuilderFactory;

    @Autowired
    public PaySalaryJobConfiguration(
            JobBuilderFactory jobBuilderFactory,
            StepBuilderFactory stepBuilderFactory
    ) {
        this.jobBuilderFactory = jobBuilderFactory;
        this.stepBuilderFactory = stepBuilderFactory;
    }

    @Bean
    @StepScope
    public ItemProcessor<Employee, Employee> employeeProcessor() {
        return item -> {
            log.info("Process item: {}", item);
            return item;
        };
    }

    @Bean
    @StepScope
    public JpaPagingItemReader<Employee> someoneReader(
            EntityManagerFactory entityManagerFactory
    ) {
        return new JpaPagingItemReaderBuilder<Employee>()
                .name("someone-reader")
                .entityManagerFactory(entityManagerFactory)
                .queryString("Select e from Employee e order by e.id asc")
                .pageSize(1)
                .build();
    }

    @Bean
    @StepScope
    public FlatFileItemWriter<Employee> csvWriter(
            @Value("#{jobParameters['output.path.csv']}") String outputPath
    ) {
        DelimitedLineAggregator<Employee> lineAggregator = new DelimitedLineAggregator<>();
        lineAggregator.setDelimiter(",");

        BeanWrapperFieldExtractor<Employee> fieldExtractor = new BeanWrapperFieldExtractor<>();
        fieldExtractor.setNames(new String[]{"id", "firstName", "lastName", "salary"});
        lineAggregator.setFieldExtractor(fieldExtractor);

        return new FlatFileItemWriterBuilder<Employee>()
                .name("csv-salary-writer")
                .resource(new FileSystemResource(outputPath))
                .lineAggregator(lineAggregator)
                .encoding("UTF-8")
                .build();
    }

    @Bean
    public Step writeSalarySlipToCsv (
            JpaPagingItemReader<Employee> someoneReader,
            FlatFileItemWriter<Employee> csvWriter
    ) {
        return stepBuilderFactory
                .get("retrieve-salary-slip-step")
                .<Employee, Employee>chunk(1)
                .reader(someoneReader)
                .processor(employeeProcessor())
                .writer(csvWriter)
                .stream(someoneReader)
                .build();
    }

    @Bean
    public Job paySalaryJob(
            Step writeSalarySlipToCsv
    ) {
        return jobBuilderFactory
                .get("pay-salary-job")
                .incrementer(new RunIdIncrementer())
                .start(writeSalarySlipToCsv)
                .build();
    }
}

下面是我的测试类：

@Slf4j
@SpringBatchTest
@RunWith(SpringRunner.class)
@EnableAutoConfiguration
@ContextConfiguration(classes = PaySalaryJobConfiguration.class)
public class StepScopeIntegrationTest {

    @Autowired
    private JpaPagingItemReader<Employee> someoneReader;

    @Autowired
    private JdbcTemplate jdbcTemplate;

    @Before
    public void setUp() {
        log.info("Before execution we have {} entries", jdbcTemplate.queryForObject("SELECT COUNT(*) FROM EMPLOYEE", Integer.class ));
    }

    public StepExecution getStepExecution() {
        log.info("Step Execution !");
        StepExecution stepExecution = MetaDataInstanceFactory.createStepExecution("retrieve-salary-slip-step", 1564L);
        log.info("Context = {} / Step = {} ", stepExecution.getExecutionContext(), stepExecution.getStepName());
        return stepExecution;
    }

    @Test
    public void testReader() throws Exception {
        log.info("Page = {}", someoneReader.getPage());
        log.info("PageSize = {}", someoneReader.getPageSize());
        Assert.assertNotNull(someoneReader.read());
    }
}

问题出现在第192行的JPAPagingItemReader类中：

@Override
@SuppressWarnings("unchecked")
protected void doReadPage() {

    EntityTransaction tx = null;

    if (transacted) {
        tx = entityManager.getTransaction(); // EntityManager is null...
        tx.begin();

        entityManager.flush();
        entityManager.clear();
    }//end if

下面是来自测试执行的stacktrace：

2019-06-12 17:17:32.845-info-[main]{b.g.t.StepScopeIntegrationTest 45}-->分步执行！2019-06-12 17:17:32.845-Info-[main]{B.G.T.StepScopeIntegrationTest 47}-->上下文={}/Step=检索-工资-工资单-步骤2019-06-12 17:17:32.970-Info-[main]{B.G.T.StepScopeIntegrationTest 41}-->在执行之前，我们有4个条目2019-06-12 17:17:33.002-Info-[main]{B.G.T.StepScopeIntegrationTest 53}-->页面=0 2019-06-12 17:17:33.002-Info-[main]{B.G.T.StepScopeIntegrationTest 54}

java.lang.NullPointerException（位于org.springframework.batch.item.database.jpapagingitemreader.doReadPage（位于org.springframework.batch.item.database.abstractpagingitemreader.java:192)（位于org.springframework.batch.item.database.abstractpagingitemreader.doRead(abstractpagingitemreader.java:108)（位于nvoke(DelegatingIntroductionInterceptor.java:124)位于org.springframework.aop.framework.cglibaopproxy$DynamicAdvisedInterceptor.Intercept(Cglibaopproxy.java:688)位于org.springframework.batch.item.database.jpapagingitemreader$$enhancerbyspringcglib$$b998315d.read()位于VA:47)位于org.junit.internal.runners.statements.invokeMethod.evaluate(invokeMethod.Java:17)位于org.springframework.test.context.junit4.statements.runbeforestexecutioncallbacks.evaluate(runbeforestexecutioncallbacks.Java:73)位于org.springframework.test.context.junit4.statements.runaftertestexecutioncallbacks.evaluatet.context.junit4.SpringJunit4ClassRunner.Runchild(SpringJunit4ClassRunner.:97）在org.junit.runners.ParentRunner$3。Run（ParentRunner.Java:290)在org.junit.runners.ParentRunner.Runchildre(ParentRunner.Java:290)在org.junit.runners.ParentRunner.Runchildre(ParentRunner.Java:288)在org.junit.runners.ParentRunner.Access$000（.rt.execution.junit.IdeateStrunner$repeater.startrunnerWithArgs（ideateStrunner.java:47)位于com.intellij.rt.execution.junit.junitstarter.PrepareStreamsandStart（junitstarter.java:242)位于com.intellij.rt.execution.junit.junit.main（junitstarter.java:70)

我使用的是h2数据库，在启动时我用import.sql和schema-h2.html" target="_blank">sql脚本提供的数据填充数据库。我正在使用spring-boot 2.0.5版和spring批处理核心4.1.2版。

Answer 1

这是不正确的。您在测试范围之外也会遇到同样的问题。正如本JIRA bug中所评论的，您需要定义bean来返回ItemStreamReader对象，而不是JPapAgingItemReader。

以下是吉米·普雷特的原话

没有调用writer的open（）方法，因为writer没有注册为流。您的bean定义方法应该返回一个ItemStreamReader而不是ItemReader，这样它就会自动注册为一个流。

下面是参考https://jira.spring.io/si/jira.issueViews:issue-html/batch-2256/batch-2256.html

Answer 2

正如Mahmoud Ben Hassine在评论中所建议的，我将测试类更改为：

@Slf4j
//@SpringBootTest
@SpringBatchTest
@RunWith(SpringRunner.class)
@EnableAutoConfiguration
@ContextConfiguration(classes = {
        PaySalaryJobConfiguration.class
})
@ActiveProfiles("test")
public class StepScopeIntegrationTest {

    @Autowired
    private JpaPagingItemReader<Employee> someoneReader;

    @Autowired
    private JdbcTemplate jdbcTemplate;

    private StepExecution stepExecution;

    @Before
    public void setUp() {
        log.info("Before execution we have {} entries", jdbcTemplate.queryForObject("SELECT COUNT(*) FROM EMPLOYEE", Integer.class ));
    }

    public StepExecution getStepExecution() {
        log.info("Step Execution !");
        stepExecution = MetaDataInstanceFactory.createStepExecution();
        log.info("Context = {} / Step = {} ", stepExecution.getExecutionContext(), stepExecution.getStepName());
        return stepExecution;
    }

    @Test
    public void testReader() throws Exception {
        log.info("Page = {}", someoneReader.getPage());
        log.info("PageSize = {}", someoneReader.getPageSize());
        someoneReader.open(stepExecution.getExecutionContext());
        Assert.assertNotNull(someoneReader.read());
    }
}

现在测试工作很好。