Spring Boot Batch设计咨询：条件执行步骤与周调度实现

完全可以实现！这在Spring Boot Batch里是非常典型的条件执行场景，我来给你拆解下最优的实现思路和具体代码：

要实现「仅在步骤1查询到数据时才执行步骤2」，关键是用**JobExecutionDecider**做流程决策。它能读取步骤1的执行结果，动态决定后续流程走向——要么执行删除步骤，要么直接结束Job。

具体逻辑是：

1. 步骤1：查询数据库目标数据的数量，把结果存入Job的执行上下文（JobExecutionContext）
2. 决策器：从上下文读取查询结果，判断是否存在数据，返回对应的状态标记
3. Job配置：根据决策器的标记，路由到删除步骤或直接结束流程

1. 步骤1：查询数据的Tasklet #

用Tasklet实现简单的查询逻辑，把结果存入上下文：

@Component
public class CheckDataTasklet implements Tasklet {

    private final JdbcTemplate jdbcTemplate;

    public CheckDataTasklet(JdbcTemplate jdbcTemplate) {
        this.jdbcTemplate = jdbcTemplate;
    }

    @Override
    public RepeatStatus execute(StepContribution contribution, ChunkContext chunkContext) throws Exception {
        // 替换成你的查询条件
        int recordCount = jdbcTemplate.queryForObject(
            "SELECT COUNT(*) FROM your_target_table WHERE your_condition", 
            Integer.class
        );
        // 把是否存在数据的标记存入Job上下文
        chunkContext.getStepContext().getJobExecutionContext()
            .put("hasTargetData", recordCount > 0);
        return RepeatStatus.FINISHED;
    }
}

2. 流程决策器 #

自定义JobExecutionDecider，根据上下文的标记返回不同状态：

@Component
public class DataExistsDecider implements JobExecutionDecider {

    @Override
    public FlowExecutionStatus decide(JobExecution jobExecution, StepExecution stepExecution) {
        boolean hasData = (boolean) jobExecution.getExecutionContext().get("hasTargetData");
        return hasData ? new FlowExecutionStatus("DATA_EXISTS") : new FlowExecutionStatus("NO_DATA");
    }
}

3. Job与Step配置 #

把步骤、决策器组装成完整的Job：

@Configuration
@EnableBatchProcessing
public class BatchJobConfig {

    private final JobBuilderFactory jobBuilderFactory;
    private final StepBuilderFactory stepBuilderFactory;
    private final CheckDataTasklet checkDataTasklet;
    private final DataExistsDecider dataExistsDecider;
    private final JdbcTemplate jdbcTemplate;

    // 构造注入依赖
    public BatchJobConfig(JobBuilderFactory jobBuilderFactory, StepBuilderFactory stepBuilderFactory,
                          CheckDataTasklet checkDataTasklet, DataExistsDecider dataExistsDecider,
                          JdbcTemplate jdbcTemplate) {
        this.jobBuilderFactory = jobBuilderFactory;
        this.stepBuilderFactory = stepBuilderFactory;
        this.checkDataTasklet = checkDataTasklet;
        this.dataExistsDecider = dataExistsDecider;
        this.jdbcTemplate = jdbcTemplate;
    }

    // 步骤1：检查数据
    @Bean
    public Step checkDataStep() {
        return stepBuilderFactory.get("checkDataStep")
                .tasklet(checkDataTasklet)
                .build();
    }

    // 步骤2：删除数据（如果数据量大，建议用Chunk模式替代Tasklet）
    @Bean
    public Step deleteDataStep() {
        return stepBuilderFactory.get("deleteDataStep")
                .tasklet((contribution, chunkContext) -> {
                    // 替换成你的删除逻辑
                    jdbcTemplate.update(
                        "DELETE FROM your_target_table WHERE your_condition"
                    );
                    return RepeatStatus.FINISHED;
                })
                .build();
    }

    // 完整Job流程
    @Bean
    public Job dataCleanupJob() {
        return jobBuilderFactory.get("dataCleanupJob")
                .start(checkDataStep())
                .next(dataExistsDecider)
                // 存在数据则执行删除步骤
                .on("DATA_EXISTS").to(deleteDataStep())
                // 无数据则直接结束Job
                .on("NO_DATA").end()
                .end()
                .build();
    }
}

用Spring的@EnableScheduling实现定时触发，配合Cron表达式设置每周执行：

@Configuration
@EnableScheduling
public class BatchScheduler {

    private final JobLauncher jobLauncher;
    private final Job dataCleanupJob;

    public BatchScheduler(JobLauncher jobLauncher, Job dataCleanupJob) {
        this.jobLauncher = jobLauncher;
        this.dataCleanupJob = dataCleanupJob;
    }

    // 示例：每周日凌晨2点执行，可根据需求调整Cron表达式
    @Scheduled(cron = "0 0 2 * * SUN")
    public void runDataCleanupJob() throws Exception {
        // 必须添加唯一参数（比如时间戳），避免Spring Batch认为是重复Job实例
        JobParameters jobParameters = new JobParametersBuilder()
                .addLong("executionTimestamp", System.currentTimeMillis())
                .toJobParameters();
        jobLauncher.run(dataCleanupJob, jobParameters);
    }
}

• 如果删除的数据量较大，建议把步骤2改成Chunk模式（ItemReader+ItemProcessor+ItemWriter），利用Spring Batch的事务管理和批处理优化能力
• 确保JobParameters每次都唯一，否则Spring Batch会跳过重复的Job实例
• 可以结合Spring Boot Actuator监控Batch任务的执行状态和历史记录

内容的提问来源于stack exchange，提问作者Amit Kumar