Conditional Flow in Spring Batch

I just announced the new Learn Spring course, focused on the fundamentals of Spring 5 and Spring Boot 2:

CHECK OUT THE COURSE

1. Introduction

We use Spring Batch to compose jobs from multiple steps that read, transform, and write data. If the steps in a job have multiple paths, similar to using an if statement in our code, we say that the job flow is conditional.

In this tutorial, we'll look at two ways to create Spring Batch jobs with a conditional flow.

2. Exit Status and Batch Status

When we specify a conditional step with Spring's Batch framework, we're using the exit status of a step or job. Therefore, we need to understand the difference between Batch Status and Exit Status in our steps and jobs:

  • BatchStatus is an Enum representing the status of a step/job and is used by the Batch framework internally
  • Possible values are: ABANDONED, COMPLETED, FAILED, STARTED, STARTING, STOPPED, STOPPING, UNKNOWN
  • ExitStatus is the status of a step when execution is completed and is used to conditionally determine the flow

By default, the ExitStatus of a step or job is the same as its BatchStatus. We can also set a custom ExitStatus to drive flow.

3. Conditional Flow

Let's say we have an IOT device that sends us measurements. Our device measurements are arrays of integers, and we need to send notifications if any of our measurements contain positive integers.

In other words, we need to send a notification when we detect a positive measurement.

3.1. ExitStatus

Importantly, we use the exit status of a step to drive conditional flow.

To set the exit status of a step, we need to use the StepExecution object's setExitStatus method. In order to do that, we need to create an ItemProcessor that extends ItemListenerSupport and gets the step's StepExecution.

We use this to set our step's exit status to NOTIFY when we find a positive number. When we determine our exit status based on data within the batch job, we can use an ItemProcessor.

Let's look at our NumberInfoClassifier to see the three methods we need:

public class NumberInfoClassifier extends ItemListenerSupport<NumberInfo, Integer>
implements ItemProcessor<NumberInfo, Integer> { private StepExecution stepExecution; @BeforeStep
public void beforeStep(StepExecution stepExecution) {
this.stepExecution = stepExecution;
this.stepExecution.setExitStatus(new ExitStatus(QUIET));
} @Override
public Integer process(NumberInfo numberInfo) throws Exception {
return Integer.valueOf(numberInfo.getNumber());
} @Override
public void afterProcess(NumberInfo item, Integer result) {
super.afterProcess(item, result);
if (item.isPositive()) {
stepExecution.setExitStatus(new ExitStatus(NOTIFY));
}
}
}

Note: we use the ItemProcessor to set the ExitStatus in this example, but we could just as easily do it in our step's ItemReader or ItemWriter.

Finally, when we create our job, we tell our JobBuilderFactory to send notifications for any step that exits with a status of NOTIFY:

jobBuilderFactory.get("Number generator - second dataset")
.start(dataProviderStep)
.on("NOTIFY").to(notificationStep)
.end()
.build();

Also note that when we have additional conditional branches and multiple exit codes, we can add them to our job with the from and on methods of the JobBuilderFacotry:

jobBuilderFactory.get("Number generator - second dataset")
.start(dataProviderStep)
.on("NOTIFY").to(notificationStep)
.from(step).on("LOG_ERROR").to(errorLoggingStep)
.end()
.build();

Now, any time our ItemProcessor sees a positive number, it will direct our job to run the notificationStep, which simply prints a message to System.out:

Second Dataset Processor 11
Second Dataset Processor -2
Second Dataset Processor -3
[Number generator - second dataset] contains interesting data!!

If we had a data set without a positive number, we would not see our notificationStep message:

Second Dataset Processor -1
Second Dataset Processor -2
Second Dataset Processor -3

3.2. Programmatic Branching With JobExecutionDecider

Alternatively, we can use a class that implements JobExecutionDecider to determine job flow. This is especially useful if we have external factors for determining execution flow.

To use this method, we first need to modify our ItemProcessor to remove the ItemListenerSupport interface and @BeforeStep method:

public class NumberInfoClassifierWithDecider extends ItemListenerSupport<NumberInfo, Integer>
implements ItemProcessor<NumberInfo, Integer> { @Override
public Integer process(NumberInfo numberInfo) throws Exception {
return Integer.valueOf(numberInfo.getNumber());
}
}

Next, we create a decider class that determines the notification status of our step:

public class NumberInfoDecider implements JobExecutionDecider {

    private boolean shouldNotify() {
return true;
} @Override
public FlowExecutionStatus decide(JobExecution jobExecution, StepExecution stepExecution) {
if (shouldNotify()) {
return new FlowExecutionStatus(NOTIFY);
} else {
return new FlowExecutionStatus(QUIET);
}
}
}

Then, we set up our Job to use the decider in the flow:

jobBuilderFactory.get("Number generator - third dataset")
.start(dataProviderStep)
.next(new NumberInfoDecider()).on("NOTIFY").to(notificationStep)
.end()
.build();

4. Conclusion

In this quick tutorial, we explored two options for implementing conditional flows with Spring Batch. First, we looked at how to use the ExitStatus to control the flow of our job.

Then we took a look at how we can control flow programmatically by defining our own JobExecutionDecider.

As always, the full source code of the article is available over on GitHub.

Spring Batch(0)——控制Step执行流程的更多相关文章

  1. spring batch中控制step的走向

    1.顺序执行step: <job id="job"> <step id="stepA" parent="s1" next= ...

  2. Spring Security 案例实现和执行流程剖析

    Spring Security Spring Security 是 Spring 社区的一个顶级项目,也是 Spring Boot 官方推荐使用的安全框架.除了常规的认证(Authentication ...

  3. Spring 文件上传MultipartFile 执行流程分析

    在了解Spring 文件上传执行流程之前,我们必须知道两点: 1.Spring 文件上传是基于common-fileUpload 组件的,所以,文件上传必须引入此包 2.Spring 文件上传需要在X ...

  4. Spring MVC必须知道的执行流程

    Spring MVC的执行流程 一.名词解释 1.前端控制器(DispatcherServlet) 接收请求,响应结果,相当于转发器,中央处理器 2.处理器映射器(HandlerMapping) 根据 ...

  5. Struts 2 Spring Hibernate三大框架的执行流程以及原理

    Struts2框架 一.简介 Struts2是一个相当强大的Java Web开源框架,是一个基于POJO的Action的MVC Web框架.它基于当年的WebWork和XWork框架,继承其优点,同时 ...

  6. Spring MVC 原理介绍(执行流程)

    Spring MVC工作流程图   图一   图二    Spring工作流程描述       1. 用户向服务器发送请求,请求被Spring 前端控制Servelt DispatcherServle ...

  7. ASP.NET MVC3 中整合 NHibernate3.3、Spring.NET2.0 使用AOP执行事务处理

    方法1 <object id="ServiceOperation" type="Spring.Aop.Support.SdkRegularExpressionMet ...

  8. Spring Batch在大型企业中的最佳实践

    在大型企业中,由于业务复杂.数据量大.数据格式不同.数据交互格式繁杂,并非所有的操作都能通过交互界面进行处理.而有一些操作需要定期读取大批量的数据,然后进行一系列的后续处理.这样的过程就是" ...

  9. Spring Batch实践

    Spring Batch在大型企业中的最佳实践 在大型企业中,由于业务复杂.数据量大.数据格式不同.数据交互格式繁杂,并非所有的操作都能通过交互界面进行处理.而有一些操作需要定期读取大批量的数据,然后 ...

随机推荐

  1. 三. 为什么要用Promise

    # 三. 为什么要用Promise /* 1.指定回调函数的方式更加灵活: 旧的:必须在启动异步任务前指定 promise:启动异步任务 => 返回promie对象 => 给promise ...

  2. 开发中常见的@NotNull,@NotBlank,@NotEmpty注解的区别

    开发中常看见@NotNull,@NotBlank,@NotEmpty三个注解,但却没有深入了解过,下面介绍一下他们的应用场景和区别 @NotNull:主要用在基本数据类型上(Int,Integer,D ...

  3. [第三章]c++学习笔记2(静态成员变量)

    静态成员:在说明前加了static关键字的对象 使用例: 基本概念 普通成员变量每个对象有各自的一份,而静态成员变量总共只有一份,为所有对象共享. 普通成员函数必须具体作用与某个对象,而静态成员函数并 ...

  4. Python基础(list与tuple)

    #list 类似于数组的概念 classmates = ['傻狗1','傻狗2','傻狗3'] # print(classmates) # print(len(classmates)) # print ...

  5. Python-Unittest多线程生成报告

    前言 selenium多线程跑用例,这个前面一篇已经解决了,如何生成一个测试报告这个是难点,刚好在github上有个大神分享了BeautifulReport,完美的结合起来,就能生成报告了. 环境必备 ...

  6. 关于Cefsharp无法拖动Dom元素的解决方法

    如图所显示,Cefsharp在嵌入网页,页面有对Dom元素的拖动的操作,独立在浏览器上对网页元素的拖动是没有问题的,但是嵌入到Cefsharp上显示禁用的图标.排查了H5的代码,没有写入禁用拖动的操作 ...

  7. web渗透工程师学习

    职位描述: 对公司网站.业务系统进行安全评估测试(黑盒.白盒测试): 对公司各类系统进行安全加固: 对公司安全事件进行响应,清理后门,根据日志分析攻击途径: 安全技术研究,包括安全防范技术,黑客技术等 ...

  8. Debugging and Running MPI in Xcode

    1.安装 mpich2 与 Xcode mpich2安装地址:/usr/local/Cellar/mpich2/3.1.4/ Xcode 版本:Version 6.2 (6C131e) 2.新建工程 ...

  9. 认识Influxdb时序数据库及Influxdb基础命令操作

    认识Influxdb时序数据库及Influxdb基础命令操作 一.什么是Influxdb,什么又是时序数据库 Influxdb是一个用于存储时间序列,事件和指标的开源数据库,由Go语言编写而成,无需外 ...

  10. MySQL_集群

    管理节点:192.168.31.66 sql节点1+data1节点:192.168.31.42 sql节点2+data2节点:192.168.31.128 llll