MongoDB
 sql >> Base de Dados >  >> NoSQL >> MongoDB

criando lógica dentro do Spring Batch usando Tasklet ou dentro do CompositeItemWriter?


Você precisa usar o ClassifierCompositeItemWriter para classificar os itens e escrever cada tipo em seu arquivo correspondente. Aqui está um exemplo rápido que você pode tentar:
import java.util.Arrays;

import org.springframework.batch.core.Job;
import org.springframework.batch.core.JobParameters;
import org.springframework.batch.core.Step;
import org.springframework.batch.core.configuration.annotation.EnableBatchProcessing;
import org.springframework.batch.core.configuration.annotation.JobBuilderFactory;
import org.springframework.batch.core.configuration.annotation.StepBuilderFactory;
import org.springframework.batch.core.launch.JobLauncher;
import org.springframework.batch.item.ItemReader;
import org.springframework.batch.item.ItemWriter;
import org.springframework.batch.item.file.FlatFileItemWriter;
import org.springframework.batch.item.file.builder.FlatFileItemWriterBuilder;
import org.springframework.batch.item.file.transform.PassThroughLineAggregator;
import org.springframework.batch.item.support.ClassifierCompositeItemWriter;
import org.springframework.batch.item.support.ListItemReader;
import org.springframework.classify.Classifier;
import org.springframework.context.ApplicationContext;
import org.springframework.context.annotation.AnnotationConfigApplicationContext;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.core.io.FileSystemResource;

@Configuration
@EnableBatchProcessing
public class MyJob {

    private JobBuilderFactory jobBuilderFactory;

    private StepBuilderFactory stepBuilderFactory;

    public MyJob(JobBuilderFactory jobBuilderFactory, StepBuilderFactory stepBuilderFactory) {
        this.jobBuilderFactory = jobBuilderFactory;
        this.stepBuilderFactory = stepBuilderFactory;
    }

    @Bean
    public ItemReader<Person> itemReader() {
        Person foo1 = new Person();foo1.setId(1);foo1.setName("foo1");
        Person foo2 = new Person();foo2.setId(2);foo2.setName("foo2");
        Person bar1 = new Person();bar1.setId(3);bar1.setName("bar1");
        Person bar2 = new Person();bar2.setId(4);bar2.setName("bar2");
        return new ListItemReader<>(Arrays.asList(foo1, foo2, bar1, bar2));
    }

    @Bean
    public ClassifierCompositeItemWriter<Person> classifierCompositeItemWriter(ItemWriter<Person> fooItemWriter, ItemWriter<Person> barItemWriter) {
        ClassifierCompositeItemWriter<Person> classifierCompositeItemWriter = new ClassifierCompositeItemWriter<>();
        classifierCompositeItemWriter.setClassifier((Classifier<Person, ItemWriter<? super Person>>) person -> {
            if (person.getName().startsWith("foo")) {
                return fooItemWriter;
            } else {
                return barItemWriter;
            }
        });
        return classifierCompositeItemWriter;
    }

    @Bean
    public FlatFileItemWriter<Person> fooItemWriter() {
        return new FlatFileItemWriterBuilder<Person>()
                .name("fooItemWriter")
                .resource(new FileSystemResource("foos.txt"))
                .lineAggregator(new PassThroughLineAggregator<>())
                .build();
    }

    @Bean
    public FlatFileItemWriter<Person> barItemWriter() {
        return new FlatFileItemWriterBuilder<Person>()
                .name("barItemWriter")
                .resource(new FileSystemResource("bars.txt"))
                .lineAggregator(new PassThroughLineAggregator<>())
                .build();
    }

    @Bean
    public Step dataExtractionStep() {
        return stepBuilderFactory.get("dataExtractionStep")
                .<Person, Person>chunk(2)
                .reader(itemReader())
                .writer(classifierCompositeItemWriter(fooItemWriter(), barItemWriter()))
                .stream(fooItemWriter())
                .stream(barItemWriter())
                .build();
    }

    @Bean
    public Job dataExtractionJob() {
        return jobBuilderFactory.get("dataExtractionJob")
                .start(dataExtractionStep())
                .build();
    }

    public static void main(String[] args) throws Exception {
        ApplicationContext context = new AnnotationConfigApplicationContext(MyJob.class);
        JobLauncher jobLauncher = context.getBean(JobLauncher.class);
        Job job = context.getBean(Job.class);
        jobLauncher.run(job, new JobParameters());
    }

    public static class Person {

        private int id;

        private String name;

        public Person() {
        }

        public Person(int id, String name) {
            this.id = id;
            this.name = name;
        }

        public int getId() {
            return id;
        }

        public void setId(int id) {
            this.id = id;
        }

        public String getName() {
            return name;
        }

        public void setName(String name) {
            this.name = name;
        }

        @Override
        public String toString() {
            return "Person{" +
                    "id=" + id +
                    ", name='" + name + '\'' +
                    '}';
        }
    }

}

Esta amostra lê algum Person itens e escreve aqueles com o nome foo* para foos.txt e aqueles com nome bar* para bars.txt .

Espero que isto ajude.