Spring Batch FlatFileItemWriter仅在第二次运行后才处理数据

Question

我有一个春季批处理，可以在步骤1中通过sftp下载多个csv文件，然后在步骤2中对其进行一些处理。

一切正常，但仅当我两次启动该程序时（以便在启动该程序时已存在下载的文件）。

似乎软件在启动时不在文件中就看不到文件。

我想这与在执行step1()之前读取@Value 。

@Configuration
@EnableBatchProcessing
public class BatchConfiguration {
    private static final Logger log = LoggerFactory.getLogger(BillProcessor.class);
    @Autowired
    public JobBuilderFactory jobBuilderFactory;

    @Autowired
    public StepBuilderFactory stepBuilderFactory;

    @Bean
    public Job mainJob() throws IOException, InterruptedException {
        return jobBuilderFactory
                .get("mainJob")
                .incrementer(new RunIdIncrementer())
                .start(step1())
                .next(step2())
                .build();
    }

    @Bean
    protected Step step1() {
        return stepBuilderFactory
                .get("step1")
                .tasklet(sftpTransferer())
                .build();
    }

    @Bean
    public Step step2() throws IOException, InterruptedException {
        return stepBuilderFactory.get("step2").<Bill, Bill>chunk(1000)
                .reader(multiResourceItemReader())
                .writer(writer())
                .build();
    }

    @Bean
    public SftpTransferer sftpTransferer() {
        SftpTransferer tasklet = new SftpTransferer();
        return tasklet;
    }

    @Bean
    public FlatFileItemReader<Bill> reader() throws IOException, InterruptedException {
        log.info("start reader");
        ...
        return reader;
    }

    @Value("file:*.tsv")
    private Resource[] inputResources;

    @Bean
    public MultiResourceItemReader<Bill> multiResourceItemReader() throws InterruptedException, IOException {
        log.info("start MultiResourceItemReader");
        MultiResourceItemReader<Bill> resourceItemReader = new MultiResourceItemReader<Bill>();
        resourceItemReader.setResources(inputResources);
        resourceItemReader.setDelegate(reader());
        return resourceItemReader;
    }

    @Bean
    public BillProcessor processor() {
        log.info("start processor");
        return new BillProcessor();
    }


    @Bean
    public FlatFileItemWriter<Bill> writer() throws IOException {
        log.info("start writer");

        FlatFileItemWriter<Bill> writer = new FlatFileItemWriter<>();
        writer.setResource(new FileSystemResource("_accounting.csv"));
        log.info("set append to true");
        writer.setAppendAllowed(true);
        HeaderWriter headerWriter = new HeaderWriter("id;path;processed;records");
        writer.setHeaderCallback(headerWriter);

        writer.setLineAggregator(new DelimitedLineAggregator<Bill>() {{
            setDelimiter(";");
            setFieldExtractor(new BeanWrapperFieldExtractor<Bill>() {{
                setNames(new String[]{"id", "path", "processed", "records"});
            }});
        }});
        return writer;
    }
}

我在网上找不到任何东西。 任何帮助表示赞赏！

Answer 1

我建议使bean multiresourceItemReader()逐步作用域，并使其接受inputResources作为参数：

    @Bean
    public Step step2() throws IOException, InterruptedException {
        return stepBuilderFactory.get("step2").<Bill, Bill>chunk(1000)
                .reader(multiResourceItemReader(null))
                .writer(writer())
                .build();
    }

...

    @StepScope
    @Bean
    public MultiResourceItemReader<Bill> multiResourceItemReader(@Value("file:*.tsv") Resource[] inputResources) throws InterruptedException, IOException {
        log.info("start MultiResourceItemReader");
        MultiResourceItemReader<Bill> resourceItemReader = new MultiResourceItemReader<Bill>();
        resourceItemReader.setResources(inputResources);
        resourceItemReader.setDelegate(reader());
        return resourceItemReader;
    }

这样，收集inputResources时间将推迟到执行step1之后，并且文件实际存在。