Spring 批量分页问题

Question

我正在处理 Spring 批处理，我正在使用分页从数据库中获取数据。这是它第一次按预期触发查询，但对于第二页，它又添加了一个意想不到的参数。我不明白为什么。

我在 XML 文件中的查询是

<bean id="summaryAppReader" scope="step" autowire-candidate="false"
    class="org.springframework.batch.item.database.JdbcPagingItemReader">
    <property name="dataSource" ref="dataSource" />
    <property name="rowMapper">
        <bean class="com.majesco.nyl.batch.mapper.SummaryApplicationRowMapper" />
    </property>
    <property name="queryProvider">
        <bean
            class="org.springframework.batch.item.database.support.SqlPagingQueryProviderFactoryBean">
            <property name="dataSource" ref="dataSource" />
            <property name="fromClause" value="VIEW_SUMMARY_APPLICATION" />
            <property name="selectClause"
                value="application_id, upper(application_status)  as application_status,relationship_code, create_date,submission_date,transmission_date,TRANSMISSION_STATUS,tpa_notes,site_details_id,transaction_num,PROCESS_STATUS,process_date,application_status_date,first_name,middle_name,last_name,suffix,prefix,APPROVED_REJECTED_DATE,action_by,ass_acronym,user_name,applicant_name,site_name,ASSOCIATION_ID, (TO_DATE('#{jobParameters['runDate']}', 'yyyy-mm-dd') - trunc(create_date)) as elapsedDays" />
            <property name="sortKeys">
                <map>
                    <entry key="transaction_num" value="ASCENDING" />
                </map>
            </property>
            <property name="whereClause"
                value="(upper(application_status)='PENDING' OR (upper(application_status) IN ('APPROVED','DECLINED') AND trunc(application_status_date) = TO_DATE('#{jobParameters['runDate']}', 'yyyy-mm-dd'))) and TRANSACTION_NUM &gt;= :minId and TRANSACTION_NUM &lt;= :maxId" />
        </bean>
    </property>
    <property name="parameterValues">
        <map>
            <entry key="minId" value="#{stepExecutionContext[minValue]}" />
            <entry key="maxId" value="#{stepExecutionContext[maxValue]}" />
        </map>
    </property>
</bean>

为了获取第一页的数据，它的触发查询是这样的，它是预期的

Reading page 0
SQL used for reading first page: [SELECT * FROM (SELECT application_id, upper(application_status)  as application_status,relationship_code, create_date,submission_date,transmission_date,TRANSMISSION_STATUS,tpa_notes,site_details_id,transaction_num,PROCESS_STATUS,process_date,application_status_date,first_name,middle_name,last_name,suffix,prefix,APPROVED_REJECTED_DATE,action_by,ass_acronym,user_name,applicant_name,site_name,ASSOCIATION_ID, (TO_DATE('2016-06-29', 'yyyy-mm-dd') - trunc(create_date)) as elapsedDays FROM VIEW_SUMMARY_APPLICATION WHERE (upper(application_status)='PENDING' OR (upper(application_status) IN ('APPROVED','DECLINED') AND trunc(application_status_date) = TO_DATE('2016-06-29', 'yyyy-mm-dd'))) and TRANSACTION_NUM >= :minId and TRANSACTION_NUM <= :maxId ORDER BY transaction_num ASC) WHERE ROWNUM <= 10]
Using parameterMap:{minId=100000002630, maxId=100000002663}

但是对于第二页，它在查询 _TRANSACTION_NUM 中又添加了一个参数，这与

一样出乎意料

Reading page 1
SQL used for reading remaining pages: [SELECT * FROM (SELECT application_id, upper(appication_status)  as application_status,relationship_code, create_date,submission_date,transmission_date,TRANSMISSION_STATUS,tpa_notes,site_details_id,transaction_num,PROCESS_STATUS,process_date,application_status_date,first_name,middle_name,last_name,suffix,prefix,APPROVED_REJECTED_DATE,action_by,ass_acronym,user_name,applicant_name,site_name,ASSOCIATION_ID, (TO_DATE('2016-06-29', 'yyyy-mm-dd') - trunc(create_date)) as elapsedDays FROM VIEW_SUMMARY_APPLICATION WHERE (upper(application_status)='PENDING' OR (upper(application_status) IN ('APPROVED','DECLINED') AND trunc(application_status_date) = TO_DATE('2016-06-29', 'yyyy-mm-dd'))) and TRANSACTION_NUM >= :minId and TRANSACTION_NUM <= :maxId ORDER BY transaction_num ASC) WHERE ROWNUM <= 10 AND ((transaction_num > :_transaction_num))]
Using parameterMap:{minId=100000002596, maxId=100000002629, _transaction_num=100000002622}

仅供参考。 TRANSACTION_NUM 不是唯一的。

Answer 1

这是由于您的 sortKeys。您定义了以下内容：

        <property name="sortKeys">
            <map>
                <entry key="transaction_num" value="ASCENDING" />
            </map>
        </property>

Spring 批处理正在使用第 0 页中最高的 transaction_num，并表示下一页的 transaction_num 必须高于该值。

框架将在每个页面递增绑定变量，以确保您始终获得唯一的结果。这里的问题是您实际上需要一个主键才能正确分页，否则您的 pageSize（默认值 = 10 并看到它在您的查询中添加 WHERE ROWNUM <= 10 的位置）将切断您的一些记录。

Spring 批量分页问题

Spring Batch Paging Issue

spring

batch-processing

spring-batch