Question

假设我希望使用JOIN过滤掉一个带有LATERAL VIEW的巨大表格。例如：

SELECT things.*
FROM a_few_objects fo join
 (SELECT an_id,
        object.id as object_id,
        object.big_thing
  FROM big_table
  LATERAL VIEW EXPLODE(outer_things) ot AS outer_thing
  LATERAL VIEW EXPLODE(objects) o AS object) things
ON fo.objectid = things.object_id

是否有更有效的方法来实例化“事物”子查询（我认为在磁盘上放置一个巨大的表）？

我可以将a_few_objects的行放在内部查询的WHERE子句中，但是我想在Hive中指定a_few_objects，所以我不必一直更改查询。

如果我将连接放在子查询中，Hive就会感到困惑。我正在使用Hive 0.13。

Answer 1

Hive 0.13有CTE（公用表格式）。

WITH q1 AS ( 
    SELECT an_id,
        object.id AS object_id,
        object.big_thing
    FROM big_table
    LATERAL VIEW EXPLODE(outer_things) ot AS outer_thing
    LATERAL VIEW EXPLODE(objects) o AS object) 

SELECT *
FROM q1
JOIN big_table fo
ON q1.objectid = fo.object_id

如何过滤蜂巢中的侧视图？

1 个答案: