Question

我有一个包含结构数组的表。有没有办法使用like运算符来过滤此列中的记录？

hive> desc location;
location_list           array<struct<city:string,state:string>>

hive> select * from location;
row1 : [{"city":"Hudson","state":"NY"},{"city":"San Jose","state":"CA"},{"city":"Albany","state":"NY"}]
row2 : [{"city":"San Jose","state":"CA"},{"city":"San Diego","state":"CA"}]

我正在尝试运行这样的查询，只过滤那些具有“NY”状态的记录。

hive> select * from location where location_list like '%"NY"%';
FAILED: SemanticException [Error 10014]: Line 1:29 Wrong arguments ''%"NY"%'': No matching method for class org.apache.hadoop.hive.ql.udf.UDFLike with (array<struct<city:string,state:string>>, string). Possible choices: _FUNC_(string, string)

注意：我可以通过做侧视图来做到这一点。爆炸这个结构列。但是试图避免它，因为我需要将这个表与另一个不接受横向视图的表联系起来。

Answer 1

不错的问题，你可以用以下有效（和漂亮）的方式来做。

select * from location 
where array_contains(location_list.state, 'NY');

在这种情况下，location_list.state将创建一个字符串数组（在您的情况下为状态），因此您可以使用UDF array_contains进行值检查。这将寻找准确的值，您将无法像like运算符那样执行匹配，但您应该能够实现您正在寻找的内容

Answer 2

array_contains的演示：

select my_array  
from
( --emulation of your dataset. Just replace this subquery with your table
 select array(named_struct("city","Hudson","state","NY"),named_struct("city","San Jose","state","CA"),named_struct("city","Albany","state","NY")) as my_array
 union all
 select array(named_struct("city","San Jose","state","CA"),named_struct("city","San Diego","state","CA")) as my_array
)s
where array_contains(my_array.state,'NY') 
;

结果：

OK
[{"city":"Hudson","state":"NY"},{"city":"San Jose","state":"CA"},{"city":"Albany","state":"NY"}]
Time taken: 34.055 seconds, Fetched: 1 row(s)

对STRUCT数据类型使用like运算符

2 个答案: