2017-12-27 142 views
2

我有表A中,其中列中的一个是像转换重复记录到重复的字符串数组

   +- children: record (repeated) 
       | |- name: string 
       | |- gender: string 
       | |- age: integer 

我有一个表B,其中所述柱中的一个是STRING一个重复RECORD(重复)

   +- names : string (repeated) 

寻找选项将名称列表从RECORD中的表A移动到表B的字符串数组。

任何建议都会有很大的帮助

回答

2

可以使用ARRAY功能。试试这个:

#standardSQL 
SELECT 
    ARRAY_TO_STRING(
    ARRAY(SELECT name FROM UNNEST(children)) 
) AS names 
FROM `dataset.table` 

它创建正好从结构内的name领域的新阵列,然后将数组转换为字符串。

0

下面是BigQuery的标准SQL

如果你希望得到阵列您可以使用下面

#standardSQL 
SELECT ARRAY(SELECT name FROM UNNEST(children)) AS names 
FROM `yourproject.yourdataset.yourtable` 

您可以测试/使用虚拟数据作为

#standardSQL 
WITH `yourproject.yourdataset.yourtable` AS (
    SELECT [STRUCT<name STRING, gender STRING, age INT64>('abc1','m',12),('xyz1','m',13),('uvw1','f',14)] children UNION ALL 
    SELECT [STRUCT<name STRING, gender STRING, age INT64>('abc2','f',12),('xyz2','m',13),('uvw2','f',14)] 
) 
SELECT ARRAY(SELECT name FROM UNNEST(children)) AS names 
FROM `yourproject.yourdataset.yourtable` 
它玩

输出为

Row names  
1 abc1  
    xyz1  
    uvw1  
2 abc2  
    xyz2  
    uvw2  

在情况下,如果你期望的字符串

#standardSQL 
SELECT (SELECT STRING_AGG(name) FROM UNNEST(children)) AS names 
FROM `yourproject.yourdataset.yourtable` 

您可以测试/使用相同的虚拟数据

#standardSQL 
WITH `yourproject.yourdataset.yourtable` AS (
    SELECT [STRUCT<name STRING, gender STRING, age INT64>('abc1','m',12),('xyz1','m',13),('uvw1','f',14)] children UNION ALL 
    SELECT [STRUCT<name STRING, gender STRING, age INT64>('abc2','f',12),('xyz2','m',13),('uvw2','f',14)] 
) 
SELECT (SELECT STRING_AGG(name) FROM UNNEST(children)) AS names 
FROM `yourproject.yourdataset.yourtable` 

和输出现在是

Row names  
1 abc1,xyz1,uvw1 
2 abc2,xyz2,uvw2 
它玩