识别多个表 Snowflake SQL 之间的唯一列

Posted 2023-02-26

【中文标题】识别多个表 Snowflake SQL 之间的唯一列

【英文标题】：Identifying unique columns between multiple table Snowflake SQL

【发布时间】：2021-06-23 14:58:35

【问题描述】：

在执行“UNION ALL”时，我必须识别多个表之间的唯一列。

例如：Employee_1、Employee_2、Employee_3

Table1: Employee_1

|Emp_Id|Joined_Mnth|Store_NO | Marked_YR | 
+------+-----------+---------+-----------+
| 1    |March      |100020   | 2018      |
| 2    |April      |120004   | 2018      |
| 3    |January    |100032   | 2019      |
| 4    |October    |231009   | 2019      |

Table2: Employee_2

|Emp_Id|Store_NO | Marked_YR | 
+------+---------+-----------+
| 1    |100020   | 2018      |
| 2    |120004   | 2018      |

Table3: Employee_3

|Emp_Id|Joined_Mnth|Store_NO | Closed_YR | 
+------+-----------+---------+-----------+
| 1    |March      |100020   | 2020      |
| 2    |April      |120004   | 2018      |
| 7    |January    |100032   | 2021      |
| 8    |October    |231009   | 2019      |

视图中的输出：

CREATE OR REPLACE VIEW Employee AS
SELECT Emp_Id,Store_NO FROM Employee_1
UNION ALL
SELECT Emp_Id,Store_NO FROM Employee_2
UNION ALL
SELECT Emp_Id,Store_NO FROM Employee_3

|Emp_Id|Store_NO |  ==> Common columns between Employee_1,Employee_2,Employee_3
+------+---------+
| 1    |100020   |
| 2    |120004   |
| 3    |100032   |
| 4    |231009   |
| 1    |100020   |
| 2    |120004   |
| 1    |100020   |
| 2    |120004   |
| 7    |100032   |
| 8    |231009   |

如何识别上述所有表之间的公共列？

【参考方案1】：

你可以使用information_schema.columns:

select column_name
from information_schema.columns
where table_name in ('Employee_1', 'Employee_2', 'Employee_3')
group by column_name
having count(*) = 3;

【参考方案2】：

您可以使用INFORMATION_SCHEMA.COLUMNS 获取列列表：

SELECT COLUMN_NAME
FROM INFORMATION_SCHEMA.COLUMNS
WHERE TABLE_NAME ILIKE ANY ('Employee_1', 'Employee_2', 'Employee_3')
GROUP BY COLUMN_NAME
HAVING COUNT(*) = 3;

---

编辑：

可以使用NATURAL FULL JOIN组合多个表（感谢Lukas Eder，related post）：

CREATE OR REPLACE VIEW Employee AS
SELECT *
FROM (SELECT *, 'Employee_1' AS tab FROM Employee_1) sub1
NATURAL FULL JOIN (SELECT *, 'Employee_2' AS tab FROM Employee_2) sub2
NATURAL FULL JOIN (SELECT *, 'Employee_3' AS tab FROM Employee_3) sub3;

这是一种OUTER UNION CORRESPONDING 实现。

【讨论】：

有没有办法避免不匹配的列？

@K.Tom 使用NATURAL FULL JOIN，没有。