How to normalize table1 in postgres sql?

Question

我需要通过拆分“错误发生在”列来规范化（以技术方式）表 1，使其与表 2 中的一样。

是否有任何 DML 函数可以帮助我解决这个问题？这样就可以在不使用超前滞后功能的情况下传输其余信息？

我尝试使用 string_to_array 和转置函数。

Answer 1

WITH split_dates AS (
  SELECT other_columns..., unnest(string_to_array(error_occured_at, ', '))::date AS date
  FROM table1
  WHERE error_occured_at RLIKE "," -- Only considering rows that actually do contain a comma
)
INSERT INTO table1 (other_columns..., error_occured_at)
SELECT other_columns..., date
FROM split_dates;

如果你想删除

error_occured_at

是旧格式的行，你可以这样做：

DELETE FROM table1
WHERE error_occured_at RLIKE ","

Answer 2

您可以将

unnest

与

string_to_array

结合起来，如下所示：

SELECT *
FROM   mytable, unnest(string_to_array(error_occured_at, ', ')) error_occured_at