BigQuery SPLIT 手动创建表

Posted 2023-03-24

【中文标题】BigQuery SPLIT 手动创建表

【英文标题】：BigQuery SPLIT manual creation of table

【发布时间】：2017-07-13 09:36:45

【问题描述】：

参考这个question，我想手动创建多个列。

SELECT SPLIT(Titles) AS Title 
FROM (SELECT 'Title 1,Title 2,Title 3,Title 4' AS Titles)

我尝试过像这样简单地添加新列：

SELECT SPLIT(Titles) AS Title, SPLIT(Names) AS Name,FROM (SELECT 'Title 1,Title 2,Title 3,Title 4' AS Titles, 'Name 1,Name 2,Name 3,Name 4' AS NAMES)

但是 BQ 向我显示以下错误：

Error: Cannot output multiple independently repeated fields at the same time. Found Title and Name

我认为这可能与 BQ 如何使结果变平有关，我发现了类似的问题 here。不幸的是，我无法转换我的代码。我只能使用旧版 SQL。

编辑： 预期的表格应如下所示：

-- +---------+--------+
-- | Title   | Name   |
-- +---------+--------+
-- | Title 1 | Name 1 |
-- | Title 2 | Name 2 | 
-- | Title 3 | Name 3 |
-- | Title 4 | Name 4 |
-- +---------+--------+

【问题讨论】：

您能否提供您期望的确切结果，因为它可能会有所不同

我编辑了这篇文章。基本上它应该添加具有相同行数的列。

【参考方案1】：

以下是 BigQuery 标准 SQL

#standardSQL
WITH data AS (
  SELECT 'Title 1,Title 2,Title 3,Title 4' AS Titles, 'Name 1,Name 2,Name 3,Name 4' AS Names
)
SELECT 
  Title, 
  Name
FROM data, 
  UNNEST(SPLIT(Titles)) AS Title WITH OFFSET AS pos1, 
  UNNEST(SPLIT(Names)) AS Name WITH OFFSET AS pos2
WHERE pos1 = pos2  
ORDER BY Title  

BigQuery Legacy SQL 中的相同想法看起来更加繁琐

#legacySQL
SELECT
  Title, Name
FROM FLATTEN((
  SELECT Title,  POSITION(Title) AS pos1
  FROM (
    SELECT SPLIT(Titles) AS Title
    FROM (SELECT 'Title 1,Title 2,Title 3,Title 4' AS Titles, 'Name 1,Name 2,Name 3,Name 4' AS Names)
  )
), pos1) AS titles
JOIN FLATTEN((
  SELECT Name, POSITION(Name) AS pos2
  FROM (
    SELECT SPLIT(Names) AS Name
    FROM (SELECT 'Title 1,Title 2,Title 3,Title 4' AS Titles, 'Name 1,Name 2,Name 3,Name 4' AS Names)
  )
), pos2) AS names
ON pos1 = pos2

【讨论】：

@johnsnow - 你有机会尝试吗？