Question

对于我的一个大型数据库项目，我采用了一种结构，该结构使用嵌套结构构建表，并将这些表组合在一起（再次使用结构）到一个大表中。我发现这种方法可以极大地干燥我的代码，并减少了多次执行各种任务的需要。

但是，我正在努力修改结构中的内容，特别是在尝试使用解析函数rank和percent_rank时。这个问题基于this previous question of mine，我在下面添加了一个可重现的示例：

#standardSQL
WITH wide_stats AS (
  SELECT 'joe' name, 'bills' team,
    STRUCT(STRUCT(7 AS fga, 5 AS fgm) AS o, STRUCT(8 AS fga, 3 AS fgm) AS d) AS t1,
    STRUCT(STRUCT(3 AS fga, 4 AS fgm) AS o, STRUCT(9 AS fga, 2 AS fgm) AS d) AS t2 UNION ALL 
  SELECT 'nick' name, 'jets' team,
    STRUCT(STRUCT(12 AS fga, 7 AS fgm) AS o, STRUCT(13 AS fga, 7 AS fgm) AS d) AS t1,
    STRUCT(STRUCT(15 AS fga, 7 AS fgm) AS o, STRUCT(22 AS fga, 7 AS fgm) AS d) AS t2
)
SELECT * REPLACE (
  (SELECT AS STRUCT t1.* REPLACE (
    (SELECT AS STRUCT 
      t1.o.*,
      SAFE_DIVIDE(wide_stats.t1.o.fgm, wide_stats.t1.o.fga) AS fg_pct,
      rank() over(order by wide_stats.t1.o.fgm asc) as fgm_rank
    ) 
    AS o))
  AS t1)
FROM wide_stats

由于以SELECT without FROM clause cannot use analytic functions at [13:6]开头的行，因此返回以下错误rank() over ...

我喜欢在BQ中使用STRUCT，但是无法在以后的查询中使用解析函数来更新结构，这是目前对我要完成的工作的重大限制。任何帮助，我们将不胜感激！

谢谢！

Answer 1

以下应该可以解决问题（BigQuery Standard SQL）

#standardSQL
WITH wide_stats AS (
  SELECT 'joe' name, 'bills' team,
    STRUCT(STRUCT(7 AS fga, 5 AS fgm) AS o, STRUCT(8 AS fga, 3 AS fgm) AS d) AS t1,
    STRUCT(STRUCT(3 AS fga, 4 AS fgm) AS o, STRUCT(9 AS fga, 2 AS fgm) AS d) AS t2 UNION ALL 
  SELECT 'nick' name, 'jets' team,
    STRUCT(STRUCT(12 AS fga, 7 AS fgm) AS o, STRUCT(13 AS fga, 7 AS fgm) AS d) AS t1,
    STRUCT(STRUCT(15 AS fga, 7 AS fgm) AS o, STRUCT(22 AS fga, 7 AS fgm) AS d) AS t2
), temp AS (
  SELECT *, RANK() OVER(ORDER BY wide_stats.t1.o.fgm ASC) AS fgm_rank 
  FROM wide_stats
)
SELECT * EXCEPT(fgm_rank) REPLACE (
  (SELECT AS STRUCT t1.* REPLACE (
    (SELECT AS STRUCT 
      t1.o.*,
      SAFE_DIVIDE(t1.o.fgm, t1.o.fga) AS fg_pct,
      fgm_rank
    ) 
    AS o))
  AS t1)    
FROM temp

BigQuery无法在SELECT * REPLACE中使用解析功能

1 个答案: