对连续记录进行分组

Question

这是我的记录（输入）。 field2 总是从 100 开始，之后它可以取任何大于 100 的值。

field1   field2
===============
val1     100
val2     110
------------
val3     100
val4     110
val3     130
val3     140
------------
val1     100

我需要对以 100 开头的连续记录进行分组，之后它可以是任何非 100 的数字。对于上面的示例，我需要的输出是

field 1                    field2
===================================================
(val1, val2)                (100, 110)
(val3, val4, val3, val3)    (100, 110, 130, 140)
(val1)                      (100)

我如何实现这一目标

Answer 1

让我假设您有一个指定排序的列。然后您可以通过计算每条记录之前的“100”条记录的数量来识别组，然后使用array_agg()：

select array_agg(field1 order by id) as field1s,
       array_agg(field2 order by id) as field2s
from (select t.*,
             sum(case when field2 = 100 then 1 else 0 end) over (order by id) as grp
      from t
     ) t
group by grp;

注意：MySQL 中的解决方案看起来非常非常不同。但是，它仍然以select 开头。

Answer 2

您必须有一些可用于定义订单的字段。
在下面的示例中，我假设它是id 字段

下面应该做你所期望的

#standardSQL
SELECT 
  CONCAT('(',STRING_AGG(field1 ORDER BY id), ')') AS field1,
  CONCAT('(',STRING_AGG(CAST(field2 AS STRING) ORDER BY id), ')') AS field2
FROM (
  SELECT 
    id, field1, field2,
    COUNTIF(field2 = 100) OVER (ORDER BY id) AS grp
  FROM yourTable
) t
GROUP BY grp
ORDER BY MIN(id)   

您可以使用以下来自您问题的虚拟数据进行测试/尝试

#standardSQL
WITH yourTable AS (
  SELECT 1 AS id, 'val1' AS field1, 100 AS field2 UNION ALL
  SELECT 2 AS id, 'val2' AS field1, 110 AS field2 UNION ALL
  SELECT 3 AS id, 'val3' AS field1, 100 AS field2 UNION ALL
  SELECT 4 AS id, 'val4' AS field1, 110 AS field2 UNION ALL
  SELECT 5 AS id, 'val3' AS field1, 130 AS field2 UNION ALL
  SELECT 6 AS id, 'val3' AS field1, 140 AS field2 UNION ALL
  SELECT 7 AS id, 'val1' AS field1, 100 AS field2 
)
SELECT 
  CONCAT('(',STRING_AGG(field1 ORDER BY id), ')') AS field1,
  CONCAT('(',STRING_AGG(CAST(field2 AS STRING) ORDER BY id), ')') AS field2
FROM (
  SELECT 
    id, field1, field2,
    COUNTIF(field2 = 100) OVER (ORDER BY id) AS grp
  FROM yourTable
) 
GROUP BY grp
ORDER BY MIN(id) 

输出是

field1                  field2   
------                  ------   
(val1,val2)             (100,110)    
(val3,val4,val3,val3)   (100,110,130,140)    
(val1)                  (100)    

想知道这是否可能（不使用订单列）？

如果您的表中确实只有这两个字段 - 您很可能不走运，需要重新访问使用用于排序的附加字段填充此表的逻辑（作为时间线）

作为最后的手段 - 您可以尝试下面的示例，其中此类列是即时生成的 - 但请理解，绝对不能保证获得您期望的订单

#standardSQL
WITH yourTable AS (
  SELECT 'val1' AS field1, 100 AS field2 UNION ALL
  SELECT 'val2' AS field1, 110 AS field2 UNION ALL
  SELECT 'val3' AS field1, 100 AS field2 UNION ALL
  SELECT 'val4' AS field1, 110 AS field2 UNION ALL
  SELECT 'val3' AS field1, 130 AS field2 UNION ALL
  SELECT 'val3' AS field1, 140 AS field2 UNION ALL
  SELECT 'val1' AS field1, 100 AS field2 
),
tempTable AS (
  SELECT field1, field2, ROW_NUMBER() OVER() AS id  
  FROM yourTable
)
SELECT 
  CONCAT('(',STRING_AGG(field1 ORDER BY id), ')') AS field1,
  CONCAT('(',STRING_AGG(CAST(field2 AS STRING) ORDER BY id), ')') AS field2
FROM (
  SELECT 
    id, field1, field2,
    COUNTIF(field2 = 100) OVER (ORDER BY id) AS grp
  FROM tempTable
) 
GROUP BY grp
ORDER BY MIN(id) 

输出是相同的 - 但同样 - 不能保证！

field1                  field2   
------                  ------   
(val1,val2)             (100,110)    
(val3,val4,val3,val3)   (100,110,130,140)    
(val1)                  (100)