BigQuery 可以在响应 ROLLUP 查询时删除小计行吗

Question

当我使用 ROLLUP 通过潜在的大量分组条件字段查询 BQ 时（在本例中 campaign_group_id）

例如：

SELECT
  campaign_group_id AS campaign_group_id,
  DATE(DATE_ADD(TIME, 3, 'HOUR')) AS DAY,
  SUM(impressions) AS imps
FROM
  [browser_traffic.2016_05_28],
  [browser_traffic.2016_05_29]
WHERE
  ( DATE_ADD( TIME, 3, "HOUR") >= '2016-05-28 00:00:00'
    AND DATE_ADD( TIME, 3, "HOUR") < '2016-05-30 00:00:00' )
GROUP EACH BY ROLLUP (campaign_group_id, DAY)
ORDER BY DAY ASC, campaign_group_id ASC
LIMIT 500

BQ returns 很多行的小计不适用于我的用例：

+-------------------+------+-----------+
| campaign_group_id | day  |   imps    |
+-------------------+------+-----------+
|              NULL | NULL | 158423933 |
|                61 | NULL |         0 |
|               496 | NULL |     79870 |
|               497 | NULL |     10492 |
|               809 | NULL |         0 |
|               936 | NULL |      2451 |
|               937 | NULL |         0 |
|               940 | NULL |      6844 |
|               942 | NULL |    207685 |
|               946 | NULL |         0 |
|               961 | NULL |         0 |
|               975 | NULL |     16167 |
|               976 | NULL |     15767 |
|              1018 | NULL |         0 |
|              1020 | NULL |         0 |
|              1022 | NULL |    766875 |
|              1039 | NULL |    355765 |
...

我需要以某种方式减少结果中的小计行，但保留完整的总计行（这是上面结果中的第一行）

BQ returns 是否可能只有所选字段的总计行？

Answer 1

如果您对总计感兴趣 - 您很可能不需要 ROLLUP

相反，您可以考虑如下常规 GROUP BY

SELECT
  DATE(DATE_ADD(TIME, 3, 'HOUR')) AS DAY,
  SUM(impressions) AS imps
FROM
  [browser_traffic.2016_05_28],
  [browser_traffic.2016_05_29]
WHERE
  ( DATE_ADD( TIME, 3, "HOUR") >= '2016-05-28 00:00:00'
    AND DATE_ADD( TIME, 3, "HOUR") < '2016-05-30 00:00:00' )
GROUP 1
ORDER BY DAY ASC
LIMIT 500

Answer 2

您可以使用另一个 SELECT 语句过滤查询结果：

SELECT campaign_group_id, day, imps 
FROM (
    ... your rollup query with LIMIT removed ...
)
WHERE (day IS NOT NULL) OR (campaign_group_id IS NULL)
LIMIT 500

Answer 3

请注意 GROUPING() 函数也存在，它将帮助您：

SELECT year, name, SUM(number) s,
       GROUPING(year) is_grouping_year, 
       GROUPING(name) is_grouping_name
FROM [bigquery-public-data:usa_names.usa_1910_2013] 
WHERE name IN ('John', 'Jovana')
AND year BETWEEN 2012 AND 2013
GROUP BY ROLLUP(name, year)
ORDER BY year, name

year    name    s       is_grouping_year    is_grouping_name     
null    null    21182   1                   1    
null    John    21164   1                   0    
null    Jovana  18      1                   0    
2012    John    10576   0                   0    
2012    Jovana  18      0                   0    
2013    John    10588   0                   0

基本上，您要求的是 is_grouping_year 和 is_grouping_name 均为 0 或 1 的行。

来自文档：

When using the ROLLUP function, you can use the GROUPING function to distinguish between rows that were added because of the ROLLUP function and rows that actually have a NULL value for the group key.

https://cloud.google.com/bigquery/query-reference#groupby

BigQuery 可以在响应 ROLLUP 查询时删除小计行吗

Can BigQuery drop subtotals rows WHEN responding to ROLLUP query

rollup

google-bigquery