在 BigQuery 中对日期进行分组时出现 DATE_ADD 或 DATE_DIFF 错误

DATE_ADD or DATE_DIFF error when grouping dates in BigQuery

尽我所能尝试搜索,我仍然无法弄清楚,尽管 https://cloud.google.com/bigquery/docs/reference/standard-sql/date_functions#date_add 提供了一些帮助,但我仍然卡住了。我正在尝试将日期分组为数周,但不断收到代码下方的两个错误之一。

day         bitcoin_total   dash_total
2009-01-03  1               0
2009-01-09  14              0
2009-01-10  61              0

理想的结果是一周的开始日期(可以是星期一或星期日,以哪个为准)

day         bitcoin_total   dash_total
2008-12-28  1               0
2009-01-04  75              0

这似乎是一个常见问题,但大多数答案都是针对 T-SQL 而不是标准 SQL。我的日期列是 Date 类型,但那是 return 类型,所以这应该不是问题。

DATE_ADD(week, DATE_DIFF(week, 0, day), 0) Date
FROM
my_table
GROUP BY
DATE_ADD(week, DATE_DIFF(week, 0, day), 0) 
ORDER BY
DATE_ADD(week, DATE_DIFF(week, 0, day), 0)

如果我将 DATE_ADD 函数中的 date_expression 更改为 DATE "2009-01-01",则上述代码或 Error: Expected INTERVAL expression at [2:29] 出现 Unrecognized name: week at [2:10] 错误]

以下适用于 BigQuery 标准 SQL

#standardSQL
SELECT DATE_TRUNC(day, WEEK) AS day, 
  SUM(bitcoin_total) AS bitcoin_total, 
  SUM(dash_total) AS dash_total
FROM `project.dataset.table`
GROUP BY day   

如果要应用于您问题中的样本数据,如下例所示

#standardSQL
WITH `project.dataset.table` AS (
  SELECT DATE '2009-01-03' day, 1 bitcoin_total, 0 dash_total UNION ALL
  SELECT '2009-01-09', 14, 0 UNION ALL
  SELECT '2009-01-10', 61, 0 
)
SELECT DATE_TRUNC(day, WEEK) AS day, 
  SUM(bitcoin_total) AS bitcoin_total, 
  SUM(dash_total) AS dash_total
FROM `project.dataset.table`
GROUP BY day   

输出将是

Row day         bitcoin_total   dash_total   
1   2008-12-28  1               0    
2   2009-01-04  75              0