在sql中按日期小时分组

Question

晕，我想添加 id 然后 group by 产品和日期时间（不将日期时间更改为 varchar）。数据是：

datetime               product    sold
2020-10-27 08:51:00       A        50
2020-10-27 08:52:00       A        10
2020-10-27 08:52:00       B        20
2020-10-27 09:52:00       B        30

我的查询是：

select newId() as id, datehour(datetime) datehour, product, sum(sold)
from table_a
group by datehour(datetime), product

但没用

我的预期结果：

注意：日期时间应该是 datetime 类型而不是 varchar

id   datehour               product    sold
1    2020-10-27 08:00:00       A        60
2    2020-10-27 08:00:00       B        20
3    2020-10-27 09:00:00       B        30

注意：日期时间应该是 datetime 类型而不是 varchar

有什么想法吗？

Answer 1

尚不清楚您要使用newId() 实现什么目标。除此之外，您可以使用DATE_FORMAT 和STR_TO_DATE 从时间中去除分钟和秒以获取datehour 值，从而获得所需的结果；然后你可以 GROUP BY 和 product 和 SUM sold 值：

SELECT STR_TO_DATE(DATE_FORMAT(`datetime`, '%Y-%m-%d %H'), '%Y-%m-%d %H') AS datehour,
       product,
       SUM(sold) AS sold
FROM table_a
GROUP BY datehour, product

输出：

datehour                sold    product
2020-10-27 08:00:00     60      A
2020-10-27 08:00:00     20      B
2020-10-27 09:00:00     30      B

Demo on db-fiddle

在生成id 值方面，如果您使用的是 MySQL 8+，则可以使用ROW_NUMBER()：

SELECT ROW_NUMBER() OVER (ORDER BY datehour, product) AS id,
       datehour,
       product,
       sold
FROM (
  SELECT STR_TO_DATE(DATE_FORMAT(`datetime`, '%Y-%m-%d %H'), '%Y-%m-%d %H') AS datehour,
         product,
         SUM(sold) AS sold
  FROM table_a
  GROUP BY datehour, product
) a
ORDER BY id

对于 MySQL 5.x，您可以使用变量模拟 ROW_NUMBER()：

SELECT @rn := @rn + 1 AS id,
       datehour,
       product,
       sold
FROM (
  SELECT STR_TO_DATE(DATE_FORMAT(`datetime`, '%Y-%m-%d %H'), '%Y-%m-%d %H') AS datehour,
         product,
         SUM(sold) AS sold
  FROM table_a
  GROUP BY datehour, product
  ORDER BY datehour, product
) a
CROSS JOIN (SELECT @rn := 0) i

在这两种情况下，输出都是：

id  datehour                product     sold
1   2020-10-27 08:00:00     A           60
2   2020-10-27 08:00:00     B           20
3   2020-10-27 09:00:00     B           30

Demo on db-fiddle