SQL 查询以查找给定日期之前的最后一次更改

Question

我有 2 个表，一个包含产品更改信息，另一个包含测试信息，我正在尝试构建一个查询以查找测试前的最后一个产品更改信息。

例如，在下面的数据中，我想要返回的每个测试都类似于：

product_sn	testtype	testdate	part_spec	changedate
ABC0001	hot	2021-01-01 13:00:00	nylon	2021-01-01 00:00:00
ABC0001	hot	2021-01-01 16:00:00	steel	2021-01-01 15:00:00
ABC1234	hot	2021-01-01 18:00:00	steel	2021-01-01 17:00:00
ABC1234	cold	2021-01-01 18:10:00	steel	2021-01-01 17:00:00
ABC1234	mid	2021-01-01 18:20:00	steel	2021-01-01 17:00:00
ABC0001	hot	2021-01-02 09:00:00	brass	2021-01-01 20:00:00
ABC1234	cold	2021-01-01 19:00:00	steel	2021-01-01 17:00:00

我在 sql 方面有点菜鸟，不太清楚在 stackoverflow 的其他地方搜索的最佳术语 :)

CREATE OR REPLACE TABLE Scratchpad.product
    (`ID` INT64,
    `product_sn` STRING,
    `part_spec` STRING,
    `changedate` datetime
    )
;
    
INSERT INTO Scratchpad.product
    (`ID`, `product_sn`, `part_spec`, `changedate`)
VALUES
    (1, 'ABC0001', 'nylon','2021-01-01 00:00:00'),
    (2, 'ABC0001', 'steel','2021-01-01 15:00:00'),
    (3, 'ABC0001', 'brass','2021-01-01 20:00:00'),
    (4, 'ABC1234', 'steel','2021-01-01 17:00:00'),
    (5, 'ABC1234', 'nylon','2021-02-01 00:00:00')
;


CREATE OR REPLACE TABLE Scratchpad.test
    (`ID` INT64,
     `product_sn` STRING,
      `testtype` STRING,
      `testdate` datetime
    )
;
    
INSERT INTO Scratchpad.test
    (`ID`, `product_sn`, `testtype`, `testdate`)
VALUES
    (1, 'ABC0001', 'hot','2021-01-01 13:00:00'),
    (2, 'ABC0001', 'hot','2021-01-01 16:00:00'),
    (3, 'ABC1234', 'hot','2021-01-01 18:00:00'),
    (4, 'ABC1234', 'cold','2021-01-01 18:10:00'),
    (5, 'ABC1234', 'mid','2021-01-01 18:20:00'),
    (6, 'ABC0001', 'hot','2021-01-02 09:00:00'),
    (7, 'ABC1234', 'cold','2021-01-01 19:00:00')
;

Answer 1

感谢https://stackoverflow.com/a/45041970/274349这里的另一个答案@

#standardSQL
SELECT 
  t.ID AS ID, 
  t.product_sn,
  t.testdate as date, 
  t.testtype,

  ARRAY_AGG(r.part_spec ORDER BY r.changedate desc LIMIT 1)[SAFE_OFFSET(0)] AS part_spec,
  ARRAY_AGG(r.changedate ORDER BY r.changedate desc LIMIT 1)[SAFE_OFFSET(0)] AS part_change

FROM `Scratchpad.test` AS t
JOIN `Scratchpad.product` AS r
ON t.product_sn = r.product_sn 
AND r.changedate <= t.testdate
GROUP BY ID,product_sn, date, testtype

Answer 2

这在 BQ 中可能是一个棘手的问题。一种解决方案是：

select t.*,
       (select as struct p.part_spec, p.change_date
        from `Scratchpad.product` p
        where p.product_sn = t.product_sn and
              p.changedate < t.testdate
        order by p.changedate desc
        limit 1
       )
from `Scratchpad.test` t;

这可能有效，但 BigQuery 有时对左连接和子查询中的不等式很挑剔。

更详细的方法是：

select array_agg(t order by p.change_date desc limit 1)[ordinal(1)].*,
       array_agg(p order by p.change_date desc limit 1)[ordinal(1)].*
from `Scratchpad.test` t left join
     `Scratchpad.product` p
     on p.product_sn = t.product_sn and
        p.changedate < t.testdate
group by t.id;