SQL Oracle - 按 ID、任务 ID、最小和最大时间戳分组

Posted 2023-04-12

【中文标题】SQL Oracle - 按 ID、任务 ID、最小和最大时间戳分组

【英文标题】：SQL Oracle - Group by ID, task ID, min and max timestamp

【发布时间】：2021-06-16 11:28:20

【问题描述】：

我有用户执行不同任务的数据。 我想根据用户 ID 和任务 ID 对这些数据进行分组，以获取每个任务的开始和结束时间。当员工切换到另一个任务时，应该有一个新的行，其中包含新的开始和结束时间。

简化数据集示例：

userid	taskid	date_time_stamp (ascending)
1	task-A	16/6/2021 04:17:00
1	task-A	16/6/2021 04:19:00
1	task-A	16/6/2021 04:27:00
1	task-B	16/6/2021 04:31:00
1	task-B	16/6/2021 04:33:00
1	task-B	16/6/2021 04:36:00
1	task-A	16/6/2021 04:42:00
1	task-A	16/6/2021 04:44:00

示例结果

userid	taskid	first_dtm	last_dtm
1	task-A	16/6/2021 04:17:00	16/6/2021 04:27:00
1	task-B	16/6/2021 04:31:00	16/6/2021 04:36:00
1	task-A	16/6/2021 04:42:00	16/6/2021 04:44:00

我知道我应该使用一些结合 GROUP BY 的 min() 和 max() 函数。但是，在此示例中，按 userid 和 taskid 分组将导致 task-A 仅占一行。

【问题讨论】：

这是常见的“差距和岛屿”任务。 PS。 OracleDB 或 mysql，而不是两者都选 - 选择一个并编辑标签列表。

【参考方案1】：

从 Oracle 12 开始，您可以使用MATCH_RECOGNIZE：

SELECT *
FROM   table_name
MATCH_RECOGNIZE(
  PARTITION BY userid
  ORDER BY date_time_stamp
  MEASURES
    FIRST(taskid) AS taskid,
    FIRST(date_time_stamp) AS start_date,
    LAST(date_time_stamp) AS end_date
  ONE ROW PER MATCH
  PATTERN ( same_task+ )
  DEFINE same_task AS FIRST(taskid) = taskid
)

在此之前，可以使用ROW_NUMBER解析函数并聚合：

SELECT userid,
       taskid,
       MIN(date_time_stamp) AS start_date,
       MAX(date_time_stamp) AS end_date
FROM   (
  SELECT t.*,
         ROW_NUMBER() OVER ( PARTITION BY userid ORDER BY date_time_stamp )
           - ROW_NUMBER() OVER ( PARTITION BY userid, taskid ORDER BY date_time_stamp )
           AS grp
  FROM   table_name t
)
GROUP BY userid, taskid, grp
ORDER BY userid, start_date

其中，对于您的示例数据：

CREATE TABLE table_name ( userid, taskid, date_time_stamp ) AS
SELECT 1, 'task-A', DATE '2021-06-16' + INTERVAL '04:17:00' HOUR TO SECOND FROM DUAL UNION ALL
SELECT 1, 'task-A', DATE '2021-06-16' + INTERVAL '04:19:00' HOUR TO SECOND FROM DUAL UNION ALL
SELECT 1, 'task-A', DATE '2021-06-16' + INTERVAL '04:27:00' HOUR TO SECOND FROM DUAL UNION ALL
SELECT 1, 'task-B', DATE '2021-06-16' + INTERVAL '04:31:00' HOUR TO SECOND FROM DUAL UNION ALL
SELECT 1, 'task-B', DATE '2021-06-16' + INTERVAL '04:33:00' HOUR TO SECOND FROM DUAL UNION ALL
SELECT 1, 'task-B', DATE '2021-06-16' + INTERVAL '04:36:00' HOUR TO SECOND FROM DUAL UNION ALL
SELECT 1, 'task-A', DATE '2021-06-16' + INTERVAL '04:42:00' HOUR TO SECOND FROM DUAL UNION ALL
SELECT 1, 'task-A', DATE '2021-06-16' + INTERVAL '04:44:00' HOUR TO SECOND FROM DUAL

两个输出：

USERID	TASKID	START_DATE	END_DATE
1	task-A	2021-06-16 04:17:00	2021-06-16 04:27:00
1	task-B	2021-06-16 04:31:00	2021-06-16 04:36:00
1	task-A	2021-06-16 04:42:00	2021-06-16 04:44:00

db小提琴here