我坚持这个，我不知道该怎么做。
我有这张桌子
create table events(event_datetime datetime,event_code varchar(40),val varchar(40));
insert into events (event_datetime, event_code, val)
values ('2018-01-01 06:00:00', '50', '1'),
('2018-01-01 07:00:00', '54', null),
('2018-01-01 11:00:00', '50', '2'),
('2018-01-01 13:00:00', '54', null),
('2018-01-02 07:00:00', '50', '1'),
('2018-01-02 23:00:00', '54', null),
('2018-01-24 07:00:00', '50', '1'),
('2018-01-25 23:00:00', '54', null),
('2018-02-02 10:00:00', '50', '1'),
('2018-02-02 12:00:00', '54',null)
Event_code 50表示事件开始。 Event_code 54表示事件停止。
我想总计一天54到50之间的小时数（有时一天有50-54多个） 比我想做AVG的1天，四次。
例如，上面第1周的数据：
2018-01-01 07:00:00 - 2018-01-01 06:00:00 = 1hour
2018-01-01 13:00:00 - 2018-01-01 11:00:00 = 2hours
2018-01-02 23:00:00 - 2018-01-02 07:00:00 = 16hours
Day 1 = 3hours
Day 2 = 16hours
Day 3-7 = 0hours
Week 1 = AVG(days) = 19/7
输出类似于：
| Month | Week 1 | Week 2 | Week 3 | Week 4 |
|---------|------------|------------|-------------|------------|
| Jan |avghours/day|avghours/day|avghours/day |avghours/day|
| Feb |avghours/day|avghours/day|avghours/day |avghours/day|
编辑：
事件50总是直接跟随54
事件永远不会超过一天。
编辑2：
我创建了一个存储过程：
DELIMITER #
CREATE PROCEDURE tablediff()
BEGIN
DECLARE done int default false;
DECLARE v_day DATETIME;
DECLARE v_code VARCHAR(40);
DECLARE v_day_1 DATETIME;
DECLARE v_code_1 VARCHAR(40);
DECLARE start_time DATETIME;
DECLARE stop_time DATETIME;
DECLARE code VARCHAR(40);
DECLARE duration TIME;
DECLARE cur1 CURSOR FOR
SELECT event_datetime, event_code as code, event_param_1
FROM events
AND event_code =50
ORDER BY event_datetime ASC;
DECLARE cur2 CURSOR FOR
SELECT event_datetime, event_code as code, event_param_1
FROM events
AND event_code = 50
ORDER BY event_datetime ASC;
DECLARE CONTINUE HANDLER FOR NOT FOUND SET done = 1;
OPEN cur1;
OPEN cur2;
CREATE TEMPORARY TABLE tmp
(
start_time DATETIME;
stop_time DATETIME;
code VARCHAR(40);
duration TIME;
);
FETCH cur2 INTO v_day_1, v_code_1;
forLoop: LOOP
FETCH cur1 INTO v_day, v_code;
FETCH cur2 INTO v_day_1, v_code_1;
CASE
WHEN v_code = 50 THEN
IF v_code_1 = 54 THEN
SET start_time = v_day;
SET stop_time = v_day_1;
SET code = v_code;
SET duration = duration + TIMEDIFF(stop_time, start_time);
ELSE
SET start_time = null;
SET stop_time = null;
SET code = null;
SET duration = duration;
END IF;
END CASE;
INSERT INTO tmp VALUES (start_time, stop_time, code, duration);
IF done
THEN
LEAVE forLoop;
END IF;
END LOOP;
CLOSE cur1;
CLOSE cur2;
END#
即使没有54也可以显示开始和停止时间，但我需要进行时间计算，持续时间不起作用，每行显示为空。
答案 0 :(得分：2)
代码之后的ID字段（作为键）和事件标识字段是理想的。
如果无法做到这一点，函数会帮助你：
DELIMITER $$
CREATE FUNCTION event_stoptime (start_time datetime)
RETURNS datetime
BEGIN
DECLARE stop_time datetime;
SET stop_time = (SELECT MIN(event_datetime) FROM events WHERE
event_datetime > start_time);
RETURN stop_time;
END $$
DELIMITER ;
构建函数后，您可以将其与MySQL日期函数结合使用以获取所需的数据：
SELECT event_datetime as event_start,
event_stoptime(event_datetime) as event_stop,
TIMEDIFF(event_stoptime(event_datetime), event_datetime) as event_duration,
WEEK(event_datetime) as event_week,
MONTH(event_datetime) as event_month
from events
where event_code = 50;
我应该得到的结果是
| event_start | event_stop | duration | week| month|
'2018-01-01 06:00:00', '2018-01-01 07:00:00', '01:00:00', '0', '1'
'2018-01-01 11:00:00', '2018-01-01 13:00:00', '02:00:00', '0', '1'
'2018-01-02 07:00:00', '2018-01-02 23:00:00', '16:00:00', '0', '1'
'2018-01-24 07:00:00', '2018-01-25 23:00:00', '40:00:00', '3', '1'
'2018-02-02 10:00:00', '2018-02-02 12:00:00', '02:00:00', '4', '2'
GREAT BIG CAVEAT：如果给定的50没有相应的54事件（由于错误等），时间将严重偏离。
这是一个更好的解决方案使用ID字段。为了便于测试，我在插入语句中包含了ID：
CREATE TABLE `events` (
`ID` int(11) NOT NULL AUTO_INCREMENT,
`event_datetime` datetime DEFAULT NULL,
`event_code` varchar(40) DEFAULT NULL,
`val` varchar(40) DEFAULT NULL,
`event_id` int(11) DEFAULT NULL,
PRIMARY KEY (`ID`)
);
INSERT INTO `events`
VALUES (1,'2018-01-01 06:00:00','50','1',NULL),
(2,'2018-01-01 07:00:00','54',NULL,1),
(3,'2018-01-01 11:00:00','50','2',NULL),
(4,'2018-01-01 13:00:00','54',NULL,3),
(5,'2018-01-01 09:00:00','50',NULL,NULL),
(6,'2018-01-02 23:00:00','54',NULL,5),
(7,'2018-01-24 07:00:00','50','1',NULL),
(8,'2018-01-25 23:00:00','54','3',7),
(9,'2018-02-02 10:00:00','50','1',NULL),
(10,'2018-02-02 12:00:00','54',NULL,9),
(11,'2018-01-30 23:00:00','54','3',NULL);
视图将表连接到自身以选择正确的＆＃34; 54＆＃34;至＆＃34; 50＆＃34;。请注意创建日期以查找该月的第一周：
CREATE VIEW `event_list` AS
select `events`.`ID` AS `ID`,`events`.`event_datetime` AS `event_start`,
`end_events`.`event_datetime` AS `event_stop`,
timediff(`end_events`.`event_datetime`,`events`.`event_datetime`) AS `event_duration`,
week(`events`.`event_datetime`) AS `event_week`,
month(`events`.`event_datetime`) AS `month_number`,
monthname(`events`.`event_datetime`) AS `event_month`,
WEEK(STR_TO_DATE(concat(year(events.event_datetime), '-', month(events.event_datetime), '-', '01'), '%Y-%m-%d')) as first_week
from (`events` left join `events` `end_events` on((`events`.`ID` = `end_events`.`event_id`)))
where (`events`.`event_code` = '50');
select语句使用IF为周摘要选择正确的数据。将计算持续时间的时间分解为秒，然后重新组装以供显示;
select event_month,
sec_to_time(sum(time_to_sec(if(event_week = first_week, event_duration, 0)))) as week_one,
sec_to_time(sum(time_to_sec(if(event_week = first_week + 1, event_duration, 0)))) as week_two,
sec_to_time(sum(time_to_sec(if(event_week = first_week + 2, event_duration, 0)))) as week_three,
sec_to_time(sum(time_to_sec(if(event_week = first_week + 3, event_duration, 0)))) as week_four,
sec_to_time(sum(time_to_sec(if(event_week = first_week + 4, event_duration, 0)))) as week_five
from event_list
group by event_month
order by month_number
不太优雅，但有效。