Question

我有桌子：

+--------+------+------------+
| id     | type | timestamp  |
+--------+------+------------+
| 927061 |  573 | 1520284933 |
| 927062 |  573 | 1520284933 |
| 927063 |  573 | 1520284933 |
| 927064 |  573 | 1520284933 |
| 927347 |  573 | 1520285483 |
| 928796 |  573 | 1520287665 |
| 928799 |  573 | 1520287672 |
| 928801 |  573 | 1520287676 |
| 928802 |  573 | 1520287680 |
| 928806 |  573 | 1520287684 |
| 928821 |  573 | 1520287735 |
| 928822 |  573 | 1520287735 |
| 928823 |  573 | 1520287735 |
| 928824 |  573 | 1520287735 |
| 928825 |  573 | 1520287735 |
| 928826 |  573 | 1520287735 |
| 928827 |  573 | 1520287736 |
| 928828 |  573 | 1520287736 |
| 928829 |  573 | 1520287736 |
| 928830 |  573 | 1520287736 |
+--------+------+------------+

该表现在有4,134,798行并且计数...并且记录有318种类型并且正在计数。我想根据时间戳创建每种类型的已用时间报告。每条记录都意味着某种类型的活动。

类似于的报告：

type - elapsed time
573 - 3 days 4 hours 3 minutes
103 - 1 days 1 hours 1 minutes

我必须遍历所有记录，添加某种类型事件之间的经过时间，存储某类事件的总耗用时间，并按类型显示已用时间。

有人愿意给我一些见解吗？

这是我的疑问：

SET @total = 0;
SET @curr_timestamp = 0;
SET @curr_id = 0;
SET @lineitem = 0;
SELECT CONCAT(FLOOR(HOUR(SEC_TO_TIME(Dedication)) / 24), ' días', ' ',TIME_FORMAT(SEC_TO_TIME(Dedication % (24 * 60 * 60)), '%H'), ' horas',' ',TIME_FORMAT(SEC_TO_TIME(Dedication % (24 * 60 * 60)), '%i'), ' minutos') As Dedicacion FROM
(
    SELECT type,
        (@lineitem:=@lineitem+1) line,
        (@prev_id:=@curr_id),
        (@curr_id:=id),
        (@prev_timestamp:=@curr_timestamp),
        (@curr_timestamp:=timestamp),
        (@newvalue:=IF(@prev_id>@curr_id,@prev_timestamp+@curr_timestamp,ABS(@prev_timestamp-@curr_timestamp))) newval,
        (@total:=@total+@newvalue) Dedication
    FROM mdl_logstore_standard_log WHERE (DATEDIFF(NOW(),FROM_UNIXTIME(timestamp) ) < 7) AND type=573
) A WHERE line = @lineitem;

查询基于：

https://dba.stackexchange.com/questions/87159/how-to-calculate-a-total-by-comparing-current-and-next-row

谢谢，豪尔赫·达维拉。

Answer 1

以下是如何完成所需内容的示例：

SELECT
    type,
    TIMESTAMPDIFF(YEAR, MIN(timestamp), MAX(timestamp)) AS in_years,
    TIMESTAMPDIFF(MONTH, MIN(timestamp), MAX(timestamp)) AS in_month,
    TIMESTAMPDIFF(DAY, MIN(timestamp), MAX(timestamp)) AS in_days,
    TIMESTAMPDIFF(MINUTE, MIN(timestamp), MAX(timestamp)) AS in_minutes,
    TIMESTAMPDIFF(SECOND, MIN(timestamp), MAX(timestamp)) AS in_seconds
FROM
    activities
GROUP BY type

Answer 2

您的时间戳似乎是Unix所代表的秒数。您可以在几秒钟内获得差异：

select type, max(timestamp) - min(timestamp) as diff_in_seconds
from activities 
group by type;

将此转换为时间取决于您使用的数据库的功能。将其转换为十进制天或小时非常容易。例如：

select type, (max(timestamp) - min(timestamp)) / (60 * 60) as diff_hours
from activities 
group by type;

在时间段内经过的时间量

2 个答案: