Question

我有一个查询，它给我输出

select distinct(d.data_name) ,h.workstation_name, h.type_id, convert(varchar(19),(h.insert_date),120) as start
from data d, process_data_history h 
where h.data_id = d.data_id
and h.type_id between '500' and '5001'
and h.workstation_name like 'PL1W7D-105120'
and CONVERT(VARCHAR(25), h.insert_date, 126) LIKE '2016-04%'
order by d.data_name, start ;

输出看起来像这样

data_name                     workstation_name  type_id        start
WE228JP_2016_04_05__10_03_03    PL1W7D-105120   500     2016-04-05 12:45:38
WE228JP_2016_04_05__10_03_03    PL1W7D-105120   501     2016-04-05 12:45:38
WE228JP_2016_04_05__10_03_03    PL1W7D-105120   600     2016-04-05 12:45:39
WE228JP_2016_04_05__10_03_03    PL1W7D-105120   601     2016-04-05 12:45:39
WE228JP_2016_04_05__10_03_03    PL1W7D-105120   4001    2016-04-05 14:06:04
WE228JP_2016_04_05__10_03_03    PL1W7D-105120   4000    2016-04-05 14:05:55
WE228JP_2016_04_05__10_03_03    PL1W7D-105120   4100    2016-04-05 14:06:05
WE228JP_2016_04_05__10_03_03    PL1W7D-105120   4101    2016-04-05 14:10:43
WE228JP_2016_04_05__10_03_03    PL1W7D-105120   5000    2016-04-05 14:10:43
WE228JP_2016_04_05__10_03_03    PL1W7D-105120   5001    2016-04-05 14:10:45
WE228JP_2016_04_06__10_33_03    PL1W7D-105120   500     2016-04-06 14:10:46
WE228JP_2016_04_06__10_33_03    PL1W7D-105120   600     2016-04-06 14:10:47
WE228JP_2016_04_06__10_33_03    PL1W7D-105120   501     2016-04-06 14:10:47
WE228JP_2016_04_06__10_33_03    PL1W7D-105120   601     2016-04-06 14:10:48
WE228JP_2016_04_06__10_33_03    PL1W7D-105120   4000    2016-04-06 15:30:07
WE228JP_2016_04_06__10_33_03    PL1W7D-105120   4001    2016-04-06 15:30:18
WE228JP_2016_04_06__10_33_03    PL1W7D-105120   4100    2016-04-06 15:30:18
WE228JP_2016_04_06__10_33_03    PL1W7D-105120   4101    2016-04-06 15:35:01
WE228JP_2016_04_06__10_33_03    PL1W7D-105120   5000    2016-04-06 15:35:01
WE228JP_2016_04_06__10_33_03    PL1W7D-105120   5001    2016-04-06 15:35:03
WE228JP_2016_04_15__10_03_03    PL1W7D-105122   500     2016-04-05 12:45:38
WE228JP_2016_04_15__10_03_03    PL1W7D-105122   501     2016-04-05 12:45:38
WE228JP_2016_04_15__10_03_03    PL1W7D-105122   600     2016-04-05 12:45:39
WE228JP_2016_04_15__10_03_03    PL1W7D-105120   601     2016-04-05 12:45:39
WE228JP_2016_04_15__10_03_03    PL1W7D-105122   4001    2016-04-05 14:06:04
WE228JP_2016_04_15__10_03_03    PL1W7D-105122   5000    2016-04-05 14:10:43
WE228JP_2016_04_15__10_03_03    PL1W7D-105122   5001    2016-04-05 14:10:45

我也知道

ID 500 600 ... 5000表示步骤（程序）的开始时间

ID 501 601 ... 5001表示步骤（程序）的结束时间

所以我需要知道工作站每天工作多长时间：

所以期望的输出就像这样（例子）：

data_name                     workstation_name        Hours
2016-04-05                    PL1W7D-105120     22 hours
2016-04-06                    PL1W7D-105120     21 hours
2016-04-15                    PL1W7D-105122     17 hours

任何暗示我怎么能这样做？什么都行;）

Answer 1

我认为你只想要聚合。这是一种方法：

select d.data_name, h.workstation_name, h.type_id,
       datediff(hour, min(insert_date), max(insert_date)) as hours_sortof
from data d join
     process_data_history h 
     on h.data_id = d.data_id
where h.type_id between '500' and '5001' and 
      h.workstation_name like 'PL1W7D-105120'
      h.insert_date >= '2016-04-01' and h.insert_date < '2016-05-01'
group by d.date_name, h.workstation_name,
order by d.data_name, start ;

注意：

学习使用显式JOIN语法。从不在FROM子句中使用逗号。
出于比较目的，请勿将日期转换为字符串。 SQL Server具有内置于系统中的非常好的日期/时间功能。
仅对字符串常量使用LIKE。隐式转换可能是问题的主要来源。
hours_sortof基于SQL Server处理时间差异的方式。它计算小时边界而不是实际小时数。要进一步细化，请使用较小的增量，例如分钟或秒。

Answer 2

将输出数据放入表（或临时表）中，这应该可行。您的数据有一个在结束前开始的事件！

;with  x as
(
    select  data_name, workstation_name, event_id = type_id/10, endEventIndicator = type_id%10, timestamp = start
from Data
)
select Date = cast(starts.timestamp as date), starts.workstation_name, sum(DateDiff(hour, starts.timestamp, ends.timestamp))
from x starts inner join x as ends
on starts.event_id =  ends.event_id
and starts.endEventIndicator = 0 and ends.endEventIndicator = 1
and starts.workstation_name = ends.workstation_name
group by cast(starts.timestamp as date), starts.workstation_name

SQL如何确定每天的日期和工作站的小时数

2 个答案: