T SQL查询格式构建动态表

Question

我需要帮助来创建查询以生成我正在寻找的结果。

我正在创建一个应用程序来根据现有数据库表跟踪员工出勤情况。用户从日历中选择一个日期或一组日期（比如10/1/18、10/2/18 和10/3/18）。他们点击提交，我需要生成一个表格，显示每位员工，如果他们当天在场，则在日期列中带有复选标记。

该表名为 History，有 2 个主要列：EmployeeID；和交易日期。每次员工走过一扇门（包括入口）时，都会创建一个历史事务（NFC 徽章），该事务会添加一个带有员工 ID 和日期/时间戳的新行。每个员工每天可能有几次滑动，但我真正需要知道的是那天是否有一次滑动。我正在发布表格的外观以及查询结果和我的表格的外观的图片...

表格：

结果：

我可以这样查询：

select employeeid, TranDate from History 
where (CAST(trandate as DATE) = '2018-10-1' or CAST(trandate as DATE) = '2018-10-2' or CAST(trandate as DATE) = '2018-10-3' ) 
order by employeeid, TranDate

并以编程方式对其进行排序，但我觉得有一种更有效的方法来查询我正在寻找的结果。

非常感谢任何帮助。如果需要更好的解释，请告诉我。

Answer 1

你想做的就是所谓的pivot

尝试使用此代码

SELECT employeeid
      ,case when [2018-10-1] > 0 then '1' else '' end as [2018-10-1]
      ,case when [2018-10-2] > 0 then '1' else '' end as [2018-10-2]
      ,case when [2018-10-3] > 0 then '1' else '' end as [2018-10-3]
FROM(
SELECT employeeid
     , TranDate
     , 1 as num 
FROM History 
WHERE (CAST(trandate as DATE) = '2018-10-1' 
       or CAST(trandate as DATE) = '2018-10-2' 
       or CAST(trandate as DATE) = '2018-10-3' ) 
)
PIVOT(
COUNT(num)
FOR TranDate IN ([2018-10-1],[2018-10-2],[2018-10-3])
) as pvt

您可以在PIVOT, UNPIVOT SQL 上查看关于 PIVOT 和 UNPIVOT 的 sql 文档，此外，如果您想要更动态的转换字段，您可以查看 Dynamic Pivot

Answer 2

你说：“我真正需要的是知道那天是否有单次滑动。”

答案 - 聚合查询

select employeeid, TranDate, count(TranDate) as count
from History 
where -- look below
Group by employeeid, TranDate

在WHERE 中输入什么？ - 依靠。如果你想要月份，那就做

TranDate between '2018-10-1' and '2018-10-31 23:59:59.999'

如果您想要特定日期，请执行

CAST(trandate as DATE) in ('2018-10-1', '2018-10-11', '2018-10-11')

在后一种情况下，您必须动态构建它

但在这种情况下，您将面临一个新问题 - 如果您的人从未进过门怎么办？此人不会出现在列表中。然后你需要letf join这个有一个包含所有员工的表

现在因为我不是 TSql 超级大师，但我喜欢编码，我向您展示工作代码如何以编程方式完成。下面是几个可以组合成一个的sn-ps代码

-- SETUP
create table recs (id int, dt date );
insert into recs values(1, '2018-10-1');
insert into recs values(1, '2018-10-1');
insert into recs values(2, '2018-10-2');
insert into recs values(2, '2018-10-2');
insert into recs values(2, '2018-10-3');
insert into recs values(3, '2018-10-3');
insert into recs values(3, '2018-10-3');
insert into recs values(3, '2018-10-4');
insert into recs values(3, '2018-10-1'); 


-- Prepare Date Grid table
DECLARE crs_dt CURSOR FOR SELECT dt FROM recs group by dt order by dt;
declare @createTbl varchar(1200) = 'create table tbl (id int, ';
declare @fetched varchar(20);

OPEN crs_dt;  
FETCH NEXT FROM crs_dt into @fetched;  

WHILE @@FETCH_STATUS = 0  
BEGIN  
   set @createTbl = @createTbl + '['+@fetched+'] int'
   FETCH NEXT FROM crs_dt into @fetched;
   if @@FETCH_STATUS = 0
   begin set @createTbl = @createTbl + ','; end
END  

set @createTbl = @createTbl + ')';

CLOSE crs_dt;  
DEALLOCATE crs_dt;

select @createTbl; -- just a test
execute (@createTbl)
SELECT * FROM tbl; -- just a test


-- LOAD date grid table with data
DECLARE crs_rec CURSOR FOR 
SELECT id, dt, FORMAT ( dt, 'yyyy-MM-dd' ) colName, 
(case count(dt)  when 0 then 0 else 1 end) cnt
FROM recs group by id, dt order by dt;

declare @createInsert varchar(1200);
declare @id int, @dt date, @colName varchar(20),@yesNo int;

OPEN crs_rec;  
FETCH NEXT FROM crs_rec into @id, @dt, @colName, @yesNo;  

WHILE @@FETCH_STATUS = 0  
BEGIN
   if exists(select 1 from tbl where id = @id)
       set @createInsert = 'update tbl set ['+@colName+']='+cast(@yesNo as varchar(1))+ ' where id='+ cast(@id as varchar(1000));
   else  
       set @createInsert = 
           'insert into tbl (id,['+@colName+']) values ('+cast(@id as varchar(1000))+','+cast(@yesNo as varchar(1))+')';
   select @createInsert; -- just a test
   execute (@createInsert);

   FETCH NEXT FROM crs_rec into  @id, @dt, @colName, @yesNo;
END  
CLOSE crs_rec;  
DEALLOCATE crs_rec;

commit;

-- Lets Load data 
SELECT * FROM tbl

结果是……

不是最快或最高效的代码，但肯定很有趣。而且，1-这是完全动态的。 2-如果你看结果，现在回到我之前提到的问题，你可以将此表加入到完整的员工列表中，并获取整组数据，包括在所选时间段内没有刷卡的员工。

Answer 3

所以我想出了一种方法来获得我正在寻找的结果。

首先我需要获取不同的员工 ID 并将它们存储到临时表中：

create table #TempTable
(
    EmployeeID varchar(25)
)

insert into #TempTable (EmployeeID)

select distinct employeeID from History order by employeeID;

从这里，我可以将我的临时表加入到我的历史表中，并准确地选择我要查找的内容：

SELECT distinct M.EmployeeID AS Employee, 
(SELECT CASE WHEN EXISTS
(SELECT 1 FROM HISTORY WHERE TAGID = M.EmployeeID AND (CAST(TRANDATE AS DATE) = '10/12/2018')) 
THEN CAST (1 AS BIT) ELSE CAST (0 AS BIT) END) AS '10/12/2018'           
FROM #TempTable AS M left JOIN HISTORY AS H ON M.EmployeeID = H.TAGID where m.EmployeeID = '000000000000000000000001' order BY Employee

最后删除临时表：

drop table #TempTable

对于更多日期，我只需添加一个新列，将日期作为标题。 感谢大家在这个问题上的帮助。

Answer 4

使用@T.S.创建表：

create table recs (id int, dt date );
insert into recs values(1, '2018-10-1');
insert into recs values(1, '2018-10-1');
insert into recs values(2, '2018-10-2');
insert into recs values(2, '2018-10-2');
insert into recs values(2, '2018-10-3');
insert into recs values(3, '2018-10-3');
insert into recs values(3, '2018-10-3');
insert into recs values(3, '2018-10-4');
insert into recs values(3, '2018-10-1'); 

declare @dts as varchar(max) =''
declare @dtsSelect as varchar(max) =''
select  @dts =  @dts + ',[' +dt +']',
             @dtsSelect =  @dtsSelect + ',Case WHEN [' +dt+']>0 then 1 else 0 end ['+dt+']'
From (select distinct cast(dt as varchar(100)) dt from recs) recs
//we use some variables to create columns should apear in pivot list and 
// also to check if we have a value for column then put 1 instead
set @dts = Stuff(@dts,1,1,'')
set @dtsSelect = Stuff(@dtsSelect,1,1,'')
//delete the first comma

declare @sql nvarchar(max) = '
select id,'+@dtsSelect+' from recs 
pivot(Count(dt) For dt in('+@dts+'))PV

'
exec sp_executesql @stme = @sql

假设日期并不总是相同

Answer 5

如果枢轴列是固定的，您可以使用它

编辑 - 我猜它也可以是动态的，因为您可以动态创建 SQL 语句。

Microsoft Docs : FROM - Using PIVOT and UNPIVOT

例子

select * from History
PIVOT  
(
    COUNT(TranDate)
    FOR TranDate IN ([2018-10-01], [2018-10-02], [2018-10-03])  
) AS PivotTable;