How to account for double entries in a card swipe table?

问题

How do I account for accidental swipes in a card swipe table? I tried selecting the max(time_cst) by grouping them on entry. It did not solve the problem. http://www.sqlfiddle.com/#!18/06cc8/2

EmpID   Enter/Exit  Time 
 2999   Entry   06:00AM 
 2999   Entry   06:01AM 
 2999   Exit    12:00PM
 2999   Entry   01:00PM 
 2999   Exit    03:00PM 
 2999   Entry   04:00PM
 2999   Exit    06:00PM

This is how I want it to show up. http://www.sqlfiddle.com/#!18/5cfbb/2

EmpID   EntryTime  ExitTime
 2999   06:01AM    12:00PM
 2999   01:00PM    03:00PM
 2999   04:00PM    06:00PM

回答1:

I've added a fiddle to show this working but I'll copy the query here for future reference.

http://www.sqlfiddle.com/#!18/87593fe/12/0

Below is very close to the fiddle version, just using a temp table instead of a permanent table.

Also, I edited the sample data as there were extra spaces between the names on the last two records. I assume this was in error.

DROP TABLE IF EXISTS #Table1
CREATE TABLE #Table1
    ([Name] varchar(9), [TimeATT] int, [DateTime] varchar(19))
;

INSERT INTO #Table1
    ([Name], [TimeATT], [DateTime])
VALUES
    ('jane doe', 1, '2019-04-23T08:00:00'),
    ('jane doe', 1, '2019-04-23T08:01:01'),
    ('jane doe', 2, '2019-04-23T12:00:00'),
    ('jane doe', 1, '2019-04-23T12:05:00'),

    ('john doe', 1, '2019-04-23T08:00:00'),
    ('john doe', 2, '2019-04-23T09:00:01'),
    ('john doe', 1, '2019-04-23T09:05:00'),
    ('john doe', 2, '2019-04-23T12:00:00')

SELECT [Name],[EntryTime], [ExitTime]
 FROM ( -- sub query to get matching exit time for each entry if it exists
        SELECT 
         [Name],
         [DateTime] as EntryTime,
         LEAD([DateTime], 1, NULL) OVER(PARTITION BY [Name] ORDER BY [DateTime]) AS ExitTime
         ,TimeATT, [DateTime]
        FROM ( -- subquery to exclude duplicate records
                SELECT * FROM ( -- subquery to identify records to ignore
                              SELECT 
                                  [Name], [TimeATT], [DateTime],
                                  CASE LEAD(TimeATT, 1, 0) OVER(PARTITION BY [Name] ORDER BY [DateTime]) 
                                    WHEN TimeATT THEN 1
                                    ELSE 0
                                  END AS Exclude
                              FROM #Table1) a
                              WHERE Exclude = 0
              ) t
        ) z
  WHERE [TimeATT] = 1 -- filter so left column is always entry time.
  ORDER BY [Name], [DateTime]

Note The LEAD function needs to partitioned by Name to avoid excluding rows where the TimeATT column is the same on two consecutive rows but for different people.

I used LEAD instead of LAG as your sample returned the 2nd "Entry" record when duplicates were found.

This could be written more elegantly with CTEs but this does work.

Here is the final result..

回答2:

You can use the LAG function to look at the previous row and see if the TimeATT value is the same (ordering by Name and then time).

This requires that there never be a possibility of overlapping transactions. Also note the order by in the LAG function and in your query have to be the same in this instance.

  select [Name], [TimeATT], [DateTime],
  case when LAG(TimeATT)OVER(order by [Name],[DateTime]) = TimeATT then 'Dup'
  else ''
  end as Dup
  from Table1
  order by [Name],[DateTime]

http://www.sqlfiddle.com/#!18/06cc8/14/0

You can then use the results of this to do whatever processing needed. One example would be to put the above in a common table expression and use another query. Note that you have to add "TOP XXX" to get order by to work in a CTE, so we have to force it since it is required for the data.

  WITH DupsMarked as (
  select top 100 PERCENT [Name], [TimeATT], [DateTime],
  case when LAG(TimeATT)OVER(order by [Name],[DateTime]) = TimeATT then 'Dup'
  else ''
  end as Dup
  from Table1
  order by [Name],[DateTime]
  )
  SELECT * FROM DupsMarked where Dup = ''

来源：https://stackoverflow.com/questions/55909621/how-to-account-for-double-entries-in-a-card-swipe-table

标签

sql

sql-server-2012

ssrs-2012