SSMS 2018 - Find Gaps in Dates and Flag the Gaps - sql

Question

I have reviewed many posts about how to find gaps in dates and believe that I am close to figuring it out but need just a little extra help. Per my query I am pulling distinct days with a record count for each distinct day. I have added a "Gap_Days" column which should return a zero if no gap from previous date OR the number of days since the previous date. As you can see all of my Gap_Days are zero when in fact I am missing 10/24 and 10/25. Therefore on 10/26 there should be a gap of 2 since the previous date is 10/23.
Thanks in advance for pointing out what I am probably looking right at.
SELECT DISTINCT Run_Date, COUNT(Run_Date) AS Daily_Count,
Gap_Days = Coalesce(DateDiff(Day,Lag(Run_Date) Over (partition by Run_Date order by Run_Date DESC), Run_Date)-1,0)
FROM tblUnitsOfWork
WHERE (Run_Date >= '2022-10-01')
GROUP BY Run_Date
ORDER BY Run_Date DESC;
Run_Date Daily_Count Gap_Days
2022-10-29 00:00:00.000 8431 0
2022-10-28 00:00:00.000 8204 0
2022-10-27 00:00:00.000 8705 0
2022-10-26 00:00:00.000 7885 0
2022-10-23 00:00:00.000 7485 0
2022-10-22 00:00:00.000 8699 0
2022-10-21 00:00:00.000 9212 0
2022-10-20 00:00:00.000 9220 0

Answer 1

First let's set up some demo data:
DECLARE #table TABLE (ID INT IDENTITY, date DATE)
DECLARE #dt DATE
WHILE (SELECT COUNT(*) FROM #table) < 30
BEGIN
SET #dt = DATEADD(DAY,(ROUND(((50 - 1 -1) * RAND() + 1), 0) - 1)-25,CURRENT_TIMESTAMP)
IF NOT EXISTS (SELECT 1 FROM #table WHERE date = #dt) INSERT INTO #table (date) SELECT #dt
END
ID date
--------
1 2022-11-10
2 2022-11-15
3 2022-10-20
...
28 2022-10-14
29 2022-11-13
30 2022-11-21
This gives us a table variable with 30 random dates in a 50 day window. Now let's look for missing dates:
SELECT *, CASE WHEN ROW_NUMBER() OVER (ORDER BY date) > 1 AND LAG(date,1) OVER (ORDER BY date) <> DATEADD(DAY,-1,date) THEN 'GAP! ' + CAST(DATEDIFF(DAY,LAG(date,1) OVER (ORDER BY date),date)-1 AS NVARCHAR) + ' DAYS MISSING!' END
FROM #table
ORDER BY date
All we're doing here is ignoring the first date (since it's expected there wouldn't be one before then) and from then on comparing the last date (using lag ordered by date) to the current date. If it is not a day before the case statement will produce a message with how many days were missing.
ID date MissingDatesFlag
----------------------------
1 2022-10-08 NULL
4 2022-10-09 NULL
25 2022-10-10 NULL
28 2022-10-11 NULL
22 2022-10-15 GAP! 4 DAYS MISSING!
2 2022-10-18 GAP! 3 DAYS MISSING!
12 2022-10-19 NULL
24 2022-10-20 NULL
....
15 2022-11-18 GAP! 3 DAYS MISSING!
29 2022-11-21 GAP! 3 DAYS MISSING!
20 2022-11-22 NULL
Since the demo data is randomly selected your results may vary, but they should be similar.