Question

This is a bit tricky to explain, but I will do my best.

Say I have two tables:

Dates (DateColumn datetime)

SomeData (ID1 int, ID2 int, SomeDate datetime)

---

Dates ALWAYS contains the following dates:

1/1/2012
1/2/2012
1/3/2012

---

SomeData contains data similar to this:

ID  ID1 ID2 SomeDate
1   1   4   1/1/2012
2   1   4   1/2/2012
3   2   3   1/1/2012
4   2   3   1/3/2012
6   5   1   1/2/2012

---

Now, notice how in SomeData, some of the rows have the same ID1 and ID2 combinations. Also notice how some of them are missing certain dates found in the Dates table. I need to somehow join each of those combinations with EVERY date in the Dates table, to find the missing dates.

So on joining, the data would end up looking like this:

ID      ID1 ID2 SomeDate
1       1   4   1/1/2012
2       1   4   1/2/2012
NULL    1   4   1/3/2012 <-- missing from SomeData table

3       2   3   1/1/2012
NULL    2   3   1/2/2012 <-- missing from SomeData table
4       2   3   1/3/2012

NULL    5   1   1/1/2012 <-- missing from SomeData table
6       5   1   1/2/2012
NULL    5   1   1/3/2012 <-- missing from SomeData table

Notice how each ID1+ID2 combination each returns 3 records, using all the dates in the Dates table, and also notice how each “missing” record maintains the ID1+ID2 combination. Any idea how I can pull this off?

Thanks in advance!

Answer 1

CROSS JOIN is your friend here. Here’s a full worked example

DECLARE @Dates TABLE (DateColumn DATETIME)
INSERT @Dates VALUES ('20120101'), ('20120102'), ('20120103')

DECLARE @SomeData TABLE (ID INTEGER IDENTITY(1,1), ID1 INTEGER, ID2 INTEGER, 
    SomeDate DATETIME)
INSERT @SomeData VALUES (1,4,'20120101'),(1,4,'20120102'),(2,3,'20120101'),
    (2,3,'20120103'), (5,1,'20120102')


SELECT sd2.ID, sd1.ID1, sd1.ID2, d.DateColumn
FROM
(
    SELECT ID1, ID2
    FROM @SomeData
    GROUP BY ID1, ID2
) sd1
    CROSS JOIN @Dates d
    LEFT JOIN @SomeData sd2 ON sd1.ID1 = sd2.ID1 
            AND sd1.ID2 = sd2.ID2 
            AND sd2.SomeDate = d.DateColumn

So to break it down:

1. “sd1” subquery is getting a distinct set of all ID1 and ID2 combinations
2. This is then CROSS APPLY’d to the Dates table. The result of this, is for each ID1+ID2 combination, it will result in a row being returned for each date in the Dates table
3. the final LEFT JOIN then tries to match on the combination of ID1+ID2+Date to return the ID value. This returns NULL as the ID if no match is found, or the ID if it is found.