SQL 使用不同的条件和多列分组

Question

I have an Addresses Table, with columns being: A,B,C,D......Address, Number, State, City, Zip **

I need to get a count of distinct addresses that can be grouped by 2 different conditions:

(Address, Number, State, City) OR (Address, Number, Zip)

The reason for the 2 conditions is that Zip can be null in some cases with City and State provided and vice versa. However, in cases where all 3 are present, the final table should just have 1 row for that address

select State, City, Address, Number
into T1
from Main Table
group by State, City, Address, Number

select Zip, Address, Number
into T2
from Main Table
group by Zip, Address, Number

select a.*, b.Zip into JoinT1T2 from T1 a inner join T2 b 
on a.Address=b.Address AND a.Number=b.Number 

select count(*) from JoinT1T2

But this is not giving me the desired result.

Answer 1

You can create a new field called TypeOfDir, using the CASE instruction to get a value from 1 to 3 for the differents conditions that you have. Then you can group by that new field.

Answer 2

What you need to do is define the condition as an extra column. You don't need to add this to the table (although it will help performance), you can do this on the fly if you want, by using CROSS APPLY .

SELECT COUNT(*) FROM (
    SELECT DISTINCT HasZip, State, City, Zip, Address, Number
    FROM Table t
    CROSS APPLY (VALUES (CASE WHEN Zip IS NULL THEN 0 ELSE 1 END) ) v(HasZip)
) t