sum from two tables insert if not exists

Question

I have a situation where I want to sum difference between two tables. problem is that a row can exist in the second table and then I want to insert it as new row.

Pseudo

SELECT T1.seller, T1.code, T1.amount - T2.amount

Look at the image for an explanation

LINK

Answer 1

 DECLARE @T1 TABLE(
    seller VARCHAR(10),
    code VARCHAR(3) NULL,
    amount MONEY
)

 DECLARE @T2 TABLE(
    seller VARCHAR(10),
    code VARCHAR(3) NULL,
    amount MONEY
)

INSERT INTO @T1 VALUES
('VL',NULL,1),
('VL','317',70005.6)

INSERT INTO @T2 VALUES
('VL',NULL,0.5),
('VL','500',4450)

SELECT seller,code,SUM(amount) [amount] FROM 
(
SELECT * FROM @T1
UNION ALL
SELECT seller,code,-amount as amount FROM @T2
) T
GROUP BY seller,code

Answer 2

You'll need to make a full outer join - then sum it up. If you run only the inner query you will get each possible row combination (exclusive in t1, exist both in t1 and t2, exlclusive in t2) - you then group it up and do the sum.

SELECT  Seller ,
        Code ,
        SUM(Tab1_amt - Tab2_amt) AS Amount
FROM    ( SELECT    COALESCE(tab1.Seller, tab2.Seller) AS Seller ,
                    COALESCE(tab1.code, tab2.code) AS Code ,
                    COALESCE(tab1.amount, 0) AS tab1_amt ,
                    COALESCE(tab2.amount, 0) AS tab2_amt
          FROM      tab1
                    FULL OUTER JOIN tab2 ON tab1.seller = tab2.seller
                                            AND tab1.code = tab2.code
        ) AS Tbl
GROUP BY Seller ,
        Code

See SQLFiddle Demo

Answer 3

Something like this? your not telling what table to insert into "therefore tableX"

insert into tableX (seller,code,amount) values (T1.seller, T1.code, T1.amount -T2.amount)
select count(*) from table2 having count(*) < 1

notice you put the filter as a select statement after your insert to apply filter you didn't specify what type of sql you need, so i cant say if this works

Answer 4

I think you need a FULL JOIN (unless you actually want the row inserting to the first table)

SELECT  COALESCE(t1.Seller, t2.Seller) AS Seller,
        COALESCE(t1.Code, t2.Code) AS Code,
        COALESCE(t1.Amount, 0) - COALESCE(t2.Amount, 0) AS Amount
FROM    Table1 t1
        FULL JOIN Table2 t2
            ON t1.Seller = t2.Seller
            AND COALESCE(t1.Code, 0) = COALESCE(t2.Code, 0);

If you do need the row inserting to table 1 you will need to do 2 operations, first insert, then select:

INSERT Table1 (Seller, Code, Amount)
SELECT  t2.Seller, t2.Code, 0 AS Amount
FROM    Table2 t2
WHERE   NOT EXISTS
        (   SELECT  1
            FROM    Table1 t1
            WHERE   t1.Seller = t2.Seller
            AND     COALESCE(t1.Code, 0) = COALESCE(t2.Code, 0)
        );

SELECT  t1.Seller, 
        t1.Code, 
        t1.Amount - COALESCE(t2.Amount, 0) AS Amount
FROM    Table1 t1
        LEFT JOIN Table2 t2
            ON t1.Seller = t2.Seller
            AND COALESCE(t1.Code, 0) = COALESCE(t2.Code, 0);

EDIT

IF rows in each table are not unqiue and you need to sum them then you will need to do the sums in subqueries as the JOIN will introduce cross joining:

Consider this data

Table1
Seller  Code    Amount
VL      500     10
VL      500     20

Table2
Seller  Code    Amount
VL      500     30
VL      500     5

When you join this you will get:

t1.Seller   t1.Code t1.Amount   t2.Seller   t2.Code t2.Amount
VL          500     10          VL          500     30
VL          500     10          VL          500     5
VL          500     20          VL          500     30
VL          500     20          VL          500     5

The sum of the difference is then -10 instead of -5.

SELECT  COALESCE(t1.Seller, t2.Seller) AS Seller,
        COALESCE(t1.Code, t2.Code) AS Code,
        COALESCE(t1.Amount, 0) - COALESCE(t2.Amount, 0) AS Amount
FROM    (   SELECT  Seller, Code, SUM(Amount) AS Amount
            FROM    Table1 
            GROUP BY Seller, Code
        ) t1
        FULL JOIN 
        (   SELECT  Seller, Code, SUM(Amount) AS Amount
            FROM    Table2 
            GROUP BY Seller, Code
        ) t2
            ON t1.Seller = t2.Seller
            AND COALESCE(t1.Code, 0) = COALESCE(t2.Code, 0);

EDIT 2

The UNION method in Daniel's answer will perform much better than the FULL JOIN:

SELECT  Seller, Code, Amount = SUM(Amount)
FROM    (   SELECT  Seller, Code, Amount
            FROM    Table1
            UNION
            SELECT  Seller, Code, -Amount
            FROM    Table2
        ) t
GROUP BY Seller, Code