Monday, February 8, 2010

Union Query: Part 2


More Union Queries
Last time in Union Queries: Part 1, I discussed simple Union queries and some of their uses. This time I want to talk about some advanced topics, starting with Union All.

Union Vs. Union All

As you remember from last time, a Union query takes the result of one query and "appends" it to another. At the same time, it removes duplicates and orders the records on the first field in the field list.

The Union All does the same thing as the Union, but without removing the duplicates or sorting the resultset, so:

SELECT * FROM TableA
UNION ALL
SELECT * FROM TableB


Would return:

CustomerID CreditLimit
---------- -----------
1001       $500
1010       $350
1017       $375
1020       $425
1017       $375
1008       $600


The order is not determined by the base tables. The result has no order, so the records could be returned in any order.

Sorting Union Queries

So what if you want to sort your Union All or want a different sort order for your Union? Like any Select query, you can add an Order By clause to the end.

SELECT CustomerID, CreditLimit
FROM TableA
UNION ALL
SELECT CustomerID, CreditLimit
FROM TableB
ORDER BY CustomerID;


Will return:

CustomerID CreditLimit
---------- -----------
1001       $500
1008       $600
1010       $350
1020       $425
1017       $375
1017       $375


The Order By applies to the entire result. If the column names are different between queries (which, you'll remember, I said was allowable), the column names from the *first* query must be used. Order By clauses in any of the other component queries will be ignored. Only the last one will be applied.

Restricting Rows

Like any other Select query, you can restrict rows from the resultset with a Where clause. Unlike the Order By, however, each component query can have its own criteria. If you put a Where clause on the last query, it will apply to ONLY the results of the last query. So this:

SELECT CustomerID, CreditLimit
FROM TableA
UNION
SELECT CustomerID, CreditLimit
FROM TableB
WHERE CreditLimit > 500


Produces this:

CustomerID CreditLimit
---------- -----------
1001       $500
1008       $600
1010       $350
1017       $375


Since 1010 and 1017 are in TableA, they can be less than 500, whereas 1020 is in TableB, so it is restricted. If you want a Where to apply to the entire resultset, you have to put it in each query.

SELECT CustomerID, CreditLimit
FROM TableA
WHERE CreditLimit > 500
UNION
SELECT CustomerID, CreditLimit
FROM TableB
WHERE CreditLimit > 500


Tables With Unequal Numbers Of Columns

As I said, the individual queries don't have to have the same column names, nor do they have to be the same datatype, but the *number* of columns in the field list must be equal.

What if your tables have unequal columns? In that case, you need to create false columns with an alias.

So suppose you have two tables which hold much the same data, but have slightly different structures.*

TABLE A
CustomerID CreditLimit
---------- -----------
1001       $500
1010       $350
1017       $375


TABLE B
CustomerNum Credit Active
---------- ------ ------
1008       $600   Y
1017       $375   Y
1020       $425   Y


*Please Note: I am not advocating this as a good database design. In a properly designed database, you would never have two tables which stored the same type of data. However, you may find yourself in this situation if you are fixing a poorly designed database or if you are merging separate databases.

If I don't know the value of "Active" for TableA, I can leave the value NULL.

SELECT CustomerID, CreditLimit, NULL AS Active
FROM TableA
UNION
SELECT CustomerNum, Credit, Active
FROM TableB;


CustomerID CreditLimit Active
---------- ----------- ------
1001       $500
1008       $600        Y
1010       $350
1017       $375        Y
1020       $425        Y


However, if I assume all of the customers in TableA are active

SELECT CustomerID, CreditLimit, "Y" AS Active
FROM TableA
UNION
SELECT CustomerNum, Credit, Active
FROM TableB;


CustomerID CreditLimit Active
---------- ----------- ------
1001       $500        Y
1008       $600        Y
1010       $350        Y
1017       $375        Y
1020       $425        Y


Because Union Queries cannot be viewed in the Query Builder, they are often over looked by Access novices, but they are a powerful tool to have in your SQL arsenal.

.

2 comments:

Khiel said...

I am really struggling trying to use a Union Query to input information in a form. I read your article "Recordset not updatable. Why?", and I am now puzzled with WHY NOT. It would be so useful to me to use the result of the Union Query as a querry for inputing into two tables at the same time, but this is simply impossible. So, why not? Is there a way to get around this?

Roger Carlson said...

Useful or not, it's one of the limitations of Access SQL.

As for solutions, there are probably several, depending on the specifics of your problem.

But this is not the best venue for support. I suggest asking the question on Utter Access Forums
www.utteraccess.com