Sams Teach Yourself SQL in 24 Hours

Recommended Book 

Types of Joins

While different implementations have many ways of joining tables, you concentrate on the most common joins in this lesson. The types of joins that you learn are

EQUIJOINS 
NATURAL JOINS
NON-EQUIJOINS
OUTER JOINS
SELF JOINS

Component Locations of a Join Condition

As you have learned from previous hours, the SELECT and FROM clauses are both required SQL statement elements; the WHERE clause is a required element of a SQL statement when joining tables. The tables being joined are listed in the FROM clause. The join is performed in the WHERE clause. Several operators can be used to join tables, such as =, <, >, <>, <=, >=, !=, BETWEEN, LIKE, and NOT; they can all be used to join tables. However, the most common operator is the equal symbol.

Joins of Equality

Perhaps the most used and important of the joins is the EQUIJOIN, also referred to as an INNER JOIN. The EQUIJOIN joins two tables with a common column in which each is usually the primary key.

The syntax for an EQUIJOIN is

   
SELECT TABLE1.COLUMN1, TABLE2.COLUMN2...
FROM TABLE1, TABLE2 [, TABLE3 ]
WHERE TABLE1.COLUMN_NAME = TABLE2.COLUMN_NAME
[ AND TABLE1.COLUMN_NAME = TABLE3.COLUMN_NAME ]

Look at the following example:

SELECT EMPLOYEE_TBL.EMP_ID,
       EMPLOYEE_PAY_TBL.DATE_HIRE
FROM EMPLOYEE_TBL,
       EMPLOYEE_PAY_TBL
WHERE EMPLOYEE_TBL.EMP_ID = EMPLOYEE_PAY_TBL.EMP_ID;

This SQL statement returns the employee identification and the employee's date of hire. The employee identification is selected from the EMPLOYEE_TBL (although it exists in both tables, you must specify one table), whereas the hire date is selected from the EMPLOYEE_PAY_TBL. Because the employee identification exists in both tables, both columns must be justified with the table name. By justifying the columns with the table names, you tell the database server where to get the data.

Data in the following example is selected from tables EMPLOYEE_TBL and EMPLOYEE_PAY_TBL tables because desired data resides in each of the two tables. An equality join is used.

   

   SELECT EMPLOYEE_TBL.EMP_ID, EMPLOYEE_TB
   L.LAST_NAME,
       
   EMPLOYEE_PAY_TBL.POSITION

   FROM EMPLOYEE_TBL, EMPLOYEE_PAY_TBL

   WHERE EMPLOYEE_TBL.EMP_ID = EMPLOYEE_PAY_TBL.EMP_ID;

   
EMP_ID    LAST_NAM POSITION
--------- -------- -------------
311549902 STEPHENS MARKETING
442346889 PLEW     TEAM LEADER
213764555 GLASS    SALES MANAGER
313782439 GLASS    SALESMAN
220984332 WALLACE  SHIPPER
443679012 SPURGEON SHIPPER

6 rows selected.

Notice that each column in the SELECT clause is preceded by the associated table name in order to identify each column. This is called qualifying columns in a query. Qualifying columns is only necessary for columns that exist in more than one table referenced by a query. You usually qualify all columns for consistency and to avoid any questions when debugging or modifying SQL code.

Natural Joins

A NATURAL JOIN is nearly the same as the EQUIJOIN; however, the NATURAL JOIN differs from the EQUIJOIN by eliminating duplicate columns in the joining columns. The JOIN condition is the same, but the columns selected differ.

The syntax is as follows:

   
SELECT TABLE1.*, TABLE2.COLUMN_NAME
       [ TABLE3.COLUMN_NAME ]
FROM TABLE1, TABLE2 [ TABLE3 ]
WHERE TABLE1.COLUMN_NAME = TABLE2.COLUMN_NAME
[ AND TABLE1.COLUMN_NAME = TABLE3.COLUMN ]

Look at the following example:

SELECT EMPLOYEE_TBL.*, EMPLOYEE_PAY_TBL.SALARY 
FROM EMPLOYEE_TBL,
     EMPLOYEE_PAY_TBL
WHERE EMPLOYEE_TBL.EMP_ID = EMPLOYEE_PAY_TBL.EMP_ID;

This SQL statement returns all columns from EMPLOYEE_TBL and SALARY from the EMPLOYEE_PAY_TBL. The EMP_ID is in both tables, but is retrieved only from the EMPLOYEE_TBL because both contain the same information and do not need to be selected.

The following example selects all columns from the EMPLOYEE_TBL table and only one column from the EMPLOYEE_PAY_TBL table. Remember that the asterisk (*) represents all columns of a table.

   

   SELECT EMPLOYEE_TBL.*, EMPLOYEE_PAY_TBL
   .POSITION

   FROM EMPLOYEE_TBL, EMPLOYEE_PAY_TBL

   WHERE EMPLOYEE_TBL.EMP_ID = EMPLOYEE_PAY_TBL.EMP_ID;

   
EMP_ID    LAST_NAM FIRST_NA M ADDRESS       CITY         ST ZIP   PHONE
--------- -------- -------- - ------------- ------------ -- ----- ----------
PAGER      POSITION
---------- --------------
311549902 STEPHENS TINA     D RR 3 BOX 17A  GREENWOOD    IN 47890 3178784465
           MARKETING

442346889 PLEW     LINDA    C 3301 BEACON   INDIANAPOLIS IN 46224 3172978990
           TEAM LEADER

213764555 GLASS    BRANDON  S 1710 MAIN ST  WHITELAND    IN 47885 3178984321
3175709980 SALES MANAGER

313782439 GLASS    JACOB    3789 RIVER BLVD INDIANAPOLIS IN 45734 3175457676
8887345678 SALESMAN

220984332 WALLACE  MARIAH   7889 KEYSTONE   INDIANAPOLIS IN 46741 3173325986
           SHIPPER

443679012 SPURGEON TIFFANY  5 GEORGE COURT  INDIANAPOLIS IN 46234 3175679007
           SHIPPER


6 rows selected.

Using Table Aliases

The use of table aliases means to rename a table in a particular SQL statement. The renaming is a temporary change. The actual table name does not change in the database. As you will learn later in this hour, giving the tables aliases is a necessity for the SELF JOIN. Giving tables aliases is most often used to save keystrokes, which results in the SQL statement being shorter and easier to read. In addition, fewer keystrokes means fewer keystroke errors. Also, programming errors are typically less frequent if you can refer to an alias, which is often shorter in length and more descriptive of the data with which you are working. Giving tables aliases also means that the columns being selected must be qualified with the table alias. The following are some examples of table aliases and the corresponding columns:

SELECT E.EMP_ID, EP.SALARY, EP.DATE_HIRE, E.LAST_NAME 
FROM EMPLOYEE_TBL E,
     EMPLOYEE_PAY_TBL EP
WHERE E.EMP_ID = EP.EMP_ID
AND EP.SALARY > 20000;

The tables have been given aliases in the preceding SQL statement. The EMPLOYEE_TBL has been renamed E. The EMPLOYEE_PAY_TBL has been renamed EP. The choice of what to rename the tables is arbitrary. The letter E is chosen because the EMPLOYEE_TBL starts with E. Because the EMPLOYEE_PAY_TBL also begins with the letter E, you could not use E again. Instead, the first letter (E) and the first letter of the second word in the name (PAY) are used as the alias. The selected columns were justified with the corresponding table alias. Note that SALARY was used in the WHERE clause and must also be justified with the table alias.

Joins of Non-Equality

NON-EQUIJOIN joins two or more tables based on a specified column value not equaling a specified column value in another table. The syntax for the NON-EQUIJOIN is

   
FROM TABLE1, TABLE2 [, TABLE3 ]
WHERE TABLE1.COLUMN_NAME != TABLE2.COLUMN_NAME
[ AND TABLE1.COLUMN_NAME != TABLE2.COLUMN_NAME ]

An example is as follows:

SELECT EMPLOYEE_TBL.EMP_ID, EMPLOYEE_PAY_TBL.DATE_HIRE 
FROM EMPLOYEE_TBL,
     EMPLOYEE_PAY_TBL
WHERE EMPLOYEE_TBL.EMP_ID != EMPLOYEE_PAY_TBL.EMP_ID;

The preceding SQL statement returns the employee identification and the date of hire for all employees who do not have a corresponding record in both tables. The following example is a join of non-equality:

   

   SELECT E.EMP_ID, E.LAST_NAME, P.POSITIO
   N

   FROM EMPLOYEE_TBL E,
     
   EMPLOYEE_PAY_TBL P

   WHERE E.EMP_ID <> P.EMP_ID;

   
EMP_ID    LAST_NAM POSITION
--------- -------- -------------
442346889 PLEW     MARKETING
213764555 GLASS    MARKETING
313782439 GLASS    MARKETING
220984332 WALLACE  MARKETING
443679012 SPURGEON MARKETING
311549902 STEPHENS TEAM LEADER
213764555 GLASS    TEAM LEADER
313782439 GLASS    TEAM LEADER
220984332 WALLACE  TEAM LEADER
443679012 SPURGEON TEAM LEADER
311549902 STEPHENS SALES MANAGER
442346889 PLEW     SALES MANAGER
313782439 GLASS    SALES MANAGER
220984332 WALLACE  SALES MANAGER
443679012 SPURGEON SALES MANAGER
311549902 STEPHENS SALESMAN
442346889 PLEW     SALESMAN
213764555 GLASS    SALESMAN
220984332 WALLACE  SALESMAN
443679012 SPURGEON SALESMAN
311549902 STEPHENS SHIPPER
442346889 PLEW     SHIPPER
213764555 GLASS    SHIPPER
313782439 GLASS    SHIPPER
443679012 SPURGEON SHIPPER
311549902 STEPHENS SHIPPER
442346889 PLEW     SHIPPER
213764555 GLASS    SHIPPER
313782439 GLASS    SHIPPER
220984332 WALLACE  SHIPPER

30 rows selected.

You may be curious why 30 rows were retrieved when only 6 rows exist in each table. For every record in EMPLOYEE_TBL, there is a corresponding record in EMPLOYEE_PAY_TBL. Because non-equality was tested in the join of the two tables, each row in the first table is paired with all rows from the second table, except for its own corresponding row. This means that each of the 6 rows are paired with 5 unrelated rows in the second table; 6 rows multiplied by 5 rows equals 30 rows total.

In the previous section's test for equality example, each of the six rows in the first table were paired with only one row in the second table (each row's corresponding row); six rows multiplied by one row yields a total of six rows.

Outer Joins

An OUTER JOIN is used to return all rows that exist in one table, even though corresponding rows do not exist in the joined table. The (+) symbol is used to denote an OUTER JOIN in a query. The (+) is placed at the end of the table name in the WHERE clause. The table with the (+) should be the table that does not have matching rows. In many implementations, the OUTER JOIN is broken down into joins called LEFT OUTER JOIN, RIGHT OUTER JOIN, and FULL OUTER JOIN. The OUTER JOIN in these implementations is normally optional.

The general syntax is

   
FROM TABLE1
{RIGHT | LEFT | FULL} [OUTER] JOIN
ON TABLE2

The Oracle syntax is

FROM TABLE1, TABLE2 [, TABLE3 ] 
WHERE TABLE1.COLUMN_NAME[(+)] = TABLE2.COLUMN_NAME[(+)]
[ AND TABLE1.COLUMN_NAME[(+)] = TABLE3.COLUMN_NAME[(+)]]

The concept of the OUTER JOIN is explained in the next two examples. In the first example, the product description and the quantity ordered are selected; both values are extracted from two separate tables. One important factor to keep in mind is that there may not be a corresponding record in the ORDERS_TBL table for every product. A regular join of equality is performed:

   

   SELECT P.PROD_DESC, O.QTY

   FROM PRODUCTS_TBL P,
     
   ORDERS_TBL O

   WHERE P.PROD_ID = O.PROD_ID;

   
PROD_DESC                        QTY
-------------------------------- ---
WITCHES COSTUME                    1
PLASTIC PUMPKIN 18 INCH           25
PLASTIC PUMPKIN 18 INCH            2
LIGHTED LANTERNS                  10
FALSE PARAFFIN TEETH              20
KEY CHAIN                          1

6 rows selected.

Only six rows were selected, but there are 10 distinct products. You want to display all products, whether the products have been placed on order or not.

The next example accomplishes the desired output through the use of an OUTER JOIN. Oracle's syntax is used for the OUTER JOIN.

   
   

   SELECT P.PROD_DESC, O.QTY

   FROM PRODUCTS_TBL P,
     
   ORDERS_TBL O

   WHERE P.PROD_ID = O.PROD_ID(+);


   
PROD_DESC                        QTY
-------------------------------- ---
WITCHES COSTUME                    1
ASSORTED MASKS
FALSE PARAFFIN TEETH              20
ASSORTED COSTUMES
PLASTIC PUMPKIN 18 INCH           25
PLASTIC PUMPKIN 18 INCH            2
PUMPKIN CANDY
PLASTIC SPIDERS
CANDY CORN
LIGHTED LANTERNS                  10
KEY CHAIN                          1
OAK BOOKSHELF

12 rows selected.

All products were returned by the query, even though they may not have had a quantity ordered. The outer join is inclusive of all rows of data in the PRODUCTS_TBL table, whether a corresponding row exists in the ORDERS_TBL table or not.

Self Joins

The SELF JOIN is used to join a table to itself, as if the table were two tables, temporarily renaming at least one table in the SQL statement. The syntax is as follows:

   
SELECT A.COLUMN_NAME, B.COLUMN_NAME, [ C.COLUMN_NAME ]
FROM TABLE1 A, TABLE2 B [, TABLE3 C ]
WHERE A.COLUMN_NAME = B.COLUMN_NAME
[ AND A.COLUMN_NAME = C.COLUMN_NAME ]

The following is an example:

SELECT A.LAST_NAME, B.LAST_NAME, A.FIRST_NAME 
FROM EMPLOYEE_TBL A,
     EMPLOYEE_TBL B
WHERE A.LAST_NAME = B.LAST_NAME;

The preceding SQL statement returns the employees' first name for all the employees with the same last name from the EMPLOYEE_TBL. Self joins are useful when all of the data you want to retrieve resides in one table, but you must somehow compare records in the table to other records in the table.

Another common example used to explain a self join is as follows. Suppose you have a table that stores an employee identification number, the employee's name, and the employee identification number of the employee's manager. You may want to produce a list of all employees and their managers' names. The problem is that the manager name does not exist in the table, only the employee name:

SELECT * FROM EMP; 

ID   NAME      MGR_ID
---- --------- ------
1    JOHN      0
2    MARY      1
3    STEVE     1
4    JACK      2
5    SUE       2

In the following example, we have included the table EMP twice in the FROM clause of the query, giving the table two aliases for the purpose of the query. By providing two aliases, it is as if you are selecting from two distinct tables. All managers are also employees, so the join condition between the two tables compares the value of the employee identification number from the first table with the manager identification number in the second table. The first table acts as a table that stores employee information, whereas the second table acts as a table that stores manager information:

SELECT E1.NAME, E2.NAME 
FROM EMP E1, EMP E2
WHERE E1.MGR_ID = E2.ID;

NAME      NAME
--------- ---------
MARY      JOHN
STEVE     JOHN
JACK      MARY
SUE       MARY

Joining on Multiple Keys

Most join operations involve the merging of data based on a key in one table and a key in another table. Depending on how your database has been designed, you may have to join on more than one key field to accurately depict that data in your database. You may have a table that has a primary key that is comprised of more than one column. You may also have a foreign key in a table that consists of more than one column, which references the multiple column primary key.

Consider the following Oracle tables that are used here for examples only:

SQL> desc prod 
 Name                                      Null?    Type
 ----------------------------------------- -------- ----------------------------
 SERIAL_NUMBER                             NOT NULL NUMBER(10)
 VENDOR_NUMBER                             NOT NULL NUMBER(10)
 PRODUCT_NAME                              NOT NULL VARCHAR2(30)
 COST                                      NOT NULL NUMBER(8,2)

SQL> desc ord
 Name                                      Null?    Type
 ----------------------------------------- -------- ----------------------------
 ORD_NO                                    NOT NULL NUMBER(10)
 PROD_NUMBER                               NOT NULL NUMBER(10)
 VENDOR_NUMBER                             NOT NULL NUMBER(10)
 QUANTITY                                  NOT NULL NUMBER(5)
 ORD_DATE                                  NOT NULL DATE

The primary key in PROD is the combination of the columns SERIAL_NUMBER and VENDOR_NUMBER. Perhaps two products can have the same serial number within the distribution company, but each serial number is unique per vendor.

The foreign key in ORD is also the combination of the columns SERIAL_NUMBER and VENDOR_NUMBER.

When selecting data from both tables (PROD and ORD), the join operation may appear as follows:

SELECT P.PRODUCT_NAME, O.ORD_DATE, O.QUANTITY 
FROM PROD P, ORD O
WHERE P.SERIAL_NUMBER = O.SERIAL_NUMBER
  AND P.VENDOR_NUMBER = O.VENDOR_NUMBER;