Handout 9 CS-605 Spring 18 Page 1 of 8. Handout 9. SQL Select -- Multi Table Queries. Joins and Nested Subqueries.

Transcription

1 Handout 9 CS-605 Spring 18 Page 1 of 8 Handout 9 SQL Select -- Multi Table Queries. Joins and Nested Subqueries. Joins In Oracle Many queries need to access data from more than one table. To combine (or join) two or more tables in a query: equi-join - connects tuples based on the equality of the attributes of rows in two tables. Most often it involves the equality between a foreign key of one table to a primary key of another. Example: Display the names of all employees along with the name of the department in which they are working. SELECT emp_fname emp_lname, dept_name FROM employee, dept WHERE emp_dept= dept_nbr; emp_deptis the foreign key in the EMPLOYEE table that references the dept_nbr pri- mary key of the DEPT_table. Aletrnative syntax (uses the JOIN operation in the FROM clause) SELECT E.emp_fname ' ' E.emp_lname, D.dept_name FROM employee JOIN dept ON emp_dept = dept_nbr; In the queries above emp_dept= dept_nbr is the join condition, it ties together the rows of the two tables. Note that the first query can be written in the following way, where each column name is prefixed with the name of the table from which the column is taken: or as SELECT EMPLOYEE.emp_fname EMPLOYEE.emp_lname, DEPT.dept_name FROM EMPLOYEE, DEPT WHERE EMPLOYEE.emp_dept= DEPT.dept_nbr; SELECT E.emp_fname E.emp_lname, D.dept_name FROM EMPLOYEE E, DEPT D WHERE E.emp_dept = D.dept_nbr; Such prefix is necessary if two tables have columns with the same name.

2 Handout 9 CS-605 Spring 18 Page 2 of 8 When the WHERE clause is missing or does not contain a join condition, the result is a cross join (a.k.a. Cartesian product) of the two tables: a set of rows that includes every possible row combination from the two tables. In other words, each row from one table is conjoined with each row of the other. Example: what will be produced by the following query? How many rows will be selected assuming that the EMPLOYEE table has 16 rows the DEPT table has 7 rows? SELECT emp_fname emp_lname, dept_name FROM EMPLOYEE, DEPT Note that the WHERE clause may contain conditions other than the join condition. Example: Display the names of those employees that work in building G SELECT emp_fname emp_lname FROM EMPLOYEE, DEPT WHERE emp_dept= dept_nbr and dept_building = G ; More than two tables can be joined together. Example: Display the names of those customer representatives who work with customers from California. For each representative also display the building in which they work. SELECT emp_fname emp_lname, dept_building FROM EMPLOYEE, DEPT, CUSTOMER WHERE cust_state = CA and cust_rep = emp_nbr and emp_dept= dept_nbr ORDER BY cust_rep; Q: Is the result going to possibly contain duplicates? How would you verify correctness of this query? Here is a more complex examples that demonstrates multi-table grouping: Example: Display the number and name of each customer and the total amount of last year s sales to that customer. SELECT cust_nbr, cust_name, SUM (sales_last_year) FROM SALES, CUSTOMER WHERE cust_nbr = sales_cust_nbr GROUP BY cust_nbr, cust_name; Q: Are the customers with no sales records (i.e. 0 sales) going to be included in the result?

3 Handout 9 CS-605 Spring 18 Page 3 of 8 Practice problems: 1. What result is generated by the following queries? select Emp_nbr, Emp_lname, Emp_fname from EMPLOYEE, CUSTOMER, SALES where Emp_nbr = Cust_rep and Cust_nbr = Sales_Cust_nbr and Sales_prod_nbr = 60; select Cust_name, Cust_phone, sum(sales_last_year) from CUSTOMER, SALES, PRODUCT where Cust_nbr = Sales_cust_nbr and Sales_prod_nbr = Prod_nbr and Prod_prodline between 10 and 19 group by Cust_nbr, Cust_name, Cust_phone having count(sales_last_year) > 3 order by count(sales_last_year) desc; 2. Print department name, department manager name and department phone number for all departments. 3. Display the customer number and phone of those customers who bought products from product lines 21, 22 and 23. A table can be joined with itself (self-join, sometimes necessary). Then, the FROM clause must list the same table twice, giving each table instance a separate name, so that two diff t rows in the same table could be referenced Example: Display all employees who are married to another employee, together with the name of their spouse: SELECT E1.emp_fname E1.emp_lname, E2.emp_fname E2.emp_lname FROM EMPLOYEE E1, EMPLOYEE E2, MARRIAGE M WHERE M.emp1_nbr = E1.emp_nbr and M.emp2_nbr = E2.emp_nbr;

4 Handout 9 CS-605 Spring 18 Page 4 of 8 Here s another example where aliasing is necessary: Example: Display all pairs of customers that have the same customer representative: SELECT C1.cust_name, C2.cust_name FROM CUSTOMER C1, CUSTOMER C2 WHERE C1.cust_rep = C2.cust_rep Practice problem: What would be the result if customer table was not aliased, i.e. SELECT cust_name, cust_name FROM CUSTOMER WHERE cust_rep = cust_rep; Nested Subqueries Some SELECT statements include nested SELECT subqueries in the WHERE, or HAVING clause. The result of the subquery is either 1. a number, generated by an aggregate fn. (COUNT, SUM, AVE, MIN, MAX). Example: Display the names of those employees that have the highest number of dependents. Hint: an employee has the highest number of dependents if his number of dependents equals the maximum number of dependents across the entire company. SELECT emp_fname ' ' emp_lname, emp_nbr_of_dependents FROM EMPLOYEE WHERE emp_nbr_of_dependents = (SELECT MAX(emp_nbr_of_dependents) FROM EMPLOYEE) Example: Display the names of those customers who bought the most expensive product Hint: generate the price of the most expensive products; select those customers who have sales records that refer to a product from that list. SELECT DISTINCT cust_name FROM CUSTOMER, SALES,PRODUCT WHERE cust_nbr = sales cust_nbr AND sales prod_nbr = prod_nbr AND prod_price = (SELECT MAX (prod_price) FROM PRODUCT) 2. a set of rows. Example Display the names of all customer representatives SELECT emp_fname emp_lname FROM EMPLOYEE WHERE emp_nbr IN (SELECT cust_rep FROM CUSTOMER )

5 Handout 9 CS-605 Spring 18 Page 5 of 8 Note the use of operator ALL and the subquery in the next example: Example Display the name of the customer representative(s) that has(ve) the greatest number of customers SELECT cust_rep, emp_fname emp_lname As "NAME", count(cust_nbr) As "Total customers" FROM CUSTOMER,EMPLOYEE WHERE emp_nbr = cust_rep GROUP BY cust_rep, emp_lname, emp_fname HAVING count(cust_nbr) >= ALL ( SELECT count(*) FROM CUSTOMER GROUP BY cust_rep) ; The above examples are examples in which the inner query does not refer to any column values from the outer query - these are called non-correlated subqueries. There are also correlated subqueries - those that refer to a column value from the outer query: Example: Display the names of those customers who bought the most expensive product in the product line. Hint: join the customer and sales tables with the product table and pick records with most expensive product per product line. SELECT DISTINCT cust_name FROM CUSTOMER, SALES, PRODUCT P1 WHERE cust_nbr = sales_cust_nbr AND sales_prod_nbr = P1.prod_nbr AND P1.prod_price = (SELECT MAX(P2.prod_price) FROM PRODUCT P2 WHERE P2.prod_prodline = P1.prod_prodline ); And, the same as above, but also showing the product line and the price of the most expensive item in it using a subquery in the SELECT clause SELECT DISTINCT cust_name, P1.prod_prodline, (SELECT MAX(P2.prod_price) FROM PRODUCT P2 WHERE P2.prod_prodline = P1.prod_prodline ) AS "Max Price" FROM CUSTOMER, SALES, PRODUCT P1 WHERE cust_nbr = sales_cust_nbr AND sales_prod_nbr = P1.prod_nbr AND P1.prod_price = (SELECT MAX(P2.prod_price) FROM PRODUCT P2 WHERE P2.prod_prodline = P1.prod_prodline );

6 Handout 9 CS-605 Spring 18 Page 6 of 8 Example: Display the names of those employees that have the highest number of dependents in their department. Hint: an employee has the highest number of dependents if his number of dependents equals the maximum number of dependents across other employees in his/her department. Practice problems: SELECT E1.emp_fname E1.emp_lname, E1.emp_nbr_of_dependents FROM EMPLOYEE E1 WHERE E1.emp_nbr_of_dependents = ( SELECT MAX(E2.emp_nbr_of_dependents) FROM EMPLOYEE E2 WHERE E2.emp_dept= E1.emp_dept) 1. What is produced by the following queries: select Cust_nbr, Cust_name from CUSTOMER where Cust_nbr NOT IN (select sales_cust_nbr from SALES); select Prod_nbr, Prod_name, Prodline_name from PRODUCT P1, PRODLINE where Prod_prodline = Prodline_nbr and Prod_price < 0.5* (SELECT avg(prod_price) FROM PRODUCT P2 WHERE P2.prod_prodline = P1.prod_prodline); 3. Display the names of all employees that are not department managers. 4. Display the names and phone numbers of those customers who had the smallest number of the total sales in their customer group.

7 Handout 9 CS-605 Spring 18 Page 7 of 8 Set Operations Build result sets based on two or multiple separate queries. Three operations UNION Every row from either table will be included in the answer set INTERSECT Every row that is in both tables will be included in the answer set MINUS Every row that is in the first table but not in the second table will be included in the answer set Requirements for set compatibility The same number of columns The same data types Returns result sets of each select sequentially Outer Join Outer Joins - an operation on two tables that includes rows that do not have matching values in the joined table as well as those that have. LEFT OUTER JOIN - includes non-matching rows from the first table. RIGHT OUTER JOIN - includes non-matching rows from the second table. Syntax of the outer joins varies by RDBMS (and version). The same result can always be derived using combination of UNION and NOT IN operators. Example: common join (a.k.a equi-join, a.k.a inner join ): Display customer representatives numbers, names along with the customers they serve. SELECT emp_nbr, emp_fname emp_lname, cust_nbr, cust_name FROM employee, customer WHERE emp_nbr = cust_rep Same query, but alternative syntax - note the FROM clause has the join condition SELECT emp_nbr, emp_fname emp_lname, cust_nbr, cust_name FROM employee JOIN customer ON emp_nbr = cust_rep Example: LEFT OUTER JOIN: Display all employee numbers and names along with the customers they serve, including information on non-customer representative employees as well. Notice the use of (+). SELECT emp_nbr, emp_fname ' ' emp_lname, cust_name FROM employee, customer WHERE emp_nbr = cust_rep (+)

8 Handout 9 CS-605 Spring 18 Page 8 of 8 Alternative syntax for the LEFT OUTER JOIN. Same query as above. SELECT emp_nbr, emp_fname ' ' emp_lname, cust_name FROM EMPLOYEE LEFT OUTER JOIN Customer ON emp_nbr = cust_rep Alternative way of generating the same information, uses UNION and NOT IN operators. The first SELECT-statement generates the equi-join result, second SELECTstatement generates the rows without matching values SELECT emp_nbr, emp_fname ' ' emp_lname, cust_name FROM employee, customer WHERE emp_nbr = cust_rep UNION SELECT emp_nbr, emp_fname ' ' emp_lname, NULL FROM employee WHERE emp_nbr NOT IN (SELECT cust_rep FROM customer) Practice problem: Display the total number of sales records for each product (there are products for which there is no record in the Sales table).