1. SQL: Advanced Options, Updates and Views Lecturer: Dr Pavle Mogin

2. Plan for Adv. Options, Updates and Views
Advanced options of the query language
Joined tables,
Outer join,
Aggregate functions
Grouping
Update commands:
Insert,
Delete, and
Update
SQL views
Additional features of SQL
Readings:
Chapters 4 and 5

3. Joined Tables
To avoid mixing conditional expressions and join conditions in the WHERE clause, it is possible to define join in the FROM clause
Retrieve first name, course id and corresponding grades of the student with Student Id =
SELECT FName, CourId, Grade FROM Student NATURAL JOIN Grades WHERE StudId = ;
(…FROM Student s JOIN Grades g ON
s.StudId = g.StudId … (equi join))
It can also be any kind of a theta join, or any kind of an outer join
Student Natural JOIN Grades is just one joined table. There can be more joined tables defined in the FROM clause.

4. Outer Join
Outer join has been defined to overcome problems with null values and tuples of the referenced table being not referenced
A conventional join:
SELECT * FROM R1 JOIN R2 ON R1.B = R2.B; R1 R2
R1 JOIN R2 A B C a1 ω b1 c1 a2 b2 b3

5. Left Outer Join A a left outer join:
SELECT * FROM R1 LEFT JOIN R2 ON R1.B = R2.B; R1 R2
R1 LEFT JOIN R2 A B C a1 ω b1 c1 a2 b2 b3

6. Right Outer Join A right outer join:
SELECT * FROM R1 RIGHT JOIN R2 ON R1.B = R2.B; R1 R2
R1 RIGHT JOIN R2 A B C a1 ω b1 c1 a2 b2 b3

7. Full Outer Join A full outer join:
SELECT * FROM R1 FULL JOIN R2 ON R1.B = R2.B; R1 R2
R1 RIGHT JOIN R2 A B C a1 ω b1 c1 a2 b2 b3

8. Aggregate Functions
SQL supports COUNT, SUM, AVG, MIN, MAX and some other aggregate functions that can be used in a SELECT clause
e.g.
What is the total point value of the courses in the COMP dept?
SELECT SUM(Points ) FROM Course WHERE Dept = 'Comp';
How many different first names do COMP majors have?
SELECT COUNT(DISTINCT FName ) AS NoOfNames FROM Student WHERE Major = 'Comp';
sum_Points 52
COUNT(*) returns the number of tuples in the query result,
COUNT(〈atribute_list〉) returns the number of rows having not null values in the columns of the attribute list,
while COUNT(DISTINCT〈atribute_list〉) returns the number rows having distinct values that are not null
SUM and AVERAGE can be applied to numeric attributes, only. MIN and MAX can be applied on the attributes with a total ordering. It is a question what happens if a SUM (or AVERAGE) attribute contains a null value.
NoOfNames 1

9. Grouping
GROUP BY clause is used to apply an aggregate function to groups of tuples
Groups are determined by means of a grouping attribute list and all attributes from that list have to appear in the query result
e.g.
For each student, retrieve the number of courses passed
SELECT StudId, COUNT(*) FROM Grades WHERE Grade IS NOT NULL GROUP BY StudId ;
StudId
COUNT(*)
007007 4
131313 1
555555

10. HAVING Clause
HAVING clause is used to choose from groups according to a condition specified on aggregate function values
Whereas a conditional expression in the WHERE clause filters individual tuples, the HAVING clause filters groups of tuples
e.g.
Retrieve the number of courses passed for students that passed at least two courses
SELECT StudId, COUNT(*) FROM Student s NATURAL JOIN Grades g WHERE Grades IS NOT NULL GROUP BY s.StudId HAVING COUNT(*) > 1;
Whereas selection condition in the WHERE clause chooses between tuples, condition in the HAVING clause chooses between groups of tuples
StudId
COUNT(*)
007007 4

11. Summary of SQL queries SELECT 〈attribute_and_function_list〉
FROM 〈table_list〉
[WHERE 〈condition〉]
[GROUP BY 〈grouping_attribute_list〉]
[HAVING 〈group_condition〉]
[ORDER BY 〈attribute_list〉]
In principle, we can say the following (although a query optimizer performs it differently):
The FROM clause is executed first to identify all relations involved and, to materialize any join table, this materialization is done by executing a Cartesian product of the tables listed in the FROM clause,
Then is WHERE clause evaluated for every tuple from the Cartesian product, the tuples that evaluate true are inserted into temporary query result,
Tuples from temporary query result are projected onto attributes from SELECT clause,
Then is GROUP BY applied,
Follows HAVING, and
Finally ORDER BY clause.

12. Database Update Commands
INSERT command to insert new tuples in the relation,
But, DBMS checks for duplicates first
DELETE command to find and delete tuples from relation, and
UPDATE command to find, modify and store again tuples in the relation

13. INSERT Command INSERT INTO 〈table_name〉 [〈attribute_list〉]
( VALUES (〈value_list〉) | SELECT …)
INSERT INTO Student VALUES (111111, 'Bole', 'Ann', Math);
The values in VALUES have to appear in the same order as the attributes in the corresponding CREATE TABLE command,
INSERT INTO Student (FName, LName, StudId )
VALUES('Ann', 'Bole', );
Useful when values of attributes:
Declared as NULL, or
Having DEFAULT value
are missing
Not allowed if the missing attribute is declared NOT NULL

14. INSERT Command (continued)
A form of INSERT command that is suitable for creation of materialized vies and temporary tables
CREATE TABLE StudentInfo (
StudId INT PRIMARY KEY,
LName CHAR(15) NOT NULL,
NoOfCourses INT);
INSERT INTO StudentInfo
SELECT s.StudId, LName, COUNT(*) AS NoOfCourses
FROM Student s, Grades g
WHERE s.StudId = g.StudId
GROUP BY StudId, LName ;
The table we wont
to populate by data
If a query is to complex, it is advisable to implement it step by step by creating and populating a series of tables with intermediate results.
If DBMS doesn't support nested queries, the above approach is the only available to simulate nested queries.

15. DELETE Command DELETE FROM 〈table_name〉 [WHERE 〈condition〉]
DELETE FROM Student
WHERE FName = 'Susan';
WHERE StudId IN (SELECT s.StudId FROM Student s, Grades g WHERE s.StudId = g.StudId AND CourId = 'C302');
DELETE FROM Student ;
First DELETE command deletes all Student tuples where the FName is 'Susan'.
Second DELETE command deletes all students that enrolled or passed Comp302
The third DELETE command deletes all Student tuples, but lives the table itself in the Catalog.

16. UPDATE Command UPDATE〈table_name〉
SET 〈attribute_name〉=〈value_expression〉 {, 〈attribute_name〉=〈value_expression〉}
[WHERE 〈condition〉]
e.g.
UPDATE Grades
SET Grade = 'A+'
WHERE CourId = 'C302';
Again, it is possible to update a set of tuples by one command, if all the tuples should be modified by the same value
Again, the condition of the WHERE clause can be simple, or may contain nested query.
If the WHERE clause is omitted, all the relation tuples are updated

17. Views in SQL
A SQL view is a virtual table that is derived from other base or virtual tables
Base tables are defined by CREATE TABLE command and are permanently stored in a database
Virtual tables are defined by the CREATE VIEW command to avoid defining complex SQL retrieval expressions repeatedly
The definition of a view is stored in the Catalog, but its data are not stored in the database itself, so it is computed every time it is used in a query
Contrast the up to date status of a temporary table and the corresponding view

18. SQL Views
Assume the course table contains the attribute Hours (number of lecture and tutorial hours per week)
A view definition
CREATE VIEW StudOccupied AS
SELECT g.StudId, SUM(Hours ) AS Occupied
FROM Grades g, Course p
WHERE g.CourId = p.CourId AND Grade IS NULL
GROUP BY StudId ;
Deleting a view
DROP VIEW StudOccupied ;
It is supposed that Grades relation contains null values in Grade column for student – course pairs if the student currently attends classes for courses, and accordingly hasn't any grade

19. Additional Features of SQL
Assertions as general constraints (CREATE ASSERTION – a DDL command that may use DML SELECT command)
Triggers as procedures stored with tables
GRANT and REVOKE commands to deal with database user privileges
Embedded SQL and CURSOR
SQL transaction control commands (COMMIT, ROLLBACK)
User Defined Functions (UDF):
SQL Functions
Procedural Language (C, PL/pgSQL, Java) Functions
Assertions as general constraints
Complex database constraints are declared by means of CREATE ASSERTION statement
Triggers as procedures stored with tables
Trigger contains specification of an event and corresponding action
Event can be constraint violation or a pre specified database state
Action is either to prevent constraint violation, or to inform user about the database state
The concept of the trigger is crucial for active databases
GRANT and REVOKE commands to deal with database user privileges
Embedded SQL and CURSOR
SQL statements are embedded into "host language" program, so that all the traffic between program and the database is accomplished via SQL
Cursor is the result of a SQL query that is considered by the host program as a file with sequential access mode

20. Summary The relational database language has commands to define:
Database structure (schema, domain, table, and constraints) (CREATE SCHEMA, CREATE DOMAIN, CREATE TABLE)
Queries (SELECT… FROM… WHERE… GROUP BY…HAVING… ORDER BY…)
Update operations (INSERT, DELETE, UPDATE)
views (CREATE VIEW)
Additional features (ASSERTION, TRIGGER, CURSOR, GRANT, REVOKE, COMMIT, ROLLBACK, DEFINE FUNCTION)
Although SQL is defined by a standard, implementations have some dialects and exceptions

21. Next Lecture
Relational Algebra