Query Languages: How to build or interrogate a relational database

1

Query Languages: How to build or interrogate a relational database

Structured Query Language (SQL)

2

SQL

SQL is a query language for relational databases. Contains:

Data Definition Language to define databases Data Manipulation Language to manipulate

databases. SQL is widely accepted and is used by most

relational DBMSs. Is being standardized.

3

The importance of SQL

Since SQL is used in almost all relational databases, once you know SQL you can probably construct and manipulate databases in all RDBMs.

Knowing SQL makes you a (beginning) ORACLE, Informix, SyBase, AdaBas, and so on programmer!

4

Functionalities of SQL

SQL provides On-line and embedded use. Precompilation of embedded queries. Dynamic database definition and alteration. Maintenance of indexes View mechanism Authorization mechanism Automatic concurrency control Logging and database recovery Report formatting

5

Tables in SQL

SQL recognizes Base Tables

real tables that physically exist in the database. There are physically stored records and possibly physically stored indexes directly corresponding to the table

Views virtual tables that do not physically exist but look to the

user as if they do

6

Data Definition

An SQL database consist of Database Spaces Base tables Indexes Views

7

Database Spaces

DBSpace is a section of physical disk. It consists of

Base tables Indices Views

All can be dynamically dropped from DBSpaces. DBSpaces allow the DB administrator to distribute

data accesses over different disks.

8

Indexes

As we know, indexes can improve search performance.

Cost: more space needed and slower insertion. Indexes can be defined over any combination of

attributes in a base table. Automatically maintained in SQL. Users never directly use an index.

9

Views

Correspond to external schemas. Derived from one or more base tables or views. Computed dynamically.

10

Operations in SQL

For tables: CREATE, ALTER, DROP

For indexes CREATE, DROP

For views: CREATE, DROP

11

Creating tables

CREATE <table name>

(<coldecl> [,<coldecl>*], [, <pkdef> [, <fkdef>*]);

<coldecl> :=

<col><type>[NOT NULL]

<type> :=

integer|smallint|float(p)|

decimal(p,q)|char(n)|

varchar(n)|long varchar|

date|time

12

Creating tables continued

<pkdef> :=

PRIMARY KEY (<colname>

[,<colname>*]

<fkdef> :=

FOREIGN KEY (<colname>[,<colname>*])

REFERENCES <table>

[ ON DELETE <effect>]

13

More on creating tables

<effect> :=

RESTRICT | CASCADE | SET NULL

What happens when the tuple in the referenced table with that value is deleted RESTRICT: Do not delete as long as there tuples

in other table with that foreign key value CASCADE: Delete all tuples with that foreign key

value SET NULL: Set value of foreign key to NULL.

(Note violates referential integrity).

14

Example 1

CREATE TABLE Student

(sid CHAR(5) NOT NULL,

sname VARCHAR(20),

address VARCHAR(70),

PRIMARY KEY (sid));

OR

CREATE TABLE Student

(sid CHAR(5) PRIMARY KEY,

sname VARCHAR(20),

address VARCHAR(70));

15

Example 2

CREATE TABLE Enrol

(sid CHAR(5) NOT NULL,

cid CHAR(5) NOT NULL,

grade INT,

PRIMARY KEY(sid, cid),

FOREIGN KEY (sid)

REFERENCES Student

ON DELETE CASCADE

FOREIGN KEY (cid)

REFERENCES Course

ON DELETE RESTRICT);

16

Altering tables I

ALTER TABLE <table name>

ADD {<coldecl>|

<pkdef>|

<fkdef>};

ALTER TABLE Enrol

ADD edate DATE;

adds a new column to the table grade. For existing tuples, the value is set to NULL.

17

Altering tables II

ALTER TABLE <table name>

DROP {PRIMARY KEY|

<fkname>};

Note that care must be taken when dropping columns.

18

Dropping tables

Tables can be dropped at any time. Dropping a table deletes both the definition and

data. Also, all views, indexes and foreign key

definitions referring to this table are dropped.

DROP TABLE <table name>;

19

Creating indexes

CREATE [UNIQUE] INDEX

<index> ON <table>

(<colname> [<order>]

[,<colname> [<order>]*]);

<order>:= ASC | DESC

Creates an index on named columns. With UNIQUE, no two tuples can have the same values for the indexes columns.

Example:CREATE INDEX Course

ON Enrol (cid);

20

Data manipulation

Having created the tables, indexes and views, we now need to populate the database and retrieve information from it.

In other words, we want to manipulate the data.

21

Retrieval

SELECT [DISTINCT] <items>

FROM <table> [, <table>*]

[WHERE <pred>]

[GROUP BY <attrs>

[HAVING <pred>]]

[ORDER BY <attrs> ];

Corresponds to a JOIN-SELECT-PROJECT expression in relational algebra.

22

Predicates

The predicate <pred> is a condition formed by parentheses and boolean operators AND, OR and NOT.

A condition has the form <attr><op>{<value>|<attr>}

and an operator is one of < | =< | > | >= | = | !=

23

WHERE clauses

In general, WHERE clauses are constructed as in relational algebra, but with some additions

LIKE string May contain wildcard characters %, which matches

any string, and _, which matches a single character.

IN (set of values) Tests for set membership

BETWEEN c1 AND c2

24

Example

Find Student IDs and grades for those students who read CS51T

SELECT sid, grade

FROM Enrol

WHERE cid = ‘CS51T’;

Compare

sid, grade ( cid = ‘CS51T’(Enrol))

25

Example continued

We can embellish the way in which the result appears by including format strings in the SELECT

Example

SELECT Student as sid, grade

FROM Enrol

WHERE cid = ‘CS51T’;

26

DISTINCT

DISTINCT is used to make sure that we do not get any duplicate values.

ExampleSELECT DISTINCT cid

FROM Enrol

WHERE grade > 70;

First, find the various course numbers that qualify and then remove duplicates.

27

More examples

The use of * in the SELECT returns all attributesSELECT *

FROM Enrol

WHERE cid = ‘CS51T’;

Find all students who obtained 60 or more for CS51TSELECT sid

FROM Enrol

WHERE cid = ‘CS51T’

AND grade >= 60;

28

Yet more examples

Find all results for either or CS51T or CS51SSELECT *

FROM Enrol

WHERE cid IN

(‘CS51S’, ‘CS51T’);

Find results for CS coursesSELECT *

FROM Enrol

WHERE cid LIKE ‘CS%’;

29

Ordering results

Get all results for CS51S and CS51T but order them by result

SELECT sid, cid, grade

FROM Enrol

WHERE cid IN

(‘CS51S’, ‘CS51T’)

ORDER BY grade DESC;

30

Subqueries

Notice that the result of a SELECT clause is a table which can be used in another WHERE clause.

Find course titles of the courses for which 123 was registeredSELECT title

FROM Course

WHERE cid IN

(SELECT cid FROM Enrol

WHERE sid = ‘123’);

31

Table labels

Sometimes we need to interrogate the same table twice.

We use table labels Example: Get IDs from those students who did both

CS51S and CS51T

SELECT DISTINCT sid

FROM Enrol as E1, Enrol as E2

WHERE E1.Sid = E2.Sid

AND E1.Cid = ‘CS51S’

AND E2.Cid = ‘CS51T’;

32

Table labels can usually be avoided

We could formulate the same query as

SELECT sid

FROM Enrol

WHERE cid = ‘CS51S’

AND sid IN

(SELECT sid

FROM Enrol WHERE cid = ‘CS51T’);

33

Use of ALL in WHERE clauses

Queries that look at all tuples satisfying a particular predicate.

Get the IDs of the students all of whose results are over 70.SELECT sidFROM Enrol as E1WHERE 70 < ALL (SELECT grade FROM Enrol as E2 WHERE E1.sid = E2.sid);

Forms of ALL:< ALL, <= ALL, = ALL, >= ALL, > ALL

34

Union

Union allows one to union tuples from different tables.

Get Student IDs for all students whose name starts with a ‘J’ or who obtained an A for CS51T.

SELECT sid FROM Student

WHERE sname LIKE ‘J%’

UNION

SELECT sid FROM Enrol

WHERE cid = ‘CS51T’

AND grade > 70;

35

Intersect

Allows one to intersect Get all IDs for students whose name begins with

a ‘J’ and who obtained an A for CS51S

SELECT sid FROM Student

WHERE sname LIKE ‘J%’

UNION

SELECT sid FROM Enrol

WHERE cid = ‘CS51S’

AND grade > 70;

36

EXISTS and NOT EXISTS

Counterpart of ALL Find name of students who have not obtained an

A for any course

SELECT sname FROM Student

WHERE NOT EXISTS

(SELECT * FROM Enrol

WHERE sid = Student.sid

AND grade > 70);

37

Analysis of data

In order to help do some primitive analysis of data, SQL has some built-in functions COUNT(*) COUNT(DISTINCT <attr>) SUM([DISTINCT]<item>)

where <item> may be an abstraction and does not need to be a single attribute.

AVG([DISTINCT]<item>) MAX(<item>) MIN(<item>)

38

Some simple examples of data analysis in SQL How many students are registered for at least one

courseSELECT COUNT(DISTINCT sid)

FROM Enrol;

Find the average grade for CS51SSELECT AVG(grade)

FROM Enrol

WHERE cid = ‘CS51S’;

39

Another example

How many students were above the average for CS51T?

SELECT COUNT(*)

FROM Enrol

WHERE grade >

(SELECT AVG(grade)

FROM Enrol

WHERE cid = ‘CS51T’);

40

Yet another example

What is the name of the student who got the best mark for CS51T?

SELECT sname

FROM Student

WHERE sid IN

(SELECT sid

FROM Enrol

WHERE grade =

(SELECT MAX(grade)

FROM Enrol

WHERE cid = ‘CS51T’));

41

GROUP BY

A relation can be partitioned into groups according to some value. Analysis can then be done on these groups.

What are the averages for the various courses?

SELECT cid, AVG(grade)

FROM Enrol

GROUP BY cid;

42

HAVING

After partitioning, we can disqualify groups. What is average results for courses with

enrollment of more than 10? SELECT cid, AVG(grade) FROM Enrol

GROUP BY cid

HAVING COUNT(*) > 10;

COUNT is applied to each group separately.

43

Insertion

INSERT INTO {<table>|<view>}

[(<attr>] [,<attr>*])]

{VALUES (<items>|

<select statement>)};

Example

INSERT INTO Enrol

(cid, sid, grade)

VALUES (‘CS51T’, ‘123’, 67);

44

Insertion through a SELECT statement For each course, get the average and insert into a

RES table

INSERT INTO RES (cid, average)

SELECT cid, AVG(grade)

FROM Enrol

GROUP BY cid;

45

Deletion

DELETE FROM <table>

[<WHERE clause>];

Example DELETE FROM Enrol

WHERE cid = ‘CS51T’;

Difference between DELETE and DROP DELETE FROM Enrol;

DELETE empties the table but leaves the table and indexes.

46

Updating tables

UPDATE <table>

SET <attr> = <expr>

[, <attr> = <expr>*]

[<WHERE CLAUSE>];

Example: Give everybody 10 extra marks for CS35A

UPDATE Enrol

SET grade = grade + 10

WHERE cid = ‘CS51T’;

47

Views

Views are derived tables whose definition is stored and whose content is computed.

Can be used as base table for retrieval and view definition.

Exact condition for updating an open problem. Currently only update iff

derived form single base table and, has rows and attributes corresponding to a

unique and distinct row in base table.

48

Advantages of views

Views are SQL’s external schemas. They are useful Users are immune to database growth Users are immune to database restructuring

(logical data independence) Simplified user perception Different views of same data for different users Automatic security for hidden data.

49

Creation and deletion of views

CREATE VIEW <view>

[(<colname>[,<colname>*])]

AS select-statement;

ExampleCREATE VIEW CourseAvg

(Cid,Average)

AS SELECT cid, Avg(grade)

FROM Enrol

GROUP BY cid;

DeletionDELETE VIEW <view>;

DELETE VIEW CourseAvg;

50

The view update problem

The view CourseAvg as defined above cannot be updated, as any updates cannot be translated into the base table.

The DB administrator should decide whether a view is updatable.

51

Authorization Mechanism

The authorization mechanism allows one to give other users permission to access and update data in a view or table.

The owner must explicitly grant necessary privileges to others, as by default the owner has all privileges and others have none.

52

GRANT and REVOKE

GRANT <privilege>

ON <table> | <view>

TO <user> [,<user>*] |

PUBLIC

[WITH GRANT OPTION];

REVOKE <privilege>

ON <table> | <view>

FROM <user> [,<user>*] |

PUBLIC;

53

Grantable privileges

These privileges are allowed: SELECT INSERT UPDATE DELETE ALTER INDEX

permission to create or drop indexes on a table. ALL

54

Use of SQL

Most users, other than the database manager, will not directly interact with SQL.

Typically, one sets up some graphical interfaces (forms) for user interaction. However, underlying the forms are SQL queries.