Download sql - structured query language

Sheffield Hallam University
SCHOOL OF COMPUTING AND MANAGEMENT SCIENCES
___________________________________________________________________
SQL - STRUCTURED QUERY LANGUAGE
WORKBOOK
NOTE: This workbook is based upon the one written for teaching Oracle SQL to
undergraduate students of CMS. Special thanks to the original authors for allowing
me to use their work as the basis for this document.
The target audience for this version is students on postgraduate courses at CMS,
but who have not had SQL experience, or are very rusty. It is a cut down version of
the original and is intended only to give some early exposure to the language at the
heart of the database elements of the course.
The other significant difference is that all examples and questions have been written
such that they will run in a variety of environments. Wherever possible standard ISO
SQL is used. Where there can be differences, examples are given in both Oracle
and MySQL format.
Topic
Section
Introduction and selecting data from tables
A
Joining tables
B
Simple functions and grouping data
C
Subqueries (Oracle users only)
D
Creating and populating tables and views
E
Managing the database, access and security
F
Section A
INTRODUCTION
The philosophy of Relational Database Systems is that they are easy to understand
and use. The standard end-user language for controlling and accessing the
database is also simple to learn and use. However, on its own it has limited
functionality and sophistication, and leaves much to be desired in terms of enforcing
data integrity and can encourage very inefficient use of system resources.
Traditionally, a database language comprises 3 different components:
DDL - Data Definition/Description Language –
used to define and alter the database structure and format
DML - Data Manipulation Language –
used to manipulate the data values in the database
DCL - Data Control Language –
used to control user access and processing of the database
In Relational Systems, all three components are part of the same language – SQL
(Structured Query Language). SQL provides for database definition, population,
selection, alteration, structural change and security.
Example statements are:
DDL - CREATE, DROP, ALTER
DML - SELECT, INSERT, UPDATE, DELETE
DCL - GRANT, REVOKE
(In addition, there are a set of SQL 'commands' - e.g. to set the run-time
environment, interrogate the data dictionary etc...).
28/04/17
2
SQL primer workbook 1999/2000
Section A
The DATA DICTIONARY
Relational databases will have a data dictionary that holds details of the database
definition and its environment. This dictionary is implemented using differently with
different RDBMS.
Some useful commands are:-
e.g.
DESCRIBE table_name
DESCRIBE EMP
used to display a table definition
The command can be abbreviated to DESC
 Oracle sees this is an SQL ‘command’ and hence does not need a ; terminator.
 MySQL, however, does require a closing ;
In Oracle you can view the tables you currently own by:
SELECT * FROM TAB;
In MySQL its:
SHOW TABLES;
RUNNING SQL(Oracle)
The easiest way to run SQL commands, all bar the simplest, is to create text files of
SQL command(s) with an editor.
If these text files are given an extension of sql (in lower case) e.g. myfile.sql, you
can execute the file by typing @ filename or RUN filename at the sql prompt. You
can then use the text editor to correct or modify the statements if necessary, re-save
the file and execute as required.
RUNNING SQL(MySQL)
Similarly, one can type commands directly in a the mysql> prompt, or save them for
batch mode running in a text file. You run these text files from the command shell
using the greater than sign:
C:\mysql\bin>
mysql <somecode.sql
This will carry out the SQL command contained in the file somecode.sql if it is also
located in the bin folder.
28/04/17
3
SQL primer workbook 1999/2000
Section A
COMMENTING SQL CODE
Comments can be included to clarify and explain an sqlplus statement or procedure.
The comment starts with /* and ends with */ e.g.
SELECT NAME FROM CUST
ORDER BY NAME
/* this is a piece of simple sql */
/* this sorts the output */ ;
 These comments work for both Oracle and MySQL.
Throughout this workbook the exercises are based on a personnel system and the
contents of the tables used are given on the following page.
The examples in the text are based on the three tables below, a simple accounts
system.
The tables represent the fact that a customer may have many accounts, and that an
account may be held jointly by more than one customer.
CUST
Owns
CUSTACC
Allocated
ACC
CUST
REFNO
NAME
ADDRESS
AREA
A123
J Doe
1 High Street
Sheffield
A124
J Smith
2 West Street
Sheffield
B127
R Best
4 East Row
Rotherham
B128
J Best
4 East Row
Rotherham
C371
R Done
23 Middle Avenue
Barnsley
CUSTACC
REFNO
ACCNO
A123
1245890
A123
1494315
B127
5418490
B128
5418490
ACC
ACCNO
BALANCE
BRANCH
1245890
234.50
1494315
0.50
5418490
1789.40
28/04/17
Broomhill
Tinsley
Broomhill
4
SQL primer workbook 1999/2000
Section A
THE SAMPLE TABLES
EMP
EMPNO
ENAME
JOB
MGR
HIREDATE
SAL
7369
7499
7521
7566
7654
7698
7782
7788
7839
7844
7876
7900
7902
7934
SMITH
ALLEN
WARD
JONES
MARTIN
BLAKE
CLARK
SCOTT
KING
TURNER
ADAMS
JAMES
FORD
MILLER
CLERK
SALESMAN
SALESMAN
MANAGER
SALESMAN
MANAGER
MANAGER
ANALYST
PRESIDENT
SALESMAN
CLERK
CLERK
ANALYST
CLERK
7902
7698
7698
7839
7698
7839
7839
7566
17-DEC-80
20-FEB-81
22-FEB-81
02-APR-81
28-SEP-81
01-MAY-81
09-JUN-81
09-DEC-82
17-NOV-81
08-SEP-81
12-JAN-83
03-DEC-81
03-DEC-81
23-JAN-82
800.00
1600.00
1250.00
2975.00
1200.00
2850.00
2450.00
3000.00
5000.00
1500.00
1100.00
950.00
3000.00
1300.00
7698
7788
7698
7566
7782
COMM
300.00
500.00
1250.00
0.00
DEPTNO
20
30
30
20
30
30
10
20
30
20
30
20
10
DEPT
DEPTNO
10
20
30
40
DNAME
ACCOUNTING
RESEARCH
SALES
OPERATIONS
LOC
NEW YORK
DALLAS
CHICAGO
BOSTON
SALGRADE
28/04/17
GRADE
LOSAL
HISAL
1
2
3
4
5
700.00
1201.00
1401.00
2001.00
3001.00
1200.00
1400.00
2000.00
3000.00
9999.00
5
SQL primer workbook 1999/2000
13.3.1
Section A
The SELECT statement
Has the form :
SELECT
Specification of desired columns
FROM
Specification of table or tables
WHERE
Selection criteria for rows
GROUP BY
Formation of groups with identical values in specified column
HAVING
Selection of specific groups
ORDER BY
Sorting sequence of rows of result table
The first two arguments SELECT and FROM are mandatory.
Selecting data from a single table:SELECT * FROM table-name;
e.g.
displays all columns
SELECT * FROM CUST;
It is also possible to manipulate the column values before displaying them:SELECT
FROM
e.g.
column, column2*2.5, column3+column4 alias1, etc..
table-name;
displays specified columns, performs arithmetic on columns and
specifies an alias to use as a column heading on output
SELECT ACCNO, BALANCE*0.1 BONUS FROM ACC;
will produce the following:ACCNO
BONUS
1245890
1494315
5418490
28/04/17
23.45
0.05
178.94
6
SQL primer workbook 1999/2000
13.3.1
Section A
EXERCISES
A1
Display the contents of the DEPT, EMP and SALGRADE tables.
[Check your results against the tables listed.]
Select * from Emp;
Select
Select
To produce an answer to the request "Give a list of all employees showing
their employee number, name & department number." will require the
following SQL statement,
Select EmpNo, Ename, DeptNo from Emp;
which will produce the results table shown below.
EMPNO
7369
7499
7521
7566
7654
7698
7782
7788
7839
7844
7876
7900
7902
7934
ENAME
SMITH
ALLEN
WARD
JONES
MARTIN
BLAKE
CLARK
SCOTT
KING
TURNER
ADAMS
JAMES
FORD
MILLER
DEPTNO
20
30
30
20
30
30
10
20
30
20
30
20
10
For all the following exercises record the statement that will give
you the required output. Check against the given tables to ensure
that the correct rows are selected and save the statements for
future reference.
(For each section the output from some of the queries is reproduced for you, but
after that you are on your own!)
A2
28/04/17
Display the name and commission of all the employees.
7
SQL primer workbook 1999/2000
13.3.1
Section A
Select
ENAME
SMITH
ALLEN
WARD
JONES
MARTIN
BLAKE
CLARK
SCOTT
KING
TURNER
ADAMS
JAMES
FORD
MILLER
A3
COMM
300.00
500.00
1250.00
0.00
Display the name and commission of all the employees together with another
column that shows their commission increased by 10%.
Select
ENAME
SMITH
ALLEN
WARD
JONES
MARTIN
BLAKE
CLARK
SCOTT
KING
TURNER
ADAMS
JAMES
FORD
MILLER
A4
28/04/17
COMM
NEW_COMM
300.00
500.00
330.00
550.00
1250.00
1375.00
0.00
0.00
Display the job title of all the employees.
8
SQL primer workbook 1999/2000
13.3.1
Section A
Select
JOB
CLERK
SALESMAN
SALESMAN
MANAGER
SALESMAN
MANAGER
MANAGER
ANALYST
PRESIDENT
SALESMAN
CLERK
CLERK
ANALYST
CLERK
To remove duplicates within the result table, use:
SELECT DISTINCT column1, column2, etc
A5
FROM table-name;
Display all the different job titles which currently exist in the company.
Select
JOB
ANALYST
CLERK
MANAGER
PRESIDENT
SALESMAN
28/04/17
9
SQL primer workbook 1999/2000
13.3.1
Section A
SELECT from a single table using the WHERE condition
WHERE
=, <>, >, >=, <, <=, (!= means not equal) comparison operators
NOT, AND, OR
logical operators
WHERE
column BETWEEN field value AND field value
WHERE
column IN (field value, field value, ..., ...)
WHERE
column LIKE '%charstring%'
(% means skip n characters; use _ for a single character)
WHERE
column IS NULL
(All these operators can be negated by using NOT)
A6
Display the employee number, name and current job of all those who work in
Department 30.
Select
From
Where
A7
EMPNO
ENAME
JOB
7499
7521
7654
7698
7844
7900
ALLEN
WARD
MARTIN
BLAKE
TURNER
JAMES
SALESMAN
SALESMAN
SALESMAN
MANAGER
SALESMAN
CLERK
Display the names of all the clerks, showing their employee number and that
of their manager. (String matching is case sensitive.)
Select
A8
Display details of all clerks, analysts and salesmen.
Select
28/04/17
10
SQL primer workbook 1999/2000
13.3.1
A9
Section A
Display details of employees who are not clerks, analysts or salesmen.
Select
A10
Display details of all employees whose commission is greater than salary.
Select
A11
Display employee name, job and department number for employees whose
names begin with ‘M’.
Select
A12
Display details of employees whose salaries are not between £1,200 and
£1,400.
Select
A13
Display details of salesmen and managers in dept 30, whose salary is greater
than or equal to £1,500.
(Note: with logical operators - AND precedes OR)
Select
28/04/17
11
SQL primer workbook 1999/2000
Using the ORDER BY statement to control the display order of selected records
ORDER BY column1, column2, etc ASC/DESC;
ORDER BY n;
(being the nth column)
or
Note: The default ordering is ascending.
Null values are always displayed first regardless of sort sequence.
A14
Display the employee number, current job and salary of all those who work in
Department 30, with the output in ascending salary order.
Select
From
Where
Order By
EMPNO
7900
7521
7654
7844
7499
7698
A15
JOB
SAL
CLERK
SALESMAN
SALESMAN
SALESMAN
SALESMAN
MANAGER
950.00
1250.00
1250.00
1500.00
1600.00
2850.00
Display the employee name and current job of all those who work in
Department 30, with the output in descending salary order.
Select
A16
Display the employee name and current job of all those who work in
Department 30, with the output in descending salary order within each job.
Select
28/04/17
12
SQL primer workbook 1999/2000
A17
Display employee details for departments 10 and 30, in name order, within
each department
Select
28/04/17
13
SQL primer workbook 1999/2000
13.3.3
Section C
JOINING TABLES
To SELECT data that is held in more than one table the tables need to be “joined”
together. The important thing here is the WHERE clause which specifies how the
tables are to be joined.
SELECT
FROM
WHERE
AND...
select-list
table_name1, table_name2, , ,
join-condition
Example 1: SELECT
FROM
WHERE
ACCNO, NAME
CUST, CUSTACC
CUST.REFNO = CUSTACC.REFNO;
This will list every account number and the name of the holder(s)
ACCNO
1245890
1494315
5418490
5418490
Example 2: SELECT
FROM
WHERE
AND
NAME
J Doe
J Doe
R Best
J Best
C.REFNO, NAME, A.ACCNO, BALANCE
ACC A, CUSTACC B, CUST C
A.ACCNO = B.ACCNO
B.REFNO = C.REFNO;
This is similar to the first example but also displays the account balance:REFNO
A123
A123
B127
B128
NAME
J Doe
J Doe
R Best
J Best
C.ACCNO BALANCE
1245890
234.50
1494315
0.50
5418490
1789.40
5418490
1789.40
Note, in the second example A, B and C are temporary labels, called aliases, for
the tables and are a useful shorthand in this example.
If any column name occurs in more than one table, and that column is used
anywhere in the query then it must be identified by specifying the table name as a
prefix. If this is not done then an error message indicating that the column name is
ambiguous will be returned.
28/04/17
14
SQL primer workbook 1999/2000
13.3.3
Section C
Simple Joins (also called an equi-join) are the most common form and return rows
from the two tables based upon the equality condition in the WHERE clause (see
the examples above)
WHERE table1.column_a = table2.column_b
(although we could equally well replace the = with a < or >)
B1
Predict what results will be obtained by joining the EMP and DEPT tables
without specifying any selection criteria, then try it out. How many rows does
the result produce?
Select
B2
Repeat B1 but this time joining EMP to itself. Look carefully at the output to
make sure you understand it. (Ensure that you do have results table and not
an error message!!
Select
B3
Find the name and salary of employees in Dallas.
Select
From
Where
And
ENAME
SMITH
JONES
SCOTT
ADAMS
FORD
28/04/17
SAL
800.00
2975.00
3000.00
1100.00
3000.00
15
SQL primer workbook 1999/2000
13.3.3
B4
Section C
Produce a list that will show the salary grade each employee is on. Display
the name, job, salary and grade with the output in ascending order of salary,
with those on the same salary ordered alphabetically.
Select
From
Where
Order By
B5
Display employee name, job, salary and department name for those on grade 3.
Select
B6
Display details of employees in ACCOUNTING. (Don’t answer this by finding
who is in department number 10.)
Select
B7
Show the name, employee number and manager's name of those who are
managed by either Blake or Jones. Do not skip this question.
Select
28/04/17
16
SQL primer workbook 1999/2000
13.3.3
B8
Section C
Display the employee name, location and department name of those whose
salary is greater than £1500.
Select
How many rows have you selected?
Outer Joins
Sometimes you want to return all elements of one side of a join, even if there is no
entry in the other side. For this we need an outer join. An outer join uses a (+) on
the side of the operator where we want to have nulls returned when there is no
value match.
Consider the Dept table. If you wanted to list all the departments, and list the name
of employees who worked in them you could do:
SELECT D.DName, E.Ename
FROM EMP E, DEPT D
WHERE D.DeptNo = E.DeptNo ;
But this isn’t a complete list of departments. To get that, using an outer join, one
would use the (+) notation thus:
SELECT D.DName, E.Ename
FROM EMP E, DEPT D
WHERE D.DeptNo = E.DeptNo (+) ;
Try both and see for yourself. When you come to answering the questions in
section C you will need to remember this.
28/04/17
17
SQL primer workbook 1999/2000
13.3.3
Section C
SIMPLE FUNCTIONS.
SQL provides some simple aggregating functions that enable us to derive
information about the rows in a table.
COUNT(*)
returns a single value, the number of rows in the table
MAX(attribute)
returns the largest value for the attribute
MIN(attribute)
returns the smallest value for the attribute
AVG(attribute)
works out the average value for this attribute
SUM(attribute)
works out the total value for this attribute
SELECT COUNT(*) FROM CUST;
returns a single value, the number of rows in the customer table
SELECT MIN(BALANCE) FROM ACC;
displays the lowest value held in the balance field from all the rows in the
accounts table
Any of these can be in an SQL statement that includes constraints
e.g.
SELECT
FROM
WHERE
COUNT(*)
CUST
AREA = ‘Sheffield';
tells us how many customers are based in Sheffield.
For all the following exercises you should check your results carefully against
the table listings in section A.
C1
What are the lowest and highest basic salaries within the company?
Select
Minimum_Sal
800
28/04/17
Maximum_Sal
5000
18
SQL primer workbook 1999/2000
13.3.3
C2
Section C
How many people have a salary greater than £2000?
Select
From
Where
How many people are there in ACCOUNTING? Don’t work out how many
people there are in department 10!
C3
Select
NULL Values
If a column has no value for any particular row, it is referred to as a NULL value. This
is not the same as a zero in a numeric column, or a space in a character column.
The NVL function can be used to convert a null into a specified value e.g.
NVL(COMM, 0) will return the actual value of COMM if it has a value, or a zero if
COMM is null. This is often necessary when performing calculations, or formatting
for output since null values will be totally ignored by many functions and operations.
 MySQL has a function called COALESCE which allows you do a similar thing. It
actually returns the first non-null value in a list, but used as below enables a null
to be treated as a specific number:
Select ename COALESCE(comm,0) from emp;
A column can be tested using IS NULL or IS NOT NULL e.g.
SELECT ...FROM... WHERE column_name IS NOT NULL;
28/04/17
19
SQL primer workbook 1999/2000
13.3.3
C4
Section C
What are the highest and lowest incomes (i.e. to include commission) in the
Sales Department?
Select
28/04/17
20
SQL primer workbook 1999/2000
GROUPING DATA
The GROUP BY clause splits the table into specified groups, returning one summary
row for each group which is then used in the SELECT clause.
The following will list the sum of the account balances for each customer (identified
by the reference number.)
SELECT
FROM
WHERE
GROUP BY
REFNO
A123
B127
B128
REFNO, SUM(BALANCE) TOTAL
CUSTACC C, ACC A
C.ACCNO = A.ACCNO
REFNO;
TOTAL
235.00
1789.40
1789.40
You can only specify fields in the SELECT clause if they are specified in one of the
grouping clauses or are part of an aggregating function.
What will be the result of the following query?
SELECT
FROM
GROUP
AREA, COUNT(*)
CUST
BY AREA;
In each of the following queries, you may treat the President as not an employee (as
he isn’t given a deptno). However, here’s a question for the sharper ones: how your
queries change if you did need to include him as an employee?
C5
How many people are there in each department?
Select
From
Where
Group By
C6
How many people are there in each type of job in each department?
Select
28/04/17
21
SQL primer workbook 1999/2000
C7
For each department, find the average salary and the total salary bill
excluding commission.
Select
C8
Find the maximum commission earned, and the number of people in each
department.
Select
THE HAVING CLAUSE.
The HAVING clause is used to specify which GROUPS are to be displayed.
WHERE restricts which rows a SELECT works on and HAVING restricts which
groups.
Example:
28/04/17
SELECT
FROM
WHERE
AND
GROUP BY
HAVING
AREA, SUM(BALANCE) TOTAL
CUST, CUSTACC, ACC
CUST.REFNO = CUSTACC.REFNO
CUSTACC.ACCNO = ACC.ACCNO
AREA
SUM(BALANCE) > 1000;
22
SQL primer workbook 1999/2000
C9
Display the department name and number of employees in departments with
fewer than 6 employees.
Select
From
Where
Group By
Having
C10
Display the names of all the people in a management position, their
department name and the number of staff for whom they have direct
responsibility. (This is not simply asking for departmental managers.)
Select
28/04/17
23
SQL primer workbook 1999/2000
NESTED SUBQUERIES
 Please note: MySQL (version 3.23.nn) does not yet support nested subqueries. If
you are using MySQL you will just need to use your imagination for this section!
If we take an example such as to find out which account has the largest balance,
thinking about this question logically, it falls into 2 parts.
Firstly we need to find out what the largest balance is; secondly we need to find
which account this is against.
SQL supports this strategy for querying the database through the use of subqueries.
The subquery is processed first; the result of the subquery is then substituted into the
WHERE clause of the outer query.
SELECT
FROM
WHERE
ACCNO, BALANCE
ACC
BALANCE = (SELECT MAX(BALANCE) FROM ACC);
The inner statement will return a single value, in this example £1,789.40. That value
is then tested against the value in the balance field of each of the rows in the ACC
table. Every row with that value will be returned in the results table:ACCNO
5418490
BALANCE
1,789.40
Care needs to be taken with the WHERE clause. In the example above, the nested
SELECT query returns a single value and so the ‘WHERE BALANCE =’ is valid.
If the inner SELECT can return several rows then the condition would need to take
this into account e.g. WHERE column IN (SELECT…..).
You can also use the construct WHERE column > ANY {or > ALL};
In the logic of SQL, WHERE
is processed before
GROUP BY before
HAVING.
D1
Who has the highest basic salary?
Select
From
Where
= (Select
From
28/04/17
);
24
SQL primer workbook 1999/2000
D2
Which salesman earns the most, including commission?
Select
From
Where
And
= (Select
From
Where
);
D3
Who in DALLAS has the highest salary?
D4
List the names of the people who work with Jones (Employee number 7566)
in his department.
Select
From
Where
And
=(Select
From
Where
28/04/17
25
);
SQL primer workbook 1999/2000
13.4.4
Section E
CREATING AND POPULATING TABLES AND VIEWS (DDL)
The CREATE statement
Example of a simple table create statement :
CREATE TABLE DEPT
(DEPTNO
DNAME
LOC
NUMBER(2) ,
CHAR(12) ,
CHAR(12)) ;
 NOTE: MYSQL users: replace Number(2) with SMALLINT
This simple version of the CREATE statement does not define integrity constraints
(neither entity integrity nor referential integrity).
Data Types
Data types are undoubtedly the most troublesome aspect of inter-database
operations. Even with the simplest of statement, as above, it is quite possible that
what works for one database will not work in another. In this case MySQL does not
have a NUMBER type.
If you find yourself needing to write scripts which work on multiple databases, you will
need to find the common types where possible, and then translate to the most
appropriate data type for those non-common data types.
Oracle would expect you to code the above SQL statement as it is. However, Oracle
also recognises most ANSI standard data types and converts them internally. This
means that you could equally well have coded the same thing thus:
CREATE TABLE DEPT2
(DEPTNO
SMALLINT ,
DNAME
CHAR(12) ,
LOC
CHAR(12)) ;
This causes no error since Oracle converts SMALLINT to Number(38). You can
prove this by asking Oracle to describe the table (DESC DEPT2) and you will see the
data type is not defined as SMALLINT, even though that is how you coded it.
Knowing that SMALLINT is valid in both MySQL and Oracle environments means
you could save the latter code and use it in both databases. This is an example of
finding the common data types. Do not get too worried about this, however. Most
Oracle professionals will spend most of their time using Oracle only, and will never
need to translate between databases.
Each column must have a name and a data type assigned to it. The declaration of
the data type determines the operations that may be performed on the data. E.g.
28/04/17
26
SQL primer workbook 1999/2000
13.4.4
Section E
string variables will be sorted alphabetically, i.e. in ASCII code order, and numeric
data will be sorted numerically.
The main data types supported by the version of SQL used on the course are shown
in the table below:Oracle Data Type
MySQL Data Type
CHAR(n)
CHAR(n)
CHARACTER(n)
CHARACTER(n)
VARCHAR2(n)
VARCHAR(n)
Variable length character string, having a maximum
length of n. n<2001, default 1. Needs less storage if
strings are of mixed lengths, but not ANSI standard
LONG
LONGTEXT
Variable length character string, having a maximum
length of 2 gigabytes (both databases)
NUMBER(p, s)
NUMERIC(p,s)
Number having size p and s decimal places.
NUMBER(n)
TINYINT
Integer of size n (Oracle), Tinyint = 1 byte
SMALLINT
SMALLINT
Translated to Number in Oracle, 2 bytes in Mysql
BIGINT
8 bytes
DATE
Dates ranging from 1/1/4712BC to 31/12/4712AD. Oracle
uses a single data type for both dates and time, so even
when you declare as a date, you are actually storing:
DATE
TIME
DATETIME
28/04/17
Description
Fixed length character string of n characters. n<256,
default n=1. Always padded to n.
century, year, month, day, hour, minute,
second
MySQL does allow you to store DATE only,
TIME only, or DATETIME.
27
SQL primer workbook 1999/2000
14.2, 14.3
Section E
The CONSTRAINT clause
The flexibility and ease of use of Relational Databases is achieved at a cost - data
integrity is sacrificed. It is possible to create databases which allow users to delete
important cross-referenced data, to create inconsistencies within the data values and
to produce invalid output from an SQL procedure.
Semantic integrity constraints may be applied to :
a single column in a table
e.g. Salary must be between 1000 and 30000
Sex code must be 'M' or 'F'
multiple columns in a table
e.g. Start date must be greater than Birth date
multiple rows in a table
e.g. Salary of employee must be less than
salary of manager
Existence constraints apply to 2 or more rows in different tables:
A row in an Order table cannot be created unless there is a corresponding row in the
Customer table.
The constraint clause can be applied to a column (or group of columns) as part of
the CREATE and ALTER TABLE statements. The constraint is checked and applied
when an INSERT, UPDATE or DELETE statement is run against that table.
Constraints may impose the following combinations of rules :
.
require that the value of a column(s) is NOT NULL
.
require that the value of a column(s) is UNIQUE within the table
.
identify a column(s) as the PRIMARY KEY
.
require the value of a column(s) to exist in another table as a FOREIGN KEY
.
require the value in a column to conform to an expression (CHECK)
Note :
PRIMARY KEY is a superset of UNIQUE
UNIQUE requires NOT NULL to be specified
UNIQUE cannot be applied to PRIMARY KEY
28/04/17
28
SQL primer workbook 1999/2000
14.2, 14.3
Section E
Each constraint may be optionally named so that it can be located in the data
dictionary. If a constraint is not named then it cannot be identified to be used within
a MODIFY statement.
The easiest way to implement constraints is as part of the field declarations in the
CREATE statement e.g.
CREATE TABLE Dept
(Deptno
NUMBER
Dname
Loc
CHAR(12)
CHAR(12)
PRIMARY KEY
CHECK (Deptno BETWEEN 10 AND 99),
CHECK (Dname = UPPER(DNAME)),
CHECK (Loc IN
('DALLAS', 'BOSTON', 'NEW YORK')));
The constraints may also be defined as separate clauses at the end of the CREATE
statement but still within it e.g.
CREATE TABLE Dept2
( Deptno
NUMBER,
Dname
CHAR(12),
Loc
CHAR(12),
PRIMARY KEY (DEPTNO),
CONSTRAINT dept_pk
CHECK (Deptno BETWEEN 10 AND 99),
CONSTRAINT dname_ck
CHECK (Dname = UPPER(DNAME)),
CONSTRAINT loc_ck
CHECK (Loc IN ('DALLAS', 'BOSTON', 'NEW YORK')));
Oracle automatically enforces NOT NULL on column(s) defined as Primary Keys.
MySQL need you to defined the column as NOT NULL first:
SAME EXAMPLE in MYSQL:
CREATE TABLE Dept2
( Deptno
Integer NOT NULL,
Dname
CHAR(12),
Loc
CHAR(12),
PRIMARY KEY (DEPTNO),
CONSTRAINT dept_pk
CHECK (Deptno BETWEEN 10 AND 99),
CONSTRAINT dname_ck
CHECK (Dname = UPPER(DNAME)),
CONSTRAINT loc_ck
CHECK (Loc IN ('DALLAS', 'BOSTON', 'NEW YORK')));
28/04/17
29
SQL primer workbook 1999/2000
14.2, 14.3
Section E
This method is the only way to handle compound primary keys:CREATE TABLE Order_Detail
(Order_Id
NUMBER,
Part_Number
NUMBER,
Quantity
NUMBER,
PRIMARY KEY (Order_Id, Part_Number));
MySQL Version:
CREATE TABLE Order_Detail
(Order_Id INTEGER NOT NULL,
Part_NUMBER INTEGER NOT NULL,
Quantity INTEGER,
PRIMARY KEY (Order_Id, Part_NUMBER));
To establish referential integrity the column(s) in the related table must have foreign
key references defined on them:CREATE TABLE Order_detail2
(Order_id
NUMBER,
Part_number
NUMBER,
Quantity
NUMBER,
PRIMARY KEY (Order_id, Part_number),
FOREIGN KEY (Order_id)
REFERENCES Order(Order_id),
FOREIGN KEY (Part_number) REFERENCES Part(Part_number));
This table must be created after the tables named Order and Part, both of which
must have the referenced columns defined as primary keys.
 NOTE, MySQL users: The FOREIGN KEY syntax in MySQL exists only for
compatibility with other SQL vendors' CREATE TABLE commands; it doesn't do
anything.
There are real performance issues around foreign keys. You should use them by
default, and certainly in the majority of cases they are sensible. However, whilst it is
hard to give rules for when not to use them, please remember that they can be a
place to look for improvements if performance becomes an issue. If you can
guarantee that data that is being entered is consistent with the relationships (and
perhaps able to enforce that at the client end) you will save the database doing a lot
of work during inserts and updates on tables which contain these constraints.
28/04/17
30
SQL primer workbook 1999/2000
14.2, 14.3
E1
Section E
Create the three tables which implement the Customer database that was
described in Section A. Ensure that you have implemented the primary keys
correctly and set up the referential integrity. (If you need to delete a table the
command is DROP TABLE table_name.)
CREATE
CREATE
CREATE
The ALTER statement
An existing table can be altered at any time by adding table elements, modifying
column definitions and dropping constraints.
E.g.
Or
Or
Or
28/04/17
ALTER TABLE Order_detail
ADD Order_date DATE;
Creates a new column
ALTER TABLE Order_detail
MODIFY Part_number CHAR(5);
Changes the data type
ALTER TABLE Dept
DROP CONSTRAINT dname_ck;
Removes the format constraint
ALTER TABLE Order_detail
MODIFY (Quantity NOT NULL);
Requires a value in the column
31
SQL primer workbook 1999/2000
13.3.10
Section E
The INSERT statement
Values are inserted into a table one row at a time e.g.
INSERT INTO CUST (REFNO, NAME, ADDRESS. AREA)
VALUES ('A123', ‘J Doe’, ‘1 High Street’, ‘Sheffield’) ;
(Note the mandatory single quote marks around character strings, this also applies to
the entering of dates).
If we are supplying a value for every field in the row then we can omit the references
to the field names, provided the values are presented in the same sequence as the
fields appear in the table.
INSERT INTO ACC
VALUES (1245890, 234.5, ’Broomhill’)
We can also insert rows into a table with data values missing, provided we don’t try
to omit values which are defined as mandatory (not null) e.g. the primary key, but this
will require us to specify the columns into which we are entering data.
INSERT INTO ACC (ACCNO)
VALUES (1255546);
will insert a new row with an account number set up and the remaining columns null.
E2
Populate the tables that you created in Exercise E1 with the data as given in
Section A. Give an example of each insert statement in the boxes below.
Insert
Insert
Insert
The DELETE statement
28/04/17
32
SQL primer workbook 1999/2000
13.3.10
Section E
The DELETE statement is used to remove rows from a table.
WHERE clause to specify which rows are to be deleted:DELETE
FROM
WHERE
This requires a
CUST
REFNO = ‘A123’;
Unlike the INSERT which operates on a row at a time, DELETE will remove all the
rows that satisfy the ‘where’ condition - so beware!
E3
R Best has decided to transfer her business from the bank. Delete all the
information about her from the database.
The UPDATE statement
To change the data values in the database use the UPDATE statement
UPDATE is similar to DELETE in that it can operate on a set of rows that meet a
specific condition, not just a single row.
UPDATE
SET
WHERE
E4
ACC
BALANCE = 500
ACCNO = 1245890;
J Best has moved to 31 Hanover Street, Chapeltown. Amend the data in the
Customer table to reflect this change.
In Summary:Adding rows
28/04/17
INSERT INTO table_name (column1, ....)
33
SQL primer workbook 1999/2000
13.3.10
Section E
VALUES
(field_value1,....);
Updating / Modifying
UPDATE
SET
WHERE
table_name
column = field_value
test condition;
Deleting / Removing
DELETE
FROM
WHERE
table_name
test condition;
28/04/17
34
SQL primer workbook 1999/2000
14.4
Section F
The COMMIT statement
This may be issued to make permanent on the database all changes caused by the
last transaction. A transaction is 'a logical unit of work' and a 'unit of integrity' in that
it takes the database from one known state to another. In between these states, the
database may be partially updated and inconsistent.
A normal QUIT from ORACLE or the execution of a DDL statement will automatically
cause a commit. A commit also releases the transaction's locks (this is concurrency
control).
As an alternative to having to issue a commit, this can be performed automatically
after each successful sql statement by setting AUTOCOMMIT ON. The current state
of autocommit can be seen by the SHOW AUTOCOMMIT statement.
 By default, MySQL runs in autocommit mode. This means that as soon as you
execute an update, MySQL will store the update on disk.
The SAVEPOINT statement (Oracle Only)
If autocommit is off a savepoint can be issued at a suitable stage during a
transaction (a set of sql statements):e. g. SAVEPOINT stage 2;
This is a useful facility if the sql is embedded within an application program that may
contain several functions. If an error occurs it is possible to return the database state
to that just prior to the function that failed, rather than have to re-run the entire
application.
The ROLLBACK statement (Oracle Only)
This will undo the effects of the last transaction:e.g. ROLLBACK
We can also rollback to a specified savepoint:e.g. ROLLBACK TO SAVEPOlNT stage_2;
28/04/17
35
SQL primer workbook 1999/2000
Appendix
INDEXES
ORACLE uses indexes to improve performance when
 accessing tables in the index column order
 searching for rows with specified index column values.
However; an index slows down insertions; deletions and changes in indexed column
values. Several indexes may be created on the same table using different columns.
Null values are not indexed.
CREATE INDEX
ON
name
table-name (column,...... );
or
CREATE UNIQUE INDEX name
ON
table-name (column,...... );
Example
CREATE INDEX Customers ON CUST (Area, RefNo)
This will create an index on the table CUST that will order the records alphabetically
by area and then by reference number within each area.
i.e:REFNO
NAME
ADDRESS
AREA
C371
R Done
23 Middle Avenue
Barnsley
B127
R Best
4 East Row
Rotherham
B128
J Best
4 East Row
Rotherham
A123
J Doe
1 High Street
Sheffield
A124
J Smith
2 West Street
Sheffield
Indexes are used to a) speed up retrieval of rows from the table
and/or
b) enforce uniqueness on values in a column.
F3
Create a unique index on the EMP table using the column ename and then try
to enter the following record.
EMPNO
7599
28/04/17
ENAME
JONES
JOB
CLERK
MGR
7839
36
HIREDATE
02-APR-81
SAL COMM
975
DEPTNO
20
SQL primer workbook 1999/2000