1. Digital recordkeeping and preservation I
Databases : SQL
ARK2100
Digital recordkeeping and preservation I
2016
Thomas Sødring
P48-R407

2. Structured Query Language (SQL)
SQL is a database language designed for managing data in relational databases
Strictly speaking SQL includes DML, DDL and DCL commands
Select * from Car;
Car
registrationNr
chassisNr
colour
manufacturer
model
LH12984
Red
Volkswagen
Golf
DK23491
Blue
Toyota
Yaris
BP12349
Green
Skoda
Fabia
ZT97495
White
Seat
Leon

3. SELECT Attribute(s) FROM Relation;

4. SQL Examples SELECT * FROM Car; SELECT registrationNr FROM Car;
SELECT registrationNr, colour FROM Car;
SELECT * FROM Airport;
SELECT IntCode FROM Airport;
SELECT * FROM Building;

5. The previous examples gave us all the rows in the table, but you often only want a specific row or field in the database
To filter we use
WHERE

6. SELECT Attribute(s) FROM Relation WHERE Attribute = 'value';

7. SELECT with WHERE SELECT * FROM Car WHERE colour = 'red';
SELECT * FROM Car WHERE colour = 'red' AND
model = 'Golf';
SELECT registrationNr, colour FROM Car
WHERE registrationNr = 'ZH10045';
SELECT name FROM Airport
WHERE country = 'Norway';

8. SQL wildcards
SQL supports wildcards that can be used to increase the number of rows in a query
They make the search 'fuzzy'
SQL uses % and _ as wildcards
% means any text up to this point
%land - Ireland, England, Finland, Iceland
_ means any character at this point
_an - van, ban, man
Can be useful if you only remember part of a string
The equivalents for '%' and '_' in operating system terminals are '*' and '?'

9. Wildcard examples
Search for any registration number that begins with LH
LIKE 'LH%'
Search for any registration number that contains the number '12'
LIKE '%12%'
Search for any registration number that begins with 'L' and the number section contains '12345'
LIKE 'L_12345'

10. SELECT with wildcards SELECT * FROM Car WHERE
registrationNr LIKE 'LH%';
registrationNr LIKE '%12%';
registrationNr LIKE 'L_12345';

11. SQL DISTINCT
From the data in your tables, you may want a list of unique or distinct values in an attribute / table
The Car table will have many cars from many different manufacturers
Volkswagen, Skoda, Renault
Create a list of all the distinct (unique) car manufacturers that are in the database
Imagine trying to do this manually
Very good example of why we digitise

12. SELECT with DISTINCT Create a list of the different car manufacturers
Create a list of the different car models
SELECT DISTINCT manufacturer
FROM Car;
SELECT DISTINCT model
FROM Car;
List of manufacturers and models?

13. SELECT and LIMIT
You can limit the result set and specify the maximum number of rows you want back
You can also limit the result set by ranges, e.g. ignore the first two rows and fetch the next 5
SELECT * FROM Car LIMIT 3;
SELECT * FROM Car LIMIT 2,5;

14. SELECT with ORDER BY
Remember a database will retrieve data a row at a time
The DBMS returns results in the same order as it retrieves data from the table
Sometimes it is desirable to sort these results

15. WHERE Attribute = 'value'
SELECT Attribute(s)
FROM Relation
WHERE Attribute = 'value'
ORDER BY Attribute
[ ASC | DESC ] ;

16. SELECT with ORDER BY SELECT * FROM Car WHERE colour = 'red' LIMIT 20;
ORDER BY chassisNr ASC LIMIT 20;
ORDER BY model DESC LIMIT 20;
WHERE colour= 'red'
ORDER BY manufacturer LIMIT 20;

17. SQL GROUP BY
A DBMS supports a number of built-in calculation functions
These can also be used in conjunction with GROUP BY clause
AVG ()
MAX ()
MIN ()
COUNT ()
MySQL supports far more than these statistics

18. SQL GROUP BY GROUP BY can answer questions like:
How many Cars are there of each colour?
How many 'Volkswagen Golf' are there of each colour?
SELECT colour, COUNT( * ) AS 'Total' FROM
Car GROUP BY colour;
SELECT colour, COUNT( * ) AS 'Golf colours' FROM
Car WHERE model = 'Golf' GROUP BY colour;

19. Crime Scene Investigators
A red car with LH in the registration number was observed driving away from the scene of an accident
Witnesses claim the car was either a Volkswagen Golf or Renault Megane
With your advanced database skills can you help the police narrow down the search and provide a list of potential cars
The solution can be several SQL commands

20. or CSI solution SELECT * FROM Car WHERE registrationNr LIKE 'LH%' AND
colour = 'red' AND model = 'Golf';
colour = 'red' AND model = 'Megane'; or
SELECT * FROM Car WHERE
registrationNr LIKE 'LH%' AND
colour = 'red' AND
(model = 'Megane' OR model = 'Golf');

21. null Null has a special meaning in a database
It means no value has been assigned
In phpmyadmin, a blank field that has null will typically be denoted with the word NULL in italics
You can also explicitly state that a field is blank in an insert statement
But null and 'null' are NOT the same thing
Using 'null' results in an assignment of the textual string 'null' to a field and as such is not null

22. null So we use null to say that we do not have a value
The following is incorrect:
INSERT INTO Car VALUES (
"RQ10000", " ", null, null, "A3" );
INSERT INTO Car VALUES (
"RQ10000", " ", "null", "null", "A3" );

23. SQL and JOINs
So far we have explored SQL where the data is contained within a single table
This really only corresponds to flat files and does not reflect the relational model with data across tables
Data in tables can be joined in different ways
Inner / Natural join
Right outer join
Left outer join
Full outer join
Anti join

24. SQL and JOINs We introduce two new tables, Employee and Department
empNr
firstname
surname
department
department
manager 1
John
Smith
Sales
Sales
Celine 2
Jim
Hansen
Finance
Admin
Oliver 3
Frank
Burton
Sales
Finance
George 4
Mary
Abbot
Finance
Departments.Department is a foreign key that refers to Employee.Department

25. SQL INNER Join
A SQL INNER JOIN between two tables should return all rows that match on a common attribute
SELECT * FROM Employee INNER JOIN
Department ON Employee.department =
Department.department;

26. SQL INNER JOIN
Employee
Department

27. ⋈ SQL INNER Join Jim Finance John Sales Mary Shipping Frank 1 2 3 4
Employee
firstname
department
empNr
Hansen
Smith
Abbot
Burton
surname
Admin
Sales
Department
department
Oliver
Celine
manager
Finance
George ⋈
Jim
Finance
John
Sales
Frank 1 2 3
Employee ⋈ Department
firstname
department
empNr
Hansen
Smith
Burton
surname
George
Celine
manager

28. SQL NATURAL Join
A SQL NATURAL JOIN is like an INNER JOIN but the common column is not repeated
Joins on columns with the same name and same data type
SELECT * FROM Employee NATURAL JOIN
Department;

29. SQL LEFT OUTER JOIN
A SQL LEFT OUTER JOIN between two tables, A and B, should return all rows in A and all rows in B that match rows in A on a common attribute
SELECT * FROM Employee LEFT OUTER JOIN
Department ON Employee.department =
Department.department;

30. SQL LEFT OUTER JOIN
Employee
Department

31. ⟕ SQL LEFT OUTER JOIN Jim Finance John Sales Mary Shipping Frank 1 2 34
Employee
firstName
department
empNr
Hansen
Smith
Abbot
Burton
surname
Admin
Sales
Department
department
Oliver
Celine
manager
Finance
George ⟕
Employee ⋈ Department
empNr
firstname
surname
department
manager 1
John
Smith
Sales
Celine 2
Jim
Hansen
Finance
George 3
Frank
Burton
Sales
Celine 4
Mary
Abbot
null
null

32. SQL RIGHT OUTER JOIN
A SQL RIGHT OUTER JOIN between two tables, A and B, should return all rows in B and all rows in A that match rows in B on a common attribute
SELECT * FROM Employee RIGHT OUTER JOIN
Department ON Employee.department =
Department.department;

33. SQL RIGHT OUTER JOIN
Employee
Department

34. ⟖ SQL RIGHT OUTER JOIN Jim Finance John Sales Mary Shipping Frank 1 23 4
Employee
firstName
department
empNr
Hansen
Smith
Abbot
Burton
surname
Admin
Sales
Department
department
Oliver
Celine
manager
Finance
George ⟖
Employee ⋈ Department
empNr
firstname
surname
department
manager 1
John
Smith
Sales
Celine 2
Jim
Hansen
Finance
George 3
Frank
Burton
Sales
Celine
null
null
null
Admin
Oliver

35. SQL FULL OUTER JOIN Employee Department
A SQL FULL OUTER JOIN between two tables, A and B, should return all rows in A, all rows in B and all rows in A and B that match on a common attribute
Employee
Department

36. SQL FULL OUTER JOIN MySQL does not support a FULL OUTER JOIN
Can be emulated by combining a left and right outer join
SELECT * FROM Employee LEFT OUTER JOIN
Department ON Employee.department =
Department.department
UNION
SELECT * FROM Employee RIGHT OUTER JOIN
Department.department;

37. ⟕ SQL FULL OUTER JOIN Jim Finance John Sales Mary Shipping Frank 1 2 34
Employee
firstName
department
empNr
Hansen
Smith
Abbot
Burton
surname
Admin
Sales
Department
department
Oliver
Celine
manager
Finance
George ⟕
Employee ⋈ Department
empNr
firstname
surname
department
manager 1
John
Smith
Sales
Celine 2
Jim
Hansen
Finance
George 3
Frank
Burton
Sales
Celine 4
Mary
Abbot
null
null
null
null
null
Admin
Oliver

38. SQL ANTI JOIN
A SQL Anti Join between two tables, A and B, is the set of all rows in A that do not have a matching row in B on a common attribute
Can answer the following question
Which Employee has no Manager / Department?
SELECT * FROM Employee
WHERE ( department ) NOT IN (
SELECT department FROM Department );

39. ∆ SQL ANTI JOIN Jim Finance John Sales Mary Shipping Frank 1 2 3 4
Employee
firstName
department
empNr
Hansen
Smith
Abbot
Burton
surname
Admin
Department
Oliver
Celine
manager
George ∆
firstname
Employee Departments

40. Views
A view is a stored SQL query that looks like a table and becomes a sort of virtual table
A view can be used to
define a subset of tables / data
link data from multiple tables (JOIN)
You can limit a users access to a view
Views can be specified with read or write access to data

41. View with read access CREATE VIEW AllData AS
A view that combines the Employee and Department for a table / view
CREATE VIEW AllData AS
SELECT empNr, firstname, surname,
Employee.department, manager
FROM Employee, Department
WHERE
Employee.department =
Department.department;

42. Index
If you want to search for data from a particular attribute (column) and the table has large amounts of data, a search can take considerable time (in seconds)
There are 2.9 million cars in Norway, and if you search for chassisNr then you potentially have to search through all 2.9 million rows
The car you are looking for was in row number 2,899,999
A DBMS supports indexes that makes it significantly faster to find a specific row
The difference is anywhere from minutes / seconds to under one second depending on database size
The primary key is indexed automatically
Each index increases the size of the database on disk

43. ALTER TABLE CREATE INDEX
ALTER TABLE Car
ADD INDEX
chassis_index (chassisNr);
# or
ADD UNIQUE