1 SQL n Structured Query Language n Declarative l Specify the properties that should hold in the result, not how to obtain the result l Complex queries.

1

SQL Structured Query Language Declarative

Specify the properties that should hold in the result, not how to obtain the result

Complex queries have procedural elements International Standard

SQL1 (1986) SQL2 (SQL-92) SQL3 (pieces have started to appear)

Two components DDL statements DML statements

2

Basic SQL structure SELECT A1, A2, …, An

FROM R1, R2, …, Rm

WHERE P where

Ai are attributes from…

Ri which are relations

P is a predicate* can replace Ai’s if all attributes are to be retrieved

SQL Queries

3

What Kind of Predicates?

Simple predicates: Expression Value where Expression can be an

attribute or an arithmetic expression involving attributes = {<, >, =, <=, >=, <>} and Value can be from one of the data types

Example: Name = ‘J. Doe’ (Age + 30) >= 65

Compound predicates Simple predicates combined with logical connectives

AND, OR, NOT

4

Example Simple Queries

List names of all branches in the bank.SELECT B-nameFROM branch

List names of all customers who have an account together with their account numbers.

SELECT C-name, Acc-numberFROM deposit

Find all cities where at least one customer lives.SELECT DISTINCT CityFROM customer

branch (B-name, Address, City, Assets)customer(C-name, Street, City)deposit(Acc-number, C-name, B-name, Balance)borrow(Acc-number, C-name, B-name, Amount)

5

Queries With Predicates

Find the names of all branches with assets greater than $2,500,000.

SELECT B-nameFROM branchWHERE Assets > 2500000

Find the names of all branches in Edmonton with assets greater than $2,500,000.

SELECT B-nameFROM branch’WHERE Assets>2500000 AND City=‘Edmonton’


6

Ordering the Results

Find the names and assets of all branches with assets greater than $2,500,000 and order the result in ascending order of asset values.

SELECT B-name, AssetsFROM branchWHERE Assets > 2500000ORDER BY Assets

Default is ascending order, but descending order can be specified by the DESC keyword.


7

Queries Over Multiple Relations

List the name and cities of all customers who has an account with balance greater than $2,000.

SELECT C-name, CityFROM customer, depositWHERE Balance > 2000AND customer.C-name = deposit.C-name

Find the name and assets of all branches which have customers living in Jasper.

SELECT B-name, AssetsFROM branch,customer,depositWHERE customer.City = ‘Jasper’ AND customer.C-name = deposit.C-nameAND deposit.B-name = branch.B-name


8

Queries Over Multiple Relations

List the pairs of customer names who have loans from the same bank branch.

SELECT P1.C-name, P2.C-nameFROM borrow P1, borrow P2 WHERE P1.B-name = P2.B-name


9

UNION, MINUS, INTERSECT Operands may be specified by

create temp relationsASSIGN TO temp-1 ASSIGN TO temp-1SELECT A1, ..., An SELECT B1, ..., Bp

FROM R1, ..., Rm FROM S1, ..., Sr

WHERE P; WHERE Q;temp-1 UNION temp-2;

use parentheses(SELECT A1, ..., An

FROM R1, ..., Rm WHERE P)

UNION(SELECTB1, ..., Bp

FROM S1, ..., Sr

WHERE Q);

Queries Involving Set Operators

10

Queries With Set Operators

Find all cities where there is either a customer or a branch.(SELECT CityFROM customer)UNION(SELECT CityFROM branch)

Find all cities that has a branch but no customers.(SELECT CityFROM branch)MINUS(SELECT CityFROM customer)


11

Queries With Set Operators

Find all the cities that have both customers and bank branches.

(SELECT CityFROM customer)INTERSECT(SELECT CityFROM branch)


12

Queries within the WHERE clause of an outer query

SELECTFROMWHERE OPERATOR

(SELECT FROM WHERE)

There can be multiple levels of nesting These can usually be written using other constructs

(UNION, JOIN, etc). Three operators: IN, (NOT) EXISTS, CONTAINS

Queries With Nested Structures

13

SELECTFROMWHERE A1, ..., An IN

(SELECT A1, ..., An FROM R1, ..., Rm WHERE P)

Semantics: Values of attributes A1, ..., An are found in the relation that is returned as a result of the calculation of the inner query.

Other comparison operators {<, <=, >, >=, <>} can be used in place of IN.

“IN” Construct

14

“IN” Construct Example

Find all the customers who have a deposit at some branch at which John Doe has a deposit.

SELECTC-nameFROM depositWHERE B-name IN

(SELECT B-name FROM deposit WHERE C-name = ‘John

Doe’)


15


Find all the customers who have either a deposit or a loan and their cities.

SELECT C-name, CityFROM customerWHERE C-name IN (SELECT C-name

FROM deposit )OR

C-name IN (SELECT C-name FROM borrow)


16


Find all the branches that have assets greater than all branches in Calgary.

SELECT B-nameFROM branchWHERE Assets > ALL

(SELECT Assets FROM branch WHERE City = ‘Calgary’)


17

SELECTFROMWHERE (NOT) EXISTS

(SELECT *

FROM R1, ..., Rm WHERE P)

Semantics: For each tuple of the outer query, execute the inner query; if there is (no) at least one tuple in the result of the inner query, then retrieve that tuple of the outer query.

This accounts for the “there exists” type of queries

“EXISTS” Construct

18

“EXISTS” Construct Example

Find the names of customers who live in a city which has a bank branch.

SELECT C-nameFROM customerWHERE EXISTS (SELECT *

FROM branch WHERE customer.City=branch.City)

Find the names of customers who live in a city with no bank branches.

SELECT C-nameFROM customerWHERE NOT EXISTS (SELECT *

FROM branch WHERE customer.City=branch.City)


19


Find the names and addresses of all the customers who have no deposit in the bank.

SELECT C-name, Street, CityFROM customerWHERE NOT EXISTS

(SELECT * FROM deposit WHERE customer.C-name=deposit.C-name)


20


Find all the customers who have deposits in every branch of the bank. Find all customers such that there is no bank branch where they

do not have depositsSELECT *FROM customerWHERE NOT EXISTS

(SELECT B-name FROM branch WHERE NOT EXISTS

(SELECT * FROM deposit

WHERE branch.B-name=deposit.B-name AND customer.C-name = deposit.C-name))


21

“CONTAINS” Construct

(SELECT A1, ..., An

FROM R1, ..., Rm WHERE P)CONTAINS

(SELECT B1, ..., Bp

FROM S1, ..., Sr

WHERE Q)

Semantics: Compare the the result relations of the two queries and return TRUE if the second one is contained in the first.

22

“CONTAINS” Construct Example

Find all the customers who have a loan from every branch in Edmonton.

SELECT *FROM customerWHERE ( (SELECT B-name

FROM borrow WHERE customer.C-name=borrow.C-name)CONTAINS(SELECT B-name FROM branch WHERE City = ‘Edmonton’))


23

Specify a function that calculates a numeric value from a given relation. The function is usually applied to an attribute. COUNT, SUM, MAX, MIN, AVG

SELECT AggFunc(Ai), …, AggFunc(Aj)

FROM R1, ..., Rm

WHERE P

Aggregate Functions

24

Aggregate Query Examples

Find the total assets of branches in Edmonton.SELECT SUM(Assets)FROM branchWHERE City = ‘Edmonton’

Find the number of cities where John Doe has an account with any bank branch.

SELECT COUNT(DISTINCT City)FROM branch, depositWHERE branch.B-name = deposit.B-nameAND deposit.C-name = ‘John Doe’


25

Aggregate Query Examples

Find the names of branches which have assets greater than the average assets of all bank branches.

SELECT B-nameFROM branchWHERE Assets >

(SELECT AVG(Assets) FROM branch)


26

Grouping Queries Group the results according to a set of attributes

SELECT Ai, …, An

FROM R1, ..., Rm

WHERE PGROUP BY Aj …, Ak

Rules: All of the attributes in the SELECT clause that are not

involved in an aggregation operation have to be included in the GROUP BY clause.

GROUP BY can have more attributes (k n)

27

Grouping Query Examples

For each city, find the names of branches.SELECT City, B-nameFROM branchGROUP BY City, B-name

For each branch at each city, list the account numbers of the customer’s loans over $100,000.

SELECT City, C-name, Acc-numberFROM customer, borrowWHERE customer.C-name = borrow.C-nameAND Amount > 100000GROUP BY City,B-name,C-name,Acc-number


28

Predicates on Groups Group the results according to a set of attributes if

they satisfy a certain conditionSELECT Ai, …, An

FROM R1, ..., Rm WHERE PGROUP BY Aj …, Ak

HAVING Q Rules:

Q must have a single value per group. An attribute in Q has to either appear in an aggregation

operator (in the SELECT clause) or be listed in the GROUP BY

29


Find the cities with more than two bank branches.

SELECT CityFROM branchGROUP BY CityHAVING COUNT(*) > 2


30


For each city where there are more than two branches, find the names of branches.

SELECT City, B-nameFROM branchWHERE City IN

(SELECT City FROM branch GROUP BY City HAVING COUNT(*) > 2)


31


Find the branches where the average account balance is greater than $1,200.

SELECT B-name, AVG(Balance)FROM depositGROUP BY B-nameHAVING AVG(Balance) > 1200


32


For each branch, list the names and number of loans of customers who have more than 2 loans over $100,000.

SELECT B-name, C-name, COUNT(Acc-number)FROM borrowWHERE Amount > 100000GROUP BY B-name, C-nameHAVING COUNT(*) > 2


33

Update Commands

Assume R(A1, …, An)

INSERT Form

INSERT INTO RVALUES (value(A1), …, value(An))

You can explicitly specify attribute names

INSERT INTO R(Ai, …, Ak)VALUES (value(Ai), …, value(Ak))

DELETEDELETE FROM RWHERE P

34

Update Commands

Assume R(A1, …, An)

UPDATEUPDATE RSET Ai=value, …, Ak=valueWHERE P

35

Update Command Examples

Insert a new customer record for John Smith who lives on 345 Jasper Avenue, Edmonton.

INSERT INTO customerVALUES (‘John Smith’,‘345 Jasper

Avenue’,‘Edmonton’) Delete all the customers who have less than $1,000 in their

account.

DELETE FROM customerWHERE C-name IN

(SELECT C-name FROM deposit WHERE Balance < 1000)


36

Update Command Examples

Increase by 5% the balances of customers who live in Edmonton and have a balance of more than $5,000.

UPDATE depositSET Balance = Balance*1.05WHERE Balance > 5000AND C-name IN

(SELECT C-name FROM customer WHERE City = ‘Edmonton’)


37

View Definition

General formCREATE VIEW V[(Ai, …, Ak)]

AS SELECT A1, …, An

FROM R1, …, Rm

WHERE P

In queries, treat view V as any base relationSELECT Aj, …, Al

FROM V

WHERE Q

Updates of views may be restricted

38

View Definition Example

Create a view jasper-customer of customers who live in Jasper.

CREATE VIEW jasper-customerAS SELECT *

FROM customerWHERE City = ‘Jasper’


39

View Definition Example

Create a view total-balance that gives, for each customer at each city, the number of accounts owned by the customer and the total balance.

CREATE VIEW total-balance(Name, City, Number, Total)

AS SELECT C.City,C.C-name,COUNT(Acc-number), SUM(Balance)

FROM deposit D, customer CWHERE D.C-name=C.C-nameGROUP BY C.C-name,City


40

View Update

A view with a single defining table is updatable if the view attributes contain the primary key or some other candidate key.

Views defined on multiple tables using joins are generally not updatable.

Views defined using aggregate functions are not updatable.

41

View Management

Two strategies for view management Query modification

A view is a virtual relation Multiple queries on complex views Query processor modifies a query on a view with the view

definition

View materialization View is a materialized table in the database Incrementally update the view (update propagation)

42

Handling Nulls

Attribute values can be NULL NULL is not a constant, nor can it be used as an operand

NAME = NULL Wrong! NULL + 30 Wrong!

Operating on variables with NULL values If x is NULL

x <arithmetic op.> constant is NULL x <arithmetic op.> y is NULL

Comparing a NULL value with any other value returns UNKNOWN (three-valued logic).

43

Outer Join

Ensures that tuples from one or both relations that do not satisfy the join condition still appear in the final result with other relation’s attribute values set to NULL

Only available in SQL2 via join expression R [NATURAL] JOIN S [ON <condition>]

Alternatives R [NATURAL] FULL OUTER JOIN S [ON <condition>] R [NATURAL] LEFT OUTER JOIN S [ON <condition>] R [NATURAL] RIGHT OUTER JOIN S [ON <condition>]

44

SQL DDL Statements

Create schemaCREATE SCHEMA Schema_Name AUTHORIZATION User_Name

Create table Specify a new relation scheme General form

CREATE TABLE <Table_Name>

(Attribute_1 <Type>[DEFAULT <value>][<Null constraint>],

Attribute_2 <Type>[DEFAULT <value>][<Null constraint>],

…

Attribute_n <Type>[DEFAULT <value>][<Null constraint>],

[<Constraints>])

45

Example

Givenbranch (B-name, Address, City, Assets)customer(C-name, Street, City)deposit(Account-number, C-name, B-name,

Balance) Definition of branch

CREATE TABLE branch( B_name CHAR(15),

Address VARCHAR(20),City CHAR(9),Assets DECIMAL(10,2)DEFAULT 0.00);

Assume customer is defined similarly

46

Allowable Data Types Numeric

INT, SHORTINT REAL (FLOAT), DOUBLE PRECISION DECIMAL(i,j)

Character-string CHAR(n), VARCHAR(n)

Bit-string BIT(n), BIT VARYING(n)

Date YYYY-MM-DD

Time HH:MM:SS

Timestamp both DATE and TIME fields plus a minimum of six positions for

fractions of seconds

47

User-Defined Types

Create a DOMAINCREATE DOMAIN <domain_name> AS <primitive_type>

ExampleCREATE DOMAIN Gender AS CHAR(1)

Generally useful only for easy modification of type specification.

The value can be defaulted by the DEFAULT specification

48

NOT NULL specifies that an attribute cannot contain null values should be specified for all primary keys

PRIMARY KEY designates a set of columns as the table’s primary key

UNIQUE specifies the alternate keys

FOREIGN KEY designates a set of columns as the foreign key in a referential

constraint

Key Constraints

49

Example

Givenbranch (B-name, Address, City, Assets)customer(C-name, Street, City)deposit(Account-number, C-name, B-name,

Balance)

Enhance the previous definition of branch:CREATE TABLE branch

( B_name CHAR(15)NOT NULL,

Address VARCHAR(20),

City CHAR(9),

Assets DECIMAL(10,2)DEFAULT 0.00,

PRIMARY KEY (B_name));

50

REFERENTIALLY TRIGGERED ACTION Referential integrity: The primary key of one relation appears as an

attribute (foreign key) of another relation. Example:

customer(C-name, Street, City)deposit(Acc-number, C-name, B-name, Balance) Deletion or update of primary key tuple requires action on the foreign

key tuple. Specify constraint on delete or update. How to manage?

reject cascade: (on delete) automatically remove foreign keys if a referenced key is

removed or (on update) change the foreign key value to the new value of the referenced key

set null set default

Referential Constraints

51

Example

branch (B-name, Address, City, Assets)customer(C-name, Street, City)deposit(Acc-number, C-name, B-name, Balance)

Definition of depositCREATE TABLE deposit( Acc-number CHAR(9)NOT NULL,

C-name VARCHAR(15),B-name VARCHAR(15),Balance DECIMAL(10,2)DEFAULT 0.00,

PRIMARY KEY (Acc-number),FOREIGN KEY (C-name) REFERENCES customer(C-name)ON DELETE SET NULL ON UPDATE CASCADE,

FOREIGN KEY (B-name) REFERENCES branch(B-name));

52

Attribute & Domain Constraints

CHECK specifies the condition that each row has to satisfy

Enhance the previous definition of branch:CREATE TABLE branch(B_name CHAR(15)NOT NULL,

Address VARCHAR(20), City CHAR(9), Assets DECIMAL(10,2)DEFAULT 0.00

CHECK (Assets >= 0),PRIMARY KEY (B_name));

Can be specified on domains as domain constraintsCREATE DOMAIN Gender AS CHAR(1) CHECK (VALUE IN (‘F’, ‘M’));

53

Tuple Constraints Use CHECK command and specify condition The condition is checked every time a tuple is inserted or

updated and the action rejected if the condition is false Example

CREATE TABLE deposit(Acc-number CHAR(9)NOT NULL,

C-name VARCHAR(15), B-name VARCHAR(15), Balance DECIMAL(10,2)DEFAULT 0.00,

PRIMARY KEY (Acc-number),FOREIGN KEY (C-name) REFERENCES customer(C-name)ON DELETE SET NULL ON UPDATE CASCADE,

FOREIGN KEY (B-name) REFERENCES branch(B-name))CHECK (NOT(Acc-number<36300) OR Balance>100000);

54

Assertions

Global constraints of the form CREATE ASSERTION name CHECK (condition)

Example: No branch can have more than two customers who have more than 2 loans over $100,000.

CREATE ASSERTION deposit CHECK

(NOT EXISTS

(SELECT COUNT(Acc-number) FROM borrow WHERE Amount > 100000 GROUP BY B-name, C-name HAVING COUNT(*) > 2));

55

Other SQL DDL Commands

DROP SCHEMA Delete an entire schema CASCADE: delete all the tables in the schema RESTRICT: delete only if empty

DROP TABLE RESTRICT: delete only if not referenced in a

constraint

ALTER TABLE change definition

56

Embedded SQL

Ability to access a database from within an application program.

Issues Interface

The same as ad hoc SQL but with EXEC SQL command Using query results within the application program

Define shared variables using the format allowed in host language Shared variables can be used in INSERT, DELETE, UPDATE as well

as regular queries and schema modification statements Retrieval queries require care

If the result relation has a single tuple: Each value in the SELECT statement requires a shared variable

If the result relation has a set of tuples: Cursor has to be defined

57

Example for Update

Consider an example in which will transfer $50,000 from a customer’s account in one branch to another customer’s account in another branch.

…EXEC SQL BEGIN DECLARE SECTION;int cname1, cname2; /* two customer names */char bname1[15], bname2[15]; /* branch names */int amount; /* amount to be transferred */

EXEC SQL END DECLARE SECTION;/* Code (omitted) to read the customer and branch names and amount */

EXEC SQL UPDATE deposit SET Balance = Balance + :amount WHERE C-name = :cname2 AND B-name = :bname2;

EXEC SQL UPDATE depositSET Balance = Balance - :amountWHERE C-name = :cname1 AND B-name = :bname1;

…

58

Retrieval Queries

If the result is only only one tuple, follow the same approach. Example: For a given branch, retrieve the names and number of

loans of customers who have more than 2 loans over $100,000.…EXEC SQL BEGIN DECLARE SECTION;

char bname[15], cname[15]; /* branch & customer name */int no-acct; /* number of accounts */

EXEC SQL END DECLARE SECTION;…/* Code (omitted) to get the branch name from use */EXEC SQL SELECT C-name, COUNT(Acc-number)INTO :cname, :no-acctFROM borrowWHERE B-name = :bname AND Amount > 100000GROUP BY C-nameHAVING COUNT(*) > 2;/* Code (omitted) to print the total balance */…

59

More General Retrieval Queries…EXEC SQL BEGIN DECLARE SECTION;

<shared variable declarations>EXEC SQL END DECLARE SECTION;…EXEC SQL DECLARE <cursor-name> [options] CURSOR FOR

<query> [options]…EXEC SQL OPEN <cursor-name>…while(condition) {

EXEC SQL FETCH FROM <cursor-name> INTO <shared-variable(s)>if(tuple exists) process itelse break

}EXEC SQL CLOSE <cursor-name>…

60

Cursors Use when the embedded SQL query is expected to return a relation with

more than one tuple Think about it almost as a pointer to successive tuples in the result

relation General form

EXEC SQL DECLARE <cursor-name> [INSENSITIVE][SCROLL] CURSOR FOR <query>ORDER BY <attribute(s)>[FOR READ ONLY];

INSENSITIVE: The cursor is insensitive to changes in the relations referred to in the query while the cursor is open.

SCROLL: Allows the subsequent FETCH command to retrieve tuples other than the default which is moving forward. FETCH [NEXT|PRIOR|FIRST|LAST|RELATIVE [+|-]n|ABSOLUTE [+|-]n]

ORDER BY: Orders the results of the query according to some attribute(s).FOR READ ONLY: Ensures that accesses to the underlying relation(s) via this cursor will not change the

relation.

61

Cursor Example For each branch, retrieve the names and number of loans of customers

who have more than 2 loans over $100,000.…EXEC SQL BEGIN DECLARE SECTION;

char bname[15], cname[15]; /* branch & customer name */int no-acct; /* number of accounts */

EXEC SQL END DECLARE SECTION;…EXEC SQL DECLARE loans CURSOR FOR

SELECT B-name, C-name, COUNT(Acc-number)FROM borrowWHERE B-name = :bname AND Amount > 100000GROUP BY C-nameHAVING COUNT(*) > 2;

…EXEC SQL OPEN loans;…while(1) {

EXEC SQL FETCH FROM loans INTO :bname, :cname, :no-acctif(strcmp(SQLSTATE, “02000”) then breakelse print the info

}EXEC SQL CLOSE loans…

1 SQL n Structured Query Language n Declarative l Specify the properties that should hold in the result, not how to obtain the result l Complex queries.

Documents

n queries

n union

predicates n

n fromr

results n

n operands

n simple predicates

set operators n