Design Patterns

advertisement
Design Patterns and Refactoring
12-Mar-16
History

A Pattern Language: Towns, Buildings, Construction,
Christopher Alexander, 1977

The Timeless Way of Building, Christopher Alexander, 1979

Using Pattern Languages for Object-Oriented Programs (a paper
at the OOPSLA-87 conference), Ward Cunningham and Kent
Beck, 1987

Design Patterns, Erich Gamma, Richard Helm, John Vlissides,
and Ralph Johnson (known as the “Gang of Four”, or GoF), 1994

Refactoring: Improving the Design of Existing Code, Martin
Fowler, 2000
2
Buzzwords


Design Patterns describe the higher-level organization of
solutions to common problems
UML is a diagramming language designed for Object-Oriented
programming


Refactoring is restructuring code in a series of small, semanticspreserving transformations (i.e. the code keeps working) in order
to make the code easier to maintain and modify



Design Patterns are always described in UML notation
Refactoring often modifies or introduces Design Patterns
Extreme Programming is a form of Agile Programming that
emphasizes refactoring
Unit testing is testing classes in isolation


Unit testing is an essential component of Extreme Programming
Unit testing is supported by JUnit
3
Design Patterns

Design Patterns describe the higher-level organization
of solutions to common problems



Design Patterns are a current hot topic in O-O design
UML is always used for describing Design Patterns
Design Patterns are used to describe refactorings
4
UML



UML stands for Unified Modeling Language
UML is a big, complicated diagramming language designed for
Object-Oriented programming
UML comprises at least seven or eight different kinds of
diagrams that can be used to describe:






the organization of a program
how a program executes
how a program is used
how a program is deployed over a network
…and more
This talk will cover just a tiny bit of one kind of diagram, the
class diagram
5
UML class diagrams
Example:
Name of the class
Variables [optional]
Methods

Card
cardId:int
-copy:boolean=false
«constructor» Card(int id)
+isKind(desiredKind:int)
+isSharable():boolean
+toString():String
Key:





+ means public visibility
# means protected visibility
- means private visibility
<blank> means default (package) visibility
static variables are underlined
6
UML relationships
A
A
C
Factory
1..4
B
Class B
extends
class A
B
Class B
implements
interface A
D
Class C
contains
1 to 4 objects
of class D
creates
Product
Other kinds of
relations
7
Sample Design Patterns


There are a few dozen Design Patterns described in the
GoF book
I’ll only talk about a couple, to give some of the flavor
of what they are all about


I’ll try to use examples that are relevant to problems you have
dealt with in this class (plus some I’m particularly fond of)
I’m using UML diagrams in only a few examples,
because they just take too much time to draw in
PowerPoint

Specialized tools, such as Rational Rose, Together, and
ArgoUML are much better for this
8
Problem: Uncertain delegation


Much of the point of polymorphism is that you can just send a
message to an object, and the object does the right thing,
depending on its type
However, if the object might be null, you have to be careful not
to send it any message


Examples:




if (myObject != null) myObject.doSomething();
You have an Ocean, represented by a sparse array containing a few Fish
You have a TrafficGrid, some of which contains Cars and Trucks
You want to send output to somewhere, possibly to /dev/null
If you do a lot with this object, you code can end up cluttered
with tests for null
9
Solution: Null Object

Create another kind of object, a “null object,”
representing the absence of any other kind of object





Example: An Ocean might contain Inhabitants, where
Inhabitant is subclassed by BigFish, LittleFish, Algae, and
NothingButWater
This way, no location in the Ocean is null
If Inhabitant contains a method reproduce(), the subclass
NothingButWater could implement this method with an
empty method body
If appropriate, some methods of the null object could throw an
Exception
Ideally, the superclass (Inhabitant, in this example)
should be abstract
10
Refactoring: Introduce Null Object




The general idea is simple: Instead of having some variables
(locations in the array) be null, have them be “null objects”
However, this requires numerous changes in the code
It’s hazardous to change working code—you introduce bugs that
it can take days to find
Refactoring is all about:




doing an operation like this in small steps,
having an automated set of unit tests, and
running unit tests frequently, so that if an error occurs you can pinpoint it
immediately
This approach makes refactoring much safer and protects against
hard-to-find bugs

As a result, programmers are far more willing to refactor
11
Introduce Null Object: In detail, I
1.
Create a subclass of the source class to act as a null version of
the class. Create an isNull operation on the source class and the
null class. For the source class it should return false, for the null
class it should return true.


2.
3.
You may find it useful to create an explicitly nullable interface for the
isNull method.
As an alternative you can use a testing interface to test for nullness
Compile.
Find all places that can give out a null when asked for a source
object. Replace them to give out a null object instead.
12
Introduce Null Object: In detail, II
4.
Find all places that compare a variable of the source type with
null and replace them with a call to isNull.


You may be able to do this by replacing one source and its clients at a
time and compiling and testing between working on sources.
A few assertions that check for null in places where you should no
longer see it can be useful.
5. Compile and test.
6. Look for cases in which clients invoke an operation if not null
and do some alternative behavior if null.
7. For each of these cases override the operation in the null class
with the alternative behavior.
8. Remove the condition check for those that use the overridden
behavior, compile, and test.
13
Refactoring details




The details of Introduce Null Object were copied directly from
Fowler, pages 261-262
I am not going into this much detail in any of the remaining
examples
Notice, however, that with this list of “baby steps” in front of
you, you can do the refactoring a little at a time, with wellmarked places to do testing, so that it’s very easy to catch and
correct errors
Note also that to do this, you need a good set of totally automated
tests—otherwise the testing you have to do is just too much work,
and you won’t do it


Unless, that is, you have a superhuman amount of discipline
JUnit (now built in to BlueJ) is a great start
14
Scenario: Big fish and little fish

The scenario: “big fish” and “little fish” move
around in an “ocean”



Fish move about randomly
A big fish can move to where a little fish is (and eat it)
A little fish will not move to where a big fish is
Fish
<<abstract>>move()
BigFish
LittleFish
move()
move()
15
Problem: Similar methods in subclasses

Here we have a Fish class with two subclasses, BigFish
and LittleFish





The two kinds move the same way
To avoid code duplication, the move method ought to be in
the superclass Fish
However, a LittleFish won’t move to some locations where a
BigFish will move
The test for whether it is OK to move really ought to be in the
move method
More generally, you want to have almost the same
method in two or more sibling classes
16
Solution: Template method

Note: The Design Pattern is called “Template Method”; the
refactoring is called “Form Template Method”



In the superclass, write the common method, but call an auxiliary
method (such as okToMove) to perform the part of the logic that
needs to differ
Write the auxiliary method as an abstract method



We won’t bother making this distinction in the remainder of the lecture
This in turn requires that the superclass be abstract
In each subclass, implement the auxiliary method according to
the needs of that subclass
When a subclass instance executes the common method, it will
use its own auxiliary method as needed
17
The move() method

General outline of the method:


public void move() {
choose a random direction;
// same for both
find the location in that direction; // same for both
check if it’s ok to move there;
// different
if it’s ok, make the move;
// same for both
}
To refactor:




Extract the check on whether it’s ok to move
In the Fish class, put the actual (template) move() method
Create an abstract okToMove() method in the Fish class
Implement okToMove() in each subclass
18
The Fish refactoring
Fish

<<abstract>>move()
BigFish
LittleFish
move()
move()

Fish
move()
<<abstract>>okToMove(locn):boolean

Note how this works:
When a BigFish tries
to move, it uses the
move() method in
Fish
But the move()
method in Fish uses
the okToMove(locn)
method in BigFish
And similarly for
LittleFish
BigFish
BigFish
okToMove(locn):boolean
okToMove(locn):boolean
19
Problem: Constructors create objects


Constructors make objects. Only constructors can make objects.
When you call a constructor of a class, you will get an instance of
that class.
Sometimes you want more flexibility than that—




You may want to guarantee that you can never have more than one object
of a given class
You may want to create an object only if you don’t already have an
equivalent object
You may want to create an object without being sure exactly what kind of
object you want
The key insight is that, although only constructors make objects,
you don’t have to call constructors directly—you can call a
method that calls the constructor for you

Several “creational” Design Patterns are based on this observation
20
Singleton

A Singleton is a class that can have only one instance



You may want just one instance of a null object, which you use in many
places
You may want to create just one AudioStream, so you can only play one
tune at a time
class Singleton {
private static Singleton instance = new Singleton();
// don’t let Java give you a default public constructor
private Singleton() { }
}
Singleton getInstance() {
return instance;
}
...
21
The Factory Method Design Pattern

Suppose you write a class that works with several
different kinds of objects




You can do this if the classes all have a common interface
You may want to be able to create objects, without being
dependent on the kind of object
A factory methods can create instances of different classes,
depending (say) on its parameters
Example:

Image createImage (String ext) {
if (ext.equals("gif")) return new GIFImage();
if (ext.equals("jpg")) return new JPEGImage();
...
}
22
Problem: Reducing interdependencies

Suppose you have an application that provides multiple services





Suppose further that the application consists of a large number of classes
You want to provide access to those services, without requiring
the user to know all the internal details
For example, you have a simulation of an ocean, containing
assorted kinds of fish, and you want to provide access to that
simulation
Just to make the problem interesting, suppose that you have two
or more such applications (say, Oracle, MySql, and Access 2000)
and you want to write a program that works with any of them
Solution: Use the Façade Design Pattern
23
The Façade Design Pattern

Create a class that accepts many different kinds of requests, and
“forwards” them to the appropriate internal class


Example:


If the back ends vary, you man need to write a separate Interface class for
each (all implementing the same interface), but the users of your Facade
class don’t need to change
class Ocean {
public void setBigFishGestationPeriod(int period) {
BigFish.setGestationPeriod(period);
}
public int getNumberOfAlgae() {
return Algae.getCount();
}
...
}
Of course, the Facade class can do other things as well
24
The Immutable Design Pattern


There are many benefits to objects that cannot be
changed after they have been created
Such objects are called immutable




Objects that refer to an immutable object never have to worry
about whether that object has been changed
Immutable objects are thread-safe—this is a significant
efficiency concern, because synchronization is expensive
Example: Strings in Java
It’s easy to make immutable objects in Java:


Make all instance variables private, and
Provide no methods that change those variables
25
Delegation (or, when not to use inheritance)

When you create a subclass, you agree to inherit all its (nonprivate) fields and methods


Example: A Vector can do everything that a Stack should be
able to do, and much, much more





What if you don’t want them all?
You may want to inherit just some of the functionality, and probably add
some of your own
Inheritance doesn’t let you do that—at least, not easily
If your class wants to hide variables or methods inherited from a
superclass, it shouldn’t inherit from that superclass
If an object needs to be a different subclass at different times
(say, a LittleFish turning into a BigFish), then it shouldn’t be a
subclass of that class in the first place
Instead of inheriting, just use an instance of that class, and
delegate to it
26
Example: Stacks

class Stack {
Vector contents = new Vector();
public void push(Object o) {
contents.add(o); // delegate to the Vector
}
public Object pop() {
return contents.remove(contents.size() – 1);
}
...
}
27
An example





Some time ago I was working
on code to evalute
expressions
Expressions can be parsed
into a tree structure
Now what?
You could walk the tree and,
at each node, use a switch
statement to do the right thing
I “discovered” a better
solution (basically, a simple
form of the Command
Design Pattern)
+
2
*
5
x
Tree for 2 + 5 * x
Command
lhs:Command
rhs:Command
0..2 value:int
evaluate():int
28
Using my “Command” pattern



class Add extends Command {
int evaluate( ) {
int v1 = lhs.evaluate().value;
int v2 = rhs.evaluate().value;
value = v1 + v2;
return value;
}
}
Command
lhs:Command
rhs:Command
0..2 value:int
evaluate():int
To evaluate the entire tree, evaluate the root node
This is just a rough description; there are a lot of other details to
consider



Some operands are unary
You have to look up the values of variables
Etc.
29
The Command Design Pattern

Reasons for using the Command Design Pattern:




You want to control if, when, and in what order the
commands are executed
You want to keep a log of commands executed
Popular reason: You want to manage undo and redo
operations
Possible class organization (from GoF):




AbstractCommand with doIt() and undoIt() methods
ConcreteCommand subclasses of AbstractCommand
Invoker is a class that creates ConcreteCommand objects if it
needs to invoke a command
CommandManager to decide what, when, and how to
execute and undo commands
30
Refactoring

Refactoring is:




Refactoring is not just any old restructuring




restructuring (rearranging) code...
...in a series of small, semantics-preserving transformations (i.e. the code
keeps working)...
...in order to make the code easier to maintain and modify
You need to keep the code working
You need small steps that preserve semantics
You need to have unit tests to prove the code works
There are numerous well-known refactoring techniques

You should be at least somewhat familiar with these before inventing your
own
31
When to refactor

You should refactor:

Any time that you see a better way to do things


You can do so without breaking the code


“Better” means making the code easier to understand and to modify in
the future
Unit tests are essential for this
You should not refactor:


Stable code (code that won’t ever need to change)
Someone else’s code

Unless you’ve inherited it (and now it’s yours)
32
Design vs. coding




“Design” is the process of determining, in detail, what
the finished product will be and how it will be put
together
“Coding” is following the plan
In traditional engineering (building bridges), design is
perhaps 15% of the total effort
In software engineering, design is 85-90% of the total
effort

By comparison, coding is cheap
33
The refactoring environment


Traditional software engineering is modeled after traditional
engineering practices (= design first, then code)
Assumptions:



“Agile” software engineering is based on different assumptions:




The desired end product can be determined in advance
Workers of a given type (plumbers, electricians, etc.) are interchangeable
Requirements (and therefore design) change as users become acquainted
with the software
Programmers are professionals with varying skills and knowledge
Programmers are in the best position for making design decisions
Refactoring is fundamental to agile programming

Refactoring is sometimes necessary in a traditional process, when the
design is found to be flawed
34
A personal view

In my opinion,

Design, because it is a lot more creative than simple coding,
is also a lot more fun



Most small to medium-sized projects could benefit from an
agile programming approach


Admittedly, “more fun” is not necessarily “better”
...but it does help you retain good programmers
We don’t yet know about large projects
Most programming methodologies attempt to turn everyone
into a mediocre programmer


Sadly, this is probably an improvement in general
These methodologies work less well when you have some very good
programmers
35
Back to refactoring

When should you refactor?



Any time you find that you can improve the design of existing
code
You detect a “bad smell” (an indication that something is
wrong) in the code
When can you refactor?


You should be in a supportive environment (agile
programming team, or doing your own work)
You should have an adequate set of automatic unit tests
36
Example 1: switch statements

switch statements are very rare in properly designed
object-oriented code




Therefore, a switch statement is a simple and easily detected
“bad smell”
Of course, not all uses of switch are bad
A switch statement should not be used to distinguish between
various kinds of object
There are several well-defined refactorings for this case

The simplest is the creation of subclasses
37
Example 1, continued

class Animal {
final int MAMMAL = 0, BIRD = 1, REPTILE = 2;
int myKind; // set in constructor
...
String getSkin() {
switch (myKind) {
case MAMMAL: return "hair";
case BIRD: return "feathers";
case REPTILE: return "scales";
default: return "integument";
}
}
}
38
Example 1, improved

class Animal {
String getSkin() { return "integument"; }
}
class Mammal extends Animal {
String getSkin() { return "hair"; }
}
class Bird extends Animal {
String getSkin() { return "feathers"; }
}
class Reptile extends Animal {
String getSkin() { return "scales"; }
}
39
How is this an improvement?




Adding a new animal type, such as Amphibian, does
not require revising and recompiling existing code
Mammals, birds, and reptiles are likely to differ in other
ways, and we’ve already separated them out (so we
won’t need more switch statements)
We’ve gotten rid of the flags we needed to tell one kind
of animal from another
Basically, we’re now using Objects the way they were
meant to be used
40
JUnit tests





As we refactor, we need to run JUnit tests to ensure that we
haven’t introduced errors
public void testGetSkin() {
assertEquals("hair", myMammal.getSkin());
assertEquals("feathers", myBird.getSkin());
assertEquals("scales", myReptile.getSkin());
assertEquals("integument", myAnimal.getSkin());
}
This should work equally well with either implementation
The setUp() method of the test fixture may need to be modified
A (beta) version of JUnit is now included in BlueJ
41
Bad Smell Examples


We should refactor any time we detect a “bad smell” in
the code
Examples of bad smells include:








Duplicate Code
Long Methods
Large Classes
Long Parameter Lists
Multi location code changes
Feature Envy
Data Clumps
Primitive Obsession
42
Structure of Observer Pattern
Subject
for all observers obs
{
obs->update()
}
+notify()
+attach(in Observer)
+detach(in Observer)
Observer
*
+update()
1
ConcreteObserver
ConcreteSubject
-subjectSate
+getState()
return subjectState
*
+update()
1
observerState = subject->getState()
Three Types of Patterns

Creational patterns:


Structural patterns:



Deal with initializing and configuring classes and objects
Deal with decoupling interface and implementation of classes and
objects
Composition of classes or objects
Behavioral patterns:


Deal with dynamic interactions among societies of classes and
objects
How they distribute responsibility
Creational Patterns





Abstract Factory:
 Factory for building related objects
Builder:
 Factory for building complex objects incrementally
Factory Method:
 Method in a derived class creates associates
Prototype:
 Factory for cloning new instances from a prototype
Singleton:
 Factory for a singular (sole) instance
Structural patterns

Describe ways to assemble objects to realize new
functionality



Added flexibility inherent in object composition due to ability to
change composition at run-time
not possible with static class composition
Example: Proxy

Proxy: acts as convenient surrogate or placeholder for another
object.
 Remote Proxy: local representative for object in a different address
space
 Virtual Proxy: represent large object that should be loaded on
demand
 Protected Proxy: protect access to the original object
Structural Patterns







Adapter:
 Translator adapts a server interface for a client
Bridge:
 Abstraction for binding one of many implementations
Composite:
 Structure for building recursive aggregations
Decorator:
 Decorator extends an object transparently
Facade:
 Simplifies the interface for a subsystem
Flyweight:
 Many fine-grained objects shared efficiently.
Proxy:
 One object approximates another
Behavioral Patterns






Chain of Responsibility:
 Request delegated to the responsible service provider
Command:
 Request or Action is first-class object, hence re-storable
Iterator:
 Aggregate and access elements sequentially
Interpreter:
 Language interpreter for a small grammar
Mediator:
 Coordinates interactions between its associates
Memento:
 Snapshot captures and restores object states privately
Which ones do you think you have seen somewhere?
Behavioral Patterns (cont.)





Observer:
 Dependents update automatically when subject changes
State:
 Object whose behavior depends on its state
Strategy:
 Abstraction for selecting one of many algorithms
Template Method:
 Algorithm with some steps supplied by a derived class
Visitor:
 Operations applied to elements of a heterogeneous object structure
The End
50
Download