diff options
Diffstat (limited to 'LectureNotesOnOOAD.tex')
| -rw-r--r-- | LectureNotesOnOOAD.tex | 1725 |
1 files changed, 1725 insertions, 0 deletions
diff --git a/LectureNotesOnOOAD.tex b/LectureNotesOnOOAD.tex new file mode 100644 index 0000000..5a5279d --- /dev/null +++ b/LectureNotesOnOOAD.tex @@ -0,0 +1,1725 @@ +\documentclass[landscape,25pt]{foils}
+\usepackage{amsmath}
+\usepackage{amssymb}
+\usepackage{amsthm}
+\usepackage{color}
+\usepackage{subfigure}
+\usepackage{graphicx}
+\usepackage{hyperref}
+\usepackage{textcomp}
+\usepackage{listings}
+\lstset{
+ language=Python,
+ basicstyle=\footnotesize,
+ frame=single,
+ showstringspaces=false,
+ formfeed=newpage,
+ tabsize=4,
+ commentstyle=\it,
+ morekeywords={models, lambda, forms}
+}
+
+
+\title{Introduction to Object-oriented Analysis and Design}
+
+\author{}
+\date{}
+\MyLogo{Copyright \copyright 2018, 2019 Hui Lan}
+
+\begin{document}
+\maketitle
+
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Course information (Fall 2019)}
+
+
+This full-semester course introduces terminology of the object-oriented (OO) \underline{paradigm}, such as classes, superclasses (or parent classes), subclasses (or child classes), class variables, objects, encapsulation, abstraction, simple inheritance, multiple inheritance, polymorphism, duck-typing, exceptions, and abstract base classes. The students will learn the most useful tools in the object-oriented principles by practicing OO analysis, OO design and OO programming (using python 3) for an existing open publishing service. The object-oriented programming paradigm may at first appear quite strange for people familiar with the procedural paradigm, but can turn out to be quite natural for many problems. This course aims to train students to view these problems and come up with solutions in an object-oriented fashion. Previous programming experience, though desirable, is not required. This is a rather {\em practical} course, in which concepts introduced in lectures will be soon applied in the labs as well as in the course project. Finally, we will analyze whether several claimed benefits of the OO paradigm, such as reliability, productivity, reuse, ease of modification, indeed exist.
+
+{\bf Paradigm}: a pattern or model.
+
+{\bf Recommended textbook}: Dusty Phillips. (2015) Python 3 Object Oriented Programming. Second Edition.
+
+A significant portion of the lecture notes is built on the above book.
+
+{\bf Recommended reference book}: Craig Larman. (2004) Applying UML and Patterns: An Introduction to Object-Oriented Analysis and Design and Iterative Development. Third Edition.
+
+
+
+{\bf Grading policy}:
+
+\begin{tabular}{ | l | l |}
+ \hline
+ {\bf Component} & {\bf Weight} \\ \hline
+ Quizzes & 10 \\ \hline
+ Labs & 20 \\ \hline
+ Course Project & 20 \\ \hline
+ Final exam & 50 \\ \hline
+\end{tabular}
+
+
+{\bf Software freely available for this course}:
+
+\begin{itemize}
+\item Python 3.7.0 interpreter:
+
+ https://www.python.org/downloads/release/python-370/
+
+ [scroll to bottom to find a installer for your operating system.]
+
+
+\item Wing IDE 101: http://wingware.com/downloads/wing-101
+
+\item Flask: https://pypi.org/project/Flask/
+\end{itemize}
+
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Essential Elements of Object-oriented Analysis and Design}
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Object}
+
+Object - a physical thing that we can sense, feel and manipulate (e.g., wallets, toys, babies, apples, oranges, etc).
+
+Put simply, an object is a tangible thing.
+
+It does not have to be physical.
+
+\begin{itemize}
+\item A {\bf file} can be an object.
+\item An {\bf operating system} can be an object.
+\item A {\bf job position} can be an object.
+\end{itemize}
+
+An object has {\bf data} and {\bf behaviors} (usually).
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Oriented}
+
+Orient's dictionary definition. 1. align or position (something) relative to the points of a compass or other specified positions. Find one's position in relation to new and strange surroundings. 2. adjust or tailor (something) to specified circumstances or needs.
+
+Oriented - {\em directed towards}.
+
+Object-oriented. Specify the style of software development: we are going to {\bf model objects and their interactions} (if any). Functionally directed toward modeling objects.
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{The object-oriented umbrella}
+
+Object-oriented umbrella - OO analysis, OO design, and OO programming.
+
+As we have learned from our {\em Software Engineering} course, design happens before programming, and analysis before design.
+
+In reality, this order is not that strict.
+
+In fact, iteration is common.
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Object-oriented analysis}
+
+{\em What}?
+
+DO THE RIGHT THING.
+
+Analyze the requirements first. Work product: {\bf use cases}.
+
+Understand the problem. Happen after the Software Requirements stage.
+
+Understand what needs to be done (and what needs {\em not} to be done), by looking at a task and identifying objects and their interactions.
+
+Object-oriented exploration - interview customers, study their processes, and eliminate possibilities (i.e., define scope).
+
+
+
+For example, an online food store:
+
+\begin{itemize}
+\item {\em review} our {\bf history}
+\item {\em apply} for {\bf jobs}
+\item {\em browse}, {\em compare}, and {\em order} {\bf products}
+\end{itemize}
+
+
+Tasks:
+
+\begin{itemize}
+\item Identify objects, e.g., Plane, Flight, Pilot.
+\item Organize the objects by creating a object model diagram.
+\item Describe how the objects interact/collaborate.
+\end{itemize}
+
+\href{https://www.tutorialspoint.com/object\_oriented\_analysis\_design/ooad\_tutorial.pdf}{A useful tutorial on OOAD}
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Object-oriented design}
+
+
+{\em How}?
+
+DO THE THING RIGHT.
+
+
+
+Present a conceptual solution.
+
+Figure out {\em how} things should be done.
+
+More specifically, {\bf assign responsibilities} to classes and collaborate objects.
+
+{\bf Responsibility assignment} is an important skill we need to master.
+
+{\bf Responsibility-driven design}.
+
+Convert analysis of requirements into implementation specification (classes and interfaces).
+
+Tasks:
+
+\begin{itemize}
+\item Name the objects.
+\item Specify which objects interact with other objects.
+\item Identify constraints.
+\item Define object attributes -- data, e.g., tailNumber in Plane.
+\item Define object actions -- behaviors, e.g., getFlightHistory.
+\end{itemize}
+
+
+Class Responsibility Collaborator Models. You should be able to draw CRC cards and move them around on a table. A design technique. An effective tool for conceptual modeling and detailed design.
+
+\begin{itemize}
+\item Class - class name (a singular noun)
+\item Responsibility - what the class {\bf knows} or {\bf does}.
+\item Collaborator - collaborate/interact with other classes. Need help from other classes. For example, Seminar could be Student's collaborator for checking space availability and for enrollment.
+\end{itemize}
+
+http://www.agilemodeling.com/artifacts/crcModel.htm
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Object-oriented programming}
+
+Smalltalk, Java, Python.
+
+
+Convert design into a working program using an OO programming language.
+
+Smalltalk (too ancient), C++ (too complex), Java (too verbose), Python (just right).
+
+Some classic designs are called {\bf design patterns}.
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{The real world is MURKY}
+
+Murky dictionary definition: dark and gloomy, especially due to thick mist. Not fully explained or understood.
+
+No matter how hard we try to separate these stages (OOA, OOD and OOP), we will always
+find things that need further analysis while we are designing. When we are programming, we find features
+that need clarification in the design.
+
+Remember the Agile Process we've talked in our Software Engineering course? A series of short development cycles.
+
+{\bf Development is iterative}. How iterative it is depends on the team's experience, and the completeness of requirements.
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Iterative and incremental development}
+
+IID - iterative and incremental development, dates back to late 1950s.
+
+45\% of the features in Waterfall requirements are never used.
+
+A typical software project experienced a 25\% change in requirements [Boehm and Papaccio].
+
+Work on a series of {\bf mini-projects} to get a series of {\bf partial systems}, each being a growing subset of the final system.
+
+1. Get some requirements.
+
+2. Plan.
+
+3. {\bf Build} a partial system.
+
+4. Ask for {\bf feedback} (requirements clarification, marketplace change).
+
+5. Analyze and {\bf incorporate} feedback.
+
+Repeat 2-5 for about 10-15 iterations.
+
+See picture iterative-development.png
+
+UP - Unified Process (Extreme Programming - Test-driven development and pair programming, Risk-driven development, Client-driven development, Scrum - war room, daily stand-up meeting, three special questions to be answered by each team member, Refactoring and CI)
+
+
+
+Monday: one-hour meeting, review last iteration's diagrams, whiteboards, pseudocode and design notes. No rush to code.
+
+No overly-detailed design.
+
+Short {\bf timeboxed} iterations are preferred. Why?
+
+{\bf Agile methods}: Each iteration refines requirements, plans and design. {\em Whatever works. }
+
+3 weeks better than 6 weeks. Why?
+
+Too much work in the current iteration? De-scope iteration goals.
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Waterfall thinking}
+
+
+Up-front analysis and modeling.
+
+BUT software development is a {\bf high-change} area.
+
+Figure 2.3 Larman's book PDF page 63.
+
+``Let's write all the use cases before starting to program.''
+
+``Let's do many detailed OO models in UML before starting to program.''
+
+
+Consequences: 45\% of the features in waterfall requirements are never
+used, and early waterfall schedules and estimates vary up to 400\%
+from the final actuals.
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Evolutionary analysis and design}
+
+For example, a project needs 20 iterations, each timeboxed in 3 weeks.
+
+The first 5 iterations include requirements {\bf workshops} and
+critical prototypes, stabilizing 90\% of the requirements, but only
+completing 10\% of the whole software. That is, about 20\% of the
+whole project time is devoted to requirements.
+
+Figure 2.4 Larman's book PDF page 69.
+
+Starting coding near Day One of the project is also bad. We need a middle way.
+
+Iterative and evolutionary requirements analysis combined with early {\bf timeboxed iterative development} and {\bf frequent stakeholder participation}, evaluation, and {\bf feedback} on partial results.
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Agile modeling}
+
+The purpose of modeling (sketching UML, ...) is primarily to understand, not to document.
+
+Treat it lightly.
+
+Doing UML is not to create many detailed UML diagrams for programmers,
+but to quickly explore alternatives and paths to a good OO design.
+
+Misconception: people translate UML diagrams mechanically to code.
+
+Prefer sketching UML on whiteboards, and taking a picture for it.
+
+Don't do it alone.
+
+``Good enough'', simple notations.
+
+Quick creative flow and change.
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Agile Unified Process}
+
+Phase plan - high-level
+
+Iteration plan - detailed, adaptive
+
+Tackle high-risk and high-value issues in early iterations.
+
+Engage users for evaluation, feedback and requirements.
+
+Test early, often, and realistically.
+
+Manage requirements.
+
+{\bf Artifacts}
+
+Domain model - noteworthy concepts in the application domain
+
+Use-Case model and Supplementary Specification - functional and non-functional requirements. Glossary. Vision. Business Rules.
+
+Design model - objects.
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Objects and classes}
+
+A class usually has {\bf attributes} and {\bf behaviors}.
+
+Kind of objects is {\bf class}. In the class, we summarize their common attributes (the attributes we are interested in).
+
+Classes describe objects. A class definition is like a {\bf blueprint} for creating objects.
+
+This orange of 50 grams belongs to \texttt{Orange} class, and that orange of 100 grams belongs to \texttt{Orange} class. The weight is one of the {\em infinitely many} characteristics that are shared by all oranges. These infinitely many characteristics include farm, pick date, best before date, etc. But any application needs only a finite set of characteristics, we must {\bf ignore irrelevant characteristics} while modeling. This selecting and ignoring process is called ABSTRACTION.
+
+An object is called an {\bf instance} of a class. This object instance has its own set of data and behaviors.
+
+We can make arbitrary number of objects from a class.
+
+An Orange class may have three attributes: weight, orchard and date picked. You can use it to
+describe/represent real oranges sold in the market, each having different weight, orchards and pick dates.
+
+\lstinputlisting{./oop_prep/Orange.py}
+
+Use \texttt{cheap\_orange.\_\_dict\_\_} to show attributes and their values. Try also \texttt{Orange.\_\_dict\_\_}.
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{UML - Unified Modeling Language}
+
+Boxes and lines to intuitively illustrate classes and the association between them.
+
+A useful communication tool during OO analysis and design.
+
+A useful reference for myself in the future. {\em Why did I do that?}
+
+{\bf Caution}: best used only when needed. Don't be lost among UML details. Hide uninteresting details.
+
+{\bf Reality}: the initial diagrams become outdated very soon. Because these diagrams are subject to change in subsequent iterations, some people think drawing UML class diagrams is a waste of time (if you spend too much time on it). Don't make it too formal in the beginning.
+
+Most useful diagrams: class diagrams, use case diagrams, activity diagrams, and sequence diagrams.
+
+\begin{itemize}
+
+\item Class diagrams. A box represents a class. A line between two boxes represents a relationship.
+
+\item Sequence diagrams. Model the interactions (step-by-step) among objects in a Use Case. Vertical lifeline (the dashed line hanging from each object), activation bars, horizontal arrows (messages, methods, return values).
+
+\item Use case diagrams.
+
+\item Activity diagrams.
+%https://creately.com/blog/diagrams/sequence-diagram-tutorial/
+
+\end{itemize}
+
+
+%%% More stuff on UML diagrams
+%%% https://holub.com/uml/
+%%% http://agilemodeling.com/essays/umlDiagrams.htm
+%%% https://www.pl.cs.jhu.edu/oose/lectures/uml.shtml
+%%% http://www.agilemodeling.com/artifacts/useCaseDiagram.htm
+%%% http://www.agilemodeling.com/artifacts/sequenceDiagram.htm
+%%% http://www.agilemodeling.com/artifacts/classDiagram.htm
+%%% https://www.tutorialspoint.com/uml/uml_standard_diagrams.htm
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Useful names and resources}
+
+{\bf Martin Fowler} - UML Distilled, Refactoring.
+
+{\bf Scott Ambler} - \href{http://agilemodeling.com/essays/agileRequirements.htm}{Agile Modeling}.
+
+
+{\bf Gang of Four} - Design Patterns. Not an introductory book.
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Procedural versus object-oriented}
+
+A wallet for money deposit and withdrawal.
+
+Procedural:
+
+\lstinputlisting{./oop_prep/money_procedural.py}
+
+Problem: the above code only works for a {\em single} wallet. We can add a \texttt{create} function which returns a dictionary (e.g., \{'money':0\}) for each wallet created.
+
+Procedural:
+
+\lstinputlisting{./oop_prep/money_dictionary.py}
+
+Object-oriented:
+
+{\bf Think in Objects}
+
+\lstinputlisting{./oop_prep/money_oo.py}
+
+\begin{lstlisting}
+
+ w = Wallet(0)
+ w.check()
+ 'The wallet has 0 RMB.'
+
+ w.deposit(1000)
+ w.check()
+ 'The wallet has 1000 RMB.'
+
+ w.withdraw(500)
+ w.check()
+ 'The wallet has 500 RMB.'
+
+\end{lstlisting}
+
+
+%\begin{center}
+%\includegraphics[width=0.85\textwidth]{fig/BoehmCurve_real}
+%\end{center}
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Chapter 1 Object-oriented design}
+
+Abstraction
+
+Classes
+
+Encapsulation
+
+Inheritance
+
+UML
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Example: Dice Game}
+
+
+Roll two dices and check whether their face values sum to 7.
+
+
+{\bf Use case}: Player rolls the dice. System returns results. If the dice face values totals 7, play wins; otherwise, player loses.
+
+{\bf Domain model, or conceptual object model}
+
+{\bf Interaction diagrams} We could use a sequence diagram: flow of message, invocation of methods.
+
+{\bf Class diagrams}.
+
+Visual Modeling is a good thing. Use UML as a sketch. Don't let too many uninteresting details get in the way.
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Everything in Python is a class}
+
+We have seen a few examples of customary classes. Note that the internal python data types such as numbers, strings, modules, and even functions, are classes too.
+
+\begin{lstlisting}
+>>> import random
+>>> type(random)
+ <class 'module'>
+
+>>> x = 123
+>>> type(x)
+ <class 'int'>
+
+>>> x = '123'
+>>> type(x)
+ <class 'str'>
+
+>>> x = [1,2,3]
+>>> type(x)
+ <class 'list'>
+
+>>> def f():
+ pass
+>>> type(f)
+ <class 'function'>
+
+>>> class Minimalism:
+ pass
+
+>>> x = Minimalism()
+>>> type(x)
+ <class '__main__.Minimalism'>
+\end{lstlisting}
+
+I fail to see why not everything in the world cannot be described as a class.
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Specifying attributes and behaviors}
+
+Objects are {\bf instances} of classes.
+
+Each object of a class has its {\bf own} set of data, and methods
+dealing with these data. With OOP, we in principle don't access these
+class attributes directly, but {\em only} via class methods.
+
+\begin{itemize}
+\item Data describe objects.
+
+ Data represents the individual characteristics of a certain object.
+
+ Attributes - values
+
+ All objects instantiated from a class have the same attributes but may have different values.
+
+ Attributes are sometimes called members or properties (usually read-only).
+
+\item Behaviors are actions.
+
+ Behaviors are actions ({\bf methods}) that can occur on an object.
+
+ We can think of methods as functions which have access to all the data associated with this object.
+
+ Methods accept parameters and return values.
+
+\end{itemize}
+
+OOA and OOD are all about identifying objects and specifying their interactions.
+
+%\begin{center}
+%\includegraphics[width=0.98\textwidth]{fig/classes_publisher}
+%\end{center}
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Interacting objects}
+
+How do we make object interact? Pass objects as arguments to object methods.
+
+\lstinputlisting{./oop_prep/orange_basket.py}
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Composition and aggregation}
+
+
+{\bf Composition}: collecting several objects to make a new one.
+
+{\bf Aggregation} is closely related to composition. Main difference: the aggregate objects may exist independently (they won't be
+destroyed after the container object is gone).
+
+Composite and aggregate objects have different lifespan.
+
+The main difference is whether the child could {\bf exist independently} from the parent.
+
+A class has a list of students. \texttt{Students} could exist independently from \texttt{Class}. Aggregation.
+
+A house has a number of rooms. \texttt{Room} could not exist independently from \texttt{House}. Composition.
+
+The difference is not very important in practice.
+
+\begin{itemize}
+\item A car is \underline{composed of} an engine, transmission, starter,
+headlights and windshield. The engine comprises many parts. We can
+decompose the parts further if needed. {\em has a} relationship.
+
+\item How about abstract components? For example, names, titles, accounts,
+appointments and payments.
+\end{itemize}
+
+Model chess game.
+
+Two {\bf player}s - a player may be a human or a computer.
+
+One {\bf chess set} - a {\bf board} with 64 {\bf
+ positions}, 32 {\bf piece}s including pawns, rooks, bishops,
+knights, king and queen). Each piece has a shape and a unique move rule.
+
+The pieces have an aggregate relationship with the chess set. If the board is destroyed, we can still use the pieces.
+
+The positions have a composite relationship with the chess set. If the board is destroyed, we cannot re-use positions anymore (because positions are part of the board).
+
+Man - Leg - Shoes
+
+\lstinputlisting{./oop_prep/aggregation_vs_composition.py}
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Simple inheritance}
+
+This is the {\bf most famous} object-oriented principle.
+
+For creating {\em is a } relationship.
+
+Abstract common logic into superclasses and manage specific details in the subclass.
+
+Queen {\em is a} Piece.
+
+So are Pawns, Bishops, Rooks, Knights and King.
+
+This is inheritance. Inheritance is useful for sharing code (and for avoiding duplicate code).
+
+Everything in python is inherited (derived) from the most {\em base} class, {\bf object}.
+
+Check the output of \texttt{help(object)} and \texttt{dir(object)}.
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Overriding methods}
+
+Re-defining a method of the superclass (the method name unchanged) in the subclass. We can override special methods (such as \texttt{\_\_init\_\_}, \texttt{\_\_str\_\_}) too.
+
+\lstinputlisting{./oop_prep/Piece.py}
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{super()}
+
+The super() function returns an object instantiated from the parent class, allowing us to call the parent methods directly.
+
+The super() function can be called anywhere in any method in the subclass.
+
+\lstinputlisting{./oop_prep/Friend.py}
+
+It makes sense to order something from a supplier (Supplier) but not from my friends (Friend). So Supplier has the method order() while Friend does not have this method, although both subclasses are derived from the same parent class, Contact.
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Class variables}
+
+In class \texttt{Contact}, \texttt{all\_contacts} is a {\bf class variable}.
+
+What is special about the class variable? It is shared by all instances of this class.
+
+In the above example, whenever we create an object (from Contact, Friend, or Supplier), this object is appended to \texttt{all\_contacts}.
+
+We access the class variable via: \texttt{Contact.all\_contacts}, \texttt{Friend.all\_contacts}, or \texttt{f.all\_contacts}.
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Extending built-ins}
+
+\begin{itemize}
+
+ \item Add a search method to the built-in type \texttt{list}.
+
+\lstinputlisting{./oop_prep/ContactList.py}
+
+In fact, [] is {\bf syntax sugar} for list().
+
+\item Add a longest\_key method to the built-in type \texttt{dict}.
+
+ \lstinputlisting{./oop_prep/ContactList.py}
+\end{itemize}
+
+%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%% \foilhead{Keyword arguments - kwargs}
+
+%% **kwargs is a dictionary that collects any keyword arguments that have not been consumed in the parameter list.
+
+%% \lstinputlisting{./oop_prep/kwargs.py}
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Polymorphism}
+
+Polymorphism - many forms of (a function).
+
+A fancy name.
+
+Treat a class differently depending on which {\em subclass} is implemented.
+
+Different behaviors happen depending on which {\em subclass} is being used, without having to explicitly know what the subclass actually is.
+
+Mostly talking about method overriding. A concept based on inheritance.
+
+Same method name, but different actions (function definitions).
+
+\lstinputlisting{./oop_prep/Polymorphism2.py}
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Duck-typing}
+
+
+\lstinputlisting{./oop_prep/ducktyping.py}
+
+This sort of polymorphism in Python is typically called {\bf ducking typing}: ``If it walks like a duck or swims like a duck, it is a duck''.
+
+No inheritance is involved. We don't really care if it really {\em is a} duck object, as long as it can fly (i.e., has the \texttt{fly()} method).
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Multiple inheritance}
+
+A subclass inherits from more than one superclasses.
+
+\begin{lstlisting}
+class RicohAficio(Copier, Printer, Scanner, Faxer):
+ pass
+\end{lstlisting}
+
+
+Not used very often as it can accidentally create the {\bf Diamond Problem} (or Diamond Inheritance): ambiguity in deciding which parent method (with the same name) to use.
+
+\lstinputlisting{./oop_prep/MRO.py}
+
+
+One potential consequence of the diamond problem is that the base class can be called twice. The following code demonstrates that.
+
+\lstinputlisting{./oop_prep/MRO2.py}
+
+Therefore, to avoid that, we need {\bf Method Resolution Order} (MRO) (with super()).
+\lstinputlisting{./oop_prep/MRO1.py}
+
+``next'' method versus ``parent'' method.
+
+Many expert programmers recommend against using it because it will make our code messy and hard to debug. Alternative: composition, instead of inheritance. Include an object from the superclass and use the methods in that object.
+
+{\bf mixin}. A mixin is a superclass that provides extra functionality.
+
+\lstinputlisting{./oop_prep/EmailableContact.py}
+
+In the above example, MailSender is a mixin superclass.
+
+\lstinputlisting{./oop_prep/AddressHolder.py}
+
+
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%\foilhead{Instructors - Students - Courses}
+
+%\lstinputlisting{./oop_prep/teaching.py}
+
+
+
+%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%% \foilhead{Library catalog}
+
+%% \lstinputlisting{./oop_prep/catalog.py}
+
+
+%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%% \foilhead{A shopping mall}
+
+%% What common objects (classes) can we identify in a shopping mall (e.g., Yorkdale, Eaton, Wanda)? Nouns (class names, class attributes) and verbs (class methods).
+%% When can composition or inheritance be used?
+
+%% \begin{enumerate}
+%% \item Mall name
+%% \item Location
+%% \item Transit access
+%% \item Opening date
+%% \item Owner
+%% \item Number of visitors
+%% \item Architect
+%% \item Hours, holiday hours
+%% \item Total area
+%% \item Entrances (main, side)
+%% \item Underground, parking garages(parking)
+%% \item Fire exits (doors)
+%% \item Escalators
+%% \item Washrooms
+%% \item Cinema
+%% \item Bank machines (ATMs)
+%% \item Floor maps, floors
+%% \item Mall map
+%% \item Stores may have sales, promotions
+%% \item News
+%% \item Events
+%% \item Customer services
+%% \item Tourist information
+%% \item Shop card
+%% \item Social media
+%% \item Number of stores
+%% \item Books and Music store
+%% \item Food courts (fast food, restaurants)
+%% \item Food stores, Apparel stores (men's, ladies, unisex, children), kitchen, housewares, pharmacies and drug stores, footwear stores, Athleticwear stores (asics, adidas, nike), leather and luggage stores, gift stores, stationary stores, health and beauty stores, toys stores, jewelery stores, the apple store (electronics, computers and telephones), the Microsoft store, coffee shops
+%% \item Customers, kids, Sales representatives, building staff (security guards, cleaners), cashiers, pharmacist, optometrist
+%% \item Products and their associated information
+%% \item Interactions between customers and store sales persons.
+%% \end{enumerate}
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Hiding details and creating public interface}
+
+Determine the {\bf public interface}. Make it stable.
+
+Interface: the collection of attributes and methods that other objects can use to interact with that object.
+
+As class users/clients, it is good enough to just know the interface (API documentation) without needing to worry about its internal workings. As class designers/programmers, they should keep the interface stable while making changes to its internals so that users' code can still work (without modification).
+
+
+The remote control is our interface to the Television. Each button is like a method that can be called on the TV.
+
+We don't care:
+
+\begin{itemize}
+\item Signal transmission from antenna/cable/satellite
+\item How the signals are converted to pictures and sound.
+\item Signal sent to adjust the volume
+\end{itemize}
+
+Vendor machines, %TVs (remote control),
+cellphones, Microwaves, Cars, and Jets.
+
+{\bf Information hiding:} the process of hiding functional details. Sometimes loosely called {\bf encapsulation}.
+
+In python, we don't have or need {\em true} information hiding.
+
+
+
+We should focus on the level of detail most appropriate to a given task and ignore irrelevant details while designing a class. This is called {\bf abstraction}.
+
+{\bf Abstraction} is an object-oriented principle related information hiding and encapsulation. Abstraction is the process of encapsulating information with separate
+public and private interfaces. The private information can be subject to information hiding.
+
+A car driver has a different task domain from a car mechanic.
+
+Driver needs access to brakes, gas pedal and should be able to steer, change gears and apply brake.
+
+Mechanic needs access to disc brakes, fuel injected engine, automatic transmission and should be able to adjust brake and change oil.
+
+So a car can have different abstraction levels, depending on who operates it.
+
+
+Design tips:
+
+\begin{itemize}
+\item Keep the interface simple.
+\item When abstracting interfaces, model exactly what needs to be modeled and nothing more.
+\item Imagine that the object has a strong preference for privacy.
+\end{itemize}
+
+%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%% \foilhead{Abstraction, encapsulation and information hiding}
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Packages, modules, classes and methods}
+
+Usually, methods are organized in a class, classes in a module (a file), and modules in a package.
+
+{\bf A module is a file} containing class/function definitions.
+
+For small projects, just put all classes in one file. So we got only one module. For example, Lab3.py.
+
+
+A package is a folder containing a few modules. We must create an empty \texttt{\_\_init\_\_} file under that folder to make the folder a package.
+
+
+There are two options for importing {\bf modules}: import and from-import.
+
+Use simple imports if the imported modules are under the same folder as the importing file, or in system path.
+
+\begin{lstlisting}
+import database
+
+db = database.Database()
+
+
+from database import Database
+db = Database()
+
+
+from database import Database as DB
+db = DB()
+\end{lstlisting}
+
+
+\begin{itemize}
+
+ \item import
+
+\begin{lstlisting}
+import package.module # use period operator to separate packages or modules.
+
+c = package.module.UserClass()
+\end{lstlisting}
+
+For example,
+
+\begin{lstlisting}
+import math
+
+math.sqrt(4)
+\end{lstlisting}
+
+Can you do \texttt{import math.sqrt}? No. sqrt is not a module.
+
+We can do \texttt{from math import sqrt}.
+
+\item from-import, or from-import-as.
+
+\begin{lstlisting}
+from package.module import UserClass
+
+c = UserClass()
+\end{lstlisting}
+
+For example,
+
+\begin{lstlisting}
+from math import sqrt
+
+sqrt(4)
+
+\end{lstlisting}
+
+
+\end{itemize}
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Organizing modules to packages and properly importing them}
+
+\begin{lstlisting}
+ proj/
+ main.py
+ ecommerce/
+ __init__.py
+ database.py
+ products.py
+ payments/
+ __init__.py
+ paypal.py
+ creditcard.py
+\end{lstlisting}
+
+
+How to use the module paypal.py in main.py? Use {\bf absolute imports}, which specify the complete path.
+
+Each module in the package can use {\bf absolute imports} too. (But it won't work if we want to run this module as main program. One solution is move the whole package to a system path called \texttt{site-packages}. We can get all system paths using \texttt{sys.path}.)
+
+Module-level code will be executed immediately when the module is imported.
+
+\begin{lstlisting}
+import ecommerce.payments.paypal
+ecommerce.payments.paypal.pay()
+
+from ecommerce.payments.paypal import pay
+pay()
+
+from ecommerce.payments import paypal
+paypal.pay()
+\end{lstlisting}
+
+
+main.py:
+
+\lstinputlisting{./oop_prep/main.py}
+
+ecommerce/products.py:
+
+\lstinputlisting{./oop_prep/ecommerce/products.py}
+
+ecommerce/payments/paypal.py:
+
+\lstinputlisting{./oop_prep/ecommerce/payments/paypal.py}
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Each module has a special, hidden variable called \texttt{\_\_name\_\_}}
+
+Use \texttt{print(\_\_name\_\_)} to check its value.
+
+If you run that module as a main program (note that each module is a file), then \texttt{\_\_name\_\_} is equal to \texttt{'\_\_main\_\_'}.
+
+We say running a module as a main program if we type in the command line like this: \texttt{python module\_name.py}.
+
+If you import that module, then that module's \texttt{\_\_name\_\_} contains its actual file name. For example, \texttt{paypal.py}'s \texttt{\_\_name\_\_} is ``\texttt{ecommerce.payments.paypal}'' when we import the module \texttt{paypal.py} using \texttt{from ecommerce.payments import paypal}.
+
+When we import a module, the module's code will be executed. But since that module's \texttt{\_\_name\_\_} is not \texttt{'\_\_main\_\_'}, then the code under that module's \texttt{if \_\_name\_\_ == '\_\_main\_\_':} won't be executed.
+
+So it is a good idea to put \texttt{if \_\_name\_\_ == '\_\_main\_\_':} in the end of every module (main module or not) and put the test code for that module after it.
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Organizing module contents}
+
+A typical order in a Python module:
+
+\begin{lstlisting}
+ class UsefulClass:
+ ''' This class might be useful to other modules.'''
+ pass
+
+
+ def main():
+ ''' Do something with it for our module. '''
+ u = UsefulClass()
+ print(u)
+
+ if __name__ == '__main__':
+ main()
+\end{lstlisting}
+
+
+Inner classes and inner functions. Usually used as an {\em one-off} helper, not to be used by other methods or other modules.
+
+\lstinputlisting{./oop_prep/formatter_string.py}
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Public, protected and private attributes}
+
+
+\lstinputlisting{./oop_prep/wallet.py}
+
+Attribute names with two underscores are not visible.
+
+\lstinputlisting{./oop_prep/wallet_test.py}
+
+%\texttt{builtins.AttributeError: 'Wallet' object has no attribute '\_\_gbp'}
+
+%% \begin{lstinputlisting}
+%% >>> Wallet.__dict__
+%% mappingproxy({'__module__': 'wallet', '__doc__': ' For demostrating public, protected and private attributes. ', '__init__': <function Wallet.__init__ at 0x0000000002A95598>, 'deposit': <function Wallet.deposit at 0x0000000002A95488>, 'withdraw': <function Wallet.withdraw at 0x0000000002A95400>, 'check': <function Wallet.check at 0x0000000002A95F28>, '__dict__': <attribute '__dict__' of 'Wallet' objects>, '__weakref__': <attribute '__weakref__' of 'Wallet' objects>})
+%% >>> w.__dict__
+%% {'rmb': 0, '_cad': 0, '_Wallet__gbp': 0}
+%% \end{lstinputlisting}
+
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Abstract methods and interfaces}
+
+No \texttt{interface} keyword in Python. We use ABCs (Abstract Base Classes) instead.
+
+All subclasses derived from an abstract base class {\bf must implement} the abstract methods (marked by @abstractmethod). This is forced. It is like a contract between class users and class implementers.
+
+We cannot instantiate an abstract base class. We cannot instantiate a subclass of abstract class without defining all its abstract methods.
+
+
+Specify method names in abstract class, and implement these methods in subclasses.
+
+\lstinputlisting{./oop_prep/AbstractBaseClass2.py}
+
+Duck-typing and isinstance.
+
+\lstinputlisting{./oop_prep/ducktyping2.py}
+
+@ is called decorator.
+
+%\texttt{cls} is class name, and \textt{C} subclass name.
+
+%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%% \foilhead{The iterator design pattern}
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Expecting the Unexpected}
+
+An exception is not expected to happen often. Use exception handling for really exceptional cases.
+
+Cleaner code; more efficient.
+
+{\bf Look before you leap.} We can use the \texttt{if-elif-else} clause, why do we bother with exception?
+
+{\bf Ask forgiveness rather than permission.} Exception is not a bad thing, not something to avoid. It is a powerful way to communicate information (pass messages).
+
+What does an exception class look like?
+
+\begin{lstlisting}
+class IndexError(LookupError)
+ | Sequence index out of range.
+ |
+ | Method resolution order:
+ | IndexError
+ | LookupError
+ | Exception
+ | BaseException
+ | object
+\end{lstlisting}
+
+
+The exception hierarchy.
+
+\begin{lstlisting}
+
+ BaseException
+ ^
+ SystemExit KeyboardInterrupt Exception
+ ^
+ Most Other Exception
+
+\end{lstlisting}
+
+
+Other built-in errors: ValueError, TypeError, KeyError, ZeroDivisionError and AttributeError.
+
+\lstinputlisting{./oop_prep/EvenOnly.py}
+
+try-except:
+\lstinputlisting{./oop_prep/EvenOnly2.py}
+
+\lstinputlisting{./oop_prep/IndexErrorException.py}
+
+We can omit ``Exception as e'' or use LookupError or IndexError instead. Using IndexError is best as we are explicit here which exception we want to catch (and then handle).
+
+\lstinputlisting{./oop_prep/noreturn.py}
+
+Good floor:
+\lstinputlisting{./oop_prep/goodfloor.py}
+
+Stack exception clauses.
+
+\lstinputlisting{./oop_prep/funnydivision.py}
+
+try-except-else-finally:
+
+\lstinputlisting{./oop_prep/IndexErrorException2.py}
+
+\lstinputlisting{./oop_prep/finally.py}
+
+Things under \texttt{finally} will executed no matter what happens (a good place to put clean-up statements). Extremely useful for
+\begin{itemize}
+\item Cleaning up an open database connection
+\item Closing an open file
+\end{itemize}
+We can use try-finally without the except clause.
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Customized exceptions}
+
+Inherit from class Exception. Add information to the exception.
+
+\lstinputlisting{./oop_prep/InvalidWithdrawal.py}
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Getters, setters and @property decorator}
+
+Data encapsulation.
+
+Make methods look like attributes.
+
+``Blurring the distinction between behavior and data''.
+
+No change to client code. Backward compatible.
+
+The \texttt{Color} example, starting from page 130 in Dusty's book.
+
+\lstinputlisting{./oop_prep/ManHeight.py}
+% https://www.programiz.com/python-programming/property
+% https://www.programiz.com/python-programming/decorator
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{More on @property}
+
+property is in fact a special function that returns a property object.
+
+\begin{lstlisting}
+ property(fget=None, fset=None, fdel=None, doc=None)
+\end{lstlisting}
+
+
+\begin{lstlisting}
+ p = property()
+ dir(p) # attributes [..., 'fdel', 'fget', 'fset', 'getter', 'setter']
+ help(p)
+\end{lstlisting}
+
+\begin{lstlisting}
+ Man2.height # <property object at 0x0000000002A2FE58>
+ Man2.height.fget(m)
+ Man2.height.fset(m, 123)
+ Man2.height.__getattribute__('getter')
+ Man2.height.__getattribute__('settter')
+\end{lstlisting}
+
+\lstinputlisting{./oop_prep/property_example.py}
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{More on decorators}
+
+{\bf A decorator is a function} which adds some toppings to the decorated function.
+
+\lstinputlisting{./oop_prep/decorator2.py}
+
+In the above example, @steamed\_milk is a decorator.
+steamed\_milk is a function that accepts one argument, the name of the decorated
+function. steamed\_milk returns its inner function, decor. The inner
+function adds some ``toppings'' (Steamed milk).
+
+With overriding (single inheritance) we can add some toppings.
+
+With mixin (multiple inheritance) we can also add some toppings.
+
+We can stack decorators.
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Design patterns}
+
+Bridges (stone arch, suspension, cantilever, etc).
+
+We have proven bridge structures that work.
+
+Standard solutions to design for frequently encountered problems.
+
+{\bf Pattern}: an example for others to follow.
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Design patterns - the iterator pattern}
+
+An iterator object usually has a \texttt{next} method and a \texttt{done} method.
+
+\begin{lstlisting}
+ while not iterator.done():
+ item = iterator.next()
+ # do sth with the item
+\end{lstlisting}
+
+In Python,
+
+\begin{itemize}
+\item We have a special method called \texttt{\_\_next\_\_}, accessible by \texttt{next(iterator)}.
+\item There is no \texttt{done} method. Raise exception \texttt{StopIteration} instead.
+\end{itemize}
+
+\lstinputlisting{./oop_prep/iterator_iterable.py}
+
+Internally, a \texttt{for} loop is actually a \texttt{while} loop.
+
+\begin{lstlisting}
+iter([1,2,3])
+<list_iterator object at 0x0000000002C40B00>
+
+iter('123')
+<str_iterator object at 0x0000000002C45E10>
+
+iter({'a':1, 'b':2})
+<dict_keyiterator object at 0x0000000002AD7098>
+\end{lstlisting}
+
+
+Two abstract base classes in \texttt{collections.abc}:
+
+\begin{itemize}
+\item Iterable must define \texttt{\_\_iter\_\_}.
+\item Iterator must define \texttt{\_\_next\_\_} and \texttt{\_\_iter\_\_}, collectively called the iterator protocol.
+\end{itemize}
+
+File objects are iterators too. It has method \texttt{\_\_next\_\_} and \texttt{\_\_iter\_\_} (inherited from \_IOBase)
+
+If there is no StopIetration, we have an infinite iterator.
+
+\lstinputlisting{./oop_prep/RandomIterator.py}
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Comprehensions}
+
+Convert/map a list of items of this form to a list of items of that
+form. Each item from the original list can be passed into a function
+and the return value of that function becomes the new item in the
+converted list.
+
+Usually done in {\bf one} line of code.
+
+Benefits: very concise.
+
+We have list comprehensions, set comprehensions and dictionary comprehensions.
+
+``List comprehensions are far faster than \texttt{for} loops when looping over a huge number of items.'' - I don't agree as of November 2018. In fact, they have similar performance.
+
+\lstinputlisting{./oop_prep/comprehension.py}
+
+\begin{lstlisting}
+
+Time used [comprehension]: 5.93
+Time used [for]: 6.19
+Time used [comprehension]: 0.31
+Time used [for]: 0.30
+
+\end{lstlisting}
+
+We can also make set comprehensions or dictionary comprehensions. To do that, we use braces instead of brackets.
+
+\lstinputlisting{./oop_prep/comprehension_set.py}
+
+Lists, sets and dictionaries are called {\em containers}.
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Generators}
+
+Sometimes the data is too large, e.g., GB or TB, and we don't need it
+at once, so we don't need to load everything into computer memory.
+
+
+We can use a \texttt{for} loop.
+
+We can also use a comprehension-like expression, called a generator.
+
+\lstinputlisting{./oop_prep/apache-samples/generator1.py}
+
+Use \texttt{yield} inside a function to make a generator.
+
+\lstinputlisting{./oop_prep/apache-samples/generator2.py}
+
+What is the type of filter? It is a generator, which can be used to
+generate many values (one-the-fly). A generator is an iterator (since
+it has methods \texttt{\_\_next\_\_} and \texttt{\_\_iter\_\_}), which
+will be exhausted after one pass.
+
+What is going on inside looks like:
+
+\lstinputlisting{./oop_prep/apache-samples/generator2b.py}
+
+The generator is created without executing the code in the function
+body. When we put the generator in a for loop (which internally call
+the method \texttt{\_\_next\_\_}), each iteration will stop after the
+\texttt{yield} statement.
+
+\lstinputlisting{./oop_prep/apache-samples/generator3.py}
+
+$x$ is 0, $y$ is 1 and $z$ is 2.
+
+Yield data from another generator using \texttt{yield from}.
+
+\lstinputlisting{./oop_prep/apache-samples/generator3b.py}
+
+List all files and directories under a directory:
+
+\lstinputlisting{./oop_prep/apache-samples/generator4.py}
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Coroutines}
+
+`''Generators with a bit extra syntax.''
+
+Execution order:
+
+\begin{itemize}
+\item \texttt{yield} occurs and the generator pauses.
+\item \texttt{send()} occurs and the generator resumes.
+\item The value sent in is assigned (to the left side of the \texttt{yield} statement).
+\item The generator continues until the next \texttt{yield} statement.
+\end{itemize}
+
+
+\lstinputlisting{./oop_prep/coroutine.py}
+
+\lstinputlisting{./oop_prep/coroutine_echo.py}
+
+\lstinputlisting{./oop_prep/tally.py}
+
+Linux kernel log parsing:
+
+{\tiny
+\begin{verbatim}
+unrelated log messages
+sd 0:0:0:0 Attached Disk Drive
+unrelated log messages
+sd 0:0:0:0 (SERIAL=ZZ12345)
+unrelated log messages
+sd 0:0:0:0 [sda] Options
+unrelated log messages
+XFS ERROR [sda]
+
+unrelated log messages
+sd 2:0:0:1 Attached Disk Drive
+unrelated log messages
+sd 2:0:0:1 (SERIAL=ZZ67890)
+unrelated log messages
+sd 2:0:0:1 [sdb] Options
+unrelated log messages
+
+sd 3:0:1:8 Attached Disk Drive
+unrelated log messages
+sd 3:0:1:8 (SERIAL=WW11111)
+unrelated log messages
+sd 3:0:1:8 [sdc] Options
+unrelated log messages
+XFS ERROR [sdc]
+unrelated log messages
+\end{verbatim}
+}
+
+\lstinputlisting{./oop_prep/coroutine_logparse.py}
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Design patterns - the observer pattern}
+
+The observers are told (updated with) changes occurred to the {\bf core}
+object.
+
+(The observers can then take their own actions.)
+
+Useful for making redundant backup (in database, remote host, local file, etc).
+
+Useful for broadcasting an announcement (using email, text messages, and other instant messaging systems).
+
+\lstinputlisting{./oop_prep/Design_pattern_observer.py}
+
+In the above example, whenever we change the properties \texttt{quantity} and \texttt{product}, the observers get updated (because we called the method \texttt{\_update\_observers}).
+
+We can use \texttt{o()} because we have defined the special method \texttt{\_\_call\_\_}.
+
+The main point of using the observer pattern is that we can add (attach) different observers (having different behaviors) upon any change in the core object.
+
+Different behaviors:
+
+\begin{itemize}
+\item Back up the data in a file.
+\item Back up the data in a database.
+\item Back up the data in an Internet application.
+\item Back up the data in a tape.
+\end{itemize}
+
+Main benefit: code detachment. We don't have to mix code handling {\em multiple behaviors} code in the observed object. Instead, we put such code in individual observers, facilitating maintainability.
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Design patterns - the strategy pattern}
+
+Provide different solutions to a single problem, each in a different object.
+
+\lstinputlisting{./oop_prep/Design_pattern_strategy.py}
+
+Each class has a single method, and the method names are the same in all three classes.
+
+Each class does nothing else except provide a single function.
+
+We can replace \texttt{make\_background} with \texttt{\_\_call\_\_} to make the object directly callable (for example, \texttt{strategy('trump.jpg', (2400, 1200))}).
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Design patterns - the state pattern}
+
+See page 314 in the textbook.
+
+The XML file \texttt{book.xml} is shown below.\\
+
+\lstinputlisting{./oop_prep/book.xml}
+
+Each class represents a state (see page 316 for the state transition diagram).
+
+Each state class has the same method called \texttt{process}, which
+takes the context object \texttt{parser} as an argument.
+\texttt{process} consumes the input string, edits the tree, and modifies \texttt{parser}'s state.
+
+Note that the context class \texttt{Parser} also has method
+\texttt{process}, which invokes state object's \texttt{process} for
+consuming the remaining string, and recursively calls itself.\\
+
+
+\lstinputlisting{./oop_prep/Design_pattern_state.py}
+
+I have to say the above design is quite clever.
+
+Parse \texttt{book.xml} with a generator. \\
+
+\lstinputlisting{./oop_prep/coroutine_parse_xml.py}
+
+Output\\
+
+\lstinputlisting{./oop_prep/parse_xml_output.txt}
+
+
+\foilhead{Design patterns - the decorator pattern}
+
+
+Purpose: wrap an object to make it look different. In other words, add extra things.
+
+Decorate a string. \\
+
+\lstinputlisting{./oop_prep/Design_pattern_decorator1.py}
+
+We have produced {\em altered} strings by wrapping it in a class and rewriting the \texttt{\_\_str\_\_} method.
+
+Gift wrapping (many options). \\
+
+\lstinputlisting{./oop_prep/Design_pattern_decorator0.py}
+
+The interface is not changed. Recall that we learned property decorators before. \\
+
+\lstinputlisting{./oop_prep/Design_pattern_decorator.py}
+
+%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%% \foilhead{Monkey-patching}
+
+%% %Use functions (function names) as attributes.
+
+%% Monkey-patching: replacing or adding methods at run time.
+
+%% Class-level and object-level.
+
+%% A ``messy hack''.
+
+%% Useful for mock testing. For example, replace a transaction method \texttt{draw\_money} which connects to a remote, in-service database and makes changes there with a dummy transaction method which operates on a local database dedicated for testing.
+
+%% Use it sparingly. \\
+
+
+%% \lstinputlisting{./oop_prep/fakeprint.py}
+
+%% \texttt{dir(my\_print)} gives us a special method \texttt{\_\_call\_\_}.
+
+%% We can create {\em callable objects} by adding a special method \texttt{\_\_call\_\_} to the class definition, as we did for \texttt{a2()} and \texttt{a3()}.
+
+%% First-class functions. Functions can be treated as objects, i.e., can be assigned to a variable or used as function arguments.
+
+%% {\em Functions are objects}. (1) We can set attributes on functions. (2) We can pass functions (as arguments) around.
+
+%% \lstinputlisting{./oop_prep/functions_are_objects.py}
+
+
+%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%% \foilhead{Design patterns - the composite pattern}
+
+%% See page 351 in the textbook for the UML diagram.
+
+%% \lstinputlisting{./oop_prep/Design_pattern_composite.py}
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Design patterns - the template pattern}
+
+The {\bf Don't Repeat Yourself} principle.
+
+Useful in the situation where several tasks share a common subset of steps.
+
+For example, query a SQLite database using different query constructions (statements).
+
+Base class: a sequence of common steps.
+
+Subclasses: overriding one or several of the above steps to provide customized behaviors.
+
+Refer to the picture in Dusty's book (page 325).
+
+\lstinputlisting{./oop_prep/Design_pattern_template.py}
+
+In the two subclasses, we create \texttt{self.query}, which is {\bf not declared but assumed to be there} in the superclass QueryTemplate.
+
+We don't have to declare all attributes in the \texttt{\_\_init\_\_} method. Instead, we can add attributes ``on the fly'' (in class methods or even after we've created an object from that class).
+
+This design pattern can minimize change if we change to other SQL engines, such MySQL. We only need to change QueryTemplate, while leaving its subclasses not affected.
+
+We can override \texttt{format\_results} in UserGrossQuery if we want an HTML file instead of a text file.
+
+
+%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%% \foilhead{Design patterns - the adapter pattern}
+
+%% An adapter object sits between two different interfaces, performing translation.
+
+%% Translate argument format. Translate order of arguments.
+
+%% \lstinputlisting{./oop_prep/Design_pattern_adapter.py}
+
+
+%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%% \foilhead{Design patterns - the facade pattern}
+
+%% This pattern provides a simple interface (wrapper) that encapsulates the typical usage of a complex system.
+
+%% Easy-to-comprehend and easy-to-use interfaces to complicated tasks.
+
+%% %This pattern abstracts a simpler interface out of a complex one.
+
+%% \lstinputlisting{./oop_prep/Design_pattern_facade.py}
+
+%% Now we can use the above simple facade class to send and receive
+%% messages much easier than if we have to interact directly with the
+%% complex libraries (\texttt{smtplib} and \texttt{imaplib}).
+
+%% In fact, the library \texttt{smtplib} itself is an abstraction for the SMTP (Simple Mail Transfer Protocol).
+
+%% Other facades:
+%% \begin{itemize}
+%% \item \texttt{for} loops
+%% \item \texttt{list} comprehensions
+%% \item Generators
+%% \end{itemize}
+
+
+%% %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+%% \foilhead{Design patterns - the singleton pattern}
+
+%% SKIP. SKIP. SKIP. SKIP. SKIP. SKIP.
+
+%% Only one instance of class is created (not matter how many times we create objects of that class).
+
+%% \lstinputlisting{./oop_prep/singleton_pattern.py}
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Responsibility-driven design}
+
+Nine principles in Craig Larman's GRASP (General Responsibility Assignment Software Patterns).
+
+These principles are a learning aid for object-oriented design with responsibilities.
+
+Use methods to fulfill responsibilities.
+
+\begin{itemize}
+\item Information Expert. \newline \newline
+ PROBLEM: What is a basic principle by which to assign responsibilities
+ to an object? \newline
+ SOLUTION: Assign a responsibility to the class that has the information
+ needed to respond to it. \newline
+ Examples:
+ \begin{itemize}
+ \item Assign a {\em search} responsibility to a Catalog object, since the Catalog object has the information for all books in the library.
+ \item Assign a {\em locate} responsibility to a LibraryItem object, since the LibraryItem object has the information for the DDS number for the book.
+ \end{itemize}
+\item Creator. \newline \newline
+ PROBLEM: Who creates an object A?\newline
+ SOLUTION: Assign class B the responsibility to create
+ an instance of class A if one of these is true:
+ \begin{itemize}
+ \item B ``contains'' or completely aggregates A
+ \item B records A
+ \item B closely uses A
+ \item B has the initializing data for A
+ \end{itemize}
+ Example: \newline
+ \begin{itemize}
+ \item Assign Catalog the responsibility to create LibraryItem objects, since Catalog ``contains'' library items.
+ \item Assign Notebook the responsibility to create Note objects, since Notebook ``contains'' notes.
+ \end{itemize}
+\item Controller. \newline \newline
+ PROBLEM: What first object (beyond the UI layer) receives and coordinates a System Operation?\newline
+ SOLUTION: Assign the responsibility to an object representing one of
+ these choices:
+ \begin{itemize}
+ \item Represents the overall ``system'', a root object.
+ \item Represents a use case scenario within which the system operation occurs.
+ \end{itemize}
+ Example: \newline
+ \begin{itemize}
+ \item Assign the responsibility to Notebook object. The UI layer is represented by class Menu (page 55 in Dusty Phillips book).
+ \end{itemize}
+\item Low Coupling
+\item High Cohesion
+\item Polymorphism. \newline \newline
+ PROBLEM: How to handle alternatives based on type? \newline
+ SOLUTION: Giving the same name to similar or related services. \newline
+ Example: \newline
+ \begin{itemize}
+ \item LibraryItem is inherited by Book, CD, and DVD, each having the same service {\em locate}.
+ \end{itemize}
+\item Pure Fabrication. \newline \newline
+ PROBLEM: Sometimes assigning responsibilities only to domain
+layer software classes leads to problems like poor
+cohesion or coupling, or low reuse potential. \newline
+ SOLUTION: Assign a highly cohesive set of responsibilities to a convenience class that does not represent a
+domain concept. \newline
+ Example: \newline
+ \begin{itemize}
+ \item Make a fabricated class called Cup that holds the dice, roll them, and know their total. Cup can be reused in other applications.
+ \end{itemize}
+\item Indirection
+\item Protected Variations
+\end{itemize}
+
+
+
+%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
+\foilhead{Test-driven development}
+
+It forces writing tests before writing code.
+
+It helps understand requirements better.
+
+A good reading: Yamaura, Tsuneo. "How to Design Practical Test Cases." IEEE Software (November/December 1998): 30-36.
+
+It ensures that code continues working after we make changes.
+
+Note that changes to one part of the code can inadvertently break other parts.
+
+
+\end{document}
|