When you sit down to write a module of code, there is a very strong probability that you are NOT the first person in the world to need that module. In most of these cases, the functionality you need has already been written and available for you to use. This saves you time and the module usually has been rigorously tested for errors.
The phrase that comes to mind for this situation is:
Python Standard Library - This library comes with every release of Python. There are so many modules we could spend a full 2 weeks getting familiar with them. A non-exhaustive list includes packages for:
Example: You need to calculate absolute values of numbers in your code module. Should you write a function to do that from scratch? Are you the first software engineer to need an absolute value? Of course not! This functionality exists in the appropriately titled "math" package.
The math module documentation provides the following information:
math.fabs(x)
# Return the absolute value of x.Which you access by importing the module by typing the important statement in the header of your Python script and later in your code call fabs(x) with:
import math
...
y = math.fabs(x)Python Package Index (PyPI) - While the Python Standard Library is rich, it does not include every possible need. Generous individuals, groups and institutions have created and made public packages of modules designed to work with Python. There are more than 40,000 Python packages in PyPI! There are additional Python packages that are not included in PyPI. We will not cover all of these today.
Before you write a module of code, I recommend you browse PyPI https://pypi.python.org/pypi?:action=browse and search the internet for pre-existing implementations.
In order to download and install other modules or packages, you need to install the "pip" program. Instructions on the website walk you through it.
Picking a good package is a challenge. You may have many to choose from and their quality and features will vary. Do research and test before you commit to develop code with a particular package.
About Frameworks: Some packages you see in pip are "Frameworks". A framework is different than a typical package.
A framework is a package with a set of functions for a particular task. A framework hides low level details of the functions which typically speeds development time and improves readability. We will not explore Frameworks further in this course.
The best known Framework for Python is called Django. It is used to create web applications.
You or someone in your organization wrote code in the past that is applicable to a current requirement. So it is an easy task to reuse this code, right?
WRONG!
Code reuse is a very challenging proposition for organizations. Why? The basic concept is that code is written to solve a particular set of problems under a deadline. Code reuse runs into the following types of challenges.
a. Organization. A package intended for reuse must be fully featured and meet more than the narrow needs of the original authors. This requires planning and strong support from the organization`s leadership.
b. Economics. A code reuse strategy requires additional work. The time and funding to package code for reuse may not be available.
c. Administration. Code to be reused must be available in a central location, maintained, documented and its existence communicated to the entire organization.
d. Politics. Some organizations have hostile internal politics and use of code written by other work groups might be resisted due to a loss of prestige or perceived loss of control of a project.
e. Psychology. For a variety of reasons, individual developers may resent using key components for their project that they did not write.
If your organization has the leadership to overcome these issues, great! You are very lucky and your development will be eased. Because code reuse is a difficult proposition, this rarely happens.
Thus code is typically reused in an ad hoc fashion, obtained from developers in your local work group. So before you code, ask around. In this case, the source is typically copied to the new project and modified and retested.
Design Patterns: The purpose of software is to solve problems! If code you need to solve a problem has been written before, surely the problem has been solved before. Or at least some of the problem.
The factory pattern creates objects that inherit from an abstract class. The actual subclass is not decided until runtime. As long as the subclasses implement the required interface(s), they can be handled identically by the rest of the program while retaining the characteristics of the subclass.
Example: An audio editing program that needs to handle different audio formats with the same set of functions. So regardless of the audio type (.au, .wav, .mp3, etc.), you can get a generic audio object that can be handled by the program in the same way.
The strategy pattern allows the behavior of an function to be changeable at runtime. This is done by defining a common interface that a family of functions implements. As the state of the program or an object changes, its function implementation can be changed appropriately.
Example: Scientific application that processes different types of datasets. The manner in which the distance between individual data examples is measured is different depending on the dataset. A family of distance(example_a, example_b) functions can be defined and the appropriate one chosen when the type of dataset is determined.
Exercise 1: Write a script to calculate the weekly pay for employees. Where appropriate, use applicable Python library modules.
Program should print out "employee name: pay" for each Employee
The pay scale depends on number of years of service:
0 to <2 = 500
4 to <8 = 1000
8 to <12 = 1100
12+ = 1200
you may assume 365 days = 1 year
Some code to get you started. It might need modification as you discover useful library modules:
class Employee():
def __init__(self, name, start_date):
self.name = name
self.start_date = start_date
Andrew = Employee("Andrew", "2012/08/01")
Barbara = Employee("Barbara", "2004/01/19")
Clare = Employee("Clare", "2007/11/30")
Daniel = Employee("Daniel", "2001/05/23")
Edith = Employee("Edith", "2010/07/04")
Exercise 2: You are planning a large software system to be written in Python. The system will monitor Twitter for tweets that mention particular types of activty. These are environmental changes, damage or pollution. The system will plot the location of the activity on Google Maps.
Search PyPi for at least three useful packages that might assist you.
python-twitter1.3.1, nltk, pykml 0.1.0