Libraries and Modules in Python
Error detection and exceptions¶
We all make mistakes¶
Mistakes and errors happen in computer programs as much as in real life. Like life, how you handle an error in your program shows your level of professionalism, and gives others evidence that they can trust that you have written a program that will work well.
In the last section we indicated errors in the Person.setHeight function by printing a message to the screen and returning False to indicate that the call to setHeight had failed.
class Person:
"""Class that holds a person's height"""
def __init__(self):
"""Construct a person who has zero height"""
self._height = 0
def setHeight(self, height):
"""Set the person's height to 'height', returning whether or
not the height was set successfully
"""
if height < 0 or height > 300:
print("This is an invalid height! %s" % height)
return False
else:
self._height = height
return True
def getHeight(self):
"""Return the person's height"""
return self._height
p = Person()
p.setHeight(-20)
Why is this a problem?¶
This is not a good way of indicating an error. The issues with this are;
- How does the person calling
getHeightknow to check whether the call returnsTrueorFalse - What if we wanted to return something else? Should we return the error state and the value we want together?
- If the error state is not checked, and nobody reads the error message printed to the screen, then the program is broken, as the person has been created with a height of 0.
The solution is to send something to the programmer that they cannot ignore, which indicates that there is an error. That something is called an "exception".
Take a look at this simple code that sets the height...
def setHeight(height):
if height < 0 or height > 300:
raise ValueError("Invalid height: %s. This should be between 0 and 300" % height)
print("Height is set to %s" % height)
setHeight(-5)
Inform and assist your users¶
When we try to use an invalid value for the height, we raise (or throw) a ValueError exception. This stops the function from continuing, and gives us a very helpful print out of what went wrong, and where.
ValueError is just a class. The name of the class provides us with useful information (there was an error with a value in the program). You choose what error you want to raise. Python provides a set of usefully named error classes that you can use:
IOError: Error raised when you have a problem with IO, e.g. opening or closing filesZeroDivisionError: Error raised when you divide by zeroTypeError: Error raised when you are using the wrong type, e.g. maybe setting the height to a stringIndexError: Error raised when you are using an invalid index to access a list or other similar containerKeyError: Error raised when you are using an invalid key to access a dictionary or other similar container
A full list of standard Python exceptions is available here.
You are free to raise any exception class you want. It is your job as a programmer to choose the one that is most sensible, e.g.
def setHeight(height):
if height < 0 or height > 300:
raise ZeroDivisionError("Invalid height: %s. This should be between 0 and 300" % height)
print("Height is set to %s" % height)
setHeight(400)
Using a ZeroDivisionError is a bad choice, as the error has nothing to do with division by zero. A ValueError is the right choice as the error relates to an invalid value passed to the function.
You are free to create your own exception classes.
class InvalidHeightError(Exception):
pass
def setHeight(height):
if height < 0 or height > 300:
raise InvalidHeightError("Invalid height: %s. This should be between 0 and 300" % height)
print("Height is set to %s" % height)
setHeight(-10)
Your own exception classes must be declared as derived from type Exception, hence why you have to write class InvalidHeightError(Exception):. As the class doesn't need to do anything else, you can use pass to say that nothing else needs to be added. Note that you can call your error class anything you want. By convention, it is good to end the class name with Error so that other programmers know what it is for.
Exercise¶
Here is an extended version of the Person code from the previous lesson.
class Person:
"""Class that holds a person's height"""
def __init__(self, height=0, weight=0):
"""Construct a person with the specified name, height and weight"""
self.setHeight(height)
self.setWeight(weight)
def setHeight(self, height):
"""Set the person's height in meters"""
self._height = height
def setWeight(self, weight):
"""Set the person's weight in kilograms"""
self._weight = weight
def getHeight(self):
"""Return the person's height in meters"""
return self._height
def getWeight(self):
"""Return the person's weight in kilograms"""
return self._weight
def bmi(self):
"""Return the person's body mass index (bmi)"""
return self.getWeight() / self.getHeight()**2