Session Seven: Testing, More OO

Multiple Inheritance,
Class and Static Methods,
Special (Magic) Methods


Review of Previous Class

  • Did anyone look more deeply into Unicode?
    • Any questions about that?
  • Object Oriented Programming
    • Questions about concept?
    • Questions about Python implimentation?

Homework review

Homework Questions?

How is progress going on the HTML Renderer?

A Quick Note

One issue that seems vexing is how to make a script “executable”

Have you seen something like this:

$ ./
-bash: ./ Permission denied

The problem is that the file is not “executable”:

$ ls -l
-rw-r--r--  1 cewing  staff  5015 Dec 10 21:18

The fix for this is to add the executable bit to the permissions for the file:

$ chmod u+x
$ ls -l
-rwxr--r--  1 cewing  staff  5015 Dec 10 21:18

You can also do this with a numeric file-mode designation:

$ chmod 744
$ ls -l
-rwxr--r--  1 cewing  staff  5015 Dec 10 21:18


You’ve already seen some a very basic testing strategy.

You’ve written some tests using that strategy.

These tests were pretty basic, and a bit awkward in places (testing error conditions in particular).

It gets better

Test Runners

So far our tests have been limited to code in an if __name__ == "__main__": block.

  • They are run only when the file is executed
  • They are always run when the file is executed
  • You can’t do anything else when the file is executed without running tests.

This is not optimal.

Python provides testing systems to help.

The original testing system in Python.

You write subclasses of the unittest.TestCase class:

# in
import unittest

class MyTests(unittest.TestCase):
    def test_tautology(self):
        self.assertEquals(1, 1)

Then you run the tests by using the main function from the unittest module:

# in
if __name__ == '__main__':

This way, you can write your code in one file and test it from another:

# in
def my_func(val1, val2):
    return val1 * val2

# in
import unittest
from my_mod import my_func

class MyFuncTestCase(unittest.TestCase):
    def test_my_func(self):
        test_vals = (2, 3)
        expected = reduce(lambda x, y: x * y, test_vals)
        actual = my_func(*test_vals)
        self.assertEquals(expected, actual)

if __name__ == '__main__':

The unittest module is pretty full featured

It comes with the standard Python distribution, no installation required.

It provides a wide variety of assertions for testing all sorts of situations.

It allows for a setup and tear down workflow both before and after all tests and before and after each test.

It’s well known and well understood.

It’s Object Oriented, and quite heavy.

It was modeled after Java’s junit and it shows...

It uses the framework design pattern, so knowing how to use the features means learning what to override.

Needing to override means you have to be cautious.

Test discovery is both inflexible and brittle.

There are several other options for running tests in Python.

  • Nose
  • pytest
  • ... (many frameworks supply their own test runners)

We are going to play today with pytest

The first step is to install the package:

(cff2py)$ pip install pytest

You may need to use ‘sudo’ to get that to work.

Once this is complete, you should have a py.test command you can run at the command line:

(cff2py)$ py.test

If you have any tests in your repository, that will find and run them.

Do you?

I’ve added two files to the Examples/Session07 folder, along with a python source code file called

The results you should have seen when you ran py.test above come partly from these files.

Let’s take a few minutes to look these files over.


When you run the py.test command, pytest starts in your current working directory and searches the filesystem for things that might be tests.

It follows some simple rules:

  • Any python file that starts with test_ or _test is imported.
  • Any functions in them that start with test_ are run as tests.
  • Any classes that start with Test are treated similarly, with methods that begin with test_ treated as tests.

This test running framework is simple, flexible and configurable.

Read the documentation for more information.

What we’ve just done here is the first step in what is called Test Driven Development.

A bunch of tests exist, but the code to make them pass does not yet exist.

The red we see in the terminal when we run our tests is a goad to us to write the code that fixes these tests.

Let’s do that next!

[lab time!]

More on Subclassing

Watch This Video:

( I pointed you to it last week, but Seriously, well worth the time. )

What’s a Subclass For?

The most salient points from that video are as follows:

Subclassing is not for Specialization

Subclassing is for Reusing Code

Bear in mind that the subclass is in charge

Is any of this starting to make sense with the HTML builder example?

Multiple Inheritance

Multiple inheritance: Inheriting from more than one class

Simply provide more than one parent.

class Combined(Super1, Super2, Super3):
    def __init__(self, something, something else):
        # some custom initialization here.
        Super1.__init__(self, ......)
        Super2.__init__(self, ......)
        Super3.__init__(self, ......)
        # possibly more custom initialization

(calls to the super class __init__ are optional – case dependent)

Now you have one class with functionaility of ALL the superclasess!

But what if the same attribute exists in more than one superclass?

class Combined(Super1, Super2, Super3)

Attributes are located bottom-to-top, left-to-right

  • Is it an instance attribute ?
  • Is it a class attribute ?
  • Is it a superclass attribute ?
    • is the it an attribute of the left-most superclass?
    • is the it an attribute of the next superclass?
    • and so on up the hierarchy...
  • Is it a super-superclass attribute ?
  • ... also left to right ...

(This is not at all simple!)

Why would you want multiple inheritance? – one reason is mix-ins.

Provides an subset of expected functionality in a re-usable package.

Hierarchies are not always simple:

  • Animal
    • Mammal
      • GiveBirth()
    • Bird
      • LayEggs()

Where do you put a Platypus?

Real World Example: FloatCanvas

Careful About This Pattern

All the class definitions we’ve been showing inherit from object.

This is referred to as a “new style” class.

They were introduced in python2.2 to better merge types and classes, and clean up a few things.

There are differences in method resolution order and properties.

Always Make New-Style Classes.

The differences are subtle, and may not appear until they jump up to bite you.

(which they will the rest of this class session!)

super(): use it to call a superclass method, rather than explicitly calling the unbound method on the superclass.

instead of:

class A(B):
    def __init__(self, *args, **kwargs)
        B.__init__(self, *args, **kwargs)

You can do:

class A(B):
    def __init__(self, *args, **kwargs)
        super(A, self).__init__(*args, **kwargs)

Caution: There are some subtle differences with multiple inheritance.

One difference is the syntax: need to think hard to understand all that:

super(A, self).__init__(*args, **kwargs)

This means something like:

“create a super object for the superclass of class A, with this instance. Then call __init__ on that object.”

Important note: super() does not return the superclass object!

But you can use explicit calling to ensure that the ‘right’ method is called.

Two seminal articles about super():

“Super Considered Harmful” – James Knight

“super() considered super!” – Raymond Hettinger}

(Both worth reading....)

While appearing to be contradictory, they both have the same final message...

super() issues...

Both articles actually say similar things:

  • The method being called by super() needs to exist
  • Every occurrence of the method needs to use super():
    • Use it consistently, and document that you use it, as it is part of the external interface for your class, like it or not.

The caller and callee need to have a matching argument signature:

Never call super with anything but the exact arguments you received, unless you really know what you’re doing.

If you add one or more optional arguments, always accept:

*args, **kwargs

and call super like:

super(MyClass, self).method(args_declared, *args, **kwargs)


One of the strengths of Python is lack of clutter.

Attributes are simple and concise:

In [5]: class C(object):
        def __init__(self):
                self.x = 5
In [6]: c = C()
In [7]: c.x
Out[7]: 5
In [8]: c.x = 8
In [9]: c.x
Out[9]: 8

Getter and Setters?

But what if you need to add behavior later?

  • do some calculation
  • check data validity
  • keep things in sync
In [5]: class C(object):
   ...:     def __init__(self):
   ...:         self.x = 5
   ...:     def get_x(self):
   ...:         return self.x
   ...:     def set_x(self, x):
   ...:         self.x = x
In [6]: c = C()
In [7]: c.get_x()
Out[7]: 5
In [8]: c.set_x(8)
In [9]: c.get_x()
Out[9]: 8

<shudder> This is ugly and verbose – Java?

When (and if) you need them:

class C(object):
    def __init__(self, x=5):
        self._x = x
    def _getx(self):
        return self._x
    def _setx(self, value):
        self._x = value
    def _delx(self):
        del self._x
    x = property(_getx, _setx, _delx, doc="docstring")

Now the interface is still like simple attribute access!

[demo: Examples/Session07/]

Not all the arguments to property are required.

You can use this to create attributes that are “read only”:

In [11]: class D(object):
   ....:     def __init__(self, x=5):
   ....:         self._x = 5
   ....:     def getx(self):
   ....:         return self._x
   ....:     x = property(getx, doc="I am read only")
In [12]: d = D()
In [13]: d.x
Out[13]: 5
In [14]: d.x = 6
AttributeError                            Traceback (most recent call last)
<ipython-input-14-c83386d97be3> in <module>()
----> 1 d.x = 6
AttributeError: can't set attribute

This imperative style of adding a property to you class is clear, but it’s still a little verbose.

It also has the effect of leaving all those defined method objects laying around:

In [19]: d.x
Out[19]: 5
In [20]: d.getx
Out[20]: <bound method D.getx of <__main__.D object at 0x1043a4a10>>
In [21]: d.getx()
Out[21]: 5

Python provides us with a way to solve both these issues at once, using a syntactic feature called decorators (more about these next session):

In [22]: class E(object):
   ....:     def __init__(self, x=5):
   ....:         self._x = x
   ....:     @property
   ....:     def x(self):
   ....:         return self._x
   ....:     @x.setter
   ....:     def x(self, value):
   ....:         self._x = value
In [23]: e = E()
In [24]: e.x
Out[24]: 5
In [25]: e.x = 6
In [26]: e.x
Out[26]: 6

Static and Class Methods

You’ve seen how methods of a class are bound to an instance when it is created.

And you’ve seen how the argument self is then automatically passed to the method when it is called.

And you’ve seen how you can call unbound methods on a class object so long as you pass an instance of that class as the first argument.

But what if you don’t want or need an instance?

Static Methods

A static method is a method that doesn’t get self:

In [36]: class StaticAdder(object):
   ....:     def add(a, b):
   ....:         return a + b
   ....:     add = staticmethod(add)

In [37]: StaticAdder.add(3, 6)
Out[37]: 9

[demo: Examples/Session07/]

Like properties, static methods can be written declaratively using the staticmethod built-in as a decorator:

class StaticAdder(object):
    def add(a, b):
        return a + b

Where are static methods useful?

Usually they aren’t

99% of the time, it’s better just to write a module-level function

An example from the Standard Library (

class TarInfo(object):
    # ...
    def _create_payload(payload):
        """Return the string payload filled with zero bytes
           up to the next 512 byte border.
        blocks, remainder = divmod(len(payload), BLOCKSIZE)
        if remainder > 0:
            payload += (BLOCKSIZE - remainder) * NUL
        return payload

Class Methods

A class method gets the class object, rather than an instance, as the first argument

In [41]: class Classy(object):
   ....:     x = 2
   ....:     def a_class_method(cls, y):
   ....:         print(u"in a class method: ", cls)
   ....:         return y ** cls.x
   ....:     a_class_method = classmethod(a_class_method)
In [42]: Classy.a_class_method(4)
in a class method:  <class '__main__.Classy'>
Out[42]: 16

[demo: Examples/Session07/]

Once again, the classmethod built-in can be used as a decorator for a more declarative style of programming:

class Classy(object):
    x = 2
    def a_class_method(cls, y):
        print(u"in a class method: ", cls)
        return y ** cls.x

Unlike static methods, class methods are quite common.

They have the advantage of being friendly to subclassing.

Consider this:

In [44]: class SubClassy(Classy):
   ....:     x = 3

In [45]: SubClassy.a_class_method(4)
in a class method:  <class '__main__.SubClassy'>
Out[45]: 64

Because of this friendliness to subclassing, class methods are often used to build alternate constructors.

Consider the case of wanting to build a dictionary with a given iterable of keys:

In [57]: d = dict([1,2,3])
TypeError                                 Traceback (most recent call last)
<ipython-input-57-50c56a77d95f> in <module>()
----> 1 d = dict([1,2,3])

TypeError: cannot convert dictionary update sequence element #0 to a sequence

The stock constructor for a dictionary won’t work this way. So the dict object implements an alternate constructor that can.

def fromkeys(cls, iterable, value=None):
    '''OD.fromkeys(S[, v]) -> New ordered dictionary with keys from S.
    If not specified, the value defaults to None.

    self = cls()
    for key in iterable:
        self[key] = value
    return self

(this is actually from the OrderedDict implementation in

See also, etc....

Properties, Static Methods and Class Methods are powerful features of Pythons OO model.

They are implemented using an underlying structure called descriptors

Here is a low level look at how the descriptor protocol works.

The cool part is that this mechanism is available to you, the programmer, as well.

Kicking the Tires

Copy the file Example/Session07/ to your student folder. (we used it for our testing try out...)

In it, update the simple “Circle” class:

In [13]: c = Circle(3)
In [15]: c.diameter
Out[15]: 6.0
In [16]: c.diameter = 8
In [17]: c.radius
Out[17]: 4.0
In [18]: c.area
Out[18]: 50.26548245743669

Use properties so you can keep the radius and diameter in sync, and the area computed on the fly.

Extra Credit: use a class method to make an alternate constructor that takes the diameter instead.

Also copy the file to your student folder.

As you work, run the tests:

(cff2py)$ py.test

As each of the requirements from above are fulfilled, you’ll see tests ‘turn green’.

When all your tests are passing, you’ve completed the job.

(This clear finish line is another of the advantages of TDD)

Special Methods

Special methods (also called magic methods) are the secret sauce to Python’s Duck typing.

Defining the appropriate special methods in your classes is how you make your class act like standard classes.

What’s in a Name?

We’ve seen at least one special method so far:


It’s all in the double underscores...

Pronounced “dunder” (or “under-under”)

try: dir(2) or dir(list)

The set of special methods needed to emulate a particular type of Python object is called a protocol.

Your classes can “become” like Python built-in classes by implementing the methods in a given protocol.

Remember, these are more guidelines than laws. Implement what you need.

Do you want your class to behave like a number? Implement these methods:

object.__add__(self, other)
object.__sub__(self, other)
object.__mul__(self, other)
object.__floordiv__(self, other)
object.__mod__(self, other)
object.__divmod__(self, other)
object.__pow__(self, other[, modulo])
object.__lshift__(self, other)
object.__rshift__(self, other)
object.__and__(self, other)
object.__xor__(self, other)
object.__or__(self, other)

Want to make a container type? Here’s what you need:

object.__getitem__(self, key)
object.__setitem__(self, key, value)
object.__delitem__(self, key)
object.__contains__(self, item)
object.__getslice__(self, i, j)
object.__setslice__(self, i, j, sequence)
object.__delslice__(self, i, j)

Each of these methods supports a common Python operation.

For example, to make ‘+’ work with a sequence type in a vector-like fashion, implement __add__:

def __add__(self, v):
    """return the element-wise vector sum of self and v
    assert len(self) == len(v)
    return Vector([x1 + x2 for x1, x2 in zip(self, v)])

[a more complete example: Examples/Session07/>]

You only need to define the special methods that will be used by your class.

However, even in the absence of wanting to duck-type, you should almost always define these:

Called by the str() built-in function and by the print statement to compute the informal string representation of an object.
Called by the unicode() built-in function. This converts an object to an informal unicode representation.

Called by the repr() built-in function and by string conversions (reverse quotes) to compute the official string representation of an object.

(ideally: eval( repr(something) ) == something)

Use special methods when you want your class to act like a “standard” class in some way.

Look up the special methods you need and define them.

There’s more to read about the details of implementing these methods:

Be a bit cautious about the code examples in that last one. It uses quite a bit of old-style class definitions, which should not be emulated.

Kicking the Tires

Extend your “Circle” class:

  • Add __str__ and __repr__ methods
  • Write an __add__ method so you can add two circles
  • Make it so you can multiply a circle by a number....
In [22]: c1 = Circle(3)
In [23]: c2 = Circle(4)
In [24]: c3 = c1+c2
In [25]: c3.radius
Out[25]: 7
In [26]: c1*3
Out[26]: Circle(9)

If you have time: compare them... (c1 > c2 , etc)

As you work, run the tests in

(cff2py)$ py.test

As each of the requirements from above are fulfilled, you’ll see tests ‘turn green’.

When all your tests are passing, you’ve completed the job.