Purpose

Generic functions are functions that we can create the are able to work on a variety of different objects. For example, if we have a Square class and a Rectangle class and we want to find the area of either, we should be able to make one function that takes in either of the objects, figures out what the type of the object is, and calls the right function to find the area for that object.

Type Tag Function

Let’s start be making and understanding out type_tag function. We need some way of inspecting what type our input is, so we will be using the type function that takes in any object and returns the class that the object is an instance of. Take a look at the examples below:

>>> t = []
>>> type(t)
<class 'list'>

We can also compare the type of something by using the normal == operator.

>>> t = []
>>> type(t) == list
True

Our type_tag function is going to take in a generic object, look up its type, and then find that type in our type_tag.tags dictionary that we will define later. Here’s what our type_tag function will look like.

def type_tag(generic_object):
    return type_tag.tags[type(generic_object)]

Let’s break this down. First, we are checking the type of our object. Then we are using the type of our object as the key that we are looking up in the type_tag.tags dictionary and we will get back its value. The function itself will return this value that it finds by looking up the type.

We’re going to use this to map our class names to some broader name that can represent a bunch of different classes. For this, we are going to use the following scenario. Say we have a bunch of different types of animals, all of which have a different reproduction rate. Cats reproduce at the rate of 0.1n, Birds reproduce at the rate of n2 and Rabbits reproduce at the rate of 2n, where n is the number of years. However, we can have different types of Birds such as a Parrot, a BlueJay, and a Pigeon. Each of these birds have the same reproduction rate, however, we can now have different instances of the Bird class. Let’s define our classes as such.

class Bird(object):
    wings = 2

class Cat(object):
    meows = 2

class Rabbit(object):
    carrots = 100

class Parrot(Bird):
    colors = 'lots'

class BlueJay(Bird):
    colors = 'blue'

class Pigeon(Bird):
    colors = 'grey'

Creating Our Dictionaries

The body of the classes don’t matter that much for this example. So we need to construct some sort of dictionary that maps our class names to our shorter broader names. We want to do this because we only want to define one function that computes the reproduction rate for Birds. We don’t want to have one for each of the Parrot, BlueJay, and Pigeon classes because they are all Birds and all Birds have the same reproduction rate. Our dictionary will be defined as follows:

type_tag.tags = {Bird: 'bird', Cat: 'cat', Rabbit: 'rabbit', Parrot: 'bird', BlueJay: 'bird', Pigeon: 'bird'}

Now for a test, follow along with the code below and make sure you understand it. We will be making a series of objects and then get its type tag by passing it through the type_tag function.

>>> p = Parrot()
>>> pi = Pigeon()
>>> c = Cat()
>>> r = Rabbit()
>>> type_tag(r)
'rabbit'
>>> type_tag(p)
'bird'
>>> type_tag(c)
'cat'
>>> type_tag(pi)
'bird'

Now, we need to construct a second dictionary, this time mapping our tags to our functions that work for each tag. For the Cat class, we need to make a function that takes in a number_of_years and correctly computes the number of Cats we will have. We also need to do the same for the Rabbit and Bird classes. Let us define it as follows:

>>> reproduction_rates = {}
>>> reproduction_rates['cat'] = lambda years: 0.1 * years
>>> reproduction_rates['bird'] = lambda years: years ** 2
>>> reproduction_rates['rabbit'] = lambda years: 2 ** years

Putting it all Together

Great! Now we want to construct our generic function, children_on_the_nth_year(obj, n). This funtion is going to take in a generic object (some sort of animal that we defined above) and a number n that will calculate the amount of children that the animal will have on the nth year based on its reproduction rate. For now, let’s pretend that we have a dictionary that maps our short names (or tags) to functions that correctly compute the number of children for each object. We need some way to figure out what type our object is, look it up in this dictionary, then call that function on our input n. We define it as follows:

def children_on_the_nth_year(obj, n):
    tag = type_tag(obj)
    return reproduction_rates[tag](n)

Does this work? Let’s try some examples to see if it does indeed work.

>>> p = Parrot()
>>> children_on_the_nth_year(p, 2)
4
>>> r = Rabbit()
>>> children_on_the_nth_year(r, 3)
8

It seems to be working. Now let’s try to understand why it’s working.

We first create a Parrot object and call it p. Then we call children_on_the_nth_year with the object as our first argument and the number 2 as our second argument. What will tag equal after executing the first line of the function? Remember that type_tag looks up the class name for our object, in this case it would be looking up Parrot, and in the type_tag.tags dictionary we would lookup the key Parrot and get back the value 'bird'. We store that as the value tag. Then, on the next line, we lookup that tag that we just found in our reproduction_rates dictionary. What type of object will this dictionary return? It will give us back a function. What does this function do? It takes in some number, which represents the number of years, and returns how many children that animal will have that year. In this instance, we lookup 'bird' and get back the lambda function lambda years: years ** 2. With the input now being n which is the value 2 in this example, the function returns that value 2 ** 2 which is just 4. The overall function will now return 4 and that is our answer. The similar process applies to the Rabbit, however, this time its tag is ‘rabbit’ and the function will be for the rabbit and not for the bird.

I don't claim to be perfect so if you find an error on this page, please send me an email preferably with a link to this page so that I know what I need to fix!

comments powered by Disqus