Linked Lists

We've learned that a Python list is one way to store sequential values. Another type of list is a linked list. A Python list stores all of its elements in a single object, and each element can be accessed by using its index. A linked list, on the other hand, is a recursive object that only stores two things: its first value and a reference to the rest of the list, which is another linked list.

We can implement a class, Link, that represents a linked list object. Each instance of Link has two instance attributes, first and rest.

class Link:
    """A linked list.

    >>> s = Link(1)
    >>> s.first
    1
    >>> s.rest is Link.empty
    True
    >>> s = Link(2, Link(3, Link(4)))
    >>> s.first = 5
    >>> s.rest.first = 6
    >>> s.rest.rest = Link.empty
    >>> s                                    # Displays the contents of repr(s)
    Link(5, Link(6))
    >>> s.rest = Link(7, Link(Link(8, Link(9))))
    >>> s
    Link(5, Link(7, Link(Link(8, Link(9)))))
    >>> print(s)                             # Prints str(s)
    <5 7 <8 9>>
    """
    empty = ()

    def __init__(self, first, rest=empty):
        assert rest is Link.empty or isinstance(rest, Link)
        self.first = first
        self.rest = rest

    def __repr__(self):
        if self.rest is not Link.empty:
            rest_repr = ', ' + repr(self.rest)
        else:
            rest_repr = ''
        return 'Link(' + repr(self.first) + rest_repr + ')'

    def __str__(self):
        string = '<'
        while self.rest is not Link.empty:
            string += str(self.first) + ' '
            self = self.rest
        return string + str(self.first) + '>'

A valid linked list can be one of the following:

1. An empty linked list (Link.empty)
2. A Link object containing the first value of the linked list and a reference to the rest of the linked list

What makes a linked list recursive is that the rest attribute of a single Link instance is another linked list! In the big picture, each Link instance stores a single value of the list. When multiple Links are linked together through each instance's rest attribute, an entire sequence is formed.

Note: This definition means that the rest attribute of any Link instance must be either Link.empty or another Link instance! This is enforced in Link.__init__, which raises an AssertionError if the value passed in for rest is neither of these things.

To check if a linked list is empty, compare it against the class attribute Link.empty. For example, the function below prints out whether or not the link it is handed is empty:

def test_empty(link):
    if link is Link.empty:
        print('This linked list is empty!')
    else:
        print('This linked list is not empty!')

Motivation: Why linked lists

Since you are already familiar with Python's built-in lists, you might be wondering why we are teaching you another list representation. There are historical reasons, along with practical reasons. Later in the course, you'll be programming in Scheme, which is a programming language that uses linked lists for almost everything.

For now, let's compare linked lists and Python lists by looking at two common sequence operations: inserting an item and indexing.

Python's built-in list is like a sequence of containers with indices on them:

Linked lists are a list of items pointing to their neighbors. Notice that there's no explicit index for each item.

Suppose we want to add an item at the head of the list.

• With Python's built-in list, if you want to put an item into the container labeled with index 0, you must move all the items in the list into its neighbor containers to make room for the first item;

• With a linked list, you tell Python that the neighbor of the new item is the old beginning of the list.

Now, let's take a look at indexing. Say we want the item at index 3 from a list.

• In the built-in list, you can use Python list indexing, e.g. lst[3], to easily get the item at index 3.
• In the linked list, you need to start at the first item and repeatedly follow the rest attribute, e.g. link.rest.rest.first. How does this scale if the index you were trying to access was very large?

Can you think of situations where you would want to use one type of list over another? In this class, we aren't too worried about performance. However, in future computer science courses, you'll learn how to make performance tradeoffs in your programs by choosing your data structures carefully.

Trees

Recall that a tree is a recursive abstract data type that has a label (the value stored in the root of the tree) and branches (a list of trees directly underneath the root).

We saw one way to implement the tree ADT -- using constructor and selector functions that treat trees as lists. Another, more formal, way to implement the tree ADT is with a class. Here is part of the class definition for Tree, which can be found in lab07.py:

class Tree:
    """
    >>> t = Tree(3, [Tree(2, [Tree(5)]), Tree(4)])
    >>> t.label
    3
    >>> t.branches[0].label
    2
    >>> t.branches[1].is_leaf()
    True
    """
    def __init__(self, label, branches=[]):
        for b in branches:
            assert isinstance(b, Tree)
        self.label = label
        self.branches = list(branches)

    def is_leaf(self):
        return not self.branches

Even though this is a new implementation, everything we know about the tree ADT remains true. That means that solving problems involving trees as objects uses the same techniques that we developed when first studying the tree ADT (e.g. we can still use recursion on the branches!). The main difference, aside from syntax, is that tree objects are mutable.

Here is a summary of the differences between the tree ADT implemented using functions and lists vs. implemented using a class: