A short article with all you need to know about sequence indexing in Python – and a bit more.

(If you are new here and have no idea what a Pydon't is, you may want to read the Pydon't Manifesto.)

Introduction

Sequences in Python, like strings, lists, and tuples, are objects that support indexing: a fairly simple operation that we can use to access specific elements. This short article will cover the basics of how sequence indexing works and then give you some tips regarding anti-patterns to avoid when using indices in your Python code.

In this article you will:

• learn the basic syntax for indexing sequences;
• learn how negative indices work;
• see some tools that are often used to work with sequences and indices;
• learn a couple of tricks and things to avoid when indexing;

Sequence indexing

First and foremost, I am talking about sequence indexing here to distinguish the type of indexing you do to access the values of a dictionary, where you use keys to index into the dictionary and retrieve its values. In this article we will be talking about using integers to index linear sequences, that is, sequences that we can traverse from one end to the other, in an ordered fashion.

A very simple example of such a sequence is a string:

>>> s = "Indexing is easy!"
>>> s
'Indexing is easy!'

To index a specific character of this string I just use square brackets and the integer that corresponds to the character I want. Python is 0-indexed, which means it starts counting indices at 0. Therefore, the very first element of a sequence can be obtained with [0]. In our example, this should give a capital "I":

>>> s = "Indexing is easy!"
>>> s[0]
'I'

Then, each following character is obtained by increasing the index by 1:

>>> s = "Indexing is easy!"
>>> s[1]
'n'
>>> s[2]
'd'
>>> s[3]
'e'

Here is a figure that shows how to look at a sequence and figure out which index corresponds to each element:

Imagine vertical bars that separate consecutive elements, and then number each of those vertical bars, starting with the leftmost bar. Each element gets the index associated with the bar immediately to its left:

Maximum legal index and index errors

Because indices start at 0, the last legal index to a sequence is the index that is equal to the length of the sequence, minus one:

>>> s = "Indexing is easy!"
>>> len(s)
17
>>> s[16]
'!'
>>> s[17]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
IndexError: string index out of range

As you can see above, if you use an index that is too large (read: greater than or equal to the length of the sequence) Python will raise an IndexError, warning you about your usage of an integer that is too large for that specific indexing operation.

Negative indices

If the last legal index is the length of the sequence minus 1, then there is an obvious way to access the last item of a sequence:

>>> s = "Indexing is easy!"
>>> s[len(s)-1]
'!'
>>> l = [12, 45, 11, 89, 0, 99]
>>> l[len(l)-1] 
99

However, Python provides this really interesting feature where you can use negative indices to count from the end of the sequence. In order to figure out which negative index corresponds to which element, think about writing the sequence to the left of itself:

Then you just have to continue the numbering from the right to the left, therefore making use of negative numbers:

From the figure above you can see that the index -1 refers to the last element of the sequence, the index -2 refers to the second to last, etc:

>>> s = "Indexing is easy!"
>>> s[-1]
'!'
>>> s[-2]
'y'

We can also take a look at all the negative indices that work for our specific sequence:

Another way to look at negative indices is to pretend there is a len(s) to their left:

Negative index		Corresponding positive index
-1		len(s) - 1
-2		len(s) - 2
-3		len(s) - 3
...		...
-len(s)		len(s) - len(s) (same as 0)

And a couple of examples:

>>> s = "Indexing is easy!"
>>> s[-5]
'e'
>>> s[len(s)-5]
'e'
>>> s[-13]
'x'
>>> s[len(s)-13]
'x'
>>> len(s)
17
>>> s[-17]
'I'
>>> s[len(s)-17]
'I'

Indexing idioms

Having seen the basic syntax for indexing, there are a couple of indices that would be helpful if you were able to read them immediately for what they are, without having to think about them:

Index operation		Interpretation
s[0]		First element of s
s[1]		Second element of s
s[-1]		Last element of s
s[-2]		Second to last element of s

To index or not to index?

Just a quick note on something that I trip over every now and then.

Python has many useful built-ins and built-in data types. Of them, strings, lists and tuples are indexable with integers. Sets are not.

You should also be careful about things that you think are like lists, but really are not. These include enumerate, zip, map, and other objects. None of these are indexable, none of these have a len value, etc. Pay attention to that!

>>> l = [0, 1, 2, 3, 4, 5, 6, 7, 8, 9]
>>> e = enumerate(l)
>>> e[3]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: 'enumerate' object is not subscriptable
>>> z = zip(l)
>>> z[3]
# ...
TypeError: 'zip' object is not subscriptable
>>> m = map(str, l)
>>> m[3]
# ...
TypeError: 'map' object is not subscriptable

Best practices in code

A looping pattern with range

Because of the way both range and indices work, one can understand that range(len(s)) will generate all the legal indices for s:

>>> s = "Indexing is easy!"
>>> list(range(len(s)))     # use list() to print the values
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16]
>>> s[0]
'I'
>>> s[16]
'!'

A consequence of this is that beginners, and people who are more distracted or used to other programming languages, end up employing a very common anti-pattern in for loops.

To exemplify this, suppose we wanted to write a fairly naïve program to find the unique letters in our string. Here is what the anti-pattern would look like:

>>> s = "Indexing is easy!"
>>> uniques = []
>>> for idx in range(len(s)):
...     if s[idx] not in uniques:
...             uniques.append(s[idx])
... 
>>> uniques
['I', 'n', 'd', 'e', 'x', 'i', 'g', ' ', 's', 'a', 'y', '!']

The problem here is that the for loop is being done in a roundabout way: we have access to a sequence (the string) that we could iterate over, but instead we find its length, so that we can use range to compute its legal indices, which we then iterate over, only to then access the elements of the sequence through their indices.

This way of writing for loops is similar to the way one would write for loops in other programming languages, if you were to iterate over the elements of an array.

However, we are using Python, not any other language. One of the things I enjoy the most about Python's for loops is that you can access directly the consecutive elements of a sequence. Hence, we can actually rewrite our for loop slightly, but in a way that makes it much more elegant:

>>> s = "Indexing is easy!"
>>> uniques = []
>>> for letter in s:
...     if letter not in uniques:
...             uniques.append(letter)
... 
>>> uniques
['I', 'n', 'd', 'e', 'x', 'i', 'g', ' ', 's', 'a', 'y', '!']

What I really like about these types of loops is that if your variables are named correctly, the statements express your intent very clearly. The line for letter in s: is read as

“For each letter in (the string) s...”

This type of for loop iterates directly over the values you care about, which is often what you want. If you care about the indices, then be my guest and use range(len(s))!

Large expressions as indices

When you are dealing with sequences and with indices for those sequences, you may end up needing to perform some calculations to compute new indices that interest you. For example, suppose you want the middle element of a string and you don't know about // yet:

>>> s = "Indexing is easy!"
>>> s[len(s)/2]     # len(s)/2 isn't an integer!!
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: string indices must be integers
>>> len(s)/2
8.5
>>> import math
>>> s[math.floor(len(s)/2)]
' '
>>> s[len(s)//2]    # Pro-tip: the operation // is ideal here
' '

Where am I going with this?

Take a look at the expression you just used:

s[math.floor(len(s)/2)]

Maybe it is me getting old, but I struggle a bit to read that because of the [] enclosing the expression which then has a couple of () that I also have to parse, to figure out what goes where.

If you have large expressions to compute indices (and here, large will be subjective), inserting those expressions directly inside [] may lead to long lines of code that are then complicated to read and understand. If you have lines that are hard to understand, then you probably need to comment them, creating even more lines of code.

Another alternative is to create a well-named variable to hold the result of the computation of the new index:

>>> s = "Indexing is easy!"
>>> mid_char_idx = math.floor(len(s)/2)
>>> s[mid_char_idx]
' '

For this silly example, notice that the new variable name is almost as long as the expression itself! However, s[mid_char_idx] is very, very, easy to read and does not need any further comments.

So, if you have large expressions to compute indices, think twice before using them to index directly into the sequence at hands and consider using an intermediate variable with a descriptive name.

Unpacking with indexing

You will find yourself often working with small groups of data, for example pairs of things that you keep together in a small list for ease of use. For example, the first and last names of a person:

>>> names = ["Mary", "Doe"]

Now you have this little function that creates a formal or informal greeting for a given name:

>>> names = ["Mary", "Doe"]
>>> def greet(names, formal):
...     if formal:
...             return "Hello Miss " + names[1]
...     else:
...             return "Hey there " + names[0]
... 
>>> greet(names, True)
'Hello Miss Doe'
>>> greet(names, False)
'Hey there Mary'

Something you might consider and that adds a bit of clarity to your code is unpacking the names before you reach the if statement:

def greet(names, formal):
    first, last = names
    if formal:
        return "Hello Miss " + last
    else:
        return "Hey there " + first

Why would this be preferable, if I just added a line of code? It makes the intent of the code much more obvious. Just from looking at the function as is, you can see from the first line first, last = names that names is supposed to be a pair with the first and last names of a person and then the if: ... else: ... is very, very easy to follow because we see immediately that we want to use the last name if we need a formal greeting, and otherwise (else) we use the first name.

Furthermore, the action of unpacking (like so:)

first, last = names

forces your greet function to expect pairs as the names variable, because a list with less or more elements will raise an error:

>>> first, last = ["Mary", "Anne", "Doe"] 
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: too many values to unpack (expected 2)

We are assuming we really are working with pairs, so if the greet function gets something that is not a pair, this error is useful in spotting a problem in our code. Maybe someone didn't understand how to use the function and called it with the first name of the person?

>>> greet("Mary", True)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 2, in greet
ValueError: too many values to unpack (expected 2)

This would help you find a location where the greet function was not being properly used.

Conclusion

Here's the main takeaway of this article, for you, on a silver platter:

“Indexing is simple and powerful, but sometimes when indexing looks like the answer, there is another Python feature waiting to be used.”

This Pydon't showed you that:

• Indexing in Python is 0-based;
• Python allows negative indices in sequences;
• Using indices in a for loop to access the elements of a sequence is an anti-pattern in Python;
• Using large expressions when indexing bloats your code and you are better off with a descriptive variable, even if that variable has a long name;
• If you know the exact structure of the sequence you are dealing with, unpacking might be preferable to indexing.

If you liked this Pydon't be sure to leave a reaction below and share this with your friends and fellow Pythonistas. Also, don't forget to subscribe to the newsletter so you don't miss a single Pydon't!

Become a better Python 🐍 developer 🚀

+35 chapters. +400 pages. Hundreds of examples. Over 30,000 readers!

My book “Pydon'ts” teaches you how to write elegant, expressive, and Pythonic code, to help you become a better developer. >>> Download it here 🐍🚀.