This article covers the basics of sequence slicing in Python and teaches you some idiomatic slicing patterns to write more elegant code.

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

Introduction

Last time we went over sequence indexing in Python to cover the basics for our next topic of discussion: sequence slicing. Slicing is a “more advanced” way of accessing portions of sequences (like lists and tuples). I say it is more advanced just because indexing is the simplest form of accessing sequence items; as you will see, indexing isn't that complicated either.

As it turns out, much can be said about sequence slicing, so I will split all of the contents into two Pydon'ts, this and the next one.

In this Pydon't you will:

• learn the slicing syntax;
• learn how slicing works with 1 and 2 parameters;
• relate slices to the range built-in;
• master slicing with negative indices;
• learn to write Pythonic and idiomatic slices; and
• a couple of common use cases where slicing is not the way to go.

In the next Pydon't we will continue on this train of thought and cover the more advanced material related to slicing. In particular, you will

• learn about the step parameter in slicing;
• learn about slice assignment;
• see how slicing can be used to copy sequences;
• learn some more idiomatic slicing patterns;
• uncover the two layers of syntactic sugar surrounding list slicing; and
• learn how to implement slicing for your custom objects.

Throughout both Pydon'ts we will try to keep an eye out for how slices are actually used in real-world Python code, namely in the Python Standard Library.

If you don't want to miss the next Pydon't on the more advanced slicing topics, you can either subscribe to the Pydon'ts newsletter or grab your copy of the Pydon'ts book right now.

Slicing syntax

Slicing in Python is the act of accessing a sequence of elements that are extracted from successive positions of a larger sequence. Just think of an actual knife cutting through the sequence you are working with (which could be a string, list, tuple, etc) and extracting a smaller piece of your sequence.

For example, if we were working with the string "Slicing is easy!", which I present below.

Together with the characters of the string, we have the little numbers indicating the index of each character. Each little number gives the index for the box right in front of it. This is the representation I go to in my head whenever I have to reason about indices in Python, especially when I am working with negative indices. (Just take a quick look at this Pydon't if you need to jog your memory on how indexing is done in Python.)

Now, we could be interested in extracting the portion "icing" from the string:

How would we do that in Python? If you didn't know how slicing worked, you could come up with a solution involving a for loop and a range:

>>> s = "Slicing is easy!"
>>> subs = ""
>>> for i in range(2, 7):
...     subs += s[i]
...
>>> subs
'icing'

This is all good, but there is a much shorter syntax for this type of operation, the slicing syntax.

When you want to slice a sequence, you need to use brackets [] and a colon : to separate the start and end points. The key here is in figuring out what the start and end points are, but that is just a matter of looking at the figure above or at the solution with the range(2, 7):

>>> s = "Slicing is easy!"
>>> s[2:7]
'icing'

This is the very first important point to make about slicing: the start and end points give you the bars that enclose what you will extract, which, in other words, means that the start point (2, in the previous example) is the index of the first element that is included in the slice, whereas the end point is the index of the first element that is not included in the slice:

>>> s = "Slicing is easy!"
>>> s[2:7]
'icing'
>>> s[7]
' '

Now is a good time to fire up your Python interpreter, define s as the string "Slicing is easy!", and work out a couple of slices for yourself.

What to slice?

Just in case it wasn't clear earlier, here are just some of the things that you can slice in Python:

>>> "Hello"[1:3]                    # strings
'el'
>>> [True, False, 1, "hey"][1:3]    # lists
[False, 1]
>>> (True, False, 1, "hey")[1:3]    # tuples
(False, 1)
>>> range(10)[1:3]                  # ranges
range(1, 3)
>>> # etc...

However, we will be using string examples for most of the Pydon't, just for the sake of consistency.

Slicing from the beginning

Now assume that we wanted to extract "Slicing" from our string.

![](_slicing_slicing.svg "The substring “Slicing” within “Slicing is easy!”)

If we go back to our naïve range solution, most of us would write the following:

>>> s = "Slicing is easy!"
>>> subs = ""
>>> for i in range(7):
...     subs += s[i]
...
>>> subs
'Slicing'

Notice that, unlike when we used range(2, 7) for "icing", now our range only has one argument, the end point. That is because range interprets the missing starting index as 0.

When we are slicing, we can do a similar thing! If we want to extract a portion from the beginning of a sequence, the Pythonic way of writing that slice is without specifying the explicit 0 as a start point. Therefore, both alternatives below work, but the second one is the preferred.

>>> s = "Slicing is easy!"
>>> s[0:7]      # Works ...
'Slicing'
>>> s[:7]       # ... but this is preferred!
'Slicing'

In terms of the figures I have been sharing, think of it like this: it's like you never tell Python where the slicing starts, so the bar that is hovering the string ends up covering the whole beginning of the string, stopping at the position you indicate.

Slicing until the end

Similarly to omitting the start point, you can omit the end point of your slice. To predict what will happen if we do so, we just have to look at the figure above and create a new one with the bar pointing in the other direction:

Therefore, if we don't indicate the end point for the slice, we extract all elements from the point specified, onwards. Naturally, we can specify the end point of the slice to be the length of the sequence, but that adds too much visual noise:

>>> s = "Slicing is easy!"
>>> s[7:len(s)]     # Works...
' is easy!'
>>> s[7:]           # ... but this is preferred!
' is easy!'

Slicing with negative indices

If you've read my previous Pydon't, you will have seen how indexing works with negative indices. Slicing can also use them, and the logic remains the same: we just draw the bar that selects the elements that are between the relevant indices.

To illustrate this, here is the representation of the negative indices of the string we have been using so far:

Now, regardless of the fact that the numbers are negative, if you had to tell me where to draw two vertical bars in order to enclose the substring "icing", what positions would you point to? You would probably tell me “Draw the bars on positions -14 and -9”, and that would be absolutely correct!

In fact, using -14 and -9 would work in my naïve range solution but also – and most importantly – with the slice syntax:

>>> s = "Slicing is easy!"
>>> subs = ""
>>> for i in range(-14, -9):
...     subs += s[i]
...
>>> subs
'icing'
>>> s[-14:-9]       # Also works and is preferred!
'icing'

Slicing and range

At this point you should start to notice a pattern, and that is that the parameters you insert in your slices seem to be governing the indices that Python uses to fetch elements from your sequence, if those indices were generated with the range function. If you are looking at a slice and you have no clue what items are going to be picked up by it, try thinking about the slice in this way, with the range. It might help you.

Idiomatic slicing patterns

Now that you have taken a look at some basic slicing with positive and negative indices, and now that you know you can omit the first or the last parameters of your slices, you should really learn about four different slice patterns that are really idiomatic. Don't worry, I'll show you which four patterns I am talking about.

Suppose you have a variable n that is a positive integer (it may help to think of it as a small integer, like 1 or 2), and suppose s is some sequence that supports slicing. Here are the four idiomatic slicing patterns I am talking about:

• s[n:]
• s[-n:]
• s[:n]
• s[:-n]

Why are these “idiomatic” slicing patterns? These are idiomatic because, with a little practice, you stop looking at them as “slice s starting at position blah and ending at position blah blah”, and you will start looking at them for their semantic meaning.

Open your Python interpreter, set s = "Slicing is easy!" and n = 2, and see what the four slices above return. Experiment with other values of n. Can you give an interpretation for what each slice means?

Go ahead...

Here is what the slicing patterns mean.

s[n:]

If n is not negative (so 0 or more), then s[n:] means “skip the first n elements of s”:

>>> s = "Slicing is easy!"
>>> s[2:]
'icing is easy!'
>>> s[3:]
'cing is easy!'
>>> s[4:]
'ing is easy!'

s[-n:]

If n is positive (so 1 or more), then s[-n:] means “the last n elements of s”:

>>> s = "Slicing is easy!"
>>> s[-2:]
'y!'
>>> s[-3:]
'sy!'
>>> s[-4:]
'asy!'

>>> s = "Slicing is easy!"
>>> s[-0:]
'Slicing is easy!''

s[:n]

If n is not negative (so 0 or more), then s[:n] can be read as “the first n elements of s”:

>>> s = "Slicing is easy!"
>>> s[:2]
'Sl'
>>> s[:3]
'Sli'
>>> s[:4]
'Slic'

s[:-n]

Finally, if n is positive (so 1 or more), then s[:-n] means “drop the last n elements of s”:

>>> s = "Slicing is easy!"
>>> s[:-2]
'Slicing is eas'
>>> s[:-3]
'Slicing is ea'
>>> s[:-4]
'Slicing is e'

Summary

Here is a short summary with the four slicing patterns next to each other:

# Four idiomatic uses of sequence slicing in Python.
>>> s = "Slicing is easy!"
>>> s[4:]           # Drop the first 4 characters.
'ing is easy!'
>>> s[-4:]          # Take the last 4 characters.
'asy!'
>>> s[:4]           # Take the first 4 characters.
'Slic'
>>> s[:-4]          # Drop the last 4 characters.
'Slicing is e'

Empty slices

Something worthy of note that may confuse some but not others, is the fact that if you get your start and end points mixed up, you will end up with empty slices, because your start point is to the right of the end point... And because of negative indices, it is not enough to check if the start point is less than the end point.

Take a look at the figure below:

Now try to work out why all of these slices are empty:

>>> s = "Slicing is easy!"
>>> s[10:5]
''
>>> s[-6:-10]
''
>>> s[-9:3]
''
>>> s[10:-10]
''

All it takes is looking at the figure above, and realising that the end point is relative to an index that is to the left of the start point.

More empty slices

Another really interesting thing about slicing is that if you use numbers that are too high or too low, slicing still works and doesn't raise IndexError or something like that. In a way, this makes sense, and goes in line with the interpretation of the idioms we presented above.

If s[50:] is “skip the first 50 elements of s”, and if s only has 16 elements, how many will there be left? Zero, naturally, so it is no surprise that s[50:] returns an empty string when s = "Slicing is easy!":

>>> s = "Slicing is easy!"
>>> s[50:]
''

This segues nicely into the first common usage pattern of slices.

Examples in code

Ensuring at most n elements

Imagine someone is writing a spellchecker, and they have a function that takes a misspelled word and returns the top 5 closest suggestions for what the user meant to type.

Here is what that function could look like:

def compute_top_suggestions(misspelled, k, corpus):
    similar = find_similar(misspelled, corpus)
    ordered = rank_suggestions_by_similarity(misspelled, similar)
    top_k = []
    for i in range(min(k, len(ordered))):
        top_k.append(ordered[i])
    return top_k

The final loop there is to make sure you return at most k results. However, the person who wrote this piece of code did not read this Pydon't! Because if they had, they would know that you can use slicing to extract at most k elements from ordered:

def compute_top_suggestions(misspelled, k, corpus):
    similar = find_similar(misspelled, corpus)
    ordered = rank_suggestions_by_similarity(misspelled, similar)
    return ordered[:k]
    # ^ Idiom! Read as “return at most `k` from beginning”

A very similar usage pattern arises when you want to return at most k from the end, but you already knew that because you read about the four slicing idioms I shared earlier.

This usage pattern of slicing can show up in many ways, as this is just us employing slicing because of the semantic meaning this particular idiom has. Above, we have seen four different idioms, so just keep those in mind with working with sequences!

Start of a string

Slicing is great, I hope I already convinced you of that, but slicing is not the answer to all of your problems!

A common use case for slices is to check if a given sequence starts with a predefined set of values. For example, we might want to know if a string starts with the four characters ">>> ", which are the characters that mark the REPL Python prompt. The doctest Python module, for example, does a similar check, so we will be able to compare our solution to doctest's.

You just learned about slicing and you know that s[:4] can be read idiomatically as “the first four characters of s”, so maybe you would write something like

def check_prompt(line):
    if line[:4] == ">>> ":
        return True
    return False

or, much more elegantly,

def check_prompt(line):
    return line[:4] == ">>> "

However, it is important to note that this is not the best solution possible, because Python strings have an appropriate method for this type of check: the startswith function.

Therefore, the best solution would be

def check_prompt(line):
    return line.startswith(">>> ")

This is better because this is a tested and trusted function that does exactly what you need, so the code expresses very clearly what you want. What is more, if you later change the prompt, you don't need to remember to also change the index used in the slice.

If we take a look at the actual source code for doctest, what they write is

# Inside _check_prefix from Lib/doctest.py for Python 3.9.2
# ...
if line and not line.startswith(prefix):
    # ...

As we can see here, they are using the startswith method to see if line starts with the prefix given as argument.

Removing prefixes and suffixes

Similar to the example from above, another common usage pattern for slices is to remove prefixes or suffixes from sequences, more notably from strings. For example, most of the code I present in these Pydon'ts starts with ">>> " because of the REPL prompt. How could I write a short function to strip a line of code of this prompt? I am really hyped about slicing, so obviously I'll do something like

>>> def strip_prompt(line):
...     if line.startswith(">>> "):
...         return line[4:]
...     else:
...         return line
...
>>> strip_prompt(">>> 3 + 3")
'3 + 3'
>>> strip_prompt("6")
'6'

Or even better, I might do it in a generic way and suppress the else:

>>> def strip_prefix(line, prefix):
...     if line.startswith(prefix)
...         return line[len(prefix):]
...     return line
...
>>> prompt = ">>> "
>>> strip_prefix(">>> 3 + 3", prompt)
'3 + 3'
>>> strip_prefix("6", prompt)
'6'

However, I already have Python 3.9 installed on my machine, so I should be using the string methods that Python provides me with:

def strip_prefix(line, prefix):
    return line.removeprefix(prefix)

Of course, at this point, defining my own function is redundant and I would just go with

>>> prompt = ">>> "
>>> ">>> 3 + 3".removeprefix(prompt)
'3 + 3'
>>> "6".removeprefix(prompt)
'6'

In case you are interested, Python 3.9 also added a removesuffix method that does the analogous, but at the end of strings.

This just goes to show that it is nice to try and stay more or less on top of the features that get added to your favourite/most used programming languages. Also (!), this also shows that one has to be careful when looking for code snippets online, e.g. on StackOverflow. StackOverflow has amazing answers... that get outdated, so always pay attention to the most voted answers, but also the most recent ones, those could contain the more modern approaches.

Conclusion

Here's the main takeaway of this Pydon't, for you, on a silver platter:

“The relationship between slicing and indexing means there are four really nice idiomatic usages of slices that are well-worth knowing.”

This Pydon't showed you that:

• slicing sequences lets you access series of consecutive elements;
• you can slice strings, lists, tuples, ranges, and more;
• if the start parameter is omitted, the slice starts from the beginning of the sequence;
• if the end parameter is omitted, the slice ends at the end of the sequence;
• slicing is the same as selecting elements with a for loop and a range with the same parameters;
• much like with plain indexing, negative integers can be used and those count from the end of the sequence;
• s[n:], s[-n:], s[:n], and s[:-n] are four idiomatic slicing patterns that have a clear semantic meaning:
  • s[n:] is “skip the first n elements of s”;
  • s[-n:] is “the last n elements of s”;
  • s[:n] is “the first n elements of s”;
  • s[:-n] is “skip the last n elements of s”;
• slices with parameters that are too large produce empty sequences;
• if the parameters are in the wrong order, empty sequences are produced; and
• some operations that seem to ask for slicing might have better alternatives, for example using startswith, endswith, removeprefix, and removesuffix with strings.

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 🐍🚀.

References

• Python 3 Documentation, The Python Language Reference, Expressions – Slicings, https://docs.python.org/3/reference/expressions.html#slicings [last acessed 20-04-2021];
• Python 3 Documentation, The Python Language Reference, Data Model – The Standard Type Hierarchy, https://docs.python.org/3/reference/datamodel.html#the-standard-type-hierarchy [last accessed 20-04-2021];
• Python 3 Documentation, The Python Language Reference, Data Model – Emulating Container Types, https://docs.python.org/3/reference/datamodel.html#emulating-container-types [last accessed 20-04-2021];
• Python 3 Documentation, The Python Language Reference, Built-in Functions, slice, https://docs.python.org/3/library/functions.html#slice [last accessed 20-04-2021];
• “Effective Python – 90 Specific Ways to Write Better Python”; Slatkin, Brett; ISBN 9780134853987.