Why you should use Catch-All Unpacking and not Slicing in Python

Francis Ali 👨‍🎨 — Sun, 31 Jan 2021 00:42:13 +0000

When it comes to programming, there is the correct way of doing something, and then there is the right way of doing it. There are many cases within the Python language where this is true.

Lists are a fundamental data structure in Python, with a wide variety of use-cases. There are many instances where we require algorithms to split a list into "first" and "rest" pairs. Which is usually done with the use of indexing and slicing. For example:

names = ["Alice", "Bob", "Carol", "Dave", "Elon"]
alice = names[0]
bob = names[1]
everyone_else = names[2:]

The above code snippet is correct however not right for a number of reasons:

we've taken up three lines of code in order to accomplish something so simple
we need to know the ordering in advance
error prone as it can lead to off-by-one errors

Why is it error prone? 😔

What could be the case in the future is that we may want to change the boundaries or the index values, and forget to do so for the others. This of course can lead to errors or unpredictable results.

Catch-All Unpacking to the rescue 🔥

To remedy this situation, Python supports catch-all unpacking through the use of a starred expression. This will allow us to achieve the same result as above, without having to make use of indexing or slicing.

names = ["Alice", "Bob", "Carol", "Dave", "Elon"]
alice, bob, *everyone_else = names

🤯 Now isn't that a thing of beauty! We have now:

shortened it to one line
made it easier to read
no longer have to deal with the error prone brittleness of boundary indexes that must be kept in sync between lines.

More on starred expressions ⭐️

The great thing about starred expressions is that we can place them in any position. This way you can get the benefits of catch-all unpacking anytime you need to extract one slice:

names = ["Alice", "Bob", "Carol", "Dave", "Elon"]
alice, *everyone_else, elon = names

Some Gotchas 🧐

There are two gotchas when using catch-all unpacking. The first one is to remember that you must have at least one required part, otherwise you will face a SyntaxError. For example:

*everyone = names

Will result in:

SyntaxError: starred assignment target must be in a list or tuple

The second is that you cannot make use of multiple catch-all expressions in a single-level unpacking structure:

alice, *some_people, *other_people, elon = names

Which will result in:

SyntaxError: multiple starred expressions in assignment

Catch-All Unpacking with a multi-level structure 💪

The second gotcha is true for single-level structures (lists, tuples). However this plays a little differently if we're using multi-level structures.
For example:

classrooms = {
    "classroom 1": ["Alice", "Bob", "Carol"],
    "classroom 2": ["David", "Elon"],
}
(
    (class_1, (best_student_1, *other_students_1)),
    (class_2, (*other_students_2, best_student_2)),
) = classrooms.items()

print(
    f"The best student in {class_1} is {best_student_1} not the other {len(other_students_1)}"
)
print(
    f"The best student in {class_2} is {best_student_2} not the other {len(other_students_2)}"
)

With the output being:

The best student in classroom 1 is Alice not the other 2
The best student in classroom 2 is Elon not the other 1

This may look a bit freakish with the brackets, however it is much more readable and cleaner than if we were to use slicing and indexing.

Other things to note 📝

No more unpacking

Used correctly, starred expressions will always result in lists. However if there are no more items to unpack from a sequence, it will result in an empty list [].

books = ["Harry Potter", "Narnia"]
harry_potter, narnia, *other_books = books
print(harry_potter)
print(narnia)
print(other_books)

Output

Harry Potter
Narnia
[]

Catch-All Unpacking on Iterators

The other great thing about Catch-All Unpacking is that we can also use them on iterators. For example:

movies = iter(["Star Wars", "The Godfather", "The Matrix", "Goodfellas"])
star_wars, the_godfather, *other_movies = movies

Although the above snippet is very basic and should in reality just be replaced with a standard list. The idea however is that you could extend this if you were dealing with a CSV file (for example).

Final takeaways

This more-or-less sums up why you should prefer Catch-All Unpacking over Slicing and Indexing. The key takeaways here are:

catch-all is visually cleaner compared to slicing and indexing
it is a lot less error prone
starred expressions can appear in any position and will always result in a list, containing zero or more elements

If you enjoyed this post. Be sure to follow me on Twitter: @sixfwa
GitHub: @sixfwa

DEV Community: Francis Ali 👨‍🎨