Today I learned how many soft keywords Python has and what they are.

Python's soft keywords

Python 3.12 has 4 soft keywords:

  1. match – became a soft keyword in Python 3.10 when structural pattern matching was introduced;
  2. case – see above;
  3. _ – surprisingly, _ was also turned into a soft keyword because of structural pattern matching (this short read will give you a bit more context); and
  4. type – with the introduction of type statements in Python 3.12, type became a soft keyword (it is also a built-in!)

In older versions of Python (I think Python 3.5 and 3.6), async and await were soft keywords too, but that's no longer the case.

What are soft keywords?

Loosely speaking, soft keywords are keywords that can be assigned to, in opposition to hard keywords that can never be assigned to.

if is an example of a hard keyword. No matter what you do, if will always be used to represent the beginning of a conditional statement and you definitely cannot assign to if under no circumstance, as writing something like if = 3 will raise a SyntaxError.

On the other hand, a keyword like match is a soft keyword because you can use it for structural pattern matching, but it's also a legal variable name elsewhere: match = re.match(...) works perfectly well.

How to determine all of Python's soft keywords?

EDIT: After sharing my original method and posting this article on X, Pablo Galindo Salgado – a Python core dev – pointed out I could've simply done the following:

>>> import keyword
>>> keyword.softkwlist
['_', 'case', 'match', 'type']

In order to determine Python's soft keywords, I started by opening the Python grammar (at the time of writing, this file was the most recent grammar version).

  • I opened the Python REPL;
  • I loaded the file; and
  • I used a regular expression to look for all keywords:
>>> import re
>>> pat = re.compile(r'\'.{,15}\'|".{,15}?"')
>>> matches = set(pat.findall(gram))
>>> matches
    "':' &('}'|','",
    "'=', or '!', or '",
    "' | '**'",
    "'+' | '-'",
    "'except' ':'",
    # ...

As you can see, that gives plenty of false positives, but that's ok. We have 222 possible keywords to go through:

>>> len(matches)

However, if we get rid of all of the ones that contain whitespace, only 105 matches are left:

>>> matches = {match for match in matches if " " not in match}
>>> len(matches)

After this, I painstakingly went through all 105 matches and checked manually whether they were soft keywords or not. I took none for granted. I even went through the trouble of checking if and for, for example, so that I could be 100% certain I wouldn't miss a single soft keyword.

Then, I was doing some research on structural pattern matching and I was reading PEP 634 that introduces it, when I read this sentence referring to the Python grammar:

β€œBy convention, hard keywords use single quotes while soft keywords use double quotes.”

This means that I could've used a simpler pattern:

>>> pat = re.compile(r'".{,15}?"')
>>> matches = set(pat.findall(gram))
>>> len(matches)
>>> matches
    '"expected \':\'"',
    '", "',

I was also quite intriguided by the false positive "dsfsdf", but now I can confirm that it is definitely not a hidden keyword in the language!

Become a better Python 🐍 developer πŸš€

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

My book β€œPydon'ts” teaches you how to write elegant, expressive, and Pythonic code, to help you become a better developer. >>> Download it here πŸπŸš€.

Previous Post Next Post

Blog Comments powered by Disqus.