With print, that dictionary is less readable:

>>> print(user)
{'name': 'Trey', 'languages': ['Python', 'JavaScript'], 'address': {'city': 'San Diego', 'state': 'CA'}}

Pretty-printing is great for data structures that are too long for one line. It's especially helpful for nested data structures.

🧵 (2/4)

The pprint.pprint function sorts dictionary keys, but pprint.pp keeps them in their original order.

PDB's pp command pretty-prints too (sorting keys, like pprint does):

(Pdb) pp user
{'address': {'city': 'San Diego', 'state': 'CA'},
'languages': ['Python', 'JavaScript'],
'name': 'Trey'}

🧵 (3/4)

If, like me, you wish pretty printed output in Python was formatted the way that code is usually formatted, try Python 3.15's new "expand=True" argument:

>>> pp(user, expand=True)
{
'name': 'Trey',
'languages': ['Python', 'JavaScript'],
'address': {'city': 'San Diego', 'state': 'CA'},
}

That shows the same (very common) shape that Black & ruff use.

🧵 (4/4)

Python Tip #290 (of 365):

Pretty-print data structures with pp.

>>> from pprint import pp
>>> pp(user)
{'name': 'Trey',
'languages': ['Python', 'JavaScript'],
'address': {'city': 'San Diego', 'state': 'CA'}}

🧵 (1/4)

#Python #DailyPythonTip

Python 3.10 reached end of life a couple weeks ago.

If you maintain a library and you're wondering what's in Python 3.11 that you can use, here's a cheat sheet.

Upgrading to Python 3.14 because you like to stay one Python version behind at all times? This cheat sheet's for you too.

It's nowhere near comprehensive. I've tried to make short but useful.

The printable version shows more features and a support timeline More cheat sheets at http://pym.dev/cheat-sheets

#Python #Python315

Python 3.15 is now out! 🐍🎉

My favorite new features include:

• UTF-8 default file encoding everywhere (less buggy cross-platform reads/writes!)
• a new sampling profiler with nearly no overhead
• explicit lazy imports: lazy import json
• unpacking in comprehensions: [*sub for sub in nested]
• a built-in sentinel type
• frozendict
• pprint(data, expand=True)
• re.prefixmatch
• more helpful error messages & more color

A short demo of all my favorites:
https://youtu.be/OD0QObnCZmA

#Python #Python315

It's not a standalone binary, though: the system it's on still needs Python installed.

Note that third-party dependencies AREN'T bundled automatically. You'll need to install them into src first, with "pip install --target src".

And packages with C extensions (like NumPy) can't be imported from a zip file at all.

More on this:
https://docs.python.org/3/library/zipapp.html

🧵 (3/3)

Python runs that file, just like "python3 -m greet" runs "greet/__main__.py".

Running "python3 greet.pyz" works for any ".pyz" file.

The "-p" option adds a shebang line and makes the file executable (on Linux and macOS), so "./greet.pyz" works too.

On Windows, run "py greet.pyz" instead.

The ".pyz" extension is just a convention. If you rename it to greet and put it in a directory on your PATH, you'll have a greet command instead.

🧵 (2/3)

Python Tip #281 (of 365):

Bundle your package into one runnable file with zipapp.

$ python3 -m zipapp src -m "greet.cli:main" -p "/usr/bin/env python3" -o greet.pyz
$ python3 greet.pyz
Hello from greet!
$ ./greet.pyz
Hello from greet!

That ".pyz" file is a zip file holding everything in src (here, yesterday's greet package) plus a __main__‍.py file that zipapp writes at its root:

# -*- coding: utf-8 -*-
import greet.cli
greet.cli.main()

🧵 (1/3)

#Python #DailyPythonTip

Python 3.6 added the "secrets" module to Python about 10 years ago. At that point, it had a "random" module for many years already.

If you've ever wondered why Python has two different modules for generating random numbers, you might enjoy this article I wrote about it for LWN.

It's subscriber-only at the moment, but this link will give you access to read it: https://lwn.net/SubscriberLink/1097468/337c54eb66d5ba97/

#Python

Python's own packages do this too. The zipfile package has a __main__‍.py file, which is why "python3 -m zipfile" works.

And as of Python 3.14, "python3 -m json" pretty-prints JSON the way "python3 -m json.tool" always has. I opened the PR for that small change.

To see more packages that do this within Python itself, see my article on Python's many command-line utilities: https://pym.dev/cli-tools/

🧵 (3/3)

A __main__‍.py file is usually tiny. It imports the real entry point and calls it:

from .cli import main

main()

Putting the actual code in a submodule (like cli‍.py) means it can also be IMPORTED without running anything (see tip #253 on import side effects).

https://mastodon.social/@treyhunner/117248061679877066

🧵 (2/3)

Python Tip #280 (of 365):

Make your package runnable with a __main__‍.py file.

greet
├── cli‍.py
├── __init__‍.py
└── __main__‍.py

When you run a package with "python3 -m", Python runs its __main__‍.py file:

$ python3 -m greet
Hello from greet!

Without that file, the package can't be run directly:

$ python3 -m food_utils
No module named food_utils.__main__; 'food_utils' is a package and cannot be directly executed

🧵 (1/3)

#Python #DailyPythonTip

Specifying positional-only arguments is rare. But a method that accepts arbitrary keyword arguments is one of the few cases when it's particularly helpful.

🧵 (3/3)

Python got multiple values for self because the self argument is ALWAYS specified for every method.

But we were trying to ADDITIONALLY specify self for capturing in our kwargs dictionary.

We can fix this by putting a / between self and "**kwargs", forcing self to be positional-only:

def __init__(self, /, **kwargs):

Now a self keyword argument is captured into our kwargs dictionary instead:

>>> space = Namespace(other=2, self=1)
>>> space.self
1

🧵 (2/3)

Python Tip #276 (of 365):

Make self positional-only when accepting "**kwargs".

This class accepts arbitrary keyword arguments... except for self:

>>> class Namespace:
... def __init__(self, **kwargs):
... for name, value in kwargs.items():
... setattr(self, name, value)
...
>>> space = Namespace(other=2, self=1)
Traceback (most recent call last):
...
TypeError: Namespace.__init__() got multiple values for argument 'self'

🧵 (1/3)

#Python #DailyPythonTip

Do you use / prefer package-relative imports or absolute ones?

Which do you prefer?

RE: https://hachyderm.io/@nedbat/117283309206194845

I love playing with dates like this to see what weird facts they reveal.

Asking questions about dates and durations is one of the few arithmetic things (that I often do) that's hard with a calculator but super easy in a Python REPL.

Another option is to import the whole module:

import datetime

datetime‍.datetime.‍now()

There's no ambiguity, but I really hate reading datetime.datetime... it's annoyingly repetitive.

You could instead import the whole datetime module but rename it to dt:

import datetime as dt

dt‍.‍datetime.‍now()
dt‍.date.‍today()
dt.timedelta(days=3)

There's no variable shadowing concern, since date and time are still namespaced. And dt‍.datetime.‍now() is pretty readable.

🧵 (2/3)

Python's documentation recently switched to this style: every datetime example in the docs now starts with "import datetime as dt", and the docs recommend it over the other two options:

https://docs.python.org/3/library/datetime.html#timedelta-objects

Thanks to Adam Johnson (@adamchainz), whose 2019 article inspired the discussion that led to that change:

https://adamj.eu/tech/2019/09/12/how-i-import-pythons-datetime-module/

And thanks to Hugo van Kemenade (@hugovk) for updating the Python docs:

https://github.com/python/cpython/pull/145315

🧵 (3/3)