jtpavlock/nbapy

Python API for stats.nba.com

nbapy - stats.nba.com API for python

Summary

A python facing API for stats.nba.com

Warning stats.nba.com is notorious for being extremely unreliable. Please report any issues you find.

Usage

All data is returned as a pandas dataframe (check out the starter docs if you're new to pandas). For example:

from nbapy import game
import pandas as pd

game_id = '0021900017'  # taken from 'https://stats.nba.com/game/0021900017/'
stats = game.BoxScore(game_id).players_stats()
print(stats)

If you want to cache results so you don't have to reach the api every time, you can use requests-cache

from nbapy import game
import pandas as pd
import requests_cache

requests_cache.install_cache('nbapy_cache')

game_id = '0021900017'
stats = game.BoxScore(game_id).players_stats()
print(stats)

Documentation

A constant work-in-progress, but check them out!

Installation

To install from pypi:

$ python -m pip install nbapy

Contributing

1. Fork the repository and create a feature/bug fix branch

2. Install development requirements

You will to first install poetry if you don't already have it.

$ python -m pip install poetry

After that, just run poetry install inside of your fork.

3. Hack away

Coding conventions

• black for formatting
• google docstrings
• flake8 for linting
• mypy for static typing analysis
• conventional commits for commit style.
• isort for import organization.

Whew, that's a lot, but I'm a big fan of clean code, and I hope you are too! The good news is that if you follow the following advice, you'll find these aren't too hard to manage 😄

Optional (but recommended)

nbapy has a pre-commit file that you can install to automatically enforce these conventions prior to committing via a git hook.

To install: $ pre-commit install

You can also use $ pre-commit run -a to run the checks manually.

For commit messages, I recommend using commitizen. It is automatically installed in the dev dependencies, so to commit, you just run cz c and follow the prompts.

If you're using pre-commit, and either the black or isort check fails, the good news is it will fix the problems for you. However, it won't continue the commit automatically so that you get a chance to look over the changes. You'll have to re-stage the changes, and then you can run cz c --retry so you don't lose that nice commit message you just wrote.

4. Create some tests

5. Make sure everything looks good

$ pytest --cov*

$ sphinx-build -n -W -q -b html docs docs/_build/html

$ pre-commit run -a (if you didn't install the pre-commit git hook)

* note the first time you run this, it may take a few minutes. However, the requests will cache, and subsequent runs should be much faster.

6. Submit a pull request

Other ways to contribute involve submitting any issues or adding some documentation!

To-Do

• Finish Jupyter Notebook documentation

Authors

This is orginally based off of https://github.com/seemethere/nba_py so a lot of the work was done by those guys. My goal with this project is to clean up the code, add some proper documentation, and keep it up to date.

nbapy's Issues

Request Time out for any API call

Hey I am picking back up my project and was using this code again. I notice now that any API call I make times out. Are you seeing that also?

from nbapy import game
import pandas as pd

game_id = '0021900017' # taken from 'https://stats.nba.com/game/0021900017/'
stats = game.BoxScore(game_id).players_stats()
print(stats)

Need Help Looping thru BoxScores

i am trying just to get 2nd dataframe in the return, and loop thru all games

from nbapy import game
game_id = '0021900017'
stats = game.BoxScore(game_id).players_stats()

New to Python but have question for you

Hey man thanks so much for sharing this repo! It's been great for me doing some exploratory analysis. I am running into a timeout issue though which I think might be a function of the NBA stats side. I am trying to get the player information for all NBA players. Your documentation allowed me to get the information for an individual player very easily. I am looping through to get for all players and am consistently running into a timeout issue. Do you know if the NBA stats prohibits this type of behavior? Or could I change the timeout setting myself?

summary = Summary(1629027)
summary.info()

Running Box Score example

When executing
pd.DataFrame(game.BoxScore(game_id).players_stats())
there seems to be an exceptionally long execution time with no output.
Is this package still useable?