from box import Box

movie_box = Box({ "Robin Hood: Men in Tights": { "imdb stars": 6.7, "length": 104 } })

movie_box.Robin_Hood_Men_in_Tights.imdb_stars
# 6.7

Box will automatically make otherwise inaccessible keys safe to access as an attribute. You can always pass conversion_box=False to Box to disable that behavior. Also, all new dict and lists added to a Box or BoxList object are converted automatically.

There are over a half dozen ways to customize your Box and make it work for you.

Check out the new Box github wiki for more details and examples!

Install

Version Pin Your Box!

If you aren't in the habit of version pinning your libraries, it will eventually bite you. Box has a list of breaking change between major versions you should always check out before updating.

requirements.txt

python-box[all]~=8.0

As Box adheres to semantic versioning (aka API changes will only occur on between major version), it is best to use Compatible release matching using the ~= clause.

Install from command line

python -m pip install --upgrade pip
pip install "python-box[all]~=8.0" --upgrade

Install with selected dependencies

Box does not install external dependencies such as yaml and toml writers. Instead you can specify which you want, for example, [all] is shorthand for:

pip install "python-box[ruamel.yaml,toml,msgpack]~=8.0" --upgrade

But you can also sub out ruamel.yaml for PyYAML. To use tomli-w for TOML output, select the tomli extra instead of toml.

Check out more details on installation details.

Box 8 is tested on python 3.11+, if you are upgrading from previous versions, please look through any breaking changes and new features.

Optimized Version

Box has introduced Cython optimizations for major platforms by default. Loading large data sets can be up to 10x faster!

If you are not on a x86_64 supported system you will need to do some extra work to install the optimized version. There will be an warning of "WARNING: Cython not installed, could not optimize box" during install. You will need python development files, system compiler, and the python packages Cython, setuptools and wheel.

Linux Example:

First make sure you have python development files installed (python3-dev or python3-devel in most repos). You will then need Cython, setuptools and wheel installed and then install (or re-install with --force) python-box. As Cython is not a build requirement, the build has to run without isolation so it can see your installed copy.

pip install Cython "setuptools>=77" wheel
pip install "python-box[all]~=8.0" --upgrade --force --no-binary python-box --no-build-isolation

If you have any issues please open a github issue with the error you are experiencing!

Overview

Box is designed to be a near transparent drop in replacements for dictionaries that add dot notation access and other powerful feature.

There are a lot of types of boxes to customize it for your needs, as well as handy converters!

Keep in mind any sub dictionaries or ones set after initiation will be automatically converted to a Box object, and lists will be converted to BoxList, all other objects stay intact.

Check out the Quick Start for more in depth details.

Box can be instantiated the same ways as dict.

Box({'data': 2, 'count': 5})
Box(data=2, count=5)
Box({'data': 2, 'count': 1}, count=5)
Box([('data', 2), ('count', 5)])

# All will create
# <Box: {'data': 2, 'count': 5}>

Box is a subclass of dict which overrides some base functionality to make sure everything stored in the dict can be accessed as an attribute or key value.

small_box = Box({'data': 2, 'count': 5})
small_box.data == small_box['data'] == getattr(small_box, 'data')

All dicts (and lists) added to a Box will be converted on insertion to a Box (or BoxList), allowing for recursive dot notation access.

Box also includes helper functions to transform it back into a dict, as well as into JSON, YAML, TOML, or msgpack strings or files.

Internal Attributes on Subclasses

If you subclass Box and need attributes that are not part of the data (a source path, a cache, a logger...), list their names in _box_internal_attrs. Setting or deleting those names with attribute access stores them on the instance, like a normal Python object, instead of in the dictionary.

from box import Box

class Config(Box):
    _box_internal_attrs = ("source_path", "_cache")

    def __init__(self, *args, source_path=None, **kwargs):
        super().__init__(*args, **kwargs)
        self.source_path = source_path
        self._cache = {}

c = Config({"name": "app"}, source_path="/etc/app.yml")
c.source_path        # '/etc/app.yml'
c.to_dict()          # {'name': 'app'}
"source_path" in c   # False

Internal attributes are excluded from to_dict, to_json, to_yaml, to_toml, to_msgpack, keys(), iteration, in, len, equality and hashing. They survive pickle, and are carried over by copy(), copy.copy (shared, shallow) and copy.deepcopy (deep copied). They can be set and deleted even on a frozen_box, since they are not data.

Things to keep in mind:

• Nested boxes are created through your subclass (as box_class defaults to it), so every nested box gets its own fresh internal attributes from __init__, they are not shared with the parent.
• Only attribute access is affected. c["source_path"] still refers to the dictionary data, and if a data key and an internal attribute share a name, attribute access returns the internal attribute.

Thanks

A huge thank you to everyone that has given features and feedback over the years to Box! Check out everyone that has contributed.

A big thanks to Python Software Foundation, and PSF-Trademarks Committee, for official approval to use the Python logo on the Box logo!

Also special shout-out to PythonBytes, who featured Box on their podcast.

License

MIT License, Copyright (c) 2017-2026 Chris Griffith. See LICENSE file.