
Security News
MCP Community Begins Work on Official MCP Metaregistry
The MCP community is launching an official registry to standardize AI tool discovery and let agents dynamically find and install MCP servers.
Bunch is a dictionary that supports attribute-style access, a la JavaScript.
b = Bunch() b.hello = 'world' b.hello 'world' b['hello'] += "!" b.hello 'world!' b.foo = Bunch(lol=True) b.foo.lol True b.foo is b['foo'] True
A Bunch is a subclass of dict
; it supports all the methods a dict
does:
b.keys() ['foo', 'hello']
Including update()
:
b.update({ 'ponies': 'are pretty!' }, hello=42) print repr(b) Bunch(foo=Bunch(lol=True), hello=42, ponies='are pretty!')
As well as iteration:
[ (k,b[k]) for k in b ] [('ponies', 'are pretty!'), ('foo', Bunch(lol=True)), ('hello', 42)]
And "splats":
"The {knights} who say {ni}!".format(**Bunch(knights='lolcats', ni='can haz')) 'The lolcats who say can haz!'
Bunches happily and transparently serialize to JSON and YAML.
b = Bunch(foo=Bunch(lol=True), hello=42, ponies='are pretty!') import json json.dumps(b) '{"ponies": "are pretty!", "foo": {"lol": true}, "hello": 42}'
If JSON support is present (json
or simplejson
), Bunch
will have a toJSON()
method which returns the object as a JSON string.
If you have PyYAML_ installed, Bunch attempts to register itself with the various YAML Representers so that Bunches can be transparently dumped and loaded.
b = Bunch(foo=Bunch(lol=True), hello=42, ponies='are pretty!') import yaml yaml.dump(b) '!bunch.Bunch\nfoo: !bunch.Bunch {lol: true}\nhello: 42\nponies: are pretty!\n' yaml.safe_dump(b) 'foo: {lol: true}\nhello: 42\nponies: are pretty!\n'
In addition, Bunch instances will have a toYAML()
method that returns the YAML string using yaml.safe_dump()
. This method also replaces __str__
if present, as I find it far more readable. You can revert back to Python's default use of __repr__
with a simple assignment: Bunch.__str__ = Bunch.__repr__
. The Bunch class will also have a static method Bunch.fromYAML()
, which loads a Bunch out of a YAML string.
Finally, Bunch converts easily and recursively to (unbunchify()
, Bunch.toDict()
) and from (bunchify()
, Bunch.fromDict()
) a normal dict
, making it easy to cleanly serialize them in other formats.
It is safe to import *
from this module. You'll get: Bunch
, bunchify
, and unbunchify
.
Ample doctests::
$ python -m bunch.test $ python -m bunch.test -v | tail -n22 1 items had no tests: bunch.fromYAML 16 items passed all tests: 8 tests in bunch 13 tests in bunch.Bunch 7 tests in bunch.Bunch.contains 4 tests in bunch.Bunch.delattr 7 tests in bunch.Bunch.getattr 3 tests in bunch.Bunch.repr 5 tests in bunch.Bunch.setattr 2 tests in bunch.Bunch.fromDict 2 tests in bunch.Bunch.toDict 5 tests in bunch.bunchify 2 tests in bunch.from_yaml 3 tests in bunch.toJSON 6 tests in bunch.toYAML 3 tests in bunch.to_yaml 3 tests in bunch.to_yaml_safe 4 tests in bunch.unbunchify 77 tests in 17 items. 77 passed and 0 failed. Test passed.
Open a ticket / fork the project on GitHub_, or send me an email at dsc@less.ly
_.
.. _PyYAML: http://pyyaml.org/wiki/PyYAML .. _GitHub: http://github.com/dsc/bunch .. _dsc@less.ly: mailto:dsc@less.ly
FAQs
A dot-accessible dictionary (a la JavaScript objects)
We found that bunch-py3 demonstrated a healthy version release cadence and project activity because the last version was released less than a year ago. It has 1 open source maintainer collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
The MCP community is launching an official registry to standardize AI tool discovery and let agents dynamically find and install MCP servers.
Research
Security News
Socket uncovers an npm Trojan stealing crypto wallets and BullX credentials via obfuscated code and Telegram exfiltration.
Research
Security News
Malicious npm packages posing as developer tools target macOS Cursor IDE users, stealing credentials and modifying files to gain persistent backdoor access.