graph TD;
A[just]-->B["use()"];
B --> C["Path()"]
B --> D["URL()"]
B --> E[packages]
B --> F[git]
To install, enter python -m pip install justuse in a commandline, then you can import use in your code and simply use() stuff. Check for examples below and our Showcase!
Over the years, many times I've come across various situations where Python's import statement just didn't work the way I needed. There were projects where I felt that a central module from where to expand functionality would be the simplest, most elegant approach, but that would only work with simple modules and libs, not with functionality that required access to the main state of the application. In those situations the first thing to try would be "import B" in module A and "import A" in module B - a classical circular import, which comes with a lot of headaches and often results in overly convoluted code. All this could be simplified if it was possible to pass some module-level global variables to the about-to-be-imported module before its actual execution, but how the heck could that work with an import statement?
Later, I worked and experimented a lot in jupyter notebooks. Usually I would have some regular python module only with pure functions open in an editor, import that module in jupyter and have all the state there. Most of the time I'd spend in jupyter, but from time to time I'd edit the file and have to manually reload the module in jupyter - why couldn't it just automatically reload the module when my file is saved? Also, in the past reload() was a builtin function, now I have to import it extra from importlib, another hoop to jump through..
Then there were situations where I found a cool module on github, a single file in a package I wanted to import - but why is it necessary to download it manually or maybe pip install an outdated package? Shouldn't it be possible to import just this exact file and somehow make it secure enough that I don't have to worry about man-in-the-middle attacks or some github-hack?
On my journey I came across many blogposts, papers and presentation-notebooks with code that just 'import library' but mentions nowhere which actual version is being used, so trying to copy and paste this code in order to reproduce the presented experiment or solution is doomed to failure.
I also remember how I had some code in a jupyter notebook that did 'import opencv' but I had not noted which actual version I had initially used. When I tried to run this notebook after a year again, it failed in a subtle way: the call signature of an opencv-function had slightly changed. It took quite a while to track down what my code was expecting and when this change occured until I figured out how to fix this issue. This could've been avoided or at least made a lot simpler if my imports were somehow annotated and checked in a pythonic way. After I complained about this in IRC, nedbat suggested an alternative functional way for imports with assignments: mod = import("name", version) which I found very alien and cumbersome at first sight - after all, we have an import statement for imports, and there should be one and preferrably only one way to do it - right?
Well, those shortcomings of the import statement kept bugging me. And when I stumbled over the word 'use' as a name for whatever I was conceiving, I thought "what the heck, let's try it! - how hard could it be?!" Turns out, some challenges like actual, working hot-reloading are pretty hard! But by now use() can cover all the original usecases and there's even more to come!
Here are a few tidbits on how to use() stuff to wet your appetite, for a more in-depth overview, check out our Showcase!
import use
np = use("numpy", version="1.19.2") corresponds to import numpy as np; if np.version != "1.19.2": warn()
use("pprint").pprint(some_dict) corresponds to a one-off from pprint import pprint; pprint(some_dict) without importing it into the global namespace
tools = use(use.Path("/media/sf_Dropbox/code/tools.py"), reloading=True) impossible to realize with classical imports
test = use("functions", initial_globals={"foo":34, "bar":49}) also impossible with the classical import statement, although importlib can help
utils = use(use.URL("https://raw.githubusercontent.com/PIA-Group/BioSPPy/7696d682dc3aafc898cd9161f946ea87db4fed7f/biosppy/utils.py"), hashes={"95f98f25ef8cfa0102642ea5babbe6dde3e3a19d411db9164af53a9b4cdcccd8"}) no chance with classical imports
np = use("numpy", version="1.21.0rc2", hashes={"3c90b0bb77615bda5e007cfa4c53eb6097ecc82e247726e0eb138fcda769b45d"}, modes=use.auto_install) inline installation of packages and importing the same package with different versions in parallel in the same code - most people wouldn't even dream of that!
Nope. SHA256 hashes normally are pretty long (64 characters per hexdigest) and we require them defined within regular python code. Additionally, if you want to support multiple platforms, you need to supply a hash for every platform - which can add up to huge blocks of visual noise. Since Python 3 supports unicode by default, why not use the whole range of printable characters to encode those hashes? It's easier said than done - turns out emojis don't work well across different systems and editors - however, it is feasible to merge the Japanese, ASCII, Chinese and Korean alphabets into a single, big one we call JACK - which can be used to reliably encode those hashes in merely 18 characters. Since humans aren't supposed to manually type those hashes but simply copy&paste them anyway, there is only the question how to get them if you only have hexdigests at hand for some reason. Simply do use.hexdigest_as_JACK(H) and you're ready to go. Of course we also support classical hexdigests as fallback.
Inspired by the q package/module, use() is a subclass of ModuleType, which is a callable class that replaces the module on import, so that only 'import use' is needed to be able to call use() on things.
Probably the most magical thing about use() is that it does not return the plain module you wanted but a ProxyModule instead which adds a layer of abstraction. This allows things like automatic and transparent reloading without any intervention needed on your part. ProxyModules also add operations on modules like aspectizing via mod @ (check, pattern, decorator) syntax, which would not be possible with the classical import machinery.
FAQs
a pure-python alternative to import
We found that justuse 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
In this segment of the Risky Business podcast, Feross Aboukhadijeh and Patrick Gray discuss the challenges of tracking malware discovered in open source softare.
Research
Security News
A threat actor's playbook for exploiting the npm ecosystem was exposed on the dark web, detailing how to build a blockchain-powered botnet.
Security News
NVD’s backlog surpasses 20,000 CVEs as analysis slows and NIST announces new system updates to address ongoing delays.