How to import a module that is installed in the venv?

[marcsyp]

I am writing an extension and would like to use kneed/KneeLocator. I have successfully gotten it installed using the install.py script of my extension, but I can't for the life of me figure out how to import the module into my code. New to this type of dev environment, coming from a long C#/VS background and just getting into the groove in Python/A1111, so any help appreciated.

Thanks!

[w-e-w]

so I have a feeling that you may not know what a venv is
so I'll give you a quick crash course that's all you really need to know

a venv it's like the isolated environment that you can install python packages for a project so that it doesn't interfere with others project
to use a venv you need to first "enter" it

the simplest way to do so is in your terminal a command like

PS C:\> <venv>\Scripts\Activate.ps1

this is for powershell different terminals would have different commands see how to activated venv

if you use an IDE or have some other integration configured, the most likely automatically will enter the venv for you so you don't need to do it yourself

once you're inside of venv you can use commands like pip to install packages
venv is python version dependent, so if you want to run 2 version of python say 3.10 and 3.11 you need 2 separate venv
any python script executed in the venv will be automatically have access to the packages installed in it just using normal import XXX in you .py script
there's no special code that you need to use the import packages from within the venv it is essentially transparent to python interpreter

if you wish to know more you can read the docs
https://docs.python.org/3/library/venv.html

[marcsyp]

Thanks for the explanation @w-e-w -- I do know what a venv is but don't have a lot of practical experience working with it within my own plugin development, this is the first time I have needed to add a dependency to my own extension. A few notes:

1. I am using VS Code and the venv does seem to be registered properly, as I can import any of the preexisting modules that have been installed by SD or other extensions. For instance, Shift Attention installs imageio-ffmpeg using the following install.py:

import launch

if not launch.is_installed("imageio-ffmpeg"):
launch.run_pip("install imageio-ffmpeg", "requirements for Shift Attention")

And I can import in my extensions module with no problem:

2. I am following the same pattern and installing kneed via install.py:
   import launch

if not launch.is_installed("kneed"):
launch.run_pip("install kneed", "requirements for Space Explorer")

After running SD once and observing that kneed has been installed into the venv:

For some reason I cannot import kneed, or the KneeLocator class from within it, there is just no recognition that it exists and if I just brute force type it in and run my code, it results in a module not found:

I have never had to activate my venv environment for VS Code to register modules in the IDE, I've only ever done that when manually installing a dependency or updating torch or other such activities.

What am I missing?

[w-e-w]

that's really odd
from what I can tell it should work
I prepared a minimum test and it did

for importing using either methods works
you can see the breakpoint has been hit (it wouldn't if there's imported error)

this is my test code

install.py

import launch
if not launch.is_installed('kneed'):
    launch.run_pip('install kneed')

scripts/what_ever.py

import kneed
print(kneed.DataGenerator, kneed.KneeLocator)
from kneed import DataGenerator, KneeLocator
print(DataGenerator, KneeLocator)

if your code is on GitHub then maybe I can take a look that's all I can offer

[marcsyp]

Turns out ChatGPT was able to help me solve the problem. Turns out my VS Code was pointed at the wrong venv -- a backup copy. Here is the response that GPT gave me:

It sounds like you're encountering an issue with module visibility within your virtual environment in VS Code. Here are a few steps you can take to troubleshoot and potentially resolve the problem:

Verify Installation: First, double-check that kneed was indeed installed correctly within your virtual environment. You can do this by navigating to your virtual environment directory and checking if kneed is listed in the Lib/site-packages directory.
Restart VS Code: Sometimes, VS Code might not recognize newly installed packages until it's restarted. Close VS Code and reopen it to see if the issue persists.
Check Python Interpreter: Ensure that VS Code is using the correct Python interpreter associated with your virtual environment. You can do this by clicking on the Python interpreter version in the bottom-left corner of the VS Code window. Select your virtual environment from the list if it's not already selected.
Check Python Path: Verify that the Python path in VS Code points to the correct virtual environment. You can do this by opening a terminal in VS Code and running which python or where python depending on your operating system. It should point to the python executable within your virtual environment.
Install Dependencies Manually: Instead of relying on the install.py script, try installing kneed manually within your virtual environment using the terminal in VS Code. Run pip install kneed and check if you can import it afterwards.
Check for Typos: Ensure that you're importing the module and class correctly in your code. Even a small typo could lead to a "module not found" error.
Restart Kernel: If you're using a Jupyter notebook or an interactive Python session in VS Code, try restarting the kernel to see if that resolves the issue.
Recreate Virtual Environment: If none of the above steps work, consider recreating your virtual environment. Sometimes, virtual environments can become corrupted or misconfigured, leading to unexpected behavior.
By following these steps, you should be able to diagnose and resolve the issue with importing kneed and its KneeLocator class within your virtual environment in VS Code.

#3 was the right answer -- and it revealed that I should probably explocitly define my venv path in my startup script to ensure everything is synchronized.

Regardless, thanks for the time and effort you put trying to help! Much appreciated.

[w-e-w]

nice