Import resolution not respecting virtual environments

[ITProKyle]

Describe the bug

When updating from 1.1.321 to 1.1.327, I am receiving reportMissingImports in all of my projects. I use poetry to manage python dependencies & virtual environment for both local deployment and pipeline execution.

EDIT: Tested with 1.1.326 and this bug is not present so it was introduced in 1.1.327

When running pyright, I run npm exec --no -- pyright --venvpath ./ with venv = ".venv" set in my pyproject.toml. I typically (especially in pipeline executions) do not activate the virtual environment prior to running pyright.

For all versions prior to 1.1.327, this worked fine. However, after updating pyright raises reportMissingImports for all dependencies that exist only in my virtual environment (e.g. pytest for my tests).

If I activate my python virtual environment prior to running pyright, it is able to resolve the imports correctly.

All repos I am seeing this in are private or I would provide a link to an example.

VS Code extension or command-line

Identified via command line.

[erictraut]

@rchiodo, I'm guessing this is related to your real path changes. Could you investigate?

[rchiodo]

Sure let me try to reproduce the problem. It sounds like it should reproduce with a pip venv too.

[rchiodo]

Yeah, I can reproduce. I think it has something to do with translating paths on the command line.

If you put the 'venvFolder' into the pyrightconfig.json, I believe that will fix the problem temporarily.

At least if I have this in pyrightconfig.json, I can't reproduce:

{
   "venv": ".venv",
   "venvPath": "."
}

But with this:

{
   "venv": ".venv",
}

and passing --venvpath on the command line, then it reproduces.

[rchiodo]

Oh whoops, I shouldn't put 'Fixed' in the titles of the PRs.

[fzakaria]

I am seeing this also on 1.1.327 as well (upgraded from 1.1.325) -- doesn't matter if i specify --venvpath or --pythonpath

[fzakaria]

Seems like it has to do something with symlinks ?
FWIW I also use Pyright with Nix (https://nixos.org), specifically poetry2nix (https://github.com/nix-community/poetry2nix)
which creates a symlink tree of a site-packages directory (see below)

❯ ls -l /nix/store/kkr5h7331csxybk2fvzrqzdrv6gfrhra-python3-3.10.12-env/lib/python3.10/site-packages | head
dr-xr-xr-x   - fmzakari  1 Jan  1970 __pycache__
lrwxrwxrwx 114 fmzakari  1 Jan  1970 _black_version.py -> /nix/store/19635i1dahlgnp6hqav96hzf7yn49cr4-python3.10-black-23.9.1/lib/python3.10/site-packages/_black_version.py
lrwxrwxrwx 117 fmzakari  1 Jan  1970 _distutils_hack -> /nix/store/hxqf8h7yq8d8nchhkac9lqihdqj9h0ml-python3.10-setuptools-67.4.0/lib/python3.10/site-packages/_distutils_hack
lrwxrwxrwx 104 fmzakari  1 Jan  1970 _pytest -> /nix/store/g063887s1sa392l2i15s2lr2fqxw3c4i-python3.10-pytest-7.4.2/lib/python3.10/site-packages/_pytest
lrwxrwxrwx 130 fmzakari  1 Jan  1970 _sysconfigdata__linux_x86_64-linux-gnu.py -> /nix/store/bc45k1n0pkrdkr3xa6w84w1xhkl1kkyp-python3-3.10.12/lib/python3.10/site-packages/_sysconfigdata__linux_x86_64-linux-gnu.py
lrwxrwxrwx 102 fmzakari  1 Jan  1970 apsw -> /nix/store/jvliqw7rxkwdf2hdkzfvp3gq4ksrdckb-python3.10-apsw-3.43.1.0/lib/python3.10/site-packages/apsw
lrwxrwxrwx 121 fmzakari  1 Jan  1970 apsw-3.43.1.0.dist-info -> /nix/store/jvliqw7rxkwdf2hdkzfvp3gq4ksrdckb-python3.10-apsw-3.43.1.0/lib/python3.10/site-packages/apsw-3.43.1.0.dist-info
lrwxrwxrwx 102 fmzakari  1 Jan  1970 black -> /nix/store/19635i1dahlgnp6hqav96hzf7yn49cr4-python3.10-black-23.9.1/lib/python3.10/site-packages/black
lrwxrwxrwx 119 fmzakari  1 Jan  1970 black-23.9.1.dist-info -> /nix/store/19635i1dahlgnp6hqav96hzf7yn49cr4-python3.10-black-23.9.1/lib/python3.10/site-packages/black-23.9.1.dist-info
lrwxrwxrwx 103 fmzakari  1 Jan  1970 blackd -> /nix/store/19635i1dahlgnp6hqav96hzf7yn49cr4-python3.10-black-23.9.1/lib/python3.10/site-packages/blackd

[erictraut]

Yes, this is related to symlinks. There is already a fix in place, and it will be included in next week's pyright release. In the meantime, you can work around the issue by downgrading to 1.1.326. Apologies for the regression.

[fzakaria]

Thank you for the tool and the work.

[erictraut]

This is addressed in pyright 1.1.328, which I just published. It will also be included in a future release of pylance.

[mattiacolombomc]

This is still present today. Could you please help?

Just so you know, the script runs perfectly.

[erictraut]

@mattiacolombomc, this issue has been fixed and closed. If you're experiencing a similar issue, then it's different than the original. If you think that you are seeing a bug, please file a new issue with repro steps as described in the bug template.