"dag" is not a known attribute of module "networkx"

[AbdealiLoKo]

I am using python 3.9 and am seeing pyright throwing an error with networkx in CLI - but NOT in vscode.

I am unable to figure out what the difference is ! Any suggestions ?

Things I have tried:

• Tried explicitly setting useLibraryCodeForTypes: true
• Tried with python 3.8

---

Here is a MRE file I am testing with:

$ cat run.py
from __future__ import annotations

import networkx as nx


graph = nx.DiGraph({})
nx.dag.topological_sort(graph)

And the CLI command:

$ venv/bin/pyright run.py
/home/ali/run.py
  /home/ali/run.py:7:4 - error: "dag" is not a known attribute of module "networkx" (reportAttributeAccessIssue)
1 error, 0 warnings, 0 informations

But in vscode, there is no error (and intellisense and so on work fine)

And intellisense works too:

This is the pyright_config:

$ cat pyrightconfig.json
{
  "pythonVersion": "3.7",
  "exclude": ["**/dist", "**/build", "**/venv*"],
  "venvPath": ".",
  "venv": "venv"
}

Versions:

$ venv/bin/python -V
Python 3.9.19
$ venv/bin/pyright --version
pyright 1.1.382
$ venv/bin/pip list | grep networkx
networkx                  3.2.1

[rchiodo]

My guess is this is because Pylance ships a bunch of bundled stubs and Pyright doesn't. You can read about the differences here:
https://github.com/microsoft/pylance-release/blob/main/USING_WITH_PYRIGHT.md

[rchiodo]

Transferred. Going to make an issue out of this.

[rchiodo]

Created issue #6508

[AbdealiLoKo]

Interesting, I was so sure this was a usage issue on my part :D

I also see a similar thing with pandas then.
When I have the code:

import pandas as pd
val = 1
if pd.api.types.is_scalar(val) and pd.isna(val):
    print("scalar")

I get:

$ venv/bin/pyright run.py
/home/ali/run.py
  /home/ali/run.py:7:11 - error: Cannot access attribute "types" for class "type[Panel]"
    Attribute "types" is unknown (reportAttributeAccessIssue)
  /home/ali/run.py:7:11 - error: Cannot access attribute "types" for class "type[datetime]"
    Attribute "types" is unknown (reportAttributeAccessIssue)
  /home/ali/run.py:7:11 - error: Cannot access attribute "types" for class "type[SparseArray]"
    Attribute "types" is unknown (reportAttributeAccessIssue)
  /home/ali/run.py:7:17 - error: "is_scalar" is not a known attribute of module "numpy.testing" (reportAttributeAccessIssue)
  /home/ali/run.py:7:17 - error: Cannot access attribute "is_scalar" for class "type[NoseTester]"
    Attribute "is_scalar" is unknown (reportAttributeAccessIssue)

[rchiodo]

Are you sure you saved the file? I don't get any errors myself when doing the same thing.

[AbdealiLoKo]

Yes - I am sure I saved the file
I tried creating a MRE for the pandas part and realized that when I use pandas 1.3.5 - I get the error.
But with latest pandas - 2.2.3 I don't get the error

Is there something I can do to make the older pandas work correctly ?

Steps to reproduce:

$ docker run --rm -it redhat/ubi9
# curl -LsSf https://astral.sh/uv/install.sh | sh
# source ~/.bashrc
# uv venv venv --python 3.9 --seed
# cat <<EOF | tee run.py
import pandas as pd

val = 1
if pd.api.types.is_scalar(val) and pd.isna(val):
    print("scalar")

pd.testing.assert_frame_equal(
    pd.DataFrame({}),
    pd.DataFrame({}),
    check_dtype=True,
    check_index_type=False,
    check_column_type=True,
    check_frame_type=False,
    check_names=False,
    check_categorical=True,
)
EOF
# cat <<EOF | tee pyrightconfig.json
{
  "pythonVersion": "3.7",
  "venv": "venv",
  "venvPath": "."
}
EOF
# uv pip install pyright pandas==1.3.5 --python venv
# venv/bin/pyright run.py
/run.py
  /run.py:4:4 - error: Invalid conditional operand of type "Unknown | ndarray[Unknown, Unknown] | ExtensionArraySupportsAnyAll | Any | NDArray[Any] | Series | DataFrame | Literal[False]"
    Method __bool__ for type "Series" returns type "NoReturn" rather than "bool"
    Method __bool__ for type "DataFrame" returns type "NoReturn" rather than "bool" (reportGeneralTypeIssues)
  /run.py:4:11 - error: Cannot access attribute "types" for class "type[datetime]"
    Attribute "types" is unknown (reportAttributeAccessIssue)
  /run.py:4:11 - error: "types" is not a known attribute of module "numpy" (reportAttributeAccessIssue)
  /run.py:4:11 - error: Cannot access attribute "types" for class "type[SparseArray]"
    Attribute "types" is unknown (reportAttributeAccessIssue)
  /run.py:4:36 - error: Invalid conditional operand of type "Unknown | ndarray[Unknown, Unknown] | ExtensionArraySupportsAnyAll | Any | NDArray[Any] | Series | DataFrame | Literal[False]"
    Method __bool__ for type "Series" returns type "NoReturn" rather than "bool"
    Method __bool__ for type "DataFrame" returns type "NoReturn" rather than "bool" (reportGeneralTypeIssues)
  /run.py:7:12 - error: Cannot access attribute "assert_frame_equal" for class "type[datetime]"
    Attribute "assert_frame_equal" is unknown (reportAttributeAccessIssue)
  /run.py:7:12 - error: "assert_frame_equal" is not a known attribute of module "numpy" (reportAttributeAccessIssue)
  /run.py:7:12 - error: Cannot access attribute "assert_frame_equal" for class "type[SparseArray]"
    Attribute "assert_frame_equal" is unknown (reportAttributeAccessIssue)
8 errors, 0 warnings, 0 informations