JuliaPy / JuliaPy/PythonCall.jl

Unable to import Python module

未关闭
#757 1 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看
question
主要语言
Julia
星标
1.1k
派生
86
平均合并
1 天 22 小时
30 天内合并 PR
3

描述

To be honest, I'm a little perplexed at how this is failing.

For a background, I need to import a python file that I wrote and is stored in the local directory (LHIPA.py). That file, in turn, needs to import PyWavelets, which has been installed in the .Condapkg environment. I have confirmed in my Julia program as well as in the python file that it is using the python environment created by PythonCall.jl. I've also confirmed, in the Python program, that PyWavelets is an installed package, yet I get this error message: `ERROR: Python: ModuleNotFoundError: No module named 'pywavelets'`

First the Julia program as well as its output:
```
...

ENV["JULIA_CONDAPKG_BACKEND"] = "Null"
ENV["JULIA_PYTHONCALL_EXE"] = raw"/Users/MattPetersonsAccount/.julia/environments/v1.11/.CondaPkg/.pixi/envs/default/lib/python3.11'"

using PythonCall
println("~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~")
println("\n PythonCall is using", PythonCall.python_executable_path())
println("~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~")

pth = (@__DIR__) # Path where the python routines are
pyimport("sys").path.append(pth) # append path

println("sys.path = ", pyimport("sys").path)
println("~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~")
println("ENV['JULIA_CONDAPKG_BACKEND'] = ", ENV["JULIA_CONDAPKG_BACKEND"] )
println("~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~")
println("ENV['JULIA_PYTHONCALL_EXE'] = ", ENV["JULIA_PYTHONCALL_EXE"] )

LH = pyimport("LHIPA")
```

And the output looks as expected:

```
PythonCall is using/Users/MattPetersonsAccount/.julia/environments/v1.11/.CondaPkg/.pixi/envs/default/bin/python
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
sys.path = ['/Users/MattPetersonsAccount/.julia/packages/PythonCall/83z4q/pysrc',
'/Users/MattPetersonsAccount/.julia/environments/v1.11/.CondaPkg/.pixi/envs/default/lib/python311.zip',
'/Users/MattPetersonsAccount/.julia/environments/v1.11/.CondaPkg/.pixi/envs/default/lib/python3.11',
'/Users/MattPetersonsAccount/.julia/environments/v1.11/.CondaPkg/.pixi/envs/default/lib/python3.11/lib-dynload',
'/Users/MattPetersonsAccount/.julia/environments/v1.11/.CondaPkg/.pixi/envs/default/lib/python3.11/site-packages',
'/Users/MattPetersonsAccount/Documents/Development/PsychoPy/Spatial n-back 2']
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
ENV['JULIA_CONDAPKG_BACKEND'] = Null
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
ENV['JULIA_PYTHONCALL_EXE'] = /Users/MattPetersonsAccount/.julia/environments/v1.11/.CondaPkg/.pixi/envs/default/lib/python3.11'
```

The Python file, LHIPA.py, looks like this:

```
from platform import python_version
print("LHIPA python version is: ",python_version())

import importlib_metadata
dists = importlib_metadata.distributions()
for dist in dists:
name = dist.metadata["Name"]
version = dist.version
print(f'found distribution {name}=={version}')
import sys
print("\n",sys.prefix,"\n")

import math
import PyWavelets
```

And again, the output is as expected:

```
LHIPA python version is: 3.11.0
found distribution setuptools==82.0.1
found distribution wheel==0.46.3
found distribution importlib_metadata==8.7.1
found distribution pip==26.0.1
found distribution packaging==26.0
found distribution numpy==2.4.4
found distribution PyWavelets==1.9.0
found distribution zipp==3.23.0

/Users/MattPetersonsAccount/.julia/environments/v1.11/.CondaPkg/.pixi/envs/default
```

So, I've confirmed that Julia and Python are using the same environment (unless one of them is reporting incorrectly, or there is an error in front of my eyes I am missing) and I've confirmed that PyWavelets in the Python environment. Yet, I get an import error.

Any ideas? Is there something blatantly obvious that I'm missing?

贡献指南

这个仓库没有索引到贡献指南

调研方向

Start with LHIPA.py and the Julia program shown in the issue, then reproduce the ModuleNotFoundError in the displayed PythonCall environment. Compare the reported sys.path and installed distributions with the import attempt; done means identifying whether PythonCall or the module/package import is responsible and documenting a verified resolution.

由索引模型根据 Issue 内容生成。

评估

技术栈
python
领域
tooling
Issue 类型
缺陷
难度
3/5
预计耗时
1-2 天
活跃度
冷清
描述清晰度
需要澄清
新手友好度
35/100

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。