如何在 Linux 服务器中设置新用户以使用 Python？

Question

I have added a new user in a linux server.

While I can run simple python script, the new user get errors to import package that I can upload.

For instance

import matplotlib as plt

returns

no module named 'matplotlib'

Answer 1

That is because you probably did :

pip install --user matplotlib

to install matplotlib on your side.

You have multiple solutions, here a non-exhaustive one (docker would be probably overkilled for that purpose imo) ordered from the worse to the best :

• Install packages for everyone and avoid using --user when using pip

• Create a requirements.txt with all the needed package names and all your users can do pip install -r requirements.txt --user

• Create a requirements.txt and use virtual environments ( https://docs.python.org/3/library/venv.html or https://docs.conda.io/projects/conda/en/latest/user-guide/tasks/manage-environments.html )

Answer 2

This can be for multiple reasons.

matplotlib installed as user

If you installed matplotlib with pip install --user package in a different user account. Don't expect another user to have that package available, it won't happen.

Install the package with pip install package to be available system wide, for all users.

Note. Depending on your specific setup you may need elevated permissions.

matplotlib is installed in different python version

Irrespectively of how you installed the package (see above), you might have installed it in a different version than the one you're trying to run.

To be sure you can import the package, as in python -m package , use python -m pip install package . This will ensure the same python interpreter is used to install as is used to execute.

Better approaches

one more standard approach

Learn and use virtual environments. They are made to handle a python interpreter installation per project, rather than per system. Allowing you to keep dependent packages isolated from your system's python packages.

one slightly more advance approach

I'll use the opportunity to share a little Makefile that handles that crap for me. Manages a virtual environment in the development folder, and keeps it up to date to with whatever packages are set in requirements.txt .

Invoking make test is sufficient to run the test suite (or any other thing you may configure) in an environment with the correct packages.

.PHONY: test

# directory to store virtual environment
VENV_NAME=venv

# python runtime version
PYTHON_VER=3.7

# python executble
PYTHON=${VENV_NAME}/bin/python${PYTHON_VER}

# pip requirements file
REQUIREMENTS=requirements.txt

venv:           ## Recreates the virtual environment if needed.
venv: $(VENV_NAME)/bin/activate
$(VENV_NAME)/bin/activate: ${REQUIREMENTS}
    test -d $(VENV_NAME) || virtualenv -p python${PYTHON_VER} $(VENV_NAME)
    ${PYTHON} -m pip install -U pip
    ${PYTHON} -m pip install -r ${REQUIREMENTS}
    touch $@

test:           ## Runs the test suite.
test: venv
    $(PYTHON) -m pytest tests