公司电脑无root权限下Python依赖安装及pip3配置重置问题

First, let's break down the root issue: your pip configuration was pre-set to use your company's private JFrog PyPI repository, which triggers an authentication prompt that's failing your dependency installs and virtualenv setup. Your error log makes this clear:

Looking in indexes: https://pypi.org/simple, https://firstname.lastname:jFrog12345@companydev.jfrog.io/companydev/api/pypi/pypi/simple

Here's a straightforward, step-by-step solution to reset pip and get your ML dependencies installed locally without touching system directories:

1. Check Current Pip Configuration #

First, confirm the problematic repo settings exist:

pip3 config list

You’ll likely see entries like global.index-url or global.extra-index-url pointing to the company JFrog address.

2. Reset Pip to Default Settings #

We’ll clear the user-level pip config (since you don’t have root access to modify system-wide settings):

• Delete the user pip config file directly:

  rm ~/.config/pip/pip.conf
  

  Note: If this file doesn’t exist, check ~/.pip/pip.conf instead—location can vary slightly.
• Alternatively, unset the problematic config entries one by one:

  pip3 config unset global.index-url
  pip3 config unset global.extra-index-url
  

3. Verify the Reset #

Run pip3 config list again. You should see no repo-related entries, meaning pip will now default to the official PyPI.

4. Install Dependencies Locally with --user #

Now you can install numpy, gym, and other packages directly to your user-specific site-packages directory—no system permissions needed:

pip3 install --user numpy gym

These will land in ~/.local/lib/python3.6/site-packages (matching your Python 3.6 setup), which is entirely safe and isolated from system files.

5. Fix Virtualenv Setup (If You Still Want It) #

If you prefer using a virtual environment to isolate your ML project, follow these steps:

1. First upgrade your local virtualenv to avoid the built-in download that triggers the JFrog prompt:

   pip3 install --user --upgrade virtualenv
   

2. Create the virtual environment with the --no-download flag (this uses your local copies of setuptools/pip/wheel instead of trying to fetch them from the company repo):

   python3 -m virtualenv --no-download env
   

3. Activate the environment:

   source env/bin/activate
   

4. Install dependencies inside the virtual environment—pip here will use the default PyPI, no authentication required:

   pip install numpy gym
   

This should resolve all authentication errors and let you work on your ML code without system directory modifications or dealing with the company repo.

内容的提问来源于stack exchange，提问作者Leviathan