Installing Python Modules & Using `pip`¶

Python features an enormous ecosystem of open-source third-party libraries available on the Python Package Index (PyPI). The standard tool for installing and managing these libraries is pip.

The Standard Installation Command¶

Always invoke pip using the python -m pip syntax to ensure packages are installed into the active Python environment:

python -m pip install requests

To install a specific version or range:

# Exact version
python -m pip install requests==2.31.0

# Minimum version
python -m pip install "fastapi>=0.100.0"

# Compatible release (matches >= 1.4.0, < 2.0.0)
python -m pip install "pydantic~=1.4"

Managing Project Dependencies with `requirements.txt`¶

For reproducible deployments and team development, record all project dependencies in a requirements.txt file:

# requirements.txt
fastapi>=0.110.0
uvicorn[standard]>=0.28.0
sqlalchemy>=2.0.0
pydantic>=2.6.0
python-dotenv>=1.0.0

Install all listed dependencies in one command:

python -m pip install -r requirements.txt

Freezing Exact Pinned Versions¶

To create an exact snapshot of every installed package (including transitive sub-dependencies):

python -m pip freeze > requirements.lock

Wheels (`.whl`) vs Source Distributions (`sdist`)¶

When pip downloads a package, it looks for two formats:

               PyPI Package Artifacts
                      |
        +-------------+-------------+
        |                           |
        v                           v
   Wheel (.whl)            Source Dist (.tar.gz)
   - Pre-built binary      - Raw Python & C source code
   - No compiler required  - Requires gcc/clang/MSVC
   - Installs in ms        - Takes minutes to compile

Wheel (.whl): A pre-compiled binary package. It requires no C compilers or external header files on your machine. Packages like numpy, cryptography, and pandas install instantly.
Source Distribution (.tar.gz / sdist): Contains raw source code. If the package contains C or Rust extensions, your computer must have the appropriate compiler tools installed.

Editable Installs for Local Development (`-e`)¶

When developing your own Python package or library, you want your code changes to take effect immediately without having to run pip install after every edit.

Use editable mode (-e):

# Inside the root of your package repo (where pyproject.toml lives)
python -m pip install -e .

This adds a direct link (via a .pth file) into site-packages pointing to your local repository directory.

Installing from Git Repositories¶

You can install packages directly from GitHub or GitLab branches or commit hashes:

python -m pip install git+https://github.com/psf/requests.git@main

Upgrading and Uninstalling¶

# Upgrade an existing package to the latest release
python -m pip install --upgrade requests

# Uninstall a package
python -m pip uninstall requests

# Inspect package metadata and install location
python -m pip show requests

Installing Python Modules & Using pip¶