pip.md

Installing Packages with pip

pip is a tool to install Python packages and resolve dependencies automatically. This section lists a couple of things you can do with pip:

Install a package

To install a Python package, call pip with the package name:

:::bash
pip install pandas

You can specify the exact version of a package:

:::bash
pip install pandas==0.25.0

---

Install many packages

First, create a file requirements.txt in your project directory. The file should look similar to this:

:::text
pandas==0.25
numpy>=1.17
scikit-learn
requests

Second, ask pip to install everything:

:::bash
pip -r requirements.txt

---

Install from a git repo

If a repository has a setup.py file, you could install directly from git. This is useful to install branches, forks and other work in progress:

:::bash
pip install git+https://github.com/pandas-dev/pandas.git

---

Install a package you are developing

When developing, you might want to pip-install a working copy. This allows you to import your package (e.g. for testing). Changes to the code directly take effect in the installation.

For the following to work, your project folder needs to have a setup.py:

:::bash
pip install --editable .

---

List all installed packages

This one prints all packages you have installed and their versions:

:::bash
pip freeze

To search for a pacakge, use grep:

:::bash
pip freeze | grep pandas

---

Uninstall a package

pip also removes packages:

:::bash
pip uninstall pandas

---

Where does pip store its files?

Usually, packages are stored in the site_packages/ folder. Where this one is depends on your distribution and your virtual environment.

You might want to check your PYTHONPATH environment variable. To do so from Python, use:

:::python3
import sys
print(sys.path)

---

Also see

• The conda program (part of the Anaconda distribution) is often a viable alternative to pip
• You find all installable packages on the Python Package Index pypi.org