Empower your script with auto_venv: Say Goodbye to Manual Setup or Install!

With auto_venv, effortlessly automate virtual environment creation and pip-installs for your script's requirements. No more tedious manual setups, no more hassle with requirement installations, and certainly, no extra steps for executing your script. No need to install simple scripts, just copy your python files and run.

auto_venv sets up a virtual environment in a temporary location upon your script's first run, seamlessly installing all specified requirements via pip.

Usage

Use a recent Python 3 (3.6+) version:

#!/bin/env python3

Download and bundle a copy of auto_venv.py with your main script. Import auto_venv and call its init function passing these arguments:

1. filename (str)            : calling script's filename including full path.
2. requires (list[str])      : list of requirements to install
3. fancy    (bool, optional) : fancy/ANSI printout.  (Defaults to False)
4. quiet    (bool, optional) : quiet (no) messages.  (Defaults to False)
5. dot_pth  (bool, optional) : create .pth link to the .venv  (Defaults to False)

from pathlib import Path

import auto_venv

__filename__ = str(Path(__file__).resolve())
__requires__ = [
    "beautifulsoup4==4.12.3",
    "requests==2.31.0",
    "pyyaml==6.0.1",
]

auto_venv.init(__filename__, __requires__, True, False, True)

Import required packages, and continue with the rest of your script.

import bs4       # noqa: E402
import requests  # noqa: E402
import yaml      # noqa: E402

print(f"bs4 version      : {bs4.__version__}")
print(f"requests version : {requests.__version__}")
print(f"yaml version     : {yaml.__version__}\n")

soup = bs4.BeautifulSoup(requests.get("https://pypi.org").content, "html.parser")
print(yaml.dump({"header": soup.find("h1").get_text()}))

📝 Note: because not all our imports are at the top of the script, we need to add # noqa: E402 to the end of each import to ignore Lint error: Module level import not at top of the file.

Virtual Environment

By default virtual environments are created under `/tmp``, which is NOT persistent and will be removed upon machine reboot. You can change the default location by setting VENV_AUTO_ROOT environment variable.

To avoid collision with other users, scripts, ... the name of the virtual environment is: .venv_<UUID>_<USER>_<FILENAME>

UUID        unique id generated based on the full script path (filename) passed to init
USER        is the username
FILENAME    is the basename of the script without full path

VSCode and Intellisense

To be able to debug/navigate the installed packages, auto_venv will create a .pth file under /home/${USER}/.local/lib/python<VERSION>/site-packages/ with name venv_<UUID>_<USER>_<FILENAME>.pth that points to the created virtual environment. This will add the path to the virtual environment to the PYTHON_PATH. This way, VSCode and other IDEs would be able to find the packages installed in this virtual enviroment. If multiple dependencies with different versions are installed in different virtual envirnment, which version will be picked up is not very clear. For this reason, only enable this option, dot_pth or AUTO_VENV_DOT_PTH, when debugging using IDE.

Environment variables

auto_venv uses the following environment variables to override the fancy/quiet arguments.

VENV_AUTO_ROOT       path to location for creating the virtual environment, default is /tmp

AUTO_VENV_FANCY      fancy is set to True if set to '1', 'y', 'yes' or 'true', otherwise False
AUTO_VENV_QUIET      quiet is set to True if set to '1', 'y', 'yes' or 'true', otherwise False
AUTO_VENV_DOT_PTH    dot_pth is set to True if set to '1', 'y', 'yes' or 'true', otherwise False

Example output

First run (fancy=True, quiet=False, dot_pth=True)

$ example.py
╭────────────────────────────────────────────────────────────────────────────────
│ ➜ First run or some requirements were not found.  Reinstalling ...
│ ➜ Creating virtual environment ...
│     '/tmp/.venv_da65983c-2c58-5629-94e6-e157b6565a0d_${USER}_example.py'
│ ➜ Setting up PYTHONPATH ...
│     '/home/${USER}/.local/lib/python3.12/site-packages/venv_dafede3c-2c58-5629-94e6-e157b6565a0d_${USER}_example.py.pth'
│ ➜ Upgrading 'pip' ...
│ ➜ Installing requirements:
│   ✓ beautifulsoup4==4.12.3
│   ✓ requests==2.31.0
│   ✓ pyyaml==6.0.1
│ ➜ Continuing with the script ...
╰────────────────────────────────────────────────────────────────────────────────
bs4 version      : 4.12.3
requests version : 2.31.0
yaml version     : 6.0.1

header: Find, install and publish Python packages with the Python Package Index

Later runs (fancy=True, quiet=False, dot_pth=True):

$ example.py
╭────────────────────────────────────────────────────────────────────────────────
│ ➜ Virtual environment already exists ...
│     '/tmp/.venv_dafede3c-2c58-5629-94e6-e157b6565a0d_${USER}_example.py'
│ ➜ All requirements are met.
│ ➜ Continuing with the script ...
╰────────────────────────────────────────────────────────────────────────────────
bs4 version      : 4.12.3
requests version : 2.31.0
yaml version     : 6.0.1

header: Find, install and publish Python packages with the Python Package Index

📝 Notes

Once the requirements are installed, if you update the requirement versions, there is no automatic way to check if the versions have changed. You just need to delete the virtual environment (and .pth file) and rerun the script with the new requirements.

References

• Inspied by pip.wtf
• pip - Installing Python Modules
• venv - Creation of virtual environments
• Inline Script Metadata
• PEP 723 – Inline script metadata
• PEP 704 – Require virtual environments by default for package installers
• REJECTED: PEP 582 - Python local packages directory
• REJECTED: PEP 722 – Dependency specification for single-file scripts