/vulnerablecode

A work-in-progress towards a free and open vulnerabilities database and the packages they impact. And the tools to aggregate and correlate these vulnerabilities. Sponsored by NLnet https://nlnet.nl/project/vulnerabilitydatabase/ for https://www.aboutcode.org/ Chat at https://gitter.im/aboutcode-org/vulnerablecode

Primary LanguagePythonApache License 2.0Apache-2.0

VulnerableCode

Build Status License Python 3.8 stability-wip Gitter chat

VulnerableCode is a free and open database of FOSS software package vulnerabilities and the tools to create and keep the data current.

It is made by the FOSS community to improve and secure the open source software ecosystem.

README.gif

Why?

The existing solutions are commercial proprietary vulnerability databases, which in itself does not make sense because the data is about FOSS (Free and Open Source Software).

The National Vulnerability Database which is a primary centralized data source for known vulnerabilities is not particularly well suited to address FOSS security issues because:

  1. It predates the explosion of FOSS software usage
  2. It's data format reflects a commercial vendor-centric point of view in part due to the usage of CPE to map vulnerabilities to existing packages.
  3. CPEs are just not designed to map FOSS to vulnerabilities owing to their vendor-product centric semantics. This makes it really hard to answer the fundamental questions "Is package foo vulnerable" and "Is package foo vulnerable to vulnerability bar?"

How

VulnerableCode independently aggregates many software vulnerability data sources and supports data re-creation in a decentralized fashion. These data sources (see complete list here) include security advisories published by Linux and BSD distributions, application software package managers and package repositories, FOSS projects, GitHub and more. Thanks to this approach, the data is focused on specific ecosystems yet aggregated in a single database that enables querying a richer graph of relations between multiple incarnations of a package. Being specific increases the accuracy and validity of the data as the same version of an upstream package across different ecosystems may or may not be vulnerable to the same vulnerability.

The packages are identified using Package URL PURL as primary identifiers rather than CPEs. This makes answers to questions such as "Is package foo vulnerable to vulnerability bar?" much more accurate and easy to interpret.

The primary access to the data is through a REST API.

In addition, an emerging web interface goal is to support vulnerabilities data browsing and search and progressively to enable community curation of the data with the addition of new packages and vulnerabilities, and reviewing and updating their relationships.

We also plan to mine for vulnerabilities which didn't receive any exposure due to various reasons like but not limited to the complicated procedure to receive CVE ID or not able to classify a bug as a security compromise.

Recent presentations:

Setting up VulnerableCode

First clone the source code:

git clone https://github.com/nexB/vulnerablecode.git
cd vulnerablecode

Using Docker Compose

Please find the docker documentation in Docker Installation

TL;DR

git clone https://github.com/nexB/vulnerablecode.git && cd vulnerablecode
make envfile
docker-compose up

Go to http://localhost:8000/ on a web browser to access the web UI.

Without Docker Compose

System requirements

  • Python 3.8+
  • PostgreSQL 9+
  • Compiler toolchain and development files for Python and PostgreSQL

On Debian-based distros, these can be installed with:

sudo apt-get install python3-venv python3-dev postgresql libpq-dev build-essential

Database configuration

  • Create a user named vulnerablecode. Use vulnerablecode as password when prompted:

    sudo -u postgres createuser --no-createrole --no-superuser --login \
        --inherit --createdb --pwprompt vulnerablecode``
    
  • Create a databased named vulnerablecode:

    createdb --encoding=utf-8 --owner=vulnerablecode  --user=vulnerablecode \
        --password --host=localhost --port=5432 vulnerablecode
    

Application dependencies

Create a virtualenv, install dependencies, generate static files and run the database migrations:

make envfile
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
python manage.py collectstatic
python manage.py migrate

For a production mode, an environment variable named SECRET_KEY needs to be set. The recommended way to generate this key is to use the code Django includes for this purpose:

SECRET_KEY=$(python -c "from django.core.management import utils; print(utils.get_random_secret_key())")

You will also need to setup the ALLOWED_HOSTS array inside vulnerablecode/settings.py according to [django specifications](https://docs.djangoproject.com/en/3.2/ref/settings/#allowed-hosts). One example would be: .. code-block:: python

ALLOWED_HOSTS = ['vulnerablecode.your.domain.example.com']

You can specify several hosts by separating them with a comma (,)

Using Nix

You can install VulnerableCode with Nix (Flake support is needed):

cd etc/nix
nix-shell -p nixFlakes --run "nix --print-build-logs flake check " # build & run tests

There are several options to use the Nix version:

# Enter an interactive environment with all dependencies set up.
cd etc/nix
nix develop
> ../../manage.py ... # invoke the local checkout
> vulnerablecode-manage.py ... # invoke manage.py as installed in the nix store

# Test the import prodecure using the Nix version.
etc/nix/test-import-using-nix.sh --all # import everything
# Test the import using the local checkout.
INSTALL_DIR=. etc/nix/test-import-using-nix.sh ruby # import ruby only

Keeping the Nix setup in sync

The Nix installation uses mach-nix to handle Python dependencies because some dependencies are currently not available as Nix packages. All Python dependencies are automatically fetched from ./requirements.txt. If the mach-nix-based installation fails, you might need to update mach-nix itself and the pypi-deps-db version in use (see etc/nix/flake.nix:inputs.machnix and machnixFor.pypiDataRev).

Non-Python dependencies are curated in:

etc/nix/flake.nix:vulnerablecode.propagatedBuildInputs

Run Tests

Make sure to install dev dependencies by running pip install -r requirements-dev.txt Use these commands to run code style checks and the test suite:

black -l 100 --check .
python -m pytest

Data import

Some data importers use the GitHub APIs. For this, export the GH_TOKEN environment variable with:

export GH_TOKEN=yourgithubtoken

If you are running behind a proxy, you will need to setup the standard https_proxy variable.

export https_proxy=https?://<proxy>:<port>

See GitHub docs for instructions on how to obtain your GitHub token.

To run all data importers use:

python manage.py import --all

To list available importers use:

python manage.py import --list

To run specific importers:

python manage.py import rust npm

REST API access

Start the webserver:

python manage.py runserver

For full documentation about API endpoints use this URL:

http://127.0.0.1:8000/api/docs

Continuous periodic Data import

If you want to run the import periodically, you can use a systemd timer:

$ cat ~/.config/systemd/user/vulnerablecode.service

[Unit]
Description=Update vulnerability database

[Service]
Type=oneshot
ExecStart=/path/to/venv/bin/python /path/to/vulnerablecode/manage.py import --all

$ cat ~/.config/systemd/user/vulnerablecode.timer

[Unit]
Description=Periodically update vulnerability database

[Timer]
OnCalendar=daily

[Install]
WantedBy=multi-user.target

Start this "timer" with:

systemctl --user daemon-reload
systemctl --user start vulnerablecode.timer