This code analyzes the behavior of Chrome extensions. It makes use of Selenium Webdriver to get a list of extensions from the Chrome Web Store, download their source code, install them in a browser and then open a series of websites.
It uses a modified version of a Python proxy to log all connections that the browser makes when opening the websites, and stores this in a report. If an extension sends data to a remote host every time a URL is openend (or batch-wise after a few opens), you will see it in the logs.
First, clone the repository. Then, install the dependencies (Chromium browser, Chromedriver, Selenium and Pip).
On Ubuntu, you can use apt-get for this:
sudo apt-get install chromium-browser chromium-chromedriver python3-pip
Create a virtual environment (optional):
virtualenv --python python3 venv
source venv/bin/activate
Then, install the Python requirements:
pip install -r requirements.txt
You should now be all set. Depending on the installation location of your chromedriver binary, you might need to tweak the executable path given to it in the scripts.
There are three scripts to get a list of the extensions, download their source code and test them.
To get a list of extensions from the Google Web Store, run get_extension_list.py:
python get_extension_list.py data/extensions.json
This will scroll down the page until you hit a button on the keyboard, then it will parse all extensions that it has seen and store them in a JSON object.
After getting the extension list, you can download the extensions:
python download_extensions data/extensions.json data/extensions
Finally, to analyze the downloaded extensions:
python test_extensions data/extensions.json data/extensions data/reports
This will test all extensions in data/extension.json for which there exists a downloaded version in data/extensions, and store the result in data/reports. If an extension already has been analyzed, it will skip it.
The script contains an additional parameter baseline, which is set to False by default. If you set it to True, in addition to checking the browser behavior with the given extension installed, the script will check behavior without it as well, in order to provide a baseline for the connections made by opening the list of websites you provided. This is useful to distinguish "normal" tracking from extension-based tracking.
The data directory contains example data for extension source code, lists and reports.
Feel free to contact me.
This code is public domain, use it in whichever way you like (please note the different license for the Proxy server).