CookieBlock is a browser extension that automatically enforces your GDPR consent preferences for browser cookies. It classifies cookies on-the-fly into four distinct categories, and deletes those that the user did not consent to.
This helps enforce user privacy without having to rely on the website hosting the cookies.
CookieBlock is an extension that allows the user to apply their cookie consent preferences to any website, no matter if the website has a cookie banner. The user specifies their consent options once when the extension is first installed, and then CookieBlock will try to remove any cookies that do not align with the user's policy as they are being created.
This is intended to ensure that the privacy of the user is preserved. One can reject any of the following categories:
Note that CookieBlock does not handle the cookie banner itself. In order to remove these annoying banners, we recommend using the Consent-O-Matic extension:
CookieBlock is compatible with both Firefox and Chromium-based browsers, and it is available on the following addon stores:
If you would like to submit feedback, or report websites that break because of the addon, you can open an issue on this Github page, or alternatively use this Google Forms document.
No requirements outside of what is contained in this repository is needed to build CookieBlock. Simply pack the contents of the subfolder src into a zip file, and you can install it into your browser. More information here.
Alternatively, you can also install npm and use the web-ext command-line tool, with the command web-ext build.
The model files are constructed in the following process:
xgb_train.py) found in this repository.xgboost_small_dump.py) to transform the XGBoost model into a minified JSON tree structure. This script produces the four model files forest_class0.json to forest_class3.json../src/ext_data/model/And replace the existing forest class files. Make sure to preserve their names as is.
The policy enforcement process is a background script that executes every time a cookie event is raised in the browser. If this event indicates that a cookie was added or updated, the extension will store the cookie in a local history of cookie updates, and then perform a classification for that cookie.
The category for each cookie is predicted using a forest of decision trees model trained via the XGBoost classifier, and a set of feature extraction steps. First, the cookie is turned into a numerical vector, which is then provided as an input to the forest of trees. This produces a score for each class, and the best score is the class that gets assigned to the cookie.
Available cookie categories are:
Granularity is intentionally kept low to make the decision as simple as possible for the user. Note that "strictly necessary" cookies cannot be rejected, as this is the class of cookies that is required to make the website work. Without them, essential services such as logins would stop working.
The feature extractor can be found in the subfolder nodejs-feature-extractor/. This is used to extract the features for the training data set.
For the classifier training, see:
https://github.com/dibollinger/CookieBlock-Consent-Classifier
The classifier is not completely accurate. It may occur that certain functions on some sites are broken because essential cookies get misclassified. This is hard to resolve without gathering more cookie data to train on. As such, the approach has its limits.
To resolve these problems, we maintain a list of known cookie categories. This is a JSON file storing cookie labels for known cookie identifiers. If a cookie is contained in this file. the prediction is skipped, and the known class is applied.
By reporting broken websites, you can help us keep an updated list of cookie exceptions. This makes the extension more useable for everyone while also keeping a high level of privacy.
node-feature-extractor/: Contains the feature extractor implemented using npm. Used to extract features with the same JavaScript code as the extension.
/modules/: Contains code used to perform the feature extraction and prediction./outputs/: Output directory for the feature extraction./training_data/: Path for cookie data in json format, used for extracting features./validation_data/: Path for cookie features, extracted in libsvm format, used for prediction and verifying model accuracy./cli.js: Command-line script used to run the feature extraction.logo/: Contains the original CookieBlock logo files.promotional/: Contains store page text, localizations and promo images used for those stores.src/: Source code for the CookieBlock extension.
_locales/: Contains all human-readable strings displayed on the extension interface, with translations to different languages.background/: JavaScript code and HTML for the extension background process. Currently only Manifest v2 compatible.ext_data/: All external data required to perform the feature extraction and class label prediction.
model/: Extracted CART prediction tree forests, one for each class of cookies.resources/: Resources used with the feature extraction.default_config.json: Defines default storage values used in the extension.features.json: Defines how the feature extraction operates, and which individual feature are enabled.known_cookies.json: Defines default categorizations for some known cookies. Used to grant exceptions to potential website breakages.icons/: Browser extension icons.modules/: Contains scripts that handle certain aspects of the feature extraction and prediction.
third_party/: Third party code libraries.options/: Contains the options and first time setup page of the extension.popup/: Contains code for the extension popup.credits.txt: Links to the third-party libraries and credits to the respective authors.LICENSE: License of the extension.CookieBlock logo designed by Charmaine Coates.
Czech translation provided by Karel Kubicek.
Japanese translation provided by Shitennouji.
Spanish translation provided by @6ig6oy.
Automated localization in store pages performed using DeepL.
CookieBlock includes code from the following libraries and projects:
This repository was created as part of the master thesis "Analyzing Cookies Compliance with the GDPR", which can be found at:
https://www.research-collection.ethz.ch/handle/20.500.11850/477333
as well as the paper "Automating Cookie Consent and GDPR Violation Detection", which can be found at:
https://karelkubicek.github.io/post/cookieblock.html
Thesis Supervision and co-authors:
See also the following repositories for other components that were developed as part of the thesis:
Copyright © 2021-2022 Dino Bollinger, Department of Computer Science at ETH Zürich, Information Security Group
MIT License, see included LICENSE file