OpenWPM

OpenWPM

OpenWPM is a web privacy measurement framework which makes it easy to collect data for privacy studies on a scale of thousands to millions of websites. OpenWPM is built on top of Firefox, with automation provided by Selenium. It includes several hooks for data collection. Check out the instrumentation section below for more details.

Table of Contents

• Installation
  • Pre-requisites
  • Install
  • Mac OSX
• Quick Start
• Troubleshooting
• Documentation
• Advice for Measurement Researchers
• Developer instructions
• Instrumentation and Configuration
• Storage
  • Local Storage
  • Remote storage
• Docker Deployment for OpenWPM
  • Building the Docker Container
  • Running Measurements from inside the Container
  • MacOS GUI applications in Docker
• Citation
• License

Installation

OpenWPM is tested on Ubuntu 24.04 via GitHub Actions and is commonly used via the docker container that this repo builds, which is based on Ubuntu 22.04. Although we don't officially support other platforms, conda is a cross platform utility and the install script can be expected to work on OSX and other linux distributions.

OpenWPM does not support windows: https://github.com/openwpm/OpenWPM/issues/503

Pre-requisites

The main pre-requisite for OpenWPM is conda, a fast cross-platform package management tool.

Conda is open-source and can be installed from https://docs.conda.io/projects/conda/en/latest/user-guide/install/index.html.

Install

An installation script, install.sh is included to: install the conda environment, install unbranded firefox, and build the instrumentation extension.

All installation is confined to your conda environment and should not affect your machine. The installation script will, however, override any existing conda environment named openwpm.

To run the install script, run

./install.sh

After running the install script, activate your conda environment by running:

conda activate openwpm

Mac OSX

You may need to install make / gcc in order to build the extension. The necessary packages are part of xcode: xcode-select --install

We do not run CI tests for Mac, so new issues may arise. We welcome PRs to fix these issues and add full CI testing for Mac.

Running Firefox with xvfb on OSX is untested and will require the user to install an X11 server. We suggest XQuartz. This setup has not been tested, we welcome feedback as to whether this is working.

Quick Start

Once installed, it is very easy to run a quick test of OpenWPM. Check out demo.py for an example. This will use the default setting specified in openwpm/config.py::ManagerParams and openwpm/config.py::BrowserParams, with the exception of the changes specified in demo.py.

The demo script also includes a sample of how to use the Tranco top sites list via the optional command line flag demo.py --tranco. Note that since this is a real top sites list it will include NSFW websites, some of which will be highly ranked.

More information on the instrumentation and configuration parameters is given below.

The docs provide a more in-depth tutorial, and a description of the methods of data collection available.

Troubleshooting

1. WebDriverException: Message: The browser appears to have exited before we could connect...

   This error indicates that Firefox exited during startup (or was prevented from starting). There are many possible causes of this error:

   • If you are seeing this error for all browser spawn attempts check that:

     • Both selenium and Firefox are the appropriate versions. Run the following commands and check that the versions output match the required versions in install.sh and environment.yaml. If not, re-run the install script.
     bash复制

     cd firefox-bin/
     firefox --version

     and

     bash复制

       conda list selenium

     • If you are running in a headless environment (e.g. a remote server), ensure that all browsers have display_mode set to "headless" before launching.

   • If you are seeing this error randomly during crawls it can be caused by an overtaxed system, either memory or CPU usage. Try lowering the number of concurrent browsers.

2. In older versions of firefox (pre 74) the setting to enable extensions was called extensions.legacy.enabled. If you need to work with earlier firefox, update the setting name extensions.experiments.enabled in openwpm/deploy_browsers/configure_firefox.py.

3. Make sure your conda environment is activated (conda activate openwpm). You can see you environments and the activate one by running conda env list the active environment will have a * by it.

4. make / gcc may need to be installed in order to build the web extension. On Ubuntu, this is achieved with apt-get install make. On OSX the necessary packages are part of xcode: xcode-select --install.

5. On a very sparse operating system additional dependencies may need to be installed. See the Dockerfile for more inspiration, or open an issue if you are still having problems.

6. If you see errors related to incompatible or non-existing python packages, try re-running the file with the environment variable PYTHONNOUSERSITE set. E.g., PYTHONNOUSERSITE=True python demo.py. If that fixes your issues, you are experiencing issue 689, which can be fixed by clearing your python user site packages directory, by prepending PYTHONNOUSERSITE=True to a specific command, or by setting the environment variable for the session (e.g., export PYTHONNOUSERSITE=True in bash). Please also add a comment to that issue to let us know you ran into this problem.

Documentation

Further information is available at OPENWPM's Documentation Page.

Advice for Measurement Researchers

OpenWPM is often used for web measurement research. We recommend the following for researchers using the tool:

Use a versioned release. We aim to follow Firefox's release cadence, which is roughly once every four weeks. If we happen to fall behind on checking in new releases, please file an issue. Versions more than a few months out of date will use unsupported versions of Firefox, which are likely to have known security vulnerabilities. Versions less than v0.10.0 are from a previous architecture and should not be used.

Include the OpenWPM version number in your publication. As of v0.10.0 OpenWPM pins all python, npm, and system dependencies. Including this information alongside your work will allow other researchers to contextualize the results, and can be helpful if future versions of OpenWPM have instrumentation bugs that impact results.

Developer instructions

If you want to contribute to OpenWPM have a look at our CONTRIBUTING.md

Instrumentation and Configuration

OpenWPM provides a breadth of configuration options which can be found in Configuration.md More detail on the output is available below.

Storage

OpenWPM distinguishes between two types of data, structured and unstructured. Structured data is all data captured by the instrumentation or emitted by the platform. Generally speaking all data you download is unstructured data.

For each of the data classes we offer a variety of storage providers, and you are encouraged to implement your own, should the provided backends not be enough for you.

We have an outstanding issue to enable saving content generated by commands, such as screenshots and page dumps to unstructured storage (see #232).
For now, they get saved to manager_params.data_directory.

Local Storage

For storing structured data locally we offer two StorageProviders:

• The SQLiteStorageProvider which writes all data into a SQLite database
  • This is the recommended approach for getting started as the data is easily explorable
• The LocalArrowProvider which stores the data into Parquet files.
  • This method integrates well with NumPy/Pandas
  • It might be harder to ad-hoc process

For storing unstructured data locally we also offer two solutions:

• The LevelDBProvider which stores all data into a LevelDB
  • This is the recommended approach
• The LocalGzipProvider that gzips and stores the files individually on disk
  • Please note that file systems usually don't like thousands of files in one folder
  • Use with care or for single site visits

Remote storage

When running in the cloud, saving records to disk is not a reasonable thing to do. So we offer a remote StorageProviders for S3 (See #823) and GCP. Currently, all remote StorageProviders write to the respective object storage service (S3/GCS). The structured providers use the Parquet format.

NOTE: The Parquet and SQL schemas should be kept in sync except output-specific columns (e.g., instance_id in the Parquet output). You can compare the two schemas by running diff -y openwpm/storage/schema.sql openwpm/storage/parquet_schema.py.

Docker Deployment for OpenWPM

OpenWPM can be run in a Docker container. This is similar to running OpenWPM in a virtual machine, only with less overhead.

Building the Docker Container

Step 1: install Docker on your system. Most Linux distributions have Docker in their repositories. It can also be installed from docker.com. For Ubuntu you can use: sudo apt-get install docker.io

You can test the installation with: sudo docker run hello-world

Note, in order to run Docker without root privileges, add your user to the docker group (sudo usermod -a -G docker $USER). You will have to logout-login for the change to take effect, and possibly also restart the Docker service.

Step 2: to build the image, run the following command from a terminal within the root OpenWPM directory:

    docker build -f Dockerfile -t openwpm .

After a few minutes, the container is ready to use.

Running Measurements from inside the Container

You can run the demo measurement from inside the container, as follows:

First of all, you need to give the container permissions on your local X-server. You can do this by running: xhost +local:docker

Then you can run the demo script using:

    mkdir -p docker-volume && docker run -v $PWD/docker-volume:/opt/OpenWPM/datadir \
    -e DISPLAY=$DISPLAY -v /tmp/.X11-unix:/tmp/.X11-unix --shm-size=2g \
    -it --init openwpm

Note: the --shm-size=2g parameter is required, as it increases the amount of shared memory available to Firefox. Without this parameter you can expect Firefox to crash on 20-30% of sites.

This command uses bind-mounts to share scripts and output between the container and host, as explained below (note the paths in the command assume it's being run f