A web service for scanning media hosted on a Matrix media repository.

This project requires libolm development headers, as well as libmagic to be installed on the system. On Debian/Ubuntu:

sudo apt install libolm-dev libmagic1

Then, preferably in a virtual environment, install the Matrix Content Scanner:

pip install matrix-content-scanner

Copy and edit the sample configuration file. Each key is documented in this file.

Then run the content scanner (from within your virtual environment if one was created):

python -m matrix_content_scanner.mcs -c CONFIG_FILE

Where CONFIG_FILE is the path to your configuration file.

See the API documentation for information about how clients are expected to interact with the Matrix Content Scanner.

Because it uses the same APIs and Olm pickle format as the legacy Matrix Content Scanner, this project can be used as a drop-in replacement. The only change (apart from the deployment instructions) is the configuration format:

• the server section is renamed web
• scan.tempDirectory is renamed scan.temp_directory
• scan.baseUrl is renamed download.base_homeserver_url (and becomes optional)
• scan.doNotCacheExitCodes is renamed scan.do_not_cache_exit_codes
• scan.directDownload is removed. Direct download always happens when download.base_homeserver_url is absent from the configuration file, and setting a value for it will always cause files to be downloaded from the server configured.
• proxy is renamed download.proxy
• middleware.encryptedBody.pickleKey is renamed crypto.pickle_key
• middleware.encryptedBody.picklePath is renamed crypto.pickle_path
• acceptedMimeType is renamed allowed_mimetypes
• requestHeader is renamed download.additional_headers and turned into a dictionary.

Note that the format of the cryptographic pickle file and key are compatible between this project and the legacy Matrix Content Scanner. If no file exist at that path one will be created automatically.

In a virtual environment with pip ≥ 21.1, run

pip install -e .[dev]

To run the unit tests, you can either use:

tox -e py

or

trial tests

To run the linters and mypy type checker, use ./scripts-dev/lint.sh.

The exact steps for releasing will vary; but this is an approach taken by the Synapse developers (assuming a Unix-like shell):

1. Set a shell variable to the version you are releasing (this just makes subsequent steps easier):

   version=X.Y.Z

2. Update setup.cfg so that the version is correct.

3. Stage the changed files and commit.

   git add -u
   git commit -m v$version -n

4. Push your changes.

   git push

5. When ready, create a signed tag for the release:

   git tag -s v$version

   Base the tag message on the changelog.

6. Push the tag.

   git push origin tag v$version

7. Create a release, based on the tag you just pushed, on GitHub or GitLab.

8. Create a source distribution and upload it to PyPI:

   python -m build
   twine upload dist/matrix_content_scanner-$version*