Dashboard

Before you can run anything, you need to install all dependencies. It's recommended to do this in a virtual environment:

python3.12 -m venv env
source env/bin/activate
python -m pip install -r requirements.txt

After you run this once, next time you only need to run

source env/bin/activate

before you can use the available scripts. Note that Python 3.12 is required.

You can start the backend with gunicorn app:app. In addition to the API, the backend does the following:

• On startup, if match-ids.txt or config.py is newer than out-min/dashboard.json, the server will automatically update all the datasets.
• Every midnight PST, the datasets are refreshed (after enough time, older matches will be excluded from some stats).

Each API endpoint is registered via add_url_rule in app.py.

All data is derived from either tracker.gg or the unofficial VALORANT API. samples contain example JSON from these sources. Since bot detection has improved, scraping from tracker.gg is now deprecated, so data from there can no longer (reliably) be scraped. Consequently, previously scraped data needs to be manually acquired, and all future matches will be fetched via the unofficial API.

Code for fetching and parsing from each data source can be found in data_providers.

tracker-match-ids.txt contains all the matches that were scraped from tracker.gg and will no longer be updated. match-ids.txt contains all matches fetched from the unofficial API and will be updated.

Each line in tracker-data.jsonl and api-data.jsonl is mapped to a Match object, which exposes all data with an API (see Match.py). To see examples of how to use it, see notebooks/example.ipynb.

From the main scraped data, we generate smaller datasets with derived statistics that we're interested in. Each of these datasets corresponds to a DatasetGenerator implementation in dataset_generators. Each DatasetGenerator needs to implement three methods:

Once a DatasetGenerator is implemented, it needs to be included somewhere in generate_datasets.py.

All datasets used for dashboards can be found in the out-min directory. To generate them manually, run generate_datasets.py to generate all the smaller datasets used by the front-end. They will appear in out-min.

There is a simple integration test in integ-test.py which tests the codebase against a database with one match to handle. API endpoints can be tested using Bruno; there's already configuration for it included.

New players should be registered as an enum in constants/player.py and all their aliases in config.py. In addition to changes in this codebase, we also need to register them in the frontend (profile picture, colors, current tag, etc.).