process-mining-dfg-plugin

Native Apache Superset chart plugin that renders a Directly-Follows Graph (DFG) for process mining, computed by a pm4py-backed Flask Blueprint. See the project README for the full architecture.

This package ships only the compiled frontend bundle (dist/remoteEntry.js). It requires the companion backend Blueprint (extension/backend/process_mining/) to be installed and mounted on the same Superset instance — the chart calls POST /process-mining/api/dfg-from-rows and expects the bundle itself to be served under /process-mining/plugin/.

Installing on your own Superset instance

1. Install the backend on the machine/image running Superset:

   • Copy extension/backend/process_mining/ onto the Python path Superset uses (e.g. mount it and add it via sys.path.insert(0, ...) in superset_config_docker.py / superset_config.py).

   • Install its dependencies: pm4py>=2.7,<3.0, pandas>=2.0, pydantic>=2.0 (see extension/backend/requirements.txt).

   • Register the Blueprint and exempt it from CSRF:

     from process_mining.api import process_mining_bp
     from flask_wtf.csrf import CSRFProtect
     
     BLUEPRINTS = [process_mining_bp]
     csrf.exempt(process_mining_bp)

2. Enable dynamic plugins and point Superset at this package's bundle:

   FEATURE_FLAGS = {"DYNAMIC_PLUGINS": True}

   Then register the plugin (via the DYNAMIC_PLUGINS admin UI or the dynamic_plugin API) with a bundle URL served by your Blueprint, e.g. https://<your-superset-host>/process-mining/plugin/remoteEntry.js, or https://unpkg.com/process-mining-dfg-plugin@<version>/dist/remoteEntry.js if you are not self-hosting the bundle.

3. Restart Superset. The chart "Process Mining" should appear in the chart type picker under the "Other" category.

Column auto-detection

The chart auto-detects case:concept:name, concept:name and time:timestamp (and common variants) as Case ID / Activity / Timestamp columns, but any columns can be selected manually in the control panel.

License

Apache-2.0 — see LICENSE.