Added Configuration UI and REST Endpoints

[dominiquekleeven]

Closes #14
Closes #9

---

Summary

This PR introduces the following changes:

---

REST API

ModelConfig Management Endpoints

• Full CRUD support following REST Level 2 principles.
• Tenancy-aware logic.
• Request body validation using Pydantic.
• External validation via OpenRemote when modifying/creating a config:
  • Verifies asset existence.
  • Validates realm correctness.

OpenRemote Utility Endpoints

• Retrieve assets with datapoint attributes.
• Retrieve assets by a list of IDs (for display in the overview).
• Retrieve realm config (minimal schema with styling only).

UI Bundle Endpoint

• Serves the built front-end bundle from the environment variable: ML_WEB_DIST_DIR
• The bundle is served at /ui

---

Web UI Frontend

Note: Currently does not have localization! Still needs to be figured out how we want to handle this for services.

Tech Stack

• LIT for extending OR components and building the pages.
• Rspack as a modern Webpack alternative.
• Vaadin Router for routing.
• OpenRemote Web Components.

---

Pages

Model Configs Overview

• Displays all model configs for the current realm.
• Embedded in the WIP Services page of the Manager UI.

Model Config Editor

• Large form for creating or editing a model config.
• Uses native HTML5 form validation.
• Searchable dropdown for selecting valid assets.
• Attribute dropdown updates dynamically based on selected asset.
• Also embedded in the WIP Services page.

---

Bug Fixes

• Fixed a bug where regressors with identical attribute names caused dataset creation to fail.

Other

• Updated the CI workflow to include a quality check flow for the front-end, uses eslint and prettier.
• Updated the Dockerfile to build the front-end and include it in the image, so that the back-end can serve it.
• DIR Paths are now constants, and cannot be overriden by ENV.
• Model files are now properly cleaned up after a config has been deleted.
• Changed some of the default values for the prophet model configuration.
• ML_WEBSERVER_ORIGINS was added to specify the allowed origins.
• Added GZIP middleware to compress responses over 1kb in size

---

Testing

Automated (Back-End)

• Tests added for:
  • ModelConfig CRUD endpoints.
  • OpenRemote utility endpoints.
  • Validation of request bodies.

Manual (Front-End)

• Currently tested manually.
• Uses OpenRemote components.
• E2E testing could be added later if needed.

---

Local Testing Instructions

I've updated the README.md with additional setup instructions.

Things to make sure of:

• Active OR instance (Demo Setup) is running locally
• Create a service user with Super Admin permissions
• Pass the credentials via the ML_OR_SERVICE_USER and ML_OR_SERVICE_SECRET environment variables.
  You can also create a .env file in the root directory
  

And then just follow the README instructions for starting the back-end and front-end.

---

It is also deployed on test3 and is connected to the OR instance on test3 You can find it in the services page.

---

[dominiquekleeven]

Test3 is up to date with the changes from this PR. I've added a Services page to the Manager that loads the UI as an iframe from the service.

Notably test3 also includes changes from #30, but this doesn't change any of the functionality besides introducing protected endpoints/SSO auth

Services page with the iframe loaded

[MartinaeyNL]

I've taken a look at your frontend code, and made a few remarks.
Feel free to check them out or ask for clarification 👍

Great work!

[dominiquekleeven]

@MartinaeyNL Thanks for the review, appreciate it.

I've addressed all the requested changes, feel free to take a look again when you have the chance. Re-requested the review :)

[MartinaeyNL]

I'll walk through the changes one more time. Give me a bit of time 😉

[MartinaeyNL]

Made 1 comment on the Rspack dependency. When addressed, the PR is OK to merge 👍