1. git clone https://github.com/simra/FRCModeling
2. cd FRCModeling
3. git checkout simra/2025
4. Install Anaconda Python from here: https://www.anaconda.com/download
5. In the VS Code terminal, conda create --name frcmodeling python=3.10
6. conda activate frcmodeling
7. pip install -r requirements.txt
8. cd current

1. python fetchMatches.py --year 2024 --reset

   This will pull all the data for 2024.

   You can optionally pass --reset to re-pull all the data, and --year YEAR to pull data for a year other than 2024.

2. runScoutingReport.py

- backend\  - python flask app - see deploy.sh for build/deploy steps
- frontend\ - react front-end

1. Install node.js on your machine from https://nodejs.org/en/download/
2. Open a command prompt cd current\frontend
3. run npm install and then npm run build
4. Delete backend\static\build if it exists
5. Copy the contents of build to backend\static\build
6. cd ..\backend
7. pip install -r requirements.txt
8. python -m flask run

Model generation is in OPR.py. See runScoutingReport for an example of how it's used.

The backend uses an incremental, file-based cache (matches_<district>_<year>.pkl) to avoid redundant round-trips to The Blue Alliance (TBA).

1. On the first run the full events list and all per-event matches/teams are fetched from TBA and written to disk.
2. On subsequent runs the existing pickle file is loaded and the HTTP If-Modified-Since header is used for every request:
   • The events-list endpoint gets the global Last-Modified value saved with the cache.
   • Each event's matches/teams endpoint gets a per-event Last-Modified value stored in result['event_last_modified']. Events that TBA has not updated since the last fetch return HTTP 304 Not Modified and keep their cached data untouched.
3. A backup of the previous pickle is kept at <file>.bak before each write.
4. The /model/<key>/refresh API endpoint triggers a full incremental re-fetch and rebuilds the in-memory model.

Pass --reset (CLI) or call fetch_all_matches(reset=True) to force a complete re-fetch, ignoring all cached timestamps.

The pickle cache is a local file, which is ephemeral on most PaaS platforms (e.g., Azure App Service restarts wipe the local filesystem). To retain the cache between deployments, choose one of the following strategies:

The simplest production path is to mount an Azure Files share and set DATA_FOLDER=/mnt/data so the existing pickle-based logic works without any code change.