9. Using it as a library

For anything beyond "run once, get files" — batch runs, custom output processing, coupling with other tools — use the Python API directly instead of the CLI:

from vflood_rri.model import load_model
from vflood_rri.engine import run
 
model = load_model("RRI_Input.txt")
print(model.river.count, "river cells")
 
def on_output(snap):
    print(snap.step, snap.hs.max())
 
final_state = run(model, on_output=on_output)

load_model() does all the static setup (grid, indexing, channel geometry — no physics yet); run() does the actual time-stepping and calls on_output/on_storage callbacks whenever there's a new snapshot to hand off. See cli.py for a complete, working example of wiring those callbacks up to real files.