scripts/01_parse_emdat.py

"""Step 1: Parse EMDAT data and create event catalog with train/test splits."""

from src.data.emdat_parser import parse_emdat, save_events, split_events
from src.llm.client import load_config


def main():
    config = load_config()
    paths = config["paths"]

    print("Parsing EMDAT data...")
    events = parse_emdat(config)
    print(f"  Found {len(events)} European flood events")

    print("Splitting into train/test sets...")
    train, test = split_events(events, config)
    print(f"  Train (2023-2024): {len(train)} events")
    print(f"  Test (2025):       {len(test)} events")

    print("Saving...")
    save_events(events, paths["events_catalog"])
    save_events(train, paths["train_events"])
    save_events(test, paths["test_events"])

    print(f"  Catalog:  {paths['events_catalog']}")
    print(f"  Train:    {paths['train_events']}")
    print(f"  Test:     {paths['test_events']}")
    print("Done.")


if __name__ == "__main__":
    main()