"""Sport name normalisation."""

import re

_MAPPING: dict[str, str] = {
    # cycling variants (FIT enums, GPX types, Strava API/CSV types)
    "cycling": "cycling",
    "biking": "cycling",
    "bike": "cycling",
    "ride": "cycling",
    "road_biking": "cycling",
    "road_cycling": "cycling",
    "mountain_biking": "cycling",
    "mountain_bike_ride": "cycling",
    "gravel_cycling": "cycling",
    "gravel_ride": "cycling",
    "cyclocross": "cycling",
    "indoor_cycling": "cycling",
    "indoor_ride": "cycling",
    "virtual_ride": "cycling",
    "e_biking": "cycling",
    "ebikeride": "cycling",
    "e_bike_ride": "cycling",
    "ebike_ride": "cycling",
    "handcycle": "cycling",
    "velomobile": "cycling",
    # running
    "running": "running",
    "run": "running",
    "trail_running": "running",
    "trail_run": "running",
    "treadmill_running": "running",
    "treadmill": "running",
    "virtual_run": "running",
    "outdoor_run": "running",
    "indoor_run": "running",
    "track_run": "running",
    # hiking
    "hiking": "hiking",
    "hike": "hiking",
    "outdoor_hike": "hiking",
    # walking
    "walking": "walking",
    "walk": "walking",
    "outdoor_walk": "walking",
    # skiing
    "skiing": "skiing",
    "cross_country_skiing": "skiing",
    "nordic_skiing": "skiing",
    "downhill_skiing": "skiing",
    "alpine_skiing": "skiing",
    "skate_skiing": "skiing",
    "backcountry_skiing": "skiing",
    # swimming
    "swimming": "swimming",
    "swim": "swimming",
    "open_water_swimming": "swimming",
    "lap_swimming": "swimming",
}

BAS_SPORTS = {"cycling", "running", "hiking", "walking", "swimming", "skiing", "other"}


def normalise_sport(raw: object) -> str:
    if raw is None:
        return "other"
    key = str(raw).lower().strip().replace(" ", "_").replace("-", "_")
    # Strip leading date-like prefixes e.g. "20231117outdoor_run" → "outdoor_run"
    key = re.sub(r"^\d+", "", key)
    return _MAPPING.get(key, "other")