trying to get sub label showed properly

bincio/extract/sport.py

@@ -47,10 +47,13 @@ _MAPPING: dict[str, str] = {
     "skiing": "skiing",
     "cross_country_skiing": "skiing",
     "nordic_skiing": "skiing",
+    "nordic_ski": "skiing",
     "downhill_skiing": "skiing",
     "alpine_skiing": "skiing",
+    "alpine_ski": "skiing",
     "skate_skiing": "skiing",
     "backcountry_skiing": "skiing",
+    "backcountry_ski": "skiing",
     # swimming
     "swimming": "swimming",
     "swim": "swimming",
@@ -58,13 +61,63 @@ _MAPPING: dict[str, str] = {
     "lap_swimming": "swimming",
 }
 
+_SUB_SPORT_MAPPING: dict[str, str] = {
+    # cycling
+    "road_biking":        "road",
+    "road_cycling":       "road",
+    "mountain_biking":    "mountain",
+    "mountain_bike_ride": "mountain",
+    "gravel_cycling":     "gravel",
+    "gravel_ride":        "gravel",
+    "cyclocross":         "gravel",
+    "indoor_cycling":     "indoor",
+    "indoor_ride":        "indoor",
+    "virtual_ride":       "indoor",
+    # running
+    "trail_running":      "trail",
+    "trail_run":          "trail",
+    "treadmill_running":  "indoor",
+    "treadmill":          "indoor",
+    "indoor_run":         "indoor",
+    "virtual_run":        "indoor",
+    "track_run":          "track",
+    # skiing
+    "cross_country_skiing":  "nordic",
+    "nordic_skiing":         "nordic",
+    "nordic_ski":            "nordic",
+    "skate_skiing":          "nordic",
+    "backcountry_skiing":    "nordic",
+    "backcountry_ski":       "nordic",
+    "downhill_skiing":       "alpine",
+    "alpine_skiing":         "alpine",
+    "alpine_ski":            "alpine",
+    # swimming
+    "open_water_swimming":   "open_water",
+    "lap_swimming":          "pool",
+}
+
 BAS_SPORTS = {"cycling", "running", "hiking", "walking", "swimming", "skiing", "other"}
 
 
+def _normalise_key(raw: object) -> str:
+    key = str(raw).strip()
+    # CamelCase → snake_case  ("MountainBikeRide" → "mountain_bike_ride")
+    key = re.sub(r"([A-Z])", r"_\1", key).lower().lstrip("_")
+    key = key.replace(" ", "_").replace("-", "_")
+    return re.sub(r"^\d+", "", key)
+
+
 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")
+    return _MAPPING.get(_normalise_key(raw), "other")
+
+
+def normalise_sub_sport(raw: object) -> str | None:
+    """Infer sub_sport from a raw sport type string (e.g. 'mountain_bike_ride' → 'mountain').
+
+    Returns None when no sub_sport is implied (e.g. plain 'ride', 'run').
+    """
+    if raw is None:
+        return None
+    return _SUB_SPORT_MAPPING.get(_normalise_key(raw))