Handle citygml-tools directory outputs and doc cjio paths

scripts/split_semantics.py

@@ -41,8 +41,8 @@ def parse_args(argv: Iterable[str] | None = None) -> argparse.Namespace:
     )
     parser.add_argument(
         "--pattern",
-        default="*.city.json",
-        help="Glob pattern for input files (defaults to all .city.json files).",
+        default="**/*.json",
+        help="Glob pattern for input files (defaults to any .json under the input dir).",
     )
     return parser.parse_args(argv)
 
@@ -65,9 +65,26 @@ def base_name(path: Path) -> str:
         return name[: -len(".tri.city.json")]
     if name.endswith(".city.json"):
         return name[: -len(".city.json")]
+    if name.endswith(".json"):
+        return name[: -len(".json")]
     return path.stem
 
 
+def resolve_input_file(path: Path) -> Path | None:
+    """Handle both flat files and citygml-tools style output directories."""
+    if path.is_file():
+        return path
+    if path.is_dir():
+        # citygml-tools writes <name>.city.json/<name>.json
+        candidate = path / f"{path.stem}.json"
+        if candidate.is_file():
+            return candidate
+        matches = list(path.glob("*.json"))
+        if len(matches) == 1:
+            return matches[0]
+    return None
+
+
 def allowed_semantic_indices(semantics: dict[str, Any], allowed_types: set[str]) -> set[int]:
     surfaces = semantics.get("surfaces") or []
     return {idx for idx, surface in enumerate(surfaces) if surface.get("type") in allowed_types}
@@ -156,7 +173,12 @@ def filter_cityjson(cityjson: dict[str, Any], allowed_types: set[str]) -> dict[s
 
 
 def process_file(path: Path, targets: list[str], output_dir: Path) -> int:
-    cityjson = read_json(path)
+    resolved = resolve_input_file(path)
+    if not resolved:
+        print(f"[skip] cannot resolve CityJSON file for {path}", file=sys.stderr)
+        return 0
+
+    cityjson = read_json(resolved)
     written = 0
     base = base_name(path)