feat: provision regional Kempen buildings

backend/app/services/storage_service.py

@@ -95,6 +95,39 @@ class StorageService:
         }
         return metadata
 
+    @staticmethod
+    def persist_dataset_file_from_path(
+        project_id: str,
+        dataset_id: str,
+        dataset_type: str,
+        original_filename: str,
+        source_path: str | Path,
+        content_type: str | None,
+    ) -> dict[str, Any]:
+        source = Path(source_path).resolve()
+        if not source.is_file():
+            raise FileNotFoundError(f"Dataset source artifact does not exist: {source}")
+
+        normalized_type = StorageService.normalize_dataset_type(dataset_type)
+        file_path = Path(StorageService.dataset_file_path(project_id, dataset_id, normalized_type, original_filename))
+        file_path.parent.mkdir(parents=True, exist_ok=True)
+        digest = hashlib.sha256()
+        size_bytes = 0
+        with source.open("rb") as input_stream, file_path.open("wb") as output_stream:
+            for chunk in iter(lambda: input_stream.read(8 * 1024 * 1024), b""):
+                output_stream.write(chunk)
+                digest.update(chunk)
+                size_bytes += len(chunk)
+
+        return {
+            "original_filename": StorageService._safe_filename(original_filename),
+            "stored_filename": file_path.name,
+            "content_type": content_type or "application/octet-stream",
+            "size_bytes": size_bytes,
+            "checksum_sha256": digest.hexdigest(),
+            "storage_path": str(file_path),
+        }
+
     @staticmethod
     def persist_file(
         storage_path: str,