Ensure a dataset is downloaded and ready to use
(dataset_name)
| 265 | |
| 266 | |
| 267 | def ensure_dataset_available(dataset_name): |
| 268 | """Ensure a dataset is downloaded and ready to use""" |
| 269 | # Create datasets directory if it doesn't exist |
| 270 | datasets_dir = "datasets" |
| 271 | if not os.path.exists(datasets_dir): |
| 272 | os.makedirs(datasets_dir) |
| 273 | print(f"Created datasets directory at {os.path.abspath(datasets_dir)}") |
| 274 | |
| 275 | # Create dataset-specific directory if it doesn't exist |
| 276 | dataset_dir = os.path.join(datasets_dir, dataset_name) |
| 277 | if not os.path.exists(dataset_dir): |
| 278 | os.makedirs(dataset_dir) |
| 279 | print(f"Created dataset directory at {os.path.abspath(dataset_dir)}") |
| 280 | |
| 281 | # Check if dataset files exist |
| 282 | if not any(f.endswith(".parquet") for f in os.listdir(dataset_dir)): |
| 283 | print(f"Dataset {dataset_name} not found. Downloading...") |
| 284 | prepare_dataset(dataset_name) |
| 285 | else: |
| 286 | print(f"Dataset {dataset_name} is already available at {dataset_dir}") |
| 287 | |
| 288 | |
| 289 | def prompt_and_get_dataset_metadata(): |
no test coverage detected