()
| 333 | self.stats.print_failure_summary() |
| 334 | |
| 335 | def main(): |
| 336 | if len(sys.argv) < 3: |
| 337 | print("Usage: python crawlUrls.py <input_file> <target_directory> [max_retries]") |
| 338 | sys.exit(1) |
| 339 | |
| 340 | input_file = sys.argv[1] |
| 341 | target_dir = sys.argv[2] |
| 342 | |
| 343 | # Parse additional arguments |
| 344 | max_retries = 3 # Default |
| 345 | |
| 346 | # Check for max_retries |
| 347 | if len(sys.argv) > 3: |
| 348 | max_retries = int(sys.argv[3]) |
| 349 | |
| 350 | # Read URLs |
| 351 | with open(input_file) as f: |
| 352 | urls = f.readlines() |
| 353 | |
| 354 | # Create and run crawler |
| 355 | crawler = SimpleCrawler( |
| 356 | target_dir=target_dir, |
| 357 | max_retries=max_retries |
| 358 | ) |
| 359 | |
| 360 | print("Starting crawler with sequential processing") |
| 361 | print(f"Max retries: {max_retries}") |
| 362 | |
| 363 | crawler.crawl_urls(urls) |
| 364 | |
| 365 | if __name__ == "__main__": |
| 366 | main() |
no test coverage detected