Download file from URL and cache it locally Args: url: File URL to download media_type: 'image' or 'video' Returns: Local cache filename
(self, url: str, media_type: str)
| 277 | return message |
| 278 | |
| 279 | async def download_and_cache(self, url: str, media_type: str) -> str: |
| 280 | """ |
| 281 | Download file from URL and cache it locally |
| 282 | |
| 283 | Args: |
| 284 | url: File URL to download |
| 285 | media_type: 'image' or 'video' |
| 286 | |
| 287 | Returns: |
| 288 | Local cache filename |
| 289 | """ |
| 290 | filename = self._generate_cache_filename(url, media_type) |
| 291 | file_path = self.cache_dir / filename |
| 292 | download_lock = self._download_locks.setdefault(filename, asyncio.Lock()) |
| 293 | |
| 294 | async with download_lock: |
| 295 | # Check if already cached and not expired |
| 296 | if file_path.exists(): |
| 297 | if self._is_cleanup_disabled(): |
| 298 | return filename |
| 299 | file_age = time.time() - file_path.stat().st_mtime |
| 300 | if file_age < self.default_timeout: |
| 301 | debug_logger.log_info(f"Cache hit: {filename}") |
| 302 | return filename |
| 303 | try: |
| 304 | file_path.unlink() |
| 305 | except Exception: |
| 306 | pass |
| 307 | |
| 308 | # Download file |
| 309 | debug_logger.log_info(f"Downloading file from: {url}") |
| 310 | |
| 311 | fingerprint = self._get_request_fingerprint() |
| 312 | proxy_url = await self._resolve_download_proxy(media_type, fingerprint=fingerprint) |
| 313 | headers = self._build_download_headers(media_type, fingerprint=fingerprint) |
| 314 | |
| 315 | # Try method 1: curl_cffi with browser impersonation |
| 316 | try: |
| 317 | async with AsyncSession() as session: |
| 318 | response = await session.get( |
| 319 | url, |
| 320 | timeout=60, |
| 321 | proxy=proxy_url, |
| 322 | headers=headers, |
| 323 | impersonate="chrome120", |
| 324 | verify=False |
| 325 | ) |
| 326 | |
| 327 | if response.status_code == 200 and response.content: |
| 328 | self._write_cached_content(file_path, response.content) |
| 329 | debug_logger.log_info( |
| 330 | f"File cached (curl_cffi): {filename} ({len(response.content)} bytes)" |
| 331 | ) |
| 332 | return filename |
| 333 | debug_logger.log_warning( |
| 334 | f"curl_cffi failed with HTTP {response.status_code}, trying wget..." |
| 335 | ) |
| 336 |
no test coverage detected