Checks SFDA (Saudi Arabia) for medical device safety updates. Two sources: 1. NCMDR Weekly Alert page (weekly PDF reports) -- parsed from HTML table 2. SFDA RSS feed (general medical device news) -- filtered for device content
| 406 | # --------------------------------------------------------------------------- |
| 407 | |
| 408 | class SFDAChecker: |
| 409 | """Checks SFDA (Saudi Arabia) for medical device safety updates. |
| 410 | |
| 411 | Two sources: |
| 412 | 1. NCMDR Weekly Alert page (weekly PDF reports) -- parsed from HTML table |
| 413 | 2. SFDA RSS feed (general medical device news) -- filtered for device content |
| 414 | """ |
| 415 | |
| 416 | def __init__(self, session: requests.Session, state: dict, |
| 417 | seed_mode: bool = False): |
| 418 | self.session = session |
| 419 | self.state = state |
| 420 | self.seed_mode = seed_mode |
| 421 | |
| 422 | def check(self, source_id: str, source: dict) -> Optional[dict]: |
| 423 | check_sub = source.get("check_sub", "weekly_alert") |
| 424 | if check_sub == "rss": |
| 425 | return self._check_rss(source_id, source) |
| 426 | return self._check_weekly_alert(source_id, source) |
| 427 | |
| 428 | def _check_weekly_alert(self, source_id: str, source: dict) -> Optional[dict]: |
| 429 | url = source["url"] |
| 430 | prev = self.state.get(source_id, {}) |
| 431 | prev_ids = set(prev.get("seen_ids", [])) |
| 432 | |
| 433 | try: |
| 434 | resp = self.session.get(url, timeout=30) |
| 435 | resp.raise_for_status() |
| 436 | except Exception as e: |
| 437 | print(f" ERROR SFDA weekly: {e}") |
| 438 | return None |
| 439 | |
| 440 | entries = self._parse_weekly_page(resp.text) |
| 441 | if not entries: |
| 442 | print(" WARNING: No entries from SFDA weekly alert page") |
| 443 | return None |
| 444 | |
| 445 | print(f" INFO: Parsed {len(entries)} weekly reports from SFDA") |
| 446 | |
| 447 | new_items = [] |
| 448 | all_ids = list(prev_ids) |
| 449 | |
| 450 | for entry in entries: |
| 451 | eid = entry.get("id", entry.get("title", "")) |
| 452 | if not eid or eid in prev_ids: |
| 453 | continue |
| 454 | all_ids.append(eid) |
| 455 | new_items.append(entry) |
| 456 | |
| 457 | self.state[source_id] = { |
| 458 | "url": url, |
| 459 | "last_checked": datetime.now().isoformat(), |
| 460 | "seen_ids": all_ids[-200:], |
| 461 | } |
| 462 | |
| 463 | if new_items and (prev_ids or self.seed_mode): |
| 464 | if self.seed_mode and not prev_ids: |
| 465 | new_items = new_items[:5] |