| 616 | # --------------------------------------------------------------------------- |
| 617 | |
| 618 | class HTTPChecker: |
| 619 | def __init__(self, session: requests.Session, state: dict): |
| 620 | self.session = session |
| 621 | self.state = state |
| 622 | |
| 623 | def check(self, source_id: str, source: dict) -> Optional[dict]: |
| 624 | url = source["url"] |
| 625 | prev = self.state.get(source_id, {}) |
| 626 | headers = {} |
| 627 | if prev.get("etag"): |
| 628 | headers["If-None-Match"] = prev["etag"] |
| 629 | if prev.get("last_modified"): |
| 630 | headers["If-Modified-Since"] = prev["last_modified"] |
| 631 | try: |
| 632 | resp = self.session.head(url, headers=headers, timeout=15, allow_redirects=True) |
| 633 | new_state = {"url": url, "last_checked": datetime.now().isoformat(), |
| 634 | "status_code": resp.status_code} |
| 635 | if resp.headers.get("ETag"): |
| 636 | new_state["etag"] = resp.headers["ETag"] |
| 637 | if resp.headers.get("Last-Modified"): |
| 638 | new_state["last_modified"] = resp.headers["Last-Modified"] |
| 639 | self.state[source_id] = new_state |
| 640 | if resp.status_code == 304: |
| 641 | return None |
| 642 | elif resp.status_code == 200: |
| 643 | if prev.get("etag") and new_state.get("etag") == prev["etag"]: |
| 644 | return None |
| 645 | return _make_update(source_id, source, "http_head", |
| 646 | "Page may have been updated - manual review required") |
| 647 | else: |
| 648 | print(f" WARNING: HTTP {resp.status_code}") |
| 649 | return None |
| 650 | except requests.RequestException as e: |
| 651 | print(f" ERROR: {e}") |
| 652 | return None |
| 653 | |
| 654 | |
| 655 | # --------------------------------------------------------------------------- |