(self, source_id: str, source: dict)
| 894 | return new_items, domain_filtered, title_filtered |
| 895 | |
| 896 | def check(self, source_id: str, source: dict) -> Optional[dict]: |
| 897 | query = source.get("google_query") |
| 898 | if not query or not self.available: |
| 899 | if not self.available: |
| 900 | print(f" SKIP (Vertex AI Search not configured)") |
| 901 | return None |
| 902 | skip_domain = source.get("skip_domain_filter", False) |
| 903 | site_domain = None if skip_domain else self._extract_site_domain(query) |
| 904 | title_filter = source.get("title_filter") |
| 905 | exclude_domains = source.get("exclude_domains") |
| 906 | prev = self.state.get(source_id, {}) |
| 907 | prev_titles = set(prev.get("seen_titles", [])) |
| 908 | |
| 909 | items = self.search(query) |
| 910 | new_items, domain_filtered, tf_count = self._filter_items( |
| 911 | items, prev_titles, site_domain, title_filter, exclude_domains) |
| 912 | |
| 913 | if not new_items and domain_filtered > 0 and site_domain: |
| 914 | fallback_query = re.sub(r"site:\S+\s*", "", query).strip() |
| 915 | print(f" INFO: {domain_filtered} results from wrong domain; " |
| 916 | f"retrying without site: prefix") |
| 917 | items = self.search(fallback_query) |
| 918 | new_items, domain_filtered, tf_count = self._filter_items( |
| 919 | items, prev_titles, site_domain, title_filter, exclude_domains) |
| 920 | |
| 921 | if domain_filtered: |
| 922 | print(f" INFO: {domain_filtered} results filtered by domain ({site_domain})") |
| 923 | if tf_count: |
| 924 | print(f" INFO: {tf_count} results filtered by title_filter") |
| 925 | |
| 926 | all_titles = list(prev_titles) + [i["title"] for i in new_items] |
| 927 | self.state[source_id] = {"url": source["url"], |
| 928 | "last_checked": datetime.now().isoformat(), |
| 929 | "seen_titles": all_titles[-100:]} |
| 930 | if new_items and (prev_titles or self.seed_mode): |
| 931 | if self.seed_mode and not prev_titles: |
| 932 | new_items = new_items[:10] |
| 933 | print(f" INFO: Seed mode -- returning top {len(new_items)} results as initial news") |
| 934 | result = _make_update(source_id, source, "vertex_ai_search", |
| 935 | f"{len(new_items)} new result(s) via Vertex AI Search") |
| 936 | result["new_items"] = new_items |
| 937 | result["query"] = query |
| 938 | return result |
| 939 | elif not prev_titles: |
| 940 | if not new_items and items: |
| 941 | print(f" INFO: All {len(items)} results filtered out (domain/title mismatch)") |
| 942 | else: |
| 943 | print(f" INFO: Baseline established ({len(items)} results indexed)") |
| 944 | return None |
| 945 | |
| 946 | |
| 947 | # --------------------------------------------------------------------------- |
nothing calls this directly
no test coverage detected