(blacklist_auto_path: str, blacklist_manual_path: str)
| 81 | |
| 82 | # ===================== 黑名单/纠错字典处理 ===================== |
| 83 | def load_blacklist(blacklist_auto_path: str, blacklist_manual_path: str) -> set: |
| 84 | def _extract_black_urls(file_path): |
| 85 | lines = read_txt(file_path) |
| 86 | urls = [] |
| 87 | for line in lines: |
| 88 | if "," in line: |
| 89 | url = line.split(',')[1].strip() |
| 90 | if url: |
| 91 | urls.append(url) |
| 92 | return urls |
| 93 | auto_urls = _extract_black_urls(blacklist_auto_path) |
| 94 | manual_urls = _extract_black_urls(blacklist_manual_path) |
| 95 | combined = set(auto_urls + manual_urls) |
| 96 | print(f"[INFO] 合并黑名单URL数: {len(combined)}") |
| 97 | return combined |
| 98 | |
| 99 | def load_corrections(corrections_path: str) -> dict: |
| 100 | corrections = {} |
no test coverage detected