diff --git a/scripts/tender-fetcher/fetch-xkgroup.py b/scripts/tender-fetcher/fetch-xkgroup.py index fe376a9..d990fdd 100755 --- a/scripts/tender-fetcher/fetch-xkgroup.py +++ b/scripts/tender-fetcher/fetch-xkgroup.py @@ -41,7 +41,16 @@ NOTICE_TYPES = { "D": "流标公告", "B": "结果公示", } -DEFAULT_TYPES = ["C"] # 默认只抓取寻源公告 +DEFAULT_TYPES = ["C", "D"] # 抓取寻源公告 + 流标公告 + +# 多次询价标题正则 (匹配 二次/三次/四次/.../九次) +RETRY_PATTERN = re.compile(r"[二三四五六七八九]次") + +# 发布组织白名单 (新矿供应链平台关注的发布方) +ORG_WHITELIST = ( + "新疆宝地矿业股份有限公司", + "新疆火烧云铅锌矿有限责任公司", +) OUTPUT_BASE = os.path.expanduser("~/desktop/招标公告") @@ -301,6 +310,16 @@ def run(): filtered_count += 1 continue + # 多次询价标题过滤 (二次/三次/四次...) + if not RETRY_PATTERN.search(notice["title"]): + filtered_count += 1 + continue + + # 发布组织白名单 + if not any(org_name in notice["org"] for org_name in ORG_WHITELIST): + filtered_count += 1 + continue + notice_id = notice["id"] # 去重