""" 海运运输分类:只认 Port to Port、其他。 本文件职责:从原话收成标准词、生成补问清单。 禁止:发明第三项;把分类带去 TMS;改六态。 线程:纯函数,可在 Worker 槽调用。 """ from __future__ import annotations SEA_CLASSES = ("Port to Port", "其他") # 长词优先,避免 P2P 误伤其它词;门到门不在名单里。 _ALIASES = ( ("port to port", "Port to Port"), ("港到港", "Port to Port"), ("p2p", "Port to Port"), ("其它", "其他"), ("其他", "其他"), ) def canonicalize_sea_class(raw: str) -> str: """ 把销售原话收成 Port to Port 或 其他;对不上返回空串。 「门到门」不算。先匹配更长别名。副作用:无。 """ text = (raw or "").strip() if not text: return "" lowered = text.lower() for needle, canon in _ALIASES: if needle in lowered or needle in text: return canon stripped = text.replace("运输分类", "").replace(":", "").replace(":", "").strip() if stripped in SEA_CLASSES: return stripped return "" def harvest_sea_class(text: str, current: dict[str, str] | None = None) -> dict[str, str]: """ 当前句能认则覆盖;认不出则保留已有标准值。 参数:本轮原文、已有 facts。返回合并后的 facts 副本。 非法旧值会丢掉,避免「门到门」一直占着必填位。 """ merged = dict(current or {}) found = canonicalize_sea_class(text or "") if found: merged["运输分类"] = found return merged kept = canonicalize_sea_class(str(merged.get("运输分类") or "")) if kept: merged["运输分类"] = kept else: merged.pop("运输分类", None) return merged # 运输分类用语,不能冒充贸易条款。销售原话写了「贸易条款」才保留。 _DOOR_TRADE_TERMS = frozenset( { "门到门", "door to door", "door-to-door", "doortodoor", "d2d", } ) def drop_invented_door_trade_terms(text: str, facts: dict[str, str] | None = None) -> dict[str, str]: """ 丢掉冒充贸易条款的「门到门」。 测服 WO202609200013:原话没写贸易条款,确认卡却出现「贸易条款:门到门」。 门到门只是运输分类说法,不是 FOB/CIF 这类条款;原话没有「贸易条款」四字时不准收下。 销售明确写了「贸易条款:门到门」则保留原文,不改六态。 """ merged = dict(facts or {}) trade = str(merged.get("贸易条款") or "").strip() if not trade: return merged if "贸易条款" in (text or ""): return merged if trade.lower().replace(" ", "") in {x.replace(" ", "") for x in _DOOR_TRADE_TERMS} or trade in _DOOR_TRADE_TERMS: merged.pop("贸易条款", None) return merged def sea_class_prompt() -> str: """缺运输分类时只列两项,不再追问「其他是什么」。""" return "运输分类(必填):请回复 Port to Port 或 其他。"