空运人工价锁舱现拉补全舱位并拦清缺参；成交/报价填表与工单陆运费率展示同步收紧。

Co-authored-by: Cursor <cursoragent@cursor.com>

inquiry-agent/agent/llm/extract_text.py

@@ -757,6 +757,30 @@ _RE_MONEY = re.compile(
     r"|(?:USD|CNY|RMB|美金|美元)\s*(\d+(?:\.\d+)?)",
     re.I,
 )
+# 陆运集拼：「2-500KG 单价CNY990」是分档单价，不是「燃油附加费=2」。
+_RE_WEIGHT_BAND_PRICE = re.compile(
+    r"(?P<min>\d+)\s*[-~～—–到至]\s*(?P<max>\d+)\s*K\s*G"
+    r"(?:\s*以上)?"
+    r"\s*(?:单价)?"
+    r"\s*(?:(?P<ccy>USD|CNY|RMB|美金|美元|人民币)\s*)?"
+    r"(?P<price>\d+(?:\.\d+)?)",
+    re.I,
+)
+_RE_WEIGHT_BAND_OPEN = re.compile(
+    r"(?P<min>\d+)\s*K\s*G\s*以上"
+    r"\s*(?:单价)?"
+    r"\s*(?:(?P<ccy>USD|CNY|RMB|美金|美元|人民币)\s*)?"
+    r"(?P<price>\d+(?:\.\d+)?)",
+    re.I,
+)
+_RE_FUEL_SURCHARGE_RATE = re.compile(
+    r"燃油附加费率\s*[:：=]?\s*(?P<rate>\d+(?:\.\d+)?)\s*%?",
+    re.I,
+)
+_RE_TAX_RATE = re.compile(
+    r"(?<![A-Za-z\u4e00-\u9fff])税率\s*[:：=]?\s*(?P<rate>\d+(?:\.\d+)?)\s*%?",
+    re.I,
+)
 _RE_FEE_LINE = re.compile(
     r"(?P<name>[\u4e00-\u9fffA-Za-z][\u4e00-\u9fffA-Za-z0-9/()（）+＋\-－]{0,24}费|[A-Za-z]{2,10})"
     r"\s*(?:改为|改成|调整为|调整成)?"
@@ -867,6 +891,8 @@ def extract_product_quote(text: str, *, injected: Optional[dict[str, Any]] = Non
     认不出价格返回空 dict，调用方回「没有识别到报价」。
     群里常见「码头操作费100CNY / 报关费改为300CNY」，不能只把第一笔数当成海运费。
     「空运费 单价…金额…」必须保留空运费，金额用「金额」后面的数，不用单价。
+    陆运集拼：「2-500KG 单价CNY990，燃油附加费率1%，税率6%」走分档单价+费率，
+    禁止把公斤档下限「2」当成燃油附加费金额（WO021）。
     """
     if isinstance(injected, dict) and (injected.get("total") or injected.get("fee_rows")):
         out = dict(injected)
@@ -879,16 +905,22 @@ def extract_product_quote(text: str, *, injected: Optional[dict[str, Any]] = Non
     compact = stripped.replace(" ", "").replace("　", "")
     if re.fullmatch(r"WO\d{12}", compact, re.I):
         return {}
-    rows = _parse_fee_rows(raw)
+    band = _parse_weight_band_quote(raw)
+    if band:
+        return band
+    # 费率句若仍落到普通费用行，会把「燃油附加费」配上句首公斤数。先剥掉再抽。
+    scrubbed = _strip_rate_and_band_noise(raw)
+    rows = _parse_fee_rows(scrubbed) if scrubbed.strip() else []
     if rows:
         ccy = rows[0].get("currency") or _currency_of(raw)
-        total = rows[0]["amount"]
-        if len({x.get("currency") for x in rows}) == 1:
+        money_rows = [x for x in rows if "%" not in str(x.get("amount") or "")]
+        total = (money_rows[0] if money_rows else rows[0])["amount"]
+        if money_rows and len({x.get("currency") for x in money_rows}) == 1:
             try:
-                total = str(sum(float(x["amount"]) for x in rows))
+                total = str(sum(float(x["amount"]) for x in money_rows))
             except ValueError:
-                total = rows[0]["amount"]
-        return {
+                total = money_rows[0]["amount"]
+        out = {
             "total": total,
             "currency": ccy,
             "source_label": "产品报价",
@@ -896,9 +928,12 @@ def extract_product_quote(text: str, *, injected: Optional[dict[str, Any]] = Non
             "fee_rows": rows,
             "fee_lines": list(rows),
         }
-    hit = _RE_MONEY.search(_ascii_amount(raw))
+        out.update(_extract_rate_fields(raw))
+        return out
+    hit = _RE_MONEY.search(_ascii_amount(scrubbed or raw))
     if not hit:
-        return {}
+        rates = _extract_rate_fields(raw)
+        return rates if rates else {}
     amount = _ascii_amount(hit.group(1) or hit.group(2) or "")
     if not amount:
         return {}
@@ -906,12 +941,13 @@ def extract_product_quote(text: str, *, injected: Optional[dict[str, Any]] = Non
     # 句里已经点了费用名（空运费）时，不能退回默认海运费。
     named = re.search(
         r"([\u4e00-\u9fffA-Za-z][\u4e00-\u9fffA-Za-z0-9/()（）+＋\-－]{0,24}费)",
-        raw,
+        scrubbed or raw,
     )
     fee_name = (named.group(1) if named else "") or "海运费"
-    if fee_name in {"费用"}:
-        fee_name = "海运费"
-    return {
+    if fee_name in {"费用", "燃油附加费"}:
+        # 「燃油附加费率」里的「燃油附加费」不是费用行名；无其它费名时用运费。
+        fee_name = "海运费" if fee_name == "费用" else "陆运费"
+    out = {
         "total": amount,
         "currency": ccy,
         "source_label": "产品报价",
@@ -919,3 +955,113 @@ def extract_product_quote(text: str, *, injected: Optional[dict[str, Any]] = Non
         "fee_rows": [{"name": fee_name, "amount": amount, "currency": ccy}],
         "fee_lines": [{"name": fee_name, "amount": amount, "currency": ccy}],
     }
+    out.update(_extract_rate_fields(raw))
+    return out
+
+
+def _format_percent_rate(raw: str) -> str:
+    """费率展示统一带 %；已有 % 保留。"""
+    text = (raw or "").strip()
+    if not text:
+        return ""
+    if text.endswith("%"):
+        return text
+    return f"{text}%"
+
+
+def _extract_rate_fields(text: str) -> dict[str, str]:
+    """抽出燃油附加费率 / 税率；只返回出现过的键。"""
+    out: dict[str, str] = {}
+    fuel = _RE_FUEL_SURCHARGE_RATE.search(text or "")
+    if fuel:
+        out["fuel_surcharge_rate"] = _format_percent_rate(fuel.group("rate") or "")
+        out["fuelSurchargeRate"] = out["fuel_surcharge_rate"]
+    tax = _RE_TAX_RATE.search(text or "")
+    if tax:
+        out["tax_rate"] = _format_percent_rate(tax.group("rate") or "")
+        out["taxRate"] = out["tax_rate"]
+    return out
+
+
+def _strip_rate_and_band_noise(text: str) -> str:
+    """去掉分档价与费率短语，避免后续费用行/金额回退误伤。"""
+    cleaned = _ascii_amount(text or "")
+    cleaned = _RE_WEIGHT_BAND_PRICE.sub(" ", cleaned)
+    cleaned = _RE_WEIGHT_BAND_OPEN.sub(" ", cleaned)
+    cleaned = _RE_FUEL_SURCHARGE_RATE.sub(" ", cleaned)
+    cleaned = _RE_TAX_RATE.sub(" ", cleaned)
+    return cleaned
+
+
+def _parse_weight_band_quote(text: str) -> dict[str, Any]:
+    """
+    陆运集拼分档单价：2-500KG 单价CNY990 + 可选燃油附加费率/税率。
+
+    产出 weightPrices，供主账出单按「2-500KG」表头对档；费率进顶层字段填 Excel。
+    """
+    raw = _ascii_amount(text or "")
+    tiers: list[dict[str, Any]] = []
+    rows: list[dict[str, str]] = []
+    ccy = ""
+    for hit in _RE_WEIGHT_BAND_PRICE.finditer(raw):
+        lo = int(hit.group("min") or 0)
+        hi = int(hit.group("max") or 0)
+        price = _ascii_amount(hit.group("price") or "")
+        if not price:
+            continue
+        line_ccy = _currency_of(raw, hit.group("ccy") or "")
+        ccy = ccy or line_ccy
+        label = f"{lo}-{hi}KG"
+        tiers.append({"min": lo, "max": hi, "price": price})
+        rows.append(
+            {
+                "name": label,
+                "amount": price,
+                "currency": line_ccy,
+                "unit_price": price,
+            }
+        )
+    for hit in _RE_WEIGHT_BAND_OPEN.finditer(raw):
+        lo = int(hit.group("min") or 0)
+        price = _ascii_amount(hit.group("price") or "")
+        if not price:
+            continue
+        line_ccy = _currency_of(raw, hit.group("ccy") or "")
+        ccy = ccy or line_ccy
+        label = f"{lo}KG以上"
+        tiers.append({"min": lo, "max": None, "price": price})
+        rows.append(
+            {
+                "name": label,
+                "amount": price,
+                "currency": line_ccy,
+                "unit_price": price,
+            }
+        )
+    if not tiers:
+        return {}
+    rates = _extract_rate_fields(raw)
+    # 费率行进明细方便群摘要展示；金额带 %，合计时只加数字价。
+    if rates.get("fuel_surcharge_rate"):
+        rows.append(
+            {
+                "name": "燃油附加费率",
+                "amount": rates["fuel_surcharge_rate"],
+                "currency": "",
+            }
+        )
+    if rates.get("tax_rate"):
+        rows.append({"name": "税率", "amount": rates["tax_rate"], "currency": ""})
+    # 分档是单价，不能把多档相加冒充费用合计。
+    out: dict[str, Any] = {
+        "total": tiers[0]["price"],
+        "currency": ccy or "CNY",
+        "source_label": "产品报价",
+        "raw_text": (text or "").strip(),
+        "fee_rows": rows,
+        "fee_lines": list(rows),
+        "weightPrices": tiers,
+        "price_per_kg": tiers[0]["price"],
+    }
+    out.update(rates)
+    return out