qizhong/inquiry_robot
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

产品调价继承工单币种（港币不再默成美元）；陆运集拼空分档只填询价重量那一档，跨境整车按已选线路写起讫。

Browse Source 
Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
linmianxun
2026-10-12 19:22:44 +08:00
co-authored by Cursor
parent 3a4c440354
commit 97f0219dfd
8 changed files with 678 additions and 47 deletions
Download Patch File Download Diff File Expand all files Collapse all files
inquiry-agent/agent/llm/extract_text.py
+57 -24
View File
@@ -11,14 +11,17 @@ import logging
import re
from typing import Any, Optional
from agent.schema.field_validate import QUOTE_DATE_SYNONYMS
try:
from agent.schema.field_validate import QUOTE_DATE_SYNONYMS
except ImportError: # 测服旧 release 未导出同义词时仍可抽产品价
QUOTE_DATE_SYNONYMS = ()
logger = logging.getLogger(__name__)
# 报价日期同义词。长的在前，避免短词截断。货好时间、提货地址不在这里。
_QUOTE_DATE_LABEL_RE = "|".join(
re.escape(name)
for name in sorted(("报价日期",) + QUOTE_DATE_SYNONYMS, key=len, reverse=True)
for name in sorted(("报价日期",) + tuple(QUOTE_DATE_SYNONYMS or ()), key=len, reverse=True)
)
_MODE_WORDS = (
@@ -829,9 +832,11 @@ def extract_air_collab_fields(
return {k: got[k] for k in keys if str(got.get(k) or "").strip()}
# 金额旁币种词：含港币；禁止只认 USD/CNY 把港币单漏成美元。
_CCY_TOKEN = r"USD|CNY|RMB|HKD|EUR|JPY|GBP|美金|美元|人民币|港币|港元|欧元|日元|英镑"
_RE_MONEY = re.compile(
r"(?:USD|CNY|RMB|美金|美元|人民币)?\s*(\d+(?:\.\d+)?)\s*(?:USD|CNY|RMB|美金|美元|人民币|元)?"
r"|(?:USD|CNY|RMB|美金|美元)\s*(\d+(?:\.\d+)?)",
rf"(?:{_CCY_TOKEN}|元)?\s*(\d+(?:\.\d+)?)\s*(?:{_CCY_TOKEN}|元)?"
rf"|(?:{_CCY_TOKEN})\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」。
@@ -839,14 +844,14 @@ _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*)?"
rf"\s*(?:(?P<ccy>{_CCY_TOKEN})\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*)?"
rf"\s*(?:(?P<ccy>{_CCY_TOKEN})\s*)?"
r"(?P<price>\d+(?:\.\d+)?)",
re.I,
)
@@ -862,9 +867,9 @@ _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)?"
r"\s*[:：=]?"
r"\s*(?:(?P<ccy1>USD|CNY|RMB|美金|美元|人民币)\s+)?"
rf"\s*(?:(?P<ccy1>{_CCY_TOKEN})\s+)?"
r"(?P<amount>\d+(?:\.\d+)?)"
r"\s*(?P<ccy2>USD|CNY|RMB|美金|美元|人民币|元)?",
rf"\s*(?P<ccy2>{_CCY_TOKEN}|元)?",
re.I,
)
# 「文件发改为：500」漏了「费」、中间还有冒号，不能丢掉。
@@ -873,19 +878,21 @@ _RE_FEE_CHANGE = re.compile(
r"\s*(?:改为|改成|调整为|调整成)"
r"\s*[:：=]?"
r"\s*(?P<amount>\d+(?:\.\d+)?)"
r"\s*(?P<ccy>USD|CNY|RMB|美金|美元|人民币|元)?",
rf"\s*(?P<ccy>{_CCY_TOKEN}|元)?",
re.I,
)
_SKIP_FEE_NAMES = frozenset({"usd", "cny", "rmb", "美金", "美元", "人民币", "元"})
_SKIP_FEE_NAMES = frozenset(
{"usd", "cny", "rmb", "hkd", "eur", "jpy", "gbp", "美金", "美元", "人民币", "港币", "港元", "元"}
)
# 航线常写「空运费 单价CNY99，数量100，金额CNY199」。单价和金额不是同一个数，必须认金额。
_RE_FEE_PRICED = re.compile(
r"(?P<name>[\u4e00-\u9fffA-Za-z][\u4e00-\u9fffA-Za-z0-9/()（）+＋\-－]{0,24}费)"
r"\s*单价\s*"
r"(?:(?P<unit_ccy>USD|CNY|RMB|美金|美元|人民币)\s*)?"
rf"(?:(?P<unit_ccy>{_CCY_TOKEN})\s*)?"
r"(?P<unit>\d+(?:\.\d+)?)"
r"(?:\s*[，,、]\s*数量\s*(?P<qty>\d+(?:\.\d+)?))?"
r"(?:\s*[，,、]\s*金额\s*"
r"(?:(?P<amt_ccy>USD|CNY|RMB|美金|美元|人民币)\s*)?"
rf"(?:(?P<amt_ccy>{_CCY_TOKEN})\s*)?"
r"(?P<amount>\d+(?:\.\d+)?))?",
re.I,
)
@@ -898,21 +905,40 @@ def _ascii_amount(raw: str) -> str:
def _currency_of(text: str, hint: str = "") -> str:
"""
从正文/提示抽三字码。正文没写币种时只回传 hint，禁止默认编造 USD。
港币必须先于「元」判断，避免「港元」被误判成 CNY。
WO202610090107：TMS 是 HKD，产品只写「运输费调整为1100」时必须继承 HKD。
"""
blob = f"{hint} {text}".upper()
if any(x in blob for x in ("CNY", "RMB", "人民币", "元")):
raw = f"{hint} {text}"
if any(x in blob for x in ("HKD", "港币", "港元", "HK$")) or "港币" in raw or "港元" in raw:
return "HKD"
if any(x in blob for x in ("CNY", "RMB", "人民币")) or (
"元" in raw and "港元" not in raw and "港币" not in raw
):
return "CNY"
if any(x in blob for x in ("USD", "美金", "美元")):
return "USD"
return hint or "USD"
if any(x in blob for x in ("EUR", "欧元")):
return "EUR"
if any(x in blob for x in ("JPY", "日元")):
return "JPY"
if any(x in blob for x in ("GBP", "英镑")):
return "GBP"
return str(hint or "").strip().upper()
def _parse_fee_rows(text: str) -> list[dict[str, str]]:
def _parse_fee_rows(text: str, baseline_currency: str = "") -> list[dict[str, str]]:
"""
按行抽费用：码头操作费100CNY、报关费改为300CNY。
只认带「费」的中文名或短英文费用码，避免把工单号当金额。
句里无币种时继承 baseline_currency（通常是当前 TMS/主账币种），禁止默认 USD。
"""
raw = _ascii_amount(text or "")
base_ccy = str(baseline_currency or "").strip().upper()
rows: list[dict[str, str]] = []
seen: set[str] = set()
@@ -925,10 +951,11 @@ def _parse_fee_rows(text: str) -> list[dict[str, str]]:
if name in seen:
return
seen.add(name)
line_ccy = _currency_of(raw, ccy or base_ccy) or base_ccy
row = {
"name": name,
"amount": amount,
"currency": _currency_of(raw, ccy),
"currency": line_ccy,
}
for key, val in extra.items():
if val:
@@ -986,6 +1013,7 @@ def extract_product_quote(
injected: Optional[dict[str, Any]] = None,
business_line: str = "",
transport_mode: str = "",
baseline_currency: str = "",
) -> dict[str, Any]:
"""
产品文字报价：先按费用行抽，再退回合计/币别。注入优先。
@@ -996,7 +1024,9 @@ def extract_product_quote(
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）。
无费用名时按 business_line/transport_mode 默认陆运费/空运费/海运费，禁止陆运硬套海运费。
正文未写币种时继承 baseline_currency（当前 TMS/主账币种），禁止默认 USD（WO107）。
"""
base_ccy = str(baseline_currency or "").strip().upper()
if isinstance(injected, dict) and (injected.get("total") or injected.get("fee_rows")):
out = dict(injected)
out.setdefault("source_label", "产品报价")
@@ -1008,14 +1038,14 @@ def extract_product_quote(
compact = stripped.replace(" ", "").replace("　", "")
if re.fullmatch(r"WO\d{12}", compact, re.I):
return {}
band = _parse_weight_band_quote(raw)
band = _parse_weight_band_quote(raw, baseline_currency=base_ccy)
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 []
rows = _parse_fee_rows(scrubbed, baseline_currency=base_ccy) if scrubbed.strip() else []
if rows:
ccy = rows[0].get("currency") or _currency_of(raw)
ccy = rows[0].get("currency") or _currency_of(raw, base_ccy) or base_ccy
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:
@@ -1040,7 +1070,7 @@ def extract_product_quote(
amount = _ascii_amount(hit.group(1) or hit.group(2) or "")
if not amount:
return {}
ccy = _currency_of(raw)
ccy = _currency_of(raw, base_ccy) or base_ccy
default_fee = _default_product_fee_name(business_line, transport_mode)
# 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(
@@ -1066,6 +1096,7 @@ def extract_product_quote(
return out
def _format_percent_rate(raw: str) -> str:
"""费率展示统一带 %；已有 % 保留。"""
text = (raw or "").strip()
@@ -1100,13 +1131,15 @@ def _strip_rate_and_band_noise(text: str) -> str:
return cleaned
def _parse_weight_band_quote(text: str) -> dict[str, Any]:
def _parse_weight_band_quote(text: str, baseline_currency: 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。
无币种时优先继承 baseline，再退回 CNY（集拼常见人民币报价）。
"""
raw = _ascii_amount(text or "")
base_ccy = str(baseline_currency or "").strip().upper()
tiers: list[dict[str, Any]] = []
rows: list[dict[str, str]] = []
ccy = ""
@@ -1116,7 +1149,7 @@ def _parse_weight_band_quote(text: str) -> dict[str, Any]:
price = _ascii_amount(hit.group("price") or "")
if not price:
continue
line_ccy = _currency_of(raw, hit.group("ccy") or "")
line_ccy = _currency_of(raw, hit.group("ccy") or base_ccy) or base_ccy
ccy = ccy or line_ccy
label = f"{lo}-{hi}KG"
tiers.append({"min": lo, "max": hi, "price": price})
@@ -1133,7 +1166,7 @@ def _parse_weight_band_quote(text: str) -> dict[str, Any]:
price = _ascii_amount(hit.group("price") or "")
if not price:
continue
line_ccy = _currency_of(raw, hit.group("ccy") or "")
line_ccy = _currency_of(raw, hit.group("ccy") or base_ccy) or base_ccy
ccy = ccy or line_ccy
label = f"{lo}KG以上"
tiers.append({"min": lo, "max": None, "price": price})
@@ -1162,7 +1195,7 @@ def _parse_weight_band_quote(text: str) -> dict[str, Any]:
# 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",
"currency": ccy or base_ccy or "CNY",
"source_label": "产品报价",
"raw_text": (text or "").strip(),
"fee_rows": rows,