Files
group_fqcd_jr/app/service/risk_natural_language.py
T

137 lines
4.2 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
"""风控预警自然语言筛选条件解析。"""
from __future__ import annotations
import re
from datetime import UTC, date, datetime, time, timedelta
from zoneinfo import ZoneInfo
from app.core.risk_contracts import RiskAlertQuery
SHANGHAI = ZoneInfo("Asia/Shanghai")
def parse_risk_alert_filters(
message: str,
*,
now: datetime | None = None,
) -> dict[str, str]:
"""把常见中文筛选表达转换为风控查询工具参数。"""
current = now or datetime.now(SHANGHAI)
if current.tzinfo is None:
current = current.replace(tzinfo=SHANGHAI)
filters: dict[str, str] = {}
level = re.search(r"(低|中|高)(?:风险)?", message)
if level:
filters["risk_level"] = level.group(1)
rule = re.search(r"RW-\d{3}", message, re.IGNORECASE)
if rule:
filters["rule_code"] = rule.group(0).upper()
customer = re.search(
r"(?:客户编号|客户号|客户账户|客户)\s*[::#]?\s*"
r"([A-Za-z0-9_-]{1,64})",
message,
)
if customer:
filters["customer_no"] = customer.group(1)
elif matched := re.search(r"\b(CUST-\d+)\b", message, re.IGNORECASE):
filters["customer_no"] = matched.group(1).upper()
product_code = re.search(
r"(?:产品编号|产品代码|产品编码)\s*[::#]?\s*"
r"([A-Za-z0-9_-]{1,64})",
message,
)
if product_code:
filters["product_code"] = product_code.group(1)
elif matched := re.search(r"\b(P-[A-Za-z0-9_-]{1,60})\b", message):
filters["product_code"] = matched.group(1)
product_name = re.search(
r"产品名称\s*[::#]?\s*([^\s,,。;;?!]{1,128})",
message,
)
if product_name is None:
product_name = re.search(
r"产品\s*[::]\s*([^\s,,。;;?!]{1,128})",
message,
)
if product_name:
candidate = product_name.group(1).strip()
if candidate not in {"编号", "代码", "编码", "名称"}:
filters["product_name"] = candidate
start_time, end_time = _parse_time_range(message, current)
if start_time is not None:
filters["start_time"] = _to_utc_iso(start_time)
if end_time is not None:
filters["end_time"] = _to_utc_iso(end_time)
validated = RiskAlertQuery.model_validate(filters)
return validated.model_dump(mode="json", exclude_none=True)
def _parse_time_range(
message: str,
current: datetime,
) -> tuple[datetime | None, datetime | None]:
explicit = re.search(
r"(\d{4}-\d{2}-\d{2})\s*(?:至|到|~|—)\s*(\d{4}-\d{2}-\d{2})",
message,
)
if explicit:
start_date = _parse_date(explicit.group(1))
end_date = _parse_date(explicit.group(2))
if start_date is not None and end_date is not None:
return (
datetime.combine(start_date, time.min, tzinfo=SHANGHAI),
datetime.combine(end_date, time.max, tzinfo=SHANGHAI),
)
if "今天" in message or "今日" in message:
start = current.replace(hour=0, minute=0, second=0, microsecond=0)
return start, current
if "本月" in message:
start = current.replace(
day=1,
hour=0,
minute=0,
second=0,
microsecond=0,
)
return start, current
recent = re.search(r"(?:近|最近)\s*(\d+)\s*(天|日)", message)
if recent:
days = int(recent.group(1))
return current - timedelta(days=days), current
since = re.search(r"(\d{4}-\d{2}-\d{2})\s*(?:以来|之后|起)", message)
if since:
start_date = _parse_date(since.group(1))
if start_date is not None:
return datetime.combine(start_date, time.min, tzinfo=SHANGHAI), current
until = re.search(r"(?:截至|截止)\s*(\d{4}-\d{2}-\d{2})", message)
if until:
end_date = _parse_date(until.group(1))
if end_date is not None:
return None, datetime.combine(end_date, time.max, tzinfo=SHANGHAI)
return None, None
def _to_utc_iso(value: datetime) -> str:
return value.astimezone(UTC).replace(tzinfo=None).isoformat()
def _parse_date(value: str) -> date | None:
try:
return date.fromisoformat(value)
except ValueError:
return None