feat:客服agent接入nl2sql

2026-09-13 20:48:42 +08:00
parent b133dd58c4
commit dd281b3361
11 changed files with 778 additions and 18 deletions
@@ -0,0 +1,139 @@
+"""按 fin_product 为指定客户生成 fin_holdings 持仓数据（本地开发/演示用）。
+
+用法：
+    python scripts/seed_holdings_for_customer.py --customer-id 18
+    python scripts/seed_holdings_for_customer.py --customer-id 18 --count 5 --force
+
+规则：
+- 只挑选 fin_product 中 status=在售 的产品，按 id 升序取前 count 只；
+- 每笔持仓的买入净值 = 当前净值 × (1 - 浮动)，浮动由固定 seed 生成，保证可复现；
+- shares = cost_amount / 买入净值（4 位小数），current_value = shares × 当前净值；
+- profit_loss / profit_ratio 由上述字段推导；
+- 客户已有持仓时默认拒绝，需 --force 才会追加。
+"""
+from __future__ import annotations
+
+import argparse
+import asyncio
+import random
+import sys
+from decimal import Decimal, ROUND_HALF_UP
+from pathlib import Path
+
+sys.path.insert(0, str(Path(__file__).resolve().parents[1]))
+
+from sqlalchemy import select
+
+from config.database.mysql import get_session_factory
+from model.fin_holdings import FinHoldings
+from model.fin_product import FinProduct
+from model.sys_user import SysUser
+
+_MONEY = Decimal("0.01")
+_SHARES = Decimal("0.0001")
+_RATIO = Decimal("0.0001")
+
+
+def _money(value: Decimal) -> Decimal:
+    return value.quantize(_MONEY, rounding=ROUND_HALF_UP)
+
+
+async def seed(customer_id: int, *, count: int, force: bool) -> list[dict]:
+    session_factory = get_session_factory()
+    async with session_factory() as db:
+        user = await db.get(SysUser, customer_id)
+        if user is None or user.user_type != "CUSTOMER":
+            raise SystemExit(f"customer_id={customer_id} 不是客户用户或不存在")
+
+        existing = (
+            await db.execute(
+                select(FinHoldings).where(FinHoldings.customer_id == customer_id)
+            )
+        ).scalars().all()
+        if existing and not force:
+            raise SystemExit(
+                f"客户 {customer_id} 已有 {len(existing)} 条持仓，如需追加请加 --force"
+            )
+
+        products = (
+            await db.execute(
+                select(FinProduct)
+                .where(FinProduct.status == "在售")
+                .order_by(FinProduct.id)
+                .limit(count)
+            )
+        ).scalars().all()
+        if not products:
+            raise SystemExit("fin_product 中没有在售产品")
+
+        rng = random.Random(f"holdings-{customer_id}")
+        created: list[dict] = []
+        for product in products:
+            nav = product.nav or Decimal("1.000000")
+            cost_amount = _money(Decimal(rng.randint(5_000, 50_000)))
+            # 买入净值在当前净值的 88%~97% 之间浮动，形成有涨有跌的持仓
+            buy_nav = _money(nav * Decimal(str(round(rng.uniform(0.88, 0.97), 6))))
+            if buy_nav <= 0:
+                buy_nav = Decimal("1.0000")
+            shares = (cost_amount / buy_nav).quantize(_SHARES, rounding=ROUND_HALF_UP)
+            current_value = _money(shares * nav)
+            profit_loss = _money(current_value - cost_amount)
+            profit_ratio = (profit_loss / cost_amount).quantize(_RATIO, rounding=ROUND_HALF_UP)
+
+            db.add(
+                FinHoldings(
+                    customer_id=customer_id,
+                    product_id=product.id,
+                    shares=shares,
+                    cost_amount=cost_amount,
+                    current_value=current_value,
+                    profit_loss=profit_loss,
+                    profit_ratio=profit_ratio,
+                    status="持有中",
+                )
+            )
+            created.append(
+                {
+                    "product_id": product.id,
+                    "product_name": product.product_name,
+                    "buy_nav": str(buy_nav),
+                    "nav": str(nav),
+                    "shares": str(shares),
+                    "cost_amount": str(cost_amount),
+                    "current_value": str(current_value),
+                    "profit_loss": str(profit_loss),
+                    "profit_ratio": str(profit_ratio),
+                }
+            )
+
+        await db.commit()
+        return created
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(description="按 fin_product 生成客户持仓数据")
+    parser.add_argument("--customer-id", type=int, required=True)
+    parser.add_argument("--count", type=int, default=8, help="生成持仓条数（默认 8）")
+    parser.add_argument(
+        "--force", action="store_true", help="客户已有持仓时仍允许追加"
+    )
+    args = parser.parse_args()
+
+    created = asyncio.run(seed(args.customer_id, count=args.count, force=args.force))
+    total_cost = sum(Decimal(item["cost_amount"]) for item in created)
+    total_value = sum(Decimal(item["current_value"]) for item in created)
+    print(f"客户 {args.customer_id} 新增持仓 {len(created)} 条：")
+    for item in created:
+        print(
+            f"  [{item['product_id']}] {item['product_name'][:24]}… "
+            f"买入净值={item['buy_nav']} 当前净值={item['nav']} "
+            f"份额={item['shares']} 成本={item['cost_amount']} "
+            f"市值={item['current_value']} 盈亏={item['profit_loss']} "
+            f"({item['profit_ratio']})"
+        )
+    print(f"合计：成本 {total_cost} 元，市值 {total_value} 元，"
+          f"盈亏 {total_value - total_cost} 元")
+
+
+if __name__ == "__main__":
+    main()
@@ -1,7 +1,18 @@
-"""读取当前 MySQL 元数据并同步到 Milvus。"""
+"""同步 NL2SQL 元数据到 Milvus（表名单以 model/ 目录的 ORM 为主）。
+
+主数据源策略：
+- 表名单 = model/ 下 SQLAlchemy ORM 定义的全部表（Base.metadata），
+  不再与数据库求交集——数据库尚未建表的 ORM 表也纳入元数据，
+  其表/字段信息从 ORM 定义合成，并在表说明标注"尚未建表"；
+- 数据库里真实存在的 ORM 表，表/列中文注释仍取自 information_schema
+  （列注释只存在于库中，ORM 代码里没有字段级注释）；
+- 名单之外（库里有、model/ 没定义）的表元数据 chunk 会被从 Milvus 删除。
+"""
 from __future__ import annotations

 import asyncio
+import importlib
+import pkgutil
 import sys
 from pathlib import Path

@@ -14,7 +25,8 @@ from config.database import mysql
 from config.database.milvus import client as milvus_client
 from config.database.mysql import get_session_factory
 from config.settings import settings
-from nl2sql.metadata_sync import sync_metadata
+from model.base import Base
+from nl2sql.metadata_sync import merge_orm_metadata_rows, sync_metadata


 TABLES_SQL = text(
@@ -36,6 +48,15 @@ COLUMNS_SQL = text(
 )


+def collect_orm_tables() -> set[str]:
+    """导入 model/ 全部模块，从 Base.metadata 收集 ORM 表名。"""
+    import model
+
+    for module_info in pkgutil.iter_modules(model.__path__):
+        importlib.import_module(f"model.{module_info.name}")
+    return set(Base.metadata.tables.keys())
+
+
 async def load_information_schema() -> tuple[list[dict], list[dict]]:
    async with get_session_factory()() as session:
        tables = [
@@ -49,14 +70,43 @@ async def load_information_schema() -> tuple[list[dict], list[dict]]:
    return tables, columns


-async def synchronize() -> int:
+def _table_name(row: dict) -> str:
+    return str(row.get("TABLE_NAME") or row.get("table_name") or "").strip()
+
+
+async def synchronize() -> tuple[int, set[str], set[str], set[str]]:
+    """返回 (upserted, allowed_tables, 名单外表, 未建表的 ORM 表)。"""
+    orm_tables = collect_orm_tables()
    try:
        tables, columns = await load_information_schema()
-        return await sync_metadata(milvus_client(), tables, columns)
+        db_tables = {_table_name(row) for row in tables}
+        missing_tables = sorted(orm_tables - db_tables)
+        table_rows, column_rows = merge_orm_metadata_rows(
+            tables,
+            columns,
+            allowed_tables=orm_tables,
+            missing_table_objects=(
+                Base.metadata.tables[name] for name in missing_tables
+            ),
+        )
+        upserted = await sync_metadata(
+            milvus_client(),
+            table_rows,
+            column_rows,
+            allowed_tables=orm_tables,
+        )
+        dropped = db_tables - orm_tables
+        return upserted, orm_tables, dropped, set(missing_tables)
    finally:
        await mysql.dispose()
        await milvus_db.dispose()


 if __name__ == "__main__":
-    print(f"upserted {asyncio.run(synchronize())} NL2SQL metadata chunks")
+    upserted, allowed, dropped, missing = asyncio.run(synchronize())
+    print(f"ORM 表名单（model/ 全量）: {len(allowed)} 张")
+    print(f"其中数据库尚未建表（用 ORM 定义合成元数据）: {len(missing)} 张")
+    if missing:
+        print(f"  {sorted(missing)}")
+    print(f"已排除的非 ORM 表: {sorted(dropped) if dropped else '无'}")
+    print(f"upserted {upserted} NL2SQL metadata chunks")