Simon
2da234220a
- domain：市场数据实体（Stock / 交易日历 / 日线 / 复权 / 财务含 announce_date）+ Repository 与 MarketDataProvider Protocol
- 数据源：TushareProvider（归一化、重试、鉴权错误归类）、SinaProvider（备用，明确前复权口径与能力边界）、FailoverProvider + SyncLog 审计（禁止静默切换）
- 持久化：SQLAlchemy 2.x Models + Repository 实现（按业务键幂等 upsert、as_of_date 防未来函数过滤）+ Alembic 迁移
- CLI：uv run python -m app.cli.sync {basic|calendar|daily|financial|verify}，支持 --resume 断点续传
- 真实 Tushare 验证：stock 5556 / 交易日历 366 / daily+factor 242 / 财务 55；sync_log 审计完整
- 测试：38 passed（domain / provider / failover / repository / 未来函数 / 迁移），ruff clean
220 lines
7.5 KiB
Python
220 lines
7.5 KiB
Python
"""Tushare Provider —— 首选数据源实现。
|
||
|
||
依赖注入：pro 客户端（tushare.pro.client 或测试 Fake）。真实运行时惰性加载
|
||
tushare 库（pyproject optional：uv sync --extra datasource-tushare）。
|
||
归一化函数只依赖 list[dict]，便于无 pandas 环境下单测。
|
||
"""
|
||
|
||
from __future__ import annotations
|
||
|
||
import importlib
|
||
from datetime import date, datetime
|
||
from decimal import Decimal
|
||
from typing import Any
|
||
|
||
from app.domain.entities.market import (
|
||
AdjustFactor,
|
||
DailyBar,
|
||
FinancialIndicator,
|
||
Stock,
|
||
TradingCalendar,
|
||
)
|
||
from app.infrastructure.data_sources.errors import (
|
||
DataSourceAuthenticationError,
|
||
DataSourceError,
|
||
)
|
||
|
||
_TS_DATE = "%Y%m%d"
|
||
|
||
|
||
def _to_date(value: str | None) -> date | None:
|
||
if value is None or value == "":
|
||
return None
|
||
return datetime.strptime(str(value)[:10], _TS_DATE).date()
|
||
|
||
|
||
def _to_decimal(value) -> Decimal | None:
|
||
if value is None:
|
||
return None
|
||
try:
|
||
num = float(value)
|
||
except (ValueError, TypeError):
|
||
return None
|
||
if num != num: # NaN
|
||
return None
|
||
return Decimal(str(num))
|
||
|
||
|
||
class TushareProvider:
|
||
"""封装 Tushare Pro（ts.pro_api）。所有输出已归一化为领域实体。"""
|
||
|
||
name = "tushare"
|
||
|
||
def __init__(
|
||
self,
|
||
token: str = "",
|
||
*,
|
||
pro: object | None = None,
|
||
max_retries: int = 3,
|
||
) -> None:
|
||
self._pro = pro if pro is not None else _build_pro(token)
|
||
self._max_retries = max_retries
|
||
|
||
# ---- 归一化（纯函数，输入 list[dict]，可单测） ----
|
||
|
||
@staticmethod
|
||
def normalize_stock(records: list[dict[str, Any]]) -> list[Stock]:
|
||
stocks: list[Stock] = []
|
||
for rec in records:
|
||
stocks.append(
|
||
Stock(
|
||
symbol=str(rec.get("ts_code") or rec.get("symbol") or ""),
|
||
name=str(rec.get("name") or ""),
|
||
industry=rec.get("industry"),
|
||
area=rec.get("area"),
|
||
market=rec.get("market"),
|
||
exchange=rec.get("exchange"),
|
||
list_date=_to_date(rec.get("list_date")) or date.min,
|
||
delist_date=_to_date(rec.get("delist_date")),
|
||
status=str(rec.get("status") or "L"),
|
||
)
|
||
)
|
||
return stocks
|
||
|
||
@staticmethod
|
||
def normalize_calendar(records: list[dict[str, Any]]) -> list[TradingCalendar]:
|
||
return [
|
||
TradingCalendar(
|
||
calendar_date=_to_date(rec.get("cal_date")) or date.min,
|
||
is_open=bool(rec.get("is_open")),
|
||
)
|
||
for rec in records
|
||
]
|
||
|
||
@staticmethod
|
||
def normalize_daily(records: list[dict[str, Any]]) -> list[DailyBar]:
|
||
bars: list[DailyBar] = []
|
||
for rec in records:
|
||
vol = _to_decimal(rec.get("vol"))
|
||
amount = _to_decimal(rec.get("amount"))
|
||
bars.append(
|
||
DailyBar(
|
||
symbol=str(rec.get("ts_code") or ""),
|
||
trade_date=_to_date(rec.get("trade_date")) or date.min,
|
||
open=_to_decimal(rec.get("open")),
|
||
high=_to_decimal(rec.get("high")),
|
||
low=_to_decimal(rec.get("low")),
|
||
close=_to_decimal(rec.get("close")),
|
||
volume=vol * 100 if vol is not None else None,
|
||
amount=amount * 1000 if amount is not None else None,
|
||
)
|
||
)
|
||
return bars
|
||
|
||
@staticmethod
|
||
def normalize_adj_factor(records: list[dict[str, Any]]) -> list[AdjustFactor]:
|
||
return [
|
||
AdjustFactor(
|
||
symbol=str(rec.get("ts_code") or ""),
|
||
trade_date=_to_date(rec.get("trade_date")) or date.min,
|
||
factor=_to_decimal(rec.get("adj_factor")) or Decimal(1),
|
||
)
|
||
for rec in records
|
||
]
|
||
|
||
@staticmethod
|
||
def normalize_financial(records: list[dict[str, Any]]) -> list[FinancialIndicator]:
|
||
rows: list[FinancialIndicator] = []
|
||
for rec in records:
|
||
rows.append(
|
||
FinancialIndicator(
|
||
symbol=str(rec.get("ts_code") or ""),
|
||
report_date=_to_date(rec.get("end_date")) or date.min,
|
||
announce_date=_to_date(rec.get("ann_date")) or date.min,
|
||
eps=_to_decimal(rec.get("eps")),
|
||
roe=_to_decimal(rec.get("roe")),
|
||
net_profit=_to_decimal(rec.get("n_income_attr_p")),
|
||
gross_margin=_to_decimal(rec.get("grossprofit_margin")),
|
||
)
|
||
)
|
||
return rows
|
||
|
||
# ---- 接口调用 ----
|
||
|
||
def get_stock_basic(self) -> list[Stock]:
|
||
records = self._call(
|
||
"stock_basic",
|
||
fields="ts_code,symbol,name,area,industry,market,exchange,list_date,delist_date,status",
|
||
)
|
||
return self.normalize_stock(records)
|
||
|
||
def get_trade_cal(self, start: date, end: date) -> list[TradingCalendar]:
|
||
records = self._call(
|
||
"trade_cal",
|
||
exchange="SSE",
|
||
start_date=start.strftime(_TS_DATE),
|
||
end_date=end.strftime(_TS_DATE),
|
||
is_open="",
|
||
)
|
||
return self.normalize_calendar(records)
|
||
|
||
def get_daily(self, symbol: str, start: date, end: date) -> list[DailyBar]:
|
||
records = self._call(
|
||
"daily",
|
||
ts_code=symbol,
|
||
start_date=start.strftime(_TS_DATE),
|
||
end_date=end.strftime(_TS_DATE),
|
||
)
|
||
return self.normalize_daily(records)
|
||
|
||
def get_adjust_factor(self, symbol: str, start: date, end: date) -> list[AdjustFactor]:
|
||
records = self._call(
|
||
"adj_factor",
|
||
ts_code=symbol,
|
||
start_date=start.strftime(_TS_DATE),
|
||
end_date=end.strftime(_TS_DATE),
|
||
)
|
||
return self.normalize_adj_factor(records)
|
||
|
||
def get_financial(self, symbol: str) -> list[FinancialIndicator]:
|
||
records = self._call("fina_indicator", ts_code=symbol)
|
||
return self.normalize_financial(records)
|
||
|
||
# ---- 内部 ----
|
||
|
||
def _call(self, api: str, **kwargs) -> list[dict[str, Any]]:
|
||
last_error: Exception | None = None
|
||
for _ in range(self._max_retries):
|
||
try:
|
||
fn = getattr(self._pro, api)
|
||
result = fn(**kwargs)
|
||
if result is None:
|
||
return []
|
||
if hasattr(result, "to_dict"):
|
||
return result.to_dict("records")
|
||
if isinstance(result, list):
|
||
return result
|
||
return []
|
||
except Exception as exc: # noqa: BLE001 —— tushare 异常无统一类型，逐一归类
|
||
last_error = exc
|
||
msg = str(exc)
|
||
if "权限" in msg or "积分" in msg or "token" in msg.lower():
|
||
raise DataSourceAuthenticationError(msg) from exc
|
||
raise DataSourceError(
|
||
f"tushare.{api} 重试 {self._max_retries} 次仍失败: {last_error}"
|
||
) from last_error
|
||
|
||
|
||
def _build_pro(token: str):
|
||
if not token:
|
||
raise DataSourceAuthenticationError(
|
||
"缺少 TUSHARE_TOKEN：请 cp .env.example .env 并填入 Tushare Pro token"
|
||
)
|
||
try:
|
||
ts = importlib.import_module("tushare")
|
||
except ImportError as exc: # pragma: no cover —— 环境相关
|
||
raise DataSourceError(
|
||
"未安装 tushare 客户端：cd backend && uv sync --extra datasource-tushare"
|
||
) from exc
|
||
return ts.pro_api(token)
|