mirror of
https://github.com/shy3130/tick-stock-panel.git
synced 2026-11-02 10:26:13 +00:00
shy3130
eab70c402f
部分行情源的 get_instruments 只返回代码/名称/市场/类型, 没有股本与交易 规则字段。维表同步整表覆盖后 float_shares/total_shares 变 null, enriched 的 turnover_rate 随之全空, 依赖换手率与流通市值的策略 (起爆、5连阳等) 基础过滤会筛出 0 只 —— 换数据源即策略结果不同。 改为按 symbol 逐行 coalesce: provider 有值用 provider, 为空才回落既有 维表值; 新上市标的保持 null 不伪造; provider 自带字段时以其为准。 provider 完全缺字段且无历史维表时打 warning 提示换手率将为空。 验证: 新增 5 个用例; 真实维表同步后 5580 只仅 3 只新标的为空。
126 lines
4.8 KiB
Python
126 lines
4.8 KiB
Python
"""标的维表同步: provider 不提供股本时不覆盖已有值。
|
|
|
|
回归背景: 部分行情源的 get_instruments 只返回代码/名称/市场/类型, 没有
|
|
股本与交易规则字段。旧实现整表覆盖后 float_shares/total_shares 变 null,
|
|
enriched 的 turnover_rate 随之全空, 依赖换手率与流通市值的策略 (起爆/5连阳)
|
|
直接筛出 0 只 —— 同一个策略换数据源结果就不同。
|
|
"""
|
|
from __future__ import annotations
|
|
|
|
from pathlib import Path
|
|
|
|
import polars as pl
|
|
import pytest
|
|
|
|
from app.services import instrument_sync as isync
|
|
|
|
|
|
def _write_existing(data_dir: Path) -> None:
|
|
out = data_dir / "instruments"
|
|
out.mkdir(parents=True, exist_ok=True)
|
|
pl.DataFrame({
|
|
"symbol": ["600000.SH", "000001.SZ"],
|
|
"name": ["浦发银行", "平安银行"],
|
|
"code": ["600000", "000001"],
|
|
"exchange": ["SH", "SZ"],
|
|
"region": ["CN", "CN"],
|
|
"type": ["stock", "stock"],
|
|
"listing_date": ["1999-11-10", "1991-04-03"],
|
|
"total_shares": [2.9e10, 1.9e10],
|
|
"float_shares": [2.9e10, 1.9e10],
|
|
"tick_size": [0.01, 0.01],
|
|
"limit_up": [11.0, 12.0],
|
|
"limit_down": [9.0, 10.0],
|
|
}).write_parquet(out / "instruments.parquet")
|
|
|
|
|
|
def _provider_rows(*symbols: str) -> list[dict]:
|
|
"""模拟只提供标识字段的行情源 (无股本/交易规则)。"""
|
|
return [
|
|
{
|
|
"symbol": s, "name": s, "code": s.split(".")[0],
|
|
"exchange": s.split(".")[1], "region": "CN", "type": "stock",
|
|
"listing_date": None, "total_shares": None, "float_shares": None,
|
|
"tick_size": None, "limit_up": None, "limit_down": None,
|
|
}
|
|
for s in symbols
|
|
]
|
|
|
|
|
|
def test_share_fields_restored_from_existing_table(tmp_path, monkeypatch) -> None:
|
|
_write_existing(tmp_path)
|
|
monkeypatch.setattr(
|
|
isync, "_fetch_instruments_via_provider",
|
|
lambda: _provider_rows("600000.SH", "000001.SZ"),
|
|
)
|
|
|
|
assert isync.sync_instruments(tmp_path) == 2
|
|
|
|
df = pl.read_parquet(tmp_path / "instruments" / "instruments.parquet")
|
|
by_symbol = {r["symbol"]: r for r in df.to_dicts()}
|
|
assert by_symbol["600000.SH"]["float_shares"] == pytest.approx(2.9e10)
|
|
assert by_symbol["600000.SH"]["total_shares"] == pytest.approx(2.9e10)
|
|
assert by_symbol["600000.SH"]["tick_size"] == pytest.approx(0.01)
|
|
assert by_symbol["000001.SZ"]["float_shares"] == pytest.approx(1.9e10)
|
|
|
|
|
|
def test_new_symbol_keeps_null_shares_instead_of_faking(tmp_path, monkeypatch) -> None:
|
|
"""新上市标的没有历史值 → 保持 null, 不伪造股本。"""
|
|
_write_existing(tmp_path)
|
|
monkeypatch.setattr(
|
|
isync, "_fetch_instruments_via_provider",
|
|
lambda: _provider_rows("600000.SH", "301999.SZ"),
|
|
)
|
|
|
|
isync.sync_instruments(tmp_path)
|
|
|
|
df = pl.read_parquet(tmp_path / "instruments" / "instruments.parquet")
|
|
new = df.filter(pl.col("symbol") == "301999.SZ")
|
|
assert new.height == 1
|
|
assert new["float_shares"][0] is None
|
|
assert new["total_shares"][0] is None
|
|
|
|
|
|
def test_provider_fields_win_when_present(tmp_path, monkeypatch) -> None:
|
|
"""provider 自带股本时以 provider 为准 (不复用旧值)。"""
|
|
_write_existing(tmp_path)
|
|
rows = _provider_rows("600000.SH")
|
|
rows[0]["float_shares"] = 1.0e9
|
|
rows[0]["total_shares"] = 2.0e9
|
|
monkeypatch.setattr(isync, "_fetch_instruments_via_provider", lambda: rows)
|
|
|
|
isync.sync_instruments(tmp_path)
|
|
|
|
df = pl.read_parquet(tmp_path / "instruments" / "instruments.parquet")
|
|
assert df["float_shares"][0] == pytest.approx(1.0e9)
|
|
assert df["total_shares"][0] == pytest.approx(2.0e9)
|
|
|
|
|
|
def test_no_existing_table_is_tolerated(tmp_path, monkeypatch) -> None:
|
|
"""首次同步且无历史维表: 不报错, 字段保持 null。"""
|
|
monkeypatch.setattr(
|
|
isync, "_fetch_instruments_via_provider",
|
|
lambda: _provider_rows("600000.SH"),
|
|
)
|
|
|
|
assert isync.sync_instruments(tmp_path) == 1
|
|
|
|
df = pl.read_parquet(tmp_path / "instruments" / "instruments.parquet")
|
|
assert df["float_shares"][0] is None
|
|
|
|
|
|
def test_partial_provider_fields_filled_per_row(tmp_path, monkeypatch) -> None:
|
|
"""provider 只给部分标的股本时, 也要逐行补全 (不能整列跳过)。"""
|
|
_write_existing(tmp_path)
|
|
rows = _provider_rows("600000.SH", "000001.SZ")
|
|
rows[0]["float_shares"] = 5.0e8 # 第一只有值, 第二只仍为空
|
|
rows[0]["total_shares"] = 6.0e8
|
|
monkeypatch.setattr(isync, "_fetch_instruments_via_provider", lambda: rows)
|
|
|
|
isync.sync_instruments(tmp_path)
|
|
|
|
df = pl.read_parquet(tmp_path / "instruments" / "instruments.parquet")
|
|
by_symbol = {r["symbol"]: r for r in df.to_dicts()}
|
|
assert by_symbol["600000.SH"]["float_shares"] == pytest.approx(5.0e8)
|
|
assert by_symbol["000001.SZ"]["float_shares"] == pytest.approx(1.9e10)
|