mirror of
https://github.com/shy3130/tick-stock-panel.git
synced 2026-11-02 10:26:13 +00:00
shy3130
d43a3021da
- 日K/分时批量单请求上限改为系统设置可配 (默认 400), 后端守卫与前端分块 读同一偏好值, 不再发出必然被拒的请求; 两档独立 (分时成本约为日K 20 倍) - _gzip_payload 懒净化: 无 NaN/inf 常态跳过 _json_safe 整树深拷贝, 峰值 -65% (40→14MB/块), CPU -25% - minute-batch/daily-batch 全程持有 Polars 帧直出 JSON (write_json 行式), 不再构建 Python dict: 96k 根序列化 1222ms→26ms, 峰值 69→18MB; 字节级等价由 tests/test_kline_batch_frames_json.py 锁定 (含 inf/NaN/None) - 白盒测试改为解析 Response body (HTTP 真实契约) - 随 kline.py 同文件: minute-batch 门控在全量分钟服务健康时放行本地读, 补拉限速改走 resolve_limit (×0.8, 与盘后同步同 rpm 桶) 实测: 300 只分时(7.2万根)整请求 191ms; 20 并发尖峰 +728MB→+60MB Signed-off-by: shy3130 <shy3130@qq.com>
109 lines
4.3 KiB
Python
109 lines
4.3 KiB
Python
"""批量端点 Polars 直出 JSON 的字节级等价契约。
|
|
|
|
_batch_frames_json 取代 to_dicts()+json.dumps (对象图与序列化耗时一起消失,
|
|
见 kline.py 该函数 docstring 的实测数字)。直出的任何格式漂移 —— 键序、浮点
|
|
表示、日期格式、NaN/inf 语义、extra 键拼接 —— 都会破坏前端与已冻结扩展
|
|
bundle 的解析 (前端 since 增量还按 datetime 字符串字典序比较), 用字节级
|
|
断言锁定与旧 dict 路径完全一致。
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import gzip
|
|
import json
|
|
from datetime import date, datetime, timedelta
|
|
from types import SimpleNamespace
|
|
|
|
import polars as pl
|
|
|
|
from app.api.kline import _batch_frames_json, _gzip_raw, _json_safe
|
|
|
|
|
|
def _minute_frame(sym: str) -> pl.DataFrame:
|
|
"""真实分时列集合; 含典型价格浮点、inf/NaN/None 三种非有限与缺失形态。"""
|
|
start = datetime(2026, 10, 23, 9, 31)
|
|
return pl.DataFrame({
|
|
"symbol": [sym] * 5,
|
|
"datetime": [start + timedelta(minutes=i) for i in range(5)],
|
|
"open": [10.5, 3.33, 0.01, 9999.99, 1.5],
|
|
"high": [10.6, 3.34, 0.02, 10000.0, float("inf")],
|
|
"low": [10.4, 3.32, 0.005, 9990.0, float("nan")],
|
|
"close": [10.55, 3.335, 0.015, 9995.55, 2.5],
|
|
"volume": [123456, 0, 999999999, 1, 42],
|
|
"amount": [1.5e8, 0.0, 2.5e6, 9999.5, None],
|
|
})
|
|
|
|
|
|
def test_frames_json_byte_identical_to_dict_path():
|
|
"""直出与 to_dicts()+dumps (与 _gzip_payload 同参 + _json_safe 净化) 逐字节一致。"""
|
|
parts = {
|
|
"600000.SH": _minute_frame("600000.SH"),
|
|
"000001.SZ": _minute_frame("000001.SZ"),
|
|
}
|
|
new = _batch_frames_json(parts, "datetime", "%Y-%m-%dT%H:%M:%S")
|
|
|
|
def _iso(o):
|
|
return o.isoformat() if hasattr(o, "isoformat") else str(o)
|
|
|
|
old = json.dumps(
|
|
_json_safe({"data": {s: sub.to_dicts() for s, sub in parts.items()}}),
|
|
ensure_ascii=False, separators=(",", ":"), allow_nan=False, default=_iso,
|
|
).encode()
|
|
assert new == old
|
|
|
|
|
|
def test_frames_json_extra_keys_spliced_into_root():
|
|
"""extra 键并入根对象 (而非拼成兄弟对象), 键序与旧 dict 路径一致。"""
|
|
raw = _batch_frames_json(
|
|
{"600000.SH": _minute_frame("600000.SH")},
|
|
"datetime", "%Y-%m-%dT%H:%M:%S",
|
|
{"full_minute_local": True, "incremental": True, "truncated": ["600000.SH"]},
|
|
)
|
|
payload = json.loads(raw)
|
|
assert payload["full_minute_local"] is True
|
|
assert payload["incremental"] is True
|
|
assert payload["truncated"] == ["600000.SH"]
|
|
assert list(payload) == ["data", "full_minute_local", "incremental", "truncated"]
|
|
|
|
|
|
def test_frames_json_daily_date_format():
|
|
"""日K的 date 列输出 ISO 日期串, 与 date.isoformat 语义一致。"""
|
|
df = pl.DataFrame({
|
|
"symbol": ["000001.SZ"] * 2,
|
|
"date": [date(2026, 10, 22), date(2026, 10, 23)],
|
|
"open": [1.0, 2.0], "high": [1.1, 2.1], "low": [0.9, 1.9],
|
|
"close": [1.05, 2.05], "volume": [100, 200],
|
|
})
|
|
payload = json.loads(_batch_frames_json({"000001.SZ": df}, "date", "%Y-%m-%d"))
|
|
assert [r["date"] for r in payload["data"]["000001.SZ"]] == ["2026-10-22", "2026-10-23"]
|
|
|
|
|
|
def test_gzip_raw_negotiation_and_threshold(monkeypatch):
|
|
"""_gzip_raw 与 _gzip_payload 同规则: 偏好 / 客户端协商 / 1024 阈值。"""
|
|
from app.services import preferences
|
|
|
|
monkeypatch.setattr(preferences, "get_minute_batch_compress", lambda: True)
|
|
monkeypatch.setattr(preferences, "get_daily_batch_compress", lambda: False)
|
|
|
|
big = b'{"data":{"x":"' + b"y" * 2048 + b'"}}'
|
|
req_gz = SimpleNamespace(headers={"accept-encoding": "gzip"})
|
|
req_plain = SimpleNamespace(headers={})
|
|
|
|
resp = _gzip_raw(req_gz, big, pref_key="minute_batch_compress")
|
|
assert resp.headers["content-encoding"] == "gzip"
|
|
assert gzip.decompress(resp.body) == big
|
|
|
|
# 客户端不接受 gzip → 原样 (尊重协商)
|
|
resp = _gzip_raw(req_plain, big, pref_key="minute_batch_compress")
|
|
assert "content-encoding" not in resp.headers
|
|
assert resp.body == big
|
|
|
|
# 偏好关 → 原样
|
|
resp = _gzip_raw(req_gz, big, pref_key="daily_batch_compress")
|
|
assert "content-encoding" not in resp.headers
|
|
|
|
# 未超 1024 阈值 → 原样
|
|
resp = _gzip_raw(req_gz, b'{"data":{}}', pref_key="minute_batch_compress")
|
|
assert "content-encoding" not in resp.headers
|
|
assert resp.body == b'{"data":{}}'
|