数据源统一为 bzzoiro,移除 Understat 与 injuries:
- 删除 src/data/understat.py / injuries.py 及相关测试
- 删除 injuries 模型与表;扩展 match_stats(xG 之外增加 big_chances/fouls)
- 新增 standings 表(联赛积分榜:位置/积分/xG/走势/分区)
- matches 表增加 source_event_id 血缘列,支撑统计回填
采集管线(bzzoiro 三条管线):
- events:赛程/比分(/events/),记录 source_event_id
- standings:积分榜快照(/leagues/{id}/standings/)
- stats:已完赛比赛详细统计回填(/events/{id}/stats/)
预测增强:
- standings_slice 替代 injuries_slice;积分榜专家替代阵容完整性专家
- AGENT_META runtime_config 同步更新
管理后台:
- ingest 路由重写为单一 bzzoiro 入口 + task 参数(events/standings/stats/all)
- 新增 /admin/data-completeness 数据完整性分析 API
- 数据源状态页简化为 bzzoiro 单源
前端:
- 采集页重构为任务驱动(比赛/积分榜/统计回填/全量)
- 新增「数据完整性」可视化页(覆盖率矩阵/字段完整率/健康摘要)
- 新增主站积分榜页(/standings)与比赛详情完整统计面板
- agent 名称同步更新(injuries→standings)
迁移 0015_bzzoiro_single_source 已在容器内验证通过,后端测试全部通过。
Co-Authored-By: new-provider/LongCat-2.0 <<EMAIL>>
128 lines
5.4 KiB
Python
128 lines
5.4 KiB
Python
"""回归测试: match_stats.available_at 回测防泄漏语义。
|
|
|
|
验证:
|
|
1. _is_stats_available: available_at is None + cutoff 不为 None → 不可用
|
|
2. _is_stats_available: available_at is None + cutoff is None(实盘) → 可用(兼容旧数据)
|
|
3. _is_stats_available: available_at > cutoff → 不可用
|
|
4. _is_stats_available: available_at <= cutoff → 可用
|
|
5. 写入策略: bzzoiro 写入 available_at 使用 match_date + 2h 缓冲
|
|
6. cutoff 在缓冲内时不可用(available_at > cutoff → False)
|
|
7. stats 回填(bzzoiro event stats)也使用 2h 缓冲
|
|
"""
|
|
from __future__ import annotations
|
|
|
|
from datetime import datetime, timezone, timedelta
|
|
from unittest.mock import MagicMock
|
|
|
|
import pytest
|
|
|
|
from src.llm.context_builder import _is_stats_available
|
|
|
|
|
|
def _make_stats(available_at):
|
|
s = MagicMock()
|
|
s.available_at = available_at
|
|
return s
|
|
|
|
|
|
class TestIsStatsAvailable:
|
|
"""_is_stats_available 回测防泄漏语义。"""
|
|
|
|
def test_none_before_allows_none_available_at(self):
|
|
"""实盘模式(before=None): available_at 为 None 时允许(兼容旧数据)。"""
|
|
stats = _make_stats(available_at=None)
|
|
assert _is_stats_available(stats, before=None) is True
|
|
|
|
def test_cutoff_with_none_available_at_is_unavailable(self):
|
|
"""回测模式(before=cutoff): available_at 为 None → 不可用(保守)。
|
|
|
|
这是核心修复:防止无血缘时间的后期回填数据进入回测。
|
|
"""
|
|
stats = _make_stats(available_at=None)
|
|
cutoff = datetime(2026, 1, 13, 20, 0, tzinfo=timezone.utc)
|
|
assert _is_stats_available(stats, before=cutoff) is False
|
|
|
|
def test_available_at_after_cutoff_is_unavailable(self):
|
|
"""available_at 在 cutoff 之后 → 不可用。"""
|
|
cutoff = datetime(2026, 1, 13, 20, 0, tzinfo=timezone.utc)
|
|
stats = _make_stats(available_at=cutoff + timedelta(hours=1))
|
|
assert _is_stats_available(stats, before=cutoff) is False
|
|
|
|
def test_available_at_before_cutoff_is_available(self):
|
|
"""available_at 在 cutoff 之前 → 可用。"""
|
|
cutoff = datetime(2026, 1, 13, 20, 0, tzinfo=timezone.utc)
|
|
stats = _make_stats(available_at=cutoff - timedelta(hours=1))
|
|
assert _is_stats_available(stats, before=cutoff) is True
|
|
|
|
def test_available_at_equals_cutoff_is_available(self):
|
|
"""available_at == cutoff → 可用(边界包含)。"""
|
|
cutoff = datetime(2026, 1, 13, 20, 0, tzinfo=timezone.utc)
|
|
stats = _make_stats(available_at=cutoff)
|
|
assert _is_stats_available(stats, before=cutoff) is True
|
|
|
|
def test_real_world_scenario_backtest_avoids_future_data(self):
|
|
"""真实场景:回测时,赛后才生成的统计数据不应出现。
|
|
|
|
比赛:2026-01-15 20:00
|
|
cutoff(回测):2026-01-14 20:00(赛前 1 天)
|
|
stats 在赛后才生成(available_at=2026-01-15 22:00)
|
|
→ 不可用
|
|
"""
|
|
cutoff = datetime(2026, 1, 14, 20, 0, tzinfo=timezone.utc)
|
|
stats = _make_stats(available_at=datetime(2026, 1, 15, 22, 0, tzinfo=timezone.utc))
|
|
assert _is_stats_available(stats, before=cutoff) is False
|
|
|
|
|
|
class TestWriteBufferStrategy:
|
|
"""验证 bzzoiro(events + stats 回填)写入 available_at 使用 match_date + 2h 缓冲。"""
|
|
|
|
def test_bzzoirot_new_match_available_at_is_two_hours_after_kickoff(self):
|
|
"""bzzoiro 新建比赛(stats 回填创建 MatchStats)时 available_at 应为开球 + 2 小时。"""
|
|
import inspect
|
|
from src.data import bzzoiro
|
|
|
|
source = inspect.getsource(bzzoiro)
|
|
assert 'timedelta(hours=2)' in source, \
|
|
"bzzoiro 应使用 match_date + timedelta(hours=2) 作为 available_at"
|
|
|
|
def test_bzzoirot_multiple_writes_use_two_hour_buffer(self):
|
|
"""bzzoiro 多处写入(创建/更新)都应使用 2 小时缓冲。"""
|
|
import inspect
|
|
from src.data import bzzoiro
|
|
|
|
source = inspect.getsource(bzzoiro)
|
|
count = source.count('timedelta(hours=2)')
|
|
assert count >= 2, f"期望至少 2 处 timedelta(hours=2),实际 {count} 处"
|
|
|
|
def test_cutoff_within_buffer_makes_stats_unavailable(self):
|
|
"""cutoff 在 2 小时缓冲内时,统计学不可用(回测防泄漏)。
|
|
|
|
开球:2026-01-15 20:00
|
|
available_at:2026-01-15 22:00(开球 + 2h)
|
|
cutoff:2026-01-15 21:00(开赛后 1h, statistics 尚未可用)
|
|
→ 不可用
|
|
"""
|
|
kickoff = datetime(2026, 1, 15, 20, 0, tzinfo=timezone.utc)
|
|
available_at = kickoff + timedelta(hours=2) # 22:00
|
|
cutoff = kickoff + timedelta(hours=1) # 21:00,在缓冲内
|
|
|
|
stats = _make_stats(available_at=available_at)
|
|
assert _is_stats_available(stats, before=cutoff) is False, \
|
|
"cutoff 在 2h 缓冲内时应不可用(available_at > cutoff)"
|
|
|
|
def test_cutoff_after_buffer_makes_stats_available(self):
|
|
"""cutoff 超过 2 小时缓冲后,统计变为可用。
|
|
|
|
开球:2026-01-15 20:00
|
|
available_at:2026-01-15 22:00(开球 + 2h)
|
|
cutoff:2026-01-16 20:00(开赛后 1 天,超过缓冲)
|
|
→ 可用
|
|
"""
|
|
kickoff = datetime(2026, 1, 15, 20, 0, tzinfo=timezone.utc)
|
|
available_at = kickoff + timedelta(hours=2) # 22:00
|
|
cutoff = kickoff + timedelta(days=1) # 2026-01-16 20:00
|
|
|
|
stats = _make_stats(available_at=available_at)
|
|
assert _is_stats_available(stats, before=cutoff) is True, \
|
|
"cutoff 超过 2h 缓冲后应可用(available_at < cutoff)"
|