Profeto/tests/test_prediction_unique_constraint.py

"""回归测试: 预测唯一约束修复 —— live 与 backtest 可共存。

验证:
  1. 唯一约束包含 mode + run_type
  2. 同一场比赛 live 与 backtest 预测可共存,互不覆盖
  3. _insert_or_find_by_fingerprint 正确区分 run_type
"""
from __future__ import annotations

import inspect
from pathlib import Path

from pydantic import BaseModel
import pytest
from sqlalchemy import Index

from src.db.models import Prediction, UniqueConstraint, CheckConstraint

# 仓库根目录下的 alembic 迁移目录 —— 相对本测试文件解析,
# 避免硬编码某台机器/CI 上的绝对路径(见 tests/test_regressions.py 的 _read 约定)。
REPO_ROOT = Path(__file__).resolve().parent.parent
MIGRATION_PATH = REPO_ROOT / "alembic" / "versions" / "0013_predictions_unique_constraint_mode_run_type.py"


class TestUniqueConstraint:
    """P0-03: 验证幂等指纹唯一索引(替代旧 (match, provider, model, mode, run_type) 唯一约束)。"""

    def test_input_hash_partial_unique_index(self):
        """P0-03: input_hash 非空时必须唯一(同指纹 → 返回已有行,不 UPDATE/INSERT)。"""
        idx = [
            i for i in Prediction.__table__.indexes
            if i.unique and "input_hash" in i.name
        ]
        assert len(idx) == 1, f"缺少 input_hash partial unique 索引,现有 indexes: {[i.name for i in Prediction.__table__.indexes]}"
        # partial unique: postgresql_where 必须限制 input_hash IS NOT NULL
        assert idx[0].dialect_kwargs.get("postgresql_where") is not None

    def test_old_unique_constraint_removed(self):
        """P0-03: 旧 (match, provider, model, mode, run_type) 唯一约束必须已移除。"""
        from sqlalchemy import UniqueConstraint

        old = [
            c for c in Prediction.__table__.constraints
            if isinstance(c, UniqueConstraint) and c.name == "uq_predictions_match_provider_model_mode_run_type"
        ]
        assert len(old) == 0, f"旧约束必须已移除,但仍存在: {[c.name for c in old]}"

    def test_run_type_check_constraint(self):
        """应有 run_type 的 check constraint。"""
        from sqlalchemy import CheckConstraint

        cc = [
            c for c in Prediction.__table__.constraints
            if isinstance(c, CheckConstraint) and "run_type" in c.name
        ]
        assert len(cc) == 1

    def test_run_type_column_exists(self):
        """run_type 列应存在且 NOT NULL,默认 'live'。"""
        cols = {c.name: c for c in Prediction.__table__.columns}
        assert "run_type" in cols
        assert cols["run_type"].nullable is False
        # 默认值
        assert cols["run_type"].default.arg == "live" if cols["run_type"].default else True


class TestUpsertPredictionSignature:
    """验证 _insert_or_find_by_fingerprint 签名(P0-03 指纹模式)。"""

    def test_signature_uses_values_dict(self):
        """P0-03: 新接口通过 values dict 接收全部字段(含 run_type/match_id/...)。"""
        from src.llm.predict import _insert_or_find_by_fingerprint

        sig = inspect.signature(_insert_or_find_by_fingerprint)
        params = sig.parameters
        assert "session" in params
        assert "values" in params  # 所有业务字段走 values dict

    def test_signature_has_backtest_in_predict_match(self):
        from src.llm.predict import predict_match

        sig = inspect.signature(predict_match)
        assert "backtest" in sig.parameters

    def test_signature_has_backtest_in_predict_multi(self):
        from src.llm.agents.orchestrator import predict_match_multi

        sig = inspect.signature(predict_match_multi)
        assert "backtest" in sig.parameters


class TestMigration:
    """验证迁移文件存在且内容正确。"""

    def test_migration_exists(self):
        assert MIGRATION_PATH.is_file(), f"迁移文件不存在: {MIGRATION_PATH}"

    def test_migration_adds_column_and_constraint(self):
        content = MIGRATION_PATH.read_text(encoding="utf-8")

        assert 'run_type' in content
        assert 'uq_predictions_match_provider_model_mode_run_type' in content
        assert 'backtest' in content
        assert 'live' in content
        # 验证数据回填逻辑
        assert "UPDATE predictions SET run_type = 'live'" in content


class TestLiveBacktestCoexist:
    """验证 live 与 backtest 可共存(逻辑验证,无需数据库)。"""

    def test_different_run_type_allow_coexistence(self):
        """同一 match_id + provider + model + mode,不同 run_type 应可共存。

        这是核心修复:之前唯一约束只有 (match_id, provider, model),
        backtest 会覆盖 live 预测。
        """
        # 模拟两行数据
        class FakeRow:
            def __init__(self, **kw):
                for k, v in kw.items():
                    setattr(self, k, v)

        live = FakeRow(match_id=1, provider="openai", model="gpt-4o", mode="single", run_type="live")
        backtest = FakeRow(match_id=1, provider="openai", model="gpt-4o", mode="single", run_type="backtest")

        # 两者唯一键不同(因为 run_type 不同)
        live_key = (live.match_id, live.provider, live.model, live.mode, live.run_type)
        backtest_key = (backtest.match_id, backtest.provider, backtest.model, backtest.mode, backtest.run_type)

        assert live_key != backtest_key, "live 与 backtest 应有不同的唯一键"
        assert live_key == (1, "openai", "gpt-4o", "single", "live")
        assert backtest_key == (1, "openai", "gpt-4o", "single", "backtest")

    def test_same_run_type_prevents_duplicate(self):
        """相同 run_type 的重复预测仍应被约束阻止。"""
        key1 = (1, "openai", "gpt-4o", "single", "live")
        key2 = (1, "openai", "gpt-4o", "single", "live")
        assert key1 == key2, "相同 run_type 应有相同唯一键,应被约束阻止"