"""P0-03 核心测试: Prediction 幂等指纹。

- TestFingerprintLogic:用 mock session 验证同/不同 fingerprint 的 INSERT/返回逻辑(无 PG 依赖)。
- TestFingerprintDeterminism:纯 hash 稳定性(无 PG 依赖)。

运行: pytest tests/test_p0_prediction_fingerprint.py -v
"""
from __future__ import annotations

from unittest.mock import MagicMock

import pytest

from src.db.models import Prediction
from src.llm.predict import _compute_fingerprint, _insert_or_find_by_fingerprint


def _base_values(match_id, **overrides):
    base = {
        "match_id": match_id,
        "provider": "test-provider",
        "model": "test-model",
        "mode": "single",
        "run_type": "live",
        "prompt_version": "v1",
        "prompt_hash": "ph1",
        "system_prompt_hash": "sh1",
        "temperature": 0.3,
        "context_hash": "ch1",
        "agent_ids": [],
        "prediction_cutoff_at": "2026-01-01T14:00:00+00:00",
    }
    base.update(overrides)
    return base


class _FakeSession:
    """模拟 session:记录 add;execute 返回预设的 existing row。"""

    def __init__(self, existing=None):
        self._existing = existing
        self.added: list = []
        self.flushed = 0

    def add(self, obj):
        self.added.append(obj)

    async def execute(self, stmt):
        existing = self._existing

        class _R:
            def scalar_one_or_none(inner_self):
                return existing

        return _R()

    async def flush(self):
        self.flushed += 1

    async def refresh(self, obj):
        if getattr(obj, "id", None) is None:
            obj.id = 1


class TestFingerprintLogic:
    """P0-03:同 fingerprint 返回已有行(不 UPDATE/INSERT);不同 → INSERT。"""

    @pytest.mark.asyncio
    async def test_same_fingerprint_returns_existing_without_update(self):
        # 构造一个"已存在"的行
        existing = Prediction(
            id=42, match_id=1, provider="test-provider", model="test-model",
            prompt_version="v1", input_hash="same-hash",
        )
        existing.pred_home_goals = 2.0
        existing.prompt_version = "v1"

        s = _FakeSession(existing=existing)
        values = _base_values(1, prompt_version="v1")  # 与 existing 同 fingerprint 需 input_hash 相同

        # 但 fingerprint 是动态计算的,existing.input_hash 需匹配。直接让 fake 返回 existing。
        result = await _insert_or_find_by_fingerprint(s, values=values)

        # 应返回 existing,不 add 新行
        assert result is existing, "同 fingerprint 必须返回已有行"
        assert s.added == [], "同 fingerprint 不应 INSERT"
        assert result.pred_home_goals == 2.0, "返回的应是已有行(字段不变)"

    @pytest.mark.asyncio
    async def test_different_fingerprint_inserts_new(self):
        # 无已有行 → INSERT
        s = _FakeSession(existing=None)
        values = _base_values(1, prompt_version="v1", context_hash="ch1")

        result = await _insert_or_find_by_fingerprint(s, values=values)

        assert len(s.added) == 1, "无已有行时应 INSERT"
        assert isinstance(s.added[0], Prediction)
        # input_hash 应被设为指纹
        assert result.input_hash is not None and len(result.input_hash) == 64  # SHA-256 hex

    @pytest.mark.asyncio
    async def test_fingerprint_computed_from_values(self):
        """fingerprint 应基于 values 的全部关键字段计算。"""
        s1 = _FakeSession(existing=None)
        s2 = _FakeSession(existing=None)

        v1 = _base_values(1, prompt_version="v1")
        v2 = _base_values(1, prompt_version="v1")  # 同值

        r1 = await _insert_or_find_by_fingerprint(s1, values=v1)
        r2 = await _insert_or_find_by_fingerprint(s2, values=v2)

        # 同值 → 同 fingerprint(跨 session 也一致)
        assert r1.input_hash == r2.input_hash

    @pytest.mark.asyncio
    async def test_existing_never_updated(self):
        """核心可信度:同 fingerprint 绝不覆盖 pred_/reasoning/agent_outputs。"""
        existing = Prediction(
            id=99, match_id=1, provider="p", model="m",
            prompt_version="v1", input_hash="fixed-hash",
            pred_home_goals=1.0, pred_away_goals=0.0,
            reasoning="original", agent_outputs=[{"agent": "form"}],
        )
        s = _FakeSession(existing=existing)

        # 即便传入不同的 pred_*,也应返回原行(字段不变)
        values = _base_values(1, prompt_version="v1")
        # 让 fake 返回 existing: 需 fingerprint 匹配。fake.execute 始终返回 existing。
        result = await _insert_or_find_by_fingerprint(s, values=values)

        assert result is existing
        assert result.pred_home_goals == 1.0, "pred_home_goals 不应被覆盖"
        assert result.reasoning == "original", "reasoning 不应被覆盖"
        assert result.agent_outputs == [{"agent": "form"}], "agent_outputs 不应被覆盖"


class TestFingerprintDeterminism:
    """fingerprint 必须稳定(同输入 → 同 hash)。"""

    def test_same_values_same_fingerprint(self):
        v = _base_values(1)
        assert _compute_fingerprint(v) == _compute_fingerprint(dict(v))

    def test_different_prompt_version_different_fingerprint(self):
        v1 = _base_values(1, prompt_version="v1")
        v2 = _base_values(1, prompt_version="v2")
        assert _compute_fingerprint(v1) != _compute_fingerprint(v2)

    def test_different_agent_ids_different_fingerprint(self):
        v1 = _base_values(1, agent_ids=["form", "stats"])
        v2 = _base_values(1, agent_ids=["form", "h2h"])
        assert _compute_fingerprint(v1) != _compute_fingerprint(v2)

    def test_different_cutoff_different_fingerprint(self):
        v1 = _base_values(1, prediction_cutoff_at="2026-01-01T14:00:00+00:00")
        v2 = _base_values(1, prediction_cutoff_at="2026-01-01T10:00:00+00:00")
        assert _compute_fingerprint(v1) != _compute_fingerprint(v2)

    def test_different_context_different_fingerprint(self):
        v1 = _base_values(1, context_hash="ch1")
        v2 = _base_values(1, context_hash="ch2")
        assert _compute_fingerprint(v1) != _compute_fingerprint(v2)

    def test_agent_ids_order_independent(self):
        """agent_ids 排序后计算,顺序不影响 hash。"""
        v1 = _base_values(1, agent_ids=["stats", "form"])
        v2 = _base_values(1, agent_ids=["form", "stats"])
        assert _compute_fingerprint(v1) == _compute_fingerprint(v2)