"""PR-RAG-Time-1: freshness decay 단위 테스트.

가드레일 (plan §10):
  1. news recent vs old      — recent 가 final score 높음
  2. law_monitor recent vs old
  3. non-target source unaffected — manual 등은 score 변화 0
  4. missing created_at unaffected — decay None, score 변화 0
  5. future created_at clamped — age_days = 0
  6. floor behavior — 매우 오래되어도 multiplier 0.7 미만 안 떨어짐
"""

from __future__ import annotations

import math
import os
import sys
from datetime import datetime, timedelta, timezone
from types import SimpleNamespace

# tests/ → 프로젝트 루트 → app/
sys.path.insert(0, os.path.join(os.path.dirname(__file__), "..", "app"))

import pytest

from services.search.freshness_decay import (
    DECAY_FLOOR,
    HALF_LIFE_DAYS,
    _DocMeta,
    adjusted_score,
    apply_freshness_decay,
    compute_age_days,
    compute_decay,
    freshness_policy,
)


NOW = datetime(2026, 5, 3, 12, 0, 0, tzinfo=timezone.utc)


def _meta(channel: str | None, *, days_ago: float | None = 30.0, origin: str | None = None,
          material_type: str | None = None) -> _DocMeta:
    if days_ago is None:
        created = None
    elif days_ago < 0:
        # future: now + |days_ago|
        created = NOW + timedelta(days=-days_ago)
    else:
        created = NOW - timedelta(days=days_ago)
    return _DocMeta(source_channel=channel, content_origin=origin, created_at=created,
                    material_type=material_type)


# ─── policy dispatcher ────────────────────────────────────────────


def test_policy_news():
    assert freshness_policy(_meta("news")) == "news_90d"


def test_policy_law_monitor_now_unaffected():
    # C-1 후속: law_365d 폐기 → law_monitor 비적용 (현행성은 version_status 가 처리)
    assert freshness_policy(_meta("law_monitor")) is None


def test_policy_incident():
    # C-1 후속: 재해사례/사망사고(material_type='incident') → news_90d 흡수 (source 무관)
    assert freshness_policy(_meta("crawl", material_type="incident")) == "news_90d"
    assert freshness_policy(_meta("inbox_route", material_type="incident")) == "news_90d"


def test_policy_manual_unaffected():
    assert freshness_policy(_meta("manual")) is None


def test_policy_drive_sync_unaffected():
    assert freshness_policy(_meta("drive_sync")) is None


def test_policy_inbox_route_unaffected():
    assert freshness_policy(_meta("inbox_route")) is None


def test_policy_memo_unaffected():
    assert freshness_policy(_meta("memo")) is None


def test_policy_ai_drafted_skipped_even_for_news():
    # 가드 2: content_origin='ai_drafted' 면 source_channel 무관 비적용
    assert freshness_policy(_meta("news", origin="ai_drafted")) is None


def test_policy_meta_none():
    assert freshness_policy(None) is None


def test_policy_unknown_source():
    # 가드 6: unknown source_channel → no decay
    assert freshness_policy(_meta("future_channel")) is None


# ─── compute_age_days ─────────────────────────────────────────────


def test_age_normal():
    age = compute_age_days(NOW - timedelta(days=10), now=NOW)
    assert age == pytest.approx(10.0)


def test_age_missing_date():
    # 가드 6: missing date → None
    assert compute_age_days(None, now=NOW) is None


def test_age_future_clamped_to_zero():
    # 가드 6: future date → age_days = 0 (음수 금지)
    future = NOW + timedelta(days=5)
    assert compute_age_days(future, now=NOW) == 0.0


def test_age_naive_datetime_treated_as_utc():
    # DB는 timestamptz 라 naive 안 오지만 mock 안전망
    naive = datetime(2026, 4, 23, 12, 0, 0)  # 10일 전
    age = compute_age_days(naive, now=NOW)
    assert age == pytest.approx(10.0)


# ─── compute_decay ────────────────────────────────────────────────


def test_decay_at_half_life_news():
    # age=90 → decay = 0.5
    assert compute_decay(90.0, "news_90d") == pytest.approx(0.5, rel=1e-6)


def test_decay_law_365d_removed_returns_one():
    # C-1 후속: law_365d 폐기 → HALF_LIFE_DAYS 미등록 policy → decay 1.0 (no-op)
    assert compute_decay(365.0, "law_365d") == 1.0


def test_decay_age_zero_full():
    assert compute_decay(0.0, "news_90d") == pytest.approx(1.0)


def test_decay_news_30_days():
    expected = math.exp(-math.log(2) * 30 / 90)  # ~0.794
    assert compute_decay(30.0, "news_90d") == pytest.approx(expected, rel=1e-6)


def test_decay_policy_none_returns_one():
    assert compute_decay(100.0, None) == 1.0


def test_decay_age_none_returns_one():
    assert compute_decay(None, "news_90d") == 1.0


# ─── adjusted_score ───────────────────────────────────────────────


def test_adjusted_score_decay_one_no_change():
    # decay=1 → final = base * 1.0
    assert adjusted_score(0.85, 1.0) == pytest.approx(0.85)


def test_adjusted_score_decay_zero_floor():
    # 가드 11: decay=0 → final = base * FLOOR (0.7), 그 아래로 안 내려감
    assert adjusted_score(0.85, 0.0) == pytest.approx(0.85 * DECAY_FLOOR)


def test_adjusted_score_old_news_floor_bound():
    # 매우 오래된 뉴스 (age >> half_life) — multiplier 가 0.7 미만 절대 아님
    decay = compute_decay(10000.0, "news_90d")  # ~0
    score = adjusted_score(1.0, decay)
    assert score >= DECAY_FLOOR
    assert score == pytest.approx(DECAY_FLOOR, abs=0.01)


# ─── apply_freshness_decay (integration with mock session) ────────


class _MockSession:
    """SQLAlchemy AsyncSession mock — execute(text, params) 호출만 받음."""

    def __init__(self, meta_rows: list[dict]):
        self._rows = meta_rows
        self.last_ids: list[int] | None = None

    async def execute(self, _stmt, params):
        self.last_ids = list(params["ids"])
        # SearchPipeline 의 row._mapping 패턴이 아니라 row.field 패턴이라
        # SimpleNamespace 로 충분.
        return [SimpleNamespace(**row) for row in self._rows if row["id"] in self.last_ids]


def _result(doc_id: int, score: float):
    """SearchResult 흉내. freshness_decay 는 .id / .score / .freshness_debug 만 만짐."""
    return SimpleNamespace(id=doc_id, score=score, freshness_debug=None)


@pytest.mark.asyncio
async def test_apply_news_recent_vs_old_recent_higher():
    # 가드 1: news recent vs old → recent 가 final score 높음
    base = 0.50  # 동일 base
    rows = [
        {"id": 1, "source_channel": "news", "content_origin": "extracted",
         "created_at": NOW - timedelta(days=3)},   # 매우 최근
        {"id": 2, "source_channel": "news", "content_origin": "extracted",
         "created_at": NOW - timedelta(days=365)},  # 1년 전
    ]
    session = _MockSession(rows)
    results = [_result(1, base), _result(2, base)]
    out = await apply_freshness_decay(results, session, now=NOW)
    # 재정렬: 최근(id=1) 이 위
    assert out[0].id == 1
    assert out[0].score > out[1].score
    # debug
    assert out[0].freshness_debug["freshness_policy"] == "news_90d"
    assert out[0].freshness_debug["base_score"] == base
    assert out[0].freshness_debug["age_days"] == 3
    assert out[0].freshness_debug["freshness_date_source"] == "created_at"
    assert out[1].freshness_debug["age_days"] == 365


@pytest.mark.asyncio
async def test_apply_law_monitor_now_unaffected():
    # C-1 후속: law_monitor freshness 폐기 → recent/old 동일 score (재정렬 없음)
    base = 0.50
    rows = [
        {"id": 1, "source_channel": "law_monitor", "content_origin": "extracted",
         "material_type": "law", "created_at": NOW - timedelta(days=10)},
        {"id": 2, "source_channel": "law_monitor", "content_origin": "extracted",
         "material_type": "law", "created_at": NOW - timedelta(days=730)},
    ]
    session = _MockSession(rows)
    results = [_result(1, base), _result(2, base)]
    out = await apply_freshness_decay(results, session, now=NOW)
    assert out[0].score == base and out[1].score == base
    assert out[0].freshness_debug["freshness_policy"] is None


@pytest.mark.asyncio
async def test_apply_incident_recent_vs_old_recent_higher():
    # C-1 후속: 재해사례(incident) recent 가 위 (news_90d 흡수, source_channel='crawl')
    base = 0.50
    rows = [
        {"id": 1, "source_channel": "crawl", "content_origin": "extracted",
         "material_type": "incident", "created_at": NOW - timedelta(days=5)},
        {"id": 2, "source_channel": "crawl", "content_origin": "extracted",
         "material_type": "incident", "created_at": NOW - timedelta(days=400)},
    ]
    session = _MockSession(rows)
    results = [_result(1, base), _result(2, base)]
    out = await apply_freshness_decay(results, session, now=NOW)
    assert out[0].id == 1
    assert out[0].score > out[1].score
    assert out[0].freshness_debug["freshness_policy"] == "news_90d"


@pytest.mark.asyncio
async def test_apply_non_target_unaffected():
    # 가드 3: manual 은 score 변화 0
    base = 0.42
    rows = [
        {"id": 1, "source_channel": "manual", "content_origin": "extracted",
         "created_at": NOW - timedelta(days=1000)},
    ]
    session = _MockSession(rows)
    results = [_result(1, base)]
    out = await apply_freshness_decay(results, session, now=NOW)
    assert out[0].score == base
    assert out[0].freshness_debug["freshness_policy"] is None
    assert out[0].freshness_debug["base_score"] == base
    assert out[0].freshness_debug["decay_factor"] is None


@pytest.mark.asyncio
async def test_apply_missing_created_at_unaffected():
    # 가드 4: missing date → score 변화 0
    base = 0.50
    rows = [
        {"id": 1, "source_channel": "news", "content_origin": "extracted",
         "created_at": None},
    ]
    session = _MockSession(rows)
    results = [_result(1, base)]
    out = await apply_freshness_decay(results, session, now=NOW)
    assert out[0].score == base
    # policy 는 'news_90d' 인데 age None 이라 decay 1.0 → adjusted_score 호출 자체가 안 됨 (policy 없는 길 아님)
    # freshness_decay 코드: policy True + age None → decay=1.0 → adjusted_score(base, 1.0) = base * 1.0
    # 결과적으로 score 변화 0 — 가드 4 통과
    assert out[0].freshness_debug["freshness_policy"] == "news_90d"
    assert out[0].freshness_debug["age_days"] is None


@pytest.mark.asyncio
async def test_apply_future_created_at_clamped():
    # 가드 5: future created_at → age_days = 0 → decay = 1.0 → score 그대로
    base = 0.50
    rows = [
        {"id": 1, "source_channel": "news", "content_origin": "extracted",
         "created_at": NOW + timedelta(days=10)},
    ]
    session = _MockSession(rows)
    results = [_result(1, base)]
    out = await apply_freshness_decay(results, session, now=NOW)
    assert out[0].freshness_debug["age_days"] == 0
    assert out[0].score == pytest.approx(base * 1.0)


@pytest.mark.asyncio
async def test_apply_floor_never_below_0_7():
    # 가드 6 (floor): 매우 오래된 news → multiplier 0.7 미만 절대 아님
    base = 1.0
    rows = [
        {"id": 1, "source_channel": "news", "content_origin": "extracted",
         "created_at": NOW - timedelta(days=10000)},  # 27년 ㅋ
    ]
    session = _MockSession(rows)
    results = [_result(1, base)]
    out = await apply_freshness_decay(results, session, now=NOW)
    assert out[0].score >= base * DECAY_FLOOR
    assert out[0].score == pytest.approx(base * DECAY_FLOOR, abs=0.001)


@pytest.mark.asyncio
async def test_apply_ai_drafted_news_skipped():
    # 가드 2 통합: source=news 라도 content_origin='ai_drafted' 면 비적용
    base = 0.50
    rows = [
        {"id": 1, "source_channel": "news", "content_origin": "ai_drafted",
         "created_at": NOW - timedelta(days=300)},
    ]
    session = _MockSession(rows)
    results = [_result(1, base)]
    out = await apply_freshness_decay(results, session, now=NOW)
    assert out[0].score == base
    assert out[0].freshness_debug["freshness_policy"] is None


@pytest.mark.asyncio
async def test_apply_empty_results_noop():
    session = _MockSession([])
    out = await apply_freshness_decay([], session, now=NOW)
    assert out == []


@pytest.mark.asyncio
async def test_apply_resort_by_adjusted_score():
    # base score 가 같아도 freshness 가 정렬을 바꾼다.
    rows = [
        {"id": 1, "source_channel": "news", "content_origin": "extracted",
         "created_at": NOW - timedelta(days=200)},   # decay ~0.21
        {"id": 2, "source_channel": "news", "content_origin": "extracted",
         "created_at": NOW - timedelta(days=10)},   # decay ~0.926
        {"id": 3, "source_channel": "manual", "content_origin": "extracted",
         "created_at": NOW - timedelta(days=10000)},  # 비적용 → base 그대로
    ]
    session = _MockSession(rows)
    # 일부러 base 동일 0.5 → 정렬은 freshness 만으로 결정
    results = [_result(1, 0.5), _result(2, 0.5), _result(3, 0.5)]
    out = await apply_freshness_decay(results, session, now=NOW)
    # manual 은 base 0.5 그대로, news 10일 전 은 0.5 * (0.7 + 0.3*0.926) ≈ 0.489
    # news 200일 전 은 0.5 * (0.7 + 0.3*0.21) ≈ 0.382
    # 정렬: manual(0.5) > news_recent(0.489) > news_old(0.382)
    assert out[0].id == 3  # manual
    assert out[1].id == 2  # news 10일
    assert out[2].id == 1  # news 200일