xiaoxia-saas/tests/unit/test_smart_asset_selector.py

"""SmartAssetSelector 智能素材选择服务单元测试."""

from __future__ import annotations

import unittest
from dataclasses import dataclass

from app.services.smart_asset_selector import (
    SmartAssetSelector,
    _MEDIUM_BUCKET_MAX,
    _SHORT_BUCKET_MAX,
)


@dataclass
class MockAsset:
    """模拟 Asset 实体."""

    id: str
    quality_score: float | None = None
    width: int | None = None
    height: int | None = None
    duration: float | None = None
    file_size: int = 0
    mime_type: str = "video/mp4"
    status: str = "ready"

    @property
    def status_value(self) -> str:
        return self.status


class TestSmartAssetSelectorScoring(unittest.TestCase):
    """评分维度测试."""

    def setUp(self):
        self.selector = SmartAssetSelector()

    def test_quality_score_normalization(self):
        """质量分正确归一化到 0-1."""
        asset_high = MockAsset(id="1", quality_score=90.0)
        asset_low = MockAsset(id="2", quality_score=30.0)
        asset_none = MockAsset(id="3", quality_score=None)

        detail_high = self.selector._score_asset(asset_high)
        detail_low = self.selector._score_asset(asset_low)
        detail_none = self.selector._score_asset(asset_none)

        # 90分 → 0.9 × 0.5权重 = 0.45 基础贡献
        self.assertAlmostEqual(detail_high.quality_score, 0.9, delta=0.01)
        # 30分 → 0.3 × 0.5权重 = 0.15 基础贡献
        self.assertAlmostEqual(detail_low.quality_score, 0.3, delta=0.01)
        # 无质量分给默认 0.5
        self.assertAlmostEqual(detail_none.quality_score, 0.5, delta=0.01)

    def test_resolution_score_1080p_full(self):
        """1080p 分辨率得满分."""
        asset = MockAsset(id="1", width=1920, height=1080)
        detail = self.selector._score_asset(asset)
        self.assertAlmostEqual(detail.resolution_score, 1.0, delta=0.01)

    def test_resolution_score_4k_full(self):
        """4K 也得满分（高于目标分辨率不扣分）."""
        asset = MockAsset(id="1", width=3840, height=2160)
        detail = self.selector._score_asset(asset)
        self.assertAlmostEqual(detail.resolution_score, 1.0, delta=0.01)

    def test_resolution_score_720p_lower(self):
        """720p 低于 1080p，得分低于 1."""
        asset = MockAsset(id="1", width=1280, height=720)
        detail = self.selector._score_asset(asset)
        self.assertLess(detail.resolution_score, 1.0)
        self.assertGreater(detail.resolution_score, 0.3)

    def test_resolution_score_none(self):
        """分辨率未知给中评分."""
        asset = MockAsset(id="1", width=None, height=None)
        detail = self.selector._score_asset(asset)
        self.assertAlmostEqual(detail.resolution_score, 0.5, delta=0.01)

    def test_duration_score_optimal(self):
        """最佳时长区间内得满分."""
        asset = MockAsset(id="1", duration=10.0)
        detail = self.selector._score_asset(asset)
        self.assertAlmostEqual(detail.duration_score, 1.0, delta=0.01)

    def test_duration_score_too_short(self):
        """时长过短扣分."""
        asset = MockAsset(id="1", duration=1.0)
        detail = self.selector._score_asset(asset)
        self.assertLess(detail.duration_score, 1.0)

    def test_duration_score_too_long(self):
        """时长过长扣分."""
        asset = MockAsset(id="1", duration=120.0)
        detail = self.selector._score_asset(asset)
        self.assertLess(detail.duration_score, 1.0)

    def test_duration_score_none(self):
        """时长未知给中评分."""
        asset = MockAsset(id="1", duration=None)
        detail = self.selector._score_asset(asset)
        self.assertAlmostEqual(detail.duration_score, 0.5, delta=0.01)

    def test_total_score_weighted_sum(self):
        """总分是各维度的加权和."""
        asset = MockAsset(
            id="1",
            quality_score=100.0,  # 1.0 × 0.5 = 0.5
            width=1920,  # 1.0 × 0.2 = 0.2
            height=1080,
            duration=10.0,  # 1.0 × 0.2 = 0.2
            file_size=10_000_000,  # ~8Mbps，10秒 → 约 1.0 × 0.1 = 0.1
        )
        detail = self.selector._score_asset(asset)
        # 理论上接近 1.0
        self.assertGreater(detail.total_score, 0.85)
        self.assertLessEqual(detail.total_score, 1.0)


class TestSmartAssetSelectorSelection(unittest.TestCase):
    """选择逻辑测试."""

    def setUp(self):
        self.selector = SmartAssetSelector(min_quality_score=0)  # 测试时关闭质量门槛

    def _make_assets(self, count: int, base_quality: float = 80.0) -> list[MockAsset]:
        assets = []
        for i in range(count):
            assets.append(
                MockAsset(
                    id=f"asset_{i}",
                    quality_score=base_quality - i * 5,  # 质量递减
                    width=1920,
                    height=1080,
                    duration=10.0 + i,
                    file_size=5_000_000 + i * 100_000,
                )
            )
        return assets

    def test_select_all_when_count_zero(self):
        """count=0 时返回全部符合条件的."""
        assets = self._make_assets(10)
        result = self.selector.select(assets, count=0)
        self.assertEqual(len(result.selected_ids), 10)
        self.assertEqual(result.total_candidates, 10)

    def test_select_top_n(self):
        """返回指定数量的 top N."""
        assets = self._make_assets(10)
        result = self.selector.select(assets, count=3)
        self.assertEqual(len(result.selected_ids), 3)
        # 最高分的应该是 asset_0（质量分最高）
        self.assertEqual(result.selected_ids[0], "asset_0")

    def test_select_more_than_available(self):
        """请求数量超过候选数量时返回全部."""
        assets = self._make_assets(5)
        result = self.selector.select(assets, count=10)
        self.assertEqual(len(result.selected_ids), 5)

    def test_filter_non_ready(self):
        """非 ready 状态的素材被过滤."""
        assets = [
            MockAsset(id="1", quality_score=90.0, status="ready"),
            MockAsset(id="2", quality_score=80.0, status="processing"),
            MockAsset(id="3", quality_score=70.0, status="ready"),
        ]
        result = self.selector.select(assets, count=0)
        self.assertEqual(len(result.selected_ids), 2)
        self.assertIn("1", result.selected_ids)
        self.assertIn("3", result.selected_ids)
        self.assertNotIn("2", result.selected_ids)

    def test_filter_non_video(self):
        """非视频素材被过滤."""
        assets = [
            MockAsset(id="1", quality_score=90.0, mime_type="video/mp4"),
            MockAsset(id="2", quality_score=80.0, mime_type="image/jpeg"),
            MockAsset(id="3", quality_score=70.0, mime_type="video/quicktime"),
        ]
        result = self.selector.select(assets, count=0)
        self.assertEqual(len(result.selected_ids), 2)

    def test_min_quality_filter(self):
        """最低质量分门槛过滤."""
        selector = SmartAssetSelector(min_quality_score=60.0)
        assets = [
            MockAsset(id="1", quality_score=90.0),
            MockAsset(id="2", quality_score=50.0),  # 低于门槛
            MockAsset(id="3", quality_score=70.0),
            MockAsset(id="4", quality_score=30.0),  # 低于门槛
        ]
        result = selector.select(assets, count=0)
        self.assertEqual(len(result.selected_ids), 2)
        self.assertEqual(result.filtered_out, 2)
        self.assertIn("1", result.selected_ids)
        self.assertIn("3", result.selected_ids)

    def test_empty_input(self):
        """空输入返回空结果."""
        result = self.selector.select([], count=5)
        self.assertEqual(result.selected_ids, [])
        self.assertEqual(result.total_candidates, 0)
        self.assertEqual(result.avg_score, 0.0)

    def test_sorted_by_score_descending(self):
        """结果按总分降序排列."""
        assets = self._make_assets(5)
        result = self.selector.select(assets, count=0, ensure_diversity=False)
        scores = [d.total_score for d in result.details]
        # 应该是降序
        self.assertEqual(scores, sorted(scores, reverse=True))


class TestSmartAssetSelectorDiversity(unittest.TestCase):
    """多样性选择测试."""

    def setUp(self):
        self.selector = SmartAssetSelector(min_quality_score=0)

    def _make_assets(self, count: int, base_quality: float = 80.0) -> list[MockAsset]:
        assets = []
        for i in range(count):
            assets.append(
                MockAsset(
                    id=f"asset_{i}",
                    quality_score=base_quality - i * 5,
                    width=1920,
                    height=1080,
                    duration=10.0 + i,
                    file_size=5_000_000 + i * 100_000,
                )
            )
        return assets

    def test_diversity_all_short(self):
        """全是短素材时不报错，正常返回."""
        assets = []
        for i in range(10):
            assets.append(
                MockAsset(
                    id=f"short_{i}",
                    quality_score=80.0 + i,
                    width=1920,
                    height=1080,
                    duration=2.0 + i * 0.1,  # 都 < 5s
                    file_size=1_000_000,
                )
            )
        result = self.selector.select(assets, count=5, ensure_diversity=True)
        self.assertEqual(len(result.selected_ids), 5)

    def test_diversity_mixed_buckets(self):
        """混合时长素材时，各桶都有代表."""
        assets = []
        # 短素材（质量分高）
        for i in range(5):
            assets.append(
                MockAsset(
                    id=f"short_{i}",
                    quality_score=95.0 - i,
                    width=1920,
                    height=1080,
                    duration=3.0,
                    file_size=2_000_000,
                )
            )
        # 中素材（质量分中等）
        for i in range(5):
            assets.append(
                MockAsset(
                    id=f"medium_{i}",
                    quality_score=85.0 - i,
                    width=1920,
                    height=1080,
                    duration=10.0,
                    file_size=5_000_000,
                )
            )
        # 长素材（质量分低）
        for i in range(5):
            assets.append(
                MockAsset(
                    id=f"long_{i}",
                    quality_score=75.0 - i,
                    width=1920,
                    height=1080,
                    duration=60.0,
                    file_size=20_000_000,
                )
            )

        result = self.selector.select(assets, count=6, ensure_diversity=True)
        selected = result.selected_ids

        # 6个素材，每个桶至少有1个（基础配额 max(1, 6//3)=2）
        short_count = sum(1 for sid in selected if sid.startswith("short_"))
        medium_count = sum(1 for sid in selected if sid.startswith("medium_"))
        long_count = sum(1 for sid in selected if sid.startswith("long_"))

        # 每个桶至少1个
        self.assertGreaterEqual(short_count, 1)
        self.assertGreaterEqual(medium_count, 1)
        self.assertGreaterEqual(long_count, 1)
        self.assertEqual(len(selected), 6)

    def test_diversity_disabled_returns_top(self):
        """关闭多样性时，直接返回 top N（可能全是短素材）."""
        assets = []
        # 短素材（质量分最高）
        for i in range(10):
            assets.append(
                MockAsset(
                    id=f"short_{i}",
                    quality_score=95.0 - i,
                    width=1920,
                    height=1080,
                    duration=3.0,
                    file_size=2_000_000,
                )
            )
        # 长素材（质量分低）
        for i in range(5):
            assets.append(
                MockAsset(
                    id=f"long_{i}",
                    quality_score=70.0,
                    width=1920,
                    height=1080,
                    duration=60.0,
                    file_size=20_000_000,
                )
            )

        result = self.selector.select(assets, count=5, ensure_diversity=False)
        selected = result.selected_ids
        # 全是短素材（因为质量分高）
        self.assertTrue(all(s.startswith("short_") for s in selected))

    def test_avg_score_calculated(self):
        """平均分正确计算."""
        assets = self._make_assets(3)
        result = self.selector.select(assets, count=3, ensure_diversity=False)
        expected_avg = sum(d.total_score for d in result.details) / 3
        self.assertAlmostEqual(result.avg_score, expected_avg, delta=0.001)


class TestSmartAssetSelectorEdgeCases(unittest.TestCase):
    """边界情况测试."""

    def setUp(self):
        self.selector = SmartAssetSelector(min_quality_score=0)

    def _make_assets(self, count: int, base_quality: float = 80.0) -> list[MockAsset]:
        assets = []
        for i in range(count):
            assets.append(
                MockAsset(
                    id=f"asset_{i}",
                    quality_score=base_quality - i * 5,
                    width=1920,
                    height=1080,
                    duration=10.0 + i,
                    file_size=5_000_000 + i * 100_000,
                )
            )
        return assets

    def test_single_asset(self):
        """单个素材正常返回."""
        assets = [MockAsset(id="1", quality_score=80.0, width=1920, height=1080, duration=10.0)]
        result = self.selector.select(assets, count=1)
        self.assertEqual(len(result.selected_ids), 1)
        self.assertEqual(result.selected_ids[0], "1")

    def test_zero_width_height(self):
        """宽高为0时按未知处理."""
        asset = MockAsset(id="1", width=0, height=0)
        detail = self.selector._score_asset(asset)
        self.assertAlmostEqual(detail.resolution_score, 0.5, delta=0.01)

    def test_negative_duration(self):
        """负时长按未知处理."""
        asset = MockAsset(id="1", duration=-5.0)
        detail = self.selector._score_asset(asset)
        self.assertAlmostEqual(detail.duration_score, 0.5, delta=0.01)

    def test_zero_file_size_with_duration(self):
        """文件大小为0时码率评分中等."""
        asset = MockAsset(id="1", file_size=0, duration=10.0)
        detail = self.selector._score_asset(asset)
        self.assertAlmostEqual(detail.bitrate_score, 0.5, delta=0.01)

    def test_bitrate_score_optimal(self):
        """最佳码率范围得满分."""
        # 5 Mbps × 10秒 = 6.25 MB → file_size = 6,250,000 bytes
        asset = MockAsset(id="1", file_size=6_250_000, duration=10.0)
        detail = self.selector._score_asset(asset)
        self.assertAlmostEqual(detail.bitrate_score, 1.0, delta=0.01)

    def test_details_match_selected_ids(self):
        """details 列表和 selected_ids 顺序一致."""
        assets = self._make_assets(5)
        result = self.selector.select(assets, count=3, ensure_diversity=False)
        self.assertEqual(len(result.details), 3)
        for i, aid in enumerate(result.selected_ids):
            self.assertEqual(result.details[i].asset_id, aid)


if __name__ == "__main__":
    unittest.main()