xiaoxia-saas/tests/unit/test_duplicate_rate.py

"""Tests for duplicate_rate computation and API response."""

import json
import sys
from pathlib import Path
from unittest.mock import MagicMock, patch

import pytest

# Mock cv2 and numpy before any imports that need them
sys.modules.setdefault("cv2", MagicMock())

ROOT = Path(__file__).resolve().parents[2]
sys.path.insert(0, str(ROOT / "apps" / "api"))
sys.path.insert(0, str(ROOT / "packages"))
sys.path.insert(0, str(ROOT / "apps" / "worker"))

# Issue #1658: 归一化颜色直方图（96 维 = 3 通道 × 32 bins，sum=1.0）。
# 相同归一化直方图的 Bhattacharyya 系数 ≈ 1.0（颜色完全一致），
# 用于验证 pHash + 颜色直方图融合后的 duplicate_rate。
_NORM_HISTOGRAM = [1.0 / 96] * 96


class TestComputeDuplicateRate:
    """Test VideoDeduplicator.compute_duplicate_rate."""

    def _make_fingerprint(self, md5="abc123", phashes=None, histograms=None):
        from video_processing.dedup import VideoFingerprint

        phashes = phashes or ["ff00ff00ff00ff00"]
        if histograms is None:
            histograms = [_NORM_HISTOGRAM] * len(phashes)
        return VideoFingerprint(
            md5=md5,
            keyframe_phashes=phashes,
            color_histograms=histograms,
            duration=10.0,
            resolution=(1920, 1080),
        )

    def _make_existing_video(self, vid, fingerprint_dict):
        from packages.domain import GeneratedVideo

        return GeneratedVideo(
            id=vid,
            project_id="proj1",
            generation_task_id="task1",
            name=f"video-{vid}",
            file_url=f"https://example.com/{vid}.mp4",
            file_size=1000,
            duration=10.0,
            width=1920,
            height=1080,
            fps=25.0,
            video_fingerprint=fingerprint_dict,
        )

    def test_no_existing_videos_returns_zero(self):
        from video_processing.dedup import VideoDeduplicator

        deduplicator = VideoDeduplicator()
        fingerprint = self._make_fingerprint()
        session = MagicMock()

        with patch("video_processing.dedup.SQLAlchemyGeneratedVideoRepository") as MockRepo:
            mock_repo = MockRepo.return_value
            query_mock = MagicMock()
            query_mock.filter.return_value = query_mock
            query_mock.order_by.return_value.limit.return_value.all.return_value = []
            session.query.return_value = query_mock
            rate = deduplicator.compute_duplicate_rate(fingerprint, "proj1", "vid1", session)

        assert rate == 0.0

    def test_md5_match_returns_100(self):
        from video_processing.dedup import VideoDeduplicator

        from packages.adapters.sqlalchemy_impl.models import GeneratedVideoModel

        deduplicator = VideoDeduplicator()
        fingerprint = self._make_fingerprint(md5="exact_match_md5")
        session = MagicMock()

        existing = self._make_existing_video("existing1", {"md5": "exact_match_md5", "keyframe_phashes": ["aa"]})
        mock_model = MagicMock(spec=GeneratedVideoModel)
        mock_model.id = existing.id
        mock_model.project_id = existing.project_id
        mock_model.video_fingerprint = existing.video_fingerprint
        mock_model.generated_at = "2026-01-01"

        with patch("video_processing.dedup.SQLAlchemyGeneratedVideoRepository") as MockRepo:
            mock_repo = MockRepo.return_value
            mock_repo._to_domain.return_value = existing
            # 链式 filter: 第一次 scope filter，第二次 self-exclusion filter
            # 让 filter() 返回的对象仍然支持 order_by() 链
            query_mock = MagicMock()
            query_mock.filter.return_value = query_mock  # filter → filter chainable
            query_mock.order_by.return_value.limit.return_value.all.return_value = [mock_model]
            session.query.return_value = query_mock
            rate = deduplicator.compute_duplicate_rate(fingerprint, "proj1", "vid1", session)

        assert rate == 100.0

    def test_phash_similarity_computed(self):
        from video_processing.dedup import VideoDeduplicator

        from packages.adapters.sqlalchemy_impl.models import GeneratedVideoModel

        deduplicator = VideoDeduplicator()
        fingerprint = self._make_fingerprint(md5="different_md5", phashes=["ff00ff00ff00ff00"])
        session = MagicMock()

        existing = self._make_existing_video(
            "existing1",
            {
                "md5": "other_md5",
                "keyframe_phashes": ["ff00ff00ff00ff03"],
                # Issue #1658: 提供归一化直方图（颜色一致 → hist_sim≈1.0）
                "color_histograms": [_NORM_HISTOGRAM],
            },
        )
        mock_model = MagicMock(spec=GeneratedVideoModel)
        mock_model.id = existing.id
        mock_model.project_id = existing.project_id
        mock_model.video_fingerprint = existing.video_fingerprint
        mock_model.generated_at = "2026-01-01"

        with patch("video_processing.dedup.SQLAlchemyGeneratedVideoRepository") as MockRepo:
            mock_repo = MockRepo.return_value
            mock_repo._to_domain.return_value = existing
            query_mock = MagicMock()
            query_mock.filter.return_value = query_mock
            query_mock.order_by.return_value.limit.return_value.all.return_value = [mock_model]
            session.query.return_value = query_mock
            rate = deduplicator.compute_duplicate_rate(fingerprint, "proj1", "vid1", session)

        # Issue #1658 融合公式：hamming distance = 2 → phash_sim = 1 - 2/64；
        # hist_sim ≈ 1.0 → fusion = 0.7*(1-2/64) + 0.3*1.0 = 0.978125 → 97.81
        assert rate == pytest.approx(97.81, abs=0.1)

    def test_excludes_self_video(self):
        from video_processing.dedup import VideoDeduplicator

        from packages.adapters.sqlalchemy_impl.models import GeneratedVideoModel

        deduplicator = VideoDeduplicator()
        fingerprint = self._make_fingerprint(md5="same_md5")
        session = MagicMock()

        self_video = self._make_existing_video("vid1", {"md5": "same_md5", "keyframe_phashes": ["aa"]})
        mock_model = MagicMock(spec=GeneratedVideoModel)
        mock_model.id = self_video.id
        mock_model.project_id = self_video.project_id
        mock_model.video_fingerprint = self_video.video_fingerprint
        mock_model.generated_at = "2026-01-01"

        with patch("video_processing.dedup.SQLAlchemyGeneratedVideoRepository") as MockRepo:
            mock_repo = MockRepo.return_value
            mock_repo._to_domain.return_value = self_video
            query_mock = MagicMock()
            query_mock.filter.return_value = query_mock
            query_mock.order_by.return_value.limit.return_value.all.return_value = [mock_model]
            session.query.return_value = query_mock
            rate = deduplicator.compute_duplicate_rate(fingerprint, "proj1", "vid1", session)

        assert rate == 0.0

    def test_takes_max_similarity(self):
        from video_processing.dedup import VideoDeduplicator

        from packages.adapters.sqlalchemy_impl.models import GeneratedVideoModel

        deduplicator = VideoDeduplicator()
        fingerprint = self._make_fingerprint(md5="new_md5", phashes=["ff00ff00ff00ff00"])
        session = MagicMock()

        existing1 = self._make_existing_video(
            "e1",
            {
                "md5": "md5_1",
                "keyframe_phashes": ["ff00ff00ff00ff0f"],
                # Issue #1658: 颜色直方图一致 → hist_sim≈1.0
                "color_histograms": [_NORM_HISTOGRAM],
            },
        )
        existing2 = self._make_existing_video(
            "e2",
            {
                "md5": "md5_2",
                "keyframe_phashes": ["ff00ff00ff00ff01"],
                "color_histograms": [_NORM_HISTOGRAM],
            },
        )
        mock_model1 = MagicMock(spec=GeneratedVideoModel)
        mock_model1.id = existing1.id
        mock_model1.project_id = existing1.project_id
        mock_model1.video_fingerprint = existing1.video_fingerprint
        mock_model1.generated_at = "2026-01-02"
        mock_model2 = MagicMock(spec=GeneratedVideoModel)
        mock_model2.id = existing2.id
        mock_model2.project_id = existing2.project_id
        mock_model2.video_fingerprint = existing2.video_fingerprint
        mock_model2.generated_at = "2026-01-01"

        with patch("video_processing.dedup.SQLAlchemyGeneratedVideoRepository") as MockRepo:
            mock_repo = MockRepo.return_value
            mock_repo._to_domain.side_effect = [existing1, existing2]
            query_mock = MagicMock()
            query_mock.filter.return_value = query_mock
            query_mock.order_by.return_value.limit.return_value.all.return_value = [
                mock_model1,
                mock_model2,
            ]
            session.query.return_value = query_mock
            rate = deduplicator.compute_duplicate_rate(fingerprint, "proj1", "vid1", session)

        # Issue #1658 融合公式：e1 distance=4 → 0.7*(1-4/64)+0.3 ≈ 0.95625 → 95.62；
        # e2 distance=1 → 0.7*(1-1/64)+0.3 = 0.9890625 → 98.91。取最大值 e2。
        assert rate == pytest.approx(98.91, abs=0.1)

    def test_user_id_scope_cross_project(self):
        """传 user_id 时应跨项目查询，而非仅当前项目."""
        from video_processing.dedup import VideoDeduplicator

        from packages.adapters.sqlalchemy_impl.models import GeneratedVideoModel

        deduplicator = VideoDeduplicator()
        fingerprint = self._make_fingerprint(md5="cross_proj_md5")
        session = MagicMock()

        # 模拟一个不同项目但同一用户的视频
        existing = self._make_existing_video(
            "existing_other_proj", {"md5": "cross_proj_md5", "keyframe_phashes": ["aa"]}
        )
        existing.project_id = "proj2"  # 不同项目
        existing.user_id = "user1"

        mock_model = MagicMock(spec=GeneratedVideoModel)
        mock_model.id = existing.id
        mock_model.project_id = existing.project_id
        mock_model.user_id = existing.user_id
        mock_model.video_fingerprint = existing.video_fingerprint
        mock_model.generated_at = "2026-01-01"

        with patch("video_processing.dedup.SQLAlchemyGeneratedVideoRepository") as MockRepo:
            mock_repo = MockRepo.return_value
            mock_repo._to_domain.return_value = existing

            query_mock = MagicMock()
            query_mock.filter.return_value = query_mock
            query_mock.order_by.return_value.limit.return_value.all.return_value = [mock_model]
            session.query.return_value = query_mock

            rate = deduplicator.compute_duplicate_rate(
                fingerprint,
                "proj1",
                "vid1",
                session,
                user_id="user1",
            )

        # 应通过 user_id 过滤，且匹配到跨项目视频
        assert rate == 100.0

    def test_user_id_empty_falls_back_to_project(self):
        """user_id 为空时应回退到 project_id 过滤."""
        from video_processing.dedup import VideoDeduplicator

        deduplicator = VideoDeduplicator()
        fingerprint = self._make_fingerprint()
        session = MagicMock()

        with patch("video_processing.dedup.SQLAlchemyGeneratedVideoRepository") as MockRepo:
            mock_repo = MockRepo.return_value
            query_mock = MagicMock()
            query_mock.filter.return_value = query_mock
            query_mock.order_by.return_value.limit.return_value.all.return_value = []
            session.query.return_value = query_mock

            rate = deduplicator.compute_duplicate_rate(
                fingerprint,
                "proj1",
                "vid1",
                session,
                user_id="",
            )

        assert rate == 0.0
        # 验证使用的是 project_id 过滤（回退路径）
        # 通过检查 filter 被调用时的参数来间接验证


class TestDuplicateRateAPI:
    """Test that duplicate_rate is returned in API responses."""

    def test_video_item_response_has_duplicate_rate(self):
        from app.schemas.video_center import VideoItemResponse

        resp = VideoItemResponse(
            id="v1",
            project_id="p1",
            generation_task_id="t1",
            name="test.mp4",
            file_url="https://example.com/test.mp4",
            file_size=1000,
            duration=10.0,
            width=1920,
            height=1080,
            fps=25.0,
            duplicate_rate=75.5,
        )
        assert resp.duplicate_rate == 75.5

    def test_video_item_response_duplicate_rate_default_none(self):
        from app.schemas.video_center import VideoItemResponse

        resp = VideoItemResponse(
            id="v1",
            project_id="p1",
            generation_task_id="t1",
            name="test.mp4",
            file_url="https://example.com/test.mp4",
            file_size=1000,
            duration=10.0,
            width=1920,
            height=1080,
            fps=25.0,
        )
        assert resp.duplicate_rate is None