fix: 修复 AI Code Review 阻塞级问题

1. assets.py: 修复高频排除回退逻辑 min(limit,1) → limit
2. assets.py: high_use_excluded 改用 set 提升查找效率
3. dedup.py: compute_duplicate_rate 限制查询最近100个视频避免OOM
4. 更新单元测试适配新的查询方式

apps/api/app/api/routes/assets.py

@@ -619,7 +619,7 @@ def smart_match_assets(
                     asset_ids=asset_ids,
                     recent_video_count=5,
                 )
-                high_use_excluded = []
+                high_use_excluded = set()
                 for r in filtered_results:
                     aid = getattr(r.asset, "id", "")
                     count = use_counts.get(aid, 0)
@@ -628,12 +628,12 @@ def smart_match_assets(
                             "smart-match 排除高频使用素材: asset_id=%s use_count=%d limit=%d",
                             aid, count, MAX_RECENT_USE_COUNT,
                         )
-                        high_use_excluded.append(r)
+                        high_use_excluded.add(id(r))
                     else:
                         pass
                 # 如果排除后不够 limit，放宽到不限制
-                remaining = [r for r in filtered_results if r not in high_use_excluded]
-                if len(remaining) >= min(request.limit, 1):
+                remaining = [r for r in filtered_results if id(r) not in high_use_excluded]
+                if len(remaining) >= request.limit:
                     filtered_results = remaining
                 else:
                     logger.info("smart-match 高频排除后素材不足(%d<%d)，保留全部", len(remaining), request.limit)

apps/worker/video_processing/dedup.py

@@ -332,8 +332,17 @@ class VideoDeduplicator:
         Returns:
             duplicate_rate: 0~100 的浮点数
         """
+        # 限制查询最近 100 个视频，避免大项目内存溢出
+        from packages.adapters.sqlalchemy_impl.models import GeneratedVideoModel
+        recent_models = (
+            session.query(GeneratedVideoModel)
+            .filter(GeneratedVideoModel.project_id == project_id)
+            .order_by(GeneratedVideoModel.generated_at.desc())
+            .limit(100)
+            .all()
+        )
         video_repo = SQLAlchemyGeneratedVideoRepository(session)
-        existing_videos = video_repo.list_by_project(project_id)
+        existing_videos = [video_repo._to_domain(m) for m in recent_models]
 
         max_similarity = 0.0
         for existing in existing_videos:

tests/unit/test_duplicate_rate.py

@@ -1,5 +1,4 @@
 """Tests for duplicate_rate computation and API response."""
-
 import json
 import sys
 from pathlib import Path
@@ -21,7 +20,6 @@ class TestComputeDuplicateRate:
 
     def _make_fingerprint(self, md5="abc123", phashes=None):
         from video_processing.dedup import VideoFingerprint
-
         return VideoFingerprint(
             md5=md5,
             keyframe_phashes=phashes or ["ff00ff00ff00ff00"],
@@ -32,7 +30,6 @@ class TestComputeDuplicateRate:
 
     def _make_existing_video(self, vid, fingerprint_dict):
         from packages.domain import GeneratedVideo
-
         return GeneratedVideo(
             id=vid,
             project_id="proj1",
@@ -49,37 +46,44 @@ class TestComputeDuplicateRate:
 
     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
-            mock_repo.list_by_project.return_value = []
+            session.query.return_value.filter.return_value.order_by.return_value.limit.return_value.all.return_value = []
             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"]})
+        # Create a mock model with the domain attributes
+        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.list_by_project.return_value = [existing]
+            mock_repo._to_domain.return_value = existing
+            # Mock the session.query chain
+            session.query.return_value.filter.return_value.order_by.return_value.limit.return_value.all.return_value = [mock_model]
             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()
@@ -88,10 +92,16 @@ class TestComputeDuplicateRate:
             "existing1",
             {"md5": "other_md5", "keyframe_phashes": ["ff00ff00ff00ff03"]},
         )
+        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.list_by_project.return_value = [existing]
+            mock_repo._to_domain.return_value = existing
+            session.query.return_value.filter.return_value.order_by.return_value.limit.return_value.all.return_value = [mock_model]
             rate = deduplicator.compute_duplicate_rate(fingerprint, "proj1", "vid1", session)
 
         # hamming distance = 2, similarity = (1 - 2/64) * 100 = 96.875
@@ -99,33 +109,54 @@ class TestComputeDuplicateRate:
 
     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.list_by_project.return_value = [self_video]
+            mock_repo._to_domain.return_value = self_video
+            session.query.return_value.filter.return_value.order_by.return_value.limit.return_value.all.return_value = [mock_model]
             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"]})
-        existing2 = self._make_existing_video("e2", {"md5": "md5_2", "keyframe_phashes": ["ff00ff00ff00ff01"]})
+        existing1 = self._make_existing_video(
+            "e1", {"md5": "md5_1", "keyframe_phashes": ["ff00ff00ff00ff0f"]}
+        )
+        existing2 = self._make_existing_video(
+            "e2", {"md5": "md5_2", "keyframe_phashes": ["ff00ff00ff00ff01"]}
+        )
+        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.list_by_project.return_value = [existing1, existing2]
+            mock_repo._to_domain.side_effect = [existing1, existing2]
+            session.query.return_value.filter.return_value.order_by.return_value.limit.return_value.all.return_value = [mock_model1, mock_model2]
             rate = deduplicator.compute_duplicate_rate(fingerprint, "proj1", "vid1", session)
 
         # max similarity: e2 distance=1, (1-1/64)*100 = 98.4375
@@ -137,35 +168,19 @@ class TestDuplicateRateAPI:
 
     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,
+            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,
+            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