feat(asset-dedup): 素材去重机制6项优化 — 调参+警告+贪心分配 (#1611)

Co-authored-by: xiaoxia <dev@xiaoxiajianji.com>
Co-committed-by: xiaoxia <dev@xiaoxiajianji.com>

apps/api/app/api/routes/templates_editor/clips.py

@@ -582,6 +582,47 @@ def _get_mediakit_recommendations(
         return {}
 
 
+def _calc_plan_internal_duplicate_rate(clips_data: list[dict]) -> float:
+    """估算单条成片内部重复率（%）.
+
+    检查本条成片中同一素材是否有重叠的时间区间。
+    重叠时长 / 成片总时长 * 100 = 内部重复率。
+    这是一个轻量估算，不依赖视频指纹；完整查重由 worker 异步完成。
+    """
+    if not clips_data:
+        return 0.0
+
+    # 按素材分组
+    by_asset: dict[str, list[tuple[float, float]]] = {}
+    total_duration = 0.0
+    for c in clips_data:
+        aid = c.get("asset_id", "")
+        if not aid:
+            continue
+        start = c.get("start_time", 0.0)
+        end = start + c.get("duration", 0.0)
+        by_asset.setdefault(aid, []).append((start, end))
+        total_duration += c.get("duration", 0.0)
+
+    if total_duration <= 0:
+        return 0.0
+
+    # 检查同素材内的区间重叠
+    overlap_duration = 0.0
+    for segments in by_asset.values():
+        if len(segments) < 2:
+            continue
+        segments_sorted = sorted(segments, key=lambda s: s[0])
+        last_end = segments_sorted[0][1]
+        for start, end in segments_sorted[1:]:
+            overlap = max(0.0, min(end, last_end) - start)
+            if overlap > 0:
+                overlap_duration += overlap
+            last_end = max(last_end, end)
+
+    return round(overlap_duration / total_duration * 100, 1)
+
+
 @router.post("/clips/from-assets", response_model=ClipsFromAssetsResponse)
 def create_clips_from_assets_editor(
     template_id: str,
@@ -662,19 +703,31 @@ def create_clips_from_assets_editor(
             return False
         return reused_durations.get(aid, 0.0) / assigned > REUSE_RATIO_LIMIT
 
+    # 素材耗尽标志：某轮循环中所有素材均被跳过时为 True
+    all_assets_exhausted = False
+
     for i, (_seg_order, dur_min, dur_max) in enumerate(segments):
         # 在 segment 的 duration_min ~ duration_max 之间随机取值（保留一位小数）
         raw_duration = random.uniform(dur_min, dur_max)
 
-        # 轮询分配素材：跳过时长缺失、复用占比已超 15% 阈值的素材；
+        # 贪心分配素材：按"已使用次数"升序排列候选素材（使用最少的优先），
+        # 同次数随机打散，避免"A-B-C-D"的固定组合反复出现。
+        # 跳过时长缺失、复用占比已超 10% 阈值的素材；
         # 选中后计算起点，若该素材可用区间耗尽且复用被闸门拒绝（calc 返回 None），
-        # 继续轮询下一个素材
+        # 继续尝试下一个素材
         asset_id = ""
         clip_duration = 0.0
         start_time: float | None = None
-        n_assets = len(asset_ids)
-        for offset in range(n_assets):
-            candidate = asset_ids[(i + offset) % n_assets]
+        # 动态按使用次数排序：优先选使用最少的素材，同次数随机打散
+        asset_use_counts = {
+            aid: len(used_segments.get(aid, []))
+            for aid in asset_ids
+        }
+        sorted_candidates = sorted(
+            asset_ids,
+            key=lambda aid: (asset_use_counts.get(aid, 0), random.random()),
+        )
+        for candidate in sorted_candidates:
             candidate_total = asset_durations.get(candidate, 0.0)
             if candidate_total <= 0:
                 continue
@@ -690,7 +743,7 @@ def create_clips_from_assets_editor(
                 continue
             # 随机起始时间（不调用 MediaKit，保证接口快速返回）；100 次避不开
             # 历史区间时走受控复用回调（复用片段累加 reused_durations，回调内部
-            # 预判复用后占比超 15% 则拒绝并返回 None）
+            # 预判复用后占比超 10% 则拒绝并返回 None）
             candidate_start = _calc_random_start_time(
                 candidate,
                 candidate_duration,
@@ -712,6 +765,7 @@ def create_clips_from_assets_editor(
 
         if not asset_id or start_time is None:
             # 所有素材时长缺失、复用占比超阈值，或区间耗尽且复用被拒 → 素材可切区间不足
+            all_assets_exhausted = True
             raise HTTPException(
                 status_code=status.HTTP_400_BAD_REQUEST,
                 detail="素材可切区间不足，请补充新素材",
@@ -754,11 +808,31 @@ def create_clips_from_assets_editor(
         unique_asset_ids,
     )
 
-    # 6. 立即返回响应
+    # 6. 估算成片内部重复率（本条成片中同一素材的重叠片段时长占比）
+    dup_rate = _calc_plan_internal_duplicate_rate(clips_data)
+    duplicate_warning = None
+    if dup_rate > 50:
+        duplicate_warning = f"查重率 {dup_rate:.1f}% 超过50%，建议更换素材或模板"
+        logger.warning(
+            "from-assets 成片查重率超标: plan_id=%s dup_rate=%.1f%%",
+            plan_id, dup_rate,
+        )
+
+    # 7. 素材耗尽提示
+    exhaustion_warning = None
+    if all_assets_exhausted and created_count < len(segments):
+        exhaustion_warning = (
+            "素材可切区间不足，部分片段使用了复用素材。"
+            "建议：1) 补充更多素材到素材库 2) 使用不同的素材组合生成"
+        )
+
+    # 8. 立即返回响应
     return ClipsFromAssetsResponse(
         created_count=created_count,
         plan_id=plan_id,
         clip_ids=[],
+        duplicate_warning=duplicate_warning,
+        exhaustion_warning=exhaustion_warning,
     )
 
 

apps/api/app/api/routes/templates_editor/schemas.py

@@ -189,6 +189,8 @@ class ClipsFromAssetsResponse(BaseModel):
     plan_id: str = ""
     message: str = ""
     clip_ids: List[str] = Field(default_factory=list, description="创建的片段ID列表")
+    duplicate_warning: Optional[str] = Field(default=None, description="查重率超标警告")
+    exhaustion_warning: Optional[str] = Field(default=None, description="素材耗尽警告")
 
 
 # ── 封面配置 ────────────────────────────────────────────────────────────────