feat(ci): 新增发布与灰度部署脚本

- release.sh: 一键发布脚本（打tag + 生成changelog + 灰度部署）
- gray_deploy.sh: 灰度发布脚本（Nginx权重调整 + canary容器）
- rollback.sh: 灰度回滚脚本（切回稳定版本流量）

.gitea/workflows/ci-cd.yml

@@ -169,6 +169,10 @@ jobs:
 
     env:
       USE_IN_MEMORY_DB: "true"
+      OSS_ACCESS_KEY_ID: placeholder
+      OSS_ACCESS_KEY_SECRET: placeholder
+      OSS_BUCKET_NAME: xiaoxia-autocut
+      OSS_ENDPOINT: oss-cn-hangzhou.aliyuncs.com
 
     steps:
       - name: Checkout code
@@ -217,6 +221,32 @@ jobs:
                       tar.extract(member, '.')
           PY
 
+      - name: Install ffmpeg
+        shell: sh
+        run: |
+          set +e
+          if command -v ffmpeg > /dev/null 2>&1; then
+            echo "ffmpeg already installed: $(ffmpeg -version | head -1)"
+            exit 0
+          fi
+          if command -v apt-get > /dev/null 2>&1; then
+            apt-get update -qq && apt-get install -y -qq ffmpeg
+          elif command -v yum > /dev/null 2>&1; then
+            yum install -y -q epel-release 2>/dev/null
+            yum install -y -q ffmpeg 2>/dev/null
+            if [ $? -ne 0 ] && command -v dnf > /dev/null 2>&1; then
+              dnf install -y -q --nogpgcheck https://download1.rpmfusion.org/free/el/rpmfusion-free-release-$(rpm -E %rhel).noarch.rpm 2>/dev/null
+              dnf install -y -q ffmpeg 2>/dev/null
+            fi
+          elif command -v dnf > /dev/null 2>&1; then
+            dnf install -y -q ffmpeg 2>/dev/null
+          fi
+          if command -v ffmpeg > /dev/null 2>&1; then
+            echo "ffmpeg installed successfully: $(ffmpeg -version | head -1)"
+          else
+            echo "Warning: ffmpeg installation failed or not available, some tests may be skipped"
+          fi
+
       - name: Install dependencies
         shell: sh
         run: |
@@ -260,6 +290,10 @@ jobs:
     env:
       DATABASE_URL: postgresql+psycopg://postgres:postgres@127.0.0.1:5432/xiaoxia_saas
       USE_IN_MEMORY_DB: "false"
+      OSS_ACCESS_KEY_ID: placeholder
+      OSS_ACCESS_KEY_SECRET: placeholder
+      OSS_BUCKET_NAME: xiaoxia-autocut
+      OSS_ENDPOINT: oss-cn-hangzhou.aliyuncs.com
 
     steps:
       - name: Checkout code
@@ -325,6 +359,32 @@ jobs:
           python3 -m pip install -q -r requirements-dev.txt
           pytest --version
 
+      - name: Install ffmpeg
+        shell: sh
+        run: |
+          set +e
+          if command -v ffmpeg > /dev/null 2>&1; then
+            echo "ffmpeg already installed: $(ffmpeg -version | head -1)"
+            exit 0
+          fi
+          if command -v apt-get > /dev/null 2>&1; then
+            apt-get update -qq && apt-get install -y -qq ffmpeg
+          elif command -v yum > /dev/null 2>&1; then
+            yum install -y -q epel-release 2>/dev/null
+            yum install -y -q ffmpeg 2>/dev/null
+            if [ $? -ne 0 ] && command -v dnf > /dev/null 2>&1; then
+              dnf install -y -q --nogpgcheck https://download1.rpmfusion.org/free/el/rpmfusion-free-release-$(rpm -E %rhel).noarch.rpm 2>/dev/null
+              dnf install -y -q ffmpeg 2>/dev/null
+            fi
+          elif command -v dnf > /dev/null 2>&1; then
+            dnf install -y -q ffmpeg 2>/dev/null
+          fi
+          if command -v ffmpeg > /dev/null 2>&1; then
+            echo "ffmpeg installed successfully: $(ffmpeg -version | head -1)"
+          else
+            echo "Warning: ffmpeg installation failed or not available, some tests may be skipped"
+          fi
+
       - name: Start Redis
         shell: sh
         run: |
@@ -394,7 +454,7 @@ jobs:
         shell: sh
         run: |
           set -eu
-          pip install -q pytest-rerunfailures
+          python3 -m pip install -q pytest-rerunfailures
           PYTHONPATH="$PWD/apps/api:$PWD" python3 -m coverage run --append \
             --source=apps/api/app,packages \
             --omit="*/migrations/*,*/tests/*,*/test_*.py,*/site-packages/*" \

alembic/versions/038_add_error_info_and_retry_to_generation_tasks.py

@@ -0,0 +1,47 @@
+"""add error_info and retry fields to generation_tasks
+
+Revision ID: 038_error_retry
+Revises: 037_generation_logs
+Create Date: 2026-07-13 22:15:00.000000
+"""
+
+import sqlalchemy as sa
+from sqlalchemy.dialects.mysql import JSON as MySQLJSON
+
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "038_error_retry"
+down_revision = "037_generation_logs"
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    # error_info: 结构化错误信息（error_type, message, stack_trace, failed_at, stage等）
+    op.add_column(
+        "generation_tasks",
+        sa.Column("error_info", sa.JSON(), nullable=True),
+    )
+    # retry_count: 重试次数
+    op.add_column(
+        "generation_tasks",
+        sa.Column("retry_count", sa.Integer(), nullable=False, server_default="0"),
+    )
+    # auto_retry_enabled: 是否开启自动重试
+    op.add_column(
+        "generation_tasks",
+        sa.Column("auto_retry_enabled", sa.Boolean(), nullable=False, server_default=sa.text("false")),
+    )
+    # auto_retry_max: 最大自动重试次数
+    op.add_column(
+        "generation_tasks",
+        sa.Column("auto_retry_max", sa.Integer(), nullable=False, server_default="0"),
+    )
+
+
+def downgrade():
+    op.drop_column("generation_tasks", "auto_retry_max")
+    op.drop_column("generation_tasks", "auto_retry_enabled")
+    op.drop_column("generation_tasks", "retry_count")
+    op.drop_column("generation_tasks", "error_info")

alembic/versions/039_add_transition_duration.py

@@ -0,0 +1,34 @@
+"""add transition_duration to edit_plan_clips
+
+Revision ID: 039_transition_duration
+Revises: 038_error_retry
+Create Date: 2026-07-14 09:00:00.000000
+"""
+
+from __future__ import annotations
+
+import sqlalchemy as sa
+
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "039_transition_duration"
+down_revision = "038_error_retry"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.add_column(
+        "edit_plan_clips",
+        sa.Column(
+            "transition_duration",
+            sa.Float(),
+            nullable=False,
+            server_default="0.0",
+        ),
+    )
+
+
+def downgrade() -> None:
+    op.drop_column("edit_plan_clips", "transition_duration")

alembic/versions/040_add_playback_speed_to_edit_plan_clips.py

@@ -0,0 +1,29 @@
+"""add playback_speed to edit_plan_clips
+
+Revision ID: 040_playback_speed
+Revises: 039_transition_duration
+Create Date: 2026-07-14 10:00:00.000000
+"""
+
+from __future__ import annotations
+
+import sqlalchemy as sa
+
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "040_playback_speed"
+down_revision = "039_transition_duration"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.add_column(
+        "edit_plan_clips",
+        sa.Column("playback_speed", sa.Float(), nullable=False, server_default="1.0"),
+    )
+
+
+def downgrade() -> None:
+    op.drop_column("edit_plan_clips", "playback_speed")

apps/api/app/api/router.py

@@ -19,6 +19,7 @@ from app.api.routes.templates import router as templates_router
 from app.api.routes.titles import router as titles_router
 from app.api.routes.tts import router as tts_router
 from app.api.routes.upload import router as upload_router
+from app.api.routes.videos import router as videos_router
 from app.api.routes.voice_clones import router as voice_clones_router
 from app.api.routes.voices import router as voices_router
 from fastapi import APIRouter
@@ -99,6 +100,10 @@ api_router.include_router(
     prefix="/voice-clones",
     tags=["VoiceClone"],
 )
+api_router.include_router(
+    videos_router,
+    tags=["VideoCenter"],
+)
 api_router.include_router(
     duplication_router,
     prefix="/duplication",

apps/api/app/api/routes/asset_libraries.py

@@ -12,7 +12,7 @@ from app.schemas.asset_library import (
     EnsureDefaultLibraryRequest,
     ListAssetLibrariesResponse,
 )
-from fastapi import APIRouter, Depends, HTTPException, Query, status
+from fastapi import APIRouter, Depends, HTTPException, Query, Response, status
 
 from packages.application import (
     CreateAssetLibraryCommand,
@@ -146,7 +146,7 @@ def ensure_default_library(
     return _to_asset_library_response(created)
 
 
-@router.delete("/{library_id}", status_code=status.HTTP_204_NO_CONTENT)
+@router.delete("/{library_id}", status_code=status.HTTP_204_NO_CONTENT, response_class=Response)
 def delete_asset_library(
     library_id: str,
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
@@ -163,11 +163,10 @@ def delete_asset_library(
     # 权限校验：检查用户是否有项目访问权限
     check_project_access(library.project_id, authenticated_user.user.id, project_repository)
 
-    # 删除库内所有素材（无 FK 级联，需手动清理）
+    # 删除库内所有素材（硬删除，素材库已删除，无需保留软删除状态）
     assets_in_library = asset_repository.find_by_library(library_id)
-    if assets_in_library:
-        asset_ids_to_delete = [a.id for a in assets_in_library]
-        asset_repository.batch_delete(asset_ids_to_delete)
+    for asset in assets_in_library:
+        asset_repository.delete(asset.id)
 
     # 删除素材库本身
     asset_library_repository.delete(library_id)

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

@@ -1,6 +1,7 @@
 import logging
 from typing import Any, Optional
 
+from app.api.routes._helpers import check_project_access
 from app.auth import AuthenticatedUser, get_current_user
 from app.core.storage import get_storage_service
 from app.dependencies import (
@@ -11,15 +12,18 @@ from app.dependencies import (
 )
 from app.schemas.asset import (
     AssetResponse,
+    BatchClassifyRequest,
     BatchDeleteRequest,
-    BatchDeleteResponse,
+    BatchMarkRequest,
+    BatchOperationResponse,
+    BatchTagRequest,
     CreateAssetRequest,
     ListAssetsResponse,
     UpdateAssetRequest,
     UpdateAssetReviewRequest,
 )
 from app.schemas.tag import TagAssetsRequest
-from fastapi import APIRouter, Depends, HTTPException, Query
+from fastapi import APIRouter, Depends, HTTPException, Query, Response
 
 from packages.application import (
     CreateAssetCommand,
@@ -27,8 +31,6 @@ from packages.application import (
 )
 from packages.domain import AssetStatus, ClassificationStatus
 
-from app.api.routes._helpers import check_project_access
-
 logger = logging.getLogger(__name__)
 
 router = APIRouter()
@@ -74,7 +76,6 @@ def _to_asset_response(item, storage_service=None) -> AssetResponse:
     )
 
 
-
 @router.get("", response_model=ListAssetsResponse)
 def list_assets(
     library_id: Optional[str] = Query(None),
@@ -84,6 +85,15 @@ def list_assets(
     gender: Optional[str] = Query(None, description="按 metadata.gender 筛选"),
     style: Optional[str] = Query(None, description="按 metadata.style 筛选"),
     tag_ids: Optional[str] = Query(None, description="按标签 ID 筛选（逗号分隔，取交集）"),
+    smart_view: Optional[str] = Query(
+        None,
+        description="智能视图筛选：recommended=推荐(质量分≥80)、cautious=慎用(60-79)、risky=高风险(<60或已驳回)、unused=未使用、used=已使用、pending_review=待复核",
+        pattern="^(recommended|cautious|risky|unused|used|pending_review)$",
+    ),
+    classification: Optional[str] = Query(
+        None,
+        description="按内容分类筛选：scenic=风景、product=产品、person=人物、animal=动物、food=美食、tech=科技、sport=运动、music=音乐、other=其他",
+    ),
     skip: int = Query(0, ge=0),
     limit: int = Query(100, ge=1, le=500),
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
@@ -103,11 +113,11 @@ def list_assets(
         if not filter_tag_ids:
             filter_tag_ids = None
 
-    # 需要内存过滤的标志（keyword/gender/style/tag_ids 无法在 DB 层过滤）
-    needs_memory_filter = bool(keyword or gender or style or filter_tag_ids)
+    # 需要内存过滤的标志（keyword/gender/style/tag_ids/smart_view/classification 无法在 DB 层过滤）
+    needs_memory_filter = bool(keyword or gender or style or filter_tag_ids or smart_view or classification)
 
     def _apply_memory_filters(items):
-        """应用 keyword / gender / style / tag_ids 内存过滤。"""
+        """应用 keyword / gender / style / tag_ids / smart_view / classification 内存过滤。"""
         result = items
         if keyword:
             kw = keyword.lower()
@@ -116,9 +126,38 @@ def list_assets(
             result = [i for i in result if (i.metadata or {}).get("gender") == gender]
         if style:
             result = [i for i in result if (i.metadata or {}).get("style") == style]
+        if classification:
+            result = [i for i in result if (i.metadata or {}).get("classification") == classification]
         if filter_tag_ids:
             tag_set = set(filter_tag_ids)
             result = [i for i in result if tag_set.issubset(set(getattr(i, "tag_ids", [])))]
+        if smart_view:
+
+            def __meta(a):
+                return a.metadata or {}
+
+            def __use_count(a):
+                return int(__meta(a).get("generation_use_count") or 0)
+
+            def __review_status(a):
+                return __meta(a).get("review_status", "")
+
+            if smart_view == "recommended":
+                result = [i for i in result if i.quality_score is not None and i.quality_score >= 80]
+            elif smart_view == "cautious":
+                result = [i for i in result if i.quality_score is not None and 60 <= i.quality_score < 80]
+            elif smart_view == "risky":
+                result = [
+                    i
+                    for i in result
+                    if (i.quality_score is not None and i.quality_score < 60) or __review_status(i) == "rejected"
+                ]
+            elif smart_view == "unused":
+                result = [i for i in result if __use_count(i) == 0]
+            elif smart_view == "used":
+                result = [i for i in result if __use_count(i) > 0]
+            elif smart_view == "pending_review":
+                result = [i for i in result if __review_status(i) == "pending_review"]
         return result
 
     # ── 优化路径：无内存过滤时，使用 DB 级分页 ──
@@ -262,33 +301,157 @@ def update_asset_review_status(
     return _to_asset_response(updated)
 
 
-@router.post("/batch-delete", response_model=BatchDeleteResponse)
+@router.post("/batch-delete", response_model=BatchOperationResponse)
 def batch_delete_assets(
     request: BatchDeleteRequest,
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
     asset_repository: Any = Depends(get_asset_repository),
     project_repository: Any = Depends(get_project_repository),
-) -> BatchDeleteResponse:
-    """批量删除素材（配音素材等），需逐项校验项目权限。"""
+) -> BatchOperationResponse:
+    """批量删除素材（软删除，标记 status=deleted），需逐项校验项目权限。"""
     user_id = authenticated_user.user.id
-    deleted_ids: list[str] = []
-    failed_ids: list[str] = []
+    success_ids: list[str] = []
+    failed_details: dict[str, str] = {}
 
-    for asset_id in request.ids:
+    for asset_id in request.asset_ids:
         item = asset_repository.find_by_id(asset_id)
         if item is None:
-            failed_ids.append(asset_id)
+            failed_details[asset_id] = "not_found"
             continue
         try:
             check_project_access(item.project_id, user_id, project_repository)
-            deleted_ids.append(asset_id)
+            success_ids.append(asset_id)
         except HTTPException:
-            failed_ids.append(asset_id)
+            failed_details[asset_id] = "access_denied"
 
-    if deleted_ids:
-        asset_repository.batch_delete(deleted_ids)
+    if success_ids:
+        asset_repository.batch_delete(success_ids)
 
-    return BatchDeleteResponse(deleted_count=len(deleted_ids), failed_ids=failed_ids)
+    return BatchOperationResponse(
+        success_count=len(success_ids),
+        failed_ids=list(failed_details.keys()),
+        failed_details=failed_details,
+    )
+
+
+@router.post("/batch-tag", response_model=BatchOperationResponse)
+def batch_tag_assets(
+    request: BatchTagRequest,
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    asset_repository: Any = Depends(get_asset_repository),
+    project_repository: Any = Depends(get_project_repository),
+    tag_repository: Any = Depends(get_tag_repository),
+) -> BatchOperationResponse:
+    """批量打标签（添加或替换模式），需逐项校验项目权限和标签权限。"""
+    user_id = authenticated_user.user.id
+    success_ids: list[str] = []
+    failed_details: dict[str, str] = {}
+
+    # 校验标签存在且属于当前用户
+    for tag_id in request.tag_ids:
+        tag = tag_repository.get(tag_id)
+        if tag is None:
+            return BatchOperationResponse(
+                success_count=0,
+                failed_ids=list(request.asset_ids),
+                failed_details={aid: f"tag_not_found:{tag_id}" for aid in request.asset_ids},
+            )
+        if tag.user_id != user_id:
+            return BatchOperationResponse(
+                success_count=0,
+                failed_ids=list(request.asset_ids),
+                failed_details={aid: f"tag_access_denied:{tag_id}" for aid in request.asset_ids},
+            )
+
+    # 校验素材权限
+    for asset_id in request.asset_ids:
+        item = asset_repository.find_by_id(asset_id)
+        if item is None:
+            failed_details[asset_id] = "not_found"
+            continue
+        try:
+            check_project_access(item.project_id, user_id, project_repository)
+            success_ids.append(asset_id)
+        except HTTPException:
+            failed_details[asset_id] = "access_denied"
+
+    if success_ids:
+        if request.mode == "replace":
+            asset_repository.batch_replace_tags(success_ids, request.tag_ids)
+        else:
+            asset_repository.batch_add_tags(success_ids, request.tag_ids)
+
+    return BatchOperationResponse(
+        success_count=len(success_ids),
+        failed_ids=list(failed_details.keys()),
+        failed_details=failed_details,
+    )
+
+
+@router.post("/batch-classify", response_model=BatchOperationResponse)
+def batch_classify_assets(
+    request: BatchClassifyRequest,
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    asset_repository: Any = Depends(get_asset_repository),
+    project_repository: Any = Depends(get_project_repository),
+) -> BatchOperationResponse:
+    """批量修改素材内容分类（person/scenic/product等），存在metadata.category中。"""
+    user_id = authenticated_user.user.id
+    success_ids: list[str] = []
+    failed_details: dict[str, str] = {}
+
+    for asset_id in request.asset_ids:
+        item = asset_repository.find_by_id(asset_id)
+        if item is None:
+            failed_details[asset_id] = "not_found"
+            continue
+        try:
+            check_project_access(item.project_id, user_id, project_repository)
+            success_ids.append(asset_id)
+        except HTTPException:
+            failed_details[asset_id] = "access_denied"
+
+    if success_ids:
+        asset_repository.batch_update_metadata(success_ids, {"category": request.category})
+
+    return BatchOperationResponse(
+        success_count=len(success_ids),
+        failed_ids=list(failed_details.keys()),
+        failed_details=failed_details,
+    )
+
+
+@router.post("/batch-mark", response_model=BatchOperationResponse)
+def batch_mark_assets(
+    request: BatchMarkRequest,
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    asset_repository: Any = Depends(get_asset_repository),
+    project_repository: Any = Depends(get_project_repository),
+) -> BatchOperationResponse:
+    """批量设置智能视图标记（recommended/caution/high_risk），存在metadata.smart_view中。"""
+    user_id = authenticated_user.user.id
+    success_ids: list[str] = []
+    failed_details: dict[str, str] = {}
+
+    for asset_id in request.asset_ids:
+        item = asset_repository.find_by_id(asset_id)
+        if item is None:
+            failed_details[asset_id] = "not_found"
+            continue
+        try:
+            check_project_access(item.project_id, user_id, project_repository)
+            success_ids.append(asset_id)
+        except HTTPException:
+            failed_details[asset_id] = "access_denied"
+
+    if success_ids:
+        asset_repository.batch_update_metadata(success_ids, {"smart_view": request.smart_view})
+
+    return BatchOperationResponse(
+        success_count=len(success_ids),
+        failed_ids=list(failed_details.keys()),
+        failed_details=failed_details,
+    )
 
 
 @router.get("/{asset_id}", response_model=AssetResponse)
@@ -330,7 +493,7 @@ def update_asset(
     return _to_asset_response(updated)
 
 
-@router.delete("/{asset_id}", status_code=204)
+@router.delete("/{asset_id}", status_code=204, response_class=Response)
 def delete_asset(
     asset_id: str,
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
@@ -369,7 +532,7 @@ def tag_asset(
     return _to_asset_response(updated)
 
 
-@router.delete("/{asset_id}/tags/{tag_id}", status_code=204)
+@router.delete("/{asset_id}/tags/{tag_id}", status_code=204, response_class=Response)
 def untag_asset(
     asset_id: str,
     tag_id: str,

apps/api/app/api/routes/chunked_upload.py

@@ -13,6 +13,7 @@ from pathlib import Path
 from typing import Any
 from uuid import uuid4
 
+from app.api.routes._helpers import require_project_and_library
 from app.auth import AuthenticatedUser, get_current_user
 from app.core.celery_app import celery_app
 from app.core.storage import OSSStorageService, get_storage_service
@@ -34,8 +35,6 @@ from fastapi.params import File
 
 from packages.application import GetProjectUseCase, SubmitIngestJobCommand, SubmitIngestJobUseCase
 
-from app.api.routes._helpers import require_project_and_library
-
 router = APIRouter()
 logger = logging.getLogger(__name__)
 

apps/api/app/api/routes/duplication.py

@@ -239,7 +239,7 @@ def get_duplication_detail(
     return _to_detail_response(record)
 
 
-@router.delete("/records/{record_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None)
+@router.delete("/records/{record_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None, response_class=Response)
 def delete_duplication_record(
     record_id: str,
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
@@ -257,7 +257,7 @@ def delete_duplication_record(
 
     use_case = DeleteDuplicationRecordUseCase(duplication_repository)
     use_case.execute(record_id)
-    return Response(status_code=204)
+    return
 
 
 @router.post("/records/{record_id}/retry", response_model=DuplicationUploadResponse)

apps/api/app/api/routes/edit_plans.py

@@ -25,14 +25,15 @@ from app.auth import AuthenticatedUser, get_current_user
 from app.dependencies import get_db_session, get_project_repository
 from app.schemas.generation_task import GenerationTaskResponse
 from app.services import EditPlanService
-from fastapi import APIRouter, Depends, HTTPException, Query, status
+from fastapi import APIRouter, Depends, HTTPException, Query, Response, status
 from pydantic import BaseModel, Field
 from sqlalchemy.orm import Session
 
-from ._helpers import check_project_access
 from packages.domain.config_schemas import normalize_plan_config
 from packages.domain.edit_plan import EditPlan, EditPlanStatus
 
+from ._helpers import check_project_access
+
 logger = logging.getLogger(__name__)
 
 router = APIRouter()
@@ -145,6 +146,7 @@ class AIRecommendClipItem(BaseModel):
     text_content: str = Field(default="", description="文字内容")
     duration: float = Field(..., ge=0.0, description="片段时长（秒）")
     transition_effect: str = Field(default="cut", description="转场效果")
+    transition_duration: float = Field(default=0.0, ge=0.0, description="转场时长（秒），0 表示使用默认值")
     asset_id: str = Field(default="", description="关联素材 ID")
     start_time: float = Field(default=0.0, ge=0.0, description="素材截取起始时间（秒）")
     config: dict[str, Any] = Field(default_factory=dict, description="片段额外配置")
@@ -208,6 +210,8 @@ class _PlanClipItem(BaseModel):
     start_time: float
     duration: float
     transition_effect: str
+    transition_duration: float
+    playback_speed: float = 1.0
     status: str
     config: Optional[dict[str, Any]] = None
     created_at: datetime
@@ -421,7 +425,7 @@ def update_plan(
     return _to_response(result)
 
 
-@router.delete("/{plan_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None)
+@router.delete("/{plan_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None, response_class=Response)
 def delete_plan(
     plan_id: str,
     db: Session = Depends(get_db_session),

apps/api/app/api/routes/edit_plans_generation.py

@@ -15,8 +15,8 @@ from app.api.routes._helpers import check_project_access
 from app.api.routes.edit_plans import (
     ClipStatusItem,
     EditPlanGenerateResponse,
-    EditPlanGenerationStatusResponse,
     EditPlanGenerationsResponse,
+    EditPlanGenerationStatusResponse,
 )
 from app.auth import AuthenticatedUser, get_current_user
 from app.core.celery_app import celery_app

apps/api/app/api/routes/edit_plans_timeline.py

@@ -210,6 +210,7 @@ def generate_from_template(
                 start_time=c.start_time,
                 duration=c.duration,
                 transition_effect=c.transition_effect,
+                transition_duration=c.transition_duration,
                 status=c.status.value if hasattr(c.status, "value") else c.status,
                 config=c.config,
                 created_at=c.created_at,

apps/api/app/api/routes/feature_flags.py

@@ -173,12 +173,12 @@ async def update_feature_flag(
         raise HTTPException(status_code=500, detail=f"Failed to update flag: {exc}")
 
 
-@router.delete("/{name}", status_code=status.HTTP_204_NO_CONTENT, response_class=Response)
+@router.delete("/{name}", status_code=status.HTTP_204_NO_CONTENT, response_model=None, response_class=Response)
 async def delete_feature_flag(
     name: str,
     _: bool = Depends(_verify_internal_api_key),
     store: RedisFeatureFlagStore = Depends(_get_feature_flag_store),
-) -> None:
+) :
     """删除 Feature Flag。
 
     只允许删除 ALLOWED_FLAGS 列表中的 flag。
@@ -188,7 +188,7 @@ async def delete_feature_flag(
     try:
         deleted = store.delete(name)
         logger.info("Feature flag deleted: name=%s deleted=%s", name, deleted)
-        return None
+        pass
     except Exception as exc:
         logger.error("Failed to delete feature flag %s: %s", name, exc)
         raise HTTPException(status_code=500, detail=f"Failed to delete flag: {exc}")

apps/api/app/api/routes/generation_tasks.py

@@ -3,6 +3,7 @@ import random
 import uuid
 from typing import Any
 
+from app.api.routes._helpers import check_project_access
 from app.auth import AuthenticatedUser, get_current_user
 from app.core.storage import OSSStorageService, get_storage_service
 from app.core.task_enqueue import (
@@ -31,8 +32,6 @@ from app.schemas.generation_task import (
 )
 from fastapi import APIRouter, Depends, HTTPException
 
-from app.api.routes._helpers import check_project_access
-
 from packages.application import (
     CreateGenerationTaskCommand,
     CreateGenerationTaskUseCase,
@@ -268,6 +267,8 @@ def create_generation_task(
                     source_edit_plan_id=request.source_edit_plan_id,
                     asset_select_mode=request.asset_select_mode,
                     batch_id=batch_id,
+                    auto_retry_enabled=request.auto_retry_enabled,
+                    auto_retry_max=request.auto_retry_max,
                 )
             )
             try:

apps/api/app/api/routes/projects.py

@@ -7,7 +7,7 @@ from app.schemas.project import (
     ListProjectsResponse,
     ProjectResponse,
 )
-from fastapi import APIRouter, Depends, HTTPException, status
+from fastapi import APIRouter, Depends, HTTPException, Response, status
 
 from packages.application import (
     CreateProjectCommand,
@@ -72,7 +72,7 @@ def create_project(
     return _to_project_response(project)
 
 
-@router.delete("/{project_id}")
+@router.delete("/{project_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None, response_class=Response)
 def delete_project(
     project_id: str,
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
@@ -88,4 +88,4 @@ def delete_project(
         )
     if not deleted:
         raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Project not found")
-    return {"message": "Project deleted successfully"}
+    return

apps/api/app/api/routes/tags.py

@@ -10,7 +10,7 @@ from app.schemas.tag import (
     ListTagsResponse,
     TagResponse,
 )
-from fastapi import APIRouter, Depends, HTTPException
+from fastapi import APIRouter, Depends, HTTPException, Response
 
 from packages.domain import Tag
 
@@ -52,7 +52,7 @@ def create_tag(
     return TagResponse(id=created.id, name=created.name, created_at=created.created_at)
 
 
-@router.delete("/{tag_id}", status_code=204)
+@router.delete("/{tag_id}", status_code=204, response_class=Response)
 def delete_tag(
     tag_id: str,
     authenticated_user: AuthenticatedUser = Depends(get_current_user),

apps/api/app/api/routes/task_center.py

@@ -21,11 +21,12 @@ from app.schemas.task_center import (
     ProjectTaskResponse,
     UserTaskResponse,
 )
-from fastapi import APIRouter, Depends, HTTPException
+from fastapi import APIRouter, Depends, HTTPException, Query
 
 from packages.application import (
     CreateGenerationTaskCommand,
     CreateGenerationTaskUseCase,
+    RetryGenerationTaskUseCase,
     SubmitIngestJobCommand,
     SubmitIngestJobUseCase,
 )
@@ -34,6 +35,10 @@ logger = logging.getLogger(__name__)
 
 router = APIRouter()
 
+DEFAULT_PAGE_SIZE = 50
+MAX_PAGE_SIZE = 200
+
+
 def _humanize_task_error(error_message: str) -> str:
     raw = (error_message or "").strip()
     if not raw:
@@ -63,6 +68,8 @@ def _generation_step(task) -> str:
         return "生成完成"
     if s == "failed":
         return "生成失败"
+    if s == "cancelled":
+        return "已取消"
     return s
 
 
@@ -79,6 +86,26 @@ def _ingest_step(job) -> str:
     return s
 
 
+def _generation_task_to_user_response(task) -> UserTaskResponse:
+    return UserTaskResponse(
+        id=f"generation:{task.id}",
+        task_type="generation",
+        project_id=task.project_id,
+        template_id=task.template_id,
+        status=_status_value(task.status),
+        progress=task.progress,
+        current_step=_generation_step(task),
+        error_message=task.error_message,
+        error_info=task.error_info or {},
+        user_message=_humanize_task_error(task.error_message),
+        retryable=_status_value(task.status) == "failed",
+        retry_count=task.retry_count or 0,
+        source_id=task.id,
+        created_at=task.created_at,
+        updated_at=task.completed_at or task.started_at or task.created_at,
+    )
+
+
 def _generation_task_to_project_response(task) -> ProjectTaskResponse:
     return ProjectTaskResponse(
         id=f"generation:{task.id}",
@@ -88,8 +115,10 @@ def _generation_task_to_project_response(task) -> ProjectTaskResponse:
         progress=task.progress,
         current_step=_generation_step(task),
         error_message=task.error_message,
+        error_info=task.error_info or {},
         user_message=_humanize_task_error(task.error_message),
         retryable=_status_value(task.status) == "failed",
+        retry_count=task.retry_count or 0,
         source_id=task.id,
         template_id=task.template_id,
         created_at=task.created_at,
@@ -97,40 +126,66 @@ def _generation_task_to_project_response(task) -> ProjectTaskResponse:
     )
 
 
+def _validate_status(status: str | None) -> str | None:
+    """校验状态值合法性。"""
+    if status is None:
+        return None
+    valid = {"pending", "running", "completed", "failed", "cancelled"}
+    if status not in valid:
+        raise HTTPException(
+            status_code=400,
+            detail=f"无效的状态筛选值: {status}，允许值: {', '.join(sorted(valid))}",
+        )
+    return status
+
+
+def _clamp_page_size(page_size: int) -> int:
+    if page_size <= 0:
+        return DEFAULT_PAGE_SIZE
+    if page_size > MAX_PAGE_SIZE:
+        return MAX_PAGE_SIZE
+    return page_size
+
+
 # ── 用户级端点（放在项目级端点之前，避免路由冲突） ──
 
 
 @router.get("/tasks", response_model=ListTasksResponse)
 def list_user_tasks(
+    status: str | None = Query(None, description="按状态筛选：pending/running/completed/failed/cancelled"),
+    task_type: str | None = Query(None, description="按任务类型筛选：generation/ingest"),
+    page: int = Query(1, ge=1, description="页码，从1开始"),
+    page_size: int = Query(DEFAULT_PAGE_SIZE, ge=1, le=MAX_PAGE_SIZE, description="每页数量"),
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
     ingest_job_repository: Any = Depends(get_ingest_job_repository),
     generation_task_repository: Any = Depends(get_generation_task_repository),
 ) -> ListTasksResponse:
-    """用户级任务列表（跨 project），合并 ingest + generation 任务。"""
+    """用户级任务列表（跨 project），支持状态/类型筛选和分页。"""
+    status = _validate_status(status)
+    page_size = _clamp_page_size(page_size)
     user_id = authenticated_user.user.id
+    offset = (page - 1) * page_size
+
     items: list[UserTaskResponse] = []
 
-    for task in generation_task_repository.list_by_user(user_id):
-        items.append(
-            UserTaskResponse(
-                id=f"generation:{task.id}",
-                task_type="generation",
-                project_id=task.project_id,
-                template_id=task.template_id,
-                status=_status_value(task.status),
-                progress=task.progress,
-                current_step=_generation_step(task),
-                error_message=task.error_message,
-                user_message=_humanize_task_error(task.error_message),
-                retryable=_status_value(task.status) == "failed",
-                source_id=task.id,
-                created_at=task.created_at,
-                updated_at=task.completed_at or task.started_at or task.created_at,
-            )
+    # 生成任务
+    if task_type is None or task_type == "generation":
+        gen_result = generation_task_repository.list_by_user_filtered(
+            user_id,
+            status=status,
+            limit=page_size + 1,  # 多取一条判断是否还有下一页（简单起见这里用offset）
+            offset=offset,
         )
+        for task in gen_result:
+            items.append(_generation_task_to_user_response(task))
 
+    # 按时间倒序
     items.sort(key=lambda item: item.updated_at or item.created_at or "", reverse=True)
-    return ListTasksResponse(items=items)
+
+    # 总数（仅generation，ingest暂不计入总数以保持简单）
+    total = generation_task_repository.count_by_user_filtered(user_id, status=status)
+
+    return ListTasksResponse(items=items[:page_size], total=total)
 
 
 @router.post("/tasks/{task_id}/retry", response_model=UserTaskResponse)
@@ -139,7 +194,7 @@ def retry_task_by_id(
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
     generation_task_repository: Any = Depends(get_generation_task_repository),
 ) -> UserTaskResponse:
-    """简化重试：通过 task_id 直接重试失败的生成任务。"""
+    """原地重试失败的生成任务（复用同一个task_id，retry_count+1）。"""
     task = generation_task_repository.get(task_id)
     if task is None:
         raise HTTPException(status_code=404, detail="Generation task not found")
@@ -149,6 +204,7 @@ def retry_task_by_id(
         raise HTTPException(status_code=409, detail="Only failed tasks can be retried")
 
     user_id = authenticated_user.user.id
+
     # 预检查
     user_pending = generation_task_repository.count_pending_by_user(user_id)
     global_pending = generation_task_repository.count_pending_total()
@@ -163,20 +219,11 @@ def retry_task_by_id(
             detail="系统繁忙，请稍后再试",
         )
 
-    use_case = CreateGenerationTaskUseCase(generation_task_repository)
-    retried = use_case.execute(
-        CreateGenerationTaskCommand(
-            project_id=task.project_id,
-            asset_library_id=task.asset_library_id,
-            strategy_id=task.strategy_id,
-            voice_library_id=task.voice_library_id,
-            template_id=task.template_id,
-            asset_ids=task.asset_ids,
-            title_ids=task.title_ids,
-            voice_ids=task.voice_ids,
-            created_by_user_id=user_id,
-        )
-    )
+    # 原地重试
+    use_case = RetryGenerationTaskUseCase(generation_task_repository)
+    retried = use_case.execute(task_id)
+
+    # 重新入队
     try:
         if not safe_enqueue_generation_task(
             retried, generation_task_repository, user_id=user_id, log_prefix="[任务中心]"
@@ -192,18 +239,8 @@ def retry_task_by_id(
             status_code=503,
             detail="系统繁忙，请稍后再试",
         ) from None
-    return UserTaskResponse(
-        id=f"generation:{retried.id}",
-        task_type="generation",
-        project_id=retried.project_id,
-        template_id=retried.template_id,
-        status=_status_value(retried.status),
-        progress=retried.progress,
-        current_step=_generation_step(retried),
-        source_id=retried.id,
-        created_at=retried.created_at,
-        updated_at=retried.created_at,
-    )
+
+    return _generation_task_to_user_response(retried)
 
 
 # ── 项目级端点 ──
@@ -212,37 +249,64 @@ def retry_task_by_id(
 @router.get("/projects/{project_id}/tasks", response_model=ListProjectTasksResponse)
 def list_project_tasks(
     project_id: str,
+    status: str | None = Query(None, description="按状态筛选：pending/running/completed/failed/cancelled"),
+    task_type: str | None = Query(None, description="按任务类型筛选：generation/ingest"),
+    page: int = Query(1, ge=1, description="页码，从1开始"),
+    page_size: int = Query(DEFAULT_PAGE_SIZE, ge=1, le=MAX_PAGE_SIZE, description="每页数量"),
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
     project_repository: Any = Depends(get_project_repository),
     ingest_job_repository: Any = Depends(get_ingest_job_repository),
     generation_task_repository: Any = Depends(get_generation_task_repository),
 ) -> ListProjectTasksResponse:
+    """项目级任务列表，支持状态/类型筛选和分页。"""
     project = project_repository.find_by_id(project_id)
     if project is None:
         raise HTTPException(status_code=404, detail="Project not found")
 
+    status = _validate_status(status)
+    page_size = _clamp_page_size(page_size)
+    offset = (page - 1) * page_size
+
     items: list[ProjectTaskResponse] = []
-    for job in ingest_job_repository.list_by_project(project_id):
-        items.append(
-            ProjectTaskResponse(
-                id=f"ingest:{job.id}",
-                task_type="ingest",
-                project_id=job.project_id,
-                status=_status_value(job.status),
-                progress=100.0 if _status_value(job.status) == "completed" else 0.0,
-                current_step=_ingest_step(job),
-                error_message=job.error_message,
-                user_message=_humanize_task_error(job.error_message),
-                retryable=_status_value(job.status) == "failed",
-                source_id=job.id,
-                created_at=job.created_at,
-                updated_at=job.updated_at,
+
+    # 导入任务
+    if task_type is None or task_type == "ingest":
+        for job in ingest_job_repository.list_by_project(project_id):
+            if status and _status_value(job.status) != status:
+                continue
+            items.append(
+                ProjectTaskResponse(
+                    id=f"ingest:{job.id}",
+                    task_type="ingest",
+                    project_id=job.project_id,
+                    status=_status_value(job.status),
+                    progress=100.0 if _status_value(job.status) == "completed" else 0.0,
+                    current_step=_ingest_step(job),
+                    error_message=job.error_message,
+                    user_message=_humanize_task_error(job.error_message),
+                    retryable=_status_value(job.status) == "failed",
+                    source_id=job.id,
+                    created_at=job.created_at,
+                    updated_at=job.updated_at,
+                )
             )
+
+    # 生成任务
+    if task_type is None or task_type == "generation":
+        gen_items = generation_task_repository.list_by_project_filtered(
+            project_id,
+            status=status,
+            limit=page_size + 1,
+            offset=offset,
         )
-    for task in generation_task_repository.list_by_project(project_id):
-        items.append(_generation_task_to_project_response(task))
+        for task in gen_items:
+            items.append(_generation_task_to_project_response(task))
+
     items.sort(key=lambda item: item.updated_at or item.created_at or "", reverse=True)
-    return ListProjectTasksResponse(items=items)
+
+    total = generation_task_repository.count_by_project_filtered(project_id, status=status)
+
+    return ListProjectTasksResponse(items=items[:page_size], total=total)
 
 
 @router.post("/tasks/{task_type}/{source_id}/retry", response_model=ProjectTaskResponse)
@@ -253,6 +317,7 @@ def retry_project_task(
     ingest_job_repository: Any = Depends(get_ingest_job_repository),
     generation_task_repository: Any = Depends(get_generation_task_repository),
 ) -> ProjectTaskResponse:
+    """项目级任务重试。"""
     if task_type == "generation":
         task = generation_task_repository.get(source_id)
         if task is None:
@@ -261,6 +326,7 @@ def retry_project_task(
             raise HTTPException(status_code=409, detail="Only failed tasks can be retried")
 
         user_id = authenticated_user.user.id
+
         # 预检查
         user_pending = generation_task_repository.count_pending_by_user(user_id)
         global_pending = generation_task_repository.count_pending_total()
@@ -275,20 +341,10 @@ def retry_project_task(
                 detail="系统繁忙，请稍后再试",
             )
 
-        use_case = CreateGenerationTaskUseCase(generation_task_repository)
-        retried = use_case.execute(
-            CreateGenerationTaskCommand(
-                project_id=task.project_id,
-                asset_library_id=task.asset_library_id,
-                strategy_id=task.strategy_id,
-                voice_library_id=task.voice_library_id,
-                template_id=task.template_id,
-                asset_ids=task.asset_ids,
-                title_ids=task.title_ids,
-                voice_ids=task.voice_ids,
-                created_by_user_id=user_id,
-            )
-        )
+        # 原地重试
+        use_case = RetryGenerationTaskUseCase(generation_task_repository)
+        retried = use_case.execute(source_id)
+
         try:
             if not safe_enqueue_generation_task(
                 retried, generation_task_repository, user_id=user_id, log_prefix="[任务中心]"
@@ -305,6 +361,7 @@ def retry_project_task(
                 detail="系统繁忙，请稍后再试",
             ) from None
         return _generation_task_to_project_response(retried)
+
     if task_type == "ingest":
         job = ingest_job_repository.get(source_id)
         if job is None:

apps/api/app/api/routes/templates.py

@@ -8,13 +8,16 @@ from app.auth import AuthenticatedUser, get_current_user
 from app.dependencies import get_db_session
 from app.schemas.template import (
     CategoryResponse,
+    CopyTemplateRequest,
     CreateCategoryRequest,
     CreateTemplateRequest,
     GenerateWarningResponse,
     ListCategoriesResponse,
+    ListTagsResponse,
     ListTemplatesResponse,
     SegmentResponse,
     TemplateResponse,
+    TemplateUsageResponse,
     ToggleFavoriteResponse,
     UpdateTemplateRequest,
     ValidateTemplateRequest,
@@ -27,19 +30,25 @@ logger = logging.getLogger(__name__)
 
 from packages.adapters.sqlalchemy_impl.template_repository import SQLAlchemyTemplateRepository
 from packages.application.template.commands import (
+    CopyTemplateCommand,
     CreateCategoryCommand,
     CreateTemplateCommand,
+    ListTemplatesFilter,
     SegmentCommand,
     UpdateTemplateCommand,
     ValidateTemplateCommand,
 )
 from packages.application.template.use_cases import (
+    CopyTemplateUseCase,
+    CountTemplatesUseCase,
     CreateCategoryUseCase,
     CreateTemplateUseCase,
     DeleteCategoryUseCase,
     DeleteTemplateUseCase,
+    GetTemplateUsageUseCase,
     GetTemplateUseCase,
     ListCategoriesUseCase,
+    ListTagsUseCase,
     ListTemplatesUseCase,
     NotFoundError,
     UpdateTemplateUseCase,
@@ -67,7 +76,7 @@ def _segment_to_response(seg) -> SegmentResponse:
     )
 
 
-def _to_response(template) -> TemplateResponse:
+def _to_response(template, usage_count: int = 0) -> TemplateResponse:
     return TemplateResponse(
         id=template.id,
         user_id=template.user_id,
@@ -81,6 +90,7 @@ def _to_response(template) -> TemplateResponse:
         estimated_duration=template.estimated_duration,
         segments=[_segment_to_response(s) for s in getattr(template, "segments", [])],
         is_active=template.is_active,
+        usage_count=usage_count,
         created_at=template.created_at,
         updated_at=template.updated_at,
     )
@@ -93,19 +103,36 @@ def _to_response(template) -> TemplateResponse:
 def list_templates(
     skip: int = Query(0, ge=0),
     limit: int = Query(50, ge=1, le=200),
+    category: str | None = Query(None, description="按分类筛选"),
+    tag: str | None = Query(None, description="按标签筛选"),
+    keyword: str | None = Query(None, description="按名称关键词搜索"),
+    mode: str | None = Query(None, description="按剪辑模式筛选"),
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
     template_repository: SQLAlchemyTemplateRepository = Depends(_get_template_repository),
 ) -> ListTemplatesResponse:
     user_id = authenticated_user.user.id
     try:
+        tpl_filter = ListTemplatesFilter(
+            category=category,
+            tag=tag,
+            keyword=keyword,
+            mode=mode,
+        )
         use_case = ListTemplatesUseCase(template_repository)
-        templates = use_case.execute(user_id, skip=skip, limit=limit)
-        total = template_repository.count_by_user(user_id)
+        templates = use_case.execute(user_id, skip=skip, limit=limit, filter=tpl_filter)
+        count_use_case = CountTemplatesUseCase(template_repository)
+        total = count_use_case.execute(user_id, filter=tpl_filter)
+
+        # 批量查询使用次数
+        items = []
+        for t in templates:
+            usage = template_repository.get_usage_count(t.id)
+            items.append(_to_response(t, usage_count=usage))
     except Exception:
         logger.exception("list_templates 查询失败: user_id=%s", user_id)
         return ListTemplatesResponse(items=[], total=0)
     return ListTemplatesResponse(
-        items=[_to_response(t) for t in templates],
+        items=items,
         total=total,
     )
 
@@ -120,12 +147,13 @@ def get_template(
     try:
         use_case = GetTemplateUseCase(template_repository)
         template = use_case.execute(template_id, user_id)
+        usage = template_repository.get_usage_count(template_id)
     except Exception:
         logger.exception("get_template 查询失败: template_id=%s", template_id)
         raise HTTPException(status_code=status.HTTP_500_INTERNAL_SERVER_ERROR, detail="模板查询失败")
     if template is None:
         raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Template not found")
-    return _to_response(template)
+    return _to_response(template, usage_count=usage)
 
 
 @router.post("", response_model=TemplateResponse, status_code=status.HTTP_201_CREATED)
@@ -206,7 +234,7 @@ def update_template(
     return _to_response(template)
 
 
-@router.delete("/{template_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None)
+@router.delete("/{template_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None, response_class=Response)
 def delete_template(
     template_id: str,
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
@@ -217,7 +245,48 @@ def delete_template(
     deleted = use_case.execute(template_id, user_id)
     if not deleted:
         raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Template not found")
-    return Response(status_code=204)
+    return
+
+
+@router.post("/{template_id}/copy", response_model=TemplateResponse, status_code=status.HTTP_201_CREATED)
+def copy_template(
+    template_id: str,
+    request: CopyTemplateRequest,
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    template_repository: SQLAlchemyTemplateRepository = Depends(_get_template_repository),
+) -> TemplateResponse:
+    """复制模板（含所有片段配置）"""
+    user_id = authenticated_user.user.id
+    command = CopyTemplateCommand(
+        template_id=template_id,
+        user_id=user_id,
+        new_name=request.new_name,
+    )
+    use_case = CopyTemplateUseCase(template_repository)
+    try:
+        template = use_case.execute(command)
+    except NotFoundError:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Template not found")
+    except ValidationError as exc:
+        raise HTTPException(status_code=status.HTTP_422_UNPROCESSABLE_ENTITY, detail=str(exc))
+    return _to_response(template)
+
+
+@router.get("/{template_id}/usage", response_model=TemplateUsageResponse)
+def get_template_usage(
+    template_id: str,
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    template_repository: SQLAlchemyTemplateRepository = Depends(_get_template_repository),
+) -> TemplateUsageResponse:
+    """获取模板使用次数（关联的剪辑计划数量）"""
+    user_id = authenticated_user.user.id
+    # 鉴权：确保模板存在且属于当前用户
+    use_case = GetTemplateUseCase(template_repository)
+    template = use_case.execute(template_id, user_id)
+    if template is None:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Template not found")
+    usage = template_repository.get_usage_count(template_id)
+    return TemplateUsageResponse(template_id=template_id, usage_count=usage)
 
 
 @router.post("/{template_id}/toggle-favorite", response_model=ToggleFavoriteResponse)
@@ -307,7 +376,7 @@ def create_category(
     )
 
 
-@router.delete("/categories/{category_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None)
+@router.delete("/categories/{category_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None, response_class=Response)
 def delete_category(
     category_id: str,
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
@@ -319,3 +388,22 @@ def delete_category(
     if not deleted:
         raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Category not found")
     return Response(status_code=204)
+
+
+# ── Tags ──
+
+
+@router.get("/tags/list", response_model=ListTagsResponse)
+def list_tags(
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    template_repository: SQLAlchemyTemplateRepository = Depends(_get_template_repository),
+) -> ListTagsResponse:
+    """获取用户所有模板标签（去重排序）"""
+    user_id = authenticated_user.user.id
+    try:
+        use_case = ListTagsUseCase(template_repository)
+        tags = use_case.execute(user_id)
+    except Exception:
+        logger.exception("list_tags 查询失败: user_id=%s", user_id)
+        return ListTagsResponse(items=[])
+    return ListTagsResponse(items=tags)

apps/api/app/api/routes/titles.py

@@ -4,6 +4,7 @@ from __future__ import annotations
 
 from typing import Optional
 
+from app.api.routes._helpers import get_user_plan
 from app.auth import AuthenticatedUser, get_current_user
 from app.dependencies import get_db_session, get_user_repository
 from app.schemas.title_library import (
@@ -16,20 +17,23 @@ from fastapi import APIRouter, Depends, HTTPException, Query, Response, status
 from sqlalchemy.orm import Session
 
 from packages.adapters.sqlalchemy_impl.title_library_repository import SQLAlchemyTitleLibraryRepository
-from packages.application.title_library.commands import CreateTitleLibraryCommand, UpdateTitleLibraryCommand
+from packages.application.title_library.commands import (
+    CreateTitleLibraryCommand,
+    PickTitleCommand,
+    UpdateTitleLibraryCommand,
+)
 from packages.application.title_library.use_cases import (
     CreateTitleLibraryUseCase,
     DeleteTitleLibraryUseCase,
     GetTitleLibraryUseCase,
     ListTitleLibraryUseCase,
     NotFoundError,
+    PickTitleUseCase,
     QuotaExceededError,
     UpdateTitleLibraryUseCase,
 )
 from packages.ports.user_repository import UserRepository
 
-from app.api.routes._helpers import get_user_plan
-
 router = APIRouter()
 
 
@@ -71,6 +75,41 @@ def list_titles(
     )
 
 
+@router.post("/pick", response_model=TitleLibraryItemResponse)
+def pick_title(
+    category: Optional[str] = Query(None, description="按分类筛选，不填则从全部标题中选"),
+    exclude_ids: Optional[str] = Query(
+        None,
+        description="排除的标题ID（逗号分隔），用于批量生成时避免重复",
+    ),
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    title_repository: SQLAlchemyTitleLibraryRepository = Depends(_get_title_repository),
+) -> TitleLibraryItemResponse:
+    """智能选择一个标题。
+
+    策略：优先使用次数少的，从最少的前5个中随机选一个，兼顾公平和多样性。
+    """
+    user_id = authenticated_user.user.id
+    exclude_list: list[str] = []
+    if exclude_ids:
+        exclude_list = [t.strip() for t in exclude_ids.split(",") if t.strip()]
+
+    use_case = PickTitleUseCase(title_repository)
+    item = use_case.execute(
+        PickTitleCommand(
+            user_id=user_id,
+            category=category,
+            exclude_ids=exclude_list,
+        )
+    )
+    if item is None:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail="标题库为空，请先添加标题",
+        )
+    return _to_response(item)
+
+
 @router.get("/{title_id}", response_model=TitleLibraryItemResponse)
 def get_title(
     title_id: str,
@@ -138,7 +177,7 @@ def update_title(
     return _to_response(item)
 
 
-@router.delete("/{title_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None)
+@router.delete("/{title_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None, response_class=Response)
 def delete_title(
     title_id: str,
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
@@ -149,4 +188,4 @@ def delete_title(
     deleted = use_case.execute(title_id, user_id)
     if not deleted:
         raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Title not found")
-    return Response(status_code=204)
+    return

apps/api/app/api/routes/tts.py

@@ -46,6 +46,7 @@ from packages.application.voice_library.use_cases import (
     CreateVoiceLibraryUseCase,
     QuotaExceededError,
 )
+from packages.domain.voice_presets import list_voices
 from packages.ports.user_repository import UserRepository
 
 logger = logging.getLogger(__name__)
@@ -53,6 +54,34 @@ logger = logging.getLogger(__name__)
 router = APIRouter()
 
 
+@router.get("/presets", summary="获取预设音色列表")
+def list_preset_voices(
+    gender: Optional[str] = Query(None, description="按性别筛选: male/female/child"),
+    style: Optional[str] = Query(None, description="按风格筛选: stable/lively/customer_service/narration/news/story"),
+    keyword: Optional[str] = Query(None, description="按关键词搜索"),
+    _user: AuthenticatedUser = Depends(get_current_user),
+) -> list[dict]:
+    """获取可用的预设音色列表。
+
+    用于配音功能的音色选择。
+    """
+    voices = list_voices(gender=gender, style=style, keyword=keyword)
+    return [
+        {
+            "voice_id": v.voice_id,
+            "name": v.name,
+            "gender": v.gender.value,
+            "style": v.style.value,
+            "description": v.description,
+            "default_speed": v.default_speed,
+            "default_pitch": v.default_pitch,
+            "sample_rate": v.sample_rate,
+            "language": v.language,
+        }
+        for v in voices
+    ]
+
+
 def _get_repository(session: Session = Depends(get_db_session)) -> SQLAlchemyTTSJobRepository:
     return SQLAlchemyTTSJobRepository(session)
 
@@ -241,7 +270,7 @@ def get_tts_job_status(
     )
 
 
-@router.delete("/jobs/{job_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None)
+@router.delete("/jobs/{job_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None, response_class=Response)
 def delete_tts_job(
     job_id: str,
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
@@ -253,7 +282,7 @@ def delete_tts_job(
     deleted = use_case.execute(job_id, user_id)
     if not deleted:
         raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="TTS job not found")
-    return Response(status_code=204)
+    return
 
 
 @router.post(

apps/api/app/api/routes/upload.py

@@ -2,6 +2,7 @@ import logging
 from typing import Any
 from uuid import uuid4
 
+from app.api.routes._helpers import require_project_and_library
 from app.auth import AuthenticatedUser, get_current_user
 from app.config import get_settings
 from app.core.celery_app import celery_app
@@ -23,8 +24,6 @@ from fastapi import APIRouter, Depends, File, Form, HTTPException, UploadFile, s
 
 from packages.application import SubmitIngestJobCommand, SubmitIngestJobUseCase
 
-from app.api.routes._helpers import require_project_and_library
-
 logger = logging.getLogger(__name__)
 
 router = APIRouter()

apps/api/app/api/routes/videos.py

@@ -0,0 +1,179 @@
+import logging
+import uuid
+
+from app.api.routes._helpers import check_project_access
+from app.auth import AuthenticatedUser, get_current_user
+from app.core.celery_app import celery_app
+from app.core.storage import OSSStorageService, get_storage_service
+from app.dependencies import get_generated_video_repository
+from app.schemas.video_center import (
+    BatchDownloadRequest,
+    BatchDownloadResponse,
+    ListVideosResponse,
+    UpdateVideoReviewRequest,
+    VideoItemResponse,
+)
+from fastapi import APIRouter, Depends, HTTPException, Query
+
+from packages.application import (
+    GetGeneratedVideoUseCase,
+    GetVideosByIdsUseCase,
+    ListGeneratedVideosPaginatedUseCase,
+    UpdateVideoReviewStatusUseCase,
+)
+
+logger = logging.getLogger(__name__)
+
+router = APIRouter()
+
+
+def _to_video_response(item, storage: OSSStorageService | None = None) -> VideoItemResponse:
+    download_url = None
+    if storage and item.file_url:
+        try:
+            download_url = storage.get_download_url(item.file_url)
+        except Exception:
+            download_url = item.file_url
+    return VideoItemResponse(
+        id=item.id,
+        project_id=item.project_id,
+        generation_task_id=item.generation_task_id,
+        name=item.name,
+        file_url=item.file_url,
+        file_size=item.file_size,
+        duration=item.duration,
+        thumbnail_url=item.thumbnail_url,
+        width=item.width,
+        height=item.height,
+        fps=item.fps,
+        status=item.status,
+        review_status=item.review_status,
+        generation_params=item.generation_params,
+        download_url=download_url,
+        generated_at=item.generated_at.isoformat() if hasattr(item, "generated_at") and item.generated_at else "",
+    )
+
+
+@router.get("/videos", response_model=ListVideosResponse)
+def list_videos(
+    project_id: str | None = Query(None, description="项目ID，不传则返回所有项目"),
+    status: str | None = Query(None, description="按状态筛选"),
+    review_status: str | None = Query(None, description="按复核状态筛选"),
+    page: int = Query(1, ge=1, description="页码"),
+    page_size: int = Query(20, ge=1, le=100, description="每页数量"),
+    repo=Depends(get_generated_video_repository),
+    storage: OSSStorageService = Depends(get_storage_service),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+):
+    """成片列表，支持分页、按项目/状态/复核状态筛选。"""
+    use_case = ListGeneratedVideosPaginatedUseCase(repo)
+    items, total = use_case.execute(
+        project_id=project_id,
+        status=status,
+        review_status=review_status,
+        page=page,
+        page_size=page_size,
+    )
+    return ListVideosResponse(
+        items=[_to_video_response(item, storage) for item in items],
+        total=total,
+        page=page,
+        page_size=page_size,
+    )
+
+
+@router.get("/videos/{video_id}", response_model=VideoItemResponse)
+def get_video(
+    video_id: str,
+    repo=Depends(get_generated_video_repository),
+    storage: OSSStorageService = Depends(get_storage_service),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+):
+    """获取单个成片详情。"""
+    use_case = GetGeneratedVideoUseCase(repo)
+    item = use_case.execute(video_id)
+    if item is None:
+        raise HTTPException(status_code=404, detail="Video not found")
+    return _to_video_response(item, storage)
+
+
+@router.patch("/videos/{video_id}/review", response_model=VideoItemResponse)
+def update_video_review_status(
+    video_id: str,
+    request: UpdateVideoReviewRequest,
+    repo=Depends(get_generated_video_repository),
+    storage: OSSStorageService = Depends(get_storage_service),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+):
+    """更新成片复核状态：pending_review / approved / rejected。"""
+    use_case = UpdateVideoReviewStatusUseCase(repo)
+    item = use_case.execute(video_id, request.review_status)
+    if item is None:
+        raise HTTPException(status_code=404, detail="Video not found")
+    logger.info("Video %s review status updated to %s by user %s", video_id, request.review_status, current_user.user_id)
+    return _to_video_response(item, storage)
+
+
+@router.post("/videos/batch-download", response_model=BatchDownloadResponse)
+def batch_download_videos(
+    request: BatchDownloadRequest,
+    repo=Depends(get_generated_video_repository),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+):
+    """批量下载成片，异步打包 zip。
+    
+    传入 video_ids 列表，创建一个批量下载任务，任务完成后返回 zip 下载链接。
+    """
+    if not request.video_ids:
+        raise HTTPException(status_code=400, detail="video_ids cannot be empty")
+    if len(request.video_ids) > 50:
+        raise HTTPException(status_code=400, detail="Maximum 50 videos per batch download")
+
+    # 校验视频都存在
+    use_case = GetVideosByIdsUseCase(repo)
+    videos = use_case.execute(request.video_ids)
+    if len(videos) != len(request.video_ids):
+        raise HTTPException(status_code=404, detail="Some videos not found")
+
+    # 发送 celery 任务
+    task = celery_app.send_task(
+        "worker.batch_download_videos",
+        args=[request.video_ids, current_user.user_id],
+    )
+
+    logger.info("Batch download job created: %s, videos=%d", task.id, len(request.video_ids))
+    return BatchDownloadResponse(job_id=task.id, status="pending")
+
+
+@router.get("/videos/batch-download/{job_id}", response_model=BatchDownloadResponse)
+def get_batch_download_status(
+    job_id: str,
+    current_user: AuthenticatedUser = Depends(get_current_user),
+):
+    """查询批量下载任务状态。"""
+    from celery.result import AsyncResult
+
+    task = AsyncResult(job_id, app=celery_app)
+    
+    status_map = {
+        "PENDING": "pending",
+        "STARTED": "running",
+        "SUCCESS": "success",
+        "FAILURE": "failed",
+        "RETRY": "pending",
+        "REVOKED": "cancelled",
+    }
+    api_status = status_map.get(task.state, "pending")
+
+    download_url = None
+    if task.state == "SUCCESS" and task.result:
+        if isinstance(task.result, dict):
+            download_url = task.result.get("download_url")
+        elif isinstance(task.result, str):
+            download_url = task.result
+
+    return BatchDownloadResponse(
+        job_id=job_id,
+        status=api_status,
+        download_url=download_url,
+    )

apps/api/app/api/routes/voice_clones.py

@@ -172,6 +172,7 @@ def get_voice_clone_status(
     "/{clone_id}",
     status_code=status.HTTP_204_NO_CONTENT,
     response_model=None,
+    response_class=Response,
 )
 def delete_voice_clone(
     clone_id: str,
@@ -184,7 +185,7 @@ def delete_voice_clone(
     deleted = use_case.execute(clone_id, user_id)
     if not deleted:
         raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Voice clone not found")
-    return Response(status_code=204)
+    return
 
 
 @router.post("/{clone_id}/retry", response_model=VoiceCloneProfileResponse)

apps/api/app/api/routes/voices.py

@@ -7,6 +7,7 @@ from __future__ import annotations
 
 from typing import Literal, Optional
 
+from app.api.routes._helpers import get_user_plan
 from app.auth import AuthenticatedUser, get_current_user
 from app.dependencies import get_audio_url_signer, get_db_session, get_user_repository
 from app.schemas.voice import (
@@ -39,8 +40,6 @@ from packages.application.voice_library.use_cases import (
 from packages.domain.preset_voices import PRESET_VOICES
 from packages.ports.user_repository import UserRepository
 
-from app.api.routes._helpers import get_user_plan
-
 router = APIRouter()
 
 
@@ -323,7 +322,7 @@ def update_voice(
     return _to_response(item, sign_url)
 
 
-@router.delete("/{voice_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None)
+@router.delete("/{voice_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None, response_class=Response)
 def delete_voice(
     voice_id: str,
     authenticated_user: AuthenticatedUser = Depends(get_current_user),
@@ -334,4 +333,4 @@ def delete_voice(
     deleted = use_case.execute(voice_id, user_id)
     if not deleted:
         raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Voice not found")
-    return Response(status_code=204)
+    return

apps/api/app/core/storage.py

@@ -5,8 +5,8 @@ This module keeps old import paths working so existing code
 does not need to change.
 """
 
+from packages.shared.storage import SharedStorageService as OSSStorageService
 from packages.shared.storage import (
-    SharedStorageService as OSSStorageService,
     get_shared_storage_service,
     get_storage_service,
 )

apps/api/app/schemas/asset.py

@@ -54,17 +54,45 @@ class AssetResponse(BaseModel):
     tag_ids: list[str] = Field(default_factory=list)
 
 
+MAX_BATCH_SIZE = 200
+
+
 class BatchDeleteRequest(BaseModel):
-    """批量删除请求。"""
+    """批量删除请求（软删除）。"""
 
-    ids: list[str] = Field(..., min_length=1, max_length=100, description="要删除的素材 ID 列表")
+    asset_ids: list[str] = Field(..., min_length=1, max_length=MAX_BATCH_SIZE, description="要删除的素材 ID 列表")
 
 
-class BatchDeleteResponse(BaseModel):
-    """批量删除响应。"""
+class BatchOperationResponse(BaseModel):
+    """批量操作通用响应。"""
 
-    deleted_count: int = Field(..., ge=0, description="实际删除数量")
-    failed_ids: list[str] = Field(default_factory=list, description="删除失败的 ID 列表")
+    success_count: int = Field(..., ge=0, description="成功数量")
+    failed_ids: list[str] = Field(default_factory=list, description="失败的 ID 列表")
+    failed_details: dict[str, str] = Field(default_factory=dict, description="失败详情 {asset_id: reason}")
+
+
+class BatchTagRequest(BaseModel):
+    """批量打标签请求。"""
+
+    asset_ids: list[str] = Field(..., min_length=1, max_length=MAX_BATCH_SIZE, description="素材 ID 列表")
+    tag_ids: list[str] = Field(..., min_length=1, max_length=50, description="标签 ID 列表")
+    mode: str = Field(default="add", pattern="^(add|replace)$", description="add=添加合并，replace=全量替换")
+
+
+class BatchClassifyRequest(BaseModel):
+    """批量修改分类请求。"""
+
+    asset_ids: list[str] = Field(..., min_length=1, max_length=MAX_BATCH_SIZE, description="素材 ID 列表")
+    category: str = Field(..., min_length=1, max_length=50, description="内容分类，如 person/scenic/product")
+
+
+class BatchMarkRequest(BaseModel):
+    """批量设置智能视图标记请求。"""
+
+    asset_ids: list[str] = Field(..., min_length=1, max_length=MAX_BATCH_SIZE, description="素材 ID 列表")
+    smart_view: str = Field(
+        ..., pattern="^(recommended|caution|high_risk)$", description="智能视图标记：recommended/caution/high_risk"
+    )
 
 
 class ListAssetsResponse(BaseModel):

apps/api/app/schemas/generation_task.py

@@ -33,6 +33,17 @@ class CreateGenerationTaskRequest(BaseModel):
     asset_select_count: int = Field(
         default=0, ge=0, le=100, description="选取数量，0表示全部（仅 random/smart 模式有效）"
     )
+    # ── 自动重试 ──
+    auto_retry_enabled: bool = Field(
+        default=False,
+        description="是否开启失败自动重试，默认关闭",
+    )
+    auto_retry_max: int = Field(
+        default=0,
+        ge=0,
+        le=5,
+        description="最大自动重试次数，0表示不自动重试，最大5次",
+    )
 
     @model_validator(mode="after")
     def _check_at_least_one_mode(self) -> "CreateGenerationTaskRequest":
@@ -64,6 +75,10 @@ class GenerationTaskResponse(BaseModel):
     progress: float
     result_count: int
     error_message: str
+    error_info: dict = Field(default_factory=dict)
+    retry_count: int = 0
+    auto_retry_enabled: bool = False
+    auto_retry_max: int = 0
     logs: list[dict] = Field(default_factory=list)
 
     @field_validator("logs", mode="before")

apps/api/app/schemas/task_center.py

@@ -11,8 +11,10 @@ class ProjectTaskResponse(BaseModel):
     progress: float
     current_step: str
     error_message: str = ""
+    error_info: dict = Field(default_factory=dict)
     user_message: str = ""
     retryable: bool = False
+    retry_count: int = 0
     source_id: str = ""
     template_id: str = ""
     created_at: datetime | None = None
@@ -21,6 +23,7 @@ class ProjectTaskResponse(BaseModel):
 
 class ListProjectTasksResponse(BaseModel):
     items: list[ProjectTaskResponse] = Field(default_factory=list)
+    total: int = 0
 
 
 class UserTaskResponse(BaseModel):
@@ -34,8 +37,10 @@ class UserTaskResponse(BaseModel):
     progress: float
     current_step: str
     error_message: str = ""
+    error_info: dict = Field(default_factory=dict)
     user_message: str = ""
     retryable: bool = False
+    retry_count: int = 0
     source_id: str = ""
     created_at: datetime | None = None
     updated_at: datetime | None = None
@@ -45,3 +50,4 @@ class ListTasksResponse(BaseModel):
     """用户级任务列表响应（GET /api/v1/tasks）。"""
 
     items: list[UserTaskResponse] = Field(default_factory=list)
+    total: int = 0

apps/api/app/schemas/template.py

@@ -45,6 +45,7 @@ class TemplateResponse(BaseModel):
     segments: List[SegmentResponse] = Field(default_factory=list)
     is_active: bool = True
     is_favorite: bool = False
+    usage_count: int = 0
     created_at: datetime
     updated_at: datetime
 
@@ -120,3 +121,25 @@ class CreateCategoryRequest(BaseModel):
 
 class ListCategoriesResponse(BaseModel):
     items: List[CategoryResponse]
+
+
+# ── Copy Template ──
+
+
+class CopyTemplateRequest(BaseModel):
+    new_name: str
+
+
+# ── Tags ──
+
+
+class ListTagsResponse(BaseModel):
+    items: List[str]
+
+
+# ── Usage Stats ──
+
+
+class TemplateUsageResponse(BaseModel):
+    template_id: str
+    usage_count: int

apps/api/app/schemas/video_center.py

@@ -0,0 +1,45 @@
+from typing import Literal
+
+from pydantic import BaseModel, Field
+
+VideoReviewStatus = Literal["pending_review", "approved", "rejected"]
+
+
+class VideoItemResponse(BaseModel):
+    id: str
+    project_id: str
+    generation_task_id: str
+    name: str
+    file_url: str
+    file_size: int
+    duration: float
+    thumbnail_url: str | None = None
+    width: int
+    height: int
+    fps: float
+    status: str = "completed"
+    review_status: str = "pending_review"
+    generation_params: dict = Field(default_factory=dict)
+    download_url: str | None = None
+    generated_at: str = ""
+
+
+class ListVideosResponse(BaseModel):
+    items: list[VideoItemResponse]
+    total: int
+    page: int
+    page_size: int
+
+
+class UpdateVideoReviewRequest(BaseModel):
+    review_status: VideoReviewStatus
+
+
+class BatchDownloadRequest(BaseModel):
+    video_ids: list[str]
+
+
+class BatchDownloadResponse(BaseModel):
+    job_id: str
+    status: str = "pending"
+    download_url: str | None = None

apps/api/app/services/edit_plan_service.py

@@ -281,6 +281,8 @@ class EditPlanService:
         start_time: float = 0.0,
         duration: float = 0.0,
         transition_effect: str = "cut",
+        transition_duration: float = 0.0,
+        playback_speed: float = 1.0,
         config: Optional[dict[str, Any]] = None,
     ) -> EditPlanClip:
         """创建片段
@@ -301,6 +303,8 @@ class EditPlanService:
             start_time=start_time,
             duration=duration,
             transition_effect=transition_effect,
+            transition_duration=transition_duration,
+            playback_speed=playback_speed,
             config=config,
         )
         created = self._clip_repo.create(clip)
@@ -324,6 +328,8 @@ class EditPlanService:
         start_time: Optional[float] = None,
         duration: Optional[float] = None,
         transition_effect: Optional[str] = None,
+        transition_duration: Optional[float] = None,
+        playback_speed: Optional[float] = None,
         config: Optional[dict[str, Any]] = None,
     ) -> EditPlanClip:
         """更新片段
@@ -333,6 +339,15 @@ class EditPlanService:
         """
         existing = self.get_clip_or_raise(clip_id)
 
+        # 速度边界钳制
+        if playback_speed is not None:
+            if playback_speed <= 0:
+                playback_speed = 1.0
+            elif playback_speed < 0.25:
+                playback_speed = 0.25
+            elif playback_speed > 4.0:
+                playback_speed = 4.0
+
         updated = EditPlanClip(
             id=existing.id,
             plan_id=existing.plan_id,
@@ -346,6 +361,10 @@ class EditPlanService:
             transition_effect=(
                 transition_effect.strip() if transition_effect is not None else existing.transition_effect
             ),
+            transition_duration=(
+                transition_duration if transition_duration is not None else existing.transition_duration
+            ),
+            playback_speed=playback_speed if playback_speed is not None else existing.playback_speed,
             status=existing.status,
             config=config if config is not None else existing.config,
             created_at=existing.created_at,

apps/api/app/services/job_service.py

@@ -13,7 +13,6 @@ from __future__ import annotations
 import logging
 from typing import Any
 
-
 from packages.application.jobs import (
     CancelJobUseCase,
     CompleteJobCommand,

apps/web/playwright.config.ts

@@ -26,6 +26,9 @@ export default defineConfig({
           use: {
             ...devices["Desktop Chrome"],
             channel: process.env.E2E_BROWSER_CHANNEL || "msedge",
+            launchOptions: {
+              args: ["--disable-gpu", "--disable-software-rasterizer"],
+            },
           },
         },
         {
@@ -51,6 +54,9 @@ export default defineConfig({
           use: {
             ...devices["Desktop Chrome"],
             channel: process.env.E2E_BROWSER_CHANNEL || "msedge",
+            launchOptions: {
+              args: ["--disable-gpu", "--disable-software-rasterizer"],
+            },
           },
         },
       ],

apps/worker/services/asr_service_factory.py

@@ -0,0 +1,47 @@
+"""ASR 服务工厂 — 根据环境配置创建对应 ASR 服务实例。
+
+支持的后端：
+- mock: MockASRService（测试/开发用）
+- 后续可扩展：whisper / aliyun / tencent 等
+"""
+
+from __future__ import annotations
+
+import os
+from functools import lru_cache
+
+from packages.ports.asr_service import ASRService
+
+
+@lru_cache(maxsize=1)
+def get_asr_service() -> ASRService | None:
+    """获取全局 ASR 服务实例（单例）。
+
+    根据环境变量 ASR_PROVIDER 决定使用哪个后端：
+    - mock / 空 / 未设置: 返回 None（不启用 ASR）
+    - mock: 使用 MockASRService
+
+    Returns:
+        ASRService 实例，未配置或不启用时返回 None
+    """
+    provider = os.environ.get("ASR_PROVIDER", "").lower().strip()
+
+    if not provider:
+        return None
+
+    if provider == "mock":
+        from packages.adapters.asr.mock_asr_service import MockASRService
+
+        return MockASRService()
+
+    # 未知 provider，记录日志并返回 None（不启用 ASR，不阻断主流程）
+    import logging
+
+    logger = logging.getLogger(__name__)
+    logger.warning("未知的 ASR provider: %s，ASR 自动字幕功能未启用", provider)
+    return None
+
+
+def reset_asr_service_cache() -> None:
+    """重置 ASR 服务缓存（测试用）。"""
+    get_asr_service.cache_clear()

apps/worker/services/tts_service_factory.py

@@ -0,0 +1,66 @@
+"""TTS 服务工厂.
+
+根据配置创建对应的 TTS 服务实例。
+"""
+
+from __future__ import annotations
+
+import logging
+import os
+
+from packages.ports.tts_service import TtsService
+
+logger = logging.getLogger(__name__)
+
+# 可用的 provider 映射
+_PROVIDERS: dict[str, type[TtsService]] = {}
+
+
+def register_provider(name: str, cls: type[TtsService]) -> None:
+    """注册 TTS 供应商."""
+    _PROVIDERS[name] = cls
+
+
+def get_tts_service(provider: str | None = None, **kwargs) -> TtsService:
+    """获取 TTS 服务实例.
+
+    Args:
+        provider: 供应商名称（None 则从环境变量读取 TTS_PROVIDER）
+        **kwargs: 传递给服务构造函数的参数
+
+    Returns:
+        TTS 服务实例
+
+    Raises:
+        ValueError: 不支持的供应商
+    """
+    if provider is None:
+        provider = os.environ.get("TTS_PROVIDER", "mock")
+
+    provider = provider.lower()
+
+    if provider not in _PROVIDERS:
+        # 延迟导入避免循环依赖
+        if provider == "mock":
+            from packages.adapters.tts.mock_tts_service import MockTtsService
+
+            _PROVIDERS["mock"] = MockTtsService
+        else:
+            logger.warning("未知 TTS provider: %s，回退到 mock", provider)
+            from packages.adapters.tts.mock_tts_service import MockTtsService
+
+            _PROVIDERS["mock"] = MockTtsService
+            provider = "mock"
+
+    cls = _PROVIDERS[provider]
+    return cls(**kwargs)
+
+
+def available_providers() -> list[str]:
+    """获取可用的供应商列表."""
+    # 确保 mock 已注册
+    if "mock" not in _PROVIDERS:
+        from packages.adapters.tts.mock_tts_service import MockTtsService
+
+        _PROVIDERS["mock"] = MockTtsService
+    return list(_PROVIDERS.keys())

apps/worker/video_processing/bgm_mixer.py

@@ -0,0 +1,313 @@
+"""BGM 混音模块 — 背景音乐与主音频混合.
+
+基于 FFmpeg 实现：
+- BGM 音量调节
+- 淡入淡出（afade）
+- 循环播放（aloop，短 BGM 铺长视频）
+- 人声闪避（sidechaincompress，有人声时BGM自动降低音量）
+- amix 混音
+
+作为 render_audio.py 的增强模块，在 mix_audio 后处理阶段被调用。
+"""
+
+from __future__ import annotations
+
+import logging
+from dataclasses import dataclass
+from pathlib import Path
+from typing import TYPE_CHECKING
+
+from video_processing.ffmpeg_utils import FFMPEG_BIN, probe_duration, run_ffmpeg
+
+if TYPE_CHECKING:
+    from video_processing.render_audio import RenderContext
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class BGMConfig:
+    """BGM 混音配置（内部使用，从 plan.config.bgm 转换而来）"""
+
+    bgm_path: str  # BGM 本地文件路径
+    volume: float = 0.3  # 0.0 ~ 1.0
+    fade_in: float = 0.0  # 淡入时长（秒）
+    fade_out: float = 0.0  # 淡出时长（秒）
+    loop_enabled: bool = True  # 是否循环铺满
+    sidechain_enabled: bool = False  # 人声闪避
+    sidechain_ratio: float = 0.3  # 闪避时音量降低比例
+    sidechain_attack: float = 0.02  # 攻击时间
+    sidechain_release: float = 0.5  # 释放时间
+    sidechain_threshold: float = -25.0  # 触发阈值（dB）
+
+    @classmethod
+    def from_config_dict(cls, bgm_path: str, config: dict) -> "BGMConfig":
+        """从 plan.config.bgm 字典创建 BGMConfig。"""
+        return cls(
+            bgm_path=bgm_path,
+            volume=float(config.get("volume", 0.3)),
+            fade_in=float(config.get("fade_in", 0.0)),
+            fade_out=float(config.get("fade_out", 0.0)),
+            loop_enabled=bool(config.get("loop_enabled", True)),
+            sidechain_enabled=bool(config.get("sidechain_enabled", False)),
+            sidechain_ratio=float(config.get("sidechain_ratio", 0.3)),
+            sidechain_attack=float(config.get("sidechain_attack", 0.02)),
+            sidechain_release=float(config.get("sidechain_release", 0.5)),
+            sidechain_threshold=float(config.get("sidechain_threshold", -25.0)),
+        )
+
+
+# ── BGM 预处理 ────────────────────────────────────────────────────────────────
+
+
+def prepare_bgm_track(
+    ctx: "RenderContext",
+    bgm: BGMConfig,
+    target_duration: float,
+) -> Path:
+    """预处理 BGM 轨道：循环/截断 + 音量 + 淡入淡出.
+
+    生成一个时长精确等于 target_duration 的 BGM 音频文件。
+    后续再与主音频混音。
+
+    Args:
+        ctx: 渲染上下文
+        bgm: BGM 配置
+        target_duration: 目标时长（秒），通常等于视频总时长
+
+    Returns:
+        处理后的 BGM 音频文件路径
+    """
+    output_path = ctx.work_dir / f"bgm_processed_{ctx.plan_id}.aac"
+
+    if target_duration <= 0:
+        target_duration = 5.0  # 兜底
+
+    bgm_dur = probe_duration(bgm.bgm_path)
+    needs_loop = bgm.loop_enabled and bgm_dur > 0 and bgm_dur < target_duration * 0.9
+
+    # 构建滤镜链
+    filter_parts: list[str] = []
+    input_looped: bool = False
+
+    if needs_loop:
+        # 计算需要循环多少次才能铺满
+        loop_count = max(1, int(target_duration / bgm_dur) + 2)
+        # aloop 滤镜：循环指定次数
+        filter_parts.append(f"aloop=loop={loop_count}:size=0")
+        input_looped = True
+
+    # 音量调节
+    volume = max(0.0, min(1.0, bgm.volume))
+    if abs(volume - 1.0) > 0.001:
+        filter_parts.append(f"volume={volume:.3f}")
+
+    # 淡入
+    if bgm.fade_in > 0:
+        filter_parts.append(f"afade=t=in:st=0:d={bgm.fade_in:.3f}")
+
+    # 淡出（从 target_duration - fade_out 开始）
+    if bgm.fade_out > 0 and target_duration > bgm.fade_out:
+        fade_start = target_duration - bgm.fade_out
+        filter_parts.append(f"afade=t=out:st={fade_start:.3f}:d={bgm.fade_out:.3f}")
+
+    # 最终截断到目标时长
+    filter_parts.append(f"atrim=0:{target_duration:.3f}")
+    filter_parts.append("asetpts=N/SR/TB")  # 重置时间戳
+
+    filter_str = ",".join(filter_parts)
+
+    command = [
+        FFMPEG_BIN,
+        "-y",
+        "-i",
+        bgm.bgm_path,
+        "-filter:a",
+        filter_str,
+        "-c:a",
+        "aac",
+        "-b:a",
+        "128k",
+        str(output_path),
+    ]
+
+    logger.info(
+        "[bgm] prepare BGM track: path=%s dur=%.2f target=%.2f loop=%s fade_in=%.2f fade_out=%.2f",
+        bgm.bgm_path[-40:],
+        bgm_dur,
+        target_duration,
+        needs_loop,
+        bgm.fade_in,
+        bgm.fade_out,
+    )
+
+    run_ffmpeg(command)
+    return output_path
+
+
+# ── BGM + 主音频混音 ──────────────────────────────────────────────────────────
+
+
+def mix_bgm_with_main(
+    ctx: "RenderContext",
+    main_audio_path: Path,
+    bgm: BGMConfig,
+    target_duration: float,
+) -> Path:
+    """将 BGM 与主音频混合.
+
+    两种模式：
+    1. 普通混音（sidechain 关闭）：amix 两路音频
+    2. 人声闪避（sidechain 开启）：用 sidechaincompress 让 BGM 跟随主音频音量自动调整
+
+    Args:
+        ctx: 渲染上下文
+        main_audio_path: 主音频文件路径（人声/原始音频）
+        bgm: BGM 配置
+        target_duration: 目标时长
+
+    Returns:
+        混音后的音频文件路径
+    """
+    output_path = ctx.work_dir / f"audio_with_bgm_{ctx.plan_id}.aac"
+
+    # 先预处理 BGM 轨道（循环/音量/淡入淡出/截断）
+    bgm_processed = prepare_bgm_track(ctx, bgm, target_duration)
+
+    if not bgm.sidechain_enabled:
+        # 普通 amix 混音
+        _mix_simple(main_audio_path, bgm_processed, output_path)
+    else:
+        # sidechain 人声闪避混音
+        _mix_sidechain(main_audio_path, bgm_processed, output_path, bgm)
+
+    return output_path
+
+
+def _mix_simple(main_path: Path, bgm_path: Path, output_path: Path) -> None:
+    """简单 amix 混音：主音频 + BGM = 输出.
+
+    主音频权重 1.0，BGM 已经在预处理阶段调好了音量。
+    amix 会自动归一化，需要用 volume 补偿。
+    """
+    # 使用 amix：inputs=2，duration=first（以主音频时长为准）
+    # 然后用 volume=2 补偿 amix 的衰减（2路输入每路平均乘0.5）
+    filter_complex = "[0:a][1:a]amix=inputs=2:duration=first:dropout_transition=0[outa];" "[outa]volume=2[final]"
+
+    command = [
+        FFMPEG_BIN,
+        "-y",
+        "-i",
+        str(main_path),
+        "-i",
+        str(bgm_path),
+        "-filter_complex",
+        filter_complex,
+        "-map",
+        "[final]",
+        "-c:a",
+        "aac",
+        "-b:a",
+        "128k",
+        str(output_path),
+    ]
+
+    logger.info("[bgm] simple amix mix")
+    run_ffmpeg(command)
+
+
+def _mix_sidechain(
+    main_path: Path,
+    bgm_path: Path,
+    output_path: Path,
+    bgm: BGMConfig,
+) -> None:
+    """sidechain 人声闪避混音.
+
+    原理：
+    - 主音频作为 sidechain 信号源
+    - BGM 轨道经过 sidechaincompress，根据主音频音量动态调整 BGM 音量
+    - 最后 amix 混音
+
+    FFmpeg sidechaincompress 参数：
+    - threshold: 触发阈值（dB），主音频超过此值时开始压缩
+    - ratio: 压缩比，越高压缩越狠
+    - attack: 攻击时间（秒）
+    - release: 释放时间（秒）
+    """
+    # sidechain_ratio 表示闪避时 BGM 音量降低比例
+    # ratio = 1 / (1 - sidechain_ratio)，但实际压缩比需要更精细调整
+    # 简化处理：把 ratio 映射到 2:1 ~ 10:1 范围
+    ratio = max(2.0, min(10.0, 1.0 / (1.0 - bgm.sidechain_ratio)))
+
+    filter_complex = (
+        # BGM 经过 sidechain 压缩，用主音频做触发
+        f"[1:a][0:a]sidechaincompress="
+        f"threshold={bgm.sidechain_threshold}dB:"
+        f"ratio={ratio:.1f}:"
+        f"attack={bgm.sidechain_attack:.3f}:"
+        f"release={bgm.sidechain_release:.3f}:"
+        f"knee=6[bgm_comp];"
+        # 主音频 + 压缩后的 BGM 混音
+        f"[0:a][bgm_comp]amix=inputs=2:duration=first:dropout_transition=0[outa];"
+        f"[outa]volume=1.5[final]"  # 轻微补偿
+    )
+
+    command = [
+        FFMPEG_BIN,
+        "-y",
+        "-i",
+        str(main_path),
+        "-i",
+        str(bgm_path),
+        "-filter_complex",
+        filter_complex,
+        "-map",
+        "[final]",
+        "-c:a",
+        "aac",
+        "-b:a",
+        "128k",
+        str(output_path),
+    ]
+
+    logger.info(
+        "[bgm] sidechain mix: threshold=%.1fdB ratio=%.1f attack=%.3f release=%.3f",
+        bgm.sidechain_threshold,
+        ratio,
+        bgm.sidechain_attack,
+        bgm.sidechain_release,
+    )
+    run_ffmpeg(command)
+
+
+# ── 纯 BGM 模式（无主音频） ──────────────────────────────────────────────────
+
+
+def build_bgm_only(
+    ctx: "RenderContext",
+    bgm: BGMConfig,
+    target_duration: float,
+) -> Path:
+    """只有 BGM、没有主音频时，直接生成 BGM 音频.
+
+    Args:
+        ctx: 渲染上下文
+        bgm: BGM 配置
+        target_duration: 目标时长
+
+    Returns:
+        BGM 音频文件路径
+    """
+    output_path = ctx.work_dir / f"bgm_only_{ctx.plan_id}.aac"
+
+    if target_duration <= 0:
+        target_duration = 5.0
+
+    bgm_processed = prepare_bgm_track(ctx, bgm, target_duration)
+
+    # 直接复制
+    import shutil
+
+    shutil.copy2(bgm_processed, output_path)
+    return output_path

apps/worker/video_processing/chroma_key_engine.py

@@ -0,0 +1,248 @@
+"""绿幕抠像引擎 — 基于 FFmpeg colorkey / chromakey 滤镜.
+
+支持将指定颜色（默认绿色）变为透明，可用于虚拟背景、画中画背景替换等场景。
+
+使用方式：
+    config = ChromaKeyConfig(key_color="#00FF00", similarity=0.3, blend=0.1)
+    engine = ChromaKeyEngine(config)
+    filter_str = engine.build_filter(input_label, output_label)
+    # 结果: [in]colorkey=color=0x00FF00:similarity=0.3:blend=0.1[out]
+
+降级策略：
+    - 参数越界自动钳制
+    - 素材格式不支持时跳过（调用方捕获异常）
+"""
+
+from __future__ import annotations
+
+import logging
+import re
+from dataclasses import dataclass
+from typing import Optional
+
+logger = logging.getLogger(__name__)
+
+
+# ── 配置模型 ──────────────────────────────────────────────────────────────────
+
+
+@dataclass
+class ChromaKeyConfig:
+    """绿幕抠像配置。
+
+    Attributes:
+        enabled: 是否启用抠像
+        key_color: 要抠除的颜色，支持 hex 格式（如 "#00FF00"）或颜色名
+        similarity: 颜色相似度阈值 0.01~1.0，值越大抠除范围越大
+        blend: 边缘平滑/混合度 0.0~1.0，值越大边缘越柔和
+        spill_suppress: 溢色抑制 0.0~1.0，减少边缘的绿幕反光
+    """
+
+    enabled: bool = False
+    key_color: str = "#00FF00"
+    similarity: float = 0.3
+    blend: float = 0.1
+    spill_suppress: float = 0.0
+
+    @classmethod
+    def from_dict(cls, data: dict | None) -> "ChromaKeyConfig":
+        """从字典解析配置，参数越界自动钳制。"""
+        if not data or not data.get("enabled", False):
+            return cls(enabled=False)
+
+        key_color = str(data.get("key_color", "#00FF00")).strip()
+
+        def _safe_float(val, default):
+            try:
+                return float(val)
+            except (TypeError, ValueError):
+                return default
+
+        similarity = _safe_float(data.get("similarity", 0.3), 0.3)
+        blend = _safe_float(data.get("blend", 0.1), 0.1)
+        spill_suppress = _safe_float(data.get("spill_suppress", 0.0), 0.0)
+
+        # 钳制到合法范围
+        similarity = max(0.01, min(1.0, similarity))
+        blend = max(0.0, min(1.0, blend))
+        spill_suppress = max(0.0, min(1.0, spill_suppress))
+
+        return cls(
+            enabled=True,
+            key_color=key_color,
+            similarity=similarity,
+            blend=blend,
+            spill_suppress=spill_suppress,
+        )
+
+    def has_effect(self) -> bool:
+        """判断是否有实际抠像效果。"""
+        return self.enabled and self.similarity > 0
+
+
+# ── 预设配置 ──────────────────────────────────────────────────────────────────
+
+# 常见绿幕/蓝幕预设
+CHROMA_KEY_PRESETS = {
+    "green_screen": {
+        "key_color": "#00FF00",
+        "similarity": 0.3,
+        "blend": 0.1,
+        "spill_suppress": 0.5,
+    },
+    "blue_screen": {
+        "key_color": "#0000FF",
+        "similarity": 0.3,
+        "blend": 0.1,
+        "spill_suppress": 0.5,
+    },
+    "red_screen": {
+        "key_color": "#FF0000",
+        "similarity": 0.3,
+        "blend": 0.1,
+        "spill_suppress": 0.0,
+    },
+    "precise_green": {
+        "key_color": "#00FF00",
+        "similarity": 0.2,
+        "blend": 0.05,
+        "spill_suppress": 0.3,
+    },
+    "soft_green": {
+        "key_color": "#00FF00",
+        "similarity": 0.45,
+        "blend": 0.2,
+        "spill_suppress": 0.5,
+    },
+}
+
+
+# ── 引擎实现 ──────────────────────────────────────────────────────────────────
+
+
+class ChromaKeyEngine:
+    """绿幕抠像引擎。
+
+    基于 FFmpeg colorkey 滤镜实现，将指定颜色变为透明。
+    适用于绿幕/蓝幕视频的背景去除，配合画中画或 overlay 实现虚拟背景。
+    """
+
+    def __init__(self, config: ChromaKeyConfig):
+        self.config = config
+
+    @staticmethod
+    def _normalize_color(color_str: str) -> str:
+        """将颜色字符串转为 FFmpeg colorkey 接受的格式。
+
+        支持:
+          - "#RRGGBB" / "#RRGGBBAA" → 0xRRGGBB
+          - "0xRRGGBB" → 直接使用
+          - 颜色名（green/blue/red/black/white 等）→ 直接透传
+        """
+        color = color_str.strip()
+
+        # hex 格式
+        hex_match = re.match(r"^#?([0-9a-fA-F]{6})([0-9a-fA-F]{2})?$", color)
+        if hex_match:
+            return f"0x{hex_match.group(1).upper()}"
+
+        # 已经是 0x 格式
+        if color.lower().startswith("0x"):
+            return color.upper()
+
+        # 颜色名直接透传（FFmpeg 支持常见颜色名）
+        return color
+
+    def build_filter(self, input_label: str, output_label: str) -> str:
+        """构建 colorkey 滤镜字符串。
+
+        Args:
+            input_label: 输入标签，如 "[0:v]" 或 "[v0]"
+            output_label: 输出标签，如 "[ck0]"
+
+        Returns:
+            FFmpeg 滤镜字符串，如 "[v0]colorkey=color=0x00FF00:similarity=0.3:blend=0.1[ck0]"
+
+        Raises:
+            ValueError: 配置无效时抛出（调用方应捕获并降级）
+        """
+        if not self.config.has_effect():
+            # 无效果，直接直通
+            return f"{input_label}copy{output_label}"
+
+        color = self._normalize_color(self.config.key_color)
+        similarity = self.config.similarity
+        blend = self.config.blend
+
+        # 基础 colorkey 滤镜
+        parts = [f"colorkey=color={color}:similarity={similarity}:blend={blend}"]
+
+        # 溢色抑制（通过 colorchannelmixer 降低绿色通道增益）
+        if self.config.spill_suppress > 0:
+            # 降低绿通道增益，减少绿幕反光溢出
+            spill = self.config.spill_suppress
+            # 绿通道增益 = 1 - spill_factor
+            g_gain = max(0.3, 1.0 - spill * 0.7)
+            # 同时稍微提升红和蓝来补偿色偏
+            r_gain = 1.0 + spill * 0.15
+            b_gain = 1.0 + spill * 0.15
+            parts.append(f"colorchannelmixer=" f"rr={r_gain}:" f"gg={g_gain}:" f"bb={b_gain}:" f"aa=1")
+
+        filter_str = f"{input_label}{','.join(parts)}{output_label}"
+        return filter_str
+
+    def build_filter_chromakey(self, input_label: str, output_label: str) -> str:
+        """使用 chromakey 滤镜（更高级的版本，支持更多参数）。
+
+        注意：并非所有 FFmpeg 版本都支持 chromakey 滤镜，
+        优先使用 colorkey（兼容性更好）。
+
+        Args:
+            input_label: 输入标签
+            output_label: 输出标签
+
+        Returns:
+            FFmpeg 滤镜字符串
+        """
+        if not self.config.has_effect():
+            return f"{input_label}copy{output_label}"
+
+        color = self._normalize_color(self.config.key_color)
+        similarity = self.config.similarity
+        blend = self.config.blend
+
+        return f"{input_label}" f"chromakey=color={color}:similarity={similarity}:blend={blend}" f"{output_label}"
+
+
+def apply_chroma_key_if_needed(
+    clip_config: dict | None,
+    input_label: str,
+    output_label: str,
+) -> Optional[str]:
+    """便捷函数：根据 clip 配置判断是否需要应用绿幕抠像。
+
+    Args:
+        clip_config: clip 的 config 字典
+        input_label: 输入标签
+        output_label: 输出标签
+
+    Returns:
+        滤镜字符串，不需要抠像时返回 None
+    """
+    if not clip_config:
+        return None
+
+    chroma_key_data = clip_config.get("chroma_key")
+    if not chroma_key_data:
+        return None
+
+    try:
+        config = ChromaKeyConfig.from_dict(chroma_key_data)
+        if not config.has_effect():
+            return None
+
+        engine = ChromaKeyEngine(config)
+        return engine.build_filter(input_label, output_label)
+    except Exception as e:
+        logger.warning("[chroma-key] 应用抠像失败，跳过: %s", e)
+        return None

apps/worker/video_processing/color_grade_engine.py

@@ -0,0 +1,416 @@
+"""滤镜调色引擎 — 基于 FFmpeg eq + colorbalance + hue + curves 滤镜组合实现画面色彩调整.
+
+支持能力：
+- 基础调色参数：亮度、对比度、饱和度、色温、色调
+- 8种风格预设：清新、日系、复古、电影、胶片、黑白、暖色、冷色
+- 分段应用：每个 clip 可独立设置不同滤镜
+- 降级策略：参数越界自动钳制，不阻断渲染
+"""
+
+from __future__ import annotations
+
+import logging
+from dataclasses import dataclass, field
+from typing import Any
+
+logger = logging.getLogger(__name__)
+
+
+# ── 预设滤镜包 ────────────────────────────────────────────────────────────────
+
+# 预设名称常量
+PRESET_FRESH = "fresh"  # 清新
+PRESET_JAPANESE = "japanese"  # 日系
+PRESET_VINTAGE = "vintage"  # 复古
+PRESET_CINEMA = "cinema"  # 电影
+PRESET_FILM = "film"  # 胶片
+PRESET_BW = "black_white"  # 黑白
+PRESET_WARM = "warm"  # 暖色
+PRESET_COOL = "cool"  # 冷色
+
+VALID_PRESETS = {
+    PRESET_FRESH,
+    PRESET_JAPANESE,
+    PRESET_VINTAGE,
+    PRESET_CINEMA,
+    PRESET_FILM,
+    PRESET_BW,
+    PRESET_WARM,
+    PRESET_COOL,
+}
+
+# 预设名称 → 中文显示名
+PRESET_DISPLAY_NAMES = {
+    PRESET_FRESH: "清新",
+    PRESET_JAPANESE: "日系",
+    PRESET_VINTAGE: "复古",
+    PRESET_CINEMA: "电影",
+    PRESET_FILM: "胶片",
+    PRESET_BW: "黑白",
+    PRESET_WARM: "暖色",
+    PRESET_COOL: "冷色",
+}
+
+# 预设参数配置
+# 每个预设包含：brightness, contrast, saturation, temperature, hue
+# 取值范围：brightness/contrast/temperature -100~100, saturation 0~200, hue -180~180
+PRESET_PARAMS: dict[str, dict[str, float]] = {
+    PRESET_FRESH: {
+        # 清新：提亮、高饱和、偏冷、微微调
+        "brightness": 8,
+        "contrast": 10,
+        "saturation": 120,
+        "temperature": -8,
+        "hue": 5,
+    },
+    PRESET_JAPANESE: {
+        # 日系：低对比、低饱和、偏暖、偏黄绿
+        "brightness": 12,
+        "contrast": -15,
+        "saturation": 70,
+        "temperature": 10,
+        "hue": -5,
+    },
+    PRESET_VINTAGE: {
+        # 复古：低饱和、偏黄、对比度适中、偏暖
+        "brightness": -5,
+        "contrast": 5,
+        "saturation": 60,
+        "temperature": 25,
+        "hue": -8,
+    },
+    PRESET_CINEMA: {
+        # 电影：高对比、低饱和、偏冷蓝、暗角感
+        "brightness": -8,
+        "contrast": 20,
+        "saturation": 75,
+        "temperature": -15,
+        "hue": -3,
+    },
+    PRESET_FILM: {
+        # 胶片：中对比、饱和适中、偏暖、颗粒感（这里只用调色模拟）
+        "brightness": -3,
+        "contrast": 12,
+        "saturation": 95,
+        "temperature": 15,
+        "hue": -2,
+    },
+    PRESET_BW: {
+        # 黑白：饱和度为0，对比度略高
+        "brightness": 0,
+        "contrast": 15,
+        "saturation": 0,
+        "temperature": 0,
+        "hue": 0,
+    },
+    PRESET_WARM: {
+        # 暖色：高色温、偏红黄
+        "brightness": 5,
+        "contrast": 8,
+        "saturation": 110,
+        "temperature": 30,
+        "hue": -5,
+    },
+    PRESET_COOL: {
+        # 冷色：低色温、偏蓝青
+        "brightness": 3,
+        "contrast": 8,
+        "saturation": 105,
+        "temperature": -25,
+        "hue": 8,
+    },
+}
+
+
+# ── 参数范围 ──────────────────────────────────────────────────────────────────
+
+PARAM_RANGES = {
+    "brightness": (-100.0, 100.0),
+    "contrast": (-100.0, 100.0),
+    "saturation": (0.0, 200.0),
+    "temperature": (-100.0, 100.0),
+    "hue": (-180.0, 180.0),
+}
+
+# 默认值（零调整）
+DEFAULT_PARAMS = {
+    "brightness": 0.0,
+    "contrast": 0.0,
+    "saturation": 100.0,
+    "temperature": 0.0,
+    "hue": 0.0,
+}
+
+
+# ── 数据模型 ──────────────────────────────────────────────────────────────────
+
+
+@dataclass
+class ColorGradeConfig:
+    """色彩调色配置.
+
+    优先级：自定义参数 > 预设参数
+    即：先加载预设的基础参数，再用 custom 中显式指定的参数覆盖
+    """
+
+    enabled: bool = False
+    preset: str = ""  # 预设名称，空表示不使用预设
+    # 自定义参数覆盖（None 表示不覆盖，使用预设值或默认值）
+    brightness: float | None = None
+    contrast: float | None = None
+    saturation: float | None = None
+    temperature: float | None = None
+    hue: float | None = None
+
+    def resolve_params(self) -> dict[str, float]:
+        """解析最终调色参数（预设 + 自定义覆盖 + 边界钳制）.
+
+        Returns:
+            包含 brightness, contrast, saturation, temperature, hue 的参数字典
+        """
+        # 1. 从默认值开始
+        params = dict(DEFAULT_PARAMS)
+
+        # 2. 应用预设
+        if self.preset and self.preset in PRESET_PARAMS:
+            params.update(PRESET_PARAMS[self.preset])
+
+        # 3. 应用自定义覆盖
+        if self.brightness is not None:
+            params["brightness"] = self.brightness
+        if self.contrast is not None:
+            params["contrast"] = self.contrast
+        if self.saturation is not None:
+            params["saturation"] = self.saturation
+        if self.temperature is not None:
+            params["temperature"] = self.temperature
+        if self.hue is not None:
+            params["hue"] = self.hue
+
+        # 4. 边界钳制
+        for key, (min_val, max_val) in PARAM_RANGES.items():
+            params[key] = max(min_val, min(max_val, params[key]))
+
+        return params
+
+    def has_effect(self) -> bool:
+        """判断是否有实际调色效果（所有参数都是默认值则无效果）.
+
+        用于优化：无效果时跳过滤镜，不浪费性能。
+        """
+        params = self.resolve_params()
+        for key, default in DEFAULT_PARAMS.items():
+            if abs(params[key] - default) > 0.001:
+                return True
+        return False
+
+    @classmethod
+    def from_dict(cls, data: dict[str, Any] | None) -> "ColorGradeConfig":
+        """从字典解析配置."""
+        if not data or not data.get("enabled", False):
+            return cls(enabled=False)
+
+        preset = data.get("preset", "")
+        if preset and preset not in VALID_PRESETS:
+            logger.warning("未知的调色预设: %s，忽略预设", preset)
+            preset = ""
+
+        def _get_float(key: str) -> float | None:
+            val = data.get(key)
+            if val is None:
+                return None
+            try:
+                return float(val)
+            except (ValueError, TypeError):
+                return None
+
+        try:
+            return cls(
+                enabled=True,
+                preset=preset,
+                brightness=_get_float("brightness"),
+                contrast=_get_float("contrast"),
+                saturation=_get_float("saturation"),
+                temperature=_get_float("temperature"),
+                hue=_get_float("hue"),
+            )
+        except Exception as e:
+            logger.warning("调色配置解析失败: %s，使用默认配置", e)
+            return cls(enabled=False)
+
+
+# ── 调色引擎 ──────────────────────────────────────────────────────────────────
+
+
+class ColorGradeEngine:
+    """滤镜调色引擎 — 生成 FFmpeg 调色滤镜链.
+
+    滤镜组合策略：
+    1. eq 滤镜：调整亮度(brightness)、对比度(contrast)、饱和度(saturation)
+    2. colorbalance 滤镜：调整色温（通过调整红/青、黄/蓝平衡）
+    3. hue 滤镜：调整色调
+
+    所有参数转换公式：
+    - brightness: 用户值 -100~100 → FFmpeg eq brightness -1.0~1.0
+    - contrast: 用户值 -100~100 → FFmpeg eq contrast -1000~1000（非线性映射）
+    - saturation: 用户值 0~200 → FFmpeg eq saturation 0.0~2.0
+    - temperature: 用户值 -100~100 → colorbalance 红/蓝通道偏移
+    - hue: 用户值 -180~180 → FFmpeg hue H -180~180（度）
+    """
+
+    @staticmethod
+    def _map_brightness(value: float) -> float:
+        """用户亮度值 → FFmpeg eq brightness.
+
+        用户范围 -100~100 → FFmpeg范围 -1.0~1.0
+        """
+        return value / 100.0
+
+    @staticmethod
+    def _map_contrast(value: float) -> float:
+        """用户对比度值 → FFmpeg eq contrast.
+
+        用户范围 -100~100 → FFmpeg范围 -2.0~2.0
+        注：FFmpeg eq 的 contrast 公式为 linear gain，1.0 为原始
+        -2 ~ 2 的范围对应 ~-1000 ~ 1000 的老式定义的约 -66% ~ +100%
+        """
+        if value >= 0:
+            # 正向：0~100 → 1.0~2.0
+            return 1.0 + value / 100.0
+        else:
+            # 负向：-100~0 → 0.0~1.0
+            return 1.0 + value / 100.0  # value为负数，相当于 1.0 - |value|/100
+
+    @staticmethod
+    def _map_saturation(value: float) -> float:
+        """用户饱和度 → FFmpeg eq saturation.
+
+        用户范围 0~200 → FFmpeg范围 0.0~2.0
+        """
+        return value / 100.0
+
+    @staticmethod
+    def _map_temperature(value: float) -> tuple[float, float, float]:
+        """用户色温值 → colorbalance 三个通道参数.
+
+        返回：(red, green, blue) — 每个通道 -1.0~1.0 的偏移
+
+        色温为正（暖）：增加红、减蓝
+        色温为负（冷）：减红、加蓝
+        """
+        # -100~100 → -0.5~0.5
+        normalized = value / 200.0
+
+        if normalized >= 0:
+            # 暖色调：红+，绿微+，蓝-
+            red = normalized * 0.8
+            green = normalized * 0.3
+            blue = -normalized * 0.8
+        else:
+            # 冷色调：红-，绿微+，蓝+
+            red = normalized * 0.8  # 负数
+            green = -normalized * 0.2  # 正数（冷色也加点绿让它偏青）
+            blue = -normalized * 0.8  # 正数
+
+        return (red, green, blue)
+
+    @staticmethod
+    def _map_hue(value: float) -> float:
+        """用户色调值 → FFmpeg hue滤镜角度.
+
+        用户范围 -180~180 → FFmpeg H -180~180
+        """
+        return value
+
+    @classmethod
+    def build_filter(cls, config: ColorGradeConfig, input_label: str = "", output_label: str = "") -> str:
+        """构建调色滤镜字符串.
+
+        Args:
+            config: 调色配置
+            input_label: 输入标签（带方括号，如 "[0:v]"），空则无
+            output_label: 输出标签（带方括号，如 "[graded]"），空则无
+
+        Returns:
+            FFmpeg 滤镜字符串，如 "[0:v]eq=brightness=0.1:contrast=1.2,hue=H=10[graded]"
+        """
+        if not config.enabled or not config.has_effect():
+            # 无效果时直通
+            if input_label and output_label:
+                return f"{input_label}copy{output_label}"
+            return ""
+
+        params = config.resolve_params()
+        filters: list[str] = []
+
+        # 1. eq 滤镜：亮度 + 对比度 + 饱和度
+        eq_parts: list[str] = []
+        brightness = cls._map_brightness(params["brightness"])
+        contrast = cls._map_contrast(params["contrast"])
+        saturation = cls._map_saturation(params["saturation"])
+
+        if abs(brightness) > 0.001:
+            eq_parts.append(f"brightness={brightness:.3f}")
+        if abs(contrast - 1.0) > 0.001:
+            eq_parts.append(f"contrast={contrast:.3f}")
+        if abs(saturation - 1.0) > 0.001:
+            eq_parts.append(f"saturation={saturation:.3f}")
+
+        if eq_parts:
+            filters.append(f"eq={':'.join(eq_parts)}")
+
+        # 2. colorbalance 滤镜：色温
+        if abs(params["temperature"]) > 0.001:
+            red, green, blue = cls._map_temperature(params["temperature"])
+            cb_parts = []
+            # 调整阴影/中间调/高光的平衡（简化：全部统一调整）
+            if abs(red) > 0.001:
+                cb_parts.append(f"rs={red:.3f}")
+                cb_parts.append(f"rm={red:.3f}")
+                cb_parts.append(f"rh={red:.3f}")
+            if abs(green) > 0.001:
+                cb_parts.append(f"gs={green:.3f}")
+                cb_parts.append(f"gm={green:.3f}")
+                cb_parts.append(f"gh={green:.3f}")
+            if abs(blue) > 0.001:
+                cb_parts.append(f"bs={blue:.3f}")
+                cb_parts.append(f"bm={blue:.3f}")
+                cb_parts.append(f"bh={blue:.3f}")
+            if cb_parts:
+                filters.append(f"colorbalance={':'.join(cb_parts)}")
+
+        # 3. hue 滤镜：色调
+        if abs(params["hue"]) > 0.001:
+            hue_val = cls._map_hue(params["hue"])
+            filters.append(f"hue=h={hue_val:.1f}")
+
+        if not filters:
+            # 理论上不会到这里（has_effect 已判断），保险起见
+            if input_label and output_label:
+                return f"{input_label}copy{output_label}"
+            return ""
+
+        filter_str = ",".join(filters)
+        if input_label:
+            filter_str = f"{input_label}{filter_str}"
+        if output_label:
+            filter_str = f"{filter_str}{output_label}"
+
+        return filter_str
+
+
+# ── 便捷函数 ──────────────────────────────────────────────────────────────────
+
+
+def get_preset_names() -> list[tuple[str, str]]:
+    """获取所有预设名称列表.
+
+    Returns:
+        [(preset_key, display_name), ...]
+    """
+    return [(key, PRESET_DISPLAY_NAMES.get(key, key)) for key in PRESET_PARAMS.keys()]
+
+
+def get_preset_params(preset: str) -> dict[str, float] | None:
+    """获取指定预设的参数."""
+    return PRESET_PARAMS.get(preset)

apps/worker/video_processing/cover_generator.py

@@ -0,0 +1,431 @@
+"""视频封面生成器 — 从视频中提取/生成封面图.
+
+支持能力：
+- 指定时间点抽帧（默认第1秒）
+- 智能封面：抽取多帧选最清晰的一帧
+- 自定义上传封面图（直接返回路径）
+- 生成的封面图保存为 JPEG 格式，可复用
+"""
+
+from __future__ import annotations
+
+import logging
+import subprocess
+from pathlib import Path
+from typing import Any
+
+from video_processing.ffmpeg_utils import FFMPEG_BIN, probe_video_info, run_ffmpeg
+
+logger = logging.getLogger(__name__)
+
+
+# ── 配置常量 ──────────────────────────────────────────────────────────────────
+
+# 智能封面抽帧数量
+SMART_COVER_FRAME_COUNT = 3
+
+# 默认抽帧时间点（秒）
+DEFAULT_COVER_TIME = 1.0
+
+# 封面输出尺寸（宽x高）
+DEFAULT_COVER_WIDTH = 1080
+DEFAULT_COVER_HEIGHT = 1920
+
+# 封面质量（JPEG quality 1-31，越小质量越高）
+DEFAULT_COVER_QUALITY = 5
+
+
+# ── 数据模型 ──────────────────────────────────────────────────────────────────
+
+
+class CoverGenerator:
+    """视频封面生成器.
+
+    三种模式：
+    1. 指定时间点抽帧：从视频指定时间提取一帧
+    2. 智能封面：抽取3帧，用 blur 检测选最清晰的
+    3. 自定义上传：直接使用用户上传的图片
+    """
+
+    @staticmethod
+    def extract_frame(
+        video_path: str | Path,
+        output_path: str | Path,
+        *,
+        time_sec: float = DEFAULT_COVER_TIME,
+        width: int = DEFAULT_COVER_WIDTH,
+        height: int = DEFAULT_COVER_HEIGHT,
+        quality: int = DEFAULT_COVER_QUALITY,
+    ) -> Path:
+        """从视频指定时间点提取一帧作为封面.
+
+        Args:
+            video_path: 视频文件路径
+            output_path: 输出图片路径
+            time_sec: 抽帧时间点（秒）
+            width: 输出宽度
+            height: 输出高度
+            quality: JPEG 质量（1-31，越小越好）
+
+        Returns:
+            封面图片路径
+
+        Raises:
+            FileNotFoundError: 视频文件不存在
+            subprocess.CalledProcessError: FFmpeg 执行失败
+        """
+        video_path = Path(video_path)
+        output_path = Path(output_path)
+
+        if not video_path.exists():
+            raise FileNotFoundError(f"视频文件不存在: {video_path}")
+
+        # 确保输出目录存在
+        output_path.parent.mkdir(parents=True, exist_ok=True)
+
+        # 安全钳制时间
+        info = probe_video_info(str(video_path))
+        duration = info.get("duration", 0.0)
+        if duration > 0 and time_sec >= duration:
+            # 超过视频长度，取中间帧
+            time_sec = max(0, duration / 2)
+        if time_sec < 0:
+            time_sec = 0
+
+        # scale + crop 实现 cover 裁剪（铺满输出尺寸）
+        vf = f"scale={width}:{height}:force_original_aspect_ratio=increase," f"crop={width}:{height}"
+
+        command = [
+            FFMPEG_BIN,
+            "-y",
+            "-ss",
+            f"{time_sec:.3f}",
+            "-i",
+            str(video_path),
+            "-vframes",
+            "1",
+            "-vf",
+            vf,
+            "-q:v",
+            str(quality),
+            "-f",
+            "mjpeg",
+            str(output_path),
+        ]
+
+        logger.info("抽取视频封面: video=%s time=%.2fs output=%s", video_path.name, time_sec, output_path.name)
+        run_ffmpeg(command)
+
+        if not output_path.exists() or output_path.stat().st_size == 0:
+            raise RuntimeError(f"封面生成失败: {output_path}")
+
+        return output_path
+
+    @staticmethod
+    def extract_smart_cover(
+        video_path: str | Path,
+        output_path: str | Path,
+        *,
+        frame_count: int = SMART_COVER_FRAME_COUNT,
+        width: int = DEFAULT_COVER_WIDTH,
+        height: int = DEFAULT_COVER_HEIGHT,
+        quality: int = DEFAULT_COVER_QUALITY,
+        work_dir: str | Path | None = None,
+    ) -> Path:
+        """智能封面：抽取多帧，选最清晰的一帧.
+
+        清晰度判断：使用拉普拉斯方差（Variance of Laplacian），
+        方差越大表示图像边缘越丰富，越清晰。
+
+        Args:
+            video_path: 视频文件路径
+            output_path: 最终输出封面路径
+            frame_count: 抽帧数量（均匀分布在视频中）
+            width: 输出宽度
+            height: 输出高度
+            quality: JPEG 质量
+            work_dir: 临时工作目录（默认输出目录的父目录）
+
+        Returns:
+            最佳封面图片路径
+        """
+        video_path = Path(video_path)
+        output_path = Path(output_path)
+
+        if not video_path.exists():
+            raise FileNotFoundError(f"视频文件不存在: {video_path}")
+
+        # 获取视频时长
+        info = probe_video_info(str(video_path))
+        duration = info.get("duration", 0.0)
+
+        if duration <= 0 or frame_count <= 1:
+            # 无法获取时长或只有1帧，退化为普通抽帧
+            return CoverGenerator.extract_frame(
+                video_path,
+                output_path,
+                time_sec=min(DEFAULT_COVER_TIME, max(0, duration / 2)),
+                width=width,
+                height=height,
+                quality=quality,
+            )
+
+        # 临时目录
+        if work_dir is None:
+            work_dir = output_path.parent
+        work_dir = Path(work_dir)
+        work_dir.mkdir(parents=True, exist_ok=True)
+
+        # 均匀分布抽帧时间点（跳过首尾5%）
+        start_pct = 0.05
+        end_pct = 0.95
+        if frame_count == 1:
+            time_points = [duration * 0.5]
+        else:
+            step = (end_pct - start_pct) / (frame_count - 1)
+            time_points = [duration * (start_pct + step * i) for i in range(frame_count)]
+
+        # 抽取候选帧
+        candidate_frames: list[tuple[float, Path]] = []
+        for i, t in enumerate(time_points):
+            frame_path = work_dir / f"cover_candidate_{i}.jpg"
+            try:
+                CoverGenerator.extract_frame(
+                    video_path,
+                    frame_path,
+                    time_sec=t,
+                    width=width,
+                    height=height,
+                    quality=quality,
+                )
+                candidate_frames.append((t, frame_path))
+            except Exception as e:
+                logger.warning("智能封面抽帧失败（t=%.2fs）: %s", t, e)
+                continue
+
+        if not candidate_frames:
+            # 全部失败，退化到普通抽帧
+            logger.warning("智能封面所有候选帧抽取失败，退化为普通抽帧")
+            return CoverGenerator.extract_frame(
+                video_path,
+                output_path,
+                time_sec=min(DEFAULT_COVER_TIME, duration / 2),
+                width=width,
+                height=height,
+                quality=quality,
+            )
+
+        if len(candidate_frames) == 1:
+            # 只有一帧，直接用
+            import shutil
+
+            shutil.copy2(candidate_frames[0][1], output_path)
+            return output_path
+
+        # 计算每帧清晰度（用 FFmpeg 的 stats 滤镜或简化处理）
+        # 简化方案：比较文件大小（同一尺寸下，JPEG文件越大通常细节越丰富、越清晰）
+        # 更准确的方案是用拉普拉斯方差，但需要额外依赖
+        # 这里用文件大小作为近似指标
+        best_frame = max(candidate_frames, key=lambda x: x[1].stat().st_size)
+
+        # 复制最佳帧到输出路径
+        import shutil
+
+        shutil.copy2(best_frame[1], output_path)
+
+        logger.info(
+            "智能封面生成完成: 候选%d帧, 最佳t=%.2fs, 大小=%d字节",
+            len(candidate_frames),
+            best_frame[0],
+            output_path.stat().st_size,
+        )
+
+        # 清理临时文件
+        for _, fp in candidate_frames:
+            try:
+                fp.unlink()
+            except OSError:
+                pass
+
+        return output_path
+
+    @staticmethod
+    def process_custom_cover(
+        image_path: str | Path,
+        output_path: str | Path,
+        *,
+        width: int = DEFAULT_COVER_WIDTH,
+        height: int = DEFAULT_COVER_HEIGHT,
+        quality: int = DEFAULT_COVER_QUALITY,
+    ) -> Path:
+        """处理用户自定义上传的封面图.
+
+        调整尺寸、格式转换为标准封面格式。
+
+        Args:
+            image_path: 用户上传的图片路径
+            output_path: 输出封面路径
+            width: 目标宽度
+            height: 目标高度
+            quality: JPEG 质量
+
+        Returns:
+            处理后的封面图片路径
+        """
+        image_path = Path(image_path)
+        output_path = Path(output_path)
+
+        if not image_path.exists():
+            raise FileNotFoundError(f"封面图片不存在: {image_path}")
+
+        output_path.parent.mkdir(parents=True, exist_ok=True)
+
+        # scale + crop 实现 cover 裁剪
+        vf = f"scale={width}:{height}:force_original_aspect_ratio=increase," f"crop={width}:{height}"
+
+        command = [
+            FFMPEG_BIN,
+            "-y",
+            "-i",
+            str(image_path),
+            "-vf",
+            vf,
+            "-q:v",
+            str(quality),
+            "-f",
+            "mjpeg",
+            str(output_path),
+        ]
+
+        logger.info("处理自定义封面: input=%s output=%s", image_path.name, output_path.name)
+
+        try:
+            run_ffmpeg(command)
+        except subprocess.CalledProcessError:
+            # 处理失败，直接复制原图
+            logger.warning("自定义封面处理失败，使用原图")
+            import shutil
+
+            shutil.copy2(image_path, output_path)
+
+        return output_path
+
+    @staticmethod
+    def generate_cover(
+        video_path: str | Path,
+        output_path: str | Path,
+        *,
+        mode: str = "smart",  # smart / time / custom
+        time_sec: float = DEFAULT_COVER_TIME,
+        custom_image: str | Path | None = None,
+        width: int = DEFAULT_COVER_WIDTH,
+        height: int = DEFAULT_COVER_HEIGHT,
+        quality: int = DEFAULT_COVER_QUALITY,
+    ) -> Path:
+        """统一封面生成入口.
+
+        Args:
+            video_path: 视频文件路径
+            output_path: 输出封面路径
+            mode: 模式 - smart（智能选帧）/ time（指定时间）/ custom（自定义图片）
+            time_sec: time 模式下的抽帧时间点
+            custom_image: custom 模式下的自定义图片路径
+            width: 输出宽度
+            height: 输出高度
+            quality: JPEG 质量
+
+        Returns:
+            封面图片路径
+        """
+        if mode == "custom" and custom_image:
+            return CoverGenerator.process_custom_cover(
+                custom_image,
+                output_path,
+                width=width,
+                height=height,
+                quality=quality,
+            )
+        elif mode == "time":
+            return CoverGenerator.extract_frame(
+                video_path,
+                output_path,
+                time_sec=time_sec,
+                width=width,
+                height=height,
+                quality=quality,
+            )
+        else:
+            # 默认智能封面
+            return CoverGenerator.extract_smart_cover(
+                video_path,
+                output_path,
+                width=width,
+                height=height,
+                quality=quality,
+            )
+
+
+# ── 便捷函数 ──────────────────────────────────────────────────────────────────
+
+
+def generate_cover_from_plan(
+    plan: Any,
+    video_path: str | Path,
+    output_dir: str | Path,
+) -> Path | None:
+    """从 EditPlan 配置生成封面图.
+
+    配置读取：plan.config.cover_config
+    支持字段：
+      - mode: smart / time / custom
+      - time_sec: 抽帧时间（time模式）
+      - custom_image_url: 自定义图片URL（需要先下载到本地）
+
+    Args:
+        plan: EditPlan 对象
+        video_path: 渲染后的视频路径
+        output_dir: 封面输出目录
+
+    Returns:
+        封面图片路径，或 None（不需要生成封面时）
+    """
+    config = getattr(plan, "config", None) or {}
+    cover_config = config.get("cover_config") if isinstance(config, dict) else None
+
+    if not cover_config:
+        return None
+
+    mode = cover_config.get("mode", "smart")
+    output_dir = Path(output_dir)
+    output_dir.mkdir(parents=True, exist_ok=True)
+    output_path = output_dir / f"cover_{plan.id}.jpg"
+
+    try:
+        if mode == "custom":
+            # 自定义封面：需要先有本地图片路径
+            custom_path = cover_config.get("custom_image_path")
+            if custom_path and Path(custom_path).exists():
+                return CoverGenerator.process_custom_cover(
+                    custom_path,
+                    output_path,
+                )
+            else:
+                logger.warning("自定义封面图片路径无效，退化为智能封面")
+                mode = "smart"
+
+        if mode == "time":
+            time_sec = float(cover_config.get("time_sec", DEFAULT_COVER_TIME))
+            return CoverGenerator.extract_frame(
+                video_path,
+                output_path,
+                time_sec=time_sec,
+            )
+        else:
+            # smart
+            return CoverGenerator.extract_smart_cover(
+                video_path,
+                output_path,
+            )
+    except Exception as e:
+        logger.warning("封面生成失败: %s", e)
+        return None

apps/worker/video_processing/dedup_helpers.py

@@ -75,6 +75,19 @@ def create_video_record_and_dedup(
         video_repo = SQLAlchemyGeneratedVideoRepository(session)
         video_repo.create(generated_video)
 
+        # 生成封面缩略图
+        thumbnail_storage_key = f"generated/projects/{project_id}/thumbnails/{video_id}.jpg"
+        try:
+            from video_processing.thumbnail_generator import generate_and_upload_thumbnail
+
+            thumbnail_url = generate_and_upload_thumbnail(video_path, thumbnail_storage_key)
+            if thumbnail_url:
+                generated_video.thumbnail_url = thumbnail_url
+                video_repo.update_thumbnail(video_id, thumbnail_url)
+                logger.info("Thumbnail generated for video %s: %s", video_id, thumbnail_url)
+        except Exception as thumb_err:
+            logger.warning("Thumbnail generation failed for %s: %s", video_id, thumb_err)
+
         # 计算视频指纹
         deduplicator = VideoDeduplicator()
         try:

apps/worker/video_processing/ffmpeg_utils.py

@@ -25,8 +25,14 @@ DEFAULT_FPS = 25
 
 # xfade 转场映射：transition_effect 名称 → FFmpeg xfade transition 名称
 # 键同时支持 TransitionEffect 枚举值和字符串名称（向后兼容）
+# "cut" 为特殊值：硬切，不使用 xfade（由调用方特殊处理）
 XFADE_TRANSITION_MAP: dict[str, str] = {
+    # 基础
     "fade": "fade",
+    "dissolve": "dissolve",
+    "crossfade": "dissolve",
+    "crossdissolve": "dissolve",
+    # 滑入系列
     "slideleft": "slideleft",
     "slide_left": "slideleft",
     "slideright": "slideright",
@@ -35,9 +41,22 @@ XFADE_TRANSITION_MAP: dict[str, str] = {
     "slide_up": "slideup",
     "slidedown": "slidedown",
     "slide_down": "slidedown",
-    "dissolve": "dissolve",
-    "wipe": "wipeleft",
+    "slide": "slideleft",  # 默认向左滑
+    # 缩放
+    "zoom": "zoomin",
+    "zoomin": "zoomin",
+    "zoomout": "zoomout",
+    # 擦除系列
+    "wipe": "wipeleft",  # 默认向左擦
     "wipeleft": "wipeleft",
+    "wiperight": "wiperight",
+    "wipeup": "wipeup",
+    "wipedown": "wipedown",
+    # 特殊效果
+    "circlecrop": "circlecrop",
+    "circle": "circlecrop",
+    "rectcrop": "rectcrop",
+    "rect": "rectcrop",
 }
 
 DEFAULT_TRANSITION_DURATION = 0.5

apps/worker/video_processing/intro_outro_engine.py

@@ -0,0 +1,421 @@
+"""片头片尾引擎 — 视频包装与品牌标识.
+
+支持：
+- 片头：视频片段 或 纯文字片头（背景色 + 标题 + 副标题）
+- 片尾：视频片段 或 关注引导片尾
+- 自动与正片拼接（xfade 转场）
+- 时长可配置
+"""
+
+from __future__ import annotations
+
+import logging
+import subprocess
+from dataclasses import dataclass
+from pathlib import Path
+from typing import Any
+
+from video_processing.ffmpeg_utils import FFMPEG_BIN, run_ffmpeg
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class IntroOutroConfig:
+    """片头片尾配置.
+
+    type: "video" 视频片段 | "text" 纯文字 | "none" 不启用
+    """
+
+    enabled: bool = False
+
+    # 片头
+    intro_type: str = "none"  # none | video | text
+    intro_video_path: str = ""  # 视频片段路径
+    intro_duration: float = 3.0  # 片头时长（秒）
+
+    # 文字片头配置
+    intro_background: str = "#000000"  # 背景色
+    intro_title: str = ""
+    intro_subtitle: str = ""
+    intro_title_color: str = "white"
+    intro_title_size: int = 48
+    intro_subtitle_color: str = "gray"
+    intro_subtitle_size: int = 24
+
+    # 片尾
+    outro_type: str = "none"  # none | video | text | follow
+    outro_video_path: str = ""  # 视频片段路径
+    outro_duration: float = 3.0  # 片尾时长（秒）
+
+    # 文字片尾配置
+    outro_background: str = "#000000"
+    outro_title: str = "感谢观看"
+    outro_subtitle: str = "点赞关注不迷路"
+    outro_title_color: str = "white"
+    outro_title_size: int = 48
+    outro_subtitle_color: str = "gray"
+    outro_subtitle_size: int = 24
+
+    # 转场
+    transition_effect: str = "fade"
+    transition_duration: float = 0.5
+
+    @classmethod
+    def from_dict(cls, data: dict[str, Any] | None) -> IntroOutroConfig:
+        """从字典构造."""
+        if not data:
+            return cls()
+
+        enabled = data.get("enabled", False)
+        if not enabled:
+            return cls()
+
+        intro = data.get("intro", {}) or {}
+        outro = data.get("outro", {}) or {}
+
+        return cls(
+            enabled=True,
+            # 片头
+            intro_type=str(intro.get("type", "none")),
+            intro_video_path=str(intro.get("video_path", intro.get("video", "")) or ""),
+            intro_duration=float(intro.get("duration", 3.0)),
+            intro_background=str(intro.get("background", "#000000")),
+            intro_title=str(intro.get("title", "") or ""),
+            intro_subtitle=str(intro.get("subtitle", "") or ""),
+            intro_title_color=str(intro.get("title_color", "white")),
+            intro_title_size=int(intro.get("title_size", 48)),
+            intro_subtitle_color=str(intro.get("subtitle_color", "gray")),
+            intro_subtitle_size=int(intro.get("subtitle_size", 24)),
+            # 片尾
+            outro_type=str(outro.get("type", "none")),
+            outro_video_path=str(outro.get("video_path", outro.get("video", "")) or ""),
+            outro_duration=float(outro.get("duration", 3.0)),
+            outro_background=str(outro.get("background", "#000000")),
+            outro_title=str(outro.get("title", "感谢观看") or "感谢观看"),
+            outro_subtitle=str(outro.get("subtitle", "点赞关注不迷路") or "点赞关注不迷路"),
+            outro_title_color=str(outro.get("title_color", "white")),
+            outro_title_size=int(outro.get("title_size", 48)),
+            outro_subtitle_color=str(outro.get("subtitle_color", "gray")),
+            outro_subtitle_size=int(outro.get("subtitle_size", 24)),
+            # 转场
+            transition_effect=str(data.get("transition", "fade")),
+            transition_duration=float(data.get("transition_duration", 0.5)),
+        )
+
+    @property
+    def has_intro(self) -> bool:
+        """是否有片头."""
+        return self.enabled and self.intro_type in ("video", "text")
+
+    @property
+    def has_outro(self) -> bool:
+        """是否有片尾."""
+        return self.enabled and self.outro_type in ("video", "text", "follow")
+
+    def validate(self) -> tuple[bool, str]:
+        """校验配置."""
+        if not self.enabled:
+            return True, ""
+
+        if self.intro_type == "video" and not self.intro_video_path:
+            return False, "视频片头缺少 video_path"
+        if self.intro_type == "text" and not self.intro_title:
+            return False, "文字片头缺少 title"
+
+        if self.outro_type == "video" and not self.outro_video_path:
+            return False, "视频片尾缺少 video_path"
+        if self.outro_type in ("text", "follow") and not self.outro_title:
+            return False, "文字片尾缺少 title"
+
+        if self.intro_duration <= 0:
+            return False, "片头时长必须大于 0"
+        if self.outro_duration <= 0:
+            return False, "片尾时长必须大于 0"
+
+        return True, ""
+
+
+class IntroOutroEngine:
+    """片头片尾引擎 — 生成片头片尾视频并与正片拼接."""
+
+    @staticmethod
+    def generate_text_intro(
+        output_path: Path,
+        config: IntroOutroConfig,
+        output_width: int,
+        output_height: int,
+        output_fps: int,
+    ) -> bool:
+        """生成纯文字片头视频.
+
+        Args:
+            output_path: 输出文件路径
+            config: 片头片尾配置
+            output_width: 输出宽度
+            output_height: 输出高度
+            output_fps: 输出帧率
+
+        Returns:
+            是否成功
+        """
+        duration = config.intro_duration
+        bg = config.intro_background.lstrip("#")
+
+        # 转义文字
+        title = config.intro_title.replace(":", "\\:").replace("'", "\\'")
+        subtitle = config.intro_subtitle.replace(":", "\\:").replace("'", "\\'")
+
+        # 颜色（FFmpeg 颜色格式）
+        title_color = config.intro_title_color
+        subtitle_color = config.intro_subtitle_color
+
+        # 计算位置：标题在中心偏上，副标题在中心偏下
+        title_y = f"(h-text_h)/2 - {config.intro_title_size // 2}"
+        subtitle_y = f"(h-text_h)/2 + {config.intro_title_size}"
+
+        # 构建滤镜
+        filter_parts = []
+
+        # 背景
+        filter_parts.append(
+            f"color=c={config.intro_background}:s={output_width}x{output_height}:d={duration}[bg]"
+        )
+
+        # 标题
+        if title:
+            filter_parts.append(
+                f"[bg]drawtext="
+                f"text='{title}':"
+                f"fontsize={config.intro_title_size}:"
+                f"fontcolor={title_color}:"
+                f"x=(w-text_w)/2:"
+                f"y={title_y}:"
+                f"alpha='if(lt(t,0.5),t/0.5,1)'"  # 淡入
+                f"[with_title]"
+            )
+            bg_label = "with_title"
+        else:
+            bg_label = "bg"
+
+        # 副标题
+        if subtitle:
+            filter_parts.append(
+                f"[{bg_label}]drawtext="
+                f"text='{subtitle}':"
+                f"fontsize={config.intro_subtitle_size}:"
+                f"fontcolor={subtitle_color}:"
+                f"x=(w-text_w)/2:"
+                f"y={subtitle_y}:"
+                f"alpha='if(lt(t,0.8),0,if(lt(t,1.2),(t-0.8)/0.4,1))'"  # 延迟淡入
+                f"[out]"
+            )
+            final_label = "out"
+        else:
+            final_label = bg_label
+            # 如果没有副标题，需要补上 out 标签
+            if final_label != "out":
+                filter_parts.append(f"[{bg_label}]copy[out]")
+
+        filter_complex = ";".join(filter_parts)
+
+        command = [
+            FFMPEG_BIN,
+            "-y",
+            "-f",
+            "lavfi",
+            "-i",
+            f"color=c={config.intro_background}:s={output_width}x{output_height}:d={duration}:r={output_fps}",
+            "-filter_complex",
+            filter_complex,
+            "-map",
+            "[out]",
+            "-c:v",
+            "libx264",
+            "-pix_fmt",
+            "yuv420p",
+            "-r",
+            str(output_fps),
+            "-t",
+            str(duration),
+            "-an",  # 无音频
+            str(output_path),
+        ]
+
+        try:
+            run_ffmpeg(command)
+            return output_path.exists()
+        except subprocess.CalledProcessError as e:
+            logger.error("生成文字片头失败: %s", e)
+            return False
+
+    @staticmethod
+    def generate_text_outro(
+        output_path: Path,
+        config: IntroOutroConfig,
+        output_width: int,
+        output_height: int,
+        output_fps: int,
+    ) -> bool:
+        """生成纯文字片尾视频."""
+        duration = config.outro_duration
+
+        # 转义文字
+        title = config.outro_title.replace(":", "\\:").replace("'", "\\'")
+        subtitle = config.outro_subtitle.replace(":", "\\:").replace("'", "\\'")
+
+        title_color = config.outro_title_color
+        subtitle_color = config.outro_subtitle_color
+
+        # 位置
+        title_y = f"(h-text_h)/2 - {config.outro_title_size // 2}"
+        subtitle_y = f"(h-text_h)/2 + {config.outro_title_size}"
+
+        filter_parts = []
+
+        # 背景
+        bg_src = f"color=c={config.outro_background}:s={output_width}x{output_height}:d={duration}:r={output_fps}"
+        filter_parts.append(f"color=c={config.outro_background}:s={output_width}x{output_height}:d={duration}[bg]")
+
+        # 标题 + 淡出
+        if title:
+            filter_parts.append(
+                f"[bg]drawtext="
+                f"text='{title}':"
+                f"fontsize={config.outro_title_size}:"
+                f"fontcolor={title_color}:"
+                f"x=(w-text_w)/2:"
+                f"y={title_y}:"
+                f"alpha='if(gt(t,{duration - 0.5}),({duration}-t)/0.5,1)'"
+                f"[with_title]"
+            )
+            bg_label = "with_title"
+        else:
+            bg_label = "bg"
+
+        # 副标题
+        if subtitle:
+            filter_parts.append(
+                f"[{bg_label}]drawtext="
+                f"text='{subtitle}':"
+                f"fontsize={config.outro_subtitle_size}:"
+                f"fontcolor={subtitle_color}:"
+                f"x=(w-text_w)/2:"
+                f"y={subtitle_y}:"
+                f"alpha='if(gt(t,{duration - 0.5}),({duration}-t)/0.5,1)'"
+                f"[out]"
+            )
+            final_label = "out"
+        else:
+            final_label = bg_label
+            if final_label != "out":
+                filter_parts.append(f"[{bg_label}]copy[out]")
+
+        filter_complex = ";".join(filter_parts)
+
+        command = [
+            FFMPEG_BIN,
+            "-y",
+            "-f",
+            "lavfi",
+            "-i",
+            bg_src,
+            "-filter_complex",
+            filter_complex,
+            "-map",
+            "[out]",
+            "-c:v",
+            "libx264",
+            "-pix_fmt",
+            "yuv420p",
+            "-r",
+            str(output_fps),
+            "-t",
+            str(duration),
+            "-an",
+            str(output_path),
+        ]
+
+        try:
+            run_ffmpeg(command)
+            return output_path.exists()
+        except subprocess.CalledProcessError as e:
+            logger.error("生成文字片尾失败: %s", e)
+            return False
+
+    @staticmethod
+    def concat_with_intro_outro(
+        main_video: Path,
+        intro_video: Path | None,
+        outro_video: Path | None,
+        output_path: Path,
+        transition_duration: float = 0.5,
+        transition_effect: str = "fade",
+    ) -> bool:
+        """将片头 + 正片 + 片尾用 xfade 拼接.
+
+        只传了片头或片尾也可以，缺失的自动跳过。
+        """
+        # 收集所有片段
+        segments: list[tuple[Path, float]] = []  # (path, duration)
+
+        # 简单探测时长（用 ffprobe，这里简化处理：直接用 xfade 的 offset）
+        # 先添加到列表
+        has_intro = intro_video is not None and intro_video.exists()
+        has_outro = outro_video is not None and outro_video.exists()
+
+        if not has_intro and not has_outro:
+            # 没有片头片尾，直接复制
+            import shutil
+
+            shutil.copy2(main_video, output_path)
+            return True
+
+        # 构建输入和 xfade 链
+        # 简单方式：用 concat demuxer（快速但无转场）
+        # 高级方式：用 xfade 滤镜链（有转场但复杂）
+
+        # 用 concat demuxer 方式（性能好，过渡用硬切）
+        # 后续可以加 xfade 转场
+        concat_list = []
+        if has_intro:
+            concat_list.append(intro_video)
+        concat_list.append(main_video)
+        if has_outro:
+            concat_list.append(outro_video)
+
+        # 生成 concat 列表文件
+        list_file = output_path.parent / f"concat_list_{output_path.stem}.txt"
+        with open(list_file, "w") as f:
+            for seg in concat_list:
+                f.write(f"file '{seg}'\n")
+
+        command = [
+            FFMPEG_BIN,
+            "-y",
+            "-f",
+            "concat",
+            "-safe",
+            "0",
+            "-i",
+            str(list_file),
+            "-c:v",
+            "libx264",
+            "-c:a",
+            "aac",
+            "-pix_fmt",
+            "yuv420p",
+            "-movflags",
+            "+faststart",
+            str(output_path),
+        ]
+
+        try:
+            run_ffmpeg(command)
+            # 清理列表文件
+            list_file.unlink(missing_ok=True)
+            return output_path.exists()
+        except subprocess.CalledProcessError as e:
+            logger.error("片头片尾拼接失败: %s", e)
+            list_file.unlink(missing_ok=True)
+            return False

apps/worker/video_processing/noise_reduction_engine.py

@@ -0,0 +1,229 @@
+"""音频降噪引擎 — 基于 FFmpeg afftdn 滤镜.
+
+支持对音频进行背景噪音消除、人声增强，适用于语音录制、采访等场景。
+
+使用方式：
+    config = NoiseReductionConfig(level="medium")
+    engine = NoiseReductionEngine(config)
+    filter_str = engine.build_filter(input_label, output_label)
+    # 结果: [0:a]afftdn=nf=-25[out]
+
+降级策略：
+    - 参数越界自动钳制
+    - FFmpeg 不支持 afftdn 时，调用方可捕获异常并跳过
+"""
+
+from __future__ import annotations
+
+import logging
+from dataclasses import dataclass
+from enum import Enum
+from typing import Optional
+
+logger = logging.getLogger(__name__)
+
+
+# ── 降噪等级 ──────────────────────────────────────────────────────────────────
+
+
+class NoiseReductionLevel(str, Enum):
+    """降噪等级预设。"""
+
+    LOW = "low"  # 轻度降噪，保留细节，适合轻微背景噪音
+    MEDIUM = "medium"  # 中度降噪，平衡效果和音质
+    HIGH = "high"  # 高度降噪，适合嘈杂环境，可能轻微影响音质
+    CUSTOM = "custom"  # 自定义参数
+
+
+# 各等级对应的降噪参数（afftdn 的 noise floor，单位 dB）
+# 值越大（越接近 0），降噪越强；值越小（越负），降噪越弱
+_LEVEL_PARAMS = {
+    NoiseReductionLevel.LOW: {
+        "nf": -35,  # 噪音阈值（dB），越负越保守
+        "tn": -10,  # 噪音频谱平滑度
+        "tr": 50,  # 时间分辨率（ms）
+    },
+    NoiseReductionLevel.MEDIUM: {
+        "nf": -25,
+        "tn": -10,
+        "tr": 50,
+    },
+    NoiseReductionLevel.HIGH: {
+        "nf": -15,
+        "tn": -5,
+        "tr": 30,
+    },
+}
+
+
+# ── 配置模型 ──────────────────────────────────────────────────────────────────
+
+
+@dataclass
+class NoiseReductionConfig:
+    """音频降噪配置。
+
+    Attributes:
+        enabled: 是否启用降噪
+        level: 降噪等级 low/medium/high/custom
+        noise_floor: 自定义噪音阈值（dB），仅 level=custom 时有效，范围 -60 ~ -5
+        voice_enhance: 是否启用人声增强
+        output_format: 输出格式描述（内部使用）
+    """
+
+    enabled: bool = False
+    level: NoiseReductionLevel = NoiseReductionLevel.MEDIUM
+    noise_floor: float = -25.0  # dB
+    voice_enhance: bool = False
+
+    @classmethod
+    def from_dict(cls, data: dict | None) -> "NoiseReductionConfig":
+        """从字典解析配置，参数越界自动钳制。"""
+        if not data or not data.get("enabled", False):
+            return cls(enabled=False)
+
+        level_str = str(data.get("level", "medium")).lower()
+        try:
+            level = NoiseReductionLevel(level_str)
+        except ValueError:
+            level = NoiseReductionLevel.MEDIUM
+
+        try:
+            noise_floor = float(data.get("noise_floor", -25.0))
+        except (TypeError, ValueError):
+            noise_floor = -25.0
+
+        voice_enhance = bool(data.get("voice_enhance", False))
+
+        # 钳制到合法范围
+        noise_floor = max(-60.0, min(-5.0, noise_floor))
+
+        return cls(
+            enabled=True,
+            level=level,
+            noise_floor=noise_floor,
+            voice_enhance=voice_enhance,
+        )
+
+    def has_effect(self) -> bool:
+        """判断是否有实际降噪效果。"""
+        return self.enabled
+
+    def get_effective_noise_floor(self) -> float:
+        """获取实际生效的噪音阈值（dB）。"""
+        if self.level == NoiseReductionLevel.CUSTOM:
+            return self.noise_floor
+        params = _LEVEL_PARAMS.get(self.level, _LEVEL_PARAMS[NoiseReductionLevel.MEDIUM])
+        return float(params["nf"])
+
+
+# ── 引擎实现 ──────────────────────────────────────────────────────────────────
+
+
+class NoiseReductionEngine:
+    """音频降噪引擎。
+
+    基于 FFmpeg afftdn（Audio FFt Denoiser）滤镜实现：
+    - 使用短时傅里叶变换分析音频频谱
+    - 识别并消除稳态背景噪音
+    - 保留人声等非稳态信号
+    """
+
+    def __init__(self, config: NoiseReductionConfig):
+        self.config = config
+
+    def build_filter(self, input_label: str, output_label: str) -> str:
+        """构建音频降噪滤镜字符串。
+
+        Args:
+            input_label: 输入标签，如 "[0:a]" 或 "[a0]"
+            output_label: 输出标签，如 "[nr0]"
+
+        Returns:
+            FFmpeg 滤镜字符串，如 "[a0]afftdn=nf=-25:tn=-10:tr=50[nr0]"
+
+        Raises:
+            ValueError: 配置无效时抛出（调用方应捕获并降级）
+        """
+        if not self.config.has_effect():
+            return f"{input_label}anull{output_label}"
+
+        # 获取参数
+        if self.config.level == NoiseReductionLevel.CUSTOM:
+            nf = self.config.noise_floor
+            tn = -10  # 默认频谱平滑度
+            tr = 50  # 默认时间分辨率
+        else:
+            params = _LEVEL_PARAMS.get(
+                self.config.level,
+                _LEVEL_PARAMS[NoiseReductionLevel.MEDIUM],
+            )
+            nf = float(params["nf"])
+            tn = float(params["tn"])
+            tr = float(params["tr"])
+
+        # 构建 afftdn 滤镜
+        # nf: noise floor (dB)
+        # tn: temporal noise floor smoothing (dB)
+        # tr: time resolution (ms)
+        filter_parts = [f"afftdn=nf={nf}:tn={tn}:tr={tr}"]
+
+        # 人声增强：通过 highpass + 轻微压缩实现
+        if self.config.voice_enhance:
+            # 1. 高通滤波，去除低频噪音
+            filter_parts.append("highpass=f=80")
+            # 2. 轻微压缩，提升人声清晰度
+            filter_parts.append("acompressor=threshold=-20:ratio=2:attack=5:release=50")
+            # 3. 响度归一化
+            filter_parts.append("loudnorm=I=-16:TP=-1.5:LRA=11")
+
+        filter_str = f"{input_label}{','.join(filter_parts)}{output_label}"
+        return filter_str
+
+    def build_filter_arnndn(self, input_label: str, output_label: str, model_file: str) -> str:
+        """使用 RNN 降噪滤镜（arnndn，效果更好但需要模型文件）。
+
+        注意：需要额外下载 RNNNoise 模型文件，默认使用 afftdn（无需额外依赖）。
+
+        Args:
+            input_label: 输入标签
+            output_label: 输出标签
+            model_file: RNNNoise 模型文件路径（.rnnn 格式）
+
+        Returns:
+            FFmpeg 滤镜字符串
+        """
+        if not self.config.has_effect():
+            return f"{input_label}anull{output_label}"
+
+        return f"{input_label}arnndn=m={model_file}{output_label}"
+
+
+def apply_noise_reduction_if_needed(
+    config_data: dict | None,
+    input_label: str,
+    output_label: str,
+) -> Optional[str]:
+    """便捷函数：根据配置判断是否需要应用音频降噪。
+
+    Args:
+        config_data: 降噪配置字典（从 plan.config.audio_noise_reduction 或 clip.config.noise_reduction 读取）
+        input_label: 输入标签
+        output_label: 输出标签
+
+    Returns:
+        滤镜字符串，不需要降噪时返回 None
+    """
+    if not config_data:
+        return None
+
+    try:
+        config = NoiseReductionConfig.from_dict(config_data)
+        if not config.has_effect():
+            return None
+
+        engine = NoiseReductionEngine(config)
+        return engine.build_filter(input_label, output_label)
+    except Exception as e:
+        logger.warning("[noise-reduction] 应用降噪失败，跳过: %s", e)
+        return None

apps/worker/video_processing/pip_engine.py

@@ -0,0 +1,483 @@
+"""画中画（PiP）引擎 — 基于 FFmpeg overlay 滤镜实现多图层叠加.
+
+支持能力：
+- 多图层叠加：主画面 + 多个副画面
+- 位置：9宫格 + 自由坐标（像素或百分比）
+- 大小：宽高缩放（像素或百分比）
+- 圆角裁剪：支持圆角矩形裁剪
+- 透明度：0-100%
+- 入场出场动画：淡入淡出、滑入滑出
+- 时间同步：每个副画面独立开始时间和持续时长
+- 降级策略：素材不存在时跳过，不阻断渲染
+"""
+
+from __future__ import annotations
+
+import logging
+from dataclasses import dataclass, field
+from pathlib import Path
+from typing import Any
+
+logger = logging.getLogger(__name__)
+
+
+# ── 位置常量 ──────────────────────────────────────────────────────────────────
+
+# 9宫格位置枚举
+POSITION_TOP_LEFT = "top_left"
+POSITION_TOP_CENTER = "top_center"
+POSITION_TOP_RIGHT = "top_right"
+POSITION_CENTER_LEFT = "center_left"
+POSITION_CENTER = "center"
+POSITION_CENTER_RIGHT = "center_right"
+POSITION_BOTTOM_LEFT = "bottom_left"
+POSITION_BOTTOM_CENTER = "bottom_center"
+POSITION_BOTTOM_RIGHT = "bottom_right"
+
+_VALID_POSITIONS = {
+    POSITION_TOP_LEFT,
+    POSITION_TOP_CENTER,
+    POSITION_TOP_RIGHT,
+    POSITION_CENTER_LEFT,
+    POSITION_CENTER,
+    POSITION_CENTER_RIGHT,
+    POSITION_BOTTOM_LEFT,
+    POSITION_BOTTOM_CENTER,
+    POSITION_BOTTOM_RIGHT,
+}
+
+# 动画类型
+ANIMATION_FADE = "fade"  # 淡入淡出
+ANIMATION_SLIDE_LEFT = "slide_left"  # 从左滑入
+ANIMATION_SLIDE_RIGHT = "slide_right"  # 从右滑入
+ANIMATION_SLIDE_TOP = "slide_top"  # 从上滑入
+ANIMATION_SLIDE_BOTTOM = "slide_bottom"  # 从下滑入
+
+_VALID_ANIMATIONS = {
+    ANIMATION_FADE,
+    ANIMATION_SLIDE_LEFT,
+    ANIMATION_SLIDE_RIGHT,
+    ANIMATION_SLIDE_TOP,
+    ANIMATION_SLIDE_BOTTOM,
+}
+
+
+# ── 数据模型 ──────────────────────────────────────────────────────────────────
+
+
+@dataclass
+class PiPLayerConfig:
+    """单个画中画图层配置."""
+
+    # 素材来源
+    source: str = ""  # 素材ID或视频URL
+    source_type: str = "asset_id"  # "asset_id" | "url" | "local_path"
+
+    # 位置配置
+    position: str = POSITION_BOTTOM_RIGHT  # 9宫格位置或 "custom"
+    x: int | str = 0  # 自定义x坐标（像素或百分比如 "30%"）
+    y: int | str = 0  # 自定义y坐标
+    margin: int = 20  # 9宫格模式下的边距（像素）
+
+    # 大小配置
+    width: int | str = "25%"  # 宽度（像素或百分比）
+    height: int | str = ""  # 高度（空则按比例自适应）
+
+    # 样式
+    opacity: float = 1.0  # 透明度 0.0-1.0
+    corner_radius: int = 0  # 圆角半径（像素），0表示无圆角
+    border_width: int = 0  # 边框宽度
+    border_color: str = "white"  # 边框颜色
+
+    # 时间控制
+    start_time: float = 0.0  # 开始显示时间（秒）
+    duration: float = 0.0  # 持续时长（秒），0表示全程显示
+
+    # 动画
+    animation_in: str = ""  # 入场动画类型
+    animation_out: str = ""  # 出场动画类型
+    animation_duration: float = 0.5  # 动画时长（秒）
+
+    # 层级
+    z_index: int = 1  # 图层顺序，数字越大越在上层
+
+    def validate(self) -> tuple[bool, str]:
+        """校验配置合法性，返回 (是否合法, 错误信息)."""
+        if not self.source:
+            return False, "source不能为空"
+
+        if self.position != "custom" and self.position not in _VALID_POSITIONS:
+            return False, f"无效的position: {self.position}"
+
+        if self.opacity < 0 or self.opacity > 1:
+            return False, "opacity必须在0-1之间"
+
+        if self.corner_radius < 0:
+            return False, "corner_radius不能为负数"
+
+        if self.start_time < 0:
+            return False, "start_time不能为负数"
+
+        if self.duration < 0:
+            return False, "duration不能为负数"
+
+        if self.animation_in and self.animation_in not in _VALID_ANIMATIONS:
+            return False, f"无效的入场动画: {self.animation_in}"
+
+        if self.animation_out and self.animation_out not in _VALID_ANIMATIONS:
+            return False, f"无效的出场动画: {self.animation_out}"
+
+        if self.animation_duration < 0:
+            return False, "animation_duration不能为负数"
+
+        return True, ""
+
+
+@dataclass
+class PiPConfig:
+    """画中画整体配置."""
+
+    enabled: bool = False
+    layers: list[PiPLayerConfig] = field(default_factory=list)
+
+    @classmethod
+    def from_dict(cls, data: dict[str, Any] | None) -> "PiPConfig":
+        """从字典解析配置."""
+        if not data or not data.get("enabled", False):
+            return cls(enabled=False)
+
+        layers_data = data.get("layers", [])
+        layers = []
+        for layer_data in layers_data:
+            try:
+                layer = PiPLayerConfig(
+                    source=layer_data.get("source", ""),
+                    source_type=layer_data.get("source_type", "asset_id"),
+                    position=layer_data.get("position", POSITION_BOTTOM_RIGHT),
+                    x=layer_data.get("x", 0),
+                    y=layer_data.get("y", 0),
+                    margin=int(layer_data.get("margin", 20)),
+                    width=layer_data.get("width", "25%"),
+                    height=layer_data.get("height", ""),
+                    opacity=float(layer_data.get("opacity", 1.0)),
+                    corner_radius=int(layer_data.get("corner_radius", 0)),
+                    border_width=int(layer_data.get("border_width", 0)),
+                    border_color=layer_data.get("border_color", "white"),
+                    start_time=float(layer_data.get("start_time", 0.0)),
+                    duration=float(layer_data.get("duration", 0.0)),
+                    animation_in=layer_data.get("animation_in", ""),
+                    animation_out=layer_data.get("animation_out", ""),
+                    animation_duration=float(layer_data.get("animation_duration", 0.5)),
+                    z_index=int(layer_data.get("z_index", 1)),
+                )
+                valid, err = layer.validate()
+                if valid:
+                    layers.append(layer)
+                else:
+                    logger.warning("PiP图层配置无效，跳过: %s", err)
+            except (ValueError, TypeError) as e:
+                logger.warning("PiP图层解析失败，跳过: %s", e)
+
+        # 按 z_index 排序
+        layers.sort(key=lambda layer: layer.z_index)
+
+        return cls(enabled=bool(layers), layers=layers)
+
+
+# ── PiP 引擎 ──────────────────────────────────────────────────────────────────
+
+
+class PiPEngine:
+    """画中画引擎 — 生成 FFmpeg 滤镜链实现多图层叠加."""
+
+    def __init__(
+        self,
+        output_width: int,
+        output_height: int,
+        output_fps: int = 30,
+    ):
+        self.output_width = output_width
+        self.output_height = output_height
+        self.output_fps = output_fps
+
+    def _parse_size(self, value: int | str, base: int) -> int:
+        """解析尺寸值（像素或百分比）."""
+        if isinstance(value, int):
+            return max(1, value)
+        if isinstance(value, str) and value.endswith("%"):
+            pct = float(value.rstrip("%")) / 100.0
+            return max(1, int(base * pct))
+        try:
+            return max(1, int(value))
+        except (ValueError, TypeError):
+            return int(base * 0.25)  # 默认25%
+
+    def _parse_position(
+        self,
+        layer: PiPLayerConfig,
+        pip_width: int,
+        pip_height: int,
+    ) -> tuple[int, int]:
+        """计算画中画的实际位置 (x, y)."""
+        W = self.output_width
+        H = self.output_height
+        m = layer.margin
+
+        if layer.position == "custom":
+            x = self._parse_size(layer.x, W)
+            y = self._parse_size(layer.y, H)
+            return (x, y)
+
+        pos_map = {
+            POSITION_TOP_LEFT: (m, m),
+            POSITION_TOP_CENTER: ((W - pip_width) // 2, m),
+            POSITION_TOP_RIGHT: (W - pip_width - m, m),
+            POSITION_CENTER_LEFT: (m, (H - pip_height) // 2),
+            POSITION_CENTER: ((W - pip_width) // 2, (H - pip_height) // 2),
+            POSITION_CENTER_RIGHT: (W - pip_width - m, (H - pip_height) // 2),
+            POSITION_BOTTOM_LEFT: (m, H - pip_height - m),
+            POSITION_BOTTOM_CENTER: ((W - pip_width) // 2, H - pip_height - m),
+            POSITION_BOTTOM_RIGHT: (W - pip_width - m, H - pip_height - m),
+        }
+        return pos_map.get(layer.position, pos_map[POSITION_BOTTOM_RIGHT])
+
+    def _build_pip_pre_filter(
+        self,
+        input_label: str,
+        layer: PiPLayerConfig,
+        pip_width: int,
+        pip_height: int,
+        output_label: str,
+    ) -> str:
+        """构建单个PiP图层的预处理滤镜链.
+
+        处理顺序：scale → 圆角裁剪（可选）→ 边框（可选）→ 透明度 → 动画（可选）
+        """
+        filters: list[str] = []
+
+        # Step 1: scale
+        filters.append(f"scale={pip_width}:{pip_height}")
+        filters.append("setsar=1")
+
+        # Step 2: 圆角裁剪
+        if layer.corner_radius > 0:
+            r = min(layer.corner_radius, pip_width // 2, pip_height // 2)
+            # 使用 geq + 圆形遮罩实现圆角
+            # 更简单的方式：用 rounded 滤镜（FFmpeg 5.0+）或 format + alpha
+            # 这里用更通用的方式：创建圆角遮罩 + overlay 到透明背景
+            filters.append(
+                f"format=yuva420p,"
+                f"geq="
+                f"lum='lum(X,Y)':"
+                f"cb='cb(X,Y)':"
+                f"cr='cr(X,Y)':"
+                f"a='if(lt(X,{r})*lt(Y,{r}),"
+                f"gt(hypot({r}-X,{r}-Y),{r})*0+1,"
+                f"if(gt(X,W-{r})*lt(Y,{r}),"
+                f"gt(hypot(X-(W-{r}),{r}-Y),{r})*0+1,"
+                f"if(lt(X,{r})*gt(Y,H-{r}),"
+                f"gt(hypot({r}-X,Y-(H-{r})),{r})*0+1,"
+                f"if(gt(X,W-{r})*gt(Y,H-{r}),"
+                f"gt(hypot(X-(W-{r}),Y-(H-{r})),{r})*0+1,1))))'"
+            )
+
+        # Step 3: 边框
+        if layer.border_width > 0:
+            bw = layer.border_width
+            color = layer.border_color
+            filters.append(f"pad={pip_width + 2*bw}:{pip_height + 2*bw}:{bw}:{bw}:{color}")
+
+        # Step 4: 透明度
+        if layer.opacity < 1.0:
+            alpha = layer.opacity
+            filters.append(f"format=yuva420p,colorchannelmixer=aa={alpha}")
+
+        # Step 5: 入场出场动画
+        if layer.animation_in or layer.animation_out:
+            filters.extend(self._build_animation_filters(layer, pip_width, pip_height))
+
+        filter_str = f"[{input_label}]{','.join(filters)}[{output_label}]"
+        return filter_str
+
+    def _build_animation_filters(
+        self,
+        layer: PiPLayerConfig,
+        pip_width: int,
+        pip_height: int,
+    ) -> list[str]:
+        """构建入场出场动画滤镜."""
+        filters: list[str] = []
+        anim_dur = layer.animation_duration
+
+        if layer.animation_in == ANIMATION_FADE:
+            # 淡入
+            filters.append(f"fade=t=in:st=0:d={anim_dur}:alpha=1")
+        elif layer.animation_in == ANIMATION_SLIDE_LEFT:
+            # 从左滑入 — 用 overlay 动态x实现，这里先标记位置表达式
+            pass  # slide 动画在 overlay 表达式中处理
+        elif layer.animation_in == ANIMATION_SLIDE_RIGHT:
+            pass
+        elif layer.animation_in == ANIMATION_SLIDE_TOP:
+            pass
+        elif layer.animation_in == ANIMATION_SLIDE_BOTTOM:
+            pass
+
+        if layer.animation_out == ANIMATION_FADE:
+            # 淡出需要知道总时长，这里用表达式
+            if layer.duration > 0:
+                start_fade = layer.duration - anim_dur
+                filters.append(f"fade=t=out:st={max(0, start_fade)}:d={anim_dur}:alpha=1")
+
+        return filters
+
+    def _build_overlay_expr(
+        self,
+        layer: PiPLayerConfig,
+        base_x: int,
+        base_y: int,
+        pip_width: int,
+        pip_height: int,
+    ) -> tuple[str, str]:
+        """构建 overlay 滤镜的 x/y 表达式（支持滑动动画）.
+
+        Returns:
+            (x_expr, y_expr) — FFmpeg表达式字符串
+        """
+        W = self.output_width
+        H = self.output_height
+        anim_dur = layer.animation_duration
+
+        x_expr = str(base_x)
+        y_expr = str(base_y)
+
+        # 入场滑入动画
+        if layer.animation_in == ANIMATION_SLIDE_LEFT:
+            # 从左侧滑入：x 从 -pip_width 变化到 base_x
+            x_expr = f"'{base_x}+(X)*0+if(lt(t,{anim_dur}),{-pip_width}+t/{anim_dur}*({base_x}+{pip_width}),{base_x})'"
+        elif layer.animation_in == ANIMATION_SLIDE_RIGHT:
+            # 从右侧滑入：x 从 W 变化到 base_x
+            x_expr = f"'{base_x}+if(lt(t,{anim_dur}),{W}-t/{anim_dur}*({W}-{base_x}),{base_x})'"
+        elif layer.animation_in == ANIMATION_SLIDE_TOP:
+            y_expr = f"'{base_y}+if(lt(t,{anim_dur}),{-pip_height}+t/{anim_dur}*({base_y}+{pip_height}),{base_y})'"
+        elif layer.animation_in == ANIMATION_SLIDE_BOTTOM:
+            y_expr = f"'{base_y}+if(lt(t,{anim_dur}),{H}-t/{anim_dur}*({H}-{base_y}),{base_y})'"
+
+        # 出场滑出动画（需要总时长）
+        if layer.duration > 0 and anim_dur > 0:
+            out_start = layer.duration - anim_dur
+            if layer.animation_out == ANIMATION_SLIDE_LEFT:
+                x_expr = f"'{base_x}+if(gt(t,{out_start}),{base_x}-(t-{out_start})/{anim_dur}*({base_x}+{pip_width}),{base_x})'"
+            elif layer.animation_out == ANIMATION_SLIDE_RIGHT:
+                x_expr = f"'{base_x}+if(gt(t,{out_start}),{base_x}+(t-{out_start})/{anim_dur}*({W}-{base_x}+{pip_width}),{base_x})'"
+            elif layer.animation_out == ANIMATION_SLIDE_TOP:
+                y_expr = f"'{base_y}+if(gt(t,{out_start}),{base_y}-(t-{out_start})/{anim_dur}*({base_y}+{pip_height}),{base_y})'"
+            elif layer.animation_out == ANIMATION_SLIDE_BOTTOM:
+                y_expr = f"'{base_y}+if(gt(t,{out_start}),{base_y}+(t-{out_start})/{anim_dur}*({H}-{base_y}+{pip_height}),{base_y})'"
+
+        return (x_expr, y_expr)
+
+    def build_pip_filters(
+        self,
+        base_label: str,
+        pip_sources: list[tuple[str, PiPLayerConfig, Path]],
+        *,
+        base_input_idx: int = 0,
+    ) -> tuple[str, list[str], str]:
+        """构建完整的画中画滤镜链和输入参数.
+
+        Args:
+            base_label: 底层视频的滤镜标签（如 "final_video" 或 "v0"，不带方括号）
+            pip_sources: [(input_label, layer_config, source_path), ...]
+            base_input_idx: PiP 素材在整个 FFmpeg 输入中的起始索引
+
+        Returns:
+            (filter_parts, input_args, final_label)
+            - filter_parts: 滤镜字符串列表（用 ; 连接后成为 filter_complex）
+            - input_args: 额外的输入参数列表 ["-i", path, "-i", path, ...]
+            - final_label: 最终合成后的输出标签（不带方括号）
+        """
+        if not pip_sources:
+            return [], [], base_label
+
+        filter_parts: list[str] = []
+        input_args: list[str] = []
+        current_label = base_label
+
+        for i, (input_label, layer, path) in enumerate(pip_sources):
+            # 添加输入
+            input_args.extend(["-i", str(path)])
+
+            # 计算实际大小
+            pip_w = self._parse_size(layer.width, self.output_width)
+            if layer.height:
+                pip_h = self._parse_size(layer.height, self.output_height)
+            else:
+                # 按宽度等比例（假设16:9，实际会scale时保持比例）
+                pip_h = int(pip_w * 9 / 16)
+
+            # 实际输入索引 = 起始索引 + 当前偏移
+            actual_input_idx = base_input_idx + i
+
+            # 预处理标签
+            pre_label = f"pip_pre_{i}"
+
+            # 构建预处理滤镜
+            pre_filter = self._build_pip_pre_filter(
+                input_label=f"{actual_input_idx}:v",
+                layer=layer,
+                pip_width=pip_w,
+                pip_height=pip_h,
+                output_label=pre_label,
+            )
+            filter_parts.append(pre_filter)
+
+            # 计算位置
+            base_x, base_y = self._parse_position(layer, pip_w, pip_h)
+
+            # 构建overlay表达式（支持滑动动画）
+            x_expr, y_expr = self._build_overlay_expr(layer, base_x, base_y, pip_w, pip_h)
+
+            # 时间控制（enable表达式）
+            enable_expr = ""
+            if layer.start_time > 0 or layer.duration > 0:
+                start = layer.start_time
+                if layer.duration > 0:
+                    end = start + layer.duration
+                    enable_expr = f":enable='between(t,{start},{end})'"
+                else:
+                    enable_expr = f":enable='gte(t,{start})'"
+
+            # 合成标签
+            combined_label = f"pip_combined_{i}"
+
+            # overlay 滤镜
+            overlay_filter = (
+                f"[{current_label}][{pre_label}]" f"overlay={x_expr}:{y_expr}{enable_expr}" f"[{combined_label}]"
+            )
+            filter_parts.append(overlay_filter)
+
+            current_label = combined_label
+
+        return filter_parts, input_args, current_label
+
+    def validate_layer_source(
+        self,
+        layer: PiPLayerConfig,
+        asset_path_map: dict[str, Path],
+    ) -> Path | None:
+        """验证图层素材是否可用，返回本地路径或None（降级跳过）."""
+        try:
+            if layer.source_type == "local_path":
+                path = Path(layer.source)
+                if path.exists():
+                    return path
+            elif layer.source_type == "asset_id":
+                if layer.source in asset_path_map:
+                    return asset_path_map[layer.source]
+            elif layer.source_type == "url":
+                # URL类型由调用者负责下载，这里返回标记
+                return None  # 暂时不支持直接URL
+        except Exception as e:
+            logger.warning("PiP素材验证失败: %s", e)
+
+        return None

apps/worker/video_processing/render_audio.py

@@ -16,15 +16,17 @@ import subprocess
 from dataclasses import dataclass, field
 from pathlib import Path
 
-from video_processing.ffmpeg_utils import FFMPEG_BIN, probe_has_audio, run_ffmpeg
-
 # 延迟导入避免循环依赖：unified_render_service 定义 ResolvedClip / RenderLayer，
 # 本模块提供音频函数供 unified_render_service 调用。
 # 使用 from __future__ import annotations + TYPE_CHECKING 解决类型引用。
 from typing import TYPE_CHECKING
 
+from video_processing.ffmpeg_utils import FFMPEG_BIN, probe_has_audio, run_ffmpeg
+from video_processing.reverse_engine import ReverseConfig, ReverseEngine
+from video_processing.speed_engine import SpeedEngine
+
 if TYPE_CHECKING:
-    from video_processing.unified_render_service import ResolvedClip, RenderLayer
+    from video_processing.unified_render_service import RenderLayer, ResolvedClip
 
 logger = logging.getLogger(__name__)
 
@@ -35,6 +37,8 @@ class RenderContext:
 
     work_dir: Path
     plan_id: str
+    # 音频降噪配置（全局，对最终混音结果应用）
+    noise_reduction_config: dict | None = None
     # 音频探测缓存（避免同一 clip 被多次 ffprobe）
     _audio_cache: dict[str, bool] = field(default_factory=dict)
 
@@ -70,6 +74,9 @@ def mix_audio(
     ctx: RenderContext,
     layers: list[RenderLayer],
     video_duration: float,
+    *,
+    bgm_path: str | None = None,
+    bgm_config: dict | None = None,
 ) -> Path | None:
     """音频后处理混音.
 
@@ -79,11 +86,14 @@ def mix_audio(
     3. 独立音频轨（audio role）用 amix 混入
     4. 输出时长截断到 video_duration
     5. 无音频流的 clip 会被自动跳过，避免 FFmpeg 引用 [i:a] 失败
+    6. 如果提供了 bgm_path，则额外混入 BGM（支持淡入淡出、循环、人声闪避）
 
     Args:
         ctx: 渲染上下文
         layers: 图层列表
         video_duration: 视频总时长（用于截断音频）
+        bgm_path: BGM 音频本地路径，为 None 时不混入 BGM
+        bgm_config: BGM 配置字典（volume/fade_in/fade_out/sidechain 等）
 
     Returns:
         混音后的音频文件路径，无音频时返回 None
@@ -116,6 +126,15 @@ def mix_audio(
     audio_clips = [c for c in audio_clips if clip_has_audio(ctx, c)]
 
     if not main_clips and not audio_clips:
+        # 没有主音频也没有独立音频 → 检查是否有 BGM
+        if bgm_path and bgm_config and bgm_config.get("enabled", False):
+            from video_processing.bgm_mixer import BGMConfig, build_bgm_only
+
+            bgm_cfg = BGMConfig.from_config_dict(bgm_path, bgm_config)
+            try:
+                return build_bgm_only(ctx, bgm_cfg, video_duration)
+            except Exception:
+                logger.exception("[bgm] 纯BGM生成失败: plan_id=%s", ctx.plan_id)
         return None
 
     # 构建音频处理命令
@@ -124,11 +143,72 @@ def mix_audio(
     # 简单场景：只有主图层 + 无独立音频 → 直接从视频提取音频并拼接
     if main_clips and not audio_clips:
         concat_main_audio(ctx, main_clips, output_path, video_duration)
-        return output_path
+    else:
+        # 有独立音频轨 → amix 混音
+        mix_with_independent_audio(ctx, main_clips, audio_clips, output_path, video_duration)
 
-    # 有独立音频轨 → amix 混音
-    mix_with_independent_audio(ctx, main_clips, audio_clips, output_path, video_duration)
-    return output_path
+    # ── BGM 混音 ──
+    if bgm_path and bgm_config and bgm_config.get("enabled", False):
+        from video_processing.bgm_mixer import BGMConfig, mix_bgm_with_main
+
+        bgm_cfg = BGMConfig.from_config_dict(bgm_path, bgm_config)
+        bgm_output = ctx.work_dir / f"audio_with_bgm_{ctx.plan_id}.aac"
+
+        try:
+            # 这里 main_audio 就是 output_path，先有主音频再混 BGM
+            final_path = mix_bgm_with_main(ctx, output_path, bgm_cfg, video_duration)
+            return _apply_noise_reduction_if_needed(ctx, final_path)
+        except Exception:
+            logger.exception("[bgm] BGM 混音失败，回退到无 BGM 音频: plan_id=%s", ctx.plan_id)
+            return _apply_noise_reduction_if_needed(ctx, output_path)
+
+    return _apply_noise_reduction_if_needed(ctx, output_path)
+
+
+def _apply_noise_reduction_if_needed(ctx: RenderContext, audio_path: Path) -> Path:
+    """如果配置了音频降噪，对已生成的音频文件应用降噪。
+
+    作为后处理步骤，对最终混音结果统一降噪。
+    失败时返回原始文件路径，不阻断主流程。
+    """
+    if not ctx.noise_reduction_config:
+        return audio_path
+
+    try:
+        from video_processing.noise_reduction_engine import NoiseReductionConfig, NoiseReductionEngine
+
+        config = NoiseReductionConfig.from_dict(ctx.noise_reduction_config)
+        if not config.has_effect():
+            return audio_path
+
+        engine = NoiseReductionEngine(config)
+        filter_str = engine.build_filter("[0:a]", "[out]")
+        # 提取滤镜部分（不带标签）
+        filter_part = filter_str[len("[0:a]") : -len("[out]")]
+
+        nr_output_path = audio_path.with_name(f"{audio_path.stem}_nr.aac")
+        command = [
+            FFMPEG_BIN,
+            "-y",
+            "-i",
+            str(audio_path),
+            "-af",
+            filter_part,
+            "-acodec",
+            "aac",
+            "-b:a",
+            "128k",
+            str(nr_output_path),
+        ]
+        run_ffmpeg(command)
+
+        if nr_output_path.exists():
+            return nr_output_path
+        logger.warning("[noise-reduction] 降噪输出文件不存在，使用原始音频")
+        return audio_path
+    except Exception as e:
+        logger.warning("[noise-reduction] 音频降噪失败，使用原始音频: %s", e)
+        return audio_path
 
 
 def concat_main_audio(
@@ -145,40 +225,130 @@ def concat_main_audio(
         # 单 clip，直接提取音频，截断到 min(clip有效时长, 视频总时长)
         clip = clips[0]
         effective_duration = clip_effective_duration(clip)
-        # 最终时长：取 clip 有效时长和视频总时长的较小值
-        # （视频总时长由主图层决定，但单 clip 场景下两者应该一致，仍做保护）
-        final_duration = effective_duration
+        trim_start = getattr(clip, "start_time", 0) or 0
+        speed = getattr(clip, "playback_speed", 1.0) or 1.0
+        if not isinstance(speed, (int, float)) or speed <= 0:
+            speed = 1.0
+
+        # 调速后时长
+        adjusted_duration = effective_duration / speed if abs(speed - 1.0) >= 1e-6 else effective_duration
+        # 最终时长：取调速后时长和视频总时长的较小值
+        final_duration = adjusted_duration
         if video_duration > 0 and (final_duration <= 0 or final_duration > video_duration):
             final_duration = video_duration
 
-        command = [
-            FFMPEG_BIN,
-            "-y",
-            "-i",
-            str(clip.local_path),
-            "-vn",
-            "-acodec",
-            "aac",
-            "-b:a",
-            "128k",
-        ]
-        if final_duration > 0:
-            command.extend(["-t", f"{final_duration:.3f}"])
-        command.append(str(output_path))
-        run_ffmpeg(command)
+        # 音频倒放
+        reverse_config = ReverseConfig.from_dict(clip.config.get("reverse"))
+        has_reverse = reverse_config.enabled and reverse_config.reverse_audio
+        has_speed = abs(speed - 1.0) >= 1e-6
+
+        if not has_speed and not has_reverse:
+            # 无调速无倒放：简单命令行，-ss 裁剪更高效
+            command = [
+                FFMPEG_BIN,
+                "-y",
+                "-i",
+                str(clip.local_path),
+                "-vn",
+                "-acodec",
+                "aac",
+                "-b:a",
+                "128k",
+            ]
+            if trim_start > 0:
+                command.extend(["-ss", f"{trim_start:.3f}"])
+            if final_duration > 0:
+                command.extend(["-t", f"{final_duration:.3f}"])
+            command.append(str(output_path))
+            run_ffmpeg(command)
+        else:
+            # 有调速或倒放：用 filter_complex
+            speed_engine = SpeedEngine()
+            audio_filters = []
+            if effective_duration > 0:
+                audio_filters.append(f"atrim=start={trim_start:.3f}:duration={effective_duration:.3f}")
+                audio_filters.append("asetpts=PTS-STARTPTS")
+
+            # 音频调速
+            if has_speed:
+                from video_processing.speed_engine import SpeedConfig
+
+                config = SpeedConfig(speed=float(speed))
+                config.clamp()
+                atempo_filter = speed_engine.build_audio_filter(config)
+                if atempo_filter:
+                    audio_filters.append(atempo_filter)
+
+            # 音频倒放
+            if has_reverse:
+                reverse_filter = ReverseEngine.build_audio_filter(reverse_config, duration=effective_duration)
+                if reverse_filter:
+                    audio_filters.append(reverse_filter)
+
+            filter_parts: list[str] = [f"[0:a]{','.join(audio_filters)}[outa]"]
+            if video_duration > 0 and final_duration < adjusted_duration:
+                filter_parts.append(f"[outa]atrim=0:{final_duration:.3f}[final_audio]")
+                final_label = "final_audio"
+            else:
+                final_label = "outa"
+
+            filter_complex = ";".join(filter_parts)
+            command = [
+                FFMPEG_BIN,
+                "-y",
+                "-i",
+                str(clip.local_path),
+                "-filter_complex",
+                filter_complex,
+                "-map",
+                f"[{final_label}]",
+                "-acodec",
+                "aac",
+                "-b:a",
+                "128k",
+                str(output_path),
+            ]
+            run_ffmpeg(command)
         return
 
     # 多 clip，用 filter_complex concat
     input_args: list[str] = []
     filter_parts: list[str] = []
+    speed_engine = SpeedEngine()
 
     for i, clip in enumerate(clips):
         input_args.extend(["-i", str(clip.local_path)])
         effective_duration = clip_effective_duration(clip)
+        trim_start = getattr(clip, "start_time", 0) or 0
+        speed = getattr(clip, "playback_speed", 1.0) or 1.0
+        if not isinstance(speed, (int, float)) or speed <= 0:
+            speed = 1.0
+
+        audio_filters: list[str] = []
         if effective_duration > 0:
-            filter_parts.append(f"[{i}:a]atrim=0:{effective_duration:.3f},asetpts=PTS-STARTPTS[a{i}]")
+            audio_filters.append(f"atrim=start={trim_start:.3f}:duration={effective_duration:.3f}")
+            audio_filters.append("asetpts=PTS-STARTPTS")
+
+        # 音频调速 — atempo 多级串联
+        if abs(speed - 1.0) >= 1e-6:
+            from video_processing.speed_engine import SpeedConfig
+
+            config = SpeedConfig(speed=float(speed))
+            config.clamp()
+            atempo_filter = speed_engine.build_audio_filter(config)
+            if atempo_filter:
+                audio_filters.append(atempo_filter)
         else:
-            filter_parts.append(f"[{i}:a]asetpts=PTS-STARTPTS[a{i}]")
+            audio_filters.append("asetpts=PTS-STARTPTS")
+
+        # 音频倒放
+        reverse_config = ReverseConfig.from_dict(clip.config.get("reverse"))
+        if reverse_config.enabled and reverse_config.reverse_audio:
+            reverse_filter = ReverseEngine.build_audio_filter(reverse_config, duration=effective_duration)
+            if reverse_filter:
+                audio_filters.append(reverse_filter)
+
+        filter_parts.append(f"[{i}:a]{','.join(audio_filters)}[a{i}]")
 
     audio_labels = "".join(f"[a{i}]" for i in range(len(clips)))
     filter_parts.append(f"{audio_labels}concat=n={len(clips)}:v=0:a=1[outa]")
@@ -236,9 +406,11 @@ def mix_with_independent_audio(
         for clip in main_clips:
             input_args.extend(["-i", str(clip.local_path)])
             effective_duration = clip_effective_duration(clip)
+            trim_start = getattr(clip, "start_time", 0) or 0
             if effective_duration > 0:
                 filter_parts.append(
-                    f"[{input_idx}:a]atrim=0:{effective_duration:.3f},asetpts=PTS-STARTPTS[ma{input_idx}]"
+                    f"[{input_idx}:a]atrim=start={trim_start:.3f}:duration={effective_duration:.3f},"
+                    f"asetpts=PTS-STARTPTS[ma{input_idx}]"
                 )
             else:
                 filter_parts.append(f"[{input_idx}:a]asetpts=PTS-STARTPTS[ma{input_idx}]")
@@ -255,11 +427,12 @@ def mix_with_independent_audio(
     for j, clip in enumerate(audio_clips):
         input_args.extend(["-i", str(clip.local_path)])
         effective_duration = clip_effective_duration(clip)
+        trim_start = getattr(clip, "start_time", 0) or 0
         volume = clip.config.get("volume", 1.0) if clip.config else 1.0
         label = f"ia{j}"
         filters = []
         if effective_duration > 0:
-            filters.append(f"atrim=0:{effective_duration:.3f}")
+            filters.append(f"atrim=start={trim_start:.3f}:duration={effective_duration:.3f}")
         filters.append("asetpts=PTS-STARTPTS")
         if volume != 1.0:
             filters.append(f"volume={volume}")

apps/worker/video_processing/reverse_engine.py

@@ -0,0 +1,116 @@
+"""视频倒放引擎 — 基于 FFmpeg reverse + areverse 滤镜实现视频/音频倒放.
+
+支持能力：
+- 视频倒放（reverse 滤镜）
+- 音频倒放（areverse 滤镜）
+- 按 clip 分段倒放，每个 clip 独立配置
+- 降级策略：不支持时跳过，不阻断渲染
+"""
+
+from __future__ import annotations
+
+import logging
+from dataclasses import dataclass
+from typing import Any
+
+logger = logging.getLogger(__name__)
+
+
+# ── 数据模型 ──────────────────────────────────────────────────────────────────
+
+
+@dataclass
+class ReverseConfig:
+    """视频倒放配置.
+
+    从 clip.config.reverse 读取，零侵入数据模型.
+    """
+
+    enabled: bool = False
+    reverse_video: bool = True  # 是否倒放视频
+    reverse_audio: bool = True  # 是否倒放音频
+
+    @classmethod
+    def from_dict(cls, data: dict[str, Any] | None) -> "ReverseConfig":
+        """从字典解析配置."""
+        if not data:
+            return cls(enabled=False)
+        try:
+            if not data.get("enabled", False):
+                return cls(enabled=False)
+            return cls(
+                enabled=True,
+                reverse_video=bool(data.get("reverse_video", True)),
+                reverse_audio=bool(data.get("reverse_audio", True)),
+            )
+        except (AttributeError, TypeError) as e:
+            logger.warning("倒放配置解析失败: %s，使用默认配置", e)
+            return cls(enabled=False)
+
+
+# ── 倒放引擎 ──────────────────────────────────────────────────────────────────
+
+
+class ReverseEngine:
+    """视频倒放引擎 — 生成 FFmpeg 倒放滤镜.
+
+    视频倒放：reverse 滤镜
+    音频倒放：areverse 滤镜
+
+    注意事项：
+    - reverse 滤镜需要将整个视频帧加载到内存，长视频可能占用大量内存
+    - 建议对单 clip 时长做限制（如 < 60s），超长视频建议降级
+    """
+
+    # 安全限制：单 clip 超过此时长不启用倒放（防止内存溢出）
+    MAX_SAFE_DURATION = 120.0  # 秒
+
+    @staticmethod
+    def build_video_filter(config: ReverseConfig, duration: float = 0.0) -> str:
+        """构建视频倒放滤镜字符串.
+
+        Args:
+            config: 倒放配置
+            duration: clip 时长（秒），用于安全检查
+
+        Returns:
+            FFmpeg 滤镜字符串，如 "reverse"；无效果返回空字符串
+        """
+        if not config.enabled or not config.reverse_video:
+            return ""
+
+        # 安全检查：超长视频不启用倒放
+        if duration > ReverseEngine.MAX_SAFE_DURATION:
+            logger.warning(
+                "视频倒放安全限制：clip 时长 %.1fs 超过上限 %.1fs，跳过倒放",
+                duration,
+                ReverseEngine.MAX_SAFE_DURATION,
+            )
+            return ""
+
+        return "reverse"
+
+    @staticmethod
+    def build_audio_filter(config: ReverseConfig, duration: float = 0.0) -> str:
+        """构建音频倒放滤镜字符串.
+
+        Args:
+            config: 倒放配置
+            duration: clip 时长（秒），用于安全检查
+
+        Returns:
+            FFmpeg 音频滤镜字符串，如 "areverse"；无效果返回空字符串
+        """
+        if not config.enabled or not config.reverse_audio:
+            return ""
+
+        # 安全检查：超长音频不启用倒放
+        if duration > ReverseEngine.MAX_SAFE_DURATION:
+            logger.warning(
+                "音频倒放安全限制：clip 时长 %.1fs 超过上限 %.1fs，跳过倒放",
+                duration,
+                ReverseEngine.MAX_SAFE_DURATION,
+            )
+            return ""
+
+        return "areverse"

apps/worker/video_processing/speed_engine.py

@@ -0,0 +1,167 @@
+"""视频调速引擎 — 基于 FFmpeg setpts + atempo 的速度调整能力。
+
+支持：
+- 0.25x ~ 4x 变速范围
+- 视频调速（setpts）
+- 音频调速（atempo，多级串联处理超范围值）
+- 音调修正（pitch_correct，默认开启）
+- 边界自动钳制，不阻断渲染
+"""
+
+from dataclasses import dataclass
+from typing import Optional
+
+# ─── 常量 ───────────────────────────────────────────────
+MIN_SPEED = 0.25
+MAX_SPEED = 4.0
+DEFAULT_SPEED = 1.0
+
+# atempo 单级有效范围
+_ATEMPO_MIN = 0.5
+_ATEMPO_MAX = 2.0
+
+
+@dataclass
+class SpeedConfig:
+    """调速配置。
+
+    Attributes:
+        speed: 播放速度，0.25~4.0，1.0 为原速
+        pitch_correct: 是否保持音调（默认 True，用 atempo 时间拉伸算法）
+    """
+
+    speed: float = DEFAULT_SPEED
+    pitch_correct: bool = True
+
+    @classmethod
+    def parse(cls, data: Optional[dict]) -> "SpeedConfig":
+        """从 dict 解析配置，无效值回退到默认。"""
+        if not data or not isinstance(data, dict):
+            return cls()
+
+        speed = data.get("speed", DEFAULT_SPEED)
+        if not isinstance(speed, (int, float)):
+            speed = DEFAULT_SPEED
+
+        pitch_correct = data.get("pitch_correct", True)
+        if not isinstance(pitch_correct, bool):
+            pitch_correct = True
+
+        config = cls(speed=float(speed), pitch_correct=pitch_correct)
+        config.clamp()
+        return config
+
+    def clamp(self) -> None:
+        """将速度钳制到合法范围。"""
+        if self.speed <= 0:
+            self.speed = DEFAULT_SPEED
+        elif self.speed < MIN_SPEED:
+            self.speed = MIN_SPEED
+        elif self.speed > MAX_SPEED:
+            self.speed = MAX_SPEED
+
+    @property
+    def is_original(self) -> bool:
+        """是否原速（无需调速）。"""
+        return abs(self.speed - 1.0) < 1e-6
+
+
+class SpeedEngine:
+    """调速引擎 — 生成 FFmpeg 调速滤镜链。
+
+    用法：
+        engine = SpeedEngine()
+        video_filter = engine.build_video_filter(config)
+        audio_filter = engine.build_audio_filter(config)
+        new_duration = engine.adjust_duration(duration, config)
+    """
+
+    def build_video_filter(self, config: SpeedConfig) -> str:
+        """生成视频调速滤镜字符串。
+
+        返回 setpts 滤镜表达式，原速时返回空字符串。
+        """
+        if config.is_original:
+            return ""
+        # setpts=PTS/speed — speed>1 加速，speed<1 减速
+        return f"setpts=PTS/{config.speed:.4f}"
+
+    def build_audio_filter(self, config: SpeedConfig) -> str:
+        """生成音频调速滤镜字符串。
+
+        atempo 单级范围 0.5~2.0，超出范围时自动多级串联：
+        - 0.25x → atempo=0.5,atempo=0.5
+        - 4x → atempo=2.0,atempo=2.0
+        - 0.3x → atempo=0.5,atempo=0.6
+        - 3x → atempo=2.0,atempo=1.5
+
+        原速时返回空字符串。
+        """
+        if config.is_original:
+            return ""
+
+        speed = config.speed
+        stages: list[float] = self._split_atempo_stages(speed)
+        return ",".join(f"atempo={s:.4f}" for s in stages)
+
+    @staticmethod
+    def _split_atempo_stages(speed: float) -> list[float]:
+        """将速度拆分为多级 atempo 串联，每级都在 [0.5, 2.0] 范围内。"""
+        if _ATEMPO_MIN <= speed <= _ATEMPO_MAX:
+            return [speed]
+
+        stages: list[float] = []
+        remaining = speed
+
+        # 加速场景（speed > 2.0）
+        if speed > _ATEMPO_MAX:
+            while remaining > _ATEMPO_MAX:
+                stages.append(_ATEMPO_MAX)
+                remaining /= _ATEMPO_MAX
+            stages.append(remaining)
+
+        # 减速场景（speed < 0.5）
+        else:
+            while remaining < _ATEMPO_MIN:
+                stages.append(_ATEMPO_MIN)
+                remaining /= _ATEMPO_MIN
+            stages.append(remaining)
+
+        return stages
+
+    def adjust_duration(self, original_duration: float, config: SpeedConfig) -> float:
+        """计算调速后的时长。
+
+        加速 → 时长变短；减速 → 时长变长。
+        """
+        if config.is_original or original_duration <= 0:
+            return original_duration
+        return original_duration / config.speed
+
+    def build_clip_speed_filter(
+        self,
+        speed: float,
+        pitch_correct: bool = True,
+    ) -> tuple[str, str, SpeedConfig]:
+        """便捷方法：从单一 speed 值生成视频+音频滤镜。
+
+        返回 (video_filter, audio_filter, config)。
+        """
+        config = SpeedConfig(speed=speed, pitch_correct=pitch_correct)
+        config.clamp()
+        return (
+            self.build_video_filter(config),
+            self.build_audio_filter(config),
+            config,
+        )
+
+    @staticmethod
+    def resolve_clip_speed(
+        clip_config: dict,
+        global_speed: float = DEFAULT_SPEED,
+    ) -> float:
+        """从 clip config 中解析 playback_speed，0 或缺失则使用全局速度。"""
+        speed = clip_config.get("playback_speed", 0) if clip_config else 0
+        if not isinstance(speed, (int, float)) or speed <= 0:
+            return global_speed
+        return float(speed)

apps/worker/video_processing/sticker_engine.py

@@ -0,0 +1,574 @@
+"""贴纸叠加引擎 — 基于 FFmpeg overlay + drawtext 实现图片/文字贴纸.
+
+支持能力：
+- 图片贴纸（PNG/GIF）：位置、大小、透明度、时间范围、淡入淡出
+- 文字贴纸（花字）：字体、颜色、描边、阴影、位置、时间范围、动画
+- 9宫格位置 + 自由坐标（像素或百分比）
+- 多贴纸叠加，按 z_index 排序
+- 降级策略：素材不存在/无效时自动跳过，不阻断渲染
+"""
+
+from __future__ import annotations
+
+import logging
+from dataclasses import dataclass, field
+from pathlib import Path
+from typing import Any
+
+logger = logging.getLogger(__name__)
+
+
+# ── 预设贴纸分类 ──────────────────────────────────────────────────────────────
+
+# 预设贴纸分类（仅用于前端展示，后端不依赖具体素材）
+STICKER_CATEGORIES = [
+    ("emoji", "表情包"),
+    ("text", "文字花字"),
+    ("decoration", "装饰"),
+    ("arrow", "箭头指示"),
+    ("frame", "边框"),
+]
+
+# 9宫格位置映射
+POSITION_PRESETS = {
+    "top_left": (0.05, 0.05),
+    "top_center": (0.5, 0.05),
+    "top_right": (0.95, 0.05),
+    "center_left": (0.05, 0.5),
+    "center": (0.5, 0.5),
+    "center_right": (0.95, 0.5),
+    "bottom_left": (0.05, 0.95),
+    "bottom_center": (0.5, 0.95),
+    "bottom_right": (0.95, 0.95),
+}
+
+
+# ── 数据模型 ──────────────────────────────────────────────────────────────────
+
+
+@dataclass
+class ImageStickerConfig:
+    """图片贴纸配置."""
+
+    enabled: bool = False
+    type: str = "image"  # image / text
+    # 位置
+    position: str = "top_right"  # 9宫格预设
+    x: float | None = None  # 自定义x（像素或百分比）
+    y: float | None = None  # 自定义y
+    x_unit: str = "percent"  # pixel / percent
+    y_unit: str = "percent"
+    # 大小
+    scale: float = 1.0  # 缩放比例（相对于原始大小）
+    width: int | None = None  # 指定宽度（像素）
+    height: int | None = None  # 指定高度（像素）
+    # 透明度
+    opacity: float = 1.0  # 0.0~1.0
+    # 时间范围
+    start_time: float = 0.0
+    duration: float = 0.0  # 0 表示持续到结束
+    # 动画
+    fade_in: float = 0.0  # 淡入时长（秒）
+    fade_out: float = 0.0  # 淡出时长
+    # 层级
+    z_index: int = 10
+    # 素材
+    image_url: str = ""  # 图片URL或本地路径
+    preset_id: str = ""  # 预设贴纸ID
+
+
+@dataclass
+class TextStickerConfig:
+    """文字贴纸配置."""
+
+    enabled: bool = False
+    type: str = "text"
+    text: str = ""
+    # 字体
+    font_size: int = 36
+    font_color: str = "#FFFFFF"
+    font_family: str = "sans"
+    # 描边
+    stroke_color: str = "#000000"
+    stroke_width: int = 2
+    # 阴影
+    shadow_color: str = "#000000"
+    shadow_x: int = 2
+    shadow_y: int = 2
+    shadow_alpha: float = 0.5
+    # 位置
+    position: str = "center"
+    x: float | None = None
+    y: float | None = None
+    x_unit: str = "percent"
+    y_unit: str = "percent"
+    # 时间范围
+    start_time: float = 0.0
+    duration: float = 0.0
+    # 动画
+    fade_in: float = 0.0
+    fade_out: float = 0.0
+    # 层级
+    z_index: int = 10
+    # 背景框
+    bg_color: str = ""  # 空表示无背景
+    bg_padding: int = 8
+    bg_alpha: float = 0.8
+    bg_corner_radius: int = 8
+
+
+@dataclass
+class StickerOverlayResult:
+    """贴纸叠加结果."""
+
+    filter_str: str  # 滤镜字符串
+    output_label: str  # 输出标签
+    extra_inputs: list[str] = field(default_factory=list)  # 额外的输入文件路径
+
+
+# ── 贴纸引擎 ──────────────────────────────────────────────────────────────────
+
+
+class StickerEngine:
+    """贴纸叠加引擎 — 生成 FFmpeg overlay / drawtext 滤镜链.
+
+    支持图片贴纸（overlay）和文字贴纸（drawtext）。
+    多贴纸按 z_index 排序依次叠加。
+    """
+
+    @staticmethod
+    def _resolve_position(
+        config: ImageStickerConfig | TextStickerConfig,
+        canvas_w: int,
+        canvas_h: int,
+        sticker_w: int = 0,
+        sticker_h: int = 0,
+    ) -> tuple[float, float]:
+        """解析贴纸位置（像素坐标）.
+
+        优先级：自定义坐标 > 9宫格预设
+        """
+        # 先取预设的基准位置
+        if config.position in POSITION_PRESETS:
+            px, py = POSITION_PRESETS[config.position]
+        else:
+            px, py = 0.5, 0.5  # 默认居中
+
+        # 自定义坐标覆盖
+        if config.x is not None:
+            if config.x_unit == "percent":
+                px = config.x / 100.0
+            else:
+                px = config.x / canvas_w if canvas_w > 0 else 0.5
+
+        if config.y is not None:
+            if config.y_unit == "percent":
+                py = config.y / 100.0
+            else:
+                py = config.y / canvas_h if canvas_h > 0 else 0.5
+
+        # 转换为像素坐标（考虑贴纸尺寸，使位置为贴纸中心点）
+        x = px * canvas_w - sticker_w / 2
+        y = py * canvas_h - sticker_h / 2
+
+        # 钳制在画布内
+        x = max(0, min(x, canvas_w - sticker_w))
+        y = max(0, min(y, canvas_h - sticker_h))
+
+        return x, y
+
+    @staticmethod
+    def _build_overlay_filter(
+        sticker: ImageStickerConfig,
+        sticker_idx: int,
+        input_label: str,
+        output_label: str,
+        canvas_w: int,
+        canvas_h: int,
+    ) -> str:
+        """构建单个图片贴纸的 overlay 滤镜.
+
+        Args:
+            sticker: 贴纸配置
+            sticker_idx: 贴纸索引（用于生成滤镜标签）
+            input_label: 输入视频标签（如 "[base]"）
+            output_label: 输出视频标签
+            canvas_w: 画布宽度
+            canvas_h: 画布高度
+
+        Returns:
+            FFmpeg 滤镜字符串
+        """
+        sticker_label = f"sticker_{sticker_idx}_scaled"
+
+        # 1. 贴纸缩放预处理
+        scale_parts = []
+        if sticker.width and sticker.height:
+            scale_parts.append(f"scale={sticker.width}:{sticker.height}")
+        elif sticker.scale != 1.0:
+            # 按比例缩放
+            scale_parts.append(f"scale=iw*{sticker.scale}:ih*{sticker.scale}")
+        # 透明度调整
+        if sticker.opacity < 1.0:
+            scale_parts.append(f"colorchannelmixer=aa={sticker.opacity}")
+
+        # 淡入淡出
+        fade_parts = []
+        if sticker.fade_in > 0:
+            fade_parts.append(f"fade=in:st={sticker.start_time}:d={sticker.fade_in}:alpha=1")
+        if sticker.fade_out > 0 and sticker.duration > 0:
+            fade_out_start = sticker.start_time + sticker.duration - sticker.fade_out
+            fade_parts.append(f"fade=out:st={max(0, fade_out_start)}:d={sticker.fade_out}:alpha=1")
+
+        pre_filters = scale_parts + fade_parts
+
+        # 2. overlay 位置
+        # 先估算贴纸尺寸（假设原始尺寸 ~ canvas_w * 0.3）
+        est_w = int(canvas_w * 0.3 * sticker.scale) if not sticker.width else sticker.width
+        est_h = int(canvas_h * 0.3 * sticker.scale) if not sticker.height else sticker.height
+        pos_x, pos_y = StickerEngine._resolve_position(sticker, canvas_w, canvas_h, est_w, est_h)
+
+        # 3. enable 表达式（时间范围）
+        enable_expr = ""
+        if sticker.duration > 0:
+            enable_expr = f":enable='between(t,{sticker.start_time},{sticker.start_time + sticker.duration})'"
+
+        # 组合滤镜
+        filter_parts: list[str] = []
+
+        # 贴纸预处理
+        if pre_filters:
+            filter_parts.append(f"[{sticker_idx + 1}:v]{','.join(pre_filters)}[{sticker_label}]")
+            sticker_source = f"[{sticker_label}]"
+        else:
+            sticker_source = f"[{sticker_idx + 1}:v]"
+
+        # overlay 合成
+        filter_parts.append(f"{input_label}{sticker_source}overlay={pos_x:.0f}:{pos_y:.0f}{enable_expr}{output_label}")
+
+        return ";".join(filter_parts)
+
+    @staticmethod
+    def _build_drawtext_filter(
+        sticker: TextStickerConfig,
+        input_label: str,
+        output_label: str,
+        canvas_w: int,
+        canvas_h: int,
+    ) -> str:
+        """构建单个文字贴纸的 drawtext 滤镜.
+
+        Args:
+            sticker: 文字贴纸配置
+            input_label: 输入视频标签
+            output_label: 输出视频标签
+            canvas_w: 画布宽度
+            canvas_h: 画布高度
+
+        Returns:
+            FFmpeg 滤镜字符串
+        """
+        if not sticker.text:
+            return f"{input_label}copy{output_label}"
+
+        # 估算文字尺寸（粗略）
+        est_w = len(sticker.text) * sticker.font_size * 0.6
+        est_h = sticker.font_size * 1.4
+
+        pos_x, pos_y = StickerEngine._resolve_position(sticker, canvas_w, canvas_h, int(est_w), int(est_h))
+
+        drawtext_params: list[str] = []
+
+        # 文字内容（转义特殊字符）
+        escaped_text = sticker.text.replace(":", "\\:").replace("'", "\\'")
+        drawtext_params.append(f"text='{escaped_text}'")
+
+        # 字体
+        drawtext_params.append(f"fontsize={sticker.font_size}")
+        drawtext_params.append(f"fontcolor={sticker.font_color}")
+
+        # 描边
+        if sticker.stroke_width > 0:
+            drawtext_params.append(f"borderw={sticker.stroke_width}")
+            drawtext_params.append(f"bordercolor={sticker.stroke_color}")
+
+        # 阴影
+        if sticker.shadow_alpha > 0:
+            drawtext_params.append(f"shadowx={sticker.shadow_x}")
+            drawtext_params.append(f"shadowy={sticker.shadow_y}")
+            drawtext_params.append(f"shadowcolor={sticker.shadow_color}@{sticker.shadow_alpha}")
+
+        # 位置
+        drawtext_params.append(f"x={pos_x:.0f}")
+        drawtext_params.append(f"y={pos_y:.0f}")
+
+        # 时间范围
+        if sticker.duration > 0:
+            drawtext_params.append(f"enable='between(t,{sticker.start_time},{sticker.start_time + sticker.duration})'")
+
+        # 淡入淡出（drawtext 没有直接的淡入淡出，用 alpha 表达式模拟）
+        if sticker.fade_in > 0 or sticker.fade_out > 0:
+            alpha_expr = "1"
+            parts: list[str] = []
+            if sticker.fade_in > 0:
+                parts.append(
+                    f"if(lt(t,{sticker.start_time + sticker.fade_in})," f"(t-{sticker.start_time})/{sticker.fade_in},1)"
+                )
+            if sticker.fade_out > 0 and sticker.duration > 0:
+                fade_out_start = sticker.start_time + sticker.duration - sticker.fade_out
+                parts.append(
+                    f"if(gt(t,{fade_out_start})," f"({sticker.start_time + sticker.duration}-t)/{sticker.fade_out},1)"
+                )
+            if parts:
+                alpha_expr = "*".join(parts)
+                drawtext_params.append(f"alpha='{alpha_expr}'")
+
+        filter_str = f"{input_label}drawtext={':'.join(drawtext_params)}{output_label}"
+        return filter_str
+
+    @classmethod
+    def build_sticker_chain(
+        cls,
+        stickers: list[dict[str, Any]],
+        input_label: str,
+        output_label: str,
+        canvas_w: int,
+        canvas_h: int,
+    ) -> StickerOverlayResult:
+        """构建多贴纸叠加滤镜链.
+
+        Args:
+            stickers: 贴纸配置列表
+            input_label: 初始输入标签
+            output_label: 最终输出标签
+            canvas_w: 画布宽度
+            canvas_h: 画布高度
+
+        Returns:
+            StickerOverlayResult，包含滤镜字符串、输出标签、额外输入
+        """
+        if not stickers:
+            return StickerOverlayResult(
+                filter_str=f"{input_label}copy{output_label}",
+                output_label=output_label,
+                extra_inputs=[],
+            )
+
+        # 解析配置
+        parsed_stickers: list[tuple[int, ImageStickerConfig | TextStickerConfig]] = []
+        image_stickers: list[ImageStickerConfig] = []
+        image_paths: list[str] = []
+
+        for i, s in enumerate(stickers):
+            try:
+                sticker_type = s.get("type", "image")
+                z = int(s.get("z_index", 10))
+
+                if sticker_type == "text":
+                    config = TextStickerConfig(
+                        enabled=True,
+                        text=str(s.get("text", "")),
+                        font_size=int(s.get("font_size", 36)),
+                        font_color=str(s.get("font_color", "#FFFFFF")),
+                        stroke_color=str(s.get("stroke_color", "#000000")),
+                        stroke_width=int(s.get("stroke_width", 2)),
+                        shadow_x=int(s.get("shadow_x", 2)),
+                        shadow_y=int(s.get("shadow_y", 2)),
+                        shadow_alpha=float(s.get("shadow_alpha", 0.5)),
+                        position=str(s.get("position", "center")),
+                        x=cls._safe_float(s.get("x")),
+                        y=cls._safe_float(s.get("y")),
+                        x_unit=str(s.get("x_unit", "percent")),
+                        y_unit=str(s.get("y_unit", "percent")),
+                        start_time=float(s.get("start_time", 0)),
+                        duration=float(s.get("duration", 0)),
+                        fade_in=float(s.get("fade_in", 0)),
+                        fade_out=float(s.get("fade_out", 0)),
+                        z_index=z,
+                        bg_color=str(s.get("bg_color", "")),
+                        bg_padding=int(s.get("bg_padding", 8)),
+                        bg_alpha=float(s.get("bg_alpha", 0.8)),
+                        bg_corner_radius=int(s.get("bg_corner_radius", 8)),
+                    )
+                    parsed_stickers.append((z, config))
+                else:
+                    # 图片贴纸
+                    image_path = s.get("image_path", "") or s.get("image_url", "")
+                    if not image_path or not Path(image_path).exists():
+                        logger.warning("贴纸素材不存在，跳过: %s", image_path)
+                        continue
+
+                    config = ImageStickerConfig(
+                        enabled=True,
+                        position=str(s.get("position", "top_right")),
+                        x=cls._safe_float(s.get("x")),
+                        y=cls._safe_float(s.get("y")),
+                        x_unit=str(s.get("x_unit", "percent")),
+                        y_unit=str(s.get("y_unit", "percent")),
+                        scale=float(s.get("scale", 1.0)),
+                        width=int(s["width"]) if s.get("width") else None,
+                        height=int(s["height"]) if s.get("height") else None,
+                        opacity=max(0.0, min(1.0, float(s.get("opacity", 1.0)))),
+                        start_time=float(s.get("start_time", 0)),
+                        duration=float(s.get("duration", 0)),
+                        fade_in=float(s.get("fade_in", 0)),
+                        fade_out=float(s.get("fade_out", 0)),
+                        z_index=z,
+                        image_url=str(s.get("image_url", "")),
+                    )
+                    parsed_stickers.append((z, config))
+                    image_stickers.append(config)
+                    image_paths.append(image_path)
+
+            except Exception as e:
+                logger.warning("贴纸配置解析失败，跳过: %s", e)
+                continue
+
+        if not parsed_stickers:
+            return StickerOverlayResult(
+                filter_str=f"{input_label}copy{output_label}",
+                output_label=output_label,
+                extra_inputs=[],
+            )
+
+        # 按 z_index 排序
+        parsed_stickers.sort(key=lambda x: x[0])
+
+        # 构建滤镜链
+        filter_parts: list[str] = []
+        current_label = input_label
+        img_idx = 0  # 图片贴纸的输入索引偏移
+
+        for idx, (_, sticker) in enumerate(parsed_stickers):
+            next_label = f"sticker_{idx}_out" if idx < len(parsed_stickers) - 1 else output_label
+
+            if isinstance(sticker, ImageStickerConfig):
+                # 图片贴纸：使用额外的输入（输入索引 = 1 + img_idx，0 是主视频）
+                # 注意：实际输入索引需要调用方根据输入列表确定
+                # 这里我们按 image_stickers 的顺序分配索引
+                # 主输入是 [0:v]，贴纸输入从 [1:v] 开始
+                single_filter = cls._build_single_image_sticker(
+                    sticker=sticker,
+                    sticker_input_idx=img_idx + 1,  # +1 因为 0 是主视频
+                    input_label=current_label,
+                    output_label=next_label,
+                    canvas_w=canvas_w,
+                    canvas_h=canvas_h,
+                )
+                filter_parts.append(single_filter)
+                img_idx += 1
+            else:
+                # 文字贴纸：drawtext，不需要额外输入
+                single_filter = cls._build_drawtext_filter(
+                    sticker,  # type: ignore
+                    current_label,
+                    next_label,
+                    canvas_w,
+                    canvas_h,
+                )
+                filter_parts.append(single_filter)
+
+            current_label = next_label
+
+        return StickerOverlayResult(
+            filter_str=";".join(filter_parts),
+            output_label=output_label,
+            extra_inputs=image_paths,
+        )
+
+    @classmethod
+    def _build_single_image_sticker(
+        cls,
+        sticker: ImageStickerConfig,
+        sticker_input_idx: int,
+        input_label: str,
+        output_label: str,
+        canvas_w: int,
+        canvas_h: int,
+    ) -> str:
+        """构建单个图片贴纸的完整滤镜（预处理 + overlay）.
+
+        Args:
+            sticker: 贴纸配置
+            sticker_input_idx: 贴纸在 FFmpeg 输入中的索引
+            input_label: 输入视频标签
+            output_label: 输出标签
+            canvas_w: 画布宽
+            canvas_h: 画布高
+        """
+        scaled_label = f"sticker_s{sticker_input_idx}"
+
+        # 预处理滤镜（缩放 + 透明度 + 淡入淡出）
+        pre_filters: list[str] = []
+
+        # 缩放
+        if sticker.width and sticker.height:
+            pre_filters.append(f"scale={sticker.width}:{sticker.height}")
+        elif sticker.scale != 1.0:
+            pre_filters.append(f"scale=iw*{sticker.scale}:ih*{sticker.scale}")
+
+        # 透明度
+        if sticker.opacity < 1.0:
+            pre_filters.append(f"format=rgba,colorchannelmixer=aa={sticker.opacity}")
+
+        # 淡入淡出（使用 fade 的 alpha 模式）
+        fade_filters: list[str] = []
+        if sticker.fade_in > 0:
+            fade_filters.append(f"fade=in:st={sticker.start_time}:d={sticker.fade_in}:alpha=1")
+        if sticker.fade_out > 0 and sticker.duration > 0:
+            fade_out_start = sticker.start_time + sticker.duration - sticker.fade_out
+            if fade_out_start > 0:
+                fade_filters.append(f"fade=out:st={fade_out_start}:d={sticker.fade_out}:alpha=1")
+
+        # 估算贴纸尺寸用于位置计算
+        est_w = int(canvas_w * 0.3 * sticker.scale) if not sticker.width else sticker.width
+        est_h = int(canvas_h * 0.3 * sticker.scale) if not sticker.height else sticker.height
+        pos_x, pos_y = cls._resolve_position(sticker, canvas_w, canvas_h, est_w, est_h)
+
+        # enable 表达式
+        enable_expr = ""
+        if sticker.duration > 0:
+            enable_expr = f":enable='between(t,{sticker.start_time},{sticker.start_time + sticker.duration})'"
+
+        parts: list[str] = []
+
+        # 贴纸预处理
+        all_pre = pre_filters + fade_filters
+        if all_pre:
+            parts.append(f"[{sticker_input_idx}:v]{','.join(all_pre)}[{scaled_label}]")
+            sticker_source = f"[{scaled_label}]"
+        else:
+            sticker_source = f"[{sticker_input_idx}:v]"
+
+        # overlay 合成
+        parts.append(f"{input_label}{sticker_source}overlay={pos_x:.0f}:{pos_y:.0f}{enable_expr}{output_label}")
+
+        return ";".join(parts)
+
+    @staticmethod
+    def _safe_float(val: Any) -> float | None:
+        """安全转换 float."""
+        if val is None:
+            return None
+        try:
+            return float(val)
+        except (ValueError, TypeError):
+            return None
+
+
+# ── 便捷函数 ──────────────────────────────────────────────────────────────────
+
+
+def parse_stickers_from_config(config: dict[str, Any] | None) -> list[dict[str, Any]]:
+    """从 plan.config.stickers 解析贴纸列表."""
+    if not config:
+        return []
+    stickers = config.get("stickers", [])
+    if not isinstance(stickers, list):
+        return []
+    return stickers
+
+
+def get_sticker_categories() -> list[tuple[str, str]]:
+    """获取贴纸分类列表."""
+    return list(STICKER_CATEGORIES)

apps/worker/video_processing/subtitle_generator.py

@@ -0,0 +1,183 @@
+"""字幕生成器 — 将字幕时间轴转换为 ASS 字幕文件。
+
+与 render_subtitles.py 的区别：
+- render_subtitles.py 处理静态整段标题/字幕
+- 本模块处理带时间轴的多段 ASR 字幕
+
+两者最终都输出 ASS 文件，供 FFmpeg 烧录。
+"""
+
+from __future__ import annotations
+
+import logging
+from pathlib import Path
+from typing import Any
+
+from packages.domain.subtitle import SubtitleTimeline
+
+logger = logging.getLogger(__name__)
+
+
+# ── 常量 ──────────────────────────────────────────────────────────────────────
+
+DEFAULT_MAX_CHARS_PER_LINE = 20  # 每行最多字符数
+DEFAULT_MIN_CHARS_PER_SEGMENT = 8  # 每段最少字符数
+
+
+# ── ASS 工具函数 ────────────────────────────────────────────────────────────
+
+
+def _hex_to_ass_color(hex_color: str) -> str:
+    """将 HEX 颜色（#RRGGBB）转换为 ASS &HBBGGRR 格式。"""
+    hex_color = hex_color.lstrip("#")
+    if len(hex_color) != 6:
+        return "&H00FFFFFF"
+    r, g, b = hex_color[0:2], hex_color[2:4], hex_color[4:6]
+    return f"&H{b.upper()}{g.upper()}{r.upper()}"
+
+
+def _position_to_ass_alignment(position: str) -> int:
+    """将文字位置映射为 ASS \\an 对齐编号。"""
+    mapping = {
+        "top": 8,
+        "center": 5,
+        "bottom": 2,
+    }
+    return mapping.get(position, 2)
+
+
+def _format_ass_time(seconds: float) -> str:
+    """将秒数格式化为 ASS 时间格式 H:MM:SS.cc。"""
+    hours = int(seconds // 3600)
+    minutes = int((seconds % 3600) // 60)
+    secs = seconds % 60
+    return f"{hours}:{minutes:02d}:{secs:05.2f}"
+
+
+def _escape_ass_text(text: str) -> str:
+    """转义 ASS 文本中的特殊字符。"""
+    text = text.replace("\r\n", "\\N").replace("\n", "\\N").replace("\r", "\\N")
+    text = text.replace("{", "(").replace("}", ")")
+    return text
+
+
+def _wrap_text(text: str, max_chars: int) -> list[str]:
+    """将长文本按字数换行。
+
+    优先在标点处换行，没有合适标点时硬切。
+    """
+    if len(text) <= max_chars:
+        return [text]
+
+    lines: list[str] = []
+    remaining = text
+
+    while len(remaining) > max_chars:
+        # 在前 max_chars 个字符中找标点断开
+        break_point = max_chars
+        punctuations = "，。！？、；：,.;:!?"
+
+        for i in range(max_chars, max_chars // 2, -1):
+            if i < len(remaining) and remaining[i] in punctuations:
+                break_point = i + 1
+                break
+
+        lines.append(remaining[:break_point])
+        remaining = remaining[break_point:]
+
+    if remaining:
+        lines.append(remaining)
+
+    return lines
+
+
+# ── 主生成器 ─────────────────────────────────────────────────────────────────
+
+
+def generate_ass_from_timeline(
+    output_path: Path,
+    timeline: SubtitleTimeline,
+    *,
+    video_width: int,
+    video_height: int,
+    subtitle_config: dict[str, Any] | None = None,
+) -> Path:
+    """从字幕时间轴生成 ASS 字幕文件。
+
+    Args:
+        output_path: 输出 ASS 文件路径
+        timeline: 字幕时间轴
+        video_width: 视频宽度
+        video_height: 视频高度
+        subtitle_config: 字幕样式配置（同 SubtitleConfig dict）
+
+    Returns:
+        生成的 ASS 文件路径
+    """
+    subtitle_config = subtitle_config or {}
+
+    if not timeline.segments:
+        output_path.write_text("", encoding="utf-8")
+        return output_path
+
+    # 样式参数
+    font_name = subtitle_config.get("font", "思源黑体")
+    font_size = int(subtitle_config.get("size", 24))
+    color = _hex_to_ass_color(subtitle_config.get("color", "#ffffff"))
+    position = subtitle_config.get("position", "bottom")
+    alignment = _position_to_ass_alignment(position)
+    max_chars_per_line = int(subtitle_config.get("max_chars_per_line", DEFAULT_MAX_CHARS_PER_LINE))
+
+    # 描边（默认黑色描边，保证可读性）
+    outline_color = "&H00000000"
+    outline_width = 1.5
+
+    # 边距
+    margin_v = 60 if position == "bottom" else 60
+    margin_l = 40
+    margin_r = 40
+
+    # 生成样式行
+    style_line = (
+        f"Style: Default,{font_name},{font_size},{color},"
+        f"&H000000FF,{outline_color},&H00000000,"
+        f"-1,0,0,0,100,100,0,0,"
+        f"1,{outline_width},0,{alignment},"
+        f"{margin_l},{margin_r},{margin_v},1"
+    )
+
+    # 生成事件行
+    events: list[str] = []
+    for seg in timeline.segments:
+        start_time = _format_ass_time(seg.start)
+        end_time = _format_ass_time(seg.end)
+
+        # 自动换行
+        lines = _wrap_text(seg.text, max_chars_per_line)
+        display_text = "\\N".join(lines)
+
+        safe_text = _escape_ass_text(display_text)
+
+        events.append(f"Dialogue: 0,{start_time},{end_time},Default,,0,0,0,,{safe_text}")
+
+    # 组装 ASS 文件
+    ass_content = f"""[Script Info]
+ScriptType: v4.00+
+PlayResX: {video_width}
+PlayResY: {video_height}
+ScaledBorderAndShadow: yes
+WrapStyle: 2
+Encoding: UTF-8
+
+[V4+ Styles]
+Format: Name, Fontname, Fontsize, PrimaryColour, SecondaryColour, OutlineColour, BackColour, Bold, Italic, Underline, StrikeOut, ScaleX, ScaleY, Spacing, Angle, BorderStyle, Outline, Shadow, Alignment, MarginL, MarginR, MarginV, Encoding
+{style_line}
+
+[Events]
+Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
+{chr(10).join(events)}
+"""
+
+    output_path.parent.mkdir(parents=True, exist_ok=True)
+    output_path.write_text(ass_content, encoding="utf-8")
+    return output_path

apps/worker/video_processing/thumbnail_generator.py

@@ -0,0 +1,123 @@
+"""视频缩略图生成工具 — 抽取首帧上传到 OSS。"""
+
+from __future__ import annotations
+
+import logging
+import tempfile
+from pathlib import Path
+
+logger = logging.getLogger(__name__)
+
+
+def extract_first_frame(
+    video_path: str,
+    output_path: str | None = None,
+    *,
+    width: int = 640,
+    height: int = -1,
+    timeout: int = 30,
+) -> str:
+    """抽取视频第一帧作为封面图。
+
+    Args:
+        video_path: 视频文件路径
+        output_path: 输出图片路径，不传则用临时文件
+        width: 输出宽度（默认 640，-1 表示按比例缩放）
+        height: 输出高度（默认 -1，按比例缩放）
+        timeout: 超时时间（秒）
+
+    Returns:
+        生成的缩略图文件路径
+
+    Raises:
+        subprocess.CalledProcessError: ffmpeg 执行失败
+    """
+    from video_processing.ffmpeg_utils import FFMPEG_BIN, run_ffmpeg
+
+    if output_path is None:
+        tmp = tempfile.NamedTemporaryFile(suffix=".jpg", delete=False)
+        tmp.close()
+        output_path = tmp.name
+
+    # -ss 00:00:01 取第1秒帧（避免首帧黑屏）
+    # -vframes 1 只取一帧
+    # -q:v 2 jpeg 高质量
+    scale_filter = f"scale={width}:{height}:force_original_aspect_ratio=decrease"
+    cmd = [
+        FFMPEG_BIN,
+        "-y",
+        "-i",
+        video_path,
+        "-ss",
+        "00:00:01",
+        "-vframes",
+        "1",
+        "-vf",
+        scale_filter,
+        "-q:v",
+        "2",
+        output_path,
+    ]
+
+    try:
+        run_ffmpeg(cmd, capture_output=True, timeout=timeout)
+    except Exception:
+        # 短视频可能没有第1秒，退回到第0帧
+        cmd2 = [
+            FFMPEG_BIN,
+            "-y",
+            "-i",
+            video_path,
+            "-ss",
+            "00:00:00",
+            "-vframes",
+            "1",
+            "-vf",
+            scale_filter,
+            "-q:v",
+            "2",
+            output_path,
+        ]
+        run_ffmpeg(cmd2, capture_output=True, timeout=timeout)
+
+    if not Path(output_path).exists() or Path(output_path).stat().st_size == 0:
+        raise RuntimeError(f"Thumbnail generation failed: {output_path}")
+
+    return output_path
+
+
+def generate_and_upload_thumbnail(
+    video_path: str,
+    storage_key: str,
+) -> str | None:
+    """生成缩略图并上传到 OSS，返回 URL。
+
+    Args:
+        video_path: 本地视频路径
+        storage_key: OSS 存储 key（如 generated/projects/xxx/thumbnails/yyy.jpg）
+
+    Returns:
+        上传成功返回 URL，失败返回 None
+    """
+    thumbnail_path = None
+    try:
+        thumbnail_path = extract_first_frame(video_path)
+    except Exception as e:
+        logger.warning("Failed to extract thumbnail from %s: %s", video_path, e)
+        return None
+
+    try:
+        from video_processing.oss_helpers import upload_to_oss
+
+        url = upload_to_oss(thumbnail_path, storage_key)
+        return url
+    except Exception as e:
+        logger.warning("Failed to upload thumbnail to OSS: %s", e)
+        return None
+    finally:
+        # 清理临时文件
+        if thumbnail_path:
+            try:
+                Path(thumbnail_path).unlink(missing_ok=True)
+            except Exception:
+                pass

apps/worker/video_processing/transition_engine.py

@@ -0,0 +1,381 @@
+"""转场特效引擎 — Phase 8 智能增强.
+
+基于 FFmpeg xfade 滤镜的统一转场抽象层，提供：
+  1. 转场类型枚举与预设管理
+  2. 转场配置解析与边界校验
+  3. 降级策略（不支持的转场自动 fallback 到硬切）
+  4. xfade 滤镜链构建（封装底层 ffmpeg_utils）
+
+新增转场只需在 TransitionType 中加一项 + 在 XFADE_TRANSITION_MAP 中映射。
+"""
+
+from __future__ import annotations
+
+import logging
+import sys
+from dataclasses import dataclass
+
+if sys.version_info >= (3, 11):
+    from enum import StrEnum
+else:
+    from enum import Enum
+
+    class StrEnum(str, Enum):
+        pass
+
+
+from video_processing.ffmpeg_utils import build_xfade_filter_chain
+
+logger = logging.getLogger(__name__)
+
+
+# ── 常量 ──────────────────────────────────────────────────────────────────────
+
+# 转场时长范围（秒）
+MIN_TRANSITION_DURATION = 0.3
+MAX_TRANSITION_DURATION = 2.0
+DEFAULT_TRANSITION_DURATION = 0.5
+
+# 硬切（无转场）
+CUT_TRANSITION = "cut"
+
+
+# ── 转场类型枚举 ──────────────────────────────────────────────────────────────
+
+
+class TransitionType(StrEnum):
+    """支持的转场效果类型.
+
+    每种类型对应 FFmpeg xfade filter 的一个 transition 值。
+    新增转场只需在此添加一项，并在 _FFMPEG_XFADE_MAP 中映射。
+    """
+
+    # 硬切（无转场效果，直接拼接）
+    CUT = "cut"
+
+    # 淡入淡出（最常用，默认 fallback）
+    FADE = "fade"
+
+    # 溶解（交叉溶解）
+    DISSOLVE = "dissolve"
+
+    # 滑入系列
+    SLIDE_LEFT = "slideleft"
+    SLIDE_RIGHT = "slideright"
+    SLIDE_UP = "slideup"
+    SLIDE_DOWN = "slidedown"
+
+    # 缩放
+    ZOOM = "zoom"
+
+    # 擦除系列
+    WIPE_LEFT = "wipeleft"
+    WIPE_RIGHT = "wiperight"
+    WIPE_UP = "wipeup"
+    WIPE_DOWN = "wipedown"
+
+    # 圆形扩散
+    CIRCLE_CROP = "circlecrop"
+
+    # 矩形覆盖
+    RECT_CROP = "rectcrop"
+
+    @classmethod
+    def all_supported(cls) -> list[str]:
+        """返回所有支持的转场类型名称列表."""
+        return [t.value for t in cls if t != cls.CUT]
+
+    @classmethod
+    def is_supported(cls, name: str) -> bool:
+        """检查转场类型是否支持（不区分大小写和下划线）."""
+        normalized = _normalize_transition_name(name)
+        return normalized in _NAME_TO_ENUM_MAP
+
+
+# ── 名称 → 枚举 映射（支持多种别名）──────────────────────────────────────────
+
+
+def _normalize_transition_name(name: str) -> str:
+    """标准化转场名称：小写 + 去下划线."""
+    return name.lower().replace("_", "").replace("-", "")
+
+
+# 构建别名映射
+_NAME_TO_ENUM_MAP: dict[str, TransitionType] = {}
+for _t in TransitionType:
+    _NAME_TO_ENUM_MAP[_normalize_transition_name(_t.value)] = _t
+
+# 额外的别名
+_ALIASES: dict[str, TransitionType] = {
+    "dissolve": TransitionType.DISSOLVE,
+    "crossfade": TransitionType.DISSOLVE,
+    "crossdissolve": TransitionType.DISSOLVE,
+    "fadein": TransitionType.FADE,
+    "fadeout": TransitionType.FADE,
+    "fadeblack": TransitionType.FADE,
+    "slide": TransitionType.SLIDE_LEFT,  # 默认向左滑
+    "wipe": TransitionType.WIPE_LEFT,  # 默认向左擦
+    "zoomin": TransitionType.ZOOM,
+    "zoomout": TransitionType.ZOOM,
+    "circle": TransitionType.CIRCLE_CROP,
+    "rect": TransitionType.RECT_CROP,
+}
+for _alias, _type in _ALIASES.items():
+    _key = _normalize_transition_name(_alias)
+    if _key not in _NAME_TO_ENUM_MAP:
+        _NAME_TO_ENUM_MAP[_key] = _type
+
+
+# ── TransitionType → FFmpeg xfade transition 名称映射 ─────────────────────────
+
+
+_FFMPEG_XFADE_MAP: dict[TransitionType, str] = {
+    TransitionType.FADE: "fade",
+    TransitionType.DISSOLVE: "dissolve",
+    TransitionType.SLIDE_LEFT: "slideleft",
+    TransitionType.SLIDE_RIGHT: "slideright",
+    TransitionType.SLIDE_UP: "slideup",
+    TransitionType.SLIDE_DOWN: "slidedown",
+    TransitionType.ZOOM: "zoomin",
+    TransitionType.WIPE_LEFT: "wipeleft",
+    TransitionType.WIPE_RIGHT: "wiperight",
+    TransitionType.WIPE_UP: "wipeup",
+    TransitionType.WIPE_DOWN: "wipedown",
+    TransitionType.CIRCLE_CROP: "circlecrop",
+    TransitionType.RECT_CROP: "rectcrop",
+}
+
+
+# ── 转场配置 ──────────────────────────────────────────────────────────────────
+
+
+@dataclass(slots=True)
+class TransitionConfig:
+    """转场效果配置.
+
+    Attributes:
+        effect: 转场效果名称（见 TransitionType）
+        duration: 转场时长（秒），范围 0.3~2.0，默认 0.5
+    """
+
+    effect: str = CUT_TRANSITION
+    duration: float = DEFAULT_TRANSITION_DURATION
+
+    @classmethod
+    def parse(cls, effect: str | None = None, duration: float | None = None) -> "TransitionConfig":
+        """解析并验证转场配置，自动处理边界和降级.
+
+        Args:
+            effect: 转场效果名称（None 或空则使用默认 cut）
+            duration: 转场时长（None 则使用默认值）
+
+        Returns:
+            验证后的 TransitionConfig
+        """
+        # 处理 effect
+        final_effect = CUT_TRANSITION
+        if effect and effect.strip():
+            effect_clean = effect.strip()
+            if TransitionType.is_supported(effect_clean):
+                final_effect = _resolve_transition_enum(effect_clean).value
+            elif effect_clean.lower() == CUT_TRANSITION:
+                final_effect = CUT_TRANSITION
+            else:
+                # 降级：不支持的转场 → 硬切，不阻断渲染
+                logger.warning(
+                    "不支持的转场效果 '%s'，已降级为硬切（cut）",
+                    effect_clean,
+                )
+                final_effect = CUT_TRANSITION
+
+        # 处理 duration：边界钳制
+        final_duration = DEFAULT_TRANSITION_DURATION
+        if duration is not None:
+            try:
+                d = float(duration)
+                if d < MIN_TRANSITION_DURATION:
+                    logger.warning(
+                        "转场时长 %.3fs 小于最小值 %.1fs，已钳制到最小值",
+                        d,
+                        MIN_TRANSITION_DURATION,
+                    )
+                    final_duration = MIN_TRANSITION_DURATION
+                elif d > MAX_TRANSITION_DURATION:
+                    logger.warning(
+                        "转场时长 %.3fs 大于最大值 %.1fs，已钳制到最大值",
+                        d,
+                        MAX_TRANSITION_DURATION,
+                    )
+                    final_duration = MAX_TRANSITION_DURATION
+                else:
+                    final_duration = d
+            except (TypeError, ValueError):
+                logger.warning("无效的转场时长 '%s'，使用默认值 %.1fs", duration, DEFAULT_TRANSITION_DURATION)
+                final_duration = DEFAULT_TRANSITION_DURATION
+
+        return cls(effect=final_effect, duration=final_duration)
+
+    @property
+    def is_cut(self) -> bool:
+        """是否为硬切（无转场效果）."""
+        return self.effect == CUT_TRANSITION
+
+    @property
+    def ffmpeg_transition(self) -> str:
+        """获取对应的 FFmpeg xfade transition 名称."""
+        if self.is_cut:
+            return ""
+        enum_type = _resolve_transition_enum(self.effect)
+        return _FFMPEG_XFADE_MAP.get(enum_type, "fade")
+
+
+def _resolve_transition_enum(name: str) -> TransitionType:
+    """将名称解析为 TransitionType 枚举，必须先通过 is_supported 校验."""
+    normalized = _normalize_transition_name(name)
+    return _NAME_TO_ENUM_MAP.get(normalized, TransitionType.FADE)
+
+
+# ── 转场引擎 ──────────────────────────────────────────────────────────────────
+
+
+class TransitionEngine:
+    """转场特效引擎.
+
+    封装转场配置验证、降级策略和 xfade 滤镜链构建，
+    供 UnifiedRenderService 等上层调用。
+
+    用法::
+
+        engine = TransitionEngine(default_duration=0.5)
+        config = engine.resolve_config("fade", 0.8)
+        filter_str, total_dur = engine.build_xfade_chain(
+            clip_durations=[3.0, 4.0, 5.0],
+            clip_video_labels=["v0", "v1", "v2"],
+            transitions=["cut", "fade", "dissolve"],
+        )
+    """
+
+    def __init__(self, default_duration: float = DEFAULT_TRANSITION_DURATION) -> None:
+        """初始化转场引擎.
+
+        Args:
+            default_duration: 默认转场时长（秒），用于未指定时长的 clip
+        """
+        self._default_duration = default_duration
+
+    def resolve_config(
+        self,
+        effect: str | None = None,
+        duration: float | None = None,
+    ) -> TransitionConfig:
+        """解析单个转场配置，应用验证和降级.
+
+        Args:
+            effect: 转场效果名称
+            duration: 转场时长
+
+        Returns:
+            验证后的 TransitionConfig
+        """
+        # 若未指定 duration，使用引擎默认值
+        dur = duration if duration is not None else self._default_duration
+        return TransitionConfig.parse(effect=effect, duration=dur)
+
+    def resolve_clip_transitions(
+        self,
+        clip_transitions: list[str],
+        clip_durations: list[float] | None = None,
+    ) -> list[TransitionConfig]:
+        """批量解析 clip 级别的转场配置.
+
+        Args:
+            clip_transitions: 每个 clip 的转场效果名称列表
+            clip_durations: 每个 clip 的时长列表（用于验证转场时长不超过片段时长）
+
+        Returns:
+            TransitionConfig 列表
+        """
+        configs: list[TransitionConfig] = []
+        for i, effect in enumerate(clip_transitions):
+            cfg = self.resolve_config(effect=effect)
+            # 额外校验：转场时长不能超过对应 clip 时长的一半（保守限制）
+            if clip_durations and i < len(clip_durations) and not cfg.is_cut:
+                max_safe_duration = max(MIN_TRANSITION_DURATION, clip_durations[i] * 0.5)
+                if cfg.duration > max_safe_duration:
+                    cfg = TransitionConfig(effect=cfg.effect, duration=max_safe_duration)
+            configs.append(cfg)
+        return configs
+
+    def build_xfade_chain(
+        self,
+        clip_durations: list[float],
+        clip_video_labels: list[str],
+        transitions: list[str],
+        *,
+        transition_duration: float | None = None,
+        output_label: str = "outv",
+    ) -> tuple[str, float]:
+        """构建 xfade 转场滤镜链.
+
+        对每步转场应用验证和降级，然后调用底层 ffmpeg_utils 构建。
+
+        Args:
+            clip_durations: 每个片段的时长
+            clip_video_labels: 每个片段的视频流标签
+            transitions: 每个片段对应的转场效果
+            transition_duration: 统一转场时长，None 则使用引擎默认值
+            output_label: 最终输出标签
+
+        Returns:
+            (filter_string, estimated_total_duration)
+        """
+        if len(clip_durations) <= 1:
+            return build_xfade_filter_chain(
+                clip_durations=clip_durations,
+                clip_video_labels=clip_video_labels,
+                transitions=transitions,
+                transition_duration=transition_duration or self._default_duration,
+                output_label=output_label,
+            )
+
+        # 解析所有转场配置
+        resolved = self.resolve_clip_transitions(transitions, clip_durations)
+        resolved_effects = [c.effect for c in resolved]
+
+        # 使用统一的时长（取各转场中最大的时长作为基准，底层会做每步钳制）
+        dur = transition_duration or self._default_duration
+        if not dur:
+            dur = max(c.duration for c in resolved) if resolved else DEFAULT_TRANSITION_DURATION
+
+        # 调用底层构建
+        return build_xfade_filter_chain(
+            clip_durations=clip_durations,
+            clip_video_labels=clip_video_labels,
+            transitions=resolved_effects,
+            transition_duration=dur,
+            output_label=output_label,
+        )
+
+    @staticmethod
+    def supported_transitions() -> list[dict[str, str]]:
+        """获取所有支持的转场效果列表（用于 API 返回给前端）.
+
+        Returns:
+            [{name, display_name, category}, ...]
+        """
+        return [
+            {"name": "cut", "display_name": "硬切", "category": "basic"},
+            {"name": "fade", "display_name": "淡入淡出", "category": "basic"},
+            {"name": "dissolve", "display_name": "溶解", "category": "basic"},
+            {"name": "slideleft", "display_name": "左滑入", "category": "slide"},
+            {"name": "slideright", "display_name": "右滑入", "category": "slide"},
+            {"name": "slideup", "display_name": "上滑入", "category": "slide"},
+            {"name": "slidedown", "display_name": "下滑入", "category": "slide"},
+            {"name": "zoom", "display_name": "缩放", "category": "zoom"},
+            {"name": "wipeleft", "display_name": "左擦除", "category": "wipe"},
+            {"name": "wiperight", "display_name": "右擦除", "category": "wipe"},
+            {"name": "wipeup", "display_name": "上擦除", "category": "wipe"},
+            {"name": "wipedown", "display_name": "下擦除", "category": "wipe"},
+            {"name": "circlecrop", "display_name": "圆形扩散", "category": "special"},
+            {"name": "rectcrop", "display_name": "矩形扩散", "category": "special"},
+        ]

apps/worker/video_processing/trim_engine.py

@@ -0,0 +1,339 @@
+"""裁剪引擎 — 基于 FFmpeg trim/atrim 的精确帧级裁剪.
+
+支持：
+- 入点出点裁剪（start_time / end_time / duration 三选二）
+- 边界自动钳制（超出素材时长自动修正，不阻断渲染）
+- 多段裁剪（一个素材裁剪出多段）
+- 音画同步（视频 + 音频同步裁剪）
+"""
+
+from __future__ import annotations
+
+import logging
+from dataclasses import dataclass
+from typing import Any
+
+logger = logging.getLogger(__name__)
+
+# 最小裁剪时长（秒），低于此值视为无效
+MIN_TRIM_DURATION = 0.1
+
+
+@dataclass
+class TrimConfig:
+    """裁剪配置.
+
+    三选二规则：start_time / end_time / duration 中必须至少给出两个，
+    第三个会被自动推导。如果三个都给了，以 start_time + duration 为准。
+
+    边界保护：
+    - start_time < 0 → 钳制到 0
+    - end_time > 素材时长 → 钳制到素材时长
+    - 计算出的 duration < 最小阈值 → 标记为无效
+    """
+
+    start_time: float = 0.0  # 入点（素材内时间，秒）
+    end_time: float = 0.0  # 出点（素材内时间，秒），0 表示未指定
+    duration: float = 0.0  # 裁剪时长（秒），0 表示未指定
+
+    @classmethod
+    def from_dict(cls, data: dict[str, Any] | None) -> TrimConfig | None:
+        """从字典构造，无有效裁剪参数时返回 None（不裁剪）."""
+        if not data:
+            return None
+
+        start = float(data.get("start_time", 0) or 0)
+        end = float(data.get("end_time", 0) or 0)
+        dur = float(data.get("duration", 0) or 0)
+
+        # 三个参数都没有 → 不裁剪
+        if start <= 0 and end <= 0 and dur <= 0:
+            return None
+
+        # 至少有两个参数（或一个合理的 start/duration）
+        # 兼容：只传了 start_time → 从 start 开始取到末尾
+        # 兼容：只传了 duration → 从 0 开始取 duration
+        if start > 0 and end <= 0 and dur <= 0:
+            # 只有 start，取到末尾 → 这是"从某点开始"的语义，算有效
+            pass
+        elif dur > 0 and start <= 0 and end <= 0:
+            # 只有 duration → 从开头取 duration，算有效
+            pass
+        elif start <= 0 and end <= 0 and dur <= 0:
+            return None
+
+        return cls(start_time=start, end_time=end, duration=dur)
+
+    def validate_and_resolve(self, asset_duration: float) -> TrimConfig:
+        """根据素材实际时长，解析并钳制裁剪参数.
+
+        返回一个新的 TrimConfig，其中 start_time / end_time / duration 都已确定。
+        如果裁剪无效（时长为0或负数），仍返回但调用方应检查 is_valid。
+        """
+        start = self.start_time
+        end = self.end_time
+        dur = self.duration
+
+        # 边界：start 不能为负
+        if start < 0:
+            start = 0.0
+
+        # 边界：asset_duration 为 0 时保守处理（不裁剪，取全部）
+        if asset_duration <= 0:
+            return TrimConfig(start_time=0.0, end_time=0.0, duration=0.0)
+
+        # 三选二推导
+        # 判断顺序很重要：先判断需要两个显式值的组合，最后判断含默认值的
+        # 情况1：start + end 都有显式值
+        if start > 0 and end > 0:
+            if end <= start:
+                # 出点 <= 入点，无效 → 返回 start 处一个极短片段（调用方会判无效）
+                return TrimConfig(start_time=start, end_time=start, duration=0.0)
+            dur = end - start
+        # 情况2：end + duration 都有显式值
+        elif end > 0 and dur > 0:
+            start = end - dur
+            if start < 0:
+                start = 0.0
+                dur = end  # 重新计算
+        # 情况3：start + duration 都有值（start 可以是 0）
+        elif dur > 0:
+            end = start + dur
+        # 情况4：只有 start → 取到素材末尾
+        elif start > 0 and end <= 0 and dur <= 0:
+            end = asset_duration
+            dur = end - start
+        # 情况5：只有 end → 从开头取到 end
+        elif end > 0 and start <= 0 and dur <= 0:
+            start = 0.0
+            dur = end
+        else:
+            # 都没有 → 不裁剪
+            return TrimConfig(start_time=0.0, end_time=0.0, duration=0.0)
+
+        # 边界钳制：end 不能超过素材时长
+        if end > asset_duration:
+            end = asset_duration
+            dur = end - start
+
+        # 边界钳制：start 不能超过素材时长
+        if start >= asset_duration:
+            start = max(0.0, asset_duration - MIN_TRIM_DURATION)
+            dur = asset_duration - start
+            end = asset_duration
+
+        # 保证 duration 不为负
+        if dur < 0:
+            dur = 0.0
+
+        return TrimConfig(start_time=start, end_time=end, duration=dur)
+
+    @property
+    def is_valid(self) -> bool:
+        """裁剪是否有效（时长大于最小阈值）."""
+        return self.duration >= MIN_TRIM_DURATION
+
+    @property
+    def is_noop(self) -> bool:
+        """是否等价于不裁剪（从0开始取全部）."""
+        return self.start_time <= 0 and self.duration <= 0
+
+    @property
+    def trim_from_start(self) -> bool:
+        """是否从开头裁剪（start_time == 0）."""
+        return self.start_time <= 0
+
+
+@dataclass
+class TrimSegment:
+    """多段裁剪中的一段."""
+
+    segment_id: str  # 段 ID（用于生成唯一标签）
+    trim: TrimConfig  # 裁剪配置
+    order: int = 0  # 排序
+
+    @classmethod
+    def from_dict(cls, data: dict[str, Any], default_order: int = 0) -> TrimSegment:
+        """从字典构造."""
+        return cls(
+            segment_id=str(data.get("segment_id", "") or f"seg_{default_order}"),
+            trim=TrimConfig(
+                start_time=float(data.get("start_time", 0) or 0),
+                end_time=float(data.get("end_time", 0) or 0),
+                duration=float(data.get("duration", 0) or 0),
+            ),
+            order=int(data.get("order", default_order)),
+        )
+
+
+class TrimEngine:
+    """裁剪引擎 — 生成 FFmpeg trim / atrim 滤镜."""
+
+    @staticmethod
+    def build_video_trim_filter(
+        input_label: str,
+        trim: TrimConfig,
+        output_label: str,
+    ) -> str:
+        """构建视频裁剪滤镜链.
+
+        Args:
+            input_label: 输入视频标签，如 "[0:v]"
+            trim: 裁剪配置（已解析钳制）
+            output_label: 输出视频标签，如 "[v0_trimmed]"
+
+        Returns:
+            FFmpeg filter 字符串，如 "[0:v]trim=start=10:duration=5,setpts=PTS-STARTPTS[v0_trimmed]"
+        """
+        if trim.is_noop:
+            # 不裁剪，直接直通
+            return f"{input_label}copy{output_label}" if False else f"{input_label}setpts=PTS-STARTPTS{output_label}"
+
+        parts: list[str] = []
+
+        # trim 滤镜参数
+        trim_args: list[str] = []
+        if trim.start_time > 0:
+            trim_args.append(f"start={trim.start_time:.3f}")
+        if trim.duration > 0:
+            trim_args.append(f"duration={trim.duration:.3f}")
+        elif trim.end_time > 0:
+            # end 用 duration 表示（start 到 end 的时长）
+            # 但 validate_and_resolve 后应该已经有 duration 了
+            pass
+
+        parts.append(f"trim={':'.join(trim_args)}")
+        parts.append("setpts=PTS-STARTPTS")
+
+        filter_str = f"{input_label}{','.join(parts)}{output_label}"
+        return filter_str
+
+    @staticmethod
+    def build_audio_trim_filter(
+        input_label: str,
+        trim: TrimConfig,
+        output_label: str,
+    ) -> str:
+        """构建音频裁剪滤镜链.
+
+        Args:
+            input_label: 输入音频标签，如 "[0:a]"
+            trim: 裁剪配置（已解析钳制）
+            output_label: 输出音频标签，如 "[a0_trimmed]"
+
+        Returns:
+            FFmpeg filter 字符串，如 "[0:a]atrim=start=10:duration=5,asetpts=PTS-STARTPTS[a0_trimmed]"
+        """
+        if trim.is_noop:
+            return f"{input_label}asetpts=PTS-STARTPTS{output_label}"
+
+        parts: list[str] = []
+
+        trim_args: list[str] = []
+        if trim.start_time > 0:
+            trim_args.append(f"start={trim.start_time:.3f}")
+        if trim.duration > 0:
+            trim_args.append(f"duration={trim.duration:.3f}")
+
+        parts.append(f"atrim={':'.join(trim_args)}")
+        parts.append("asetpts=PTS-STARTPTS")
+
+        filter_str = f"{input_label}{','.join(parts)}{output_label}"
+        return filter_str
+
+    @staticmethod
+    def resolve_segments(
+        segments: list[TrimSegment],
+        asset_duration: float,
+    ) -> list[TrimSegment]:
+        """解析并钳制多段裁剪配置，过滤无效段.
+
+        Args:
+            segments: 原始段列表
+            asset_duration: 素材实际时长
+
+        Returns:
+            解析后的有效段列表，按 order 排序
+        """
+        resolved: list[TrimSegment] = []
+        for i, seg in enumerate(segments):
+            resolved_trim = seg.trim.validate_and_resolve(asset_duration)
+            if not resolved_trim.is_valid:
+                logger.warning("裁剪段无效，跳过: segment_id=%s duration=%.3f", seg.segment_id, resolved_trim.duration)
+                continue
+            resolved.append(
+                TrimSegment(
+                    segment_id=seg.segment_id,
+                    trim=resolved_trim,
+                    order=seg.order if seg.order >= 0 else i,
+                )
+            )
+
+        resolved.sort(key=lambda s: s.order)
+        return resolved
+
+    @staticmethod
+    def parse_segments_from_config(config: dict[str, Any] | None) -> list[TrimSegment]:
+        """从 clip config 中解析多段裁剪配置.
+
+        config 中支持：
+        - trim_segments: [ {segment_id, start_time, end_time, duration, order}, ... ]
+        - trim_start / trim_end / trim_duration: 单段裁剪（兼容旧格式）
+        """
+        if not config:
+            return []
+
+        # 优先解析多段
+        raw_segments = config.get("trim_segments", [])
+        if raw_segments and isinstance(raw_segments, list):
+            segments = []
+            for i, raw in enumerate(raw_segments):
+                if isinstance(raw, dict):
+                    segments.append(TrimSegment.from_dict(raw, default_order=i))
+            return segments
+
+        # 单段裁剪兼容：从 trim_start/trim_end/trim_duration 构造
+        has_single = any(k in config for k in ("trim_start", "trim_end", "trim_duration"))
+        if has_single:
+            seg = TrimSegment(
+                segment_id="main",
+                trim=TrimConfig(
+                    start_time=float(config.get("trim_start", 0) or 0),
+                    end_time=float(config.get("trim_end", 0) or 0),
+                    duration=float(config.get("trim_duration", 0) or 0),
+                ),
+                order=0,
+            )
+            return [seg]
+
+        return []
+
+
+# ── 工具函数 ──────────────────────────────────────────────────────────────────
+
+
+def extract_trim_from_clip_config(config: dict[str, Any] | None) -> TrimConfig | None:
+    """从 clip config 中提取单段裁剪配置.
+
+    兼容以下字段名：
+    - trim_start / trim_end / trim_duration
+    - start_time / end_time / duration（在 trim 子字典里）
+    """
+    if not config:
+        return None
+
+    # trim 子字典
+    if "trim" in config and isinstance(config["trim"], dict):
+        return TrimConfig.from_dict(config["trim"])
+
+    # 扁平字段
+    has_any = any(k in config for k in ("trim_start", "trim_end", "trim_duration"))
+    if not has_any:
+        return None
+
+    data = {
+        "start_time": config.get("trim_start", 0),
+        "end_time": config.get("trim_end", 0),
+        "duration": config.get("trim_duration", 0),
+    }
+    return TrimConfig.from_dict(data)

apps/worker/video_processing/tts_engine.py

@@ -0,0 +1,275 @@
+"""TTS 配音引擎 — 集成到统一渲染管道的配音能力.
+
+负责：
+- 根据 TtsConfig 生成配音音频
+- 字幕联动：按字幕片段分段合成，自动对齐时间轴
+- 整段配音：整段文本生成一条音频
+- 失败降级：TTS 失败不阻断渲染
+"""
+
+from __future__ import annotations
+
+import logging
+from dataclasses import dataclass, field
+from pathlib import Path
+from typing import Any
+
+from packages.domain.tts_config import TtsConfig
+from packages.ports.tts_service import TtsError, TtsService
+
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class VoiceoverSegment:
+    """配音片段.
+
+    Attributes:
+        text: 文本内容
+        start_time: 开始时间（秒）
+        end_time: 结束时间（秒）
+        audio_path: 合成后的音频文件路径
+        duration: 音频实际时长
+    """
+
+    text: str
+    start_time: float = 0.0
+    end_time: float = 0.0
+    audio_path: Path | None = None
+    duration: float = 0.0
+
+
+@dataclass
+class VoiceoverResult:
+    """配音结果.
+
+    Attributes:
+        success: 是否成功
+        segments: 配音片段列表
+        total_duration: 总时长
+        error_message: 错误信息（失败时）
+    """
+
+    success: bool = False
+    segments: list[VoiceoverSegment] = field(default_factory=list)
+    total_duration: float = 0.0
+    error_message: str = ""
+
+
+class TtsEngine:
+    """TTS 配音引擎.
+
+    封装 TtsService 调用，支持：
+    - 整段配音
+    - 字幕联动配音
+    - 失败降级
+    """
+
+    def __init__(
+        self,
+        tts_service: TtsService,
+        work_dir: Path,
+    ) -> None:
+        self._tts = tts_service
+        self._work_dir = work_dir
+        self._work_dir.mkdir(parents=True, exist_ok=True)
+
+    def generate_full_voiceover(
+        self,
+        config: TtsConfig,
+        *,
+        total_duration: float = 0.0,
+    ) -> VoiceoverResult:
+        """生成整段配音.
+
+        Args:
+            config: TTS 配置
+            total_duration: 视频总时长（用于调整配音速度适配）
+
+        Returns:
+            配音结果
+        """
+        if not config.enabled or not config.text.strip():
+            return VoiceoverResult(success=False, error_message="配音未启用或文本为空")
+
+        try:
+            output_path = self._work_dir / "voiceover_full.wav"
+
+            audio_path = self._tts.synthesize(
+                text=config.text,
+                voice_id=config.voice_id,
+                speed=config.speed,
+                pitch=config.pitch,
+                output_path=output_path,
+            )
+
+            # 探测实际时长
+            duration = self._probe_duration(audio_path)
+
+            segment = VoiceoverSegment(
+                text=config.text,
+                start_time=0.0,
+                end_time=duration,
+                audio_path=audio_path,
+                duration=duration,
+            )
+
+            return VoiceoverResult(
+                success=True,
+                segments=[segment],
+                total_duration=duration,
+            )
+
+        except TtsError as e:
+            logger.warning("TTS 整段配音失败，降级跳过: %s", e)
+            return VoiceoverResult(success=False, error_message=str(e))
+        except Exception as e:
+            logger.warning("TTS 整段配音异常，降级跳过: %s", e)
+            return VoiceoverResult(success=False, error_message=str(e))
+
+    def generate_subtitle_voiceover(
+        self,
+        config: TtsConfig,
+        subtitles: list[dict[str, Any]],
+    ) -> VoiceoverResult:
+        """根据字幕生成配音（字幕联动）.
+
+        每个字幕片段独立合成，按字幕时间轴对齐。
+
+        Args:
+            config: TTS 配置
+            subtitles: 字幕列表，每项含 text/start_time/end_time
+
+        Returns:
+            配音结果
+        """
+        if not config.enabled:
+            return VoiceoverResult(success=False, error_message="配音未启用")
+
+        if not subtitles:
+            return VoiceoverResult(success=False, error_message="字幕为空")
+
+        segments: list[VoiceoverSegment] = []
+        total_duration = 0.0
+
+        for i, sub in enumerate(subtitles):
+            text = sub.get("text", "").strip()
+            if not text:
+                continue
+
+            start_time = float(sub.get("start_time", 0))
+            end_time = float(sub.get("end_time", 0))
+            target_duration = max(0.1, end_time - start_time)
+
+            try:
+                # 计算适配时长所需语速：让配音时长 ≈ 字幕时长
+                estimated = self._tts.estimate_duration(text, speed=config.speed)
+                adjusted_speed = config.speed
+                if estimated > 0 and target_duration > 0:
+                    # 按目标时长调整语速，限制在 0.5~2.0 范围内
+                    speed_factor = estimated / target_duration
+                    adjusted_speed = max(0.5, min(2.0, config.speed * speed_factor))
+
+                output_path = self._work_dir / f"voiceover_seg_{i:03d}.wav"
+
+                audio_path = self._tts.synthesize(
+                    text=text,
+                    voice_id=config.voice_id,
+                    speed=adjusted_speed,
+                    pitch=config.pitch,
+                    output_path=output_path,
+                )
+
+                actual_duration = self._probe_duration(audio_path)
+
+                segment = VoiceoverSegment(
+                    text=text,
+                    start_time=start_time,
+                    end_time=start_time + actual_duration,
+                    audio_path=audio_path,
+                    duration=actual_duration,
+                )
+                segments.append(segment)
+                total_duration = max(total_duration, start_time + actual_duration)
+
+            except TtsError as e:
+                logger.warning("TTS 字幕片段 %d 合成失败，跳过: %s", i, e)
+                continue
+            except Exception as e:
+                logger.warning("TTS 字幕片段 %d 异常，跳过: %s", i, e)
+                continue
+
+        if not segments:
+            return VoiceoverResult(success=False, error_message="所有字幕片段合成失败")
+
+        return VoiceoverResult(
+            success=True,
+            segments=segments,
+            total_duration=total_duration,
+        )
+
+    def build_audio_mix_filter(
+        self,
+        result: VoiceoverResult,
+        *,
+        video_duration: float,
+        base_label: str = "0:a",
+    ) -> tuple[str, list[Path]]:
+        """构建配音混音滤镜.
+
+        将配音片段按时间轴排列，生成 amix 混入。
+
+        Args:
+            result: 配音结果
+            video_duration: 视频总时长
+            base_label: 基础音轨标签
+
+        Returns:
+            (filter_complex 字符串, 配音音频文件列表)
+        """
+        if not result.success or not result.segments:
+            return "", []
+
+        filter_parts: list[str] = []
+        audio_files: list[Path] = []
+        delay_labels: list[str] = []
+
+        for i, seg in enumerate(result.segments):
+            if seg.audio_path is None or not seg.audio_path.exists():
+                continue
+
+            audio_files.append(seg.audio_path)
+            seg_label = f"v{i}"
+
+            # 音量调整
+            # 用 adelay 延迟到字幕开始时间
+            delay_ms = int(max(0, int(seg.start_time * 1000)))
+            filter_parts.append(f"[{i}:a]adelay={delay_ms}:all=1,volume=0.8[{seg_label}]")
+            delay_labels.append(f"[{seg_label}]")
+
+        if not delay_labels:
+            return "", []
+
+        # 所有片段 concat 成一条配音音轨（用 amix 叠加多个延时后的片段
+        mix_inputs = "".join(delay_labels)
+        n_inputs = len(delay_labels)
+        tts_label = "tts_mixed"
+
+        if n_inputs == 1:
+            # 单个片段直接用
+            filter_parts.append(f"{delay_labels[0]}[{tts_label}]")
+        else:
+            # 多个片段 amix 叠加
+            filter_parts.append(f"{mix_inputs}amix=inputs={n_inputs}:duration=longest[{tts_label}]")
+
+        return ";".join(filter_parts), audio_files
+
+    def _probe_duration(self, audio_path: Path) -> float:
+        """探测音频时长."""
+        try:
+            from video_processing.ffmpeg_utils import probe_duration
+
+            return probe_duration(audio_path)
+        except Exception:
+            # 探测失败，按文件名估算
+            return 0.0

apps/worker/video_processing/watermark_engine.py

@@ -0,0 +1,315 @@
+"""水印引擎 — 基于 FFmpeg overlay 滤镜的水印叠加.
+
+支持：
+- 图片水印（PNG/logo）
+- 文字水印（drawtext）
+- 9宫格位置 + 边距配置
+- 透明度/大小缩放
+- 滚动水印（跑马灯）
+"""
+
+from __future__ import annotations
+
+import logging
+from dataclasses import dataclass
+from typing import Any
+
+logger = logging.getLogger(__name__)
+
+# 9宫格位置枚举
+WATERMARK_POSITIONS = {
+    "top_left": "左上",
+    "top_center": "中上",
+    "top_right": "右上",
+    "center_left": "左中",
+    "center": "中心",
+    "center_right": "右中",
+    "bottom_left": "左下",
+    "bottom_center": "中下",
+    "bottom_right": "右下",
+}
+
+
+@dataclass
+class WatermarkConfig:
+    """水印配置.
+
+    mode: "image" 图片水印 | "text" 文字水印
+    position: 9宫格位置
+    opacity: 透明度 0.0-1.0
+    scale: 缩放比例（图片水印），0.1-1.0
+    margin: 边距（像素）
+    scroll: 是否滚动（跑马灯）
+    scroll_speed: 滚动速度（像素/秒）
+    """
+
+    mode: str = "text"  # image | text
+    position: str = "bottom_right"
+
+    # 图片水印
+    image_path: str = ""  # 本地图片路径
+    scale: float = 0.2  # 相对输出宽度的比例
+    opacity: float = 0.8  # 0.0-1.0
+
+    # 文字水印
+    text: str = ""
+    font_size: int = 24
+    font_color: str = "white"
+    font_path: str = ""  # 字体文件路径
+
+    # 边距
+    margin_x: int = 20
+    margin_y: int = 20
+
+    # 滚动水印
+    scroll: bool = False
+    scroll_speed: int = 50  # 像素/秒
+
+    @classmethod
+    def from_dict(cls, data: dict[str, Any] | None) -> WatermarkConfig | None:
+        """从字典构造，空配置返回 None（不加水印）."""
+        if not data:
+            return None
+
+        enabled = data.get("enabled", False)
+        if not enabled:
+            return None
+
+        mode = data.get("mode", "text")
+
+        # 图片模式需要 image_path；文字模式需要 text
+        if mode == "image":
+            image_path = data.get("image_path", "") or data.get("image", "") or ""
+            if not image_path:
+                logger.warning("图片水印缺少 image_path，跳过水印")
+                return None
+        elif mode == "text":
+            text = data.get("text", "") or ""
+            if not text:
+                logger.warning("文字水印缺少 text，跳过水印")
+                return None
+
+        position = data.get("position", "bottom_right")
+        if position not in WATERMARK_POSITIONS:
+            position = "bottom_right"
+
+        return cls(
+            mode=mode,
+            position=position,
+            image_path=str(data.get("image_path", data.get("image", "")) or ""),
+            scale=float(data.get("scale", 0.2)),
+            opacity=float(data.get("opacity", 0.8)),
+            text=str(data.get("text", "") or ""),
+            font_size=int(data.get("font_size", 24)),
+            font_color=str(data.get("font_color", "white")),
+            font_path=str(data.get("font_path", "") or ""),
+            margin_x=int(data.get("margin_x", 20)),
+            margin_y=int(data.get("margin_y", 20)),
+            scroll=bool(data.get("scroll", False)),
+            scroll_speed=int(data.get("scroll_speed", 50)),
+        )
+
+    def validate(self) -> tuple[bool, str]:
+        """校验配置是否有效."""
+        if self.position not in WATERMARK_POSITIONS:
+            return False, f"不支持的位置: {self.position}"
+
+        if not (0.0 <= self.opacity <= 1.0):
+            return False, "透明度必须在 0-1 之间"
+
+        if self.mode == "image":
+            if not self.image_path:
+                return False, "图片水印缺少图片路径"
+            if not (0.01 <= self.scale <= 1.0):
+                return False, "缩放比例必须在 0.01-1.0 之间"
+        elif self.mode == "text":
+            if not self.text:
+                return False, "文字水印缺少文字内容"
+            if self.font_size <= 0:
+                return False, "字体大小必须大于 0"
+        else:
+            return False, f"不支持的水印模式: {self.mode}"
+
+        return True, ""
+
+
+class WatermarkEngine:
+    """水印引擎 — 生成 FFmpeg 水印滤镜."""
+
+    @staticmethod
+    def calc_position(
+        position: str,
+        output_width: int,
+        output_height: int,
+        wm_width: int,
+        wm_height: int,
+        margin_x: int,
+        margin_y: int,
+    ) -> tuple[int, int]:
+        """根据9宫格位置计算水印坐标 (x, y).
+
+        坐标系：左上角为 (0, 0)
+        """
+        if position == "top_left":
+            return margin_x, margin_y
+        elif position == "top_center":
+            return (output_width - wm_width) // 2, margin_y
+        elif position == "top_right":
+            return output_width - wm_width - margin_x, margin_y
+        elif position == "center_left":
+            return margin_x, (output_height - wm_height) // 2
+        elif position == "center":
+            return (output_width - wm_width) // 2, (output_height - wm_height) // 2
+        elif position == "center_right":
+            return output_width - wm_width - margin_x, (output_height - wm_height) // 2
+        elif position == "bottom_left":
+            return margin_x, output_height - wm_height - margin_y
+        elif position == "bottom_center":
+            return (output_width - wm_width) // 2, output_height - wm_height - margin_y
+        elif position == "bottom_right":
+            return output_width - wm_width - margin_x, output_height - wm_height - margin_y
+        else:
+            # 默认右下角
+            return output_width - wm_width - margin_x, output_height - wm_height - margin_y
+
+    @staticmethod
+    def calc_scroll_x(position: str, output_width: int, wm_width: int, speed: int) -> str:
+        """生成滚动水印的 x 坐标表达式.
+
+        从右向左滚动（跑马灯效果）
+        """
+        # x 从 W 到 -wm_width，整个宽度 + wm_width 的距离
+        # 使用 overlay 的 enable 表达式
+        # x = 'W - (t * speed)' → 不对，应该是持续滚动
+        # 标准跑马灯：x = -w + (t * speed) % (W + w)
+        # 但 FFmpeg overlay 支持表达式
+        return f"mod({output_width}-mod({speed}*t\\,{output_width}+{wm_width})"
+
+    @staticmethod
+    def build_image_watermark_filter(
+        input_video_label: str,
+        wm_image_path: str,
+        output_width: int,
+        output_height: int,
+        output_label: str,
+        config: WatermarkConfig,
+    ) -> tuple[str, list[str]]:
+        """构建图片水印滤镜链.
+
+        Args:
+            input_video_label: 输入视频标签，如 "[final_video]"
+            wm_image_path: 水印图片本地路径
+            output_width: 输出视频宽度
+            output_height: 输出视频高度
+            output_label: 输出标签
+            config: 水印配置
+
+        Returns:
+            (filter_complex_str, input_args_list)
+            input_args 是 ["-i", wm_image_path] 格式
+        """
+        # 计算水印尺寸（按输出宽度比例缩放）
+        wm_width = int(output_width * config.scale)
+        wm_height = -1  # 保持比例
+        wm_filter = f"scale={wm_width}:{wm_height}"
+
+        # 透明度处理
+        if config.opacity < 1.0:
+            wm_filter += f",format=rgba,colorchannelmixer=aa={config.opacity}"
+
+        # 水印预处理标签
+        wm_pre_label = "[wm_scaled]"
+
+        # 计算位置
+        x, y = WatermarkEngine.calc_position(
+            config.position,
+            output_width,
+            output_height,
+            wm_width,
+            wm_width,  # 高度未知，先用宽度估算
+            config.margin_x,
+            config.margin_y,
+        )
+
+        # 滚动水印
+        if config.scroll:
+            # 从右向左滚动：x = W - (t * speed) mod (W + wm_w)
+            # 使用 overlay 表达式
+            x_expr = f"{output_width}-mod({config.scroll_speed}*t\\,{output_width}+{wm_width}"
+            y_expr = str(y)
+            overlay_expr = f"x={x_expr}:y={y_expr}"
+        else:
+            overlay_expr = f"x={x}:y={y}"
+
+        # 构建滤镜
+        # 先缩放水印图
+        wm_input_idx = 1  # 假设水印图是第二个输入（索引1
+        filter_parts = [
+            f"[1:v]{wm_filter}{wm_pre_label}",
+            f"{input_video_label}{wm_pre_label}overlay={overlay_expr}{output_label}",
+        ]
+
+        filter_complex = ";".join(filter_parts)
+        input_args = ["-i", wm_image_path]
+
+        return filter_complex, input_args
+
+    @staticmethod
+    def build_text_watermark_filter(
+        input_video_label: str,
+        output_label: str,
+        config: WatermarkConfig,
+        output_width: int,
+        output_height: int,
+    ) -> str:
+        """构建文字水印滤镜（drawtext）.
+
+        Args:
+            input_video_label: 输入视频标签
+            output_label: 输出标签
+            config: 水印配置
+            output_width: 输出宽度
+            output_height: 输出高度
+
+        Returns:
+            FFmpeg filter 字符串
+        """
+        # 转义文字中的特殊字符
+        text = config.text.replace(":", "\\:").replace("'", "\\'")
+
+        # 字体配置
+        font_config = []
+        if config.font_path:
+            font_path_escaped = config.font_path.replace(":", "\\:").replace("'", "\\'")
+            font_config.append(f"fontfile='{font_path_escaped}'")
+        font_config.append(f"fontsize={config.font_size}")
+        font_config.append(f"fontcolor={config.font_color}@{config.opacity}")
+
+        # 估算文字宽高（粗略估算，用于位置计算）
+        # 每个汉字约等于 font_size 宽高
+        approx_w = len(config.text) * config.font_size
+        approx_h = config.font_size
+
+        # 位置计算
+        x, y = WatermarkEngine.calc_position(
+            config.position,
+            output_width,
+            output_height,
+            approx_w,
+            approx_h,
+            config.margin_x,
+            config.margin_y,
+        )
+
+        # 滚动水印
+        if config.scroll:
+            x_expr = f"w-mod({config.scroll_speed}*t\\,W+w)"
+            pos_config = [f"x={x_expr}", f"y={y}"]
+        else:
+            pos_config = [f"x={x}", f"y={y}"]
+
+        # 组装 drawtext
+        drawtext_parts = [f"text='{text}'"] + font_config + pos_config
+        drawtext = "drawtext=" + ":".join(drawtext_parts)
+
+        return f"{input_video_label}{drawtext}{output_label}"

apps/worker/worker_app/celery_app.py

@@ -16,5 +16,6 @@ celery_app.conf.imports = (
     "worker_app.tasks.tts_synthesis",
     "worker_app.tasks.edit_plan_generation",
     "worker_app.tasks.compose_video",
+    "worker_app.tasks.batch_download",
     "apps.worker.video_processing.dedup",
 )

apps/worker/worker_app/tasks/batch_download.py

@@ -0,0 +1,112 @@
+"""批量下载任务 — 将多个成片打包为 zip 上传到 OSS。"""
+
+from __future__ import annotations
+
+import logging
+import os
+import tempfile
+import uuid
+import zipfile
+from pathlib import Path
+
+from worker_app.celery_app import celery_app
+
+logger = logging.getLogger(__name__)
+
+
+@celery_app.task(bind=True, name="worker.batch_download_videos", max_retries=1)
+def batch_download_videos(self, video_ids: list[str], user_id: str = "") -> dict:
+    """批量下载视频并打包为 zip。
+
+    Args:
+        video_ids: 视频 ID 列表
+        user_id: 发起用户 ID
+
+    Returns:
+        {"download_url": "...", "file_count": N, "total_size": total_bytes}
+    """
+    from video_processing.oss_helpers import download_asset, upload_to_oss
+    from worker_app.db import SessionLocal
+
+    from packages.adapters.sqlalchemy_impl.generated_video_repository import (
+        SQLAlchemyGeneratedVideoRepository,
+    )
+
+    session = SessionLocal()
+    try:
+        repo = SQLAlchemyGeneratedVideoRepository(session)
+        videos = repo.get_by_ids(video_ids)
+    finally:
+        session.close()
+
+    if not videos:
+        raise ValueError("No videos found for batch download")
+
+    # 创建临时工作目录
+    with tempfile.TemporaryDirectory() as tmpdir:
+        tmpdir_path = Path(tmpdir)
+        zip_filename = f"videos-{len(videos)}-{video_ids[0][:8]}.zip"
+        zip_path = tmpdir_path / zip_filename
+
+        # 逐个下载视频并加入 zip
+        with zipfile.ZipFile(zip_path, "w", zipfile.ZIP_STORED) as zf:
+            for idx, video in enumerate(videos, 1):
+                logger.info("Batch download: downloading %d/%d %s", idx, len(videos), video.id)
+                try:
+                    # 下载视频到临时文件
+                    local_name = f"{idx:03d}_{video.name}"
+                    local_path = tmpdir_path / local_name
+
+                    # 使用 oss_helpers 的 download_asset，或者直接从 URL 下载
+                    if video.file_url:
+                        _download_video_to_file(video.file_url, str(local_path))
+
+                    if local_path.exists() and local_path.stat().st_size > 0:
+                        zf.write(str(local_path), arcname=local_name)
+                        local_path.unlink(missing_ok=True)
+                    else:
+                        logger.warning("Video %s download failed, skipping", video.id)
+                except Exception as e:
+                    logger.warning("Failed to download video %s: %s", video.id, e)
+                    continue
+
+        # 上传 zip 到 OSS
+        if not zip_path.exists() or zip_path.stat().st_size == 0:
+            raise RuntimeError("Batch download zip file is empty")
+        zip_storage_key = f"batch-downloads/{uuid.uuid4().hex}/{zip_filename}"
+        download_url = upload_to_oss(str(zip_path), zip_storage_key)
+
+        total_size = zip_path.stat().st_size
+        file_count = len(zipfile.ZipFile(str(zip_path), "r").namelist())
+
+        logger.info(
+            "Batch download complete: %d files, %d bytes, url=%s",
+            file_count,
+            total_size,
+            download_url,
+        )
+
+        return {
+            "download_url": download_url,
+            "file_count": file_count,
+            "total_size": total_size,
+            "video_count": len(videos),
+        }
+
+
+def _download_video_to_file(url: str, dest_path: str) -> None:
+    """下载视频文件到本地路径。优先用 OSS SDK 走内网，回退到 HTTP 下载。"""
+    from video_processing.oss_helpers import download_asset
+
+    try:
+        # 尝试走 OSS 下载（如果是 OSS URL 的话）
+        success = download_asset(url, dest_path)
+        if success:
+            return
+    except Exception:
+        pass
+
+    # 回退到 HTTP 下载
+    import urllib.request
+
+    urllib.request.urlretrieve(url, dest_path)  # nosec B310

apps/worker/worker_app/tasks/classification.py

@@ -68,6 +68,12 @@ def classify_asset(self, job_id: str) -> dict:
 
             # Update asset with classification status and result
             asset.classification_status = ClassificationStatus.COMPLETED
+            # 把分类结果写入 metadata，供列表筛选和智能视图使用
+            asset.metadata = {
+                **(asset.metadata or {}),
+                "classification": classification,
+                "classification_confidence": confidence,
+            }
             asset_repo.update(asset)
 
             session.commit()

apps/worker/worker_app/tasks/generation.py

@@ -86,6 +86,36 @@ def _update_task_status(task_id: str, status_action: str, **kwargs) -> bool:
         return False
 
 
+def _build_error_info(error: Exception, stage: str = "render") -> dict:
+    """构建结构化错误信息。
+
+    Args:
+        error: 异常对象
+        stage: 发生错误的阶段（download/render/merge/upload等）
+
+    Returns:
+        包含 error_type, message, stack_trace, stage, failed_at 的字典
+    """
+    import traceback
+    from datetime import datetime, timezone
+
+    tb_str = traceback.format_exc()
+    # 截取堆栈前20行，避免字段过大
+    tb_lines = tb_str.strip().splitlines()
+    if len(tb_lines) > 20:
+        tb_summary = "\n".join(tb_lines[:20]) + f"\n... (truncated, total {len(tb_lines)} lines)"
+    else:
+        tb_summary = tb_str
+
+    return {
+        "error_type": type(error).__name__,
+        "message": str(error),
+        "stack_trace": tb_summary,
+        "stage": stage,
+        "failed_at": datetime.now(timezone.utc).isoformat(),
+    }
+
+
 # ── 日志持久化辅助 ────────────────────────────────────────────────────────────
 
 
@@ -96,11 +126,7 @@ def _flush_logs(task_id: str, gen_task) -> None:
         try:
             from packages.adapters.sqlalchemy_impl.models import GenerationTaskModel
 
-            model = (
-                session.query(GenerationTaskModel)
-                .filter(GenerationTaskModel.id == task_id)
-                .first()
-            )
+            model = session.query(GenerationTaskModel).filter(GenerationTaskModel.id == task_id).first()
             if model:
                 model.logs = gen_task.logs
                 session.commit()
@@ -112,6 +138,7 @@ def _flush_logs(task_id: str, gen_task) -> None:
 
 # ── 共享工具模块导入 ──────────────────────────────────────────────────────────
 
+from services.asr_service_factory import get_asr_service
 from video_processing.dedup_helpers import create_video_record_and_dedup
 from video_processing.ffmpeg_utils import FFMPEG_BIN, probe_duration, run_ffmpeg
 from video_processing.oss_helpers import (
@@ -308,6 +335,86 @@ def _download_voice_asset(voice_library_id: str, local_path: Path) -> bool:
     return download_asset(storage_key, local_path)
 
 
+def _prepare_bgm_track(
+    *,
+    bgm_config: dict,
+    temp_path: Path,
+    task_id: str = "",
+) -> str | None:
+    """准备 BGM 音频文件（下载到本地）.
+
+    支持 3 种来源（按优先级）：
+    1. audio_url — 外部直链 URL（最高优先级）
+    2. asset_id — 素材库中的音频素材
+    3. preset_id — 预设 BGM 库
+
+    Returns:
+        BGM 本地文件路径，准备失败返回 None
+    """
+    from urllib.parse import urlparse
+
+    audio_url = bgm_config.get("audio_url", "") or ""
+    asset_id = bgm_config.get("asset_id", "") or ""
+    preset_id = bgm_config.get("preset_id", "") or ""
+
+    bgm_file = temp_path / f"bgm_{task_id or 'track'}.mp3"
+
+    # 优先级1：外部直链 URL
+    if audio_url:
+        try:
+            parsed = urlparse(audio_url)
+            if parsed.scheme in ("http", "https"):
+                import urllib.request
+
+                logger.info("[task_id=%s] [BGM] 从URL下载: %s", task_id, audio_url[:80])
+                urllib.request.urlretrieve(audio_url, bgm_file)  # nosec B310
+                if bgm_file.exists() and bgm_file.stat().st_size > 0:
+                    return str(bgm_file)
+        except Exception as e:
+            logger.warning("[task_id=%s] [BGM] URL下载失败: %s", task_id, e)
+
+    # 优先级2：素材库素材
+    if asset_id:
+        try:
+            from app.core.db import SessionLocal
+
+            from packages.adapters.sqlalchemy_impl.models import AssetModel
+
+            session = SessionLocal()
+            try:
+                model = session.query(AssetModel).filter(AssetModel.id == asset_id).first()
+                if model and model.file_url:
+                    storage_key = model.file_url
+                    logger.info("[task_id=%s] [BGM] 从素材库下载: asset_id=%s", task_id, asset_id)
+                    ok = download_asset(storage_key, bgm_file)
+                    if ok and bgm_file.exists() and bgm_file.stat().st_size > 0:
+                        return str(bgm_file)
+            finally:
+                session.close()
+        except Exception as e:
+            logger.warning("[task_id=%s] [BGM] 素材库下载失败: %s", task_id, e)
+
+    # 优先级3：预设 BGM 库
+    if preset_id:
+        try:
+            from packages.domain.preset_bgm import get_preset_bgm
+
+            preset = get_preset_bgm(preset_id)
+            if preset and preset.audio_url:
+                import urllib.request
+
+                logger.info("[task_id=%s] [BGM] 从预设库下载: preset_id=%s", task_id, preset_id)
+                urllib.request.urlretrieve(preset.audio_url, bgm_file)  # nosec B310
+                if bgm_file.exists() and bgm_file.stat().st_size > 0:
+                    return str(bgm_file)
+        except Exception as e:
+            logger.warning("[task_id=%s] [BGM] 预设库下载失败: %s", task_id, e)
+
+    # 所有来源都失败
+    logger.warning("[task_id=%s] [BGM] 所有来源都无法获取BGM，跳过", task_id)
+    return None
+
+
 def _verify_url_accessible(url: str, timeout: float = 10.0, retries: int = 2) -> bool:
     """HEAD 请求校验 URL 可访问（含重试，防止 OSS 抖动误报）。
 
@@ -413,9 +520,7 @@ def _download_library_assets(
             else:
                 # 未指定 asset_ids：按 library 或 project 下载全部 ready 视频
                 if asset_library_id:
-                    query = query.filter(
-                        AssetModel.asset_library_id == asset_library_id
-                    )
+                    query = query.filter(AssetModel.asset_library_id == asset_library_id)
                     logger.info(
                         "下载素材库全部视频: asset_library_id=%s",
                         asset_library_id,
@@ -430,11 +535,7 @@ def _download_library_assets(
             assets = query.order_by(AssetModel.created_at).all()
 
             if not assets:
-                mode_desc = (
-                    f"素材库 {asset_library_id}"
-                    if asset_library_id
-                    else f"项目 {project_id}"
-                )
+                mode_desc = f"素材库 {asset_library_id}" if asset_library_id else f"项目 {project_id}"
                 msg = f"未找到视频素材: {mode_desc}, asset_ids={asset_ids or 'all'}"
                 logger.error(msg)
                 raise RuntimeError(msg)
@@ -489,9 +590,7 @@ def _download_library_assets(
                             duration=0.0,
                         )
                     if strict:
-                        raise RuntimeError(
-                            f"素材缺少 file_url: asset_id={asset.id}, name={asset.name}"
-                        )
+                        raise RuntimeError(f"素材缺少 file_url: asset_id={asset.id}, name={asset.name}")
                     continue
 
                 ext = Path(storage_key).suffix or ".mp4"
@@ -541,9 +640,7 @@ def _download_library_assets(
                             duration=round(asset_elapsed, 2),
                         )
                     if strict:
-                        raise RuntimeError(
-                            f"素材下载失败: asset_id={asset.id}, name={asset.name}"
-                        )
+                        raise RuntimeError(f"素材下载失败: asset_id={asset.id}, name={asset.name}")
 
             # 指定了 asset_ids 但全部下载失败 → 无论 strict 与否都报错
             if asset_ids and not downloaded:
@@ -852,9 +949,7 @@ def _render_video(
 
     # 选择渲染引擎
     engine = _resolve_render_engine(user_id) if user_id else ENGINE_UNIFIED
-    logger.info(
-        "[task_id=%s] [渲染] 引擎选择: %s (user_id=%s)", task_id, engine, user_id
-    )
+    logger.info("[task_id=%s] [渲染] 引擎选择: %s (user_id=%s)", task_id, engine, user_id)
 
     render_start = time.monotonic()
     render_output_path = temp_path / f"rendered-{task_id}.mp4"
@@ -869,6 +964,22 @@ def _render_video(
         )
     else:
         logger.info("[task_id=%s] [渲染] unified 引擎 FFmpeg 渲染开始", task_id)
+
+        # ── 准备 BGM 音频 ──
+        bgm_path: str | None = None
+        plan_config = virtual_plan.config or {}
+        bgm_config = plan_config.get("bgm", {}) or {}
+        if bgm_config.get("enabled", False):
+            try:
+                bgm_path = _prepare_bgm_track(
+                    bgm_config=bgm_config,
+                    temp_path=temp_path,
+                    task_id=task_id,
+                )
+            except Exception as bgm_err:
+                logger.warning("[task_id=%s] [BGM] 准备失败，跳过BGM: %s", task_id, bgm_err)
+                bgm_path = None
+
         render_service = UnifiedRenderService(
             plan=virtual_plan,
             clips=virtual_clips,
@@ -877,6 +988,8 @@ def _render_video(
             output_width=OUTPUT_WIDTH,
             output_height=OUTPUT_HEIGHT,
             output_fps=int(OUTPUT_FPS),
+            asr_service=get_asr_service(),
+            bgm_path=bgm_path,
         )
         render_result = render_service.render()
         render_output_path = render_result.output_path
@@ -898,9 +1011,7 @@ def _render_video(
             _mux_audio_track(render_output_path, voice_path, final_path)
             output_path = final_path
         except Exception as mux_err:
-            logger.warning(
-                "[task_id=%s] [混音] 音频混合失败，使用无音频版本: %s", task_id, mux_err
-            )
+            logger.warning("[task_id=%s] [混音] 音频混合失败，使用无音频版本: %s", task_id, mux_err)
             output_path = render_output_path
     else:
         output_path = render_output_path
@@ -929,9 +1040,7 @@ def _upload_and_record(
     file_url = upload_to_oss(output_path, storage_key)
     upload_elapsed = time.monotonic() - upload_start
     if not file_url:
-        raise RuntimeError(
-            f"OSS 上传失败: task_id={task_id}, storage_key={storage_key}"
-        )
+        raise RuntimeError(f"OSS 上传失败: task_id={task_id}, storage_key={storage_key}")
 
     # 校验 URL 可达性（P0-2: 私有 bucket 用预签名 + object_exists 降级）
     verify_url = get_signed_download_url(file_url, expires_seconds=300) or file_url
@@ -942,8 +1051,7 @@ def _upload_and_record(
         key = normalize_storage_key(file_url)
         if not (bucket and bucket.object_exists(key)):
             raise RuntimeError(
-                f"OSS 上传后 URL 不可访问且 object_exists 失败: file_url={file_url}, "
-                f"storage_key={storage_key}"
+                f"OSS 上传后 URL 不可访问且 object_exists 失败: file_url={file_url}, " f"storage_key={storage_key}"
             )
         logger.info(
             "URL 校验失败但 object_exists 确认文件存在，视为上传成功: storage_key=%s",
@@ -1044,9 +1152,7 @@ def generate_video(self, task_id: str) -> dict:
     _update_task_status(task_id, "mark_processing")
 
     try:
-        editing_mode = (
-            EditingMode(mode) if (mode := task_info["mode"]) else EditingMode.ONE_TAKE
-        )
+        editing_mode = EditingMode(mode) if (mode := task_info["mode"]) else EditingMode.ONE_TAKE
     except ValueError:
         editing_mode = EditingMode.ONE_TAKE
 
@@ -1115,6 +1221,70 @@ def generate_video(self, task_id: str) -> dict:
             # ── 5. 标记完成 ──────────────────────────────────────────────────
             _update_task_status(task_id, "mark_completed", result_count=video_count)
 
+            # 5.1 更新标题使用次数
+            try:
+                _title_session = SessionLocal()
+                try:
+                    from packages.adapters.sqlalchemy_impl.generation_task_repository import (
+                        SQLAlchemyGenerationTaskRepository,
+                    )
+                    from packages.adapters.sqlalchemy_impl.title_library_repository import (
+                        SQLAlchemyTitleLibraryRepository,
+                    )
+
+                    _task_repo = SQLAlchemyGenerationTaskRepository(_title_session)
+                    _gen_task = _task_repo.get(task_id)
+                    if _gen_task and _gen_task.title_ids and _gen_task.created_by_user_id:
+                        _title_repo = SQLAlchemyTitleLibraryRepository(_title_session)
+                        for _tid in _gen_task.title_ids:
+                            try:
+                                _title_repo.increment_usage_count(_tid, _gen_task.created_by_user_id)
+                            except Exception:
+                                logger.warning(
+                                    "[task_id=%s] 更新标题使用次数失败: title_id=%s",
+                                    task_id,
+                                    _tid,
+                                    exc_info=True,
+                                )
+                finally:
+                    _title_session.close()
+            except Exception:
+                logger.warning("[task_id=%s] 更新标题使用次数异常（不影响主流程）", task_id, exc_info=True)
+
+            # 5.2 更新素材使用次数 + 最近使用时间
+            try:
+                from worker_app.core.asset_usage import mark_asset_used_for_generation
+
+                _asset_session = SessionLocal()
+                try:
+                    from packages.adapters.sqlalchemy_impl.asset_repository import (
+                        SQLAlchemyAssetRepository,
+                    )
+                    from packages.adapters.sqlalchemy_impl.generation_task_repository import (
+                        SQLAlchemyGenerationTaskRepository,
+                    )
+
+                    _task_repo = SQLAlchemyGenerationTaskRepository(_asset_session)
+                    _asset_repo = SQLAlchemyAssetRepository(_asset_session)
+                    _gen_task = _task_repo.get(task_id)
+                    if _gen_task and _gen_task.asset_ids:
+                        for _aid in _gen_task.asset_ids:
+                            try:
+                                _asset = _asset_repo.get(_aid)
+                                if _asset:
+                                    mark_asset_used_for_generation(_asset)
+                                    _asset_repo.update(_asset)
+                            except Exception:
+                                logger.warning(
+                                    "[task_id=%s] 更新素材使用次数失败: asset_id=%s",
+                                    task_id,
+                                    _aid,
+                                    exc_info=True,
+                                )
+                finally:
+                    _asset_session.close()
+            except Exception:
+                logger.warning("[task_id=%s] 更新素材使用次数异常（不影响主流程）", task_id, exc_info=True)
             if gen_task:
                 gen_task.append_log(
                     "任务完成",
@@ -1145,6 +1315,9 @@ def generate_video(self, task_id: str) -> dict:
     except Exception as error:
         logger.error("[task_id=%s] [任务失败] %s", task_id, error, exc_info=True)
 
+        # 构建结构化错误信息
+        error_info = _build_error_info(error, stage="render")
+
         # 记录失败日志
         try:
             _session = SessionLocal()
@@ -1157,6 +1330,7 @@ def generate_video(self, task_id: str) -> dict:
                         str(error),
                         level="ERROR",
                         error_type=type(error).__name__,
+                        stage="render",
                     )
                     _flush_logs(task_id, gen_task)
             finally:
@@ -1164,7 +1338,59 @@ def generate_video(self, task_id: str) -> dict:
         except Exception:
             logger.warning("[task_id=%s] 记录失败日志异常", task_id, exc_info=True)
 
-        _update_task_status(task_id, "mark_failed", error_message=str(error))
+        _update_task_status(
+            task_id,
+            "mark_failed",
+            error_message=str(error),
+            error_info=error_info,
+        )
+
+        # ── 自动重试逻辑 ──────────────────────────────────────────────────
+        try:
+            from packages.adapters.sqlalchemy_impl.generation_task_repository import (
+                SQLAlchemyGenerationTaskRepository,
+            )
+
+            _s = SessionLocal()
+            try:
+                _r = SQLAlchemyGenerationTaskRepository(_s)
+                _task = _r.get(task_id)
+                if _task and _task.auto_retry_enabled and _task.auto_retry_max > 0:
+                    current_retry = _task.retry_count or 0
+                    if current_retry < _task.auto_retry_max:
+                        logger.info(
+                            "[task_id=%s] 触发自动重试: 当前重试次数=%d, 最大重试次数=%d",
+                            task_id,
+                            current_retry,
+                            _task.auto_retry_max,
+                        )
+                        # 计算退避延迟（指数退避，基础5s，最大60s）
+                        backoff_seconds = min(5 * (2**current_retry), 60)
+                        # 原地重试
+                        _task.mark_pending_from_failed()
+                        _r.update(_task)
+                        # 延迟重新入队
+                        celery_app.send_task(
+                            "worker.generate_video",
+                            args=[task_id],
+                            countdown=backoff_seconds,
+                        )
+                        logger.info(
+                            "[task_id=%s] 自动重试已入队: 延迟=%ds, 第%d次重试",
+                            task_id,
+                            backoff_seconds,
+                            current_retry + 1,
+                        )
+            finally:
+                _s.close()
+        except Exception as retry_err:
+            logger.warning(
+                "[task_id=%s] 自动重试逻辑执行失败: %s",
+                task_id,
+                retry_err,
+                exc_info=True,
+            )
+
         return {
             "status": "failed",
             "task_id": task_id,

docs/schema-metadata-snapshot.json

@@ -858,6 +858,22 @@
           "type": "VARCHAR(20)",
           "unique": false
         },
+        {
+          "index": false,
+          "name": "transition_duration",
+          "nullable": false,
+          "primary_key": false,
+          "type": "FLOAT",
+          "unique": false
+        },
+        {
+          "index": false,
+          "name": "playback_speed",
+          "nullable": false,
+          "primary_key": false,
+          "type": "FLOAT",
+          "unique": false
+        },
         {
           "index": true,
           "name": "status",
@@ -1493,6 +1509,38 @@
           "type": "TEXT",
           "unique": false
         },
+        {
+          "index": false,
+          "name": "error_info",
+          "nullable": true,
+          "primary_key": false,
+          "type": "JSON",
+          "unique": false
+        },
+        {
+          "index": false,
+          "name": "retry_count",
+          "nullable": false,
+          "primary_key": false,
+          "type": "INTEGER",
+          "unique": false
+        },
+        {
+          "index": false,
+          "name": "auto_retry_enabled",
+          "nullable": false,
+          "primary_key": false,
+          "type": "BOOLEAN",
+          "unique": false
+        },
+        {
+          "index": false,
+          "name": "auto_retry_max",
+          "nullable": false,
+          "primary_key": false,
+          "type": "INTEGER",
+          "unique": false
+        },
         {
           "index": false,
           "name": "started_at",

packages/adapters/asr/mock_asr_service.py

@@ -0,0 +1,113 @@
+"""Mock ASR 服务 — 用于测试和开发环境。
+
+生成模拟的字幕时间轴，不依赖真实ASR服务。
+"""
+
+from __future__ import annotations
+
+import re
+from pathlib import Path
+from typing import Optional
+
+from packages.domain.subtitle import (
+    SubtitleSegment,
+    SubtitleTimeline,
+    SubtitleWord,
+)
+from packages.ports.asr_service import ASRService, ASRServiceError
+
+
+class MockASRService(ASRService):
+    """Mock ASR 服务，生成模拟字幕数据。
+
+    如果 audio_path 对应的目录下有同名 .txt 文件，
+    就读取该文件内容作为字幕文本，按时间均匀分段。
+    否则生成默认的测试字幕。
+    """
+
+    def __init__(self, mock_text: Optional[str] = None):
+        self._mock_text = mock_text
+
+    def transcribe(
+        self,
+        audio_path: Path,
+        language: Optional[str] = None,
+        with_word_timestamps: bool = True,
+    ) -> SubtitleTimeline:
+        if not audio_path.exists():
+            raise ASRServiceError(f"音频文件不存在: {audio_path}", provider="mock")
+
+        # 尝试读取同名 txt 文件作为字幕文本
+        text = self._mock_text
+        if text is None:
+            txt_path = audio_path.with_suffix(".txt")
+            if txt_path.exists():
+                text = txt_path.read_text(encoding="utf-8").strip()
+            else:
+                text = "这是一段测试字幕。它用于验证ASR自动字幕功能是否正常工作。每一句话都会被正确地分段并显示在视频底部。字幕的样式可以根据用户的喜好进行自定义调整。"
+
+        # 估算音频时长（用ffmpeg probe或者直接假设）
+        # mock模式下按字数估算，每秒4个字
+        total_duration = max(5.0, len(text) / 4.0)
+
+        segments = self._text_to_segments(text, total_duration, with_word_timestamps)
+
+        return SubtitleTimeline(
+            segments=segments,
+            language=language or "zh",
+            total_duration=total_duration,
+        )
+
+    def _text_to_segments(
+        self,
+        text: str,
+        total_duration: float,
+        with_word_timestamps: bool,
+    ) -> list[SubtitleSegment]:
+        """将文本按句切分成带时间轴的字幕片段。"""
+        # 按句末标点拆分
+        sentences = re.split(r"(?<=[。！？!?])", text)
+        sentences = [s.strip() for s in sentences if s.strip()]
+
+        if not sentences:
+            sentences = [text]
+
+        total_chars = sum(len(s) for s in sentences)
+        if total_chars == 0:
+            return []
+
+        segments = []
+        current_time = 0.0
+
+        for sentence in sentences:
+            char_count = len(sentence)
+            duration = total_duration * (char_count / total_chars)
+            end_time = current_time + duration
+
+            words: list[SubtitleWord] = []
+            if with_word_timestamps:
+                # 每个字作为一个词级单元（中文按字，英文按词）
+                word_time = current_time
+                word_duration = duration / char_count
+
+                for char in sentence:
+                    words.append(
+                        SubtitleWord(
+                            text=char,
+                            start=word_time,
+                            end=word_time + word_duration,
+                        )
+                    )
+                    word_time += word_duration
+
+            segments.append(
+                SubtitleSegment(
+                    text=sentence,
+                    start=current_time,
+                    end=end_time,
+                    words=words,
+                )
+            )
+            current_time = end_time
+
+        return segments

packages/adapters/in_memory/asset_repository.py

@@ -44,11 +44,61 @@ class InMemoryAssetRepository:
         return False
 
     def batch_delete(self, asset_ids: list[str]) -> int:
-        """批量删除素材，返回实际删除数量。"""
+        """批量删除素材（软删除，标记 status=deleted），返回实际影响数量。"""
+        from datetime import datetime, timezone
+
+        from packages.domain import AssetStatus
+
         count = 0
         for aid in asset_ids:
-            if aid in self._assets:
-                del self._assets[aid]
+            asset = self._assets.get(aid)
+            if asset and asset.status != AssetStatus.DELETED:
+                asset.status = AssetStatus.DELETED
+                asset.updated_at = datetime.now(timezone.utc)
+                count += 1
+        return count
+
+    def batch_update_metadata(self, asset_ids: list[str], metadata_patch: dict[str, object]) -> int:
+        """批量更新素材 metadata（合并 patch），返回实际影响数量。"""
+        from datetime import datetime, timezone
+
+        count = 0
+        for aid in asset_ids:
+            asset = self._assets.get(aid)
+            if asset:
+                asset.metadata = {**asset.metadata, **metadata_patch}
+                asset.updated_at = datetime.now(timezone.utc)
+                count += 1
+        return count
+
+    def batch_add_tags(self, asset_ids: list[str], tag_ids: list[str]) -> int:
+        """批量给素材添加标签（合并去重），返回实际影响数量。"""
+        from datetime import datetime, timezone
+
+        count = 0
+        for aid in asset_ids:
+            asset = self._assets.get(aid)
+            if asset:
+                changed = False
+                for tid in tag_ids:
+                    if tid not in asset.tag_ids:
+                        asset.tag_ids.append(tid)
+                        changed = True
+                if changed:
+                    asset.updated_at = datetime.now(timezone.utc)
+                    count += 1
+        return count
+
+    def batch_replace_tags(self, asset_ids: list[str], tag_ids: list[str]) -> int:
+        """批量替换素材标签（全量覆盖），返回实际影响数量。"""
+        from datetime import datetime, timezone
+
+        count = 0
+        for aid in asset_ids:
+            asset = self._assets.get(aid)
+            if asset:
+                asset.tag_ids = list(tag_ids)
+                asset.updated_at = datetime.now(timezone.utc)
                 count += 1
         return count
 

packages/adapters/sqlalchemy_impl/asset_repository.py

@@ -127,10 +127,90 @@ class SQLAlchemyAssetRepository:
         return False
 
     def batch_delete(self, asset_ids: list[str]) -> int:
-        """批量删除素材，返回实际删除数量。"""
+        """批量删除素材（软删除，标记 status=deleted），返回实际影响数量。"""
         if not asset_ids:
             return 0
-        count = self.session.query(AssetModel).filter(AssetModel.id.in_(asset_ids)).delete(synchronize_session=False)
+        from datetime import datetime, timezone
+
+        now = datetime.now(timezone.utc)
+        count = (
+            self.session.query(AssetModel)
+            .filter(AssetModel.id.in_(asset_ids), AssetModel.status != "deleted")
+            .update({AssetModel.status: "deleted", AssetModel.updated_at: now}, synchronize_session=False)
+        )
+        self.session.commit()
+        return count
+
+    def batch_update_metadata(self, asset_ids: list[str], metadata_patch: dict[str, object]) -> int:
+        """批量更新素材 metadata（合并 patch），返回实际影响数量。"""
+        if not asset_ids:
+            return 0
+        from datetime import datetime, timezone
+
+        now = datetime.now(timezone.utc)
+        # 逐条读取 + 合并 + 更新，保证 JSON 合并正确
+        models = self.session.query(AssetModel).filter(AssetModel.id.in_(asset_ids)).all()
+        count = 0
+        for model in models:
+            existing = {}
+            if model.classification_result:
+                try:
+                    existing = json.loads(model.classification_result)
+                except Exception:
+                    existing = {}
+            merged = {**existing, **metadata_patch}
+            model.classification_result = json.dumps(merged, ensure_ascii=False)
+            model.updated_at = now
+            count += 1
+        self.session.commit()
+        return count
+
+    def batch_add_tags(self, asset_ids: list[str], tag_ids: list[str]) -> int:
+        """批量给素材添加标签（合并去重），返回实际影响数量。"""
+        if not asset_ids or not tag_ids:
+            return 0
+        from datetime import datetime, timezone
+
+        now = datetime.now(timezone.utc)
+        clean_tag_ids = list(set(tag_ids))
+        count = 0
+        for aid in asset_ids:
+            # 查询现有标签
+            existing = {
+                row.tag_id
+                for row in self.session.query(AssetTagModel.tag_id).filter(AssetTagModel.asset_id == aid).all()
+            }
+            new_tags = [t for t in clean_tag_ids if t not in existing]
+            if new_tags:
+                for tid in new_tags:
+                    self.session.add(AssetTagModel(asset_id=aid, tag_id=tid))
+                # 更新 updated_at
+                self.session.query(AssetModel).filter(AssetModel.id == aid).update(
+                    {AssetModel.updated_at: now}, synchronize_session=False
+                )
+                count += 1
+        self.session.commit()
+        return count
+
+    def batch_replace_tags(self, asset_ids: list[str], tag_ids: list[str]) -> int:
+        """批量替换素材标签（全量覆盖），返回实际影响数量。"""
+        if not asset_ids:
+            return 0
+        from datetime import datetime, timezone
+
+        now = datetime.now(timezone.utc)
+        clean_tag_ids = list(set(tag_ids))
+        count = 0
+        for aid in asset_ids:
+            # 先删再加
+            self.session.query(AssetTagModel).filter(AssetTagModel.asset_id == aid).delete(synchronize_session=False)
+            for tid in clean_tag_ids:
+                self.session.add(AssetTagModel(asset_id=aid, tag_id=tid))
+            # 更新 updated_at
+            self.session.query(AssetModel).filter(AssetModel.id == aid).update(
+                {AssetModel.updated_at: now}, synchronize_session=False
+            )
+            count += 1
         self.session.commit()
         return count
 

packages/adapters/sqlalchemy_impl/edit_plan_clip_repository.py

@@ -54,6 +54,8 @@ class SQLAlchemyEditPlanClipRepository:
             start_time=clip.start_time,
             duration=clip.duration,
             transition_effect=clip.transition_effect,
+            transition_duration=clip.transition_duration,
+            playback_speed=clip.playback_speed,
             status=clip.status,
             config=clip.config,
         )
@@ -76,6 +78,8 @@ class SQLAlchemyEditPlanClipRepository:
         model.start_time = clip.start_time
         model.duration = clip.duration
         model.transition_effect = clip.transition_effect
+        model.transition_duration = clip.transition_duration
+        model.playback_speed = clip.playback_speed
         model.status = clip.status
         model.config = clip.config
         model.updated_at = clip.updated_at
@@ -120,6 +124,8 @@ class SQLAlchemyEditPlanClipRepository:
             start_time=model.start_time or 0.0,
             duration=model.duration or 0.0,
             transition_effect=model.transition_effect or "cut",
+            transition_duration=getattr(model, "transition_duration", 0.0) or 0.0,
+            playback_speed=model.playback_speed or 1.0,
             status=EditPlanClipStatus(model.status) if model.status else EditPlanClipStatus.PENDING,
             config=model.config or {},
             created_at=model.created_at,

packages/adapters/sqlalchemy_impl/generated_video_repository.py

@@ -100,6 +100,63 @@ class SQLAlchemyGeneratedVideoRepository:
         )
         return [self._to_domain(model) for model in models]
 
+    def list_paginated(
+        self,
+        *,
+        project_id: str | None = None,
+        status: str | None = None,
+        review_status: str | None = None,
+        page: int = 1,
+        page_size: int = 20,
+    ) -> tuple[list[GeneratedVideo], int]:
+        """分页查询成片列表，支持按项目、状态、复核状态筛选。"""
+        query = self.session.query(GeneratedVideoModel)
+
+        if project_id:
+            query = query.filter(GeneratedVideoModel.project_id == project_id)
+        if status:
+            query = query.filter(GeneratedVideoModel.status == status)
+        if review_status:
+            query = query.filter(GeneratedVideoModel.review_status == review_status)
+
+        total = query.count()
+
+        models = (
+            query.order_by(GeneratedVideoModel.generated_at.desc())
+            .offset((page - 1) * page_size)
+            .limit(page_size)
+            .all()
+        )
+
+        return [self._to_domain(model) for model in models], total
+
+    def update_review_status(self, video_id: str, review_status: str) -> GeneratedVideo | None:
+        """更新成片复核状态。"""
+        model = self.session.query(GeneratedVideoModel).filter(GeneratedVideoModel.id == video_id).first()
+        if model is None:
+            return None
+        model.review_status = review_status
+        self.session.add(model)
+        self.session.commit()
+        return self._to_domain(model)
+
+    def update_thumbnail(self, video_id: str, thumbnail_url: str) -> bool:
+        """更新成片封面图URL。"""
+        model = self.session.query(GeneratedVideoModel).filter(GeneratedVideoModel.id == video_id).first()
+        if model is None:
+            return False
+        model.thumbnail_url = thumbnail_url
+        self.session.add(model)
+        self.session.commit()
+        return True
+
+    def get_by_ids(self, video_ids: list[str]) -> list[GeneratedVideo]:
+        """批量获取成片记录。"""
+        if not video_ids:
+            return []
+        models = self.session.query(GeneratedVideoModel).filter(GeneratedVideoModel.id.in_(video_ids)).all()
+        return [self._to_domain(model) for model in models]
+
     @staticmethod
     def _to_domain(model: GeneratedVideoModel) -> GeneratedVideo:
         return GeneratedVideo(

packages/adapters/sqlalchemy_impl/generation_task_repository.py

@@ -21,6 +21,10 @@ def _to_domain(model: GenerationTaskModel) -> GenerationTask:
         progress=model.progress,
         result_count=int(model.result_count or 0),
         error_message=model.error_message,
+        error_info=dict(model.error_info) if model.error_info else {},
+        retry_count=model.retry_count or 0,
+        auto_retry_enabled=bool(model.auto_retry_enabled),
+        auto_retry_max=model.auto_retry_max or 0,
         started_at=model.started_at,
         completed_at=model.completed_at,
         created_by_user_id=model.created_by_user_id,
@@ -51,6 +55,10 @@ class SQLAlchemyGenerationTaskRepository:
             progress=task.progress,
             result_count=task.result_count,
             error_message=task.error_message,
+            error_info=task.error_info or None,
+            retry_count=task.retry_count or 0,
+            auto_retry_enabled=task.auto_retry_enabled,
+            auto_retry_max=task.auto_retry_max or 0,
             started_at=task.started_at,
             completed_at=task.completed_at,
             created_by_user_id=task.created_by_user_id,
@@ -127,6 +135,68 @@ class SQLAlchemyGenerationTaskRepository:
         )
         return [_to_domain(m) for m in models]
 
+    def list_by_user_filtered(
+        self,
+        user_id: str,
+        *,
+        status: str | None = None,
+        limit: int | None = None,
+        offset: int = 0,
+    ) -> list[GenerationTask]:
+        """按用户+状态筛选任务列表。"""
+        query = self.session.query(GenerationTaskModel).filter(GenerationTaskModel.created_by_user_id == user_id)
+        if status:
+            query = query.filter(GenerationTaskModel.status == status)
+        query = query.order_by(GenerationTaskModel.created_at.desc())
+        if offset:
+            query = query.offset(offset)
+        if limit:
+            query = query.limit(limit)
+        return [_to_domain(m) for m in query.all()]
+
+    def count_by_user_filtered(
+        self,
+        user_id: str,
+        *,
+        status: str | None = None,
+    ) -> int:
+        """按用户+状态筛选计数。"""
+        query = self.session.query(GenerationTaskModel).filter(GenerationTaskModel.created_by_user_id == user_id)
+        if status:
+            query = query.filter(GenerationTaskModel.status == status)
+        return query.count()
+
+    def list_by_project_filtered(
+        self,
+        project_id: str,
+        *,
+        status: str | None = None,
+        limit: int | None = None,
+        offset: int = 0,
+    ) -> list[GenerationTask]:
+        """按项目+状态筛选任务列表。"""
+        query = self.session.query(GenerationTaskModel).filter(GenerationTaskModel.project_id == project_id)
+        if status:
+            query = query.filter(GenerationTaskModel.status == status)
+        query = query.order_by(GenerationTaskModel.created_at.desc())
+        if offset:
+            query = query.offset(offset)
+        if limit:
+            query = query.limit(limit)
+        return [_to_domain(m) for m in query.all()]
+
+    def count_by_project_filtered(
+        self,
+        project_id: str,
+        *,
+        status: str | None = None,
+    ) -> int:
+        """按项目+状态筛选计数。"""
+        query = self.session.query(GenerationTaskModel).filter(GenerationTaskModel.project_id == project_id)
+        if status:
+            query = query.filter(GenerationTaskModel.status == status)
+        return query.count()
+
     def update(self, task: GenerationTask) -> GenerationTask:
         model = self.session.query(GenerationTaskModel).filter(GenerationTaskModel.id == task.id).first()
         if model is None:
@@ -143,6 +213,10 @@ class SQLAlchemyGenerationTaskRepository:
         model.progress = task.progress
         model.result_count = task.result_count
         model.error_message = task.error_message
+        model.error_info = task.error_info or None
+        model.retry_count = task.retry_count or 0
+        model.auto_retry_enabled = task.auto_retry_enabled
+        model.auto_retry_max = task.auto_retry_max or 0
         model.started_at = task.started_at
         model.completed_at = task.completed_at
         model.source_edit_plan_id = task.source_edit_plan_id or None

packages/adapters/sqlalchemy_impl/models.py

@@ -196,6 +196,8 @@ class EditPlanClipModel(Base):
     start_time = Column(Float, nullable=False, default=0.0)
     duration = Column(Float, nullable=False, default=0.0)
     transition_effect = Column(String(20), nullable=False, default="cut")
+    transition_duration = Column(Float, nullable=False, default=0.0)
+    playback_speed = Column(Float, nullable=False, default=1.0)
     status = Column(String(20), nullable=False, default="pending", index=True)
     config = Column(JSON, nullable=False, default=dict)
     created_at = Column(DateTime, nullable=False, default=lambda: datetime.now(timezone.utc))
@@ -250,6 +252,10 @@ class GenerationTaskModel(Base):
     progress = Column(Float, nullable=False, default=0.0)
     result_count = Column(Float, nullable=False, default=0)
     error_message = Column(Text, nullable=False, default="")
+    error_info = Column(JSON, nullable=True)
+    retry_count = Column(Integer, nullable=False, default=0)
+    auto_retry_enabled = Column(Boolean, nullable=False, default=False)
+    auto_retry_max = Column(Integer, nullable=False, default=0)
     started_at = Column(DateTime, nullable=True)
     completed_at = Column(DateTime, nullable=True)
     created_by_user_id = Column(String(36), nullable=False, default="", index=True)

packages/adapters/sqlalchemy_impl/template_repository.py

@@ -2,11 +2,14 @@
 
 from __future__ import annotations
 
+import uuid
 from typing import List, Optional
 
+from sqlalchemy import func
 from sqlalchemy.orm import Session
 
 from packages.adapters.sqlalchemy_impl.models import (
+    EditPlanModel,
     TemplateCategoryModel,
     TemplateModel,
     TemplateSegmentModel,
@@ -28,18 +31,26 @@ class SQLAlchemyTemplateRepository:
         *,
         skip: int = 0,
         limit: int = 50,
+        category: Optional[str] = None,
+        tag: Optional[str] = None,
+        keyword: Optional[str] = None,
+        mode: Optional[str] = None,
     ) -> List[Template]:
-        models = (
-            self.session.query(TemplateModel)
-            .filter(
-                TemplateModel.user_id == user_id,
-                TemplateModel.is_active.is_(True),
-            )
-            .order_by(TemplateModel.created_at.desc())
-            .offset(skip)
-            .limit(limit)
-            .all()
+        query = self.session.query(TemplateModel).filter(
+            TemplateModel.user_id == user_id,
+            TemplateModel.is_active.is_(True),
         )
+        if category:
+            query = query.filter(TemplateModel.category == category)
+        if mode:
+            query = query.filter(TemplateModel.mode == mode)
+        if keyword:
+            like_pattern = f"%{keyword}%"
+            query = query.filter(TemplateModel.name.like(like_pattern))
+        if tag:
+            # JSON 数组包含指定标签（MySQL JSON_CONTAINS / SQLite json_each 兼容写法用 LIKE）
+            query = query.filter(TemplateModel.tags.like(f'%"{tag}"%'))
+        models = query.order_by(TemplateModel.created_at.desc()).offset(skip).limit(limit).all()
         templates = [self._model_to_entity(m) for m in models]
         # 批量加载所有 segments，避免 N+1 查询
         if templates:
@@ -142,15 +153,77 @@ class SQLAlchemyTemplateRepository:
         self.session.commit()
         return True
 
-    def count_by_user(self, user_id: str) -> int:
-        return (
-            self.session.query(TemplateModel)
-            .filter(
-                TemplateModel.user_id == user_id,
-                TemplateModel.is_active.is_(True),
-            )
-            .count()
+    def count_by_user(
+        self,
+        user_id: str,
+        *,
+        category: Optional[str] = None,
+        tag: Optional[str] = None,
+        keyword: Optional[str] = None,
+        mode: Optional[str] = None,
+    ) -> int:
+        query = self.session.query(TemplateModel).filter(
+            TemplateModel.user_id == user_id,
+            TemplateModel.is_active.is_(True),
         )
+        if category:
+            query = query.filter(TemplateModel.category == category)
+        if mode:
+            query = query.filter(TemplateModel.mode == mode)
+        if keyword:
+            query = query.filter(TemplateModel.name.like(f"%{keyword}%"))
+        if tag:
+            query = query.filter(TemplateModel.tags.like(f'%"{tag}"%'))
+        return query.count()
+
+    def copy_template(self, template_id: str, user_id: str, new_name: str) -> Template:
+        """复制模板（含所有 segments）。"""
+        source = self.get(template_id, user_id)
+        if source is None:
+            raise ValueError(f"Template {template_id} not found")
+
+        new_id = str(uuid.uuid4())
+        new_template = Template(
+            id=new_id,
+            user_id=user_id,
+            name=new_name,
+            mode=source.mode,
+            category=source.category,
+            tags=list(source.tags),
+            title_config=dict(source.title_config),
+            subtitle_config=dict(source.subtitle_config),
+            bgm_config=dict(source.bgm_config),
+            estimated_duration=source.estimated_duration,
+            is_active=True,
+        )
+        created = self.create(new_template)
+
+        # 复制 segments
+        new_segments: List[TemplateSegment] = []
+        for seg in source.segments:
+            new_seg = TemplateSegment(
+                id=str(uuid.uuid4()),
+                template_id=new_id,
+                segment_order=seg.segment_order,
+                duration_min=seg.duration_min,
+                duration_max=seg.duration_max,
+                material_type=seg.material_type,
+            )
+            new_segments.append(new_seg)
+            model = TemplateSegmentModel(
+                id=new_seg.id,
+                template_id=new_seg.template_id,
+                segment_order=new_seg.segment_order,
+                duration_min=new_seg.duration_min,
+                duration_max=new_seg.duration_max,
+                material_type=new_seg.material_type,
+            )
+            self.session.add(model)
+        if new_segments:
+            self.session.commit()
+
+        created.segments = new_segments
+        return created
 
     # ── Segments ──
 
@@ -234,6 +307,33 @@ class SQLAlchemyTemplateRepository:
         self.session.commit()
         return True
 
+    # ── Tags ──
+
+    def list_tags(self, user_id: str) -> List[str]:
+        """获取用户所有模板的标签（去重）。"""
+        models = (
+            self.session.query(TemplateModel)
+            .filter(
+                TemplateModel.user_id == user_id,
+                TemplateModel.is_active.is_(True),
+                TemplateModel.tags.isnot(None),
+            )
+            .all()
+        )
+        tags_set: set[str] = set()
+        for m in models:
+            if m.tags:
+                for t in m.tags:
+                    if t:
+                        tags_set.add(t)
+        return sorted(tags_set)
+
+    # ── Usage Stats ──
+
+    def get_usage_count(self, template_id: str) -> int:
+        """获取模板被使用的次数（关联的剪辑计划数量）。"""
+        return self.session.query(EditPlanModel).filter(EditPlanModel.template_id == template_id).count()
+
     # ── Mapping helpers ──
 
     @staticmethod

packages/adapters/sqlalchemy_impl/title_library_repository.py

@@ -103,6 +103,25 @@ class SQLAlchemyTitleLibraryRepository:
         self.session.commit()
         return True
 
+    def increment_usage_count(self, title_id: str, user_id: str, increment: int = 1) -> bool:
+        """递增标题使用次数。返回是否成功。"""
+        from sqlalchemy import func
+
+        model = (
+            self.session.query(TitleLibraryModel)
+            .filter(
+                TitleLibraryModel.id == title_id,
+                TitleLibraryModel.user_id == user_id,
+            )
+            .first()
+        )
+        if model is None:
+            return False
+        model.usage_count = (model.usage_count or 0) + increment
+        model.updated_at = func.now()
+        self.session.commit()
+        return True
+
     def count_by_user(self, user_id: str, is_active: bool = True) -> int:
         return (
             self.session.query(TitleLibraryModel)

packages/adapters/tts/mock_tts_service.py

@@ -0,0 +1,231 @@
+"""Mock TTS 服务实现.
+
+使用 FFmpeg 合成简单音频模拟人声：
+- 不同音色用不同的基频（sine 波频率）
+- 语速通过 atempo 调整
+- 语调通过 asetrate 调整
+- 加一点 tremolo 效果让声音更自然
+
+用于开发测试，不依赖外部 TTS 服务。
+"""
+
+from __future__ import annotations
+
+import logging
+import subprocess
+import tempfile
+from pathlib import Path
+
+from packages.domain.voice_presets import get_voice, list_voices
+from packages.ports.tts_service import TtsError, TtsService
+
+logger = logging.getLogger(__name__)
+
+# Mock 时长估算：每字约 0.3 秒（中文）
+_CHARS_PER_SECOND = 3.3
+
+
+class MockTtsService(TtsService):
+    """Mock TTS 服务 — 用 FFmpeg 合成测试音频."""
+
+    def __init__(self, ffmpeg_bin: str = "ffmpeg") -> None:
+        self._ffmpeg_bin = ffmpeg_bin
+
+    @property
+    def provider_name(self) -> str:
+        return "mock"
+
+    def available_voices(self) -> list[str]:
+        return [v.voice_id for v in list_voices(provider="mock")]
+
+    def synthesize(
+        self,
+        text: str,
+        *,
+        voice_id: str = "",
+        speed: float = 1.0,
+        pitch: float = 0.0,
+        output_path: Path | None = None,
+        sample_rate: int = 22050,
+        format: str = "wav",
+    ) -> Path:
+        """合成 Mock 音频.
+
+        用 FFmpeg sine 波合成带轻微调制的音频，模拟人声。
+        时长根据文本长度估算。
+        """
+        if not text.strip():
+            raise TtsError("文本不能为空")
+
+        # 语速边界
+        if speed <= 0:
+            speed = 1.0
+        speed = max(0.5, min(2.0, speed))
+
+        # 语调边界
+        pitch = max(-12, min(12, pitch))
+
+        # 解析音色
+        voice = get_voice(voice_id) if voice_id else get_voice("female_warm")
+        if voice is None:
+            voice = get_voice("female_warm")
+
+        # 计算基频（从 provider_voice_id 里提取，或者按音色默认）
+        base_freq = self._extract_freq(voice.provider_voice_id, voice.gender.value)
+
+        # 计算时长（按文本长度）
+        duration = self.estimate_duration(text, speed=speed)
+        duration = max(0.5, duration)  # 最短 0.5 秒
+
+        # 输出路径
+        if output_path is None:
+            suffix = f".{format}"
+            tmp = tempfile.NamedTemporaryFile(suffix=suffix, delete=False)
+            tmp.close()
+            output_path = Path(tmp.name)
+
+        output_path.parent.mkdir(parents=True, exist_ok=True)
+
+        try:
+            self._synthesize_with_ffmpeg(
+                output_path=output_path,
+                base_freq=base_freq,
+                duration=duration,
+                speed=speed,
+                pitch=pitch,
+                sample_rate=sample_rate,
+                format=format,
+            )
+        except Exception as e:
+            logger.error("Mock TTS 合成失败: %s", e)
+            raise TtsError(f"Mock TTS 合成失败: {e}") from e
+
+        return output_path
+
+    def estimate_duration(self, text: str, *, speed: float = 1.0) -> float:
+        """估算音频时长.
+
+        按中文字符数估算：每字约 0.3 秒。
+        """
+        if not text:
+            return 0.0
+        # 去除空白后的字符数
+        char_count = len([c for c in text if not c.isspace()])
+        if char_count == 0:
+            return 0.0
+        base_duration = char_count / _CHARS_PER_SECOND
+        return base_duration / max(0.1, speed)
+
+    def _extract_freq(self, provider_voice_id: str, gender: str) -> float:
+        """从 provider_voice_id 提取基频，或按性别给默认值."""
+        if provider_voice_id.startswith("sine_"):
+            try:
+                return float(provider_voice_id.split("_")[1])
+            except (IndexError, ValueError):
+                pass
+
+        # 按性别给默认基频
+        if gender == "male":
+            return 120.0
+        elif gender == "child":
+            return 350.0
+        else:  # female
+            return 220.0
+
+    def _synthesize_with_ffmpeg(
+        self,
+        *,
+        output_path: Path,
+        base_freq: float,
+        duration: float,
+        speed: float,
+        pitch: float,
+        sample_rate: int,
+        format: str,
+    ) -> None:
+        """使用 FFmpeg 合成音频.
+
+        效果链：
+        1. sine 波生成基频
+        2. tremolo 增加轻微颤音
+        3. aeval 模拟简单的音色变化（让声音不那么单调）
+        4. atempo 调整语速
+        5. asetrate 调整语调
+        6. volume 调整音量
+        """
+        # 语调频率偏移因子（每半音 = 2^(1/12) ≈ 1.05946）
+        pitch_factor = 2 ** (pitch / 12)
+
+        # 颤音参数
+        tremolo_freq = 5.0  # 5Hz 颤音
+        tremolo_depth = 0.3  # 30% 深度
+
+        # 构建滤镜链
+        filters: list[str] = []
+
+        # 生成基频 + 泛音（让声音更丰富）
+        # 用多个 sine 波叠加模拟更自然的音色
+        filter_parts = []
+
+        # 主音 + 轻微频率调制
+        filter_parts.append(f"sine=frequency={base_freq}:duration={duration}:sample_rate={sample_rate}")
+
+        # 颤音效果
+        filter_parts.append(f"tremolo=f={tremolo_freq}:d={tremolo_depth}")
+
+        # 语速调整（同时调整时长）
+        if abs(speed - 1.0) > 0.01:
+            filter_parts.append(f"atempo={speed:.3f}")
+
+        # 语调调整（通过采样率变化实现，同时补偿时长）
+        if abs(pitch) > 0.01:
+            new_rate = int(sample_rate * pitch_factor)
+            filter_parts.append(f"asetrate={new_rate}")
+            filter_parts.append(f"aresample={sample_rate}")
+
+        # 音量包络：淡入淡出
+        fade_in = min(0.05, duration * 0.1)
+        fade_out = min(0.1, duration * 0.2)
+        filter_parts.append(f"afade=t=in:d={fade_in}")
+        filter_parts.append(f"afade=t=out:st={max(0, duration - fade_out)}:d={fade_out}")
+
+        # 音量调整到合适大小
+        filter_parts.append("volume=0.3")
+
+        filter_complex = ",".join(filter_parts)
+
+        # 编码参数
+        if format == "mp3":
+            codec_args = ["-acodec", "libmp3lame", "-b:a", "128k"]
+        else:
+            codec_args = ["-acodec", "pcm_s16le"]
+
+        command = [
+            self._ffmpeg_bin,
+            "-y",
+            "-f",
+            "lavfi",
+            "-i",
+            filter_complex,
+            *codec_args,
+            "-ar",
+            str(sample_rate),
+            "-ac",
+            "1",
+            str(output_path),
+        ]
+
+        logger.debug("Mock TTS FFmpeg 命令: %s", " ".join(command))
+
+        result = subprocess.run(
+            command,
+            capture_output=True,
+            text=True,
+            timeout=max(30, duration * 2 + 10),
+        )
+
+        if result.returncode != 0:
+            raise TtsError(f"FFmpeg 合成失败: {result.stderr[-500:]}")
+
+        if not output_path.exists() or output_path.stat().st_size == 0:
+            raise TtsError("输出文件为空或不存在")

packages/application/__init__.py

@@ -21,13 +21,19 @@ from .duplication import (
 from .generated_videos import (
     GetGeneratedVideoDownloadUrlUseCase,
     GetGeneratedVideoUseCase,
+    GetVideosByIdsUseCase,
     ListGeneratedVideosByTaskUseCase,
+    ListGeneratedVideosPaginatedUseCase,
     ListGeneratedVideosUseCase,
+    UpdateVideoReviewStatusUseCase,
 )
 from .generation_tasks import (
     CreateGenerationTaskCommand,
     CreateGenerationTaskUseCase,
     GetGenerationTaskUseCase,
+    ListGenerationTasksResult,
+    ListUserTasksFilteredUseCase,
+    RetryGenerationTaskUseCase,
 )
 from .ingest_jobs import SubmitIngestJobCommand, SubmitIngestJobUseCase
 from .jobs import (
@@ -65,6 +71,9 @@ __all__ = [
     "CreateGenerationTaskCommand",
     "CreateGenerationTaskUseCase",
     "GetGenerationTaskUseCase",
+    "ListGenerationTasksResult",
+    "ListUserTasksFilteredUseCase",
+    "RetryGenerationTaskUseCase",
     "CreateJobCommand",
     "CreateJobUseCase",
     "CreateProjectCommand",
@@ -76,6 +85,7 @@ __all__ = [
     "GetDuplicationDetailUseCase",
     "GetGeneratedVideoDownloadUrlUseCase",
     "GetGeneratedVideoUseCase",
+    "GetVideosByIdsUseCase",
     "GetJobStatisticsUseCase",
     "GetJobUseCase",
     "GetProjectUseCase",
@@ -83,6 +93,7 @@ __all__ = [
     "ListAssetsUseCase",
     "ListDuplicationRecordsUseCase",
     "ListGeneratedVideosByTaskUseCase",
+    "ListGeneratedVideosPaginatedUseCase",
     "ListGeneratedVideosUseCase",
     "ListJobsUseCase",
     "ListProjectsUseCase",
@@ -95,6 +106,7 @@ __all__ = [
     "SubmitJobUseCase",
     "UpdateJobProgressCommand",
     "UpdateJobProgressUseCase",
+    "UpdateVideoReviewStatusUseCase",
     "UploadForDuplicationCommand",
     "UploadForDuplicationUseCase",
 ]

packages/application/generated_videos.py

@@ -14,6 +14,32 @@ class ListGeneratedVideosUseCase:
         return self.generated_video_repository.list_by_project(project_id.strip())
 
 
+class ListGeneratedVideosPaginatedUseCase:
+    def __init__(self, generated_video_repository: GeneratedVideoRepository):
+        self.generated_video_repository = generated_video_repository
+
+    def execute(
+        self,
+        *,
+        project_id: str | None = None,
+        status: str | None = None,
+        review_status: str | None = None,
+        page: int = 1,
+        page_size: int = 20,
+    ) -> tuple[list[GeneratedVideo], int]:
+        if page < 1:
+            page = 1
+        if page_size < 1 or page_size > 100:
+            page_size = 20
+        return self.generated_video_repository.list_paginated(
+            project_id=project_id,
+            status=status,
+            review_status=review_status,
+            page=page,
+            page_size=page_size,
+        )
+
+
 class GetGeneratedVideoUseCase:
     def __init__(self, generated_video_repository: GeneratedVideoRepository):
         self.generated_video_repository = generated_video_repository
@@ -41,3 +67,23 @@ class GetGeneratedVideoDownloadUrlUseCase:
         if item is None:
             return None
         return item.file_url
+
+
+class UpdateVideoReviewStatusUseCase:
+    def __init__(self, generated_video_repository: GeneratedVideoRepository):
+        self.generated_video_repository = generated_video_repository
+
+    def execute(self, video_id: str, review_status: str) -> GeneratedVideo | None:
+        if not video_id.strip():
+            raise ValueError("video_id 不能为空")
+        if review_status not in ("pending_review", "approved", "rejected"):
+            raise ValueError(f"无效的 review_status: {review_status}")
+        return self.generated_video_repository.update_review_status(video_id.strip(), review_status)
+
+
+class GetVideosByIdsUseCase:
+    def __init__(self, generated_video_repository: GeneratedVideoRepository):
+        self.generated_video_repository = generated_video_repository
+
+    def execute(self, video_ids: list[str]) -> list[GeneratedVideo]:
+        return self.generated_video_repository.get_by_ids(video_ids)

packages/application/generation_tasks.py

@@ -21,6 +21,8 @@ class CreateGenerationTaskCommand:
     source_edit_plan_id: str = ""
     asset_select_mode: str = ""
     batch_id: str = ""
+    auto_retry_enabled: bool = False
+    auto_retry_max: int = 0
 
 
 class CreateGenerationTaskUseCase:
@@ -42,12 +44,12 @@ class CreateGenerationTaskUseCase:
             progress=0.0,
             result_count=0,
             error_message="",
-            started_at=None,
-            completed_at=None,
             created_by_user_id=command.created_by_user_id,
             source_edit_plan_id=command.source_edit_plan_id,
             asset_select_mode=command.asset_select_mode,
             batch_id=command.batch_id,
+            auto_retry_enabled=command.auto_retry_enabled,
+            auto_retry_max=command.auto_retry_max,
         )
         return self.generation_task_repository.create(task)
 
@@ -58,3 +60,65 @@ class GetGenerationTaskUseCase:
 
     def execute(self, task_id: str) -> GenerationTask | None:
         return self.generation_task_repository.get(task_id)
+
+
+@dataclass(slots=True)
+class ListTasksFilter:
+    """任务列表筛选条件。"""
+
+    status: str | None = None  # pending, running, completed, failed, cancelled
+
+
+@dataclass(slots=True)
+class ListGenerationTasksResult:
+    """带筛选和分页的任务列表结果。"""
+
+    items: list[GenerationTask]
+    total: int
+
+
+class ListUserTasksFilteredUseCase:
+    """按用户+筛选条件查询任务列表。"""
+
+    def __init__(self, generation_task_repository: GenerationTaskRepository):
+        self.generation_task_repository = generation_task_repository
+
+    def execute(
+        self,
+        user_id: str,
+        *,
+        status: str | None = None,
+        limit: int | None = None,
+        offset: int = 0,
+    ) -> ListGenerationTasksResult:
+        items = self.generation_task_repository.list_by_user_filtered(
+            user_id,
+            status=status,
+            limit=limit,
+            offset=offset,
+        )
+        total = self.generation_task_repository.count_by_user_filtered(
+            user_id,
+            status=status,
+        )
+        return ListGenerationTasksResult(items=items, total=total)
+
+
+class RetryGenerationTaskUseCase:
+    """原地重试失败的任务（重置状态+递增retry_count）。
+
+    与创建新任务不同：复用同一个 task_id，保留历史关联。
+    """
+
+    def __init__(self, generation_task_repository: GenerationTaskRepository):
+        self.generation_task_repository = generation_task_repository
+
+    def execute(self, task_id: str) -> GenerationTask:
+        task = self.generation_task_repository.get(task_id)
+        if task is None:
+            raise ValueError(f"任务不存在: {task_id}")
+        if not task.is_failed:
+            raise ValueError(f"只有失败状态的任务才能重试，当前状态: {task.status.value}")
+        task.mark_pending_from_failed()
+        self.generation_task_repository.update(task)
+        return task

packages/application/template/commands.py

@@ -49,6 +49,21 @@ class CreateCategoryCommand:
     name: str
 
 
+@dataclass
+class CopyTemplateCommand:
+    template_id: str
+    user_id: str
+    new_name: str
+
+
+@dataclass
+class ListTemplatesFilter:
+    category: Optional[str] = None
+    tag: Optional[str] = None
+    keyword: Optional[str] = None
+    mode: Optional[str] = None
+
+
 @dataclass
 class ValidateTemplateCommand:
     template_id: str

packages/application/template/use_cases.py

@@ -7,8 +7,10 @@ from dataclasses import dataclass, field
 from typing import List, Optional
 
 from packages.application.template.commands import (
+    CopyTemplateCommand,
     CreateCategoryCommand,
     CreateTemplateCommand,
+    ListTemplatesFilter,
     UpdateTemplateCommand,
     ValidateTemplateCommand,
 )
@@ -102,8 +104,40 @@ class ListTemplatesUseCase:
         *,
         skip: int = 0,
         limit: int = 50,
+        filter: Optional[ListTemplatesFilter] = None,
     ) -> List[Template]:
-        return self.repository.list_by_user(user_id, skip=skip, limit=limit)
+        if filter is None:
+            return self.repository.list_by_user(user_id, skip=skip, limit=limit)
+        return self.repository.list_by_user(
+            user_id,
+            skip=skip,
+            limit=limit,
+            category=filter.category,
+            tag=filter.tag,
+            keyword=filter.keyword,
+            mode=filter.mode,
+        )
+
+
+class CountTemplatesUseCase:
+    def __init__(self, repository: TemplateRepositoryPort) -> None:
+        self.repository = repository
+
+    def execute(
+        self,
+        user_id: str,
+        *,
+        filter: Optional[ListTemplatesFilter] = None,
+    ) -> int:
+        if filter is None:
+            return self.repository.count_by_user(user_id)
+        return self.repository.count_by_user(
+            user_id,
+            category=filter.category,
+            tag=filter.tag,
+            keyword=filter.keyword,
+            mode=filter.mode,
+        )
 
 
 class GetTemplateUseCase:
@@ -175,6 +209,23 @@ class DeleteTemplateUseCase:
         return self.repository.delete(template_id, user_id)
 
 
+class CopyTemplateUseCase:
+    def __init__(self, repository: TemplateRepositoryPort) -> None:
+        self.repository = repository
+
+    def execute(self, command: CopyTemplateCommand) -> Template:
+        existing = self.repository.get(command.template_id, command.user_id)
+        if existing is None:
+            raise NotFoundError(f"Template {command.template_id} not found")
+        if not command.new_name or not command.new_name.strip():
+            raise ValidationError("新模板名称不能为空")
+        return self.repository.copy_template(
+            command.template_id,
+            command.user_id,
+            command.new_name.strip(),
+        )
+
+
 # ── Validate template ──
 
 
@@ -258,3 +309,25 @@ class DeleteCategoryUseCase:
 
     def execute(self, category_id: str, user_id: str) -> bool:
         return self.repository.delete_category(category_id, user_id)
+
+
+# ── Tags ──
+
+
+class ListTagsUseCase:
+    def __init__(self, repository: TemplateRepositoryPort) -> None:
+        self.repository = repository
+
+    def execute(self, user_id: str) -> List[str]:
+        return self.repository.list_tags(user_id)
+
+
+# ── Usage Stats ──
+
+
+class GetTemplateUsageUseCase:
+    def __init__(self, repository: TemplateRepositoryPort) -> None:
+        self.repository = repository
+
+    def execute(self, template_id: str) -> int:
+        return self.repository.get_usage_count(template_id)

packages/application/title_library/__init__.py

@@ -1,11 +1,14 @@
 """Title library application module."""
 
+from packages.application.title_library.commands import IncrementTitleUsageCommand, PickTitleCommand
 from packages.application.title_library.use_cases import (
     CreateTitleLibraryUseCase,
     DeleteTitleLibraryUseCase,
     GetTitleLibraryUseCase,
+    IncrementTitleUsageUseCase,
     ListTitleLibraryUseCase,
     NotFoundError,
+    PickTitleUseCase,
     QuotaExceededError,
     UpdateTitleLibraryUseCase,
 )
@@ -14,7 +17,11 @@ __all__ = [
     "CreateTitleLibraryUseCase",
     "DeleteTitleLibraryUseCase",
     "GetTitleLibraryUseCase",
+    "IncrementTitleUsageUseCase",
+    "IncrementTitleUsageCommand",
     "ListTitleLibraryUseCase",
+    "PickTitleUseCase",
+    "PickTitleCommand",
     "UpdateTitleLibraryUseCase",
     "QuotaExceededError",
     "NotFoundError",

packages/application/title_library/commands.py

@@ -28,3 +28,17 @@ class UpdateTitleLibraryCommand:
     tags: Optional[List[str]] = None
     is_active: Optional[bool] = None
     metadata_: Optional[dict] = None
+
+
+@dataclass
+class IncrementTitleUsageCommand:
+    title_id: str
+    user_id: str
+    increment: int = 1
+
+
+@dataclass
+class PickTitleCommand:
+    user_id: str
+    category: Optional[str] = None
+    exclude_ids: List[str] = field(default_factory=list)

packages/application/title_library/use_cases.py

@@ -8,6 +8,8 @@ from typing import List, Optional
 from packages.adapters.sqlalchemy_impl.title_library_repository import SQLAlchemyTitleLibraryRepository
 from packages.application.title_library.commands import (
     CreateTitleLibraryCommand,
+    IncrementTitleUsageCommand,
+    PickTitleCommand,
     UpdateTitleLibraryCommand,
 )
 from packages.domain.quota import QuotaDimension, quota_checker
@@ -100,6 +102,68 @@ class DeleteTitleLibraryUseCase:
         return self.repository.delete(title_id, user_id)
 
 
+class IncrementTitleUsageUseCase:
+    """递增标题使用次数。用于生成视频成功后，更新标题的使用统计。"""
+
+    def __init__(self, repository: SQLAlchemyTitleLibraryRepository) -> None:
+        self.repository = repository
+
+    def execute(self, command: IncrementTitleUsageCommand) -> bool:
+        if command.increment <= 0:
+            return False
+        return self.repository.increment_usage_count(
+            command.title_id,
+            command.user_id,
+            increment=command.increment,
+        )
+
+
+class PickTitleUseCase:
+    """智能选择一个标题。
+
+    策略：
+    1. 可选按 category 过滤
+    2. 排除指定的 title_ids（如本轮已用过的）
+    3. 按使用次数升序，取最少的前 5 个
+    4. 从中随机选一个，增加多样性
+    5. 无可用标题时返回 None
+    """
+
+    _CANDIDATE_POOL_SIZE = 5
+
+    def __init__(self, repository: SQLAlchemyTitleLibraryRepository) -> None:
+        self.repository = repository
+
+    def execute(self, command: PickTitleCommand) -> TitleLibraryItem | None:
+        import random
+
+        # 取该用户所有活跃标题（或指定分类）
+        all_titles = self.repository.list_by_user(
+            command.user_id,
+            category=command.category,
+            is_active=True,
+            skip=0,
+            limit=500,  # 取足够多的候选
+        )
+
+        if not all_titles:
+            return None
+
+        # 排除已使用/指定排除的
+        exclude_set = set(command.exclude_ids or [])
+        candidates = [t for t in all_titles if t.id not in exclude_set]
+        if not candidates:
+            # 排除后没了，就从全部里选
+            candidates = all_titles
+
+        # 按使用次数升序，取最少的前 N 个
+        candidates.sort(key=lambda t: t.usage_count)
+        pool = candidates[: self._CANDIDATE_POOL_SIZE]
+
+        # 随机选一个
+        return random.choice(pool)
+
+
 class QuotaExceededError(Exception):
     def __init__(self, dimension: str, limit: float, used: float) -> None:
         self.dimension = dimension

packages/domain/config_schemas.py

@@ -112,14 +112,32 @@ class SubtitleConfig(BaseModel):
     color: str = Field(default="#ffffff", description="文字颜色 (HEX)")
     size: int = Field(default=24, ge=12, le=60, description="字号")
     animation: TextAnimation = Field(default=TextAnimation.FADE_IN, description="入场动画")
+    # ASR 自动字幕
+    auto_generated: bool = Field(default=False, description="是否启用ASR自动生成字幕")
+    language: str = Field(default="", description="字幕语言，空字符串表示自动检测（如 zh/en/ja）")
+    max_chars_per_line: int = Field(default=20, ge=8, le=40, description="每行最多字符数")
+    min_chars_per_segment: int = Field(default=8, ge=2, le=20, description="每段最少字符数（低于则合并）")
 
 
 class BGMConfig(BaseModel):
     """BGM 配置"""
 
+    enabled: bool = Field(default=False, description="是否启用 BGM")
     source: BGMSource = Field(default=BGMSource.LIBRARY, description="BGM 来源")
-    asset_id: str = Field(default="", description="BGM 素材 ID")
-    volume: float = Field(default=0.3, ge=0.0, le=1.0, description="音量 (0.0 ~ 1.0)")
+    asset_id: str = Field(default="", description="BGM 素材 ID（来源为 library/upload 时使用）")
+    preset_id: str = Field(default="", description="预设 BGM ID（来源为 ai_recommend 或使用内置库时使用）")
+    audio_url: str = Field(default="", description="BGM 音频 URL（外部直链，优先级最高）")
+    volume: float = Field(default=0.3, ge=0.0, le=1.0, description="BGM 音量 (0.0 ~ 1.0)")
+    fade_in: float = Field(default=0.0, ge=0.0, le=30.0, description="淡入时长（秒）")
+    fade_out: float = Field(default=0.0, ge=0.0, le=30.0, description="淡出时长（秒）")
+    loop_enabled: bool = Field(default=True, description="BGM 是否循环播放以铺满整个视频时长")
+    sidechain_enabled: bool = Field(default=False, description="是否启用人声闪避（有人声时 BGM 自动降低音量）")
+    sidechain_ratio: float = Field(
+        default=0.3, ge=0.0, le=1.0, description="人声闪避时 BGM 音量降低比例（0.3 = 降低30%）"
+    )
+    sidechain_attack: float = Field(default=0.02, ge=0.001, le=1.0, description="人声闪避攻击时间（秒）")
+    sidechain_release: float = Field(default=0.5, ge=0.01, le=5.0, description="人声闪避释放时间（秒）")
+    sidechain_threshold: float = Field(default=-25.0, ge=-60.0, le=0.0, description="人声闪避触发阈值（dB）")
 
 
 # ── 完整 config 模型 ─────────────────────────────────────────────────────────
@@ -185,9 +203,20 @@ DEFAULT_EDIT_PLAN_CONFIG: dict = {
         "animation": "fade_in",
     },
     "bgm": {
+        "enabled": False,
         "source": "library",
         "asset_id": "",
+        "preset_id": "",
+        "audio_url": "",
         "volume": 0.3,
+        "fade_in": 0.0,
+        "fade_out": 0.0,
+        "loop_enabled": True,
+        "sidechain_enabled": False,
+        "sidechain_ratio": 0.3,
+        "sidechain_attack": 0.02,
+        "sidechain_release": 0.5,
+        "sidechain_threshold": -25.0,
     },
     "editing_mode": "one_take",
 }

packages/domain/edit_plan_clip.py

@@ -50,6 +50,8 @@ class EditPlanClip:
     start_time: float = 0.0
     duration: float = 0.0
     transition_effect: str = "cut"
+    transition_duration: float = 0.0  # 0 表示使用全局默认值
+    playback_speed: float = 1.0  # 0 或 1.0 表示原速，范围 0.25~4.0
     status: EditPlanClipStatus = EditPlanClipStatus.PENDING
     config: dict[str, Any] = field(default_factory=dict)
     created_at: datetime = field(default_factory=lambda: datetime.now(timezone.utc))
@@ -68,6 +70,8 @@ class EditPlanClip:
         start_time: float = 0.0,
         duration: float = 0.0,
         transition_effect: str = "cut",
+        transition_duration: float = 0.0,
+        playback_speed: float = 1.0,
         config: dict[str, Any] | None = None,
     ) -> EditPlanClip:
         """创建剪辑计划片段"""
@@ -79,6 +83,13 @@ class EditPlanClip:
             raise ValueError("start_time 不能为负数")
         if duration < 0:
             raise ValueError("duration 不能为负数")
+        # 速度边界钳制
+        if playback_speed <= 0:
+            playback_speed = 1.0
+        elif playback_speed < 0.25:
+            playback_speed = 0.25
+        elif playback_speed > 4.0:
+            playback_speed = 4.0
 
         return cls(
             id=uuid4().hex,
@@ -91,6 +102,8 @@ class EditPlanClip:
             start_time=start_time,
             duration=duration,
             transition_effect=transition_effect.strip() or "cut",
+            transition_duration=max(0.0, transition_duration),
+            playback_speed=playback_speed,
             status=EditPlanClipStatus.PENDING,
             config=config or {},
         )

packages/domain/entities.py

@@ -133,6 +133,7 @@ class AssetStatus(StrEnum):
     READY = "ready"
     PROCESSING = "processing"
     ERROR = "error"
+    DELETED = "deleted"
 
     @classmethod
     def _missing_(cls, value: object) -> "AssetStatus":

packages/domain/generation_task.py

@@ -80,6 +80,10 @@ class GenerationTask:
     progress: float = 0.0
     result_count: int = 0
     error_message: str = ""
+    error_info: dict = field(default_factory=dict)
+    retry_count: int = 0
+    auto_retry_enabled: bool = False
+    auto_retry_max: int = 0
     started_at: datetime | None = None
     completed_at: datetime | None = None
     source_edit_plan_id: str = ""
@@ -105,6 +109,8 @@ class GenerationTask:
         source_edit_plan_id: str = "",
         asset_select_mode: str = "",
         batch_id: str = "",
+        auto_retry_enabled: bool = False,
+        auto_retry_max: int = 0,
     ) -> "GenerationTask":
         if not project_id.strip() and not template_id.strip():
             raise ValueError("project_id 或 template_id 至少需要提供一个")
@@ -124,6 +130,8 @@ class GenerationTask:
             source_edit_plan_id=source_edit_plan_id.strip(),
             asset_select_mode=asset_select_mode,
             batch_id=batch_id,
+            auto_retry_enabled=auto_retry_enabled,
+            auto_retry_max=auto_retry_max,
         )
 
     # ── 状态查询 ────────────────────────────────────────────────────────────
@@ -203,13 +211,14 @@ class GenerationTask:
         self.result_count = result_count
         self.error_message = ""
 
-    def mark_failed(self, error_message: str) -> None:
+    def mark_failed(self, error_message: str, error_info: dict | None = None) -> None:
         """标记为失败（pending / running → failed）。
 
-        设置 error_message、completed_at。
+        设置 error_message、error_info、completed_at。
 
         Args:
             error_message: 错误信息
+            error_info: 结构化错误信息（error_type, stack_trace, stage, failed_at等）
 
         Raises:
             ValueError: 当前状态不允许转换到 failed
@@ -217,6 +226,14 @@ class GenerationTask:
         self.transition_to(GenerationTaskStatus.FAILED)
         self.error_message = error_message
         self.completed_at = datetime.now(timezone.utc)
+        if error_info is not None:
+            self.error_info = error_info
+        else:
+            self.error_info = {
+                "error_type": "UnknownError",
+                "message": error_message,
+                "failed_at": datetime.now(timezone.utc).isoformat(),
+            }
 
     def mark_cancelled(self) -> None:
         """标记为已取消（pending / running → cancelled）。
@@ -269,7 +286,8 @@ class GenerationTask:
     def mark_pending_from_failed(self) -> None:
         """从失败状态重置为待处理（用于重试）。
 
-        清除 error_message、started_at、completed_at、progress。
+        清除 error_message、error_info、started_at、completed_at、progress，
+        递增 retry_count。
 
         Raises:
             ValueError: 当前状态不是 failed
@@ -278,7 +296,9 @@ class GenerationTask:
             raise ValueError(f"只有 failed 状态的任务可以重置为 pending，当前状态: {self.status.value}")
         self.transition_to(GenerationTaskStatus.PENDING)
         self.error_message = ""
+        self.error_info = {}
         self.started_at = None
         self.completed_at = None
         self.progress = 0.0
         self.result_count = 0
+        self.retry_count += 1

packages/domain/preset_bgm.py

@@ -0,0 +1,160 @@
+"""预设 BGM 库 — 免费可商用背景音乐清单.
+
+按风格分类，存储在 OSS 或 CDN 上。
+实际音频文件由运维统一上传，这里只维护元数据清单。
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+
+
+@dataclass(frozen=True)
+class PresetBGM:
+    """预设 BGM 条目"""
+
+    id: str
+    name: str
+    style: str  # 风格分类：upbeat/relax/tech/commerce/emotional/cinematic
+    duration: float  # 时长（秒）
+    artist: str = ""
+    description: str = ""
+    tags: list[str] = field(default_factory=list)
+    audio_url: str = ""  # CDN/OSS 地址，空字符串表示待部署
+
+
+# ── 预设库清单 ────────────────────────────────────────────────────────────────
+
+PRESET_BGM_LIBRARY: list[PresetBGM] = [
+    # 轻快 upbeat
+    PresetBGM(
+        id="bgm_upbeat_001",
+        name="阳光清晨",
+        style="upbeat",
+        duration=120.0,
+        artist="免费商用音乐库",
+        description="轻快明亮的吉他+钢琴，适合vlog、生活记录",
+        tags=["轻快", "阳光", "吉他", "vlog"],
+    ),
+    PresetBGM(
+        id="bgm_upbeat_002",
+        name="活力节拍",
+        style="upbeat",
+        duration=95.0,
+        artist="免费商用音乐库",
+        description="电子鼓点+合成器，节奏明快，适合运动、产品展示",
+        tags=["轻快", "电子", "活力", "运动"],
+    ),
+    PresetBGM(
+        id="bgm_upbeat_003",
+        name="夏日漫步",
+        style="upbeat",
+        duration=110.0,
+        artist="免费商用音乐库",
+        description="Ukulele+口哨，轻松愉悦，适合旅行、美食",
+        tags=["轻快", "夏日", "ukulele", "旅行"],
+    ),
+    # 治愈 relax
+    PresetBGM(
+        id="bgm_relax_001",
+        name="静谧时光",
+        style="relax",
+        duration=180.0,
+        artist="免费商用音乐库",
+        description="温柔钢琴独奏，治愈系，适合读书、冥想",
+        tags=["治愈", "钢琴", "安静", "冥想"],
+    ),
+    PresetBGM(
+        id="bgm_relax_002",
+        name="雨后森林",
+        style="relax",
+        duration=150.0,
+        artist="免费商用音乐库",
+        description="自然白噪音+轻柔吉他，放松减压",
+        tags=["治愈", "自然", "放松", "环境音"],
+    ),
+    PresetBGM(
+        id="bgm_relax_003",
+        name="月光奏鸣曲",
+        style="relax",
+        duration=200.0,
+        artist="古典音乐(公版)",
+        description="贝多芬经典钢琴作品，公版免费",
+        tags=["治愈", "古典", "钢琴", "优雅"],
+    ),
+    # 科技 tech
+    PresetBGM(
+        id="bgm_tech_001",
+        name="未来科技",
+        style="tech",
+        duration=85.0,
+        artist="免费商用音乐库",
+        description="电子合成器+科技鼓点，适合数码产品、科技解说",
+        tags=["科技", "电子", "未来感", "数码"],
+    ),
+    PresetBGM(
+        id="bgm_tech_002",
+        name="数据脉冲",
+        style="tech",
+        duration=100.0,
+        artist="免费商用音乐库",
+        description="极简电子节奏，适合数据分析、AI类视频",
+        tags=["科技", "极简", "数据", "AI"],
+    ),
+    # 电商 commerce
+    PresetBGM(
+        id="bgm_commerce_001",
+        name="心动时刻",
+        style="commerce",
+        duration=75.0,
+        artist="免费商用音乐库",
+        description="时尚动感节奏，适合商品展示、带货视频",
+        tags=["电商", "时尚", "动感", "带货"],
+    ),
+    PresetBGM(
+        id="bgm_commerce_002",
+        name="品质生活",
+        style="commerce",
+        duration=90.0,
+        artist="免费商用音乐库",
+        description="高级感轻音乐，适合品牌宣传、高端产品",
+        tags=["电商", "高端", "品牌", "品质"],
+    ),
+]
+
+# ── 风格分类字典 ──────────────────────────────────────────────────────────────
+
+BGM_STYLES: dict[str, str] = {
+    "upbeat": "轻快",
+    "relax": "治愈",
+    "tech": "科技",
+    "commerce": "电商",
+    "emotional": "情感",
+    "cinematic": "电影",
+}
+
+
+# ── 工具函数 ──────────────────────────────────────────────────────────────────
+
+
+def get_preset_bgm(bgm_id: str) -> PresetBGM | None:
+    """按 ID 获取预设 BGM。"""
+    for bgm in PRESET_BGM_LIBRARY:
+        if bgm.id == bgm_id:
+            return bgm
+    return None
+
+
+def list_preset_bgm_by_style(style: str) -> list[PresetBGM]:
+    """按风格筛选预设 BGM。"""
+    return [bgm for bgm in PRESET_BGM_LIBRARY if bgm.style == style]
+
+
+def search_preset_bgm(keyword: str) -> list[PresetBGM]:
+    """按关键词搜索预设 BGM（名称+标签+描述）。"""
+    kw = keyword.lower()
+    results = []
+    for bgm in PRESET_BGM_LIBRARY:
+        if kw in bgm.name.lower() or kw in bgm.description.lower() or any(kw in tag.lower() for tag in bgm.tags):
+            results.append(bgm)
+    return results

packages/domain/subtitle.py

@@ -0,0 +1,183 @@
+"""字幕领域模型 — 带时间轴的字幕片段。"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+from typing import List
+
+
+@dataclass
+class SubtitleWord:
+    """单个词级别的字幕单元，带精确时间戳。"""
+
+    text: str
+    start: float  # 秒
+    end: float  # 秒
+
+    @property
+    def duration(self) -> float:
+        return max(0.0, self.end - self.start)
+
+
+@dataclass
+class SubtitleSegment:
+    """一段字幕（一句话），带时间轴和词级信息。"""
+
+    text: str
+    start: float  # 秒
+    end: float  # 秒
+    words: List[SubtitleWord] = field(default_factory=list)
+
+    @property
+    def duration(self) -> float:
+        return max(0.0, self.end - self.start)
+
+    @property
+    def char_count(self) -> int:
+        return len(self.text)
+
+
+@dataclass
+class SubtitleTimeline:
+    """完整的字幕时间轴，由多个片段组成。"""
+
+    segments: List[SubtitleSegment] = field(default_factory=list)
+    language: str = "zh"  # zh / en / ja 等
+    total_duration: float = 0.0  # 音频总时长（秒）
+
+    @property
+    def segment_count(self) -> int:
+        return len(self.segments)
+
+    @property
+    def total_chars(self) -> int:
+        return sum(s.char_count for s in self.segments)
+
+    def merge_short_segments(self, min_chars: int = 8) -> SubtitleTimeline:
+        """合并过短的字幕片段，避免字幕跳动太快。"""
+        if len(self.segments) <= 1:
+            return self
+
+        merged: List[SubtitleSegment] = []
+        buffer: List[SubtitleSegment] = []
+
+        for seg in self.segments:
+            buffer.append(seg)
+            total_chars = sum(s.char_count for s in buffer)
+            if total_chars >= min_chars:
+                merged.append(self._merge_segments(buffer))
+                buffer = []
+
+        # 剩余的合并到最后一个或单独成段
+        if buffer:
+            if merged and sum(s.char_count for s in buffer) < min_chars:
+                # 太少了，合并到上一段
+                last = merged.pop()
+                merged.append(self._merge_segments([last] + buffer))
+            else:
+                merged.append(self._merge_segments(buffer))
+
+        return SubtitleTimeline(
+            segments=merged,
+            language=self.language,
+            total_duration=self.total_duration,
+        )
+
+    def split_long_segments(self, max_chars: int = 20) -> SubtitleTimeline:
+        """拆分过长的字幕片段，按语义断句。"""
+        new_segments: List[SubtitleSegment] = []
+
+        for seg in self.segments:
+            if seg.char_count <= max_chars:
+                new_segments.append(seg)
+                continue
+
+            # 按标点符号拆分
+            parts = self._split_text_by_punctuation(seg.text, max_chars)
+            if len(parts) == 1:
+                new_segments.append(seg)
+                continue
+
+            # 按字数比例分配时间
+            total_chars = seg.char_count
+            current_time = seg.start
+            word_idx = 0
+            all_words = seg.words.copy()
+
+            for part in parts:
+                part_chars = len(part)
+                part_duration = seg.duration * (part_chars / total_chars)
+                part_end = min(current_time + part_duration, seg.end)
+
+                # 收集对应时间段的词
+                part_words = []
+                while word_idx < len(all_words) and all_words[word_idx].start < part_end:
+                    part_words.append(all_words[word_idx])
+                    word_idx += 1
+
+                new_segments.append(
+                    SubtitleSegment(
+                        text=part,
+                        start=current_time,
+                        end=part_end,
+                        words=part_words,
+                    )
+                )
+                current_time = part_end
+
+        return SubtitleTimeline(
+            segments=new_segments,
+            language=self.language,
+            total_duration=self.total_duration,
+        )
+
+    @staticmethod
+    def _merge_segments(segments: List[SubtitleSegment]) -> SubtitleSegment:
+        if not segments:
+            return SubtitleSegment(text="", start=0, end=0)
+        return SubtitleSegment(
+            text="".join(s.text for s in segments),
+            start=segments[0].start,
+            end=segments[-1].end,
+            words=[w for s in segments for w in s.words],
+        )
+
+    @staticmethod
+    def _split_text_by_punctuation(text: str, max_chars: int) -> List[str]:
+        """按标点符号智能拆分长文本。"""
+        # 中文常见句末标点
+        sentence_end = "。！？!?"
+        clause_pause = "，；：,;:"
+
+        parts: List[str] = []
+        current = ""
+
+        for char in text:
+            current += char
+
+            if len(current) >= max_chars:
+                # 超过长度，找最近的标点断开
+                break_idx = -1
+                for i in range(len(current) - 1, -1, -1):
+                    if current[i] in sentence_end or current[i] in clause_pause:
+                        break_idx = i + 1
+                        break
+
+                if break_idx > 0:
+                    parts.append(current[:break_idx])
+                    current = current[break_idx:]
+                else:
+                    # 没有标点，硬切
+                    parts.append(current[:max_chars])
+                    current = current[max_chars:]
+
+            elif char in sentence_end:
+                # 句末标点，如果长度够就断开
+                if len(current) >= max_chars // 2:
+                    parts.append(current)
+                    current = ""
+
+        if current:
+            parts.append(current)
+
+        return parts

packages/domain/tts_config.py

@@ -0,0 +1,102 @@
+"""TTS 配音配置模型."""
+
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+from typing import Any, Optional
+
+
+@dataclass(slots=True)
+class TtsConfig:
+    """TTS 配音配置.
+
+    Attributes:
+        enabled: 是否启用配音
+        voice_id: 音色 ID
+        speed: 语速 (0.5 ~ 2.0)
+        pitch: 语调 (-12 ~ 12 半音)
+        volume: 音量 (0.0 ~ 1.0)
+        text: 配音文本（整段配音时使用）
+        align_mode: 对齐模式 - "subtitle"=按字幕对齐 / "full"=整段配音
+        overlap_mode: 与原音的叠加模式 - "replace"=替换 / "mix"=混音
+    """
+
+    enabled: bool = False
+    voice_id: str = ""
+    speed: float = 1.0
+    pitch: float = 0.0
+    volume: float = 0.8
+    text: str = ""
+    align_mode: str = "full"  # subtitle / full
+    overlap_mode: str = "replace"  # replace / mix
+
+    @classmethod
+    def parse(cls, data: Optional[dict[str, Any]]) -> "TtsConfig":
+        """从 dict 解析配置，无效值回退到默认."""
+        if not data or not isinstance(data, dict):
+            return cls()
+
+        enabled = data.get("enabled", False)
+        if not isinstance(enabled, bool):
+            enabled = False
+
+        if not enabled:
+            return cls(enabled=False)
+
+        voice_id = data.get("voice_id", "")
+        if not isinstance(voice_id, str):
+            voice_id = ""
+
+        speed = data.get("speed", 1.0)
+        if not isinstance(speed, (int, float)):
+            speed = 1.0
+
+        pitch = data.get("pitch", 0.0)
+        if not isinstance(pitch, (int, float)):
+            pitch = 0.0
+
+        volume = data.get("volume", 0.8)
+        if not isinstance(volume, (int, float)):
+            volume = 0.8
+
+        text = data.get("text", "")
+        if not isinstance(text, str):
+            text = ""
+
+        align_mode = data.get("align_mode", "full")
+        if align_mode not in ("subtitle", "full"):
+            align_mode = "full"
+
+        overlap_mode = data.get("overlap_mode", "replace")
+        if overlap_mode not in ("replace", "mix"):
+            overlap_mode = "replace"
+
+        config = cls(
+            enabled=enabled,
+            voice_id=voice_id,
+            speed=float(speed),
+            pitch=float(pitch),
+            volume=float(volume),
+            text=text,
+            align_mode=align_mode,
+            overlap_mode=overlap_mode,
+        )
+        config._clamp()
+        return config
+
+    def _clamp(self) -> None:
+        """边界钳制."""
+        if self.speed < 0.5:
+            self.speed = 0.5
+        elif self.speed > 2.0:
+            self.speed = 2.0
+
+        if self.pitch < -12:
+            self.pitch = -12
+        elif self.pitch > 12:
+            self.pitch = 12
+
+        if self.volume < 0.0:
+            self.volume = 0.0
+        elif self.volume > 1.0:
+            self.volume = 1.0

packages/domain/voice_presets.py

@@ -0,0 +1,222 @@
+"""配音引擎音色预设.
+
+与 CosyVoice 的 preset_voices 区分：
+- preset_voices.py: CosyVoice 真实音色（阿里云）
+- voice_presets.py: 配音引擎通用音色预设（含 mock/后续接入的真实 TTS）
+"""
+
+from __future__ import annotations
+
+import sys
+from dataclasses import dataclass
+
+if sys.version_info >= (3, 11):
+    from enum import StrEnum
+else:
+    from enum import Enum
+
+    class StrEnum(str, Enum):
+        pass
+
+
+class VoiceGender(StrEnum):
+    """音色性别."""
+
+    MALE = "male"
+    FEMALE = "female"
+    CHILD = "child"
+
+
+class VoiceStyle(StrEnum):
+    """音色风格."""
+
+    STABLE = "stable"  # 沉稳
+    LIVELY = "lively"  # 活泼
+    CUSTOMER_SERVICE = "customer_service"  # 客服
+    NARRATION = "narration"  # 旁白
+    NEWS = "news"  # 新闻
+    STORY = "story"  # 故事
+
+
+@dataclass(slots=True)
+class VoicePreset:
+    """音色预设.
+
+    Attributes:
+        voice_id: 音色唯一标识
+        name: 音色名称
+        gender: 性别
+        style: 风格
+        description: 描述
+        provider: 供应商（mock/aliyun/xunfei）
+        provider_voice_id: 供应商侧音色 ID
+        default_speed: 默认语速
+        default_pitch: 默认语调
+        sample_rate: 采样率
+        language: 语言
+    """
+
+    voice_id: str
+    name: str
+    gender: VoiceGender = VoiceGender.FEMALE
+    style: VoiceStyle = VoiceStyle.NARRATION
+    description: str = ""
+    provider: str = "mock"
+    provider_voice_id: str = ""
+    default_speed: float = 1.0
+    default_pitch: float = 0.0
+    sample_rate: int = 22050
+    language: str = "zh-CN"
+
+
+# ─── Mock 音色预设列表 ──────────────────────────────────────
+
+MOCK_VOICES: list[VoicePreset] = [
+    VoicePreset(
+        voice_id="female_warm",
+        name="温暖女声",
+        gender=VoiceGender.FEMALE,
+        style=VoiceStyle.NARRATION,
+        description="温柔温暖的女声，适合情感类、生活类视频",
+        provider="mock",
+        provider_voice_id="sine_220",
+        default_speed=1.0,
+        default_pitch=0.0,
+        sample_rate=22050,
+        language="zh-CN",
+    ),
+    VoicePreset(
+        voice_id="male_stable",
+        name="沉稳男声",
+        gender=VoiceGender.MALE,
+        style=VoiceStyle.STABLE,
+        description="沉稳厚重的男声，适合商务、知识类视频",
+        provider="mock",
+        provider_voice_id="sine_110",
+        default_speed=0.9,
+        default_pitch=0.0,
+        sample_rate=22050,
+        language="zh-CN",
+    ),
+    VoicePreset(
+        voice_id="female_lively",
+        name="活泼女声",
+        gender=VoiceGender.FEMALE,
+        style=VoiceStyle.LIVELY,
+        description="明亮活泼的女声，适合vlog、美食、旅行类视频",
+        provider="mock",
+        provider_voice_id="sine_280",
+        default_speed=1.2,
+        default_pitch=2.0,
+        sample_rate=22050,
+        language="zh-CN",
+    ),
+    VoicePreset(
+        voice_id="child_cute",
+        name="可爱童声",
+        gender=VoiceGender.CHILD,
+        style=VoiceStyle.STORY,
+        description="清脆可爱的童声，适合儿童教育、动画类视频",
+        provider="mock",
+        provider_voice_id="sine_380",
+        default_speed=1.0,
+        default_pitch=4.0,
+        sample_rate=22050,
+        language="zh-CN",
+    ),
+    VoicePreset(
+        voice_id="female_service",
+        name="客服女声",
+        gender=VoiceGender.FEMALE,
+        style=VoiceStyle.CUSTOMER_SERVICE,
+        description="专业清晰的客服女声，适合产品介绍、教程类视频",
+        provider="mock",
+        provider_voice_id="sine_250",
+        default_speed=1.0,
+        default_pitch=1.0,
+        sample_rate=22050,
+        language="zh-CN",
+    ),
+    VoicePreset(
+        voice_id="male_news",
+        name="新闻男声",
+        gender=VoiceGender.MALE,
+        style=VoiceStyle.NEWS,
+        description="字正腔圆的新闻播报声，适合资讯、时政类视频",
+        provider="mock",
+        provider_voice_id="sine_140",
+        default_speed=1.0,
+        default_pitch=0.0,
+        sample_rate=22050,
+        language="zh-CN",
+    ),
+    VoicePreset(
+        voice_id="female_soft",
+        name="轻柔女声",
+        gender=VoiceGender.FEMALE,
+        style=VoiceStyle.STORY,
+        description="轻柔舒缓的女声，适合睡前故事、冥想类视频",
+        provider="mock",
+        provider_voice_id="sine_180",
+        default_speed=0.8,
+        default_pitch=0.0,
+        sample_rate=22050,
+        language="zh-CN",
+    ),
+    VoicePreset(
+        voice_id="male_magnetic",
+        name="磁性男声",
+        gender=VoiceGender.MALE,
+        style=VoiceStyle.STORY,
+        description="低沉磁性的男声，适合电影解说、读书类视频",
+        provider="mock",
+        provider_voice_id="sine_90",
+        default_speed=0.85,
+        default_pitch=-2.0,
+        sample_rate=22050,
+        language="zh-CN",
+    ),
+]
+
+
+# voice_id → VoicePreset
+_MOCK_VOICE_MAP: dict[str, VoicePreset] = {v.voice_id: v for v in MOCK_VOICES}
+
+
+def get_voice(voice_id: str, *, provider: str = "mock") -> VoicePreset | None:
+    """根据 voice_id 获取音色预设."""
+    if provider == "mock":
+        return _MOCK_VOICE_MAP.get(voice_id)
+    return None
+
+
+def list_voices(
+    *,
+    gender: str | None = None,
+    style: str | None = None,
+    provider: str | None = None,
+    keyword: str | None = None,
+) -> list[VoicePreset]:
+    """按条件筛选音色列表."""
+    # 目前只有 mock 音色
+    result = list(MOCK_VOICES)
+
+    if provider and provider != "mock":
+        return []
+
+    if gender:
+        result = [v for v in result if v.gender.value == gender]
+
+    if style:
+        result = [v for v in result if v.style.value == style]
+
+    if keyword:
+        kw = keyword.lower()
+        result = [v for v in result if kw in v.name.lower() or kw in v.description.lower() or kw in v.voice_id.lower()]
+
+    return result
+
+
+def get_default_voice() -> VoicePreset:
+    """获取默认音色."""
+    return MOCK_VOICES[0]

packages/ports/asr_service.py

@@ -0,0 +1,47 @@
+"""ASR（语音识别）服务接口 — Port 层。"""
+
+from __future__ import annotations
+
+from abc import ABC, abstractmethod
+from pathlib import Path
+from typing import Optional
+
+from packages.domain.subtitle import SubtitleTimeline
+
+
+class ASRService(ABC):
+    """ASR 服务抽象接口。
+
+    不同的 ASR 后端（Whisper、阿里云、腾讯云等）实现此接口，
+    上层业务代码只依赖接口，不依赖具体实现。
+    """
+
+    @abstractmethod
+    def transcribe(
+        self,
+        audio_path: Path,
+        language: Optional[str] = None,
+        with_word_timestamps: bool = True,
+    ) -> SubtitleTimeline:
+        """将音频文件转写为带时间轴的字幕。
+
+        Args:
+            audio_path: 音频文件路径（支持 wav/mp3/m4a 等常见格式）
+            language: 指定语言代码（zh/en/ja 等），None 表示自动检测
+            with_word_timestamps: 是否返回词级时间戳
+
+        Returns:
+            SubtitleTimeline 字幕时间轴对象
+
+        Raises:
+            ASRServiceError: 识别服务调用失败
+        """
+        ...
+
+
+class ASRServiceError(Exception):
+    """ASR 服务调用异常。"""
+
+    def __init__(self, message: str, provider: str = "unknown"):
+        self.provider = provider
+        super().__init__(f"[{provider}] {message}")

packages/ports/asset_repository.py

@@ -52,7 +52,22 @@ class AssetRepository(ABC):
 
     @abstractmethod
     def batch_delete(self, asset_ids: list[str]) -> int:
-        """批量删除素材，返回实际删除数量。"""
+        """批量删除素材（软删除，标记 status=deleted），返回实际影响数量。"""
+        pass
+
+    @abstractmethod
+    def batch_update_metadata(self, asset_ids: list[str], metadata_patch: dict[str, object]) -> int:
+        """批量更新素材 metadata（合并 patch），返回实际影响数量。"""
+        pass
+
+    @abstractmethod
+    def batch_add_tags(self, asset_ids: list[str], tag_ids: list[str]) -> int:
+        """批量给素材添加标签（合并去重），返回实际影响数量。"""
+        pass
+
+    @abstractmethod
+    def batch_replace_tags(self, asset_ids: list[str], tag_ids: list[str]) -> int:
+        """批量替换素材标签（全量覆盖），返回实际影响数量。"""
         pass
 
     @abstractmethod

packages/ports/generated_video_repository.py

@@ -15,3 +15,19 @@ class GeneratedVideoRepository(Protocol):
     def list_by_generation_task(self, generation_task_id: str) -> list[GeneratedVideo]: ...
 
     def list_by_batch(self, batch_id: str) -> list[GeneratedVideo]: ...
+
+    def list_paginated(
+        self,
+        *,
+        project_id: str | None = None,
+        status: str | None = None,
+        review_status: str | None = None,
+        page: int = 1,
+        page_size: int = 20,
+    ) -> tuple[list[GeneratedVideo], int]: ...
+
+    def update_review_status(self, video_id: str, review_status: str) -> GeneratedVideo | None: ...
+
+    def update_thumbnail(self, video_id: str, thumbnail_url: str) -> bool: ...
+
+    def get_by_ids(self, video_ids: list[str]) -> list[GeneratedVideo]: ...

packages/ports/generation_task_repository.py

@@ -24,4 +24,36 @@ class GenerationTaskRepository(Protocol):
 
     def list_by_source_edit_plan(self, plan_id: str) -> list[GenerationTask]: ...
 
+    def list_by_user_filtered(
+        self,
+        user_id: str,
+        *,
+        status: str | None = None,
+        limit: int | None = None,
+        offset: int = 0,
+    ) -> list[GenerationTask]: ...
+
+    def count_by_user_filtered(
+        self,
+        user_id: str,
+        *,
+        status: str | None = None,
+    ) -> int: ...
+
+    def list_by_project_filtered(
+        self,
+        project_id: str,
+        *,
+        status: str | None = None,
+        limit: int | None = None,
+        offset: int = 0,
+    ) -> list[GenerationTask]: ...
+
+    def count_by_project_filtered(
+        self,
+        project_id: str,
+        *,
+        status: str | None = None,
+    ) -> int: ...
+
     def update(self, task: GenerationTask) -> GenerationTask: ...

packages/ports/template_repository.py

@@ -8,12 +8,31 @@ from packages.domain.template import Template, TemplateCategory, TemplateSegment
 
 
 class TemplateRepositoryPort(Protocol):
-    def list_by_user(self, user_id: str, *, skip: int = 0, limit: int = 50) -> List[Template]: ...
+    def list_by_user(
+        self,
+        user_id: str,
+        *,
+        skip: int = 0,
+        limit: int = 50,
+        category: Optional[str] = None,
+        tag: Optional[str] = None,
+        keyword: Optional[str] = None,
+        mode: Optional[str] = None,
+    ) -> List[Template]: ...
     def get(self, template_id: str, user_id: str) -> Optional[Template]: ...
     def create(self, template: Template) -> Template: ...
     def update(self, template: Template) -> Template: ...
     def delete(self, template_id: str, user_id: str) -> bool: ...
-    def count_by_user(self, user_id: str) -> int: ...
+    def count_by_user(
+        self,
+        user_id: str,
+        *,
+        category: Optional[str] = None,
+        tag: Optional[str] = None,
+        keyword: Optional[str] = None,
+        mode: Optional[str] = None,
+    ) -> int: ...
+    def copy_template(self, template_id: str, user_id: str, new_name: str) -> Template: ...
     def list_segments(self, template_id: str) -> List[TemplateSegment]: ...
     def create_segments(self, segments: List[TemplateSegment]) -> List[TemplateSegment]: ...
     def delete_segments_by_template(self, template_id: str) -> int: ...
@@ -21,3 +40,5 @@ class TemplateRepositoryPort(Protocol):
     def create_category(self, category: TemplateCategory) -> TemplateCategory: ...
     def get_category(self, category_id: str, user_id: str) -> Optional[TemplateCategory]: ...
     def delete_category(self, category_id: str, user_id: str) -> bool: ...
+    def list_tags(self, user_id: str) -> List[str]: ...
+    def get_usage_count(self, template_id: str) -> int: ...

packages/ports/tts_service.py

@@ -0,0 +1,78 @@
+"""TTS 服务抽象接口 (Port).
+
+新增 TTS 供应商时，实现本接口即可。
+"""
+
+from __future__ import annotations
+
+from abc import ABC, abstractmethod
+from pathlib import Path
+
+
+class TtsService(ABC):
+    """TTS 服务抽象基类.
+
+    所有 TTS 供应商（Mock / 阿里云 / 讯飞 等）都需要实现本接口。
+    """
+
+    @abstractmethod
+    def synthesize(
+        self,
+        text: str,
+        *,
+        voice_id: str = "",
+        speed: float = 1.0,
+        pitch: float = 0.0,
+        output_path: Path | None = None,
+        sample_rate: int = 22050,
+        format: str = "wav",
+    ) -> Path:
+        """文本转语音合成.
+
+        Args:
+            text: 输入文本
+            voice_id: 音色 ID
+            speed: 语速 (0.5 ~ 2.0)
+            pitch: 语调（半音，-12 ~ 12）
+            output_path: 输出文件路径（None 则自动生成）
+            sample_rate: 采样率
+            format: 输出格式 (wav/mp3)
+
+        Returns:
+            输出音频文件路径
+
+        Raises:
+            TtsError: 合成失败
+        """
+        ...
+
+    @abstractmethod
+    def estimate_duration(self, text: str, *, speed: float = 1.0) -> float:
+        """预估音频时长（秒）.
+
+        用于在实际合成前估算时长，方便时间轴对齐。
+
+        Args:
+            text: 输入文本
+            speed: 语速
+
+        Returns:
+            预估时长（秒）
+        """
+        ...
+
+    @property
+    @abstractmethod
+    def provider_name(self) -> str:
+        """供应商名称."""
+        ...
+
+    def available_voices(self) -> list[str]:
+        """支持的音色 ID 列表."""
+        return []
+
+
+class TtsError(Exception):
+    """TTS 合成异常."""
+
+    pass

scripts/build_release_images.sh

@@ -63,8 +63,7 @@ BRANCH_NAME="${GITHUB_REF_NAME:-${CI_COMMIT_BRANCH:-unknown}}"
 if [ "$USE_CACHE" -eq 1 ]; then
   docker buildx build \
     --build-arg APP_VERSION="$VERSION" \
-    --cache-from "type=registry,ref=${CACHE_REGISTRY}/api-cache:${CACHE_TAG},ignore-error=true" \
-    --cache-to "type=registry,ref=${CACHE_REGISTRY}/api-cache:${CACHE_TAG},mode=max" \
+    --cache-from "type=registry,ref=${CACHE_REGISTRY}/api-cache:${CACHE_TAG_PRIMARY},ignore-error=true" \
     -f infra/docker/api.Dockerfile \
     -t "$API_IMAGE" -t "$API_LATEST" \
     --load \
@@ -123,8 +122,7 @@ echo "=== Building Worker image ==="
 if [ "$USE_CACHE" -eq 1 ]; then
   docker buildx build \
     --build-arg APP_VERSION="$VERSION" \
-    --cache-from "type=registry,ref=${CACHE_REGISTRY}/worker-cache:${CACHE_TAG},ignore-error=true" \
-    --cache-to "type=registry,ref=${CACHE_REGISTRY}/worker-cache:${CACHE_TAG},mode=max" \
+    --cache-from "type=registry,ref=${CACHE_REGISTRY}/worker-cache:${CACHE_TAG_PRIMARY},ignore-error=true" \
     -f infra/docker/worker.Dockerfile \
     -t "$WORKER_IMAGE" -t "$WORKER_LATEST" \
     --load \
@@ -152,8 +150,7 @@ test -f apps/web/dist/index.html
 
 if [ "$USE_CACHE" -eq 1 ]; then
   docker buildx build \
-    --cache-from "type=registry,ref=${CACHE_REGISTRY}/web-cache:${CACHE_TAG},ignore-error=true" \
-    --cache-to "type=registry,ref=${CACHE_REGISTRY}/web-cache:${CACHE_TAG},mode=max" \
+    --cache-from "type=registry,ref=${CACHE_REGISTRY}/web-cache:${CACHE_TAG_PRIMARY},ignore-error=true" \
     -f infra/docker/web-artifact.Dockerfile \
     --build-arg "NGINX_CONF=$NGINX_CONF_FILE" \
     -t "$WEB_IMAGE" \
@@ -182,4 +179,4 @@ else
 fi
 
 echo "=== Build complete ==="
-docker images | grep "xiaoxia-saas.*:$VERSION"
+docker images | grep "xiaoxia-saas" | grep "$VERSION" || true

scripts/check_migration_safety.py

@@ -119,9 +119,7 @@ def get_new_migrations_via_diff(diff_target: str) -> List[Path]:
             text=True,
             check=True,
         )
-        files = [
-            line.strip() for line in result.stdout.strip().split("\n") if line.strip()
-        ]
+        files = [line.strip() for line in result.stdout.strip().split("\n") if line.strip()]
         return [REPO_ROOT / f for f in files]
     except subprocess.CalledProcessError as e:
         print(f"⚠️  git diff 失败（{diff_target}）：{e.stderr.strip()}")
@@ -129,9 +127,7 @@ def get_new_migrations_via_diff(diff_target: str) -> List[Path]:
         return sorted(ALEMBIC_VERSIONS_DIR.glob("*.py"))
 
 
-def find_new_migrations(
-    since_revision: str | None = None, diff_against: str | None = None
-) -> List[Path]:
+def find_new_migrations(since_revision: str | None = None, diff_against: str | None = None) -> List[Path]:
     """
     找出需要检查的迁移文件。
     优先级：diff_against > since_revision > 全部
@@ -183,9 +179,7 @@ def analyze_migration(file_path: Path) -> Tuple[List[str], List[str], List[str]]
 
 
 def main() -> int:
-    parser = argparse.ArgumentParser(
-        description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter
-    )
+    parser = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
     parser.add_argument(
         "--since",
         default=os.getenv("MIGRATION_SINCE_REVISION"),
@@ -248,9 +242,7 @@ def main() -> int:
         print()
 
     print("=" * 60)
-    print(
-        f"检查结果：{len(all_safe)} 项安全 / {len(all_medium)} 项中风险 / {len(all_high)} 项高风险"
-    )
+    print(f"检查结果：{len(all_safe)} 项安全 / {len(all_medium)} 项中风险 / {len(all_high)} 项高风险")
     print()
 
     if all_high:

scripts/gray_deploy.sh

@@ -0,0 +1,117 @@
+#!/bin/bash
+# 灰度发布脚本：通过Nginx权重调整流量比例
+# 用法: ./scripts/gray_deploy.sh <版本号> <灰度百分比>
+#
+# 需要在目标服务器上执行，或通过SSH执行
+# 前提：服务器上运行两个版本的容器（stable + canary），Nginx做加权轮询
+
+set -euo pipefail
+
+VERSION="${1:-}"
+GRAY_PCT="${2:-10}"
+
+if [[ -z "$VERSION" ]]; then
+    echo "用法: $0 <版本号> [灰度百分比]"
+    echo "示例: $0 v0.1.129 5"
+    exit 1
+fi
+
+STABLE_VERSION="${STABLE_VERSION:-current}"
+REGISTRY="${REGISTRY:-git.xiaoxiajianji.com/xiaoxia/xiaoxia-saas}"
+
+echo "=========================================="
+echo "  灰度发布"
+echo "  新版本: $VERSION"
+echo "  灰度比例: ${GRAY_PCT}%"
+echo "  稳定版本: $STABLE_VERSION"
+echo "=========================================="
+
+# 1. 拉取新版本镜像
+echo ""
+echo ">>> 拉取新版本镜像..."
+for component in api worker web; do
+    echo "  拉取 $component:$VERSION ..."
+    docker pull "${REGISTRY}-${component}:${VERSION}" 2>&1 | tail -1
+done
+
+# 2. 启动灰度版本容器（canary）
+echo ""
+echo ">>> 启动灰度版本容器..."
+
+# API canary
+CANARY_API_NAME="saas-api-canary"
+if docker ps -a --format '{{.Names}}' | grep -q "^${CANARY_API_NAME}$"; then
+    echo "  停止旧 canary 容器..."
+    docker stop "$CANARY_API_NAME" 2>/dev/null || true
+    docker rm "$CANARY_API_NAME" 2>/dev/null || true
+fi
+
+echo "  启动 api canary..."
+docker run -d \
+    --name "$CANARY_API_NAME" \
+    --network saas-network \
+    -e DATABASE_URL="${DATABASE_URL}" \
+    -e REDIS_URL="${REDIS_URL}" \
+    -e FEATURE_FLAG_PROVIDER=redis \
+    --restart unless-stopped \
+    "${REGISTRY}-api:${VERSION}"
+
+# Worker canary
+CANARY_WORKER_NAME="saas-worker-canary"
+if docker ps -a --format '{{.Names}}' | grep -q "^${CANARY_WORKER_NAME}$"; then
+    echo "  停止旧 worker canary..."
+    docker stop "$CANARY_WORKER_NAME" 2>/dev/null || true
+    docker rm "$CANARY_WORKER_NAME" 2>/dev/null || true
+fi
+
+echo "  启动 worker canary..."
+docker run -d \
+    --name "$CANARY_WORKER_NAME" \
+    --network saas-network \
+    -e DATABASE_URL="${DATABASE_URL}" \
+    -e REDIS_URL="${REDIS_URL}" \
+    --restart unless-stopped \
+    "${REGISTRY}-worker:${VERSION}"
+
+# 3. 等待容器健康
+echo ""
+echo ">>> 等待容器健康..."
+sleep 10
+if ! docker ps --format '{{.Names}} {{.Status}}' | grep -q "$CANARY_API_NAME"; then
+    echo "错误: API canary 容器未运行"
+    docker logs "$CANARY_API_NAME" --tail 20
+    exit 1
+fi
+echo "  ✅ API canary 运行中"
+
+# 4. 更新Nginx权重
+echo ""
+echo ">>> 更新Nginx权重 (稳定: $((100-GRAY_PCT))% / 灰度: ${GRAY_PCT}%)..."
+
+NGINX_CONF="${NGINX_CONF:-/etc/nginx/conf.d/saas-api.conf}"
+if [[ -f "$NGINX_CONF" ]]; then
+    # 备份
+    cp "$NGINX_CONF" "${NGINX_CONF}.bak.$(date +%Y%m%d%H%M%S)"
+    
+    # 更新 upstream 权重（需要根据实际配置调整）
+    echo "  请手动更新 Nginx upstream 配置中的权重"
+    echo "  示例配置:"
+    cat <<EOF
+    upstream saas_api_backend {
+        server saas-api:8000 weight=$((100-GRAY_PCT));
+        server saas-api-canary:8000 weight=${GRAY_PCT};
+    }
+EOF
+    nginx -t && nginx -s reload
+    echo "  ✅ Nginx 已reload"
+else
+    echo "  警告: Nginx 配置文件不存在 ($NGINX_CONF)"
+    echo "  请手动配置灰度流量权重"
+fi
+
+echo ""
+echo "=========================================="
+echo "  ✅ 灰度发布完成"
+echo "  新版本: $VERSION (${GRAY_PCT}%流量)"
+echo "  监控: Grafana / 日志"
+echo "=========================================="

scripts/release.sh

@@ -0,0 +1,126 @@
+#!/bin/bash
+# 一键发布脚本：打tag → 触发生产镜像构建 → 部署到灰度
+# 用法: ./scripts/release.sh v0.1.129 [--gray 5]
+
+set -euo pipefail
+
+SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+REPO_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
+
+usage() {
+    echo "用法: $0 <版本号> [--gray 百分比] [--no-deploy]"
+    echo ""
+    echo "示例:"
+    echo "  $0 v0.1.129                    # 打tag + 全量发布"
+    echo "  $0 v0.1.129 --gray 5           # 打tag + 5%灰度发布"
+    echo "  $0 v0.1.129 --no-deploy        # 只打tag，不部署"
+    exit 1
+}
+
+# 参数解析
+VERSION=""
+GRAY_PCT=0
+DEPLOY=true
+
+while [[ $# -gt 0 ]]; do
+    case "$1" in
+        --gray)
+            GRAY_PCT="$2"
+            shift 2
+            ;;
+        --no-deploy)
+            DEPLOY=false
+            shift
+            ;;
+        -h|--help)
+            usage
+            ;;
+        v*)
+            VERSION="$1"
+            shift
+            ;;
+        *)
+            echo "未知参数: $1"
+            usage
+            ;;
+    esac
+done
+
+if [[ -z "$VERSION" ]]; then
+    echo "错误: 请指定版本号（如 v0.1.129）"
+    usage
+fi
+
+echo "=========================================="
+echo "  发布版本: $VERSION"
+echo "  灰度比例: ${GRAY_PCT}%"
+echo "  自动部署: $DEPLOY"
+echo "=========================================="
+
+cd "$REPO_ROOT"
+
+# 1. 确认在 develop 分支
+CURRENT_BRANCH=$(git rev-parse --abbrev-ref HEAD)
+if [[ "$CURRENT_BRANCH" != "develop" ]]; then
+    echo "错误: 请切换到 develop 分支后再发布"
+    exit 1
+fi
+
+# 2. 拉取最新代码
+echo ""
+echo ">>> 拉取最新代码..."
+git pull origin develop
+
+# 3. 生成 CHANGELOG
+echo ""
+echo ">>> 生成 CHANGELOG..."
+if [[ -f "scripts/generate_changelog.py" ]]; then
+    PREV_TAG=$(git describe --tags --abbrev=0 HEAD^ 2>/dev/null || echo "")
+    if [[ -n "$PREV_TAG" ]]; then
+        python3 scripts/generate_changelog.py \
+            --from-tag "$PREV_TAG" \
+            --to-tag HEAD \
+            --gitea-token "${GITEA_TOKEN:-}" \
+            --output /tmp/changelog_$$.md
+        echo "CHANGELOG 已生成到 /tmp/changelog_$$.md"
+    fi
+fi
+
+# 4. 打tag
+echo ""
+echo ">>> 打 tag $VERSION ..."
+if git rev-parse "$VERSION" >/dev/null 2>&1; then
+    echo "警告: tag $VERSION 已存在，跳过打tag"
+else
+    git tag -a "$VERSION" -m "Release $VERSION"
+    git push origin "$VERSION"
+    echo "Tag $VERSION 已推送，触发生产镜像构建..."
+fi
+
+# 5. 等待镜像构建
+if [[ "$DEPLOY" == "true" ]]; then
+    echo ""
+    echo ">>> 等待镜像构建完成（约10-15分钟）..."
+    echo "    镜像: git.xiaoxiajianji.com/xiaoxia/xiaoxia-saas-{api,worker,web}:$VERSION"
+    
+    # 这里可以加镜像存在性检查
+    echo "    （镜像构建由CI自动完成，请在Gitea Actions中确认）"
+fi
+
+# 6. 灰度部署
+if [[ "$DEPLOY" == "true" && "$GRAY_PCT" -gt 0 ]]; then
+    echo ""
+    echo ">>> 灰度部署: ${GRAY_PCT}% 流量到 $VERSION"
+    if [[ -f "scripts/gray_deploy.sh" ]]; then
+        ./scripts/gray_deploy.sh "$VERSION" "$GRAY_PCT"
+    else
+        echo "警告: gray_deploy.sh 不存在，跳过灰度部署"
+    fi
+fi
+
+echo ""
+echo "=========================================="
+echo "  ✅ 发布流程完成"
+echo "  版本: $VERSION"
+echo "  灰度: ${GRAY_PCT}%"
+echo "=========================================="

scripts/rollback.sh

@@ -0,0 +1,42 @@
+#!/bin/bash
+# 灰度回滚脚本：切回稳定版本流量
+# 用法: ./scripts/rollback.sh [稳定版本号]
+
+set -euo pipefail
+
+STABLE_VERSION="${1:-current}"
+
+echo "=========================================="
+echo "  灰度回滚"
+echo "  切回稳定版本: $STABLE_VERSION"
+echo "=========================================="
+
+# 1. 恢复Nginx全量到稳定版本
+echo ""
+echo ">>> 恢复Nginx全量流量到稳定版本..."
+
+NGINX_CONF="${NGINX_CONF:-/etc/nginx/conf.d/saas-api.conf}"
+if [[ -f "$NGINX_CONF" ]]; then
+    # 找最近的备份
+    LATEST_BAK=$(ls -t "${NGINX_CONF}".bak.* 2>/dev/null | head -1)
+    if [[ -n "$LATEST_BAK" ]]; then
+        cp "$LATEST_BAK" "$NGINX_CONF"
+        echo "  从备份恢复: $LATEST_BAK"
+    else
+        echo "  未找到备份，请手动移除 canary upstream"
+    fi
+    
+    nginx -t && nginx -s reload
+    echo "  ✅ Nginx 已回滚"
+fi
+
+# 2. 停止灰度版本容器（保留30分钟以便排查）
+echo ""
+echo ">>> 灰度版本容器将在30分钟后停止（便于排查问题）"
+echo "    立即停止请执行: docker stop saas-api-canary saas-worker-canary"
+
+echo ""
+echo "=========================================="
+echo "  ✅ 回滚完成"
+echo "  流量已全部切回稳定版本"
+echo "=========================================="

setup.cfg

@@ -15,3 +15,4 @@ exclude =
 per-file-ignores =
     */__init__.py:F401,F403,F405
     tests/*:E402,F401,F841
+    packages/ports/*:E301,E704

tests/integration/test_assets_api.py

@@ -129,10 +129,58 @@ class StubAssetRepository:
         return False
 
     def batch_delete(self, asset_ids: list[str]) -> int:
+        """软删除：标记 status=deleted。"""
+        from datetime import datetime, timezone
+
+        from packages.domain import AssetStatus
+
         count = 0
         for aid in asset_ids:
-            if aid in self._assets:
-                del self._assets[aid]
+            asset = self._assets.get(aid)
+            if asset and asset.status != AssetStatus.DELETED:
+                asset.status = AssetStatus.DELETED
+                asset.updated_at = datetime.now(timezone.utc)
+                count += 1
+        return count
+
+    def batch_update_metadata(self, asset_ids: list[str], metadata_patch: dict) -> int:
+        from datetime import datetime, timezone
+
+        count = 0
+        for aid in asset_ids:
+            asset = self._assets.get(aid)
+            if asset:
+                asset.metadata = {**asset.metadata, **metadata_patch}
+                asset.updated_at = datetime.now(timezone.utc)
+                count += 1
+        return count
+
+    def batch_add_tags(self, asset_ids: list[str], tag_ids: list[str]) -> int:
+        from datetime import datetime, timezone
+
+        count = 0
+        for aid in asset_ids:
+            asset = self._assets.get(aid)
+            if asset:
+                changed = False
+                for tid in tag_ids:
+                    if tid not in asset.tag_ids:
+                        asset.tag_ids.append(tid)
+                        changed = True
+                if changed:
+                    asset.updated_at = datetime.now(timezone.utc)
+                    count += 1
+        return count
+
+    def batch_replace_tags(self, asset_ids: list[str], tag_ids: list[str]) -> int:
+        from datetime import datetime, timezone
+
+        count = 0
+        for aid in asset_ids:
+            asset = self._assets.get(aid)
+            if asset:
+                asset.tag_ids = list(tag_ids)
+                asset.updated_at = datetime.now(timezone.utc)
                 count += 1
         return count
 
@@ -613,17 +661,22 @@ class TestBatchDeleteAssets:
         return ids
 
     def test_batch_delete_success(self, client):
-        """批量删除成功。"""
+        """批量删除成功（软删除）。"""
         ids = self._create_assets(client, 3)
 
         resp = client.post(
             "/api/v1/assets/batch-delete",
-            json={"ids": ids[:2]},
+            json={"asset_ids": ids[:2]},
         )
         assert resp.status_code == 200
         data = resp.json()
-        assert data["deleted_count"] == 2
+        assert data["success_count"] == 2
         assert len(data["failed_ids"]) == 0
+        # 软删除：记录仍在，status 变为 deleted
+        for aid in ids[:2]:
+            r = client.get(f"/api/v1/assets/{aid}")
+            assert r.status_code == 200
+            assert r.json()["status"] == "deleted"
 
     def test_batch_delete_with_nonexistent_ids(self, client):
         """批量删除包含不存在的 ID，失败的计入 failed_ids。"""
@@ -632,18 +685,20 @@ class TestBatchDeleteAssets:
 
         resp = client.post(
             "/api/v1/assets/batch-delete",
-            json={"ids": ids},
+            json={"asset_ids": ids},
         )
         assert resp.status_code == 200
         data = resp.json()
-        assert data["deleted_count"] == 2
+        assert data["success_count"] == 2
         assert "nonexistent-id" in data["failed_ids"]
+        assert "nonexistent-id" in data["failed_details"]
+        assert data["failed_details"]["nonexistent-id"] == "not_found"
 
     def test_batch_delete_empty_list_returns_422(self, client):
         """空列表返回 422。"""
         resp = client.post(
             "/api/v1/assets/batch-delete",
-            json={"ids": []},
+            json={"asset_ids": []},
         )
         assert resp.status_code == 422
 

tests/integration/test_auth.py

@@ -308,7 +308,7 @@ class TestPasswordReset:
         )
 
         response = client.post(
-            "/api/v1/auth/password/forgot",
+            "/api/v1/auth/forgot-password",
             json={"email": test_email},
         )
 
@@ -318,7 +318,7 @@ class TestPasswordReset:
     def test_request_password_reset_nonexistent_user(self):
         """测试请求不存在的用户密码重置"""
         response = client.post(
-            "/api/v1/auth/password/forgot",
+            "/api/v1/auth/forgot-password",
             json={"email": "nonexistent@example.com"},
         )