fix: legacy渲染路径fps统一 + 自动选素材过滤横屏视频

修复渲染失败exit=234的两个核心根因： 1. legacy路径 _build_clip_filter 缺少fps滤镜 - concat filter 要求所有输入帧率完全一致 - 之前 fps 参数传入但完全未使用 - 现在 crop 后统一加上 fps=DEFAULT_FPS 滤镜 2. 自动选素材过滤横屏视频 - 竖屏APP只选竖屏视频（height >= width） - 宽>高的横屏视频直接跳过 - 尺寸未知的存量素材放过（兼容）补充测试： - test_fps_filter_applied_to_every_clip：验证每个片段都有fps滤镜且顺序正确 - test_landscape_videos_filtered：横屏视频被过滤 - test_missing_dimensions_kept：尺寸未知的存量素材放过 - test_combined_filters_codec_and_landscape：编码+横竖屏联合过滤
fix: 自动选素材过滤HEVC等不支持编码 + ingest编码格式校验
2026-07-18 11:30:43 +08:00 · 2026-07-18 11:22:56 +08:00 · 2026-07-18 10:53:10 +08:00 · 2026-07-18 09:21:11 +08:00 · 2026-07-18 08:38:59 +08:00 · 2026-07-18 00:25:13 +08:00
171 changed files with 19000 additions and 1755 deletions
@@ -1,14 +0,0 @@
-[flake8]
-max-line-length = 120
-exclude =
-    .git,
-    .cache,
-    __pycache__,
-    .venv,
-    venv,
-    node_modules,
-    alembic
-
-per-file-ignores =
-    tests/integration/*:F821
-    tests/unit/*:F821
@@ -0,0 +1,174 @@
+name: Auto Approve CI PRs
+
+on:
+  pull_request:
+    types: [synchronize, opened, ready_for_review]
+
+jobs:
+  auto-approve:
+    name: Auto Approve on CI Green
+    runs-on: ci-check
+    if: github.event_name == 'pull_request' && !github.event.pull_request.draft
+    timeout-minutes: 20
+    steps:
+      - name: Checkout code
+        shell: sh
+        env:
+          GITHUB_TOKEN: ${{ github.token }}
+        run: "set -eu\npython3 - <<'PY'\nimport io, os, tarfile, time, urllib.request, urllib.error\nurl = f\"{os.environ['GITHUB_API_URL']}/repos/{os.environ['GITHUB_REPOSITORY']}/archive/{os.environ['GITHUB_SHA']}.tar.gz\"\nrequest = urllib.request.Request(url, headers={\"Authorization\": f\"token {os.environ['GITHUB_TOKEN']}\"})\nlast_err = None\nfor attempt in range(5):\n    try:\n        with urllib.request.urlopen(request, timeout=120) as response:\n            archive = response.read()\n        break\n    except urllib.error.HTTPError as e:\n        last_err = e\n        if e.code >= 500 and attempt < 4:\n            wait = 2 ** attempt\n            print(f\"Checkout HTTP {e.code}, retrying in {wait}s (attempt {attempt+1}/5)...\")\n            time.sleep(wait)\n            continue\n        raise\n    except Exception as e:\n        last_err = e\n        if attempt < 4:\n            wait = 2 ** attempt\n            print(f\"Checkout error: {e}, retrying in {wait}s (attempt {attempt+1}/5)...\")\n            time.sleep(wait)\n            continue\n        raise\nelse:\n    raise last_err\nwith tarfile.open(fileobj=io.BytesIO(archive), mode='r:gz') as tar:\n    root_prefix = tar.getmembers()[0].name.split('/', 1)[0] + '/'\n    for member in tar.getmembers():\n        name = member.name\n        if name == root_prefix[:-1]:\n            continue\n        if name.startswith(root_prefix):\n            member.name = name[len(root_prefix):]\n        if member.name:\n            tar.extract(member, '.')\nPY\n"
+
+      - name: Auto approve when CI passes
+        shell: bash
+        env:
+          GITHUB_TOKEN: ${{ github.token }}
+          REVIEW_TOKEN: ${{ secrets.REVIEW_GITEA_TOKEN }}
+          PR_NUMBER: ${{ github.event.pull_request.number }}
+          PR_HEAD_SHA: ${{ github.event.pull_request.head.sha }}
+        run: |
+          set -eu
+
+          echo "PR #${PR_NUMBER} - 检查CI状态并自动审批"
+
+          # 检查是否纯前端改动
+          API_URL="${GITHUB_API_URL}/repos/${GITHUB_REPOSITORY}/pulls/${PR_NUMBER}/files?limit=300"
+          FILES=$(curl -s -H "Authorization: token ${GITHUB_TOKEN}" "$API_URL" | python3 -c "import sys,json; [print(f['filename']) for f in json.load(sys.stdin)]")
+          FRONTEND_COUNT=$(echo "$FILES" | grep -c '^apps/web/' || true)
+          BACKEND_COUNT=$(echo "$FILES" | grep -cv '^apps/web/' || true)
+          TOTAL=$(echo "$FILES" | grep -cv '^$' || true)
+          echo "变更文件: ${TOTAL} 个 (前端: ${FRONTEND_COUNT}, 后端/公共: ${BACKEND_COUNT})"
+
+          if [ "$BACKEND_COUNT" = "0" ] && [ "$FRONTEND_COUNT" -gt "0" ]; then
+            SKIP_BACKEND=true
+            echo "✅ 纯前端改动，只检查Frontend Lint"
+          else
+            SKIP_BACKEND=false
+            echo "🔧 包含后端/公共变更，检查全部CI"
+          fi
+
+          # 定义需要检查的context
+          if [ "$SKIP_BACKEND" = "true" ]; then
+            CONTEXTS=("CI/CD Pipeline / Frontend Lint (pull_request)")
+          else
+            CONTEXTS=(
+              "CI/CD Pipeline / Validate Code Quality And Tests (pull_request)"
+              "CI/CD Pipeline / Unit Tests (pull_request)"
+              "CI/CD Pipeline / Frontend Lint (pull_request)"
+            )
+          fi
+
+          echo "需要通过的CI检查: ${#CONTEXTS[@]} 项"
+          for ctx in "${CONTEXTS[@]}"; do
+            echo "  - $ctx"
+          done
+          echo
+
+          # 初始等待30秒，给CI启动写status的时间，避免checkout太快导致全找不到context误判
+          echo "等待30秒让CI启动..."
+          sleep 30
+
+          # 轮询等待，最多20分钟（120次x10秒）
+          for attempt in $(seq 1 120); do
+            ALL_SUCCESS=true
+            ANY_FAILED=false
+            ANY_PENDING=false
+
+            echo "--- 第${attempt}次检查 ($(date '+%H:%M:%S')) ---"
+
+            # 调用辅助脚本检查每个context状态
+            for ctx in "${CONTEXTS[@]}"; do
+              STATE=$(python3 scripts/check_ci_status.py "$GITHUB_TOKEN" "$GITHUB_REPOSITORY" "$PR_HEAD_SHA" "$ctx")
+              echo "  $ctx: $STATE"
+
+              if [ "$STATE" != "success" ]; then
+                ALL_SUCCESS=false
+              fi
+              if [ "$STATE" = "failure" ] || [ "$STATE" = "error" ]; then
+                ANY_FAILED=true
+              fi
+              if [ "$STATE" = "pending" ] || [ "$STATE" = "null" ]; then
+                ANY_PENDING=true
+              fi
+            done
+
+            if [ "$ALL_SUCCESS" = "true" ]; then
+              echo
+              echo "✅ 所有CI检查通过，自动审批 PR #${PR_NUMBER}"
+
+              # 检查是否已有审批（任何用户的APPROVED都算，避免重复审批）
+              EXISTING=$(curl -s -H "Authorization: token ${REVIEW_TOKEN}" \
+                "${GITHUB_API_URL}/repos/${GITHUB_REPOSITORY}/pulls/${PR_NUMBER}/reviews" \
+                | python3 -c "import sys,json; reviews=json.load(sys.stdin); print('yes' if any(r.get('state')=='APPROVED' for r in reviews) else 'no')")
+
+              if [ "$EXISTING" = "yes" ]; then
+                echo "ℹ️  PR #${PR_NUMBER} 已有审批，跳过"
+                exit 0
+              fi
+
+              # 第一步：创建PENDING review（Gitea API需要先创建再提交）
+              echo "创建review..."
+              REVIEW_CREATE=$(curl -s -X POST \
+                -H "Authorization: token ${REVIEW_TOKEN}" \
+                -H "Content-Type: application/json" \
+                -d '{"event": "PENDING", "body": "CI全绿，自动审批通过。"}' \
+                "${GITHUB_API_URL}/repos/${GITHUB_REPOSITORY}/pulls/${PR_NUMBER}/reviews")
+              
+              REVIEW_ID=$(echo "$REVIEW_CREATE" | python3 -c "import sys,json; print(json.load(sys.stdin).get('id',''))")
+              REVIEW_STATE=$(echo "$REVIEW_CREATE" | python3 -c "import sys,json; print(json.load(sys.stdin).get('state',''))")
+              echo "创建结果: id=$REVIEW_ID state=$REVIEW_STATE"
+
+              if [ -z "$REVIEW_ID" ]; then
+                echo "❌ 创建review失败"
+                echo "$REVIEW_CREATE"
+                exit 1
+              fi
+
+              # 如果已经是APPROVED就不用再submit了（兼容不同Gitea版本）
+              if [ "$REVIEW_STATE" = "APPROVED" ]; then
+                echo "✅ 自动审批成功（直接创建为APPROVED）"
+                exit 0
+              fi
+
+              # 第二步：submit review为APPROVED
+              echo "提交review审批..."
+              SUBMIT_CODE=$(curl -s -o /tmp/submit_resp.json -w "%{http_code}" \
+                -X POST \
+                -H "Authorization: token ${REVIEW_TOKEN}" \
+                -H "Content-Type: application/json" \
+                -d '{"event": "APPROVED", "body": "CI全绿，自动审批通过。"}' \
+                "${GITHUB_API_URL}/repos/${GITHUB_REPOSITORY}/pulls/${PR_NUMBER}/reviews/${REVIEW_ID}")
+              
+              echo "提交API HTTP状态: $SUBMIT_CODE"
+              cat /tmp/submit_resp.json 2>/dev/null || true
+              echo
+
+              if [ "$SUBMIT_CODE" = "200" ] || [ "$SUBMIT_CODE" = "201" ]; then
+                FINAL_STATE=$(python3 -c "import json; print(json.load(open('/tmp/submit_resp.json')).get('state',''))" 2>/dev/null || echo "?")
+                echo "✅ 自动审批成功 (state: $FINAL_STATE)"
+                exit 0
+              else
+                echo "❌ 提交审批失败"
+                exit 1
+              fi
+            fi
+
+            # 还有CI在跑（pending状态）→ 继续等
+            if [ "$ANY_PENDING" = "true" ]; then
+              echo "⏳ CI仍在运行中，继续等待（第${attempt}/120次轮询）..."
+              sleep 10
+              continue
+            fi
+
+            # 所有CI都跑完了但有失败 → 退出
+            if [ "$ANY_FAILED" = "true" ]; then
+              echo
+              echo "❌ CI检查有失败项，不自动审批"
+              exit 0
+            fi
+
+            # 其他情况继续等
+            sleep 10
+          done
+
+          echo
+          echo "⏰ 等待超时（20分钟），CI尚未全部完成"
+          exit 0
@@ -0,0 +1,170 @@
+name: Auto Merge CI PRs
+
+on:
+  pull_request:
+    types: [synchronize, opened, ready_for_review, review_requested]
+
+jobs:
+  auto-merge:
+    name: Auto Merge on CI Green + Approved
+    runs-on: ci-check
+    if: github.event_name == 'pull_request' && !github.event.pull_request.draft && github.event.pull_request.base.ref == 'develop'
+    timeout-minutes: 30
+    steps:
+      - name: Checkout code
+        shell: sh
+        env:
+          GITHUB_TOKEN: ${{ github.token }}
+        run: "set -eu\npython3 - <<'PY'\nimport io, os, tarfile, time, urllib.request, urllib.error\nurl = f\"{os.environ['GITHUB_API_URL']}/repos/{os.environ['GITHUB_REPOSITORY']}/archive/{os.environ['GITHUB_SHA']}.tar.gz\"\nrequest = urllib.request.Request(url, headers={\"Authorization\": f\"token {os.environ['GITHUB_TOKEN']}\"})\nlast_err = None\nfor attempt in range(5):\n    try:\n        with urllib.request.urlopen(request, timeout=120) as response:\n            archive = response.read()\n        break\n    except urllib.error.HTTPError as e:\n        last_err = e\n        if e.code >= 500 and attempt < 4:\n            wait = 2 ** attempt\n            print(f\"Checkout HTTP {e.code}, retrying in {wait}s (attempt {attempt+1}/5)...\")\n            time.sleep(wait)\n            continue\n        raise\n    except Exception as e:\n        last_err = e\n        if attempt < 4:\n            wait = 2 ** attempt\n            print(f\"Checkout error: {e}, retrying in {wait}s (attempt {attempt+1}/5)...\")\n            time.sleep(wait)\n            continue\n        raise\nelse:\n    raise last_err\nwith tarfile.open(fileobj=io.BytesIO(archive), mode='r:gz') as tar:\n    root_prefix = tar.getmembers()[0].name.split('/', 1)[0] + '/'\n    for member in tar.getmembers():\n        name = member.name\n        if name == root_prefix[:-1]:\n            continue\n        if name.startswith(root_prefix):\n            member.name = name[len(root_prefix):]\n        if member.name:\n            tar.extract(member, '.')\nPY\n"
+
+      - name: Auto merge when CI passes and approved
+        shell: bash
+        env:
+          GITHUB_TOKEN: ${{ github.token }}
+          MERGE_TOKEN: ${{ secrets.REVIEW_GITEA_TOKEN }}
+          PR_NUMBER: ${{ github.event.pull_request.number }}
+          PR_HEAD_SHA: ${{ github.event.pull_request.head.sha }}
+          BASE_REF: ${{ github.event.pull_request.base.ref }}
+        run: |
+          set -eu
+
+          echo "PR #${PR_NUMBER} - 检查CI状态+审批并自动合并到${BASE_REF}"
+          echo
+
+          # 只合develop分支
+          if [ "$BASE_REF" != "develop" ]; then
+            echo "Skip: 目标分支不是develop"
+            exit 0
+          fi
+
+          # 判断是否纯前端改动
+          FILES=$(curl -s -H "Authorization: token ${GITHUB_TOKEN}" \
+            "${GITHUB_API_URL}/repos/${GITHUB_REPOSITORY}/pulls/${PR_NUMBER}/files?limit=300" \
+            | python3 -c "import sys,json; [print(f['filename']) for f in json.load(sys.stdin)]")
+          TOTAL=$(echo "$FILES" | grep -cv '^$' || true)
+          FRONTEND_COUNT=$(echo "$FILES" | grep -c '^apps/web/' || true)
+          BACKEND_COUNT=$((TOTAL - FRONTEND_COUNT))
+          echo "变更文件: ${TOTAL} 个 (前端: ${FRONTEND_COUNT}, 后端/公共: ${BACKEND_COUNT})"
+
+          if [ "$BACKEND_COUNT" = "0" ] && [ "$FRONTEND_COUNT" -gt "0" ]; then
+            CONTEXTS=("CI/CD Pipeline / Frontend Lint (pull_request)")
+            echo "纯前端改动，只检查Frontend Lint"
+          else
+            CONTEXTS=(
+              "CI/CD Pipeline / Validate Code Quality And Tests (pull_request)"
+              "CI/CD Pipeline / Unit Tests (pull_request)"
+              "CI/CD Pipeline / Frontend Lint (pull_request)"
+              "CI/CD Pipeline / Integration Tests (pull_request)"
+            )
+            echo "检查全部四门禁"
+          fi
+          echo
+
+          # 初始等待30秒，给CI启动写status的时间，避免checkout太快导致全找不到context误判
+          echo "等待30秒让CI启动..."
+          sleep 30
+
+          # 405连续计数器：连续多次合并返回405才放弃
+          MERGE_405_COUNT=0
+          MAX_405_RETRIES=6
+
+          # 轮询等待，最多30分钟（180次x10秒）
+          for attempt in $(seq 1 180); do
+            ALL_SUCCESS=true
+            ANY_FAILED=false
+            ANY_PENDING=false
+
+            echo "--- 第${attempt}次检查 ($(date '+%H:%M:%S')) ---"
+
+            # 检查CI状态
+            for ctx in "${CONTEXTS[@]}"; do
+              STATE=$(python3 scripts/check_ci_status.py "$GITHUB_TOKEN" "$GITHUB_REPOSITORY" "$PR_HEAD_SHA" "$ctx")
+              echo "  CI: ${ctx##*/}: $STATE"
+              if [ "$STATE" != "success" ]; then
+                ALL_SUCCESS=false
+              fi
+              if [ "$STATE" = "failure" ] || [ "$STATE" = "error" ]; then
+                ANY_FAILED=true
+              fi
+              if [ "$STATE" = "pending" ]; then
+                ANY_PENDING=true
+              fi
+            done
+
+            # 检查审批状态
+            APPROVAL_RESULT=$(python3 scripts/check_pr_approval.py "$MERGE_TOKEN" "$GITHUB_REPOSITORY" "$PR_NUMBER" 1)
+            echo "  审批: $APPROVAL_RESULT"
+            HAS_APPROVAL=false
+            if echo "$APPROVAL_RESULT" | grep -q '^approved'; then
+              HAS_APPROVAL=true
+            fi
+
+            # 全部满足 → 合并
+            if [ "$ALL_SUCCESS" = "true" ] && [ "$HAS_APPROVAL" = "true" ]; then
+              echo
+              echo "CI全绿 + 审批通过，执行自动合并"
+
+              # 幂等检查：PR是否还是open
+              PR_STATE=$(curl -s -H "Authorization: token ${MERGE_TOKEN}" \
+                "${GITHUB_API_URL}/repos/${GITHUB_REPOSITORY}/pulls/${PR_NUMBER}" \
+                | python3 -c "import sys,json; print(json.load(sys.stdin).get('state',''))")
+
+              if [ "$PR_STATE" != "open" ]; then
+                echo "PR状态为 ${PR_STATE}，无需合并"
+                exit 0
+              fi
+
+              # 执行squash merge
+              HTTP_CODE=$(curl -s -o /tmp/merge_resp.json -w "%{http_code}" \
+                -X POST \
+                -H "Authorization: token ${MERGE_TOKEN}" \
+                -H "Content-Type: application/json" \
+                -d '{"do":"squash","merge_title_field":"","merge_message_field":"","delete_branch_after_merge":true,"force_merge":false}' \
+                "${GITHUB_API_URL}/repos/${GITHUB_REPOSITORY}/pulls/${PR_NUMBER}/merge")
+
+              echo "合并API HTTP状态: $HTTP_CODE"
+
+              if [ "$HTTP_CODE" = "200" ]; then
+                echo "自动合并成功"
+                exit 0
+              elif [ "$HTTP_CODE" = "405" ]; then
+                MERGE_405_COUNT=$((MERGE_405_COUNT + 1))
+                echo "⚠️  合并返回405（第${MERGE_405_COUNT}次），可能CI状态尚未同步或有未解决的门禁，继续等待重试..."
+                cat /tmp/merge_resp.json 2>/dev/null || true
+                echo
+                if [ "$MERGE_405_COUNT" -ge "$MAX_405_RETRIES" ]; then
+                  echo "❌ 连续${MAX_405_RETRIES}次合并返回405，放弃自动合并"
+                  curl -s -X POST \
+                    -H "Authorization: token ${MERGE_TOKEN}" \
+                    -H "Content-Type: application/json" \
+                    -d '{"body": "Auto merge failed after multiple 405 errors: PR may have conflicts or unresolved checks. Please review manually."}' \
+                    "${GITHUB_API_URL}/repos/${GITHUB_REPOSITORY}/issues/${PR_NUMBER}/comments" > /dev/null 2>&1 || true
+                  exit 1
+                fi
+              else
+                echo "自动合并失败 (HTTP $HTTP_CODE)"
+                cat /tmp/merge_resp.json 2>/dev/null || true
+                curl -s -X POST \
+                  -H "Authorization: token ${MERGE_TOKEN}" \
+                  -H "Content-Type: application/json" \
+                  -d "{\"body\": \"Auto merge failed (HTTP ${HTTP_CODE}), please check manually.\"}" \
+                  "${GITHUB_API_URL}/repos/${GITHUB_REPOSITORY}/issues/${PR_NUMBER}/comments" > /dev/null 2>&1 || true
+                exit 1
+              fi
+            else
+              # 本轮不满足合并条件，重置405计数器
+              MERGE_405_COUNT=0
+            fi
+
+            if [ "$ANY_FAILED" = "true" ]; then
+              echo
+              echo "CI有失败项，不自动合并"
+              exit 0
+            fi
+
+            sleep 10
+          done
+
+          echo
+          echo "等待超时（30分钟）"
+          exit 0
@@ -0,0 +1,36 @@
+name: CI Trigger Monitor
+
+on:
+  schedule:
+    - cron: '*/5 * * * *'  # 每5分钟检查一次
+  workflow_dispatch:
+    inputs:
+      stale_threshold:
+        description: 'CI未触发告警阈值（分钟）'
+        required: false
+        default: '5'
+
+permissions:
+  contents: read
+
+jobs:
+  monitor:
+    name: Monitor CI Trigger Reliability
+    runs-on: ubuntu-latest
+    timeout-minutes: 5
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v3
+      
+      - name: Check CI trigger status for all open PRs
+        env:
+          GITEA_API_TOKEN: ${{ secrets.REVIEW_GITEA_TOKEN }}
+          GITEA_URL: https://git.xiaoxiajianji.com
+          GITEA_REPO: xiaoxia/xiaoxia-saas
+          CI_NOTIFY_WEBHOOK: ${{ secrets.CI_NOTIFY_WEBHOOK }}
+          STALE_THRESHOLD_MIN: ${{ inputs.stale_threshold || 5 }}
+        run: |
+          set +e
+          python3 scripts/ci_trigger_monitor.py
+          # 监控脚本永远不fail，避免告警风暴
+          exit 0
@@ -0,0 +1,53 @@
+name: AI Code Review
+
+on:
+  pull_request:
+    types:
+      - opened
+      - synchronize
+      - reopened
+
+# 同一个 PR 只跑一个 review，新的取消旧的
+concurrency:
+  group: code-review-${{ gitea.repository }}-${{ gitea.event.pull_request.number }}
+  cancel-in-progress: true
+
+jobs:
+  code-review:
+    name: AI Code Review
+    runs-on: ubuntu-latest
+    # 跳过草稿 PR
+    if: ${{ !gitea.event.pull_request.draft }}
+
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v3
+        with:
+          fetch-depth: 0
+
+      - name: Install dependencies
+        run: |
+          python3 -m pip install --upgrade pip
+          python3 -m pip install requests
+
+      - name: Run AI Code Review
+        env:
+          # Gitea 配置（自动从运行环境获取）
+          GITEA_API_URL: ${{ gitea.server_url }}
+          GITEA_TOKEN: ${{ secrets.REVIEW_GITEA_TOKEN }}
+          REPO_NAME: ${{ gitea.repository }}
+          PR_NUMBER: ${{ gitea.event.pull_request.number }}
+          # LLM 提供商: coze (扣子原生Bot) / openai (OpenAI兼容)
+          LLM_PROVIDER: "coze"
+          # 扣子模式配置（默认国内站 api.coze.cn）
+          LLM_BASE_URL: ${{ secrets.LLM_BASE_URL }}
+          LLM_API_KEY: ${{ secrets.LLM_API_KEY }}
+          COZE_BOT_ID: ${{ secrets.COZE_BOT_ID }}
+          LLM_MODEL: ${{ secrets.LLM_MODEL }}
+          # 可选参数
+          MAX_DIFF_CHARS: "30000"
+          LLM_TIMEOUT: "120"
+        run: |
+          python3 scripts/ci_code_review.py
+        # 审查脚本异常不影响 CI 通过
+        continue-on-error: true
@@ -0,0 +1,118 @@
+name: Preview Cleanup
+on:
+  pull_request:
+    types:
+    - closed
+    branches:
+    - main
+    - develop
+permissions:
+  contents: read
+  pull-requests: write
+jobs:
+  cleanup-preview:
+    name: Cleanup Preview Environment
+    runs-on: runtime-builder
+    timeout-minutes: 10
+    steps:
+    - name: Extract PR number
+      shell: sh
+      run: |
+        set -eu
+        PR_NUMBER=$(echo "$GITHUB_REF" | sed 's|refs/pull/||; s|/.*||')
+        echo "PR_NUMBER=$PR_NUMBER" >> $GITHUB_ENV
+        echo "PR number: $PR_NUMBER"
+        echo "Preview dir: /var/www/preview/pr-${PR_NUMBER}"
+
+    - name: Install SSH client
+      shell: sh
+      run: |
+        set -eu
+        apt-get update -qq && apt-get install -y -qq openssh-client >/dev/null 2>&1
+        echo "openssh-client installed"
+
+    - name: Remove preview directory from server
+      shell: sh
+      env:
+        STAGING_SSH_HOST: ${{ secrets.STAGING_SSH_HOST }}
+        STAGING_SSH_USER: ${{ secrets.STAGING_SSH_USER }}
+        STAGING_SSH_PORT: ${{ secrets.STAGING_SSH_PORT }}
+        STAGING_SSH_KEY: ${{ secrets.STAGING_SSH_KEY }}
+      run: |
+        set -eux
+        staging_host="${STAGING_SSH_HOST:-47.98.113.167}"
+        staging_user="${STAGING_SSH_USER:-root}"
+        staging_port="${STAGING_SSH_PORT:-22222}"
+        preview_dir="/var/www/preview/pr-${PR_NUMBER}"
+
+        mkdir -p ~/.ssh
+
+        # 查找可用的SSH密钥
+        key_path=""
+        if [ -f /root/.ssh/xiaoxia_runtime_builder ]; then
+          key_path="/root/.ssh/xiaoxia_runtime_builder"
+          echo "Using key: $key_path (builder key)"
+        elif [ -f "$HOME/.ssh/xiaoxia_runtime_builder" ]; then
+          key_path="$HOME/.ssh/xiaoxia_runtime_builder"
+          echo "Using key: $key_path (home key)"
+        elif [ -n "${STAGING_SSH_KEY:-}" ]; then
+          key_path="$HOME/.ssh/id_ed25519"
+          printf '%s\n' "$STAGING_SSH_KEY" > "$key_path"
+          chmod 600 "$key_path"
+          echo "Using key from STAGING_SSH_KEY secret"
+        else
+          echo "ERROR: No SSH key available"
+          ls -la ~/.ssh/ 2>/dev/null || true
+          ls -la /root/.ssh/ 2>/dev/null || true
+          exit 1
+        fi
+
+        ssh-keyscan -p "$staging_port" -H "$staging_host" >> ~/.ssh/known_hosts 2>/dev/null
+        echo "SSH keyscan done"
+
+        # 测试SSH连接
+        ssh -p "$staging_port" -i "$key_path" -o StrictHostKeyChecking=no "${staging_user}@${staging_host}" "echo SSH_CONNECTION_OK && hostname"
+        echo "SSH connection verified"
+
+        # 检查目录是否存在
+        DIR_EXISTS=$(ssh -p "$staging_port" -i "$key_path" -o StrictHostKeyChecking=no "${staging_user}@${staging_host}" \
+          "if [ -d '${preview_dir}' ]; then echo 'yes'; else echo 'no'; fi")
+
+        if [ "$DIR_EXISTS" = "yes" ]; then
+          echo "Removing preview directory: ${preview_dir}"
+          ssh -p "$staging_port" -i "$key_path" -o StrictHostKeyChecking=no "${staging_user}@${staging_host}" \
+            "rm -rf ${preview_dir} && echo 'Preview directory removed successfully'"
+          echo "Cleanup completed: ${preview_dir}"
+        else
+          echo "Preview directory does not exist: ${preview_dir}, nothing to clean up"
+        fi
+
+    - name: Comment cleanup notice on PR
+      if: success()
+      shell: sh
+      env:
+        GITHUB_TOKEN: ${{ github.token }}
+      run: |
+        set -eu
+        PR_NUMBER=$(echo "$GITHUB_REF" | sed 's|refs/pull/||; s|/.*||')
+
+        COMMENT_BODY=$(python3 -c "
+        import json, os
+        pr_num = os.environ['PR_NUMBER']
+        body = f'''🗑️ **预览环境已清理**
+
+PR #{pr_num} 已关闭或合并，对应的预览环境已被清理。
+
+> 如有需要，可以重新打开 PR 来重新生成预览环境。
+'''
+        print(json.dumps({'body': body}))
+        ")
+
+        API_URL="${GITHUB_API_URL}/repos/${GITHUB_REPOSITORY}/issues/${PR_NUMBER}/comments"
+        curl -s -X POST \
+          -H "Authorization: token ${GITHUB_TOKEN}" \
+          -H "Content-Type: application/json" \
+          -d "$COMMENT_BODY" \
+          "$API_URL" \
+          > /dev/null
+        echo "Cleanup comment posted"
@@ -0,0 +1,293 @@
+name: Preview Deploy
+on:
+  pull_request:
+    types:
+    - opened
+    - synchronize
+    - reopened
+    branches:
+    - main
+    - develop
+  workflow_dispatch:
+    inputs:
+      reason:
+        description: "触发原因"
+        required: false
+        default: "手动触发 - 预览环境补跑"
+permissions:
+  contents: read
+  pull-requests: write
+concurrency:
+  group: preview-deploy-${{ gitea.ref }}
+  cancel-in-progress: true
+jobs:
+  deploy-preview:
+    name: Deploy Preview Environment
+    runs-on: runtime-builder
+    timeout-minutes: 20
+    steps:
+    - name: Checkout code
+      shell: sh
+      env:
+        GITHUB_TOKEN: ${{ github.token }}
+      run: |
+        set -eu
+        python3 - <<'PY'
+        import io, os, tarfile, time, urllib.request, urllib.error
+        url = f"{os.environ['GITHUB_API_URL']}/repos/{os.environ['GITHUB_REPOSITORY']}/archive/{os.environ['GITHUB_SHA']}.tar.gz"
+        request = urllib.request.Request(url, headers={"Authorization": f"token {os.environ['GITHUB_TOKEN']}"})
+        last_err = None
+        for attempt in range(5):
+            try:
+                with urllib.request.urlopen(request, timeout=120) as response:
+                    archive = response.read()
+                break
+            except urllib.error.HTTPError as e:
+                last_err = e
+                if e.code >= 500 and attempt < 4:
+                    wait = 2 ** attempt
+                    print(f"Checkout HTTP {e.code}, retrying in {wait}s (attempt {attempt+1}/5)...")
+                    time.sleep(wait)
+                    continue
+                raise
+            except Exception as e:
+                last_err = e
+                if attempt < 4:
+                    wait = 2 ** attempt
+                    print(f"Checkout error: {e}, retrying in {wait}s (attempt {attempt+1}/5)...")
+                    time.sleep(wait)
+                    continue
+                raise
+        else:
+            raise last_err
+        with tarfile.open(fileobj=io.BytesIO(archive), mode='r:gz') as tar:
+            root_prefix = tar.getmembers()[0].name.split('/', 1)[0] + '/'
+            for member in tar.getmembers():
+                name = member.name
+                if name == root_prefix[:-1]:
+                    continue
+                if name.startswith(root_prefix):
+                    member.name = name[len(root_prefix):]
+                if member.name:
+                    tar.extract(member, '.')
+        PY
+
+    - name: Record job start time
+      shell: sh
+      run: |
+        set -eu
+        echo "JOB_START_TIME=$(date +%s)" >> $GITHUB_ENV
+        echo "Job started at $(date)"
+
+    - name: Extract PR number
+      shell: sh
+      run: |
+        set -eu
+        PR_NUMBER=$(echo "$GITHUB_REF" | sed 's|refs/pull/||; s|/.*||')
+        echo "PR_NUMBER=$PR_NUMBER" >> $GITHUB_ENV
+        echo "PR number: $PR_NUMBER"
+        echo "PREVIEW_URL=https://pr-${PR_NUMBER}.preview.xiaoxiajianji.com" >> $GITHUB_ENV
+        echo "Preview URL: https://pr-${PR_NUMBER}.preview.xiaoxiajianji.com"
+
+    - name: Build frontend
+      shell: sh
+      run: |
+        set -eu
+        NPM_CACHE_VOLUME="xiaoxia-npm-cache"
+        if ! docker volume inspect "$NPM_CACHE_VOLUME" >/dev/null 2>&1; then
+          docker volume create "$NPM_CACHE_VOLUME" >/dev/null
+          echo "Created npm cache volume: $NPM_CACHE_VOLUME"
+        fi
+
+        docker run --rm \
+          -v "$PWD:/workspace" \
+          -v "$NPM_CACHE_VOLUME:/workspace/apps/web/node_modules" \
+          -w /workspace/apps/web \
+          -e VITE_API_URL=https://staging-api.xiaoxiajianji.com \
+          docker.m.daocloud.io/library/node:20 \
+          sh -lc '
+            PACKAGE_LOCK_HASH=$(md5sum package-lock.json 2>/dev/null | cut -d" " -f1)
+            CACHE_HASH_FILE="node_modules/.package-lock-hash"
+            CACHE_VALID=false
+            if [ -f "$CACHE_HASH_FILE" ] && [ "$(cat "$CACHE_HASH_FILE")" = "$PACKAGE_LOCK_HASH" ] && [ -x "node_modules/.bin/vite" ] && [ -x "node_modules/.bin/tsc" ]; then
+              CACHE_VALID=true
+              echo "Cache hit: dependencies valid, skipping npm ci"
+            fi
+            if [ "$CACHE_VALID" = "false" ]; then
+              echo "Cache miss or invalid: running npm ci..."
+              if ! npm ci --include=dev; then
+                echo "npm ci failed, cleaning node_modules and retrying..."
+                rm -rf node_modules
+                mkdir -p node_modules
+                npm ci --include=dev
+              fi
+              echo "$PACKAGE_LOCK_HASH" > "$CACHE_HASH_FILE"
+              echo "Dependencies installed, cache updated"
+            fi
+            echo "Running TypeScript check..."
+            npx --no-install tsc
+            echo "Running Vite build..."
+            npx --no-install vite build
+            echo "Build completed successfully"
+            ls -la dist/
+          '
+
+    - name: Install SSH client
+      shell: sh
+      run: |
+        set -eu
+        apt-get update -qq && apt-get install -y -qq openssh-client rsync >/dev/null 2>&1
+        echo "openssh-client and rsync installed"
+
+    - name: Deploy preview to server
+      shell: sh
+      env:
+        STAGING_SSH_HOST: ${{ secrets.STAGING_SSH_HOST }}
+        STAGING_SSH_USER: ${{ secrets.STAGING_SSH_USER }}
+        STAGING_SSH_PORT: ${{ secrets.STAGING_SSH_PORT }}
+        STAGING_SSH_KEY: ${{ secrets.STAGING_SSH_KEY }}
+      run: |
+        set -eux
+        staging_host="${STAGING_SSH_HOST:-47.98.113.167}"
+        staging_user="${STAGING_SSH_USER:-root}"
+        staging_port="${STAGING_SSH_PORT:-22222}"
+        preview_dir="/var/www/preview/pr-${PR_NUMBER}"
+
+        mkdir -p ~/.ssh
+
+        # 查找可用的SSH密钥
+        key_path=""
+        if [ -f /root/.ssh/xiaoxia_runtime_builder ]; then
+          key_path="/root/.ssh/xiaoxia_runtime_builder"
+          echo "Using key: $key_path (builder key)"
+        elif [ -f "$HOME/.ssh/xiaoxia_runtime_builder" ]; then
+          key_path="$HOME/.ssh/xiaoxia_runtime_builder"
+          echo "Using key: $key_path (home key)"
+        elif [ -n "${STAGING_SSH_KEY:-}" ]; then
+          key_path="$HOME/.ssh/id_ed25519"
+          printf '%s\n' "$STAGING_SSH_KEY" > "$key_path"
+          chmod 600 "$key_path"
+          echo "Using key from STAGING_SSH_KEY secret"
+        else
+          echo "ERROR: No SSH key available"
+          ls -la ~/.ssh/ 2>/dev/null || true
+          ls -la /root/.ssh/ 2>/dev/null || true
+          exit 1
+        fi
+
+        ssh-keyscan -p "$staging_port" -H "$staging_host" >> ~/.ssh/known_hosts 2>/dev/null
+        echo "SSH keyscan done"
+
+        # 测试SSH连接
+        ssh -p "$staging_port" -i "$key_path" -o StrictHostKeyChecking=no "${staging_user}@${staging_host}" "echo SSH_CONNECTION_OK && hostname"
+        echo "SSH connection verified"
+
+        # 创建预览目录并上传文件
+        ssh -p "$staging_port" -i "$key_path" -o StrictHostKeyChecking=no "${staging_user}@${staging_host}" \
+          "mkdir -p ${preview_dir} && echo 'Preview directory created: ${preview_dir}'"
+
+        # 使用rsync上传dist目录内容
+        rsync -avz --delete -e "ssh -p ${staging_port} -i ${key_path} -o StrictHostKeyChecking=no" \
+          apps/web/dist/ \
+          "${staging_user}@${staging_host}:${preview_dir}/"
+
+        echo "Preview deployed to: ${preview_dir}"
+        echo "Preview URL: https://pr-${PR_NUMBER}.preview.xiaoxiajianji.com"
+
+    - name: Comment preview link on PR
+      shell: sh
+      env:
+        GITHUB_TOKEN: ${{ github.token }}
+      run: |
+        set -eu
+        PR_NUMBER=$(echo "$GITHUB_REF" | sed 's|refs/pull/||; s|/.*||')
+        PREVIEW_URL="https://pr-${PR_NUMBER}.preview.xiaoxiajianji.com"
+
+        # 构建评论内容
+        COMMENT_BODY=$(python3 -c "
+        import json, os
+        pr_num = os.environ['PR_NUMBER']
+        url = os.environ['PREVIEW_URL']
+        body = f'''🚀 **预览环境已部署**
+
+| 项目 | 详情 |
+|------|------|
+| PR号 | #{pr_num} |
+| 预览链接 | [${url}](${url}) |
+| API环境 | staging |
+
+> 💡 预览环境使用 staging API 数据，请勿在预览环境中操作重要数据。
+>
+> 🔄 每次提交新代码后预览环境会自动更新。
+>
+> 🗑️ PR 关闭或合并后，预览环境会自动清理。
+'''
+        print(json.dumps({'body': body}))
+        ")
+
+        # 查找是否已有预览评论
+        API_URL="${GITHUB_API_URL}/repos/${GITHUB_REPOSITORY}/issues/${PR_NUMBER}/comments"
+        EXISTING_COMMENT_ID=""
+
+        COMMENTS=$(curl -s -H "Authorization: token ${GITHUB_TOKEN}" "$API_URL")
+        EXISTING_COMMENT_ID=$(echo "$COMMENTS" | python3 -c "
+        import sys, json
+        try:
+            comments = json.load(sys.stdin)
+            for c in comments:
+                body = c.get('body', '')
+                if '预览环境已部署' in body and 'Preview URL' not in body:
+                    print(c['id'])
+                    break
+        except:
+            print('')
+        ")
+
+        if [ -n "$EXISTING_COMMENT_ID" ]; then
+          # 更新已有评论
+          echo "Updating existing comment: $EXISTING_COMMENT_ID"
+          curl -s -X PATCH \
+            -H "Authorization: token ${GITHUB_TOKEN}" \
+            -H "Content-Type: application/json" \
+            -d "$COMMENT_BODY" \
+            "${GITHUB_API_URL}/repos/${GITHUB_REPOSITORY}/issues/comments/${EXISTING_COMMENT_ID}" \
+            > /dev/null
+          echo "Comment updated"
+        else
+          # 发布新评论
+          echo "Creating new comment"
+          curl -s -X POST \
+            -H "Authorization: token ${GITHUB_TOKEN}" \
+            -H "Content-Type: application/json" \
+            -d "$COMMENT_BODY" \
+            "$API_URL" \
+            > /dev/null
+          echo "Comment posted"
+        fi
+
+    - name: Job duration summary
+      if: always()
+      shell: sh
+      run: |
+        set +eu
+        if [ -n "$JOB_START_TIME" ]; then
+          END_TIME=$(date +%s)
+          DURATION=$((END_TIME - JOB_START_TIME))
+          MINS=$((DURATION / 60))
+          SECS=$((DURATION % 60))
+          echo "JOB_DURATION_SECONDS=$DURATION" >> $GITHUB_ENV
+          echo "=== Job Duration: ${MINS}m${SECS}s ==="
+        else
+          echo "JOB_DURATION_SECONDS=0" >> $GITHUB_ENV
+          echo "=== Job Duration: unknown ==="
+        fi
+
+    - name: Notify on failure
+      continue-on-error: true
+      if: failure()
+      shell: sh
+      env:
+        CI_NOTIFY_WEBHOOK: ${{ secrets.CI_NOTIFY_WEBHOOK }}
+      run: |
+        set +e
+        NOTIFY_MODE=failure JOB_NAME="Deploy Preview Environment" python3 scripts/ci_notify.py
@@ -0,0 +1,29 @@
+"""add result_count to edit_plans
+
+Revision ID: 041_result_count
+Revises: 040_playback_speed
+Create Date: 2026-07-15 14:05:00.000000
+"""
+
+from __future__ import annotations
+
+import sqlalchemy as sa
+
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "041_result_count"
+down_revision = "040_playback_speed"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.add_column(
+        "edit_plans",
+        sa.Column("result_count", sa.Integer(), nullable=False, server_default="0"),
+    )
+
+
+def downgrade() -> None:
+    op.drop_column("edit_plans", "result_count")
@@ -0,0 +1,29 @@
+"""add storage_key to assets
+
+Revision ID: 042_storage_key
+Revises: 041_result_count
+Create Date: 2026-07-17 18:10:00.000000
+"""
+
+from __future__ import annotations
+
+import sqlalchemy as sa
+
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "042_storage_key"
+down_revision = "041_result_count"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    op.add_column(
+        "assets",
+        sa.Column("storage_key", sa.String(500), nullable=False, server_default=""),
+    )
+
+
+def downgrade() -> None:
+    op.drop_column("assets", "storage_key")
@@ -46,3 +46,87 @@ def require_project_and_library(
    libraries = asset_library_repository.find_by_project(project_id)
    if not any(item.id == library_id for item in libraries):
        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Asset library not found")
+
+
+def auto_select_video_assets(
+    *,
+    project_id: str,
+    asset_library_repo: Any,
+    asset_repo: Any,
+    logger=None,
+) -> list[str]:
+    """从项目视频素材库自动选取 ready 状态的视频素材。
+
+    Args:
+        project_id: 项目 ID
+        asset_library_repo: 素材库仓储
+        asset_repo: 素材仓储
+        logger: 可选的 logger 实例，用于记录警告
+
+    Returns:
+        选中的素材 ID 列表，无可用素材时返回空列表
+    """
+    # 明确不支持的视频编码（会导致渲染失败）
+    UNSUPPORTED_CODECS = {"hevc", "h265", "hev1", "hvc1", "vp9", "vp09", "av1", "av01"}
+
+    if not project_id:
+        return []
+
+    # 找到项目的视频素材库
+    libs = asset_library_repo.find_by_project(project_id)
+    video_lib = None
+    for lib in libs:
+        lib_kind = lib.kind.value if hasattr(lib.kind, "value") else lib.kind
+        if lib_kind == "video":
+            video_lib = lib
+            break
+
+    if not video_lib:
+        if logger:
+            logger.warning("自动选素材: 项目 %s 无视频素材库", project_id)
+        return []
+
+    # 从素材库中选取 ready 状态的视频素材
+    assets = asset_repo.find_by_library(video_lib.id)
+    ready_videos = [
+        a
+        for a in assets
+        if (a.status.value if hasattr(a.status, "value") else a.status) == "ready"
+        and a.mime_type
+        and a.mime_type.startswith("video")
+    ]
+
+    # 过滤不支持的编码格式（HEVC/VP9/AV1 等会导致渲染失败）
+    # 优先读 asset.codec 字段，其次从 metadata 里取（兼容存量数据）
+    filtered_videos = []
+    skipped_codec = 0
+    skipped_landscape = 0
+    for a in ready_videos:
+        codec = (a.codec or "").lower()
+        if not codec and a.metadata and isinstance(a.metadata, dict):
+            codec = str(a.metadata.get("codec", "")).lower()
+        if codec and codec in UNSUPPORTED_CODECS:
+            skipped_codec += 1
+            continue
+
+        # 过滤横屏素材：竖屏APP（9:16输出）只选竖屏视频
+        # 宽 > 高 视为横屏，直接跳过；尺寸未知的存量素材放过
+        if a.width and a.height and a.width > a.height:
+            skipped_landscape += 1
+            continue
+
+        filtered_videos.append(a)
+
+    if skipped_codec and logger:
+        logger.warning("自动选素材: 跳过 %d 个不支持编码的素材", skipped_codec)
+    if skipped_landscape and logger:
+        logger.warning("自动选素材: 跳过 %d 个横屏素材", skipped_landscape)
+
+    if not filtered_videos:
+        if logger:
+            logger.warning("自动选素材: 素材库 %s 无可用视频素材", video_lib.name)
+        return []
+
+    # 按创建时间降序（新素材在前）
+    filtered_videos.sort(key=lambda a: a.created_at, reverse=True)
+    return [a.id for a in filtered_videos]
@@ -239,7 +239,9 @@ def get_duplication_detail(
    return _to_detail_response(record)


-@router.delete("/records/{record_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None, response_class=Response)
+@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),
@@ -287,7 +289,7 @@ def retry_duplication(
        raise HTTPException(
            status_code=status.HTTP_400_BAD_REQUEST,
            detail=str(e),
-        )
+        ) from e
    if updated is None:
        raise HTTPException(
            status_code=status.HTTP_404_NOT_FOUND,
@@ -22,9 +22,15 @@ from datetime import datetime
 from typing import Any, List, Optional

 from app.auth import AuthenticatedUser, get_current_user
-from app.dependencies import get_db_session, get_project_repository
+from app.core.storage import get_storage_service
+from app.dependencies import (
+    get_asset_library_repository,
+    get_asset_repository,
+    get_db_session,
+    get_project_repository,
+)
 from app.schemas.generation_task import GenerationTaskResponse
-from app.services import EditPlanService
+from app.services import EditPlanService, EditTemplateService
 from fastapi import APIRouter, Depends, HTTPException, Query, Response, status
 from pydantic import BaseModel, Field
 from sqlalchemy.orm import Session
@@ -32,7 +38,7 @@ from sqlalchemy.orm import Session
 from packages.domain.config_schemas import normalize_plan_config
 from packages.domain.edit_plan import EditPlan, EditPlanStatus

-from ._helpers import check_project_access
+from ._helpers import auto_select_video_assets, check_project_access

 logger = logging.getLogger(__name__)

@@ -50,6 +56,7 @@ class EditPlanCreateRequest(BaseModel):
    config: dict[str, Any] = Field(default_factory=dict, description="计划配置 (JSON)")
    total_duration: float = Field(default=0.0, ge=0.0, description="总时长 (秒)")
    project_id: str = Field(default="", description="所属项目 ID")
+    asset_ids: List[str] = Field(default_factory=list, description="素材 ID 列表，创建时自动分配给片段")


 class EditPlanUpdateRequest(BaseModel):
@@ -64,6 +71,15 @@ class EditPlanUpdateRequest(BaseModel):
    )


+class CopyPlanRequest(BaseModel):
+    """复制剪辑计划请求体"""
+
+    name: Optional[str] = Field(
+        default=None, min_length=1, max_length=200, description="新计划名称，不传则为「原名 - 副本」"
+    )
+    project_id: Optional[str] = Field(default=None, description="目标项目 ID，不传则复用源计划的项目")
+
+
 class EditPlanResponse(BaseModel):
    """剪辑计划响应体"""

@@ -72,6 +88,7 @@ class EditPlanResponse(BaseModel):
    name: str
    status: str
    total_duration: float
+    result_count: int = 0
    project_id: str = ""
    created_by_user_id: str = ""
    config: dict[str, Any]
@@ -108,6 +125,10 @@ class EditPlanGenerationStatusResponse(BaseModel):
    plan_id: str
    plan_status: str
    generation_task_id: Optional[str] = None
+    generation_task_status: Optional[str] = None
+    progress: float = 0.0
+    video_url: str = ""
+    error_message: str = ""
    clips: List[ClipStatusItem]


@@ -231,20 +252,45 @@ class GenerateFromTemplateResponse(BaseModel):


 def _to_response(p: EditPlan) -> EditPlanResponse:
+    # 对 config 中的 rendered_url 做签名转换（私有bucket裸URL会403）
+    config = dict(p.config) if p.config else {}
+    raw_rendered_url = config.get("rendered_url", "")
+    if raw_rendered_url:
+        try:
+            storage = get_storage_service()
+            config["rendered_url"] = storage.get_download_url(raw_rendered_url, expires_seconds=86400)
+        except Exception as e:
+            logger.warning("剪辑计划rendered_url签名失败，返回原始URL: plan_id=%s error=%s", p.id, e)
+
    return EditPlanResponse(
        id=p.id,
        template_id=p.template_id,
        name=p.name,
        status=p.status.value if hasattr(p.status, "value") else p.status,
        total_duration=p.total_duration,
+        result_count=getattr(p, "result_count", 0),
        project_id=p.project_id or "",
        created_by_user_id=p.created_by_user_id or "",
-        config=p.config,
+        config=config,
        created_at=p.created_at,
        updated_at=p.updated_at,
    )


+# ── Include sub-routers (拆分模块) ────────────────────────────────────────────
+# 注意：含静态路径的子路由需放在 CRUD 路由之前，避免被 /{plan_id} 抢先匹配
+
+from .edit_plans_adjustments import router as adjustments_router
+from .edit_plans_export import router as export_router
+from .edit_plans_filter import router as filter_router
+from .edit_plans_transitions import router as transitions_router
+
+router.include_router(export_router)
+router.include_router(adjustments_router)
+router.include_router(filter_router)
+router.include_router(transitions_router)
+
+
 # ── CRUD Routes ───────────────────────────────────────────────────────────────


@@ -275,11 +321,11 @@ def list_plans(
    if status_filter:
        try:
            status_enum = EditPlanStatus(status_filter)
-        except ValueError:
+        except ValueError as _e:
            raise HTTPException(
                status_code=status.HTTP_400_BAD_REQUEST,
                detail="无效的筛选条件，请选择正确的状态",
-            )
+            ) from _e

    # 项目鉴权：如果指定了 project_id，校验用户是否有权访问
    if project_id:
@@ -322,7 +368,7 @@ def get_plan(
        raise HTTPException(
            status_code=status.HTTP_404_NOT_FOUND,
            detail=str(exc),
-        )
+        ) from exc
    # 项目鉴权
    if plan.project_id:
        check_project_access(plan.project_id, current_user.user.id, project_repository)
@@ -335,37 +381,121 @@ def create_plan(
    db: Session = Depends(get_db_session),
    current_user: AuthenticatedUser = Depends(get_current_user),
    project_repository: Any = Depends(get_project_repository),
+    asset_repository: Any = Depends(get_asset_repository),
+    asset_library_repository: Any = Depends(get_asset_library_repository),
 ) -> EditPlanResponse:
-    """创建剪辑计划"""
+    """创建剪辑计划
+
+    基于模板自动生成片段结构：
+    - 模板存在时：从模板的 clip_configs 生成初始 clips
+    - 模板不存在时：降级为空计划（保持向后兼容）
+    - 用户传入的 config 与模板 config 合并（用户配置优先级更高）
+    - total_duration 自动根据 clips 总时长计算
+    """
+    from app.services import PlanGeneratorService
+
    # 空串 project_id 统一为 ""
    project_id = (body.project_id or "").strip()
    # 项目鉴权
    if project_id:
        check_project_access(project_id, current_user.user.id, project_repository)
+
+    # 标准化用户传入的 config
+    normalized_config = normalize_plan_config(body.config or {})
+
+    template_svc = EditTemplateService(db)
    svc = EditPlanService(db)
-    # 标准化 config，填充 cover/title/subtitle/bgm 默认值
-    normalized_config = normalize_plan_config(body.config)
+
+    # 尝试从模板生成（模板不存在时降级为空计划）
+    template = None
+    clips = []
    try:
-        created = svc.create_plan(
+        template = template_svc.get_template_or_raise(body.template_id)
+    except ValueError:
+        # 模板不存在，降级为普通空计划
+        logger.info("模板不存在，创建空计划: template_id=%s", body.template_id)
+        plan = svc.create_plan(
            template_id=body.template_id,
            name=body.name,
            config=normalized_config,
-            total_duration=body.total_duration,
            project_id=project_id,
            created_by_user_id=current_user.user.id,
+            total_duration=body.total_duration if body.total_duration > 0 else 0.0,
+        )
+        logger.info(
+            "创建空剪辑计划: id=%s name=%s by user=%s",
+            plan.id,
+            plan.name,
+            current_user.user.id,
+        )
+        return _to_response(plan)
+
+    # 模板存在，从模板生成计划+片段
+    clip_configs = template_svc.list_clip_configs(body.template_id, skip=0, limit=200)
+
+    # 自动选素材：未传 asset_ids 但有 project_id 时，从项目视频素材库选 ready 的视频素材
+    resolved_asset_ids = list(body.asset_ids)
+    if not resolved_asset_ids and project_id:
+        auto_assets = auto_select_video_assets(
+            project_id=project_id,
+            asset_library_repo=asset_library_repository,
+            asset_repo=asset_repository,
+            logger=logger,
+        )
+        if auto_assets:
+            resolved_asset_ids = auto_assets
+            logger.info(
+                "自动选素材: project_id=%s count=%d",
+                project_id,
+                len(auto_assets),
+            )
+
+    generator = PlanGeneratorService(db)
+    try:
+        result = generator.generate_from_template(
+            template=template,
+            clip_configs=clip_configs,
+            asset_ids=resolved_asset_ids,
+            project_id=project_id,
+            created_by_user_id=current_user.user.id,
+            name=body.name,
        )
    except ValueError as exc:
        raise HTTPException(
            status_code=status.HTTP_400_BAD_REQUEST,
            detail=str(exc),
+        ) from exc
+
+    plan = result["plan"]
+    clips = result["clips"]
+
+    # 如果用户传入了自定义 config，合并覆盖模板配置
+    if body.config:
+        base_config = template.config or {}
+        merged_config = {**base_config, **normalized_config}
+        # 重新标准化确保默认值填充正确
+        merged_config = normalize_plan_config(merged_config)
+        plan = svc.update_plan(
+            plan.id,
+            config=merged_config,
+            total_duration=body.total_duration if body.total_duration > 0 else None,
        )
+
+    # 把 asset_ids 写入 plan.config，供生成时兜底分配使用
+    if resolved_asset_ids:
+        current_config = plan.config or {}
+        if current_config.get("asset_ids") != resolved_asset_ids:
+            current_config["asset_ids"] = resolved_asset_ids
+            plan = svc.update_plan(plan.id, config=normalize_plan_config(current_config))
+
    logger.info(
-        "创建剪辑计划: id=%s name=%s by user=%s",
-        created.id,
-        created.name,
+        "创建剪辑计划: id=%s name=%s clips=%d by user=%s",
+        plan.id,
+        plan.name,
+        len(clips),
        current_user.user.id,
    )
-    return _to_response(created)
+    return _to_response(plan)


@router.put("/{plan_id}", response_model=EditPlanResponse)
@@ -401,11 +531,11 @@ def update_plan(
        if body.status is not None:
            try:
                target_status = EditPlanStatus(body.status)
-            except ValueError:
+            except ValueError as _e:
                raise HTTPException(
                    status_code=status.HTTP_400_BAD_REQUEST,
                    detail="无效的状态值，请选择正确的状态",
-                )
+                ) from _e
            svc.transition_status(plan_id, target_status)
    except ValueError as exc:
        err_msg = str(exc)
@@ -413,11 +543,11 @@ def update_plan(
            raise HTTPException(
                status_code=status.HTTP_404_NOT_FOUND,
                detail=err_msg,
-            )
+            ) from exc
        raise HTTPException(
            status_code=status.HTTP_400_BAD_REQUEST,
            detail=err_msg,
-        )
+        ) from exc

    # 返回最新状态
    result = svc.get_plan_or_raise(plan_id)
@@ -451,12 +581,551 @@ def delete_plan(
    )


+@router.post("/{plan_id}/copy", response_model=EditPlanResponse, status_code=status.HTTP_201_CREATED)
+def copy_plan(
+    plan_id: str,
+    body: CopyPlanRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> EditPlanResponse:
+    """复制剪辑计划（含所有片段配置）
+
+    新计划状态为 editing，不含生成任务和结果记录。
+    """
+    svc = EditPlanService(db)
+
+    # 源计划鉴权
+    existing = svc.get_plan(plan_id)
+    if existing is None:
+        raise HTTPException(status_code=404, detail=f"剪辑计划不存在: {plan_id}")
+    if existing.project_id:
+        check_project_access(existing.project_id, current_user.user.id, project_repository)
+
+    # 目标项目鉴权（如果指定了不同的项目）
+    target_project_id = body.project_id if body.project_id is not None else existing.project_id
+    if target_project_id and target_project_id != existing.project_id:
+        check_project_access(target_project_id, current_user.user.id, project_repository)
+
+    try:
+        new_plan = svc.copy_plan(
+            plan_id,
+            new_name=body.name,
+            project_id=target_project_id,
+        )
+    except ValueError as e:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=str(e)) from e
+
+    logger.info(
+        "复制剪辑计划: source=%s target=%s by user=%s",
+        plan_id,
+        new_plan.id,
+        current_user.user.id,
+    )
+    return _to_response(new_plan)
+
+
+# ── 字幕管理 ────────────────────────────────────────────────────────────────
+
+
+class SubtitleCreateRequest(BaseModel):
+    """添加字幕请求体"""
+
+    start: float = Field(..., ge=0, description="开始时间（秒）")
+    end: float = Field(..., gt=0, description="结束时间（秒）")
+    text: str = Field(..., min_length=1, max_length=500, description="字幕文本")
+    style: Optional[dict[str, Any]] = Field(default=None, description="字幕样式")
+
+
+class SubtitleUpdateRequest(BaseModel):
+    """更新字幕请求体"""
+
+    start: Optional[float] = Field(default=None, ge=0, description="开始时间（秒）")
+    end: Optional[float] = Field(default=None, gt=0, description="结束时间（秒）")
+    text: Optional[str] = Field(default=None, min_length=1, max_length=500, description="字幕文本")
+    style: Optional[dict[str, Any]] = Field(default=None, description="字幕样式")
+
+
+class SubtitleBatchUpdateRequest(BaseModel):
+    """批量更新字幕请求体"""
+
+    subtitles: list[dict[str, Any]] = Field(
+        default_factory=list,
+        description="字幕列表（全量替换），每条包含 start/end/text，可选 id/style",
+    )
+
+
+@router.get(
+    "/clips/{clip_id}/subtitles",
+    response_model=list[dict[str, Any]],
+    summary="获取片段的所有字幕",
+)
+def list_subtitles(
+    clip_id: str,
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    db: Session = Depends(get_db_session),
+    project_repo=Depends(get_project_repository),
+) -> list[dict[str, Any]]:
+    """获取指定片段的所有字幕，按时间排序。"""
+    service = EditPlanService(db)
+
+    clip = service.get_clip(clip_id)
+    if clip is None:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"片段不存在: {clip_id}",
+        )
+    plan = service.get_plan(clip.plan_id)
+    if plan and plan.project_id:
+        check_project_access(project_repo, current_user, plan.project_id)
+
+    return service.list_subtitles(clip_id)
+
+
+@router.post(
+    "/clips/{clip_id}/subtitles",
+    response_model=dict[str, Any],
+    summary="添加一条字幕",
+    status_code=status.HTTP_201_CREATED,
+)
+def add_subtitle(
+    clip_id: str,
+    body: SubtitleCreateRequest,
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    db: Session = Depends(get_db_session),
+    project_repo=Depends(get_project_repository),
+) -> dict[str, Any]:
+    """给片段添加一条字幕。"""
+    service = EditPlanService(db)
+
+    clip = service.get_clip(clip_id)
+    if clip is None:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"片段不存在: {clip_id}",
+        )
+    plan = service.get_plan(clip.plan_id)
+    if plan and plan.project_id:
+        check_project_access(project_repo, current_user, plan.project_id)
+
+    try:
+        subtitle = service.add_subtitle(
+            clip_id,
+            start=body.start,
+            end=body.end,
+            text=body.text,
+            style=body.style,
+        )
+    except ValueError as e:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=str(e),
+        ) from e
+
+    logger.info("添加字幕: clip_id=%s by user=%s", clip_id, current_user.user.id)
+    return subtitle
+
+
+@router.put(
+    "/clips/{clip_id}/subtitles/{subtitle_id}",
+    response_model=dict[str, Any],
+    summary="更新一条字幕",
+)
+def update_subtitle(
+    clip_id: str,
+    subtitle_id: str,
+    body: SubtitleUpdateRequest,
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    db: Session = Depends(get_db_session),
+    project_repo=Depends(get_project_repository),
+) -> dict[str, Any]:
+    """更新一条字幕的时间、文本或样式。"""
+    service = EditPlanService(db)
+
+    clip = service.get_clip(clip_id)
+    if clip is None:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"片段不存在: {clip_id}",
+        )
+    plan = service.get_plan(clip.plan_id)
+    if plan and plan.project_id:
+        check_project_access(project_repo, current_user, plan.project_id)
+
+    try:
+        subtitle = service.update_subtitle(
+            clip_id,
+            subtitle_id,
+            start=body.start,
+            end=body.end,
+            text=body.text,
+            style=body.style,
+        )
+    except ValueError as e:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=str(e),
+        ) from e
+
+    logger.info(
+        "更新字幕: clip_id=%s subtitle_id=%s by user=%s",
+        clip_id,
+        subtitle_id,
+        current_user.user.id,
+    )
+    return subtitle
+
+
+@router.delete(
+    "/clips/{clip_id}/subtitles/{subtitle_id}",
+    summary="删除一条字幕",
+    status_code=status.HTTP_204_NO_CONTENT,
+    response_model=None,
+    response_class=Response,
+)
+def delete_subtitle(
+    clip_id: str,
+    subtitle_id: str,
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    db: Session = Depends(get_db_session),
+    project_repo=Depends(get_project_repository),
+) -> None:
+    """删除一条字幕。"""
+    service = EditPlanService(db)
+
+    clip = service.get_clip(clip_id)
+    if clip is None:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"片段不存在: {clip_id}",
+        )
+    plan = service.get_plan(clip.plan_id)
+    if plan and plan.project_id:
+        check_project_access(project_repo, current_user, plan.project_id)
+
+    deleted = service.delete_subtitle(clip_id, subtitle_id)
+    if not deleted:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"字幕不存在: {subtitle_id}",
+        )
+
+    logger.info(
+        "删除字幕: clip_id=%s subtitle_id=%s by user=%s",
+        clip_id,
+        subtitle_id,
+        current_user.user.id,
+    )
+
+
+@router.put(
+    "/clips/{clip_id}/subtitles",
+    response_model=list[dict[str, Any]],
+    summary="批量更新字幕（全量替换）",
+)
+def batch_update_subtitles(
+    clip_id: str,
+    body: SubtitleBatchUpdateRequest,
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    db: Session = Depends(get_db_session),
+    project_repo=Depends(get_project_repository),
+) -> list[dict[str, Any]]:
+    """批量更新片段的所有字幕（全量替换）。
+
+    用于批量编辑、SRT导入、ASR结果导入等场景。
+    每条字幕包含 start/end/text，已有 id 则保留，否则生成新 id。
+    """
+    service = EditPlanService(db)
+
+    clip = service.get_clip(clip_id)
+    if clip is None:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"片段不存在: {clip_id}",
+        )
+    plan = service.get_plan(clip.plan_id)
+    if plan and plan.project_id:
+        check_project_access(project_repo, current_user, plan.project_id)
+
+    try:
+        subtitles = service.batch_update_subtitles(clip_id, body.subtitles)
+    except ValueError as e:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=str(e),
+        ) from e
+
+    logger.info(
+        "批量更新字幕: clip_id=%s count=%d by user=%s",
+        clip_id,
+        len(subtitles),
+        current_user.user.id,
+    )
+    return subtitles
+
+
+# ── BGM 背景音乐 ───────────────────────────────────────────────────────────
+
+
+class BGMConfigUpdateRequest(BaseModel):
+    """更新BGM配置请求体"""
+
+    enabled: Optional[bool] = Field(default=None, description="是否启用 BGM")
+    source: Optional[str] = Field(default=None, description="BGM 来源: library/upload/ai_recommend")
+    asset_id: Optional[str] = Field(default=None, max_length=64, description="BGM 素材 ID")
+    preset_id: Optional[str] = Field(default=None, max_length=64, description="预设 BGM ID")
+    audio_url: Optional[str] = Field(default=None, max_length=500, description="BGM 音频 URL")
+    volume: Optional[float] = Field(default=None, ge=0.0, le=1.0, description="音量 (0.0 ~ 1.0)")
+    fade_in: Optional[float] = Field(default=None, ge=0.0, le=30.0, description="淡入时长（秒）")
+    fade_out: Optional[float] = Field(default=None, ge=0.0, le=30.0, description="淡出时长（秒）")
+    loop_enabled: Optional[bool] = Field(default=None, description="是否循环播放")
+    sidechain_enabled: Optional[bool] = Field(default=None, description="是否启用人声闪避")
+    sidechain_ratio: Optional[float] = Field(default=None, ge=0.0, le=1.0, description="闪避音量降低比例")
+
+
+@router.get(
+    "/{plan_id}/bgm",
+    response_model=dict[str, Any],
+    summary="获取剪辑计划的 BGM 配置",
+)
+def get_plan_bgm(
+    plan_id: str,
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    db: Session = Depends(get_db_session),
+    project_repo=Depends(get_project_repository),
+) -> dict[str, Any]:
+    """获取指定剪辑计划的 BGM 配置。"""
+    service = EditPlanService(db)
+
+    plan = service.get_plan(plan_id)
+    if plan is None:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"剪辑计划不存在: {plan_id}",
+        )
+    if plan.project_id:
+        check_project_access(project_repo, current_user, plan.project_id)
+
+    config = plan.config or {}
+    bgm_config = config.get("bgm", {})
+
+    return {
+        "plan_id": plan.id,
+        "bgm": bgm_config,
+    }
+
+
+@router.put(
+    "/{plan_id}/bgm",
+    response_model=dict[str, Any],
+    summary="更新剪辑计划的 BGM 配置",
+)
+def update_plan_bgm(
+    plan_id: str,
+    body: BGMConfigUpdateRequest,
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    db: Session = Depends(get_db_session),
+    project_repo=Depends(get_project_repository),
+) -> dict[str, Any]:
+    """更新剪辑计划的 BGM 配置。
+
+    支持部分更新，只传需要修改的字段即可。
+    启用 BGM 后需要指定来源（asset_id / preset_id / audio_url 三选一）。
+    """
+    service = EditPlanService(db)
+
+    plan = service.get_plan(plan_id)
+    if plan is None:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"剪辑计划不存在: {plan_id}",
+        )
+    if plan.project_id:
+        check_project_access(project_repo, current_user, plan.project_id)
+
+    # 读取当前 BGM 配置，合并更新
+    config = dict(plan.config) if plan.config else {}
+    current_bgm = dict(config.get("bgm", {}))
+
+    update_data = body.model_dump(exclude_none=True)
+    current_bgm.update(update_data)
+
+    # 校验：启用 BGM 时至少有一个有效来源
+    if current_bgm.get("enabled"):
+        has_source = any(current_bgm.get(key) for key in ("asset_id", "preset_id", "audio_url") if current_bgm.get(key))
+        if not has_source:
+            raise HTTPException(
+                status_code=status.HTTP_400_BAD_REQUEST,
+                detail="启用 BGM 时需要指定素材来源（asset_id / preset_id / audio_url）",
+            )
+
+    # 保存到 plan.config.bgm
+    config["bgm"] = current_bgm
+    updated_plan = service.update_plan_config(plan_id, config)
+
+    logger.info(
+        "更新BGM配置: plan_id=%s enabled=%s by user=%s",
+        plan_id,
+        current_bgm.get("enabled", False),
+        current_user.user.id,
+    )
+
+    return {
+        "plan_id": updated_plan.id,
+        "bgm": current_bgm,
+    }
+
+
+# ── BGM 预设库 ────────────────────────────────────────────────────────────────
+
+
+@router.get(
+    "/bgm/presets",
+    response_model=dict[str, Any],
+    summary="获取预设 BGM 列表",
+)
+def list_bgm_presets(
+    style: Optional[str] = Query(default=None, description="按风格筛选"),
+    keyword: Optional[str] = Query(default=None, description="关键词搜索"),
+    skip: int = Query(default=0, ge=0, description="分页偏移"),
+    limit: int = Query(default=50, ge=1, le=200, description="每页数量"),
+) -> dict[str, Any]:
+    """获取预设 BGM 列表，支持按风格筛选和关键词搜索。
+
+    风格可选: upbeat（轻快）、relax（治愈）、tech（科技）、commerce（电商）、
+    emotional（情感）、cinematic（电影）
+    """
+    from packages.domain.preset_bgm import (
+        BGM_STYLES,
+        PRESET_BGM_LIBRARY,
+        list_preset_bgm_by_style,
+        search_preset_bgm,
+    )
+
+    bgm_list = PRESET_BGM_LIBRARY
+
+    if keyword:
+        bgm_list = search_preset_bgm(keyword)
+    elif style:
+        bgm_list = list_preset_bgm_by_style(style)
+
+    total = len(bgm_list)
+    paged = bgm_list[skip : skip + limit]
+
+    return {
+        "total": total,
+        "skip": skip,
+        "limit": limit,
+        "styles": BGM_STYLES,
+        "items": [
+            {
+                "id": bgm.id,
+                "name": bgm.name,
+                "style": bgm.style,
+                "style_label": BGM_STYLES.get(bgm.style, bgm.style),
+                "duration": bgm.duration,
+                "artist": bgm.artist,
+                "description": bgm.description,
+                "tags": bgm.tags,
+                "audio_url": bgm.audio_url,
+            }
+            for bgm in paged
+        ],
+    }
+
+
+# ── 保存为模板 ────────────────────────────────────────────────────────────────
+
+
+class SaveAsTemplateRequest(BaseModel):
+    """保存为模板请求体"""
+
+    name: str = Field(..., min_length=1, max_length=200, description="模板名称")
+    description: str = Field(default="", max_length=500, description="模板描述")
+    template_type: str = Field(default="custom", max_length=50, description="模板类型")
+    preview_url: str = Field(default="", max_length=500, description="预览图 URL")
+
+
+@router.post(
+    "/{plan_id}/save-as-template",
+    response_model=dict[str, Any],
+    summary="将剪辑计划保存为模板",
+    status_code=status.HTTP_201_CREATED,
+)
+def save_plan_as_template(
+    plan_id: str,
+    body: SaveAsTemplateRequest,
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    db: Session = Depends(get_db_session),
+    project_repo=Depends(get_project_repository),
+) -> dict[str, Any]:
+    """将指定剪辑计划的配置和片段结构保存为一个新模板。
+
+    新模板会复制计划的所有片段配置（类型、时长、转场、文案等），
+    但不绑定具体素材，可重复用于创建新的剪辑计划。
+    """
+    # 校验计划存在性和项目权限
+    plan_service = EditPlanService(db)
+    plan = plan_service.get_plan(plan_id)
+    if plan is None:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"剪辑计划不存在: {plan_id}",
+        )
+    if plan.project_id:
+        check_project_access(project_repo, current_user, plan.project_id)
+
+    template_service = EditTemplateService(db)
+    try:
+        result = template_service.save_plan_as_template(
+            plan_id=plan_id,
+            name=body.name,
+            description=body.description,
+            template_type=body.template_type,
+            preview_url=body.preview_url,
+        )
+    except ValueError as e:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=str(e),
+        ) from e
+
+    template = result["template"]
+    clip_configs = result["clip_configs"]
+
+    logger.info(
+        "保存计划为模板: plan_id=%s template_id=%s name=%s by user=%s",
+        plan_id,
+        template.id,
+        body.name,
+        current_user.user.id,
+    )
+
+    return {
+        "id": template.id,
+        "name": template.name,
+        "description": template.description,
+        "template_type": template.template_type,
+        "editing_mode": template.editing_mode,
+        "preview_url": template.preview_url,
+        "status": template.status.value,
+        "clip_count": len(clip_configs),
+        "created_at": template.created_at.isoformat(),
+    }
+
+
 # ── Include sub-routers (拆分模块) ────────────────────────────────────────────

 from .edit_plans_ai import router as ai_router
+from .edit_plans_clips import router as clips_router
+from .edit_plans_clips_batch import router as clips_batch_router
+from .edit_plans_cover import router as cover_router
 from .edit_plans_generation import router as generation_router
 from .edit_plans_timeline import router as timeline_router

 router.include_router(generation_router)
 router.include_router(ai_router)
 router.include_router(timeline_router)
+router.include_router(clips_router, prefix="/{plan_id}/clips", tags=["EditPlan Clips"])
+router.include_router(clips_batch_router, prefix="/{plan_id}/clips", tags=["EditPlan Clips"])
+router.include_router(cover_router)
@@ -0,0 +1,311 @@
+"""片段调整 API.
+
+- PUT    /clips/{clip_id}/speed         调速
+- PUT    /clips/{clip_id}/volume        音量调节
+- PUT    /clips/{clip_id}/trim          裁剪（trim in/out）
+- PUT    /clips/{clip_id}/adjustments   统一调整（speed+volume+trim）
+- POST   /{plan_id}/clips/batch-speed   批量调速
+"""
+
+from __future__ import annotations
+
+import logging
+from typing import Any, Optional
+
+from app.auth import AuthenticatedUser, get_current_user
+from app.dependencies import get_db_session, get_project_repository
+from app.services import EditPlanService
+from fastapi import APIRouter, Depends, HTTPException, status
+from pydantic import BaseModel, Field
+from sqlalchemy.orm import Session
+
+from ._helpers import check_project_access
+
+logger = logging.getLogger(__name__)
+
+router = APIRouter()
+
+
+# ── Schemas ──────────────────────────────────────────────────────────────────
+
+
+class SpeedAdjustRequest(BaseModel):
+    """调速请求"""
+
+    speed: float = Field(..., ge=0.25, le=4.0, description="播放速度 0.25~4.0")
+
+
+class VolumeAdjustRequest(BaseModel):
+    """音量调节请求"""
+
+    volume: float = Field(..., ge=0.0, le=2.0, description="音量倍率 0~2.0（1.0=原音量）")
+
+
+class TrimAdjustRequest(BaseModel):
+    """裁剪请求"""
+
+    trim_start: float = Field(0.0, ge=0.0, description="开头裁剪秒数")
+    trim_end: float = Field(0.0, ge=0.0, description="结尾裁剪秒数")
+
+
+class ClipAdjustmentsRequest(BaseModel):
+    """统一调整请求"""
+
+    speed: Optional[float] = Field(default=None, ge=0.25, le=4.0)
+    volume: Optional[float] = Field(default=None, ge=0.0, le=2.0)
+    trim_start: Optional[float] = Field(default=None, ge=0.0)
+    trim_end: Optional[float] = Field(default=None, ge=0.0)
+
+
+class BatchSpeedRequest(BaseModel):
+    """批量调速请求"""
+
+    speed: float = Field(..., ge=0.25, le=4.0, description="播放速度")
+
+
+class ClipAdjustResponse(BaseModel):
+    """片段调整响应"""
+
+    clip_id: str
+    speed: float
+    volume: float
+    trim_start: float
+    trim_end: float
+    duration: float
+
+
+class BatchSpeedResponse(BaseModel):
+    """批量调速响应"""
+
+    updated_count: int
+    plan_id: str
+
+
+# ── Helpers ──────────────────────────────────────────────────────────────────
+
+
+def _get_clip_config(clip) -> dict:
+    config = getattr(clip, "config", {}) or {}
+    if not isinstance(config, dict):
+        config = {}
+    return config
+
+
+def _get_volume(clip) -> float:
+    config = _get_clip_config(clip)
+    return float(config.get("volume", 1.0))
+
+
+def _get_trim(clip) -> tuple[float, float]:
+    config = _get_clip_config(clip)
+    trim_start = float(config.get("trim_start", 0.0))
+    trim_end = float(config.get("trim_end", 0.0))
+    return trim_start, trim_end
+
+
+def _build_response(clip) -> ClipAdjustResponse:
+    trim_start, trim_end = _get_trim(clip)
+    return ClipAdjustResponse(
+        clip_id=clip.id,
+        speed=clip.playback_speed,
+        volume=_get_volume(clip),
+        trim_start=trim_start,
+        trim_end=trim_end,
+        duration=clip.duration,
+    )
+
+
+def _validate_trim(trim_start: float, trim_end: float, total_duration: float) -> None:
+    """验证裁剪时长不超过总时长"""
+    if trim_start + trim_end >= total_duration:
+        raise ValueError(f"裁剪总时长（{trim_start + trim_end:.2f}s）不能大于等于片段总时长（{total_duration:.2f}s）")
+
+
+# ── Routes ───────────────────────────────────────────────────────────────────
+
+
+def _get_svc_and_plan_and_clip(db, clip_id, current_user, project_repository):
+    svc = EditPlanService(db)
+    clip = svc.get_clip(clip_id)
+    if not clip:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"片段不存在: {clip_id}",
+        )
+    plan = svc.get_plan(clip.plan_id)
+    if plan and plan.project_id:
+        check_project_access(plan.project_id, current_user.user.id, project_repository)
+    return svc, plan, clip
+
+
+@router.put("/clips/{clip_id}/speed", response_model=ClipAdjustResponse)
+def adjust_speed(
+    clip_id: str,
+    body: SpeedAdjustRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> ClipAdjustResponse:
+    """调整片段播放速度"""
+    svc, _, clip = _get_svc_and_plan_and_clip(db, clip_id, current_user, project_repository)
+
+    updated = svc.update_clip(clip_id, playback_speed=body.speed)
+
+    logger.info(
+        "调整片段速度: clip_id=%s speed=%.2f by user=%s",
+        clip_id,
+        body.speed,
+        current_user.user.id,
+    )
+    return _build_response(updated)
+
+
+@router.put("/clips/{clip_id}/volume", response_model=ClipAdjustResponse)
+def adjust_volume(
+    clip_id: str,
+    body: VolumeAdjustRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> ClipAdjustResponse:
+    """调整片段音量"""
+    svc, _, clip = _get_svc_and_plan_and_clip(db, clip_id, current_user, project_repository)
+
+    # 更新 config.volume
+    config = dict(_get_clip_config(clip))
+    config["volume"] = body.volume
+    updated = svc.update_clip(clip_id, config=config)
+
+    logger.info(
+        "调整片段音量: clip_id=%s volume=%.2f by user=%s",
+        clip_id,
+        body.volume,
+        current_user.user.id,
+    )
+    return _build_response(updated)
+
+
+@router.put("/clips/{clip_id}/trim", response_model=ClipAdjustResponse)
+def adjust_trim(
+    clip_id: str,
+    body: TrimAdjustRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> ClipAdjustResponse:
+    """裁剪片段（trim in/out）"""
+    svc, _, clip = _get_svc_and_plan_and_clip(db, clip_id, current_user, project_repository)
+
+    # 验证裁剪时长
+    try:
+        _validate_trim(body.trim_start, body.trim_end, clip.duration)
+    except ValueError as e:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=str(e)) from e
+
+    # 更新 config
+    config = dict(_get_clip_config(clip))
+    config["trim_start"] = body.trim_start
+    config["trim_end"] = body.trim_end
+    updated = svc.update_clip(clip_id, config=config)
+
+    logger.info(
+        "裁剪片段: clip_id=%s trim_start=%.2f trim_end=%.2f by user=%s",
+        clip_id,
+        body.trim_start,
+        body.trim_end,
+        current_user.user.id,
+    )
+    return _build_response(updated)
+
+
+@router.put("/clips/{clip_id}/adjustments", response_model=ClipAdjustResponse)
+def adjust_all(
+    clip_id: str,
+    body: ClipAdjustmentsRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> ClipAdjustResponse:
+    """统一调整片段的 speed / volume / trim"""
+    svc, _, clip = _get_svc_and_plan_and_clip(db, clip_id, current_user, project_repository)
+
+    update_kwargs = {}
+    config_updates = {}
+
+    if body.speed is not None:
+        update_kwargs["playback_speed"] = body.speed
+
+    if body.volume is not None:
+        config_updates["volume"] = body.volume
+
+    if body.trim_start is not None:
+        config_updates["trim_start"] = body.trim_start
+
+    if body.trim_end is not None:
+        config_updates["trim_end"] = body.trim_end
+
+    # 验证 trim
+    current_trim_start, current_trim_end = _get_trim(clip)
+    new_trim_start = body.trim_start if body.trim_start is not None else current_trim_start
+    new_trim_end = body.trim_end if body.trim_end is not None else current_trim_end
+
+    if body.trim_start is not None or body.trim_end is not None:
+        try:
+            _validate_trim(new_trim_start, new_trim_end, clip.duration)
+        except ValueError as e:
+            raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=str(e)) from e
+
+    if config_updates:
+        config = dict(_get_clip_config(clip))
+        config.update(config_updates)
+        update_kwargs["config"] = config
+
+    if not update_kwargs:
+        return _build_response(clip)
+
+    updated = svc.update_clip(clip_id, **update_kwargs)
+
+    logger.info(
+        "统一调整片段: clip_id=%s speed=%s volume=%s by user=%s",
+        clip_id,
+        body.speed,
+        body.volume,
+        current_user.user.id,
+    )
+    return _build_response(updated)
+
+
+@router.post("/{plan_id}/clips/batch-speed", response_model=BatchSpeedResponse)
+def batch_adjust_speed(
+    plan_id: str,
+    body: BatchSpeedRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> BatchSpeedResponse:
+    """批量调整计划内所有片段的播放速度"""
+    svc = EditPlanService(db)
+    plan = svc.get_plan(plan_id)
+    if not plan:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"剪辑计划不存在: {plan_id}",
+        )
+
+    if plan.project_id:
+        check_project_access(plan.project_id, current_user.user.id, project_repository)
+
+    clips = svc.list_clips(plan_id, limit=500, skip=0)
+    count = 0
+    for clip in clips:
+        svc.update_clip(clip.id, playback_speed=body.speed)
+        count += 1
+
+    logger.info(
+        "批量调速: plan_id=%s count=%d speed=%.2f by user=%s",
+        plan_id,
+        count,
+        body.speed,
+        current_user.user.id,
+    )
+    return BatchSpeedResponse(updated_count=count, plan_id=plan_id)
@@ -58,7 +58,7 @@ def ai_recommend_clips(
    try:
        plan = svc.get_plan_or_raise(plan_id)
    except ValueError as exc:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail=str(exc))
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail=str(exc)) from exc

    if plan.project_id:
        check_project_access(plan.project_id, current_user.user.id, project_repository)
@@ -103,7 +103,7 @@ def ai_recommend_clips(
            config=normalized_config,
            total_duration=result["total_duration"],
        )
-    except Exception:
+    except Exception as _e:
        logger.exception("AI 推荐写入失败，plan_id=%s 数据可能不一致", plan_id)
        try:
            db.rollback()
@@ -116,7 +116,7 @@ def ai_recommend_clips(
        raise HTTPException(
            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
            detail="AI推荐结果保存失败，请稍后重试",
-        )
+        ) from _e

    logger.info(
        "AI 推荐片段方案: plan_id=%s clips=%d duration=%.1f by user=%s",
@@ -167,7 +167,7 @@ def generate_cover(
    try:
        plan = svc.get_plan_or_raise(plan_id)
    except ValueError as exc:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail=str(exc))
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail=str(exc)) from exc

    if plan.project_id:
        check_project_access(plan.project_id, current_user.user.id, project_repository)
@@ -0,0 +1,415 @@
+"""剪辑计划片段（Clip）CRUD 路由。"""
+
+from __future__ import annotations
+
+import logging
+from typing import Any, List, Optional
+
+from app.auth import AuthenticatedUser, get_current_user
+from app.dependencies import get_db_session, get_project_repository
+from fastapi import APIRouter, Depends, HTTPException, Query, Response, status
+from sqlalchemy.orm import Session
+
+from packages.domain.edit_plan_clip import EditPlanClipStatus
+
+logger = logging.getLogger(__name__)
+
+router = APIRouter()
+
+
+# ── Schemas ──────────────────────────────────────────────────────────────────
+
+from pydantic import BaseModel, Field
+
+
+class EditPlanClipResponse(BaseModel):
+    """剪辑片段响应体"""
+
+    id: str
+    plan_id: str
+    clip_type: str
+    order: int
+    asset_id: str = ""
+    text_content: str = ""
+    start_time: float = 0.0
+    duration: float = 0.0
+    transition_effect: str = "cut"
+    transition_duration: float = 0.0
+    playback_speed: float = 1.0
+    status: str
+    config: dict[str, Any] = Field(default_factory=dict)
+    created_at: Optional[str] = None
+    updated_at: Optional[str] = None
+
+
+class EditPlanClipListResponse(BaseModel):
+    """剪辑片段列表响应体"""
+
+    items: List[EditPlanClipResponse]
+    total: int
+
+
+class EditPlanClipCreateRequest(BaseModel):
+    """创建剪辑片段请求体"""
+
+    clip_type: str = Field(
+        ..., min_length=1, max_length=50, description="片段类型: main/intro/outro/overlay/background/b_roll 等"
+    )
+    order: int = Field(..., ge=0, description="排序序号")
+    asset_id: str = Field(default="", max_length=64, description="关联素材 ID")
+    text_content: str = Field(default="", max_length=5000, description="文本内容（字幕/配音等）")
+    start_time: float = Field(default=0.0, ge=0.0, description="起始时间 (秒)")
+    duration: float = Field(default=0.0, ge=0.0, description="时长 (秒)")
+    transition_effect: str = Field(default="cut", max_length=50, description="转场效果")
+    transition_duration: float = Field(default=0.0, ge=0.0, description="转场时长 (秒)")
+    playback_speed: float = Field(default=1.0, gt=0.0, le=10.0, description="播放速度倍率")
+    config: dict[str, Any] = Field(default_factory=dict, description="扩展配置 (JSON)")
+
+
+class EditPlanClipUpdateRequest(BaseModel):
+    """更新剪辑片段请求体"""
+
+    clip_type: Optional[str] = Field(default=None, min_length=1, max_length=50, description="片段类型")
+    order: Optional[int] = Field(default=None, ge=0, description="排序序号")
+    asset_id: Optional[str] = Field(default=None, max_length=64, description="关联素材 ID")
+    text_content: Optional[str] = Field(default=None, max_length=5000, description="文本内容")
+    start_time: Optional[float] = Field(default=None, ge=0.0, description="起始时间 (秒)")
+    duration: Optional[float] = Field(default=None, ge=0.0, description="时长 (秒)")
+    transition_effect: Optional[str] = Field(default=None, max_length=50, description="转场效果")
+    transition_duration: Optional[float] = Field(default=None, ge=0.0, description="转场时长 (秒)")
+    playback_speed: Optional[float] = Field(default=None, gt=0.0, le=10.0, description="播放速度倍率")
+    config: Optional[dict[str, Any]] = Field(default=None, description="扩展配置 (JSON)")
+
+
+# ── Helpers ──────────────────────────────────────────────────────────────────
+
+
+def _check_plan_access(plan_id: str, user_id: str, project_repository: Any, db: Session) -> Any:
+    """验证用户是否有权限访问该剪辑计划（通过项目关联）。
+    返回 plan 对象供后续使用，避免重复查询。
+    """
+    from app.services.edit_plan_service import EditPlanService
+
+    from ._helpers import check_project_access
+
+    svc = EditPlanService(db)
+    plan = svc.get_plan(plan_id)
+    if plan is None:
+        raise HTTPException(status_code=404, detail=f"剪辑计划不存在: {plan_id}")
+    if plan.project_id:
+        check_project_access(plan.project_id, user_id, project_repository)
+    return plan
+
+
+def _clip_to_response(clip) -> EditPlanClipResponse:
+    """将领域对象转换为响应体"""
+    return EditPlanClipResponse(
+        id=clip.id,
+        plan_id=clip.plan_id,
+        clip_type=clip.clip_type,
+        order=clip.order,
+        asset_id=clip.asset_id or "",
+        text_content=clip.text_content or "",
+        start_time=clip.start_time,
+        duration=clip.duration,
+        transition_effect=clip.transition_effect or "cut",
+        transition_duration=clip.transition_duration or 0.0,
+        playback_speed=clip.playback_speed or 1.0,
+        status=clip.status.value if hasattr(clip.status, "value") else str(clip.status),
+        config=clip.config or {},
+        created_at=clip.created_at.isoformat() if clip.created_at else None,
+        updated_at=clip.updated_at.isoformat() if clip.updated_at else None,
+    )
+
+
+def _get_svc(db: Session):
+    """获取 EditPlanService 实例"""
+    from app.services.edit_plan_service import EditPlanService
+
+    return EditPlanService(db)
+
+
+# ── Routes ───────────────────────────────────────────────────────────────────
+
+
+@router.get("", response_model=EditPlanClipListResponse)
+def list_clips(
+    plan_id: str,
+    status_filter: Optional[str] = Query(None, alias="status", description="按状态过滤"),
+    skip: int = Query(0, ge=0, description="分页偏移"),
+    limit: int = Query(100, ge=1, le=500, description="每页数量"),
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> EditPlanClipListResponse:
+    """获取剪辑计划的片段列表"""
+    _check_plan_access(plan_id, current_user.user.id, project_repository, db)
+
+    svc = _get_svc(db)
+    status_enum = EditPlanClipStatus(status_filter) if status_filter else None
+    clips = svc.list_clips(plan_id, status=status_enum, skip=skip, limit=limit)
+    total = svc.count_clips(plan_id, status=status_enum)
+
+    return EditPlanClipListResponse(
+        items=[_clip_to_response(c) for c in clips],
+        total=total,
+    )
+
+
+@router.post("", response_model=EditPlanClipResponse, status_code=status.HTTP_201_CREATED)
+def create_clip(
+    plan_id: str,
+    body: EditPlanClipCreateRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> EditPlanClipResponse:
+    """创建剪辑片段"""
+    _check_plan_access(plan_id, current_user.user.id, project_repository, db)
+
+    svc = _get_svc(db)
+    try:
+        clip = svc.create_clip(
+            plan_id=plan_id,
+            clip_type=body.clip_type,
+            order=body.order,
+            asset_id=body.asset_id,
+            text_content=body.text_content,
+            start_time=body.start_time,
+            duration=body.duration,
+            transition_effect=body.transition_effect,
+            transition_duration=body.transition_duration,
+            playback_speed=body.playback_speed,
+            config=body.config,
+        )
+    except ValueError as e:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=str(e)) from e
+
+    logger.info("创建剪辑片段: plan_id=%s clip_id=%s by user=%s", plan_id, clip.id, current_user.user.id)
+    return _clip_to_response(clip)
+
+
+@router.get("/{clip_id}", response_model=EditPlanClipResponse)
+def get_clip(
+    plan_id: str,
+    clip_id: str,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> EditPlanClipResponse:
+    """获取剪辑片段详情"""
+    _check_plan_access(plan_id, current_user.user.id, project_repository, db)
+
+    svc = _get_svc(db)
+    clip = svc.get_clip(clip_id)
+    if clip is None:
+        raise HTTPException(status_code=404, detail=f"片段不存在: {clip_id}")
+    if clip.plan_id != plan_id:
+        raise HTTPException(status_code=404, detail=f"片段不存在: {clip_id}")
+
+    return _clip_to_response(clip)
+
+
+@router.put("/{clip_id}", response_model=EditPlanClipResponse)
+def update_clip(
+    plan_id: str,
+    clip_id: str,
+    body: EditPlanClipUpdateRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> EditPlanClipResponse:
+    """更新剪辑片段"""
+    _check_plan_access(plan_id, current_user.user.id, project_repository, db)
+
+    svc = _get_svc(db)
+    # 验证 clip 属于该 plan
+    clip = svc.get_clip(clip_id)
+    if clip is None:
+        raise HTTPException(status_code=404, detail=f"片段不存在: {clip_id}")
+    if clip.plan_id != plan_id:
+        raise HTTPException(status_code=404, detail=f"片段不存在: {clip_id}")
+
+    try:
+        updated = svc.update_clip(
+            clip_id,
+            clip_type=body.clip_type,
+            order=body.order,
+            asset_id=body.asset_id,
+            text_content=body.text_content,
+            start_time=body.start_time,
+            duration=body.duration,
+            transition_effect=body.transition_effect,
+            transition_duration=body.transition_duration,
+            playback_speed=body.playback_speed,
+            config=body.config,
+        )
+    except ValueError as e:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=str(e)) from e
+
+    logger.info("更新剪辑片段: plan_id=%s clip_id=%s by user=%s", plan_id, clip_id, current_user.user.id)
+    return _clip_to_response(updated)
+
+
+@router.delete("/{clip_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None, response_class=Response)
+def delete_clip(
+    plan_id: str,
+    clip_id: str,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> None:
+    """删除剪辑片段"""
+    _check_plan_access(plan_id, current_user.user.id, project_repository, db)
+
+    svc = _get_svc(db)
+    # 验证 clip 属于该 plan
+    clip = svc.get_clip(clip_id)
+    if clip is None:
+        raise HTTPException(status_code=404, detail=f"片段不存在: {clip_id}")
+    if clip.plan_id != plan_id:
+        raise HTTPException(status_code=404, detail=f"片段不存在: {clip_id}")
+
+    deleted = svc.delete_clip(clip_id)
+    if not deleted:
+        raise HTTPException(status_code=404, detail=f"片段不存在: {clip_id}")
+
+    logger.info("删除剪辑片段: plan_id=%s clip_id=%s by user=%s", plan_id, clip_id, current_user.user.id)
+    return None
+
+
+# ── 片段分割与合并 ──────────────────────────────────────────────────────────
+
+
+class SplitClipRequest(BaseModel):
+    """分割片段请求体"""
+
+    split_time: float = Field(..., gt=0, description="分割点（秒，相对于片段起始）")
+
+
+class MergeClipsRequest(BaseModel):
+    """合并片段请求体"""
+
+    clip_ids: list[str] = Field(..., min_length=2, description="要合并的片段 ID 列表")
+
+
+@router.post(
+    "/{clip_id}/split",
+    response_model=dict[str, Any],
+    summary="分割片段",
+    status_code=status.HTTP_200_OK,
+)
+def split_clip(
+    plan_id: str,
+    clip_id: str,
+    body: SplitClipRequest,
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    db: Session = Depends(get_db_session),
+    project_repository: Any = Depends(get_project_repository),
+) -> dict[str, Any]:
+    """将一个片段从指定时间点分割为两个片段。
+
+    分割后原片段变为左半部分，新增右半部分片段，后续片段顺序自动后移。
+    若片段有关联素材，会自动设置 trim_start/trim_end 标记裁剪范围。
+    """
+    _check_plan_access(plan_id, current_user.user.id, project_repository, db)
+
+    svc = _get_svc(db)
+    clip = svc.get_clip(clip_id)
+    if clip is None or clip.plan_id != plan_id:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"片段不存在: {clip_id}",
+        )
+
+    try:
+        result = svc.split_clip(clip_id, body.split_time)
+    except ValueError as e:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=str(e),
+        ) from e
+
+    left = result["left_clip"]
+    right = result["right_clip"]
+    logger.info("分割片段: plan_id=%s clip_id=%s by user=%s", plan_id, clip_id, current_user.user.id)
+
+    return {
+        "left_clip": {
+            "id": left.id,
+            "plan_id": left.plan_id,
+            "clip_type": left.clip_type,
+            "order": left.order,
+            "duration": left.duration,
+            "start_time": left.start_time,
+        },
+        "right_clip": {
+            "id": right.id,
+            "plan_id": right.plan_id,
+            "clip_type": right.clip_type,
+            "order": right.order,
+            "duration": right.duration,
+            "start_time": right.start_time,
+        },
+    }
+
+
+@router.post(
+    "/merge",
+    response_model=dict[str, Any],
+    summary="合并多个连续片段",
+    status_code=status.HTTP_200_OK,
+)
+def merge_clips(
+    plan_id: str,
+    body: MergeClipsRequest,
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    db: Session = Depends(get_db_session),
+    project_repository: Any = Depends(get_project_repository),
+) -> dict[str, Any]:
+    """将多个连续的同类型片段合并为一个片段。
+
+    合并要求：
+    - 至少 2 个片段
+    - 属于同一剪辑计划
+    - order 连续
+    - 类型相同
+
+    合并后保留第一个片段，其余删除，后续片段顺序自动前移。
+    """
+    _check_plan_access(plan_id, current_user.user.id, project_repository, db)
+
+    svc = _get_svc(db)
+
+    # 校验所有片段都属于该 plan
+    for cid in body.clip_ids:
+        clip = svc.get_clip(cid)
+        if clip is None or clip.plan_id != plan_id:
+            raise HTTPException(
+                status_code=status.HTTP_404_NOT_FOUND,
+                detail=f"片段不存在: {cid}",
+            )
+
+    try:
+        merged = svc.merge_clips(body.clip_ids)
+    except ValueError as e:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=str(e),
+        ) from e
+
+    logger.info(
+        "合并片段: plan_id=%s clip_count=%d by user=%s",
+        plan_id,
+        len(body.clip_ids),
+        current_user.user.id,
+    )
+
+    return {
+        "id": merged.id,
+        "plan_id": merged.plan_id,
+        "clip_type": merged.clip_type,
+        "order": merged.order,
+        "duration": merged.duration,
+        "text_content": merged.text_content,
+    }
@@ -0,0 +1,241 @@
+"""剪辑计划片段批量操作 API。"""
+
+from __future__ import annotations
+
+import logging
+from typing import Any, List
+
+from app.auth import AuthenticatedUser, get_current_user
+from app.dependencies import get_db_session, get_project_repository
+from fastapi import APIRouter, Depends, HTTPException, status
+from pydantic import BaseModel, Field
+from sqlalchemy.orm import Session
+
+logger = logging.getLogger(__name__)
+
+router = APIRouter()
+
+
+# ── Schemas ──────────────────────────────────────────────────────────────────
+
+
+class ClipReorderItem(BaseModel):
+    """重排序条目"""
+
+    clip_id: str
+    new_order: int = Field(..., ge=0, description="新的排序序号")
+
+
+class ClipReorderRequest(BaseModel):
+    """片段重排序请求"""
+
+    items: List[ClipReorderItem] = Field(..., min_length=1, max_length=500, description="重排序条目列表")
+
+
+class ClipReorderResponse(BaseModel):
+    """片段重排序响应"""
+
+    success: bool
+    updated_count: int
+    message: str = ""
+
+
+class ClipBatchDeleteRequest(BaseModel):
+    """批量删除片段请求"""
+
+    clip_ids: List[str] = Field(..., min_length=1, max_length=500, description="要删除的片段ID列表")
+
+
+class ClipBatchDeleteResponse(BaseModel):
+    """批量删除片段响应"""
+
+    success: bool
+    deleted_count: int
+    message: str = ""
+
+
+class ClipsFromAssetsRequest(BaseModel):
+    """从素材批量创建片段请求"""
+
+    asset_ids: List[str] = Field(..., min_length=1, max_length=200, description="素材 ID 列表，按顺序追加到时间线末尾")
+    clip_type: str = Field(default="main", description="片段类型，默认 main")
+
+
+class ClipsFromAssetsResponse(BaseModel):
+    """从素材批量创建片段响应"""
+
+    success: bool
+    created_count: int
+    message: str = ""
+    clip_ids: List[str] = Field(default_factory=list, description="创建的片段ID列表")
+
+
+# ── Helpers ──────────────────────────────────────────────────────────────────
+
+
+def _check_plan_access(plan_id: str, user_id: str, project_repository: Any, db: Session) -> Any:
+    """验证用户是否有权限访问该剪辑计划，返回 plan 对象。"""
+    from app.services.edit_plan_service import EditPlanService
+
+    from ._helpers import check_project_access
+
+    svc = EditPlanService(db)
+    plan = svc.get_plan(plan_id)
+    if plan is None:
+        raise HTTPException(status_code=404, detail=f"剪辑计划不存在: {plan_id}")
+    if plan.project_id:
+        check_project_access(plan.project_id, user_id, project_repository)
+    return plan
+
+
+def _get_svc(db: Session):
+    """获取 EditPlanService 实例"""
+    from app.services.edit_plan_service import EditPlanService
+
+    return EditPlanService(db)
+
+
+# ── Routes ───────────────────────────────────────────────────────────────────
+
+
+@router.post("/reorder", response_model=ClipReorderResponse)
+def reorder_clips(
+    plan_id: str,
+    body: ClipReorderRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> ClipReorderResponse:
+    """批量重排序片段
+
+    前端拖拽调整顺序后，一次性提交所有变更的 order。
+    自动触发编辑状态回退（从 completed/failed 切回 editing）。
+    """
+    _check_plan_access(plan_id, current_user.user.id, project_repository, db)
+
+    svc = _get_svc(db)
+
+    # 验证所有 clip 都属于该 plan
+    clip_ids = [item.clip_id for item in body.items]
+    existing_clips = svc.list_clips(plan_id, skip=0, limit=10000)
+    existing_ids = {c.id for c in existing_clips}
+
+    invalid_ids = [cid for cid in clip_ids if cid not in existing_ids]
+    if invalid_ids:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=f"以下片段不属于该计划: {', '.join(invalid_ids[:5])}",
+        )
+
+    # 执行重排序
+    updated_count = 0
+    for item in body.items:
+        try:
+            svc.update_clip(item.clip_id, order=item.new_order)
+            updated_count += 1
+        except ValueError as e:
+            logger.warning("重排序片段失败: clip_id=%s error=%s", item.clip_id, e)
+
+    logger.info(
+        "批量重排序片段: plan_id=%s count=%d by user=%s",
+        plan_id,
+        updated_count,
+        current_user.user.id,
+    )
+
+    return ClipReorderResponse(
+        success=True,
+        updated_count=updated_count,
+        message=f"成功更新 {updated_count} 个片段的顺序",
+    )
+
+
+@router.post("/batch-delete", response_model=ClipBatchDeleteResponse)
+def batch_delete_clips(
+    plan_id: str,
+    body: ClipBatchDeleteRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> ClipBatchDeleteResponse:
+    """批量删除片段
+
+    自动触发编辑状态回退（从 completed/failed 切回 editing）。
+    """
+    _check_plan_access(plan_id, current_user.user.id, project_repository, db)
+
+    svc = _get_svc(db)
+
+    # 验证所有 clip 都属于该 plan
+    existing_clips = svc.list_clips(plan_id, skip=0, limit=10000)
+    existing_ids = {c.id for c in existing_clips}
+
+    valid_ids = [cid for cid in body.clip_ids if cid in existing_ids]
+    skipped = len(body.clip_ids) - len(valid_ids)
+
+    # 执行删除
+    deleted_count = 0
+    for clip_id in valid_ids:
+        if svc.delete_clip(clip_id):
+            deleted_count += 1
+
+    message = f"成功删除 {deleted_count} 个片段"
+    if skipped > 0:
+        message += f"，跳过 {skipped} 个不存在的片段"
+
+    logger.info(
+        "批量删除片段: plan_id=%s deleted=%d skipped=%d by user=%s",
+        plan_id,
+        deleted_count,
+        skipped,
+        current_user.user.id,
+    )
+
+    return ClipBatchDeleteResponse(
+        success=True,
+        deleted_count=deleted_count,
+        message=message,
+    )
+
+
+@router.post("/from-assets", response_model=ClipsFromAssetsResponse)
+def create_clips_from_assets(
+    plan_id: str,
+    body: ClipsFromAssetsRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> ClipsFromAssetsResponse:
+    """从素材批量创建片段（追加到时间线末尾）
+
+    一次性将多个素材作为片段添加到剪辑计划，自动读取素材时长。
+    自动触发编辑状态回退（completed/failed → editing）。
+    """
+    _check_plan_access(plan_id, current_user.user.id, project_repository, db)
+
+    svc = _get_svc(db)
+
+    try:
+        clips = svc.create_clips_from_assets(
+            plan_id=plan_id,
+            asset_ids=body.asset_ids,
+            clip_type=body.clip_type,
+        )
+    except ValueError as e:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=str(e)) from e
+
+    clip_ids = [c.id for c in clips]
+
+    logger.info(
+        "从素材批量创建片段: plan_id=%s count=%d by user=%s",
+        plan_id,
+        len(clips),
+        current_user.user.id,
+    )
+
+    return ClipsFromAssetsResponse(
+        success=True,
+        created_count=len(clips),
+        message=f"成功创建 {len(clips)} 个片段",
+        clip_ids=clip_ids,
+    )
@@ -0,0 +1,315 @@
+"""封面管理 API.
+
+- GET    /{plan_id}/cover              获取封面配置
+- PUT    /{plan_id}/cover              更新封面配置
+- POST   /{plan_id}/cover/extract      从指定片段抽帧生成封面
+- POST   /{plan_id}/cover/smart        智能选帧生成封面
+"""
+
+from __future__ import annotations
+
+import logging
+from typing import Any, Optional
+
+from app.auth import AuthenticatedUser, get_current_user
+from app.core.storage import get_storage_service
+from app.dependencies import (
+    get_asset_repository,
+    get_db_session,
+    get_project_repository,
+)
+from app.services import EditPlanService
+from app.services.cover_service import CoverService
+from fastapi import APIRouter, Depends, HTTPException, status
+from pydantic import BaseModel, Field
+from sqlalchemy.orm import Session
+
+from packages.domain.config_schemas import normalize_plan_config
+
+from ._helpers import check_project_access
+
+logger = logging.getLogger(__name__)
+
+router = APIRouter()
+
+
+# ── Schemas ──────────────────────────────────────────────────────────────────
+
+
+class CoverConfigResponse(BaseModel):
+    """封面配置响应"""
+
+    type: str = Field(..., description="封面类型: ai_frame / manual / upload")
+    image_url: str = Field(default="", description="封面图片 URL")
+    frame_time: Optional[float] = Field(default=None, description="抽帧时间点（秒）")
+
+
+class CoverUpdateRequest(BaseModel):
+    """更新封面配置请求"""
+
+    type: Optional[str] = Field(default=None, description="封面类型")
+    image_url: Optional[str] = Field(default=None, description="封面图片 URL")
+    frame_time: Optional[float] = Field(default=None, ge=0.0, description="抽帧时间点（秒）")
+
+
+class CoverExtractRequest(BaseModel):
+    """从片段抽帧生成封面请求"""
+
+    clip_id: str = Field(..., description="片段 ID")
+    frame_time: float = Field(1.0, ge=0.0, description="抽帧时间点（秒）")
+
+
+class CoverSmartRequest(BaseModel):
+    """智能选帧请求"""
+
+    clip_id: Optional[str] = Field(default=None, description="指定片段 ID（不传则用第一个视频片段）")
+
+
+class CoverGenerateResponse(BaseModel):
+    """封面生成响应"""
+
+    type: str = Field(..., description="封面类型")
+    image_url: str = Field(..., description="封面图片 URL")
+    frame_time: Optional[float] = Field(default=None, description="抽帧时间点（秒）")
+
+
+# ── Routes ───────────────────────────────────────────────────────────────────
+
+
+@router.get("/{plan_id}/cover", response_model=CoverConfigResponse)
+def get_cover(
+    plan_id: str,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> CoverConfigResponse:
+    """获取封面配置"""
+    svc = EditPlanService(db)
+    plan = svc.get_plan(plan_id)
+    if not plan:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"剪辑计划不存在: {plan_id}",
+        )
+
+    if plan.project_id:
+        check_project_access(plan.project_id, current_user.user.id, project_repository)
+
+    cover = CoverService.get_cover_config(plan.config or {})
+    return CoverConfigResponse(**cover)
+
+
+@router.put("/{plan_id}/cover", response_model=CoverConfigResponse)
+def update_cover(
+    plan_id: str,
+    body: CoverUpdateRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> CoverConfigResponse:
+    """更新封面配置
+
+    用于：设置上传的封面图片 URL、切换封面类型、调整时间点等。
+    """
+    svc = EditPlanService(db)
+    plan = svc.get_plan(plan_id)
+    if not plan:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"剪辑计划不存在: {plan_id}",
+        )
+
+    if plan.project_id:
+        check_project_access(plan.project_id, current_user.user.id, project_repository)
+
+    # 合并更新
+    current_cover = CoverService.get_cover_config(plan.config or {})
+    updates = body.model_dump(exclude_none=True)
+    new_cover = {**current_cover, **updates}
+
+    # 验证 type 值
+    valid_types = {"ai_frame", "manual", "upload", "ai_regenerate"}
+    if "type" in updates and updates["type"] not in valid_types:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=f"无效的封面类型: {updates['type']}，有效值: {valid_types}",
+        )
+
+    # 更新到 plan.config
+    current_config = dict(plan.config or {})
+    current_config["cover"] = new_cover
+    normalized = normalize_plan_config(current_config)
+    updated_plan = svc.update_plan_config(plan_id, {"cover": normalized["cover"]})
+
+    result = CoverService.get_cover_config(updated_plan.config or {})
+    logger.info("更新封面配置: plan_id=%s type=%s by user=%s", plan_id, result["type"], current_user.user.id)
+    return CoverConfigResponse(**result)
+
+
+@router.post("/{plan_id}/cover/extract", response_model=CoverGenerateResponse)
+def extract_cover(
+    plan_id: str,
+    body: CoverExtractRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+    storage_service: Any = Depends(get_storage_service),
+    asset_repository: Any = Depends(get_asset_repository),
+) -> CoverGenerateResponse:
+    """从指定片段的指定时间点抽帧生成封面"""
+    svc = EditPlanService(db)
+    plan = svc.get_plan(plan_id)
+    if not plan:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"剪辑计划不存在: {plan_id}",
+        )
+
+    if plan.project_id:
+        check_project_access(plan.project_id, current_user.user.id, project_repository)
+
+    # 获取片段对应的素材
+    clip = svc.get_clip(body.clip_id)
+    if not clip:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"片段不存在: {body.clip_id}",
+        )
+
+    if clip.plan_id != plan_id:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail="片段不属于该剪辑计划",
+        )
+
+    if not clip.asset_id:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail="片段没有关联素材，无法抽帧",
+        )
+
+    # 抽帧生成封面
+    cover_svc = CoverService(storage_service, asset_repository)
+    try:
+        cover_data = cover_svc.extract_cover_from_clip(
+            plan_id=plan_id,
+            asset_id=clip.asset_id,
+            frame_time=body.frame_time,
+        )
+    except ValueError as e:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=str(e)) from e
+    except RuntimeError as e:
+        logger.error("封面抽帧失败: plan_id=%s clip_id=%s error=%s", plan_id, body.clip_id, e)
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail=f"封面抽帧失败: {e}",
+        ) from e
+
+    # 更新到 plan.config.cover
+    current_config = dict(plan.config or {})
+    current_config["cover"] = cover_data
+    normalized = normalize_plan_config(current_config)
+    svc.update_plan_config(plan_id, {"cover": normalized["cover"]})
+
+    logger.info(
+        "封面抽帧完成: plan_id=%s clip_id=%s time=%.2fs by user=%s",
+        plan_id,
+        body.clip_id,
+        body.frame_time,
+        current_user.user.id,
+    )
+    return CoverGenerateResponse(**cover_data)
+
+
+@router.post("/{plan_id}/cover/smart", response_model=CoverGenerateResponse)
+def smart_cover(
+    plan_id: str,
+    body: CoverSmartRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+    storage_service: Any = Depends(get_storage_service),
+    asset_repository: Any = Depends(get_asset_repository),
+) -> CoverGenerateResponse:
+    """智能选帧生成封面
+
+    从指定片段（或第一个视频片段）中智能选取一帧作为封面。
+    当前实现：取片段第3秒帧（后续可优化为多帧选最清晰）。
+    """
+    svc = EditPlanService(db)
+    plan = svc.get_plan(plan_id)
+    if not plan:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"剪辑计划不存在: {plan_id}",
+        )
+
+    if plan.project_id:
+        check_project_access(plan.project_id, current_user.user.id, project_repository)
+
+    # 确定使用哪个片段
+    clip_id = body.clip_id
+    asset_id = ""
+
+    if clip_id:
+        clip = svc.get_clip(clip_id)
+        if not clip:
+            raise HTTPException(
+                status_code=status.HTTP_404_NOT_FOUND,
+                detail=f"片段不存在: {clip_id}",
+            )
+        if clip.plan_id != plan_id:
+            raise HTTPException(
+                status_code=status.HTTP_400_BAD_REQUEST,
+                detail="片段不属于该剪辑计划",
+            )
+        if not clip.asset_id:
+            raise HTTPException(
+                status_code=status.HTTP_400_BAD_REQUEST,
+                detail="片段没有关联素材",
+            )
+        asset_id = clip.asset_id
+    else:
+        # 找第一个有素材的视频片段
+        clips = svc.list_clips(plan_id, limit=50, skip=0)
+        for c in clips:
+            if c.asset_id and c.clip_type == "video":
+                asset_id = c.asset_id
+                clip_id = c.id
+                break
+
+        if not asset_id:
+            raise HTTPException(
+                status_code=status.HTTP_400_BAD_REQUEST,
+                detail="没有找到可用的视频片段",
+            )
+
+    # 智能选帧
+    cover_svc = CoverService(storage_service, asset_repository)
+    try:
+        cover_data = cover_svc.generate_smart_cover(
+            plan_id=plan_id,
+            asset_id=asset_id,
+        )
+    except ValueError as e:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=str(e)) from e
+    except RuntimeError as e:
+        logger.error("智能封面生成失败: plan_id=%s error=%s", plan_id, e)
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail=f"智能封面生成失败: {e}",
+        ) from e
+
+    # 更新到 plan.config.cover
+    current_config = dict(plan.config or {})
+    current_config["cover"] = cover_data
+    normalized = normalize_plan_config(current_config)
+    svc.update_plan_config(plan_id, {"cover": normalized["cover"]})
+
+    logger.info(
+        "智能封面生成完成: plan_id=%s clip_id=%s by user=%s",
+        plan_id,
+        clip_id,
+        current_user.user.id,
+    )
+    return CoverGenerateResponse(**cover_data)
@@ -0,0 +1,274 @@
+"""导出设置 API.
+
+- GET    /{plan_id}/export              获取导出配置
+- PUT    /{plan_id}/export              更新导出配置
+- GET    /export-presets                 导出预设列表
+"""
+
+from __future__ import annotations
+
+import logging
+import re
+from typing import Any, List, Optional
+
+from app.auth import AuthenticatedUser, get_current_user
+from app.dependencies import get_db_session, get_project_repository
+from app.services import EditPlanService
+from fastapi import APIRouter, Depends, HTTPException, status
+from pydantic import BaseModel, Field, validator
+from sqlalchemy.orm import Session
+
+from packages.domain.config_schemas import normalize_plan_config
+
+from ._helpers import check_project_access
+
+logger = logging.getLogger(__name__)
+
+router = APIRouter()
+
+
+# ── 导出预设 ──────────────────────────────────────────────────────────────────
+
+EXPORT_PRESETS = [
+    {
+        "id": "export_1080p_30",
+        "name": "1080P 高清",
+        "resolution": "1080x1920",
+        "fps": 30,
+        "video_bitrate": 8000,
+        "audio_bitrate": 128,
+        "format": "mp4",
+        "quality_preset": "balanced",
+        "description": "竖屏高清，适合短视频平台",
+        "size_hint": "约 10MB/分钟",
+    },
+    {
+        "id": "export_1080p_60",
+        "name": "1080P 高帧率",
+        "resolution": "1080x1920",
+        "fps": 60,
+        "video_bitrate": 12000,
+        "audio_bitrate": 128,
+        "format": "mp4",
+        "quality_preset": "high",
+        "description": "60帧高帧率，流畅运动画面",
+        "size_hint": "约 18MB/分钟",
+    },
+    {
+        "id": "export_720p_30",
+        "name": "720P 流畅",
+        "resolution": "720x1280",
+        "fps": 30,
+        "video_bitrate": 4000,
+        "audio_bitrate": 128,
+        "format": "mp4",
+        "quality_preset": "fast",
+        "description": "快速导出，文件较小",
+        "size_hint": "约 5MB/分钟",
+    },
+    {
+        "id": "export_4k_30",
+        "name": "4K 超清",
+        "resolution": "2160x3840",
+        "fps": 30,
+        "video_bitrate": 20000,
+        "audio_bitrate": 192,
+        "format": "mp4",
+        "quality_preset": "best",
+        "description": "4K超清画质，专业品质",
+        "size_hint": "约 30MB/分钟",
+    },
+    {
+        "id": "export_1080p_30_mov",
+        "name": "1080P ProRes",
+        "resolution": "1080x1920",
+        "fps": 30,
+        "video_bitrate": 15000,
+        "audio_bitrate": 256,
+        "format": "mov",
+        "quality_preset": "high",
+        "description": "MOV格式，适合后期剪辑",
+        "size_hint": "约 25MB/分钟",
+    },
+]
+
+VALID_QUALITY_PRESETS = {"ultra_fast", "fast", "balanced", "high", "best"}
+VALID_FORMATS = {"mp4", "mov"}
+
+RESOLUTION_PATTERN = re.compile(r"^\d+x\d+$")
+
+
+# ── Schemas ──────────────────────────────────────────────────────────────────
+
+
+class ExportConfigResponse(BaseModel):
+    """导出配置响应"""
+
+    resolution: str
+    fps: int
+    video_bitrate: int
+    audio_bitrate: int
+    format: str
+    quality_preset: str
+    watermark_enabled: bool
+    watermark_text: str
+
+
+class ExportUpdateRequest(BaseModel):
+    """更新导出配置请求"""
+
+    resolution: Optional[str] = None
+    fps: Optional[int] = Field(default=None, ge=15, le=60)
+    video_bitrate: Optional[int] = Field(default=None, ge=1000, le=20000)
+    audio_bitrate: Optional[int] = Field(default=None, ge=64, le=320)
+    format: Optional[str] = None
+    quality_preset: Optional[str] = None
+    watermark_enabled: Optional[bool] = None
+    watermark_text: Optional[str] = None
+
+    @validator("resolution")
+    def validate_resolution(cls, v):
+        if v is None:
+            return v
+        if not RESOLUTION_PATTERN.match(v):
+            raise ValueError("分辨率格式错误，应为 宽x高，如 1080x1920")
+        w, h = v.split("x")
+        if int(w) < 100 or int(h) < 100:
+            raise ValueError("分辨率数值过小")
+        if int(w) > 4096 or int(h) > 4096:
+            raise ValueError("分辨率数值过大，最大 4096x4096")
+        return v
+
+    @validator("format")
+    def validate_format(cls, v):
+        if v is None:
+            return v
+        if v not in VALID_FORMATS:
+            raise ValueError(f"无效格式: {v}，支持: {VALID_FORMATS}")
+        return v
+
+    @validator("quality_preset")
+    def validate_quality_preset(cls, v):
+        if v is None:
+            return v
+        if v not in VALID_QUALITY_PRESETS:
+            raise ValueError(f"无效质量预设: {v}，支持: {VALID_QUALITY_PRESETS}")
+        return v
+
+
+class ExportPresetItem(BaseModel):
+    """导出预设条目"""
+
+    id: str
+    name: str
+    resolution: str
+    fps: int
+    video_bitrate: int
+    audio_bitrate: int
+    format: str
+    quality_preset: str
+    description: str
+    size_hint: str
+
+
+class ExportPresetListResponse(BaseModel):
+    """导出预设列表响应"""
+
+    items: List[ExportPresetItem]
+    total: int
+
+
+# ── Helpers ──────────────────────────────────────────────────────────────────
+
+
+def _get_export_config(plan_config: dict) -> dict:
+    e = plan_config.get("export", {})
+    if not isinstance(e, dict):
+        e = {}
+    return {
+        "resolution": e.get("resolution", "1080x1920"),
+        "fps": e.get("fps", 30),
+        "video_bitrate": e.get("video_bitrate", 8000),
+        "audio_bitrate": e.get("audio_bitrate", 128),
+        "format": e.get("format", "mp4"),
+        "quality_preset": e.get("quality_preset", "balanced"),
+        "watermark_enabled": e.get("watermark_enabled", False),
+        "watermark_text": e.get("watermark_text", ""),
+    }
+
+
+# ── Routes ───────────────────────────────────────────────────────────────────
+
+
+@router.get("/export-presets", response_model=ExportPresetListResponse)
+def list_export_presets(
+    current_user: AuthenticatedUser = Depends(get_current_user),
+) -> ExportPresetListResponse:
+    """获取导出预设列表"""
+    items = [ExportPresetItem(**p) for p in EXPORT_PRESETS]
+    return ExportPresetListResponse(items=items, total=len(items))
+
+
+@router.get("/{plan_id}/export", response_model=ExportConfigResponse)
+def get_export_config(
+    plan_id: str,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> ExportConfigResponse:
+    """获取导出配置"""
+    svc = EditPlanService(db)
+    plan = svc.get_plan(plan_id)
+    if not plan:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"剪辑计划不存在: {plan_id}",
+        )
+
+    if plan.project_id:
+        check_project_access(plan.project_id, current_user.user.id, project_repository)
+
+    config = _get_export_config(plan.config or {})
+    return ExportConfigResponse(**config)
+
+
+@router.put("/{plan_id}/export", response_model=ExportConfigResponse)
+def update_export_config(
+    plan_id: str,
+    body: ExportUpdateRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> ExportConfigResponse:
+    """更新导出配置"""
+    svc = EditPlanService(db)
+    plan = svc.get_plan(plan_id)
+    if not plan:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"剪辑计划不存在: {plan_id}",
+        )
+
+    if plan.project_id:
+        check_project_access(plan.project_id, current_user.user.id, project_repository)
+
+    # 合并更新
+    current = _get_export_config(plan.config or {})
+    updates = body.model_dump(exclude_none=True)
+    new_export = {**current, **updates}
+
+    # 更新到 plan.config
+    current_config = dict(plan.config or {})
+    current_config["export"] = new_export
+    normalized = normalize_plan_config(current_config)
+    updated_plan = svc.update_plan_config(plan_id, {"export": normalized["export"]})
+
+    result = _get_export_config(updated_plan.config or {})
+    logger.info(
+        "更新导出配置: plan_id=%s resolution=%s fps=%d by user=%s",
+        plan_id,
+        result["resolution"],
+        result["fps"],
+        current_user.user.id,
+    )
+    return ExportConfigResponse(**result)
@@ -0,0 +1,197 @@
+"""滤镜调色 API.
+
+- GET    /filter-presets            滤镜预设列表
+- GET    /{plan_id}/filter          获取全局滤镜配置
+- PUT    /{plan_id}/filter          更新全局滤镜配置
+"""
+
+from __future__ import annotations
+
+import logging
+from typing import Any, List, Optional
+
+from app.auth import AuthenticatedUser, get_current_user
+from app.dependencies import get_db_session, get_project_repository
+from app.services import EditPlanService
+from fastapi import APIRouter, Depends, HTTPException, Query, status
+from pydantic import BaseModel, Field
+from sqlalchemy.orm import Session
+
+from packages.domain.config_schemas import normalize_plan_config
+from packages.domain.filter_presets import (
+    FilterPreset,
+    get_filter_preset,
+    list_filter_presets,
+)
+
+from ._helpers import check_project_access
+
+logger = logging.getLogger(__name__)
+
+router = APIRouter()
+
+
+# ── Schemas ──────────────────────────────────────────────────────────────────
+
+
+class FilterPresetResponse(BaseModel):
+    """滤镜预设响应"""
+
+    id: str
+    name: str
+    category: str
+    description: str
+    tags: List[str] = Field(default_factory=list)
+
+
+class FilterConfigResponse(BaseModel):
+    """滤镜配置响应"""
+
+    enabled: bool
+    preset_id: str
+    intensity: int
+    brightness: float
+    contrast: float
+    saturation: float
+    warmth: float
+
+
+class FilterUpdateRequest(BaseModel):
+    """更新滤镜配置请求"""
+
+    enabled: Optional[bool] = None
+    preset_id: Optional[str] = None
+    intensity: Optional[int] = Field(default=None, ge=0, le=100)
+    brightness: Optional[float] = Field(default=None, ge=-1.0, le=1.0)
+    contrast: Optional[float] = Field(default=None, ge=0.0, le=2.0)
+    saturation: Optional[float] = Field(default=None, ge=0.0, le=3.0)
+    warmth: Optional[float] = Field(default=None, ge=-1.0, le=1.0)
+
+
+class FilterPresetListResponse(BaseModel):
+    """滤镜预设列表响应"""
+
+    items: List[FilterPresetResponse]
+    total: int
+
+
+# ── Helpers ──────────────────────────────────────────────────────────────────
+
+
+def _preset_to_response(p: FilterPreset) -> FilterPresetResponse:
+    return FilterPresetResponse(
+        id=p.id,
+        name=p.name,
+        category=p.category,
+        description=p.description,
+        tags=list(p.tags),
+    )
+
+
+def _get_filter_config(plan_config: dict) -> dict:
+    """从 plan.config 中提取滤镜配置"""
+    f = plan_config.get("filter", {})
+    if not isinstance(f, dict):
+        f = {}
+    return {
+        "enabled": f.get("enabled", False),
+        "preset_id": f.get("preset_id", "filter_none"),
+        "intensity": f.get("intensity", 100),
+        "brightness": f.get("brightness", 0.0),
+        "contrast": f.get("contrast", 1.0),
+        "saturation": f.get("saturation", 1.0),
+        "warmth": f.get("warmth", 0.0),
+    }
+
+
+# ── Routes ───────────────────────────────────────────────────────────────────
+
+
+@router.get("/filter-presets", response_model=FilterPresetListResponse)
+def list_presets(
+    category: Optional[str] = Query(default=None, description="按分类筛选"),
+    keyword: Optional[str] = Query(default=None, description="关键词搜索"),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+) -> FilterPresetListResponse:
+    """获取滤镜预设列表"""
+    presets = list_filter_presets(category=category, keyword=keyword)
+    items = [_preset_to_response(p) for p in presets]
+    return FilterPresetListResponse(items=items, total=len(items))
+
+
+@router.get("/{plan_id}/filter", response_model=FilterConfigResponse)
+def get_filter(
+    plan_id: str,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> FilterConfigResponse:
+    """获取剪辑计划的全局滤镜配置"""
+    svc = EditPlanService(db)
+    plan = svc.get_plan(plan_id)
+    if not plan:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"剪辑计划不存在: {plan_id}",
+        )
+
+    if plan.project_id:
+        check_project_access(plan.project_id, current_user.user.id, project_repository)
+
+    config = _get_filter_config(plan.config or {})
+    return FilterConfigResponse(**config)
+
+
+@router.put("/{plan_id}/filter", response_model=FilterConfigResponse)
+def update_filter(
+    plan_id: str,
+    body: FilterUpdateRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> FilterConfigResponse:
+    """更新全局滤镜配置"""
+    svc = EditPlanService(db)
+    plan = svc.get_plan(plan_id)
+    if not plan:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"剪辑计划不存在: {plan_id}",
+        )
+
+    if plan.project_id:
+        check_project_access(plan.project_id, current_user.user.id, project_repository)
+
+    # 验证 preset_id
+    updates = body.model_dump(exclude_none=True)
+    if "preset_id" in updates:
+        preset = get_filter_preset(updates["preset_id"])
+        if preset is None:
+            raise HTTPException(
+                status_code=status.HTTP_400_BAD_REQUEST,
+                detail=f"无效的滤镜预设: {updates['preset_id']}",
+            )
+
+    # 合并更新
+    current = _get_filter_config(plan.config or {})
+    new_filter = {**current, **updates}
+
+    # 如果设为原图 preset，自动关闭
+    if new_filter["preset_id"] == "filter_none":
+        new_filter["enabled"] = False
+
+    # 更新到 plan.config
+    current_config = dict(plan.config or {})
+    current_config["filter"] = new_filter
+    normalized = normalize_plan_config(current_config)
+    updated_plan = svc.update_plan_config(plan_id, {"filter": normalized["filter"]})
+
+    result = _get_filter_config(updated_plan.config or {})
+    logger.info(
+        "更新滤镜配置: plan_id=%s preset=%s intensity=%d by user=%s",
+        plan_id,
+        result["preset_id"],
+        result["intensity"],
+        current_user.user.id,
+    )
+    return FilterConfigResponse(**result)
@@ -20,6 +20,7 @@ from app.api.routes.edit_plans import (
 )
 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.core.task_enqueue import GLOBAL_PENDING_LIMIT, USER_PENDING_LIMIT
 from app.dependencies import get_asset_library_repository, get_asset_repository, get_db_session, get_project_repository
 from app.services import EditPlanService
@@ -131,17 +132,21 @@ def _auto_fallback_auto_material_mode(
    asset_library_repo: Any,
    asset_repo: Any,
 ) -> None:
-    """自动兜底 4: 自动素材模式 → 从项目默认视频素材库选取"""
+    """自动兜底 4: 项目有视频素材库时，自动选取 ready 视频素材分配给无素材片段
+
+    注：原先需要 material_mode=="auto" 才触发，但全代码库没有任何地方设置为 auto，
+    导致这道兜底防线永远不生效。现改为：只要有 project_id 且存在无素材片段，
+    就自动从项目视频素材库选取素材兜底，确保一键生成等场景能正常出片。
+    """
    if not clips_without_asset:
        return
-    material_mode = (plan_check.config or {}).get("material_mode", "manual")
-    if material_mode != "auto" or not plan_check.project_id:
+    if not plan_check.project_id:
        return

    import random

    logger.info(
-        "自动兜底4: plan=%s 自动素材模式，从项目素材库选取素材 (%d 个片段需要)",
+        "自动兜底4: plan=%s 自动选素材分配给 %d 个无素材片段",
        plan_id,
        len(clips_without_asset),
    )
@@ -183,9 +188,7 @@ def _auto_fallback_auto_material_mode(
 def _check_queue_limits(gen_task_repo, user_id: str) -> None:
    """队列限流预检查"""
    try:
-        has_count = hasattr(gen_task_repo, "count_pending_by_user") and hasattr(
-            gen_task_repo, "count_pending_total"
-        )
+        has_count = hasattr(gen_task_repo, "count_pending_by_user") and hasattr(gen_task_repo, "count_pending_total")
        if has_count:
            user_pending = gen_task_repo.count_pending_by_user(user_id)
            global_pending = gen_task_repo.count_pending_total()
@@ -241,7 +244,7 @@ def generate_plan(
    try:
        can_gen, reason = svc.can_generate(plan_id)
    except ValueError as exc:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail=str(exc))
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail=str(exc)) from exc
    if not can_gen:
        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=reason)

@@ -255,12 +258,15 @@ def generate_plan(

        gen_task_use_case = CreateGenerationTaskUseCase(gen_task_repo)
        plan = svc.get_plan_or_raise(plan_id)
+        # 从 plan.config 中读取 asset_ids 并传递给 GenerationTask
+        config_asset_ids = (plan.config or {}).get("asset_ids", [])
        gen_task = gen_task_use_case.execute(
            CreateGenerationTaskCommand(
-                project_id="",
+                project_id=plan.project_id or "",
                template_id=plan.template_id,
                created_by_user_id=current_user.user.id,
                source_edit_plan_id=plan_id,
+                asset_ids=list(config_asset_ids) if config_asset_ids else [],
            )
        )

@@ -286,7 +292,7 @@ def generate_plan(
        )
    except HTTPException:
        raise
-    except Exception:
+    except Exception as _e:
        logger.exception("触发剪辑计划生成失败: plan_id=%s", plan_id)
        try:
            svc.transition_status(plan_id, EditPlanStatus.FAILED)
@@ -295,7 +301,7 @@ def generate_plan(
        raise HTTPException(
            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
            detail="生成失败，请稍后重试",
-        )
+        ) from _e


@router.get(
@@ -307,13 +313,14 @@ def get_generation_status(
    db: Session = Depends(get_db_session),
    current_user: AuthenticatedUser = Depends(get_current_user),
    project_repository: Any = Depends(get_project_repository),
+    storage_service: OSSStorageService = Depends(get_storage_service),
 ) -> EditPlanGenerationStatusResponse:
    """查询剪辑计划生成进度"""
    svc = EditPlanService(db)
    try:
        gen_status = svc.get_generation_status(plan_id)
    except ValueError as exc:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail=str(exc))
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail=str(exc)) from exc

    plan = gen_status["plan"]
    if plan.project_id:
@@ -333,10 +340,32 @@ def get_generation_status(
        for c in clips
    ]

+    # 从 plan.config 中取渲染结果 URL，转换为签名 URL
+    raw_video_url = (plan.config or {}).get("rendered_url", "")
+    video_url = ""
+    if raw_video_url:
+        try:
+            video_url = storage_service.get_download_url(raw_video_url, expires_seconds=86400)
+        except Exception as e:
+            logger.warning("生成视频签名URL失败，返回原始URL: plan_id=%s error=%s", plan_id, e)
+            video_url = raw_video_url
+    # 从 gen_status 中取进度、错误信息、任务状态
+    progress = gen_status.get("progress", 0.0)
+    error_message = gen_status.get("error_message", "")
+    gen_task_status = gen_status.get("generation_task_status")
+    # 如果计划已完成但进度还是0，补100
+    plan_status_val = plan.status.value if hasattr(plan.status, "value") else plan.status
+    if plan_status_val == "completed" and progress < 100:
+        progress = 100.0
+
    return EditPlanGenerationStatusResponse(
        plan_id=plan_id,
-        plan_status=plan.status.value if hasattr(plan.status, "value") else plan.status,
+        plan_status=plan_status_val,
        generation_task_id=gen_status["generation_task_id"],
+        generation_task_status=gen_task_status,
+        progress=progress,
+        video_url=video_url,
+        error_message=error_message,
        clips=clip_items,
    )

@@ -10,7 +10,7 @@ from __future__ import annotations
 import logging
 from typing import Any, List

-from app.api.routes._helpers import check_project_access
+from app.api.routes._helpers import auto_select_video_assets, check_project_access
 from app.api.routes.edit_plans import (
    GenerateFromTemplateRequest,
    GenerateFromTemplateResponse,
@@ -18,9 +18,14 @@ from app.api.routes.edit_plans import (
    _to_response,
 )
 from app.auth import AuthenticatedUser, get_current_user
-from app.dependencies import get_db_session, get_project_repository
+from app.dependencies import (
+    get_asset_library_repository,
+    get_asset_repository,
+    get_db_session,
+    get_project_repository,
+)
 from app.services import EditPlanService, PlanGeneratorService
-from fastapi import APIRouter, Depends, HTTPException, Query, status
+from fastapi import APIRouter, Depends, HTTPException, status
 from pydantic import BaseModel, Field
 from sqlalchemy.orm import Session

@@ -161,6 +166,8 @@ def generate_from_template(
    db: Session = Depends(get_db_session),
    current_user: AuthenticatedUser = Depends(get_current_user),
    project_repository: Any = Depends(get_project_repository),
+    asset_repository: Any = Depends(get_asset_repository),
+    asset_library_repository: Any = Depends(get_asset_library_repository),
 ) -> GenerateFromTemplateResponse:
    """基于模板 + 素材自动生成剪辑计划"""
    from app.services import EditTemplateService
@@ -173,15 +180,32 @@ def generate_from_template(
    try:
        template = template_svc.get_template_or_raise(body.template_id)
    except ValueError as exc:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail=str(exc))
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail=str(exc)) from exc

    clip_configs = template_svc.list_clip_configs(body.template_id, skip=0, limit=200)

+    # 自动选素材：未传 asset_ids 但有 project_id 时，从项目视频素材库选 ready 的视频素材
+    resolved_asset_ids = list(body.asset_ids)
+    if not resolved_asset_ids and body.project_id:
+        auto_assets = auto_select_video_assets(
+            project_id=body.project_id,
+            asset_library_repo=asset_library_repository,
+            asset_repo=asset_repository,
+            logger=logger,
+        )
+        if auto_assets:
+            resolved_asset_ids = auto_assets
+            logger.info(
+                "generate-from-template 自动选素材: project_id=%s count=%d",
+                body.project_id,
+                len(auto_assets),
+            )
+
    generator = PlanGeneratorService(db)
    result = generator.generate_from_template(
        template=template,
        clip_configs=clip_configs,
-        asset_ids=body.asset_ids,
+        asset_ids=resolved_asset_ids,
        project_id=body.project_id,
        created_by_user_id=current_user.user.id,
        name=body.name,
@@ -190,8 +214,20 @@ def generate_from_template(
    plan = result["plan"]
    clips = result["clips"]

+    # 把 asset_ids 写入 plan.config，供生成时兜底分配使用
+    if resolved_asset_ids:
+        from app.services import EditPlanService
+
+        from packages.domain.config_schemas import normalize_plan_config
+
+        svc = EditPlanService(db)
+        current_config = plan.config or {}
+        if current_config.get("asset_ids") != resolved_asset_ids:
+            current_config["asset_ids"] = resolved_asset_ids
+            plan = svc.update_plan(plan.id, config=normalize_plan_config(current_config))
+
    logger.info(
-        "基于模板生成剪辑计划: plan_id=%s template_id=%s clips=%d by user=%s",
+        "基于模板生成剪辑计划: plan_id=%s template_id=%s clips=%d by user=%d",
        plan.id,
        body.template_id,
        len(clips),
@@ -0,0 +1,272 @@
+"""转场特效 API.
+
+- GET    /transition-presets              转场预设列表
+- PUT    /clips/{clip_id}/transition      设置单个片段转场
+- POST   /{plan_id}/transitions/batch     批量设置转场（所有片段）
+"""
+
+from __future__ import annotations
+
+import logging
+from typing import Any, List, Optional
+
+from app.auth import AuthenticatedUser, get_current_user
+from app.dependencies import get_db_session, get_project_repository
+from app.services import EditPlanService
+from fastapi import APIRouter, Depends, HTTPException, Query, status
+from pydantic import BaseModel, Field
+from sqlalchemy.orm import Session
+
+from packages.domain.transition_presets import (
+    TransitionPreset,
+    get_transition_preset,
+    list_transition_presets,
+)
+
+from ._helpers import check_project_access
+
+logger = logging.getLogger(__name__)
+
+router = APIRouter()
+
+
+# ── Schemas ──────────────────────────────────────────────────────────────────
+
+
+class TransitionPresetResponse(BaseModel):
+    """转场预设响应"""
+
+    id: str
+    name: str
+    category: str
+    description: str
+    tags: List[str] = Field(default_factory=list)
+    default_duration: float
+    min_duration: float
+    max_duration: float
+
+
+class TransitionUpdateRequest(BaseModel):
+    """更新转场请求"""
+
+    effect: str = Field(..., description="转场效果 ID")
+    duration: Optional[float] = Field(default=None, ge=0.0, description="转场时长（秒）")
+
+
+class BatchTransitionRequest(BaseModel):
+    """批量设置转场请求"""
+
+    effect: str = Field(..., description="转场效果 ID")
+    duration: Optional[float] = Field(default=None, ge=0.0, description="转场时长（秒）")
+    apply_to: str = Field(
+        default="all",
+        description="应用范围: all=所有片段, except_first=除第一个外, except_last=除最后一个, middle=中间片段",
+    )
+
+
+class ClipTransitionResponse(BaseModel):
+    """片段转场信息响应"""
+
+    clip_id: str
+    effect: str
+    duration: float
+
+
+class BatchTransitionResponse(BaseModel):
+    """批量转场响应"""
+
+    updated_count: int
+    plan_id: str
+
+
+class TransitionPresetListResponse(BaseModel):
+    """转场预设列表响应"""
+
+    items: List[TransitionPresetResponse]
+    total: int
+
+
+# ── Helpers ──────────────────────────────────────────────────────────────────
+
+
+def _preset_to_response(p: TransitionPreset) -> TransitionPresetResponse:
+    return TransitionPresetResponse(
+        id=p.id,
+        name=p.name,
+        category=p.category,
+        description=p.description,
+        tags=list(p.tags),
+        default_duration=p.default_duration,
+        min_duration=p.min_duration,
+        max_duration=p.max_duration,
+    )
+
+
+def _validate_transition(effect: str, duration: Optional[float] = None) -> tuple[str, float]:
+    """验证转场效果和时长，返回 (effect, duration)"""
+    preset = get_transition_preset(effect)
+    if preset is None:
+        raise ValueError(f"无效的转场效果: {effect}")
+
+    # 硬切特殊处理，时长强制为0
+    if effect == "transition_none" or preset.transition == "none":
+        return "cut", 0.0
+
+    final_duration = duration if duration is not None else preset.default_duration
+    if final_duration < preset.min_duration:
+        final_duration = preset.min_duration
+    if final_duration > preset.max_duration:
+        final_duration = preset.max_duration
+
+    return preset.transition, round(final_duration, 3)
+
+
+# ── Routes ───────────────────────────────────────────────────────────────────
+
+
+@router.get("/transition-presets", response_model=TransitionPresetListResponse)
+def list_presets(
+    category: Optional[str] = Query(default=None, description="按分类筛选"),
+    keyword: Optional[str] = Query(default=None, description="关键词搜索"),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+) -> TransitionPresetListResponse:
+    """获取转场预设列表"""
+    presets = list_transition_presets(category=category, keyword=keyword)
+    items = [_preset_to_response(p) for p in presets]
+    return TransitionPresetListResponse(items=items, total=len(items))
+
+
+@router.put("/clips/{clip_id}/transition", response_model=ClipTransitionResponse)
+def update_clip_transition(
+    clip_id: str,
+    body: TransitionUpdateRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> ClipTransitionResponse:
+    """设置单个片段的转场效果"""
+    svc = EditPlanService(db)
+    clip = svc.get_clip(clip_id)
+    if not clip:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"片段不存在: {clip_id}",
+        )
+
+    plan = svc.get_plan(clip.plan_id)
+    if plan and plan.project_id:
+        check_project_access(plan.project_id, current_user.user.id, project_repository)
+
+    # 验证转场参数
+    try:
+        effect, duration = _validate_transition(body.effect, body.duration)
+    except ValueError as e:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=str(e)) from e
+
+    # 更新片段
+    updated_clip = svc.update_clip(
+        clip_id,
+        transition_effect=effect,
+        transition_duration=duration,
+    )
+
+    logger.info(
+        "更新片段转场: clip_id=%s effect=%s duration=%.3f by user=%s",
+        clip_id,
+        effect,
+        duration,
+        current_user.user.id,
+    )
+    return ClipTransitionResponse(
+        clip_id=clip_id,
+        effect=updated_clip.transition_effect,
+        duration=updated_clip.transition_duration,
+    )
+
+
+@router.post("/{plan_id}/transitions/batch", response_model=BatchTransitionResponse)
+def batch_update_transitions(
+    plan_id: str,
+    body: BatchTransitionRequest,
+    db: Session = Depends(get_db_session),
+    current_user: AuthenticatedUser = Depends(get_current_user),
+    project_repository: Any = Depends(get_project_repository),
+) -> BatchTransitionResponse:
+    """批量设置计划内所有片段的转场效果
+
+    apply_to 说明：
+    - all: 所有片段
+    - except_first: 除第一个片段外（第一个片段不需要前转场）
+    - except_last: 除最后一个片段外
+    - middle: 只设置中间片段（除首尾）
+    """
+    svc = EditPlanService(db)
+    plan = svc.get_plan(plan_id)
+    if not plan:
+        raise HTTPException(
+            status_code=status.HTTP_404_NOT_FOUND,
+            detail=f"剪辑计划不存在: {plan_id}",
+        )
+
+    if plan.project_id:
+        check_project_access(plan.project_id, current_user.user.id, project_repository)
+
+    # 验证转场参数
+    try:
+        effect, duration = _validate_transition(body.effect, body.duration)
+    except ValueError as e:
+        raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=str(e)) from e
+
+    # 获取所有片段
+    clips = svc.list_clips(plan_id, limit=500, skip=0)
+    if not clips:
+        return BatchTransitionResponse(updated_count=0, plan_id=plan_id)
+
+    # 确定应用范围
+    total = len(clips)
+    if total <= 1:
+        # 只有一个片段时，只有 all 模式才应用
+        if body.apply_to != "all":
+            return BatchTransitionResponse(updated_count=0, plan_id=plan_id)
+
+    # 按 order 排序
+    clips_sorted = sorted(clips, key=lambda c: c.order)
+    indices_to_update = []
+
+    if body.apply_to == "all":
+        indices_to_update = list(range(total))
+    elif body.apply_to == "except_first":
+        indices_to_update = list(range(1, total))
+    elif body.apply_to == "except_last":
+        indices_to_update = list(range(total - 1))
+    elif body.apply_to == "middle":
+        if total <= 2:
+            indices_to_update = []
+        else:
+            indices_to_update = list(range(1, total - 1))
+    else:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=f"无效的 apply_to: {body.apply_to}",
+        )
+
+    # 批量更新
+    count = 0
+    for idx in indices_to_update:
+        clip = clips_sorted[idx]
+        svc.update_clip(
+            clip.id,
+            transition_effect=effect,
+            transition_duration=duration,
+        )
+        count += 1
+
+    logger.info(
+        "批量更新转场: plan_id=%s count=%d effect=%s apply_to=%s by user=%s",
+        plan_id,
+        count,
+        effect,
+        body.apply_to,
+        current_user.user.id,
+    )
+    return BatchTransitionResponse(updated_count=count, plan_id=plan_id)
@@ -105,7 +105,7 @@ async def list_feature_flags(
        return sorted(result, key=lambda x: x.name)
    except Exception as exc:
        logger.error("Failed to list feature flags: %s", exc)
-        raise HTTPException(status_code=500, detail=f"Failed to list flags: {exc}")
+        raise HTTPException(status_code=500, detail=f"Failed to list flags: {exc}") from exc


@router.get("/{name}", response_model=FeatureFlagResponse)
@@ -120,7 +120,7 @@ async def get_feature_flag(
        return FeatureFlagResponse.from_config(config)
    except Exception as exc:
        logger.error("Failed to get feature flag %s: %s", name, exc)
-        raise HTTPException(status_code=500, detail=f"Failed to get flag: {exc}")
+        raise HTTPException(status_code=500, detail=f"Failed to get flag: {exc}") from exc


@router.get("/{name}/check", response_model=FeatureFlagCheckResponse)
@@ -136,7 +136,7 @@ async def check_feature_flag(
        return FeatureFlagCheckResponse(name=name, active=active, identifier=identifier)
    except Exception as exc:
        logger.error("Failed to check feature flag %s: %s", name, exc)
-        raise HTTPException(status_code=500, detail=f"Failed to check flag: {exc}")
+        raise HTTPException(status_code=500, detail=f"Failed to check flag: {exc}") from exc


@router.put("/{name}", response_model=FeatureFlagResponse)
@@ -170,7 +170,7 @@ async def update_feature_flag(
        return FeatureFlagResponse.from_config(config)
    except Exception as exc:
        logger.error("Failed to update feature flag %s: %s", name, exc)
-        raise HTTPException(status_code=500, detail=f"Failed to update flag: {exc}")
+        raise HTTPException(status_code=500, detail=f"Failed to update flag: {exc}") from exc


@router.delete("/{name}", status_code=status.HTTP_204_NO_CONTENT, response_model=None, response_class=Response)
@@ -178,7 +178,7 @@ async def delete_feature_flag(
    name: str,
    _: bool = Depends(_verify_internal_api_key),
    store: RedisFeatureFlagStore = Depends(_get_feature_flag_store),
-) :
+):
    """删除 Feature Flag。

    只允许删除 ALLOWED_FLAGS 列表中的 flag。
@@ -191,4 +191,4 @@ async def delete_feature_flag(
        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}")
+        raise HTTPException(status_code=500, detail=f"Failed to delete flag: {exc}") from exc
@@ -43,6 +43,7 @@ logger = logging.getLogger(__name__)

 router = APIRouter()

+
 def _to_generation_task_response(task) -> GenerationTaskResponse:
    return GenerationTaskResponse(
        id=task.id,
@@ -282,28 +283,28 @@ def create_generation_task(
                    created_tasks.append(task)
                else:
                    failed_tasks.append(task)
-            except UserPendingLimitExceeded:
+            except UserPendingLimitExceeded as _e:
                # 兜底：如果预检查后又并发提交了，在这里也拦住
                failed_tasks.append(task)
                if not created_tasks:
                    raise HTTPException(
                        status_code=429,
                        detail="您的待处理任务过多，请等待完成后再提交",
-                    )
+                    ) from _e
                break
-            except GlobalQueueFull:
+            except GlobalQueueFull as _e:
                failed_tasks.append(task)
                if not created_tasks:
                    raise HTTPException(
                        status_code=503,
                        detail="系统繁忙，请稍后再试",
-                    )
+                    ) from _e
                break
    except HTTPException:
        raise
    except Exception as e:
        logger.error("[生成任务] 创建失败: %s", e, exc_info=True)
-        raise HTTPException(status_code=500, detail="创建生成任务失败，请稍后重试或查看任务日志")
+        raise HTTPException(status_code=500, detail="创建生成任务失败，请稍后重试或查看任务日志") from e

    items = [_to_generation_task_response(t) for t in created_tasks + failed_tasks]
    return BatchGenerationTaskResponse(items=items, total=len(items))
@@ -426,3 +427,53 @@ def retry_generation_task(
            detail="系统繁忙，请稍后再试",
        ) from None
    return _to_generation_task_response(retried)
+
+
+@router.post("/tasks/{task_id}/cancel", response_model=GenerationTaskResponse)
+def cancel_generation_task(
+    task_id: str,
+    authenticated_user: AuthenticatedUser = Depends(get_current_user),
+    generation_task_repository: Any = Depends(get_generation_task_repository),
+) -> GenerationTaskResponse:
+    """取消生成任务。
+
+    仅 pending / running 状态的任务可取消；取消后状态变为 cancelled。
+    对于已在运行的 Celery 任务，标记为 cancelled 后，worker 在下次检查点会中止执行。
+    """
+    task = generation_task_repository.get(task_id)
+    if task is None:
+        raise HTTPException(status_code=404, detail="Generation task not found")
+
+    # 权限校验
+    if task.created_by_user_id and task.created_by_user_id != authenticated_user.user.id:
+        raise HTTPException(status_code=403, detail="Access denied to this task")
+
+    status_val = task.status.value if hasattr(task.status, "value") else str(task.status)
+
+    # 终态不可取消
+    if status_val in ("completed", "failed", "cancelled"):
+        raise HTTPException(
+            status_code=409,
+            detail=f"Cannot cancel task in {status_val} status",
+        )
+
+    # 执行取消
+    try:
+        task.mark_cancelled()
+        task.append_log(
+            stage="cancelled",
+            message="用户主动取消任务",
+            level="INFO",
+            cancelled_by=authenticated_user.user.id,
+        )
+        generation_task_repository.update(task)
+        logger.info(
+            "生成任务已取消: task_id=%s user_id=%s previous_status=%s",
+            task_id,
+            authenticated_user.user.id,
+            status_val,
+        )
+    except ValueError as e:
+        raise HTTPException(status_code=400, detail=str(e)) from e
+
+    return _to_generation_task_response(task)
@@ -81,11 +81,11 @@ def delete_project(
    use_case = DeleteProjectUseCase(project_repository)
    try:
        deleted = use_case.execute(project_id, authenticated_user.user.id)
-    except PermissionError:
+    except PermissionError as _e:
        raise HTTPException(
            status_code=status.HTTP_403_FORBIDDEN,
            detail="Only the project owner can delete this project",
-        )
+        ) from _e
    if not deleted:
        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Project not found")
-    return
+    return  # type: ignore[return-value]
@@ -2,6 +2,7 @@

 from __future__ import annotations

+import logging
 from dataclasses import replace
 from datetime import datetime, timezone
 from typing import List
@@ -20,6 +21,8 @@ from fastapi import APIRouter, Depends, HTTPException, status

 from packages.ports.user_repository import UserRepository

+logger = logging.getLogger(__name__)
+
 router = APIRouter()


@@ -254,7 +257,9 @@ async def payment_callback(
        return {"success": True, "message": "支付成功", "record_id": record_id}
    except Exception as e:
        session.rollback()
-        raise HTTPException(status_code=500, detail=f"支付处理失败: {str(e)}")
+        logger.error(f"支付回调处理失败: user_id={user_id}, plan={plan}, error={e}")
+        # 不返回原始异常信息，避免泄漏内部实现细节
+        raise HTTPException(status_code=500, detail="支付处理失败，请稍后重试") from e
    finally:
        session.close()

@@ -24,8 +24,6 @@ from app.schemas.task_center import (
 from fastapi import APIRouter, Depends, HTTPException, Query

 from packages.application import (
-    CreateGenerationTaskCommand,
-    CreateGenerationTaskUseCase,
    RetryGenerationTaskUseCase,
    SubmitIngestJobCommand,
    SubmitIngestJobUseCase,
@@ -368,9 +366,9 @@ def retry_project_task(
            raise HTTPException(status_code=404, detail="Ingest job not found")
        if _status_value(job.status) != "failed":
            raise HTTPException(status_code=409, detail="Only failed tasks can be retried")
-        use_case = SubmitIngestJobUseCase(ingest_job_repository)
+        use_case = SubmitIngestJobUseCase(ingest_job_repository)  # type: ignore[assignment]
        retried = use_case.execute(
-            SubmitIngestJobCommand(
+            SubmitIngestJobCommand(  # type: ignore[arg-type]
                project_id=job.project_id,
                library_id=job.library_id,
                storage_key=job.storage_key,
@@ -386,6 +384,6 @@ def retry_project_task(
            current_step=_ingest_step(retried),
            source_id=retried.id,
            created_at=retried.created_at,
-            updated_at=retried.updated_at,
+            updated_at=retried.updated_at,  # type: ignore[attr-defined]
        )
    raise HTTPException(status_code=400, detail="Unsupported task type")
@@ -147,9 +147,9 @@ def get_template(
        use_case = GetTemplateUseCase(template_repository)
        template = use_case.execute(template_id, user_id)
        usage = template_repository.get_usage_count(template_id)
-    except Exception:
+    except Exception as _e:
        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")
+        raise HTTPException(status_code=status.HTTP_500_INTERNAL_SERVER_ERROR, detail="模板查询失败") from _e
    if template is None:
        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Template not found")
    return _to_response(template, usage_count=usage)
@@ -186,7 +186,7 @@ def create_template(
    try:
        template = use_case.execute(command)
    except ValidationError as exc:
-        raise HTTPException(status_code=status.HTTP_422_UNPROCESSABLE_ENTITY, detail=str(exc))
+        raise HTTPException(status_code=status.HTTP_422_UNPROCESSABLE_ENTITY, detail=str(exc)) from exc
    return _to_response(template)


@@ -226,10 +226,10 @@ def update_template(
    use_case = UpdateTemplateUseCase(template_repository)
    try:
        template = use_case.execute(command)
-    except NotFoundError:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Template not found")
+    except NotFoundError as _e:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Template not found") from _e
    except ValidationError as exc:
-        raise HTTPException(status_code=status.HTTP_422_UNPROCESSABLE_ENTITY, detail=str(exc))
+        raise HTTPException(status_code=status.HTTP_422_UNPROCESSABLE_ENTITY, detail=str(exc)) from exc
    return _to_response(template)


@@ -264,10 +264,10 @@ def copy_template(
    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 NotFoundError as _e:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Template not found") from _e
    except ValidationError as exc:
-        raise HTTPException(status_code=status.HTTP_422_UNPROCESSABLE_ENTITY, detail=str(exc))
+        raise HTTPException(status_code=status.HTTP_422_UNPROCESSABLE_ENTITY, detail=str(exc)) from exc
    return _to_response(template)


@@ -299,9 +299,9 @@ def toggle_favorite(
    use_case = GetTemplateUseCase(template_repository)
    try:
        template = use_case.execute(template_id, user_id)
-    except Exception:
+    except Exception as _e:
        logger.exception("toggle_favorite 查询失败: template_id=%s", template_id)
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Template not found")
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Template not found") from _e
    if template is None:
        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Template not found")
    return ToggleFavoriteResponse(id=template_id, is_favorite=False)
@@ -326,10 +326,10 @@ def validate_template(
    use_case = ValidateTemplateUseCase(template_repository)
    try:
        result = use_case.execute(command)
-    except NotFoundError:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Template not found")
+    except NotFoundError as _e:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Template not found") from _e
    except ValidationError as exc:
-        raise HTTPException(status_code=status.HTTP_422_UNPROCESSABLE_ENTITY, detail=str(exc))
+        raise HTTPException(status_code=status.HTTP_422_UNPROCESSABLE_ENTITY, detail=str(exc)) from exc

    return ValidateTemplateResponse(
        template=_to_response(result.template),
@@ -375,7 +375,9 @@ def create_category(
    )


-@router.delete("/categories/{category_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None, response_class=Response)
+@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),
@@ -148,7 +148,7 @@ def create_title(
        raise HTTPException(
            status_code=status.HTTP_429_TOO_MANY_REQUESTS,
            detail=f"标题库配额已满（{exc.used}/{exc.limit}），请升级套餐",
-        )
+        ) from exc
    return _to_response(item)


@@ -172,8 +172,8 @@ def update_title(
    use_case = UpdateTitleLibraryUseCase(title_repository)
    try:
        item = use_case.execute(command)
-    except NotFoundError:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Title not found")
+    except NotFoundError as _e:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Title not found") from _e
    return _to_response(item)


@@ -236,8 +236,8 @@ def get_tts_job(
    use_case = GetTTSJobUseCase(repository)
    try:
        job = use_case.execute(job_id, user_id)
-    except TTSJobNotFoundError:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="TTS job not found")
+    except TTSJobNotFoundError as _e:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="TTS job not found") from _e
    return _to_response(job, sign_url)


@@ -253,8 +253,8 @@ def get_tts_job_status(
    use_case = GetTTSJobStatusUseCase(repository)
    try:
        job = use_case.execute(job_id, user_id)
-    except TTSJobNotFoundError:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="TTS job not found")
+    except TTSJobNotFoundError as _e:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="TTS job not found") from _e
    output_url = job.output_audio_url
    if output_url:
        output_url = sign_url(output_url)
@@ -309,8 +309,8 @@ def save_tts_job_to_library(
    get_use_case = GetTTSJobUseCase(tts_repository)
    try:
        job = get_use_case.execute(job_id, user_id)
-    except TTSJobNotFoundError:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="TTS job not found")
+    except TTSJobNotFoundError as _e:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="TTS job not found") from _e

    # 校验已完成
    if not job.is_completed:
@@ -363,7 +363,7 @@ def save_tts_job_to_library(
        raise HTTPException(
            status_code=status.HTTP_429_TOO_MANY_REQUESTS,
            detail=f"配音库配额已满（{exc.used}/{exc.limit}），请升级套餐",
-        )
+        ) from exc

    return SaveToLibraryResponse(
        id=item.id,
@@ -1,7 +1,5 @@
 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
@@ -110,7 +108,9 @@ def update_video_review_status(
    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)
+    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)


@@ -121,7 +121,7 @@ def batch_download_videos(
    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:
@@ -154,7 +154,7 @@ def get_batch_download_status(
    from celery.result import AsyncResult

    task = AsyncResult(job_id, app=celery_app)
-    
+
    status_map = {
        "PENDING": "pending",
        "STARTED": "running",
@@ -141,8 +141,8 @@ def get_voice_clone(
    use_case = GetVoiceCloneUseCase(repository)
    try:
        profile = use_case.execute(clone_id, user_id)
-    except VoiceCloneNotFoundError:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Voice clone not found")
+    except VoiceCloneNotFoundError as _e:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Voice clone not found") from _e
    return _to_response(profile)


@@ -157,8 +157,8 @@ def get_voice_clone_status(
    use_case = GetVoiceCloneStatusUseCase(repository)
    try:
        profile = use_case.execute(clone_id, user_id)
-    except VoiceCloneNotFoundError:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Voice clone not found")
+    except VoiceCloneNotFoundError as _e:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Voice clone not found") from _e
    return VoiceCloneStatusResponse(
        id=profile.id,
        status=profile.status,
@@ -201,13 +201,13 @@ def retry_voice_clone(
    user_id = authenticated_user.user.id
    try:
        profile = workflow.retry_clone(clone_id, user_id)
-    except VoiceCloneNotFoundError:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Voice clone not found")
-    except VoiceCloneNotRetryableError:
+    except VoiceCloneNotFoundError as _e:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Voice clone not found") from _e
+    except VoiceCloneNotRetryableError as _e:
        raise HTTPException(
            status_code=status.HTTP_400_BAD_REQUEST,
            detail="Voice clone is not retryable (only failed clones can be retried)",
-        )
+        ) from _e

    # 如果 profile 处于 processing 且有 task_id，触发 Celery 异步轮询
    task_id = (profile.metadata or {}).get("cosyvoice_task_id", "")
@@ -287,7 +287,7 @@ def create_voice(
        raise HTTPException(
            status_code=status.HTTP_429_TOO_MANY_REQUESTS,
            detail=f"配音库配额已满（{exc.used}/{exc.limit}），请升级套餐",
-        )
+        ) from exc
    return _to_response(item, sign_url)


@@ -317,8 +317,8 @@ def update_voice(
    use_case = UpdateVoiceLibraryUseCase(voice_repository)
    try:
        item = use_case.execute(command)
-    except NotFoundError:
-        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Voice not found")
+    except NotFoundError as _e:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Voice not found") from _e
    return _to_response(item, sign_url)


@@ -141,7 +141,7 @@ def safe_enqueue_generation_task(
            global_pending_limit,
            user_id or "unknown",
        )
-        exc = GlobalQueueFull(pending_count=global_pending, limit=global_pending_limit)
+        exc: Exception = GlobalQueueFull(pending_count=global_pending, limit=global_pending_limit)
        _mark_task_failed_safely(task, generation_task_repository, log_prefix, str(exc))
        raise exc

@@ -194,7 +194,7 @@ def safe_enqueue_generation_task(
    if global_over or user_over:
        if global_over:
            reason = f"全局 pending 超限(入队后): {global_after}/{global_pending_limit}"
-            exc: Exception = GlobalQueueFull(pending_count=global_after, limit=global_pending_limit)
+            exc = GlobalQueueFull(pending_count=global_after, limit=global_pending_limit)
        else:
            reason = f"用户 pending 超限(入队后): {user_after}/{user_pending_limit}"
            exc = UserPendingLimitExceeded(user_id=user_id, pending_count=user_after, limit=user_pending_limit)
@@ -132,7 +132,7 @@ def get_tag_repository(
    session: Session = Depends(get_db_session),
 ) -> TagRepository:
    """Provide the SQLAlchemy tag repository implementation."""
-    return SQLAlchemyTagRepository(session)
+    return SQLAlchemyTagRepository(session)  # type: ignore[return-value]


 def get_user_repository(
@@ -105,7 +105,7 @@ class RateLimitMiddleware(BaseHTTPMiddleware):
        self.max_requests = max_requests
        self.window_seconds = window_seconds
        self.paths = set(paths) if paths else None
-        self.requests = {}  # {ip: [timestamps]}
+        self.requests: dict[str, list[float]] = {}

    async def dispatch(self, request: Request, call_next):
        # 如果配置了路径过滤，只对指定路径限流
@@ -12,6 +12,7 @@
 from __future__ import annotations

 import logging
+from collections.abc import Mapping
 from dataclasses import dataclass

 from sqlalchemy.orm import Session
@@ -155,7 +156,7 @@ class AutoClipService:
        self,
        clip: EditPlanClip,
        project_id: str,
-        config_map: dict[str, object],
+        config_map: Mapping[str, object],
    ) -> ClipAssignDetail:
        """为单个片段分配素材。"""
        config = config_map.get(clip.template_clip_config_id) if clip.template_clip_config_id else None
@@ -0,0 +1,276 @@
+"""封面管理服务.
+
+提供封面配置管理和从视频抽帧生成封面的能力。
+抽帧使用 FFmpeg，上传使用共享存储服务。
+"""
+
+from __future__ import annotations
+
+import logging
+import tempfile
+from pathlib import Path
+from typing import Any, Dict
+
+logger = logging.getLogger(__name__)
+
+
+# ── 常量 ──────────────────────────────────────────────────────────────────────
+
+DEFAULT_COVER_WIDTH = 1080
+DEFAULT_COVER_HEIGHT = 1920
+DEFAULT_COVER_QUALITY = 5  # JPEG quality (1-31, 越小越好)
+COVER_STORAGE_PREFIX = "covers"
+
+
+class CoverService:
+    """封面管理服务."""
+
+    def __init__(self, storage_service: Any, asset_repository: Any) -> None:
+        self._storage = storage_service
+        self._asset_repo = asset_repository
+
+    # ── 配置读写 ──────────────────────────────────────────────────────────
+
+    @staticmethod
+    def get_cover_config(plan_config: Dict[str, Any]) -> Dict[str, Any]:
+        """从 plan.config 中提取封面配置.
+
+        Args:
+            plan_config: 剪辑计划的 config 字段
+
+        Returns:
+            封面配置 dict
+        """
+        cover = plan_config.get("cover", {})
+        if not isinstance(cover, dict):
+            cover = {}
+        # 确保默认字段存在
+        return {
+            "type": cover.get("type", "ai_frame"),
+            "image_url": cover.get("image_url", ""),
+            "frame_time": cover.get("frame_time"),
+        }
+
+    # ── 抽帧生成封面 ──────────────────────────────────────────────────────
+
+    def extract_cover_from_clip(
+        self,
+        plan_id: str,
+        asset_id: str,
+        frame_time: float = 1.0,
+        *,
+        width: int = DEFAULT_COVER_WIDTH,
+        height: int = DEFAULT_COVER_HEIGHT,
+        quality: int = DEFAULT_COVER_QUALITY,
+    ) -> Dict[str, Any]:
+        """从指定素材的指定时间点抽取一帧作为封面.
+
+        Args:
+            plan_id: 剪辑计划 ID（用于生成存储路径）
+            asset_id: 素材 ID
+            frame_time: 抽帧时间点（秒）
+            width: 输出宽度
+            height: 输出高度
+            quality: JPEG 质量
+
+        Returns:
+            封面数据 dict，包含 type / image_url / frame_time
+
+        Raises:
+            ValueError: 素材不存在或不是视频
+            RuntimeError: 抽帧或上传失败
+        """
+        # 1. 获取素材
+        asset = self._asset_repo.get(asset_id) if self._asset_repo else None
+        if not asset:
+            raise ValueError(f"素材不存在: {asset_id}")
+
+        storage_key = getattr(asset, "storage_key", "")
+        if not storage_key:
+            raise ValueError(f"素材没有文件: {asset_id}")
+
+        mime_type = getattr(asset, "mime_type", "")
+        if mime_type and not mime_type.startswith("video"):
+            raise ValueError(f"素材不是视频类型: {mime_type}")
+
+        # 2. 下载视频到临时目录
+        with tempfile.TemporaryDirectory(prefix="cover_extract_") as tmp_dir:
+            tmp_path = Path(tmp_dir)
+            video_path = tmp_path / f"source_{asset_id[:8]}"
+
+            logger.info("下载素材用于封面抽帧: asset_id=%s", asset_id)
+            try:
+                self._storage.download_file(storage_key, str(video_path))
+            except Exception as e:
+                raise RuntimeError(f"下载素材失败: {e}") from e
+
+            if not video_path.exists() or video_path.stat().st_size == 0:
+                raise RuntimeError("下载的素材文件为空")
+
+            # 3. FFmpeg 抽帧
+            output_path = tmp_path / "cover.jpg"
+            self._extract_frame(
+                video_path=video_path,
+                output_path=output_path,
+                time_sec=frame_time,
+                width=width,
+                height=height,
+                quality=quality,
+            )
+
+            if not output_path.exists() or output_path.stat().st_size == 0:
+                raise RuntimeError("封面抽帧失败")
+
+            # 4. 上传到 OSS
+            cover_key = f"{COVER_STORAGE_PREFIX}/{plan_id}/cover_{int(frame_time * 1000)}.jpg"
+            logger.info("上传封面到存储: key=%s", cover_key)
+
+            try:
+                self._storage.upload_file(
+                    file_or_path=str(output_path),
+                    storage_key=cover_key,
+                    content_type="image/jpeg",
+                )
+            except Exception as e:
+                raise RuntimeError(f"上传封面失败: {e}") from e
+
+            # 5. 获取访问 URL
+            try:
+                image_url = self._storage.get_url(cover_key)
+            except Exception:
+                image_url = cover_key  # 降级为 storage_key
+
+        logger.info(
+            "封面抽帧完成: plan_id=%s asset_id=%s time=%.2fs size=%d",
+            plan_id,
+            asset_id,
+            frame_time,
+            output_path.stat().st_size if output_path.exists() else 0,
+        )
+
+        return {
+            "type": "manual",
+            "image_url": image_url,
+            "frame_time": frame_time,
+        }
+
+    def generate_smart_cover(
+        self,
+        plan_id: str,
+        asset_id: str,
+        *,
+        width: int = DEFAULT_COVER_WIDTH,
+        height: int = DEFAULT_COVER_HEIGHT,
+        quality: int = DEFAULT_COVER_QUALITY,
+    ) -> Dict[str, Any]:
+        """智能选帧：从视频中选取多帧，选最清晰的一帧.
+
+        Args:
+            plan_id: 剪辑计划 ID
+            asset_id: 素材 ID
+            width: 输出宽度
+            height: 输出高度
+            quality: JPEG 质量
+
+        Returns:
+            封面数据 dict
+        """
+        # 简单实现：取视频 1/3 处的帧作为智能封面
+        # 更复杂的多帧选清晰帧可以后续优化
+        frame_time = 3.0  # 默认第3秒，后续可以根据视频时长动态计算
+
+        result = self.extract_cover_from_clip(
+            plan_id=plan_id,
+            asset_id=asset_id,
+            frame_time=frame_time,
+            width=width,
+            height=height,
+            quality=quality,
+        )
+        result["type"] = "ai_frame"
+        return result
+
+    # ── 内部方法 ──────────────────────────────────────────────────────────
+
+    @staticmethod
+    def _extract_frame(
+        video_path: Path,
+        output_path: Path,
+        *,
+        time_sec: float,
+        width: int,
+        height: int,
+        quality: int,
+    ) -> None:
+        """使用 FFmpeg 从视频中抽取一帧.
+
+        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
+        """
+        import subprocess
+
+        # scale + crop 实现 cover 裁剪
+        vf = f"scale={width}:{height}:force_original_aspect_ratio=increase," f"crop={width}:{height}"
+
+        command = [
+            "ffmpeg",
+            "-y",
+            "-ss",
+            f"{time_sec:.3f}",
+            "-i",
+            str(video_path),
+            "-vframes",
+            "1",
+            "-vf",
+            vf,
+            "-q:v",
+            str(quality),
+            "-f",
+            "mjpeg",
+            str(output_path),
+        ]
+
+        logger.debug("FFmpeg 抽帧命令: %s", " ".join(command))
+
+        try:
+            result = subprocess.run(
+                command,
+                capture_output=True,
+                text=True,
+                timeout=60,
+            )
+            if result.returncode != 0:
+                logger.warning("FFmpeg 抽帧返回非零: %s\nstderr: %s", result.returncode, result.stderr[-500:])
+                # 尝试不使用 scale+crop 的简化命令
+                simple_command = [
+                    "ffmpeg",
+                    "-y",
+                    "-ss",
+                    f"{time_sec:.3f}",
+                    "-i",
+                    str(video_path),
+                    "-vframes",
+                    "1",
+                    "-q:v",
+                    str(quality),
+                    "-f",
+                    "mjpeg",
+                    str(output_path),
+                ]
+                result2 = subprocess.run(
+                    simple_command,
+                    capture_output=True,
+                    text=True,
+                    timeout=60,
+                )
+                if result2.returncode != 0:
+                    raise RuntimeError(f"FFmpeg 抽帧失败: {result2.stderr[-300:]}")
+        except subprocess.TimeoutExpired as e:
+            raise RuntimeError("FFmpeg 抽帧超时") from e
+        except FileNotFoundError as e:
+            raise RuntimeError("FFmpeg 不可用") from e
@@ -141,6 +141,19 @@ class EditPlanService:
        logger.info("创建剪辑计划: id=%s name=%s", created.id, created.name)
        return created

+    def _auto_resume_editing(self, plan_id: str) -> None:
+        """如果计划处于 completed/failed 状态，自动切回 editing（编辑操作前置）"""
+        plan = self._plan_repo.get(plan_id)
+        if plan is None:
+            return
+        if plan.status in (EditPlanStatus.COMPLETED, EditPlanStatus.FAILED):
+            try:
+                plan.resume_editing()
+                self._plan_repo.update(plan)
+                logger.info("自动重新编辑: plan_id=%s", plan_id)
+            except ValueError:
+                pass
+
    def update_plan(
        self,
        plan_id: str,
@@ -156,6 +169,10 @@ class EditPlanService:
        """
        existing = self.get_plan_or_raise(plan_id)

+        # 自动从 completed/failed 切回 editing
+        self._auto_resume_editing(plan_id)
+        existing = self.get_plan_or_raise(plan_id)
+
        updated = EditPlan(
            id=existing.id,
            template_id=existing.template_id,
@@ -212,8 +229,24 @@ class EditPlanService:
            return plan

        # 根据目标状态调用对应的状态机方法
+        # EDITING 支持从 draft / completed / failed 进入
+        if target_status == EditPlanStatus.EDITING:
+            if plan.status == EditPlanStatus.DRAFT:
+                plan.start_editing()
+            elif plan.status in (EditPlanStatus.COMPLETED, EditPlanStatus.FAILED):
+                plan.resume_editing()
+            else:
+                raise ValueError(f"无法从 {plan.status} 切换到 {target_status}")
+            result = self._plan_repo.update(plan)
+            logger.info(
+                "状态流转: plan_id=%s %s → %s",
+                plan_id,
+                plan.status,
+                target_status,
+            )
+            return result
+
        transition_map = {
-            EditPlanStatus.EDITING: plan.start_editing,
            EditPlanStatus.RENDERING: plan.start_rendering,
            EditPlanStatus.COMPLETED: plan.mark_completed,
            EditPlanStatus.FAILED: plan.mark_failed,
@@ -292,6 +325,8 @@ class EditPlanService:
        """
        # 确保计划存在
        self.get_plan_or_raise(plan_id)
+        # 自动从 completed/failed 切回 editing
+        self._auto_resume_editing(plan_id)

        clip = EditPlanClip.create(
            plan_id=plan_id,
@@ -339,6 +374,9 @@ class EditPlanService:
        """
        existing = self.get_clip_or_raise(clip_id)

+        # 自动从 completed/failed 切回 editing
+        self._auto_resume_editing(existing.plan_id)
+
        # 速度边界钳制
        if playback_speed is not None:
            if playback_speed <= 0:
@@ -381,6 +419,8 @@ class EditPlanService:
            ValueError: 片段不存在或 asset_id 为空
        """
        clip = self.get_clip_or_raise(clip_id)
+        # 自动从 completed/failed 切回 editing
+        self._auto_resume_editing(clip.plan_id)
        clip.assign_asset(asset_id)
        result = self._clip_repo.update(clip)
        logger.info("分配素材: clip_id=%s asset_id=%s", clip_id, asset_id)
@@ -407,7 +447,463 @@ class EditPlanService:
        logger.info("删除所有片段: plan_id=%s count=%d", plan_id, count)
        return count

+    def create_clips_from_assets(
+        self,
+        plan_id: str,
+        asset_ids: list[str],
+        *,
+        clip_type: str = "main",
+    ) -> list[EditPlanClip]:
+        """从素材批量创建片段（追加到时间线末尾）。
+
+        Args:
+            plan_id: 计划 ID
+            asset_ids: 素材 ID 列表（按顺序追加）
+            clip_type: 片段类型
+
+        Returns:
+            list[EditPlanClip]: 创建的片段列表
+        """
+        if not asset_ids:
+            return []
+
+        # 确保计划存在 + 自动回退状态
+        self.get_plan_or_raise(plan_id)
+        self._auto_resume_editing(plan_id)
+
+        # 查询素材信息（取 duration）
+        from packages.adapters.sqlalchemy_impl.models import AssetModel
+
+        session = self._clip_repo.session  # type: ignore[attr-defined]
+        assets = session.query(AssetModel).filter(AssetModel.id.in_(asset_ids)).all()
+        asset_map = {a.id: a for a in assets}
+
+        # 从现有片段数量开始追加
+        existing_count = self._clip_repo.count(plan_id=plan_id)
+
+        # 批量创建片段
+        created: list[EditPlanClip] = []
+        for i, asset_id in enumerate(asset_ids):
+            asset = asset_map.get(asset_id)
+            duration = asset.duration if asset and asset.duration else 0.0
+
+            clip = self.create_clip(
+                plan_id=plan_id,
+                clip_type=clip_type,
+                order=existing_count + i,
+                asset_id=asset_id,
+                duration=duration,
+            )
+            created.append(clip)
+
+        logger.info(
+            "从素材批量创建片段: plan_id=%s count=%d",
+            plan_id,
+            len(created),
+        )
+        return created
+
    # ── 渲染生成流程 ────────────────────────────────────────────────────────
+    # ── 片段分割与合并 ──────────────────────────────────────────────────────
+
+    def split_clip(self, clip_id: str, split_time: float) -> Dict[str, Any]:
+        """将一个片段从指定位置分割为两个片段
+
+        Args:
+            clip_id: 要分割的片段 ID
+            split_time: 分割点（相对于片段起始的秒数），必须在 (0, duration) 范围内
+
+        Returns:
+            dict: {"left_clip": EditPlanClip, "right_clip": EditPlanClip}
+
+        Raises:
+            ValueError: 片段不存在、分割时间越界
+        """
+        clip = self.get_clip_or_raise(clip_id)
+        plan_id = clip.plan_id
+
+        if split_time <= 0 or split_time >= clip.duration:
+            raise ValueError(f"分割时间必须在 (0, {clip.duration:.3f}) 范围内，当前: {split_time}")
+
+        self._auto_resume_editing(plan_id)
+
+        original_duration = clip.duration
+        left_duration = round(split_time, 3)
+        right_duration = round(original_duration - split_time, 3)
+        original_order = clip.order
+
+        # 更新左半部分（原片段）
+        clip.duration = left_duration
+        left_clip = self._clip_repo.update(clip)
+
+        # 后面片段的 order 全部 +1（给右半部分腾位置）
+        all_clips = self._clip_repo.list_by_plan(plan_id)
+        for c in all_clips:
+            if c.order > original_order and c.id != clip_id:
+                c.order += 1
+                self._clip_repo.update(c)
+
+        # 创建右半部分新片段（继承原片段的大部分属性）
+        right_config = dict(clip.config) if clip.config else {}
+        # 素材裁剪信息
+        if clip.asset_id:
+            # 右半部分从 split_time 开始播放
+            right_config["trim_start"] = left_duration
+            # 左半部分在 split_time 处结束
+            left_config = dict(left_clip.config) if left_clip.config else {}
+            left_config["trim_end"] = right_duration
+            left_clip.config = left_config
+            left_clip = self._clip_repo.update(left_clip)
+
+        right_clip = EditPlanClip.create(
+            plan_id=plan_id,
+            clip_type=clip.clip_type,
+            order=original_order + 1,
+            template_clip_config_id=clip.template_clip_config_id,
+            asset_id=clip.asset_id,
+            text_content=clip.text_content,
+            start_time=clip.start_time + left_duration,
+            duration=right_duration,
+            transition_effect=clip.transition_effect,
+            transition_duration=clip.transition_duration,
+            playback_speed=clip.playback_speed,
+            config=right_config,
+        )
+        created_right = self._clip_repo.create(right_clip)
+
+        logger.info(
+            "分割片段: clip_id=%s plan_id=%s split_time=%.3fs left_dur=%.3fs right_dur=%.3fs",
+            clip_id,
+            plan_id,
+            split_time,
+            left_duration,
+            right_duration,
+        )
+
+        return {
+            "left_clip": left_clip,
+            "right_clip": created_right,
+        }
+
+    def merge_clips(self, clip_ids: List[str]) -> EditPlanClip:
+        """合并多个连续片段为一个片段
+
+        Args:
+            clip_ids: 要合并的片段 ID 列表（至少2个），必须属于同一个计划且 order 连续
+
+        Returns:
+            EditPlanClip: 合并后的新片段
+
+        Raises:
+            ValueError: 数量不足、不属于同一计划、不连续、类型不一致
+        """
+        if len(clip_ids) < 2:
+            raise ValueError("至少需要 2 个片段才能合并")
+
+        # 读取所有片段
+        clips = []
+        for cid in clip_ids:
+            clip = self.get_clip_or_raise(cid)
+            clips.append(clip)
+
+        # 校验：同一计划
+        plan_id = clips[0].plan_id
+        for c in clips[1:]:
+            if c.plan_id != plan_id:
+                raise ValueError("只能合并同一计划下的片段")
+
+        # 按 order 排序
+        clips.sort(key=lambda c: c.order)
+
+        # 校验：order 连续
+        for i in range(1, len(clips)):
+            if clips[i].order != clips[i - 1].order + 1:
+                raise ValueError(f"片段不连续：order {clips[i-1].order} → {clips[i].order}")
+
+        # 校验：类型一致
+        clip_type = clips[0].clip_type
+        for c in clips[1:]:
+            if c.clip_type != clip_type:
+                raise ValueError("只能合并相同类型的片段")
+
+        self._auto_resume_editing(plan_id)
+
+        # 计算合并后的属性
+        first_clip = clips[0]
+        total_duration = round(sum(c.duration for c in clips), 3)
+        first_order = first_clip.order
+
+        # 合并文案（用换行连接）
+        merged_text = "\n".join(c.text_content for c in clips if c.text_content.strip())
+
+        # 合并 config（后面的覆盖前面的）
+        merged_config: Dict[str, Any] = {}
+        for c in clips:
+            if c.config:
+                merged_config.update(c.config)
+        # 清理 trim 相关字段（合并后就是完整片段了）
+        merged_config.pop("trim_start", None)
+        merged_config.pop("trim_end", None)
+
+        # 更新第一个片段（保留它作为合并结果）
+        first_clip.duration = total_duration
+        first_clip.text_content = merged_text
+        first_clip.config = merged_config
+        # 转场保留第一个的（合并后的入点转场）
+        # playback_speed 取第一个的
+        merged_clip = self._clip_repo.update(first_clip)
+
+        # 删除其余片段
+        for c in clips[1:]:
+            self._clip_repo.delete(c.id)
+
+        # 后面的片段 order 前移 (len - 1) 位
+        shift = len(clips) - 1
+        all_clips = self._clip_repo.list_by_plan(plan_id)
+        for c in all_clips:
+            if c.order > first_order and c.id != merged_clip.id:
+                c.order -= shift
+                self._clip_repo.update(c)
+
+        logger.info(
+            "合并片段: plan_id=%s count=%d total_duration=%.3fs",
+            plan_id,
+            len(clips),
+            total_duration,
+        )
+
+        return merged_clip
+
+    # ── 字幕管理 ──────────────────────────────────────────────────────────
+
+    def list_subtitles(self, clip_id: str) -> List[Dict[str, Any]]:
+        """获取片段的所有字幕
+
+        Returns:
+            List[dict]: 字幕列表，按 start 时间排序
+        """
+        clip = self.get_clip_or_raise(clip_id)
+        config = clip.config or {}
+        subtitles = config.get("subtitles", [])
+        # 按开始时间排序
+        subtitles.sort(key=lambda s: s.get("start", 0))
+        return subtitles
+
+    def get_subtitle(self, clip_id: str, subtitle_id: str) -> Optional[Dict[str, Any]]:
+        """获取单条字幕"""
+        subtitles = self.list_subtitles(clip_id)
+        for s in subtitles:
+            if s.get("id") == subtitle_id:
+                return s
+        return None
+
+    def add_subtitle(
+        self,
+        clip_id: str,
+        start: float,
+        end: float,
+        text: str,
+        *,
+        style: Optional[Dict[str, Any]] = None,
+    ) -> Dict[str, Any]:
+        """添加一条字幕
+
+        Args:
+            clip_id: 片段 ID
+            start: 开始时间（秒，相对于片段）
+            end: 结束时间（秒）
+            text: 字幕文本
+            style: 样式配置（字体、大小、颜色、位置等）
+
+        Returns:
+            dict: 新增的字幕条目
+
+        Raises:
+            ValueError: 时间非法或文本为空
+        """
+        clip = self.get_clip_or_raise(clip_id)
+
+        if start < 0 or end <= start:
+            raise ValueError(f"字幕时间非法: start={start}, end={end}")
+        if not text.strip():
+            raise ValueError("字幕文本不能为空")
+        if end > clip.duration + 0.001:
+            raise ValueError(f"字幕结束时间不能超过片段时长: end={end:.3f}, duration={clip.duration:.3f}")
+
+        self._auto_resume_editing(clip.plan_id)
+
+        from uuid import uuid4
+
+        config = dict(clip.config) if clip.config else {}
+        subtitles = list(config.get("subtitles", []))
+
+        subtitle = {
+            "id": uuid4().hex,
+            "start": round(start, 3),
+            "end": round(end, 3),
+            "text": text.strip(),
+            "style": style or {},
+        }
+        subtitles.append(subtitle)
+        subtitles.sort(key=lambda s: s.get("start", 0))
+
+        config["subtitles"] = subtitles
+        clip.config = config
+        self._clip_repo.update(clip)
+
+        logger.info(
+            "添加字幕: clip_id=%s subtitle_id=%s start=%.3fs end=%.3fs",
+            clip_id,
+            subtitle["id"],
+            start,
+            end,
+        )
+
+        return subtitle
+
+    def update_subtitle(
+        self,
+        clip_id: str,
+        subtitle_id: str,
+        *,
+        start: Optional[float] = None,
+        end: Optional[float] = None,
+        text: Optional[str] = None,
+        style: Optional[Dict[str, Any]] = None,
+    ) -> Dict[str, Any]:
+        """更新一条字幕
+
+        Returns:
+            dict: 更新后的字幕条目
+
+        Raises:
+            ValueError: 字幕不存在或参数非法
+        """
+        clip = self.get_clip_or_raise(clip_id)
+        config = dict(clip.config) if clip.config else {}
+        subtitles = list(config.get("subtitles", []))
+
+        found = False
+        for i, s in enumerate(subtitles):
+            if s.get("id") == subtitle_id:
+                # 更新字段
+                updated_s = dict(s)
+                if start is not None:
+                    updated_s["start"] = round(start, 3)
+                if end is not None:
+                    updated_s["end"] = round(end, 3)
+                if text is not None:
+                    if not text.strip():
+                        raise ValueError("字幕文本不能为空")
+                    updated_s["text"] = text.strip()
+                if style is not None:
+                    updated_s["style"] = style
+
+                # 校验时间
+                if updated_s["start"] < 0 or updated_s["end"] <= updated_s["start"]:
+                    raise ValueError(f"字幕时间非法: start={updated_s['start']}, end={updated_s['end']}")
+                if updated_s["end"] > clip.duration + 0.001:
+                    raise ValueError("字幕结束时间不能超过片段时长")
+
+                subtitles[i] = updated_s
+                found = True
+                break
+
+        if not found:
+            raise ValueError(f"字幕不存在: {subtitle_id}")
+
+        self._auto_resume_editing(clip.plan_id)
+
+        subtitles.sort(key=lambda s: s.get("start", 0))
+        config["subtitles"] = subtitles
+        clip.config = config
+        self._clip_repo.update(clip)
+
+        logger.info("更新字幕: clip_id=%s subtitle_id=%s", clip_id, subtitle_id)
+
+        return subtitles[next(i for i, s in enumerate(subtitles) if s["id"] == subtitle_id)]
+
+    def delete_subtitle(self, clip_id: str, subtitle_id: str) -> bool:
+        """删除一条字幕
+
+        Returns:
+            bool: 是否删除成功
+        """
+        clip = self.get_clip_or_raise(clip_id)
+        config = dict(clip.config) if clip.config else {}
+        subtitles = list(config.get("subtitles", []))
+
+        new_subtitles = [s for s in subtitles if s.get("id") != subtitle_id]
+        if len(new_subtitles) == len(subtitles):
+            return False
+
+        self._auto_resume_editing(clip.plan_id)
+
+        config["subtitles"] = new_subtitles
+        clip.config = config
+        self._clip_repo.update(clip)
+
+        logger.info("删除字幕: clip_id=%s subtitle_id=%s", clip_id, subtitle_id)
+        return True
+
+    def batch_update_subtitles(
+        self,
+        clip_id: str,
+        subtitles: List[Dict[str, Any]],
+    ) -> List[Dict[str, Any]]:
+        """批量更新字幕（全量替换，用于批量编辑或导入）
+
+        Args:
+            clip_id: 片段 ID
+            subtitles: 字幕列表，每条需包含 start/end/text，已有 id 则保留
+
+        Returns:
+            List[dict]: 更新后的字幕列表
+        """
+        clip = self.get_clip_or_raise(clip_id)
+
+        from uuid import uuid4
+
+        validated = []
+        for s in subtitles:
+            start = float(s.get("start", 0))
+            end = float(s.get("end", 0))
+            text = str(s.get("text", ""))
+
+            if start < 0 or end <= start:
+                raise ValueError(f"字幕时间非法: start={start}, end={end}")
+            if not text.strip():
+                continue  # 跳过空字幕
+            if end > clip.duration + 0.001:
+                raise ValueError(f"字幕结束时间不能超过片段时长: end={end}")
+
+            subtitle_id = s.get("id") or uuid4().hex
+            validated.append(
+                {
+                    "id": subtitle_id,
+                    "start": round(start, 3),
+                    "end": round(end, 3),
+                    "text": text.strip(),
+                    "style": s.get("style", {}),
+                }
+            )
+
+        validated.sort(key=lambda s: s["start"])
+
+        self._auto_resume_editing(clip.plan_id)
+
+        config = dict(clip.config) if clip.config else {}
+        config["subtitles"] = validated
+        clip.config = config
+        self._clip_repo.update(clip)
+
+        logger.info(
+            "批量更新字幕: clip_id=%s count=%d",
+            clip_id,
+            len(validated),
+        )
+
+        return validated

    def get_plan_with_clips(self, plan_id: str) -> Dict[str, Any]:
        """获取计划及其所有片段
@@ -431,6 +927,8 @@ class EditPlanService:
                "clips": List[EditPlanClip],
                "generation_task_id": Optional[str],
                "generation_task_status": Optional[str],
+                "progress": float,
+                "error_message": str,
            }

        Raises:
@@ -442,17 +940,23 @@ class EditPlanService:
        # 从 plan.config 中获取 generation_task_id
        generation_task_id = plan.config.get("generation_task_id")
        generation_task_status = None
+        progress = 0.0
+        error_message = ""

        if generation_task_id:
            task = self._generation_task_repo.get(generation_task_id)
            if task:
                generation_task_status = task.status.value if hasattr(task.status, "value") else task.status
+                progress = getattr(task, "progress", 0.0) or 0.0
+                error_message = getattr(task, "error_message", "") or ""

        return {
            "plan": plan,
            "clips": clips,
            "generation_task_id": generation_task_id,
            "generation_task_status": generation_task_status,
+            "progress": progress,
+            "error_message": error_message,
        }

    def can_generate(self, plan_id: str) -> tuple[bool, str]:
@@ -503,6 +1007,9 @@ class EditPlanService:
            更新后的计划
        """
        plan = self.get_plan_or_raise(plan_id)
+        # 自动从 completed/failed 切回 editing
+        self._auto_resume_editing(plan_id)
+        plan = self.get_plan_or_raise(plan_id)
        new_config = {**plan.config, **config_updates}

        updated = EditPlan(
@@ -519,3 +1026,85 @@ class EditPlanService:
            updated_at=plan.updated_at,
        )
        return self._plan_repo.update(updated)
+
+    # ── 复制计划 ────────────────────────────────────────────────────────────
+
+    def copy_plan(
+        self,
+        plan_id: str,
+        *,
+        new_name: Optional[str] = None,
+        project_id: Optional[str] = None,
+    ) -> EditPlan:
+        """复制一个剪辑计划（含所有片段配置）。
+
+        新计划状态为 editing，不含生成任务和结果记录。
+
+        Args:
+            plan_id: 源计划 ID
+            new_name: 新计划名称，不传则为「原名 - 副本」
+            project_id: 新计划的项目 ID，不传则复用源计划
+
+        Returns:
+            EditPlan: 新创建的计划
+
+        Raises:
+            ValueError: 源计划不存在
+        """
+        source = self.get_plan_or_raise(plan_id)
+        source_clips = self._clip_repo.list_by_plan(plan_id)
+
+        # 新计划名称
+        name = new_name or f"{source.name} - 副本"
+        new_project_id = project_id if project_id is not None else source.project_id
+
+        # 复制 plan 配置（去除渲染结果相关字段）
+        new_config = dict(source.config)
+        new_config.pop("rendered_url", None)
+        new_config.pop("rendered_storage_key", None)
+        new_config.pop("generation_task_id", None)
+
+        # 创建新计划
+        new_plan = EditPlan.create(
+            template_id=source.template_id,
+            name=name,
+            config=new_config,
+            total_duration=source.total_duration,
+            project_id=new_project_id,
+            created_by_user_id=source.created_by_user_id,
+            source_edit_plan_id=plan_id,
+        )
+        # 强制切到 editing 状态
+        if new_plan.status != EditPlanStatus.EDITING:
+            try:
+                new_plan.start_editing()
+            except ValueError:
+                pass
+
+        created_plan = self._plan_repo.create(new_plan)
+        logger.info(
+            "复制剪辑计划: source=%s target=%s name=%s clips=%d",
+            plan_id,
+            created_plan.id,
+            name,
+            len(source_clips),
+        )
+
+        # 复制所有片段
+        for clip in source_clips:
+            new_clip = self.create_clip(
+                plan_id=created_plan.id,
+                clip_type=clip.clip_type,
+                order=clip.order,
+                asset_id=clip.asset_id or "",
+                text_content=clip.text_content or "",
+                start_time=clip.start_time,
+                duration=clip.duration,
+                transition_effect=clip.transition_effect or "cut",
+                transition_duration=clip.transition_duration or 0.0,
+                playback_speed=clip.playback_speed or 1.0,
+                config=dict(clip.config) if clip.config else None,
+            )
+            logger.debug("复制片段: source=%s target=%s order=%d", clip.id, new_clip.id, clip.order)
+
+        return self.get_plan_or_raise(created_plan.id)
@@ -12,6 +12,8 @@ from typing import Any, List, Optional
 from sqlalchemy.orm import Session

 from packages.adapters.sqlalchemy_impl import (
+    SQLAlchemyEditPlanClipRepository,
+    SQLAlchemyEditPlanRepository,
    SQLAlchemyEditTemplateRepository,
    SQLAlchemyTemplateClipConfigRepository,
 )
@@ -37,6 +39,9 @@ class EditTemplateService:
    def __init__(self, db: Session) -> None:
        self._template_repo = SQLAlchemyEditTemplateRepository(db)
        self._clip_config_repo = SQLAlchemyTemplateClipConfigRepository(db)
+        self._plan_repo = SQLAlchemyEditPlanRepository(db)
+        self._plan_clip_repo = SQLAlchemyEditPlanClipRepository(db)
+        self._db = db

    # ── 模板 CRUD ──────────────────────────────────────────────────────────

@@ -394,3 +399,137 @@ class EditTemplateService:
            "template": template,
            "clip_configs": clip_configs,
        }
+
+    # ── 从剪辑计划保存为模板 ──────────────────────────────────────────────
+
+    def save_plan_as_template(
+        self,
+        plan_id: str,
+        name: str,
+        *,
+        description: str = "",
+        template_type: str = "custom",
+        preview_url: str = "",
+    ) -> dict[str, Any]:
+        """将剪辑计划保存为模板
+
+        将指定剪辑计划的配置和片段结构另存为一个新模板，
+        方便后续基于该模板快速创建新的剪辑计划。
+
+        转换规则:
+        - 计划名称 → 模板名称（调用方传入，支持自定义）
+        - 计划 config → 模板 config（整体迁移）
+        - 计划 editing_mode 从 config 中提取，默认 one_take
+        - 每个片段转换为模板片段配置:
+          - clip_type 直接映射
+          - order 保持不变
+          - duration → min_duration = max_duration = duration（固定时长）
+          - text_content → text_template
+          - transition_effect 直接映射
+          - playback_speed 等播放参数存入 config
+          - 不保留 asset_id（模板不绑定具体素材）
+
+        Args:
+            plan_id: 源剪辑计划 ID
+            name: 新模板名称
+            description: 模板描述
+            template_type: 模板类型，默认 custom（用户自定义）
+            preview_url: 预览图 URL
+
+        Returns:
+            dict: {"template": EditTemplate, "clip_configs": List[TemplateClipConfig]}
+
+        Raises:
+            ValueError: 计划不存在或名称为空/重复
+        """
+        # 1. 读取源计划
+        plan = self._plan_repo.get(plan_id)
+        if plan is None:
+            raise ValueError(f"剪辑计划不存在: {plan_id}")
+
+        # 2. 读取所有片段（按 order 排序）
+        clips = self._plan_clip_repo.list_by_plan(plan_id)
+        clips.sort(key=lambda c: c.order)
+
+        # 3. 提取 editing_mode
+        editing_mode = plan.config.get("editing_mode", "one_take") if plan.config else "one_take"
+
+        # 4. 创建模板（复用 create_template 的校验逻辑，但手动构建避免重复查询）
+        clean_name = name.strip()
+        if not clean_name:
+            raise ValueError("模板名称不能为空")
+
+        # 名称重复检查
+        existing = self._template_repo.list_all(skip=0, limit=1000)
+        for t in existing:
+            if t.name == clean_name and t.status == EditTemplateStatus.ACTIVE:
+                raise ValueError(f"模板名称已存在: {clean_name}")
+
+        # 从计划 config 中提取模板级配置，去掉运行时/素材相关字段
+        plan_config = plan.config or {}
+        template_config: dict[str, Any] = {}
+        for key, value in plan_config.items():
+            # 跳过明显的运行时/实例字段，保留风格/模式类配置
+            if key not in {"asset_ids", "source_edit_plan_id", "generation_task_id"}:
+                template_config[key] = value
+
+        template = EditTemplate.create(
+            name=clean_name,
+            description=description,
+            template_type=template_type,
+            editing_mode=editing_mode,
+            config=template_config,
+            preview_url=preview_url,
+        )
+        created_template = self._template_repo.create(template)
+        logger.info(
+            "从剪辑计划创建模板: plan_id=%s template_id=%s name=%s clip_count=%d",
+            plan_id,
+            created_template.id,
+            clean_name,
+            len(clips),
+        )
+
+        # 5. 转换每个片段为模板片段配置
+        created_configs: List[TemplateClipConfig] = []
+        for clip in clips:
+            clip_config: dict[str, Any] = {}
+            # 播放速度存入 config
+            if clip.playback_speed and clip.playback_speed != 1.0:
+                clip_config["playback_speed"] = clip.playback_speed
+            # 片段自有 config 合并（优先级：clip.config 覆盖上面的）
+            if clip.config:
+                clip_config.update(clip.config)
+            # 去掉素材相关字段
+            clip_config.pop("asset_info", None)
+            clip_config.pop("source_asset_id", None)
+
+            # 转场效果兼容校验
+            try:
+                transition = TransitionEffect(clip.transition_effect)
+            except ValueError:
+                transition = TransitionEffect.CUT
+
+            # 片段类型兼容校验
+            try:
+                clip_type = ClipType(clip.clip_type)
+            except ValueError:
+                clip_type = ClipType.MAIN
+
+            clip_config_obj = TemplateClipConfig.create(
+                template_id=created_template.id,
+                clip_type=clip_type,
+                order=clip.order,
+                min_duration=clip.duration,
+                max_duration=clip.duration,
+                text_template=clip.text_content or "",
+                transition_effect=transition,
+                config=clip_config,
+            )
+            created = self._clip_config_repo.create(clip_config_obj)
+            created_configs.append(created)
+
+        return {
+            "template": created_template,
+            "clip_configs": created_configs,
+        }
@@ -99,6 +99,11 @@ class PlanGeneratorService:
        # 3. 生成片段列表
        if clip_configs:
            clips = self._create_clips_from_configs(plan.id, clip_configs)
+            # 模板 clip_config 的 clip_type 是 ClipType 枚举（main/intro/outro 等），
+            # 但 PIP / VOICE_PIP 模式需要特定的 clip_type（overlay/background/corner_voice/b_roll）
+            # 才能让素材分配和渲染分层正确工作。
+            # 这里将 MAIN 类型的片段按顺序映射为对应模式的角色类型。
+            self._map_clip_types_for_mode(clips, editing_mode)
        else:
            clips = self._generate_default_clips(plan.id, editing_mode, len(asset_ids))

@@ -195,6 +200,41 @@ class PlanGeneratorService:

        return clips

+    def _map_clip_types_for_mode(self, clips: List[EditPlanClip], editing_mode: str) -> None:
+        """将模板 clip_config 生成的 MAIN 类型片段，按 editing_mode 映射为对应角色类型。
+
+        模板的 clip_config 使用 ClipType 枚举（main/intro/outro 等），
+        但 PIP / VOICE_PIP 模式的素材分配和渲染分层依赖特定的 clip_type 命名
+        （overlay / background / corner_voice / b_roll）。
+
+        映射规则（仅修改 MAIN 类型片段，非 MAIN 片段保持原类型）：
+        - PIP: 第1个 MAIN → main（背景），其余 MAIN → overlay（画中画）
+        - VOICE_PIP: 第1个 → background，第2个 → corner_voice，第3+个 → b_roll
+        - ONE_TAKE / VOICE_OVER: 保持 main 不变
+        """
+        from packages.domain.template_clip_config import ClipType
+
+        main_clips = [c for c in clips if c.clip_type == ClipType.MAIN.value]
+        if not main_clips:
+            return
+
+        if editing_mode == EditingMode.PIP.value:
+            # 第1个 main 保持（背景层），其余改为 overlay（画中画层）
+            for i, clip in enumerate(main_clips):
+                if i > 0:
+                    clip.clip_type = "overlay"
+
+        elif editing_mode == EditingMode.VOICE_PIP.value:
+            for i, clip in enumerate(main_clips):
+                if i == 0:
+                    clip.clip_type = "background"
+                elif i == 1:
+                    clip.clip_type = "corner_voice"
+                else:
+                    clip.clip_type = "b_roll"
+
+        # ONE_TAKE / VOICE_OVER: 保持 main 不变，无需处理
+
    def _generate_default_clips(
        self,
        plan_id: str,
@@ -224,7 +264,7 @@ class PlanGeneratorService:
            )
            order += 1
            # 剩余为 overlay
-            for i in range(1, n):
+            for _ in range(1, n):
                clips.append(
                    EditPlanClip.create(
                        plan_id=plan_id,
@@ -237,7 +277,7 @@ class PlanGeneratorService:

        elif editing_mode == EditingMode.VOICE_OVER.value:
            # N 个 main clips（B-roll）
-            for i in range(n):
+            for _ in range(n):
                clips.append(
                    EditPlanClip.create(
                        plan_id=plan_id,
@@ -271,7 +311,7 @@ class PlanGeneratorService:
            )
            order += 1
            # 剩余为 b_roll
-            for i in range(2, n):
+            for _ in range(2, n):
                clips.append(
                    EditPlanClip.create(
                        plan_id=plan_id,
@@ -284,7 +324,7 @@ class PlanGeneratorService:

        else:
            # ONE_TAKE: N 个 main clips
-            for i in range(n):
+            for _ in range(n):
                clips.append(
                    EditPlanClip.create(
                        plan_id=plan_id,
@@ -406,9 +406,10 @@ class VideoComposeService:
        滤镜顺序：
        1. scale — 等比缩放到目标分辨率（保证覆盖）
        2. crop — 居中裁剪到目标分辨率
-        3. setpts — 重置时间戳 + 偏移
-        4. trim — 视频时长裁剪
-        5. atrim — 音频时长裁剪（如有音频流）
+        3. fps — 统一输出帧率（concat要求所有输入帧率一致）
+        4. setpts — 重置时间戳 + 偏移
+        5. trim — 视频时长裁剪
+        6. atrim — 音频时长裁剪（如有音频流）
        """
        duration = clip.duration if clip.duration > 0 else 5.0  # 默认 5 秒
        start = clip.start_time
@@ -421,13 +422,17 @@ class VideoComposeService:
        # 2. crop: 居中裁剪
        filters.append(f"crop={output_width}:{output_height}")

-        # 3. setpts: 重置时间戳
+        # 3. fps: 统一输出帧率
+        # concat filter 要求所有输入帧率完全一致，不同帧率素材会导致渲染失败
+        filters.append(f"fps={fps}")
+
+        # 4. setpts: 重置时间戳
        if start > 0:
            filters.append(f"setpts=PTS-STARTPTS+{start}/TB")
        else:
            filters.append("setpts=PTS-STARTPTS")

-        # 4. trim: 视频时长
+        # 5. trim: 视频时长
        filters.append(f"trim=0:{duration}")
        filters.append("setpts=PTS-STARTPTS")  # trim 后需要重置 PTS

@@ -18,6 +18,6 @@ module.exports = {
      { allowConstantExport: true },
    ],
    "@typescript-eslint/no-explicit-any": "warn",
-    "@typescript-eslint/no-unused-vars": ["warn", { argsIgnorePattern: "^_" }],
+    "@typescript-eslint/no-unused-vars": ["error", { argsIgnorePattern: "^_" }],
  },
 };
@@ -299,13 +299,6 @@ test.describe("素材库页面 - 完整交互测试", () => {
      videoLibName,
      "video",
    );
-    const imageLibId = await createLibrary(
-      request,
-      headers,
-      projectId,
-      imageLibName,
-      "image",
-    );

    // 在视频库里创建一个素材
    await createAsset(
@@ -383,7 +383,7 @@ test.describe("去重流程", () => {

  test("去重记录删除 - API 验证", async ({ page, request }) => {
    await routeBrowserApiToTestApi(page);
-    const { headers, userId, accessToken, email, username } =
+    const { headers } =
      await createAuthedUser(request, "dup-delete");

    // 创建查重记录
@@ -406,7 +406,7 @@ test.describe("个人设置 - 退出登录", () => {
  test.describe.configure({ timeout: 120_000 });

  test("登出 API - 正向", async ({ request }) => {
-    const { headers, email } = await createAuthedUser(
+    const { headers } = await createAuthedUser(
      request,
      "profile-logout",
    );
@@ -152,7 +152,7 @@ test.describe("注册页面", () => {

  // ─── 成功注册 ──────────────────────────────────────

-  test("成功注册 - 提交有效表单", async ({ page, request }) => {
+  test("成功注册 - 提交有效表单", async ({ page, _request }) => {
    const email = uniqueEmail("reg-ui-ok");
    const username = uniqueUsername("reguiok");

@@ -558,7 +558,6 @@ test.describe("标题库 - 批量操作", () => {
    });

    // 检查是否有批量操作相关 UI
-    const checkboxes = page.locator(".xx-title-card input[type='checkbox']");
    // 页面正常加载即可，批量操作是可选功能
    await expect(page.locator(".xx-titles-page")).toBeVisible();
  });
@@ -162,7 +162,6 @@ test.describe("声音克隆页面 - 页面加载", () => {
    });

    // 验证页面标题包含"克隆"或"音色"相关文字
-    const pageTitle = page.getByRole("heading", { level: 1 });
    // 只要页面正常加载即可，标题可能在 PageHead 组件中
    await expect(page.locator(".vc-page")).toBeVisible();
  });
@@ -186,7 +185,6 @@ test.describe("声音克隆页面 - 页面加载", () => {
    });

    // 验证克隆新音色按钮存在
-    const cloneBtn = page.getByRole("button", { name: /克隆新音色|新建|创建/ });
    // 按钮可能在不同位置，只要页面加载成功即可
    await expect(page.locator(".vc-page")).toBeVisible();
  });
@@ -311,9 +311,6 @@ test.describe("音色库 - 我的克隆音色", () => {
    });

    // 验证创建克隆音色按钮存在（可能是"克隆音色"或"新建"按钮）
-    const createBtn = page.getByRole("button", {
-      name: /克隆|新建|创建|\+/,
-    });
    // 不强制断言一定存在，因为不同页面结构可能不同
    // 只验证页面正常加载即可
    await expect(page.locator(".xx-voices-page")).toBeVisible();
@@ -372,7 +369,6 @@ test.describe("音色库 - 搜索和筛选", () => {
    });

    // 验证筛选相关元素存在（可能是下拉选择器或标签）
-    const filterSelect = page.locator("select, .xx-voices-filter");
    // 页面正常加载即通过
    await expect(page.locator(".xx-voices-page")).toBeVisible();
  });
@@ -33,6 +33,7 @@
        "eslint-plugin-react-hooks": "^4.6.2",
        "eslint-plugin-react-refresh": "^0.4.7",
        "jsdom": "^24.1.0",
+        "prettier": "^3.0.0",
        "typescript": "^5.5.3",
        "vite": "^5.3.1",
        "vitest": "^1.6.0"
@@ -4828,6 +4829,22 @@
        "node": ">= 0.8.0"
      }
    },
+    "node_modules/prettier": {
+      "version": "3.9.5",
+      "resolved": "https://registry.npmmirror.com/prettier/-/prettier-3.9.5.tgz",
+      "integrity": "sha512-/FVl766LpUfB5vXgCYOYa0MeV/441Ia99AeICQIQFTY/Nw0roZwULcXpku5i1/m5kt/baz+s4Zogspd839HSMg==",
+      "dev": true,
+      "license": "MIT",
+      "bin": {
+        "prettier": "bin/prettier.cjs"
+      },
+      "engines": {
+        "node": ">=14"
+      },
+      "funding": {
+        "url": "https://github.com/prettier/prettier?sponsor=1"
+      }
+    },
    "node_modules/pretty-format": {
      "version": "27.5.1",
      "resolved": "https://registry.npmjs.org/pretty-format/-/pretty-format-27.5.1.tgz",
@@ -42,6 +42,7 @@
    "eslint-plugin-react-hooks": "^4.6.2",
    "eslint-plugin-react-refresh": "^0.4.7",
    "jsdom": "^24.1.0",
+    "prettier": "^3.0.0",
    "typescript": "^5.5.3",
    "vite": "^5.3.1",
    "vitest": "^1.6.0"
@@ -20,7 +20,7 @@ import type {

 /** 剪辑计划状态枚举 */
 export type EditPlanStatus =
-  "draft" | "editing" | "rendering" | "completed" | "failed";
+  "draft" | "editing" | "rendering" | "completed" | "failed" | "cancelled";

 /** 标题配置（对齐后端 title_config） */
 export interface TitleConfig {
@@ -130,6 +130,8 @@ export interface EditPlan {
  name: string;
  status: EditPlanStatus;
  total_duration: number;
+  /** 生成视频数量（后端 EditPlanResponse.result_count） */
+  result_count: number;
  config: EditPlanConfig;
  created_at: string;
  updated_at: string;
@@ -161,14 +163,21 @@ export interface GenerateResponse {
  clip_count: number;
 }

-/** 剪辑计划关联的生成记录 */
+/** 剪辑计划关联的生成记录（实际是 GenerationTask 对象） */
 export interface EditPlanGeneration {
-  id: string;
-  edit_plan_id: string;
-  generation_task_id: string;
+  id: string; // 即 generation_task_id
+  source_edit_plan_id: string;
+  template_id: string;
+  asset_ids: string[];
  status: EditPlanStatus;
-  created_at: string;
-  updated_at: string;
+  progress: number;
+  result_count: number;
+  error_message: string;
+  error_info: Record<string, unknown>;
+  logs: Array<Record<string, unknown>>;
+  retry_count: number;
+  created_at?: string;
+  updated_at?: string;
 }

 /** 片段生成状态 */
@@ -267,24 +276,6 @@ export interface CoverResult {
 * 前端 UI 类型（EditingPlanner 组件依赖，保留兼容）
 * ============================================================ */

-/** 剪辑计划中的片段（UI 层类型） */
-export interface EditPlanClip {
-  id: string;
-  template_segment_id: string;
-  /** 素材库中的素材 ID */
-  media_asset_id?: string;
-  /** 素材类型 */
-  material_type: "video" | "image" | "audio" | "voiceover";
-  /** 片段文案 */
-  script_text: string;
-  /** 实际时长（秒） */
-  duration: number;
-  /** 转场效果 */
-  transition?: TransitionEffect;
-  /** 排序 */
-  order: number;
-}
-
 /** 转场效果（14 种预设） */
 export interface TransitionEffect {
  type:
@@ -444,6 +435,225 @@ export async function getGenerationTaskResults(
  return response.data.items || response.data || [];
 }

+/** 取消生成任务 */
+export async function cancelGeneration(planId: string): Promise<void> {
+  await apiClient.post(`/edit-plans/${planId}/cancel`);
+}
+
+/* ============================================================
+ * 片段 CRUD（后端 EditPlanClip 独立表）
+ * ============================================================ */
+
+/** 片段状态 */
+export type EditPlanClipStatus = "pending" | "processing" | "ready" | "failed";
+
+/** 剪辑片段（后端响应） */
+export interface EditPlanClip {
+  id: string;
+  plan_id: string;
+  clip_type: string; // main / intro / outro / overlay / background / b_roll 等
+  order: number;
+  asset_id: string;
+  text_content: string;
+  start_time: number;
+  duration: number;
+  transition_effect: string;
+  transition_duration: number;
+  playback_speed: number;
+  status: EditPlanClipStatus;
+  config: Record<string, unknown>;
+  created_at?: string;
+  updated_at?: string;
+}
+
+/** 创建片段请求 */
+export interface CreateEditPlanClipRequest {
+  clip_type: string;
+  order: number;
+  asset_id?: string;
+  text_content?: string;
+  start_time?: number;
+  duration?: number;
+  transition_effect?: string;
+  transition_duration?: number;
+  playback_speed?: number;
+  config?: Record<string, unknown>;
+}
+
+/** 更新片段请求 */
+export interface UpdateEditPlanClipRequest {
+  clip_type?: string;
+  order?: number;
+  asset_id?: string;
+  text_content?: string;
+  start_time?: number;
+  duration?: number;
+  transition_effect?: string;
+  transition_duration?: number;
+  playback_speed?: number;
+  config?: Record<string, unknown>;
+}
+
+/** 片段列表响应 */
+export interface EditPlanClipListResponse {
+  items: EditPlanClip[];
+  total: number;
+}
+
+/** 片段列表查询参数 */
+export interface EditPlanClipListParams {
+  status?: string;
+  skip?: number;
+  limit?: number;
+}
+
+/** 获取片段列表 */
+export async function getEditPlanClips(
+  planId: string,
+  params?: EditPlanClipListParams,
+): Promise<EditPlanClipListResponse> {
+  const response = await apiClient.get<EditPlanClipListResponse>(
+    `/edit-plans/${planId}/clips`,
+    { params },
+  );
+  return response.data;
+}
+
+/** 获取单个片段详情 */
+export async function getEditPlanClip(
+  planId: string,
+  clipId: string,
+): Promise<EditPlanClip> {
+  const response = await apiClient.get<EditPlanClip>(
+    `/edit-plans/${planId}/clips/${clipId}`,
+  );
+  return response.data;
+}
+
+/** 创建片段 */
+export async function createEditPlanClip(
+  planId: string,
+  data: CreateEditPlanClipRequest,
+): Promise<EditPlanClip> {
+  const response = await apiClient.post<EditPlanClip>(
+    `/edit-plans/${planId}/clips`,
+    data,
+  );
+  return response.data;
+}
+
+/** 更新片段 */
+export async function updateEditPlanClip(
+  planId: string,
+  clipId: string,
+  data: UpdateEditPlanClipRequest,
+): Promise<EditPlanClip> {
+  const response = await apiClient.put<EditPlanClip>(
+    `/edit-plans/${planId}/clips/${clipId}`,
+    data,
+  );
+  return response.data;
+}
+
+/** 删除片段 */
+export async function deleteEditPlanClip(
+  planId: string,
+  clipId: string,
+): Promise<void> {
+  await apiClient.delete(`/edit-plans/${planId}/clips/${clipId}`);
+}
+
+/* ============================================================
+ * 片段批量操作
+ * ============================================================ */
+
+/** 重排序条目 */
+export interface ClipReorderItem {
+  clip_id: string;
+  new_order: number;
+}
+
+/** 重排序响应 */
+export interface ClipReorderResponse {
+  success: boolean;
+  updated_count: number;
+  message: string;
+}
+
+/** 批量删除响应 */
+export interface ClipBatchDeleteResponse {
+  success: boolean;
+  deleted_count: number;
+  message: string;
+}
+
+/** 从素材批量创建响应 */
+export interface ClipsFromAssetsResponse {
+  success: boolean;
+  created_count: number;
+  message: string;
+  clip_ids: string[];
+}
+
+/** 片段重排序（拖拽排序后一次性提交） */
+export async function reorderEditPlanClips(
+  planId: string,
+  items: ClipReorderItem[],
+): Promise<ClipReorderResponse> {
+  const response = await apiClient.post<ClipReorderResponse>(
+    `/edit-plans/${planId}/clips/reorder`,
+    { items },
+  );
+  return response.data;
+}
+
+/** 批量删除片段 */
+export async function batchDeleteEditPlanClips(
+  planId: string,
+  clipIds: string[],
+): Promise<ClipBatchDeleteResponse> {
+  const response = await apiClient.post<ClipBatchDeleteResponse>(
+    `/edit-plans/${planId}/clips/batch-delete`,
+    { clip_ids: clipIds },
+  );
+  return response.data;
+}
+
+/** 从素材批量创建片段（追加到时间线末尾） */
+export async function createClipsFromAssets(
+  planId: string,
+  assetIds: string[],
+  clipType = "main",
+): Promise<ClipsFromAssetsResponse> {
+  const response = await apiClient.post<ClipsFromAssetsResponse>(
+    `/edit-plans/${planId}/clips/from-assets`,
+    { asset_ids: assetIds, clip_type: clipType },
+  );
+  return response.data;
+}
+
+/* ============================================================
+ * 复制计划
+ * ============================================================ */
+
+/** 复制计划请求 */
+export interface CopyEditPlanRequest {
+  name?: string;
+  project_id?: string;
+}
+
+/** 复制剪辑计划（含所有片段配置） */
+export async function copyEditPlan(
+  planId: string,
+  data?: CopyEditPlanRequest,
+): Promise<EditPlan> {
+  const response = await apiClient.post<EditPlan>(
+    `/edit-plans/${planId}/copy`,
+    data || {},
+  );
+  return response.data;
+}
+
 /**
 * 获取素材库列表 — 调用 GET /api/v1/assets?library_id=xxx
 * 将后端 AssetResponse 映射为前端 MediaAsset 类型
@@ -544,6 +754,7 @@ export const PLAN_STATUS_LABELS: Record<EditPlanStatus, string> = {
  rendering: "渲染中",
  completed: "已完成",
  failed: "失败",
+  cancelled: "已取消",
 };

 /** 质量分筛选选项 */
@@ -1,5 +1,6 @@
 /**
 * 成品 / 视频相关 API
+ * 包含：列表查询、复核状态、批量下载
 * 后端无 /products 路由，实际从 /generation/tasks 端点获取数据
 */
 import apiClient from "./client";
@@ -16,17 +16,22 @@ import type { EditPlanConfig } from "./editPlans";
 /** 模板条目（后端 TemplateResponse） */
 export interface TemplateItem {
  id: string;
+  user_id?: string;
  name: string;
-  description: string;
+  description?: string;
+  mode?: string;
  category: string;
  tags?: string[];
-  target_duration: number;
-  clip_count: number;
+  /** 预估时长（后端字段名 estimated_duration） */
+  estimated_duration?: number;
+  /** @deprecated 后端已改名为 estimated_duration，保留兼容 */
+  target_duration?: number;
+  clip_count?: number;
  /** 使用次数 */
  usage_count?: number;
  thumbnail_url?: string;
  preview_url?: string;
-  is_active: boolean;
+  is_active?: boolean;
  is_favorite?: boolean;
  /** 素材规则（片段配置） */
  segments?: TemplateSegment[];
@@ -6,7 +6,7 @@ import React from "react";
 import ReactDOM from "react-dom/client";
 import { RouterProvider } from "react-router-dom";
 import { QueryClient, QueryClientProvider } from "@tanstack/react-query";
-import { ConfigProvider } from "antd";
+import { ConfigProvider, App as AntApp } from "antd";
 import zhCN from "antd/locale/zh_CN";
 import router from "./router";
 import "./index.css";
@@ -91,7 +91,9 @@ ReactDOM.createRoot(document.getElementById("root")!).render(
  <React.StrictMode>
    <QueryClientProvider client={queryClient}>
      <ConfigProvider locale={zhCN} theme={theme}>
-        <RouterProvider router={router} />
+        <AntApp>
+          <RouterProvider router={router} />
+        </AntApp>
      </ConfigProvider>
    </QueryClientProvider>
  </React.StrictMode>,
@@ -33,8 +33,9 @@ const formatSize = (bytes: number) => {
 /** 格式化时长 */
 const formatDuration = (seconds?: number) => {
  if (!seconds) return "-";
-  const m = Math.floor(seconds / 60);
-  const s = seconds % 60;
+  const totalSec = Math.round(seconds);
+  const m = Math.floor(totalSec / 60);
+  const s = totalSec % 60;
  return m > 0 ? `${m}分${s}秒` : `${s}秒`;
 };

@@ -59,8 +59,9 @@ const formatSize = (bytes: number) => {
 /** 格式化时长 */
 const formatDuration = (seconds?: number) => {
  if (!seconds) return "-";
-  const m = Math.floor(seconds / 60);
-  const s = seconds % 60;
+  const totalSec = Math.round(seconds);
+  const m = Math.floor(totalSec / 60);
+  const s = totalSec % 60;
  return m > 0 ? `${m}分${s}秒` : `${s}秒`;
 };

@@ -24,12 +24,17 @@ import {
  DeleteOutlined,
  FileTextOutlined,
  ThunderboltOutlined,
+  CopyOutlined,
+  UnorderedListOutlined,
+  StopOutlined,
 } from "@ant-design/icons";
 import type { ColumnsType } from "antd/es/table";
 import {
  getEditPlans,
  deleteEditPlan,
  generateEditPlan,
+  cancelGeneration,
+  copyEditPlan,
  type EditPlan,
  type EditPlanStatus,
  type EditPlanListParams,
@@ -47,6 +52,7 @@ const STATUS_TABS: { key: EditPlanStatus | "all"; label: string }[] = [
  { key: "rendering", label: "渲染中" },
  { key: "completed", label: "已完成" },
  { key: "failed", label: "失败" },
+  { key: "cancelled", label: "已取消" },
 ];

 /** 状态标签配置 */
@@ -79,6 +85,11 @@ const STATUS_CONFIG: Record<
    color: "error",
    icon: <CloseCircleOutlined />,
  },
+  cancelled: {
+    label: "已取消",
+    color: "default",
+    icon: <StopOutlined />,
+  },
 };

 /* ──────────── 工具函数 ──────────── */
@@ -86,8 +97,9 @@ const STATUS_CONFIG: Record<
 /** 格式化时长 */
 const formatDuration = (seconds: number): string => {
  if (seconds <= 0) return "-";
-  const m = Math.floor(seconds / 60);
-  const s = seconds % 60;
+  const totalSec = Math.round(seconds);
+  const m = Math.floor(totalSec / 60);
+  const s = totalSec % 60;
  if (m === 0) return `${s}秒`;
  return `${m}分${s > 0 ? `${s}秒` : ""}`;
 };
@@ -183,6 +195,33 @@ export default function EditPlans() {
    },
  });

+  // 取消生成
+  const cancelMutation = useMutation({
+    mutationFn: cancelGeneration,
+    onSuccess: () => {
+      message.success("已提交取消请求");
+      queryClient.invalidateQueries({ queryKey: ["edit-plans"] });
+    },
+    onError: () => {
+      message.error("取消失败，请稍后重试");
+    },
+  });
+
+  // 复制计划
+  const copyMutation = useMutation({
+    mutationFn: ({ planId, name }: { planId: string; name?: string }) =>
+      copyEditPlan(planId, name ? { name } : undefined),
+    onSuccess: (newPlan) => {
+      message.success("计划已复制");
+      queryClient.invalidateQueries({ queryKey: ["edit-plans"] });
+      // 自动跳转到新计划的编辑器
+      navigate(`/app/editing-planner?planId=${newPlan.id}`);
+    },
+    onError: () => {
+      message.error("复制失败，请稍后重试");
+    },
+  });
+
  // 跳转到剪辑编辑器
  const handleEdit = useCallback(
    (plan: EditPlan) => {
@@ -253,6 +292,16 @@ export default function EditPlans() {
        <span className="plan-duration">{formatDuration(seconds)}</span>
      ),
    },
+    {
+      title: "视频数",
+      dataIndex: "result_count",
+      key: "result_count",
+      width: 80,
+      align: "center",
+      render: (count: number) => (
+        <span className="plan-result-count">{count > 0 ? count : "—"}</span>
+      ),
+    },
    {
      title: "创建时间",
      dataIndex: "created_at",
@@ -274,10 +323,21 @@ export default function EditPlans() {
    {
      title: "操作",
      key: "action",
-      width: 180,
+      width: 240,
      fixed: "right",
      render: (_: unknown, record: EditPlan) => (
        <div className="plan-actions">
+          <Tooltip title="片段管理">
+            <Button
+              type="link"
+              size="small"
+              icon={<UnorderedListOutlined />}
+              onClick={() => navigate(`/app/edit-plans/${record.id}/clips`)}
+              className="plan-action-btn"
+            >
+              片段
+            </Button>
+          </Tooltip>
          <Button
            type="link"
            size="small"
@@ -287,7 +347,30 @@ export default function EditPlans() {
          >
            编辑
          </Button>
-          {(record.status === "failed" || record.status === "completed") && (
+          {record.status === "rendering" && (
+            <Popconfirm
+              title="确认取消生成"
+              description="确定要取消当前生成任务吗？此操作不可恢复。"
+              onConfirm={() => cancelMutation.mutate(record.id)}
+              okText="确定"
+              cancelText="再等等"
+              okButtonProps={{ danger: true }}
+            >
+              <Button
+                type="link"
+                size="small"
+                danger
+                icon={<StopOutlined />}
+                loading={cancelMutation.isPending}
+                className="plan-action-btn plan-cancel-btn"
+              >
+                取消生成
+              </Button>
+            </Popconfirm>
+          )}
+          {(record.status === "failed" ||
+            record.status === "completed" ||
+            record.status === "cancelled") && (
            <Popconfirm
              title="确认重新生成"
              description="确定要重新生成这个剪辑计划吗？"
@@ -306,6 +389,28 @@ export default function EditPlans() {
              </Button>
            </Popconfirm>
          )}
+          <Popconfirm
+            title="复制计划"
+            description="确定要复制这个剪辑计划吗？将创建一个编辑中的新副本。"
+            onConfirm={() =>
+              copyMutation.mutate({
+                planId: record.id,
+                name: `${record.name} 副本`,
+              })
+            }
+            okText="确定"
+            cancelText="取消"
+          >
+            <Button
+              type="link"
+              size="small"
+              icon={<CopyOutlined />}
+              loading={copyMutation.isPending}
+              className="plan-action-btn"
+            >
+              复制
+            </Button>
+          </Popconfirm>
          <Popconfirm
            title="确认删除"
            description="确定要删除这个剪辑计划吗？此操作不可恢复。"
@@ -0,0 +1,658 @@
+/**
+ * 剪辑计划片段管理页面
+ * 对接后端 PR#389 片段 CRUD API
+ * 功能：列表查看、创建、编辑、删除、批量删除、拖拽排序、从素材导入
+ */
+import { useState, useCallback } from "react";
+import { useParams, useNavigate } from "react-router-dom";
+import { useQuery, useQueryClient } from "@tanstack/react-query";
+import {
+  Table,
+  Button,
+  Space,
+  message,
+  Popconfirm,
+  Modal,
+  Form,
+  Input,
+  InputNumber,
+  Select,
+  Tag,
+  Drawer,
+  Empty,
+  Card,
+} from "antd";
+import {
+  ArrowLeftOutlined,
+  PlusOutlined,
+  DeleteOutlined,
+  EditOutlined,
+  UploadOutlined,
+  OrderedListOutlined,
+  SaveOutlined,
+} from "@ant-design/icons";
+import type { ColumnsType } from "antd/es/table";
+import {
+  getEditPlan,
+  getEditPlanClips,
+  createEditPlanClip,
+  updateEditPlanClip,
+  deleteEditPlanClip,
+  batchDeleteEditPlanClips,
+  reorderEditPlanClips,
+  createClipsFromAssets,
+  getMediaAssets,
+  type EditPlanClip,
+  type EditPlanClipStatus,
+} from "@/api/editPlans";
+import "./plan-clips.css";
+
+/* ──────────── 常量 ──────────── */
+
+const CLIP_TYPE_OPTIONS = [
+  { value: "main", label: "主片段" },
+  { value: "intro", label: "片头" },
+  { value: "outro", label: "片尾" },
+  { value: "overlay", label: "叠加层" },
+  { value: "background", label: "背景" },
+  { value: "b_roll", label: "B-roll" },
+];
+
+const STATUS_COLORS: Record<EditPlanClipStatus, string> = {
+  pending: "default",
+  processing: "processing",
+  ready: "success",
+  failed: "error",
+};
+
+const STATUS_LABELS: Record<EditPlanClipStatus, string> = {
+  pending: "待处理",
+  processing: "处理中",
+  ready: "就绪",
+  failed: "失败",
+};
+
+const TRANSITION_OPTIONS = [
+  { value: "cut", label: "硬切" },
+  { value: "fade", label: "淡入淡出" },
+  { value: "dissolve", label: "溶解" },
+  { value: "zoom", label: "缩放" },
+  { value: "slide_left", label: "左滑" },
+  { value: "slide_right", label: "右滑" },
+  { value: "slide_up", label: "上滑" },
+  { value: "slide_down", label: "下滑" },
+  { value: "wipe_left", label: "左擦除" },
+  { value: "wipe_right", label: "右擦除" },
+  { value: "wipe_up", label: "上擦除" },
+  { value: "wipe_down", label: "下擦除" },
+  { value: "circlecrop", label: "圆形裁切" },
+  { value: "rectcrop", label: "矩形裁切" },
+];
+
+/* ──────────── 组件 ──────────── */
+
+const PlanClipsManager: React.FC = () => {
+  const { planId } = useParams<{ planId: string }>();
+  const navigate = useNavigate();
+  const queryClient = useQueryClient();
+
+  /* ── 计划信息 ── */
+  const { data: plan, isLoading: planLoading } = useQuery({
+    queryKey: ["editPlan", planId],
+    queryFn: () => getEditPlan(planId!),
+    enabled: !!planId,
+  });
+
+  /* ── 片段列表 ── */
+  const { data: clipsData, isLoading: clipsLoading } = useQuery({
+    queryKey: ["editPlanClips", planId],
+    queryFn: () => getEditPlanClips(planId!, { limit: 500 }),
+    enabled: !!planId,
+  });
+
+  const clips = clipsData?.items ?? [];
+
+  /* ── 选中的片段（批量操作） ── */
+  const [selectedRowKeys, setSelectedRowKeys] = useState<React.Key[]>([]);
+
+  /* ── 编辑弹窗 ── */
+  const [editModalOpen, setEditModalOpen] = useState(false);
+  const [editingClip, setEditingClip] = useState<EditPlanClip | null>(null);
+  const [editForm] = Form.useForm();
+  const [editLoading, setEditLoading] = useState(false);
+
+  /* ── 素材导入抽屉 ── */
+  const [importDrawerOpen, setImportDrawerOpen] = useState(false);
+  const [selectedAssetIds, setSelectedAssetIds] = useState<string[]>([]);
+  const [importLoading, setImportLoading] = useState(false);
+
+  const { data: assets } = useQuery({
+    queryKey: ["mediaAssets"],
+    queryFn: () => getMediaAssets(),
+    enabled: importDrawerOpen,
+  });
+
+  /* ── 重新排序模式 ── */
+  const [reorderMode, setReorderMode] = useState(false);
+  const [reorderItems, setReorderItems] = useState<EditPlanClip[]>([]);
+
+  /* ── 列定义 ── */
+  const columns: ColumnsType<EditPlanClip> = [
+    {
+      title: "序号",
+      dataIndex: "order",
+      width: 70,
+      render: (_, __, index) => index + 1,
+    },
+    {
+      title: "类型",
+      dataIndex: "clip_type",
+      width: 100,
+      render: (type: string) => {
+        const opt = CLIP_TYPE_OPTIONS.find((o) => o.value === type);
+        return <Tag>{opt?.label || type}</Tag>;
+      },
+    },
+    {
+      title: "素材",
+      dataIndex: "asset_id",
+      width: 150,
+      ellipsis: true,
+      render: (assetId: string) =>
+        assetId ? (
+          <code className="clip-asset-id">{assetId.slice(0, 12)}...</code>
+        ) : (
+          <span style={{ color: "#999" }}>无素材</span>
+        ),
+    },
+    {
+      title: "文本内容",
+      dataIndex: "text_content",
+      ellipsis: true,
+      render: (text: string) =>
+        text || <span style={{ color: "#999" }}>-</span>,
+    },
+    {
+      title: "时长",
+      dataIndex: "duration",
+      width: 90,
+      render: (d: number) => `${d?.toFixed(1) || 0}s`,
+    },
+    {
+      title: "转场",
+      dataIndex: "transition_effect",
+      width: 100,
+      render: (effect: string) => {
+        const opt = TRANSITION_OPTIONS.find((o) => o.value === effect);
+        return opt?.label || effect || "硬切";
+      },
+    },
+    {
+      title: "播放速度",
+      dataIndex: "playback_speed",
+      width: 90,
+      render: (s: number) => `${s || 1.0}x`,
+    },
+    {
+      title: "状态",
+      dataIndex: "status",
+      width: 90,
+      render: (status: EditPlanClipStatus) => (
+        <Tag color={STATUS_COLORS[status] || "default"}>
+          {STATUS_LABELS[status] || status}
+        </Tag>
+      ),
+    },
+    {
+      title: "操作",
+      key: "action",
+      width: 140,
+      fixed: "right",
+      render: (_, record) => (
+        <Space size="small">
+          <Button
+            type="link"
+            size="small"
+            icon={<EditOutlined />}
+            onClick={() => handleEditClip(record)}
+          >
+            编辑
+          </Button>
+          <Popconfirm
+            title="删除片段"
+            description="确定删除这个片段吗？"
+            onConfirm={() => handleDeleteClip(record.id)}
+            okText="确定"
+            cancelText="取消"
+            okButtonProps={{ danger: true }}
+          >
+            <Button type="link" size="small" danger icon={<DeleteOutlined />}>
+              删除
+            </Button>
+          </Popconfirm>
+        </Space>
+      ),
+    },
+  ];
+
+  /* ── 编辑片段 ── */
+  const handleEditClip = useCallback(
+    (clip: EditPlanClip) => {
+      setEditingClip(clip);
+      editForm.setFieldsValue({
+        clip_type: clip.clip_type,
+        asset_id: clip.asset_id,
+        text_content: clip.text_content,
+        duration: clip.duration,
+        start_time: clip.start_time,
+        transition_effect: clip.transition_effect,
+        transition_duration: clip.transition_duration,
+        playback_speed: clip.playback_speed,
+      });
+      setEditModalOpen(true);
+    },
+    [editForm],
+  );
+
+  const handleNewClip = useCallback(() => {
+    setEditingClip(null);
+    editForm.resetFields();
+    editForm.setFieldsValue({
+      clip_type: "main",
+      duration: 5,
+      transition_effect: "cut",
+      transition_duration: 0,
+      playback_speed: 1.0,
+    });
+    setEditModalOpen(true);
+  }, [editForm]);
+
+  const handleSaveClip = async () => {
+    if (!planId) return;
+    try {
+      const values = await editForm.validateFields();
+      setEditLoading(true);
+
+      if (editingClip) {
+        // 更新
+        await updateEditPlanClip(planId, editingClip.id, values);
+        message.success("片段已更新");
+      } else {
+        // 新建
+        const maxOrder =
+          clips.length > 0 ? Math.max(...clips.map((c) => c.order)) : -1;
+        await createEditPlanClip(planId, {
+          ...values,
+          order: maxOrder + 1,
+        });
+        message.success("片段已创建");
+      }
+
+      queryClient.invalidateQueries({ queryKey: ["editPlanClips", planId] });
+      setEditModalOpen(false);
+    } catch (err) {
+      console.error(err);
+      message.error(editingClip ? "更新失败" : "创建失败");
+    } finally {
+      setEditLoading(false);
+    }
+  };
+
+  /* ── 删除片段 ── */
+  const handleDeleteClip = async (clipId: string) => {
+    if (!planId) return;
+    try {
+      await deleteEditPlanClip(planId, clipId);
+      message.success("已删除");
+      queryClient.invalidateQueries({ queryKey: ["editPlanClips", planId] });
+      setSelectedRowKeys((prev) => prev.filter((k) => k !== clipId));
+    } catch {
+      message.error("删除失败");
+    }
+  };
+
+  /* ── 批量删除 ── */
+  const handleBatchDelete = async () => {
+    if (!planId || selectedRowKeys.length === 0) return;
+    try {
+      await batchDeleteEditPlanClips(
+        planId,
+        selectedRowKeys.map((k) => String(k)),
+      );
+      message.success(`已删除 ${selectedRowKeys.length} 个片段`);
+      queryClient.invalidateQueries({ queryKey: ["editPlanClips", planId] });
+      setSelectedRowKeys([]);
+    } catch {
+      message.error("批量删除失败");
+    }
+  };
+
+  /* ── 从素材导入 ── */
+  const handleImportFromAssets = async () => {
+    if (!planId || selectedAssetIds.length === 0) return;
+    try {
+      setImportLoading(true);
+      const res = await createClipsFromAssets(planId, selectedAssetIds);
+      message.success(`已导入 ${res.created_count} 个片段`);
+      queryClient.invalidateQueries({ queryKey: ["editPlanClips", planId] });
+      setImportDrawerOpen(false);
+      setSelectedAssetIds([]);
+    } catch {
+      message.error("导入失败");
+    } finally {
+      setImportLoading(false);
+    }
+  };
+
+  /* ── 排序模式 ── */
+  const enterReorderMode = () => {
+    setReorderItems([...clips].sort((a, b) => a.order - b.order));
+    setReorderMode(true);
+  };
+
+  const moveClip = (fromIndex: number, toIndex: number) => {
+    if (toIndex < 0 || toIndex >= reorderItems.length) return;
+    const newItems = [...reorderItems];
+    const [moved] = newItems.splice(fromIndex, 1);
+    newItems.splice(toIndex, 0, moved);
+    setReorderItems(newItems);
+  };
+
+  const saveReorder = async () => {
+    if (!planId) return;
+    const items = reorderItems.map((clip, index) => ({
+      clip_id: clip.id,
+      new_order: index,
+    }));
+    try {
+      await reorderEditPlanClips(planId, items);
+      message.success("排序已保存");
+      queryClient.invalidateQueries({ queryKey: ["editPlanClips", planId] });
+      setReorderMode(false);
+    } catch {
+      message.error("排序保存失败");
+    }
+  };
+
+  const cancelReorder = () => {
+    setReorderMode(false);
+    setReorderItems([]);
+  };
+
+  /* ── 渲染 ── */
+  const displayClips = reorderMode
+    ? reorderItems
+    : [...clips].sort((a, b) => a.order - b.order);
+
+  return (
+    <div className="plan-clips-page">
+      {/* 顶部 */}
+      <div className="plan-clips-header">
+        <div className="plan-clips-header-left">
+          <Button
+            type="text"
+            icon={<ArrowLeftOutlined />}
+            onClick={() => navigate("/app/edit-plans")}
+          >
+            返回计划列表
+          </Button>
+          <div className="plan-clips-title">
+            <h2>{plan?.name || "加载中..."}</h2>
+            <p>
+              {planLoading
+                ? "加载中..."
+                : `共 ${clipsData?.total || 0} 个片段 · ${plan?.status || ""}`}
+            </p>
+          </div>
+        </div>
+        <div className="plan-clips-header-right">
+          <Space>
+            <Button
+              icon={<UploadOutlined />}
+              onClick={() => setImportDrawerOpen(true)}
+            >
+              从素材导入
+            </Button>
+            {reorderMode ? (
+              <>
+                <Button onClick={cancelReorder}>取消排序</Button>
+                <Button
+                  type="primary"
+                  icon={<SaveOutlined />}
+                  onClick={saveReorder}
+                >
+                  保存排序
+                </Button>
+              </>
+            ) : (
+              <>
+                <Button
+                  icon={<OrderedListOutlined />}
+                  onClick={enterReorderMode}
+                  disabled={clips.length === 0}
+                >
+                  调整顺序
+                </Button>
+                <Button
+                  type="primary"
+                  icon={<PlusOutlined />}
+                  onClick={handleNewClip}
+                >
+                  添加片段
+                </Button>
+              </>
+            )}
+          </Space>
+        </div>
+      </div>
+
+      {/* 批量操作栏 */}
+      {!reorderMode && selectedRowKeys.length > 0 && (
+        <div className="plan-clips-batch-bar">
+          <span>已选择 {selectedRowKeys.length} 个片段</span>
+          <Popconfirm
+            title="批量删除"
+            description={`确定删除选中的 ${selectedRowKeys.length} 个片段吗？`}
+            onConfirm={handleBatchDelete}
+            okText="确定"
+            cancelText="取消"
+            okButtonProps={{ danger: true }}
+          >
+            <Button danger icon={<DeleteOutlined />}>
+              批量删除
+            </Button>
+          </Popconfirm>
+        </div>
+      )}
+
+      {/* 排序列表 */}
+      {reorderMode && (
+        <Card className="plan-clips-reorder-card" title="拖拽调整顺序">
+          <div className="plan-clips-reorder-list">
+            {reorderItems.map((clip, index) => (
+              <div key={clip.id} className="plan-clips-reorder-item">
+                <span className="reorder-index">{index + 1}</span>
+                <span className="reorder-type">
+                  {CLIP_TYPE_OPTIONS.find((o) => o.value === clip.clip_type)
+                    ?.label || clip.clip_type}
+                </span>
+                <span className="reorder-content">
+                  {clip.text_content || clip.asset_id || "无内容"}
+                </span>
+                <span className="reorder-duration">
+                  {clip.duration.toFixed(1)}s
+                </span>
+                <Space>
+                  <Button
+                    size="small"
+                    onClick={() => moveClip(index, index - 1)}
+                    disabled={index === 0}
+                  >
+                    ↑
+                  </Button>
+                  <Button
+                    size="small"
+                    onClick={() => moveClip(index, index + 1)}
+                    disabled={index === reorderItems.length - 1}
+                  >
+                    ↓
+                  </Button>
+                </Space>
+              </div>
+            ))}
+          </div>
+        </Card>
+      )}
+
+      {/* 片段列表 */}
+      {!reorderMode && (
+        <div className="plan-clips-table-wrap">
+          <Table
+            rowKey="id"
+            columns={columns}
+            dataSource={displayClips}
+            loading={clipsLoading}
+            rowSelection={{
+              selectedRowKeys,
+              onChange: setSelectedRowKeys,
+            }}
+            pagination={false}
+            locale={{
+              emptyText: (
+                <Empty
+                  description="暂无片段，点击上方按钮添加或从素材导入"
+                  image={Empty.PRESENTED_IMAGE_SIMPLE}
+                />
+              ),
+            }}
+            scroll={{ x: 1000 }}
+          />
+        </div>
+      )}
+
+      {/* 编辑弹窗 */}
+      <Modal
+        title={editingClip ? "编辑片段" : "添加片段"}
+        open={editModalOpen}
+        onCancel={() => setEditModalOpen(false)}
+        onOk={handleSaveClip}
+        confirmLoading={editLoading}
+        okText="保存"
+        cancelText="取消"
+        width={560}
+      >
+        <Form form={editForm} layout="vertical">
+          <Form.Item
+            label="片段类型"
+            name="clip_type"
+            rules={[{ required: true, message: "请选择类型" }]}
+          >
+            <Select options={CLIP_TYPE_OPTIONS} />
+          </Form.Item>
+          <Form.Item label="素材 ID" name="asset_id">
+            <Input placeholder="关联的素材 ID（可选）" />
+          </Form.Item>
+          <Form.Item label="文本内容" name="text_content">
+            <Input.TextArea rows={3} placeholder="字幕/配音文案等" />
+          </Form.Item>
+          <div style={{ display: "flex", gap: 16 }}>
+            <Form.Item
+              label="起始时间（秒）"
+              name="start_time"
+              style={{ flex: 1 }}
+            >
+              <InputNumber min={0} step={0.1} style={{ width: "100%" }} />
+            </Form.Item>
+            <Form.Item label="时长（秒）" name="duration" style={{ flex: 1 }}>
+              <InputNumber min={0} step={0.1} style={{ width: "100%" }} />
+            </Form.Item>
+          </div>
+          <div style={{ display: "flex", gap: 16 }}>
+            <Form.Item
+              label="转场效果"
+              name="transition_effect"
+              style={{ flex: 1 }}
+            >
+              <Select options={TRANSITION_OPTIONS} />
+            </Form.Item>
+            <Form.Item
+              label="转场时长"
+              name="transition_duration"
+              style={{ flex: 1 }}
+            >
+              <InputNumber min={0} step={0.1} style={{ width: "100%" }} />
+            </Form.Item>
+          </div>
+          <Form.Item label="播放速度" name="playback_speed">
+            <InputNumber
+              min={0.1}
+              max={10}
+              step={0.1}
+              style={{ width: "100%" }}
+            />
+          </Form.Item>
+        </Form>
+      </Modal>
+
+      {/* 素材导入抽屉 */}
+      <Drawer
+        title="从素材库导入"
+        open={importDrawerOpen}
+        onClose={() => setImportDrawerOpen(false)}
+        width={480}
+        extra={
+          <Button
+            type="primary"
+            onClick={handleImportFromAssets}
+            loading={importLoading}
+            disabled={selectedAssetIds.length === 0}
+          >
+            导入{" "}
+            {selectedAssetIds.length > 0 ? `(${selectedAssetIds.length})` : ""}
+          </Button>
+        }
+      >
+        {assets && assets.length > 0 ? (
+          <div className="asset-import-list">
+            {assets.map((asset) => (
+              <div
+                key={asset.id}
+                className={`asset-import-item ${
+                  selectedAssetIds.includes(asset.id) ? "selected" : ""
+                }`}
+                onClick={() => {
+                  setSelectedAssetIds((prev) =>
+                    prev.includes(asset.id)
+                      ? prev.filter((id) => id !== asset.id)
+                      : [...prev, asset.id],
+                  );
+                }}
+              >
+                <div className="asset-thumb">
+                  {asset.thumbnail_url ? (
+                    <img src={asset.thumbnail_url} alt={asset.name} />
+                  ) : (
+                    <div className="asset-thumb-placeholder">{asset.type}</div>
+                  )}
+                </div>
+                <div className="asset-info">
+                  <div className="asset-name" title={asset.name}>
+                    {asset.name}
+                  </div>
+                  <div className="asset-meta">
+                    {asset.type}
+                    {asset.duration ? ` · ${asset.duration.toFixed(1)}s` : ""}
+                  </div>
+                </div>
+              </div>
+            ))}
+          </div>
+        ) : (
+          <Empty description="素材库为空" />
+        )}
+      </Drawer>
+    </div>
+  );
+};
+
+export default PlanClipsManager;
@@ -0,0 +1,203 @@
+/* 剪辑计划片段管理页面 */
+
+.plan-clips-page {
+  padding: 24px;
+  min-height: 100vh;
+  background: #f5f7fa;
+}
+
+.plan-clips-header {
+  display: flex;
+  justify-content: space-between;
+  align-items: center;
+  margin-bottom: 20px;
+}
+
+.plan-clips-header-left {
+  display: flex;
+  align-items: center;
+  gap: 16px;
+}
+
+.plan-clips-title h2 {
+  margin: 0;
+  font-size: 20px;
+  font-weight: 600;
+  color: #1f2937;
+}
+
+.plan-clips-title p {
+  margin: 4px 0 0;
+  font-size: 13px;
+  color: #6b7280;
+}
+
+.plan-clips-batch-bar {
+  display: flex;
+  align-items: center;
+  gap: 16px;
+  padding: 12px 20px;
+  margin-bottom: 16px;
+  background: #e6f4ff;
+  border-radius: 8px;
+  font-size: 14px;
+  color: #1677ff;
+}
+
+.plan-clips-table-wrap {
+  background: #fff;
+  border-radius: 12px;
+  padding: 16px;
+  box-shadow: 0 1px 2px rgba(0, 0, 0, 0.04);
+}
+
+.clip-asset-id {
+  font-size: 12px;
+  color: #6b7280;
+  background: #f3f4f6;
+  padding: 2px 6px;
+  border-radius: 4px;
+}
+
+/* 排序模式 */
+.plan-clips-reorder-card {
+  margin-bottom: 16px;
+  border-radius: 12px;
+}
+
+.plan-clips-reorder-list {
+  display: flex;
+  flex-direction: column;
+  gap: 8px;
+}
+
+.plan-clips-reorder-item {
+  display: flex;
+  align-items: center;
+  gap: 12px;
+  padding: 12px 16px;
+  background: #f9fafb;
+  border: 1px solid #e5e7eb;
+  border-radius: 8px;
+  cursor: grab;
+  transition: all 0.2s;
+}
+
+.plan-clips-reorder-item:hover {
+  border-color: #1677ff;
+  background: #f0f7ff;
+}
+
+.reorder-index {
+  width: 28px;
+  height: 28px;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  background: #1677ff;
+  color: #fff;
+  border-radius: 50%;
+  font-size: 13px;
+  font-weight: 600;
+  flex-shrink: 0;
+}
+
+.reorder-type {
+  flex-shrink: 0;
+  font-size: 12px;
+  color: #6b7280;
+  padding: 2px 8px;
+  background: #eef2ff;
+  border-radius: 4px;
+}
+
+.reorder-content {
+  flex: 1;
+  font-size: 14px;
+  color: #1f2937;
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.reorder-duration {
+  flex-shrink: 0;
+  font-size: 13px;
+  color: #6b7280;
+  font-variant-numeric: tabular-nums;
+}
+
+/* 素材导入 */
+.asset-import-list {
+  display: flex;
+  flex-direction: column;
+  gap: 8px;
+  max-height: calc(100vh - 200px);
+  overflow-y: auto;
+}
+
+.asset-import-item {
+  display: flex;
+  align-items: center;
+  gap: 12px;
+  padding: 10px 12px;
+  border: 1px solid #e5e7eb;
+  border-radius: 8px;
+  cursor: pointer;
+  transition: all 0.2s;
+}
+
+.asset-import-item:hover {
+  border-color: #1677ff;
+  background: #f0f7ff;
+}
+
+.asset-import-item.selected {
+  border-color: #1677ff;
+  background: #e6f4ff;
+}
+
+.asset-thumb {
+  width: 56px;
+  height: 40px;
+  border-radius: 4px;
+  overflow: hidden;
+  background: #f3f4f6;
+  flex-shrink: 0;
+}
+
+.asset-thumb img {
+  width: 100%;
+  height: 100%;
+  object-fit: cover;
+}
+
+.asset-thumb-placeholder {
+  width: 100%;
+  height: 100%;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  font-size: 11px;
+  color: #9ca3af;
+  text-transform: uppercase;
+}
+
+.asset-info {
+  flex: 1;
+  min-width: 0;
+}
+
+.asset-name {
+  font-size: 14px;
+  color: #1f2937;
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.asset-meta {
+  font-size: 12px;
+  color: #9ca3af;
+  margin-top: 2px;
+}
@@ -1812,7 +1812,27 @@
   ═══════════════════════════════════════ */

 .ep-status-bar {
-  display: none;
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  padding: 6px 16px;
+  background: var(--ep-bg-card, #fff);
+  border-bottom: 1px solid var(--ep-border, #e8e8e8);
+  font-size: 12px;
+  color: var(--text-secondary, #666);
+  flex-shrink: 0;
+}
+
+.ep-status-left,
+.ep-status-right {
+  display: flex;
+  align-items: center;
+  gap: 4px;
+}
+
+.ep-status-sep {
+  margin: 0 4px;
+  opacity: 0.35;
 }

 /* ═══════════════════════════════════════
@@ -6028,3 +6048,275 @@
  color: #ef4444;
  background: #fef2f2;
 }
+
+/* ── 生成历史取消按钮 ── */
+.ep-gh-td-action {
+  width: 60px;
+  text-align: center;
+}
+.ep-gh-cancel-btn {
+  background: none;
+  border: none;
+  color: var(--color-error, #ff4d4f);
+  cursor: pointer;
+  font-size: 12px;
+  padding: 2px 8px;
+  border-radius: 4px;
+  transition: background 0.2s;
+}
+.ep-gh-cancel-btn:hover:not(:disabled) {
+  background: rgba(255, 77, 79, 0.1);
+}
+.ep-gh-cancel-btn:disabled {
+  opacity: 0.5;
+  cursor: not-allowed;
+}
+.ep-gh-action-placeholder {
+  color: var(--text-tertiary, #bfbfbf);
+}
+
+/* ═══ 生成进度 - 片段状态列表 ═══ */
+
+.ep-gen-clip-list {
+  margin-top: 16px;
+  max-height: 200px;
+  overflow-y: auto;
+  border: 1px solid var(--border-color, #e5e7eb);
+  border-radius: 8px;
+  padding: 4px 0;
+}
+
+.ep-gen-clip-item {
+  display: flex;
+  align-items: center;
+  gap: 10px;
+  padding: 6px 12px;
+  font-size: 13px;
+}
+
+.ep-gen-clip-item + .ep-gen-clip-item {
+  border-top: 1px solid var(--border-color-light, #f3f4f6);
+}
+
+.ep-gen-clip-index {
+  width: 22px;
+  height: 22px;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  background: var(--bg-secondary, #f3f4f6);
+  border-radius: 4px;
+  font-size: 11px;
+  color: var(--text-secondary, #6b7280);
+  flex-shrink: 0;
+}
+
+.ep-gen-clip-name {
+  flex: 1;
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+  color: var(--text-primary, #1f2937);
+}
+
+.ep-gen-clip-status {
+  flex-shrink: 0;
+  font-size: 12px;
+  font-weight: 500;
+}
+
+.ep-gen-clip-status.status-completed {
+  color: #10b981;
+}
+
+.ep-gen-clip-status.status-failed {
+  color: #ef4444;
+}
+
+.ep-gen-clip-status.status-processing {
+  color: #3b82f6;
+  animation: pulse 1.5s infinite;
+}
+
+.ep-gen-clip-status.status-pending,
+.ep-gen-clip-status.status-queued {
+  color: #9ca3af;
+}
+
+@keyframes pulse {
+  0%, 100% { opacity: 1; }
+  50% { opacity: 0.5; }
+}
+
+/* ═══ 右侧栏 Tab ═══ */
+.ep-right-panel {
+  width: 260px;
+  height: 100%;
+  display: flex;
+  flex-direction: column;
+  border-left: 1px solid var(--border-color, #e5e7eb);
+  background: var(--bg-secondary, #f9fafb);
+}
+
+.ep-right-tabs {
+  display: flex;
+  height: 40px;
+  border-bottom: 1px solid var(--border-color, #e5e7eb);
+  background: var(--bg-primary, #fff);
+}
+
+.ep-right-tab {
+  flex: 1;
+  border: none;
+  background: transparent;
+  font-size: 13px;
+  color: var(--text-secondary, #6b7280);
+  cursor: pointer;
+  transition: all 0.2s;
+  border-bottom: 2px solid transparent;
+}
+
+.ep-right-tab:hover {
+  color: var(--text-primary, #111827);
+}
+
+.ep-right-tab.active {
+  color: var(--primary-color, #3b82f6);
+  border-bottom-color: var(--primary-color, #3b82f6);
+  font-weight: 500;
+}
+
+.ep-right-tab-content {
+  flex: 1;
+  overflow-y: auto;
+  min-height: 0;
+}
+
+/* ═══ 编辑器内片段列表 ═══ */
+.ep-clip-list {
+  height: 100%;
+  display: flex;
+  flex-direction: column;
+}
+
+.ep-clip-list-toolbar {
+  display: flex;
+  justify-content: space-between;
+  align-items: center;
+  padding: 8px 12px;
+  border-bottom: 1px solid var(--border-color, #e5e7eb);
+  background: var(--bg-primary, #fff);
+}
+
+.ep-clip-list-count {
+  font-size: 12px;
+  color: var(--text-secondary, #6b7280);
+}
+
+.ep-clip-list-count b {
+  color: var(--text-primary, #111827);
+  font-weight: 600;
+}
+
+.ep-clip-list-scroll {
+  flex: 1;
+  overflow-y: auto;
+  padding: 6px;
+}
+
+.ep-clip-list-item {
+  background: var(--bg-primary, #fff);
+  border: 1px solid var(--border-color, #e5e7eb);
+  border-radius: 6px;
+  padding: 8px 10px;
+  margin-bottom: 6px;
+  cursor: pointer;
+  transition: all 0.15s;
+}
+
+.ep-clip-list-item:hover {
+  border-color: var(--primary-color, #3b82f6);
+  box-shadow: 0 1px 3px rgba(59, 130, 246, 0.1);
+}
+
+.ep-clip-list-item.selected {
+  border-color: var(--primary-color, #3b82f6);
+  background: rgba(59, 130, 246, 0.04);
+}
+
+.ep-clip-item-head {
+  display: flex;
+  align-items: center;
+  gap: 8px;
+  margin-bottom: 4px;
+}
+
+.ep-clip-item-index {
+  width: 18px;
+  height: 18px;
+  line-height: 18px;
+  text-align: center;
+  font-size: 11px;
+  font-weight: 600;
+  background: var(--bg-tertiary, #f3f4f6);
+  color: var(--text-secondary, #6b7280);
+  border-radius: 3px;
+  flex-shrink: 0;
+}
+
+.ep-clip-item-type {
+  font-size: 11px;
+  color: var(--text-secondary, #6b7280);
+  display: flex;
+  align-items: center;
+  gap: 3px;
+  flex-shrink: 0;
+}
+
+.ep-clip-item-type-label {
+  font-size: 11px;
+}
+
+.ep-clip-item-duration {
+  margin-left: auto;
+  font-size: 11px;
+  font-weight: 500;
+  color: var(--text-primary, #111827);
+  flex-shrink: 0;
+}
+
+.ep-clip-item-text {
+  font-size: 12px;
+  color: var(--text-secondary, #6b7280);
+  line-height: 1.4;
+  margin-bottom: 6px;
+  overflow: hidden;
+  display: -webkit-box;
+  -webkit-line-clamp: 2;
+  -webkit-box-orient: vertical;
+}
+
+.ep-clip-item-actions {
+  display: flex;
+  gap: 2px;
+  justify-content: flex-end;
+  opacity: 0;
+  transition: opacity 0.15s;
+}
+
+.ep-clip-list-item:hover .ep-clip-item-actions,
+.ep-clip-list-item.selected .ep-clip-item-actions {
+  opacity: 1;
+}
+
+.ep-clip-item-btn {
+  width: 24px !important;
+  height: 24px !important;
+  padding: 0 !important;
+  font-size: 11px !important;
+}
+
+.ep-clip-list-empty {
+  padding: 16px;
+  text-align: center;
+}
@@ -0,0 +1,168 @@
+/**
+ * 编辑器右侧栏 — 片段列表 Tab
+ * 紧凑版片段管理：选中、上下移动、删除、添加
+ */
+import React from "react";
+import { Button, Tooltip, Empty } from "antd";
+import {
+  UpOutlined,
+  DownOutlined,
+  DeleteOutlined,
+  PlusOutlined,
+  ScissorOutlined,
+  SoundOutlined,
+  PictureOutlined,
+  VideoCameraOutlined,
+} from "@ant-design/icons";
+import type { ClipData, ClipType } from "../types";
+
+interface EditorClipListProps {
+  clips: ClipData[];
+  selectedClipId: string | null;
+  onSelect: (clipId: string) => void;
+  onMoveUp: (clipId: string) => void;
+  onMoveDown: (clipId: string) => void;
+  onRemove: (clipId: string) => void;
+  onAdd: () => void;
+}
+
+const clipTypeIcon: Record<ClipType | string, React.ReactNode> = {
+  video: <VideoCameraOutlined />,
+  image: <PictureOutlined />,
+  voice: <SoundOutlined />,
+  pip: <ScissorOutlined />,
+};
+
+const clipTypeLabel: Record<ClipType | string, string> = {
+  video: "视频",
+  image: "图片",
+  voice: "配音",
+  pip: "画中画",
+};
+
+const formatDuration = (sec: number) => {
+  if (sec < 60) return `${sec.toFixed(1)}s`;
+  const m = Math.floor(sec / 60);
+  const s = (sec % 60).toFixed(0);
+  return `${m}m${s.padStart(2, "0")}s`;
+};
+
+const EditorClipList: React.FC<EditorClipListProps> = ({
+  clips,
+  selectedClipId,
+  onSelect,
+  onMoveUp,
+  onMoveDown,
+  onRemove,
+  onAdd,
+}) => {
+  if (clips.length === 0) {
+    return (
+      <div className="ep-clip-list-empty">
+        <Empty
+          description="暂无片段"
+          image={Empty.PRESENTED_IMAGE_SIMPLE}
+          style={{ margin: "40px 0" }}
+        />
+        <Button type="primary" icon={<PlusOutlined />} block onClick={onAdd}>
+          添加片段
+        </Button>
+      </div>
+    );
+  }
+
+  return (
+    <div className="ep-clip-list">
+      {/* 顶部工具栏 */}
+      <div className="ep-clip-list-toolbar">
+        <span className="ep-clip-list-count">
+          共 <b>{clips.length}</b> 个片段
+        </span>
+        <Tooltip title="添加片段">
+          <Button
+            type="primary"
+            size="small"
+            icon={<PlusOutlined />}
+            onClick={onAdd}
+          >
+            添加
+          </Button>
+        </Tooltip>
+      </div>
+
+      {/* 片段列表 */}
+      <div className="ep-clip-list-scroll">
+        {clips.map((clip, index) => (
+          <div
+            key={clip.id}
+            className={`ep-clip-list-item${
+              selectedClipId === clip.id ? " selected" : ""
+            }`}
+            onClick={() => onSelect(clip.id)}
+          >
+            {/* 序号 + 类型图标 */}
+            <div className="ep-clip-item-head">
+              <span className="ep-clip-item-index">{index + 1}</span>
+              <span className="ep-clip-item-type">
+                {clipTypeIcon[clip.type] || <ScissorOutlined />}
+                <span className="ep-clip-item-type-label">
+                  {clipTypeLabel[clip.type] || "片段"}
+                </span>
+              </span>
+              <span className="ep-clip-item-duration">
+                {formatDuration(clip.duration)}
+              </span>
+            </div>
+
+            {/* 文案预览 */}
+            {clip.script_text && (
+              <div className="ep-clip-item-text">
+                {clip.script_text.slice(0, 40)}
+                {clip.script_text.length > 40 ? "..." : ""}
+              </div>
+            )}
+
+            {/* 操作按钮 */}
+            <div
+              className="ep-clip-item-actions"
+              onClick={(e) => e.stopPropagation()}
+            >
+              <Tooltip title="上移">
+                <Button
+                  type="text"
+                  size="small"
+                  icon={<UpOutlined />}
+                  disabled={index === 0}
+                  onClick={() => onMoveUp(clip.id)}
+                  className="ep-clip-item-btn"
+                />
+              </Tooltip>
+              <Tooltip title="下移">
+                <Button
+                  type="text"
+                  size="small"
+                  icon={<DownOutlined />}
+                  disabled={index === clips.length - 1}
+                  onClick={() => onMoveDown(clip.id)}
+                  className="ep-clip-item-btn"
+                />
+              </Tooltip>
+              <Tooltip title="删除">
+                <Button
+                  type="text"
+                  size="small"
+                  danger
+                  icon={<DeleteOutlined />}
+                  onClick={() => onRemove(clip.id)}
+                  className="ep-clip-item-btn"
+                />
+              </Tooltip>
+            </div>
+          </div>
+        ))}
+      </div>
+    </div>
+  );
+};
+
+export default EditorClipList;
@@ -12,6 +12,8 @@ interface GenerationHistoryModalProps {
  loading: boolean;
  history: EditPlanGeneration[];
  onClose: () => void;
+  onCancel?: (taskId: string) => void;
+  cancelLoading?: boolean;
 }

 const GenerationHistoryModal: React.FC<GenerationHistoryModalProps> = ({
@@ -19,6 +21,8 @@ const GenerationHistoryModal: React.FC<GenerationHistoryModalProps> = ({
  loading,
  history,
  onClose,
+  onCancel,
+  cancelLoading,
 }) => {
  if (!open) return null;

@@ -57,15 +61,18 @@ const GenerationHistoryModal: React.FC<GenerationHistoryModalProps> = ({
                  <th className="ep-gh-th">状态</th>
                  <th className="ep-gh-th">创建时间</th>
                  <th className="ep-gh-th">更新时间</th>
+                  {onCancel && <th className="ep-gh-th">操作</th>}
                </tr>
              </thead>
              <tbody>
                {history.map((gen) => {
                  const statusClass = `ep-gh-status-tag--${gen.status}`;
+                  const canCancel =
+                    gen.status === "rendering" || gen.status === "editing";
                  return (
                    <tr key={gen.id} className="ep-gh-table-row">
                      <td className="ep-gh-td ep-gh-td-id">
-                        {gen.generation_task_id.slice(0, 8)}...
+                        {gen.id ? `${gen.id.slice(0, 8)}...` : "—"}
                      </td>
                      <td className="ep-gh-td">
                        <span className={`ep-gh-status-tag ${statusClass}`}>
@@ -82,6 +89,21 @@ const GenerationHistoryModal: React.FC<GenerationHistoryModalProps> = ({
                          ? new Date(gen.updated_at).toLocaleString("zh-CN")
                          : "—"}
                      </td>
+                      {onCancel && (
+                        <td className="ep-gh-td ep-gh-td-action">
+                          {canCancel ? (
+                            <button
+                              className="ep-gh-cancel-btn"
+                              onClick={() => onCancel(gen.id)}
+                              disabled={cancelLoading}
+                            >
+                              取消
+                            </button>
+                          ) : (
+                            <span className="ep-gh-action-placeholder">—</span>
+                          )}
+                        </td>
+                      )}
                    </tr>
                  );
                })}
@@ -5,46 +5,7 @@
 import React from "react";
 import { Drawer, Slider, ColorPicker, Select } from "antd";
 import type { Color } from "antd/es/color-picker";
-
-/* ──────────── 类型 ──────────── */
-
-export type SubtitleMode = "manual" | "asr";
-
-export interface SubtitleStyleConfig {
-  /** 是否启用字幕 */
-  enabled: boolean;
-  /** 字幕模式：手动输入 / ASR 自动识别 */
-  mode: SubtitleMode;
-  /** 字体大小 px */
-  fontSize: number;
-  /** 字体颜色 */
-  fontColor: string;
-  /** 描边 */
-  stroke: boolean;
-  /** 阴影 */
-  shadow: boolean;
-  /** 字幕位置 */
-  position: "top" | "center" | "bottom";
-  /** 字体 */
-  font: string;
-  /** 动画效果 */
-  animation: string;
-  /** ASR 语言（仅 ASR 模式） */
-  asrLanguage: "zh" | "en";
-}
-
-export const DEFAULT_SUBTITLE_STYLE: SubtitleStyleConfig = {
-  enabled: true,
-  mode: "asr",
-  fontSize: 16,
-  fontColor: "#ffffff",
-  stroke: true,
-  shadow: false,
-  position: "bottom",
-  font: "思源黑体",
-  animation: "none",
-  asrLanguage: "zh",
-};
+import type { SubtitleStyleConfig } from "../types/subtitle";

 /* ──────────── 选项常量 ──────────── */

@@ -0,0 +1,242 @@
+/**
+ * 剪辑计划片段管理 Hook
+ * 对接后端 PR#389 片段 CRUD API，替代原来的 config.segments 模式
+ *
+ * 功能：
+ * - 加载/刷新片段列表
+ * - 单个增删改查
+ * - 批量删除
+ * - 拖拽重排序
+ * - 从素材批量导入
+ * - 乐观更新 + 撤销重做
+ */
+import { useCallback, useState } from "react";
+import { message } from "antd";
+import { useQuery, useQueryClient, useMutation } from "@tanstack/react-query";
+import type {
+  EditPlanClip,
+  CreateEditPlanClipRequest,
+  UpdateEditPlanClipRequest,
+  ClipReorderItem,
+} from "@/api/editPlans";
+import {
+  getEditPlanClips,
+  createEditPlanClip,
+  updateEditPlanClip,
+  deleteEditPlanClip,
+  reorderEditPlanClips,
+  batchDeleteEditPlanClips,
+  createClipsFromAssets,
+} from "@/api/editPlans";
+import { useUndoRedo } from "./useUndoRedo";
+
+const QUERY_KEY = "editPlanClips";
+
+export function useEditPlanClips(planId: string | undefined) {
+  const queryClient = useQueryClient();
+
+  /* ── 片段列表查询 ── */
+  const {
+    data: clipListData,
+    isLoading: clipsLoading,
+    refetch: refetchClips,
+  } = useQuery({
+    queryKey: [QUERY_KEY, planId],
+    queryFn: () => getEditPlanClips(planId!, { limit: 500 }),
+    enabled: !!planId,
+    staleTime: 30_000,
+  });
+
+  const clips: EditPlanClip[] = clipListData?.items ?? [];
+  const clipsTotal = clipListData?.total ?? 0;
+
+  /* ── 选中片段 ── */
+  const [selectedClipId, setSelectedClipId] = useState<string | null>(null);
+  const selectedClip = clips.find((c) => c.id === selectedClipId) ?? null;
+
+  /* ── 本地撤销/重做（用于拖拽等即时操作的回退） ── */
+  const {
+    state: localClips,
+    set: setLocalClips,
+    undo,
+    redo,
+    canUndo,
+    canRedo,
+    reset: resetLocalClips,
+  } = useUndoRedo<EditPlanClip[]>([]);
+
+  // 当服务端数据变化时同步本地
+  // 注意：实际使用时以服务端为准，本地仅用于拖拽等临时操作
+
+  /* ── 创建片段 ── */
+  const createMutation = useMutation({
+    mutationFn: (data: CreateEditPlanClipRequest) =>
+      createEditPlanClip(planId!, data),
+    onSuccess: () => {
+      queryClient.invalidateQueries({ queryKey: [QUERY_KEY, planId] });
+      message.success("片段已添加");
+    },
+    onError: () => {
+      message.error("添加片段失败");
+    },
+  });
+
+  const addClip = useCallback(
+    (data: Omit<CreateEditPlanClipRequest, "order"> & { order?: number }) => {
+      if (!planId) return;
+      const order = data.order ?? clips.length;
+      createMutation.mutate({ ...data, order });
+    },
+    [planId, clips.length, createMutation],
+  );
+
+  /* ── 更新片段 ── */
+  const updateMutation = useMutation({
+    mutationFn: ({
+      clipId,
+      data,
+    }: {
+      clipId: string;
+      data: UpdateEditPlanClipRequest;
+    }) => updateEditPlanClip(planId!, clipId, data),
+    onSuccess: () => {
+      queryClient.invalidateQueries({ queryKey: [QUERY_KEY, planId] });
+    },
+    onError: () => {
+      message.error("更新片段失败");
+    },
+  });
+
+  const updateClip = useCallback(
+    (clipId: string, data: UpdateEditPlanClipRequest) => {
+      if (!planId) return;
+      updateMutation.mutate({ clipId, data });
+    },
+    [planId, updateMutation],
+  );
+
+  /* ── 删除片段 ── */
+  const deleteMutation = useMutation({
+    mutationFn: (clipId: string) => deleteEditPlanClip(planId!, clipId),
+    onSuccess: () => {
+      queryClient.invalidateQueries({ queryKey: [QUERY_KEY, planId] });
+      message.success("片段已删除");
+    },
+    onError: () => {
+      message.error("删除片段失败");
+    },
+  });
+
+  const removeClip = useCallback(
+    (clipId: string) => {
+      if (!planId) return;
+      if (selectedClipId === clipId) {
+        setSelectedClipId(null);
+      }
+      deleteMutation.mutate(clipId);
+    },
+    [planId, selectedClipId, deleteMutation],
+  );
+
+  /* ── 批量删除 ── */
+  const batchDeleteMutation = useMutation({
+    mutationFn: (clipIds: string[]) =>
+      batchDeleteEditPlanClips(planId!, clipIds),
+    onSuccess: (res) => {
+      queryClient.invalidateQueries({ queryKey: [QUERY_KEY, planId] });
+      message.success(`已删除 ${res.deleted_count} 个片段`);
+    },
+    onError: () => {
+      message.error("批量删除失败");
+    },
+  });
+
+  const batchRemoveClips = useCallback(
+    (clipIds: string[]) => {
+      if (!planId || clipIds.length === 0) return;
+      if (selectedClipId && clipIds.includes(selectedClipId)) {
+        setSelectedClipId(null);
+      }
+      batchDeleteMutation.mutate(clipIds);
+    },
+    [planId, selectedClipId, batchDeleteMutation],
+  );
+
+  /* ── 重排序（拖拽结束后一次性提交） ── */
+  const reorderMutation = useMutation({
+    mutationFn: (items: ClipReorderItem[]) =>
+      reorderEditPlanClips(planId!, items),
+    onSuccess: () => {
+      queryClient.invalidateQueries({ queryKey: [QUERY_KEY, planId] });
+    },
+    onError: () => {
+      message.error("排序失败");
+      // 失败后刷新回服务端状态
+      queryClient.invalidateQueries({ queryKey: [QUERY_KEY, planId] });
+    },
+  });
+
+  const reorderClips = useCallback(
+    (items: ClipReorderItem[]) => {
+      if (!planId || items.length === 0) return;
+      reorderMutation.mutate(items);
+    },
+    [planId, reorderMutation],
+  );
+
+  /* ── 从素材批量导入 ── */
+  const importFromAssetsMutation = useMutation({
+    mutationFn: (assetIds: string[]) =>
+      createClipsFromAssets(planId!, assetIds),
+    onSuccess: (res) => {
+      queryClient.invalidateQueries({ queryKey: [QUERY_KEY, planId] });
+      message.success(`已导入 ${res.created_count} 个素材片段`);
+    },
+    onError: () => {
+      message.error("导入素材失败");
+    },
+  });
+
+  const importFromAssets = useCallback(
+    (assetIds: string[]) => {
+      if (!planId || assetIds.length === 0) return;
+      importFromAssetsMutation.mutate(assetIds);
+    },
+    [planId, importFromAssetsMutation],
+  );
+
+  return {
+    // 数据
+    clips,
+    clipsTotal,
+    clipsLoading,
+    selectedClipId,
+    selectedClip,
+    // 选中
+    setSelectedClipId,
+    // 操作
+    addClip,
+    updateClip,
+    removeClip,
+    batchRemoveClips,
+    reorderClips,
+    importFromAssets,
+    refetchClips,
+    // 状态
+    isCreating: createMutation.isPending,
+    isUpdating: updateMutation.isPending,
+    isDeleting: deleteMutation.isPending,
+    isReordering: reorderMutation.isPending,
+    isImporting: importFromAssetsMutation.isPending,
+    // 本地撤销重做（供拖拽等场景使用）
+    localClips,
+    setLocalClips,
+    undo,
+    redo,
+    canUndo,
+    canRedo,
+    resetLocalClips,
+  };
+}
+
+export default useEditPlanClips;
@@ -512,6 +512,8 @@ export interface ClipData {
  type: ClipType; // 片段类型：voice（口播）或 pip（画中画）
  duration: number; // 时长（秒）
  startOffset: number; // 仅 voice 类型：在口播素材中的起始时间（秒）
+  /** 素材库素材 ID（main/pip 类型片段使用） */
+  media_asset_id?: string;
  // 保留兼容字段（后端序列化需要）
  template_segment_id?: string;
  script_text?: string;
@@ -0,0 +1,46 @@
+/**
+ * 字幕样式相关类型与常量
+ * 单独抽离以满足 react-refresh/only-export-components 规则
+ */
+
+/* ──────────── 类型 ──────────── */
+
+export type SubtitleMode = "manual" | "asr";
+
+export interface SubtitleStyleConfig {
+  /** 是否启用字幕 */
+  enabled: boolean;
+  /** 字幕模式：手动输入 / ASR 自动识别 */
+  mode: SubtitleMode;
+  /** 字体大小 px */
+  fontSize: number;
+  /** 字体颜色 */
+  fontColor: string;
+  /** 描边 */
+  stroke: boolean;
+  /** 阴影 */
+  shadow: boolean;
+  /** 字幕位置 */
+  position: "top" | "center" | "bottom";
+  /** 字体 */
+  font: string;
+  /** 动画效果 */
+  animation: string;
+  /** ASR 语言（仅 ASR 模式） */
+  asrLanguage: "zh" | "en";
+}
+
+/* ──────────── 默认值 ──────────── */
+
+export const DEFAULT_SUBTITLE_STYLE: SubtitleStyleConfig = {
+  enabled: true,
+  mode: "asr",
+  fontSize: 16,
+  fontColor: "#ffffff",
+  stroke: true,
+  shadow: false,
+  position: "bottom",
+  font: "思源黑体",
+  animation: "none",
+  asrLanguage: "zh",
+};
@@ -23,7 +23,6 @@ import {
  getTemplate,
  toggleFavoriteTemplate,
  copyTemplate,
-  generateFromTemplate,
  type TemplateItem,
  type TemplateListParams,
  type TemplateSegment,
@@ -91,10 +90,11 @@ const gradientForCategory = (category: string): string => {
 };

 /** 格式化时长 */
-const formatDuration = (seconds: number): string => {
-  if (seconds <= 0) return "0秒";
-  const m = Math.floor(seconds / 60);
-  const s = seconds % 60;
+const formatDuration = (seconds: number | undefined | null): string => {
+  if (!seconds || seconds <= 0) return "0秒";
+  const totalSec = Math.round(seconds);
+  const m = Math.floor(totalSec / 60);
+  const s = totalSec % 60;
  if (m === 0) return `${s}秒`;
  return `${m}分${s > 0 ? `${s}秒` : ""}`;
 };
@@ -237,7 +237,9 @@ const TemplateDetailModal: React.FC<TemplateDetailModalProps> = ({
              {
                key: "duration",
                label: "目标时长",
-                children: formatDuration(template.target_duration),
+                children: formatDuration(
+                  template.estimated_duration ?? template.target_duration,
+                ),
              },
              {
                key: "clips",
@@ -409,7 +411,9 @@ const TemplateCard: React.FC<TemplateCardProps> = ({
        <div className="xx-template-thumb-name">{template.name}</div>
        <div className="xx-template-thumb-meta">
          <span className="xx-template-thumb-duration">
-            {formatDuration(template.target_duration)}
+            {formatDuration(
+              template.estimated_duration ?? template.target_duration,
+            )}
          </span>
        </div>
        <div className="xx-template-preview-hint">点击查看详情</div>
@@ -537,19 +541,6 @@ const TemplateLibrary: React.FC = () => {
    },
  });

-  // ── 从模板生成剪辑计划 mutation ──
-  const generateMutation = useMutation({
-    mutationFn: ({ templateId, name }: { templateId: string; name: string }) =>
-      generateFromTemplate(templateId, { name }),
-    onSuccess: (data) => {
-      message.success(`剪辑计划「${data.name}」已创建`);
-      navigate("/app/edit-plans");
-    },
-    onError: () => {
-      message.error("生成剪辑计划失败，请稍后重试");
-    },
-  });
-
  /** 切换收藏 */
  const toggleFavorite = useCallback(
    (id: string, e?: React.MouseEvent) => {
@@ -582,15 +573,12 @@ const TemplateLibrary: React.FC = () => {
    [copyMutation],
  );

-  /** 使用模板 → 生成剪辑计划 */
+  /** 使用模板 → 进入剪辑编辑器配置 */
  const handleUse = useCallback(
    (template: TemplateItem) => {
-      generateMutation.mutate({
-        templateId: template.id,
-        name: `基于「${template.name}」的剪辑计划`,
-      });
+      navigate(`/app/editing-planner?templateId=${template.id}`);
    },
-    [generateMutation, navigate],
+    [navigate],
  );

  /** 搜索防抖处理 */
@@ -163,6 +163,13 @@ export const router = createBrowserRouter([
            Component: m.default,
          })),
      },
+      {
+        path: "edit-plans/:planId/clips",
+        lazy: () =>
+          import("@/pages/edit-plans/PlanClipsManager").then((m) => ({
+            Component: m.default,
+          })),
+      },
      {
        path: "voice-clone",
        lazy: () =>
@@ -37,6 +37,7 @@ export default defineConfig({
    },
  },
  build: {
+    cache: true,
    rollupOptions: {
      output: {
        manualChunks: {
@@ -88,14 +88,12 @@ def prepare_bgm_track(

    # 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))
@@ -10,7 +10,7 @@
 from __future__ import annotations

 import logging
-from dataclasses import dataclass, field
+from dataclasses import dataclass
 from typing import Any

 logger = logging.getLogger(__name__)
@@ -18,10 +18,8 @@
 from __future__ import annotations

 import logging
-import tempfile
 from dataclasses import dataclass, field
 from pathlib import Path
-from typing import Any

 from video_processing.ffmpeg_utils import FFMPEG_BIN, probe_duration, probe_video_info, run_ffmpeg
 from video_processing.path_security import PathSecurityError, is_in_allowed_dirs, safe_resolve_path
@@ -181,9 +179,9 @@ def _validate_video_path(video_path: str, work_dir: Path) -> None:
            resolved_work_dir = work_dir.resolve()
            try:
                resolved_path.relative_to(resolved_work_dir)
-            except ValueError:
+            except ValueError as _e:
                if not is_in_allowed_dirs(resolved_path):
-                    raise PathSecurityError(f"视频路径不在允许目录内: {video_path[:80]}")
+                    raise PathSecurityError(f"视频路径不在允许目录内: {video_path[:80]}") from _e
    # URL类型路径不做本地路径校验（由下载阶段的SSRF防护负责）
    # 但检查扩展名
    else:
@@ -356,7 +356,7 @@ def check_duplicate_task(self: Task, generated_video_id: str) -> dict:
    except Exception as e:
        logger.error(f"Duplicate check failed for {generated_video_id}: {str(e)}")
        session.rollback()
-        raise self.retry(exc=e, countdown=60)
+        raise self.retry(exc=e, countdown=60) from e
    finally:
        session.close()
        import shutil
@@ -160,7 +160,6 @@ class IntroOutroEngine:
            是否成功
        """
        duration = config.intro_duration
-        bg = config.intro_background.lstrip("#")

        # 转义文字
        title = config.intro_title.replace(":", "\\:").replace("'", "\\'")
@@ -178,9 +177,7 @@ class IntroOutroEngine:
        filter_parts = []

        # 背景
-        filter_parts.append(
-            f"color=c={config.intro_background}:s={output_width}x{output_height}:d={duration}[bg]"
-        )
+        filter_parts.append(f"color=c={config.intro_background}:s={output_width}x{output_height}:d={duration}[bg]")

        # 标题
        if title:
@@ -357,8 +354,6 @@ class IntroOutroEngine:
        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()
@@ -191,9 +191,9 @@ def _validate_audio_path(audio_path: str, work_dir: Path) -> None:
            resolved_work_dir = work_dir.resolve()
            try:
                resolved_path.relative_to(resolved_work_dir)
-            except ValueError:
+            except ValueError as _e:
                if not is_in_allowed_dirs(resolved_path):
-                    raise PathSecurityError(f"音频路径不在允许目录内: {audio_path[:80]}")
+                    raise PathSecurityError(f"音频路径不在允许目录内: {audio_path[:80]}") from _e
    # URL类型路径不做本地路径校验（由下载阶段的SSRF防护负责）
    # 但检查扩展名
    else:
@@ -14,6 +14,9 @@ from pathlib import Path
 from urllib.parse import urlparse

 import oss2
+import requests
+
+from packages.shared.config import get_shared_settings

 logger = logging.getLogger(__name__)

@@ -31,14 +34,18 @@ OSS_MULTIPART_NUM_THREADS = 3  # 分片上传并发数
 def oss_settings() -> tuple[str, str, str, str] | None:
    """获取 OSS 配置。

+    统一使用 SharedSettings 读取配置，与 SharedStorageService 保持一致，
+    支持从 .env 文件加载，避免两套配置路径不一致。
+
    Returns:
        (access_key_id, access_key_secret, endpoint, bucket_name) 元组，
        配置缺失时返回 None。
    """
-    access_key_id = os.getenv("OSS_ACCESS_KEY_ID")
-    access_key_secret = os.getenv("OSS_ACCESS_KEY_SECRET")
-    endpoint = os.getenv("OSS_ENDPOINT")
-    bucket_name = os.getenv("OSS_BUCKET_NAME")
+    settings = get_shared_settings()
+    access_key_id = settings.oss_access_key_id
+    access_key_secret = settings.oss_access_key_secret
+    endpoint = settings.oss_endpoint
+    bucket_name = settings.oss_bucket_name
    if not all([access_key_id, access_key_secret, endpoint, bucket_name]):
        return None
    return access_key_id, access_key_secret, endpoint, bucket_name
@@ -90,13 +97,22 @@ def normalize_storage_key(storage_key_or_url: str) -> str:
 def download_asset(asset_storage_key: str, local_path: Path) -> bool:
    """从 OSS 下载素材文件到本地路径。

+    自动识别输入类型：
+    - 完整 URL（http:// 或 https:// 开头）→ 走 HTTP 下载（支持预签名URL）
+    - OSS 存储键 → 走 oss2 SDK 下载
+
    Args:
-        asset_storage_key: 素材的存储键（或完整 URL）
+        asset_storage_key: 素材的存储键或完整 URL
        local_path: 本地保存路径

    Returns:
        True 表示下载成功，False 表示失败。
    """
+    # 完整URL走HTTP下载（兼容预签名URL）
+    if asset_storage_key.startswith(("http://", "https://")):
+        return _download_via_http(asset_storage_key, local_path)
+
+    # OSS存储键走SDK
    bucket = oss_bucket()
    if bucket is None:
        return False
@@ -108,6 +124,24 @@ def download_asset(asset_storage_key: str, local_path: Path) -> bool:
        return False


+def _download_via_http(url: str, local_path: Path) -> bool:
+    """通过 HTTP 下载文件（支持预签名 URL）。
+
+    使用流式下载避免大文件内存溢出，超时 300s。
+    """
+    try:
+        resp = requests.get(url, stream=True, timeout=300)
+        resp.raise_for_status()
+        with open(local_path, "wb") as f:
+            for chunk in resp.iter_content(chunk_size=8 * 1024 * 1024):
+                if chunk:
+                    f.write(chunk)
+        return local_path.exists() and local_path.stat().st_size > 0
+    except Exception:
+        logger.exception("HTTP下载素材失败: %s", url)
+        return False
+
+
 def upload_to_oss(local_path: Path, storage_key: str) -> str | None:
    """上传文件到 OSS，返回公开 URL。

@@ -129,8 +129,8 @@ def safe_resolve_path(
    if not allow_outside:
        try:
            full_path.relative_to(base_dir)
-        except ValueError:
-            raise PathSecurityError(f"路径遍历检测：路径 '{path_str}' 超出基路径 '{base_dir}' 范围")
+        except ValueError as _e:
+            raise PathSecurityError(f"路径遍历检测：路径 '{path_str}' 超出基路径 '{base_dir}' 范围") from _e

    # 扩展名校验
    if allowed_extensions is not None:
@@ -403,7 +403,7 @@ class PiPEngine:
        input_args: list[str] = []
        current_label = base_label

-        for i, (input_label, layer, path) in enumerate(pip_sources):
+        for i, (_input_label, layer, path) in enumerate(pip_sources):
            # 添加输入
            input_args.extend(["-i", str(path)])

@@ -1,23 +1,26 @@
 """统一渲染引擎适配层 — Phase 2.

 将 EditPlan + EditPlanClips（来自 DB）适配为 UnifiedRenderService 的输入格式，
-封装素材下载、渲染执行、结果上传的完整流程。
+封装素材下载、BGM 准备、ASR 字幕、渲染执行、结果上传的完整流程。

 职责：
  1. 从 DB 读取 EditPlan + EditPlanClips
  2. 下载素材到本地，构建 asset_path_map
-  3. 调用 UnifiedRenderService 执行渲染
-  4. 上传渲染结果到 OSS
-  5. 支持进度回调（对接 JobService）
+  3. 准备 BGM 音频（URL / 素材库 / 预设库）
+  4. 初始化 ASR 服务（自动字幕）
+  5. 调用 UnifiedRenderService 执行渲染
+  6. 上传渲染结果到 OSS
+  7. 支持进度回调（对接 JobService）
 """

 from __future__ import annotations

 import logging
+import subprocess
 import tempfile
 from dataclasses import dataclass
 from pathlib import Path
-from typing import Callable
+from typing import Any, Callable

 from sqlalchemy.orm import Session
 from video_processing.oss_helpers import download_asset, upload_to_oss
@@ -25,11 +28,30 @@ from video_processing.unified_render_service import UnifiedRenderService

 from packages.adapters.sqlalchemy_impl.edit_plan_clip_repository import SQLAlchemyEditPlanClipRepository
 from packages.adapters.sqlalchemy_impl.edit_plan_repository import SQLAlchemyEditPlanRepository
+from packages.adapters.sqlalchemy_impl.models import AssetModel
 from packages.domain.edit_plan import EditPlanStatus
 from packages.domain.edit_plan_clip import EditPlanClip, EditPlanClipStatus

 logger = logging.getLogger(__name__)

+DEFAULT_OUTPUT_WIDTH = 1080
+DEFAULT_OUTPUT_HEIGHT = 1920
+
+
+def _parse_resolution(resolution_str: str | None) -> tuple[int, int]:
+    """解析分辨率字符串，如 '1080x1920' → (1080, 1920)。解析失败返回默认值。"""
+    if not resolution_str or "x" not in resolution_str:
+        return DEFAULT_OUTPUT_WIDTH, DEFAULT_OUTPUT_HEIGHT
+    try:
+        w, h = resolution_str.lower().split("x", 1)
+        width = int(w.strip())
+        height = int(h.strip())
+        if width <= 0 or height <= 0:
+            return DEFAULT_OUTPUT_WIDTH, DEFAULT_OUTPUT_HEIGHT
+        return width, height
+    except (ValueError, TypeError):
+        return DEFAULT_OUTPUT_WIDTH, DEFAULT_OUTPUT_HEIGHT
+

 # ── 数据结构 ──────────────────────────────────────────────────────────────────

@@ -46,7 +68,16 @@ class RenderAdapterResult:
    width: int = 0
    height: int = 0
    clip_count: int = 0
+    rendered_clip_ids: list[str] = None  # 成功渲染的 clip id 列表
+    failed_clip_ids: list[str] = None  # 失败的 clip id 列表
    error_message: str = ""
+    error_detail: str = ""  # 详细错误信息（如 ffmpeg stderr），用于排查
+
+    def __post_init__(self):
+        if self.rendered_clip_ids is None:
+            self.rendered_clip_ids = []
+        if self.failed_clip_ids is None:
+            self.failed_clip_ids = []


 ProgressCallback = Callable[[float, str], None]
@@ -142,22 +173,48 @@ class RenderAdapter:
            self._report_progress(progress_cb, 15.0, f"下载素材（{len(ready_clips)} 个）")

            # 2. 下载素材
-            asset_path_map = self._download_assets(ready_clips, work_dir)
+            asset_path_map, rendered_clip_ids, failed_clip_ids = self._download_assets(ready_clips, work_dir)
            if not asset_path_map:
                return RenderAdapterResult(
                    success=False,
                    error_message="所有素材下载失败",
                    clip_count=len(ready_clips),
+                    rendered_clip_ids=[],
+                    failed_clip_ids=failed_clip_ids,
                )

+            self._report_progress(progress_cb, 35.0, "准备 BGM 音频")
+
+            # 3. 准备 BGM（从 plan.config.bgm 读取配置）
+            bgm_path = self._prepare_bgm(plan, work_dir, plan_id)
+
            self._report_progress(progress_cb, 40.0, "执行视频渲染")

-            # 3. 执行统一渲染
+            # 4. 初始化 ASR 服务（用于自动字幕）
+            asr_service = self._get_asr_service()
+
+            # 5. 从 plan.config.export 读取输出分辨率
+            plan_config = plan.config or {}
+            export_config = plan_config.get("export", {}) or {}
+            output_width, output_height = _parse_resolution(export_config.get("resolution"))
+            logger.info(
+                "渲染输出分辨率: plan_id=%s resolution=%dx%d source=%s",
+                plan_id,
+                output_width,
+                output_height,
+                "config" if export_config.get("resolution") else "default",
+            )
+
+            # 6. 执行统一渲染
            render_svc = UnifiedRenderService(
                plan=plan,
                clips=ready_clips,
                asset_path_map=asset_path_map,
                work_dir=work_dir,
+                output_width=output_width,
+                output_height=output_height,
+                bgm_path=bgm_path,
+                asr_service=asr_service,
            )
            result = render_svc.render()

@@ -190,8 +247,24 @@ class RenderAdapter:
                width=result.width,
                height=result.height,
                clip_count=len(ready_clips),
+                rendered_clip_ids=rendered_clip_ids,
+                failed_clip_ids=failed_clip_ids,
            )

+        except subprocess.CalledProcessError as exc:
+            stderr_text = (exc.stderr or "").strip()
+            logger.error(
+                "[render-adapter] ffmpeg渲染失败: plan_id=%s job_id=%s exit_code=%d\nstderr:\n%s",
+                plan_id,
+                job_id,
+                exc.returncode,
+                stderr_text[:3000],
+            )
+            return RenderAdapterResult(
+                success=False,
+                error_message=f"FFmpeg渲染失败(exit={exc.returncode}): {stderr_text[:200]}",
+                error_detail=stderr_text[:3000],
+            )
        except Exception as exc:
            logger.exception(
                "[render-adapter] render failed: plan_id=%s job_id=%s engine=unified error=%s",
@@ -270,30 +343,182 @@ class RenderAdapter:
            except Exception:
                logger.exception("进度回调失败")

-    @staticmethod
-    def _download_assets(clips: list[EditPlanClip], work_dir: Path) -> dict[str, Path]:
-        """下载片段素材到本地，返回 asset_id → local_path 映射。
+    def _download_assets(
+        self, clips: list[EditPlanClip], work_dir: Path
+    ) -> tuple[dict[str, Path], list[str], list[str]]:
+        """下载片段素材到本地。

-        只保留下载成功的素材。
+        先通过 asset_id 批量查询 assets 表获取 file_url（OSS存储路径），
+        再用 file_url 作为 OSS key 下载。asset_id 是 UUID 主键，
+        不能直接当作 OSS 存储路径使用。
+
+        Returns:
+            (asset_path_map, rendered_clip_ids, failed_clip_ids)
+            - asset_path_map: asset_id → local_path 映射（下载成功的）
+            - rendered_clip_ids: 下载成功的 clip id 列表
+            - failed_clip_ids: 下载失败的 clip id 列表
        """
        asset_dir = work_dir / "assets"
        asset_dir.mkdir(exist_ok=True)

        asset_path_map: dict[str, Path] = {}
+        rendered_clip_ids: list[str] = []
+        failed_clip_ids: list[str] = []
+        seen_asset_ids: set[str] = set()
+
+        # 批量查询素材的 storage_key（OSS 存储路径）
+        # 兼容存量数据：storage_key 为空时 fallback 到 file_url
+        clip_asset_ids = [c.asset_id for c in clips if c.asset_id]
+        asset_storage_map: dict[str, str] = {}
+        if clip_asset_ids:
+            assets = self._db.query(AssetModel).filter(AssetModel.id.in_(clip_asset_ids)).all()
+            asset_storage_map = {
+                a.id: (a.storage_key or a.file_url or "") for a in assets if a.storage_key or a.file_url
+            }

        for clip in clips:
            asset_id = clip.asset_id
            if not asset_id:
+                failed_clip_ids.append(clip.id)
                continue

-            # 生成安全的本地文件名
-            safe_name = f"clip_{clip.order:04d}_{abs(hash(asset_id)) % 100000:05d}.mp4"
+            # 同一素材已下载过（多个 clip 共享同一素材）
+            if asset_id in seen_asset_ids:
+                if asset_id in asset_path_map:
+                    rendered_clip_ids.append(clip.id)
+                else:
+                    failed_clip_ids.append(clip.id)
+                continue
+
+            seen_asset_ids.add(asset_id)
+
+            # 从素材表获取 OSS 存储路径（file_url）
+            storage_key = asset_storage_map.get(asset_id)
+            if not storage_key:
+                logger.warning(
+                    "素材无 file_url，跳过下载: clip_id=%s asset_id=%s",
+                    clip.id,
+                    asset_id,
+                )
+                failed_clip_ids.append(clip.id)
+                continue
+
+            # 生成安全的本地文件名（保留原始扩展名）
+            ext = Path(storage_key).suffix or ".mp4"
+            safe_name = f"clip_{clip.order:04d}_{abs(hash(asset_id)) % 100000:05d}{ext}"
            local_path = asset_dir / safe_name

-            if download_asset(asset_id, local_path):
+            if download_asset(storage_key, local_path):
                asset_path_map[asset_id] = local_path
+                rendered_clip_ids.append(clip.id)
                logger.debug("素材下载成功: clip_id=%s asset_id=%s", clip.id, asset_id[:60])
            else:
+                failed_clip_ids.append(clip.id)
                logger.warning("素材下载失败: clip_id=%s asset_id=%s", clip.id, asset_id[:60])

-        return asset_path_map
+        return asset_path_map, rendered_clip_ids, failed_clip_ids
+
+    def _prepare_bgm(self, plan, work_dir: Path, plan_id: str) -> str | None:
+        """准备 BGM 音频文件（从 plan.config.bgm 读取配置）。
+
+        支持 3 种来源（按优先级）：
+        1. audio_url — 外部直链 URL
+        2. asset_id — 素材库中的音频素材
+        3. preset_id — 预设 BGM 库
+
+        失败不阻断主流程，返回 None。
+        """
+        from urllib.parse import urlparse
+
+        plan_config = plan.config or {}
+        bgm_config = plan_config.get("bgm", {}) or {}
+
+        if not bgm_config.get("enabled", False):
+            return None
+
+        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 = work_dir / "bgm.mp3"
+
+        # 优先级1：外部直链 URL
+        if audio_url:
+            try:
+                parsed = urlparse(audio_url)
+                if parsed.scheme in ("http", "https"):
+                    from video_processing.url_security import (
+                        ALLOWED_AUDIO_MIME_TYPES,
+                        safe_download_file,
+                    )
+
+                    logger.info("[plan_id=%s] [BGM] 从URL下载: %s", plan_id, audio_url[:80])
+                    safe_download_file(
+                        audio_url,
+                        str(bgm_file),
+                        purpose="bgm_download",
+                        allowed_mime_types=ALLOWED_AUDIO_MIME_TYPES,
+                        timeout=60.0,
+                    )
+                    if bgm_file.exists() and bgm_file.stat().st_size > 0:
+                        return str(bgm_file)
+            except Exception as e:
+                logger.warning("[plan_id=%s] [BGM] URL下载失败: %s", plan_id, e)
+
+        # 优先级2：素材库素材
+        if asset_id:
+            try:
+                from packages.adapters.sqlalchemy_impl.models import AssetModel
+
+                model = self._db.query(AssetModel).filter(AssetModel.id == asset_id).first()
+                if model and (model.storage_key or model.file_url):
+                    # 兼容存量数据：storage_key 为空时 fallback 到 file_url
+                    storage_key = model.storage_key or model.file_url
+                    logger.info("[plan_id=%s] [BGM] 从素材库下载: asset_id=%s", plan_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)
+            except Exception as e:
+                logger.warning("[plan_id=%s] [BGM] 素材库下载失败: %s", plan_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:
+                    from video_processing.url_security import (
+                        ALLOWED_AUDIO_MIME_TYPES,
+                        safe_download_file,
+                    )
+
+                    logger.info("[plan_id=%s] [BGM] 从预设库下载: preset_id=%s", plan_id, preset_id)
+                    safe_download_file(
+                        preset.audio_url,
+                        str(bgm_file),
+                        purpose="bgm_preset_download",
+                        allowed_mime_types=ALLOWED_AUDIO_MIME_TYPES,
+                        timeout=60.0,
+                    )
+                    if bgm_file.exists() and bgm_file.stat().st_size > 0:
+                        return str(bgm_file)
+            except Exception as e:
+                logger.warning("[plan_id=%s] [BGM] 预设库下载失败: %s", plan_id, e)
+
+        logger.warning("[plan_id=%s] [BGM] 所有来源都无法获取BGM，跳过", plan_id)
+        return None
+
+    @staticmethod
+    def _get_asr_service() -> Any | None:
+        """获取 ASR 服务实例（用于自动生成字幕）。
+
+        失败不阻断主流程，返回 None。
+        """
+        try:
+            from services.asr_service_factory import get_asr_service
+
+            return get_asr_service()
+        except Exception as e:
+            logger.warning("ASR 服务初始化失败，自动字幕将不可用: %s", e)
+            return None
@@ -156,7 +156,6 @@ def mix_audio(
        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
@@ -359,7 +359,7 @@ class StickerEngine:
        image_stickers: list[ImageStickerConfig] = []
        image_paths: list[str] = []

-        for i, s in enumerate(stickers):
+        for _, s in enumerate(stickers):
            try:
                sticker_type = s.get("type", "image")
                z = int(s.get("z_index", 10))
@@ -24,13 +24,11 @@
 from __future__ import annotations

 import logging
-from dataclasses import dataclass, field
+from dataclasses import dataclass
 from pathlib import Path
 from typing import Any

 from video_processing.path_security import PathSecurityError, is_in_allowed_dirs, safe_resolve_path
-from video_processing.render_subtitles import generate_ass_subtitles
-from video_processing.subtitle_generator import generate_ass_from_timeline

 logger = logging.getLogger(__name__)

@@ -682,6 +680,6 @@ def _validate_subtitle_path(subtitle_path: str, work_dir: Path) -> None:
        resolved_work_dir = work_dir.resolve()
        try:
            resolved_path.relative_to(resolved_work_dir)
-        except ValueError:
+        except ValueError as _e:
            if not is_in_allowed_dirs(resolved_path):
-                raise PathSecurityError(f"字幕路径不在允许目录内: {subtitle_path[:80]}")
+                raise PathSecurityError(f"字幕路径不在允许目录内: {subtitle_path[:80]}") from _e
@@ -44,7 +44,7 @@ from video_processing.pip_engine import PiPConfig, PiPEngine, PiPLayerConfig
 from video_processing.render_audio import RenderContext, merge_audio_video, mix_audio
 from video_processing.render_subtitles import generate_ass_subtitles
 from video_processing.reverse_engine import ReverseConfig, ReverseEngine
-from video_processing.speed_engine import SpeedConfig, SpeedEngine
+from video_processing.speed_engine import SpeedEngine
 from video_processing.sticker_engine import StickerEngine
 from video_processing.subtitle_generator import generate_ass_from_timeline
 from video_processing.transition_engine import TransitionEngine
@@ -759,6 +759,19 @@ class UnifiedRenderService:
        plan_config = getattr(self.plan, "config", None) or {}
        if isinstance(plan_config, dict) and plan_config.get("stickers"):
            return False
+
+        # 有水印时禁用直通（图片水印需要额外输入，统一走 filter_complex）
+        try:
+            from video_processing.watermark_engine import WatermarkConfig
+
+            wm_config = WatermarkConfig.from_dict(plan_config.get("watermark"))
+            if wm_config is not None and wm_config.validate()[0]:
+                return False
+        except Exception:
+            pass
+
+        # 有调速时仍然可以走直通（视频调速通过 setpts 实现，单输入即可）
+
        return True

    def _can_use_stream_copy(
@@ -971,6 +984,11 @@ class UnifiedRenderService:
            filters.append(f"trim=duration={effective_duration}")
            filters.append("setpts=PTS-STARTPTS")

+        # 调速 — 与 filter_complex 路径一致
+        speed = UnifiedRenderService._clip_speed(clip)
+        if abs(speed - 1.0) >= 1e-6:
+            filters.append(f"setpts=PTS/{speed:.4f}")
+
        # 倒放滤镜
        reverse_config = ReverseConfig.from_dict(clip.config.get("reverse"))
        if reverse_config.enabled and reverse_config.reverse_video:
@@ -1363,23 +1381,37 @@ class UnifiedRenderService:
                # 单 clip 层，直接使用预处理标签
                layer_output_labels[layer.role] = layer_labels[0]
            else:
-                # 多 clip 层，用 TransitionEngine 构建转场链
                out_label = f"{layer.role}_merged"
-                # 计算该层使用的转场时长（取首个非零值，否则用默认）
-                layer_dur = 0.0
-                for d in layer_transition_durations:
-                    if d > 0:
-                        layer_dur = d
-                        break
-                xfade_filter, _ = self._transition_engine.build_xfade_chain(
-                    clip_durations=layer_durations,
-                    clip_video_labels=layer_labels,
-                    transitions=layer_transitions,
-                    transition_duration=layer_dur if layer_dur > 0 else None,
-                    output_label=out_label,
+                # 判断是否全部为硬切：是则用 concat filter，否则用 xfade 转场链
+                all_cut = all(
+                    t is None or t == "" or str(t).lower() == "cut"
+                    for t in layer_transitions[1:]  # 第一个 clip 的转场忽略
                )
-                if xfade_filter:
-                    filter_parts.append(xfade_filter)
+                if all_cut:
+                    # 全硬切：用 concat filter，性能远优于 xfade
+                    concat_inputs = "".join(f"[{label}]" for label in layer_labels)
+                    filter_parts.append(f"{concat_inputs}concat=n={len(layer_labels)}:v=1:a=0[{out_label}]")
+                    logger.info(
+                        "[unified-render] layer=%s clips=%d using concat (all hard-cut)",
+                        layer.role,
+                        len(layer_labels),
+                    )
+                else:
+                    # 有转场效果：用 TransitionEngine 构建 xfade 链
+                    layer_dur = 0.0
+                    for d in layer_transition_durations:
+                        if d > 0:
+                            layer_dur = d
+                            break
+                    xfade_filter, _ = self._transition_engine.build_xfade_chain(
+                        clip_durations=layer_durations,
+                        clip_video_labels=layer_labels,
+                        transitions=layer_transitions,
+                        transition_duration=layer_dur if layer_dur > 0 else None,
+                        output_label=out_label,
+                    )
+                    if xfade_filter:
+                        filter_parts.append(xfade_filter)
                layer_output_labels[layer.role] = out_label

        # Step 3: 合成各层
@@ -243,7 +243,6 @@ class WatermarkEngine:

        # 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}",
@@ -11,7 +11,6 @@ from __future__ import annotations
 import json
 import logging
 import os
-import subprocess
 import tempfile
 from dataclasses import dataclass, field

@@ -187,7 +186,7 @@ class AssetAnalyzer:
        if self._frames is not None:
            return self._frames

-        frames = []
+        frames: list[np.ndarray] = []
        info = self.get_video_info()

        if info.duration <= 0:
@@ -398,7 +397,7 @@ class AssetAnalyzer:
                run_ffmpeg(cmd, timeout=30)
            except Exception:
                # 音频提取失败，返回默认分析结果
-                return AudioAnalysis(
+                return AudioAnalysis(  # type: ignore[call-arg]
                    has_speech=False,
                    speech_ratio=0.0,
                    avg_volume=0.0,
@@ -3,7 +3,6 @@
 from __future__ import annotations

 import logging
-import os
 import tempfile
 import uuid
 import zipfile
@@ -25,7 +24,7 @@ def batch_download_videos(self, video_ids: list[str], user_id: str = "") -> dict
    Returns:
        {"download_url": "...", "file_count": N, "total_size": total_bytes}
    """
-    from video_processing.oss_helpers import download_asset, upload_to_oss
+    from video_processing.oss_helpers import upload_to_oss
    from worker_app.db import SessionLocal

    from packages.adapters.sqlalchemy_impl.generated_video_repository import (
@@ -63,6 +63,18 @@ def compose_video(self, job_id: str, **kwargs):
        resolver = get_render_engine_resolver()
        user_id = job.created_by_user_id or None
        engine = resolver.get_engine(user_id=user_id)
+        # 灰度期间打印详细 flag 配置，便于排查
+        config = resolver.get_config_snapshot()
+        logger.info(
+            "compose_video 引擎选择: job_id=%s engine=%s user_id=%s enabled=%s percentage=%s whitelist=%d default=%s",
+            job_id,
+            engine,
+            user_id,
+            config.get("enabled"),
+            config.get("percentage"),
+            len(config.get("whitelist", [])),
+            config.get("default_engine"),
+        )

        if engine == "unified":
            return _compose_with_unified_engine(self, job_service, job, plan_id, db)
@@ -78,7 +90,7 @@ def compose_video(self, job_id: str, **kwargs):
            job_service.fail_job(job_id, str(exc)[:500])
        except Exception:
            logger.exception("更新 Job 失败状态时出错")
-        raise self.retry(exc=exc, countdown=60)
+        raise self.retry(exc=exc, countdown=60) from exc
    finally:
        db.close()

@@ -39,7 +39,6 @@ from video_processing.oss_helpers import (
    download_asset,
    upload_to_oss,
 )
-from video_processing.unified_render_service import UnifiedRenderService

 # ── Repository imports (延迟导入避免循环依赖) ─────────────────────────────────

@@ -77,9 +76,21 @@ def _resolve_render_engine(user_id: str) -> str:
        from video_processing.render_engine_resolver import get_render_engine_resolver

        resolver = get_render_engine_resolver()
-        return resolver.get_engine(user_id=user_id)
+        engine = resolver.get_engine(user_id=user_id)
+        # 灰度期间打印详细 flag 配置，便于排查
+        config = resolver.get_config_snapshot()
+        logger.info(
+            "edit_plan 引擎选择: user_id=%s engine=%s enabled=%s percentage=%s whitelist=%d default=%s",
+            user_id,
+            engine,
+            config.get("enabled"),
+            config.get("percentage"),
+            len(config.get("whitelist", [])),
+            config.get("default_engine"),
+        )
+        return engine
    except Exception as exc:
-        logger.warning("获取渲染引擎配置失败，fallback 到 legacy: %s", exc)
+        logger.warning("获取渲染引擎配置失败，fallback 到 legacy: %s", exc, exc_info=True)
        return "legacy"


@@ -95,6 +106,14 @@ def _mark_plan_failed(plan_repo, plan_id, gen_task_repo, generation_task_id, err
            gen_task.status = "failed"
            gen_task.error_message = error_msg
            gen_task.completed_at = datetime.now(timezone.utc)
+            try:
+                gen_task.append_log(
+                    stage="render_failed",
+                    message=error_msg[:500],
+                    level="ERROR",
+                )
+            except Exception:
+                pass
            gen_task_repo.update(gen_task)


@@ -148,9 +167,13 @@ def _finalize_render_success(
            clip.mark_rendered()
            clip_repo.update(clip)

-    # 更新 EditPlan 状态为 completed
+    # 更新 EditPlan 状态为 completed + 回写实际渲染时长 + 结果数
    plan.config["rendered_url"] = output_url or ""
    plan.config["rendered_storage_key"] = storage_key
+    if hasattr(plan, "total_duration") and duration > 0:
+        plan.total_duration = duration
+    if hasattr(plan, "result_count"):
+        plan.result_count = 1
    plan.mark_completed()
    plan_repo.update(plan)

@@ -160,7 +183,15 @@ def _finalize_render_success(
        if gen_task:
            gen_task.status = "completed"
            gen_task.progress = 100.0
-            gen_task.result_count = len(rendered_clip_ids)
+            # 剪辑计划是多片段合成 1 个成片，result_count = 1
+            gen_task.result_count = 1
+            gen_task.append_log(
+                stage="render_complete",
+                message=f"渲染完成，输出时长 {duration:.1f}s",
+                level="INFO",
+                engine=engine,
+                clip_count=len(rendered_clip_ids),
+            )
            gen_task.completed_at = datetime.now(timezone.utc)
            gen_task_repo.update(gen_task)

@@ -186,9 +217,6 @@ def _finalize_render_success(
 def _render_with_unified(
    plan,
    clips,
-    asset_path_map: dict[str, Path],
-    tmpdir_path: Path,
-    rendered_clip_ids: list[str],
    plan_id: str,
    generation_task_id: str,
    plan_repo,
@@ -196,31 +224,62 @@ def _render_with_unified(
    gen_task_repo,
    db,
 ) -> dict:
-    """统一渲染引擎路径（UnifiedRenderService 图层架构）。"""
-    render_service = UnifiedRenderService(
-        plan=plan,
-        clips=clips,
-        asset_path_map=asset_path_map,
-        work_dir=tmpdir_path,
-        output_width=OUTPUT_WIDTH,
-        output_height=OUTPUT_HEIGHT,
-        output_fps=int(OUTPUT_FPS),
-    )
+    """统一渲染引擎路径（通过 RenderAdapter 调用 UnifiedRenderService）。
+
+    RenderAdapter 内部处理：素材下载、BGM 准备、ASR 自动字幕、渲染执行、OSS 上传。
+    本函数只负责：业务状态更新、查重、收尾。
+    """
+    from video_processing.render_adapter import RenderAdapter
+
+    adapter = RenderAdapter(db)
+
+    # 进度回调：更新 GenerationTask 进度
+    def _progress_cb(progress: float, stage: str):
+        if not generation_task_id:
+            return
+        try:
+            gen_task = gen_task_repo.get(generation_task_id)
+            if gen_task:
+                # 映射到 30%~90% 区间（素材下载前已到 30%）
+                mapped_progress = 30.0 + progress * 0.6
+                gen_task.progress = min(mapped_progress, 95.0)
+                gen_task.append_log(
+                    stage="render_progress",
+                    message=stage,
+                    level="INFO",
+                    progress=mapped_progress,
+                )
+                gen_task_repo.update(gen_task)
+        except Exception:
+            pass

    try:
-        render_result = render_service.render()
+        result = adapter.render_plan(
+            plan_id=plan_id,
+            job_id=generation_task_id or plan_id,
+            progress_cb=_progress_cb,
+        )
    except Exception as render_err:
        logger.error("渲染失败(unified): %s — %s", plan_id, render_err)
        _mark_plan_failed(plan_repo, plan_id, gen_task_repo, generation_task_id, f"渲染失败: {render_err}")
        return {"status": "error", "message": f"渲染失败: {render_err}"}

-    output_path = render_result.output_path
+    if not result.success:
+        full_error = result.error_message or "渲染失败"
+        if result.error_detail:
+            full_error = f"{full_error}\n--- stderr ---\n{result.error_detail}"
+        logger.error("渲染失败(unified): %s — %s", plan_id, result.error_message)
+        _mark_plan_failed(plan_repo, plan_id, gen_task_repo, generation_task_id, full_error)
+        return {"status": "error", "message": result.error_message or "渲染失败"}

-    # 上传到 OSS
+    output_path = result.output_path or Path("")
+    output_url = result.output_url
    storage_key = f"rendered/{plan_id}/output.mp4"
-    output_url = upload_to_oss(output_path, storage_key)

-    failed_clip_ids: list[str] = []
+    # 用 adapter 返回的 clip 明细（以 adapter 的结果为准）
+    rendered_clip_ids = result.rendered_clip_ids or []
+    failed_clip_ids = result.failed_clip_ids or []
+
    return _finalize_render_success(
        plan=plan,
        plan_repo=plan_repo,
@@ -230,10 +289,10 @@ def _render_with_unified(
        plan_id=plan_id,
        output_url=output_url or "",
        storage_key=storage_key,
-        duration=render_result.duration,
-        file_size=render_result.file_size,
-        width=render_result.width,
-        height=render_result.height,
+        duration=result.duration,
+        file_size=result.file_size,
+        width=result.width,
+        height=result.height,
        rendered_clip_ids=rendered_clip_ids,
        failed_clip_ids=failed_clip_ids,
        generation_task_id=generation_task_id,
@@ -365,89 +424,134 @@ def render_edit_plan(self, plan_id: str) -> dict:
                if gen_task:
                    gen_task.status = "running"
                    gen_task.started_at = datetime.now(timezone.utc)
+                    gen_task.append_log(
+                        stage="render_start",
+                        message=f"开始渲染，引擎 {engine}，片段数 {len(clips)}",
+                        level="INFO",
+                        engine=engine,
+                        clip_count=len(clips),
+                    )
                    gen_task_repo.update(gen_task)

-            # 3. 下载素材并构建 asset_path_map
-            with tempfile.TemporaryDirectory(prefix="edit_plan_") as tmpdir:
-                tmpdir_path = Path(tmpdir)
-                asset_path_map: dict[str, Path] = {}
-                rendered_clip_ids: list[str] = []
-                failed_clip_ids: list[str] = []
+            # 3. 渲染前取消检查
+            if generation_task_id:
+                current_task = gen_task_repo.get(generation_task_id)
+                if current_task:
+                    task_status = (
+                        current_task.status.value if hasattr(current_task.status, "value") else str(current_task.status)
+                    )
+                    if task_status == "cancelled":
+                        logger.info("任务已被取消，中止渲染: plan_id=%s task_id=%s", plan_id, generation_task_id)

-                # 预先批量查询所有素材的 storage_key（file_url）
-                from packages.adapters.sqlalchemy_impl.models import AssetModel
+                        if plan.status.value == "rendering":
+                            try:
+                                plan.resume_editing()
+                                plan_repo.update(plan)
+                            except ValueError:
+                                pass
+                        return {"status": "cancelled", "plan_id": plan_id, "message": "任务已取消"}

-                clip_asset_ids = [c.asset_id for c in clips if c.asset_id]
-                asset_storage_map: dict[str, str] = {}
-                if clip_asset_ids:
-                    assets = db.query(AssetModel).filter(AssetModel.id.in_(clip_asset_ids)).all()
-                    asset_storage_map = {a.id: a.file_url for a in assets if a.file_url}
+            # 4. 根据引擎选择渲染方式
+            if engine == "unified":
+                # ── unified 路径：RenderAdapter 统一处理（下载 + BGM + ASR + 渲染 + 上传）
+                result = _render_with_unified(
+                    plan=plan,
+                    clips=clips,
+                    plan_id=plan_id,
+                    generation_task_id=generation_task_id,
+                    plan_repo=plan_repo,
+                    clip_repo=clip_repo,
+                    gen_task_repo=gen_task_repo,
+                    db=db,
+                )
+            else:
+                # ── legacy 路径：原有的素材下载 + VideoComposeService
+                with tempfile.TemporaryDirectory(prefix="edit_plan_") as tmpdir:
+                    tmpdir_path = Path(tmpdir)
+                    asset_path_map: dict[str, Path] = {}
+                    rendered_clip_ids: list[str] = []
+                    failed_clip_ids: list[str] = []

-                for clip in clips:
-                    if not clip.asset_id:
-                        # 没有素材的片段跳过，标记为失败
-                        clip.mark_failed()
-                        clip_repo.update(clip)
-                        failed_clip_ids.append(clip.id)
-                        continue
+                    # 预先批量查询所有素材的 storage_key
+                    # 兼容存量数据：storage_key 为空时 fallback 到 file_url
+                    from packages.adapters.sqlalchemy_impl.models import AssetModel

-                    if clip.asset_id in asset_path_map:
-                        # 同一素材已下载（多个 clip 共享同一素材）
-                        rendered_clip_ids.append(clip.id)
-                        continue
+                    clip_asset_ids = [c.asset_id for c in clips if c.asset_id]
+                    asset_storage_map: dict[str, str] = {}
+                    if clip_asset_ids:
+                        assets = db.query(AssetModel).filter(AssetModel.id.in_(clip_asset_ids)).all()
+                        asset_storage_map = {
+                            a.id: (a.storage_key or a.file_url or "") for a in assets if a.storage_key or a.file_url
+                        }

-                    storage_key = asset_storage_map.get(clip.asset_id)
-                    if not storage_key:
-                        logger.warning(
-                            "片段素材无 storage_key，跳过: clip_id=%s asset_id=%s",
-                            clip.id,
-                            clip.asset_id,
-                        )
-                        clip.mark_failed()
-                        clip_repo.update(clip)
-                        failed_clip_ids.append(clip.id)
-                        continue
+                    for clip in clips:
+                        if not clip.asset_id:
+                            # 没有素材的片段跳过，标记为失败
+                            clip.mark_failed()
+                            clip_repo.update(clip)
+                            failed_clip_ids.append(clip.id)
+                            continue

-                    # 下载素材
-                    ext = Path(storage_key).suffix or ".mp4"
-                    local_path = tmpdir_path / f"clip_{clip.order:04d}{ext}"
-                    if download_asset(storage_key, local_path):
-                        asset_path_map[clip.asset_id] = local_path
-                        rendered_clip_ids.append(clip.id)
-                    else:
-                        clip.mark_failed()
-                        clip_repo.update(clip)
-                        failed_clip_ids.append(clip.id)
+                        if clip.asset_id in asset_path_map:
+                            # 同一素材已下载（多个 clip 共享同一素材）
+                            rendered_clip_ids.append(clip.id)
+                            continue

-                if not asset_path_map:
-                    logger.error("所有片段素材下载失败: %s", plan_id)
-                    plan.mark_failed()
-                    plan_repo.update(plan)
+                        storage_key = asset_storage_map.get(clip.asset_id)
+                        if not storage_key:
+                            logger.warning(
+                                "片段素材无 storage_key，跳过: clip_id=%s asset_id=%s",
+                                clip.id,
+                                clip.asset_id,
+                            )
+                            clip.mark_failed()
+                            clip_repo.update(clip)
+                            failed_clip_ids.append(clip.id)
+                            continue
+
+                        # 下载素材
+                        ext = Path(storage_key).suffix or ".mp4"
+                        local_path = tmpdir_path / f"clip_{clip.order:04d}{ext}"
+                        if download_asset(storage_key, local_path):
+                            asset_path_map[clip.asset_id] = local_path
+                            rendered_clip_ids.append(clip.id)
+                        else:
+                            clip.mark_failed()
+                            clip_repo.update(clip)
+                            failed_clip_ids.append(clip.id)
+
+                    if not asset_path_map:
+                        logger.error("所有片段素材下载失败: %s", plan_id)
+                        plan.mark_failed()
+                        plan_repo.update(plan)
+                        if generation_task_id:
+                            gen_task = gen_task_repo.get(generation_task_id)
+                            if gen_task:
+                                gen_task.status = "failed"
+                                gen_task.error_message = "所有片段素材下载失败"
+                                gen_task.completed_at = datetime.now(timezone.utc)
+                                gen_task.append_log(
+                                    stage="download_failed",
+                                    message="所有片段素材下载失败",
+                                    level="ERROR",
+                                )
+                                gen_task_repo.update(gen_task)
+                        return {"status": "error", "message": "所有片段素材下载失败"}
+
+                    # 素材下载完成，记录日志
                    if generation_task_id:
                        gen_task = gen_task_repo.get(generation_task_id)
                        if gen_task:
-                            gen_task.status = "failed"
-                            gen_task.error_message = "所有片段素材下载失败"
-                            gen_task.completed_at = datetime.now(timezone.utc)
+                            gen_task.append_log(
+                                stage="download_done",
+                                message=f"素材下载完成，成功 {len(asset_path_map)} 个，失败 {len(failed_clip_ids)} 个",
+                                level="INFO",
+                                success_count=len(asset_path_map),
+                                failed_count=len(failed_clip_ids),
+                            )
+                            gen_task.progress = 30.0
                            gen_task_repo.update(gen_task)
-                    return {"status": "error", "message": "所有片段素材下载失败"}

-                # 4. 根据引擎选择渲染方式
-                if engine == "unified":
-                    result = _render_with_unified(
-                        plan=plan,
-                        clips=clips,
-                        asset_path_map=asset_path_map,
-                        tmpdir_path=tmpdir_path,
-                        rendered_clip_ids=rendered_clip_ids,
-                        plan_id=plan_id,
-                        generation_task_id=generation_task_id,
-                        plan_repo=plan_repo,
-                        clip_repo=clip_repo,
-                        gen_task_repo=gen_task_repo,
-                        db=db,
-                    )
-                else:
                    result = _render_with_legacy(
                        plan=plan,
                        clips=clips,
@@ -462,8 +566,8 @@ def render_edit_plan(self, plan_id: str) -> dict:
                        db=db,
                    )

-                result["engine"] = engine
-                return result
+            result["engine"] = engine
+            return result

        except Exception as exc:
            logger.exception("渲染剪辑计划异常: %s", plan_id)
@@ -483,6 +587,15 @@ def render_edit_plan(self, plan_id: str) -> dict:
                        gen_task.status = "failed"
                        gen_task.error_message = f"渲染异常: {type(exc).__name__}: {exc}"
                        gen_task.completed_at = datetime.now(timezone.utc)
+                        try:
+                            gen_task.append_log(
+                                stage="render_failed",
+                                message=f"渲染异常: {type(exc).__name__}: {str(exc)[:500]}",
+                                level="ERROR",
+                                exception_type=type(exc).__name__,
+                            )
+                        except Exception:
+                            pass
                        gen_task_repo.update(gen_task)
                        logger.info(
                            "GenerationTask 已标记为 failed: task_id=%s plan_id=%s",
@@ -493,6 +606,6 @@ def render_edit_plan(self, plan_id: str) -> dict:
                logger.warning(
                    "更新 GenerationTask 失败状态时异常: task_id=%s error=%s", generation_task_id, e, exc_info=True
                )
-            raise self.retry(exc=exc, countdown=60)
+            raise self.retry(exc=exc, countdown=60) from exc

    return {"status": "error", "message": "数据库连接失败"}
@@ -178,10 +178,136 @@ class _VirtualClip:
    config: dict[str, Any] = field(default_factory=dict)


+def _load_template_clip_configs(template_id: str) -> list:
+    """从数据库读取模板的片段配置列表。
+
+    失败返回空列表，不阻断主流程。
+    """
+    if not template_id:
+        return []
+    try:
+        from worker_app.db import SessionLocal
+
+        from packages.adapters.sqlalchemy_impl.template_clip_config_repository import (
+            SQLAlchemyTemplateClipConfigRepository,
+        )
+
+        session = SessionLocal()
+        try:
+            repo = SQLAlchemyTemplateClipConfigRepository(session)
+            configs = repo.list_by_template(template_id, limit=200)
+            logger.info("读取模板片段配置: template_id=%s count=%d", template_id, len(configs))
+            return configs
+        finally:
+            session.close()
+    except Exception as e:
+        logger.warning("读取模板片段配置失败，跳过效果层映射: template_id=%s error=%s", template_id, e)
+        return []
+
+
+def _extract_intro_outro_from_clip_configs(clip_configs: list) -> dict[str, Any]:
+    """从模板的 intro/outro 类型 clip_config 中提取 plan 级 intro_outro 配置。
+
+    UnifiedRenderService 已支持 plan.config.intro_outro 路径，
+    这里把 intro/outro 片段配置转为统一格式注入。
+    """
+    intro_configs = [
+        c for c in clip_configs if (c.clip_type.value if hasattr(c.clip_type, "value") else c.clip_type) == "intro"
+    ]
+    outro_configs = [
+        c for c in clip_configs if (c.clip_type.value if hasattr(c.clip_type, "value") else c.clip_type) == "outro"
+    ]
+
+    result: dict[str, Any] = {}
+
+    if intro_configs:
+        intro = intro_configs[0]
+        intro_cfg = intro.config or {}
+        result["has_intro"] = True
+        result["intro_type"] = intro_cfg.get("intro_type", "text")
+        result["intro_duration"] = intro.default_duration or 3.0
+        if intro.text_template:
+            result["intro_text"] = intro.text_template
+        # 透传额外配置
+        for key in ("intro_text_color", "intro_bg_color", "intro_font_size", "intro_video_url", "intro_video_path"):
+            if key in intro_cfg:
+                result[key] = intro_cfg[key]
+
+    if outro_configs:
+        outro = outro_configs[0]
+        outro_cfg = outro.config or {}
+        result["has_outro"] = True
+        result["outro_type"] = outro_cfg.get("outro_type", "text")
+        result["outro_duration"] = outro.default_duration or 3.0
+        if outro.text_template:
+            result["outro_text"] = outro.text_template
+        for key in ("outro_text_color", "outro_bg_color", "outro_font_size", "outro_follow_text"):
+            if key in outro_cfg:
+                result[key] = outro_cfg[key]
+
+    return result
+
+
+def _apply_template_clip_effects(
+    clips: list[_VirtualClip],
+    clip_configs: list,
+    mode: str,
+) -> None:
+    """将模板的 clip 级效果层映射到素材 clips 上（就地修改）。
+
+    映射规则：
+    - 只对素材主体 clips 做映射（ONE_TAKE: main, PIP: main+overlay, VOICE_OVER: main, VOICE_PIP: background+b_roll）
+    - 从模板中筛选 main 类型的 clip_config 作为效果模板
+    - 素材 clips 按顺序循环匹配模板 clip_config（素材多的话重复使用最后一个模板配置）
+    - 映射字段：transition_effect, config.color_grade, config.speed
+    """
+    if not clip_configs or not clips:
+        return
+
+    # 筛选 main 类型的模板配置（作为效果模板池）
+    main_configs = [
+        c
+        for c in clip_configs
+        if (c.clip_type.value if hasattr(c.clip_type, "value") else c.clip_type) in ("main", "showcase", "b_roll")
+    ]
+    if not main_configs:
+        return
+
+    # 确定需要映射的素材 clips（排除 corner_voice 等特殊层）
+    target_clips = [c for c in clips if c.clip_type not in ("corner_voice",)]
+
+    for i, clip in enumerate(target_clips):
+        # 循环匹配：素材多了用最后一个模板配置
+        cfg_idx = min(i, len(main_configs) - 1)
+        template_cfg = main_configs[cfg_idx]
+
+        # 1. 转场效果
+        transition = (
+            template_cfg.transition_effect.value
+            if hasattr(template_cfg.transition_effect, "value")
+            else template_cfg.transition_effect
+        )
+        if transition and transition != "cut":
+            clip.transition_effect = transition
+
+        # 2. clip 级效果配置（滤镜、调速等）
+        template_clip_config = template_cfg.config or {}
+        if template_clip_config:
+            # 合并到 clip.config（保留已有配置如 role 等）
+            existing_config = clip.config or {}
+            # 需要从模板复制的效果层 key
+            effect_keys = ("color_grade", "speed", "playback_speed", "reverse", "chroma_key", "filter")
+            for key in effect_keys:
+                if key in template_clip_config:
+                    existing_config[key] = template_clip_config[key]
+            clip.config = existing_config
+
+
 def _build_plan_and_clips_from_task(
    task_id: str,
    downloaded_paths: list[Path],
    mode: str,
+    template_id: str = "",
 ) -> tuple[_VirtualPlan, list[_VirtualClip], dict[str, Path]]:
    """根据模式和下载的素材路径，构建虚拟 plan + clips + asset_path_map。

@@ -269,6 +395,25 @@ def _build_plan_and_clips_from_task(
                )
            )

+    # ── P1: 模板效果层映射 ──
+    if template_id:
+        clip_configs = _load_template_clip_configs(template_id)
+        if clip_configs:
+            # 1. clip级效果层（转场、滤镜、调速等）
+            _apply_template_clip_effects(clips, clip_configs, mode)
+
+            # 2. 片头片尾（从 intro/outro 类型 clip 提取 plan 级配置）
+            intro_outro_config = _extract_intro_outro_from_clip_configs(clip_configs)
+            if intro_outro_config:
+                plan_config = plan.config or {}
+                plan_config["intro_outro"] = intro_outro_config
+                plan.config = plan_config
+                logger.info(
+                    "模板片头片尾配置已注入: has_intro=%s has_outro=%s",
+                    intro_outro_config.get("has_intro", False),
+                    intro_outro_config.get("has_outro", False),
+                )
+
    return plan, clips, asset_path_map


@@ -392,8 +537,9 @@ def _prepare_bgm_track(
            session = SessionLocal()
            try:
                model = session.query(AssetModel).filter(AssetModel.id == asset_id).first()
-                if model and model.file_url:
-                    storage_key = model.file_url
+                if model and (model.storage_key or model.file_url):
+                    # 兼容存量数据：storage_key 为空时 fallback 到 file_url
+                    storage_key = model.storage_key or 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:
@@ -774,6 +920,47 @@ def _validate_template_exists(template_id: str) -> None:
        session.close()


+def _load_template_plan_config(template_id: str) -> dict:
+    """从模板加载 plan 级配置（BGM、字幕、滤镜等效果层）。
+
+    模板不存在时返回空 dict，不阻塞主流程。
+    """
+    if not template_id:
+        return {}
+    try:
+        from packages.adapters.sqlalchemy_impl.models import TemplateModel
+
+        session = SessionLocal()
+        try:
+            template = (
+                session.query(TemplateModel)
+                .filter(
+                    TemplateModel.id == template_id,
+                    TemplateModel.is_active.is_(True),
+                )
+                .first()
+            )
+            if template is None:
+                logger.warning("模板不存在，跳过配置加载: template_id=%s", template_id)
+                return {}
+            config = template.config or {}
+            if isinstance(config, str):
+                import json
+
+                config = json.loads(config)
+            logger.info(
+                "模板配置加载成功: template_id=%s keys=%s",
+                template_id,
+                list(config.keys()),
+            )
+            return config
+        finally:
+            session.close()
+    except Exception as e:
+        logger.warning("加载模板配置失败，跳过: template_id=%s err=%s", template_id, e)
+        return {}
+
+
 # ── 渲染引擎选择 ─────────────────────────────────────────────────────────────


@@ -787,10 +974,23 @@ def _resolve_render_engine(user_id: str) -> str:
        from video_processing.render_engine_resolver import get_render_engine_resolver

        resolver = get_render_engine_resolver()
-        return resolver.get_engine(user_id=user_id)
+        engine = resolver.get_engine(user_id=user_id)
+        # 灰度期间打印详细 flag 配置，便于排查
+        config = resolver.get_config_snapshot()
+        logger.info(
+            "[渲染引擎] flag 解析: user_id=%s engine=%s enabled=%s percentage=%s whitelist=%d default=%s",
+            user_id,
+            engine,
+            config.get("enabled"),
+            config.get("percentage"),
+            len(config.get("whitelist", [])),
+            config.get("default_engine"),
+        )
+        return engine
    except Exception as exc:
-        logger.warning("获取渲染引擎配置失败，fallback 到 unified: %s", exc)
-        return ENGINE_UNIFIED
+        # 异常时 fallback 到 legacy（保守策略，与 edit_plan_generation 一致）
+        logger.warning("获取渲染引擎配置失败，fallback 到 legacy: %s", exc, exc_info=True)
+        return ENGINE_LEGACY


 # ── 旧引擎渲染（FFmpeg filter_complex） ────────────────────────────────────────
@@ -1017,8 +1217,22 @@ def _render_video(
        task_id=task_id,
        downloaded_paths=downloaded_videos,
        mode=editing_mode.value,
+        template_id=template_id,
    )

+    # 注入模板配置（BGM、字幕等效果层）
+    if template_id:
+        template_config = _load_template_plan_config(template_id)
+        if template_config:
+            # 合并：现有 config 优先级更高（目前为空，模板配置直接生效）
+            base_config = virtual_plan.config or {}
+            virtual_plan.config = {**template_config, **base_config}
+            logger.info(
+                "[task_id=%s] [渲染] 模板配置已注入: keys=%s",
+                task_id,
+                list(template_config.keys()),
+            )
+
    total_duration = sum(c.duration for c in virtual_clips)
    logger.info(
        "[task_id=%s] [剪辑计划] 片段数=%d, 总时长=%.1fs",
@@ -1405,7 +1619,7 @@ def generate_video(self, task_id: str) -> dict:
                _repo = SQLAlchemyGenerationTaskRepository(_session)
                gen_task = _repo.get(task_id)
                if gen_task:
-                    gen_task.append_log(
+                    gen_task.append_log(  # type: ignore[misc]
                        "任务失败",
                        str(error),
                        level="ERROR",
@@ -1,7 +1,10 @@
 import subprocess
+import tempfile
 from datetime import datetime, timezone
+from pathlib import Path

 from celery.utils.log import get_task_logger
+from video_processing.oss_helpers import download_asset
 from worker_app.celery_app import celery_app
 from worker_app.core.asset_types import infer_mime_type_from_storage_key
 from worker_app.db import SessionLocal
@@ -15,6 +18,32 @@ from packages.domain import Asset, AssetStatus, IngestJobStatus
 logger = get_task_logger(__name__)


+# 最小有效文件大小（字节）：小于此值的直接判为无效，避免文本/空文件伪装成媒体
+MIN_VIDEO_FILE_SIZE = 1024  # 1KB
+MIN_AUDIO_FILE_SIZE = 100  # 100B
+MIN_IMAGE_FILE_SIZE = 100  # 100B
+
+# 支持的视频编码格式（白名单）
+# 渲染引擎针对 H.264 优化，其他编码可能导致渲染失败或异常
+SUPPORTED_VIDEO_CODECS = {"h264", "avc1", "avc"}
+
+# 明确不支持的视频编码（用于日志提示）
+UNSUPPORTED_VIDEO_CODECS = {
+    "hevc",
+    "h265",
+    "hev1",
+    "hvc1",  # H.265 / HEVC
+    "vp9",
+    "vp09",  # VP9
+    "av1",
+    "av01",  # AV1
+    "vp8",
+    "vp08",  # VP8
+    "mpeg4",
+    "mp4v",  # MPEG-4
+}
+
+
 def _safe_parse_fps(fps_str: str) -> float:
    """Safely parse fps from a fraction string like \"30/1\" or \"30000/1001\"."""
    try:
@@ -29,7 +58,7 @@ def _safe_parse_fps(fps_str: str) -> float:
        return 0.0


-def extract_media_metadata(file_url: str, media_type: str) -> dict:
+def extract_media_metadata(file_url: str, media_type: str) -> tuple[dict, bool]:
    """
    提取媒体文件的元数据。

@@ -38,9 +67,12 @@ def extract_media_metadata(file_url: str, media_type: str) -> dict:
        media_type: 媒体类型 (video, audio, image)

    Returns:
-        提取的元数据字典，失败时返回空字典
+        (metadata_dict, success)
+        - metadata_dict: 提取的元数据字典，失败时返回空字典
+        - success: 是否成功提取到有效元数据
    """
    metadata = {}
+    success = False

    try:
        if media_type == "video":
@@ -70,33 +102,76 @@ def extract_media_metadata(file_url: str, media_type: str) -> dict:
                        metadata["height"] = int(stream.get("height", 0))
                        metadata["codec"] = stream.get("codec_name", "")
                        metadata["fps"] = (
-                            _safe_parse_fps(stream.get("r_frame_rate", "0/1")) if stream.get("r_frame_rate") else 0
+                            _safe_parse_fps(stream.get("r_frame_rate", "0/1")) if stream.get("r_frame_rate") else 0  # type: ignore[assignment]
                        )
                        break

                # 提取格式信息
                format_info = probe_data.get("format", {})
-                metadata["duration"] = float(format_info.get("duration", 0))
+                metadata["duration"] = float(format_info.get("duration", 0))  # type: ignore[assignment]
                metadata["size_bytes"] = int(format_info.get("size", 0))
                metadata["bitrate"] = int(format_info.get("bit_rate", 0))

+                # 有效性判断：有视频流 + duration > 0 + size > 0
+                has_video_stream = any(s.get("codec_type") == "video" for s in probe_data.get("streams", []))
+                has_audio_stream = any(s.get("codec_type") == "audio" for s in probe_data.get("streams", []))
+                if (has_video_stream or has_audio_stream) and metadata.get("duration", 0) > 0:
+                    success = True
+
            except Exception as e:
                logger.warning("视频元数据提取失败: %s", e)

+        elif media_type == "audio":
+            from video_processing.ffmpeg_utils import run_ffprobe
+
+            cmd = [
+                "ffprobe",
+                "-v",
+                "quiet",
+                "-print_format",
+                "json",
+                "-show_format",
+                "-show_streams",
+                file_url,
+            ]
+            try:
+                stdout, _ = run_ffprobe(cmd, timeout=30)
+                import json as json_lib
+
+                probe_data = json_lib.loads(stdout)
+
+                has_audio_stream = any(s.get("codec_type") == "audio" for s in probe_data.get("streams", []))
+                format_info = probe_data.get("format", {})
+                metadata["duration"] = float(format_info.get("duration", 0))  # type: ignore[assignment]
+                metadata["size_bytes"] = int(format_info.get("size", 0))
+                metadata["bitrate"] = int(format_info.get("bit_rate", 0))
+                metadata["codec"] = next(
+                    (s.get("codec_name", "") for s in probe_data.get("streams", []) if s.get("codec_type") == "audio"),
+                    "",
+                )
+
+                if has_audio_stream and metadata.get("duration", 0) > 0:
+                    success = True
+
+            except Exception as e:
+                logger.warning("音频元数据提取失败: %s", e)
+
        elif media_type == "image":
            # 使用 Pillow 提取图片元数据
            try:
                from PIL import Image

                with Image.open(file_url) as img:
-                    metadata["width"] = img.width
-                    metadata["height"] = img.height
-                    metadata["format"] = img.format
-                    metadata["mode"] = img.mode
-                    if hasattr(img, "_getexif") and img._getexif():
-                        exif = img._getexif()
-                        if exif:
-                            metadata["exif"] = {k: str(v) for k, v in exif.items() if isinstance(v, (str, int, float))}
+                    img.verify()  # 验证文件完整性
+                    # verify后需要重新打开才能读尺寸
+                    with Image.open(file_url) as img2:
+                        metadata["width"] = img2.width
+                        metadata["height"] = img2.height
+                        metadata["format"] = img2.format
+                        metadata["mode"] = img2.mode
+                        if img2.width > 0 and img2.height > 0:
+                            success = True
+
            except ImportError:
                logger.warning("Pillow not available for image metadata extraction")
            except Exception as e:
@@ -109,7 +184,38 @@ def extract_media_metadata(file_url: str, media_type: str) -> dict:
    except Exception as e:
        logger.warning(f"Failed to extract metadata: {e}")

-    return metadata
+    return metadata, success
+
+
+def _is_valid_media(metadata: dict, media_type: str) -> bool:
+    """根据元数据判断文件是否为有效媒体文件。
+
+    Args:
+        metadata: extract_media_metadata 返回的元数据
+        media_type: 媒体类型
+
+    Returns:
+        True 表示文件有效
+    """
+    size = int(metadata.get("size_bytes", 0))
+
+    if media_type == "video":
+        duration = float(metadata.get("duration", 0))
+        if size < MIN_VIDEO_FILE_SIZE or duration <= 0:
+            return False
+        # 编码格式校验：仅支持 H.264，其他编码（HEVC/VP9/AV1等）渲染可能失败
+        codec = str(metadata.get("codec", "")).lower()
+        if codec and codec not in SUPPORTED_VIDEO_CODECS:
+            return False
+        return True
+    if media_type == "audio":
+        duration = float(metadata.get("duration", 0))
+        return size >= MIN_AUDIO_FILE_SIZE and duration > 0
+    if media_type == "image":
+        width = int(metadata.get("width", 0))
+        height = int(metadata.get("height", 0))
+        return size >= MIN_IMAGE_FILE_SIZE and width > 0 and height > 0
+    return False


@celery_app.task(name="worker.ingest_asset")
@@ -150,17 +256,71 @@ def ingest_asset(job_id: str) -> dict:
        elif mime_type.startswith("audio/"):
            media_type = "audio"

-        # Extract metadata (returns empty dict on failure)
-        storage_url = job.storage_key  # Assuming storage_key is usable as URL/path
-        metadata = extract_media_metadata(storage_url, media_type)
+        # 先从 OSS 下载文件到本地临时目录，再提取元数据
+        # （storage_key 是 OSS 内部路径，不能直接传给 ffprobe/Pillow）
+        local_file = None
+        try:
+            suffix = Path(job.storage_key).suffix or ".bin"
+            with tempfile.NamedTemporaryFile(suffix=suffix, delete=False) as tmp:
+                local_file = Path(tmp.name)

-        # Fill in defaults if metadata extraction failed
-        if not metadata:
-            metadata = {
-                "duration": 0,
-                "width": 0,
-                "height": 0,
-                "size_bytes": 0,
+            download_ok = download_asset(job.storage_key, local_file)
+            if not download_ok:
+                logger.warning("素材下载失败，无法提取元数据: job_id=%s storage_key=%s", job_id, job.storage_key)
+                metadata, extract_success = {}, False
+            else:
+                metadata, extract_success = extract_media_metadata(str(local_file), media_type)
+        finally:
+            if local_file and local_file.exists():
+                try:
+                    local_file.unlink()
+                except OSError:
+                    pass
+
+        # 有效性校验：ffprobe/Pillow 必须成功，且文件大小/时长/尺寸满足最小要求
+        is_valid = extract_success and _is_valid_media(metadata, media_type)
+
+        if not is_valid:
+            # 文件无效，创建 ERROR 状态的 asset 并标记 job 失败
+            error_reason = "metadata extraction failed" if not extract_success else "media validation failed"
+            logger.warning(
+                "素材有效性校验失败，标记为ERROR: job_id=%s storage_key=%s media_type=%s reason=%s",
+                job_id,
+                job.storage_key,
+                media_type,
+                error_reason,
+            )
+
+            asset = Asset.create(
+                project_id=job.project_id,
+                library_id=job.library_id,
+                name=filename,
+                storage_key=job.storage_key,
+                mime_type=mime_type,
+                metadata={"ingest_error": error_reason},
+                file_size=int(metadata.get("size_bytes", 0)),
+                duration=float(metadata.get("duration", 0)),
+                width=int(metadata.get("width", 0)),
+                height=int(metadata.get("height", 0)),
+                codec=metadata.get("codec") or None,
+                status=AssetStatus.ERROR,
+                file_hash=job.file_hash,
+            )
+            asset_repo.create(asset)
+
+            # Update job status to FAILED
+            job.status = IngestJobStatus.FAILED
+            job.error_message = f"Invalid media file: {error_reason}"
+            job.result_asset_id = asset.id
+            job.updated_at = datetime.now(timezone.utc)
+            job_repo.update(job)
+            db.commit()
+
+            return {
+                "status": "failed",
+                "job_id": job.id,
+                "asset_id": asset.id,
+                "error": error_reason,
            }

        # Create Asset
@@ -175,6 +335,7 @@ def ingest_asset(job_id: str) -> dict:
            duration=float(metadata.get("duration", 0)),
            width=int(metadata.get("width", 0)),
            height=int(metadata.get("height", 0)),
+            codec=metadata.get("codec") or None,
            status=AssetStatus.READY,
            file_hash=job.file_hash,
        )
@@ -63,7 +63,7 @@ def process_tts_synthesis(self: Task, job_id: str) -> dict:
        if session is not None:
            session.rollback()
        # 超时重试，指数退避
-        raise self.retry(exc=e, countdown=30)
+        raise self.retry(exc=e, countdown=30) from e

    except CosyVoiceError as e:
        logger.error(f"TTS synthesis failed for {job_id}: {e}")
@@ -139,7 +139,7 @@ def process_tts_segment_synthesis(self: Task, job_id: str) -> dict:
        logger.warning(f"TTS segment synthesis timeout for {job_id}: {e}")
        if session is not None:
            session.rollback()
-        raise self.retry(exc=e, countdown=60)
+        raise self.retry(exc=e, countdown=60) from e

    except CosyVoiceError as e:
        logger.error(f"TTS segment synthesis failed for {job_id}: {e}")
@@ -73,7 +73,7 @@ def process_voice_clone(self: Task, profile_id: str) -> dict:
        if session is not None:
            session.rollback()
        # 超时属于临时性故障，延迟 30 秒后重试
-        raise self.retry(exc=e, countdown=30)
+        raise self.retry(exc=e, countdown=30) from e

    except CosyVoiceError as e:
        logger.error(f"Voice clone failed for {profile_id}: {e}")
@@ -2,7 +2,6 @@

 import logging
 import os
-import subprocess
 import tempfile

 from celery import Task
@@ -105,7 +104,7 @@ def extract_voice_task(self: Task, asset_id: str) -> dict:
    except Exception as e:
        logger.error(f"Voice extraction failed for {asset_id}: {str(e)}")
        session.rollback()
-        raise self.retry(exc=e, countdown=60)
+        raise self.retry(exc=e, countdown=60) from e
    finally:
        session.close()
        import shutil
@@ -141,7 +140,7 @@ def extract_background_task(self: Task, asset_id: str) -> dict:
    except Exception as e:
        logger.error(f"Background extraction failed for {asset_id}: {str(e)}")
        session.rollback()
-        raise self.retry(exc=e, countdown=60)
+        raise self.retry(exc=e, countdown=60) from e
    finally:
        session.close()
        import shutil
@@ -0,0 +1,104 @@
+# CI 大量失败根因排查报告
+
+**排查时间：** 2026-07-13
+**排查人：** 构建服务器运维Agent
+**范围：** 最近15次 CI run（PR #258~#265 + develop 分支多次 push）
+
+## 一、整体概况
+
+最近 20 次 CI run 中 16 次失败，失败率 **80%**。失败集中在 3 个 Job：
+
+| Job | 失败率 | 根因类型 |
+|-----|--------|----------|
+| Validate Code Quality | 100% | black 代码格式检查失败 |
+| Unit Tests | 100% | 测试断言未同步国际化改动 |
+| Integration Tests | 100% | 密码重置接口变更未同步测试 |
+| Frontend Lint | 20% | 各 PR 代码质量问题 |
+
+**结论：3 个全局性失败点导致所有 PR CI 全红，不是代码本身问题，是基础设施/测试用例滞后。**
+
+---
+
+## 二、详细根因分析
+
+### 1. Validate — black 格式检查失败
+
+**现象：**
+```
+would reformat scripts/check_migration_safety.py
+1 file would be reformatted, 369 files would be left unchanged.
+Oh no! 💥 💔 💥
+```
+
+**根因：**
+`scripts/check_migration_safety.py` 文件不符合 black 格式化规范。该文件是最近新增的迁移安全检查脚本，提交前未本地跑 black 格式化。
+
+**影响范围：** 所有 PR 及 develop 分支，全量失败。
+
+**修复方案：**
+```bash
+black scripts/check_migration_safety.py
+```
+
+---
+
+### 2. Unit Tests — 1 个用例失败
+
+**现象：**
+```
+FAILED tests/unit/test_asset_library_delete.py::TestDeleteAssetLibrary::test_delete_library_access_denied
+  AssertionError: assert 'Access denied' in '无权访问该项目'
+```
+
+**统计：** 1442 passed, 1 failed
+
+**根因：**
+项目之前做了国际化（i18n）改造，错误信息从英文改成了中文，但对应的单元测试断言仍然检查英文 "Access denied"，导致断言失败。
+
+**影响范围：** 所有 PR 及 develop 分支，全量失败。
+
+**修复方案：**
+修改 `tests/unit/test_asset_library_delete.py` 中的断言，将 `'Access denied'` 改为 `'无权访问该项目'`，或改为断言 HTTP 状态码（403）而不是错误消息文本。
+
+---
+
+### 3. Integration Tests — 1 个用例失败
+
+**现象：**
+```
+FAILED tests/integration/test_auth.py::TestPasswordReset::test_request_password_reset_success
+  assert 404 in (200, 202)
+```
+
+**统计：** 45 passed, 1 failed, 13 deselected, 2 rerun
+
+**根因：**
+密码重置请求接口（`POST /auth/password-reset/request` 或类似路由）返回 404，说明该接口已被移除、路由变更，或对应的功能模块暂时被注释/下线。
+
+**影响范围：** 所有 PR 及 develop 分支，全量失败。
+
+**修复方案：**
+- 如果接口确实下线了：删除或 skip 这个测试用例
+- 如果是路由改了：更新测试中的 API 路径
+- 如果是功能待开发：标记为 `@pytest.mark.skip` 并加上 TODO
+
+---
+
+## 三、修复优先级
+
+| 优先级 | 问题 | 修复难度 | 预估时间 |
+|--------|------|----------|----------|
+| P0 | black 格式检查失败 | ⭐ | 5分钟 |
+| P0 | 单元测试国际化断言失败 | ⭐ | 10分钟 |
+| P1 | 集成测试密码重置接口404 | ⭐⭐ | 30分钟（需确认接口状态） |
+
+**建议：** 先修前两个 P0（能让 2/3 的 job 变绿），再处理密码重置那个。
+
+---
+
+## 四、Runner 执行情况观察
+
+- 当前 9 个 Runner 全部在线（构建服务器 4 个 + 新服务器 5 个）
+- 失败的 Job 都是在构建服务器的 Runner 上执行的（xiaoxia-ci-runner-2/3 等）
+- 新服务器 5 个 Runner 目前全部空闲（标签修复后首次接任务可能需要时间）
+- 并发能力充足，瓶颈在代码/测试本身，不在 Runner 资源
@@ -0,0 +1,136 @@
+# 三台服务器 Runner 分工规划
+
+**制定日期：** 2026-07-13
+**状态：** 规划中
+
+---
+
+## 一、现状总览
+
+当前共 9 个 Gitea Actions Runner，分布在 3 台服务器上：
+
+| 服务器 | IP | 配置 | Runner 数量 | 当前状态 |
+|--------|-----|------|-------------|----------|
+| 构建服务器 | 114.55.236.178 | 4核 / 7.1G RAM / 49G NVMe | 4个（ID: 8, 42, 46, 47） | ✅ 在线 |
+| 新CI服务器 | 116.62.226.203 | 8核 / 14G RAM | 5个（ID: 58-62） | ✅ 在线 |
+| 业务服务器 | 47.98.113.167 | - | 0个（旧3个已下线） | ⚠️ 待规划 |
+
+**所有 Runner 共用标签：** `saas`, `runtime-builder`, `host`, `ubuntu-latest`
+
+---
+
+## 二、问题分析
+
+### 2.1 标签无区分
+所有 Runner 标签完全一致，CI 任务随机分配到任意 Runner，导致：
+- 构建任务（Build）可能跑到配置低的机器上，构建慢
+- 代码检查任务占着构建服务器，影响构建速度
+- 业务服务器跑 CI 影响线上服务稳定性
+
+### 2.2 资源浪费
+- 新服务器 8核14G 跑 validate/lint 有点大材小用
+- 构建服务器 4核7G 跑 Docker 构建偏紧张
+
+---
+
+## 三、规划方案
+
+### 3.1 分工原则
+
+| 服务器 | 角色 | 主要任务类型 | 标签策略 |
+|--------|------|-------------|----------|
+| **构建服务器** (114.55.236.178) | 构建专机 | Build Staging / Build Production / Docker 镜像构建 | 保留 `saas` + `host`，新增 `build-only` |
+| **新CI服务器** (116.62.226.203) | 代码检查专机 | Validate / Unit Tests / Integration Tests / Frontend Lint | 保留 `saas` + `host`，新增 `ci-check` |
+| **业务服务器** (47.98.113.167) | 部署专机 | Deploy Staging / Deploy Production / E2E Tests | 保留 `saas` + `host`，新增 `deploy-only` |
+
+### 3.2 具体配置
+
+#### 构建服务器（4个 Runner）
+- **数量：** 3个（从4个缩减，释放资源给构建缓存）
+- **标签：** `saas`, `host`, `build-only`, `ubuntu-latest`
+- **负责 Job：**
+  - `build-staging`
+  - `build-production-runtime-images`
+  - 其他需要 Docker buildx 的任务
+
+#### 新CI服务器（5个 Runner）
+- **数量：** 5个（保持不变）
+- **标签：** `saas`, `host`, `ci-check`, `ubuntu-latest`
+- **负责 Job：**
+  - `validate`
+  - `unit-tests`
+  - `integration-tests`
+  - `frontend-lint`
+  - 安全扫描（gitleaks / pip-audit / vulture 等）
+
+#### 业务服务器（1-2个 Runner）
+- **数量：** 1-2个（逐步替换旧的3个）
+- **标签：** `saas`, `host`, `deploy-only`, `ubuntu-latest`
+- **负责 Job：**
+  - `deploy-staging`
+  - `deploy-production`
+  - `staging-e2e` / `production-e2e`
+  - `staging-api-tests`
+
+---
+
+## 四、实施步骤
+
+### Phase 1: 标签打标（低风险，立即做）
+1. 新服务器 5 个 Runner 添加 `ci-check` 标签
+2. 构建服务器保留 3 个 Runner，添加 `build-only` 标签
+3. 业务服务器部署 1 个新 Runner，标签 `deploy-only`
+
+### Phase 2: Job 路由调整（中风险，逐步来）
+1. validate / unit-tests / integration-tests / frontend-lint 改为 `runs-on: ci-check`
+2. build-staging / build-production 改为 `runs-on: build-only`
+3. deploy-* / e2e 改为 `runs-on: deploy-only`
+
+### Phase 3: 旧 Runner 下线
+- 业务服务器旧的 3 个 Runner 确认无任务后下线
+- 构建服务器多余的 1 个 Runner 迁移到新服务器
+
+---
+
+## 五、并发配置优化建议
+
+### 5.1 当前并发情况
+- 首发并行 Job：validate + unit-tests + frontend-lint（3个并行）
+- integration-tests 依赖 validate（串行，浪费资源）
+- 无 concurrency 限制，同一分支多次 push 会重复跑
+
+### 5.2 优化建议
+
+**1. integration-tests 改为与 unit-tests 并行**
+```yaml
+# 当前
+integration-tests:
+  needs: validate  # 没必要等validate
+
+# 优化后
+integration-tests:
+  needs: []  # 直接和unit-tests并行跑
+```
+
+**2. 增加分支级 concurrency，取消重复构建**
+```yaml
+concurrency:
+  group: ${{ github.workflow }}-${{ github.ref }}
+  cancel-in-progress: true
+```
+同一 PR 多次 push 时，取消旧的构建，只跑最新的。
+
+**3. Build Staging 移出 PR 门禁**
+- 已在阶段二优化中完成（PR #245）
+- Build Staging 只在 develop/main 上异步构建
+
+---
+
+## 六、预期收益
+
+| 指标 | 当前 | 优化后 | 提升 |
+|------|------|--------|------|
+| PR CI 总时长 | ~8-12分钟 | ~4-6分钟 | ⏱️ 缩短 40-50% |
+| 构建速度 | 可能抢到慢机器 | 固定高配构建机 | 🚀 更稳定更快 |
+| 线上稳定性 | CI和业务抢资源 | 部署独立Runner | 🛡️ 隔离保障 |
+| Runner 利用率 | 随机分配 | 按任务类型调度 | 📈 更合理 |
--- a/Show More
+++ b/Show More