Byted Las Video Inpaint

{
  "skill_name": "byted-las-video-inpaint",
  "evals": [
    {
      "id": 1,
      "prompt": "我有一个带水印的视频（tos://bucket/input/watermark.mp4），帮我用 lasutil 提交修复任务并告诉我怎么轮询拿到输出的 inpainted_video_path。输出要包含完整命令和关键字段说明。",
      "expected_output": "给出 data.json 模板、lasutil submit/poll 命令，解释 COMPLETED 后从 data.inpainted_video_path 获取结果。",
      "files": []
    },
    {
      "id": 2,
      "prompt": "我有一个 TOS 目录 tos://bucket/input/ 下面很多 mp4，需要批量去水印。请给出可分批的 batch-submit/batch-poll 流程（含 workdir/batch_001 目录组织），并提醒我用 -h 查看参数含义。",
      "expected_output": "给出 tos-ls 生成总清单、按批次子目录切分、batch-submit/batch-poll 示例命令，说明 jobs/interval/progress-interval/max-seconds 的含义与用法。",
      "files": []
    },
    {
      "id": 3,
      "prompt": "我只有本地文件 ./input.mp4，想要最终输出落到本地 ./output.mp4，但算子只能写 TOS。请给出上传、提交、轮询、下载的完整步骤。",
      "expected_output": "使用 lasutil tos-upload 上传到 TOS，中转 output_tos_path，完成后用 lasutil tos-download 下载到本地。",
      "files": []
    }
  ]
}

# Copyright (c) 2025 Beijing Volcano Engine Technology Co., Ltd. and/or its affiliates.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

#!/bin/bash
# ==============================================================================
# 视频格式预检查
# Usage: scripts/check_format.sh <file_path>
# ==============================================================================

FILE_PATH="$1"

if [ -z "$FILE_PATH" ]; then
  echo "❌ 错误: 请提供文件路径"
  exit 1
fi

# 获取文件扩展名并转小写
EXT=$(echo "$FILE_PATH" | awk -F. '{print tolower($NF)}')

# 允许的视频格式
ALLOWED_FORMATS="mp4 wmv avi mkv mov flv webm mpg mpeg"

if [[ " $ALLOWED_FORMATS " =~ " $EXT " ]]; then
  echo "✅ 格式检查通过: $EXT"
  exit 0
else
  echo "⚠️  警告: 文件扩展名 '$EXT' 不在推荐格式列表中"
  echo "   推荐格式: $ALLOWED_FORMATS"
  echo "   请确认输入格式正确"
  exit 1
fi

#!/bin/bash

# ==============================================================================
# LAS SDK & Skill 环境初始化与更新脚本
# 用法: source scripts/env_init.sh <operator_id>
# ==============================================================================

OPERATOR_ID=$1

if [ -z "$OPERATOR_ID" ]; then
  echo "❌ 错误: 必须提供 operator_id"
  return 1 2>/dev/null || exit 1
fi

# 1. 虚拟环境隔离
# Find project root dynamically, then check for .las_venv
# skill structure: <project_root>/skills/<skill>/scripts/env_init.sh
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
PROJECT_ROOT="$(cd "${SCRIPT_DIR}/../../.." && pwd)"

# 优先复用项目根目录的虚拟环境，不存在才创建新的
if [ -d "${PROJECT_ROOT}/.las_venv" ]; then
  source "${PROJECT_ROOT}/.las_venv/bin/activate"
elif [ ! -d ".las_venv" ]; then
  python3 -m venv .las_venv
  source .las_venv/bin/activate
else
  source .las_venv/bin/activate
fi

# 2. 获取远程 manifest
manifest_url="https://las-ai-cn-beijing-online.tos-cn-beijing.volces.com/operator_cards_serving/public/skills/sdk/manifest.json"
manifest=$(curl -sf "$manifest_url" || echo '{}')

# 3. 比较 SDK 版本并自动更新
local_ver=$(lasutil --version 2>/dev/null | grep -oE '[0-9]+\.[0-9]+\.[0-9]+' || echo "0.0.0")
remote_ver=$(echo "$manifest" | jq -r '.sdk_version // "unknown"')

if [ "$local_ver" = "0.0.0" ] || [ "$local_ver" != "$remote_ver" ]; then
  echo "📦 SDK 更新: $local_ver → $remote_ver，正在安装..."
  pip install --quiet --upgrade https://las-ai-cn-beijing-online.tos-cn-beijing.volces.com/operator_cards_serving/public/skills/sdk/las_sdk-0.2.0-py3-none-any.whl
fi

# 4. 检查当前算子的参数变更
op_changes=$(echo "$manifest" | jq -r ".operators.${OPERATOR_ID}.changes_since // {} | to_entries[] | \"  自 \(.key) 起: \(.value[])\"" 2>/dev/null)
if [ -n "$op_changes" ]; then
  echo "⚠️ 算子 ${OPERATOR_ID} 有参数更新:"
  echo "$op_changes"
  echo "建议使用 find-skills 检索并更新此 skill 到最新版本。"
fi

# 5. 初始化工作目录
export LAS_WORKDIR=$(mktemp -d /tmp/las_work_XXXXXX)
echo "✅ 环境初始化完成，工作目录: $LAS_WORKDIR"

#!/bin/bash
# ==============================================================================
# 生成结果展示 markdown 模板，自动包含计费声明
# Usage: scripts/generate_result.md.sh <task_id> <output_dir> <estimated_price>
# ==============================================================================

TASK_ID="$1"
OUTPUT_DIR="$2"
ESTIMATED_PRICE="$3"

if [ -z "$TASK_ID" ] || [ -z "$OUTPUT_DIR" ]; then
  echo "❌ 错误: 参数不全"
  exit 1
fi

STATUS=$(cat "${OUTPUT_DIR}/COMPLETED" 2>/dev/null || echo "UNKNOWN")

cat <<EOF
# ASR 识别结果

## 任务信息
- **任务 ID**: \`${TASK_ID}\`
- **最终状态**: \`${STATUS}\`
EOF

if [ -n "$ESTIMATED_PRICE" ]; then
cat <<EOF
- **预估计费**: ${ESTIMATED_PRICE} 元
EOF
fi

if [ -f "${OUTPUT_DIR}/transcript.txt" ]; then
  TEXT_LEN=$(wc -c < "${OUTPUT_DIR}/transcript.txt")
  DURATION_MS=$(cat "${OUTPUT_DIR}/result.json" | jq -r '.data.audio_info.duration // 0')
  DURATION_MIN=$(echo "scale=1; $DURATION_MS / 1000 / 60" | bc)
  LANG=$(cat "${OUTPUT_DIR}/result.json" | jq -r '.data.result.additions.lid_lang // "unknown"')

cat <<EOF

## 识别统计
- **音频时长**: ${DURATION_MIN} 分钟
- **识别语种**: ${LANG}
- **总字数**: ${TEXT_LEN} 字符

## 文本预览（前 500 字）

\`\`\`
$(head -c 500 "${OUTPUT_DIR}/transcript.txt")
\`\`\`
EOF
fi

cat <<EOF

## 结果文件位置
\`${OUTPUT_DIR}/\`
- \`result.json\` - 完整 API 响应
- \`transcript.txt\` - 完整识别文本
EOF

if [ -f "${OUTPUT_DIR}/utterances.csv" ]; then
cat <<EOF
- \`utterances.csv\` - 分句说话人信息（CSV格式）
- \`utterances.json\` - 分句原始数据
EOF
fi

cat <<'EOF'

---

本方式的计费均为预估计费，与实际费用有差距，实际费用以运行后火山产生的账单为准。计费说明请参考 [Volcengine LAS 定价](https://www.volcengine.com/docs/6492/1544808)。
EOF

#!/bin/bash
# ==============================================================================
# 后台轮询脚本 - 动态间隔
# Usage: scripts/poll_background.sh <task_id> [output_dir]
# ==============================================================================

TASK_ID="$1"
OUTPUT_DIR="${2:-output/${TASK_ID}}"

if [ -z "$TASK_ID" ]; then
  echo "❌ 错误: 请提供 task_id"
  exit 1
fi

# 加载环境
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
PROJECT_ROOT="$(cd "${SCRIPT_DIR}/../../.." && pwd)"

if [ -f "${PROJECT_ROOT}/.env" ]; then
  source "${PROJECT_ROOT}/.env"
  export LAS_API_KEY LAS_REGION
fi

# 激活虚拟环境
if [ -d "${PROJECT_ROOT}/.las_venv" ]; then
  source "${PROJECT_ROOT}/.las_venv/bin/activate"
fi

OPERATOR_ID="las_video_inpaint"
mkdir -p "${OUTPUT_DIR}"

echo "$(date): Starting background polling for task ${TASK_ID}" >> "${OUTPUT_DIR}/poll.log"

# 轮询计数，用于动态调整间隔
ATTEMPT=0

# Poll until terminal state
while true; do
  ((ATTEMPT++))

  # 动态调整轮询间隔:
  # - 前 5 次: 60秒
  # - 5-10 次: 120秒
  # - 10次之后: 300秒（长视频理解时间更长）
  if [ $ATTEMPT -le 5 ]; then
    SLEEP=60
  elif [ $ATTEMPT -le 10 ]; then
    SLEEP=120
  else
    SLEEP=300
  fi

  # Save full response first, then extract status
  FULL_OUTPUT=$(lasutil poll "${OPERATOR_ID}" "${TASK_ID}")
  echo "$FULL_OUTPUT" > "${OUTPUT_DIR}/last_poll.json"

  # Extract status
  STATUS=$(echo "$FULL_OUTPUT" | python3 -c "
import json
import sys
try:
  data = json.load(sys.stdin)
  print(data['metadata']['task_status'])
except Exception as e:
  print('ERROR')
")

  echo "$(date): Attempt $ATTEMPT, status = ${STATUS}, next poll in ${SLEEP}s" >> "${OUTPUT_DIR}/poll.log"

  if [ "$STATUS" = "COMPLETED" ] || [ "$STATUS" = "FAILED" ] || [ "$STATUS" = "TIMEOUT" ] || [ "$STATUS" = "CANCELED" ]; then
    # Save final result
    echo "$FULL_OUTPUT" > "${OUTPUT_DIR}/result.json"

    # Create completion flag
    echo "$STATUS" > "${OUTPUT_DIR}/COMPLETED"
    echo "$(date): Poll completed, final status = ${STATUS}" >> "${OUTPUT_DIR}/poll.log"
    exit 0
  fi

  sleep "$SLEEP"
done