name: "MinerU PDF Parser"
description: "调用远端 MinerU FastAPI 服务解析 PDF，提取 Markdown 文本与图片。支持 Base64 内联、本地保存或上传至外部图床服务。"
language: "python"
raw_id: "MinerUPdfParser"
version: "1.0.0"
vendor: "DeepData"
modal: "text"
inputs: "text"
outputs: "text"
types:
  - 'collect'

release:
  - "首次发布：支持远程 MinerU API 调用"
  - "新增图片外部上传图床功能"
  - "优化图片处理优先级与容错机制"

runtime:
  memory: 524288000
  cpu: 0.1
  gpu: 0
  npu: 0
  storage: 50MB

metrics:
  - name: "吞吐量"
    metric: "Depends on API latency"
  - name: "准确率"
    metric: "Depends on MinerU version"

settings:
  mineruApi:
    name: "MinerU API 地址"
    description: "MinerU 服务基础 URL (如 http://mineru.ailab.ccoop.cc:18585)"
    type: "input"
    defaultVal: "http://mineru.ailab.ccoop.cc:18585"
    required: true
  server_url:
    name: "兼容服务地址"
    description: "VLM/Hybrid 后端所需的 OpenAI 兼容地址 (可选)"
    type: "input"
    defaultVal: ""
    required: false
  backend:
    name: "解析后端"
    description: "选择解析模式"
    type: "select"
    defaultVal: "pipeline"
    required: true
    options:
      - label: "本地推理 (Pipeline)"
        value: "pipeline"
      - label: "远程 VLM (HTTP Client)"
        value: "vlm-http-client"
      - label: "远程混合 (Hybrid HTTP Client)"
        value: "hybrid-http-client"
  lang_list:
    name: "语言列表"
    description: "输入语言代码，多个语言用逗号分隔 (如 ch,en)"
    type: "input"
    defaultVal: "ch"
    required: true
  parse_method:
    name: "解析方法"
    description: "PDF 解析方式"
    type: "select"
    defaultVal: "auto"
    required: true
    options:
      - label: "自动"
        value: "auto"
      - label: "OCR"
        value: "ocr"
      - label: "文本"
        value: "txt"
  formula_enable:
    name: "启用公式识别"
    description: "是否解析数学公式"
    type: "switch"
    defaultVal: "true"
    checkedLabel: "是"
    unCheckedLabel: "否"
  table_enable:
    name: "启用表格解析"
    description: "是否解析文档中的表格"
    type: "switch"
    defaultVal: "true"
    checkedLabel: "是"
    unCheckedLabel: "否"
  return_images:
    name: "返回图片数据"
    description: "是否将提取的图片数据包含在响应中（关闭可节省带宽）"
    type: "switch"
    defaultVal: "true"
    checkedLabel: "是"
    unCheckedLabel: "否"
  timeout:
    name: "请求超时 (秒)"
    description: "HTTP 请求最大等待时间"
    type: "slider"
    defaultVal: 900
    min: 60
    max: 3600
    step: 60
  embed_images_base64:
    name: "嵌入 Base64 图片"
    description: "是否将提取的图片以 Base64 格式内联到 Markdown 文本中"
    type: "switch"
    defaultVal: "true"
    checkedLabel: "是"
    unCheckedLabel: "否"
    
  # ================= 新增：外部图床上传配置 =================
  upload_images:
    name: "上传图片到外部服务"
    description: "开启后，将自动把提取的图片上传至下方指定的 API，并替换为可访问的网络链接"
    type: "switch"
    defaultVal: "true"
    checkedLabel: "开启上传"
    unCheckedLabel: "使用 Base64/本地"
  upload_api_url:
    name: "图片上传 API 地址"
    description: "接收图片的 POST 接口 (支持 multipart/form-data)，例如: http://example.com/upload/form"
    type: "input"
    defaultVal: "http://mfs.ailab.ccoop.cc:18585/upload/form"
    required: false
  preview_base_url:
    name: "图片预览基础地址"
    description: "用于拼接上传返回的相对路径，生成最终 URL。例如: http://example.com"
    type: "input"
    defaultVal: "http://mfs.ailab.ccoop.cc:18585/"
    required: false