法律文本格式化助手

YAML 源码

app:
  description: 将法律文本转换为规范的 Markdown 格式
  icon: 🤖
  icon_background: '#FFEAD5'
  mode: workflow
  name: 法律文本格式化助手
  use_icon_as_answer_icon: false
kind: app
version: 0.1.5
workflow:
  conversation_variables: []
  environment_variables: []
  features:
    file_upload:
      allowed_file_extensions: []
      allowed_file_types: []
      allowed_file_upload_methods: []
      enabled: false
      fileUploadConfig:
        audio_file_size_limit: 50
        batch_count_limit: 5
        file_size_limit: 15
        image_file_size_limit: 10
        video_file_size_limit: 100
        workflow_file_upload_limit: 10
      image:
        enabled: false
        number_limits: 3
        transfer_methods: []
      number_limits: 3
    opening_statement: ''
    retriever_resource:
      enabled: false
    sensitive_word_avoidance:
      enabled: false
    speech_to_text:
      enabled: false
    suggested_questions: []
    suggested_questions_after_answer:
      enabled: false
    text_to_speech:
      enabled: false
      language: ''
      voice: ''
  graph:
    edges:
    - data:
        isInIteration: false
        sourceType: start
        targetType: llm
      id: 1735626727052-source-1735632132331-target
      source: '1735626727052'
      sourceHandle: source
      target: '1735632132331'
      targetHandle: target
      type: custom
      zIndex: 0
    - data:
        isInIteration: false
        sourceType: llm
        targetType: code
      id: 1735632132331-source-1735629665760-target
      source: '1735632132331'
      sourceHandle: source
      target: '1735629665760'
      targetHandle: target
      type: custom
      zIndex: 0
    - data:
        isInIteration: true
        iteration_id: '1735632653899'
        sourceType: iteration-start
        targetType: llm
      id: 1735632653899start-source-1735632981287-target
      source: 1735632653899start
      sourceHandle: source
      target: '1735632981287'
      targetHandle: target
      type: custom
      zIndex: 1002
    - data:
        isInIteration: false
        sourceType: code
        targetType: end
      id: 1735633258714-source-1735628174959-target
      source: '1735633258714'
      sourceHandle: source
      target: '1735628174959'
      targetHandle: target
      type: custom
      zIndex: 0
    - data:
        isInIteration: false
        sourceType: code
        targetType: iteration
      id: 1735629665760-source-1735632653899-target
      source: '1735629665760'
      sourceHandle: source
      target: '1735632653899'
      targetHandle: target
      type: custom
      zIndex: 0
    - data:
        isInIteration: false
        sourceType: iteration
        targetType: code
      id: 1735632653899-source-1735633258714-target
      source: '1735632653899'
      sourceHandle: source
      target: '1735633258714'
      targetHandle: target
      type: custom
      zIndex: 0
    - data:
        isInIteration: false
        sourceType: code
        targetType: tool
      id: 1735633258714-source-1735802619533-target
      source: '1735633258714'
      sourceHandle: source
      target: '1735802619533'
      targetHandle: target
      type: custom
      zIndex: 0
    nodes:
    - data:
        desc: ''
        selected: false
        title: 开始
        type: start
        variables:
        - label: legal_text
          max_length: 80000
          options: []
          required: true
          type: paragraph
          variable: legal_text
      height: 88
      id: '1735626727052'
      position:
        x: 27.937620977678364
        y: 292
      positionAbsolute:
        x: 27.937620977678364
        y: 292
      selected: false
      sourcePosition: right
      targetPosition: left
      type: custom
      width: 243
    - data:
        desc: ''
        outputs:
        - value_selector:
          - '1735633258714'
          - final_text
          variable: text
        selected: false
        title: 结束
        type: end
      height: 88
      id: '1735628174959'
      position:
        x: 1198.1485031082693
        y: 292
      positionAbsolute:
        x: 1198.1485031082693
        y: 292
      selected: false
      sourcePosition: right
      targetPosition: left
      type: custom
      width: 243
    - data:
        code: "import json\nimport re\n\ndef main(text, analysis_result):\n    # 处理\
          \ analysis_result，移除可能的 markdown 标记\n    analysis_text = analysis_result.strip()\n\
          \    if analysis_text.startswith('```json'):\n        analysis_text = analysis_text[7:]\n\
          \    if analysis_text.endswith('```'):\n        analysis_text = analysis_text[:-3]\n\
          \    \n    # 使用 json.loads 而不是 eval\n    analysis = json.loads(analysis_text.strip())\n\
          \    text_type = analysis.get('text_type')\n    structure = analysis.get('structure',\
          \ {})\n    target_length = 2000  # 保持2000字符\n    max_length = target_length\
          \ * 1.1  # 允许超出10%\n    \n    def get_split_point_checker(text_type, structure):\n\
          \        if text_type == 'law_text':\n            def is_split_point(line,\
          \ current_length):\n                # 严格按照章节分割\n                return line.strip().startswith('第')\
          \ and ('章' in line or '编' in line)\n        else:\n            def is_split_point(line,\
          \ current_length):\n                # 严格按照案例分割\n                return (line.strip().startswith('案例')\
          \ or\n                       line.strip().endswith('案') or\n           \
          \            line.strip().endswith('判决书') or\n                       line.strip().endswith('裁定书'))\n\
          \                \n        return is_split_point\n    \n    is_split_point\
          \ = get_split_point_checker(text_type, structure)\n    \n    # 预处理：移除多余的空行，保留单个空行\n\
          \    raw_lines = text.split('\\n')\n    lines = []\n    prev_line_empty\
          \ = True\n    \n    for line in raw_lines:\n        line = line.rstrip()\n\
          \        is_empty = not line.strip()\n        \n        # 保留有意义的空行，但避免连续空行\n\
          \        if not is_empty or not prev_line_empty:\n            lines.append(line)\n\
          \        prev_line_empty = is_empty\n    \n    raw_segments = []  # 改名为\
          \ raw_segments\n    current_segment = []\n    current_length = 0\n    \n\
          \    for i, line in enumerate(lines):\n        line_length = len(line.strip())\n\
          \        \n        # 如果当前段落为空，直接添加\n        if not current_segment:\n  \
          \          current_segment.append(line)\n            current_length = line_length\n\
          \            continue\n        \n        # 检查是否是分割点\n        if is_split_point(line,\
          \ current_length):\n            # 在分割点处分割\n            if current_segment:\n\
          \                raw_segments.append('\\n'.join(current_segment))\n    \
          \        current_segment = [line]\n            current_length = line_length\n\
          \            continue\n        \n        # 如果超过最大长度，强制分割\n        if current_length\
          \ + line_length > max_length:\n            raw_segments.append('\\n'.join(current_segment))\n\
          \            current_segment = [line]\n            current_length = line_length\n\
          \        else:\n            current_segment.append(line)\n            current_length\
          \ += line_length\n    \n    # 添加最后一个段落\n    if current_segment:\n      \
          \  raw_segments.append('\\n'.join(current_segment))\n    \n    # 确保至少返回一个段落\n\
          \    if not raw_segments:\n        raw_segments = [text]\n    \n    # 格式化为字符串数组\n\
          \    segments = []\n    for segment in raw_segments:\n        if segment.strip():\
          \  # 确保段落不为空\n            segments.append(segment.strip())\n    \n    #\
          \ 如果没有有效段落，返回包含整个文本的单个段落\n    if not segments:\n        segments = [text.strip()]\n\
          \    \n    # 返回包含 segments 数组的字典\n    return {\n        \"segments\": segments\n\
          \    }"
        code_language: python3
        desc: '代码

          用于文本的分割，适应大模型一次性输出长度

          同时需要输出数组提供给迭代模块用于并行处理'
        outputs:
          segments:
            children: null
            type: array[string]
        selected: false
        title: 文本分割
        type: code
        variables:
        - value_selector:
          - '1735626727052'
          - legal_text
          variable: text
        - value_selector:
          - '1735632132331'
          - text
          variable: analysis_result
      height: 144
      id: '1735629665760'
      position:
        x: 730.6999387678718
        y: 58.97775019592794
      positionAbsolute:
        x: 730.6999387678718
        y: 58.97775019592794
      selected: false
      sourcePosition: right
      targetPosition: left
      type: custom
      width: 243
    - data:
        context:
          enabled: true
          variable_selector:
          - '1735626727052'
          - legal_text
        desc: 用LLM分析文本是法条还是案例，返回json
        model:
          completion_params:
            temperature: 0.3
          mode: chat
          name: deepseek-chat
          provider: deepseek
        prompt_template:
        - edition_type: basic
          id: 8a14522d-d436-4e39-96cc-770cd9daee9e
          role: system
          text: "你是一个专业的法律文本分析助手。请分析输入的文本，判断其是法律条文还是法律案例，并分析其结构特征。\n\n请直接返回JSON格式（不要包含任何markdown标记），必须包含以下字段：\n\
            {\n  \"text_type\": \"law_text/case_text\",  // 文本类型：law_text为法律条文，case_text为法律案例\n\
            \  \"confidence\": number,                // 判断的置信度（0-1）\n  \"reason\"\
            : \"string\",                 // 判断依据\n  \"structure\": {            \
            \          // 结构特征\n    \"has_chapters\": boolean,          // 是否有章节划分\n\
            \    \"has_articles\": boolean,          // 是否有条款编号\n    \"has_case_number\"\
            : boolean,       // 是否有案号\n    \"has_court_parts\": boolean,       //\
            \ 是否有法院审理部分\n    \"main_sections\": [\"string\"]       // 主要段落类型\n  },\n\
            \  \"split_strategy\": ...（过长已截断）