kg_agent/knowledge_v2/visualization/workflow_visualization.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
'''
知识获取工作流可视化
1. 读取知识获取工作流详细过程数据文件（参考 .cache/9f510b2a8348/execution_record.json）
2. 文件路径可设置，每个文件表示一个输入信息执行过程，可能有多个。多个文件路径硬编码在代码中，用list表示
3. 将知识获取工作流用html页面展示出来
4. HTML 文件输出路径在当前目录下，文件名称为 workflow_visualization_datetime.html
'''

import json
import os
from datetime import datetime
from pathlib import Path


# 硬编码的文件路径列表（相对于项目根目录）
DATA_FILE_PATHS = [
    "../.cache/557e5ce27c1c/execution_record.json",
    "../.cache/d0608df43f77/execution_record.json",
    # 可以在这里添加更多文件路径
]


def load_data_files(file_paths):
    """读取并解析JSON文件，返回数据和文件夹名称的列表"""
    data_list = []
    script_dir = Path(__file__).parent
    for file_path in file_paths:
        # 将相对路径转换为绝对路径
        abs_path = (script_dir / file_path).resolve()
        if not abs_path.exists():
            print(f"警告: 文件不存在: {abs_path}")
            continue
        try:
            with open(abs_path, 'r', encoding='utf-8') as f:
                data = json.load(f)
                # 从文件路径中提取文件夹名称（如 a588c7a380ee）
                # 路径格式：../.cache/a588c7a380ee/execution_record.json
                folder_name = abs_path.parent.name
                data_list.append((data, folder_name))
        except (json.JSONDecodeError, IOError) as e:
            print(f"错误: 读取文件失败 {abs_path}: {e}")
    return data_list


def parse_workflow_data(data, folder_name=''):
    """解析工作流数据，提取关键信息"""
    workflow = {
        'input': {},
        'steps': [],
        'output': {},
        'folder_name': folder_name
    }
    
    # 提取输入信息（字符串格式）
    if 'input' in data:
        input_data = data['input']
        if isinstance(input_data, str):
            # 直接保存字符串
            workflow['input'] = {'text': input_data}
        else:
            # 兼容旧格式：转换为字符串
            question = input_data.get('question', '')
            post_info = input_data.get('post_info', '')
            persona_info = input_data.get('persona_info', '')
            input_text = f'问题：{question}\n' if question else ''
            if post_info:
                input_text += f'{post_info}\n'
            if persona_info:
                input_text += f'账号人设信息：{persona_info}\n'
            workflow['input'] = {'text': input_text.strip()}
    
    # 提取执行流程（新格式：execution 直接包含各步骤，不再有 modules.function_knowledge）
    if 'execution' in data:
        execution = data['execution']
        
        # 步骤1: 生成query（暂时隐藏，不添加到步骤列表）
        # if 'generate_query' in execution:
        #     generate_query = execution['generate_query']
        #     workflow['steps'].append({
        #         'step': 'generate_query',
        #         'name': '生成查询',
        #         'query': generate_query.get('query', '') or generate_query.get('response', ''),
        #         'prompt': generate_query.get('prompt', '')
        #     })
        
        # 步骤1: 选择工具（原步骤2，现在变成步骤1）
        if 'select_tool' in execution:
            select_tool = execution['select_tool']
            response = select_tool.get('response', {})
            workflow['steps'].append({
                'step': 'select_tool',
                'name': '选择工具',
                'prompt': select_tool.get('prompt', ''),
                'tool_name': response.get('工具名', '') if isinstance(response, dict) else '',
                'tool_id': response.get('工具调用ID', '') if isinstance(response, dict) else '',
                'tool_usage': response.get('使用方法', '') if isinstance(response, dict) else '',
                'match_reason': response.get('匹配理由', '') if isinstance(response, dict) else '',
                'application_scenario': response.get('应用场景', '') if isinstance(response, dict) else ''
            })
        
        # 判断是否选择了工具（如果 response 为空字典或没有工具信息，或工具名为"无工具匹配"，则没有选择到工具）
        has_tool = False
        if 'select_tool' in execution:
            select_tool = execution['select_tool']
            response = select_tool.get('response', {})
            if isinstance(response, dict):
                tool_name = response.get('工具名', '')
                # 检查工具名是否存在且不是"无工具匹配"
                if tool_name and tool_name != '无工具匹配':
                    has_tool = True
        
        # 如果选择了工具，执行工具调用流程
        if has_tool:
            # 步骤3: 提取参数
            if 'extract_params' in execution:
                extract_params = execution['extract_params']
                workflow['steps'].append({
                    'step': 'extract_params',
                    'name': '提取参数',
                    'prompt': extract_params.get('prompt', ''),
                    'params': extract_params.get('params', {})
                })
            
            # 步骤4: 执行工具（新格式：tool_call 替代 execute_tool）
            if 'tool_call' in execution:
                tool_call = execution['tool_call']
                # 优先使用 result，如果没有则使用 response
                result = tool_call.get('result', '')
                if not result and tool_call.get('response'):
                    # 如果 response 是字典，尝试提取其中的 result
                    response = tool_call.get('response', {})
                    if isinstance(response, dict):
                        result = response.get('result', response)
                    else:
                        result = response
                workflow['steps'].append({
                    'step': 'execute_tool',
                    'name': '执行工具',
                    'response': result or tool_call.get('response', '')
                })
            
            # 步骤5: 工具结果评估
            if 'evaluate_tool_result' in execution:
                evaluate_tool_result = execution['evaluate_tool_result']
                eval_result = evaluate_tool_result.get('eval_result', {})
                workflow['steps'].append({
                    'step': 'evaluate_tool_result',
                    'name': '工具结果评估',
                    'prompt': evaluate_tool_result.get('prompt', ''),
                    'can_answer': eval_result.get('是否可以回答', ''),
                    'reason': eval_result.get('理由', '')
                })
        
        # 如果没有选择到工具，进行知识搜索流程
        else:
            if 'knowledge_search' in execution:
                knowledge_search = execution['knowledge_search']
                
                # 步骤3: LLM搜索（大模型+search 渠道的搜索过程）
                if 'llm_search' in knowledge_search:
                    llm_search = knowledge_search['llm_search']
                    # 处理 generated_queries（生成查询）
                    generated_queries = llm_search.get('generated_queries', {})
                    queries = generated_queries.get('queries', [])
                    search_results = llm_search.get('search_results', [])
                    # 处理 merge（合并结果）
                    merge = llm_search.get('merge', {})
                    workflow['steps'].append({
                        'step': 'llm_search',
                        'name': 'LLM搜索',
                        'prompt': generated_queries.get('prompt', ''),
                        'queries': queries,
                        'search_results': search_results,
                        'merge_prompt': merge.get('prompt', ''),
                        'merge_response': merge.get('response', ''),
                        'sources_count': merge.get('sources_count', 0)
                    })
                
                # 步骤4: 多渠道搜索结果整合
                if 'multi_search_merge' in knowledge_search:
                    multi_search_merge = knowledge_search['multi_search_merge']
                    workflow['steps'].append({
                        'step': 'multi_search_merge',
                        'name': '多渠道搜索结果整合',
                        'prompt': multi_search_merge.get('prompt', ''),
                        'response': multi_search_merge.get('response', ''),
                        'sources_count': multi_search_merge.get('sources_count', 0),
                        'valid_sources_count': multi_search_merge.get('valid_sources_count', 0)
                    })
                
                # 步骤5: 发现新工具
                if 'extra_tools' in knowledge_search:
                    extra_tools = knowledge_search['extra_tools']
                    match_tool_response = extra_tools.get('match_tool_response', {})
                    selected_tools = match_tool_response.get('selected_tools', [])
                    workflow['steps'].append({
                        'step': 'extra_tools',
                        'name': '发现新工具',
                        'prompt': extra_tools.get('match_tool_prompt', ''),
                        'selected_tools': selected_tools,
                        'analysis_summary': match_tool_response.get('analysis_summary', '')
                    })
    
    # 提取输出信息
    if 'output' in data:
        workflow['output'] = {
            'result': data['output'].get('result', '')
        }
    
    return workflow


def escape_html(text):
    """转义HTML特殊字符"""
    if not isinstance(text, str):
        text = str(text)
    return (text.replace('&', '&')
                .replace('<', '&lt;')
                .replace('>', '&gt;')
                .replace('"', '&quot;')
                .replace("'", '&#39;'))


def format_json_for_display(obj):
    """格式化JSON对象用于显示"""
    if isinstance(obj, dict):
        return json.dumps(obj, ensure_ascii=False, indent=2)
    elif isinstance(obj, str):
        try:
            parsed = json.loads(obj)
            return json.dumps(parsed, ensure_ascii=False, indent=2)
        except (json.JSONDecodeError, ValueError):
            return obj
    return str(obj)


def generate_html(workflows):
    """生成HTML页面"""
    html = '''<!DOCTYPE html>
<html lang="zh-CN">
<head>
    <meta charset="UTF-8">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <title>知识获取工作流可视化</title>
    <style>
        * {
            margin: 0;
            padding: 0;
            box-sizing: border-box;
        }
        
        body {
            font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', 'PingFang SC', 'Hiragino Sans GB', 'Microsoft YaHei', sans-serif;
            background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
            color: #333;
            line-height: 1.6;
            min-height: 100vh;
        }
        
        .container {
            max-width: 1400px;
            margin: 0 auto;
            padding: 30px 20px;
        }
        
        h1 {
            text-align: center;
            color: white;
            margin-bottom: 40px;
            font-size: 32px;
            font-weight: 600;
            text-shadow: 0 2px 10px rgba(0,0,0,0.2);
            letter-spacing: 1px;
        }
        
        .tabs {
            display: flex;
            background: white;
            border-radius: 12px 12px 0 0;
            box-shadow: 0 4px 20px rgba(0,0,0,0.15);
            overflow-x: auto;
            padding: 5px;
        }
        
        .tab {
            padding: 16px 28px;
            cursor: pointer;
            border: none;
            background: transparent;
            color: #666;
            font-size: 14px;
            transition: all 0.3s cubic-bezier(0.4, 0, 0.2, 1);
            white-space: nowrap;
            border-radius: 8px;
            margin: 0 4px;
            position: relative;
            font-weight: 500;
        }
        
        .tab:hover {
            background: #f0f0f0;
            color: #333;
        }
        
        .tab.active {
            background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
            color: white;
            box-shadow: 0 4px 12px rgba(102, 126, 234, 0.4);
        }
        
        .tab-content {
            display: none;
            background: white;
            padding: 40px;
            border-radius: 0 0 12px 12px;
            box-shadow: 0 4px 20px rgba(0,0,0,0.15);
            margin-bottom: 20px;
            animation: fadeIn 0.3s ease-in;
        }
        
        @keyframes fadeIn {
            from { opacity: 0; transform: translateY(10px); }
            to { opacity: 1; transform: translateY(0); }
        }
        
        .tab-content.active {
            display: block;
        }
        
        .input-section {
            background: linear-gradient(135deg, #f5f7fa 0%, #c3cfe2 100%);
            padding: 28px;
            border-radius: 12px;
            margin-bottom: 35px;
            box-shadow: 0 4px 15px rgba(0,0,0,0.1);
            border: 1px solid rgba(255,255,255,0.5);
        }
        
        .input-section h3 {
            color: #2c3e50;
            margin-bottom: 20px;
            font-size: 20px;
            font-weight: 600;
            display: flex;
            align-items: center;
            gap: 10px;
        }
        
        .input-section h3::before {
            content: '📋';
            font-size: 24px;
        }
        
        .input-item {
            margin-bottom: 16px;
            padding: 12px;
            background: rgba(255,255,255,0.7);
            border-radius: 8px;
            transition: all 0.3s;
        }
        
        .input-item:hover {
            background: rgba(255,255,255,0.9);
            transform: translateX(5px);
        }
        
        .input-item strong {
            color: #495057;
            display: inline-block;
            width: 110px;
            font-weight: 600;
        }
        
        .input-item .placeholder {
            color: #999;
            font-style: italic;
        }
        
        .workflow {
            position: relative;
        }
        
        .workflow-step {
            background: white;
            border: 2px solid #e0e0e0;
            border-radius: 12px;
            margin-bottom: 25px;
            transition: all 0.3s cubic-bezier(0.4, 0, 0.2, 1);
            overflow: hidden;
            box-shadow: 0 2px 8px rgba(0,0,0,0.08);
        }
        
        .workflow-step.active {
            border-color: #667eea;
            box-shadow: 0 8px 24px rgba(102, 126, 234, 0.25);
            transform: translateY(-2px);
        }
        
        .step-header {
            padding: 20px 24px;
            background: linear-gradient(135deg, #f8f9fa 0%, #e9ecef 100%);
            cursor: pointer;
            display: flex;
            justify-content: space-between;
            align-items: center;
            user-select: none;
            transition: all 0.3s;
        }
        
        .step-header:hover {
            background: linear-gradient(135deg, #e9ecef 0%, #dee2e6 100%);
        }
        
        .workflow-step.active .step-header {
            background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
            color: white;
        }
        
        .workflow-step.active .step-name {
            color: white;
        }
        
        .workflow-step.active .step-toggle {
            color: white;
        }
        
        .step-title {
            display: flex;
            align-items: center;
            gap: 15px;
        }
        
        .step-number {
            display: inline-flex;
            align-items: center;
            justify-content: center;
            width: 36px;
            height: 36px;
            background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
            color: white;
            border-radius: 50%;
            font-size: 16px;
            font-weight: bold;
            box-shadow: 0 4px 12px rgba(102, 126, 234, 0.3);
        }
        
        .workflow-step.active .step-number {
            background: white;
            color: #667eea;
            box-shadow: 0 4px 12px rgba(255,255,255,0.3);
        }
        
        .step-name {
            font-size: 18px;
            font-weight: 600;
            color: #2c3e50;
        }
        
        .step-toggle {
            color: #6c757d;
            font-size: 20px;
            transition: transform 0.3s cubic-bezier(0.4, 0, 0.2, 1);
        }
        
        .step-toggle.expanded {
            transform: rotate(180deg);
        }
        
        .step-content {
            padding: 0 20px;
            max-height: 0;
            overflow: hidden;
            transition: max-height 0.3s ease-out, padding 0.3s;
        }
        
        .step-content.expanded {
            max-height: 5000px;
            padding: 20px;
        }
        
        .step-detail {
            margin-bottom: 20px;
        }
        
        .step-detail-label {
            font-weight: 600;
            color: #495057;
            margin-bottom: 10px;
            display: block;
            font-size: 14px;
            text-transform: uppercase;
            letter-spacing: 0.5px;
        }
        
        .step-detail-content {
            background: #f8f9fa;
            padding: 16px;
            border-radius: 8px;
            border-left: 4px solid #667eea;
            font-size: 14px;
            line-height: 1.8;
            white-space: pre-wrap;
            word-wrap: break-word;
            max-height: 400px;
            overflow-y: auto;
            box-shadow: 0 2px 8px rgba(0,0,0,0.05);
        }
        
        .json-content {
            font-family: 'SF Mono', 'Monaco', 'Courier New', monospace;
            background: #1e1e1e;
            color: #d4d4d4;
            padding: 20px;
            border-radius: 8px;
            overflow-x: auto;
            border-left: 4px solid #667eea;
            box-shadow: 0 4px 12px rgba(0,0,0,0.15);
        }
        
        .prompt-toggle-btn {
            background: linear-gradient(135deg, #667eea 0%, #764ba2 100%);
            color: white;
            border: none;
            padding: 10px 20px;
            border-radius: 6px;
            cursor: pointer;
            font-size: 13px;
            font-weight: 500;
            margin-top: 15px;
            transition: all 0.3s;
            box-shadow: 0 2px 8px rgba(102, 126, 234, 0.3);
        }
        
        .prompt-toggle-btn:hover {
            transform: translateY(-2px);
            box-shadow: 0 4px 12px rgba(102, 126, 234, 0.4);
        }
        
        .prompt-content {
            display: none;
            margin-top: 15px;
            padding: 16px;
            background: #fff3cd;
            border-radius: 8px;
            border-left: 4px solid #ffc107;
            font-size: 13px;
            line-height: 1.8;
            white-space: pre-wrap;
            word-wrap: break-word;
            max-height: 500px;
            overflow-y: auto;
        }
        
        .prompt-content.show {
            display: block;
            animation: slideDown 0.3s ease-out;
        }
        
        @keyframes slideDown {
            from {
                opacity: 0;
                max-height: 0;
            }
            to {
                opacity: 1;
                max-height: 500px;
            }
        }
        
        .output-section {
            background: linear-gradient(135deg, #e0f2fe 0%, #bae6fd 100%);
            padding: 28px;
            border-radius: 12px;
            margin-top: 35px;
            border-left: 4px solid #0ea5e9;
            box-shadow: 0 4px 15px rgba(14, 165, 233, 0.2);
        }
        
        .output-section h3 {
            color: #0369a1;
            margin-bottom: 20px;
            font-size: 20px;
            font-weight: 600;
            display: flex;
            align-items: center;
            gap: 10px;
        }
        
        .output-section h3::before {
            content: '✨';
            font-size: 24px;
        }
        
        .arrow {
            text-align: center;
            color: #667eea;
            font-size: 32px;
            margin: -15px 0;
            position: relative;
            z-index: 1;
            filter: drop-shadow(0 2px 4px rgba(102, 126, 234, 0.3));
        }
        
        .arrow::before {
            content: '↓';
        }
        
        @media (max-width: 768px) {
            .container {
                padding: 10px;
            }
            
            .tab {
                padding: 12px 15px;
                font-size: 13px;
            }
            
            .tab-content {
                padding: 20px;
            }
        }
    </style>
</head>
<body>
    <div class="container">
        <h1>知识获取工作流可视化</h1>
        <div class="tabs" id="tabs">
'''
    
    # 生成Tab标签
    for i, workflow in enumerate(workflows):
        # 使用文件夹名称作为Tab标题
        folder_name = workflow.get('folder_name', f'问题 {i+1}')
        active_class = 'active' if i == 0 else ''
        html += f'            <button class="tab {active_class}" onclick="switchTab({i})">{escape_html(folder_name)}</button>\n'
    
    html += '        </div>\n'
    
    # 生成Tab内容
    for i, workflow in enumerate(workflows):
        active_class = 'active' if i == 0 else ''
        html += f'        <div class="tab-content {active_class}" id="tab-{i}">\n'
        
        # 输入信息
        html += '            <div class="input-section">\n'
        html += '                <h3>输入信息</h3>\n'
        input_text = workflow['input'].get('text', '')
        if input_text:
            html += f'                <div class="input-item" style="white-space: pre-wrap; line-height: 1.8;">{escape_html(input_text)}</div>\n'
        else:
            html += '                <div class="input-item"><span class="placeholder">（无）</span></div>\n'
        html += '            </div>\n'
        
        # 工作流程
        html += '            <div class="workflow">\n'
        for j, step in enumerate(workflow['steps']):
            step_id = f"step-{i}-{j}"
            html += f'                <div class="workflow-step" id="{step_id}">\n'
            html += '                    <div class="step-header" onclick="toggleStep(\'' + step_id + '\')">\n'
            html += '                        <div class="step-title">\n'
            html += f'                            <span class="step-number">{j+1}</span>\n'
            html += f'                            <span class="step-name">{escape_html(step["name"])}</span>\n'
            html += '                        </div>\n'
            html += '                        <span class="step-toggle">▼</span>\n'
            html += '                    </div>\n'
            html += '                    <div class="step-content" id="content-' + step_id + '">\n'
            
            # 根据步骤类型显示不同内容（prompt放在最后，默认隐藏）
            prompt_id = f"prompt-{step_id}"
            
            if step['step'] == 'generate_query':
                if step.get('query'):
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">生成的Query:</span>\n'
                    html += f'                            <div class="step-detail-content">{escape_html(step["query"])}</div>\n'
                    html += '                        </div>\n'
            
            elif step['step'] == 'select_tool':
                # 判断是否选择了工具
                tool_name = step.get('tool_name', '')
                if tool_name and tool_name != '无工具匹配':
                    # 有工具选择时显示工具信息
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">工具名称:</span>\n'
                    html += f'                            <div class="step-detail-content">{escape_html(tool_name)}</div>\n'
                    html += '                        </div>\n'
                    if step.get('tool_id'):
                        html += '                        <div class="step-detail">\n'
                        html += '                            <span class="step-detail-label">工具调用ID:</span>\n'
                        html += f'                            <div class="step-detail-content">{escape_html(step["tool_id"])}</div>\n'
                        html += '                        </div>\n'
                    if step.get('match_reason'):
                        html += '                        <div class="step-detail">\n'
                        html += '                            <span class="step-detail-label">匹配理由:</span>\n'
                        html += f'                            <div class="step-detail-content">{escape_html(step["match_reason"])}</div>\n'
                        html += '                        </div>\n'
                    if step.get('application_scenario'):
                        html += '                        <div class="step-detail">\n'
                        html += '                            <span class="step-detail-label">应用场景:</span>\n'
                        html += f'                            <div class="step-detail-content">{escape_html(step["application_scenario"])}</div>\n'
                        html += '                        </div>\n'
                    if step.get('tool_usage'):
                        html += '                        <div class="step-detail">\n'
                        html += '                            <span class="step-detail-label">使用方法:</span>\n'
                        html += f'                            <div class="step-detail-content">{escape_html(step["tool_usage"])}</div>\n'
                        html += '                        </div>\n'
                else:
                    # 无工具选择时显示提示和匹配理由、应用场景
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">选择结果:</span>\n'
                    html += '                            <div class="step-detail-content" style="color: #dc3545; font-weight: 500;">无匹配工具</div>\n'
                    html += '                        </div>\n'
                    if step.get('match_reason'):
                        html += '                        <div class="step-detail">\n'
                        html += '                            <span class="step-detail-label">匹配理由:</span>\n'
                        html += f'                            <div class="step-detail-content">{escape_html(step["match_reason"])}</div>\n'
                        html += '                        </div>\n'
                    if step.get('application_scenario'):
                        html += '                        <div class="step-detail">\n'
                        html += '                            <span class="step-detail-label">应用场景:</span>\n'
                        html += f'                            <div class="step-detail-content">{escape_html(step["application_scenario"])}</div>\n'
                        html += '                        </div>\n'
            
            elif step['step'] == 'extract_params':
                if step.get('params'):
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">提取的参数:</span>\n'
                    params_str = format_json_for_display(step['params'])
                    html += f'                            <div class="step-detail-content json-content">{escape_html(params_str)}</div>\n'
                    html += '                        </div>\n'
            
            elif step['step'] == 'execute_tool':
                if step.get('response'):
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">执行结果:</span>\n'
                    response_str = format_json_for_display(step['response'])
                    html += f'                            <div class="step-detail-content json-content">{escape_html(response_str)}</div>\n'
                    html += '                        </div>\n'
            
            elif step['step'] == 'evaluate_tool_result':
                can_answer = step.get('can_answer', '')
                reason = step.get('reason', '')
                if can_answer:
                    # 根据是否可以回答设置颜色
                    answer_color = '#28a745' if can_answer == '是' else '#dc3545'
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">是否可以回答输入需求:</span>\n'
                    html += f'                            <div class="step-detail-content" style="color: {answer_color}; font-weight: 500;">{escape_html(can_answer)}</div>\n'
                    html += '                        </div>\n'
                if reason:
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">理由:</span>\n'
                    html += f'                            <div class="step-detail-content">{escape_html(reason)}</div>\n'
                    html += '                        </div>\n'
            
            elif step['step'] == 'llm_search':
                # 显示生成的查询
                queries = step.get('queries', [])
                if queries:
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">生成的查询:</span>\n'
                    for idx, query in enumerate(queries, 1):
                        html += '                            <div style="margin-bottom: 8px; padding: 8px; background: #e8f4f8; border-radius: 4px; border-left: 3px solid #4a90e2;">\n'
                        html += f'                                <div style="font-weight: 500; color: #2c3e50;">查询 {idx}: {escape_html(query)}</div>\n'
                        html += '                            </div>\n'
                    html += '                        </div>\n'
                
                # 显示搜索结果
                search_results = step.get('search_results', [])
                if search_results:
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">搜索结果:</span>\n'
                    for idx, result in enumerate(search_results, 1):
                        query = result.get('query', '')
                        content = result.get('content', '')
                        html += '                            <div style="margin-bottom: 15px; padding: 12px; background: #f0f8ff; border-radius: 6px; border-left: 3px solid #4a90e2;">\n'
                        html += f'                                <div style="font-weight: 600; color: #2c3e50; margin-bottom: 8px;">查询 {idx}: {escape_html(query)}</div>\n'
                        html += f'                                <div style="color: #555; line-height: 1.6; white-space: pre-wrap;">{escape_html(content)}</div>\n'
                        html += '                            </div>\n'
                    html += '                        </div>\n'
                
                # 显示合并结果
                merge_response = step.get('merge_response', '')
                if merge_response:
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">合并结果:</span>\n'
                    html += f'                            <div class="step-detail-content">{escape_html(merge_response)}</div>\n'
                    html += '                        </div>\n'
                
                # 显示来源统计
                sources_count = step.get('sources_count', 0)
                if sources_count > 0:
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">来源统计:</span>\n'
                    html += f'                            <div class="step-detail-content">来源数: {sources_count}</div>\n'
                    html += '                        </div>\n'
            
            elif step['step'] == 'multi_search_merge':
                if step.get('sources_count') is not None:
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">来源统计:</span>\n'
                    html += f'                            <div class="step-detail-content">总来源数: {step.get("sources_count", 0)}, 有效来源数: {step.get("valid_sources_count", 0)}</div>\n'
                    html += '                        </div>\n'
                if step.get('response'):
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">整合结果:</span>\n'
                    response_str = format_json_for_display(step['response'])
                    html += f'                            <div class="step-detail-content">{escape_html(response_str)}</div>\n'
                    html += '                        </div>\n'
            
            elif step['step'] == 'extra_tools':
                selected_tools = step.get('selected_tools', [])
                if selected_tools:
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">发现的新工具:</span>\n'
                    for idx, tool in enumerate(selected_tools, 1):
                        tool_name = tool.get('tool_name', '')
                        tool_function = tool.get('function', '')
                        html += '                            <div style="margin-bottom: 15px; padding: 12px; background: #fff3cd; border-radius: 6px; border-left: 3px solid #ffc107;">\n'
                        html += f'                                <div style="font-weight: 600; color: #856404; margin-bottom: 6px;">工具 {idx}: {escape_html(tool_name)}</div>\n'
                        html += f'                                <div style="color: #856404; line-height: 1.6;">{escape_html(tool_function)}</div>\n'
                        html += '                            </div>\n'
                    html += '                        </div>\n'
                else:
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">发现结果:</span>\n'
                    html += '                            <div class="step-detail-content" style="color: #6c757d;">未发现合适的新工具</div>\n'
                    html += '                        </div>\n'
                if step.get('analysis_summary'):
                    html += '                        <div class="step-detail">\n'
                    html += '                            <span class="step-detail-label">分析摘要:</span>\n'
                    html += f'                            <div class="step-detail-content">{escape_html(step["analysis_summary"])}</div>\n'
                    html += '                        </div>\n'
            
            # Prompt放在最后，默认隐藏（generate_query 步骤不显示 prompt）
            if step.get('prompt') and step['step'] != 'generate_query':
                html += f'                        <button class="prompt-toggle-btn" onclick="togglePrompt(\'{prompt_id}\')">显示 Prompt</button>\n'
                html += f'                        <div class="prompt-content" id="{prompt_id}">{escape_html(step["prompt"])}</div>\n'
            
            html += '                    </div>\n'
            html += '                </div>\n'
            
            # 添加箭头（除了最后一步）
            if j < len(workflow['steps']) - 1:
                html += '                <div class="arrow"></div>\n'
        
        html += '            </div>\n'
        
        # 输出信息（已隐藏）
        # if workflow['output'].get('result'):
        #     html += '            <div class="output-section">\n'
        #     html += '                <h3>最终输出</h3>\n'
        #     result_str = format_json_for_display(workflow['output']['result'])
        #     html += f'                <div class="step-detail-content json-content">{escape_html(result_str)}</div>\n'
        #     html += '            </div>\n'
        
        html += '        </div>\n'
    
    html += '''    </div>
    
    <script>
        function switchTab(index) {
            // 隐藏所有tab内容
            const contents = document.querySelectorAll('.tab-content');
            contents.forEach(content => content.classList.remove('active'));
            
            // 移除所有tab的active状态
            const tabs = document.querySelectorAll('.tab');
            tabs.forEach(tab => tab.classList.remove('active'));
            
            // 显示选中的tab内容
            document.getElementById('tab-' + index).classList.add('active');
            tabs[index].classList.add('active');
        }
        
        function toggleStep(stepId) {
            const step = document.getElementById(stepId);
            const content = document.getElementById('content-' + stepId);
            const toggle = step.querySelector('.step-toggle');
            
            if (content.classList.contains('expanded')) {
                content.classList.remove('expanded');
                toggle.classList.remove('expanded');
                step.classList.remove('active');
            } else {
                content.classList.add('expanded');
                toggle.classList.add('expanded');
                step.classList.add('active');
            }
        }
        
        function togglePrompt(promptId) {
            const promptContent = document.getElementById(promptId);
            const btn = promptContent.previousElementSibling;
            
            if (promptContent.classList.contains('show')) {
                promptContent.classList.remove('show');
                btn.textContent = '显示 Prompt';
            } else {
                promptContent.classList.add('show');
                btn.textContent = '隐藏 Prompt';
            }
        }
        
        // 页面加载时高亮第一个步骤
        window.addEventListener('load', function() {
            const firstSteps = document.querySelectorAll('.workflow-step');
            firstSteps.forEach((step, index) => {
                if (index === 0 || index % (firstSteps.length / document.querySelectorAll('.tab-content').length) === 0) {
                    step.classList.add('active');
                    const content = step.querySelector('.step-content');
                    const toggle = step.querySelector('.step-toggle');
                    if (content) {
                        content.classList.add('expanded');
                        toggle.classList.add('expanded');
                    }
                }
            });
        });
    </script>
</body>
</html>'''
    
    return html


def main():
    """主函数"""
    # 获取当前脚本所在目录
    script_dir = Path(__file__).parent
    os.chdir(script_dir)
    
    # 读取数据文件
    print("正在读取数据文件...")
    data_list = load_data_files(DATA_FILE_PATHS)
    
    if not data_list:
        print("错误: 没有成功读取任何数据文件")
        return
    
    print(f"成功读取 {len(data_list)} 个数据文件")
    
    # 解析工作流数据
    print("正在解析工作流数据...")
    workflows = [parse_workflow_data(data, folder_name) for data, folder_name in data_list]
    
    # 生成HTML
    print("正在生成HTML页面...")
    html = generate_html(workflows)
    
    # 保存HTML文件
    timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
    output_filename = f"workflow_visualization_{timestamp}.html"
    
    with open(output_filename, 'w', encoding='utf-8') as f:
        f.write(html)
    
    print(f"HTML页面已生成: {output_filename}")
    print(f"文件路径: {os.path.abspath(output_filename)}")


if __name__ == '__main__':
    main()