minnan-website/scripts/ExportToAi/ich.ts

import { writeFile } from 'fs/promises';
import CommonContent, { GetContentListParams, type GetContentDetailItem, type GetContentListItem } from '../../api/CommonContent';
import ProjectsContent from '../../api/inheritor/ProjectsContent';
import InheritorContent from '../../api/inheritor/InheritorContent';
import SeminarContent from '../../api/inheritor/SeminarContent';
import UnitContent from '../../api/inheritor/UnitContent';
import UnmoveableContent from '../../api/inheritor/UnmoveableContent';
import path from 'path';
import fs from 'fs';
import { argv, cwd } from 'process';

const data = [] as Array<GetContentListItem & { detail?: GetContentDetailItem }>;

// HTML转Markdown的简单实现
function htmlToMarkdown(html: string): string {
  if (!html) return '';
  
  // 处理标题
  html = html.replace(/<h1[^>]*>(.*?)<\/h1>/gi, '# $1\n\n');
  html = html.replace(/<h2[^>]*>(.*?)<\/h2>/gi, '## $1\n\n');
  html = html.replace(/<h3[^>]*>(.*?)<\/h3>/gi, '### $1\n\n');
  
  // 处理段落
  html = html.replace(/<p[^>]*>(.*?)<\/p>/gi, '$1');
  
  // 处理加粗
  html = html.replace(/<strong[^>]*>(.*?)<\/strong>/gi, '**$1**');
  html = html.replace(/<b[^>]*>(.*?)<\/b>/gi, '**$1**');
  
  // 处理斜体
  html = html.replace(/<em[^>]*>(.*?)<\/em>/gi, '*$1*');
  html = html.replace(/<i[^>]*>(.*?)<\/i>/gi, '*$1*');
  
  // 处理列表
  html = html.replace(/<ul[^>]*>([\s\S]*?)<\/ul>/gi, (match, content) => {
    return content.replace(/<li[^>]*>(.*?)<\/li>/gi, '- $1\n') + '\n';
  });
  
  html = html.replace(/<ol[^>]*>([\s\S]*?)<\/ol>/gi, (match, content, index) => {
    let count = 1;
    return content.replace(/<li[^>]*>(.*?)<\/li>/gi, () => {
      return `${count++}. $1\n`;
    }) + '\n';
  });
  
  // 处理图片
  html = html.replace(/<img[^>]*src="([^"]*)"[^>]*alt="([^"]*)"[^>]*>/gi, '![$2]($1)');
  
  // 处理链接
  html = html.replace(/<a[^>]*href="([^"]*)"[^>]*>(.*?)<\/a>/gi, '[$2]($1)');
  
  // 处理换行
  html = html.replace(/<br[^>]*>/gi, ' ');
  
  // 去除所有HTML标签
  html = html.replace(/<[^>]*>/g, '');
  
  // 处理多余的换行
  html = html.replace(/\n\s*\n/g, '\n\n');
  
  return html.trim();
}
function sanitizeFileNameAdvanced(fileName: string, replacement = '_') {
  if (typeof fileName !== 'string') return '';
  
  // 1. 定义跨平台非法字符正则（核心非法字符+不可见控制字符）
  const illegalRegex = /[<>:"/\\|?*\x00-\x1F]/g; // \x00-\x1F 是不可见ASCII控制字符
  // 2. 定义Windows保留文件名（不区分大小写）
  const windowsReservedNames = new Set([
    'con', 'nul', 'prn', 'aux', 'com1', 'com2', 'com3', 'com4',
    'com5', 'com6', 'com7', 'com8', 'com9', 'lpt1', 'lpt2',
    'lpt3', 'lpt4', 'lpt5', 'lpt6', 'lpt7', 'lpt8', 'lpt9'
  ]);
  
  // 步骤1：替换非法字符
  let safeFileName = fileName.replace(illegalRegex, replacement);
  
  // 步骤2：清理首尾空格、句点，压缩连续替换符（避免多个下划线连在一起）
  safeFileName = safeFileName
    .trim()
    .replace(/\.$/, '') // 移除结尾句点
    .replace(new RegExp(`${replacement}+`, 'g'), replacement); // 压缩连续替换符
  
  // 步骤3：处理Windows保留文件名
  const fileNameWithoutExt = safeFileName.split('.')[0].toLowerCase();
  if (windowsReservedNames.has(fileNameWithoutExt)) {
    safeFileName = `${safeFileName}_${Date.now()}`; // 追加时间戳避免冲突
  }
  
  // 步骤4：限制文件名长度（Windows最大255字符，这里取200字符留有余地）
  const maxFileNameLength = 200;
  if (safeFileName.length > maxFileNameLength) {
    const ext = safeFileName.includes('.') ? safeFileName.split('.').pop() : '';
    const name = safeFileName.slice(0, maxFileNameLength - (ext ? ext.length + 1 : 0));
    safeFileName = ext ? `${name}.${ext}` : name;
  }
  
  // 步骤5：兜底空文件名
  return safeFileName || 'unnamed_file';
}

// 生成Markdown文本
async function generateMarkdownIch(subDir: string, type: string) {
  
  for (const item of data) {
    
    let md = '---\n';    
    
    function addMeta(key: string, value: any) {
      if (value)
        md += `${key}: ${value}\n`;
    }

    addMeta('level', item.levelText);
    addMeta('crType', item.crTypeText);
    addMeta('region', item.regionText);
    addMeta('batch', item.batchText);
    addMeta('ichType', item.ichTypeText);
    addMeta('type', type);
    addMeta('unit', item.detail?.unit);
    addMeta('name', item.title);
    addMeta('address', item.address);

    md += `---\n\n`;

    // 基本信息
    md += `# ${item.title}\n\n`;
    if (item.desc)
      md += `${item.desc}\n\n`;
    md += `${item.title}类型：${type}\n`;

    function addRow(key: string, value: any) {
      if (value)
        md += `- ${key}: ${value}\n`;
    }
    addRow(`${item.title}非遗级别`, item.levelText);
    addRow(`${item.title}非遗类别`, item.ichTypeText);
    addRow(`${item.title}地区`, item.regionText);
    addRow(`${item.title}批次`, item.batchText);
    addRow(`${item.title}保护单位`, item.unit);
    addRow(`${item.title}地址`, item.address);
    addRow(`${item.title}字号名称`, item.fontName);
    addRow(`${item.title}认定类型`, item.brandType);
    md += `${item.title}数据库索引ID: 类型: intangible ID: ${item.id || '无'}\n\n`;
    
    // 详细信息
    if (item.detail) {
      const detail = item.detail as GetContentDetailItem;
      // 简介
      if (detail.intro)
        md += htmlToMarkdown(detail.intro) + '\n\n';
      // 内容
      if (detail.content)
        md += htmlToMarkdown(detail.content) + '\n\n';
      
      // 传承人
      if (detail.inheritorsList && detail.inheritorsList.length > 0) {
        md += `${item.title}相关非遗传承人：${detail.inheritorsList.map(inheritor => inheritor.title).join('、')}\n\n`;
        if (detail.inheritor) 
          md += htmlToMarkdown(detail.inheritor) + '\n\n';
      }
      
      // 传习所
      if (detail.ichSitesList && detail.ichSitesList.length > 0)
        md += `${item.title}相关非遗传习所: ${detail.ichSitesList.map(site => site.title).join('、')}\n\n`; 

      // 同级别项目
      if (detail.otherLevel && detail.otherLevel.length > 0)
        md += `${item.title}其他级别: ` + detail.otherLevel.map(project => project.levelText).join('、') + `\n\n`;

      // 传承谱系
      if (detail.pedigree) {
        md += `${item.title}的传承谱系: ` + htmlToMarkdown(detail.pedigree as string) + '\n\n';
      }
    }

    try {
      await writeFile(path.join(subDir, `${sanitizeFileNameAdvanced(item.title) || item.id}.md`), md);
    } catch (err) {
      console.error(`写入文件 ${item.title || item.id}.md 失败:`, err);
    }
  }
}
async function generateMarkdownInheritor(subDir: string) {
  
  for (const item of data) {
    
    let md = '---\n';    
    
    function addMeta(key: string, value: any) {
      if (value)
        md += `${key}: ${value}\n`;
    }

    addMeta('level', item.levelText);
    addMeta('type', '非遗传承人');
    addMeta('unit', item.detail?.unit);
    addMeta('sex', item.detail?.gender == '1'? '男' : '女');
    addMeta('name', item.title);

    md += `---\n\n`;

    // 基本信息
    md += `# ${item.title}\n\n`;
    if (item.desc)
      md += `${item.desc}\n\n`;
    md += `## 基本信息\n\n`;
    md += `${item.title}类型：非遗传承人\n`;
    function addRow(key: string, value: any) {
      if (value)
        md += `- ${key}: ${value}\n`;
    }
    addRow('民族', item.detail?.nation);
    addRow('性别', item.detail?.gender == '1'? '男' : '女');
    addRow('出生日期', item.detail?.dateBirth);
    addRow('出生地区', item.detail?.birthplace);
    addRow('单位', item.detail?.unit);
    addRow('传承项目', item.detail?.associationMeList[0]?.title);
    addRow('传承人级别', item.detail?.batchText);
    addRow('公布批次', item.detail?.batchText);

    md += `\n## ${item.title}数据库索引ID\n\n`;
    md += `- 类型: inheritor\n`;
    md += `- ID: ${item.id || '无'}\n\n`;
    
    // 详细信息
    if (item.detail) {
      const detail = item.detail as GetContentDetailItem;
      // 简介
      if (detail.intro) {
        md += `## ${item.title}简介\n\n`;
        md += htmlToMarkdown(detail.intro) + '\n\n';
      }
      if (detail.content) {
        md += htmlToMarkdown(detail.content) + '\n\n';
      }
      // 奖项
      if (detail.prize) {
        md += htmlToMarkdown(detail.prize as string) + '\n\n';
      }
      // 相关项目
      if (detail.associationMeList && detail.associationMeList.length > 0)
        md += `${item.title}相关项目${detail.associationMeList.map(project => project.title).join('、')}\n\n`;
      // 传习所
      if (detail.ichSitesList && detail.ichSitesList.length > 0)
        md += `${item.title}相关非遗传习所: ${detail.ichSitesList.map(site => site.title).join('、')}\n\n`;
    }

    await writeFile(path.join(subDir, `${sanitizeFileNameAdvanced(item.title) ||item.id}.md`), md);
  }
}
async function generateMarkdownArtifact(subDir: string) {
  
  for (const item of data) {
    
    let md = '---\n';    
    
    function addMeta(key: string, value: any) {
      if (value)
        md += `${key}: ${value}\n`;
    }

    addMeta('name', item.title);
    addMeta('crType', item.crTypeText);
    addMeta('region', item.regionText);
    addMeta('level', item.levelText);
    addMeta('address', item.address);
    addMeta('age', item.age);
    addMeta('type', '非遗文物');
    addMeta('unit', item.detail?.unit);

    md += `---\n\n`;

    // 基本信息
    md += `# ${item.title}\n\n`;
    if (item.desc)
      md += `${item.desc}\n\n`;
    md += `## 基本信息\n\n`;
    md += `类型：非遗文物\n`;

    function addRow(key: string, value: any) {
      if (value)
        md += `- ${key}: ${value}\n`;
    }
    addRow('开放时间', item.detail?.openStatusText);
    addRow('年代', item.age);
    addRow('级别', item.levelText);
    addRow('所属区域', item.regionText);
    addRow('文物类型', item.crTypeText);
    addRow('单位', item.detail?.unit);

    md += `\n## 数据库索引ID\n\n`;
    md += `- 类型: artifact\n`;
    md += `- ID: ${item.id || '无'}\n\n`;

    if (item.video) {
      md += `## 视频\n\n`;
      md += `![${item.title}视频](${item.video})\n\n`;
    }
    
    // 详细信息
    if (item.detail) {
      const detail = item.detail as GetContentDetailItem;
      // 简介
      if (detail.intro) {
        md += `## 简介\n\n`;
        md += htmlToMarkdown(detail.intro) + '\n\n';
      }
      if (detail.content) {
        md += `## 详情\n\n`;
        md += htmlToMarkdown(detail.content) + '\n\n';
      }
      // 奖项
      if (detail.protectedArea) {
        md += `## 保护范围\n\n`;
        md += htmlToMarkdown(detail.protectedArea as string) + '\n\n';
      }
      if (detail.environment) {
        md += `## 建筑环境\n\n`;
        md += htmlToMarkdown(detail.environment as string) + '\n\n';
      }
      if (detail.价值评估) {
        md += `## 价值评估\n\n`;
        md += htmlToMarkdown(detail.价值评估 as string) + '\n\n';
      }
    }

    await writeFile(path.join(subDir, `${sanitizeFileNameAdvanced(item.title) ||item.id}.md`), md);
  }
}

async function main() {

  const type = argv[2];

  function makeDir(nanme: string) {
    const dir = path.join(cwd(), `dist/${nanme}`);
    if (!fs.existsSync(dir)) 
      fs.mkdirSync(dir, { recursive: true });
    return dir;
  }

  switch (type) {
    case 'ich': {
      const dir = makeDir('ich');
      (await ProjectsContent.getContentList(new GetContentListParams(), 1, 1000)).list.forEach(item => {
        data.push(item);
      });
      for (const item of data)
        item.detail = (await ProjectsContent.getContentDetail(item.id)) as GetContentDetailItem;
      generateMarkdownIch(dir, '非遗项目');
      break;
    }
    case 'seminar': {
      const dir = makeDir('seminar');
      (await SeminarContent.getContentList(new GetContentListParams(), 1, 1000)).list.forEach(item => {
        data.push(item);
      });
      for (const item of data)
        item.detail = (await SeminarContent.getContentDetail(item.id)) as GetContentDetailItem;
      generateMarkdownIch(dir, '非遗传习所');
      break;
    }
    case 'old': {
      const dir = makeDir('old');
      (await CommonContent.getContentList(new GetContentListParams()
        .setModelId(17)
        .setMainBodyColumnId(312)
      , 1, 1000)).list.forEach(item => {
        data.push(item);
      });
      for (const item of data)
        item.detail = (await CommonContent.getContentDetail(item.id)) as GetContentDetailItem;
      generateMarkdownIch(dir, '老字号');
      break;
    }
    case 'unit': {
      const dir = makeDir('unit');
      (await UnitContent.getContentList(new GetContentListParams(), 1, 1000)).list.forEach(item => {
        data.push(item);
      });
      for (const item of data)
        item.detail = (await UnitContent.getContentDetail(item.id)) as GetContentDetailItem;
      generateMarkdownIch(dir, '非遗保护单位');
      break;
    }
    case 'inheritor': {
      const dir = makeDir('inheritor');
      (await InheritorContent.getContentList(new GetContentListParams(), 1, 1000)).list.forEach(item => {
        data.push(item);
      });
      for (const item of data)
        item.detail = (await InheritorContent.getContentDetail(item.id)) as GetContentDetailItem;
      generateMarkdownInheritor(dir);
      break;
    }
    case 'artifact': {
      const dir = makeDir('artifact');
      (await UnmoveableContent.getContentList(new GetContentListParams(), 1, 1000)).list.forEach(item => {
        data.push(item);
      });
      for (const item of data)
        item.detail = (await UnmoveableContent.getContentDetail(item.id)) as GetContentDetailItem;
      generateMarkdownArtifact(dir);
      break;
    }
    case 'publishToDify': {
      const datasetId = argv[3];
      const localDir = argv[4];
      if (!datasetId || !localDir) {
        console.error('请提供 datasetId 和 localDir 参数');
        console.error('用法: node ich.js publishToDify <datasetId> <localDir>');
        break;
      }
      await publishToDify(datasetId, localDir);
      break;
    }
    default:
      console.log('不支持的类型');
      break;
  }
}

async function publishToDify(datasetId: string, localDir: string) {
  // Dify API 配置
  const DIFY_API_KEY = process.env.DIFY_API_KEY || 'dataset-ZELjB79MnbcvCeyjEVMEuTmB';
  const DIFY_API_URL = 'http://localhost:8089/v1/datasets';

  if (!DIFY_API_KEY) {
    console.error('请设置 DIFY_API_KEY 环境变量');
    return;
  }

  if (!fs.existsSync(localDir)) {
    console.error(`本地目录不存在: ${localDir}`);
    return;
  }

  console.log(`开始上传文档到 Dify 知识库 (ID: ${datasetId})...`);
  console.log(`本地目录: ${localDir}`);

  try {
    // 读取本地目录中的所有 md 文件
    const files = fs.readdirSync(localDir);
    const mdFiles = files.filter(file => file.endsWith('.md'));

    if (mdFiles.length === 0) {
      console.log('目录中没有找到 .md 文件');
      return;
    }

    console.log(`找到 ${mdFiles.length} 个 .md 文件`);

    // 逐个上传文件
    for (const file of mdFiles) {
      console.log(`正在上传: ${file}`);

      const filePath = path.join(localDir, file);
      const content = fs.readFileSync(filePath, 'utf8');

      // 构建请求数据
      const requestData = JSON.stringify({
        name: file,
        text: content,
        indexing_technique: 'high_quality',  
        doc_language: 'zh',
      });

      // 发送上传请求
      const response = await fetch(`${DIFY_API_URL}/${datasetId}/document/create-by-text`, {
        method: 'POST',
        headers: {
          'Content-Type': 'application/json',
          'Authorization': `Bearer ${DIFY_API_KEY}`
        },
        body: requestData
      });

      if (response.ok) {
         await response.json();
        console.log(`✅ 上传成功: ${file})`);
      } else {
        const error = await response.json();
        console.error(`❌ 上传失败: ${file}`);
        console.error(`   错误信息: ${(error as Error).message || response.statusText}`);
      }
    }

    console.log('\n🎉 所有文档上传完成！');
  } catch (error) {
    console.error('上传过程中发生错误:', error);
  }
}

main();