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

    function addRow(key: string, value: any) {
      if (value)
        md += `- ${key}: ${value}\n`;
    }

    addRow('级别', item.levelText);
    addRow('类别', item.ichTypeText);
    addRow('地区', item.regionText);
    addRow('批次', item.batchText);
    addRow('保护单位', item.unit);
    addRow('地址', item.address);
    addRow('字号名称', item.fontName);
    addRow('认定类型', item.brandType);
    addRow('其他级别保护单位', item.detail?.otherLevel && item.detail.otherLevel.length > 0 ? `${item.detail.otherLevel.length}个` : '');
    
    md += `\n## 数据库索引ID\n\n`;
    md += `- 类型: intangible\n`;
    md += `- ID: ${item.id || '无'}\n\n`;

    // 图片
    if (item.images && item.images.length > 0) {
      md += `## 图片\n\n`;
      item.images.forEach(image => {
        md += `![${item.title}图片](${image})\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';
      }
      // 内容
      if (detail.content) {
        md += `## 内容\n\n`;
        md += htmlToMarkdown(detail.content) + '\n\n';
      }
      // 传承谱系
      if (detail.pedigree) {
        md += `## 传承谱系\n\n`;
        md += htmlToMarkdown(detail.pedigree as string) + '\n\n';
      }
      // 视频
      if (detail.video) {
        md += `## 视频\n\n`;
        md += `[视频](${detail.video})\n\n`;
      }
      if (detail.publishVideo)
        md += `[介绍视频](${detail.publishVideo})\n\n`;
      
      // 传承人
      if (detail.inheritorsList && detail.inheritorsList.length > 0) {
        md += `## 相关传承人\n\n`;
        if (detail.inheritor) {
          md += htmlToMarkdown(detail.inheritor) + '\n\n';
        }
        
        detail.inheritorsList.forEach(inheritor => {
          md += `### ${inheritor.title}\n\n`;
          md += `级别：${inheritor.levelLext || '无'}\n\n`;

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

          if (inheritor.image) {
            md += `![传承人头像](${inheritor.image})\n\n`;
          }
        });
      }
      
      // 传习所
      if (detail.ichSitesList && detail.ichSitesList.length > 0) {
        md += `## 相关传习所\n\n`;
        detail.ichSitesList.forEach(site => { 
          md += `### ${site.title}\n\n`;
          md += `级别：${site.levelLext || '无'}\n\n`;
          md += `地址：${site.address || '无'}\n\n`;

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

          if (site.image) {
            md += `![传习所图片](${site.image})\n\n`;
          }
        });
      }

      // 同级别项目
      if (detail.otherLevel && detail.otherLevel.length > 0) {
        md += `## 其他级别非遗项目\n\n`;
        detail.otherLevel.forEach(project => { 
          md += `### ${project.title}\n\n`;
          md += `级别：${project.levelLext || '无'}\n\n`;
          md += `保护单位：${project.unit || '无'}\n\n`;

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

    await writeFile(path.join(subDir, `${item.id}.md`), md);
  }
}
async function generateMarkdownInheritor(subDir: string) {
  
  for (const item of data) {
    
    let md = '';
    // 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：非遗传承人\n\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## 数据库索引ID\n\n`;
    md += `- 类型: inheritor\n`;
    md += `- ID: ${item.id || '无'}\n\n`;

    // 图片
    if (item.images && item.images.length > 0) {
      md += `## 图片\n\n`;
      item.images.forEach(image => {
        md += `![${item.title}图片](${image})\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.prize) {
        md += `## 奖项\n\n`;
        md += htmlToMarkdown(detail.prize as string) + '\n\n';
      }
      // 相关项目
      if (detail.associationMeList && detail.associationMeList.length > 0) {
        md += `## 相关项目\n\n`;        
        detail.associationMeList.forEach(inheritor => {
          md += `### ${inheritor.title}\n\n`;

          md += `#### 数据库索引ID\n\n`;
          md += `- 类型: intangible\n`;
          md += `- ID: ${inheritor.id || '无'}\n\n`;
        });
      }   
      // 传习所
      if (detail.ichSitesList && detail.ichSitesList.length > 0) {
        md += `## 相关传习所\n\n`;
        detail.ichSitesList.forEach(site => { 
          md += `### ${site.title}\n\n`;
          md += `级别：${site.levelLext || '无'}\n\n`;
          md += `地址：${site.address || '无'}\n\n`;

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

          if (site.image) {
            md += `![传习所图片](${site.image})\n\n`;
          }
        });
      }
    }

    await writeFile(path.join(subDir, `${item.id}.md`), md);
  }
}
async function generateMarkdownArtifact(subDir: string) {
  
  for (const item of data) {
    
    let md = '';
    // 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：非遗传承人\n\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`;

    // 图片
    if (item.images && item.images.length > 0) {
      md += `## 图片\n\n`;
      item.images.forEach(image => {
        md += `![${item.title}图片](${image})\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, `${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;
    }
    default:
      console.log('不支持的类型');
      break;
  }
}

main();