hugo-mistergeek/scripts/generate-content.js

const fs = require('fs');
const path = require('path');
const he = require('he');

const DATA_DIR = path.join(__dirname, '..', 'data', 'wordpress');
const CONTENT_DIR = path.join(__dirname, '..', 'content');
const PAGES_DIR = path.join(CONTENT_DIR, 'pages');

function generateContent() {
  const posts = JSON.parse(fs.readFileSync(path.join(DATA_DIR, 'posts.json'), 'utf8'));
  const pages = JSON.parse(fs.readFileSync(path.join(DATA_DIR, 'pages.json'), 'utf8'));
  
  // Ensure content directories exist
  if (!fs.existsSync(CONTENT_DIR)) {
    fs.mkdirSync(CONTENT_DIR, { recursive: true });
  }
  
  if (!fs.existsSync(PAGES_DIR)) {
    fs.mkdirSync(PAGES_DIR, { recursive: true });
  }

  // Process posts - only include published posts
  posts.filter(post => post.status === 'publish').forEach(post => {
    const slug = post.slug;
    const date = new Date(post.date);
    const year = date.getFullYear();
    const month = String(date.getMonth() + 1).padStart(2, '0');

    // Get the primary category (first category in the list)
    const primaryCategory = post._embedded?.['wp:term']?.[0]?.[0];
    const categorySlug = primaryCategory ? primaryCategory.slug : 'non-classe';

    const contentDir = path.join(CONTENT_DIR, categorySlug, `${year}-${month}-${slug}`);
    // const contentDir = path.join(CONTENT_DIR, `${year}-${month}-${slug}`);

    if (!fs.existsSync(contentDir)) {
      fs.mkdirSync(contentDir, { recursive: true });
    }

    const frontmatter = {
      title: he.decode(post.title.rendered),
      date: post.date,
      draft: false,
      slug: slug,
      wordpress_id: post.id,
      excerpt: he.decode(post.excerpt.rendered.replace(/<[^>]*>/g, '')),
      featured_image: post._embedded?.['wp:featuredmedia']?.[0]?.source_url || '',
      author: post._embedded?.author?.[0]?.name || 'Unknown',
      categories: (post._embedded?.['wp:term']?.[0] || []).map(cat => cat.name || 'Non classé'),
      tags: (post._embedded?.['wp:term']?.[1] || []).map(cat => cat.name || 'Non classé'),
      section: categorySlug
    };

    // Decode HTML entities in the content and clean up HTML tags
    let contentHtml = he.decode(post.content.rendered);

    // Convert absolute URLs in a href to relative URLs (only for wp.mistergeek.net)
    contentHtml = contentHtml.replace(/<a\s+[^>]*href="([^"]+)"[^>]*>/g, (match, href) => {
      // Check if the href is an absolute URL containing wp.mistergeek.net
      if (href.startsWith('http://') || href.startsWith('https://')) {
        try {
          const url = new URL(href);
          if (url.hostname === 'wp.mistergeek.net' || url.hostname === 'www.wp.mistergeek.net') {
            // Only convert wp.mistergeek.net URLs to relative paths
            return match.replace(href, url.pathname);
          }
        } catch (error) {
          // If URL parsing fails, return the original href
          console.warn('Failed to parse URL:', href, error);
        }
      }
      return match;
    });

    contentHtml = contentHtml
      .replace(/<p>\s*<\/p>/g, '') // Remove empty paragraphs
      .replace(/<\/p>\s*<p>/g, '\n\n'); // Replace paragraph breaks with newlines

    const content = `---
${Object.entries(frontmatter)
  .map(([key, value]) => `${key}: ${JSON.stringify(value)}`)
  .join('\n')}
---
${contentHtml.trim()}`;

    fs.writeFileSync(path.join(contentDir, 'index.md'), content);
  });

  // Process pages - only include published pages
  pages.filter(page => page.status === 'publish').forEach(page => {
    const slug = page.slug;
    
    const contentDir = path.join(PAGES_DIR, slug);

    if (!fs.existsSync(contentDir)) {
      fs.mkdirSync(contentDir, { recursive: true });
    }

    const frontmatter = {
      title: he.decode(page.title.rendered),
      slug: slug,
      type: "pages",
      layout: "single",
      wordpress_id: page.id,
      date: page.date,
      modified: page.modified,
      draft: false,
      aliases: [`/${slug}/`]
    };

    // Decode HTML entities in the content and clean up HTML tags
    let contentHtml = he.decode(page.content.rendered);

    // Convert absolute URLs in a href to relative URLs (only for wp.mistergeek.net)
    contentHtml = contentHtml.replace(/<a\s+[^>]*href="([^"]+)"[^>]*>/g, (match, href) => {
      // Check if the href is an absolute URL containing wp.mistergeek.net
      if (href.startsWith('http://') || href.startsWith('https://')) {
        try {
          const url = new URL(href);
          if (url.hostname === 'wp.mistergeek.net' || url.hostname === 'www.wp.mistergeek.net') {
            // Only convert wp.mistergeek.net URLs to relative paths
            return match.replace(href, url.pathname);
          }
        } catch (error) {
          // If URL parsing fails, return the original href
          console.warn('Failed to parse URL:', href, error);
        }
      }
      return match;
    });

    contentHtml = contentHtml
      .replace(/<p>\s*<\/p>/g, '') // Remove empty paragraphs
      .replace(/<\/p>\s*<p>/g, '\n\n'); // Replace paragraph breaks with newlines

    const content = `---
${Object.entries(frontmatter)
  .map(([key, value]) => `${key}: ${JSON.stringify(value)}`)
  .join('\n')}
---
${contentHtml.trim()}`;

    fs.writeFileSync(path.join(contentDir, 'index.md'), content);
  });

  const publishedPosts = posts.filter(post => post.status === 'publish');
  const publishedPages = pages.filter(page => page.status === 'publish');
  
  // Generate author directories and index pages
  generateAuthorDirectories(publishedPosts);
  
  console.log(`✅ Generated ${publishedPosts.length} content files`);
  console.log(`✅ Generated ${publishedPages.length} page files`);
}

function generateAuthorDirectories(posts) {
  const AUTHORS_DIR = path.join(CONTENT_DIR, 'author');
  
  // Ensure authors directory exists
  if (!fs.existsSync(AUTHORS_DIR)) {
    fs.mkdirSync(AUTHORS_DIR, { recursive: true });
  }
  
  // Group posts by author
  const postsByAuthor = {};
  
  posts.forEach(post => {
    const authorName = post._embedded?.author?.[0]?.name || 'Unknown';
    const authorSlug = authorName.toLowerCase().replace(/\s+/g, '-').replace(/[^\w-]/g, '');
    
    if (!postsByAuthor[authorSlug]) {
      postsByAuthor[authorSlug] = {
        name: authorName,
        slug: authorSlug,
        posts: []
      };
    }
    
    postsByAuthor[authorSlug].posts.push(post);
  });
  
  // Create author directories and index pages
  Object.values(postsByAuthor).forEach(author => {
    const authorDir = path.join(AUTHORS_DIR, author.slug);
    
    if (!fs.existsSync(authorDir)) {
      fs.mkdirSync(authorDir, { recursive: true });
    }
    
    // Generate author index page
    const frontmatter = {
      title: `Lise des articles de ${author.name}`,
      type: 'authors',
      layout: 'list',
      author: author.name,
      author_slug: author.slug
    };
    
    const content = `---
${Object.entries(frontmatter)
  .map(([key, value]) => `${key}: ${JSON.stringify(value)}`)
  .join('\n')}
---
`;
    
    fs.writeFileSync(path.join(authorDir, '_index.md'), content);
    
    console.log(`✅ Generated author directory: ${author.name} (${author.posts.length} posts)`);
  });
}

generateContent();