pengzhan.dev/server/services/content.js

const fs = require('fs');
const path = require('path');
const matter = require('gray-matter');
const { createMarkdownRenderer, extractToc } = require('./markdown');
const searchEngine = require('./search-engine');
const pluginManager = require('./plugin-manager');

class ContentService {
  constructor() {
    this.contentDir = '';
    this.postsBySlug = new Map();       // slug -> { defaultPost, variants: { lang -> post } }
    this.allPostsList = [];             // Chronological list of default language posts
    this.pagesBySlug = new Map();       // slug -> page object
    this.taxonomy = {
      categories: new Map(),            // name.toLowerCase() -> { name, count, posts }
      tags: new Map()                   // name.toLowerCase() -> { name, slug, count, posts }
    };
    this.markdownRenderer = null;
  }

  init(contentDir) {
    this.contentDir = contentDir;
    this.markdownRenderer = createMarkdownRenderer(pluginManager);
    this.reload();
  }

  reload() {
    console.log('[ContentService] Scanning content directory:', this.contentDir);
    this.postsBySlug.clear();
    this.allPostsList = [];
    this.pagesBySlug.clear();
    this.taxonomy.categories.clear();
    this.taxonomy.tags.clear();

    // Re-initialize markdown renderer in case plugins changed
    this.markdownRenderer = createMarkdownRenderer(pluginManager);

    this.loadPosts();
    this.loadPages();

    // Update search engine with all searchable post variants
    const allSearchable = [];
    for (const entry of this.postsBySlug.values()) {
      allSearchable.push(entry.defaultPost);
      for (const [lang, variant] of Object.entries(entry.variants)) {
        if (lang !== entry.defaultPost.lang) {
          allSearchable.push(variant);
        }
      }
    }
    searchEngine.setPosts(allSearchable);

    // Sync to SQLite database & FTS5
    const dbService = require('./db');
    dbService.syncPosts(allSearchable);

    // Invalidate super page cache on content refresh
    const pageCache = require('../middleware/cache');
    pageCache.clear();

    console.log(`[ContentService] Loaded ${this.allPostsList.length} blog posts and ${this.pagesBySlug.size} pages.`);
  }

  loadPosts() {
    const postsDir = path.join(this.contentDir, 'posts');
    if (!fs.existsSync(postsDir)) {
      console.warn(`[ContentService] Posts directory not found at ${postsDir}`);
      return;
    }

    const entries = fs.readdirSync(postsDir, { withFileTypes: true });

    for (const entry of entries) {
      if (!entry.isDirectory()) continue;
      const slug = entry.name;
      const postDirPath = path.join(postsDir, slug);

      // Find all markdown files in the post folder
      const files = fs.readdirSync(postDirPath);
      const mdFiles = files.filter(f => f.endsWith('.md'));
      if (mdFiles.length === 0) continue;

      const variants = {};
      let defaultPost = null;

      for (const file of mdFiles) {
        const filePath = path.join(postDirPath, file);
        const fileContent = fs.readFileSync(filePath, 'utf8');
        const { data: frontmatter, content: rawMarkdown } = matter(fileContent);

        // Determine language:
        // if file is 'index.md', lang defaults to frontmatter.lang or 'en'
        // if file is 'zh.md', lang is 'zh'
        const baseName = path.basename(file, '.md');
        const lang = frontmatter.lang || (baseName === 'index' ? 'en' : baseName);

        const renderedHtml = this.markdownRenderer.render(rawMarkdown);
        const toc = extractToc(rawMarkdown);

        // Compute excerpt if not explicitly given
        let excerpt = frontmatter.excerpt;
        if (!excerpt) {
          const firstParaMatch = rawMarkdown.replace(/#+\s+.+/g, '').trim().split(/\n\s*\n/)[0];
          excerpt = firstParaMatch ? firstParaMatch.replace(/[*_`#[\]]/g, '').slice(0, 180) : '';
        }

        // Resolve cover image path relative to web root
        let cover = frontmatter.cover || null;
        if (cover && !cover.startsWith('http') && !cover.startsWith('/')) {
          cover = `/posts-assets/${slug}/${cover}`;
        }

        // Tags normalization
        const rawTags = Array.isArray(frontmatter.tags)
          ? frontmatter.tags
          : (frontmatter.tags ? String(frontmatter.tags).split(',').map(s => s.trim()) : []);

        const postObj = {
          slug,
          lang,
          title: frontmatter.title || slug,
          date: frontmatter.date ? new Date(frontmatter.date) : new Date(),
          author: frontmatter.author || 'Pengzhan Hao',
          category: frontmatter.category || 'Tech',
          tags: rawTags,
          cover,
          excerpt,
          rawContent: rawMarkdown,
          contentHtml: renderedHtml,
          toc,
          isDefault: (baseName === 'index'),
          availableLanguages: []
        };

        pluginManager.applyPostHooks(postObj);

        variants[lang] = postObj;
        if (baseName === 'index' || !defaultPost) {
          defaultPost = postObj;
        }
      }

      if (defaultPost) {
        const availableLangs = Object.keys(variants);
        defaultPost.availableLanguages = availableLangs;
        for (const v of Object.values(variants)) {
          v.availableLanguages = availableLangs;
        }

        this.postsBySlug.set(slug, {
          slug,
          defaultPost,
          variants
        });

        this.allPostsList.push(defaultPost);

        // Index Dynamic Taxonomy from default post
        this.indexTaxonomy(defaultPost);
      }
    }

    // Sort posts list in reverse chronological order
    this.allPostsList.sort((a, b) => b.date.getTime() - a.date.getTime());
  }

  indexTaxonomy(post) {
    // 1. Category indexing
    if (post.category) {
      const catKey = post.category.trim().toLowerCase();
      if (!this.taxonomy.categories.has(catKey)) {
        this.taxonomy.categories.set(catKey, {
          name: post.category.trim(),
          count: 0,
          posts: []
        });
      }
      const catObj = this.taxonomy.categories.get(catKey);
      catObj.count++;
      catObj.posts.push(post);
    }

    // 2. Tag indexing
    if (Array.isArray(post.tags)) {
      for (const tag of post.tags) {
        if (!tag) continue;
        const tagName = String(tag).trim();
        const tagKey = tagName.toLowerCase();
        const tagSlug = encodeURIComponent(tagKey.replace(/\s+/g, '-'));

        if (!this.taxonomy.tags.has(tagKey)) {
          this.taxonomy.tags.set(tagKey, {
            name: tagName,
            slug: tagSlug,
            count: 0,
            posts: []
          });
        }
        const tagObj = this.taxonomy.tags.get(tagKey);
        tagObj.count++;
        tagObj.posts.push(post);
      }
    }
  }

  loadPages() {
    const pagesDir = path.join(this.contentDir, 'pages');
    if (!fs.existsSync(pagesDir)) {
      return;
    }

    const files = fs.readdirSync(pagesDir);
    for (const file of files) {
      if (!file.endsWith('.md') && !file.endsWith('.html')) continue;
      const filePath = path.join(pagesDir, file);
      const fileContent = fs.readFileSync(filePath, 'utf8');
      const { data: frontmatter, content: rawContent } = matter(fileContent);

      const baseName = path.basename(file, path.extname(file));
      const slug = frontmatter.permalink || (baseName === 'home' ? '/' : `/${baseName}`);

      let contentHtml = '';
      if (file.endsWith('.html')) {
        contentHtml = rawContent;
      } else {
        contentHtml = this.markdownRenderer.render(rawContent);
      }

      const pageObj = {
        slug,
        title: frontmatter.title || baseName,
        layout: frontmatter.layout || (baseName === 'home' ? 'home' : 'page'),
        frontmatter,
        rawContent,
        contentHtml
      };

      this.pagesBySlug.set(slug, pageObj);
    }
  }

  getAllPosts() {
    return this.allPostsList;
  }

  getPost(slug, lang) {
    const entry = this.postsBySlug.get(slug);
    if (!entry) return null;

    if (lang && entry.variants[lang]) {
      return entry.variants[lang];
    }
    return entry.defaultPost;
  }

  getPage(slug) {
    return this.pagesBySlug.get(slug) || null;
  }

  getTaxonomy() {
    // Return structured taxonomy with top tags sorted by count
    const categories = Array.from(this.taxonomy.categories.values())
      .sort((a, b) => b.count - a.count);

    const tags = Array.from(this.taxonomy.tags.values())
      .sort((a, b) => b.count - a.count);

    return { categories, tags };
  }

  getAdjacentPosts(currentSlug) {
    const index = this.allPostsList.findIndex(p => p.slug === currentSlug);
    if (index === -1) return { prev: null, next: null };

    // List is sorted descending by date:
    // next is newer (index - 1), prev is older (index + 1)
    return {
      next: index > 0 ? this.allPostsList[index - 1] : null,
      prev: index < this.allPostsList.length - 1 ? this.allPostsList[index + 1] : null
    };
  }
}

module.exports = new ContentService();