Wordpress XML Export 2 JSON + HTML + MD

package.json

{
  "dependencies": {
    "turndown": "^7.2.0",
    "xml2js": "^0.6.2"
  }
}

parser.js

const fs = require('fs');
const xml2js = require('xml2js');
const path = require('path');
const TurndownService = require('turndown');

async function parseWordpressXML(filename) {
    try {
        const xmlData = fs.readFileSync(filename, 'utf8');

        const parser = new xml2js.Parser({
            explicitArray: false,
            mergeAttrs: true,
            trim: true
        });

        const result = await new Promise((resolve, reject) => {
            parser.parseString(xmlData, (err, parsed) => {
                if (err) reject(err);
                resolve(parsed);
            });
        });

        const turndownService = new TurndownService();

        const outputDir = path.join(process.cwd(), path.basename(filename) + '_wordpress_export');
        const postsDir = path.join(outputDir, 'posts');
        const metadataDir = path.join(outputDir, 'metadata');

        if (!fs.existsSync(outputDir)) fs.mkdirSync(outputDir);
        if (!fs.existsSync(postsDir)) fs.mkdirSync(postsDir);
        if (!fs.existsSync(metadataDir)) fs.mkdirSync(metadataDir);

        const authorsMap = {};
        const authors = Array.isArray(result.rss.channel['wp:author'])
            ? result.rss.channel['wp:author']
            : [result.rss.channel['wp:author']];

        const processedAuthors = authors.map(author => {
            const authorData = {
                id: author['wp:author_id'],
                login: author['wp:author_login'],
                email: author['wp:author_email'],
                firstName: author['wp:author_first_name'],
                lastName: author['wp:author_last_name'],
                displayName: author['wp:author_display_name'],
                fullName: `${author['wp:author_first_name']} ${author['wp:author_last_name']}`.trim()
            };

            authorsMap[author['wp:author_login']] = authorData;
            return authorData;
        });

        fs.writeFileSync(
            path.join(metadataDir, 'authors.json'),
            JSON.stringify(processedAuthors, null, 2)
        );

        const categories = Array.isArray(result.rss.channel['wp:category'])
            ? result.rss.channel['wp:category']
            : [result.rss.channel['wp:category']];

        const processedCategories = categories.map(cat => ({
            id: cat['wp:term_id'],
            name: cat['wp:cat_name'],
            slug: cat['wp:category_nicename'],
            description: cat['wp:category_description'],
            parent: cat['wp:category_parent']
        }));

        fs.writeFileSync(
            path.join(metadataDir, 'categories.json'),
            JSON.stringify(processedCategories, null, 2)
        );

        const tags = Array.isArray(result.rss.channel['wp:tag'])
            ? result.rss.channel['wp:tag']
            : [result.rss.channel['wp:tag']];

        const processedTags = tags.map(tag => ({
            id: tag['wp:term_id'],
            name: tag['wp:tag_name'],
            slug: tag['wp:tag_slug'],
            description: tag['wp:tag_description']
        }));

        fs.writeFileSync(
            path.join(metadataDir, 'tags.json'),
            JSON.stringify(processedTags, null, 2)
        );

        const posts = Array.isArray(result.rss.channel.item)
            ? result.rss.channel.item.filter(item => item['wp:post_type'] === 'post')
            : [];

        const processedPosts = posts.map(post => {
            const authorLogin = post['dc:creator'];

            var processedPost = {
                title: post.title,
                pubDate: post.pubDate,
                creator: authorLogin,
                description: post.description,
                post_id: post['wp:post_id'],
                post_date_gmt: post['wp:post_date_gmt'],
                post_modified_gmt: post['wp:post_modified_gmt'],
                status: post['wp:status'],
            };

            if (post.category && (Array.isArray(post.category) ? post.category.filter(c => c.domain === 'category').map(c => c._).length > 0 : [post.category._].filter(c => c.domain === 'category').length > 0)) {
                processedPost.categories = Array.isArray(post.category) ? post.category.filter(c => c.domain === 'category').map(c => c._) : [post.category._].filter(c => c.domain === 'category');
            }

            if (post.category && (Array.isArray(post.category) ? post.category.filter(t => t.domain === 'post_tag').map(t => t._).length > 0 : [post.category._].filter(t => t.domain === 'post_tag').length > 0))
                processedPost.tags = Array.isArray(post.category) ? post.category.filter(t => t.domain === 'post_tag').map(t => t._) : [post.category._].filter(t => t.domain === 'post_tag');

            const sanitizedTitle = post.title
                .normalize('NFD')                   // Normalize Unicode characters
                .replace(/[\u0300-\u036f]/g, '')    // Remove diacritical marks
                .replace(/[^a-z0-9\s-]/gi, '')      // Remove special characters except spaces and hyphens
                .replace(/\s+/g, '-')               // Replace spaces with hyphens
                .toLowerCase()
                .substring(0, 100);                 // Optional: limit filename length

            const htmlFilename = path.join(postsDir, `${sanitizedTitle}_${post['wp:post_id']}.html`);
            fs.writeFileSync(htmlFilename, post['content:encoded']);

            const markdownFilename = path.join(postsDir, `${sanitizedTitle}_${post['wp:post_id']}.md`);
            const markdownContent = turndownService.turndown(post['content:encoded']);
            fs.writeFileSync(markdownFilename, markdownContent);

            return processedPost;
        });

        const jsonFilename = path.join(outputDir, 'wordpress_posts.json');
        fs.writeFileSync(jsonFilename, JSON.stringify(processedPosts, null, 2));

        console.log(`Processed ${processedPosts.length} posts`);
        console.log(`Processed ${processedAuthors.length} authors`);
        console.log(`Processed ${processedCategories.length} categories`);
        console.log(`Processed ${processedTags.length} tags`);
        console.log(`Output saved to ${outputDir}`);

        return {
            posts: processedPosts,
            authors: processedAuthors,
            categories: processedCategories,
            tags: processedTags
        };
    } catch (error) {
        console.error('Error parsing WordPress XML:', error);
        throw error;
    }
}

const filename = process.argv[2];

if (!filename) {
    console.error('Please provide a WordPress XML export file');
    console.error('Usage: node script.js <filename>');
    process.exit(1);
}

parseWordpressXML(filename)
    .then(() => console.log('Export complete'))
    .catch(err => console.error('Export failed:', err));