src/lib/scraper/detector.test.ts

/**
 * Tests for URL detector
 */

import { detectPlatform, isPlatform, extractPostId } from './detector';

describe('URL Detector', () => {
  describe('detectPlatform', () => {
    it('should detect Instagram posts', () => {
      const result = detectPlatform('https://www.instagram.com/p/Abc123xyz/');
      expect(result.platform).toBe('instagram');
      expect(result.postId).toBe('Abc123xyz');
    });

    it('should detect Instagram reels', () => {
      const result = detectPlatform('https://instagram.com/reel/Abc123/');
      expect(result.platform).toBe('instagram');
    });

    it('should detect Twitter/X tweets', () => {
      const result = detectPlatform('https://twitter.com/user/status/1234567890');
      expect(result.platform).toBe('twitter');
      expect(result.postId).toBe('1234567890');
    });

    it('should detect X.com tweets', () => {
      const result = detectPlatform('https://x.com/user/status/9876543210');
      expect(result.platform).toBe('twitter');
    });

    it('should detect TikTok videos', () => {
      const result = detectPlatform('https://www.tiktok.com/@user/video/1234567890');
      expect(result.platform).toBe('tiktok');
    });

    it('should detect TikTok short URLs', () => {
      const result = detectPlatform('https://vm.tiktok.com/Abc123/');
      expect(result.platform).toBe('tiktok');
    });

    it('should detect Threads posts', () => {
      const result = detectPlatform('https://www.threads.net/@user/post/1234567890');
      expect(result.platform).toBe('threads');
    });

    it('should detect generic web URLs', () => {
      const result = detectPlatform('https://example.com/article');
      expect(result.platform).toBe('web');
    });

    it('should handle URLs without protocol', () => {
      const result = detectPlatform('www.instagram.com/p/Abc123/');
      expect(result.platform).toBe('instagram');
    });

    it('should handle invalid URLs', () => {
      const result = detectPlatform('not a real url at all');
      expect(result.platform).toBe('unknown');
    });
  });

  describe('isPlatform', () => {
    it('should check if URL is from a specific platform', () => {
      expect(isPlatform('https://instagram.com/p/Abc/', 'instagram')).toBe(true);
      expect(isPlatform('https://instagram.com/p/Abc/', 'twitter')).toBe(false);
    });
  });

  describe('extractPostId', () => {
    it('should extract post ID from Instagram URL', () => {
      const id = extractPostId('https://instagram.com/p/MyPostId123/');
      expect(id).toBe('MyPostId123');
    });

    it('should extract tweet ID from Twitter URL', () => {
      const id = extractPostId('https://twitter.com/user/status/1234567890');
      expect(id).toBe('1234567890');
    });

    it('should return undefined for URLs without IDs', () => {
      const id = extractPostId('https://example.com');
      expect(id).toBeUndefined();
    });
  });
});