LLM Researcher

import { describe, it, expect } from 'vitest'; import { DuckDuckGoSearcher } from '@/search.js'; import { readFileSync } from 'fs'; import { join } from 'path'; describe('DuckDuckGoSearcher - HTML with both Previous and Next buttons', () => { it('should correctly parse both Previous and Next pagination info', () => { const searcher = new DuckDuckGoSearcher(); // Read the HTML with both Previous and Next buttons const html = readFileSync(join(__dirname, 'fixtures', 'ddg_with_both_buttons.html'), 'utf8'); const paginationInfo = searcher.parsePaginationInfo(html); console.log('Both buttons pagination info:', paginationInfo); // Should have both Previous and Next expect(paginationInfo.hasNext).toBe(true); expect(paginationInfo.hasPrevious).toBe(true); // Check Previous page parameters expect(paginationInfo.previousPageParams).toBeDefined(); expect(paginationInfo.previousPageParams?.s).toBe('15'); expect(paginationInfo.previousPageParams?.dc).toBe('-39'); expect(paginationInfo.previousPageParams?.q).toBe('"codehex"'); expect(paginationInfo.previousPageParams?.vqd).toBe('4-322476139742216406546201285543403802044'); // Check Next page parameters expect(paginationInfo.nextPageParams).toBeDefined(); expect(paginationInfo.nextPageParams?.s).toBe('80'); expect(paginationInfo.nextPageParams?.dc).toBe('41'); expect(paginationInfo.nextPageParams?.q).toBe('"codehex"'); expect(paginationInfo.nextPageParams?.vqd).toBe('4-322476139742216406546201285543403802044'); }); it('should construct correct URLs for both Previous and Next', () => { const searcher = new DuckDuckGoSearcher(); const html = readFileSync(join(__dirname, 'fixtures', 'ddg_with_both_buttons.html'), 'utf8'); const paginationInfo = searcher.parsePaginationInfo(html); // Test Previous URL construction if (paginationInfo.previousPageParams) { const prevParams = new URLSearchParams(paginationInfo.previousPageParams); const prevUrl = `https://duckduckgo.com/html/?${prevParams.toString()}`; console.log('Previous URL:', prevUrl); // Key parameters should be present expect(prevUrl).toContain('s=15'); expect(prevUrl).toContain('dc=-39'); expect(prevUrl).toContain('q=%22codehex%22'); } // Test Next URL construction if (paginationInfo.nextPageParams) { const nextParams = new URLSearchParams(paginationInfo.nextPageParams); const nextUrl = `https://duckduckgo.com/html/?${nextParams.toString()}`; console.log('Next URL:', nextUrl); // Key parameters should be present expect(nextUrl).toContain('s=80'); expect(nextUrl).toContain('dc=41'); expect(nextUrl).toContain('q=%22codehex%22'); } }); it('should verify critical differences between Previous and Next parameters', () => { const searcher = new DuckDuckGoSearcher(); const html = readFileSync(join(__dirname, 'fixtures', 'ddg_with_both_buttons.html'), 'utf8'); const paginationInfo = searcher.parsePaginationInfo(html); // Both should exist expect(paginationInfo.previousPageParams).toBeDefined(); expect(paginationInfo.nextPageParams).toBeDefined(); if (paginationInfo.previousPageParams && paginationInfo.nextPageParams) { // Critical differences expect(paginationInfo.previousPageParams.s).toBe('15'); expect(paginationInfo.nextPageParams.s).toBe('80'); expect(paginationInfo.previousPageParams.dc).toBe('-39'); expect(paginationInfo.nextPageParams.dc).toBe('41'); // Same values expect(paginationInfo.previousPageParams.q).toBe(paginationInfo.nextPageParams.q); expect(paginationInfo.previousPageParams.vqd).toBe(paginationInfo.nextPageParams.vqd); expect(paginationInfo.previousPageParams.kl).toBe(paginationInfo.nextPageParams.kl); } }); });