@apify-scrapers/shared/src/utils/platform-detector.js

Shared utilities and constants for Apify scrapers

import { PLATFORMS, PLATFORM_DOMAINS } from '../constants/platforms.js';

/**
 * Detect platform from URL
 * @param {string} url - The URL to analyze
 * @returns {string} Platform name (shopee, lazada, or unknown)
 */
export function detectPlatform(url) {
    if (!url) return PLATFORMS.UNKNOWN;
    
    const urlObj = new URL(url);
    const hostname = urlObj.hostname.toLowerCase();
    
    // Check Shopee domains
    if (PLATFORM_DOMAINS[PLATFORMS.SHOPEE].some(domain => hostname.includes(domain))) {
        return PLATFORMS.SHOPEE;
    }
    
    // Check Lazada domains
    if (PLATFORM_DOMAINS[PLATFORMS.LAZADA].some(domain => hostname.includes(domain))) {
        return PLATFORMS.LAZADA;
    }
    
    return PLATFORMS.UNKNOWN;
}

/**
 * Extract country code from URL
 * @param {string} url - The URL to analyze
 * @returns {string} Country code (SG, MY, ID, etc.)
 */
export function extractCountryFromUrl(url) {
    if (!url) return null;
    
    const urlObj = new URL(url);
    const hostname = urlObj.hostname.toLowerCase();
    
    // Extract country from subdomain or path
    const countryMatch = hostname.match(/\.([a-z]{2})\./);
    if (countryMatch) {
        return countryMatch[1].toUpperCase();
    }
    
    // Fallback: check common patterns
    if (hostname.includes('.sg')) return 'SG';
    if (hostname.includes('.my')) return 'MY';
    if (hostname.includes('.id')) return 'ID';
    if (hostname.includes('.ph')) return 'PH';
    if (hostname.includes('.th')) return 'TH';
    if (hostname.includes('.vn')) return 'VN';
    
    return null;
}

/**
 * Generate next page URL for pagination
 * @param {string} currentUrl - Current page URL
 * @param {number} nextPage - Next page number
 * @param {string} platform - Platform name
 * @returns {string|null} Next page URL or null if error
 */
export function getNextPageUrl(currentUrl, nextPage, platform) {
    try {
        const url = new URL(currentUrl);
        
        if (platform === PLATFORMS.LAZADA) {
            // Lazada uses different pagination structure
            url.searchParams.set('page', nextPage.toString());
        } else {
            // Shopee pagination
            url.searchParams.set('page', nextPage.toString());
        }
        
        return url.toString();
    } catch (error) {
        console.error('Error creating next page URL:', error);
        return null;
    }
}