cannaiq/backend/archive/debug-dutchie-selectors.ts

import { createStealthBrowser, createStealthContext, waitForPageLoad, isCloudflareChallenge, waitForCloudflareChallenge } from './src/utils/stealthBrowser';
import { getRandomProxy } from './src/utils/proxyManager';
import { pool } from './src/db/migrate';
import * as fs from 'fs/promises';

async function debugDutchieSelectors() {
  console.log('🔍 Debugging Dutchie page structure...\n');

  const url = 'https://dutchie.com/dispensary/sol-flower-dispensary';

  // Get proxy
  const proxy = await getRandomProxy();
  console.log(`Using proxy: ${proxy?.server || 'none'}\n`);

  const browser = await createStealthBrowser({ proxy: proxy || undefined, headless: true });

  try {
    const context = await createStealthContext(browser, { state: 'Arizona' });
    const page = await context.newPage();

    console.log(`Loading: ${url}`);
    await page.goto(url, { waitUntil: 'domcontentloaded', timeout: 60000 });

    // Check for Cloudflare
    if (await isCloudflareChallenge(page)) {
      console.log('🛡️  Cloudflare detected, waiting...');
      await waitForCloudflareChallenge(page, 60000);
    }

    await waitForPageLoad(page);

    // Wait for content
    await page.waitForTimeout(5000);

    console.log('\n📸 Taking screenshot...');
    await page.screenshot({ path: '/tmp/dutchie-page.png', fullPage: true });

    console.log('💾 Saving HTML...');
    const html = await page.content();
    await fs.writeFile('/tmp/dutchie-page.html', html);

    console.log('\n🔎 Looking for common React/product patterns...\n');

    // Try to find product containers by various methods
    const patterns = [
      // React data attributes
      'a[href*="/product/"]',
      '[data-testid*="product"]',
      '[data-cy*="product"]',
      '[data-test*="product"]',

      // Common class patterns
      '[class*="ProductCard"]',
      '[class*="product-card"]',
      '[class*="Product_"]',
      '[class*="MenuItem"]',
      '[class*="menu-item"]',

      // Semantic HTML
      'article',
      '[role="article"]',
      '[role="listitem"]',

      // Link patterns
      'a[href*="/menu/"]',
      'a[href*="/products/"]',
      'a[href*="/item/"]',
    ];

    for (const selector of patterns) {
      const count = await page.locator(selector).count();
      if (count > 0) {
        console.log(`✓ ${selector}: ${count} elements`);

        // Get details of first element
        try {
          const first = page.locator(selector).first();
          const html = await first.evaluate(el => el.outerHTML.substring(0, 500));
          const classes = await first.getAttribute('class');
          const testId = await first.getAttribute('data-testid');

          console.log(`  Classes: ${classes || 'none'}`);
          console.log(`  Data-testid: ${testId || 'none'}`);
          console.log(`  HTML preview: ${html}...`);
          console.log('');
        } catch (e) {
          console.log(`  (Could not get element details)`);
        }
      }
    }

    // Try to extract actual product links
    console.log('\n🔗 Looking for product links...\n');
    const links = await page.locator('a[href*="/product/"], a[href*="/menu/"], a[href*="/item/"]').all();

    if (links.length > 0) {
      console.log(`Found ${links.length} potential product links:`);
      for (let i = 0; i < Math.min(5, links.length); i++) {
        const href = await links[i].getAttribute('href');
        const text = await links[i].textContent();
        console.log(`  ${i + 1}. ${href}`);
        console.log(`     Text: ${text?.substring(0, 100)}`);
      }
    }

    // Check page title and URL
    console.log(`\n📄 Page title: ${await page.title()}`);
    console.log(`📍 Final URL: ${page.url()}`);

    // Try to find the main content container
    console.log('\n🎯 Looking for main content container...\n');
    const mainPatterns = ['main', '[role="main"]', '#root', '#app', '[id*="app"]'];
    for (const selector of mainPatterns) {
      const count = await page.locator(selector).count();
      if (count > 0) {
        console.log(`✓ ${selector}: found`);
        const classes = await page.locator(selector).first().getAttribute('class');
        console.log(`  Classes: ${classes || 'none'}`);
      }
    }

    console.log('\n✅ Debug complete!');
    console.log('📸 Screenshot saved to: /tmp/dutchie-page.png');
    console.log('💾 HTML saved to: /tmp/dutchie-page.html');

  } catch (error) {
    console.error('❌ Error:', error);
  } finally {
    await browser.close();
    await pool.end();
  }
}

debugDutchieSelectors();