Files
cannaiq/backend/new-scrapers/scrape-deeply-rooted-inventory-by-brand.ts
Kelly 66e07b2009 fix(monitor): remove non-existent worker columns from job_run_logs query
The job_run_logs table tracks scheduled job orchestration, not individual
worker jobs. Worker info (worker_id, worker_hostname) belongs on
dispensary_crawl_jobs, not job_run_logs.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-12-03 18:45:05 -07:00

184 lines
6.2 KiB
TypeScript

import { chromium as playwright } from 'playwright-extra';
import StealthPlugin from 'puppeteer-extra-plugin-stealth';
import fs from 'fs/promises';
import path from 'path';
playwright.use(StealthPlugin());
type Product = {
name: string;
brand?: string;
price?: number;
size?: string;
category?: string;
url?: string;
imageUrl?: string;
inStock: boolean;
};
type BrandGroup = {
brand: string;
products: Product[];
};
const TARGET_URL = 'https://dutchie.com/embedded-menu/AZ-Deeply-Rooted';
const OUTPUT_DIR = path.join(process.cwd(), 'scrape-output', 'deeply-rooted');
const JSON_PATH = path.join(OUTPUT_DIR, 'inventory-by-brand.json');
async function ensureDirs(): Promise<void> {
await fs.mkdir(OUTPUT_DIR, { recursive: true });
}
async function waitForCloudflare(page: any, maxWaitMs = 60000): Promise<boolean> {
const start = Date.now();
while (Date.now() - start < maxWaitMs) {
const title = await page.title().catch(() => '');
const content = await page.content().catch(() => '');
const challenge =
title.includes('Attention Required') ||
title.includes('Just a moment') ||
content.includes('challenge-platform') ||
content.includes('cf-challenge');
if (!challenge) return true;
await page.waitForTimeout(2000);
}
return false;
}
async function loadAllProducts(page: any): Promise<void> {
const maxScrolls = 40;
for (let i = 0; i < maxScrolls; i++) {
const beforeCount = await page.$$eval('[data-testid*="product"], [data-testid*="card"]', (els) => els.length);
await page.mouse.wheel(0, 1400);
await page.waitForTimeout(900);
const afterCount = await page.$$eval('[data-testid*="product"], [data-testid*="card"]', (els) => els.length);
if (afterCount <= beforeCount) break;
}
await page.evaluate(() => window.scrollTo({ top: 0 }));
}
async function extractProducts(page: any): Promise<Product[]> {
const script = `
(() => {
function parsePrice(text) {
if (!text) return undefined;
const match = text.match(/\\$?(\\d+(?:\\.\\d{1,2})?)/);
return match ? parseFloat(match[1]) : undefined;
}
function pickImage(card) {
const imgEl =
card.querySelector('img[src^="http"]') ||
card.querySelector('source[srcset]');
if (imgEl && imgEl.src && imgEl.src.startsWith('http')) {
return imgEl.src;
}
if (imgEl && imgEl.srcset) {
const first = imgEl.srcset.split(',')[0]?.trim().split(' ')[0];
if (first && first.startsWith('http')) return first;
}
const dataSrc = card.querySelector('img[data-src]')?.getAttribute('data-src');
if (dataSrc && dataSrc.startsWith('http')) return dataSrc;
return undefined;
}
const cards = Array.from(
document.querySelectorAll('[data-testid="product-list-item"], [data-testid="card-link"], [data-testid*="product-card"]')
);
return cards
.map((card) => {
const name =
card.querySelector('[data-testid="product-card-name"]')?.innerText?.trim() ||
card.querySelector('[data-testid="product-name"]')?.innerText?.trim() ||
card.querySelector('h3, h4')?.innerText?.trim() ||
(card.textContent || '').split('\\n').map((t) => t.trim()).find((t) => t.length > 3) ||
'';
const brand =
card.querySelector('[data-testid="product-card-brand"]')?.innerText?.trim() ||
card.querySelector('[data-testid="product-brand"]')?.innerText?.trim() ||
undefined;
const priceText =
card.querySelector('[data-testid="product-card-price"]')?.innerText ||
card.textContent ||
'';
const price = parsePrice(priceText);
const size =
card.querySelector('[data-testid*="size"]')?.innerText?.trim() ||
card.querySelector('[data-testid*="weight"]')?.innerText?.trim() ||
undefined;
const category =
card.querySelector('[data-testid*="category"]')?.innerText?.trim() ||
undefined;
const link = card.querySelector('a[href*="/product/"]');
const url = link?.href;
const imageUrl = pickImage(card);
const cardText = (card.textContent || '').toLowerCase();
const inStock = !(cardText.includes('sold out') || cardText.includes('out of stock'));
return { name, brand, price, size, category, url, imageUrl, inStock };
})
.filter((p) => p.name);
})();
`;
return page.evaluate(script);
}
function groupByBrand(products: Product[]): BrandGroup[] {
const map = new Map<string, Product[]>();
for (const p of products) {
const key = p.brand || 'Unknown';
if (!map.has(key)) map.set(key, []);
map.get(key)!.push(p);
}
return Array.from(map.entries()).map(([brand, prods]) => ({ brand, products: prods }));
}
async function main() {
await ensureDirs();
const browser = await playwright.launch({
headless: true,
args: ['--no-sandbox', '--disable-setuid-sandbox', '--disable-blink-features=AutomationControlled'],
});
const page = await browser.newPage({
viewport: { width: 1300, height: 900 },
userAgent:
'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36',
});
try {
console.log(`Navigating to ${TARGET_URL}...`);
await page.goto(TARGET_URL, { waitUntil: 'domcontentloaded', timeout: 90000 });
const cfOk = await waitForCloudflare(page, 60000);
if (!cfOk) throw new Error('Cloudflare challenge not passed in time');
await page.waitForSelector('[data-testid*="product"], [data-testid*="card"]', { timeout: 60000 }).catch(() => undefined);
await loadAllProducts(page);
const products = await extractProducts(page);
const grouped = groupByBrand(products);
await fs.writeFile(JSON_PATH, JSON.stringify(grouped, null, 2));
console.log(`Found ${products.length} products across ${grouped.length} brands`);
console.log(`Saved grouped inventory to ${JSON_PATH}`);
} catch (err) {
console.error('Inventory scrape failed:', err);
process.exitCode = 1;
} finally {
await page.context().browser()?.close();
}
}
main();