doordash

name	doordash
description	Proven scraping playbook for doordash.com. React SPA behind Cloudflare with server-rendered deal carousels on /browse/deals. Cloudflare challenge clears automatically with puppeteer-extra stealth (no captcha, no proxies, no login needed). Activate for any doordash.com target. Covers carousel selectors, price-token parsing, geo-fenced deals, and known virtualization pitfalls.
metadata	{"author":"torch","version":"1.0.0"}

DoorDash (doordash.com)

React SPA served through Cloudflare. Deal pages are server-side rendered with all items inlined in the initial HTML — don't bother with the GraphQL endpoints. A headed stealth Puppeteer session clears the Cloudflare challenge automatically; no captcha solver or proxy needed.

Detection

Signal	Value
CDN	Cloudflare (cf-ray, cf-mitigated, "Just a moment..." on raw curl)
Framework	React SPA with SSR bootstrap
Anti-bot	Cloudflare Managed Challenge — clears via stealth plugin, no CAPTCHA
Rendering	SSR for deals carousels; GraphQL for personalized/auth surfaces
Auth	Not required for /browse/deals
Geo	Results vary by geoIP — default US egress lands in SF 94107

Strategy

Puppeteer (stealth, headed)
  → wait for networkidle2
  → scroll to trigger lazy sections
  → parse rendered HTML with Cheerio

Do NOT hit the GraphQL endpoints:

• /graphql/homeLandingPageStores, getGeoByIP, etc. fire on page load but none return the deals feed.
• The typical facetFeed endpoint only fires on authenticated/personalized surfaces, not /browse/deals.
• The deals feed is inlined in the SSR HTML — parse that directly.

Stealth config that works

import puppeteer from "puppeteer-extra";
import StealthPlugin from "puppeteer-extra-plugin-stealth";
puppeteer.use(StealthPlugin());

const browser = await puppeteer.launch({
  headless: false,
  args: [
    "--no-sandbox",
    "--disable-setuid-sandbox",
    "--disable-blink-features=AutomationControlled",
  ],
});

const page = await browser.newPage();
await page.goto("https://www.doordash.com/browse/deals", { waitUntil: "networkidle2" });

Standard Chrome UA is fine — no rotation needed. Cloudflare clears on its own once JS runs.

Verify you're past the challenge before parsing:

const title = await page.title();
if (title !== "Deals | Doordash") {
  throw new Error(`Unexpected page title: ${title} — Cloudflare may still be challenging`);
}

Selectors

All selectors use data-testid attributes (stable):

Element	Selector
Carousel row	[data-testid="LegoStandardCarouselContainer"]
Carousel header link	First <a href> inside the carousel — text like "Frozen Deals", URL /convenience/category?store_id=<id>&l1_id=<dept>&filter_key=deals_deals
Carousel item	[data-testid="LegoFlexibleItemSquareContainer"]
Item product id	data-item-id attribute on the item
Item image	img[alt] when rendered; fallback picture > source[srcset]
Discount label	[data-testid^="percent_discount"] or [data-testid^="dollar_discount"]
Store CTA	[data-testid="ActionCard"] → href="/convenience/store/<id>"

Price parsing — CRITICAL

Prices are split across three separate text nodes inside each item. Do not regex a single string. Join the text node sequence:

['$', '3', '99', '$4.69', '15% off', 'Product Name']
  → sale = $3.99
  → original = $4.69
  → discount = "15% off"
  → name = last entry

Member-only deals look like:

['$', '7', '69', '$5.99 with store membership', 'Product Name']

The unit suffix ("ea", "lb") appears right after the cents token when present.

const items = $('[data-testid="LegoFlexibleItemSquareContainer"]').map((_, el) => {
  const textNodes = [];
  $(el).find('*').contents().each((_, n) => {
    if (n.type === 'text' && n.data.trim()) textNodes.push(n.data.trim());
  });

  // Join $, dollars, cents into sale_price
  const dollarIdx = textNodes.indexOf('$');
  const salePrice = dollarIdx >= 0
    ? parseFloat(`${textNodes[dollarIdx + 1]}.${textNodes[dollarIdx + 2]}`)
    : null;

  // Name is the last non-price text node
  const name = textNodes[textNodes.length - 1];

  // ... extract original_price, discount, etc.
  return { item_id: $(el).attr('data-item-id'), name, sale_price: salePrice };
}).get();

Anti-blocking summary

Technique	Needed?	Notes
Stealth plugin	✅ Yes	Required to clear Cloudflare challenge
Headed mode	✅ Yes	headless: false — challenge fails in headless
Realistic UA	✅ Standard Chrome	No rotation needed
Residential proxies	❌ No	Datacenter IPs work
CAPTCHA solver	❌ No	Cloudflare shows "Just a moment" but auto-clears
Cookie persistence	❌ No	Single-session works
Login	❌ No	/browse/deals is public

Data shape

{
  "source": "https://www.doordash.com/browse/deals",
  "scraped_at": "2026-04-09T12:00:00Z",
  "carousel_count": 12,
  "item_count": 118,
  "carousels": [
    {
      "department": "Drinks Deals",
      "store_id": "1741509",
      "l1_id": "751",
      "category_url": "https://www.doordash.com/convenience/category?store_id=1741509&l1_id=751&filter_key=deals_deals&show_store_header=true",
      "item_count": 10
    }
  ],
  "items": [
    {
      "item_id": "11547785152",
      "name": "Signature Select Refreshe Purified Drinking Water Bottles (16.9 oz x 24 ct)",
      "sale_price": 3.99,
      "sale_unit": null,
      "original_price": 4.69,
      "discount_label": "15% off",
      "member_price_note": null,
      "image_url": "https://img.cdn4dd.com/...",
      "department": "Drinks Deals",
      "store_id": "1741509",
      "l1_id": "751",
      "category_url": "https://www.doordash.com/convenience/category?..."
    }
  ]
}

Pagination

None on /browse/deals — one SSR response returns ~10 items × 12 carousels = ~120 items. To see more:

1. Open each carousel's category_url individually (store × department deals filter).
2. Set a delivery address first — the page is keyed on geoIP otherwise.

Do NOT click the in-carousel next buttons. DoorDash virtualizes the carousel and newly-injected items re-parent into sibling DOM nodes, so you end up attributing items to the wrong department/store.

Gotchas

1. Geo-fenced deals — results depend on geoIP. Default US egress lands in SF 94107. To target a specific region, drive the address picker in the UI before scraping.

2. Label vs reality mismatch — "Frozen Deals" / "Produce Deals" labels come from DoorDash's department slice, but the actual items inside can be mixed (the feed reorders by availability). Trust store_id + l1_id over the human label.

3. Lazy images — only ~35% of items have a rendered <img> at SSR time; the rest are still in <picture> placeholders. Fall back to picture > source[srcset].

4. Three-node price split — prices are split across $, dollars, cents text nodes. Joining them is required; regex on a single string will miss.

5. Member-only prices — look like $5.99 with store membership after the cents token instead of a strikethrough original.

6. Page title check — "Deals | Doordash" confirms you landed past Cloudflare. Any other title means the challenge didn't clear.

7. Carousel virtualization — as above, don't click next. Navigate via category_url instead.

Output

./output/doordash.json