Chedraui Mexico Supermarket Price Scraper

Read prices from Chedraui (one of Mexico's large supermarket chains, on the VTEX commerce platform) by keyword or by department: the selling price, the list price and the price before a promotion in MXN, the discount percent, whether the product is in stock and how many, the unit it is sold by, EAN barcode, brand, category, image and the product URL, one flat row per SKU. Run it on a schedule and diff the dataset to get a Mexican supermarket price watch: what a litre of milk, a kilo of tomatoes or a television costs at Chedraui today against last week, and what went out of stock.

No account, no API key, no login, no browser. One request returns 50 products.

Input

Field	Notes
searchQueries	Search queries, one per line (leche, aceite, televisor). Spanish and brand names work. A SKU found by two queries is returned once, under the first.
categories	Departments or categories to read, by name or id, one per line. The 16 departments are 1 Supermercado, 2 Vinos y licores, 3 Bebes, 4 Farmacia, 5 Moda y belleza, 6 Mascotas, 7 Hogar y jardin, 8 Tecnologia, 9 Catalogo extendido, 10 Jugueteria, 11 Cuidado e higiene personal, 12 Belleza, 13 Electrodomesticos y linea blanca, 14 Papeleria y oficina, 15 Deportes, 16 Automoviles y motos. Sub-categories work: Supermercado/Lacteos y huevo, or its id 105.
maxItems	Rows to read per query, per category, or per query-and-category pair (default 100, up to 100,000).
inStockOnly	Skip SKUs Chedraui reports as unavailable (not returned, not charged). Off by default.
proxyConfiguration	Apify datacenter proxy by default.

Give queries, categories, or both. At least one is required. With both, every query is searched inside every category (leche in Supermercado), which is how you keep a loose word from matching a toy.

Picking a category

A category is matched by id (1, 105, 10501; ids are unique across the whole tree), by a path of ids (1/105), by name, or by a path of names (Supermercado/Lacteos y huevo). Names ignore case and accents. The actor reads Chedraui's live category tree (776 categories, three levels deep) to do it, so a category added tomorrow works without a new release, and a name that matches nothing or matches two categories fails the run before anything is charged, with the list of what exists. A bare name that several categories carry resolves to the shallowest one (Mascotas is department 6; there is another under Catalogo extendido), and the id is always exact.

Output

Prices are numbers in Mexican pesos; currency is always MXN.

Field	Notes
query, position	The query that first found the SKU (null when it was found by walking a category), and where its product sat in the result list of the request that returned it (1 is the first result). Sibling SKUs share a position.
product_id, sku_id	Chedraui's product id and the SKU id (VTEX itemId). sku_id is unique per row in a run.
name, sku_name	The product name and the SKU's full name.
ean, brand	The SKU's barcode (leading zeros kept) and the brand. Null when Chedraui sends none.
url, image_url	The https://www.chedraui.com.mx/<slug>/p product page, and the first photo.
category, category_path	The most specific category and its breadcrumb (["Supermercado", "Lacteos y huevo", "Leche"]).
price, list_price, price_without_discount	The selling price now (Price), the reference price before the promotion (ListPrice) and PriceWithoutDiscount. Null when Chedraui has no price for the SKU (see below).
on_sale, discount_percent	on_sale is true when list_price is above price. discount_percent is worked out from the two prices (one decimal place) and is null when the SKU is not on sale, never 0.
in_stock, stock_quantity	Chedraui's availability flag and AvailableQuantity for the chosen seller's offer. stock_quantity is 99999 for stock the seller does not count (190 of 315 offers in a sample, nearly all groceries); it is passed through as sent.
measurement_unit, unit_multiplier	VTEX's measurementUnit (un for a unit, kg for produce and meat sold by weight) and unitMultiplier, as sent and not interpreted.
seller_id, seller_name, sold_by_chedraui	The seller whose offer the row shows. Id 1 is Chedraui itself; any other is a marketplace seller (televisions and furniture mostly).

Example row (leche, position 1):

{
  "query": "leche",
  "position": 1,
  "product_id": "3855683",
  "sku_id": "3855683",
  "name": "Leche Bové Leche con Café Orgánica 330ml",
  "sku_name": "Leche Bové Leche con Café Orgánica 330ml",
  "ean": "7501639313538",
  "brand": "Bové",
  "url": "https://www.chedraui.com.mx/leche-bove-leche-con-cafe-organica-330ml-3855683/p",
  "image_url": "https://chedrauimx.vteximg.com.br/arquivos/ids/78650497/7501639313538_00.jpg?v=639271041376600000",
  "category": "Leche",
  "category_path": ["Supermercado", "Lacteos y huevo", "Leche"],
  "currency": "MXN",
  "price": 30.0,
  "list_price": 30.0,
  "price_without_discount": 30.0,
  "on_sale": false,
  "discount_percent": null,
  "in_stock": true,
  "stock_quantity": 99999,
  "measurement_unit": "un",
  "unit_multiplier": 1.0,
  "seller_id": "1",
  "seller_name": "Chedraui",
  "sold_by_chedraui": true
}

Why one row per SKU

VTEX nests a product over its SKUs, and the SKU is what has a barcode, a price and a stock level. On Chedraui that is almost always the same thing as a product: none of 301 products sampled (groceries, televisions, phones) had a second SKU. The row is still a SKU, so a product that does have several is not reported under one price.

Honest nulls

A field Chedraui did not send is null, never 0 or "".

• Unpriced SKUs. Chedraui sends Price: 0 and ListPrice: 0 for a SKU it cannot currently sell (16 of 315 offers in a sample). The actor reports a null price, not a price of zero pesos, and still returns the row with in_stock: false, because a stock watch needs it. Set inStockOnly to skip them. A few SKUs keep their price while flagged unavailable (a phone with stock 0 still lists 23,499); both are reported as sent.
• Several sellers. A marketplace SKU lists Chedraui's own seller beside the marketplace one, unpriced. The row shows the seller Chedraui marks as the default (when none is marked, the first seller with a price), because the default is the one the site sells from.
• Prices are the website's default, not a store's. Chedraui's catalogue API carries no store or postcode, so these are the prices the site shows before a delivery location is chosen. A store's own shelf price can differ.

Pricing

Pay per event:

• $0.002 per run (actor-start), charged only after your input is validated and your categories are found, so a run that fails on bad input costs nothing.
• $0.00056 per row written (product-scraped), i.e. $0.56 per 1,000 rows. A row that is not written is not charged.

How it fetches

It calls the VTEX catalog JSON behind Chedraui's own search and category pages (www.chedraui.com.mx/api/catalog_system/pub/products/search?ft=<query> and fq=C:/<department>/<category>/, _from=0&_to=49), 50 products per request, and the category tree at /api/catalog_system/pub/category/tree/3. Measured 2026-10-19: HTTP 200 or 206 and JSON from a plain home connection, an Apify datacenter exit and a residential exit. HTTP 206 Partial Content is the normal answer for a full page; the last page is 200. A refusal (403, 429, 202, 5xx), or a 200 that is not the product array, is treated as the address being blocked, and the actor rotates to a fresh one and retries. Chedraui occasionally answers a good request with a 500 (The operation was canceled., or a Solr queue overflow); that is retried the same way.

Things worth knowing, all measured:

• Depth ceiling: 2,550 products per search. VTEX refuses _from above 2,500 (HTTP 400) on this store as on any other, so one search can be read to 51 pages of 50 however many products it matches. The way past it is the category tree: when a category holds more than that and maxItems asks for more, the actor reads its sub-categories instead (Supermercado's 13 hold 20,560 products against its own 20,561, and Despensa's 10,401 split again). The split is logged. A search that is still over the ceiling with nothing left to split, such as a loose query like leche across the whole store, logs a warning so you can narrow it (add a brand or size, or pick a sub-category). At the default maxItems of 100 nothing is split: one page is enough.
• The match count moves. leche read 1,150 matches on one request and 1,151 on the next; Supermercado read 20,559 to 20,562 within a minute. The count in the log is informational.
• Page size. VTEX serves at most 50 per page (_to above 50 is HTTP 400), so the actor never asks for more.
• No matches. A query with no results is a normal empty answer ([]), not an error. It is logged and the run continues with the next one. A page past the end of a smaller result set is also [].
• Heavy pages. One page of 50 products is 0.8 MB of JSON, so a whole department of 20,000 products moves about 330 MB.
• Not offered. Sort orders, price filters and store selection. An empty query with no category would list the whole 66,000-product catalogue and is refused.