Paris.cl scraper: Chile product prices, offers and Paris card prices

This Paris.cl scraper reads the category pages of paris.cl, Chile's Paris department store, and returns one flat row per product: the price in Chilean pesos, the list ("normal") price, the Paris card (Tarjeta Cencosud) price, the percent off, the campaign tags (instalments without interest, free delivery), brand, seller, rating and review count, colour and size, the product link and the photo. Run it on a schedule and diff the dataset to get a Chilean retail price and promotion watch: which iPhone, TV or sofa dropped, which went on offer, and how much lower the Paris card price is.

No account, no API key, no login, no browser. One request returns one category's first 30 products.

Input

Field	Notes
categories	One or more Paris categories, one per line: the path below paris.cl (tecnologia, electro/television, mujer) or the page URL (https://www.paris.cl/electro/audio/audifonos/). Up to 200. Product pages (ending .html) and the search page are not supported.
maxItems	Stop after this many products across all categories (default 100, up to 6,000). A category page lists 30, so 30 reads one category.
proxyConfiguration	Apify proxy by default; a refused request is retried from a fresh address and, after three refusals in a row, through the residential proxy. Switch the proxy off, or supply your own, to use exactly that.

Example input:

{
  "categories": ["tecnologia", "electro/television", "electro/audio/audifonos"],
  "maxItems": 90
}

A product that two categories both list is returned once, under the first.

Output

Prices are whole Chilean pesos as integers; currency is CLP. One row is one product as the category page lists it, i.e. its main variant: a shirt sold in twelve sizes is one row, and variant_count says how many other variants exist. product_id is the unique key within a run.

Field	Notes
category, position	The category the row was read from and the place in its page, counting from 1. Paris' default order ("relevance") drifts a little between requests.
product_id, sku, name, brand	Paris' id for the product (the number or code that ends its address), the sku of the main variant, the name and the brand as Paris shows them.
url, image_url	The product page (https://www.paris.cl/<name>-<id>.html) and its first photo.
currency	CLP.
price	What the shopper pays today without the store card: the offer price when there is one, else the list price.
list_price	Paris' regular (list, "normal") price. Null when the page shows none.
card_price	The price with Paris' store card (Tarjeta Cencosud). Null when the product has none: 20 of the 30 products of tecnologia and 66 of the 150 read while building this actor had one.
on_sale, discount_percent, card_discount_percent	True when price is below list_price (false when equal, null when there is no list price), and how far price and card_price are below list_price in percent (10.53 is 10.53% off), as Paris states them. It is what the page shows, not a comparison against any price history. The percentages are null when the product has no offer or card price.
promo_tags	The campaign tags Paris attaches to the product, as sent (tag-10-cuotas-sin-interes-7, tag-despacho-gratis-sobre-19990-1P-13). They name promotions such as instalments without interest or free delivery; the page does not say what each one is worth. An empty list when there are none.
rating, review_count	The average star rating (1 to 5, two decimals) and the number of reviews. Rating is null for a product with no reviews, and review_count is then 0.
seller	Paris for Paris itself, otherwise the marketplace seller's name.
color, size, variant_count	The colour and size of the main variant (size is null on products that have none) and how many other variants the product has.
category_total, scraped_at	How many products Paris says the whole category holds (47201 for tecnologia on 2026-11-05; the page lists only the first 30), and when the row was read, UTC.

Example row (real values, 2026-11-05, tecnologia):

{
  "category": "tecnologia",
  "position": 2,
  "product_id": "837803999",
  "sku": "837803999",
  "name": "iPhone 16 128GB Negro Liberado",
  "brand": "Apple",
  "url": "https://www.paris.cl/iphone-16-128gb-negro-liberado-837803999.html",
  "image_url": "https://cl-dam-resizer.ecomm.cencosud.com/unsafe/adaptive-fit-in/792x1068/paris/837803999/variant/images/2cf11c09-019c-4272-9e00-a3c043f3c4ec/837803999-0000-001.jpg",
  "currency": "CLP",
  "price": 849990,
  "list_price": 949990,
  "card_price": 799990,
  "on_sale": true,
  "discount_percent": 10.53,
  "card_discount_percent": 15.79,
  "promo_tags": [
    "tag-10-cuotas-sin-interes-7",
    "tag-10-cuotas-sin-interes-10",
    "tag-10-cuotas-sin-interes-8",
    "tag-despacho-gratis-sobre-19990-1P-13"
  ],
  "rating": 4.44,
  "review_count": 45,
  "seller": "Paris",
  "color": "Único Color",
  "size": null,
  "variant_count": 0,
  "category_total": 47201,
  "scraped_at": "2026-11-05T09:15:42Z"
}

Honest nulls, and what the page does not carry

A field the page did not show is null, never 0 or ""; promo_tags is an empty list when a product has none.

• 30 products per category address, no second page. Paris' category page shows the first 30 products and loads more from the browser. Asking the page for ?page=2&sortby=relevance (the address its own "load more" link carries), or for another sortby, returned the first page again (measured from the Apify platform, 2026-11-05), so this actor cannot page. List several subcategories (electro/television, electro/audio, electro/audio/audifonos) to watch more products: each is its own 30.
• Order is Paris' "relevance", which moves between requests; sort the dataset by price or discount_percent yourself.
• Rows read from the schema.org copy. Every page carries its 30 products twice, in the page's own data and as schema.org markup. If Paris changes the first, the actor reads the second, which has the name, link, photo, brand, offer price and rating but no list price, card price, seller, colour, size or tags: those columns are null and the run log says how many pages were read that way.
• A category Paris does not have is HTTP 404 and is skipped at no charge.

Pricing

Pay per event:

• $0.002 per run (actor-start), charged only after your input is validated, so a run that fails on bad input costs nothing.
• $0.0021 per product written (product-scraped), i.e. $2.10 per 1,000 products. A row that is not written is not charged: a product already returned for an earlier category, an empty category and a category that fails are free.

How it fetches

paris.cl is a Next.js site behind CloudFront and AWS WAF, and a category page is 2.8 to 5.1 MB of HTML. From the machine this actor was built on every request, static files included, was answered HTTP 405 ("Human Verification"). From inside the Apify platform, with a Chrome TLS fingerprint, the page came back on 12 of 31 datacenter-proxy requests (the rest HTTP 403 from CloudFront) and 19 of 31 residential ones (most of the rest HTTP 405). The actor therefore starts on the Apify datacenter proxy, retries each refused request from a fresh address, and after three refusals in a row continues through the residential proxy (four tries) and stays there. A refusal page served with HTTP 200 is treated as a refusal as well.

The products are read from the page's own data (productData: regular and offer prices and the paymentMethod card price, all centAmount in whole pesos), not from the page's visible HTML. If Paris changes the page the run fails with a message, or reads the schema.org copy as described above, instead of returning empty rows.