MUJI Product Scraper & Price Tracker (US, Canada, Australia)

A MUJI scraper and MUJI price tracker for the three MUJI web shops that run on Shopify: www.muji.us (USD), muji.ca (CAD) and muji.com.au (AUD). Pick a storefront, read its whole catalogue or any collection (mens-new-arrivals, apparel-clearance, new-arrivals) and get one flat row per product: the current price and its range in the shop's own currency, the price before markdown, the largest discount in percent, whether it is in stock and how many sizes are, and every size or colour variant with its own SKU, price and stock flag. Run it on a schedule for MUJI price, markdown and new-product alerts: what went live this week, what went on sale since yesterday and by how much, and what the same product costs in each country.

No account, no API key, no login, no browser. One request returns 250 products; the whole catalogue is 20 requests on the US and Canadian shops and 8 on the Australian one. The Japanese shop (muji.com/jp) is a different platform and is not covered.

Input

Field	Notes
storefront	us (default), ca or au. Each is its own shop with its own catalogue, prices and currency; one run reads one storefront, so run it once per country you track.
collections	Collections of that storefront to read instead of the whole catalogue, one per line: the name from the address bar (mens-new-arrivals, apparel-clearance) or the link (https://www.muji.us/collections/mens-new-arrivals). A link from a different storefront is refused. Each is read in the shop's own order. A product found by two collections is returned once, under the first. Leave empty to read the whole catalogue, newest first. A collection name the shop does not have fails at no charge.
publishedAfter	Only products that went live after this moment: 24 hours, 7 days, 2 weeks, a date (2026-10-01) or a date-time (2026-10-01T12:00:00Z). On the whole catalogue the run stops at the first older product, since the feed is newest-published first. Leave empty for no cut-off.
onlyOnSale	Return only products where at least one variant is priced below its price before markdown (on_sale). Others are read, not returned and not charged, and do not count toward maxItems.
maxItems	Products to return per collection, or for the whole catalogue (default 250, up to 25,000). 250 per page.
proxyConfiguration	Apify datacenter proxy by default.

Example input, products that went live on the Canadian shop in the last week:

{
  "storefront": "ca",
  "publishedAfter": "7 days",
  "maxItems": 500
}

Example input, everything currently marked down in the US shop:

{
  "storefront": "us",
  "onlyOnSale": true,
  "maxItems": 1500
}

Output

Prices are numbers in the storefront's currency; currency on every row is USD, CAD or AUD. It is fixed per storefront, not read per run: each shop's own /meta.json and /cart.js said so on 2026-10-20. A field the shop did not send is null, never 0 or "".

Field	Notes
query, position	The shop URL (whole catalogue) or collection URL that first found the product, and its place in that list (1 is first, 250 per page). The whole catalogue is newest-published first.
store	www.muji.us, muji.ca or muji.com.au: the storefront the row came from.
product_id, handle, url	Shopify's product id (a string), its URL handle, and the product link on that storefront's own domain. One product is usually one colourway of a style, with its sizes as variants.
title, vendor, product_type	As the shop sends them (`Adjustable Handle Hard Shell Suitcase 75L
tags	The shop's own merchandising tags (season codes such as 25AW, promotions, categories), as sent. Not a standard vocabulary.
currency	USD, CAD or AUD, by storefront.
price, price_max	The lowest and highest current variant price. They differ when only some sizes or colours are reduced.
original_price, original_price_max	The lowest and highest price before markdown: per variant its compare_at_price when that is above its price, otherwise its price. Equal to the current price when nothing is reduced.
on_sale	True when at least one variant's compare_at_price is above its price.
discount_percent	The largest reduction of any variant, (1 - price / compare_at_price) * 100, one decimal place. Worked out from the prices, not sent. Null when nothing is reduced, never 0.
available, variant_count, available_variant_count	Whether any variant can be bought now, how many variants there are, and how many can be bought: the restock signal.
option_names	The product's options, e.g. ["Color", "Size"]; ["Title"] for a product with no options. The shops spell it Color, COLOR, Colour, SIZE as they please.
variants	Every variant: variant_id, title (Off White), sku, price, compare_at_price (null when the shop sends none), available, and options (the variant's values in option_names order).
created_at, published_at, updated_at	ISO 8601 times with the shop's offset. published_at is when the product went live, which is what publishedAfter reads; created_at can be many months older. updated_at is rewritten in bulk, so it is not a price-change time.
image_url, image_count	The first photo on cdn.shopify.com, as sent, and how many photos there are.

Example row (real values from the US shop, 2026-10-20; tags and variants shortened here to two and one entries):

{
  "query": "https://www.muji.us",
  "position": 81,
  "store": "www.muji.us",
  "product_id": "8560568729790",
  "title": "Adjustable Handle Hard Shell Suitcase 75L | Check-In",
  "handle": "adjustable-handle-hard-shell-suitcase-75l-check-in-eee46a5a",
  "url": "https://www.muji.us/products/adjustable-handle-hard-shell-suitcase-75l-check-in-eee46a5a",
  "vendor": "MUJI",
  "product_type": "Suitcases",
  "tags": [
    "25AW",
    "Best Seller Travel"
  ],
  "created_at": "2025-08-21T15:15:31-04:00",
  "published_at": "2026-10-10T09:23:49-04:00",
  "updated_at": "2026-10-19T16:22:32-04:00",
  "image_url": "https://cdn.shopify.com/s/files/1/0556/8066/3742/files/75L_White_Suitcase.jpg?v=1775757407",
  "image_count": 103,
  "currency": "USD",
  "price": 209.99,
  "price_max": 249.0,
  "original_price": 249.0,
  "original_price_max": 249.0,
  "on_sale": true,
  "discount_percent": 15.7,
  "available": false,
  "variant_count": 5,
  "available_variant_count": 0,
  "option_names": [
    "Color"
  ],
  "variants": [
    {
      "variant_id": "45329538580670",
      "title": "Off White",
      "sku": "4550584949944",
      "price": 209.99,
      "compare_at_price": 249.0,
      "available": false,
      "options": [
        "Off White"
      ]
    }
  ]
}

Honest nulls and what is not here

• No ratings, reviews or stock counts. The catalogue feed carries none. available says a variant can be bought, nothing more.
• discount_percent is null when nothing is reduced. A compare_at_price equal to or below the price is not a reduction: 7 US variants send one equal to the price and 118 send one below it (a US sweatpant at 39.90 with a compare_at_price of 29.90), and neither counts as a reduction.
• sku is null where the shop sends none (143 Australian variants).
• A price of 0.0 is the shop's. One US and Canadian product (a skincare set) is priced 0.00 in the feed; it is returned as sent.

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.
• $0.00175 per product row written (product-scraped), i.e. $1.75 per 1,000 products. A row that is not written is not charged: rows filtered by onlyOnSale or publishedAfter, and repeats of a product another collection already returned, are free.

How it fetches

It reads /products.json, the public catalogue feed every Shopify shop serves (and /collections/<name>/products.json for a collection). Measured 2026-10-20: HTTP 200 JSON, 1.6 to 3.2 MB per 250 products, no key or login needed. A refusal (403, 429, 430, 5xx), a 200 that is not the feed, or a dropped connection is treated as the address being blocked, and the actor rotates to a fresh one and retries.

Things worth knowing, all measured on 2026-10-20 by walking each whole feed:

Storefront	Host	Currency	Products	Pages	Reduced	Published in the last 7 days
us	www.muji.us	USD	4,896	20	933	79
ca	muji.ca	CAD	4,964	20	745	86
au	muji.com.au	AUD	1,786	8	145	36

• Three shops, three catalogues. Only 2,383 handles are on both the US and Canadian shops, 76 on both the US and Australian ones and 42 on all three, so rows are never merged across storefronts and store and currency are on every row.
• The US shop prices by country; the actor pins US. www.muji.us ships to 44 countries and re-prices its feed for the visitor (a sneaker 79.90 for the US, 117.00 for Canada). Every request carries localization=US, so USD holds whatever address the run exits from. The Canadian and Australian shops ship to one country each and priced identically with the cookie, a wrong one or none.
• 250 per page is the ceiling. A larger limit is silently capped at 250.
• The end is an empty page. After the last page the feed answers HTTP 200 with {"products":[]} (US page 21, Canada page 21, Australia page 9); the actor stops there.
• Newest-published first. published_at never increased down any of the three feeds, which is what makes publishedAfter stop early. created_at is not in order.
• Links use the shop's own domain. www.muji.ca and www.muji.com.au answer the feed but redirect every product page to muji.ca and muji.com.au, so rows link there.
• Collections differ per storefront. The US shop has mens-new-arrivals (100 by its count), apparel-clearance (228) and many *-sale lists; Canada has mens-new-arrivals and home-clearance; Australia has new-arrivals (62) and mens-sale. A collection's count includes products that are not published, so its feed can hold fewer.
• A *-new-arrivals collection is curated by hand, not by date. The first product of the US mens-new-arrivals was published on 2026-07-10 while the whole catalogue's newest was from 2026-10-16. For new products use publishedAfter on the whole catalogue, not a new-arrivals collection.
• An unknown collection looks empty. The feed answers an unknown name with an empty list; the actor asks the collection's own record, and a name that does not exist fails instead of being reported as empty.

If MUJI changes its feed, the run fails with a message instead of returning empty rows.