Shopify Collection Product & Variant Exporter

Export products and variants from publicly accessible Shopify collection handles. The Actor reads only the public /collections/<handle>/products.json feed, checks robots.txt first, and enforces hard limits on pages, products and rows.

Published on Apify Store. This Actor is priced per output event. Check the current Pricing tab and run limits before starting. Start with the small collection-variant Example.

What makes it different from a general Shopify scraper

• Collection-scoped. You name the collections you need (for example new-arrivals, sale). It does not crawl a whole store.
• Variant-level rows. One row per variant with SKU, option values, price, compare-at price, a computed discount_percent and availability. Or switch to one summary row per product (price range, available variant count, best discount).
• Cross-collection de-duplication. A product listed in several collections of the same store is exported once.
• Bounded by design. Caps on products per collection, pages per collection, rows per run, response size, and a minimum delay between requests.
• Run summary. A SUMMARY record in the key-value store lists, per collection, the status (ok, empty, not_found, blocked_by_robots, error), pages fetched, duplicates skipped and whether the export was truncated by your limits.

Sample input

Run this small prefilled public-store sample, or replace it with a store and collection you are authorized to access.

{
    "shopDomain": "www.allbirds.com",
    "collectionHandles": ["all"],
    "exportMode": "variants",
    "maxProductsPerCollection": 2,
    "maxPagesPerCollection": 1,
    "maxTotalRows": 500,
    "pageSize": 2,
    "onlyAvailable": false,
    "dedupeAcrossCollections": true,
    "includeDescription": false,
    "requestDelaySeconds": 1
}

You can also pass full URLs (several stores allowed, 25 collections maximum per run):

{
    "collectionUrls": [
        "https://store-one.com/collections/summer",
        "https://store-two.com/collections/best-sellers"
    ]
}

Input reference

Field	Default	Meaning
shopDomain + collectionHandles	-	One store and its collection handles.
collectionUrls	-	Alternative: full https collection URLs.
exportMode	variants	variants or products.
maxProductsPerCollection	50	1-1000.
maxPagesPerCollection	4	1-10 feed requests per collection.
pageSize	50	1-250 products per feed page.
maxTotalRows	500	1-10000 rows for the whole run.
onlyAvailable	false	Skip unavailable variants/products.
dedupeAcrossCollections	true	Export each product once per store.
includeDescription / descriptionMaxChars	false / 500	Plain-text description, truncated.
requestDelaySeconds	1	0.5-30 seconds between requests to one store.

Output

Variant mode row (illustrative structure with placeholder values, not a real result):

{
    "row_type": "variant",
    "shop_domain": "your-store.com",
    "collection_handle": "sale",
    "product_id": 1001,
    "product_title": "Sample Jacket",
    "product_url": "https://your-store.com/products/sample-jacket",
    "variant_id": 2001,
    "variant_title": "M / Blue",
    "sku": "JKT-M-BLU",
    "option_values": {"Size": "M", "Color": "Blue"},
    "price": 80.0,
    "compare_at_price": 100.0,
    "discount_percent": 20.0,
    "available": true
}

Product mode adds variant_count, available_variant_count, min_price, max_price, min_compare_at_price, max_discount_percent and skus instead of variant fields.

Safety and compliance

• Only https URLs on standard port 443; credentials, IP addresses, local/reserved names and non-public DNS results are rejected.
• Redirects are never followed.
• robots.txt is fetched per store and applied, including wildcard rules. If it cannot be read (other than 404/410), the store is skipped (fail closed).
• Responses are size-capped; retries are limited to two, for HTTP 429/5xx only.
• No proxies, no login, no personal customer data. Only public catalog fields are read.
• You are responsible for complying with each store's terms and applicable law.

Billing note

The code emits one export-row event per row actually written to the dataset. Failed collections, robots blocks, empty results, skipped duplicates and errors write no rows and are therefore never charged. No payable pricing plan is configured in this project; any price must be set separately after real cost measurement.

Limitations and sample-size caveats

• Stores can disable or restrict the public feed; those collections return an error or empty status.
• The feed has no currency field; prices are the store's raw numbers (you must know the store currency). Prices shown are what the feed exposes, which may differ from checkout prices (taxes, markets, discounts).
• Output is a bounded snapshot. With default limits (50 products, 4 pages) a large collection is truncated; check truncated in SUMMARY. A single run is not evidence of full-catalog coverage or of price trends.
• Collection ordering and contents change over time; repeated runs may differ.
• A domain is checked against DNS before connecting, but a hostile DNS rebinding race cannot be fully excluded.
• Some feeds omit fields (for example barcodes, inventory counts); those are not exported.
• Product-type and tag values depend on how each merchant fills them in.
• Tests in this project run offline against synthetic fixtures. They do not prove behavior against any live store.

Local development

python -m unittest discover -s tests

(The tests need no network and do not import the Apify SDK.)