# Changelog of AliExpress Scraper - Products & Sellers (`parsebird/aliexpress-scraper`) Actor

- **URL**: https://apify.com/parsebird/aliexpress-scraper/changelog.md
- **Full Actor documentation**: https://apify.com/parsebird/aliexpress-scraper.md

## Changelog

### v1.2 — 2026-09-27

- Default `maxResults` raised from 20 to 60: one AliExpress search page holds 60 products and costs the same to load as 20, so a default run now returns a full page
- Default memory lowered to 1 GB (runs use under 500 MB), making runs cheaper

### v1.1 — 2026-09-02

- **Fixed near-empty listing records.** AliExpress moved the search grid out of
  the old signed `mtop` XHR and into the page's embedded `_dida_config_`
  data blob, so the actor was only recovering full details for ~1 in 20
  products (the rest came back as just an ID + URL). The grid is now read
  straight from that blob — title, sale/original price, discount %, star
  rating and "N sold" for every product on the page.
- **Big cost cut**: the browser now blocks images, media, fonts, and
  stylesheets plus common trackers. A default 5-record run was pulling ~9.5 MB
  of residential-proxy bandwidth (mostly the image-heavy homepage warm-up and
  search grid) — more than the run earned. Everything the actor extracts comes
  from JSON/DOM text, so this is a ~70% bandwidth cut with no data loss.
- Default `maxResults` raised from 5 to 20 — the fixed browser + session
  warm-up cost only pays off across a reasonable number of results.
- Lowered the actor's max memory to 3 GB so a run can't silently use 4 GB of
  compute.
- The `detailed-product` event is now only charged when the product page
  actually adds data the listing card doesn't have (specs, SKU variants,
  seller, or description) — not for a page visit that only re-confirmed the
  price.
- `detailedItems` fails fast: AliExpress currently captcha-walls the
  product-detail API, so the actor now detects that on the first page, keeps
  the page-level fields it can read, and stops visiting product pages after 2
  misses instead of paying ~4 MB of residential bandwidth per blocked page.

### v1.0.9 — 2026-08-29

- Added a wall-clock time budget: the run now stops itself and flushes collected
  records well before the platform timeout instead of being killed with a
  half-finished dataset
- Added a circuit breaker: when AliExpress's anti-bot blocks several pages in a
  row, the run stops early with a clear status message rather than retrying
  every remaining task to the timeout
- Capped total proxy-session rotations per run and lowered per-page retries from
  5 to 3
- When too many product pages are blocked, the run falls back to listing-only
  records for the remainder (no further detailed-product work)
- Runs that produce zero records now finish as FAILED with an explanatory
  message instead of SUCCEEDED with an empty dataset
- Lowered default run memory to 2 GB and default timeout to 60 minutes

### v1.0 — 2026-08-22

- Initial release
- Search AliExpress by keyword, or supply search, category, or product URLs directly (URL type is auto-detected)
- Listing records with title, price, currency, rating, orders, and image straight from the search results page
- Optional full product-detail enrichment: specifications, SKU variants, full image gallery, description, category path, and seller info
- Residential proxy support pinned to the selected ship-to country, with automatic session rotation on anti-bot blocks
- Regional site and currency selection
- Pay-per-event pricing (Result + Detailed Product) with spending limit enforcement