Amazon Best Sellers Scraper, Amazon URLs from 9 different Amazon domains: .com, com.au, .in, co.jp, .de. .co.uk, .fr, .es and .it. Scrape the 100 top-selling items on Amazon
fix(parser): Correct comma-decimal and Japanese scale-first ratings across existing marketplaces; interpret decimal and lowercase K/M review counts and Japanese 万 counts by their numeric multiplier. Preserve raw prices and existing field shapes.
test(parser): Resolve all 22 regional numeric expected failures and retain historical TS captures with explicit, source-backed numeric corrections.
fix(proxy): Force Amazon UAE managed proxy routes to the US, including retries and fallback, while retaining AE language/currency and every other marketplace routing policy.
fix(proxy): Restart ranking GET/ACP 503 failures with a fresh proxy assignment, client, and connection pool within the configured page-attempt limit.
feat(locales): Add UAE, Saudi Arabia, Netherlands, Poland, and Turkey with regional currency, language cookies, proxy country, and ranking URL validation.
fix(locales): Interpret Turkish scale-first stars, Arabic decimal/grouping signs and bidi controls, and Polish/Turkish/Arabic review-count formats in localized parsing mode.
test(locales): Extend HTML/ACP, request/session, proxy routing, and live numeric-audit coverage for the five marketplaces.
feat(locales): Add Amazon Mexico, Singapore, and Brazil ranking URLs with MXN, SGD, and BRL currencies and regional request/proxy settings.
fix(locales): Parse comma-decimal stars and grouped or abbreviated review counts for the three new marketplaces; preserve raw price strings and existing-marketplace compatibility.
test(locales): Cover initial HTML and ACP fields, missing values, input validation, request language/currency, proxy country, and URLs with or without www .
1.0.1 - 2026-10-08
refactor(runtime): Rebuild the Actor in Python.
compatibility: Preserve existing inputs and outputs, so users can continue using the Actor without any migration.
0.1.10 - 2026-07-02
fix(nav): Restore children extraction for Amazon Best Sellers left navigation when the current node and links are wrapped by span.a-list-item in the latest DOM.
test(nav): Add fixture-based regression coverage for pages with child categories and leaf categories using the latest saved Amazon HTML.
fix(input): If categoryUrls is provided (non-empty), it takes precedence and urls is ignored (prevents urls prefills/defaults from overriding legacy input).
0.1.1
fix(nav): More general left-side category tree parsing; compatible with differing DOM structures between Best Sellers and New Releases (aria-current + sibling nodes), and fixes missing captures for deep parents.
fix(currency): Corrects the UK site currency code to GBP.
test: Adds regression tests for the category tree parser and cookie handling.
feat: All Categories page scraping with recursive category discovery.
fix: Request Timeout
0.0.35
fix: If there are fewer than 50 items on the first page, do not continue scraping the next page.