Klook Reviews Scraper — Activity Reviews & Ratings
Pricing
Pay per usage
Klook Reviews Scraper — Activity Reviews & Ratings
WIP: cloud transport-discovery probe, not the final actor yet.
Pricing
Pay per usage
Rating
0.0
(0)
Developer
Iakov Iudin
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
8 days ago
Last modified
Categories
Share
Klook Reviews Scraper — Activity Reviews & Ratings (DRAFT, BLOCKED)
Status: BLOCKED — see ACCESS_PROBE.md. klook.com fronts its activity
pages with a DataDome CAPTCHA challenge that curl_cffi, Camoufox, and — on
the issue #82 retry — patchright + real Chrome + Xvfb (the most-trusted free
browser fingerprint in the fleet) all fail to clear in-house; DataDome
escalates to the same interactive puzzle widget every time. No in-house
transport reaches review data. This directory holds the cloud
discovery-probe used to establish that, kept as the investigation record. It
is not a working scraper, does not pass the cloud gate, and is not
published.
Intended shape (champion-parity target, issue #82)
Input: urls[] (Klook activity page URLs), max_review_per_page, sort_by,
language, proxyConfiguration.
Output (10 champion-parity fields, per
knagymate/klook-reviews-scraper):
page_id, page_url, review_id, time, content_original, content_translate, score, score_description, images[], like_countparse_confidence, warnings.
Why it's blocked
See ACCESS_PROBE.md for the full rung-by-rung diagnosis. Short version:
every activity-page navigation (curl_cffi direct, Camoufox browser render,
patchright + real Chrome + Xvfb, and a bounded set of guessed
sibling-service JSON endpoints) is served a DataDome CAPTCHA instead of
content — patchright's real-Chrome fingerprint still gets escalated to the
same interactive puzzle widget, on 3 separate cloud runs with fresh proxy
sessions. The one confirmed-working keyless internal API (reused from the
sibling klook-activity-price-tracker actor) only returns a review
summary (star rating + count), not individual records.
What would unblock it
A managed-access service capable of clearing DataDome's interactive CAPTCHA — a COGS decision reserved for Bond, not made here. This is now the strongest BD candidate among the retry set: the free stack's best fingerprint (real Chrome + Xvfb) was actually tried and still escalated.
