Anjuke Real Estate Properties and Details Scraper

This Actor contains an offline parser for Anjuke second-hand property listings. It recognizes structured data and common listing-card HTML, retaining listing URLs, titles, rooms, area, price in 万/RMB, unit price when exposed, and source-page evidence.

Dataset rows require a stable listing identifier or listing URL plus at least one substantive property fact (such as rooms, community, floor, area, or price). A page title and link by themselves are rejected.

The current Anjuke user agreement prohibits commercial use of platform data without prior written consent, rejects crawlers, and prohibits scraping or other use without explicit written permission (agreement version effective 2025-04-10). No permission is recorded for this Actor. By default it makes zero source requests, finishes SUCCEEDED with collectionStatus: SKIPPED_POLICY, and writes no dataset rows. The checkbox is only an operator attestation; it does not grant permission. Do not enable collection unless current written permission covers it. The Actor does not solve verification challenges and does not follow redirects.

The OUTPUT summary uses only SUCCEEDED or FAILED for status and runStatus. It includes bounded, generic failureReasons and a policyReason when collection is skipped. A valid empty search completes successfully without placeholder rows. A source or parsing failure before any listing is saved reports FAILED; if a later request or storage write fails after a real listing was saved, the run remains SUCCEEDED with collectionStatus: PARTIAL and the saved-row count. Errors and access challenges are never written as dataset records.

Input

• authorizationConfirmed: default false; set true only if prior written permission explicitly covers automated collection and copying.
• startUrls: HTTPS Anjuke sale-listing URLs; the default target is Beijing resale listings.
• maxItems: maximum listing rows across all supplied pages.
• timeoutSecs: per-request timeout.

Offline validation

npm install
apify validate-schema
npm test

The parser tests use synthetic HTML and do not verify current source data. No live run is documented here because written permission has not been recorded. OUTPUT reports request count, real listing count, status, completion time, and generic failure reasons without diagnostic records.

Dependency remediation — October 3, 2026

This Actor uses native HTTP and parser code, with a small readable default-storage adapter and the pinned official Apify logger. The unused SDK/crawler/proxy/cache dependency graph was removed because its current http-cache-semantics advisory has no released fix. Cloud storage uses the documented Apify REST interface and bearer headers; local writes append indexed JSON without overwriting historical rows. Storage failures propagate and ambiguous POST writes are not retried. Native exit codes and source authorization/access gates are preserved. Node 22.19+ is required. The source remains unqualified until its existing permission/access prerequisites are met.

Relevant advisory: https://github.com/advisories/GHSA-ch52-4w7c-c8xp. Runtime regression checks: node --test test/platform.test.js.