Current Active Pricing (04/11/2026): apify-actor-start $0.005 Free/Bronze, $0.004 Silver+; listing-scraped $0.0020 Free/Bronze, $0.0014 Silver+; detail-scraped $0.0040 Free/Bronze, $0.0032 Silver+, charged only when includeDetails is true and maxDetails above 0. Older entries below are history; where they disagree, this block wins. Silver+ rows cost $0.0014: 200,000 rows a month is $280.12 with 30 daily starts, under a $300 budget.
- Reinforced resilient pagination in
main.py: warn and continue to next page when a search page encounters a network error or missing __NEXT_DATA__ blob, preventing premature scan termination.
- Flagship SEO integration: 10 comprehensive sections, zero H1 tags, zero nested objects.
- 10 Public Tasks synchronized with custom SEO metadata and Google PAA FAQs.
- Windows UTF-8 encoding fixes across all test fixtures and utility scripts.
- Added secret protection for
stateKey parameter (isSecret: true).
- Added 100% real snapshot pre-warming script (
prewarm_snapshots.py) supporting all 10 public tasks with zero synthetic data.
- Standardized 93 flat columns preserved.
- README lists the ten ready-made searches with the input each one stores, and sample links carry ?v=38.
- Sample links carry a version tag, so an edge copy cannot serve an old file.
- Latitude and longitude come out in the right order even when the portal prints the pair reversed.
- Bedroom, price, area and rent-frequency filters are applied and declared in the input schema.
belowCommunityMedianPercent now carries the published sign: positive means cheaper than the median.
- README states the real phone coverage: 9/35 sale, 4/10 off-plan, 34/35 rent, and rent numbers mostly share one agency line.
- When the page prints no contact field, a phone number written in the listing description fills
agent1Phone.
- Sample CSV cells that hold an object or a list are JSON, so
json.loads parses them.
- Dataset width is now fixed at 93 keys in both modes:
detail always ships (empty in list mode) and internal bookkeeping flags no longer reach the dataset. A CSV header no longer moves between a 92-key list row and a 94-key detail row.
- The eight public sample files migrated to 93 columns.
- Detail mode:
includeDetails and maxDetails. Each enriched row ships the full description, amenities, agency licence, property reference, verification state, DLD validation link, developer, handover date and payment plan.
- New pay-per-event
detail-scraped: 0.0040 USD per enriched row, 0.0032 USD from Silver. Charged in detail mode only, already-collected rows included; zero when detail mode is off.
- Row budget is split evenly across the search URLs, so one URL cannot take the whole run.
- Tests: parser, detail page, billing, columns, OpSec and flow checks.
- Ninety-two columns: identity, price, size, location, project, seller, media, time, DLD and joined rows.
- The store gets
areaSqft so every row carries a six-part unit key.
- Snapshot then state batch: first seen, days on market and the other portals' price.
- Read pages through the UNBLOCKER proxy, parse
__NEXT_DATA__, no Algolia key.