Dev

[shahar-brd]

Introducing three entirely new services natively, fixes broken internal dataset tests, adds strong typing to dataset records, and thoroughly documents the SDK's architecture.

New Features & Services:

• Discover API (client.discover() & client.discoverTrigger()):
  • Brings native AI-powered web search with intent-based relevance ranking.
  • Implemented DiscoverJob and DiscoverResult with a generalized pollUntilReady loop (configured for 'done' status).
  • Added Zod schemas for strict input and response validation.
• Scraper Studio (client.scraperStudio):
  • Added native support for custom scrapers (DCA API).
  • Built a custom fetch-or-202 retry loop using a Deadline and transient NetworkError fallbacks (max 3 retries).
  • Added JobStatusResponseSchema with mixed-case field validation to normalize Bright Data API responses (Id/Status/Success_rate).
• Browser API (client.browser.getConnectUrl()):
  • Synchronous URL builder for CDP WebSockets (for Playwright/Puppeteer integration).
  • Features lazy credential validation (fails on access, not construction), port validation (capped at 65535), and automatic encodeURIComponent parsing for special characters.

Datasets & Maintenance:

• Dataset Probe Tests Fixed: Updated all 16 dataset probe test filters to use the correct structured format ({ name, operator, value }) expected by the /datasets/filter endpoint.
• Dataset Typing: Created AmazonBestSellerRecord interface mapping all 44 API fields, serving as the baseline pattern for typing the remaining 125 dataset records.

Architecture & Documentation:

• Architecture Docs: Added 15 behavioral traces across 6 categories (lifecycle, data transformation, error recovery, etc.) to map out the previously undocumented codebase.
• Scoped Planning Docs: Added full architectural planning and critic traces for the new services to /docs/scoped (Tasks 3, 4, 5, and 6).
• README Update: Added code examples for Discover API, Scraper Studio, and Browser API. Corrected dataset count (19 -> 126) and added Pinterest to the platforms list.

Testing & Validation:

• Added 49 new unit tests (19 Discover, 20 Scraper Studio, 10 Browser).
• Added 3 new Probe test scripts mirroring Python notebook scenarios (Discover, Scraper Studio, Browser API).
• Validated all 16 dataset filters against the live API.