- Replace deep imports with top-level omniread exports in tests - Ensure tests validate only the supported public API surface - Align HTML and PDF tests with documented library usage
- Add smart httpx MockTransport routing based on endpoint paths - Render HTML fixtures via Jinja templates populated from JSON data - Introduce explicit, typed HTML parsers for semantic and table-based content - Add end-to-end tests covering scraper → content → parser → Pydantic models - Enforce explicit output contracts and avoid default dict-based parsing
