NEWS

rxref 0.5.0

Added cache management helpers: rxref_cache(), rxref_cache_info(), rxref_cache_clear(), rxref_cache_use_memory(), rxref_cache_use_disk(), and rxref_cache_disable(). These helpers make it easier to inspect, clear, persist, or disable rxref's memoised RxNorm/RxClass API-response cache.
Revised vignettes to better distinguish use cases for rxref, cleaned up vignette data bundling and removed potentially confusing avoidance of API calls during vignette building from the vignettes themselves.

Improved API error-handling consistency across rxref. Centralized rxref API condition classes are now used more consistently for connection errors, rate limits, server errors, empty responses, malformed JSON, and other unsuccessful HTTP responses from the NLM.
Added internal optional-API handling helpers to return fallback values with informative warnings as opposed to silent API failures.
Updated NDC status retrieval, clinical attribute enrichment, historical clinical fallback, and product-search fallback logic to distinguish required API calls from optional enrichment calls.

Added support for retrieving active and historical RxNorm product concepts in product discovery workflows, including products_for_ingredients() and related search helpers.
Added concept_status support to allow users to request active concepts only or active plus historical RxNorm concepts, where supported.
Added exclude_products_with_ingredients(), a product-level filtering helper for removing RxNorm products that contain user-specified ingredients.
exclude_products_with_ingredients() supports fixed-dose combination products represented with semicolon-delimited ingredient names or RxCUIs, such as "atenolol; chlorthalidone".
Added safer internal handling for historical concept verification and lookups, reducing failures when RxNorm returns incomplete, empty, or unexpected API responses.

Improved retrieval of clinical product attributes for historical RxCUIs by attempting to recover dose form, dose form group, route, ingredient count, and ingredient metadata when available.
Improved route filtering workflows for products that include historical RxNorm concepts.
Improved fixed-dose combination formatting in get_clinical_attributes(). Ingredient fields such as ingredient_rxcui, ingredient_name, and ingredient_tty now use semicolon-delimited strings instead of pseudo-vector strings.
Improved consistency of product-level ingredient metadata to support cleaner downstream joins and filtering.
Improved reuse of cached API responses in product discovery, historical concept lookup, and clinical attribute workflows to reduce repeated calls to the RxNorm and RxClass APIs.
Added or improved progress reporting for longer-running product discovery workflows, including API retrieval steps where appropriate.
Updated route-specific vignette example datasets so package vignettes do not rely on live API calls during package checks.
Updated vignette example datasets to avoid duplicated .x / .y join columns and to use current, singular column naming conventions.
Updated README examples and package vignettes to reflect current function behavior, argument names, output columns, and recommended workflows.
Improved internal helper functions for normalizing product ingredient fields and handling fixed-dose combination products.

Fixed issues where historical or obsolete RxCUIs could return incomplete clinical attributes and be unintentionally dropped from route-filtered workflows.
Fixed cases where historical product concepts could produce incorrect or inflated ingredient counts.
Fixed validation and test coverage for concept_status arguments.
Fixed issues arising from incomplete or malformed historical concept lookups.
Fixed vignette data-generation issues that could create unexpected NA rows in pre-built example datasets.
Fixed CRAN submission issues related to API name formatting, web references, and package citation handling.

Updated package documentation to describe active and historical RxNorm concept support.
Updated documentation for product discovery, clinical attribute retrieval, route filtering, and fixed-dose combination handling.
Updated NEWS, README, vignettes, and CRAN-facing metadata for the 0.4.0 release.
Updated package citation handling to avoid errors when the package is not installed.

Removed the experimental set_backend() helper. rxref currently uses the RxNorm/RxClass APIs directly.