FUME language guide
FUME mappings are written in the FUME language.
That language gives you one continuous authoring model for selecting data, computing values, coordinating steps, and writing structured output across extraction, orchestration, and FHIR resource construction.
Mental model
- FUME uses JSONata-based expressions for selection, computation, and orchestration.
- FLASH is available when you need to write structured FHIR output.
- Comment rules apply throughout authoring, across both expression-heavy mappings and FLASH blocks.
- A FUME mapping can return extracted or transformed values directly, construct custom JSON structures, or produce FHIR-shaped, validated output.
Language foundations
The language guide starts with the shared rules that apply before any output-specific pattern:
Readers who are new to FUME should move through the pages in that order. Readers who already know the expression layer can jump directly into FLASH once the shared syntax rules are familiar.
Continue deeper
The language guide teaches the mental model and the core authoring patterns.
When you need exact syntax or execution detail, continue into the adjacent lookup pages:
- Path operators for navigation and selection behavior
- Comparison operators, Boolean operators, and Numeric operators for expression mechanics
- Async execution for array-constructor parallelism and async-capable mappings
- Functions for built-in helper syntax, categories, and execution badges
Use the guide to learn the model, then use those focused pages when you need precise syntax or exact helper behavior.