DRAFT Validation Algorithm

  1. Collect initial schema set:
  • Get schema from resourceType
  • Get schemas from meta.profile
  • Get external schemas passed to validator

For the following patient resource:

{
  "resourceType": "Patient",
  "id": "some-patient-1",
  "meta": {
    "profile": [
      "http://hl7.org/fhir/us/core/StructureDefinition/us-core-patient"
    ]
  },
  "name": [{
    "family": "Smith",
    "given": ["John"]
  }]
}

The following initial schema-set is built:

["Base" "Resource" "DomainResource" "Patient" "us-core-patient"]
  1. Iterate over collected schema set to check top-level rules:
  • Check required and excluded directives
  • Check poly-roots directive
    • Type mutual exclusiveness check
    • Check that selected type is contained in choices
  • Evaluate fhirpath constraints
  1. Start iteration over input data to obtain key-value pairs:
  • For each obtained key, iterate over the schema set and obtain sub-schemas for this key

For the following patient resource and key name:

{
  "resourceType": "Patient",
  "id": "some-patient-1",
  "meta": {
    "profile": [
      "http://hl7.org/fhir/us/core/StructureDefinition/us-core-patient"
    ]
  },
  "name": [{
    "family": "Smith",
    "given": ["John"]
  }]
}

The following set of sub-schemas will be obtained:

["Element" "HumanName"]

  • If no schema is found, raise an unknown-key error

  • Do pre-recursive-descent checks:

    • Check shape - array or scalar
    • If array - check cardinality and slicings

  • Perform recursive call with new value and sub-schemas set. Preserve current path in data, preserve current path in schema, and preserve from which schema the recursive hop was made.

    3.1 Primitive extensions processing:

    • If key starts with _, try to find a key with an actual value (same key, but without _) and build a value wrapper that holds the Element sub-part (id, extension). The resulting wrapper contains value, sub-schemas, and Element sub-part (id, extension)
    • If key doesn't start with _, try to find the _ sub-part for this key and build a value wrapper
    • Perform a check that a primitive extension is added for a primitive value, not a complex one
  1. Slicing processing (Mark/Sweep like)
  • Iterate over sub-schemas and obtain slicing directive, which describes slices and their match directives.
  • Mark every collection element with slices that are matched against it. Some slices (such as constraining slices) may not contain match directives; we process them during the sweep phase.
  • Do sweep, find remaining unprocessed slices and check if they are valid (constraining slices), do cardinality checks, process @default slice.