NRL-1472 add validation for structured document contentTypes

Author: katebobyn-nhs

layer/nrlf/consumer/fhir/r4/model.py

@@ -246,11 +246,19 @@ class NRLFormatCode(Coding):
         Field(description="The system URL for the NRLF Format Code."),
     ]
     code: Annotated[
-        Literal["urn:nhs-ic:record-contact", "urn:nhs-ic:unstructured"],
+        Literal[
+            "urn:nhs-ic:record-contact",
+            "urn:nhs-ic:unstructured",
+            "urn:nhs-ic:structured",
+        ],
         Field(description="The code representing the format of the document."),
     ]
     display: Annotated[
-        Literal["Contact details (HTTP Unsecured)", "Unstructured Document"],
+        Literal[
+            "Contact details (HTTP Unsecured)",
+            "Unstructured Document",
+            "Structured Document",
+        ],
         Field(description="The display text for the code."),
     ]
 

layer/nrlf/core/constants.py

@@ -674,3 +674,16 @@ def coding_value(self):
     "https://fhir.nhs.uk/England/CodeSystem/England-NRLContentStability"
 )
 CONTENT_FORMAT_CODE_URL = "https://fhir.nhs.uk/England/CodeSystem/England-NRLFormatCode"
+CONTENT_FORMAT_CODE_MAP = {
+    "urn:nhs-ic:record-contact": "Contact details (HTTP Unsecured)",
+    "urn:nhs-ic:unstructured": "Unstructured Document",
+    "urn:nhs-ic:structured": "Structured Document",
+}
+
+ATTACHMENT_CONTENT_TYPES = {
+    "application/pdf",
+    "text/html",
+    "application/json",
+    "application/fhir+json",
+    "application/json+fhir",
+}

layer/nrlf/core/tests/test_validators.py

@@ -1157,6 +1157,38 @@ def test_validate_content_format_invalid_code_for_unstructured_document():
     }
 
 
+def test_validate_content_format_invalid_code_for_structured_document():
+    validator = DocumentReferenceValidator()
+    document_ref_data = load_document_reference_json("Y05868-736253002-Valid")
+
+    document_ref_data["content"][0]["format"] = {
+        "system": "https://fhir.nhs.uk/England/CodeSystem/England-NRLFormatCode",
+        "code": "urn:nhs-ic:record-contact",
+        "display": "Contact details (HTTP Unsecured)",
+    }
+
+    result = validator.validate(document_ref_data)
+
+    assert result.is_valid is False
+    assert result.resource.id == "Y05868-99999-99999-999999"
+    assert len(result.issues) == 1
+    assert result.issues[0].model_dump(exclude_none=True) == {
+        "severity": "error",
+        "code": "business-rule",
+        "details": {
+            "coding": [
+                {
+                    "system": "https://fhir.nhs.uk/ValueSet/Spine-ErrorOrWarningCode-1",
+                    "code": "UNPROCESSABLE_ENTITY",
+                    "display": "Unprocessable Entity",
+                }
+            ]
+        },
+        "diagnostics": "Invalid content format code: urn:nhs-ic:record-contact format code must be 'urn:nhs-ic:structured' for Structured Document attachments.",
+        "expression": ["content[0].format.code"],
+    }
+
+
 def test_validate_content_format_invalid_code_for_contact_details():
     validator = DocumentReferenceValidator()
     document_ref_data = load_document_reference_json("Y05868-736253002-Valid")

layer/nrlf/core/validators.py

@@ -7,7 +7,9 @@
 from nrlf.consumer.fhir.r4.model import RequestQueryCategory
 from nrlf.core.codes import SpineErrorConcept
 from nrlf.core.constants import (
+    ATTACHMENT_CONTENT_TYPES,
     CATEGORY_ATTRIBUTES,
+    CONTENT_FORMAT_CODE_MAP,
     ODS_SYSTEM,
     PRACTICE_SETTING_VALUE_SET_URL,
     REQUIRED_CREATE_FIELDS,
@@ -483,6 +485,21 @@ def _validate_content_format(self, model: DocumentReference):
                     diagnostics=f"Invalid content format code: {content.format.code} format code must be 'urn:nhs-ic:unstructured' for Unstructured Document attachments.",
                     field=f"content[{i}].format.code",
                 )
+            elif (
+                content.attachment.contentType
+                in {
+                    "application/json",
+                    "application/fhir+json",
+                    "application/json+fhir",
+                }
+                and content.format.code != "urn:nhs-ic:structured"
+            ):
+                self.result.add_error(
+                    issue_code="business-rule",
+                    error_code="UNPROCESSABLE_ENTITY",
+                    diagnostics=f"Invalid content format code: {content.format.code} format code must be 'urn:nhs-ic:structured' for Structured Document attachments.",
+                    field=f"content[{i}].format.code",
+                )
 
     def _validate_content_extension(self, model: DocumentReference):
         """
@@ -613,28 +630,23 @@ def _validate_practiceSetting(self, model: DocumentReference):
 
     def _validate_content(self, model: DocumentReference):
         """
-        Validate that the contentType is present and is either 'application/pdf' or 'text/html'.
+        Validate that the contentType is present and supported.
         """
         logger.log(LogReference.VALIDATOR001, step="content")
 
-        format_code_display_map = {
-            "urn:nhs-ic:record-contact": "Contact details (HTTP Unsecured)",
-            "urn:nhs-ic:unstructured": "Unstructured Document",
-        }
-
         for i, content in enumerate(model.content):
-            if content.attachment.contentType not in ["application/pdf", "text/html"]:
+            if content.attachment.contentType not in ATTACHMENT_CONTENT_TYPES:
                 self.result.add_error(
                     issue_code="business-rule",
                     error_code="UNPROCESSABLE_ENTITY",
-                    diagnostics=f"Invalid contentType: {content.attachment.contentType}. Must be 'application/pdf' or 'text/html'",
+                    diagnostics=f"Invalid contentType: {content.attachment.contentType}. Must be 'application/pdf', 'text/html' or 'application/fhir+json'",
                     field=f"content[{i}].attachment.contentType",
                 )
 
             # Validate NRLFormatCode
             format_code = content.format.code
             format_display = content.format.display
-            expected_display = format_code_display_map.get(format_code)
+            expected_display = CONTENT_FORMAT_CODE_MAP.get(format_code)
             if expected_display and format_display != expected_display:
                 self.result.add_error(
                     issue_code="business-rule",

layer/nrlf/producer/fhir/r4/model.py

@@ -1,6 +1,6 @@
 # generated by datamodel-codegen:
 #   filename:  swagger.yaml
-#   timestamp: 2025-02-07T14:10:35+00:00
+#   timestamp: 2025-05-22T18:53:21+00:00
 
 from __future__ import annotations
 
@@ -290,11 +290,19 @@ class NRLFormatCode(Coding):
         Field(description="The system URL for the NRLF Format Code."),
     ]
     code: Annotated[
-        Literal["urn:nhs-ic:record-contact", "urn:nhs-ic:unstructured"],
+        Literal[
+            "urn:nhs-ic:record-contact",
+            "urn:nhs-ic:unstructured",
+            "urn:nhs-ic:structured",
+        ],
         Field(description="The code representing the format of the document."),
     ]
     display: Annotated[
-        Literal["Contact details (HTTP Unsecured)", "Unstructured Document"],
+        Literal[
+            "Contact details (HTTP Unsecured)",
+            "Unstructured Document",
+            "Structured Document",
+        ],
         Field(description="The display text for the code."),
     ]
 

layer/nrlf/producer/fhir/r4/strict_model.py

@@ -1,6 +1,6 @@
 # generated by datamodel-codegen:
 #   filename:  swagger.yaml
-#   timestamp: 2025-02-07T14:10:37+00:00
+#   timestamp: 2025-05-22T18:53:21+00:00
 
 from __future__ import annotations
 
@@ -261,11 +261,19 @@ class NRLFormatCode(Coding):
         Field(description="The system URL for the NRLF Format Code."),
     ]
     code: Annotated[
-        Literal["urn:nhs-ic:record-contact", "urn:nhs-ic:unstructured"],
+        Literal[
+            "urn:nhs-ic:record-contact",
+            "urn:nhs-ic:unstructured",
+            "urn:nhs-ic:structured",
+        ],
         Field(description="The code representing the format of the document."),
     ]
     display: Annotated[
-        Literal["Contact details (HTTP Unsecured)", "Unstructured Document"],
+        Literal[
+            "Contact details (HTTP Unsecured)",
+            "Unstructured Document",
+            "Structured Document",
+        ],
         Field(description="The display text for the code."),
     ]