834 message parser meta data

containers/message-parser/app/main.py

@@ -7,6 +7,7 @@
 from app.utils import (
  load_parsing_schema,
  get_parsers,
+ get_metadata,
  convert_to_fhir,
  get_credential_manager,
  search_for_required_values,
@@ -69,6 +70,10 @@ class ParseMessageInput(BaseModel):
  "FHIR converter when conversion to FHIR is required.",
  default=None,
  )
+ include_metadata: Optional[Literal["true", "false"]] = Field(
+ description="Boolean to include metadata in the response.",
+ default=None,
+ )
  message: Union[str, dict] = Field(description="The message to be parsed.")
 
  @root_validator
@@ -200,7 +205,8 @@ async def parse_message_endpoint(
  )
  values.append(value)
  parsed_values[field] = values
-
+ if input.include_metadata == "true":
+ parsed_values = get_metadata(parsed_values, parsing_schema)
  return {"message": "Parsing succeeded!", "parsed_values": parsed_values}
 
 

containers/message-parser/app/utils.py

@@ -12,6 +12,7 @@
 from phdi.cloud.azure import AzureCredentialManager
 from phdi.cloud.core import BaseCredentialManager
 from phdi.cloud.gcp import GcpCredentialManager
+import re
 
 
 @cache
@@ -99,6 +100,34 @@ def get_parsers(extraction_schema: frozendict) -> frozendict:
  return frozendict(parsers)
 
 
+def get_metadata(parsed_values: dict, schema):
+ data = {}
+ for key, value in parsed_values.items():
+ if key not in schema:
+ data[key] = field_metadata(value=value)
+ else:
+ fhir_path = schema[key]["fhir_path"] if "fhir_path" in schema[key] else ""
+ match = re.search(r"resourceType\s*=\s*'([^']+)'", fhir_path)
+ resource_type = match.group(1) if match and match.group(1) else ""
+ data_type = schema[key]["data_type"] if "data_type" in schema[key] else ""
+ data[key] = field_metadata(
+ value=value,
+ fhir_path=fhir_path,
+ data_type=data_type,
+ resource_type=resource_type,
+ )
+ return data
+
+
+def field_metadata(value="", fhir_path="", data_type="", resource_type=""):
+ return {
+ "value": value,
+ "fhir_path": fhir_path,
+ "data_type": data_type,
+ "resource_type": resource_type,
+ }
+
+
 def search_for_required_values(input: dict, required_values: list) -> str:
  """
  Search for required values in the input dictionary and the environment.

containers/message-parser/tests/test_parse_message_endpoint.py

@@ -47,6 +47,51 @@
  },
 }
 
+expected_successful_response_with_meta_data = {
+ "message": "Parsing succeeded!",
+ "parsed_values": {
+ "first_name": {
+ "value": "John ",
+ "fhir_path": "Bundle.entry.resource.where(resourceType = 'Patient').name"
+ + ".first().given.first()",
+ "data_type": "string",
+ "resource_type": "Patient",
+ },
+ "last_name": {
+ "value": "doe",
+ "fhir_path": "Bundle.entry.resource.where(resourceType = 'Patient').name"
+ + ".first().family",
+ "data_type": "string",
+ "resource_type": "Patient",
+ },
+ "latitude": {
+ "value": None,
+ "fhir_path": "Bundle.entry.resource.where(resourceType = 'Patient')"
+ + ".address.extension.where"
+ + "(url='http://hl7.org/fhir/StructureDefinition/geolocation').extension."
+ + "where(url='latitude').valueDecimal",
+ "data_type": "float",
+ "resource_type": "Patient",
+ },
+ "longitude": {
+ "value": None,
+ "fhir_path": "Bundle.entry.resource.where(resourceType = 'Patient').address"
+ + ".extension.where"
+ + "(url='http://hl7.org/fhir/StructureDefinition/geolocation').extension"
+ + ".where(url='longitude').valueDecimal",
+ "data_type": "float",
+ "resource_type": "Patient",
+ },
+ "active_problems": {
+ "value": [],
+ "fhir_path": "Bundle.entry.resource.where(resourceType='Condition')"
+ + ".where(category.coding.code='problem-item-list')",
+ "data_type": "array",
+ "resource_type": "Condition",
+ },
+ },
+}
+
 expected_successful_response_floats = {
  "message": "Parsing succeeded!",
  "parsed_values": {
@@ -58,6 +103,51 @@
  },
 }
 
+expected_successful_response_floats_with_meta_data = {
+ "message": "Parsing succeeded!",
+ "parsed_values": {
+ "first_name": {
+ "value": "John ",
+ "fhir_path": "Bundle.entry.resource.where(resourceType = 'Patient').name"
+ + ".first().given.first()",
+ "data_type": "string",
+ "resource_type": "Patient",
+ },
+ "last_name": {
+ "value": "doe",
+ "fhir_path": "Bundle.entry.resource.where(resourceType = 'Patient').name"
+ + ".first().family",
+ "data_type": "string",
+ "resource_type": "Patient",
+ },
+ "latitude": {
+ "value": "34.58002",
+ "fhir_path": "Bundle.entry.resource.where(resourceType = 'Patient').address"
+ + ".extension"
+ + ".where(url='http://hl7.org/fhir/StructureDefinition/geolocation')"
+ + ".extension.where(url='latitude').valueDecimal",
+ "data_type": "float",
+ "resource_type": "Patient",
+ },
+ "longitude": {
+ "value": "-118.08925",
+ "fhir_path": "Bundle.entry.resource.where(resourceType = 'Patient').address"
+ + ".extension"
+ + ".where(url='http://hl7.org/fhir/StructureDefinition/geolocation')"
+ + ".extension.where(url='longitude').valueDecimal",
+ "data_type": "float",
+ "resource_type": "Patient",
+ },
+ "active_problems": {
+ "value": [],
+ "fhir_path": "Bundle.entry.resource.where(resourceType='Condition')"
+ + ".where(category.coding.code='problem-item-list')",
+ "data_type": "array",
+ "resource_type": "Condition",
+ },
+ },
+}
+
 
 def test_parse_message_success_internal_schema():
  test_request = {
@@ -82,6 +172,30 @@ def test_parse_message_success_internal_schema():
  assert actual_response2.json() == expected_successful_response_floats
 
 
+def test_parse_message_success_internal_schema_with_metadata():
+ test_request = {
+ "message_format": "fhir",
+ "parsing_schema_name": "test_schema.json",
+ "include_metadata": "true",
+ "message": fhir_bundle,
+ }
+
+ actual_response = client.post("/parse_message", json=test_request)
+ assert actual_response.status_code == 200
+ assert actual_response.json() == expected_successful_response_with_meta_data
+
+ test_request2 = {
+ "message_format": "fhir",
+ "include_metadata": "true",
+ "parsing_schema_name": "test_schema.json",
+ "message": fhir_bundle_w_float,
+ }
+
+ actual_response2 = client.post("/parse_message", json=test_request2)
+ assert actual_response2.status_code == 200
+ assert actual_response2.json() == expected_successful_response_floats_with_meta_data
+
+
 def test_parse_message_success_external_schema():
  request = {
  "message_format": "fhir",

containers/message-parser/tests/test_utils.py

@@ -11,6 +11,8 @@
  search_for_required_values,
  convert_to_fhir,
  freeze_parsing_schema,
+ field_metadata,
+ get_metadata,
 )
 from app.config import get_settings
 
@@ -170,3 +172,49 @@ def test_freeze_parsing_schema():
  for key in test_schema:
  for subkey in test_schema[key]:
  assert test_schema[key][subkey] == frozen_schema[key][subkey]
+
+
+def test_field_metadata():
+ expected_result = {
+ "value": "foo",
+ "fhir_path": "bar",
+ "data_type": "biz",
+ "resource_type": "baz",
+ }
+ assert (
+ field_metadata(
+ value="foo", fhir_path="bar", data_type="biz", resource_type="baz"
+ )
+ == expected_result
+ )
+ expected_result2 = {
+ "value": "",
+ "fhir_path": "",
+ "data_type": "",
+ "resource_type": "",
+ }
+ assert field_metadata() == expected_result2
+
+
+def test_get_metadata():
+ example_parsed_values = {"foo": "bar", "fiz": "biz", "baz": "Null"}
+ example_schema = {
+ "foo": {
+ "fhir_path": "Bundle.entry.resource.where(resourceType='Foo').biz",
+ "data_type": "string",
+ "nullable": False,
+ },
+ "baz": {},
+ }
+ expected_result = {
+ "foo": {
+ "value": "bar",
+ "fhir_path": "Bundle.entry.resource.where(resourceType='Foo').biz",
+ "data_type": "string",
+ "resource_type": "Foo",
+ },
+ "fiz": {"value": "biz", "fhir_path": "", "data_type": "", "resource_type": ""},
+ "baz": {"value": "Null", "fhir_path": "", "data_type": "", "resource_type": ""},
+ }
+ result = get_metadata(parsed_values=example_parsed_values, schema=example_schema)
+ assert result == expected_result