Adding distribution files

README.md

@@ -33,11 +33,13 @@ For a more detailed Demo, clone repo into Databricks and refer to the notebook [
 
 ## Usage: Read & Analyze a FHIR Bundle
 
-### 1. FHIR representations
+### 1. FHIR representations & Versions
 
 ``` python 
 from  dbignite.fhir_mapping_model import FhirSchemaModel
-fhir_schema = FhirSchemaModel()
+fhir_schema = FhirSchemaModel(schema_version="ci-build") #this is the default and contains latest, unfinalized changes 
+fhir_schema = FhirSchemaModel(schema_version="r4") #OR use this for FHIR R4
+fhir_schema = FhirSchemaModel(schema_version="r5") #OR use this for FHIR R5
 
 #list all supported FHIR resources
 sorted(fhir_schema.list_keys()) # ['Account', 'ActivityDefinition', 'ActorDefinition'...
@@ -67,7 +69,7 @@ sample_data = "./sampledata/*json"
 bundle = read_from_directory(sample_data)
 
 #Read all the bundles and parse
-bundle.entry()
+bundle.entry() #OR specify a schema version here -> bundle.entry(schemas =  FhirSchemaModel(schema_version="r4"))
 
 #Show the total number of patient resources in all bundles
 bundle.count_resource_type("Patient").show() 
@@ -106,6 +108,8 @@ bundle.count_resource_type("Patient").show()
 ```
 
 ## SQL on FHIR
+> [!TIP] 
+> For very large batches of FHIR, use bundle.entry().cache() before calling bulk_table_write for best performance
 
 ``` python
 %python
@@ -117,6 +121,7 @@ bundle.bulk_table_write(location="hls_healthcare.hls_dev"
   ,write_mode="overwrite"
   ,columns=["Patient", "Claim"]) #if columns is not specified, all columns of the dataframe are written (157 resources are written with default settings)
 ```
+
 ``` SQL
 %sql
 -- Select claim line detailed information
@@ -303,18 +308,22 @@ result.map(lambda x: json.loads(x)).foreach(lambda x: print(json.dumps(x, indent
     "resourceType": "Bundle",
     "entry": [
         {
-            "resourceType": "Claim",
-            "id": "CLM123"
+            "resource": {
+                "resourceType": "Claim",
+                "id": "CLM123"
+            }
         },
         {
-            "resourceType": "Patient",
-            "id": "PAT01",
-            "identifier": [
-                {
-                    "system": "<url of a hardcoded system reference>",
-                    "value": "COH123"
-                }
-            ]
+            "resource": {
+                "resourceType": "Patient",
+                "id": "PAT01",
+                "identifier": [
+                    {
+                        "system": "<url of a hardcoded system reference>",
+                        "value": "COH123"
+                    }
+                ]
+            }
         }
     ]
 }
@@ -323,18 +332,22 @@ result.map(lambda x: json.loads(x)).foreach(lambda x: print(json.dumps(x, indent
     "resourceType": "Bundle",
     "entry": [
         {
-            "resourceType": "Claim",
-            "id": "CLM345"
+            "resource": {
+                "resourceType": "Claim",
+                "id": "CLM345"
+            }
         },
         {
-            "resourceType": "Patient",
-            "id": "PAT02",
-            "identifier": [
-                {
-                    "system": "<url of a hardcoded system reference>",
-                    "value": "COH123"
-                }
-            ]
+            "resource": {
+                "resourceType": "Patient",
+                "id": "PAT02",
+                "identifier": [
+                    {
+                        "system": "<url of a hardcoded system reference>",
+                        "value": "COH123"
+                    }
+                ]
+            }
         }
     ]
 }

dbignite/fhir_mapping_model.py

@@ -9,28 +9,29 @@ class FhirSchemaModel:
     #
     # Class that manages access to FHIR resourceType ->  Spark Schema mapping
     #
-    def __init__(
-        self, fhir_resource_map: Optional[dict[str, StructType]] = None
-    ) -> None:
+    def __init__(self,
+                 fhir_resource_map: Optional[dict[str, StructType]] = None,
+                 schema_version = "ci-build") -> None:
+        #schema version can also be r4, or r5. This will change what directory resources are read from under dbignite/schemas/<directory>
+        self.schema_version = schema_version 
         self.__fhir_resource_map = (
             {
-                resource_type: FhirSchemaModel.__read_schema(schema_path)
-                for resource_type, schema_path in FhirSchemaModel.__get_schema_paths()
+                resource_type: FhirSchemaModel._read_schema(schema_path)
+                for resource_type, schema_path in self._get_schema_paths()
             }
             if fhir_resource_map is None
             else fhir_resource_map
         )
 
 
     @classmethod
-    def __read_schema(cls, path: str) -> StructType:
+    def _read_schema(cls, path: str) -> StructType:
         with open(path, "r") as f:
             return StructType.fromJson(json.load(f))
 
 
-    @classmethod
-    def __get_schema_paths(cls) -> list[tuple[str, str]]:
-        schema_dir = str(files("dbignite")) + "/schemas"
+    def _get_schema_paths(self) -> list[tuple[str, str]]:
+        schema_dir = str(files("dbignite")) + "/schemas/" + self.schema_version
         return [
             (os.path.splitext(p)[0], os.path.join(schema_dir, p))
             for p in os.listdir(schema_dir)
@@ -96,11 +97,10 @@ def us_core_fhir_resource_mapping(cls):
     #
     # Load supplied subset of FHIR resources into one dictionary
     #
-    @classmethod
-    def custom_fhir_resource_mapping(cls, resource_list: list[str]) -> "FhirSchemaModel":
+    def custom_fhir_resource_mapping(self, resource_list: list[str]) -> "FhirSchemaModel":
         custom_mapping = {
-            resource_type: FhirSchemaModel.__read_schema(schema_path)
-            for resource_type, schema_path in FhirSchemaModel.__get_schema_paths()
+            resource_type: FhirSchemaModel._read_schema(schema_path)
+            for resource_type, schema_path in self._get_schema_paths()
             if resource_type in resource_list
         }
         return FhirSchemaModel(fhir_resource_map=custom_mapping)
@@ -115,7 +115,7 @@ def list_keys(self):
     # Return all keys of FHIR Resources packaged
     #
     def list_packaged_data(self):
-        return [resource_type for resource_type, _ in FhirSchemaModel.__get_schema_paths()]
+        return [resource_type for resource_type, _ in self._get_schema_paths()]
 
     #
     # Allow searching at the metadata level contained in the spark schema

dbignite/schemas/README.md

@@ -1 +1 @@
-All FHIR Resources definitions from HL7 are translated in this directory. The version of schemas in this repository is [IG 6.0.0](https://hl7.org/fhir/us/core/history.html). To change schemas to a different version see [this](https://github.com/databricks-industry-solutions/json2spark-schema/blob/main/01_healthcare_FHIR_demo.py) public notebook.
+All FHIR Resources definitions from HL7 are translated in this directory. The versions of latest countinous integration (ci-build), r4, and r5 are built from [this](https://github.com/databricks-industry-solutions/json2spark-schema/blob/main/01_healthcare_FHIR_demo.py) notebook