Experimental XML and YAML serialization and deserialization supports. See [Advanced Usages] section!
Powered by pydantic, all FHIR Resources are available as python class with built-in
data validation, faster in performance and optionally orjson support has been included as a performance booster! Written in modern python.
- Easy to construct, easy to extended validation, easy to export.
- By inheriting behaviour from pydantic, compatible with ORM.
- Full support of FHIR® Extensibility for Primitive Data Types are available.
- Previous release of FHIR® Resources are available.
- Free software: BSD license
FHIR® (Release R4, version 4.0.1) is available as default. Also previous versions are available as Python sub-package
(each release name string becomes sub-package name, i.e STU3 ).
Available Previous Versions:
STU3(3.0.2)DSTU2(1.0.2) [see issue#13][don't have full tests coverage]
Just a simple pip install fhir.resources or easy_install fhir.resources is enough. But if you want development
version, just clone from https://github.com/nazrulworld/fhir.resources and pip install -e .[dev].
Example: 1: Construct Resource Model object:
>>> from fhir.resources.organization import Organization
>>> from fhir.resources.address import Address
>>> data = {
... "id": "f001",
... "active": True,
... "name": "Acme Corporation",
... "address": [{"country": "Switzerland"}]
... }
>>> org = Organization(**data)
>>> org.resource_type == "Organization"
True
>>> isinstance(org.address[0], Address)
True
>>> org.address[0].country == "Switzerland"
True
>>> org.dict()['active'] is True
True
Example: 2: Resource object created from json string:
>>> from fhir.resources.organization import Organization
>>> from fhir.resources.address import Address
>>> json_str = '''{"resourceType": "Organization",
... "id": "f001",
... "active": True,
... "name": "Acme Corporation",
... "address": [{"country": "Switzerland"}]
... }'''
>>> org = Organization.parse_raw(json_str)
>>> isinstance(org.address[0], Address)
True
>>> org.address[0].country == "Switzerland"
True
>>> org.dict()['active'] is True
True
Example: 3: Resource object created from json object(py dict):
>>> from fhir.resources.patient import Patient
>>> from fhir.resources.humanname import HumanName
>>> from datetime import date
>>> json_obj = {"resourceType": "Patient",
... "id": "p001",
... "active": True,
... "name": [
... {"text": "Adam Smith"}
... ],
... "birthDate": "1985-06-12"
... }
>>> pat = Patient.parse_obj(json_obj)
>>> isinstance(pat.name[0], HumanName)
True
>>> pat.birthDate == date(year=1985, month=6, day=12)
True
>>> pat.active is True
True
Example: 4: Construct Resource object from json file:
>>> from fhir.resources.patient import Patient
>>> import os
>>> import pathlib
>>> filename = pathlib.Path("foo/bar.json")
>>> pat = Patient.parse_file(filename)
>>> pat.resource_type == "Patient"
True
Example: 5: Construct resource object in python way:
>>> from fhir.resources.organization import Organization
>>> from fhir.resources.address import Address
>>> json_obj = {"resourceType": "Organization",
... "id": "f001",
... "active": True,
... "name": "Acme Corporation",
... "address": [{"country": "Switzerland"}]
... }
>>> org = Organization.construct()
>>> org.id = "f001"
>>> org.active = True
>>> org.name = "Acme Corporation"
>>> org.address = list()
>>> address = Address.construct()
>>> address.country = "Switzerland"
>>> org.address.append(address)
>>> org.dict() == json_obj
True
Note
Please note that due to the way the validation works, you will run into issues if you are using construct() to create
resources that have more than one mandatory field. See this comment in issue#56 for details.
Example: 4: Using Resource Factory Function:
>>> from fhir.resources import construct_fhir_element
>>> json_dict = {"resourceType": "Organization",
... "id": "mmanu",
... "active": True,
... "name": "Acme Corporation",
... "address": [{"country": "Switzerland"}]
... }
>>> org = construct_fhir_element('Organization', json_dict)
>>> org.address[0].country == "Switzerland"
True
>>> org.dict()['active'] is True
True
Example: 5: Auto validation while providing wrong datatype:
>>> try:
... org = Organization({"id": "fmk", "address": ["i am wrong type"]})
... raise AssertionError("Code should not come here")
... except ValueError:
... pass
It is possible to add comments inside json like xml, but need to follow some convention, what is suggested by Grahame Grieve; is implemented here.
Also it is possible to generate json string output without comments.
Examples:
>>> observation_str = b"""{
... "resourceType": "Observation",
... "id": "f001",
... "fhir_comments": [
... " a specimen identifier - e.g. assigned when the specimen was taken by the orderer/placer use the accession number for the filling lab ",
... " Placer ID "
... ],
... "text": {
... "fhir_comments": [
... " a specimen identifier - e.g. assigned when the specimen was taken by the orderer/placer use the accession number for the filling lab ",
... " Placer ID "
... ],
... "status": "generated",
... "div": "<div xmlns=\"http://www.w3.org/1999/xhtml\">.........</div>"
... },
... "identifier": [
... {
... "use": "official",
... "system": "http://www.bmc.nl/zorgportal/identifiers/observations",
... "value": "6323"
... }
... ],
... "status": "final",
... "_status": {
... "fhir_comments": [
... " EH: Note to balloters - lots of choices for whole blood I chose this. "
... ]
... },
... "code": {
... "coding": [
... {
... "system": "http://loinc.org",
... "code": "15074-8",
... "display": "Glucose [Moles/volume] in Blood"
... }
... ]
... },
... "subject": {
... "reference": "Patient/f001",
... "display": "P. van de Heuvel"
... },
... "effectivePeriod": {
... "start": "2013-04-02T09:30:10+01:00"
... },
... "issued": "2013-04-03T15:30:10+01:00",
... "performer": [
... {
... "reference": "Practitioner/f005",
... "display": "A. Langeveld"
... }
... ],
... "valueQuantity": {
... "value": 6.3,
... "unit": "mmol/l",
... "system": "http://unitsofmeasure.org",
... "code": "mmol/L"
... },
... "interpretation": [
... {
... "coding": [
... {
... "system": "http://terminology.hl7.org/CodeSystem/v3-ObservationInterpretation",
... "code": "H",
... "display": "High"
... }
... ]
... }
... ],
... "referenceRange": [
... {
... "low": {
... "value": 3.1,
... "unit": "mmol/l",
... "system": "http://unitsofmeasure.org",
... "code": "mmol/L"
... },
... "high": {
... "value": 6.2,
... "unit": "mmol/l",
... "system": "http://unitsofmeasure.org",
... "code": "mmol/L"
... }
... }
... ]
... }"""
>>> from fhir.resources.observation import Observation
>>> obj = Observation.parse_raw(observation_str)
>>> "fhir_comments" in obj.json()
>>> # Test comments filtering
>>> "fhir_comments" not in obj.json(exclude_comments=True)
In some cases, implementers might
find that they do not have appropriate data for an element with minimum cardinality = 1.
In this case, the element must be present, but unless the resource or a profile on it has made the
actual value of the primitive data type mandatory, it is possible to provide an extension that
explains why the primitive value is not present.
Example (required intent element is missing but still valid because of extension):
>>> json_str = b"""{
... "resourceType": "MedicationRequest",
... "id": "1620518",
... "meta": {
... "versionId": "1",
... "lastUpdated": "2020-10-27T11:04:42.215+00:00",
... "source": "#z072VeAlQWM94jpc",
... "tag": [
... {
... "system": "http://www.alpha.alp/use-case",
... "code": "EX20"
... }
... ]
... },
... "status": "completed",
... "_intent": {
... "extension": [
... {
... "url": "http://hl7.org/fhir/StructureDefinition/data-absent-reason",
... "valueCode": "unknown"
... }
... ]
... },
... "medicationReference": {
... "reference": "Medication/1620516",
... "display": "Erythromycin 250 MG Oral Tablet"
... },
... "subject": {
... "reference": "Patient/1620472"
... },
... "encounter": {
... "reference": "Encounter/1620506",
... "display": "Follow up encounter"
... },
... "authoredOn": "2018-06-16",
... "requester": {
... "reference": "Practitioner/1620502",
... "display": "Dr. Harold Hippocrates"
... },
... "reasonReference": [
... {
... "reference": "Condition/1620514",
... "display": "Otitis Media"
... }
... ],
... "dosageInstruction": [
... {
... "text": "250 mg 4 times per day for 10 days",
... "timing": {
... "repeat": {
... "boundsDuration": {
... "value": 10,
... "unit": "day",
... "system": "http://unitsofmeasure.org",
... "code": "d"
... },
... "frequency": 4,
... "period": 1,
... "periodUnit": "d"
... }
... },
... "doseAndRate": [
... {
... "doseQuantity": {
... "value": 250,
... "unit": "mg",
... "system": "http://unitsofmeasure.org",
... "code": "mg"
... }
... }
... ]
... }
... ],
... "priorPrescription": {
... "reference": "MedicationRequest/1620517",
... "display": "Amoxicillin prescription"
... }
... }"""
>>> from fhir.resources.medicationrequest import MedicationRequest
>>> obj = MedicationRequest.parse_raw(json_str)
>>> "intent" not in obj.dict()
fhir.resources is providing the extensive API to create and attach custom validator into any model. See more about root validator
Some convention you have to follow though, while creating a root validator.
- Number of arguments are fixed, as well as names are also. i.e
(cls, values). - Should return
values, unless any exception need to be raised. - Validator should be attached only one time for individual Model. Update [from now, it's not possible to attach multiple time same name validator on same class]
Example 1: Validator for Patient:
from typing import Dict
from fhir.resources.patient import Patient
import datetime
def validate_birthdate(cls, values: Dict):
if not values:
return values
if "birthDate" not in values:
raise ValueError("Patient's ``birthDate`` is required.")
minimum_date = datetime.date(2002, 1, 1)
if values["birthDate"] > minimum_date:
raise ValueError("Minimum 18 years patient is allowed to use this system.")
return values
# we want this validator to execute after data evaluating by individual field validators.
Patient.add_root_validator(validate_gender, pre=False)
Example 2: Validator for Patient from Validator Class:
from typing import Dict
from fhir.resources.patient import Patient
import datetime
class MyValidator:
@classmethod
def validate_birthdate(cls, values: Dict):
if not values:
return values
if "birthDate" not in values:
raise ValueError("Patient's ``birthDate`` is required.")
minimum_date = datetime.date(2002, 1, 1)
if values["birthDate"] > minimum_date:
raise ValueError("Minimum 18 years patient is allowed to use this system.")
return values
# we want this validator to execute after data evaluating by individual field validators.
Patient.add_root_validator(MyValidator.validate_gender, pre=False)
important notes It is possible add root validator into any base class like DomainResource.
In this case you have to make sure root validator is attached before any import of derived class, other
than validator will not trigger for successor class (if imported before) by nature.
fhir.resources is providing API for enum constraint for each field (where applicable), but it-self doesn't
enforce enum based validation! see discussion here.
If you want to enforce enum constraint, you have to create a validator for that.
Example: Gender Enum:
from typing import Dict
from fhir.resources.patient import Patient
def validate_gender(cls, values: Dict):
if not values:
return values
enums = cls.__fields__["gender"].field_info.extra["enum_values"]
if "gender" in values and values["gender"] not in enums:
raise ValueError("write your message")
return values
Patient.add_root_validator(validate_gender, pre=True)
fhir.resources is also providing enum like list of permitted resource types through field property enum_reference_types.
You can get that list by following above (Enum) approaches resource_types = cls.__fields__["managingOrganization"].field_info.extra["enum_reference_types"]
orjson is one of the fastest Python library for JSON and is more correct than the standard json library (according to their docs).
Good news is that fhir.resource has an extensive support for orjson and it's too easy to enable it automatically. What you need to do, just make orjson as your project dependency!
pydantic Field Type Support
All available fhir resources (types) can be use as pydantic's Field's value types. See issue#46 Support for FastAPI pydantic response models.
The module fhirtypes.py contains all fhir resources related types and should trigger validator automatically.
There are a lots of discussion here here i.) https://bit.ly/360HksL ii.) https://bit.ly/3o1fZgl about the length of Resource.Id's value.
Based on those discussions, we recommend that keep your Resource.Id size within 64 letters (for the seek of intercompatibility with third party system), but we are also providing freedom
about the length of Id, in respect with others opinion that 64 chr length is not sufficient. fhirtypes.Id.configure_constraints()
is offering to customize as your own requirement.
- Examples::
>>> from fhir.resources.fhirtypes import Id >>> Id.configure_constraints(min_length=16, max_length=128)
Note: when you will change that behaviour, that would impact into your whole project.
Along side with JSON string export, it is possible to export as XML string!
Before using this feature, make sure associated dependent library is installed. Use fhir.resources[xml] or fhir.resources[all] as
your project requirements.
XML schema validator! It is possible to provide custom xmlparser, during load from file or string, meaning that you can validate data against FHIR xml schema(and/or your custom schema).
- Example-1 Export::
>>> from fhir.resources.patient import Patient >>> data = {"active": True, "gender": "male", "birthDate": "2000-09-18", "name": [{"text": "Primal Kons"}]} >>> patient_obj = Patient(**data) >>> xml_str = patient_obj.xml(pretty_print=True) >>> print(xml_str) <?xml version='1.0' encoding='utf-8'?> <Patient xmlns="http://hl7.org/fhir"> <active value="true"/> <name> <text value="Primal Kons"/> </name> <gender value="male"/> <birthDate value="2000-09-18"/> </Patient>
- Example-2 Import from string::
>>> from fhir.resources.patient import Patient >>> data = {"active": True, "gender": "male", "birthDate": "2000-09-18", "name": [{"text": "Primal Kons"}]} >>> patient_obj = Patient(**data) >>> xml_str = patient_obj.xml(pretty_print=True) >>> print(xml_str) >>> data = b"""<?xml version='1.0' encoding='utf-8'?> ... <Patient xmlns="http://hl7.org/fhir"> ... <active value="true"/> ... <name> ... <text value="Primal Kons"/> ... </name> ... <gender value="male"/> ... <birthDate value="2000-09-18"/> ... </Patient>""" >>> patient = Patient.parse_raw(data, content_type="text/xml") >>> print(patient.json(indent=2)) { "resourceType": "Patient", "active": true, "name": [ { "text": "Primal Kons", "family": "Kons", "given": [ "Primal" ] } ], "gender": "male", "birthDate": "2000-09-18" }
>>> with xml parser >>> import lxml >>> schema = lxml.etree.XMLSchema(file=str(FHIR_XSD_DIR / "patient.xsd")) >>> xmlparser = lxml.etree.XMLParser(schema=schema) >>> patient2 = Patient.parse_raw(data, content_type="text/xml", xmlparser=xmlparser) >>> patient2 == patient True
- Example-3 Import from file::
>>> patient3 = Patient.parse_file("Patient.xml") >>> patient3 == patient and patient3 == patient2 True
XML FAQ
- Although generated XML is validated against
FHIR/patient.xsdandFHIR/observation.xsdin tests, but we suggest you check output of your production data.- Comment feature is included, but we recommend you check in your complex usages.
Although there is no official support for YAML documented in FHIR specification, but as an experimental feature, we add this support.
Now it is possible export/import YAML strings.
Before using this feature, make sure associated dependent library is installed. Use fhir.resources[yaml] or fhir.resources[all] as
your project requirements.
- Example-1 Export::
>>> from fhir.resources.patient import Patient >>> data = {"active": True, "gender": "male", "birthDate": "2000-09-18", "name": [{"text": "Primal Kons", "family": "Kons", "given": ["Primal"]}]} >>> patient_obj = Patient(**data) >>> yml_str = patient_obj.yaml(indent=True) >>> print(yml_str) resourceType: Patient active: true name: - text: Primal Kons family: Kons given: - Primal gender: male birthDate: 2000-09-18
- Example-2 Import from YAML string::
>>> from fhir.resources.patient import Patient >>> data = b""" ... resourceType: Patient ... active: true ... name: ... - text: Primal Kons ... family: Kons ... given: ... - Primal ... gender: male ... birthDate: 2000-09-18 ... """ >>> patient_obj = Patient.parse_raw(data, content_type="text/yaml") >>> json_str = patient_obj.json(indent=True) >>> print(json_str) { "resourceType": "Patient", "active": true, "name": [ { "text": "Primal Kons", "family": "Kons", "given": [ "Primal" ] } ], "gender": "male", "birthDate": "2000-09-18" }
- Example-3 Import from YAML file::
>>> from fhir.resources.patient import Patient >>> patient_obj = Patient.parse_file("Patient.yml") >>> json_str = patient_obj.json(indent=True) >>> print(json_str) { "resourceType": "Patient", "active": true, "name": [ { "text": "Primal Kons", "family": "Kons", "given": [ "Primal" ] } ], "gender": "male", "birthDate": "2000-09-18" }
YAML FAQ
- We are using https://pyyaml.org/ PyYAML library, for serialization/deserialization but if we find more faster library, we could use that. you are welcome to provide us your suggestion.
- YAML based comments is not supported yet, instead json comments syntax is used! Of course this comment feature is in our todo list.
Although this is not good practice to allow empty string value against FHIR primitive data type String. But
we in real life scenario, is it unavoidable sometimes.
- Examples::
Place this code inside your __init__.py module or any place, just to make sure that this fragment of codes is runtime executed.
>>> from fhir.resources.fhirtypes import String >>> String.configure_empty_str(allow=True)
This migration guide states some underlying changes of API and replacement, those are commonly used from later than 6.X.X version.
Replacement: fhir.resources.construct_fhir_element
- First parameter value is same as previous, the Resource name.
- Second parameter is more flexible than previous! it is possible to provide not only json
dictbut also json string or json file path. - No third parameter, what was in previous version.
Replacement: fhir.resources.fhirabstractmodel.FHIRAbstractModel::parse_obj<classmethod>
- First parameter value is same as previous, json dict.
- No second parameter, what was in previous version.
Replacement: fhir.resources.fhirabstractmodel.FHIRAbstractModel::dict
- Output are almost same previous, but there has some difference in case of some date type, for example py date, datetime, Decimal are in object representation.
- It is possible to use
fhir.resources.fhirabstractmodel.FHIRAbstractModel::jsonas replacement, when json string is required (so not need further, json dumps from dict)
Note:
All resources/classes are derived from fhir.resources.fhirabstractmodel.FHIRAbstractModel what was previously
from fhir.resources.fhirabstractbase.FHIRAbstractBase.
Starting from version 5.0.0 we are following our own release policy and we although follow Semantic Versioning scheme like FHIR® version.
Unlike previous statement (bellow), releasing now is not dependent on FHIR®.
removed statement
This package is following FHIR® release and versioning policy, for example say, FHIR releases next version 4.0.1, we also release same version here.
All FHIR® Resources (python classes) are generated using fhir-parser which is forked from https://github.com/smart-on-fhir/fhir-parser.git.
This package skeleton was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.
© Copyright HL7® logo, FHIR® logo and the flaming fire are registered trademarks owned by Health Level Seven International