fhirsnake: turn static files to a FHIR server

What is it?

fhirsnake is a minimalistic FHIR server that serve yaml and json files as FHIR resources

How it works?

The server reads all yaml and json files from an input directory (resources by default). The input directory should have subdirectories with names equal to resource types:

resources/
├── Patient/
│   ├── example.yaml
│   ├── john-doe.json
│   └── patient3.yaml
├── Questionnaire/
│   ├── questionnaire1.yaml
│   ├── questionnaire2.yaml
│   └── subdir/
│       └── questionnaire3.yaml

Use the --input flag to specify a custom input directory. For export and watch commands, --input can be passed multiple times to load resources from several directories.

Environment variable substitution

To use environment variables in resources, you can use the syntax ${VAR_NAME}. In the case of missing environment variable, the exception will be raised.

NOTE: The syntax $VAR without braces is not supported because it might be used in resources.

Supported operations

Important

All created and updated resources persist in runtime only. All changes are vanished after service restart

• read, create and update operations are supported
• search - limied support without any search params
• GET /$index operation returns a map of all resources in format <resource_type>:<id>

Usage

1. Organize resources in a directory

Server

docker run \
  -p 8002:8000 \
  -v ./resources:/resources \
  bedasoftware/fhirsnake server --input /resources

Or build a custom image with Dockerfile.resources:

docker build -t bedasoftware/fhirsnake -f Dockerfile.resources .
docker run -p 8000:8000 fhirsnake-resources

Export

Export resources as .json (Bundle), .ndjson, or .ndjson.gz:

docker run \
  -v ./resources:/resources \
  -v ./output:/output \
  bedasoftware/fhirsnake export \
    --input /resources \
    --output /output/seeds.ndjson.gz

Multiple input directories:

docker run \
  -v ./resources1:/resources1 \
  -v ./resources2:/resources2 \
  -v ./output:/output \
  bedasoftware/fhirsnake export \
    --input /resources1 \
    --input /resources2 \
    --output /output/seeds.ndjson.gz

With external FCE->FHIR converter:

docker run \
  -v ./resources:/resources \
  -v ./output:/output \
  bedasoftware/fhirsnake export \
    --input /resources \
    --output /output/seeds.ndjson.gz \
    --external-questionnaire-fce-fhir-converter-url http://host.docker.internal:3000/to-fhir

Watch

Watch resources for changes and send as PUT requests to an external FHIR server:

docker run \
  -v ./resources:/resources \
  bedasoftware/fhirsnake watch \
    --input /resources \
    --external-fhir-server-url http://host.docker.internal:8080

With auth headers:

docker run \
  -v ./resources:/resources \
  bedasoftware/fhirsnake watch \
    --input /resources \
    --external-fhir-server-url http://host.docker.internal:8080 \
    --external-fhir-server-header "Authorization: Token token"

Multiple input directories:

docker run \
  -v ./resources1:/resources1 \
  -v ./resources2:/resources2 \
  bedasoftware/fhirsnake watch \
    --input /resources1 \
    --input /resources2 \
    --external-fhir-server-url http://host.docker.internal:8080

External questionnaire FCE->FHIR converter

The image bedasoftware/questionnaire-fce-fhir-converter:latest provides a /to-fhir endpoint that can be used with the --external-questionnaire-fce-fhir-converter-url flag in export and watch commands.

Contribution and feedback

Please, use Issues

Development

Format and fix with ruff

ruff format
ruff check . --fix

Issue new version - run semantic release locally

npx semantic-release --no-ci

Publish to pypi

poetry build
poetry publish