The fhirpath-lab is a dedicated tool for testing out fhirpath expressions against the various fhirpath execution engines available - dotnet (Firely), java (HAPI) and javascript (nlm)
It also has support for browsing StructureDefinitions, SearchParameters and Questionnaires to be able to test out the various expressions in those contexts.
It will leverage Library resource instances to store fhirpath expressions, along with references to test resource data to evaluate with them
| Resource | Usage Notes |
|---|---|
| Library | FhirPath Expression and references to test data/outcomes |
| StructureDefinition | Source resources for invariant fhirpath expressions |
| Questionnaire | Source resource for validation, pre-population, answerExpressions |
| SearchParameter | Source resource for search parameter expressions |
| StructureMap | Source resource for maps between formats/resource types |
| SubscriptionTopic | Source resource for subscription topic expressions used to detect changes in resources |
| List | favorites data (future use) |
What can I do with it:
- Are there any issues with this expression?
- update the expression to fix the issue
- create a new expression to return the patient's MRN identifier value
- create a new expression to return the patient's name
- correct the coding where clause to handle the presence of multiple types
- help me with this OperationOutcome against this resource (will provide ability to copy the resource into the test resource page, and the outcome expressions into the context)
| Field | Value | Description |
|---|---|---|
| API Key | sk-xxxx |
Your OpenAI API key |
| Base Path | - | Leave this field blank - uses the default OpenAI API endpoint |
| API Version | - | Leave this field blank - only used by Azure deployments |
| Model | gpt-4 |
The model to use for the AI responses |
| Fast Model | gpt-35-turbo |
The model to use for the fast responses |
Note: You can replace the model name(s) with any you have access to and want to try out.
| Field | Value | Description |
|---|---|---|
| API Key | xxxx |
Your Azure Resource endpoint key |
| Base Path | https://xxxx.openai.azure.com/openai/deployments/ |
The base path to the Azure OpenAI deployment - replace the xxxx with your Azure resource name. |
| API Version | 2024-02-01 |
The version of the API to use - recommend using the latest version available https://learn.microsoft.com/en-us/azure/ai-services/openai/reference |
| Model | gpt-4 |
The model to use for the AI responses |
| Fast Model | gpt-35-turbo |
The model to use for the fast responses |
Note: You can replace the model name(s) with any you have access to and want to try out.
| Field | Value | Description |
|---|---|---|
| API Key | - | Leave this field blank - local ollama installation doesn't require an api key! |
| Base Path | http://localhost:11434/v1 |
The base path to the Ollama deployment |
| API Version | - | Leave this field blank - only used by Azure deployments |
| Model | llama3 |
The model to use for the AI responses |
| Fast Model | llama3 |
The model to use for the fast responses |
You will also need to update your ollama configuration to permit the fhirpath-lab to access it. This is done through a system environment variable
OLLAMA_ORIGINS=https://dev-fhirpath-lab.com;https://fhirpath-lab.com
Note: You can replace the model name(s) with any you have access to and want to try out.
Direct access to launch the test page can be done and pass through test data as parameters:
| Parameter | Description |
|---|---|
| expression | (fhirpath expression) The expression to evaluate |
| context | (optional fhirpath expression) A start point within the resource to execute the expression. If this expression returns multiple elements, the expression will be evaluated against each line individually |
| resource | (optional) The resource id of a test resource to evaluate the expression against. If a relative id is provided then it will use the FHIR server selected in your settings. Must contain the resource type e.g. Patient/example |
| resourceJson | (optional) A compressed/encoded test resource JSON |
| libaryId | (optional) A specific fhirpath library instance to load which has all the data |
| exampletype | (optional) The resource type that the expression will be evaluated against (used to select an example resource if the resource or resourceJson is not provided) This is usually used when trying to test a StructureDefinition or other instance where you don't have a specific instance to test with, but you do know what the resource type the expression should be run against The tester will select a resource with the id example on the fhir server selected in your settings |
| engine | (optional) Which fhirpath engine to select (.NET (firely) , fhirpath.js , java (HAPI), java (IBM)) |
| terminologyServer | (optional) Only used by the .net engine for the experimental terminology functionality |
| parameters | A compressed/encoded json object with all the above data in it |
# install dependencies
$ npm install
# serve with hot reload at localhost:3000
$ npm run dev
# build for production and launch server
$ npm run build
$ npm run start
# generate static project
$ npm run generateFor detailed explanation on how things work, check out the documentation.
You can create the following extra directories, some of which have special behaviors. Only pages is required; you can delete them if you don't want to use their functionality.
The assets directory contains your uncompiled assets such as Stylus or Sass files, images, or fonts. More information about the usage of this directory in the documentation.
The components directory contains your Vue.js components. Components make up the different parts of your page and can be reused and imported into your pages, layouts and even other components.
More information about the usage of this directory in the documentation.
This directory contains your application views and routes. Nuxt will read all the *.vue files inside this directory and setup Vue Router automatically.
More information about the usage of this directory in the documentation.
The plugins directory contains JavaScript plugins that you want to run before instantiating the root Vue.js Application. This is the place to add Vue plugins and to inject functions or constants. Every time you need to use Vue.use(), you should create a file in plugins/ and add its path to plugins in nuxt.config.js.
More information about the usage of this directory in the documentation.
This directory contains your static files. Each file inside this directory is mapped to /.
Example: /static/robots.txt is mapped as /robots.txt.
More information about the usage of this directory in the documentation.
This directory contains a custom JSON Parser built using ANTLR4. The grammar file JSON5.g4 is the only source file, the others are all generated.
The grammar file comes from: https://github.com/antlr/grammars-v4/blob/master/json5/JSON5.g4
Followed the instructions here to actually generate the parser: https://github.com/antlr/antlr4/blob/master/doc/typescript-target.md
# Install the antlr tools using python (only need python local)
$ pip install antlr4-tools
# Generate the parser files (run in the /json-parser directory)
$ antlr4 -Dlanguage=TypeScript JSON5.g4/* a multi-line comment - FHIR syntax highlighers */
1 // integer
| -2 // negative integer
| 3.141592650 // decimal
| @2020-12-01 // date
| @2015-02-04T14:34:28+09:00 // datetime
| @T12:45:28 // time portion
| 'some string' // string
| '\tsom\\e \'st\'r\ning' // string with a quote inside it (and newline)
| '\u00A9' // unicode copyright symbol
| 'string' & ' another string' // string contactenation
| 1 weeks // quantity
| 1 'mg' // quantity
| gender // simple identifier
| meta.versionId // backbone property
| name[0].text // array indexer
| today() // top level function
| 1.toString() // const calling a string
| birthDate.toString() // type conversion(function) on property
| `deceased` // quoted identifier
| text.`div` // identifier that is a reserved word
| _testme // idenfier starting with an underscore (not a fhir property)
| deceased.ofType(Boolean) // type selector
// | deceased as boolean // 'as' cast operator
// | deceased.as(boolean) // `as(` cast function
| (deceased is Boolean)
| name.where($this.use = 'usual').text // $this keyword in use
| iif(true, 'iif true', 'iif false') // iif function
| %ucum // an environment variable (from the spec)
| %`ucum` + '/test' // delimitted environment variable
| 5-6 // a numeric expression
| iif(true or false xor true, 'boolean tv', 'boolean fv') // booleans
| 6*6 //numeric multiplication
| name.select('x: '+ family.combine(given).join(', '))The project uses the custom json parser to parse FHIR resources and create a tree structure containing all the nodes in the resource:
- Node text e.g.
family,name - Simple FHIPath based path e.g.
Patient.name[1].family - Datatype of the node e.g.
HumanName,string - Children of the node (if any) (for objects or arrays)
- Position of the node in the source JSON
Note: The value of the nodes is not stored in the tree (not the purpose of the tree - just use JSON.parse if you want that)
The tree is used in the fhirpath-lab to enable the navigation from a given simple fhirpath expression to the node in the tree that it represents. Currently this is in the fhirpath test page and questionnaire test page.
Note: This functionality may be used in the future to enhance the ace editor for the fhir expression editor.
https://github.com/antlr/antlr4/blob/master/doc/ace-javascript-target.md
(this set of issues are known but not on the main issues list as don't plan to fix unless have legitimate use cases that need them)
- Environment Variables using the quoted format cannot include escaped values in them e.g.
%`tes\`t` - Environmant Variable processing in the JAVA engine doesn't support resource framgents
- Sharing an expression via the share link doesn't include resource JSON or environment variable values (plan to support this only via saved Library resources in the future)
- json object tree path not accurate for primitive extensions
(these should be cleaned up in the exitObject function)
e.g.
Patient._birthDate.extension[0].valueDateTime