nazrulworld / fhirpath Goto Github PK

• fhirpath version: 0.10.4
• Python version: 3.9.0
• Operating System: MacOS

Description

This is not a bug report but a question.

What I Did

I created a Patient resource with the following content

{
  "resourceType": "Patient",
  "id": "1",
  "name": [
    {
      "given": [
        "John"
      ],
      "family": "Mercer"
    },
    {
      "given": [
        "Alex"
      ],
      "family": "Connor"
    }
  ]
}

Then, I tried the following search

search_context = SearchContext(engine, "Patient")
# equivalent to /Patient?name=John&name=Connor
params = (("name", "John"), ("name", "Connor"), )
search_tool = Search(search_context, params=params)
bundle = search_tool()
assert bundle.total > 0

No documents are matched due to the nested wrapper on the top-most GroupTerm.

A similar query on HAPI Test Server returns a non-zero response

curl -X GET "http://hapi.fhir.org/baseR4/Patient?name=Ajay&name=Mishra&_pretty=true"

I could not any relevant documentation or rules around such matches on FHIR Search.

Is this behavior server specific or something that I may have missed in FHIR specification?

Description

The search is not yet able to handle the _include modifier which allows to "join" resources between them.
For instance, the query string /Observation?category=laboratory&_include=Observation:encounter which is equivalent to

search_context = SearchContext(engine, "Observation")
params = (("category", "laboratory"), ("_include", "Observation:encounter"))
fhir_search = Search(search_context,  params=params)

should output a Bundle like:

{
    "id": "1",
    "meta": {},
    "entry": [
        {
            "fullUrl": "Observation/123",
            "resource": {
                "...": "...",
                "resourceType": "Observation"
            },
            "search": {
                "mode": "match"
            }
        },
        {
            "fullUrl": "Encounter/456",
            "resource": {
                "...": "...",
                "resourceType": "Encounter"
            },
            "search": {
                "mode": "include"
            }
        }
    ],
    "link": [
        {
            "relation": "self",
            "url": "https://some.url"
        }
    ],
    "total": 1,
    "type": "searchset",
    "resourceType": "Bundle"
}

I'm not sure how we should implement "relations" between documents but the parent-child relationship using the join datatype seems to be a good fit.
A first implementation could also consist in running multiple ES queries and construct a bundle from multiple results.

https://www.hl7.org/fhir/search.html

The project is currently licensed under GPLv3, which (if I'm not mistaken) makes dependent projects using GPLv3 as well (see https://choosealicense.com/licenses/gpl-3.0/).

We are currently building open-source software on top of fhirpath (and fhir.resources) but we are using the more permissive Apache2 license (like most projects hosted at https://github.com/FHIR or popular projects like https://github.com/hapifhir/hapi-fhir or https://github.com/IBM/FHIR).
Apache2 details (and TLDR) can be found here: https://choosealicense.com/licenses/apache-2.0/
Note that we would not be impacted by other permissive licenses like the MIT License.

I know we already discussed the subject and you seemed to be open to changing the licensing model but this is becoming an issue on our end.

Related: arkhn/fhir-river#281

Description

Hey @nazrulworld and congratulations for the amazing work you've done here! I'm currently building a FHIR api in python and fhirpath seems to supports many features i'm interested in (search params, chaining, many modifiers etc..) so I was thinking of giving it a go! However I already have my own "provider" as you call it, and I would like to write an integration for my provider to work with fhirpath (just like what you did with fhirpath_guillotina).

The library I use is called fhirstore and it is based on mongoDB (storage) and elasticsearch (indexing) so I was thinking that maybe it could be easy to write my own integration but I don't really know which interfaces I need to implement and I did not find more detailed documentation. I started digging in the code of fhirpath_guillotina but there is a lot of called imported from the guillotina framework and I don't fully understand what's going on.

Would you be willing to share more details on how to write a new provider ?

The company I work with dedicates some of its resources to develop the FHIR ecosystem (mostly python libs) and would be happy to contribute to fhirpath!

Cheers :)

• fhirpath version: 0.7.1
• Python version: 3.7
• Operating System: MacOS

Description

I need to search for Task those statuses are not in completed, cancelled, rejected.

What I Did

query = {
            "status:not": "completed,cancelled,rejected",
            "code:not": "|consentTask",
        }

Didn’t get the expected result, but when do query positive term like "status": "ready,in-progress", it's working!

Sometimes there would no parameters for example ´@fhir/Task´ that should support (of course have to maintain maximum result limit)

Right now that is not possible.

Description

In #7, we added support for search on several resource types but forgot to test this functionality. It would be good to do so as soon as possible.