datasets/fullquery behaves differently with or without Elasticsearch

[jl-wynen]

Differences between instances with and without Elasitcsearch

There are some differences when querying between SciCat instances that run Elasticsearch vs those that don't run it. I tested with staging.scicat.ess.eu, scicat.ess.eu, and a locally deployed instance (latest GH package) w/o Elasticsearch.

Different behaviour

When querying ESS staging for datasets, it ignores some fields in the query. E.g.

import json
import httpx

response = httpx.get(
        "https://staging.scicat.ess.eu/api/v3/datasets/fullquery",
        params={
            "fields": json.dumps(
                {
                    "proposalId": "123"
                }
            ),
        },
    )
datasets = response.json()
print(len(datasets))
print(datasets[0]['proposalId'])

prints

100
67JH32

In fact, it does not respect proposalId but returns all datasets.

But when running the same query against https://scicat.ess.eu, I get 0 results as this proposal does not exist.

It even ignores unknown fields:

import json
import httpx

response = httpx.get(
        "https://staging.scicat.ess.eu/api/v3/datasets/fullquery",
        params={
            "fields": json.dumps(
                {
                    "asd": ["ess"]
                }
            ),
        },
    )
print(response)
print(len(response.json()))

prints

<Response [200 OK]>
100

Different syntax

import json
import httpx

response = httpx.get(
        "https://staging.scicat.ess.eu/api/v3/datasets/fullquery",
        params={
            "fields": json.dumps(
                {
                    "ownerGroup": "ess"
                }
            ),
        },
    )
print(response)

prints

<Response [400 Bad Request]>

but on production, I get <Response [200 OK]> and a correct list of datasets.

To fix this, I need to change it to

"fields": json.dumps(
    {
        "ownerGroup": ["ess"]
    }
),

on staging.

[nitrosx]

They should behave exactly the same with or without ES

[Junjiequan]

This issue should be fixd by #1228 and #1184.