Nice!

@Herman Slagman - Perhaps we will meet again and we can have a vigorous debate! Your turn to buy ;)

I am somewhat reluctant to show this as the performance is not great. We are working on that. Also, keep in mind that the expression language is JSON Path Language which was made an ISO Standard in the 2016 SQL Standard. We have extended it a bit and also implemented it in a way that allows it to be used outside of SQL which is not part of the Standard.

In this snippet, the lvar 'j' is assigned the value of the JSON posted by the OP. I modified it to make it valid JSON (just added quotes here and there).

do ##class(%ASQ.SetUtils).pp(j.apply("$.items[*].identifiers[*]?(@.typeDiscriminator == 'ClassifiedId' && @.type.term.en_US == 'Scopus Author ID')"))

The result:

[
  {
    "typeDiscriminator": "ClassifiedId",
    "pureId": "xxxxxxxx",
    "id": "xxxxxxxx",
    "type": {
      "uri": "/dk/atira/pure/person/personsources/scopusauthor",
      "term": {
        "en_US": "Scopus Author ID"
      }
    }
  }
]

The performance of apply() can be improved significantly by parsing the expression first and then passing the parsed result to apply(). For one-off executes that isn't helpful but if you are applying the same expression to different data then the improvement is significant.

LATEST:USER>set ast = ##class(%ASQ.Parser).parse("$.items[*].identifiers[*]?(@.typeDiscriminator == 'ClassifiedId' && @.type.term.en_US == 'Scopus Author ID')")

LATEST:USER>set result = j.apply(ast)

LATEST:USER>do ##class(%ASQ.SetUtils).pp(result)
[
  {
    "typeDiscriminator": "ClassifiedId",
    "pureId": "xxxxxxxx",
    "id": "xxxxxxxx",
    "type": {
      "uri": "/dk/atira/pure/person/personsources/scopusauthor",
      "term": {
        "en_US": "Scopus Author ID"
      }
    }
  }
]

You can also do this in SQL but it is quite verbose. You have to map the JSON values to columns in the SELECT. I can provide an example if it would be helpful.