解析 Snowflake 中嵌套的 JSON 字段

Question

I am pretty new to Snowflake and I am now trying to parse a JSON field and pull its attributes to return in the response.

I tried a few variations but every time, the attribute is populating as null .

attributes column in my table has this JSON :

{
    "Status": [
        "ACTIVE"
    ],
    "Coverence": [
        {
            "Sub": [
                {
                    "EndDate": [
                        "2020-06-22"
                    ],
                    "Source": [
                        "Test"
                    ],
                    "Id": [
                        "CovId1"
                    ],
                    "Type": [
                        "CovType1"
                    ],
                    "StartDate": [
                        "2019-06-22"
                    ],
                    "Status": [
                        "ACTIVE"
                    ]
                }
            ]
        }
    ]
}

What I tried :

SELECT DISTINCT *
    from
        (
            TRIM(mt."attributes":Status, '[""]')::string as STATUS,
            TRIM(r.value:"Sub"."Id", '[""]')::string as ID,
            TRIM(r.value:"Sub"."Source", '[""]')::string as SOURCE
            from "myTable" mt,
            lateral flatten ( input => mt."attributes":"Coverence", outer => true) r
        )
    GROUP BY
        STATUS,
        ID,
        SOURCE;

Later I tried :

SELECT DISTINCT *
    from
        (
            TRIM(mt."attributes":Status, '[""]')::string as STATUS,
            TRIM(r.value:"Id", '[""]')::string as ID,
            TRIM(r.value:"Source", '[""]')::string as SOURCE
            from "myTable" mt,
            lateral flatten ( input => mt."attributes":"Coverence":"Sub", outer => true) r
        )
    GROUP BY
        STATUS,
        ID,
        SOURCE;

But nothing worked. The STATUS is populating as expected. But ID and SOURCE are populating null .

Am I missing something or have I done something dumb? Please shed some light.

Answer 1

Assuming that Coverence could contain multiple Sub , therefore FLATTEN twice. At lowest level only first element is chosen ( EndDate[0] , Source[0] etc):

SELECT 
   mt."attributes":Status[0]::TEXT AS Status
  ,r2.value:EndDate[0]::TEXT AS EndDate
  ,r2.value:Source[0]::TEXT AS Source
  ,r2.value:Id[0]::TEXT AS Id
FROM myTable AS mt,
LATERAL FLATTEN(input => mt."attributes",
                path => 'Coverence',
                outer => true) r1,
LATERAL FLATTEN(input => r1.value,
                path => 'Sub',
                outer => true) r2; 

Output:

Answer 2

All your elements are array type and the overall JSON is not making much sense... so to access all your individual elements, you have to use [] notation and then you can access the element values. You don't need to use flatten also, if you just have to access individual elements via index.