雪花 JSON FLATTEN with ORDER BY

Question

I have a working query that flattens a nested JSON object into rows of data. What I would like to do, however, is preserve the original order of one array of objects which is nested several layers in.

I have tried to use ROW_NUMBER with an ORDER BY NULL and an ORDER BY (SELECT NULL) and neither seem to preserve the order.

Any ideas on how to accomplish that? Examples below. I chose to mask the real data, but the important parts of the structure are the same. The data in JSON format comes through with no rank-identifying information, but I used numbers as examples here to show the strange results.

Original structure (masked):

{
    "topNode: {
        "childNode": {
            "list": [
                {
                    "title": "example title 1",
                },
                {
                    "title": "example title 2",
                },
                {
                    "title": "example title 3",
                },
                {
                    "title": "example title 4",
                },
                {
                    "title": "example title 5",
                }
            ]
        }
    }
}

Example query (masked):

SELECT 
    A.VALUE:"title"::VARCHAR AS "TITLE",
    ROW_NUMBER() OVER(ORDER BY NULL) AS RANK
FROM 
    DB.SCHEMA.TABLE as A,
    lateral flatten(input=>A.JSON:topNode.childNode.list) "list_flatten"

Example output:

TITLE                     RANK
"example title 3"         1
"example title 5"         2
"example title 2"         3
"example title 1"         4
"example title 4"         5

Answer 1

It is possible with INDEX , which returns index of element in array:

SELECT A.VALUE:"title"::VARCHAR AS "TITLE",
       "list_flatten".index AS "RANK"
FROM DB.SCHEMA.TABLE as A,
lateral flatten(input=>A.JSON:topNode.childNode.list) "list_flatten"

Answer 2

I have a working query that flattens a nested JSON object into rows of data. What I would like to do, however, is preserve the original order of one array of objects which is nested several layers in.

I have tried to use ROW_NUMBER with an ORDER BY NULL and an ORDER BY (SELECT NULL) and neither seem to preserve the order.

Any ideas on how to accomplish that? Examples below. I chose to mask the real data, but the important parts of the structure are the same. The data in JSON format comes through with no rank-identifying information, but I used numbers as examples here to show the strange results.

Original structure (masked):

{
    "topNode: {
        "childNode": {
            "list": [
                {
                    "title": "example title 1",
                },
                {
                    "title": "example title 2",
                },
                {
                    "title": "example title 3",
                },
                {
                    "title": "example title 4",
                },
                {
                    "title": "example title 5",
                }
            ]
        }
    }
}

Example query (masked):

SELECT 
    A.VALUE:"title"::VARCHAR AS "TITLE",
    ROW_NUMBER() OVER(ORDER BY NULL) AS RANK
FROM 
    DB.SCHEMA.TABLE as A,
    lateral flatten(input=>A.JSON:topNode.childNode.list) "list_flatten"

Example output:

TITLE                     RANK
"example title 3"         1
"example title 5"         2
"example title 2"         3
"example title 1"         4
"example title 4"         5