Snowflake - Get value of key in an object within another object - json

Question

I'm working with Snowflake and loading json files from a Staging environment to an ODS environment.
The structure of the json data looks like this:
{"address": "921 Pearl St",
"attributes": {
"Alcohol": "'beer_and_wine'",
"Ambience": "{'touristy': False, 'hipster': False, 'romantic': False, 'divey': False, 'intimate': False, 'trendy': False, 'upscale': False, 'classy': False, 'casual': True}",
"BikeParking": "True",
"BusinessAcceptsBitcoin": "False",
"BusinessAcceptsCreditCards": "True",
"BusinessParking": "{'garage': False, 'street': True, 'validated': False, 'lot': False, 'valet': False}"
}
Now I'm trying to query my data in Snowflake like this:
SELECT json_data:attributes.BusinessParking.garage
FROM my_json_table;
but all I get are null values.
Any idea how do I get to the bottom levels in the "BusinessParking" field?
thanks!!!

Answer 1

Trying to parse the JSON in online tool I got:
Error: Parse error on line 9:
..., 'valet': False}"}
----------------------^
Expecting '}', ',', got 'EOF'
You could check if it is correct by running TRY_PARSE_JSON:
SELECT TRY_PARSE_JSON(json_data) from my_json_table

Answer 2

Firstly your object is missing a closing }
So if I rewrite you "JSON" to be valid then the select form you use works:
select parse_json('{
"address": "921 Pearl St",
"attributes": {
"Alcohol": "beer_and_wine",
"Ambience": {
"touristy": False,
"hipster": False,
"romantic": False,
"divey": False,
"intimate": False,
"trendy": False,
"upscale": False,
"classy": False,
"casual": True
},
"BikeParking": True,
"BusinessAcceptsBitcoin": "False",
"BusinessAcceptsCreditCards": "True",
"BusinessParking": {
"garage": False,
"street": True,
"validated": False,
"lot": False,
"valet": False
}
}
}') as json_data,
json_data:attributes.BusinessParking.garage,
json_data:attributes:BusinessParking:garage
;
but as that implies your JSON is not valid.
looking at your Ambience it a stringfied JSON sub-object, not an object, thus the embedded ' inside it. So escaping your '`' in the sub strings, to it parses correctly we will need to PARSE_JSON the sub-objects, thus:
select parse_json('{"address": "921 Pearl St",
"attributes": {
"Alcohol": "\'beer_and_wine\'",
"Ambience": "{\'touristy\': False, \'hipster\': False, \'romantic\': False, \'divey\': False, \'intimate\': False, \'trendy\': False, \'upscale\': False, \'classy\': False, \'casual\': True}",
"BikeParking": "True",
"BusinessAcceptsBitcoin": "False",
"BusinessAcceptsCreditCards": "True",
"BusinessParking": "{\'garage\': False, \'street\': True, \'validated\': False, \'lot\': False, \'valet\': False}"
}}') as json_data,
PARSE_JSON(json_data:attributes.BusinessParking):garage
gives the answer false that you are looking for.
EDIT
to add more details, you object has json objects nestsed which are Stringify'ed
to put replicate that the above (and below) SQL uses escape tokens \ to allow single quotes into the data. But after that we can select json_data:attributes.BusinessParking and see it presented like you have it shown in your example, which mean if we then PARSE_JSON on that, we get a valid JSON object again, which we can then access via :garage
select parse_json('{"address": "921 Pearl St",
"attributes": {
"Alcohol": "\'beer_and_wine\'",
"Ambience": "{\'touristy\': False, \'hipster\': False, \'romantic\': False, \'divey\': False, \'intimate\': False, \'trendy\': False, \'upscale\': False, \'classy\': False, \'casual\': True}",
"BikeParking": "True",
"BusinessAcceptsBitcoin": "False",
"BusinessAcceptsCreditCards": "True",
"BusinessParking": "{\'garage\': False, \'street\': True, \'validated\': False, \'lot\': False, \'valet\': False}"
}}') as json_data,
json_data:attributes.BusinessParking as it_looks_the_same,
PARSE_JSON(it_looks_the_same) as valid_json,
valid_json:garage
which gives:

Answer 3

After a lot of research I'm able to query the fields the way I needed to.
I found out that I need to use the "try_parse_json" functions with those fields as they are defined as strings not actual json objects.
After using the function I can parse specific fields just like this:
SELECT
TRY_PARSE_JSON(
jsondata: attributes.BusinessParking
): garage AS garage
FROM
"RESTAURANT"."STAGING"."JSON_TABLE_BUSINESS"
Snowflake's doc isn't that great but somehow it explains what you can do: https://docs.snowflake.com/en/sql-reference/functions/try_parse_json.html
This is another way you could transform string values to columns in case you need it: https://sqlkover.com/cool-stuff-in-snowflake-part-3-split-and-flatten/
Thanks for all the comments #simeon-pilgrim!