I have 2 tables company and contacts. Contacts has addresses JSONB column. I tried a select statement with a join on contacts.linked_to_company and using jsonb_array_elements(company.addresses) but I get error 'cannot extract elements from a scalar' which I understand is because some entries do have [null] in column address. I have seen answers to use coalesce or a CASE statement. Coalesce I could get to not work and CASE example is in the select statement how do use it in a join? Here is the sql

SELECT company.id, trading_name, nature_of_business, t.id contactID, address->>'PostCode' Postcode, position_in_company FROM contact t FULL JOIN company ON (t.company_linked_to = company.id ), jsonb_array_elements(t.addresses) address WHERE t.company_linked_to ='407381'; 

here is example jsonb

[{"PostCode":"BN7788","Address":"South Street","AddressFull":"","Types":[{"Type":"Collection"}]}] 

1 Answer

You can try one of these (instead of jsonb_array_elements(t.addresses) address):

jsonb_array_elements( case jsonb_typeof(addresses) when 'array' then addresses else '[]' end ) as address -- or jsonb_array_elements( case jsonb_typeof(addresses) when 'array' then addresses else '[{"PostCode": null}]' end ) as address 

The first one hides rows with improper json format of the column, the second one gives null for them.

However, the problem actually stems from that one or more values in the column is not a json array. You can easily fix it with the command:

update contact set addresses = '[null]' -- or -- set addresses = '[{"PostCode": null}]' where jsonb_typeof(addresses) <> 'array' or addresses = '[]'; 

After this correction you won't need case in jsonb_array_elements().

3

Your Answer

Sign up or log in

Sign up using Google Sign up using Facebook Sign up using Email and Password

Post as a guest

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.