I have a table on PostgreSQL 10 that looks like:
CREATE TABLE forms (id serial, form jsonb);
INSERT INTO forms (id, form) VALUES (
1, '"permissions": [ {"name": "perm-name-1", "who":["group1", "group2",]},
{"name": "perm-name-2", "who":["group1", "group3"]}]');
INSERT INTO forms (id, form) VALUES (
2, '"permissions": [ {"name": "perm-name-1", "who":["group5", "group6"]},
{"name": "perm-name-2", "who":["group7", "group8"]}]');
The JSON for column "form" has more key-value pairs which are irrelevant here.
I need to query based on the permissions "name" and one or more "who" values e.g. :
- Get all rows whose permission "name" is "perm-name-1" and "who" is any of "group1", "group5" should match all the rows of the example above.
- Get all rows whose permission "name" is "perm-name-1" and "who" is just "group1" should match row with id 1.
Also, any insight/comment on a JSON schema which would help the database to perform better is welcomed!
Best Answer
I think the first level of arrays is not useful, especially if you are going to retrieve permissions by name.
I would store it like this:
Then both your queries are quite easy to write:
For the first one you can use the "any" operator
?|
:For the second one, the
?
operator is enough:You didn't ask for it, but you can also query for rows where a specific permissions contains all specified groups. That can be done using the
?&
operator. The following would return the row with ID=1The following would return nothing: