JSONB

Clear offers JSONB functions through Clear::SQL::JSONB helper and the Expression Engine.

JSONB is a great No-SQL mapping under PostgreSQL. It helps you to store value and documents which otherwise would be difficult

Let's imaging a table events where you store the events of differents suppliers:

Postgres limitation and Clear's answer

The main limitation of JSONB is the "simple" syntax is not indexable. For example:

  SELECT * FROM events WHERE payload->'source'->>'name' = 'Asana'

The code above will not use any indexes and will do a sequencial scan over your table.

However, using the @> operator and a gin index on your column will improve drastically the performances:

  SELECT * FROM events WHERE payload @> '{"source": {"name": "Asana"}}'

Obviously, the second syntax is more complex and error prone. Clear offers leverage and simplicity:

  Event.query.where{ payload.jsonb("source.name") == "asana" }
  #  SELECT * FROM events WHERE payload @> '{"source": {"name": "Asana"}}'

Expression Engine

jsonb

calling node.jsonb(key) on expression node will resolve to:

• node->'key_elm1'->'key_elm...n'

Using equality testing between a jsonb path and a literal will use the indexable notation @> :

where{ data.jsonb('a.b.c') == 1 }
#output:
# data @> '{"a":{"b":{"c":1}}}'

In the case the operation is not indexable (e.g. the value is variable, operator is not equality...), Clear will automatically switch back to the arrow -> notation:

where{ data.jsonb('a.b.c') == raw("NOW()") }
# output:
# data->'a'->'b'->'c' = NOW()

Casting

You can cast the element using cast after your expression:

where{ data.jsonb("a.b").cast("text") == "o" }
# output:
# data->'a'->'b'::text == 'o'

Note: If you cast the jsonb, clear will never use @> operator

From path to arrow notation

Clear::SQL::JSONB.jsonb_resolve("data", "a.b.c", "text")
# output:
# data->'a'->'b'->'c'::text

Use outside Expression Engine (@> operator)

Clear::SQL::JSONB.jsonb_eq(data, "a.b.c", "value")
#output:
# data @> {"a":{"b":{"c":"value"}}}