对象化中的复杂查询

Question

Consider a more complex entity structure like

class Entity {
    Float valueA;
    Float valueB;
    List<Property> properties;
}

class Property {
    Long id;
    Float value;
}

I'm now trying to figure out an index structure to do queries like (Pseudo sql):

select entity where valueA >= x and valueB < y and properties contains
((id = a, value >= b) and (id = c, value = d))

The problems I'm facing are:

• I can't put multible inequality filters in one query.
• How would I add constraints for the list property objects

So far I have one idea that came accross my mind:

I could make use of the relation index pattern for all properties. Eg. create the following entities:

class ValueA/ValueB {
   @Parent
   Key<Entity> parent;
   @Id
   Long id = 1L;
   @Index
   Float minValue;
}

and change the Property class like

class Property {
    @Parent
    Key<Entity> parent;
    @Id
    Long id;
    @Index
    Float value;
}

Then I could do a query for every related index class and keep the parent keys of those entities that mach all criteria.

This would be hard to do efficiently and could easily get very expensive.

Are there any other solutions I could try out?

Thanks in advance!

Answer 1

I would store a denormalized copy of my Entity in the Search Api, which allows for much more flexible querying.

Make sure that your result from the Search Api contains the Id of the entities.

Finally, do a keys() query with Objectify to get to the actual entities of your result.