EnumSet lookup : Java Glossary

You probably could use an EnumSet as a key to lookup records in a HashSet, however that is unlikely what you want. Let us say you had an EnumSet of tree attributes: Deciduous, Rainforest, Desert, Temperate, Hardwood…. You might want to see all the records in your collection what were Deciduous Temperate or Deciduous Temperate but not Hardwood. Java does not have built-in tools for doing this. What could you do?

Probably the easiest way to handle this is to put all your data in an SQL (Standard Query Language) engine and leave it up to the designers of the engine to optimise the searches.

Specifying

You could define a tri-bit string 0=must not be present 1=must be present 2=does not matter, to describe the combinations you are looking for.

You could use two EnumSets, one for what you definitely want, and one for what you definitely don’t want. Any enum not mentioned in either is irrelevant.

Implementation

Here are some possible approaches to implementation:

• Perhaps the most Javaesque way to implement it would be to use a Map<${PropertyType},Set<Person>> per property. Depending on application logic you would build these just once when reading the data or update indexes whenever you update fields. Querying would use set operations to reduce the set of results. This would be fairly slow and memory hogging since the information is stored in the presence or absence of an Object is a Set, rather than just a single bit.
• You could store the EnumSets for each candidate is your collection in an array of longs. You would have to sequentially process each candidate with:

  // wanted, candidate notWanted and candidate are longs.
  boolean take = ( wanted & candidate & ~( notWanted & candidate ) ) != 0;

• If your list of candidates does not change, you could extract a BitSet for each enum, each n long where n is the number of candidates in the collection. You would do logical operations on the relevant long bit strings to create a bit string with a 1 in each slot for each candidate that meets all the criteria.
• You could use LINQ in C#.