Creating MongoDB case insensitive index in Spring based project

Example problem to solve

One of the most popular class in our services is User.
The user may enter his nickname during registration.
Let’s assume that our business requirement is:

• user should be unique by his nickname

Solution in MongoDB - case insensitive index

If we use MongoDB, we can use unique index with appropriate option to achieve that requirement.

The appropriate option is collation with with set fields locale and strength.

Supported values for this fields you can find here:

• locale
• strength

For this article let’s assume that we want to focus on language English (United States) so we will use en_US value for locale.

For strength we should use values 1 or 2 because this levels means that values comparison will ignore case.

So, our solution directly on MongoDB will look like:

db.users.createIndex( { nickname: 1}, { unique: true, collation: { locale: 'en_US', strength: 1 } } )

How to create case insensitive index in MongoDB using Spring Framework?

Usually for creating indexes we can use annotations:

• @Indexed on field
• @CompoundIndex on class

They have option for uniqueness to set but not for collation.

Fortunately, such option is available if we create index directly via MongoTemplate. So, with such simple code we can meet the requirement.

@Configuration
@DependsOn("mongoTemplate")
class CollectionConfig(@Autowired private val mongoTemplate: MongoTemplate) {

  @PostConstruct
  fun ensureIndexes() {
    mongoTemplate.indexOps(User::class.java).ensureIndex(
      Index("username", Sort.Direction.ASC)
        .unique()
        .collation(
          of(Locale.US).strength(ComparisonLevel.primary())
        )
    )
  }

}

Issue in spring-boot-mongodb project

I’ve created issue to add parameters in @Indexed and @CompoundIndex annotations for setting collation value. You can watch this issue if you are interested in what will happen next.

Sources

• Case Insensitive Index in MongoDB documentation.
• Question on StackOverflow with mine answer as well.