Spring Data MongoDB 3.1 发布

The Spring Data MongoDB 3.1 release is one of the modules that highly benefited from the recent changes in the Spring Data Commons module, by leveraging the infrastructure built there to bring reactive features like auditing and SpEL.

The following snippet gives you an impression of what this means for declarative MongoDB queries using SpEL:

@Query("{ 'supervisor' : ?#{ hasRole('ROLE_ADMIN') ? new Document('$regex', '*') : principal.name } }")
Flux<Person> findAllFilteredByRole();COPY

@EnableReactiveMongoAuditing uses the common infrastructure so you can keep track of changes easily.

@Configuration
@EnableReactiveMongoAuditing
static class AuditingConfig extends AbstractReactiveMongoConfiguration {
   
    @Override
     protected String getDatabaseName() {
        return "database";
    }
    
    @Bean
    public ReactiveAuditorAware<User> auditorProvider() {
        return () -> ReactiveSecurityContextHolder.getContext()
        .map(it -> ...);
    }
}COPY

Another focus of the current release was to keep up with the changes and enhancements around the MongoDB Aggregation Framework that lets you specify all sorts of data aggregation operations now, even including the possibility to run JavaScript functions as part of the pipeline (as shown in the example), recreating the average books calculation from the MongoDB reference documentation, below:

Accumulator accumulator = accumulatorBuilder()
	.init("function() { return { count: 0, sum: 0 } }")
	.accumulate("function(state, numCopies) { return { count: state.count + 1, sum: state.sum + numCopies } }")
	.accumulateArgs("$copies")
	.merge("function(state1, state2) { return { count: state1.count + state2.count, sum: state1.sum + state2.sum } }")
	.finalize("function(state) { return (state.sum / state.count) }");COPY

To improve performance of aggregations operating on large datasets, MongoDB lets you add index hints to the pipeline by using AggregationOptions.builder().hint("...").

Speaking of indices, although we highly recommend taking programmatic control over index creation, we also understand the convenience of declarative index definition holds. Therefore, we added partial filter expressions to the @Indexed and @CompoundIndex annotations to make index setup a little more convenient, as shown below:

@CompoundIndex(name = "idx", def = "...", partialFilter = "{'ssn': {'$exists': true}}")
public class Person {
	
	//...
	@Indexed(partialFilter = "{'rating': { $lt: 10 }}") 
	String lastname;
}COPY

So now that we have sped up aggregations and indexes, it is about time we talk about counting elements. As you know, we switched MongoOperations.count(...) (which previously worked on cached collection statistics) to run an aggregation for precise counting as part of our efforts on Multi Document Transactions.
Now, in some cases, one might not need the exact document count but can live with a close enough approximation. For those cases, you can now get an estimate from MongoOperations.estimatedCount(...).

Last but not least, I wanted to mention the incubating GraalVM native image support for Spring Data MongoDB. Make sure to have Docker installed (just for convenience, as this will get you going without the need to install the GraalVM nor MongoDB) and check out the samples demonstrating instant startup for your classic or reactive MongoDB application.

$ git clone https://github.com/spring-projects-experimental/spring-graalvm-native.git
$ cd spring-graalvm-native
$ git checkout 0.8.2
$ ./run-dev-container.sh
...@docker: cd spring-graalvm-native-samples/data-mongo
...@docker: ./build.sh
...@docker: ./target/data-mongodbCOPY