Count operation returns 1 instead of 0 after grouping array in Spring Data MongoDB Aggregation

Question

I need to create advanced aggregation using Spring Data MongoDB having model like that:

@Getter
@Setter
@Document
public class Library {

  @Id
  @JsonSerialize(using = ToStringSerializer.class)
  private ObjectId id;

  private Address address;

  private String workingHours;

  ...

}

@Getter
@Setter
@Document
public class Book {

  @Id
  @JsonSerialize(using = ToStringSerializer.class)
  private ObjectId id;

  private Boolean published;

  private Boolean hidden;

  private String title;

  @JsonSerialize(using = ToStringSerializer.class)
  private ObjectId libraryId;

  ...

}

pom.xml

<dependency>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-starter-data-mongodb</artifactId>
</dependency>
<dependency>
        <groupId>org.springframework.data</groupId>
        <artifactId>spring-data-mongodb</artifactId>
        <version>2.2.0</version>
</dependency>

Libraries collection:

{ 
    "_id" : ObjectId("5f45440ee89590218e83a697"), 
    "workingHours" : "8:00 PM - 8:00 AM",
    "address" : DBRef("addresses", ObjectId("5f4544198da452a5523e3d11"))
}

Books collection:

{ 
    "_id" : ObjectId("5f454423be823729015661ed"), 
    "published": true,
    "hidden": false,
    "title": "The Hobbit, or There and Back Again"
    "libraryId": ObjectId("5f45440ee89590218e83a697")
},
{ 
    "_id" : ObjectId("5f45445b876d08649b88ed5a"), 
    "published": true,
    "hidden": false,
    "title": "Harry Potter and the Philosopher's Stone"
    "libraryId": ObjectId("5f45440ee89590218e83a697")
},
{ 
    "_id" : ObjectId("5f45446c7e33ca70363f629a"), 
    "published": true,
    "hidden": false,
    "title": "Harry Potter and the Cursed Child"
    "libraryId": ObjectId("5f45440ee89590218e83a697")
},
{ 
    "_id" : ObjectId("5f45447285f9b3e4cb8739ad"), 
    "published": true,
    "hidden": false,
    "title": "Fantastic Beasts and Where to Find Them"
    "libraryId": ObjectId("5f45440ee89590218e83a697")
},
{ 
    "_id" : ObjectId("5f45449fc121a20afa4fbb96"), 
    "published": false,
    "hidden": false,
    "title": "Universal Parks & Resorts"
    "libraryId": ObjectId("5f45440ee89590218e83a697")
},
{ 
    "_id" : ObjectId("5f4544a5f13839bbe89edb23"), 
    "published": false,
    "hidden": true,
    "title": "Ministry of Dawn"
    "libraryId": ObjectId("5f45440ee89590218e83a697")
}

Depending on the context of the user, I have to return a different count of books that can be filtered based on startsWith() or like() principle.

Assuming that I have 4 published books, one more added by normal user and one more hidden.

1. Administrator should know about all of them, so he will see booksCount as 6 .
2. The regular user sees only published or added by himself, so he will see booksCount as 5 .
3. There can be some other conditions in the future.

I came up with aggregation like this:

Criteria criteria = Criteria.where("_id").ne(null).and("address.city").is("Chicago");

MatchOperation matchOperation = Aggregation.match(criteria);
            
LookupOperation lookupOperation = LookupOperation.newLookup().from("books").localField("_id").foreignField("topicId").as("books");

UnwindOperation unwindOperation = Aggregation.unwind("books", true);

MatchOperation secondMatchOperation = Aggregation.match(Criteria.where("books.published").is(Boolean.TRUE).orOperator(Criteria.where("creator.userId").is(context.getUserId()));

AggregationOperation group = Aggregation.group("_id")
            .first("_id").as("id")
            .first("published").as("published")
            .first("title").as("title")
            .push("books").as("books").count().as("booksCount");

Aggregation aggregation = !isAdministrator() ?
Aggregation.newAggregation(matchOperation, lookupOperation, unwindOperation, secondMatchOperation, group) : 
Aggregation.newAggregation(matchOperation, lookupOperation, unwindOperation, group);
            
mongoTemplate.aggregate(aggregation, "libraries", Document.class).getRawResults().get("results");

Everything works fine instead of count() operation.

1. If books array size is 0 it always returns 1.
2. If books array size is greater than 0 it returns the proper amount.
3. I've tried using newBuilder(GroupOps.SUM, null, 0) instead of count() , but now it always return 0.
4. If I use newBuilder(GroupOps.SUM, null, 2) it returns size + 2 . I don't know what is going on.

My quesions:

1. Can anyone tell me what I am doing wrong and how to correct it?
2. In addition I need to parse "booksCount" from Integer to String . Is it possible in group stage?

Thank you in advance.

Answer 1

It happens because of Aggregation.unwind("books", true); . When there is no join, this statement keeps as the document unless you make it as Aggregation.unwind("books") . Default behavior is false . Then when you counting, the document is counted as a document. That's why its giving you the 1 as output. Example with wrong output

So what you can do is, you can count the size in the next stage.

 project("_id", "published", "title", "books")
   .and(ArrayOperators.Size.lengthOfArray(ConditionalOperators.ifNull("books").then(Collections.emptyList()))).as("booksCount")

Working Mongo playground with correct answer