Spring Data MongoTemplate多字段去重聚合查询遇空结果求助

问题背景 #

作为MongoDB聚合查询新手，需在userDetails集合中对userId和user字段执行去重操作。在Robo 3T中执行以下聚合语句可得到正确结果，但对应的Spring Data MongoTemplate代码返回空结果：

原MongoDB聚合语句 #

db.userDetails.aggregate([
    {"$match" : {zip:"52034", city:"Delhi"}}, 
    // 按userId和user分组实现去重
    {"$group" : {_id : {userId:"$userId", user:"$user"}}}, 
    // 整理输出结构
    {"$project" : {_id:0, userId:"$_id.userId", user:"$_id.user"}}
])

出错的Java代码 #

MatchOperation matchOperation = Aggregation.match(
          new Criteria("zip").is(zipCode).and("city").is(cityName));
GroupOperation groupOperation = new GroupOperation(Fields.fields("userId", "user"));
ProjectionOperation projectionOperation = Aggregation.project().andExclude("_id").and("userId").nested(
        Fields.fields("_id.userId"))
        .and("user").nested(Fields.fields("_id.user"));
Aggregation  aggregation = Aggregation.newAggregation(matchOperation, groupOperation, projectionOperation);
AggregationResults<UserDTO> result = mongoTemplate.aggregate(aggregation, UserDetails.class, UserDTO.class);

版本信息：Spring Data MongoDB 3.0.7，Driver 4.0.4

问题原因 #

出错核心在于ProjectionOperation的写法错误：使用nested()方法会创建嵌套文档结构，而非从_id中提取字段值映射到顶级字段，导致最终返回的文档结构与UserDTO不匹配，因此无法正确映射得到结果。

修正后的代码 #

MatchOperation matchOperation = Aggregation.match(
        new Criteria("zip").is(zipCode).and("city").is(cityName));

// GroupOperation写法正确，无需修改
GroupOperation groupOperation = new GroupOperation(Fields.fields("userId", "user"));

// 修正ProjectionOperation：直接提取_id中的字段映射为顶级字段
ProjectionOperation projectionOperation = Aggregation.project()
        .andExclude("_id")
        .and("_id.userId").as("userId")
        .and("_id.user").as("user");

Aggregation aggregation = Aggregation.newAggregation(matchOperation, groupOperation, projectionOperation);
AggregationResults<UserDTO> result = mongoTemplate.aggregate(aggregation, UserDetails.class, UserDTO.class);

说明 #

修正后的ProjectionOperation会生成与原MongoDB聚合语句一致的$project阶段：

{"$project" : {_id:0, userId:"$_id.userId", user:"$_id.user"}}

这样返回的文档结构能正确匹配UserDTO的字段，即可得到预期的去重结果。

内容的提问来源于stack exchange，提问作者user2068260