摘要

mongo aggregation是mongo 的一個輕量級的map-reduce框架，可以實現一些count,sum,groupby的聚合。本身的意義就在於減少查詢次數，將原本需要客戶端的多次查詢能夠整合到db端的基於pipeline的查詢。這樣帶來的好處就是減少請求資料庫次數，壞處就是這個查詢佔用的資料庫記憶體大，時間查，有可能拖垮其他查詢。這個在本篇文章不討論。

概述

常用的階段函式

• $match： 就是where查詢

• $group : 就是group by

• $project：新增新欄位/重置已有欄位，比如group階段的key值會變成_id就需要

• $unwind :將陣列拆分為幾個欄位

• $sort 排序

• $limit 限制條數

• $skip 跳過條數

例子

1. java 配置 mongo 連線

   @Configuration
   @EnableMongoRepositories(basePackages = "com.fs.mongo.dao")
   public class MongoConfig extends AbstractMongoConfiguration {
   
       @Bean
       public MongoDbFactory mongoDbFactory() throws UnknownHostException {
           MongoClientOptions options = new
    
    MongoClientOptions.Builder().connectionsPerHost(8).build();
           return new SimpleMongoDbFactory(new MongoClient("localhost",options ),"test");
       }
       // 預設資料庫會生成_class欄位，需要更改mappingMongoConverter的typeMappper屬性值
       @Bean
       public MongoTemplate mongoTemplate() throws Exception {
           return
    
    new MongoTemplate(mongoDbFactory(),mappingMongoConverter());
       }
   
       protected String getDatabaseName() {
           return "test";
       }
   
       @Bean
       public Mongo mongo() throws Exception {
           return new MongoClient();
       }
   
       @Bean
       @Override
       public MappingMongoConverter mappingMongoConverter() throws Exception {
           MappingMongoConverter converter = new         MappingMongoConverter(mongoDbFactory(),this.mongoMappingContext());
           converter.setTypeMapper(new DefaultMongoTypeMapper(null));
           return converter;
       }
   }

2. 基於mongoTemplate 傳送aggregate請求

   @Component
   public class PostDb {
       @Autowired
       private MongoTemplate mongoTemplate;
   
       public PostDb(MongoTemplate mongoTemplate) {
           this.mongoTemplate = mongoTemplate;
       }
       public List<Post> getPeopleAggregationByCategory(long userId) {
           MatchOperation match = Aggregation.match(new Criteria("userId").is(userId));
           GroupOperation group  = Aggregation.group("category").count().as("total");
           // 注group key category會對映成_id，所以要利用project階段映射回category
           ProjectionOperation project =  Aggregation.project("total").and("_id").as("category");
           Aggregation aggregation = Aggregation.newAggregation(match,group,project);
           AggregationResults<Post> results = mongoTemplate.aggregate(aggregation,"post",Post.class);
           return results.getMappedResults();
       }
   }

掃碼關注NoSQL開發公眾號，第一時間接收文章推送