Mongodb sum 求和

Mongodb 統計某個字段的和，可以使用聚合函數。

數據結構如下：

> db.message.findOne()
{
    "_class" : "com.cyou.appwiki.bean.Message",
    "_id" : ObjectId("53997f80c451bf833288b882"),
    "content" : "power9li的回答2",
    "fromNick" : "power9li",
    "fromUid" : "5176923980",
    "pic1" : null,
    "pic2" : null,
    "pic3" : null,
    "qid" : "53997f4dc451bf833288b880",
    "time" : NumberLong("1402568587464"),
    "toUid" : "F34780D6A3CA543DB020D584A91CC8BB",
    "unreadCount" : 2
}

我要統計的是某個toUid的unreadCount的總數

我發現有兩種方法可以實現，第一種是mapReduce的方式，我在《MongodDB權威指南》書裏看到的：

 

db.runCommand({"group":{
    "ns":"message",
    "key":"toUid",
    "initial":{"total":0},
    "$reduce" : function(doc,prev){
        prev.total += doc.unreadCount;
    },
    "condition":{"toUid":"F34780D6A3CA543DB020D584A91CC8BB"}
}});

 

 返回的結果格式：

{
    "retval" : [
        {
            "total" : 13
        }
    ],
    "count" : 5,
    "keys" : 1,
    "ok" : 1
}

 

第二種方法是在 http://www.w3cschool.cc/mongodb/mongodb-aggregate.html 看到的：

db.message.aggregate([
    { $match : { "toUid" : "F34780D6A3CA543DB020D584A91CC8BB"}},
    { $group : { _id : "$toUid", num_tutorial : {$sum : "$unreadCount"} }}
]);

 

返回的結果：

{
    "result" : [
        {
            "_id" : "F34780D6A3CA543DB020D584A91CC8BB",
            "num_tutorial" : 13
        }
    ],
    "ok" : 1
}

 在使用spring-mongo進行開發的時候，也對應兩種方法

1：（老方法，支持spring-mongo所有版本）

public long findNewMessage(Integer appId, String userId) {
        Long total = 0l;
        String reduce = "function(doc, aggr){" +
                "            aggr.total += doc.unreadCount;" +
                "        }";
        Query query = Query.query(Criteria.where("toUid").is(userId).and("appId").is(appId));
        DBObject result = mongoTemplate.getCollection("message").group(new BasicDBObject("toUid", 1),
                query.getQueryObject(),
                new BasicDBObject("total", total),
                reduce);

        Map<String,BasicDBObject> map = result.toMap();
        if(map.size() > 0){
            BasicDBObject bdbo = map.get("0");
            if(bdbo != null && bdbo.get("total") != null)
                total = bdbo.getLong("total");
        }
        return total;
    }

 

方法2（只有spring-mongo 3.2以上才支持)

public long getNewMessageCount(Integer appId, String userId){
        Long total = 0l;
        Aggregation aggregation = Aggregation.newAggregation(
            match(Criteria.where("toUid").is(userId).and("appId").is(appId)),
            group("toUid").sum("unreadCount").as("total")
        );

        AggregationResults<StateStats> ar = mongoTemplate.aggregate(aggregation, "message", StateStats.class);
        List<StateStats> list = ar.getMappedResults();
        if(list.size() > 0){
            total = list.get(0).total;
        }
        return total;
    }