结合MongoDB开发LBS应用——附近的人

"The palest ink is better than the best memory"——好记性不如烂笔头。2013~2015补记

题记：

2014.7月业务数据增长，附近人脉功能模块导致Mysql数据库成为服务器瓶颈

关键词：

附近的人、LBS、Geo、geoNear、maxDistance、distanceMultiplier、2dsphere

方案：

1.基于Mysql数据库,ABS函数（当前出现问题）
2.基于Mysql采用GeoHash索引(生产环境5.1版本不支持，全库升级比较麻烦)
3.基于Solr的geodist和geohash（上家公司做附近商铺实现过，目前还没引入Solr）
4.基于Mongo的geo(当前正在使用Mongo，值得尝试一下，和Solr比较比较)

MongoDB原生支持地理位置索引，可以直接用于位置距离计算和查询。查询结果默认将会由近到远排序，而且查询结果也包含目标点对象、距离目标点的距离等信息

代码:

1.mongo库中数据格式：

 {
   "_id": ObjectId("582c13293e5467eefbd94ad9"),
   "uid": NumberInt(2448286),
   "gps": {
     "lon": 121.470238,
     "lat": 31.272408
  },
   "flag": NumberInt(1),
   "createTime": NumberLong(1479440591566)
}   

2.需要添加geo索引（https://docs.mongodb.com/manual/core/geospatial-indexes/）：

db.mydb.createIndex( { "gps": "2d" } )

3.核心java实现代码（建议先熟悉官方文档，写好shell）：

    /**
     * 保存或更新GPS
     * @param uid
     * @param longitude
     * @param latitude
     * @param flag
     * @param createTime
     * @return
     * @throws Exception
     */
@Override
public int saveOrUpdateGPS(int uid, double longitude, double latitude,
            int flag,long createTime) throws Exception {

        if(longitude < -180 || longitude > 180 || latitude < -90 || latitude > 90){
            return -1;
        }

        DBObject where = new BasicDBObject("uid", uid);

        DBCollection dbc = this.getUserGPSCollection();

        DBObject gps = new BasicDBObject();
        gps.put("lon", longitude);
        gps.put("lat", latitude);

        DBObject doc = new BasicDBObject();
        doc.put("uid", uid);
        doc.put("gps", gps);
        doc.put("flag", flag);
        doc.put("createTime", createTime);

        DBObject newDoc = new BasicDBObject();
        newDoc.put("$set", doc);

        return dbc.update(where, newDoc, true, true).getN();
    }

    /**
     * 获取附近用户
     * <p>
     * 
     * <Strong>MongoDB实现参考官方资料：</Strong>
     * <br>
     * 1.<link>http://docs.mongodb.org/manual/core/aggregation-introduction/</link>
     * 2.<link>http://docs.mongodb.org/manual/reference/operator/aggregation/geoNear/</link>
     * 
     * <p>
     * @param uid
     * @param startTime
     * @param longitude
     * @param latitude
     * @param maxDistance(单位KM)
     * @param maxNum
     * @param page
     * @param pageSize
     * @return
     * @throws Exception
     */
@Override
public List<JSONObject> getNearUIDS(int uid,long startTime, double longitude,
            double latitude, double maxDistance,int maxNum ,int page,int pageSize) throws Exception {

        DBObject near = new BasicDBObject("lon", longitude).append("lat", latitude);

        DBObject where = new BasicDBObject();
        where.put("flag", 1);
        where.put("uid", new BasicDBObject(QueryOperators.NE,uid));
        where.put("createTime", new BasicDBObject(QueryOperators.GTE, startTime));

        DBObject geoFields = new BasicDBObject();
        geoFields.put("geoNear", "user_gps");
        geoFields.put("near", near);
        geoFields.put("spherical", "true");
        geoFields.put("maxDistance", maxDistance/6371);
        geoFields.put("query", where);
        geoFields.put("distanceMultiplier", 6371);//注意
        geoFields.put("distanceField", "dis");
        geoFields.put("num", maxNum);

        DBObject geo = new BasicDBObject("$geoNear",geoFields);

        DBObject projectFields = new BasicDBObject();
        projectFields.put("_id", 0);
        projectFields.put("uid", 1);
        projectFields.put("dis", 1);
        projectFields.put("createTime", 1);

        DBObject project = new BasicDBObject("$project", projectFields );

//      DBObject orderBy = new BasicDBObject("dis",1);
//      orderBy.put("createTime", -1);
//      DBObject sort = new BasicDBObject("$sort",orderBy);//整体sort可以去掉

        DBObject skip = new BasicDBObject("$skip",(page - 1) * pageSize);

        DBObject limit = new BasicDBObject("$limit",pageSize);

        DBCollection dbc = this.getUserGPSCollection();

//      AggregationOutput out = dbc.aggregate(geo,project,sort,skip,limit);
        AggregationOutput out = dbc.aggregate(geo,project,skip,limit);
        Iterable<DBObject> res = out.results();
        Iterator<DBObject> iterator =  res.iterator();

        List<JSONObject> resList = new LinkedList<JSONObject>();

        while(iterator.hasNext()){
            DBObject one = iterator.next();
            JSONObject json = JSONObject.fromObject(one.toString());
            resList.add(json);
        }

        return resList;
    }

实际产品：

（不显示实际具体距离，是有意而为之）

参考：

1.http://docs.mongodb.org/manual/core/aggregation-introduction/
2.http://docs.mongodb.org/manual/reference/operator/aggregation/geoNear/
3.http://www.infoq.com/cn/articles/depth-study-of-Symfony2
4.https://wiki.apache.org/solr/SpatialSearch/