比如有一个复杂对象：
Map[Int, Map[Int, Double]]
需要将其转为JSON保存，之后再读取使用，试了几种方法，最后的方案是：
1、定义case class
2、所有的数据类型都转为String（避免不必要的麻烦，至少Map的key都要为String，不然会报错scala.MatchError）
3、Map必须是immutable.Map

代码如下

1、转json

import org.json4s.JsonDSL._
import org.json4s.jackson.JsonMethods._
import org.json4s.jackson.Serialization._
import org.json4s.jackson.Serialization
case class WOE(col: String, woe: Map[String, String])
implicit val formats = Serialization.formats(NoTypeHints)
val testMap = Map[String, String]()
testMap += ("1" -> "1.1")
val a = WOE("1", immutable.Map(testMap.toList:_*))
println(write(a))

输出{"col":"1","woe":{"1":"1.1"}}

2、解析json

implicit val formats = Serialization.formats(NoTypeHints)
val js =
"""
{"col":"1","woe":{"1":"1.1"}}
"""
val ab = parse(js).extract[WOE]
println(write(ab))

如果是List也可以

implicit val formats = Serialization.formats(NoTypeHints)

    val b = new ListBuffer[WOE]
    val testMap = Map[String, String]()
    testMap += ("1" -> "1.1")
    b += WOE("1", immutable.Map(testMap.toList:_*))
    b += WOE("3", immutable.Map(testMap.toList:_*))
    println(write(b))

val js =
      """
 [{"col":"1","woe":{"1":"1.1"}},{"col":"3","woe":{"1":"1.1"}}]
      """
    val ab = parse(js).extract[List[WOE]]
    println(ab.toString)

之前整理的一些方法
1、scala自带的Json解析
scala 2.10（以上，其他版本不清楚）自带Json解析，scala.util.parsing.json.JSON
object转json

val testMap = Map[String, String]()
    testMap += ("1" -> "2.034")
    testMap += ("2" -> "2.0134")
    println(scala.util.parsing.json.JSONObject(scala.collection.immutable.Map(testMap.toList: _*)))

但好像只能处理map，且map要转成immutable

2、fastjson

解析json

import com.alibaba.fastjson.JSON
object JsonDemo {
  def main(args: Array[String]) {
    val text = "{\"name\":\"name1\", \"age\":55}"
    val json = JSON.parseObject(text)
    println(json.get("name"))
    println(json.get("age"))
  }
}

再例如

import com.alibaba.fastjson.JSON

object Json {
  def main(args: Array[String]): Unit = {
    val str2 = "{\"et\":\"kanqiu_client_join\",\"vtm\":1435898329434,\"body\":{\"client\":\"866963024862254\",\"client_type\":\"android\",\"room\":\"NBA_HOME\",\"gid\":\"\",\"type\":\"\",\"roomid\":\"\"},\"time\":1435898329}"
       val json=JSON.parseObject(str2)
       //获取成员
       val fet=json.get("et")
       //返回字符串成员
       val etString=json.getString("et")
       //返回整形成员
       val vtm=json.getInteger("vtm")
       println(vtm)
       //返回多级成员
       val client=json.getJSONObject("body").get("client")
       println(client)

在spark-steaming中,使用fast-json更加稳定,json-lib经常出现莫名问题,而且fastjson的解析速度更快.

object转json，首先必须要显式的定义参数，否则会报错

ambiguous reference to overloaded definition
1
例如：

val testMap = Map[String, String]()
testMap += ("1" -> "2.034")
testMap += ("2" -> "2.0134")
val a = JSON.toJSONString(testMap, true)
println(a)

不会报错，但是输出结果是奇怪的

{
    "empty":false,
    "sizeMapDefined":false,
    "traversableAgain":true
}

3、json4s

object转json

val testMap = Map[String, String]()
testMap += ("1" -> "2.034")
testMap += ("2" -> "2.0134")
val jj = compact(render(testMap))
println(jj)

输出

[{"2":"2.0134"},{"1":"2.034"}]

如果都是String，复杂的Map结构也可以解析

val testMap = Map[String, Map[String, String]]()
val subMap = Map[String, String]()
subMap += ("1" -> "1.1")
testMap += ("1" -> subMap)
println(write(testMap))

输出{"1":{"1":"1.1"}}
但这样的形式不利于解析

再例如

implicit val formats = Serialization.formats(NoTypeHints)
val m = Map(
      "name" -> "john doe",
      "age" -> 18,
      "hasChild" -> true,
      "childs" -> List(
        Map("name" -> "dorothy", "age" -> 5, "hasChild" -> false),
        Map("name" -> "bill", "age" -> 8, "hasChild" -> false)))
    val mm = Map(
      "1" -> Map ("1"->"1.2")
    )
println(write(a))

TEST

package com.dfssi.dataplatform.analysis.exhaust.alarm

import java.sql.Timestamp
import java.util

import com.alibaba.fastjson.serializer.SerializerFeature
import org.apache.spark.Logging
import org.json4s.NoTypeHints


//将要解析得数据
case class NeedEntity(val vin: String,
                      val downoutput: Double,
                      val collectTime: Long,
                      val lon: Double,
                      val lat: Double,
                      val failureList: java.util.List[Integer] = new util.ArrayList[Integer]()
                     ) extends Serializable

//管理状态
//这是事件管理得 按照每个事件来处理



class OverLimitEvent(var vin: String,
                     var startTime: Long,
                     var startLon: Double,
                     var startLat: Double,
                     var eventType:String="overlimit",
                      var endTime: Long = 0,
                      var endLon: Double = 0.0,
                      var endLat: Double = 0.0,
                      var minValue: Double = 0.0,
                      var maxValue: Double = 0.0
                         ) extends Serializable with Logging{

  def getInsertMap(): Map[String, Any] = {
    Map(
      "vin" -> vin,
      "startTime" -> new Timestamp(startTime),
      "startLon" -> startLon,
      "startLat" -> startLat
    )
  }

  def getUpdateMap(): Map[String, Any] = {
    Map(
      "vin" -> vin,
      "startTime" -> new Timestamp(startTime),
      "endTime" -> new Timestamp(startTime),
      "endLon" -> startLon,
      "endLat" -> startLat,
      "maxValue" -> maxValue,
      "minValue" -> minValue
    )
  }

  def updateByEntity(entity: NeedEntity) = {
    this.endTime = entity.collectTime
    this.endLat = entity.lat
    this.endLon = entity.lon
    if (this.maxValue != null && this.maxValue < entity.downoutput) {
      this.maxValue = entity.downoutput
    }
    if (this.minValue != null && this.minValue > entity.downoutput) {
      this.minValue = entity.downoutput
    }

  }

  override  def toString(): String ={
    import org.json4s.jackson.Serialization._
    import org.json4s.jackson.Serialization
    implicit val formats = Serialization.formats(NoTypeHints)
    write(this)
  }

}



object OverLimitEvent {
  val ID_FIELD = Array("vin", "startTime")
  def apply(
             vin: String,
             startTime: Long,
             startLon: Double,
             startLat: Double,
             endTime: Long,
             endLon: Double,
             endLat: Double,
             minValue: Double,
             maxValue: Double
           ): OverLimitEvent = {
    val event = new OverLimitEvent(vin, startTime, startLon, startLat)
    event.endTime = endTime
    event.endLat = endLat
    event.endLon = endLon
    event.maxValue = maxValue
    event.minValue = minValue
    event
  }

  def buildByEntity(entity: NeedEntity): OverLimitEvent = {
    new OverLimitEvent(entity.vin, entity.collectTime, entity.lon, entity.lat)
  }

  def buildByJson(json: String): OverLimitEvent = {
    com.alibaba.fastjson.JSON.parseObject(json, classOf[OverLimitEvent])
  }

  override  def toString(): String ={
    import org.json4s.jackson.Serialization._
    import org.json4s.jackson.Serialization
    implicit val formats = Serialization.formats(NoTypeHints)
    write(this)
  }

}

case class ExhaustAlarmStatus(val vin: String, var overLimitEvent: OverLimitEvent=null,var faultEvent:Map[String,OverLimitEvent]=null, var lastTime: Long) {
  override  def toString(): String ={
    import org.json4s.jackson.Serialization._
    import org.json4s.jackson.Serialization
    implicit val formats = Serialization.formats(NoTypeHints)
    write(this)
  }
}

object ExhaustAlarmStatus {
  def buildByJson(json: String): ExhaustAlarmStatus = {
    if(json!=null){
        com.alibaba.fastjson.JSON.parseObject(json,
        classOf[ExhaustAlarmStatus])
    }else{
      null
    }
  }

  def toJSON(state: ExhaustAlarmStatus): String = com.alibaba.fastjson.JSON.toJSONString(state, SerializerFeature.PrettyFormat)

  def main(args: Array[String]): Unit = {
    val json = "{\"vin\":\"222\", \"OverLimitEvent\":{ \"vin\":\"222\",  \"startTime\":123456789, \"startLon\":1.0, \"startLat\":1.0, \"endTime\":123456789, \"endLon\":1.0, \"endLat\":1.0, \"minValue\":1.0, \"maxValue\":1.0 },\"lastTime\":1556441242000}";
    val state  = com.alibaba.fastjson.JSON.parseObject(json,
      classOf[ExhaustAlarmStatus])
    println(state.overLimitEvent)
    import org.json4s.JsonDSL._
    import org.json4s.jackson.JsonMethods._
    import org.json4s.jackson.Serialization._
    import org.json4s.jackson.Serialization
    implicit val formats = Serialization.formats(NoTypeHints)
    val jsonstr = write(state)
    println(jsonstr)
  }


}