yaml vs json

yaml vs json

我們知道yaml是json的超集，關於yaml和json的介紹我們可以看wiki yaml 和wiki json

yaml scala

yaml scala語言沒太好用的wrapper，最終使用的java提供的snakeyaml

定義

由於將yml格式的文件轉scala對象的時候需要符合JavaBean的規範，所以scala的對象定義的比較醜陋，最後就是這樣了：

package com.ximalaya.sparkcommon.model

import scala.beans.BeanProperty
/**
  * Created by todd.chen on 6/4/16.
  * email : [e-mail]
  */
class EmailConfigYaml {
  @BeanProperty var host: String = _
  @BeanProperty var username: String = _
  @BeanProperty var password: String = _
  @BeanProperty var auth: Boolean = _
  @BeanProperty var fromEmail: String = _
  override def toString = s"EmailConfig($host,$username,$password,$auth,$fromEmail)"
}

讀取和寫入：

  def getEmailConfigYaml = {
    val source =
      """
        |--- !!com.ximalaya.sparkcommon.model.EmailConfigYaml
        |host : smtp.sina.cn
        |username : cjuexuan
        |password : password
        |auth : true
        |fromEmail: [email protected]
        |
      """.stripMargin
    val yml = new Yaml()
    val email = yml.load(source).asInstanceOf[EmailConfigYaml]
    println(s"yaml2email:$email")
    val emailString = yml.dump(email)
    println(s"email2yaml:$emailString")
  }

json scala

現在我json最喜歡用的庫在scala中是play-jso
定義和讀寫也相當簡單，簡單過一下

定義

package com.ximalaya.sparkcommon.model

import play.api.libs.json.Json
/**
  * Created by todd.chen on 6/4/16.
  * email : [e-mail]
  */
case class EmailConfigJson(
                          host:String,
                          username:String,
                          password:String,
                          auth:String,
                          fromEmail:String
                          )
object EmailConfigJson{
  val format = Json.format[EmailConfigJson]
}

讀寫

  def getEmailConfigJson = {
    val source =
      """
        |{
        | "host" : "smtp.sina.cn",
        | "username" : "cjuexuan",
        | "password" : "password",
        | "auth" : "true",
        | "fromEmail" : "[email protected]"
        |}
      """.stripMargin
    val json = Json.parse(source)
    implicit val format = EmailConfigJson.format
    val email = json.as[EmailConfigJson]
    println(s"json2email:$email")
    val emailString = Json.toJson(email)
    println(s"email2json:$emailString")
  }

讀寫測試測試用例

  def test(func: ⇒ Unit) = {
    val start = Instant.now()
    func
    val end = Instant.now()
    val duration = Duration.between(start, end)
    println(s"duration time :${duration.toNanos}}")
  }

  def main(args: Array[String]): Unit = {
    test(getEmailConfigYaml)
    test(getEmailConfigJson)
  }

結果和結論

json2email:EmailConfigJson(smtp.sina.cn,cjuexuan,password,true,[e-mail])
email2json:{"host":"smtp.sina.cn","username":"cjuexuan","password":"password","auth":"true","fromEmail":"[email protected]"}
duration time :1586000000}
yaml2email:EmailConfig(smtp.sina.cn,cjuexuan,password,true,[e-mail])
email2yaml:!!com.ximalaya.sparkcommon.model.EmailConfigYaml {auth: true, fromEmail: [e-mail],
  host: smtp.sina.cn, password: password, username: cjuexuan}

duration time :253000000}

Process finished with exit code 0

交換順序進行測試

yaml2email:EmailConfig(smtp.sina.cn,cjuexuan,password,true,[e-mail])
email2yaml:!!com.ximalaya.sparkcommon.model.EmailConfigYaml {auth: true, fromEmail: [e-mail],
  host: smtp.sina.cn, password: password, username: cjuexuan}

duration time :676000000}
json2email:EmailConfigJson(smtp.sina.cn,cjuexuan,password,true,[e-mail])
email2json:{"host":"smtp.sina.cn","username":"cjuexuan","password":"password","auth":"true","fromEmail":"[email protected]"}
duration time :994000000}

Process finished with exit code 0

發現無論哪種方式下yaml的解析還是要快於json的，但使用bean對象的時候yaml的定義複雜於json，所以沒啥明顯性能問題的時候我還是傾向json