c# 解析JSON的几种办法

欲成为海洋大师，必知晓海中每一滴水的真名。

刚开始只是想找一个转换JSON数组的方法，结果在MSDN翻到一大把。

搜索过程中免不了碰到一大堆名词：WCF => DataContract => DataMember => DataContractJsonSerializer，然后就是不停的引入命名空间。

这一段经历，立即让我就联想到了上面引用的这句存在于某小说里面巫师的话语。静态语言真有点令人抓狂，不停的做准备，然后才可能开始工作。

对比

.NET下几种常见的解析JSON方法

主要类	命名空间	限制	内建LINQ支持
DataContractJsonSerializer	System.Runtime.Serialization.Json	通用	否
JavaScriptSerializer	System.Web.Script.Serialization	只能在Web环境使用	否
JsonArray、JsonObject、JsonValue	System.Json	只能在Silverlight中使用	是
JsonConvert、JArray、JObject、JValue、JProperty	Newtonsoft.Json	通用	是

准备数据

实体类：

    [DataContract]
    public class Person
    {
        [DataMember(Order = 0, IsRequired = true)]
        public string Name { get; set; }

        [DataMember(Order = 1)]
        public int Age { get; set; }

        [DataMember(Order = 2)]
        public bool Alive { get; set; }

        [DataMember(Order = 3)]
        public string[] FavoriteFilms { get; set; }

        [DataMember(Order = 4)]
        public Person Child { get; set; }
    }

定义：

Action<object> log = o => Console.WriteLine(o);
Func<int, int, int> add = (x, y) => x + y;

var p1 = new Person {
    Age = 12,
    Alive = true,
    Name = "lj",
    FavoriteFilms = new[] { "Up", "Avatar" }
};
var p2 = new Person() { Age = 28, Name = "cy", Child = p1 };
            

使用DataContractJsonSerializer

帮助类：

  // using System.Runtime.Serialization.Json;
    
    /// <summary>
    /// 解析JSON，仿Javascript风格
    /// </summary>
    public static class JSON
    {

        public static T parse<T>(string jsonString)
        {
            using (var ms = new MemoryStream(Encoding.UTF8.GetBytes(jsonString)))
            {
                return (T)new DataContractJsonSerializer(typeof(T)).ReadObject(ms);
            }
        }

        public static string stringify(object jsonObject)
        {
            using (var ms = new MemoryStream())
            {
                new DataContractJsonSerializer(jsonObject.GetType()).WriteObject(ms, jsonObject);
                return Encoding.UTF8.GetString(ms.ToArray());
            }
        }
    }

用法：

    // 序列化
    var jsonString = JSON.stringify(new[] { p1, p2 });
    log(jsonString == JSON.stringify(new List<Person>() { p1, p2 }));   //true
    log(jsonString);
    // 反序列化，泛型集合
    JSON.parse<List<Person>>(jsonString);
    // 数组转换            
    JSON.parse<Person[]>(jsonString);

输出：

[{"Name":"lj","Age":12,"Alive":true,"FavoriteFilms":["Up","Avatar"],"Child":null
},{"Name":"cy","Age":28,"Alive":false,"FavoriteFilms":null,"Child":{"Name":"lj",
"Age":12,"Alive":true,"FavoriteFilms":["Up","Avatar"],"Child":null}}]

使用JavaScriptSerializer

  // using System.Web.Script.Serialization;
    
    var jser    = new JavaScriptSerializer();
    var json    = jser.Serialize(new List<Person>() { p1, p2 });
    var persons = jser.Deserialize<List<Person>>(json);

使用Silverlight

  // using System.Json
    
    var css = "{ \"#header\" : {background:\"red\"}, layout : [5,4,1],color:\"cyan\" }";
    
    var style = JsonObject.Parse(css) as JsonObject;    
    
    (
    from s in style
    where s.Key == "color"
    select (string)s.Value
    ).First().ToString();	
    // "cyan"
    
    
    // 更多操作
    style["layout"][0] = 22;
    
    var hd = style["#header"];
    style["body>div+p"] = hd;
    style.Remove("#header");
    
    var bd = new JsonObject();
    bd["border"] = "1px solid cyan";
    style["body>div+p"]["#meta"] = bd;
    style.ToString();	
    // {"layout":[22,4,1],"color":"cyan","body>div+p":{"background":"red","#meta":{"border":"1px solid cyan"}}}

使用JSON.NET

  // using Newtonsoft.Json;
    
    var json = JsonConvert.SerializeObject(new[] { p1, p2 });
    var persons = JsonConvert.DeserializeObject<List<Person>>(json);
    var ja = JArray.Parse(jsonString);            
    log(ja);	//注意，格式化过的输出

输出：

[
  {
    "Name": "lj",
    "Age": 12,
    "Alive": true,
    "FavoriteFilms": [
      "Up",
      "Avatar"
    ],
    "Child": null
  },
  {
    "Name": "cy",
    "Age": 28,
    "Alive": false,
    "FavoriteFilms": null,
    "Child": {
      "Name": "lj",
      "Age": 12,
      "Alive": true,
      "FavoriteFilms": [
        "Up",
        "Avatar"
      ],
      "Child": null
    }
  }
]

LINQ：

    var ageCount = ja.Select(j => (int)j["Age"]).Aggregate(add);    
    var q = from j in ja
            where !j["Name"].Value<string>().Equals("lj")
            select (int)j["Age"];
    
    log(q.Aggregate(add) == ageCount);  //false

其他：

  // 与Linq to XML 相似的嵌套构造函数：
    var jo = new JObject(
                    new JProperty("age", persons.Select( p => p.Age)),
                    new JProperty("funny", true),
                    new JProperty("array", new JArray(new[] { 2, 4, 1 }))
                    );
    log(jo);
    
    // JObject 操作
    var css = "{ \"#header\" : {background:\"red\"}, layout : [5,4,1] }";
    var style = JObject.Parse(css);

    var bd = new JObject();
    bd["color"] = "1px solid cyan";

    style["border"] = bd;

    var hd = style["#header"];
    style["body>div+p"] = hd;

    hd.Parent.Remove();

    style["layout"][0] = 22;
    log(style);

输出：

    {
      "age": [
        12,
        28
      ],
      "funny": true,
      "array": [
        2,
        4,
        1
      ]
    }
    {
      "layout": [
        22,
        4,
        1
      ],
      "border": {
        "color": "1px solid cyan"
      },
      "body>div+p": {
        "background": "red"
      }
    }