如何在JSON中为空的DataTable包含列元数据

Question

我正在寻找适当地描述JSON中的列元数据的方法，稍后由Newtonsoft对其进行解析以构建C＃DataTable。这样，我希望解决一个DataTable没有行或没有列的a的问题，但是即使我传递一个空表，我也需要使用标签和希望的数据类型来创建列。

标准输入的示例：

{
    "BrokerID": "998",
    "AccountID": "1313",
    "Packages": [
        {
            "PackageID": 226,
            "Amount": 15000,
            "Auto_sync": true,
            "Color": "BLUE"
        },
        {
            "PackageID": 500,
            "Amount": 15000,
            "Auto_sync": true,
            "Color": "PEACH"
        }
    ]
}

输入表为空的示例：

{
    "BrokerID" : "998",
    "AccountID" : "1313",
    "Packages":[]
}

当我使用解析此内容时JsonConvert.DeserializeObject<DataTable>(params["Packages"]);，没有行，而且显然没有列。我正在寻找一种描述JSON正文中的列元数据的方法。

Answer 1

该DataTableConverter附带Json.Net不输出列的元数据，即使你设置TypeNameHandling到All。但是，没有什么可以阻止您制作自己的自定义转换器来执行此操作，而是使用该自定义转换器。以下是我根据您的需求拼凑而成的一种：

class CustomDataTableConverter : JsonConverter
{
    public override bool CanConvert(Type objectType)
    {
        return (objectType == typeof(DataTable));
    }

    public override void WriteJson(JsonWriter writer, object value, JsonSerializer serializer)
    {
        DataTable dt = (DataTable)value;
        JObject metaDataObj = new JObject();
        foreach (DataColumn col in dt.Columns)
        {
            metaDataObj.Add(col.ColumnName, col.DataType.AssemblyQualifiedName);
        }
        JArray rowsArray = new JArray();
        rowsArray.Add(metaDataObj);
        foreach (DataRow row in dt.Rows)
        {
            JObject rowDataObj = new JObject();
            foreach (DataColumn col in dt.Columns)
            {
                rowDataObj.Add(col.ColumnName, JToken.FromObject(row[col]));
            }
            rowsArray.Add(rowDataObj);
        }
        rowsArray.WriteTo(writer);
    }

    public override object ReadJson(JsonReader reader, Type objectType, object existingValue, JsonSerializer serializer)
    {
        JArray rowsArray = JArray.Load(reader);
        JObject metaDataObj = (JObject)rowsArray.First();
        DataTable dt = new DataTable();
        foreach (JProperty prop in metaDataObj.Properties())
        {
            dt.Columns.Add(prop.Name, Type.GetType((string)prop.Value, throwOnError: true));
        }
        foreach (JObject rowDataObj in rowsArray.Skip(1))
        {
            DataRow row = dt.NewRow();
            foreach (DataColumn col in dt.Columns)
            {
                if (rowDataObj[col.ColumnName].Type != JTokenType.Null)//Skip if the Value is Null/Missing, especially for a non-nullable type.
                    row[col] = rowDataObj[col.ColumnName].ToObject(col.DataType);
            }
            dt.Rows.Add(row);
        }
        return dt;
    }
}

这是一个演示。请注意，序列化表时，列类型被写为JSON中数组的第一行。反序列化时，即使没有其他行，此元数据也将用于使用正确的列类型和名称重建表。（您可以通过在顶部将行数据添加到表中的两行注释掉来验证这一点。）

class Program
{
    static void Main(string[] args)
    {
        DataTable dt = new DataTable();
        dt.Columns.Add("PackageID", typeof(int));
        dt.Columns.Add("Amount", typeof(int));
        dt.Columns.Add("Auto_sync", typeof(bool));
        dt.Columns.Add("Color", typeof(string));
        // Comment out these two lines to see the table with no data.
        // Test with a null Value for a Non-Nullable DataType.
        dt.Rows.Add(new object[] { 226,  null, true, "BLUE" });
        dt.Rows.Add(new object[] { 500, 15000, true, "PEACH" });

        Foo foo = new Foo
        {
            BrokerID = "998",
            AccountID = "1313",
            Packages = dt
        };

        JsonSerializerSettings settings = new JsonSerializerSettings();
        settings.Converters.Add(new CustomDataTableConverter());
        settings.Formatting = Formatting.Indented;

        string json = JsonConvert.SerializeObject(foo, settings);
        Console.WriteLine(json);
        Console.WriteLine();

        Foo foo2 = JsonConvert.DeserializeObject<Foo>(json, settings);
        Console.WriteLine("BrokerID: " + foo2.BrokerID);
        Console.WriteLine("AccountID: " + foo2.AccountID);
        Console.WriteLine("Packages table:");
        Console.WriteLine("  " + string.Join(", ", 
            foo2.Packages.Columns
                .Cast<DataColumn>()
                .Select(c => c.ColumnName + " (" + c.DataType.Name + ")")));

        foreach (DataRow row in foo2.Packages.Rows)
        {
            Console.WriteLine("  " + string.Join(", ", row.ItemArray
                .Select(v => v != null ? v.ToString() : "(null)")));
        }
    }
}

class Foo
{
    public string BrokerID { get; set; }
    public string AccountID { get; set; }
    public DataTable Packages { get; set; }
}

输出：

{
  "BrokerID": "998",
  "AccountID": "1313",
  "Packages": [
    {
      "PackageID": "System.Int32, mscorlib, Version=4.0.0.0, Culture=neutral, PublicKeyToken=b77a5c561934e089",
      "Amount": "System.Int32, mscorlib, Version=4.0.0.0, Culture=neutral, PublicKeyToken=b77a5c561934e089",
      "Auto_sync": "System.Boolean, mscorlib, Version=4.0.0.0, Culture=neutral, PublicKeyToken=b77a5c561934e089",
      "Color": "System.String, mscorlib, Version=4.0.0.0, Culture=neutral, PublicKeyToken=b77a5c561934e089"
    },
    {
      "PackageID": 226,
      "Amount": null,
      "Auto_sync": true,
      "Color": "BLUE"
    },
    {
      "PackageID": 500,
      "Amount": 15000,
      "Auto_sync": true,
      "Color": "PEACH"
    }
  ]
}

BrokerID: 998
AccountID: 1313
Packages table:
  PackageID (Int32), Amount (Int32), Auto_sync (Boolean), Color (String)
  226, , True, BLUE
  500, 15000, True, PEACH

小提琴：https :
//dotnetfiddle.net/GGrn9z