序列化之Apache Thrift

本文轉載自查看原文 2014-01-05 01:53 2964

Apache Thrift和ProtoBuf一樣，都是為優化序列化而生，Thrift是一個通信框架，最初有FaceBook開發，后交由Apache管理，目前Facebook也在使用。Thrift與ProtoBuf同樣是跨平台多語言的，不過Thrift幾乎支持現下的所有流行的語言，而ProtoBuf.net是.net的移植，相比而來Thrift支持更廣。我這里做的是對Thrift序列化數據在效率做實驗

1.定義接口文件

相當於數據契約。Thrift的契約定義用了自己的一套語法(IDL)，這樣做的目的是讓他可以跨語言，同一套契約可以在任何語言下使用，關於IDL方法可以去參考Thrift的白皮書。

和上一篇文章相同，有以下這樣兩個類

public class Person
    {
        public int Id { get; set; }]
        public string Name { get; set; }
        public Address Address { get; set; }
    }


    public class Address
    {
        public string Line1 { get; set; }
        public string Line2 { get; set; }
    }

用Thrift 的IDL語言定義后如下，文件保存為test.idl

struct Address
{
        1: string Line1,
        2: string Line2
}
struct Person
{
        1: i32 Id,
        2: string Name,
        3: Address Address
}

2. 使用Thrift提供的windows編譯器生成文件

使用thrift-0.9.1.exe生成文件，將test.idl與thrift-0.9.1.exe放在同級目錄下，在控制台是執行命令

thrift-0.9.1 --gen csharp test.idl

文件生成在目錄下的gen-csharp文件夾中

Address.cs/** * Autogenerated by Thrift Compiler (0.9.1) * * DO NOT EDIT UNLESS YOU ARE SURE THAT YOU KNOW WHAT YOU ARE DOING * @generated */
using System;
using System.Collections;
using System.Collections.Generic;
using System.Text;
using System.IO;
using Thrift;
using Thrift.Collections;
using System.Runtime.Serialization;
using Thrift.Protocol;
using Thrift.Transport;


#if !SILVERLIGHT
[Serializable]
#endif
public partial class Address : TBase
{
  private string _Line1;
  private string _Line2;

  public string Line1
  {
    get
    {
      return _Line1;
    }
    set
    {
      __isset.Line1 = true;
      this._Line1 = value;
    }
  }

  public string Line2
  {
    get
    {
      return _Line2;
    }
    set
    {
      __isset.Line2 = true;
      this._Line2 = value;
    }
  }


  public Isset __isset;
  #if !SILVERLIGHT
  [Serializable]
  #endif
  public struct Isset {
    public bool Line1;
    public bool Line2;
  }

  public Address() {
  }

  public void Read (TProtocol iprot)
  {
    TField field;
    iprot.ReadStructBegin();
    while (true)
    {
      field = iprot.ReadFieldBegin();
      if (field.Type == TType.Stop) { 
        break;
      }
      switch (field.ID)
      {
        case 1:
          if (field.Type == TType.String) {
            Line1 = iprot.ReadString();
          } else { 
            TProtocolUtil.Skip(iprot, field.Type);
          }
          break;
        case 2:
          if (field.Type == TType.String) {
            Line2 = iprot.ReadString();
          } else { 
            TProtocolUtil.Skip(iprot, field.Type);
          }
          break;
        default: 
          TProtocolUtil.Skip(iprot, field.Type);
          break;
      }
      iprot.ReadFieldEnd();
    }
    iprot.ReadStructEnd();
  }

  public void Write(TProtocol oprot) {
    TStruct struc = new TStruct("Address");
    oprot.WriteStructBegin(struc);
    TField field = new TField();
    if (Line1 != null && __isset.Line1) {
      field.Name = "Line1";
      field.Type = TType.String;
      field.ID = 1;
      oprot.WriteFieldBegin(field);
      oprot.WriteString(Line1);
      oprot.WriteFieldEnd();
    }
    if (Line2 != null && __isset.Line2) {
      field.Name = "Line2";
      field.Type = TType.String;
      field.ID = 2;
      oprot.WriteFieldBegin(field);
      oprot.WriteString(Line2);
      oprot.WriteFieldEnd();
    }
    oprot.WriteFieldStop();
    oprot.WriteStructEnd();
  }

  public override string ToString() {
    StringBuilder sb = new StringBuilder("Address(");
    sb.Append("Line1: ");
    sb.Append(Line1);
    sb.Append(",Line2: ");
    sb.Append(Line2);
    sb.Append(")");
    return sb.ToString();
  }

}

Person.cs/** * Autogenerated by Thrift Compiler (0.9.1) * * DO NOT EDIT UNLESS YOU ARE SURE THAT YOU KNOW WHAT YOU ARE DOING * @generated */
using System;
using System.Collections;
using System.Collections.Generic;
using System.Text;
using System.IO;
using Thrift;
using Thrift.Collections;
using System.Runtime.Serialization;
using Thrift.Protocol;
using Thrift.Transport;


#if !SILVERLIGHT
[Serializable]
#endif
public partial class Person : TBase
{
  private int _Id;
  private string _Name;
  private Address _Address;

  public int Id
  {
    get
    {
      return _Id;
    }
    set
    {
      __isset.Id = true;
      this._Id = value;
    }
  }

  public string Name
  {
    get
    {
      return _Name;
    }
    set
    {
      __isset.Name = true;
      this._Name = value;
    }
  }

  public Address Address
  {
    get
    {
      return _Address;
    }
    set
    {
      __isset.Address = true;
      this._Address = value;
    }
  }


  public Isset __isset;
  #if !SILVERLIGHT
  [Serializable]
  #endif
  public struct Isset {
    public bool Id;
    public bool Name;
    public bool Address;
  }

  public Person() {
  }

  public void Read (TProtocol iprot)
  {
    TField field;
    iprot.ReadStructBegin();
    while (true)
    {
      field = iprot.ReadFieldBegin();
      if (field.Type == TType.Stop) { 
        break;
      }
      switch (field.ID)
      {
        case 1:
          if (field.Type == TType.I32) {
            Id = iprot.ReadI32();
          } else { 
            TProtocolUtil.Skip(iprot, field.Type);
          }
          break;
        case 2:
          if (field.Type == TType.String) {
            Name = iprot.ReadString();
          } else { 
            TProtocolUtil.Skip(iprot, field.Type);
          }
          break;
        case 3:
          if (field.Type == TType.Struct) {
            Address = new Address();
            Address.Read(iprot);
          } else { 
            TProtocolUtil.Skip(iprot, field.Type);
          }
          break;
        default: 
          TProtocolUtil.Skip(iprot, field.Type);
          break;
      }
      iprot.ReadFieldEnd();
    }
    iprot.ReadStructEnd();
  }

  public void Write(TProtocol oprot) {
    TStruct struc = new TStruct("Person");
    oprot.WriteStructBegin(struc);
    TField field = new TField();
    if (__isset.Id) {
      field.Name = "Id";
      field.Type = TType.I32;
      field.ID = 1;
      oprot.WriteFieldBegin(field);
      oprot.WriteI32(Id);
      oprot.WriteFieldEnd();
    }
    if (Name != null && __isset.Name) {
      field.Name = "Name";
      field.Type = TType.String;
      field.ID = 2;
      oprot.WriteFieldBegin(field);
      oprot.WriteString(Name);
      oprot.WriteFieldEnd();
    }
    if (Address != null && __isset.Address) {
      field.Name = "Address";
      field.Type = TType.Struct;
      field.ID = 3;
      oprot.WriteFieldBegin(field);
      Address.Write(oprot);
      oprot.WriteFieldEnd();
    }
    oprot.WriteFieldStop();
    oprot.WriteStructEnd();
  }

  public override string ToString() {
    StringBuilder sb = new StringBuilder("Person(");
    sb.Append("Id: ");
    sb.Append(Id);
    sb.Append(",Name: ");
    sb.Append(Name);
    sb.Append(",Address: ");
    sb.Append(Address== null ? "<null>" : Address.ToString());
    sb.Append(")");
    return sb.ToString();
  }

}

3. 添加Thrift.dll

下載thrift-0.9.1.tar.gz解壓后找到對應的C#代碼編譯后就得到Thrift.dll了

4. 序列化數據

創建一個C#控制台應用程序，將生成的兩個代碼文件添加到項目中，再添加Thrift.dll引用，和上次一樣序列化同樣的1000條數據，得到的數據大小為48.7 KB (49,890 字節)

    class Program
    {
        static void Main(string[] args)
        {
            Person person = new Person { Address = new Address { Line1 = "Line1", Line2 = "Line2" }, Id = 1, Name = "Name" };
            using (System.IO.Stream stream = System.IO.File.Create("Person.dat"))
            {
                Thrift.Transport.TTransport transport = new Thrift.Transport.TStreamTransport(null,stream);
                Thrift.Protocol.TProtocol protocol = new Thrift.Protocol.TBinaryProtocol(transport);
                List<Person> list = new List<Person>();
                for (int i = 0; i < 1000; i++)
                {
                    Person item = new Person { Address = new Address { Line1 = "Line1", Line2 = "Line2" }, Id = i, Name = "Name" + i };
                    item.Write(protocol);
                }
            }

        }
    }

Demo下載

對比上次的ProtoBuf大了不少，和C#的原始二進制序列化只差5kb。當然我這次對比幾個數據也不能說明什么問題，只能做個大概了解。網絡通信是個復雜的過程。序列化數據大小直接影響網絡帶寬，序列化與反序列化的效率也直接影響着服務器資源。當數據量大小上升到不同數量級時ProtoBuf和Thrift到底誰更占上風我就法驗證了。

Thrift是個網絡通信框架，提供了多種數據序列化方式，並且支持多種語言，通過Thrift的IDL語言定義過的契約可以在所有支持的語言上使用，且與平台無關，這就是他NB的地方。從Thrift的Tutorial看來對現將通信協議遷移也不是件事。

免責聲明！

本站轉載的文章為個人學習借鑒使用，本站對版權不負任何法律責任。如果侵犯了您的隱私權益，請聯系本站郵箱yoyou2525@163.com刪除。

猜您在找 Apache Thrift系列詳解(三) - 序列化機制 Thrift 個人實戰--Thrift 的序列化機制 Thrift序列化與反序列化由淺入深了解Thrift之服務模型和序列化機制 JAVA RPC (四) 之thrift序列化普通對象 Thrift序列化與反序列化的實現機制分析 java序列化框架（protobuf、thrift、kryo、fst、fastjson、Jackson、gson、hessian）性能對比 Apache Shiro反序列化漏洞復現 Vulhub-Apache-Shiro反序列化漏洞 Apache Commons Collections反序列化研究