【C#】通过Thrift操作HBase
题外话:C# 调用 Java 的几种方法
1.将Java端的接口通过WebService方式发布,C#可以方便的调用
2.先使用C++ 通过 JNI 调用 Java,C# 调用C++的接口
3.使用开源的库直接使用C#调用Java ,详细信息请点击
4.使用IKVM实现C#调用Java,参考:http://www.ikvm.net/
之所以说这些,是因为自己这边客户端要调用HBase接口(Java实现),刚开始我是使用WS方式实现调用,这种方式很简单,而且通用性好。之后一段时间发现了上面所说的第三种方式,并成功调用,但是写这个库的哥们,好像没有维护自己写的这个库,里面有几个很明显的BUG,而且在循环调用的时间,会报内存错误,由于对JNI不太熟悉,也就放弃了这种方式,如果对这种方式感兴趣的童鞋可以给他完善一下,再提个醒,这个开源库依赖jvm.dll,只有32位的JavaJDK才行。至于第二种和第四种方式没有深入研究,在这也就不说了。
最终我并没有采用上面的任何一种方式,而我采用的是Thrift方式,虽然比Java API 慢一点,但也在可接受的范围之内。接下来就要进入正题了:
准备阶段:
1. 下载 Thrift 的源代码包,http://thrift.apache.org/
2. 下载 Thrift compiler for Windows ,http://www.apache.org/dyn/closer.cgi?path=/thrift/0.9.0/thrift-0.9.0.exe
生成Thrfit接口类:
1. 从HBase包中,得到HBase.Thrift文件。(..\hbase-0.94.6.1\src\main\resources\org\apache\hadoop\hbase\thrift 在此目录下)
2. 将Thrift-0.9.0.exe 与 HBase.Thrift文件放到同意目录下(当然也可以不在同一目录)
3. 进入命令行, Thrift-0.9.0.exe -gen CSharp HBase.Thrift此目录下就成了名为gen-csharp的文件夹
构建解决方案
该准备的项目代码都已完成,新建VS Project , 将Thrift 的源代码项目与刚刚生成的接口接口类引入。

开始集群的Thrift服务
hbase-daemon.sh start thrift 端口号默认9090
编写测试代码

第一篇(查看)已经介绍了关于通过Thrift访问HBase基本使用,这一篇说一些关于通过Thrift对HBase进行查询的操作。通过HBase.Thrift 所生成的接口类中,其中Hbase.cs为核心类,所有客户端操作HBase的接口的定义和实现都在此类中,如果想查看Thrift服务端的代码,请参考HBase源代码。
以下是类中所定义的查询接口:
1 List<TRowResult> getRow(byte[] tableName, byte[] row, Dictionary<byte[], byte[]> attributes); 2 List<TRowResult> getRowWithColumns(byte[] tableName, byte[] row, List<byte[]> columns, Dictionary<byte[], byte[]> attributes); 3 List<TRowResult> getRowTs(byte[] tableName, byte[] row, long timestamp, Dictionary<byte[], byte[]> attributes); 4 List<TRowResult> getRowWithColumnsTs(byte[] tableName, byte[] row, List<byte[]> columns, long timestamp, Dictionary<byte[], byte[]> attributes); 5 List<TRowResult> getRows(byte[] tableName, List<byte[]> rows, Dictionary<byte[], byte[]> attributes); 6 List<TRowResult> getRowsWithColumns(byte[] tableName, List<byte[]> rows, List<byte[]> columns, Dictionary<byte[], byte[]> attributes); 7 List<TRowResult> getRowsTs(byte[] tableName, List<byte[]> rows, long timestamp, Dictionary<byte[], byte[]> attributes); 8 List<TRowResult> getRowsWithColumnsTs(byte[] tableName, List<byte[]> rows, List<byte[]> columns, long timestamp, Dictionary<byte[], byte[]> attributes); 9 int scannerOpenWithScan(byte[] tableName, TScan scan, Dictionary<byte[], byte[]> attributes); 10 int scannerOpen(byte[] tableName, byte[] startRow, List<byte[]> columns, Dictionary<byte[], byte[]> attributes); 11 int scannerOpenWithStop(byte[] tableName, byte[] startRow, byte[] stopRow, List<byte[]> columns, Dictionary<byte[], byte[]> attributes); 12 int scannerOpenWithPrefix(byte[] tableName, byte[] startAndPrefix, List<byte[]> columns, Dictionary<byte[], byte[]> attributes); 13 int scannerOpenTs(byte[] tableName, byte[] startRow, List<byte[]> columns, long timestamp, Dictionary<byte[], byte[]> attributes); 14 int scannerOpenWithStopTs(byte[] tableName, byte[] startRow, byte[] stopRow, List<byte[]> columns, long timestamp, Dictionary<byte[], byte[]> attributes); 15 List<TRowResult> scannerGet(int id); 16 List<TRowResult> scannerGetList(int id, int nbRows); 17 void scannerClose(int id);
结合项目的应用介绍几个比较常用的接口(其实只要用过HBaseAPI的童鞋,上面这些接口就不在话下了):
1.getRow(这类查询简单,通过rowkey获取数据,接口的大部分参数类型为字节数组)

结果:

2.scannerOpenWithStop(通过RowKey的范围获取数据)
结果:

3.scannerOpenWithPrefix(通过RowKey的前缀进行匹配查询)
/// <summary>
/// 通过Rowkey前缀Fliter
/// </summary>
/// <param name="tablename"></param>
/// <param name="startrowkey"></param>
/// <param name="endrowkey"></param>
static void GetDataFromHBaseThroughRowKeyPrefix(string tablename, string Prefixrowkey)
{
transport.Open();
int ScannerID = client.scannerOpenWithPrefix(Encoding.UTF8.GetBytes(tablename), Encoding.UTF8.GetBytes(Prefixrowkey), new List<byte[]> { Encoding.UTF8.GetBytes("i:Data") }, null);
/*
* scannerGetList(string ID),源码中其实调用scannerGetList(string ID,int nbRow)方法,nbRow传值为1
*/
List<TRowResult> reslut = client.scannerGetList(ScannerID,100);
foreach (var key in reslut)
{
Console.WriteLine(Encoding.UTF8.GetString(key.Row));
foreach (var k in key.Columns)
{
Console.Write(Encoding.UTF8.GetString(k.Key) + "\t");
Console.WriteLine(Encoding.UTF8.GetString(k.Value.Value));
Console.WriteLine("++++++++++++++++++++++++++++++++++++++");
}
}
}
/调用
static void Main(string[] args)
bsp; {
GetDataFromHBaseThroughRowKeyPrefix("HStudy", "00");
}
4.scannerOpenWithScan(通过过滤器进行查询)
这个接口是所有查询接口中最麻烦的一个吧,因为它用到了过滤器,也就是HBaseAPI中的Filter。这个接口的参数中有一个参数类型为TScan,基本结构如下:
public partial class TScan : TBase
{
private byte[] _startRow;
private byte[] _stopRow;
private long _timestamp;
private List<byte[]> _columns;
private int _caching;
private byte[] _filterString;
}
前面的几个参数不多说,这里说一下_filterString (关于HaseAPI中各种Filter这里就不多说),以常见的SingleColumnValueFilter为例,如果我想定义一个查询PatientName为小红的一个过滤器:
stringfilterString = "SingleColumnValueFilter('s','PatientName',=,'substring:小红')";
byte[]_filterString = Encoding.UTF8.GetBytes(filterString);
如果要定义多个过滤器,过滤器之间用‘AND’连接。
/// <summary>
/// 通过Filter进行数据的Scanner
/// </summary>
/// <param name="tablename"></param>
/// <param name="filterString"></param>
static void GetDataFromHBaseThroughFilter(string tablename, string filterString,List<byte[]> _cols)
{
TScan _scan = new TScan();
//SingleColumnValueFilter('i', 'Data', =, '2')
_scan.FilterString =Encoding.UTF8.GetBytes(filterString);
_scan.Columns = _cols;
transport.Open();
int ScannerID = client.scannerOpenWithScan(Encoding.UTF8.GetBytes(tablename), _scan,null);
List<TRowResult> reslut = client.scannerGetList(ScannerID, 100);
foreach (var key in reslut)
{
Console.WriteLine(Encoding.UTF8.GetString(key.Row));
foreach (var k in key.Columns)
{
Console.Write(Encoding.UTF8.GetString(k.Key) + "\t");
Console.WriteLine(Encoding.UTF8.GetString(k.Value.Value));
Console.WriteLine("++++++++++++++++++++++++++++++++++++++");
}
}
}
static void Main(string[] args)
{
////GetDataFromHBaseThroughRowKeyRange("HImages", "123.456.1", "123.456.9");
List<byte[]> _byte = new List<byte[]>();
_byte.Add(Encoding.UTF8.GetBytes("s:PatientName"));
_byte.Add(Encoding.UTF8.GetBytes("s:StudyInstanceUID"));
_byte.Add(Encoding.UTF8.GetBytes("s:PatientSex"));
////string filterString = "((SingleColumnValueFilter('s','PatientName',=,'substring:Jim')) AND (SingleColumnValueFilter('s','PatientSex',=,'substring:10')))";
string filterString = "SingleColumnValueFilter('s','PatientName',=,'substring:小红')";
GetDataFromHBaseThroughFilter("HStudy", filterString, _byte);
Console.ReadLine();
}
关于通过Thrift查询HBase就说到这。
#最近无论工作还是生活都弄的一团糟,可能是我太浮躁了,整天游戏人生浑浑噩噩的过。算了不说这些扫兴的事情了。
关于C#访问Thrift的文章,这应该是第三篇了,也是最后一篇了吧。Thrfit 最麻烦的应该是查询那部分,第二篇我也详细的说了一下,这一篇文章,说一下Thrift中使用很平凡的API(新建表,删除表,插入数据,更新数据,删除数据),最后发一下,为了方便使用Thrift写的一个Helper类。
值得注意的是,Thrift 插入行和更新行使用的同一函数(mutateRow等一类函数),使用过HBaseAPI的童鞋,这点不足为奇。这几个API都比较简单,下面我就直接贴出Helper类及简单的测试类。
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using ThriftHelper;
using IThrift;
using Thrift.Transport;
using Thrift.Protocol;
namespace Test
{
class Program
{
/*
* 表名: HTest
* 列簇: i
* 子列: Data
*/
static void Main(string[] args)
{
#region Test
Helper.Open();
Printer("CreateTable:");
ColumnDescriptor _cd = new ColumnDescriptor();
_cd.Name = Encoding.UTF8.GetBytes("i");
if (Helper.CreateTable("ITest", new List<ColumnDescriptor> { _cd }))
Printer("CreateTable is Success");
else
Printer("CreateTable Occurred Error");
Printer("++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++");
Printer("MutateRowHBase:");
Mutation _mutation = new Mutation();
_mutation.Column = Encoding.UTF8.GetBytes("i:one");
_mutation.Value = Encoding.UTF8.GetBytes("1");
if (Helper.MutateRowHBase("ITest", "001", new List<Mutation> { _mutation }))
Printer("MutateRowHBase is Success");
else
Printer("MutateRowHBase Occurred Error");
Printer("++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++");
Printer("GetDataFromHBase:");
List<TRowResult> _result = Helper.GetDataFromHBase("ITest", "001");
Printer(_result);
Printer("++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++");
Printer("MutateRowHBase:");
_mutation = new Mutation();
_mutation.Column = Encoding.UTF8.GetBytes("i:one");
_mutation.Value = Encoding.UTF8.GetBytes("-1");
if (Helper.MutateRowHBase("ITest", "001", new List<Mutation> { _mutation }))
Printer("MutateRowHBase is Success");
else
Printer("MutateRowHBase Occurred Error");
Printer("++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++");
Printer("GetDataFromHBase:");
_result = Helper.GetDataFromHBase("ITest", "001");
Printer(_result);
Printer("++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++");
Printer("DeleteRow:");
if (Helper.DeleteAllRow("ITest", "001"))
Printer("DeleteAllRow is Success");
else
Printer("DeleteAllRow Occurred Error");
Helper.Close();
#endregion
Console.ReadKey();
}
static void Printer(List<TRowResult> reslut)
{
if (reslut.Count == 0)
return;
foreach (var key in reslut)
{
Console.WriteLine(Encoding.UTF8.GetString(key.Row));
foreach (var k in key.Columns)
{
Console.Write(Encoding.UTF8.GetString(k.Key) + "\t");
Console.WriteLine(Encoding.UTF8.GetString(k.Value.Value));
Console.WriteLine("++++++++++++++++++++++++++++++++++++++");
}
}
}
static void Printer(string conent)
{
Console.Write(conent);
}
}
}
Helper类:
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using Thrift;
using IThrift;
using Thrift.Transport;
using Thrift.Protocol;
namespace ThriftHelper
{
public static class Helper
{
static TTransport transport = new TSocket("192.168.2.200", 9090);
static TProtocol tProtocol = new TBinaryProtocol(transport);
static Hbase.Client client = new Hbase.Client(tProtocol);
public static void Open()
{
transport.Open();
}
public static void Close()
{
transport.Close();
}
/// <summary>
/// 通过rowkey获取数据
/// </summary>
/// <param name="tablename"></param>
/// <param name="rowkey"></param>
public static List<TRowResult> GetDataFromHBase(string tablename, string rowkey)
{
List<TRowResult> reslut = client.getRow(Encoding.UTF8.GetBytes(tablename), Encoding.UTF8.GetBytes(rowkey), null);
return reslut;
}
/// <summary>
/// 通过Rowkey前缀Fliter
/// </summary>
/// <param name="tablename"></param>
/// <param name="startrowkey"></param>
/// <param name="endrowkey"></param>
public static List<TRowResult> GetDataFromHBaseThroughRowKeyPrefix(string tablename, string Prefixrowkey,List<string> _cols)
{
List<byte[]> _bytes = new List<byte[]>();
foreach (string str in _cols)
_bytes.Add(Encoding.UTF8.GetBytes(str));
int ScannerID = client.scannerOpenWithPrefix(Encoding.UTF8.GetBytes(tablename), Encoding.UTF8.GetBytes(Prefixrowkey),
_bytes, null);
/*
* scannerGetList(string ID),源码中其实调用scannerGetList(string ID,int nbRow)方法,nbRow传值为1
*/
List<TRowResult> reslut = client.scannerGetList(ScannerID, 100);
return reslut;
}
/// <summary>
/// 通过RowKey的范围获取数据
/// </summary>
/// <param name="tablename"></param>
/// <param name="stRowkey"></param>
/// <param name="?"></param>
/// <remarks>结果集包含StartRowKey列值,不包含EndRowKey的列值</remarks>
public static List<TRowResult> GetDataFromHBaseThroughRowKeyRange(string tablename,
string stRowkey, string endRowkey,List<string> _cols)
{
List<byte[]> _bytes = new List<byte[]>();
foreach (string str in _cols)
_bytes.Add(Encoding.UTF8.GetBytes(str));
int ScannerID = client.scannerOpenWithStop(Encoding.UTF8.GetBytes(tablename),
Encoding.UTF8.GetBytes(stRowkey), Encoding.UTF8.GetBytes(endRowkey),
_bytes, null);
List<TRowResult> reslut = client.scannerGetList(ScannerID, 100);
return reslut;
}
/// <summary>
/// 通过Filter进行数据的Scanner
/// </summary>
/// <param name="tablename"></param>
/// <param name="filterString"></param>
public static List<TRowResult> GetDataFromHBaseThroughFilter(string tablename, string filterString, List<byte[]> _cols)
{
TScan _scan = new TScan();
//SingleColumnValueFilter('i', 'Data', =, '2')
_scan.FilterString = Encoding.UTF8.GetBytes(filterString);
_scan.Columns = _cols;
int ScannerID = client.scannerOpenWithScan(Encoding.UTF8.GetBytes(tablename), _scan, null);
List<TRowResult> reslut = client.scannerGetList(ScannerID, 100);
return reslut;
}
public static bool MutateRowHBase(string tablename, string rowkey, List<Mutation> _mutations)
{
try
{
client.mutateRow(Encoding.UTF8.GetBytes(tablename), Encoding.UTF8.GetBytes(rowkey), _mutations, null);
return true;
}
catch (Exception e)
{
return false;
}
}
public static bool MutateRowsHBase(string tablename, List<BatchMutation> _BatchMutation)
{
try
{
client.mutateRows(Encoding.UTF8.GetBytes(tablename), _BatchMutation, null);
return true;
}
catch (Exception e)
{
return false;
}
}
public static bool DeleteRowHBase(string tablename, string rowkey, string column)
{
try
{
client.deleteAll(Encoding.UTF8.GetBytes(tablename), Encoding.UTF8.GetBytes(rowkey),
Encoding.UTF8.GetBytes(column), null);
return true;
}
catch (Exception e)
{
return false;
}
}
public static bool DeleteAllRow(string tablename, string rowkey)
{
try
{
client.deleteAllRow(Encoding.UTF8.GetBytes(tablename), Encoding.UTF8.GetBytes(rowkey), null);
return true;
}
catch (Exception e)
{
return false;
}
}
public static bool DeleteTable(string tablename)
{
try
{
client.deleteTable(Encoding.UTF8.GetBytes(tablename));
return true;
}
catch (Exception e)
{
return false;
}
}
public static bool CreateTable(string tablename, List<ColumnDescriptor> _cols)
{
try
{
client.createTable(Encoding.UTF8.GetBytes(tablename), _cols);
return true;
}
catch (Exception e)
{
return false;
}
}
}
}
好了,关于Thrift的基本操作就写到这,当然Thrift也支持Hbase中比较高级的操作,在以后的博客会不断更新。谢谢大家,个人水平有限,不足之处请谅解。

浙公网安备 33010602011771号