HBase Filter – Comparision Filters | 程序员的自我修养

HBase的Get和Scan实例可以调用setFilter()来设置过滤器，HBase的过滤器种类繁多，以满足不同的过滤需求。Filter作用于各个RegionServer，通过使用过滤器可以高效的获取数据。HBase Filter主要分为三大类，用户也可以通过继承FilterBase或实现Filter接口来自定义Filter。若需要多个Filter结合使用，可以通过FilterList来满足。

以下主要对各个Filter进行简单的了解。

数据准备

rowkey	cf1:col1	cf1:col2	cf2:col1	cf2:col2
101	10086	qwe1	1352288xxxx	wer1
102	10000	sdf2	1820160xxxx	ert2
103	10001	cxv3	1531308xxxx	mnb3
104	12306	jhg4	1387223xxxx	kji4
105	12580	nju5	1580101xxxx	nbv5

代码准备

    private HTable table;
    @Before
    public void setUp() throws IOException {
        Configuration conf = HBaseConfiguration.create();
        conf.set("hbase.zookeeper.quorum", "192.168.0.126");
        conf.set("hbase.zookeeper.property.clientPort", "2181");
        conf.set("mapred.task.timeout", "0");
        table = new HTable(conf, "pigstore");
    }

    @After
    public void tearDown() throws IOException {
        table.close();
    }

    private void setFilterAndPrint(Filter filter) throws IOException {
        Scan scan = new Scan();
        scan.setFilter(filter);
        print(scan);
    }

    private void print(Scan scan) throws IOException {
        ResultScanner resultScanner = table.getScanner(scan);
        for(Result res : resultScanner) {
            for(KeyValue kv : res.raw()){
                System.out.println("Key: " + kv + ", Value: " + Bytes.toString(kv.getValue()));
            }
        }
    }

private HTable table;

@Before

public void setUp() throws IOException {

Configuration conf = HBaseConfiguration.create();

conf.set("hbase.zookeeper.quorum", "192.168.0.126");

conf.set("hbase.zookeeper.property.clientPort", "2181");

conf.set("mapred.task.timeout", "0");

table = new HTable(conf, "pigstore");

}

@After

public void tearDown() throws IOException {

table.close();

}

private void setFilterAndPrint(Filter filter) throws IOException {

Scan scan = new Scan();

scan.setFilter(filter);

print(scan);

}

private void print(Scan scan) throws IOException {

ResultScanner resultScanner = table.getScanner(scan);

for(Result res : resultScanner) {

for(KeyValue kv : res.raw()){

System.out.println("Key: " + kv + ", Value: " + Bytes.toString(kv.getValue()));

}

Comparision Filters(比较过滤器)

RowFilter

使用示例

@Test
//过滤rowkey="101"
public void testRowFilter() throws IOException {
	Filter filter = new RowFilter(CompareFilter.CompareOp.EQUAL,
		new BinaryComparator(Bytes.toBytes("101")));
	setFilterAndPrint(filter);
}

@Test

//过滤rowkey="101"

public void testRowFilter() throws IOException {

Filter filter = new RowFilter(CompareFilter.CompareOp.EQUAL,

new BinaryComparator(Bytes.toBytes("101")));

setFilterAndPrint(filter);

}

结果

Key: 101/cf1:col1/1404704499698/Put/vlen=5/ts=0, Value: 10086
Key: 101/cf1:col2/1404704499698/Put/vlen=4/ts=0, Value: qwe1
Key: 101/cf2:col1/1404704499698/Put/vlen=11/ts=0, Value: 1352288xxxx
Key: 101/cf2:col2/1404704499698/Put/vlen=4/ts=0, Value: wer1

其它

RowFilter，行过滤器，实际上称之为行键过滤器更合适，用于行键的各种过滤上。构造函数为RowFilter(CompareFilter.CompareOp rowCompareOp, WritableByteArrayComparable rowComparator)。结合两个构造参数可以实现基于rowkey的不同过滤方式。

假设根据rowkey，数据分为3部分A、B、C，其中B为通过rowComparator选出的数据，接着通过rowCompareOp来确定最终的数据：若为EQUAL则选择B，若为LESS_OR_EQUAL则选择A+B，如此类推。

FamilyFilter

使用示例

@Test
//过滤列族小于cf2
public void testFamilyFilter() throws IOException {
	Filter filter = new FamilyFilter(CompareFilter.CompareOp.LESS, 
		new BinaryComparator(Bytes.toBytes("cf2")));
	setFilterAndPrint(filter);
}

@Test

//过滤列族小于cf2

public void testFamilyFilter() throws IOException {

Filter filter = new FamilyFilter(CompareFilter.CompareOp.LESS,

new BinaryComparator(Bytes.toBytes("cf2")));

setFilterAndPrint(filter);

}

结果

Key: 101/cf1:col1/1404704499698/Put/vlen=5/ts=0, Value: 10086
Key: 101/cf1:col2/1404704499698/Put/vlen=4/ts=0, Value: qwe1
Key: 102/cf1:col1/1404704499699/Put/vlen=5/ts=0, Value: 10000
Key: 102/cf1:col2/1404704499699/Put/vlen=4/ts=0, Value: sdf2
Key: 103/cf1:col1/1404704499699/Put/vlen=5/ts=0, Value: 10001
Key: 103/cf1:col2/1404704499699/Put/vlen=4/ts=0, Value: cxv3
Key: 104/cf1:col1/1404704499699/Put/vlen=5/ts=0, Value: 12306
Key: 104/cf1:col2/1404704499699/Put/vlen=4/ts=0, Value: jhg4
Key: 105/cf1:col1/1404704499699/Put/vlen=5/ts=0, Value: 12580
Key: 105/cf1:col2/1404704499699/Put/vlen=4/ts=0, Value: nju5

其它

FamilyFilter，列族过滤器，用于列族的过滤。其余特性与RowFilter类似，只是范围由rowkey变成了列族名。

QualifierFilter

使用示例

@Test
//过滤列名大于col1
public void testQualifierFilter() throws IOException {
	Filter filter = new QualifierFilter(CompareFilter.CompareOp.GREATER, 
		new BinaryComparator(Bytes.toBytes("col1")));
	setFilterAndPrint(filter);
}

@Test

//过滤列名大于col1

public void testQualifierFilter() throws IOException {

Filter filter = new QualifierFilter(CompareFilter.CompareOp.GREATER,

new BinaryComparator(Bytes.toBytes("col1")));

setFilterAndPrint(filter);

}

结果

Key: 101/cf1:col2/1404704499698/Put/vlen=4/ts=0, Value: qwe1
Key: 101/cf2:col2/1404704499698/Put/vlen=4/ts=0, Value: wer1
Key: 102/cf1:col2/1404704499699/Put/vlen=4/ts=0, Value: sdf2
Key: 102/cf2:col2/1404704499699/Put/vlen=4/ts=0, Value: ert2
Key: 103/cf1:col2/1404704499699/Put/vlen=4/ts=0, Value: cxv3
Key: 103/cf2:col2/1404704499699/Put/vlen=4/ts=0, Value: mnb3
Key: 104/cf1:col2/1404704499699/Put/vlen=4/ts=0, Value: jhg4
Key: 104/cf2:col2/1404704499699/Put/vlen=4/ts=0, Value: kji4
Key: 105/cf1:col2/1404704499699/Put/vlen=4/ts=0, Value: nju5
Key: 105/cf2:col2/1404704499699/Put/vlen=4/ts=0, Value: nbv5

其它

QualifierFilter，列名过滤器，用于列名的过滤。其余特性与RowFilter类似，只是范围由rowkey变成了列名。

ValueFilter

使用示例

@Test
//过滤Value的前缀为100的数据
public void testValueFilter() throws IOException {
	Filter filter = new ValueFilter(CompareFilter.CompareOp.EQUAL,
		new SubstringComparator("100"));
	setFilterAndPrint(filter);
}

@Test

//过滤Value的前缀为100的数据

public void testValueFilter() throws IOException {

Filter filter = new ValueFilter(CompareFilter.CompareOp.EQUAL,

new SubstringComparator("100"));

setFilterAndPrint(filter);

}

结果

Key: 101/cf1:col1/1404704499698/Put/vlen=5/ts=0, Value: 10086
Key: 102/cf1:col1/1404704499699/Put/vlen=5/ts=0, Value: 10000
Key: 103/cf1:col1/1404704499699/Put/vlen=5/ts=0, Value: 10001

其它

ValueFilter，值过滤器，用于值的过滤。其余特性与RowFilter类似，只是范围由rowkey变成了值。

DependentColumnFilter

使用示例

@Test
//过滤前缀等于1000的数据
public void testDependentColumnFilter() throws IOException {
	//第三个参数为false时，不输出cf1:col1这一列，即不输出参考列
	Filter filter = new DependentColumnFilter(Bytes.toBytes("cf1"), 
		Bytes.toBytes("col1"),false,CompareFilter.CompareOp.EQUAL,
		new BinaryPrefixComparator(Bytes.toBytes("1000")));
	setFilterAndPrint(filter);
}

@Test

//过滤前缀等于1000的数据

public void testDependentColumnFilter() throws IOException {

//第三个参数为false时，不输出cf1:col1这一列，即不输出参考列

Filter filter = new DependentColumnFilter(Bytes.toBytes("cf1"),

Bytes.toBytes("col1"),false,CompareFilter.CompareOp.EQUAL,

new BinaryPrefixComparator(Bytes.toBytes("1000")));

setFilterAndPrint(filter);

}

结果

Key: 102/cf1:col1/1404704499699/Put/vlen=5/ts=0, Value: 10000
Key: 102/cf1:col2/1404704499699/Put/vlen=4/ts=0, Value: sdf2
Key: 102/cf2:col1/1404704499699/Put/vlen=11/ts=0, Value: 1820160xxxx
Key: 102/cf2:col2/1404704499699/Put/vlen=4/ts=0, Value: ert2
Key: 103/cf1:col1/1404704499699/Put/vlen=5/ts=0, Value: 10001
Key: 103/cf1:col2/1404704499699/Put/vlen=4/ts=0, Value: cxv3
Key: 103/cf2:col1/1404704499699/Put/vlen=11/ts=0, Value: 1531308xxxx
Key: 103/cf2:col2/1404704499699/Put/vlen=4/ts=0, Value: mnb3

其它

初步理解：与ValueFilter类似，只不过ValueFilter只输出满足条件的列，DependentColumnFilter则将满足条件的列所在的行全部输出。

（转载本站文章请注明作者和出处程序员的自我修养 – SelfUp.cn ，请勿用于任何商业用途）

Anonymous:
Anonymous:
Anonymous: 运行报错：java.lang.Abstr actMethodError
Anonymous:
Anonymous:
Anonymous:
Anonymous:
zzg: 请问zz你问题核实了没� ��我也遇到了这个奇葩� ��问题，我们也是基�...
Anonymous:
kuyuzasur: what do i tell doctor to get viagra. edhelprie.mdhelpserv.com – how long we can take viagra,
Anonymous: 这个问题可以忽略。不影响
Anonymous:
Anonymous: 牛
Anonymous: 楼主你好，我偶尔也会遇到Reconnect due to socket error: java.nio.channels.ClosedCha...
Anonymous: sdfs
Anonymous:
Anonymous: java.io.NotSerializableExcepti on: DStream checkpointing has been enabled but the DStreams with their...
wick: HI,请问一下，U,S,V得到� ��，怎么得到近似矩阵� ��(用spark java)，谢谢。
Michael Whitaker: Thank you for this blog, it was very helpful in troubleshooting my own issues. It seems that no...
Anonymous: