/**
* Copyright 2008 The Apache Software Foundation
*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.hadoop.hbase.thrift;
import java.io.UnsupportedEncodingException;
import java.nio.ByteBuffer;
import java.nio.charset.CharacterCodingException;
import java.nio.charset.Charset;
import java.nio.charset.CharsetDecoder;
import java.text.NumberFormat;
import java.util.ArrayList;
import java.util.List;
import java.util.Map;
import java.util.TreeMap;
import java.util.SortedMap;
import org.apache.hadoop.hbase.thrift.generated.AlreadyExists;
import org.apache.hadoop.hbase.thrift.generated.ColumnDescriptor;
import org.apache.hadoop.hbase.thrift.generated.Hbase;
import org.apache.hadoop.hbase.thrift.generated.IOError;
import org.apache.hadoop.hbase.thrift.generated.IllegalArgument;
import org.apache.hadoop.hbase.thrift.generated.Mutation;
import org.apache.hadoop.hbase.thrift.generated.TCell;
import org.apache.hadoop.hbase.thrift.generated.TRowResult;
import org.apache.thrift.TException;
import org.apache.thrift.protocol.TBinaryProtocol;
import org.apache.thrift.protocol.TProtocol;
import org.apache.thrift.transport.TSocket;
import org.apache.thrift.transport.TTransport;
/*
* Instructions:
* 1. Run Thrift to generate the java module HBase
* thrift --gen java ../../../src/main/resources/org/apache/hadoop/hbase/thrift/Hbase.thrift
* 2. Acquire a jar of compiled Thrift java classes. As of this writing, HBase ships
* with this jar (libthrift-[VERSION].jar). If this jar is not present, or it is
* out-of-date with your current version of thrift, you can compile the jar
* yourself by executing {ant} in {$THRIFT_HOME}/lib/java.
* 3. Compile and execute this file with both the libthrift jar and the gen-java/
* directory in the classpath. This can be done on the command-line with the
* following lines: (from the directory containing this file and gen-java/)
*
* javac -cp /path/to/libthrift/jar.jar:gen-java/ DemoClient.java
* mv DemoClient.class gen-java/org/apache/hadoop/hbase/thrift/
* java -cp /path/to/libthrift/jar.jar:gen-java/ org.apache.hadoop.hbase.thrift.DemoClient
*
*/
public class DemoClient {
static protected int port;
static protected String host;
CharsetDecoder decoder = null;
public static void main(String[] args)
throws IOError, TException, UnsupportedEncodingException, IllegalArgument, AlreadyExists {
if (args.length != 2) {
System.out.println("Invalid arguments!");
System.out.println("Usage: DemoClient host port");
System.exit(-1);
}
port = Integer.parseInt(args[1]);
host = args[0];
DemoClient client = new DemoClient();
client.run();
}
DemoClient() {
decoder = Charset.forName("UTF-8").newDecoder();
}
// Helper to translate byte[]'s to UTF8 strings
private String utf8(byte[] buf) {
try {
return decoder.decode(ByteBuffer.wrap(buf)).toString();
} catch (CharacterCodingException e) {
return "[INVALID UTF-8]";
}
}
// Helper to translate strings to UTF8 bytes
private byte[] bytes(String s) {
try {
return s.getBytes("UTF-8");
} catch (UnsupportedEncodingException e) {
e.printStackTrace();
return null;
}
}
private void run() throws IOError, TException, IllegalArgument,
AlreadyExists {
TTransport transport = new TSocket(host, port);
TProtocol protocol = new TBinaryProtocol(transport, true, true);
Hbase.Client client = new Hbase.Client(protocol);
transport.open();
byte[] t = bytes("demo_table");
//
// Scan all tables, look for the demo table and delete it.
//
System.out.println("scanning tables...");
for (ByteBuffer name : client.getTableNames()) {
System.out.println(" found: " + utf8(name.array()));
if (utf8(name.array()).equals(utf8(t))) {
if (client.isTableEnabled(name)) {
System.out.println(" disabling table: " + utf8(name.array()));
client.disableTable(name);
}
System.out.println(" deleting table: " + utf8(name.array()));
client.deleteTable(name);
}
}
//
// Create the demo table with two column families, entry: and unused:
//
ArrayList<ColumnDescriptor> columns = new ArrayList<ColumnDescriptor>();
ColumnDescriptor col = null;
col = new ColumnDescriptor();
col.name = ByteBuffer.wrap(bytes("entry:"));
col.maxVersions = 10;
columns.add(col);
col = new ColumnDescriptor();
col.name = ByteBuffer.wrap(bytes("unused:"));
columns.add(col);
System.out.println("creating table: " + utf8(t));
try {
client.createTable(ByteBuffer.wrap(t), columns);
} catch (AlreadyExists ae) {
System.out.println("WARN: " + ae.message);
}
System.out.println("column families in " + utf8(t) + ": ");
Map<ByteBuffer, ColumnDescriptor> columnMap = client.getColumnDescriptors(ByteBuffer.wrap(t));
for (ColumnDescriptor col2 : columnMap.values()) {
System.out.println(" column: " + utf8(col2.name.array()) + ", maxVer: " + Integer.toString(col2.maxVersions));
}
//
// Test UTF-8 handling
//
byte[] invalid = {(byte) 'f', (byte) 'o', (byte) 'o', (byte) '-', (byte) 0xfc, (byte) 0xa1, (byte) 0xa1, (byte) 0xa1, (byte) 0xa1};
byte[] valid = {(byte) 'f', (byte) 'o', (byte) 'o', (byte) '-', (byte) 0xE7, (byte) 0x94, (byte) 0x9F, (byte) 0xE3, (byte) 0x83, (byte) 0x93, (byte) 0xE3, (byte) 0x83, (byte) 0xBC, (byte) 0xE3, (byte) 0x83, (byte) 0xAB};
ArrayList<Mutation> mutations;
// non-utf8 is fine for data
mutations = new ArrayList<Mutation>();
mutations.add(new Mutation(false, ByteBuffer.wrap(bytes("entry:foo")), ByteBuffer.wrap(invalid)));
client.mutateRow(ByteBuffer.wrap(t), ByteBuffer.wrap(bytes("foo")), mutations);
// try empty strings
mutations = new ArrayList<Mutation>();
mutations.add(new Mutation(false, ByteBuffer.wrap(bytes("entry:")), ByteBuffer.wrap(bytes(""))));
client.mutateRow(ByteBuffer.wrap(t), ByteBuffer.wrap(bytes("")), mutations);
// this row name is valid utf8
mutations = new ArrayList<Mutation>();
mutations.add(new Mutation(false, ByteBuffer.wrap(bytes("entry:foo")), ByteBuffer.wrap(valid)));
client.mutateRow(ByteBuffer.wrap(t), ByteBuffer.wrap(valid), mutations);
// non-utf8 is now allowed in row names because HBase stores values as binary
ByteBuffer bf = ByteBuffer.wrap(invalid);
mutations = new ArrayList<Mutation>();
mutations.add(new Mutation(false, ByteBuffer.wrap(bytes("entry:foo")), ByteBuffer.wrap(invalid)));
client.mutateRow(ByteBuffer.wrap(t), ByteBuffer.wrap(invalid), mutations);
// Run a scanner on the rows we just created
ArrayList<ByteBuffer> columnNames = new ArrayList<ByteBuffer>();
columnNames.add(ByteBuffer.wrap(bytes("entry:")));
System.out.println("Starting scanner...");
int scanner = client.scannerOpen(ByteBuffer.wrap(t), ByteBuffer.wrap(bytes("")), columnNames);
while (true) {
List<TRowResult> entry = client.scannerGet(scanner);
if (entry.isEmpty()) {
break;
}
printRow(entry);
}
//
// Run some operations on a bunch of rows
//
for (int i = 100; i >= 0; --i) {
// format row keys as "00000" to "00100"
NumberFormat nf = NumberFormat.getInstance();
nf.setMinimumIntegerDigits(5);
nf.setGroupingUsed(false);
byte[] row = bytes(nf.format(i));
mutations = new ArrayList<Mutation>();
mutations.add(new Mutation(false, ByteBuffer.wrap(bytes("unused:")), ByteBuffer.wrap(bytes("DELETE_ME"))));
client.mutateRow(ByteBuffer.wrap(t), ByteBuffer.wrap(row), mutations);
printRow(client.getRow(ByteBuffer.wrap(t), ByteBuffer.wrap(row)));
client.deleteAllRow(ByteBuffer.wrap(t), ByteBuffer.wrap(row));
mutations = new ArrayList<Mutation>();
mutations.add(new Mutation(false, ByteBuffer.wrap(bytes("entry:num")), ByteBuffer.wrap(bytes("0"))));
mutations.add(new Mutation(false, ByteBuffer.wrap(bytes("entry:foo")), ByteBuffer.wrap(bytes("FOO"))));
client.mutateRow(ByteBuffer.wrap(t), ByteBuffer.wrap(row), mutations);
printRow(client.getRow(ByteBuffer.wrap(t), ByteBuffer.wrap(row)));
Mutation m = null;
mutations = new ArrayList<Mutation>();
m = new Mutation();
m.column = ByteBuffer.wrap(bytes("entry:foo"));
m.isDelete = true;
mutations.add(m);
m = new Mutation();
m.column = ByteBuffer.wrap(bytes("entry:num"));
m.value = ByteBuffer.wrap(bytes("-1"));
mutations.add(m);
client.mutateRow(ByteBuffer.wrap(t), ByteBuffer.wrap(row), mutations);
printRow(client.getRow(ByteBuffer.wrap(t), ByteBuffer.wrap(row)));
mutations = new ArrayList<Mutation>();
mutations.add(new Mutation(false, ByteBuffer.wrap(bytes("entry:num")), ByteBuffer.wrap(bytes(Integer.toString(i)))));
mutations.add(new Mutation(false, ByteBuffer.wrap(bytes("entry:sqr")), ByteBuffer.wrap(bytes(Integer.toString(i * i)))));
client.mutateRow(ByteBuffer.wrap(t), ByteBuffer.wrap(row), mutations);
printRow(client.getRow(ByteBuffer.wrap(t), ByteBuffer.wrap(row)));
// sleep to force later timestamp
try {
Thread.sleep(50);
} catch (InterruptedException e) {
// no-op
}
mutations.clear();
m = new Mutation();
m.column = ByteBuffer.wrap(bytes("entry:num"));
m.value= ByteBuffer.wrap(bytes("-999"));
mutations.add(m);
m = new Mutation();
m.column = ByteBuffer.wrap(bytes("entry:sqr"));
m.isDelete = true;
client.mutateRowTs(ByteBuffer.wrap(t), ByteBuffer.wrap(row), mutations, 1); // shouldn't override latest
printRow(client.getRow(ByteBuffer.wrap(t), ByteBuffer.wrap(row)));
List<TCell> versions = client.getVer(ByteBuffer.wrap(t), ByteBuffer.wrap(row), ByteBuffer.wrap(bytes("entry:num")), 10);
printVersions(ByteBuffer.wrap(row), versions);
if (versions.isEmpty()) {
System.out.println("FATAL: wrong # of versions");
System.exit(-1);
}
List<TCell> result = client.get(ByteBuffer.wrap(t), ByteBuffer.wrap(row), ByteBuffer.wrap(bytes("entry:foo")));
if (result.isEmpty() == false) {
System.out.println("FATAL: shouldn't get here");
System.exit(-1);
}
System.out.println("");
}
// scan all rows/columnNames
columnNames.clear();
for (ColumnDescriptor col2 : client.getColumnDescriptors(ByteBuffer.wrap(t)).values()) {
System.out.println("column with name: " + new String(col2.name.array()));
System.out.println(col2.toString());
columnNames.add(col2.name);
}
System.out.println("Starting scanner...");
scanner = client.scannerOpenWithStop(ByteBuffer.wrap(t), ByteBuffer.wrap(bytes("00020")), ByteBuffer.wrap(bytes("00040")),
columnNames);
while (true) {
List<TRowResult> entry = client.scannerGet(scanner);
if (entry.isEmpty()) {
System.out.println("Scanner finished");
break;
}
printRow(entry);
}
transport.close();
}
private final void printVersions(ByteBuffer row, List<TCell> versions) {
StringBuilder rowStr = new StringBuilder();
for (TCell cell : versions) {
rowStr.append(utf8(cell.value.array()));
rowStr.append("; ");
}
System.out.println("row: " + utf8(row.array()) + ", values: " + rowStr);
}
private final void printRow(TRowResult rowResult) {
// copy values into a TreeMap to get them in sorted order
TreeMap<String, TCell> sorted = new TreeMap<String, TCell>();
for (Map.Entry<ByteBuffer, TCell> column : rowResult.columns.entrySet()) {
sorted.put(utf8(column.getKey().array()), column.getValue());
}
StringBuilder rowStr = new StringBuilder();
for (SortedMap.Entry<String, TCell> entry : sorted.entrySet()) {
rowStr.append(entry.getKey());
rowStr.append(" => ");
rowStr.append(utf8(entry.getValue().value.array()));
rowStr.append("; ");
}
System.out.println("row: " + utf8(rowResult.row.array()) + ", cols: " + rowStr);
}
private void printRow(List<TRowResult> rows) {
for (TRowResult rowResult : rows) {
printRow(rowResult);
}
}
}