使用 FileSystem JAVA API 对 HDFS 进行读、写、删除等操作

Hadoop文件系统 
基本的文件系统命令操作, 通过hadoop fs -help可以获取所有的命令的详细帮助文件。 

Java抽象类org.apache.hadoop.fs.FileSystem定义了hadoop的一个文件系统接口。该类是一个抽象类，通过以下两种静态工厂方法可以过去FileSystem实例： 
public static FileSystem.get(Configuration conf) throws IOException 
public static FileSystem.get(URI uri, Configuration conf) throws IOException 

具体方法实现： 
1、public boolean mkdirs(Path f) throws IOException 
一次性新建所有目录（包括父目录）， f是完整的目录路径。 

2、public FSOutputStream create(Path f) throws IOException 
创建指定path对象的一个文件，返回一个用于写入数据的输出流 
create()有多个重载版本，允许我们指定是否强制覆盖已有的文件、文件备份数量、写入文件缓冲区大小、文件块大小以及文件权限。 

3、public boolean copyFromLocal(Path src, Path dst) throws IOException 
将本地文件拷贝到文件系统 

4、public boolean exists(Path f) throws IOException 
检查文件或目录是否存在 

5、public boolean delete(Path f, Boolean recursive) 
永久性删除指定的文件或目录，如果f是一个空目录或者文件，那么recursive的值就会被忽略。只有recursive＝true时，一个非空目录及其内容才会被删除。 

6、FileStatus类封装了文件系统中文件和目录的元数据，包括文件长度、块大小、备份、修改时间、所有者以及权限信息。 

通过"FileStatus.getPath（）"可查看指定HDFS中某个目录下所有文件。 

01	package hdfsTest;

02

03	import java.io.IOException;

04

05	import org.apache.hadoop.conf.Configuration;

06	import org.apache.hadoop.fs.FSDataOutputStream;

07	import org.apache.hadoop.fs.FileStatus;

08	import org.apache.hadoop.fs.FileSystem;

09	import org.apache.hadoop.fs.Path;

10

11	public class OperatingFiles {

12	//initialization

13	static Configuration conf = new Configuration();

14	static FileSystem hdfs;

15

static {

16	String path = "/usr/java/hadoop-1.0.3/conf/";

17	conf.addResource(new Path(path + "core-site.xml"));

18	conf.addResource(new Path(path + "hdfs-site.xml"));

19	conf.addResource(new Path(path + "mapred-site.xml"));

20	path = "/usr/java/hbase-0.90.3/conf/";

21	conf.addResource(new Path(path + "hbase-site.xml"));

22

try {

23	hdfs = FileSystem.get(conf);

24	} catch (IOException e) {

25	e.printStackTrace();

26

}

27

}

28

29	//create a direction

30	public void createDir(String dir) throws IOException {

31	Path path = new Path(dir);

32	hdfs.mkdirs(path);

33	System.out.println("new dir \t" + conf.get("fs.default.name") + dir);

34

}

35

36	//copy from local file to HDFS file

37	public void copyFile(String localSrc, String hdfsDst) throws IOException{

38	Path src = new Path(localSrc);

39	Path dst = new Path(hdfsDst);

40	hdfs.copyFromLocalFile(src, dst);

41

42	//list all the files in the current direction

43	FileStatus files[] = hdfs.listStatus(dst);

44	System.out.println("Upload to \t" + conf.get("fs.default.name") + hdfsDst);

45	for (FileStatus file : files) {

46	System.out.println(file.getPath());

47

}

48

}

49

50	//create a new file

51	public void createFile(String fileName, String fileContent) throws IOException {

52	Path dst = new Path(fileName);

53	byte[] bytes = fileContent.getBytes();

54	FSDataOutputStream output = hdfs.create(dst);

55	output.write(bytes);

56	System.out.println("new file \t" + conf.get("fs.default.name") + fileName);

57

}

58

59	//list all files

60	public void listFiles(String dirName) throws IOException {

61	Path f = new Path(dirName);

62	FileStatus[] status = hdfs.listStatus(f);

63	System.out.println(dirName + " has all files:");

64	for (int i = 0; i< status.length; i++) {

65	System.out.println(status[i].getPath().toString());

66

}

67

}

68

69	//judge a file existed? and delete it!

70	public void deleteFile(String fileName) throws IOException {

71	Path f = new Path(fileName);

72	boolean isExists = hdfs.exists(f);

73	if (isExists) { //if exists, delete

74	boolean isDel = hdfs.delete(f,true);

75	System.out.println(fileName + "  delete? \t" + isDel);

76

} else {

77	System.out.println(fileName + "  exist? \t" + isExists);

78

}

79

}

80

81	public static void main(String[] args) throws IOException {

82	OperatingFiles ofs = new OperatingFiles();

83	System.out.println("\n=======create dir=======");

84	String dir = "/test";

85	ofs.createDir(dir);

86	System.out.println("\n=======copy file=======");

87	String src = "/home/ictclas/Configure.xml";

88	ofs.copyFile(src, dir);

89	System.out.println("\n=======create a file=======");

90	String fileContent = "Hello, world! Just a test.";

91	ofs.createFile(dir+"/word.txt", fileContent);

92

}

93

}

Using HDFS in java (0.20.0)

Below is a code sample of how to read from and write to HDFS in java. 

1. Creating a configuration object:   To be able to read from or write to HDFS, you need to create a Configuration object and pass configuration parameter to it using hadoop configuration files.  
  
    // Conf object will read the HDFS configuration parameters from these   XML
    // files. You may specify the parameters for your own if you want. 

    Configuration conf = new Configuration(); 
    conf.addResource(new Path("/opt/hadoop-0.20.0/conf/core-site.xml")); 
    conf.addResource(new Path("/opt/hadoop-0.20.0/conf/hdfs-site.xml")); 

    If you do not assign the configurations to conf object (using hadoop xml file) your HDFS operation will be performed on the local file system and not on the HDFS. 

2. Adding file to HDFS: Create a FileSystem object and use a file stream to add a file. 

    FileSystem fileSystem = FileSystem.get(conf);
    
    // Check if the file already exists
    Path path = new Path("/path/to/file.ext");
    if (fileSystem.exists(path)) {
        System.out.println("File " + dest + " already exists");
        return;
    }

    // Create a new file and write data to it.
    FSDataOutputStream out = fileSystem.create(path);
    InputStream in = new BufferedInputStream(new FileInputStream(
        new File(source)));

    byte[] b = new byte[1024];
    int numBytes = 0;
    while ((numBytes = in.read(b)) > 0) {
        out.write(b, 0, numBytes);
    }

    // Close all the file descripters
    in.close();
    out.close();
    fileSystem.close();

3. Reading file from HDFS: Create a file stream object to a file in HDFS and read it. 

    FileSystem fileSystem = FileSystem.get(conf);

    Path path = new Path("/path/to/file.ext"); 
    if (!fileSystem.exists(path)) { 
        System.out.println("File does not exists"); 
        return; 
    }

    FSDataInputStream in = fileSystem.open(path); 

    String filename = file.substring(file.lastIndexOf('/') + 1,
        file.length()); 

    OutputStream out = new BufferedOutputStream(new FileOutputStream(
        new File(filename))); 

    byte[] b = new byte[1024]; 
    int numBytes = 0; 
    while ((numBytes = in.read(b)) > 0) { 
        out.write(b, 0, numBytes); 
    } 

    in.close(); 
    out.close(); 
    fileSystem.close(); 

3. Deleting file from HDFS: Create a file stream object to a file in HDFS and delete it. 

    FileSystem fileSystem = FileSystem.get(conf); 

    Path path = new Path("/path/to/file.ext"); 
    if (!fileSystem.exists(path)) { 
        System.out.println("File does not exists"); 
        return; 
    }

    // Delete file
    fileSystem.delete(new Path(file), true); 

    fileSystem.close(); 

3. Create dir in HDFS: Create a file stream object to a file in HDFS and read it. 

    FileSystem fileSystem = FileSystem.get(conf); 

    Path path = new Path(dir); 
    if (fileSystem.exists(path)) { 
        System.out.println("Dir " + dir + " already not exists"); 
        return; 
    }

    // Create directories
    fileSystem.mkdirs(path); 

    fileSystem.close(); 

Code:

001	import java.io.BufferedInputStream;

002	import java.io.BufferedOutputStream;

003	import java.io.File;

004	import java.io.FileInputStream;

005	import java.io.FileOutputStream;

006	import java.io.IOException;

007	import java.io.InputStream;

008	import java.io.OutputStream;

009

010	import org.apache.hadoop.conf.Configuration;

011	import org.apache.hadoop.fs.FSDataInputStream;

012	import org.apache.hadoop.fs.FSDataOutputStream;

013	import org.apache.hadoop.fs.FileSystem;

014	import org.apache.hadoop.fs.Path;

015

016	public class HDFSClient {

017	public HDFSClient() {

018

019

}

020

021	public void addFile(String source, String dest) throws IOException {

022	Configuration conf = new Configuration();

023

024	// Conf object will read the HDFS configuration parameters from these

025	// XML files.

026	conf.addResource(new Path("/opt/hadoop-0.20.0/conf/core-site.xml"));

027	conf.addResource(new Path("/opt/hadoop-0.20.0/conf/hdfs-site.xml"));

028

029	FileSystem fileSystem = FileSystem.get(conf);

030

031	// Get the filename out of the file path

032	String filename = source.substring(source.lastIndexOf('/') + 1,

033	source.length());

034

035	// Create the destination path including the filename.

036	if (dest.charAt(dest.length() - 1) != '/') {

037	dest = dest + "/" + filename;

038

} else {

039	dest = dest + filename;

040

}

041

042	// System.out.println("Adding file to " + destination);

043

044	// Check if the file already exists

045	Path path = new Path(dest);

046	if (fileSystem.exists(path)) {

047	System.out.println("File " + dest + " already exists");

048

return;

049

}

050

051	// Create a new file and write data to it.

052	FSDataOutputStream out = fileSystem.create(path);

053	InputStream in = new BufferedInputStream(new FileInputStream(

054	new File(source)));

055

056	byte[] b = new byte[1024];

057	int numBytes = 0;

058	while ((numBytes = in.read(b)) > 0) {

059	out.write(b, 0, numBytes);

060

}

061

062	// Close all the file descripters

063	in.close();

064	out.close();

065	fileSystem.close();

066

}

067

068	public void readFile(String file) throws IOException {

069	Configuration conf = new Configuration();

070	conf.addResource(new Path("/opt/hadoop-0.20.0/conf/core-site.xml"));

071

072	FileSystem fileSystem = FileSystem.get(conf);

073

074	Path path = new Path(file);

075	if (!fileSystem.exists(path)) {

076	System.out.println("File " + file + " does not exists");

077

return;

078

}

079

080	FSDataInputStream in = fileSystem.open(path);

081

082	String filename = file.substring(file.lastIndexOf('/') + 1,

083	file.length());

084

085	OutputStream out = new BufferedOutputStream(new FileOutputStream(

086	new File(filename)));

087

088	byte[] b = new byte[1024];

089	int numBytes = 0;

090	while ((numBytes = in.read(b)) > 0) {

091	out.write(b, 0, numBytes);

092

}

093

094	in.close();

095	out.close();

096	fileSystem.close();

097

}

098

099	public void deleteFile(String file) throws IOException {

100	Configuration conf = new Configuration();

101	conf.addResource(new Path("/opt/hadoop-0.20.0/conf/core-site.xml"));

102

103	FileSystem fileSystem = FileSystem.get(conf);

104

105	Path path = new Path(file);

106	if (!fileSystem.exists(path)) {

107	System.out.println("File " + file + " does not exists");

108

return;

109

}

110

111	fileSystem.delete(new Path(file), true);

112

113	fileSystem.close();

114

}

115

116	public void mkdir(String dir) throws IOException {

117	Configuration conf = new Configuration();

118	conf.addResource(new Path("/opt/hadoop-0.20.0/conf/core-site.xml"));

119

120	FileSystem fileSystem = FileSystem.get(conf);

121

122	Path path = new Path(dir);

123	if (fileSystem.exists(path)) {

124	System.out.println("Dir " + dir + " already not exists");

125

return;

126

}

127

128	fileSystem.mkdirs(path);

129

130	fileSystem.close();

131

}

132

133	public static void main(String[] args) throws IOException {

134

135	if (args.length < 1) {

136	System.out.println("Usage: hdfsclient add/read/delete/mkdir" +

137	" [<local_path> <hdfs_path>]");

138	System.exit(1);

139

}

140

141	HDFSClient client = new HDFSClient();

142	if (args[0].equals("add")) {

143	if (args.length < 3) {

144	System.out.println("Usage: hdfsclient add <local_path> " +

145	"<hdfs_path>");

146	System.exit(1);

147

}

148

149	client.addFile(args[1], args[2]);

150	} else if (args[0].equals("read")) {

151	if (args.length < 2) {

152	System.out.println("Usage: hdfsclient read <hdfs_path>");

153	System.exit(1);

154

}

155

156	client.readFile(args[1]);

157	} else if (args[0].equals("delete")) {

158	if (args.length < 2) {

159	System.out.println("Usage: hdfsclient delete <hdfs_path>");

160	System.exit(1);

161

}

162

163	client.deleteFile(args[1]);

164	} else if (args[0].equals("mkdir")) {

165	if (args.length < 2) {

166	System.out.println("Usage: hdfsclient mkdir <hdfs_path>");

167	System.exit(1);

168

}

169

170	client.mkdir(args[1]);

171

} else {

172	System.out.println("Usage: hdfsclient add/read/delete/mkdir" +

173	" [<local_path> <hdfs_path>]");

174	System.exit(1);

175

}

176

177	System.out.println("Done!");

178

}

179

}

from：http://smallwildpig.iteye.com/blog/1705039   Java对HDFS的操作

http://blog.rajeevsharma.in/2009/06/using-hdfs-in-java-0200.html   Using HDFS in java (0.20.0)