1 | package org.hathitrust.extractedfeatures;
|
---|
2 |
|
---|
3 | import java.io.BufferedInputStream;
|
---|
4 | import java.io.BufferedOutputStream;
|
---|
5 | import java.io.BufferedReader;
|
---|
6 | import java.io.BufferedWriter;
|
---|
7 | import java.io.FileOutputStream;
|
---|
8 | import java.io.IOException;
|
---|
9 | import java.io.InputStreamReader;
|
---|
10 | import java.io.OutputStreamWriter;
|
---|
11 | import java.net.URI;
|
---|
12 | import java.net.URISyntaxException;
|
---|
13 |
|
---|
14 | import org.apache.commons.compress.compressors.CompressorException;
|
---|
15 | import org.apache.commons.compress.compressors.CompressorInputStream;
|
---|
16 | import org.apache.commons.compress.compressors.CompressorOutputStream;
|
---|
17 | import org.apache.commons.compress.compressors.CompressorStreamFactory;
|
---|
18 | import org.apache.hadoop.conf.Configuration;
|
---|
19 | import org.apache.hadoop.fs.FSDataInputStream;
|
---|
20 | import org.apache.hadoop.fs.FSDataOutputStream;
|
---|
21 | import org.apache.hadoop.fs.FileSystem;
|
---|
22 | import org.apache.hadoop.fs.Path;
|
---|
23 |
|
---|
24 | public class ClusterFileIO {
|
---|
25 |
|
---|
26 |
|
---|
27 | public static void memory_usage(String prefix)
|
---|
28 | {
|
---|
29 | Runtime runtime = Runtime.getRuntime();
|
---|
30 |
|
---|
31 | java.text.NumberFormat format = java.text.NumberFormat.getInstance();
|
---|
32 |
|
---|
33 | StringBuilder sb = new StringBuilder();
|
---|
34 | long maxMemory = runtime.maxMemory();
|
---|
35 | long allocatedMemory = runtime.totalMemory();
|
---|
36 | long freeMemory = runtime.freeMemory();
|
---|
37 |
|
---|
38 | sb.append(prefix+" free memory: " + format.format(freeMemory / 1024) + "\n");
|
---|
39 | sb.append(prefix+" allocated memory: " + format.format(allocatedMemory / 1024) + "\n");
|
---|
40 | sb.append(prefix+" max memory: " + format.format(maxMemory / 1024) + "\n");
|
---|
41 | sb.append(prefix+" total free memory: " + format.format((freeMemory + (maxMemory - allocatedMemory)) / 1024) + "\n");
|
---|
42 |
|
---|
43 | System.out.print(sb.toString());
|
---|
44 | }
|
---|
45 |
|
---|
46 |
|
---|
47 | protected static FileSystem getFileSystemInstance(String input_file_or_dir)
|
---|
48 | {
|
---|
49 | FileSystem fs = null;
|
---|
50 |
|
---|
51 | try {
|
---|
52 | Configuration conf = new Configuration();
|
---|
53 | URI uri = new URI(input_file_or_dir);
|
---|
54 | fs = FileSystem.newInstance(uri,conf);
|
---|
55 | }
|
---|
56 | catch (URISyntaxException e) {
|
---|
57 | e.printStackTrace();
|
---|
58 | }
|
---|
59 | catch (IOException e) {
|
---|
60 | e.printStackTrace();
|
---|
61 | }
|
---|
62 |
|
---|
63 | return fs;
|
---|
64 | }
|
---|
65 |
|
---|
66 | public static boolean isHDFS(String fileIn)
|
---|
67 | {
|
---|
68 | return fileIn.startsWith("hdfs://");
|
---|
69 | }
|
---|
70 |
|
---|
71 | public static boolean exists(String file)
|
---|
72 | {
|
---|
73 | FileSystem fs = getFileSystemInstance(file);
|
---|
74 |
|
---|
75 | boolean exists = false;
|
---|
76 |
|
---|
77 | try {
|
---|
78 | Path path = new Path(file);
|
---|
79 | exists = fs.exists(path);
|
---|
80 | } catch (IllegalArgumentException e) {
|
---|
81 | exists = false;
|
---|
82 | } catch (IOException e) {
|
---|
83 | exists = false;
|
---|
84 | }
|
---|
85 |
|
---|
86 | return exists;
|
---|
87 | }
|
---|
88 |
|
---|
89 | public static String removeSuffix(String file,String suffix)
|
---|
90 | {
|
---|
91 | return file.substring(0,file.length() - suffix.length());
|
---|
92 | }
|
---|
93 |
|
---|
94 | public static boolean createDirectoryAll(String dir)
|
---|
95 | {
|
---|
96 | FileSystem fs = getFileSystemInstance(dir);
|
---|
97 | boolean created_dir = false;
|
---|
98 |
|
---|
99 | if (!exists(dir)) {
|
---|
100 | try {
|
---|
101 | URI uri = new URI(dir);
|
---|
102 | Path path = new Path(uri);
|
---|
103 | fs.mkdirs(path);
|
---|
104 | created_dir = true;
|
---|
105 | } catch (URISyntaxException e) {
|
---|
106 | e.printStackTrace();
|
---|
107 | } catch (IOException e) {
|
---|
108 | e.printStackTrace();
|
---|
109 | }
|
---|
110 | }
|
---|
111 |
|
---|
112 | return created_dir;
|
---|
113 | }
|
---|
114 |
|
---|
115 | public static BufferedInputStream getBufferedInputStream(String fileIn)
|
---|
116 | throws IOException
|
---|
117 | {
|
---|
118 | FileSystem fs = getFileSystemInstance(fileIn);
|
---|
119 |
|
---|
120 | BufferedInputStream bis = null;
|
---|
121 |
|
---|
122 | if (isHDFS(fileIn)) {
|
---|
123 | URI uri = URI.create (fileIn);
|
---|
124 | //Configuration conf = new Configuration();
|
---|
125 | //FileSystem file = FileSystem.get(uri, conf);
|
---|
126 | //FSDataInputStream fin = file.open(new Path(uri));
|
---|
127 |
|
---|
128 | //FSDataInputStream fin = _fs.open(new Path(fileIn));
|
---|
129 | FSDataInputStream fin = fs.open(new Path(uri));
|
---|
130 |
|
---|
131 | bis = new BufferedInputStream(fin);
|
---|
132 | }
|
---|
133 | else {
|
---|
134 | /*
|
---|
135 | // Trim 'file://' off the front
|
---|
136 | String local_file_in = fileIn;
|
---|
137 | if (local_file_in.startsWith("file://")) {
|
---|
138 | local_file_in = fileIn.substring("file://".length());
|
---|
139 | }
|
---|
140 | FileInputStream fin = new FileInputStream(local_file_in);
|
---|
141 | bis = new BufferedInputStream(fin);
|
---|
142 | */
|
---|
143 |
|
---|
144 | FSDataInputStream fin = fs.open(new Path(fileIn));
|
---|
145 |
|
---|
146 | bis = new BufferedInputStream(fin);
|
---|
147 | }
|
---|
148 |
|
---|
149 | return bis;
|
---|
150 | }
|
---|
151 |
|
---|
152 | public static BufferedOutputStream getBufferedOutputStream(String fileOut)
|
---|
153 | throws IOException
|
---|
154 | {
|
---|
155 | BufferedOutputStream bos = null;
|
---|
156 |
|
---|
157 | if (fileOut.startsWith("hdfs://")) {
|
---|
158 | URI uri = URI.create (fileOut);
|
---|
159 | Configuration conf = new Configuration();
|
---|
160 | FileSystem file = FileSystem.get(uri, conf);
|
---|
161 | FSDataOutputStream fout = file.create(new Path(uri));
|
---|
162 |
|
---|
163 | bos = new BufferedOutputStream(fout);
|
---|
164 | }
|
---|
165 | else {
|
---|
166 | // Trim 'file://' off the front
|
---|
167 | String local_file_out = fileOut;
|
---|
168 | if (local_file_out.startsWith("file://")) {
|
---|
169 | local_file_out = fileOut.substring("file://".length());
|
---|
170 | }
|
---|
171 | FileOutputStream fout = new FileOutputStream(local_file_out);
|
---|
172 | bos = new BufferedOutputStream(fout);
|
---|
173 | }
|
---|
174 |
|
---|
175 | return bos;
|
---|
176 | }
|
---|
177 |
|
---|
178 | public static BufferedReader getBufferedReaderForCompressedFile(String fileIn)
|
---|
179 | throws IOException, CompressorException
|
---|
180 | {
|
---|
181 | BufferedInputStream bis = getBufferedInputStream(fileIn);
|
---|
182 | CompressorInputStream cis = new CompressorStreamFactory().createCompressorInputStream(bis);
|
---|
183 | BufferedReader br = new BufferedReader(new InputStreamReader(cis,"UTF8"));
|
---|
184 | return br;
|
---|
185 | }
|
---|
186 |
|
---|
187 | public static BufferedWriter getBufferedWriterForCompressedFile(String fileOut)
|
---|
188 | throws IOException, CompressorException
|
---|
189 | {
|
---|
190 | BufferedOutputStream bos = getBufferedOutputStream(fileOut);
|
---|
191 | CompressorStreamFactory csf = new CompressorStreamFactory();
|
---|
192 | CompressorOutputStream cos = csf.createCompressorOutputStream(CompressorStreamFactory.BZIP2,bos);
|
---|
193 | BufferedWriter bw = new BufferedWriter(new OutputStreamWriter(cos,"UTF8"));
|
---|
194 | return bw;
|
---|
195 | }
|
---|
196 |
|
---|
197 | }
|
---|