001/** 002 * Licensed to the Apache Software Foundation (ASF) under one 003 * or more contributor license agreements. See the NOTICE file 004 * distributed with this work for additional information 005 * regarding copyright ownership. The ASF licenses this file 006 * to you under the Apache License, Version 2.0 (the 007 * "License"); you may not use this file except in compliance 008 * with the License. You may obtain a copy of the License at 009 * 010 * http://www.apache.org/licenses/LICENSE-2.0 011 * 012 * Unless required by applicable law or agreed to in writing, software 013 * distributed under the License is distributed on an "AS IS" BASIS, 014 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 015 * See the License for the specific language governing permissions and 016 * limitations under the License. 017 */ 018 019package org.apache.hadoop.mapred; 020 021import java.io.IOException; 022import java.text.NumberFormat; 023 024import org.apache.hadoop.classification.InterfaceAudience; 025import org.apache.hadoop.classification.InterfaceStability; 026import org.apache.hadoop.fs.FileSystem; 027import org.apache.hadoop.fs.Path; 028import org.apache.hadoop.io.compress.CompressionCodec; 029import org.apache.hadoop.mapred.FileAlreadyExistsException; 030import org.apache.hadoop.mapreduce.security.TokenCache; 031import org.apache.hadoop.util.Progressable; 032 033/** A base class for {@link OutputFormat}. */ 034@InterfaceAudience.Public 035@InterfaceStability.Stable 036public abstract class FileOutputFormat<K, V> implements OutputFormat<K, V> { 037 038 /** 039 * Set whether the output of the job is compressed. 040 * @param conf the {@link JobConf} to modify 041 * @param compress should the output of the job be compressed? 042 */ 043 public static void setCompressOutput(JobConf conf, boolean compress) { 044 conf.setBoolean(org.apache.hadoop.mapreduce.lib.output. 045 FileOutputFormat.COMPRESS, compress); 046 } 047 048 /** 049 * Is the job output compressed? 050 * @param conf the {@link JobConf} to look in 051 * @return <code>true</code> if the job output should be compressed, 052 * <code>false</code> otherwise 053 */ 054 public static boolean getCompressOutput(JobConf conf) { 055 return conf.getBoolean(org.apache.hadoop.mapreduce.lib.output. 056 FileOutputFormat.COMPRESS, false); 057 } 058 059 /** 060 * Set the {@link CompressionCodec} to be used to compress job outputs. 061 * @param conf the {@link JobConf} to modify 062 * @param codecClass the {@link CompressionCodec} to be used to 063 * compress the job outputs 064 */ 065 public static void 066 setOutputCompressorClass(JobConf conf, 067 Class<? extends CompressionCodec> codecClass) { 068 setCompressOutput(conf, true); 069 conf.setClass(org.apache.hadoop.mapreduce.lib.output. 070 FileOutputFormat.COMPRESS_CODEC, codecClass, 071 CompressionCodec.class); 072 } 073 074 /** 075 * Get the {@link CompressionCodec} for compressing the job outputs. 076 * @param conf the {@link JobConf} to look in 077 * @param defaultValue the {@link CompressionCodec} to return if not set 078 * @return the {@link CompressionCodec} to be used to compress the 079 * job outputs 080 * @throws IllegalArgumentException if the class was specified, but not found 081 */ 082 public static Class<? extends CompressionCodec> 083 getOutputCompressorClass(JobConf conf, 084 Class<? extends CompressionCodec> defaultValue) { 085 Class<? extends CompressionCodec> codecClass = defaultValue; 086 087 String name = conf.get(org.apache.hadoop.mapreduce.lib.output. 088 FileOutputFormat.COMPRESS_CODEC); 089 if (name != null) { 090 try { 091 codecClass = 092 conf.getClassByName(name).asSubclass(CompressionCodec.class); 093 } catch (ClassNotFoundException e) { 094 throw new IllegalArgumentException("Compression codec " + name + 095 " was not found.", e); 096 } 097 } 098 return codecClass; 099 } 100 101 public abstract RecordWriter<K, V> getRecordWriter(FileSystem ignored, 102 JobConf job, String name, 103 Progressable progress) 104 throws IOException; 105 106 public void checkOutputSpecs(FileSystem ignored, JobConf job) 107 throws FileAlreadyExistsException, 108 InvalidJobConfException, IOException { 109 // Ensure that the output directory is set and not already there 110 Path outDir = getOutputPath(job); 111 if (outDir == null && job.getNumReduceTasks() != 0) { 112 throw new InvalidJobConfException("Output directory not set in JobConf."); 113 } 114 if (outDir != null) { 115 FileSystem fs = outDir.getFileSystem(job); 116 // normalize the output directory 117 outDir = fs.makeQualified(outDir); 118 setOutputPath(job, outDir); 119 120 // get delegation token for the outDir's file system 121 TokenCache.obtainTokensForNamenodes(job.getCredentials(), 122 new Path[] {outDir}, job); 123 124 // check its existence 125 if (fs.exists(outDir)) { 126 throw new FileAlreadyExistsException("Output directory " + outDir + 127 " already exists"); 128 } 129 } 130 } 131 132 /** 133 * Set the {@link Path} of the output directory for the map-reduce job. 134 * 135 * @param conf The configuration of the job. 136 * @param outputDir the {@link Path} of the output directory for 137 * the map-reduce job. 138 */ 139 public static void setOutputPath(JobConf conf, Path outputDir) { 140 outputDir = new Path(conf.getWorkingDirectory(), outputDir); 141 conf.set(org.apache.hadoop.mapreduce.lib.output. 142 FileOutputFormat.OUTDIR, outputDir.toString()); 143 } 144 145 /** 146 * Set the {@link Path} of the task's temporary output directory 147 * for the map-reduce job. 148 * 149 * <p><i>Note</i>: Task output path is set by the framework. 150 * </p> 151 * @param conf The configuration of the job. 152 * @param outputDir the {@link Path} of the output directory 153 * for the map-reduce job. 154 */ 155 156 static void setWorkOutputPath(JobConf conf, Path outputDir) { 157 outputDir = new Path(conf.getWorkingDirectory(), outputDir); 158 conf.set(JobContext.TASK_OUTPUT_DIR, outputDir.toString()); 159 } 160 161 /** 162 * Get the {@link Path} to the output directory for the map-reduce job. 163 * 164 * @return the {@link Path} to the output directory for the map-reduce job. 165 * @see FileOutputFormat#getWorkOutputPath(JobConf) 166 */ 167 public static Path getOutputPath(JobConf conf) { 168 String name = conf.get(org.apache.hadoop.mapreduce.lib.output. 169 FileOutputFormat.OUTDIR); 170 return name == null ? null: new Path(name); 171 } 172 173 /** 174 * Get the {@link Path} to the task's temporary output directory 175 * for the map-reduce job 176 * 177 * <h4 id="SideEffectFiles">Tasks' Side-Effect Files</h4> 178 * 179 * <p><i>Note:</i> The following is valid only if the {@link OutputCommitter} 180 * is {@link FileOutputCommitter}. If <code>OutputCommitter</code> is not 181 * a <code>FileOutputCommitter</code>, the task's temporary output 182 * directory is same as {@link #getOutputPath(JobConf)} i.e. 183 * <tt>${mapreduce.output.fileoutputformat.outputdir}$</tt></p> 184 * 185 * <p>Some applications need to create/write-to side-files, which differ from 186 * the actual job-outputs. 187 * 188 * <p>In such cases there could be issues with 2 instances of the same TIP 189 * (running simultaneously e.g. speculative tasks) trying to open/write-to the 190 * same file (path) on HDFS. Hence the application-writer will have to pick 191 * unique names per task-attempt (e.g. using the attemptid, say 192 * <tt>attempt_200709221812_0001_m_000000_0</tt>), not just per TIP.</p> 193 * 194 * <p>To get around this the Map-Reduce framework helps the application-writer 195 * out by maintaining a special 196 * <tt>${mapreduce.output.fileoutputformat.outputdir}/_temporary/_${taskid}</tt> 197 * sub-directory for each task-attempt on HDFS where the output of the 198 * task-attempt goes. On successful completion of the task-attempt the files 199 * in the <tt>${mapreduce.output.fileoutputformat.outputdir}/_temporary/_${taskid}</tt> (only) 200 * are <i>promoted</i> to <tt>${mapreduce.output.fileoutputformat.outputdir}</tt>. Of course, the 201 * framework discards the sub-directory of unsuccessful task-attempts. This 202 * is completely transparent to the application.</p> 203 * 204 * <p>The application-writer can take advantage of this by creating any 205 * side-files required in <tt>${mapreduce.task.output.dir}</tt> during execution 206 * of his reduce-task i.e. via {@link #getWorkOutputPath(JobConf)}, and the 207 * framework will move them out similarly - thus she doesn't have to pick 208 * unique paths per task-attempt.</p> 209 * 210 * <p><i>Note</i>: the value of <tt>${mapreduce.task.output.dir}</tt> during 211 * execution of a particular task-attempt is actually 212 * <tt>${mapreduce.output.fileoutputformat.outputdir}/_temporary/_{$taskid}</tt>, and this value is 213 * set by the map-reduce framework. So, just create any side-files in the 214 * path returned by {@link #getWorkOutputPath(JobConf)} from map/reduce 215 * task to take advantage of this feature.</p> 216 * 217 * <p>The entire discussion holds true for maps of jobs with 218 * reducer=NONE (i.e. 0 reduces) since output of the map, in that case, 219 * goes directly to HDFS.</p> 220 * 221 * @return the {@link Path} to the task's temporary output directory 222 * for the map-reduce job. 223 */ 224 public static Path getWorkOutputPath(JobConf conf) { 225 String name = conf.get(JobContext.TASK_OUTPUT_DIR); 226 return name == null ? null: new Path(name); 227 } 228 229 /** 230 * Helper function to create the task's temporary output directory and 231 * return the path to the task's output file. 232 * 233 * @param conf job-configuration 234 * @param name temporary task-output filename 235 * @return path to the task's temporary output file 236 * @throws IOException 237 */ 238 public static Path getTaskOutputPath(JobConf conf, String name) 239 throws IOException { 240 // ${mapred.out.dir} 241 Path outputPath = getOutputPath(conf); 242 if (outputPath == null) { 243 throw new IOException("Undefined job output-path"); 244 } 245 246 OutputCommitter committer = conf.getOutputCommitter(); 247 Path workPath = outputPath; 248 TaskAttemptContext context = 249 new TaskAttemptContextImpl(conf, 250 TaskAttemptID.forName(conf.get( 251 JobContext.TASK_ATTEMPT_ID))); 252 if (committer instanceof FileOutputCommitter) { 253 workPath = ((FileOutputCommitter)committer).getWorkPath(context, 254 outputPath); 255 } 256 257 // ${mapred.out.dir}/_temporary/_${taskid}/${name} 258 return new Path(workPath, name); 259 } 260 261 /** 262 * Helper function to generate a name that is unique for the task. 263 * 264 * <p>The generated name can be used to create custom files from within the 265 * different tasks for the job, the names for different tasks will not collide 266 * with each other.</p> 267 * 268 * <p>The given name is postfixed with the task type, 'm' for maps, 'r' for 269 * reduces and the task partition number. For example, give a name 'test' 270 * running on the first map o the job the generated name will be 271 * 'test-m-00000'.</p> 272 * 273 * @param conf the configuration for the job. 274 * @param name the name to make unique. 275 * @return a unique name accross all tasks of the job. 276 */ 277 public static String getUniqueName(JobConf conf, String name) { 278 int partition = conf.getInt(JobContext.TASK_PARTITION, -1); 279 if (partition == -1) { 280 throw new IllegalArgumentException( 281 "This method can only be called from within a Job"); 282 } 283 284 String taskType = (conf.getBoolean(JobContext.TASK_ISMAP, true)) ? "m" : "r"; 285 286 NumberFormat numberFormat = NumberFormat.getInstance(); 287 numberFormat.setMinimumIntegerDigits(5); 288 numberFormat.setGroupingUsed(false); 289 290 return name + "-" + taskType + "-" + numberFormat.format(partition); 291 } 292 293 /** 294 * Helper function to generate a {@link Path} for a file that is unique for 295 * the task within the job output directory. 296 * 297 * <p>The path can be used to create custom files from within the map and 298 * reduce tasks. The path name will be unique for each task. The path parent 299 * will be the job output directory.</p>ls 300 * 301 * <p>This method uses the {@link #getUniqueName} method to make the file name 302 * unique for the task.</p> 303 * 304 * @param conf the configuration for the job. 305 * @param name the name for the file. 306 * @return a unique path accross all tasks of the job. 307 */ 308 public static Path getPathForCustomFile(JobConf conf, String name) { 309 return new Path(getWorkOutputPath(conf), getUniqueName(conf, name)); 310 } 311} 312