| 1 | {{{ |
| 2 | #!html |
| 3 | <div style="text-align: center; color: blue"><big |
| 4 | style="font-weight: bold;"><big><big>Example</big></big></big></div> |
| 5 | }}} |
| 6 | |
| 7 | [[PageOutline]] |
| 8 | |
| 9 | = 個別編譯程式 = |
| 10 | |
| 11 | == 1 mapper.java == |
| 12 | |
| 13 | 1. new |
| 14 | |
| 15 | || File -> || new -> || mapper || |
| 16 | [[Image(wiki:waue/2009/0617:file-new-mapper.png)]] |
| 17 | |
| 18 | ----------- |
| 19 | |
| 20 | 2. create |
| 21 | |
| 22 | [[Image(wiki:waue/2009/0617:3-1.png)]] |
| 23 | {{{ |
| 24 | #!sh |
| 25 | source folder-> 輸入: icas/src |
| 26 | Package : Sample |
| 27 | Name -> : mapper |
| 28 | }}} |
| 29 | ---------- |
| 30 | |
| 31 | 3. modify |
| 32 | |
| 33 | {{{ |
| 34 | #!java |
| 35 | package Sample; |
| 36 | |
| 37 | import java.io.IOException; |
| 38 | import java.util.StringTokenizer; |
| 39 | |
| 40 | import org.apache.hadoop.io.IntWritable; |
| 41 | import org.apache.hadoop.io.LongWritable; |
| 42 | import org.apache.hadoop.io.Text; |
| 43 | import org.apache.hadoop.mapred.MapReduceBase; |
| 44 | import org.apache.hadoop.mapred.Mapper; |
| 45 | import org.apache.hadoop.mapred.OutputCollector; |
| 46 | import org.apache.hadoop.mapred.Reporter; |
| 47 | |
| 48 | public class mapper extends MapReduceBase implements Mapper<LongWritable, Text, Text, IntWritable> { |
| 49 | private final static IntWritable one = new IntWritable(1); |
| 50 | private Text word = new Text(); |
| 51 | |
| 52 | public void map(LongWritable key, Text value, OutputCollector<Text, IntWritable> output, Reporter reporter) throws IOException { |
| 53 | String line = value.toString(); |
| 54 | StringTokenizer tokenizer = new StringTokenizer(line); |
| 55 | while (tokenizer.hasMoreTokens()) { |
| 56 | word.set(tokenizer.nextToken()); |
| 57 | output.collect(word, one); |
| 58 | } |
| 59 | } |
| 60 | } |
| 61 | |
| 62 | }}} |
| 63 | |
| 64 | 建立mapper.java後,貼入程式碼 |
| 65 | [[Image(wiki:waue/2009/0617:3-2.png)]] |
| 66 | |
| 67 | ------------ |
| 68 | |
| 69 | == 2 reducer.java == |
| 70 | |
| 71 | 1. new |
| 72 | |
| 73 | * File -> new -> reducer |
| 74 | [[Image(wiki:waue/2009/0617:file-new-reducer.png)]] |
| 75 | |
| 76 | ------- |
| 77 | 2. create |
| 78 | [[Image(wiki:waue/2009/0617:3-3.png)]] |
| 79 | |
| 80 | {{{ |
| 81 | #!sh |
| 82 | source folder-> 輸入: icas/src |
| 83 | Package : Sample |
| 84 | Name -> : reducer |
| 85 | }}} |
| 86 | |
| 87 | ----------- |
| 88 | |
| 89 | 3. modify |
| 90 | |
| 91 | {{{ |
| 92 | #!java |
| 93 | package Sample; |
| 94 | |
| 95 | import java.io.IOException; |
| 96 | import java.util.Iterator; |
| 97 | |
| 98 | import org.apache.hadoop.io.IntWritable; |
| 99 | import org.apache.hadoop.io.Text; |
| 100 | import org.apache.hadoop.mapred.MapReduceBase; |
| 101 | import org.apache.hadoop.mapred.OutputCollector; |
| 102 | import org.apache.hadoop.mapred.Reducer; |
| 103 | import org.apache.hadoop.mapred.Reporter; |
| 104 | |
| 105 | public class reducer extends MapReduceBase implements Reducer<Text, IntWritable, Text, IntWritable> { |
| 106 | public void reduce(Text key, Iterator<IntWritable> values, OutputCollector<Text, IntWritable> output, Reporter reporter) throws IOException { |
| 107 | int sum = 0; |
| 108 | while (values.hasNext()) { |
| 109 | sum += values.next().get(); |
| 110 | } |
| 111 | output.collect(key, new IntWritable(sum)); |
| 112 | } |
| 113 | } |
| 114 | }}} |
| 115 | |
| 116 | * File -> new -> Map/Reduce Driver |
| 117 | [[Image(wiki:waue/2009/0617:file-new-mr-driver.png)]] |
| 118 | ---------- |
| 119 | |
| 120 | == 3 WordCount.java (main function) == |
| 121 | |
| 122 | 1. new |
| 123 | |
| 124 | 建立WordCount.java,此檔用來驅動mapper 與 reducer,因此選擇 Map/Reduce Driver |
| 125 | |
| 126 | |
| 127 | [[Image(wiki:waue/2009/0617:3-4.png)]] |
| 128 | ------------ |
| 129 | |
| 130 | 2. create |
| 131 | |
| 132 | {{{ |
| 133 | #!sh |
| 134 | source folder-> 輸入: icas/src |
| 135 | Package : Sample |
| 136 | Name -> : WordCount.java |
| 137 | }}} |
| 138 | |
| 139 | ------- |
| 140 | 3. modify |
| 141 | |
| 142 | {{{ |
| 143 | #!java |
| 144 | package Sample; |
| 145 | import org.apache.hadoop.fs.Path; |
| 146 | import org.apache.hadoop.io.IntWritable; |
| 147 | import org.apache.hadoop.io.Text; |
| 148 | import org.apache.hadoop.mapred.FileInputFormat; |
| 149 | import org.apache.hadoop.mapred.FileOutputFormat; |
| 150 | import org.apache.hadoop.mapred.JobClient; |
| 151 | import org.apache.hadoop.mapred.JobConf; |
| 152 | import org.apache.hadoop.mapred.TextInputFormat; |
| 153 | import org.apache.hadoop.mapred.TextOutputFormat; |
| 154 | |
| 155 | public class WordCount { |
| 156 | |
| 157 | public static void main(String[] args) throws Exception { |
| 158 | JobConf conf = new JobConf(WordCount.class); |
| 159 | conf.setJobName("wordcount"); |
| 160 | |
| 161 | conf.setOutputKeyClass(Text.class); |
| 162 | conf.setOutputValueClass(IntWritable.class); |
| 163 | |
| 164 | conf.setMapperClass(mapper.class); |
| 165 | conf.setCombinerClass(reducer.class); |
| 166 | conf.setReducerClass(reducer.class); |
| 167 | |
| 168 | conf.setInputFormat(TextInputFormat.class); |
| 169 | conf.setOutputFormat(TextOutputFormat.class); |
| 170 | |
| 171 | FileInputFormat.setInputPaths(conf, new Path("/user/hadooper/input")); |
| 172 | FileOutputFormat.setOutputPath(conf, new Path("lab5_out2")); |
| 173 | |
| 174 | JobClient.runJob(conf); |
| 175 | } |
| 176 | } |
| 177 | }}} |
| 178 | |
| 179 | 三個檔完成後並存檔後,整個程式建立完成 |
| 180 | [[Image(wiki:waue/2009/0617:3-5.png)]] |
| 181 | |
| 182 | ------- |
| 183 | |
| 184 | * 三個檔都存檔後,可以看到icas專案下的src,bin都有檔案產生,我們用指令來check |
| 185 | |
| 186 | {{{ |
| 187 | $ cd workspace/icas |
| 188 | $ ls src/Sample/ |
| 189 | mapper.java reducer.java WordCount.java |
| 190 | $ ls bin/Sample/ |
| 191 | mapper.class reducer.class WordCount.class |
| 192 | }}} |
| 193 | |
| 194 | |
| 195 | = eclipse 可以產生出jar檔 = |
| 196 | |
| 197 | File -> Export -> java -> JAR file [[br]] |
| 198 | -> next -> |
| 199 | -------- |
| 200 | 選擇要匯出的專案 -> |
| 201 | jarfile: /home/hadooper/mytest.jar -> [[br]] |
| 202 | next -> |
| 203 | -------- |
| 204 | next -> |
| 205 | -------- |
| 206 | main class: 選擇有Main的class -> [[br]] |
| 207 | Finish |
| 208 | -------- |
| 209 | |
| 210 | * 以上的步驟就可以在/home/hadooper/ 產生出你的 mytest.jar |
| 211 | |
| 212 | |
| 213 | = 用Makefile 來更快速編譯 = |
| 214 | * 程式常常修改,每次都做這些動作也很累很煩,讓我們來體驗一下'''用指令比用圖形介面操作還方便'''吧 |
| 215 | |
| 216 | ==1 產生Makefile 檔 == |
| 217 | |
| 218 | {{{ |
| 219 | $ cd /home/hadooper/workspace/icas/ |
| 220 | $ gedit Makefile |
| 221 | }}} |
| 222 | |
| 223 | * 輸入以下Makefile的內容 (注意 ":" 後面要接 "tab" 而不是 "空白") |
| 224 | {{{ |
| 225 | JarFile="sample-0.1.jar" |
| 226 | MainFunc="Sample.WordCount" |
| 227 | LocalOutDir="/tmp/output" |
| 228 | HADOOP_BIN="/opt/hadoop/bin" |
| 229 | |
| 230 | all:jar run output clean |
| 231 | |
| 232 | jar: |
| 233 | jar -cvf ${JarFile} -C bin/ . |
| 234 | |
| 235 | run: |
| 236 | ${HADOOP_BIN}/hadoop jar ${JarFile} ${MainFunc} input output |
| 237 | |
| 238 | clean: |
| 239 | ${HADOOP_BIN}/hadoop fs -rmr output |
| 240 | |
| 241 | output: |
| 242 | rm -rf ${LocalOutDir} |
| 243 | ${HADOOP_BIN}/hadoop fs -get output ${LocalOutDir} |
| 244 | gedit ${LocalOutDir}/part-r-00000 & |
| 245 | |
| 246 | help: |
| 247 | @echo "Usage:" |
| 248 | @echo " make jar - Build Jar File." |
| 249 | @echo " make clean - Clean up Output directory on HDFS." |
| 250 | @echo " make run - Run your MapReduce code on Hadoop." |
| 251 | @echo " make output - Download and show output file" |
| 252 | @echo " make help - Show Makefile options." |
| 253 | @echo " " |
| 254 | @echo "Example:" |
| 255 | @echo " make jar; make run; make output; make clean" |
| 256 | }}} |
| 257 | |
| 258 | * 或是直接下載 [http://trac.nchc.org.tw/cloud/raw-attachment/wiki/Hadoop_Lab5/Makefile Makefile] 吧 |
| 259 | {{{ |
| 260 | $ cd /home/hadooper/workspace/icas/ |
| 261 | $ wget http://trac.nchc.org.tw/cloud/raw-attachment/wiki/Hadoop_Lab5/Makefile |
| 262 | }}} |
| 263 | |
| 264 | == 4.2.2 執行 == |
| 265 | |
| 266 | * 執行Makefile,可以到該目錄下,執行make [參數],若不知道參數為何,可以打make 或 make help |
| 267 | * make 的用法說明 |
| 268 | |
| 269 | {{{ |
| 270 | $ cd /home/hadooper/workspace/icas/ |
| 271 | $ make |
| 272 | Usage: |
| 273 | make jar - Build Jar File. |
| 274 | make clean - Clean up Output directory on HDFS. |
| 275 | make run - Run your MapReduce code on Hadoop. |
| 276 | make output - Download and show output file |
| 277 | make help - Show Makefile options. |
| 278 | |
| 279 | Example: |
| 280 | make jar; make run; make output; make clean |
| 281 | }}} |
| 282 | |
| 283 | * 下面提供各種make 的參數 |
| 284 | |
| 285 | == make jar == |
| 286 | * 1. 編譯產生jar檔 |
| 287 | |
| 288 | {{{ |
| 289 | $ make jar |
| 290 | }}} |
| 291 | |
| 292 | == make run == |
| 293 | * 2. 跑我們的wordcount 於hadoop上 |
| 294 | |
| 295 | {{{ |
| 296 | $ make run |
| 297 | }}} |
| 298 | |
| 299 | * make run基本上能正確無誤的運作到結束,因此代表我們在eclipse編譯的程式可以順利在hadoop0.18.3的平台上運行。 |
| 300 | |
| 301 | * 而回到eclipse視窗,我們可以看到下方視窗run完的job會呈現出來;左方視窗也多出output資料夾,part-r-00000就是我們的結果檔 |
| 302 | |
| 303 | [[Image(wiki:waue/2009/0617:4-1.png)]] |
| 304 | ------ |
| 305 | * 因為有設定完整的javadoc, 因此可以得到詳細的解說與輔助 |
| 306 | [[Image(wiki:waue/2009/0617:4-2.png)]] |
| 307 | |
| 308 | == make output == |
| 309 | * 3. 這個指令是幫助使用者將結果檔從hdfs下載到local端,並且用gedit來開啟你的結果檔 |
| 310 | |
| 311 | {{{ |
| 312 | $ make output |
| 313 | }}} |
| 314 | |
| 315 | == make clean == |
| 316 | * 4. 這個指令用來把hdfs上的output資料夾清除。如果你還想要在跑一次make run,請先執行make clean,否則hadoop會告訴你,output資料夾已經存在,而拒絕工作喔! |
| 317 | |
| 318 | {{{ |
| 319 | $ make clean |
| 320 | }}} |
| 321 | |
| 322 | |
| 323 | = 練習:匯入專案 = |
| 324 | * 將 [http://trac.nchc.org.tw/cloud/raw-attachment/wiki/Hadoop_Lab5/hadoop_sample_codes.zip nchc-sample] 給匯入到eclipse 內開發吧! |