◢ <實作三> | <回課程大綱> ▲ | <實作五> ◣

實作四 Lab4

MapReduce 程式編譯
Compiling Hadoop MapReduce Java Program

Practice 1 : Word Count (Basic)

• 上傳內容到 HDFS 內
  upload data to HDFS

$ mkdir lab6_input
$ echo "I like NCTU Cloud Course." > lab6_input/input1
$ echo "I like nctu Cloud Course, and we enjoy this course." > lab6_input/input2
$ hadoop fs -put lab6_input lab6_input
$ hadoop fs -ls lab6_input
Found 2 items
-rw-r--r--   2 hXXXX supergroup         26 2011-04-19 10:07 /user/hXXXX/lab6_input/input1
-rw-r--r--   2 hXXXX supergroup         52 2011-04-19 10:07 /user/hXXXX/lab6_input/input2

• 下載 ​WordCount.java 並存到家目錄;
  Download ​WordCount.java and save to your home directory

  ~$ wget http://hadoop.nchc.org.tw/WordCount.java
  

• 運作程式
  Compile WordCount.java and run it by hadoop jar command

$ mkdir MyJava
$ ln -s /usr/lib/hadoop/hadoop-*-core.jar hadoop-core.jar
$ javac -classpath hadoop-core.jar -d MyJava WordCount.java
$ jar -cvf wordcount.jar -C MyJava .
$ hadoop jar wordcount.jar WordCount lab6_input/ lab6_out1/
$ hadoop fs -cat lab6_out1/part-00000

• lab6_out1 執行結果
  You should see results like this :

  Cloud 2
  Course, 1
  Course. 1
  I 2
  NCTU  1
  and 1
  course. 1
  enjoy 1
  like  2
  nctu  1
  this  1
  we  1
  

---

Practice 2 : Word Count (Advanced)

$ echo "\." >pattern.txt && echo "\," >>pattern.txt
$ hadoop fs -put pattern.txt .
$ mkdir -p MyJava2

• 下載 ​WordCount2.java 並存到家目錄;
  Download ​WordCount2.java to home directory

  ~$ wget http://hadoop.nchc.org.tw/WordCount2.java
  

$ javac -classpath hadoop-core.jar -d MyJava2 WordCount2.java
$ jar -cvf wordcount2.jar -C MyJava2 .
$ hadoop jar wordcount2.jar WordCount2 lab6_input lab6_out2 -skip pattern.txt
$ hadoop fs -cat lab6_out2/part-00000

• lab6_out2 執行結果
  You should see results like this:

  Cloud 2
  Course  2
  I 2
  NCTU  1
  and 1
  course  1
  enjoy 1
  like  2
  nctu  1
  this  1
  we  1
  

• Let's given case insensitive and ignore pattern for this example

  $ hadoop jar wordcount2.jar WordCount2 -Dwordcount.case.sensitive=false lab6_input lab6_out3 -skip pattern.txt
  $ hadoop fs -cat lab6_out3/part-00000
  

• lab6_out3 執行結果
  You should see results like this:

  and 1
  cloud 2
  course  3
  enjoy 1
  i 2
  like  2
  nctu  2
  this  1
  we  1