cleanup

5 files changed, 75 insertions, 0 deletions

diff --git a/hadoop/streaming/mapper/input b/hadoop/streaming/mapper/input
new file mode 100644
index 0000000..338fd87
--- /dev/null
+++ b/hadoop/streaming/mapper/input
@@ -0,0 +1,10 @@
+0	1	a	c
+0	2	b	c
+0	3	c	c
+1	4	a	c
+1	5	b	c
+1	6	c	c
+1	7	d	c
+2	8	a	c
+2	9	b	c
+2	10	c	c
diff --git a/hadoop/streaming/mapper/mapper_test.py b/hadoop/streaming/mapper/mapper_test.py
new file mode 100644
index 0000000..d358bda
--- /dev/null
+++ b/hadoop/streaming/mapper/mapper_test.py
@@ -0,0 +1,9 @@
+#!/usr/bin/env python2
+
+import sys
+
+
+if __name__ == "__main__":
+  for line in sys.stdin:
+    print line.upper()
+
diff --git a/hadoop/streaming/mapper/mapper_test.sh b/hadoop/streaming/mapper/mapper_test.sh
new file mode 100755
index 0000000..f0c5da3
--- /dev/null
+++ b/hadoop/streaming/mapper/mapper_test.sh
@@ -0,0 +1,23 @@
+#!/bin/sh
+
+hadoop dfs -put input input
+
+HADOOP_HOME=/usr/lib/hadoop
+HADOOP_VERSION=0.20.2-cdh3u1
+JAR=contrib/streaming/hadoop-streaming-$HADOOP_VERSION.jar
+HSTREAMING="$HADOOP_HOME/bin/hadoop jar $HADOOP_HOME/$JAR"
+
+OUT=mapper_test_out
+
+$HSTREAMING \
+    -input input \
+    -output $OUT \
+    -mapper "python mapper.py" \
+    -file mapper_test.py \
+    -reducer org.apache.hadoop.mapred.lib.IdentityReducer \
+    -jobconf mapred.reduce.tasks=3
+
+hadoop dfs -get $OUT .
+hadoop dfs -rmr $OUT
+hadoop dfs -rm input
+
diff --git a/hadoop/streaming/mapper/mapper_test1.py b/hadoop/streaming/mapper/mapper_test1.py
new file mode 100644
index 0000000..79c8aa6
--- /dev/null
+++ b/hadoop/streaming/mapper/mapper_test1.py
@@ -0,0 +1,9 @@
+#!/usr/bin/env python2
+
+import sys
+
+
+if __name__ == "__main__":
+  for line in sys.stdin:
+    print "MYKEY\t%s"%line.upper()
+
diff --git a/hadoop/streaming/mapper/mapper_test1.sh b/hadoop/streaming/mapper/mapper_test1.sh
new file mode 100755
index 0000000..475699f
--- /dev/null
+++ b/hadoop/streaming/mapper/mapper_test1.sh
@@ -0,0 +1,24 @@
+#!/bin/sh
+
+
+hadoop dfs -put input input
+
+HADOOP_HOME=/usr/lib/hadoop
+HADOOP_VERSION=0.20.2-cdh3u1
+JAR=contrib/streaming/hadoop-streaming-$HADOOP_VERSION.jar
+HSTREAMING="$HADOOP_HOME/bin/hadoop jar $HADOOP_HOME/$JAR"
+
+OUT=mapper_test1_out
+
+$HSTREAMING \
+    -input input \
+    -output $OUT \
+    -mapper "python mapper1.py" \
+    -file mapper_test1.py \
+    -reducer org.apache.hadoop.mapred.lib.IdentityReducer \
+    -jobconf mapred.reduce.tasks=3
+
+hadoop dfs -get $OUT .
+hadoop dfs -rmr $OUT
+hadoop dfs -rm input
+