do not store nulls in hbase. see issue 5

1 files changed, 13 insertions, 15 deletions

diff --git a/src/main/scala/parallelai/spyglass/hbase/HBaseConversions.scala b/src/main/scala/parallelai/spyglass/hbase/HBaseConversions.scala
index b6d5742..31ed3ea 100644
--- a/src/main/scala/parallelai/spyglass/hbase/HBaseConversions.scala
+++ b/src/main/scala/parallelai/spyglass/hbase/HBaseConversions.scala
@@ -10,15 +10,14 @@ import org.apache.hadoop.hbase.util.Bytes
 import cascading.tuple.TupleEntry
 
 class HBasePipeWrapper (pipe: Pipe) {
-   def toBytesWritable(f: Fields): Pipe = {
+    def toBytesWritable(f: Fields): Pipe = {
 	  asList(f)
-     .foldLeft(pipe){ (p, f) => {
-	    p.map(f.toString -> f.toString){ from: String => {
-	      new ImmutableBytesWritable(Bytes.toBytes(
-	          if (from == null) "" else from))
-	    }}
-	  }} 
-	}
+        .foldLeft(pipe){ (p, f) => {
+	      p.map(f.toString -> f.toString){ from: String =>
+            Option(from).map(x => new ImmutableBytesWritable(Bytes.toBytes(x))).getOrElse(null)
+          }}
+      }
+    }
 
 //   def toBytesWritable : Pipe = {
 //	  asList(Fields.ALL.asInstanceOf[TupleEntry].getFields()).foldLeft(pipe){ (p, f) => {
@@ -30,13 +29,12 @@ class HBasePipeWrapper (pipe: Pipe) {
 
 	def fromBytesWritable(f: Fields): Pipe = {
 	  asList(f)
-	  .foldLeft(pipe) { (p, fld) =>
-	    p.map(fld.toString -> fld.toString) { from: ImmutableBytesWritable => {
-	    	Bytes.toString(from.get)
-	      }
-	    }
-	  }
-	}
+	    .foldLeft(pipe) { (p, fld) => {
+	      p.map(fld.toString -> fld.toString) { from: ImmutableBytesWritable =>
+            Option(from).map(x => Bytes.toString(x.get)).getOrElse(null)
+          }
+        }}
+    }
 
 //	def fromBytesWritable : Pipe = {
 //	  asList(Fields.ALL.asInstanceOf[TupleEntry].getFields()).foldLeft(pipe) { (p, fld) =>