diff --git a/src/main/scala/etl.scala b/src/main/scala/etl.scala
index 965ebf1..8b853a3 100644
--- a/src/main/scala/etl.scala
+++ b/src/main/scala/etl.scala
@@ -3,8 +3,80 @@
   */
 package cn.piflow
 
-class SparkETLProcess extends Process{
-  override def run(pc: ProcessContext): Unit = {
+import java.util.concurrent.atomic.AtomicInteger
 
+import org.apache.spark.sql.DataFrame
+import org.apache.spark.sql.execution.datasources.{FileFormat => SparkFileFormat}
+
+import scala.collection.mutable.{ArrayBuffer, Map => MMap}
+
+class SparkETLProcess extends Process {
+  val ends = ArrayBuffer[() => Unit]();
+  val idgen = new AtomicInteger();
+
+  override def run(pc: ProcessContext): Unit = {
+    ends.foreach(_.apply());
   }
+
+  abstract class AbstractStream extends Stream {
+    val id = idgen.incrementAndGet();
+
+    override def getId(): Int = id;
+  }
+
+  def loadStream(streamSource: StreamSource): Stream = {
+    return new AbstractStream() {
+      override def getDataFrame(): DataFrame = {
+        println("load");
+        null;
+      }
+    }
+  }
+
+  def writeStream(stream: Stream, streamSink: StreamSink): Unit = {
+    ends += { () => {
+      println("write");
+      stream.getDataFrame().show();
+    }
+    };
+  }
+
+  def transform(stream: Stream, transformer: Transformer): Stream = {
+    return new AbstractStream() {
+      override def getDataFrame(): DataFrame = {
+        println("transform");
+        stream.getDataFrame();
+      }
+    }
+  }
+}
+
+trait Stream {
+  def getId(): Int;
+
+  def getDataFrame(): DataFrame;
+}
+
+trait StreamSource {
+
+}
+
+trait Transformer {
+
+}
+
+trait StreamSink {
+
+}
+
+case class TextFile(path: String, format: String) extends StreamSource with StreamSink {
+
+}
+
+case class DoMap(mapFunc: String) extends Transformer {
+
+}
+
+case class DoFlatMap(mapFunc: String) extends Transformer {
+
 }
\ No newline at end of file
diff --git a/src/test/scala/FlowTest.scala b/src/test/scala/FlowTest.scala
index c64bacc..87943d5 100644
--- a/src/test/scala/FlowTest.scala
+++ b/src/test/scala/FlowTest.scala
@@ -50,21 +50,20 @@ class FlowTest {
   @Test
   def test2() {
     val fg = new SparkETLProcess();
-    val node1 = fg.createNode(DoLoad(TextFile("./out/honglou.txt")));
-    val node2 = fg.createNode(DoMap(
+    val s1 = fg.loadStream(TextFile("./out/honglou.txt", "text"));
+    val s2 = fg.transform(s1, DoMap(
       """
       function(s){
         return s.replaceAll("[\\x00-\\xff]|，|。|：|．|“|”|？|！|　", "");
       }"""));
 
-    val node3 = fg.createNode(DoFlatMap(
+    val s3 = fg.transform(s2, DoFlatMap(
       """
       function(s){
         return s.zip(s.drop(1)).map(t => "" + t._1 + t._2);
       }"""));
 
-    val node4 = fg.createNode(DoWrite(JsonFile("./out/wordcount")));
-    fg.pipe(node1, node2, node3, node4);
+    fg.writeStream(s3, TextFile("./out/wordcount", "json"));
 
     _testFlow1(fg);
   }
diff --git a/src/test/scala/SparkETLTest.scala b/src/test/scala/SparkETLTest.scala
new file mode 100644
index 0000000..1e1b3e5
--- /dev/null
+++ b/src/test/scala/SparkETLTest.scala
@@ -0,0 +1,29 @@
+/**
+  * Created by bluejoe on 2018/5/6.
+  */
+
+import cn.piflow._
+import org.junit.Test
+
+class SparkETLTest {
+  @Test
+  def test1(): Unit = {
+    val fg = new SparkETLProcess();
+    val s1 = fg.loadStream(TextFile("./out/honglou.txt", "text"));
+    val s2 = fg.transform(s1, DoMap(
+      """
+      function(s){
+        return s.replaceAll("[\\x00-\\xff]|，|。|：|．|“|”|？|！|　", "");
+      }"""));
+
+    val s3 = fg.transform(s2, DoFlatMap(
+      """
+      function(s){
+        return s.zip(s.drop(1)).map(t => "" + t._1 + t._2);
+      }"""));
+
+    fg.writeStream(s3, TextFile("./out/wordcount", "json"));
+
+    fg.run(null);
+  }
+}