Switch to getting resources as streams instead of files (databricks#101)

Replace calls to getResource (which look for resource files only on the local disk) with calls to getResourceAsStream, which uses the classloader to search for resource files anywhere in the classpath.

Author: smurching

src/test/scala/com/databricks/sparkdl/ImageUtilsSuite.scala

@@ -32,8 +32,8 @@ object ImageUtilsSuite {
 
   /** Read image data into a BufferedImage, then use our utility method to convert to a row image */
   def getImageRow(resourcePath: String): Row = {
-    val resourceFilename = getClass.getResource(resourcePath).getFile
-    val imageBuffer = ImageIO.read(new File(resourceFilename))
+    val resourceStream = getClass.getResourceAsStream(resourcePath)
+    val imageBuffer = ImageIO.read(resourceStream)
     ImageUtils.spImageFromBufferedImage(imageBuffer)
   }
 
@@ -50,8 +50,8 @@ class ImageUtilsSuite extends FunSuite {
   test("Test spImage resize.") {
     def javaResize(imagePath: String, tgtWidth: Int, tgtHeight: Int): Row = {
       // Read BufferedImage directly from file
-      val resourceFilename = getClass.getResource(imagePath).getFile
-      val srcImg = ImageIO.read(new File(resourceFilename))
+      val resourceStream = getClass.getResourceAsStream(imagePath)
+      val srcImg = ImageIO.read(resourceStream)
       val tgtImg = new BufferedImage(tgtWidth, tgtHeight, srcImg.getType)
       // scaledImg is a java.awt.Image which supports drawing but not pixel lookup by index.
       val scaledImg = srcImg.getScaledInstance(tgtWidth, tgtHeight, Image.SCALE_AREA_AVERAGING)