supporting fields with NULL types as sql NullType

src/main/scala/com/databricks/spark/avro/SchemaConverters.scala

@@ -52,6 +52,7 @@ object SchemaConverters {
       case LONG => SchemaType(LongType, nullable = false)
       case FIXED => SchemaType(BinaryType, nullable = false)
       case ENUM => SchemaType(StringType, nullable = false)
+      case NULL => SchemaType(NullType, nullable = false)
 
       case RECORD =>
         val fields = avroSchema.getFields.asScala.map { f =>
@@ -150,7 +151,7 @@ object SchemaConverters {
           (item: AnyRef) => if (item == null) null else item.toString
         // Byte arrays are reused by avro, so we have to make a copy of them.
         case (IntegerType, INT) | (BooleanType, BOOLEAN) | (DoubleType, DOUBLE) |
-             (FloatType, FLOAT) | (LongType, LONG) =>
+             (FloatType, FLOAT) | (LongType, LONG) | (NullType, NULL) | (NullType, STRING) =>
           identity
         case (BinaryType, FIXED) =>
           (item: AnyRef) =>

src/test/scala/com/databricks/spark/avro/AvroSuite.scala

@@ -109,9 +109,8 @@ class AvroSuite extends FunSuite with BeforeAndAfterAll {
       dataFileWriter.flush()
       dataFileWriter.close()
 
-      intercept[IncompatibleSchemaException] {
-        spark.read.avro(s"$dir.avro")
-      }
+      val df = spark.read.avro(s"$dir.avro")
+      assert(df.count() == 1)
     }
   }