[SPARK-8621] [SQL] support empty string as column name

improve the empty check in `parseAttributeName` so that we can allow empty string as column name.
Close #7117

Author: Wenchen Fan <cloud0fan@outlook.com>

Closes #7149 from cloud-fan/8621 and squashes the following commits:

efa9e3e [Wenchen Fan] support empty string

sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/plans/logical/LogicalPlan.scala

@@ -161,7 +161,7 @@ abstract class LogicalPlan extends QueryPlan[LogicalPlan] with Logging {
           if (tmp.nonEmpty) throw e
           inBacktick = true
         } else if (char == '.') {
-          if (tmp.isEmpty) throw e
+          if (name(i - 1) == '.' || i == name.length - 1) throw e
           nameParts += tmp.mkString
           tmp.clear()
         } else {
@@ -170,7 +170,7 @@ abstract class LogicalPlan extends QueryPlan[LogicalPlan] with Logging {
       }
       i += 1
     }
-    if (tmp.isEmpty || inBacktick) throw e
+    if (inBacktick) throw e
     nameParts += tmp.mkString
     nameParts.toSeq
   }

sql/core/src/test/scala/org/apache/spark/sql/DataFrameSuite.scala

@@ -730,4 +730,11 @@ class DataFrameSuite extends QueryTest {
     val res11 = ctx.range(-1).select("id")
     assert(res11.count == 0)
   }
+
+  test("SPARK-8621: support empty string column name") {
+    val df = Seq(Tuple1(1)).toDF("").as("t")
+    // We should allow empty string as column name
+    df.col("")
+    df.col("t.``")
+  }
 }