[SPARK-8301] [SQL] Improve UTF8String substring/startsWith/endsWith/contains performance

Jira: https://issues.apache.org/jira/browse/SPARK-8301

Added the private method startsWith(prefix, offset) to implement startsWith, endsWith and contains without copying the array

I hope that the component SQL is still correct. I copied it from the Jira ticket.

Author: Tarek Auel <tarek.auel@googlemail.com>
Author: Tarek Auel <tarek.auel@gmail.com>

Closes #6804 from tarekauel/SPARK-8301 and squashes the following commits:

f5d6b9a [Tarek Auel] fixed parentheses and annotation
6d7b068 [Tarek Auel] [SPARK-8301] removed null checks
9ca0473 [Tarek Auel] [SPARK-8301] removed null checks
1c327eb [Tarek Auel] [SPARK-8301] removed new
9f17cc8 [Tarek Auel] [SPARK-8301] fixed conversion byte to string in codegen
3a0040f [Tarek Auel] [SPARK-8301] changed call of UTF8String.set to UTF8String.from
e4530d2 [Tarek Auel] [SPARK-8301] changed call of UTF8String.set to UTF8String.from
a5f853a [Tarek Auel] [SPARK-8301] changed visibility of set to protected. Changed annotation of bytes from Nullable to Nonnull
d2fb05f [Tarek Auel] [SPARK-8301] added additional null checks
79cb55b [Tarek Auel] [SPARK-8301] null check. Added test cases for null check.
b17909e [Tarek Auel] [SPARK-8301] removed unnecessary copying of UTF8String. Added a private function startsWith(prefix, offset) to implement the check for startsWith, endsWith and contains.

2 files changed, 4 insertions, 6 deletions

diff --git a/sql/catalyst/src/main/java/org/apache/spark/sql/catalyst/expressions/UnsafeRow.java b/sql/catalyst/src/main/java/org/apache/spark/sql/catalyst/expressions/UnsafeRow.java
index c4b7f8490a..ed04d2e50e 100644
--- a/sql/catalyst/src/main/java/org/apache/spark/sql/catalyst/expressions/UnsafeRow.java
+++ b/sql/catalyst/src/main/java/org/apache/spark/sql/catalyst/expressions/UnsafeRow.java
@@ -312,7 +312,6 @@ public final class UnsafeRow extends BaseMutableRow {
 
   public UTF8String getUTF8String(int i) {
     assertIndexIsValid(i);
-    final UTF8String str = new UTF8String();
     final long offsetToStringSize = getLong(i);
     final int stringSizeInBytes =
       (int) PlatformDependent.UNSAFE.getLong(baseObject, baseOffset + offsetToStringSize);
@@ -324,8 +323,7 @@ public final class UnsafeRow extends BaseMutableRow {
       PlatformDependent.BYTE_ARRAY_OFFSET,
       stringSizeInBytes
     );
-    str.set(strBytes);
-    return str;
+    return UTF8String.fromBytes(strBytes);
   }
 
   @Override
diff --git a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/Cast.scala b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/Cast.scala
index b20086bcc4..ad920f2878 100644
--- a/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/Cast.scala
+++ b/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/expressions/Cast.scala
@@ -438,17 +438,17 @@ case class Cast(child: Expression, dataType: DataType) extends UnaryExpression w
 
       case (BinaryType, StringType) =>
         defineCodeGen (ctx, ev, c =>
-          s"new ${ctx.stringType}().set($c)")
+          s"${ctx.stringType}.fromBytes($c)")
       case (DateType, StringType) =>
         defineCodeGen(ctx, ev, c =>
-          s"""new ${ctx.stringType}().set(
+          s"""${ctx.stringType}.fromString(
                 org.apache.spark.sql.catalyst.util.DateUtils.toString($c))""")
       // Special handling required for timestamps in hive test cases since the toString function
       // does not match the expected output.
       case (TimestampType, StringType) =>
         super.genCode(ctx, ev)
       case (_, StringType) =>
-        defineCodeGen(ctx, ev, c => s"new ${ctx.stringType}().set(String.valueOf($c))")
+        defineCodeGen(ctx, ev, c => s"${ctx.stringType}.fromString(String.valueOf($c))")
 
       // fallback for DecimalType, this must be before other numeric types
       case (_, dt: DecimalType) =>