Github user fhueske commented on a diff in the pull request:
https://github.com/apache/flink/pull/1465#discussion_r50721087
--- Diff:
flink-libraries/flink-table/src/main/scala/org/apache/flink/api/table/typeinfo/NullAwareComparator.scala
---
@@ -0,0 +1,231 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.flink.api.table.typeinfo
+
+import org.apache.flink.api.common.typeutils.{CompositeTypeComparator,
TypeComparator}
+import org.apache.flink.core.memory.{DataInputView, DataOutputView,
MemorySegment}
+
+import scala.collection.JavaConversions._
+import scala.collection.mutable.ArrayBuffer
+
+/**
+ * Null-aware comparator that wraps a comparator which does not support
null references.
+ *
+ * NOTE: This class assumes to be used within a composite type comparator
(such
+ * as [[RowComparator]]) that handles serialized comparison and calls
+ * [[NullAwareComparator#extractKeys()]] or
[[NullAwareComparator#extractKeysFromNull()]]
+ * respectively.
+ */
+class NullAwareComparator[T](
+ val wrappedComparator: TypeComparator[T],
+ val order: Boolean)
+ extends TypeComparator[T] {
+
+ // stores the null for reference comparison
+ private var nullReference = false
+
+ override def hash(record: T): Int = {
+ if (record != null) {
+ wrappedComparator.hash(record)
+ }
+ else {
+ 0
+ }
+ }
+
+ override def getNormalizeKeyLen: Int = {
+ val len = wrappedComparator.getNormalizeKeyLen
+ if (len == Integer.MAX_VALUE) {
+ Integer.MAX_VALUE
+ }
+ else {
+ len + 1 // add one for a null byte
+ }
+ }
+
+ override def putNormalizedKey(
+ record: T,
+ target: MemorySegment,
+ offset: Int,
+ numBytes: Int)
+ : Unit = {
+ if (numBytes > 0) {
+ // write a null byte with padding
+ if (record == null) {
+ target.putBoolean(offset, false)
+ // write padding
+ var j = 0
+ while (j < numBytes - 1) {
+ target.put(offset + 1 + j, 0.toByte)
+ j += 1
+ }
+ }
+ // write a non-null byte with key
+ else {
+ target.putBoolean(offset, true)
+ // write key
+ wrappedComparator.putNormalizedKey(record, target, offset + 1,
numBytes - 1)
+ }
+ }
+ }
+
+ override def invertNormalizedKey(): Boolean =
wrappedComparator.invertNormalizedKey()
+
+ override def supportsSerializationWithKeyNormalization(): Boolean = false
+
+ override def writeWithKeyNormalization(record: T, target:
DataOutputView): Unit =
+ throw new UnsupportedOperationException("Record serialization with
leading normalized keys" +
+ " not supported.")
+
+ override def readWithKeyDenormalization(reuse: T, source:
DataInputView): T =
+ throw new UnsupportedOperationException("Record deserialization with
leading normalized keys" +
+ " not supported.")
+
+ override def isNormalizedKeyPrefixOnly(keyBytes: Int): Boolean =
+ wrappedComparator.isNormalizedKeyPrefixOnly(keyBytes - 1)
+
+ override def setReference(toCompare: T): Unit = {
+ if (toCompare == null) {
+ nullReference = true
+ }
+ else {
+ nullReference = false
+ wrappedComparator.setReference(toCompare)
+ }
+ }
+
+ override def compare(first: T, second: T): Int = {
+ // both values are null -> equality
+ if (first == null && second == null) {
+ 0
+ }
+ // first value is null -> inequality
+ // but order is considered
+ else if (first == null) {
+ if (order) -1 else 1
+ }
+ // second value is null -> inequality
+ // but order is considered
+ else if (second == null) {
+ if (order) 1 else -1
+ }
+ // no null values
+ else {
+ wrappedComparator.compare(first, second)
+ }
+ }
+
+ override def compareToReference(referencedComparator:
TypeComparator[T]): Int = {
+ val otherComparator =
referencedComparator.asInstanceOf[NullAwareComparator[T]]
+ val otherNullReference = otherComparator.nullReference
+ // both values are null -> equality
+ if (nullReference && otherNullReference) {
+ 0
+ }
+ // first value is null -> inequality
+ // but order is considered
+ else if (nullReference) {
+ if (order) 1 else -1
+ }
+ // second value is null -> inequality
+ // but order is considered
+ else if (otherNullReference) {
+ if (order) -1 else 1
+ }
+ // no null values
+ else {
+
wrappedComparator.compareToReference(otherComparator.wrappedComparator)
+ }
+ }
+
+ override def supportsNormalizedKey(): Boolean =
wrappedComparator.supportsNormalizedKey()
+
+ override def equalToReference(candidate: T): Boolean = {
+ // both values are null
+ if (candidate == null && nullReference) {
+ true
+ }
+ // one value is null
+ else if (candidate == null || nullReference) {
+ false
+ }
+ // no null value
+ else {
+ wrappedComparator.equalToReference(candidate)
+ }
+ }
+
+ override def duplicate(): TypeComparator[T] = {
+ new NullAwareComparator[T](wrappedComparator.duplicate(), order)
+ }
+
+ /**
+ * This method does not support records that are null. Use
[[extractKeysFromNull()]] instead.
--- End diff --
We can get the number of fields to extract by calling
`getFlatComparators()` in the constructor of the `NullAwareComparator` and use
this number to fill the target array with nulls if `record` is `null`.
Then we do not need the special `extractKeysFromNull` method and avoid some
overhead.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
