Provide kafka unbounded reader to checkpoint mark when offset based deduplication is supported. (#34669)

Author: tomstepp

sdks/java/io/kafka/src/main/java/org/apache/beam/sdk/io/kafka/KafkaCheckpointMark.java

@@ -35,6 +35,7 @@
  */
 @DefaultCoder(AvroCoder.class)
 public class KafkaCheckpointMark implements UnboundedSource.CheckpointMark {
+  private static final long OFFSET_DEDUP_PARTITIONS_PER_SPLIT = 1;
 
   private List<PartitionMark> partitions;
 
@@ -44,8 +45,6 @@ public class KafkaCheckpointMark implements UnboundedSource.CheckpointMark {
   @SuppressWarnings("initialization") // Avro will set the fields by breaking abstraction
   private KafkaCheckpointMark() {} // for Avro
 
-  private static final long OFFSET_DEDUP_PARTITIONS_PER_SPLIT = 1;
-
   public KafkaCheckpointMark(
       List<PartitionMark> partitions, Optional<KafkaUnboundedReader<?, ?>> reader) {
     this.partitions = partitions;

sdks/java/io/kafka/src/main/java/org/apache/beam/sdk/io/kafka/KafkaUnboundedReader.java

@@ -268,7 +268,9 @@ public CheckpointMark getCheckpointMark() {
                         p.nextOffset,
                         p.lastWatermark.getMillis()))
             .collect(Collectors.toList()),
-        source.getSpec().isCommitOffsetsInFinalizeEnabled() ? Optional.of(this) : Optional.empty());
+        source.getSpec().isCommitOffsetsInFinalizeEnabled() || offsetBasedDeduplicationSupported()
+            ? Optional.of(this)
+            : Optional.empty());
   }
 
   @Override