HBASE-29696 Fix incorrect TableInputFormat split boundaries (#8491)

Signed-off-by: Junegunn Choi <junegunn@apache.org>
Reviewed-by: Andrew Olson <930946+noslowerdna@users.noreply.github.com>

Author: Sigma-Ma

hbase-mapreduce/src/main/java/org/apache/hadoop/hbase/mapreduce/TableInputFormatBase.java

@@ -407,6 +407,9 @@ protected List<InputSplit> createNInputSplitsUniform(InputSplit split, int n)
 
     // Split Region into n chunks evenly
     byte[][] splitKeys = Bytes.split(startRow, endRow, true, n - 1);
+    // Restore the original boundaries after using synthetic ones to calculate the split keys.
+    splitKeys[0] = ts.getStartRow();
+    splitKeys[splitKeys.length - 1] = ts.getEndRow();
     for (int i = 0; i < splitKeys.length - 1; i++) {
       // In the table input format for single table we do not need to
       // store the scan object in table split because it can be memory intensive and redundant

hbase-mapreduce/src/test/java/org/apache/hadoop/hbase/mapreduce/TestTableInputFormatBase.java

@@ -17,6 +17,7 @@
  */
 package org.apache.hadoop.hbase.mapreduce;
 
+import static org.junit.jupiter.api.Assertions.assertArrayEquals;
 import static org.junit.jupiter.api.Assertions.assertEquals;
 import static org.mockito.ArgumentMatchers.any;
 import static org.mockito.ArgumentMatchers.anyBoolean;
@@ -27,6 +28,7 @@
 import java.net.Inet6Address;
 import java.net.InetAddress;
 import java.net.UnknownHostException;
+import java.util.List;
 import java.util.Map;
 import java.util.TreeMap;
 import java.util.concurrent.ExecutorService;
@@ -52,6 +54,7 @@
 import org.apache.hadoop.hbase.testclassification.SmallTests;
 import org.apache.hadoop.hbase.util.Bytes;
 import org.apache.hadoop.hbase.util.Pair;
+import org.apache.hadoop.mapreduce.InputSplit;
 import org.apache.hadoop.mapreduce.JobContext;
 import org.junit.jupiter.api.Tag;
 import org.junit.jupiter.api.Test;
@@ -62,6 +65,24 @@
 @Tag(SmallTests.TAG)
 public class TestTableInputFormatBase {
 
+  @Test
+  public void testCreateNInputSplitsUniformPreservesOriginalBoundaries() throws IOException {
+    TableInputFormat inputFormat = new TableInputFormat();
+    for (byte[] startRow : new byte[][] { HConstants.EMPTY_START_ROW, Bytes.toBytes("start") }) {
+      TableSplit split =
+        new TableSplit(TableName.valueOf("test"), startRow, HConstants.EMPTY_END_ROW, "localhost");
+
+      List<InputSplit> splits = inputFormat.createNInputSplitsUniform(split, 2);
+
+      assertEquals(2, splits.size());
+      TableSplit first = (TableSplit) splits.get(0);
+      TableSplit last = (TableSplit) splits.get(1);
+      assertArrayEquals(startRow, first.getStartRow());
+      assertArrayEquals(first.getEndRow(), last.getStartRow());
+      assertArrayEquals(HConstants.EMPTY_END_ROW, last.getEndRow());
+    }
+  }
+
   @Test
   public void testReuseRegionSizeCalculator() throws IOException {
     JobContext context = mock(JobContext.class);