Fix a bug that caused the RCF to be written incorrectly during multip…

…le rounds

deltacat/compute/compactor_v2/private/compaction_utils.py

@@ -585,7 +585,9 @@ def _process_merge_results(
             f"as or greater than params.num_rounds, which is {params.num_rounds}"
         )
         hb_id_to_entry_indices_range[str(mat_result.task_index)] = (
-            file_index,
+            hb_id_to_entry_indices_range.get(str(mat_result.task_index), [file_index])[
+                0
+            ],
             file_index + mat_result.pyarrow_write_result.files,
         )
 

deltacat/tests/compute/test_compact_partition_incremental.py

@@ -328,6 +328,16 @@ def _incremental_compaction_setup():
         **compaction_audit_obj
     )
 
+    # assert if RCF covers all files
+    if compactor_version != CompactorVersion.V1.value:
+        previous_end = None
+        for start, end in round_completion_info.hb_index_to_entry_range.values():
+            assert (previous_end is None and start == 0) or start == previous_end
+            previous_end = end
+        assert (
+            previous_end == round_completion_info.compacted_pyarrow_write_result.files
+        )
+
     tables = ds.download_delta(
         compacted_delta_locator, storage_type=StorageType.LOCAL, **ds_mock_kwargs
     )

deltacat/tests/compute/test_compact_partition_multiple_rounds.py

@@ -309,6 +309,16 @@ def test_compact_partition_rebase_multiple_rounds_same_source_and_destination(
             **compaction_audit_obj
         )
 
+        # assert if RCF covers all files
+        # multiple rounds feature is only supported in V2 compactor
+        previous_end = None
+        for start, end in round_completion_info.hb_index_to_entry_range.values():
+            assert (previous_end is None and start == 0) or start == previous_end
+            previous_end = end
+        assert (
+            previous_end == round_completion_info.compacted_pyarrow_write_result.files
+        )
+
         # Assert not in-place compacted
         assert (
             execute_compaction_result_spy.call_args.args[-1] is False

deltacat/tests/compute/test_compact_partition_rebase.py

@@ -299,6 +299,17 @@ def test_compact_partition_rebase_same_source_and_destination(
             round_completion_info.compaction_audit_url
         )
 
+        # assert if RCF covers all files
+        if compactor_version != CompactorVersion.V1.value:
+            previous_end = None
+            for start, end in round_completion_info.hb_index_to_entry_range.values():
+                assert (previous_end is None and start == 0) or start == previous_end
+                previous_end = end
+            assert (
+                previous_end
+                == round_completion_info.compacted_pyarrow_write_result.files
+            )
+
         compaction_audit_obj: Dict[str, Any] = read_s3_contents(
             s3_resource, audit_bucket, audit_key
         )

deltacat/tests/compute/test_compact_partition_rebase_then_incremental.py

@@ -355,6 +355,16 @@ def test_compact_partition_rebase_then_incremental(
     compacted_delta_locator_incremental: DeltaLocator = (
         round_completion_info.compacted_delta_locator
     )
+    # assert if RCF covers all files
+    if compactor_version != CompactorVersion.V1.value:
+        previous_end = None
+        for start, end in round_completion_info.hb_index_to_entry_range.values():
+            assert (previous_end is None and start == 0) or start == previous_end
+            previous_end = end
+        assert (
+            previous_end == round_completion_info.compacted_pyarrow_write_result.files
+        )
+
     audit_bucket, audit_key = round_completion_info.compaction_audit_url.replace(
         "s3://", ""
     ).split("/", 1)