Fixed SELECTing from ReplacingMergeTree with do_not_merge_across_partitions_select_final

src/Processors/QueryPlan/ReadFromMergeTree.cpp

@@ -1038,7 +1038,8 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsFinal(
             /// MergeTreeReadPool and MergeTreeThreadSelectProcessor for parallel select.
             if (num_streams > 1 && settings.do_not_merge_across_partitions_select_final &&
                 std::distance(parts_to_merge_ranges[range_index], parts_to_merge_ranges[range_index + 1]) == 1 &&
-                parts_to_merge_ranges[range_index]->data_part->info.level > 0)
+                parts_to_merge_ranges[range_index]->data_part->info.level > 0
+                && data.merging_params.is_deleted_column.empty())
             {
                 sum_marks_in_lonely_parts += parts_to_merge_ranges[range_index]->getMarksCount();
                 lonely_parts.push_back(std::move(*parts_to_merge_ranges[range_index]));
@@ -1094,7 +1095,8 @@ Pipe ReadFromMergeTree::spreadMarkRangesAmongStreamsFinal(
         /// with level > 0 then we won't postprocess this part
         if (settings.do_not_merge_across_partitions_select_final &&
             std::distance(parts_to_merge_ranges[range_index], parts_to_merge_ranges[range_index + 1]) == 1 &&
-            parts_to_merge_ranges[range_index]->data_part->info.level > 0)
+            parts_to_merge_ranges[range_index]->data_part->info.level > 0 &&
+            data.merging_params.is_deleted_column.empty())
         {
             partition_pipes.emplace_back(Pipe::unitePipes(std::move(pipes)));
             continue;

tests/queries/0_stateless/02814_ReplacingMergeTree_fix_select_final_on_single_partition.reference

@@ -0,0 +1,29 @@
+--- Based on https://github.com/ClickHouse/ClickHouse/issues/49685
+--- Verify that ReplacingMergeTree properly handles _is_deleted:
+--- SELECT FINAL should take _is_deleted in consideration when there is only one partition.
+-- { echoOn }
+
+CREATE TABLE t
+(
+    `account_id` UInt64,
+    `_is_deleted` UInt8,
+    `_version` UInt64
+)
+ENGINE = ReplacingMergeTree(_version, _is_deleted)
+ORDER BY (account_id);
+INSERT INTO t SELECT number, 0, 1 FROM numbers(1e3);
+-- Mark first 100 rows as deleted.
+INSERT INTO t SELECT number, 1, 1 FROM numbers(1e2);
+-- Put everything is in one partition
+OPTIMIZE TABLE t FINAL;
+SELECT count() FROM t;
+1000
+SELECT count() FROM t FINAL;
+900
+-- Both should produce same number of rows.
+-- Previously `do_not_merge_across_partitions_select_final = 1` showed more rows, 
+-- as if no rows were deleted.
+SELECT count() FROM t FINAL SETTINGS do_not_merge_across_partitions_select_final = 1;
+900
+SELECT count() FROM t FINAL SETTINGS do_not_merge_across_partitions_select_final = 0;
+900

tests/queries/0_stateless/02814_ReplacingMergeTree_fix_select_final_on_single_partition.sql

@@ -0,0 +1,30 @@
+--- Based on https://github.com/ClickHouse/ClickHouse/issues/49685
+--- Verify that ReplacingMergeTree properly handles _is_deleted:
+--- SELECT FINAL should take _is_deleted in consideration when there is only one partition.
+-- { echoOn }
+
+CREATE TABLE t
+(
+    `account_id` UInt64,
+    `_is_deleted` UInt8,
+    `_version` UInt64
+)
+ENGINE = ReplacingMergeTree(_version, _is_deleted)
+ORDER BY (account_id);
+
+INSERT INTO t SELECT number, 0, 1 FROM numbers(1e3);
+-- Mark first 100 rows as deleted.
+INSERT INTO t SELECT number, 1, 1 FROM numbers(1e2);
+
+-- Put everything is in one partition
+OPTIMIZE TABLE t FINAL;
+
+SELECT count() FROM t;
+SELECT count() FROM t FINAL;
+
+-- Both should produce same number of rows.
+-- Previously `do_not_merge_across_partitions_select_final = 1` showed more rows, 
+-- as if no rows were deleted.
+SELECT count() FROM t FINAL SETTINGS do_not_merge_across_partitions_select_final = 1;
+SELECT count() FROM t FINAL SETTINGS do_not_merge_across_partitions_select_final = 0;
+