Merge pull request #53000 from ClickHouse/vdimir/join_filter_set_sparse

ClickHouse · Aug 7, 2023 · 490dfc9 · 490dfc9
2 parents b1f0cb8 + e89bd29
commit 490dfc9
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 2 deletions.
diff --git a/src/Processors/Transforms/CreateSetAndFilterOnTheFlyTransform.cpp b/src/Processors/Transforms/CreateSetAndFilterOnTheFlyTransform.cpp
@@ -8,6 +8,7 @@
 #include <Common/formatReadable.h>
 #include <Common/logger_useful.h>
 #include <Columns/IColumn.h>
+#include <Columns/ColumnSparse.h>
 #include <Core/ColumnWithTypeAndName.h>
 #include <base/types.h>
 
@@ -35,7 +36,11 @@ Columns getColumnsByIndices(const Chunk & chunk, const std::vector<size_t> & ind
     Columns columns;
     const Columns & all_cols = chunk.getColumns();
     for (const auto & index : indices)
-        columns.push_back(all_cols.at(index));
+    {
+        auto col = recursiveRemoveSparse(all_cols.at(index));
+        columns.push_back(std::move(col));
+    }
+
     return columns;
 }
 
@@ -149,7 +154,7 @@ IProcessor::Status FilterBySetOnTheFlyTransform::prepare()
             LOG_DEBUG(log, "Finished {} by [{}]: consumed {} rows in total, {} rows bypassed, result {} rows, {:.2f}% filtered",
                 Poco::toLower(getDescription()), fmt::join(column_names, ", "),
                 stat.consumed_rows, stat.consumed_rows_before_set, stat.result_rows,
-                100 - 100.0 * stat.result_rows / stat.consumed_rows);
+                stat.consumed_rows > 0 ? (100 - 100.0 * stat.result_rows / stat.consumed_rows) : 0);
         }
         else
         {

diff --git a/tests/queries/0_stateless/02841_join_filter_set_sparse.reference b/tests/queries/0_stateless/02841_join_filter_set_sparse.reference
@@ -0,0 +1,2 @@
+3428033
+3428033
diff --git a/tests/queries/0_stateless/02841_join_filter_set_sparse.sql b/tests/queries/0_stateless/02841_join_filter_set_sparse.sql
@@ -0,0 +1,22 @@
+
+DROP TABLE IF EXISTS t1;
+DROP TABLE IF EXISTS t2;
+
+CREATE TABLE t1 (s String) ENGINE = MergeTree ORDER BY s
+SETTINGS ratio_of_defaults_for_sparse_serialization = 0.5;
+
+INSERT INTO t1 SELECT if (number % 13 = 0, toString(number), '') FROM numbers(2000);
+
+CREATE TABLE t2 (s String) ENGINE = MergeTree ORDER BY s
+SETTINGS ratio_of_defaults_for_sparse_serialization = 0.5;
+
+INSERT INTO t2 SELECT if (number % 14 = 0, toString(number), '') FROM numbers(2000);
+
+SELECT countIf(ignore(*) == 0) FROM t1 JOIN t2 ON t1.s = t2.s;
+
+SET join_algorithm = 'full_sorting_merge', max_rows_in_set_to_optimize_join = 100_000;
+
+SELECT countIf(ignore(*) == 0) FROM t1 JOIN t2 ON t1.s = t2.s;
+
+DROP TABLE t1;
+DROP TABLE t2;