Merge pull request ClickHouse#100110 from hanfei1991/hanfei/skip-estimator

Skip estimator if it is not a CNF

Author: hanfei1991

src/Interpreters/InterpreterSelectQuery.cpp

@@ -841,9 +841,13 @@ InterpreterSelectQuery::InterpreterSelectQuery(
                         parts_for_estimator = *parts;
                 }
 
-                /// Just attempting to read statistics files on disk can increase query latencies
-                /// First check the in-memory metadata if statistics are present at all
-                auto estimator = storage_snapshot->metadata->hasStatistics()
+                /// Just attempting to read statistics files on disk can increase query latencies.
+                /// First check the in-memory metadata if statistics are present at all.
+                /// Also, statistics are only used to reorder conditions, so skip if there is just one.
+                const auto * where_function = query.where()->as<ASTFunction>();
+                const bool has_multiple_conditions = where_function && where_function->name == "and";
+                const bool has_statistics = storage_snapshot->metadata->hasStatistics();
+                auto estimator = (has_statistics && has_multiple_conditions)
                                     ? storage->getConditionSelectivityEstimator(parts_for_estimator, queried_columns, context)
                                     : nullptr;
 

src/Processors/QueryPlan/Optimizations/optimizePrewhere.cpp

@@ -181,10 +181,15 @@ void optimizePrewhere(QueryPlan::Node & parent_node, const bool remove_unused_co
 
     const auto & queried_columns = source_step_with_filter->requiredSourceColumns();
 
+    /// Statistics are only used to reorder conditions, so skip if there is just one.
+    const auto & filter_root_node = filter_step->getExpression().findInOutputs(filter_step->getFilterColumnName());
+    const bool has_multiple_conditions = filter_root_node.type == ActionsDAG::ActionType::FUNCTION
+        && filter_root_node.function_base && filter_root_node.function_base->getName() == "and";
+
     MergeTreeWhereOptimizer where_optimizer{
         std::move(column_compressed_sizes),
         storage_snapshot,
-        read_from_merge_tree_step ? read_from_merge_tree_step->getConditionSelectivityEstimator(queried_columns) : nullptr,
+        (has_multiple_conditions && read_from_merge_tree_step) ? read_from_merge_tree_step->getConditionSelectivityEstimator(queried_columns) : nullptr,
         queried_columns,
         storage.supportedPrewhereColumns(),
         getLogger("QueryPlanOptimizePrewhere")};

tests/parallel_replicas_blacklist.txt

@@ -310,6 +310,7 @@
 00443_optimize_final_vertical_merge
 01660_join_or_any
 01801_s3_cluster
+03707_statistics_cache
 
     Produces different result:
 03658_joined_block_split_single_row_bytes

tests/queries/0_stateless/03707_statistics_cache.reference

@@ -1,3 +1,3 @@
 1
-1
+0
 1

tests/queries/0_stateless/03707_statistics_cache.sql

@@ -1,20 +1,14 @@
 -- Tags: no-fasttest
 
-SET allow_statistics = 1;
-SET use_statistics = 1;
-SET log_queries = 1;
-SET log_query_settings = 1;
-SET mutations_sync = 2;
-SET max_execution_time = 60;
 SET optimize_move_to_prewhere = 1, query_plan_optimize_prewhere = 1;
+SET enable_analyzer = 1;
+SET use_statistics = 1;
+
 -- Statistics loading happens inside optimizePrewhere(), which requires
 -- the Filter→ReadFromMergeTree pattern. query_plan_merge_expressions = 0
 -- inserts an intermediate Expression step that breaks this pattern.
 SET query_plan_merge_expressions = 1;
 
--- test rely on local execution, - force parallel replicas to genearate local plan
-SET parallel_replicas_local_plan=1;
-
 DROP TABLE IF EXISTS sc_core SYNC;
 
 CREATE TABLE sc_core