add test

src/common/exception/Cargo.toml

@@ -30,7 +30,6 @@ serde = { workspace = true }
 serde_json = { workspace = true }
 tantivy = { workspace = true }
 thiserror = { workspace = true }
-tokio = "1.39.2"
 tonic = { workspace = true }
 
 [package.metadata.cargo-machete]

src/query/sql/Cargo.toml

@@ -73,8 +73,8 @@ serde = { workspace = true }
 sha2 = { workspace = true }
 simsearch = "0.2"
 time = "0.3.14"
-url = "2.3.1"
 tokio = "1.39.2"
+url = "2.3.1"
 
 [lints]
 workspace = true

src/query/sql/src/planner/optimizer/dynamic_sample/dynamic_sample.rs

@@ -17,7 +17,6 @@ use std::time::Duration;
 
 use databend_common_base::base::tokio::time::Instant;
 use databend_common_catalog::table_context::TableContext;
-use databend_common_exception::ErrorCode;
 use databend_common_exception::Result;
 
 use crate::optimizer::dynamic_sample::filter_selectivity_sample::filter_selectivity_sample;
@@ -80,9 +79,10 @@ pub async fn dynamic_sample(
             join_selectivity_sample(ctx, metadata, s_expr, sample_executor).await
         }
         RelOperator::Scan(_) => s_expr.plan().derive_stats(&RelExpr::with_s_expr(s_expr)),
-        _ => Err(ErrorCode::Unimplemented(format!(
-            "derive_cardinality_by_sample for {:?} is not supported yet",
-            s_expr.plan()
-        ))),
+        // Todo: add more operators here, and support more query patterns.
+        _ => {
+            let rel_expr = RelExpr::with_s_expr(s_expr);
+            rel_expr.derive_cardinality()
+        }
     }
 }

src/query/sql/src/planner/optimizer/dynamic_sample/filter_selectivity_sample.rs

@@ -60,17 +60,19 @@ pub async fn filter_selectivity_sample(
 
         // Calculate sample size (0.2% of total data)
         let sample_size = (num_rows as f64 * 0.002).ceil();
-
-        scan.sample = Some(Sample {
-            sample_level: SampleLevel::ROW,
-            sample_conf: SampleConfig::RowsNum(sample_size),
-        });
-
-        let new_child = SExpr::create_leaf(Arc::new(RelOperator::Scan(scan)));
-        let mut new_s_expr = s_expr.replace_children(vec![Arc::new(new_child)]);
-        let collect_statistics_optimizer =
-            CollectStatisticsOptimizer::new(ctx.clone(), metadata.clone());
-        new_s_expr = collect_statistics_optimizer.run(&new_s_expr).await?;
+        let mut new_s_expr = s_expr.clone();
+        // If the table is too small, we don't need to sample.
+        if sample_size >= 10.0 {
+            scan.sample = Some(Sample {
+                sample_level: SampleLevel::ROW,
+                sample_conf: SampleConfig::RowsNum(sample_size),
+            });
+            let new_child = SExpr::create_leaf(Arc::new(RelOperator::Scan(scan)));
+            new_s_expr = s_expr.replace_children(vec![Arc::new(new_child)]);
+            let collect_statistics_optimizer =
+                CollectStatisticsOptimizer::new(ctx.clone(), metadata.clone());
+            new_s_expr = collect_statistics_optimizer.run(&new_s_expr).await?;
+        }
 
         new_s_expr = SExpr::create_unary(
             Arc::new(create_count_aggregate(AggregateMode::Partial).into()),
@@ -91,12 +93,12 @@ pub async fn filter_selectivity_sample(
             if let Some(count) = block.get_last_column().as_number() {
                 if let Some(number_scalar) = count.index(0) {
                     // Compute and return selectivity
-                    let selectivity = number_scalar.to_f64().to_f64().unwrap() / sample_size as f64;
+                    let selectivity = number_scalar.to_f64().to_f64().unwrap() / sample_size;
                     let mut statistics = child_rel_expr.derive_cardinality()?.statistics.clone();
                     let mut sb = SelectivityEstimator::new(&mut statistics, HashSet::new());
                     sb.update_other_statistic_by_selectivity(selectivity);
                     let stat_info = Arc::new(StatInfo {
-                        cardinality: selectivity * num_rows as f64,
+                        cardinality: (selectivity * num_rows as f64).ceil(),
                         statistics,
                     });
                     *s_expr.stat_info.lock().unwrap() = Some(stat_info.clone());

src/query/sql/src/planner/optimizer/dynamic_sample/mod.rs

@@ -12,6 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+#[allow(clippy::module_inception)]
 mod dynamic_sample;
 mod filter_selectivity_sample;
 mod join_selectivity_sample;

src/query/sql/src/planner/optimizer/join/single_to_inner.rs

@@ -46,7 +46,7 @@ impl SingleToInnerOptimizer {
         let mut children = Vec::with_capacity(s_expr.arity());
         for child in s_expr.children() {
             let new_child = Self::single_to_inner(child)?;
-            if !new_child.eq(&child) {
+            if !new_child.eq(child) {
                 children_changed = true;
             }
             children.push(Arc::new(new_child));