x

Author: forsaken628

src/query/sql/test-support/src/optimizer/mod.rs

@@ -112,6 +112,27 @@ pub struct HistogramStats {
     pub avg_spacing: Option<f64>,
 }
 
+impl HistogramStats {
+    pub fn to_histogram(&self) -> Result<Histogram> {
+        let buckets = self
+            .buckets
+            .iter()
+            .map(|bucket| {
+                HistogramBucket::try_from_bounds(
+                    bucket.lower_bound.clone(),
+                    bucket.upper_bound.clone(),
+                    bucket.num_values,
+                    bucket.num_distinct,
+                )
+                .map_err(|err| ErrorCode::Internal(format!("invalid histogram bucket: {err}")))
+            })
+            .collect::<Result<Vec<_>>>()?;
+
+        Histogram::try_from_buckets(self.accuracy, buckets, self.avg_spacing)
+            .map_err(|err| ErrorCode::Internal(format!("invalid histogram: {err}")))
+    }
+}
+
 #[derive(Debug, Serialize, Deserialize, Clone)]
 pub struct HistogramBucketStats {
     pub lower_bound: Datum,
@@ -556,7 +577,7 @@ impl StatsApplier<'_> {
             {
                 let full_name = format!("{table_name}.{column_name}");
                 if let Some(stats) = self.histogram_stats.get(&full_name) {
-                    result.insert(*column_index, Some(histogram_from_stats(stats)?));
+                    result.insert(*column_index, Some(stats.to_histogram()?));
                 }
             }
         }
@@ -565,25 +586,6 @@ impl StatsApplier<'_> {
     }
 }
 
-pub(crate) fn histogram_from_stats(stats: &HistogramStats) -> Result<Histogram> {
-    let buckets = stats
-        .buckets
-        .iter()
-        .map(|bucket| {
-            HistogramBucket::try_from_bounds(
-                bucket.lower_bound.clone(),
-                bucket.upper_bound.clone(),
-                bucket.num_values,
-                bucket.num_distinct,
-            )
-            .map_err(|err| ErrorCode::Internal(format!("invalid histogram bucket: {err}")))
-        })
-        .collect::<Result<Vec<_>>>()?;
-
-    Histogram::try_from_buckets(stats.accuracy, buckets, stats.avg_spacing)
-        .map_err(|err| ErrorCode::Internal(format!("invalid histogram: {err}")))
-}
-
 fn write_result<F>(mint: &mut Mint, name: &str, f: F) -> Result<()>
 where F: FnOnce(&mut dyn Write) -> Result<()> {
     let mut file = mint.new_goldenfile(name).unwrap();

src/query/sql/test-support/src/optimizer/replay.rs

@@ -26,7 +26,6 @@ use databend_common_statistics::Histogram;
 use serde::Deserialize;
 
 use super::HistogramStats;
-use super::histogram_from_stats;
 
 #[derive(Debug, Clone, Deserialize)]
 pub struct ReplayInput {
@@ -159,7 +158,8 @@ impl ReplayInput {
                             return None;
                         }
                         column.histogram.as_ref().map(|histogram| {
-                            histogram_from_stats(histogram)
+                            histogram
+                                .to_histogram()
                                 .map(|histogram| (column.column_name.clone(), histogram))
                         })
                     })

src/query/sql/tests/it/optimizer/union_all.rs

@@ -30,9 +30,8 @@ use databend_common_sql::plans::Operator;
 use databend_common_sql::plans::Plan;
 use databend_common_sql::plans::RelOperator;
 use databend_common_sql_test_support::ColumnStats;
-use databend_common_statistics::Datum;
+use databend_common_sql_test_support::HistogramStats;
 use databend_common_statistics::Histogram;
-use databend_common_statistics::HistogramBuilder;
 
 use crate::framework::LiteTableContext;
 use crate::framework::golden::open_golden_file;
@@ -304,6 +303,12 @@ fn column_stat(json: &str) -> Result<BasicColumnStatistics> {
     Ok(stats.to_basic_column_statistics())
 }
 
+fn histogram_stat(json: &str) -> Result<Histogram> {
+    let stats: HistogramStats = serde_json::from_str(json)
+        .map_err(|err| ErrorCode::Internal(format!("invalid histogram statistics JSON: {err}")))?;
+    stats.to_histogram()
+}
+
 fn cases() -> Result<Vec<UnionCase>> {
     Ok(vec![
         UnionCase {
@@ -408,11 +413,15 @@ fn cases() -> Result<Vec<UnionCase>> {
                     )]),
                     histograms: HashMap::from([(
                         "k".to_string(),
-                        HistogramBuilder::from_ndv(
-                            10,
-                            100,
-                            Some((Datum::Int(0), Datum::Int(9))),
-                            2,
+                        histogram_stat(
+                            r#"{
+                                "accuracy": false,
+                                "buckets": [
+                                    {"lower_bound": {"Int": 0}, "upper_bound": {"Int": 4}, "num_values": 50.0, "num_distinct": 5.0},
+                                    {"lower_bound": {"Int": 5}, "upper_bound": {"Int": 9}, "num_values": 50.0, "num_distinct": 5.0}
+                                ],
+                                "avg_spacing": 4.5
+                            }"#,
                         )?,
                     )]),
                 },
@@ -425,11 +434,15 @@ fn cases() -> Result<Vec<UnionCase>> {
                     )]),
                     histograms: HashMap::from([(
                         "k".to_string(),
-                        HistogramBuilder::from_ndv(
-                            10,
-                            100,
-                            Some((Datum::Int(5), Datum::Int(14))),
-                            2,
+                        histogram_stat(
+                            r#"{
+                                "accuracy": false,
+                                "buckets": [
+                                    {"lower_bound": {"Int": 5}, "upper_bound": {"Int": 9}, "num_values": 50.0, "num_distinct": 5.0},
+                                    {"lower_bound": {"Int": 10}, "upper_bound": {"Int": 14}, "num_values": 50.0, "num_distinct": 5.0}
+                                ],
+                                "avg_spacing": 4.5
+                            }"#,
                         )?,
                     )]),
                 },