x

Author: forsaken628

src/common/statistics/src/histogram.rs

@@ -157,10 +157,10 @@ impl Histogram {
 
     pub fn bucket_iter(&self) -> HistogramBucketIter<'_> {
         match self {
-            Self::Int(histogram) => HistogramBucketIter::Int(histogram.buckets_iter()),
-            Self::UInt(histogram) => HistogramBucketIter::UInt(histogram.buckets_iter()),
-            Self::Float(histogram) => HistogramBucketIter::Float(histogram.buckets_iter()),
-            Self::Bytes(histogram) => HistogramBucketIter::Bytes(histogram.buckets_iter()),
+            Self::Int(histogram) => HistogramBucketIter::Int(histogram.buckets.iter()),
+            Self::UInt(histogram) => HistogramBucketIter::UInt(histogram.buckets.iter()),
+            Self::Float(histogram) => HistogramBucketIter::Float(histogram.buckets.iter()),
+            Self::Bytes(histogram) => HistogramBucketIter::Bytes(histogram.buckets.iter()),
         }
     }
 

src/common/statistics/src/typed_histogram.rs

@@ -93,10 +93,6 @@ impl<T> TypedHistogram<T> {
         self.buckets.iter().map(|bucket| bucket.num_distinct).sum()
     }
 
-    pub fn buckets_iter(&self) -> std::slice::Iter<'_, TypedHistogramBucket<T>> {
-        self.buckets.iter()
-    }
-
     pub fn scale_counts(&mut self, selectivity: f64) {
         for bucket in &mut self.buckets {
             bucket.num_values *= selectivity;
@@ -846,7 +842,7 @@ mod tests {
     fn test_typed_histogram_builder_preserves_max_bound_when_range_not_divisible() {
         let histogram =
             TypedHistogramBuilder::from_ndv(738, 738, Some((0_u64, 737_u64)), 100).unwrap();
-        let buckets = histogram.buckets_iter().collect::<Vec<_>>();
+        let buckets = histogram.buckets.iter().collect::<Vec<_>>();
 
         assert_eq!(buckets.last().unwrap().upper_bound(), &737);
     }
@@ -855,7 +851,8 @@ mod tests {
     fn test_typed_histogram_builder_partitions_discrete_synthetic_bounds() {
         let histogram = TypedHistogramBuilder::from_ndv(10, 10, Some((0_u64, 9_u64)), 10).unwrap();
         let bounds = histogram
-            .buckets_iter()
+            .buckets
+            .iter()
             .map(|bucket| (*bucket.lower_bound(), *bucket.upper_bound()))
             .collect::<Vec<_>>();
 

src/common/statistics/src/typed_histogram/join.rs

@@ -47,8 +47,8 @@ impl<T: Value> TypedHistogram<T> {
     pub fn estimate_join(&self, other: &TypedHistogram<T>) -> JoinEstimation {
         let mut estimation = JoinEstimation::zero();
 
-        for left_bucket in self.buckets_iter() {
-            for right_bucket in other.buckets_iter() {
+        for left_bucket in &self.buckets {
+            for right_bucket in &other.buckets {
                 estimation = estimation.add(left_bucket.estimate_join_contribution(right_bucket));
             }
         }

src/query/expression/src/function/stat_distribution.rs

@@ -18,6 +18,7 @@ pub use databend_common_statistics::Ndv;
 use crate::Domain;
 use crate::Scalar;
 use crate::types::DataType;
+use crate::types::boolean::BooleanDomain;
 use crate::types::nullable::NullableDomain;
 
 #[derive(Debug, Clone)]
@@ -105,6 +106,18 @@ impl<'a> ArgStat<'a> {
 }
 
 impl ReturnStat {
+    pub fn boolean(true_count: StatEstimate) -> Self {
+        Self {
+            domain: Domain::Boolean(BooleanDomain {
+                has_true: true,
+                has_false: true,
+            }),
+            ndv: Ndv::Stat(2.0),
+            null_count: 0,
+            distribution: OwnedDistribution::Boolean(BooleanDistribution { true_count }),
+        }
+    }
+
     pub fn histogram(&self) -> Option<&Histogram> {
         self.distribution.as_histogram()
     }
@@ -297,9 +310,3 @@ impl StatEstimate {
 pub struct BooleanDistribution {
     pub true_count: StatEstimate,
 }
-
-impl BooleanDistribution {
-    pub fn new(true_count: StatEstimate) -> Self {
-        Self { true_count }
-    }
-}