fix: http handler cut block to pages using memory size after gc. (#19071)

* Datablock add fn memory_size_with_options

* fix: http handler cut block to pages using memory size after gc.

* Column::memory_size add arg gc.

* Column::memory_size add arg gc.

Author: youngsofun

src/common/column/src/binview/mod.rs

@@ -314,8 +314,12 @@ impl<T: ViewType + ?Sized> BinaryViewColumnGeneric<T> {
             .get_or_init(|| self.views.iter().map(|v| v.length as usize).sum::<usize>())
     }
 
-    pub fn memory_size(&self) -> usize {
-        self.total_buffer_len() + self.len() * 16
+    pub fn memory_size(&self, gc: bool) -> usize {
+        if gc {
+            self.total_bytes_len() + self.len() * 16
+        } else {
+            self.total_buffer_len() + self.len() * 16
+        }
     }
 
     fn total_unshared_buffer_len(&self) -> usize {

src/common/column/tests/it/binview/mod.rs

@@ -151,15 +151,15 @@ fn test_slice() {
     ];
 
     let array: Utf8ViewColumn = data.into_iter().collect();
-    assert_eq!(array.memory_size(), 156);
+    assert_eq!(array.memory_size(false), 156);
 
     let a3 = array.sliced(2, 3);
     assert_eq!(a3.into_iter().collect::<Vec<_>>(), vec![
         "databend",
         "yyyyyyyyyyyyyyyyyyyyy",
         "zzzzzzzzzzzzzzzzzzzzz",
     ]);
-    assert_eq!(a3.memory_size(), 108);
+    assert_eq!(a3.memory_size(false), 108);
 }
 
 #[test]

src/query/expression/src/block.rs

@@ -98,7 +98,14 @@ impl BlockEntry {
     pub fn memory_size(&self) -> usize {
         match self {
             BlockEntry::Const(scalar, _, _) => scalar.as_ref().memory_size(),
-            BlockEntry::Column(column) => column.memory_size(),
+            BlockEntry::Column(column) => column.memory_size(false),
+        }
+    }
+
+    pub fn memory_size_with_options(&self, gc: bool) -> usize {
+        match self {
+            BlockEntry::Const(scalar, _, _) => scalar.as_ref().memory_size(),
+            BlockEntry::Column(column) => column.memory_size(gc),
         }
     }
 

src/query/expression/src/types.rs

@@ -607,7 +607,8 @@ pub trait AccessType: Debug + Clone + PartialEq + Sized + 'static {
         std::mem::size_of::<Self::Scalar>()
     }
 
-    fn column_memory_size(col: &Self::Column) -> usize {
+    fn column_memory_size(col: &Self::Column, gc: bool) -> usize {
+        let _ = gc;
         Self::column_len(col) * std::mem::size_of::<Self::Scalar>()
     }
 

src/query/expression/src/types/any.rs

@@ -83,8 +83,8 @@ impl AccessType for AnyType {
         scalar.memory_size()
     }
 
-    fn column_memory_size(col: &Self::Column) -> usize {
-        col.memory_size()
+    fn column_memory_size(col: &Self::Column, gc: bool) -> usize {
+        col.memory_size(gc)
     }
 
     #[inline(always)]

src/query/expression/src/types/array.rs

@@ -101,11 +101,11 @@ impl<T: AccessType> AccessType for ArrayType<T> {
     }
 
     fn scalar_memory_size(scalar: &Self::ScalarRef<'_>) -> usize {
-        <T as AccessType>::column_memory_size(scalar)
+        <T as AccessType>::column_memory_size(scalar, false)
     }
 
-    fn column_memory_size(col: &Self::Column) -> usize {
-        col.memory_size()
+    fn column_memory_size(col: &Self::Column, gc: bool) -> usize {
+        col.memory_size(gc)
     }
 
     fn compare(a: T::Column, b: T::Column) -> Ordering {
@@ -269,8 +269,8 @@ impl<T: AccessType> ArrayColumn<T> {
         }
     }
 
-    pub fn memory_size(&self) -> usize {
-        T::column_memory_size(&self.underlying_column()) + self.offsets.len() * 8
+    pub fn memory_size(&self, gc: bool) -> usize {
+        T::column_memory_size(&self.underlying_column(), gc) + self.offsets.len() * 8
     }
 
     // Note: if the array column has been sliced, the number of values may not match the offsets.

src/query/expression/src/types/binary.rs

@@ -100,7 +100,7 @@ impl AccessType for BinaryType {
         scalar.len()
     }
 
-    fn column_memory_size(col: &Self::Column) -> usize {
+    fn column_memory_size(col: &Self::Column, _gc: bool) -> usize {
         col.data().len() + col.offsets().len() * 8
     }
 

src/query/expression/src/types/bitmap.rs

@@ -100,7 +100,7 @@ impl AccessType for BitmapType {
         scalar.len()
     }
 
-    fn column_memory_size(col: &Self::Column) -> usize {
+    fn column_memory_size(col: &Self::Column, _gc: bool) -> usize {
         col.memory_size()
     }
 

src/query/expression/src/types/compute_view.rs

@@ -111,7 +111,7 @@ where
         std::mem::size_of::<F>()
     }
 
-    fn column_memory_size(col: &Self::Column) -> usize {
+    fn column_memory_size(col: &Self::Column, _gc: bool) -> usize {
         F::column_len(col) * std::mem::size_of::<F>()
     }
 

src/query/expression/src/types/generic.rs

@@ -86,8 +86,8 @@ impl<const INDEX: usize> AccessType for GenericType<INDEX> {
         scalar.memory_size()
     }
 
-    fn column_memory_size(col: &Self::Column) -> usize {
-        col.memory_size()
+    fn column_memory_size(col: &Self::Column, gc: bool) -> usize {
+        col.memory_size(gc)
     }
 
     #[inline(always)]