fix

databendlabs · Oct 15, 2024 · 2e2e5f6 · 2e2e5f6
1 parent 0c1473b
commit 2e2e5f6
Show file tree

Hide file tree

Showing 34 changed files with 52 additions and 202 deletions.
diff --git a/src/query/expression/src/converts/arrow2/from.rs b/src/query/expression/src/converts/arrow2/from.rs
@@ -31,7 +31,6 @@ use super::ARROW_EXT_TYPE_EMPTY_MAP;
 use super::ARROW_EXT_TYPE_GEOMETRY;
 use super::ARROW_EXT_TYPE_VARIANT;
 use crate::types::array::ArrayColumn;
-use crate::types::binary;
 use crate::types::binary::BinaryColumn;
 use crate::types::binary::BinaryColumnBuilder;
 use crate::types::decimal::DecimalColumn;
@@ -365,7 +364,7 @@ impl Column {
                         );
                     Column::Binary(binary_array_to_binary_column(arrow_col))
                 }
-                (DataType::Binary, ArrowDataType::FixedSizeBinary(size)) => {
+                (DataType::Binary, ArrowDataType::FixedSizeBinary(_)) => {
                     let arrow_col = arrow_col
                     .as_any()
                     .downcast_ref::<databend_common_arrow::arrow::array::FixedSizeBinaryArray>()

diff --git a/src/query/expression/src/filter/filter_executor.rs b/src/query/expression/src/filter/filter_executor.rs
@@ -125,7 +125,7 @@ impl FilterExecutor {
             if self.keep_order && self.has_or {
                 self.true_selection[0..result_count].sort();
             }
-            data_block.take(&self.true_selection[0..result_count], &mut None)
+            data_block.take(&self.true_selection[0..result_count])
         }
     }
 

diff --git a/src/query/expression/src/filter/select_value/select_column_scalar.rs b/src/query/expression/src/filter/select_value/select_column_scalar.rs
@@ -242,9 +242,9 @@ impl<'a> Selector<'a> {
                             for idx in 0u32..count as u32 {
                                 let ret = if NOT {
                                     validity.get_bit_unchecked(idx as usize)
-                                        && !searcher
+                                        && searcher
                                             .search(column.index_unchecked_bytes(idx as usize))
-                                            .is_some()
+                                            .is_none()
                                 } else {
                                     validity.get_bit_unchecked(idx as usize)
                                         && searcher
@@ -273,9 +273,9 @@ impl<'a> Selector<'a> {
                         if let LikePattern::SurroundByPercent(searcher) = like_pattern {
                             for idx in 0u32..count as u32 {
                                 let ret = if NOT {
-                                    !searcher
+                                    searcher
                                         .search(column.index_unchecked_bytes(idx as usize))
-                                        .is_some()
+                                        .is_none()
                                 } else {
                                     searcher
                                         .search(column.index_unchecked_bytes(idx as usize))

diff --git a/src/query/expression/src/kernels/concat.rs b/src/query/expression/src/kernels/concat.rs
@@ -23,7 +23,6 @@ use itertools::Itertools;
 
 use crate::copy_continuous_bits;
 use crate::kernels::take::BIT_MASK;
-use crate::kernels::utils::copy_advance_aligned;
 use crate::kernels::utils::set_vec_len_by_ptr;
 use crate::store_advance_aligned;
 use crate::types::array::ArrayColumnBuilder;

diff --git a/src/query/expression/src/kernels/group_by_hash/method_single_string.rs b/src/query/expression/src/kernels/group_by_hash/method_single_string.rs
@@ -13,7 +13,6 @@
 // limitations under the License.
 
 use databend_common_arrow::arrow::array::BinaryViewArray;
-use databend_common_arrow::arrow::buffer::Buffer;
 use databend_common_exception::Result;
 use databend_common_hashtable::hash_join_fast_string_hash;
 

diff --git a/src/query/expression/src/kernels/group_by_hash/utils.rs b/src/query/expression/src/kernels/group_by_hash/utils.rs
@@ -15,9 +15,7 @@
 use ethnum::i256;
 
 use crate::kernels::utils::copy_advance_aligned;
-use crate::kernels::utils::set_vec_len_by_ptr;
 use crate::kernels::utils::store_advance;
-use crate::kernels::utils::store_advance_aligned;
 use crate::types::binary::BinaryColumn;
 use crate::types::binary::BinaryColumnBuilder;
 use crate::types::decimal::DecimalColumn;

diff --git a/src/query/expression/src/kernels/scatter.rs b/src/query/expression/src/kernels/scatter.rs
@@ -33,24 +33,9 @@ impl DataBlock {
 
         let scatter_indices = Self::divide_indices_by_scatter_size(indices, scatter_size);
 
-        let has_string_column = self
-            .columns()
-            .iter()
-            .any(|col| col.data_type.is_string_column());
-        let mut string_items_buf = if has_string_column {
-            let max_num_rows = scatter_indices
-                .iter()
-                .map(|indices| indices.len())
-                .max()
-                .unwrap();
-            Some(vec![(0, 0); max_num_rows])
-        } else {
-            None
-        };
-
         let mut results = Vec::with_capacity(scatter_size);
         for indices in scatter_indices.iter().take(scatter_size) {
-            let block = self.take(indices, &mut string_items_buf)?;
+            let block = self.take(indices)?;
             results.push(block);
         }
 

diff --git a/src/query/expression/src/kernels/sort.rs b/src/query/expression/src/kernels/sort.rs
@@ -115,7 +115,7 @@ impl DataBlock {
         }
 
         let permutations = sort_compare.take_permutation();
-        DataBlock::take(block, &permutations, &mut None)
+        DataBlock::take(block, &permutations)
     }
 }
 

diff --git a/src/query/expression/src/kernels/take.rs b/src/query/expression/src/kernels/take.rs
@@ -19,8 +19,6 @@ use databend_common_arrow::arrow::bitmap::Bitmap;
 use databend_common_arrow::arrow::buffer::Buffer;
 use databend_common_exception::Result;
 
-use crate::kernels::utils::copy_advance_aligned;
-use crate::kernels::utils::set_vec_len_by_ptr;
 use crate::types::binary::BinaryColumn;
 use crate::types::nullable::NullableColumn;
 use crate::types::string::StringColumn;
@@ -34,19 +32,13 @@ use crate::Value;
 pub const BIT_MASK: [u8; 8] = [1, 2, 4, 8, 16, 32, 64, 128];
 
 impl DataBlock {
-    pub fn take<I>(
-        &self,
-        indices: &[I],
-        string_items_buf: &mut Option<Vec<(u64, usize)>>,
-    ) -> Result<Self>
-    where
-        I: databend_common_arrow::arrow::types::Index,
-    {
+    pub fn take<I>(&self, indices: &[I]) -> Result<Self>
+    where I: databend_common_arrow::arrow::types::Index {
         if indices.is_empty() {
             return Ok(self.slice(0..0));
         }
 
-        let mut taker = TakeVisitor::new(indices, string_items_buf);
+        let mut taker = TakeVisitor::new(indices);
 
         let after_columns = self
             .columns()
@@ -73,17 +65,15 @@ struct TakeVisitor<'a, I>
 where I: databend_common_arrow::arrow::types::Index
 {
     indices: &'a [I],
-    string_items_buf: &'a mut Option<Vec<(u64, usize)>>,
     result: Option<Value<AnyType>>,
 }
 
 impl<'a, I> TakeVisitor<'a, I>
 where I: databend_common_arrow::arrow::types::Index
 {
-    fn new(indices: &'a [I], string_items_buf: &'a mut Option<Vec<(u64, usize)>>) -> Self {
+    fn new(indices: &'a [I]) -> Self {
         Self {
             indices,
-            string_items_buf,
             result: None,
         }
     }

diff --git a/src/query/expression/src/kernels/take_chunks.rs b/src/query/expression/src/kernels/take_chunks.rs
@@ -22,8 +22,6 @@ use databend_common_hashtable::RowPtr;
 use itertools::Itertools;
 
 use crate::kernels::take::BIT_MASK;
-use crate::kernels::utils::copy_advance_aligned;
-use crate::kernels::utils::set_vec_len_by_ptr;
 use crate::types::array::ArrayColumnBuilder;
 use crate::types::binary::BinaryColumn;
 use crate::types::bitmap::BitmapType;
@@ -112,7 +110,6 @@ impl DataBlock {
         build_columns_data_type: &[DataType],
         indices: &[RowPtr],
         result_size: usize,
-        binary_items_buf: &mut Option<Vec<(u64, usize)>>,
     ) -> Self {
         let num_columns = build_columns.len();
         let result_columns = (0..num_columns)
@@ -123,7 +120,6 @@ impl DataBlock {
                     data_type.clone(),
                     indices,
                     result_size,
-                    binary_items_buf,
                 );
                 BlockEntry::new(data_type.clone(), Value::Column(column))
             })
@@ -632,7 +628,6 @@ impl Column {
         data_type: DataType,
         indices: &[RowPtr],
         result_size: usize,
-        binary_items_buf: &mut Option<Vec<(u64, usize)>>,
     ) -> Column {
         match &columns {
             ColumnVec::Null { .. } => Column::Null { len: result_size },
@@ -656,12 +651,12 @@ impl Column {
             ColumnVec::Boolean(columns) => {
                 Column::Boolean(Self::take_block_vec_boolean_types(columns, indices))
             }
-            ColumnVec::Binary(columns) => BinaryType::upcast_column(
-                Self::take_block_vec_binary_types(columns, indices, binary_items_buf.as_mut()),
-            ),
-            ColumnVec::String(columns) => StringType::upcast_column(
-                Self::take_block_vec_string_types(columns, indices, binary_items_buf.as_mut()),
-            ),
+            ColumnVec::Binary(columns) => {
+                BinaryType::upcast_column(Self::take_block_vec_binary_types(columns, indices))
+            }
+            ColumnVec::String(columns) => {
+                StringType::upcast_column(Self::take_block_vec_string_types(columns, indices))
+            }
             ColumnVec::Timestamp(columns) => {
                 let builder = Self::take_block_vec_primitive_types(columns, indices);
                 let ts = <NumberType<i64>>::upcast_column(<NumberType<i64>>::column_from_vec(
@@ -714,25 +709,23 @@ impl Column {
                     columns, builder, indices,
                 )
             }
-            ColumnVec::Bitmap(columns) => BitmapType::upcast_column(
-                Self::take_block_vec_binary_types(columns, indices, binary_items_buf.as_mut()),
-            ),
+            ColumnVec::Bitmap(columns) => {
+                BitmapType::upcast_column(Self::take_block_vec_binary_types(columns, indices))
+            }
             ColumnVec::Nullable(columns) => {
                 let inner_data_type = data_type.as_nullable().unwrap();
                 let inner_column = Self::take_column_vec_indices(
                     &columns.column,
                     *inner_data_type.clone(),
                     indices,
                     result_size,
-                    binary_items_buf,
                 );
 
                 let inner_bitmap = Self::take_column_vec_indices(
                     &columns.validity,
                     DataType::Boolean,
                     indices,
                     result_size,
-                    binary_items_buf,
                 );
 
                 NullableColumn::new_column(
@@ -751,25 +744,22 @@ impl Column {
                             ty.clone(),
                             indices,
                             result_size,
-                            binary_items_buf,
                         )
                     })
                     .collect();
 
                 Column::Tuple(fields)
             }
-            ColumnVec::Variant(columns) => VariantType::upcast_column(
-                Self::take_block_vec_binary_types(columns, indices, binary_items_buf.as_mut()),
-            ),
-            ColumnVec::Geometry(columns) => GeometryType::upcast_column(
-                Self::take_block_vec_binary_types(columns, indices, binary_items_buf.as_mut()),
-            ),
+            ColumnVec::Variant(columns) => {
+                VariantType::upcast_column(Self::take_block_vec_binary_types(columns, indices))
+            }
+            ColumnVec::Geometry(columns) => {
+                GeometryType::upcast_column(Self::take_block_vec_binary_types(columns, indices))
+            }
             ColumnVec::Geography(columns) => {
                 let columns = columns.iter().map(|x| x.0.clone()).collect::<Vec<_>>();
                 GeographyType::upcast_column(GeographyColumn(Self::take_block_vec_binary_types(
-                    &columns,
-                    indices,
-                    binary_items_buf.as_mut(),
+                    &columns, indices,
                 )))
             }
         }
@@ -785,11 +775,7 @@ impl Column {
         builder
     }
 
-    pub fn take_block_vec_binary_types(
-        col: &[BinaryColumn],
-        indices: &[RowPtr],
-        binary_items_buf: Option<&mut Vec<(u64, usize)>>,
-    ) -> BinaryColumn {
+    pub fn take_block_vec_binary_types(col: &[BinaryColumn], indices: &[RowPtr]) -> BinaryColumn {
         let mut builder = BinaryColumnBuilder::with_capacity(indices.len(), 0);
         for row_ptr in indices {
             unsafe {
@@ -802,11 +788,7 @@ impl Column {
         builder.build()
     }
 
-    pub fn take_block_vec_string_types(
-        cols: &[StringColumn],
-        indices: &[RowPtr],
-        binary_items_buf: Option<&mut Vec<(u64, usize)>>,
-    ) -> StringColumn {
+    pub fn take_block_vec_string_types(cols: &[StringColumn], indices: &[RowPtr]) -> StringColumn {
         let binary_cols = cols
             .iter()
             .map(|col| col.clone().into())
@@ -815,7 +797,6 @@ impl Column {
             StringColumn::from_binary_unchecked(Self::take_block_vec_binary_types(
                 &binary_cols,
                 indices,
-                binary_items_buf,
             ))
         }
     }

diff --git a/src/query/expression/src/kernels/take_compact.rs b/src/query/expression/src/kernels/take_compact.rs
@@ -18,7 +18,6 @@ use databend_common_exception::Result;
 
 use crate::kernels::utils::copy_advance_aligned;
 use crate::kernels::utils::set_vec_len_by_ptr;
-use crate::kernels::utils::store_advance_aligned;
 use crate::types::binary::BinaryColumn;
 use crate::types::nullable::NullableColumn;
 use crate::types::string::StringColumn;

diff --git a/src/query/expression/src/kernels/take_ranges.rs b/src/query/expression/src/kernels/take_ranges.rs
@@ -22,7 +22,6 @@ use databend_common_exception::Result;
 
 use crate::copy_continuous_bits;
 use crate::kernels::take::BIT_MASK;
-use crate::kernels::utils::copy_advance_aligned;
 use crate::kernels::utils::set_vec_len_by_ptr;
 use crate::kernels::utils::store_advance_aligned;
 use crate::types::binary::BinaryColumn;

diff --git a/src/query/expression/src/types/binary.rs b/src/query/expression/src/types/binary.rs
@@ -14,16 +14,11 @@
 
 use std::cmp::Ordering;
 use std::iter::once;
-use std::marker::PhantomData;
 use std::ops::Range;
 
-use databend_common_arrow::arrow::array::Array;
 use databend_common_arrow::arrow::array::BinaryViewArray;
 use databend_common_arrow::arrow::array::MutableBinaryViewArray;
-use databend_common_arrow::arrow::buffer::Buffer;
 use databend_common_arrow::arrow::trusted_len::TrustedLen;
-use databend_common_exception::ErrorCode;
-use databend_common_exception::Result;
 use serde::Deserialize;
 use serde::Serialize;
 
@@ -33,7 +28,6 @@ use crate::types::DataType;
 use crate::types::DecimalSize;
 use crate::types::GenericMap;
 use crate::types::ValueType;
-use crate::utils::arrow::buffer_into_mut;
 use crate::values::Column;
 use crate::values::Scalar;
 use crate::ColumnBuilder;

diff --git a/src/query/expression/src/types/geography.rs b/src/query/expression/src/types/geography.rs
@@ -15,13 +15,11 @@
 use std::cmp::Ordering;
 use std::fmt::Debug;
 use std::hash::Hash;
-use std::marker::PhantomData;
 use std::ops::Range;
 
 use borsh::BorshDeserialize;
 use borsh::BorshSerialize;
 use databend_common_arrow::arrow::trusted_len::TrustedLen;
-use databend_common_arrow::parquet::encoding::plain_byte_array::BinaryIter;
 use databend_common_exception::Result;
 use databend_common_io::geography::*;
 use databend_common_io::wkb::make_point;

diff --git a/src/query/expression/src/types/string.rs b/src/query/expression/src/types/string.rs
@@ -13,12 +13,9 @@
 // limitations under the License.
 
 use std::cmp::Ordering;
-use std::iter::once;
 use std::ops::Range;
 
-use databend_common_arrow::arrow::array::Array;
 use databend_common_arrow::arrow::array::BinaryViewArray;
-use databend_common_arrow::arrow::buffer::Buffer;
 use databend_common_arrow::arrow::trusted_len::TrustedLen;
 use databend_common_exception::ErrorCode;
 use databend_common_exception::Result;
@@ -34,7 +31,6 @@ use crate::types::DataType;
 use crate::types::DecimalSize;
 use crate::types::GenericMap;
 use crate::types::ValueType;
-use crate::utils::arrow::buffer_into_mut;
 use crate::values::Column;
 use crate::values::Scalar;
 use crate::ColumnBuilder;