Merge pull request #1606 from zhyass/refactor_datavalues

Remove DataType::Utf8

common/arrow/Cargo.toml

@@ -15,8 +15,8 @@ simd = ["arrow/simd"]
 # Workspace dependencies
 
 # Github dependencies
-arrow = { package = "arrow2", git="https://github.com/datafuse-extras/arrow2", rev = "7765067" }
-arrow-flight = { git="https://github.com/datafuse-extras/arrow2", rev = "7765067" }
+arrow = { package = "arrow2", git="https://github.com/zhyass/arrow2", rev = "23682f0" }
+arrow-flight = { git="https://github.com/zhyass/arrow2", rev = "23682f0" }
 parquet = {package = "parquet2", git = "https://github.com/datafuse-extras/parquet2", rev = "d28330f"}
 # Crates.io dependencies
 

common/datablocks/src/kernels/data_block_concat_test.rs

@@ -23,7 +23,7 @@ use crate::*;
 fn test_data_block_concat() -> Result<()> {
     let schema = DataSchemaRefExt::create(vec![
         DataField::new("a", DataType::Int64, false),
-        DataField::new("b", DataType::Utf8, false),
+        DataField::new("b", DataType::String, false),
     ]);
 
     let blocks = vec![

common/datablocks/src/kernels/data_block_group_by_hash.rs

@@ -141,7 +141,7 @@ impl HashMethodKind {
     }
     pub fn data_type(&self) -> DataType {
         match self {
-            HashMethodKind::Serializer(_) => DataType::Binary,
+            HashMethodKind::Serializer(_) => DataType::String,
             HashMethodKind::KeysU8(_) => DataType::UInt8,
             HashMethodKind::KeysU16(_) => DataType::UInt16,
             HashMethodKind::KeysU32(_) => DataType::UInt32,
@@ -155,7 +155,7 @@ pub struct HashMethodSerializer {}
 
 impl HashMethodSerializer {
     #[inline]
-    pub fn get_key(&self, array: &DFBinaryArray, row: usize) -> Vec<u8> {
+    pub fn get_key(&self, array: &DFStringArray, row: usize) -> Vec<u8> {
         let v = array.inner().value(row);
         v.to_owned()
     }

common/datablocks/src/kernels/data_block_group_by_hash_test.rs

@@ -23,7 +23,7 @@ fn test_data_block_group_by_hash() -> Result<()> {
         DataField::new("a", DataType::Int8, false),
         DataField::new("b", DataType::Int8, false),
         DataField::new("c", DataType::Int8, false),
-        DataField::new("x", DataType::Utf8, false),
+        DataField::new("x", DataType::String, false),
     ]);
 
     let block = DataBlock::create_by_array(schema.clone(), vec![

common/datablocks/src/kernels/data_block_group_by_test.rs

@@ -21,7 +21,7 @@ use crate::*;
 fn test_data_block_group_by() -> Result<()> {
     let schema = DataSchemaRefExt::create(vec![
         DataField::new("a", DataType::Int8, false),
-        DataField::new("b", DataType::Utf8, false),
+        DataField::new("b", DataType::String, false),
     ]);
 
     let block = DataBlock::create_by_array(schema.clone(), vec![

common/datablocks/src/kernels/data_block_sort_test.rs

@@ -21,7 +21,7 @@ use crate::*;
 fn test_data_block_sort() -> Result<()> {
     let schema = DataSchemaRefExt::create(vec![
         DataField::new("a", DataType::Int64, false),
-        DataField::new("b", DataType::Utf8, false),
+        DataField::new("b", DataType::String, false),
     ]);
 
     let raw = DataBlock::create_by_array(schema.clone(), vec![
@@ -77,7 +77,7 @@ fn test_data_block_sort() -> Result<()> {
 fn test_data_block_merge_sort() -> Result<()> {
     let schema = DataSchemaRefExt::create(vec![
         DataField::new("a", DataType::Int64, false),
-        DataField::new("b", DataType::Utf8, false),
+        DataField::new("b", DataType::String, false),
     ]);
 
     let raw1 = DataBlock::create_by_array(schema.clone(), vec![

common/datablocks/src/kernels/data_block_take_test.rs

@@ -21,7 +21,7 @@ use crate::*;
 fn test_data_block_take() -> Result<()> {
     let schema = DataSchemaRefExt::create(vec![
         DataField::new("a", DataType::Int64, false),
-        DataField::new("b", DataType::Utf8, false),
+        DataField::new("b", DataType::String, false),
     ]);
 
     let raw = DataBlock::create_by_array(schema.clone(), vec![

common/datavalues/src/arrays/arithmetic.rs

@@ -312,24 +312,23 @@ where
     }
 }
 
-fn concat_strings(l: &str, r: &str) -> String {
-    // fastest way to concat strings according to https://github.com/hoodie/concatenation_benchmarks-rs
-    let mut s = String::with_capacity(l.len() + r.len());
-    s.push_str(l);
-    s.push_str(r);
+fn concat_strings(l: &[u8], r: &[u8]) -> Vec<u8> {
+    let mut s = Vec::with_capacity(l.len() + r.len());
+    s.extend_from_slice(l);
+    s.extend_from_slice(r);
     s
 }
 
-impl Add for &DFUtf8Array {
-    type Output = Result<DFUtf8Array>;
+impl Add for &DFStringArray {
+    type Output = Result<DFStringArray>;
 
     fn add(self, rhs: Self) -> Self::Output {
         // broadcasting path
         if rhs.len() == 1 {
             let rhs = rhs.get(0);
             return match rhs {
                 Some(rhs) => self.add(rhs),
-                None => Ok(DFUtf8Array::full_null(self.len())),
+                None => Ok(DFStringArray::full_null(self.len())),
             };
         }
 
@@ -346,18 +345,18 @@ impl Add for &DFUtf8Array {
     }
 }
 
-impl Add for DFUtf8Array {
-    type Output = Result<DFUtf8Array>;
+impl Add for DFStringArray {
+    type Output = Result<DFStringArray>;
 
     fn add(self, rhs: Self) -> Self::Output {
         (&self).add(&rhs)
     }
 }
 
-impl Add<&str> for &DFUtf8Array {
-    type Output = Result<DFUtf8Array>;
+impl Add<&[u8]> for &DFStringArray {
+    type Output = Result<DFStringArray>;
 
-    fn add(self, rhs: &str) -> Self::Output {
+    fn add(self, rhs: &[u8]) -> Self::Output {
         Ok(match self.null_count() {
             0 => self
                 .into_no_null_iter()
@@ -401,5 +400,5 @@ where
 }
 
 impl Pow for DFBooleanArray {}
-impl Pow for DFUtf8Array {}
+impl Pow for DFStringArray {}
 impl Pow for DFListArray {}