feat: Implement BinaryType and BinaryVector

src/common/src/bytes.rs

@@ -0,0 +1,7 @@
+/// Bytes buffer.
+#[derive(Debug, Default, Clone)]
+pub struct Bytes(Vec<u8>);
+
+/// String buffer with arbitrary encoding.
+#[derive(Debug, Default, Clone)]
+pub struct StringBytes(Vec<u8>);

src/common/src/lib.rs

@@ -1,7 +1 @@
-/// Bytes buffer.
-#[derive(Debug, Default, Clone)]
-pub struct Bytes(Vec<u8>);
-
-/// String buffer with arbitrary encoding.
-#[derive(Debug, Default, Clone)]
-pub struct StringBytes(Vec<u8>);
+pub mod bytes;

src/datatypes/src/data_type.rs

@@ -4,7 +4,7 @@ use crate::type_id::LogicalTypeId;
 use crate::value::Value;
 
 /// Data type abstraction.
-pub trait DataType: std::fmt::Debug {
+pub trait DataType: std::fmt::Debug + Send + Sync {
     /// Name of this data type.
     fn name(&self) -> &str;
 

src/datatypes/src/lib.rs

@@ -1,6 +1,11 @@
 mod data_type;
+pub mod prelude;
 mod schema;
 pub mod type_id;
 mod types;
 pub mod value;
 pub mod vectors;
+
+use arrow2::array::BinaryArray;
+
+pub type LargeBinaryArray = BinaryArray<i64>;

src/datatypes/src/prelude.rs

@@ -0,0 +1,3 @@
+pub use crate::data_type::{DataType, DataTypeRef};
+pub use crate::type_id::LogicalTypeId;
+pub use crate::value::Value;

src/datatypes/src/types.rs

@@ -1,2 +1,3 @@
+pub mod binary_type;
 pub mod primitive_traits;
 pub mod primitive_type;

src/datatypes/src/types/binary_type.rs

@@ -0,0 +1,30 @@
+use std::sync::Arc;
+
+use common::bytes::StringBytes;
+
+use crate::data_type::{DataType, DataTypeRef};
+use crate::type_id::LogicalTypeId;
+use crate::value::Value;
+
+#[derive(Debug, Default)]
+pub struct BinaryType;
+
+impl BinaryType {
+    pub fn arc() -> DataTypeRef {
+        Arc::new(Self)
+    }
+}
+
+impl DataType for BinaryType {
+    fn name(&self) -> &str {
+        "Binary"
+    }
+
+    fn logical_type_id(&self) -> LogicalTypeId {
+        LogicalTypeId::String
+    }
+
+    fn default_value(&self) -> Value {
+        StringBytes::default().into()
+    }
+}

src/datatypes/src/value.rs

@@ -1,4 +1,4 @@
-use common::{Bytes, StringBytes};
+use common::bytes::{Bytes, StringBytes};
 
 /// Value holds a single arbitrary value of any [DataType](crate::data_type::DataType).
 #[derive(Debug)]
@@ -57,3 +57,5 @@ impl_from!(Int32, i32);
 impl_from!(Int64, i64);
 impl_from!(Float32, f32);
 impl_from!(Float64, f64);
+impl_from!(String, StringBytes);
+impl_from!(Binary, Bytes);

src/datatypes/src/vectors.rs

@@ -1,3 +1,4 @@
+pub mod binary;
 pub mod primitive;
 
 use std::any::Any;
@@ -8,6 +9,8 @@ use crate::data_type::DataTypeRef;
 /// Vector of data values.
 pub trait Vector: Send + Sync {
     /// Returns the data type of the vector.
+    ///
+    /// This may require heap allocation.
     fn data_type(&self) -> DataTypeRef;
 
     /// Returns the vector as [Any](std::any::Any) so that it can be

src/datatypes/src/vectors/binary.rs

@@ -0,0 +1,26 @@
+use std::any::Any;
+
+use crate::data_type::DataTypeRef;
+use crate::types::binary_type::BinaryType;
+use crate::vectors::Vector;
+use crate::LargeBinaryArray;
+
+/// Vector of binary strings.
+#[derive(Debug)]
+pub struct BinaryVector {
+    array: LargeBinaryArray,
+}
+
+impl Vector for BinaryVector {
+    fn data_type(&self) -> DataTypeRef {
+        BinaryType::arc()
+    }
+
+    fn as_any(&self) -> &dyn Any {
+        self
+    }
+
+    fn len(&self) -> usize {
+        self.array.len()
+    }
+}