refactor: Use re-exported arrow mod from datatypes crate (GreptimeTea…

…m#571)
dut3062796s · Nov 18, 2022 · 22ae983 · 22ae983
1 parent e1f3262
commit 22ae983
Showing 31 changed files with 57 additions and 155 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/src/common/function/Cargo.toml b/src/common/function/Cargo.toml
@@ -21,20 +21,6 @@ paste = "1.0"
 snafu = { version = "0.7", features = ["backtraces"] }
 statrs = "0.15"
 
-[dependencies.arrow]
-features = [
-    "io_csv",
-    "io_json",
-    "io_parquet",
-    "io_parquet_compression",
-    "io_ipc",
-    "ahash",
-    "compute",
-    "serde_types",
-]
-package = "arrow2"
-version = "0.10"
-
 [dev-dependencies]
 ron = "0.7"
 serde = { version = "1.0", features = ["derive"] }
diff --git a/src/common/function/src/scalars/numpy/interp.rs b/src/common/function/src/scalars/numpy/interp.rs
@@ -14,9 +14,9 @@
 
 use std::sync::Arc;
 
-use arrow::array::PrimitiveArray;
-use arrow::compute::cast::primitive_to_primitive;
-use arrow::datatypes::DataType::Float64;
+use datatypes::arrow::array::PrimitiveArray;
+use datatypes::arrow::compute::cast::primitive_to_primitive;
+use datatypes::arrow::datatypes::DataType::Float64;
 use datatypes::data_type::DataType;
 use datatypes::prelude::ScalarVector;
 use datatypes::type_id::LogicalTypeId;

diff --git a/src/common/function/src/scalars/timestamp/from_unixtime.rs b/src/common/function/src/scalars/timestamp/from_unixtime.rs
@@ -17,11 +17,11 @@
 use std::fmt;
 use std::sync::Arc;
 
-use arrow::compute::arithmetics;
-use arrow::datatypes::DataType as ArrowDatatype;
-use arrow::scalar::PrimitiveScalar;
 use common_query::error::{IntoVectorSnafu, UnsupportedInputDataTypeSnafu};
 use common_query::prelude::{Signature, Volatility};
+use datatypes::arrow::compute::arithmetics;
+use datatypes::arrow::datatypes::DataType as ArrowDatatype;
+use datatypes::arrow::scalar::PrimitiveScalar;
 use datatypes::prelude::ConcreteDataType;
 use datatypes::vectors::{TimestampVector, VectorRef};
 use snafu::ResultExt;

diff --git a/src/common/grpc/Cargo.toml b/src/common/grpc/Cargo.toml
@@ -22,20 +22,6 @@ tokio = { version = "1.0", features = ["full"] }
 tonic = "0.8"
 tower = "0.4"
 
-[dependencies.arrow]
-package = "arrow2"
-version = "0.10"
-features = [
-    "io_csv",
-    "io_json",
-    "io_parquet",
-    "io_parquet_compression",
-    "io_ipc",
-    "ahash",
-    "compute",
-    "serde_types",
-]
-
 [dev-dependencies]
 criterion = "0.4"
 rand = "0.8"

diff --git a/src/common/grpc/src/physical/plan.rs b/src/common/grpc/src/physical/plan.rs
@@ -18,8 +18,6 @@ use std::sync::Arc;
 
 use api::v1::codec::physical_plan_node::PhysicalPlanType;
 use api::v1::codec::{MockInputExecNode, PhysicalPlanNode, ProjectionExecNode};
-use arrow::array::{PrimitiveArray, Utf8Array};
-use arrow::datatypes::{DataType, Field, Schema};
 use async_trait::async_trait;
 use datafusion::execution::runtime_env::RuntimeEnv;
 use datafusion::field_util::SchemaExt;
@@ -29,6 +27,8 @@ use datafusion::physical_plan::{
     ExecutionPlan, PhysicalExpr, SendableRecordBatchStream, Statistics,
 };
 use datafusion::record_batch::RecordBatch;
+use datatypes::arrow::array::{PrimitiveArray, Utf8Array};
+use datatypes::arrow::datatypes::{DataType, Field, Schema, SchemaRef};
 use snafu::{OptionExt, ResultExt};
 
 use crate::error::{
@@ -162,11 +162,11 @@ impl ExecutionPlan for MockExecution {
         self
     }
 
-    fn schema(&self) -> arrow::datatypes::SchemaRef {
+    fn schema(&self) -> SchemaRef {
         let field1 = Field::new("id", DataType::UInt32, false);
         let field2 = Field::new("name", DataType::Utf8, false);
         let field3 = Field::new("age", DataType::UInt32, false);
-        Arc::new(arrow::datatypes::Schema::new(vec![field1, field2, field3]))
+        Arc::new(Schema::new(vec![field1, field2, field3]))
     }
 
     fn output_partitioning(&self) -> datafusion::physical_plan::Partitioning {

diff --git a/src/common/grpc/src/select.rs b/src/common/grpc/src/select.rs
@@ -19,12 +19,12 @@ use api::result::{build_err_result, ObjectResultBuilder};
 use api::v1::codec::SelectResult;
 use api::v1::column::{SemanticType, Values};
 use api::v1::{Column, ObjectResult};
-use arrow::array::{Array, BooleanArray, PrimitiveArray};
 use common_base::BitVec;
 use common_error::prelude::ErrorExt;
 use common_error::status_code::StatusCode;
 use common_query::Output;
 use common_recordbatch::{util, RecordBatches, SendableRecordBatchStream};
+use datatypes::arrow::array::{Array, BooleanArray, PrimitiveArray};
 use datatypes::arrow_array::{BinaryArray, StringArray};
 use datatypes::schema::SchemaRef;
 use snafu::{OptionExt, ResultExt};
@@ -136,7 +136,8 @@ pub fn null_mask(arrays: &Vec<Arc<dyn Array>>, row_count: usize) -> Vec<u8> {
 }
 
 macro_rules! convert_arrow_array_to_grpc_vals {
-    ($data_type: expr, $arrays: ident,  $(($Type: pat, $CastType: ty, $field: ident, $MapFunction: expr)), +) => {
+    ($data_type: expr, $arrays: ident,  $(($Type: pat, $CastType: ty, $field: ident, $MapFunction: expr)), +) => {{
+        use datatypes::arrow::datatypes::{DataType, TimeUnit};
         match $data_type {
             $(
                 $Type => {
@@ -155,7 +156,7 @@ macro_rules! convert_arrow_array_to_grpc_vals {
             )+
             _ => unimplemented!(),
         }
-    };
+    }};
 }
 
 pub fn values(arrays: &[Arc<dyn Array>]) -> Result<Values> {
@@ -164,7 +165,6 @@ pub fn values(arrays: &[Arc<dyn Array>]) -> Result<Values> {
     }
     let data_type = arrays[0].data_type();
 
-    use arrow::datatypes::DataType;
     convert_arrow_array_to_grpc_vals!(
         data_type, arrays,
 
@@ -192,19 +192,18 @@ pub fn values(arrays: &[Arc<dyn Array>]) -> Result<Values> {
         (DataType::Date32,        PrimitiveArray<i32>,    date_values,    |x| {*x as i32}),
         (DataType::Date64,        PrimitiveArray<i64>,    datetime_values,|x| {*x as i64}),
 
-        (DataType::Timestamp(arrow::datatypes::TimeUnit::Millisecond, _), PrimitiveArray<i64>, ts_millis_values, |x| {*x})
+        (DataType::Timestamp(TimeUnit::Millisecond, _), PrimitiveArray<i64>, ts_millis_values, |x| {*x})
     )
 }
 
 #[cfg(test)]
 mod tests {
     use std::sync::Arc;
 
-    use arrow::array::{Array, BooleanArray, PrimitiveArray};
-    use arrow::datatypes::{DataType, Field};
     use common_recordbatch::{RecordBatch, RecordBatches};
     use datafusion::field_util::SchemaExt;
-    use datatypes::arrow::datatypes::Schema as ArrowSchema;
+    use datatypes::arrow::array::{Array, BooleanArray, PrimitiveArray};
+    use datatypes::arrow::datatypes::{DataType, Field, Schema as ArrowSchema};
     use datatypes::arrow_array::StringArray;
     use datatypes::schema::Schema;
     use datatypes::vectors::{UInt32Vector, VectorRef};

diff --git a/src/common/query/Cargo.toml b/src/common/query/Cargo.toml
@@ -18,10 +18,6 @@ datatypes = { path = "../../datatypes" }
 snafu = { version = "0.7", features = ["backtraces"] }
 statrs = "0.15"
 
-[dependencies.arrow]
-package = "arrow2"
-version = "0.10"
-
 [dev-dependencies]
 common-base = { path = "../base" }
 tokio = { version = "1.0", features = ["full"] }
diff --git a/src/common/query/src/columnar_value.rs b/src/common/query/src/columnar_value.rs
@@ -80,7 +80,7 @@ impl From<ColumnarValue> for DfColumnarValue {
 mod tests {
     use std::sync::Arc;
 
-    use arrow::datatypes::DataType as ArrowDataType;
+    use datatypes::arrow::datatypes::DataType as ArrowDataType;
     use datatypes::vectors::BooleanVector;
 
     use super::*;

diff --git a/src/common/query/src/error.rs b/src/common/query/src/error.rs
@@ -14,9 +14,9 @@
 
 use std::any::Any;
 
-use arrow::datatypes::DataType as ArrowDatatype;
 use common_error::prelude::*;
 use datafusion_common::DataFusionError;
+use datatypes::arrow::datatypes::DataType as ArrowDatatype;
 use datatypes::error::Error as DataTypeError;
 use datatypes::prelude::ConcreteDataType;
 use statrs::StatsError;
@@ -180,7 +180,7 @@ impl From<BoxedError> for Error {
 
 #[cfg(test)]
 mod tests {
-    use arrow::error::ArrowError;
+    use datatypes::arrow::error::ArrowError;
     use snafu::GenerateImplicitData;
 
     use super::*;

diff --git a/src/common/query/src/function.rs b/src/common/query/src/function.rs
@@ -14,8 +14,8 @@
 
 use std::sync::Arc;
 
-use arrow::datatypes::DataType as ArrowDataType;
 use datafusion_expr::ReturnTypeFunction as DfReturnTypeFunction;
+use datatypes::arrow::datatypes::DataType as ArrowDataType;
 use datatypes::prelude::{ConcreteDataType, DataType};
 use datatypes::vectors::VectorRef;
 use snafu::ResultExt;

diff --git a/src/common/query/src/logical_plan/accumulator.rs b/src/common/query/src/logical_plan/accumulator.rs
@@ -17,10 +17,10 @@
 use std::fmt::Debug;
 use std::sync::Arc;
 
-use arrow::array::ArrayRef;
 use common_time::timestamp::TimeUnit;
 use datafusion_common::Result as DfResult;
 use datafusion_expr::Accumulator as DfAccumulator;
+use datatypes::arrow::array::ArrayRef;
 use datatypes::prelude::*;
 use datatypes::value::ListValue;
 use datatypes::vectors::{Helper as VectorHelper, VectorRef};
@@ -266,9 +266,9 @@ fn try_convert_list_value(list: ListValue) -> Result<ScalarValue> {
 
 #[cfg(test)]
 mod tests {
-    use arrow::datatypes::DataType;
     use common_base::bytes::{Bytes, StringBytes};
     use datafusion_common::ScalarValue;
+    use datatypes::arrow::datatypes::DataType;
     use datatypes::value::{ListValue, OrderedFloat};
 
     use super::*;

diff --git a/src/common/query/src/logical_plan/mod.rs b/src/common/query/src/logical_plan/mod.rs
@@ -72,12 +72,12 @@ pub fn create_aggregate_function(
 mod tests {
     use std::sync::Arc;
 
-    use arrow::array::BooleanArray;
-    use arrow::datatypes::DataType;
     use datafusion_expr::{
         ColumnarValue as DfColumnarValue, ScalarUDF as DfScalarUDF,
         TypeSignature as DfTypeSignature,
     };
+    use datatypes::arrow::array::BooleanArray;
+    use datatypes::arrow::datatypes::DataType;
     use datatypes::prelude::*;
     use datatypes::vectors::{BooleanVector, VectorRef};
 

diff --git a/src/common/query/src/logical_plan/udaf.rs b/src/common/query/src/logical_plan/udaf.rs
@@ -19,11 +19,11 @@
 use std::fmt::{self, Debug, Formatter};
 use std::sync::Arc;
 
-use arrow::datatypes::DataType as ArrowDataType;
 use datafusion_expr::{
     AccumulatorFunctionImplementation as DfAccumulatorFunctionImplementation,
     AggregateUDF as DfAggregateUdf, StateTypeFunction as DfStateTypeFunction,
 };
+use datatypes::arrow::datatypes::DataType as ArrowDataType;
 use datatypes::prelude::*;
 
 use crate::function::{

diff --git a/src/common/query/src/physical_plan.rs b/src/common/query/src/physical_plan.rs
@@ -199,7 +199,6 @@ impl DfPhysicalPlan for DfPhysicalPlanAdapter {
 
 #[cfg(test)]
 mod test {
-    use arrow::datatypes::{DataType, Field, Schema as ArrowSchema};
     use common_recordbatch::{RecordBatch, RecordBatches};
     use datafusion::arrow_print;
     use datafusion::datasource::TableProvider as DfTableProvider;
@@ -209,6 +208,7 @@ mod test {
     use datafusion::prelude::ExecutionContext;
     use datafusion_common::field_util::SchemaExt;
     use datafusion_expr::Expr;
+    use datatypes::arrow::datatypes::{DataType, Field, Schema as ArrowSchema};
     use datatypes::schema::Schema;
     use datatypes::vectors::Int32Vector;
 

diff --git a/src/common/query/src/signature.rs b/src/common/query/src/signature.rs
@@ -15,9 +15,9 @@
 //! Signature module contains foundational types that are used to represent signatures, types,
 //! and return types of functions.
 //! Copied and modified from datafusion.
-use arrow::datatypes::DataType as ArrowDataType;
 pub use datafusion::physical_plan::functions::Volatility;
 use datafusion_expr::{Signature as DfSignature, TypeSignature as DfTypeSignature};
+use datatypes::arrow::datatypes::DataType as ArrowDataType;
 use datatypes::data_type::DataType;
 use datatypes::prelude::ConcreteDataType;
 
@@ -143,7 +143,7 @@ impl From<Signature> for DfSignature {
 
 #[cfg(test)]
 mod tests {
-    use arrow::datatypes::DataType;
+    use datatypes::arrow::datatypes::DataType;
 
     use super::*;
 

diff --git a/src/datanode/Cargo.toml b/src/datanode/Cargo.toml
@@ -54,20 +54,6 @@ tower = { version = "0.4", features = ["full"] }
 tower-http = { version = "0.3", features = ["full"] }
 frontend = { path = "../frontend" }
 
-[dependencies.arrow]
-package = "arrow2"
-version = "0.10"
-features = [
-  "io_csv",
-  "io_json",
-  "io_parquet",
-  "io_parquet_compression",
-  "io_ipc",
-  "ahash",
-  "compute",
-  "serde_types",
-]
-
 [dev-dependencies]
 axum-test-helper = { git = "https://github.com/sunng87/axum-test-helper.git", branch = "patch-1" }
 client = { path = "../client" }
@@ -77,17 +63,3 @@ datafusion = { git = "https://github.com/apache/arrow-datafusion.git", branch =
 ] }
 datafusion-common = { git = "https://github.com/apache/arrow-datafusion.git", branch = "arrow2" }
 tempdir = "0.3"
-
-[dev-dependencies.arrow]
-package = "arrow2"
-version = "0.10"
-features = [
-  "io_csv",
-  "io_json",
-  "io_parquet",
-  "io_parquet_compression",
-  "io_ipc",
-  "ahash",
-  "compute",
-  "serde_types",
-]