mirror of
https://github.com/GreptimeTeam/greptimedb.git
synced 2026-05-24 17:00:37 +00:00
chore: upgrade DataFusion family (#7558)
* chore: upgrade DataFusion family Signed-off-by: luofucong <luofc@foxmail.com> * use main proto Signed-off-by: luofucong <luofc@foxmail.com> * fix ci Signed-off-by: luofucong <luofc@foxmail.com> --------- Signed-off-by: luofucong <luofc@foxmail.com>
This commit is contained in:
@@ -27,13 +27,14 @@ common-recordbatch.workspace = true
|
||||
common-runtime.workspace = true
|
||||
common-telemetry.workspace = true
|
||||
datafusion.workspace = true
|
||||
datafusion-datasource.workspace = true
|
||||
datafusion-orc.workspace = true
|
||||
datatypes.workspace = true
|
||||
futures.workspace = true
|
||||
lazy_static.workspace = true
|
||||
object-store.workspace = true
|
||||
object_store_opendal.workspace = true
|
||||
orc-rust = { version = "0.6.3", default-features = false, features = ["async"] }
|
||||
orc-rust = { version = "0.7", default-features = false, features = ["async"] }
|
||||
parquet.workspace = true
|
||||
paste.workspace = true
|
||||
regex.workspace = true
|
||||
|
||||
@@ -14,7 +14,7 @@
|
||||
|
||||
use arrow::record_batch::RecordBatch;
|
||||
use async_trait::async_trait;
|
||||
use datafusion::parquet::format::FileMetaData;
|
||||
use parquet::file::metadata::ParquetMetaData;
|
||||
|
||||
use crate::error::Result;
|
||||
|
||||
@@ -24,5 +24,5 @@ pub trait DfRecordBatchEncoder {
|
||||
|
||||
#[async_trait]
|
||||
pub trait ArrowWriterCloser {
|
||||
async fn close(mut self) -> Result<FileMetaData>;
|
||||
async fn close(mut self) -> Result<ParquetMetaData>;
|
||||
}
|
||||
|
||||
@@ -40,7 +40,6 @@ use datafusion::datasource::physical_plan::{
|
||||
use datafusion::error::{DataFusionError, Result as DataFusionResult};
|
||||
use datafusion::physical_plan::SendableRecordBatchStream;
|
||||
use datafusion::physical_plan::metrics::ExecutionPlanMetricsSet;
|
||||
use datatypes::arrow::datatypes::SchemaRef;
|
||||
use futures::{StreamExt, TryStreamExt};
|
||||
use object_store::ObjectStore;
|
||||
use object_store_opendal::OpendalStore;
|
||||
@@ -303,24 +302,20 @@ where
|
||||
pub async fn file_to_stream(
|
||||
store: &ObjectStore,
|
||||
filename: &str,
|
||||
file_schema: SchemaRef,
|
||||
file_source: Arc<dyn FileSource>,
|
||||
projection: Option<Vec<usize>>,
|
||||
compression_type: CompressionType,
|
||||
) -> Result<DfSendableRecordBatchStream> {
|
||||
let df_compression: DfCompressionType = compression_type.into();
|
||||
let config = FileScanConfigBuilder::new(
|
||||
ObjectStoreUrl::local_filesystem(),
|
||||
file_schema,
|
||||
file_source.clone(),
|
||||
)
|
||||
.with_file_group(FileGroup::new(vec![PartitionedFile::new(
|
||||
filename.to_string(),
|
||||
0,
|
||||
)]))
|
||||
.with_projection(projection)
|
||||
.with_file_compression_type(df_compression)
|
||||
.build();
|
||||
let config =
|
||||
FileScanConfigBuilder::new(ObjectStoreUrl::local_filesystem(), file_source.clone())
|
||||
.with_file_group(FileGroup::new(vec![PartitionedFile::new(
|
||||
filename.to_string(),
|
||||
0,
|
||||
)]))
|
||||
.with_projection_indices(projection)
|
||||
.with_file_compression_type(df_compression)
|
||||
.build();
|
||||
|
||||
let store = Arc::new(OpendalStore::new(store.clone()));
|
||||
let file_opener = file_source
|
||||
|
||||
@@ -440,14 +440,11 @@ mod tests {
|
||||
.await
|
||||
.unwrap(),
|
||||
);
|
||||
let csv_source = CsvSource::new(true, b',', b'"')
|
||||
.with_schema(schema.clone())
|
||||
.with_batch_size(8192);
|
||||
let csv_source = CsvSource::new(schema).with_batch_size(8192);
|
||||
|
||||
let stream = file_to_stream(
|
||||
&store,
|
||||
compressed_file_path_str,
|
||||
schema.clone(),
|
||||
csv_source.clone(),
|
||||
None,
|
||||
compression_type,
|
||||
|
||||
@@ -347,14 +347,11 @@ mod tests {
|
||||
.await
|
||||
.unwrap(),
|
||||
);
|
||||
let json_source = JsonSource::new()
|
||||
.with_schema(schema.clone())
|
||||
.with_batch_size(8192);
|
||||
let json_source = JsonSource::new(schema).with_batch_size(8192);
|
||||
|
||||
let stream = file_to_stream(
|
||||
&store,
|
||||
compressed_file_path_str,
|
||||
schema.clone(),
|
||||
json_source.clone(),
|
||||
None,
|
||||
compression_type,
|
||||
|
||||
@@ -18,15 +18,15 @@ use std::sync::Arc;
|
||||
use arrow::record_batch::RecordBatch;
|
||||
use arrow_schema::Schema;
|
||||
use async_trait::async_trait;
|
||||
use datafusion::datasource::physical_plan::{FileMeta, ParquetFileReaderFactory};
|
||||
use datafusion::datasource::physical_plan::ParquetFileReaderFactory;
|
||||
use datafusion::error::Result as DatafusionResult;
|
||||
use datafusion::parquet::arrow::async_reader::AsyncFileReader;
|
||||
use datafusion::parquet::arrow::{ArrowWriter, parquet_to_arrow_schema};
|
||||
use datafusion::parquet::errors::{ParquetError, Result as ParquetResult};
|
||||
use datafusion::parquet::file::metadata::ParquetMetaData;
|
||||
use datafusion::parquet::format::FileMetaData;
|
||||
use datafusion::physical_plan::SendableRecordBatchStream;
|
||||
use datafusion::physical_plan::metrics::ExecutionPlanMetricsSet;
|
||||
use datafusion_datasource::PartitionedFile;
|
||||
use datatypes::schema::SchemaRef;
|
||||
use futures::StreamExt;
|
||||
use futures::future::BoxFuture;
|
||||
@@ -100,11 +100,11 @@ impl ParquetFileReaderFactory for DefaultParquetFileReaderFactory {
|
||||
fn create_reader(
|
||||
&self,
|
||||
_partition_index: usize,
|
||||
file_meta: FileMeta,
|
||||
partitioned_file: PartitionedFile,
|
||||
_metadata_size_hint: Option<usize>,
|
||||
_metrics: &ExecutionPlanMetricsSet,
|
||||
) -> DatafusionResult<Box<dyn AsyncFileReader + Send>> {
|
||||
let path = file_meta.location().to_string();
|
||||
let path = partitioned_file.path().to_string();
|
||||
let object_store = self.object_store.clone();
|
||||
|
||||
Ok(Box::new(LazyParquetFileReader::new(object_store, path)))
|
||||
@@ -180,7 +180,7 @@ impl DfRecordBatchEncoder for ArrowWriter<SharedBuffer> {
|
||||
|
||||
#[async_trait]
|
||||
impl ArrowWriterCloser for ArrowWriter<SharedBuffer> {
|
||||
async fn close(self) -> Result<FileMetaData> {
|
||||
async fn close(self) -> Result<ParquetMetaData> {
|
||||
self.close().context(error::EncodeRecordBatchSnafu)
|
||||
}
|
||||
}
|
||||
|
||||
@@ -67,14 +67,14 @@ impl Test<'_> {
|
||||
async fn test_json_opener() {
|
||||
let store = test_store("/");
|
||||
let schema = test_basic_schema();
|
||||
let file_source = Arc::new(JsonSource::new()).with_batch_size(test_util::TEST_BATCH_SIZE);
|
||||
let file_source = Arc::new(JsonSource::new(schema)).with_batch_size(test_util::TEST_BATCH_SIZE);
|
||||
|
||||
let path = &find_workspace_path("/src/common/datasource/tests/json/basic.json")
|
||||
.display()
|
||||
.to_string();
|
||||
let tests = [
|
||||
Test {
|
||||
config: scan_config(schema.clone(), None, path, file_source.clone()),
|
||||
config: scan_config(None, path, file_source.clone()),
|
||||
file_source: file_source.clone(),
|
||||
expected: vec![
|
||||
"+-----+-------+",
|
||||
@@ -87,7 +87,7 @@ async fn test_json_opener() {
|
||||
],
|
||||
},
|
||||
Test {
|
||||
config: scan_config(schema, Some(1), path, file_source.clone()),
|
||||
config: scan_config(Some(1), path, file_source.clone()),
|
||||
file_source,
|
||||
expected: vec![
|
||||
"+-----+------+",
|
||||
@@ -112,13 +112,11 @@ async fn test_csv_opener() {
|
||||
.display()
|
||||
.to_string();
|
||||
|
||||
let file_source = CsvSource::new(true, b',', b'"')
|
||||
.with_batch_size(test_util::TEST_BATCH_SIZE)
|
||||
.with_schema(schema.clone());
|
||||
let file_source = CsvSource::new(schema).with_batch_size(test_util::TEST_BATCH_SIZE);
|
||||
|
||||
let tests = [
|
||||
Test {
|
||||
config: scan_config(schema.clone(), None, path, file_source.clone()),
|
||||
config: scan_config(None, path, file_source.clone()),
|
||||
file_source: file_source.clone(),
|
||||
expected: vec![
|
||||
"+-----+-------+---------------------+----------+------------+",
|
||||
@@ -131,7 +129,7 @@ async fn test_csv_opener() {
|
||||
],
|
||||
},
|
||||
Test {
|
||||
config: scan_config(schema, Some(1), path, file_source.clone()),
|
||||
config: scan_config(Some(1), path, file_source.clone()),
|
||||
file_source,
|
||||
expected: vec![
|
||||
"+-----+------+---------------------+----------+------------+",
|
||||
@@ -158,10 +156,10 @@ async fn test_parquet_exec() {
|
||||
.display()
|
||||
.to_string();
|
||||
|
||||
let parquet_source = ParquetSource::default()
|
||||
let parquet_source = ParquetSource::new(schema)
|
||||
.with_parquet_file_reader_factory(Arc::new(DefaultParquetFileReaderFactory::new(store)));
|
||||
|
||||
let config = scan_config(schema, None, path, Arc::new(parquet_source));
|
||||
let config = scan_config(None, path, Arc::new(parquet_source));
|
||||
let exec = DataSourceExec::from_data_source(config);
|
||||
let ctx = SessionContext::new();
|
||||
|
||||
@@ -197,11 +195,11 @@ async fn test_orc_opener() {
|
||||
|
||||
let store = test_store("/");
|
||||
let schema = Arc::new(OrcFormat.infer_schema(&store, path).await.unwrap());
|
||||
let file_source = Arc::new(OrcSource::default());
|
||||
let file_source = Arc::new(OrcSource::new(schema.into()));
|
||||
|
||||
let tests = [
|
||||
Test {
|
||||
config: scan_config(schema.clone(), None, path, file_source.clone()),
|
||||
config: scan_config(None, path, file_source.clone()),
|
||||
file_source: file_source.clone(),
|
||||
expected: vec![
|
||||
"+----------+-----+-------+------------+-----+-----+-------+--------------------+------------------------+-----------+---------------+------------+----------------+---------------+-------------------+--------------+---------------+---------------+----------------------------+-------------+",
|
||||
@@ -216,7 +214,7 @@ async fn test_orc_opener() {
|
||||
],
|
||||
},
|
||||
Test {
|
||||
config: scan_config(schema.clone(), Some(1), path, file_source.clone()),
|
||||
config: scan_config(Some(1), path, file_source.clone()),
|
||||
file_source,
|
||||
expected: vec![
|
||||
"+----------+-----+------+------------+---+-----+-------+--------------------+------------------------+-----------+---------------+------------+----------------+---------------+-------------------+--------------+---------------+---------------+-------------------------+-------------+",
|
||||
|
||||
@@ -80,7 +80,6 @@ pub fn csv_basic_schema() -> SchemaRef {
|
||||
}
|
||||
|
||||
pub(crate) fn scan_config(
|
||||
file_schema: SchemaRef,
|
||||
limit: Option<usize>,
|
||||
filename: &str,
|
||||
file_source: Arc<dyn FileSource>,
|
||||
@@ -89,7 +88,7 @@ pub(crate) fn scan_config(
|
||||
let filename = &filename.replace('\\', "/");
|
||||
let file_group = FileGroup::new(vec![PartitionedFile::new(filename.clone(), 4096)]);
|
||||
|
||||
FileScanConfigBuilder::new(ObjectStoreUrl::local_filesystem(), file_schema, file_source)
|
||||
FileScanConfigBuilder::new(ObjectStoreUrl::local_filesystem(), file_source)
|
||||
.with_file_group(file_group)
|
||||
.with_limit(limit)
|
||||
.build()
|
||||
@@ -109,7 +108,7 @@ pub async fn setup_stream_to_json_test(origin_path: &str, threshold: impl Fn(usi
|
||||
|
||||
let size = store.read(origin_path).await.unwrap().len();
|
||||
|
||||
let config = scan_config(schema, None, origin_path, Arc::new(JsonSource::new()));
|
||||
let config = scan_config(None, origin_path, Arc::new(JsonSource::new(schema)));
|
||||
let stream = FileStream::new(
|
||||
&config,
|
||||
0,
|
||||
@@ -151,10 +150,8 @@ pub async fn setup_stream_to_csv_test(
|
||||
|
||||
let schema = csv_basic_schema();
|
||||
|
||||
let csv_source = CsvSource::new(true, b',', b'"')
|
||||
.with_schema(schema.clone())
|
||||
.with_batch_size(TEST_BATCH_SIZE);
|
||||
let config = scan_config(schema, None, origin_path, csv_source.clone());
|
||||
let csv_source = CsvSource::new(schema).with_batch_size(TEST_BATCH_SIZE);
|
||||
let config = scan_config(None, origin_path, csv_source.clone());
|
||||
let size = store.read(origin_path).await.unwrap().len();
|
||||
|
||||
let csv_opener = csv_source.create_file_opener(
|
||||
|
||||
@@ -104,7 +104,8 @@ mod tests {
|
||||
assert!(matches!(f.signature(),
|
||||
datafusion_expr::Signature {
|
||||
type_signature: datafusion_expr::TypeSignature::Uniform(1, valid_types),
|
||||
volatility: datafusion_expr::Volatility::Immutable
|
||||
volatility: datafusion_expr::Volatility::Immutable,
|
||||
..
|
||||
} if valid_types == &ConcreteDataType::numerics().into_iter().map(|dt| { use datatypes::data_type::DataType; dt.as_arrow_type() }).collect::<Vec<_>>()));
|
||||
}
|
||||
|
||||
|
||||
@@ -331,7 +331,8 @@ mod tests {
|
||||
assert!(matches!(f.signature(),
|
||||
datafusion_expr::Signature {
|
||||
type_signature: datafusion_expr::TypeSignature::Uniform(1, valid_types),
|
||||
volatility: datafusion_expr::Volatility::Immutable
|
||||
volatility: datafusion_expr::Volatility::Immutable,
|
||||
..
|
||||
} if valid_types == &vec![ArrowDataType::Utf8]));
|
||||
}
|
||||
|
||||
|
||||
@@ -145,7 +145,8 @@ mod tests {
|
||||
assert!(matches!(f.signature(),
|
||||
datafusion_expr::Signature {
|
||||
type_signature: datafusion_expr::TypeSignature::OneOf(sigs),
|
||||
volatility: datafusion_expr::Volatility::Immutable
|
||||
volatility: datafusion_expr::Volatility::Immutable,
|
||||
..
|
||||
} if sigs.len() == 2));
|
||||
}
|
||||
|
||||
|
||||
@@ -341,6 +341,7 @@ impl AggregateUDFImpl for StateWrapper {
|
||||
name: acc_args.name,
|
||||
is_distinct: acc_args.is_distinct,
|
||||
exprs: acc_args.exprs,
|
||||
expr_fields: acc_args.expr_fields,
|
||||
};
|
||||
self.inner.accumulator(acc_args)?
|
||||
};
|
||||
|
||||
@@ -650,7 +650,7 @@ async fn test_last_value_order_by_udaf() {
|
||||
DataType::Timestamp(arrow_schema::TimeUnit::Millisecond, None),
|
||||
true
|
||||
), // ordering field is added to state fields too
|
||||
Field::new("is_set", DataType::Boolean, true)
|
||||
Field::new("last_value[last_value_is_set]", DataType::Boolean, true)
|
||||
]
|
||||
.into()
|
||||
),
|
||||
@@ -735,7 +735,7 @@ async fn test_last_value_order_by_udaf() {
|
||||
DataType::Timestamp(arrow_schema::TimeUnit::Millisecond, None),
|
||||
true,
|
||||
),
|
||||
Field::new("is_set", DataType::Boolean, true),
|
||||
Field::new("last_value[last_value_is_set]", DataType::Boolean, true),
|
||||
]
|
||||
.into(),
|
||||
vec![
|
||||
|
||||
@@ -122,7 +122,8 @@ mod tests {
|
||||
matches!(f.signature(),
|
||||
Signature {
|
||||
type_signature: TypeSignature::OneOf(sigs),
|
||||
volatility: Volatility::Immutable
|
||||
volatility: Volatility::Immutable,
|
||||
..
|
||||
} if sigs.len() == 15),
|
||||
"{:?}",
|
||||
f.signature()
|
||||
|
||||
@@ -193,7 +193,8 @@ mod tests {
|
||||
assert!(matches!(f.signature(),
|
||||
Signature {
|
||||
type_signature: TypeSignature::OneOf(sigs),
|
||||
volatility: Volatility::Immutable
|
||||
volatility: Volatility::Immutable,
|
||||
..
|
||||
} if sigs.len() == 6));
|
||||
}
|
||||
|
||||
|
||||
@@ -120,7 +120,8 @@ mod tests {
|
||||
matches!(f.signature(),
|
||||
Signature {
|
||||
type_signature: TypeSignature::OneOf(sigs),
|
||||
volatility: Volatility::Immutable
|
||||
volatility: Volatility::Immutable,
|
||||
..
|
||||
} if sigs.len() == 15),
|
||||
"{:?}",
|
||||
f.signature()
|
||||
|
||||
@@ -25,7 +25,6 @@ use datafusion_common::arrow::array::{
|
||||
};
|
||||
use datafusion_common::arrow::datatypes::DataType;
|
||||
use datafusion_common::{DataFusionError, Result};
|
||||
use datafusion_expr::type_coercion::aggregates::STRINGS;
|
||||
use datafusion_expr::{ColumnarValue, ScalarFunctionArgs, Signature, Volatility};
|
||||
use datatypes::arrow_array::{int_array_value_at_index, string_array_value_at_index};
|
||||
use datatypes::json::JsonStructureSettings;
|
||||
@@ -519,7 +518,7 @@ impl Default for JsonGetObject {
|
||||
DataType::LargeBinary,
|
||||
DataType::BinaryView,
|
||||
],
|
||||
STRINGS.to_vec(),
|
||||
vec![DataType::UInt8, DataType::LargeUtf8, DataType::Utf8View],
|
||||
),
|
||||
}
|
||||
}
|
||||
|
||||
@@ -99,7 +99,8 @@ mod tests {
|
||||
assert!(matches!(rate.signature(),
|
||||
Signature {
|
||||
type_signature: TypeSignature::Uniform(2, valid_types),
|
||||
volatility: Volatility::Immutable
|
||||
volatility: Volatility::Immutable,
|
||||
..
|
||||
} if valid_types == NUMERICS
|
||||
));
|
||||
let values = vec![1.0, 3.0, 6.0];
|
||||
|
||||
@@ -19,8 +19,10 @@ use datafusion_common::DataFusionError;
|
||||
use datafusion_common::arrow::array::{Array, AsArray, StringViewBuilder};
|
||||
use datafusion_common::arrow::compute;
|
||||
use datafusion_common::arrow::datatypes::DataType;
|
||||
use datafusion_expr::type_coercion::aggregates::BINARYS;
|
||||
use datafusion_expr::{ColumnarValue, ScalarFunctionArgs, Signature, TypeSignature, Volatility};
|
||||
use datafusion_common::types::logical_binary;
|
||||
use datafusion_expr::{
|
||||
Coercion, ColumnarValue, ScalarFunctionArgs, Signature, TypeSignatureClass, Volatility,
|
||||
};
|
||||
use datatypes::types::vector_type_value_to_string;
|
||||
|
||||
use crate::function::{Function, extract_args};
|
||||
@@ -35,11 +37,10 @@ pub struct VectorToStringFunction {
|
||||
impl Default for VectorToStringFunction {
|
||||
fn default() -> Self {
|
||||
Self {
|
||||
signature: Signature::one_of(
|
||||
vec![
|
||||
TypeSignature::Uniform(1, vec![DataType::BinaryView]),
|
||||
TypeSignature::Uniform(1, BINARYS.to_vec()),
|
||||
],
|
||||
signature: Signature::coercible(
|
||||
vec![Coercion::new_exact(TypeSignatureClass::Native(
|
||||
logical_binary(),
|
||||
))],
|
||||
Volatility::Immutable,
|
||||
),
|
||||
}
|
||||
|
||||
@@ -15,10 +15,10 @@
|
||||
use std::fmt::Display;
|
||||
|
||||
use datafusion::arrow::datatypes::DataType;
|
||||
use datafusion::logical_expr::ColumnarValue;
|
||||
use datafusion::logical_expr::{Coercion, ColumnarValue, TypeSignature, TypeSignatureClass};
|
||||
use datafusion_common::ScalarValue;
|
||||
use datafusion_expr::type_coercion::aggregates::{BINARYS, STRINGS};
|
||||
use datafusion_expr::{ScalarFunctionArgs, Signature, TypeSignature, Volatility};
|
||||
use datafusion_common::types::{logical_binary, logical_string};
|
||||
use datafusion_expr::{ScalarFunctionArgs, Signature, Volatility};
|
||||
use nalgebra::DVectorView;
|
||||
|
||||
use crate::function::Function;
|
||||
@@ -36,9 +36,12 @@ impl Default for ElemAvgFunction {
|
||||
Self {
|
||||
signature: Signature::one_of(
|
||||
vec![
|
||||
TypeSignature::Uniform(1, STRINGS.to_vec()),
|
||||
TypeSignature::Uniform(1, BINARYS.to_vec()),
|
||||
TypeSignature::Uniform(1, vec![DataType::BinaryView]),
|
||||
TypeSignature::Coercible(vec![Coercion::new_exact(
|
||||
TypeSignatureClass::Native(logical_binary()),
|
||||
)]),
|
||||
TypeSignature::Coercible(vec![Coercion::new_exact(
|
||||
TypeSignatureClass::Native(logical_string()),
|
||||
)]),
|
||||
],
|
||||
Volatility::Immutable,
|
||||
),
|
||||
|
||||
@@ -15,10 +15,10 @@
|
||||
use std::fmt::Display;
|
||||
|
||||
use datafusion::arrow::datatypes::DataType;
|
||||
use datafusion::logical_expr::ColumnarValue;
|
||||
use datafusion::logical_expr_common::type_coercion::aggregates::{BINARYS, STRINGS};
|
||||
use datafusion::logical_expr::{Coercion, ColumnarValue, TypeSignature, TypeSignatureClass};
|
||||
use datafusion_common::ScalarValue;
|
||||
use datafusion_expr::{ScalarFunctionArgs, Signature, TypeSignature, Volatility};
|
||||
use datafusion_common::types::{logical_binary, logical_string};
|
||||
use datafusion_expr::{ScalarFunctionArgs, Signature, Volatility};
|
||||
use nalgebra::DVectorView;
|
||||
|
||||
use crate::function::Function;
|
||||
@@ -49,9 +49,12 @@ impl Default for ElemProductFunction {
|
||||
Self {
|
||||
signature: Signature::one_of(
|
||||
vec![
|
||||
TypeSignature::Uniform(1, STRINGS.to_vec()),
|
||||
TypeSignature::Uniform(1, BINARYS.to_vec()),
|
||||
TypeSignature::Uniform(1, vec![DataType::BinaryView]),
|
||||
TypeSignature::Coercible(vec![Coercion::new_exact(
|
||||
TypeSignatureClass::Native(logical_binary()),
|
||||
)]),
|
||||
TypeSignature::Coercible(vec![Coercion::new_exact(
|
||||
TypeSignatureClass::Native(logical_string()),
|
||||
)]),
|
||||
],
|
||||
Volatility::Immutable,
|
||||
),
|
||||
|
||||
@@ -15,9 +15,9 @@
|
||||
use std::fmt::Display;
|
||||
|
||||
use datafusion::arrow::datatypes::DataType;
|
||||
use datafusion::logical_expr::ColumnarValue;
|
||||
use datafusion::logical_expr::{Coercion, ColumnarValue, TypeSignatureClass};
|
||||
use datafusion_common::ScalarValue;
|
||||
use datafusion_expr::type_coercion::aggregates::{BINARYS, STRINGS};
|
||||
use datafusion_common::types::{logical_binary, logical_string};
|
||||
use datafusion_expr::{ScalarFunctionArgs, Signature, TypeSignature, Volatility};
|
||||
use nalgebra::DVectorView;
|
||||
|
||||
@@ -36,9 +36,12 @@ impl Default for ElemSumFunction {
|
||||
Self {
|
||||
signature: Signature::one_of(
|
||||
vec![
|
||||
TypeSignature::Uniform(1, STRINGS.to_vec()),
|
||||
TypeSignature::Uniform(1, BINARYS.to_vec()),
|
||||
TypeSignature::Uniform(1, vec![DataType::BinaryView]),
|
||||
TypeSignature::Coercible(vec![Coercion::new_exact(
|
||||
TypeSignatureClass::Native(logical_binary()),
|
||||
)]),
|
||||
TypeSignature::Coercible(vec![Coercion::new_exact(
|
||||
TypeSignatureClass::Native(logical_string()),
|
||||
)]),
|
||||
],
|
||||
Volatility::Immutable,
|
||||
),
|
||||
|
||||
@@ -15,9 +15,9 @@
|
||||
use std::fmt::Display;
|
||||
|
||||
use datafusion::arrow::datatypes::DataType;
|
||||
use datafusion::logical_expr::ColumnarValue;
|
||||
use datafusion::logical_expr_common::type_coercion::aggregates::{BINARYS, STRINGS};
|
||||
use datafusion::logical_expr::{Coercion, ColumnarValue, TypeSignatureClass};
|
||||
use datafusion_common::ScalarValue;
|
||||
use datafusion_common::types::{logical_binary, logical_string};
|
||||
use datafusion_expr::{ScalarFunctionArgs, Signature, TypeSignature, Volatility};
|
||||
|
||||
use crate::function::Function;
|
||||
@@ -49,8 +49,12 @@ impl Default for VectorDimFunction {
|
||||
Self {
|
||||
signature: Signature::one_of(
|
||||
vec![
|
||||
TypeSignature::Uniform(1, STRINGS.to_vec()),
|
||||
TypeSignature::Uniform(1, BINARYS.to_vec()),
|
||||
TypeSignature::Coercible(vec![Coercion::new_exact(
|
||||
TypeSignatureClass::Native(logical_binary()),
|
||||
)]),
|
||||
TypeSignature::Coercible(vec![Coercion::new_exact(
|
||||
TypeSignatureClass::Native(logical_string()),
|
||||
)]),
|
||||
],
|
||||
Volatility::Immutable,
|
||||
),
|
||||
|
||||
@@ -15,9 +15,9 @@
|
||||
use std::fmt::Display;
|
||||
|
||||
use datafusion::arrow::datatypes::DataType;
|
||||
use datafusion::logical_expr::ColumnarValue;
|
||||
use datafusion::logical_expr_common::type_coercion::aggregates::{BINARYS, STRINGS};
|
||||
use datafusion::logical_expr::{Coercion, ColumnarValue, TypeSignatureClass};
|
||||
use datafusion_common::ScalarValue;
|
||||
use datafusion_common::types::{logical_binary, logical_string};
|
||||
use datafusion_expr::{ScalarFunctionArgs, Signature, TypeSignature, Volatility};
|
||||
use nalgebra::DVectorView;
|
||||
|
||||
@@ -52,9 +52,12 @@ impl Default for VectorNormFunction {
|
||||
Self {
|
||||
signature: Signature::one_of(
|
||||
vec![
|
||||
TypeSignature::Uniform(1, STRINGS.to_vec()),
|
||||
TypeSignature::Uniform(1, BINARYS.to_vec()),
|
||||
TypeSignature::Uniform(1, vec![DataType::BinaryView]),
|
||||
TypeSignature::Coercible(vec![Coercion::new_exact(
|
||||
TypeSignatureClass::Native(logical_binary()),
|
||||
)]),
|
||||
TypeSignature::Coercible(vec![Coercion::new_exact(
|
||||
TypeSignatureClass::Native(logical_string()),
|
||||
)]),
|
||||
],
|
||||
Volatility::Immutable,
|
||||
),
|
||||
|
||||
@@ -106,7 +106,8 @@ mod tests {
|
||||
assert!(matches!(f.signature(),
|
||||
datafusion_expr::Signature {
|
||||
type_signature: datafusion_expr::TypeSignature::Uniform(1, valid_types),
|
||||
volatility: datafusion_expr::Volatility::Immutable
|
||||
volatility: datafusion_expr::Volatility::Immutable,
|
||||
..
|
||||
} if valid_types == &vec![ArrowDataType::Utf8]));
|
||||
}
|
||||
|
||||
|
||||
@@ -103,10 +103,11 @@ impl FlightEncoder {
|
||||
FlightMessage::RecordBatch(record_batch) => {
|
||||
let (encoded_dictionaries, encoded_batch) = self
|
||||
.data_gen
|
||||
.encoded_batch(
|
||||
.encode(
|
||||
&record_batch,
|
||||
&mut self.dictionary_tracker,
|
||||
&self.write_options,
|
||||
&mut Default::default(),
|
||||
)
|
||||
.expect("DictionaryTracker configured above to not fail on replacement");
|
||||
|
||||
|
||||
@@ -7,7 +7,6 @@ license.workspace = true
|
||||
[dependencies]
|
||||
common-error.workspace = true
|
||||
common-macro.workspace = true
|
||||
prost.workspace = true
|
||||
snafu.workspace = true
|
||||
tokio.workspace = true
|
||||
|
||||
|
||||
Reference in New Issue
Block a user