1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223
// Licensed to the Apache Software Foundation (ASF) under one
// or more contributor license agreements. See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership. The ASF licenses this file
// to you under the Apache License, Version 2.0 (the
// "License"); you may not use this file except in compliance
// with the License. You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing,
// software distributed under the License is distributed on an
// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, either express or implied. See the License for the
// specific language governing permissions and limitations
// under the License.
//! Execution plan for writing data to [`DataSink`]s
use super::expressions::PhysicalSortExpr;
use super::{
DisplayAs, DisplayFormatType, ExecutionPlan, Partitioning, SendableRecordBatchStream,
Statistics,
};
use arrow::datatypes::SchemaRef;
use arrow::record_batch::RecordBatch;
use arrow_array::{ArrayRef, UInt64Array};
use arrow_schema::{DataType, Field, Schema};
use async_trait::async_trait;
use core::fmt;
use datafusion_common::Result;
use datafusion_physical_expr::PhysicalSortRequirement;
use futures::StreamExt;
use std::any::Any;
use std::fmt::Debug;
use std::sync::Arc;
use crate::physical_plan::stream::RecordBatchStreamAdapter;
use crate::physical_plan::Distribution;
use datafusion_common::DataFusionError;
use datafusion_execution::TaskContext;
/// `DataSink` implements writing streams of [`RecordBatch`]es to
/// user defined destinations.
///
/// The `Display` impl is used to format the sink for explain plan
/// output.
#[async_trait]
pub trait DataSink: DisplayAs + Debug + Send + Sync {
// TODO add desired input ordering
// How does this sink want its input ordered?
/// Writes the data to the sink, returns the number of values written
///
/// This method will be called exactly once during each DML
/// statement. Thus prior to return, the sink should do any commit
/// or rollback required.
async fn write_all(
&self,
data: SendableRecordBatchStream,
context: &Arc<TaskContext>,
) -> Result<u64>;
}
/// Execution plan for writing record batches to a [`DataSink`]
///
/// Returns a single row with the number of values written
pub struct InsertExec {
/// Input plan that produces the record batches to be written.
input: Arc<dyn ExecutionPlan>,
/// Sink to whic to write
sink: Arc<dyn DataSink>,
/// Schema describing the structure of the data.
schema: SchemaRef,
}
impl fmt::Debug for InsertExec {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
write!(f, "InsertExec schema: {:?}", self.schema)
}
}
impl InsertExec {
/// Create a plan to write to `sink`
pub fn new(input: Arc<dyn ExecutionPlan>, sink: Arc<dyn DataSink>) -> Self {
Self {
input,
sink,
schema: make_count_schema(),
}
}
}
impl ExecutionPlan for InsertExec {
/// Return a reference to Any that can be used for downcasting
fn as_any(&self) -> &dyn Any {
self
}
/// Get the schema for this execution plan
fn schema(&self) -> SchemaRef {
self.schema.clone()
}
fn output_partitioning(&self) -> Partitioning {
Partitioning::UnknownPartitioning(1)
}
fn output_ordering(&self) -> Option<&[PhysicalSortExpr]> {
None
}
fn required_input_distribution(&self) -> Vec<Distribution> {
vec![Distribution::SinglePartition]
}
fn required_input_ordering(&self) -> Vec<Option<Vec<PhysicalSortRequirement>>> {
// Require that the InsertExec gets the data in the order the
// input produced it (otherwise the optimizer may chose to reorder
// the input which could result in unintended / poor UX)
//
// More rationale:
// https://github.com/apache/arrow-datafusion/pull/6354#discussion_r1195284178
vec![self
.input
.output_ordering()
.map(PhysicalSortRequirement::from_sort_exprs)]
}
fn maintains_input_order(&self) -> Vec<bool> {
vec![false]
}
fn children(&self) -> Vec<Arc<dyn ExecutionPlan>> {
vec![self.input.clone()]
}
fn with_new_children(
self: Arc<Self>,
children: Vec<Arc<dyn ExecutionPlan>>,
) -> Result<Arc<dyn ExecutionPlan>> {
Ok(Arc::new(Self {
input: children[0].clone(),
sink: self.sink.clone(),
schema: self.schema.clone(),
}))
}
/// Execute the plan and return a stream of `RecordBatch`es for
/// the specified partition.
fn execute(
&self,
partition: usize,
context: Arc<TaskContext>,
) -> Result<SendableRecordBatchStream> {
if partition != 0 {
return Err(DataFusionError::Internal(
format!("Invalid requested partition {partition}. InsertExec requires a single input partition."
)));
}
// Execute each of our own input's partitions and pass them to the sink
let input_partition_count = self.input.output_partitioning().partition_count();
if input_partition_count != 1 {
return Err(DataFusionError::Internal(format!(
"Invalid input partition count {input_partition_count}. \
InsertExec needs only a single partition."
)));
}
let data = self.input.execute(0, context.clone())?;
let schema = self.schema.clone();
let sink = self.sink.clone();
let stream = futures::stream::once(async move {
sink.write_all(data, &context).await.map(make_count_batch)
})
.boxed();
Ok(Box::pin(RecordBatchStreamAdapter::new(schema, stream)))
}
fn fmt_as(
&self,
t: DisplayFormatType,
f: &mut std::fmt::Formatter,
) -> std::fmt::Result {
match t {
DisplayFormatType::Default | DisplayFormatType::Verbose => {
write!(f, "InsertExec: sink=")?;
self.sink.fmt_as(t, f)
}
}
}
fn statistics(&self) -> Statistics {
Statistics::default()
}
}
/// Create a output record batch with a count
///
/// ```text
/// +-------+,
/// | count |,
/// +-------+,
/// | 6 |,
/// +-------+,
/// ```
fn make_count_batch(count: u64) -> RecordBatch {
let array = Arc::new(UInt64Array::from(vec![count])) as ArrayRef;
RecordBatch::try_from_iter_with_nullable(vec![("count", array, false)]).unwrap()
}
fn make_count_schema() -> SchemaRef {
// define a schema.
Arc::new(Schema::new(vec![Field::new(
"count",
DataType::UInt64,
false,
)]))
}