CTTY commented on code in PR #1511: URL: https://github.com/apache/iceberg-rust/pull/1511#discussion_r2220568216
########## crates/integrations/datafusion/src/physical_plan/commit.rs: ########## @@ -0,0 +1,269 @@ +// 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. + +use std::any::Any; +use std::fmt::{Debug, Formatter}; +use std::sync::Arc; + +use datafusion::arrow::array::{Array, ArrayRef, RecordBatch, StringArray, UInt64Array}; +use datafusion::arrow::datatypes::{ + DataType, Field, Schema as ArrowSchema, SchemaRef as ArrowSchemaRef, +}; +use datafusion::common::{DataFusionError, Result as DFResult}; +use datafusion::execution::{SendableRecordBatchStream, TaskContext}; +use datafusion::physical_expr::{EquivalenceProperties, Partitioning}; +use datafusion::physical_plan::execution_plan::{Boundedness, EmissionType}; +use datafusion::physical_plan::stream::RecordBatchStreamAdapter; +use datafusion::physical_plan::{ + DisplayAs, DisplayFormatType, ExecutionPlan, PlanProperties, execute_stream_partitioned, +}; +use futures::StreamExt; +use iceberg::Catalog; +use iceberg::spec::{DataFile, deserialize_data_file_from_json}; +use iceberg::table::Table; +use iceberg::transaction::Transaction; + +use crate::to_datafusion_error; + +/// IcebergCommitExec is responsible for collecting results from multiple IcebergWriteExec +/// instances and using Transaction::fast_append to commit the data files written. +pub(crate) struct IcebergCommitExec { + table: Table, + catalog: Arc<dyn Catalog>, + write_plan: Arc<dyn ExecutionPlan>, + schema: ArrowSchemaRef, + count_schema: ArrowSchemaRef, + plan_properties: PlanProperties, +} + +impl IcebergCommitExec { + pub fn new( + table: Table, + catalog: Arc<dyn Catalog>, + write_plan: Arc<dyn ExecutionPlan>, + schema: ArrowSchemaRef, + ) -> Self { + let plan_properties = Self::compute_properties(schema.clone()); + + Self { + table, + catalog, + write_plan, + schema, + count_schema: Self::make_count_schema(), + plan_properties, + } + } + + // Compute the plan properties for this execution plan + fn compute_properties(schema: ArrowSchemaRef) -> PlanProperties { + PlanProperties::new( + EquivalenceProperties::new(schema), + Partitioning::UnknownPartitioning(1), + EmissionType::Incremental, + Boundedness::Bounded, + ) + } + + // Create a record batch with just the count of rows written + fn make_count_batch(count: u64) -> DFResult<RecordBatch> { + let count_array = Arc::new(UInt64Array::from(vec![count])) as ArrayRef; + + RecordBatch::try_from_iter_with_nullable(vec![("count", count_array, false)]).map_err(|e| { + DataFusionError::ArrowError(e, Some("Failed to make count batch!".to_string())) + }) + } + + fn make_count_schema() -> ArrowSchemaRef { + // Define a schema. + Arc::new(ArrowSchema::new(vec![Field::new( + "count", + DataType::UInt64, + false, + )])) + } +} + +impl Debug for IcebergCommitExec { + fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result { + write!(f, "IcebergCommitExec") + } +} + +impl DisplayAs for IcebergCommitExec { + fn fmt_as(&self, t: DisplayFormatType, f: &mut Formatter) -> std::fmt::Result { + match t { + DisplayFormatType::Default => { + write!(f, "IcebergCommitExec: table={}", self.table.identifier()) + } + DisplayFormatType::Verbose => { + write!( + f, + "IcebergCommitExec: table={}, schema={:?}", + self.table.identifier(), + self.schema + ) + } + DisplayFormatType::TreeRender => { + write!(f, "IcebergCommitExec: table={}", self.table.identifier()) + } + } + } +} + +impl ExecutionPlan for IcebergCommitExec { + fn name(&self) -> &str { + "IcebergCommitExec" + } + + fn as_any(&self) -> &dyn Any { + self + } + + fn properties(&self) -> &PlanProperties { + &self.plan_properties + } + + fn children(&self) -> Vec<&Arc<dyn ExecutionPlan>> { + vec![&self.write_plan] + } + + fn with_new_children( + self: Arc<Self>, + children: Vec<Arc<dyn ExecutionPlan>>, + ) -> DFResult<Arc<dyn ExecutionPlan>> { + if children.len() != 1 { + return Err(DataFusionError::Internal( + "IcebergCommitExec expects exactly one child".to_string(), + )); + } + + Ok(Arc::new(IcebergCommitExec::new( + self.table.clone(), + self.catalog.clone(), + children[0].clone(), + self.schema.clone(), + ))) + } + + fn execute( + &self, + partition: usize, + context: Arc<TaskContext>, + ) -> DFResult<SendableRecordBatchStream> { + // IcebergCommitExec only has one partition (partition 0) + if partition != 0 { + return Err(DataFusionError::Internal(format!( + "IcebergCommitExec only has one partition, but got partition {}", + partition + ))); + } + + let table = self.table.clone(); + let input_plan = self.write_plan.clone(); + let count_schema = Arc::clone(&self.count_schema); + + // todo revisit this + let spec_id = self.table.metadata().default_partition_spec_id(); + let partition_type = self.table.metadata().default_partition_type().clone(); + let current_schema = self.table.metadata().current_schema().clone(); + + let _catalog = Arc::clone(&self.catalog); + + // Process the input streams from all partitions and commit the data files + let stream = futures::stream::once(async move { + let mut data_files: Vec<DataFile> = Vec::new(); + let mut total_record_count: u64 = 0; + + // Execute and collect results from all partitions of the input plan + let batches = execute_stream_partitioned(input_plan, context)?; + + // Collect all data files from this partition's stream + for mut batch_stream in batches { Review Comment: Good point! I've added `CoalescePartititionsExec` in `IcebergTableProvider::insert_into` as a wrapper of `IcebergWriteExec` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org For additional commands, e-mail: issues-h...@iceberg.apache.org
