Struct PositionDeleteFileWriter

pub struct PositionDeleteFileWriter<B: FileWriterBuilder> { /* private fields */ }

Writer used to write position delete files.

Position delete files mark specific rows in data files as deleted by their position (row number). The schema is fixed with two columns:

• file_path: The path to the data file
• pos: The row position (0-indexed) in that file

Example

use std::sync::Arc;
use arrow_array::{RecordBatch, StringArray, Int64Array};
use iceberg::spec::DataFileFormat;
use iceberg::writer::base_writer::position_delete_writer::{
    PositionDeleteFileWriterBuilder, PositionDeleteWriterConfig
};
use iceberg::writer::file_writer::ParquetWriterBuilder;
use iceberg::writer::IcebergWriterBuilder;

// Create position delete writer
let config = PositionDeleteWriterConfig::new(None, 0, None);
let mut writer = PositionDeleteFileWriterBuilder::new(parquet_builder, config)
    .build()
    .await?;

// Write delete records
let file_paths = StringArray::from(vec!["s3://bucket/data/file1.parquet"]);
let positions = Int64Array::from(vec![42]); // Delete row 42
let batch = RecordBatch::try_new(
    PositionDeleteWriterConfig::arrow_schema(),
    vec![Arc::new(file_paths), Arc::new(positions)],
)?;

writer.write(batch).await?;
let data_files = writer.close().await?;

Trait Implementations

impl<B: Debug + FileWriterBuilder> Debug for PositionDeleteFileWriter<B>

where B::R: Debug,

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<B: FileWriterBuilder> IcebergWriter for PositionDeleteFileWriter<B>

fn write<'life0, 'async_trait>( &'life0 mut self, batch: RecordBatch, ) -> Pin<Box<dyn Future<Output = Result<()>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Write data to iceberg table.

fn close<'life0, 'async_trait>( &'life0 mut self, ) -> Pin<Box<dyn Future<Output = Result<Vec<DataFile>>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Close the writer and return the written data files. If close failed, the data written before maybe be lost. User may need to recreate the writer and rewrite the data again.