Struct ParquetAsyncReader

pub struct ParquetAsyncReader { /* private fields */ }

Available on crate features parquet and cloud only.

A Parquet reader on top of the async object_store API. Only the batch reader is implemented since parquet files on cloud storage tend to be big and slow to access.

Implementations

impl ParquetAsyncReader

pub async fn from_uri( uri: &str, cloud_options: Option<&CloudOptions>, metadata: Option<FileMetadataRef>, ) -> PolarsResult<ParquetAsyncReader>

pub async fn with_arrow_schema_projection( self, first_schema: &Arc<ArrowSchema>, projected_arrow_schema: Option<&ArrowSchema>, allow_missing_columns: bool, ) -> PolarsResult<Self>

pub async fn schema(&mut self) -> PolarsResult<ArrowSchemaRef>

pub async fn num_rows(&mut self) -> PolarsResult<usize>

pub fn with_slice(self, slice: Option<(usize, usize)>) -> Self

Only positive offsets are supported for simplicity - the caller should translate negative offsets into the positive equivalent.

pub fn with_row_index(self, row_index: Option<RowIndex>) -> Self

pub fn set_rechunk(self, rechunk: bool) -> Self

pub fn with_projection(self, projection: Option<Vec<usize>>) -> Self

pub fn with_predicate(self, predicate: Option<Arc<dyn PhysicalIoExpr>>) -> Self

pub fn use_statistics(self, toggle: bool) -> Self

Use statistics in the parquet to determine if pages can be skipped from reading.

pub fn with_hive_partition_columns(self, columns: Option<Vec<Series>>) -> Self

pub fn with_include_file_path( self, include_file_path: Option<(PlSmallStr, Arc<str>)>, ) -> Self

pub fn read_parallel(self, parallel: ParallelStrategy) -> Self

pub async fn batched( self, chunk_size: usize, ) -> PolarsResult<BatchedParquetReader>

pub async fn get_metadata(&mut self) -> PolarsResult<&FileMetadataRef>

pub async fn finish(self) -> PolarsResult<DataFrame>