Struct DataChunk 

pub struct DataChunk { /* private fields */ }

A batch of rows in column-major format.

Columns are ValueVectors — either flat byte buffers from storage or owned TypedValue vecs from operators. A SelectionVector maps logical row indices to physical positions, enabling zero-copy filtering.

Implementations

impl DataChunk

pub fn from_vectors( columns: Vec<ValueVector>, selection: SelectionVector, ) -> Self

Create from ValueVector columns + SelectionVector (scan path).

pub fn new(columns: Vec<Vec<TypedValue>>) -> Self

Create a new DataChunk from owned TypedValue columns (backward compat).

pub fn new_with_row_count( columns: Vec<Vec<TypedValue>>, num_rows: usize, ) -> Self

Create a DataChunk with a specific row count (for zero-column chunks).

pub fn empty(num_columns: usize) -> Self

Create an empty DataChunk with the given number of columns.

pub fn with_capacity(num_columns: usize, row_capacity: usize) -> Self

Create an empty DataChunk with pre-allocated column capacity.

pub fn single_empty_row(num_columns: usize) -> Self

Create a DataChunk with a single row of Null values.

pub fn from_rows(rows: &[Vec<TypedValue>], num_columns: usize) -> Self

Create a DataChunk from rows (row-major → column-major).

pub fn num_columns(&self) -> usize

pub fn num_rows(&self) -> usize

pub fn is_empty(&self) -> bool

pub fn get_value(&self, row_idx: usize, col_idx: usize) -> TypedValue

Get a single value by (logical) row and column index.

pub fn get_row(&self, row_idx: usize) -> Vec<TypedValue>

Get a row as a vector across all columns.

pub fn append_row(&mut self, row: &[TypedValue])

Append a single row. Only works when columns are Owned.

pub fn append(&mut self, other: &DataChunk)

Append all rows from another chunk. Materializes self into Owned columns if needed (e.g., when self has Flat columns from storage).

pub fn selection(&self) -> &SelectionVector

Access the selection vector.

pub fn column(&self, col_idx: usize) -> &ValueVector

Direct access to a column’s ValueVector.

pub fn compact_column(&self, col_idx: usize) -> ValueVector

Extract a column compacted through the selection vector. When the selection is identity, clones the column directly (cheap for FlatVector/Owned). When non-identity, materializes selected values.

pub fn select_columns(self, indices: &[usize]) -> Self

Create a new DataChunk containing only the specified columns (by move).

pub fn take_column(&mut self, col_idx: usize) -> ValueVector

Move a column out of the chunk (replacing it with an empty Owned vec). For identity selection, avoids cloning the underlying data. For non-identity, materializes selected values.

pub fn with_selection(self, selection: SelectionVector) -> Self

Replace selection, keeping columns intact. Consumes self.

pub fn row_ref(&self, row_idx: usize) -> RowRef<'_>

Get a zero-copy row reference for expression evaluation.

pub fn append_row_from_chunk(&mut self, source: &DataChunk, row_idx: usize)

Copy a single row from a source chunk. Target must use Owned columns.

pub fn append_row_from_chunk_with_extra( &mut self, source: &DataChunk, row_idx: usize, extra: TypedValue, )

Copy a row from source chunk and append an extra value as the last column.

Trait Implementations

impl Clone for DataChunk

fn clone(&self) -> DataChunk

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for DataChunk

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

Formats the value using the given formatter.