Struct ArrowSheet 

pub struct ArrowSheet {
    pub name: Arc<str>,
    pub date_system: DateSystem,
    pub columns: Vec<ArrowColumn>,
    pub nrows: u32,
    pub chunk_starts: Vec<usize>,
    pub chunk_rows: usize,
}

Fields

name: Arc<str>

date_system: DateSystem

Serial encoding used by every temporal lane in this sheet.

Changing this value requires re-encoding all temporal cells; engine and workbook constructors therefore set it once when the sheet is created.

columns: Vec<ArrowColumn>

nrows: u32

chunk_starts: Vec<usize>

chunk_rows: usize

Preferred chunk size (rows) for capacity growth operations.

For Arrow-ingested sheets this matches the ingest chunk_rows. For sparse/overlay-created sheets this defaults to 32k to avoid creating thousands of tiny chunks during growth.

Implementations

impl ArrowSheet

pub fn new_sparse( sheet_name: &str, ncols: usize, nrows: usize, chunk_rows: usize, ) -> Self

Create a logical sheet whose cells are initially implicit empty values.

Columns start with no materialized chunks; callers can populate only touched column/chunk pairs via set_sparse_overlay_value. Missing chunks remain observable as empty cells through scalar and range reads.

pub fn new_sparse_with_date_system( sheet_name: &str, ncols: usize, nrows: usize, chunk_rows: usize, date_system: DateSystem, ) -> Self

Create a sparse sheet using an explicitly selected workbook date system.

pub fn set_sparse_overlay_value( &mut self, abs_row: usize, abs_col: usize, value: OverlayValue, ) -> isize

Populate a single sparse cell using the overlay cascade while preserving implicit-empty chunks elsewhere. This is intended for sparse initial ingest, not for user edits; no graph/changelog side effects are triggered here.

pub fn shape(&self) -> Vec<ColumnShape>

Return a summary of each column’s chunk counts, total rows, and lane presence.

pub fn range_view( &self, sr: usize, sc: usize, er: usize, ec: usize, ) -> RangeView<'_>

pub fn get_cell_value(&self, abs_row: usize, abs_col: usize) -> LiteralValue

Fast single-cell read (0-based row/col) with overlay precedence.

This avoids constructing a 1x1 RangeView and is intended for tight read loops.

pub fn ensure_row_capacity(&mut self, target_rows: usize)

Ensure capacity to address at least target_rows rows by extending the row chunk map.

This updates chunk_starts/nrows but does not eagerly densify all columns with new empty chunks. Missing chunks are treated as all-empty and can be materialized lazily.

pub fn ensure_column_chunk_mut( &mut self, col_idx: usize, ch_idx: usize, ) -> Option<&mut ColumnChunk>

Ensure a mutable chunk for a given column/chunk index.

If the chunk is beyond the column’s dense chunk vector, it is stored in sparse_chunks.

pub fn chunk_of_row(&self, abs_row: usize) -> Option<(usize, usize)>

Return (chunk_idx, in_chunk_offset) for absolute 0-based row.

pub fn maybe_compact_chunk( &mut self, col_idx: usize, ch_idx: usize, abs_threshold: usize, frac_den: usize, ) -> usize

Heuristic compaction: rebuilds a chunk’s base arrays by applying its overlay when overlay density crosses thresholds. Returns true if a rebuild occurred.

pub fn compact_computed_overlay_chunk( &mut self, col_idx: usize, ch_idx: usize, ) -> usize

Compact a dense chunk’s computed overlay into its base arrays, freeing overlay memory while preserving the data. Returns the number of bytes freed.

This is the computed-overlay counterpart of maybe_compact_chunk (which compacts user-edit overlays). The read cascade is overlay → computed_overlay → base, so folding computed overlay entries into base arrays is transparent: the overlay layer (user edits) is left untouched and still takes precedence on reads.

pub fn compact_computed_overlay_sparse_chunk( &mut self, col_idx: usize, ch_idx: usize, ) -> usize

Compact a sparse chunk’s computed overlay into its base arrays. Equivalent to compact_computed_overlay_chunk but for sparse chunks.

pub fn insert_rows(&mut self, before: usize, count: usize)

Insert count rows before absolute 0-based row before.

pub fn delete_rows(&mut self, start: usize, count: usize)

Delete count rows starting from absolute 0-based row start.

pub fn insert_columns(&mut self, before: usize, count: usize)

Insert count columns before absolute 0-based column before with empty chunks.

pub fn delete_columns(&mut self, start: usize, count: usize)

Delete count columns starting at absolute 0-based column start.

Trait Implementations

impl Clone for ArrowSheet

fn clone(&self) -> ArrowSheet

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ArrowSheet

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

Formats the value using the given formatter.