Struct GroupIndex 

pub struct GroupIndex {
    pub groups: Vec<GroupMeta>,
    pub key_names: Vec<String>,
}

A deterministic group index built from a TidyView.

Groups are created in first-occurrence order: the first time a key combination appears (scanning visible rows in ascending base-row order), a new group entry is appended. This guarantees a stable, reproducible output ordering regardless of hash-map iteration order.

No column buffers are copied during group construction.

Fields

groups: Vec<GroupMeta>

Groups in first-occurrence order.

key_names: Vec<String>

The column names used as group keys (projected names).

Implementations

impl GroupIndex

pub fn build( base: &DataFrame, key_col_indices: &[usize], visible_rows: &[usize], key_names: Vec<String>, ) -> Self

Build a GroupIndex from a materialized set of visible rows.

key_col_indices are indices into base.columns. visible_rows are base-frame row indices in ascending order.

impl GroupIndex

pub fn build_fast( base: &DataFrame, key_col_indices: &[usize], visible_rows: &[usize], key_names: Vec<String>, ) -> Self

Build a GroupIndex using a BTree-accelerated lookup.

Semantics: identical to GroupIndex::build(). First-occurrence group ordering is preserved. The only difference is O(N log G) vs O(N × G).

Trait Implementations

impl Clone for GroupIndex

fn clone(&self) -> GroupIndex

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for GroupIndex

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

Formats the value using the given formatter.