Struct LazyView 

pub struct LazyView { /* private fields */ }

A lazy view that captures operations without executing them.

Implementations

impl LazyView

pub fn from_df(df: DataFrame) -> Self

Create from a DataFrame (takes ownership, wraps in Rc).

pub fn from_rc(df: Rc<DataFrame>) -> Self

Create from an Rc.

pub fn filter(self, predicate: DExpr) -> Self

Filter rows by a DExpr predicate.

pub fn select(self, columns: Vec<String>) -> Self

Project to a subset of columns.

pub fn mutate(self, assignments: Vec<(String, DExpr)>) -> Self

Add or replace columns via expressions.

pub fn arrange(self, keys: Vec<ArrangeKey>) -> Self

Sort rows by keys.

pub fn group_summarise( self, group_keys: Vec<String>, aggregations: Vec<(String, TidyAgg)>, ) -> Self

Group by keys and aggregate.

pub fn distinct(self, columns: Vec<String>) -> Self

Keep distinct rows by columns.

pub fn join( self, right: LazyView, on: Vec<(String, String)>, kind: JoinType, ) -> Self

Join with another LazyView.

pub fn collect(self) -> Result<TidyFrame, TidyError>

Optimize and execute the plan, returning a TidyFrame.

pub fn plan(&self) -> &ViewNode

Inspect the plan tree (for testing/debugging).

pub fn optimized_plan(self) -> ViewNode

Consume and return the optimized plan without executing (for testing).

impl LazyView

pub fn collect_batched(self) -> Result<TidyFrame, TidyError>

Optimize and execute the plan using batch processing.

This is an alternative to collect() that processes data in 2048-row batches, fusing chains of streamable operations (Filter, Select, Mutate) into a single pass per batch.

Pipeline breakers (Arrange, GroupSummarise, Distinct, Join) cause full materialization at that point.

The output is identical to collect() – this is purely an execution strategy optimization.