Struct QueryOptimizer 

pub struct QueryOptimizer { /* private fields */ }

Query optimizer

Implementations

impl QueryOptimizer

pub fn new() -> Self

pub fn initialize(&mut self) -> Result<(), StatisticalError>

pub fn optimization_rules(&self) -> &[OptimizationRule]

Returns the list of optimization rules currently registered.

pub fn add_optimization_rule(&mut self, rule: OptimizationRule)

Add an optimization rule if it is not already present.

pub fn has_rule(&self, rule: &OptimizationRule) -> bool

Returns true when the given rule is registered.

pub fn estimate_cost(&self, operation: &QueryOperation) -> CostModel

Estimate the cost of a single query operation based on its type and the amount of data it operates on. Costs are dimensionless weights chosen so that cheaper operations (Filter, Project, Limit) sort before expensive ones (Scan, Join, Aggregate, Sort).

pub fn optimize_with_cost( &mut self, operations: &[QueryOperation], ) -> Result<ExecutionPlan, StatisticalError>

Optimize a sequence of operations by reordering them to minimize total cost. Uses a simple greedy strategy: estimate each operation’s cost and execute cheapest-first. The resulting ExecutionPlan is stored on the optimizer and also returned.

pub fn get_execution_plan(&self) -> Option<&ExecutionPlan>

Returns the most recently optimized execution plan, if any.

pub fn optimize( &self, operations: Vec<QueryOperation>, ) -> Result<QueryPlan, StatisticalError>

Optimize a sequence of operations into a QueryPlan.

Applies three rewrite rules:

1. Predicate pushdown — filters are moved ahead of joins so rows are reduced before the expensive join.
2. Join-type selection — HashJoin is selected when both join inputs are ≥ 1000 rows; NestedLoop when both are < 100; otherwise the caller-supplied join type is retained.
3. Limit-last — Limit is always the final step.

After reordering, per-step cost and output row count are estimated using a simple cost model that tracks the running row count through the plan.