Struct Interpreter 

pub struct Interpreter {
    pub max_depth: usize,
    pub max_terminals: usize,
    pub seed: u64,
    /* private fields */
}

The CGA Shape Grammar derivation engine.

Rules are registered by name, then derive is called with a root scope and root rule name. The engine expands rules breadth-first until every branch terminates with an I(mesh) terminal.

Stochastic rules with multiple weighted variants use the engine’s seed for reproducible randomness — the same seed always yields the same building.

Fields

max_depth: usize

Hard cap on rule-derivation recursion depth. Defaults to MAX_DEPTH (64). Exceeding it returns ShapeError::DepthLimitExceeded.

max_terminals: usize

Hard cap on the number of terminals a single derivation may emit. Defaults to MAX_TERMINALS (100 000). Exceeding it returns ShapeError::CapacityOverflow.

seed: u64

Seed for stochastic rule selection. Each call to Interpreter::derive constructs a fresh Pcg64 from this seed, so re-running with the same seed produces a bit-identical ShapeModel. Default 0.

Implementations

impl Interpreter

pub fn new() -> Self

pub fn rules(&self) -> &HashMap<String, Vec<WeightedVariant>>

Returns a reference to the full rule table (rule name → weighted variants).

pub fn set_variants( &mut self, name: impl Into<String>, variants: Vec<WeightedVariant>, )

Directly inserts a pre-built variant list for name, bypassing weight validation.

Intended for restoring snapshots produced by crate::genetics::ShapeGenotype::to_interpreter.

pub fn add_rule(&mut self, name: impl Into<String>, ops: Vec<ShapeOp>)

Registers a deterministic production rule.

pub fn add_weighted_rules( &mut self, name: impl Into<String>, variants: Vec<(f64, Vec<ShapeOp>)>, ) -> Result<(), ShapeError>

Registers a stochastic rule with multiple weighted alternatives.

variants is a list of (relative_weight, ops) pairs. Weights need not sum to 1.0 — they are normalised internally during selection.

Returns Err(InvalidNumericValue) if any weight is non-finite or negative.

pub fn has_rule(&self, name: &str) -> bool

Returns true if a rule with name is registered.

pub fn derive( &self, root_scope: Scope, root_rule: impl Into<String>, ) -> Result<ShapeModel, ShapeError>

Derives the shape model starting from root_scope and root_rule.

Uses a breadth-first work queue to expand rules until all branches terminate via I(mesh_id) or an unknown rule name (implicit terminal). A fresh RNG seeded from self.seed is created for each call, making derivations reproducible for the same seed value.

Trait Implementations

impl Default for Interpreter

fn default() -> Self

Returns the “default value” for a type.