Crate csp_solver 

A generalized CSP (Constraint Satisfaction Problem) solver.

This is the sole solver in the workspace. Supports:

• Backtracking search with configurable pruning and variable ordering
• AC-3 (Maintaining Arc Consistency) propagation
• Forward checking and the AC-FC hybrid
• GAC all-different (Régin 1994), default-ON
• Lattice domains for monotonic fixed-point propagation

The optional PyO3 (feature = "py") and wasm bindings expose this same core to Python and JavaScript — they wrap it, they do not mirror a separate implementation.

The crate-root surface is assembled from three internal modules: [config] (solve vocabulary + the Csp<D> container), [csp] (the builder methods), and [csp::solve] (propagation + search dispatch).

Re-exports

pub use builder::assignment::AssignmentBuilder;

pub use builder::assignment::AssignmentError;

pub use builder::assignment::AssignmentSolution;

pub use builder::assignment::SENTINEL;

pub use builder::assignment::assignment;

pub use cancel::CancelToken;

pub use error::CspError;

pub use puzzles::sudoku;

Modules

builder

Fluent builders for common CSP / COP shapes.

cancel

Cooperative cancellation token.

constraint

Constraint trait, built-in implementations, and devirtualized dispatch.

domain

Domain trait and built-in implementations.

error

Unified error family for the solver’s public API surface.

ordering

Variable ordering heuristics.

puzzles

Domain-specific puzzle modules.

solver

Solver dispatch: AC-3 propagation, forward checking, and the unified monomorphized search kernel.

variable

CSP variable with domain tracking and pruned-value undo log.

Structs

Csp

The main CSP solver struct.

SolveConfig

Solve configuration, isomorphic to Python’s CSP constructor arguments.

SolveStats

Solver statistics.

Unsatisfiable

Error type for unsatisfiable constraints.

Enums

OptimizationMode

Optimization mode for the solver.

PropagationStrategy

Propagation strategy for propagate_with().

Pruning

Pruning strategy for backtracking search.