1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330
//! Track value bindings and constraints during the evaluation process. //! //! State is the imperative core of each logic program. It manages the updates //! to the relationships between values while delegating the actual storage to a //! type specific [`Domain`](crate::domains). //! //! In general, it is preferred to deal with State indirectly through //! [goals](crate::goal). They are essentially equivalent in capability, and //! their declarative, higher level nature makes them much easier to use. //! Notably, goal functions provide automatic [value](crate::value) wrapping //! through [`IntoVal`](crate::value::IntoVal). //! //! An open [State] is the initial struct that you will start with (explicitly //! or implicitly through a [goal](crate::goal)). Iterating through the //! potentially results will yield zero or more //! [`ResolvedStates`](ResolvedState). pub mod constraints; mod impls; mod iter_resolved; mod resolved; use super::util::multikeymultivaluemap::MKMVMap; use crate::domains::{Domain, DomainType}; use crate::unify::UnifyIn; use crate::value::{ LVarId, Val, Val::{Resolved, Var}, }; #[doc(hidden)] pub use constraints::Constraint; pub use iter_resolved::{IterResolved, ResolvedStateIter}; pub use resolved::ResolvedState; use std::fmt::Debug; use std::iter::once; use std::rc::Rc; /// Type alias for an [`Iterator`] of [`States`](crate::state::State) pub type StateIter<'s, D> = Box<dyn Iterator<Item = State<'s, D>> + 's>; type ConstraintFns<'s, D> = MKMVMap<LVarId, Rc<dyn Constraint<'s, D> + 's>>; /// The core struct used to contain and manage [value](crate::value) bindings. /// /// An open [State] can be updated in a few different ways. Most update methods /// return an `Option<State<D>>` to reflect the fact each new constraint /// invalidate the state. This gives you the ability to quickly short circuit as /// soon the state hits a dead end. /// /// In general, it is most ergonomic to manipulate a state inside a function /// that returns an `Option<State<D>>` to allow the use of the question mark /// operator (Note that the [`.apply()`](State::apply()) function makes it easy /// to do this). /// /// ``` /// use canrun::{State, val, var}; /// use canrun::domains::example::I32; /// /// fn my_fn<'a>() -> Option<State<'a, I32>> { /// let x = var(); /// let y = var(); /// let state: State<I32> = State::new(); /// let maybe: Option<State<I32>> = state.unify(&val!(x), &val!(1)); /// maybe?.unify(&val!(x), &val!(y)) /// } /// assert!(my_fn().is_some()); /// ``` #[derive(Clone)] pub struct State<'a, D: Domain<'a> + 'a> { domain: D, constraints: ConstraintFns<'a, D>, forks: im_rc::Vector<Rc<dyn Fork<'a, D> + 'a>>, } impl<'a, D: Domain<'a> + 'a> State<'a, D> { /// Create a new, empty state. /// /// This often does not need to be used directly as you can /// [`.query()`](crate::goal::Goal::query()) a [`Goal`](crate::goal::Goal) /// directly, which handles the state creation internally. /// /// However, there are use cases for creating and managing a state /// independently of any goals. /// /// # Example: /// ``` /// use canrun::{State, var}; /// use canrun::domains::example::I32; /// /// let state: State<I32> = State::new(); /// ``` pub fn new() -> Self { State { domain: D::new(), constraints: MKMVMap::new(), forks: im_rc::Vector::new(), } } /// Apply an arbitrary function to a state. /// /// This is primarily a helper to make it easier to get into a function /// where you can use the question mark operator while applying multiple /// updates to a state. /// /// # Example: /// ``` /// use canrun::{State, Query, val, var}; /// use canrun::domains::example::I32; /// /// let s: State<I32> = State::new(); /// let x = var(); /// let s = s.apply(|s| { /// s.unify(&val!(x), &val!(1))? /// .unify(&val!(1), &val!(x)) /// }); /// let results: Vec<i32> = s.query(x).collect(); /// assert_eq!(results, vec![1]); /// ``` pub fn apply<F>(self, func: F) -> Option<Self> where F: Fn(Self) -> Option<Self>, { func(self) } fn iter_forks(mut self) -> StateIter<'a, D> { let fork = self.forks.pop_front(); match fork { None => Box::new(once(self)), Some(fork) => Box::new(fork.fork(self).flat_map(State::iter_forks)), } } /// Recursively resolve a [`Val`](crate::value::Val) as far as the currently /// known variable bindings allow. /// /// This will return either the final [`Val::Resolved`] (if found) or the /// last [`Val::Var`] it attempted to resolve. It will not force /// [`forks`](State::fork()) to enumerate, so potential bindings are not /// considered. /// /// # Example: /// ``` /// use canrun::{State, Query, val, var}; /// use canrun::domains::example::I32; /// /// # fn test() -> Option<()> { /// let state: State<I32> = State::new(); /// /// let x = val!(var()); /// assert_eq!(state.resolve_val(&x), &x); /// /// let state = state.unify(&x, &val!(1))?; /// assert_eq!(state.resolve_val(&x), &val!(1)); /// # Some(()) /// # } /// # test(); /// ``` pub fn resolve_val<'r, T>(&'r self, val: &'r Val<T>) -> &'r Val<T> where T: Debug, D: DomainType<'a, T>, { self.domain.resolve(val) } /// Attempt to [unify](module@crate::unify) two values with each other. /// /// If the unification fails, [`None`](std::option::Option::None) will be /// returned. [`Val::Var`]s will be checked against relevant /// [constraints](State::constrain), which can also cause a state to fail. /// /// # Examples: /// ``` /// use canrun::{State, Query, val, var}; /// use canrun::domains::example::I32; /// /// let x = val!(var()); /// /// let state: State<I32> = State::new(); /// let state = state.unify(&x, &val!(1)); /// assert!(state.is_some()); /// ``` /// ``` /// # use canrun::{State, Query, val}; /// # use canrun::domains::example::I32; /// let state: State<I32> = State::new(); /// let state = state.unify(&val!(1), &val!(2)); /// assert!(state.is_none()); /// ``` pub fn unify<T>(mut self, a: &Val<T>, b: &Val<T>) -> Option<Self> where T: UnifyIn<'a, D> + Debug, D: DomainType<'a, T>, { let a = self.resolve_val(a); let b = self.resolve_val(b); match (a, b) { (Resolved(a), Resolved(b)) => { let a = a.clone(); let b = b.clone(); UnifyIn::unify_resolved(self, a, b) } (Var(a), Var(b)) if a == b => Some(self), (Var(var), val) | (val, Var(var)) => { let key = *var; let value = val.clone(); // TODO: Add occurs check? self.domain.update(key, value); // check constraints matching newly assigned lvar if let Some(constraints) = self.constraints.extract(&key.id) { constraints .into_iter() .try_fold(self, |state, func| state.constrain(func)) } else { Some(self) } } } } /// Add a constraint to the store that can be reevaluated as variables are /// resolved. /// /// Some logic is not easy or even possible to express until the resolved /// values are available. `.constrain()` provides a low level way to run /// custom imperative code whenever certain bindings are updated. /// /// See the [`Constraint` trait](constraints::Constraint) for more /// information. pub fn constrain(mut self, constraint: Rc<dyn Constraint<'a, D> + 'a>) -> Option<Self> { match constraint.attempt(&self) { Ok(resolve) => resolve(self), Err(watch) => { self.constraints.add(watch.0, constraint); Some(self) } } } /// Add a potential fork point to the state. /// /// If there are many possibilities for a certain value or set of values, /// this method allows you to add a [`Fork`] object that can enumerate those /// possible alternate states. /// /// While this is not quite as finicky as the /// [`Constraints`](State::constrain()), you still probably want to use the /// [`any`](crate::goal::any!) or [`either`](crate::goal::either()) goals. /// /// [Unification](State::unify()) is performed eagerly as soon as it is /// called. [Constraints](State::constrain()) are run as variables are /// resolved. Forking is executed lazily at the end, when /// [`.iter_resolved()`](crate::state::IterResolved::iter_resolved()) (or /// [`.query()](crate::query::Query::query())) is called. pub fn fork(mut self, fork: Rc<dyn Fork<'a, D> + 'a>) -> Option<Self> { self.forks.push_back(fork); Some(self) } } /// Fork a [`State`] into zero or more alternate states. /// /// Added to a [`State`] with [`.fork()`](crate::state::State::fork()). /// /// # Example: /// ``` /// use canrun::{val, var, Fork, Query, State, StateIter, Val}; /// use canrun::domains::example::I32; /// use std::rc::Rc; /// /// #[derive(Debug)] /// struct Is1or2 { /// x: Val<i32>, /// } /// /// impl<'a> Fork<'a, I32> for Is1or2 { /// fn fork(&self, state: State<'a, I32>) -> StateIter<'a, I32> { /// let s1 = state.clone().unify(&self.x, &val!(1)); /// let s2 = state.unify(&self.x, &val!(2)); /// Box::new(s1.into_iter().chain(s2.into_iter())) /// } /// } /// /// # fn main() { /// let x = var(); /// let state: State<I32> = State::new(); /// let state = state.fork(Rc::new(Is1or2 { x: val!(x) })); /// let results: Vec<i32> = state.query(x).collect(); /// assert_eq!(results, vec![1, 2]); /// # } /// ``` pub trait Fork<'a, D: Domain<'a>>: Debug { /// Given a [`State`], return an iterator of states that result from the /// fork operation. fn fork(&self, state: State<'a, D>) -> StateIter<'a, D>; } #[cfg(test)] mod test { use crate::domains::example::I32; use crate::{val, var, Fork, Query, State, StateIter, Val}; use std::rc::Rc; #[derive(Debug)] struct Is1or2 { x: Val<i32>, } impl<'a> Fork<'a, I32> for Is1or2 { fn fork(&self, state: State<'a, I32>) -> StateIter<'a, I32> { let s1 = state.clone().unify(&self.x, &val!(1)); let s2 = state.unify(&self.x, &val!(2)); Box::new(s1.into_iter().chain(s2.into_iter())) } } #[test] fn doctest() { let x = var(); let state: State<I32> = State::new(); let state = state.fork(Rc::new(Is1or2 { x: val!(x) })); let results: Vec<i32> = state.query(x).collect(); assert_eq!(results, vec![1, 2]); } }