rexact-core 0.2.0

//! Limits manager (architecture section 6, RULE-08, FR-G05).
//!
//! Contract: one [`Limits`] value is passed explicitly to every computation;
//! a [`Budget`] tracks consumption and records *which* limit stopped a run.
//! Invariants: counters use checked arithmetic and saturate into a limit error,
//! never wrap (NFR-C05). Failure mode: every breach maps to a catalogue code
//! whose exit class is "unknown" (2) or a usage error (3).

use rexact_syntax::error::ErrorCode;
use std::time::Duration;

/// Monotonic clock that is inert on WebAssembly (where `Instant` panics):
/// there the time limit is not enforced and other limits still apply.
#[derive(Debug, Clone, Copy)]
pub struct Clock {
    #[cfg(not(target_arch = "wasm32"))]
    start: std::time::Instant,
}

impl Clock {
    /// Start now.
    pub fn now() -> Self {
        Clock {
            #[cfg(not(target_arch = "wasm32"))]
            start: std::time::Instant::now(),
        }
    }
    /// Time since start (always zero on WebAssembly).
    pub fn elapsed(&self) -> Duration {
        #[cfg(not(target_arch = "wasm32"))]
        {
            self.start.elapsed()
        }
        #[cfg(target_arch = "wasm32")]
        {
            Duration::ZERO
        }
    }
}

/// Configured limits. All fields are public so they can be reported verbatim
/// in results.
#[derive(Debug, Clone, PartialEq, Eq)]
pub struct Limits {
    /// Maximum automaton / product states explored (`RX-L001`).
    pub states: u64,
    /// Wall-clock budget (`RX-L002`).
    pub time: Duration,
    /// Estimated memory ceiling in bytes (`RX-L003`).
    pub memory_bytes: u64,
    /// Maximum counted-repetition expansion (`RX-L004`).
    pub repeat: u64,
    /// Maximum pattern length in Unicode scalar values (`RX-L005`).
    pub length: u64,
    /// Maximum nesting depth (`RX-L006`).
    pub depth: u32,
    /// Maximum enumerated items for witnesses / synthesis (`RX-L007`).
    pub enumerate: u64,
}

impl Default for Limits {
    fn default() -> Self {
        Self {
            states: 200_000,
            time: Duration::from_millis(5_000),
            memory_bytes: 1024 * 1024 * 1024,
            repeat: 10_000,
            length: 1_000_000,
            depth: 250,
            enumerate: 100_000,
        }
    }
}

/// Tracks consumption against [`Limits`] and remembers the first breach.
#[derive(Debug)]
pub struct Budget {
    limits: Limits,
    start: Clock,
    states: u64,
    memory: u64,
    hit: Option<ErrorCode>,
}

impl Budget {
    /// Start a budget; the clock starts now.
    pub fn new(limits: Limits) -> Self {
        Self {
            limits,
            start: Clock::now(),
            states: 0,
            memory: 0,
            hit: None,
        }
    }

    /// The configured limits (reported in results).
    pub fn limits(&self) -> &Limits {
        &self.limits
    }

    /// The first limit that stopped the run, if any (`--trace-limits`).
    pub fn limit_hit(&self) -> Option<ErrorCode> {
        self.hit
    }

    fn trip(&mut self, code: ErrorCode) -> Result<(), ErrorCode> {
        // Keep the first breach: later ones are consequences of it.
        self.hit.get_or_insert(code);
        Err(code)
    }

    /// Account for `n` new states. Overflow counts as a breach.
    pub fn add_states(&mut self, n: u64) -> Result<(), ErrorCode> {
        match self.states.checked_add(n) {
            Some(total) if total <= self.limits.states => {
                self.states = total;
                Ok(())
            }
            _ => self.trip(ErrorCode::L001),
        }
    }

    /// Account for `bytes` of estimated memory.
    pub fn add_memory(&mut self, bytes: u64) -> Result<(), ErrorCode> {
        match self.memory.checked_add(bytes) {
            Some(total) if total <= self.limits.memory_bytes => {
                self.memory = total;
                Ok(())
            }
            _ => self.trip(ErrorCode::L003),
        }
    }

    /// Cooperative time check; call at loop boundaries.
    pub fn check_time(&mut self) -> Result<(), ErrorCode> {
        if self.start.elapsed() > self.limits.time {
            self.trip(ErrorCode::L002)
        } else {
            Ok(())
        }
    }

    /// Check a counted-repetition bound before expanding it.
    pub fn check_repeat(&mut self, count: u64) -> Result<(), ErrorCode> {
        if count > self.limits.repeat {
            self.trip(ErrorCode::L004)
        } else {
            Ok(())
        }
    }

    /// Check pattern length.
    pub fn check_length(&mut self, len: u64) -> Result<(), ErrorCode> {
        if len > self.limits.length {
            self.trip(ErrorCode::L005)
        } else {
            Ok(())
        }
    }

    /// Check nesting depth.
    pub fn check_depth(&mut self, depth: u32) -> Result<(), ErrorCode> {
        if depth > self.limits.depth {
            self.trip(ErrorCode::L006)
        } else {
            Ok(())
        }
    }

    /// Check an enumeration count.
    pub fn check_enumerate(&mut self, count: u64) -> Result<(), ErrorCode> {
        if count > self.limits.enumerate {
            self.trip(ErrorCode::L007)
        } else {
            Ok(())
        }
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    fn small() -> Limits {
        Limits {
            states: 10,
            repeat: 5,
            length: 8,
            depth: 3,
            enumerate: 4,
            ..Limits::default()
        }
    }

    #[test]
    fn states_breach_reports_l001_and_remembers_it() {
        let mut b = Budget::new(small());
        assert!(b.add_states(10).is_ok());
        assert_eq!(b.add_states(1), Err(ErrorCode::L001));
        assert_eq!(b.limit_hit(), Some(ErrorCode::L001));
    }

    #[test]
    fn first_breach_wins() {
        let mut b = Budget::new(small());
        let _ = b.check_depth(99);
        let _ = b.check_repeat(99);
        assert_eq!(b.limit_hit(), Some(ErrorCode::L006));
    }

    #[test]
    fn overflow_is_a_breach_not_a_wrap() {
        let mut b = Budget::new(Limits {
            states: u64::MAX,
            ..small()
        });
        assert!(b.add_states(u64::MAX).is_ok());
        assert_eq!(b.add_states(1), Err(ErrorCode::L001));
        let mut m = Budget::new(Limits {
            memory_bytes: u64::MAX,
            ..small()
        });
        assert!(m.add_memory(u64::MAX).is_ok());
        assert_eq!(m.add_memory(1), Err(ErrorCode::L003));
    }

    #[test]
    fn boundaries_are_inclusive() {
        let mut b = Budget::new(small());
        assert!(b.check_repeat(5).is_ok());
        assert_eq!(b.check_repeat(6), Err(ErrorCode::L004));
        let mut b = Budget::new(small());
        assert!(b.check_length(8).is_ok());
        assert_eq!(b.check_length(9), Err(ErrorCode::L005));
        let mut b = Budget::new(small());
        assert!(b.check_depth(3).is_ok());
        assert_eq!(b.check_depth(4), Err(ErrorCode::L006));
        let mut b = Budget::new(small());
        assert!(b.check_enumerate(4).is_ok());
        assert_eq!(b.check_enumerate(5), Err(ErrorCode::L007));
    }

    #[test]
    fn time_limit_trips_l002() {
        let mut b = Budget::new(Limits {
            time: Duration::ZERO,
            ..small()
        });
        std::thread::sleep(Duration::from_millis(2));
        assert_eq!(b.check_time(), Err(ErrorCode::L002));
    }

    #[test]
    fn memory_breach_reports_l003() {
        let mut b = Budget::new(Limits {
            memory_bytes: 100,
            ..small()
        });
        assert!(b.add_memory(100).is_ok());
        assert_eq!(b.add_memory(1), Err(ErrorCode::L003));
    }
}