Changeset - 2810f86816cd
[Not reviewed]
default
0 2 0
Laman - 10 months ago 2024-06-29 00:16:42

added an explicit fail state, removed its special handling from the code
2 files changed with 25 insertions and 11 deletions:
0 comments (0 inline, 0 general)
regexp.py
Show inline comments
 
@@ -272,12 +272,16 @@ class Regexp:
 
					index[target_tup] = new_target
 
					compact_rules.extend([-1] * n)
 
					stack.append(target_tup)
 
				compact_rules[index[multistate]*n + alphabet_index[c]] = index[target_tup]
 
				if any(st in self.end_states for st in target_set):
 
					end_states.add(index[target_tup])
 

	
 
		fail = len(index)
 
		compact_rules = [(st if st >= 0 else fail) for st in compact_rules]
 
		compact_rules.extend([fail] * n)
 
		
 
		return (compact_rules, end_states, alphabet_index)
 

	
 

	
 
class RegexpDFA:
 
	def __init__(self, rules, end_states, alphabet_index):
 
@@ -292,15 +296,16 @@ class RegexpDFA:
 

	
 
		return cls(rules, end_states, alphabet_index)
 

	
 
	def match(self, s):
 
		st = 0
 
		n = len(self.alphabet_index)
 
		fail = len(self.rules) // n
 

	
 
		for c in s:
 
			if c not in self.alphabet_index or st < 0:
 
			if c not in self.alphabet_index or st == fail:
 
				return False
 
			key = (st*n + self.alphabet_index[c])
 
			st = self.rules[key]
 

	
 
		return st in self.end_states
 

	
 
@@ -309,23 +314,23 @@ class RegexpDFA:
 
		(rules, end_states) = self._collapse_states(equivalents)
 

	
 
		return RegexpDFA(rules, end_states, self.alphabet_index)
 

	
 
	def normalize(self):
 
		n = len(self.alphabet_index)
 
		index = {-1: -1, 0: 0}
 
		index = {0: 0}
 
		queue = deque([0])
 

	
 
		rules = []
 

	
 
		while queue:
 
			si = queue.popleft()
 
			row = self.rules[si*n:(si+1)*n]
 
			for sj in row:
 
				if sj not in index:
 
					index[sj] = len(index)-1
 
					index[sj] = len(index)
 
					queue.append(sj)
 
			rules.extend(index[sj] for sj in row)
 
		
 
		end_states = {index[si] for si in self.end_states}
 

	
 
		return RegexpDFA(rules, end_states, self.alphabet_index)
 
@@ -355,13 +360,13 @@ class RegexpDFA:
 
		rules = []
 

	
 
		eq_mapping = dict()
 
		for (s1, s2) in equivalents:
 
			eq_mapping[s2] = min(s1, eq_mapping.get(s2, math.inf))
 

	
 
		discard_mapping = {-1: -1}
 
		discard_mapping = dict()
 
		discard_count = 0
 

	
 
		for i in range(0, len(self.rules), n):
 
			si = i//n
 
			if si in eq_mapping:
 
				discard_count += 1
src/regexp.rs
Show inline comments
 
@@ -3,13 +3,12 @@ use std::{collections::{HashMap, HashSet
 
mod token;
 
pub use token::ParsingError;
 
use token::parse;
 

	
 
const START_NFA: usize = usize::MAX;
 
const START_DFA: usize = 0;
 
const FAIL: usize = usize::MAX>>1;
 

	
 
fn encode_set(set: &HashSet<usize>) -> String {
 
	let mut v = Vec::from_iter(set.iter());
 
	v.sort();
 
	let res: Vec<String> = v.into_iter().map(|x| x.to_string()).collect();
 
	return res.join(",");
 
@@ -77,12 +76,13 @@ impl Regexp {
 
		}
 

	
 
		return multistate.iter().any(|x| self.end_states.contains(x));
 
	}
 

	
 
	pub fn determinize(&self) -> RegexpDFA {
 
		const FAIL: usize = usize::MAX;
 
		let alphabet_index: HashMap<char, usize> = self.alphabet.iter().enumerate().map(|(i, c)| (*c, i)).collect();
 
		let n = alphabet_index.len();
 
		let mut compact_rules = vec![FAIL; n];
 
		let mut end_states: HashSet<usize> = HashSet::new();
 
		if self.end_states.contains(&START_NFA) {end_states.insert(START_DFA);}
 

	
 
@@ -121,12 +121,16 @@ impl Regexp {
 
				if is_end {
 
					end_states.insert(index_new[&target_hash]);
 
				}
 
			}
 
		}
 

	
 
		let fail = index_new.len();
 
		compact_rules = compact_rules.into_iter().map(|st| if st != FAIL {st} else {fail}).collect();
 
		compact_rules.extend(iter::repeat(fail).take(n));
 

	
 
		return RegexpDFA{rules: compact_rules, end_states, alphabet_index};
 
	}
 
}
 

	
 
pub struct RegexpDFA {
 
	rules: Vec<usize>,
 
@@ -134,21 +138,25 @@ pub struct RegexpDFA {
 
	alphabet_index: HashMap<char, usize>
 
}
 

	
 
impl RegexpDFA {
 
	pub fn eval(&self, s: String) -> bool {
 
		let n = self.alphabet_index.len();
 
		if n == 0 {
 
			return s.len() == 0;
 
		}
 
		let fail = self.rules.len() / n;
 
		let mut state = START_DFA;
 

	
 
		for c in s.chars() {
 
			if let Some(ci) = self.alphabet_index.get(&c) {
 
				state = self.rules[state*n + ci];
 
			} else {
 
				return false;
 
			}
 
			if state == FAIL {
 
			if state == fail {
 
				return false;
 
			}
 
		}
 

	
 
		return self.end_states.contains(&state);
 
	}
 
@@ -164,30 +172,31 @@ impl RegexpDFA {
 
	pub fn normalize(&self) -> RegexpDFA {
 
		let n = self.alphabet_index.len();
 
		if n == 0 {
 
			return RegexpDFA{rules: self.rules.clone(), end_states: self.end_states.clone(), alphabet_index: self.alphabet_index.clone()}; 
 
		}
 
		let m = self.rules.len()/n;
 
		let mut index: Vec<usize> = vec![FAIL;m];
 
		let fail = m;
 
		let mut index: Vec<usize> = vec![fail;m];
 
		index[0] = 0;
 
		let mut queue = VecDeque::from([START_DFA]);
 

	
 
		let mut rules = vec![];
 
		let mut k = 1;
 

	
 
		while !queue.is_empty() {
 
			let si = queue.pop_front().unwrap();
 
			let row = &self.rules[si*n..(si+1)*n];
 
			for &sj in row {
 
				if sj != FAIL && index[sj] == FAIL {
 
				if sj != fail && index[sj] == fail {
 
					index[sj] = k;
 
					k += 1;
 
					queue.push_back(sj);
 
				}
 
			}
 
			rules.extend(row.iter().map(|&st| if st != FAIL {index[st]} else {FAIL}));
 
			rules.extend(row.iter().map(|&st| index[st]));
 
		}
 

	
 
		let end_states = self.end_states.iter().map(|st| index[*st]).collect();
 
		
 
		return RegexpDFA{rules, end_states, alphabet_index: self.alphabet_index.clone()};
 
	}
 
@@ -236,15 +245,15 @@ impl RegexpDFA {
 
		for si in 0..m {
 
			if eq_mapping[si] != si {
 
				discard_count += 1;
 
				continue;
 
			}
 
			discard_mapping[si] = si-discard_count;
 
			rules.extend(self.rules[si*n..(si+1)*n].iter().map(|&st| if st!=FAIL {eq_mapping[st]} else {FAIL}));
 
			rules.extend(self.rules[si*n..(si+1)*n].iter().map(|&st| eq_mapping[st]));
 
		}
 

	
 
		rules = rules.into_iter().map(|st| if st!=FAIL {discard_mapping[st]} else {FAIL}).collect();
 
		rules = rules.into_iter().map(|st| discard_mapping[st]).collect();
 
		let end_states = self.end_states.iter().map(|st| discard_mapping[eq_mapping[*st]]).collect();
 

	
 
		return RegexpDFA{rules, end_states, alphabet_index: self.alphabet_index.clone()};
 
	}
 
}
0 comments (0 inline, 0 general)