Store all states visited in a single run of 'addStateToList()' in a slice

regex/compile.go

@@ -816,7 +816,7 @@ func thompson(re []postfixNode) (Reg, error) {
 	// In these cases, we will return an NFA with 1 state, with an assertion that is always true.
 	if len(re) == 0 {
 		start := zeroLengthMatchState()
-		nfa = append(nfa, &start)
+		nfa = append(nfa, start)
 	}
 
 	for _, c := range re {
@@ -1068,14 +1068,14 @@ func thompson(re []postfixNode) (Reg, error) {
 					nfa = append(nfa, s2)
 				}
 				tmp := zeroLengthMatchState()
-				s2 = &tmp
+				s2 = tmp
 			}
 			if err1 != nil || (s1.groupBegin && s1.numTransitions() == 0) { // Doesn't exist, or its just an LPAREN
 				if err1 == nil { // See above for explanation
 					nfa = append(nfa, s1)
 				}
 				tmp := zeroLengthMatchState()
-				s1 = &tmp
+				s1 = tmp
 			}
 			s3 := alternate(s1, s2)
 			nfa = append(nfa, s3)

regex/matching.go

@@ -251,38 +251,40 @@ func (regex Reg) FindAllSubmatch(str string) []Match {
 	return indices
 }
 
-func addStateToList(str []rune, idx int, list []nfaState, state nfaState, threadGroups []Group) []nfaState {
-	if stateExists(list, state) {
+func addStateToList(str []rune, idx int, list []nfaState, state nfaState, threadGroups []Group, visited []nfaState) []nfaState {
+	if stateExists(list, state) || stateExists(visited, state) {
 		return list
 	}
+	visited = append(visited, state)
+
 	if state.isKleene || state.isQuestion {
 		copyThread(state.splitState, state)
-		list = addStateToList(str, idx, list, *state.splitState, threadGroups)
+		list = addStateToList(str, idx, list, *state.splitState, threadGroups, visited)
 		copyThread(state.next, state)
-		list = addStateToList(str, idx, list, *state.next, threadGroups)
+		list = addStateToList(str, idx, list, *state.next, threadGroups, visited)
 		return list
 	}
 	if state.isAlternation {
 		copyThread(state.next, state)
-		list = addStateToList(str, idx, list, *state.next, threadGroups)
+		list = addStateToList(str, idx, list, *state.next, threadGroups, visited)
 		copyThread(state.splitState, state)
-		list = addStateToList(str, idx, list, *state.splitState, threadGroups)
+		list = addStateToList(str, idx, list, *state.splitState, threadGroups, visited)
 		return list
 	}
 	state.threadGroups = append([]Group{}, threadGroups...)
 	if state.assert != noneAssert {
 		if state.checkAssertion(str, idx) {
 			copyThread(state.next, state)
-			return append(list, addStateToList(str, idx, list, *state.next, state.threadGroups)...)
+			return addStateToList(str, idx, list, *state.next, state.threadGroups, visited)
 		}
 	}
 	if state.groupBegin {
 		state.threadGroups[state.groupNum].StartIdx = idx
-		return append(list, addStateToList(str, idx, list, *state.next, state.threadGroups)...)
+		return addStateToList(str, idx, list, *state.next, state.threadGroups, visited)
 	}
 	if state.groupEnd {
 		state.threadGroups[state.groupNum].EndIdx = idx
-		return append(list, addStateToList(str, idx, list, *state.next, state.threadGroups)...)
+		return addStateToList(str, idx, list, *state.next, state.threadGroups, visited)
 	}
 	return append(list, state)
 
@@ -344,7 +346,7 @@ func findAllSubmatchHelper(start *nfaState, str []rune, offset int, numGroups in
 
 	start.threadGroups = newMatch(numGroups + 1)
 	start.threadGroups[0].StartIdx = i
-	currentStates = addStateToList(str, i, currentStates, *start, start.threadGroups)
+	currentStates = addStateToList(str, i, currentStates, *start, start.threadGroups, nil)
 	var match Match = nil
 	//	var isEmptyAndNoAssertion bool
 	// Main loop
@@ -366,7 +368,7 @@ func findAllSubmatchHelper(start *nfaState, str []rune, offset int, numGroups in
 				break
 			} else if !currentState.isAlternation && !currentState.isKleene && !currentState.isQuestion && !currentState.groupBegin && !currentState.groupEnd { // Normal character or assertion
 				if currentState.contentContains(str, idx) {
-					nextStates = addStateToList(str, idx+1, nextStates, *currentState.next, currentState.threadGroups)
+					nextStates = addStateToList(str, idx+1, nextStates, *currentState.next, currentState.threadGroups, nil)
 				}
 			}
 

regex/nfa.go

@@ -402,19 +402,20 @@ func newState() nfaState {
 }
 
 // Creates and returns a state that _always_ has a zero-length match.
-func zeroLengthMatchState() nfaState {
-	start := newState()
+func zeroLengthMatchState() *nfaState {
+	start := &nfaState{}
 	start.content = newContents(epsilon)
 	start.isEmpty = true
 	start.assert = alwaysTrueAssert
+	start.output = append([]*nfaState{}, start)
 	return start
 }
 
 func (s nfaState) equals(other nfaState) bool {
-	return slices.Equal(s.content, other.content) &&
-		s.isEmpty == other.isEmpty &&
+	return s.isEmpty == other.isEmpty &&
 		s.isLast == other.isLast &&
 		slices.Equal(s.output, other.output) &&
+		slices.Equal(s.content, other.content) &&
 		s.next == other.next &&
 		s.isKleene == other.isKleene &&
 		s.isQuestion == other.isQuestion &&