From bc11777ad527007ee389c300fc3a45d77c3a5614 Mon Sep 17 00:00:00 2001
From: Aadhavan Srinivasan <aadhavan@twomorecents.org>
Date: Tue, 22 Oct 2024 17:07:01 -0400
Subject: [PATCH] Fixed Kleene Star matching

---
 main.go     | 23 ++++++++++++-----------
 matching.go | 26 +++++++++++++++++---------
 nfa.go      |  4 ++--
 3 files changed, 31 insertions(+), 22 deletions(-)

diff --git a/main.go b/main.go
index e885475..b802b15 100644
--- a/main.go
+++ b/main.go
@@ -102,8 +102,8 @@ func shuntingYard(re string) string {
 
 // Thompson's algorithm. Constructs Finite-State Automaton from given string.
 // Returns start state.
-func thompson(re string) State {
-	nfa := make([]State, 0) // Stack of states
+func thompson(re string) *State {
+	nfa := make([]*State, 0) // Stack of states
 	for _, c := range re {
 		if isAlphaNum(c) {
 			state := State{}
@@ -112,7 +112,7 @@ func thompson(re string) State {
 			state.output = make([]*State, 0)
 			state.output = append(state.output, &state)
 			state.isEmpty = false
-			nfa = append(nfa, state)
+			nfa = append(nfa, &state)
 		}
 		// Must be an operator if it isn't alphanumeric
 		switch c {
@@ -120,31 +120,31 @@ func thompson(re string) State {
 			s2 := pop(&nfa)
 			s1 := pop(&nfa)
 			for i := range s1.output {
-				s1.output[i].transitions[s2.content] = append(s1.output[i].transitions[s2.content], &s2)
+				s1.output[i].transitions[s2.content] = append(s1.output[i].transitions[s2.content], s2)
 			}
 			s1.output = s2.output
 			nfa = append(nfa, s1)
 		case '*':
 			s1 := pop(&nfa)
 			for i := range s1.output {
-				s1.output[i].transitions[s1.content] = append(s1.output[i].transitions[s1.content], &s1)
+				s1.output[i].transitions[s1.content] = append(s1.output[i].transitions[s1.content], s1)
 			}
 			// Reset output to s1 (in case s1 was a union operator state, which has multiple outputs)
 			s1.output = nil
-			s1.output = append(s1.output, &s1)
+			s1.output = append(s1.output, s1)
 			nfa = append(nfa, s1)
 		case '|':
 			s1 := pop(&nfa)
 			s2 := pop(&nfa)
 			s3 := State{}
 			s3.transitions = make(map[int][]*State)
-			s3.output = append(s3.output, &s1, &s2)
-			s3.transitions[s1.content] = append(s3.transitions[s1.content], &s1)
-			s3.transitions[s2.content] = append(s3.transitions[s2.content], &s2)
+			s3.output = append(s3.output, s1, s2)
+			s3.transitions[s1.content] = append(s3.transitions[s1.content], s1)
+			s3.transitions[s2.content] = append(s3.transitions[s2.content], s2)
 			s3.content = UNION
 			s3.isEmpty = true
 
-			nfa = append(nfa, s3)
+			nfa = append(nfa, &s3)
 		}
 	}
 	if len(nfa) != 1 {
@@ -160,13 +160,14 @@ func thompson(re string) State {
 func main() {
 	if len(os.Args) < 3 {
 		fmt.Println("ERROR: Missing cmdline args")
+		os.Exit(22)
 	}
 	var re string
 	re = os.Args[1]
 	re_postfix := shuntingYard(re)
 	fmt.Println(re_postfix)
 	start := thompson(re_postfix)
-	s, e, matched := match(&start, os.Args[2])
+	s, e, matched := match(start, os.Args[2])
 	if matched {
 		fmt.Printf("Matched from %d to %d\n", s, e)
 	} else {
diff --git a/matching.go b/matching.go
index 4678cdf..a55a14d 100644
--- a/matching.go
+++ b/matching.go
@@ -31,22 +31,30 @@ func match(start *State, str string) (startIdx int, endIdx int, matched bool) {
 		for _, state := range currentStates {
 			if len(state.transitions[int(str[i])]) > 0 {
 				tempStates = append(tempStates, state.transitions[int(str[i])]...)
+			} else {
+				// This enables the 'greedy' behavior - last-state status is only checked if we can't match anything else
+				if state.isLast {
+					endIdx = i
+					return startIdx, endIdx, true
+				}
 			}
 		}
 		copy(currentStates, tempStates)
 		tempStates = nil
 
-		// If any of the current-states is a last state, return true.
-		for _, state := range currentStates {
-			if state.isLast {
-				endIdx = i
-				return startIdx, endIdx, true
-			}
-		}
 		i++
 	}
 
-	// We don't seem to have reached a last-state. Return fals
-	return -1, -1, false
+	// End-of-string reached. Check if any of our states is in the end position.
+	for _, state := range currentStates {
+		if state.isLast {
+			endIdx = i
+			return startIdx, endIdx, true
+		} else {
+			return -1, -1, false
+		}
+	}
 
+	// Default
+	return -1, -1, false
 }
diff --git a/nfa.go b/nfa.go
index 030d21c..53cf37a 100644
--- a/nfa.go
+++ b/nfa.go
@@ -39,6 +39,6 @@ func verifyLastStatesHelper(state *State, visited map[*State]bool) {
 }
 
 // verifyLastStates penables the 'isLast' flag for the leaf nodes (last states)
-func verifyLastStates(start []State) {
-	verifyLastStatesHelper(&start[0], make(map[*State]bool))
+func verifyLastStates(start []*State) {
+	verifyLastStatesHelper(start[0], make(map[*State]bool))
 }