From 21142e6e1384caa2c02c4076567eaad3bb6aad02 Mon Sep 17 00:00:00 2001
From: Aadhavan Srinivasan <aadhavan@twomorecents.org>
Date: Sun, 3 Nov 2024 14:37:38 -0500
Subject: [PATCH] Wrote function to clone the NFA starting at a given state,
 and a function to find question mark operator (a? == (a|))

---
 nfa.go | 62 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 61 insertions(+), 1 deletion(-)

diff --git a/nfa.go b/nfa.go
index 65b30fc..dd2a2b6 100644
--- a/nfa.go
+++ b/nfa.go
@@ -25,6 +25,52 @@ type State struct {
 	zeroMatchFound bool             // Whether or not the state has been used for a zero-length match - only relevant for zero states
 }
 
+// Clones the NFA starting from the given state.
+func cloneState(start *State) *State {
+	return cloneStateHelper(start, make(map[*State]*State))
+}
+
+// Helper function for clone. The map is used to keep track of which states have
+// already been copied, and which ones haven't.
+// This function was created using output from Llama3.1:405B.
+func cloneStateHelper(state *State, cloneMap map[*State]*State) *State {
+	// Base case - if the clone exists in our map, return it.
+	if clone, exists := cloneMap[state]; exists {
+		return clone
+	}
+	// Recursive case - if the clone doesn't exist, create it, add it to the map,
+	// and recursively call for each of the transition states.
+	clone := &State{
+		content:        append([]int{}, state.content...),
+		isEmpty:        state.isEmpty,
+		isLast:         state.isLast,
+		output:         make([]*State, len(state.output)),
+		transitions:    make(map[int][]*State),
+		isKleene:       state.isKleene,
+		assert:         state.assert,
+		zeroMatchFound: state.zeroMatchFound,
+	}
+	cloneMap[state] = clone
+	for i, s := range state.output {
+		if s == state {
+			clone.output[i] = clone
+		} else {
+			clone.output[i] = cloneStateHelper(s, cloneMap)
+		}
+	}
+	for k, v := range state.transitions {
+		clone.transitions[k] = make([]*State, len(v))
+		for i, s := range v {
+			if s == state {
+				clone.transitions[k][i] = clone
+			} else {
+				clone.transitions[k][i] = cloneStateHelper(s, cloneMap)
+			}
+		}
+	}
+	return clone
+}
+
 // Checks if the given state's assertion is true. Returns true if the given
 // state doesn't have an assertion.
 func (s State) checkAssertion(str string, idx int) bool {
@@ -123,9 +169,13 @@ func verifyLastStates(start []*State) {
 	verifyLastStatesHelper(start[0], make(map[*State]bool))
 }
 
+// Concatenates s1 and s2, returns the start of the concatenation.
 func concatenate(s1 *State, s2 *State) *State {
+	if s1 == nil {
+		return s2
+	}
 	for i := range s1.output {
-		for _, c := range s2.content { // Create transitions for every element in s2's content to s2'
+		for _, c := range s2.content { // Create transitions for every element in s1's content to s2'
 			s1.output[i].transitions[c], _ = unique_append(s1.output[i].transitions[c], s2)
 		}
 	}
@@ -172,3 +222,13 @@ func alternate(s1 *State, s2 *State) *State {
 
 	return toReturn
 }
+
+func question(s1 *State) *State { // Use the fact that ab? == a(b|)
+	s2 := &State{}
+	s2.transitions = make(map[int][]*State)
+	s2.content = newContents(EPSILON)
+	s2.output = append(s2.output, s2)
+	s2.isEmpty = true
+	s3 := alternate(s1, s2)
+	return s3
+}