From 4966a222f9735b6aaaa9ee6c1caed8fabc7d66ca Mon Sep 17 00:00:00 2001
From: Aadhavan Srinivasan <aadhavan@twomorecents.org>
Date: Sat, 25 Jan 2025 12:44:40 -0500
Subject: [PATCH] Added detection of empty parentheses, as zero-length matches

---
 compile.go | 35 ++++++++++++++++++++++++++++-------
 1 file changed, 28 insertions(+), 7 deletions(-)

diff --git a/compile.go b/compile.go
index e57ab66..0566133 100644
--- a/compile.go
+++ b/compile.go
@@ -764,15 +764,36 @@ func thompson(re []postfixNode) (Reg, error) {
 			// and then some other node.
 			// These three nodes (LPAREN, the middle node and RPAREN) are extracted together, concatenated
 			// and added back in.
+			// If the middle node doesn't exist (ie. something like '()' ), that's fine, I just connect the LPAREN
+			// and RPAREN nodes.
+			// If neither node exists, that's a problem so I return an error.
 			if c.nodetype == RPAREN {
 				s.groupEnd = true
-				middleNode := mustPop(&nfa)
-				lparenNode := mustPop(&nfa)
-				s.groupNum = lparenNode.groupNum
-				tmp := concatenate(lparenNode, middleNode)
-				to_add := concatenate(tmp, s)
-				nfa = append(nfa, to_add)
-
+				middleNode, err1 := pop(&nfa)
+				lparenNode, err2 := pop(&nfa)
+				if err1 != nil && err2 != nil {
+					return Reg{}, fmt.Errorf("Imbalanced parentheses.")
+				} else if err2 != nil { // There was no third node. ie. something like '()'
+					lparenNode = middleNode
+					if lparenNode.groupBegin != true { // There are only two nodes, but the first one isn't an LPAREN.
+						return Reg{}, fmt.Errorf("Imbalanced parentheses.")
+					}
+					s.groupNum = lparenNode.groupNum
+					to_add := concatenate(lparenNode, s)
+					nfa = append(nfa, to_add)
+				} else {
+					// At this point, we assume all three nodes are valid ('lparenNode', 'middleNode' and 's')
+					if lparenNode.groupBegin {
+						s.groupNum = lparenNode.groupNum
+					} else if middleNode.groupBegin { // Something like 'a()'
+						s.groupNum = middleNode.groupNum
+					} else { // A middleNode and lparenNode exist, but neither is actually an LPAREN.
+						return Reg{}, fmt.Errorf("Imbalanced parentheses.")
+					}
+					tmp := concatenate(lparenNode, middleNode)
+					to_add := concatenate(tmp, s)
+					nfa = append(nfa, to_add)
+				}
 			}
 		}
 		if c.nodetype == CHARCLASS { // A Character class consists of all the nodes in it, alternated