From 6a96c98d04abe8b60ce5b60b84b5caa697bd0ebb Mon Sep 17 00:00:00 2001
From: Aadhavan Srinivasan <aadhavan@twomorecents.org>
Date: Sat, 1 Feb 2025 19:20:33 -0500
Subject: [PATCH] Fixed bug where the regex '(()|.)(b)' wouldn't compile

---
 regex/compile.go | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/regex/compile.go b/regex/compile.go
index fe324be..eb78a93 100644
--- a/regex/compile.go
+++ b/regex/compile.go
@@ -949,7 +949,9 @@ func thompson(re []postfixNode) (Reg, error) {
 			// 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 the middle node exists but is itself the start of a group, then that _must_ be the opening paren for
+			// the closing paren that I'm on. I put the third node back (because it isn't involved in the capturing group), then
+			// I concatenate those two and add them. If neither node exists, that's a problem so I return an error.
 			if c.nodetype == rparenNode {
 				s.groupEnd = true
 				middleNode, err1 := pop(&nfa)
@@ -964,6 +966,11 @@ func thompson(re []postfixNode) (Reg, error) {
 					s.groupNum = lparenNode.groupNum
 					to_add := concatenate(lparenNode, s)
 					nfa = append(nfa, to_add)
+				} else if middleNode.groupBegin && len(middleNode.transitions) == 0 { // The middle node is a lone lparen - something like '(())', and I'm looking at the first closing parentheses
+					nfa = append(nfa, lparenNode)
+					s.groupNum = middleNode.groupNum // In this case, the 'middle' node is actually a paren node
+					to_add := concatenate(middleNode, s)
+					nfa = append(nfa, to_add)
 				} else {
 					// At this point, we assume all three nodes are valid ('lparenNode', 'middleNode' and 's')
 					if lparenNode.groupBegin {