From 6ea4f64e564e901ae78b3d40d3dd01fa01810c3b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jonathan=20D=C3=B6nszelmann?= <jonathan@donsz.nl>
Date: Mon, 30 Sep 2024 12:17:43 +0200
Subject: [PATCH] fix regex matcher

---
 scopegraphs-regular-expressions/src/regex.rs |  8 +++++--
 scopegraphs/tests/test_regex.rs              | 23 ++++++++++++++++++++
 2 files changed, 29 insertions(+), 2 deletions(-)

diff --git a/scopegraphs-regular-expressions/src/regex.rs b/scopegraphs-regular-expressions/src/regex.rs
index a400f7c..0db0aa9 100644
--- a/scopegraphs-regular-expressions/src/regex.rs
+++ b/scopegraphs-regular-expressions/src/regex.rs
@@ -299,8 +299,12 @@ fn normalize_or(l: &Rc<Regex>, r: &Rc<Regex>, ab: &AlphabetOrder) -> Rc<Regex> {
         // ~0 | a => ~0
         (Regex::Complement(c), _) if c.deref() == &Regex::EmptySet => l,
         // (a | (b | c)) => (b | (a | c)) if b < a
-        (_, Regex::Or(il, ir)) if l.compare(il, ab) > 0 => {
-            normalize_or(il, &normalize_or(&l, ir, ab), ab)
+        (_, Regex::Or(il, ir)) => {
+            if l.compare(il, ab) > 0 {
+                normalize_or(il, &normalize_or(&l, ir, ab), ab)
+            } else {
+                Regex::Or(l, r).into()
+            }
         }
         // a | b => b | a if b < a
         _ if l.compare(&r, ab) > 0 => normalize_or(&r, &l, ab),
diff --git a/scopegraphs/tests/test_regex.rs b/scopegraphs/tests/test_regex.rs
index 4844e39..45a0087 100644
--- a/scopegraphs/tests/test_regex.rs
+++ b/scopegraphs/tests/test_regex.rs
@@ -85,6 +85,29 @@ fn test_or_3() {
     assert!(Machine::new().accepts([C]));
 }
 
+#[test]
+fn test_repeat_or_2() {
+    use Alphabet::*;
+
+    compile_regex!(type Machine<Alphabet> = A* (A | B));
+
+    assert!(!Machine::new().accepts([A; 0]));
+    assert!(Machine::new().accepts([A]));
+    assert!(Machine::new().accepts([B]));
+}
+
+#[test]
+fn test_repeat_or_3() {
+    use Alphabet::*;
+
+    compile_regex!(type Machine<Alphabet> = A* (A | B | C));
+
+    assert!(!Machine::new().accepts([A; 0]));
+    assert!(Machine::new().accepts([A]));
+    assert!(Machine::new().accepts([B]));
+    assert!(Machine::new().accepts([C]));
+}
+
 #[test]
 fn test_negate() {
     use Alphabet::*;