From efc34086b047acf0b8381680e7f47d6e4f1c5d58 Mon Sep 17 00:00:00 2001
From: matt rice <ratmice@gmail.com>
Date: Sat, 2 May 2026 17:42:01 -0700
Subject: [PATCH] Stop trimming escaped spaces off the end regex

---
 lrlex/src/lib/parser.rs                  | 49 +++++++++++++++++++++++-
 lrpar/cttests/src/regex_trailing_ws.test | 19 +++++++++
 2 files changed, 67 insertions(+), 1 deletion(-)
 create mode 100644 lrpar/cttests/src/regex_trailing_ws.test

diff --git a/lrlex/src/lib/parser.rs b/lrlex/src/lib/parser.rs
index 784a83f11..170e929a6 100644
--- a/lrlex/src/lib/parser.rs
+++ b/lrlex/src/lib/parser.rs
@@ -490,7 +490,7 @@ where
 
         if !dupe {
             let (start_states, re_str) =
-                self.parse_start_states(i, line[..rspace].trim_end_matches(matches_whitespace))?;
+                self.parse_start_states(i, trim_end_unescaped(&line[..rspace]))?;
             let rules_len = self.rules.len();
             let tok_id = LexerTypesT::StorageT::try_from(rules_len)
                            .unwrap_or_else(|_| panic!("StorageT::try_from \
@@ -685,6 +685,20 @@ where
     }
 }
 
+fn trim_end_unescaped(s: &str) -> &str {
+    let trimmed = s.trim_end_matches(matches_whitespace);
+    if trimmed.len() == s.len() {
+        return s;
+    }
+    // If the number of backslashes is odd then the first space in the trimmed portion is escaped so re-add it.
+    if trimmed.chars().rev().take_while(|&c| c == '\\').count() % 2 == 1 {
+        // Panic safety: the trimmed portion is at least one char long.
+        &s[..trimmed.len() + s[trimmed.len()..].chars().next().unwrap().len_utf8()]
+    } else {
+        trimmed
+    }
+}
+
 #[cfg(test)]
 mod test {
     use super::*;
@@ -1826,4 +1840,37 @@ b "A"
             18,
         );
     }
+
+    #[test]
+    fn unescaped_trim() {
+        let escapes = [
+            (r#"\ "#, r#"\ "#),
+            (r#"\  "#, r#"\ "#),
+            (r#"\\ "#, r#"\\"#),
+            (r#"\\  "#, r#"\\"#),
+            (r#"\\\ "#, r#"\\\ "#),
+            (r#"\\\  "#, r#"\\\ "#),
+            (r#"\\\\ "#, r#"\\\\"#),
+            (r#"\\\\  "#, r#"\\\\"#),
+            (r#"x"#, r#"x"#),
+            (r#"x\ "#, r#"x\ "#),
+            (r#"x\  "#, r#"x\ "#),
+            (r#"x\\ "#, r#"x\\"#),
+            (r#"x\\  "#, r#"x\\"#),
+            (r#"x\\\ "#, r#"x\\\ "#),
+            (r#"x\\\  "#, r#"x\\\ "#),
+            (r#"x\\\\ "#, r#"x\\\\"#),
+            (r#"x\\\\  "#, r#"x\\\\"#),
+            (r#"x\ y "#, r#"x\ y"#),
+            (r#"x\ y\ "#, r#"x\ y\ "#),
+            (r#"x\ y\\ "#, r#"x\ y\\"#),
+            (r#"x\ y  "#, r#"x\ y"#),
+            (r#"x\ y\  "#, r#"x\ y\ "#),
+            (r#"x\ y\\  "#, r#"x\ y\\"#),
+        ];
+        for (escaped, expected) in escapes {
+            let trimmed = trim_end_unescaped(escaped);
+            assert_eq!(expected, trimmed)
+        }
+    }
 }
diff --git a/lrpar/cttests/src/regex_trailing_ws.test b/lrpar/cttests/src/regex_trailing_ws.test
new file mode 100644
index 000000000..ce5b880c6
--- /dev/null
+++ b/lrpar/cttests/src/regex_trailing_ws.test
@@ -0,0 +1,19 @@
+name: Test with regex containing trailing ws
+grammar: |
+    %grmtools {
+        yacckind: Original(YaccOriginalActionKind::NoAction),
+        recoverer: RecoveryKind::None,
+        test_files: ["*.input_trailing_ws"],
+    }
+    %start Expr
+    %%
+    Expr: "trailing";
+
+lexer: |
+    %%
+    [a-zA-Z]\  "trailing"
+    [\n\t] ;
+
+extra_files:
+  input1.input_trailing_ws: |
+    a