From e84841b1f52c8f925fda9c04f041f992d276bdbd Mon Sep 17 00:00:00 2001
From: David Gisser <dgisser@gmail.com>
Date: Sun, 21 Dec 2025 20:22:42 -0800
Subject: [PATCH] fix censor behavior for japanese diactritics

---
 src/censor.rs | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)

diff --git a/src/censor.rs b/src/censor.rs
index 2294047..abf65bb 100644
--- a/src/censor.rs
+++ b/src/censor.rs
@@ -164,10 +164,12 @@ impl<I: Iterator<Item = char>> Censor<I> {
         fn filter_char(c: &char) -> bool {
             use finl_unicode::categories::{CharacterCategories, MinorCategory};
             let category = c.get_minor_category();
+            // Preserve Japanese dakuten/handakuten so kana aren't turned into their unvoiced forms.
+            let preserve_japanese = matches!(*c, '\u{3099}' | '\u{309A}');
             let nok = matches!(
                 category,
                 MinorCategory::Cn | MinorCategory::Co | MinorCategory::Mn
-            );
+            ) && !preserve_japanese;
 
             !(nok || BANNED.deref().deref().contains(*c))
         }
@@ -1275,6 +1277,15 @@ mod tests {
         );
     }
 
+    #[test]
+    #[serial]
+    fn japanese_diacritics_preserved() {
+        assert_eq!("パピプペポ", "パピプペポ".censor());
+        assert_eq!("バビブベボ", "バビブベボ".censor());
+        assert_eq!("ぱぴぷぺぽ", "ぱぴぷぺぽ".censor());
+        assert_eq!("ばびぶべぼ", "ばびぶべぼ".censor());
+    }
+
     #[test]
     #[serial]
     fn bandwidth() {