add proptests to test against panics

add proptests to test against panics
handle voiced kana

Author: PSeitz

Cargo.toml

@@ -32,6 +32,9 @@ path = "src/bin/to_kana.rs"
 name = "to_romaji"
 path = "src/bin/to_romaji.rs"
 
+[dev-dependencies]
+proptest = "1.5.0"
+
 # docs.rs-specific configuration
 [package.metadata.docs.rs]
 # document all features

src/lib.rs

@@ -90,11 +90,53 @@ pub mod traits;
 pub use traits::{ConvertJapanese, IsJapaneseChar, IsJapaneseStr};
 
 #[cfg(test)]
-mod tests {
+mod proptests {
     use super::*;
 
     #[test]
     fn test_regression_11() {
         assert_eq!("シークヮーサー".to_hiragana(), "しいくゎあさあ");
     }
+
+    use proptest::prelude::*;
+
+    proptest! {
+        #[test]
+        fn to_kana_does_not_panic(input in ".*") {
+            let _ = input.to_kana();
+        }
+
+        #[test]
+        fn to_romaji_does_not_panic(input in ".*") {
+            let _ = &input.to_romaji();
+        }
+
+        #[test]
+        fn to_romaji_does_not_panic_kana_katakana_input(
+            input in r"[ぁ-ゟァ-ヿー]{1,100}"
+        ) {
+                let _ = &input.to_romaji();
+        }
+
+        #[test]
+        fn to_katakana_does_not_panic(input in ".*") {
+            let _ = &input.to_katakana();
+        }
+
+        #[test]
+        fn to_hiragana_does_not_panic(input in ".*") {
+            let _ = &input.to_hiragana();
+        }
+
+        #[test]
+        fn mixed_romaji_kana_katakana_input_does_not_panic(
+            input in r"[A-Za-zぁ-ゟァ-ヿー]{1,100}"
+        ) {
+            let _ = input.to_kana();
+            let _ = input.to_romaji();
+            let _ = input.to_katakana();
+            let _ = input.to_hiragana();
+        }
+
+    }
 }

src/to_romaji.rs

@@ -232,4 +232,9 @@ mod tests {
     fn check_panic_issue_13() {
         assert_eq!(to_romaji("ウーッー"), "uu");
     }
+
+    #[test]
+    fn check_panic_proptest1() {
+        assert_eq!(to_romaji("ヷー"), "waa");
+    }
 }

src/to_romaji_node_tree.rs

@@ -42,6 +42,13 @@ impl Node {
 lazy_static! {
     pub(crate) static ref TO_ROMAJI_NODE_TREE: Node = {
         let transitions = Some(vec![
+            (
+                'ヷ',
+                Node {
+                    transitions: None,
+                    output: "va",
+                },
+            ),
             (
                 '\u{3000}',
                 Node {

src/utils/katakana_to_hiragana.rs

@@ -59,17 +59,12 @@ pub(crate) fn katakana_to_hiragana_with_opt(input: &str, is_destination_romaji:
             (previous_kana, is_char_inner_long_dash(input_char, index))
         {
             // Transform previous_kana back to romaji, and slice off the vowel
-            let romaji = TO_ROMAJI_NODE_TREE
-                .find_transition_node(previous_kana)
-                .unwrap_or_else(|| {
-                    panic!(
-                        "Failed to find transition node for previous kana {}",
-                        previous_kana
-                    )
-                })
-                .output;
+            let Some(node) = TO_ROMAJI_NODE_TREE.find_transition_node(previous_kana) else {
+                hira.push(input_char);
+                continue;
+            };
 
-            let romaji_opt = romaji.chars().last();
+            let romaji_opt = node.output.chars().last();
             // However, ensure 'オー' => 'おお' => 'oo' if this is a transform on the way to romaji
             if let Some(prev_char) = input.chars().nth(index - 1) {
                 if is_char_katakana(prev_char) && romaji_opt == Some('o') && is_destination_romaji {
@@ -82,9 +77,20 @@ pub(crate) fn katakana_to_hiragana_with_opt(input: &str, is_destination_romaji:
                 hira.push(*hit);
             }
         } else if !is_char_long_dash(input_char) && is_char_katakana(input_char) {
-            // Shift charcode.
-            let code = input_char as i32 + (HIRAGANA_START as i32 - KATAKANA_START as i32);
-            let hira_char = std::char::from_u32(code as u32).unwrap();
+            let hira_char = match input_char {
+                // rare special cases
+                'ヷ' => 'わ', // wa with a voiced mark
+                'ヸ' => 'ゐ', // wi with a voiced mark
+                'ヹ' => 'ゑ', // we with a voiced mark
+                'ヺ' => 'を', // wo with a voiced mark
+                _ => {
+                    // Shift charcode.
+                    let code = input_char as i32 + (HIRAGANA_START as i32 - KATAKANA_START as i32);
+                    // the fallback shouldn't normally happen
+                    std::char::from_u32(code as u32).unwrap_or(input_char)
+                }
+            };
+
             hira.push(hira_char);
             previous_kana = Some(hira_char);
         } else {