Merge pull request #8 from aynumosir/fix-edge-cases

fix: Fix edge cases for kana converter

Author: neet

ainu-utils/src/kana/kana.rs

@@ -1,7 +1,7 @@
 use super::constants::{CONSONANTS, SPECIAL_CONSONANTS, VOWELS};
 use super::linking::link;
 use super::maps::{TABLE_1, TABLE_2};
-use super::symbols::symbols;
+use super::symbols::map_symbols;
 use unicode_normalization::char::is_combining_mark;
 use unicode_normalization::UnicodeNormalization;
 
@@ -20,7 +20,7 @@ pub fn to_kana(input: &str) -> String {
 
     input = normalize(input);
     input = link(input);
-    input = symbols(input);
+    input = map_symbols(input);
 
     let chars: Vec<char> = input.chars().collect();
 
@@ -43,7 +43,7 @@ pub fn to_kana(input: &str) -> String {
 
             if CONSONANTS.contains(&current) {
                 if let Some(&next) = next {
-                    if current == next && current != 'n' {
+                    if current == next && !['n', 'y', 'w'].contains(&current) {
                         kana.push_str(TABLE_1.get("t").unwrap());
                         index += 1;
                         continue;
@@ -75,6 +75,11 @@ pub fn to_kana(input: &str) -> String {
                 }
             }
 
+            if '\'' == current {
+                index += 1;
+                continue;
+            }
+
             kana.push(current);
             index += 1;
         }

ainu-utils/src/kana/kana_test.rs

@@ -168,3 +168,14 @@ fn test_k_prefix() {
 fn test_diacritics() {
     assert_eq!(to_kana("kamúy"), "カムイ")
 }
+
+#[test]
+fn test_yy_and_ww() {
+    assert_eq!(to_kana("kamuyyukar"), "カムイユカㇻ");
+    assert_eq!(to_kana("eawwo"), "エアウウォ");
+}
+
+#[test]
+fn test_glottal_stop() {
+    assert_eq!(to_kana("hioy'oy"), "ヒオイオイ");
+}

ainu-utils/src/kana/symbols.rs

@@ -1,4 +1,4 @@
-static SYMBOLS: [(&str, &str); 16] = [
+static SYMBOLS: [(&str, &str); 15] = [
     ("-", ""),
     ("=", ""),
     (" ", "　"),
@@ -13,11 +13,10 @@ static SYMBOLS: [(&str, &str); 16] = [
     (".", "。"),
     ("!", "！"),
     ("?", "？"),
-    ("'", ""),
     ("`", ""),
 ];
 
-pub fn symbols(input: String) -> String {
+pub fn map_symbols(input: String) -> String {
     let mut input = input;
 
     for (from, to) in SYMBOLS.iter() {