feat: phone_tone_to_kana

crates/sbv2_core/src/jtalk.rs

@@ -1,5 +1,5 @@
 use crate::error::{Error, Result};
-use crate::mora::{MORA_KATA_TO_MORA_PHONEMES, VOWELS};
+use crate::mora::{CONSONANTS, MORA_KATA_TO_MORA_PHONEMES, MORA_PHONEMES_TO_MORA_KATA, VOWELS};
 use crate::norm::{replace_punctuation, PUNCTUATIONS};
 use jpreprocess::{kind, DefaultTokenizer, JPreprocess, SystemDictionaryConfig, UserDictionary};
 use once_cell::sync::Lazy;
@@ -76,6 +76,30 @@ static MORA_PATTERN: Lazy<Vec<String>> = Lazy::new(|| {
 });
 static LONG_PATTERN: Lazy<Regex> = Lazy::new(|| Regex::new(r"(\w)(ー*)").unwrap());
 
+fn phone_tone_to_kana(phones: Vec<String>, tones: Vec<i32>) {
+    let mut results = Vec::new();
+    let mut current_mora = String::new();
+    for ((phone, next_phone), (tone, next_tone)) in phones
+        .iter()
+        .zip(phones.iter().skip(1))
+        .zip(tones.iter().zip(tones.iter().skip(1)))
+    {
+        if PUNCTUATIONS.contains(&phone.clone().as_str()) {
+            results.push((phone, tone));
+            continue;
+        }
+        if CONSONANTS.contains(&phone.clone()) {
+            assert_eq!(current_mora, "");
+            assert_eq!(tone, next_tone);
+            current_mora = phone.to_string()
+        } else {
+            current_mora += phone;
+            results.push((MORA_PHONEMES_TO_MORA_KATA.get(&current_mora).unwrap(), tone));
+            current_mora = String::new();
+        }
+    }
+}
+
 pub struct JTalkProcess {
     jpreprocess: Arc<JPreprocessType>,
     parsed: Vec<String>,

crates/sbv2_core/src/mora.rs

@@ -25,6 +25,21 @@ static MORA_LIST_ADDITIONAL: Lazy<Vec<Mora>> = Lazy::new(|| {
     data.additional
 });
 
+pub static MORA_PHONEMES_TO_MORA_KATA: Lazy<HashMap<String, String>> = Lazy::new(|| {
+    let mut map = HashMap::new();
+    for mora in MORA_LIST_MINIMUM.iter() {
+        map.insert(
+            format!(
+                "{}{}",
+                mora.consonant.clone().unwrap_or("".to_string()),
+                mora.vowel
+            ),
+            mora.mora.clone(),
+        );
+    }
+    map
+});
+
 pub static MORA_KATA_TO_MORA_PHONEMES: Lazy<HashMap<String, (Option<String>, String)>> =
     Lazy::new(|| {
         let mut map = HashMap::new();
@@ -37,4 +52,12 @@ pub static MORA_KATA_TO_MORA_PHONEMES: Lazy<HashMap<String, (Option<String>, Str
         map
     });
 
+pub static CONSONANTS: Lazy<Vec<String>> = Lazy::new(|| {
+    let consonants = MORA_KATA_TO_MORA_PHONEMES
+        .values()
+        .filter_map(|(consonant, _)| consonant.clone())
+        .collect::<Vec<_>>();
+    consonants
+});
+
 pub const VOWELS: [&str; 6] = ["a", "i", "u", "e", "o", "N"];