Rust: 한글 종성 유니코드 변경하기

할 것: 종성이 'ㄲ'인 것을 'ㄴ'으로 바꾸기

(pdf를 txt로 변환하다 보니 "깐"인데 "깎" 이런 식으로 되어 있어서 만들게 됨)

 

한글 유니코드:

초성: (음절 - 0xAC00) / (21 * 28)

중성: (음절 - 0xAC00) % (21 * 28) / 28

종성: (음절 - 0xAC00) % 28

음절의 종성이 'ㄲ'(02)인 경우, 이를 'ㄴ'(04)으로 바꿔야 한다.

 

fn main() {
    let input = "갂깡깔깐깎...";
    let result = replace_gg_with_n(input);
    println!("Result: {:}", result); // Result: 간깡깔깐깐...
}

/// ㄲ 받침 -> ㄴ으로 바꿈
fn replace_gg_with_n(text: &str) -> String {
    let mut output = String::new();

    for ch in text.chars() {
        let unicode_value = ch as u32;

        if unicode_value >= 0xAC00 && unicode_value <= 0xD7A3 {
            let initial = (unicode_value - 0xAC00) / (21 * 28);
            let medial = (unicode_value - 0xAC00) % (21 * 28) / 28;
            let final_consonant = (unicode_value - 0xAC00) % 28;

            if final_consonant == 2 {
                let new_final_consonant = 4;
                let new_unicode_value = 0xAC00 + (initial * 21 * 28) + (medial * 28) + new_final_consonant;
                output.push(std::char::from_u32(new_unicode_value).unwrap());
            } else {
                output.push(ch);
            }
        } else {
            output.push(ch);
        }
    }

    output
}

 

좀 더 응용하면 원하는 종성을 지정해서 아래처럼 만들 수 있을 것 같다.

use std::collections::HashMap;

fn get_jongseong_index(jongseong: char) -> Option {
    let jongseong_map: HashMap<char, u16> = [
        (' ', 0), ('ㄱ', 1), ('ㄲ', 2), ('ㄳ', 3), 
        ('ㄴ', 4), ('ㄵ', 5), ('ㄶ', 6), ('ㄷ', 7), 
        ('ㄹ', 8), ('ㄺ', 9), ('ㄻ', 10), ('ㄼ', 11),
        ('ㄽ', 12), ('ㄾ', 13), ('ㄿ', 14), ('ㅀ', 15),
        ('ㅁ', 16), ('ㅂ', 17), ('ㅄ', 18), ('ㅅ', 19),
        ('ㅆ', 20), ('ㅇ', 21), ('ㅈ', 22), ('ㅊ', 23),
        ('ㅋ', 24), ('ㅌ', 25), ('ㅍ', 26), ('ㅎ', 27),
    ].iter().cloned().collect();

    jongseong_map.get(&jongseong).cloned()
}


fn replace_jongseong(s: &str, original: char, replacement: char) -> String {
    let original_index = get_jongseong_index(original).unwrap();
    let replacement_index = get_jongseong_index(replacement).unwrap();
    
    s.chars()
        .map(|c| {
            let code = c as u32;
            // Check if it's a Hangul syllable
            if 0xAC00 <= code && code <= 0xD7A3 {
                // Decompose the syllable
                let syllable_index = code - 0xAC00;
                let jongseong_index = syllable_index as u16 % 28;
                if jongseong_index == original_index {
                    // If the jongseong matches the original, replace it
                    let new_code = code - (original_index as u32) + (replacement_index as u32);
                    std::char::from_u32(new_code).unwrap()
                } else {
                    // If the jongseong doesn't match, leave it alone
                    c
                }
            } else {
                // If it's not a Hangul syllable, leave it alone
                c
            }
        })
        .collect()
}

fn main() {
    let hey = "깔깘깎꾼";
    let result = replace_jongseong(hey, 'ㄽ', 'ㅋ');
    let result = replace_jongseong(&result, 'ㄲ', 'ㄴ');
    println!("{result}"); // 깔깤깐꾼
}