hanja/lib/translate.js

Sino-Korean, aka Hanja, Utilities for Korean Language Processing

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.split = exports.translate = void 0;
const hanjaTable = require("./data/hanjaeum.json");
function isHanja(char) {
    return char in hanjaTable;
}
function split(text) {
    let prev = null;
    let segment = null;
    const result = [];
    for (const char of text) {
        if (prev === isHanja(char)) {
            if (segment !== null)
                segment += char;
            continue;
        }
        if (segment !== null)
            result.push(segment);
        prev = !prev;
        segment = "";
        segment += char;
    }
    if (segment !== null)
        result.push(segment);
    return result;
}
exports.split = split;
function dueum(char, context) {
    var _a;
    const chars = char.normalize('NFKD');
    // 모음이나 ㄴ 받침 뒤에 이어지는 '렬, 률'은 '열, 율'로 발음한다.
    if ('렬률'.includes(char)) {
        const prevChars = (_a = context[context.length - 1]) === null || _a === void 0 ? void 0 : _a.normalize('NFKD');
        if (prevChars === undefined || prevChars[2] === undefined || prevChars[2] === '안'.normalize('NFKD')[2]) {
            return ('ㅇ' + chars.slice(1)).normalize('NFKC');
        }
    }
    if (context !== '')
        return char;
    // 한자음 '녀, 뇨, 뉴, 니', '랴, 려, 례, 료, 류, 리'가 단어 첫머리에 올 때
    // '여, 요, 유, 이', '야, 여, 예, 요, 유, 이'로 발음한다.
    if (chars[0] === 'ㄴ'.normalize('NFKD') && 'ㅑㅕㅛㅠㅣㅖ'.normalize('NFKD').includes(chars[1])) {
        return ('ㅇ' + chars.slice(1)).normalize('NFKC');
    }
    if (chars[0] === 'ㄹ'.normalize('NFKD') && 'ㅑㅕㅛㅠㅣㅖ'.normalize('NFKD').includes(chars[1])) {
        return ('ㅇ' + chars.slice(1)).normalize('NFKC');
    }
    // 한자음 '라, 래, 로, 뢰, 루, 르'가 단어 첫머리에 올 때 '나, 내, 노, 뇌,
    // 누, 느'로 발음한다.
    if (chars[0] === 'ㄹ'.normalize('NFKD') && 'ㅏㅗㅜㅡㅐㅚ'.normalize('NFKD').includes(chars[1])) {
        return ('ㄴ' + chars.slice(1)).normalize('NFKC');
    }
    return char;
}
function translate(text, mode) {
    if (mode === 'SUBSTITUTION') {
        let result = "";
        for (const char of text) {
            if (char in hanjaTable)
                result += dueum(hanjaTable[char], result);
            else
                result += char;
        }
        return result;
    }
    else if (typeof mode === 'function') {
        const fn = mode || function (hanja) { return hanja; };
        let result = "";
        const segments = split(text);
        for (const segment of segments) {
            if (!isHanja(segment[0])) {
                result += segment;
            }
            else {
                result += fn(segment, translate(segment, 'SUBSTITUTION'));
            }
        }
        return result;
    }
    else if (mode === 'PARENTHESIS_HANGUL') {
        return translate(text, (hanja, hangul) => `${hanja}(${hangul})`);
    }
    else if (mode === 'PARENTHESIS_HANJA') {
        return translate(text, (hanja, hangul) => `${hangul}(${hanja})`);
    }
    return text;
}
exports.translate = translate;