taipa/lib/client.js

Taiwanese morphological parsing library

"use strict";
Object.defineProperty(exports, "__esModule", { value: true });
exports.Client = exports.TokenAnalysis = void 0;
const init_1 = require("./tonal/init");
const analyzer_1 = require("./unchange/analyzer");
const init_2 = require("./kana/init");
const analyzer_2 = require("./kana/analyzer");
const unit_1 = require("./unit");
const metaplasm_1 = require("./unchange/metaplasm");
class TokenAnalysis {
    /** Analyzed token. */
    word = new unit_1.Word();
    /** Base forms of the word. */
    lemmas = new Array();
    /** Inflectional suffix. */
    inflectionalEnding = '';
    /** sound sequences. */
    soundSequences = new Array();
    /**
     * Syllabic block or syllabogram sequences with tone letters.
     * Multiple sequences could be pushed into this array of strings.
     */
    blockSequences = [];
    /** Standalone form sequences. */
    standaloneSequences = new Array(); // standalone form sequences
}
exports.TokenAnalysis = TokenAnalysis;
class Client {
    processKana(str) {
        (0, init_2.checkNumberOfLettersKana)();
        // kana
        let ta = new TokenAnalysis();
        if (str) {
            const ka = analyzer_2.kanaLemmatizationAnalyzer;
            const morphemes = ka.morphAnalyze(str);
            ta.blockSequences = (0, init_2.getKanaBlocks)(morphemes);
            for (const m of morphemes) {
                ta.soundSequences.push(m.sounds);
            }
        }
        return ta;
    }
    processTonal(str) {
        (0, init_1.checkNumberOfLetterTonal)();
        // tonal lurzmafjiz
        let ta = new TokenAnalysis();
        if (str) {
            const tla = analyzer_1.tonalLemmatizationAnalyzer;
            const morphemes = tla.morphAnalyze(str, new metaplasm_1.TonalStandaloneForms([]));
            const lexeme = tla.lexAnalyze(morphemes);
            ta.word = lexeme.word;
            ta.lemmas = lexeme.getLemmas();
            ta.inflectionalEnding = lexeme.getInflectionalEnding();
            ta.blockSequences = (0, init_1.getBlocks)(morphemes);
            for (const m of morphemes) {
                ta.soundSequences.push(m.sounds);
                // TODO: first free tone to fourth. first checked tone to eighth
                ta.standaloneSequences.push(m.getForms().map((it) => it.literal));
            }
        }
        return ta;
    }
}
exports.Client = Client;
//# sourceMappingURL=client.js.map