rosaenlg-filter/dist/LanguageFilterEnglish.js

Filtering feature of RosaeNLG

"use strict";
/**
 * @license
 * Copyright 2019 Ludan Stoecklé
 * SPDX-License-Identifier: Apache-2.0
 */
var __importDefault = (this && this.__importDefault) || function (mod) {
    return (mod && mod.__esModule) ? mod : { "default": mod };
};
Object.defineProperty(exports, "__esModule", { value: true });
exports.LanguageFilterEnglish = void 0;
const LanguageFilter_1 = require("./LanguageFilter");
const english_a_an_1 = require("english-a-an");
const aan_json_1 = __importDefault(require("english-a-an-list/dist/aan.json"));
const better_title_case_1 = __importDefault(require("better-title-case"));
class LanguageFilterEnglish extends LanguageFilter_1.LanguageFilter {
    constructor(languageCommon) {
        super(languageCommon);
        this.cleanSpacesPunctuationDoDefault = true;
    }
    protectRawNumbers(input) {
        let res = input;
        const regexNumber = new RegExp(`([^\\d])${this.constants.stdBeforeWithParenthesis}((\\d{1,3}(?:\\,\\d{3})*|(?:\\d+))(?:\\.\\d+)?)`, 'g');
        res = res.replace(regexNumber, (_match, before1, before2, content) => {
            return before1 + before2 + '<protect>' + content + '</protect>';
        });
        return res;
    }
    beforeProtect(input) {
        let res = input;
        res = this.aAnGeneric(res, true);
        res = this.enPossessivesBeforeProtect(res);
        return res;
    }
    justBeforeUnprotect(input) {
        let res = input;
        res = this.aAnGeneric(res, false);
        res = this.enPossessives(res);
        return res;
    }
    enPossessives(input) {
        let res = input;
        // the <b>earrings</b> 's size => The <b>earrings</b>' size
        const regexSS = new RegExp("s([☞☜\\s]*)'s([^" + this.constants.tousCaracteresMinMajRe + '])', 'g');
        res = res.replace(regexSS, (_match, between, after) => {
            return `s${between}'${after}`;
        });
        return res;
    }
    enPossessivesBeforeProtect(input) {
        let res = input;
        const regexSS = new RegExp("(s\\s*§[\\s¤]*'s)([^" + this.constants.tousCaracteresMinMajRe + '])', 'g');
        res = res.replace(regexSS, (_corresp, _first, second) => {
            return `s§' ${second}`;
        });
        return res;
    }
    aAnGeneric(input, beforeProtect) {
        let res = input;
        const regexA = new RegExp(`([^${this.constants.tousCaracteresMinMajRe}])([aA])${this.constants.stdBetweenWithParenthesis}(${this.constants.getInBetween(beforeProtect)})([${this.constants.tousCaracteresMinMajRe}]*)`, 'g');
        res = res.replace(regexA, (match, before, aA, between, beforeWord, word) => {
            if (word != null && word != '') {
                // can be null when orphan "a" at the very end of a text
                const newAa = this.redoCapitalization(aA, (0, english_a_an_1.getAAn)(this.dictManager.getAdjsWordsData(), aan_json_1.default, word));
                return `${before}${newAa}${between}${beforeWord}${word}`;
            }
            else {
                return match;
            }
        });
        return res;
    }
    redoCapitalization(initial, replacement) {
        if (initial === 'A') {
            return replacement.substring(0, 1).toUpperCase() + replacement.substring(1); // A or An...
        }
        else {
            return replacement;
        }
    }
    titlecase(input) {
        return (0, better_title_case_1.default)(input);
    }
    cleanSpacesPunctuationCorrect(input) {
        let res = input;
        // ['the phone \'s', 'The phone\'s'],
        res = res.replace(/\s*'/g, "'");
        return res;
    }
}
exports.LanguageFilterEnglish = LanguageFilterEnglish;
//# sourceMappingURL=LanguageFilterEnglish.js.map