rosaenlg-filter/dist/punctuation.js

Filtering feature of RosaeNLG

"use strict";
/**
 * @license
 * Copyright 2019 Ludan Stoecklé
 * SPDX-License-Identifier: Apache-2.0
 */
Object.defineProperty(exports, "__esModule", { value: true });
exports.addCaps = exports.quotes = exports.parenthesis = exports.cleanSpacesPunctuation = exports.duplicatePunctuation = exports.EATSPACE = void 0;
const rosaenlg_commons_1 = require("rosaenlg-commons");
exports.EATSPACE = 'EATSPACE';
// warning: we use both allPunctList and stdPunctList to manage Spanish properly
function duplicatePunctuation(input, languageFilter) {
    let res = input;
    // ['bla ...', 'bla…'],
    res = res.replace(/\.\.\./g, '…');
    // ['bla ! . bla', 'Bla! Bla'],
    const regexDoublePunct = new RegExp(`([${languageFilter.constants.allPunctList}])((?:${languageFilter.constants.spaceOrNonBlockingClass}*[${rosaenlg_commons_1.Constants.stdPunctList}])*)`, 'g');
    res = res.replace(regexDoublePunct, function (_match, firstPunct, otherStuff) {
        const regexRemovePunct = new RegExp(`[${rosaenlg_commons_1.Constants.stdPunctList}]`, 'g');
        const removedPunct = otherStuff.replace(regexRemovePunct, () => {
            return '';
        });
        return `${firstPunct}${removedPunct}`;
    });
    return res;
}
exports.duplicatePunctuation = duplicatePunctuation;
function cleanSpacesPunctuation(input, languageFilter) {
    let res = input;
    // 2 spaces or more
    res = res.replace(/\s{2,}/g, ' ');
    res = languageFilter.cleanSpacesPunctuation(res);
    if (languageFilter.cleanSpacesPunctuationDoDefault) {
        const regexPunct = new RegExp(
        // stdPunctList and not allPunctList: on purpose, as special Spanish is managed just before
        `(${languageFilter.constants.spaceOrNonBlockingClass}*)([${rosaenlg_commons_1.Constants.stdPunctList}])(${languageFilter.constants.spaceOrNonBlockingClass}*)`, 'g');
        res = res.replace(regexPunct, (_match, before, punct, after) => {
            return `${before.replace(/\s/g, '')}${punct} ${after.replace(/\s/g, '')}`;
        });
    }
    res = res.replace(/\s+☚/g, '☚');
    // ['bla  .   </p>', 'bla.</p>']
    res = res.replace(/☛\s+/g, '☛');
    res = res.replace(/\s+☚/g, '☚');
    // spaces at the very end
    res = res.trim();
    // eat spaces
    const eatspaceRe = new RegExp(`[\\s¤]*${exports.EATSPACE}[\\s¤]*`, 'g');
    res = res.replace(eatspaceRe, '');
    res = languageFilter.cleanSpacesPunctuationCorrect(res);
    return res;
}
exports.cleanSpacesPunctuation = cleanSpacesPunctuation;
function parenthesis(input, languageFilter) {
    let res = input;
    // remove spaces after '(' or before ')'
    res = res.replace(/\(\s+/g, '(');
    res = res.replace(/\s+\)/g, ')');
    // add spaces before '(' or after ')'
    const regexSpaceBeforePar = new RegExp('[' + languageFilter.constants.tousCaracteresMinMajRe + ']\\(', 'g');
    res = res.replace(regexSpaceBeforePar, (match) => {
        return match.charAt(0) + ' (';
    });
    const regexSpaceAfterPar = new RegExp('\\)[' + languageFilter.constants.tousCaracteresMinMajRe + ']', 'g');
    res = res.replace(regexSpaceAfterPar, (match) => {
        return ') ' + match.charAt(1);
    });
    return res;
}
exports.parenthesis = parenthesis;
function quotes(input) {
    let res = input;
    let alreadyStarted = false;
    res = res.replace(/(\s*)"(\s*)/g, () => {
        if (!alreadyStarted) {
            alreadyStarted = true;
            return ' "';
        }
        else {
            alreadyStarted = false;
            return '" ';
        }
    });
    // trigger a warning if an end is missing
    if (alreadyStarted) {
        console.log(`WARNING: did find a starting " but not the ending one`);
    }
    // mixes of quotes and parenthesis
    res = res.replace(/\(\s*"/g, () => {
        return ' ("';
    });
    res = res.replace(/"\s*\)/g, () => {
        return '") ';
    });
    return res;
}
exports.quotes = quotes;
function addCaps(input, languageFilter) {
    let res = input;
    {
        const triggerCapsWithSpace = '[\\.!\\?¡¿]';
        const regexCapsAfterDot = new RegExp(`(${triggerCapsWithSpace})(${languageFilter.constants.spaceOrNonBlockingClass}*)([${languageFilter.constants.tousCaracteresMinMajRe}])`, 'g');
        res = res.replace(regexCapsAfterDot, (_match, punct, before, firstWord) => {
            return `${punct}${before.replace(/\s/g, '')} ${firstWord.toUpperCase()}`;
        });
    }
    res = languageFilter.addCapsSpecific(res);
    {
        const regexCapsAfterP = new RegExp(`([☛☚])(${languageFilter.constants.spaceOrNonBlockingClass}*)([${languageFilter.constants.tousCaracteresMinMajRe}])`, 'g');
        res = res.replace(regexCapsAfterP, (_match, start, between, char) => {
            return `${start}${between.replace(/ /g, '')}${char.toUpperCase()}`;
        });
    }
    return res;
}
exports.addCaps = addCaps;
//# sourceMappingURL=punctuation.js.map