Grammalecte  Diff

// Grammalecte - text formatter
/*jslint esversion: 6*/
/*global exports*/

"use strict";

${map}


// Latin letters: http://unicode-table.com/fr/

    "nbsp_repair":                [ [/([\[(])[   ]([!?:;])/g, "$1$2"],
                                    [/(https?|ftp)[   ]:\/\//g, "$1://"],
                                    [/&([a-z]+)[   ];/g, "&$1;"],
                                    [/&#([0-9]+|x[0-9a-fA-F]+)[   ];/g, "&#$1;"] ],
    //// missing spaces
    "add_space_after_punctuation":[ [/[;!…](?=[a-zA-Zà-ö0-9À-Öø-ÿØ-ßĀ-ʯ])/g, "$& "],
                                    [/[?](?=[A-ZÀ-ÖØ-ßĀ-ʯ])/g, "? "],
                                    [/\.(?=[a-zA-Zà-öÀ-Öø-ÿØ-ßĀ-ʯ][a-zA-Zà-ö0-9À-Öø-ÿØ-ßĀ-ʯ])/g, ". "],
                                    [/\.(?=[A-ZÀ-ÖØ-ßĀ-ʯ][a-zA-Zà-ö0-9À-Öø-ÿØ-ßĀ-ʯ])/g, ". "],
                                    [/\.(?=À)/g, ". "],
                                    [/[,:](?=[a-zA-Zà-öÀ-Öø-ÿØ-ßĀ-ʯ])/g, "$& "],
                                    [/([a-zA-Zà-öÀ-Öø-ÿØ-ßĀ-ʯ]),(?=[0-9])/g, "$1, "] ],
    "add_space_around_hyphens":   [ [/ [-–—](?=[a-zA-Zà-ö0-9À-Öø-ÿØ-ßĀ-ʯ"«“'‘])/g, "$& "],
                                    [/([a-zA-Zà-ö0-9À-Öø-ÿØ-ßĀ-ʯ"»”'’])([-–—]) /g, "$1 $2 "] ],
    "add_space_repair":           [ [/DnT, ([wA])\b/g, "DnT,$1"] ],
    //// erase

                                    [/^(qu|lorsqu|puisqu|quoiqu|presqu|jusqu|aujourd|entr|quelqu|prud) /ig, "$1’"] ],
    "ma_1letter_lowercase":       [ [/[  ]([ldjnmtscç]) (?=[aàeéêiîoôuyhAÀEÉÊIÎOÔUYH])/g, "$1’"] ],
    "ma_1letter_uppercase":       [ [/[  ]([LDJNMTSCÇ]) (?=[aàeéêiîoôuyhAÀEÉÊIÎOÔUYH])/g, "$1’"],
                                    [/^([LDJNMTSCÇ]) (?=[aàeéêiîoôuyhAÀEÉÊIÎOÔUYH])/g, "$1’"] ]
};


const dDefaultOptions = new Map ([
const dTFDefaultOptions = new Map ([
    ["ts_units", true],
    ["start_of_paragraph", true],
    ["end_of_paragraph", true],
    ["between_words", true],
    ["before_punctuation", true],
    ["within_parenthesis", true],
    ["within_square_brackets", true],

    ["etc", true],
    ["missing_hyphens", true],
    ["ma_word", true],
    ["ma_1letter_lowercase", false],
    ["ma_1letter_uppercase", false]
]);

const dOptions = dDefaultOptions._shallowCopy();
const dTFOptions = dTFDefaultOptions.gl_shallowCopy();


class TextFormatter {

    constructor () {
        this.sLang = "fr";
    };
    }

    formatText (sText, dOpt=null) {
        if (dOpt !== null) {
            dOptions._updateOnlyExistingKeys(dOpt);
            dTFOptions.gl_updateOnlyExistingKeys(dOpt);
        }
        for (let [sOptName, bVal] of dOptions) {
        for (let [sOptName, bVal] of dTFOptions) {
            if (bVal && oReplTable.has(sOptName)) {
                for (let [zRgx, sRep] of oReplTable[sOptName]) {
                    sText = sText.replace(zRgx, sRep);
                }
            }
        }
        return sText;
    };

    }

    getDefaultOptions () {
        return dDefaultOptions;
        return dTFDefaultOptions;
    }
}


if (typeof(exports) !== 'undefined') {
    exports.TextFormatter = TextFormatter;
    exports.oReplTable = oReplTable;
}