Grammalecte  Diff

// Grammalecte - text formatter

/* jshint esversion:6, -W097 */
/* jslint esversion:6 */
/* global exports, console */

"use strict";

//!${map}


// Latin letters: http://unicode-table.com/fr/
// 0-9
// A-Z
// a-z
// À-Ö     00C0-00D6   (upper case)

    "erase_non_breaking_hyphens": [ [/­/g, ""] ],
    //// typographic signs
    "ts_apostrophe":              [ [/\b([ldnjmtscç])['´‘′`](?=[a-zA-Zà-ö0-9À-Öø-ÿØ-ßĀ-ʯ])/ig, "$1’"],
                                    [/\b(qu|jusqu|lorsqu|puisqu|quoiqu|quelqu|presqu|entr|aujourd|prud)['´‘′`]/ig, "$1’"] ],
    "ts_ellipsis":                [ [/\.\.\./g, "…"],
                                    [/…\.\./g, "……"],
                                    [/…\.(?!\.)/g, "…"] ],
    "ts_n_dash_middle":           [ [/ [-—] /g, " – "],
                                    [/ [-—],/g, " –,"] ],
    "ts_m_dash_middle":           [ [/ [-–] /g, " — "],
                                    [/ [-–],/g, " —,"] ],
    "ts_n_dash_start":            [ [/^[-—][  ]/gm, "– "],
                                    [/^– /gm, "– "],
                                    [/^[-–—](?=[a-zA-Zà-ö0-9À-Öø-ÿØ-ßĀ-ʯ.…])/gm, "– "] ],
    "ts_m_dash_start":            [ [/^[-–][  ]/gm, "— "],

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




class TextFormatter {

    constructor (bDebug=false) {
        this.sLang = "fr";
        this.bDebug = bDebug;
        //don't change this in external ;)
        this._optionsUsed = dTFDefaultOptions.gl_shallowCopy();
    }

    formatText (sText, dOpt=null) {
        if (dOpt !== null) {
            this._optionsUsed.gl_updateOnlyExistingKeys(dOpt);
        }
        for (let [sOptName, bVal] of this._optionsUsed) {
            //console.log(oReplTable);
            if (bVal && oReplTable[sOptName]) {
                for (let [zRgx, sRep] of oReplTable[sOptName]) {
                    sText = sText.replace(zRgx, sRep);
                }
            }
        }
        return sText;
    }

    formatTextCount (sText, dOpt=null) {
        let nCount = 0;
        if (dOpt !== null) {
            this._optionsUsed.gl_updateOnlyExistingKeys(dOpt);
        }
        for (let [sOptName, bVal] of this._optionsUsed) {
            if (bVal && oReplTable[sOptName]) {
                for (let [zRgx, sRep] of oReplTable[sOptName]) {
                    nCount += (sText.match(zRgx) || []).length;
                    sText = sText.replace(zRgx, sRep);
                }
            }
        }
        return [sText, nCount];
    }

    formatTextRule (sText, sRuleName) {
        if (oReplTable[sRuleName]) {
            for (let [zRgx, sRep] of oReplTable[sRuleName]) {
                sText = sText.replace(zRgx, sRep);
            }
        } else if (this.bDebug){
            console.log("# Error. TF: there is no option “" + sRuleName+ "”.");
        }
        return sText;
    }

    formatTextRuleCount (sText, sRuleName) {
        let nCount = 0;
        if (oReplTable[sRuleName]) {
            for (let [zRgx, sRep] of oReplTable[sRuleName]) {
                nCount += (sText.match(zRgx) || []).length;
                sText = sText.replace(zRgx, sRep);
            }
        } else if (this.bDebug){
            console.log("# Error. TF: there is no option “" + sRuleName+ "”.");
        }
        return [sText, nCount];
    }

    removeHyphenAtEndOfParagraphs (sText) {
        sText = sText.replace(/-[  ]*\n/gm, "");
        return sText;
    }

    removeHyphenAtEndOfParagraphsCount (sText) {
        let nCount = (sText.match(/-[  ]*\n/gm) || []).length;
        sText = sText.replace(/-[  ]*\n/gm, "");
        return [sText, nCount];
    }

    mergeContiguousParagraphs (sText) {
        sText = sText.replace(/^[  ]+$/gm, ""); // clear empty paragraphs
        let s = "";
        for (let sParagraph of this.getParagraph(sText)) {
            if (sParagraph === "") {
                s += "\n";
            } else {
                s += sParagraph + " ";
            }
        }
        s = s.replace(/  +/gm, " ").replace(/ $/gm, "");
        return s;
    }

    mergeContiguousParagraphsCount (sText) {
        let nCount = 0;
        sText = sText.replace(/^[  ]+$/gm, ""); // clear empty paragraphs
        let s = "";
        for (let sParagraph of this.getParagraph(sText)) {
            if (sParagraph === "") {
                s += "\n";
            } else {
                s += sParagraph + " ";
                nCount += 1;
            }
        }
        s = s.replace(/  +/gm, " ").replace(/ $/gm, "");
        return [s, nCount];
    }

    * getParagraph (sText, sSep="\n") {
        // generator: returns paragraphs of text
        let iStart = 0;
        let iEnd = 0;
        while ((iEnd = sText.indexOf(sSep, iStart)) !== -1) {
            yield sText.slice(iStart, iEnd);
            iStart = iEnd + 1;
        }
        yield sText.slice(iStart);
    }

    getDefaultOptions () {
        //we return a copy to make sure they are no modification in external
        return dTFDefaultOptions.gl_shallowCopy();
    }

    getUsedOptions () {
        //we return a copy to make sure they are no modification in external
        return this._optionsUsed.gl_shallowCopy();
    }

    setUsedOptions (dOpt=null) {
        if (dOpt !== null) {
            this._optionsUsed.gl_updateOnlyExistingKeys(dOpt);
        } else if (this.bDebug){
            console.log("# Error. TF: no option to change.");
        }
    }

    getReplTable(){
        return oReplTable;
    }
}


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

}