Grammalecte  Diff

// Grammalecte - Lexicographe
// License: MPL 2
/*jslint esversion: 6*/
/*global require,exports*/

"use strict";

${string}
${map}


if (typeof(require) !== 'undefined') {
    var helpers = require("resource://grammalecte/helpers.js");

}

const _dTAGS = new Map ([
    [':G', "[mot grammatical]"],
    [':N', " nom,"],
    [':A', " adjectif,"],
    [':M1', " prénom,"],
    [':M2', " patronyme,"],

class Lexicographe {

    constructor (oDict) {
        this.oDict = oDict;
        this._zElidedPrefix = new RegExp ("^([dljmtsncç]|quoiqu|lorsqu|jusqu|puisqu|qu)['’](.+)", "i");
        this._zCompoundWord = new RegExp ("([a-zA-Zà-ö0-9À-Öø-ÿØ-ßĀ-ʯ]+)-((?:les?|la)-(?:moi|toi|lui|[nv]ous|leur)|t-(?:il|elle|on)|y|en|[mts][’'](?:y|en)|les?|l[aà]|[mt]oi|leur|lui|je|tu|ils?|elles?|on|[nv]ous)$", "i");
        this._zTag = new RegExp ("[:;/][a-zA-Zà-ö0-9À-Öø-ÿØ-ßĀ-ʯ*][^:;/]*", "g");

    }

    getInfoForToken (oToken) {
        // Token: .sType, .sValue, .nStart, .nEnd
        // return a list [type, token_string, values]
        let m = null;
        try {
            switch (oToken.sType) {

                    break;
                case 'WORD': 
                    if (oToken.sValue.gl_count("-") > 4) {
                        return { sType: "COMPLEX", sValue: oToken.sValue, aLabel: ["élément complexe indéterminé"] };
                    }
                    else if (this.oDict.isValidToken(oToken.sValue)) {
                        let lMorph = this.oDict.getMorph(oToken.sValue);
                        let aElem = [];
                        for (let s of lMorph){
                            if (s.includes(":"))  aElem.push( this._formatTags(s) );
                        }
                        return { sType: oToken.sType, sValue: oToken.sValue, aLabel: aElem};
                    }
                    else if (m = this._zCompoundWord.exec(oToken.sValue)) {
                        // mots composés
                        let lMorph = this.oDict.getMorph(m[1]);
                        let aElem = [];
                        for (let s of lMorph){
                            if (s.includes(":"))  aElem.push( this._formatTags(s) );
                        }
                        aElem.push("-" + m[2] + ": " + this._formatSuffix(m[2].toLowerCase()));
                        return { sType: oToken.sType, sValue: oToken.sValue, aLabel: aElem };
                    }
                    else {
                        return { sType: "UNKNOWN", sValue: oToken.sValue, aLabel: ["inconnu du dictionnaire"] };
                    }
                    break;
            }
        }
        catch (e) {
            helpers.logerror(e);
        }
        return null;

    }

    _formatTags (sTags) {
        let sRes = "";
        sTags = sTags.replace(/V([0-3][ea]?)[itpqnmr_eaxz]+/, "V$1");
        let m;
        while ((m = this._zTag.exec(sTags)) !== null) {
            sRes += _dTAGS.get(m[0]);
            if (sRes.length > 100) {
                break;
            }
        }
        if (sRes.startsWith(" verbe") && !sRes.endsWith("infinitif")) {
            sRes += " [" + sTags.slice(1, sTags.indexOf(" ")) + "]";
        }
        if (!sRes) {
            sRes = "#Erreur. Étiquette inconnue : [" + sTags + "]";
            helpers.echo(sRes);
            return sRes;
        }
        return sRes.gl_trimRight(",");

    }

    _formatSuffix (s) {
        if (s.startsWith("t-")) {
            return "“t” euphonique +" + _dAD.get(s.slice(2));
        }
        if (!s.includes("-")) {
            return _dAD.get(s.replace("’", "'"));
        }
        if (s.endsWith("ous")) {
            s += '2';
        }
        let nPos = s.indexOf("-");
        return _dAD.get(s.slice(0, nPos)) + " +" + _dAD.get(s.slice(nPos+1));

    }
}


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