Grammalecte  Diff

        }
        /*
            Properties:
            sName, nCompressionMethod, sHeader, lArcVal, nArcVal, sByDic, sLang, nChar, nBytesArc, nBytesNodeAddress,
            nEntry, nNode, nArc, nAff, cStemming, nTag, dChar, nBytesOffset,
        */

        /*
            Bug workaround.
            Mozilla’s JS parser sucks. Can’t read file bigger than 4 Mb!
            So we convert huge hexadecimal string to list of numbers…
            https://github.com/mozilla/addons-linter/issues/1361
        */
        let lTemp = [];
        for (let i = 0;  i < this.sByDic.length;  i+=2) {
            lTemp.push(parseInt(this.sByDic.slice(i, i+2), 16));
        }
        this.byDic = lTemp;
        //this.byDic = new Uint8Array(lTemp);  // not quicker, even slower
        /* end of bug workaround */

        if (!(this.sHeader.startsWith("/grammalecte-fsa/") || this.sHeader.startsWith("/pyfsa/"))) {
            throw TypeError("# Error. Not a grammalecte-fsa binary dictionary. Header: " + this.sHeader);
        }
        if (!(this.nCompressionMethod == 1 || this.nCompressionMethod == 2 || this.nCompressionMethod == 3)) {
            throw RangeError("# Error. Unknown dictionary compression method: " + this.nCompressionMethod);
        }
        // <dChar> to get the value of an arc, <dCharVal> to get the char of an arc with its value
        this.dChar = helpers.objectToMap(this.dChar);
        this.dCharVal = this.dChar.gl_reverse();
        this.a2grams = (this.l2grams) ? new Set(this.l2grams) : null;

        if (this.cStemming == "S") {
            this.funcStemming = str_transform.changeWordWithSuffixCode;
        } else if (this.cStemming == "A") {
            this.funcStemming = str_transform.changeWordWithAffixCode;
        } else {
            this.funcStemming = str_transform.noStemming;
        }

        /*
            Bug workaround.
            Mozilla’s JS parser sucks. Can’t read file bigger than 4 Mb!
            So we convert huge hexadecimal string to list of numbers…
            https://github.com/mozilla/addons-linter/issues/1361
        */
        /*
            Performance trick:
            Instead of converting bytes to integers each times we parse the binary dictionary,
            we do it once, then parse the array
        */
        let nAcc = 0;
        this._arcMask = (2 ** ((this.nBytesArc * 8) - 3)) - 1;
        this._finalNodeMask = 1 << ((this.nBytesArc * 8) - 1);
        this._lastArcMask = 1 << ((this.nBytesArc * 8) - 2);


        let lBytesBuffer = [];
        let lTemp = [];
        let nDivisor = (this.nBytesArc + this.nBytesNodeAddress) / 2;
        for (let i = 0;  i < this.sByDic.length;  i+=2) {
            lBytesBuffer.push(parseInt(this.sByDic.slice(i, i+2), 16));
            if (nAcc == (this.nBytesArc - 1)) {
                lTemp.push(this._convBytesToInteger(lBytesBuffer));
                lBytesBuffer = [];
            }
            else if (nAcc == (this.nBytesArc + this.nBytesNodeAddress - 1)) {
                lTemp.push(Math.round(this._convBytesToInteger(lBytesBuffer) / nDivisor));  // Math.round should be useless, BUT with JS who knowns what can happen…
                lBytesBuffer = [];
                nAcc = -1;
            }
        // Configuring DAWG functions according to nCompressionMethod
        switch (this.nCompressionMethod) {
            case 1:
            nAcc = nAcc + 1;
                this.morph = this._morph1;
                this.stem = this._stem1;
                this._lookupArcNode = this._lookupArcNode1;
        }
        this.byDic = lTemp;
        /* end of bug workaround */

                this._getArcs = this._getArcs1;
                this._writeNodes = this._writeNodes1;
                break;
            case 2:
                this.morph = this._morph2;
                this.stem = this._stem2;
                this._lookupArcNode = this._lookupArcNode2;
        this._arcMask = (2 ** ((this.nBytesArc * 8) - 3)) - 1;
        this._finalNodeMask = 1 << ((this.nBytesArc * 8) - 1);
        this._lastArcMask = 1 << ((this.nBytesArc * 8) - 2);
                this._getArcs = this._getArcs2;
                this._writeNodes = this._writeNodes2;
                break;
            case 3:
                this.morph = this._morph3;
                this.stem = this._stem3;
                this._lookupArcNode = this._lookupArcNode3;
                this._getArcs = this._getArcs3;
                this._writeNodes = this._writeNodes3;
                break;
            default:
                throw ValueError("# Error: unknown code: " + this.nCompressionMethod);
        }

        //console.log(this.getInfo());
        this.bAcronymValid = true;
        this.bNumAtLastValid = false;

        // lexicographer module ?
        this.lexicographer = null;
        // JS still sucks: we’ll try importation when importation will be available in Workers. Still waiting...
        if (self && self.hasOwnProperty("lexgraph_"+this.sLangCode)) { // self is the Worker
            this.lexicographer = self["lexgraph_"+this.sLangCode];
        }

    }

    getInfo () {
        return  `  Language: ${this.sLangName}   Lang code: ${this.sLangCode}   Dictionary name: ${this.sDicName}\n` +
                `  Compression method: ${this.nCompressionMethod}   Date: ${this.sDate}   Stemming: ${this.cStemming}FX\n` +
                `  Arcs values:  ${this.nArcVal} = ${this.nChar} characters,  ${this.nAff} affixes,  ${this.nTag} tags\n` +
                `  Dictionary: ${this.nEntry} entries,    ${this.nNode} nodes,   ${this.nArc} arcs\n` +

                return false;
            }
            iAddr = this._lookupArcNode(this.dChar.get(c), iAddr);
            if (iAddr === null) {
                return false;
            }
        }
        return Boolean(this._convBytesToInteger(this.byDic.slice(iAddr, iAddr+this.nBytesArc)) & this._finalNodeMask);
        return Boolean(this.byDic[iAddr] & this._finalNodeMask);
    }

    getMorph (sWord) {
        // retrieves morphologies list, different casing allowed
        if (!sWord) {
            return [];
        }
        sWord = str_transform.spellingNormalization(sWord);
        let l = this.morph(sWord);
        let l = this._morph(sWord);
        if (sWord[0].gl_isUpperCase()) {
            l.push(...this.morph(sWord.toLowerCase()));
            l.push(...this._morph(sWord.toLowerCase()));
            if (sWord.gl_isUpperCase() && sWord.length > 1) {
                l.push(...this.morph(sWord.gl_toCapitalize()));
                l.push(...this._morph(sWord.gl_toCapitalize()));
            }
        }
        return l;
    }

    suggest (sWord, nSuggLimit=10, bSplitTrailingNumbers=false) {
        // returns a array of suggestions for <sWord>

            }
        }
    }

    _suggest (oSuggResult, sRemain, nMaxSwitch=0, nMaxDel=0, nMaxHardRepl=0, nMaxJump=0, nDist=0, nDeep=0, iAddr=0, sNewWord="", bAvoidLoop=false) {
        // returns a set of suggestions
        // recursive function
        if (this._convBytesToInteger(this.byDic.slice(iAddr, iAddr+this.nBytesArc)) & this._finalNodeMask) {
        if (this.byDic[iAddr] & this._finalNodeMask) {
            if (sRemain == "") {
                oSuggResult.addSugg(sNewWord);
                for (let sTail of this._getTails(iAddr)) {
                    oSuggResult.addSugg(sNewWord+sTail);
                }
                return;
            }

    }

    _getTails (iAddr, sTail="", n=2) {
        // return a list of suffixes ending at a distance of <n> from <iAddr>
        let aTails = new Set();
        for (let [nVal, jAddr] of this._getArcs(iAddr)) {
            if (nVal <= this.nChar) {
                if (this._convBytesToInteger(this.byDic.slice(jAddr, jAddr+this.nBytesArc)) & this._finalNodeMask) {
                if (this.byDic[jAddr] & this._finalNodeMask) {
                    aTails.add(sTail + this.dCharVal.get(nVal));
                }
                if (n && aTails.size == 0) {
                    aTails.gl_update(this._getTails(jAddr, sTail+this.dCharVal.get(nVal), n-1));
                }
            }
        }

            zFlexPattern = (sFlexPattern !== "") ? new RegExp(sFlexPattern) : null;
            zTagsPattern = (sTagsPattern !== "") ? new RegExp(sTagsPattern) : null;
        }
        catch (e) {
            console.log("Error in regex pattern");
            console.log(e.message);
        }
        yield* this._select1(zFlexPattern, zTagsPattern, 0, "");
        yield* this._select(zFlexPattern, zTagsPattern, 0, "");
    }

    // VERSION 1

    * _select1 (zFlexPattern, zTagsPattern, iAddr, sWord) {
    * _select (zFlexPattern, zTagsPattern, iAddr, sWord) {
        // recursive generator
        for (let [nVal, jAddr] of this._getArcs1(iAddr)) {
        for (let [nVal, jAddr] of this._getArcs(iAddr)) {
            if (nVal <= this.nChar) {
                // simple character
                yield* this._select1(zFlexPattern, zTagsPattern, jAddr, sWord + this.lArcVal[nVal]);
                yield* this._select(zFlexPattern, zTagsPattern, jAddr, sWord + this.lArcVal[nVal]);
            } else {
                if (!zFlexPattern || zFlexPattern.test(sWord)) {
                    let sStem = this.funcStemming(sWord, this.lArcVal[nVal]);
                    for (let [nMorphVal, _] of this._getArcs1(jAddr)) {
                    for (let [nMorphVal, _] of this._getArcs(jAddr)) {
                        if (!zTagsPattern || zTagsPattern.test(this.lArcVal[nMorphVal])) {
                            yield [sWord, sStem, this.lArcVal[nMorphVal]];
                        }
                    }
                }
            }
        }
    }

    _morph1 (sWord) {
    _morph (sWord) {
        // returns morphologies of sWord
        let iAddr = 0;
        for (let c of sWord) {
            if (!this.dChar.has(c)) {
                return [];
            }
            iAddr = this._lookupArcNode(this.dChar.get(c), iAddr);
            if (iAddr === null) {
                return [];
            }
        }
        if (this._convBytesToInteger(this.byDic.slice(iAddr, iAddr+this.nBytesArc)) & this._finalNodeMask) {
        if (this.byDic[iAddr] & this._finalNodeMask) {
            let l = [];
            let nRawArc = 0;
            while (!(nRawArc & this._lastArcMask)) {
                let iEndArcAddr = iAddr + this.nBytesArc;
                nRawArc = this._convBytesToInteger(this.byDic.slice(iAddr, iEndArcAddr));
                let iEndArcAddr = iAddr + 1;
                nRawArc = this.byDic[iAddr];
                let nArc = nRawArc & this._arcMask;
                if (nArc > this.nChar) {
                    // This value is not a char, this is a stemming code
                    let sStem = ">" + this.funcStemming(sWord, this.lArcVal[nArc]);
                    // Now , we go to the next node and retrieve all following arcs values, all of them are tags
                    let iAddr2 = this._convBytesToInteger(this.byDic.slice(iEndArcAddr, iEndArcAddr+this.nBytesNodeAddress));
                    let iAddr2 = this.byDic[iEndArcAddr];
                    let nRawArc2 = 0;
                    while (!(nRawArc2 & this._lastArcMask)) {
                        let iEndArcAddr2 = iAddr2 + this.nBytesArc;
                        nRawArc2 = this._convBytesToInteger(this.byDic.slice(iAddr2, iEndArcAddr2));
                        let iEndArcAddr2 = iAddr2 + 1;
                        nRawArc2 = this.byDic[iAddr2];
                        l.push(sStem + "/" + this.lArcVal[nRawArc2 & this._arcMask]);
                        iAddr2 = iEndArcAddr2+this.nBytesNodeAddress;
                        iAddr2 = iEndArcAddr2 + 1;
                    }
                }
                iAddr = iEndArcAddr + this.nBytesNodeAddress;
                iAddr = iEndArcAddr + 1;
            }
            return l;
        }
        return [];
    }

    _stem1 (sWord) {
    _stem (sWord) {
        // returns stems list of sWord
        let iAddr = 0;
        for (let c of sWord) {
            if (!this.dChar.has(c)) {
                return [];
            }
            iAddr = this._lookupArcNode(this.dChar.get(c), iAddr);
            if (iAddr === null) {
                return [];
            }
        }
        if (this._convBytesToInteger(this.byDic.slice(iAddr, iAddr+this.nBytesArc)) & this._finalNodeMask) {
        if (this.byDic[iAddr] & this._finalNodeMask) {
            let l = [];
            let nRawArc = 0;
            while (!(nRawArc & this._lastArcMask)) {
                let iEndArcAddr = iAddr + this.nBytesArc;
                nRawArc = this._convBytesToInteger(this.byDic.slice(iAddr, iEndArcAddr));
                let iEndArcAddr = iAddr + 1;
                nRawArc = this.byDic[iAddr];
                let nArc = nRawArc & this._arcMask;
                if (nArc > this.nChar) {
                    // This value is not a char, this is a stemming code
                    l.push(this.funcStemming(sWord, this.lArcVal[nArc]));
                }
                iAddr = iEndArcAddr + this.nBytesNodeAddress;
                iAddr = iEndArcAddr + 1;
            }
            return l;
        }
        return [];
    }

    _lookupArcNode1 (nVal, iAddr) {
    _lookupArcNode (nVal, iAddr) {
        // looks if nVal is an arc at the node at iAddr, if yes, returns address of next node else None
        while (true) {
            let iEndArcAddr = iAddr+this.nBytesArc;
            let nRawArc = this._convBytesToInteger(this.byDic.slice(iAddr, iEndArcAddr));
            let iEndArcAddr = iAddr+1;
            let nRawArc = this.byDic[iAddr];
            if (nVal == (nRawArc & this._arcMask)) {
                // the value we are looking for
                // we return the address of the next node
                return this._convBytesToInteger(this.byDic.slice(iEndArcAddr, iEndArcAddr+this.nBytesNodeAddress));
                return this.byDic[iEndArcAddr];
            }
            else {
                // value not found
                if (nRawArc & this._lastArcMask) {
                    return null;
                }
                iAddr = iEndArcAddr + this.nBytesNodeAddress;
                iAddr = iEndArcAddr + 1;
            }
        }
    }

    * _getArcs1 (iAddr) {
    * _getArcs (iAddr) {
        // generator: return all arcs at <iAddr> as tuples of (nVal, iAddr)
        while (true) {
            let iEndArcAddr = iAddr+this.nBytesArc;
            let nRawArc = this._convBytesToInteger(this.byDic.slice(iAddr, iEndArcAddr));
            yield [nRawArc & this._arcMask, this._convBytesToInteger(this.byDic.slice(iEndArcAddr, iEndArcAddr+this.nBytesNodeAddress))];
            let iEndArcAddr = iAddr+1;
            let nRawArc = this.byDic[iAddr];
            yield [nRawArc & this._arcMask, this.byDic[iEndArcAddr]];
            if (nRawArc & this._lastArcMask) {
                break;
            }
            iAddr = iEndArcAddr+this.nBytesNodeAddress;
            iAddr = iEndArcAddr+1;
        }
    }

    // VERSION 2
    _morph2 (sWord) {
        // to do
    }

    _stem2 (sWord) {
        // to do
    }

    _lookupArcNode2 (nVal, iAddr) {
        // to do
    }


    // VERSION 3
    _morph3 (sWord) {
        // to do
    }

    _stem3 (sWord) {
        // to do
    }

    _lookupArcNode3 (nVal, iAddr) {
        // to do
    }
}


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