Grammalecte  Check-in [a25340628a]

            }
        }
        this.finish();
        this.countNodes();
        this.countArcs();
        this.sortNodeArcs(dValOccur);
        this.displayInfo();
        //this.writeInfo();
        //this.oRoot.display(0, this.lArcVal, true);
    }

    // BUILD DAWG
    insert (aEntry) {
        if (aEntry < this.aPreviousEntry) {
            throw "Error: Words must be inserted in alphabetical order.";

        for (let s of this.lArcVal) {
            console.log(i + ": " + s);
            i++;
        }
    }

    // BINARY CONVERSION
    createBinary (nMethod) {
        console.log("Write DAWG as an indexable binary dictionary [method: "+nMethod+"]");
        if (nMethod == 1) {
            this.nBytesArc = Math.floor( (this.nArcVal.toString(2).length + 2) / 8 ) + 1;     // We add 2 bits. See DawgNode.convToBytes1()
            this._calcNumBytesNodeAddress()
            this._calcNodesAddress1()
        } else {
            console.log("Error: unknown compression method");
        }
        console.log("Arc values (chars, affixes and tags): " + this.nArcVal);
        console.log("Arc size: "+this.nBytesArc+" bytes, Address size: "+this.nBytesNodeAddress+" bytes");

            - Section Values:
                    * a list of strings encoded in binary from utf-8, each value separated with a tabulation
            
            - Section Word Graph (nodes / arcs)
                    * A list of nodes which are a list of arcs with an address of the next node.
                      See DawgNode.convToBytes() for details.
        */






        let sByDic = "";
        if (nMethod == 1) {
            sByDic = this.oRoot.convToBytes1(this.nBytesArc, this.nBytesNodeAddress);
            for (let oNode of this.dMinimizedNodes.values()) {
                sByDic += oNode.convToBytes1(this.nBytesArc, this.nBytesNodeAddress);
            }
        }

        let oJSON = {
            "sName": this.sName,
            "nVersion": this.nMethod,
            "sHeader": this.sHeader,
            "lArcVal": this.lArcVal,
            "nArcVal": this.nArcVal,
            "byDic": sByDic,
            "sLang": this.sLang,
            "nChar": this.nChar,
            "nBytesArc": this.nBytesArc,
            "nBytesNodeAddress": this.nBytesNodeAddress,
            "nEntries": this.nEntry,
            "nNode": this.nNode,
            "nArc": this.nArc,
            "nAff": this.nAff,
            "cStemming": this.cStemming,
            "nTag": this.nTag,
            "dChar": this.dChar,
            "_arcMask": this._arcMask,
            "_finalNodeMask": this._finalNodeMask,
            "_lastArcMask": this._lastArcMask,
            "_addrBitMask": this._addrBitMask,
            "nBytesOffset": this.nBytesOffset
        };
        return oJSON;
    }
}


const oNodeCounter = {
    nNextId: 0,

              |  \___ if 1, last arc of this node
               \_____ if 1, this node is final (only on the first arc)
        */
        let nArc = this.arcs.size;
        let nFinalNodeMask = 1 << ((nBytesArc*8)-1);
        let nFinalArcMask = 1 << ((nBytesArc*8)-2);
        if (this.arcs.size == 0) {
            let nVal = nFinalNodeMask | nFinalArcMask;
            let sBinary = this.convValueToHexString(nVal, nBytesArc);
            sBinary += this.convValueToHexString(0, nBytesNodeAddress);
            return sBinary;
        }
        let sBinary = "";
        let i = 1;
        for (let arc of this.arcs.keys()) {
            let nVal = arc;
            if (i == 1 && this.final) {
                nVal = nVal | nFinalNodeMask;
            }
            if (i == nArc) {
                nVal = nVal | nFinalArcMask;
            }
            i++;
            sBinary += this.convValueToHexString(nVal, nBytesArc);
            sBinary += this.convValueToHexString(this.arcs.get(arc).addr, nBytesNodeAddress);
        }
        return sBinary;
    }




    convValueToHexString (nVal, nByte) {
        // nVal: value to convert, nByte: number of bytes
        let sHexVal = nVal.toString(16); // conversion to hexadecimal string
        //console.log(`value: ${nVal} in ${nByte} bytes`);
        if (sHexVal.length < (nByte*2)) {
            return "0".repeat((nByte*2) - sHexVal.length) + sHexVal;
        } else if (sHexVal.length == (nByte*2)) {
            return sHexVal
        } else {
            throw "Conversion to byte string: value bigger than allowed.";
        }
    }
}



// Another attempt to sort node arcs

const _dCharOrder = new Map([ ["", new Map()] ]);
// key: previous char, value: dictionary of chars {c: nValue}