Grammalecte  Check-in [702572913c]

                    }
                }
            }
        }
    }

    // BINARY CONVERSION
    createBinary (nCompressionMethod) {
        console.log("Write DAWG as an indexable binary dictionary [method: "+nCompressionMethod+"]");
        if (nCompressionMethod == 1) {
            this.nBytesArc = Math.floor( (this.nArcVal.toString(2).length + 2) / 8 ) + 1;     // We add 2 bits. See DawgNode.convToBytes1()
            this.nBytesOffset = 0;
            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");
        console.log("-> " + this.nBytesArc+this.nBytesNodeAddress + " * " + this.nArc + " = " + (this.nBytesArc+this.nBytesNodeAddress)*this.nArc + " bytes");
        return this._createJSON(nCompressionMethod);
    }

    _calcNumBytesNodeAddress () {
        // how many bytes needed to store all nodes/arcs in the binary dictionary
        this.nBytesNodeAddress = 1;
        while (((this.nBytesArc + this.nBytesNodeAddress) * this.nArc) > (2 ** (this.nBytesNodeAddress * 8))) {
            this.nBytesNodeAddress += 1;
        }
    }

    _calcNodesAddress1 () {
        let nBytesNode = this.nBytesArc + this.nBytesNodeAddress;
        let iAddr = this.oRoot.arcs.size * nBytesNode;
        for (let oNode of this.dMinimizedNodes.values()) {
            oNode.addr = iAddr;
            iAddr += Math.max(oNode.arcs.size, 1) * nBytesNode;
        }
    }

    _createJSON (nCompressionMethod) {
        let sByDic = "";
        if (nCompressionMethod == 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,
            "nCompressionMethod": nCompressionMethod,
            "sDate": this._getDate(),
            "sHeader": this.sHeader + nCompressionMethod + "/",
            "lArcVal": this.lArcVal,
            "nArcVal": this.nArcVal,
            "byDic": sByDic,  // binary word graph
            "sLang": this.sLang,
            "nChar": this.nChar,
            "nBytesArc": this.nBytesArc,
            "nBytesNodeAddress": this.nBytesNodeAddress,

            Object.assign(this, oData);
        }
        catch (e) {
            throw Error("# Error. File not found or not loadable.\n" + e.message + "\n");
        }
        /*
            Properties:
            sName, nCompressionMethod, sHeader, lArcVal, nArcVal, byDic, sLang, nChar, nBytesArc, nBytesNodeAddress,
            nEntries, 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.byDic.length;  i+=2) {
            lTemp.push(parseInt(this.byDic.slice(i, i+2), 16));
        }
        this.byDic = lTemp;
        /* end of bug workaround */

        if (!this.sHeader.startsWith("/pyfsa/")) {
            throw TypeError("# Error. Not a pyfsa 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.byDic = new Uint8Array(this.byDic);  // not quicker, even slower

        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;
        }

        this._arcMask = (2 ** ((this.nBytesArc * 8) - 3)) - 1;
        this._finalNodeMask = 1 << ((this.nBytesArc * 8) - 1);
        this._lastArcMask = 1 << ((this.nBytesArc * 8) - 2);


        // Configuring DAWG functions according to nCompressionMethod
        switch (this.nCompressionMethod) {
            case 1:
                this.morph = this._morph1;
                this.stem = this._stem1;
                this._lookupArcNode = this._lookupArcNode1;
                this._getArcs = this._getArcs1;
                this._writeNodes = this._writeNodes1;
                break;

                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.bOptNumSigle = true;
        this.bOptNumAtLast = false;
    }

    getInfo () {
        return  `  Language: ${this.sLang}   Version: ${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.nEntries} entries,    ${this.nNode} nodes,   ${this.nArc} arcs\n` +
                `  Address size: ${this.nBytesNodeAddress} bytes,  Arc size: ${this.nBytesArc} bytes\n`;
    }

    isValidToken (sToken) {
        // checks if sToken is valid (if there is hyphens in sToken, sToken is split, each part is checked)

                sEntry = sWord + "\t" + self.funcStemming(sWord, self.lArcVal[nVal])
                for nMorphVal, _ in oNextNode.arcs.items():
                    if not zPattern or zPattern.search(self.lArcVal[nMorphVal]):
                        yield sEntry + "\t" + self.lArcVal[nMorphVal]


    # BINARY CONVERSION
    def createBinary (self, sPathFile, nCompressionMethod, bDebug=False):
        print(" > Write DAWG as an indexable binary dictionary [method: %d]" % nCompressionMethod)
        if nCompressionMethod == 1:
            self.nBytesArc = ( (self.nArcVal.bit_length() + 2) // 8 ) + 1   # We add 2 bits. See DawgNode.convToBytes1()
            self.nBytesOffset = 0
            self._calcNumBytesNodeAddress()
            self._calcNodesAddress1()
        elif nCompressionMethod == 2:
            self.nBytesArc = ( (self.nArcVal.bit_length() + 3) // 8 ) + 1   # We add 3 bits. See DawgNode.convToBytes2()
            self.nBytesOffset = 0
            self._calcNumBytesNodeAddress()
            self._calcNodesAddress2()
        elif nCompressionMethod == 3:
            self.nBytesArc = ( (self.nArcVal.bit_length() + 3) // 8 ) + 1   # We add 3 bits. See DawgNode.convToBytes3()
            self.nBytesOffset = 1
            self.nMaxOffset = (2 ** (self.nBytesOffset * 8)) - 1
            self._calcNumBytesNodeAddress()
            self._calcNodesAddress3()
        else:
            print(" # Error: unknown compression method")
        print("   Arc values (chars, affixes and tags): {}  ->  {} bytes".format( self.nArcVal, len("\t".join(self.lArcVal).encode("utf-8")) ))
        print("   Arc size: {} bytes, Address size: {} bytes   ->   {} * {} = {} bytes".format( self.nBytesArc, self.nBytesNodeAddress, \
                                                                                                self.nBytesArc+self.nBytesNodeAddress, self.nArc, \
                                                                                                (self.nBytesArc+self.nBytesNodeAddress)*self.nArc ))
        self._writeBinary(sPathFile, nCompressionMethod)
        self._writeAsJSObject(sPathFile, nCompressionMethod)
        if bDebug:
            self._writeNodes(sPathFile, nCompressionMethod)

    def _calcNumBytesNodeAddress (self):
        "how many bytes needed to store all nodes/arcs in the binary dictionary"
        self.nBytesNodeAddress = 1
        while ((self.nBytesArc + self.nBytesNodeAddress) * self.nArc) > (2 ** (self.nBytesNodeAddress * 8)):
            self.nBytesNodeAddress += 1

                for oNextNode in self.lSortedNodes[i].arcs.values():
                    if 1 < (oNextNode.addr - self.lSortedNodes[i].addr) < self.nMaxOffset:
                        nSize -= nDiff
                if self.lSortedNodes[i].size != nSize:
                    self.lSortedNodes[i].size = nSize
                    bEnd = False

    def _writeAsJSObject (self, spfDst, nCompressionMethod, bInJSModule=False, bBinaryDictAsHexString=True):
        if not spfDst.endswith(".json"):
            spfDst += "."+str(nCompressionMethod)+".json"
        byDic = b""
        if nCompressionMethod == 1:
            byDic = self.oRoot.convToBytes1(self.nBytesArc, self.nBytesNodeAddress)
            for oNode in self.lMinimizedNodes:
                byDic += oNode.convToBytes1(self.nBytesArc, self.nBytesNodeAddress)
        elif nCompressionMethod == 2:
            byDic = self.oRoot.convToBytes2(self.nBytesArc, self.nBytesNodeAddress)
            for oNode in self.lSortedNodes:
                byDic += oNode.convToBytes2(self.nBytesArc, self.nBytesNodeAddress)
        elif nCompressionMethod == 3:
            byDic = self.oRoot.convToBytes3(self.nBytesArc, self.nBytesNodeAddress, self.nBytesOffset)
            for oNode in self.lSortedNodes:
                byDic += oNode.convToBytes3(self.nBytesArc, self.nBytesNodeAddress, self.nBytesOffset)

        with open(spfDst, "w", encoding="utf-8", newline="\n") as hDst:
            if bInJSModule:
                hDst.write('// JavaScript\n// Generated data (do not edit)\n\n"use strict";\n\nconst dictionary = ')
            hDst.write(json.dumps({
                            "sName": "todo",
                            "nCompressionMethod": nCompressionMethod,
                            "sDate": str(datetime.datetime.now())[:-7],
                            "sHeader": "/pyfsa/"+str(nCompressionMethod)+"/",
                            "lArcVal": self.lArcVal,
                            "nArcVal": self.nArcVal,
                            # JavaScript is a pile of shit, so Mozilla’s JS parser don’t like file bigger than 4 Mb!
                            # So, if necessary, we use an hexadecimal string, that we will convert later in Firefox’s extension.
                            # https://github.com/mozilla/addons-linter/issues/1361
                            "byDic": byDic.hex()  if bBinaryDictAsHexString  else [ e  for e in byDic ],
                            "sLang": self.sLang,

                            "nTag": self.nTag,
                            "dChar": self.dChar,
                            "nBytesOffset": self.nBytesOffset
                        }, ensure_ascii=False))
            if bInJSModule:
                hDst.write(";\n\nexports.dictionary = dictionary;\n")

    def _writeBinary (self, sPathFile, nCompressionMethod):
        """
        Format of the binary indexable dictionary:
        Each section is separated with 4 bytes of \0
        
        - Section Header:
            /pyfsa/[compression method]
                * compression method is an ASCII string
        
        - Section Informations:
            /[tag_lang]
            /[number of chars]
            /[number of bytes for each arc]
            /[number of bytes for each address node]
            /[number of entries]

                * 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.
        """
        if not sPathFile.endswith(".bdic"):
            sPathFile += "."+str(nCompressionMethod)+".bdic"
        with open(sPathFile, 'wb') as hDst:
            # header
            hDst.write("/pyfsa/{}/".format(nCompressionMethod).encode("utf-8"))
            hDst.write(b"\0\0\0\0")
            # infos
            hDst.write("{}/{}/{}/{}/{}/{}/{}/{}/{}".format(self.sLang, self.nChar, self.nBytesArc, self.nBytesNodeAddress, \
                                                           self.nEntry, self.nNode, self.nArc, self.nAff, self.cStemming).encode("utf-8"))
            hDst.write(b"\0\0\0\0")
            # lArcVal
            hDst.write("\t".join(self.lArcVal).encode("utf-8"))
            hDst.write(b"\0\0\0\0")
            # DAWG: nodes / arcs
            if nCompressionMethod == 1:
                hDst.write(self.oRoot.convToBytes1(self.nBytesArc, self.nBytesNodeAddress))
                for oNode in self.lMinimizedNodes:
                    hDst.write(oNode.convToBytes1(self.nBytesArc, self.nBytesNodeAddress))
            elif nCompressionMethod == 2:
                hDst.write(self.oRoot.convToBytes2(self.nBytesArc, self.nBytesNodeAddress))
                for oNode in self.lSortedNodes:
                    hDst.write(oNode.convToBytes2(self.nBytesArc, self.nBytesNodeAddress))
            elif nCompressionMethod == 3:
                hDst.write(self.oRoot.convToBytes3(self.nBytesArc, self.nBytesNodeAddress, self.nBytesOffset))
                for oNode in self.lSortedNodes:
                    hDst.write(oNode.convToBytes3(self.nBytesArc, self.nBytesNodeAddress, self.nBytesOffset))
            hDst.close()

    def _writeNodes (self, sPathFile, nCompressionMethod):
        "for debugging only"
        print(" > Write nodes")
        with open(sPathFile+".nodes."+str(nCompressionMethod)+".txt", 'w', encoding='utf-8', newline="\n") as hDst:
            if nCompressionMethod == 1:
                hDst.write(self.oRoot.getTxtRepr1(self.nBytesArc, self.nBytesNodeAddress, self.lArcVal)+"\n")
                #hDst.write( ''.join( [ "%02X " %  z  for z in self.oRoot.convToBytes1(self.nBytesArc, self.nBytesNodeAddress) ] ).strip() )
                for oNode in self.lMinimizedNodes:
                    hDst.write(oNode.getTxtRepr1(self.nBytesArc, self.nBytesNodeAddress, self.lArcVal)+"\n")
            if nCompressionMethod == 2:
                hDst.write(self.oRoot.getTxtRepr2(self.nBytesArc, self.nBytesNodeAddress, self.lArcVal)+"\n")
                for oNode in self.lSortedNodes:
                    hDst.write(oNode.getTxtRepr2(self.nBytesArc, self.nBytesNodeAddress, self.lArcVal)+"\n")
            if nCompressionMethod == 3:
                hDst.write(self.oRoot.getTxtRepr3(self.nBytesArc, self.nBytesNodeAddress, self.nBytesOffset, self.lArcVal)+"\n")
                #hDst.write( ''.join( [ "%02X " %  z  for z in self.oRoot.convToBytes3(self.nBytesArc, self.nBytesNodeAddress, self.nBytesOffset) ] ).strip() )
                for oNode in self.lSortedNodes:
                    hDst.write(oNode.getTxtRepr3(self.nBytesArc, self.nBytesNodeAddress, self.nBytesOffset, self.lArcVal)+"\n")
            hDst.close()
    
    def writeResults (self, sPathFile):

            raise ValueError("# Error. Unknown dictionary version: {}".format(self.by[7:8]))
        try:
            header, info, values, bdic = self.by.split(b"\0\0\0\0", 3)
        except Exception:
            raise Exception

        self.sName = sDicName
        self.nCompressionMethod = int(self.by[7:8].decode("utf-8"))
        self.sHeader = header.decode("utf-8")
        self.lArcVal = values.decode("utf-8").split("\t")
        self.nArcVal = len(self.lArcVal)
        self.byDic = bdic

        l = info.decode("utf-8").split("/")
        self.sLang = l[0]

        self._arcMask = (2 ** ((self.nBytesArc * 8) - 3)) - 1
        self._finalNodeMask = 1 << ((self.nBytesArc * 8) - 1)
        self._lastArcMask = 1 << ((self.nBytesArc * 8) - 2)
        self._addrBitMask = 1 << ((self.nBytesArc * 8) - 3)  # version 2

        self.nBytesOffset = 1 # version 3

        # Configuring DAWG functions according to nCompressionMethod
        if self.nCompressionMethod == 1:
            self.morph = self._morph1
            self.stem = self._stem1
            self._lookupArcNode = self._lookupArcNode1
            self._getArcs = self._getArcs1
            self._writeNodes = self._writeNodes1
        elif self.nCompressionMethod == 2:
            self.morph = self._morph2
            self.stem = self._stem2
            self._lookupArcNode = self._lookupArcNode2
            self._getArcs = self._getArcs2
            self._writeNodes = self._writeNodes2
        elif self.nCompressionMethod == 3:
            self.morph = self._morph3
            self.stem = self._stem3
            self._lookupArcNode = self._lookupArcNode3
            self._getArcs = self._getArcs3
            self._writeNodes = self._writeNodes3
        else:
            raise ValueError("  # Error: unknown code: {}".format(self.nCompressionMethod))

        self.bOptNumSigle = False
        self.bOptNumAtLast = False

    def getInfo (self):
        return  "  Language: {0.sLang:>10}   Version: {0.nCompressionMethod:>2}   Date: {0.sDate}   Stemming: {0.cStemming}FX\n" \
                "  Arcs values:  {0.nArcVal:>10,} = {0.nChar:>5,} characters,  {0.nAff:>6,} affixes,  {0.nTag:>6,} tags\n" \
                "  Dictionary: {0.nEntries:>12,} entries,    {0.nNode:>11,} nodes,   {0.nArc:>11,} arcs\n" \
                "  Address size: {0.nBytesNodeAddress:>1} bytes,  Arc size: {0.nBytesArc:>1} bytes\n".format(self)

    def writeAsJSObject (self, spfDest, bInJSModule=False, bBinaryDictAsHexString=False):
        "write IBDAWG as a JavaScript object in a JavaScript module"
        import json
        with open(spfDest, "w", encoding="utf-8", newline="\n") as hDst:
            if bInJSModule:
                hDst.write('// JavaScript\n// Generated data (do not edit)\n\n"use strict";\n\nconst dictionary = ')
            hDst.write(json.dumps({
                            "sName": self.sName,
                            "nCompressionMethod": self.nCompressionMethod,
                            "sDate": str(datetime.datetime.now())[:-7],
                            "sHeader": self.sHeader,
                            "lArcVal": self.lArcVal,
                            "nArcVal": self.nArcVal,
                            # JavaScript is a pile of shit, so Mozilla’s JS parser don’t like file bigger than 4 Mb!
                            # So, if necessary, we use an hexadecimal string, that we will convert later in Firefox’s extension.
                            # https://github.com/mozilla/addons-linter/issues/1361