Grammalecte  Check-in [86250e8e6c]

Overview
Comment:[graphspell] ibdawg: code cleaning, remove old code, useless compression versions
Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | descendants | both | graphspell | dict2
Files: files | file ages | folders
SHA3-256: 86250e8e6cc98d64ac880a7d26c35ce787811fc88e9f19d72a9afed0ad538676
User & Date: olr on 2020-11-04 12:02:12
Other Links: branch diff | manifest | tags
Context
2020-11-04
12:21
[graphspell][py] ibdawg: remove binary dict support check-in: 866ec22f7d user: olr tags: dict2, graphspell
12:02
[graphspell] ibdawg: code cleaning, remove old code, useless compression versions check-in: 86250e8e6c user: olr tags: dict2, graphspell
11:37
[build][graphspell][lo] dictionary: drop support for binary file -> use JSON check-in: 05fb167483 user: olr tags: build, dict2, graphspell, lo
Changes

Modified graphspell-js/dawg.js from [525275df92] to [bb21108d9e].

340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
...
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
...
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
                    }
                }
            }
        }
    }

    // BINARY CONVERSION
    createBinaryJSON (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 = {
            "sHeader": "/grammalecte-fsa/",
            "sLangCode": this.sLangCode,
            "sLangName": this.sLangName,
            "sDicName": this.sDicName,
            "sDescription": this.sDescription,
................................................................................
            for (let oNode of this.arcs.values()) {
                oNode.display(nTab+1, lArcVal, bRecur);
            }
        }
    }

    // VERSION 1 =====================================================================================================
    convToBytes1 (nBytesArc, nBytesNodeAddress) {
        /*
            Node scheme:
            - Arc length is defined by nBytesArc
            - Address length is defined by nBytesNodeAddress

            |                Arc                |                         Address of next node                          |
            |                                   |                                                                       |







|
|
<
|
|
|
|
<
<
<







 







|








|
<
<
|
|
|
<







 







|







340
341
342
343
344
345
346
347
348

349
350
351
352



353
354
355
356
357
358
359
...
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376


377
378
379

380
381
382
383
384
385
386
...
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
                    }
                }
            }
        }
    }

    // BINARY CONVERSION
    createBinaryJSON (nCompressionMethod=1) {
        console.log("Write DAWG as an indexable binary dictionary");

        this.nBytesArc = Math.floor( (this.nArcVal.toString(2).length + 2) / 8 ) + 1;     // We add 2 bits. See DawgNode.convToBytes()
        this.nBytesOffset = 0;
        this._calcNumBytesNodeAddress();
        this._calcNodesAddress();



        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;
        }
    }

    _calcNodesAddress () {
        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=1) {


        let sByDic = this.oRoot.convToBytes(this.nBytesArc, this.nBytesNodeAddress);
        for (let oNode of this.dMinimizedNodes.values()) {
            sByDic += oNode.convToBytes(this.nBytesArc, this.nBytesNodeAddress);

        }
        let oJSON = {
            "sHeader": "/grammalecte-fsa/",
            "sLangCode": this.sLangCode,
            "sLangName": this.sLangName,
            "sDicName": this.sDicName,
            "sDescription": this.sDescription,
................................................................................
            for (let oNode of this.arcs.values()) {
                oNode.display(nTab+1, lArcVal, bRecur);
            }
        }
    }

    // VERSION 1 =====================================================================================================
    convToBytes (nBytesArc, nBytesNodeAddress) {
        /*
            Node scheme:
            - Arc length is defined by nBytesArc
            - Address length is defined by nBytesNodeAddress

            |                Arc                |                         Address of next node                          |
            |                                   |                                                                       |

Modified graphspell/dawg.py from [c083d6a347] to [8c6420c0dc].

354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
...
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
...
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
...
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
...
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
                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 _calculateBinary (self, nCompressionMethod):
        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 ))

    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

    def _calcNodesAddress1 (self):
        nBytesNode = self.nBytesArc + self.nBytesNodeAddress
        iAddr = len(self.oRoot.arcs) * nBytesNode
        for oNode in self.lMinimizedNodes:
            oNode.addr = iAddr
            iAddr += max(len(oNode.arcs), 1) * nBytesNode

    def _calcNodesAddress2 (self):
        nBytesNode = self.nBytesArc + self.nBytesNodeAddress
        iAddr = len(self.oRoot.arcs) * nBytesNode
        for oNode in self.lSortedNodes:
            oNode.addr = iAddr
            iAddr += max(len(oNode.arcs), 1) * nBytesNode
            for oNextNode in oNode.arcs.values():
                if (oNode.pos + 1) == oNextNode.pos:
                    iAddr -= self.nBytesNodeAddress
                    #break

    def _calcNodesAddress3 (self):
        nBytesNode = self.nBytesArc + self.nBytesNodeAddress
        # theorical nodes size if only addresses and no offset
        self.oRoot.size = len(self.oRoot.arcs) * nBytesNode
        for oNode in self.lSortedNodes:
            oNode.size = max(len(oNode.arcs), 1) * nBytesNode
        # rewind and calculate dropdown from the end, several times
        nDiff = self.nBytesNodeAddress - self.nBytesOffset
        bEnd = False
        while not bEnd:
            bEnd = True
            # recalculate addresses
            iAddr = self.oRoot.size
            for oNode in self.lSortedNodes:
                oNode.addr = iAddr
                iAddr += oNode.size
            # rewind and calculate dropdown from the end, several times
            for i in range(self.nNode-1, -1, -1):
                nSize = max(len(self.lSortedNodes[i].arcs), 1) * nBytesNode
                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 getBinaryAsJSON (self, nCompressionMethod=1, bBinaryDictAsHexString=True):
        "return a JSON string containing all necessary data of the dictionary (compressed as a binary string)"
        self._calculateBinary(nCompressionMethod)
        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)
        return {
            "sHeader": "/grammalecte-fsa/",
            "sLangCode": self.sLangCode,
            "sLangName": self.sLangName,
            "sDicName": self.sDicName,
            "sDescription": self.sDescription,
            "sFileName": self.sFileName,
................................................................................
        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(self.getBinaryAsJSON(nCompressionMethod, bBinaryDictAsHexString), ensure_ascii=False) )
            if bInJSModule:
                hDst.write(";\n\nexports.dictionary = dictionary;\n")

    def writeBinary (self, sPathFile, nCompressionMethod, bDebug=False):
        """
        Save as a binary file.

        Format of the binary indexable dictionary:
        Each section is separated with 4 bytes of \0

        - Section Header:
            /grammalecte-fsa/[compression method]
                * compression method is an ASCII string

        - Section Informations:
            /[lang code]
            /[lang name]
            /[dictionary name]
            /[date creation]
            /[number of chars]
            /[number of bytes for each arc]
            /[number of bytes for each address node]
            /[number of entries]
            /[number of nodes]
            /[number of arcs]
            /[number of affixes]
                * each field is a ASCII string
            /[stemming code]
                * "S" means stems are generated by /suffix_code/,
                  "A" means they are generated by /affix_code/
                  See defineSuffixCode() and defineAffixCode() for details.
                  "N" means no stemming

        - 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.

        - Section 2grams:
                * A list of 2grams (as strings: 2 chars) encoded in binary from utf-8, each value separated with a tabulation
        """
        self._calculateBinary(nCompressionMethod)
        if not sPathFile.endswith(".bdic"):
            sPathFile += "."+str(nCompressionMethod)+".bdic"
        with open(sPathFile, 'wb') as hDst:
            # header
            hDst.write("/grammalecte-fsa/{}/".format(nCompressionMethod).encode("utf-8"))
            hDst.write(b"\0\0\0\0")
            # infos
            sInfo = "{}//{}//{}//{}//{}//{}//{}//{}//{}//{}//{}//{}//{}".format(self.sLangCode, self.sLangName, self.sDicName, self.sDescription, self._getDate(), \
                                                                                self.nChar, self.nBytesArc, self.nBytesNodeAddress, \
                                                                                self.nEntry, self.nNode, self.nArc, self.nAff, self.cStemming)
            hDst.write(sInfo.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")
            # 2grams
            hDst.write("\t".join(self.a2grams).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))
        if bDebug:
            self._writeNodes(sPathFile, nCompressionMethod)

    def _getDate (self):
        return time.strftime("%Y-%m-%d %H:%M:%S")

    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.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.lArcVal)+"\n")
            if nCompressionMethod == 2:
                hDst.write(self.oRoot.getTxtRepr2(self.nBytesArc, self.lArcVal)+"\n")
                for oNode in self.lSortedNodes:
                    hDst.write(oNode.getTxtRepr2(self.nBytesArc, self.lArcVal)+"\n")
            if nCompressionMethod == 3:
                hDst.write(self.oRoot.getTxtRepr3(self.nBytesArc, 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.nBytesOffset, self.lArcVal)+"\n")



class DawgNode:
    """Node of the word graph"""

    NextId = 0
................................................................................
        self.arcs = collections.OrderedDict(sorted(self.arcs.items(), key=lambda t: dValOccur.get(t[0], 0), reverse=True))

    def sortArcs2 (self, dValOccur, lArcVal):
        "sort arcs of each node depending on the previous char"
        self.arcs = collections.OrderedDict(sorted(self.arcs.items(), key=lambda t: dValOccur.get(lArcVal[t[0]], 0), reverse=True))

    # VERSION 1 =====================================================================================================
    def convToBytes1 (self, nBytesArc, nBytesNodeAddress):
        """
        Convert to bytes (method 1).

        Node scheme:
        - Arc length is defined by nBytesArc
        - Address length is defined by nBytesNodeAddress

................................................................................
                val = val | nFinalNodeMask
            if i == nArc:
                val = val | nFinalArcMask
            by += val.to_bytes(nBytesArc, byteorder='big')
            by += self.arcs[arc].addr.to_bytes(nBytesNodeAddress, byteorder='big')
        return by

    def getTxtRepr1 (self, nBytesArc, lVal):
        "return representation as string of node (method 1)"
        nArc = len(self.arcs)
        nFinalNodeMask = 1 << ((nBytesArc*8)-1)
        nFinalArcMask = 1 << ((nBytesArc*8)-2)
        s = "i{:_>10} -- #{:_>10}\n".format(self.i, self.addr)
        if not nArc:
            s += "  {:<20}  {:0>16}  i{:_>10}   #{:_>10}\n".format("", bin(nFinalNodeMask | nFinalArcMask)[2:], "0", "0")
................................................................................
            val = arc
            if i == 1 and self.final:
                val = val | nFinalNodeMask
            if i == nArc:
                val = val | nFinalArcMask
            s += "  {:<20}  {:0>16}  i{:_>10}   #{:_>10}\n".format(lVal[arc], bin(val)[2:], self.arcs[arc].i, self.arcs[arc].addr)
        return s

    # VERSION 2 =====================================================================================================
    def convToBytes2 (self, nBytesArc, nBytesNodeAddress):
        """
        Convert to bytes (method 2).

        Node scheme:
        - Arc length is defined by nBytesArc
        - Address length is defined by nBytesNodeAddress

        |                Arc                |                         Address of next node                          |
        |                                   |                                                                       |
         ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓
         ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃
         ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛
         [...]
         ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓
         ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃
         ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛
          ^ ^ ^
          ┃ ┃ ┃
          ┃ ┃ ┗━━ if 1, caution, no address: next node is the following node
          ┃ ┗━━━━ if 1, last arc of this node
          ┗━━━━━━ if 1, this node is final (only on the first arc)
        """
        nArc = len(self.arcs)
        nFinalNodeMask = 1 << ((nBytesArc*8)-1)
        nFinalArcMask = 1 << ((nBytesArc*8)-2)
        nNextNodeMask = 1 << ((nBytesArc*8)-3)
        if not nArc:
            val = nFinalNodeMask | nFinalArcMask
            by = val.to_bytes(nBytesArc, byteorder='big')
            by += (0).to_bytes(nBytesNodeAddress, byteorder='big')
            return by
        by = b""
        for i, arc in enumerate(self.arcs, 1):
            val = arc
            if i == 1 and self.final:
                val = val | nFinalNodeMask
            if i == nArc:
                val = val | nFinalArcMask
            if (self.pos + 1) == self.arcs[arc].pos and self.i != 0:
                val = val | nNextNodeMask
                by += val.to_bytes(nBytesArc, byteorder='big')
            else:
                by += val.to_bytes(nBytesArc, byteorder='big')
                by += self.arcs[arc].addr.to_bytes(nBytesNodeAddress, byteorder='big')
        return by

    def getTxtRepr2 (self, nBytesArc, lVal):
        "return representation as string of node (method 2)"
        nArc = len(self.arcs)
        nFinalNodeMask = 1 << ((nBytesArc*8)-1)
        nFinalArcMask = 1 << ((nBytesArc*8)-2)
        nNextNodeMask = 1 << ((nBytesArc*8)-3)
        s = "i{:_>10} -- #{:_>10}\n".format(self.i, self.addr)
        if not nArc:
            s += "  {:<20}  {:0>16}  i{:_>10}   #{:_>10}\n".format("", bin(nFinalNodeMask | nFinalArcMask)[2:], "0", "0")
            return s
        for i, arc in enumerate(self.arcs, 1):
            val = arc
            if i == 1 and self.final:
                val = val | nFinalNodeMask
            if i == nArc:
                val = val | nFinalArcMask
            if (self.pos + 1) == self.arcs[arc].pos  and self.i != 0:
                val = val | nNextNodeMask
                s += "  {:<20}  {:0>16}\n".format(lVal[arc], bin(val)[2:])
            else:
                s += "  {:<20}  {:0>16}  i{:_>10}   #{:_>10}\n".format(lVal[arc], bin(val)[2:], self.arcs[arc].i, self.arcs[arc].addr)
        return s

    # VERSION 3 =====================================================================================================
    def convToBytes3 (self, nBytesArc, nBytesNodeAddress, nBytesOffset):
        """
        Convert to bytes (method 3).

        Node scheme:
        - Arc length is defined by nBytesArc
        - Address length is defined by nBytesNodeAddress
        - Offset length is defined by nBytesOffset

        |                Arc                |            Address of next node  or  offset to next node              |
        |                                   |                                                                       |
         ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓
         ┃1┃0┃0┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃
         ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛
         [...]
         ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓
         ┃0┃0┃1┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃     Offsets are shorter than addresses
         ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛
         ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓ ┏━━━━━━━━━━━━━━━┓
         ┃0┃1┃0┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃ ┃
         ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛ ┗━━━━━━━━━━━━━━━┛

          ^ ^ ^
          ┃ ┃ ┃
          ┃ ┃ ┗━━ if 1, offset instead of address of next node
          ┃ ┗━━━━ if 1, last arc of this node
          ┗━━━━━━ if 1, this node is final (only on the first arc)
        """
        nArc = len(self.arcs)
        nFinalNodeMask = 1 << ((nBytesArc*8)-1)
        nFinalArcMask = 1 << ((nBytesArc*8)-2)
        nNextNodeMask = 1 << ((nBytesArc*8)-3)
        nMaxOffset = (2 ** (nBytesOffset * 8)) - 1
        if not nArc:
            val = nFinalNodeMask | nFinalArcMask
            by = val.to_bytes(nBytesArc, byteorder='big')
            by += (0).to_bytes(nBytesNodeAddress, byteorder='big')
            return by
        by = b""
        for i, arc in enumerate(self.arcs, 1):
            val = arc
            if i == 1 and self.final:
                val = val | nFinalNodeMask
            if i == nArc:
                val = val | nFinalArcMask
            if 1 < (self.arcs[arc].addr - self.addr) < nMaxOffset and self.i != 0:
                val = val | nNextNodeMask
                by += val.to_bytes(nBytesArc, byteorder='big')
                by += (self.arcs[arc].addr-self.addr).to_bytes(nBytesOffset, byteorder='big')
            else:
                by += val.to_bytes(nBytesArc, byteorder='big')
                by += self.arcs[arc].addr.to_bytes(nBytesNodeAddress, byteorder='big')
        return by

    def getTxtRepr3 (self, nBytesArc, nBytesOffset, lVal):
        "return representation as string of node (method 3)"
        nArc = len(self.arcs)
        nFinalNodeMask = 1 << ((nBytesArc*8)-1)
        nFinalArcMask = 1 << ((nBytesArc*8)-2)
        nNextNodeMask = 1 << ((nBytesArc*8)-3)
        nMaxOffset = (2 ** (nBytesOffset * 8)) - 1
        s = "i{:_>10} -- #{:_>10}  ({})\n".format(self.i, self.addr, self.size)
        if not nArc:
            s += "  {:<20}  {:0>16}  i{:_>10}   #{:_>10}\n".format("", bin(nFinalNodeMask | nFinalArcMask)[2:], "0", "0")
            return s
        for i, arc in enumerate(self.arcs, 1):
            val = arc
            if i == 1 and self.final:
                val = val | nFinalNodeMask
            if i == nArc:
                val = val | nFinalArcMask
            if 1 < (self.arcs[arc].addr - self.addr) < nMaxOffset and self.i != 0:
                val = val | nNextNodeMask
                s += "  {:<20}  {:0>16}  i{:_>10}   +{:_>10}\n".format(lVal[arc], bin(val)[2:], self.arcs[arc].i, self.arcs[arc].addr - self.addr)
            else:
                s += "  {:<20}  {:0>16}  i{:_>10}   #{:_>10}\n".format(lVal[arc], bin(val)[2:], self.arcs[arc].i, self.arcs[arc].addr)
        return s



# Another attempt to sort node arcs

_dCharOrder = {
    # key: previous char, value: dictionary of chars {c: nValue}
    "": {}







|
|
<
|
|
|
|
<
<
<
<
<
<
<
<
<
<
<
<
<











|






<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<




<
|
|
|
<
<
<
<
<
<
<
<







 







<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<



|



<
|
|
|
|
<
<
<
<
<
<
<
<
<







 







|







 







|







 







<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<







354
355
356
357
358
359
360
361
362

363
364
365
366













367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384





































385
386
387
388

389
390
391








392
393
394
395
396
397
398
...
425
426
427
428
429
430
431











































































432
433
434
435
436
437
438

439
440
441
442









443
444
445
446
447
448
449
...
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
...
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
...
558
559
560
561
562
563
564























































































































































565
566
567
568
569
570
571
                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 _calculateBinary (self, nCompressionMethod=1):
        print(" > Write DAWG as an indexable binary dictionary")

        self.nBytesArc = ( (self.nArcVal.bit_length() + 2) // 8 ) + 1   # We add 2 bits. See DawgNode.convToBytes()
        self.nBytesOffset = 0
        self._calcNumBytesNodeAddress()
        self._calcNodesAddress()













        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 ))

    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

    def _calcNodesAddress (self):
        nBytesNode = self.nBytesArc + self.nBytesNodeAddress
        iAddr = len(self.oRoot.arcs) * nBytesNode
        for oNode in self.lMinimizedNodes:
            oNode.addr = iAddr
            iAddr += max(len(oNode.arcs), 1) * nBytesNode






































    def getBinaryAsJSON (self, nCompressionMethod=1, bBinaryDictAsHexString=True):
        "return a JSON string containing all necessary data of the dictionary (compressed as a binary string)"
        self._calculateBinary(nCompressionMethod)
        byDic = b""

        byDic = self.oRoot.convToBytes(self.nBytesArc, self.nBytesNodeAddress)
        for oNode in self.lMinimizedNodes:
            byDic += oNode.convToBytes(self.nBytesArc, self.nBytesNodeAddress)








        return {
            "sHeader": "/grammalecte-fsa/",
            "sLangCode": self.sLangCode,
            "sLangName": self.sLangName,
            "sDicName": self.sDicName,
            "sDescription": self.sDescription,
            "sFileName": self.sFileName,
................................................................................
        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(self.getBinaryAsJSON(nCompressionMethod, bBinaryDictAsHexString), ensure_ascii=False) )
            if bInJSModule:
                hDst.write(";\n\nexports.dictionary = dictionary;\n")












































































    def _getDate (self):
        return time.strftime("%Y-%m-%d %H:%M:%S")

    def _writeNodes (self, sPathFile, nCompressionMethod=1):
        "for debugging only"
        print(" > Write nodes")
        with open(sPathFile+".nodes."+str(nCompressionMethod)+".txt", 'w', encoding='utf-8', newline="\n") as hDst:

            hDst.write(self.oRoot.getTxtRepr(self.nBytesArc, self.lArcVal)+"\n")
            #hDst.write( ''.join( [ "%02X " %  z  for z in self.oRoot.convToBytes(self.nBytesArc, self.nBytesNodeAddress) ] ).strip() )
            for oNode in self.lMinimizedNodes:
                hDst.write(oNode.getTxtRepr(self.nBytesArc, self.lArcVal)+"\n")












class DawgNode:
    """Node of the word graph"""

    NextId = 0
................................................................................
        self.arcs = collections.OrderedDict(sorted(self.arcs.items(), key=lambda t: dValOccur.get(t[0], 0), reverse=True))

    def sortArcs2 (self, dValOccur, lArcVal):
        "sort arcs of each node depending on the previous char"
        self.arcs = collections.OrderedDict(sorted(self.arcs.items(), key=lambda t: dValOccur.get(lArcVal[t[0]], 0), reverse=True))

    # VERSION 1 =====================================================================================================
    def convToBytes (self, nBytesArc, nBytesNodeAddress):
        """
        Convert to bytes (method 1).

        Node scheme:
        - Arc length is defined by nBytesArc
        - Address length is defined by nBytesNodeAddress

................................................................................
                val = val | nFinalNodeMask
            if i == nArc:
                val = val | nFinalArcMask
            by += val.to_bytes(nBytesArc, byteorder='big')
            by += self.arcs[arc].addr.to_bytes(nBytesNodeAddress, byteorder='big')
        return by

    def getTxtRepr (self, nBytesArc, lVal):
        "return representation as string of node (method 1)"
        nArc = len(self.arcs)
        nFinalNodeMask = 1 << ((nBytesArc*8)-1)
        nFinalArcMask = 1 << ((nBytesArc*8)-2)
        s = "i{:_>10} -- #{:_>10}\n".format(self.i, self.addr)
        if not nArc:
            s += "  {:<20}  {:0>16}  i{:_>10}   #{:_>10}\n".format("", bin(nFinalNodeMask | nFinalArcMask)[2:], "0", "0")
................................................................................
            val = arc
            if i == 1 and self.final:
                val = val | nFinalNodeMask
            if i == nArc:
                val = val | nFinalArcMask
            s += "  {:<20}  {:0>16}  i{:_>10}   #{:_>10}\n".format(lVal[arc], bin(val)[2:], self.arcs[arc].i, self.arcs[arc].addr)
        return s

























































































































































# Another attempt to sort node arcs

_dCharOrder = {
    # key: previous char, value: dictionary of chars {c: nValue}
    "": {}