Grammalecte  Check-in [f3f1cc9e30]

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

Overview
Comment:[graphspell] better quicker spelling suggestions: check 2grams existence
Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | descendants | both | trunk | graphspell
Files: files | file ages | folders
SHA3-256:f3f1cc9e3090ebc5b7e59c28517cb6ccb4a8015c052b8b8fbd7eff2e0f505eee
User & Date: olr 2018-10-06 13:46:58
Context
2018-10-09
08:54
[graphspell][js] performance tests check-in: 14ed269c7c user: olr tags: graphspell, trunk
2018-10-06
13:46
[graphspell] better quicker spelling suggestions: check 2grams existence check-in: f3f1cc9e30 user: olr tags: graphspell, trunk
13:46
[fr] passation de pouvoir check-in: 4292f525f4 user: olr tags: fr, trunk
Changes

Changes to graphspell-js/dawg.js.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
..
37
38
39
40
41
42
43
44
45
46
47
48
49
50


51
52



53
54
55
56
57
58
59
..
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
..
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
...
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
...
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
...
271
272
273
274
275
276
277

278
279
280
281
282
283
284
...
392
393
394
395
396
397
398
399

400
401
402
403
404
405
406
...
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
// JavaScript

// FSA DICTIONARY BUILDER
//
// by Olivier R.
// License: MPL 2
//
// This tool encodes lexicon into an indexable binary dictionary 
// Input files MUST be encoded in UTF-8.

"use strict";


if (typeof(require) !== 'undefined') {
    var str_transform = require("resource://grammalecte/graphspell/str_transform.js");
................................................................................
            case "S":
                funcStemmingGen = str_transform.defineSuffixCode; break;
            case "N":
                funcStemmingGen = str_transform.noStemming; break;
            default:
                throw "Error. Unknown stemming code: " + cStemming;
        }
        
        let lEntry = [];
        let lChar = [''],  dChar = new Map(),  nChar = 1,  dCharOccur = new Map();
        let lAff  = [],    dAff  = new Map(),  nAff  = 0,  dAffOccur = new Map();
        let lTag  = [],    dTag  = new Map(),  nTag  = 0,  dTagOccur = new Map();
        let nErr = 0;
        


        // read lexicon
        for (let [sFlex, sStem, sTag] of lEntrySrc) {



            addWordToCharDict(sFlex);
            // chars
            for (let c of sFlex) {
                if (!dChar.get(c)) {
                    dChar.set(c, nChar);
                    lChar.push(c);
                    nChar += 1;
................................................................................
        if (lEntry.length == 0) {
            throw "Error. Empty lexicon";
        }

        lEntry = [...new Set(lEntry.map(e => JSON.stringify(e)))].map(s => JSON.parse(s));
        // Set can’t distinguish similar lists, so we transform list item in string given to the Set
        // then we transform items in list a new.
        
        // Preparing DAWG
        console.log(" > Preparing list of words");
        let lVal = lChar.concat(lAff).concat(lTag);
        let lWord = [];
        for (let [sFlex, iAff, iTag] of lEntry) {
            let lTemp = [];
            for (let c of sFlex) {
................................................................................
                lTemp.push(dChar.get(c));
            }
            lTemp.push(iAff+nChar);
            lTemp.push(iTag+nChar+nAff)
            lWord.push(lTemp);
        }
        lEntry.length = 0; // clear the array
        
        // Dictionary of arc values occurrency, to sort arcs of each node
        let lKeyVal = [];
        for (let c of dChar.keys()) { lKeyVal.push([dChar.get(c), dCharOccur.get(c)]); }
        for (let sAff of dAff.keys()) { lKeyVal.push([dAff.get(sAff)+nChar, dAffOccur.get(sAff)]); }
        for (let sTag of dTag.keys()) { lKeyVal.push([dTag.get(sTag)+nChar+nAff, dTagOccur.get(sTag)]); }
        let dValOccur = new Map(lKeyVal);
        lKeyVal.length = 0; // clear the array
................................................................................
        if (cStemming == "A") {
            this.funcStemming = str_transform.changeWordWithAffixCode;
        } else if (cStemming == "S") {
            this.funcStemming = str_transform.changeWordWithSuffixCode;
        } else {
            this.funcStemming = str_transform.noStemming;
        }
        
        // build
        lWord.sort();
        if (xProgressBarNode) {
            xProgressBarNode.value = 0;
            xProgressBarNode.max = lWord.length;
        }
        let i = 1;
................................................................................

    countArcs () {
        this.nArc = 0;
        for (let oNode of this.dMinimizedNodes.values()) {
            this.nArc += oNode.arcs.size;
        }
    }
    
    sortNodeArcs (dValOccur) {
        console.log(" > Sort node arcs");
        this.oRoot.sortArcs(dValOccur);
        for (let oNode of this.dMinimizedNodes.values()) {
            oNode.sortArcs(dValOccur);
        }
    }
................................................................................
        console.log("Entries: " + this.nEntry);
        console.log("Characters: " + this.nChar);
        console.log("Affixes: " + this.nAff);
        console.log("Tags: " + this.nTag);
        console.log("Arc values: " + this.nArcVal);
        console.log("Nodes: " + this.nNode);
        console.log("Arcs: " + this.nArc);

        console.log("Stemming: " + this.cStemming + "FX");
    }

    getArcStats () {
        let d = new Map();
        for (let oNode of this.dMinimizedNodes.values()) {
            let n = oNode.arcs.size;
................................................................................
            "nArc": this.nArc,
            "lArcVal": this.lArcVal,
            "nArcVal": this.nArcVal,
            "nCompressionMethod": nCompressionMethod,
            "nBytesArc": this.nBytesArc,
            "nBytesNodeAddress": this.nBytesNodeAddress,
            "nBytesOffset": this.nBytesOffset,
            "sByDic": sByDic    // binary word graph

        };
        return oJSON;
    }

    _getDate () {
        let oDate = new Date();
        let sMonth = (oDate.getMonth() + 1).toString().padStart(2, "0"); // Month+1: Because JS always sucks somehow.
................................................................................

    // VERSION 1 =====================================================================================================
    convToBytes1 (nBytesArc, nBytesNodeAddress) {
        /*
            Node scheme:
            - Arc length is defined by nBytesArc
            - Address length is defined by nBytesNodeAddress
                                           
            |                Arc                |                         Address of next node                          |
            |                                   |                                                                       |
             /---------------\ /---------------\ /---------------\ /---------------\ /---------------\ /---------------\
             | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
             \---------------/ \---------------/ \---------------/ \---------------/ \---------------/ \---------------/
             [...]
             /---------------\ /---------------\ /---------------\ /---------------\ /---------------\ /---------------\







|







 







|





|
>
>


>
>
>







 







|







 







|







 







|







 







|







 







>







 







|
>







 







|







1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
..
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
..
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
...
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
...
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
...
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
...
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
...
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
...
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
// JavaScript

// FSA DICTIONARY BUILDER
//
// by Olivier R.
// License: MPL 2
//
// This tool encodes lexicon into an indexable binary dictionary
// Input files MUST be encoded in UTF-8.

"use strict";


if (typeof(require) !== 'undefined') {
    var str_transform = require("resource://grammalecte/graphspell/str_transform.js");
................................................................................
            case "S":
                funcStemmingGen = str_transform.defineSuffixCode; break;
            case "N":
                funcStemmingGen = str_transform.noStemming; break;
            default:
                throw "Error. Unknown stemming code: " + cStemming;
        }

        let lEntry = [];
        let lChar = [''],  dChar = new Map(),  nChar = 1,  dCharOccur = new Map();
        let lAff  = [],    dAff  = new Map(),  nAff  = 0,  dAffOccur = new Map();
        let lTag  = [],    dTag  = new Map(),  nTag  = 0,  dTagOccur = new Map();
        let nErr = 0;

        this.a2grams = new Set();

        // read lexicon
        for (let [sFlex, sStem, sTag] of lEntrySrc) {
            for (let s2grams of str_transform.getNgrams(sFlex)) {
                this.a2grams.add(s2grams);
            }
            addWordToCharDict(sFlex);
            // chars
            for (let c of sFlex) {
                if (!dChar.get(c)) {
                    dChar.set(c, nChar);
                    lChar.push(c);
                    nChar += 1;
................................................................................
        if (lEntry.length == 0) {
            throw "Error. Empty lexicon";
        }

        lEntry = [...new Set(lEntry.map(e => JSON.stringify(e)))].map(s => JSON.parse(s));
        // Set can’t distinguish similar lists, so we transform list item in string given to the Set
        // then we transform items in list a new.

        // Preparing DAWG
        console.log(" > Preparing list of words");
        let lVal = lChar.concat(lAff).concat(lTag);
        let lWord = [];
        for (let [sFlex, iAff, iTag] of lEntry) {
            let lTemp = [];
            for (let c of sFlex) {
................................................................................
                lTemp.push(dChar.get(c));
            }
            lTemp.push(iAff+nChar);
            lTemp.push(iTag+nChar+nAff)
            lWord.push(lTemp);
        }
        lEntry.length = 0; // clear the array

        // Dictionary of arc values occurrency, to sort arcs of each node
        let lKeyVal = [];
        for (let c of dChar.keys()) { lKeyVal.push([dChar.get(c), dCharOccur.get(c)]); }
        for (let sAff of dAff.keys()) { lKeyVal.push([dAff.get(sAff)+nChar, dAffOccur.get(sAff)]); }
        for (let sTag of dTag.keys()) { lKeyVal.push([dTag.get(sTag)+nChar+nAff, dTagOccur.get(sTag)]); }
        let dValOccur = new Map(lKeyVal);
        lKeyVal.length = 0; // clear the array
................................................................................
        if (cStemming == "A") {
            this.funcStemming = str_transform.changeWordWithAffixCode;
        } else if (cStemming == "S") {
            this.funcStemming = str_transform.changeWordWithSuffixCode;
        } else {
            this.funcStemming = str_transform.noStemming;
        }

        // build
        lWord.sort();
        if (xProgressBarNode) {
            xProgressBarNode.value = 0;
            xProgressBarNode.max = lWord.length;
        }
        let i = 1;
................................................................................

    countArcs () {
        this.nArc = 0;
        for (let oNode of this.dMinimizedNodes.values()) {
            this.nArc += oNode.arcs.size;
        }
    }

    sortNodeArcs (dValOccur) {
        console.log(" > Sort node arcs");
        this.oRoot.sortArcs(dValOccur);
        for (let oNode of this.dMinimizedNodes.values()) {
            oNode.sortArcs(dValOccur);
        }
    }
................................................................................
        console.log("Entries: " + this.nEntry);
        console.log("Characters: " + this.nChar);
        console.log("Affixes: " + this.nAff);
        console.log("Tags: " + this.nTag);
        console.log("Arc values: " + this.nArcVal);
        console.log("Nodes: " + this.nNode);
        console.log("Arcs: " + this.nArc);
        console.log("2grams: " + this.a2grams.size);
        console.log("Stemming: " + this.cStemming + "FX");
    }

    getArcStats () {
        let d = new Map();
        for (let oNode of this.dMinimizedNodes.values()) {
            let n = oNode.arcs.size;
................................................................................
            "nArc": this.nArc,
            "lArcVal": this.lArcVal,
            "nArcVal": this.nArcVal,
            "nCompressionMethod": nCompressionMethod,
            "nBytesArc": this.nBytesArc,
            "nBytesNodeAddress": this.nBytesNodeAddress,
            "nBytesOffset": this.nBytesOffset,
            "sByDic": sByDic,    // binary word graph
            "l2grams": Array.from(this.a2grams)
        };
        return oJSON;
    }

    _getDate () {
        let oDate = new Date();
        let sMonth = (oDate.getMonth() + 1).toString().padStart(2, "0"); // Month+1: Because JS always sucks somehow.
................................................................................

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

            |                Arc                |                         Address of next node                          |
            |                                   |                                                                       |
             /---------------\ /---------------\ /---------------\ /---------------\ /---------------\ /---------------\
             | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
             \---------------/ \---------------/ \---------------/ \---------------/ \---------------/ \---------------/
             [...]
             /---------------\ /---------------\ /---------------\ /---------------\ /---------------\ /---------------\

Changes to graphspell-js/ibdawg.js.

44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
...
135
136
137
138
139
140
141

142
143
144
145
146
147
148
...
210
211
212
213
214
215
216
217

218
219
220
221
222
223
224
...
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
...
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
...
393
394
395
396
397
398
399







400
401
402
403
404
405
406
                    this.dSugg.set(nDist, []);
                }
                this.dSugg.get(nDist).push(sSugg);
                this.aSugg.add(sSugg);
                if (nDist < this.nMinDist) {
                    this.nMinDist = nDist;
                }
                this.nDistLimit = Math.min(this.nDistLimit, this.nMinDist+2);
            }
        }
    }

    getSuggestions (nSuggLimit=10, nDistLimit=-1) {
        // return a list of suggestions
        let lRes = [];
................................................................................
        }
        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();


        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;
................................................................................
            "nArc": this.nArc,
            "lArcVal": this.lArcVal,
            "nArcVal": this.nArcVal,
            "nCompressionMethod": this.nCompressionMethod,
            "nBytesArc": this.nBytesArc,
            "nBytesNodeAddress": this.nBytesNodeAddress,
            "nBytesOffset": this.nBytesOffset,
            "sByDic": this.sByDic    // binary word graph

        };
        return oJSON;
    }

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

        let cCurrent = sRemain.slice(0, 1);
        for (let [cChar, jAddr] of this._getCharArcs(iAddr)) {
            if (char_player.d1to1.gl_get(cCurrent, cCurrent).indexOf(cChar) != -1) {
                this._suggest(oSuggResult, sRemain.slice(1), nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, jAddr, sNewWord+cChar);
            }
            else if (!bAvoidLoop) {
                if (nMaxHardRepl) {
                    this._suggest(oSuggResult, sRemain.slice(1), nMaxSwitch, nMaxDel, nMaxHardRepl-1, nMaxJump, nDist+1, nDeep+1, jAddr, sNewWord+cChar, true);
                }
                if (nMaxJump) {
                    this._suggest(oSuggResult, sRemain, nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump-1, nDist+1, nDeep+1, jAddr, sNewWord+cChar, true);
                }
            }
        }
................................................................................
            if (sRemain.length > 1) {
                if (cCurrent == sRemain.slice(1, 2)) {
                    // same char, we remove 1 char without adding 1 to <sNewWord>
                    this._suggest(oSuggResult, sRemain.slice(1), nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord);
                }
                else {
                    // switching chars
                    if (nMaxSwitch > 0) {
                        this._suggest(oSuggResult, sRemain.slice(1, 2)+sRemain.slice(0, 1)+sRemain.slice(2), nMaxSwitch-1, nMaxDel, nMaxHardRepl, nMaxJump, nDist+1, nDeep+1, iAddr, sNewWord, true);
                    }
                    // delete char
                    if (nMaxDel > 0) {
                        this._suggest(oSuggResult, sRemain.slice(1), nMaxSwitch, nMaxDel-1, nMaxHardRepl, nMaxJump, nDist+1, nDeep+1, iAddr, sNewWord, true);
                    }
                }
                // Phonetic replacements
                for (let sRepl of char_player.get1toXReplacement(sNewWord.slice(-1), cCurrent, sRemain.slice(1,2))) {
                    this._suggest(oSuggResult, sRepl + sRemain.slice(1), nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, true);
                }
................................................................................
                this._suggest(oSuggResult, "", nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, true); // remove last char and go on
                for (let sRepl of char_player.dFinal1.gl_get(sRemain, [])) {
                    this._suggest(oSuggResult, sRepl, nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, true);
                }
            }
        }
    }








    * _getCharArcs (iAddr) {
        // generator: yield all chars and addresses from node at address <iAddr>
        for (let [nVal, jAddr] of this._getArcs(iAddr)) {
            if (nVal <= this.nChar) {
                yield [this.dCharVal.get(nVal), jAddr];
            }







|







 







>







 







|
>







 







|







 







|



|







 







>
>
>
>
>
>
>







44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
...
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
...
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
...
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
...
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
...
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
                    this.dSugg.set(nDist, []);
                }
                this.dSugg.get(nDist).push(sSugg);
                this.aSugg.add(sSugg);
                if (nDist < this.nMinDist) {
                    this.nMinDist = nDist;
                }
                this.nDistLimit = Math.min(this.nDistLimit, this.nMinDist+1);
            }
        }
    }

    getSuggestions (nSuggLimit=10, nDistLimit=-1) {
        // return a list of suggestions
        let lRes = [];
................................................................................
        }
        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 = new Set(this.l2grams);

        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;
................................................................................
            "nArc": this.nArc,
            "lArcVal": this.lArcVal,
            "nArcVal": this.nArcVal,
            "nCompressionMethod": this.nCompressionMethod,
            "nBytesArc": this.nBytesArc,
            "nBytesNodeAddress": this.nBytesNodeAddress,
            "nBytesOffset": this.nBytesOffset,
            "sByDic": this.sByDic,  // binary word graph
            "l2grams": this.l2grams
        };
        return oJSON;
    }

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

        let cCurrent = sRemain.slice(0, 1);
        for (let [cChar, jAddr] of this._getCharArcs(iAddr)) {
            if (char_player.d1to1.gl_get(cCurrent, cCurrent).indexOf(cChar) != -1) {
                this._suggest(oSuggResult, sRemain.slice(1), nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, jAddr, sNewWord+cChar);
            }
            else if (!bAvoidLoop) {
                if (nMaxHardRepl  &&  this.isNgramsOK(cChar+sRemain.slice(1,2))) {
                    this._suggest(oSuggResult, sRemain.slice(1), nMaxSwitch, nMaxDel, nMaxHardRepl-1, nMaxJump, nDist+1, nDeep+1, jAddr, sNewWord+cChar, true);
                }
                if (nMaxJump) {
                    this._suggest(oSuggResult, sRemain, nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump-1, nDist+1, nDeep+1, jAddr, sNewWord+cChar, true);
                }
            }
        }
................................................................................
            if (sRemain.length > 1) {
                if (cCurrent == sRemain.slice(1, 2)) {
                    // same char, we remove 1 char without adding 1 to <sNewWord>
                    this._suggest(oSuggResult, sRemain.slice(1), nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord);
                }
                else {
                    // switching chars
                    if (nMaxSwitch > 0  &&  this.isNgramsOK(sNewWord.slice(-1)+sRemain.slice(1,2))  &&  this.isNgramsOK(sRemain.slice(1,2)+sRemain.slice(0,1))) {
                        this._suggest(oSuggResult, sRemain.slice(1, 2)+sRemain.slice(0, 1)+sRemain.slice(2), nMaxSwitch-1, nMaxDel, nMaxHardRepl, nMaxJump, nDist+1, nDeep+1, iAddr, sNewWord, true);
                    }
                    // delete char
                    if (nMaxDel > 0  &&  this.isNgramsOK(sNewWord.slice(-1)+sRemain.slice(1,2))) {
                        this._suggest(oSuggResult, sRemain.slice(1), nMaxSwitch, nMaxDel-1, nMaxHardRepl, nMaxJump, nDist+1, nDeep+1, iAddr, sNewWord, true);
                    }
                }
                // Phonetic replacements
                for (let sRepl of char_player.get1toXReplacement(sNewWord.slice(-1), cCurrent, sRemain.slice(1,2))) {
                    this._suggest(oSuggResult, sRepl + sRemain.slice(1), nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, true);
                }
................................................................................
                this._suggest(oSuggResult, "", nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, true); // remove last char and go on
                for (let sRepl of char_player.dFinal1.gl_get(sRemain, [])) {
                    this._suggest(oSuggResult, sRepl, nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, true);
                }
            }
        }
    }

    isNgramsOK (sChars) {
        if (sChars.length != 2) {
            return true;
        }
        return this.a2grams.has(sChars);
    }

    * _getCharArcs (iAddr) {
        // generator: yield all chars and addresses from node at address <iAddr>
        for (let [nVal, jAddr] of this._getArcs(iAddr)) {
            if (nVal <= this.nChar) {
                yield [this.dCharVal.get(nVal), jAddr];
            }

Changes to graphspell/dawg.py.

59
60
61
62
63
64
65


66
67
68
69
70
71
72
..
73
74
75
76
77
78
79

80
81
82
83
84
85
86
...
281
282
283
284
285
286
287

288
289
290
291
292
293
294
...
444
445
446
447
448
449
450
451

452
453
454
455
456
457
458
...
494
495
496
497
498
499
500



501
502
503
504
505
506
507
...
511
512
513
514
515
516
517



518
519
520
521
522
523
524

        aEntry = set()
        lChar = ['']; dChar = {}; nChar = 1; dCharOccur = {}
        lAff  = [];   dAff  = {}; nAff  = 0; dAffOccur = {}
        lTag  = [];   dTag  = {}; nTag  = 0; dTagOccur = {}
        nErr = 0



        try:
            zFilter = re.compile(sSelectFilterRegex)  if sSelectFilterRegex  else None
        except:
            print(" # Error. Wrong filter regex. Filter ignored.")
            traceback.print_exc()
            zFilter = None

................................................................................
        # read lexicon
        if type(src) is str:
            iterable = readFile(src)
        else:
            iterable = src
        for sFlex, sStem, sTag in iterable:
            if not zFilter or zFilter.search(sTag):

                addWordToCharDict(sFlex)
                # chars
                for c in sFlex:
                    if c not in dChar:
                        dChar[c] = nChar
                        lChar.append(c)
                        nChar += 1
................................................................................
        print(" * {:<12} {:>16,}".format("Entries:", self.nEntry))
        print(" * {:<12} {:>16,}".format("Characters:", self.nChar))
        print(" * {:<12} {:>16,}".format("Affixes:", self.nAff))
        print(" * {:<12} {:>16,}".format("Tags:", self.nTag))
        print(" * {:<12} {:>16,}".format("Arc values:", self.nArcVal))
        print(" * {:<12} {:>16,}".format("Nodes:", self.nNode))
        print(" * {:<12} {:>16,}".format("Arcs:", self.nArc))

        print(" * {:<12} {:>16}".format("Stemming:", self.cStemming + "FX"))

    def getArcStats (self):
        "return a string with statistics about nodes and arcs"
        d = {}
        for oNode in self.lMinimizedNodes:
            n = len(oNode.arcs)
................................................................................
            "nCompressionMethod": nCompressionMethod,
            "nBytesArc": self.nBytesArc,
            "nBytesNodeAddress": self.nBytesNodeAddress,
            "nBytesOffset": self.nBytesOffset,
            # 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
            "sByDic": byDic.hex()  if bBinaryDictAsHexString  else [ e  for e in byDic ]

        }

    def writeAsJSObject (self, spfDst, nCompressionMethod, bInJSModule=False, bBinaryDictAsHexString=True):
        "write a file (JSON or JS module) with all the necessary data"
        if not spfDst.endswith(".json"):
            spfDst += "."+str(nCompressionMethod)+".json"
        with open(spfDst, "w", encoding="utf-8", newline="\n") as hDst:
................................................................................

        - 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.



        """
        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"))
................................................................................
                                                                              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")



            # 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))







>
>







 







>







 







>







 







|
>







 







>
>
>







 







>
>
>







59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
..
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
...
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
...
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
...
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
...
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535

        aEntry = set()
        lChar = ['']; dChar = {}; nChar = 1; dCharOccur = {}
        lAff  = [];   dAff  = {}; nAff  = 0; dAffOccur = {}
        lTag  = [];   dTag  = {}; nTag  = 0; dTagOccur = {}
        nErr = 0

        self.a2grams = set()

        try:
            zFilter = re.compile(sSelectFilterRegex)  if sSelectFilterRegex  else None
        except:
            print(" # Error. Wrong filter regex. Filter ignored.")
            traceback.print_exc()
            zFilter = None

................................................................................
        # read lexicon
        if type(src) is str:
            iterable = readFile(src)
        else:
            iterable = src
        for sFlex, sStem, sTag in iterable:
            if not zFilter or zFilter.search(sTag):
                self.a2grams.update(st.getNgrams(sFlex))
                addWordToCharDict(sFlex)
                # chars
                for c in sFlex:
                    if c not in dChar:
                        dChar[c] = nChar
                        lChar.append(c)
                        nChar += 1
................................................................................
        print(" * {:<12} {:>16,}".format("Entries:", self.nEntry))
        print(" * {:<12} {:>16,}".format("Characters:", self.nChar))
        print(" * {:<12} {:>16,}".format("Affixes:", self.nAff))
        print(" * {:<12} {:>16,}".format("Tags:", self.nTag))
        print(" * {:<12} {:>16,}".format("Arc values:", self.nArcVal))
        print(" * {:<12} {:>16,}".format("Nodes:", self.nNode))
        print(" * {:<12} {:>16,}".format("Arcs:", self.nArc))
        print(" * {:<12} {:>16,}".format("2grams:", len(self.a2grams)))
        print(" * {:<12} {:>16}".format("Stemming:", self.cStemming + "FX"))

    def getArcStats (self):
        "return a string with statistics about nodes and arcs"
        d = {}
        for oNode in self.lMinimizedNodes:
            n = len(oNode.arcs)
................................................................................
            "nCompressionMethod": nCompressionMethod,
            "nBytesArc": self.nBytesArc,
            "nBytesNodeAddress": self.nBytesNodeAddress,
            "nBytesOffset": self.nBytesOffset,
            # 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
            "sByDic": byDic.hex()  if bBinaryDictAsHexString  else [ e  for e in byDic ],
            "l2grams": list(self.a2grams)
        }

    def writeAsJSObject (self, spfDst, nCompressionMethod, bInJSModule=False, bBinaryDictAsHexString=True):
        "write a file (JSON or JS module) with all the necessary data"
        if not spfDst.endswith(".json"):
            spfDst += "."+str(nCompressionMethod)+".json"
        with open(spfDst, "w", encoding="utf-8", newline="\n") as hDst:
................................................................................

        - 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"))
................................................................................
                                                                              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))

Changes to graphspell/ibdawg.py.

58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
...
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164

165
166
167
168
169
170
171
172
173
...
185
186
187
188
189
190
191

192
193
194
195
196
197
198
...
223
224
225
226
227
228
229
230

231
232
233
234
235
236
237
...
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
...
349
350
351
352
353
354
355





356
357
358
359
360
361
362
            if nDist <= self.nDistLimit:
                if nDist not in self.dSugg:
                    self.dSugg[nDist] = []
                self.dSugg[nDist].append(sSugg)
                self.aSugg.add(sSugg)
                if nDist < self.nMinDist:
                    self.nMinDist = nDist
                self.nDistLimit = min(self.nDistLimit, self.nMinDist+2)

    def getSuggestions (self, nSuggLimit=10):
        "return a list of suggestions"
        if self.dSugg[0]:
            # we sort the better results with the original word
            self.dSugg[0].sort(key=lambda sSugg: st.distanceDamerauLevenshtein(self.sWord, sSugg))
        lRes = self.dSugg.pop(0)
................................................................................
    def _initBinary (self):
        "initialize with binary structure file"
        if self.by[0:17] != b"/grammalecte-fsa/":
            raise TypeError("# Error. Not a grammalecte-fsa binary dictionary. Header: {}".format(self.by[0:9]))
        if not(self.by[17:18] == b"1" or self.by[17:18] == b"2" or self.by[17:18] == b"3"):
            raise ValueError("# Error. Unknown dictionary version: {}".format(self.by[17:18]))
        try:
            header, info, values, bdic = self.by.split(b"\0\0\0\0", 3)
        except Exception:
            raise Exception

        self.nCompressionMethod = int(self.by[17:18].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.sLangCode = l.pop(0)
        self.sLangName = l.pop(0)
        self.sDicName = l.pop(0)
        self.sDate = l.pop(0)
        self.nChar = int(l.pop(0))
        self.nBytesArc = int(l.pop(0))
        self.nBytesNodeAddress = int(l.pop(0))
................................................................................
        self.nBytesOffset = 1 # version 3

    def _initJSON (self, oJSON):
        "initialize with a JSON text file"
        self.__dict__.update(oJSON)
        self.byDic = binascii.unhexlify(self.sByDic)
        self.dCharVal = { v: k  for k, v in self.dChar.items() }


    def getInfo (self):
        "return string about the IBDAWG"
        return  "  Language: {0.sLangName}   Lang code: {0.sLangCode}   Dictionary name: {0.sDicName}" \
                "  Compression method: {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.nEntry:>12,} entries,    {0.nNode:>11,} nodes,   {0.nArc:>11,} arcs\n" \
................................................................................
                "nCompressionMethod": self.nCompressionMethod,
                "nBytesArc": self.nBytesArc,
                "nBytesNodeAddress": self.nBytesNodeAddress,
                "nBytesOffset": self.nBytesOffset,
                # 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
                "sByDic": self.byDic.hex()  if bBinaryDictAsHexString  else [ e  for e in self.byDic ]

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

    def isValidToken (self, sToken):
        "checks if <sToken> is valid (if there is hyphens in <sToken>, <sToken> is split, each part is checked)"
        sToken = cp.spellingNormalization(sToken)
................................................................................
        if nDist > oSuggResult.nDistLimit:
            return
        cCurrent = sRemain[0:1]
        for cChar, jAddr in self._getCharArcs(iAddr):
            if cChar in cp.d1to1.get(cCurrent, cCurrent):
                self._suggest(oSuggResult, sRemain[1:], nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, jAddr, sNewWord+cChar)
            elif not bAvoidLoop:
                if nMaxHardRepl:
                    self._suggest(oSuggResult, sRemain[1:], nMaxSwitch, nMaxDel, nMaxHardRepl-1, nMaxJump, nDist+1, nDeep+1, jAddr, sNewWord+cChar, True)
                if nMaxJump:
                    self._suggest(oSuggResult, sRemain, nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump-1, nDist+1, nDeep+1, jAddr, sNewWord+cChar, True)
        if not bAvoidLoop: # avoid infinite loop
            if len(sRemain) > 1:
                if cCurrent == sRemain[1:2]:
                    # same char, we remove 1 char without adding 1 to <sNewWord>
                    self._suggest(oSuggResult, sRemain[1:], nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord)
                else:
                    # switching chars
                    if nMaxSwitch:
                        self._suggest(oSuggResult, sRemain[1:2]+sRemain[0:1]+sRemain[2:], nMaxSwitch-1, nMaxDel, nMaxHardRepl, nMaxJump, nDist+1, nDeep+1, iAddr, sNewWord, True)
                    # delete char
                    if nMaxDel:
                        self._suggest(oSuggResult, sRemain[1:], nMaxSwitch, nMaxDel-1, nMaxHardRepl, nMaxJump, nDist+1, nDeep+1, iAddr, sNewWord, True)
                # Phonetic replacements
                for sRepl in cp.get1toXReplacement(sNewWord[-1:], cCurrent, sRemain[1:2]):
                    self._suggest(oSuggResult, sRepl + sRemain[1:], nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True)
                for sRepl in cp.d2toX.get(sRemain[0:2], ()):
                    self._suggest(oSuggResult, sRepl + sRemain[2:], nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True)
            # end of word
................................................................................
                for sRepl in cp.dFinal2.get(sRemain, ()):
                    self._suggest(oSuggResult, sRepl, nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True)
            elif len(sRemain) == 1:
                self._suggest(oSuggResult, "", nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True) # remove last char and go on
                for sRepl in cp.dFinal1.get(sRemain, ()):
                    self._suggest(oSuggResult, sRepl, nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True)






    #@timethis
    def suggest2 (self, sWord, nSuggLimit=10):
        "returns a set of suggestions for <sWord>"
        sWord = cp.spellingNormalization(sWord)
        sPfx, sWord, sSfx = cp.cut(sWord)
        oSuggResult = SuggResult(sWord)
        self._suggest2(oSuggResult)







|







 







|




|
|

|
>

|







 







>







 







|
>







 







|










|


|







 







>
>
>
>
>







58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
...
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
...
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
...
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
...
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
...
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
            if nDist <= self.nDistLimit:
                if nDist not in self.dSugg:
                    self.dSugg[nDist] = []
                self.dSugg[nDist].append(sSugg)
                self.aSugg.add(sSugg)
                if nDist < self.nMinDist:
                    self.nMinDist = nDist
                self.nDistLimit = min(self.nDistLimit, self.nMinDist+1)

    def getSuggestions (self, nSuggLimit=10):
        "return a list of suggestions"
        if self.dSugg[0]:
            # we sort the better results with the original word
            self.dSugg[0].sort(key=lambda sSugg: st.distanceDamerauLevenshtein(self.sWord, sSugg))
        lRes = self.dSugg.pop(0)
................................................................................
    def _initBinary (self):
        "initialize with binary structure file"
        if self.by[0:17] != b"/grammalecte-fsa/":
            raise TypeError("# Error. Not a grammalecte-fsa binary dictionary. Header: {}".format(self.by[0:9]))
        if not(self.by[17:18] == b"1" or self.by[17:18] == b"2" or self.by[17:18] == b"3"):
            raise ValueError("# Error. Unknown dictionary version: {}".format(self.by[17:18]))
        try:
            byHeader, byInfo, byValues, by2grams, byDic = self.by.split(b"\0\0\0\0", 4)
        except Exception:
            raise Exception

        self.nCompressionMethod = int(self.by[17:18].decode("utf-8"))
        self.sHeader = byHeader.decode("utf-8")
        self.lArcVal = byValues.decode("utf-8").split("\t")
        self.nArcVal = len(self.lArcVal)
        self.byDic = byDic
        self.a2grams = set(by2grams.decode("utf-8").split("\t"))

        l = byInfo.decode("utf-8").split("//")
        self.sLangCode = l.pop(0)
        self.sLangName = l.pop(0)
        self.sDicName = l.pop(0)
        self.sDate = l.pop(0)
        self.nChar = int(l.pop(0))
        self.nBytesArc = int(l.pop(0))
        self.nBytesNodeAddress = int(l.pop(0))
................................................................................
        self.nBytesOffset = 1 # version 3

    def _initJSON (self, oJSON):
        "initialize with a JSON text file"
        self.__dict__.update(oJSON)
        self.byDic = binascii.unhexlify(self.sByDic)
        self.dCharVal = { v: k  for k, v in self.dChar.items() }
        self.a2grams = set(self.l2grams)

    def getInfo (self):
        "return string about the IBDAWG"
        return  "  Language: {0.sLangName}   Lang code: {0.sLangCode}   Dictionary name: {0.sDicName}" \
                "  Compression method: {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.nEntry:>12,} entries,    {0.nNode:>11,} nodes,   {0.nArc:>11,} arcs\n" \
................................................................................
                "nCompressionMethod": self.nCompressionMethod,
                "nBytesArc": self.nBytesArc,
                "nBytesNodeAddress": self.nBytesNodeAddress,
                "nBytesOffset": self.nBytesOffset,
                # 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
                "sByDic": self.byDic.hex()  if bBinaryDictAsHexString  else [ e  for e in self.byDic ],
                "l2grams": list(self.a2grams)
            }, ensure_ascii=False))
            if bInJSModule:
                hDst.write(";\n\nexports.dictionary = dictionary;\n")

    def isValidToken (self, sToken):
        "checks if <sToken> is valid (if there is hyphens in <sToken>, <sToken> is split, each part is checked)"
        sToken = cp.spellingNormalization(sToken)
................................................................................
        if nDist > oSuggResult.nDistLimit:
            return
        cCurrent = sRemain[0:1]
        for cChar, jAddr in self._getCharArcs(iAddr):
            if cChar in cp.d1to1.get(cCurrent, cCurrent):
                self._suggest(oSuggResult, sRemain[1:], nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, jAddr, sNewWord+cChar)
            elif not bAvoidLoop:
                if nMaxHardRepl and self.isNgramsOK(cChar+sRemain[1:2]):
                    self._suggest(oSuggResult, sRemain[1:], nMaxSwitch, nMaxDel, nMaxHardRepl-1, nMaxJump, nDist+1, nDeep+1, jAddr, sNewWord+cChar, True)
                if nMaxJump:
                    self._suggest(oSuggResult, sRemain, nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump-1, nDist+1, nDeep+1, jAddr, sNewWord+cChar, True)
        if not bAvoidLoop: # avoid infinite loop
            if len(sRemain) > 1:
                if cCurrent == sRemain[1:2]:
                    # same char, we remove 1 char without adding 1 to <sNewWord>
                    self._suggest(oSuggResult, sRemain[1:], nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord)
                else:
                    # switching chars
                    if nMaxSwitch and self.isNgramsOK(sNewWord[-1:]+sRemain[1:2]) and self.isNgramsOK(sRemain[1:2]+sRemain[0:1]):
                        self._suggest(oSuggResult, sRemain[1:2]+sRemain[0:1]+sRemain[2:], nMaxSwitch-1, nMaxDel, nMaxHardRepl, nMaxJump, nDist+1, nDeep+1, iAddr, sNewWord, True)
                    # delete char
                    if nMaxDel and self.isNgramsOK(sNewWord[-1:]+sRemain[1:2]):
                        self._suggest(oSuggResult, sRemain[1:], nMaxSwitch, nMaxDel-1, nMaxHardRepl, nMaxJump, nDist+1, nDeep+1, iAddr, sNewWord, True)
                # Phonetic replacements
                for sRepl in cp.get1toXReplacement(sNewWord[-1:], cCurrent, sRemain[1:2]):
                    self._suggest(oSuggResult, sRepl + sRemain[1:], nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True)
                for sRepl in cp.d2toX.get(sRemain[0:2], ()):
                    self._suggest(oSuggResult, sRepl + sRemain[2:], nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True)
            # end of word
................................................................................
                for sRepl in cp.dFinal2.get(sRemain, ()):
                    self._suggest(oSuggResult, sRepl, nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True)
            elif len(sRemain) == 1:
                self._suggest(oSuggResult, "", nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True) # remove last char and go on
                for sRepl in cp.dFinal1.get(sRemain, ()):
                    self._suggest(oSuggResult, sRepl, nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True)

    def isNgramsOK (self, sChars):
        if len(sChars) != 2:
            return True
        return sChars in self.a2grams

    #@timethis
    def suggest2 (self, sWord, nSuggLimit=10):
        "returns a set of suggestions for <sWord>"
        sWord = cp.spellingNormalization(sWord)
        sPfx, sWord, sSfx = cp.cut(sWord)
        oSuggResult = SuggResult(sWord)
        self._suggest2(oSuggResult)