author | Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com> |
Thu, 27 May 2010 12:45:19 +0300 | |
changeset 37 | fd64c38c277d |
parent 24 | 0ba2181d7c28 |
permissions | -rw-r--r-- |
24
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
1 |
/* |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
2 |
* Copyright (c) 2004-2009 Nokia Corporation and/or its subsidiary(-ies). |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
3 |
* All rights reserved. |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
4 |
* This component and the accompanying materials are made available |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
5 |
* under the terms of "Eclipse Public License v1.0" |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
6 |
* which accompanies this distribution, and is available |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
7 |
* at the URL "http://www.eclipse.org/legal/epl-v10.html". |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
8 |
* |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
9 |
* Initial Contributors: |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
10 |
* Nokia Corporation - initial contribution. |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
11 |
* |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
12 |
* Contributors: |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
13 |
* |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
14 |
* Description: |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
15 |
* |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
16 |
*/ |
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
17 |
|
0
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
18 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
19 |
#ifndef __T_RNDUTILS_H__ |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
20 |
#define __T_RNDUTILS_H__ |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
21 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
22 |
|
24
0ba2181d7c28
Revision: 201007
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
0
diff
changeset
|
23 |
#include "t_utils.h" |
0
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
24 |
#include "t_utils2.h" |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
25 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
26 |
const TInt KDigramNumberOfNonLetters = 3; // space, -, null |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
27 |
const TInt KDigramNumberOfLetters = 26 ; |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
28 |
const TInt KDigramNumberOfChars = KDigramNumberOfLetters*2+ KDigramNumberOfNonLetters; |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
29 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
30 |
/** Used to create randomly generated words using n-grams found in the specified data file. |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
31 |
Using n-grams lead to much more realistic words and thus makes it obvious a descriptor |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
32 |
is data and not noise. |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
33 |
This implementation only supports digrams */ |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
34 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
35 |
class CWordNgrams : public CBase, public MRandomWordGenerator |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
36 |
{ |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
37 |
public: |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
38 |
HBufC* WordLC(); |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
39 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
40 |
protected: |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
41 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
42 |
TBool AddLetter(TDes& aWord); |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
43 |
void ConstructL(const TDesC& aFilename, TInt aNgramSize = 2, TBool aUseNarrowChar = ETrue); |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
44 |
virtual TChar NextLetter(const TDesC& aWord) = 0; |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
45 |
virtual void SetNgram(const TDesC8& aNgram, TUint aNumber) = 0; |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
46 |
virtual void SetNgram(const TDesC16& aNgram, TUint aNumber) = 0; |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
47 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
48 |
}; |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
49 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
50 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
51 |
/** Implementation of CWordNgrams which uses digram frequencies (letter pairs) |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
52 |
to generate random words */ |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
53 |
class CWordDigrams : public CWordNgrams |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
54 |
{ |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
55 |
public: |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
56 |
static CWordDigrams* NewLC(const TDesC& aFilename, TInt64& aRandSeed ); |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
57 |
virtual ~CWordDigrams(); |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
58 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
59 |
protected: |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
60 |
void ConstructL(const TDesC& aFilename); |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
61 |
void SetNgram(const TDesC8& aNgram, TUint aNumber); |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
62 |
void SetNgram(const TDesC16& aNgram, TUint aNumber); // empty function |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
63 |
TChar NextLetter(const TDesC& aWord); |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
64 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
65 |
private: |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
66 |
CWordDigrams(TInt64& aRandSeed); |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
67 |
static TInt Index(const TChar c); |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
68 |
static TChar Char(TInt i); |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
69 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
70 |
private: |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
71 |
TInt64& iRandSeed; |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
72 |
TUint8 iDigrams[KDigramNumberOfChars][KDigramNumberOfChars]; |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
73 |
TUint16 iSizes[KDigramNumberOfChars]; |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
74 |
}; |
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
75 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
76 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
77 |
|
e686773b3f54
Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
78 |
#endif |