charconvfw/Charconv/ongoing/Source/foreign/builtin/UnicodeLittle.cpp
author Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
Fri, 16 Apr 2010 16:55:07 +0300
changeset 16 56cd22a7a1cb
parent 0 1fb32624e06b
permissions -rw-r--r--
Revision: 201011 Kit: 201015
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
0
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     1
/*
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     2
* Copyright (c) 1997-2000 Nokia Corporation and/or its subsidiary(-ies). 
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     3
* All rights reserved.
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     4
* This component and the accompanying materials are made available
16
56cd22a7a1cb Revision: 201011
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents: 0
diff changeset
     5
* under the terms of the License "Eclipse Public License v1.0"
0
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     6
* which accompanies this distribution, and is available
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     7
* at the URL "http://www.eclipse.org/legal/epl-v10.html".
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     8
*
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     9
* Initial Contributors:
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    10
* Nokia Corporation - initial contribution.
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    11
*
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    12
* Contributors:
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    13
*
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    14
* Description:      
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    15
*
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    16
*/
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    17
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    18
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    19
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    20
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    21
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    22
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    23
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    24
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    25
#include <e32std.h>
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    26
#include <convdata.h>
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    27
#include "Unicode.h"
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    28
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    29
#define ARRAY_LENGTH(aArray) (sizeof(aArray)/sizeof((aArray)[0]))
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    30
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    31
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    32
GLREF_D const SCnvConversionData unicodeConversionDataLittle=
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    33
	{
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    34
	SCnvConversionData::EFixedLittleEndian,
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    35
		{
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    36
		ARRAY_LENGTH(unicodeVariableByteDataRanges),
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    37
		unicodeVariableByteDataRanges
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    38
		},
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    39
		{
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    40
		ARRAY_LENGTH(unicodeTounicodeDataRanges),
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    41
		unicodeTounicodeDataRanges
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    42
		},
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    43
		{
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    44
		ARRAY_LENGTH(unicodeTounicodeDataRanges),
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    45
		unicodeTounicodeDataRanges
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    46
		},
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    47
	NULL,
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    48
	NULL
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    49
	};
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    50
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    51
GLREF_C void IsCharacterSetUnicodeLittle(TInt& aConfidenceLevel, const TDesC8& aSample)
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    52
	{
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    53
	
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    54
	TInt sampleLength = aSample.Length();
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    55
	aConfidenceLevel =70;
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    56
	if (sampleLength < 2)
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    57
		return;
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    58
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    59
	if (aSample[0]==0xff)
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    60
		{
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    61
		// The first byte is a possible ByteOrderMark
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    62
		// Try matching the next character 
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    63
		if(aSample[1]==0xfe)
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    64
			{
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    65
			// the byte order mark could be 0xFEFF or 0xFFFE depending on 
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    66
			// endianness of the sample text.
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    67
			aConfidenceLevel=100;
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    68
			}
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    69
		}
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    70
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    71
	for (TInt i = 0; i < sampleLength-1; ++i)
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    72
		{
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    73
		if (aSample[i] == 0x0d)
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    74
 			{
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    75
 			if (aSample[i+1] == 0x0a)
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    76
 				{
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    77
 				// reduce the confidence level
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    78
 				aConfidenceLevel -= 25;
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    79
 				}
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    80
 			}
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    81
		}
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    82
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    83
	// if not 100% confident already, check if most odd bytes zero 
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    84
	#define MAX_SAMPLE_LENGTH 2048
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    85
	if ( aConfidenceLevel < 100 )
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    86
		{	
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    87
		TInt repeat=0;
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    88
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    89
		// only check the first MAX_SAMPLE_LENGTH if big sample
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    90
		TInt length =( sampleLength > MAX_SAMPLE_LENGTH ? MAX_SAMPLE_LENGTH : sampleLength);
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    91
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    92
		// start from 1 and check the odd bytes
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    93
		for (TInt i = 1; i < length-1; i+=2)
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    94
			{
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    95
			if (aSample[i] == 0x0) 
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    96
				repeat ++;
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    97
			}
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    98
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    99
		// if more than 80% odd bytes zero, then this IS little Endian
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
   100
		if ( (repeat * 100) /  (length * 5) >= 8)
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
   101
			aConfidenceLevel  = 100;
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
   102
		}
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
   103
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
   104
	aConfidenceLevel =(aConfidenceLevel >0)? ((aConfidenceLevel > 100)? 100: aConfidenceLevel): 0;
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
   105
	}
1fb32624e06b Revision: 201003
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
   106