|
sl@0
|
1 |
/*
|
|
sl@0
|
2 |
* Copyright (c) 1997-2009 Nokia Corporation and/or its subsidiary(-ies).
|
|
sl@0
|
3 |
* All rights reserved.
|
|
sl@0
|
4 |
* This component and the accompanying materials are made available
|
|
sl@0
|
5 |
* under the terms of "Eclipse Public License v1.0"
|
|
sl@0
|
6 |
* which accompanies this distribution, and is available
|
|
sl@0
|
7 |
* at the URL "http://www.eclipse.org/legal/epl-v10.html".
|
|
sl@0
|
8 |
*
|
|
sl@0
|
9 |
* Initial Contributors:
|
|
sl@0
|
10 |
* Nokia Corporation - initial contribution.
|
|
sl@0
|
11 |
*
|
|
sl@0
|
12 |
* Contributors:
|
|
sl@0
|
13 |
*
|
|
sl@0
|
14 |
* Description:
|
|
sl@0
|
15 |
*
|
|
sl@0
|
16 |
*/
|
|
sl@0
|
17 |
|
|
sl@0
|
18 |
|
|
sl@0
|
19 |
#include <e32std.h>
|
|
sl@0
|
20 |
#include <convdata.h>
|
|
sl@0
|
21 |
|
|
sl@0
|
22 |
#define ARRAY_LENGTH(aArray) (sizeof(aArray)/sizeof((aArray)[0]))
|
|
sl@0
|
23 |
|
|
sl@0
|
24 |
LOCAL_D const SCnvConversionData::SVariableByteData::SRange iso88591VariableByteDataRanges[]=
|
|
sl@0
|
25 |
{
|
|
sl@0
|
26 |
{
|
|
sl@0
|
27 |
0x00,
|
|
sl@0
|
28 |
0xff,
|
|
sl@0
|
29 |
0,
|
|
sl@0
|
30 |
0
|
|
sl@0
|
31 |
}
|
|
sl@0
|
32 |
};
|
|
sl@0
|
33 |
|
|
sl@0
|
34 |
LOCAL_D const SCnvConversionData::SOneDirectionData::SRange iso88591ToUnicodeDataRanges[]=
|
|
sl@0
|
35 |
{
|
|
sl@0
|
36 |
{
|
|
sl@0
|
37 |
0x00,
|
|
sl@0
|
38 |
0x7f,
|
|
sl@0
|
39 |
SCnvConversionData::SOneDirectionData::SRange::EDirect,
|
|
sl@0
|
40 |
0,
|
|
sl@0
|
41 |
0,
|
|
sl@0
|
42 |
{
|
|
sl@0
|
43 |
0,
|
|
sl@0
|
44 |
0
|
|
sl@0
|
45 |
}
|
|
sl@0
|
46 |
},
|
|
sl@0
|
47 |
{
|
|
sl@0
|
48 |
0x81,
|
|
sl@0
|
49 |
0xff,
|
|
sl@0
|
50 |
SCnvConversionData::SOneDirectionData::SRange::EDirect,
|
|
sl@0
|
51 |
0,
|
|
sl@0
|
52 |
0,
|
|
sl@0
|
53 |
{
|
|
sl@0
|
54 |
0,
|
|
sl@0
|
55 |
0
|
|
sl@0
|
56 |
}
|
|
sl@0
|
57 |
},
|
|
sl@0
|
58 |
{
|
|
sl@0
|
59 |
0x80,
|
|
sl@0
|
60 |
0x80,
|
|
sl@0
|
61 |
SCnvConversionData::SOneDirectionData::SRange::EOffset,
|
|
sl@0
|
62 |
0,
|
|
sl@0
|
63 |
0,
|
|
sl@0
|
64 |
{
|
|
sl@0
|
65 |
STATIC_CAST(TUint, 8236)
|
|
sl@0
|
66 |
}
|
|
sl@0
|
67 |
}
|
|
sl@0
|
68 |
};
|
|
sl@0
|
69 |
|
|
sl@0
|
70 |
LOCAL_D const SCnvConversionData::SOneDirectionData::SRange unicodeToIso88591DataRanges[]=
|
|
sl@0
|
71 |
{
|
|
sl@0
|
72 |
{
|
|
sl@0
|
73 |
0x0000,
|
|
sl@0
|
74 |
0x007f,
|
|
sl@0
|
75 |
SCnvConversionData::SOneDirectionData::SRange::EDirect,
|
|
sl@0
|
76 |
1,
|
|
sl@0
|
77 |
0,
|
|
sl@0
|
78 |
{
|
|
sl@0
|
79 |
0,
|
|
sl@0
|
80 |
0
|
|
sl@0
|
81 |
}
|
|
sl@0
|
82 |
},
|
|
sl@0
|
83 |
{
|
|
sl@0
|
84 |
0x0081,
|
|
sl@0
|
85 |
0x00ff,
|
|
sl@0
|
86 |
SCnvConversionData::SOneDirectionData::SRange::EDirect,
|
|
sl@0
|
87 |
1,
|
|
sl@0
|
88 |
0,
|
|
sl@0
|
89 |
{
|
|
sl@0
|
90 |
0,
|
|
sl@0
|
91 |
0
|
|
sl@0
|
92 |
}
|
|
sl@0
|
93 |
},
|
|
sl@0
|
94 |
{
|
|
sl@0
|
95 |
0x20ac,
|
|
sl@0
|
96 |
0x20ac,
|
|
sl@0
|
97 |
SCnvConversionData::SOneDirectionData::SRange::EOffset,
|
|
sl@0
|
98 |
1,
|
|
sl@0
|
99 |
0,
|
|
sl@0
|
100 |
{
|
|
sl@0
|
101 |
STATIC_CAST(TUint, -8236)
|
|
sl@0
|
102 |
}
|
|
sl@0
|
103 |
}
|
|
sl@0
|
104 |
};
|
|
sl@0
|
105 |
|
|
sl@0
|
106 |
GLREF_D const SCnvConversionData iso88591ConversionData=
|
|
sl@0
|
107 |
{
|
|
sl@0
|
108 |
SCnvConversionData::EUnspecified,
|
|
sl@0
|
109 |
{
|
|
sl@0
|
110 |
ARRAY_LENGTH(iso88591VariableByteDataRanges),
|
|
sl@0
|
111 |
iso88591VariableByteDataRanges
|
|
sl@0
|
112 |
},
|
|
sl@0
|
113 |
{
|
|
sl@0
|
114 |
ARRAY_LENGTH(iso88591ToUnicodeDataRanges),
|
|
sl@0
|
115 |
iso88591ToUnicodeDataRanges
|
|
sl@0
|
116 |
},
|
|
sl@0
|
117 |
{
|
|
sl@0
|
118 |
ARRAY_LENGTH(unicodeToIso88591DataRanges),
|
|
sl@0
|
119 |
unicodeToIso88591DataRanges
|
|
sl@0
|
120 |
},
|
|
sl@0
|
121 |
NULL,
|
|
sl@0
|
122 |
NULL
|
|
sl@0
|
123 |
};
|
|
sl@0
|
124 |
|
|
sl@0
|
125 |
GLREF_C void IsCharacterSetISO88591(TInt& aConfidenceLevel, const TDesC8& aSample)
|
|
sl@0
|
126 |
{
|
|
sl@0
|
127 |
TInt sampleLength = aSample.Length();
|
|
sl@0
|
128 |
aConfidenceLevel = 75;
|
|
sl@0
|
129 |
|
|
sl@0
|
130 |
for (TInt i=0; i<sampleLength; ++i)
|
|
sl@0
|
131 |
{
|
|
sl@0
|
132 |
// ISO88591 includes ASCII as well
|
|
sl@0
|
133 |
// first check if the char is in the range 0x80 - 0x9f (controls codes)
|
|
sl@0
|
134 |
// If it is in that range then it's not ISO88591
|
|
sl@0
|
135 |
if ((aSample[i] >= 0x80) && (aSample[i] <= 0x9f))
|
|
sl@0
|
136 |
{
|
|
sl@0
|
137 |
aConfidenceLevel=0;
|
|
sl@0
|
138 |
break;
|
|
sl@0
|
139 |
}
|
|
sl@0
|
140 |
// 0xf7 is the division symbol in ISO88591.
|
|
sl@0
|
141 |
// 0xd7 is the division symbol in ISO88591.If char on either side of the division
|
|
sl@0
|
142 |
// symbol is a number then the confidence that it's ISO88591 increases
|
|
sl@0
|
143 |
if( i>0 && ((aSample[i]==0xf7) || (aSample[i]==0xd7)) && ((i+1)<sampleLength) )
|
|
sl@0
|
144 |
{
|
|
sl@0
|
145 |
if ( (aSample[i-1] >= 0x30) && (aSample[i-1] <= 0x39) && // char before is a number
|
|
sl@0
|
146 |
(aSample[i+1] >= 0x30) && (aSample[i+1] <= 0x39) ) // char after is a number
|
|
sl@0
|
147 |
{
|
|
sl@0
|
148 |
aConfidenceLevel+=5;
|
|
sl@0
|
149 |
}
|
|
sl@0
|
150 |
}
|
|
sl@0
|
151 |
// Can also use the currency symbol to increase confidence if the char after a
|
|
sl@0
|
152 |
// currency symbol is numeric
|
|
sl@0
|
153 |
if((aSample[i]>=0xa2) && (aSample[i] <= 0xa5) && ((i+1)<sampleLength))
|
|
sl@0
|
154 |
{
|
|
sl@0
|
155 |
if ((aSample[i+1] >= 0x30) && (aSample[i+1] <= 0x39))
|
|
sl@0
|
156 |
{
|
|
sl@0
|
157 |
aConfidenceLevel+=5;
|
|
sl@0
|
158 |
}
|
|
sl@0
|
159 |
}
|
|
sl@0
|
160 |
} // for loop
|
|
sl@0
|
161 |
aConfidenceLevel =(aConfidenceLevel >0)? ((aConfidenceLevel > 100)? 100: aConfidenceLevel): 0;
|
|
sl@0
|
162 |
}
|