|
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 TUint16 keyedTables16OfIndexedTables16_indexedEntries_codePage1252ToUnicode_1[]=
|
|
sl@0
|
25 |
{
|
|
sl@0
|
26 |
0x201a,
|
|
sl@0
|
27 |
0x0192,
|
|
sl@0
|
28 |
0x201e,
|
|
sl@0
|
29 |
0x2026,
|
|
sl@0
|
30 |
0x2020,
|
|
sl@0
|
31 |
0x2021,
|
|
sl@0
|
32 |
0x02c6,
|
|
sl@0
|
33 |
0x2030,
|
|
sl@0
|
34 |
0x0160,
|
|
sl@0
|
35 |
0x2039,
|
|
sl@0
|
36 |
0x0152
|
|
sl@0
|
37 |
};
|
|
sl@0
|
38 |
|
|
sl@0
|
39 |
LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_codePage1252ToUnicode_2[]=
|
|
sl@0
|
40 |
{
|
|
sl@0
|
41 |
0x2018,
|
|
sl@0
|
42 |
0x2019,
|
|
sl@0
|
43 |
0x201c,
|
|
sl@0
|
44 |
0x201d,
|
|
sl@0
|
45 |
0x2022,
|
|
sl@0
|
46 |
0x2013,
|
|
sl@0
|
47 |
0x2014,
|
|
sl@0
|
48 |
0x02dc,
|
|
sl@0
|
49 |
0x2122,
|
|
sl@0
|
50 |
0x0161,
|
|
sl@0
|
51 |
0x203a,
|
|
sl@0
|
52 |
0x0153
|
|
sl@0
|
53 |
};
|
|
sl@0
|
54 |
|
|
sl@0
|
55 |
LOCAL_D const TUint16 keyedTables16OfIndexedTables16_indexedEntries_codePage1252ToUnicode_3[]=
|
|
sl@0
|
56 |
{
|
|
sl@0
|
57 |
0x017e,
|
|
sl@0
|
58 |
0x0178
|
|
sl@0
|
59 |
};
|
|
sl@0
|
60 |
|
|
sl@0
|
61 |
LOCAL_D const SCnvConversionData::SOneDirectionData::SRange::UData::SKeyedTable16OfIndexedTables16::SKeyedEntry keyedTables16OfIndexedTables16_keyedEntries_codePage1252ToUnicode_1[]=
|
|
sl@0
|
62 |
{
|
|
sl@0
|
63 |
{
|
|
sl@0
|
64 |
0x82,
|
|
sl@0
|
65 |
0x8c,
|
|
sl@0
|
66 |
keyedTables16OfIndexedTables16_indexedEntries_codePage1252ToUnicode_1
|
|
sl@0
|
67 |
},
|
|
sl@0
|
68 |
{
|
|
sl@0
|
69 |
0x91,
|
|
sl@0
|
70 |
0x9c,
|
|
sl@0
|
71 |
keyedTables16OfIndexedTables16_indexedEntries_codePage1252ToUnicode_2
|
|
sl@0
|
72 |
},
|
|
sl@0
|
73 |
{
|
|
sl@0
|
74 |
0x9e,
|
|
sl@0
|
75 |
0x9f,
|
|
sl@0
|
76 |
keyedTables16OfIndexedTables16_indexedEntries_codePage1252ToUnicode_3
|
|
sl@0
|
77 |
}
|
|
sl@0
|
78 |
};
|
|
sl@0
|
79 |
|
|
sl@0
|
80 |
LOCAL_D const SCnvConversionData::SOneDirectionData::SRange::UData::SKeyedTable1616::SEntry keyedTable1616_unicodeToCodePage1252_1[]=
|
|
sl@0
|
81 |
{
|
|
sl@0
|
82 |
{
|
|
sl@0
|
83 |
0x0152,
|
|
sl@0
|
84 |
0x8c
|
|
sl@0
|
85 |
},
|
|
sl@0
|
86 |
{
|
|
sl@0
|
87 |
0x0153,
|
|
sl@0
|
88 |
0x9c
|
|
sl@0
|
89 |
},
|
|
sl@0
|
90 |
{
|
|
sl@0
|
91 |
0x0160,
|
|
sl@0
|
92 |
0x8a
|
|
sl@0
|
93 |
},
|
|
sl@0
|
94 |
{
|
|
sl@0
|
95 |
0x0161,
|
|
sl@0
|
96 |
0x9a
|
|
sl@0
|
97 |
},
|
|
sl@0
|
98 |
{
|
|
sl@0
|
99 |
0x0178,
|
|
sl@0
|
100 |
0x9f
|
|
sl@0
|
101 |
},
|
|
sl@0
|
102 |
{
|
|
sl@0
|
103 |
0x017d,
|
|
sl@0
|
104 |
0x8e
|
|
sl@0
|
105 |
},
|
|
sl@0
|
106 |
{
|
|
sl@0
|
107 |
0x017e,
|
|
sl@0
|
108 |
0x9e
|
|
sl@0
|
109 |
},
|
|
sl@0
|
110 |
{
|
|
sl@0
|
111 |
0x0192,
|
|
sl@0
|
112 |
0x83
|
|
sl@0
|
113 |
},
|
|
sl@0
|
114 |
{
|
|
sl@0
|
115 |
0x02c6,
|
|
sl@0
|
116 |
0x88
|
|
sl@0
|
117 |
},
|
|
sl@0
|
118 |
{
|
|
sl@0
|
119 |
0x02dc,
|
|
sl@0
|
120 |
0x98
|
|
sl@0
|
121 |
},
|
|
sl@0
|
122 |
{
|
|
sl@0
|
123 |
0x2013,
|
|
sl@0
|
124 |
0x96
|
|
sl@0
|
125 |
},
|
|
sl@0
|
126 |
{
|
|
sl@0
|
127 |
0x2014,
|
|
sl@0
|
128 |
0x97
|
|
sl@0
|
129 |
},
|
|
sl@0
|
130 |
{
|
|
sl@0
|
131 |
0x2018,
|
|
sl@0
|
132 |
0x91
|
|
sl@0
|
133 |
},
|
|
sl@0
|
134 |
{
|
|
sl@0
|
135 |
0x2019,
|
|
sl@0
|
136 |
0x92
|
|
sl@0
|
137 |
},
|
|
sl@0
|
138 |
{
|
|
sl@0
|
139 |
0x201a,
|
|
sl@0
|
140 |
0x82
|
|
sl@0
|
141 |
},
|
|
sl@0
|
142 |
{
|
|
sl@0
|
143 |
0x201c,
|
|
sl@0
|
144 |
0x93
|
|
sl@0
|
145 |
},
|
|
sl@0
|
146 |
{
|
|
sl@0
|
147 |
0x201d,
|
|
sl@0
|
148 |
0x94
|
|
sl@0
|
149 |
},
|
|
sl@0
|
150 |
{
|
|
sl@0
|
151 |
0x201e,
|
|
sl@0
|
152 |
0x84
|
|
sl@0
|
153 |
},
|
|
sl@0
|
154 |
{
|
|
sl@0
|
155 |
0x2020,
|
|
sl@0
|
156 |
0x86
|
|
sl@0
|
157 |
},
|
|
sl@0
|
158 |
{
|
|
sl@0
|
159 |
0x2021,
|
|
sl@0
|
160 |
0x87
|
|
sl@0
|
161 |
},
|
|
sl@0
|
162 |
{
|
|
sl@0
|
163 |
0x2022,
|
|
sl@0
|
164 |
0x95
|
|
sl@0
|
165 |
},
|
|
sl@0
|
166 |
{
|
|
sl@0
|
167 |
0x2026,
|
|
sl@0
|
168 |
0x85
|
|
sl@0
|
169 |
},
|
|
sl@0
|
170 |
{
|
|
sl@0
|
171 |
0x2030,
|
|
sl@0
|
172 |
0x89
|
|
sl@0
|
173 |
},
|
|
sl@0
|
174 |
{
|
|
sl@0
|
175 |
0x2039,
|
|
sl@0
|
176 |
0x8b
|
|
sl@0
|
177 |
},
|
|
sl@0
|
178 |
{
|
|
sl@0
|
179 |
0x203a,
|
|
sl@0
|
180 |
0x9b
|
|
sl@0
|
181 |
},
|
|
sl@0
|
182 |
{
|
|
sl@0
|
183 |
0x20ac,
|
|
sl@0
|
184 |
0x80
|
|
sl@0
|
185 |
},
|
|
sl@0
|
186 |
{
|
|
sl@0
|
187 |
0x2122,
|
|
sl@0
|
188 |
0x99
|
|
sl@0
|
189 |
}
|
|
sl@0
|
190 |
};
|
|
sl@0
|
191 |
|
|
sl@0
|
192 |
LOCAL_D const SCnvConversionData::SVariableByteData::SRange codePage1252VariableByteDataRanges[]=
|
|
sl@0
|
193 |
{
|
|
sl@0
|
194 |
{
|
|
sl@0
|
195 |
0x00,
|
|
sl@0
|
196 |
0xff,
|
|
sl@0
|
197 |
0,
|
|
sl@0
|
198 |
0
|
|
sl@0
|
199 |
}
|
|
sl@0
|
200 |
};
|
|
sl@0
|
201 |
|
|
sl@0
|
202 |
LOCAL_D const SCnvConversionData::SOneDirectionData::SRange codePage1252ToUnicodeDataRanges[]=
|
|
sl@0
|
203 |
{
|
|
sl@0
|
204 |
{
|
|
sl@0
|
205 |
0x00,
|
|
sl@0
|
206 |
0x7f,
|
|
sl@0
|
207 |
SCnvConversionData::SOneDirectionData::SRange::EDirect,
|
|
sl@0
|
208 |
0,
|
|
sl@0
|
209 |
0,
|
|
sl@0
|
210 |
{
|
|
sl@0
|
211 |
0,
|
|
sl@0
|
212 |
0
|
|
sl@0
|
213 |
}
|
|
sl@0
|
214 |
},
|
|
sl@0
|
215 |
{
|
|
sl@0
|
216 |
0xa0,
|
|
sl@0
|
217 |
0xff,
|
|
sl@0
|
218 |
SCnvConversionData::SOneDirectionData::SRange::EDirect,
|
|
sl@0
|
219 |
0,
|
|
sl@0
|
220 |
0,
|
|
sl@0
|
221 |
{
|
|
sl@0
|
222 |
0,
|
|
sl@0
|
223 |
0
|
|
sl@0
|
224 |
}
|
|
sl@0
|
225 |
},
|
|
sl@0
|
226 |
{
|
|
sl@0
|
227 |
0x80,
|
|
sl@0
|
228 |
0x80,
|
|
sl@0
|
229 |
SCnvConversionData::SOneDirectionData::SRange::EOffset,
|
|
sl@0
|
230 |
0,
|
|
sl@0
|
231 |
0,
|
|
sl@0
|
232 |
{
|
|
sl@0
|
233 |
STATIC_CAST(TUint, 8236),
|
|
sl@0
|
234 |
0
|
|
sl@0
|
235 |
}
|
|
sl@0
|
236 |
},
|
|
sl@0
|
237 |
{
|
|
sl@0
|
238 |
0x8e,
|
|
sl@0
|
239 |
0x8e,
|
|
sl@0
|
240 |
SCnvConversionData::SOneDirectionData::SRange::EOffset,
|
|
sl@0
|
241 |
0,
|
|
sl@0
|
242 |
0,
|
|
sl@0
|
243 |
{
|
|
sl@0
|
244 |
STATIC_CAST(TUint, 239),
|
|
sl@0
|
245 |
0
|
|
sl@0
|
246 |
}
|
|
sl@0
|
247 |
},
|
|
sl@0
|
248 |
{
|
|
sl@0
|
249 |
0x82,
|
|
sl@0
|
250 |
0x9f,
|
|
sl@0
|
251 |
SCnvConversionData::SOneDirectionData::SRange::EKeyedTable16OfIndexedTables16,
|
|
sl@0
|
252 |
0,
|
|
sl@0
|
253 |
0,
|
|
sl@0
|
254 |
{
|
|
sl@0
|
255 |
UData_SKeyedTable16OfIndexedTables16(keyedTables16OfIndexedTables16_keyedEntries_codePage1252ToUnicode_1)
|
|
sl@0
|
256 |
}
|
|
sl@0
|
257 |
}
|
|
sl@0
|
258 |
};
|
|
sl@0
|
259 |
|
|
sl@0
|
260 |
LOCAL_D const SCnvConversionData::SOneDirectionData::SRange unicodeToCodePage1252DataRanges[]=
|
|
sl@0
|
261 |
{
|
|
sl@0
|
262 |
{
|
|
sl@0
|
263 |
0x0000,
|
|
sl@0
|
264 |
0x007f,
|
|
sl@0
|
265 |
SCnvConversionData::SOneDirectionData::SRange::EDirect,
|
|
sl@0
|
266 |
1,
|
|
sl@0
|
267 |
0,
|
|
sl@0
|
268 |
{
|
|
sl@0
|
269 |
0,
|
|
sl@0
|
270 |
0
|
|
sl@0
|
271 |
}
|
|
sl@0
|
272 |
},
|
|
sl@0
|
273 |
{
|
|
sl@0
|
274 |
0x00a0,
|
|
sl@0
|
275 |
0x00ff,
|
|
sl@0
|
276 |
SCnvConversionData::SOneDirectionData::SRange::EDirect,
|
|
sl@0
|
277 |
1,
|
|
sl@0
|
278 |
0,
|
|
sl@0
|
279 |
{
|
|
sl@0
|
280 |
0,
|
|
sl@0
|
281 |
0
|
|
sl@0
|
282 |
}
|
|
sl@0
|
283 |
},
|
|
sl@0
|
284 |
{
|
|
sl@0
|
285 |
0x0152,
|
|
sl@0
|
286 |
0x2122,
|
|
sl@0
|
287 |
SCnvConversionData::SOneDirectionData::SRange::EKeyedTable1616,
|
|
sl@0
|
288 |
1,
|
|
sl@0
|
289 |
0,
|
|
sl@0
|
290 |
{
|
|
sl@0
|
291 |
UData_SKeyedTable1616(keyedTable1616_unicodeToCodePage1252_1)
|
|
sl@0
|
292 |
}
|
|
sl@0
|
293 |
}
|
|
sl@0
|
294 |
};
|
|
sl@0
|
295 |
|
|
sl@0
|
296 |
GLREF_D const SCnvConversionData codePage1252ConversionData=
|
|
sl@0
|
297 |
{
|
|
sl@0
|
298 |
SCnvConversionData::EUnspecified,
|
|
sl@0
|
299 |
{
|
|
sl@0
|
300 |
ARRAY_LENGTH(codePage1252VariableByteDataRanges),
|
|
sl@0
|
301 |
codePage1252VariableByteDataRanges
|
|
sl@0
|
302 |
},
|
|
sl@0
|
303 |
{
|
|
sl@0
|
304 |
ARRAY_LENGTH(codePage1252ToUnicodeDataRanges),
|
|
sl@0
|
305 |
codePage1252ToUnicodeDataRanges
|
|
sl@0
|
306 |
},
|
|
sl@0
|
307 |
{
|
|
sl@0
|
308 |
ARRAY_LENGTH(unicodeToCodePage1252DataRanges),
|
|
sl@0
|
309 |
unicodeToCodePage1252DataRanges
|
|
sl@0
|
310 |
},
|
|
sl@0
|
311 |
NULL,
|
|
sl@0
|
312 |
NULL
|
|
sl@0
|
313 |
};
|
|
sl@0
|
314 |
|
|
sl@0
|
315 |
GLREF_C void IsCharacterSetCP1252(TInt& aConfidenceLevel, const TDesC8& aSample)
|
|
sl@0
|
316 |
{
|
|
sl@0
|
317 |
aConfidenceLevel = 60;
|
|
sl@0
|
318 |
TInt sampleLength = aSample.Length();
|
|
sl@0
|
319 |
|
|
sl@0
|
320 |
for (TInt i=0; i<sampleLength; ++i)
|
|
sl@0
|
321 |
{
|
|
sl@0
|
322 |
// CP1252 includes ASCII as well
|
|
sl@0
|
323 |
// first check if the char is in the range 0x80 - 0x9f (controls codes in ISO88591)
|
|
sl@0
|
324 |
// If it is in that range then the likelihood that it's CP1252 is a bit higher
|
|
sl@0
|
325 |
if ((aSample[i] >= 0x80) && (aSample[i] <= 0x9f))
|
|
sl@0
|
326 |
{
|
|
sl@0
|
327 |
if((aSample[i]==0x81)||(aSample[i]==0x8D)||(aSample[i]==0x8f)||
|
|
sl@0
|
328 |
(aSample[i]==0x90)||(aSample[i]==0x9d))
|
|
sl@0
|
329 |
{
|
|
sl@0
|
330 |
// These code values are not supported by the Codepage CP1252
|
|
sl@0
|
331 |
aConfidenceLevel = 0;
|
|
sl@0
|
332 |
break;
|
|
sl@0
|
333 |
}
|
|
sl@0
|
334 |
else
|
|
sl@0
|
335 |
{
|
|
sl@0
|
336 |
// problem: UTF8 uses the values 0x80-0x9f in more than 50% of it's multibyte representation
|
|
sl@0
|
337 |
// so if the text was UTF8 .... the confidence here would hit the roof. Could check to make
|
|
sl@0
|
338 |
// sure that this is not UTF8
|
|
sl@0
|
339 |
aConfidenceLevel+=1;
|
|
sl@0
|
340 |
}
|
|
sl@0
|
341 |
}
|
|
sl@0
|
342 |
TInt increment1 = i+1;
|
|
sl@0
|
343 |
TInt decrement1 = i-1;
|
|
sl@0
|
344 |
// 0xf7 is the division symbol in CP1252.
|
|
sl@0
|
345 |
// 0xd7 is the division symbol in CP1252.If char on either side of the division
|
|
sl@0
|
346 |
// symbol is a number then the confidence that it's ISO88591 increases
|
|
sl@0
|
347 |
if( decrement1>= 0 && ((aSample[i]==0xf7) || (aSample[i]==0xd7)) && increment1<sampleLength)
|
|
sl@0
|
348 |
{
|
|
sl@0
|
349 |
|
|
sl@0
|
350 |
if (increment1 >= sampleLength)
|
|
sl@0
|
351 |
break;
|
|
sl@0
|
352 |
if ( (aSample[decrement1] >= 0x30) && (aSample[decrement1] <= 0x39) && // char before is a number
|
|
sl@0
|
353 |
(aSample[increment1] >= 0x30) && (aSample[increment1] <= 0x39) ) // char after is a number
|
|
sl@0
|
354 |
{
|
|
sl@0
|
355 |
aConfidenceLevel+=5;
|
|
sl@0
|
356 |
}
|
|
sl@0
|
357 |
}
|
|
sl@0
|
358 |
// Can also use the currency symbol to increase confidence if the char after a
|
|
sl@0
|
359 |
// currency symbol is numeric
|
|
sl@0
|
360 |
if((aSample[i]>=0xa2) && (aSample[i] <= 0xa5) && increment1<sampleLength)
|
|
sl@0
|
361 |
{
|
|
sl@0
|
362 |
if ((aSample[increment1] >= 0x30) && (aSample[increment1] <= 0x39))
|
|
sl@0
|
363 |
{
|
|
sl@0
|
364 |
aConfidenceLevel+=5;
|
|
sl@0
|
365 |
}
|
|
sl@0
|
366 |
}
|
|
sl@0
|
367 |
} // for loop
|
|
sl@0
|
368 |
aConfidenceLevel =(aConfidenceLevel >0)? ((aConfidenceLevel > 100)? 100: aConfidenceLevel): 0;
|
|
sl@0
|
369 |
}
|