charguess 1.0
Sign up to get free protection for your applications and to get access to all the features.
- data/History.txt +4 -0
- data/Manifest.txt +134 -0
- data/PostInstall.txt +7 -0
- data/README.rdoc +26 -0
- data/Rakefile +25 -0
- data/ext/charguess/charguess.c +29 -0
- data/ext/charguess/extconf.rb +11 -0
- data/ext/libcharguess/cpp/.deps/EUCJPProber.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/EUCJPProber.Po +87 -0
- data/ext/libcharguess/cpp/.deps/EUCKRProber.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/EUCKRProber.Po +85 -0
- data/ext/libcharguess/cpp/.deps/EUCTWProber.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/EUCTWProber.Po +85 -0
- data/ext/libcharguess/cpp/.deps/EscCharsetProber.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/EscCharsetProber.Po +83 -0
- data/ext/libcharguess/cpp/.deps/EscSM.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/EscSM.Po +77 -0
- data/ext/libcharguess/cpp/.deps/GB2312Prober.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/GB2312Prober.Po +85 -0
- data/ext/libcharguess/cpp/.deps/LangBulgarianModel.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/LangBulgarianModel.Po +78 -0
- data/ext/libcharguess/cpp/.deps/LangCyrillicModel.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/LangCyrillicModel.Po +78 -0
- data/ext/libcharguess/cpp/.deps/LangGreekModel.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/LangGreekModel.Po +78 -0
- data/ext/libcharguess/cpp/.deps/LangHungarianModel.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/LangHungarianModel.Po +78 -0
- data/ext/libcharguess/cpp/.deps/LangThaiModel.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/LangThaiModel.Po +78 -0
- data/ext/libcharguess/cpp/.deps/Latin1Prober.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/Latin1Prober.Po +78 -0
- data/ext/libcharguess/cpp/.deps/MBCSGroupProber.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/MBCSGroupProber.Po +102 -0
- data/ext/libcharguess/cpp/.deps/MBCSSM.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/MBCSSM.Po +77 -0
- data/ext/libcharguess/cpp/.deps/SBCSGroupProber.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/SBCSGroupProber.Po +80 -0
- data/ext/libcharguess/cpp/.deps/SBCharsetProber.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/SBCharsetProber.Po +78 -0
- data/ext/libcharguess/cpp/.deps/SJISProber.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/SJISProber.Po +86 -0
- data/ext/libcharguess/cpp/.deps/UTF8Prober.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/UTF8Prober.Po +82 -0
- data/ext/libcharguess/cpp/.deps/big5Prober.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/big5Prober.Po +84 -0
- data/ext/libcharguess/cpp/.deps/charDistribution.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/charDistribution.Po +87 -0
- data/ext/libcharguess/cpp/.deps/chardet.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/chardet.Po +84 -0
- data/ext/libcharguess/cpp/.deps/charguess.Po +77 -0
- data/ext/libcharguess/cpp/.deps/jpCntx.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/jpCntx.Po +75 -0
- data/ext/libcharguess/cpp/.deps/universal.Plo +1 -0
- data/ext/libcharguess/cpp/.deps/universal.Po +111 -0
- data/ext/libcharguess/cpp/AUTHORS +3 -0
- data/ext/libcharguess/cpp/Big5Freq.tab +928 -0
- data/ext/libcharguess/cpp/COPYING +340 -0
- data/ext/libcharguess/cpp/COPYRIGHT +20 -0
- data/ext/libcharguess/cpp/ChangeLog +0 -0
- data/ext/libcharguess/cpp/EUCJPProber.cpp +80 -0
- data/ext/libcharguess/cpp/EUCJPProber.h +58 -0
- data/ext/libcharguess/cpp/EUCKRFreq.tab +615 -0
- data/ext/libcharguess/cpp/EUCKRProber.cpp +80 -0
- data/ext/libcharguess/cpp/EUCKRProber.h +54 -0
- data/ext/libcharguess/cpp/EUCTWFreq.tab +448 -0
- data/ext/libcharguess/cpp/EUCTWProber.cpp +79 -0
- data/ext/libcharguess/cpp/EUCTWProber.h +53 -0
- data/ext/libcharguess/cpp/EscCharsetProber.cpp +89 -0
- data/ext/libcharguess/cpp/EscCharsetProber.h +49 -0
- data/ext/libcharguess/cpp/EscSM.cpp +244 -0
- data/ext/libcharguess/cpp/GB2312Freq.tab +476 -0
- data/ext/libcharguess/cpp/GB2312Prober.cpp +84 -0
- data/ext/libcharguess/cpp/GB2312Prober.h +56 -0
- data/ext/libcharguess/cpp/INSTALL +229 -0
- data/ext/libcharguess/cpp/JISFreq.tab +574 -0
- data/ext/libcharguess/cpp/LICENSE +504 -0
- data/ext/libcharguess/cpp/LangBulgarianModel.cpp +230 -0
- data/ext/libcharguess/cpp/LangCyrillicModel.cpp +340 -0
- data/ext/libcharguess/cpp/LangGreekModel.cpp +229 -0
- data/ext/libcharguess/cpp/LangHungarianModel.cpp +228 -0
- data/ext/libcharguess/cpp/LangThaiModel.cpp +206 -0
- data/ext/libcharguess/cpp/Latin1Prober.cpp +190 -0
- data/ext/libcharguess/cpp/Latin1Prober.h +49 -0
- data/ext/libcharguess/cpp/MBCSGroupProber.cpp +186 -0
- data/ext/libcharguess/cpp/MBCSGroupProber.h +58 -0
- data/ext/libcharguess/cpp/MBCSSM.cpp +610 -0
- data/ext/libcharguess/cpp/Makefile.am +45 -0
- data/ext/libcharguess/cpp/Makefile.in +608 -0
- data/ext/libcharguess/cpp/NEWS +0 -0
- data/ext/libcharguess/cpp/README +0 -0
- data/ext/libcharguess/cpp/SBCSGroupProber.cpp +244 -0
- data/ext/libcharguess/cpp/SBCSGroupProber.h +54 -0
- data/ext/libcharguess/cpp/SBCharsetProber.cpp +100 -0
- data/ext/libcharguess/cpp/SBCharsetProber.h +89 -0
- data/ext/libcharguess/cpp/SJISProber.cpp +86 -0
- data/ext/libcharguess/cpp/SJISProber.h +60 -0
- data/ext/libcharguess/cpp/UTF8Prober.cpp +75 -0
- data/ext/libcharguess/cpp/UTF8Prober.h +46 -0
- data/ext/libcharguess/cpp/aclocal.m4 +1008 -0
- data/ext/libcharguess/cpp/autogen.sh +153 -0
- data/ext/libcharguess/cpp/big5Prober.cpp +76 -0
- data/ext/libcharguess/cpp/big5Prober.h +53 -0
- data/ext/libcharguess/cpp/charDistribution.cpp +90 -0
- data/ext/libcharguess/cpp/charDistribution.h +219 -0
- data/ext/libcharguess/cpp/charguess.cpp +56 -0
- data/ext/libcharguess/cpp/charguess.h +23 -0
- data/ext/libcharguess/cpp/charsetProber.h +50 -0
- data/ext/libcharguess/cpp/codingStateMachine.h +92 -0
- data/ext/libcharguess/cpp/config.h +36 -0
- data/ext/libcharguess/cpp/config.h.in +35 -0
- data/ext/libcharguess/cpp/config.status +1075 -0
- data/ext/libcharguess/cpp/configure +5226 -0
- data/ext/libcharguess/cpp/configure.in +49 -0
- data/ext/libcharguess/cpp/depcomp +472 -0
- data/ext/libcharguess/cpp/fix_copyright +32 -0
- data/ext/libcharguess/cpp/install-sh +294 -0
- data/ext/libcharguess/cpp/jpCntx.cpp +194 -0
- data/ext/libcharguess/cpp/jpCntx.h +100 -0
- data/ext/libcharguess/cpp/missing +336 -0
- data/ext/libcharguess/cpp/mkinstalldirs +111 -0
- data/ext/libcharguess/cpp/pkgInt.h +72 -0
- data/ext/libcharguess/cpp/stamp-h1 +1 -0
- data/ext/libcharguess/cpp/test/test.cpp +78 -0
- data/ext/libcharguess/cpp/types.h +41 -0
- data/ext/libcharguess/cpp/universal.cpp +273 -0
- data/ext/libcharguess/cpp/universal.h +65 -0
- data/script/console +9 -0
- data/script/destroy +14 -0
- data/script/generate +14 -0
- data/tasks/extconf/charguess.rake +47 -0
- data/tasks/extconf.rake +13 -0
- data/test/test_charguess.rb +7 -0
- data/test/test_charguess_extn.rb +10 -0
- data/test/test_helper.rb +3 -0
- metadata +219 -0
@@ -0,0 +1,610 @@
|
|
1
|
+
/*
|
2
|
+
libcharguess - Guess the encoding/charset of a string
|
3
|
+
Copyright (C) 2003 Stephane Corbe <noubi@users.sourceforge.net>
|
4
|
+
Based on Mozilla sources
|
5
|
+
|
6
|
+
This library is free software; you can redistribute it and/or
|
7
|
+
modify it under the terms of the GNU Lesser General Public
|
8
|
+
License as published by the Free Software Foundation; either
|
9
|
+
version 2.1 of the License, or (at your option) any later version.
|
10
|
+
|
11
|
+
This library is distributed in the hope that it will be useful,
|
12
|
+
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
13
|
+
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
14
|
+
Lesser General Public License for more details.
|
15
|
+
|
16
|
+
You should have received a copy of the GNU Lesser General Public
|
17
|
+
License along with this library; if not, write to the Free Software
|
18
|
+
Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
19
|
+
*/
|
20
|
+
|
21
|
+
#include "codingStateMachine.h"
|
22
|
+
|
23
|
+
/*
|
24
|
+
Modification from frank tang's original work:
|
25
|
+
. 0x00 is allowed as a legal character. Since some web pages contains this char in
|
26
|
+
text stream.
|
27
|
+
*/
|
28
|
+
|
29
|
+
// BIG5
|
30
|
+
|
31
|
+
static PRUint32 BIG5_cls [ 256 / 8 ] = {
|
32
|
+
//PCK4BITS(0,1,1,1,1,1,1,1), // 00 - 07
|
33
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 00 - 07 //allow 0x00 as legal value
|
34
|
+
PCK4BITS(1,1,1,1,1,1,0,0), // 08 - 0f
|
35
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 10 - 17
|
36
|
+
PCK4BITS(1,1,1,0,1,1,1,1), // 18 - 1f
|
37
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 20 - 27
|
38
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 28 - 2f
|
39
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 30 - 37
|
40
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 38 - 3f
|
41
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 40 - 47
|
42
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 48 - 4f
|
43
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 50 - 57
|
44
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 58 - 5f
|
45
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 60 - 67
|
46
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 68 - 6f
|
47
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 70 - 77
|
48
|
+
PCK4BITS(2,2,2,2,2,2,2,1), // 78 - 7f
|
49
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 80 - 87
|
50
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 88 - 8f
|
51
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 90 - 97
|
52
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 98 - 9f
|
53
|
+
PCK4BITS(4,3,3,3,3,3,3,3), // a0 - a7
|
54
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // a8 - af
|
55
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // b0 - b7
|
56
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // b8 - bf
|
57
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // c0 - c7
|
58
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // c8 - cf
|
59
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // d0 - d7
|
60
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // d8 - df
|
61
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // e0 - e7
|
62
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // e8 - ef
|
63
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // f0 - f7
|
64
|
+
PCK4BITS(3,3,3,3,3,3,3,0) // f8 - ff
|
65
|
+
};
|
66
|
+
|
67
|
+
|
68
|
+
static PRUint32 BIG5_st [ 3] = {
|
69
|
+
PCK4BITS(eError,eStart,eStart, 3,eError,eError,eError,eError),//00-07
|
70
|
+
PCK4BITS(eError,eError,eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eError),//08-0f
|
71
|
+
PCK4BITS(eError,eStart,eStart,eStart,eStart,eStart,eStart,eStart) //10-17
|
72
|
+
};
|
73
|
+
|
74
|
+
static const PRUint32 Big5CharLenTable[] = {0, 1, 1, 2, 0};
|
75
|
+
|
76
|
+
SMModel Big5SMModel = {
|
77
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, BIG5_cls },
|
78
|
+
5,
|
79
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, BIG5_st },
|
80
|
+
Big5CharLenTable,
|
81
|
+
"Big5",
|
82
|
+
};
|
83
|
+
|
84
|
+
static PRUint32 EUCJP_cls [ 256 / 8 ] = {
|
85
|
+
//PCK4BITS(5,4,4,4,4,4,4,4), // 00 - 07
|
86
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 00 - 07
|
87
|
+
PCK4BITS(4,4,4,4,4,4,5,5), // 08 - 0f
|
88
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 10 - 17
|
89
|
+
PCK4BITS(4,4,4,5,4,4,4,4), // 18 - 1f
|
90
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 20 - 27
|
91
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 28 - 2f
|
92
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 30 - 37
|
93
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 38 - 3f
|
94
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 40 - 47
|
95
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 48 - 4f
|
96
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 50 - 57
|
97
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 58 - 5f
|
98
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 60 - 67
|
99
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 68 - 6f
|
100
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 70 - 77
|
101
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 78 - 7f
|
102
|
+
PCK4BITS(5,5,5,5,5,5,5,5), // 80 - 87
|
103
|
+
PCK4BITS(5,5,5,5,5,5,1,3), // 88 - 8f
|
104
|
+
PCK4BITS(5,5,5,5,5,5,5,5), // 90 - 97
|
105
|
+
PCK4BITS(5,5,5,5,5,5,5,5), // 98 - 9f
|
106
|
+
PCK4BITS(5,2,2,2,2,2,2,2), // a0 - a7
|
107
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // a8 - af
|
108
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b0 - b7
|
109
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b8 - bf
|
110
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // c0 - c7
|
111
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // c8 - cf
|
112
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d0 - d7
|
113
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d8 - df
|
114
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // e0 - e7
|
115
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // e8 - ef
|
116
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // f0 - f7
|
117
|
+
PCK4BITS(0,0,0,0,0,0,0,5) // f8 - ff
|
118
|
+
};
|
119
|
+
|
120
|
+
|
121
|
+
static PRUint32 EUCJP_st [ 5] = {
|
122
|
+
PCK4BITS( 3, 4, 3, 5,eStart,eError,eError,eError),//00-07
|
123
|
+
PCK4BITS(eError,eError,eError,eError,eItsMe,eItsMe,eItsMe,eItsMe),//08-0f
|
124
|
+
PCK4BITS(eItsMe,eItsMe,eStart,eError,eStart,eError,eError,eError),//10-17
|
125
|
+
PCK4BITS(eError,eError,eStart,eError,eError,eError, 3,eError),//18-1f
|
126
|
+
PCK4BITS( 3,eError,eError,eError,eStart,eStart,eStart,eStart) //20-27
|
127
|
+
};
|
128
|
+
|
129
|
+
static const PRUint32 EUCJPCharLenTable[] = {2, 2, 2, 3, 1, 0};
|
130
|
+
|
131
|
+
SMModel EUCJPSMModel = {
|
132
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, EUCJP_cls },
|
133
|
+
6,
|
134
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, EUCJP_st },
|
135
|
+
EUCJPCharLenTable,
|
136
|
+
"EUC-JP",
|
137
|
+
};
|
138
|
+
|
139
|
+
static PRUint32 EUCKR_cls [ 256 / 8 ] = {
|
140
|
+
//PCK4BITS(0,1,1,1,1,1,1,1), // 00 - 07
|
141
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 00 - 07
|
142
|
+
PCK4BITS(1,1,1,1,1,1,0,0), // 08 - 0f
|
143
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 10 - 17
|
144
|
+
PCK4BITS(1,1,1,0,1,1,1,1), // 18 - 1f
|
145
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 20 - 27
|
146
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 28 - 2f
|
147
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 30 - 37
|
148
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 38 - 3f
|
149
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 40 - 47
|
150
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 48 - 4f
|
151
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 50 - 57
|
152
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 58 - 5f
|
153
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 60 - 67
|
154
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 68 - 6f
|
155
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 70 - 77
|
156
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 78 - 7f
|
157
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 80 - 87
|
158
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 88 - 8f
|
159
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 90 - 97
|
160
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 98 - 9f
|
161
|
+
PCK4BITS(0,2,2,2,2,2,2,2), // a0 - a7
|
162
|
+
PCK4BITS(2,2,2,2,2,3,3,3), // a8 - af
|
163
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b0 - b7
|
164
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b8 - bf
|
165
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // c0 - c7
|
166
|
+
PCK4BITS(2,3,2,2,2,2,2,2), // c8 - cf
|
167
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d0 - d7
|
168
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d8 - df
|
169
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // e0 - e7
|
170
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // e8 - ef
|
171
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // f0 - f7
|
172
|
+
PCK4BITS(2,2,2,2,2,2,2,0) // f8 - ff
|
173
|
+
};
|
174
|
+
|
175
|
+
|
176
|
+
static PRUint32 EUCKR_st [ 2] = {
|
177
|
+
PCK4BITS(eError,eStart, 3,eError,eError,eError,eError,eError),//00-07
|
178
|
+
PCK4BITS(eItsMe,eItsMe,eItsMe,eItsMe,eError,eError,eStart,eStart) //08-0f
|
179
|
+
};
|
180
|
+
|
181
|
+
static const PRUint32 EUCKRCharLenTable[] = {0, 1, 2, 0};
|
182
|
+
|
183
|
+
SMModel EUCKRSMModel = {
|
184
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, EUCKR_cls },
|
185
|
+
4,
|
186
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, EUCKR_st },
|
187
|
+
EUCKRCharLenTable,
|
188
|
+
"EUC-KR",
|
189
|
+
};
|
190
|
+
|
191
|
+
static PRUint32 EUCTW_cls [ 256 / 8 ] = {
|
192
|
+
//PCK4BITS(0,2,2,2,2,2,2,2), // 00 - 07
|
193
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 00 - 07
|
194
|
+
PCK4BITS(2,2,2,2,2,2,0,0), // 08 - 0f
|
195
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 10 - 17
|
196
|
+
PCK4BITS(2,2,2,0,2,2,2,2), // 18 - 1f
|
197
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 20 - 27
|
198
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 28 - 2f
|
199
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 30 - 37
|
200
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 38 - 3f
|
201
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 40 - 47
|
202
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 48 - 4f
|
203
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 50 - 57
|
204
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 58 - 5f
|
205
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 60 - 67
|
206
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 68 - 6f
|
207
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 70 - 77
|
208
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 78 - 7f
|
209
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 80 - 87
|
210
|
+
PCK4BITS(0,0,0,0,0,0,6,0), // 88 - 8f
|
211
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 90 - 97
|
212
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 98 - 9f
|
213
|
+
PCK4BITS(0,3,4,4,4,4,4,4), // a0 - a7
|
214
|
+
PCK4BITS(5,5,1,1,1,1,1,1), // a8 - af
|
215
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // b0 - b7
|
216
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // b8 - bf
|
217
|
+
PCK4BITS(1,1,3,1,3,3,3,3), // c0 - c7
|
218
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // c8 - cf
|
219
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // d0 - d7
|
220
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // d8 - df
|
221
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // e0 - e7
|
222
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // e8 - ef
|
223
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // f0 - f7
|
224
|
+
PCK4BITS(3,3,3,3,3,3,3,0) // f8 - ff
|
225
|
+
};
|
226
|
+
|
227
|
+
|
228
|
+
static PRUint32 EUCTW_st [ 6] = {
|
229
|
+
PCK4BITS(eError,eError,eStart, 3, 3, 3, 4,eError),//00-07
|
230
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eItsMe,eItsMe),//08-0f
|
231
|
+
PCK4BITS(eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eError,eStart,eError),//10-17
|
232
|
+
PCK4BITS(eStart,eStart,eStart,eError,eError,eError,eError,eError),//18-1f
|
233
|
+
PCK4BITS( 5,eError,eError,eError,eStart,eError,eStart,eStart),//20-27
|
234
|
+
PCK4BITS(eStart,eError,eStart,eStart,eStart,eStart,eStart,eStart) //28-2f
|
235
|
+
};
|
236
|
+
|
237
|
+
static const PRUint32 EUCTWCharLenTable[] = {0, 0, 1, 2, 2, 2, 3};
|
238
|
+
|
239
|
+
SMModel EUCTWSMModel = {
|
240
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, EUCTW_cls },
|
241
|
+
7,
|
242
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, EUCTW_st },
|
243
|
+
EUCTWCharLenTable,
|
244
|
+
"x-euc-tw",
|
245
|
+
};
|
246
|
+
|
247
|
+
/* obsolete GB2312 by gb18030
|
248
|
+
static PRUint32 GB2312_cls [ 256 / 8 ] = {
|
249
|
+
//PCK4BITS(0,1,1,1,1,1,1,1), // 00 - 07
|
250
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 00 - 07
|
251
|
+
PCK4BITS(1,1,1,1,1,1,0,0), // 08 - 0f
|
252
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 10 - 17
|
253
|
+
PCK4BITS(1,1,1,0,1,1,1,1), // 18 - 1f
|
254
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 20 - 27
|
255
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 28 - 2f
|
256
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 30 - 37
|
257
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 38 - 3f
|
258
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 40 - 47
|
259
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 48 - 4f
|
260
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 50 - 57
|
261
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 58 - 5f
|
262
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 60 - 67
|
263
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 68 - 6f
|
264
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 70 - 77
|
265
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 78 - 7f
|
266
|
+
PCK4BITS(1,0,0,0,0,0,0,0), // 80 - 87
|
267
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 88 - 8f
|
268
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 90 - 97
|
269
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 98 - 9f
|
270
|
+
PCK4BITS(0,2,2,2,2,2,2,2), // a0 - a7
|
271
|
+
PCK4BITS(2,2,3,3,3,3,3,3), // a8 - af
|
272
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b0 - b7
|
273
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b8 - bf
|
274
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // c0 - c7
|
275
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // c8 - cf
|
276
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d0 - d7
|
277
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d8 - df
|
278
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // e0 - e7
|
279
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // e8 - ef
|
280
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // f0 - f7
|
281
|
+
PCK4BITS(2,2,2,2,2,2,2,0) // f8 - ff
|
282
|
+
};
|
283
|
+
|
284
|
+
|
285
|
+
static PRUint32 GB2312_st [ 2] = {
|
286
|
+
PCK4BITS(eError,eStart, 3,eError,eError,eError,eError,eError),//00-07
|
287
|
+
PCK4BITS(eItsMe,eItsMe,eItsMe,eItsMe,eError,eError,eStart,eStart) //08-0f
|
288
|
+
};
|
289
|
+
|
290
|
+
static const PRUint32 GB2312CharLenTable[] = {0, 1, 2, 0};
|
291
|
+
|
292
|
+
SMModel GB2312SMModel = {
|
293
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, GB2312_cls },
|
294
|
+
4,
|
295
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, GB2312_st },
|
296
|
+
GB2312CharLenTable,
|
297
|
+
"GB2312",
|
298
|
+
};
|
299
|
+
*/
|
300
|
+
|
301
|
+
// the following state machine data was created by perl script in
|
302
|
+
// intl/chardet/tools. It should be the same as in PSM detector.
|
303
|
+
static PRUint32 GB18030_cls [ 256 / 8 ] = {
|
304
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 00 - 07
|
305
|
+
PCK4BITS(1,1,1,1,1,1,0,0), // 08 - 0f
|
306
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 10 - 17
|
307
|
+
PCK4BITS(1,1,1,0,1,1,1,1), // 18 - 1f
|
308
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 20 - 27
|
309
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 28 - 2f
|
310
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // 30 - 37
|
311
|
+
PCK4BITS(3,3,1,1,1,1,1,1), // 38 - 3f
|
312
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 40 - 47
|
313
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 48 - 4f
|
314
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 50 - 57
|
315
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 58 - 5f
|
316
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 60 - 67
|
317
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 68 - 6f
|
318
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 70 - 77
|
319
|
+
PCK4BITS(2,2,2,2,2,2,2,4), // 78 - 7f
|
320
|
+
PCK4BITS(5,6,6,6,6,6,6,6), // 80 - 87
|
321
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // 88 - 8f
|
322
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // 90 - 97
|
323
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // 98 - 9f
|
324
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // a0 - a7
|
325
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // a8 - af
|
326
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // b0 - b7
|
327
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // b8 - bf
|
328
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // c0 - c7
|
329
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // c8 - cf
|
330
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // d0 - d7
|
331
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // d8 - df
|
332
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // e0 - e7
|
333
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // e8 - ef
|
334
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // f0 - f7
|
335
|
+
PCK4BITS(6,6,6,6,6,6,6,0) // f8 - ff
|
336
|
+
};
|
337
|
+
|
338
|
+
|
339
|
+
static PRUint32 GB18030_st [ 6] = {
|
340
|
+
PCK4BITS(eError,eStart,eStart,eStart,eStart,eStart, 3,eError),//00-07
|
341
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eItsMe,eItsMe),//08-0f
|
342
|
+
PCK4BITS(eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eError,eError,eStart),//10-17
|
343
|
+
PCK4BITS( 4,eError,eStart,eStart,eError,eError,eError,eError),//18-1f
|
344
|
+
PCK4BITS(eError,eError, 5,eError,eError,eError,eItsMe,eError),//20-27
|
345
|
+
PCK4BITS(eError,eError,eStart,eStart,eStart,eStart,eStart,eStart) //28-2f
|
346
|
+
};
|
347
|
+
|
348
|
+
// To be accurate, the length of class 6 can be either 2 or 4.
|
349
|
+
// But it is not necessary to discriminate between the two since
|
350
|
+
// it is used for frequency analysis only, and we are validing
|
351
|
+
// each code range there as well. So it is safe to set it to be
|
352
|
+
// 2 here.
|
353
|
+
static const PRUint32 GB18030CharLenTable[] = {0, 1, 1, 1, 1, 1, 2};
|
354
|
+
|
355
|
+
SMModel GB18030SMModel = {
|
356
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, GB18030_cls },
|
357
|
+
7,
|
358
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, GB18030_st },
|
359
|
+
GB18030CharLenTable,
|
360
|
+
"GB18030",
|
361
|
+
};
|
362
|
+
|
363
|
+
// sjis
|
364
|
+
|
365
|
+
static PRUint32 SJIS_cls [ 256 / 8 ] = {
|
366
|
+
//PCK4BITS(0,1,1,1,1,1,1,1), // 00 - 07
|
367
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 00 - 07
|
368
|
+
PCK4BITS(1,1,1,1,1,1,0,0), // 08 - 0f
|
369
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 10 - 17
|
370
|
+
PCK4BITS(1,1,1,0,1,1,1,1), // 18 - 1f
|
371
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 20 - 27
|
372
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 28 - 2f
|
373
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 30 - 37
|
374
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 38 - 3f
|
375
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 40 - 47
|
376
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 48 - 4f
|
377
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 50 - 57
|
378
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 58 - 5f
|
379
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 60 - 67
|
380
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 68 - 6f
|
381
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // 70 - 77
|
382
|
+
PCK4BITS(2,2,2,2,2,2,2,1), // 78 - 7f
|
383
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // 80 - 87
|
384
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // 88 - 8f
|
385
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // 90 - 97
|
386
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // 98 - 9f
|
387
|
+
//0xa0 is illegal in sjis encoding, but some pages does
|
388
|
+
//contain such byte. We need to be more error forgiven.
|
389
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // a0 - a7
|
390
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // a8 - af
|
391
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b0 - b7
|
392
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // b8 - bf
|
393
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // c0 - c7
|
394
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // c8 - cf
|
395
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d0 - d7
|
396
|
+
PCK4BITS(2,2,2,2,2,2,2,2), // d8 - df
|
397
|
+
PCK4BITS(3,3,3,3,3,3,3,3), // e0 - e7
|
398
|
+
PCK4BITS(3,3,3,3,3,4,4,4), // e8 - ef
|
399
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // f0 - f7
|
400
|
+
PCK4BITS(4,4,4,4,4,0,0,0) // f8 - ff
|
401
|
+
};
|
402
|
+
|
403
|
+
|
404
|
+
static PRUint32 SJIS_st [ 3] = {
|
405
|
+
PCK4BITS(eError,eStart,eStart, 3,eError,eError,eError,eError),//00-07
|
406
|
+
PCK4BITS(eError,eError,eError,eError,eItsMe,eItsMe,eItsMe,eItsMe),//08-0f
|
407
|
+
PCK4BITS(eItsMe,eItsMe,eError,eError,eStart,eStart,eStart,eStart) //10-17
|
408
|
+
};
|
409
|
+
|
410
|
+
static const PRUint32 SJISCharLenTable[] = {0, 1, 1, 2, 0, 0};
|
411
|
+
|
412
|
+
SMModel SJISSMModel = {
|
413
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, SJIS_cls },
|
414
|
+
6,
|
415
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, SJIS_st },
|
416
|
+
SJISCharLenTable,
|
417
|
+
"Shift_JIS",
|
418
|
+
};
|
419
|
+
|
420
|
+
|
421
|
+
static PRUint32 UCS2BE_cls [ 256 / 8 ] = {
|
422
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 00 - 07
|
423
|
+
PCK4BITS(0,0,1,0,0,2,0,0), // 08 - 0f
|
424
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 10 - 17
|
425
|
+
PCK4BITS(0,0,0,3,0,0,0,0), // 18 - 1f
|
426
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 20 - 27
|
427
|
+
PCK4BITS(0,3,3,3,3,3,0,0), // 28 - 2f
|
428
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 30 - 37
|
429
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 38 - 3f
|
430
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 40 - 47
|
431
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 48 - 4f
|
432
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 50 - 57
|
433
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 58 - 5f
|
434
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 60 - 67
|
435
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 68 - 6f
|
436
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 70 - 77
|
437
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 78 - 7f
|
438
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 80 - 87
|
439
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 88 - 8f
|
440
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 90 - 97
|
441
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 98 - 9f
|
442
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // a0 - a7
|
443
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // a8 - af
|
444
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // b0 - b7
|
445
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // b8 - bf
|
446
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // c0 - c7
|
447
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // c8 - cf
|
448
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // d0 - d7
|
449
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // d8 - df
|
450
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // e0 - e7
|
451
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // e8 - ef
|
452
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // f0 - f7
|
453
|
+
PCK4BITS(0,0,0,0,0,0,4,5) // f8 - ff
|
454
|
+
};
|
455
|
+
|
456
|
+
|
457
|
+
static PRUint32 UCS2BE_st [ 7] = {
|
458
|
+
PCK4BITS( 5, 7, 7,eError, 4, 3,eError,eError),//00-07
|
459
|
+
PCK4BITS(eError,eError,eError,eError,eItsMe,eItsMe,eItsMe,eItsMe),//08-0f
|
460
|
+
PCK4BITS(eItsMe,eItsMe, 6, 6, 6, 6,eError,eError),//10-17
|
461
|
+
PCK4BITS( 6, 6, 6, 6, 6,eItsMe, 6, 6),//18-1f
|
462
|
+
PCK4BITS( 6, 6, 6, 6, 5, 7, 7,eError),//20-27
|
463
|
+
PCK4BITS( 5, 8, 6, 6,eError, 6, 6, 6),//28-2f
|
464
|
+
PCK4BITS( 6, 6, 6, 6,eError,eError,eStart,eStart) //30-37
|
465
|
+
};
|
466
|
+
|
467
|
+
static const PRUint32 UCS2BECharLenTable[] = {2, 2, 2, 0, 2, 2};
|
468
|
+
|
469
|
+
SMModel UCS2BESMModel = {
|
470
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, UCS2BE_cls },
|
471
|
+
6,
|
472
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, UCS2BE_st },
|
473
|
+
UCS2BECharLenTable,
|
474
|
+
"UTF-16BE",
|
475
|
+
};
|
476
|
+
|
477
|
+
static PRUint32 UCS2LE_cls [ 256 / 8 ] = {
|
478
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 00 - 07
|
479
|
+
PCK4BITS(0,0,1,0,0,2,0,0), // 08 - 0f
|
480
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 10 - 17
|
481
|
+
PCK4BITS(0,0,0,3,0,0,0,0), // 18 - 1f
|
482
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 20 - 27
|
483
|
+
PCK4BITS(0,3,3,3,3,3,0,0), // 28 - 2f
|
484
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 30 - 37
|
485
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 38 - 3f
|
486
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 40 - 47
|
487
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 48 - 4f
|
488
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 50 - 57
|
489
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 58 - 5f
|
490
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 60 - 67
|
491
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 68 - 6f
|
492
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 70 - 77
|
493
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 78 - 7f
|
494
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 80 - 87
|
495
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 88 - 8f
|
496
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 90 - 97
|
497
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // 98 - 9f
|
498
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // a0 - a7
|
499
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // a8 - af
|
500
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // b0 - b7
|
501
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // b8 - bf
|
502
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // c0 - c7
|
503
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // c8 - cf
|
504
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // d0 - d7
|
505
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // d8 - df
|
506
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // e0 - e7
|
507
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // e8 - ef
|
508
|
+
PCK4BITS(0,0,0,0,0,0,0,0), // f0 - f7
|
509
|
+
PCK4BITS(0,0,0,0,0,0,4,5) // f8 - ff
|
510
|
+
};
|
511
|
+
|
512
|
+
|
513
|
+
static PRUint32 UCS2LE_st [ 7] = {
|
514
|
+
PCK4BITS( 6, 6, 7, 6, 4, 3,eError,eError),//00-07
|
515
|
+
PCK4BITS(eError,eError,eError,eError,eItsMe,eItsMe,eItsMe,eItsMe),//08-0f
|
516
|
+
PCK4BITS(eItsMe,eItsMe, 5, 5, 5,eError,eItsMe,eError),//10-17
|
517
|
+
PCK4BITS( 5, 5, 5,eError, 5,eError, 6, 6),//18-1f
|
518
|
+
PCK4BITS( 7, 6, 8, 8, 5, 5, 5,eError),//20-27
|
519
|
+
PCK4BITS( 5, 5, 5,eError,eError,eError, 5, 5),//28-2f
|
520
|
+
PCK4BITS( 5, 5, 5,eError, 5,eError,eStart,eStart) //30-37
|
521
|
+
};
|
522
|
+
|
523
|
+
static const PRUint32 UCS2LECharLenTable[] = {2, 2, 2, 2, 2, 2};
|
524
|
+
|
525
|
+
SMModel UCS2LESMModel = {
|
526
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, UCS2LE_cls },
|
527
|
+
6,
|
528
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, UCS2LE_st },
|
529
|
+
UCS2LECharLenTable,
|
530
|
+
"UTF-16LE",
|
531
|
+
};
|
532
|
+
|
533
|
+
|
534
|
+
static PRUint32 UTF8_cls [ 256 / 8 ] = {
|
535
|
+
//PCK4BITS(0,1,1,1,1,1,1,1), // 00 - 07
|
536
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 00 - 07 //allow 0x00 as a legal value
|
537
|
+
PCK4BITS(1,1,1,1,1,1,0,0), // 08 - 0f
|
538
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 10 - 17
|
539
|
+
PCK4BITS(1,1,1,0,1,1,1,1), // 18 - 1f
|
540
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 20 - 27
|
541
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 28 - 2f
|
542
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 30 - 37
|
543
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 38 - 3f
|
544
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 40 - 47
|
545
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 48 - 4f
|
546
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 50 - 57
|
547
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 58 - 5f
|
548
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 60 - 67
|
549
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 68 - 6f
|
550
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 70 - 77
|
551
|
+
PCK4BITS(1,1,1,1,1,1,1,1), // 78 - 7f
|
552
|
+
PCK4BITS(2,2,2,2,3,3,3,3), // 80 - 87
|
553
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 88 - 8f
|
554
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 90 - 97
|
555
|
+
PCK4BITS(4,4,4,4,4,4,4,4), // 98 - 9f
|
556
|
+
PCK4BITS(5,5,5,5,5,5,5,5), // a0 - a7
|
557
|
+
PCK4BITS(5,5,5,5,5,5,5,5), // a8 - af
|
558
|
+
PCK4BITS(5,5,5,5,5,5,5,5), // b0 - b7
|
559
|
+
PCK4BITS(5,5,5,5,5,5,5,5), // b8 - bf
|
560
|
+
PCK4BITS(0,0,6,6,6,6,6,6), // c0 - c7
|
561
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // c8 - cf
|
562
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // d0 - d7
|
563
|
+
PCK4BITS(6,6,6,6,6,6,6,6), // d8 - df
|
564
|
+
PCK4BITS(7,8,8,8,8,8,8,8), // e0 - e7
|
565
|
+
PCK4BITS(8,8,8,8,8,9,8,8), // e8 - ef
|
566
|
+
PCK4BITS(10,11,11,11,11,11,11,11), // f0 - f7
|
567
|
+
PCK4BITS(12,13,13,13,14,15,0,0) // f8 - ff
|
568
|
+
};
|
569
|
+
|
570
|
+
|
571
|
+
static PRUint32 UTF8_st [ 26] = {
|
572
|
+
PCK4BITS(eError,eStart,eError,eError,eError,eError, 12, 10),//00-07
|
573
|
+
PCK4BITS( 9, 11, 8, 7, 6, 5, 4, 3),//08-0f
|
574
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//10-17
|
575
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//18-1f
|
576
|
+
PCK4BITS(eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eItsMe),//20-27
|
577
|
+
PCK4BITS(eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eItsMe,eItsMe),//28-2f
|
578
|
+
PCK4BITS(eError,eError, 5, 5, 5, 5,eError,eError),//30-37
|
579
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//38-3f
|
580
|
+
PCK4BITS(eError,eError,eError, 5, 5, 5,eError,eError),//40-47
|
581
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//48-4f
|
582
|
+
PCK4BITS(eError,eError, 7, 7, 7, 7,eError,eError),//50-57
|
583
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//58-5f
|
584
|
+
PCK4BITS(eError,eError,eError,eError, 7, 7,eError,eError),//60-67
|
585
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//68-6f
|
586
|
+
PCK4BITS(eError,eError, 9, 9, 9, 9,eError,eError),//70-77
|
587
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//78-7f
|
588
|
+
PCK4BITS(eError,eError,eError,eError,eError, 9,eError,eError),//80-87
|
589
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//88-8f
|
590
|
+
PCK4BITS(eError,eError, 12, 12, 12, 12,eError,eError),//90-97
|
591
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//98-9f
|
592
|
+
PCK4BITS(eError,eError,eError,eError,eError, 12,eError,eError),//a0-a7
|
593
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//a8-af
|
594
|
+
PCK4BITS(eError,eError, 12, 12, 12,eError,eError,eError),//b0-b7
|
595
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError),//b8-bf
|
596
|
+
PCK4BITS(eError,eError,eStart,eStart,eStart,eStart,eError,eError),//c0-c7
|
597
|
+
PCK4BITS(eError,eError,eError,eError,eError,eError,eError,eError) //c8-cf
|
598
|
+
};
|
599
|
+
|
600
|
+
static const PRUint32 UTF8CharLenTable[] = {0, 1, 0, 0, 0, 0, 2, 3,
|
601
|
+
3, 3, 4, 4, 5, 5, 6, 6 };
|
602
|
+
|
603
|
+
SMModel UTF8SMModel = {
|
604
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, UTF8_cls },
|
605
|
+
16,
|
606
|
+
{eIdxSft4bits, eSftMsk4bits, eBitSft4bits, eUnitMsk4bits, UTF8_st },
|
607
|
+
UTF8CharLenTable,
|
608
|
+
"UTF-8",
|
609
|
+
};
|
610
|
+
|
@@ -0,0 +1,45 @@
|
|
1
|
+
#----------------------------------------------------------------------
|
2
|
+
#
|
3
|
+
|
4
|
+
## Process this file with automake to produce Makefile.in
|
5
|
+
|
6
|
+
AM_CPPFLAGS = $(all_includes)
|
7
|
+
|
8
|
+
lib_LIBRARIES = libcharguess.a
|
9
|
+
|
10
|
+
# libcharguess_a_LIBADD =
|
11
|
+
# libcharguess_a_LDFLAGS = $(all_libraries) -version-info 0.1 -no-undefined
|
12
|
+
|
13
|
+
libcharguess_a_SOURCES = \
|
14
|
+
types.h \
|
15
|
+
charsetProber.h \
|
16
|
+
codingStateMachine.h pkgInt.h \
|
17
|
+
charguess.cpp charguess.h \
|
18
|
+
EscSM.cpp \
|
19
|
+
MBCSSM.cpp \
|
20
|
+
GB2312Prober.cpp GB2312Prober.h \
|
21
|
+
Latin1Prober.cpp Latin1Prober.h \
|
22
|
+
SJISProber.cpp SJISProber.h \
|
23
|
+
big5Prober.cpp big5Prober.h \
|
24
|
+
EUCJPProber.cpp EUCJPProber.h \
|
25
|
+
jpCntx.cpp jpCntx.h \
|
26
|
+
MBCSGroupProber.cpp MBCSGroupProber.h \
|
27
|
+
EUCKRProber.cpp EUCKRProber.h \
|
28
|
+
charDistribution.cpp charDistribution.h \
|
29
|
+
universal.cpp universal.h \
|
30
|
+
EUCTWProber.cpp EUCTWProber.h \
|
31
|
+
SBCharsetProber.cpp SBCharsetProber.h \
|
32
|
+
EscCharsetProber.cpp EscCharsetProber.h \
|
33
|
+
SBCSGroupProber.cpp SBCSGroupProber.h \
|
34
|
+
UTF8Prober.cpp UTF8Prober.h \
|
35
|
+
LangHungarianModel.cpp \
|
36
|
+
LangThaiModel.cpp \
|
37
|
+
LangCyrillicModel.cpp \
|
38
|
+
LangBulgarianModel.cpp \
|
39
|
+
LangGreekModel.cpp
|
40
|
+
|
41
|
+
METASOURCES = AUTO
|
42
|
+
|
43
|
+
CLEANFILES = *.bak *~
|
44
|
+
|
45
|
+
|