annotate libguess/guess.c @ 2301:1db6ecbedb27 trunk

[svn] Change /home/js/.audacious to /home/js/.config/audacious
author js
date Tue, 09 Jan 2007 07:45:01 -0800
parents a9bc621d6b1b
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
1 /*
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
2 * This code is derivative of guess.c of Gauche-0.8.3.
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
3 * The following is the original copyright notice.
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
4 */
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
5
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
6 /*
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
7 * guess.c - guessing character encoding
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
8 *
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
9 * Copyright (c) 2000-2003 Shiro Kawai, All rights reserved.
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
10 *
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
11 * Redistribution and use in source and binary forms, with or without
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
12 * modification, are permitted provided that the following conditions
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
13 * are met:
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
14 *
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
15 * 1. Redistributions of source code must retain the above copyright
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
16 * notice, this list of conditions and the following disclaimer.
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
17 *
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
18 * 2. Redistributions in binary form must reproduce the above copyright
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
19 * notice, this list of conditions and the following disclaimer in the
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
20 * documentation and/or other materials provided with the distribution.
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
21 *
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
22 * 3. Neither the name of the authors nor the names of its contributors
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
23 * may be used to endorse or promote products derived from this
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
24 * software without specific prior written permission.
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
25 *
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
26 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
27 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
28 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
29 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
30 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
31 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
32 * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
33 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
34 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
35 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
36 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
37 *
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
38 */
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
39
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
40 #include "libguess.h"
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
41 #define NULL ((void *)0)
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
42
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
43 /* take precedence if scores are same. you can customize the order as: */
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
44 /* ORDER_** &highest, &second, ... &lowest */
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
45 #define ORDER_JP &utf8, &sjis, &eucj
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
46 #define ORDER_TW &utf8, &big5
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
47 #define ORDER_CN &utf8, &gb2312, &gb18030
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
48 #define ORDER_KR &utf8, &euck, &johab
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
49
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
50 /* workaround for that glib's g_convert can't convert
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
51 properly from UCS-2BE/LE trailing after BOM. */
1595
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
52 #define WITH_G_CONVERT 1
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
53 /* #undef WITH_G_CONVERT */
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
54
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
55 #ifdef WITH_G_CONVERT
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
56 const char UCS_2BE[] = "UTF-16";
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
57 const char UCS_2LE[] = "UTF-16";
1595
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
58 #else
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
59 const char UCS_2BE[] = "UCS-2BE";
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
60 const char UCS_2LE[] = "UCS-2LE";
1595
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
61 #endif
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
62
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
63 /* data types */
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
64 typedef struct guess_arc_rec
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
65 {
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
66 unsigned int next; /* next state */
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
67 double score; /* score */
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
68 } guess_arc;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
69
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
70 typedef struct guess_dfa_rec
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
71 {
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
72 signed char (*states)[256];
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
73 guess_arc *arcs;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
74 int state;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
75 double score;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
76 } guess_dfa;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
77
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
78 /* macros */
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
79 #define DFA_INIT(st, ar) \
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
80 { st, ar, 0, 1.0 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
81
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
82 #define DFA_NEXT(dfa, ch) \
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
83 do { \
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
84 int arc__; \
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
85 if (dfa.state >= 0) { \
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
86 arc__ = dfa.states[dfa.state][ch]; \
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
87 if (arc__ < 0) { \
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
88 dfa.state = -1; \
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
89 } else { \
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
90 dfa.state = dfa.arcs[arc__].next; \
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
91 dfa.score *= dfa.arcs[arc__].score; \
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
92 } \
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
93 } \
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
94 } while (0)
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
95
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
96 #define DFA_ALIVE(dfa) (dfa.state >= 0)
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
97
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
98 /* include DFA table generated by guess.scm */
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
99 #include "guess_tab.c"
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
100
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
101 const char *guess_jp(const char *buf, int buflen)
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
102 {
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
103 int i;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
104 guess_dfa eucj = DFA_INIT(guess_eucj_st, guess_eucj_ar);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
105 guess_dfa sjis = DFA_INIT(guess_sjis_st, guess_sjis_ar);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
106 guess_dfa utf8 = DFA_INIT(guess_utf8_st, guess_utf8_ar);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
107 guess_dfa *top = NULL;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
108
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
109 guess_dfa *order[] = { ORDER_JP, NULL };
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
110
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
111 for (i = 0; i < buflen; i++) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
112 int c = (unsigned char) buf[i];
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
113
1595
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
114 /* special treatment of iso-2022 escape sequence */
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
115 if (c == 0x1b) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
116 if (i < buflen - 1) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
117 c = (unsigned char) buf[++i];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
118 if (c == '$' || c == '(')
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
119 return "ISO-2022-JP";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
120 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
121 }
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
122
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
123 /* special treatment of BOM */
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
124 if (i == 0 && c == 0xff) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
125 if (i < buflen - 1) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
126 c = (unsigned char) buf[i + 1];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
127 if (c == 0xfe)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
128 return UCS_2LE;
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
129 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
130 }
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
131 if (i == 0 && c == 0xfe) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
132 if (i < buflen - 1) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
133 c = (unsigned char) buf[i + 1];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
134 if (c == 0xff)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
135 return UCS_2BE;
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
136 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
137 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
138
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
139 if (DFA_ALIVE(eucj)) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
140 if (!DFA_ALIVE(sjis) && !DFA_ALIVE(utf8))
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
141 return "EUC-JP";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
142 DFA_NEXT(eucj, c);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
143 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
144 if (DFA_ALIVE(sjis)) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
145 if (!DFA_ALIVE(eucj) && !DFA_ALIVE(utf8))
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
146 return "SJIS";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
147 DFA_NEXT(sjis, c);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
148 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
149 if (DFA_ALIVE(utf8)) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
150 if (!DFA_ALIVE(sjis) && !DFA_ALIVE(eucj))
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
151 return "UTF-8";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
152 DFA_NEXT(utf8, c);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
153 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
154
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
155 if (!DFA_ALIVE(eucj) && !DFA_ALIVE(sjis) && !DFA_ALIVE(utf8)) {
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
156 /* we ran out the possibilities */
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
157 return NULL;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
158 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
159 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
160
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
161 /* Now, we have ambigous code. Pick the highest score. If more than
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
162 one candidate tie, pick the default encoding. */
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
163 for (i = 0; order[i] != NULL; i++) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
164 if (order[i]->state >= 0) { //DFA_ALIVE()
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
165 if (top == NULL || order[i]->score > top->score)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
166 top = order[i];
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
167 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
168 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
169
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
170 if (top == &eucj)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
171 return "EUC-JP";
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
172 if (top == &utf8)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
173 return "UTF-8";
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
174 if (top == &sjis)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
175 return "SJIS";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
176 return NULL;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
177 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
178
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
179 const char *guess_tw(const char *buf, int buflen)
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
180 {
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
181 int i;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
182 guess_dfa big5 = DFA_INIT(guess_big5_st, guess_big5_ar);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
183 guess_dfa utf8 = DFA_INIT(guess_utf8_st, guess_utf8_ar);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
184 guess_dfa *top = NULL;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
185
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
186 guess_dfa *order[] = { ORDER_TW, NULL };
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
187
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
188 for (i = 0; i < buflen; i++) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
189 int c = (unsigned char) buf[i];
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
190
1595
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
191 /* special treatment of iso-2022 escape sequence */
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
192 if (c == 0x1b) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
193 if (i < buflen - 1) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
194 c = (unsigned char) buf[++i];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
195 if (c == '$' || c == '(')
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
196 return "ISO-2022-TW";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
197 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
198 }
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
199
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
200 /* special treatment of BOM */
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
201 if (i == 0 && c == 0xff) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
202 if (i < buflen - 1) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
203 c = (unsigned char) buf[i + 1];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
204 if (c == 0xfe)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
205 return UCS_2LE;
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
206 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
207 }
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
208 if (i == 0 && c == 0xfe) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
209 if (i < buflen - 1) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
210 c = (unsigned char) buf[i + 1];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
211 if (c == 0xff)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
212 return UCS_2BE;
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
213 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
214 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
215
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
216 if (DFA_ALIVE(big5)) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
217 if (!DFA_ALIVE(utf8))
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
218 return "BIG5";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
219 DFA_NEXT(big5, c);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
220 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
221 if (DFA_ALIVE(utf8)) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
222 if (!DFA_ALIVE(big5))
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
223 return "UTF-8";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
224 DFA_NEXT(utf8, c);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
225 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
226
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
227 if (!DFA_ALIVE(big5) && !DFA_ALIVE(utf8)) {
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
228 /* we ran out the possibilities */
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
229 return NULL;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
230 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
231 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
232
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
233 /* Now, we have ambigous code. Pick the highest score. If more than
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
234 one candidate tie, pick the default encoding. */
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
235 for (i = 0; order[i] != NULL; i++) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
236 if (order[i]->state >= 0) { //DFA_ALIVE()
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
237 if (top == NULL || order[i]->score > top->score)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
238 top = order[i];
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
239 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
240 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
241
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
242 if (top == &big5)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
243 return "BIG5";
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
244 if (top == &utf8)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
245 return "UTF-8";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
246 return NULL;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
247 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
248
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
249 const char *guess_cn(const char *buf, int buflen)
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
250 {
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
251 int i;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
252 guess_dfa gb2312 = DFA_INIT(guess_gb2312_st, guess_gb2312_ar);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
253 guess_dfa utf8 = DFA_INIT(guess_utf8_st, guess_utf8_ar);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
254 guess_dfa gb18030 = DFA_INIT(guess_gb18030_st, guess_gb18030_ar);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
255 guess_dfa *top = NULL;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
256
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
257 guess_dfa *order[] = { ORDER_CN, NULL };
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
258
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
259 for (i = 0; i < buflen; i++) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
260 int c = (unsigned char) buf[i];
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
261 int c2;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
262
1595
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
263 /* special treatment of iso-2022 escape sequence */
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
264 if (c == 0x1b) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
265 if (i < buflen - 1) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
266 c = (unsigned char) buf[i + 1];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
267 c2 = (unsigned char) buf[i + 2];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
268 if (c == '$' && (c2 == ')' || c2 == '+'))
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
269 return "ISO-2022-CN";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
270 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
271 }
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
272
1595
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
273 /* special treatment of BOM */
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
274 if (i == 0 && c == 0xff) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
275 if (i < buflen - 1) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
276 c = (unsigned char) buf[i + 1];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
277 if (c == 0xfe)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
278 return UCS_2LE;
1595
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
279 }
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
280 }
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
281 if (i == 0 && c == 0xfe) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
282 if (i < buflen - 1) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
283 c = (unsigned char) buf[i + 1];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
284 if (c == 0xff)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
285 return UCS_2BE;
1595
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
286 }
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
287 }
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
288
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
289 if (DFA_ALIVE(gb2312)) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
290 if (!DFA_ALIVE(utf8) && !DFA_ALIVE(gb18030))
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
291 return "GB2312";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
292 DFA_NEXT(gb2312, c);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
293 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
294 if (DFA_ALIVE(utf8)) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
295 if (!DFA_ALIVE(gb2312) && !DFA_ALIVE(gb18030))
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
296 return "UTF-8";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
297 DFA_NEXT(utf8, c);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
298 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
299 if (DFA_ALIVE(gb18030)) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
300 if (!DFA_ALIVE(utf8) && !DFA_ALIVE(gb2312))
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
301 return "GB18030";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
302 DFA_NEXT(gb18030, c);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
303 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
304
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
305 if (!DFA_ALIVE(gb2312) && !DFA_ALIVE(utf8) && !DFA_ALIVE(gb18030)) {
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
306 /* we ran out the possibilities */
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
307 return NULL;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
308 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
309 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
310
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
311 /* Now, we have ambigous code. Pick the highest score. If more than
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
312 one candidate tie, pick the default encoding. */
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
313 for (i = 0; order[i] != NULL; i++) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
314 if (order[i]->state >= 0) { //DFA_ALIVE()
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
315 if (top == NULL || order[i]->score > top->score)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
316 top = order[i];
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
317 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
318 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
319
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
320 if (top == &gb2312)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
321 return "GB2312";
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
322 if (top == &utf8)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
323 return "UTF-8";
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
324 if (top == &gb18030)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
325 return "GB18030";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
326 return NULL;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
327 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
328
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
329 const char *guess_kr(const char *buf, int buflen)
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
330 {
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
331 int i;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
332 guess_dfa euck = DFA_INIT(guess_euck_st, guess_euck_ar);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
333 guess_dfa utf8 = DFA_INIT(guess_utf8_st, guess_utf8_ar);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
334 guess_dfa johab = DFA_INIT(guess_johab_st, guess_johab_ar);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
335 guess_dfa *top = NULL;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
336
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
337 guess_dfa *order[] = { ORDER_KR, NULL };
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
338
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
339 for (i = 0; i < buflen; i++) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
340 int c = (unsigned char) buf[i];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
341 int c2;
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
342
1595
e6cc84e06444 [svn] - libguess update: workaround for that g_convert can't convert properly from UCS-2BE/LE trailing after BOM.
yaz
parents: 1105
diff changeset
343 /* special treatment of iso-2022 escape sequence */
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
344 if (c == 0x1b) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
345 if (i < buflen - 1) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
346 c = (unsigned char) buf[i + 1];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
347 c2 = (unsigned char) buf[i + 2];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
348 if (c == '$' && c2 == ')')
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
349 return "ISO-2022-KR";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
350 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
351 }
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
352
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
353 /* special treatment of BOM */
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
354 if (i == 0 && c == 0xff) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
355 if (i < buflen - 1) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
356 c = (unsigned char) buf[i + 1];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
357 if (c == 0xfe)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
358 return UCS_2LE;
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
359 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
360 }
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
361 if (i == 0 && c == 0xfe) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
362 if (i < buflen - 1) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
363 c = (unsigned char) buf[i + 1];
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
364 if (c == 0xff)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
365 return UCS_2BE;
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
366 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
367 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
368
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
369 if (DFA_ALIVE(euck)) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
370 if (!DFA_ALIVE(johab) && !DFA_ALIVE(utf8))
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
371 return "EUC-KR";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
372 DFA_NEXT(euck, c);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
373 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
374 if (DFA_ALIVE(johab)) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
375 if (!DFA_ALIVE(euck) && !DFA_ALIVE(utf8))
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
376 return "JOHAB";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
377 DFA_NEXT(johab, c);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
378 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
379 if (DFA_ALIVE(utf8)) {
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
380 if (!DFA_ALIVE(euck) && !DFA_ALIVE(johab))
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
381 return "UTF-8";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
382 DFA_NEXT(utf8, c);
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
383 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
384
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
385 if (!DFA_ALIVE(euck) && !DFA_ALIVE(johab) && !DFA_ALIVE(utf8)) {
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
386 /* we ran out the possibilities */
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
387 return NULL;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
388 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
389 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
390
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
391 /* Now, we have ambigous code. Pick the highest score. If more than
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
392 one candidate tie, pick the default encoding. */
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
393 for (i = 0; order[i] != NULL; i++) {
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
394 if (order[i]->state >= 0) { //DFA_ALIVE()
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
395 if (top == NULL || order[i]->score > top->score)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
396 top = order[i];
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
397 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
398 }
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
399
2297
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
400 if (top == &euck)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
401 return "EUC-KR";
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
402 if (top == &utf8)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
403 return "UTF-8";
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
404 if (top == &johab)
a9bc621d6b1b [svn] libguess update:
yaz
parents: 1595
diff changeset
405 return "JOHAB";
1105
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
406 return NULL;
4be4d74db123 [svn] automatic character encoding detector for id3 metadata. --enable-chardet enables this feature.
yaz
parents:
diff changeset
407 }