| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
|
|
2
|
|
|
|
|
|
|
/* $Id: sjis_imode2.c 5246 2008-01-17 08:47:46Z hio $ */ |
|
3
|
|
|
|
|
|
|
|
|
4
|
|
|
|
|
|
|
#include "Japanese.h" |
|
5
|
|
|
|
|
|
|
#include |
|
6
|
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
#ifndef __cplusplus |
|
8
|
|
|
|
|
|
|
#undef bool |
|
9
|
|
|
|
|
|
|
#undef true |
|
10
|
|
|
|
|
|
|
#undef false |
|
11
|
|
|
|
|
|
|
typedef enum bool { false, true, } bool; |
|
12
|
|
|
|
|
|
|
#endif |
|
13
|
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
#define DISP_U2S 0 |
|
15
|
|
|
|
|
|
|
#define DISP_S2U 0 |
|
16
|
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
#if DISP_U2S |
|
18
|
|
|
|
|
|
|
#define ECHO_U2S(arg) fprintf arg |
|
19
|
|
|
|
|
|
|
#define ON_U2S(cmd) cmd |
|
20
|
|
|
|
|
|
|
#else |
|
21
|
|
|
|
|
|
|
#define ECHO_U2S(arg) |
|
22
|
|
|
|
|
|
|
#define ON_U2S(cmd) |
|
23
|
|
|
|
|
|
|
#endif |
|
24
|
|
|
|
|
|
|
#if DISP_S2U |
|
25
|
|
|
|
|
|
|
#define ECHO_S2U(arg) fprintf arg |
|
26
|
|
|
|
|
|
|
#define ON_S2U(cmd) cmd |
|
27
|
|
|
|
|
|
|
#else |
|
28
|
|
|
|
|
|
|
#define ECHO_S2U(arg) |
|
29
|
|
|
|
|
|
|
#define ON_S2U(cmd) |
|
30
|
|
|
|
|
|
|
#endif |
|
31
|
|
|
|
|
|
|
|
|
32
|
|
|
|
|
|
|
/* ---------------------------------------------------------------------------- |
|
33
|
|
|
|
|
|
|
* SV* sv_utf8 = xs_sjis_imode2_utf8(SV* sv_sjis) |
|
34
|
|
|
|
|
|
|
* convert sjis(imode2) into utf8. |
|
35
|
|
|
|
|
|
|
* ------------------------------------------------------------------------- */ |
|
36
|
|
|
|
|
|
|
EXTERN_C |
|
37
|
|
|
|
|
|
|
SV* |
|
38
|
262
|
|
|
|
|
|
xs_sjis_imode2_utf8(SV* sv_str) |
|
39
|
|
|
|
|
|
|
{ |
|
40
|
|
|
|
|
|
|
UJ_UINT8* src; |
|
41
|
|
|
|
|
|
|
STRLEN len; |
|
42
|
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
SV_Buf result; |
|
44
|
|
|
|
|
|
|
const UJ_UINT8* src_end; |
|
45
|
|
|
|
|
|
|
|
|
46
|
262
|
50
|
|
|
|
|
if( sv_str==&PL_sv_undef ) |
|
47
|
|
|
|
|
|
|
{ |
|
48
|
0
|
|
|
|
|
|
return newSVsv(&PL_sv_undef); |
|
49
|
|
|
|
|
|
|
} |
|
50
|
262
|
50
|
|
|
|
|
if( SvGMAGICAL(sv_str) ) |
|
51
|
|
|
|
|
|
|
{ |
|
52
|
0
|
|
|
|
|
|
mg_get(sv_str); |
|
53
|
|
|
|
|
|
|
} |
|
54
|
262
|
50
|
|
|
|
|
if( !SvOK(sv_str) ) |
|
55
|
|
|
|
|
|
|
{ |
|
56
|
0
|
|
|
|
|
|
return newSVsv(&PL_sv_undef); |
|
57
|
|
|
|
|
|
|
} |
|
58
|
|
|
|
|
|
|
|
|
59
|
262
|
|
|
|
|
|
src = (UJ_UINT8*)SvPV(sv_str, len); |
|
60
|
|
|
|
|
|
|
#if DISP_S2U |
|
61
|
|
|
|
|
|
|
fprintf(stderr,"Unicode::Japanese::(xs)sjis_utf8_imode2\n",len); |
|
62
|
|
|
|
|
|
|
bin_dump("in ",src,len); |
|
63
|
|
|
|
|
|
|
#endif |
|
64
|
262
|
50
|
|
|
|
|
SV_Buf_init(&result,len*3/2+4); |
|
|
|
50
|
|
|
|
|
|
|
65
|
262
|
|
|
|
|
|
src_end = src+len; |
|
66
|
|
|
|
|
|
|
|
|
67
|
524
|
100
|
|
|
|
|
while( src
|
|
68
|
|
|
|
|
|
|
{ |
|
69
|
|
|
|
|
|
|
const UJ_UINT8* ptr; |
|
70
|
262
|
50
|
|
|
|
|
if( src[0]<0x80 ) |
|
71
|
|
|
|
|
|
|
{ /* ASCII */ |
|
72
|
|
|
|
|
|
|
ECHO_U2S((stderr,"ascii: %02x\n",src[0])); |
|
73
|
0
|
0
|
|
|
|
|
if( src[0]=='&' && src+3
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
74
|
|
|
|
|
|
|
{ /* check "ddddd;" */ |
|
75
|
0
|
|
|
|
|
|
int num = 0; |
|
76
|
0
|
|
|
|
|
|
UJ_UINT8* ptr = src+2; |
|
77
|
0
|
|
|
|
|
|
const UJ_UINT8* ptr_end = ptr+8
|
|
78
|
0
|
0
|
|
|
|
|
for( ; ptr
|
|
79
|
|
|
|
|
|
|
{ |
|
80
|
0
|
0
|
|
|
|
|
if( *ptr==';' ) break; |
|
81
|
0
|
0
|
|
|
|
|
if( *ptr<'0' || '9'<*ptr ) break; |
|
|
|
0
|
|
|
|
|
|
|
82
|
0
|
|
|
|
|
|
num = num*10 + *ptr-'0'; |
|
83
|
|
|
|
|
|
|
} |
|
84
|
0
|
0
|
|
|
|
|
if( ptr
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
85
|
|
|
|
|
|
|
{ /* yes, this is "ddddd;" */ |
|
86
|
0
|
|
|
|
|
|
const UJ_UINT8* emoji = (UJ_UINT8*)&g_ei2u2_table[num&0x1ff]; |
|
87
|
0
|
0
|
|
|
|
|
if( emoji[3] ) |
|
88
|
|
|
|
|
|
|
{ |
|
89
|
|
|
|
|
|
|
/*fprintf(stderr,"utf8-len: [%d]\n",4); */ |
|
90
|
0
|
0
|
|
|
|
|
SV_Buf_append_mem(&result, emoji, 4); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
91
|
0
|
|
|
|
|
|
src = ptr+1; |
|
92
|
0
|
|
|
|
|
|
continue; |
|
93
|
|
|
|
|
|
|
} |
|
94
|
|
|
|
|
|
|
} |
|
95
|
|
|
|
|
|
|
} |
|
96
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,*src); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
97
|
0
|
|
|
|
|
|
++src; |
|
98
|
0
|
|
|
|
|
|
continue; |
|
99
|
262
|
100
|
|
|
|
|
}else if( 0xa1<=src[0] && src[0]<=0xdf ) |
|
|
|
50
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
{ /* half-width katakana (ja:半角カナ) */ |
|
101
|
|
|
|
|
|
|
ECHO_U2S((stderr,"kana: %02x\n",src[0])); |
|
102
|
0
|
|
|
|
|
|
ptr = (UJ_UINT8*)&g_s2u_table[(src[0]-0xa1)*3]; |
|
103
|
0
|
|
|
|
|
|
++src; |
|
104
|
262
|
50
|
|
|
|
|
}else if( src+1
|
|
|
|
50
|
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
105
|
1
|
|
|
|
|
|
{ /* a double-byte letter (ja:2バイト文字) */ |
|
106
|
1
|
|
|
|
|
|
const UJ_UINT16 sjis = (src[0]<<8)+src[1]; /* ntohs */ |
|
107
|
|
|
|
|
|
|
ECHO_U2S((stderr,"sjis.dbcs#1: %04x\n",sjis)); |
|
108
|
1
|
|
|
|
|
|
ptr = (UJ_UINT8*)&g_s2u_table[(sjis - 0x8100 + 0x3f)*3]; |
|
109
|
1
|
|
|
|
|
|
src += 2; |
|
110
|
261
|
50
|
|
|
|
|
}else if( src+1
|
|
|
|
100
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
111
|
0
|
|
|
|
|
|
{ /* i-mode emoji */ |
|
112
|
|
|
|
|
|
|
const UJ_UINT32* ptr32; |
|
113
|
|
|
|
|
|
|
ECHO_S2U((stderr,"code: %02x %02x\n", src[0],src[1])); |
|
114
|
261
|
|
|
|
|
|
ptr32 = &g_ei2u2_table[((src[0]&1)<<8)|src[1]]; |
|
115
|
261
|
50
|
|
|
|
|
if( ((char*)ptr32)[3]!=0 ) |
|
116
|
|
|
|
|
|
|
{ |
|
117
|
261
|
50
|
|
|
|
|
SV_Buf_append_ch4(&result, *ptr32); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
118
|
261
|
|
|
|
|
|
src += 2; |
|
119
|
261
|
|
|
|
|
|
continue; |
|
120
|
0
|
0
|
|
|
|
|
}else if( *ptr32 ) |
|
121
|
|
|
|
|
|
|
{ |
|
122
|
0
|
0
|
|
|
|
|
SV_Buf_append_mem(&result, ptr32, strlen((char*)ptr32)); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
123
|
0
|
|
|
|
|
|
src += 2; |
|
124
|
0
|
|
|
|
|
|
continue; |
|
125
|
|
|
|
|
|
|
}else |
|
126
|
|
|
|
|
|
|
{ |
|
127
|
0
|
|
|
|
|
|
const UJ_UINT16 sjis = (src[0]<<8)+src[1]; /* ntohs */ |
|
128
|
|
|
|
|
|
|
ECHO_U2S((stderr,"sjis.dbcs#2: %04x\n",sjis)); |
|
129
|
0
|
|
|
|
|
|
ptr = &g_s2u_table[(sjis- 0xe000 + 0x1f3f)*3]; |
|
130
|
0
|
|
|
|
|
|
src += 2; |
|
131
|
|
|
|
|
|
|
} |
|
132
|
0
|
0
|
|
|
|
|
}else if( src+1
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
133
|
0
|
|
|
|
|
|
{ /* a double-byte letter (ja:2バイト文字) */ |
|
134
|
0
|
|
|
|
|
|
const UJ_UINT16 sjis = ntohs(*(UJ_UINT16*)src); |
|
135
|
|
|
|
|
|
|
ECHO_U2S((stderr,"sjis.dbcs#2: %04x\n",sjis)); |
|
136
|
0
|
|
|
|
|
|
ptr = &g_s2u_table[(sjis- 0xe000 + 0x1f3f)*3]; |
|
137
|
0
|
|
|
|
|
|
src += 2; |
|
138
|
|
|
|
|
|
|
}else |
|
139
|
|
|
|
|
|
|
{ /* unknown */ |
|
140
|
|
|
|
|
|
|
/*fprintf(stderr,"unknown: %02x\n",src[0]); */ |
|
141
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
142
|
0
|
|
|
|
|
|
++src; |
|
143
|
0
|
|
|
|
|
|
continue; |
|
144
|
|
|
|
|
|
|
} |
|
145
|
|
|
|
|
|
|
|
|
146
|
|
|
|
|
|
|
ECHO_U2S((stderr,"offset: 0x%04x\n",ptr-g_s2u_table)); |
|
147
|
|
|
|
|
|
|
ECHO_U2S((stderr,"utf8-char : %02x %02x %02x\n",ptr[0],ptr[1],ptr[2])); |
|
148
|
1
|
50
|
|
|
|
|
if( ptr[2] ) |
|
149
|
|
|
|
|
|
|
{ |
|
150
|
|
|
|
|
|
|
/*fprintf(stderr,"utf8-len: [%d]\n",3); */ |
|
151
|
1
|
50
|
|
|
|
|
SV_Buf_append_mem(&result, ptr, 3); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
152
|
0
|
0
|
|
|
|
|
}else if( ptr[1] ) |
|
153
|
|
|
|
|
|
|
{ |
|
154
|
|
|
|
|
|
|
/*fprintf(stderr,"utf8-len: [%d]\n",2); */ |
|
155
|
0
|
0
|
|
|
|
|
SV_Buf_append_mem(&result, ptr, 2); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
156
|
0
|
0
|
|
|
|
|
}else if( ptr[0] ) |
|
157
|
|
|
|
|
|
|
{ |
|
158
|
|
|
|
|
|
|
/*fprintf(stderr,"utf8-len: [%d]\n",1); */ |
|
159
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,*ptr); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
160
|
|
|
|
|
|
|
}else |
|
161
|
|
|
|
|
|
|
{ |
|
162
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
163
|
|
|
|
|
|
|
} |
|
164
|
|
|
|
|
|
|
} |
|
165
|
|
|
|
|
|
|
#if DISP_S2U |
|
166
|
|
|
|
|
|
|
ON_S2U( bin_dump("out",SV_Buf_getBegin(&result),SV_Buf_getLength(&result)) ); |
|
167
|
|
|
|
|
|
|
#endif |
|
168
|
262
|
|
|
|
|
|
SV_Buf_setLength(&result); |
|
169
|
|
|
|
|
|
|
|
|
170
|
262
|
|
|
|
|
|
return SV_Buf_getSv(&result); |
|
171
|
|
|
|
|
|
|
} |
|
172
|
|
|
|
|
|
|
|
|
173
|
|
|
|
|
|
|
|
|
174
|
|
|
|
|
|
|
/* --------------------------------------------------------------------------- |
|
175
|
|
|
|
|
|
|
* utf8 ==> imode2 |
|
176
|
|
|
|
|
|
|
* ------------------------------------------------------------------------- */ |
|
177
|
|
|
|
|
|
|
EXTERN_C |
|
178
|
|
|
|
|
|
|
SV* |
|
179
|
35
|
|
|
|
|
|
xs_utf8_sjis_imode2(SV* sv_str) |
|
180
|
|
|
|
|
|
|
{ |
|
181
|
|
|
|
|
|
|
UJ_UINT8* src; |
|
182
|
|
|
|
|
|
|
STRLEN len; |
|
183
|
|
|
|
|
|
|
SV_Buf result; |
|
184
|
|
|
|
|
|
|
const UJ_UINT8* src_end; |
|
185
|
|
|
|
|
|
|
|
|
186
|
35
|
50
|
|
|
|
|
if( sv_str==&PL_sv_undef ) |
|
187
|
|
|
|
|
|
|
{ |
|
188
|
0
|
|
|
|
|
|
return newSVsv(&PL_sv_undef); |
|
189
|
|
|
|
|
|
|
} |
|
190
|
35
|
50
|
|
|
|
|
if( SvGMAGICAL(sv_str) ) |
|
191
|
|
|
|
|
|
|
{ |
|
192
|
0
|
|
|
|
|
|
mg_get(sv_str); |
|
193
|
|
|
|
|
|
|
} |
|
194
|
35
|
50
|
|
|
|
|
if( !SvOK(sv_str) ) |
|
195
|
|
|
|
|
|
|
{ |
|
196
|
0
|
|
|
|
|
|
return newSVsv(&PL_sv_undef); |
|
197
|
|
|
|
|
|
|
} |
|
198
|
35
|
|
|
|
|
|
src = (UJ_UINT8*)SvPV(sv_str, len); |
|
199
|
|
|
|
|
|
|
|
|
200
|
|
|
|
|
|
|
ECHO_U2S((stderr,"Unicode::Japanese::(xs)utf8_sjis_imode1\n")); |
|
201
|
|
|
|
|
|
|
ON_U2S( bin_dump("in ",src,len) ); |
|
202
|
|
|
|
|
|
|
|
|
203
|
35
|
50
|
|
|
|
|
SV_Buf_init(&result,len+4); |
|
|
|
50
|
|
|
|
|
|
|
204
|
35
|
|
|
|
|
|
src_end = src+len; |
|
205
|
|
|
|
|
|
|
|
|
206
|
72
|
100
|
|
|
|
|
while( src
|
|
207
|
|
|
|
|
|
|
{ |
|
208
|
|
|
|
|
|
|
UJ_UINT32 ucs; |
|
209
|
|
|
|
|
|
|
const UJ_UINT8* sjis_ptr; |
|
210
|
|
|
|
|
|
|
|
|
211
|
37
|
50
|
|
|
|
|
if( *src<=0x7f ) |
|
212
|
0
|
|
|
|
|
|
{ |
|
213
|
|
|
|
|
|
|
/* ascii chars sequence (ja:ASCIIはまとめて追加〜) */ |
|
214
|
0
|
|
|
|
|
|
int len = 1; |
|
215
|
0
|
0
|
|
|
|
|
while( src+len
|
|
|
|
0
|
|
|
|
|
|
|
216
|
|
|
|
|
|
|
{ |
|
217
|
0
|
|
|
|
|
|
++len; |
|
218
|
|
|
|
|
|
|
} |
|
219
|
0
|
0
|
|
|
|
|
SV_Buf_append_mem(&result,src,len); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
220
|
0
|
|
|
|
|
|
src+=len; |
|
221
|
0
|
|
|
|
|
|
continue; |
|
222
|
|
|
|
|
|
|
} |
|
223
|
|
|
|
|
|
|
|
|
224
|
|
|
|
|
|
|
/* non-ascii */ |
|
225
|
37
|
50
|
|
|
|
|
if( 0xe0<=*src && *src<=0xef ) |
|
|
|
100
|
|
|
|
|
|
|
226
|
2
|
|
|
|
|
|
{ /* 3byte range. mostly enter here. */ |
|
227
|
2
|
|
|
|
|
|
const int utf8_len = 3; |
|
228
|
2
|
|
|
|
|
|
const UJ_UINT32 ucs_min = 0x800; |
|
229
|
2
|
|
|
|
|
|
const UJ_UINT32 ucs_max = 0xffff; |
|
230
|
|
|
|
|
|
|
ECHO_U2S((stderr,"utf8-len: [%d]\n",utf8_len)); |
|
231
|
|
|
|
|
|
|
/* check length */ |
|
232
|
2
|
50
|
|
|
|
|
if( src+utf8_len<=src_end ) |
|
233
|
|
|
|
|
|
|
{ /* noop */ |
|
234
|
|
|
|
|
|
|
}else |
|
235
|
|
|
|
|
|
|
{ /* no enough sequence */ |
|
236
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
237
|
0
|
|
|
|
|
|
++src; |
|
238
|
0
|
|
|
|
|
|
continue; |
|
239
|
|
|
|
|
|
|
} |
|
240
|
|
|
|
|
|
|
/* check follow sequences */ |
|
241
|
2
|
50
|
|
|
|
|
if( 0x80<=src[1] && src[1]<=0xbf && 0x80<=src[2] && src[2]<=0xbf ) |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
242
|
|
|
|
|
|
|
{ /* noop */ |
|
243
|
|
|
|
|
|
|
}else |
|
244
|
|
|
|
|
|
|
{ |
|
245
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
246
|
0
|
|
|
|
|
|
++src; |
|
247
|
0
|
|
|
|
|
|
continue; |
|
248
|
|
|
|
|
|
|
} |
|
249
|
|
|
|
|
|
|
|
|
250
|
|
|
|
|
|
|
/* compute code point */ |
|
251
|
2
|
|
|
|
|
|
ucs = ((src[0] & 0x0F)<<12)|((src[1] & 0x3F)<<6)|(src[2] & 0x3F); |
|
252
|
2
|
|
|
|
|
|
src += utf8_len; |
|
253
|
2
|
50
|
|
|
|
|
if( ucs_min<=ucs && ucs<=ucs_max ) |
|
|
|
50
|
|
|
|
|
|
|
254
|
|
|
|
|
|
|
{ /* noop */ |
|
255
|
|
|
|
|
|
|
}else |
|
256
|
|
|
|
|
|
|
{ /* illegal sequence */ |
|
257
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
258
|
0
|
|
|
|
|
|
continue; |
|
259
|
|
|
|
|
|
|
} |
|
260
|
|
|
|
|
|
|
/* ok. */ |
|
261
|
35
|
50
|
|
|
|
|
}else if( 0xf0<=*src && *src<=0xf7 ) |
|
|
|
50
|
|
|
|
|
|
|
262
|
0
|
|
|
|
|
|
{ |
|
263
|
35
|
|
|
|
|
|
const int utf8_len = 4; |
|
264
|
35
|
|
|
|
|
|
const UJ_UINT32 ucs_min = 0x010000; |
|
265
|
35
|
|
|
|
|
|
const UJ_UINT32 ucs_max = 0x10ffff; |
|
266
|
|
|
|
|
|
|
ECHO_U2S((stderr,"utf8-len: [%d]\n",utf8_len)); |
|
267
|
|
|
|
|
|
|
/* check length */ |
|
268
|
35
|
50
|
|
|
|
|
if( src+utf8_len<=src_end ) |
|
269
|
|
|
|
|
|
|
{ /* noop */ |
|
270
|
|
|
|
|
|
|
}else |
|
271
|
|
|
|
|
|
|
{ /* no enough sequence */ |
|
272
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
273
|
0
|
|
|
|
|
|
++src; |
|
274
|
0
|
|
|
|
|
|
continue; |
|
275
|
|
|
|
|
|
|
} |
|
276
|
|
|
|
|
|
|
/* check follow sequences */ |
|
277
|
35
|
50
|
|
|
|
|
if( 0x80<=src[1] && src[1]<=0xbf && 0x80<=src[2] && src[2]<=0xbf |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
278
|
35
|
50
|
|
|
|
|
&& 0x80<=src[3] && src[3]<=0xbf ) |
|
|
|
50
|
|
|
|
|
|
|
279
|
|
|
|
|
|
|
{ /* noop */ |
|
280
|
|
|
|
|
|
|
}else |
|
281
|
|
|
|
|
|
|
{ |
|
282
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
283
|
0
|
|
|
|
|
|
++src; |
|
284
|
0
|
|
|
|
|
|
continue; |
|
285
|
|
|
|
|
|
|
} |
|
286
|
|
|
|
|
|
|
|
|
287
|
|
|
|
|
|
|
/* compute code point */ |
|
288
|
35
|
|
|
|
|
|
ucs = ((src[0] & 0x07)<<18)|((src[1] & 0x3F)<<12)| |
|
289
|
35
|
|
|
|
|
|
((src[2] & 0x3f) << 6)|(src[3] & 0x3F); |
|
290
|
35
|
|
|
|
|
|
src += utf8_len; |
|
291
|
35
|
50
|
|
|
|
|
if( ucs_min<=ucs && ucs<=ucs_max ) |
|
|
|
50
|
|
|
|
|
|
|
292
|
|
|
|
|
|
|
{ /* noop */ |
|
293
|
|
|
|
|
|
|
}else |
|
294
|
|
|
|
|
|
|
{ /* illegal sequence */ |
|
295
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
296
|
0
|
|
|
|
|
|
continue; |
|
297
|
|
|
|
|
|
|
} |
|
298
|
|
|
|
|
|
|
/* private area: block emoji */ |
|
299
|
35
|
50
|
|
|
|
|
if( 0x0f0000<=ucs && ucs<=0x0fffff ) |
|
|
|
50
|
|
|
|
|
|
|
300
|
35
|
|
|
|
|
|
{ |
|
301
|
|
|
|
|
|
|
const UJ_UINT16* sjis16; |
|
302
|
|
|
|
|
|
|
const UJ_UINT8* sjis8; |
|
303
|
35
|
50
|
|
|
|
|
if( ucs<0x0fe000 ) |
|
304
|
|
|
|
|
|
|
{ /* unknown area. */ |
|
305
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
306
|
0
|
|
|
|
|
|
continue; |
|
307
|
|
|
|
|
|
|
} |
|
308
|
|
|
|
|
|
|
/* imode */ |
|
309
|
35
|
|
|
|
|
|
sjis16 = &g_eu2i2_table[ucs - 0x0fe000]; |
|
310
|
35
|
|
|
|
|
|
sjis8 = (UJ_UINT8*)sjis16; |
|
311
|
35
|
50
|
|
|
|
|
if( sjis8[1]!=0 ) |
|
312
|
|
|
|
|
|
|
{ /* double-byte char */ |
|
313
|
35
|
50
|
|
|
|
|
SV_Buf_append_ch2(&result, *sjis16); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
314
|
0
|
0
|
|
|
|
|
}else if( sjis8[0]!=0 ) |
|
315
|
|
|
|
|
|
|
{ /* single-byte char, is it exists?? */ |
|
316
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result, *sjis8); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
317
|
|
|
|
|
|
|
}else |
|
318
|
|
|
|
|
|
|
{ /* no mapping */ |
|
319
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
320
|
|
|
|
|
|
|
} |
|
321
|
35
|
|
|
|
|
|
continue; |
|
322
|
|
|
|
|
|
|
} |
|
323
|
|
|
|
|
|
|
|
|
324
|
|
|
|
|
|
|
/* > U+10FFFF not supported by UTF-8 (RFC 3629). */ |
|
325
|
0
|
0
|
|
|
|
|
if( ucs>0x10FFFF ) |
|
326
|
|
|
|
|
|
|
{ |
|
327
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
328
|
0
|
|
|
|
|
|
continue; |
|
329
|
|
|
|
|
|
|
} |
|
330
|
0
|
0
|
|
|
|
|
}else if( 0xc0<=*src && *src<=0xdf ) |
|
|
|
0
|
|
|
|
|
|
|
331
|
0
|
|
|
|
|
|
{ |
|
332
|
0
|
|
|
|
|
|
const int utf8_len = 2; |
|
333
|
0
|
|
|
|
|
|
const UJ_UINT32 ucs_min = 0x80; |
|
334
|
0
|
|
|
|
|
|
const UJ_UINT32 ucs_max = 0x7ff; |
|
335
|
|
|
|
|
|
|
ECHO_U2S((stderr,"utf8-len: [%d]\n",utf8_len)); |
|
336
|
|
|
|
|
|
|
/* check length */ |
|
337
|
0
|
0
|
|
|
|
|
if( src+utf8_len<=src_end ) |
|
338
|
|
|
|
|
|
|
{ /* noop */ |
|
339
|
|
|
|
|
|
|
}else |
|
340
|
|
|
|
|
|
|
{ /* no enough sequence */ |
|
341
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
342
|
0
|
|
|
|
|
|
++src; |
|
343
|
0
|
|
|
|
|
|
continue; |
|
344
|
|
|
|
|
|
|
} |
|
345
|
|
|
|
|
|
|
/* check follow sequences */ |
|
346
|
0
|
0
|
|
|
|
|
if( 0x80<=src[1] && src[1]<=0xbf ) |
|
|
|
0
|
|
|
|
|
|
|
347
|
|
|
|
|
|
|
{ /* noop */ |
|
348
|
|
|
|
|
|
|
}else |
|
349
|
|
|
|
|
|
|
{ |
|
350
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
351
|
0
|
|
|
|
|
|
++src; |
|
352
|
0
|
|
|
|
|
|
continue; |
|
353
|
|
|
|
|
|
|
} |
|
354
|
|
|
|
|
|
|
|
|
355
|
|
|
|
|
|
|
/* compute code point */ |
|
356
|
0
|
|
|
|
|
|
ucs = ((src[0] & 0x1F)<<6)|(src[1] & 0x3F); |
|
357
|
0
|
|
|
|
|
|
src += utf8_len; |
|
358
|
0
|
0
|
|
|
|
|
if( ucs_min<=ucs && ucs<=ucs_max ) |
|
|
|
0
|
|
|
|
|
|
|
359
|
|
|
|
|
|
|
{ /* noop */ |
|
360
|
|
|
|
|
|
|
}else |
|
361
|
|
|
|
|
|
|
{ /* illegal sequence */ |
|
362
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
363
|
0
|
|
|
|
|
|
continue; |
|
364
|
|
|
|
|
|
|
} |
|
365
|
|
|
|
|
|
|
|
|
366
|
|
|
|
|
|
|
/* ok. */ |
|
367
|
0
|
0
|
|
|
|
|
}else if( 0xf8<=*src && *src<=0xfb ) |
|
|
|
0
|
|
|
|
|
|
|
368
|
0
|
|
|
|
|
|
{ |
|
369
|
0
|
|
|
|
|
|
const int utf8_len = 5; |
|
370
|
|
|
|
|
|
|
ECHO_U2S((stderr,"utf8-len: [%d]\n",utf8_len)); |
|
371
|
|
|
|
|
|
|
/* check length */ |
|
372
|
0
|
0
|
|
|
|
|
if( src+utf8_len<=src_end ) |
|
373
|
|
|
|
|
|
|
{ /* noop */ |
|
374
|
|
|
|
|
|
|
}else |
|
375
|
|
|
|
|
|
|
{ /* no enough sequence */ |
|
376
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
377
|
0
|
|
|
|
|
|
++src; |
|
378
|
0
|
|
|
|
|
|
continue; |
|
379
|
|
|
|
|
|
|
} |
|
380
|
|
|
|
|
|
|
/* check follow sequences */ |
|
381
|
0
|
0
|
|
|
|
|
if( 0x80<=src[1] && src[1]<=0xbf && 0x80<=src[2] && src[2]<=0xbf |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
382
|
0
|
0
|
|
|
|
|
&& 0x80<=src[3] && src[3]<=0xbf && 0x80<=src[4] && src[4]<=0xbf ) |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
383
|
|
|
|
|
|
|
{ /* noop */ |
|
384
|
|
|
|
|
|
|
}else |
|
385
|
|
|
|
|
|
|
{ |
|
386
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
387
|
0
|
|
|
|
|
|
++src; |
|
388
|
0
|
|
|
|
|
|
continue; |
|
389
|
|
|
|
|
|
|
} |
|
390
|
|
|
|
|
|
|
|
|
391
|
|
|
|
|
|
|
/* compute code point */ |
|
392
|
|
|
|
|
|
|
/* > U+10FFFF not supported by UTF-8 (RFC 3629). */ |
|
393
|
0
|
|
|
|
|
|
src += utf8_len; |
|
394
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
395
|
0
|
|
|
|
|
|
continue; |
|
396
|
0
|
0
|
|
|
|
|
}else if( 0xfc<=*src && *src<=0xfd ) |
|
|
|
0
|
|
|
|
|
|
|
397
|
0
|
|
|
|
|
|
{ |
|
398
|
0
|
|
|
|
|
|
const int utf8_len = 6; |
|
399
|
|
|
|
|
|
|
ECHO_U2S((stderr,"utf8-len: [%d]\n",utf8_len)); |
|
400
|
|
|
|
|
|
|
/* check length */ |
|
401
|
0
|
0
|
|
|
|
|
if( src+utf8_len<=src_end ) |
|
402
|
|
|
|
|
|
|
{ /* noop */ |
|
403
|
|
|
|
|
|
|
}else |
|
404
|
|
|
|
|
|
|
{ /* no enough sequence */ |
|
405
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
406
|
0
|
|
|
|
|
|
++src; |
|
407
|
0
|
|
|
|
|
|
continue; |
|
408
|
|
|
|
|
|
|
} |
|
409
|
|
|
|
|
|
|
/* check follow sequences */ |
|
410
|
0
|
0
|
|
|
|
|
if( 0x80<=src[1] && src[1]<=0xbf && 0x80<=src[2] && src[2]<=0xbf |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
411
|
0
|
0
|
|
|
|
|
&& 0x80<=src[3] && src[3]<=0xbf && 0x80<=src[4] && src[4]<=0xbf |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
412
|
0
|
0
|
|
|
|
|
&& 0x80<=src[5] && src[5]<=0xbf ) |
|
|
|
0
|
|
|
|
|
|
|
413
|
|
|
|
|
|
|
{ /* noop */ |
|
414
|
|
|
|
|
|
|
}else |
|
415
|
|
|
|
|
|
|
{ |
|
416
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
417
|
0
|
|
|
|
|
|
++src; |
|
418
|
0
|
|
|
|
|
|
continue; |
|
419
|
|
|
|
|
|
|
} |
|
420
|
|
|
|
|
|
|
|
|
421
|
|
|
|
|
|
|
/* compute code point */ |
|
422
|
|
|
|
|
|
|
/* > U+10FFFF not supported by UTF-8 (RFC 3629). */ |
|
423
|
0
|
|
|
|
|
|
src += utf8_len; |
|
424
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
425
|
0
|
|
|
|
|
|
continue; |
|
426
|
|
|
|
|
|
|
}else |
|
427
|
|
|
|
|
|
|
{ |
|
428
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
429
|
0
|
|
|
|
|
|
++src; |
|
430
|
0
|
|
|
|
|
|
continue; |
|
431
|
|
|
|
|
|
|
} |
|
432
|
|
|
|
|
|
|
|
|
433
|
|
|
|
|
|
|
/* ucs => sjis */ |
|
434
|
|
|
|
|
|
|
ECHO_U2S((stderr,"ucs [%04x]\n",ucs)); |
|
435
|
2
|
50
|
|
|
|
|
if( ucs<=0x9FFF ) |
|
436
|
|
|
|
|
|
|
{ |
|
437
|
2
|
|
|
|
|
|
sjis_ptr = g_u2s_table + ucs*2; |
|
438
|
0
|
0
|
|
|
|
|
}else if( 0xF900<=ucs && ucs<=0xFFFF ) |
|
|
|
0
|
|
|
|
|
|
|
439
|
|
|
|
|
|
|
{ |
|
440
|
0
|
|
|
|
|
|
sjis_ptr = g_u2s_table + (ucs - 0xF900 + 0xA000)*2; |
|
441
|
0
|
0
|
|
|
|
|
}else if( 0x0FE000<=ucs && ucs<=0x0FFFFF ) |
|
|
|
0
|
|
|
|
|
|
|
442
|
|
|
|
|
|
|
{ |
|
443
|
0
|
|
|
|
|
|
sjis_ptr = (UJ_UINT8*)"?"; /* exactly 2byte: "?\0" */ |
|
444
|
|
|
|
|
|
|
}else |
|
445
|
|
|
|
|
|
|
{ |
|
446
|
0
|
|
|
|
|
|
sjis_ptr = (UJ_UINT8*)"\0"; /* exactly 2byte: "\0\0" */ |
|
447
|
|
|
|
|
|
|
} |
|
448
|
2
|
100
|
|
|
|
|
if( sjis_ptr[0]!=0 || sjis_ptr[1]!=0 ) |
|
|
|
50
|
|
|
|
|
|
|
449
|
|
|
|
|
|
|
{ /* mapping dest exists. */ |
|
450
|
1
|
50
|
|
|
|
|
if( sjis_ptr[1]!=0 ) |
|
451
|
|
|
|
|
|
|
{ |
|
452
|
1
|
50
|
|
|
|
|
SV_Buf_append_mem(&result, sjis_ptr, 2); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
453
|
|
|
|
|
|
|
}else |
|
454
|
|
|
|
|
|
|
{ |
|
455
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,sjis_ptr[0]); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
456
|
|
|
|
|
|
|
} |
|
457
|
1
|
50
|
|
|
|
|
}else if( ucs<=0x7F ) |
|
458
|
|
|
|
|
|
|
{ |
|
459
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,(UJ_UINT8)ucs); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
460
|
|
|
|
|
|
|
}else |
|
461
|
|
|
|
|
|
|
{ |
|
462
|
1
|
50
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
463
|
|
|
|
|
|
|
} |
|
464
|
|
|
|
|
|
|
} /* while */ |
|
465
|
|
|
|
|
|
|
|
|
466
|
|
|
|
|
|
|
ON_U2S( bin_dump("out",SV_Buf_getBegin(&result),SV_Buf_getLength(&result)) ); |
|
467
|
35
|
|
|
|
|
|
SV_Buf_setLength(&result); |
|
468
|
|
|
|
|
|
|
|
|
469
|
35
|
|
|
|
|
|
return SV_Buf_getSv(&result); |
|
470
|
|
|
|
|
|
|
} |
|
471
|
|
|
|
|
|
|
|
|
472
|
|
|
|
|
|
|
/* ---------------------------------------------------------------------------- |
|
473
|
|
|
|
|
|
|
* End of File. |
|
474
|
|
|
|
|
|
|
* ------------------------------------------------------------------------- */ |
|
475
|
|
|
|
|
|
|
|