| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
/* ---------------------------------------------------------------------------- |
|
2
|
|
|
|
|
|
|
* utf16_utf8.c |
|
3
|
|
|
|
|
|
|
* ---------------------------------------------------------------------------- |
|
4
|
|
|
|
|
|
|
* Mastering programed by YAMASHINA Hio |
|
5
|
|
|
|
|
|
|
* ---------------------------------------------------------------------------- |
|
6
|
|
|
|
|
|
|
* $Id: utf16.c 4654 2006-07-03 01:33:16Z hio $ |
|
7
|
|
|
|
|
|
|
* ------------------------------------------------------------------------- */ |
|
8
|
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
#include "Japanese.h" |
|
11
|
|
|
|
|
|
|
|
|
12
|
|
|
|
|
|
|
#define ENABLE_SURROGATE_PAIR 1 |
|
13
|
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
/* ---------------------------------------------------------------------------- |
|
15
|
|
|
|
|
|
|
* convert utf-16 into utf-8 |
|
16
|
|
|
|
|
|
|
* ------------------------------------------------------------------------- */ |
|
17
|
|
|
|
|
|
|
EXTERN_C |
|
18
|
|
|
|
|
|
|
SV* |
|
19
|
5
|
|
|
|
|
|
xs_utf16_utf8(SV* sv_str) |
|
20
|
|
|
|
|
|
|
{ |
|
21
|
|
|
|
|
|
|
UJ_UINT8* src; |
|
22
|
|
|
|
|
|
|
STRLEN len; |
|
23
|
|
|
|
|
|
|
SV_Buf result; |
|
24
|
|
|
|
|
|
|
const UJ_UINT8* src_end; |
|
25
|
|
|
|
|
|
|
union { |
|
26
|
|
|
|
|
|
|
UJ_UINT32 u32_val; |
|
27
|
|
|
|
|
|
|
UJ_UINT16 u16_val; |
|
28
|
|
|
|
|
|
|
UJ_UINT8 u8_val[4]; |
|
29
|
|
|
|
|
|
|
} buf; |
|
30
|
|
|
|
|
|
|
|
|
31
|
5
|
50
|
|
|
|
|
if( sv_str==&PL_sv_undef ) |
|
32
|
|
|
|
|
|
|
{ |
|
33
|
0
|
|
|
|
|
|
return newSVpvn("",0); |
|
34
|
|
|
|
|
|
|
} |
|
35
|
5
|
50
|
|
|
|
|
if( SvGMAGICAL(sv_str) ) |
|
36
|
|
|
|
|
|
|
{ |
|
37
|
0
|
|
|
|
|
|
mg_get(sv_str); |
|
38
|
|
|
|
|
|
|
} |
|
39
|
5
|
50
|
|
|
|
|
if( !SvOK(sv_str) ) |
|
40
|
|
|
|
|
|
|
{ |
|
41
|
0
|
|
|
|
|
|
return newSVpvn("",0); |
|
42
|
|
|
|
|
|
|
} |
|
43
|
|
|
|
|
|
|
|
|
44
|
5
|
|
|
|
|
|
src = (UJ_UINT8*)SvPV(sv_str, len); |
|
45
|
5
|
|
|
|
|
|
src_end = src+(len&~1); |
|
46
|
|
|
|
|
|
|
/*fprintf(stderr,"Unicode::Japanese::(xs)utf16_utf8\n",len);*/ |
|
47
|
|
|
|
|
|
|
/*bin_dump("in ",src,len);*/ |
|
48
|
5
|
50
|
|
|
|
|
SV_Buf_init(&result,len*3/2+4); |
|
|
|
50
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
|
|
50
|
5
|
50
|
|
|
|
|
if( len&1 ) |
|
51
|
|
|
|
|
|
|
{ |
|
52
|
0
|
|
|
|
|
|
Perl_croak(aTHX_ "Unicode::Japanese::utf16_utf8, invalid length (not 2*n)"); |
|
53
|
|
|
|
|
|
|
} |
|
54
|
|
|
|
|
|
|
|
|
55
|
10
|
100
|
|
|
|
|
for(; src
|
|
56
|
|
|
|
|
|
|
{ |
|
57
|
5
|
|
|
|
|
|
const UJ_UINT16 utf16 = (src[0]<<8)+src[1]; /* ntohs */ |
|
58
|
5
|
50
|
|
|
|
|
if( utf16<0x80 ) |
|
59
|
|
|
|
|
|
|
{ |
|
60
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,(UJ_UINT8)utf16); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
61
|
5
|
50
|
|
|
|
|
}else if( utf16<0x800 ) |
|
62
|
|
|
|
|
|
|
{ |
|
63
|
0
|
|
|
|
|
|
buf.u8_val[0] = 0xC0 | (utf16 >> 6); |
|
64
|
0
|
|
|
|
|
|
buf.u8_val[1] = 0x80 | (utf16 & 0x3F); |
|
65
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch2(&result, buf.u16_val); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
66
|
5
|
50
|
|
|
|
|
}else if( !(0xd800 <= utf16 && utf16 <= 0xdfff) ) |
|
|
|
50
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
{ /* normal char (non surrogate pair) */ |
|
68
|
0
|
|
|
|
|
|
buf.u8_val[0] = 0xE0 | (utf16 >> 12); |
|
69
|
0
|
|
|
|
|
|
buf.u8_val[1] = 0x80 | ((utf16 >> 6) & 0x3F); |
|
70
|
0
|
|
|
|
|
|
buf.u8_val[2] = 0x80 | (utf16 & 0x3F); |
|
71
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch3(&result, buf.u32_val); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
72
|
|
|
|
|
|
|
}else |
|
73
|
|
|
|
|
|
|
{ /* surrogate pair */ |
|
74
|
5
|
50
|
|
|
|
|
if( src+2
|
|
75
|
|
|
|
|
|
|
{ |
|
76
|
5
|
|
|
|
|
|
const UJ_UINT16 utf16a = (src[2]<<8)+src[3]; /* ntohs */ |
|
77
|
5
|
50
|
|
|
|
|
if( utf16<=0xdbff && 0xdc00 <= utf16a && utf16a <= 0xdfff ) |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
78
|
5
|
|
|
|
|
|
{ |
|
79
|
|
|
|
|
|
|
#if ENABLE_SURROGATE_PAIR |
|
80
|
5
|
|
|
|
|
|
const UJ_UINT32 ucs4 = ((utf16&0x03FF)<<10|(utf16a&0x03FF))+0x010000; |
|
81
|
5
|
|
|
|
|
|
src += 2; |
|
82
|
5
|
50
|
|
|
|
|
if( 0x010000<=ucs4 && ucs4<=0x10FFFF ) |
|
|
|
50
|
|
|
|
|
|
|
83
|
|
|
|
|
|
|
{ |
|
84
|
5
|
|
|
|
|
|
buf.u8_val[0] = 0xF0 | ((ucs4>>18) & 0x3F); |
|
85
|
5
|
|
|
|
|
|
buf.u8_val[1] = 0x80 | ((ucs4>>12) & 0x3F); |
|
86
|
5
|
|
|
|
|
|
buf.u8_val[2] = 0x80 | ((ucs4>>6) & 0x3F); |
|
87
|
5
|
|
|
|
|
|
buf.u8_val[3] = 0x80 | (ucs4 & 0x3F); |
|
88
|
5
|
50
|
|
|
|
|
SV_Buf_append_ch4(&result, buf.u32_val); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
89
|
|
|
|
|
|
|
}else |
|
90
|
|
|
|
|
|
|
{ |
|
91
|
|
|
|
|
|
|
/* utf8 not support >= U+10FFFF */ |
|
92
|
|
|
|
|
|
|
/* or illegal representation */ |
|
93
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
94
|
|
|
|
|
|
|
} |
|
95
|
|
|
|
|
|
|
#else |
|
96
|
|
|
|
|
|
|
{ |
|
97
|
|
|
|
|
|
|
/* surrogate pair disabled. */ |
|
98
|
|
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
99
|
|
|
|
|
|
|
} |
|
100
|
|
|
|
|
|
|
#endif |
|
101
|
|
|
|
|
|
|
}else |
|
102
|
|
|
|
|
|
|
{ |
|
103
|
|
|
|
|
|
|
/* invalid surrogate */ |
|
104
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
105
|
|
|
|
|
|
|
} |
|
106
|
|
|
|
|
|
|
}else |
|
107
|
|
|
|
|
|
|
{ |
|
108
|
|
|
|
|
|
|
/* no trail surrogate */ |
|
109
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch(&result,'?'); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
110
|
|
|
|
|
|
|
} |
|
111
|
|
|
|
|
|
|
} |
|
112
|
|
|
|
|
|
|
} |
|
113
|
|
|
|
|
|
|
|
|
114
|
|
|
|
|
|
|
/*bin_dump("out",result.getBegin(),result.getLength()); */ |
|
115
|
5
|
|
|
|
|
|
SV_Buf_setLength(&result); |
|
116
|
|
|
|
|
|
|
|
|
117
|
5
|
|
|
|
|
|
return SV_Buf_getSv(&result); |
|
118
|
|
|
|
|
|
|
} |
|
119
|
|
|
|
|
|
|
|
|
120
|
|
|
|
|
|
|
/* ---------------------------------------------------------------------------- |
|
121
|
|
|
|
|
|
|
* convert utf-8 into utf-16 |
|
122
|
|
|
|
|
|
|
* ------------------------------------------------------------------------- */ |
|
123
|
|
|
|
|
|
|
EXTERN_C |
|
124
|
|
|
|
|
|
|
SV* |
|
125
|
6
|
|
|
|
|
|
xs_utf8_utf16(SV* sv_str) |
|
126
|
|
|
|
|
|
|
{ |
|
127
|
|
|
|
|
|
|
UJ_UINT8* src; |
|
128
|
|
|
|
|
|
|
STRLEN len; |
|
129
|
|
|
|
|
|
|
SV_Buf result; |
|
130
|
|
|
|
|
|
|
const UJ_UINT8* src_end; |
|
131
|
|
|
|
|
|
|
|
|
132
|
6
|
50
|
|
|
|
|
if( sv_str==&PL_sv_undef ) |
|
133
|
|
|
|
|
|
|
{ |
|
134
|
0
|
|
|
|
|
|
return newSVpvn("",0); |
|
135
|
|
|
|
|
|
|
} |
|
136
|
6
|
50
|
|
|
|
|
if( SvGMAGICAL(sv_str) ) |
|
137
|
|
|
|
|
|
|
{ |
|
138
|
0
|
|
|
|
|
|
mg_get(sv_str); |
|
139
|
|
|
|
|
|
|
} |
|
140
|
6
|
50
|
|
|
|
|
if( !SvOK(sv_str) ) |
|
141
|
|
|
|
|
|
|
{ |
|
142
|
0
|
|
|
|
|
|
return newSVpvn("",0); |
|
143
|
|
|
|
|
|
|
} |
|
144
|
|
|
|
|
|
|
|
|
145
|
6
|
|
|
|
|
|
src = (UJ_UINT8*)SvPV(sv_str, len); |
|
146
|
6
|
|
|
|
|
|
src_end = src+len; |
|
147
|
|
|
|
|
|
|
/*fprintf(stderr,"Unicode::Japanese::(xs)utf8_utf16\n",len); */ |
|
148
|
|
|
|
|
|
|
/*bin_dump("in ",src,len); */ |
|
149
|
6
|
50
|
|
|
|
|
SV_Buf_init(&result,len*2); |
|
|
|
50
|
|
|
|
|
|
|
150
|
|
|
|
|
|
|
|
|
151
|
12
|
100
|
|
|
|
|
while( src
|
|
152
|
|
|
|
|
|
|
{ |
|
153
|
|
|
|
|
|
|
UJ_UINT32 ucs; |
|
154
|
6
|
100
|
|
|
|
|
if( *src<=0x7f ) |
|
155
|
|
|
|
|
|
|
{ |
|
156
|
1
|
50
|
|
|
|
|
SV_Buf_append_ch2(&result,htons(*src)); |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
157
|
1
|
|
|
|
|
|
++src; |
|
158
|
1
|
|
|
|
|
|
continue; |
|
159
|
|
|
|
|
|
|
} |
|
160
|
5
|
50
|
|
|
|
|
if( 0xc0<=*src && *src<=0xdf ) |
|
|
|
100
|
|
|
|
|
|
|
161
|
0
|
|
|
|
|
|
{ /* length [2] */ |
|
162
|
1
|
|
|
|
|
|
const int utf8_len = 2; |
|
163
|
1
|
|
|
|
|
|
const UJ_UINT32 ucs_min = 0x80; |
|
164
|
1
|
|
|
|
|
|
const UJ_UINT32 ucs_max = 0x7ff; |
|
165
|
1
|
50
|
|
|
|
|
if( src+1>=src_end || |
|
166
|
1
|
50
|
|
|
|
|
src[1]<0x80 || 0xbf
|
|
|
|
50
|
|
|
|
|
|
|
167
|
|
|
|
|
|
|
{ |
|
168
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch2(&result,htons(*src)); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
169
|
0
|
|
|
|
|
|
++src; |
|
170
|
0
|
|
|
|
|
|
continue; |
|
171
|
|
|
|
|
|
|
} |
|
172
|
|
|
|
|
|
|
|
|
173
|
|
|
|
|
|
|
/* compute code point */ |
|
174
|
1
|
|
|
|
|
|
ucs = ((src[0] & 0x1F)<<6)|(src[1] & 0x3F); |
|
175
|
1
|
|
|
|
|
|
src += utf8_len; |
|
176
|
1
|
50
|
|
|
|
|
if( ucs_min<=ucs && ucs<=ucs_max ) |
|
|
|
0
|
|
|
|
|
|
|
177
|
|
|
|
|
|
|
{ /* noop */ |
|
178
|
|
|
|
|
|
|
}else |
|
179
|
|
|
|
|
|
|
{ /* illegal sequence */ |
|
180
|
1
|
50
|
|
|
|
|
SV_Buf_append_ch2(&result,htons('?')); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
181
|
1
|
|
|
|
|
|
continue; |
|
182
|
|
|
|
|
|
|
} |
|
183
|
|
|
|
|
|
|
|
|
184
|
|
|
|
|
|
|
/* ok. */ |
|
185
|
4
|
50
|
|
|
|
|
}else if( 0xe0<=*src && *src<=0xef ) |
|
|
|
100
|
|
|
|
|
|
|
186
|
0
|
|
|
|
|
|
{ /* length [3] */ |
|
187
|
1
|
|
|
|
|
|
const int utf8_len = 3; |
|
188
|
1
|
|
|
|
|
|
const UJ_UINT32 ucs_min = 0x800; |
|
189
|
1
|
|
|
|
|
|
const UJ_UINT32 ucs_max = 0xffff; |
|
190
|
1
|
50
|
|
|
|
|
if( src+2>=src_end || |
|
191
|
1
|
50
|
|
|
|
|
src[1]<0x80 || 0xbf
|
|
|
|
50
|
|
|
|
|
|
|
192
|
1
|
50
|
|
|
|
|
src[2]<0x80 || 0xbf
|
|
|
|
50
|
|
|
|
|
|
|
193
|
|
|
|
|
|
|
{ |
|
194
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch2(&result,htons(*src)); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
195
|
0
|
|
|
|
|
|
++src; |
|
196
|
0
|
|
|
|
|
|
continue; |
|
197
|
|
|
|
|
|
|
} |
|
198
|
|
|
|
|
|
|
|
|
199
|
|
|
|
|
|
|
/* compute code point */ |
|
200
|
1
|
|
|
|
|
|
ucs = ((src[0] & 0x0F)<<12)|((src[1] & 0x3F)<<6)|(src[2] & 0x3F); |
|
201
|
1
|
|
|
|
|
|
src += utf8_len; |
|
202
|
1
|
50
|
|
|
|
|
if( ucs_min<=ucs && ucs<=ucs_max ) |
|
|
|
0
|
|
|
|
|
|
|
203
|
|
|
|
|
|
|
{ /* noop */ |
|
204
|
|
|
|
|
|
|
}else |
|
205
|
|
|
|
|
|
|
{ /* illegal sequence */ |
|
206
|
1
|
50
|
|
|
|
|
SV_Buf_append_ch2(&result,htons('?')); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
207
|
1
|
|
|
|
|
|
continue; |
|
208
|
|
|
|
|
|
|
} |
|
209
|
|
|
|
|
|
|
|
|
210
|
0
|
0
|
|
|
|
|
if( ucs<0xD800 || ucs>0xDBFF ) |
|
|
|
0
|
|
|
|
|
|
|
211
|
|
|
|
|
|
|
{ /* normal char, noop */ |
|
212
|
|
|
|
|
|
|
}else |
|
213
|
|
|
|
|
|
|
{ /* delete surrogate pair range */ |
|
214
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch2(&result,htons('?')); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
215
|
0
|
|
|
|
|
|
continue; |
|
216
|
|
|
|
|
|
|
} |
|
217
|
|
|
|
|
|
|
|
|
218
|
|
|
|
|
|
|
/* ok. */ |
|
219
|
3
|
50
|
|
|
|
|
}else if( 0xf0<=*src && *src<=0xf7 ) |
|
|
|
100
|
|
|
|
|
|
|
220
|
|
|
|
|
|
|
{ /* length [4] */ |
|
221
|
1
|
|
|
|
|
|
const int utf8_len = 4; |
|
222
|
1
|
|
|
|
|
|
const UJ_UINT32 ucs_min = 0x010000; |
|
223
|
1
|
|
|
|
|
|
const UJ_UINT32 ucs_max = 0x10ffff; |
|
224
|
1
|
50
|
|
|
|
|
if( src+3>=src_end || |
|
225
|
1
|
50
|
|
|
|
|
src[1]<0x80 || 0xbf
|
|
|
|
50
|
|
|
|
|
|
|
226
|
1
|
50
|
|
|
|
|
src[2]<0x80 || 0xbf
|
|
|
|
50
|
|
|
|
|
|
|
227
|
1
|
50
|
|
|
|
|
src[3]<0x80 || 0xbf
|
|
|
|
50
|
|
|
|
|
|
|
228
|
|
|
|
|
|
|
{ |
|
229
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch2(&result,htons(*src)); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
230
|
0
|
|
|
|
|
|
++src; |
|
231
|
0
|
|
|
|
|
|
continue; |
|
232
|
|
|
|
|
|
|
} |
|
233
|
|
|
|
|
|
|
|
|
234
|
|
|
|
|
|
|
/* compute code point */ |
|
235
|
1
|
|
|
|
|
|
ucs = ((src[0] & 0x07)<<18)|((src[1] & 0x3F)<<12)| |
|
236
|
1
|
|
|
|
|
|
((src[2] & 0x3f) << 6)|(src[3] & 0x3F); |
|
237
|
1
|
|
|
|
|
|
src += utf8_len; |
|
238
|
1
|
50
|
|
|
|
|
if( ucs_min<=ucs && ucs<=ucs_max ) |
|
|
|
0
|
|
|
|
|
|
|
239
|
|
|
|
|
|
|
{ /* noop */ |
|
240
|
|
|
|
|
|
|
}else |
|
241
|
|
|
|
|
|
|
{ /* illegal sequence */ |
|
242
|
1
|
50
|
|
|
|
|
SV_Buf_append_ch2(&result,htons('?')); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
243
|
1
|
|
|
|
|
|
continue; |
|
244
|
|
|
|
|
|
|
} |
|
245
|
|
|
|
|
|
|
|
|
246
|
|
|
|
|
|
|
#if ENABLE_SURROGATE_PAIR |
|
247
|
0
|
|
|
|
|
|
{ /* encode surrogate pair */ |
|
248
|
0
|
|
|
|
|
|
const UJ_UINT32 surrogate = ucs - 0x010000; |
|
249
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch2(&result,htons(((surrogate>>10)&0x03FF)|0xD800)); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
250
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch2(&result,htons(((surrogate )&0x03FF)|0xDC00)); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
251
|
0
|
|
|
|
|
|
continue; |
|
252
|
|
|
|
|
|
|
} |
|
253
|
|
|
|
|
|
|
#else |
|
254
|
|
|
|
|
|
|
{ /* not supported */ |
|
255
|
|
|
|
|
|
|
SV_Buf_append_ch2(&result,htons('?')); |
|
256
|
|
|
|
|
|
|
continue; |
|
257
|
|
|
|
|
|
|
} |
|
258
|
|
|
|
|
|
|
#endif |
|
259
|
|
|
|
|
|
|
|
|
260
|
|
|
|
|
|
|
/* ok. */ |
|
261
|
2
|
50
|
|
|
|
|
}else if( 0xf8<=*src && *src<=0xfb ) |
|
|
|
100
|
|
|
|
|
|
|
262
|
1
|
|
|
|
|
|
{ |
|
263
|
1
|
|
|
|
|
|
const int utf8_len = 5; |
|
264
|
1
|
50
|
|
|
|
|
if( src+utf8_len<=src_end ) |
|
265
|
|
|
|
|
|
|
{ /* noop */ |
|
266
|
|
|
|
|
|
|
}else |
|
267
|
|
|
|
|
|
|
{ /* no enough sequence */ |
|
268
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch2(&result,htons('?')); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
269
|
0
|
|
|
|
|
|
++src; |
|
270
|
0
|
|
|
|
|
|
continue; |
|
271
|
|
|
|
|
|
|
} |
|
272
|
|
|
|
|
|
|
/* check follow sequences */ |
|
273
|
1
|
50
|
|
|
|
|
if( 0x80<=src[1] && src[1]<=0xbf && 0x80<=src[2] && src[2]<=0xbf |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
274
|
1
|
50
|
|
|
|
|
&& 0x80<=src[3] && src[3]<=0xbf && 0x80<=src[4] && src[4]<=0xbf ) |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
275
|
|
|
|
|
|
|
{ /* noop */ |
|
276
|
|
|
|
|
|
|
}else |
|
277
|
|
|
|
|
|
|
{ |
|
278
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch2(&result,htons('?')); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
279
|
0
|
|
|
|
|
|
++src; |
|
280
|
0
|
|
|
|
|
|
continue; |
|
281
|
|
|
|
|
|
|
} |
|
282
|
|
|
|
|
|
|
|
|
283
|
|
|
|
|
|
|
/* compute code point */ |
|
284
|
1
|
|
|
|
|
|
src += utf8_len; |
|
285
|
1
|
50
|
|
|
|
|
SV_Buf_append_ch2(&result,htons('?')); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
286
|
1
|
|
|
|
|
|
continue; |
|
287
|
1
|
50
|
|
|
|
|
}else if( 0xfc<=*src && *src<=0xfd ) |
|
|
|
50
|
|
|
|
|
|
|
288
|
1
|
|
|
|
|
|
{ |
|
289
|
1
|
|
|
|
|
|
const int utf8_len = 6; |
|
290
|
1
|
50
|
|
|
|
|
if( src+utf8_len<=src_end ) |
|
291
|
|
|
|
|
|
|
{ /* noop */ |
|
292
|
|
|
|
|
|
|
}else |
|
293
|
|
|
|
|
|
|
{ /* no enough sequence */ |
|
294
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch2(&result,htons('?')); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
295
|
0
|
|
|
|
|
|
++src; |
|
296
|
0
|
|
|
|
|
|
continue; |
|
297
|
|
|
|
|
|
|
} |
|
298
|
|
|
|
|
|
|
/* check follow sequences */ |
|
299
|
1
|
50
|
|
|
|
|
if( 0x80<=src[1] && src[1]<=0xbf && 0x80<=src[2] && src[2]<=0xbf |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
300
|
1
|
50
|
|
|
|
|
&& 0x80<=src[3] && src[3]<=0xbf && 0x80<=src[4] && src[4]<=0xbf |
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
301
|
1
|
50
|
|
|
|
|
&& 0x80<=src[5] && src[5]<=0xbf ) |
|
|
|
50
|
|
|
|
|
|
|
302
|
|
|
|
|
|
|
{ /* noop */ |
|
303
|
|
|
|
|
|
|
}else |
|
304
|
|
|
|
|
|
|
{ |
|
305
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch2(&result,htons('?')); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
306
|
0
|
|
|
|
|
|
++src; |
|
307
|
0
|
|
|
|
|
|
continue; |
|
308
|
|
|
|
|
|
|
} |
|
309
|
|
|
|
|
|
|
|
|
310
|
|
|
|
|
|
|
/* compute code point */ |
|
311
|
1
|
|
|
|
|
|
src += utf8_len; |
|
312
|
1
|
50
|
|
|
|
|
SV_Buf_append_ch2(&result,htons('?')); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
313
|
1
|
|
|
|
|
|
continue; |
|
314
|
|
|
|
|
|
|
}else |
|
315
|
|
|
|
|
|
|
{ /* invalid */ |
|
316
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch2(&result,htons(*src)); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
317
|
0
|
|
|
|
|
|
++src; |
|
318
|
0
|
|
|
|
|
|
continue; |
|
319
|
|
|
|
|
|
|
} |
|
320
|
|
|
|
|
|
|
|
|
321
|
0
|
0
|
|
|
|
|
if( ucs & ~0xFFFF ) |
|
322
|
|
|
|
|
|
|
{ /* utf16¤ÎÈϰϳ° (ucs4¤ÎÈϰÏ) */ |
|
323
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch2(&result,htons('?')); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
324
|
0
|
|
|
|
|
|
continue; |
|
325
|
|
|
|
|
|
|
} |
|
326
|
0
|
0
|
|
|
|
|
SV_Buf_append_ch2(&result,htons(ucs)); |
|
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
327
|
|
|
|
|
|
|
/*bin_dump("now",dst_begin,dst-dst_begin); */ |
|
328
|
|
|
|
|
|
|
} |
|
329
|
|
|
|
|
|
|
|
|
330
|
|
|
|
|
|
|
/*bin_dump("out",result.getBegin(),result.getLength()); */ |
|
331
|
6
|
|
|
|
|
|
SV_Buf_setLength(&result); |
|
332
|
|
|
|
|
|
|
|
|
333
|
6
|
|
|
|
|
|
return SV_Buf_getSv(&result); |
|
334
|
|
|
|
|
|
|
} |
|
335
|
|
|
|
|
|
|
|
|
336
|
|
|
|
|
|
|
/* ---------------------------------------------------------------------------- |
|
337
|
|
|
|
|
|
|
* End Of File. |
|
338
|
|
|
|
|
|
|
* ------------------------------------------------------------------------- */ |