Rev 7044 | Rev 7057 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed
Rev | Author | Line No. | Line |
---|---|---|---|
7045 | ashmew2 | 1 | #include |
5098 | clevermous | 2 | //#include |
3 | typedef unsigned int size_t; |
||
4 | #define NULL ((void*)0) |
||
3479 | leency | 5 | |
6 | typedef int conv_t; |
||
7 | typedef unsigned int ucs4_t; |
||
8 | |||
9 | typedef int iconv_t; |
||
10 | |||
11 | |||
12 | /* Return code if invalid input after a shift sequence of n bytes was read. |
||
13 | (xxx_mbtowc) */ |
||
14 | #define RET_SHIFT_ILSEQ(n) (-1-2*(n)) |
||
15 | /* Return code if invalid. (xxx_mbtowc) */ |
||
16 | #define RET_ILSEQ RET_SHIFT_ILSEQ(0) |
||
17 | /* Return code if only a shift sequence of n bytes was read. (xxx_mbtowc) */ |
||
18 | #define RET_TOOFEW(n) (-2-2*(n)) |
||
19 | |||
20 | /* Return code if invalid. (xxx_wctomb) */ |
||
21 | #define RET_ILUNI -1 |
||
22 | /* Return code if output buffer is too small. (xxx_wctomb, xxx_reset) */ |
||
23 | #define RET_TOOSMALL -2 |
||
24 | |||
25 | |||
26 | #define CP866 0 |
||
27 | #define CP1251 1 |
||
28 | #define CP1252 2 |
||
29 | #define KOI8_RU 3 |
||
30 | #define ISO8859_5 4 |
||
31 | #define UTF_8 5 |
||
32 | |||
33 | #include "cp866.h" |
||
34 | #include "cp1251.h" |
||
35 | #include "cp1252.h" |
||
36 | #include "koi8_ru.h" |
||
37 | #include "iso8859_5.h" |
||
38 | #include "utf8.h" |
||
39 | |||
7005 | ashmew2 | 40 | int encoding(char *what) { |
6892 | ashmew2 | 41 | |
7005 | ashmew2 | 42 | /* Ignore //TRANSLIT or //IGNORE for now. */ |
43 | int i; |
||
44 | for(i = 0; i < strlen(what); i++) { |
||
45 | if(what[i] == '/') { |
||
46 | what[i] = '\0'; |
||
47 | break; |
||
48 | } |
||
49 | } |
||
50 | |||
7045 | ashmew2 | 51 | if (!strcasecmp(what,"CP866")) return CP866; |
52 | if (!strcasecmp(what,"CP1251")) return CP1251; |
||
53 | if (!strcasecmp(what,"windows-1252")) return CP1252; |
||
54 | if (!strcasecmp(what,"CP1252")) return CP1252; |
||
55 | if (!strcasecmp(what,"KOI8-RU")) return KOI8_RU; |
||
56 | if (!strcasecmp(what,"ISO8859-5")) return ISO8859_5; |
||
57 | if (!strcasecmp(what,"UTF-8")) return UTF_8; |
||
3479 | leency | 58 | return -1; |
59 | } |
||
60 | |||
61 | |||
62 | iconv_t iconv_open(const char *tocode, const char *fromcode) { |
||
63 | int to, from; |
||
7044 | ashmew2 | 64 | |
3479 | leency | 65 | if ((to=encoding(tocode))==-1) return -1; |
66 | if ((from=encoding(fromcode))==-1) return -1; |
||
7044 | ashmew2 | 67 | |
3479 | leency | 68 | to=to<<16&0xFFFF0000; |
69 | from=from&0xFFFF; |
||
70 | return to+from; |
||
71 | } |
||
72 | |||
6882 | ashmew2 | 73 | int iconv_close(iconv_t icd) |
74 | { |
||
75 | return 0; |
||
76 | } |
||
77 | |||
3479 | leency | 78 | size_t iconv(iconv_t cd, const char **inbuf, size_t *inbytesleft, |
79 | char **outbuf, size_t *outbytesleft) |
||
80 | { |
||
81 | int n, to, from, count1,count2; |
||
82 | int pwc, converted,written; |
||
83 | int (*mbtowc)(conv_t, ucs4_t *, const unsigned char *, int); |
||
84 | int (*wctomb)(conv_t, ucs4_t *, const unsigned char *, int); |
||
85 | |||
86 | char *str; |
||
87 | str=*outbuf; |
||
88 | |||
89 | from=cd>>16; |
||
90 | to=cd&0xFFFF; |
||
91 | |||
92 | switch (from) |
||
93 | { |
||
94 | case CP866: mbtowc=cp866_mbtowc; break; |
||
95 | case CP1251: mbtowc=cp1251_mbtowc; break; |
||
96 | case CP1252: mbtowc=cp1252_mbtowc; break; |
||
97 | case ISO8859_5: mbtowc=iso8859_5_mbtowc; break; |
||
98 | case KOI8_RU: mbtowc=koi8_ru_mbtowc; break; |
||
99 | case UTF_8: mbtowc=utf8_mbtowc; break; |
||
100 | default: return -2; |
||
101 | } |
||
102 | |||
103 | switch (to) |
||
104 | { |
||
105 | case CP866: wctomb=cp866_wctomb; break; |
||
106 | case CP1251: wctomb=cp1251_wctomb; break; |
||
107 | case CP1252: wctomb=cp1252_wctomb; break; |
||
108 | case ISO8859_5: wctomb=iso8859_5_wctomb; break; |
||
109 | case KOI8_RU: wctomb=koi8_ru_wctomb; break; |
||
110 | case UTF_8: wctomb=utf8_wctomb; break; |
||
111 | default: return -3; |
||
112 | } |
||
6894 | ashmew2 | 113 | |
114 | if(from == to) { |
||
115 | int oc=0,ic=0; |
||
116 | |||
117 | while(*inbytesleft > 0 && *outbytesleft > 0) { |
||
118 | str[oc]=(*inbuf)[ic]; |
||
119 | ++ic; |
||
120 | ++oc; |
||
121 | (*inbytesleft)--; |
||
122 | (*outbytesleft)--; |
||
7007 | ashmew2 | 123 | (*outbuf)++; |
6894 | ashmew2 | 124 | } |
125 | |||
126 | return 0; |
||
127 | } |
||
3479 | leency | 128 | |
129 | count1=0; |
||
130 | count2=0; |
||
131 | |||
132 | while ( *inbytesleft>0 && *outbytesleft>1) |
||
133 | { |
||
134 | n=1; |
||
135 | |||
136 | do { |
||
137 | //converted= (utf8_mbtowc)(0,&pwc,((*inbuf)+count1),n); |
||
138 | // printf("%d\n",n); |
||
139 | converted= (mbtowc)(0,&pwc,((*inbuf)+count1),n); |
||
140 | |||
141 | n++; |
||
142 | } while (converted==RET_TOOFEW(0)); |
||
143 | |||
144 | if (converted<0) return -10; |
||
145 | //written= (cp866_wctomb)(0,str+count2,pwc,1); |
||
146 | written= (wctomb)(0,str+count2,pwc,1); |
||
147 | if (written<0) written=0;//return -11; |
||
148 | |||
149 | //printf("Conv:%d Wri:%d In:%d Out:%d UTF:%x UCS:%x 866:%s\n",converted, written, *inbytesleft,*outbytesleft,*((*inbuf)+count1),pwc, str); |
||
150 | |||
151 | (*inbytesleft)-=converted; |
||
152 | (*outbytesleft)-=written; |
||
7044 | ashmew2 | 153 | (*outbuf)+=written; |
3479 | leency | 154 | count1+=converted; |
155 | count2+=written; |
||
156 | } |
||
157 | *(str+count2)='\0'; |
||
158 | |||
159 | if (*inbytesleft>0 && *outbytesleft==0) return -12; |
||
160 | return 0; |
||
161 | } |
||
162 | |||
163 | |||
164 | /* |
||
165 | int main() |
||
166 | { |
||
167 | char *s;// ="вертолет"; |
||
168 | char *z; |
||
169 | //unsigned int pwc; |
||
170 | iconv_t cd; |
||
171 | int in, out; |
||
172 | |||
173 | FILE *infile; |
||
174 | char *fname = "file.txt"; |
||
175 | |||
176 | infile = fopen(fname,"r"); |
||
177 | |||
178 | fseek(infile, 0, SEEK_END); |
||
179 | size_t file_size = ftell(infile); |
||
180 | rewind(infile); |
||
181 | |||
182 | //printf ("LOL\n"); |
||
183 | |||
184 | char *buffer = (char*)malloc(file_size * sizeof(char)); |
||
185 | if (buffer == NULL) |
||
186 | { |
||
187 | fclose(infile); |
||
188 | printf("Error allocating %d bytes.\n", file_size * sizeof(char)); |
||
189 | return -1; |
||
190 | } |
||
191 | size_t bytes_read = fread(buffer, sizeof(char), file_size, infile); |
||
192 | if (bytes_read != file_size) |
||
193 | { |
||
194 | printf("Have read only %d bytes of %d.\n", bytes_read, file_size); |
||
195 | free(buffer); |
||
196 | fclose(infile); |
||
197 | return -1; |
||
198 | } |
||
199 | |||
200 | in=strlen(buffer); |
||
201 | z=malloc(in+1); |
||
202 | |||
203 | out=in+1; |
||
204 | cd=iconv_open("CP1251","CP866"); |
||
205 | // printf("%x\n",cd); |
||
206 | int t; |
||
207 | t=iconv(cd, &buffer, &in, &z, &out); |
||
208 | printf("\nResult: %d", t); |
||
209 | puts(z); |
||
210 | //for (;s |
||
211 | } |
||
212 | */ |
||
213 | |||
214 | |||
215 | typedef struct |
||
216 | { |
||
217 | char *name; |
||
218 | void *f; |
||
219 | } export_t; |
||
220 | |||
221 | char szStart[] = "START"; |
||
222 | char szVersion[] = "version"; |
||
223 | char sziconv_open[] = "iconv_open"; |
||
224 | char sziconv[] = "iconv"; |
||
225 | |||
226 | export_t EXPORTS[] __asm__("EXPORTS") = |
||
227 | { |
||
228 | { szStart, (void*)0x0 }, |
||
229 | { szVersion, (void*)0x00010001 }, |
||
230 | { sziconv_open, iconv_open }, |
||
231 | { sziconv, iconv }, |
||
232 | { NULL, NULL }, |
||
233 | };0)>0)>16&0xFFFF0000; |