Rev 7057 | Rev 7084 | Go to most recent revision | Only display areas with differences | Regard whitespace | Details | Blame | Last modification | View Log | RSS feed
Rev 7057 | Rev 7079 | ||
---|---|---|---|
1 | #include |
1 | #include |
2 | #include |
2 | #include |
3 | #include |
3 | #include |
4 | 4 | ||
5 | typedef int conv_t; |
5 | typedef int conv_t; |
6 | typedef unsigned int ucs4_t; |
6 | typedef unsigned int ucs4_t; |
7 | typedef int iconv_t; |
7 | typedef int iconv_t; |
8 | 8 | ||
9 | /* Return code if invalid input after a shift sequence of n bytes was read. |
9 | /* Return code if invalid input after a shift sequence of n bytes was read. |
10 | (xxx_mbtowc) */ |
10 | (xxx_mbtowc) */ |
11 | #define RET_SHIFT_ILSEQ(n) (-1-2*(n)) |
11 | #define RET_SHIFT_ILSEQ(n) (-1-2*(n)) |
12 | /* Return code if invalid. (xxx_mbtowc) */ |
12 | /* Return code if invalid. (xxx_mbtowc) */ |
13 | #define RET_ILSEQ RET_SHIFT_ILSEQ(0) |
13 | #define RET_ILSEQ RET_SHIFT_ILSEQ(0) |
14 | /* Return code if only a shift sequence of n bytes was read. (xxx_mbtowc) */ |
14 | /* Return code if only a shift sequence of n bytes was read. (xxx_mbtowc) */ |
15 | #define RET_TOOFEW(n) (-2-2*(n)) |
15 | #define RET_TOOFEW(n) (-2-2*(n)) |
16 | 16 | ||
17 | /* Return code if invalid. (xxx_wctomb) */ |
17 | /* Return code if invalid. (xxx_wctomb) */ |
18 | #define RET_ILUNI -1 |
18 | #define RET_ILUNI -1 |
19 | /* Return code if output buffer is too small. (xxx_wctomb, xxx_reset) */ |
19 | /* Return code if output buffer is too small. (xxx_wctomb, xxx_reset) */ |
20 | #define RET_TOOSMALL -2 |
20 | #define RET_TOOSMALL -2 |
21 | 21 | ||
22 | #define CP866 0 |
22 | #define CP866 0 |
23 | #define CP1251 1 |
23 | #define CP1251 1 |
24 | #define CP1252 2 |
24 | #define CP1252 2 |
25 | #define KOI8_RU 3 |
25 | #define KOI8_RU 3 |
26 | #define ISO8859_5 4 |
26 | #define ISO8859_5 4 |
27 | #define UTF_8 5 |
27 | #define UTF_8 5 |
28 | 28 | ||
29 | #include "cp866.h" |
29 | #include "cp866.h" |
30 | #include "cp1251.h" |
30 | #include "cp1251.h" |
31 | #include "cp1252.h" |
31 | #include "cp1252.h" |
32 | #include "koi8_ru.h" |
32 | #include "koi8_ru.h" |
33 | #include "iso8859_5.h" |
33 | #include "iso8859_5.h" |
34 | #include "utf8.h" |
34 | #include "utf8.h" |
35 | 35 | ||
36 | int encoding(const char *someencoding) { |
36 | int encoding(const char *someencoding) { |
37 | 37 | ||
38 | char *what = strdup(someencoding); |
38 | char *what = strdup(someencoding); |
39 | /* Ignore //TRANSLIT or //IGNORE for now. */ |
39 | /* Ignore //TRANSLIT or //IGNORE for now. */ |
40 | int i; |
40 | int i; |
41 | for(i = 0; i < strlen(what); i++) { |
41 | for(i = 0; i < strlen(what); i++) { |
42 | if(what[i] == '/') { |
42 | if(what[i] == '/') { |
43 | what[i] = '\0'; |
43 | what[i] = '\0'; |
44 | break; |
44 | break; |
45 | } |
45 | } |
46 | } |
46 | } |
47 | 47 | ||
48 | if (!strcasecmp(what,"CP866")) return CP866; |
48 | if (!strcasecmp(what,"CP866")) return CP866; |
49 | if (!strcasecmp(what,"CP1251")) return CP1251; |
49 | if (!strcasecmp(what,"CP1251")) return CP1251; |
50 | if (!strcasecmp(what,"windows-1251")) return CP1251; |
50 | if (!strcasecmp(what,"windows-1251")) return CP1251; |
51 | if (!strcasecmp(what,"windows-1252")) return CP1252; |
51 | if (!strcasecmp(what,"windows-1252")) return CP1252; |
52 | if (!strcasecmp(what,"CP1252")) return CP1252; |
52 | if (!strcasecmp(what,"CP1252")) return CP1252; |
53 | if (!strcasecmp(what,"KOI8-RU")) return KOI8_RU; |
53 | if (!strcasecmp(what,"KOI8-RU")) return KOI8_RU; |
- | 54 | if (!strcasecmp(what,"KOI8-R")) return KOI8_RU; |
|
54 | if (!strcasecmp(what,"ISO8859-5")) return ISO8859_5; |
55 | if (!strcasecmp(what,"ISO8859-5")) return ISO8859_5; |
55 | if (!strcasecmp(what,"UTF-8")) return UTF_8; |
56 | if (!strcasecmp(what,"UTF-8")) return UTF_8; |
56 | return -1; |
57 | return -1; |
57 | } |
58 | } |
58 | 59 | ||
59 | iconv_t iconv_open(const char *tocode, const char *fromcode) { |
60 | iconv_t iconv_open(const char *tocode, const char *fromcode) { |
60 | int to, from; |
61 | int to, from; |
61 | 62 | ||
62 | if ((to=encoding(tocode))==-1) return -1; |
63 | if ((to=encoding(tocode))==-1) return -1; |
63 | if ((from=encoding(fromcode))==-1) return -1; |
64 | if ((from=encoding(fromcode))==-1) return -1; |
64 | 65 | ||
65 | to=to<<16&0xFFFF0000; |
66 | to=to<<16&0xFFFF0000; |
66 | from=from&0xFFFF; |
67 | from=from&0xFFFF; |
67 | 68 | ||
68 | return to+from; |
69 | return to+from; |
69 | } |
70 | } |
70 | 71 | ||
71 | int iconv_close(iconv_t icd) |
72 | int iconv_close(iconv_t icd) |
72 | { |
73 | { |
73 | return 0; |
74 | return 0; |
74 | } |
75 | } |
75 | 76 | ||
76 | size_t iconv(iconv_t cd, const char **inbuf, size_t *inbytesleft, |
77 | size_t iconv(iconv_t cd, const char **inbuf, size_t *inbytesleft, |
77 | char **outbuf, size_t *outbytesleft) |
78 | char **outbuf, size_t *outbytesleft) |
78 | { |
79 | { |
79 | int n, to, from; |
80 | int n, to, from; |
80 | size_t count1,count2; |
81 | size_t count1,count2; |
81 | unsigned int pwc; |
82 | unsigned int pwc; |
82 | int converted,written; |
83 | int converted,written; |
83 | int (*mbtowc)(conv_t, ucs4_t *, const unsigned char *, int); |
84 | int (*mbtowc)(conv_t, ucs4_t *, const unsigned char *, int); |
84 | int (*wctomb)(conv_t, unsigned char *, ucs4_t, int); |
85 | int (*wctomb)(conv_t, unsigned char *, ucs4_t, int); |
85 | 86 | ||
86 | to=cd>>16; |
87 | to=cd>>16; |
87 | from=cd&0xFFFF; |
88 | from=cd&0xFFFF; |
88 | 89 | ||
89 | switch (from) |
90 | switch (from) |
90 | { |
91 | { |
91 | case CP866: mbtowc=cp866_mbtowc; break; |
92 | case CP866: mbtowc=cp866_mbtowc; break; |
92 | case CP1251: mbtowc=cp1251_mbtowc; break; |
93 | case CP1251: mbtowc=cp1251_mbtowc; break; |
93 | case CP1252: mbtowc=cp1252_mbtowc; break; |
94 | case CP1252: mbtowc=cp1252_mbtowc; break; |
94 | case ISO8859_5: mbtowc=iso8859_5_mbtowc; break; |
95 | case ISO8859_5: mbtowc=iso8859_5_mbtowc; break; |
95 | case KOI8_RU: mbtowc=koi8_ru_mbtowc; break; |
96 | case KOI8_RU: mbtowc=koi8_ru_mbtowc; break; |
96 | case UTF_8: mbtowc=utf8_mbtowc; break; |
97 | case UTF_8: mbtowc=utf8_mbtowc; break; |
97 | default: return (size_t)-1; |
98 | default: return (size_t)-1; |
98 | } |
99 | } |
99 | 100 | ||
100 | switch (to) |
101 | switch (to) |
101 | { |
102 | { |
102 | case CP866: wctomb=cp866_wctomb; break; |
103 | case CP866: wctomb=cp866_wctomb; break; |
103 | case CP1251: wctomb=cp1251_wctomb; break; |
104 | case CP1251: wctomb=cp1251_wctomb; break; |
104 | case CP1252: wctomb=cp1252_wctomb; break; |
105 | case CP1252: wctomb=cp1252_wctomb; break; |
105 | case ISO8859_5: wctomb=iso8859_5_wctomb; break; |
106 | case ISO8859_5: wctomb=iso8859_5_wctomb; break; |
106 | case KOI8_RU: wctomb=koi8_ru_wctomb; break; |
107 | case KOI8_RU: wctomb=koi8_ru_wctomb; break; |
107 | case UTF_8: wctomb=utf8_wctomb; break; |
108 | case UTF_8: wctomb=utf8_wctomb; break; |
108 | default: return (size_t)-1; |
109 | default: return (size_t)-1; |
109 | } |
110 | } |
110 | 111 | ||
111 | count1=0; |
112 | count1=0; |
112 | count2=0; |
113 | count2=0; |
113 | 114 | ||
114 | /* Convert input multibyte char to wide character by using calls to mbtowc */ |
115 | /* Convert input multibyte char to wide character by using calls to mbtowc */ |
115 | /* Convert wide character to multibyte by calls to wctomb */ |
116 | /* Convert wide character to multibyte by calls to wctomb */ |
116 | /* Handle errors as we go on converting to be as standard compliant as possible */ |
117 | /* Handle errors as we go on converting to be as standard compliant as possible */ |
117 | while(count1 < *inbytesleft) { |
118 | while(count1 < *inbytesleft) { |
118 | unsigned char mbholder[] = { 0,0,0,0,0,0 }; |
119 | unsigned char mbholder[] = { 0,0,0,0,0,0 }; |
119 | 120 | ||
120 | int numbytes = (mbtowc)(0, &pwc,((*inbuf)+count1), *inbytesleft - count1); |
121 | int numbytes = (mbtowc)(0, &pwc,((*inbuf)+count1), *inbytesleft - count1); |
121 | if(numbytes < 0) { |
122 | if(numbytes < 0) { |
122 | /* errno = EILSEQ if invalid multibyte sequence encountered in input */ |
123 | /* errno = EILSEQ if invalid multibyte sequence encountered in input */ |
123 | /* errno = EINVAL if input ends in the middle of a multibyte sequence */ |
124 | /* errno = EINVAL if input ends in the middle of a multibyte sequence */ |
124 | 125 | ||
125 | switch(numbytes) { |
126 | switch(numbytes) { |
126 | case RET_TOOFEW(0): |
127 | case RET_TOOFEW(0): |
127 | errno = EINVAL; |
128 | errno = EINVAL; |
128 | break; |
129 | break; |
129 | 130 | ||
130 | case RET_ILSEQ: |
131 | case RET_ILSEQ: |
131 | errno = EILSEQ; |
132 | errno = EILSEQ; |
132 | break; |
133 | break; |
133 | } |
134 | } |
134 | 135 | ||
135 | *inbytesleft -= count1; |
136 | *inbytesleft -= count1; |
136 | *outbytesleft -= count2; |
137 | *outbytesleft -= count2; |
137 | *inbuf += count1; |
138 | *inbuf += count1; |
138 | *outbuf += count2; |
139 | *outbuf += count2; |
139 | return (size_t) -1; |
140 | return (size_t) -1; |
140 | } |
141 | } |
141 | 142 | ||
142 | /* Convert from wide to multibyte storing result in mbholder and num converted in numbytes2 */ |
143 | /* Convert from wide to multibyte storing result in mbholder and num converted in numbytes2 */ |
143 | /* Pass the minimum amount of space we have, one from mbholder and one from remaining in outbuf */ |
144 | /* Pass the minimum amount of space we have, one from mbholder and one from remaining in outbuf */ |
144 | int minspace = sizeof(mbholder) <= (*outbytesleft - count2) ? sizeof(mbholder) : (*outbytesleft - count2); |
145 | int minspace = sizeof(mbholder) <= (*outbytesleft - count2) ? sizeof(mbholder) : (*outbytesleft - count2); |
145 | 146 | ||
146 | int numbytes2 = (wctomb)(0, &mbholder[0], pwc, minspace); |
147 | int numbytes2 = (wctomb)(0, &mbholder[0], pwc, minspace); |
147 | if(numbytes2 < 0) { |
148 | if(numbytes2 < 0) { |
148 | switch(numbytes2) { |
149 | switch(numbytes2) { |
149 | case RET_ILUNI: |
150 | case RET_ILUNI: |
150 | errno = EILSEQ; |
151 | errno = EILSEQ; |
151 | break; |
152 | break; |
152 | case RET_TOOSMALL: |
153 | case RET_TOOSMALL: |
153 | errno = E2BIG; |
154 | errno = E2BIG; |
154 | break; |
155 | break; |
155 | } |
156 | } |
156 | 157 | ||
157 | *inbytesleft -= count1; |
158 | *inbytesleft -= count1; |
158 | *outbytesleft -= count2; |
159 | *outbytesleft -= count2; |
159 | *inbuf += count1; |
160 | *inbuf += count1; |
160 | *outbuf += count2; |
161 | *outbuf += count2; |
161 | 162 | ||
162 | return (size_t) -1; |
163 | return (size_t) -1; |
163 | } |
164 | } |
164 | 165 | ||
165 | int i; |
166 | int i; |
166 | for(i = 0; i < numbytes2; i++) { |
167 | for(i = 0; i < numbytes2; i++) { |
167 | *(*outbuf + count2 + i) = mbholder[i]; |
168 | *(*outbuf + count2 + i) = mbholder[i]; |
168 | } |
169 | } |
169 | 170 | ||
170 | count1+=numbytes; |
171 | count1+=numbytes; |
171 | count2+=numbytes2; |
172 | count2+=numbytes2; |
172 | } |
173 | } |
173 | 174 | ||
174 | /* Successfully converted everything, update the variables and return number of bytes converted */ |
175 | /* Successfully converted everything, update the variables and return number of bytes converted */ |
175 | *inbytesleft -= count1; |
176 | *inbytesleft -= count1; |
176 | *outbytesleft -= count2; |
177 | *outbytesleft -= count2; |
177 | *inbuf += count1; |
178 | *inbuf += count1; |
178 | *outbuf += count2; |
179 | *outbuf += count2; |
179 | 180 | ||
180 | return count1; |
181 | return count1; |
181 | } |
182 | } |
182 | 183 | ||
183 | /* int main() */ |
184 | /* int main() */ |
184 | /* { */ |
185 | /* { */ |
185 | /* char *s;// ="вертолет"; */ |
186 | /* char *s;// ="вертолет"; */ |
186 | /* char *z; */ |
187 | /* char *z; */ |
187 | /* //unsigned int pwc; */ |
188 | /* //unsigned int pwc; */ |
188 | /* iconv_t cd; */ |
189 | /* iconv_t cd; */ |
189 | /* size_t in, out; */ |
190 | /* size_t in, out; */ |
190 | 191 | ||
191 | /* FILE *infile; */ |
192 | /* FILE *infile; */ |
192 | /* char *fname = "file3.txt"; */ |
193 | /* char *fname = "file3.txt"; */ |
193 | 194 | ||
194 | /* size_t testmax = 100; */ |
195 | /* size_t testmax = 100; */ |
195 | /* size_t test = 0; */ |
196 | /* size_t test = 0; */ |
196 | 197 | ||
197 | /* infile = fopen(fname,"r"); */ |
198 | /* infile = fopen(fname,"r"); */ |
198 | 199 | ||
199 | /* fseek(infile, 0, SEEK_END); */ |
200 | /* fseek(infile, 0, SEEK_END); */ |
200 | /* size_t file_size = ftell(infile); */ |
201 | /* size_t file_size = ftell(infile); */ |
201 | /* rewind(infile); */ |
202 | /* rewind(infile); */ |
202 | 203 | ||
203 | /* char *buffer = (char*)malloc(file_size * sizeof(char)); */ |
204 | /* char *buffer = (char*)malloc(file_size * sizeof(char)); */ |
204 | /* if (buffer == NULL) */ |
205 | /* if (buffer == NULL) */ |
205 | /* { */ |
206 | /* { */ |
206 | /* fclose(infile); */ |
207 | /* fclose(infile); */ |
207 | /* printf("Error allocating %d bytes.\n", file_size * sizeof(char)); */ |
208 | /* printf("Error allocating %d bytes.\n", file_size * sizeof(char)); */ |
208 | /* return -1; */ |
209 | /* return -1; */ |
209 | /* } */ |
210 | /* } */ |
210 | /* size_t bytes_read = fread(buffer, sizeof(char), file_size, infile); */ |
211 | /* size_t bytes_read = fread(buffer, sizeof(char), file_size, infile); */ |
211 | /* if (bytes_read != file_size) */ |
212 | /* if (bytes_read != file_size) */ |
212 | /* { */ |
213 | /* { */ |
213 | /* /\* printf("Have read only %d bytes of %d.\n", bytes_read, file_size); *\/ */ |
214 | /* /\* printf("Have read only %d bytes of %d.\n", bytes_read, file_size); *\/ */ |
214 | /* free(buffer); */ |
215 | /* free(buffer); */ |
215 | /* fclose(infile); */ |
216 | /* fclose(infile); */ |
216 | /* return -1; */ |
217 | /* return -1; */ |
217 | /* } */ |
218 | /* } */ |
218 | 219 | ||
219 | /* /\* in=strlen(buffer); *\/ */ |
220 | /* /\* in=strlen(buffer); *\/ */ |
220 | /* in = bytes_read; */ |
221 | /* in = bytes_read; */ |
221 | /* z=malloc(in+12000); */ |
222 | /* z=malloc(in+12000); */ |
222 | 223 | ||
223 | /* out=in-1000; */ |
224 | /* out=in-1000; */ |
224 | /* cd=iconv_open("UTF-8","UTF-8"); */ |
225 | /* cd=iconv_open("UTF-8","UTF-8"); */ |
225 | /* // printf("%x\n",cd); */ |
226 | /* // printf("%x\n",cd); */ |
226 | /* int t; */ |
227 | /* int t; */ |
227 | /* char *zor = z; */ |
228 | /* char *zor = z; */ |
228 | 229 | ||
229 | /* /\* for(t = 0; t < 27400; t++) *\/ */ |
230 | /* /\* for(t = 0; t < 27400; t++) *\/ */ |
230 | /* /\* printf("0x%x,", buffer[t]); *\/ */ |
231 | /* /\* printf("0x%x,", buffer[t]); *\/ */ |
231 | 232 | ||
232 | /* t=iconv(cd, &buffer, &in, &z, &out); */ |
233 | /* t=iconv(cd, &buffer, &in, &z, &out); */ |
233 | /* /\* printf("\nResult after iconv(): %d", t); *\/ */ |
234 | /* /\* printf("\nResult after iconv(): %d", t); *\/ */ |
234 | 235 | ||
235 | /* /\* for(t = 0; t < 24259; t++) *\/ */ |
236 | /* /\* for(t = 0; t < 24259; t++) *\/ */ |
236 | /* /\* printf("%c", zor[t]); *\/ */ |
237 | /* /\* printf("%c", zor[t]); *\/ */ |
237 | 238 | ||
238 | /* //for (;s |
239 | /* //for (;s |
239 | /* } */ |
240 | /* } */ |
240 | 241 | ||
241 | /* typedef struct */ |
242 | /* typedef struct */ |
242 | /* { */ |
243 | /* { */ |
243 | /* char *name; */ |
244 | /* char *name; */ |
244 | /* void *f; */ |
245 | /* void *f; */ |
245 | /* } export_t; */ |
246 | /* } export_t; */ |
246 | 247 | ||
247 | /* char szStart[] = "START"; */ |
248 | /* char szStart[] = "START"; */ |
248 | /* char szVersion[] = "version"; */ |
249 | /* char szVersion[] = "version"; */ |
249 | /* char sziconv_open[] = "iconv_open"; */ |
250 | /* char sziconv_open[] = "iconv_open"; */ |
250 | /* char sziconv[] = "iconv"; */ |
251 | /* char sziconv[] = "iconv"; */ |
251 | 252 | ||
252 | /* export_t EXPORTS[] __asm__("EXPORTS") = */ |
253 | /* export_t EXPORTS[] __asm__("EXPORTS") = */ |
253 | /* { */ |
254 | /* { */ |
254 | /* { szStart, (void*)0x0 }, */ |
255 | /* { szStart, (void*)0x0 }, */ |
255 | /* { szVersion, (void*)0x00010001 }, */ |
256 | /* { szVersion, (void*)0x00010001 }, */ |
256 | /* { sziconv_open, iconv_open }, */ |
257 | /* { sziconv_open, iconv_open }, */ |
257 | /* { sziconv, iconv }, */ |
258 | /* { sziconv, iconv }, */ |
258 | /* { NULL, NULL }, */ |
259 | /* { NULL, NULL }, */ |
259 | /* }; */>>>>=>>>16&0xFFFF0000; |
260 | /* }; */>>>>=>>>16&0xFFFF0000; |
260 | ><16&0xFFFF0000; |
261 | ><16&0xFFFF0000; |
261 | >> |
262 | >> |