1 #ifndef PLAN9
2 #include <sys/types.h>
3 #include <stdio.h>
4 #include <unistd.h>
5 #include <stdlib.h>
6 #include <fcntl.h>
7 #include <string.h>
8 #include <errno.h>
9 #include "plan9.h"
10 #else /* PLAN9 */
11 #include <u.h>
12 #include <libc.h>
13 #include <bio.h>
14 #endif /* PLAN9 */
15 #include "cyrillic.h"
16 #include "misc.h"
17 #include "ms.h"
18 #include "8859.h"
19 #include "big5.h"
20 #include "gb.h"
21 #include "hdr.h"
22 #include "conv.h"
23
24 void usage(void);
25 void list(void);
26 int squawk = 1;
27 int clean = 0;
28 int verbose = 0;
29 long ninput, noutput, nrunes, nerrors;
30 char *file = "stdin";
31 char *argv0;
32 Rune runes[N];
33 char obuf[UTFmax*N]; /* maximum bloat from N runes */
34 long tab[NRUNE];
35 #ifndef PLAN9
36 extern char version[];
37 #endif
38
39 void intable(int, long *, struct convert *);
40 void unicode_in(int, long *, struct convert *);
41 void unicode_out(Rune *, int, long *);
42
43 int
main(int argc,char ** argv)44 main(int argc, char **argv)
45 {
46 char *from = "utf";
47 char *to = "utf";
48 int fd;
49 int listem = 0;
50 struct convert *t, *f;
51
52 ARGBEGIN {
53 case 'c':
54 clean = 1;
55 break;
56 case 'f':
57 from = EARGF(usage());
58 break;
59 case 'l':
60 listem = 1;
61 break;
62 case 's':
63 squawk = 0;
64 break;
65 case 't':
66 to = EARGF(usage());
67 break;
68 case 'v':
69 verbose = 1;
70 break;
71 default:
72 usage();
73 break;
74 } ARGEND
75
76 USED(argc);
77 if(verbose)
78 squawk = 1;
79 if(listem){
80 list();
81 EXIT(0, 0);
82 }
83 if(!from || !to)
84 usage();
85 f = conv(from, 1);
86 t = conv(to, 0);
87 #define PROC {if(f->flags&Table)\
88 intable(fd, (long *)f->data, t);\
89 else\
90 ((Infn)(f->fn))(fd, (long *)0, t);}
91 if(*argv){
92 while(*argv){
93 file = *argv;
94 #ifndef PLAN9
95 if((fd = open(*argv, 0)) < 0){
96 EPR "%s: %s: %s\n", argv0, *argv, strerror(errno));
97 #else /* PLAN9 */
98 if((fd = open(*argv, OREAD)) < 0){
99 EPR "%s: %s: %r\n", argv0, *argv);
100 #endif /* PLAN9 */
101 EXIT(1, "open failure");
102 }
103 PROC
104 close(fd);
105 argv++;
106 }
107 } else {
108 fd = 0;
109 PROC
110 }
111 if(verbose)
112 EPR "%s: %ld input bytes, %ld runes, %ld output bytes (%ld errors)\n", argv0,
113 ninput, nrunes, noutput, nerrors);
114 EXIT(((nerrors && squawk)? 1:0), ((nerrors && squawk)? "conversion error":0));
115 return(0); /* shut up compiler */
116 }
117
118 void
119 usage(void)
120 {
121 EPR "Usage: %s [-slv] [-f cs] [-t cs] [file ...]\n", argv0);
122 verbose = 1;
123 list();
124 EXIT(1, "usage");
125 }
126
127 void
128 list(void)
129 {
130 struct convert *c;
131 char ch = verbose?'\t':' ';
132
133 #ifndef PLAN9
134 EPR "%s version = '%s'\n", argv0, version);
135 #endif
136 if(verbose)
137 EPR "character sets:\n");
138 else
139 EPR "cs:");
140 for(c = convert; c->name; c++){
141 if((c->flags&From) && c[1].name && (strcmp(c[1].name, c->name) == 0)){
142 EPR "%c%s", ch, c->name);
143 c++;
144 } else if(c->flags&Table)
145 EPR "%c%s", ch, c->name);
146 else if(c->flags&From)
147 EPR "%c%s(from)", ch, c->name);
148 else
149 EPR "%c%s(to)", ch, c->name);
150 if(verbose)
151 EPR "\t%s\n", c->chatter);
152 }
153 if(!verbose)
154 EPR "\n");
155 }
156
157 struct convert *
158 conv(char *name, int from)
159 {
160 struct convert *c;
161
162 for(c = convert; c->name; c++){
163 if(cistrcmp(c->name, name) != 0)
164 continue;
165 if(c->flags&Table)
166 return(c);
167 if(((c->flags&From) == 0) == (from == 0))
168 return(c);
169 }
170 EPR "%s: charset `%s' unknown\n", argv0, name);
171 EXIT(1, "unknown character set");
172 return(0); /* just shut the compiler up */
173 }
174
175 void
176 swab2(char *b, int n)
177 {
178 char *e, p;
179
180 for(e = b+n; b < e; b++){
181 p = *b;
182 *b = b[1];
183 *++b = p;
184 }
185 }
186
187 void
188 unicode_in(int fd, long *notused, struct convert *out)
189 {
190 u16int ubuf[N];
191 Rune buf[N];
192 int i, n;
193 int swabme;
194
195 USED(notused);
196 if(read(fd, (char *)ubuf, 2) != 2)
197 return;
198 ninput += 2;
199 switch(ubuf[0])
200 {
201 default:
202 buf[0] = ubuf[0];
203 OUT(out, buf, 1);
204 case 0xFEFF:
205 swabme = 0;
206 break;
207 case 0xFFFE:
208 swabme = 1;
209 break;
210 }
211 while((n = read(fd, (char *)ubuf, 2*N)) > 0){
212 ninput += n;
213 if(swabme)
214 swab2((char *)ubuf, n);
215 for(i=0; i<n/2; i++)
216 buf[i] = ubuf[i];
217 if(n&1){
218 if(squawk)
219 EPR "%s: odd byte count in %s\n", argv0, file);
220 nerrors++;
221 if(clean)
222 n--;
223 else
224 buf[n++/2] = Runeerror;
225 }
226 OUT(out, buf, n/2);
227 }
228 }
229
230 void
231 unicode_in_be(int fd, long *notused, struct convert *out)
232 {
233 int i, n;
234 u16int ubuf[N];
235 Rune buf[N], r;
236 uchar *p;
237
238 USED(notused);
239 while((n = read(fd, (char *)ubuf, 2*N)) > 0){
240 ninput += n;
241 p = (uchar*)ubuf;
242 for(i=0; i<n/2; i++){
243 r = *p++<<8;
244 r |= *p++;
245 buf[i] = r;
246 }
247 if(n&1){
248 if(squawk)
249 EPR "%s: odd byte count in %s\n", argv0, file);
250 nerrors++;
251 if(clean)
252 n--;
253 else
254 buf[n++/2] = Runeerror;
255 }
256 OUT(out, buf, n/2);
257 }
258 OUT(out, buf, 0);
259 }
260
261 void
262 unicode_in_le(int fd, long *notused, struct convert *out)
263 {
264 int i, n;
265 u16int ubuf[N];
266 Rune buf[N], r;
267 uchar *p;
268
269 USED(notused);
270 while((n = read(fd, (char *)ubuf, 2*N)) > 0){
271 ninput += n;
272 p = (uchar*)ubuf;
273 for(i=0; i<n/2; i++){
274 r = *p++;
275 r |= *p++<<8;
276 buf[i] = r;
277 }
278 if(n&1){
279 if(squawk)
280 EPR "%s: odd byte count in %s\n", argv0, file);
281 nerrors++;
282 if(clean)
283 n--;
284 else
285 buf[n++/2] = Runeerror;
286 }
287 OUT(out, buf, n/2);
288 }
289 OUT(out, buf, 0);
290 }
291
292 void
293 unicode_out(Rune *base, int n, long *notused)
294 {
295 static int first = 1;
296 u16int buf[N];
297 int i;
298
299 USED(notused);
300 nrunes += n;
301 if(first){
302 u16int x = 0xFEFF;
303 noutput += 2;
304 write(1, (char *)&x, 2);
305 first = 0;
306 }
307 noutput += 2*n;
308 for(i=0; i<n; i++)
309 buf[i] = base[i];
310 write(1, (char *)buf, 2*n);
311 }
312
313 void
314 unicode_out_be(Rune *base, int n, long *notused)
315 {
316 int i;
317 uchar *p;
318 Rune r;
319
320 USED(notused);
321 p = (uchar*)base;
322 for(i=0; i<n; i++){
323 r = base[i];
324 *p++ = r>>8;
325 *p++ = r;
326 }
327 nrunes += n;
328 noutput += 2*n;
329 write(1, (char *)base, 2*n);
330 }
331
332 void
333 unicode_out_le(Rune *base, int n, long *notused)
334 {
335 int i;
336 uchar *p;
337 Rune r;
338
339 USED(notused);
340 p = (uchar*)base;
341 for(i=0; i<n; i++){
342 r = base[i];
343 *p++ = r;
344 *p++ = r>>8;
345 }
346 nrunes += n;
347 noutput += 2*n;
348 write(1, (char *)base, 2*n);
349 }
350
351 void
352 intable(int fd, long *table, struct convert *out)
353 {
354 uchar buf[N];
355 uchar *p, *e;
356 Rune *r;
357 int n;
358 long c;
359
360 while((n = read(fd, (char *)buf, N)) > 0){
361 ninput += n;
362 r = runes;
363 for(p = buf, e = buf+n; p < e; p++){
364 c = table[*p];
365 if(c < 0){
366 if(squawk)
367 EPR "%s: bad char 0x%x near byte %ld in %s\n", argv0, *p, ninput+(p-e), file);
368 nerrors++;
369 if(clean)
370 continue;
371 c = BADMAP;
372 }
373 *r++ = c;
374 }
375 OUT(out, runes, r-runes);
376 }
377 OUT(out, runes, 0);
378 if(n < 0){
379 #ifdef PLAN9
380 EPR "%s: input read: %r\n", argv0);
381 #else
382 EPR "%s: input read: %s\n", argv0, strerror(errno));
383 #endif
384 EXIT(1, "input read error");
385 }
386 }
387
388 void
389 outtable(Rune *base, int n, long *map)
390 {
391 long c;
392 char *p;
393 int i;
394
395 nrunes += n;
396 for(i = 0; i < NRUNE; i++)
397 tab[i] = -1;
398 for(i = 0; i < 256; i++)
399 if(map[i] >= 0)
400 tab[map[i]] = i;
401 for(i = 0, p = obuf; i < n; i++){
402 c = tab[base[i]];
403 if(c < 0){
404 if(squawk)
405 EPR "%s: rune 0x%x not in output cs\n", argv0, base[i]);
406 nerrors++;
407 if(clean)
408 continue;
409 c = BADMAP;
410 }
411 *p++ = c;
412 }
413 noutput += p-obuf;
414 write(1, obuf, p-obuf);
415 }
416
417 long tabascii[256] =
418 {
419 0x00,0x01,0x02,0x03,0x04,0x05,0x06,0x07,0x08,0x09,0x0a,0x0b,0x0c,0x0d,0x0e,0x0f,
420 0x10,0x11,0x12,0x13,0x14,0x15,0x16,0x17,0x18,0x19,0x1a,0x1b,0x1c,0x1d,0x1e,0x1f,
421 0x20,0x21,0x22,0x23,0x24,0x25,0x26,0x27,0x28,0x29,0x2a,0x2b,0x2c,0x2d,0x2e,0x2f,
422 0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0x3a,0x3b,0x3c,0x3d,0x3e,0x3f,
423 0x40,0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0x4a,0x4b,0x4c,0x4d,0x4e,0x4f,
424 0x50,0x51,0x52,0x53,0x54,0x55,0x56,0x57,0x58,0x59,0x5a,0x5b,0x5c,0x5d,0x5e,0x5f,
425 0x60,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0x6a,0x6b,0x6c,0x6d,0x6e,0x6f,
426 0x70,0x71,0x72,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7a,0x7b,0x7c,0x7d,0x7e,0x7f,
427 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
428 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
429 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
430 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
431 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
432 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
433 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
434 -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1
435 };
436
437 long tabmsdos[256] = /* from jhelling@cs.ruu.nl (Jeroen Hellingman) */
438 {
439 0x00,0x01,0x02,0x03,0x04,0x05,0x06,0x07,0x08,0x09,0x0a,0x0b,0x0c,0x0d,0x0e,0x0f,
440 0x10,0x11,0x12,0x13,0x14,0x15,0x16,0x17,0x18,0x19,0x1a,0x1b,0x1c,0x1d,0x1e,0x1f,
441 0x20,0x21,0x22,0x23,0x24,0x25,0x26,0x27,0x28,0x29,0x2a,0x2b,0x2c,0x2d,0x2e,0x2f,
442 0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0x3a,0x3b,0x3c,0x3d,0x3e,0x3f,
443 0x40,0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0x4a,0x4b,0x4c,0x4d,0x4e,0x4f,
444 0x50,0x51,0x52,0x53,0x54,0x55,0x56,0x57,0x58,0x59,0x5a,0x5b,0x5c,0x5d,0x5e,0x5f,
445 0x60,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0x6a,0x6b,0x6c,0x6d,0x6e,0x6f,
446 0x70,0x71,0x72,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7a,0x7b,0x7c,0x7d,0x7e,0x7f,
447 0x00c7, 0x00fc, 0x00e9, 0x00e2, 0x00e4, 0x00e0, 0x00e5, 0x00e7, /* latin */
448 0x00ea, 0x00eb, 0x00e8, 0x00ef, 0x00ee, 0x00ec, 0x00c4, 0x00c5,
449 0x00c9, 0x00e6, 0x00c6, 0x00f4, 0x00f6, 0x00f2, 0x00fb, 0x00f9,
450 0x00ff, 0x00d6, 0x00dc, 0x00a2, 0x00a3, 0x00a5, 0x20a7, 0x0192,
451 0x00e1, 0x00ed, 0x00f3, 0x00fa, 0x00f1, 0x00d1, 0x00aa, 0x00ba,
452 0x00bf, 0x2310, 0x00ac, 0x00bd, 0x00bc, 0x00a1, 0x00ab, 0x00bb,
453 0x2591, 0x2592, 0x2593, 0x2502, 0x2524, 0x2561, 0x2562, 0x2556, /* forms */
454 0x2555, 0x2563, 0x2551, 0x2557, 0x255d, 0x255c, 0x255b, 0x2510,
455 0x2514, 0x2534, 0x252c, 0x251c, 0x2500, 0x253c, 0x255e, 0x255f,
456 0x255a, 0x2554, 0x2569, 0x2566, 0x2560, 0x2550, 0x256c, 0x2567,
457 0x2568, 0x2564, 0x2565, 0x2559, 0x2558, 0x2552, 0x2553, 0x256b,
458 0x256a, 0x2518, 0x250c, 0x2588, 0x2584, 0x258c, 0x2590, 0x2580,
459 0x03b1, 0x00df, 0x0393, 0x03c0, 0x03a3, 0x03c3, 0x00b5, 0x03c4, /* greek */
460 0x03a6, 0x0398, 0x2126, 0x03b4, 0x221e, 0x2205, 0x2208, 0x2229,
461 0x2261, 0x00b1, 0x2265, 0x2264, 0x2320, 0x2321, 0x00f7, 0x2248, /* math */
462 0x00b0, 0x2022, 0x00b7, 0x221a, 0x207f, 0x00b2, 0x220e, 0x00a0
463 };
464 long tabmsdos2[256] = /* from jhelling@cs.ruu.nl (Jeroen Hellingman) */
465 {
466 0x0000, 0x263a, 0x263b, 0x2665, 0x2666, 0x2663, 0x2660, 0x2022,
467 0x25d8, 0x25cb, 0x25d9, 0x2642, 0x2640, 0x266a, 0x266b, 0x263c,
468 0x25b6, 0x25c0, 0x2195, 0x203c, 0x00b6, 0x00a7, 0x2043, 0x21a8,
469 0x2191, 0x2193, 0x2192, 0x2190, 0x2319, 0x2194, 0x25b2, 0x25bc,
470 0x20,0x21,0x22,0x23,0x24,0x25,0x26,0x27,0x28,0x29,0x2a,0x2b,0x2c,0x2d,0x2e,0x2f,
471 0x30,0x31,0x32,0x33,0x34,0x35,0x36,0x37,0x38,0x39,0x3a,0x3b,0x3c,0x3d,0x3e,0x3f,
472 0x40,0x41,0x42,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0x4a,0x4b,0x4c,0x4d,0x4e,0x4f,
473 0x50,0x51,0x52,0x53,0x54,0x55,0x56,0x57,0x58,0x59,0x5a,0x5b,0x5c,0x5d,0x5e,0x5f,
474 0x60,0x61,0x62,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0x6a,0x6b,0x6c,0x6d,0x6e,0x6f,
475 0x70,0x71,0x72,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7a,0x7b,0x7c,0x7d,0x7e,0x7f,
476 0x00c7, 0x00fc, 0x00e9, 0x00e2, 0x00e4, 0x00e0, 0x00e5, 0x00e7, /* latin */
477 0x00ea, 0x00eb, 0x00e8, 0x00ef, 0x00ee, 0x00ec, 0x00c4, 0x00c5,
478 0x00c9, 0x00e6, 0x00c6, 0x00f4, 0x00f6, 0x00f2, 0x00fb, 0x00f9,
479 0x00ff, 0x00d6, 0x00dc, 0x00a2, 0x00a3, 0x00a5, 0x20a7, 0x0192,
480 0x00e1, 0x00ed, 0x00f3, 0x00fa, 0x00f1, 0x00d1, 0x00aa, 0x00ba,
481 0x00bf, 0x2310, 0x00ac, 0x00bd, 0x00bc, 0x00a1, 0x00ab, 0x00bb,
482 0x2591, 0x2592, 0x2593, 0x2502, 0x2524, 0x2561, 0x2562, 0x2556, /* forms */
483 0x2555, 0x2563, 0x2551, 0x2557, 0x255d, 0x255c, 0x255b, 0x2510,
484 0x2514, 0x2534, 0x252c, 0x251c, 0x2500, 0x253c, 0x255e, 0x255f,
485 0x255a, 0x2554, 0x2569, 0x2566, 0x2560, 0x2550, 0x256c, 0x2567,
486 0x2568, 0x2564, 0x2565, 0x2559, 0x2558, 0x2552, 0x2553, 0x256b,
487 0x256a, 0x2518, 0x250c, 0x2588, 0x2584, 0x258c, 0x2590, 0x2580,
488 0x03b1, 0x00df, 0x0393, 0x03c0, 0x03a3, 0x03c3, 0x00b5, 0x03c4, /* greek */
489 0x03a6, 0x0398, 0x2126, 0x03b4, 0x221e, 0x2205, 0x2208, 0x2229,
490 0x2261, 0x00b1, 0x2265, 0x2264, 0x2320, 0x2321, 0x00f7, 0x2248, /* math */
491 0x00b0, 0x2022, 0x00b7, 0x221a, 0x207f, 0x00b2, 0x220e, 0x00a0
492 };
493 struct convert convert[] =
494 { /* if two entries have the same name, put the from one first */
495 { "8859-1", "Latin-1 (Western and Northern Europe including Italian)", Table, (void *)tab8859_1 },
496 { "8859-2", "Latin-2 (Eastern Europe except Turkey and the Baltic countries)", Table, (void *)tab8859_2 },
497 { "8859-3", "Latin-3 (Mediterranean, South Africa, Esperanto)", Table, (void *)tab8859_3 },
498 { "8859-4", "Latin-4 (Scandinavia and the Baltic countries; obsolete)", Table, (void *)tab8859_4 },
499 { "8859-5", "Part 5 (Cyrillic)", Table, (void *)tab8859_5 },
500 { "8859-6", "Part 6 (Arabic)", Table, (void *)tab8859_6 },
501 { "8859-7", "Part 7 (Greek)", Table, (void *)tab8859_7 },
502 { "8859-8", "Part 8 (Hebrew)", Table, (void *)tab8859_8 },
503 { "8859-9", "Latin-5 (Turkey, Western Europe except Icelandic and Faroese)", Table, (void *)tab8859_9 },
504 { "8859-10", "Latin-6 (Northern Europe)", Table, (void *)tab8859_10 },
505 { "8859-15", "Latin-9 (Western Europe)", Table, (void *)tab8859_15 },
506 { "ascii", "7-bit ASCII", Table, (void *)tabascii },
507 { "atari", "ATARI-ST character set", Table, (void *)tabatari },
508 { "av", "Alternativnyj Variant", Table, (void *)tabav },
509 { "big5", "Big 5 (HKU)", From|Func, 0, (Fnptr)big5_in },
510 { "big5", "Big 5 (HKU)", Func, 0, (Fnptr)big5_out },
511 { "ebcdic", "EBCDIC", Table, (void *)tabebcdic }, /* 6f is recommended bad map */
512 { "euc-k", "Korean EUC: ASCII+KS C 5601 1987", From|Func, 0, (Fnptr)uksc_in },
513 { "euc-k", "Korean EUC: ASCII+KS C 5601 1987", Func, 0, (Fnptr)uksc_out },
514 { "gb2312", "GB2312-80 (Chinese)", From|Func, 0, (Fnptr)gb_in },
515 { "gb2312", "GB2312-80 (Chinese)", Func, 0, (Fnptr)gb_out },
516 { "html", "HTML", From|Func, 0, (Fnptr)html_in },
517 { "html", "HTML", Func, 0, (Fnptr)html_out },
518 { "ibm437", "IBM Code Page 437 (US)", Table, (void*)tabcp437 },
519 { "ibm720", "IBM Code Page 720 (Arabic)", Table, (void*)tabcp720 },
520 { "ibm737", "IBM Code Page 737 (Greek)", Table, (void*)tabcp737 },
521 { "ibm775", "IBM Code Page 775 (Baltic)", Table, (void*)tabcp775 },
522 { "ibm850", "IBM Code Page 850 (Multilingual Latin I)", Table, (void*)tabcp850 },
523 { "ibm852", "IBM Code Page 852 (Latin II)", Table, (void*)tabcp852 },
524 { "ibm855", "IBM Code Page 855 (Cyrillic)", Table, (void*)tabcp855 },
525 { "ibm857", "IBM Code Page 857 (Turkish)", Table, (void*)tabcp857 },
526 { "ibm858", "IBM Code Page 858 (Multilingual Latin I+Euro)", Table, (void*)tabcp858 },
527 { "ibm862", "IBM Code Page 862 (Hebrew)", Table, (void*)tabcp862 },
528 { "ibm866", "IBM Code Page 866 (Russian)", Table, (void*)tabcp866 },
529 { "ibm874", "IBM Code Page 874 (Thai)", Table, (void*)tabcp874 },
530 { "iso-2022-jp", "alias for jis-kanji (MIME)", From|Func, 0, (Fnptr)jisjis_in },
531 { "iso-2022-jp", "alias for jis-kanji (MIME)", Func, 0, (Fnptr)jisjis_out },
532 { "iso-8859-1", "alias for 8859-1 (MIME)", Table, (void *)tab8859_1 },
533 { "iso-8859-2", "alias for 8859-2 (MIME)", Table, (void *)tab8859_2 },
534 { "iso-8859-3", "alias for 8859-3 (MIME)", Table, (void *)tab8859_3 },
535 { "iso-8859-4", "alias for 8859-4 (MIME)", Table, (void *)tab8859_4 },
536 { "iso-8859-5", "alias for 8859-5 (MIME)", Table, (void *)tab8859_5 },
537 { "iso-8859-6", "alias for 8859-6 (MIME)", Table, (void *)tab8859_6 },
538 { "iso-8859-7", "alias for 8859-7 (MIME)", Table, (void *)tab8859_7 },
539 { "iso-8859-8", "alias for 8859-8 (MIME)", Table, (void *)tab8859_8 },
540 { "iso-8859-9", "alias for 8859-9 (MIME)", Table, (void *)tab8859_9 },
541 { "iso-8859-10", "alias for 8859-10 (MIME)", Table, (void *)tab8859_10 },
542 { "iso-8859-15", "alias for 8859-15 (MIME)", Table, (void *)tab8859_15 },
543 { "jis", "guesses at the JIS encoding", From|Func, 0, (Fnptr)jis_in },
544 { "jis-kanji", "ISO 2022-JP (Japanese)", From|Func, 0, (Fnptr)jisjis_in },
545 { "jis-kanji", "ISO 2022-JP (Japanese)", Func, 0, (Fnptr)jisjis_out },
546 { "koi8", "KOI-8 (GOST 19769-74)", Table, (void *)tabkoi8 },
547 { "koi8-r", "alias for koi8 (MIME)", Table, (void *)tabkoi8 },
548 { "latin1", "alias for 8859-1", Table, (void *)tab8859_1 },
549 { "macrom", "Macintosh Standard Roman character set", Table, (void *)tabmacroman },
550 { "microsoft", "alias for windows1252", Table, (void *)tabcp1252 },
551 { "ms-kanji", "Microsoft, or Shift-JIS", From|Func, 0, (Fnptr)msjis_in },
552 { "ms-kanji", "Microsoft, or Shift-JIS", Func, 0, (Fnptr)msjis_out },
553 { "msdos", "IBM PC (alias for ibm437)", Table, (void *)tabcp437 },
554 { "msdos2", "IBM PC (ibm437 with graphics in C0)", Table, (void *)tabmsdos2 },
555 { "next", "NEXTSTEP character set", Table, (void *)tabnextstep },
556 { "ov", "Osnovnoj Variant", Table, (void *)tabov },
557 { "ps2", "IBM PS/2: (alias for ibm850)", Table, (void *)tabcp850 },
558 { "sf1", "ISO-646: Finnish/Swedish SF-1 variant", Table, (void *)tabsf1 },
559 { "sf2", "ISO-646: Finnish/Swedish SF-2 variant (recommended)", Table, (void *)tabsf2 },
560 { "tis-620", "Thai+ASCII (TIS 620-1986)", Table, (void *)tabtis620 },
561 { "tune", "TUNE (Tamil)", From|Func, 0, (Fnptr)tune_in },
562 { "tune", "TUNE (Tamil)", Func, 0, (Fnptr)tune_out },
563 { "ucode", "Russian U-code", Table, (void *)tabucode },
564 { "ujis", "EUC-JX: JIS 0208", From|Func, 0, (Fnptr)ujis_in },
565 { "ujis", "EUC-JX: JIS 0208", Func, 0, (Fnptr)ujis_out },
566 { "unicode", "Unicode 1.1", From|Func, 0, (Fnptr)unicode_in },
567 { "unicode", "Unicode 1.1", Func, 0, (Fnptr)unicode_out },
568 { "unicode-be", "Unicode 1.1 big-endian", From|Func, 0, (Fnptr)unicode_in_be },
569 { "unicode-be", "Unicode 1.1 big-endian", Func, 0, (Fnptr)unicode_out_be },
570 { "unicode-le", "Unicode 1.1 little-endian", From|Func, 0, (Fnptr)unicode_in_le },
571 { "unicode-le", "Unicode 1.1 little-endian", Func, 0, (Fnptr)unicode_out_le },
572 { "us-ascii", "alias for ascii (MIME)", Table, (void *)tabascii },
573 { "utf", "FSS-UTF a.k.a. UTF-8", From|Func, 0, (Fnptr)utf_in },
574 { "utf", "FSS-UTF a.k.a. UTF-8", Func, 0, (Fnptr)utf_out },
575 { "utf1", "UTF-1 (ISO 10646 Annex A)", From|Func, 0, (Fnptr)isoutf_in },
576 { "utf1", "UTF-1 (ISO 10646 Annex A)", Func, 0, (Fnptr)isoutf_out },
577 { "utf-8", "alias for utf (MIME)", From|Func, 0, (Fnptr)utf_in },
578 { "utf-8", "alias for utf (MIME)", Func, 0, (Fnptr)utf_out },
579 { "utf-16", "alias for unicode (MIME)", From|Func, 0, (Fnptr)unicode_in },
580 { "utf-16", "alias for unicode (MIME)", Func, 0, (Fnptr)unicode_out },
581 { "utf-16be", "alias for unicode-be (MIME)", From|Func, 0, (Fnptr)unicode_in_be },
582 { "utf-16be", "alias for unicode-be (MIME)", Func, 0, (Fnptr)unicode_out_be },
583 { "utf-16le", "alias for unicode-le (MIME)", From|Func, 0, (Fnptr)unicode_in_le },
584 { "utf-16le", "alias for unicode-le (MIME)", Func, 0, (Fnptr)unicode_out_le },
585 { "viet1", "Vietnamese VSCII-1 (1993)", Table, (void *)tabviet1 },
586 { "viet2", "Vietnamese VSCII-2 (1993)", Table, (void *)tabviet2 },
587 { "vscii", "Vietnamese VISCII 1.1 (1992)", Table, (void *)tabviscii },
588 { "windows-1250", "Windows Code Page 1250 (Central Europe)", Table, (void *)tabcp1250 },
589 { "windows-1251", "Windows Code Page 1251 (Cyrillic)", Table, (void *)tabcp1251 },
590 { "windows-1252", "Windows Code Page 1252 (Latin I)", Table, (void *)tabcp1252 },
591 { "windows-1253", "Windows Code Page 1253 (Greek)", Table, (void *)tabcp1253 },
592 { "windows-1254", "Windows Code Page 1254 (Turkish)", Table, (void *)tabcp1254 },
593 { "windows-1255", "Windows Code Page 1255 (Hebrew)", Table, (void *)tabcp1255 },
594 { "windows-1256", "Windows Code Page 1256 (Arabic)", Table, (void *)tabcp1256 },
595 { "windows-1257", "Windows Code Page 1257 (Baltic)", Table, (void *)tabcp1257 },
596 { "windows-1258", "Windows Code Page 1258 (Vietnam)", Table, (void *)tabcp1258 },
597 { 0 }
598 };
599