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