1 a31db67d 2004-04-21 devnull #ifdef PLAN9
  2 a31db67d 2004-04-21 devnull #include	<u.h>
  3 a31db67d 2004-04-21 devnull #include	<libc.h>
  4 a31db67d 2004-04-21 devnull #include	<bio.h>
  5 fa059a4e 2005-12-17 devnull #ifdef PLAN9PORT
  6 fa059a4e 2005-12-17 devnull #include	<errno.h>
  7 a31db67d 2004-04-21 devnull #else
  8 fa059a4e 2005-12-17 devnull extern int errno;
  9 fa059a4e 2005-12-17 devnull #endif
 10 fa059a4e 2005-12-17 devnull #else
 11 a31db67d 2004-04-21 devnull #include	<sys/types.h>
 12 a31db67d 2004-04-21 devnull #include	<stdio.h>
 13 a31db67d 2004-04-21 devnull #include	<stdlib.h>
 14 a31db67d 2004-04-21 devnull #include	<string.h>
 15 a31db67d 2004-04-21 devnull #include	<unistd.h>
 16 a31db67d 2004-04-21 devnull #include	<errno.h>
 17 a31db67d 2004-04-21 devnull #include	"plan9.h"
 18 a31db67d 2004-04-21 devnull #endif
 19 a31db67d 2004-04-21 devnull #include	"hdr.h"
 20 35d26aa3 2005-12-26 devnull #ifndef EILSEQ
 21 35d26aa3 2005-12-26 devnull #define EILSEQ 9998
 22 35d26aa3 2005-12-26 devnull #endif
 23 a31db67d 2004-04-21 devnull 
 24 a31db67d 2004-04-21 devnull /*
 25 a31db67d 2004-04-21 devnull 	the our_* routines are implementations for the corresponding library
 26 a31db67d 2004-04-21 devnull 	routines. for a while, i tried to actually name them wctomb etc
 27 a31db67d 2004-04-21 devnull 	but stopped that after i found a system which made wchar_t an
 28 a31db67d 2004-04-21 devnull 	unsigned char.
 29 a31db67d 2004-04-21 devnull */
 30 a31db67d 2004-04-21 devnull 
 31 a31db67d 2004-04-21 devnull int our_wctomb(char *s, unsigned long wc);
 32 a31db67d 2004-04-21 devnull int our_mbtowc(unsigned long *p, char *s, unsigned n);
 33 a31db67d 2004-04-21 devnull int runetoisoutf(char *str, Rune *rune);
 34 a31db67d 2004-04-21 devnull int fullisorune(char *str, int n);
 35 a31db67d 2004-04-21 devnull int isochartorune(Rune *rune, char *str);
 36 a31db67d 2004-04-21 devnull 
 37 a31db67d 2004-04-21 devnull void
 38 a31db67d 2004-04-21 devnull utf_in(int fd, long *notused, struct convert *out)
 39 a31db67d 2004-04-21 devnull {
 40 7551b2ec 2005-03-09 devnull 	char buf[N];
 41 7551b2ec 2005-03-09 devnull 	int i, j, c, n, tot;
 42 7551b2ec 2005-03-09 devnull 	ulong l;
 43 a31db67d 2004-04-21 devnull 
 44 a31db67d 2004-04-21 devnull 	USED(notused);
 45 7551b2ec 2005-03-09 devnull 	tot = 0;
 46 7551b2ec 2005-03-09 devnull 	while((n = read(fd, buf+tot, N-tot)) >= 0){
 47 7551b2ec 2005-03-09 devnull 		tot += n;
 48 7551b2ec 2005-03-09 devnull 		for(i=j=0; i<tot; ){
 49 7551b2ec 2005-03-09 devnull 			c = our_mbtowc(&l, buf+i, tot-i);
 50 7f0d675d 2005-09-13 devnull 			if(c == -2)
 51 7551b2ec 2005-03-09 devnull 				break;
 52 7f0d675d 2005-09-13 devnull 			if(c == -1){
 53 7551b2ec 2005-03-09 devnull 				if(squawk)
 54 7551b2ec 2005-03-09 devnull 					EPR "%s: bad UTF sequence near byte %ld in input\n", argv0, ninput+i);
 55 7551b2ec 2005-03-09 devnull 				if(clean)
 56 7551b2ec 2005-03-09 devnull 					continue;
 57 7551b2ec 2005-03-09 devnull 				nerrors++;
 58 7551b2ec 2005-03-09 devnull 				l = Runeerror;
 59 7f0d675d 2005-09-13 devnull 				c = 1;
 60 a31db67d 2004-04-21 devnull 			}
 61 7551b2ec 2005-03-09 devnull 			runes[j++] = l;
 62 7551b2ec 2005-03-09 devnull 			i += c;
 63 a31db67d 2004-04-21 devnull 		}
 64 7551b2ec 2005-03-09 devnull 		OUT(out, runes, j);
 65 7551b2ec 2005-03-09 devnull 		tot -= i;
 66 7551b2ec 2005-03-09 devnull 		ninput += i;
 67 7551b2ec 2005-03-09 devnull 		if(tot)
 68 7551b2ec 2005-03-09 devnull 			memmove(buf, buf+i, tot);
 69 7551b2ec 2005-03-09 devnull 		if(n == 0)
 70 7551b2ec 2005-03-09 devnull 			break;
 71 7551b2ec 2005-03-09 devnull 	}
 72 a31db67d 2004-04-21 devnull }
 73 a31db67d 2004-04-21 devnull 
 74 a31db67d 2004-04-21 devnull void
 75 a31db67d 2004-04-21 devnull utf_out(Rune *base, int n, long *notused)
 76 a31db67d 2004-04-21 devnull {
 77 a31db67d 2004-04-21 devnull 	char *p;
 78 a31db67d 2004-04-21 devnull 	Rune *r;
 79 a31db67d 2004-04-21 devnull 
 80 a31db67d 2004-04-21 devnull 	USED(notused);
 81 a31db67d 2004-04-21 devnull 	nrunes += n;
 82 a31db67d 2004-04-21 devnull 	for(r = base, p = obuf; n-- > 0; r++){
 83 a31db67d 2004-04-21 devnull 		p += our_wctomb(p, *r);
 84 a31db67d 2004-04-21 devnull 	}
 85 a31db67d 2004-04-21 devnull 	noutput += p-obuf;
 86 a31db67d 2004-04-21 devnull 	write(1, obuf, p-obuf);
 87 a31db67d 2004-04-21 devnull }
 88 a31db67d 2004-04-21 devnull 
 89 a31db67d 2004-04-21 devnull void
 90 a31db67d 2004-04-21 devnull isoutf_in(int fd, long *notused, struct convert *out)
 91 a31db67d 2004-04-21 devnull {
 92 7551b2ec 2005-03-09 devnull 	char buf[N];
 93 7551b2ec 2005-03-09 devnull 	int i, j, c, n, tot;
 94 a31db67d 2004-04-21 devnull 
 95 a31db67d 2004-04-21 devnull 	USED(notused);
 96 7551b2ec 2005-03-09 devnull 	tot = 0;
 97 7551b2ec 2005-03-09 devnull 	while((n = read(fd, buf+tot, N-tot)) >= 0){
 98 7551b2ec 2005-03-09 devnull 		tot += n;
 99 7551b2ec 2005-03-09 devnull 		for(i=j=0; i<tot; ){
100 7551b2ec 2005-03-09 devnull 			if(!fullisorune(buf+i, tot-i))
101 7551b2ec 2005-03-09 devnull 				break;
102 7551b2ec 2005-03-09 devnull 			c = isochartorune(&runes[j], buf+i);
103 7551b2ec 2005-03-09 devnull 			if(runes[j] == Runeerror){
104 7551b2ec 2005-03-09 devnull 				if(squawk)
105 7551b2ec 2005-03-09 devnull 					EPR "%s: bad UTF sequence near byte %ld in input\n", argv0, ninput+i);
106 7551b2ec 2005-03-09 devnull 				if(clean)
107 7551b2ec 2005-03-09 devnull 					continue;
108 7551b2ec 2005-03-09 devnull 				nerrors++;
109 a31db67d 2004-04-21 devnull 			}
110 7551b2ec 2005-03-09 devnull 			j++;
111 7551b2ec 2005-03-09 devnull 			i += c;
112 a31db67d 2004-04-21 devnull 		}
113 7551b2ec 2005-03-09 devnull 		OUT(out, runes, j);
114 7551b2ec 2005-03-09 devnull 		tot -= i;
115 7551b2ec 2005-03-09 devnull 		ninput += i;
116 7551b2ec 2005-03-09 devnull 		if(tot)
117 7551b2ec 2005-03-09 devnull 			memmove(buf, buf+i, tot);
118 7551b2ec 2005-03-09 devnull 		if(n == 0)
119 7551b2ec 2005-03-09 devnull 			break;
120 7551b2ec 2005-03-09 devnull 	}
121 a31db67d 2004-04-21 devnull }
122 a31db67d 2004-04-21 devnull 
123 a31db67d 2004-04-21 devnull void
124 a31db67d 2004-04-21 devnull isoutf_out(Rune *base, int n, long *notused)
125 a31db67d 2004-04-21 devnull {
126 a31db67d 2004-04-21 devnull 	char *p;
127 a31db67d 2004-04-21 devnull 	Rune *r;
128 a31db67d 2004-04-21 devnull 
129 a31db67d 2004-04-21 devnull 	USED(notused);
130 a31db67d 2004-04-21 devnull 	nrunes += n;
131 a31db67d 2004-04-21 devnull 	for(r = base, p = obuf; n-- > 0; r++)
132 a31db67d 2004-04-21 devnull 		p += runetoisoutf(p, r);
133 a31db67d 2004-04-21 devnull 	noutput += p-obuf;
134 a31db67d 2004-04-21 devnull 	write(1, obuf, p-obuf);
135 a31db67d 2004-04-21 devnull }
136 a31db67d 2004-04-21 devnull 
137 a31db67d 2004-04-21 devnull 
138 a31db67d 2004-04-21 devnull enum
139 a31db67d 2004-04-21 devnull {
140 a31db67d 2004-04-21 devnull 	Char1	= Runeself,	Rune1	= Runeself,
141 a31db67d 2004-04-21 devnull 	Char21	= 0xA1,		Rune21	= 0x0100,
142 a31db67d 2004-04-21 devnull 	Char22	= 0xF6,		Rune22	= 0x4016,
143 a31db67d 2004-04-21 devnull 	Char3	= 0xFC,		Rune3	= 0x10000,	/* really 0x38E2E */
144 a31db67d 2004-04-21 devnull 	Esc	= 0xBE,		Bad	= Runeerror
145 a31db67d 2004-04-21 devnull };
146 a31db67d 2004-04-21 devnull 
147 a31db67d 2004-04-21 devnull static	uchar	U[256];
148 a31db67d 2004-04-21 devnull static	uchar	T[256];
149 a31db67d 2004-04-21 devnull 
150 a31db67d 2004-04-21 devnull static
151 a31db67d 2004-04-21 devnull void
152 a31db67d 2004-04-21 devnull mktable(void)
153 a31db67d 2004-04-21 devnull {
154 a31db67d 2004-04-21 devnull 	int i, u;
155 a31db67d 2004-04-21 devnull 
156 a31db67d 2004-04-21 devnull 	for(i=0; i<256; i++) {
157 a31db67d 2004-04-21 devnull 		u = i + (0x5E - 0xA0);
158 a31db67d 2004-04-21 devnull 		if(i < 0xA0)
159 a31db67d 2004-04-21 devnull 			u = i + (0xDF - 0x7F);
160 a31db67d 2004-04-21 devnull 		if(i < 0x7F)
161 a31db67d 2004-04-21 devnull 			u = i + (0x00 - 0x21);
162 a31db67d 2004-04-21 devnull 		if(i < 0x21)
163 a31db67d 2004-04-21 devnull 			u = i + (0xBE - 0x00);
164 a31db67d 2004-04-21 devnull 		U[i] = u;
165 a31db67d 2004-04-21 devnull 		T[u] = i;
166 a31db67d 2004-04-21 devnull 	}
167 a31db67d 2004-04-21 devnull }
168 a31db67d 2004-04-21 devnull 
169 a31db67d 2004-04-21 devnull int
170 a31db67d 2004-04-21 devnull isochartorune(Rune *rune, char *str)
171 a31db67d 2004-04-21 devnull {
172 a31db67d 2004-04-21 devnull 	int c, c1, c2;
173 a31db67d 2004-04-21 devnull 	long l;
174 a31db67d 2004-04-21 devnull 
175 a31db67d 2004-04-21 devnull 	if(U[0] == 0)
176 a31db67d 2004-04-21 devnull 		mktable();
177 a31db67d 2004-04-21 devnull 
178 a31db67d 2004-04-21 devnull 	/*
179 a31db67d 2004-04-21 devnull 	 * one character sequence
180 a31db67d 2004-04-21 devnull 	 *	00000-0009F => 00-9F
181 a31db67d 2004-04-21 devnull 	 */
182 a31db67d 2004-04-21 devnull 	c = *(uchar*)str;
183 a31db67d 2004-04-21 devnull 	if(c < Char1) {
184 a31db67d 2004-04-21 devnull 		*rune = c;
185 a31db67d 2004-04-21 devnull 		return 1;
186 a31db67d 2004-04-21 devnull 	}
187 a31db67d 2004-04-21 devnull 
188 a31db67d 2004-04-21 devnull 	/*
189 a31db67d 2004-04-21 devnull 	 * two character sequence
190 a31db67d 2004-04-21 devnull 	 *	000A0-000FF => A0; A0-FF
191 a31db67d 2004-04-21 devnull 	 */
192 a31db67d 2004-04-21 devnull 	c1 = *(uchar*)(str+1);
193 a31db67d 2004-04-21 devnull 	if(c < Char21) {
194 a31db67d 2004-04-21 devnull 		if(c1 >= Rune1 && c1 < Rune21) {
195 a31db67d 2004-04-21 devnull 			*rune = c1;
196 a31db67d 2004-04-21 devnull 			return 2;
197 a31db67d 2004-04-21 devnull 		}
198 a31db67d 2004-04-21 devnull 		goto bad;
199 a31db67d 2004-04-21 devnull 	}
200 a31db67d 2004-04-21 devnull 
201 a31db67d 2004-04-21 devnull 	/*
202 a31db67d 2004-04-21 devnull 	 * two character sequence
203 a31db67d 2004-04-21 devnull 	 *	00100-04015 => A1-F5; 21-7E/A0-FF
204 a31db67d 2004-04-21 devnull 	 */
205 a31db67d 2004-04-21 devnull 	c1 = U[c1];
206 a31db67d 2004-04-21 devnull 	if(c1 >= Esc)
207 a31db67d 2004-04-21 devnull 		goto bad;
208 a31db67d 2004-04-21 devnull 	if(c < Char22) {
209 a31db67d 2004-04-21 devnull 		*rune =  (c-Char21)*Esc + c1 + Rune21;
210 a31db67d 2004-04-21 devnull 		return 2;
211 a31db67d 2004-04-21 devnull 	}
212 a31db67d 2004-04-21 devnull 
213 a31db67d 2004-04-21 devnull 	/*
214 a31db67d 2004-04-21 devnull 	 * three character sequence
215 a31db67d 2004-04-21 devnull 	 *	04016-38E2D => A6-FB; 21-7E/A0-FF
216 a31db67d 2004-04-21 devnull 	 */
217 a31db67d 2004-04-21 devnull 	c2 = U[*(uchar*)(str+2)];
218 a31db67d 2004-04-21 devnull 	if(c2 >= Esc)
219 a31db67d 2004-04-21 devnull 		goto bad;
220 a31db67d 2004-04-21 devnull 	if(c < Char3) {
221 a31db67d 2004-04-21 devnull 		l = (c-Char22)*Esc*Esc + c1*Esc + c2 + Rune22;
222 a31db67d 2004-04-21 devnull 		if(l >= Rune3)
223 a31db67d 2004-04-21 devnull 			goto bad;
224 a31db67d 2004-04-21 devnull 		*rune = l;
225 a31db67d 2004-04-21 devnull 		return 3;
226 a31db67d 2004-04-21 devnull 	}
227 a31db67d 2004-04-21 devnull 
228 a31db67d 2004-04-21 devnull 	/*
229 a31db67d 2004-04-21 devnull 	 * bad decoding
230 a31db67d 2004-04-21 devnull 	 */
231 a31db67d 2004-04-21 devnull bad:
232 a31db67d 2004-04-21 devnull 	*rune = Bad;
233 a31db67d 2004-04-21 devnull 	return 1;
234 a31db67d 2004-04-21 devnull }
235 a31db67d 2004-04-21 devnull 
236 a31db67d 2004-04-21 devnull int
237 a31db67d 2004-04-21 devnull runetoisoutf(char *str, Rune *rune)
238 a31db67d 2004-04-21 devnull {
239 a31db67d 2004-04-21 devnull 	long c;
240 a31db67d 2004-04-21 devnull 
241 a31db67d 2004-04-21 devnull 	if(T[0] == 0)
242 a31db67d 2004-04-21 devnull 		mktable();
243 a31db67d 2004-04-21 devnull 
244 a31db67d 2004-04-21 devnull 	/*
245 a31db67d 2004-04-21 devnull 	 * one character sequence
246 a31db67d 2004-04-21 devnull 	 *	00000-0009F => 00-9F
247 a31db67d 2004-04-21 devnull 	 */
248 a31db67d 2004-04-21 devnull 	c = *rune;
249 a31db67d 2004-04-21 devnull 	if(c < Rune1) {
250 a31db67d 2004-04-21 devnull 		str[0] = c;
251 a31db67d 2004-04-21 devnull 		return 1;
252 a31db67d 2004-04-21 devnull 	}
253 a31db67d 2004-04-21 devnull 
254 a31db67d 2004-04-21 devnull 	/*
255 a31db67d 2004-04-21 devnull 	 * two character sequence
256 a31db67d 2004-04-21 devnull 	 *	000A0-000FF => A0; A0-FF
257 a31db67d 2004-04-21 devnull 	 */
258 a31db67d 2004-04-21 devnull 	if(c < Rune21) {
259 b5932d3d 2005-03-18 devnull 		str[0] = (char)Char1;
260 a31db67d 2004-04-21 devnull 		str[1] = c;
261 a31db67d 2004-04-21 devnull 		return 2;
262 a31db67d 2004-04-21 devnull 	}
263 a31db67d 2004-04-21 devnull 
264 a31db67d 2004-04-21 devnull 	/*
265 a31db67d 2004-04-21 devnull 	 * two character sequence
266 a31db67d 2004-04-21 devnull 	 *	00100-04015 => A1-F5; 21-7E/A0-FF
267 a31db67d 2004-04-21 devnull 	 */
268 a31db67d 2004-04-21 devnull 	if(c < Rune22) {
269 a31db67d 2004-04-21 devnull 		c -= Rune21;
270 a31db67d 2004-04-21 devnull 		str[0] = c/Esc + Char21;
271 a31db67d 2004-04-21 devnull 		str[1] = T[c%Esc];
272 a31db67d 2004-04-21 devnull 		return 2;
273 a31db67d 2004-04-21 devnull 	}
274 a31db67d 2004-04-21 devnull 
275 a31db67d 2004-04-21 devnull 	/*
276 a31db67d 2004-04-21 devnull 	 * three character sequence
277 a31db67d 2004-04-21 devnull 	 *	04016-38E2D => A6-FB; 21-7E/A0-FF
278 a31db67d 2004-04-21 devnull 	 */
279 a31db67d 2004-04-21 devnull 	c -= Rune22;
280 a31db67d 2004-04-21 devnull 	str[0] = c/(Esc*Esc) + Char22;
281 a31db67d 2004-04-21 devnull 	str[1] = T[c/Esc%Esc];
282 a31db67d 2004-04-21 devnull 	str[2] = T[c%Esc];
283 a31db67d 2004-04-21 devnull 	return 3;
284 a31db67d 2004-04-21 devnull }
285 a31db67d 2004-04-21 devnull 
286 a31db67d 2004-04-21 devnull int
287 a31db67d 2004-04-21 devnull fullisorune(char *str, int n)
288 a31db67d 2004-04-21 devnull {
289 a31db67d 2004-04-21 devnull 	int c;
290 a31db67d 2004-04-21 devnull 
291 a31db67d 2004-04-21 devnull 	if(n > 0) {
292 a31db67d 2004-04-21 devnull 		c = *(uchar*)str;
293 a31db67d 2004-04-21 devnull 		if(c < Char1)
294 a31db67d 2004-04-21 devnull 			return 1;
295 a31db67d 2004-04-21 devnull 		if(n > 1)
296 a31db67d 2004-04-21 devnull 			if(c < Char22 || n > 2)
297 a31db67d 2004-04-21 devnull 				return 1;
298 a31db67d 2004-04-21 devnull 	}
299 a31db67d 2004-04-21 devnull 	return 0;
300 a31db67d 2004-04-21 devnull }
301 a31db67d 2004-04-21 devnull 
302 a31db67d 2004-04-21 devnull enum
303 a31db67d 2004-04-21 devnull {
304 a31db67d 2004-04-21 devnull 	T1	= 0x00,
305 a31db67d 2004-04-21 devnull 	Tx	= 0x80,
306 a31db67d 2004-04-21 devnull 	T2	= 0xC0,
307 a31db67d 2004-04-21 devnull 	T3	= 0xE0,
308 a31db67d 2004-04-21 devnull 	T4	= 0xF0,
309 a31db67d 2004-04-21 devnull 	T5	= 0xF8,
310 a31db67d 2004-04-21 devnull 	T6	= 0xFC,
311 a31db67d 2004-04-21 devnull 
312 a31db67d 2004-04-21 devnull 	Bit1	= 7,
313 a31db67d 2004-04-21 devnull 	Bitx	= 6,
314 a31db67d 2004-04-21 devnull 	Bit2	= 5,
315 a31db67d 2004-04-21 devnull 	Bit3	= 4,
316 a31db67d 2004-04-21 devnull 	Bit4	= 3,
317 a31db67d 2004-04-21 devnull 	Bit5	= 2,
318 a31db67d 2004-04-21 devnull 	Bit6	= 2,
319 a31db67d 2004-04-21 devnull 
320 a31db67d 2004-04-21 devnull 	Mask1	= (1<<Bit1)-1,
321 a31db67d 2004-04-21 devnull 	Maskx	= (1<<Bitx)-1,
322 a31db67d 2004-04-21 devnull 	Mask2	= (1<<Bit2)-1,
323 a31db67d 2004-04-21 devnull 	Mask3	= (1<<Bit3)-1,
324 a31db67d 2004-04-21 devnull 	Mask4	= (1<<Bit4)-1,
325 a31db67d 2004-04-21 devnull 	Mask5	= (1<<Bit5)-1,
326 a31db67d 2004-04-21 devnull 	Mask6	= (1<<Bit6)-1,
327 a31db67d 2004-04-21 devnull 
328 a31db67d 2004-04-21 devnull 	Wchar1	= (1UL<<Bit1)-1,
329 a31db67d 2004-04-21 devnull 	Wchar2	= (1UL<<(Bit2+Bitx))-1,
330 a31db67d 2004-04-21 devnull 	Wchar3	= (1UL<<(Bit3+2*Bitx))-1,
331 a31db67d 2004-04-21 devnull 	Wchar4	= (1UL<<(Bit4+3*Bitx))-1,
332 a31db67d 2004-04-21 devnull 	Wchar5	= (1UL<<(Bit5+4*Bitx))-1
333 a31db67d 2004-04-21 devnull };
334 a31db67d 2004-04-21 devnull 
335 a31db67d 2004-04-21 devnull int
336 a31db67d 2004-04-21 devnull our_wctomb(char *s, unsigned long wc)
337 a31db67d 2004-04-21 devnull {
338 a31db67d 2004-04-21 devnull 	if(s == 0)
339 a31db67d 2004-04-21 devnull 		return 0;		/* no shift states */
340 a31db67d 2004-04-21 devnull 	if(wc & ~Wchar2) {
341 a31db67d 2004-04-21 devnull 		if(wc & ~Wchar4) {
342 a31db67d 2004-04-21 devnull 			if(wc & ~Wchar5) {
343 a31db67d 2004-04-21 devnull 				/* 6 bytes */
344 a31db67d 2004-04-21 devnull 				s[0] = T6 | ((wc >> 5*Bitx) & Mask6);
345 a31db67d 2004-04-21 devnull 				s[1] = Tx | ((wc >> 4*Bitx) & Maskx);
346 a31db67d 2004-04-21 devnull 				s[2] = Tx | ((wc >> 3*Bitx) & Maskx);
347 a31db67d 2004-04-21 devnull 				s[3] = Tx | ((wc >> 2*Bitx) & Maskx);
348 a31db67d 2004-04-21 devnull 				s[4] = Tx | ((wc >> 1*Bitx) & Maskx);
349 a31db67d 2004-04-21 devnull 				s[5] = Tx |  (wc & Maskx);
350 a31db67d 2004-04-21 devnull 				return 6;
351 a31db67d 2004-04-21 devnull 			}
352 a31db67d 2004-04-21 devnull 			/* 5 bytes */
353 a31db67d 2004-04-21 devnull 			s[0] = T5 |  (wc >> 4*Bitx);
354 a31db67d 2004-04-21 devnull 			s[1] = Tx | ((wc >> 3*Bitx) & Maskx);
355 a31db67d 2004-04-21 devnull 			s[2] = Tx | ((wc >> 2*Bitx) & Maskx);
356 a31db67d 2004-04-21 devnull 			s[3] = Tx | ((wc >> 1*Bitx) & Maskx);
357 a31db67d 2004-04-21 devnull 			s[4] = Tx |  (wc & Maskx);
358 a31db67d 2004-04-21 devnull 			return 5;
359 a31db67d 2004-04-21 devnull 		}
360 a31db67d 2004-04-21 devnull 		if(wc & ~Wchar3) {
361 a31db67d 2004-04-21 devnull 			/* 4 bytes */
362 a31db67d 2004-04-21 devnull 			s[0] = T4 |  (wc >> 3*Bitx);
363 a31db67d 2004-04-21 devnull 			s[1] = Tx | ((wc >> 2*Bitx) & Maskx);
364 a31db67d 2004-04-21 devnull 			s[2] = Tx | ((wc >> 1*Bitx) & Maskx);
365 a31db67d 2004-04-21 devnull 			s[3] = Tx |  (wc & Maskx);
366 a31db67d 2004-04-21 devnull 			return 4;
367 a31db67d 2004-04-21 devnull 		}
368 a31db67d 2004-04-21 devnull 		/* 3 bytes */
369 a31db67d 2004-04-21 devnull 		s[0] = T3 |  (wc >> 2*Bitx);
370 a31db67d 2004-04-21 devnull 		s[1] = Tx | ((wc >> 1*Bitx) & Maskx);
371 a31db67d 2004-04-21 devnull 		s[2] = Tx |  (wc & Maskx);
372 a31db67d 2004-04-21 devnull 		return 3;
373 a31db67d 2004-04-21 devnull 	}
374 a31db67d 2004-04-21 devnull 	if(wc & ~Wchar1) {
375 a31db67d 2004-04-21 devnull 		/* 2 bytes */
376 a31db67d 2004-04-21 devnull 		s[0] = T2 | (wc >> 1*Bitx);
377 a31db67d 2004-04-21 devnull 		s[1] = Tx | (wc & Maskx);
378 a31db67d 2004-04-21 devnull 		return 2;
379 a31db67d 2004-04-21 devnull 	}
380 a31db67d 2004-04-21 devnull 	/* 1 byte */
381 a31db67d 2004-04-21 devnull 	s[0] = T1 | wc;
382 a31db67d 2004-04-21 devnull 	return 1;
383 a31db67d 2004-04-21 devnull }
384 a31db67d 2004-04-21 devnull 
385 a31db67d 2004-04-21 devnull int
386 a31db67d 2004-04-21 devnull our_mbtowc(unsigned long *p, char *s, unsigned n)
387 a31db67d 2004-04-21 devnull {
388 a31db67d 2004-04-21 devnull 	uchar *us;
389 a31db67d 2004-04-21 devnull 	int c0, c1, c2, c3, c4, c5;
390 a31db67d 2004-04-21 devnull 	unsigned long wc;
391 a31db67d 2004-04-21 devnull 
392 a31db67d 2004-04-21 devnull 	if(s == 0)
393 a31db67d 2004-04-21 devnull 		return 0;		/* no shift states */
394 a31db67d 2004-04-21 devnull 
395 a31db67d 2004-04-21 devnull 	if(n < 1)
396 a31db67d 2004-04-21 devnull 		goto badlen;
397 a31db67d 2004-04-21 devnull 	us = (uchar*)s;
398 a31db67d 2004-04-21 devnull 	c0 = us[0];
399 a31db67d 2004-04-21 devnull 	if(c0 >= T3) {
400 a31db67d 2004-04-21 devnull 		if(n < 3)
401 a31db67d 2004-04-21 devnull 			goto badlen;
402 a31db67d 2004-04-21 devnull 		c1 = us[1] ^ Tx;
403 a31db67d 2004-04-21 devnull 		c2 = us[2] ^ Tx;
404 a31db67d 2004-04-21 devnull 		if((c1|c2) & T2)
405 a31db67d 2004-04-21 devnull 			goto bad;
406 a31db67d 2004-04-21 devnull 		if(c0 >= T5) {
407 a31db67d 2004-04-21 devnull 			if(n < 5)
408 a31db67d 2004-04-21 devnull 				goto badlen;
409 a31db67d 2004-04-21 devnull 			c3 = us[3] ^ Tx;
410 a31db67d 2004-04-21 devnull 			c4 = us[4] ^ Tx;
411 a31db67d 2004-04-21 devnull 			if((c3|c4) & T2)
412 a31db67d 2004-04-21 devnull 				goto bad;
413 a31db67d 2004-04-21 devnull 			if(c0 >= T6) {
414 a31db67d 2004-04-21 devnull 				/* 6 bytes */
415 a31db67d 2004-04-21 devnull 				if(n < 6)
416 a31db67d 2004-04-21 devnull 					goto badlen;
417 a31db67d 2004-04-21 devnull 				c5 = us[5] ^ Tx;
418 a31db67d 2004-04-21 devnull 				if(c5 & T2)
419 a31db67d 2004-04-21 devnull 					goto bad;
420 a31db67d 2004-04-21 devnull 				wc = ((((((((((c0 & Mask6) << Bitx) |
421 a31db67d 2004-04-21 devnull 					c1) << Bitx) | c2) << Bitx) |
422 a31db67d 2004-04-21 devnull 					c3) << Bitx) | c4) << Bitx) | c5;
423 a31db67d 2004-04-21 devnull 				if(wc <= Wchar5)
424 a31db67d 2004-04-21 devnull 					goto bad;
425 a31db67d 2004-04-21 devnull 				*p = wc;
426 a31db67d 2004-04-21 devnull 				return 6;
427 a31db67d 2004-04-21 devnull 			}
428 a31db67d 2004-04-21 devnull 			/* 5 bytes */
429 a31db67d 2004-04-21 devnull 			wc = ((((((((c0 & Mask5) << Bitx) |
430 a31db67d 2004-04-21 devnull 				c1) << Bitx) | c2) << Bitx) |
431 a31db67d 2004-04-21 devnull 				c3) << Bitx) | c4;
432 a31db67d 2004-04-21 devnull 			if(wc <= Wchar4)
433 a31db67d 2004-04-21 devnull 				goto bad;
434 a31db67d 2004-04-21 devnull 			*p = wc;
435 a31db67d 2004-04-21 devnull 			return 5;
436 a31db67d 2004-04-21 devnull 		}
437 a31db67d 2004-04-21 devnull 		if(c0 >= T4) {
438 a31db67d 2004-04-21 devnull 			/* 4 bytes */
439 a31db67d 2004-04-21 devnull 			if(n < 4)
440 a31db67d 2004-04-21 devnull 				goto badlen;
441 a31db67d 2004-04-21 devnull 			c3 = us[3] ^ Tx;
442 a31db67d 2004-04-21 devnull 			if(c3 & T2)
443 a31db67d 2004-04-21 devnull 				goto bad;
444 a31db67d 2004-04-21 devnull 			wc = ((((((c0 & Mask4) << Bitx) |
445 a31db67d 2004-04-21 devnull 				c1) << Bitx) | c2) << Bitx) |
446 a31db67d 2004-04-21 devnull 				c3;
447 a31db67d 2004-04-21 devnull 			if(wc <= Wchar3)
448 a31db67d 2004-04-21 devnull 				goto bad;
449 a31db67d 2004-04-21 devnull 			*p = wc;
450 a31db67d 2004-04-21 devnull 			return 4;
451 a31db67d 2004-04-21 devnull 		}
452 a31db67d 2004-04-21 devnull 		/* 3 bytes */
453 a31db67d 2004-04-21 devnull 		wc = ((((c0 & Mask3) << Bitx) |
454 a31db67d 2004-04-21 devnull 			c1) << Bitx) | c2;
455 a31db67d 2004-04-21 devnull 		if(wc <= Wchar2)
456 a31db67d 2004-04-21 devnull 			goto bad;
457 a31db67d 2004-04-21 devnull 		*p = wc;
458 a31db67d 2004-04-21 devnull 		return 3;
459 a31db67d 2004-04-21 devnull 	}
460 a31db67d 2004-04-21 devnull 	if(c0 >= T2) {
461 a31db67d 2004-04-21 devnull 		/* 2 bytes */
462 a31db67d 2004-04-21 devnull 		if(n < 2)
463 a31db67d 2004-04-21 devnull 			goto badlen;
464 a31db67d 2004-04-21 devnull 		c1 = us[1] ^ Tx;
465 a31db67d 2004-04-21 devnull 		if(c1 & T2)
466 a31db67d 2004-04-21 devnull 			goto bad;
467 a31db67d 2004-04-21 devnull 		wc = ((c0 & Mask2) << Bitx) |
468 a31db67d 2004-04-21 devnull 			c1;
469 a31db67d 2004-04-21 devnull 		if(wc <= Wchar1)
470 a31db67d 2004-04-21 devnull 			goto bad;
471 a31db67d 2004-04-21 devnull 		*p = wc;
472 a31db67d 2004-04-21 devnull 		return 2;
473 a31db67d 2004-04-21 devnull 	}
474 a31db67d 2004-04-21 devnull 	/* 1 byte */
475 a31db67d 2004-04-21 devnull 	if(c0 >= Tx)
476 a31db67d 2004-04-21 devnull 		goto bad;
477 a31db67d 2004-04-21 devnull 	*p = c0;
478 a31db67d 2004-04-21 devnull 	return 1;
479 a31db67d 2004-04-21 devnull 
480 a31db67d 2004-04-21 devnull bad:
481 a31db67d 2004-04-21 devnull 	errno = EILSEQ;
482 a31db67d 2004-04-21 devnull 	return -1;
483 a31db67d 2004-04-21 devnull badlen:
484 a31db67d 2004-04-21 devnull 	return -2;
485 a31db67d 2004-04-21 devnull }