1 5cedca1b 2004-05-15 devnull #include <stdlib.h>
  2 5cedca1b 2004-05-15 devnull 
  3 5cedca1b 2004-05-15 devnull /*
  4 5cedca1b 2004-05-15 devnull  * Use the FSS-UTF transformation proposed by posix.
  5 5cedca1b 2004-05-15 devnull  *	We define 7 byte types:
  6 5cedca1b 2004-05-15 devnull  *	T0	0xxxxxxx	7 free bits
  7 5cedca1b 2004-05-15 devnull  *	Tx	10xxxxxx	6 free bits
  8 5cedca1b 2004-05-15 devnull  *	T1	110xxxxx	5 free bits
  9 5cedca1b 2004-05-15 devnull  *	T2	1110xxxx	4 free bits
 10 5cedca1b 2004-05-15 devnull  *
 11 5cedca1b 2004-05-15 devnull  *	Encoding is as follows.
 12 5cedca1b 2004-05-15 devnull  *	From hex	Thru hex	Sequence		Bits
 13 5cedca1b 2004-05-15 devnull  *	00000000	0000007F	T0			7
 14 5cedca1b 2004-05-15 devnull  *	00000080	000007FF	T1 Tx			11
 15 5cedca1b 2004-05-15 devnull  *	00000800	0000FFFF	T2 Tx Tx		16
 16 5cedca1b 2004-05-15 devnull  */
 17 5cedca1b 2004-05-15 devnull 
 18 5cedca1b 2004-05-15 devnull int
 19 5cedca1b 2004-05-15 devnull mblen(const char *s, size_t n)
 20 5cedca1b 2004-05-15 devnull {
 21 5cedca1b 2004-05-15 devnull 
 22 5cedca1b 2004-05-15 devnull 	return mbtowc(0, s, n);
 23 5cedca1b 2004-05-15 devnull }
 24 5cedca1b 2004-05-15 devnull 
 25 5cedca1b 2004-05-15 devnull int
 26 5cedca1b 2004-05-15 devnull mbtowc(wchar_t *pwc, const char *s, size_t n)
 27 5cedca1b 2004-05-15 devnull {
 28 5cedca1b 2004-05-15 devnull 	int c, c1, c2;
 29 5cedca1b 2004-05-15 devnull 	long l;
 30 5cedca1b 2004-05-15 devnull 
 31 5cedca1b 2004-05-15 devnull 	if(!s)
 32 5cedca1b 2004-05-15 devnull 		return 0;
 33 5cedca1b 2004-05-15 devnull 
 34 5cedca1b 2004-05-15 devnull 	if(n < 1)
 35 5cedca1b 2004-05-15 devnull 		goto bad;
 36 5cedca1b 2004-05-15 devnull 	c = s[0] & 0xff;
 37 5cedca1b 2004-05-15 devnull 	if((c & 0x80) == 0x00) {
 38 5cedca1b 2004-05-15 devnull 		if(pwc)
 39 5cedca1b 2004-05-15 devnull 			*pwc = c;
 40 5cedca1b 2004-05-15 devnull 		if(c == 0)
 41 5cedca1b 2004-05-15 devnull 			return 0;
 42 5cedca1b 2004-05-15 devnull 		return 1;
 43 5cedca1b 2004-05-15 devnull 	}
 44 5cedca1b 2004-05-15 devnull 
 45 5cedca1b 2004-05-15 devnull 	if(n < 2)
 46 5cedca1b 2004-05-15 devnull 		goto bad;
 47 5cedca1b 2004-05-15 devnull 	c1 = (s[1] ^ 0x80) & 0xff;
 48 5cedca1b 2004-05-15 devnull 	if((c1 & 0xC0) != 0x00)
 49 5cedca1b 2004-05-15 devnull 		goto bad;
 50 5cedca1b 2004-05-15 devnull 	if((c & 0xE0) == 0xC0) {
 51 5cedca1b 2004-05-15 devnull 		l = ((c << 6) | c1) & 0x7FF;
 52 5cedca1b 2004-05-15 devnull 		if(l < 0x080)
 53 5cedca1b 2004-05-15 devnull 			goto bad;
 54 5cedca1b 2004-05-15 devnull 		if(pwc)
 55 5cedca1b 2004-05-15 devnull 			*pwc = l;
 56 5cedca1b 2004-05-15 devnull 		return 2;
 57 5cedca1b 2004-05-15 devnull 	}
 58 5cedca1b 2004-05-15 devnull 
 59 5cedca1b 2004-05-15 devnull 	if(n < 3)
 60 5cedca1b 2004-05-15 devnull 		goto bad;
 61 5cedca1b 2004-05-15 devnull 	c2 = (s[2] ^ 0x80) & 0xff;
 62 5cedca1b 2004-05-15 devnull 	if((c2 & 0xC0) != 0x00)
 63 5cedca1b 2004-05-15 devnull 		goto bad;
 64 5cedca1b 2004-05-15 devnull 	if((c & 0xF0) == 0xE0) {
 65 5cedca1b 2004-05-15 devnull 		l = ((((c << 6) | c1) << 6) | c2) & 0xFFFF;
 66 5cedca1b 2004-05-15 devnull 		if(l < 0x0800)
 67 5cedca1b 2004-05-15 devnull 			goto bad;
 68 5cedca1b 2004-05-15 devnull 		if(pwc)
 69 5cedca1b 2004-05-15 devnull 			*pwc = l;
 70 5cedca1b 2004-05-15 devnull 		return 3;
 71 5cedca1b 2004-05-15 devnull 	}
 72 5cedca1b 2004-05-15 devnull 
 73 5cedca1b 2004-05-15 devnull 	/*
 74 5cedca1b 2004-05-15 devnull 	 * bad decoding
 75 5cedca1b 2004-05-15 devnull 	 */
 76 5cedca1b 2004-05-15 devnull bad:
 77 5cedca1b 2004-05-15 devnull 	return -1;
 78 5cedca1b 2004-05-15 devnull 
 79 5cedca1b 2004-05-15 devnull }
 80 5cedca1b 2004-05-15 devnull 
 81 5cedca1b 2004-05-15 devnull int
 82 5cedca1b 2004-05-15 devnull wctomb(char *s, wchar_t wchar)
 83 5cedca1b 2004-05-15 devnull {
 84 5cedca1b 2004-05-15 devnull 	long c;
 85 5cedca1b 2004-05-15 devnull 
 86 5cedca1b 2004-05-15 devnull 	if(!s)
 87 5cedca1b 2004-05-15 devnull 		return 0;
 88 5cedca1b 2004-05-15 devnull 
 89 5cedca1b 2004-05-15 devnull 	c = wchar & 0xFFFF;
 90 5cedca1b 2004-05-15 devnull 	if(c < 0x80) {
 91 5cedca1b 2004-05-15 devnull 		s[0] = c;
 92 5cedca1b 2004-05-15 devnull 		return 1;
 93 5cedca1b 2004-05-15 devnull 	}
 94 5cedca1b 2004-05-15 devnull 
 95 5cedca1b 2004-05-15 devnull 	if(c < 0x800) {
 96 5cedca1b 2004-05-15 devnull 		s[0] = 0xC0 | (c >> 6);
 97 5cedca1b 2004-05-15 devnull 		s[1] = 0x80 | (c & 0x3F);
 98 5cedca1b 2004-05-15 devnull 		return 2;
 99 5cedca1b 2004-05-15 devnull 	}
100 5cedca1b 2004-05-15 devnull 
101 5cedca1b 2004-05-15 devnull 	s[0] = 0xE0 |  (c >> 12);
102 5cedca1b 2004-05-15 devnull 	s[1] = 0x80 | ((c >> 6) & 0x3F);
103 5cedca1b 2004-05-15 devnull 	s[2] = 0x80 |  (c & 0x3F);
104 5cedca1b 2004-05-15 devnull 	return 3;
105 5cedca1b 2004-05-15 devnull }
106 5cedca1b 2004-05-15 devnull 
107 5cedca1b 2004-05-15 devnull size_t
108 5cedca1b 2004-05-15 devnull mbstowcs(wchar_t *pwcs, const char *s, size_t n)
109 5cedca1b 2004-05-15 devnull {
110 5cedca1b 2004-05-15 devnull 	int i, d, c;
111 5cedca1b 2004-05-15 devnull 
112 5cedca1b 2004-05-15 devnull 	for(i=0; i < n; i++) {
113 5cedca1b 2004-05-15 devnull 		c = *s & 0xff;
114 5cedca1b 2004-05-15 devnull 		if(c < 0x80) {
115 5cedca1b 2004-05-15 devnull 			*pwcs = c;
116 5cedca1b 2004-05-15 devnull 			if(c == 0)
117 5cedca1b 2004-05-15 devnull 				break;
118 5cedca1b 2004-05-15 devnull 			s++;
119 5cedca1b 2004-05-15 devnull 		} else {
120 5cedca1b 2004-05-15 devnull 			d = mbtowc(pwcs, s, 3);
121 5cedca1b 2004-05-15 devnull 			if(d <= 0)
122 5cedca1b 2004-05-15 devnull 				return (size_t)((d<0) ? -1 : i);
123 5cedca1b 2004-05-15 devnull 			s += d;
124 5cedca1b 2004-05-15 devnull 		}
125 5cedca1b 2004-05-15 devnull 		pwcs++;
126 5cedca1b 2004-05-15 devnull 	}
127 5cedca1b 2004-05-15 devnull 	return i;
128 5cedca1b 2004-05-15 devnull }
129 5cedca1b 2004-05-15 devnull 
130 5cedca1b 2004-05-15 devnull size_t
131 5cedca1b 2004-05-15 devnull wcstombs(char *s, const wchar_t *pwcs, size_t n)
132 5cedca1b 2004-05-15 devnull {
133 1f72bc47 2004-05-17 devnull 	int d;
134 5cedca1b 2004-05-15 devnull 	long c;
135 5cedca1b 2004-05-15 devnull 	char *p, *pe;
136 5cedca1b 2004-05-15 devnull 	char buf[3];
137 5cedca1b 2004-05-15 devnull 
138 5cedca1b 2004-05-15 devnull 	p = s;
139 5cedca1b 2004-05-15 devnull 	pe = p+n-3;
140 5cedca1b 2004-05-15 devnull 	while(p < pe) {
141 5cedca1b 2004-05-15 devnull 		c = *pwcs++;
142 5cedca1b 2004-05-15 devnull 		if(c < 0x80)
143 5cedca1b 2004-05-15 devnull 			*p++ = c;
144 5cedca1b 2004-05-15 devnull 		else
145 5cedca1b 2004-05-15 devnull 			p += wctomb(p, c);
146 5cedca1b 2004-05-15 devnull 		if(c == 0)
147 5cedca1b 2004-05-15 devnull 			return p-s;
148 5cedca1b 2004-05-15 devnull 	}
149 5cedca1b 2004-05-15 devnull 	while(p < pe+3) {
150 5cedca1b 2004-05-15 devnull 		c = *pwcs++;
151 5cedca1b 2004-05-15 devnull 		d = wctomb(buf, c);
152 5cedca1b 2004-05-15 devnull 		if(p+d <= pe+3) {
153 5cedca1b 2004-05-15 devnull 			*p++ = buf[0];
154 5cedca1b 2004-05-15 devnull 			if(d > 1) {
155 0cadb430 2009-09-11 russcox 				*p++ = buf[1];
156 5cedca1b 2004-05-15 devnull 				if(d > 2)
157 0cadb430 2009-09-11 russcox 					*p++ = buf[2];
158 5cedca1b 2004-05-15 devnull 			}
159 5cedca1b 2004-05-15 devnull 		}
160 5cedca1b 2004-05-15 devnull 		if(c == 0)
161 5cedca1b 2004-05-15 devnull 			break;
162 5cedca1b 2004-05-15 devnull 	}
163 5cedca1b 2004-05-15 devnull 	return p-s;
164 5cedca1b 2004-05-15 devnull }