1 #include	"rune.h"
2 
3 enum
4 {
5 	Bit1	= 7,
6 	Bitx	= 6,
7 	Bit2	= 5,
8 	Bit3	= 4,
9 	Bit4	= 3,
10 
11 	T1	= ((1<<(Bit1+1))-1) ^ 0xFF,	/* 0000 0000 */
12 	Tx	= ((1<<(Bitx+1))-1) ^ 0xFF,	/* 1000 0000 */
13 	T2	= ((1<<(Bit2+1))-1) ^ 0xFF,	/* 1100 0000 */
14 	T3	= ((1<<(Bit3+1))-1) ^ 0xFF,	/* 1110 0000 */
15 	T4	= ((1<<(Bit4+1))-1) ^ 0xFF,	/* 1111 0000 */
16 
17 	Rune1	= (1<<(Bit1+0*Bitx))-1,		/* 0000 0000 0111 1111 */
18 	Rune2	= (1<<(Bit2+1*Bitx))-1,		/* 0000 0111 1111 1111 */
19 	Rune3	= (1<<(Bit3+2*Bitx))-1,		/* 1111 1111 1111 1111 */
20 
21 	Maskx	= (1<<Bitx)-1,			/* 0011 1111 */
22 	Testx	= Maskx ^ 0xFF,			/* 1100 0000 */
23 
24 	Bad	= Runeerror
25 };
26 
27 int
28 chartorune(Rune *rune, char *str)
29 {
30 	int c, c1, c2;
31 	long l;
32 
33 	/*
34 	 * one character sequence
35 	 *	00000-0007F => T1
36 	 */
37 	c = *(unsigned char*)str;
38 	if(c < Tx) {
39 		*rune = c;
40 		return 1;
41 	}
42 
43 	/*
44 	 * two character sequence
45 	 *	0080-07FF => T2 Tx
46 	 */
47 	c1 = *(unsigned char*)(str+1) ^ Tx;
48 	if(c1 & Testx)
49 		goto bad;
50 	if(c < T3) {
51 		if(c < T2)
52 			goto bad;
53 		l = ((c << Bitx) | c1) & Rune2;
54 		if(l <= Rune1)
55 			goto bad;
56 		*rune = l;
57 		return 2;
58 	}
59 
60 	/*
61 	 * three character sequence
62 	 *	0800-FFFF => T3 Tx Tx
63 	 */
64 	c2 = *(unsigned char*)(str+2) ^ Tx;
65 	if(c2 & Testx)
66 		goto bad;
67 	if(c < T4) {
68 		l = ((((c << Bitx) | c1) << Bitx) | c2) & Rune3;
69 		if(l <= Rune2)
70 			goto bad;
71 		*rune = l;
72 		return 3;
73 	}
74 
75 	/*
76 	 * bad decoding
77 	 */
78 bad:
79 	*rune = Bad;
80 	return 1;
81 }
82 
83 int
84 runetochar(char *str, Rune *rune)
85 {
86 	long c;
87 
88 	/*
89 	 * one character sequence
90 	 *	00000-0007F => 00-7F
91 	 */
92 	c = *rune;
93 	if(c <= Rune1) {
94 		str[0] = c;
95 		return 1;
96 	}
97 
98 	/*
99 	 * two character sequence
100 	 *	0080-07FF => T2 Tx
101 	 */
102 	if(c <= Rune2) {
103 		str[0] = T2 | (c >> 1*Bitx);
104 		str[1] = Tx | (c & Maskx);
105 		return 2;
106 	}
107 
108 	/*
109 	 * three character sequence
110 	 *	0800-FFFF => T3 Tx Tx
111 	 */
112 	str[0] = T3 |  (c >> 2*Bitx);
113 	str[1] = Tx | ((c >> 1*Bitx) & Maskx);
114 	str[2] = Tx |  (c & Maskx);
115 	return 3;
116 }
117 
118 int
119 runelen(long c)
120 {
121 	Rune rune;
122 	char str[10];
123 
124 	rune = c;
125 	return runetochar(str, &rune);
126 }
127 
128 int
129 fullrune(char *str, int n)
130 {
131 	int c;
132 
133 	if(n > 0) {
134 		c = *(unsigned char*)str;
135 		if(c < Tx)
136 			return 1;
137 		if(n > 1)
138 			if(c < T3 || n > 2)
139 				return 1;
140 	}
141 	return 0;
142 }