2 91c13e54 2004-02-29 devnull * The authors of this software are Rob Pike and Ken Thompson.
3 91c13e54 2004-02-29 devnull * Copyright (c) 2002 by Lucent Technologies.
4 91c13e54 2004-02-29 devnull * Permission to use, copy, modify, and distribute this software for any
5 91c13e54 2004-02-29 devnull * purpose without fee is hereby granted, provided that this entire notice
6 91c13e54 2004-02-29 devnull * is included in all copies of any software which is or includes a copy
7 91c13e54 2004-02-29 devnull * or modification of this software and in all copies of the supporting
8 91c13e54 2004-02-29 devnull * documentation for such software.
9 91c13e54 2004-02-29 devnull * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED
10 37488654 2004-12-29 devnull * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE
11 37488654 2004-12-29 devnull * ANY REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY
12 91c13e54 2004-02-29 devnull * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE.
14 91c13e54 2004-02-29 devnull #include <stdarg.h>
15 91c13e54 2004-02-29 devnull #include <string.h>
16 e5aa96ac 2004-12-26 devnull #include "plan9.h"
17 91c13e54 2004-02-29 devnull #include "utf.h"
21 91c13e54 2004-02-29 devnull Bit1 = 7,
22 91c13e54 2004-02-29 devnull Bitx = 6,
23 91c13e54 2004-02-29 devnull Bit2 = 5,
24 91c13e54 2004-02-29 devnull Bit3 = 4,
25 91c13e54 2004-02-29 devnull Bit4 = 3,
27 91c13e54 2004-02-29 devnull T1 = ((1<<(Bit1+1))-1) ^ 0xFF, /* 0000 0000 */
28 91c13e54 2004-02-29 devnull Tx = ((1<<(Bitx+1))-1) ^ 0xFF, /* 1000 0000 */
29 91c13e54 2004-02-29 devnull T2 = ((1<<(Bit2+1))-1) ^ 0xFF, /* 1100 0000 */
30 91c13e54 2004-02-29 devnull T3 = ((1<<(Bit3+1))-1) ^ 0xFF, /* 1110 0000 */
31 91c13e54 2004-02-29 devnull T4 = ((1<<(Bit4+1))-1) ^ 0xFF, /* 1111 0000 */
33 91c13e54 2004-02-29 devnull Rune1 = (1<<(Bit1+0*Bitx))-1, /* 0000 0000 0111 1111 */
34 91c13e54 2004-02-29 devnull Rune2 = (1<<(Bit2+1*Bitx))-1, /* 0000 0111 1111 1111 */
35 91c13e54 2004-02-29 devnull Rune3 = (1<<(Bit3+2*Bitx))-1, /* 1111 1111 1111 1111 */
37 91c13e54 2004-02-29 devnull Maskx = (1<<Bitx)-1, /* 0011 1111 */
38 91c13e54 2004-02-29 devnull Testx = Maskx ^ 0xFF, /* 1100 0000 */
40 cbeb0b26 2006-04-01 devnull Bad = Runeerror
44 91c13e54 2004-02-29 devnull chartorune(Rune *rune, char *str)
46 91c13e54 2004-02-29 devnull int c, c1, c2;
50 91c13e54 2004-02-29 devnull * one character sequence
51 91c13e54 2004-02-29 devnull * 00000-0007F => T1
53 91c13e54 2004-02-29 devnull c = *(uchar*)str;
54 91c13e54 2004-02-29 devnull if(c < Tx) {
55 91c13e54 2004-02-29 devnull *rune = c;
56 91c13e54 2004-02-29 devnull return 1;
60 91c13e54 2004-02-29 devnull * two character sequence
61 91c13e54 2004-02-29 devnull * 0080-07FF => T2 Tx
63 91c13e54 2004-02-29 devnull c1 = *(uchar*)(str+1) ^ Tx;
64 91c13e54 2004-02-29 devnull if(c1 & Testx)
65 91c13e54 2004-02-29 devnull goto bad;
66 91c13e54 2004-02-29 devnull if(c < T3) {
67 91c13e54 2004-02-29 devnull if(c < T2)
68 91c13e54 2004-02-29 devnull goto bad;
69 91c13e54 2004-02-29 devnull l = ((c << Bitx) | c1) & Rune2;
70 91c13e54 2004-02-29 devnull if(l <= Rune1)
71 91c13e54 2004-02-29 devnull goto bad;
72 91c13e54 2004-02-29 devnull *rune = l;
73 91c13e54 2004-02-29 devnull return 2;
77 91c13e54 2004-02-29 devnull * three character sequence
78 91c13e54 2004-02-29 devnull * 0800-FFFF => T3 Tx Tx
80 91c13e54 2004-02-29 devnull c2 = *(uchar*)(str+2) ^ Tx;
81 91c13e54 2004-02-29 devnull if(c2 & Testx)
82 91c13e54 2004-02-29 devnull goto bad;
83 91c13e54 2004-02-29 devnull if(c < T4) {
84 91c13e54 2004-02-29 devnull l = ((((c << Bitx) | c1) << Bitx) | c2) & Rune3;
85 91c13e54 2004-02-29 devnull if(l <= Rune2)
86 91c13e54 2004-02-29 devnull goto bad;
87 91c13e54 2004-02-29 devnull *rune = l;
88 91c13e54 2004-02-29 devnull return 3;
92 91c13e54 2004-02-29 devnull * bad decoding
95 91c13e54 2004-02-29 devnull *rune = Bad;
96 91c13e54 2004-02-29 devnull return 1;
100 91c13e54 2004-02-29 devnull runetochar(char *str, Rune *rune)
105 91c13e54 2004-02-29 devnull * one character sequence
106 91c13e54 2004-02-29 devnull * 00000-0007F => 00-7F
108 91c13e54 2004-02-29 devnull c = *rune;
109 91c13e54 2004-02-29 devnull if(c <= Rune1) {
110 91c13e54 2004-02-29 devnull str[0] = c;
111 91c13e54 2004-02-29 devnull return 1;
115 91c13e54 2004-02-29 devnull * two character sequence
116 91c13e54 2004-02-29 devnull * 0080-07FF => T2 Tx
118 91c13e54 2004-02-29 devnull if(c <= Rune2) {
119 91c13e54 2004-02-29 devnull str[0] = T2 | (c >> 1*Bitx);
120 91c13e54 2004-02-29 devnull str[1] = Tx | (c & Maskx);
121 91c13e54 2004-02-29 devnull return 2;
125 91c13e54 2004-02-29 devnull * three character sequence
126 91c13e54 2004-02-29 devnull * 0800-FFFF => T3 Tx Tx
128 91c13e54 2004-02-29 devnull str[0] = T3 | (c >> 2*Bitx);
129 91c13e54 2004-02-29 devnull str[1] = Tx | ((c >> 1*Bitx) & Maskx);
130 91c13e54 2004-02-29 devnull str[2] = Tx | (c & Maskx);
131 91c13e54 2004-02-29 devnull return 3;
135 91c13e54 2004-02-29 devnull runelen(long c)
137 91c13e54 2004-02-29 devnull Rune rune;
138 91c13e54 2004-02-29 devnull char str[10];
140 91c13e54 2004-02-29 devnull rune = c;
141 91c13e54 2004-02-29 devnull return runetochar(str, &rune);
145 91c13e54 2004-02-29 devnull runenlen(Rune *r, int nrune)
147 91c13e54 2004-02-29 devnull int nb, c;
150 91c13e54 2004-02-29 devnull while(nrune--) {
151 91c13e54 2004-02-29 devnull c = *r++;
152 91c13e54 2004-02-29 devnull if(c <= Rune1)
155 91c13e54 2004-02-29 devnull if(c <= Rune2)
156 91c13e54 2004-02-29 devnull nb += 2;
158 91c13e54 2004-02-29 devnull nb += 3;
160 91c13e54 2004-02-29 devnull return nb;
164 91c13e54 2004-02-29 devnull fullrune(char *str, int n)
168 91c13e54 2004-02-29 devnull if(n > 0) {
169 91c13e54 2004-02-29 devnull c = *(uchar*)str;
170 91c13e54 2004-02-29 devnull if(c < Tx)
171 91c13e54 2004-02-29 devnull return 1;
172 91c13e54 2004-02-29 devnull if(n > 1)
173 91c13e54 2004-02-29 devnull if(c < T3 || n > 2)
174 91c13e54 2004-02-29 devnull return 1;
176 91c13e54 2004-02-29 devnull return 0;