2 b2cfc4e2 2003-09-30 devnull * The authors of this software are Rob Pike and Ken Thompson.
3 b2cfc4e2 2003-09-30 devnull * Copyright (c) 2002 by Lucent Technologies.
4 b2cfc4e2 2003-09-30 devnull * Permission to use, copy, modify, and distribute this software for any
5 b2cfc4e2 2003-09-30 devnull * purpose without fee is hereby granted, provided that this entire notice
6 b2cfc4e2 2003-09-30 devnull * is included in all copies of any software which is or includes a copy
7 b2cfc4e2 2003-09-30 devnull * or modification of this software and in all copies of the supporting
8 b2cfc4e2 2003-09-30 devnull * documentation for such software.
9 b2cfc4e2 2003-09-30 devnull * THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED
10 b2cfc4e2 2003-09-30 devnull * WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES MAKE ANY
11 b2cfc4e2 2003-09-30 devnull * REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY
12 b2cfc4e2 2003-09-30 devnull * OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE.
14 b2cfc4e2 2003-09-30 devnull #include <stdarg.h>
15 b2cfc4e2 2003-09-30 devnull #include <string.h>
16 b2cfc4e2 2003-09-30 devnull #include "utf.h"
17 b2cfc4e2 2003-09-30 devnull #include "utfdef.h"
21 b2cfc4e2 2003-09-30 devnull Bit1 = 7,
22 b2cfc4e2 2003-09-30 devnull Bitx = 6,
23 b2cfc4e2 2003-09-30 devnull Bit2 = 5,
24 b2cfc4e2 2003-09-30 devnull Bit3 = 4,
25 b2cfc4e2 2003-09-30 devnull Bit4 = 3,
27 b2cfc4e2 2003-09-30 devnull T1 = ((1<<(Bit1+1))-1) ^ 0xFF, /* 0000 0000 */
28 b2cfc4e2 2003-09-30 devnull Tx = ((1<<(Bitx+1))-1) ^ 0xFF, /* 1000 0000 */
29 b2cfc4e2 2003-09-30 devnull T2 = ((1<<(Bit2+1))-1) ^ 0xFF, /* 1100 0000 */
30 b2cfc4e2 2003-09-30 devnull T3 = ((1<<(Bit3+1))-1) ^ 0xFF, /* 1110 0000 */
31 b2cfc4e2 2003-09-30 devnull T4 = ((1<<(Bit4+1))-1) ^ 0xFF, /* 1111 0000 */
33 b2cfc4e2 2003-09-30 devnull Rune1 = (1<<(Bit1+0*Bitx))-1, /* 0000 0000 0111 1111 */
34 b2cfc4e2 2003-09-30 devnull Rune2 = (1<<(Bit2+1*Bitx))-1, /* 0000 0111 1111 1111 */
35 b2cfc4e2 2003-09-30 devnull Rune3 = (1<<(Bit3+2*Bitx))-1, /* 1111 1111 1111 1111 */
37 b2cfc4e2 2003-09-30 devnull Maskx = (1<<Bitx)-1, /* 0011 1111 */
38 b2cfc4e2 2003-09-30 devnull Testx = Maskx ^ 0xFF, /* 1100 0000 */
40 b2cfc4e2 2003-09-30 devnull Bad = Runeerror,
44 b2cfc4e2 2003-09-30 devnull chartorune(Rune *rune, char *str)
46 b2cfc4e2 2003-09-30 devnull int c, c1, c2;
50 b2cfc4e2 2003-09-30 devnull * one character sequence
51 b2cfc4e2 2003-09-30 devnull * 00000-0007F => T1
53 b2cfc4e2 2003-09-30 devnull c = *(uchar*)str;
54 b2cfc4e2 2003-09-30 devnull if(c < Tx) {
55 b2cfc4e2 2003-09-30 devnull *rune = c;
56 b2cfc4e2 2003-09-30 devnull return 1;
60 b2cfc4e2 2003-09-30 devnull * two character sequence
61 b2cfc4e2 2003-09-30 devnull * 0080-07FF => T2 Tx
63 b2cfc4e2 2003-09-30 devnull c1 = *(uchar*)(str+1) ^ Tx;
64 b2cfc4e2 2003-09-30 devnull if(c1 & Testx)
65 b2cfc4e2 2003-09-30 devnull goto bad;
66 b2cfc4e2 2003-09-30 devnull if(c < T3) {
67 b2cfc4e2 2003-09-30 devnull if(c < T2)
68 b2cfc4e2 2003-09-30 devnull goto bad;
69 b2cfc4e2 2003-09-30 devnull l = ((c << Bitx) | c1) & Rune2;
70 b2cfc4e2 2003-09-30 devnull if(l <= Rune1)
71 b2cfc4e2 2003-09-30 devnull goto bad;
72 b2cfc4e2 2003-09-30 devnull *rune = l;
73 b2cfc4e2 2003-09-30 devnull return 2;
77 b2cfc4e2 2003-09-30 devnull * three character sequence
78 b2cfc4e2 2003-09-30 devnull * 0800-FFFF => T3 Tx Tx
80 b2cfc4e2 2003-09-30 devnull c2 = *(uchar*)(str+2) ^ Tx;
81 b2cfc4e2 2003-09-30 devnull if(c2 & Testx)
82 b2cfc4e2 2003-09-30 devnull goto bad;
83 b2cfc4e2 2003-09-30 devnull if(c < T4) {
84 b2cfc4e2 2003-09-30 devnull l = ((((c << Bitx) | c1) << Bitx) | c2) & Rune3;
85 b2cfc4e2 2003-09-30 devnull if(l <= Rune2)
86 b2cfc4e2 2003-09-30 devnull goto bad;
87 b2cfc4e2 2003-09-30 devnull *rune = l;
88 b2cfc4e2 2003-09-30 devnull return 3;
92 b2cfc4e2 2003-09-30 devnull * bad decoding
95 b2cfc4e2 2003-09-30 devnull *rune = Bad;
96 b2cfc4e2 2003-09-30 devnull return 1;
100 b2cfc4e2 2003-09-30 devnull runetochar(char *str, Rune *rune)
105 b2cfc4e2 2003-09-30 devnull * one character sequence
106 b2cfc4e2 2003-09-30 devnull * 00000-0007F => 00-7F
108 b2cfc4e2 2003-09-30 devnull c = *rune;
109 b2cfc4e2 2003-09-30 devnull if(c <= Rune1) {
110 b2cfc4e2 2003-09-30 devnull str[0] = c;
111 b2cfc4e2 2003-09-30 devnull return 1;
115 b2cfc4e2 2003-09-30 devnull * two character sequence
116 b2cfc4e2 2003-09-30 devnull * 0080-07FF => T2 Tx
118 b2cfc4e2 2003-09-30 devnull if(c <= Rune2) {
119 b2cfc4e2 2003-09-30 devnull str[0] = T2 | (c >> 1*Bitx);
120 b2cfc4e2 2003-09-30 devnull str[1] = Tx | (c & Maskx);
121 b2cfc4e2 2003-09-30 devnull return 2;
125 b2cfc4e2 2003-09-30 devnull * three character sequence
126 b2cfc4e2 2003-09-30 devnull * 0800-FFFF => T3 Tx Tx
128 b2cfc4e2 2003-09-30 devnull str[0] = T3 | (c >> 2*Bitx);
129 b2cfc4e2 2003-09-30 devnull str[1] = Tx | ((c >> 1*Bitx) & Maskx);
130 b2cfc4e2 2003-09-30 devnull str[2] = Tx | (c & Maskx);
131 b2cfc4e2 2003-09-30 devnull return 3;
135 b2cfc4e2 2003-09-30 devnull runelen(long c)
137 b2cfc4e2 2003-09-30 devnull Rune rune;
138 b2cfc4e2 2003-09-30 devnull char str[10];
140 b2cfc4e2 2003-09-30 devnull rune = c;
141 b2cfc4e2 2003-09-30 devnull return runetochar(str, &rune);
145 b2cfc4e2 2003-09-30 devnull runenlen(Rune *r, int nrune)
147 b2cfc4e2 2003-09-30 devnull int nb, c;
150 b2cfc4e2 2003-09-30 devnull while(nrune--) {
151 b2cfc4e2 2003-09-30 devnull c = *r++;
152 b2cfc4e2 2003-09-30 devnull if(c <= Rune1)
155 b2cfc4e2 2003-09-30 devnull if(c <= Rune2)
156 b2cfc4e2 2003-09-30 devnull nb += 2;
158 b2cfc4e2 2003-09-30 devnull nb += 3;
160 b2cfc4e2 2003-09-30 devnull return nb;
164 b2cfc4e2 2003-09-30 devnull fullrune(char *str, int n)
168 b2cfc4e2 2003-09-30 devnull if(n > 0) {
169 b2cfc4e2 2003-09-30 devnull c = *(uchar*)str;
170 b2cfc4e2 2003-09-30 devnull if(c < Tx)
171 b2cfc4e2 2003-09-30 devnull return 1;
172 b2cfc4e2 2003-09-30 devnull if(n > 1)
173 b2cfc4e2 2003-09-30 devnull if(c < T3 || n > 2)
174 b2cfc4e2 2003-09-30 devnull return 1;
176 b2cfc4e2 2003-09-30 devnull return 0;