137da2899SCharles.Forsyth #include "lib9.h"
237da2899SCharles.Forsyth
3*3cd4f1d1SCharles Forsyth #define Bit(i) (7-(i))
4*3cd4f1d1SCharles Forsyth /* N 0's preceded by i 1's, T(Bit(2)) is 1100 0000 */
5*3cd4f1d1SCharles Forsyth #define T(i) (((1 << (Bit(i)+1))-1) ^ 0xFF)
6*3cd4f1d1SCharles Forsyth /* 0000 0000 0000 0111 1111 1111 */
7*3cd4f1d1SCharles Forsyth #define RuneX(i) ((1 << (Bit(i) + ((i)-1)*Bitx))-1)
8*3cd4f1d1SCharles Forsyth
937da2899SCharles.Forsyth enum
1037da2899SCharles.Forsyth {
11*3cd4f1d1SCharles Forsyth Bitx = Bit(1),
1237da2899SCharles.Forsyth
13*3cd4f1d1SCharles Forsyth Tx = T(1), /* 1000 0000 */
14*3cd4f1d1SCharles Forsyth Rune1 = (1<<(Bit(0)+0*Bitx))-1, /* 0000 0000 0000 0000 0111 1111 */
1537da2899SCharles.Forsyth
1637da2899SCharles.Forsyth Maskx = (1<<Bitx)-1, /* 0011 1111 */
1737da2899SCharles.Forsyth Testx = Maskx ^ 0xFF, /* 1100 0000 */
1837da2899SCharles.Forsyth
19*3cd4f1d1SCharles Forsyth SurrogateMin = 0xD800,
20*3cd4f1d1SCharles Forsyth SurrogateMax = 0xDFFF,
21*3cd4f1d1SCharles Forsyth
2237da2899SCharles.Forsyth Bad = Runeerror,
2337da2899SCharles.Forsyth };
2437da2899SCharles.Forsyth
2537da2899SCharles.Forsyth int
chartorune(Rune * rune,char * str)2637da2899SCharles.Forsyth chartorune(Rune *rune, char *str)
2737da2899SCharles.Forsyth {
28*3cd4f1d1SCharles Forsyth int c[UTFmax], i;
29*3cd4f1d1SCharles Forsyth Rune l;
3037da2899SCharles.Forsyth
3137da2899SCharles.Forsyth /*
32*3cd4f1d1SCharles Forsyth * N character sequence
3337da2899SCharles.Forsyth * 00000-0007F => T1
34*3cd4f1d1SCharles Forsyth * 00080-007FF => T2 Tx
35*3cd4f1d1SCharles Forsyth * 00800-0FFFF => T3 Tx Tx
36*3cd4f1d1SCharles Forsyth * 10000-10FFFF => T4 Tx Tx Tx
3737da2899SCharles.Forsyth */
38*3cd4f1d1SCharles Forsyth
39*3cd4f1d1SCharles Forsyth c[0] = *(uchar*)(str);
40*3cd4f1d1SCharles Forsyth if(c[0] < Tx){
41*3cd4f1d1SCharles Forsyth *rune = c[0];
4237da2899SCharles.Forsyth return 1;
4337da2899SCharles.Forsyth }
44*3cd4f1d1SCharles Forsyth l = c[0];
4537da2899SCharles.Forsyth
46*3cd4f1d1SCharles Forsyth for(i = 1; i < UTFmax; i++) {
47*3cd4f1d1SCharles Forsyth c[i] = *(uchar*)(str+i);
48*3cd4f1d1SCharles Forsyth c[i] ^= Tx;
49*3cd4f1d1SCharles Forsyth if(c[i] & Testx)
5037da2899SCharles.Forsyth goto bad;
51*3cd4f1d1SCharles Forsyth l = (l << Bitx) | c[i];
52*3cd4f1d1SCharles Forsyth if(c[0] < T(i + 2)) {
53*3cd4f1d1SCharles Forsyth l &= RuneX(i + 1);
54*3cd4f1d1SCharles Forsyth if(i == 1) {
55*3cd4f1d1SCharles Forsyth if(c[0] < T(2) || l <= Rune1)
5637da2899SCharles.Forsyth goto bad;
57*3cd4f1d1SCharles Forsyth } else if(l <= RuneX(i) || l > Runemax)
58*3cd4f1d1SCharles Forsyth goto bad;
59*3cd4f1d1SCharles Forsyth if (i == 2 && SurrogateMin <= l && l <= SurrogateMax)
6037da2899SCharles.Forsyth goto bad;
6137da2899SCharles.Forsyth *rune = l;
62*3cd4f1d1SCharles Forsyth return i + 1;
6337da2899SCharles.Forsyth }
6437da2899SCharles.Forsyth }
6537da2899SCharles.Forsyth
6637da2899SCharles.Forsyth /*
6737da2899SCharles.Forsyth * bad decoding
6837da2899SCharles.Forsyth */
6937da2899SCharles.Forsyth bad:
7037da2899SCharles.Forsyth *rune = Bad;
7137da2899SCharles.Forsyth return 1;
7237da2899SCharles.Forsyth }
7337da2899SCharles.Forsyth
7437da2899SCharles.Forsyth int
runetochar(char * str,Rune * rune)7537da2899SCharles.Forsyth runetochar(char *str, Rune *rune)
7637da2899SCharles.Forsyth {
77*3cd4f1d1SCharles Forsyth int i, j;
78*3cd4f1d1SCharles Forsyth Rune c;
7937da2899SCharles.Forsyth
8037da2899SCharles.Forsyth c = *rune;
8137da2899SCharles.Forsyth if(c <= Rune1) {
8237da2899SCharles.Forsyth str[0] = c;
8337da2899SCharles.Forsyth return 1;
8437da2899SCharles.Forsyth }
8537da2899SCharles.Forsyth
8637da2899SCharles.Forsyth /*
87*3cd4f1d1SCharles Forsyth * one character sequence
88*3cd4f1d1SCharles Forsyth * 00000-0007F => 00-7F
8937da2899SCharles.Forsyth * two character sequence
9037da2899SCharles.Forsyth * 0080-07FF => T2 Tx
9137da2899SCharles.Forsyth * three character sequence
9237da2899SCharles.Forsyth * 0800-FFFF => T3 Tx Tx
93*3cd4f1d1SCharles Forsyth * four character sequence (21-bit value)
94*3cd4f1d1SCharles Forsyth * 10000-1FFFFF => T4 Tx Tx Tx
95*3cd4f1d1SCharles Forsyth * If the Rune is out of range or a surrogate half,
96*3cd4f1d1SCharles Forsyth * convert it to the error rune.
97*3cd4f1d1SCharles Forsyth * Do this test when i==3 because the error rune encodes to three bytes.
98*3cd4f1d1SCharles Forsyth * Doing it earlier would duplicate work, since an out of range
99*3cd4f1d1SCharles Forsyth * Rune wouldn't have fit in one or two bytes.
10037da2899SCharles.Forsyth */
101*3cd4f1d1SCharles Forsyth for(i = 2; i < UTFmax + 1; i++){
102*3cd4f1d1SCharles Forsyth if(i == 3){
103*3cd4f1d1SCharles Forsyth if(c > Runemax)
104*3cd4f1d1SCharles Forsyth c = Runeerror;
105*3cd4f1d1SCharles Forsyth if(SurrogateMin <= c && c <= SurrogateMax)
106*3cd4f1d1SCharles Forsyth c = Runeerror;
107*3cd4f1d1SCharles Forsyth }
108*3cd4f1d1SCharles Forsyth if (c <= RuneX(i) || i == UTFmax ) {
109*3cd4f1d1SCharles Forsyth str[0] = T(i) | (c >> (i - 1)*Bitx);
110*3cd4f1d1SCharles Forsyth for(j = 1; j < i; j++)
111*3cd4f1d1SCharles Forsyth str[j] = Tx | ((c >> (i - j - 1)*Bitx) & Maskx);
112*3cd4f1d1SCharles Forsyth return i;
113*3cd4f1d1SCharles Forsyth }
114*3cd4f1d1SCharles Forsyth }
115*3cd4f1d1SCharles Forsyth return UTFmax;
11637da2899SCharles.Forsyth }
11737da2899SCharles.Forsyth
11837da2899SCharles.Forsyth int
runelen(long c)11937da2899SCharles.Forsyth runelen(long c)
12037da2899SCharles.Forsyth {
121*3cd4f1d1SCharles Forsyth Rune rune;
122*3cd4f1d1SCharles Forsyth char str[10];
123*3cd4f1d1SCharles Forsyth
124*3cd4f1d1SCharles Forsyth rune = c;
125*3cd4f1d1SCharles Forsyth return runetochar(str, &rune);
12637da2899SCharles.Forsyth }
12737da2899SCharles.Forsyth
12837da2899SCharles.Forsyth int
runenlen(Rune * r,int nrune)129*3cd4f1d1SCharles Forsyth runenlen(Rune *r, int nrune)
13037da2899SCharles.Forsyth {
131*3cd4f1d1SCharles Forsyth int nb, i;
132*3cd4f1d1SCharles Forsyth Rune c;
13337da2899SCharles.Forsyth
134*3cd4f1d1SCharles Forsyth nb = 0;
135*3cd4f1d1SCharles Forsyth while(nrune--) {
13637da2899SCharles.Forsyth c = *r++;
137*3cd4f1d1SCharles Forsyth if(c <= Rune1){
138*3cd4f1d1SCharles Forsyth nb++;
139*3cd4f1d1SCharles Forsyth } else {
140*3cd4f1d1SCharles Forsyth for(i = 2; i < UTFmax + 1; i++)
141*3cd4f1d1SCharles Forsyth if(c <= RuneX(i) || i == UTFmax){
142*3cd4f1d1SCharles Forsyth nb += i;
143*3cd4f1d1SCharles Forsyth break;
14437da2899SCharles.Forsyth }
145*3cd4f1d1SCharles Forsyth }
146*3cd4f1d1SCharles Forsyth }
147*3cd4f1d1SCharles Forsyth return nb;
14837da2899SCharles.Forsyth }
14937da2899SCharles.Forsyth
15037da2899SCharles.Forsyth int
fullrune(char * str,int n)15137da2899SCharles.Forsyth fullrune(char *str, int n)
15237da2899SCharles.Forsyth {
153*3cd4f1d1SCharles Forsyth int i;
154*3cd4f1d1SCharles Forsyth Rune c;
15537da2899SCharles.Forsyth
156*3cd4f1d1SCharles Forsyth if(n <= 0)
157*3cd4f1d1SCharles Forsyth return 0;
15837da2899SCharles.Forsyth c = *(uchar*)str;
15937da2899SCharles.Forsyth if(c < Tx)
16037da2899SCharles.Forsyth return 1;
161*3cd4f1d1SCharles Forsyth for(i = 3; i < UTFmax + 1; i++)
162*3cd4f1d1SCharles Forsyth if(c < T(i))
163*3cd4f1d1SCharles Forsyth return n >= i - 1;
164*3cd4f1d1SCharles Forsyth return n >= UTFmax;
16537da2899SCharles.Forsyth }
166