123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162 |
- #include <u.h>
- #include <libc.h>
- enum
- {
- Bit1 = 7,
- Bitx = 6,
- Bit2 = 5,
- Bit3 = 4,
- Bit4 = 3,
- T1 = ((1<<(Bit1+1))-1) ^ 0xFF, /* 0000 0000 */
- Tx = ((1<<(Bitx+1))-1) ^ 0xFF, /* 1000 0000 */
- T2 = ((1<<(Bit2+1))-1) ^ 0xFF, /* 1100 0000 */
- T3 = ((1<<(Bit3+1))-1) ^ 0xFF, /* 1110 0000 */
- T4 = ((1<<(Bit4+1))-1) ^ 0xFF, /* 1111 0000 */
- Rune1 = (1<<(Bit1+0*Bitx))-1, /* 0000 0000 0111 1111 */
- Rune2 = (1<<(Bit2+1*Bitx))-1, /* 0000 0111 1111 1111 */
- Rune3 = (1<<(Bit3+2*Bitx))-1, /* 1111 1111 1111 1111 */
- Maskx = (1<<Bitx)-1, /* 0011 1111 */
- Testx = Maskx ^ 0xFF, /* 1100 0000 */
- Bad = Runeerror,
- };
- int
- chartorune(Rune *rune, char *str)
- {
- int c, c1, c2;
- long l;
- /*
- * one character sequence
- * 00000-0007F => T1
- */
- c = *(uchar*)str;
- if(c < Tx) {
- *rune = c;
- return 1;
- }
- /*
- * two character sequence
- * 0080-07FF => T2 Tx
- */
- c1 = *(uchar*)(str+1) ^ Tx;
- if(c1 & Testx)
- goto bad;
- if(c < T3) {
- if(c < T2)
- goto bad;
- l = ((c << Bitx) | c1) & Rune2;
- if(l <= Rune1)
- goto bad;
- *rune = l;
- return 2;
- }
- /*
- * three character sequence
- * 0800-FFFF => T3 Tx Tx
- */
- c2 = *(uchar*)(str+2) ^ Tx;
- if(c2 & Testx)
- goto bad;
- if(c < T4) {
- l = ((((c << Bitx) | c1) << Bitx) | c2) & Rune3;
- if(l <= Rune2)
- goto bad;
- *rune = l;
- return 3;
- }
- /*
- * bad decoding
- */
- bad:
- *rune = Bad;
- return 1;
- }
- int
- runetochar(char *str, Rune *rune)
- {
- long c;
- /*
- * one character sequence
- * 00000-0007F => 00-7F
- */
- c = *rune;
- if(c <= Rune1) {
- str[0] = c;
- return 1;
- }
- /*
- * two character sequence
- * 0080-07FF => T2 Tx
- */
- if(c <= Rune2) {
- str[0] = T2 | (c >> 1*Bitx);
- str[1] = Tx | (c & Maskx);
- return 2;
- }
- /*
- * three character sequence
- * 0800-FFFF => T3 Tx Tx
- */
- str[0] = T3 | (c >> 2*Bitx);
- str[1] = Tx | ((c >> 1*Bitx) & Maskx);
- str[2] = Tx | (c & Maskx);
- return 3;
- }
- int
- runelen(long c)
- {
- Rune rune;
- char str[10];
- rune = c;
- return runetochar(str, &rune);
- }
- int
- runenlen(Rune *r, int nrune)
- {
- int nb, c;
- nb = 0;
- while(nrune--) {
- c = *r++;
- if(c <= Rune1)
- nb++;
- else
- if(c <= Rune2)
- nb += 2;
- else
- nb += 3;
- }
- return nb;
- }
- int
- fullrune(char *str, int n)
- {
- int c;
- if(n > 0) {
- c = *(uchar*)str;
- if(c < Tx)
- return 1;
- if(n > 1)
- if(c < T3 || n > 2)
- return 1;
- }
- return 0;
- }
|