fix utf and rune handling in preparation for 32bit runes
This commit is contained in:
parent
6d99096136
commit
6cadd03bbe
53 changed files with 582 additions and 283 deletions
|
@ -512,12 +512,15 @@ _flagfmt(Fmt *f)
|
|||
int
|
||||
_badfmt(Fmt *f)
|
||||
{
|
||||
char x[3];
|
||||
char x[2+UTFmax];
|
||||
Rune r;
|
||||
int n;
|
||||
|
||||
r = f->r;
|
||||
x[0] = '%';
|
||||
x[1] = f->r;
|
||||
x[2] = '%';
|
||||
f->prec = 3;
|
||||
_fmtcpy(f, x, 3, 3);
|
||||
n = 1+runetochar(x+1, &r);
|
||||
x[n++] = '%';
|
||||
f->prec = n;
|
||||
_fmtcpy(f, x, n, n);
|
||||
return 0;
|
||||
}
|
||||
|
|
|
@ -8,16 +8,19 @@ enum
|
|||
Bit2 = 5,
|
||||
Bit3 = 4,
|
||||
Bit4 = 3,
|
||||
Bit5 = 2,
|
||||
|
||||
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 */
|
||||
T5 = ((1<<(Bit5+1))-1) ^ 0xFF, /* 1111 1000 */
|
||||
|
||||
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 */
|
||||
Rune1 = (1<<(Bit1+0*Bitx))-1, /* 0000 0000 0000 0000 0111 1111 */
|
||||
Rune2 = (1<<(Bit2+1*Bitx))-1, /* 0000 0000 0000 0111 1111 1111 */
|
||||
Rune3 = (1<<(Bit3+2*Bitx))-1, /* 0000 0000 1111 1111 1111 1111 */
|
||||
Rune4 = (1<<(Bit4+3*Bitx))-1, /* 0011 1111 1111 1111 1111 1111 */
|
||||
|
||||
Maskx = (1<<Bitx)-1, /* 0011 1111 */
|
||||
Testx = Maskx ^ 0xFF, /* 1100 0000 */
|
||||
|
@ -28,7 +31,7 @@ enum
|
|||
int
|
||||
chartorune(Rune *rune, char *str)
|
||||
{
|
||||
int c, c1, c2;
|
||||
int c, c1, c2, c3;
|
||||
long l;
|
||||
|
||||
/*
|
||||
|
@ -73,6 +76,25 @@ chartorune(Rune *rune, char *str)
|
|||
return 3;
|
||||
}
|
||||
|
||||
/*
|
||||
* four character sequence
|
||||
* 10000-10FFFF => T4 Tx Tx Tx
|
||||
*/
|
||||
if(UTFmax >= 4) {
|
||||
c3 = *(uchar*)(str+3) ^ Tx;
|
||||
if(c3 & Testx)
|
||||
goto bad;
|
||||
if(c < T5) {
|
||||
l = ((((((c << Bitx) | c1) << Bitx) | c2) << Bitx) | c3) & Rune4;
|
||||
if(l <= Rune3)
|
||||
goto bad;
|
||||
if(l > Runemax)
|
||||
goto bad;
|
||||
*rune = l;
|
||||
return 4;
|
||||
}
|
||||
}
|
||||
|
||||
/*
|
||||
* bad decoding
|
||||
*/
|
||||
|
@ -86,11 +108,14 @@ runetochar(char *str, Rune *rune)
|
|||
{
|
||||
long c;
|
||||
|
||||
c = *rune;
|
||||
if(c > Runemax)
|
||||
c = Runeerror;
|
||||
|
||||
/*
|
||||
* one character sequence
|
||||
* 00000-0007F => 00-7F
|
||||
*/
|
||||
c = *rune;
|
||||
if(c <= Rune1) {
|
||||
str[0] = c;
|
||||
return 1;
|
||||
|
@ -110,17 +135,29 @@ runetochar(char *str, Rune *rune)
|
|||
* 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;
|
||||
if(c <= Rune3) {
|
||||
str[0] = T3 | (c >> 2*Bitx);
|
||||
str[1] = Tx | ((c >> 1*Bitx) & Maskx);
|
||||
str[2] = Tx | (c & Maskx);
|
||||
return 3;
|
||||
}
|
||||
|
||||
/*
|
||||
* four character sequence
|
||||
* 10000-1FFFFF => T4 Tx Tx Tx
|
||||
*/
|
||||
str[0] = T4 | (c >> 3*Bitx);
|
||||
str[1] = Tx | ((c >> 2*Bitx) & Maskx);
|
||||
str[2] = Tx | ((c >> 1*Bitx) & Maskx);
|
||||
str[3] = Tx | (c & Maskx);
|
||||
return 4;
|
||||
}
|
||||
|
||||
int
|
||||
runelen(long c)
|
||||
{
|
||||
Rune rune;
|
||||
char str[10];
|
||||
char str[UTFmax];
|
||||
|
||||
rune = c;
|
||||
return runetochar(str, &rune);
|
||||
|
@ -140,7 +177,10 @@ runenlen(Rune *r, int nrune)
|
|||
if(c <= Rune2)
|
||||
nb += 2;
|
||||
else
|
||||
if(c <= Rune3 || c > Runemax)
|
||||
nb += 3;
|
||||
else
|
||||
nb += 4;
|
||||
}
|
||||
return nb;
|
||||
}
|
||||
|
@ -150,13 +190,15 @@ 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;
|
||||
if(n <= 0)
|
||||
return 0;
|
||||
c = *(uchar*)str;
|
||||
if(c < Tx)
|
||||
return 1;
|
||||
if(c < T3)
|
||||
return n >= 2;
|
||||
if(UTFmax == 3 || c < T4)
|
||||
return n >= 3;
|
||||
return n >= 4;
|
||||
}
|
||||
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue