o5r/plan9port
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

convert to 4-byte UTF-8 and 32-bit Rune

Browse source 
http://codereview.appspot.com/116075
This commit is contained in:
Russ Cox 2009-09-11 17:03:06 -04:00
parent 4dbf255619
commit 0cadb4301d
14 changed files with 86 additions and 44 deletions
Download patch file Download diff file Expand all files Collapse all files

11
src/lib9/fmt/dofmt.c
View file

@ -605,12 +605,13 @@ __flagfmt(Fmt *f)
int
__badfmt(Fmt *f)
{
char x[3];
char x[2+UTFmax];
int n;
x[0] = '%';
x[1] = f->r;
x[2] = '%';
f->prec = 3;
__fmtcpy(f, (const void*)x, 3, 3);
n = 1 + runetochar(x+1, &f->r);
x[n++] = '%';
f->prec = n;
__fmtcpy(f, (const void*)x, n, n);
return 0;
}

78
src/lib9/utf/rune.c
View file

@ -23,16 +23,19 @@ 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 */
@ -43,7 +46,7 @@ enum
int
chartorune(Rune *rune, char *str)
{
int c, c1, c2;
int c, c1, c2, c3;
long l;
/*
@ -88,6 +91,25 @@ 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
*/
@ -113,7 +135,7 @@ runetochar(char *str, Rune *rune)
/*
* two character sequence
* 0080-07FF => T2 Tx
* 00080-007FF => T2 Tx
*/
if(c <= Rune2) {
str[0] = T2 | (c >> 1*Bitx);
@ -123,12 +145,26 @@ runetochar(char *str, Rune *rune)
/*
* three character sequence
* 0800-FFFF => T3 Tx Tx
* 00800-0FFFF => 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 > Runemax)
c = Runeerror;
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
* 010000-1FFFFF => 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
@ -155,7 +191,10 @@ runenlen(Rune *r, int nrune)
if(c <= Rune2)
nb += 2;
else
if(c <= Rune3 || c > Runemax)
nb += 3;
else
nb += 4;
}
return nb;
}
@ -165,13 +204,14 @@ 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;
}