mirrors/pulseaudio
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/pulseaudio/pulseaudio.git synced 2025-11-05 13:29:57 -05:00
Code Issues Projects Releases Packages Wiki Activity Actions

Clean up the UTF-8 validation code.

Browse source 
git-svn-id: file:///home/lennart/svn/public/pulseaudio/trunk@870 fefdeb5f-60dc-0310-8127-8f9354f1896f
This commit is contained in:
Pierre Ossman 2006-05-15 12:44:44 +00:00
parent 147da3e36f
commit e91740f68c
1 changed files with 71 additions and 73 deletions
Download patch file Download diff file Expand all files Collapse all files

144
src/polypcore/utf8.c
View file

@ -29,87 +29,85 @@
#endif #endif
#include <stdlib.h> #include <stdlib.h>
#include <inttypes.h>
#include "utf8.h" #include "utf8.h"
#define UNICODE_VALID(Char) \ static inline int is_unicode_valid(uint32_t ch) {
((Char) < 0x110000 && \ if (ch >= 0x110000) /* End of unicode space */
(((Char) & 0xFFFFF800) != 0xD800) && \ return 0;
((Char) < 0xFDD0 || (Char) > 0xFDEF) && \ if ((ch & 0xFFFFF800) == 0xD800) /* Reserved area for UTF-16 */
((Char) & 0xFFFE) != 0xFFFE) return 0;
if ((ch >= 0xFDD0) && (ch <= 0xFDEF)) /* Reserved */
return 0;
#define CONTINUATION_CHAR \ if ((ch & 0xFFFE) == 0xFFFE) /* BOM (Byte Order Mark) */
do { \ return 0;
if ((*(const unsigned char *)p & 0xc0) != 0x80) /* 10xxxxxx */ \ return 1;
goto error; \ }
val <<= 6; \
val |= (*(const unsigned char *)p) & 0x3f; \
} while(0)
static inline int is_continuation_char(uint8_t ch) {
if ((ch & 0xc0) != 0x80) /* 10xxxxxx */
return 0;
return 1;
}
const char * static inline void merge_continuation_char(uint32_t *u_ch, uint8_t ch) {
pa_utf8_valid (const char *str) *u_ch <<= 6;
*u_ch |= ch & 0x3f;
}
{ const char* pa_utf8_valid (const char *str) {
unsigned val = 0; uint32_t val = 0;
unsigned min = 0; uint32_t min = 0;
const char *p; const uint8_t *p, *last;
for (p = str; *p; p++) for (p = (uint8_t*)str; *p; p++) {
{ if (*p < 128)
if (*(const unsigned char *)p < 128) /* done */;
/* done */; else {
else last = p;
{
const char *last;
last = p;
if ((*(const unsigned char *)p & 0xe0) == 0xc0) /* 110xxxxx */
{
if ( ((*(const unsigned char *)p & 0x1e) == 0))
goto error;
p++;
if ( ((*(const unsigned char *)p & 0xc0) != 0x80)) /* 10xxxxxx */
goto error;
}
else
{
if ((*(const unsigned char *)p & 0xf0) == 0xe0) /* 1110xxxx */
{
min = (1 << 11);
val = *(const unsigned char *)p & 0x0f;
goto TWO_REMAINING;
}
else if ((*(const unsigned char *)p & 0xf8) == 0xf0) /* 11110xxx */
{
min = (1 << 16);
val = *(const unsigned char *)p & 0x07;
}
else
goto error;
p++;
CONTINUATION_CHAR;
TWO_REMAINING:
p++;
CONTINUATION_CHAR;
p++;
CONTINUATION_CHAR;
if ( (val < min))
goto error;
if ( (!UNICODE_VALID(val))) if ((*p & 0xe0) == 0xc0) { /* 110xxxxx two-char seq. */
goto error; min = 128;
} val = *p & 0x1e;
goto ONE_REMAINING;
continue; } else if ((*p & 0xf0) == 0xe0) { /* 1110xxxx three-char seq.*/
min = (1 << 11);
error: val = *p & 0x0f;
return NULL; goto TWO_REMAINING;
} } else if ((*p & 0xf8) == 0xf0) { /* 11110xxx four-char seq */
min = (1 << 16);
val = *p & 0x07;
} else
goto error;
p++;
if (!is_continuation_char(*p))
goto error;
merge_continuation_char(&val, *p);
TWO_REMAINING:
p++;
if (!is_continuation_char(*p))
goto error;
merge_continuation_char(&val, *p);
ONE_REMAINING:
p++;
if (!is_continuation_char(*p))
goto error;
merge_continuation_char(&val, *p);
if (val < min)
goto error;
if (!is_unicode_valid(val))
goto error;
}
} }
return str; return str;
error:
return NULL;
} }