48 lines
1008 B
C
48 lines
1008 B
C
|
/*
|
||
|
* Stateful UTF-8 decoder
|
||
|
*/
|
||
|
|
||
|
#include <stdint.h>
|
||
|
|
||
|
#define UTF8_ACCEPT 0
|
||
|
#define UTF8_REJECT 1
|
||
|
|
||
|
|
||
|
static inline uint32_t decode(uint32_t* state, uint32_t* codep, uint32_t byte) {
|
||
|
static int state_table[32] = {
|
||
|
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, /* 0xxxxxxx */
|
||
|
1,1,1,1,1,1,1,1, /* 10xxxxxx */
|
||
|
2,2,2,2, /* 110xxxxx */
|
||
|
3,3, /* 1110xxxx */
|
||
|
4, /* 11110xxx */
|
||
|
1 /* 11111xxx */
|
||
|
};
|
||
|
|
||
|
static int mask_bytes[32] = {
|
||
|
0x7F,0x7F,0x7F,0x7F,0x7F,0x7F,0x7F,0x7F,
|
||
|
0x7F,0x7F,0x7F,0x7F,0x7F,0x7F,0x7F,0x7F,
|
||
|
0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
||
|
0x1F,0x1F,0x1F,0x1F,
|
||
|
0x0F,0x0F,
|
||
|
0x07,
|
||
|
0x00
|
||
|
};
|
||
|
|
||
|
static int next[5] = {
|
||
|
0,
|
||
|
1,
|
||
|
0,
|
||
|
2,
|
||
|
3
|
||
|
};
|
||
|
|
||
|
if (*state == UTF8_ACCEPT) {
|
||
|
*codep = byte & mask_bytes[byte >> 3];
|
||
|
*state = state_table[byte >> 3];
|
||
|
} else if (*state > 0) {
|
||
|
*codep = (byte & 0x3F) | (*codep << 6);
|
||
|
*state = next[*state];
|
||
|
}
|
||
|
return *state;
|
||
|
}
|