several optimization
This commit is contained in:
parent
070a69cf60
commit
c52b21c9ee
12 changed files with 214 additions and 111 deletions
|
|
@ -94,46 +94,60 @@ static int utf8_later(unsigned char c) {
|
|||
|
||||
#define CAST_I32(x) ((int)(x))
|
||||
|
||||
int sj4_utf8_read(const unsigned char* in, int* output) {
|
||||
u_char c = in[0];
|
||||
u_int n = 0;
|
||||
|
||||
if(c < 0x80) {
|
||||
*output = c;
|
||||
|
||||
return 1;
|
||||
} else if(0xc2 <= c && c < 0xe0) {
|
||||
if(!utf8_later(in[1])) return 0;
|
||||
if((in[0] & 0x1e) == 0) return 0;
|
||||
|
||||
n = CAST_I32(in[0] & 0x1f) << 6;
|
||||
n |= CAST_I32(in[1] & 0x3f);
|
||||
|
||||
*output = n;
|
||||
|
||||
return 2;
|
||||
} else if(0xe0 <= c && c < 0xf0) {
|
||||
if(!utf8_later(in[1]) || !utf8_later(in[2])) return 0;
|
||||
if((in[0] & 0x0f) == 0 && (in[1] & 0x20) == 0) return 0;
|
||||
|
||||
n = CAST_I32(in[0] & 0x0f) << 12;
|
||||
n |= CAST_I32(in[1] & 0x3f) << 6;
|
||||
n |= CAST_I32(in[2] & 0x3f);
|
||||
|
||||
*output = n;
|
||||
|
||||
return 3;
|
||||
} else if(0xf0 <= c && c < 0xf8) {
|
||||
if(!utf8_later(in[1]) || !utf8_later(in[2]) || !utf8_later(in[3])) return 0;
|
||||
if((in[0] & 0x07) == 0 && (in[1] & 0x30) == 0) return 0;
|
||||
|
||||
n = CAST_I32(in[0] & 0x07) << 18;
|
||||
n |= CAST_I32(in[1] & 0x3f) << 12;
|
||||
n |= CAST_I32(in[2] & 0x3f) << 6;
|
||||
n |= CAST_I32(in[3] & 0x3f);
|
||||
|
||||
*output = n;
|
||||
|
||||
return 4;
|
||||
}
|
||||
|
||||
return 0;
|
||||
}
|
||||
|
||||
int sj4_from_utf8(u_char* out, const u_char* in, int len) {
|
||||
const u_char* o_in = in;
|
||||
u_char* o_out = out;
|
||||
|
||||
while((in - o_in) < len) {
|
||||
u_char c = *in;
|
||||
u_int n = 0;
|
||||
int n;
|
||||
|
||||
if(c < 0x80) {
|
||||
n = c;
|
||||
|
||||
in += 1;
|
||||
} else if(0xc2 <= c && c < 0xe0) {
|
||||
if(!utf8_later(in[1])) return 0;
|
||||
if((in[0] & 0x1e) == 0) return 0;
|
||||
|
||||
n = CAST_I32(in[0] & 0x1f) << 6;
|
||||
n |= CAST_I32(in[1] & 0x3f);
|
||||
|
||||
in += 2;
|
||||
} else if(0xe0 <= c && c < 0xf0) {
|
||||
if(!utf8_later(in[1]) || !utf8_later(in[2])) return 0;
|
||||
if((in[0] & 0x0f) == 0 && (in[1] & 0x20) == 0) return 0;
|
||||
|
||||
n = CAST_I32(in[0] & 0x0f) << 12;
|
||||
n |= CAST_I32(in[1] & 0x3f) << 6;
|
||||
n |= CAST_I32(in[2] & 0x3f);
|
||||
|
||||
in += 3;
|
||||
} else if(0xf0 <= c && c < 0xf8) {
|
||||
if(!utf8_later(in[1]) || !utf8_later(in[2]) || !utf8_later(in[3])) return 0;
|
||||
if((in[0] & 0x07) == 0 && (in[1] & 0x30) == 0) return 0;
|
||||
|
||||
n = CAST_I32(in[0] & 0x07) << 18;
|
||||
n |= CAST_I32(in[1] & 0x3f) << 12;
|
||||
n |= CAST_I32(in[2] & 0x3f) << 6;
|
||||
n |= CAST_I32(in[3] & 0x3f);
|
||||
|
||||
in += 4;
|
||||
}
|
||||
in += sj4_utf8_read(in, &n);
|
||||
|
||||
n = ucs_to_euc(n);
|
||||
|
||||
|
|
@ -151,39 +165,45 @@ static int is_low_urr(wchar_t ch) {
|
|||
return 0xDC00 <= ch && ch < 0xE000;
|
||||
}
|
||||
|
||||
int sj4_utf16_read(const unsigned char* in, int* output) {
|
||||
const wchar_t* w_in = (const wchar_t*)in;
|
||||
|
||||
if(is_high_surr(w_in[0])) {
|
||||
if(is_low_urr(w_in[1])) {
|
||||
*output = 0x10000 + CAST_I32(w_in[0] - 0xd800) * 0x400 + CAST_I32(w_in[1] - 0xdc00);
|
||||
|
||||
return 2;
|
||||
} else if(w_in[1] == 0) {
|
||||
*output = w_in[0];
|
||||
|
||||
return 2;
|
||||
} else {
|
||||
return 0;
|
||||
}
|
||||
} else if(is_low_urr(w_in[0])) {
|
||||
if(w_in[1] == 0) {
|
||||
*output = w_in[0];
|
||||
|
||||
return 2;
|
||||
} else {
|
||||
return 0;
|
||||
}
|
||||
} else {
|
||||
*output = w_in[0];
|
||||
|
||||
return 1;
|
||||
}
|
||||
}
|
||||
|
||||
int sj4_from_utf16(u_char* out, const u_char* in, int len) {
|
||||
const wchar_t* w_in = (const wchar_t*)in;
|
||||
const wchar_t* o_in = w_in;
|
||||
u_char* o_out = out;
|
||||
|
||||
while((w_in - o_in) < len) {
|
||||
u_int n = 0;
|
||||
int n = 0;
|
||||
|
||||
if(is_high_surr(w_in[0])) {
|
||||
if(is_low_urr(w_in[1])) {
|
||||
n = 0x10000 + CAST_I32(w_in[0] - 0xd800) * 0x400 + CAST_I32(w_in[1] - 0xdc00);
|
||||
|
||||
w_in += 2;
|
||||
} else if(w_in[1] == 0) {
|
||||
n = w_in[0];
|
||||
|
||||
w_in += 2;
|
||||
} else {
|
||||
return 0;
|
||||
}
|
||||
} else if(is_low_urr(w_in[0])) {
|
||||
if(w_in[1] == 0) {
|
||||
n = w_in[0];
|
||||
|
||||
w_in += 2;
|
||||
} else {
|
||||
return 0;
|
||||
}
|
||||
} else {
|
||||
n = w_in[0];
|
||||
|
||||
w_in += 1;
|
||||
}
|
||||
w_in += sj4_utf16_read((const unsigned char*)w_in, &n);
|
||||
|
||||
n = ucs_to_euc(n);
|
||||
|
||||
|
|
@ -192,35 +212,55 @@ int sj4_from_utf16(u_char* out, const u_char* in, int len) {
|
|||
|
||||
return out - o_out;
|
||||
}
|
||||
#endif
|
||||
|
||||
static u_int eucjp_read(const u_char* in) {
|
||||
int sj4_eucjp_read(const unsigned char* in, int* output) {
|
||||
u_char c = *in;
|
||||
u_int n = 0;
|
||||
|
||||
if(c <= 0x7f) {
|
||||
n = c;
|
||||
*output = c;
|
||||
|
||||
return 1;
|
||||
} else if(c != 0x8f) {
|
||||
n = CAST_I32(in[0]) << 8;
|
||||
n |= CAST_I32(in[1]);
|
||||
|
||||
*output = n;
|
||||
|
||||
return 2;
|
||||
} else {
|
||||
n = CAST_I32(in[0]) << 16;
|
||||
n |= CAST_I32(in[1]) << 8;
|
||||
n |= CAST_I32(in[2]);
|
||||
}
|
||||
|
||||
return n;
|
||||
}
|
||||
*output = n;
|
||||
|
||||
static int eucjp_codesize(u_int n) {
|
||||
if(n <= 0xff) {
|
||||
return 1;
|
||||
} else if(n <= 0xffff) {
|
||||
return 2;
|
||||
} else {
|
||||
return 3;
|
||||
}
|
||||
}
|
||||
#endif
|
||||
|
||||
int sj4_sjis_read(const unsigned char* in, int* output) {
|
||||
u_char c;
|
||||
u_int n = 0;
|
||||
|
||||
c = in[0];
|
||||
|
||||
if(c <= 0x7f || (0xa1 <= c && c <= 0xdf)) {
|
||||
*output = c;
|
||||
|
||||
return 1;
|
||||
} else if((0x81 <= c && c <= 0x9f) || (0xe0 <= c && c <= 0xfc)) {
|
||||
n = CAST_I32(in[0]) << 8;
|
||||
n |= in[1];
|
||||
|
||||
*output = n;
|
||||
|
||||
return 2;
|
||||
}
|
||||
|
||||
return 0;
|
||||
}
|
||||
|
||||
int sj4_to_sjis(u_char* out, const u_char* in, int len) {
|
||||
unsigned char d[] = {0, 0};
|
||||
|
|
@ -235,10 +275,10 @@ int sj4_to_utf8(u_char* out, const u_char* in, int len) {
|
|||
u_char* o_out = out;
|
||||
|
||||
while((in - o_in) < len) {
|
||||
int i, j = 0;
|
||||
u_int n = eucjp_read(in);
|
||||
int i, j = 0;
|
||||
int n;
|
||||
|
||||
in += eucjp_codesize(n);
|
||||
in += sj4_eucjp_read(in, &n);
|
||||
|
||||
n = euc_to_ucs(n);
|
||||
|
||||
|
|
@ -272,10 +312,10 @@ int sj4_to_utf16(u_char* out, const u_char* in, int len) {
|
|||
wchar_t* o_out = w_out;
|
||||
|
||||
while((in - o_in) < len) {
|
||||
int i, j = 0;
|
||||
u_int n = eucjp_read(in);
|
||||
int i, j = 0;
|
||||
int n;
|
||||
|
||||
in += eucjp_codesize(n);
|
||||
in += sj4_eucjp_read(in, &n);
|
||||
|
||||
n = euc_to_ucs(n);
|
||||
|
||||
|
|
@ -322,3 +362,18 @@ int sj4_charset_from(int type, void* output, const void* input, int len) {
|
|||
|
||||
return len;
|
||||
}
|
||||
|
||||
int sj4_charset_read(int type, int* output, const void* input) {
|
||||
if(type == SJ4EUCJP) {
|
||||
return sj4_eucjp_read(input, output);
|
||||
} else if(type == SJ4SJIS) {
|
||||
return sj4_sjis_read(input, output);
|
||||
}
|
||||
#ifdef UCS
|
||||
else if(type == SJ4UTF8) {
|
||||
return sj4_utf8_read(input, output);
|
||||
} else if(type == SJ4UTF16) {
|
||||
return sj4_utf16_read(input, output);
|
||||
}
|
||||
#endif
|
||||
}
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue