/*
* $Id$
*
Copyright 2006-2014 Chung, Hyung-Hwan.
This file is part of H2.
H2 is free software: you can redistribute it and/or modify
it under the terms of the GNU Lesser General Public License as
published by the Free Software Foundation, either version 3 of
the License, or (at your option) any later version.
H2 is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with H2. If not, see .
*/
#include
/*
* from RFC 2279 UTF-8, a transformation format of ISO 10646
*
* UCS-4 range (hex.) UTF-8 octet sequence (binary)
* 1:2 00000000-0000007F 0xxxxxxx
* 2:2 00000080-000007FF 110xxxxx 10xxxxxx
* 3:2 00000800-0000FFFF 1110xxxx 10xxxxxx 10xxxxxx
* 4:4 00010000-001FFFFF 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
* inv 00200000-03FFFFFF 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
* inv 04000000-7FFFFFFF 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx
*/
struct __utf8_t
{
h2_uint32_t lower;
h2_uint32_t upper;
h2_uint8_t fbyte;
h2_uint8_t mask; /* mask for getting the fixed bits in the first byte (First byte & mask => fbyte) */
h2_uint8_t fmask; /* mask for getting the actual values bits off the first byte */
int length; /* number of bytes */
};
typedef struct __utf8_t __utf8_t;
static __utf8_t utf8_table[] =
{
{0x00000000ul, 0x0000007Ful, 0x00, 0x80, 0x7F, 1},
{0x00000080ul, 0x000007FFul, 0xC0, 0xE0, 0x1F, 2},
{0x00000800ul, 0x0000FFFFul, 0xE0, 0xF0, 0x0F, 3},
{0x00010000ul, 0x001FFFFFul, 0xF0, 0xF8, 0x07, 4},
{0x00200000ul, 0x03FFFFFFul, 0xF8, 0xFC, 0x03, 5},
{0x04000000ul, 0x7FFFFFFFul, 0xFC, 0xFE, 0x01, 6}
};
static H2_INLINE __utf8_t* get_utf8_slot (h2_wxchar_t uc)
{
__utf8_t* cur, * end;
H2_ASSERT (H2_SIZEOF(h2_mchar_t) == 1);
H2_ASSERT (H2_SIZEOF(h2_wxchar_t) >= 2);
end = utf8_table + H2_COUNTOF(utf8_table);
cur = utf8_table;
while (cur < end)
{
if (uc >= cur->lower && uc <= cur->upper) return cur;
cur++;
}
return H2_NULL; /* invalid character */
}
h2_size_t h2_uctoutf8 (h2_wxchar_t uc, h2_mchar_t* utf8, h2_size_t size)
{
__utf8_t* cur = get_utf8_slot (uc);
if (cur == H2_NULL) return 0; /* illegal character */
if (utf8 && cur->length <= size)
{
int index = cur->length;
while (index > 1)
{
/*
* 0x3F: 00111111
* 0x80: 10000000
*/
utf8[--index] = (uc & 0x3F) | 0x80;
uc >>= 6;
}
utf8[0] = uc | cur->fbyte;
}
/* small buffer is also indicated by this return value
* greater than 'size'. */
return (h2_size_t)cur->length;
}
h2_size_t h2_utf8touc (
const h2_mchar_t* utf8, h2_size_t size, h2_wxchar_t* uc)
{
__utf8_t* cur, * end;
H2_ASSERT (utf8 != H2_NULL);
H2_ASSERT (size > 0);
H2_ASSERT (H2_SIZEOF(h2_mchar_t) == 1);
H2_ASSERT (H2_SIZEOF(h2_wxchar_t) >= 2);
end = utf8_table + H2_COUNTOF(utf8_table);
cur = utf8_table;
while (cur < end)
{
if ((utf8[0] & cur->mask) == cur->fbyte)
{
/* if size is less that cur->length, the incomplete-seqeunce
* error is naturally indicated. so validate the string
* only if size is as large as cur->length. */
if (size >= cur->length)
{
int i;
if (uc)
{
h2_wxchar_t w;
w = utf8[0] & cur->fmask;
for (i = 1; i < cur->length; i++)
{
/* in utf8, trailing bytes are all
* set with 0x80.
*
* 10XXXXXX & 11000000 => 10000000
*
* if not, invalid. */
if ((utf8[i] & 0xC0) != 0x80) return 0;
w = (w << 6) | (utf8[i] & 0x3F);
}
*uc = w;
}
else
{
for (i = 1; i < cur->length; i++)
{
/* in utf8, trailing bytes are all
* set with 0x80.
*
* 10XXXXXX & 11000000 => 10000000
*
* if not, invalid. */
if ((utf8[i] & 0xC0) != 0x80) return 0;
}
}
}
/* this return value can indicate both
* the correct length (len >= cur->length)
* and
* the incomplete seqeunce error (len < cur->length).
*/
return (h2_size_t)cur->length;
}
cur++;
}
return 0; /* error - invalid sequence */
}
h2_size_t h2_utf8len (const h2_mchar_t* utf8, h2_size_t size)
{
return h2_utf8touc (utf8, size, H2_NULL);
}
h2_size_t h2_utf8lenmax (void)
{
return H2_UTF8LEN_MAX;
}