tb_long_t tb_charset_conv_cstr(tb_size_t ftype, tb_size_t ttype, tb_char_t const* cstr, tb_byte_t* data, tb_size_t size)
{
// check
tb_assert_and_check_return_val(TB_CHARSET_TYPE_OK(ftype) && TB_CHARSET_TYPE_OK(ttype) && cstr && data && size, -1);
// conv
return tb_charset_conv_data(ftype, ttype, (tb_byte_t const*)cstr, tb_strlen(cstr), data, size);
}
static tb_bool_t tb_oc_bplist_writer_func_string(tb_oc_bplist_writer_t* writer, tb_object_ref_t object, tb_size_t item_size)
{
// check
tb_assert_and_check_return_val(writer && writer->stream && object, tb_false);
#if 0
// writ utf8
return tb_oc_bplist_writer_func_rdata(writer, TB_OBJECT_BPLIST_TYPE_STRING, tb_oc_string_cstr(object), tb_oc_string_size(object), item_size);
#else
// writ utf16
tb_char_t const* utf8 = tb_oc_string_cstr(object);
tb_size_t size = tb_oc_string_size(object);
if (utf8 && size)
{
#ifdef TB_CONFIG_MODULE_HAVE_CHARSET
// done
tb_bool_t ok = tb_false;
tb_char_t* utf16 = tb_null;
tb_size_t osize = 0;
do
{
// init utf16 data
utf16 = tb_malloc_cstr((size + 1) << 2);
tb_assert_and_check_break(utf16);
// utf8 to utf16
osize = tb_charset_conv_data(TB_CHARSET_TYPE_UTF8, TB_CHARSET_TYPE_UTF16, (tb_byte_t const*)utf8, size, (tb_byte_t*)utf16, (size + 1) << 2);
tb_assert_and_check_break(osize > 0 && osize < (size + 1) << 2);
tb_assert_and_check_break(!(osize & 1));
// ok
ok = tb_true;
} while (0);
// ok?
if (ok)
{
// only ascii? writ utf8
if (osize == (size << 1)) ok = tb_oc_bplist_writer_func_rdata(writer, TB_OBJECT_BPLIST_TYPE_STRING, (tb_byte_t*)utf8, size, item_size);
// writ utf16
else ok = tb_oc_bplist_writer_func_rdata(writer, TB_OBJECT_BPLIST_TYPE_UNICODE, (tb_byte_t*)utf16, osize >> 1, item_size);
}
// exit utf16
if (utf16) tb_free(utf16);
utf16 = tb_null;
#else
// writ utf8 only
tb_bool_t ok = tb_oc_bplist_writer_func_rdata(writer, TB_OBJECT_BPLIST_TYPE_STRING, (tb_byte_t*)utf8, size, item_size);
#endif
// ok?
return ok;
}
tb_size_t tb_wcstombs(tb_char_t* s1, tb_wchar_t const* s2, tb_size_t n)
{
// check
tb_assert_and_check_return_val(s1 && s2, 0);
// init
tb_long_t r = 0;
tb_size_t l = tb_wcslen(s2);
// atow
if (l)
{
tb_size_t e = (sizeof(tb_wchar_t) == 4)? TB_CHARSET_TYPE_UCS4 : TB_CHARSET_TYPE_UCS2;
r = tb_charset_conv_data(e | TB_CHARSET_TYPE_LE, TB_CHARSET_TYPE_UTF8, (tb_byte_t const*)s2, l * sizeof(tb_wchar_t), (tb_byte_t*)s1, n);
}
// strip
if (r >= 0) s1[r] = '\0';
// ok?
return r > 0? r : -1;
}
static tb_object_ref_t tb_object_bplist_reader_func_string(tb_object_bplist_reader_t* reader, tb_size_t type, tb_size_t size, tb_size_t item_size)
{
// check
tb_assert_and_check_return_val(reader && reader->stream, tb_null);
// init
tb_char_t* utf8 = tb_null;
tb_char_t* utf16 = tb_null;
tb_object_ref_t object = tb_null;
// read
switch (type)
{
case TB_OBJECT_BPLIST_TYPE_STRING:
{
// size is too large?
if (size == 0x0f)
{
// read size
tb_long_t val = tb_object_bplist_reader_func_size(reader, item_size);
tb_assert_and_check_return_val(val >= 0, tb_null);
size = (tb_size_t)val;
}
// read string
if (size)
{
// init utf8
utf8 = tb_malloc_cstr(size + 1);
tb_assert_and_check_break(utf8);
// read utf8
if (!tb_stream_bread(reader->stream, (tb_byte_t*)utf8, size)) break;
utf8[size] = '\0';
}
// init object
object = tb_object_string_init_from_cstr(utf8);
}
break;
case TB_OBJECT_BPLIST_TYPE_UNICODE:
{
#ifdef TB_CONFIG_MODULE_HAVE_CHARSET
// size is too large?
if (size == 0x0f)
{
// read size
tb_long_t val = tb_object_bplist_reader_func_size(reader, item_size);
tb_assert_and_check_return_val(val >= 0, tb_null);
size = (tb_size_t)val;
}
// read string
if (size)
{
// init utf8 & utf16 data
utf8 = tb_malloc_cstr((size + 1) << 2);
utf16 = tb_malloc_cstr(size << 1);
tb_assert_and_check_break(utf8 && utf16);
// read utf16
if (!tb_stream_bread(reader->stream, (tb_byte_t*)utf16, size << 1)) break;
// utf16 to utf8
tb_long_t osize = tb_charset_conv_data(TB_CHARSET_TYPE_UTF16, TB_CHARSET_TYPE_UTF8, (tb_byte_t*)utf16, size << 1, (tb_byte_t*)utf8, (size + 1) << 2);
tb_assert_and_check_break(osize > 0 && osize < (tb_long_t)((size + 1) << 2));
utf8[osize] = '\0';
// init object
object = tb_object_string_init_from_cstr(utf8);
}
#else
// trace
tb_trace1_e("unicode type is not supported, please enable charset module config if you want to use it!");
#endif
}
break;
default:
break;
}
// exit
if (utf8) tb_free(utf8);
if (utf16) tb_free(utf16);
// ok?
return object;
}