char* opencc_convert_utf8(opencc_t t_opencc, const char* inbuf, size_t length) {
if (!lib_initialized) {
lib_initialize();
}
size_t actual_length = strlen(inbuf);
if ((length == (size_t)-1) || (length > actual_length)) {
length = actual_length;
}
ucs4_t* winbuf = utf8_to_ucs4(inbuf, length);
if (winbuf == (ucs4_t*)-1) {
/* Can not convert input UTF8 to UCS4 */
errnum = OPENCC_ERROR_ENCODING;
return (char*)-1;
}
/* Set up UTF8 buffer */
size_t outbuf_len = length;
size_t outsize = outbuf_len;
char* original_outbuf = (char*)malloc(sizeof(char) * (outbuf_len + 1));
char* outbuf = original_outbuf;
original_outbuf[0] = '\0';
/* Set conversion buffer */
size_t wbufsize = length + 64;
ucs4_t* woutbuf = (ucs4_t*)malloc(sizeof(ucs4_t) * (wbufsize + 1));
ucs4_t* pinbuf = winbuf;
ucs4_t* poutbuf = woutbuf;
size_t inbuf_left, outbuf_left;
inbuf_left = ucs4len(winbuf);
outbuf_left = wbufsize;
while (inbuf_left > 0) {
size_t retval = opencc_convert(t_opencc,
&pinbuf,
&inbuf_left,
&poutbuf,
&outbuf_left);
if (retval == (size_t)-1) {
free(outbuf);
free(winbuf);
free(woutbuf);
return (char*)-1;
}
*poutbuf = L'\0';
char* ubuff = ucs4_to_utf8(woutbuf, (size_t)-1);
if (ubuff == (char*)-1) {
free(outbuf);
free(winbuf);
free(woutbuf);
errnum = OPENCC_ERROR_ENCODING;
return (char*)-1;
}
size_t ubuff_len = strlen(ubuff);
while (ubuff_len > outsize) {
size_t outbuf_offset = outbuf - original_outbuf;
outsize += outbuf_len;
outbuf_len += outbuf_len;
original_outbuf =
(char*)realloc(original_outbuf, sizeof(char) * outbuf_len);
outbuf = original_outbuf + outbuf_offset;
}
strncpy(outbuf, ubuff, ubuff_len);
free(ubuff);
outbuf += ubuff_len;
*outbuf = '\0';
outbuf_left = wbufsize;
poutbuf = woutbuf;
}
free(winbuf);
free(woutbuf);
original_outbuf = (char*)realloc(original_outbuf,
sizeof(char) * (strlen(original_outbuf) + 1));
return original_outbuf;
}
char * opencc_convert_utf8(opencc_t t_opencc, const char * inbuf, size_t length)
{
if (!lib_initialized)
lib_initialize();
if (length == (size_t) -1 || length > strlen(inbuf))
length = strlen(inbuf);
/* 將輸入數據轉換爲ucs4_t字符串 */
ucs4_t * winbuf = utf8_to_ucs4(inbuf, length);
if (winbuf == (ucs4_t *) -1)
{
/* 輸入數據轉換失敗 */
errnum = OPENCC_ERROR_ENCODIND;
return (char *) -1;
}
/* 設置輸出UTF8文本緩衝區空間 */
size_t outbuf_len = length;
size_t outsize = outbuf_len;
char * original_outbuf = (char *) malloc(sizeof(char) * (outbuf_len + 1));
char * outbuf = original_outbuf;
original_outbuf[0] = '\0';
/* 設置轉換緩衝區空間 */
size_t wbufsize = length + 64;
ucs4_t * woutbuf = (ucs4_t *) malloc(sizeof(ucs4_t) * (wbufsize + 1));
ucs4_t * pinbuf = winbuf;
ucs4_t * poutbuf = woutbuf;
size_t inbuf_left, outbuf_left;
inbuf_left = ucs4len(winbuf);
outbuf_left = wbufsize;
while (inbuf_left > 0)
{
size_t retval = opencc_convert(t_opencc, &pinbuf, &inbuf_left, &poutbuf, &outbuf_left);
if (retval == (size_t) -1)
{
free(outbuf);
free(winbuf);
free(woutbuf);
return (char *) -1;
}
*poutbuf = L'\0';
char * ubuff = ucs4_to_utf8(woutbuf, (size_t) -1);
if (ubuff == (char *) -1)
{
free(outbuf);
free(winbuf);
free(woutbuf);
errnum = OPENCC_ERROR_ENCODIND;
return (char *) -1;
}
size_t ubuff_len = strlen(ubuff);
while (ubuff_len > outsize)
{
size_t outbuf_offset = outbuf - original_outbuf;
outsize += outbuf_len;
outbuf_len += outbuf_len;
original_outbuf = (char *) realloc(original_outbuf, sizeof(char) * outbuf_len);
outbuf = original_outbuf + outbuf_offset;
}
strncpy(outbuf, ubuff, ubuff_len);
free(ubuff);
outbuf += ubuff_len;
*outbuf = '\0';
outbuf_left = wbufsize;
poutbuf = woutbuf;
}
free(winbuf);
free(woutbuf);
original_outbuf = (char *) realloc(original_outbuf,
sizeof(char) * (strlen(original_outbuf) + 1));
return original_outbuf;
}