QString Converter::convert(QString & text)
{
QByteArray text_utf8 = text.toUtf8();
const char * buffer_in = text_utf8.data();
char * buffer_out = opencc_convert_utf8(handle, buffer_in, -1);
if (buffer_out == (char *) -1)
{
opencc_perror("Opencc runtime:");
opencc_close(handle);
return ""; //TODO failed
}
QString retval = QString::fromUtf8(buffer_out);
free(buffer_out);
return retval;
}
void Converter::setConfig(const char *config)
{
if (!s_loaded)
return;
if (handle != NULL)
opencc_close(handle);
handle = opencc_open(config);
if (handle == (opencc_t) -1)
{
opencc_perror("Opencc loading:");
return;
}
m_loaded = true;
}
void convert(const char * input_file, const char * output_file, const char * config_file)
{
opencc_t od = opencc_open(config_file);
if (od == (opencc_t) -1)
{
opencc_perror(_("OpenCC initialization error"));
exit(1);
}
FILE * fp = stdin;
FILE * fpo = stdout;
if (input_file)
{
fp = fopen(input_file, "r");
if (!fp)
{
fprintf(stderr, _("Can not read file: %s\n"), input_file);
exit(1);
}
}
if (output_file)
{
fpo = fopen(output_file, "w");
if (!fpo)
{
fprintf(stderr, _("Can not write file: %s\n"), output_file);
exit(1);
}
}
size_t size = BUFFER_SIZE;
char * buffer_in = NULL, * buffer_out = NULL;
buffer_in = (char *) malloc(size * sizeof(char));
while (fgets(buffer_in, size, fp) != NULL)
{
size_t freesize = size;
char * buffer_in_p = buffer_in;
size_t line_length = strlen(buffer_in_p);
while (line_length + 1 == freesize && buffer_in_p[line_length - 2] != '\n')
{
//如果一行沒讀完，則最後一個字符不是換行，且讀滿緩衝區
buffer_in_p += size - 1;
freesize = size + 1;
size += size;
size_t offset = buffer_in_p - buffer_in;
buffer_in = (char *) realloc(buffer_in, size * sizeof(char));
buffer_in_p = buffer_in + offset;
if (fgets(buffer_in_p, freesize, fp) == NULL)
break;
line_length = strlen(buffer_in_p);
}
buffer_out = opencc_convert_utf8(od, buffer_in, (size_t) -1);
if (buffer_out != (char *) -1)
{
fprintf(fpo, "%s", buffer_out);
}
else
{
opencc_perror(_("OpenCC error"));
break;
}
}
opencc_close(od);
free(buffer_in);
free(buffer_out);
fclose(fp);
fclose(fpo);
}
/**
* 该函数装载data/gbks2t.tab的简体转繁体的码表，
* 然后按码表将GBK字符转换成GBK繁体字符。
*
* WARNING： 该函数返回新分配内存字符串，请调用者
* 注意释放。
*/
char *ConvertGBKTradition2Simple(FcitxChttrans* transState, const char *strHZ)
{
if (strHZ == NULL)
return NULL;
switch (transState->engine) {
case ENGINE_OPENCC:
#ifdef _ENABLE_OPENCC
{
if (transState->odt2s == NULL) {
transState->odt2s = opencc_open(OPENCC_DEFAULT_CONFIG_TRAD_TO_SIMP);
if (transState->odt2s == NULL) {
opencc_perror(_("OpenCC initialization error"));
return NULL;
}
}
char * res = opencc_convert_utf8(transState->odt2s, strHZ, (size_t) - 1);
if (res == (char *) - 1) {
opencc_perror(_("OpenCC error"));
return NULL;
}
return res;
}
#endif
case ENGINE_NATIVE: {
FILE *fp;
char *ret;
int i, len, ret_len;
char *strBuf = NULL;
size_t bufLen = 0;
const char *ps;
if (!transState->t2s_table) {
len = 0;
fp = FcitxXDGGetFileWithPrefix("data", TABLE_GBKS2T, "r", NULL);
if (!fp) {
ret = (char *) malloc(sizeof(char) * (strlen(strHZ) + 1));
strcpy(ret, strHZ);
return ret;
}
while (getline(&strBuf, &bufLen, fp) != -1) {
simple2trad_t *t2s;
char *ps;
unsigned int wc;
ps = fcitx_utf8_get_char(strBuf, &wc);
t2s = (simple2trad_t*) malloc(sizeof(simple2trad_t));
fcitx_utf8_get_char(ps, &wc);
t2s->wc = wc;
t2s->len = fcitx_utf8_char_len(strBuf);
strncpy(t2s->str, strBuf, t2s->len);
t2s->str[t2s->len] = '\0';
HASH_ADD_INT(transState->t2s_table, wc, t2s);
}
if (strBuf)
free(strBuf);
}
i = 0;
len = fcitx_utf8_strlen(strHZ);
ret_len = 0;
ret = (char *) fcitx_utils_malloc0(sizeof(char) * (UTF8_MAX_LENGTH * len + 1));
ps = strHZ;
ret[0] = '\0';
for (; i < len; ++i) {
unsigned int wc;
simple2trad_t *t2s = NULL;
int chr_len = fcitx_utf8_char_len(ps);
char *nps;
nps = fcitx_utf8_get_char(ps , &wc);
HASH_FIND_INT(transState->t2s_table, &wc, t2s);
if (t2s) {
strcat(ret, t2s->str);
ret_len += t2s->len;
} else {
strncat(ret, ps, chr_len);
ret_len += chr_len;
}
ps = nps;
}
ret[ret_len] = '\0';
return ret;
}
}
return NULL;
}
void convert(const char * input_file, const char * output_file, const char * config_file)
{
opencc_t od = opencc_open(config_file);
if (od == (opencc_t) -1)
{
opencc_perror(_("OpenCC initialization error"));
exit(1);
}
FILE * fp = stdin;
FILE * fpo = stdout;
if (input_file)
{
fp = fopen(input_file, "r");
if (!fp)
{
fprintf(stderr, _("Can not read file: %s\n"), input_file);
exit(1);
}
}
if (output_file)
{
fpo = fopen(output_file, "w");
if (!fpo)
{
fprintf(stderr, _("Can not write file: %s\n"), output_file);
exit(1);
}
}
size_t size = BUFFER_SIZE;
char * buffer_in = NULL, * buffer_out = NULL;
buffer_in = (char *) malloc(size * sizeof(char));
while (!feof(fp))
{
size_t read = fread(buffer_in, 1, size, fp);
// If we haven't finished reading after filling the entire buffer,
// then it could be that we broke within an UTF-8 character, in
// that case we must backtrack and find the boundary
if (read == size) {
// Find the boundary of last UTF-8 character
int i;
for (i = read - 1; i >= 0; i--)
{
char c = buffer_in[i];
if (!(c & 0x80) || ((c & 0xC0) == 0xC0))
break;
}
if (i >= 0) {
buffer_in[i] = '\0';
// Since we read a bit too much, move file pointer back
// just that much so that we can continue reading
size_t n = read - i;
fseek(fp, -n, SEEK_CUR);
}
}
buffer_out = opencc_convert_utf8(od, buffer_in, (size_t) -1);
if (buffer_out != (char *) -1)
{
fprintf(fpo, "%s", buffer_out);
free(buffer_out);
}
else
{
opencc_perror(_("OpenCC error"));
break;
}
}
opencc_close(od);
free(buffer_in);
fclose(fp);
fclose(fpo);
}
