/*
*根据字串判断词库中是否有某个字/词，注意该函数会忽略拼音词组
*/
RECORD *TableFindPhrase(const TableDict* tableDict, const char *strHZ)
{
RECORD *recTemp;
char strTemp[UTF8_MAX_LENGTH + 1];
int i;
//首先，先查找第一个汉字的编码
strncpy(strTemp, strHZ, fcitx_utf8_char_len(strHZ));
strTemp[fcitx_utf8_char_len(strHZ)] = '\0';
recTemp = tableDict->tableSingleHZ[CalHZIndex(strTemp)];
if (!recTemp)
return (RECORD *) NULL;
//然后根据该编码找到检索的起始点
i = 0;
while (recTemp->strCode[0] != tableDict->recordIndex[i].cCode)
i++;
recTemp = tableDict->recordIndex[i].record;
while (recTemp != tableDict->recordHead) {
if (recTemp->strCode[0] != tableDict->recordIndex[i].cCode)
break;
if (!strcmp(recTemp->strHZ, strHZ)) {
if (recTemp->type != RECORDTYPE_PINYIN)
return recTemp;
}
recTemp = recTemp->next;
}
return (RECORD *) NULL;
}
SkinImage* LoadImageWithText(FcitxClassicUI* classicui, FcitxSkin* sc, const char* name, const char* text, int w, int h, boolean active)
{
if (!text || *text == '\0')
return NULL;
UnloadSingleImage(sc, name);
int len = fcitx_utf8_char_len(text);
if (len == 1 && text[len] && fcitx_utf8_char_len(text + len) == 1)
len = 2;
char* iconText = strndup(text, len);
FcitxLog(DEBUG, "%s", iconText);
cairo_surface_t* newsurface = cairo_image_surface_create(CAIRO_FORMAT_ARGB32, w, h);
cairo_t* c = cairo_create(newsurface);
int min = w > h? h: w;
min = min * 0.7;
cairo_set_operator(c, CAIRO_OPERATOR_SOURCE);
cairo_set_source_rgba(c ,1, 1, 1, 0.0);
cairo_paint(c);
FcitxConfigColor color;
if (sc->skinMainBar.bUseCustomTextIconColor) {
if (active)
color = sc->skinMainBar.textIconColor[0];
else
color = sc->skinMainBar.textIconColor[1];
}
else
color = sc->skinFont.menuFontColor[1];
int textw, texth;
FcitxCairoTextContext* ctc = FcitxCairoTextContextCreate(c);
FcitxCairoTextContextSet(ctc, classicui->font, min, false);
FcitxCairoTextContextStringSizeStrict(ctc, iconText, &textw, &texth);
FcitxCairoTextContextOutputString(ctc, iconText, (w - textw) * 0.5, 0, &color);
FcitxCairoTextContextFree(ctc);
cairo_destroy(c);
SkinImage* image = fcitx_utils_malloc0(sizeof(SkinImage));
image->name = strdup(name);
image->image = newsurface;
image->textIcon = true;
HASH_ADD_KEYPTR(hh, sc->imageTable, image->name, strlen(image->name), image);
return image;
}
unsigned int CalHZIndex(char *strHZ)
{
unsigned int iutf = 0;
int l = fcitx_utf8_char_len(strHZ);
unsigned char* utf = (unsigned char*) strHZ;
unsigned int *res;
int idx;
if (l == 2) {
iutf = *utf++ << 8;
iutf |= *utf++;
} else if (l == 3) {
iutf = *utf++ << 16;
iutf |= *utf++ << 8;
iutf |= *utf++;
} else if (l == 4) {
iutf = *utf++ << 24;
iutf |= *utf++ << 16;
iutf |= *utf++ << 8;
iutf |= *utf++;
}
res = bsearch(&iutf, fcitx_utf8_in_gb18030, 63360, sizeof(int), cmpi);
if (res)
idx = res - fcitx_utf8_in_gb18030;
else
idx = 63361;
return idx;
}
char* ProcessFullWidthChar(void* arg, const char* str)
{
FcitxFullWidthChar* fwchar = (FcitxFullWidthChar*)arg;
FcitxProfile* profile = FcitxInstanceGetProfile(fwchar->owner);
if (profile->bUseFullWidthChar) {
size_t i = 0, ret_len = 0, len = fcitx_utf8_strlen(str);
char* ret = (char *) fcitx_utils_malloc0(sizeof(char) * (UTF8_MAX_LENGTH * len + 1));
const char* ps = str;
ret[0] = '\0';
for (; i < len; ++i) {
int wc;
int chr_len = fcitx_utf8_char_len(ps);
char *nps;
nps = fcitx_utf8_get_char(ps , &wc);
if (chr_len == 1 && ps[0] >= '\x20' && ps[0] <= '\x7e') {
strcat(ret, sCornerTrans[ps[0] - 32]);
ret_len += strlen(sCornerTrans[ps[0] - 32]);
} else {
strncat(ret, ps, chr_len);
ret_len += chr_len;
}
ps = nps;
}
ret[ret_len] = '\0';
return ret;
}
else
return NULL;
}
/**
* @brief Get the non-utf8 cursor pos for fcitx
*
* @return int
**/
static int FcitxChewingGetRawCursorPos(char * str, int upos)
{
unsigned int i;
int pos = 0;
for (i = 0; i < upos; i++) {
pos += fcitx_utf8_char_len(fcitx_utf8_get_nth_char(str, i));
}
return pos;
}
/**
* 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
do {
if (transState->odt2s == NULL) {
OpenCCInit(transState);
if (transState->odt2s == NULL) {
break;
}
}
char * res = OpenCCConvert(transState->odt2s, strHZ, (size_t) - 1);
if (!res || res == (char *) - 1) {
return NULL;
}
return res;
} while(0);
#endif
case ENGINE_NATIVE: {
FILE *fp;
char *ret;
int i, len, ret_len;
const char *ps;
if (!transState->t2s_table) {
char *strBuf = NULL;
size_t bufLen = 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 = NULL;
char *ps;
uint32_t wc;
ps = fcitx_utf8_get_char(strBuf, &wc);
HASH_FIND_INT(transState->s2t_table, &wc, t2s);
if (t2s) {
continue;
}
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);
}
fcitx_utils_free(strBuf);
}
i = 0;
len = fcitx_utf8_strlen(strHZ);
ret_len = 0;
ret = fcitx_utils_malloc0(UTF8_MAX_LENGTH * len + 1);
ps = strHZ;
ret[0] = '\0';
for (; i < len; ++i) {
uint32_t 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;
}
/**
* 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;
}
boolean TableCreatePhraseCode(TableDict* tableDict, char *strHZ)
{
unsigned char i;
unsigned char i1, i2;
size_t iLen;
char strTemp[UTF8_MAX_LENGTH + 1] = {'\0', };
RECORD *recTemp;
boolean bCanntFindCode = false;
iLen = fcitx_utf8_strlen(strHZ);
if (iLen >= tableDict->iCodeLength) {
i2 = tableDict->iCodeLength;
i1 = 1;
} else {
i2 = iLen;
i1 = 0;
}
for (i = 0; i < tableDict->iCodeLength - 1; i++) {
if (tableDict->rule[i].iWords == i2 && tableDict->rule[i].iFlag == i1)
break;
}
if (i == tableDict->iCodeLength - 1)
return true;
int codeIdx = 0;
for (i1 = 0; i1 < tableDict->iCodeLength; i1++) {
int clen;
char* ps;
if (tableDict->rule[i].rule[i1].iFlag) {
ps = fcitx_utf8_get_nth_char(strHZ, tableDict->rule[i].rule[i1].iWhich - 1);
clen = fcitx_utf8_char_len(ps);
strncpy(strTemp, ps, clen);
} else {
ps = fcitx_utf8_get_nth_char(strHZ, iLen - tableDict->rule[i].rule[i1].iWhich);
clen = fcitx_utf8_char_len(ps);
strncpy(strTemp, ps, clen);
}
int hzIndex = CalHZIndex(strTemp);
if (tableDict->tableSingleHZ[hzIndex]) {
if (tableDict->tableSingleHZCons[hzIndex])
recTemp = tableDict->tableSingleHZCons[hzIndex];
else
recTemp = tableDict->tableSingleHZ[hzIndex];
}
else {
bCanntFindCode = true;
break;
}
if (strlen(recTemp->strCode) >= tableDict->rule[i].rule[i1].iIndex) {
tableDict->strNewPhraseCode[codeIdx] = recTemp->strCode[tableDict->rule[i].rule[i1].iIndex - 1];
codeIdx++;
}
}
return bCanntFindCode;
}
void CreatePYBase(void)
{
_PyStruct *head, *pyList, *temp, *t;
char strPY[7], strHZ[UTF8_MAX_LENGTH * 80 + 1], strMap[3];
int iIndex, iCount, i;
int iBaseCount;
int s = 0;
int tt = 0;
head = (_PyStruct *) malloc(sizeof(_PyStruct));
head->prev = head;
head->next = head;
iBaseCount = 0;
while (PYTable_template[iBaseCount].strPY[0] != '\0')
iBaseCount++;
for (iIndex = 0; iIndex < iBaseCount; iIndex++)
YY[iIndex] = 0;
iIndex = 0;
while (!feof(fps)) {
fscanf(fps, "%s", strPY);
fscanf(fps, "%s\n", strHZ);
if (MapPY(&pyconfig, strPY, strMap, PY_PARSE_INPUT_SYSTEM)) {
for (i = 0; i < iBaseCount; i++)
if ((!strcmp(PYTable_template[i].strPY, strPY)) && PYTable_template[i].control == PYTABLE_NONE)
YY[i] += 1;
iIndex++;
if (fcitx_utf8_strlen(strHZ) > 1) {
int8_t charLen = fcitx_utf8_char_len(strHZ);
fprintf(stderr, "%s length is larger that 1, truncated to ", strHZ);
strHZ[charLen] = '\0';
fprintf(stderr, "%s.\n", strHZ);
}
temp = (_PyStruct *) malloc(sizeof(_PyStruct));
strcpy(temp->strHZ, strHZ);
strcpy(temp->strPY, strMap);
pyList = head->prev;
while (pyList != head) {
if (strcmp(pyList->strPY, strMap) <= 0)
break;
pyList = pyList->prev;
}
temp->next = pyList->next;
temp->prev = pyList;
pyList->next->prev = temp;
pyList->next = temp;
} else
fprintf(stderr, "%s Error!!!!\n", strPY);
}
iCount = 0;
for (i = 0; i < iBaseCount; i++) {
if (YY[i])
iCount++;
}
fwrite(&iCount, sizeof(int), 1, fp1);
printf("Groups: %d\n", iCount);
iAllCount = iIndex;
pyList = head->next;
strcpy(strPY, pyList->strPY);
iCount = 0;
t = pyList;
while (pyList != head) {
if (!strcmp(strPY, pyList->strPY)) {
iCount++;
} else {
tt++;
fwrite(strPY, sizeof(char) * 2, 1, fp1);
fwrite(&iCount, sizeof(int), 1, fp1);
for (i = 0; i < iCount; i++) {
int8_t len = strlen(t->strHZ);
fwrite(&len, sizeof(int8_t), 1, fp1);
fwrite(t->strHZ, sizeof(char) * len , 1, fp1);
t = t->next;
}
s += iCount;
t = pyList;
iCount = 1;
strcpy(strPY, pyList->strPY);
}
pyList = pyList->next;
}
fwrite(strPY, sizeof(char) * 2, 1, fp1);
fwrite(&iCount, sizeof(int), 1, fp1);
for (i = 0; i < iCount; i++) {
int8_t len = strlen(t->strHZ);
fwrite(&len, sizeof(int8_t), 1, fp1);
fwrite(t->strHZ, sizeof(char) * len , 1, fp1);
t = t->next;
}
s += iCount;
fclose(fp1);
fclose(fps);
}
void CreatePYPhrase(void)
{
char strPY[256];
char strPhrase[256];
char strMap[256];
ParsePYStruct strTemp;
int iIndex, i, s1, s2, j, k;
_PyPhrase *phrase, *t, *tt;
FILE *f = fopen("pyERROR", "w");
FILE *fg = fopen("pyPhrase.ok", "w");
int kkk;
unsigned int uIndex, uTemp;
FcitxPinyinConfig pyconfig;
memset(&pyconfig, 0 , sizeof(pyconfig));
InitMHPY(&pyconfig.MHPY_C, MHPY_C_TEMPLATE);
InitMHPY(&pyconfig.MHPY_S, MHPY_S_TEMPLATE);
InitPYTable(&pyconfig);
s1 = 0;
s2 = 0;
uIndex = 0;
printf("Start Loading Phrase...\n");
while (!feof(fpt)) {
fscanf(fpt, "%s", strPY);
fscanf(fpt, "%s\n", strPhrase);
if (strlen(strPhrase) < 3)
continue;
ParsePY(&pyconfig, strPY, &strTemp, PY_PARSE_INPUT_SYSTEM, false);
s2++;
kkk = 0;
if (strTemp.iHZCount != fcitx_utf8_strlen(strPhrase) || (strTemp.iMode & PARSE_ABBR)) {
fprintf(f, "%s %s\n", strPY, strPhrase);
continue;
}
strMap[0] = '\0';
for (iIndex = 0; iIndex < strTemp.iHZCount; iIndex++)
strcat(strMap, strTemp.strMap[iIndex]);
for (iIndex = 0; iIndex < iPYFACount; iIndex++) {
if (!strncmp(PYFAList[iIndex].strMap, strMap, 2)) {
for (i = 0; i < PYFAList[iIndex].iHZCount; i++) {
if (!fcitx_utf8_strncmp(PYFAList[iIndex].pyBase[i].strHZ, strPhrase, 1)) {
t = PYFAList[iIndex].pyBase[i].phrase;
for (j = 0; j < PYFAList[iIndex].pyBase[i].iPhraseCount; j++) {
tt = t;
t = t->next;
if (!strcmp(t->strMap, strMap + 2) && !strcmp(t->strPhrase, strPhrase + fcitx_utf8_char_len(strPhrase))) {
printf("\n\t%d: %s %s ----->deleted.\n", s2, strPY, strPhrase);
goto _next;
}
if (strcmp(t->strMap, strMap + 2) > 0) {
t = tt;
break;
}
}
phrase = (_PyPhrase *) malloc(sizeof(_PyPhrase));
phrase->strPhrase = (char *) malloc(sizeof(char) * (strlen(strPhrase) - fcitx_utf8_char_len(strPhrase) + 1));
phrase->strMap = (char *) malloc(sizeof(char) * ((strTemp.iHZCount - 1) * 2 + 1));
phrase->uIndex = uIndex++;
strcpy(phrase->strPhrase, strPhrase + fcitx_utf8_char_len(strPhrase));
strcpy(phrase->strMap, strMap + 2);
tt = t->next;
t->next = phrase;
phrase->next = tt;
PYFAList[iIndex].pyBase[i].iPhraseCount++;
s1++;
kkk = 1;
_next:
;
}
}
}
}
if (!kkk)
fprintf(f, "%s %s %s\n", strPY, strPhrase, (char *)(strTemp.strPYParsed));
else
fprintf(fg, "%s %s\n", strPY, strPhrase);
}
printf("%d Phrases, %d Converted!\nWriting Phrase file ...", s2, s1);
for (i = 0; i < iPYFACount; i++) {
for (j = 0; j < PYFAList[i].iHZCount; j++) {
iIndex = PYFAList[i].pyBase[j].iPhraseCount;
if (iIndex) {
int8_t clen = strlen(PYFAList[i].pyBase[j].strHZ);
fwrite(&i, sizeof(int), 1, fp2);
fwrite(&clen, sizeof(int8_t), 1, fp2);
fwrite(PYFAList[i].pyBase[j].strHZ, sizeof(char) * clen, 1, fp2);
fwrite(&iIndex, sizeof(int), 1, fp2);
t = PYFAList[i].pyBase[j].phrase->next;
for (k = 0; k < PYFAList[i].pyBase[j].iPhraseCount; k++) {
int slen = strlen(t->strPhrase);
iIndex = strlen(t->strMap);
fwrite(&iIndex, sizeof(int), 1, fp2);
fwrite(t->strMap, sizeof(char), iIndex, fp2);
fwrite(&slen, sizeof(int), 1, fp2);
fwrite(t->strPhrase, sizeof(char), strlen(t->strPhrase), fp2);
uTemp = uIndex - 1 - t->uIndex;
fwrite(&uTemp, sizeof(unsigned int), 1, fp2);
t = t->next;
}
}
}
}
printf("\nOK!\n");
fclose(fp2);
fclose(fpt);
}