int DpsChineseListLoad(DPS_AGENT *Agent, DPS_CHINALIST *List, const char *charset, const char *fname) {
struct stat sb;
char *str, *data = NULL, *cur_n = NULL;
DPS_CHINAWORD chinaword;
char word[PATH_MAX];
dpsunicode_t uword[256];
DPS_CHARSET *sys_int, *fcs;
DPS_CONV to_uni;
int fd;
char savebyte;
sys_int = DpsGetCharSet("sys-int");
if (!(fcs = DpsGetCharSet(charset))) {
if (Agent->Conf->is_log_open) DpsLog(Agent, DPS_LOG_ERROR, "Charset '%s' not found or not supported", charset);
else fprintf(stderr, "Charset '%s' not found or not supported", charset);
return DPS_ERROR;
}
DpsConvInit(&to_uni, fcs, sys_int, Agent->Conf->CharsToEscape, DPS_RECODE_HTML);
if (*fname != '/') {
dps_snprintf(word, sizeof(word), "%s/%s", DpsVarListFindStr(&Agent->Conf->Vars, "EtcDir", DPS_CONF_DIR), fname);
fname = word;
}
if (stat(fname, &sb)) {
if (Agent->Conf->is_log_open)
DpsLog(Agent, DPS_LOG_ERROR, "Unable to stat FreqDic file '%s': %s", fname, strerror(errno));
else fprintf(stderr, "Unable to stat FrecDic file '%s': %s", fname, strerror(errno));
return DPS_ERROR;
}
if ((fd = open(fname, O_RDONLY)) <= 0) {
if (Agent->Conf->is_log_open)
DpsLog(Agent, DPS_LOG_ERROR, "Unable to open FreqDic file '%s': %s", fname, strerror(errno));
else fprintf(stderr, "Unable to open FreqDic file '%s': %s", fname, strerror(errno));
return DPS_ERROR;
}
if ((data = (char*)DpsMalloc(sb.st_size + 1)) == NULL) {
if (Agent->Conf->is_log_open)
DpsLog(Agent, DPS_LOG_ERROR, "Unable to alloc %d bytes", sb.st_size);
else fprintf(stderr, "Unable to alloc %ld bytes", (long)sb.st_size);
close(fd);
return DPS_ERROR;
}
if (read(fd, data, sb.st_size) != (ssize_t)sb.st_size) {
if (Agent->Conf->is_log_open)
DpsLog(Agent, DPS_LOG_ERROR, "Unable to read FreqDic file '%s': %s", fname, strerror(errno));
else fprintf(stderr, "Unable to read FreqDic file '%s': %s", fname, strerror(errno));
DPS_FREE(data);
close(fd);
return DPS_ERROR;
}
data[sb.st_size] = '\0';
str = data;
cur_n = strchr(str, NL_INT);
if (cur_n != NULL) {
cur_n++;
savebyte = *cur_n;
*cur_n = '\0';
}
close(fd);
bzero((void*)&chinaword, sizeof(chinaword));
chinaword.word = uword;
while(str != NULL) {
if(!str[0]) goto loop_continue;
if(str[0]=='#') goto loop_continue;
sscanf(str, "%d %63s ", &chinaword.freq, word );
DpsConv(&to_uni, (char*)uword, sizeof(uword), word, sizeof(word));
DpsChineseListAdd(List, &chinaword);
loop_continue:
str = cur_n;
if (str != NULL) {
*str = savebyte;
cur_n = strchr(str, NL_INT);
if (cur_n != NULL) {
cur_n++;
savebyte = *cur_n;
*cur_n = '\0';
}
}
}
DPS_FREE(data);
DpsChineseListSort(List);
{ register size_t i, j = 0;
for (i = 1; i < List->nwords; i++) {
if (cmpchinese(&List->ChiWord[j], &List->ChiWord[i]) == 0) {
List->ChiWord[j].freq += List->ChiWord[i].freq;
} else { j++;
}
}
for (i = j + 1; i < List->nwords; i++) {
DPS_FREE(List->ChiWord[i].word);
}
List->nwords = j + 1;
}
return DPS_OK;
}
static int DpsUniRegExec(const DPS_UNIREG_EXP *reg, const dpsunicode_t *string) {
const dpsunicode_t *start = string;
int match=0;
#ifdef DEBUG_UNIREG
DPS_CHARSET *k = DpsGetCharSet("koi8-r");
DPS_CHARSET *sy = DpsGetCharSet("sys-int");
DPS_CONV fromuni;
char sstr[1024];
char rstr[1024];
DpsConvInit(&fromuni, sy, k, NULL, 0);
#endif
for(start=string;*start;start++){
const dpsunicode_t *tstart=start;
size_t i;
for(i=0;i<reg->ntokens;i++){
const dpsunicode_t *s;
int inc=DPS_UNIREG_INC;
#ifdef DEBUG_UNIREG
DpsConv(&fromuni, sstr, 1024, (char*)tstart, 1024);
DpsConv(&fromuni, rstr, 1024, (char*)reg->Token[i].str, 1024);
printf("t:%d tstart='%s'\ttok='%s'\t", i, sstr, rstr);
#endif
switch(reg->Token[i].str[0]){
case '^':
if(string!=tstart){
match=0;
}else{
match=1;
}
break;
case '[':
match=0;
for(s=reg->Token[i].str+1;*s;s++){
if(*s==']'){
}else
if(*s=='^'){
inc=DPS_UNIREG_EXC;
match=1;
}else{
if((*tstart==*s)&&(inc==DPS_UNIREG_EXC)){
match=0;
break;
}
if((*tstart==*s)&&(inc==DPS_UNIREG_INC)){
match=1;
break;
}
}
}
tstart++;
break;
case '$':
if(*tstart!=0){
match=0;
}else{
match=1;
}
break;
default:
match=1;
for(s=reg->Token[i].str;(*s)&&(*tstart);s++,tstart++){
if(*s=='.'){
/* Any char */
}else
if((*s)!=(*tstart)){
match=0;
break;
}
}
if((*s)&&(!*tstart))match=0;
break;
}
#ifdef DEBUG_UNIREG
printf("match=%d\n",match);
#endif
if(!match)break;
}
if(match)break;
}
#ifdef DEBUG_UNIREG
printf("return match=%d\n",match);
#endif
return match;
}
__C_LINK int __DPSCALL DpsImportDictionary(DPS_ENV * Conf, const char *lang, const char *charset,
const char *filename, int skip_noflag, const char *first_letters){
struct stat sb;
char *str, *data = NULL, *cur_n = NULL;
char *lstr;
dpsunicode_t *ustr;
DPS_CHARSET *sys_int;
DPS_CHARSET *dict_charset;
DPS_CONV touni;
DPS_CONV fromuni;
int fd;
char savebyte;
if ((lstr = (char*) DpsMalloc(2048)) == NULL) {
DPS_FREE(str);
return DPS_ERROR;
}
if ((ustr = (dpsunicode_t*) DpsMalloc(8192)) == NULL) {
DPS_FREE(lstr);
return DPS_ERROR;
}
dict_charset = DpsGetCharSet(charset);
sys_int = DpsGetCharSet("sys-int");
if ((dict_charset == NULL) || (sys_int == NULL)) {
DPS_FREE(lstr);
DPS_FREE(ustr);
return DPS_ERROR;
}
DpsConvInit(&touni, dict_charset, sys_int, Conf->CharsToEscape, 0);
DpsConvInit(&fromuni, sys_int, dict_charset, Conf->CharsToEscape, 0);
if (stat(filename, &sb)) {
fprintf(stderr, "Unable to stat synonyms file '%s': %s", filename, strerror(errno));
DPS_FREE(lstr);
DPS_FREE(ustr);
return DPS_ERROR;
}
if ((fd = DpsOpen2(filename, O_RDONLY)) <= 0) {
fprintf(stderr, "Unable to open synonyms file '%s': %s", filename, strerror(errno));
return DPS_ERROR;
}
if ((data = (char*)DpsMalloc(sb.st_size + 1)) == NULL) {
fprintf(stderr, "Unable to alloc %ld bytes", (long)sb.st_size);
DpsClose(fd);
DPS_FREE(lstr);
DPS_FREE(ustr);
return DPS_ERROR;
}
if (read(fd, data, sb.st_size) != (ssize_t)sb.st_size) {
fprintf(stderr, "Unable to read synonym file '%s': %s", filename, strerror(errno));
DPS_FREE(data);
DpsClose(fd);
DPS_FREE(lstr);
DPS_FREE(ustr);
return DPS_ERROR;
}
data[sb.st_size] = '\0';
str = data;
cur_n = strchr(str, '\n');
if (cur_n != NULL) {
cur_n++;
savebyte = *cur_n;
*cur_n = '\0';
}
DpsClose(fd);
while(str != NULL) {
char *s;
const char *flag;
int res;
flag = NULL;
s = str;
while(*s){
if(*s == '\r') *s = '\0';
if(*s == '\n') *s = '\0';
s++;
}
if((s=strchr(str,'/'))){
*s=0;
s++;flag=s;
while(*s){
if(((*s>='A')&&(*s<='Z'))||((*s>='a')&&(*s<='z')))s++;
else{
*s=0;
break;
}
}
}else{
if(skip_noflag) goto loop_continue;
flag="";
}
res = DpsConv(&touni, (char*)ustr, 8192, str, 1024);
DpsUniStrToLower(ustr);
/* Dont load words if first letter is not required */
/* It allows to optimize loading at search time */
if(*first_letters) {
DpsConv(&fromuni, lstr, 2048, ((const char*)ustr),(size_t)res);
if(!strchr(first_letters,lstr[0]))
goto loop_continue;
}
res = DpsSpellAdd(&Conf->Spells,ustr,flag,lang);
if (res != DPS_OK) {
DPS_FREE(lstr);
DPS_FREE(ustr); DPS_FREE(data);
return res;
}
if (Conf->Flags.use_accentext) {
dpsunicode_t *af_uwrd = DpsUniAccentStrip(ustr);
if (DpsUniStrCmp(af_uwrd, ustr) != 0) {
res = DpsSpellAdd(&Conf->Spells, af_uwrd, flag, lang);
if (res != DPS_OK) {
DPS_FREE(lstr);
DPS_FREE(ustr); DPS_FREE(data); DPS_FREE(af_uwrd);
return res;
}
}
DPS_FREE(af_uwrd);
if (strncasecmp(lang, "de", 2) == 0) {
dpsunicode_t *de_uwrd = DpsUniGermanReplace(ustr);
if (DpsUniStrCmp(de_uwrd, ustr) != 0) {
res = DpsSpellAdd(&Conf->Spells, de_uwrd, flag, lang);
if (res != DPS_OK) {
DPS_FREE(lstr);
DPS_FREE(ustr); DPS_FREE(data); DPS_FREE(de_uwrd);
return res;
}
}
DPS_FREE(de_uwrd);
}
}
loop_continue:
str = cur_n;
if (str != NULL) {
*str = savebyte;
cur_n = strchr(str, '\n');
if (cur_n != NULL) {
cur_n++;
savebyte = *cur_n;
*cur_n = '\0';
}
}
}
DPS_FREE(data);
DPS_FREE(lstr);
DPS_FREE(ustr);
return DPS_OK;
}
__C_LINK int __DPSCALL DpsSynonymListLoad(DPS_ENV * Env,const char * filename){
struct stat sb;
char *str, *data = NULL, *cur_n = NULL;
char lang[64]="";
DPS_CHARSET *cs=NULL;
DPS_CHARSET *sys_int=DpsGetCharSet("sys-int");
DPS_CONV file_uni;
DPS_WIDEWORD *ww = NULL;
size_t key = 1;
int flag_th = 0;
int fd;
char savebyte;
if (stat(filename, &sb)) {
fprintf(stderr, "Unable to stat synonyms file '%s': %s", filename, strerror(errno));
return DPS_ERROR;
}
if ((fd = DpsOpen2(filename, O_RDONLY)) <= 0) {
dps_snprintf(Env->errstr,sizeof(Env->errstr)-1, "Unable to open synonyms file '%s': %s", filename, strerror(errno));
return DPS_ERROR;
}
if ((data = (char*)DpsMalloc(sb.st_size + 1)) == NULL) {
dps_snprintf(Env->errstr,sizeof(Env->errstr)-1, "Unable to alloc %d bytes", sb.st_size);
DpsClose(fd);
return DPS_ERROR;
}
if (read(fd, data, sb.st_size) != (ssize_t)sb.st_size) {
dps_snprintf(Env->errstr,sizeof(Env->errstr)-1, "Unable to read synonym file '%s': %s", filename, strerror(errno));
DPS_FREE(data);
DpsClose(fd);
return DPS_ERROR;
}
data[sb.st_size] = '\0';
str = data;
cur_n = strchr(str, '\n');
if (cur_n != NULL) {
cur_n++;
savebyte = *cur_n;
*cur_n = '\0';
}
while(str != NULL) {
if(str[0]=='#'||str[0]==' '||str[0]=='\t'||str[0]=='\r'||str[0]=='\n') goto loop_continue;
if(!strncasecmp(str,"Charset:",8)){
char * lasttok;
char * charset;
if((charset = dps_strtok_r(str + 8, " \t\n\r", &lasttok))) {
cs=DpsGetCharSet(charset);
if(!cs){
dps_snprintf(Env->errstr, sizeof(Env->errstr), "Unknown charset '%s' in synonyms file '%s'",
charset, filename);
DPS_FREE(data);
DpsClose(fd);
return DPS_ERROR;
}
DpsConvInit(&file_uni, cs, sys_int, Env->CharsToEscape, 0);
}
}else
if(!strncasecmp(str,"Language:",9)){
char * lasttok;
char * l;
if((l = dps_strtok_r(str + 9, " \t\n\r", &lasttok))) {
dps_strncpy(lang, l, sizeof(lang)-1);
}
}else
if(!strncasecmp(str, "Thesaurus:", 10)) {
char * lasttok;
char *tok = dps_strtok_r(str + 10, " \t\n\r", &lasttok);
flag_th = (strncasecmp(tok, "yes", 3) == 0) ? 1 : 0;
}else{
char *av[255];
size_t ac, i, j;
dpsunicode_t *t;
if(!cs){
dps_snprintf(Env->errstr,sizeof(Env->errstr)-1,"No Charset command in synonyms file '%s'",filename);
DpsClose(fd); DPS_FREE(data);
return DPS_ERROR;
}
if(!lang[0]){
dps_snprintf(Env->errstr,sizeof(Env->errstr)-1,"No Language command in synonyms file '%s'",filename);
DpsClose(fd); DPS_FREE(data);
return DPS_ERROR;
}
ac = DpsGetArgs(str, av, 255);
if (ac < 2) goto loop_continue;
if ((ww = (DPS_WIDEWORD*)DpsRealloc(ww, ac * sizeof(DPS_WIDEWORD))) == NULL) return DPS_ERROR;
for (i = 0; i < ac; i++) {
ww[i].word = av[i];
ww[i].len = dps_strlen(av[i]);
ww[i].uword = t = (dpsunicode_t*)DpsMalloc((3 * ww[i].len + 1) * sizeof(dpsunicode_t));
if (ww[i].uword == NULL) return DPS_ERROR;
DpsConv(&file_uni, (char*)ww[i].uword, sizeof(dpsunicode_t) * (3 * ww[i].len + 1), av[i], ww[i].len + 1);
DpsUniStrToLower(ww[i].uword);
ww[i].uword = DpsUniNormalizeNFC(NULL, ww[i].uword);
DPS_FREE(t);
}
for (i = 0; i < ac - 1; i++) {
for (j = i + 1; j < ac; j++) {
if((Env->Synonyms.nsynonyms + 1) >= Env->Synonyms.msynonyms){
Env->Synonyms.msynonyms += 64;
Env->Synonyms.Synonym = (DPS_SYNONYM*)DpsRealloc(Env->Synonyms.Synonym,
sizeof(DPS_SYNONYM)*Env->Synonyms.msynonyms);
if (Env->Synonyms.Synonym == NULL) {
Env->Synonyms.msynonyms = Env->Synonyms.nsynonyms = 0;
return DPS_ERROR;
}
}
bzero((void*)&Env->Synonyms.Synonym[Env->Synonyms.nsynonyms], sizeof(DPS_SYNONYM));
/* Add direct order */
Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].p.uword = DpsUniDup(ww[i].uword);
Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].s.uword = DpsUniDup(ww[j].uword);
Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].p.count =
Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].s.count = (size_t)((flag_th) ? key : 0);
Env->Synonyms.nsynonyms++;
bzero((void*)&Env->Synonyms.Synonym[Env->Synonyms.nsynonyms], sizeof(DPS_SYNONYM));
/* Add reverse order */
Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].p.uword = DpsUniDup(ww[j].uword);
Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].s.uword = DpsUniDup(ww[i].uword);
Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].p.count =
Env->Synonyms.Synonym[Env->Synonyms.nsynonyms].s.count = (size_t)((flag_th) ? key : 0);
Env->Synonyms.nsynonyms++;
}
}
for (i = 0; i < ac; i++) {
DPS_FREE(ww[i].uword);
}
do { key++; } while (key == 0);
}
loop_continue:
str = cur_n;
if (str != NULL) {
*str = savebyte;
cur_n = strchr(str, '\n');
if (cur_n != NULL) {
cur_n++;
savebyte = *cur_n;
*cur_n = '\0';
}
}
}
DPS_FREE(data);
DPS_FREE(ww);
DpsClose(fd);
return DPS_OK;
}