/**
* Loads a .fst2 file with the given name and type, according to the
* given language description.
*/
Elag_fst_file_in* load_elag_fst2_file(const VersatileEncodingConfig* vec,const char* fname,language_t* language) {
Elag_fst_file_in* fstf=(Elag_fst_file_in*)malloc(sizeof(Elag_fst_file_in));
if (fstf==NULL) {
fatal_alloc_error("load_elag_fst2_file");
}
fstf->name=strdup(fname);
if (fstf->name==NULL) {
fatal_alloc_error("load_elag_fst2_file");
}
if ((fstf->f=u_fopen(vec,fname,U_READ))==NULL) {
error("load_fst_file: unable to open '%s' for reading\n",fname);
goto error_fstf;
}
unichar buf[MAXBUF];
if (u_fgets(buf,MAXBUF,fstf->f)==EOF) {
error("load_fst_file: '%s' is empty\n",fname);
goto error_f;
}
if (!u_is_digit(*buf)) {
error("load_fst_file: %s: bad file format\n",fname);
goto error_f;
}
fstf->nb_automata=u_parse_int(buf);
fstf->language=language;
fstf->type=FST_GRAMMAR;
fstf->pos0=(int)ftell(fstf->f);
fstf->symbols=new_string_hash_ptr(64);
fstf->renumber=NULL;
if (load_elag_fst2_tags(fstf)==-1) {
error("load_fst_file: %s: cannot load symbols\n",fstf->name);
goto error_symbols;
}
fstf->pos=0;
return fstf;
/* If an error occurs */
error_symbols: free_string_hash_ptr(fstf->symbols,(void(*)(void*))free_symbols);
error_f: u_fclose(fstf->f);
error_fstf: free(fstf->name);
free(fstf);
return NULL;
}
/**
* \brief Reads an line of a concord.ind file.
*
* \param[in] line the unichar string containing the line
*
* The line is expected to be in the the format : n.n.n n.n.n t where n are integers and t is string
*
* \return The information read in a locate_pos structure
*/
locate_pos *read_concord_line(const unichar *line) {
locate_pos *l;
l = (locate_pos*) malloc(sizeof(locate_pos) * 1);
if (l == NULL) {
perror("malloc\n");
fprintf(stderr, "Impossible to allocate memory\n");
exit(1);
}
l->label = (unichar*) malloc(sizeof(unichar) * (u_strlen(line) + 1));
if (l->label == NULL) {
perror("malloc\n");
fprintf(stderr, "Impossible to allocate memory\n");
exit(1);
}
// format of a line : n.n.n n.n.n t where n are integers and t is string
const unichar **next;
const unichar *current = line;
next = &line; // make next not NULL
l->token_start_offset = (long)u_parse_int(current, next);
current = (*next)+1;
l->character_start_offset = (long)u_parse_int(current, next);
current = (*next)+1;
l->logical_start_offset = (long)u_parse_int(current, next);
current = (*next)+1;
l->token_end_offset = (long)u_parse_int(current, next);
current = (*next)+1;
l-> character_end_offset = (long)u_parse_int(current, next);
current = (*next)+1;
l-> logical_end_offset = (long)u_parse_int(current, next);
current = (*next)+1;
u_strcpy(l->label,current);
return l;
}
/**
* Loads and returns an automaton from the given .fst2.
* Returns NULL if there is no more automaton to load.
*/
Fst2Automaton* load_automaton(Elag_fst_file_in* fstf) {
if (fstf->pos>=fstf->nb_automata) {
return NULL;
}
Ustring* ustr=new_Ustring();
readline(ustr,fstf->f);
const unichar* p=ustr->str;
if (p[0]!='-') {
fatal_error("load_automaton: %s: bad file format\n",fstf->name);
}
p++;
int i=u_parse_int(p,&p);
if (i!=fstf->pos+1) {
/* We make sure that the automaton number is what it should be */
fatal_error("load_automaton: %s: parsing error with line '%S' ('-%d ...' expected)\n",fstf->name,ustr->str,fstf->pos+1);
}
/* Now p points on the automaton name */
p++;
Fst2Automaton* A=new_Fst2Automaton(p);
while (readline(ustr,fstf->f) && ustr->str[0]!='f') {
/* If there is a state to read */
p=ustr->str;
SingleGraphState state=add_state(A->automaton);
if (*p=='t') {
/* If necessary, we set the state final */
set_final_state(state);
}
/* We puts p on the first digit */
while (*p!='\0' && !u_is_digit(*p)) {
p++;
}
while (*p!='\0') {
/* If there is a transition to read */
int tag_number=u_parse_int(p,&p);
if (fstf->renumber!=NULL) {
tag_number=fstf->renumber[tag_number];
}
while (*p==' ') {
p++;
}
if (!u_is_digit(*p)) {
fatal_error("load_automaton: %s: bad file format (line='%S')\n",fstf->name,ustr->str);
}
int state_number=u_parse_int(p,&p);
symbol_t* tmp=(symbol_t*)fstf->symbols->value[tag_number];
if (tmp!=NULL) {
/* If it is a good symbol (successfully loaded), we add transition(s) */
if (fstf->type!=FST_TEXT) {
add_all_outgoing_transitions(state,tmp,state_number);
} else {
/* In a text automaton, we add one transition per element of
* the symbol list. For instance, if we have:
*
* tmp = "{domestique,.N:fs}" => "{domestique,.N:ms}" => NULL
*
* then we add two transitions. */
add_all_outgoing_transitions(state,tmp,state_number);
}
}
while (*p==' ') {
p++;
}
}
}
if (*ustr->str=='\0') {
fatal_error("load_automaton: unexpected end of file\n");
}
if (A->automaton->number_of_states==0) {
error("load_automaton: automaton with no state\n");
} else {
set_initial_state(A->automaton->states[0]);
}
fstf->pos++;
free_Ustring(ustr);
return A;
}
