struct fsm *fsm_read_text_file(char *filename) {
struct fsm_trie_handle *th;
char *text, *textp1, *textp2;
int lastword;
text = file_to_mem(filename);
if (text == NULL) {
return NULL;
}
textp1 = text;
th = fsm_trie_init();
for (lastword = 0 ; lastword == 0 ; textp1 = textp2+1) {
for (textp2 = textp1 ; *textp2 != '\n' && *textp2 != '\0'; textp2++) {
}
if (*textp2 == '\0') {
lastword = 1;
if (textp2 == textp1)
break;
}
*textp2 = '\0';
if (strlen(textp1) > 0)
fsm_trie_add_word(th, textp1);
}
xxfree(text);
return(fsm_trie_done(th));
}
struct fsm *fsm_read_text_file(const char *buffer) {
struct fsm_trie_handle *th;
const char *textp1, *textp2;
int lastword;
textp1 = buffer;
th = fsm_trie_init();
for (lastword = 0 ; lastword == 0 ; textp1 = textp2+1) {
for (textp2 = textp1 ; *textp2 != '\n' && *textp2 != '\0'; textp2++) {
}
if (*textp2 == '\0') {
lastword = 1;
if (textp2 == textp1)
break;
}
if (textp2-textp1 > 0) {
// we need a non-const copy to add the '\0'
int len = textp2-textp1;
char *aux = malloc((len+1)*sizeof(char));
memcpy(aux,textp1,len);
aux[len]='\0';
fsm_trie_add_word(th, aux);
free(aux);
}
}
return(fsm_trie_done(th));
}
struct fsm *fsm_read_spaced_text_file(char *filename) {
struct fsm_trie_handle *th;
char *text, *textorig, *insym, *outsym, *t1, *t2, *l1, *l2;
text = textorig = file_to_mem(filename);
if (text == NULL)
return NULL;
th = fsm_trie_init();
for (;;) {
for ( ; *text != '\0' && *text == '\n'; text++) { }
t1 = spacedtext_get_next_line(&text);
if (t1 == NULL)
break;
if (strlen(t1) == 0)
continue;
t2 = spacedtext_get_next_line(&text);
if (t2 == NULL || strlen(t2) == 0) {
for (l1 = t1; (insym = spacedtext_get_next_token(&l1)) != NULL; ) {
if (strcmp(insym, "0") == 0)
fsm_trie_symbol(th, "@_EPSILON_SYMBOL_@", "@_EPSILON_SYMBOL_@");
else if (strcmp(insym, "%0") == 0)
fsm_trie_symbol(th, "0", "0");
else
fsm_trie_symbol(th, insym, insym);
}
fsm_trie_end_word(th);
} else {
for (l1 = t1, l2 = t2; ; ) {
insym = spacedtext_get_next_token(&l1);
outsym = spacedtext_get_next_token(&l2);
if (insym == NULL && outsym == NULL)
break;
if (insym == NULL || strcmp(insym, "0") == 0)
insym = "@_EPSILON_SYMBOL_@";
if (strcmp(insym, "%0") == 0)
insym = "0";
if (outsym == NULL || strcmp(outsym, "0") == 0)
outsym = "@_EPSILON_SYMBOL_@";
if (strcmp(outsym, "%0") == 0)
outsym = "0";
fsm_trie_symbol(th, insym, outsym);
}
fsm_trie_end_word(th);
}
}
xxfree(textorig);
return(fsm_trie_done(th));
}
