Re_node parse_re(char **s, short end)
{
Stack stk = NULL, temp;
Tok_node next_token;
Re_node re = NULL;
if (s == NULL || *s == NULL) return NULL;
while (TRUE) {
next_token = get_token(s);
if (next_token == NULL) return NULL;
switch (tok_type(next_token)) {
case RPAREN:
retract_token(s);
case EOS:
if (end == tok_type(next_token)) return Top(cat2(&stk));
else return NULL;
case LPAREN:
re = parse_re(s, RPAREN);
if (Push(&stk, re) == NULL) return NULL;
if (tok_type(get_token(s)) != RPAREN || re == NULL) return NULL;
if (Size(stk) > 2) {
temp = stk->next;
stk->next = cat2(&temp); /* condense CAT nodes */
if (stk->next == NULL) return NULL;
else stk->size = stk->next->size + 1;
}
break;
case OPSTAR:
if (wrap(&stk, OPSTAR) == NULL) return NULL;
break;
case OPOPT:
if (wrap(&stk, OPOPT) == NULL) return NULL;
break;
case OPALT:
if (cat2(&stk) == NULL) return NULL;
re = parse_re(s, end);
if (re == NULL) return NULL;
if (mk_alt(&stk, re) == NULL) return NULL;
break;
case LITERAL:
if (Push(&stk, tok_val(next_token)) == NULL) return NULL;
if (Size(stk) > 2) {
temp = stk->next;
stk->next = cat2(&temp); /* condense CAT nodes */
if (stk->next == NULL) return NULL;
else stk->size = stk->next->size + 1;
}
break;
default:
printf("parse_re: unknown token type %d\n", tok_type(next_token));
break;
}
}
}
Tok_node get_token(char **s)
{
Tok_node rn = NULL;
if (s == NULL || *s == NULL) return NULL; /* error */
rn = (Tok_node) new_node(rn);
if (**s == NUL) tok_type(rn) = EOS; /* end of string */
else {
switch (**s) {
case '.': /* wildcard */
tok_type(rn) = LITERAL;
tok_val(rn) = parse_wildcard();
if (tok_val(rn) == NULL) return NULL;
break;
case '[': /* character set literal */
(*s)++;
tok_type(rn) = LITERAL;
tok_val(rn) = parse_cset(s);
if (tok_val(rn) == NULL) return NULL;
break;
case '(':
tok_type(rn) = LPAREN;
break;
case ')' :
tok_type(rn) = RPAREN;
break;
case '*' :
tok_type(rn) = OPSTAR;
break;
case '|' :
tok_type(rn) = OPALT;
break;
case '?' :
tok_type(rn) = OPOPT;
break;
case '\\': /* escaped character */
(*s)++;
default : /* must be ordinary character */
tok_type(rn) = LITERAL;
tok_val(rn) = parse_chlit(**s);
if (tok_val(rn) == NULL) return NULL;
break;
} /* switch (**s) */
(*s)++;
} /* else */
return rn;
}
/* like tok_pop() but ignore T_SPACE tokens; if sep, read until chopped */
char *tok_poptext(int sep)
{
while (tok_type() == T_SPACE)
tok_read();
tok_prev[0] = '\0';
do {
strcat(tok_prev, tok);
tok_read();
} while (tok[0] && !tok_chops(!sep));
return tok_prev[0] ? tok_prev : NULL;
}
/* if the next token is s, return zero and skip it */
int tok_jmp(char *s)
{
tok_blanks();
if (tok_get() && !s[1] && strchr("{}~^\t", s[0]) && !strcmp(s, tok_get())) {
tok_pop();
return 0;
}
if (tok_type() != T_KEYWORD || !tok_get() || strcmp(s, tok_get()))
return 1;
tok_pop();
return 0;
}
/* skip spaces */
static void tok_blanks(void)
{
while (tok_type() == T_SPACE)
tok_pop();
}