compiler.cpp

SSD6 练习4的原版正确答案

// work well for user type-in)
//
bool Compiler::compile_file(FILE *file, bool interactive)
{
	interactive_mode = interactive;
	inf = file; // so that get_token() will read from file
	if (!semantics.reset()) return false;
    error_flag = false;
    token = tokens;
    unget_count = 0;
    token->typ = TOKEN_OTHER;
    istack_top = 0;
    indentation[0] = 0;
    dedent_count = 0;
	newline_flag = false;
	line_number = 1;
    while (!machine->exit_flag && parse_decl()) ;
	return !error_flag;
}


bool Compiler::compile(char *filename)
{
    FILE *f = fopen(filename, "r");
	bool result;
    if (f) {
		result = compile_file(f, false);
		fclose(f);
	} else {
        char msg[128];
        strcpy(msg, "Compiler: can't open ");
        strcat(msg, filename);
        report_error(msg);
		result = false;
    }
	return result;
}


bool Compiler::parse_decl()
{
    if (get_token()) {
        return parse_decl_2();
    }
    return false;
}


bool Compiler::parse_decl_2()
{
    if (token->typ == TOKEN_ID) {
        if (streql(token->str, "def")) {
            return parse_method();
        } else if (streql(token->str, "var")) {
            return parse_var();
        } else if (streql(token->str, "class")) {
            return parse_class();
        } else {
            // assume we have an immediate statement
            // create a method for it
            semantics.method_start();
            semantics.method_name_is("<immediate command>");
            semantics.begin_formals();
            semantics.end_of_formals();
            if (parse_stmt_2() && semantics.method_end()) {
                // execute the method now
                Symbol_ptr s_imcd = Symbol::create("<immediate command>", machine);
                machine->global_function(s_imcd);
                if (machine->error_flag) return false;
                machine->call();
                return !machine->error_flag;
            }
            return false; // allow immediate stmt in place of decl
        }
    } else if (token->typ == TOKEN_NEWLINE) {
        return true;
    }
    return false;
}


bool Compiler::parse_id(char *msg)
{
    if (get_token() && token->typ == TOKEN_ID) return true;
    expected_id(msg);
    return false;
}


bool Compiler::parse_method()
{
    if (get_token()) {
        if (token->typ == TOKEN_ID &&
            semantics.method_start() &&
            semantics.method_name_is(token->str)) {
            if (parse_formals() &&
				parse_colon() &&
                parse_stmt_list() && 
                semantics.method_end()) {
                return true;
            }
        }
    }
    return false;
}

bool Compiler::parse_formals()
{
    semantics.begin_formals();
    if (!get_token() || !streql(token->str, "(")) {
        expected_open_paren();
        return false;
    }
    if (!parse_token()) return false;
    bool more = true;
    bool first = true; // allow ')' only on first time through loop
    while (more) {
        if (token->typ == TOKEN_ID) {
            bool ok = true;
            if ((streql(token->str, "optional") && 
                 (ok = semantics.set_formal_type(FORMAL_OPTIONAL))) ||
                (streql(token->str, "keyword") &&
                 (ok = semantics.set_formal_type(FORMAL_KEYWORD))) ||
                (streql(token->str, "rest") &&
                 (ok = semantics.set_formal_type(FORMAL_REST))) ||
                (streql(token->str, "dictionary") &&
                 (ok = semantics.set_formal_type(FORMAL_DICTIONARY))) ||
                (streql(token->str, "required") &&
                 (ok = semantics.set_formal_type(FORMAL_REQUIRED)))) {
                if (!parse_id("formal parameter")) return false;
            }
            // if a set_formal_type failed, return now:
            if (!ok) return false;
            if (!semantics.formal_is(token->str)) return false;
            // look for default value specification:
            if (get_token()) {
                if (streql(token->str, "=")) {
                    if (!parse_default_value()) return false;
                } else unget_token();
            } else unget_token();
            // look for comma
            if (get_token()) {
                more = streql(token->str, ",");
                if (more) {
                    if (!parse_token()) return false;
                }
            } else { // no token!
                expected_close_paren();
                return false;
            }
        } else if (first && streql(token->str, ")")) {
            more = false;
        } else break;
        first = false;
    }
    if (streql(token->str, ")")) {
        return semantics.end_of_formals();
    } else {
        expected_close_paren_or_formal();
        return false();
    }
}


int Compiler::hex_to_int(char h)
{
	if (isdigit(h)) return h - '0';
	else if (isxdigit(h)) return tolower(h) - 'a' + 10;
	expected_hex_constant();
	return 0;
}


int Compiler::oct_to_int(char o)
{
	if ((o >= '0') && (o <= '7')) return o - '0';
	expected_oct_constant();
	return 0;
}


int Compiler::dec_to_int(char d)
{
	if (isdigit(d)) return d - '0';
	expected_number_constant();
	return 0;
}


int64 Compiler::string_to_int(char *s)
{
	int64 result = 0;
	int sign = 1;
	if (!s || *s == 0) return result;
	if (*s == '-') {
		if (*s++ == 0) return result;
		sign = -1;
	} 
	if (s[0] == '0' && (s[1] == 'x' || s[1] == 'X')) { // hexadecimal
		for (int i = 2; s[i] && !error_flag; i++) {
			result = (result << 4) + hex_to_int(s[i]);
		}
		result *= sign;
	} else if (s[0] == '0') { // octal
		for (int i = 1; s[i] && !error_flag; i++) {
			result = (result << 3) + oct_to_int(s[i]);
		}
		result *= sign;
	} else {
		// since negative numbers can be larger (in magnitude) than positive
		// we'll compute decimal numbers in the negative form and then apply
		// the sign.
		for (int i = 0; s[i] && !error_flag; i++) {
			result = (result * 10) - dec_to_int(s[i]);
		}
		result *= -sign;
	}
	return result;
}


bool Compiler::parse_default_value()
{
    if (!parse_token()) return false;
    // assume value is in one token
    if (token->typ == TOKEN_LONG) {
		int64 val = string_to_int(token->str);
		if (error_flag) return false;
        return semantics.default_value(
            FLong::create(val, machine));
	} else if (token->typ == TOKEN_DOUBLE) {
		// make sure there is a + or - after e:
		char *pos = strpbrk(token->str, "eE");
		if (pos && pos[1] != '+' && pos[1] != '-') {
			// open up some space:
			memmove(pos + 2, pos + 1, strlen(pos + 1));
			pos[1] = '+';
		}
        return semantics.default_value(
            FDouble::create(atof(token->str), machine));
	} else if (token->typ == TOKEN_STRING) {
        return semantics.default_value(
            FString::create(token->str + 1, machine));
    } else if (token->typ == TOKEN_SYMBOL) {
        return semantics.default_value(
            Symbol::create(token->str + 1, machine));
	} else return false;
}


bool Compiler::parse_stmt()
{
    if (get_token()) {
        return parse_stmt_2();
    }
    expected_statement();
    return false;
}


// parse_stmt_2 -- parse stmt, assumes first token already gotten
//
bool Compiler::parse_stmt_2()
{
    if (streql(token->str, "class") || streql(token->str, "var") ||
        streql(token->str, "def")) {
        return parse_decl_2();
    } else if (streql(token->str, "if")) {
        return parse_if();
    } else if (streql(token->str, "while")) {
        return parse_while();
    } else if (streql(token->str, "for")) {
        return parse_for();
    } else if (streql(token->str, "return")) {
        return parse_return();
	} else if (streql(token->str, "print")) {
		return parse_print();
    } else if (streql(token->str, "load")) {
        return parse_load();
    } else if (token->typ == TOKEN_ID) {
        char id[STRING_MAX];
        strcpy(id, token->str);
        if (!parse_id_expr(id)) return false;
        if (get_token()) {
            if (streql(token->str, "=")) {
                return parse_assign();
            } else if (token->typ == TOKEN_NEWLINE) {
                semantics.end_id_expr();
                return true;
            }
        }
        expected_newline();
    } else {
        expected_statement();
    }
    return false;
}


bool Compiler::parse_if()
{
    // at this point, IF has already been parsed
    if (parse_expr()) {
        semantics.if_true();
        if (parse_colon() && parse_stmt_list()) {
			long elif_count = 0;
			while (get_token() && streql(token->str, "elif")) {
				semantics.if_else();
				if (parse_expr()) {
					semantics.if_true();
					if (parse_colon() && parse_stmt_list()) {
						elif_count++;
						trace("elif_count %d\n", elif_count);
					} else return false;
				} else return false;
			}
            if (streql(token->str, "else")) {
                semantics.if_else();
                if (parse_colon() && parse_stmt_list()) {
					// if_end called for each elif and for else
					while (elif_count >= 0) {
						semantics.if_end();
						elif_count--;
					}
					return true;
				} else return false;
            } else {
                unget_token();
				while (elif_count >= 0) {
					semantics.if_end();
					elif_count--;
				}
                return true;
            }
        }
    }
    return false;
}


bool Compiler::parse_while()
{
    return semantics.while_begin() && 
        parse_expr() && parse_colon() &&
        semantics.while_true() && 
        parse_stmt_list() && 
        semantics.while_end();
}


bool Compiler::parse_for()
{
	if (!parse_token()) return false;
	if (token->typ != TOKEN_ID) {
		expected_id(token->str);
		return false;
	}
	if (!semantics.for_var(token->str)) return false;
	if (!parse_token()) return false;
	else if (streql(token->str, "=")) {
		if (!parse_expr() || !semantics.for_init())
			return false;
		if (!get_token() || !streql(token->str, "to")) {
			expected_to();
			return false;
		}
		if (!(parse_expr() && semantics.for_to()))
			return false;
		if (!parse_token()) return false;
		if (streql(token->str, "by")) {
			if (!parse_expr() || !semantics.for_by(true))
				return false;
		} else {
			unget_token();
			if (!semantics.for_by(false)) return false;
		}
		if (!parse_colon() || !parse_stmt_list() || 
			!semantics.for_end_range())
			return false;
	} else if (streql(token->str, "in")) {
		if (!parse_expr() || !semantics.for_array() ||
			!parse_colon() || !parse_stmt_list() ||
			!semantics.for_end_elements())
			return false;
	} else {
		expected_equal_or_in();
		return false;
	}
	return true;
}


bool Compiler::parse_indent()
{
	if (get_token() && token->typ == TOKEN_INDENT) {
		return true;
	}
	expected_indent();
	return false;
}


bool Compiler::parse_newline()
{
    if (get_token() && token->typ == TOKEN_NEWLINE) {
        return true;
    }
    expected_newline();
    return false;
}


bool Compiler::parse_return()
{
    if (get_token() && token->typ == TOKEN_NEWLINE) {
		return semantics.return_null();
	}
    unget_token();
	return parse_expr() && parse_newline() && semantics.return_end();
}


bool Compiler::parse_load()
{
	return parse_expr() && parse_newline() && semantics.load_end();
}


bool Compiler::parse_print()
{
    bool print_endl = true;
    while (true) {
        // look for an expression
        if (!parse_token()) return false;
        if (token->typ == TOKEN_NEWLINE) {
            if (print_endl) return semantics.print_end();
            else return true;
        } else if (unget_token() && parse_expr()) {
            semantics.print_expr();
        } else return false; // no expression found
        print_endl = false;
        // look for a separator
        if (!parse_token()) return false;
        if (token->typ == TOKEN_NEWLINE) {
            return semantics.print_end();
        } else if (streql(token->str, ",")) {
            semantics.print_comma();
        } else if (streql(token->str, ";")) {
            // nothing
        } else expected_comma();
    }
}

bool Compiler::parse_stmt_list()
{
	if (!parse_token()) return false;
	if (token->typ != TOKEN_NEWLINE) {
		return parse_stmt_2(); // just one statement after a colon
	} 
	if (!parse_indent()) return false;
	while (true) {
		if (get_token()) {
			if (token->typ == TOKEN_DEDENT) {
				return true;
			} else if (parse_stmt_2()) {
				// continue parsing stmts
			} else break;
		} else {
			expected_statement();
			break;
		}
	}
	return false;
}


bool Compiler::parse_assign()
{
    // expr and "=" have already been parsed
    return semantics.rval2lval() && 
           parse_expr() && 
           parse_newline() && 
           semantics.assign_end();
}


/*
bool Compiler::parse_assign_id(char *id)
{
	return parse_expr() && parse_newline() && semantics.assign_id(id);
}
*/


bool Compiler::parse_expr()
{
    if (parse_conj()) {
        return parse_disjunctions();
    }
    return false;
}


bool Compiler::parse_conj()
{
    if (parse_relation()) {
        return parse_conjunctions();
    }
    return false;
}


bool Compiler::parse_relation()
{
    if (parse_sum()) {
        if (parse_relop()) {