5
\$\begingroup\$

I'm writing a basic LL(1) parser in java, but my parser class is quickly getting out of hand and becoming huge. PMD even calls it a "God class" and says that it has "Too many methods." Is there a way that I can refactor this class?

public class Parser {
    private static final APValueNum NEGATIVE_ONE = new APValueNum(
            new BigDecimal("-1"));
    LinkedList<Token> tokens = new LinkedList<>();
    Token lookahead;

    public Parser(final List<Token> tokens) {
        this.tokens.addAll(tokens);
        if (tokens.isEmpty()) {
            throw new ParserException("Cannot parse an empty file!");
        }
    }

    private void nextToken() {
        try {
            tokens.pop();
        } catch (final NoSuchElementException e) {
            throw new ParserException("Ran out of characters!", e);
        }
        // at the end of input we return an epsilon token
        if (tokens.isEmpty()) {
            lookahead = new Token(TokenType.EOF, "<EOF>");
        } else {
            lookahead = tokens.getFirst();
        }
    }

    public List<ExpressionNode> parse(final Context context) {

        final List<ExpressionNode> expressions = new ArrayList<>();

        lookahead = tokens.getFirst();

        while (lookahead.getType() != TokenType.EOF) {
            expressions.add(statement(context));
        }
        return expressions;
    }

    private ExpressionNode statement(final Context context) {
        final ExpressionNode.VariableNode expr = identifier();
        nextToken();
        if (lookahead.getType() == TokenType.EQUAL) {
            final ExpressionNode assignment = assignment(context, expr);
            return assignment;
        } else if (lookahead.getType() == TokenType.IDENTIFIER) {
            // we have a parameter
            nextToken();
            if (lookahead.getType() == TokenType.IDENTIFIER
                    || lookahead.getType() == TokenType.EQUAL) {
                final List<VariableNode> variables = new ArrayList<>();
                while (lookahead.getType() == TokenType.IDENTIFIER) {
                    variables.add(identifier());
                    nextToken();
                }
                if (lookahead.getType() == TokenType.EQUAL) {
                    nextToken();
                    final ExpressionNode expression = expression(context);
                    final Function function = new Function(expr.getName(),
                            variables, expression);
                    context.putFunction(expr.getName(), function);
                    nextToken();
                    return ExpressionNode.VOID;
                } else {
                    throw new ParserException(
                            "Expected parameter list or equal (function def)");
                }
            } else {
                throw new ParserException("Expected EQUAL");
            }
        } else if (lookahead.getType() == TokenType.OPEN_PARENS) {
            final ExpressionNode node = functionParameters(context, expr);
            nextToken();
            return node;
        } else {
            throw new ParserException(
                    "Non function call, assignment, or function def statement");
        }

    }

    private ExpressionNode assignment(final Context context,
            final ExpressionNode.VariableNode expr) {
        nextToken();
        final ExpressionNode assigned = expression(context);
        final ExpressionNode assignment = new ExpressionNode.AssignmentNode(
                expr, assigned);
        if (lookahead.getType() != TokenType.SEMI) {
            throw new ParserException("Expected semicolon, got "
                    + lookahead.getText());
        }
        nextToken();
        return assignment;
    }

    private ExpressionNode expression(final Context context) {
        if (lookahead.getType() == TokenType.IF) {
            return ifExpr(context);
        }

        final ExpressionNode expr = signedTerm(context);
        return lowOp(expr, context);
    }

    private ExpressionNode ifExpr(final Context context) {
        assert lookahead.getType() == TokenType.IF;
        nextToken();
        final ExpressionNode ifExpr = expression(context);
        final ExpressionNode thenExpr = expression(context);
        if (lookahead.getType() != TokenType.ELSE) {
            throw new ParserException("Needs else after if, was " + lookahead);
        }
        nextToken();
        final ExpressionNode elseExpr = expression(context);
        return new ExpressionNode.IfNode(ifExpr, thenExpr, elseExpr);
    }

    /*
     * Lower level precedence operations: +, -, ||
     */
    private ExpressionNode lowOp(final ExpressionNode expr,
            final Context context) {
        if (lookahead.getType() == TokenType.PLUSMINUS) {
            // sum_op -> PLUSMINUS term sum_op
            ExpressionNode sum;
            final boolean positive = lookahead.getText().equals("+");
            nextToken();
            final ExpressionNode t = signedTerm(context);
            if (positive) {
                sum = new ExpressionNode.AdditionNode(expr, t);
            } else {
                sum = new ExpressionNode.SubtractionNode(expr, t);
            }

            return lowOp(sum, context);
        } else if (lookahead.getType() == TokenType.OR) {
            nextToken();
            return lowOp(new ExpressionNode.OrNode(expr, term(context)),
                    context);
        } else if (lookahead.getType() == TokenType.LESS_THAN) {
            nextToken();
            return lowOp(new ExpressionNode.LessThanNode(expr, term(context)),
                    context);
        } else if (lookahead.getType() == TokenType.GREATER_THAN) {
            nextToken();
            return lowOp(
                    new ExpressionNode.GreaterThanNode(expr, term(context)),
                    context);
        } else if (lookahead.getType() == TokenType.LESS_THAN_EQUAL) {
            nextToken();
            return lowOp(new ExpressionNode.LessThanEqualNode(expr,
                    term(context)), context);
        } else if (lookahead.getType() == TokenType.GREATER_THAN_EQUAL) {
            nextToken();
            return lowOp(new ExpressionNode.GreaterThanEqualNode(expr,
                    term(context)), context);
        } else {
            // sum_op -> EPSILON
            return expr;
        }
    }

    private ExpressionNode term(final Context context) {
        // term -> factor term_op
        return highOp(factor(context), context);
    }

    /*
     * High precedence operations: *, /, &&
     */
    private ExpressionNode highOp(final ExpressionNode expr,
            final Context context) {
        if (lookahead.getType() == TokenType.MULDIV) {
            // term_op -> MULTDIV factor term_op
            ExpressionNode prod;

            final boolean positive = lookahead.getText().equals("*");
            nextToken();
            final ExpressionNode f = signedFactor(context);

            if (positive) {
                prod = new ExpressionNode.MultiplicationNode(expr, f);
            } else {
                prod = new ExpressionNode.DivisionNode(expr, f);
            }

            return highOp(prod, context);
        } else if (lookahead.getType() == TokenType.AND) {
            nextToken();
            return highOp(new ExpressionNode.AndNode(expr, value(context)),
                    context);
        } else {
            // term_op -> EPSILON
            return expr;
        }

    }

    private ExpressionNode signedFactor(final Context context) {
        if (lookahead.getType() == TokenType.PLUSMINUS) {
            final boolean positive = lookahead.getText().equals("+");
            nextToken();
            final ExpressionNode t = factor(context);
            if (positive) {
                return t;
            } else {
                return new ExpressionNode.ConstantNode(
                        ((APValueNum) t.getValue(context)).callMethod(
                                Methods.MULTIPLY, NEGATIVE_ONE));
            }
        } else {
            // signed_factor -> factor
            return factor(context);
        }
    }

    private ExpressionNode factor(final Context context) {
        // factor -> argument factor_op
        return factorOp(argument(context), context);
    }

    private ExpressionNode factorOp(final ExpressionNode expression,
            final Context context) {
        if (lookahead.getType() == TokenType.RAISED) {
            // factor_op -> RAISED expression
            nextToken();
            final ExpressionNode exponent = signedFactor(context);
            return new ExpressionNode.ExponentiationNode(expression, exponent);

        } else {
            // factor_op -> EPSILON
            return expression;
        }
    }

    private ExpressionNode argument(final Context context) {
        if (lookahead.getType() == TokenType.OPEN_PARENS) {
            // argument -> OPEN_BRACKET sum CLOSE_BRACKET
            nextToken();
            final ExpressionNode node = expression(context);

            if (lookahead.getType() != TokenType.CLOSE_PARENS) {
                throw new ParserException("Closing brackets expected and "
                        + lookahead.getText() + " found instead");
            }

            nextToken();
            return node;
        } else {
            // argument -> value
            return value(context);
        }
    }

    private ExpressionNode signedTerm(final Context context) {
        if (lookahead.getType() == TokenType.PLUSMINUS) {
            // signed_term -> PLUSMINUS term
            final boolean positive = lookahead.getText().equals("+");
            nextToken();
            final ExpressionNode t = term(context);
            if (positive) {
                return t;
            } else {
                return new ExpressionNode.ConstantNode(t.getValue(context)
                        .callMethod(Methods.MULTIPLY, NEGATIVE_ONE));
            }
        } else {
            // signed_term -> term
            return term(context);
        }
    }

    private ExpressionNode value(final Context context) {
        if (lookahead.getType() == TokenType.NUMBER) {
            final ExpressionNode.ConstantNode expr = new ExpressionNode.ConstantNode(
                    new APValueNum(new BigDecimal(lookahead.getText())));
            nextToken();
            return expr;
        } else if (lookahead.getType() == TokenType.BOOLEAN) {
            final ExpressionNode.ConstantNode expr = new ExpressionNode.ConstantNode(
                    new APValueBool(Boolean.parseBoolean(lookahead.getText())));
            nextToken();
            return expr;
        }

        else if (lookahead.getType() == TokenType.IDENTIFIER) {

            final ExpressionNode.VariableNode expr = identifier();
            nextToken();
            if (lookahead.getType() == TokenType.OPEN_PARENS) {
                return functionParameters(context, expr);
            }
            return expr;
        } else {
            throw new ParserException("Unexpected token " + lookahead
                    + " found");
        }
    }

    /**
     * @param context
     * @param expr
     * @return
     */
    private ExpressionNode functionParameters(final Context context,
            final ExpressionNode.VariableNode expr) {
        final List<ExpressionNode> parameters = new ArrayList<>();
        nextToken();
        if (lookahead.getType() == TokenType.CLOSE_PARENS) {
            // No params
            final ExpressionNode.FunctionCallNode node = new ExpressionNode.FunctionCallNode(
                    expr, parameters);
            nextToken();
            return node;
        }

        parameters.add(expression(context));
        if (lookahead.getType() == TokenType.CLOSE_PARENS) {
            // One parameter
            final ExpressionNode.FunctionCallNode node = new ExpressionNode.FunctionCallNode(
                    expr, parameters);
            nextToken();
            return node;
        }

        while (lookahead.getType() != TokenType.CLOSE_PARENS) {
            if (lookahead.getType() != TokenType.COMMA) {
                throw new ParserException("Expected comma, got " + lookahead);
            }
            nextToken();
            parameters.add(expression(context));
        }
        nextToken();

        return new ExpressionNode.FunctionCallNode(expr, parameters);
    }

    private ExpressionNode.VariableNode identifier() {
        return new ExpressionNode.VariableNode(lookahead.getText());
    }
}
\$\endgroup\$
4
  • \$\begingroup\$ That seems fine to me. I wouldn't refactor it just because some robot with no taste says so. \$\endgroup\$ Oct 30, 2014 at 23:40
  • \$\begingroup\$ Yes, but if this file grows even bigger, should I just keep it the same way? I'm not sure, but I feel like there should be some point when it should change. \$\endgroup\$
    – Kyranstar
    Oct 30, 2014 at 23:42
  • \$\begingroup\$ Have you considered switching to a parser generator? ANTLR is a personal favorite of mine. \$\endgroup\$ Oct 31, 2014 at 10:14
  • \$\begingroup\$ I've used it before, but I'm just doing this for learning purposes. But ANTLR is pretty great. \$\endgroup\$
    – Kyranstar
    Oct 31, 2014 at 13:51

1 Answer 1

4
\$\begingroup\$

When I look at this code, I find an immediate criticism that may help. You name many of your functions as nouns, e.g.

private ExpressionNode statement(final Context context) {
private ExpressionNode assignment(final Context context,
        final ExpressionNode.VariableNode expr) {
private ExpressionNode expression(final Context context) {
private ExpressionNode ifExpr(final Context context) {

Are you sure that these should be functions? If you are naming a bunch of nouns, maybe these should be classes, each of which could have its own parse or getNext method. This is even true of

private void nextToken() {

But the solution is probably different. Perhaps simply rename that to getNextLookahead(). Alternately, you could make a new TokenStream class with a getNext function, or a higher level ExpressionNode class like Package, Program, or File with a parse method.

Of course, if you follow this advice, you will end up with a much larger number of small classes that basically exist to wrap the parse method in each. You will probably have to pass your Parser to each the way that you already pass a Context.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.