prog-intro/java/expression/parser/ExpressionParser.java

package expression.parser;

import expression.*;
import java.util.List;

/**
 * @author Doschennikov Nikita (me@fymio.us)
 */
public class ExpressionParser implements ListParser {

    private String src;
    private int pos;
    private List<String> variables;

    @Override
    public ListExpression parse(String expression, List<String> variables) {
        this.src = expression;
        this.pos = 0;
        this.variables = variables;
        AbstractExpression result = parseMinMax();
        skipWhitespace();
        if (pos < src.length()) {
            throw new IllegalArgumentException(
                "Unexpected character '" +
                    src.charAt(pos) +
                    "' at position " +
                    pos
            );
        }
        return result;
    }

    private AbstractExpression parseMinMax() {
        AbstractExpression left = parseAddSub();
        while (true) {
            skipWhitespace();
            if (tryConsume("min")) {
                left = new Min(left, parseAddSub());
            } else if (tryConsume("max")) {
                left = new Max(left, parseAddSub());
            } else if (tryConsume("set")) {
                left = new SetBit(left, parseAddSub());
            } else if (tryConsume("clear")) {
                left = new Clear(left, parseAddSub());
            } else {
                break;
            }
        }
        return left;
    }

    private AbstractExpression parseAddSub() {
        AbstractExpression left = parseMulDiv();
        while (true) {
            skipWhitespace();
            if (pos < src.length() && src.charAt(pos) == '+') {
                pos++;
                left = new Add(left, parseMulDiv());
            } else if (
                pos < src.length() &&
                src.charAt(pos) == '-' &&
                !isUnaryContext()
            ) {
                pos++;
                left = new Subtract(left, parseMulDiv());
            } else {
                break;
            }
        }
        return left;
    }

    private AbstractExpression parseMulDiv() {
        AbstractExpression left = parseUnary();
        while (true) {
            skipWhitespace();
            if (pos < src.length() && src.charAt(pos) == '*') {
                pos++;
                left = new Multiply(left, parseUnary());
            } else if (pos < src.length() && src.charAt(pos) == '/') {
                pos++;
                left = new Divide(left, parseUnary());
            } else {
                break;
            }
        }
        return left;
    }

    private AbstractExpression parseUnary() {
        skipWhitespace();
        if (pos >= src.length()) {
            throw new IllegalArgumentException(
                "Unexpected end of expression at position " + pos
            );
        }

        if (src.charAt(pos) == '-') {
            pos++;
            if (pos < src.length() && Character.isDigit(src.charAt(pos))) {
                return parseNumber(true);
            }
            return new Negate(parseUnary());
        }

        if (tryConsume("reverse")) {
            requireSeparatorOrParen();
            return new Reverse(parseUnary());
        }
        if (tryConsume("digits")) {
            requireSeparatorOrParen();
            return new Digits(parseUnary());
        }
        if (tryConsume("floor")) {
            requireSeparatorOrParen();
            return new Floor(parseUnary());
        }
        if (tryConsume("ceiling")) {
            requireSeparatorOrParen();
            return new Ceiling(parseUnary());
        }

        return parsePrimary();
    }

    private AbstractExpression parsePrimary() {
        skipWhitespace();
        if (pos >= src.length()) {
            throw new IllegalArgumentException("Unexpected end of expression");
        }
        char c = src.charAt(pos);

        if (c == '(') {
            pos++;
            AbstractExpression inner = parseMinMax();
            skipWhitespace();
            expect();
            return inner;
        }

        if (c == '$') {
            pos++;
            return new Variable(parseIndex());
        }

        if (Character.isDigit(c)) {
            return parseNumber(false);
        }

        if (Character.isLetter(c)) {
            int start = pos;
            while (
                pos < src.length() &&
                (Character.isLetterOrDigit(src.charAt(pos)) ||
                    src.charAt(pos) == '_')
            ) {
                pos++;
            }
            String name = src.substring(start, pos);
            int idx = variables.indexOf(name);
            if (idx >= 0) {
                return new Variable(idx, name);
            }
            throw new IllegalArgumentException(
                "Unknown identifier '" + name + "' at position " + start
            );
        }

        throw new IllegalArgumentException(
            "Unexpected character '" + c + "' at position " + pos
        );
    }

    private void skipWhitespace() {
        while (pos < src.length() && Character.isWhitespace(src.charAt(pos))) {
            pos++;
        }
    }

    private boolean tryConsume(String keyword) {
        skipWhitespace();
        if (!src.startsWith(keyword, pos)) return false;
        int end = pos + keyword.length();
        if (end < src.length()) {
            char next = src.charAt(end);
            if (Character.isLetterOrDigit(next) || next == '_') return false;
        }
        if (variables != null && variables.contains(keyword)) return false;
        pos = end;
        return true;
    }

    private void requireSeparatorOrParen() {
        skipWhitespace();
    }

    private void expect() {
        if (pos >= src.length() || src.charAt(pos) != ')') {
            throw new IllegalArgumentException(
                "Expected '" +
                    ')' +
                    "' at position " +
                    pos +
                    (pos < src.length()
                        ? ", got '" + src.charAt(pos) + "'"
                        : ", got end of input")
            );
        }
        pos++;
    }

    private AbstractExpression parseNumber(boolean negative) {
        int start = pos;
        while (pos < src.length() && Character.isDigit(src.charAt(pos))) {
            pos++;
        }
        if (start == pos) {
            throw new IllegalArgumentException(
                "Expected digit at position " + pos
            );
        }
        String numStr = src.substring(start, pos);
        long val = Long.parseLong(numStr);
        if (negative) val = -val;
        if (val < Integer.MIN_VALUE || val > Integer.MAX_VALUE) {
            throw new IllegalArgumentException("Integer overflow: " + val);
        }
        return new Const((int) val);
    }

    private int parseIndex() {
        int start = pos;
        while (pos < src.length() && Character.isDigit(src.charAt(pos))) {
            pos++;
        }
        if (start == pos) {
            throw new IllegalArgumentException(
                "Expected digit after '$' at position " + pos
            );
        }
        return Integer.parseInt(src.substring(start, pos));
    }

    private boolean isUnaryContext() {
        return false;
    }
}