Support for Unary operators

[ezylang]

Support for Unary operators

[uklimaschewski]

This will require some tricky changes in the parser locgic. I will leave this issue open, maybe I will find some time, or there is a clever contributor who wants to earn eternal fame and honour ... ;-)

[yoavst]

I've wrote a parser that support unary operator in Kotlin.
Here is my project - https://gitlab.com/yoav-sternberg/algebraEval (it was done for learning how to parse expressions, but yes it is based on EvalEx.), if someone want to port the unary part, you are welcome.

I've find that is not so hard to support 1 char unary operator for start.

    private var previousToken: Token? = null
    public var index: Int = 0
    override fun next(): Token {
        if (index >= input.length()) throw IllegalStateException()
        val token = StringBuilder()
        var char = input[index]
        while (char.isWhitespace() && index < input.length()) {
            char = input[++index]
        }
        if (char.isDigit()) {
            while ((Character.isDigit(char) || char == ExpressionParser.DecimalSeparator) && (index < input.length())) {
                token.append(input.charAt(index++))
                char = if (index == input.length()) ' ' else input[index]
            }
            previousToken = Token(TokenType.Number, token.toString())
            return previousToken!!
        } else if (char in unaryOperators.keySet() && !peek().isWhitespace()
                && (previousToken == null || "(" == previousToken!!.text || "," == previousToken!!.text
                || previousToken!!.text in operators.keySet() || previousToken!!.text in unaryOperators.keySet())) {
            token.append(char)
            index++
            previousToken = Token(TokenType.UnaryOperator, token.toString())
            return previousToken!!
        } else if (char.isLetter() || char == '_') {
            while ((char.isLetter() || char.isDigit() || char == '_') && index < input.length()) {
                token.append(input[index++])
                char = if (index == input.length()) ' ' else input[index]
            }
            previousToken = Token(TokenType.Name, token.toString())
            return previousToken!!
        } else if (char == '(') {
            index++
            previousToken = Token(TokenType.LeftParen, "(")
            return previousToken!!
        } else if (char == ')') {
            index++
            previousToken = Token(TokenType.RightParen, ")")
            return previousToken!!
        } else if (char == ',') {
            index++
            previousToken = Token(TokenType.Comma, ",")
            return previousToken!!
        } else {
            while (!char.isLetter() && !char.isDigit() && char != '_' && !Character.isWhitespace(char)
                    && char != '(' && char != ')' && char != ',' && index < input.length()) {
                token.append(input[index])
                index++
                char = if (index == input.length()) ' ' else input[index]
                if (char in unaryOperators) {
                    break
                }
            }
            if (token.toString() !in operators.keySet()) {
                throw IllegalArgumentException("Unknown operator '" + token + "' at indexition " + (index - token.length() + 1))
            }
            previousToken = Token(TokenType.Operator, token.toString())
            return previousToken!!
        }

[uklimaschewski]

Thank you very much for this code snippet!

[micabytes]

'not' or '!' as a unary operator would be nice, as that is pretty common in many expressions.

[Sub6Resources]

I would love to see this added to the project. It appears it hasn't been worked on recently.