Raku Books / Using Raku / Regexes and Grammars / Using grammars

86. Basic calculator

Create a program that calculates mathematical operations with two operands, for example: 4 + 5.3 or 7.8 / 3.

In this task, we will only limit the solution for the simplest case with only one operation so that there are no issues with the precedence order or parentheses.

Let’s first make a solution with regexes and then transform it so it uses Raku’s grammars.

my $expression = "14 * 16.4";
$expression ~~ /
    (<[-\d\.]>+) \s*
    ('+' | '-' | '*' | '/') \s*
    (<[-\d\.]>+)
/;
given $1 {
    when '+' {say $0 + $2}
    when '-' {say $0 - $2}
    when '*' {say $0 * $2}
    when '/' {say $0 / $2};
}

↻ Runs in Rakudo; the in-browser engine (Raku++) doesn’t support this program yet — see the expected output above.

The regular expression in this example has three capturing parentheses. The first and the third of them match numbers with an optional minus sign and a floating-point dot. We do not check the number format too strictly to keep the program simple, however.

The construction <[...]> is a character class in Raku regexes. It lists accepted characters. In our case, those are the digits represented by \d, dots \., and minus -. Optional spaces around the operator are matched by \s*.

The operator character is an alternative: '+' | '-' | '*' | '/'.

After the successful match, the parts of $expression appear in the variables $0, $1, and $2. These are the shortcuts for the elements of the match objects: $/[0], $/[1], and $/[2]. Each element is also an object of the Match type.

The given statement chooses one of the when branches depending on the value in $1. Here, an implicit type conversion happens. First, the match object from $1 is coerced to a string before comparing it with one of the operator characters. Second, the $0 and $1 objects are converted to numbers, which are used in actual calculations. Run the program a few times with different expressions and see how it works.

Now, let us solve the problem differently, using grammars. Here is the grammar:

grammar calc {
    rule TOP {
        <value> <operator> <value> {
            given $<operator> {
                when '+' {say $<value>[0] + $<value>[1]}
                when '-' {say $<value>[0] - $<value>[1]}
                when '*' {say $<value>[0] * $<value>[1]}
                when '/' {say $<value>[0] / $<value>[1]}
            }
        }
    }
    token value {
        '-'? \d+ [ '.' \d+ ]?
    }
    token operator {
        '+' | '-' | '*' | '/'
    }
}

And, here is how you use it to parse an expression:

my $expression = "14 * 16.2";
calc.parse($expression); # Prints 226.8

The definition of the grammar contains the TOP rule and two tokens: value and operator. When the parse method is called, the grammar starts parsing the text, using the TOP rule, which says that the text should be a value, followed by the operator symbol, followed by another value.

This time, the regex for the values is a bit smarter to make sure the minus sign can appear only before digits, and the decimal point should be seen only once. Still, there is some room to improve the regex.

When using the grammar rules (such as TOP), you don’t need to explicitly care about the optional spaces before parts of the expression. It means that both '14 * 16.2' and '14*16.2' strings are accepted.

The TOP rule has three parts: <value> <operator> <value>.

After they all match, the action code block is triggered. It contains a given- when block, similar to the one used earlier in the solution with regexes.

Now, the string values of the expression parts are taken from the match object, which is treated as a hash. For example, the operator character is located in the $<operator> object, which is a shortcut for $/<operator>. However, there are two <value>s in the TOP rule. In this case, the match object contains a list of other objects, which you can access via indices:

say $<value>[0] + $<value>[1]

Again, before using the values, the match objects are converted by Raku to numbers.

Course navigation

← 85. Balanced parentheses   |   Date and Time →