evaler

Package evaler implements a simple floating point arithmetic expression evaluator.

Evaler uses Dijkstra's Shunting Yard algorithm to convert an infix expression to postfix/RPN format, then evaluates the RPN expression. The implementation is adapted from a Java implementation. The results are returned as a *big.Rat.

Usage

result, err := evaler.Eval("1+2")

Operators

The operators supported are:

+ - * / ^ ** () < > <= >= == !=

(^ and ** are both exponent operators)

Logical operators like < (less than) or > (greater than) get lowest precedence, all other precedence is as expected - BODMAS.

Logical tests like < and > tests will evaluate to 0.0 for false and 1.0 for true, allowing expressions like:

3 * (1 < 2) # returns 3.0
3 * (1 > 2) # returns 0.0

Minus implements both binary and unary operations.

See evaler_test.go for more examples of using operators.

Trigonometric Operators

The trigonometric operators supported are:

sin, cos, tan, ln, arcsin, arccos, arctan

For example:

cos(1)
sin(2-1)
sin(1)+2**2

See evaler_test.go for more examples of using trigonometric operators.

Variables

EvalWithVariables() allows variables to be passed into expressions, for example evaluate "x + 1", where x=5.

See evaler_test.go for more examples of using variables.

Issues

The math/big library doesn't have an exponent function ** and implenting one for big.Rat numbers is non-trivial. As a work around, arguments are converted to float64's, the calculation is done using the math.Pow() function, the result is converted to a big.Rat and placed back on the stack.