yardanico / nim-mathexpr

Tiny math expression evaluator library implemented in pure Nim

Mathexpr, a math expression evaluator library in Nim

Mathexpr is a relatively small mathematical expression evaluator library written in Nim without any third-party dependencies. It has a lot of predefined math functions and some constants, and you can also define your own.

Installation

To install mathexpr, simply run:

$ nimble install mathexpr

Documentation

https://yardanico.github.io/nim-mathexpr/mathexpr.html

nim-mathexpr's Issues

Incorrect precedence of power operator

https://github.com/Yardanico/nim-mathexpr/blob/master/src/mathexpr.nim#L316

Here, i can see that power operator (^) is parsed at the same level of precedence as multiplication operators, which is not correct from mathematical POV.

This means that: x * y ^ z will be evaluated as (x * y) ^ z when it should clearly be x * (y ^ z), different interpretations can lead to dramatically differing results, which may not be what user, who are familiar with math, may expect. Coincidentally, I could not find a test case that covers it -_-

The proposed solution is to either parse ^ in parseFactor level or add another function between parseTerm and parseFactor specifically for this operator.

Simple sum gives wrong result

import mathexpr
let e = newEvaluator()
echo e.eval("0.0767999999999999+0")
echo e.eval("0.076799999999999+0")

gives me as output

767999999999999.0
0.076799999999999

The first line is obviously incorrect.

negate operator precedes over higher priority operators

expression: -10^2
expected answer: -100
mathexpr: 100

tested on mathexpr 1.3.2

I wrote a library that does not have this issue

support comparison ops and returning bools

great library!

any chance of supporting e.g. : eval("a > 22", {"a": 5.0})?
I can use, e.g.:

proc gt(args: seq[float]): float =
  if args[0] <= args[1]: return 0.0
  return 1

mathexpr.functions["gt"] = gt
echo eval("gt(a, 4)", {"a": 5.0})

but it would be nice to have support for this. I see that everything is float as it is now, but thought you might have some ideas on this.
thanks.

all functions with expected len -1 are broken

functions like 'min' that checkArgs(-1) are broken because they evaluate true here.

I help myself by patching the existing functions like

e.addFunc("min", proc (a:seq[float]): float = a.min)

but I guess it would be nice (and easy) to fix CheckArgLen.

mathexpr raises `Unexpected character ''` exception when hits an undefined identifier

in REPL:

mathexpr>>>   ident 
error: Unexpected character ``

tested in mathexpr 1.3.2