#math #expression #evaluation #parser


Mathematical expression parser and evaluator

21 releases

0.8.8 Jun 20, 2019
0.8.7 Mar 6, 2019
0.8.5 Feb 18, 2019
0.8.3 May 10, 2018

#72 in Parser tooling

50 downloads per month

MIT license

953 lines

Crates.io CI


Sample use:

use asciimath::{eval, scope, CustomFn};

let my_sum: CustomFn = |args| Ok(args.iter().sum());
let scope = scope!{
  "x" => -1,
  "my_sum" => my_sum,

assert_eq!(Ok(260.0), eval("my_sum(1, ((1 - x) ^ 2 ^ 3), 3)", &scope));


  • evaluation
  • implicit multiplication
  • support for variables, both single-letter and word variables
  • easily-defined custom functions
  • compiling expressions and evaluating with different sets of variables
  • f64 output
  • Baked-in essential functions and constants

High-Level Goals

Ease of use

This means that e.g. passing in variables to expressions and defining custom functions must be possible with minimum knowledge of this library's internals and abstractions. Errors must be helpful and relevant.


Focusing just on mathematical expressions will make it easy for this library to remain slim and deliver superior ergonomics.


Extensive testing and maximum precision must be a part of all the modules to prevent bugs and ensure consistency.


The items below will be considered after stabilization:

  • Non-mathematical expressions, like strings
  • More operators (e.g. ternary ? : )
  • Ability to simplify expressions
  • Derivatives, incl. second-order and third-order
  • Integration
  • Partial differentiation
  • Vector calculus
  • Matrices and vector spaces


While some great libraries aiming for similar goals do exist, they wouldn't reward me with such a fruitful Rust learning experience and imo sorely lack ergonomics.


The parser is loosely based on Dijkstra's "shunting yard" algorithm for converting infix expressions into postfix expressions. However, instead of going from infix to postfix strings, we parse the expression straight into an Abstract Syntax Tree.