meval

This Rust crate provides a simple math expression parsing and evaluation. Its main goal is to be convenient to use, while allowing for some flexibility. Currently works only with f64 types. A typical use case is the configuration of numerical computations in Rust, think initial data and boundary conditions, via config files or command line arguments.

Documentation

• Full API documentation

Installation

Simply add the corresponding entry to your Cargo.toml dependency list:

[dependencies]
meval = "0.2"

Requires Rust 1.26.

Simple examples

fn main() {
    let r = meval::eval_str("1 + 2").unwrap();

    println!("1 + 2 = {}", r);
}

Need to define a Rust function from an expression? No problem, use Expr for this and more:

fn main() {
    let expr: meval::Expr = "sin(pi * x)".parse().unwrap();
    let func = expr.bind("x").unwrap();

    let vs: Vec<_> = (0..100+1).map(|i| func(i as f64 / 100.)).collect();

    println!("sin(pi * x), 0 <= x <= 1: {:?}", vs);
}

Custom constants and functions? Define a Context!

use meval::{Expr, Context};

let y = 1.;
let expr: Expr = "phi(-2 * zeta + x)".parse().unwrap();

// create a context with function definitions and variables
let mut ctx = Context::new(); // built-ins
ctx.func("phi", |x| x + y)
   .var("zeta", -1.);
// bind function with a custom context
let func = expr.bind_with_context(ctx, "x").unwrap();
assert_eq!(func(2.), -2. * -1. + 2. + 1.);

For functions of 2, 3, and N variables use Context::func2, Context::func3 and Context::funcn, respectively. See Context for more options.

If you need a custom function depending on mutable parameters, you will need to use a Cell:

use std::cell::Cell;
use meval::{Expr, Context};
let y = Cell::new(0.);
let expr: Expr = "phi(x)".parse().unwrap();

let mut ctx = Context::empty(); // no built-ins
ctx.func("phi", |x| x + y.get());

let func = expr.bind_with_context(ctx, "x").unwrap();
assert_eq!(func(2.), 2.);
y.set(3.);
assert_eq!(func(2.), 5.);

Supported expressions

meval supports basic mathematical operations on floating point numbers:

• binary operators: +, -, *, /, % (remainder), ^ (power)
• unary operators: +, -, ! (factorial)

It supports custom variables and functions like x, weight, C_0, f(1), etc. A variable or function name must start with [a-zA-Z_] and can contain only [a-zA-Z0-9_]. Custom functions with a variable number of arguments are also supported.

Build-ins (given by the context Context::new() and when no context provided) currently supported:

• functions implemented using functions of the same name in Rust std library:

  • sqrt, abs
  • exp, ln, log10
  • sin, cos, tan, asin, acos, atan, atan2
  • sinh, cosh, tanh, asinh, acosh, atanh
  • floor, ceil, round
  • signum

• other functions:

  • max(x, ...), min(x, ...): maximum and minimumum of 1 or more numbers

• constants:

  • pi
  • e

Deserialization

Expr supports deserialization using the serde library to make flexible configuration easy to set up, if the feature serde is enabled (disabled by default).

#[macro_use]
extern crate serde_derive;
extern crate toml;
extern crate meval;
use meval::{Expr, Context};

#[derive(Deserialize)]
struct Ode {
    #[serde(deserialize_with = "meval::de::as_f64")]
    x0: f64,
    #[serde(deserialize_with = "meval::de::as_f64")]
    t0: f64,
    f: Expr,
}

fn main() {
    let config = r#"
        x0 = "cos(1.)"
        t0 = 2
        f = "sin(x)"
    "#;
    let ode: Ode = toml::from_str(config).unwrap();

    assert_eq!(ode.x0, 1f64.cos());
    assert_eq!(ode.t0, 2f64);
    assert_eq!(ode.f.bind("x").unwrap()(2.), 2f64.sin());
}

Related projects

This is a toy project of mine for learning Rust, and to be hopefully useful when writing command line scripts. There is no plan to make this anything more than math expression -> number "converter". For more advanced uses, see:

• evalexpr -- A more complete expression evaluator that supports value types, boolean operators, strings, assignment operators, ...
• dyon -- A rusty dynamically typed scripting language
• gluon -- A static, type inferred programming language for application embedding
• rodolf0/tox -- another shunting yard expression parser

License

This project is dual-licensed under the Unlicense and MIT licenses.

You may use this code under the terms of either license.