#calculator #tokenizer #evaluator #lexer #tokenize

bin+lib xxcalc

Embeddable or standalone robust floating-point polynomial calculator

3 unstable releases

Uses old Rust 2015

0.2.1 Dec 5, 2016
0.2.0 Dec 5, 2016
0.1.1 Dec 3, 2016
0.1.0 Dec 3, 2016

#722 in Math

MIT license

135KB
1.5K SLoC

xxcalc-rs Build Status Coverage Status Rust Crate

An embeddable or a standalone robust floating-point polynomial calculator written in Rust.

This project is a Rust crate (library) which provides a floating-point numbers calculator in an easy to use API. Furthermore a basic unit of computation is a polynomial, so multiple arithmetic operations can be done using x symbols.

You can use this library in your own projects as mathematical evaluator or as a standalone, command line calculator by using xxcalc binary. Internally it uses hand-mande tokenizer and a Dijskatra's shunting-yard algorithm to convert infix form into Reverse Polish Notation, which is later evaluated. Please see the complete documentation for the implementation details.

Binary

xxcalc can be used to replace common bc Unix-utility. It has a support for addition, subtraction, multiplication, division and exponentation built-in.

Here is an example session with the calculator:

>>> (3+(4-1))*5
30
>>> 2 * x + 0.5 = 1
0.25
>>> 2x + 1 = 2(1-x)
0.25
>>> (x^3+2x-1)^3
x^9+6x^7-3x^6+12x^5-12x^4+11x^3-12x^2+6x-1
>>> bind((x^3+2x-1)^3, 1)
8
>>> log(128, 2)
7

Features

  • +, -, *, /, ^ operators on floating-point polynomials
  • scientific notation and negative numbers
  • polynomials representation using x symbol
  • solving linear equations solver using = operator
  • log(number, base), log10(number), bind(polynomial, x value) built-in functions
  • pi and e constants
  • input read from stdin (can be used for piping)
  • interactive readline-like mode with history

Installation

You need Rust programming language environment (works with stable, beta and nighly channels). If you don't have it already installed please use rustup.rs to easily install the complete environment. Then you can install packaged version from the crates.io using cargo install xxcalc --features interactive.

Compilation

The basic compilation process from the source is just running cargo in command line:

$ cargo build --release
$ ./target/release/xxcalc
2+2
4

Then you can copy the xxcalc binary wherever you prefer (like /usr/local/bin), however this builds a non-interactive version of the binary, with no support for history completion.

In order to build the calculator with support for history (stored in ~/.xxcalcrs_history), you need it enable interactive feature (it will automatically resolve dependencies to rustyline):

$ cargo build --release --features interactive
$ ./target/release/xxcalc
>>> 2+2
4

Benchmarks

Coming soon. It is at least 2.5 times faster than bc and incomparable quicker than GNU Octave.

Library

The xxcalc calculator provides a clean and well documented API. Please see its documentation.

The library builds on stable, beta and nighly channels and require no additional dependencies, unless interactive feature is enabled. Therefore you can easily add mathematical expressions evaluator into your project.

The calculator is meant to be extensible -- you can register your own functions or constants. You can even change tokenizer, parser or an evaluator into your own. Furthermore a Polynomial is a standalone type which can be used in your projects without usage of other features of the calculator.

Usage

Add xxcalc as dependency in your Cargo.toml, then just use xxcalc crate and the parts you need.

# Cargo.toml
[dependencies]
xxcalc = "0.2.1"
extern crate xxcalc;

use xxcalc::linear_solver::LinearSolver;
use xxcalc::calculator::Calculator;

fn main() {
  println!("The result is {}", LinearSolver.process("2+2").unwrap());
}

Many usage hints and scenarios are available in the documentation.

Tests

The projects is thoroughly unit tested, some examples are directly provided in the documentation. Use cargo test to run the unit tests.

If you have a Rust nighly compiler you can run some built-in bencharks using cargo bench. Some more extensive benchmarks (using large expressions) can be run using carbo bench -- --ignored.

License

The project is licensed under MIT license, the author is Amadeusz Juskowiak juskowiak@amadeusz.me, feel free to ask any questions or to hire me.

Dependencies

~105–510KB
~12K SLoC