coyorkdow/result

rust style error handling for cpp

result

Rust style error handling for cpp. It provides result::Result which supports pattern matching error handling and monadic operation.

cpp17 required

Basic Usage

#include <iostream>
#include <limits>

#include "result.hpp"
using result::Result;

// define a Result of int64, which error type is int
Result<int64_t, int> Multiply(int64_t a, int64_t b) {
    int64_t res = a * b;
    if (a != 0 && res / a != b) {
        return {result::Err{}, -1}; // return error
    }
    return res; // return result
}

int main() {
    // do pattern matching
    auto v = Multiply(1000, 1000).match(
        result::Ok()  = [](int64_t v) { return v; },
        result::Err() = [](int err) { std::cerr << "arithmetic overflow\n"; return err; }
    );
    std::cout << v << '\n';
    v = Multiply(std::numeric_limits<int64_t>::max(), 1000).match(
        result::Ok()  = [](int64_t v) { return v; },
        result::Err() = [](int err) { std::cerr << "arithmetic overflow\n"; return err; }
    );
    std::cout << v << '\n';
}

the return type of result::Result::match can be automatically deduced through the return type of the pattern handler.

The output will be

1000000
arithmetic overflow
-1

Using result::Error

result::Error is an type that designed to receive any type of the error. By using pattern matching it's easy to determine what kind of error that exactly is.

First, let's define a function template that returns Result<T, result::Error>

template <class T>
Result<T, result::Error> ElementaryArithmeticFromZeroToMillion(char op, T x, T y) {
  const T million = 1'000'000;
  if (!(x >= 0 && x <= million) || !(y >= 0 && y <= million)) {
    return {result::Err{}, result::OutOfRangeError{}};
  }
  using res_t = Result<T, result::Error>;
  T res{};
  switch (op) {
    case '+':
      res = x + y;
      return (res >= 0 && res <= million) ? res_t{res} : res_t{result::Err{}, result::RangeError{}};
    case '-':
      res = x - y;
      return (res >= 0 && res <= million) ? res_t{res} : res_t{result::Err{}, result::RangeError{}};
    case '*':
      res = x * y;
      return (res >= 0 && res <= million) ? res_t{res} : res_t{result::Err{}, result::RangeError{}};
    case '/':
      if (y == 0) {
        return {result::Err{}, result::DivideByZeroError{}};
      }
      res = x / y;
      return (res >= 0 && res <= million) ? res_t{res} : res_t{result::Err{}, result::RangeError{}};
    default:
      return {result::Err{}, result::InvalidArgumentError{}};
  }
}

auto v1 = ElementaryArithmeticFromZeroToMillion('+', 4, 3).match(
        result::Ok()  = [](int v) { return v; },
        result::Err() = [](result::Error&&) { return 0; });
std::cout << v1 << '\n'; // 7

auto v2 = ElementaryArithmeticFromZeroToMillion('-', 1.5, 10.5).match(
        result::Ok() = [](double v) { return v; },
        result::Err() =
            [](result::OutOfRangeError) {
              std::cout << "the arguments of calculating v2 out of range\n";
              return -1;
            },
        result::Err() = /* match here */
            [](result::RangeError) {
              std::cout << "the result of calculating v2 out of range\n";
              return -2;
            },
        result::Err() = /* the default error pattern. must be contained otherwise compile error */
            [](result::Error&&) {
              std::cout << "unknown error\n";
              return -3;
            });
std::cout << v2 << '\n'; // -2

result::Error itself also supports pattern matching.

auto res = ElementaryArithmeticFromZeroToMillion('^', 1, 0);
res.error().match(
    [](result::OutOfRangeError) {
      std::cout << "the arguments out of range\n";
      return 0;
    },
    [](result::InvalidArgumentError) { /* match here */
      std::cout << "the arguments are invalid\n";
      return 0;
    },
    [](result::Error) {
      std::cout << "unknown error\n";
      return 0;
    });

Monadic operation

auto res = ElementaryArithmeticFromZeroToMillion('^', 1, 0);
auto v3 = res.or_else([](const result::Error&) {
                std::cout << "op ^ failed, change to op /\n";
                return ElementaryArithmeticFromZeroToMillion('/', 1, 0);
              })
              .and_then([](int v) {
                std::cout << "op / succeed\n";
                return Result<int, result::Error>{v};
              })
              .or_else([](const result::Error&) {
                std::cout << "op / failed, change to op +\n";
                return ElementaryArithmeticFromZeroToMillion('+', 1, 0);
              })
              .and_then([](int v) {
                std::cout << "op + succeed, time the result by 10\n";
                return ElementaryArithmeticFromZeroToMillion('*', v, 10);
              })
              .or_else([](const result::Error&) {
                std::cout << "eventually failed, set value as -1\n";
                return Result<int, result::Error>{-1};
              })
              .value();
std::cout << "the final reuslt is " << v3 << '\n';

The output will be

op ^ failed, change to op /
op / failed, change to op +
op + succeed, time the result by 10
the final reuslt is 10