387b32f337
Switch algorithm to match modp_b64 for performance purposes |
||
---|---|---|
.github/workflows | ||
include | ||
scripts | ||
test | ||
.gitignore | ||
CMakeLists.txt | ||
CPPLINT.cfg | ||
LICENSE | ||
README.md |
base64
A simple approach to convert strings from and to base64. Header only c++ library (single header).
Usage
#include <iostream>
#include "base64.hpp"
int main() {
auto encoded_str = base64::to_base64("Hello, World!");
std::cout << encoded_str << std::endl; // SGVsbG8sIFdvcmxkIQ==
auto decoded_str = base64::from_base64("SGVsbG8sIFdvcmxkIQ==");
std::cout << decoded_str << std::endl; // Hello, World!
}
Notes
This library relies on C++17 but will exploit some C++20 features if available (e.g. bit_cast
).
There are many implementations available and it may be worth looking at those. A benchmark of various c/c++ base64 implementations can be found at https://github.com/gaspardpetit/base64/
This implementation here adopts the approach of Nick Galbreath's modp_b64
library also used by chromium (e.g. https://github.com/chromium/chromium/tree/main/third_party/modp_b64 ) but offers it as a c++ single header file. This choice was based on the good computational performance of the underpinning algorithm. We also decided to avoid relying on a c++ union
to perform type punning as this, while working in practice, is strictly speaking undefined behaviour in c++: https://en.wikipedia.org/wiki/Type_punning#Use_of_union
Faster c/c++ implementations exist althrough these likely exploit simd / openmp or similar acceleration techniques:
- https://github.com/aklomp/base64
- https://github.com/lemire/fastbase64 (From a blog post by the authors: "My understanding is that our good results have been integrated in Klomp’s base64 library.")
- Other implementations related to the one by lemire: https://github.com/WojciechMula/base64-avx512 and https://github.com/WojciechMula/base64simd
- https://github.com/powturbo/Turbo-Base64 (Note that this is licensed under GPL 3.0)
Many other C++ centric appraches exists although they seem to focus on readibility or genericity at the cost of performance, e.g.:
- https://github.com/matheusgomes28/base64pp (C++20 library from which we borrowed the unit test code)
- https://github.com/ReneNyffenegger/cpp-base64 (Implementation that works with older C++ versions)
- https://github.com/azawadzki/base-n (more generic baseN such as N=16 and N=32)