Improve error feedback for base64 decoding

[prince-chrismc]

This problem was very difficult to detect. Because prints were the same. But strings weren't been the same. My string has a lot of white characters inside (Not only on end and begin).

Originally posted by @hanusek in #248 (comment)

---

Ideally the error messages would include a position and element that is invalid. Currently the messages are tok generic

[ItsAMeMarcel]

Hi all,

I encountered this issue as well and came up with three potential improvements:

1. Add char to the runtime error

inline uint32_t index(const std::array<int8_t, 256>& rdata, char symbol) {
	auto index = rdata[static_cast<unsigned char>(symbol)];
	if (index <= -1) {
		std::stringstream ss;
		ss << "Invalid input: '" << symbol << "' (0x" << std::hex << std::setw(2) << std::setfill('0')
		   << static_cast<int>(symbol) << ") not within alphabet";
		throw std::runtime_error(ss.str());
	}
	return static_cast<uint32_t>(index);
}

2. Add char and offset to runtime error

inline uint32_t index(const std::array<int8_t, 256>& rdata, char symbol, size_t offset) {
	auto index = rdata[static_cast<unsigned char>(symbol)];
	if (index <= -1) {
		std::stringstream ss;
		ss << "Invalid input: '" << symbol << "' (0x" << std::hex << std::setw(2) << std::setfill('0')
		   << static_cast<int>(symbol) << ") at position " << std::dec << offset << " not within alphabet";
		throw std::runtime_error(ss.str());
	}
	return static_cast<uint32_t>(index);
}

3. Same as 2 but handover base and offset

inline uint32_t index(const std::array<int8_t, 256>& rdata, const std::string& base, size_t offset) {
	auto symbol = base[offset];
	auto index = rdata[static_cast<unsigned char>(symbol)];
	if (index <= -1) {
		std::stringstream ss;
		ss << "Invalid input: '" << symbol << "' (0x" << std::hex << std::setw(2) << std::setfill('0')
		   << static_cast<int>(symbol) << ") at position " << std::dec << offset << " not within alphabet";
		throw std::runtime_error(ss.str());
	}
	return static_cast<uint32_t>(index);
}

Personally, I prefer option 3, as it provides the most comprehensive information, which can be extremely helpful for debugging.

I am less inclined towards option 2 because it introduces a function argument solely for error handling, which might feel unnecessary unless it's strictly needed.

What are your thoughts on these approaches?